The relationship between total food intake (g) - CEBC - CNRS
The relationship between total food intake (g) - CEBC - CNRS
The relationship between total food intake (g) - CEBC - CNRS
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Université de POITIERS<br />
UFR des Sciences Fondamentales et Appliquées<br />
Thèse présentée par Olivier Lourdais<br />
Pour obtenir le grade de docteur de l’Université de POITIERS<br />
Coûts de la reproduction, gestion des ressources et<br />
fréquence des épisodes reproducteurs<br />
chez la vipère aspic (Vipera aspis)<br />
Soutenue le 19 novembre 2002 devant la commission d’examen :<br />
Miaud C Maître de conférence, <strong>CNRS</strong> Rapporteur<br />
Gasc JP Professeur de rang 1, <strong>CNRS</strong> Rapporteur<br />
Baehr JC Professeur à l’Université de Poitiers Examinateur<br />
Mazin JM Directeur de recherche au <strong>CNRS</strong> Examinateur<br />
Zuffi MAL Conservateur, Muséum de Pise Examinateur<br />
Bonnet X Chargé de recherches <strong>CNRS</strong> Directeur de thèse<br />
Invités:<br />
Shine R Co-directeur de thèse, Professeur à l’Université de Sydney<br />
Naulleau G Chargé de recherches <strong>CNRS</strong><br />
Weimerskirch H Directeur de recherches <strong>CNRS</strong><br />
1
Sommaire<br />
Remerciements 3<br />
Remarque 4<br />
Publications 5<br />
Résumé 8<br />
Introduction générale<br />
A. Sélection, évolution et traits d’histoire de vie 11<br />
B. Optimisation de l’investissement et coût de la reproduction 13<br />
C. Dimension physiologique des compromis et des systèmes de gestion<br />
de la ressource 15<br />
D. Connexion entre effort reproducteur et coût de la reproduction 18<br />
E. Intérêt d’une perspective ectothermique 21<br />
I. Présentation de l’espèce et des méthodes d’étude<br />
A. Résumé du Chapitre 26<br />
B. Position systématique 27<br />
C. Répartition de l’espèce et des populations l’études 28<br />
D. Biologie de la reproduction 29<br />
E. Méthodes d’étude 37<br />
II. Le système d’allocation de l’énergie<br />
A. Résumé du chapitre 47<br />
B. Article 1 : short-term versus long-term effects of <strong>food</strong> <strong>intake</strong> on<br />
reproductive output in a viviparous snake (Vipera aspis) 50<br />
C. Article 2 : when does a reproducing female viper (Vipera aspis) “decide"<br />
on her litter size 81<br />
D. Article 3 : capital breeding and reproductive effort in a variable<br />
environnment: a longitudinal study in the aspic viper (Vipera aspis) 98<br />
2
III. Les coûts de la reproduction: amplitude et degré de dépendance<br />
avec la fécondité<br />
A. Résumé du chapitre 124<br />
B. Article 4 : reproduction in a typical capital breeder, costs, currencies<br />
and complication in the aspic viper 128<br />
C. Article 5 : costs of anorexia during pregnancy in a viviparous snake<br />
(Vipera aspis) 156<br />
D. Article 6 : thermoregulation and metabolism in a viviparous snake,<br />
Vipera aspis: evidence for fecundity-independent costs 170<br />
E. Article 7 : what is the appropriate time scale for measuring costs of<br />
reproduction in a capital breeder such as the aspic viper ? 198<br />
IV. Les déterminants de la tendance semélipare femelle: description et<br />
implications démographiques<br />
A. Résumé du chapitre 217<br />
B. Article 8 : comparaisons des tactiques demographiques de la vipère aspic<br />
(Vipera aspis): y’a t’il un avantage a etre semélipares ? 220<br />
C. Article 9 : do sex divergences in ecophysiologie translate into sexdimorphic<br />
demographic patterns? 240<br />
V Discussion-conclusion<br />
Reproduction sur réserves, coûts de la reproduction et évolution vers la<br />
seméliparité 264<br />
Bibliographie 274<br />
Annexe<br />
Article : Natural thermal conditions influence embryonic development in a<br />
viviparous snake (Vipera aspis) 311<br />
3
Remerciements :<br />
Plutôt que d’écrire de très longs remerciements, un peu pénibles à lire, j’ai préféré<br />
opter pour une formulation plus concise. Les raisons d’un tel choix sont<br />
multifactorielles et j’offre ci-dessous des versions alternatives plus ou moins<br />
crédibles :<br />
Je tiens à remercier l’ensemble de l’humanité pour m’avoir fournit une aide et un<br />
réconfort quotidien dont l’ampleur est telle qu’une page ne suffirait évidemment<br />
pas. Je vous remercie donc tous : je vous aime.<br />
Il existe un véritable risque à l’écriture de remerciements trop longs. En effet, si je<br />
dépasse la fin de cette page 3, l’ensemble de ma pagination (rentrée à la main)<br />
va être modifiée. Les contraintes énergétiques associées à la remise à jour du<br />
sommaire sont telles que de fortes pressions de sélection favorisent des<br />
remerciements courts et non répétitifs (à tendance semélipare) : Merci.<br />
Enfin, d’un point de vue historique, au moment où je me suis engagé dans ce<br />
travail avec mon ami Xavier Bonnet, l’herpétologie n’était pas vraiment une<br />
discipline académiquement reconnue et le département d’herpétologie de Chizé<br />
était au bord de la fermeture. La bonne conduite de ce projet est donc liée à une<br />
poignée d’humains ayant joué un rôle fondamental. J’ai déjà pris soins de<br />
remercier individuellement ces personnes et en rajouter deviendrait de la<br />
flagornerie.<br />
De façon plus proximale, je tiens à remercier mon co-équipier François<br />
Brischoux pour m’avoir fournit une aide indispensable à la finition du manuscrit<br />
et au respect des délais.<br />
Merci à mes deux rapporteurs Claude Miaud et Jean Pierre Gasc, pour avoir<br />
accepté de lire et corriger un tel pavé.<br />
Merci à mon ami Hassan pour la phrase qu’il m’a dite un soir, l’air profondément<br />
dubitatif en autopsiant le cadavre d’une ancienne version de mon schéma de la<br />
page 267 : « Olivier, tu sais, la vie c’est comme ton schéma, c’est compliqué »<br />
…à mes parents et à Pierre pour avoir rendu possible cette véritable odyssée.<br />
4
Remarque :<br />
Le format de cette thèse s’inscrit dans le cadre actuel des efforts de production<br />
scientifique (publications) associés à un doctorat en écologie. Ce travail repose donc<br />
sur la présentation d’une série d’articles (7 publiés, 1 soumis et 1 en cours<br />
d’élaboration). Les articles publiés ou soumis sont présentés sous leur structure<br />
classique et rédigés en anglais. L’introduction, la présentation de l’espèce, les<br />
résultats récents ainsi que la discussion-conclusion ont été spécifiquement rédigés<br />
en français pour ce mémoire. Les différents travaux ont été regroupés en trois<br />
grands chapitres intimement connectés, qui retracent l’approche mise en oeuvre pour<br />
examiner la stratégie reproductrice de l’espèce et élaborer un scénario évolutif. Afin<br />
de faciliter le cheminement, chaque chapitre est précédé d’un résumé des principaux<br />
résultats et éléments de réflexion. Afin de limiter l’hétérogénéité associée à un tel<br />
regroupement d’article, tous les travaux ont été mis au même format et une<br />
pagination continue a été choisie. Enfin, une bibliographie globale est présentée à la<br />
fin du mémoire. J’espère que la présentation choisie sera à la hauteur d’un défi fort<br />
difficile, celui de donner à une thèse sur articles une organisation formelle cohérente<br />
et fonctionnelle.<br />
5
Publications<br />
Mes premières implications dans l’étude des serpents débutent très tôt dans mon<br />
cursus Universitaire (1996, stage de DEUG). Dès mon entrée en DEA à Chizé (1998)<br />
j’ai pu commencer à me consacrer au travail de publication. Ma position d’auteur<br />
associé dans ces travaux publiés trouve ainsi son origine dans une participation<br />
active de ma part à la fois dans la récolte des données, les analyses statistiques et la<br />
rédaction. Par la suite, je me suis impliqué de façon plus personnalisée avec la<br />
publication de quatre articles dont je suis l’auteur principal.<br />
Liste des publications sur la thèmatique de thèse :<br />
1. Lourdais O, Bonnet X, DeNardo D, Naulleau G. (2002) Does sex differences in<br />
reproductive eco-physiology translate in different demographic patterns ?<br />
Population Ecology, in press<br />
2. Lourdais O, Bonnet X, Shine R & Taylor E. (2002) When does a reproducing<br />
female viper (Vipera aspis) "decide" on her litter size? Journal of Zoology, London<br />
in press<br />
3. Lourdais O, Bonnet X, Shine R , DeNardo D, Naulleau G & Guillon M. (2002).<br />
Capital-breeding and reproductive effort in a variable environment: a longitudinal<br />
study of a viviparous snake. Journal of Animal Ecology 71 : 470-479.<br />
4. Lourdais O, Bonnet X, Doughty P. (2002). Costs of anorexia during pregnancy in<br />
a viviparous snake (Vipera aspis). Journal of Experimental Zoology 292 : 487-<br />
493.<br />
6
5. Bonnet X, Lourdais O, Shine R. & Guy Naulleau. (2002). Reproduction in a<br />
typical capital breeder : costs, currencies and complications in the aspic viper<br />
Ecology 83 : 2124-2135.<br />
6. Bonnet X., Naulleau G. & Lourdais O. (2002). <strong>The</strong> benefits of complementary<br />
techniques: using capture-recapture and physiological approaches to understand<br />
costs of reproduction in the asp viper. Biology of the Vipers, in press<br />
7. Bonnet X., Shine R, Lourdais O & Naulleau G (2002) Measures of reproductive<br />
allometry are sensitive to sampling Bias. Functionnal Ecology, in press<br />
8. Aubret F, Bonnet X, Shine R & Lourdais O. (2002) Fat is sexy for females but<br />
not males: the influence of body reserves on reproduction in snakes (Vipera<br />
aspis). Hormones and Behaviors 42 : 135-147.<br />
9. Bonnet X., Naulleau G., Shine R. & Lourdais O. (2001). Short-term versus long-<br />
term effects of <strong>food</strong> <strong>intake</strong> on reproductive output in a viviparous snake (Vipera<br />
aspis). Oikos 92 : 297-308.<br />
10. Bonnet X., Naulleau G., Shine R. & Lourdais O. (2000). What is the appropriate<br />
time scale for measuring costs of reproduction in a capital breeder?<br />
Evolutionary Ecology 13 : 485-497.<br />
11. Bonnet X., Naulleau G., Shine R. & Lourdais O. (2000). Reproductive versus<br />
ecological advantages to larger body size in female Vipera aspis. Oikos 89 : 509-<br />
518.<br />
12. Naulleau G., Bonnet X., Vacher-Vallas M, Shine R. & Lourdais O. (1999). Does<br />
less-than-annual production of offspring by female vipers (Vipera aspis) mean<br />
less-than-annual mating? Journal of Herpetology 33 : 688-691.<br />
7
13. Bonnet X., Naulleau G., Lourdais O. & Vacher-Vallas M. (1999). Growth in the<br />
asp viper (Vipera aspis L.): insights from long term field study. Current Studies in<br />
Herpetology. C. Miaud et R. Guyetant eds. pp. 63-69.<br />
Autres publications<br />
14. Bonnet X, Shine R, Lourdais O. (2002). Taxonomic Chauvinism. Trends in<br />
Ecology and Evolution 17 : 1-3.<br />
15. Bonnet X, Pearson D, Ladyman M, Lourdais O, & Bradshaw D. (2002). Heaven<br />
for serpents? A mark-recapture study of Tiger Snakes (Notechis scutatus) on<br />
Carnac Island, Western Australia. Austral Ecology 27 : 442-450.<br />
16. Pearson D., Shine R., Bonnet X., A. Williams, B. Jennings & O. Lourdais. (2000).<br />
Ecological notes on crowned snakes, Elapognathus coronatus, from the<br />
Archipelago of the Recherche in southwestern Australia. Australian Zoologist 31 :<br />
610-617.<br />
Enfin, signalons deux travaux actuellement soumis:<br />
17. Lourdais O, Shine R, Bonnet X, Guillon G, & Guy Naulleau. Natural thermal<br />
conditions influence embryonic development in a viviparous snake (Vipera aspis).<br />
Functionnal Ecology<br />
18. Ladyman M , Bonnet X, Lourdais O, Bradshaw D & Naulleau G. Gestation,<br />
thermoregulation and metabolism in a viviparous snake, Vipera aspis: evidence<br />
for fecundity-independent costs. Physiological and Biochemical Zoology<br />
8
Résumé :<br />
Le nombre d'épisodes reproducteurs au cours de l'existence d'un organisme<br />
constitue un trait d'histoire de vie majeur. On distingue ainsi des espèces<br />
semélipares (une seule reproduction et la mort de l’organisme), et d'autre itéropares<br />
(reproductions répétées). La vipère aspic occupe une position intermédiaire avec une<br />
faible fréquence de reproduction (tous les 2-4 ans) et une tendance marquée vers la<br />
seméliparité. D'un point de vue évolutif, il est légitime de s'interroger sur les<br />
avantages d’une telle stratégie où les dépenses reproductrices sont accrues et peu<br />
fréquentes.<br />
Nos travaux sur la vipère aspic suggèrent une relation directe entre la<br />
fréquence reproductrice et la nature des contraintes énergétiques et écologiques de<br />
la reproduction. Dès l’engagement dans la folliculogénèse, la vipère aspic va être<br />
confrontée à des activités très coûteuses (exposition aux prédateurs, coûts<br />
métaboliques) qui reflètent des changement profonds de la physiologie et du<br />
comportement. De façon surprenante, si ces changements sont directement liés au<br />
statut reproducteur, ils ne sont pas dépendants de l’effort reproducteur et du nombre<br />
de jeunes produits. En outre, ces coûts particuliers, s’expriment sur un pas de temps<br />
complexe impliquant des composantes directes (l’année de la reproduction), et des<br />
composantes délayées (post-reproduction). Ces résultats viennent donc confirmer<br />
l’hypothèse de Bull et Shine (1979) selon laquelle les systèmes à faible fréquence<br />
reproductrice émergent lorqu’il existe des contraintes reproductrices (coûts) dont<br />
l’amplitude est élevée et indépendante de la fécondité.<br />
Notre idée originale repose sur une connexion du modèle de Bull et Shine<br />
avec les stratégies d'acquisition et d'allocation de l'énergie. En effet, si le nombre de<br />
reproduction est réduit, l’organisme aura un intérêt évident à investir massivement<br />
9
son énergie pour garantir le succès de ses quelques opportunités de reproductions.<br />
Une possibilité de répondre à une telle demande passe par la sélection de système<br />
de gestion de la ressource particuliers, impliquant notamment le stockage de<br />
réserves corporelles. Nos résultats supportent largement l’existence d’une telle<br />
relation évolutive entre les coûts indépendants de la fécondité, les systèmes à faible<br />
fréquences de reproductions et les stratégies de capitalisation de l’énergie (“Capital-<br />
breeding”). Cette étude apporte donc des éléments de réponses pertinents sur les<br />
conditions d’émergence des systèmes de reproduction “extrêmes” et sur la transition<br />
évolutive vers la seméliparité.<br />
10
Introduction générale<br />
Un saumon rouge (Oncorhynchus nerka) mourant d’épuisement à la suite<br />
d’un épisode reproducteur unique (seméliparité)<br />
“...expenditures on reproductive processes must be in functional harmony with each<br />
other and worth costs, in relation to the long range reproductive interest; and the use<br />
of resources for somatic processes is favored to the extent that somatic survival, and<br />
perhaps growth, are important for future reproduction.”<br />
Willians 1966b: 687<br />
11
A. Sélection, évolution et traits d’histoire de vie<br />
Le monde vivant est caractérisé par une étonnante diversité qui se manifeste sur une<br />
suite de niveaux hiérarchisés, depuis l’échelon moléculaire jusqu’au fonctionnement<br />
des écosystèmes. Une telle diversité a depuis longtemps attiré l’attention des<br />
philosophes et scientifiques. C’est avec Darwin (1859) qu’une explication puissante<br />
et unificatrice - la théorie de l’évolution - a été formulée afin de comprendre et<br />
interpréter cette variabilité du vivant. L’évolution des organismes fait intervenir<br />
l’action clé de la sélection naturelle qui opère par multiplication différentielle des êtres<br />
vivants selon leurs aptitudes plus ou moins grandes à transmettre leurs gènes à la<br />
génération suivante. Les processus évolutifs s’opèrent par modifications et<br />
diversifications continuelles des organismes, face à des pressions sélectives variées.<br />
La théorie de l’évolution est la seule capable de fournir un sens commun à tous les<br />
domaines de la biologie (Mayr 1963). En outre, elle offre le fondement conceptuel de<br />
la biologie évolutive actuelle, un champs d’investigation très étendu dont la<br />
puissance explicative tire profit de l’intégration de disciplines complémentaires<br />
(génétique, écologie, physiologie).<br />
L’étude des traits d’histoire de vie (Lessells 1991; Roff 1992; Stearns 1992)<br />
est un domaine central en biologie évolutive qui connait un véritable essor depuis<br />
une trentaine d’année. Par “traits d’histoire de vie” , on désigne un ensemble<br />
complexe de caractères directement “impliqués“ dans la reproduction et la survie des<br />
organismes et donc la contribution en terme de descendance. Si tous les êtres<br />
vivants doivent survivre et se reproduire, il existe cependant des variations majeures<br />
dans certains traits comme la fécondité, la taille des jeunes, l’investissement<br />
reproducteur ou la durée de la vie. Ces variations sont particulièrement évidentes si<br />
l’on compare des espèces présentant des “stratégies d’histoire de vie” très<br />
12
contrastées comme un oiseau longévif et un insecte ayant une durée de vie<br />
inférieure à quelques jours (Stearns 1992).<br />
La théorie des traits d’histoire de vie cherche donc à fournir une explication<br />
évolutive pour interpréter la diversité et la complexité des cycles de vie entre les<br />
espèces, à élucider le mécanisme commun d’une relation générale liant l’âge, la taille<br />
et la mortalité avec les performances reproductrices. L’évolution et la partition de<br />
l’effort de reproduction constitue dans ce cadre un centre d’intérêt majeur. Pourquoi<br />
certaines espèces investissent une quantité énorme d’énergie dans la production<br />
d’un nombre très élevé de jeunes alors que d’autres espèces plus prudentes vont<br />
allouer une quantité plus réduite d’énergie dans des reproductions peu fréquentes<br />
et/ou des tailles de portées réduites? Le schéma d’investissement âge-spécifique<br />
dans la reproduction et notamment la fréquence des épisodes reproducteurs est<br />
particulièrement variable selon les espèces. Classiquement, on réalise une<br />
distinction entre les organismes dits itéropares et d’autres semélipares (Cole 1954).<br />
Alors que chez les premiers la vie, reproductrice est constituée d’une succession<br />
d’épisodes reproducteurs, il n’existe chez les seconds qu’une seule opportunité de<br />
reproduction associée à la mort systématique de l’organisme. La seméliparité est<br />
considérée comme un état dérivé de l’itéroparité. Les avantages de ces stratégies<br />
contrastées ont été largement discutés (Cole 1954, Stearns 1992) et de nombreux<br />
modèles mathématiques ont été formulés pour comprendre les facteurs favorisants<br />
l’une ou l’autre (Cole 1954 ; Gadgil & Bossert 1970 ; Bryant 1971 ; Charnov &<br />
Schaffer 1973 ; Ranta et al. 2002a,b). L’élucidation des pressions sélectives<br />
favorisant la transition entre itéroparité et seméliparité reste néanmoins une question<br />
complexe, notamment du fait de l’apparition de systèmes semélipares dans des<br />
groupes phylogénétiques très éloignés comme les insectes, les annélides, les<br />
mollusques ou encore les poissons ( Boyle 1983, 1987 ; Corkum et al. 1997 ; Andries<br />
13
2001, Crespi & Teo, 2002). La reconstitution de scénarios évolutifs satisfaisant pour<br />
comprendre la transition vers un mode de reproduction induisant la mort de<br />
l’organisme est un problème situé au cœur même de l’étude des traits d’histoire de<br />
vie (Crespi & Teo 2002). L’objectif principal du présent travail est d’apporter des<br />
éléments de réponses et proposer un scénario évolutif possible.<br />
B. optimisation de l’investissement<br />
reproducteur et coût de la reproduction<br />
On constate généralement de fortes variations dans les performances reproductrices<br />
des individus (Darwin 1859). De telles variations suggèrent l’existence de contraintes<br />
sur les traits d’histoire de vie (Williams 1966b ; Lessells 1991; Stearns 1992). Pour<br />
comprendre ces variations, il est important d’adopter une approche intégrative de<br />
l’organisme et de son environnement. En effet, le milieu de vie est caractérisé par<br />
d’importantes fluctuations ou limitations à la fois dans des facteurs biotiques<br />
(nourriture) et abiotiques (paramètres physiques: température, hygrométrie). Dans<br />
cette situation, les organismes ne vont disposer que d’une quantité limitante d’énergie<br />
qui devra être allouée dans des fonctions très différentes comme la croissance, la<br />
reproduction et la maintenance. La ressource investie dans la reproduction va ainsi<br />
entrer en conflit avec les autres fonctions. En effet, selon le principe d’allocation de<br />
Williams (1966a,b) et Levins (1968), tout investissement supplémentaire dans un<br />
aspect quelconque de la vie d’un organisme ne pourra se faire qu’au dépend d’un<br />
autre aspect. Le fait que les ressources soient généralement limitantes et que les<br />
organismes doivent investir l’énergie dans des voies concurrentielles est à la base de<br />
la notion de compromis ou “trade-off” entre les traits d’histoire de vie. Ainsi, la valeur<br />
reproductrice <strong>total</strong>e d'un organisme peut être considérée comme la somme entre le<br />
14
succès d’une reproduction donnée et celui attendu dans les reproductions futures.<br />
Selon la théorie de l’effort reproducteur (Williams 1966a,b), toute augmentation de<br />
l'investissement dans la reproduction courante se fera au détriment de la valeur<br />
reproductrice résiduelle. Dans un tel contexte, la maximisation du succès reproducteur<br />
à vie passera par une optimisation de l'investissement dans chaque épisode<br />
reproducteur. L’effort reproducteur optimal sera donc déterminé par un équilibre entre<br />
les bénéfices attendus d’une reproduction et les coûts pour les reproductions futures.<br />
L’existence d’un conflit entre le succès d’une reproduction donnée et la valeur<br />
reproductive résiduelle d’un organisme est une proposition clé, à la base du concept<br />
de “coût de la reproduction” formulé par Williams (1966b). Cette notion théorique<br />
simple et très attractive s’est très vite révélée difficile à tester dans la pratique et il en<br />
a résulté une intense controverse sur l’existence même des coûts et sur les<br />
méthodes adaptées pour les mesurer (Reznick 1985). En dépit de ces difficultés,<br />
cette notion occupe désormais une position clé dans la théorie des traits d'histoire de<br />
vie : depuis une quinzaine d'années, une littérature abondante est venue confirmer<br />
l'existence des coûts de la reproduction (Bell 1980 ; Bells et Koufopanou 1986 ;<br />
Stearns 1992 ; Lessells 1991) et leurs très fortes implications évolutives (Clutton-<br />
Brock 1998). Les coûts de la reproduction sont désormais considérés comme des<br />
contraintes majeures qui vont déterminer la réalisation de compromis adaptatifs entre<br />
les traits d’histoire de vie et favoriser l'émergence de stratégies reproductrices<br />
optimales (Williams 1966a,b ; Reznick 1992 ; Niewiarowski & Dunham 1994 ; Shine<br />
& Schwarzkopf 1992 ; Clutton Brock 1998). Le point majeur des réflexions actuelles<br />
ne repose donc plus sur l’existence des coûts de la reproduction mais plutôt sur<br />
l’importance et la valeur de ces coûts, leur origine proximale et les mécanismes<br />
physiologiques impliqués. On réalise ainsi une distinction fondamentale entre deux<br />
grands types de coûts (Calow 1979) : on parle de coûts en survie lorsque la<br />
15
eproduction affecte les probabilités de survie de l'organisme et d'autre part de coût<br />
en fécondité si l'événement reproducteur influence les capacités reproductrices<br />
futures de l'individu. Ces derniers peuvent être directs, via l’épuisement des réserves<br />
mais ils peuvent aussi s’exprimer de façon plus détournée en affectant par exemple<br />
le taux de croissance (Williams 1966b ; Calow 1979 ; Shine 1980). Cette<br />
classification est en fait très artificielle car les deux composantes des coûts sont<br />
rarement indépendantes. Ainsi, une dépense énergétique importante peut, par<br />
exemple, affecter la survie d'un organisme si l'augmentation des prospections<br />
alimentaires (nécessaires à la reconstitution des stocks de réserves corporelles)<br />
l’expose d’avantage aux prédateurs (Bauwens & Thoen 1981 ; Brodie 1989). Les<br />
coûts de la reproduction sont donc de natures multiples et un organisme réalisant un<br />
effort reproducteur donné pourra être affecté de façon complexe par des coûts<br />
d’origines variées<br />
C. Dimension physiologique des compromis et<br />
des systèmes de gestion de la ressource<br />
Ces différents éléments indiquent que la façon de répartir la ressource disponibles a<br />
d’importantes conséquences sur la valeur sélective (ou fitness) d’un organisme. Il est<br />
important de bien comprendre que la réalisation de “compromis d’allocation” va<br />
exister à l’échelle individuelle (Höglund & Sheldon 1998) et notamment au niveau de<br />
son fonctionnement physiologique (Sinervo & Licht 1991). En effet, au sein d’une<br />
espèce à reproduction sexuée, en raison d’une importante variabilité génétique, les<br />
individus vont différer par de nombreux points de leur biologie et notamment au<br />
niveau des “réglages physiologiques” fins qui soutendent l’allocation de l’énergie<br />
dans des voies concurrentielles (comme reproduction donnée et future). L’existence<br />
16
de compromis entre les traits d’histoire de vie souligne la présence de mécanismes<br />
endocrines contraignant la covariation entre ces traits (Sinervo & Litch 1991 ;<br />
Sinervo & DeNardo 1996). Les contraintes entre traits d’histoire de vie sont<br />
directement liées à des actions hormonales dépendantes de nombreux gènes de<br />
régulation ayant des effets multiples, opposés et pléïotropiques sur l’expression de<br />
un ou plusieurs des traits (Rose & Bradley 1998). L’investissement reproducteur va<br />
donc impliquer le rôle clé d’une intégration physiologique, s’exerçant au niveau de<br />
l’organisme. Les systèmes neuroendocrines vont offrir une certaine plasticité de<br />
fonctionnement. A l’échelle de l’organisme, ils vont permettrent l’intégration<br />
d’informations complexes (environnementales et/ou endogènes) et vont régir en<br />
conséquence les aspects majeurs de la reproduction comme l’engagement/<br />
désengagement, le degré d’investissement parental. Sinervo & Svensson (1998)<br />
soulignent que dans de nombreux cas, la plasticité observée dans les traits de vie<br />
est véhiculée par les mêmes mécanismes endocrines (gonadotropines, stéroïdes<br />
sexuels, glucocorticoïdes) qui soutendent les compromis adaptatif existant entre ces<br />
traits. A l’échelle de l’espèce ou de la population, l’évolution ultime de compromis<br />
adaptatifs impliquera l’action de la sélection sur ces facteurs de régulation. Ainsi, il<br />
existe des interactions évidentes entre les causes proximales et les causes ultimes<br />
des compromis (Sinervo & Svensson 1998). Les pressions de sélection vont donc<br />
agir sur les mécanismes proximaux qui seront alors “modelés” au cours des temps<br />
évolutifs. Cette approche “physiologique” et mécanistique des compromis est très<br />
pertinente et constitue un complément nécessaire à l’approche génétique et<br />
populationnelle classique (Rose & Bradley 1998).<br />
De façon conjointe, les pressions de sélection vont également agir sur le<br />
niveau global d’énergie disponible pour la reproduction et ce, en façonnant les<br />
systèmes d’acquisition et de gestion de l’énergie. En effet, pour compenser les<br />
17
esoins énergétiques particulièrement élevés de la reproduction, la plupart des<br />
organismes vont tenter d’acquérir de plus grande quantité de ressource (Jönsson<br />
1997). Deux tactiques de compensation sont généralement identifiées : les<br />
systèmes de reproduction basés sur l’alimentation courante (“income breeders”) et<br />
d’autres sur la réalisation d’un capital de réserves (“capital breeders”) (Drent & Daan<br />
1980, Jönsson 1997). Dans le premier cas, la reproduction est accompagnée d’une<br />
augmentation de l’acquisition des ressources alimentaires qui vont être directement<br />
allouées dans la reproduction. Dans le second, les ressources sont stockées sous<br />
forme de réserves qui serviront ultérieurement de support énergétique principal pour<br />
la reproduction. La formulation de Drent & Daan est en fait quelque peu artificielle car<br />
il existe un continuum entre ces stratégies (Doughty & Shine 1997). Cependant, la<br />
réalisation d’une telle dichotomie est très utile (Price et al. 1988, Martin 1995) et peut<br />
être appliquée à de nombreux organismes. En outre, elle permet de souligner un<br />
aspect important des stratégies reproductrices basée sur la dimension temporelle de<br />
l’acquisition et de l’allocation de l’énergie (Fisher 1930). En effet, les reproductions<br />
sur revenus ou sur réserves vont impliquer des voies biochimiques identiques (dans<br />
les deux cas il y a mise en réserve) mais vont différer dans la durée du stockage des<br />
molécules avant utilisation (Bonnet et al. 1998). Une telle variation repose<br />
directement sur la sélection de soubassement physiologiques qui vont orienter les<br />
compromis d’allocations entre reproduction, stockage de réserves et besoins de<br />
maintenance (Dought & Shine 1998). L’identification des stratégies d’utilisation de la<br />
ressource est évolutivement très importante, notamment parce que les avantages et<br />
inconvénients de la reproduction sur revenus ou sur réserves vont varier selon les<br />
taxons (Jönnson 1997 ; Bonnet et al. 1998). En outre, elle est indispensable lorsque<br />
l’on cherche à évaluer les coûts de reproduction, qui vont alors se manifester de<br />
18
façons différentes selon que la reproduction impose ou non une longue période de<br />
constitution de réserves.<br />
D. Connexion entre effort reproducteur et coût<br />
de la reproduction<br />
La compréhension de l’évolution des traits d’histoires de vie, tel l’investissement<br />
reproducteur (facteur ultime), implique donc l’intégration des facteurs proximaux<br />
sous-jacents (physiologie de l’organisme). Dans ce cadre, il convient d’examiner<br />
finement la manière dont les modifications éco-physiologiques de la reproduction<br />
vont pouvoir se manifester en terme de coûts démographiques. Les coûts de la<br />
reproduction peuvent avoir des origines multiples (prédation, dépenses énergétiques,<br />
effets pathologiques de taux hormonaux élevés, les effets de sénescence...). Dans<br />
les populations naturelles, les coûts peuvent être d’origine écologique, physiologique<br />
ou de façon plus rationnelle une combinaison de ces deux composantes. Pour<br />
étudier l’influence des différentes composantes des coûts de la reproduction sur la<br />
partition de l’énergie et de l’effort reproducteur, il est donc fondamental d’identifier la<br />
nature des coûts impliqués et les mécanismes par lesquels ils se manifestent<br />
(Sinervo & Svensson 1998). Dans un tel contexte, il devient crucial de bien<br />
distinguer les deux notions, a priori très proches, d’effort reproducteur et de coût de<br />
la reproduction (Tuomi et al. 1983 ; Niewiarowski & Dunham 1994). L'effort de<br />
reproduction représente l’allocation “brute” réalisée par l'organisme en temps,<br />
énergie ou matière dans la transmission de ses gènes (Clutton Brock 1991; 1998).<br />
Par extension, cet effort a souvent été assimilé à des coûts car l'animal va "allouer"<br />
de l'énergie et du temps dans la production de jeunes. Cependant, pour bien<br />
percevoir la nuance entre ces deux notions, il est nécessaire d’avoir une perspective<br />
19
à long terme intégrant la vie reproductrice de l’organisme. En effet, à la différence de<br />
l’effort reproducteur qui est simultané à la reproduction, les coûts peuvent s’exprimer<br />
avec un certain décalage temporel, via les effets démographiques sur la survie ou<br />
sur les capacités reproductrices ultérieures (Williams 1966b ; Clutton-brock 1998).<br />
Ces composantes (coût et effort) sont donc liées mais leur degré de connexion peut<br />
être variable. Ainsi, on peut envisager des situations où coût et effort sont<br />
directement connectés, c’est-à-dire où l’augmentation de la fécondité engendrera<br />
une augmentation proportionnelle de l’effort et des coûts. Ce type de situation est<br />
particulièrement fréquent dans des systèmes avec soins parentaux où le nombre de<br />
jeune va directement affecter le degré d’investissement parental (Clutton-brock<br />
1991). Dans d'autres, cas la relation peut être de type “tout ou rien”, notamment<br />
quand la reproduction impose des activités qui sont d’emblée coûteuses (comme des<br />
migrations) et ce indépendamment de l’effort reproducteur courant et du nombre de<br />
jeunes produits. La nature de la relation entre coût et effort n'est donc pas forcément<br />
linéaire ou constante (Niewiarowski & Dunham 1994). En outre, elle dépend<br />
largement de facteurs environnementaux et un effort reproducteur important ne se<br />
traduit pas nécessairement par des coûts élevés si le contexte est favorable<br />
(Congdon et al. 1982 ; Tuomi et al. 1983).<br />
La réalisation de cette distinction est importante car la relation existante entre<br />
investissement reproducteur et les coûts associés peut profondément déterminer<br />
l’évolution d’une stratégie d’acquisition et d’allocation optimale de la ressource (Bull<br />
& Shine 1979 ; Shine & Schwarzkopf 1992; Niewiarowski & Dunham 1994,<br />
1998 ; Sinervo & Svensson 1998). L’intégration des différentes composantes des<br />
coûts (dépendantes ou indépendantes de la fécondité) peut être un facteur clé dans<br />
la compréhension de l’évolution de la fréquence de reproduction dans les systèmes<br />
itéropares. En effet, si la majorité des organismes se reproduisent de façon<br />
20
annuelle, il existe néanmoins des espèces où les différents individus de la population<br />
vont “sauter” des opportunités de reproduction et se reproduire de façon asynchrone<br />
tous les deux ans ou plus (Bull & Shine 1979). Ce type de système à faible<br />
fréquence de reproduction (LFR = Low Frequency of Reproduction) est souvent<br />
observé chez des espèces dont le cycle impose des activités particulières,<br />
caractérisées par des dépenses en énergie, temps et/ou survie qui s’expriment de<br />
façon fixe c’est à dire indépendamment de la fécondité. Parmi ces activités, Bull &<br />
Shine (1979) citent les migrations de reproduction, les soins/défense des oeufs (à la<br />
ponte dans son ensemble), la rétention des oeufs dans les voies génitales<br />
(viviparité). Les migrations de reproduction imposent par leur nature, des contraintes<br />
indépendantes de la fécondité et du succès reproducteur de l’organisme. Une telle<br />
déconnexion existe aussi dans le cas de soins prodigués à la progéniture dans son<br />
ensemble (défense des pontes, gestation) au cours desquelles les femelles cessent<br />
souvent de manger et réduisent leur déplacements (Bull & Shine 1979). Si ces<br />
composantes des coûts sont très élevées, l’organisme aura avantage à se reproduire<br />
de façon alternée, éviter ainsi ces coûts ”fixes” et profiter d’une année de repos pour<br />
investir plus d’énergie et d’assurer un meilleur succès dans l’évènement reproducteur<br />
suivant. Ce système est particulièrement avantageux dans un habitat pauvre en<br />
ressource, où la dépense énergétique dans les activités de reproduction est très<br />
élevée par rapport au niveau de ressource disponible. L’évolution vers un<br />
système “LFR” permettrait alors une augmentation de la fécondité moyenne. Si<br />
l’hypothèse de Bull et Shine (1979) n’a pas été testée empiriquement, elle suggère<br />
néanmoins que l’identification des différentes composantes des coûts de la<br />
reproduction et de la relation entre ces composantes et l’effort reproducteur est très<br />
importante à réaliser. Une telle approche permet de déterminer la signification<br />
adaptative de la répartition de l’effort reproducteur dans la vie de l’organisme et doit<br />
21
donc être intégrée dans l’analyse de l’évolution des stratégies de reproduction<br />
extrêmes (seméliparité).<br />
E. Intérêt d’une perspective ectothermique<br />
Le test d’hypothèses évolutives sur les stratégies reproductrices et notamment la<br />
fréquence des reproductions va impliquer le recours à une démarche à la fois<br />
méticuleuse et intégrée. En effet, considérant la complexité des systèmes vivants et la<br />
diversité des pressions de sélection auxquelles sont soumis les organismes, l’examen<br />
de l’influence évolutive des coûts de reproduction sur l’effort reproducteur sera<br />
optimisé si l’on sélectionne des situations biologiques simples où le nombre de<br />
variables confondantes sera limité. Idéalement, pour élucider la nature de la relation<br />
entre effort de reproduction et coût, il va être nécessaire de caractériser clairement les<br />
sources d’investissements (Fisher 1930) et d’identifier leur chronologie exacte (Bonnet<br />
et al. 2000a). Malheureusement, dans la plupart des systèmes biologiques, la<br />
reproduction est un phénomène très complexe et multifactoriel. Chez les<br />
endothermes par exemple (oiseaux et mammifères), les grandes étapes d’allocation<br />
directe de l’énergie dans la progéniture (formation des oeufs, incubation, gestation)<br />
sont souvent réalisées en synchronie avec des activités complexes et moins<br />
étroitement liées à l’investissement gamétique comme par exemple des interactions<br />
sociales avec les congénères (hiérarchie de dominance, défense du territoire) ou des<br />
comportements sexuels très élaborés (construction d’un nid). Il va donc résulter une<br />
“superposition” entre ces sources de dépenses énergétiques et les aspects plus<br />
proximaux de l’investissement reproducteur. Une telle complication va rendre difficile,<br />
voir impossible, l’identification des contributions respectives des différentes activités<br />
reproductrices en terme de demande énergétique et de coûts associés. En outre, les<br />
endothermes manifestent généralement des soins parentaux très élaborés pouvant<br />
22
impliquer une collaboration complexe entre les partenaires (Lequette & Weimerskirch<br />
1990 ; Clutton-Brock 1991). La mesure des coûts de la reproduction va faire intervenir<br />
l’évaluation de ces soins et de leurs variations avec la fécondité. Ceci va<br />
sérieusement compliquer l’étude des déterminants du succès reproducteur individuel.<br />
Enfin, les endothermes sont caractérisés par de très forts besoins métaboliques et<br />
l’organisme est contraint de subvenir, en parallèle à toutes ces activités, à des<br />
besoins de maintenance très élevés (Else & Hulbert 1981; Hulbert & Else 1989). Les<br />
vertébrés ectothermes diffèrent fondamentalement des endothermes par de faible<br />
besoins énergétique pour la maintenance et des niveaux métaboliques qui ne<br />
dépassent pas 20% de ceux d’un mammifère ou d’un oiseau de taille comparable<br />
(Pough 1980). La régulation de la température corporelle est principalement réalisée<br />
de façon comportementale et n’implique pas de dépenses accrues dans la production<br />
de chaleur endogène. En conséquence, l’énergie assimilée va pouvoir être facilement<br />
convertie en biomasse (Bradshaw 1997). Cette grande efficience de conversion va<br />
faciliter d’autant l’étude du budget énergétique (plus grande lisibilité des prises<br />
énergétiques). De plus, il existe une grande diversité dans les soins parentaux chez<br />
ces organismes avec généralement des soins post-nataux très réduits, voir inexistants<br />
(Clutton-Brock 1991). Un tel contexte offre de nombreuses simplifications avec<br />
notamment une concentration de l’effort reproducteur avant la mise-bas ou la ponte<br />
Ce cadre s’avère donc très favorable pour l’identification précise des activités<br />
reproductrices et de leurs implications énergétiques et écologiques.<br />
Dans ce travail nous nous intéressons aux stratégies reproductrices<br />
“extrêmes” (effort reproducteur élevé associé à un petit nombre de reproduction de<br />
l’organisme). Notre objectif principal est d’identifier des éléments de réponse<br />
originaux sur les facteurs favorisant l’évolution vers des systèmes semélipares. Dans<br />
ce cadre théorique, il faut noter que les possibilités d’investigations sont très<br />
23
variables selon que l’on s’adresse à des modèles endothermes (oiseaux et<br />
mammifères) ou ectothermes. En effet, la grande majorité des systèmes semélipares<br />
au sens strict (mort après un seul épisode reproducteur chez un ou les deux sexes) a<br />
été observée chez des organismes ectothermes avec de nombreux exemples chez<br />
des invertébrés (Boyle 1983, 1987 ; Andries 2001) et certains vertébrés (Crespi &<br />
Teo 2002) . Les rares systèmes semélipares décrits chez les endothermes sont<br />
limités à quelques espèces de marsupiaux avec une restriction de la seméliparité aux<br />
mâles (Oakwood et al. 2001). Les vertébrés ectothermes sont des modèles<br />
particulièrement intéressants qui offrent une grande variabilité dans le “degré”<br />
d’itéroparité avec l’existence d’un continuum entre seméliparité et l’itéroparité<br />
(Schaffer & Elson 1975 ; Crespi & Teo 2002). Il existe ainsi des espèces de lézards<br />
avec des populations itéropares et d’autres semélipares (Bradshaw 1997) : un cas de<br />
figure exceptionnel pour une analyse comparative des forces sélectives et des<br />
mécanismes physiologiques impliqués dans cette transition. Le degré d’itéroparité<br />
peut aussi varier au sein d’une même population, notamment dans des systèmes où<br />
la fréquence de reproduction est faible ( “LFR”) et varie selon les individus. A de très<br />
rares occasions, on pourra ainsi observer la coexistence intra-populationnelle,<br />
d’individus à trajectoire itéropares et d’autres semélipares. C’est le cas notamment<br />
de la vipère aspic (Vipera aspis) au Nord de son aire de distribution. Une telle<br />
situation fournit des opportunités uniques pour tester l’hypothèse de Bull et Shine<br />
(1979) sur la nature des coûts de la reproduction et l’évolution de systèmes a faible<br />
fréquence des épisodes reproducteurs. Du fait des simplifications énergétiques de<br />
l’ectothermie, il devrait être possible de décrire finement le système d’allocation de<br />
l’énergie et le mettre en relation avec ces aspects démographiques. Dans cette<br />
perspective, nous avons examiné la stratégie reproductrice déployée par la vipère<br />
24
aspic femelle dans l’0uest de la France et ce travail se structure en cinq grandes<br />
parties:<br />
1. Présentation de l’espèce, écophysiologie et cycle reproducteur<br />
2. Le système d’allocation de l’énergie dans la reproduction : description et<br />
avantages dans un environnement contraignant<br />
3. Caractérisation des coûts associés à la reproduction, amplitude et relation avec la<br />
fécondité<br />
4. Examen des facteurs déterminant la seméliparité femelle et les conséquences<br />
démographiques<br />
5. Discussion-conclusion: reproduction sur réserves, coûts de la reproduction et<br />
évolution vers la seméliparité<br />
25
I. Présentation de l’espèce<br />
et des méthodes d’études<br />
Vipère aspic mon amie, qu’as tu fais de ta vie ?<br />
Noël Guillon, Poète-philosophe de Haute Saintonge.<br />
26
A. Résumé du chapitre:<br />
Si la biologie de la reproduction des serpents est encore mal connue, la famille des<br />
viperidés est probablement la plus intensément étudiée. Ce groupe très homogène est<br />
caractérisé par une morphologie trapue, un appareil venimeux très perfectionné et un<br />
mode de reproduction généralement vivipare. Le mode de vie est basé sur la chasse<br />
à l’affût et il existe une prédispostion au stockage de réserves corporelles pour la<br />
reproduction (“capital-breeding”). La vipère aspic (Vipera aspis) est un petit serpent<br />
européen qui présente, dans l’ouest de la France, une stratégie de reproduction très<br />
particulière. Dans cette région, la maturité des femelles est tardive, l’espérance de vie<br />
courte et la fréquence des reproductions faible (tous les 2-4 ans). Les réserves<br />
semblent jouer un rôle très important et il existe un seuil minimum de condition<br />
corporelle pour permettre l’engagement reproducteur. La reproduction est<br />
caractérisée par de fortes contraintes énergetiques et écologiques pendant la<br />
vitellogénèse et la gestation. L’investissement maternel est donc généralement très<br />
élevé et la masse des portées produites dépasse souvent celle de la mère après la<br />
mise bas. Le nombre des reproductions est réduit et la vie reproductrice des femelles<br />
présente une véritable tendance semélipare. Cette situation contraste beaucoup avec<br />
celle des mâles chez qui l’investissement reproducteur est graduel et repété au cours<br />
de la vie (itéroparité). En occupant une position intermédaire entre itéroparité et<br />
seméliparité, la stratégie des vipères aspic femelles offre une excellente opportunité<br />
d’examiner les facteurs favorisant un investissement reproducteur élevé et la<br />
transition vers les systèmes semélipares.<br />
27
B. Position systématique et phylogénie<br />
L’origine évolutive des serpents est discutée depuis plus de 130 ans et leur position<br />
au sein de l’orde des squamates (qui regroupe lézard et serpents) est encore l’objet<br />
de débats (Coates & Ruta 2000). Il existe actuellement plus de 2700 espèces de<br />
serpents qui se répartissent en trois groupes trés inégaux: les scolécophidiens, les<br />
anilioïdes et les macrostomates. Au sein des macrostomates sont regroupées les<br />
formes de serpents les plus “dérivées”, caracterisées par une très grande mobilité<br />
des mâchoires supérieures et inférieures permettant l’ingestion de proies de grande<br />
tailles. Ce groupe phylétique comprend la plus grande diversité des espèces<br />
actuelles avec notamment des taxons ayant développés des appareils venimeux<br />
élaborés comme la famille des viperidés, à laquelle appartient la vipère aspic (Vipera<br />
aspis, Linné 1758). Cette famille s’individualise par de nombreux critères<br />
anatomiques, ostéologiques, histologiques et son monophyllétisme n’a jamais été<br />
remis en question par les données moléculaires (Ineich 1995). Cette famille est<br />
caracterisée par une morphologie assez trapue, un mode de chasse basé sur l’affût<br />
et une prédisposition au stockage de réserves lipidiques pour la reproduction<br />
(“capital breeding”, Madsen et Shine 1992a, Brown 1993; Martin 1993). La répartiton<br />
actuelle des vipéridés suggère une origine asiatique avec une différentiation pendant<br />
l’ère Tertiaire à la fois dans la partie orientale (crotalinae) et occidentale (viperinae)<br />
du continent Eurasien (Ineich 1995). En europe occidentale, la vipère aspic (Vipera<br />
aspis), la vipère ammodytes (Vipera ammodytes, Linné 1758) et la vipère de lataste<br />
(Vipera latastei, Boscà 1879) constituent un groupe d’espèces très proches qui se<br />
sont probablement différenciées durant le Pleistocène dans chacune des grandes<br />
péninsules méditerranéennes (Saint Girons 1997). Parmis ces espèces, la vipère<br />
aspic est celle qui occupe la partie la plus nordique de l’aire ouest-paléarctique.<br />
28
C. Répartition de l’espèce et des populations<br />
d’études<br />
La vipère aspic est un petit serpent venimeux (en moyenne 55 cm de longueur<br />
<strong>total</strong>e) avec une morphologie trapue et une tête relativement bien distincte du corps.<br />
Cette espèce localement abondante, est caracterisée par un haut polymorphisme qui<br />
s’exprime dès le niveau intra-populationnel. Cinq sous-espèces sont classiquement<br />
reconnues : Vipera aspis aspis (nord et centre de la France), Vipera aspis<br />
francisiredi, (Nord et centre de l’Italie), Vipera aspis atra (Centre ouest de la Suisse,<br />
Nord-ouest de l’Italie), Vipera aspis hugyi (Sud de l’Italie et Sicile), Vipera aspis<br />
zinnekeri (Pyrénées Françaises et Espagnoles). Le statut de ces sous-espèces reste<br />
cependant incertain et fait l’objet de discussions actives (Zuffi 2002). Ce travail porte<br />
sur la sous espèce nominale (Vipera aspis aspis) dans des populations de l’Ouest de<br />
la France, au Nord de l’aire de répartition de l’espèce (carte ci-dessous).<br />
Figue 1. Répartition de la vipère aspic (d’après Naulleau 1997)<br />
29
D. Biologie de la reproduction<br />
1. Généralités chez les squamates<br />
Les squamates sont des vertébrés amniotes qui produisent des oeufs de grandes<br />
taille et chargés en vitellus (type mégalecithe). La reproduction impose de fortes<br />
contraintes à l’organisme femelle qui doit investir d’importantes quantités d’énergie<br />
dans des pontes de grande taille. Ce groupe est caratérisé par une importante<br />
diversité du mode de reproduction avec un véritable continuum entre oviparité et<br />
viviparité (Shine 1985). La rétention des oeufs dans les voies génitales et la<br />
production de jeunes (viviparité) génère des contraintes très spécifiques qui viennent<br />
augmenter la hauteur de l’investissement maternel. Pendant la gestation, il existe<br />
notamment un déplacement des préférences thermiques vers des températures<br />
élevées (Shine 1980). Les femelles vont alors présenter un changement dans leur<br />
schéma d’activité avec de longues périodes d’exposition. En outre, le<br />
développement embryonnaire va souvent affecter la mobilité des femelles qui sont<br />
ainsi plus exposées à la prédation (Shine 1980, Bauwens et Thoen 1981). Les soins<br />
parentaux post-nataux sont la plupart du temps très réduits et les jeunes sont<br />
autonomes dès la naissance. L’investissement paternel dans la reproduction est<br />
beaucoup moins complexe et il se limite souvent à la recherche et la fertilisation de<br />
partenaires sexuels (avec le cas échéant des combats avec des compétiteurs). Le<br />
système d’appariement est en général simple (Duval et al. 1992, 1993) et les<br />
investisssements énergétiques spécifiques des deux sexes ne sont pas masqués<br />
par des interactions comportementales complexes entre partenaires.<br />
L’éco-physiologie de la reproduction chez les squamates est un domaine<br />
encore très jeune et en pleine émergence (Crews & Gans 1992). Si de nombreuses<br />
30
études ont été conduites sur les lézards (physiologie comportementale notamment),<br />
la biologie de la reproduction des serpents est caracterisée par d’énormes lacunes.<br />
Ainsi, en dépit de l’étonnante diversité des modes de reproduction des serpents,<br />
l’état des connaissances actuelles ne permet pas une approche physiologique<br />
comparative des soubassements endocrines. Toute tentative de synthèse serait<br />
d’ailleurs très risquée du fait de la concentration des travaux sur des espèces de<br />
zones tempérées alors que l’immense majorité des serpents occupe des zones<br />
tropicales et équatoriales (Seigel et Ford 1987).<br />
Les vipéridés des zones tempérées sont probablement les serpents les plus<br />
intensément étudiés (Saint Girons 1949, 1952, 1957a,b, 1975 ; Fitch 1960 ; Klauber<br />
1972 ; Brown 1991 ; Madsen & Shine 1993 ; Martin 1993) et il existe de bonnes<br />
informations sur les cycles reproducteurs en général très uniformes et fortement<br />
saisonniers. La vipère aspic a fait l’objet de nombreux travaux et constitue un des<br />
serpents les mieux connus au monde (Saint Girons 1949, 1957a,b; Saint Girons &<br />
Duguy 1992; Bonnet 1996; Naulleau et al. 1999; Bonnet et al. 1999b, 2000a,b,<br />
2001b, 2002b) notamment en ce qui concerne les soubassements endocrines de la<br />
reproduction (Bonnet et al. 1994; Bonnet 1996; Bonnet et al. 2001a).<br />
2. Eco-physiologie de la vipère aspic<br />
La femelle vipère aspic peut être considérée comme un reproducteur sur réserves<br />
“typique”. L’entrée dans la reproduction (recrutement des follicules) s’effectue dès la<br />
sortie d’hivernage au début du printemps et la condition corporelle (masse ajustée<br />
par la taille) calculée à ce moment offre une bonne estimation du niveau des<br />
réserves lipidiques (corps gras). Il existe un seuil minimal de réserve pour permettre<br />
l’engagement reproducteur. Ainsi, seules les femelles ayant accumulé des réserves<br />
31
vitellogéniques supérieures à un niveau minimum vont se “lancer” dans la<br />
folliculogénèse. Il en résultera des reproductions asynchrones avec, chaque année,<br />
la coexistence d’une fraction de femelles reproductrices (40% en moyenne) avec des<br />
femelles non-reproductrices (Bonnet & Naulleau 1994,1995).<br />
L’accouplement a lieu dès la sortie d’hivernage uniquement chez les femelles<br />
avec une condition corporelle suffisante (Bonnet & Naulleau 1994). La mobilisation<br />
du capital de réserves lipidiques va fournir les substrats nécessaires pour la synthèse<br />
de vitellogénine par le foie et la déposition du jaune dans les follicules en croissance<br />
(Bonnet 1996). Il existe notamment une relation positive entre la masse des corps<br />
gras abdominaux et le nombre de follicules en croissance (Saint Girons & Naulleau<br />
1981). La vitellogénèse (synthèse du vitellus) constitue une étape clé de l’effort<br />
reproducteur et s’accompagne de profonds bouleversements physiologiques. Les<br />
travaux empiriques et expérimentaux indiquent le rôle clé de l’oestradiol dans la<br />
mobilisation des réserves et le déclenchement de la croissance folliculaire (Bonnet et<br />
al. 1994, Bonnet 1996). De très nombreux follicules sont en général recrutés et<br />
après cette phase initiale (début du printemps), la mort d’une fraction des follicules<br />
(atrésie, Méndez de la Cruz 1993) va déterminer de façon proximale le nombre<br />
d’oeufs ovulés. Lorsque la nourriture est disponible, les femelles reproductrices<br />
continuent a s’alimenter pendant cette période (Saint Girons & Naulleau 1981).<br />
L’ovulation et la fécondation ont lieu pendant la première quinzaine de juin. Les oeufs<br />
ovulés vont alors s’hydrater de façon importante et le développement embryonnaire<br />
dans les voies génitales va s’étendre ensuite jusqu’au début de l’automne.<br />
Pendant la gestation, la vie des femelles reproductrices va être affectée par de<br />
profonds changements. Elles vont presenter des optimum thermiques élevés et<br />
passer significativement plus de temps que les non reproductrices en<br />
thermorégulation (Bonnet & Naulleau 1996). En combinaison avec la<br />
32
thermorégulation les femelles reproductrices deviennent nettement plus sédentaires<br />
et leur domaine vital passe ainsi de 3000 m 2 (en début de gestation) à 300 m 2 à<br />
partir de la mi-juillet (Naulleau et al. 1996). Enfin on constate durant la gestation, une<br />
réduction, voir un arrêt des prises alimentaires (Saint-Girons 1952, 1979). La<br />
gestation entraîne donc des modifications profondes (écologie, physiologie) et non<br />
graduelles chez les individus qui ont pris la "décision" de s’engager dans la<br />
reproduction (Bull et Shine 1979). Un travail récent (Bonnet et al. 2001a) suggère<br />
l’importance de la progestérone dans le maintien de la gestation chez cette espèce.<br />
La mise bas ne dure que quelques minutes et les femelles produisent de 2 à 22<br />
vipéreaux de 6 g en moyenne. La masse de la portée représente de 30 à 120% de la<br />
masse post-parturiente femelle (Bonnet 1996) ce qui constitue un effort reproducteur<br />
énorme pour un vertébré amniote. Après la mise-bas, les femelles sont très<br />
amaigries et la condition corporelle post-partum constitue un bon indicateur de leur<br />
degré d’émaciation. Une à plusieurs années seront alors nécessaire pour accumuler<br />
un stock de réserves suffisant pour se reproduire à nouveau.<br />
La mise en place d’un suivi populationnel aux Moutiers en Retz (Loire<br />
Atlantique) depuis 1992 a permi d’apporter des informations précises sur le nombre<br />
d’opportunités de reproduction pour les femelles dans cette région. Le temps<br />
nécessaire à la constitution des réserves corporelles induit une maturité tardive (2.5<br />
à 3.5 ans avec une durée de vie de 5 ans en moyenne). Les données de suivi<br />
individuel suggèrent un faible nombre d’opportunités reproductrices avec une<br />
majorité d’individus à trajectoire semélipare (une seule reproduction) et d’une fraction<br />
d’individus itéropares. Ces derniers ne présentent en général que deux à trois<br />
reproductions dans leur vie. Pour une minorité d’individus, quatre reproductions ont<br />
été enregistrées. Dans tous les cas de figure, la fréquence des reproductions est<br />
faible avec des cycles compris entre deux et quatres ans. Nous sommes donc en<br />
33
présence d’un système à très faible fréquence de reproduction (“LFR”, Bull et Shine<br />
1979) avec une tendance semélipare.<br />
Femelles reproductrices<br />
Accoupl<br />
Alimentation<br />
Hibernation Vitellogénèse O Gestation MB<br />
Hibernation<br />
Janvier Février Mars Avril Mai Juin Juillet Août Septembre Octobre Novembre Décembre<br />
Femelles non-reproductrices<br />
Mâles<br />
Hibernation<br />
Alimentation<br />
Hibernation<br />
Janvier Février Mars Avril Mai Juin Juillet Août Septembre Octobre Novembre Décembre<br />
Hibernation<br />
Accoupl<br />
Alimentation<br />
Spermiogénèse continue?<br />
Hibernation<br />
Janvier Février Mars Avril Mai Juin Juillet Août Septembre Octobre Novembre Décembre<br />
Figure 2. Phénologie des cycles reproducteurs mâle et femelle dans l’ouest de la France. Les<br />
grandes étapes ont été représentées par des rectangles dont la position temporelle est variable selon<br />
les années et les individus. Seule l’ovulation semble être un phénomène assez fixe (première<br />
quinzaine de Juin). Chez les mâles, le cycle reproducteur est annuel et la production de<br />
spermatozoïdes semble continue. Dans la zone d’étude le cycle reproducteur des femelles est<br />
toujours supérieur à un an. Chaque année, la population est donc composée d’un fraction de femelles<br />
reproductrices et d’une autre de non reproductrices. (Accoupl: accouplement; O: ovulation, MB: mise<br />
bas).<br />
34
La situation des femelles contraste beaucoup avec celle des mâles. En effet,<br />
chez ces derniers, la spermiogénèse (maturation des spermatozoïdes) semble plus<br />
ou moins continue dans l’année avec deux vagues : la première en Avril-Mai alors<br />
que les accouplements ont déja commencé, et la seconde en Août -Octobre<br />
(Naulleau 1997). Les spermatozoïdes produits pendant l’automne sont stockés dans<br />
les canaux déférents jusqu’aux accouplements de printemps. Le système<br />
d’appariement est basé sur la polygynie simultanée et il n’existe pour ce sexe aucune<br />
forme de soin parental. La période sexuelle se limite à un ou deux mois d’activités de<br />
recherche et de fertilisation de partenaires avec parfois des combats ritualisés entre<br />
rivaux (Bonnet et al. 2002b). Pendant cette phase, les mâles deviennent<br />
anorexiques et les réserves lipidiques vont fournir le support énergétique nécessaire.<br />
Les mâles sont donc également des reproducteurs sur réserves , cependant,<br />
l’engagement dans la reproduction ne va pas être contrôlé par un seuil élevé de<br />
condition corporelle. Chez ce sexe, le système l’allocation de l’énergie est beaucoup<br />
plus graduel (Aubret et al. 2002, Bonnet et al. 2002b). Ainsi, un individu avec peu de<br />
réserves pourra s’accoupler efficacement avec une partenaire. En outre, si le niveau<br />
d’amaigrissement pendant la phase sexuelle devient trop élévé, un mâle pourra<br />
aisément se désengager de la reproduction et entreprendre des prospection<br />
alimentaires (Aubret et al. 2002). Les coûts de la gamétogénèse étant très réduits,<br />
les mâles sont donc beaucoup moins contraints que les femelles en terme<br />
d’investissement énergetique et temporel dans la reproduction. Le cycle<br />
reproducteur est généralement annuel (Vacher-Vallas 1997).<br />
35
Effort:<br />
Résumé des grandes étapes de l’effort<br />
reproducteur et des contraintes associées:<br />
I. La folliculogénèse (3 mois):<br />
Figure 3. folliculogénèse observée par résonance magnétique nucléaire (RMN)<br />
Mobilisation des réserves lipidiques, protéiques pour la synthèse de vitellus<br />
Recrutement et croissance folliculaire: allocation de temps, d’énergie et de<br />
Contraintes:<br />
matière dans la formation des oeufs<br />
Besoins thermiques élevés, thermophilie,<br />
Exposition aux prédateurs.<br />
36
Effort:<br />
II. La gestation (2 à 3 mois):<br />
Figure 4. Vipère aspic gestante en thermorégulation<br />
Déplacement vers des optimums thermiques élevés pour l’embryogénèse<br />
Epuisement des réserves et catabolisme protéique<br />
Contraintes:<br />
Figure 5. Comparaison de la partie postérieure d’une vipère aspic avant<br />
(en haut) et après (en bas) la mise-bas<br />
<strong>The</strong>rmophilie accentuée, exposition très fréquente.<br />
Diminution des déplacements et des prises alimentaires.<br />
Etat physiologique critique après la parturition<br />
37
E. Méthodes d’étude<br />
Le présent travail est basé sur la combinaison de deux approches complémentaires :<br />
le suivi longitunidal d’une population et la mise en place d’expérimentation en<br />
captivité.<br />
1. Approche longitudinale:<br />
Un suivi par capture-marquage-recapture d’une population de vipère aspic a été<br />
lancé par Guy Naulleau en 1992 aux Moutiers en Retz (47 o 03N'; 02 o 00W'). La<br />
zone d’étude de 33 hectares est une mosaîques de prairies, de plantations et de<br />
friches en régénération. Le climat est de type océanique-tempéré. Chaque année, la<br />
zone est régulièrment patrouillée par une à quatre personnes, essentiellement au<br />
printemps et en fin d’été (capture de femelles reproductrices). Ce suivi constitue la<br />
plus grande base de données existante sur la vipère aspic avec 1032 individus<br />
marqués et plus de 10000 données de captures-recaptures. L’effort de recherche<br />
dépasse 700 jours et 550 heures de terrain. Mon implication active dans cette<br />
étude (terrain et analyse de données) a débuté en 1996.<br />
Les animaux sont capturés à vue, sexés par éversion des hémipénis. Tout<br />
individu capturé est pesé au gramme près, mesuré au demi cm près (en longueur<br />
<strong>total</strong>e: “LT” et distance museau-cloaque: “SVL”) et marqué individuellement. Le<br />
marquage des adultes s’effectue à l’aide de puces electroniques (PIT-TAG : passive<br />
integrated transponder, TX1400L, Rhône Mérieux, 69002 LYON France, product of<br />
Destron/IDI Inc). Les jeunes individus sont marqués par ablation codifiée d’écailles<br />
ventrales. Chaque serpent est ensuite relaché au point exact de capture. La<br />
population d’étude est relativement bien isolée des populations avoisinantes<br />
(Vacher-Vallas 1997, Bonnet et al. 2000a). Cette espèce est très philopatrique<br />
38
(Naulleau et al. 1996) et tout individu non capturé pendant une longue période (> 2<br />
ans) est considéré comme mort (Bonnet et al. 2002a)<br />
Le statut reproducteur des femelles est déterminé selon différentes méthodes<br />
qui dépendent de la période de capture. Au début du printemps, les femelles<br />
présentant une condition corporelle supérieure au seuil sont considérées comme<br />
reproductrices (voir Bonnet & Naulleau 1994 pour des détails sur la méthode). Plus<br />
tard dans l’année, la palpation de l’abdomen permet de détecter et compter des<br />
follicules en croissance (vitellogénèse) ou des embryons (gestation) (Fitch 1987,<br />
Naulleau & Bonnet 1996). A la fin de l’été, les femelles reproductrices sont capturées<br />
et transportées au laboratoire. Elles sont alors maintenues en captivité dans des<br />
boîtes individuelles et pesées tous les deux jours jusqu’à la parturition. Cette<br />
méthode nous a permis de récolter des données détaillées sur plus de 190 mises<br />
bas. Les différents éléments de la portée sont caractérisés (nouveau-né, mort-né,<br />
embryon, oeuf non-développé) et pesés au dixième de gramme près. Les nouveaux-<br />
nés et les morts-nés sont mesurés au demi cm près (en longueur <strong>total</strong>: ‘LT” et<br />
distance museau cloaque:” SVL”). La masse tolale de la portée (“litter mass”) est<br />
calculée en sommant l’ensemble des différents élements produits. Pour calculer la<br />
masse de la portée viable (“fit littermass”) seuls les nouveaux-nés ont été considérés<br />
(Gregory et al. 1992). Une fois la mise-bas achevée, la femelle est palpée pour<br />
déterminer la présence éventuelle d’éléments non expulsés.<br />
De 1993 à 1996 certains individus ont été équipés avec des émetteurs placés<br />
par ingestion forcée dans l’estomac et suivis par télémétrie. Les animaux équipés<br />
(principalement des femelles reproductrices et quelques non-reproductrices) ont été<br />
suivies sur des périodes variables (inférieure à deux mois) jusqu’à régurgitation de<br />
l’émetteur ou capture en vue d’obtenir la mise bas. Pendant toute la période de suivi<br />
télémétrique, 1 à 4 relevés journaliers de positions et de températures ont été<br />
39
éalisés. Les données sur les déplacements n’ont pas été exploitées dans le cadre<br />
de ce travail. En revanche, les données de température nous ont permis d’étudier<br />
l’influence du statut reproducteur sur les préferences thermiques.<br />
Facteurs environnementaux examinés<br />
Pendant la période d’étude, nous avons obtenu des données précises sur des<br />
variables environnementales particulièrement importantes: l’abondance des proies et<br />
les fluctuations climatiques.<br />
Abondance des proies : la vipère aspic se nourrit essentiellement de campagnols<br />
(Microtus arvalis Pallas) dont les populations fluctuent avec un cycle typique de trois<br />
à quatre ans (Delattre et al. 1992). Ces variations d’abondance affectent directement<br />
la consommation des serpents et donc la proportion d’animaux capturés avec des<br />
indices de repas (palpation de proies, crottes avec poils). La proportion annuelle de<br />
serpents capturés avec des restes de repas est disponible pendant toute la durée<br />
d’étude à l’exception de 1992 et nous avons utilisé ce paramètre comme un indice<br />
d’abondance des proies. Cet estimateur donne des resultats très consistants avec<br />
les données de piégages de rongeur dans cette région (Bonnet et al. 2001b;<br />
Salamolard et al. 2000). De plus amples précisions et une discussion de cette<br />
méthode sont disponible dans la section matériel et méthode de l’article 3.<br />
Températures : la zone d’étude est située au nord de l’aire de répartition de l’espèce<br />
et les conditions climatiques sont contraignantes en comparaison avec les<br />
populations méridionnales. Ainsi, en Italie, les mises bas sont observées à la mi-<br />
Juillet, soit deux mois plus tôt que dans l’ouest de la France (Zuffi et al., 1999).<br />
Pendant toute la période d’étude, les maximum thermiques journaliers sous abris<br />
(°C) ont été mesurés (données Météo France, station d’enregistrement de Pornic).<br />
40
Deux périodes biologiques ont été identifiées: la saison active (de Mars à Octobre) et<br />
la gestation (de mi-Juin à fin Août).<br />
2. Approche expérimentale en captivité<br />
En parallèle avec les analyses des données des Moutiers, nous avons realisé<br />
plusieurs expérimentations dans les enclos d’élevages de la station biologique de<br />
Chizé (Forêt de Chizé, Deux-Sèvres, 46°07’ N, 00°25’ W). Dans tous les cas, les<br />
vipères ont été capturées dans des populations peu éloignées de la zone d’étude<br />
des Moutiers et principalement en Vendée. Après la capture, certaines femelles sont<br />
ramenées au laboratoire, pesées, mesurées et marquées individuellement par<br />
ablation d’écailles ventrales.<br />
Conditions de captivité<br />
Les femelles sont ensuite placées dans des enclos extérieurs (5 x 3m) recréant le<br />
milieu naturel de la vipère et exposés aux conditions climatiques de la station<br />
biologique de Chizé. Chaque enclos est équipé avec le même nombre de plaques<br />
en Fibrociment (50x50cm) qui servent de refuges. L’eau est fournie ad-libitum et la<br />
végétation (Poacées) est maintenue haute pour fournir abris et fraicheur. Dans le cas<br />
des captures au sortir de l’hivernage (i.e. au début des activités sexuelles), les<br />
accouplements ont été obtenus en captivité. Pour chaque femelle, un séjour de dix<br />
jours en contact avec de nombreux mâles a été organisé dans un enclos intérieur<br />
(2.5 x 1.5m) avec une source de chaleur (radiants 1000w) et de l’eau ad-libitum.<br />
Cette méthode permet d’obtenir des copulations très facilement avec des<br />
accouplements observés quelques heures après la capture.<br />
41
Alimentation<br />
Un fois placés dans les enclos extérieurs les animaux sont nourris avec des souris<br />
d’élevage fraichement tuées et posées, quand les conditions météorologiques sont<br />
jugées favorables, à proximité des abris. A chaque nourrissage les proies offertes<br />
sont standardisées au sein de chaque enclos (± 1g). La consommation des proies<br />
est enregistrée par observation directe ou par des moyens plus indirectes quand<br />
l’observation n’est pas possible (palpation de souris dans l’estomac ou changement<br />
brutal de masse). Les souris non ingérées sont retirées très rapidement des enclos<br />
(6 à 12 heures plus tard). En combinant la masse des proies offertes avec les indices<br />
de consommation, il est possible de connaître la quantité <strong>total</strong>e de proies (g)<br />
ingérées pour chaque serpent dans chaque enclos pendant l’experience (voir détails<br />
sur la méthode, article 2. La quantité et la masse des proies offertes par enclos<br />
varient selon les dispositifs expérimentaux et la problématique examinée. Dans tous<br />
les cas, les femelles sont réparties de façon aléatoire dans chaque enclos et lots<br />
experimentaux.<br />
Mesures<br />
Les pesées ont lieu regulièrement et varient selon les expériences. Le nombre de<br />
follicules en développement (vitellogénèse) ou d’embryons (gestation) sont<br />
déterminés par palpation. Peu de temps avant la mise bas, les femelles sont toutes<br />
recapturées et installées au laboratoire dans des boites individuelles jusqu’à la<br />
parturition. Le protocole appliqué aux mises bas est identique en tout point à celui<br />
décrit pour les femelles reproductrices des Moutiers en Retz.<br />
Afin d’étudier le métabolisme aérobie des vipères pendant la gestation, j’ai adapté la<br />
chambre calorimétrique du <strong>CEBC</strong> pour la mesure de petites consommations en<br />
42
oxygène (avec l’aide technique de Guy Merlet et Laurence Pastout). Ce travail de<br />
mise au point a fournit un support de base pour l’étude de la relation entre fécondité<br />
et niveau métabolique, menée par Mitchell Ladyman (Honours Degree 2000). Des<br />
informations détaillées sur le protocole et les caractéristiques techniques sont<br />
données dans la section matériels et méthodes de l’article 6.<br />
3. Les paramètres de la gestion de la resource : condition<br />
corporelle et changements de masse<br />
Comme nous l’avons vu, de nombreux travaux indiquent que la vipère aspic est un<br />
reproducteur sur réserves qui accumule d’importants stocks de lipides (Bonnet 1996 ;<br />
Bonnet & Naulleau 1994 ; Bonnet et al. 1994). Il existe donc un décallage temporel<br />
entre les phases d’aquisition (accumulation de l’énergie) et d’allocation dans la<br />
reproduction. Dans le cadre de notre problématique, la compréhension du système<br />
de gestion de la resource (alimentation versus réserves) est donc d’un interêt crucial<br />
pour déterminer les influences spécifiques des réserves et de la nourriture sur<br />
l’expréssion des traits d’histoire de vie de l’espèce. Dans un pareil cadre, il est<br />
nécessaire de pouvoir estimer précisement l’état et la dynamique des réserves au<br />
cours du cycle reproducteur. Ces informations vont être fournies par différentes<br />
mesures complémentaires : la condition corporelle et les variations de masses.<br />
Condition corporelle, mesure et signification<br />
D’une facon générale, la mesure de l’état physiologique ou “condition corporelle” des<br />
organismes est d’un interêt majeur en éco-physiologie en apportant des informations<br />
sur le succès en chasse et l’état des réserves (Jakob et al. 1996). La signification du<br />
terme condition corporelle est évidemment très variable selon les organismes et<br />
notamment selon le système d’allocation de l’énergie (reproduction sur revenu ou sur<br />
43
éserves). Il existe de nombreuses méthodes de mesures invasives ou non qui ont<br />
toutes pour objectif de s’affranchir de la taille corporelle lorsque l’on désire comparer<br />
la masse ou l’état de nutrition des individus.<br />
Chez la vipère aspic femelle, la condition corporelle peut être calculée sous la<br />
forme d’un ratio entre masse et taille ou en appliquant la méthode des résidus<br />
(Bonnet & Naulleau 1994 ; Bonnet 1996). Dans ce cas, on réalise une régréssion<br />
entre masse et taille corporelle après tranformation logarithmique. La distance<br />
résiduelle entre un point et la ligne de régression fournit un estimateur de condition.<br />
Cette méthode offre de nombreux avantages en permettant notamment une<br />
distinction nette entre l’effet de la taille corporelle et l’effet de la condition. C’est la<br />
méthode idéale pour tester des hypothèses sur la condition corporelle d’individus<br />
provenant d’une même population. L’estimation de la condition à partir des résidus<br />
permet souvent une interprétation biologique très précise (Jakob et al. 1996).<br />
Garcia-Berthou (2001) a récemment formulé une critique sur l’utilisation des résidus<br />
pour la comparaison de differents groupes d’individus. Cet auteur préfère alors<br />
l’utilisation d’analyses de covariance (avec la masse en variable dépendante et la<br />
taille corporelle en co-facteur) plus adaptées pour la comparaison de groupes.<br />
Dans le présent travail, l’utilisation des deux méthodes de calculs (résidus et<br />
covariances) a conduit à des résultats similaires n’affectant pas les conclusions.<br />
Chez la vipère aspic, la condition corporelle est un paramètre central qui fournit des<br />
informations très précises et très différentes selon la période biologique examinée.<br />
Nous avons ainsi distingué trois grandes périodes de mesures :<br />
Condition corporelle initiale : au début du printemps (de Mars à Avril avant les<br />
premières prises alimentaires) la condition corporelle renseigne sur l’état des<br />
réserves et notamment les stocks de corps gras pré-vitellogéniques (avant toute<br />
croissance des folliculles). Il s’agit donc de l’état “initial” des réserves réalisées les<br />
44
années précédentes. La qualité de l’estimation est largement confimée par des<br />
méthodes de quantification absolue non invasive (RMN).<br />
Condition corporelle pre-partum : la condition corporelle pre-partum est calculée juste<br />
avant la mise bas. Cet estimateur complexe intègre à la fois la masse de la portée et<br />
la masse de la “carcasse” de la femelle. La comparaison entre les masses ajustées<br />
initiales et pre-partum fournit un bonne indice de l’alimentation pendant cette période.<br />
Condition corporelle post-partum : la condition corporelle post-partum est basée sur<br />
la masse corporelle juste après la mise bas. A cette période, les stocks de réserves<br />
sont très réduits, voir absents et les femelles présentent un amaigrissement très net<br />
au niveau des muscles squelletiques. Le calcul de la condition corporelle post-partum<br />
constitue alors un bon indice du degré d’émaciation maternelle après la reproduction<br />
(Bonnet et al. 2000a).<br />
Les variations pondérales<br />
Nous bénéficions donc d’un estimateur précis de l’état physiologique des vipères au<br />
cours du cycle reproducteur. Les variations pondérales vont nous fournir des<br />
informations complémentaires sur la balance énergétique et notamment la cinétique<br />
des réserves. La vipère aspic est un vertébré ectotherme qui ne produit pas de<br />
chaleur de façon endogène. En dehors de toute contraintes énergétiques<br />
(reproduction, croissance) les prises alimentaires vont donc être associées à une<br />
augmentation proportionnelle de la masse corporelle et des stocks de lipides. Les<br />
variations pondérales observées entre le printemps et l’automne chez les individus<br />
non-reproducteurs vont donc nous renseigner très finement sur les prises<br />
alimentaires et la dynamique des réserves lipidiques (voir explications détaillées<br />
article 1). Chez les individus reproducteurs, la situation est moins simple car il existe<br />
45
des demandes énergétiques accentuées pendant la vitellogénèse et la gestation.<br />
Néanmoins, la situation énergétique d’un ectotherme est très contrastée par rapport<br />
à un organisme endotherme (Bradschaw 1997). Ainsi et en dépit des contraintes<br />
énergétiques de la reproduction, les variations pondérales des femelles vont aussi<br />
nous renseigner sur l’alimentation (Lourdais et al. 2002a,c). Il va alors être possible<br />
d’estimer la quantité de nourriture consommée par les femelles reproductrices dans<br />
l’année (différence entre la masse initiale et la masse pre-partum Lourdais et al.<br />
2002b) ou bien de façon spécifique à la vitellogénèse (différence entre la masse<br />
initiale et la masse à l’ovulation Bonnet et al. 2001b, Lourdais et al. 2002c) ou à la<br />
gestation (différence entre la masse à l’ovulation et la masse pre-partum ; Lourdais et<br />
al. 2002a).<br />
Figure 6. Mise-bas de Vipère aspic en septembre<br />
46
II. Le système d’allocation<br />
de l’énergie<br />
47
A. Résumé du Chapitre:<br />
Chez la vipère aspic, les réserves corporelles sont déterminantes pour la “décision”<br />
de reproduction (Naulleau & Saint Girons 1981 ; Bonnet & Naulleau 1994,1995 ;<br />
Bonnet 1996). Par ailleurs, cette espèce se nourrit essentiellement de campagnols<br />
dont la démographie est très fluctuante. Ce premier chapitre présente des résultats<br />
qui précisent les effets de l’acquisition de nourriture sur la reproduction à différentes<br />
échelles temporelles. Ce chapitre est composé de trois articles : le premier est basé<br />
sur des données de terrain, le second présente des résultats expérimentaux qui<br />
procurent un lien fonctionnel avec le troisième lui aussi basé sur le travail de terrain.<br />
Dans un premier temps (article 1), nous avons cherché à clarifier, en<br />
situation naturelle (aux Moutiers-en-Retz), l’influence de l’alimentation sur la<br />
fréquence de reproduction et le succès reproducteur des femelles de vipère aspic.<br />
Ce travail suggère l’existence d’un système d’allocation complexe combinant à la fois<br />
des aspects de reproduction sur réserves (”capital breeding”) et de reproduction sur<br />
revenu facultative (“facultative income breeding”). Ainsi, si la fréquence des<br />
reproductions est influencée par les réserves corporelles, la nourriture obtenue<br />
pendant la vitellogénèse va directement influencer la masse des jeunes produits. La<br />
taille des portées semble quand à elle déterminée par les influences conjuguées des<br />
réserves corporelles et la consommation de proies au cours de la reproduction. Ces<br />
résultats indiquent que si la vipère aspic est un reproducteur sur réserve “typique”<br />
pour certains aspects, d’autres aspects (dynamique folliculaire et la masse des<br />
jeunes) pourront être influencés par la nourriture obtenue pendant la vitellogénèse.<br />
Nous avons ensuite cherché à examiner expérimentalement les effets des<br />
réserves et des prise alimentaires sur le recrutement folliculaire et le nombre de<br />
follicules qui parviennent à maturité (article 2). Nous avons soumis un groupe de<br />
48
femelles reproductrices capturées en début de saison de reproduction à différentes<br />
conditions d’alimentation. Les résultats obtenus confirment l’importance de la<br />
nourriture obtenue pendant la vitellogénèse en tant que modulateur des processus<br />
de déclenchement de la reproduction et de recrutement des follicules qui sont<br />
contrôlés par les réserves corporelles initiales. En limitant le phénomène de<br />
régression folliculaire, cette source d’énergie va agir sur le nombre final de follicules<br />
qui arrivent à l’ovulation. La vipère aspic n’est non pas un reproducteur sur réserve<br />
“rigide” mais va pouvoir ajuster son investissement en fonction des ressources<br />
disponibles pendant la phase d’allocation de l’énergie. Une telle flexibilité peut être<br />
très avantageuse, particulièrement dans un système ou la nourriture fluctue<br />
beaucoup et de façon imprévisible d’une année à l’autre. Ainsi, elle permettrait aux<br />
femelles d’améliorer leur succès reproducteur en profitant d’opportunités<br />
alimentaires. Le capital de réserve minimum assure, quand à lui, une reproduction<br />
efficace, y compris lorsque la nourriture est peu abondante au cours de la<br />
reproduction et/ou les années précédentes : les réserves corporelles pouvant être<br />
stockées sur des périodes prolongées (années).<br />
Dans la troisième partie (article 3), nous avons examiné les bénéfices de<br />
cette flexibilité dans le système d’allocation en prenant appui sur dix ans de suivi<br />
longitudinal de la population des Moutiers-en-Retz ; notamment en examinant les<br />
conséquences des fortes variations trophiques et climatiques. Dans un<br />
environnement fluctuant, les avantages du système de gestion des ressources<br />
observé chez les femelles vipères apparaissent nettement : 1) La mise en place de<br />
réserves corporelles longtemps avant la reproduction va assurer décalage temporel<br />
entre acquisition et allocation en permettant aux vipères de se reproduire avec<br />
succès même les années ou la nourriture est rare. L’efficience de reproduction est<br />
donc toujours élevée et partiellement indépendante des conditions d’alimentation<br />
49
courante. 2) Lorsque la vipère aspic se reproduit une année où les rongeurs sont<br />
abondants, l’énergie acquise ne va pas être uniquement allouée dans la portée. Une<br />
fraction de cette énergie va ainsi influencer les caractéristiques maternelles en<br />
limitant le niveau d’amaigrissement après la mise bas. Enfin, les contraintes<br />
thermiques associées à la gestation sont aussi clairement révélées. Durant les étés<br />
chauds, les opportunités de thermorégulation sont accentuées et la durée moyenne<br />
de gestation est plus courte. Parallèlement, les coûts métaboliques associés au<br />
maintien de régimes thermiques élevés augmentent, ce qui accélère<br />
l’amaigrissement des femelles après la mise bas.<br />
Chez les femelles non-reproductrices, les fluctuations d’abondance en proies<br />
affectent directement la dynamique des stocks lipidiques et la croissance corporelle.<br />
La température joue aussi un rôle positif sur les gains de masse en influençant<br />
probablement les conditions de digestion. Les plus forts gains en masse et en taille<br />
sont observés l’année de plus forte abondance en campagnols (1996). De façon<br />
logique, les femelles qui s’engagent dans la reproduction en 1997 présentent des<br />
réserves corporelles très importantes. Pour cette année particulière, l’investissement<br />
reproducteur n’est donc pas limité par l’état des réserves corporelles. Les contraintes<br />
spatiales (volume abdominal) sur l’effort reproducteur sont alors révélées et pour<br />
cette année seulement on détecte une relation significative entre la taille maternelle<br />
et la taille de la portée. Cela signifie aussi que la plupart des années dans notre<br />
étude, l’état des réserves corporelles accumulées avant la reproduction combinée à<br />
la disponibilité alimentaire au cours de la reproduction constituent les principales<br />
limites de la fécondité.<br />
50
B. Article 1<br />
Short-term versus long-term effects of <strong>food</strong><br />
<strong>intake</strong> on reproductive output in a viviparous<br />
snake (Vipera aspis)<br />
X. Bonnet 1 2 , G. Naulleau 1 , R. Shine 4 & O. Lourdais<br />
1 Centre d'Etudes Biologiques de Chizé, <strong>CNRS</strong>, 79360, Villiers en Bois, France<br />
2 Conseil Général Des Deux Sèvres, Rue de L’abreuvoir, 79021, Niort, France<br />
3 University of Poitiers, 40 avenue du recteur Pineau, 86022 Poitiers, France<br />
4 School of Biological Sciences A08, University of Sydney, NSW 2006, Australia<br />
Published in OIKOS 92: 297-308<br />
(2001)<br />
51<br />
1 2 3
Abstract<br />
Feeding rates influence reproductive output in many kinds of animals, but we need to<br />
understand the timescale of this influence before we can compare reproductive<br />
energy allocation to energy <strong>intake</strong>. A central issue is the extent to which reproduction<br />
is fuelled by long-term energy stores ("capital" breeding) versus recently-acquired<br />
resources ("income" breeding). Our data on free-living aspic vipers show that there<br />
is no simple answer to this question: reproductive frequency is determined by long-<br />
term energy stores, offspring size is influenced by maternal <strong>food</strong> <strong>intake</strong> immediately<br />
prior to ovulation, and litter size is influenced by both long-term stores and short term<br />
energy acquisition. Thus, offspring size in free-living vipers reflects the mother’s<br />
energy balance over the preceding year (via a trade-off <strong>between</strong> litter size and<br />
offspring size) as well as her energy balance in the current breeding season. Hence,<br />
different components of a given reproductive output (litter) are not only functionally<br />
linked, but also respond to different temporal scales of prey availability. A female's<br />
body size has little effect on her reproductive output. Attempts to quantify<br />
reproductive energy allocation must take into account the fact that different<br />
reproductive traits (such as offspring size versus number) may respond to energy<br />
availability over different timespans. Thus, although the aspic viper is a typical<br />
"capital breeder" in terms of its reliance on stored reserves for maternal "decisions"<br />
concerning reproductive frequency, it is to some degree a facultative "income<br />
breeder" with respect to the determination of offspring size and litter size.<br />
Key words: annual variations, body condition, energy allocation, litter size, offspring<br />
size, prey availability, reproduction, trade-off, snakes, vitellogenesis.<br />
52
Introduction<br />
<strong>The</strong> relative allocation of resources <strong>between</strong> maintenance, growth, reproduction and<br />
storage is a central theme of studies on life-history evolution (e.g., Stearns 1989,<br />
1992; Roff 1992), but we still cannot provide a convincing answer to R. A. Fisher’s<br />
classic (1930) challenge: "It would be instructive to know not only by what<br />
physiological mechanism a just apportionment is made <strong>between</strong> the nutriment<br />
devoted to the gonads and that devoted to the rest of the parental organism, but also<br />
what circumstances in the life history and environment would render profitable the<br />
diversion of a greater or lesser share of the available resources towards<br />
reproduction". Modelling these kinds of allocation decisions is a relatively<br />
straightforward task, and much has been accomplished in this respect (e.g., Charnov<br />
1982). Measuring energy allocation among these pathways, in a form that is directly<br />
relevant to those life-history models, has proved to be a more challenging<br />
proposition. This is especially true for long-lived organisms living in places where<br />
<strong>food</strong> availability fluctuates on a seasonal or annual basis. Quantifying rates of <strong>food</strong><br />
<strong>intake</strong> and various expenditures is logistically feasible, but simply measuring the<br />
magnitude of these pathways (although of interest in its own right) does not enable<br />
us to answer Fisher’s question about the mechanisms of allocation. In order to<br />
answer the question, we need to understand how fluctuations in <strong>food</strong> supply<br />
influence allocation decisions by the organism.<br />
One central problem in this respect involves the timescale at which those<br />
fluctuations occur. Attempts to quantify allocation pathways necessarily invoke a<br />
timescale for measurement: for example, we must compare energy gain to energy<br />
expenditure over some specified period of time. Clearly, the relevant timescale will<br />
differ among taxa: for example, it will be longer for an animal that stores energy for a<br />
53
long period prior to expending it on reproduction. <strong>The</strong> timescale will also differ<br />
among habitats: for example, it will be longer in a habitat with low <strong>food</strong> availability,<br />
where the animal must forage for longer to gain enough energy to initiate<br />
reproduction. Such difficulties may be overcome by choosing the appropriate<br />
timescale for the system in question. However, another complication may also arise,<br />
that is more difficult to address. For a given reproductive bout by a single animal,<br />
some aspects of reproductive output may be governed by energy availability over a<br />
long timescale whereas other aspects are determined by short-term energy <strong>intake</strong>.<br />
In this paper we document such a case: a female aspic viper’s "decision" as to<br />
whether or not to reproduce, and how large a litter to produce, are largely governed<br />
by long-term energy stores; but her "decision" as to offspring size is driven by short-<br />
term rates of <strong>food</strong> <strong>intake</strong>. Unless we know about such <strong>relationship</strong>s, we cannot<br />
meaningfully compare resource availability to reproductive output, or apply concepts<br />
such as "capital" versus "income" breeding to biological systems.<br />
<strong>The</strong> division <strong>between</strong> "capital" and "income" breeders refers to the source of<br />
nutrients used to support reproductive expenditure (Drent & Daan 1980; Jönsson<br />
1997). Income breeders derive these resources directly from <strong>food</strong> consumed during<br />
the reproductive season, whereas capital breeders derive these resources from<br />
reserves that are developed prior to the reproductive season. This concept has<br />
proved to be useful in studies on birds (e.g. Chastel et al. 1995), but has been rarely<br />
explored in other vertebrates (Jönsson 1997; Doughty & Shine 1998). Several<br />
features of ectothermy preadapt reptiles to reliance upon capital breeding (Bonnet et<br />
al. 1998). Among reptiles, snakes are excellent models to study such strategies<br />
because most species provide no parental care (Shine 1988a) and embryonic<br />
development is primarily lecithotrophic, with minimal placental transfer of energy<br />
(Stewart 1992). Thus, the resources allocated to offspring are fully committed prior to<br />
54
ovulation, rather than being provided over a long period during gestation or after<br />
hatching (or birth). Because the number and size of offspring in snakes are<br />
determined during vitellogenesis, it is easier to quantify maternal investment than<br />
would be the case in many other kinds of animals. Furthermore, the wide range in<br />
body-sizes of adult females within a single snake population (Andrews 1982)<br />
provides an opportunity to examine the influence of maternal size on reproductive<br />
output.<br />
Previous studies suggest that two main factors are likely to play an important<br />
role in determining reproductive output in these animals: (i) maternal body size,<br />
because space to hold the clutch depends upon female size (Vitt & Congdon 1978;<br />
Shine 1988b; Seigel & Ford 1987; Ford & Seigel 1989a); and (ii) energy availability<br />
(Ford & Seigel 1989b; Seigel & Ford 1991). In turn, the latter factor can be separated<br />
into two components: <strong>food</strong> <strong>intake</strong> during follicular growth ("income") versus maternal<br />
reserves at the onset of vitellogenesis ("capital"). Long-term energy stores are likely<br />
to influence reproductive decisions in some snake species (Diller & Wallace 1984;<br />
Blem & Blem 1990; Brown 1991; Bonnet et al. 1994; Naulleau & Bonnet 1996), but<br />
not in others (Plummer 1983; Ford & Seigel 1989b; Whittier & Crews 1990; Naulleau<br />
& Bonnet 1995). For example, preliminary studies report a positive correlation<br />
<strong>between</strong> maternal reserves and offspring number in Vipera aspis but no such<br />
<strong>relationship</strong> in Elaphe longissima (Bonnet & Naulleau 1994; Naulleau & Bonnet<br />
1995).<br />
Published data suggest that the aspic viper, Vipera aspis, offers a classic<br />
example of a typical "capital breeder". In snakes, body-condition (mass scaled by<br />
size) reflects their long-term foraging success during the preceding year(s) (Forsman<br />
1996 ; Shine & Madsen 1997). Adult female aspic vipers need to accumulate very<br />
large body reserves to reach the body-condition threshold necessary for the induction<br />
55
of vitellogenesis, and hence postpone reproduction for a long period (up to 4 years)<br />
while they are accumulating those reserves (Naulleau & Bonnet 1996).<br />
Vitellogenesis in this species involves an intensive mobilisation of maternal reserves<br />
(Bonnet et al. 1994) and maternal energy reserves dictate whether or not a female<br />
will reproduce in a given year (Naulleau & Bonnet 1996). Captive female vipers can<br />
reproduce without feeding during the whole reproductive period (i.e., vitellogenesis<br />
plus gestation), demonstrating that reproduction can be entirely supported by the<br />
energetic "capital" stored prior to reproduction.<br />
<strong>The</strong> aim of our study is to examine the influences of maternal size, energy<br />
stores and current energy <strong>intake</strong>, on the number and size of offspring in a natural<br />
population of aspic vipers. How is the very large capital of stored energy invested<br />
(packaged) during vitellogenesis? Because female aspic vipers continue to feed<br />
during vitellogenesis, we can also examine whether the energy thus obtained has<br />
any influence on reproductive output. Does vitellogenesis rely only on maternal<br />
reserves, or does this snake adopt a more flexible strategy using supplementary<br />
"income" to improve the quality of the litter? In other words, is the aspic viper a<br />
"strict" capital breeder with respect to offspring size and litter size as well as<br />
reproductive frequency?<br />
Materials and methods<br />
Study site and animals<br />
Over a period of seven years (1992-1998), we studied a large (more than 1000 adults<br />
individually marked) population of aspic vipers, Vipera aspis, at Les Moutiers en<br />
Retz in western central France near the Atlantic Ocean (47°03'N; 02°00'W). This<br />
population is near the northern limit of the geographic range of the species. <strong>The</strong><br />
56
study area of approximately 33 ha consists of fields and paths bordered by hedges.<br />
In fields not used for farming, brambles, brushwood and small trees are common.<br />
<strong>The</strong> study population is separated from adjacent populations by several roads and a<br />
village (see Bonnet & Naulleau 1996 for a more detailed description).<br />
<strong>The</strong> aspic viper is a medium-sized (typically to 55 cm body length, 100 g), slow-<br />
moving, terrestrial, venomous snake species (Naulleau et al. 1996). Autopsies and<br />
NMR imaging confirm that in this species, good body condition (high mass relative to<br />
body length) is related to large body reserves (abdominal fat bodies: Bonnet and<br />
Naulleau 1994, 1995; Naulleau & Bonnet 1996; Villevieille 1997). Females more<br />
than 41 cm snout-vent length (= 47 cm <strong>total</strong> body length, the minimal body size<br />
where parturition has been observed) were considered sexually mature (henceforth<br />
referred to as adult).<br />
Procedures and measurements<br />
Any study to investigate the influence of maternal reserves on reproduction in free-<br />
living animals must meet the following criteria:<br />
(i) maternal condition must be recorded at the onset of vitellogenesis; this requires<br />
precise information about the timing and kinetics of follicular growth, ideally in relation<br />
to mobilisation of maternal reserves;<br />
(ii) <strong>relationship</strong>s <strong>between</strong> maternal size, body condition and energy reserves must be<br />
quantified, so that we can estimate the amount of reserves a female can devote to<br />
her clutch;<br />
(iii) quantitative data about <strong>food</strong> <strong>intake</strong> during follicular growth would be useful;<br />
failing this, we need a index of <strong>food</strong> consumption over this period,<br />
57
(iv) we need to recapture the monitored females before laying or parturition (which<br />
typically occurs 2 to 6 months after the start of vitellogenesis in snakes: Seigel & Ford<br />
1987), in order to obtain data on reproductive output.<br />
In the present study, snakes were caught by hand and individually marked by<br />
scale clipping or (later in the study:1993) with electronic tags (11±1mm length [mean<br />
± 1S.D. as in all subsequent results]; 2.1±0.1mm diameter; 0.25±0.1g <strong>total</strong><br />
mass;125KH, sterile transponder TX 1400L, Rhône Mérieux, Destron/IDI INC). <strong>The</strong><br />
study area was intensively searched almost every day throughout the active season<br />
(early February to October) by one to three people; the <strong>total</strong> searching effort<br />
represents more than 3500 h in the field. More than 500 adult female vipers have<br />
been marked since 1992, and their snout-vent lengths, <strong>total</strong> body lengths (± 0.5cm),<br />
and masses (± 1g with an electronic scale) recorded in the field. Body condition was<br />
calculated as the residual scores from the regression of the natural logarithm of body<br />
mass against that of body length (Jayne & Bennett 1990; Madsen & Shine 1992a;<br />
Naulleau & Bonnet 1996). <strong>The</strong> snakes were released at the exact point of capture<br />
within 15 min, and more rapidly after subsequent recaptures (except before<br />
parturition, see below). Measurements were made at three stages during the<br />
reproductive season: (i) at the onset of vitellogenesis, (ii) at the end of vitellogenesis<br />
(late May, i.e. close to ovulation), and (iii) before parturition (Figure 1). In the aspic<br />
viper in western central France, vitellogenesis begins in March (Bonnet et al. 1994),<br />
but a major increase in follicular size does not occur until late April (Saint Girons<br />
1957b; Saint Girons & Duguy 1992; unpublished data from nuclear magnetic<br />
resonance; Figure 1). Vipers are easier to catch at the beginning of the reproductive<br />
season, and many females were caught in March or April. Feeding activity is<br />
reduced during this first period of vitellogenesis, and there is generally a slight<br />
decrease in body mass. Thus, data gathered in March and April were pooled. Body<br />
58
masses obtained during the periods of rapid follicular growth (May and June) and<br />
pregnancy (July-August), or from individuals with a prey item in the stomach, were<br />
not used to calculate body condition.<br />
Figure 1. In the aspic viper (Vipera aspis), vitellogenesis begins in early March, with ovulation<br />
occurring three months later in early June. Follicular growth (yolk deposition) accelerates in late April<br />
and is complete in early June. Embryonic development and egg hydration (dashed zones) occur later,<br />
from June to August during gestation until parturition three months later. In the present study, 44<br />
reproductive females were captured at three different physiological states: (i) at the onset of<br />
vitellogenesis (arrow 1), (ii) at the end of vitellogenesis (arrow 2), and (iii) before parturition (arrow 3).<br />
Maternal body condition index (an estimate of previtellogenic body reserves) was calculated at the<br />
onset of vitellogenesis.<br />
Food <strong>intake</strong> and mass gain during vitellogenesis<br />
We do not have data on actual prey masses from female in the field, because we<br />
were unwilling to stress these animals by forced regurgitation. Thus, we used the<br />
increase in female body mass during vitellogenesis as an index of <strong>food</strong> <strong>intake</strong>. <strong>The</strong><br />
validity of this method was tested using two data sets:<br />
59
(i) <strong>food</strong> consumption versus change in body mass for 30 captive female vipers<br />
(collected from a variety of locations in France). <strong>The</strong> captive females were kept in<br />
individual cages (40x40x40cm, an electric bulb [60W] provided a thermal gradient,<br />
water was at libitum). <strong>The</strong> snakes were offered laboratory mice (weighed to the<br />
nearest 0.1g) each week, and were weighed and palpated regularly. Ovulation date<br />
was determined from palpation, radiography (Naulleau & Bidault 1981) and<br />
parturition dates. <strong>The</strong> changes in body mass recorded during vitellogenesis were of<br />
the same order of magnitude in captive (9.1 ± 25.0; range -33.0 to 56g) versus free-<br />
ranging female vipers (see results).<br />
(ii) meals records versus change in body mass in free-living snakes. Recent meals<br />
in snakes can be detected by palpation (e.g., Fitch 1987). Although we could not<br />
quantify precisely prey consumption in the field, we can compare mass change in<br />
snakes recorded to feed during vitellogenesis compared to those that were not<br />
recorded to feed over this period.<br />
Reproductive output<br />
One hundred and forty six gravid females were caught one to 21 days before<br />
parturition, in late August-early September (Figure 1), and placed in individual cages<br />
in the laboratory until they gave birth. To avoid a selection towards obviously gravid<br />
females during capture (distended body, and well-developed embryos easily<br />
detected by palpation), all females with identifiable items in the abdomen (detected<br />
by palpation, excluding the stomach region) were also collected. Palpation enabled<br />
us to detect objects as small as 2g (corroborated by dissection, unpublished data).<br />
Captive females were monitored every day until parturition, and weighed (± 0.1 g)<br />
every two days and immediately after parturition.<br />
60
We recorded the number, mass (± 0.1 g), and length (± 0.5 cm) of healthy<br />
offspring, stillborn and relatively undeveloped embryos, and the number and mass of<br />
unfertilised eggs (± 0.1 g). To analyse reproductive output, we made a distinction<br />
<strong>between</strong> the “classical” measures of <strong>total</strong> litter size (henceforth LS) or <strong>total</strong> litter mass<br />
by including healthy neonates, stillborn offspring and undeveloped eggs as well (Farr<br />
& Gregory 1991; Gregory et al. 1992); whilst we included only viable neonates in our<br />
measures of “effective” reproductive output. Thus, fourteen females that did not<br />
produce any healthy offspring were excluded from several analyses. <strong>The</strong> other 132<br />
females gave birth to 699 healthy offspring. <strong>The</strong> mean body mass of healthy<br />
offspring per litter was used to test the predicted negative <strong>relationship</strong> <strong>between</strong><br />
offspring number and size. To control for the effect of female body length, we used<br />
partial correlation analyses (Ford & Seigel 1989a).<br />
Statistics<br />
Data gathered in 1992 were excluded from several analyses on reproductive output<br />
because we selected five obviously gravid, and large, females this year. This bias<br />
should be important for inter-annual comparisons, but probably did not influence<br />
other analyses. Estimates of population size were obtained using the program<br />
CAPTURE (see Bonnet & Naulleau 1996 for further details on methods). In every<br />
case, the first models suggested by the goodness-of-fit tests were Mth (population<br />
estimate under individual heterogeneity in capture probabilities) or Mh (population<br />
estimate under time variation and individual heterogeneity in capture probabilities;<br />
Chao et al. 1992), and they were systematically adopted. <strong>The</strong> differences <strong>between</strong><br />
the two estimates were small, and we conserved the first selected model.<br />
Snakes are very secretive animals (Seigel 1993), and the breeding frequency of<br />
female aspic vipers is low in western central France (Bonnet & Naulleau 1996).<br />
61
Thus, despite a large initial sample size (> 300) and relatively high recapture rates,<br />
we obtained complete data for only 44 females. In this complete data set, each of<br />
these animals was captured at least three times during the 6 months reproductive<br />
period: early in vitellogenesis (before the 15th of April for any given year), close to the<br />
time of ovulation (from 15 May to 15 June), and close to parturition (3 weeks to one<br />
day before; figure 1). <strong>The</strong> low probability of recapturing a snake at three precise<br />
occasions separated by long time intervals combined with the exclusion of individuals<br />
caught with prey in the stomach were excluded from analyses, explains why only 44<br />
females were included in the "complete" data set. <strong>The</strong>se females did not differ from<br />
other females in mean snout vent length or reproductive characteristics (Table 1).<br />
However, this sub-sample of females tended to over-represent particular years<br />
particularly those with high proportions of reproductive animals, and a low feeding<br />
rate because we did not use data on body masses of animals containing recently-<br />
ingested prey.<br />
Table 1. Characteristics of 146 reproductive female aspic vipers (Vipera aspis) and their litters. SVL =<br />
snout-vent length, Fit litter size = number of healthy neonates, Offspring mass = mean mass of healthy<br />
neonates (calculated as the mean of the means gathered on 132 litters). * df = 1,130 for mean<br />
offspring body mass. "Complete data set" = females monitored throughout the entire reproductive<br />
year, from the onset of vitellogenesis until parturition. "Others" = females caught after the onset of<br />
vitellogenesis. <strong>The</strong>re was no difference <strong>between</strong> the two subsets of data ("complete" versus "others")<br />
in maternal size or in the reproductive characteristics that we measured.<br />
SVL<br />
(cm)<br />
Total litter<br />
size<br />
Fit litter<br />
size<br />
Total litter<br />
mass (g)<br />
Offspring<br />
mass (g)<br />
Complete n = 44 49.1 ± 3.2 5.8 ± 1.6 4.7 ± 2.3 32.0 ± 12.9 6.3 ± 0.9<br />
Others n = 102 48.7 ± 3.3 6.3 ± 2.2 4.9 ± 2.9 35.6 ± 16.2 6.3 ± 1.1<br />
F(1,144)* 0.45 2.0 0.10 1.64 0.15<br />
p 0.50 0.16 0.75 0.20 0.69<br />
62
To ensure that these biases did not substantially affect our conclusions, we also<br />
analysed the larger data set (n = 146) in which many more animals were<br />
incorporated, but with incomplete data for some individuals. We used these data to<br />
examine annual variation in reproductive output. Fourteen of these 146 reproductive<br />
females were represented twice (at intervals of 2 to 4 years, due to the low breeding<br />
frequency), raising the possibility of pseudoreplication. However, none of our results<br />
were modified when we randomly excluded duplicate records from these animals. In<br />
the complete data set (n = 44), every female was represented only once. <strong>The</strong><br />
following results are derived from analyses on these two data sets. All the tests were<br />
performed using STATISTICA 5.1.<br />
Results<br />
Relationship <strong>between</strong> <strong>food</strong> <strong>intake</strong>, growth and weight gain<br />
Females did not increase in body length during reproductive years, except for a slight<br />
increase (1 - 2 cm) in a few individuals in one year when <strong>food</strong> availability was<br />
exceptionally high (1996, see below). However, females gained appreciably in mass<br />
during vitellogenesis (mean mass change was +11.7 ± 15.3g, +12 ± 16% of initial<br />
mass, n = 44), with considerable variation among individuals (range -12 to +51g, -12<br />
to +61% of initial mass). Such body mass variations were not related to the female's<br />
SVL (r = 0.08, P = 0.58, n = 44), nor to her pre-vitellogenic body mass (r = -0.13, p =<br />
0.41, n = 44) or initial body condition (r = -0.14, p = 0.54, n = 44).<br />
A causal link <strong>between</strong> maternal mass gain and feeding was suggested by the<br />
data set on captive snakes: female viper's <strong>food</strong> <strong>intake</strong> during vitellogenesis was<br />
highly correlated with her change in body mass over this period (r = 0.83, n = 30, p <<br />
0.0001; Figure 2). This result was also supported by the data gathered on the 44<br />
63
free-ranging snakes. Using palpation, we recorded prey in 13 reproductive females<br />
in late April – May. All of these animals increased in body mass during vitellogenesis<br />
(mean mass change was +20.7 ± 15.0 g, +24 ± 17% of initial mass; range +2 g to<br />
+51 g, +2 to +60% of initial mass), whereas the other 31 females in which we found<br />
no evidence of feeding showed a lower gain in body mass (mean mass change +8.0<br />
± 14.0g, +8 ±14% of initial mass; range -12g to +51 g, -11 to + 54% of initial mass;<br />
comparing the two groups, F(1.42) = 7.28, p = 0.01). Although many of these 31<br />
females probably fed at some time during vitellogenesis, the difference in mass<br />
change is consistent with the hypothesis that mass change reflects feeding rate.<br />
Thus we conclude that the increase in body mass constitutes a simple and reliable<br />
index of prey consumption over the period of vitellogenesis.<br />
Figure 2. Relationship <strong>between</strong> changes in maternal body mass and <strong>food</strong> <strong>intake</strong> during vitellogenesis<br />
in 30 captive female asp vipers. Food <strong>intake</strong> was calculated as the sum of the mice consumed during<br />
the whole vitellogenic period (3 months). Change in body mass was the difference <strong>between</strong> the<br />
female's mass at ovulation minus her initial mass as recorded at the onset of vitellogenesis. Although<br />
change in body mass reflects the influences of complex phenomena (egg hydration + maternal<br />
metabolic expenditure + individual physiological differences [e.g. litter size]), <strong>food</strong> <strong>intake</strong> strongly<br />
influences maternal somatic weight changes (see text for statistics).<br />
64
Influence of maternal body length on reproductive output<br />
<strong>The</strong> first likely correlate (determinant?) of litter size and offspring size is maternal<br />
body size; numerous researchers have documented strong allometry in these traits in<br />
a wide variety of reptile species (Dunham et al. 1988; Wilbur & Morin 1988). If<br />
maternal body size strongly affects reproductive output, then we need to know this at<br />
the outset so that we can factor out allometric effects in all subsequent analyses. We<br />
used the entire data set (all females and their litters) for this analysis.<br />
Maternal snout-vent length was weakly correlated with <strong>total</strong> litter size (r= 0.17, p=<br />
0.036, n= 146), but not with the number of healthy neonates (r= 0.13, p= 0.12, n=<br />
n=146, Figure 3a). Hence, maternal body size has little effect on litter size in this<br />
population. Larger females tended to produce heavier offspring (r= 0.22, n= 0.01, p=<br />
132, Figure 3b). <strong>The</strong> effect was stronger in a partial correlation analysis where we<br />
held constant the effect of litter size on offspring size (partial correlation: r= 0.24, p =<br />
0.005, n = 132). Thus, our data suggest that a female's body size influences the<br />
mass of her young, but has less effect on the <strong>total</strong> number of neonates that she<br />
produces.<br />
<strong>The</strong> second plausible influence on offspring size and number is the <strong>relationship</strong><br />
<strong>between</strong> these two variables. Given finite resources, an increase in litter size will<br />
reduce mean offspring size. We looked for this effect in the entire data set, using<br />
mean values for offspring size for each litter.<br />
Trade-off <strong>between</strong> developing offspring for maternal resources<br />
Offspring snout-vent length and offspring mass were strongly correlated (r= 0.75, n=<br />
699, p< 0.0001); we use body mass to characterise neonatal size, rather than length,<br />
because it offers a more direct measure of maternal investment. Stillborn offspring,<br />
relatively undeveloped embryos, and unfertilised eggs weigh much less than healthy<br />
65
neonates (mean masses of healthy neonates, stillborn and undeveloped eggs were<br />
6.3 ± 1.1 g [n = 699], 4.4 ± 2.0 g [n = 80], and 2.0 ± 1.1 g [n = 103] respectively), and<br />
so were not included in our test of the influence of litter size on offspring size.<br />
Figure 3. <strong>The</strong> <strong>relationship</strong> <strong>between</strong> maternal size and reproductive output in Vipera aspis. Larger<br />
females do not produce more viable offspring (a), but tend to have slightly larger neonates (b).<br />
66
<strong>The</strong> predicted negative <strong>relationship</strong> <strong>between</strong> offspring number and the mean mass<br />
of healthy neonates was not evident from our raw data (Figure 4a). Using <strong>total</strong> litter<br />
size rather than the number of healthy neonates did not change the significance of<br />
this result (r= -0.15, p= 0.09, n = 132). However, when female body length was held<br />
constant, mean offspring mass was negatively correlated with offspring number<br />
(Figure 4b). Thus, our data demonstrate a weak trade-off <strong>between</strong> offspring size and<br />
number in aspic vipers.<br />
We now turn to the <strong>relationship</strong> <strong>between</strong> energy balance and viper reproduction.<br />
First, we use the data on the 44 females measured at each stage of their<br />
reproductive cycles (the "complete data" females) to assess <strong>relationship</strong>s <strong>between</strong><br />
energy stores, <strong>food</strong> <strong>intake</strong>, and reproductive output.<br />
Influence of pre-vitellogenic body condition on reproductive output<br />
Female vipers that were in good body condition early in vitellogenesis, tended to<br />
produce large litters (r = 0.47, n = 44, p = 0.001; Figure 5a). In contrast, early body<br />
condition had no influence on offspring size (r = 0.09, n = 42, p = 0.56, note that<br />
sample size was reduced because 2 females did not produce any living offspring;<br />
Figure 5b) even when female snout-vent length was held constant (r= -0.09, n= 42,<br />
p= 0.57).<br />
67
Figure 4. <strong>The</strong> <strong>relationship</strong> <strong>between</strong> litter size and offspring size in Vipera aspis. a. <strong>The</strong> number of<br />
viable offspring produced by a female viper is not significantly correlated with the mean mass of her<br />
offspring, unless the analysis factors out the effects of maternal body size on reproductive output (see<br />
Figure 3). If the effect of maternal size is removed (by using a partial correlation analysis), the<br />
underlying tradeoff <strong>between</strong> offspring size and litter size is revealed (b).<br />
Influence of <strong>food</strong> <strong>intake</strong> on reproductive output<br />
<strong>The</strong> change in maternal body mass during vitellogenesis (<strong>food</strong> <strong>intake</strong>) was correlated<br />
with litter size (r= 0.42, n= 44, p= 0.005, Figure 5c), even when female snout-vent<br />
length was held constant (r= 0.44, n= 44, p= 0.002). Similarly, mass gain during<br />
follicular growth strongly influenced offspring size (r= 0.39, n= 42, p= 0.01, Figure<br />
5d). Controlling for the influence of maternal snout-vent length on offspring mass did<br />
not change this result (r= 0.40, n= 42, p= 0.01). When the number of neonates was<br />
68
included as a correcting factor, the influence of maternal mass change on offspring<br />
size was strengthened (r= 0.47, n= 42, p= 0.002; and r= 0.50, n= 42, p= 0.001<br />
including maternal snout-vent length as an additional correcting factor).<br />
Figure 5. <strong>The</strong> <strong>relationship</strong> <strong>between</strong> previtellogenic maternal reserves (initial body condition) (a, b),<br />
maternal somatic weight change during vitellogenesis (an indication of <strong>food</strong> <strong>intake</strong>; see text) (c, d), and<br />
reproductive output in Vipera aspis. Females with high initial body condition produce larger litters (a)<br />
but do not have larger offspring (b). Females that increase substantially in body mass during the<br />
vitellogenic period produce larger litters (c) of larger neonates (d).<br />
<strong>The</strong>se latter results presumably reflect the fact that litter size influences the<br />
amount of energy from a given prey item that can be allocated to each offspring.<br />
<strong>The</strong> rank correlation <strong>between</strong> number and size of offspring in the "complete data" set<br />
(r= -0.20, n= 42, using number of healthy offspring) was similar to that from the larger<br />
69
data set (r= -0.14, n= 132). Because offspring size was positively correlated with<br />
changes in maternal mass during vitellogenesis, we tested for the presence of this<br />
trade-off including change in maternal body mass as a correcting factor. <strong>The</strong>se<br />
analyses confirmed the underlying negative <strong>relationship</strong> <strong>between</strong> offspring mass and<br />
number of healthy neonates (r= -0.35, n= 42, p= 0.025; and r= -0.40, n= 42, p= 0.01<br />
with female snout-vent length held constant).<br />
<strong>The</strong> body condition of a post-parturient female was affected by her change in<br />
body mass during vitellogenesis (r= 0.43, n= 44, p= 0.004). Thus, energy and<br />
materials obtained from the prey during vitellogenesis were not exclusively allocated<br />
to growing follicles, but were also directed to maternal reserves.<br />
Combined effects of maternal length, maternal reserves, <strong>food</strong> <strong>intake</strong> and offspring<br />
competition for energy on reproductive output<br />
Our data allow us to quantify the four main factors that seem likely to influence<br />
reproductive output (number and size of offspring) in female vipers. <strong>The</strong>se factors<br />
are independent of each other; for example, two of them (maternal length and<br />
maternal body condition) are independent by definition. <strong>The</strong> third variable, change in<br />
maternal body mass during the vitellogenic period (<strong>food</strong> <strong>intake</strong>), was not influenced<br />
by either of these variables (see above). <strong>The</strong> fourth independent variable is the<br />
trade-off <strong>between</strong> number versus size of growing follicles, given the finite amount of<br />
energy allocated for vitellogenesis.<br />
Stepwise multiple regression analyses were performed on the "complete data" set<br />
(n= 44) with these four independent variables, and the number or size of offspring as<br />
the dependent variables. <strong>The</strong> highest proportion of the variation in litter size was<br />
explained when all four predictor variables were included in the analyses (r 2 = 0.61,<br />
n= 42, p < 0.00001). However, variation in mean offspring mass was largely<br />
70
explained by maternal weight change and litter size in the regression model (r 2 =<br />
0.25, n = 42, p = 0.003), with no further significant increase in explained variance by<br />
including the other variables (r 2 = 0.30, n = 42, p = 0.008 with the four variables).<br />
Hence, we conclude that the reproductive output of a female aspic viper is<br />
determined by the combined effects of at least four factors: her body length, her pre-<br />
existing energy reserves, her <strong>food</strong> <strong>intake</strong> during vitellogenesis (as estimated by her<br />
weight change over this period) and the trade-off <strong>between</strong> offspring size versus<br />
number (i.e., competition among the offspring for energy).<br />
Annual variation in energy availability and viper reproduction<br />
In order to evaluate the generality of our results on 44 females, we can also examine<br />
patterns of annual variation in the traits of interest in the all females. If our results<br />
from the "complete data" sample are robust, they should enable us to understand the<br />
ways in which annual variation in prey availability influences the reproductive output<br />
of vipers.<br />
Mean maternal body size remained consistent among years, but significant<br />
annual variation occured in some of the reproductive traits (Table 2). <strong>The</strong><br />
consistency in maternal body size, combined with the lack of strong body-size effects<br />
on reproduction (Table 2, Figure 3), substantially simplifies the analysis of correlates<br />
of annual variation in reproductive biology. It also supports the notion that factors<br />
other than maternal size influence reproductive output in our study population.<br />
71
Table 2. Annual variation in the body sizes and reproductive output of female aspic vipers (Vipera<br />
aspis). SVL = snout-vent length, Fit Litter size = number of healthy neonates, Litter mass = <strong>total</strong> litter<br />
mass, Offspring mass = mean mass of healthy neonates. Data for the first four variables (columns)<br />
were analysed by one way ANOVAs with year of the study (1992-1998) as the factor. For the final trait<br />
(mean offspring mass), we used a two-factor ANOVA with litter number nested within year (so df = 6,<br />
557 for offspring mass). <strong>The</strong>se ANOVAs reveal no significant variation in maternal body size (SVL) of<br />
146 reproductive female aspic vipers, but significant variation in some of the characteristics of their<br />
litters. Significant values are indicated in bold faces. Note that excluding 1992 from analyses (see<br />
text) did not change any results.<br />
SVL<br />
(cm)<br />
Total litter<br />
size<br />
Fit litter size Litter mass<br />
F(6,139)* 0.44 3.42 1.58 2.36 7.14<br />
(g)<br />
Offspring<br />
mass (g)<br />
p 0.85 0.01 0.16 0.03 0.0001<br />
Prey availability varied substantially over the course of our study (see Bonnet<br />
et al. 2001b). Vipers feed mainly on voles, whose populations typically fluctuate in<br />
this region in a 3 to 4 year cycle (Krebs and Myers 1974; Delattre et al. 1992). Such<br />
fluctuations directly influence feeding rates of the snakes (as revealed by the<br />
proportion of snakes with a prey in the stomach, Bonnet et al. 2001b). Restricting the<br />
analyses to the vitellogenesis period and to adult females, we can divide the years of<br />
our study into three categories in this respect. Two years (1994 and 1998) were poor<br />
(9% [n = 228] and 10% [n = 81] respectively of the snakes with a prey in the<br />
stomach), three were "medium" (1993, 1995 and 1997 with 14% to 21% [99
i) Reproductive frequency showed substantial variation, but is difficult to quantify<br />
precisely because population sizes also varied (Figure 6; Chi-square = 99.9, df = 5, p<br />
< 0.0001). <strong>The</strong> number of reproductive females strongly decreased after a low <strong>food</strong><br />
year (1994), and increased after a good year (1996). <strong>The</strong> proportion of reproductive<br />
females in comparison to the <strong>total</strong> number of adult females also varied significantly<br />
among years (Chi-square = 46.3, df = 5, p < 0.0001), and was apparently related to<br />
prey availability in the preceding year (Figure 6).<br />
ii) Litter sizes varied among years (Table 2), with an increase of mean litter size<br />
during the "high-<strong>food</strong>" year (1996) and the next ones as well (Figure 7). This is what<br />
we would predict from the notion that litter sizes are mainly affected by existing<br />
energy stores, and by current feeding rates also (see above). Similarly, during a year<br />
of low prey availability (1994) we observed a reduction of mean litter size, and in the<br />
following year as well (Figure 7).<br />
iii) Offspring mass was higher in the "good" year (1996) than at any other time in the<br />
study, and a nested ANOVA (with litter number nested within years) detected<br />
significant variation in offspring body mass among years (Figure 8, Table 2).<br />
Interestingly, during the 1994 "poor" year we did not observe any reduction of<br />
offspring sizes, but such effect may have occurred in 1998 (Figure 8). Thus we<br />
speculate that females allocate additional nutrients to offspring only in exceptionally<br />
"good" years, and hence that modest annual variation in <strong>food</strong> supply will not be<br />
automatically reflected in detectable changes to mean offspring mass. Instead,<br />
changes to offspring mass will be apparent only in occasional very good years<br />
(Figure 8).<br />
73
Figure 6. Annual variation in the <strong>total</strong> number of adult females in a closed population of asp vipers<br />
(black dots ± S.D., black line). <strong>The</strong> observed number of reproductive females (a subset of the <strong>total</strong><br />
number of adult females) is represented with hatched bars (± S.D). <strong>The</strong> expected number of<br />
reproductive females (dotted circles), simply calculated as a constant proportion (33%, see Bonnet et<br />
Naulleau 1996 ) of the <strong>total</strong> number of adult females, is indicated to better visualise the annual<br />
fluctuations in the relative number of reproductive females in comparison to the <strong>total</strong> number of adult<br />
females. <strong>The</strong> arrows with the sign "-" indicates a low <strong>food</strong> availability year, the arrow with the sign "+"<br />
indicates a exceptionally high <strong>food</strong> availability year. Population estimates (± S.D.) were performed<br />
using the program CAPTURE (see text for statistics).<br />
Figure 7. Annual variation in litter size (annual mean ± S.E.) in a closed population of asp Vipers.<br />
Number above each symbol indicate sample size (number of females). <strong>The</strong> arrows with the sign "-"<br />
indicates a low <strong>food</strong> availability year, the arrow with the sign "+" indicates a exceptionally high <strong>food</strong><br />
availability year. See text for statistics.<br />
74
Figure 8. Annual variation in mean offspring mass (annual mean ± S.E.) in a closed population of asp<br />
Vipers. Number above each symbol indicate sample size (number of neonates). <strong>The</strong> arrows with the<br />
sign "-" indicates a low <strong>food</strong> availability year, the arrow with the sign "+" indicates a exceptionally high<br />
<strong>food</strong> availability year. See text for statistics.<br />
Discussion<br />
Our data suggest that the reproductive outputs of a female aspic viper (i.e., the size<br />
and number of offspring in her litter) are affected in complex ways by her energy<br />
acquisition over the months and years preceding the actual birth of the offspring. Her<br />
foraging success over an interval of one to three non-reproductive years determines<br />
the magnitude of her energy stores, and hence the "decision" as to whether or not<br />
she will reproduce in a given year (because reproduction is initiated only after<br />
females attain a precise body-condition threshold: Naulleau and Bonnet 1996). Our<br />
data show that these energy stores (measured by female body condition prior to the<br />
onset of vitellogenesis) determine litter size: females with larger energy stores initiate<br />
vitellogenesis of a greater number of follicles. To our knowledge (see below), this<br />
study demonstrates for the first time (using strict methodological criteria to calculate<br />
75
early body condition) that in snakes, initial maternal reserves gathered over years<br />
before reproduction positively influence reproductive output. Food <strong>intake</strong> after this<br />
time, but prior to ovulation, also seems to affect the female’s reproductive output not<br />
only by changing her litter size, but by modifying the mean mass of her offspring.<br />
Recruitment of follicles starts at the initiation of vitellogenesis in early March (Bonnet<br />
et al. 1994), feeding activity is rare before April, and vitellogenesis culminates in May-<br />
June (see Figure1). Thus, we might expect that <strong>food</strong> <strong>intake</strong> during vitellogenesis<br />
occurs too late to modify the number of growing follicles. However, we found a<br />
positive <strong>relationship</strong> <strong>between</strong> <strong>food</strong> <strong>intake</strong> and litter size. This counterintuitive result<br />
may be due to the influence of <strong>food</strong> <strong>intake</strong> on follicular atresia. Autopsies (Saint<br />
Girons 1957b) and nuclear magnetic resonance imaging data (unpublished) show<br />
that undersized follicles, as well as atretic follicles, are common during vitellogenesis.<br />
Well-nourished females may proceed with vitellogenesis rather than resorption of<br />
some of these smaller follicles.<br />
Importantly, the impact of additional <strong>food</strong> (= energy) on offspring mass<br />
necessarily depends on litter size also, because a larger litter means that a given<br />
amount of energy is divided among a greater number of offspring. Thus, offspring<br />
size in the aspic viper is affected by the mother’s feeding success over the preceding<br />
few years (because her accumulated energy stores will determine litter size) as well<br />
as her feeding success in the weeks immediately preceding ovulation. Similarly, litter<br />
sizes are affected by feeding rates in this vitellogenic period as well as in previous<br />
years. <strong>The</strong> end result is that the link <strong>between</strong> <strong>food</strong> supply and reproductive output is<br />
complex. We must understand short-term as well as long-term rates of <strong>food</strong> <strong>intake</strong><br />
before we can interpret variation in reproductive output in aspic vipers.<br />
Trade-offs <strong>between</strong> litter size and offspring size occur in several snake<br />
species (Ford & Seigel 1989a; Madsen & Shine 1992a), including the aspic viper<br />
76
(see above). Our study also shows that this trade-off (competition among growing<br />
follicles during the initial partitioning of energy) can be obscured by subsequent<br />
variation in <strong>food</strong> <strong>intake</strong> (Lessells 1991). Thus, although large litters should result in<br />
small follicles, this trade-off may be obscured by variation among females in the<br />
magnitude of additional energy reserves gathered during vitellogenesis (see van<br />
Noordwijk and de Jong 1986; Doughty and Shine 1997). Follicles "compete" for<br />
resources not only at the initiation of vitellogenesis, but throughout the vitellogenic<br />
period (when additional energy from recently-consumed prey becomes available).<br />
Thus, sibling ova compete for resources at two distinct levels and for two distinct<br />
energetic sources: first at the onset of vitellogenesis for the energy stored by the<br />
mother prior to reproduction, and second during vitellogenesis for the prey caught by<br />
the mother.<br />
In combination with previous work on this species (Naulleau 1965; Naulleau &<br />
Bidaut 1981; Naulleau et al. 1996), our data clarify the complex <strong>relationship</strong> <strong>between</strong><br />
maternal foraging success and reproductive activity. Non-reproductive individuals<br />
accumulate large body reserves over an interval of one to three years, by storing a<br />
large proportion of the energy they assimilate from prey consumed over that period.<br />
During that time, they remain very secretive (Bonnet and Naulleau 1996) until they<br />
exceed the body-condition threshold necessary for reproduction (Naulleau and<br />
Bonnet 1996). After this long period of abstinence, vitellogenesis takes the form of<br />
an "explosive" investment: almost all maternal reserves are allocated to reproduction<br />
to produce the greatest possible number of young. We have recorded several litter<br />
masses greater than the mass of the post-parturient female (maximum = 112%).<br />
Energy opportunistically obtained by feeding during vitellogenesis is also directed to<br />
the developing follicles, and significantly increases offspring size. After reproduction,<br />
females are very emaciated and many do not survive until the next year (Bonnet<br />
77
1996; Bonnet et al. 2000a). Thus, many female aspic vipers are semelparous,<br />
producing only one litter in their lifetimes (unpublished data). Given the low<br />
probability of breeding again, the “costs” of additional investments to an already high<br />
reproductive effort will be low in terms of decrements in future reproductive<br />
opportunities (Williams 1966a). Consequently, females should be under strong<br />
selection to maximise the effective output from any reproductive attempt, rather than<br />
conserve resources to invest in subsequent litters. This situation may be widespread<br />
in viperid snakes (Madsen & Shine 1993; W. S. Brown, pers. com.).<br />
Reptiles have increasingly been used as "model organisms" for research in<br />
this field (Seigel 1993; Shine & Bonnet 2000). However, both our methods and our<br />
results provide a contrast to most previous work on these animals. Firstly, we focus<br />
on events during vitellogenesis (the period during which most of the variation in litter<br />
size and offspring mass is generated) rather than during pregnancy (e.g., Stewart<br />
1989; Shine & Harlow 1993; Baron et al. 1996; Gregory & Skebo 1998).<br />
Vitellogenesis may extend for periods as long as pregnancy in many reptile species<br />
(including the aspic viper), and is a crucial phase in terms of maternal reproductive<br />
"decisions". Vitellogenesis can be viewed as the “explosive” phase of the energetic<br />
investment of reproductive females (Bonnet et al. 1994). Secondly, maternal body<br />
size has little effect on reproductive output in the aspic viper (Bonnet et al. 2000b), in<br />
strong contrast to most other snake species for which similar data are available (e.g.,<br />
Seigel and Ford 1987). This result is not an artefact of low statistical power; our<br />
sample size is larger than in most previous analyses (mean sample size was 32.8 ±<br />
51.3 in 61 studies reviewed by Seigel and Ford [1987]). Other studies on the aspic<br />
viper from the same region revealed a positive correlation <strong>between</strong> maternal size and<br />
litter size (Naulleau & Saint Girons 1981). <strong>The</strong> difference <strong>between</strong> these studies is<br />
probably explained by the particular methodology we employed (collection of all<br />
78
animals that were potentially reproductive). If we had searched only for obviously<br />
gravid snakes (e.g. snakes with particularly large litters), we would have ignored<br />
many females with small litters relative to their body size.<br />
Another difference <strong>between</strong> our study and previous analyses is that we have<br />
used a very strict criterion as to when in the reproductive cycle we estimate body<br />
reserves from overall body condition. Thus if measurements are taken on females<br />
during pregnancy or close to the time of ovulation, the reproductive products<br />
(follicles, eggs or embryos) constitute a significant proportion of maternal body mass.<br />
Hence, body condition at this time is likely to be a direct measure of reproductive<br />
output, not an indicator of the magnitude of body reserves such as the fat bodies and<br />
liver (Bonnet & Naulleau 1994, 1995). In such a case, a positive correlation <strong>between</strong><br />
"body condition" and reproductive output is almost inevitable, since the same variable<br />
occurs in both sides of the equation. In addition, the origin of the energy invested<br />
into the clutch cannot be determined by such a methodology, and the respective<br />
influences of <strong>food</strong> <strong>intake</strong> and maternal reserves become indistinguishable. Our data<br />
constitute a basis for developing hypotheses that can be tested by rearing snakes in<br />
controlled diets from the onset of vitellogenesis to the production of the offspring (e.g.<br />
laying or parturition).<br />
Despite these methodological problems, data from other snake species are<br />
sufficient to document substantial interspecific differences in the <strong>relationship</strong> <strong>between</strong><br />
energy acquisition and reproductive expenditure. For example, experimental studies<br />
show that <strong>food</strong> <strong>intake</strong> during vitellogenesis affects clutch sizes but not offspring sizes<br />
in two American snake taxa (Thamnophis marcianus and Elaphe guttata: Ford and<br />
Seigel 1989b; Seigel and Ford 1991). In both of these "income-breeding" genera,<br />
maternal energy reserves may play only a small role in fuelling reproductive<br />
expenditure (Whittier & Crews 1990; Naulleau & Bonnet 1995). In contrast, field data<br />
79
on a "capital-breeding" viperid snake, Vipera berus suggest that offspring size may<br />
be affected by prey availability in this species (Andren & Nilson 1983). Our own data<br />
suggest that reproductive frequency in the aspic viper is driven by long-term energy<br />
stores, but that litter sizes and offspring sizes respond in complex ways to maternal<br />
feeding rates over both the short and the long term.<br />
More generally, we predict that broad patterns will be apparent across<br />
species, depending on the extent of their reliance upon stored energy for breeding.<br />
In "capital" breeders, early body condition will determine reproductive status (Diller &<br />
Wallace 1984; Brown 1991; Naulleau & Bonnet 1996) and will largely determine<br />
offspring number (present study). In such species, breeding frequencies will often be<br />
low because long periods of time are necessary to accumulate energy reserves<br />
(Martin 1993; Bonnet & Naulleau 1996). <strong>The</strong> mass of the clutch relative to maternal<br />
mass will often be high, because there is a massive investment of body reserves to<br />
reproduction. In contrast, we predict that the reproductive decisions of "income"<br />
breeders (which can store only limited body reserves) will be less dependent on<br />
maternal body condition prior to vitellogenesis (Plummer 1983; Whittier and Crews<br />
1990; Naulleau & Bonnet 1995). Breeding frequency should be higher, with the<br />
number of offspring dependent on maternal foraging success immediately prior to<br />
breeding (Ford & Seigel 1989b). Relative clutch (litter) mass should be relatively low.<br />
<strong>The</strong>re is undoubtedly a continuum <strong>between</strong> capital and income breeders in snakes,<br />
as in other organisms (Chastel et al. 1995), so that there is ample opportunity for<br />
robust empirical tests of these predictions. This interspecific diversity in the role of<br />
energy reserves for reproduction, and the partial decoupling of control systems that<br />
regulate different aspects of reproductive output (offspring size versus number) mean<br />
that snakes offer exceptional opportunities to answer Fisher’s (1930) challenge about<br />
the ways in which animals allocate energy to reproduction.<br />
80
Acknowledgements<br />
We thank L. Schmidlin, M. A. L. Zuffi, and C. Thiburce for helping to measure, weigh,<br />
and mark several hundred neonates. We also thank S. Duret and M. Vacher, who<br />
spent many months to catch the snakes. T. Madsen, P. Doughty, D. Reznick, S. J.<br />
Hall and S. D. Bradshaw provided helpful comments on the manuscript. Financial<br />
support was provided by the Conseil Général des Deux Sèvres, the Centre National<br />
de la Recherche Scientifique (France) and the Australian Research Council. Rex<br />
Cambag, J. T. Wilmslow and Gisèle Gaboune helped to solve many technical<br />
problems.<br />
81
C. Article 2<br />
When does a reproducing female viper<br />
"decide" on her litter size?<br />
Olivier Lourdais 1 2 3 , Xavier Bonnet 1 , Richard Shine 4 and Emily N. Taylor 5<br />
1 Centre d'Etudes Biologiques de Chizé, <strong>CNRS</strong>, 79360, Villiers en Bois, France<br />
2 Conseil Général Des Deux Sèvres, Rue de L’abreuvoir, 79021, Niort, France<br />
3 University of Poitiers, 40 avenue du recteur Pineau, 86022 Poitiers, France<br />
4 Biological Sciences A08, University of Sydney, NSW 2006 Australia<br />
5 Department of Biology, Arizona State University, Tempe, AZ, 85287-1501,USA<br />
Accepted for publication in Journal of Zoology (London)<br />
82
Abstract<br />
Some organisms rely on stored energy to fuel reproductive expenditure (capital<br />
breeders) whereas others use energy gained during the reproductive bout itself<br />
(income breeders). Most species occupy intermediate positions on this continuum,<br />
but few experimental data are available on the timescale over which <strong>food</strong> <strong>intake</strong> can<br />
affect fecundity. Mark-recapture studies of free-ranging female aspic vipers have<br />
suggested that reproductive output relies not only on the energy in fat bodies<br />
accumulated in previous years, but also on <strong>food</strong> <strong>intake</strong> immediately prior to ovulation.<br />
We conducted a simple experiment to test this hypothesis, by maintaining females in<br />
captivity throughout the vitellogenic period and controlling their <strong>food</strong> <strong>intake</strong>. A<br />
female's energy input strongly influenced the amount of mass that she gained and<br />
the number of ova that she ovulated. Multiple regression showed that litter size in<br />
these animals was affected both by maternal body condition in early spring (an<br />
indicator of foraging success over previous years) and by <strong>food</strong> <strong>intake</strong> in the spring<br />
prior to ovulation. Our experimental data thus reinforce the results of descriptive<br />
studies on free-ranging snakes, and emphasise the flexibility of energy allocation<br />
patterns among vipers. Reproducing female vipers may combine energy from<br />
"capital" and "income" to maximise their litter sizes in the face of fluctuating levels of<br />
prey abundance.<br />
Key words: capital breeding, vitellogenesis, fecundity, snakes<br />
83
Introduction<br />
Reproduction requires a considerable expenditure of energy, especially in female<br />
organisms that produce large clutches or litters relative to their own body mass.<br />
Because <strong>food</strong> availability fluctuates through time for many species, coupling energy<br />
acquisition (feeding) with expenditure (reproduction) is not a trivial problem. <strong>The</strong><br />
notion of "capital" versus "income" breeding offers a useful conceptual framework in<br />
which to explore such issues (Drent & Daan 1980). Capital breeders gather the<br />
energy to fuel reproduction long before the actual reproductive event, whereas<br />
income breeders simultaneously gain and expend energy. However, these<br />
definitions probably describe the extremes of a continuum. Most kinds of organisms<br />
probably depend to some degree on both kinds of resources to support reproductive<br />
expenditure. Indeed, different facets of reproductive output within the same<br />
reproductive bout by a single female (such as offspring size versus number) may<br />
depend upon different timescales of energy acquisition (Bonnet et al. 2001b).<br />
Squamate reptiles provide good models for studies on this topic, because they<br />
display a diversity in systems of energy allocation. Some of the most extreme<br />
examples of capital breeding systems are found among viperid snakes, with the aspic<br />
viper (Vipera aspis) perhaps the most intensively studied capital-breeder (Saint<br />
Girons 1949, 1957a,b; Saint Girons & Duguy 1992; Bonnet 1996; Naulleau et al.<br />
1999; Bonnet et al. 1999a, 2000a,b, 2001b, 2002b). Female vipers typically<br />
reproduce only once every two to three years and sometimes less often (Saint Girons<br />
& Naulleau 1981). <strong>The</strong>y delay reproduction until they attain a threshold value for<br />
body condition (Naulleau & Bonnet 1996) and females can reproduce successfully<br />
even if they do not feed during the entire year in which the litter is produced (i.e.,<br />
throughout vitellogenesis plus gestation). Thus, "capital" stored prior to reproduction<br />
84
can support the entire reproductive effort of female aspic vipers. Nonetheless, field<br />
data indicate that female vipers often feed during vitellogenesis, and suggest that<br />
<strong>food</strong> acquired at this time can influence some components of reproductive output<br />
(Bonnet et al. 2001b, Lourdais et al. 2002b). <strong>The</strong>se results suggest flexibility in the<br />
system of energy allocation, whereby <strong>food</strong> <strong>intake</strong> in the current year, as well as long-<br />
term storage of energy gained during previous years, can influence a female's<br />
reproductive output.<br />
This scenario concerning the sources of energy for litter production in aspic<br />
vipers is, however, based largely on descriptive studies of free-ranging snakes. In<br />
these studies, maternal feeding rates have been inferred from maternal mass gain<br />
(Bonnet et al. 2001b). Although logic and indirect evidence support the assumption<br />
that these two traits are linked, it remains possible that other factors (such as<br />
maternal disease or metabolic expenditure) could also modify rates of gain in body<br />
mass. If so, correlations <strong>between</strong> mass gain and reproductive output might reflect<br />
such additional factors, rather than a straightforward effect of enhanced feeding rates<br />
on litter sizes. Experimental manipulation of <strong>food</strong> supplies offers a direct and<br />
powerful approach to resolving such uncertainties (Ford & Seigel 1989b; Seigel &<br />
Ford 1991,1992; Gregory & Skebo 1998). To investigate the relative influences of a<br />
female viper's initial body stores and subsequent energy <strong>intake</strong> during vitellogenesis<br />
on her reproductive output, we maintained vipers in captivity, directly modified their<br />
rates of prey consumption, and examined the effects of this manipulation on the<br />
numbers of offspring that they produced.<br />
85
Material and Methods<br />
Study species<br />
<strong>The</strong> aspic viper (V. aspis) is a small viviparous snake that is abundant in central<br />
western France. In this area, females typically reproduce on a less-than-annual<br />
schedule (Saint Girons 1957a,b; Bonnet & Naulleau 1996; Naulleau & Bonnet 1996;<br />
Naulleau et al. 1999). In reproductive females, the recruitment of ovarian follicles is<br />
controlled by endocrine factors at the onset of vitellogenesis in March (Bonnet et al.<br />
1994). After this initial phase, follicular atresia (i.e. death of ovarian follicles prior to<br />
ovulation, Méndez-De la Cruz et al. 1993) is the proximate factor controlling litter size<br />
(Saint Girons 1957b; Saint Girons & Naulleau 1981). That is, females initially begin<br />
to enlarge more follicles than they eventually ovulate, and hence are potentially able<br />
to adjust their eventual (ovulated) litter size depending on conditions that they<br />
experience prior to ovulation. Ovulation occurs during the first two weeks of June<br />
(Naulleau 1981), and parturition occurs two to three months later, from late August to<br />
late September.<br />
Capture and housing<br />
A <strong>total</strong> of 108 snakes (48 males and 60 females) was captured in spring 2000 in<br />
three adjacent localities in West-central France (Château d’Olonnes, Les Sables<br />
d’Olonnes and Rochefort). <strong>The</strong> snakes were collected from late February to mid-<br />
April, when they first emerged to bask after the winter hibernation period. Individuals<br />
were given identification marks by scale-clipping, and were measured (snout-vent<br />
length, SVL + 0.5 cm) and weighed (+ 0.1 g). Mating occurred in captivity, with each<br />
female given a 10-day period of contact with numerous males in an indoor enclosure<br />
(2.5 X 1.5m) with a heat source and water. Copulation was frequently observed in<br />
86
this enclosure. After mating (late March), the snakes were examined by abdominal<br />
palpation to detect vitellogenic follicles. This procedure revealed that thirty-nine<br />
females were reproductive and twenty-one were non-reproductive.<br />
<strong>The</strong> thirty-nine vitellogenic females were placed in eight outdoor enclosures (5 X 3m,<br />
mean density: 5 snakes/enclosure) recreating the natural habitat and exposed to the<br />
climatic conditions of the field research station of Chizé (Forêt de Chizé, Deux-<br />
Sèvres, 46°07’ N, 00°25’ W). <strong>The</strong> enclosures were located side by side and did not<br />
differ in term of orientation and sunlight exposure. Each enclosure was equipped<br />
with the same number of external dens to serve as hiding-places. Water was<br />
provided ad libitum and vegetation (mainly annual grasses, Poacae) was kept high<br />
(20 - 40 cm) to provide shade and shelter.<br />
Experimental design<br />
To examine the effects of absolute energy <strong>intake</strong> during vitellogenesis on subsequent<br />
litter sizes, the 39 females were randomly allocated to one of two diet treatments:<br />
high-<strong>intake</strong> group (n = 19, enclosures 1 to 4): one large mouse (mean mass: 25 ± 5g)<br />
per snake per feeding in each enclosure, provided on four occasions (every two<br />
weeks from mid-April to early June).<br />
low-<strong>intake</strong> group (n = 20, enclosures 4 to 8): one small mouse (14 ± 4g) per snake<br />
per enclosure, offered on two occasions only (mid-April and mid-May).<br />
Snakes of both treatment groups were fed by placing recently killed mice close their<br />
dens, early in the afternoon when climatic conditions were favourable. On each<br />
feeding occasion the mass of prey offered was the same (± 1g) for each replicate<br />
enclosure within each treatment. This method allowed us to calculate the <strong>total</strong> mass<br />
of prey consumed (g) for each snake during the experiment. Uneaten prey items<br />
were removed the next morning. Prey consumption was recorded by direct<br />
87
observation of feeding, or by less direct means if feeding was not observed (by<br />
palpation of mice inside the snake and by increase in body mass).<br />
Snakes were weighed three times during the study: early vitellogenesis (early April),<br />
mid-vitellogenesis (mid-May), and close to the time of ovulation (mid-June). Enlarged<br />
ovarian follicles were counted by palpation (Fitch 1987) at mid-vitellogenesis (mid<br />
May) and ovulation (mid June). This method enables us to detect objects as small as<br />
2 g (Bonnet et al. 2001b). One female from the low-<strong>intake</strong> group was killed by a feral<br />
cat in early June, and hence data from this individual were not used in most of our<br />
analyses.<br />
Our two treatment groups differed both in prey size and number in order to<br />
mimic the natural situation where snakes may sometimes encounter relatively few,<br />
small prey and in other years may encounter prey items that are larger and also more<br />
abundant. Thus, the treatments were designed to span the normal range of variation<br />
in <strong>food</strong> supply in terms of both prey size and prey number. Feeding frequency and<br />
relative prey sizes are important parameters of snake biology that may influence<br />
conversion efficiency (Secor & Diamonds 1995). However, the aim of our experiment<br />
was simply to modify the level of energy available for follicular growth and thus, the<br />
important concern was to generate variations in feeding opportunities comparable to<br />
annual variations in <strong>food</strong> consumption that occur in the field (Lourdais et al. 2002b).<br />
Statistics<br />
Data were analysed using Statistica 5.1. To provide an index of body condition<br />
(mass relative to length), we calculated residual values from the regression of log-<br />
transformed body mass against log-transformed body length (Jayne and Bennett<br />
1990). We compared snout vent length (SVL) of the two groups of females at the<br />
beginning of the experiment in terms of size-frequency distributions as well as mean<br />
88
values. To do so, the data for SVL were standardised (Z = (X - mean value)/SD) so<br />
that the distribution had a mean of zero and standard deviation (SD) of one. Three<br />
size classes were identified: small (Z
Reproductive females in the two <strong>food</strong>-<strong>intake</strong> groups did not differ in mean<br />
body mass (ANOVA, F(1.38) = 0.66; p = 0.42), mean SVL (ANOVA, F(1.38)= 0.50; p<br />
= 0.48), or initial body condition (ANOVA, F(1.38)= 0.25; p = 0.62). <strong>The</strong> two diet<br />
groups were also similar in term of size-frequency distributions (χ 2 = 0.025; df = 2; p<br />
= 0.98). For thirty-seven of the thirty-nine reproductive females, ovarian follicles were<br />
detected each time that we palpated the animals. For the remaining two animals,<br />
however (one in each diet treatment), palpation in mid-vitellogenesis revealed no<br />
detectable eggs. Thus, these animals commenced vitellogenesis but terminated the<br />
process prior to ovulation. <strong>The</strong>se individuals were in lower body condition (residual<br />
values < -0.14 ) that were the other females belonging to the same size class at the<br />
onset of the experiment (residual values > -0.07). Also, neither of these snakes ate<br />
the first prey item they had been offered. After deleting these two cancellations,<br />
females in the two treatments still did not differ significantly in either mean body mass<br />
(p = 0.4), mean SVL (p = 0.43) or more importantly, body condition at the beginning<br />
of the experiment (p = 0.69).<br />
Food <strong>intake</strong> and changes in body mass<br />
Among the 19 individuals of the low-<strong>intake</strong> group, eleven females ate one prey item<br />
and eight females ate two prey items. Among the 17 females of the high-<strong>intake</strong><br />
group, four females consumed only one mouse, seven females consumed two mice<br />
and six females ate three mice. Unsurprisingly, the <strong>total</strong> amount of prey ingested (g)<br />
differed significantly <strong>between</strong> the two groups (Kruskal-Wallis Test: H (1, n = 37) =<br />
15.4, p = 0.0001), with mean values (+SD) of 49 ± 19 of prey consumed by females<br />
in the high-<strong>intake</strong> group compared to 22 ± 7g for the low-<strong>intake</strong> group. <strong>The</strong> variance<br />
in <strong>food</strong> <strong>intake</strong> was also higher for the high <strong>food</strong> group (χ 2 = 0.88, df =1, p < 0.0001),<br />
reflecting the higher number of feeding opportunities within this group.<br />
90
We pooled data from the two groups to examine the <strong>relationship</strong> <strong>between</strong> absolute<br />
<strong>food</strong> <strong>intake</strong> and subsequent changes in body mass. <strong>The</strong> two variables were<br />
significantly correlated (r = 0.89, r 2 = 0.79, n = 36, F(1.35)= 89.79; p < 0.0001, see<br />
Figure 2), with energy <strong>intake</strong> explaining 79% of the observed variance in mass gain.<br />
Food <strong>intake</strong> was not related to female SVL (r = 0.22; n = 36; F(1.35)= 1.76; p = 0.20).<br />
Changes in body mass (g)<br />
40<br />
30<br />
20<br />
10<br />
0<br />
-10<br />
r 2 = 0.79, n = 36, p < 0.0001<br />
-20<br />
10 20 30 40 50 60 70 80<br />
Cumulative <strong>food</strong> <strong>intake</strong> (g)<br />
Figure 2. <strong>The</strong> <strong>relationship</strong> <strong>between</strong> <strong>total</strong> <strong>food</strong> <strong>intake</strong> (g) of 36 female aspic vipers and change in body<br />
mass (g) during the experiment. Each point represents an individual snake. <strong>The</strong> two treatment groups<br />
were pooled for this analysis.<br />
<strong>The</strong> difference in average <strong>food</strong> consumption <strong>between</strong> the two treatments generated<br />
significant differences in the amount of mass gained by females. <strong>The</strong> increment in<br />
mass midway through vitellogenesis was greater for the high-<strong>intake</strong> group (ANCOVA<br />
using mass gain as dependent variable, treatment as factor and initial body mass as<br />
covariate, F(1.34)= 4.29; p = 0.046). <strong>The</strong> magnitude of this difference was enhanced<br />
by the time of ovulation (F(1.33)= 8.00; p = 0.007).<br />
91
<strong>The</strong> influence of feeding treatment on change in maternal body mass was examined<br />
with a repeated measure ANOVA. Using SVL-adjusted body mass as the dependent<br />
variable, treatment as factor and the three consecutive records of masses as<br />
repeated measures, revealed a significant interaction (Wilk’s lambda = 0.75; F(3.29)=<br />
3.18; p = 0.038). <strong>The</strong> mass gain was significantly more marked in the high-<strong>intake</strong><br />
group (treatment effect, F (2.62)= 5.77; p < 0.005; see Figure 3).<br />
SVL- adjusted body mass<br />
125<br />
120<br />
115<br />
110<br />
105<br />
100<br />
Early<br />
vitellogenesis<br />
Mid<br />
vitellogenesis<br />
high <strong>food</strong> <strong>intake</strong><br />
low <strong>food</strong> <strong>intake</strong><br />
Ovulation<br />
Figure 3. Influence of experimental treatment (diet) on changes in body mass of female vipers at three<br />
times during the experiment. Black circle, continuous line: high <strong>food</strong> <strong>intake</strong> group; black triangle,<br />
dotted line: low <strong>food</strong> <strong>intake</strong> group.<br />
Determinants of fecundity<br />
<strong>The</strong> number of ova palpated in mid-vitellogenesis was greater in the high-<strong>intake</strong><br />
group than in the low-<strong>intake</strong> group (one-way ANCOVA with number of palpated ova<br />
as the dependent variable, treatment as the factor and SVL as the covariate;<br />
92
F(1.34)= 9.04; p < 0.005). <strong>The</strong> divergence was even greater at ovulation (F(1.33)=<br />
10.68; p < 0.0025, see Figure 4). <strong>The</strong> SVL-adjusted numbers of palpated ova were<br />
respectively 7.1 ± 2.46 (high-<strong>intake</strong> group) and 5.14 ± 1.89 (low-<strong>intake</strong> group) in mid-<br />
vitellogenesis and 6.67 ± 1.82 and 5.03 ± 1.47 at the time of ovulation.<br />
number of ovulated eggs<br />
10<br />
8<br />
6<br />
4<br />
2<br />
0<br />
non restricted<br />
p < 0.0025<br />
Treatment<br />
restricted<br />
Figure 4. Effects of experimental manipulation of <strong>food</strong> <strong>intake</strong> on the number of ova ovulated by aspic<br />
vipers. Error bar represent the standard deviation.<br />
A regression analysis pooling data from the two treatments confirmed that<br />
<strong>food</strong> <strong>intake</strong> during vitellogenesis influenced the number of ova at ovulation (r = 0.42,<br />
r 2 = 0.17, n = 36, p = 0.01). In addition to <strong>food</strong> <strong>intake</strong>, female SVL and early body<br />
condition are known to influence fecundity in aspic vipers (Bonnet et al. 2001b). We<br />
thus included all three of these variables in a multiple regression analysis. As<br />
predicted, the highest proportion of fecundity variation in our data set was explained<br />
by a model including all three predictor variables (r 2 = 0.41, see Table 1). When<br />
univariate regression analyses were conducted separately on data from the two<br />
93
treatment groups, however, strong differences were evident. Early body condition<br />
significantly affected fecundity in the high-<strong>intake</strong> females (r = 0.71, r 2 = 0.50, n = 17,<br />
p < 0.02) but not in the low-<strong>intake</strong> group (r = 0.34, n = 19, p = 0.56). Similarly,<br />
fecundity increased with maternal SVL in the high-<strong>intake</strong> animals (r = 0.56, n = 17, p<br />
= 0.018), but not in the low-<strong>intake</strong> group (r = 0.33, r 2 = 0.31, n = 19, p = 0.16). As a<br />
consequence, the combination of these two factors in a multiple regression explained<br />
a significant fraction of fecundity variation among the high-<strong>intake</strong> females (r = 0.69, r 2<br />
= 0.46, n = 17, p = 0.01), but not among the low-<strong>intake</strong> snakes (r = 0.34, r 2 = 0.11 n<br />
= 19, p = 0.56).<br />
Table 1. <strong>The</strong> effects of maternal body size (SVL), <strong>food</strong> <strong>intake</strong> during vitellogenesis (FOOD), and<br />
maternal body condition (BC) prior to vitellogenesis (residual scores from log-transformed mass<br />
versus SVL) on litter size of female aspic vipers. Results shown are from a multiple regression carried<br />
out on the entire data set (i.e., including female vipers from both treatment groups). <strong>The</strong> highest<br />
proportion of fecundity variance was explained by a model that included all three independent<br />
variables.<br />
Multiple Regression r = 0.64; r² = 0.41; n = 36; F(3.32)= 7.3265; p = 0.00071<br />
BETA Partial correlation Semi partial p-value<br />
SVL 0.36 0.36 0.42 0.012<br />
FOOD 0.39 0.38 0.44 0.008<br />
BC 0.30 0.30 0.36 0.033<br />
94
Discussion<br />
<strong>The</strong> experimental data from this study strongly support the conclusions of a recent<br />
descriptive study by Bonnet et al. (2001b) on free-ranging snakes. That is, the<br />
number of ova (litter size) produced by a female aspic viper is influenced not only by<br />
her body condition at the beginning of the year in which she will reproduce, but also<br />
by the amount of <strong>food</strong> that she consumes during the period of vitellogenesis. <strong>The</strong><br />
body size of the female snake also plays a role in determining fecundity, probably<br />
because a larger female has more abdominal space in which to accommodate the<br />
litter (Saint Girons 1957a). Thus, litter size in a female V. aspis is affected in a<br />
complex way by her body size, her pre-existing energy stores, and her <strong>food</strong> <strong>intake</strong> in<br />
the weeks immediately prior to ovulation.<br />
Experiments on the influence of energy input on reproductive output have<br />
provided valuable information in many vertebrate species (Arcese & Smith 1988,<br />
Bolton et al. 1993, Monaghan, P. & Nager. R.N. 1997 and references therein)<br />
including snakes (Ford & Seigel 1989b, 1994, Seigel & Ford 1991, 1992, 2001,<br />
Gregory & Skebo 1998). However, in a capital breeding species, the levels of pre-<br />
vitellogenic maternal reserves largely determine reproductive output (Bonnet et al.<br />
2001b). <strong>The</strong> possible effects of additional income energy (i.e. through manipulation<br />
of the diet) should be framed within this context. Unfortunately, quantitative data on<br />
the processes involved in of mobilisation of body reserves for follicular growth are<br />
available in only one snake species: Vipera aspis. Both empirical and experimental<br />
works on this species have shown that a peak in 17- β oestradiaol level triggers body<br />
reserve mobilisation and yolk deposition (Bonnet et al. 1994, Bonnet 1996). In the<br />
absence of equivalent data on the hormonal control of vitellogenesis in other species,<br />
interspecific comparisons would be premature.<br />
95
<strong>The</strong> female's "decision" as to whether or not to reproduce appears to be<br />
determined primarily by her initial body condition, making V. aspis a typical "capital<br />
breeder" in this respect (Naulleau & Bonnet 1996). However, the two "cancellations"<br />
(females that initiated but did not maintain vitellogenesis) suggest that <strong>food</strong> <strong>intake</strong><br />
during vitellogenesis might also play a role in this early "decision". That is, a female<br />
close to the energy-store threshold for reproduction may begin vitellogenesis, but<br />
abandon the process unless she obtains prey relatively soon. Resorption of follicles<br />
is probably adaptive, enabling the animal to recover resources if reproduction does<br />
not proceed (Blackburn 1998).<br />
Our experimental manipulation of <strong>food</strong> <strong>intake</strong> significantly modified not only<br />
the amount of maternal mass that was gained, but also the number of ova that were<br />
ovulated (and hence, litter size). Female body size and initial body condition also<br />
affected fecundity in aspic vipers, as they do in other species of snakes (Seigel &<br />
Ford 1987). Pooling data from the two treatment groups allowed us to detect<br />
significant influences of maternal SVL, fat stores and <strong>food</strong> <strong>intake</strong> on fecundity.<br />
However, conducting the analyses separately revealed that the effects of body<br />
reserves and body length were only statistically significant in the high-<strong>intake</strong> group.<br />
<strong>The</strong> differing importance of body length as an influence on fecundity may reflect the<br />
fact that <strong>total</strong> abdominal space (and thus, female body length) became a significant<br />
constraint on fecundity in the high-<strong>intake</strong> snakes. If so, we would expect to see<br />
larger litters in larger females than in smaller animals. In contrast, because snakes in<br />
the low-<strong>intake</strong> treatment group produced small litters fitting easily within the females'<br />
bodies, maternal body size was not a constraint (nor a correlate) of the variation in<br />
fecundity among these animals.<br />
<strong>The</strong> differing role of pre-existing energy reserves is less easy to explain, but<br />
may simply reflect the fact that the variation in <strong>food</strong> <strong>intake</strong> (and thus, the number of<br />
96
developing ova) among the high-<strong>intake</strong> females was higher than in the low-<strong>intake</strong><br />
snakes. Data collected by Saint Girons & Naulleau (1981) demonstrate that in female<br />
aspic vipers the mass of abdominal fat bodies is correlated with the number of<br />
growing follicles. Our results show that, after this initial phase, an exogenous source<br />
of energy will affect the number of ovulated eggs. <strong>The</strong> greater variation in <strong>food</strong> <strong>intake</strong><br />
and reproductive traits might explain why we detected correlates of reproductive<br />
output more easily in the high-<strong>intake</strong> group than in the low-<strong>intake</strong> snakes.<br />
Our study confirms that V. aspis is a typical capital breeder in some respects,<br />
notably in the "decision" whether or not to reproduce. Thus, capital stores (which<br />
reflect energy <strong>intake</strong> over long time periods) will ultimately determine reproductive<br />
frequencies in aspic vipers. In contrast, litter sizes will be determined not only by<br />
those pre-existing stores, but also by the female's foraging success in the weeks<br />
immediately prior to ovulation (Bonnet et al. 2001b). This flexibility in energy<br />
allocation enable the animals to adjust reproductive investment relative to local<br />
resource levels and is consistent with field data (Lourdais et al. 2002b). Capital<br />
breeding is widespread among ectotherms (Doughty & Shine 1997; Bonnet et al.<br />
1998) and may be particularly advantageous in situations of strong inter-annual<br />
fluctuations in <strong>food</strong> availability (Calow 1979). Vipera aspis is a sit-and-wait predator<br />
which feeds mainly on rodents, especially voles (Microtus arvalis) that show dramatic<br />
fluctuations in population density (Delatre et al. 1992, Lourdais et al. 2002b). In a<br />
situation where prey densities are unpredictable from one year to the next, a female<br />
aspic viper directly benefit from being able to:<br />
(1) reproduce successfully without having to depend upon <strong>food</strong> <strong>intake</strong> during the<br />
reproductive year. In a year when prey is scarce, a female relying upon "income"<br />
might fail to breed successfully, and either waste resources already invested or<br />
97
threaten her own survival. "Capital breeding" provides a mechanism for a temporal<br />
dissociation <strong>between</strong> feeding and breeding.<br />
(2) modify her reproductive output in a flexible fashion depending upon current levels<br />
of prey availability. Essentially, this flexibility allows the female to manipulate energy<br />
investment into reproduction and hence take advantage of "good" years by producing<br />
more offspring than would have been expected from the long-term average levels of<br />
prey availability in her habitat. Thus, although many aspects of reproduction in aspic<br />
vipers are driven by "capital" stores, some degree of reliance on “income” may help<br />
to fine-tune reproductive expenditure to <strong>food</strong> <strong>intake</strong> during the critical phase of egg<br />
production.<br />
Acknowledgements<br />
We thank Gwénael Beauplet for helpful comments on the manuscript. Financial<br />
support was provided by the Conseil Régional de Poitou-Charentes, Conseil Général<br />
des Deux Sèvres, the Centre National de la Recherche Scientifique (France).<br />
Manuscript preparation was supported by the Australian Research Council. We<br />
thank Pr. Oumid Popoye and Mr Celestin Le khobrato for helping in snake collecting<br />
and the construction of the electrical fence to impede feral cat predation. Finally, Mr<br />
Dujardin helped to solve many technical problems.<br />
98
D. Article 3<br />
Capital-breeding and reproductive effort in a<br />
variable environment: a longitudinal study of a<br />
viviparous snake<br />
Olivier Lourdais 1 2 3 , Xavier Bonnet 1 , Richard Shine 4 , Dale DeNardo 5 , Guy<br />
Naulleau 1 and Michael Guillon 1<br />
1 Centre d'Etudes Biologiques de Chizé, <strong>CNRS</strong>, 79360, Villiers en Bois, France<br />
2 Conseil Général Des Deux Sèvres, Rue de L’abreuvoir, 79021, Niort, France<br />
3 University of Poitiers, 40 avenue du recteur Pineau, 86022 Poitiers, France<br />
4 Biological Sciences A08, University of Sydney, NSW 2006, Australia<br />
5 Department of Biology, Arizona State University, Tempe, AZ, 85287-1501,USA<br />
Published in Journal of Animal Ecology 71 :470-479<br />
(2002)<br />
99
Summary<br />
1. We examined the ways that fluctuations in prey abundance and weather<br />
conditions can affect reproductive output in a "capital breeding" ectotherm, the aspic<br />
viper (Vipera aspis).<br />
2. Our longitudinal study confirms that female aspic vipers adjust reproductive<br />
investment by integrating allocations of energy from stores (“capital”) and facultative<br />
feeding (“income”). Thus, long-term energy storage enabled females to reproduce<br />
successfully even in years when prey were scarce.<br />
3. Not surprisingly, temporal changes in body reserves of female vipers preparing for<br />
reproduction depended upon current feeding rates. However, the mean<br />
environmental temperature during the active season also affected mass gain.<br />
4. Allometric patterns suggest that reproductive output was limited by energy<br />
availability in 8 out of the 9 years of our study. In the other year, high prey availability<br />
in the preceding season meant that reproductive output was maximised within the<br />
constraints set by maternal body size (and thus, abdominal volume).<br />
5. High summer temperatures increased basking opportunities of gravid vipers and<br />
thus accelerated gestation. However, maternal metabolic costs also increased in<br />
such situations, resulting in low post-partum body condition.<br />
Key words: ectothermy, capital breeding, environmental fluctuations, temperature,<br />
<strong>food</strong> availability<br />
100
Introduction<br />
Variation in reproductive success is a central theme in evolutionary biology.<br />
<strong>The</strong>oretical models predict that variation in quality among individuals provides the<br />
basic substrate for natural selection, resulting in a differential contribution to the<br />
number of descendants produced. Intrinsic sources of variation (including genetic<br />
effects), however, are only one source of phenotypic variation. Proximate<br />
(environmental) factors also affect reproductive performance. In many biological<br />
systems, year-to-year variation in environmental characteristics such as <strong>food</strong> supply<br />
or weather conditions can have dramatic repercussions on reproductive traits such as<br />
clutch size or reproductive frequency. Such influences have been documented in a<br />
variety of taxa (Lack 1954; Ballinger 1977, 1983; Brand & Keith 1979; Todd, Keith &<br />
Fisher 1981; Seigel & Fitch 1985).<br />
To cope with limitations and/or fluctuations of <strong>food</strong> resources, organisms have<br />
evolved a wide range of strategies for energy acquisition and allocation to<br />
reproduction. One fundamental dichotomy is <strong>between</strong> species in which reproduction<br />
is fuelled by recently acquired energy ("income breeders") and species where<br />
storage constitutes the primary energy source for reproduction ("capital breeders",<br />
Drent & Daan 1980). For income breeders, reproductive output should be closely<br />
linked to current resource availability, while in capital breeders a temporal separation<br />
should exist <strong>between</strong> the phase of energy acquisition and investment in reproduction.<br />
Capital breeding may be especially advantageous to buffer resource fluctuations<br />
among years, or if annual <strong>food</strong> levels are stable but insufficient to permit successful<br />
reproduction (Calow 1979). However, the acquisition and storage of large amounts<br />
of energy requires time and is also potentially costly (Jönsson 1997). <strong>The</strong>refore, the<br />
costs and benefits of alternative tactics of resource use (i.e., capital versus income<br />
101
eeding) will vary among species. For instance, many features associated with<br />
ectothermy pre-adapt organisms to store large reserves and to use them for<br />
reproduction (Bonnet, Bradshaw & Shine 1998). <strong>The</strong> duration of energy gathering<br />
may sometimes cover long periods (years) and therefore often results in a low<br />
frequency of reproduction (Bull & Shine 1979).<br />
In vertebrates, capital breeding systems coupled with infrequent (less-than-<br />
annual) reproduction have been observed in many reptiles (Saint Girons & Naulleau<br />
1981; Brown 1991; Brana, Gonzales & Barahona 1992; Doughty & Shine 1997;<br />
Madsen & Shine 1999). Some of the best examples are among viperid snakes<br />
(Madsen & Shine 1992a; Brown 1993; Martin 1993), with some species showing very<br />
low reproductive rates. For example, female aspic vipers (Vipera aspis Linné) in<br />
western France do not initiate vitellogenesis until they have accumulated sufficient<br />
energy stores to exceed a high body condition threshold (Naulleau & Bonnet 1996).<br />
<strong>The</strong> time necessary to accumulate body reserves entails a delayed maturity (2.5 to<br />
3.5 years of age with an average lifespan of 5 years; Bonnet et al. 1999a;<br />
unpublished data) and a low breeding frequency (once every 2 to 3 years). Due to<br />
high costs of reproduction, most females will produce only one or two litters within<br />
their lifetimes (Bonnet et al. 2000a; 2002a), and the same may be true for many<br />
temperate-zone viperid species (Madsen & Shine 1992a; Brown 1993).<br />
In west-central France, the habitat of the aspic viper is characterised by strong<br />
annual fluctuations in availability of rodents that are the snakes' main prey (Naulleau<br />
1997). <strong>The</strong>rmal conditions also vary significantly among years in this area. For an<br />
ectothermic species, variations in the thermal environment may affect the rate of<br />
processes such as digestion, metabolism and reproduction (Huey 1982; Naulleau<br />
1983a, b). In the present paper, we use data from a longitudinal study of a<br />
population of vipers, to examine how annual variation in both <strong>food</strong> availability and<br />
102
temperature affect reproductive output in a capital breeding ectotherm species.<br />
Specifically, we predicted that:<br />
<strong>The</strong> long duration of energy gathering prior to reproduction combined with the<br />
female's ability to store large amounts of energy within her body should result in a<br />
high investment per reproductive bout. Nonetheless, because "income" is also<br />
allocated to reproduction (Bonnet et al. 2001b), we expect that litter mass will be<br />
influenced by prey abundance in the year preceding reproduction as well as the<br />
current year.<br />
Capital breeding coupled with a fixed body condition threshold (Naulleau & Bonnet<br />
1996) means that all females initiating reproduction do so with substantial energy<br />
reserves and hence, the success of the litter should not be compromised by an<br />
unanticipated energy shortage. <strong>The</strong> proportion of healthy offspring versus non-viable<br />
components in a litter should be high, and independent of resource fluctuations.<br />
Among non-reproducing females (i.e., individuals preparing for reproduction), body<br />
reserves accumulated at the end of the activity period should depend upon current<br />
<strong>food</strong> levels and also be influenced by thermal conditions that determine digestion<br />
rate.<br />
<strong>The</strong>rmal conditions should directly affect the rate of embryogenesis and thereby<br />
gestation length.<br />
Material and methods<br />
Study Animals<br />
<strong>The</strong> aspic viper (Vipera aspis Linné) is a small viviparous snake of the western-<br />
Paleartic region and is locally abundant at the northern limit of its distribution in<br />
France. Females mature at 40 cm snout-vent length (SVL), which is attained in 2.5<br />
103
to 3.5 years (Bonnet et al. 1999a). In this area, females typically reproduce at a less-<br />
than-annual frequency (Saint Girons 1957a,b; Bonnet & Naulleau 1996; Naulleau &<br />
Bonnet 1996; Naulleau et al. 1999). Ovulation typically occurs during the first two<br />
weeks of June with limited geographical or altitudinal variations (Saint Girons 1957b,<br />
1973; Naulleau 1981). Parturition occurs two to three months later, from late August<br />
until late September (Bonnet et al. 2001b).<br />
Study site and methods<br />
<strong>The</strong> study site is near the village of Les Moutiers en Retz in west-central France<br />
(47 o 03N'; 02 o 00W'). It is a 33-hectare grove with a mosaic of meadows and<br />
regenerating scrubland. <strong>The</strong> site is characterised by a temperate oceanic climate.<br />
From 1992 to 2000, one to four people patrolled the site using a standardised<br />
searching method on almost every favourable day during the vipers’ annual activity<br />
period (late February to late October). <strong>The</strong> <strong>total</strong> searching effort exceeded 4,000<br />
hours. Snakes were caught by hand, sexed by eversion of the hemipenes, weighed<br />
to the nearest g with an electronic scale, and measured (SVL and <strong>total</strong> length) to the<br />
nearest 5 mm. Over 400 adult and sub-adult female vipers were marked using<br />
passive integrated transponder (PIT) tags (Sterile transponder TX1400L, Rhône<br />
Mérieux, 69002 LYON France, product of Destron/IDI Inc). Each snake was then<br />
released at its exact place of capture. Because the study site is surrounded by<br />
habitat unsuitable for vipers (Vacher-Vallas, Bonnet & Naulleau 1999) and this<br />
species is highly philopatric (Naulleau, Bonnet & Duret 1996), any snake not<br />
captured over a long period (> 2 years) had almost certainly died rather than<br />
emigrated or avoided capture.<br />
104
Initial <strong>total</strong> body length and body mass were measured in March-April (before<br />
any significant <strong>food</strong> <strong>intake</strong>). Changes in body mass and body length were calculated<br />
from March-April to August-November within a given year and from March-April to the<br />
next March-April <strong>between</strong> years. Since our data indicate that vipers did not show any<br />
significant change in either body mass or body length over hibernation, we excluded<br />
that time period from our calculations. To provide an index of body condition (mass<br />
relative to length), we calculated residual values from the regression of log-<br />
transformed body mass against log-transformed body length (Jayne & Bennett 1990).<br />
Initial body condition (calculated in March-April) provides an accurate indicator of the<br />
level of fat stores in female vipers (Bonnet 1996).<br />
Reproductive status was determined using two methods. First, any female<br />
whose initial body condition value was greater than the threshold at this time (March-<br />
April) was considered reproductive (see Bonnet, Naulleau & Mauget. 1994; Naulleau<br />
& Bonnet 1996). Second, from mid-vitellogenesis (May) to the end of gestation (late<br />
August) reproductive status was easily determined either by palpation of ova and/or<br />
embryos or by records of parturition (Fitch 1987; Naulleau & Bonnet 1996). Gravid<br />
females were captured and maintained in captivity after the first parturition of the year<br />
was recorded (generally in late August). For each year, mean annual changes in<br />
maternal mass prior to parturition were calculated as pre-partum body mass minus<br />
initial body mass. Date of capture of pregnant females had no significant effect on<br />
this parameter for two main reasons. First, all pregnant females were re-captured<br />
over a short period (one to two weeks) at the end of gestation, by which time there<br />
was little further mass change before parturition. Second, during the active season,<br />
<strong>food</strong> <strong>intake</strong> occurs mainly during vitellogenesis in spring and almost cases after<br />
ovulation (Saint Girons & Naulleau 1981, Bonnet et al. 2000a; Lourdais, Bonnet &<br />
105
Doughty 2002a). Overall, then, pre-partum mass change provides a robust indicator<br />
of <strong>food</strong> <strong>intake</strong> during reproduction.<br />
Pregnant females were maintained in separate cages in the laboratory until they<br />
gave birth one to 21 days later. Mass was recorded every two days and immediately<br />
after parturition. For each female, we calculated pre-partum and post-partum body<br />
condition. Body condition of pre-partum females includes both the litter mass and<br />
female carcass mass, whereas post-partum body condition offers an indicator of the<br />
degree of female emaciation (Bonnet et al. 2000a).<br />
Reproductive data were obtained on 173 litters from 149 different females. For<br />
most individuals (129) only a single litter was obtained, but 16 females produced two<br />
litters and 4 individuals produced three litters. Treating these successive litters by<br />
the same female as independent data may introduce problems with<br />
pseudoreplication. However, none of our conclusions from statistical tests differed<br />
depending on whether these ‘repeat” litters were included or excluded. Thus, the<br />
following analysis presents calculations based on the <strong>total</strong> data set. <strong>The</strong> components<br />
of the litter were characterised (yolked eggs, dead offspring, healthy offspring),<br />
counted, and weighed (± 0.1g). Additionally, young and stillborn were measured (±<br />
0.5 cm) and sexed. For one individual, <strong>total</strong> litter mass was not available.<br />
We could not distinguish unfertilised ova from ova that had been fertilised but had<br />
died early in embryogenesis. Hence, eggs where only yolk was visible (fertilised or<br />
not), underdeveloped embryos and stillborn were all grouped in the same “non-<br />
viable” category. Litter mass and litter size included all of these elements along with<br />
healthy offspring, whereas "fit litter mass" and "fit litter size" included healthy<br />
offspring only (Gregory, Larsen & Farr 1992). Relative litter mass (RLM) was defined<br />
as the residual score from the general linear regression of litter mass against post-<br />
106
parturient mass of the mother. Gestation periods were calculated from parturition<br />
dates, assuming a fixed ovulation date in mid June (Naulleau 1981).<br />
Environmental factors<br />
Food levels - Vipers feed mainly on voles (Microtus arvalis Pallas) whose populations<br />
typically fluctuate in a three to four year cycle (Delatre et al. 1992). Variations in vole<br />
abundance directly influence rates of feeding by the snakes, as revealed by the<br />
proportion of snakes captured with a prey item in the stomach (Bonnet et al. 2000a;<br />
2001b). Data on the annual proportions of adult snakes containing prey at the time<br />
of capture were available from 1993 to 2000. In the following analysis we used this<br />
parameter as an index of <strong>food</strong> abundance. This estimator of <strong>food</strong> levels provides<br />
results that are consistent with those from line trapping; for example, the same<br />
annual peaks in vole density are detected by both methods (Salamolard et al. 2000;<br />
Bonnet et al. 2000a). However, feeding rates of the snakes provide a more sensitive<br />
measure of prey abundance, because trapping can fail to detect voles at low<br />
population densities. We acknowledge that feeding rates may also be sensitive to<br />
environmental temperature (due to thermal effects on snake activity and foraging<br />
success) and hence, our measures of annual variation in temperatures and in <strong>food</strong><br />
supply would not be independent. However, we found no significant correlation<br />
<strong>between</strong> mean annual temperature and the proportion of snakes captured with a prey<br />
item in the stomach (see below). Because the aspic viper is a very sedentary animal<br />
occurring at a high density in our population, feeding rates are likely to be tightly<br />
linked to prey abundance.<br />
Temperatures - the study site is near the northern limit of the species' range and,<br />
therefore, climatic conditions may constrain the animals' reproductive biology. This<br />
inference is supported by the fact that parturition in southern populations occurs in<br />
107
mid July, two months earlier than in West-central France (Zuffi, Giudici & Iolae,<br />
1999). Throughout the nine years of our study, daily thermal maxima (°C) under<br />
shelter were recorded. Two biological periods were considered: the active season<br />
(March to October) and the gestation period (mid June to August).<br />
Results<br />
Fluctuation in environmental factors<br />
Food levels - feeding rates varied markedly during the course of our study (χ ² =<br />
25.53, df =7, p < 0.0006). Two years were distinguishable, with one year of very low<br />
(1994) and one of particularly high (1996) <strong>food</strong> levels (see Figure 1).<br />
Food index<br />
0.5<br />
0.4<br />
0.3<br />
0.2<br />
0.1<br />
0.0<br />
1992 1993 1994 1995 1996 1997 1998 1999 2000<br />
Year<br />
Figure 1. Annual variation in the proportion of aspic vipers containing prey items in the stomach at the<br />
time of capture. Analyses in this paper use this proportion as an index of availability of prey for the<br />
snakes. Most of these prey were voles (Microtus arvalis). See text for details on the method.<br />
108
Temperature - during the course of the study, mean daily temperature during the<br />
snakes' active season (and during the gestation period) varied significantly among<br />
years (ANOVA, F(8, 2196) = 3.88; p < 0.0001; F(8, 819) = 2.89; p < 0.0035,<br />
respectively, see Fig 2). We found no significant <strong>relationship</strong> <strong>between</strong> feeding rates<br />
and mean temperature during the active season (r = 0.27, n = 8, F(1, 6) = 0.5, p =<br />
0.50).<br />
Mean temperatures (active season)<br />
25<br />
24<br />
23<br />
22<br />
21<br />
20<br />
19<br />
1992 1993 1994 1995 1996 1997 1998 1999 2000<br />
Year<br />
Active season<br />
Gestation<br />
Figure 2. Annual variation in thermal conditions over the course of our study. Mean ambient<br />
temperatures (± S.E.) under cover items were calculated separately for the active season (March to<br />
October; black circles, continuous line) and for the gestation period (Mid-June to August; black<br />
triangles, dashed line).<br />
Influences on viper reproduction<br />
1 - Annual fluctuations in <strong>food</strong> levels and reproductive output<br />
a) Variation in litter mass and reproductive effort<br />
Reproductive investment was generally high (mean litter mass = 34 ± 15g),<br />
representing on average 52% of female post-partum body mass. Mean litter mass<br />
109<br />
28<br />
27<br />
26<br />
25<br />
24<br />
23<br />
22<br />
21<br />
20<br />
Mean temperatures (gestation)
showed significant annual fluctuations (ANCOVA, F(8, 160) = 2.26, p < 0.02; using<br />
litter mass as the dependent variable, female body length as a co-factor). Among the<br />
173 litters obtained, 15 individuals produced entirely non-viable litters (e.g., only<br />
undeveloped ova and still-born offspring) and thereby had very low litter mass values.<br />
Even when these non-viable litters were excluded from analyses, mean litter mass<br />
displayed significant annual variation (ANCOVA, F(7, 142) = 3.76, p < 0.0009, Figure<br />
3). <strong>The</strong> variation was mainly due to two consecutive years of especially high litter<br />
mass values: 1996 (a high <strong>food</strong> year) and 1997 (an intermediate <strong>food</strong> year).<br />
Size adjusted <strong>total</strong> litter mass (g)<br />
55<br />
50<br />
45<br />
40<br />
35<br />
30<br />
25<br />
5<br />
0<br />
1992 1993 1994 1995 1996 1997 1998 1999 2000<br />
Year<br />
Figure 3. Annual fluctuations in mean <strong>total</strong> litter mass (± S.E.) of female aspic vipers. Values have<br />
been scaled with female body length. Fifteen litters that were entirely non-viable were excluded from<br />
this analysis (see text for statistics).<br />
Substantial year-to-year variations in maternal characteristics were also<br />
detected (see Figure 4). Mean initial body condition of reproducing females varied<br />
110
significantly (ANOVA, F(8, 257) = 3.53, p < 0.0007) with the highest values observed<br />
in 1997 (after a year of high <strong>food</strong> abundance, see Figure 2). Pre-partum body<br />
condition also showed significant variation (ANOVA, F(8, 160) = 6.27, p
91% of the variance in mean annual litter mass (r = 0.95, n = 8, F(2, 4) = 19.74, p <<br />
0.009; Table 1).<br />
Table 1. Effects of <strong>food</strong> level in the year of (n) and the year prior to (n-1) reproduction on absolute litter<br />
mass of aspic vipers in south-western France.<br />
Multiple Regression r = 0.95; r² = 0.91; n = 7; F(2, 4) =19.745; p = 0.008<br />
Bêta Partial correlation Semi-partial p value<br />
Food year n 0.83 0.94 0.82 0.005<br />
Food year n-1 0.57 0.88 0.57 0.019<br />
b) Influence of maternal body size<br />
Because maternal body size is highly correlated with litter size or litter mass in many<br />
snakes (Seigel & Ford 1987), we looked for correlations <strong>between</strong> maternal length and<br />
reproductive output (litter size and litter mass) in our population. Combining data for<br />
all years, we detected a significant but weak influence of maternal body size on<br />
reproductive output (r = 0.20, n = 173, F(1,171) = 7.1, p
Table 2. Correlation <strong>between</strong> maternal body length and reproductive output (litter size and litter mass)<br />
for each year of the study. <strong>The</strong> power (1-β) of the analysis and the sample size required for α
was high during the study. Fit litter mass averaged 86% of <strong>total</strong> litter mass and, not<br />
surprisingly, these two values were strongly correlated (r = 0.95, n = 171, F(1, 169) =<br />
1747.0, p < 0.0001). To avoid statistical problems associated with ratios (Seigel and<br />
Ford 1987), we calculated residual values from the regression of fit litter mass<br />
against litter mass to provide an index of reproductive “efficiency”. This adjusted<br />
measure of fit litter mass peaked in 1994 (ANOVA, F(8, 160) = 3.21, p < 0.0021, see<br />
Figure 5; considering only years where data on more than 10 litters were obtained,<br />
F(3, 118) = 5.94; p < 0.0008).<br />
Adjusted fit litter mass (g)<br />
40<br />
35<br />
30<br />
25<br />
20<br />
5<br />
0<br />
1992 1993 1994 1995 1996 1997 1998 1999 2000<br />
Year<br />
Figure 5. Annual fluctuations in our measure of reproductive “efficiency” based on the regression of fit<br />
litter mass against litter mass for aspic vipers. For simplicity, the graph shows fit litter mass values<br />
adjusted for <strong>total</strong> litter mass (± S.E.).<br />
Paradoxically, reproductive “efficiency” was highest in 1994 (i.e., during a year of<br />
<strong>food</strong> scarcity) and lowest in 1996 (high <strong>food</strong> year). Considering fit litter size rather<br />
than fit litter mass did not change those results. This annual variation mostly reflected<br />
changes in the proportions of litters with at least one non-viable component (χ² =<br />
114
9.92; dl = 3; p = 0.02) rather than annual variation in proportional viability among<br />
litters with at least one non-viable component (ANOVA, F(4, 67) = 0.64, p = 0.64).<br />
Females with low reproductive success (i.e., > 60% of the litter non-viable) did not<br />
have decreased survival (χ² = 0.13, dl = 1, n = 154, p = 0.71), but they did have<br />
longer gestation periods (ANOVA, F(1, 170) =14.83, p
Annual mass change (g)<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
1992 1993 1994 1995 1996 1997 1998 1999 2000<br />
Year<br />
Growth<br />
Mass change<br />
Figure 6. Fluctuations in annual growth rates (open hexagons, dashed line) and annual mass<br />
changes (open square, dotted line) for non-reproducing female V. aspis. Values were scaled with<br />
female initial (spring) body size. Error bars represent standard errors.<br />
b) Interaction <strong>between</strong> <strong>food</strong> levels and thermal conditions<br />
A multiple regression analysis revealed a significant combined influence of <strong>food</strong><br />
levels and mean temperature during the active season on annual mass gain of non-<br />
reproducing females. <strong>The</strong>se two factors explained 96% of the variance in mean<br />
annual mass gain (see Table 3). Interestingly, we were not able to detect any<br />
significant interaction <strong>between</strong> <strong>food</strong> levels and thermal conditions in the<br />
determination of annual rates of growth in body length (partial correlation: r = 0.14, n<br />
= 8, F(1, 6) = 0.5, p = 0.76).<br />
116<br />
8<br />
7<br />
6<br />
5<br />
4<br />
3<br />
2<br />
1<br />
0<br />
Annual growth (cm)
Table 3. Combined influences of mean ambient temperature during the active season (Mean temp)<br />
and <strong>food</strong> levels on adjusted annual mass change of non-reproducing female vipers.<br />
Multiple Regression r = 0.96; r² = 0.92; n = 8; F (2, 5) = 30.9; p < 0.0015<br />
Bêta Partial correlation Semi-partial p value<br />
Mean temp 0.56 0.89 0.54 0.007<br />
Food level 0.95 0.96 0.92 0.0007<br />
3 - Influence of thermal conditions during the gestation period<br />
a) Reproductive success<br />
Mean daily temperature during pregnancy did not affect the proportion of viable<br />
neonates (r = 0.08, n=173, F(1, 171) = 1.12, p< 0.29).<br />
b) Gestation length<br />
Mean daily temperature during pregnancy strongly affected the length of gestation (r<br />
=- 0.47, n = 173, F(1, 171) = 49.36, p < 0.00001, Figure 7).<br />
Gestation length (days)<br />
105<br />
100<br />
95<br />
90<br />
85<br />
80<br />
75<br />
23.0 23.5 24.0 24.5 25.0 25.5<br />
Mean temperature during gestation (°C)<br />
Figure 7. Influence of mean environmental temperature during the gestation period (summer) on the<br />
duration of gestation in free-ranging aspic vipers. For simplicity, the graph shows mean annual<br />
gestation length. See text for detailed statistical analyses of these data.<br />
117
Because low reproductive success (i.e., > 60% of litter non-viable) also affected<br />
gestation length, we re-analysed the data excluding these litters, thus enabling us to<br />
focus on the effect of temperature (r = - 0.61, n = 138, F(1, 135) = 83.01, p <<br />
0.00001). Mean temperatures during pregnancy explained 40% of the variance in<br />
gestation length.<br />
c) Maternal condition post-partum<br />
We found a negative correlation <strong>between</strong> mean summer temperature and mean post-<br />
partum body condition (r = 0.24, n = 172, F(1, 170) =10.7, p < 0.00125). Post-partum<br />
body condition in this species is also positively influenced by energy acquired during<br />
the year of reproduction (Bonnet et al. 2001b). We found a positive correlation<br />
<strong>between</strong> mean mass changes prior to parturition and post-partum body condition, (r =<br />
0.84, n = 8, F(1, 7) =17.56, p < 0.004). We reanalysed the data holding pre-partum<br />
body condition constant (i.e., an indirect measure of energy acquired during the year<br />
of reproduction). <strong>The</strong> partial correlation analysis supported our initial finding (Table<br />
4).<br />
Table 4. Combined influences of temperature during gestation (Ges temp) and <strong>food</strong> <strong>intake</strong> during the<br />
year of reproduction (indirectly measured using pre-partum body condition value) on the post-partum<br />
body condition of female aspic vipers.<br />
Multiple Regression r = 0.45; r² = 0.21; n = 169; F(2, 166) =22.05;
Discussion<br />
Previous studies on this population have shown that the reproductive “decisions” of<br />
female aspic vipers largely rely on stored energy reserves ("capital breeders":<br />
Naulleau & Bonnet 1996; Bonnet et al. 2001b). Demographic data clearly identify<br />
prey-driven variations in the proportion of reproducing females in this population, with<br />
<strong>food</strong> levels in a given year influencing the number of reproductive females the<br />
following year (Bonnet et al. 2001b; Lourdais et al. 2002d). In the present work, we<br />
have focussed on annual variation in reproductive output to further understand the<br />
system of energy acquisition and allocation in this species. As predicted, capital<br />
breeding allowed for a high reproductive investment regardless of <strong>food</strong> availability in<br />
the current year. However, reproductive investment varied among years, with mean<br />
litter mass higher in 1996 and 1997 than in other years (Figure 3). Reproducing<br />
females gained substantially in body condition during the year of highest prey<br />
availability (1996), reflecting high energy <strong>intake</strong> during that year. High feeding rates<br />
at this time also resulted in high initial body mass of reproducing females early in the<br />
following season (1997). That is, the high prey availability of a single year (1996)<br />
increased litter masses not only that year but also in the following year as well (by<br />
increasing energy storage of non-reproducing females).<br />
This complex system of energy allocation involving both capital breeding and<br />
facultative income breeding is well illustrated in the multiple regression analysis<br />
combining <strong>food</strong> levels both in the year of reproduction and in the preceding year<br />
(Table 1). A female viper’s reproductive strategy combines both “rigid” and “flexible”<br />
components. Firstly, female vipers have to reach a fixed body condition threshold to<br />
engage in reproduction (Naulleau & Bonnet 1996). A certain level of flexibility is<br />
evident however, with some pre-reproductive females eating rapidly enough to<br />
119
“overshoot” the body condition threshold, and hence accumulating body reserves<br />
above the fixed threshold (Naulleau & Bonnet 1996). <strong>The</strong>se “extra” reserves are<br />
invested into reproduction and they positively influence litter size (Bonnet et al.<br />
2001b). After this initial phase that determines both the female’s reproductive<br />
decision and the number of follicles she recruits, facultative income breeding enables<br />
her to adjust her reproductive effort to current <strong>food</strong> levels during vitellogenesis. This<br />
composite system of energy allocation is advantageous for at least two reasons.<br />
First, reliance upon stored reserves secures a high reproductive “efficiency”<br />
independently of current prey availability. <strong>The</strong> proportion of undeveloped ova or<br />
stillborn offspring produced was low and was not related to <strong>food</strong> levels. Second,<br />
instead of a completely “rigid” capital breeding system, facultative income breeding<br />
enabled individuals to adjust reproductive investment to local resource levels<br />
reproductive and notably to take advantage of occasional periods when prey are<br />
abundant. A similar flexibility may occur in the closely related adder, Vipera berus<br />
Linné (Andren & Nilson 1983).<br />
<strong>The</strong> effects of marked fluctuations in <strong>food</strong> levels were also evident in non-<br />
reproducing females (i.e., individuals preparing for reproduction). Among these<br />
females, change in body mass and growth in body length were both highly<br />
dependent on <strong>food</strong> levels. However, variance in mass gain was linked not only to<br />
feeding rates, but also to thermal conditions during the active season. This result<br />
may explain the high rates of mass gain observed in 1997, a year when climatic<br />
conditions were particularly favourable but prey abundance was only intermediate<br />
(confirmed by trapping, Salamolard et al. 2000). <strong>The</strong>rmal conditions may influence<br />
average mass change in several ways. Favourable thermal conditions may influence<br />
average mass change by accelerating digestion rates whereas low temperatures<br />
prolong digestion and may even stimulate regurgitation of the meal (Naulleau 1983a,<br />
120
1983b). While our data show no correlation <strong>between</strong> feeding rates and thermal<br />
conditions, we can not exclude the possibility of undetected complex interactions<br />
<strong>between</strong> concurrent fluctuations in thermal factors and prey availability, and therefore<br />
we encourage further study of this <strong>relationship</strong>.<br />
Several results from our study support an energy limitation model for<br />
reproduction in this population. First, elevated rates of body growth and mass gain<br />
were observed during a year of high <strong>food</strong> levels (1996). Second, maternal body size<br />
correlated only weakly and inconsistently with reproductive output (litter size and litter<br />
mass) in our aspic vipers. This correlation is high in most other snakes (Seigel &<br />
Ford 1987), perhaps because maternal abdominal volume (rather than energy<br />
supply) generally constrains reproductive output (Shine 1988). In a capital breeder,<br />
however, body stores may provide a greater constraint on reproductive output, and<br />
may often be below the level at which the litter mass would be constrained by (and<br />
hence, correlated with) maternal body size. Under this scenario, we would expect a<br />
positive influence of female body size on fecundity to be more easily detected when<br />
energy <strong>intake</strong> is sufficient to allow maximisation of body reserves. This prediction is<br />
consistent with our results: the correlation <strong>between</strong> maternal body size and<br />
reproductive output was significant only in 1997, the "best" year in terms of energy<br />
availability.<br />
While the system of energy allocation allows for adjustment to fluctuations in<br />
<strong>food</strong> availability, some aspects of reproduction were directly dependent upon<br />
variations in thermal conditions. High midsummer temperature accelerated<br />
gestation, as has been reported previously in captive snakes (Blanchard & Blanchard<br />
1941). Our data also suggest that summer temperatures influence "costs of<br />
reproduction" for females. <strong>The</strong> maintenance of a higher body temperature and thus<br />
higher metabolic rate (Saint Girons, Naulleau & Célérier 1985) during pregnancy<br />
121
translated into negative effects on female post-partum body condition. During the<br />
course of gestation, female aspic vipers are virtually anorexic and feed only<br />
opportunistically. Embryo maintenance generates substantial fecundity-independent<br />
mass loss in females and constitutes an important component of energy expenditure<br />
(Lourdais et al. 2002a). Our data suggest that the magnitude of this metabolic cost<br />
varied from year to year, depending upon thermal conditions.<br />
This result demonstrates difficulties associated with accurate estimation of<br />
reproductive effort in viviparous ectotherms. Classically, "reproductive effort" has<br />
been quantified using simple measures of reproductive output, such as the ratio of<br />
absolute litter mass to post-partum female body mass (Seigel & Fitch 1984).<br />
However, our results, in combination with related work (Bonnet 2001b, unpublished<br />
data) suggest that the post-partum body condition of a female aspic viper is affected<br />
in complex ways by several factors including her direct investment in the litter<br />
(absolute litter mass), her <strong>food</strong> <strong>intake</strong> during the year of reproduction, and her<br />
metabolic expenditure during gestation. Although all of these factors affect female<br />
emaciation (and thus both current and future reproductive effort), they are controlled<br />
by very different variables: (1) investment in the litter depends on energy stores<br />
combined with current <strong>food</strong> <strong>intake</strong>; (2) independent of reproductive investment, <strong>food</strong><br />
<strong>intake</strong> during the year of reproduction enhances female post-partum body condition;<br />
and (3) female emaciation is also influenced by climatic conditions. Complex<br />
interactions <strong>between</strong> the two varying environmental factors (prey abundance and<br />
thermal conditions) thus are likely to affect the degree of female emaciation.<br />
Integrative approaches will be needed to identify the different means by which energy<br />
is expended during reproduction in ectotherms.<br />
In conclusion, our long-term study clarifies the influence of two major<br />
environmental factors (<strong>food</strong> availability and thermal conditions) on the reproductive<br />
122
iology of female vipers. <strong>The</strong> snakes' responses to annual fluctuations in <strong>food</strong><br />
availability support the hypothesis that reproductive output in V. aspis is determined<br />
by a combination of capital breeding (i.e., the use of energy stores) and income<br />
breeding (i.e., the use of energy from current <strong>food</strong> <strong>intake</strong>). This mixed (capital plus<br />
income) system masks but does not eliminate a trend for correlated fluctuations in<br />
prey abundance and reproductive output. <strong>The</strong> reliance on stored energy also<br />
enables a high and relatively invariant “efficiency” of reproduction (proportion of<br />
viable embryos).<br />
Annual fluctuations in thermal conditions also entailed both direct and indirect<br />
repercussions on viper reproduction. Among females preparing for reproduction,<br />
thermal conditions, in combination with <strong>food</strong> levels, significantly influenced the<br />
acquisition of energy stores. Additionally, thermal conditions have at least two direct<br />
effects on reproducing females during gestation period. First, gestation length is<br />
influenced by ambient temperatures that determine the thermal regimes experienced<br />
by pregnant females (Naulleau 1986). Second, high midsummer temperatures not<br />
only shorten gestation, but they also increase maternal metabolism and thus<br />
decrease the female's post-partum body condition. Such a decrease may well<br />
translate into a lower probability of survival, or into a delay in production of the<br />
eventual next litter (Bonnet et al. 2002a).<br />
Acknowledgements<br />
We thank Gwenaël Beauplet for comments on the manuscript. Financial support<br />
was provided by the Conseil Régional de Poitou-Charentes, Conseil Général des<br />
Deux Sèvres, the Centre National de la Recherche Scientifique (France). Special<br />
thanks to Melle, notably for being her.<br />
123
III. Les coûts de la<br />
reproduction: amplitude et<br />
degré de dépendance avec<br />
PROBABOLITY OF SURVIVAL (%)<br />
100<br />
90<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
la fécondité<br />
Reproductive<br />
124<br />
Non<br />
Reproductive
A. Résumé du chapitre:<br />
Au cours de chaque épisode reproducteur, la vipère aspic femelle déploie un effort<br />
particulièrement élevé, tout au moins en comparaison avec les autres vertébrés<br />
amniotes. Ainsi, le rapport entre la masse de la portée et celle de la mère après la<br />
parturition est très élevé, parfois supérieur à 1. Un tel investissement énergétique<br />
traduit la mobilisation massive d’importants stocks de réserves corporelles parfois<br />
associée à l’effet de prises alimentaires facultatives. Les modèles théoriques<br />
prévoient que les relations entre l’effort de reproduction, le succès reproducteur et les<br />
coûts qui les accompagnent sont des éléments primordiaux pour comprendre<br />
l’évolution des traits d’histoire de vie. Dans ce chapitre, les principales composantes<br />
des coûts de la reproduction, mortalité, fréquence de reproduction, dégradation de la<br />
condition corporelle, ralentissement de la croissance, ainsi que leurs valeurs relatives<br />
sont examinés.<br />
Dans un premier temps, le suivi de la population des Moutiers nous a permis<br />
de mesurer l’influence d’une reproduction donnée sur les probabilités de<br />
reproduction futures. Cette étude révèle l’existence de coûts écologiques très élevés<br />
(article 4) avec une très forte mortalité chez les femelles reproductrices en<br />
comparaison avec les non-reproductrices. Cette mortalité réduit dramatiquement le<br />
nombre d’épisode reproducteur dans la vie de l’animal avec une majorité d’individus<br />
“semélipares”. Outre les coûts en survie, nos résultats soulignent plusieurs sources<br />
de coût énergétiques potentiels avec notamment un amaigrissement marqué et une<br />
croissance très réduite (souvent nulle) pendant l’année de reproduction. L’essentiel<br />
des réserves corporelles est investit dans la reproduction courante et une femelle<br />
bénéficiant d’un important stock initial investira d’autant plus d’énergie dans la<br />
reproduction. A la différence des coûts en survie qui se traduisent directement par<br />
125
une baisse du succès reproducteur à vie, les conséquences des coûts énergétiques<br />
potentiels sur le succès des reproductions futures (individus itéropares) sont plus<br />
difficiles à cerner et à quantifier. Cette situation reflète une forte hétérogénéité inter<br />
individuelle dans la cinétique de reconstitution des stocks. Ainsi, indépendamment<br />
de l’effort reproducteur, ce sont les femelles qui vont reconstituer rapidement leurs<br />
réserves l’année suivante qui vont bénéficier d’une meilleure survie et d’une plus<br />
grande probabilité de reproductions futures. Ce résultat souligne l’influence<br />
déterminante de la cinétique de la reconstitution des réserves sur la vie reproductrice<br />
des femelles.<br />
Dans un second temps, nous avons cherché à préciser la nature des<br />
contraintes énergétiques imposées par les activités de reproduction. Si la<br />
vitellogénèse constitue la phase clé de l’allocation énergétique pour la production<br />
des follicules, la gestation, qui peut être considérée comme une forme de soin<br />
parental prénatal , est la période capitale du développement embryonnaire. Comme<br />
la vitellogenèse, elle entraîne de profonds changements éco-éthologiques chez les<br />
femelles reproductrices. Le second article de ce chapitre (article 5) révèle d’ailleurs<br />
l’existence de coûts énergétiques élevés et très spécifiques de la gestation. Pendant<br />
cette phase, le déplacement des préférences thermique vers des températures<br />
élevées est associé à une perte de masse marquée qui résulte de l’épuisement des<br />
réserves corporelles résiduelles à la vitellogenèse. De façon remarquable, la<br />
dépense énergétique n’est pas influencée par le nombre de jeunes en<br />
développement. Cette indépendance de la fécondité suggère clairement une<br />
amplitude “ fixe” des contraintes métaboliques de la gestation, probablement parce<br />
que les températures optimales de développement des embryons ne changent pas<br />
quel que soit leur nombre. En effet, le déplacement des préférences thermiques<br />
pendant la gestation est lié au maintien d’un optimum physiologique pour le<br />
126
développement embryonnaire qui se manifeste sous la forme d’un palier. Pendant<br />
cette période, la thermorégulation des femelles maternelles est donc directement<br />
déterminée par le statut reproducteur et non le nombre de jeunes en développement.<br />
Nos observations de terrain et en captivité confirment les publications<br />
antérieures suggérant que les femelles gestantes réduisent fortement leur prise<br />
alimentaire. Toutefois, nos analyses révèlent que ces pertes en opportunités<br />
alimentaires sont indépendantes de la fécondité. La gestation génère donc des<br />
pertes d’opportunités énergétiques qui s’ajoutent aux dépenses métaboliques<br />
maternelles. En captivité, l’alimentation pendant cette période limite l’impact<br />
métabolique de la gestation et améliore l’état des femelles post-parturientes. Notre<br />
étude sur la consommation en oxygène pendant le gestation (article 6) vient<br />
confirmer l’impact du régime thermique maintenues par la femelle reproductrice par<br />
rapport aux non-reproductrices sur le métabolisme et indique à nouveau une<br />
influence très réduite de la fécondité sur la consommation en oxygène. Par exemple,<br />
l’effet du nombre d’embryons vivants sur la consommation d’oxygène n’est détecté<br />
qu’en fin de gestation, lorsque le développement des vipéreaux s’achève et leur<br />
consommation en oxygène devient détectable .<br />
Enfin, dans un dernier volet (article 7), nous avons cherché à identifier<br />
l’échelle temporelle sur laquelle se manifestent les coûts énergétiques de la<br />
reproduction. Notamment si certains coûts s’expriment pendant la reproduction<br />
(prédation par exemple), d’autres composantes, vont pouvoir intervenir avec un<br />
certain décalage temporel étendu. Un tel décalage peut être particulièrement<br />
important pour des espèces comme la vipère aspic chez lesquelles de longues<br />
périodes sont nécessaires pour la reconstitution des réserves pour la reproduction.<br />
Nos résultats indiquent qu’une fraction importante de la mortalité s’exprime un an<br />
après la reproduction. Cette mortalité post-reproductrice est liée à l’épuisement<br />
127
physiologique des femelles après la mise bas. La vipère aspic est donc affectée par<br />
une combinaison complexe de coûts se manifestant à la fois l’année de la<br />
reproduction (“breeding cost”) mais aussi l’année suivante (“post-breeding cost”)<br />
alors que les coûts associés à la phase de capitalisation des réserves avant la<br />
reproduction (“pre-breeding cost”) semblent plutôt réduits.<br />
128
B. Article 4<br />
Reproduction in a typical capital breeder:<br />
costs, currencies and complications in the<br />
aspic viper (Vipera aspis)<br />
Xavier Bonnet 1 , Olivier Lourdais 1 2 3 , Richard Shine 4 & Guy Naulleau 1<br />
1 Centre d'Etudes Biologiques de Chizé, <strong>CNRS</strong>, 79360, Villiers en Bois, France<br />
2 Conseil Général Des Deux Sèvres, Rue de L’abreuvoir, 79021, Niort, France<br />
3 University of Poitiers, 40 avenue du recteur Pineau, 86022 Poitiers, France<br />
4 Biological Sciences A08, University of Sydney, NSW 2006, Australia<br />
Published in Ecology 83, 2124-2135.<br />
(2002)<br />
129
Abstract<br />
Female aspic vipers (Vipera aspis) are "capital breeders", and delay reproduction<br />
until they have amassed large energy reserves. Data from an eight-year mark-<br />
recapture study on free-ranging vipers suggest that potential costs of reproduction<br />
were high for these animals, in terms of survival as well as growth and energy<br />
storage. Females that reproduced experienced higher mortality rates than non-<br />
reproductive females, and hence exhibited a tendency toward semelparity, grew less,<br />
and devoted most of their energy stores to reproduction. Both the depletion of body<br />
reserves and the low survival of reproductive females translated into significant costs<br />
(decrements of LRS). However, the cessation of growth during pregnancy had no<br />
detectable effect on LRS. Most females produced only a single litter during their<br />
lifetimes. A female’s “costs” in energy terms were not negatively correlated with her<br />
future reproductive output, probably because female vipers vary considerably in the<br />
rate at which they can accumulate energy. This notion is supported by the<br />
observations that (1) females with higher initial body reserves expended more energy<br />
during reproduction, and (2) females that accumulated energy more rapidly after<br />
parturition were more likely to survive and to breed again. This kind of variation<br />
among females masks any underlying trade-off <strong>between</strong> current reproductive effort<br />
and probable future reproductive success. Despite this complication, a strong link<br />
<strong>between</strong> rates of survival and post-reproductive mass recovery suggests that<br />
changes in body reserves govern reproductive effort in this species.<br />
Key words: breeding frequency, cost of reproduction, energy storage, reproductive<br />
effort, semelparity, snakes.<br />
130
Introduction<br />
<strong>The</strong> extensive scientific literature on "costs of reproduction" falls into two main<br />
categories, with relatively little overlap. This dichotomy involves theory-based<br />
(primarily mathematical) explorations on the one hand, and empirical studies of living<br />
animals on the other. Many mathematical models in life-history theory incorporate<br />
some causal link <strong>between</strong> an animal's current reproductive expenditure and its<br />
probable future reproductive output. <strong>The</strong>se models suggest that the exact nature of<br />
that link has profound implications for the kinds of life-history strategies that will<br />
maximize lifetime reproductive success and hence, are expected to evolve under the<br />
conditions posited in the model (e.g., Williams 1966a,b; Schaffer 1974; Winkler &<br />
Wallin 1987; Shine & Schwarzkopf 1992).<br />
Unfortunately, it is difficult to translate these apparently simple notions into<br />
practicable measures of reproductive costs (Reznick 1992; Jönsson & Tuomi 1992;<br />
Reznick et al. 2000). Thus, much of the empirical literature on "costs of reproduction"<br />
relies on measuring variables that are linked only indirectly to the potential costs<br />
experienced by reproducing organisms (Stearns 1992). Such variables include<br />
measures of reproductive output (i.e. clutch sizes), reproductive output relative to<br />
maternal size (e.g., Relative Clutch Mass; Cuellar 1984; Seigel & Fitch 1984), or<br />
effects of reproduction on maternal traits (e.g., metabolism, locomotor speeds, post-<br />
parturition maternal body condition: e.g., Shine1980; Birchard et al; 1984; Seigel et<br />
al. 1987; Lee et al. 1996). However, it is not easy to determine whether or not these<br />
measures correlate with the degree to which current expenditure decreases future<br />
probable reproductive success (i.e., decrements in lifetime reproductive success [=<br />
LRS]: Williams 1966a,b). First, LRS is extraordinarily difficult to measure in mobile or<br />
long-lived animals (Clutton-Brock 1988). Second, effects of reproductive output on<br />
131
LRS can be mediated via several different processes. <strong>The</strong> most obvious dichotomy<br />
is <strong>between</strong> survival costs and energy costs (e.g., Calow 1979), but there are many<br />
subtleties even within these two broad categories. For example, higher energy<br />
expenditure on current reproduction may reduce LRS via decreased energy stores<br />
and/or by decreasing subsequent growth rates (and hence fecundity if body size<br />
enhances reproductive success). Third, these currencies are not independent<br />
(Bauwens & Thoen 1981; Brodie 1989). Fourth, the magnitude of various "costs" is<br />
likely to shift among habitats and years (Festa-Bianchet et al. 1998). Fifth, variation<br />
in levels of resource availability among individuals may generate a positive rather<br />
than negative correlation <strong>between</strong> current reproductive output and future<br />
reproductive success, thereby masking a trade-off <strong>between</strong> these two traits (Bell &<br />
Koufopanou 1986; Van Noordwijk & de Jong 1986).<br />
In order to overcome some of these difficulties, such studies should focus on<br />
various species that offer logistical advantages for measuring the relevant traits (i.e.<br />
reproductive expenditure and its consequences) (Reznick 1992; Seigel 1993; Shine<br />
& Bonnet 2000). In the current paper, we describe an eight-year study on such a<br />
system, the aspic viper (Vipera aspis). Aspic vipers are abundant, sedentary (hence,<br />
easily recaptured), and live in a relatively cool climate (so that thermoregulatory<br />
needs during vitellogenesis and gestation substantially modify patterns of movement<br />
and feeding). Perhaps more importantly, females reproduce on a less-than-annual<br />
basis, so that we can readily compare females that are in the reproductive versus<br />
non-reproductive years of their cycles (Bonnet et al. 2000b). This species is a typical<br />
capital breeder (sensu Stearns 1992). Large body reserves must be accumulated<br />
during long periods (years) before reproduction, both for the induction of<br />
vitellogenesis and to fuel most of the reproductive effort (Saint Girons 1957; Bonnet<br />
et al. 1994, 2001b; Naulleau & Bonnet 1996). <strong>The</strong> massive depletion of body<br />
132
eserves in the course of reproduction results in a low breeding frequency and hence<br />
can be considered as a typical energy cost (Naulleau & Bonnet 1996; Bonnet et al.<br />
2001b). Like other ectotherms, female aspic vipers are pre-adapted to capital<br />
breeding (Pough 1980; Bonnet et al. 1998). In contrast, endotherms are less well-<br />
suited to long-term storage of body reserves (Jönsson 1997), and these animals tend<br />
to rely on "income" rather than "capital" to fuel reproduction (Else & Hulbert 1981;<br />
Bonnet et al. 1998; Bronson 1998; Schneider et al. 2000). As a result, changes in<br />
body mass during reproduction may be a poor indicator of energy costs of<br />
reproduction in mammals and birds. First, such changes may reflect fluctuations in<br />
<strong>food</strong> availability independently of reproductive effort. Second, the high basal<br />
metabolic rate of these animals means that body reserves can be depleted rapidly<br />
during short periods (days) of starvation in non-reproductive individuals (Nagy 1987).<br />
This situation seriously complicates direct comparisons <strong>between</strong> reproductive and<br />
non-reproductive individuals. By contrast, most ectotherms can survive for long<br />
periods of time (months to years) during <strong>total</strong> starvation with minor changes in body<br />
mass (Pough 1980). Hence, any massive decrease in body mass that is temporally<br />
and physiologically associated with reproductive effort is likely to be a direct<br />
consequence of reproduction. <strong>The</strong> comparison <strong>between</strong> reproductive and non-<br />
reproductive females is thus straightforward: both survival rates and changes in body<br />
mass are likely useful candidates for measuring potential costs of reproduction in this<br />
snake species. Body reserves can influence LRS through major components such<br />
as breeding frequency (Naulleau & Bonnet 1996), current fecundity (Bonnet et al.<br />
2001a), and survival (Bonnet et al. 2000a).<br />
133
We set out to answer three main questions:<br />
(1) What form do “costs of reproduction” take in aspic vipers? We explore three<br />
potential currencies in this respect: probabilities of survival, decreases in energy<br />
reserves, and rates of growth in body length (because fecundity is generally<br />
associated with size in snakes).<br />
(2) Because we have long-term data, we can assess whether or not these kinds<br />
of “cost” indices (such as reductions in growth and body condition), or the magnitude<br />
of expenditure on current reproduction, actually translate into lower reproductive<br />
success in the future. This must be true for survival costs, although even here it is<br />
possible that the effect is trivial (e.g., if subsequent post-reproductive survival rates<br />
are so low that few females live long enough to reproduce again anyway). For<br />
energy-storage costs, is it true that an unusually emaciated female will delay the<br />
production of her next litter? For growth costs, will females forfeit fecundity<br />
increments in later litters if they grow less after their first reproduction? Are females<br />
with high reproductive output in their first litter, less likely to reproduce again or<br />
produce a small litter if they do?<br />
(3) What attributes of a female in the year after she reproduces (e.g., her rate of<br />
growth in body length, or her rate of replenishment of body condition) offer the best<br />
predictors of her subsequent reproductive output (i.e., determine the time she<br />
reproduces, how many offspring she produces, and their size)? Data on this issue<br />
can help us to identify which of these traits may offer the best currency in which to<br />
measure “costs” of reproduction in a typical ectothermic capital breeder.<br />
134
Materials and methods<br />
Animals and study site<br />
<strong>The</strong> aspic viper (Vipera aspis) is a stocky, medium-sized venomous snake species<br />
widely distributed through western Europe (Naulleau 1997). Adults in our population<br />
average 47.7 ± 3.4-cm snout-vent length (SVL), 54.3 ± 3.8cm <strong>total</strong> length. Female<br />
vipers mature at an age of approximately three years (Bonnet et al. 1999a). Mating<br />
occurs in spring (March-April: Saint Girons 1952, 1957a,b; Vacher-Vallas 1997;<br />
Naulleau et al. 1999). In our population, females are gravid over summer, and give<br />
birth to a litter of 1 to 13 large (20.7 ± 1.2 cm TL, 6.1 ± 1 g) offspring in autumn (late<br />
August-September). Most female vipers do not reproduce every year (Bonnet &<br />
Naulleau 1996). <strong>The</strong> exact frequency of reproduction depends upon thermal<br />
conditions (especially, length of the activity season) and <strong>food</strong> supply, so that<br />
reproductive frequencies differ among areas and among years (Saint Girons 1952,<br />
1957a,b, 1996). This less-than-annual frequency of reproduction results from the<br />
time taken to replenish energy stores for the next litter: females delay reproduction<br />
until they exceed a minimum body-condition threshold (Naulleau & Bonnet 1996).<br />
We studied the aspic viper in a closed population in western central France<br />
(Les Moutiers en Retz, 47 o 03N'; 02 o 00W'; Bonnet & Naulleau 1996). <strong>The</strong> study site<br />
is 33 ha in extent. It is bordered to the north and east by roads, to the south by the<br />
Atlantic Ocean, and to the west by a camping site (Vacher-Vallas et al. 1999). It is a<br />
typical parkland habitat that has not been intensively managed for 15 years. Thus,<br />
the hedges form a dense network, and bushes (especially brambles) have invaded<br />
the meadows to varying degrees. In some meadows, oak and pine plantations have<br />
recently (1993 to 1995) been established. <strong>The</strong> climate is a temperate oceanic one<br />
(see Bonnet & Naulleau 1993 for average temperatures).<br />
135
Procedure<br />
One to three people checked the area almost every sunny day from the time the<br />
snakes emerged (late February for the females) to the end of their reproductive<br />
period (September), and less frequently in late September-October. Searching effort<br />
averaged 95.3 days per annual activity season (sd = 32.2, range = 51 to 124 days)<br />
and 523.7 hours per annual activity season (sd = 198.4, range = 232 to 614 hours).<br />
Over the period 1992 to 1999, we hand-captured 469 different adult female vipers.<br />
Classification of these animals as adults is based on the minimum size we have<br />
recorded for parturition in this population (41.5 cm SVL, 47 cm <strong>total</strong> length). Each<br />
female was individually marked for future identification by scale-clipping in 1992, and<br />
fitted with a Passive Integrated Transponder (PIT) tag since 1993, measured (to the<br />
nearest 0.5 cm, SVL and TL), weighed (nearest 1 g), palpated for prey, eggs or<br />
embryos, and released at her exact place of capture. Reproductive status was<br />
determined by palpation of eggs or embryos, by records of parturition or by obviously<br />
post-parturient body condition. Immediately after giving birth, females are very<br />
emaciated, with a flaccid abdomen and extensive skin folds. Our analyses exclude<br />
body-mass data taken from individuals containing prey items or oviductal embryos.<br />
Recapture probabilities were high (see Bonnet & Naulleau 1996), but few females (N<br />
= 5) were recaptured in eight consecutive years due to the low survival rate of<br />
reproductive females.<br />
Survival<br />
We scored a female as having died if we failed to locate her on >250 days’ searching<br />
over a period of > 2 years. Given the very low vagility (5m/day on average: Naulleau<br />
et al. 1996) and high recapture rates within this closed population, we can be<br />
confident that such animals had died rather than moved away. Only one female<br />
136
escaped capture during three consecutive years (marked in 1993 and not recaptured<br />
until 1997), and only eight animals were “missed” during two consecutive years.<br />
Changes in body mass and body size<br />
We measured changes in body mass and body length from the onset of<br />
vitellogenesis to the post-parturition period. This covers the entire activity season (6-<br />
8 months per year). Our criteria for inclusion of data in the analyses were as follows:<br />
“early vitellogenesis” was defined as the period from March to April (Bonnet et al.<br />
1994), and data from snakes captured after this period were not included for<br />
analyses of change in body mass or length. However, we did include these later<br />
captures for analyses of rates of survival and future reproduction. Changes in body<br />
mass and rates of growth in body length were calculated from March-April to August-<br />
November within a given year, and (ignoring hibernation) from March-April to the next<br />
March-April <strong>between</strong> years. Our extensive data indicate that vipers did not show any<br />
significant change in either body mass or body length over the hibernation period.<br />
Reproductive output<br />
As soon as we recorded the first parturition of the year (generally in the second half<br />
of August), we collected all of the gravid females that we could locate, and held them<br />
in captivity in individual cages (for periods of up to one month) so that we could<br />
count, measure, weigh, sex and mark the offspring. Captive females were weighed<br />
every two days, and immediately after parturition. We defined Relative Clutch Mass<br />
(RCM) as the <strong>total</strong> mass of the litter (including stillborn offspring, etc.) divided by the<br />
post-parturient mass of the mother. Data were obtained on 195 litters from 157<br />
different females. Several females were captured shortly before parturition in two<br />
different years, so that we were able to quantify reproductive output on each<br />
137
occasion. Data on these animals allowed us to explore the <strong>relationship</strong> <strong>between</strong><br />
initial reproductive output and subsequent changes in body size, body condition, and<br />
reproductive output. However, for most analyses, data were not available from all of<br />
the females (e.g., data for mass change during the reproductive period were<br />
available on 301 females).<br />
Activities associated with reproduction likely to be costly in aspic vipers<br />
In combination with extensive studies in other parts of France (e.g., Saint Girons<br />
1952, 1957a,b, 1996), our studies reveal that reproduction imposes marked changes<br />
on several aspects of the biology of female vipers. <strong>The</strong> major modifications are as<br />
follows:<br />
(1) Relative to males and non-reproductive females, reproductive female vipers<br />
become more sedentary in the course of gestation: mean home ranges decrease<br />
sharply from 3,000 m 2 to 300 m 2 (Naulleau et al. 1996). (2) Gravid vipers spend<br />
more of their time in behavioral thermoregulation than do other animals within the<br />
population (Bonnet & Naulleau 1996), and hence could be more exposed to<br />
predation (mainly birds; Naulleau et al. 1997). (3) Pregnant females progressively<br />
reduce their rate of feeding, and may cease feeding in the latter stages of gestation<br />
(unpublished data, and see Saint Girons 1952, 1957a,b; 1996). (4) Female vipers<br />
show a consistent pattern of change in body mass over the course of the<br />
reproductive cycle. Body condition (mass relative to length) increases during the<br />
non-reproductive years, until it exceeds the threshold level required to initiate<br />
vitellogenesis (Naulleau & Bonnet 1996). <strong>The</strong> female's mass drops dramatically at<br />
parturition. <strong>The</strong> magnitude of this decrease in maternal mass (i.e., from the<br />
beginning to the end of the reproductive bout) offers a measure of her net energy<br />
138
expenditure over that period: (body reserves invested into the litter + metabolic<br />
expenditure) - <strong>food</strong> <strong>intake</strong>.<br />
Analyses<br />
Body condition was calculated as residual values from the regression of body mass<br />
(Log) against body size (Log) (Jayne & Benett 1990). We randomly selected a single<br />
record per female to avoid pseudo-replication bias in this analysis. However,<br />
ignoring such bias, and including 753 females where reproductive status is known<br />
(among a <strong>total</strong> of 853 “female-year” data, reproductive status was unknown on 100<br />
occasions) in the analyses did not change any results significantly . Importantly, the<br />
mean body sizes of reproductive females and non-reproductive females were similar<br />
(ANOVA with reproductive status as the factor and SVL as the dependent variable;<br />
F1, 267 = 0.90, P = 0.34), allowing us to compare these two categories of females<br />
without having to take into account possible effects of body size on rates of survival<br />
or growth (Bonnet et al. 2000b).<br />
<strong>The</strong> snakes' tendency toward semelparity greatly reduced our sample sizes for<br />
tests comparing successive reproductive events by the same female. Conclusions<br />
from such tests are problematic because of their low power to reject the null<br />
hypothesis. Statistical conventions are much more rigid with respect to α (the<br />
decision to reject null hypotheses with an error α < 0.05) than with respect to β (the<br />
type II error). Statistical textbooks generally recommend that the power of a test (1-<br />
β) should be > 0.80. Because statistical tests (especially correlation analyses) have<br />
very low power when sample size is small, we performed power analysis to estimate<br />
the ability of our statistical tests to detect “significant” effects. In all our ANOVAs,<br />
power was close to 1.0 and has not been reported. Correlation analyses are more<br />
sensitive; a low correlation <strong>between</strong> two variables inevitably (due to the structural<br />
139
trade-off <strong>between</strong> α and β error rates) leads to a low power of the analysis, even with<br />
a large sample size. Such a low power does not invalidate the analysis, but means<br />
that caution is needed in interpretation. In such cases, we calculated the sample<br />
sizes that would be required to detect a "significant" result at low α and β error rates<br />
(0.05 and 0.10). Power analyses and required sample size estimates are not a<br />
panacea to low sample sizes, but no conflict arose among our different tests. We<br />
used Statistica 5.1 and 6.0 to perform the statistical analyses.<br />
Results<br />
How high are the potential costs of reproduction for female aspic vipers?<br />
We can estimate these potential costs by comparing females in reproductive years<br />
versus non-reproductive years of their cycles. Our data provide four separate<br />
indices: (1) whether or not a female survived to produce her litter; (2) how much her<br />
energy stores (as measured by changes in body mass) decreased over the<br />
reproductive period (vitellogenesis plus gestation); (3) how much she grew in body<br />
length over this period; and (4) her body condition (mass relative to length) after<br />
giving birth. <strong>The</strong> first three variables are self-explanatory; the fourth is relevant<br />
because previous studies on two species of viviparous snakes sympatric with V.<br />
aspis have suggested that this trait (maternal post-parturition condition) may be a<br />
significant predictor of maternal survival rates (Madsen & Shine 1993; Luiselli et al.<br />
1996). Comparisons <strong>between</strong> reproductive and non-reproductive females suggest<br />
that potential costs are high in each of these currencies.<br />
1. Survival rates. We have data for 381 randomly sampled females caught<br />
<strong>between</strong> 1992 and 1997 (females caught in 1998 and 1999 are ignored), for which<br />
140
eproductive status and survival are known. Adult female vipers experienced much<br />
higher survival during non-reproductive years (123 of 145 records, = 85%) than<br />
during reproductive years (106 of 236 records, = 45%; χ 2 = 59.7, 1 df, p< 0.0001;<br />
see Figure 1). <strong>The</strong> <strong>total</strong> potential cost of reproduction in terms of survival is actually<br />
higher and more complex than this, because a female viper’s <strong>food</strong>-gathering activities<br />
during her non-reproductive years also constitute a component of reproduction<br />
(Bonnet et al. 2000a). However, the clear result is that activities directly associated<br />
with the production of the litter cause a substantial decline in annual survival. <strong>The</strong><br />
major component of the low survival of reproductive females seems to occur mainly<br />
before parturition (i.e. during vitellogenesis and gestation) with 88 of 142 females<br />
dying before parturition versus 31 of 78 females dying later, <strong>between</strong> parturition and<br />
the following spring (χ 2 = 10.0, df = 1, p = 0.0016; the sample size was reduced for<br />
this test because we selected females caught several times during a given year).<br />
Considering the 236 reproductive females, the studied population exhibited a<br />
tendency toward semelparity: 182 females became vitellogenic only once, 43 twice,<br />
and 11 were vitellogenic on three different occasions; leading to a mean number of<br />
1.28 reproductions per female during their life. Importantly, many (roughly 50%) of<br />
these females died before giving birth, and the average <strong>total</strong> number of litter per<br />
“reproductive female” was actually less than 1. For reliability, we excluded from this<br />
analysis the few females that survived over long periods but that we failed to<br />
recapture in some years (for example, caught in 1992 and in 1995 but missed in<br />
1993 and 1994), because of uncertainty about their number of reproductive episodes.<br />
2. Changes in maternal body mass. All reproductive females (regardless of whether<br />
we used the full data set or only one randomly-selected data point per female) lost<br />
body mass over the period from spring (vitellogenesis) to autumn (post-parturition).<br />
141
In contrast, non-reproductive females generally gained substantially in mass (ANOVA<br />
with reproductive status as the factor: F(1.299) = 616.8, p < 0.0001; see Figure 1).<br />
This decrease corresponds to the depletion of body reserves that are transferred in<br />
the embryos plus the metabolic costs of vitellogenesis and gestation (6 months).<br />
Factoring out the potential effect of size through ANCOVAs (reproductive status as<br />
the factor, changes in body mass as the dependent variable and body size as the<br />
covariate) leads to similar results (F(1.298)=606, P < 0.0001).<br />
3. Growth rates in body length. Most (57 of 68 = 84%) reproductive females<br />
showed no detectable growth in body length during the reproductive period, whereas<br />
non-reproductive females showed significant growth (ANOVA with reproductive<br />
status as the factor: F(1.375)= 37.2, p< 0.0001; see Figure 1).<br />
4. Body condition post-parturition. Females that reproduced were in much lower<br />
body condition after giving birth than were non-reproductive females at the same time<br />
of year (ANOVA with reproductive status as the factor: F(1.255)=229.7, p< 0.0001;<br />
see Figure 1). In fact, almost all post-parturient females were in poor body condition,<br />
as indicated by their abundant skin folds, typical of snakes with minimal body<br />
reserves (Bonnet 1996).<br />
142
PROBABOLITY OF SURVIVAL (%)<br />
GROWTH RATE<br />
(cm/year)<br />
CHANGES IN BODY MASS (g)<br />
100<br />
90<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
2.5<br />
2.0<br />
1.5<br />
1.0<br />
0.5<br />
0.0<br />
50<br />
45<br />
40<br />
35<br />
30<br />
25<br />
20<br />
15<br />
10<br />
5<br />
0<br />
Reproductive<br />
Reproductive<br />
Will<br />
Reproductive<br />
Again<br />
Non<br />
Reproductive<br />
Non<br />
Reproductive<br />
Will not<br />
CHANGES IN BODY MASS<br />
(g/year)<br />
BODY MASS ADJUSTED<br />
FOR SVL (g)<br />
CHANGES IN BODY MASS (g)<br />
30<br />
20<br />
10<br />
0<br />
-10<br />
-20<br />
-30<br />
-40<br />
110<br />
100<br />
90<br />
80<br />
70<br />
60<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
50<br />
45<br />
40<br />
35<br />
30<br />
25<br />
20<br />
15<br />
10<br />
5<br />
0<br />
Reproductive<br />
Reproductive<br />
Biennial<br />
Non<br />
Reproductive<br />
Non<br />
Reproductive<br />
Triennial<br />
Figure 1. Comparisons of survival rates (top left), change in body mass (top right), rate of growth in<br />
body length (middle left) and maternal body condition (residual scores from linear regression of ln-<br />
transformed mass versus snout-vent length) immediately after the time of parturition (middle right) in<br />
reproductive versus non-reproductive female vipers (for ease of interpretation, this graph shows body<br />
mass adjusted for maternal SVL rather than the body condition index). Female vipers that regained<br />
mass more rapidly after reproduction were more likely to reproduce again (bottom left) and did so after<br />
a briefer delay (bottom right: mean mass changes of females that had an inter-litter interval of 2<br />
(biennial) or 3 (triennial) years.<br />
143
Do these measures of “cost” translate into lower future reproductive output?<br />
Although it seems plausible that future reproductive output will be compromised by<br />
high levels of current expenditure (an index of energetic reproductive effort), this<br />
assumption requires empirical verification. To do this, we can examine the<br />
<strong>relationship</strong> <strong>between</strong> our measures of potential cost (survival rate, mass loss, growth<br />
rate, and maternal body condition post-partum) and reproductive output (Relative<br />
Clutch Mass) on the one hand, and future reproductive output on the other. That is,<br />
do high levels of potential cost or reproductive output correlate with lower levels of<br />
future output, as predicted by the “costs” hypothesis? In other words, do our<br />
measures of potential costs reveal real costs of reproduction?<br />
1. Survival rates. Obviously, females who die during reproduction are less likely<br />
to breed again than are surviving females. Our data clearly show that reproductive<br />
females have a low probability of survival. However, the degree to which this is a<br />
significant cost to LRS depends not only upon survival probabilities in the absence of<br />
reproduction, but also upon a postpartum female's probability of reproducing again<br />
even if she survives until the following active season. If most "surviving" post-<br />
parturient females in this population actually die before a second reproductive<br />
opportunity (regardless of their reproductive output in the first litter), then there may<br />
be little difference in the incidence of second litters <strong>between</strong> animals that survive to<br />
their first parturition versus those that die during their first reproductive year. If so,<br />
the high mortality associated with reproduction will no longer be a real cost. This is<br />
not the case in our population. Almost 50% (59 of 123) of the reproductive females<br />
that survived initiated reproduction a second time. Although a significant proportion<br />
of these animals undoubtedly died during their second pregnancy and thus did not<br />
actually produce two litters, females that survive to reproduction have significant<br />
144
opportunity to reproduce again. Overall, the low survival rate of reproductive<br />
females, regardless of the period during which mortality peaks (i.e. before or after<br />
parturition), entails significant costs by strongly decreasing future probabilities of<br />
reproduction.<br />
2. Changes in maternal body mass. Were females that lost less body mass over<br />
the reproductive period more likely to breed again? Our data show that this was not<br />
the case (logistic regression; χ 2 = 2.16, df = 1, n = 71, p = 0.14; Table 1). Similarly,<br />
we might expect females that lost more mass when producing their first litter to delay<br />
subsequent reproduction for a longer period in order to recoup their energy stores.<br />
No such effect was apparent in our data (χ 2 = 0.77, df = 1, N = 64, p = 0.38). Lastly,<br />
we might also predict that females who lost more mass would produce smaller-than-<br />
average litters and/or smaller-than-average neonates at their next reproductive<br />
episode. Neither of these patterns appeared in our data set (for litter size, r = 0.24, n<br />
= 20, p = 0.29; for offspring size, r = 0.18, n = 10, p = 0.64). However, we note that<br />
the sample sizes were small for these later tests, and hence that the power of these<br />
analyses was low (0.27 and 0.13 respectively), perhaps reflecting the non-<br />
significance of the results. Nonetheless, the correlations were positive rather than<br />
negative, and the sample sizes that would be required to obtain a significant effect<br />
were high (178 and 320 respectively). Thus, any “undetected” effects were probably<br />
weak or negligible.<br />
3. Growth rates in body length. <strong>The</strong> notion that growth costs decrease future<br />
reproductive output depends upon the assumptions that (i) a decrease in growth rate<br />
during reproduction will influence body size at the next reproduction; and (ii) a larger<br />
body size will allow a larger reproductive output. Neither of these assumptions is well<br />
supported by our data. In our population, body size influences reproductive output<br />
145
only slightly (see Bonnet et al. 2000b for a detailed discussion of this issue). In<br />
addition, the extent of a female's growth during reproduction did not correlate with her<br />
body size at the next reproduction (r = 0.24, n = 15, p = 0.39). <strong>The</strong> power (0.22) of<br />
this analysis was low, and we may have failed to detect a slight effect. Nonetheless,<br />
the weak influence of maternal size on reproductive output (Bonnet et al. 2000b)<br />
suggests that growth effects of reproduction probably have little effect on future<br />
fecundity. Thus, the almost <strong>total</strong> inhibition of growth by reproducing female vipers did<br />
not translate into a significant cost of reproduction.<br />
4. Relative Clutch Mass. Female vipers that produced small litters relative to<br />
their own body size were no more likely to reproduce again than were conspecifics<br />
producing larger litters (χ 2 = 0.78, df = 1, n = 137, p < 0.37; Table 1). Females that<br />
produced large first litters (high RCM) did not exhibit low rather than high RCMs in<br />
their second litters (r = 0.45, n = 14, p = 0.11). <strong>The</strong> power of this analysis was only<br />
0.51. Regardless, the positive rather than negative correlation strongly suggests<br />
(counter-intuitively) that high initial RCM does not translate into a reduced<br />
subsequent RCM. Females with high RCMs in their first litter did not produce smaller<br />
offspring (relative to other females) in their second litters (r = -0.20, n = 12, p = 0.54;<br />
but the power was low = 0.15) and did not delay their subsequent breeding attempts<br />
relative to other females (χ 2 = 0.04, df = 1, n = 63, p = 0.82). A ratio measure such<br />
as RCM facilitates intuitive understanding of reproductive output relative to maternal<br />
size, but may introduce statistical artifacts into analyses (e.g. Seigel and Ford 1987).<br />
To overcome this problem, we repeated all of these analyses using alternative<br />
measures of reproductive output: either absolute mass of the litter, or residual scores<br />
from the general linear regression of litter mass to body mass. We obtained similar<br />
results in each case.<br />
146
5. Maternal body condition post-partum : Female vipers are emaciated<br />
immediately after parturition, and have relatively low energy reserves at this time.<br />
Nonetheless, the degree of maternal emaciation did not correlate with a female’s<br />
probability of survival to the next season (logistic regression χ 2 = 1.88, 1 df, n= 119,<br />
P = 0.17), or with her probability of breeding again (analysis restricted to the females<br />
that survived, logistic regression χ 2 = 1.15, 1 df, n=71, p= 0.28; Table 1). Similarly,<br />
her litter size at the next reproduction (r = -0.25, n=14, p= 0.36), or offspring size at<br />
the next reproduction (r = 0.18, n=12, p=0.58), was not strongly influenced by a post-<br />
parturition female’s body reserves. <strong>The</strong> low power of these two later analyses (0.22<br />
and 0.14) require caution in interpretation; but the required sample sizes to obtain an<br />
α level < 0.05 were relatively high (164 and 320), suggesting that any “missed” effect<br />
was weak. Post-partum maternal body condition tended to affect the number of<br />
years’ delay until her next reproduction, but this trend did not attain the conventional<br />
level of statistical significance (χ 2 = 3.39, 1 df, n= 63, p= 0.065).<br />
Table 1. Mean values (± SD) of mass loss, growth rate, relative litter mass, and post-partum body<br />
condition recorded in the course of reproduction in wild female aspic vipers. <strong>The</strong> changes in mass and<br />
growth rate were calculated from the onset of vitellogenesis to parturition. All of the females used in<br />
this analyses survived to produce their first litter and were classified based upon whether or not they<br />
also reproduced again in the future. We found no significant differences <strong>between</strong> the two groups of<br />
females (See text for statistics; except for growth rate: ANCOVA with SVL as the co-variable and<br />
growth rate the dependent variable; F(1.70) = 2.53, p = 0.12).<br />
Trait Will breed again Will not N<br />
Mass loss (g) -37.2 ± 17 -35.1 ± 16 71<br />
Growth rate (cm year –1 ) 0.32 ± 1.46 0.87 ± 1.46 73<br />
Relative litter mass (%) -0.55 ± 0.22 -0.51 ± 0.24 137<br />
Post-partum body<br />
condition<br />
0.005 ± 0.1 -0.002 ± 0.1 119<br />
147
MASS RECOVERY<br />
g year-1<br />
What currencies of potential costs influence future reproductive output?<br />
Another way to identify the appropriate currency in which to assess “costs” is to<br />
ignore reproductive output per se, and focus instead on the long-term consequences<br />
of rates of change in the attributes that we know to be affected by reproductive<br />
expenditure. Body reserves and growth rates can readily be examined in this way.<br />
1. Maternal change in body mass after the first litter. If body reserves are<br />
important, we expect that females that recoup their energy (body mass) reserves<br />
rapidly will be more likely to breed again (and will breed sooner) than females that<br />
regain mass only slowly. Analysis supports this proposition for the female’s<br />
probability of reproducing again (χ 2 = 10.53, df = 1, n = 55, p < 0.001) and for the<br />
duration of the delay to her next reproduction (χ 2 = 27.70, df = 1, n = 42, p< 0.0001:<br />
see Figure 2). <strong>The</strong> female’s litter size and neonate size at her second reproduction<br />
showed no significant <strong>relationship</strong> with her rates of mass increase after the first litter<br />
(for litter size, r = 0.45, n = 14, p= 0.10; and r = 0.29, n = 11, p= 0.38 for neonate<br />
size). Although our sample sizes were small and associated statistical power low<br />
(0.51 and 0.22 respectively), these correlations were positive rather than negative.<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
a b<br />
-20<br />
-30 -20 -10 0 10 20 30<br />
RELATIVE LITTER MASS<br />
residuals<br />
148<br />
GROWTH RATE<br />
(residuals)<br />
6<br />
4<br />
2<br />
0<br />
-2<br />
-4<br />
-30 -20 -10 0 10 20 30<br />
RELATIVE LITTER MASS<br />
(residuals)
Figure. 2. Relationships <strong>between</strong> a female viper's reproductive output in her first litter and her<br />
subsequent rate of recovery of body reserves and growth rate in body length. Female vipers that<br />
invested more into reproduction exhibited higher rather than lower rates of mass recovery or growth, in<br />
contradiction to predictions from the "costs" hypothesis. Females with higher relative clutch mass<br />
(RCM, residuals of litter mass on post-parturient female’s mass) recovered their body reserves more<br />
rapidly during the following year (r = 0.43, n = 36, p = 0.01; the power of this analysis was 0.85; Figure<br />
2a); without trading this body reserve replenishment against growth rate the following year (r = 0.20, n<br />
= 36, p = 0.20; power = 0.32; Figure 2b). To control for the effect of size on growth rate, growth rate<br />
was calculated as the residual values of the regression of absolute gain in size (cm/year) on initial<br />
body size (SVL in cm) (Fig2b). Because mass recovery was not affected by body size, we used<br />
absolute values (Figure 2a).<br />
2. Rate of growth in body length after the first litter. Faster-growing females were<br />
more likely to breed again (χ 2 = 6.10, df = 1, n = 89, p= 0.01), and bred sooner than<br />
slower-growing animals (χ 2 = 4.1, df = 1, n = 58, p= 0.04). Taking into account the<br />
effect of a female's mean body size on her growth rate (by using residuals of the<br />
regression <strong>between</strong> absolute growth rate and initial SVL: Bonnet et al. 2000b), does<br />
not alter this conclusion (χ 2 = 8.20, df = 1, n = 67, p= 0.004; and χ 2 = 4.90, df = 1, n<br />
= 47, p= 0.003). Reproductive output in the second litter was not associated with<br />
rates of body growth in the period following production of the first litter (for litter size, r<br />
= 0.05, n = 18, p= 0.84; for offspring size, r = 0.12, n = 15, p= 0.65). Despite a low<br />
power of these analyses (0.05 and 0.07), the very weak correlations would require<br />
sample sizes of 4,198 and 725 to attain statistical "significance", and hence suggest<br />
an absence of effect.<br />
Discussion<br />
Our analyses of costs of reproduction differ from most previous studies in this field,<br />
by incorporating two steps into our analyses. First, we have measured the ‘potential<br />
costs of reproduction’ for female aspic vipers in terms of energy and survival.<br />
149
Second, we have examined the consequences of these ‘potential costs’ in terms of<br />
lifetime reproductive success. That is, we have tried to determine whether or not<br />
‘potential costs’ actually translate into a significant decrease in future reproductive<br />
success. By adopting this method, we can better tease apart two components of the<br />
life-history: reproductive effort per se, and costs of reproduction (Niewiarowski &<br />
Dunham 1994). Our data support four main conclusions:<br />
1) Reproducing female vipers commit themselves to a considerable effort in<br />
vitellogenesis and gestation, to the degree that most females produce only a single<br />
litter during their lifetime. Such reproductive effort is reflected in high rates of mass<br />
loss, a virtual cessation of growth, and a decrease in the probability of survival.<br />
2) <strong>The</strong> different components of reproductive effort do not systematically, and equally,<br />
translate into real ‘costs of reproduction’. <strong>The</strong> cessation of growth rate during<br />
reproductive years had no measurable influence on future reproduction. By contrast,<br />
the strong mobilization of maternal reserves necessary to fuel reproductive effort had<br />
a considerable impact on breeding frequency and lifetime reproductive success<br />
through distinct, but interconnected mechanisms. First, vitellogenesis and gestation<br />
entail a strong increase in basking frequency and expose females to predation<br />
(Bonnet & Naulleau 1996; Naulleau 1997). Second, many post-parturient females<br />
die from starvation; and even if a female survives, she delays reproduction for a long<br />
period of time (1-3 years) until she has restored her body reserves (Naulleau &<br />
Bonnet 1996; Bonnet et al. 2000a). <strong>The</strong> lower survival caused by such thermal and<br />
energetic requirements of reproduction, along with the long period <strong>between</strong><br />
reproductive bouts (that automatically increase mortality from other causes, Bonnet<br />
et al. 2000a) constitute a high cost. Reproductive females that die, thereby lose<br />
150
significant opportunities to reproduce again and to increase the <strong>total</strong> number of<br />
offspring they could produce during their life.<br />
3) <strong>The</strong> magnitude of reproductive effort in a given reproductive bout does not affect<br />
future reproductive success. Levels of reproductive output and our measures of<br />
reproductive effort (such as lowered survival, lower maternal body condition, loss in<br />
mass, or decrements in growth) were not negatively correlated with future<br />
reproductive success.<br />
4) Changes in maternal body reserves over the complex alternation of non-<br />
reproductive and reproductive phases play a key role in the reproductive biology of<br />
female aspic vipers. Notably, the emaciation of post-parturient females determines<br />
the low breeding frequency and a high proportion of the mortality experienced by<br />
reproducing females. Despite the difficulty of detecting trade-off <strong>between</strong><br />
reproductive investment and future reproductive success, the strong link <strong>between</strong><br />
rate of mass recovery in post-parturient females, versus the duration of delay to<br />
production of the next litter (and probability of survival to this time) offers strong<br />
evidence that energy stores are a crucial currency. Thus, the rate that a female viper<br />
can replenish her energy stores after reproduction may strongly affect her LRS.<br />
<strong>The</strong>se conclusions have several implications for studies on “costs of<br />
reproduction”. For example, they offer a challenge to simplistic attempts to use<br />
measures of reproductive output such as RCM, increased metabolism or cessation of<br />
growth as a shorthand index of “costs of reproduction” (i.e. Shine 1980; Vitt & Price<br />
1982; Birchard 1984; Seigel et al. 1987). <strong>The</strong> reality is far more complex: even if<br />
reproductive effort is high, its magnitude may not correlate with future reproductive<br />
output in simple phenotypic comparisons (Bauwens & Thoen 1981; Brodie 1989;<br />
Dunham et al. 1994; Olsson et al. 2000). Nonetheless, our study is encouraging in<br />
151
that a relatively easily measured trait (changes in maternal mass) may offer a<br />
reasonable currency in which to estimate several of the major “costs”, at least in<br />
ectothermic animals. Because they entail significant costs, the physiological<br />
mechanisms that control the allocation of body reserves during reproduction should<br />
be under strong selection (Sinervo & Svensson 1998; Bonnet et al. 2002b). Such a<br />
situation enables us to identify more precisely the ecological context that can favor<br />
the emergence of capital breeding instead of income breeding as alternative<br />
reproductive strategies (Stearns 1992; Jönsson 1997; Bonnet et al. 1998).<br />
Although a comparative approach will be needed to examine the evolution of<br />
such traits, at present we cannot compare the absolute magnitude or form of<br />
reproductive “costs” in V. aspis with that in other reptiles because most previous<br />
studies have relied on indirect measures of “cost”. Indeed, “costs of reproduction”<br />
may be manifested differently. For example, there is apparently no significant<br />
survival cost of reproduction in Orsini’s Viper, whereas such costs are high in both<br />
adders and aspic vipers (Madsen & Shine 1992, 1993; Capula et al. 1992; Baron et<br />
al. 1996; Luiselli et al. 1996). Nonetheless, it may often be true that female<br />
viviparous snakes living in cool climates experience such high mortality during<br />
reproduction that many females produce only a single litter in their lifetimes, and<br />
hence exhibit a strong tendency toward semelparity (Brown 1991). In the case of V.<br />
aspis, the mortality comes not only from starvation per se (because some females<br />
maintain sufficient reserves to avoid this threat) but also from vulnerability to<br />
predation during pregnancy (probably due to increased basking) and associated<br />
dangers such as occlusion of the oviducts by inviable embryos (Naulleau 1997).<br />
Thus, although energy stores are a crucial currency that limits a female’s<br />
reproductive output, energy limitation is not the only proximate determinant of<br />
reproduction-associated mortality in female vipers. When they engage in<br />
152
eproduction, females shift from a very secretive to a conspicuous way of life (Bonnet<br />
& Naulleau 1996). <strong>The</strong> thermal requirements of vitellogenesis and gestation result in<br />
high rates of basking in reproductive females (Shine & Harlow 1993), regardless the<br />
number of eggs/embryos they carry (unpublished). Hence, reproductive female<br />
vipers are disproportionately exposed to avian predation. Because such costs of<br />
reproduction can be independent of fecundity (Bull & Shine 1979), high energy and<br />
survival costs are often associated with the extreme reproductive effort in capital<br />
breeders such as viperid snakes. <strong>The</strong> low survival rates of reproductive female aspic<br />
vipers are not affected by fecundity in natural conditions (Bonnet et al. 2002b), and<br />
females may optimize their reproductive effort by producing the greatest number of<br />
offspring per litter in order to minimize the cost paid per neonate. Capitalizing large<br />
amount of body reserves prior to reproduction is an elegant way to produce a<br />
massive reproductive effort when the probability of experiencing more than a single<br />
reproductive bout is low.<br />
Our study revealed another classical complication in studies of costs of<br />
reproduction. <strong>The</strong> expected underlying trade-off <strong>between</strong> current versus future<br />
reproduction (as evidenced by negative correlations <strong>between</strong> energy “costs” and<br />
future reproduction) was masked. <strong>The</strong> comparisons <strong>between</strong> reproductive and non-<br />
reproductive females, and the link <strong>between</strong> rates of mass recovery and future<br />
reproduction, provide strong evidence for the existence of costs. Why, then, are they<br />
not manifested in negative correlations <strong>between</strong> output and costs on the one hand,<br />
and future reproduction on the other?<br />
<strong>The</strong> answer almost certainly lies in substantial differences among females<br />
within our population in their ability (opportunities) to gather resources (Glazier 2000).<br />
For example, females with large initial body reserves produce larger litters and have<br />
a greater output relative to their own body sizes (Bonnet et al. 2001b), but they are<br />
153
nonetheless no less likely to breed again (this study) and they do not produce a<br />
relatively smaller litter at their second bout (this study). Despite their high<br />
reproductive expenditure, they do not recoup energy reserves more slowly after<br />
parturition, and thus, they eventually reproduce again without additional delay. That<br />
is, females that invest more (higher RCM) in their first litter were not less likely to<br />
survive (logistic regression with survival as the dependent variable and RCM<br />
[residuals] as the independent variable: χ² = 1.03, df = 1, n = 141, p = 0.31), and<br />
even showed a tendency to regain mass more rapidly (Figure 2), than the “less lucky”<br />
females who exhibited lower reproductive output (and thus, who superficially appear<br />
to have paid lower “potential costs”). This situation may reflect strong differences in<br />
female “quality” (as manifested in traits such as energy reserves prior to<br />
reproduction) as has been documented in other species (e.g., Van Noordwijk and de<br />
Jong 1986; Doughty & Shine 1997; Reznick et al. 2000). It may often be true that<br />
any given level of reproductive output is a greater “cost” (e.g., to survival) for a<br />
female in poor body condition (e.g., Cichon et al. 1998), and for variation in maternal<br />
quality to generate positive rather than negative correlations <strong>between</strong> reproductive<br />
output and survival (e.g., Bell & Koufopanou 1986; Winkel & Winkel 1995). We have<br />
no data on the determinants of female “quality”, but some correlates suggest that<br />
females with the highest reproductive effort (i.e. RCM) also exhibited the highest<br />
abilities to recover after parturition (Figure 2), and hence to reproduce again. This<br />
variation might reflect underlying genetic factors, or processes acting during<br />
ontogeny (e.g., developmental temperatures; feeding opportunities early in life; low<br />
parasite numbers). Alternatively, these inter-individual variations may simply reflect<br />
the fact that some females have been luckier than others during foraging prior to,<br />
during and after reproduction. Food <strong>intake</strong> can affect reproductive body reserves,<br />
154
eproductive success and recovery at each of these phases, and these effects can<br />
interact strongly (Bonnet et al. 2001b).<br />
Despite the masking of phenotypic trade-off by variations among females, our<br />
data nonetheless provide support for the use of changes in maternal body mass as a<br />
realistic currency in which to estimate “costs of reproduction” in female vipers. We<br />
base this conclusion on several facts. First, reproduction is expensive energetically:<br />
a female viper’s reproductive output is tightly linked to her energy stores prior to<br />
vitellogenesis (Bonnet et al. 2001b). Second, the rate at which a female can recoup<br />
her energy stores (expended during the previous reproduction) is a significant<br />
predictor of her future reproductive output (this study). <strong>The</strong> high growth rate<br />
observed in females that recouped their body reserves rapidly reflects the fact that<br />
females with high <strong>food</strong> availability invested both in body-reserve recovery and<br />
growth, and changes in body mass integrate these two effects. Third, female vipers<br />
postpone reproduction until they have achieved a critical threshold level of body<br />
condition (Naulleau & Bonnet 1996). Thus, the substantial variation among post-<br />
parturient females in their rates of recovery of body condition is convincing evidence<br />
that rates of gain in body mass offer a biologically meaningful currency in which to<br />
gauge a female’s ability to survive and reproduce. Moreover, the absolute mass that<br />
post-parturient females must regain to reach the threshold for reproduction fits well<br />
with the absolute mass loss during reproduction (Figure 1b and 2b). <strong>The</strong> other<br />
currencies that we examined appear to be less useful for measuring “costs of<br />
reproduction”. Survival is obviously important and this parameter must be included in<br />
analyses, but it is difficult to integrate with energetic measures in any single currency.<br />
Also, the real survival “cost” of producing a litter involves survival rates over the entire<br />
reproductive cycle (not just the year in which the litter is produced), whereas energy<br />
155
allocation can be calculated from the “reproductive” year only. Litter sizes are<br />
predictable from energy stores, as manifested in body condition.<br />
One major advantage of using changes in maternal body mass as a currency<br />
for reproductive effort and “potential costs,” is that such changes are easy to<br />
measure under field conditions. Also, reproductive output can be quantified in the<br />
same currency, so that the two measures of investment (litter mass plus maternal<br />
loss in body loss) can simply be added together to calculate a female’s <strong>total</strong><br />
expenditure on reproduction. This cannot be done if investment in aspects other<br />
than the litter is measured in other currencies. Thus, our study is encouraging in that<br />
a logistically feasible currency can be used to quantify a female’s investment into<br />
current reproduction, in terms that can be directly translated into effects on future<br />
reproductive success. This currency may well prove to be useful for other species as<br />
well, especially those in which maternal energy reserves fuel most of the reproductive<br />
expenditure (“capital breeders”: Drent & Daan 1980).<br />
Acknowledgements<br />
M. Vacher-Vallas, S. Duret, L. Patard, and M. Pedrono assisted in field work. For<br />
comments on the manuscript, we thank R. Cambag, S. Cucullatus, B. S. White and<br />
B. Wilmslow. Financial support was provided by the Conseil Général des deux<br />
Sèvres (XB), the Centre National de la Recherche Scientifique and the Australian<br />
Research Council.<br />
156
C. Article 5<br />
Costs of anorexia during pregnancy in a<br />
viviparous snake (Vipera aspis)<br />
Olivier Lourdais 1 2 3 , Xavier Bonnet 1 , Paul Doughty 4<br />
1 Centre d'Etudes Biologiques de Chizé, <strong>CNRS</strong>, 79360, Villiers en Bois, France<br />
2 Conseil Général Des Deux Sèvres, Rue de L’abreuvoir, 79021, Niort, France<br />
3 University of Poitiers, 40 avenue du recteur Pineau, 86022 Poitiers, France<br />
4 Univertisty of Canberra, ACT, Australia<br />
Published in Journal of Experimental Zoology 292, 487-493.<br />
(2002)<br />
157
Summary<br />
Spontaneous anorexia has been documented in various animal species and is<br />
usually associated with activities competing with <strong>food</strong> <strong>intake</strong>. In natural conditions,<br />
most female aspic vipers (Vipera aspis) stop feeding during the two months of<br />
pregnancy. We carried out a simple experiment on 40 pregnant females to determine<br />
whether anorexia was obligatory or facultative, and to investigate the energetic<br />
consequence of fasting on post-partum body condition and litter traits. Three diet<br />
treatments were applied during gestation: no <strong>food</strong>, one feeding occasion, and two<br />
feeding occasions. Twelve non-pregnant un-fed females were used as a control<br />
group. Most gravid females accepted captive mice during gestation, suggesting that<br />
anorexia reported in the field was a side effect of the tremendous changes in activity<br />
pattern associated with pregnancy. Mass loss was high for un-fed reproductive<br />
females, indicating high-energy expenditure associated with embryo maintenance.<br />
Prey consumption allowed compensation for metabolic expenditure and enhanced<br />
post-partum female body condition, but had no effects on litter characteristics. <strong>The</strong><br />
magnitude of the metabolic expenditure during gestation appeared to be independent<br />
of fecundity.<br />
Keywords: snakes, costs of reproduction, life history trade-offs<br />
158
Introduction<br />
Life history theory has been largely influenced by the concept of cost of reproduction<br />
based on a possible trade-off <strong>between</strong> current reproduction and future reproductive<br />
success (Fischer 1930; Williams 1966b). This notion is supported by a substantial<br />
amount of theoretical study suggesting that the form of the <strong>relationship</strong> <strong>between</strong><br />
reproductive investment and the magnitude of associated costs influences the<br />
evolution of reproductive strategies (Williams 1966b; Bull & Shine 1979; Shine &<br />
Schwarzkopf 1992).<br />
Identification of proximate mechanisms by which costs are mediated is an<br />
important empirical challenge and two major categories of reproductive costs are<br />
classically distinguished (Calow 1979). <strong>The</strong> first component is ecological and linked<br />
with a reduction of survival probabilities associated with reproduction. <strong>The</strong> second<br />
component recognised as a “fecundity” cost involves an energy allocation trade-off:<br />
the investment in current reproduction affects the residual reproductive value through<br />
a depletion of body reserves or growth rate reduction (Williams 1966b; Shine 1980).<br />
Such a classification is somewhat artificial as the two major forms of costs are<br />
interconnected; for example, low body reserves may also affect survival. In addition,<br />
there is growing evidence that organisms can change the relative magnitude of the<br />
different components through behavioural modifications (Bauwens & Thoen 1981;<br />
Brodie 1989). Costs based on allocation trade-offs can take a diversity of forms.<br />
Most studies have been carried out on the effects of direct energy investment into<br />
reproduction (Anguiletta & Sears 2000). However, substantial expenditure may also<br />
arise indirectly from cessation or reduction of feeding during all or part of<br />
reproduction. Empirical studies indicate that feeding cessation or reduction during<br />
reproduction is a widespread phenomenon (Engelmann & Rau 1965; Batholomew<br />
159
1970; Mrosovsky & Sherry 1980; Weeks 1996). Cessation of feeding when <strong>food</strong> is<br />
available is superficially paradoxical and needs explanation. Furthermore, this<br />
indirect component of reproductive effort may translate into a substantial energetic<br />
cost of reproduction.<br />
Anorexia is associated with hibernation, migration or incubation in various<br />
endotherm vertebrates (Mrosovsky & Sherry 1980). Although less studied,<br />
ectotherm vertebrates may also serve as a a good model group for the study of<br />
feeding cessation. Endothermy is tightly linked with parental cares (Farmer 2000),<br />
and the lack of such cares among numerous ectotherm species lead to a substantial<br />
simplification as reproductive effort is sealed prior to oviposition or parturition. Such<br />
situation will facilitate the assessment of the <strong>relationship</strong> <strong>between</strong> reproductive effort,<br />
reproductive output (fecundity) and associated costs. Among squamate reptiles<br />
(lizards and snakes), reproduction entails major behavioural changes, notably<br />
decrease or cessation of <strong>food</strong> <strong>intake</strong> during gestation (Shine 1980; Madsen and<br />
Shine 1993; Gregory & Skebo 1998; Gregory et al. 1999). Such phenomena are<br />
particularly obvious for “capital breeding” species in which long term energy storage<br />
constitutes the primary source of energy for reproduction (Bonnet et al. 1998).<br />
<strong>The</strong> aspic viper (Vipera aspis) is a medium size (50 cm) viperid snake that<br />
displays those characteristics. In females, energy stores permit them to fuel the<br />
entire energetic requirements of reproduction. Depending upon prey availability, <strong>food</strong><br />
<strong>intake</strong> may occur during the egg production phase in spring (Saint Girons & Naulleau<br />
1981; Bonnet et al. 2001b). However, field data clearly indicate that many females<br />
virtually stop feeding during the two months of pregnancy. Three hypotheses about<br />
the proximate factors involved in gestational anorexia can be proposed: 1) anorexia<br />
could be a consequence of abdominal space limitation to accomodate both embryos<br />
and prey items (Saint Girons 1979). 2) cessation of feeding may be related to a loss<br />
160
of appetite intrinsically associated with gestation (i. e. due to changes in hormonal<br />
balance; Bonnet et al. 2001b). 3) fasting may simply be the result of low foraging<br />
success due to behavioural changes in gravid individuals (thermal needs, predator<br />
avoidance).<br />
For hypothesis 1 and 2, gestational anorexia is supposed to be obligatory. In the<br />
case of hypothesis 3, this phenomenon is expected to be facultative. In the present<br />
study, we conducted a simple experiment to test if gestational anorexia is obligatory<br />
or facultative and to examine to what extent feeding cessation during pregnancy<br />
translated into energetic costs.<br />
Materials and Methods<br />
Study species<br />
<strong>The</strong> aspic viper (V. aspis) is a small viviparous snake, abundant in central western<br />
France. In this area, females typically reproduce on a less-than-annual schedule<br />
(Saint Girons 1957a,b; Bonnet & Naulleau 1996; Naulleau & Bonnet 1996; Naulleau<br />
et al. 1999). Ovulation occurs during the first two weeks of June (Naulleau 1981),<br />
and parturition occurs two to three months later, from late August to late September.<br />
Captures and housing<br />
Forty reproductive females were collected in June 2000 from three localities: Château<br />
d’Olonnes, Les Sables d’Olonnes (both in Vendée district) and Rochefort (Charentes<br />
Maritimes district). Individuals were given a unique scale clip number, measured to<br />
the nearest 0.5 cm and weighed to the nearest 1g. Females were placed in six<br />
outdoor enclosures (5 X 3 m, mean density: 5 snakes/enclosure) broadly recreating<br />
the natural habitat and exposed to the climatic conditions of the field research station<br />
of Chizé (Forêt de Chizé, Deux-Sèvres, 46°07’ N, 00°25’ W). Each enclosure was<br />
161
equipped with numerous external dens to serve as hiding-places. Water was<br />
provided ad libitum and vegetation mainly composed of annual Poacae was kept high<br />
(20 - 40 cm) to provide shade and shelter.<br />
Experimental design<br />
Females were randomly assigned to one of the three feeding treatments during<br />
gestation: Group 1 (9 individuals): never fed; Group 2 (17 individuals): one prey item<br />
offered in July ; Group 3 (14 individuals): two prey were presented in July and early<br />
August. Snakes of both feeding treatment groups were fed by placing a recently<br />
killed mouse (average mass 20 g) close to their dens. Prey consumption was<br />
recorded by direct observation of feeding, or by less direct means if feeding was not<br />
observed (by palpation of mice inside the snake and by a sudden increase in body<br />
mass).<br />
As a control group, 12 non-pregnant females (un-fed during one month in the same<br />
conditions to reproductive females) were weighed to measure mass loss during<br />
fasting, independently of gestation.<br />
Records of body mass and reproductive output<br />
<strong>The</strong> snakes were all weighed at the onset of the experiment in early July (i.e. after<br />
ovulation). At this time the number of eggs was assessed via abdominal palpation<br />
(Fitch 1987, Bonnet et al. 2001b for further details on the acuracy of the method).<br />
Females were all recaptured at the end of gestation (late August); and weighed again<br />
to determine absolute mass changes <strong>between</strong> early July and late August. Daily mass<br />
change during gestation was calculated using absolute mass change (g) and time<br />
elapsed (days) <strong>between</strong> the two mass records. Snakes were then brought in the<br />
laboratory until parturition. We recorded post-partum female body mass and the<br />
162
number, mass (± 0.1 g) and length (± 0.5 cm) of healthy offspring. <strong>The</strong> number of<br />
unfertilised eggs and stillborn were also recorded. We made a distinction <strong>between</strong><br />
the <strong>total</strong> litter size including healthy neonates, still born offspring and undeveloped<br />
eggs as well (Farr & Gregory 1991; Gregory et al. 1992), and “fit” litter size where<br />
only viable neonates were considered. Five females (one in group 2 and two each in<br />
group 1 and 3) produced only unfertile eggs. Because snakes were caught after the<br />
mating season (Naulleau 1997), those individuals were removed from analysis of<br />
reproductive output.<br />
In this species, post-partum female body condition (mass adjusted by size)<br />
positively influence survival the year following reproduction and hence residual<br />
reproductive success (Bonnet et al. 2000a). We calculated body condition as the<br />
residual score from the general linear regression of log-transformed body mass value<br />
versus log-transformed snout-vent length value for all females (Jayne & Benett 1990,<br />
Bonnet et al. 2000a). Such index provides an accurate estimation of body reserves<br />
(Bonnet 1996).<br />
Results<br />
Prey consumption<br />
Pregnant females accepted prey most of the time (41 of 45 feeding occasions).<br />
Among the 17 snakes fed once, 15 (88%) ate the prey offered, and among the 14<br />
snakes fed twice, 12 (86%) ate both prey offered; two females ate only one mouse<br />
each. Hence, at the end of gestation it was possible to classify females by the actual<br />
number of prey eaten: no prey (11 individuals); one prey consumed (17 individuals);<br />
two prey consumed (12 individuals). In the following analysis, we considered both<br />
group treatments and actual number of prey eaten.<br />
163
Changes in body mass during gestation<br />
<strong>The</strong> three groups did not differ in snout-vent length (one factor ANOVA,<br />
F(2,37)=1.74; p=0.20, size-adjusted initial body mass (ANCOVA F(2,36)=1.39;<br />
p=0.26), or size-adjusted number of eggs (ANCOVA, F(2,36)=0.45; p=0.63; Table<br />
1). Group 1 females showed a significantly higher daily mass loss during gestation in<br />
comparison to the control group (0. 22 g/d versus 0.11 g/d; F(1,18)=12.58;<br />
p
Hence, using either initial group treatments or the number of prey actually consumed<br />
led to similar results. In the following analysis, we only present calculations based on<br />
the number of prey ingested during pregnancy, because it should more directly bear<br />
upon the influence of energy <strong>intake</strong> on reproductive output and post partum female<br />
body condition.<br />
Changes in body mass (grams per days)<br />
0.1<br />
0.0<br />
-0.1<br />
-0.2<br />
-0.3<br />
-0.4<br />
control group<br />
un-fed NR<br />
females<br />
p < 0.002<br />
no prey one prey<br />
two prey<br />
Figure 1. Effects of diet on female daily mass change (in grams per day scaled with initial body mass)<br />
during the course of gestation. Error bars represent standard error. See text for statistics.<br />
Litter characteristics and females post partum condition<br />
<strong>The</strong> number of prey consumed during pregnancy did not influence litter size<br />
(ANCOVA, F(2,32)= 0.21; p=0.81), litter mass (F(2,32)=1.67; p=0.23), fit litter size<br />
(F(2,32)=1.39; p=0.26) and fit litter mass (F(2,32)=1.84; p=0.21). Similarly, no<br />
difference in mean offspring snout vent length (ANCOVA, F(2,32)=0.57; p=0.57) or<br />
mean offspring mass (ANCOVA, F(2,32)=0.55; p=0.58) were detected (Table 2).<br />
165
However, females eating twice were in higher body condition than females fasting or<br />
eating only one prey (ANOVA, F(2,32)=5.54; p
Body mass (grams)<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
40<br />
O<br />
July 1<br />
(early gestation)<br />
G P<br />
August 31<br />
(late gestation)<br />
no prey<br />
one prey<br />
two prey<br />
September 15<br />
(post parturition)<br />
Figure 3. Influence of diet on pattern of mass changes during the experiment. Symbols represent<br />
size-adjusted body mass ± 1 standard error. (O: ovulation; G: gestation; P : parturition).<br />
Discussion<br />
Comparisons were made in the feeding behavior among three groups of pregnant<br />
vipers and with a control group of non pregnant snakes. Pregnant females accepted<br />
prey most of the time (41 of 45 feeding occasions). This observation clearly<br />
invalidates the hypothesis of an intrinsic (physiological or anatomical) origin of<br />
anorexia during pregnancy. Dramatic reduction of <strong>food</strong> <strong>intake</strong> reported in the field is<br />
thus a facultative phenomenon and appears to be a consequence of modification in<br />
the activity pattern. In female viviparous snakes, important behavioural changes are<br />
associated with gestation (increase in basking rate, physical burden, Birchard et al.<br />
167
1984; Seigel et al. 1987; Brodie 1989). In female aspic vipers, gestation is<br />
accompanied by a drastic reduction of the home range of monitored females from<br />
thirty to a few square meters (Naulleau et al. 1996). Pregnant females also adopt<br />
higher thermal preferences and substantially increase basking time (Saint Girons<br />
1952; Bonnet & Naulleau 1996). <strong>The</strong>rmal conditions are important to optimise<br />
embryonic developmental speed along with many offspring traits in squamate reptiles<br />
(Fox et al. 1961; Packard & Packard 1988; Shine et al. 1997). If changes in gravid<br />
female thermal preferences optimise developmental rates of the embryos,<br />
interference with other activities may also occur. Most notably, the time devoted to<br />
thermoregulation will trade off with the time spent foraging. In this study, we<br />
facilitated feeding behaviour by placing prey very close to the snakes. In support of<br />
this, anecdotal cases of gravid females with a prey in the stomach even in late<br />
gestation have been reported in the field (Naulleau 1997). Those females were<br />
extremely sedentary (i.e. not engaged in foraging activities) as indicated by<br />
extensive radio tracking data (Naulleau et al. 1996) and thus were probably “lucky” in<br />
catching voles passing very close to the basking site. <strong>The</strong> same situation has been<br />
documented in a closely related species, the adder (Vipera berus, Madsen & Shine<br />
1992a).<br />
As <strong>food</strong> was accepted, it was possible to explore the effects of the different<br />
diets. Food <strong>intake</strong> significantly affected the pattern of mass change during gestation<br />
(Fig 1). <strong>The</strong> higher body mass loss detected among un-fed pregnant females in<br />
comparison with the control group suggests that metabolic expenditure associated<br />
with embryonic development is high. Data on fasting pregnant females allowed us to<br />
assess the <strong>relationship</strong> <strong>between</strong> daily mass loss during pregnancy (i.e. metabolic<br />
expenditure) and reproductive output characteristics. Interestingly, the magnitude of<br />
mass loss was not correlated with <strong>total</strong> litter size (n=9, r=0.06, F(1,7)=0.02, p
or more importantly fit litter size, (n=9, r=0.27, F(1,7)=0.56, p
current level of <strong>food</strong> availability notably by marked years to years variations in prey<br />
(voles) density reported in the field (Delattre et al. 1992; Bonnet et al. 2001b).<br />
Despite the limited sample size (n=9), the fecundity-independent nature of<br />
metabolic cost associated with pregnancy agrees with results gathered by Madsen &<br />
Shine (1993), who reported significant fixed costs in the adder (Vipera berus). Such<br />
fecundity-independent costs are probably a widespread phenomenon influencing the<br />
evolution of life history strategies (Bull & Shine 1979). In the case of the aspic viper,<br />
the integration of those particular energetic costs will help to further understand the<br />
“all or nothing” system of energy allocation displayed by this species (Bonnet and<br />
Naulleau 1996; Bonnet et al. 2002a).<br />
Acknowledgements<br />
We thank Gwenaël Beauplet, Dale DeNardo, Olivier Dehorther and Yves Cherel for<br />
constructive comments on the manuscript. We are grateful to Patrice Quistinic and<br />
Michaël Guillon for help in snake collecting, and Hélène Blanchard, Günter and Felicy<br />
for picnic organisation. Manuscript preparation was supported by the Conseil<br />
Régional de Poitou-Charentes. Special thanks to Melle, notably for the numerous<br />
Italian coffees. Finally, Rex Cambag provided his own inimitable commentary.<br />
170
D. Article 6<br />
Gestation, thermoregulation and metabolism in<br />
a viviparous snake, Vipera aspis: evidence for<br />
fecundity-independent costs<br />
Mitchell Ladyman 1 , Xavier Bonnet 2 , Olivier Lourdais 2 3 4 , Don Bradshaw 1 and<br />
Guy Naulleau 2<br />
1 Department of Zoology, University of Western Australia, Perth, WA 6009<br />
2 Centre d'Etudes Biologiques de Chizé, <strong>CNRS</strong>, 79360, Villiers en Bois, France<br />
3 Conseil Général Des Deux Sèvres, Rue de L’abreuvoir, 79021, Niort, France<br />
4 University of Poitiers, 40 avenue du recteur Pineau, 86022 Poitiers, France<br />
Accepted for publication in Physiological and Biochemical Zoology<br />
171
Abstract<br />
Oxygen consumption of gestating Aspic vipers, Viper aspis (L.), was strongly<br />
dependent on body temperature and mass. Temperature-controlled, mass-<br />
independent oxygen consumption did not differ <strong>between</strong> pregnant and non-pregnant<br />
females. Maternal metabolism was not influenced during early gestation by the<br />
number of embryos carried, but was weakly influenced during late gestation. <strong>The</strong>se<br />
results differ from previous investigations that show an increase in mass-independent<br />
oxygen consumption in reproductive females relative to non-reproductive females<br />
and a positive <strong>relationship</strong> <strong>between</strong> metabolism and litter size. <strong>The</strong>se data also<br />
conflict with published field data on Vipera aspis that show a strong metabolic cost<br />
associated with reproduction. We propose that, under controlled conditions (i.e.<br />
females exposed to precise ambient temperatures), following the mobilisation of<br />
resources to create follicles (i.e. vitellogenesis), early gestation per se may not be an<br />
energetically expensive period in reproduction. Under natural conditions, however,<br />
the metabolic rate of reproductive females is strongly increased by a shift in thermal<br />
ecology (higher body temperature and longer basking periods), enabling pregnant<br />
females to accelerate the process of gestation. Combining both laboratory and field<br />
investigation in a viviparous snake, we suggest that reproduction entails discrete<br />
changes in the thermal ecology of females to provide optimal temperatures to the<br />
embryos, whatever their number. This results in the counterintuitive notion that<br />
metabolism may well be largely independent of fecundity during gestation, at least in<br />
an ectothermic reptile.<br />
172
Introduction<br />
Reproduction represents a major disruption to the typical day-to-day life of any<br />
female organism. <strong>The</strong> decision to reproduce, or to do so successfully, is often<br />
strongly resource orientated; mediated by <strong>food</strong> availability (Nagy et al. 1984, 1995;<br />
Boggs 1992; Meijer & Drent 1999), by body condition (Larsen 1980; Anderson &<br />
Karasov 1981; Nagy 1987; Brown 1991; Naulleau & Bonnet 1996; Meijer & Drent<br />
1999), or both (Meijer & Drent 1999; Bonnet et al. 2001b). For example, endogenous<br />
reserve levels can determine the initiation of vitellogenesis (Bonnet et al. 1994), or<br />
reproductive effort and associated costs (Madsen & Shine 1993; Erikstad et al. 1997;<br />
Festa-Bianchet et al. 1998; Bonnet et al. 2001b). <strong>The</strong> amounts of body reserves can<br />
also influence the maintenance of brooding (Chastel et al. 1995; Dearbon 2001),<br />
gestation (Boyd 1984; Cresswell et al. 1992), the intensity of parental care (Olsson<br />
1997), and hence offspring number; or can even affect the sex of the offspring.<br />
In many vertebrates, reproduction occurs when an animal is in positive energy<br />
balance as folliculogenesis maintains a low priority in the ongoing competition for<br />
energy allocation (Bronson 1998; Schneider et al. 2000). Any attempt to reproduce<br />
during periods of resource deficiency may potentially decrease both current and<br />
future reproductive success (Stearns 1992). <strong>The</strong> association <strong>between</strong> reproduction<br />
and resources is very intimate because reproduction is an energetically demanding<br />
process. An analysis of the available literature suggests that there is a clear<br />
difference <strong>between</strong> the metabolic rate of reproductive and non-reproductive<br />
individuals (Birchard et al. 1984; Speakman & McQueenie 1996; Mauget et al. 1997;<br />
Angilletta & Sears 2000). This conclusion is based on comparisons of rates of<br />
metabolism measured during discrete periods of the reproductive cycle (see Guillette<br />
1982). However, to generalise such a broad divergence in metabolism across the<br />
173
whole reproductive process could be slightly hazardous given that this process is<br />
clearly not static. For example, the physiological and behavioural mechanisms that<br />
underlie folliculogenesis, gestation, and lactation are extremely different (Thibault &<br />
Levasseur, 1991; Speakman & McQueenie 1996). Even among males, where the<br />
energetic contribution to gametic production is almost negligible in most species,<br />
metabolic rate varies with pubertal maturity (Brown et al. 1996) or with the alternation<br />
of mating and non-mating periods (Olsson et al. 1997). Moreover, the ecological<br />
implications of changes in metabolism over the reproductive period are important<br />
considering the fact that <strong>food</strong> resource and other environmental conditions<br />
(temperature, rainfall etc.), are often limited and/or often fluctuate.<br />
Whatever the taxon or reproductive mode (i.e. viviparity versus oviparity), the<br />
<strong>total</strong> metabolic effort expended over time is generally higher in reproductive females<br />
than non-reproductive females during the reproductive period. Limiting the focus to<br />
gestation, and hence considering viviparous species only, the extra metabolic<br />
demands associated with the developing foetus are higher than the normal metabolic<br />
demands for growth and repair (Hahn & Tinkle 1965; Mauget et al. 1997). Such a<br />
difference is detectable even in the late luteal phase of the menstrual cycle (Curtis<br />
1996). In mammals, this greater metabolic expenditure is often driven by a change in<br />
the hormonal balance of the reproducing individual that, in turn, provokes an increase<br />
in their basal metabolic rate through increased levels of cellular work (protein<br />
synthesis, mitosis, ion pumping etc.) over the reproductive period (Thibault and<br />
Levasseur 1991; Howe et al. 1993; Butte et al. 1999).<br />
Physiological changes also occur in viviparous ectotherms to cope with the<br />
demand of developing embryos. Birchard et al. (1982), and other authors, describe<br />
an increase in mass-independent oxygen consumption in reproductive females<br />
suggesting a modification of metabolic regulation set point, and this may also be<br />
174
associated with hormonal changes (Highfill & Mead 1975; Fergusson and Bradshaw<br />
1991; Callard et al. 1992). However, ectothermic vertebrates are limited in the extent<br />
to which the can elevate the rate of cellular work and therefore their metabolic rate<br />
through purely physiological mechanisms (Harlow & Grigg 1984; Shine et al. 1997;<br />
Cadenas et al. 2000; Wang et al. 2001). Instead, ectotherms may depend on the<br />
environment to compensate for such a physiological constraint (Bradshaw 1997),<br />
using available thermal energy (heat) to increase metabolic reaction rate, and<br />
increase development of embryos. It is a common observation that in viviparous<br />
reptiles living in temperate areas, gravid females bask for longer periods than non-<br />
gravid females (Gregory et al. 1987; Seigel & Ford 1987; Bonnet & Naulleau 1996;<br />
Shine 1998) with the accelerated development of embryos occurring as a result<br />
(Naulleau 1986). Changes in body temperature of pregnant female ectotherms is the<br />
most effective way to increase the rates of cellular activity involved with production of<br />
young and we may expect that this increase is proportional to the load of active<br />
tissues represented by the embryos. Such changes will involve a change in<br />
thermoregulatory behaviour. In addition, if we admit that variation in female metabolic<br />
rate is an integrative effect of the cellular work of the tissues of the mother plus those<br />
of developing embryos, we may expect a positive <strong>relationship</strong> <strong>between</strong> the size (and<br />
mass) of the litter and maternal mass-independent oxygen consumption. Such a<br />
<strong>relationship</strong> has been previously documented in viviparous reptiles, but the number of<br />
studies has been limited (Birchard et al. 1984; Demarco & Guillette 1992).<br />
Importantly, the form of the <strong>relationship</strong> <strong>between</strong> fecundity (litter size),<br />
reproductive effort (mass-independent oxygen consumption, materials invested into<br />
follicles, loss of feeding opportunities etc...), and the potential costs (decreased<br />
reproductive value, here simply viewed as a combination of lower survival and<br />
depletion of energy stores) has immense consequences on the evolution of life<br />
175
history traits, and on the underlying physiological regulations (Shine & Schwarzkopf<br />
1992; Stearns 1992; Niewiarowski & Dunham 1994; Sinervo & Svensson 1998).<br />
Clearly both empirical and experimental data are required in this field. For instance,<br />
increasing our knowledge in the costs/benefits <strong>relationship</strong> of pregnancy is a major<br />
prerequisite to better understanding the oviparity versus viviparity transition that has<br />
evolved independently several hundreds of times in many vertebrate taxa (fishes,<br />
amphibians, squamate reptiles; Shine 1985).<br />
Snakes are suitable models for investigations such as this, as there are<br />
oviparous and viviparous species in the same family group (Shine 1985). Because<br />
they are ectothermic, we can also experimentally control body temperature and<br />
assess the impact of this factor on metabolism in both reproductive and non-<br />
reproductive individuals (Beaupré & Duvall 1998). In addition, the broad range of<br />
fecundity among conspecific females allows us to identify any potential<br />
fecundity/reproductive effort/cost <strong>relationship</strong>s. Finally, many snake species have a<br />
less than annual breeding frequency (Saint Girons 1957b; Seigel & Ford 1987).<br />
<strong>The</strong>refore within any given year, both reproductive and non-reproductive individuals<br />
are influenced by the same variation of climate and resource availability making the<br />
two groups more comparable.<br />
In this investigation we examined the influence of the reproductive status and<br />
fecundity on the metabolic rate of viviparous female snakes. Notably, under the same<br />
thermal conditions, are pregnant females more metabolically active than non-<br />
reproductive females over the gestative period? As an aside, we demonstrate that<br />
broad generalisations on the effects of reproductive status on metabolism, based on<br />
laboratory experiments over discrete periods, may potentially limit the value of<br />
conclusions in field conditions.<br />
176
Materials and Methods<br />
Origin and Description of Study Animals<br />
In this study we used wild caught gravid female Aspic vipers (Vipera aspis). We used<br />
this species because this snake is abundant in our study area, tolerant to captivity<br />
and manipulation, and previous work provides a useful baseline on the reproductive<br />
ecology and physiology of this species (Saint Girons 1957b; Saint Girons & Duguy<br />
1992; Bonnet et al. 1994, 2000a,b, 2001a, b, 2002a, b; Aubret et al. 2002; Lourdais<br />
et al. 2002a, b; and references therein). <strong>The</strong> Aspic viper is a stocky, venomous snake<br />
species widely distributed throughout Western Europe (Naulleau 1997). Average<br />
snout-vent length (SVL) is 48.5 cm and body mass (BM) is 85.5 g. Females mature at<br />
approximately three years (Bonnet et al. 1999a), mating occurs in spring (March –<br />
April)(Saint Girons 1957b; Naulleau et al. 1999), and parturition occurs in autumn<br />
(September; Saint Girons 1957b). Clutch size varies from 1 -13 individuals (mean SVL<br />
= 17.9±1.2 cm, mean BM = 6.3±1.1 g)(Bonnet et al. 2000b). Females in western<br />
France do not breed every year (Saint Girons 1957b; Bonnet & Naulleau 1996) and<br />
delay reproduction until they exceed a minimum body condition threshold through the<br />
accumulation of large body reserves such as abdominal fat bodies (Naulleau &<br />
Bonnet 1996; Aubret et al. 2002).<br />
Individuals used in this investigation were collected by hand from Les<br />
Moutiers-en-Retz and Les Sables d’Olonne, central western France (47 o 03’ N; 02 o<br />
00’ W and 46° 30’ N; 01° 44’ W respectively). Both study sites lie within 60 km of<br />
each other. Habitats were consistent <strong>between</strong> the two study sites and the climate for<br />
the region is a temperate oceanic one (see Bonnet & Naulleau, 1993, for average<br />
temperatures).<br />
177
Specimens used for the metabolic tests were collected during early spring,<br />
after the mating period and during vitellogenesis (Saint Girons 1957b). At this time,<br />
reproductive females had already committed to the development of a number of<br />
follicles. Individuals were housed in outdoor terraria (8 – 16 m²) in our laboratory (46°<br />
08’ N; 00° 25’ W), from June to late September, and exposed to a similar climate to<br />
the field site from where they were collected. Within each enclosure snakes were<br />
provided with a mosaic of microhabitats, including many shelters and well-exposed<br />
sites to facilitate optimal thermoregulation. A unique code of ventral scale clipping<br />
identified individuals and these marks are permanent as the regenerated tissues<br />
exhibit a different colour.<br />
Resting Oxygen Consumption<br />
<strong>The</strong> oxygen consumption of 50 reproductive females and 19 non-reproductive<br />
females was measured in the laboratory under controlled environmental conditions.<br />
Subjects were tested for rates of oxygen consumption over two phases during the<br />
survey period: early gestation and late gestation. A <strong>total</strong> of 59 snakes was tested<br />
during early gestation (40 reproductive and 19 non-reproductive) and a subset of 17<br />
reproductive snakes (among the 40 reproductive females) was used for repeated-<br />
measures analysis in late gestation. An additional 12 reproductive snakes were<br />
measured in late gestation at 32 o C to increase the statistical power of the data set<br />
used to derive <strong>relationship</strong> with fecundity. Table 1 summarises the number of females<br />
sampled and re-sampled at different temperature regimes and/or at the two periods<br />
of gestation.<br />
178
Table 1: Details on the number of female Aspic vipers assayed for O2 consumption. <strong>The</strong> first number<br />
provides the <strong>total</strong> number of females used in each treatment group. <strong>The</strong> number in brackets refers to<br />
females used in repeated measures experiment over early and late gestation.<br />
Reproductive Status Gestation period 17.5 o C 25 o C 32.5 o C<br />
Reproductive Early 14 13 13<br />
Non-reproductive Early 6 5 8<br />
Reproductive Late 7 (7) None 22 (10)<br />
Rates of oxygen consumption (VO2 - mL O2 g -1 h -1 ) were measured using a<br />
flow-through respirometry system. Dry incurrent air was drawn through a small, clear<br />
Perspex metabolic chamber at a rate of 204+4 mL min -1 by a Byoblock Scientific 6 L<br />
air pump and flow was controlled using a Platon mass flow controller. <strong>The</strong> chamber,<br />
specifically built to accommodate Aspic vipers, was large enough (internal diameter:<br />
15 x 15 x 5 cm) to accommodate snakes up to 200 g, without preventing voluntary<br />
activity. Oxygen concentration was maintained at approximately 20.1 %. <strong>The</strong><br />
metabolic chamber was located within a sealed Cryosystem temperature-controlled<br />
chamber and positioned such that snakes could be observed during the trial through<br />
a small viewing port. During early-gestation the body temperature of each individual<br />
was checked prior to commencement of the test using a quick-registering<br />
thermometer (Novo) inserted 2 cm into the cloaca until stabilisation of the measure.<br />
Excurrent air was passed through two column desiccators containing drierite,<br />
then through a paramagnetic O2 transducer (Servomex Series 1100). <strong>The</strong> differential<br />
output of the oxygen analyser (ambient air minus excurrent air) was recorded,<br />
adjusted to standard temperature and pressure conditions and plotted on a standard<br />
desktop PC. <strong>The</strong> metabolic chamber was calibrated to the outside atmosphere<br />
(Pressure Indicator Druck DPI 260) and set at zero oxygen consumption by running<br />
179
an empty chamber for one hour prior to each snake being tested. Differential output<br />
was presented graphically as it was acquired and snakes were run for as long as<br />
necessary to obtain stable oxygen consumption for a period greater than half of an<br />
hour. Monitoring of snakes ensured that fluctuations in VO2 were attributed to activity<br />
and not technical perturbations. Oxygen consumption was calculated following the<br />
equation of Decopas & Hart (1957).<br />
Reproductive and non-reproductive snakes were run at 17.5 o C, 25 o C and<br />
32.5 o C degrees during early gestation, and reproductive snakes only were run at<br />
17.5 o C and 32.5 o C during late gestation. This corresponds to the range of<br />
temperatures we have recorded, using radio telemetry, in the field in females<br />
(reproductive and non-reproductive) during the pregnancy period (Naulleau et al.<br />
1996; unpublished data). Before measurements of oxygen consumption, animals<br />
were fasted for a minimum of four days and introduced to the system on the night<br />
prior to testing. Because the Aspic viper is diurnal, all experiments were conducted<br />
during the normal light phase appropriate for the time of the year and location<br />
(approximately 0600 to 1900 hours). Snakes of both reproductive statuses were<br />
tested randomly throughout the day to remove any possible influence of diel cycle.<br />
Body mass of snakes was recorded on a top-loading electronic scale (±0.1 g) before<br />
each run, and the measurements were used to derive mass-independent volume of<br />
oxygen consumed (mass-independent VO2).<br />
Measurements of Fecundity<br />
Fecundity was determined by palpation in early gestation and, in this species, follicles<br />
as small as 2.0 g can be detected. Fecundity was confirmed at parturition. Live and<br />
stillborn neonates were included in the analysis as metabolically active tissues during<br />
pregnancy. Unfertile and/or undeveloped eggs (i.e. where only yolk was identifiable)<br />
180
that are frequent in Aspic viper’s litters (Bonnet et al. 2001b) were not considered as<br />
metabolically active tissues during pregnancy as these undeveloped eggs are formed<br />
during vitellogenesis, but not pregnancy.<br />
Field Body Temperatures<br />
We collected field data to compare the temperature regimes imposed on snakes in<br />
our laboratory experiment with the far more complex situation experienced by wild<br />
vipers. Using internal temperature radio transmitters, we determined the body<br />
temperature of reproductive and non-reproductive females in the field during most of<br />
the period of reproduction (15 th April 1996 to the 15 th July 1996), encompassing the<br />
45 last days of vitellogenesis and the 45 first days of gestation. <strong>The</strong> methodology has<br />
been described elsewhere (Naulleau et al. 1996), and has also been used<br />
successfully in a closely related species, the adder (Vipera berus)(Madsen & Shine<br />
1992a; 1993).<br />
Radio tracking enabled the collection of over 1896 temperature records from<br />
21 female Aspic vipers (11 reproductive and 10 non-reproductive). <strong>The</strong> temperatures<br />
were collected, without disturbing the snake, during the day. Each animal was<br />
sampled one to three times per day: in the morning, at mid-day and in the afternoon<br />
(mean number of records per snake per day was 2.33±0.84, range 1-6 with 99.9 % of<br />
the cases comprised <strong>between</strong> 1 and 4 and 97 % below 4). Because we sampled<br />
each snake randomly irrespective of the reproductive status, the exact time elapsed<br />
<strong>between</strong> two consecutive records was also random. However, it was always greater<br />
than 3 hours and often greater than 12 hours. <strong>The</strong> duration <strong>between</strong> samples<br />
ensures that consecutive records will reflect the thermal behaviour adopted by each<br />
snake rather than the thermal inertia of the body of the snake. As body temperature<br />
is affected by ambient temperature, and ambient temperature increased over our<br />
181
sampling period, we took into account the effect of date in most analysis involving<br />
field body temperatures.<br />
Statistical Analysis<br />
Body mass (BM-grams) and oxygen consumption (VO2 – mL O2 h -1 ) were log10<br />
transformed to meet the normality assumption (Shapiro-Wilk W = 0.986, P = 0.731 for<br />
Log – BM and Shapiro-Wilk W = 0.972, P = 0.158 for Log – VO2)(Zar 1984). Oxygen<br />
consumption data were adjusted for body mass by regressing Log - VO2 on Log -<br />
BM. <strong>The</strong> residuals from this regression yielded mass-independent oxygen<br />
consumption (mass-independent VO2), which was used for several analysis (see<br />
results). We did not use ratios to scale oxygen consumption (Atchley et al. 1976;<br />
Packard and Boardman 1988). Instead we performed ANCOVAs, for example, to<br />
compare reproductive versus non-reproductive females using BM as a covariate<br />
(Garcia-Berthou 2001). Statistical analysis of VO2 was performed on the mean rate of<br />
oxygen consumption measured over a stable half hour period for each individual.<br />
Analyses were performed using Statistica 5.1 and 6.0 (Statsoft 1995, 2001).<br />
Results<br />
Reproductive Status, Body Size and Body Condition in Early Gestation<br />
At the beginning of gestation, there was no difference in SVL <strong>between</strong> reproductive<br />
and non-reproductive females (one factor ANOVA with SVL as the dependent<br />
variable and reproductive status as the factor: F(1,56)=2.74, p=0.10; Table 2).<br />
However, reproductive females were significantly heavier than non-reproductive<br />
females (same design ANOVA with BM as the dependent variable: F(1,56) = 17.16,<br />
p< 0.001; Table 2), and, logically were in better body condition than non-reproductive<br />
182
snakes (ANCOVA with Log - BM as the dependent variable and Log - SVL as the<br />
covariate: F(1,55) = 34.12, p< 0.001; Table 2). Importantly, at the beginning of<br />
gestation, the greater body condition of reproductive females relative to non-<br />
reproductive females does not mean that they possess larger body reserves. In fact,<br />
most of the extra-mass of reproductive females is represented by the litter and,<br />
despite their external appearance, pregnant females are relatively emaciated during<br />
early gestation (Bonnet et al. 2002a).<br />
Table 2: Morphometrics of 40 reproductive and 19 non-reproductive female Aspic vipers used during<br />
early-gestation. Means are expressed ± SD. Comparing reproductive versus non-reproductive<br />
females, adjusted body mass (scaled by size using SVL as a covariate) was greater in pregnant<br />
females; but not necessarily the amounts of body reserves (see text).<br />
Reproductive Status Snout-vent Length<br />
(cm)<br />
Body Mass (g) Adjusted Body<br />
Mass (g)<br />
Reproductive 48.80±6.14 99.84±38.01 93.59±3.25<br />
Non-reproductive 46.17±4.11 60.73±18.16 66.98±4.80<br />
Effect of Body Mass, Ambient Temperature and Reproductive Status on Oxygen<br />
Consumption in Early Gestation<br />
Body mass and body temperature are the two most important determinates of any<br />
ectotherm’s metabolic rate, therefore, these variables must be controlled before<br />
looking for an effect of reproductive status. We performed an ANCOVA with Log –<br />
VO2 (ml -1 h -1 ) as the dependent variable, reproductive status as the factor, Log - BM<br />
and body temperature as the covariates. <strong>The</strong> whole model (test of the sum of<br />
squares [SS] of the whole model versus SS residuals) explained a large proportion of<br />
the variance in oxygen consumption (r² = 0.60; F(3,54) = 26.59, p < 0.0001; Levene’s<br />
183
test for homogeneity of the variance, degrees of freedom for all F’s - 1, 56; for Log –<br />
VO2: F = 1.14, p= 0.29; for Log - BM: F = 0.47, p= 0.50; for Body Temperature: F =<br />
1.46, p = 0.23), reinforcing the notion that body mass and body temperatures play a<br />
major role in the oxygen consumption of ectotherms (Saint Girons et al. 1985;<br />
Beaupré & Zaidan III 2001). Although both covariates strongly and positively<br />
influenced oxygen consumption (respectively Log - BM: F(1,54) = 12.29, p < 0.001;<br />
and body temperature F(1,54)= 51.76, p < 0.0001), we did not find any significant<br />
effect caused by the reproductive status (F(1.54)= 2.82, p= 0.10)(Figure 1). Using<br />
ambient temperature (17.5 °C, 25 °C and 32.5 °C) as a second factor (due to its<br />
discontinuous nature in the experiment), instead of body temperature as a covariate,<br />
did not change the result (whole model: r² = 0.61; specific effect of ambient<br />
temperature: F(2,51)=25.07, p
Mass Adjusted Oxygen Consumption<br />
Log-VO2 (ml -1 h -1 )<br />
0.7<br />
0.6<br />
0.5<br />
0.4<br />
0.3<br />
0.2<br />
0.1<br />
0.0<br />
Reproductive<br />
Non reproductive<br />
6<br />
14<br />
13<br />
17.5°C 25.0°C 32.5°C<br />
Figure 1. Effect of reproductive status and temperature on oxygen consumption of female Aspic vipers<br />
during early-gestation. Each point represents the adjusted mean (least-squares using Log - VO2 ( mL -1<br />
h -1 ) as the dependent variable, reproductive status and ambient temperature as the factor and Log -<br />
BM as the covariate)(±SE and sample size) for each group of snakes in each category.<br />
Table 3. Comparison of the current mass-independent VO2 (ml g 1 h -1 ) data against data previously<br />
collected an ecologically similar species and an ecologically different species tested at three imposed<br />
temperature regimes (Secor & Nagy 1994). Vipera aspis and Crotalus cerastes are similar, sit-and-<br />
wait predators (capital breeders), while Masticophis flagellum is a typical active forager (income<br />
breeder).<br />
Species Temperature Category<br />
5<br />
Cold (17.5 o C) Medium (25.0 o C) Hot (32.5 o C)<br />
Vipera aspis 0.018 0.031 0.05<br />
Masticophis flagellum 0.012 0.028 0.06<br />
Crotalus cerastes 0.009 0.020 0.04<br />
185<br />
8<br />
13
Effect of Gestation Period on Oxygen Consumption<br />
In pregnant females, there was no change <strong>between</strong> the mean mass-independent VO2<br />
(residuals) for the two periods, early and late gestation. To control for possible inter-<br />
individual differences in reproductive stage, or body mass, we used the same<br />
females on the two occasions under the same temperature conditions (seven<br />
females at 17.5 °C and 10 females at 32.5 °C). <strong>The</strong> seventeen females were<br />
sampled during early gestation and again, two months later, during late gestation.<br />
Despite being sensitive enough to detect any potential difference in mass-<br />
independent VO2 <strong>between</strong> gestation periods the T-test for dependent samples<br />
provided a non-significant result (t = 1.56, df = 16, p= 0.14; Figure 2).<br />
Mass Independent Oxygen<br />
Consumption (residuals)<br />
0.3<br />
0.2<br />
0.1<br />
0.0<br />
-0.1<br />
-0.2<br />
-0.3<br />
-0.4<br />
EARLY LATE<br />
GESTATION<br />
Figure 2. <strong>The</strong> effect of early (shortly after ovulation) versus late (shortly before parturition) gestation,<br />
on oxygen consumption in 17 pregnant female vipers. Each female was plotted two times (early and<br />
late gestation) with a line connecting the two values. See text for statistics and details.<br />
186
Effect of Fecundity on Oxygen Consumption<br />
For the data collected in early gestation, a partial correlation analysis was performed<br />
to take into account the effect of body temperature, body mass and fecundity on<br />
oxygen consumption (r² = 0.51, F(3,34) = 11.97, p < 0.0001; specific effect of body<br />
temperature, F(3,34) = 4.62; p < 0.001; Log - BM, F(3,34)=2.68, p = 0.011; fecundity,<br />
F(3,34)= -0.31, p = 0.758). Overall, in early gestation, fecundity per se did not<br />
influence significantly oxygen consumption of reproductive individuals (Figure 3a).<br />
For this analysis the sample size was reduced and the three ambient temperatures<br />
used (17.5 °C, 25 °C and 32.5 °C) created discontinous data <strong>between</strong> groups of<br />
females, making the use of body temperature as an independent variable suspect.<br />
For this reason, we adjusted the oxygen consumption to the mean value obtained at<br />
32.5 °C by multiplying the values obtained under low and medium temperature by 2.8<br />
and 1.6 respectively; the factor by which oxygen consumption increased as<br />
temperature increased. Even after adjusting oxygen consumption, the result<br />
remained unchanged (r² = 0.18, F(2,35) = 3.90, p < 0.03; specific effect of fecundity,<br />
F(2,35) = 1.32, p = 0.195).<br />
However, in late gestation the data show a positive and significant <strong>relationship</strong><br />
<strong>between</strong> oxygen consumption and fecundity, measured as the number of offspring<br />
produced at parturition (Figure 3b). A partial correlation analysis was performed to<br />
take into account the effect of body mass on oxygen consumption. For this analysis<br />
the sample size was further reduced (n = 19 females) and only two ambient<br />
temperatures were used (17.5 °C and 32.5 °C), thus we used the temperature<br />
adjusted oxygen consumption (r² = 0.38, F(2,16) = 4.96, p = 0.02; specific of<br />
fecundity, F(2,16) = 2.97, p = 0.008) and, despite a small sample size, the power [1-<br />
β] of this analysis was relatively high (0.83).<br />
187
Mass Specific Temperature-Adjusted<br />
Oxygen Consumption (residuals)<br />
Mass Specific Temperature-Adjusted<br />
Oxygen Consumption (residuals)<br />
1.5<br />
1.0<br />
0.5<br />
0.0<br />
-0.5<br />
-1.0<br />
-1.5<br />
1 2 3 4 5 6 7 8 9 10 11 12 13<br />
0.5<br />
0.4<br />
0.3<br />
0.2<br />
0.1<br />
0.0<br />
-0.1<br />
-0.2<br />
-0.3<br />
-0.4<br />
Fecundity at ovulation<br />
-0.5<br />
1 2 3 4 5 6 7 8<br />
Number of viable offspring<br />
Figure 3. Relationship <strong>between</strong> oxygen consumption (residuals from the regression of Log – BM<br />
against Log - VO2, adjusted to an ambient temperature of 32.5°C, see text) and fecundity of<br />
reproductive female vipers during early gestation (N = 38, upper graph [Figure 3a], Y = 0.02X - 0.413,<br />
P = 0.227), and the number of viable neonates during late gestation (N = 19, lower graph [Figure 3b],<br />
Y = 0.610X - 0.501, P < 0.006). Fecundity at ovulation was determined by palpation (see text), and<br />
was measured as the number of viable neonates or fully developed stillborn young at parturition. A<br />
number of ovulated follicles do not develop and lead to unfertile eggs (where only yolk is identifiable),<br />
explaining the difference <strong>between</strong> the ranges of values <strong>between</strong> the two X axis (Bonnet et al. 2000b).<br />
A similar <strong>relationship</strong> was observed regressing the mass of live young<br />
produced against temperature adjusted VO2 (r 2 = 0.38, p = 0.029, n = 19, power =<br />
0.83). Interestingly, we found that post-parturient body mass of the females had no<br />
influence on the oxygen consumption measured before parturition (r² = 0.004, n = 19,<br />
188
p = 0.80), suggesting that the contribution of the fatigued maternal organisms in the<br />
<strong>total</strong> variance was minor in comparison to the contribution of the offspring they carry.<br />
Field Body Temperatures<br />
We found a strong effect of the sampling date on the mean body temperature<br />
selected by snakes in the field (ANOVA, F(89,1806) = 7.44, p < 0.0001). As<br />
expected, the mean body temperature of the snakes increased from mid-spring to<br />
mid-summer (correlation <strong>between</strong> mean body temperature of the females and date: r²<br />
= 0.16, F(1,88) = 17.23, p < 0.0001), with important daily fluctuations as expected<br />
under temperate-oceanic climate (see Figure 4). Field data showed that in the course<br />
of the reproductive period, reproductive females maintain higher body temperature<br />
(25.01±0.26 °C, N = 989) than non-reproductive females (22.03±0.27 °C, n =<br />
907)(ANCOVA with reproductive status as the factor, temperature records as the<br />
dependent variable and date as the covariate; reproductive status, F(1,1893) = 44.05,<br />
P < 0.0001; date, F(1,1893) = 59.86, p < 0.0001; Levene’s test for homogeneity of the<br />
variance; F(1,851) = 1.01, p = 0.31 and F(1,851) = 0.93, p = 0.34 for body<br />
temperature and date respectively). Such a difference was more pronounced during<br />
gestation where the body temperature of reproductive snakes was approximately<br />
3.79 °C higher (3.27 °C for date-adjusted means); the mean body temperature of<br />
reproductive and non-reproductive females being 26.39±0.29 °C (±SD, N = 605) and<br />
22.60±0.37 o C (±SD, n = 438), respectively (same design ANCOVA; reproductive<br />
status, F(1,1040) = 47.10, p < 0.0001; date, F(1,1040) = 15.75, p < 0.0001, Figure 4).<br />
Despite the time elapsed <strong>between</strong> two consecutive temperature records, each<br />
individual was represented more than once, and this may lead to spurious pseudo-<br />
replication effects. We checked for the possibility that pseudo-replication generated<br />
the observed difference in body temperature <strong>between</strong> the two classes. We used the<br />
189
Mean field-body temperature<br />
of female asp vipers (°C)<br />
mean body temperature calculated over the reproductive period for each female<br />
(generating independent data but weakening the sensitivity of the analysis) and<br />
performed an ANOVA with reproductive status as the factor and mean body<br />
temperature as the dependent variable. <strong>The</strong> above results were unchanged (ANOVA,<br />
F(1,19) = 7.06, p = 0.015), with mean body temperature of reproductive and non-<br />
reproductive females being 24.54±1.23 o C (±SD, N = 11) and 19.80±1.29 °C (±SD, N<br />
= 10), respectively.<br />
35<br />
30<br />
25<br />
20<br />
15<br />
5<br />
0<br />
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46<br />
Days of Gestation<br />
Figure 4. During gestation, reproductive females (grey symbols) maintain higher body temperatures<br />
than non-reproductive females (white symbols). Each symbol represent the mean body temperature<br />
(±SE) calculated for several females (n = 11.85±5.29 records on average, range 2 - 25). Despite<br />
marked daily fluctuations (due to environmental variations: i.e. passage of very cold fronts in late June,<br />
days 26 and 30), the mean body temperature of reproductive females was often several degrees<br />
higher compared to non-reproductive females. Body temperature was taken at the point of capture<br />
irrespective of behavioural status. See results for statistics.<br />
190
Discussion<br />
<strong>The</strong> data from this investigation suggest that, throughout the reproductive period,<br />
metabolism in both reproductive and non-reproductive snakes was strongly affected<br />
by both temperature and body mass. An important body of published data have<br />
already reported these effects in ectotherms, for temperature (Bennett & Dawson<br />
1976; Naulleau et al. 1984; Andrews & Pough 1985; Loumbourdis & Hailey 1985;<br />
Bradshaw et al. 1991; Zari 1991; Thompson & Withers 1992) and body mass<br />
(Bennett & Dawson 1976; Dmi'el 1986; Zari 1991; Thompson & Withers 1992;<br />
Beaupré & Zaidan III 2001). <strong>The</strong> expected difference in the metabolic rate <strong>between</strong><br />
reproductive and non-reproductive females, however, was not apparent in our data<br />
when the two groups of snakes were placed under the same temperature regime.<br />
Furthermore, in reproductive females, metabolism was largely independent of<br />
fecundity. With regard to these latter outcomes, our results differ from previous<br />
investigations in this field (Guillette 1982; Birchard et al. 1984; Beaupré & Duvall<br />
1998; Kunkele 2000).<br />
Why do our data conflict with other investigations and fail to support some of<br />
our initial hypotheses? Firstly, it is necessary to consider the methodology. For<br />
example, the period over which oxygen consumption was measured was different<br />
<strong>between</strong> the current and previous studies. In the live bearing lizard Sceloporus<br />
aeneus and the garter snake, Thamnophis s. sirtalis comparisons were made with<br />
reproductive females less than two weeks prior to parturition (Guillette 1982; Birchard<br />
et al. 1984). Beaupré & Duvall (1998) focused on vitellogenesis and they showed that<br />
reproductive female western diamondback rattlesnakes, Crotalus atrox, consume 1.4<br />
times the amount of oxygen as non-reproductive females; however, it is not made<br />
clear exactly when during vitellogenesis (a prolonged period in snakes; two to four<br />
191
months, or more in viperids) the data were collected. Because the metabolic process<br />
from the beginning of vitellogenesis to the end of pregnancy would be very dynamic,<br />
it is difficult to compare directly these data and previous investigations. In addition,<br />
the few snake species studied belong to very different lineages (colubrids versus<br />
viperids for instance). It may also be that our methodology lacked the capacity to<br />
detect subtle differences that may affect the outcome. Important inter-individual<br />
differences, such as sensitivity to manipulation, hormonal levels related to stress, and<br />
state of metabolically active tissue, such as the intestine epithelium, would certainly<br />
increase the variation in our data set.<br />
Nevertheless, the consistency <strong>between</strong> published and current data suggests<br />
that the system employed during this investigation lead to reasonable and<br />
interpretable results (Table 3) and important effects such as those linked to body<br />
mass or body temperature were clearly visible in our data. For instance our Q10 of<br />
2.05 is similar to the typical Q10 of <strong>between</strong> two and three reported for most reptiles<br />
(Bennett & Dawson 1976; Andrews & Pough 1985; Thompson & Withers 1992). <strong>The</strong><br />
experimental process undertaken, augmented by long term monitoring of large<br />
number of female Aspic vipers in the field, enables us to propose a number of<br />
falsifiable hypotheses to explain our unexpected results. For that, a simple approach<br />
is to follow the chronology of reproduction in female viviparous snakes: from<br />
vitellogenesis to the end of gestation.<br />
Vitellogenesis<br />
Clear differences in the metabolic effort <strong>between</strong> reproductive and non-reproductive<br />
females during vitellogenesis may be expected in capital breeders, such as female<br />
rattlesnakes or vipers. In capital breeders, non-reproductive females accumulate<br />
energy, sometimes over very prolonged periods (years in vertebrates; Bull & Shine<br />
192
1979), until a sufficient store (the capital) has been constituted following which<br />
vitellogenesis can be engaged (Drent & Daan 1980; Stearns 1992; Naulleau &<br />
Bonnet 1996; Bonnet et al. 1998). During vitellogenesis there is an extensive<br />
mobilisation of maternal body reserves to produce a large number of offspring (Bull &<br />
Shine 1979). In snakes, the vitellogenic process is a very intensive physiological<br />
event (Bonnet et al. 1994) where the energetic expenditure of reproductive females<br />
increases substantially over that of non-reproductive females, as was demonstrated<br />
by Beaupré & Duvall (1998) in the viperid Crotalus atrox. Although we have no data<br />
on the metabolic rate of Aspic vipers during vitellogenesis, indirect information<br />
supports the notion that this period corresponds to an increase of oxygen<br />
consumption for reproductive females over non-reproductive females. Firstly, large<br />
amounts of body resources are mobilised to develop follicles whilst non-vitellogenic<br />
females do not exhibit any sign of such a mobilisation at that time (Bonnet et al.<br />
1994). Second, vitellogenic females move and forage intensively at that time, mostly<br />
to supplement the energy available for follicular development which, in turn, improves<br />
their reproductive success; at the same time, non-vitellogenic females are more<br />
sedentary, probably to save energy and to minimise predation risk (Naulleau et al.<br />
1996; Bonnet et al. 2001a, 2002a). Third, vitellogenic females engage in<br />
energetically demanding acts of sexual behaviour (courtship, mating) whilst non-<br />
vitellogenic females do not (Naulleau et al. 1999; Aubret et al. 2002). Fourth,<br />
vitellogenic females bask in the sun much more frequently than non-vitellogenic<br />
females (Bonnet & Naulleau 1996) with a concomitant increase in metabolism due to<br />
the fundamental effects of increased temperature on reaction rates (see Results).<br />
Overall, development of follicles during vitellogenesis is dependant on the increased<br />
activity of metabolically active tissues such as the liver, the ovaries, the intestine<br />
epithelium, and the locomotor muscles, whilst these tissues are relatively less active<br />
193
in non-vitellogenic females (Secor et al. 1994; Singh & Ramachandran 2000;<br />
O’Sullivan et al. 2001). <strong>The</strong> completion of vitellogenesis occurs in early June<br />
(Naulleau & Bidaut 1981), and this period corresponds to the beginning of summer in<br />
our study area.<br />
Pregnancy<br />
Pregnancy immediately follows vitellogenesis. At this time follicular development and<br />
the mobilisation of maternal reserves are complete (Bonnet et al. 1994, 2001a) and<br />
the extra-metabolic cost of being pregnant in Aspic vipers may only be a small effort<br />
associated with the maintenance of follicles. Our data support this notion with the<br />
decrease in reproductive effort apparent in the lack of difference <strong>between</strong> the<br />
oxygen consumption of reproductive and non-reproductive females. However, during<br />
the early stages of pregnancy a precise comparison of oxygen consumption <strong>between</strong><br />
the two classes of female snakes is difficult due to the potentially confounding effect<br />
of the mass of the yolk follicles. Yolk is comprised mainly of fat (50%) and water,<br />
components likely to contribute little to organismal metabolic effort (Darken et al.<br />
1998). In early pregnancy developing embryos are very small (less than 5mm in <strong>total</strong><br />
length; unpublished data) and the yolk component is relatively large. This is a<br />
methodological difficulty that can only be resolved using techniques such as<br />
ultrasound (Beaupré & Duvall 1998) and Nuclear Magnetic Imaging and Proton<br />
Spectroscopy (unpublished) to measure the exact volume and chemical composition<br />
(i.e. water versus lipids using NMPS) of the follicles, from which mass and energy<br />
can be calculated in a non-destructive manner.<br />
Field data further support the idea that early gestation is not metabolically<br />
demanding. Gravid females in early gestation are often already emaciated at this<br />
stage (i.e. fat bodies represent less than 3% of the mass of the post-ovulatory<br />
194
females; Bonnet et al. 2002b). In this situation any substantial increases in<br />
metabolism would lead to a rapid depletion of these reserves before parturition and<br />
the possible failure to successfully produce viable offspring. Pregnant females could<br />
forage at this time to compensate for their low energy stores. However, field data<br />
show that this is not often the case. Instead they sharply reduce both locomotor and<br />
hunting activities during the two to three months of pregnancy, seldom capturing prey<br />
(Naulleau et al. 1996; Lourdais et al. 2002a). <strong>The</strong> limited body reserves of pregnant<br />
females, sometimes supplemented by energy from prey, provide sufficient energy to<br />
sustain the metabolism of the females until parturition (Lourdais et al. 2002a). <strong>The</strong>se<br />
experimental, field, and anatomical (comparative metabolic activity of muscles, liver,<br />
or intestinal epithelium <strong>between</strong> vitellogenesis and gestation) data suggest that early<br />
pregnancy is not a costly metabolic phenomenon per se. As a consequence the<br />
absence of a difference in mass-independent VO2 <strong>between</strong> pregnant and non-<br />
reproductive female vipers in early gestation is not surprising when temperature is<br />
controlled by the experimenters.<br />
In late gestation embryonic development was mostly complete, and each<br />
foetus could potentially contribute to the overall metabolism of the reproductive<br />
female. However, the current investigation suggests that reproductive females<br />
maintain a similar rate of mass-independent VO2 throughout pregnancy. Such a result<br />
is partly logical because VO2 was systematically scaled by maternal mass, which<br />
includes the mass of the embryo and extra-embryonic fluids (the <strong>total</strong> water content<br />
of embryo and fluids being close to 80 %; unpublished data) in late gestation. But this<br />
is not entirely satisfactory as we may expect a slight increase in the mass-<br />
independent oxygen consumption of the females in the course of pregnancy. Firstly,<br />
because the overall body composition (mother and embryos) now includes a larger<br />
proportion of active tissues (muscles of the embryos etc. versus yolk) at the end of<br />
195
gestation. Second, the oxygen consumption by developing embryos is not the only<br />
factor of pregnancy that will affect metabolism. Other energetic components, such as<br />
supplying oxygen to the foetuses and handling foetal nitrogenous waste, are also<br />
paid by the mother (Clark & Sisken 1956). Birchard et al. (1984) suggest that the<br />
contribution of foetal oxygen consumption to the overall oxygen consumption of the<br />
mother is the factor responsible for the observed difference in metabolic rate of<br />
pregnant versus non-pregnant garter snakes. Perhaps inter-individual variance<br />
rendered our data too noisy to detect such a difference that is likely to be subtle<br />
anyway. In late gestation there was a significant positive <strong>relationship</strong> <strong>between</strong><br />
fecundity and oxygen consumption, though the <strong>relationship</strong> was weak, indicating that<br />
our measurements enabled to detect the fact that the intra-uterine embryos were<br />
metabolically active. An alternative explanation may be that any detectable difference<br />
generated by the embryos may have been masked by a decrease in maternal<br />
metabolism at the end of gestation due to the fatigued body condition of the<br />
reproductive females in the later stages of pregnancy (Bonnet et al. 2002b).<br />
Field versus laboratory data<br />
Overall, when ambient temperature was imposed in the laboratory, reproductive<br />
status and fecundity did not influence, or only weakly influenced, maternal<br />
metabolism. However, field body temperatures show that a commitment to<br />
reproduction means a dramatic shift in behavioural thermoregulation with pregnant<br />
female vipers maintaining, on average, a 4 o C higher body temperature than non-<br />
reproductive females. This result is consistent with the observation that, in the field,<br />
pregnant females bask more often than non-reproductive females (Bonnet &<br />
Naulleau 1996). Given the fact that reproduction (at least vitellogenesis) is an<br />
energetically expensive process, and considering the difficulties of reproductive<br />
196
females to sustain metabolism at a higher rate than non-reproductive females<br />
independently from a shift in thermoregulatory behaviour (as they are ectothermic), it<br />
seems intuitive that the maintenance of higher preferred body temperature over<br />
prolonged time periods accelerates metabolism to the rate where development of<br />
young is completed over an appropriate time scale. Embryos are sensitive to<br />
temperature, both in terms of extremes and variations (Burger 1998; Rhen & Lang<br />
1999), and development is usually optimised by precise temperatures (Shine &<br />
Harlow 1996; Downes & Shine 1999; Elphick & Shine; Shine 1999; Shine & Downes<br />
1999; Andrews et al 2000). It is very likely that the optimal temperatures for follicular<br />
growth and for embryonic development are the same whatever the number of<br />
offspring. This may explain the absence of any <strong>relationship</strong>, in the current<br />
investigation, <strong>between</strong> mean field body temperature measured in pregnant females<br />
and fecundity.<br />
Conclusion<br />
Taken together these results suggest that, in female Asp vipers, the change of<br />
reproductive status corresponds to an all-or-nothing system in terms of energetic<br />
metabolism. Vitellogenesis and pregnancy impose a discrete modification in the day-<br />
to-day life of the females rather than progressive adjustments to the size of the litter<br />
(from zero to 13 offspring in our study zone). Capture-recapture data have shown<br />
that survival is independent of fecundity in our population (Bonnet et al. 2002b). We<br />
hypothesise that increased thermoregulatory behaviour linked to vitellogenesis and<br />
gestation substantially increase maternal metabolism, provoke a strong emaciation<br />
and result in increased exposure of the females to avian predation. As a result, the<br />
majority of the females will not survive a single reproductive event whatever their<br />
current fecundity (Bonnet et al. 2002a, b). Our field and laboratory data support the<br />
197
notion proposed by Bull & Shine (1979) that in systems where the costs of<br />
reproduction independent of fecundity are significant, the emergence of reproductive<br />
strategies based on a low breeding frequency should be favoured (which is the case<br />
for the many viperid snakes; Saint Girons 1957b; Brown 1991; Martin 1993; Bonnet &<br />
Naulleau 1996). <strong>The</strong> notion that costs of reproduction may be partly disconnected to<br />
fecundity provides immense potential for better understanding reproductive strategies<br />
such as capital breeding and or semelparity (Bull & Shine 1979; Bonnet et al. 1998;<br />
Olsson, et al. 2000). Although incomplete, our data clearly show that the true value of<br />
understanding the <strong>relationship</strong>s <strong>between</strong> fecundity, the energetics of reproduction<br />
and reproductive effort lie not in a simple comparison with non-reproductive<br />
conspecifics tested under laboratory conditions, but in the synthesis of laboratory<br />
data and field observation.<br />
Aknowledgements<br />
A special thanks to the <strong>CNRS</strong>, French Embassy in Canberra (A. Moulet, A. Littardi,<br />
and J. Mordeck), the Faculty of Science and the Zoology Department (UWA) for the<br />
financial support that made this project possible. Laurence Pastout taught the use of<br />
the metabolic chamber, Guy Merlet solved several technical problems, and Christian<br />
Thiburce cared for the animals <strong>between</strong> measurements.<br />
198
E. Article 7<br />
What is the Appropriate Timescale for<br />
Measuring Costs of Reproduction in a “Capital<br />
Breeder” such as the aspic viper?<br />
X. Bonnet 1 , G Naulleau 1 , R Shine 4 1 2 3<br />
, & O Lourdais<br />
1 Centre d'Etudes Biologiques de Chizé, <strong>CNRS</strong>, 79360, Villiers en Bois, France<br />
2 Conseil Général Des Deux Sèvres, Rue de L’abreuvoir, 79021, Niort, France<br />
3 University of Poitiers, 40 avenue du recteur Pineau, 86022 Poitiers, France<br />
4 Biological Sciences A08, University of Sydney, NSW 2006, Australia<br />
Published in Evolutionary Ecology 13: 485-497<br />
(2000)<br />
199
Abstract<br />
Before we can quantify the degree to which reproductive activities constitute a cost<br />
(i.e., depress an organism's probable future reproductive output), we need to<br />
determine the timescale over which such costs are paid. This is straightforward for<br />
species that acquire and expend resources simultaneously (income breeders), but<br />
more problematical for organisms that gather resources over a long period and then<br />
expend them in a brief reproductive phase (capital breeders). Most snakes are<br />
capital breeders; for example, female aspic vipers (Vipera aspis) in central western<br />
France exhibit a two- to three-year reproductive cycle, with females amassing energy<br />
reserves for one or more years prior to the year in which they become pregnant. We<br />
use long-term mark-recapture data on free-living vipers to quantify the appropriate<br />
timescale for studies of reproductive costs. Annual survival rates of female vipers<br />
varied significantly during their cycle, such that estimates of survival costs based only<br />
on years when the females were "reproductive" (i.e., produced offspring) substantially<br />
underestimated the true costs of reproduction. High mortality in the year after<br />
reproducing was apparently linked to reproductive output; low energy reserves (poor<br />
body condition) after parturition were associated with low survival rates in the<br />
following year. Thus, measures of cost need to consider the timescale over which<br />
resources are gathered as well as that over which they are expended in reproductive<br />
activities. Also, the timescale of measurement needs to continue for long enough<br />
into the post-reproductive period to detect delayed effects of reproductive "decisions".<br />
Key-words: body condition; capital breeder; energy stores; foraging; snake; Vipera<br />
aspis<br />
200
Introduction<br />
One of the primary aims of life-history theory is to facilitate comparisons among<br />
different kinds of organisms. Only in this way, by examining diverse taxa in a single<br />
conceptual framework, can we hope to derive general insights. One of the great<br />
successes in this respect has been the development of theory concerning costs of<br />
reproduction and the allied concept of Reproductive Effort. <strong>The</strong> field is based upon<br />
an original idea by George Williams (1966a,b): the realisation that an iteroparous<br />
organism will maximise its lifetime reproductive success not by maximising effort at<br />
the first reproductive opportunity, but by reproducing at a level that does not too<br />
greatly reduce it’s probable future output. This notion has been incorporated into<br />
elegant mathematical models(e.g., Schaffer 1974; Winkler & Wallin 198; Sibly &<br />
Calow 1984; Jönsson et al. 1995a,b), and has stimulated extensive empirical studies<br />
on a diverse array of species (e.g., Bell 1980; Bell & Koufopanou 1986; Clutton-Brock<br />
1991).<br />
In the present paper, we point out an important complication in measuring the<br />
costs of reproduction: we have to be careful about specifying what activities are<br />
included under our definition of reproduction. <strong>The</strong> significance of this superficially<br />
trivial caveat is that organisms differ in the timescale over which various components<br />
of “reproductive” activities are carried out. In particular, some animals concentrate all<br />
of the activities associated with reproduction (i.e., acquisition as well as expenditure<br />
of energy) in a single time period. For these income breeders (Drent & Daan 1980;<br />
Jönsson 1997), we can evaluate costs of reproduction by comparing energy balance<br />
and survival rates <strong>between</strong> reproductive and non-reproductive animals (e.g., Bell &<br />
Koufopanou 1986) or by documenting the effects of manipulating reproductive<br />
expenditure (e.g., Tombre & Erikstad 1996; Cichon et al. 1998). <strong>The</strong> situation is<br />
201
more complex with capital breeders that rely upon stored energy reserves to support<br />
reproductive output. For such taxa, energy acquisition and expenditure are<br />
temporally dissociated, so that measurements of cost taken during the period of<br />
reproductive expenditure will fail to include the components of cost that accrue during<br />
the energy-gathering phase.<br />
Timescales of Reproductive Costs<br />
Much of the scientific literature on costs of reproduction is based on studies of birds,<br />
and one technique that is often used is to manipulate reproductive expenditure by<br />
adding or removing eggs from the nest at the beginning of the period of parental care<br />
(e.g., Nur 1988; Tombre & Erikstad 1996; Cichon et al. 1998). Importantly, the facet<br />
of reproductive effort that is being affected by this manipulation is the level of the<br />
parents’ foraging effort to provision the nestling. If we compare this situation to that<br />
of a viviparous snake such as our study organism, the aspic viper, the contrast<br />
becomes obvious. Female aspic vipers produce litters only once every two or three<br />
years (or less often), with the intervening (“non-reproductive”) years being used to<br />
amass energy reserves that will fuel the eventual litter (Naulleau & Bonnet 1996).<br />
Females may eat relatively little during gestation; indeed, some may become<br />
completely anorexic (Saint Girons 1952, 1957a,b). For such an animal, most of the<br />
foraging effort occurs over a period of years prior to the year of reproductive output.<br />
If we evaluate costs of reproduction in such an organism by comparing females in<br />
“reproductive” versus “non-reproductive” years of their cycles, then we completely fail<br />
to assess the kind of reproductive cost (risk, etc., due to additional foraging) that has<br />
been the primary focus of studies on income-breeding species such as most birds.<br />
Any comparison of such costs <strong>between</strong> a bird and a snake – exactly the kind of<br />
202
comparison that is an aim of life-history theory - would be invalidated if we measured<br />
different components of cost in the two taxa.<br />
One interesting aspect of the bird-snake comparison is that the validity of such<br />
a comparison will depend on the currency in which costs are to be measured. If the<br />
currency is energetics, then it may be meaningful to measure <strong>total</strong> energy allocation<br />
to the reproductive event in both taxa, and to measure this trait over the period of<br />
expenditure only. Despite the fact that the bird may have gathered the energy over a<br />
period of weeks and the snake over a period of years, their allocation of energy to<br />
reproduction is still directly comparable. Thus, for example, one could compare <strong>total</strong><br />
reproductive allocation to body size <strong>between</strong> these two taxa. Unfortunately, the<br />
comparison breaks down as soon as we try to compare energy allocation to<br />
reproduction versus to other activities (maintenance, growth, etc.). At this point, the<br />
timescale becomes important – and for the reptile, the appropriate timescale is surely<br />
that over which these resources have been gathered, not just the “reproductive” year.<br />
<strong>The</strong> problem is even worse for survival rates, the other main potential currency in<br />
which costs can be assessed (and probably, the most important such currency for<br />
many kinds of animals –Shine & Schwarzkopf 1992). By analogy with the bird, the<br />
real survival cost of reproduction for a female viper involves the risks that she takes<br />
throughout the “non-reproductive” (energy acquisition) years as well as her risk<br />
during reproduction itself.<br />
Unfortunately for the logistics of assessing costs of reproduction, the vast<br />
majority of living species probably depend to a significant degree on stored reserves<br />
to fuel reproductive expenditure. This characteristic is particularly common in<br />
ectothermic species, for several reasons (Pough 1980; Bonnet et al. 1998).<br />
Ectotherms may also differ from endotherms in the timescale over which costs are<br />
expressed after the overt reproductive expenditure. Because endotherms (especially<br />
203
irds) are under strong energetic constraints, even a brief period of unfavourable<br />
energy balance may be fatal (e.g., Pough 1980). In contrast, the low metabolic<br />
requirements of ectotherms mean that any effect of reproductive expenditure on<br />
survival may not be apparent for a much longer period. For example, a bird that<br />
compromises its energy reserves or thermoregulatory efficiency due to reproductive<br />
costs may thereby die the following winter (e.g., McCleery et al. 1996; Daan et al.<br />
1996; Nilsson & Svensson 1996) whereas a reptile in the same situation can simply<br />
hibernate (which requires very little energy expenditure: Gregory 1982) over the<br />
entire winter period, and not have to face the consequences of its reduced energy<br />
reserves until the following spring. We stress, however that there is a continuum of<br />
timescales, and that some birds will resemble some reptiles in important respects.<br />
<strong>The</strong> difference is one of degree, but nonetheless may often be so substantial that we<br />
need studies on “costs” experienced by both kind of organisms. Even superficially<br />
similar phenomena in different types of organisms may differ in important respects.<br />
For example, high mortality in some passerines and small mammals in the year<br />
following breeding (i.e. Gustafsson & Sutherland 1988) is not directly comparable to<br />
the delayed post-reproductive decrease of survival in snakes. <strong>The</strong> “delayed survival<br />
costs” paid by small endotherms occur after several reproductive episodes (3 to 5<br />
clutches [litters] per year on average); but after a single reproductive episode in the<br />
snakes.<br />
Previous research on costs of reproduction in reptiles has concentrated<br />
primarily on events during the actual period of reproductive expenditure: for example,<br />
the decrease in survival rates, <strong>food</strong> <strong>intake</strong> and mobility of gravid females (e.g., Shine<br />
1980; Seigel et al. 1987; Madsen & Shine 1993). Our six-year mark-recapture study<br />
of free-ranging vipers allows us to document these costs over a longer timespan (i.e.,<br />
the female’s entire reproductive cycle, not simply the year in which she produces<br />
204
offspring). We examined the data to calculate the proportion of the <strong>total</strong> survival<br />
costs of reproduction that accrue during different phases of the female cycle. If the<br />
probability of survival is very high during “non-reproductive” years, and very low<br />
during “reproductive” years, then estimates based only on the latter timeframe may<br />
nonetheless provide a reasonable index of overall survival costs. However, if survival<br />
rates are low during other phases of the cycle, then measurements restricted to<br />
"reproductive" years may substantially underestimate the true costs of reproduction<br />
(Jönsson et al. 1995a,b).<br />
If we can quantify survival rates at each stage of the female's cycle, we can<br />
compare the magnitude of pre-breeding, breeding and post-breeding components of<br />
the overall cost of reproduction. This comparison would be impossible in an income<br />
breeder, because the costs of energy acquisition and offspring production occur<br />
simultaneously, whereas they are separated temporally in a capital breeder. <strong>The</strong><br />
relative magnitude of post-breeding costs is also likely to differ in consistent ways<br />
<strong>between</strong> ectotherms and endotherms, because the high metabolic rates of the latter<br />
group mean that over-depletion of energy reserves during reproduction is likely to<br />
cause death rapidly. Such an effect may well be postponed for a very long period in<br />
an organism with lower metabolic needs, such as a viperid snake. To evaluate<br />
whether or not reproducing vipers pay long-term costs in survival, we can use the<br />
comparison among years of a female’s cycle (above). A higher rate of mortality in<br />
the year immediately after parturition would support the notion of “delayed” survival<br />
costs. Also, we can compare survival rates in that post-partum year to a female’s<br />
body condition immediately after she has reproduced in the preceding year.<br />
205
Materials and Methods<br />
Aspic vipers (Vipera aspis) are medium-sized (average adult = 48.5 cm snout-vent<br />
length, 85.5 g) venomous snakes widely distributed through Europe. We studied a<br />
population in central western France (Les Moutiers en Retz), in a mosaic of<br />
meadowland and thicker vegetation. <strong>The</strong> snakes were hand-captured and<br />
individually-marked (scale clipping or, later in the study [1993] with electronic tags,<br />
sterile transponder TX 1400L, Rhône Mérieux, Destron/IDI INC). Recapture rates<br />
were high and emigration was extremely rare, because the snakes are very<br />
sedentary (Naulleau et al. 1996) and the 33-hectare study area is bounded by habitat<br />
unsuitable for this species. Thus, snakes that disappeared had almost certainly died<br />
rather than emigrated. To ensure that the lower catchability of non-reproductive<br />
females relative to reproductive females did not falsify our results, we waited at least<br />
two years to classify a given female as dead or not (and thus we did not score<br />
survival of females caught in 1996 or 1997).<br />
Further details on the study area and our methods are given elsewhere<br />
(Bonnet & Naulleau 1996; Naulleau & Bonnet 1996; Naulleau et al. 1996; Bonnet et<br />
al. 2000b). Female vipers in this population typically reproduce with a two to three<br />
year cycle, although many females do not live long enough to produce more than a<br />
single litter (Naulleau et al. in prep.). Litters consist of 1 to 13 large (17.9 ± 1.2 cm<br />
SVL, 6.3 ± 1.1 g) neonates. Females with a body size greater than the minimal size<br />
at which parturition has been recorded (41.5 cm SVL, 47 cm <strong>total</strong> length) were<br />
considered as adult.<br />
206
For the analysis of survival rates in each year of the reproductive cycle, we<br />
had to classify females with respect to their stage of the cycle. This procedure was<br />
straightforward for reproductive females, and for non-reproductive females one to<br />
four years after reproduction (post-reproductive), but more problematical for females<br />
that were pre-reproductive – i.e., those that were in the years prior to their first<br />
“intended” litter. <strong>The</strong> individuals allocated to this category were those that we caught<br />
one to four years before they first reproduced. In order to qualify as pre-reproductive,<br />
the animals had to be in relatively good body condition at the first capture (indicating<br />
that they were not post-parturient: e.g. absence of flaccid abdomen or extensive skin<br />
folds), and they needed to have been regularly recaptured (so that we were sure that<br />
they did not produce a litter during this period). Some of them were first caught as<br />
juveniles (based on their small size) and later recaptured after they had attained adult<br />
body size but before their first reproduction. In practice, the maternal body-condition<br />
threshold for breeding in this species is so consistent (Naulleau & Bonnet 1996) that<br />
it was possible to classify such females with confidence. Females increase steadily<br />
in condition throughout the “non-reproductive” years of their cycle (Figure 1). Each<br />
female was represented only once in the analyses. In <strong>total</strong>, data on 527 adult<br />
females was used in the following analyses.<br />
207
MATERNAL BODY CONDITION<br />
0.3<br />
0.2<br />
0.1<br />
0.0<br />
-0.1<br />
-0.2<br />
-0.3<br />
VIT<br />
1<br />
GEST<br />
PART<br />
2 2<br />
YEAR 1 YEAR 2 YEAR 3 YEAR 4<br />
VIT2<br />
GEST2<br />
3 - YEARS CYCLE<br />
2 - YEARS CYCLE<br />
Figure 1. Patterns of change in maternal body condition (residual score from the general linear<br />
regression of ln-transformed mass versus snout-vent length) over the course of the reproductive cycle<br />
in female vipers, Vipera aspis. Females lose considerable body condition at parturition (because of<br />
the mass of the litter) and must gradually recover this condition over a period of two to four years<br />
before they can reproduce again. <strong>The</strong> Figure provides data from females reproducing on a two-year<br />
(circles and black lines) or three-year (triangles and grey lines) cycle length. “Vit” = during the period<br />
of vitellogenesis; “gest” = during gestation; “part” = after parturition. Arrows indicate the onset of<br />
vitellogenesis in a given year (1 = first reproduction, 2 = second reproduction).<br />
Results<br />
Figure 2 shows that survival rates varied significantly over the course of the female’s<br />
reproductive cycle (χ 2 = 35.2, 3 df, p< 0.0001, all sample sizes indicated in Figure 2).<br />
Female vipers experienced very high mortality (46%) in years when they<br />
"reproduced" (i.e., initiated vitellogenesis and [if they survived] produced offspring).<br />
This rate of mortality was significantly higher than that exhibited by females at other<br />
208<br />
PART2<br />
VIT3<br />
GEST3<br />
PART3
stages of their reproductive cycles (versus pre-reproductive females: χ 2 = 30.7, 1 df,<br />
p < 0.0001; versus females one year after parturition: χ 2 = 4.2, 1 df, p= 0.041; versus<br />
females two years after reproduction: χ 2 = 6.9 [Yates correction], 1 df, p= 0.01).<br />
Mortality rates were also high in the year following parturition (32.5% of snakes died),<br />
but were relatively low in other years of the cycle (e.g., annual mortality rate was only<br />
16% two years after parturition). Survival was particularly high (80%) for females in<br />
the year immediately preceding reproduction, when they were in very good body<br />
condition (Figure 2).<br />
PROBABILITY<br />
OF SURVIVAL<br />
1.0<br />
0.9<br />
0.8<br />
0.7<br />
0.6<br />
0.5<br />
0.4<br />
0.3<br />
0.2<br />
0.1<br />
0.0<br />
188<br />
234<br />
80<br />
BEFORE DURING + 1 YEAR + 2 YEARS<br />
REPRODUCTIVE STAGE<br />
Figure 2. Survival rates of adult female vipers (Vipera aspis) in each year of their reproductive cycle.<br />
“Before” = pre-reproductive females; “during” refers to the year when offspring are produced; “+ 1<br />
year” means the year following litter production; “+ 2 years” means the subsequent 12-month period.<br />
Sample size are indicated above each bar. See text for explanation of criteria used, and statistical<br />
tests on these data.<br />
209<br />
25
We can use these data to estimate the relative magnitude of each component of cost.<br />
<strong>The</strong> <strong>total</strong> survival cost of reproduction for a female aspic viper can be divided into<br />
three components:<br />
(1) Pre-breeding cost. - This is the decrease in survival probability caused by the<br />
female delaying reproduction past the time when she has attained adult body size.<br />
This delay is clearly used to build up energy reserves (Figure 1); a female that was<br />
an income breeder would not need to delay for this additional year, and so would not<br />
pay this cost. Females in this phase comprise two of the groups in Figure 2: pre-<br />
reproductive animals, and females two years post-partum. For both groups, annual<br />
survival rates were approximately 82%. Assuming for simplicity that females differ<br />
only in the length of their cycle, the pre-breeding cost averaged an additional 18%<br />
probability of mortality for a female viper with a three-year reproductive cycle. For a<br />
female with a four-year cycle (also common in our study population), this component<br />
of cost is paid in two successive years as energy stores are laid down. <strong>The</strong> <strong>total</strong><br />
additional risk for such a female is thus 33% (= 1.0 - [0.82 X 0.82]). Much of this<br />
additional mortality may not be a direct consequence of reproductive-related activities<br />
such as increased foraging effort, but may be due to random mortality that also<br />
affects immature individuals and adult males. Nonetheless, the mortality is<br />
experienced because of the need to delay reproduction until females reach the<br />
reproductive threshold, and thus can legitimately be considered as a “cost” of this<br />
delay.<br />
(2) Breeding cost. - In the year that they initiated vitellogenesis, females experienced<br />
an annual survival rate of only 54%. Thus, the cost of the activities directly<br />
associated with offspring production (e.g., mating, gestation, parturition) averaged<br />
46%. As above, we note that some component of this mortality risk may be unrelated<br />
210
to reproductive activities, but nonetheless comprises part of the “costs” that are paid<br />
during the reproductive year.<br />
(3) Post-breeding cost. - Females experienced high mortality in the year immediately<br />
following parturition (Figure 2; 67% survival, = 33% cost: note above caveat).<br />
Survival rates of female vipers were lower in the immediately post-parturient year<br />
than in other “non-reproductive” years (Figure 2; comparing survival rates in the<br />
years immediately preceding versus following the “reproductive” year: χ 2 = 4.7, 1 df,<br />
p=0.031). This difference supports the notion that there are mortality risks<br />
associated with reproducing, that are not manifested until long after the litter is<br />
produced.<br />
To further test this proposition, we can compare a female’s probability of<br />
survival in that post-parturient year, to her body condition (residual score from the<br />
linear regression of ln-transformed mass to SVL) immediately following parturition in<br />
the preceding year. We used logistic regression for this analysis, because the<br />
dependent variable (survived versus died) is a dichotomous trait. As predicted, a<br />
female’s probability of survival in the year after she reproduced was significantly<br />
higher if she was in relatively better body condition immediately after giving birth the<br />
year before (χ 2 = 5.8, 1 df, p= 0.016; Figure 3). This significant association supports<br />
the interpretation that mortality in the year after litter production constitutes a delayed<br />
(post-breeding) cost of reproduction for female vipers.<br />
211
SURVIVAL<br />
1<br />
0<br />
- 0.4 - 0.3 - 0.2 - 0.1 0.0<br />
BODY CONDITION<br />
Figure 3. Logistic regression of a female viper’s probability of survival in the twelve months following<br />
her production of a litter, as a function of her post-parturient body condition. Body condition was<br />
calculated as the residual score from the general linear regression of ln-transformed mass versus<br />
snout-vent length for all females within the population; almost all values are negative because post-<br />
parturient females are always in much poorer body condition than are other (pre-reproductive)<br />
females. See text for explanation and statistical results.<br />
Discussion<br />
<strong>The</strong> central result from our analysis is a very straightforward one: the timescale over<br />
which we measure costs of reproduction needs to reflect the timescale over which an<br />
animal engages in activities that support that reproductive bout. In capital-breeding<br />
species, that timescale may well be very much greater than the actual period over<br />
which overt “reproduction” (production of offspring) occurs. <strong>The</strong> extended timescale<br />
reflects two factors: a longer pre-reproductive period of energy-gathering, and a<br />
longer post-reproductive period when effects of reproductive activities are<br />
manifested. Interspecific comparisons based on shorter timescales are likely to be<br />
212
misleading if they compare different components of reproductive cost of one kind of<br />
organism (e.g., an avian income-breeder) versus the other (a reptilian capital-<br />
breeder).<br />
Previous theoretical treatments have identified the importance of this<br />
distinction <strong>between</strong> pre- and post-breeding costs of reproduction for understanding<br />
the evolution of reproductive tactics (Sibly & Callow 1984; Stearns 1992; Jönsson et<br />
al. 1995a,b; Jönsson 1997). Our data provide strong empirical support for the<br />
assumptions that underpin these models, especially those proposed by Jönsson et<br />
al. (1995a,b). Thus, our data support the idea that optimal reproductive investment<br />
should increase when costs experienced late in the reproductive cycle (breeding plus<br />
post-breeding costs) are higher than pre-breeding costs. This situation is exactly the<br />
one that we have found in the asp viper (see above results and Bonnet et al. 1994).<br />
Our results also allow us to develop this idea further. <strong>The</strong> evolution of semelparity<br />
can be viewed as a consequence of extreme capital breeding tactics, where most of<br />
the maternal somatic resources are invested during a single reproductive bout<br />
(Bonnet et al. 1998). Semelparity may be favoured when the sum of survival costs<br />
measured over a long timescale (3 to 4 years on average in our study model) are<br />
particularly high. This extreme reproductive tactic, observed almost exclusively in<br />
ecthotherms, may also be associated with components of reproductive costs that are<br />
independent of fecundity (Bull & Shine 1979; Olson et al. 2000).<br />
In the case of female aspic vipers, costs of reproduction are so high that most<br />
females produce only a single litter during their lifetimes. Especially if energy<br />
acquisition is required over a period of two years rather than one, the annual survival<br />
rates of females are so low (Figure 2) that few females survive long enough to<br />
produce a second litter. Although the survival cost in the year of litter production is <<br />
50%, the additional risks due to pre-breeding mortality (18 to 33%, depending on<br />
213
cycle length) and post-breeding mortality (33%) combine to make semelparity the<br />
norm for female vipers in our study population (Naulleau et al., in prep.). This result<br />
emphasises the importance of understanding all components of reproductive costs,<br />
not simply those that are paid during the actual "reproductive" bout. In our study<br />
animals, these additional components (mostly post-breeding costs) sum to at least as<br />
high a cost as the overt mortality risk experienced by a female in the year in which<br />
she produces offspring.<br />
More generally, methodologies for measurement of costs need to be<br />
evaluated carefully before comparisons can be made. This caveat extends to<br />
particular techniques as well as to timescales. For example, the popular technique of<br />
assessing avian costs through clutch-size manipulation after laying does not<br />
incorporate any effects of the additional clutch size on maternal mobility prior to<br />
laying. Such effects may well occur in birds (e.g., Lee et al. 1996), and are believed<br />
to be an important component of the <strong>total</strong> costs experienced by some reptiles (e.g.,<br />
Shine 1980; Seigel et al. 1987; Sinervo & DeNardo 1996).<br />
Similarly, our logistic regression detected a significant mortality cost of<br />
reproduction associated with maternal body condition after parturition (see above,<br />
and Figure 3). Post-parturient condition also affects maternal survival in two other<br />
species of snakes that are partly sympatric with aspic vipers, but in both cases the<br />
correlation is apparent in the few months following parturition (Vipera berus –<br />
Madsen & Shine 1993; Coronella austriaca – Luiselli et al. 1996). No such link is<br />
apparent within aspic vipers over that period (post-partum body condition did not<br />
affect a female viper’s probability of survival to the next season; N = 93, p = 0.31 -<br />
Naulleau et al. in prep.): the survival cost of lowered maternal condition is only seen<br />
over the ensuing 12 months. This contrast suggests that even when species display<br />
214
similar <strong>relationship</strong>s <strong>between</strong> reproductive effort and cost, the taxa may differ in the<br />
timescale over which such effects are manifested.<br />
<strong>The</strong>se kinds of complications do not invalidate broad-scale comparisons of<br />
costs: indeed, we enthusiastically endorse attempts to do so. <strong>The</strong>re will be many<br />
species that are phylogenetically distant from each other, but for which the form and<br />
timescale of costs are sufficiently similar that comparisons are relatively<br />
straightforward. For example, although ectothermy predisposes animals to capital-<br />
breeding, there are many income-breeders within this group also (e.g., short-lived<br />
lizards, James & Whitford 1994). Similarly, some endotherms rely upon stored<br />
“capital” for reproduction (e.g., Cherel et al. 1993; Cherel 1995) and some<br />
endotherms experience relatively delayed, long-term survival costs (McCleery et al.<br />
1996, Daan et al. 1996). Thus, there are many opportunities to carry out appropriate<br />
comparisons among suitably-matched groups of species. Many of the comparisons<br />
that we have made (such as capital versus income, or timescales for energy<br />
acquisition in birds versus snakes) are clearly continuous rather than dichotomies.<br />
Future research could usefully quantify such timescales.<br />
Given the logistical difficulties of the kind discussed here, however, it may also<br />
be worth investigating the massive potential of intrageneric (and even, intraspecific)<br />
comparisons to clarify costs of alternative life-history traits. For example, many<br />
reptile lineages display variation in traits (such as mean body sizes, degrees of<br />
sexual size dimorphism, reproductive mode) at these levels (Fitch 1981; Blackburn<br />
1982, 1985). This diversity, among taxa that are otherwise very similar, offers a<br />
particularly powerful opportunity to characterise and quantify the costs associated<br />
with phylogenetic shifts in traits of interest. Ultimately, such comparisons may be<br />
more revealing than those made <strong>between</strong> taxa that differ so substantially in the form<br />
215
and timescale of costs that it is difficult to overcome the confounding variables<br />
involved.<br />
Acknowledgements<br />
We thank S. Duret, M. Vacher-Vallas and L. Patard for help during field work.<br />
Financial support was provided by the Conseil Général des Deux Sèvres, the Centre<br />
National de la Recherche Scientifique (France) and the Australian Research Council.<br />
We also thank Rex Cambag for maintenance of the electronic material.<br />
216
IV. Les déterminants de la<br />
tendance semélipare:<br />
description et implications<br />
démographiques<br />
217
A. Résumé du Chapitre:<br />
Chez la vipère aspic, l’effort de reproduction est associé à des coûts très élevés<br />
s’exprimant sur un échelle de temps complexe. Dans notre population d’étude les<br />
contraintes climatiques et énergétiques vont ralentir les possibilités de reconstitution<br />
des réserves corporelles. Ces contraintes vont affecter la survie des reproductrices et<br />
limiter le nombre d’opportunités de reproduction avec une majorité de femelles<br />
semélipares. Toutefois certaines femelles réussissent à se reproduire plusieurs fois.<br />
La population des Moutiers est donc constituée une fraction d’individus<br />
“ semélipares ” en coexistence avec une minorité d’individus “ itéropares ”. Une telle<br />
situation offre une excellente opportunité pour examiner les déterminants de ces<br />
stratégies démographique contrastées.<br />
Dans un premier temps, nous avons cherché à tester les avantages sélectifs<br />
des modes de reproduction semélipares versus itéropares dans cette population.<br />
Nous avons voulu examiner si la “stratégie semélipare” était associée à un<br />
investissement reproducteur plus élevé illustrant ainsi la concentration de l’effort<br />
reproducteur maternelle sur une opportunité unique de reproduction (article 8). Nos<br />
résultats vont clairement à l’encontre de cette hypothèse : les femelles semélipares<br />
produisent des portées moins lourdes constituées de vipéreaux moins nombreux et<br />
plus légers que les femelles itéropares. La “stratégie” itéropare est donc la plus<br />
avantageuse en assurant un meilleur succès reproducteur à vie. Cependant elle<br />
reste minoritaire ce qui suggère l’existence de fortes contraintes sur l’expression de<br />
l’itéroparité. La comparaison des caractéristiques des femelles semélipares et<br />
itéropares suggère une origine environnementale de la seméliparité. En effet, les<br />
individus des deux stratégies ne diffèrent pas en taille, masse ou condition corporelle<br />
initiale. En revanche on détecte des différences significatives de condition pré/post<br />
218
parturition en faveur des itéropares. Le meilleur succès de ces femelles semblent<br />
donc directement lié à des conditions énergétiques favorables (prises alimentaires<br />
facultatives) qui vont influencer les caractéristiques de la portée. Cet apport<br />
d’énergie va aussi améliorer la condition de la mère après la mise bas et favoriser les<br />
reproductions futures.<br />
De plus, nos données soulignent l’existence de forte variations interannuelles<br />
dans la proportion d’individus semélipares. Ces variations sont étroitement corrélées<br />
aux fluctuations en nourriture observées dans l’année faisant suite à la reproduction<br />
et donc au conditions trophiques pendant la phase cruciale de récupération. Une<br />
dernière analyse nous permet de modéliser la probabilité d’être semélipare dans la<br />
population. L’expression de la seméliparité dépend ainsi de contraintes énergétiques<br />
intimement liées que sont l’état d’émaciation post mise bas (qui reflète le niveau<br />
d’alimentation et les conditions thermiques l’année de la reproduction) et les<br />
conditions trophiques l’année faisant suite à la reproduction (phase de récupération).<br />
Ces résultats invalident donc l’existence d’une “stratégie” semélipare au sens propre<br />
dans la population. La tendance semélipare reflète en fait un système<br />
d’investissement reproducteur extrême associé à des coûts post-reproducteurs dont<br />
l’amplitude est élevée et largement contrôlée par des facteurs environnementaux<br />
fluctuants. Les variations d’amplitude des coûts illustrent un système où les<br />
possibilités de récupération énergétiques sont très précaires et déterminantes de<br />
l’espérance de vie reproductrice des femelles.<br />
Dans un dernier volet, nous avons examiné l’impact populationnel de la<br />
stratégie reproductrice femelle en comparant la démographie des deux sexes (article<br />
9). La production de vipéreaux impose des contraintes et des contributions très<br />
différentes selon les sexes. L’investissement des mâles est réduit dans le temps et<br />
réclame moins d’énergie que celui des femelles. Mâles et femelles partagent la<br />
219
même niche écologique (même type de proie, habitat similaire), cette situation offre<br />
un bonne opportunité pour examiner les conséquences démographiques de<br />
divergences éco-physiologiques entre les sexes sans avoir à tenir compte des<br />
autres facteurs.<br />
Nos résultats montrent l’existence d’un système démographique hautement<br />
dimorphe avec d’importantes fluctuations dans la taille de la population femelle qui<br />
contrastent fortement avec une certaine stabilité des mâles. Les variations dans la<br />
population femelle reflètent les contraintes reproductrices maternelles (vitellogénèse,<br />
gestation) et illustrent le pas de temps pertinent sur le lequel les coûts de la<br />
reproduction se manifestent. Ainsi, la dynamique de la population femelle est<br />
intimement liée aux fluctuations en proies qui influencent positivement le recrutement<br />
de nouveaux adultes et qui affecte l’amplitude des coûts payés pendant les phases<br />
de récupération. Chez les mâles le système d’allocation de l’énergie est graduel et<br />
on n’observe pas de fluctuation majeure dans la taille de la population. Les mâles<br />
peuvent ajuster leur effort reproducteur en fonction de leur état physiologique, par<br />
exemple en arrêtant de chercher les femelles si leur condition corporelle est trop<br />
faible. Par contre, une fois engagées, les femelles doivent poursuivre leur effort de<br />
reproduction jusqu’aux mises bas au risque de tout perdre. Cette dichotomie dans les<br />
contraintes reproductrices se traduit par des patterns démographiques contrastés.<br />
220
B. Article 8<br />
Y-a-t’il un avantage à être semélipare?<br />
Comparaisons des tactiques démographiques<br />
de la vipère aspic (Vipera aspis)<br />
O. Lourdais 1 2 3 , X. Bonnet 1 , R. Shine 4 , & G Naulleau 1<br />
1 Centre d'Etudes Biologiques de Chizé, <strong>CNRS</strong>, 79360, Villiers en Bois, France<br />
2 Conseil Général Des Deux Sèvres, Rue de L’abreuvoir, 79021, Niort, France<br />
3 University of Poitiers, 40 avenue du recteur Pineau, 86022 Poitiers, France<br />
4 Biological Sciences A08, University of Sydney, NSW 2006, Australia<br />
Manuscrit en préparation...<br />
221
Résumé<br />
Chez la vipère aspic (Vipera aspis) les populations situées au Nord de l’aire de<br />
répartition présentent une majorité (72.5%) de femelles semélipares. Toutefois, une<br />
minorité de femelles (27.5%) sont itéropares et se reproduisent entre 2 et 4 fois. De<br />
façon surprenante l’effort de reproduction n’est pas supérieur chez ces individus<br />
semélipares. Au contraire, ces femelles semblent moins bien réussir leur<br />
reproduction que les femelles itéropares qui produisent des portées plus lourdes<br />
constituées de vipéreaux dont le poid moyen est plus élevé. En se reproduisant de<br />
façon répétée, les individus itéropares produisent un nombre supérieur de jeunes et<br />
bénéficient d’un meilleur succès reproducteur.<br />
Nos données indiquent qu’il n’existe pas deux “stratégies” au sens propre<br />
dans la population : la minorité d’itéropares est en fait composée d’indivividus ayant<br />
bénéficié d’un apport énergétique supérieur pendant la reproduction. Ces individus<br />
jouissent d’une meilleure condition coprorelle post-partum, une variable déterminante<br />
des probabilités de survie et de reproductions ultérieures. L’origine environnementale<br />
de la seméliparité est par ailleurs confirmée par l’étude des variations inter-annuelles<br />
dans la proportions de semélipares. Ainsi, pour une année donnée, la fraction<br />
d’individus semélipares est fortement dépendante des conditions d’alimentation de<br />
l’année faisant suite à la reproduction. Le fort degré de seméliparité observé dans<br />
cette population illustre les faibles possibilités de récupération énergetique après la<br />
mise bas et s’explique donc par des facteurs environnementaux dévaforables et/ou<br />
fluctuants.<br />
Mots clés: seméliparité, itéroparité, succès reproducteur<br />
222
Introduction<br />
L’extraordinaire diversité dans les stratégies reproductrices déployées par les êtres<br />
vivants reflète l’existence de contraintes environnementales. Dans un contexte<br />
limitant, les êtres vivants doivent ainsi faire face à des compromis ou “trade-off” entre<br />
des activités ou voies concurentielles que sont croissance, reproduction et survie<br />
(Stearns 92). Les différentes combinaisons entre les traits d’histoire de vie sont<br />
généralement interprétées comme autant de réponses évolutives permettant aux<br />
organismes d’optimiser leur succès reproducteur à vie et donc de “répandre”<br />
efficacement leurs gènes au sein des populations. Les organismes sont ainsi soumis<br />
à des pressions de sélection qui favorisent l’évolution d’un effort reproducteur<br />
optimum (Stearns 1992, Roff 1992). Cette répartition optimale de l’effort reproducteur<br />
est le résultat d’un compromis entre le succès attendu dans la reproduction courante<br />
et celui des reproduction futures.<br />
L’existence d’une dichotomie entre les organismes semélipares et d’autres<br />
itéropares constitue un des contrastes les plus surprenants dans l’orientation des<br />
stratégies reproductrices (Cole 1954). Alors que chez les premiers la vie<br />
reproductrice est constituée d’une succession d’épisodes reproducteurs, chez les<br />
seconds la reproduction est suivie de la mort de l’organisme. Les avantages de ces<br />
différentes stratégies ont été largement discutés par le passé (Cole 1954, Stearns<br />
1992). De nombreux modèles mathématiques ont été développés pour comprendre<br />
l’évolution de la seméliparité (Cole 1954; Gadgil & Bossert 1971; Charnov & Schaffer<br />
1973). Les travaux les plus récents reposent sur la confrontation de stratégies<br />
semélipares et itéropares définies par des paramètres démographiques (Young<br />
1990; Young & Auspurger 1991; Ranta et al. 2000a,b). Dans tous les cas ces<br />
modèles permettent l’existence d’un mode de reproduction caracterisé par une mort<br />
223
programmée. Seméliparité et itéroparité sont alors considerées comme des<br />
stratégies pré-existentes, entrant en compétition au sein de populations théoriques<br />
(Ranta et al. 2000a,b). Les résultats obtenus suggèrent l’influence de paramètres<br />
démographiques et environnementaux sur la sélection de l’une ou l’autre des<br />
stratégies. Ces travaux ont apporté des informations précises sur l’influence relative<br />
des taux de mortalités adulte et juvénile sur les avantages de l’une ou l’autre des<br />
stratégies. Ainsi, dans certaines circonstances bien définies, un “mutant” semélipare<br />
pourrait envahir la population et vice-versa (Ranta et al. 2000a,b). Cependant, les<br />
modèles élaborés reposent tous sur un déterminisme très simple de la partition de<br />
l’effort reproducteur (itéropare versus semélipare) en considérant l’existence d’une<br />
stratégie “résidente” confrontée à l’invasion d’une stratégie mutante. Il est important<br />
de signaler l’absence de support empirique de telle situation dans la nature.<br />
Notamment, la détermination de l’effort reproducteur et de sa partition dans la vie de<br />
l’organisme fait intervenir des soubassements physiologiques complexes reposant<br />
sur des actions génétiques variées (Sinervo & Svensson 1998). Un tel contexte rend<br />
donc peu réaliste les postulats d’un codage simplifié du mode de parité, implicite à la<br />
construction des modèles.<br />
L’étude de l’évolution du mode de parité et les traits d’histoire de vie associés<br />
a donc fait l’objet de nombreuses approches théoriques, en dehors de tout support<br />
phylogénétique. Si cette approche a permis de mieux comprendre les avantages et<br />
inconvénients de la seméliparité, la nature des pressions évolutives responsables de<br />
transition entre l’itéroparité et la seméliparité demeurent très largement inconnue<br />
(Crespi & Teo 2002). Plus récemment, quelques travaux de comparaisons<br />
interspécifiques ont été entrepris dans un cadre phylogénétique robuste. Ces<br />
approches ont été très fructueuses en révélant à la fois chez les plantes et les<br />
animaux l’existence d’un continuum, et non d’une dichotomie, entre seméliparité et<br />
224
itéroparité (Hautekèete et al. 2001; Crespi & Teo 2002). De plus, en comparant des<br />
espèces réparties sur un gradient entre des itéropares longévifs et des semélipares<br />
strictes, ces travaux apportent des informations précises sur les changements dans<br />
les traits d’histoire de vie associés à une transition vers la seméliparité. Au sein d’un<br />
tel gradient, les espèces itéropares présentant des populations à tendance<br />
semélipares, constituent des modèles d’études particulièrement intéressants. Par<br />
tendance semélipare on considère les espèces dont le mode de reproduction est<br />
l’itéroparité mais où de nombreux individus ne se reproduirent qu’une seule fois et<br />
présentent ainsi une trajectoire semélipare. En occupant une position charnière, ces<br />
situations rendent possible le test d’hypothèses sur les déterminants de la répartition<br />
de l’effort reproducteur et sur les possibles contraintes environnementales impliquées<br />
dans la transition vers des systèmes semélipares. En effet, une telle approche n’est<br />
pas envisageable chez les espèces semélipares strictes chez qui la mort des<br />
organismes est programmée après la reproduction.<br />
De telles opportunités existent chez certaines espèces de reptiles et<br />
notamment parmi les squamates (Madsen & Shine 1992b, Brown 1993, Bonnet et al.<br />
2002a). Par exemple, chez la vipère aspic, il existe une tendance semélipare<br />
marquée dans les populations située au Nord de l’aire de répartition (Bonnet et al.<br />
2002b) alors que dans les populations du Sud, les femelles sont longévives et<br />
itéropares (Zuffi et al. 1999). Nous avons exploité les données d’un suivi à long terme<br />
d’une population de vipères à tendance semélipare en examinant les questions<br />
suivantes :<br />
1) Les individus semélipares diffèrent-ils des itéropares en terme d’effort<br />
reproducteur. Notamment, la seméliparité est-elle associée à un<br />
investissement plus massif dans la reproduction ?<br />
225
2) Existe-t’il une variabilité inter-annuelle dans la proportion d’individus<br />
semélipares. Si oui, quelle est la nature des facteurs déterminants de<br />
l’expression de la tendance semélipare au sein de cette population.<br />
Matériels et méthodes<br />
Zone et espèce d’étude<br />
La vipère aspic est un serpent venimeux de taille moyenne (55 cm). Cette espèce<br />
présente un système d’allocation de l’énergie basé sur l’accumulation d’importantes<br />
quantités de réserves lipidiques qui servent de support à la reproduction (Bonnet<br />
1996). Nous avons étudié une population de vipères aspic dans l’Ouest de la France<br />
(Les Moutiers en Retz, 47 o 03N'; 02 o 00W') à proximité de la limite Nord de l’aire de<br />
distribution de l’espèce. Dans cette région, une à quatre années sont nécessaires<br />
pour accumuler des stocks de réserves suffisants pour la reproduction et les femelles<br />
présentent typiquement un cycle reproducteur supérieur à un an. L’investissement<br />
dans la reproduction, très élevé pour un vertébré, est associé à de forts coûts<br />
écologiques et énergétiques. Le nombre d’épisodes reproducteurs est réduit et il<br />
existe une tendance marquée vers la seméliparité (Bonnet et al 2002a). Cette<br />
situation contraste fortement avec les populations situées plus au Sud où les<br />
femelles sont longévives et présentent un cycle reproducteur annuel (Zuffi et al.<br />
1999).<br />
Les adultes se nourrissent de micromammifères et essentiellement de<br />
campagnol (Microtus arvalis Pallas) dont les populations présentent d’importantes<br />
fluctuations inter-annuelles. Plusieurs travaux suggèrent une influence complexe de<br />
ces variations trophiques sur la dynamique des réserves et l’investissement<br />
reproducteur (Bonnet et al 2001b, Lourdais et al. 2002b). Le climat est de type<br />
226
océanique tempéré et avec des variations des conditions thermiques significatives<br />
(Lourdais et al. 2002b) En combinaison avec la ressource trophique les variations<br />
climatiques affectent de nombreux aspect du cycle de vie des vipères. Notamment<br />
les conditions thermiques pendant la gestation semblent contraignantes et les mises-<br />
bas sont observées fin septembre soit deux mois plus tard que dans les populations<br />
méridionales. De 1992 à 2000 la zone d’étude a été patrouillée par 1 à 4 personnes<br />
pendant la saison active (Mars à Avril). Les serpents observés pendant les phases<br />
de thermorégulation ont été capturés à vue, sexés, marqués individuellement et<br />
relachés sur le lieu exact de capture. Les indices de consommation de proies ont été<br />
relevés pour permettre l’estimation de l’abondance en campagnols. Des informations<br />
détaillées sur les méthodes de marquage et l’effort de recherche sont diponibles<br />
dans des différents travaux publiés sur cette même population (Bonnet et al. 2001b,<br />
2002a, Lourdais et al. 2002b)<br />
Mesures realisées<br />
Les analyses de cet article reposent sur un jeu de données de 2200 captures<br />
provenant de 524 femelles adultes différentes. Tous les serpents capturés ont été<br />
pesés au gramme près et mesurées à 0.5 cm près. Le statut reproducteur a été<br />
déterminé en utilisant plusieurs méthodes complémentaires : au début du printemps,<br />
les femelles présentant une condition corporelle supérieure au seuil sont considérées<br />
comme reproductrices (voir Bonnet et Naulleau 1994 pour les détails sur la<br />
méthode). Plus tard dans l’année, la palpation de l’abdomen a permis de détecter et<br />
compter des follicules en croissance (vitellogénèse) ou des embryons (gestation).<br />
Les femelles reproductrices capturées peu de temps avant la mise bas (fin de l’été)<br />
ont été ramenées en captivité afin d’obtenir des informations précises sur la<br />
reproduction (voir section Méthodes d’étude p 37).<br />
227
Estimation de la survie<br />
Les individus non-recapturés au-delà d’une période de deux ans sont considérés<br />
comme morts. Cette méthode repose sur une série d’arguments complémentaires :<br />
i) Le site d’étude est entouré par des zones défavorables aux vipères qui réduisent<br />
les possibilités d’émigration (Vacher-Vallas, Bonnet & Naulleau 1999). ii) Cette<br />
espèce est très philopatrique (Naulleau,et al. 1996) et les données de radiotracking<br />
ont révélé des déplacements très réduits par rapport au lieu de capture initiale<br />
(toujours
du suivi (9 ans d’études) et la forte tendance semélipare dans la population. Le<br />
succès reproducteur à vie des vipères a été estimé en condidérant les individus dont<br />
l’histoire reproductrice est connue et dont l’ensemble des mises bas a été obtenu en<br />
captivité. Enfin, la proportion annuelle d’individus semélipares et itéropares a été<br />
calculée de 1992 à 1998.<br />
Une fois la classification des femelles réalisée, nous avons cherché à<br />
comparer les deux stratégies en examinant notamment les traits morphologiques<br />
maternels (taille, condition corporelle) et les caractéristiques des portées (pour les<br />
femelles recapturées avant la mise bas). Pour réaliser ces comparaisons, nous<br />
avons uniquement considéré les données de première reproduction des itéropares et<br />
il n’y a donc pas de réplication de données pour un même individu. Nous avons par<br />
ailleurs comparé les dynamiques pondérales chez les femelles itéropares et<br />
semélipares depuis le printemps jusqu’à la mise bas. La population d’étude est<br />
affectée par de nombreuses variables environnementales (Bonnet et al. 2001b<br />
Lourdais et al. 2002b) et les comparaisons des stratégies reproductrices ont été<br />
réalisées en tenant compte de la variabilité inter-annuelle. Ainsi l’étude d’un trait<br />
donné (variable dépendante) a été effectuée en considérant comme facteurs 1)<br />
l’année de comparaison et 2) la nature de la stratégie (itéropare ou semélipare).<br />
Lorsque le trait étudié est influencé pas la taille maternelle, ce paramètre a été<br />
introduit comme co-facteur. Enfin nous avons testé l’influence de la nourriture sur<br />
l’expression de la stratégie reproductrice en utilisant l’indice d’abondance en<br />
campagnol déjà décrit par ailleurs (section Méthodes d’étude p 37, Lourdais et al.<br />
2002b).<br />
Résultats<br />
1) Estimation de la tendance semélipare<br />
229
Sur les 9 ans d’étude, nous avons pu dénombrer 241 individus semélipares et 91<br />
iteropares. La classe des semélipares comprend majoritairement des animaux se<br />
reproduisant l’année n et jamais revus ensuite (181). Une minorité d’individus (60) a<br />
cependant été observée l’année suivante (année n+1), principalement au printemps<br />
et plus jamais par la suite. Considérant notre incertitude sur l ‘estimation de la survie<br />
(période de 2 ans sans observation), il n’est pas possible d’identifier le moment où<br />
s’exerce la mortalité et donc de définir deux classes d’individus semélipares. Ainsi,<br />
parmis les 181 individus jamais recapturés, de nombreuses femelles ont<br />
probablement survécu jusqu’au printemps suivant en échappant à l’observation (la<br />
capturabilité des non-reproductrices étant toujours inférieure à celle des<br />
reproductrices, Lourdais et al. 2002d). Nos résultats indiquent clairement un forte<br />
tendance semélipare dans la population (72.5 % des individus). De plus, nous avons<br />
détecté des variations inter-annuelles significatives dans la proportion des individus<br />
semélipares et itéropares ( χ² de Pearson=25.16, dl=6, p=0.0003, voir Table 1).<br />
Morphométrie:<br />
2) Les individus semélipares sont-ils différents des itéropares?<br />
Nous avons tout d’abord examiné si les individus itéropares et semélipares diffèrent<br />
dans leur morphologie au sortir de l’hivernage (printemps). Nos analyses suggèrent<br />
l’absence de différences significatives en longueur corporelle initiale et en masse<br />
corporelle initiale (Table 2). En conséquence, il n’existe pas de différence<br />
significative de condition corporelle entre les deux catégories de femelles (analyse de<br />
covariance avec la masse ajustée par la taille, voir Table 2). Les femelles itéropares<br />
et semélipares sont donc indistinguables en début de reproduction en terme de<br />
morphométrie ou d’état des réserves lipidiques (condition corporelle).<br />
230
Table 1. Proportion d’individus itéropares et semélipares en fonction de l’année. Les données<br />
s’arrêtent en 1998 en raison du pas de 2 ans nécessaire pour statuer de la survie de vipères. Dans<br />
cette analyse nous avons considéré la seméliparité et l’itéroparité comme deux strategies sensu<br />
stricto et un même individu itéropare peut contribuer à plusieurs années différentes.<br />
Année Semélipares Itéropares Total Proportion<br />
1992 12 9 21 0.57<br />
1993 59 16 75 0.78<br />
1994 61 39 100 0.61<br />
1995 8 18 26 0.30<br />
1996 26 30 56 0.46<br />
1997 63 42 105 0.60<br />
1998<br />
10<br />
Investissement reproducteur<br />
10<br />
Nous avons ensuite étudié les paramètres de l’effort reproducteur selon la stratégie<br />
reproductrice. En accord avec les précédents travaux réalisés sur cette population,<br />
nos analyses indiquent l’existence de variations inter-annuelles dans la plupart des<br />
paramètres mesurés (voir Table 2). De facon suprenante, les femelles semélipares<br />
produisent des portées significativement moins lourdes que les itéropares (Figure 1,<br />
Table 2). Des résultats similaires sont obtenus en considérant la masse de vipéreaux<br />
viable produite ou la masse relative de la portée (Analyse de covariance avec<br />
comme variable dépendante la masse de la portée et comme covariable la masse<br />
post-partum de la mère). En revanche, nous ne détectons pas de différence<br />
significative dans les tailles de portées ou bien dans le nombre de vipereaux viables<br />
produits (voir Table 2). Enfin l’analyse des caractéristiques des jeunes viables<br />
produits indique que les portées des femelles itéropares sont constituées de jeunes<br />
significativement plus lourds (6.7g) que ceux des femelles semélipares (6.2g, voir<br />
231<br />
20<br />
0.50
Table 2) alors que l’on ne détecte pas de différence dans la taille des nouveau-nés<br />
produits (F(1,79)=1.68; p
Table 2. Suite<br />
Traits facteurs co-facteur F n p Valeurs<br />
iteropare semélipare<br />
Masse AN SVL 8.30 88
Succès reproducteur<br />
Les femelles semélipares ne produisent pas plus de vipéreaux que les femelles<br />
itéropares lors de leur unique mise bas. Les femelles itéropares en se reproduisant<br />
plusieurs fois dans leur vie vont donc avoir un nombre plus élevé de descendants<br />
que les semélipares (10.2 versus 4.3, ANOVA F(1, 73)=40.47, p
masses pre-partum avec des femelles itéropares plus lourdes (masse ajustée :<br />
119.8, versus 107.4; voir Table 2). De même, les femelles itéropares bénéficient<br />
d’une masse corporelle supérieure après la mise bas (masse ajustée: 68.9g versus<br />
63.6 Table 2). Les itéropares présentent donc une condition corporelle post-partum<br />
supérieure aux semélipares (résidus : 0.064 versus -0.015)<br />
Considérant ces éléments, nous avons voulu déterminer l’influence de<br />
l’abondance en proies sur les variations annuelles dans les proportions d’individus<br />
semélipares. Nos résultats suggèrent une absence de relation entre la proportion de<br />
semélipares et l’abondance en proies dans l’année de la reproduction (r=0.40,<br />
r 2 =0.16, n=7, p=0.37). Par contre, il existe une relation négative significative entre la<br />
proportion d’individus semélipares dans une année et l’abondance en nourriture<br />
l’année suivante (r= 0.84, r²=0.71, F(1,5)=12.672 p
Proportion de semelipares (année n)<br />
0.95<br />
0.85<br />
0.75<br />
0.65<br />
0.55<br />
0.45<br />
0.35<br />
0.06 0.12 0.18 0.24 0.30 0.36 0.42<br />
Abondance en proies dans l'année n+1<br />
Figure 3. Relation entre la proportion annuelle de femelles semélipares et l’abondance en proies<br />
l’année suivante.<br />
Discussion<br />
Cette étude apporte des informations paradoxales sur la tendance semélipare de la<br />
vipère aspic. Notre suivi indique que sur 9 ans d’études, cette “stratégie“ est<br />
largement dominante dans la population (72.5% des individus). Pourtant plusieurs<br />
de nos analyses indiquent que ce mode de reproduction est bien moins bénéfique<br />
que l’itéroparité et ce pour plusieurs raisons. Tout d’abord, les femelles semélipares<br />
n’investissent pas plus d’énergie dans la production d’un grand nombre de<br />
vipéreaux. Ainsi il n’existe pas de différence de taille de portée selon les stratégies.<br />
De façon surprenante, la masse des portées est supérieure chez les femelles<br />
itéropares et, par conséquent, ces dernières produisent des vipereaux<br />
significativement plus lourds et donc en meilleur condition corporelle à la naissance.<br />
Ces différences sont importantes car, chez les serpents, la condition corporelle à la<br />
naissance est un indice de l’état des réserves qui influence ultérieurement la<br />
croissance post-natale, le type de proies ingérables et peut-être la survie des<br />
236
jeunes. En dépit de l’absence d’informations sur la survie des nouveaux-nés, nos<br />
résultats suggèrent néanmoins des différences dans la qualité des jeunes produits.<br />
En parallèle avec ces variations de qualité, les femelles itéropares produisent un<br />
nombre de jeunes toujours plus élevé que les semélipares. De plus, il existe une<br />
relation positive très nette entre le nombre de reproduction et le nombre <strong>total</strong> de<br />
vipéreaux produits.<br />
Dans un tel contexte, il est légitime de s’interroger sur les avantages de la<br />
stratégie semélipare qui demeure pourtant majoritaire dans la population. Nos<br />
analyses indiquent que les individus semélipares ne diffèrent pas des itéropares à<br />
l’entrée dans la reproduction (pas de différence de condition corporelle ou de taille<br />
des portées). Le fait que l’on détecte des différences dans les masses pre et post<br />
partum ainsi que dans les masses des portées et des vipéreaux (Table 2)<br />
suggèrent des prises alimentaires chez les itéropares. Ainsi ces individus semblent<br />
bénéficier d’un apport énergétique supérieur pendant la reproduction (prises de<br />
nourriture facultatives) et par conséquent d’une meilleure condition corporelle post -<br />
partum. La condition corporelle est un très bon indice des possibilités de<br />
récupération ultérieure des femelles vipères et nos données soulignent donc le rôle<br />
majeur des prises alimentaires sur l’état d’émaciation des femelles et les possibilités<br />
de reproduction. En plus de l’état d’émaciation, il existe une relation très forte entre<br />
la proportion annuelle d’individus semélipares et la quantité de nourriture disponible<br />
l’année suivante (i.e. pendant la phase de récupération). Ces résultats démontrent<br />
donc une origine profondément environnementale de la tendance semélipare. Ainsi,<br />
il n’existe pas de stratégies semélipare et itéropare pures et la seméliparité apparaît<br />
plutôt comme le reflet des fortes contraintes énergétiques de la reproduction. Chez<br />
la vipère aspic l’investissement reproducteur est très élevé et basé sur la<br />
mobilisation des réserves lipidiques et protéiques. Les changements<br />
237
comportementaux et morphologiques (thermorégulation accentuée, déplacements<br />
réduits) associés à la reproduction aboutissent à une réduction des prises<br />
alimentaires, notamment pendant la gestation (Lourdais et al. 2002a). Dans cette<br />
population, il existe de très fortes variations inter-annuelles dans l’abondance des<br />
proies qui vont directement influencer le degré de prises alimentaires facultatives<br />
l’année de la reproduction (Lourdais et al. 2002b). La fraction d’individus itéropare<br />
correspond donc à une minorité d’individus (chanceux) qui ont pu bénéficier de<br />
prises alimentaires un peu plus nombreuses. Les différences de masse pre-partum<br />
(107.4g vs 119.4g) suggèrent un apport énergétique “réduit” probablement lié à la<br />
consommation de une, peut-être deux proies en plus. Ces quelques prises<br />
alimentaires semblent néanmoins exercer une influence déterminante sur l’état<br />
d’amaigrissement et la survie post mise bas. Outre les prises alimentaires l’année<br />
de la reproduction, les conditions trophiques l’année suivante vont avoir une<br />
influence majeure sur la proportion d’individus semélipares. Ainsi cette proportion<br />
sera élevée lorsque l’année de la reproduction est suivi par une année à faible<br />
abondance en proies. En revanche la fraction de semélipares sera plus réduite<br />
lorsque la nourriture est abondante l’année suivante. La probabilité d‘être<br />
semélipare une année est donc affectée par la combinaison de facteurs : la<br />
condition corporelle post-partum et les conditions trophiques de l’année suivant la<br />
reproduction. Des travaux récents (lourdais et al. 2002b) montrent que la condition<br />
corporelle post-partum est elle même une variable complexe qui intègre les prises<br />
alimentaires facultatives mais aussi les conditions thermiques pendant la gestation.<br />
Notamment les coûts métaboliques sont plus élevés lorsque les conditions<br />
climatiques offrent des opportunités de thermorégulation supérieures aux femelles<br />
gestantes. Ainsi, pendant les étés chauds, la durée de gestation est plus courte<br />
mais le niveau de catabolisme protéique supérieur. La probabilité d’être semélipare,<br />
238
intimement liée au niveau d’amaigrissement post-partum, est donc influencée par<br />
des interactions complexes entre trois variables environnementales :<br />
i) Les prises alimentaires pendant la reproduction en combinaison avec ii) les<br />
conditions thermiques de la gestation vont déterminer la condition corporelle post-<br />
partum. iii) L’abondance en proies l’année suivant la reproduction va directement<br />
influencer les possibilités de récupération des femelles.<br />
Ces résultats sont très informatifs pour bien comprendre la stratégie<br />
reproductrice de la vipère aspic. Tout d’abord, la proportion d’individus semélipares<br />
et itéropares est calculée à partir du nombre d’individus observés au printemps<br />
(période principale des captures, Lourdais et al. 2002d). La capturabilité des vipères<br />
est ensuite plus réduite dans l’année et la fraction d’individus recapturés pour les<br />
mises bas ne représente qu’une minorité des individus reproducteurs observés dans<br />
l’année. La mortalité des vipères reproductrices implique des coûts de la<br />
reproduction direct (prédation pendant la vitellogénèse, la gestation) combinés avec<br />
des coûts de post-reproduction (mort liée à une situation énergétique post-partum<br />
critique) (Bonnet et al. 2000a, 2002a). En l’absence d’estimation du niveau de<br />
prédation, ces deux composantes semblent a priori difficiles à départager.<br />
Cependant nos résultats indiquent clairement que les contraintes de la reproduction<br />
génèrent principalement des coûts de post-reproduction. En effet, la vitellogénèse<br />
entraine la mobilisation des réserves. La gestation, en imposant des régimes<br />
thermiques élevés, va entrainer l’épuisement des réserves et un important<br />
catabolisme protéique. Ces contraintes énergétiques s’illustrent par un fort niveau<br />
d’émaciation après la parturition. La relation très forte entre la proportion d’individus<br />
semélipares (une variable démographique) et l’abondance en proies l’année<br />
suivante indique donc que la plupart des individus survivent jusqu’à la mise bas et<br />
239
que la mortalité va s’exprimer pendant la phase de recupération au printemps<br />
suivant (la mortalité pendant l’hivernage étant très réduite, données non publiées).<br />
Notre étude suggère donc qu’il n’existe pas de stratégie semélipare ou<br />
itéropare au sens propre dans la population des Moutiers. La reproduction est<br />
associée à des coûts énergétiques relativement indépendants de la taille des<br />
portées produites (Lourdais et al. 2002a, Ladyman et al. soumis). Le fort<br />
amaigrissement qui en résulte et les faibles possibilités de récupération vont aboutir<br />
à une survie très faible après la mise bas. La tendance semélipare de cette<br />
population reflète donc l’existence de coûts énergétiques et écologiques à la fois<br />
élevés et fixes s’exprimant dans un contexte environnemental fluctuant et<br />
contraignant.<br />
240
C. Article 9<br />
Do sex divergences in reproductive eco-<br />
physiology translate into dimorphic<br />
demographic patterns?<br />
Olivier Lourdais 1 2 3 , Xavier Bonnet 1 , Dale DeNardo 4 , Guy Naulleau 1<br />
1 Centre d'Etudes Biologiques de Chizé, <strong>CNRS</strong>, 79360, Villiers en Bois, France<br />
2 Conseil Général Des Deux Sèvres, Rue de L’abreuvoir, 79021, Niort, France<br />
3 University of Poitiers, 40 avenue du recteur Pineau, 86022 Poitiers, France<br />
4 Department of Biology, Arizona State University, Tempe, AZ, 85287-1501,USA<br />
Accepted for publication in Population Ecology<br />
241
Abstract<br />
We examined the influence of sex divergences in reproductive role and physiology on<br />
catchability and demographic patterns in a closed population of aspic viper (Vipera<br />
aspis Linné). During eight years, there were 4800 captures of 988 adults. In both<br />
sexes, captures were more frequent in spring when climatic conditions and<br />
reproductive activities impose extended basking periods that make animals more<br />
detectable. On average, males were captured more than females, reflecting intense<br />
sexual activity (i.e., mate searching) in spring. Reproductive females were more<br />
catchable than non-reproductive females, illustrating a major increase in basking<br />
behaviour associated with reproduction.<br />
Estimates of population size revealed a sexually dimorphic demographic<br />
system with marked year-to-year fluctuations in females contrasting with a more<br />
stable male population. This sex difference in population dynamic reflects sex<br />
divergences in the acquisition and allocation of energy for reproduction. In both<br />
sexes reproduction is fuelled by body reserves. Females, however, need to<br />
accumulate substantial body reserves to reach a high body condition threshold prior<br />
to reproduction, while the male pattern of energy allocation is more gradual (i.e., no<br />
fixed threshold). In addition, reproduction entails major survival cost in females (i.e.,<br />
most females reproduce just once), whereas males are generally annual breeders.<br />
As a consequence of this sex divergence, <strong>food</strong> abundance, through its direct effect<br />
on body store dynamic, influenced major demographic parameters of females (e.g.,<br />
proportion of reproducing individuals, annual changes in population size) but not<br />
males.<br />
Keywords: catchability; population size; capital-breeding; snake.<br />
242
Introduction<br />
Descriptions of variations in life history traits and their causations constitute a<br />
fundamental theme in the study of evolution and adaptation (Stearns 1992). Accurate<br />
quantification of variation expressed in wild populations often requires capture and<br />
recapture of marked individuals over periods that are biologically relevant to the<br />
species’ generation time. <strong>The</strong>refore, estimates of population parameters such as<br />
growth rate, survival, and population size necessitate long-term mark-recapture<br />
techniques (Southwood 1988). While requiring extensive effort and time<br />
commitment, longitudinal approaches provide a multitude of benefits that have been<br />
extensively illustrated in different fields of ecology (Tinkle 1979). Data gathered in<br />
long-term studies permit the examination of individual reproductive success, survival<br />
and possibly life-time reproductive success (e.g., the basic raw material for natural<br />
selection, Clutton-Brock 1988). Additionally, longitudinal work provides an opportunity<br />
to correlate population characteristics with environmental factors such as <strong>food</strong><br />
availability, predator abundance, or climatic fluctuation that affect life history traits<br />
(Ballinger 1977; Seigel & Fitch 1985).<br />
Unfortunately, for logistical reasons, long-term mark-recapture studies are<br />
usually not feasible. Individuals of some species are too small to permit marking or<br />
too secretive to catch a reasonably large subset of a population. Similarly, recapture<br />
avoidance and long-distance displacement invalidate or at least seriously complicate<br />
any estimate of population size or mortality (Nichols et al. 1987; Brodie 1989; Cooper<br />
et al. 1990; Lebreton et al. 1992; Shine & Schwarzkopf 1992; Houston & Shine<br />
1994). Due to their cryptic behaviour, snakes are superficially poor models for<br />
recapture studies (Seigel 1993; St Girons 1996). This impression, however, is not<br />
necessarily valid as some species fit within the requirements outlined above.<br />
243
Temperate viperid snakes, for example, are typical sit and wait predators that can be<br />
locally abundant, and a number of field studies have been successfully conducted on<br />
these animals (Saint Girons 1949, 1952, 1957a,b, 1975; Fitch 1960; Klauber 1972;<br />
Brown 1991; Madsen & Shine 1993; Martin 1993). Snakes from temperate climates<br />
may be especially conducive to studies based on recaptures. While snakes in<br />
general are extremely secretive animals and frequently go undetected, species in<br />
cooler climates often must bask in the sun to meet the thermal requirements of major<br />
physiological processes such as digestion, ecdysis, and particularly reproduction<br />
(Huey 1982, Peterson et al. 1993). In males and females, reproduction entails<br />
alterations in activity pattern (e.g., mate searching, increased basking activities) and<br />
this may render the snakes more visible (Bonnet et al. 1999b). <strong>The</strong>refore, we<br />
hypothesise that any shift observed in catchability will be directly linked to underlying<br />
physiological changes imposed by reproduction and/or foraging activities. That is,<br />
capture rates not only provide raw data for estimating demographic parameters such<br />
as population size, but also provide information on the thermal and reproductive<br />
biology of the species under study.<br />
Another interesting feature of studying temperate viperid snakes is that they<br />
offer extreme examples of capital-breeding systems in that body reserves often<br />
constitute the primary fuel for reproductive activities (Seigel & Ford 1987; Naulleau<br />
and Bonnet 1996). Though fat stores are probably important both for males and<br />
females (Saint Girons 1957a,b; Olsson et al. 1997), proximate divergences exist in<br />
term of patterns of energy allocation towards reproduction. Most notable is the length<br />
of the reproductive cycle where females are generally pluriannual and males annual<br />
breeders (Saint Girons 1957b; Andren & Nilson 1983; Seigel & Ford 1987; Naulleau<br />
et al. 1999). Such a situation is particularly well described for the aspic viper (Vipera<br />
244
aspis Linné), probably the most intensively studied European snake (Saint Girons<br />
1952, 1996; Naulleau 1997, Zuffi et al. 1999).<br />
Marked sexual divergences in reproductive biology are likely to entail major<br />
ecological repercussions in term of sex specific catchability and possibly<br />
demographic patterns. We tested for sex-differences in catchability patterns and<br />
population dynamics of the aspic vipers using a data set from an eight-year mark-<br />
recapture study of a closed population in west-central France that is characterised by<br />
strong annual fluctuation in prey abundance (Bonnet et al. 2001b, Lourdais et al.<br />
2002b). We also examined whether any detected differences were congruent with<br />
pre-existing knowledge of the ecophysiology of this species.<br />
Material and methods<br />
Study species<br />
<strong>The</strong> aspic viper is a small viviparous snake of the western-Palearctic region and<br />
locally abundant in west-central France at the northern limit of its distribution.<br />
Females mature at 40 cm snout-vent length (SVL), which is attained in 2.5 to 3.5<br />
years (Bonnet et al. 1999a). Females are typical capital breeders that delay<br />
reproduction until they have amassed enough energy-reserves to reach a high body<br />
condition threshold (Naulleau & Bonnet 1996). In this area, the female reproductive<br />
cycle is longer than annual (Saint Girons 1957a,b; Bonnet & Naulleau 1996; Naulleau<br />
et al. 1999), leading to the coexistence of sub-populations of reproducing females<br />
and non-reproducing ones. Reproductive activity in females imposes marked<br />
behavioural changes that distinguish them from males and non-reproducing females.<br />
Notably, females spend more time basking from the onset of follicle production<br />
245
(March) through parturition (late August) to meet the metabolic requirements of<br />
vitellogenesis and gestation. In addition, females are more sedentary during<br />
gestation (Naulleau et al. 1996). <strong>The</strong>se changes infer substantial survival costs of<br />
reproduction and therefore most (>75%) female vipers reproduce no more than one<br />
time in their lifetimes (Bonnet et al. 2000a, 2002a).<br />
Male aspic vipers are also capital breeders in that fat store are the sole source<br />
of energy during the sexual vernal anorexia. In contrast with females, reproductive<br />
investment is temporally reduced (broadly six to eight weeks in spring) and mainly<br />
concentrated on mate searching activities. Males do not have to reach a fixed body<br />
condition to engage in reproductive activities, as reproductive effort is adjusted to<br />
their body reserves (Aubret et al. 2002). As a consequence, the reproductive cycle is<br />
generally annual in this sex (Vacher-Vallas et al. 1999). Since the smallest male<br />
found copulating (with sperm transmission) was 36.5 cm SVL, all individuals longer<br />
than this were considered adults.<br />
Study site and methods<br />
<strong>The</strong> study site is in west-central France near the village of Les Moutiers en Retz<br />
47 o 03N'; 02 o 00W'). It is a 33-hectare grove with a mosaic of meadows and<br />
regenerating scrubland. <strong>The</strong> site is characterised by a temperate oceanic climate.<br />
From 1992 to 1999, one to four people patrolled on almost every favourable day<br />
(sunny and partially cloudy, or cloudy with air temperature above 15°C)<br />
encompassing the vipers’ annual activity period: late February to late October.<br />
Although variations in searching effort occurred <strong>between</strong> years due to climatic<br />
fluctuations, effectiveness in locating vipers and the searching method were largely<br />
homogenous through the study period (except in 1992 due to a searching effort<br />
biased toward females and large males) thanks to the extended searching period. As<br />
246
a result, the catchability pattern observed pooling all the years is highly consistent<br />
with the catchability pattern observed within each year (unpublished). Total searching<br />
effort exceeded 4,000 hours and represented more than 670 “searching-days”.<br />
Snakes were caught by hand, sexed by eversion of the hemipenes, weighed to the<br />
nearest g with an electronic scale, and measured (SVL, and <strong>total</strong> length) to the<br />
nearest 5 mm. Over 1,000 adult and sub-adult vipers have been marked using<br />
passive integrated transponder (PIT) tags (Sterile transponder TX1400L, Rhônes<br />
Mérieux, 69002 LYON France, product of Destron/IDI Inc). Upon capture, each snake<br />
was color-marked on the back to avoid short-term recaptures and thus minimise<br />
disturbance. All snakes were released at their point of capture.<br />
Female reproductive status was determined using two methods. First, at the<br />
beginning of vitellogenesis, a female with a body condition (mass scaled by size)<br />
greater than a pre-determined threshold was considered reproductive (see Bonnet et<br />
al. 1994; Naulleau & Bonnet 1996, for validity of the method). Second, from mid-<br />
vitellogenesis (May) to the end of gestation (late August) reproductive status was<br />
easily determined either by palpation of follicles and / or embryos or by records of<br />
parturition (Fitch 1987; Naulleau & Bonnet 1996).<br />
Catchability and population size<br />
Different measurements of catchability were used in this study. First, for individuals<br />
marked at the onset of the study (1992 and 1993), we estimated long-term<br />
catchability by determining the number of consecutive years that an individual was<br />
observed. Secondly, we examined intra-annual catchability by defining eighteen<br />
successive two-week capture-recapture sessions from March to November. <strong>The</strong>se<br />
sessions were equally divided into three consecutive periods (broadly the spring,<br />
summer, and fall seasons) that match with major events in the reproductive cycle<br />
247
(see Table 1). For each individual, we calculated the mean number of captures per<br />
session for each season (seasonal capture rate) and for the entire year (annual<br />
capture rate).<br />
Table 1. Biological cycle of the aspic viper (from Bonnet 1996) and organisation of the eighteen<br />
capture sessions (RF, reproducing females; NRF, non-reproducing females; M, males).<br />
PERIOD<br />
RF<br />
NRF<br />
M<br />
Capture<br />
sessions<br />
1 SPRING 2 SUMMER 3 FALL<br />
vitellogenesis<br />
fat store recovery<br />
sexual anorexia<br />
ovulation, gestation<br />
fat store recovery<br />
fat store recovery<br />
parturition<br />
fat store recovery<br />
fat store recovery<br />
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18<br />
Month Mar Apr May Jun Jul Aug Sept Oct Nov<br />
Because capture occurrence typically follows a Poisson distribution, we tested<br />
the effect of different explanatory variables (body size, sex, and reproductive status)<br />
using multiple Poisson regression. For descriptive purpose, the effect of body size<br />
was also examined using three size classes. <strong>The</strong> variable of interest was<br />
standardised (Z = (X - mean value) / SD) so that the distribution had a mean of zero<br />
and a SD of one. <strong>The</strong> three size classes were: small (Z
Population estimates were calculated using CAPTURE (Otis et al. 1978). Data<br />
from 1992 were excluded from the analysis because of low searching effort. <strong>The</strong><br />
model used assumes a closed population (e.g., no birth, death, or migration) and is<br />
appropriate for use in a study covering a short time (Otis et al. 1978). Each two-week<br />
period was considered a capture session. <strong>The</strong> analysis was restricted to spring<br />
(March – May), since the survival rate is high (>0.8, unpublished data). We found no<br />
evidence of emigration and any snakes not captured over a long period (> two years)<br />
was considered dead (see Naulleau et al. 1996, and Vacher-Vallas et al. 1999). Birth<br />
did not influence our analysis because only adults were considered and neonates<br />
require at least 2.5 years to reach maturity (Bonnet et al. 1999a). Finally, CAPTURE<br />
provides the opportunity to test different models including heterogeneity of capture<br />
probabilities in populations (Mh), time-specific variation in probabilities of recapture<br />
(Mt), behavioural response after initial capture (Mb), and combinations of these<br />
models. In every case, the first model suggested by goodness-of-fit tests was<br />
adopted (Chao et al. 1992). Annual changes in population size (year n) were<br />
calculated from spring population size estimate in year n+1 minus spring population<br />
size estimate in year n. Estimates were performed in spring, just after hibernation.<br />
Because mortality during hibernation is extremely low in our population<br />
(unpublished), the difference <strong>between</strong> “year-n+1” and “year-n” estimates corresponds<br />
to the changes that occur over the active season (spring to hibernation) of the year n.<br />
We analysed the influences of demographic changes occurring in a year n on the<br />
operational sex ratio of the population in the following year. In the aspic vipers, sex is<br />
genetically determined, and analysis were not confounded by the effects of<br />
environmental variables on primary sex ratio. Estimates of sex ratio (SR) and<br />
operational sex ratio (OSR) were calculated as follow: SR = <strong>total</strong> number of adult<br />
males / (<strong>total</strong> number of adult males + <strong>total</strong> number of adult females); OSR = <strong>total</strong><br />
249
number of adult males / (<strong>total</strong> number of adult males + <strong>total</strong> number of reproducing<br />
females). All statistics were performed using Statistica 6.0.<br />
Results<br />
Recapture rates and long-term catchability<br />
During the course of the study, 988 adult snakes were marked (463 females and 525<br />
males). <strong>The</strong> recapture rate (e.g., the percentage of snakes contacted at least once<br />
after the initial capture) was 76.6% and the cumulative number of captures and<br />
recaptures was 4,723. <strong>The</strong> number of individuals captured varied from year to year<br />
(Table 2) with the highest number of animals observed in 1994. Considering only<br />
snakes marked at the onset of the study (1992-1993), most individuals (70.4%) were<br />
observed during a single or two consecutive years and only a limited number (
individual respectively – Marti 1990). This effect held true when sex was taken into<br />
account (interaction <strong>between</strong> sex and body size : Wald χ²=3.2, df=2, p=0.21). Annual<br />
capture rates were significantly higher in males than females (2.51 versus 2.16; Wald<br />
χ²=5.79, df=1, p
Capture rate<br />
2.0<br />
1.5<br />
1.0<br />
0.5<br />
0.0<br />
***<br />
***<br />
Spring Summer Fall<br />
Season<br />
NS<br />
Male<br />
Female<br />
Figure 1. Sex divergence in capture rate over the three seasons. Data for seasonal capture rate (e.g.,<br />
mean number of capture per individual) are pooled from all years and error bars represent the<br />
standard error (S.E.). Statistical analyses compares males versus females during the same season<br />
(NS, non-significant; ***, p
Demographic patterns<br />
In the spring, from 1993 to 1999, high recapture rates enabled us to estimate male,<br />
female, and <strong>total</strong> population size. In all cases, goodness-of-fit tests indicated that a<br />
time variation and individual heterogeneity in capture probabilities model was the<br />
best fit for our data (Mth; Chao et al. 1992). Population size estimates are illustrated<br />
in Figure 3.<br />
Estimated population size<br />
500<br />
400<br />
300<br />
200<br />
100<br />
93 94 95 96 97 98 99<br />
Year<br />
Male<br />
Female<br />
Total<br />
Figure 3. Annual fluctuation in <strong>total</strong> number of adult snakes (open squares, dashed line), males (open<br />
circles, continuous line) and females (open triangles, dotted line) in spring. Population estimates (±<br />
S.E.) were performed using the program CAPTURE (see text for statistics).<br />
Over the seven years, the average estimated adult population size in spring<br />
was 365 ± 65 individuals (coefficient of variation 0.17) with a mean density of 11<br />
snakes/ha (<strong>total</strong> of 33 ha) and a biomass of 1.1 kg/ha (given a mean adult body mass<br />
253
of 100g). Females exhibited greater fluctuation in population size than did males<br />
(coefficient of variations 0.31 versus 0.08; Figure 3). Fluctuation in population size<br />
was greatest when reproducing females were considered alone (coefficient of<br />
variation 0.45). Using a proportion of 44% of the <strong>total</strong> female population (the long-<br />
term mean proportion in the population, Table 2), the proportion of reproducing<br />
females / non-reproducing females was significantly greater in 1996 and 1997, and<br />
lower in 1995 (χ² = 60.9, dl = 6, p < 0.0001; Table 2, Figure 4).<br />
Table 2. Annual variation in captures, proportion of reproducing females and the operational sex ratio.<br />
Calculations were made as follows: Initial Captures = number of different individuals captured each<br />
year; Total Captures = initial captures + recaptures; Proportion of reproducing females = estimate of<br />
reproductive females number/ estimate of <strong>total</strong> female population size; Sex Ratio = estimate of adult<br />
males population size /(estimate of adult males population size + estimate of adult females population<br />
size); Operational Sex Ratio = estimate of adult males population size / (estimate of adult males<br />
population size + estimate of reproductive females number).<br />
Year<br />
1993<br />
1994<br />
1995<br />
1996<br />
1997<br />
1998<br />
1999<br />
mean<br />
Initial<br />
Captures<br />
284<br />
385<br />
273<br />
297<br />
341<br />
192<br />
167<br />
277<br />
Total<br />
Captures<br />
640<br />
1079<br />
649<br />
732<br />
834<br />
321<br />
279<br />
565<br />
Proportion<br />
of reproducing<br />
females<br />
0.38<br />
0.41<br />
0.20<br />
0.64<br />
0.59<br />
0.33<br />
0.53<br />
0.44<br />
Sex Ratio<br />
0.40<br />
0.41<br />
0.53<br />
0.54<br />
0.40<br />
0.53<br />
0.57<br />
0.48<br />
Operational<br />
Sex Ratio<br />
0.64<br />
0.63<br />
0.85<br />
0.65<br />
0.53<br />
0.77<br />
0.80<br />
0.69<br />
Prey availability at the site varied annually, with 1996 having high rodent<br />
abundance, 1994 having low abundance, and all other years being intermediate<br />
(Bonnet et al. 2001b, Lourdais et al. 2002b). Food abundance in a given year was<br />
independent of <strong>food</strong> abundance in the preceding year (r=0.16, n=7, F(1,5)=0.14,<br />
254
p
Table 3. Combined influences of <strong>food</strong> levels in year n (Food n) and <strong>food</strong> levels in year n-1 (Food n-1)<br />
on the proportion of reproducing females (RF).<br />
Multiple Regression r = 0.97; r² = 0.88; n = 7; F(2,4)=14.51 p < 0.014<br />
Bêta±SE Partial correlation Semi-partial p value<br />
Food n 0.74±0.17 0.90 0.73 0.013<br />
Food n-1 0.71±0.17 0.89 0.70 0.015<br />
Annual changes in <strong>total</strong> female population size appeared closely related to<br />
current <strong>food</strong> levels (r = 0.92, F(1,4) = 23.57, n = 6, p < 0.008). However, changes in<br />
female population size were better explained in a multiple regression combining<br />
current <strong>food</strong> abundance with <strong>food</strong> abundance in preceding year (model 1, Table 4).<br />
While <strong>food</strong> level in year n positively influenced changes in female population size<br />
during this year, a negative influence was detected for <strong>food</strong> level in year n-1.<br />
A similar influence was detected when replacing <strong>food</strong> level in year n-1 by the<br />
proportion of reproducing females in year n (model 2, Table 4). Hence, the annual<br />
proportion of reproducing females (year n) negatively influenced the change in<br />
female population size over this year n. For males, variations in population size were<br />
limited and no <strong>relationship</strong> was found <strong>between</strong> changes in population size and <strong>food</strong><br />
level in year n, year n-1 or a combination of both (respective p values: 0.17, 0.29 and<br />
0.27).<br />
256
Table 4. Examination of annual changes in female population size. Population changes (during a year<br />
n) were calculated from spring population size in year n+1 minus spring population size in year n.<br />
Model 1 was obtained by combining <strong>food</strong> levels in year n (Food n) and <strong>food</strong> levels in year n-1 (Food<br />
n-1) in the multiple regression. Model 2 was obtained by replacing <strong>food</strong> levels in year n-1 by the<br />
proportion of reproducing females in year n (% RF n).<br />
Model 1 r = 0.99; r² = 0.98; n = 6; F(2,3)=112.92 p
Figure 5. Relationship <strong>between</strong> annual changes in female population size (calculated as spring<br />
population size in year n+1 minus spring population size in year n) and operational sex ratio in year<br />
n+1 (see text for statistics).<br />
Operationnal sex ratio in year n+1<br />
0.90<br />
0.85<br />
0.80<br />
0.75<br />
0.70<br />
0.65<br />
0.60<br />
0.55<br />
0.50<br />
-140 -115 -90 -65 -40 -15 10 35 60 85 110 135<br />
Change in female population size (year n)<br />
Discussion<br />
Recently, it has been experimentally shown that female aspic vipers need to reach a<br />
high and precise body condition threshold to initiate reproduction and sexual<br />
behaviours, whereas males exhibit a more gradual <strong>relationship</strong> <strong>between</strong> body<br />
reserves and reproduction (Aubret et al. 2002). It has also been shown that this<br />
dichotomy is underlay by an “all-or-nothing” versus a “gradual” hormonal regulation of<br />
reproduction in females and males, respectively (Aubret et al. 2002). <strong>The</strong> present<br />
field study shows that such sex differences in the energy budget and regulation of<br />
reproduction translate into marked sex divergences of demographic characteristics.<br />
Sex differences in annual catchability patterns reflecting differences in reproductive<br />
258
activities are somewhat a classical result (Saint Girons 1949, 1952, 1957b).<br />
However, the juxtaposition of our extensive data set (>4500 captures) with published<br />
eco-physiological data provides for the first time a sequence of functional links<br />
<strong>between</strong> sex-specific reproductive roles, energy investment to reproduction,<br />
catchability characteristics, and resulting demographics patterns. <strong>The</strong> possibility to<br />
reveal the interdependence <strong>between</strong> these various traits (usually considered<br />
separately), results from the major significance attributable to the marked shifts in<br />
catchability of ectotherms.<br />
In both sexes capture rate was influenced by a combination of environmental<br />
and biological factors. Body size affected catchability with smaller-sized adults being<br />
less catchable than large adults. Body size positively affects catchability in the asp<br />
viper (Naulleau & Bonnet 1996; this study) as well as other snakes (Bonnet et al.<br />
2002c). Perhaps, small snakes adopt a more secretive behaviour in response to a<br />
size-dependent vulnerability to predators (Lima & Dill 1990; Houston & Shine 1994).<br />
Additionally, smaller-sized snakes may be less catchable due to a higher body<br />
surface to volume ratio that shortens heating times and thus reduces basking.<br />
Alternately, the thermal requirements of small snakes may be lower than in larger<br />
individuals. Whatever the case, in snakes, the major effect of body size on<br />
catchability is associated with sexual maturity. After birth, snakes remain extremely<br />
secretive until they reach maturity (unpublished data on more than 600 marked<br />
neonates; Madsen et al. 1999). <strong>The</strong> dramatic increase in catchability with maturity<br />
provides strong support to the notion that temperate snakes provide the opportunity<br />
to connect reproduction to survival costs associated with vulnerability and<br />
consequently to demographic patterns (Bonnet et al. 1999b).<br />
Though snakes from all size and sex categories were caught throughout the<br />
active season, most of the captures occurred in spring when cool ambient<br />
259
temperatures make it necessary for animals to bask in the sun to achieve and<br />
maintain optimal temperatures. Additionally, mate searching and male-male combat<br />
lead to a strong increase in the catchability of males at that time. After the relatively<br />
short mating season (March-April), males adopt more secretive behaviours and were<br />
observed only occasionally, usually during shedding episodes or digestion<br />
(unpublished obs). In females also, reproduction strongly and positively influenced<br />
catchability with increased exposure of reproducing females over prolonged time<br />
periods (March to September). <strong>The</strong> higher catchability rates of reproducing females<br />
are linked to the high thermal/metabolic requirements of vitellogenesis and gestation<br />
(Bonnet et al. 1994; Bonnet & Naulleau 1996). Each year, most of the adult males<br />
undertake reproductive activities whilst only a fraction of the females are<br />
reproductive. As a result, male captures outnumbered female captures in spring, but<br />
not later in the year when males were no longer involved in reproduction whereas the<br />
female reproductive activities continued through the end of summer. Overall, when<br />
comparing males and females, both the difference in the absolute values and the<br />
temporal shift in catchability are explainable in the light of their respective system of<br />
allocation of energy to reproduction (gradual versus threshold dependent).<br />
In the course of the study, <strong>total</strong> population size fluctuated widely and year-to-year<br />
variations in rodent abundance (voles) appeared to be an important regulator. A sex-<br />
specific analysis revealed that inter-annual fluctuations in population size were mainly<br />
driven by the female population, notably reproductive females. <strong>The</strong>se variations are<br />
linked to the annual recruitment rate of reproductive females among non-reproductive<br />
adults and sub-adults, and to the annual survival cost of reproduction. As<br />
reproduction requires a female to have attained a high body condition threshold, and<br />
since <strong>food</strong> availability influences the accumulation of body stores, <strong>food</strong> availability<br />
thus influences the proportion of reproducing females in the following year (Bonnet et<br />
260
al. 2001b). However, the proportion of reproducing females was also elevated the<br />
year of particularly high <strong>food</strong> availability (1996, Figure 4), and this absence of a<br />
temporal delay <strong>between</strong> <strong>food</strong> availability and reproduction suggests a more direct<br />
influence of prey availability on reproductive status. We hypothesise that some<br />
females (e.g., ones close to the threshold) may respond positively to high <strong>food</strong> levels<br />
(1996) and reproduce under such favourable condition. This complex reproductive<br />
decision process that involves both long-term storage and facultative <strong>food</strong> <strong>intake</strong> is<br />
well illustrated in the multiple regression analysis (Table 3).<br />
In this population, survival costs of reproduction are high and most females<br />
die during or shortly after reproduction (Bonnet et al. 2000a, 2002a). Part of this<br />
mortality is attributable to exposure to avian predation (Naulleau 1997) while part is<br />
due to the extreme emaciation of post-parturient females (Bonnet et al. 2000a). A<br />
similar effect of reproduction on survival through predation and body reserves<br />
depletion has been reported in a similar adder, Vipera berus (Madsen & Shine 1993).<br />
A direct consequence of this high mortality of reproducing females is that a<br />
fluctuation in the proportion of reproducing females will generate substantial variation<br />
in absolute annual mortality. This was clearly confirmed in our data set as the<br />
proportion of reproducing females in a given year n negatively influenced the annual<br />
changes in <strong>total</strong> female population over this year n (Table 4). However, current <strong>food</strong><br />
levels also positively affected annual changes in female population size (Table 4). As<br />
most female vipers die after reproduction, positive changes in adult population size<br />
can be attributable to the recruitment of new adult females. <strong>The</strong>refore, this positive<br />
<strong>relationship</strong> <strong>between</strong> <strong>food</strong> availability and adult population size suggests a strong<br />
linkage <strong>between</strong> <strong>food</strong> levels and maturation, and this is largely confirmed by field<br />
data. Growth rate and reserve storage are directly dependent on fluctuation in <strong>food</strong><br />
levels. In 1996, the best year in terms of <strong>food</strong> availability, we recorded the highest<br />
261
growth rates, up to 20 cm in juveniles (Bonnet et al. 1999a) and several females<br />
reached maturity in 2.5 years (instead of the typical 3.5) and reproduced in 1997<br />
(Bonnet et al. 1999a). Hence, demographic patterns in female vipers were clearly<br />
understandable in the frame of their particular reproductive biology, combining<br />
delayed reproduction (precise body condition) and a short reproductive life (tendency<br />
toward semelparity).<br />
In strong contrast to females, annual changes in male population size were<br />
limited and apparently not affected by <strong>food</strong> levels. Though <strong>food</strong> levels influence<br />
growth in this sex as well (Bonnet et al. 1999a), they did not directly effect population<br />
dynamic. In males, the gradual system of energy allocation to reproduction does not<br />
impose extended phases of energy gathering, hence most males can initiate<br />
reproductive activities even with limited body reserves. Furthermore, reproductive<br />
activities are concentrated in spring and do not entail prolonged exposure to<br />
predation as observed in reproductive females and, as a result, most males<br />
reproduce repeatedly during their life span. This gradual system of energy allocation<br />
(no body condition threshold, iteroparity) offers a likely explanation for the stable<br />
male population size.<br />
<strong>The</strong> differential effects of <strong>food</strong> abundance on each sex within a population led to<br />
marked fluctuations in both absolute and operational sex ratio (Table 3). To our<br />
knowledge, such a strong effect has never been documented in snakes. Since OSR<br />
influences sexual selection gradients and the reproductive strategy of males (Duvall<br />
et al. 1992, Madsen & Shine 1992b), it deserves further study.<br />
In conclusion, while differential catchability is classically viewed as a<br />
confounding factor complicating population size analysis; in ectotherms it rather<br />
offers opportunities to explore complex <strong>relationship</strong>s <strong>between</strong> the energy budget of<br />
reproduction, seasonal catchability, and population dynamic. Our study of aspic<br />
262
vipers shows that within the array of capital breeders and within a given population,<br />
two sub-populations (males versus females) nonetheless exhibit marked differential<br />
sequences in the links <strong>between</strong> various life-history traits. Sexual differences in<br />
ecology (i.e., diet, behaviour, and habitat selection) or demography in relation to<br />
climatic or resource fluctuations have been reported in endotherm species, notably<br />
ungulates (Clutton-Brock et al. 1987; du Toit 1995; Myseterud 2000; Oakes et al.<br />
1992; Owen-Smith 1993). <strong>The</strong> clear dichotomous system revealed in the present<br />
work, however, contrasts sharply with results gathered on endotherms. <strong>The</strong> temporal<br />
dissociation <strong>between</strong> the phases of energy acquisition and allocation to reproduction<br />
exhibited by ectotherms provides a unique opportunity to better unravel the complex<br />
effects of resource-fluctuating environments on reproductive strategies (Pough 1980;<br />
Shine & Bonnet 2000a). Finally, the present study also emphasizes the<br />
complementary aspects of ecological and physiological approaches to interpret<br />
capture-recapture data (Bonnet et al. 2002a).<br />
Acknowledgements<br />
We thank Gwénael Beauplet, Hervé Fritz, and Emily Taylor for comments on the<br />
manuscript. Financial support was provided by the Conseil Régional de Poitou-<br />
Charentes, Conseil Général des Deux-Sèvres, Centre National de la Recherche<br />
Scientifique (France). Thanks to Melle for enthralling debates on multiple regression<br />
analysis. Finally, Jean De Riboulin solved many technical problems.<br />
263
V Discussion-conclusion :<br />
reproduction sur réserves,<br />
coûts de la reproduction et<br />
évolution vers la<br />
seméliparité<br />
264
Les données obtenues dans ce travail rendent possible l’identification des<br />
contraintes liées à la reproduction chez la vipère aspic. L’étude de la contribution<br />
respective de ces contraintes constitue une approche puissante pour la<br />
compréhension de la stratégie reproductrice particulière déployée par cette espèce et<br />
basée sur un investissement reproducteur massif et peu fréquent. Plusieurs de nos<br />
résultats peuvent être replacés dans un contexte évolutif général et permettent<br />
d’envisager un scénario explicatif à l’évolution des systèmes de gestion des<br />
ressources et des stratégies démographiques extrêmes.<br />
A. Stratégie de reproduction de la vipère aspic<br />
L’investissement reproducteur<br />
La formation des œufs (vitellogénèse) représente une étape clé de la reproduction<br />
chez la vipère aspic. La synthèse et le dépôt du jaune dans les follicules va mobiliser<br />
de grandes quantités d’énergie. Les importants stocks de réserves lipidiques en<br />
combinaison avec des prises alimentaires facultatives vont fournir le support de base<br />
de l’investissement reproducteur. Une fois cette première étape d’allocation achevée,<br />
la gestation va être associée à de nouveaux types d’investissements maternel. En<br />
effet, Pendant la phase de développement embryonnaire, le budget temps de la<br />
femelle aspic va être consacré à la thermorégulation et au maintien de conditions<br />
optimales pour le développement embryonnaire.<br />
Relation entre effort reproducteur, coûts de la reproduction et fécondité<br />
Si l’investissement gamétique est dépendant du nombre de follicules, nous avons<br />
montré que les dépenses métaboliques de la gestation présentent une nature fixe et<br />
265
indépendante de la fécondité. Ainsi, quelque soit le nombre de jeunes, la femelle doit<br />
maintenir un profil thermique optimum pour assurer le développement embryonnaire.<br />
En outre, les activités de thermorégulation et la réduction des déplacements vont<br />
limiter les possibilités d’alimentation. Ces pertes en opportunités alimentaires sont<br />
par leur nature indépendantes du nombre de jeunes produits. La succession des<br />
phases d’investissement énergétique dans la vitellogénèse (investissement direct)<br />
d’une part et dans la gestation (effort métabolique indirect lié à des modifications du<br />
profil d’activité) ensuite vont avoir de profondes conséquences sur l’état<br />
physiologique des femelles après la mise bas (amaigrissement). Les faibles valeurs<br />
de condition corporelle observées chez les femelles post parturientes illustrent ces<br />
phases successives d’investissement et limitent les possibilités de récupération<br />
après la reproduction. Les contraintes énergétiques de la gestation sont donc<br />
composites en impliquant un volet métabolique (changement des préférences<br />
thermiques) accentué par les conséquences écologiques de la thermorégulation<br />
(pertes en opportunités alimentaires). Le niveau d’amaigrissement post partum des<br />
femelles et les coûts en survie seront donc déterminés par ces contraintes et peu ou<br />
pas affectés par le nombre de jeunes en développement. Du fait de leur nature fixe,<br />
les coûts énergétiques rapportés au nombre de vipéreaux produits seront d’autant<br />
plus élevés que la portée sera de taille réduite. En plus de ces aspects<br />
énergétiques, les changements éthologiques pendant la vitellogénèse et la gestation<br />
vont entraîner une augmentation brutale de l’exposition et par conséquent du risque<br />
de prédation. L’engagement dans la reproduction va donc être à l’origine de coûts<br />
écologiques directs (prédation) déterminés par le statut reproducteur et peu affectés<br />
par le nombre de jeunes.<br />
L’examen de ces différentes composantes de l’effort reproducteur et des coûts<br />
associés suggère donc l’existence d’une relation non linéaire entre ces deux<br />
266
variables : la reproduction va être associée à des coûts en survie directs (prédation)<br />
et indirects (amaigrissement et coûts de post reproduction) peu dépendants du<br />
nombre de jeunes produits. Nos données montrent, par ailleurs, que l’amplitude très<br />
élevée de ces coûts vient limiter directement le nombre d’opportunités<br />
reproductrices.<br />
Gestion de la ressource et fréquence de reproduction<br />
Face à de tels coûts fixes, Bull et Shine (1979) ont suggéré l’avantage de systèmes<br />
reproducteurs basés sur la faible fréquence des reproductions. En se reproduisant<br />
de façon alternée, l’organisme peut ainsi tirer un avantage en préparant son<br />
investissement pour la reproduction suivante. Nos données permettent de confirmer<br />
l’existence de telles situations et de connecter la proposition de Bull et Shine avec la<br />
sélection de systèmes particuliers de gestion de la ressource. Il existe, chez la<br />
vipère aspic, un ensemble de traits d’histoire de vie co-adaptés qui permettent de<br />
préparer très efficacement la reproduction. Ainsi, la maturité tardive et l’accumulation<br />
de réserves lipidiques jusqu’à un seuil élevé de condition corporelle sont des<br />
éléments qui garantissent un investissement reproducteur massif et donc une<br />
fécondité élevée, même en l’absence de source énergétique (alimentation) pendant<br />
la reproduction. Face à des coûts de reproduction élevés (directs et post<br />
reproduction), un tel système va assurer un succès reproducteur élevé et donc<br />
l’amortissement des coûts payés par vipéreaux. Cette orientation particulière de<br />
l’effort reproducteur constitue une réponse évolutive qui va influencer l’ensemble de<br />
la stratégie reproductrice (voir schéma récapitulatif page suivante).<br />
267
80% 40% 24% 12%<br />
Préparation<br />
Système de<br />
capitalisation<br />
de l’énergie<br />
+ + +<br />
Reproduction 1<br />
Investissement<br />
reproducteur<br />
élevé<br />
Contraintes environnementales :<br />
Prédation, climat, nourriture<br />
Récupération<br />
Reproduction 2<br />
Faibles possibilités de<br />
récupération : nombre<br />
de reproduction réduit<br />
Schéma récapitulatif : chez la vipère aspic femelle les coûts de reproduction et de<br />
post-reproduction sont élevés et affectés par plusieurs variables environnementales<br />
(en haut). Nos données indiquent que la probabilité de survivre va s’effondrer<br />
(pourcentages) dès la première reproduction. Ceci illustre l’importance des<br />
contraintes écologiques/énergétiques de la reproduction qui se manifestent<br />
indépendamment de la fécondité (prédation, pertes en opportunités alimentaires,<br />
dépense métabolique pendant la gestation). Dans ce contexte où les possibilités de<br />
reproduction ultérieure sont précaires, la capitalisation de l’énergie est très<br />
avantageuse. En effet, cette stratégie d’allocation permet un investissement<br />
reproducteur élevé et l’amortissement des coûts en optimisant le succès à chaque<br />
reproduction.<br />
268
Pour les vipères qui survivent jusqu’à la mise bas, la phase de récupération<br />
sera déterminée par les fluctuations d’abondance en proies qui vont influencer les<br />
possibilités de reconstitution des stocks de réserves. L’interaction entre les<br />
contraintes énergétiques de la reproduction et les limitations trophiques va rendre<br />
peu avantageuse la réalisation de compromis entre l’investissement dans la<br />
reproduction courante et dans les reproductions futures. Dans ce contexte, les<br />
pressions de sélection vont favoriser la maximisation de l’investissement dans<br />
chaque reproduction. La stratégie d’allocation de l’énergie de la vipère aspic reflète<br />
fidèlement cette orientation des compromis d’allocation. Le stockage de réserves<br />
corporelles avant la reproduction et l’existence d’un seuil de condition corporelle<br />
élevé et fixe, garantissent le succès reproducteur même lorsque les conditions<br />
trophiques sont limitantes pendant la reproduction. Cette stratégie d’allocation de<br />
l’énergie constitue potentiellement une réponse évolutive adaptée à de fortes<br />
contraintes reproductrices indépendantes de la fécondité (voir schéma récapitulatif).<br />
Nos résultats soutiennent fortement la proposition initiale de Bull et Shine (1979) et<br />
soulignent l’importance proximale des systèmes de gestion de la ressource<br />
(capitalisation) dans la concentration de l’effort reproducteur.<br />
Origine de la seméliparité<br />
Si le système d’allocation de l’énergie est, chez la vipère aspic, fortement orienté<br />
vers la capitalisation de la ressource, il est important de noter que cette espèce n’est<br />
pas un reproducteur sur réserves “rigide”. Ainsi lorsque la nourriture est abondante,<br />
les prises alimentaires facultatives seront observées pendant la reproduction. Cette<br />
entrée d’énergie va permettre l’augmentation de l’investissement reproducteur et la<br />
production de vipéreaux plus lourds (alimentation pendant la vitellogénèse). La<br />
femelle va aussi bénéficier d’une partie de cet apport qui va amortir des coûts<br />
269
énergétiques en améliorant sa condition corporelle post partum. Nos données<br />
suggèrent que dans cette population, l’apport énergétique facultatif est généralement<br />
réduit et fluctuant. Il en est de même, après la reproduction, où l’abondance en proie<br />
va limiter les possibilités de récupération.<br />
Dans notre population d’étude, l’impact de ces facteurs prend une dimension<br />
très particulière. Les mises bas tardives (contraintes climatiques) vont limiter, voir<br />
annuler, les possibilités de récupération avant l’hiver. Une fois l’hivernage achevé,<br />
les limitations trophiques dans l’année qui suit la reproduction vont également limiter<br />
les possibilités de récupération. En conséquence, les femelles ne vont se reproduire<br />
qu’une seule fois et il en résultera une forte tendance semélipare.<br />
Une approche comparative avec le cycle des mâles dans cette région où<br />
avec des populations plus méridionales apporte des informations intéressantes sur<br />
l’ampleur des contraintes reproductrices maternelles. Notamment, les mâles sont<br />
beaucoup moins contraints dans leur reproduction que les femelles et<br />
l’investissement, chez ce sexe, est beaucoup plus graduel. Cette situation est<br />
clairement illustrée par une dynamique populationnelle profondément divergente<br />
entre les sexes avec de très fortes fluctuations dans la population femelle qui<br />
contrastent avec la stabilité populationnelle des mâles. Ces fluctuations qui affectent<br />
la population dans son ensemble sont directement liées aux coûts de la reproduction<br />
spécifiques des femelles et aux influences environnementales sur l’amplitude de<br />
ces coûts.<br />
La comparaison avec des populations dans des régions climatiques plus<br />
favorables apporte des éléments de reflexion intéressant sur les facteurs influençant<br />
l’expression des coûts de la reproduction et leurs conséquences démographiques.<br />
Ainsi, le suivi de populations Italiennes (Zuffi et al. 1999 ; données non publiées)<br />
indique des reproductions annuelles et une répétition des reproductions (souvent<br />
270
supérieure à 4-5) au cours de la vie des femelles. Dans ces populations<br />
méridionales, les conditions climatiques sont favorables et n’imposent pas de<br />
longues phases d’exposition. Des conditions plus favorables vont permettre un cycle<br />
plus court, des expositions beaucoup plus limitées et des possibilités de<br />
récupération élevées l’année même de la reproduction. Ces informations sont<br />
importantes et soulignent l’influence majeure des variables environnementales<br />
(climat) qui vont moduler l’expression des coûts énergétiques et écologiques.<br />
B. Proposition d’un scénario évolutif de la<br />
transition vers la seméliparité<br />
Les populations de vipères aspic de l’Ouest de la France sont donc très<br />
intéressantes en occupant une position charnière dans le continuum entre itéroparité<br />
et seméliparité. A l’image des populations méridionales, le mode de reproduction<br />
dans ces zones périphériques est l’itéroparité. Pourtant, les contraintes<br />
environnementales sont telles que les femelles suivent des trajectoires<br />
reproductrices semélipares. Une telle situation offre une opportunité de retracer les<br />
étapes possibles de la transition vers la seméliparité.<br />
Comme chez beaucoup d’autres ectothermes (Bull and Shine 1979), la<br />
reproduction chez cette espèce est à l’origine de contraintes brutales générant des<br />
coûts élevés, indépendants de la fécondité. Le système d’allocation de l’énergie,<br />
basé sur une longue préparation de la reproduction, permet d’amortir l’amplitude des<br />
coûts payés par jeune produit. Dans les populations en limite de l’aire de répartition,<br />
les contraintes environnementales (climatiques et trophiques) renforcent l’amplitude<br />
de ces impacts sur la vie reproductrice ultérieure (la plupart des femelles meurent<br />
après la reproduction). Cependant, la situation n’est pas fixée et lorsque les<br />
271
conditions sont favorables, la récupération énergétique et l’investissement dans<br />
d’autres reproductions est possible. Cette situation peut être considérée comme le<br />
contexte initial de la transition vers un mode de reproduction semélipare. Si on<br />
considère une légère accentuation des contraintes environnementales (dégradation<br />
des conditions climatiques, limites trophiques), les chances de reproductions<br />
ultérieures vont se réduire et s’annuler complètement si l’amplitude des coûts fixes<br />
dépasse les possibilités de récupération de l’organisme. Le maintien ou non de ce<br />
type de conditions, à une échelle macro-évolutive, pourra alors avoir des<br />
conséquences majeures sur l’orientation de la stratégie reproductrice. Notamment, si<br />
les contraintes environnementales rendent extrêmement improbables de futures<br />
reproductions, les pressions de sélection vont favoriser la maximisation de<br />
l’investissement dans la reproduction, en capitalisant par exemple, de l’énergie sur<br />
de longues périodes. Une telle situation correspond en fait à une relaxation du<br />
compromis classique entre reproduction courante et future (Williams 1966b)<br />
favorisant l’orientation de la physiologie de l’organisme dans la réalisation d’un effort<br />
reproducteur ”explosif”. Une telle sélection directionnelle sur les processus<br />
d’allocation de l’énergie rend encore moins probable les chances de survie ultérieure<br />
de l’organisme. Le résultat final possible va être la fixation du système (voir scénario<br />
page suivante) avec une mort qui devient inévitable suite à l’épuisement de<br />
l’organisme ou, à l’extrême, de façon génétiquement programmée (en dehors de<br />
causes environnementales) comme chez de nombreux céphalopodes (Boyle 1983,<br />
1987) ou certains saumons (Crespi & Teo 2002).<br />
272
Scénario de transition vers la seméliparité<br />
1) La reproduction draine parfois une telle quantité d’énergie que cela entraîne la<br />
mort de organisme<br />
2) Si un tel phénomène a lieu très régulièrement, il n’existe plus de bénéfices à<br />
réduire l’effort reproducteur courant en dessous du maximum<br />
physiologiquement envisageable car il n’existe pas d’autres possibilités de<br />
reproduction.<br />
3) Ce contexte favorise l’accumulation de traits qui vont augmenter le succès<br />
reproducteur et en même temps réduire de plus en plus les chances de survie<br />
ultérieures.<br />
4) Le résultat évolutif final est l’accumulation de mutation et la fixation génétique<br />
du système avec une mort inévitable et systématique après la reproduction.<br />
Dans un tel scénario, les populations de vipère à tendance semélipare se<br />
situent à l’échelon 1. Les systèmes à mort programmée rencontrés chez les<br />
céphalopodes et les saumons se trouvent au stade 4. Un tel schéma offre une piste<br />
cohérente pour la transition évolutive vers les systèmes semélipares. Alors que<br />
seméliparité et itéroparité sont généralement étudiées et modélisées comme deux<br />
stratégies en compétition, notre schéma propose l’existence d’une seule stratégie de<br />
base : l’itéroparité. Dans certaines situations environnementales, la sélection<br />
favoriserait la concentration de l’investissement sur la première reproduction. La<br />
seméliparité constitue alors un état dérivé de l’itéroparité et résulte d’une orientation<br />
273
de l’organisme vers la réalisation d’un effort reproducteur explosif et unique et non<br />
d’une programmation rigide de la mort.<br />
Notre étude apporte des éléments de réflexion sur l’évolution des stratégies<br />
reproductrices. Dans le monde vivant, il existe en fait de nombreuses activités qui<br />
peuvent générer des coûts indépendants de la fécondité et ces composantes<br />
méritent donc une attention particulière pour la compréhension des systèmes<br />
d’allocation de l’énergie. Ce travail s’est principalement concentré sur des aspects<br />
énergétiques et quantitatifs de l’investissement reproducteur. Pourtant, si le nombre<br />
de jeunes produits est une variable clé, il est important de souligner que la qualité de<br />
la progéniture est un paramètre majeur qu’il est aussi nécessaire d’intégrer. En effet,<br />
la thermorégulation accentuée des femelles pendant la gestation explique en partie<br />
les coûts énergétiques élevés. L’origine de ces choix thermiques est directement liée<br />
à la nécessité d’assurer un développement embryonnaire optimal. Dans un travail<br />
récent (voir annexe), j’ai pu mettre en évidence de profondes influences des<br />
conditions thermiques de la gestation sur le phénotype des jeunes (longueur,<br />
écaillure) ainsi que sur la mortalité embryonnaire. Si l’allocation de l’énergie et<br />
l’optimisation de l’effort reproducteur sont des éléments importants des stratégies<br />
reproductrices, la qualité de la descendance va aussi exercer une influence majeure<br />
sur le succès reproducteur. La diversité des soins parentaux prodigués par les<br />
vertébrés ectothermes (notamment les soins pré-nataux pendant la gestation) offre<br />
un champ d’étude très fertile pour examiner l’importance de l’optimisation de la<br />
qualité des jeunes dans la compréhension des stratégies reproductrices. Je vais<br />
désormais orienter mes travaux de recherche dans ce domaine particulier.<br />
274
Bibliographie<br />
A.<br />
Anderson RA & Karasov WH (1981) Contrasts in energy <strong>intake</strong> and expenditure<br />
in sit-and-wait and widely foraging lizards. Oecologia 49 : 67-72.<br />
Andren C & Nilson G (1983) Reproductive tactics in an island population of<br />
adders, Vipera berus (L.), with a fluctuating <strong>food</strong> resource. Amphibia-Reptilia 3 :<br />
63-79.<br />
Andrews RM (1982) Patterns of growth in reptiles. In : Gans C & Pough H (eds)<br />
Biology of Reptilia 13, pp 273-312. Academic Press, London, United Kingdom,.<br />
Andrews RM & Pough FH (1985) Metabolism of squamate reptiles: allometric<br />
and ecological <strong>relationship</strong>s. Physiological Zoology 58 : 214-231.<br />
Andrews RM, Mathies T & Warner D (2000) Effect of incubation temperature on<br />
morphology, growth and survival of juvenile Sceloporus undulatus. Herpetological<br />
Monographs 14 : 420-431.<br />
Andries JC (2001) Endocrine and environmental control of reproduction in<br />
polychaeta. Canadian Journal of Zoology 79 : 254-270.<br />
Anguiletta MJ & Sears MW (2000) <strong>The</strong> metabolic cost of reproduction in an<br />
oviparous lizard. Functional Ecology 14 : 39-45.<br />
Arcese CD & Smith JNM (1988) Effects of population density and supplemental<br />
<strong>food</strong> on reproduction in song sparrow. Journal of Animal Ecology 57 : 119-136.<br />
Atchley WR, Gaskins CT & Anderson D (1976) Statistical properties of ratios I.<br />
Empirical results. Systematic Zoology 25 : 137-148.<br />
275
Aubret F, Bonnet X, Shine R & Lourdais O (2002) Fat is sexy for females but not<br />
males : the influence of body reserves on reproduction in snakes (Vipera aspis).<br />
Hormones and Behavior 42 : 135-147.<br />
B.<br />
Ballinger RE (1977) Reproductive strategies : <strong>food</strong> availability as a source of<br />
proximal variation in a lizard. Ecology 58 : 628-635.<br />
Ballinger RE (1983) Lizard Ecology : Studies on a model organism. In : Huey R,<br />
Pianka ER & Schoener TW (eds) Life history variations, pp. 241-260. Harvard<br />
University press, Cambridge, Massachusetts, USA.<br />
Baron JP, Ferrière R, Clobert J & Saint Girons H (1996) Stratégie<br />
démographique de Vipera ursinii ursinii au Mont-Ventoux (France). Comptes<br />
Rendus de l’Académie des Sciences, Série III 319 : 57-69.<br />
Bartholomew GA (1970) A model for the evolution of pinniped polygyny.<br />
Evolution 24 : 546-559.<br />
Bauwens D & Thoen C (1981) Escape tactics and vulnerability to predation<br />
associated with reproduction in the lizard Lacerta vivipara. Journal of Animal<br />
Ecology 50 : 733-743.<br />
Beaupré SJ & Duvall D (1998) Variation in oxygen consumption of the western<br />
diamondback rattlesnake (Crotalus atrox) : implications for sexual size<br />
dimorphism. Journal of Comparative Physiology 168(b) : 497-506.<br />
Beaupré SJ & Zaidan F (2001) Scaling CO2 production in the timber rattlesnake<br />
(Crotalus horridus), with comments on cost of growth in neonates and<br />
comparative patterns. Physiological and Biochemical Physiology 74 : 757-768.<br />
Bell G (1980) <strong>The</strong> costs of reproduction and their consequences. American<br />
Naturalist 116 : 45-76.<br />
Bell G & Koufopanou V (1986) <strong>The</strong> cost of reproduction. In : Dawkins R & Ridley<br />
M (eds) Oxford Surveys in Evolutionary Biology, pp. 83-131. Oxford University<br />
Press, Oxford.<br />
276
Bennett AF & Dawson WR (1976) Metabolism. In : Gans C & Dawson WR (eds)<br />
Biology of the Reptilia. Vol. 5 Physiology A, pp. 127-223. New York: Academic<br />
Press.<br />
Birchard GF, Black CP, Schuett GW & Black V (1984) Influence of pregnancy on<br />
oxygen consumption, heart rate and haematology in the Garter Snake :<br />
implications for the cost of reproduction in live bearing reptiles. Comparative<br />
Biochemistry and Physiology 77(a) : 519-523.<br />
Blackburn DG (1982) Evolutionary origins of viviparity in the Reptilia I. Sauria.<br />
Amphibia-Reptilia 3 : 185-205.<br />
Blackburn DG (1985) Evolutionary origins of viviparity in the Reptilia II.<br />
Serpentes, Amphisbaenia, and Icthyosauria. Amphibia-Reptilia 6 : 259-291.<br />
Blackburn DG (1998) Resorption of oviductal eggs and embryos in squamates.<br />
Herpetological Journal 8 : 65-71.<br />
Blanchard FN & Blanchard FC (1941) Factors determining time of birth in the<br />
garter snake Thamnophis sirtalis sirtalis (Linnaeus). Pap, Mich, Acad, Sci, Arts,<br />
Lett 26 : 161-176.<br />
Blem CR & Blem LB (1990) Lipid reserves of the brown water snake Nerodia<br />
taxispilota. Comparative Biochemistry and Physiology 97A : 367-372.<br />
Boggs CL (1992) Resource allocation : exploring connections <strong>between</strong> foraging<br />
and life history. Functional Ecology 6 : 508-518.<br />
Bolton M, Monaghan P & Houston CD (1993) Proximate determination of clutch<br />
size in lesser black-backed gulls : the roles of <strong>food</strong> supply and body condition.<br />
Canadian Journal of Zoology 71 : 273-279.<br />
Bonnet X (1996) Gestion des réserves corporelles et stratégie de reproduction<br />
chez Vipera aspis. Thèse de Doctorat, Université Claude Bernard, Lyon I.<br />
Bonnet X & Naulleau G (1993) Relationship <strong>between</strong> glycemia and seasonal<br />
activity in Vipera aspis. Amphibia Reptilia 14 : 295-306.<br />
Bonnet X & Naulleau G (1994) Utilisation d'un indice de condition corporelle<br />
(BCI) pour l'étude de la reproduction chez les serpents. Comptes Rendus de<br />
l’Académie des Sciences 317 : 34-41.<br />
277
Bonnet X & Naulleau G (1995) Estimation of body reserves in living snakes using<br />
a Body Condition Index (BCI). In : Llorente GA, Montori A, Santos X & Carretero<br />
MA (eds) Scientia Herpetologica, pp 237-240. Barcelona, Spain.<br />
Bonnet X & Naulleau G (1996) Catchability in snakes : consequences on<br />
breeding frequency estimates. Canadian Journal of Zoology 74 : 233-239.<br />
Bonnet X, Naulleau G & Mauget R (1994) <strong>The</strong> influence of body condition on 17-<br />
β Estradiol levels in relation to vitellogenesis in female Vipera aspis (Reptilia<br />
viperidae). General & Comparative Endocrinology 93 : 424-437.<br />
Bonnet X, Bradshaw SD, & Shine R (1998) Capital versus income breeding: an<br />
ectothemic perspective. Oikos 83 : 333-341.<br />
Bonnet X, Naulleau G & Shine R (1999a) <strong>The</strong> dangers of leaving home :<br />
dispersal and mortality in snakes. Biological Conservation 89 : 39-50.<br />
Bonnet X, Naulleau G, Lourdais O & Vacher-Vallas M (1999b) Growth in the asp<br />
viper (Vipera aspis L.) : insights from long term field study. In : Miaud C &<br />
Guyetant R (eds) Current Studies in Herpetology, pp 63-69. SEH, Le Bourget du<br />
Lac.<br />
Bonnet X, Naulleau G, Shine R & Lourdais O (2000a) What is the appropriate<br />
time scale for measuring costs of reproduction in a capital breeder? Evolutionary<br />
Ecology 13 : 485-497.<br />
Bonnet X, Naulleau G, Shine R & Lourdais O (2000b) Reproductive versus<br />
ecological advantages to larger body size in female Vipera aspis. Oikos 89 : 509-<br />
518<br />
Bonnet X, Naulleau G, Bradshaw D & Shine R (2001a) Changes in plasma<br />
progesterone in relation to vitellogenesis and gestation in the viviparous snake<br />
Vipera aspis. General and Comparative Endocrinology 121 : 84-94.<br />
Bonnet X, Naulleau G, Shine R & Lourdais O (2001b) Short-term versus long-<br />
term effects of <strong>food</strong> <strong>intake</strong> on reproductive output in a viviparous snake (Vipera<br />
aspis). Oikos 92 : 297-308.<br />
Bonnet X, Lourdais O, Shine R & Naulleau G (2002a) Reproduction in snakes<br />
(Vipera aspis): costs, currencies and complications. Ecology 83 : 2124-2135.<br />
278
Bonnet X, Naulleau G & Lourdais O (2002b) <strong>The</strong> benefits of complementary<br />
techniques : using capture-recapture and physiological approaches to understand<br />
costs of reproduction in the asp viper. In : Schuett GW, Hoggren M & Greene HW<br />
(eds) Biology of the Vipers. Biological Science Press, Traverse City, MI (USA) : In<br />
press.<br />
Bonnet X, Pearson D, Ladyman M, Lourdais O & Bradshaw D (2002c) Heaven<br />
for serpents? A mark-recapture study of Tiger Snakes (Notechis scutatus) on<br />
Carnac Island, Western Australia. Australian Ecology : In press.<br />
Boyd IL (1984) <strong>The</strong> <strong>relationship</strong> <strong>between</strong> body condition and the timing of<br />
implantation in pregnant Grey seals (Halichoerus grypus). Journal of Zoology 203<br />
: 113-123.<br />
Boyle PR (1983) Cephalopods life cycles vol I. In : Boyle PR (ed) Species<br />
accounts. Academic press. London<br />
Boyle PR (1987) Cephalopods life cycles vol II. In : Boyle PR (ed) Comparative<br />
reviews. Academic press. London<br />
Bradshaw SD (1997) Homeostasis in Desert Reptiles. Cloudsey-Thompson JL<br />
(ed), 213 pp. Springer, Heidelberg.<br />
Bradshaw SD, Saint Girons H & Bradshaw FJ (1991) Seasonal changes in<br />
material and energy balance associated with reproduction in the green lizard,<br />
Lacerta viridis, in western France. Amphibia Reptilia 12 : 21-32.<br />
Brana F, Gonzalez F & Barahona A (1992) Relationship <strong>between</strong> ovarian and fat<br />
body weights during vitellogenesis for three species of lacertid lizards. Journal of<br />
Herpetology 26 : 515-518.<br />
Brand CJ & Keith LB (1979) Lynx demography during a snowshoe hare decline<br />
in Alberta. Journal of Wildlife Management 43 : 827-839.<br />
Brodie EDI (1989) Behavioral modification as a means of reducing the cost of<br />
reproduction. American Naturalist 134 : 225-238.<br />
Brodie EDI & Ducey PK (1989) Allocation of reproductive investment in the<br />
redbelly snake Storeria occipitomaculata. American Midland Naturalist 122 : 51-<br />
58.<br />
279
Bronson FH (1998) Energy balance and ovulation: small cages versus natural<br />
habitats. Reproduction, Fertility and Development 10 : 127-137.<br />
Brown WS (1991) Female reproductive ecology in a northern population of timber<br />
rattlesnake (Crotalus horridus). Herpetologica 47 : 101-115.<br />
Brown WS (1993) Biology, status and management of the timber rattlesnake<br />
(Crotalus horridus) : a guide for conservation. Herpetological Circulars 22 : 1-78.<br />
Brown DC, Kelnar CJH & Wu FCW (1996) Energy, metabolism during male<br />
human puberty II. Use of testicular size in predictive equations for basal metabolic<br />
rate. Annals of Human Biology 23(4) : 281-284.<br />
Bryant EH (1971) Life history consequences of natural selection : Cole’s result.<br />
American naturalist 105 : 75-76.<br />
Bull JJ & Shine R (1979) Iteroparous animals that skip opportunities for<br />
reproduction. American Naturalist 114 : 296-316.<br />
Burger J (1998) Effects of incubation temperature on hatchling pine snakes :<br />
Implications for survival. Behavioural Ecology and Sociobiology 43 : 11-18.<br />
Butte NF, Hopkinson JM, Mehta N, Moon JK & Smith E (1999) Adjustments in<br />
energy expenditure and substrate utilisation during late pregnancy and lactation.<br />
American Journal of Clinical Nutrition 69(2) : 299-307.<br />
C.<br />
Cadenas S, Buckingham JA, St-Pierrie J, Dickinson K, Jones RB & Brand MD<br />
(2000) AMP decreases the efficiency of skeletal-muscle mitochondria.<br />
Biochemical Journal 351(2) : 307-311.<br />
Callard IP, Filete LA, Perez FE, Sorbera LA, Giannoukos G, Klosterman LL,<br />
Tsang P & McCracken JA (1992) Role of the corpus luteum and progesterone in<br />
the evolution of vertebrate viviparity. American Zoologist 32 : 264-275.<br />
Calow P. (1979) <strong>The</strong> cost of reproduction - a physiological approach. Biological<br />
Reviews 54 : 23-40.<br />
Capula M, Luiselli L & Anibaldi C (1992) Complementary study on the<br />
reproductive biology in female adder, Vipera berus (L.), from eastern Italian Alps.<br />
Vie Milieu 42 : 327-336.<br />
280
Chao A, Lee SM & Jeng SL (1992) Estimating population size for captures-<br />
recaptures data when captures probabilities vary by time and individual animal.<br />
Biometrics 48 : 201-216.<br />
Charnov EL (1982) <strong>The</strong> <strong>The</strong>ory of Sex Allocation. Princeton University Press,<br />
Princeton.<br />
Charnov EL & Schaffer WM (1973) Life history consequences of natural selection<br />
: Cole’s results revisited. American Naturalist 107 : 791-793.<br />
Chastel O, Weimerskirch H & Jouventin P (1995) <strong>The</strong> influence of body condition<br />
on reproductive decisions and reproductive success in the blue petrel. Ecology<br />
76 : 2240-2246.<br />
Cherel Y (1995) Nutrient reserve storage, energetics, and <strong>food</strong> consumption<br />
during the prebreeding and premoulting foraging periods of king penguins. Polar<br />
Biology 15 : 209-214.<br />
Cherel Y, Charassin JB & Handrich Y (1993) Comparison of body reserves build-<br />
up in prefasting chicks and adults of king penguins (Aptenodytes patagonicus).<br />
Physiological Zoology 66 : 750-770.<br />
Cichon M, Olejniczak P & Gustafsson L (1998) <strong>The</strong> effect of body condition on<br />
the cost of reproduction in female collared flycatchers Ficedula albicollis. Ibis 140<br />
: 128-130.<br />
Clark H & Sisken BF (1956) Nitrogenous excretion by embryos of the viviparous<br />
snake Thamnophis s. sirtalis. Journal of Experimental Biology 33 : 384-393.<br />
Clutton-Brock TH (1988) Reproductive Success. Studies of individual variation in<br />
contrasting breeding systems. University of Chicago Press, Chicago.<br />
Clutton-Brock TH (1991) <strong>The</strong> evolution of parental care. Princeton University<br />
Press, Princeton, New Jersey.<br />
Clutton-Brock TH (1998) Intoduction : studying reproductive cost. Oikos 83 : 421-<br />
423<br />
Clutton-Brock TH, Iason GR & Guiness FE (1987) Sexual segregation and<br />
density-related changes in habitat use in male and female red deer (Cervus<br />
elaphus). Journal of Zoology 211 : 275-289.<br />
281
Coates M & Ruta M (2000) Nice snakes, shame about the legs. Trends in<br />
Ecology and Evolution 15 : 503-507.<br />
Cole LC (1954) <strong>The</strong> population consequences of life history phenomena.<br />
Quarterly review of Biology 29 : 103-137.<br />
Congdon JD, Dunham AE & Tinkle DW (1982) Energy budgets and life histories<br />
of reptiles. In : Gans C & Pough FH (eds) Biology of the Reptilia, pp 233-271.<br />
Academic Press, New York.<br />
Cooper WE Jr, Vitt LJ, Hedges R & Huey RB (1990) Locomotor impairment and<br />
defense in gravid lizard (Eumeces laticeps): behavioral shift in activity may offset<br />
costs of reproduction in an active forager. Behavioural Ecology and Sociobiology<br />
27 : 153-157.<br />
Corkum LD, Ciborowski JJH & Poulin RG (1997) Effects of emergence date and<br />
maternal size on egg development and sizes of eggs and first-instar nymphs of a<br />
semelparous aquatic insect. Oecologia 111 : 69-75.<br />
Cossins AR & Roberts N (1996) <strong>The</strong> feast and famine. Nature 379 : 23.<br />
Cresswell WJ, Harris S, Cheeseman CL & Mallinson PJ (1992) To breed or not to<br />
breed: an analysis of the social and density-dependent constraints on the<br />
fecundity of female badgers (Meles meles). Philosophical Transactions of the<br />
Royal Society London B. 338 : 393-407.<br />
Crews D & Gans C (1992) <strong>The</strong> interaction of hormones, brain, and behavior : an<br />
emerging discipline in herpetology. In : Gans C & Crews D (eds) Biology of the<br />
Reptilia 18, pp 1-23. <strong>The</strong> university of Chicago press.<br />
Crespi JB & Teo R (2002) Comparative Phylogenetic Analysis of the evolution of<br />
semelparity and life history in Salmonid fishes. Ecology 56 : 1008-1020.<br />
Cuellar O (1984) Reproduction in a parthenogenetic lizard: with a discussion of<br />
optimal clutch size and a critique of the clutch weight/body weight ratio. American<br />
Midland Naturalist 111 : 242-258.<br />
Curtis V, Henry CJK, Birch E & Ghusain-Choueiri A (1996) Intra-individual<br />
variation in the basal metabolic rate of women : Effect of the menstrual cycle.<br />
American Journal of Human Biology 8(5) : 631-639.<br />
282
D.<br />
Darwin C (1859) On the origin of species by means of natural selection or the<br />
preservation favoured races in struggle for life. John Murray (ed). London<br />
Daan S, Deerenberg C & Dijkstra C (1996) Increased daily work precipitates<br />
natural death in the kestrel. Journal of Animal Ecology 65 : 539-544.<br />
Darken RS, Martin KLM & Fisher MC (1998) Metabolism during delayed hatching<br />
in terrestrial eggs of a marine fish, the Grunion, Leuresthes tenuis. Physiological<br />
Zoology 71(4) : 400-406.<br />
Dearbon DC (2001) Body condition and retaliation in the parental effort decisions<br />
of incubating great frigate birds (Fregata minor). Behavioral Ecology 12 : 200-206.<br />
Decopas F & Hart SS (1957) Use of a Pauling oxygen analyser for measurement<br />
of oxygen consumption of animals in open-circuit systems and in a short-lag,<br />
closed-circuit apparatus. Journal of Applied Physiology 10 : 388-392.<br />
Delattre P, Giraudoux P, Baudry J, Truchetet D, Musard P, Toussaint M, Stahl P,<br />
Poulle ML, Artois M, Damange JP, & Quere JP (1992) Land use patterns and<br />
types of Common Vole (Microtus arvalis) population kinetics. Agriculture,<br />
Ecosystem and Environment 3 : 153-169.<br />
Demarco V & Guillette LJ jr (1992) Physiological cost of pregnancy in a<br />
viviparous lizard Sceloporus jarrovi. Journal of Experimental Zoology 262(4) :<br />
383-390.<br />
Diller LV & Wallace RL (1984) Reproductive biology of the northern pacific<br />
rattlesnake (Crotalus viridis oreganus). Herpetologica 40 : 182-193.<br />
Dmi'el R (1986) Intra- and interspecific allometry of oxygen consumption in<br />
snakes. Journal of Herpetology 20 : 265-267.<br />
Doughty P & Shine R (1997) Detecting life history trade-offs : measuring energy<br />
stores in “capital” breeders reveals costs of reproduction. Oecologia 110 : 508-<br />
513.<br />
Doughty P & Shine R (1998) Energy allocation to reproduction in a viviparous<br />
lizard (Eulamprus tympanum) : the role of long-term energy stores. Ecology 79 :<br />
1073-1083.<br />
283
Downes SJ & Shine R (1999) Do incubation-induced changes in a lizard's<br />
phenotype influence its vulnerability to predators? Oecologia 120(1) : 9-18.<br />
Drent RH & Daan S (1980) <strong>The</strong> prudent parent: energetic adjustments in avian<br />
breeding. Ardea 68 : 225-252.<br />
Du Toit JT (1995) Sexual segregation in Kudu : sex differences in competitive<br />
ability, predation risk or nutritional needs? South African Journal of Wildlife<br />
Research 25 : 127-134.<br />
Dunham AE, Miles DB & Reznick DN (1988) Life history patterns in squamate<br />
reptiles. In : Gans C & Huey RB (eds), Biology of the Reptilia 16, pp 441-522.<br />
Alan R Liss Inc, New York.<br />
Duval D, Arnorld S & Schuett GW (1992) Pitviper mating system: Ecological<br />
potential, sexual selection and microevolution. In : Campbell JA & Brodie ED Jr<br />
(eds) Biology of the pitvipers, pp 321-336. Tyler, Texas.<br />
Duval D, Schuett GW & Arnorld SJ (1993) Ecology and evolution of snake mating<br />
systems. In : Seigel RA, Collins JT & Novak SS (eds) Snakes, Ecology and<br />
Behavior, pp 165-200. McGraw-Hill, Inc, New York.<br />
E.<br />
Elphick MJ & Shine R (1999) Sex differences in optimal incubation temperatures<br />
in a scincid lizard species. Oecologia 118(4) : 431-437.<br />
Else PL & Hulbert AJ (1981) Comparison of the "mammal machine" and the<br />
"reptile machine" : energy production. American Journal of Physiology 240 : R3-<br />
R9.<br />
Engelmann F & Rau I (1965) A correlation <strong>between</strong> feeding and the sexual cycle<br />
in Leucophaea maderae (Blattaria). Journal of Insect Physiology 11 : 53-64.<br />
Erikstad EK, Asheim M, Faucahld P, Dahlaug LT & Tveraa T (1997) Adjustment<br />
of parental effort in the puffin ; the roles of adult body condition and chick size.<br />
Behavioural Ecology and Sociobiology 40 : 95-100.<br />
F.<br />
Farmer CG (2000) Parental care : key to understanding endothermy and other<br />
convergent features in Birds and Mammals. American Naturalist 155 : 326-334.<br />
284
Farr DR & Gregory PT (1991) Sources of variation in estimating litter<br />
characteristics of the garter snake, Thamnophis elegans. Journal of Herpetology<br />
25 : 261-268.<br />
Fergusson B & Bradshaw SD (1991) Plasma arginine vasotocin, progesterone,<br />
and luteal development during pregnancy in the viviparous lizard Tiliqua rugosa.<br />
General and Comparative Endocrinology 82 : 140-151.<br />
Festa-Bianchet M, Gaillard JM & Jorgenson JT (1998) Mass- and density-<br />
dependent reproductive success and reproductive costs in a capital breeder.<br />
American Naturalist 152 : 367-379.<br />
Fisher RA (1930) <strong>The</strong> Genetical <strong>The</strong>ory of Natural Selection. Oxford University<br />
Press, Oxford.<br />
Fitch HS (1960) Autoecology of the copperhead. University of Kansas<br />
Publications of the Museum of Natural History 13 : 85-288.<br />
Fitch HS (1981) Sexual size differences in reptiles. Miscellaneous Publications<br />
of the Museum of Natural History, University of Kansas 70, 72.<br />
Fitch HS (1987) Collecting and life-history techniques. In : Seigel RA, Collins JT<br />
& Novak SS (eds) Snakes Ecology and Evolutionary Biology, pp 143-164.<br />
Macmillan, New York.<br />
Ford NB & Seigel RA (1989a) Relationships among body size, clutch size, and<br />
egg size in three species of oviparous snakes. Herpetologica 45 : 75-83.<br />
Ford NB & Seigel RA (1989b) Phenotypic plasticity in reproductive traits :<br />
evidence from a viviparous snake. Ecology 70 : 1768-1774.<br />
Ford NB & Seigel RA (1994) Phenotypic plasticity : implication for captive<br />
breeding and conservation programs. In : Murphy JB, Collins JT & Adler K (eds)<br />
Captive Managment and Conservation of Amphibians and Reptiles, pp 175-182.<br />
Society for the Study of Amphibians and Reptiles, Oxford, OH.<br />
Forsman A (1996) An experimental test for <strong>food</strong> effects on head size allometry in<br />
juvenile snakes. Evolution 50 : 2536-2542.<br />
Fox W, Gordon C & Fox MH (1961) Morphological effects of low temperatures<br />
during the embryonic development of the garter snake, Thamnophis elegans.<br />
Zoologica 46 : 57-71.<br />
285
G.<br />
Gadgil M & Bossert PW (1970) Life historical consequences of natural selection.<br />
American Naturalist 104 : 1-24.<br />
Garcia-Berthou E (2001) On the misuse of residuals in ecology : testing<br />
regression residuals vs. the analysis of covariance. Journal of Animal Ecology 70<br />
: 708-711.<br />
Glazier DS (2000) Is fatter fitter? Body storage and reproduction in ten<br />
populations of the freshwater amphipod Gammarus minus. Oecologia 122 : 335-<br />
345.<br />
Gregory PT (1982) Reptilian hibernation. In : Gans C & Pough FH (eds) Biology<br />
of the Reptilia, Volume 13, pp. 53-154. Academic Press, London.<br />
Gregory PT, Larsen KW & Farr DR (1992) Snake litter size = live young + dead<br />
young + yolks. Herpetologica 2 : 145-146.<br />
Gregory PT, Lisa HC & Skebo KM (1999) Conflicts and interactions among<br />
reproduction, thermoregulation and feeding in viviparous reptiles : are gravid<br />
snakes anorexic? Journal of Zoology London 248 : 231-241.<br />
Gregory PT, Macartney JM & Larsen KW (1987) Spatial patterns and<br />
movements. In : Seigel RA, Collins JT & Novak SS (eds.), Snakes : Ecology and<br />
Evolutionary Biology, pp. 366-395. Macmillan Publishing Company, New York.<br />
Gregory PT & Skebo KM (1998) Trade-offs <strong>between</strong> reproductive traits and the<br />
influence of <strong>food</strong> <strong>intake</strong> during pregnancy in the garter snake, Thamnophis<br />
elegans. American Naturalist 151 : 477-486.<br />
Guillette LJ (1982) Effects of gravidity on the metabolism of the reproductively<br />
bimodal lizard, Sceloporua avenues. <strong>The</strong> Journal of Experimental Zoology 223 :<br />
33-36.<br />
Gustafson L & Sutherland W (1988) <strong>The</strong> costs of reproduction in the collared<br />
flycatcher Ficedula albicollis. Nature 335 : 813-815.<br />
286
H.<br />
Hahn WE & Tinkle DW (1965) Fat body cycling and experimental evidence for its<br />
adaptive significance to ovarian follicle development in the lizard, Uta<br />
stansburiana. Journal of Experimental Zoology 158 : 79-86.<br />
Harlow P & Grigg G (1984) Shivering thermogenesis in the brooding diamond<br />
python, Python spilotes spilotes. Copeia 1984 : 959-965.<br />
Highfill DR & Mead RAM (1975) Function of corpora lutea of pregnancy in the<br />
viviparous Garter snake Thamnophis elegans. General and Comparative<br />
Endocrinology 27 : 401-407.<br />
Hoglünd J & Sheldon BC (1998) <strong>The</strong> cost of reproduction and sexula selection.<br />
Oïkos 83 : 478-483.<br />
Houston D & Shine R (1994) Movement and activity pattern of Arafura file snakes<br />
(Serpentes: Achrocordida) in tropical Australia. Herpetologica 50 : 349-357.<br />
Howe JC, Rumpler WV & Seale JL (1993) Energy expenditure by indirect<br />
calorimetry in premenopausal women : Variation with one menstrual cycle.<br />
Journal of Nutritional Biochemistry 4(5) : 268-273.<br />
Huey BR (1982) Temperature, Physiology, and the Ecology of Reptiles. In : Gans<br />
C & Pough HF (eds) Biology of the Reptilia 12, Physiological Ecology, pp. 25-95.<br />
Academic Press.<br />
Hulbert AJ & Else PL (1989) Evolution of mammalian endothermic metabolism :<br />
mitochondrial activity and cell composition. American Journal of Physiology 256 :<br />
R63-R69.<br />
I.<br />
Ineich I (1995) Actual knowledge on venomous snakes systematics. Bulletin<br />
de la société Herpetologique de France 75-76 : 7-24<br />
J.<br />
Jakob EM, Marshall SD & Uetz GW (1996) Estimating fitness : a comparison of<br />
body condition indices. Oikos 77 : 61-67.<br />
287
James CD & Whitford WG (1994) An experimental study of phenotypic plasticity<br />
in the clutch size of a lizard. Oikos 70 : 49-56.<br />
Jayne BC & Bennett AF (1990) Selection on locomotor performance capacity in a<br />
natural population of garter snakes. Evolution 44 : 1204-1229.<br />
Jönsson KI (1997) Capital and income breeding as alternative tactics of resource<br />
use in reproduction. Oikos 78 : 57-66.<br />
Jönsson KI & Tuomi J (1994) Costs of reproduction in a historical perspective.<br />
Trends in Ecology and Evolution 9 : 304-307.<br />
Jönsson KI, Tuomi J & Järemo J (1995a) Reproductive effort tactics : balancing<br />
pre- and postbreeding costs of reproduction. Oikos 74 : 35-44.<br />
Jönsson KI, Tuomi J & Järemo J (1995b) On the consequences of pre- and<br />
postbreeding costs in the evolution of reproductive effort tactics. Ecoscience 2 :<br />
311-320.<br />
K.<br />
King RB (1993) Determinants of offspring number and size in the brown snake,<br />
Storeria dekayi. Journal of Herpetology 27 : 175-185.<br />
Klauber LM (1972) Rattlesnakes : their habits, life histories and influence on<br />
Mankind, 2nd ed., 2 vols., University of California Press. Bekerley.<br />
Krebs CJ & Myers JH (1974) Population cycles in small mammals. Advances in<br />
Ecological Research 8 : 267-399.<br />
Kunkele J (2000) Energetics of gestation relative to lactation in a precocial<br />
rodent, the guinea pig (Cavia porcellus). Journal of Zoology (London) 250 : 533-<br />
539.<br />
L.<br />
Lack D (1954) <strong>The</strong> natural regulation of animal numbers. Oxford University<br />
Press, Oxford.<br />
Larsen LO (1980) Physiology of adult lampreys, with special regard to natural<br />
starvation, reproduction, and death after spawning. Canadian Journal of Fisheries<br />
and Aquatic Science 37 : 1762-1779.<br />
288
Lebreton JD, Burnham KP, Clobert J & Anderson DR (1992) Modelling survival<br />
and testing biological hypothesis using marked animals : a unified approach with<br />
case studies. Ecological Monograph 62 : 67-118.<br />
Lee SJ, Witter MS, Cuthill IC & Goldsmith AR (1996) Reduction in escape<br />
performance as a cost of reproduction in gravid starlings, Sturnus vulgaris.<br />
Proceedings of the Royal Society of London, Series B 263 : 619-623.<br />
Lequette B & Weimerskirch H. (1990) Influence of parental experience on the<br />
growth of wandering albatross chicks. Condor 92 : 726-731.<br />
Lessells C (1991) <strong>The</strong> evolution of life histories. In : Krebs R & Davies ND (eds)<br />
Behavioural Ecology, pp. 32-68. Blackwell, Oxford, London.<br />
Levins R (1968) Evolution in changing environments. Princeton University press,<br />
Princeton.<br />
Lima SL & Dill LM (1990) Behavioral decisions made under the risk of predation:<br />
a review and prospectus. Canadian Journal of Zoology 68 : 619-640.<br />
Loumbourdis NS & Hailey A (1985) Activity metabolism of the lizard Agama<br />
stellio stellio. Comparative Biochemistry and Physiology 82A : 687-691.<br />
Lourdais O, Bonnet X & Doughty P (2002a) Costs of anorexia during pregnancy<br />
in a viviparous snake (Vipera aspis). Journal of Experimental Zoology 292 : 487-<br />
493.<br />
Lourdais O, Bonnet X, Shine R, DeNardo D, Naulleau G & Guillon M (2002b)<br />
Capital-breeding and reproductive effort in a variable environment: a longitudinal<br />
study of a viviparous snake. Journal of Animal Ecology 71 : 470-479.<br />
Lourdais O, Bonnet X, Shine R & Taylor E. (2002c) When does a reproducing<br />
female viper (Vipera aspis) "decide" on her litter size? Journal of Zoology, in press<br />
Lourdais O, Bonnet X, DeNardo D & Naulleau G. (2002d) Does sex differences in<br />
reproductive eco-physiology translate in different demographic patterns ?<br />
Population Ecology, in press.<br />
Luiselli L, Capula M & Shine R (1996) Reproductive output, costs of reproduction,<br />
and ecology of the smooth snake, Coronella austriaca, in the eastern Italian Alps.<br />
Oecologia 106 : 100-110.<br />
289
M.<br />
Madsen T & Shine R (1992a) Determinants of reproductive success in female<br />
adders Vipera berus. Oecologia 92 : 40-47.<br />
Madsen T & Shine R (1992b) Temporal variability in sexual selection on<br />
reproductive tactics and body size in male snakes. American Naturalist 141 : 161-<br />
171.<br />
Madsen T & Shine R (1993) Costs of reproduction in a population of european<br />
adders. Oecologia 94 : 488-495.<br />
Madsen T & Shine R (1996) Determinants of reproductive output in female water<br />
pythons (Liasis fuscus: Pythonidae). Herpetologica 52 : 146-159.<br />
Madsen T & Shine R (1999) <strong>The</strong> adjustment of reproductive threshold to prey<br />
abundance in a capital breeder. Journal of Animal Ecology 68 : 571-580.<br />
Madsen T, Shine R, Olsson M & Wittzell H (1999) Restoration of an inbred adder<br />
population. Nature 402 : 34-35.<br />
Marti DM (1990) Sex and dimorphism in the barn owl, a test of mate choice. Auk<br />
107 : 246-254.<br />
Martin WH (1993) Reproduction of the timber rattlesnake (Crotalus horridus) in<br />
the Appalachian mountains. Journal of Herpetology 27 : 133-143.<br />
Martin K (1995) Patterns and mechanisms for age-dependent reproduction and<br />
survival in birds. American Zoologist 35 : 340-348.<br />
Mauget C, Mauget R & Sempéré A (1997) Metabolic rate in female european roe<br />
deer (Capreolus capreolus): Incidence of reproduction. Canadian Journal of<br />
Zoology 75(5) : 731-739.<br />
Mayr E (1963) Animal species and evolution. Harvard University press.<br />
McCleery RH, Clobert J, Julliard R & Perrins CM (1996) Nest predation and<br />
delayed cost of reproduction in the great tit. Journal of Animal Ecology 65 : 96-<br />
104.<br />
Meijer T & Drent S (1999) Re-examination of the capital and income dichotomy in<br />
breeding birds. Ibis 141 : 399-414.<br />
290
Méndez De La Cruz FR, Guillette LJ Jr & Villagran-Santa Cruz M (1993)<br />
Differential atresia of ovarian follicles and its effects on the clutch size of two<br />
populations of the viviparous lizard Sceloporus mucronatus. Functional Ecology 7<br />
: 535-540.<br />
Monaghan P & Nager RN (1997) Why don’t birds lay more eggs? Trends in<br />
Ecology and Evolution 7 : 270-274.<br />
Monney JC (1994) Note sur la reproduction et la taille des nouveau-nés chez la<br />
vipère aspic (Vipera aspis) et la vipère péliade (Vipera berus) dans l’Oberland<br />
bernois. Bulletin de la Société Fribourgeoise de Sciences Naturelles 83 : 61-74.<br />
Mrosovsky N & Sherry DF (1980) Animal anorexias. Science 207 : 837- 842.<br />
Mysterud A (2000) <strong>The</strong> <strong>relationship</strong> <strong>between</strong> ecological segregation and sexual<br />
body size dimorphism in large herbivores. Oecologia 124 : 40-54.<br />
N.<br />
Nagy KA (1987) Field metabolic rate and <strong>food</strong> requirement scaling in mammals<br />
and birds. Ecological Monograph 57 : 111-128.<br />
Nagy KA, Huey RB & Bennett AF (1984) Field energetics and foraging mode of<br />
Kalahari lacertid lizards. Ecology 65 : 588-596.<br />
Nagy KA, Meienberger C, Bradshaw SD & Wooller RD (1995) Field metabolic<br />
rate of a small marsupial mammal, the Honey Possum (Tarsipes rostratus).<br />
Journal of Mammalogy 76 : 862-866.<br />
Naulleau G (1965) La biologie et le comportement prédateur de Vipera aspis au<br />
laboratoire et dans la nature. Bulletin Biologique franco-belge 99 : 395-524.<br />
Naulleau G (1981) Détermination des périodes de l’ovulation chez Vipera aspis<br />
et Vipera berus dans l’ouest de la France, étudiée par radiographie. Bulletin de la<br />
Société des Sciences Naturelles de L’Ouest de la France 3 : 151-153.<br />
Naulleau G (1983a) Action de la température sur la digestion chez cinq espèces<br />
de vipères Européennes du genre Vipera. Bulletin de la Société Zoologique de<br />
France 108 : 47-58.<br />
Naulleau G (1983b) <strong>The</strong> effects of temperature on digestion in Vipera aspis.<br />
Journal of Herpetology 17 : 166-170.<br />
291
Naulleau G (1986) Effects of temperature on “gestation” in Vipera aspis and<br />
Vipera berus (Reptilia: Serpentes) Studies in Herpetology (ed Röcek, Z), pp. 489-<br />
494, Charles Univ., Prague.<br />
Naulleau G (1997) La vipère aspic. Eveil Nature, Saint Yrieix.<br />
Naulleau G & Bidaut C (1981) Intervalle entre l'accouplement, l'ovulation et la<br />
parturition chez Vipera aspis L. (Reptiles, Ophidiens, Vipéridés), dans différentes<br />
conditions expérimentales, étudié par radiographie. Bulletin de la Société<br />
Zoologique de France 106 : 137-143.<br />
Naulleau G & Saint Girons H (1981) Poids des nouveau-nés et reproduction de<br />
Vipera aspis (Reptilia: Viperidae), dans des conditions naturelles et artificielles.<br />
Amphibia-Reptilia 2 : 51-62.<br />
Naulleau G & Bonnet X (1995) Reproductive ecology, body fat reserves and<br />
foraging mode of two contrasted snakes species: Vipera aspis (terrestrial,<br />
viviparous) and Elaphe longissima (semi-arboreal, oviparous). Amphibia-Reptilia<br />
16 : 37-46.<br />
Naulleau G & Bonnet X (1996) Body condition threshold for breeding in a<br />
viviparous snake. Oecologia 107 : 301-306.<br />
Naulleau G, Célérier ML & Saint Girons H (1984) Influence de la température<br />
d’acclimatation sur la consommation d’oxygène de Vipera aspis (Reptilia,<br />
Viperidae). Comptes Rendus de l’Académie des Sciences, Paris, Série III 12 :<br />
511-514.<br />
Naulleau G, Bonnet X & Duret S (1996) Déplacements et domaines vitaux des<br />
femelles reproductrices de vipères aspic Vipera aspis (Reptilia, Viperidae) dans le<br />
centre ouest de la France. Bulletin de la Société Herpétologique de France 78 :<br />
5-18.<br />
Naulleau G, Verheyden C & Bonnet X (1997) Prédation spécialisée sur la Vipère<br />
aspic Vipera aspis par un couple de buses variables Buteo buteo. Alauda 65 :<br />
155- 160.<br />
Naulleau G, Bonnet X, Vacher-Vallas M, Shine R & Lourdais O (1999) Does less-<br />
than-annual production of offspring by female vipers (Vipera aspis) mean less-<br />
than-annual mating? Journal of Herpetology 33 : 688-691.<br />
292
Nichols JD, Hepp GR, Pollock KH & Hines JE (1987) <strong>The</strong> husting dilemma : a<br />
methodological note. Ecology 68 : 213-217.<br />
Niewiarowski PH & Dunham AE (1994) <strong>The</strong> evolution of reproductive effort in<br />
squamate reptiles : costs, trade-offs, and assumptions reconsidered. Evolution<br />
48:137-145.<br />
Niewiarowski PH & Dunham AE (1998) Effect of mortality risk and growth on a<br />
model of reproductive effort : why the Shine & Schwarzkopf model is not general.<br />
Evolution 52 : 1236-1241<br />
Nilsson JA & Svensson E (1996) <strong>The</strong> cost of reproduction : a new link <strong>between</strong><br />
current reproductive effort and future reproductive success. Proceedings of the<br />
Royal Society, London B 263 : 711-714.<br />
Nur N (1988) <strong>The</strong> costs of reproduction in birds : an examination of the evidence.<br />
Ardea 76: 155-168.<br />
O.<br />
Oakes EJ, Harmsen R & Erberl C (1992) Sex, age, and seasonal differences in<br />
the diets and activity budgets of muskoxen (Ovibos moschatus). Canadian<br />
Journal of Zoology 70 : 605-616.<br />
Oakwood M, Bradley AJ & Cockburn A (2001) Semelparity in a large marsupial.<br />
Proceedings of the Royal Society. Series B 268 : 407-411.<br />
Olsson M, Madsen T & Shine R (1997) Is sperm really so cheap? Costs of<br />
reproduction in male adders, Vipera berus. Proceedings of the Royal Society<br />
séries B 264 : 455-459.<br />
Olsson M, Shine R & Bak-Olsson E (2000) Locomotor impairment of gravid<br />
lizards: is the burden physical or physiological? Journal of Evolutionary Biology 13<br />
: 263-268.<br />
Olsson O (1997) Clutch abandonment: a state-dependent decision in king<br />
penguins. Journal of Avian Biology 28 : 264-267.<br />
O'Sullivan A, Martin A & Brown MA (2001) Efficient fat storage in premenopausal<br />
women and in early pregnancy: A role for estrogen. Journal of Clinical<br />
Endocrinology and Metabolism 86(10) : 4951-4956.<br />
293
Otis DL, Burnham KPCG & Anderson DR (1978) Statistical inference for capture<br />
data on closed animal populations. Wildlife Monograph 62 : 1-135.<br />
Owen-Smith N (1993) Comparative mortality rates of male and female Kudus: the<br />
costs of sexual size dimorphism. Journal of Animal Ecology 62 : 428-440.<br />
P.<br />
Packard GC & Packard MJ (1988) <strong>The</strong> physiological ecology of reptilian eggs<br />
and embryos. In : Gans C. Huey R. (eds). Biology of the Reptilia, pp 523-605.<br />
Alan R Liss, New York, USA 16.<br />
Packard GC & Boardman TJ (1988) <strong>The</strong> misuse of ratios, indices and<br />
percentages in ecophysiological research. Physiological Zoology 61 : 1-9.<br />
Peterson CR, Gibson AR & Dorcas ME (1993) Snake thermal ecology: the<br />
causes and consequences of body-temperature variation. In : Seigel RA, Collins<br />
JT & Novak SS (eds) Snakes, Ecology and Behavior, pp 241-314. McGraw-Hill,<br />
Inc, New York.<br />
Plummer MV (1983) Annual variation in stored lipids and reproduction in green<br />
snakes (Opheodrys aestivus). Copeia 1983 : 741-745.<br />
Pough FH (1980) <strong>The</strong> advantages of ectothermy for tetropods. American<br />
Naturalist 115 : 92-112.<br />
Price T, Kirkpatrick M & Arnold SJ (1988) Directionnal selection and the evolution<br />
of breeding date in birds. Science 240 : 798-799.<br />
R.<br />
Ranta E, Kaitala V, Alaja S, Tesar D (2000a) Nonlinear dynamics and the<br />
evolution of semelparous and iteropaous reproductive strategies. <strong>The</strong> American<br />
Naturalist 155 : 294-300.<br />
Ranta E, Tesar D, Alaja S & Kaitala V (2000b) Does evolution of iteroparous and<br />
semelparous reproduction call for spatially structured systems? Evolution 54 :<br />
145-150.<br />
Reznick DN (1992) Measuring costs of reproduction. Trends in Ecology and<br />
Evolution 7 : 42-45.<br />
294
Reznick DN (1985) Costs of reproduction: an evaluation of the empirical<br />
evidence. Oikos 44 : 257-267.<br />
Reznick DN, Nunney L & Tessier A (2000) Big houses, big cars, superfleas and<br />
the costs of reproduction. Trends in Ecology and Evolution 15 : 421-425.<br />
Rhen T & Lang J (1999) Incubation temperature and sex affect mass and energy<br />
reserves of hatchling snapping turtles, Chelydra serpentina. Oikos 86(2) : 311-<br />
319.<br />
Roff PA (1992) <strong>The</strong> evolution of life histories. Chapman et Hall, New York,<br />
Routledge.<br />
Rose MR & Bradley TJ (1998) Evolutionary physiolgy of the cost of reproduction.<br />
Oikos 83 : 443-451.<br />
S.<br />
Saint Girons H (1949) Les moeurs nuptiales de la Vipère aspic. Terre et Vie 96 :<br />
110-113.<br />
Saint Girons H (1952) Ecologie et éthologie des vipères de France. Annales des<br />
Sciences Naturelles, Zoologie 14 : 263-343.<br />
Saint Girons H (1957a) Croissance et fécondité de Vipera aspis (L). Vie et Milieu<br />
8 : 265-286.<br />
Saint Girons H (1957b) Le cycle sexuel chez Vipera aspis (L) dans l'ouest de la<br />
France. Bulletin Biologique de France et de Belgique 91 : 284-350.<br />
Saint Girons H (1973) Le cycle sexuel de Vipera aspis (L) en montagne. Vie et<br />
Milieu 23 : 309-328.<br />
Saint Girons H (1975) Critères d’âge, structure et dynamique des populations de<br />
reptiles. In : Bourlière F & Lamotte M (eds) Problèmes d’écologie, pp 233-252.<br />
Masson, Paris.<br />
Saint Girons H (1979) Les cycles alimentaires des vipères européennes dans<br />
des conditions semi-naturelles. Annales de Biologie Animale, de Biochimie et de<br />
Biophysique 19 : 125-134.<br />
295
Saint Girons H (1996) Structure et évolution d'une petite population de Vipera<br />
aspis (L.) dans une région de Bocage de l'ouest de la France. Terre et Vie 51 :<br />
223-241.<br />
Saint Girons H (1997) Vipera aspis. In : Gasc JP et al. (eds) Atlas of Amphibians<br />
and Reptiles in Europe, pp 386-387. SEH, MNHN, Paris.<br />
Saint Girons H & Naulleau G (1981) Poids des nouveau-nés et stratégies<br />
reproductrices des vipères européennes. Terre et vie 35 : 597-616.<br />
Saint Girons H & Duguy R (1992) Evolution de la masse corporelle et de la<br />
masse relative des corps gras, des ovaires et des œufs au cours des cycles<br />
reproducteurs chez Vipera aspis. Amphibia-Reptilia 13 : 351-364.<br />
Saint Girons H, Naulleau G & Célérier ML (1985) Le métabolisme aérobie de<br />
Vipera aspis et Vipera berus (Reptilia: Viperidae) acclimatées à des conditions<br />
naturelles. Influence de la température et comparaison des deux espèces.<br />
Amphibia-Reptilia 6 : 71-81.<br />
Salamolard M, Butet A, Leroux A & Bretagnolle V ( 2000) Responses of an avian<br />
predator to variations in prey density at a temperate latitude. Ecology 81 : 2428–<br />
2441.<br />
Schaffer WM (1974) Optimal reproductive effort in fluctuating environments.<br />
American Naturalist 964 : 783-900.<br />
Schaffer WM & Elson PF (1975) <strong>The</strong> adaptative significance of variations in life<br />
history among local populations of Atlantic salmon in North America. Ecology 56 :<br />
577-590.<br />
Schneider JE, Zhou D & Blum RB (2000) Leptin and metabolic control of<br />
reproduction. Hormones and Behavior 37 : 306-326.<br />
Secor SM & Diamond J (1995) Adaptive responses to feeding in Burmese<br />
pythons: pay before pumping. <strong>The</strong> Journal of Experimental Biology 198 : 1313-<br />
1325.<br />
Secor SM, Stein ED & Diamond JM (1994) Rapid up-regulation of snake intestine<br />
in response to feeding : a new model of intestinal adaptation. American Journal of<br />
Physiology 266 : G695-G705.<br />
296
Seigel RA (1993) Summary : future research on snakes, or how to combat “lizard<br />
envy”. In : Seigel RA, Collins JT & Novak SS (eds) Snakes, Ecology and<br />
Behavior, pp 395-402. McGraw-Hill, Inc, New York<br />
Seigel RA & Fitch HS (1984) Ecological patterns of relative clutch mass in<br />
snakes, Oecologia 62 : 293-301.<br />
Seigel RA & Fitch HS (1985) Annual variation in reproduction in snakes in a<br />
fluctuating environment. Journal of Animal Ecology 54 : 497-505.<br />
Seigel RA & Ford NB (1987) Reproductive Ecology. In : Seigel RA, Collins JT &<br />
Novak SS (eds) Snakes: Ecology and Evolutionary Biology, pp 210-252.<br />
Macmillan, New York.<br />
Seigel RA & Ford NB (1991) Phenotypic plasticity in the reproductive<br />
characteristics of an oviparous snake, Elaphe guttata: implication for life history<br />
studies. Herpetologica 47 : 301-307.<br />
Seigel RA & Ford NB (1992) Effect of energy input on variation in clutch size and<br />
offspring size in a viviparous reptile. Functionnal Ecology 6 : 382-385.<br />
Seigel RA & Ford NB (2001) Phenotypic plasticity in reproductive traits:<br />
geographical variation in plasticity in a viviparous snake. Functionnal Ecology 15 :<br />
16-42.<br />
Seigel RA, Huggins MM & Ford NB (1987) Reduction in locomotor ability as a<br />
cost of reproduction in gravid snake. Oecologia 73 : 481-485.<br />
Shine R (1980) "Costs" of reproduction in reptiles. Oecologia 46 : 92-100.<br />
Shine R (1985) <strong>The</strong> evolution of viviparity in reptiles : an ecological analysis. In :<br />
Gans C & Billet F (eds.), Biology of the Reptilia Vol. 15, p. 605-694. John Wiley<br />
and Sons, INC. New York.<br />
Shine R (1988a) Parental care in reptiles. In : Gans C & Huey RB (eds) Biology<br />
of the Reptilia 16, pp 275-329. Alan R. Liss, New York.<br />
Shine R (1988b) Constraints on reproductive investment: a comparison <strong>between</strong><br />
aquatic and terrestrial snakes. Evolution 42 : 17-27.<br />
Shine R (1991) Australian Snakes : A Natural History. Reed Publishing.<br />
297
Shine R (1999) Why is sex determined by nest temperature in many reptiles?<br />
Trends in Ecology and Evolution 14(5) : 186-189.<br />
Shine R & Schwarzkopf L (1992) <strong>The</strong> evolution of reproductive effort in lizards<br />
and snakes. Evolution 46: 62-75.<br />
Shine R & Harlow P (1993) Maternal thermoregulation influences offspring<br />
viability in a viviparous lizard. Oecologia 96 :122-127.<br />
Shine R & Harlow PS (1996) Maternal manipulation of offspring phenotypes via<br />
nest-site selection in an oviparous lizard. Ecology 77(6) : 1808-1817.<br />
Shine R & Madsen T (1997) Prey abundance and predator reproduction : rats<br />
and pythons on a tropical Australian floodplain. Ecology 78 : 1078-1086.<br />
Shine R & Downes SJ (1999) Can pregnant lizards adjust their offspring<br />
phenotypes to environmental conditions? Oecologia 119(1) : 1-8.<br />
Shine R & Bonnet X (2000) Studying the snakes : a “new model organism” in<br />
ecological research? Trends in Ecology and Evolution 15 : 221-222.<br />
Shine R, Madsen T, Elphick M & Harlow PS (1997) <strong>The</strong> influence of nest<br />
temperatures and maternal brooding on hatchling phenotypes in water pythons.<br />
Ecology 78 : 1713-1721.<br />
Sibly RM & Calow P (1984) Direct and absorption costing in the evolution of life<br />
cycles. Journal of <strong>The</strong>oretical Biology 111 : 463-473.<br />
Sinervo B & Licht P (1991) Hormonal and physiological control of clutch size,<br />
egg size, and egg shape in side-blotched lizards (Uta stansburiana) : constraints<br />
on the evolution of lizard life histories. Journal of Experimental Zoology 257 :<br />
252-264.<br />
Sinervo B & DeNardo DF (1996) Costs of reproduction in the wild: path analysis<br />
of natural selection and experimental tests of causation. Evolution 50 : 1299-<br />
1313.<br />
Sinervo B & Svensson E (1998) Mechanistic and selective causes of life history<br />
trade-offs and plasticity. Oikos 83 : 432-444.<br />
Singh D & Ramachandran AV (2000) Circannual variations in carbohydrate<br />
metabolism in relation to testicular cyclicity and the influence of pinealectomy on<br />
298
altered adrenocortical activity. Journal of Animal Morphology and Physiology<br />
47(1-2) : 33-44.<br />
Southwood TRE (1988) Tactics, strategies and templets. Oikos 52 : 3-18.<br />
Speakman JR & McQueenie J (1996) Limits to sustained metabolic rate: <strong>The</strong> link<br />
<strong>between</strong> <strong>food</strong> <strong>intake</strong>, basal metabolic rate and morphology in reproducing mice<br />
Mus musculus. Physiological Zoology 69(4) : 746-769.<br />
Statsoft (1995) Statistica. Statsoft, Tulsa.<br />
Statsoft (2001) Statistica. Statsoft, Tulsa.<br />
Stearns SC (1989) Trade-offs in life-history evolution. Functional Ecology 3 : 259-<br />
268.<br />
Stearns SC (1992) <strong>The</strong> evolution of life histories. Oxford University Press, New<br />
York.<br />
Stewart JR (1992) Placental structure and nutritional provision to embryos in<br />
predominantly lecithotrophic viviparous reptiles. American zoologist 32 : 303-312.<br />
T.<br />
Thibault C & Levasseur MC (1991) La reproduction chez les mammifères et<br />
l'homme. Ellipse Editions, INRA.<br />
Thompson GG & Withers PC (1992) Effects of body mass and temperature on<br />
standard metabolic rates for two Australian varanid lizards (Varanus gouldii and<br />
V. panoptes). Copeia 1992 : 343-350.<br />
Tinkle DW (1979) Long-term field studies. Biosciences 20 : 717.<br />
Todd AW, Keith LB & Fisher CA (1981) Population ecology of coyotes during a<br />
fluctuation of snowshoe hares. Journal of Wildlife Management 45 : 629-640.<br />
Tombre IM & Erikstad KE (1996) An experimental study of incubation effort in<br />
high-arctic barnacle geese. Journal of Animal Ecology 65 : 325-331.<br />
Tuomi J, Hakala T & Haukioja E (1983) Alternative concepts of reproductive<br />
effort, costs of reproduction and selection in life-history evolution. American<br />
Zoologist 23 : 25-34.<br />
V.<br />
299
Vacher-Vallas M (1997) Comportement sexuel, condition corporelle et stratégie<br />
de reproduction chez Vipera aspis: étude en milieu naturel. Unpublished Ph D<br />
thesis, Université François Rabelais, Tours.<br />
Vacher-Vallas M, Bonnet X & Naulleau G (1999) Relations entre les<br />
comportements sexuels et les accouplements chez Vipera aspis: étude en milieu<br />
naturel. Terre et Vie 54 : 375-391.<br />
Van Noordwijk AJ & De Jong G (1986) Acquisition and allocation of resources:<br />
their influence on variation in life-history tactics. American Naturalist 128 : 137-<br />
142.<br />
Villevieille S (1997) Application de la résonance magnétique nucléaire pour<br />
localiser et quantifier les réserves lipidiques chez un reptile vivipare, Vipera aspis.<br />
Unpublished D.E.A. manuscript, Université François Rabelais, Tours.<br />
Vitt LJ & Congdon JD (1978) Body shape, reproductive effort, and relative clutch<br />
mass in lizards: resolution of a paradox. American Naturalist 112 : 595-608.<br />
Vitt LJ & Price HJ (1982) Ecological and evolutionary determinants of relative<br />
clutch mass in lizards. Herpetologica 38 : 237-255.<br />
W.<br />
Weeks SC (1996) <strong>The</strong> hidden cost of reproduction : reduced <strong>food</strong> <strong>intake</strong> caused<br />
by spatial constraints in the body cavity. Oikos 75 : 345-349.<br />
Whittier JM & Crews D (1990) Body mass and reproduction in female red-sided<br />
garter snakes (Thamnophis sirtalis parietalis). Herpetologica 46 : 219-226.<br />
Wilbur HM & Morin PJ (1988) Life history evolution in turtles. In : Gans C & Huey<br />
RB (eds) Biology of the Reptilia 16, pp 387-439. Alan R Liss, INC, New York.<br />
Williams GC (1966a) Adaptation and Natural Selection : A Critique of Some<br />
Current Evolutionary Thought. Princeton University Press, New Jersey, USA.<br />
Williams GC (1966b) Natural selection, the costs of reproduction, and a<br />
refinement of Lack's principle. American Naturalist 100 : 687-690.<br />
Winkel W & Winkel D (1995) Costs and benefits of second broods in coal tits<br />
(Parus ater). Journal of Ornithology 136 : 29-36.<br />
300
Winkler DW & Wallin K (1987) Offspring size and number : a life history model<br />
linking effort per offspring and <strong>total</strong> effort. American Naturalist 129 : 708-720.<br />
Y.<br />
Young TP (1990) Evolution of semelparity in Mount Kenya lobelias. Evolutionnary<br />
Ecology 4 : 157-171.<br />
Young TP & Auspurger CK (1991) Ecology and evolution of long-lived<br />
semelparous plants. Trends in Ecology and Evolution 6 : 285-289.<br />
Z.<br />
Zar JH (1984) Biostatistical Analysis. Prentice-Hall, New Jersey.<br />
Zari TA (1991) <strong>The</strong> influence of body mass and temperature on the standard<br />
metabolic rate of the herbivourous desert lizard, Uromastyx microlepis. Journal of<br />
<strong>The</strong>rmal Biology 16 : 129-133.<br />
Zuffi MAL, Giudici F & Iolae P (1999) Frequency of reproduction and reproductive<br />
effort in female Vipera aspis from a southern population. Acta Oecologica 20 :<br />
633-638.<br />
Zuffi MAL (2002) A critic of the systematic position of the aspic viper subspecies<br />
Vipera aspis aspis (Linnaeus 1758), Vipera aspis atra Meisner, 1820, Vipera<br />
aspis francisciredi Laurenti, 1768, Vipera aspis hugyi Schinz, 1833 and Vipera<br />
aspis zinnikeri Kramer, 1958. Amphibia-Reptilia 23 : 191-213.<br />
301
ANNEXE<br />
Fluctuating climatic conditions affect embryonic<br />
development in a viviparous snake (Vipera aspis)<br />
OLIVIER LOURDAIS 1,2 , RICHARD SHINE 3 , XAVIER BONNET 1, 3 , MICHAËL<br />
GUILLON 1 AND GUY NAULLEAU 1<br />
1 Centre d'Etudes Biologiques de Chizé, <strong>CNRS</strong>, 79360, Villiers en Bois, France<br />
2 Conseil Général des Deux Sèvres, rue de l‘abreuvoir, 79000 Niort, France<br />
3 School of Biological Sciences, University of Sydney, Australia<br />
Corresponding author: Olivier Lourdais,<br />
<strong>CEBC</strong>-<strong>CNRS</strong>, 79360, Villiers en Bois, FRANCE<br />
Tel: (33) 5 49 09 78 79<br />
Fax: (33) 5 49 09 65 26<br />
E-mail: lourdais@cebc.cnrs.fr<br />
For consideration in: Ecology<br />
Table: 3, Figures: 4<br />
302
Abstract. Climatic conditions during embryonic development can exert profound and<br />
long-term effects on many types of organisms, but most previous research on this<br />
topic has focussed on endothermic vertebrates (birds and mammals). Although<br />
viviparity in ectothermic taxa allows the reproducing female to buffer ambient thermal<br />
variation for her developing offspring, even an actively thermoregulating female may<br />
be unable to provide optimal incubation regimes in severe weather conditions. We<br />
examined the extent to which fluctuations in natural thermal conditions during<br />
pregnancy affect reproduction in a temperate viviparous snake, the aspic viper<br />
(Vipera aspis). Data gathered from a long term field study demonstrated that<br />
ambient thermal conditions influenced (1) female body temperatures and (2)<br />
gestation length, embryo viability, and offspring phenotypes. Interestingly, thermal<br />
conditions during each of the three months of gestation affected different aspects of<br />
reproduction. Hotter weather early in gestation (June) increased ventral scale counts<br />
(= number of body segments) of neonates; hotter weather mid-gestation (July)<br />
hastened development and thus the date of parturition; and hotter weather late in<br />
gestation (August) reduced the incidence of stillborn neonates. <strong>The</strong> population that<br />
we studied is close to the northern limit of the species’ range, and embryonic thermal<br />
requirements may prevent Vipera aspis from extending into cooler conditions further<br />
north.<br />
Key words: viviparity, snakes, ectothermy, temperature, development<br />
303
INTRODUCTION<br />
<strong>The</strong> environment affects living organisms in many ways. Influences that occur early<br />
in an organism’s life – especially, while it is an embryo – typically have greater effects<br />
on its subsequent development than do influences that occur later in life (Henry and<br />
Ulijaszek 1996, Desai and Hales 1997). Recent research has documented many<br />
strong and persistent effects of environmental factors acting during embryogenesis<br />
on fitness-related traits (Lindström 1999, Lummaa and Clutton-Brock 2002). Most<br />
studies on this topic have been based on birds and mammals, reflecting the<br />
concentration of long-term individual-based studies on these taxa (Lindström 1999).<br />
However, we also need data on other kinds of animals if we are to discern valid<br />
generalities about effects of the early environment on subsequent phenotypic traits.<br />
Ectothermic vertebrates are of particular interest in this respect. Because they<br />
contain both oviparous and viviparous taxa, often without the confounding influence<br />
of post-hatching parental care (Clutton-Brock 1991), they allow us to examine the<br />
degree to which alternative modes of reproduction buffer the developing offspring<br />
from environmental fluctuations.<br />
Ambient temperatures not only fluctuate considerably through time, but they<br />
also influence ectothermic vertebrates in many ways. <strong>The</strong> most obvious influences<br />
concern variables such as the metabolic rates, activity levels and locomotor<br />
performance of adult animals (Huey 1982, Hertz, Huey and Stevenson 1993).<br />
However, temperature also influences the rates and trajectories of ontogenetic<br />
development, and embryonic sensitivity to ambient temperature may be an important<br />
(albeit, less-studied) component of ectothermic biology. For example, extensive<br />
studies on oviparous reptiles show that the rate of embryonic development depends<br />
upon thermal regimes inside the nest. High temperatures greatly accelerate<br />
304
embryogenesis and thus, shorten the incubation period (Hubert 1985). <strong>The</strong>rmal<br />
regimes during embryogenesis can also profoundly affect phenotypic traits of the<br />
offspring, such as body size, scalation, and locomotor performance (Shine and<br />
Harlow 1996, Shine et al. 1997, Downes and Shine 1999, Andrews and Mathies<br />
2000, Flatt et al. 2001, Shine and Elphick 2001, Webb, Brown and Shine 2001). In<br />
some taxa, incubation temperatures directly determine the sex of offspring (Bull<br />
1980).<br />
<strong>The</strong> sensitivity of reptilian embryogenesis to ambient temperature suggests<br />
that thermal conditions may play a major role in the population ecology of these<br />
animals. For example, thermal minima for embryonic development may constrain<br />
oviparous species from reproducing in cold climates (Mell 1929, Weekes 1933, Tinkle<br />
and Gibbons 1977). More generally, geographic distributions of oviparous species<br />
may be set by the thermal requirements for embryogenesis (Shine 1987). Plausibly,<br />
annual variation in climatic conditions may influence the attributes of offspring (size,<br />
shape, time of hatching or birth) that enter the population each year, and thus<br />
influence the relative size of different year-classes. However, we are not aware of<br />
any data to show such an effect, apart from anecdotal reports of delayed oviposition,<br />
hatching or parturition in unusually cold years (e.g., Saint Girons 1952, Pengilley<br />
1972, Olsson and Shine 1997). In contrast, a wide range of studies on endothermic<br />
vertebrates (especially mammals) have not only documented effects of climatic<br />
variation on neonatal phenotypes, but also have shown that the resulting effects have<br />
long-term consequences for survival and reproductive success of individuals from<br />
those cohorts (e.g., Albon, Guinness and Clutton-Brock 1983, Post et al. 1997,<br />
Lummaa and Clutton-Brock 2002).<br />
<strong>The</strong> influence of ambient thermal fluctuations on hatching dates and offspring<br />
phenotypes is likely to be obvious in oviparous species of reptiles, because females<br />
305
in such taxa cannot control the thermal conditions experienced by their offspring<br />
throughout post-oviposition development (except in cases of maternal egg-brooding:<br />
e.g., Vinegar 1973). We might expect that this sensitivity to ambient thermal regimes<br />
would be much lower in a viviparous (live-bearing) species. Viviparity has evolved<br />
>100 times within squamate reptiles and this transition seems to have occurred<br />
primarily in cold climates (Shine 1985, Blackburn 1985, 1999). <strong>The</strong> probable<br />
selective force for these repeated transitions has been the egg-retaining female’s<br />
ability to maintain high, relatively constant incubation temperatures for her developing<br />
offspring (Shine 1985). Because the gravid female can regulate her temperature<br />
behaviorally, moving among microhabitats to exploit thermal heterogeneity in the<br />
environment, the temperatures experienced by an offspring developing in utero will<br />
be buffered considerably from fluctuations in ambient temperature (Shine 1983a,<br />
Burger and Zappalorti 1988, Charland and Gregory 1990, Schwarzkopf and Shine<br />
1991, Peterson, Gibson and Dorcas 1993).<br />
Nonetheless, if ambient thermal conditions fluctuate considerably, even a<br />
carefully-thermoregulating viviparous female may be unable to maintain high,<br />
constant temperatures for her developing offspring. In keeping with this inference,<br />
laboratory studies that have manipulated basking opportunities for viviparous lizards<br />
have found many of the same phenomena as described above for egg-layers. That<br />
is, a viviparous female reptile’s access to basking opportunities not only determines<br />
rates of embryogenesis (and thus, the duration of her pregnancy: Naulleau 1986,<br />
Schwarzkopf and Shine 1991) but also affects many phenotypic traits of her offspring<br />
(Shine and Harlow 1993, Swain and Jones 2000, Wapstra 2000, Arnold and<br />
Peterson 2002). Unfortunately, the relevance of these results to free-ranging animals<br />
remains unknown.<br />
306
Although many authors have stressed the ability of gravid females in<br />
viviparous reptile species to provide relatively high stable temperatures for their<br />
developing offspring (see above references), less attention has been paid to<br />
situations where they may be unable to do so. That is, do severely cold or rapidly<br />
fluctuating weather conditions make it impossible for even an actively<br />
thermoregulating viviparous female to provide an effective thermal buffer for her<br />
offspring? In such a situation, we might expect to see the developmental rates and<br />
phenotypic traits of offspring respond to annual variation in weather conditions,<br />
despite the thermoregulatory efforts of their mothers. Such effects should be<br />
especially important for individuals living at the altitudinal or latitudinal margin of the<br />
geographic range of a species. We examined this possibility with data from a nine-<br />
year study of a free-ranging population of viviparous snakes at the extreme northern<br />
limit of the species’ range.<br />
MATERIAL AND METHODS<br />
Study Animals<br />
<strong>The</strong> aspic viper, Vipera aspis Linné, is a small viviparous snake of the western-<br />
Paleartic region and is locally abundant at the northern limit of its distribution in<br />
France. Females mature at 40 cm snout-vent length (SVL), which is attained in 2.5<br />
to 3.5 years (Bonnet et al. 1999a). Ovulation typically occurs during the first two<br />
weeks of June with limited geographical variation (Saint 1957b, Naulleau 1981).<br />
During gestation pregnant females display higher thermal preferenda and<br />
substantially increase basking times (Saint Girons 1952, Naulleau 1979, Bonnet and<br />
Naulleau 1996, Lourdais et al. 2003b, Ladyman et al. 2003). Parturition occurs two<br />
to three months after ovulation, from late August through late September.<br />
307
Study site and methods<br />
<strong>The</strong> study site is near the village of Les Moutiers en Retz in west-central France<br />
(47 o 03N'; 02 o 00W'). It is a 33-hectare grove with a mosaic of meadows and<br />
regenerating scrubland. Details on the field site, methods and searching effort are<br />
available in related works (Bonnet et al. 2000, 2001,2002). Previous results<br />
suggest that climatic conditions in this area not only prolong gestation by one to two<br />
months compared to warmer-climate (Mediterranean) populations, but also that the<br />
magnitude of such effects varies among years (Lourdais et al. 2002a).<br />
Gravid females were captured and maintained in captivity after the first<br />
parturition of the year was recorded in the field (generally in late August).<br />
Reproductive data were then obtained on 173 litters from 149 different females. For<br />
most individuals (127) only a single litter was obtained, but 17 females produced two<br />
litters and 4 individuals produced three litters.<br />
<strong>The</strong> components of the litter were characterized (undeveloped ova, dead<br />
embryo, fully-developed but stillborn, healthy offspring), counted, and weighed<br />
(±0.1g). Young were measured (±0.5cm) and sexed. Stillborn offspring were<br />
measured, weighed, and sexed when possible. Because we could not distinguish<br />
unfertilized ova from ova that had been fertilized but had died early in<br />
embryogenesis, these were grouped in the same category (undeveloped ova).<br />
Using this method we gathered data on 817 healthy offspring, 132 undeveloped ova,<br />
22 dead embryos and 78 stillborn offspring. From 1993 to 2000, ventral scales were<br />
counted for 136 mothers and 681 healthy neonates. Gestation period was calculated<br />
from parturition dates, assuming a fixed ovulation date of 10 June (Saint Girons<br />
1980, Naulleau 1981).<br />
308
<strong>The</strong>rmal conditions<br />
In our study area, climatic conditions constrain many aspects of aspic viper ecology<br />
(Lourdais et al 2002a). <strong>The</strong>se snakes are diurnal with prolonged basking episodes.<br />
Previous studies have revealed substantial daily variation in body temperatures, with<br />
low overnight temperatures followed by basking to attain and maintain high body<br />
temperatures during daylight hours. Pregnant females bask much more than do non-<br />
pregnant animal, and thus are more often encountered and captured (Bonnet and<br />
Naulleau 1996, Lourdais et al. 2002b). <strong>The</strong>rmal conditions fluctuate strongly from<br />
one day to the next in this temperate-oceanic climate. Does such variation affect the<br />
body temperatures of vipers, despite the buffering effects of behavioral<br />
thermoregulation? To answer this question we need measures of both ambient<br />
temperatures and viper body temperatures:<br />
1) Ambient temperatures<br />
As an index of ambient temperature we used daily maximum shaded air temperature,<br />
as measured in a standard metereological shelter 1.8 m above the ground in Pornic<br />
(47 o 06N'; 02 o 07W'), near our field site (47 o 03N'; 02 o 00W'). <strong>The</strong>se daily maxima will<br />
not necessarily have any close <strong>relationship</strong> to the actual temperatures experienced<br />
by an embryonic viper inside its mother’s uterus, but instead should provide a rough<br />
index of a viper’s opportunity for behavioral thermoregulation. <strong>The</strong>rmal maxima were<br />
generally achieved in the afternoon, one to two hours after sun zenith.<br />
Aspic vipers have a long period of embryonic development (up to three<br />
months: Hubert and Dufaure 1968) and we distinguished three periods broadly<br />
corresponding to major steps in embryogenesis (Hubert 1985, Hubert, Dufaure and<br />
Collin 1966, Hubert and Dufaure 1968): (1) Early gestation (10 to 30 June), the onset<br />
of embryogenesis (including blastulation, gastrulation, neurulation, somite<br />
309
development, differentiation of the head and circulatory system); (2) Mid-gestation (1-<br />
31 July), a period of organogenesis (development of the optic vesicles and olfactory<br />
bulb, appearance of the jaws, cloacal split, genitalia and trunk/tail scales; rapid<br />
growth of the embryo and development of spiral-coiling); and (3) Late-gestation (1-<br />
31 August), a period of embryonic growth and completion of development (including<br />
development of pigmentation, and differentiation of head scales).<br />
In our analysis, we calculated mean daily temperature maxima during each of<br />
these three periods of gestation for each year. We investigated the relative influence<br />
of each period by regressing these mean temperatures against the duration of<br />
gestation (as estimated from dates of birth: see above). <strong>The</strong>n, we examined the<br />
effect of thermal conditions during development on offspring phenotype (scalation).<br />
Finally, we examined the possibility that embryo mortality rates might be influenced<br />
by thermal conditions during gestation.<br />
2) Body temperatures of free-ranging vipers<br />
Using internal temperature radiotransmitters (See Naulleau, Bonnet and Duret 1996<br />
for the method), we monitored reproductive and non-reproductive female vipers<br />
during the gestation period (1 June to 31 August 1996). Females were sampled one<br />
to four times per day from 0800 to 2130 h, but to minimise temporal heterogeneity<br />
and pseudoreplication our analyses were based only on a single late-afternoon (1700<br />
– 20000 h) data point per female per day. Because of the 24-hour delay <strong>between</strong><br />
successive readings (during which time the snakes’ body temperatures dropped to<br />
minimum levels overnight: Naulleau 1997, pers obs), we have treated successive<br />
daily temperatures from the same individual as quasi-independent. Using this<br />
procedure 241 temperature records from 16 female Aspic vipers (9 reproductive and<br />
7 non-reproductive) were available (average number of records per individual<br />
310
=16±7). <strong>The</strong> influence of daily maxima on female body temperature was examined<br />
after accounting for female identity and reproductive status.<br />
Statistics<br />
All statistics were performed with Statistica 6.0. Influences of ambient thermal<br />
conditions on female body temperature were examined using general linear<br />
regression modeling (GLM). Reproductive status was considered as a fixed factor.<br />
Female identity was treated as a random factor, nested within reproductive status.<br />
Daily temperature maxima or sampling dates (Julian calendar) were then treated as<br />
covariates (fixed effects). <strong>The</strong> influences of thermal conditions on offspring<br />
phenotype were investigated using mixed-model ANCOVAS. To account for<br />
correlated responses among offspring of individual litters and repeated female<br />
contributions (17 females reproduced twice and 4 three times), maternal identity was<br />
included as a random factor. Neonatal traits were the dependent variable, offspring<br />
sex was a fixed factor and annual thermal conditions were treated as covariates<br />
(fixed effects).<br />
RESULTS<br />
I) Determinants of female body temperatures<br />
GLM analysis suggested that female body temperature was affected by at least three<br />
factors (see table 1). First, our use of air temperature as an index of<br />
thermoregulatory opportunities was validated by a significant influence of maximum<br />
air temperature on female body temperature (table 1). Second, a significant<br />
<strong>relationship</strong> <strong>between</strong> sampling date and daily thermal maxima reflected an increase<br />
in ambient temperatures over the summer period. Third, reproductive status also<br />
exerted a strong influence on female body temperature with pregnant females<br />
311
maintaining body temperatures than did non-reproductive females (28.7±0.5°C, n<br />
=133 versus 25.47±0.6°C, n =117; see table 1).<br />
II) Fluctuations in thermal conditions<br />
A repeated measure ANOVA (using year as the factor and gestation month within<br />
each year as the repeated factor) first indicated that mean daily air temperature<br />
significantly increased over the course of gestation (F(2, 522)=45.33; p < 0.00001)<br />
and more importantly, that mean air temperature during gestation varied significantly<br />
among years (F(8, 261)=3.33; p < 0.0012). We also detected a significant interaction<br />
<strong>between</strong> gestation month and year (F(16, 738) = 4.53; p < 0.0001), reflecting<br />
marked year to year fluctuations in thermal conditions over the three months of<br />
gestation (Fig. 1). During the study period, thermal conditions during gestation fell<br />
broadly into three patterns: parabolic (where the highest temperature was in July),<br />
sigmoid (where the predominant change was a major increase in temperature<br />
<strong>between</strong> June and July), and exponential (where the predominant change was a<br />
major temperature increase <strong>between</strong> July and August).<br />
As a consequence, we found no significant correlation <strong>between</strong> monthly mean<br />
daily temperatures across years (F(1, 7) = 1.52, n = 9, p = 0.25 for June versus July;<br />
F(1, 7) = 3.11, n = 9, p = 0.12 for June versus August; F(1, 7) = 0.01, n = 9, p = 0.93<br />
for July versus August). <strong>The</strong>refore, we consider mean temperatures during each of<br />
the three months as independent variables for our subsequent analyses.<br />
1) Duration of gestation period<br />
III) Impact on reproduction<br />
In a related study (Lourdais et al. 2002a), we showed that the duration of gestation<br />
in this population was influenced by mean temperatures during the gestation period,<br />
312
as well as by the frequencies of inviable elements in litters (i.e., undeveloped ova and<br />
stillborn offspring). For the present study, we can examine this result more closely in<br />
terms of the three phases of gestation defined above. We used stepwise multiple<br />
regression analysis for this purpose, and restricted the analysis to the 80 females that<br />
produced only viable offspring (i.e., no stillborns). Only mean daily temperatures<br />
during July (mid-gestation) were retained in the model (see Table 1), accounting for<br />
51.4% of the variance in overall gestation length.<br />
2) Offspring scalation<br />
First, we detected a strong influence of maternal identity on the number of ventral<br />
scales in newborn vipers (ANOVA, F(101, 527) = 4.13, p=0.00001). This influence<br />
was partially attributable to significant heritability of ventral scalation, as we detected<br />
a significant <strong>relationship</strong> <strong>between</strong> maternal and new-born number of ventral scales (r<br />
2 =0.11; F(1, 629) = 80.81; p < 0.0001 treating each offspring as an individual point<br />
and r 2 =0.22; F(1, 114) = 34.64; p < 0.0001 when considering mean offspring number<br />
of ventral scales per litter). <strong>The</strong> two sexes differed in mean numbers of ventral<br />
scales, with neonatal females having more scales than their brothers (149.1±0.17,<br />
n=329 versus 148.0±0.17, n=352, F(1, 563) = 11.37; p < 0.0007, mixed model<br />
ANOVA using female identity as random factor and offspring sex as a fixed factor).<br />
Interestingly, we detected significant year to year variation in the number of ventral<br />
scales in neonatal snakes (ANOVA, F(7, 673)=12.400, p=0.00001, see figure 2). This<br />
effect holds true even after accounting for maternal influence and offspring sex (F(7,<br />
114)=3.08, 0.0004, mixed model ANOVA year using female identity as a random<br />
factor, offspring sex and year as a fixed factors). Such annual variations appear to<br />
be linked to the climatic fluctuations described above. For instance we detected a<br />
significant influence of mean gestation thermal maxima on neonate number of ventral<br />
313
scales (F(1, 114)=6.94, p=0.008, mixed model ANCOVA using female identity as<br />
random factor, offspring sex as a fixed factor and mean gestation thermal maxima as<br />
a fixed covariate). Because organizational effects of temperature are likely to occur<br />
early in embryogenesis, we re-conducted the analysis by considering each of the<br />
three components of the gestation period independently. Only mean daily<br />
temperature maxima during the first period (i.e., the three weeks following ovulation)<br />
led to significant results (Table 3, Fig. 3). This influence is reflected in a significant<br />
<strong>relationship</strong> <strong>between</strong> mean thermal maxima in June and the mean number of ventral<br />
scales in neonatal vipers (r 2 =0.49; F(1,6)=7.75 p
(χ 2 = 0.24; dl = 1; p = 0.62, pooling the nine years of the study), nor did sex ratios of<br />
stillborn offspring vary significantly among years (χ 2 =10.27; dl = 8; p = 0.24).<br />
Based on these results, we looked for a possible influence of mean<br />
temperature during the latter part of gestation (July and August) on the probability of<br />
producing stillborn offspring. Excluding females producing undeveloped ova, we<br />
detected a significant negative influence of mean August temperature on the<br />
probability of observing late embryonic death (Logistic regression, χ 2 = 8.18; n =<br />
113; p = 0.0042). We also detected a significant negative <strong>relationship</strong> <strong>between</strong> mean<br />
August temperatures and the proportion of stillborn offspring (r = 0.33; r 2 = 0.11; n =<br />
113; F(1, 111) = 13.37; p < 0.0004). <strong>The</strong> same analysis using mean July or mean<br />
June daily temperatures yielded non significant results.<br />
DISCUSSION<br />
Our relatively long-term field study demonstrates that natural climatic conditions<br />
influence important aspects of embryogenesis in the aspic viper. Although gravid<br />
vipers show distinctive thermoregulatory behaviors that result in relatively high, stable<br />
body temperatures throughout pregnancy (Saint Girons 1952, Naulleau 1979, Bonnet<br />
and Naulleau 1996, Ladyman et al. 2003), they are unable to completely buffer their<br />
developing embryos from year-to-year thermal variations in this relatively northern,<br />
cool-climate area. This result runs counter to the primary emphasis of published<br />
studies on thermoregulation by gravid reptiles, which have stressed the<br />
thermoregulatory precision of such animals (e.g., Shine 1983a, Charland and<br />
Gregory 1990, Schwarzkopf and Shine 1991, Peterson, Gibson and Dorcas 1993).<br />
Clearly, this stenothermy is relative: even if gravid females maintain higher<br />
temperatures than do non-reproductive conspecifics, they may still vary enough in<br />
body temperatures to impact on the embryos.<br />
315
Unsurprisingly, high summer temperatures resulted in faster embryonic<br />
development and thus, earlier parturition dates. <strong>The</strong>rmal dependence of embryonic<br />
development is widespread in squamate reptiles (Blanchard and Blanchard 1941,<br />
Hubert 1985) and our results in this respect are consistent with an earlier<br />
experimental study conducted by Naulleau (1986) on this species and the closely<br />
related adder (Vipera berus). However, our data extend previous knowledge of this<br />
phenomenon in suggesting that the thermal sensitivity of gestation length seems to<br />
be significant mostly (or only) in the middle part of gestation. In our multiple-<br />
regression analyses, only July mean temperature significantly accelerated gestation.<br />
This period coincides with major steps in embryogenesis (i.e., organogenesis and<br />
active tissue synthesis). Recent studies on oviparous lizards reported that rates of<br />
embryogenesis were most sensitive to incubation temperature relatively soon after<br />
oviposition (Shine and Elphick 2001, Shine 2002). Given that the lizards involved in<br />
that study lay their eggs when embryos are almost one-third through their <strong>total</strong><br />
developmental period (Shine 1983b), this result fits well with those from our own<br />
study. <strong>The</strong> <strong>relationship</strong> <strong>between</strong> the thermosensitivity of gestation length and specific<br />
embryonic stages warrants further study, to test this apparent generality.<br />
Environmental conditions also had direct effects on the phenotypic traits of<br />
offspring. In keeping with laboratory studies that have manipulated basking<br />
opportunities for viviparous female reptiles and documented shifts in offspring<br />
phenotype as a result (Shine & Harlow 1993, Shine & Downes 1999, Swain & Jones<br />
2000, Wapstra 2000, Arnold and Peterson 2002), we found significant correlations<br />
<strong>between</strong> ambient temperatures and phenotypic traits of the neonatal vipers. In<br />
contrast with the experimental study of Arnold & Peterson (2002) that documented a<br />
flat reaction norm for scale count in another viviparous species (Thamnophis<br />
elegans), we detected a significant impact of temperature on ventral scalation in new<br />
316
orn aspic vipers. Notably, mean temperature during early stages in embryogenesis<br />
affected the number of ventral scales in offspring, with higher temperatures<br />
increasing scale numbers. Similar influences of developmental temperature on<br />
scalation have been reported from laboratory experiments on reptiles (Vinegar<br />
1973,1974; Osypra & Arnold 2000), and inferred from climate-correlated geographic<br />
shifts in scale counts (Klauber 1941). In most snakes, the number of ventral scales is<br />
tightly correlated with the number of vertebrae, reflecting the number of pairs of<br />
somites differentiated during early embryogenesis (Hubert 1985; Lindell 1996). <strong>The</strong><br />
number of ventral scales or body vertebrae shows considerable intraspecific variation<br />
(Lindell 1996, Lindell, Forsman and Merilä 1993). If temperature influences<br />
somitisation, we would expect that differences in scalation (reflecting vertebral<br />
number) will be correlated with differences in body length. <strong>The</strong> proximal <strong>relationship</strong><br />
<strong>between</strong> those two traits was supported in our data set by a significant <strong>relationship</strong><br />
<strong>between</strong> ventral scalation and offspring snout-vent length (r=0.20; F(1,679)=27.978;<br />
p
performance (Arnold 1988, Arnold and Bennett 1988). Hence, the influence of<br />
natural climatic conditions on embryo development may affect the quality of offspring<br />
produced by a female viper.<br />
Finally, weather conditions affected neonatal fitness directly by influencing<br />
rates of embryo mortality. Years with cool weather late in summer, close to the end<br />
of gestation (August) resulted in a high incidence of stillborn offspring. This result<br />
suggests that embryos may be particularly sensitive to low temperatures late in<br />
development, a pattern previously reported in a field study on an oviparous species<br />
(Burger and Zappalorti 1988). As our study population is close to the species’<br />
northern range limit (Stewart 1971), the sensitivity of offspring development to<br />
ambient weather conditions may be a direct result of climatic constraints on female<br />
thermoregulation. In more favorable environments (e.g., southern populations),<br />
viviparity may well allow female aspic vipers to selectively alter incubation conditions<br />
and thus provide optimal incubation regimes for their offspring.<br />
<strong>The</strong> ecological and evolutionary significance of such interactions among the<br />
environment, female thermoregulatory behavior, and offspring phenotype is a<br />
complex issue that requires further work, notably comparative studies of populations<br />
facing different climatic conditions. <strong>The</strong> high levels of embryo mortality detected in<br />
our study probably reflect the location of Les Moutiers at the northern limit of the<br />
geographic range of the species. In this area, female aspic vipers experience high<br />
survival costs of reproduction and most females reproduce only once in their lifetime<br />
(Bonnet et al. 1999b, 2002). Further north the aspic viper is replaced by a sister<br />
species, the adder (Vipera berus), with limited overlap in the distribution of the two<br />
species (Saint Girons 1975). While similar in size and appearance, these two vipers<br />
diverge in metabolic rates and in thermal requirements for digestion and gestation.<br />
Both are lower in V. berus than in V. aspis, allowing the former species to penetrate<br />
318
into cooler, more northern areas (Naulleau 1983, 1986 Saint Girons, Naulleau and<br />
Célérier 1985). Our analyses suggest that the thermal optima for embryonic<br />
development could also constrain the geographic distribution of V. aspis (as<br />
suggested for oviparous squamates by Shine 1987 and Shine, Barrot and Elphick<br />
2003).<br />
In conclusion, we found that in this northern population of snakes, natural<br />
thermal conditions significantly affected embryonic development despite active<br />
maternal thermoregulation. Our results underline the importance and complexity of<br />
ambient thermal influences on the lives of ectothermic vertebrates. Comparative<br />
studies with southern populations facing a less constraining environment would be of<br />
great interest. In addition, experimental examination of female thermoregulatory<br />
behavior during particular thermosensitive phases (such as early embryogenesis) are<br />
needed to clarify to the extent to which viviparity permits active maternal manipulation<br />
of offspring phenotypes.<br />
Acknowledgements<br />
We thank Dale DeNardo for helpful comments on the manuscript. Financial support<br />
was provided by the Conseil Général des Deux Sèvres, le Centre National de la<br />
Recherche Scientifique (France).<br />
Literature Cited<br />
Albon, S.D., Guinness, F.E. and Clutton-Brock, T.H. 1983. <strong>The</strong> influence of climatic<br />
variation on the birth weights of red deer (Cervus elephas). Journal of Zoology<br />
(London), 200: 295-297.<br />
Andrews, R.M. and Mathies, T. 2000. Natural history of reptilian development:<br />
constraints on the evolution of viviparity. Bioscience, 50: 227-238.<br />
Arnold, S.J. 1988. Quantitative genetics and selection in natural populations:<br />
microevolution of vertebral numbers in the garter snake Thamnophis elegans.<br />
Proceedings of the Second International Conference on Quantitative Genetics.<br />
319
(Eds Weir, B.S., Goodman, M.M., Eisen, E.J. and Namkong, G.) pp. 619-636.<br />
(Sinauer Assoc.: Sunderland, Mass.)<br />
Arnold, S.J. and Bennett, A.F. 1988. Behavioural variation in natural populations. V.<br />
Morphological correlates of locomotion in the garter snake Thamnophis radix.<br />
Biological Journal of the Linnean Society, 34: 175-190.<br />
Arnold, S.J. and Peterson, C.R. 2002. A model for optimal reaction norm : the case of<br />
the pregnant garter snake and her temperature-sensitive embryos. <strong>The</strong><br />
American Naturalist 160: 307-316.<br />
Blanchard, F.N. and Blanchard, F.C. 1941. Factors determining time of birth in<br />
garter snake. Pap. Michigan Acad. Sci., 26: 161-176.<br />
Blackburn, D.G. 1985. Evolutionary origins of viviparity in the Reptilia. II. Serpentes,<br />
Amphisbaenia, and Icthyosauria. Amphibia-Reptilia, 6: 259-291.<br />
Blackburn, D.G. 1999. Are viviparity and egg-guarding evolutionarily labile in<br />
squamates? Herpetologica, 55: 556-573.<br />
Bonnet, X. and Naulleau, G. 1996. Catchability in snakes : consequences on<br />
breeding frequency estimates. Canadian Journal of Zoology 74 : 233-239.<br />
Bonnet, X., Naulleau, G., Lourdais, O. and Vacher-Vallas, M. 1999a. Growth in the<br />
asp viper (Vipera aspis L.) : insights from long term field study. In : Miaud C and<br />
Guyetant R (eds) Current Studies in Herpetology, pp 63-69. SEH, Le Bourget<br />
du Lac.<br />
Bonnet, X., Naulleau, G., Shine, R. and Lourdais, O. 1999b. What is the appropriate<br />
time scale for measuring costs of reproduction in a capital breeder?<br />
Evolutionary Ecology 13 : 485-497.<br />
Bonnet, X., Naulleau, G., Shine, R. and Lourdais, O. 2000. Reproductive versus<br />
ecological advantages to larger body size in female Vipera aspis. Oikos 89 :<br />
509-518<br />
Bonnet, X., Naulleau, G., Shine, R. and Lourdais, O. 2001. Short-term versus long-<br />
term effects of <strong>food</strong> <strong>intake</strong> on reproductive output in a viviparous snake, Vipera<br />
aspis. Oikos 92: 297-308.<br />
Bonnet, X., Lourdais, O., Shine, R. and Naulleau, G., 2002. Reproduction in snakes<br />
(Vipera aspis): costs, currencies and complications. Ecology 83, 2124-2135.<br />
Bull, J. J. 1980. Sex determination in reptiles. Quarterly Review in Biology, 55: 4-<br />
21.<br />
320
Burger, J. and Zappalorti R.T. 1988. Effects of incubation temperature on sex ratios<br />
in pine snakes: differential vulnerability of males and females. <strong>The</strong> American<br />
Naturalist, 132: 492-505.<br />
Charland, M.B. and Gregory, P.T. 1990. <strong>The</strong> influence of female reproductive<br />
status on thermoregulation in a viviparous snake, Crotalus viridis. Copeia,<br />
1990: 1089-1098.<br />
Clutton-Brock, T.H. 1991. <strong>The</strong> evolution of parental care. Princeton University Press,<br />
Princeton, New Jersey.<br />
Desai, M. and Hales, C.N. 1997. Role of fetal and infant growth in programming<br />
metabolism in later life. Biological Reviews, 72: 329-342.<br />
Downes, S.J. and Shine, R. 1999. Do incubation-induced changes in a lizard’s<br />
phenotype influence its vulnerability to predators? Oecologia, 120: 9-18.<br />
Flatt, T., Shine, R., Borges-Landaez, P.A. and Downes, S.J. 2001. Phenotypic<br />
variation in an oviparous montane lizard (Bassiana duperreyi): the effects of<br />
thermal and hydric incubation environments. Biological Journal of the Linnean<br />
Society, 74: 339-350.<br />
Forsman, A. 1993. Survival in relation to body size and growth rate in the adder,<br />
Vipera berus. Journal of Animal Ecology, 62: 647-655.<br />
Forsman, A. and Lindell, L. E. 1993. <strong>The</strong> advantage of a big head: swallowing<br />
performance in adders, Vipera berus. Functional Ecology, 7: 183-189.<br />
Forsman, A. and Shine, R. 1997. Rejection of non-adaptive hypotheses for<br />
intraspecific variation in trophic morphology in gape-limited predators.<br />
Biological Journal of the Linnean Society, 62: 209– 223.<br />
Henry, C.J.K. and Ulijaszek, S.J. 1996. Long-term consequences of early<br />
environment. Cambridge Univ. Press, Cambridge.<br />
Hertz, P.E., Huey, R.B., and Stevenson, R.D. 1993. Evaluating temperature<br />
regulation by field-active ectotherms: the fallacy of the inappropriate question.<br />
American Naturalist, 142: 796-818.<br />
Hubert. J. 1985. Embryology of the squamata. Biology of the Reptilia 15. (Eds Gans,<br />
C. and Billet, F) pp. 1-55. (Wiley and sons: New York)<br />
Hubert, J. and Dufaure, J. P. 1968. Table de développement de la vipère aspic:<br />
Vipera aspis L. Bulletin de la Société Zoologique de France, 93: 135-148.<br />
Hubert, J., Dufaure, J.P. and Collin, J.P. 1966. Matériaux pour une table de<br />
321
développement de Vipera aspis L., I. La période de l’Organogénèse. Bulletin<br />
de la Société Zoologique de France, 91: 779-788.<br />
Huey, R.B. 1982. Temperature, physiology and the ecology of reptiles. Biology of the<br />
Reptilia 12. (Eds Gans, C. and Pough, F.H.) pp. 25-91. (Academic Press:<br />
London.)<br />
Klauber, L.M. 1941. <strong>The</strong> correlation <strong>between</strong> scalation and life zones in San Diego<br />
County snakes. Bull. Zool. Soc. San Diego, 17: 73-79.<br />
Lindell, L. E. 1996. Vertebral number in adders, Vipera berus: direct and indirect<br />
effects on growth. Biological journal of the Linnean society, 59: 69-85.<br />
Lindell, L. E., Forsman, A. and Merilä, J. 1993. Variations in number of ventral<br />
scales in snakes: effects on body size, growth rate and survival in the adder,<br />
Vipera berus. Journal of Zoology, London, 230: 101-115.<br />
Lindström, J. 1999. Early development and fitness in birds and mammals. Trends in<br />
Ecology and Evolution, 14: 343-348.<br />
Lourdais, O., Bonnet, X. and Doughty, P. 2002. Costs of anorexia during pregnancy<br />
in a viviparous snake (Vipera aspis). Journal of Experimental Zoology, 292:<br />
487-493.<br />
Lourdais, O., Bonnet, X., Shine, R., DeNardo, D., Naulleau, G. and Guillon, M.<br />
2002a. Capital-breeding and reproductive effort in a variable environment: a<br />
longitudinal study of a viviparous snake, Journal of Animal Ecology, 71: 470-<br />
479.<br />
Lourdais, O., Bonnet, X., DeNardo, D. and Naulleau, G., 2002b. Do sex divergences<br />
in reproductive ecophysiology translate into dimorphic demographic patterns?<br />
Population Ecology 44 3: 241-249.<br />
Lummaa, V. and Clutton-Brock, T. 2002. Early development, survival and<br />
reproduction in humans. Trends in Ecology and Evolution, 17: 141-147.<br />
Mell, R. 1929. Beitrage zur Fauna Sinica. IV. Grundzuge einer Okologie der<br />
chinesischen Reptilien und einer herpetologischen Tiergeographie Chinas.<br />
(Walter de Gruyter: Berlin.)<br />
Mushinsky, H. R. 1987. Foraging ecology. Snakes – Ecology and Evolutionary<br />
Biology (eds. Seigel, R. A., Collins, J. T., and Novak S. S.). pp. 302-334.<br />
Mazmilan Publishing Company, New York.<br />
Naulleau, G. 1971. Fertility of female Vipera aspis in captivity as a function of the<br />
periods of mating. Herpetologica, 27: 385-389.<br />
322
Naulleau, G. 1979. Etude biotélémétrique de la thermoregulation chez Vipera aspis<br />
(L.) Elevée en conditions artificielle. Journal on Herpetology 13(2): 203-208.<br />
Naulleau, G. 1981. Détermination des périodes d’ovulation chez Vipera aspis et<br />
Vipera berus dans l’ouest de la France, étudiée par radiographie. Bulletin de la<br />
Société des Sciences Naturelles de l’Ouest de la France, 3: 151-153.<br />
Naulleau, G. 1983. Action de la température sur la digestion chez cinq espèces de<br />
vipères européennes du genre Vipera. Bulletin de La Société Zoologique de<br />
France, 108: 47-58.<br />
Naulleau, G. 1986. Effect of temperature on gestation in Vipera aspic and V. berus<br />
(Reptilia: Serpentes). Studies in herpetology, Roček Z. Prague: 489-494.<br />
Naulleau, G. 1997. La vipère aspic. Eveil Nature, Saint Yrieix.<br />
Olsson, M. and Shine, R. 1997. <strong>The</strong> seasonal timing of oviposition in sand lizards<br />
(Lacerta agilis): why earlier clutches are better. Journal of Evolutionary Biology,<br />
10: 369-381.<br />
Osypka, N.M. and Arnold, S.J. 2000. <strong>The</strong> developmental effects of sex ratio on a<br />
sexually dimorphic scale count in the garter snake Thamnophis elegans.<br />
Journal of Herpetology, 34: 1-5.<br />
Post, E., Stenseth, N.C., Langvaten, R. and Fromentin, J. 1997. Global climate<br />
change and phenotypic variation among red deer cohorts. Proceedings of the<br />
Royal Society of London, Series B, 264: 1317-1324.<br />
Pengilley, R. 1972. Systematic <strong>relationship</strong>s and ecology of some lygosomine lizards<br />
from southeastern Australia. Ph.D. Dissertation. Unpubl. <strong>The</strong>sis. Australian<br />
National University.<br />
Peterson, C.R., Gibson, A.R. and Dorcas, M.E. 1993. Snake thermal ecology: <strong>The</strong><br />
causes and consequences of body-temperature variation. Snakes: Ecology and<br />
Behavior. (Eds Seigel, R.A. and Collins, J.T.) pp. 241-314. (McGraw-Hill: New<br />
York.)<br />
Saint Girons, H. 1952. Ecologie et éthologie des vipères de France. Annales des<br />
Sciences Naturelles de Zoologie, 14: 263-343.<br />
Saint Girons, H. 1957. Le cycle sexuelle chez Vipera aspis (L) dans l’ouest de la<br />
France. Bulletin Biologique de France et de Belgique, 91: 284-350.<br />
Saint Girons, H. 1975. Coexistence de Vipera aspis et de Vipera berus en Loire<br />
Atlantique: Un problème de compétition interspécifique. Revue d’Ecologie<br />
(Terre et Vie), 29: 590-613.<br />
323
Saint Girons, H. 1980. Biogéographie et évolution des vipères européennes.<br />
Compte Rendu des Scéances de la Société de biogéographie, 496: 146-172.<br />
Saint Girons, H., Naulleau, G. and Célérier, M.L. 1985. Le métabolisme aérobie de<br />
Vipera aspis et de la Vipera berus (Reptilia, Viperidae) acclimatées à des<br />
conditions naturelles. Influence de la température et comparaison des deux<br />
espèces. Amphibia-Reptilia, 6: 71-81.<br />
Schwarzkopf, L. and Shine, R. 1991. <strong>The</strong>rmal biology of reproduction in viviparous<br />
skinks, Eulamprus tympanum: Why do gravid females bask more? Oecologia,<br />
88: 562-569.<br />
Shine, R. 1983a. Reptilian viviparity in cold climates: testing the assumptions of an<br />
evolutionary hypothesis. Oecologia, 57: 397-405.<br />
Shine, R. 1983b. Reptilian reproductive modes: the oviparity-viviparity continuum.<br />
Herpetologica, 39: 1-8.<br />
Shine, R. 1985. <strong>The</strong> evolution of viviparity in reptiles: an ecological analysis. Biology<br />
of the Reptilia, Volume 15. (Eds Gans, C. and Billett, F.) pp. 605-694. (John<br />
Wiley and Sons: New York.)<br />
Shine, R. 1987. Reproductive mode may determine geographic distributions in<br />
Australian venomous snakes (Pseudechis, Elapidae). Oecologia, 71: 608-612.<br />
Shine, R. 2002. Reconstructing an adaptationist scenario: what selective forces<br />
favor the evolution of viviparity in montane reptiles? American Naturalist 160:<br />
582-593.<br />
Shine, R. and Downes, S.J. 1999. Can pregnant lizards adjust their offspring<br />
phenotypes to environmental conditions? Oecologia, 119: 1-8.<br />
Shine, R. and Elphick, M.J. 2001. <strong>The</strong> effect of short-term weather fluctuations on<br />
temperatures inside lizard nests, and upon the phenotypic traits of hatchling<br />
lizard. Biological Journal of the Linnean Society, 72: 555-565.<br />
Shine, R. and Harlow, P. 1993. Maternal thermoregulation influences offspring<br />
viability in a viviparous lizard. Oecologia, 96: 122-127.<br />
Shine, R. and Harlow, P. 1996. Maternal manipulation of offspring phenotypes via<br />
nest-site selection in an oviparous lizard. Ecology, 77: 1808-1817.<br />
Shine, R., Barrot, E.G. and Elphick, M.J. 2003. Some like it hot: effects of forest<br />
clearing on nest temperatures of montane reptiles. Ecology. In press<br />
Shine, R., Madsen, T.R.L., Elphick, M. J. and Harlow, P.H. 1997. <strong>The</strong> influence of<br />
nest temperatures and maternal brooding on hatchling phenotypes in water<br />
324
pythons. Ecology, 78: 1713-1721.<br />
Slip, D.J. and Shine, R. 1988. Reptilian endothermy: a field study of<br />
thermoregulation by brooding diamond pythons. Journal of Zoology (London)<br />
216: 367-378.<br />
Stewart, J.W. 1971. <strong>The</strong> snakes of Europe. (Fairleigh Dickinson Univ. Press:<br />
Rutherford).<br />
Swain, R. and Jones, S.M. 2000. Maternal effects associated with gestation<br />
conditions in a viviparous lizard, Niveoscincus metallicus. Herpetological<br />
Monographs, 14: 432-440.<br />
Tinkle, D. W. and Gibbons, J.W. 1977. <strong>The</strong> distribution and evolution of viviparity in<br />
reptiles, Misc. Pub.Mus. Nat. Hist. Univ. Mich. 154:1-55.<br />
Vinegar, A. 1973. <strong>The</strong> effects of temperature on the growth and development of<br />
embryos of the Indian python, Python morulus (Reptilia: Serpentes: Boidae).<br />
Copeia, 1973: 171-173.<br />
Vinegar, A. 1974. Evolutionary implications of temperature induced anomalies of<br />
development in snake embryos. Herpetologica, 30: 72-74.<br />
Wapstra, E. 2000. Maternal basking opportunity affects juvenile phenotype in a<br />
viviaprous lizard. Functional Ecology, 14: 345-352.<br />
Webb, J.K., Brown, G.P. and Shine, R. 2001. Body size, locomotor speed and<br />
antipredator behaviour in a tropical snake (Tropidonophis mairii, Colubridae):<br />
the influence of incubation environments and genetics factors. Functional<br />
Ecology, 15: 561-568.<br />
Zuffi, M.A.L., Giudici, F. and Iolae, P. 1999. Frequency and reproductive effort of<br />
reproduction in female Vipera aspis from a southern population. Acta<br />
Oecologica, 20: 633-638.<br />
325
Table 1. Determinants of female body temperature. Reproductive status<br />
(STATUS) was considered as a fixed factor. Females identity (IDENTITY) was<br />
considered as a random factor and was nested in the corresponding reproductive<br />
status. Daily temperature maxima (MAX) and Julian sampling date (DATE) were then<br />
treated as covariates (fixed effect).<br />
Effect dl MS F p value<br />
STATUS Fixed 1 831.82 29.36 0.0001<br />
DATE Fixed 1 307.55 10.85 0.001<br />
MAX Fixed 1 403.24 14.23 0.0002<br />
IDENTITY Random 14 169.54 5.98 0.0001<br />
Error 234<br />
Table 2. Influence of mean temperatures during the three months of pregnancy on<br />
the duration of gestation in female aspic vipers.<br />
Multiple Regression r = 0.71; r² = 0.51; n = 80; F(3, 76) = 26.86; p < 0.0001<br />
Bêta Partial correlation p value<br />
June 0.23 0.18 0.11<br />
July -0.77 0.62
Captions to figures<br />
Figure 1. Annual variation in thermal conditions during the three months of<br />
gestation. For simplicity, years were classified depending upon the thermal pattern<br />
observed. (Jun:June; Jul:July; Aug: August)<br />
Pattern 1 (parabolic): 1992 (open triangles down), 1994 (open circles), 1996 (open<br />
squares), 1999 (open diamonds).<br />
Pattern 2 (sigmoid): 1995 (open triangles up), 1997 (open triangles down).<br />
Pattern 3 (exponential): 1993 (open squares), 1998 (open hexagons), 2000 (open<br />
circles).<br />
Figure 2. Annual variation in number of ventral scales in the offspring of aspic vipers<br />
(± S.E.).<br />
Figure 3. Relationship <strong>between</strong> mean June maxima and mean offspring number of<br />
ventral scales over the course of the study.<br />
Figure 4 Influence of mean June maxima (°C) on offspring number of ventral scales<br />
(Offspring NVS) and body size (Snout vent length, cm).<br />
327
Mean Temperatures<br />
28<br />
27<br />
26<br />
25<br />
24<br />
23<br />
22<br />
21<br />
20<br />
Figure 1<br />
1 2 3<br />
Jun Jul Aug Jun Jul Aug Jun Jul Aug<br />
328
Number of Ventral scales<br />
151<br />
150<br />
149<br />
148<br />
147<br />
146<br />
145<br />
Figure 2<br />
1993 1994 1995 1996 1997 1998 1999 2000<br />
Year<br />
329
Mean number of ventral scales<br />
Figure 3<br />
150<br />
149<br />
148<br />
147<br />
146<br />
145<br />
21.0 21.5 22.0 22.5 23.0 23.5 24.0<br />
Mean June maxima<br />
330
Figure 4<br />
331