Ontogeny of cannibalism in larval and juvenile fishes with special ...
Ontogeny of cannibalism in larval and juvenile fishes with special ...
Ontogeny of cannibalism in larval and juvenile fishes with special ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Chapter n<strong>in</strong>e<br />
<strong>Ontogeny</strong> <strong>of</strong> <strong>cannibalism</strong><br />
<strong>in</strong> <strong>larval</strong> <strong>and</strong> <strong>juvenile</strong><br />
<strong>fishes</strong> <strong>with</strong> <strong>special</strong><br />
emphasis on Atlantic cod<br />
Arild Folkvord<br />
9 .1 INTRODUCTION<br />
Cannibalism can be def<strong>in</strong>ed as : "the act <strong>of</strong> kill<strong>in</strong>g <strong>and</strong> consum<strong>in</strong>g the whole,<br />
or major part, <strong>of</strong> an <strong>in</strong>dividual belong<strong>in</strong>g to the same species, irrespective <strong>of</strong><br />
its stage <strong>of</strong> development" (Smith <strong>and</strong> Reay, 1991) . It has been documented<br />
<strong>in</strong> a wide range <strong>of</strong> taxa, <strong>in</strong>clud<strong>in</strong>g Pisces (Smith <strong>and</strong> Reay, 1991 ; Elgar <strong>and</strong><br />
Crespi, 1992) . Cannibalism is encountered among most <strong>of</strong> the well-studied<br />
teleost families, <strong>and</strong> is classified accord<strong>in</strong>g to the developmental stage <strong>of</strong><br />
prey, genetic relationship <strong>of</strong> cannibal to prey, <strong>and</strong>/or the age relationship <strong>of</strong><br />
cannibal <strong>and</strong> prey (Smith <strong>and</strong> Reay, 1991) . In this chapter <strong>in</strong>tracohort<br />
<strong>cannibalism</strong> is def<strong>in</strong>ed as <strong>cannibalism</strong> <strong>in</strong>volv<strong>in</strong>g members <strong>of</strong> the same year<br />
class, <strong>and</strong> <strong>in</strong>tercohort <strong>cannibalism</strong> as <strong>cannibalism</strong> <strong>in</strong>volv<strong>in</strong>g members <strong>of</strong><br />
different year classes . Cannibalism among similar-aged <strong>in</strong>dividuals <strong>with</strong><strong>in</strong> a<br />
year class (<strong>in</strong> culture, for example) is termed coeval <strong>cannibalism</strong> . Aggressive<br />
behaviour may be a precursor <strong>of</strong> <strong>cannibalism</strong>, but <strong>in</strong>flicted mortality<br />
<strong>with</strong>out subsequent <strong>in</strong>gestion <strong>of</strong> the victim is not considered as <strong>cannibalism</strong><br />
<strong>in</strong> this context (Hecht <strong>and</strong> Pienaar, 1993) .<br />
Early Life History <strong>and</strong> Recruitment <strong>in</strong> Fish Populations .<br />
Edited by R . Christopher Chambers <strong>and</strong> Edward A . Trippel .<br />
Published <strong>in</strong> 1997 by Chapman & Hall, London . ISBN 0 412 64190 9 .
252 . <strong>Ontogeny</strong> <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> <strong>larval</strong> <strong>and</strong> <strong>juvenile</strong> <strong>fishes</strong><br />
Table 9 .1 List <strong>of</strong> recent reviews <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> fish <strong>and</strong> amphibians . Relevant<br />
contributions <strong>in</strong> book reviews are listed under the book editors<br />
Author(s) Taxa Emphasis<br />
Polis (1981) Several Evolution, population dynamics<br />
Hausfater <strong>and</strong> Hrdy (eds) (1984) Several Infanticide, filial <strong>cannibalism</strong><br />
Dom<strong>in</strong>ey <strong>and</strong> Blumer Pisces Systematic overview, filial<br />
<strong>cannibalism</strong><br />
Simon Amphibia Systematic overview, evolution<br />
Polis <strong>and</strong> Myers (1985) Amphibia Systematic overview (<strong>in</strong>clud<strong>in</strong>g<br />
reptiles)<br />
Smith <strong>and</strong> Reay (1991) Pisces Systematic overview<br />
Elgar <strong>and</strong> Crespi (eds) (1992) Several Systematic overview, ecology,<br />
evolution<br />
Elgar <strong>and</strong> Crespi Several Systematic overview, ecology,<br />
evolution<br />
Dong <strong>and</strong> Polis Pisces Population dynamics, forag<strong>in</strong>g<br />
Sargent Pisces Filial <strong>cannibalism</strong>, modell<strong>in</strong>g<br />
FitzGerald <strong>and</strong> Whoriskey Pisces Filial <strong>cannibalism</strong>, ecology<br />
Crump Amphibia Systematic overview, ecology<br />
Hecht <strong>and</strong> Pienaar (1993) Pisces Larviculture<br />
Cannibalism <strong>in</strong> fish is <strong>of</strong> <strong>special</strong> concern because it can <strong>in</strong>fluence both<br />
aquaculture production <strong>and</strong> fisheries . Several review papers <strong>and</strong> books<br />
have recently been published on the topic (Table 9 .1) . This review empha-<br />
sizes the role <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> the pelagic environment . Some <strong>of</strong> the<br />
processes <strong>in</strong>volved <strong>in</strong> <strong>cannibalism</strong> among amphibian larvae <strong>in</strong> ponds<br />
parallel those present <strong>in</strong> <strong>fishes</strong>, <strong>and</strong> some references from this field are<br />
<strong>in</strong>cluded as well . In both these groups, <strong>cannibalism</strong> is mostly a gape-<br />
limited process <strong>with</strong>out manipulation <strong>of</strong> the prey by external limbs .<br />
F<strong>in</strong>ally, this review is biased towards case studies from extensive <strong>juvenile</strong><br />
production <strong>and</strong> fisheries . In both these systems, temporary food shortage<br />
is expected to occur, <strong>in</strong> contrast to ongrow<strong>in</strong>g under <strong>in</strong>tensive aquaculture<br />
conditions, where <strong>cannibalism</strong> can be significantly reduced by satiation<br />
feed<strong>in</strong>g (Hecht <strong>and</strong> Pienaar, 1993) .<br />
Atlantic cod, Gadus morhua, is one <strong>of</strong> the most important species <strong>in</strong> world<br />
fisheries . Large efforts have recently been made to produce <strong>juvenile</strong> cod<br />
extensively for aquaculture <strong>and</strong> sea ranch<strong>in</strong>g purposes . Many <strong>of</strong> the<br />
examples are therefore taken from this species . Observations <strong>of</strong> <strong>cannibalism</strong><br />
<strong>in</strong> cod <strong>in</strong> outdoor enclosures were recorded <strong>in</strong> the late 1800s, <strong>and</strong> recent<br />
experimental studies have confirmed this cannibalistic propensity (Howell,<br />
1984) . Due to the lack <strong>of</strong> other plausible causes <strong>of</strong> mortality, <strong>cannibalism</strong><br />
was hypothesized to be responsible for the apparent density-dependent<br />
c<br />
m<br />
0<br />
U)<br />
350<br />
300<br />
250<br />
d 150<br />
!0<br />
200<br />
R 100<br />
N<br />
<strong>Ontogeny</strong> <strong>of</strong> coeval <strong>cannibalism</strong><br />
N<br />
- ,<br />
.g<br />
E<br />
3<br />
50<br />
Z o - , - - • I - - - I .<br />
0 1000 2000 3000 4000 5000<br />
Numbers at metamorphosis (thous<strong>and</strong>s)<br />
. . . , . 1<br />
Fig . 9 .1 Numbers <strong>of</strong> metamorphos<strong>in</strong>g cod (age 35-40 days) <strong>and</strong> correspond<strong>in</strong>g<br />
number <strong>of</strong> <strong>juvenile</strong>s harvested (about age 100-140 days) from various ponds<br />
(separate symbols for different ponds) . L<strong>in</strong>es represent 1%, 10% <strong>and</strong> 100% survival<br />
between the two periods .<br />
mortality (1 .5-5% day- ) after metamorphosis <strong>in</strong> extensive <strong>juvenile</strong><br />
production units (Fig . 9 .1 ; Oiestad, 1985) . Cannibalism has been<br />
confirmed by stomach analyses <strong>in</strong> <strong>juvenile</strong> rear<strong>in</strong>g ponds (own unpublished<br />
data), <strong>and</strong> similar reports <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> the field are well documented<br />
(Bogstad et al., 1993) . Knowledge <strong>of</strong> the mechanisms underly<strong>in</strong>g <strong>cannibalism</strong><br />
is thus essential to improve <strong>juvenile</strong> production <strong>of</strong> cod <strong>and</strong> other<br />
species, <strong>and</strong> to obta<strong>in</strong> a better underst<strong>and</strong><strong>in</strong>g <strong>of</strong> the dynamics <strong>in</strong> natural<br />
populations .<br />
9.2 ONTOGENY OF COEVAL CANNIBALISM<br />
Larval stage<br />
Cannibalism has not been observed among early cod larvae <strong>in</strong> the laboratory<br />
(Howell, 1984 ; own observations) . This is not surpris<strong>in</strong>g because the<br />
<strong>larval</strong> mouth height is significantly smaller than the <strong>larval</strong> body height at<br />
this stage (average mouth height is 0 .2-0 .4 mm at first feed<strong>in</strong>g <strong>and</strong><br />
average body height (<strong>in</strong>clud<strong>in</strong>g yolk sac) is 1-1 .2 mm) (Fig . 9 .2(A) ;<br />
Wiborg, 1948 ; Knutsen <strong>and</strong> Tilseth, 1985) . Typical widths <strong>of</strong> common<br />
prey organisms <strong>in</strong>gested dur<strong>in</strong>g the first days <strong>of</strong> exogenous feed<strong>in</strong>g are<br />
0 .1-0 .3 mm (Ellertsen et al ., 1984) . The possibility <strong>of</strong> coeval <strong>cannibalism</strong><br />
is further decreased by the relatively low <strong>in</strong>itial size variability commonly<br />
253
2 5 4 <strong>Ontogeny</strong> <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> <strong>larval</strong> <strong>and</strong> <strong>juvenile</strong> <strong>fishes</strong><br />
0<br />
a<br />
0.5-<br />
3 .0<br />
2.5 -<br />
2.0-<br />
1 .0<br />
B<br />
o +<br />
I 1 f<br />
5<br />
4<br />
I<br />
3<br />
6 ' '<br />
5 10 20 40 80 160<br />
St<strong>and</strong>ard length (mm)<br />
Fig . 9 .2 (A) Size-specific mouth height (solid curve) <strong>and</strong> body height (dashed<br />
curve) <strong>of</strong> Norwegian coastal cod (Otters <strong>and</strong> Folkvord, 1993) . Average values <strong>of</strong><br />
mouth' height are denoted by + for Arcto-Norwegian cod (Wiborg, 1948) <strong>and</strong> by<br />
squares for Pacific cod, Gadus macrocephalus (Shirota, 1970) . (B) Lowest possible<br />
predator :prey ratios based on morphological relations (mouth height <strong>of</strong> cannibal<br />
= body height <strong>of</strong> prey) . Data on cod (from Fig . 9 .2(A) ; l<strong>in</strong>e 1, Otters <strong>and</strong><br />
Folkvord, 1993), koi carp (l<strong>in</strong>e 2, van Damme et al., 1989), walleye pollock (l<strong>in</strong>e<br />
3, Sogard <strong>and</strong> Olla, 1994), sharptooth catfish (l<strong>in</strong>e 4, Hecht <strong>and</strong> Appelbaum,<br />
1988), sea bass (l<strong>in</strong>e 5, Parazo et al ., 1991) <strong>and</strong> pike (l<strong>in</strong>e 6, Bry et al ., 1992).<br />
observed among coeval conspecifics after hatch<strong>in</strong>g (Knutsen <strong>and</strong> Tilseth,<br />
1985 ; Folkvord et al ., 1994b) . Cannibalism is also more common dur<strong>in</strong>g<br />
periods <strong>of</strong> hunger <strong>and</strong> starvation (Folkvord, 1991), <strong>and</strong> the cod larvae are<br />
usually not food limited dur<strong>in</strong>g the transition to exogenous feed<strong>in</strong>g <strong>in</strong> the<br />
I<br />
<strong>Ontogeny</strong> <strong>of</strong> coeval <strong>cannibalism</strong><br />
Table 9 .2 <strong>Ontogeny</strong> <strong>of</strong> coeval <strong>cannibalism</strong> <strong>of</strong> cod <strong>in</strong> enclosures<br />
Stage Mechanism/attribute Cannibalism<br />
Larval, 4-10 mm Low <strong>in</strong>itial size variation Low<br />
Food limitation uncommon<br />
Relatively small mouth ; yolk sac <strong>in</strong>itially large<br />
Metamorphosis, Increas<strong>in</strong>g size variation ; zooplankton energy High<br />
12-30 mm<br />
unevenly distributed <strong>in</strong> size fractions<br />
Food limitation common<br />
Relatively large mouth<br />
High growth rate<br />
Incomplete wean<strong>in</strong>g, starvation<br />
Stomach not fully developed<br />
High density, patchy distribution<br />
Juvenile, Reduced growth rate Low<br />
50-150 mm Lower susceptibility to starvation<br />
Relatively small mouth<br />
Greater feed<strong>in</strong>g flexibility<br />
Completed wean<strong>in</strong>g<br />
Fully functional stomach<br />
Large relative size difference<br />
High<br />
<strong>juvenile</strong> production enclosures (Blom et al., 1991 ; Folkvord et al ., 1994b) .<br />
Thus it is reasonable to assume that coeval <strong>cannibalism</strong> among cod larvae<br />
<strong>in</strong> the laboratory <strong>and</strong> <strong>in</strong> enclosures is <strong>in</strong>significant dur<strong>in</strong>g the early <strong>larval</strong><br />
stage (Table 9 .2) .<br />
The larvae <strong>of</strong> freshwater <strong>fishes</strong> are generally larger <strong>and</strong> more developed<br />
at hatch<strong>in</strong>g than mar<strong>in</strong>e fish larvae (Balon, 1984) . It is not surpris<strong>in</strong>g<br />
therefore that coeval <strong>cannibalism</strong> <strong>in</strong> freshwater species has been reported<br />
to take place shortly after <strong>in</strong>itiation <strong>of</strong> exogenous feed<strong>in</strong>g . In the koi carp,<br />
Cypr<strong>in</strong>us carpio, <strong>cannibalism</strong> commenced one week after onset <strong>of</strong> feed<strong>in</strong>g,<br />
<strong>and</strong> was highest the follow<strong>in</strong>g three weeks (van Damme et al ., 1989) .<br />
This is presumably partly due to the relatively large mouth <strong>of</strong> this species<br />
dur<strong>in</strong>g this period (Fig . 9 .2(B)) . In African sharptooth catfish, Clarias gariep<strong>in</strong>us,<br />
the mouth is relatively small compared <strong>with</strong> the body depth at the<br />
<strong>larval</strong> stage, <strong>and</strong> complete <strong>in</strong>gestion is only observed at cannibal lengths<br />
larger than 45 mm (type ii <strong>cannibalism</strong>, Fig . 9 .2(B) ; Hecht <strong>and</strong><br />
Appelbaum, 1988) . Several accounts <strong>of</strong> coeval <strong>cannibalism</strong> are also<br />
reported among amphibian larvae (Polls <strong>and</strong> Meyers, 1985), <strong>and</strong> <strong>in</strong> some<br />
species this is due to cannibalistic morphs (Crump, 1992) . The cannibalistic<br />
<strong>larval</strong> morphs typically have enlarged dentition <strong>and</strong> mouth dimensions <strong>and</strong><br />
<strong>in</strong>creased jaw musculature compared <strong>with</strong> normal morphs .<br />
255
256 <strong>Ontogeny</strong> <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> <strong>larval</strong> <strong>and</strong> <strong>juvenile</strong> <strong>fishes</strong><br />
Metamorphosis<br />
Metamorphosis is def<strong>in</strong>ed as the stage when the larvae develop anatomical<br />
<strong>and</strong> morphological characteristics similar to those <strong>of</strong> adults (Balon,<br />
1984) . Metamorphosis <strong>in</strong> cod commences <strong>with</strong> the replacement <strong>of</strong> the<br />
<strong>larval</strong> f<strong>in</strong> fold <strong>with</strong> dorsal <strong>and</strong> anal f<strong>in</strong>s at <strong>larval</strong> lengths around 12 mm<br />
(Pedersen <strong>and</strong> Falk-Petersen, 1992) <strong>and</strong> is completed at lengths around<br />
2 5-30 mm .<br />
Around metamorphosis, the mouth morphology makes the cod a more<br />
capable predator (Fig . 9 .2, Ottera <strong>and</strong> Folkvord, 1993) . At lengths <strong>of</strong><br />
about 20mm the cannibal will theoretically need to be only 25% longer<br />
than the prey to completely <strong>in</strong>gest it (Fig . 9 .2(B)) . Pike, Esox lucius, is one<br />
<strong>of</strong> the few species that is morphologically capable <strong>of</strong> <strong>in</strong>gest<strong>in</strong>g relatively<br />
larger sibl<strong>in</strong>gs (Fig . 9 .2(B) ; Bry et al ., 1992) . The relatively large mouth<br />
size <strong>of</strong> cod at this stage may also be an adaptation to its most common<br />
prey dur<strong>in</strong>g the <strong>larval</strong> <strong>and</strong> early <strong>juvenile</strong> stage, Calanus f<strong>in</strong>marchicus<br />
(Folkvord et al., 1994a) . Prey width :mouth gape ratios <strong>in</strong> Japanese<br />
mackerel, Scomber japonicus, larvae average around 0 .3-0 .4 (Hunter <strong>and</strong><br />
Kimbrell, 1980a) . With these ratios, the cod will have to be around<br />
20 mm long to <strong>in</strong>gest the later copepodite <strong>and</strong> adult stages <strong>of</strong> C. f<strong>in</strong>marchicus<br />
(Folkvord et al ., 1994a) .<br />
The relatively large mouth <strong>of</strong> cod dur<strong>in</strong>g metamorphosis may create a<br />
<strong>cannibalism</strong> problem, e<strong>special</strong>ly when food availability <strong>and</strong> suitability are<br />
reduced . Such a reduction <strong>in</strong> zooplankton biomass is commonly observed<br />
<strong>in</strong> the <strong>juvenile</strong> rear<strong>in</strong>g ponds around metamorphosis (Blom et al., 1991,<br />
1994 ; Folkvord et al ., 1994b) . Modell<strong>in</strong>g studies on other species have<br />
shown that larvae <strong>and</strong> early <strong>juvenile</strong>s are particularly vulnerable to<br />
reduction <strong>in</strong> prey availability due to their high metabolic activity (Post,<br />
1990) .<br />
A semistarvation situation might also occur dur<strong>in</strong>g wean<strong>in</strong>g due to failure<br />
to accept formulated feed (Howell, 1984 ; Folkvord, 1991 .) . The problems <strong>of</strong><br />
accept<strong>in</strong>g formulated feed at this stage may to some extent be due to the<br />
relatively slow development <strong>of</strong> a functional stomach <strong>in</strong> cod compared <strong>with</strong><br />
other species (Pedersen <strong>and</strong> Falk-Petersen, 1992), which make this a<br />
critical stage <strong>in</strong> their ontogeny. Recent experiments <strong>with</strong> improved formulated<br />
feeds have shown, however, that survival over 90% dur<strong>in</strong>g wean<strong>in</strong>g<br />
is possible at a size <strong>of</strong> 20 mm <strong>and</strong> above (Ottera <strong>and</strong> Lie, 1991) .<br />
The size variability <strong>with</strong><strong>in</strong> a cohort is larger after metamorphosis than<br />
before metamorphosis (Folkvord et al ., 1994b) . Increased size variability<br />
has been shown to lead to <strong>in</strong>creased <strong>cannibalism</strong> <strong>in</strong> cod <strong>and</strong> other species<br />
at similar stages (DeAngelis et al., 1980 ; Katavic et al., 1989 ; Folkvord <strong>and</strong><br />
Ottera, 1993) . At this stage a max :m<strong>in</strong> body length ratio <strong>of</strong> 1 .5 :1 is<br />
required for <strong>cannibalism</strong> to occur, <strong>and</strong> <strong>cannibalism</strong> can be a major source<br />
1 .2 -<br />
<strong>Ontogeny</strong> <strong>of</strong> coeval <strong>cannibalism</strong><br />
Metamorphosis<br />
1 .0 1 1 1 i 1<br />
5 10 15 20 25 30 35 40 45 50<br />
Age (days)<br />
Fig. 9 .3 Estimated max :m<strong>in</strong> length ratios <strong>of</strong> cod cohorts <strong>in</strong> an<br />
enclosure (solid<br />
curve, cohort 1 ; dashed curve, cohort 2) (Folkvord et al ., 1994b) . Horizontal<br />
dotted l<strong>in</strong>es represent ratios required for <strong>cannibalism</strong> to occur (1 .5) <strong>and</strong> for <strong>cannibalism</strong><br />
to be the ma<strong>in</strong> mortality cause (2 .0) (Folkvord <strong>and</strong> Ottera, 1993) .<br />
<strong>of</strong> mortality at ratios above 2 :1 . Size differences <strong>of</strong> this magnitude do not<br />
normally occur <strong>with</strong><strong>in</strong> a cohort before metamorphosis (Fig . 9 .3) .<br />
The large amount <strong>of</strong> energy available for the cod around metamorphosis<br />
may itself cause a spread <strong>in</strong> size <strong>with</strong><strong>in</strong> a cohort as they reach this stage<br />
(Folkvord et al ., 1994b) . The <strong>in</strong>crease <strong>in</strong> spread can to some extent be a<br />
result <strong>of</strong> only the largest size fraction <strong>of</strong> cod (or the first cohort released)<br />
hav<strong>in</strong>g the opportunity to prey on the largest <strong>and</strong> most energy-rich<br />
zooplankton organisms, C . f<strong>in</strong>marchicus copepodite stages ' rv-vi, before the<br />
collapse <strong>in</strong> the zooplankton biomass <strong>in</strong> the ponds .<br />
The highest growth rates <strong>of</strong> <strong>juvenile</strong> cod are encountered <strong>in</strong> the period<br />
around metamorphosis (Blom et al., 1991) . If conspecifics account for a<br />
fixed proportion <strong>of</strong> the diet, the <strong>cannibalism</strong> rates would also be highest at<br />
this stage because high growth rates are accompanied by high feed<strong>in</strong>g rates<br />
(Folkvord, 1991) . Although <strong>cannibalism</strong> rates have been observed to be<br />
higher at elevated rear<strong>in</strong>g temperatures, experimental studies have not<br />
<strong>in</strong>dicated a temperature effect on <strong>cannibalism</strong> per se because survival to<br />
any given size was similar between treatments (Otterlei et al., 1994) .<br />
Cannibalism rates are therefore expected to be proportional to growth<br />
rates when other factors are equal .<br />
Increas<strong>in</strong>g spatial patch<strong>in</strong>ess <strong>of</strong> fish dur<strong>in</strong>g the late <strong>larval</strong> <strong>and</strong> <strong>juvenile</strong><br />
257
2 5 8 <strong>Ontogeny</strong> <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> <strong>larval</strong> <strong>and</strong> <strong>juvenile</strong> <strong>fishes</strong><br />
stages is common, <strong>and</strong> several experimental studies have shown <strong>cannibalism</strong><br />
to be density dependent <strong>in</strong> these stages (Li <strong>and</strong> Mathias, 1982 ;<br />
Giles et al ., 1986 ; Hecht <strong>and</strong> Appelbaum, 1988) . Density-dependent <strong>cannibalism</strong><br />
rates have also been observed among <strong>juvenile</strong> cod <strong>in</strong> the laboratory<br />
(Otterlei et al ., 1994), where only one <strong>in</strong>cidence <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> 8 weeks<br />
was observed <strong>in</strong> the low-density group (100 fish M-3), whereas 4 .9% were<br />
eaten <strong>in</strong> the high-density group (1000 fish m3 ) . The cod start shoal<strong>in</strong>g<br />
<strong>and</strong> school<strong>in</strong>g after metamorphosis, <strong>and</strong> local densities <strong>in</strong> the <strong>juvenile</strong><br />
rear<strong>in</strong>g ponds <strong>of</strong> 500-1000 fish m -3 have been estimated from dipnet<br />
catches (own unpubl . data) . These fish densities are comparable to the<br />
highest densities used <strong>in</strong> <strong>in</strong>tensive culture experiments (Otterlei et al .,<br />
1994) <strong>and</strong> are typically found among schools feed<strong>in</strong>g on zooplankton<br />
enter<strong>in</strong>g through the screens <strong>in</strong> the dam . Thus the local densities <strong>of</strong><br />
<strong>juvenile</strong> cod <strong>in</strong> the rear<strong>in</strong>g ponds are sufficiently high for <strong>cannibalism</strong> to<br />
be significant .<br />
In flat<strong>fishes</strong>, the morphological changes around metamorphosis<br />
drastically alter an <strong>in</strong>dividual's vulnerability to <strong>cannibalism</strong> <strong>and</strong> <strong>in</strong>traspecific<br />
aggression . The <strong>in</strong>creased body height post metamorphosis is not<br />
accompanied by a correspond<strong>in</strong>g <strong>in</strong>crease <strong>in</strong> mouth gape, thus reduc<strong>in</strong>g<br />
the possibility <strong>of</strong> be<strong>in</strong>g eaten by coeval conspecifics . Substantial aggression<br />
<strong>and</strong> <strong>cannibalism</strong> is observed prior to settl<strong>in</strong>g <strong>of</strong> turbot, Scophthalmus<br />
maxim us, dur<strong>in</strong>g periods <strong>of</strong> food shortage (own observations), but the<br />
mortality under culture conditions is generally low after metamorphosis .<br />
In treefrog tadpoles, Osteopilus septentrionalis, the risk <strong>of</strong> predation <strong>and</strong><br />
<strong>cannibalism</strong> is e<strong>special</strong>ly high dur<strong>in</strong>g metamorphosis, possibly because<br />
the metamorphos<strong>in</strong>g tadpole is less adapted to the aquatic habitat<br />
(Crump, 1986) . Contrary to the common situation, these metamorphos<strong>in</strong>g<br />
larvae are attacked by smaller <strong>and</strong> less-developed tadpoles .<br />
Juvenile stage<br />
The problems associated <strong>with</strong> coeval <strong>cannibalism</strong> <strong>of</strong> cod <strong>in</strong> culture are<br />
reduced later <strong>in</strong> the <strong>juvenile</strong> stage (Table 9 .2) . The growth rate is<br />
reduced to less than a third <strong>of</strong> its maximum value <strong>with</strong><strong>in</strong> 1-2 months<br />
after metamorphosis . Dur<strong>in</strong>g this period, the stomach becomes fully<br />
functional <strong>and</strong> few problems are encountered dur<strong>in</strong>g wean<strong>in</strong>g onto<br />
formulated feeds (Ottera <strong>and</strong> Lie, 1991) . Once wean<strong>in</strong>g is completed,<br />
proper management will prevent food shortage <strong>and</strong> starvation . It<br />
has been shown <strong>in</strong> several studies that the role <strong>of</strong> <strong>cannibalism</strong> is<br />
reduced <strong>in</strong> the presence <strong>of</strong> sufficient quantities <strong>of</strong> alternative food (Li <strong>and</strong><br />
Mathias, 1982 ; Katavic et al., 1989 ; Folkvord, 1991 ; Hecht <strong>and</strong> Pienaar,<br />
1993) .<br />
Due to their reduced growth rate <strong>and</strong> metabolism, larger <strong>juvenile</strong>s are<br />
Cannibalism as a selective process<br />
also more resistant to starvation (Post, 1990 ; Folkvord, 1991) . In addition,<br />
the larger fish generally have more food available ow<strong>in</strong>g to their wider<br />
range <strong>of</strong> acceptable prey sizes (Shirota, 1970 ; Hunter <strong>and</strong> Kimbrell,<br />
1980a) . The potential for <strong>cannibalism</strong> is further reduced by the relatively<br />
large predator :prey size difference needed for <strong>cannibalism</strong> to occur, <strong>and</strong><br />
the relatively small mouth size at this stage (Fig . 9 .2(B), Ottera <strong>and</strong><br />
Folkvord, 1993) . The potential for <strong>cannibalism</strong> <strong>in</strong> the <strong>juvenile</strong> stage may<br />
also <strong>in</strong>crease, however, due to the common <strong>in</strong>crease <strong>in</strong> relative size<br />
between the largest <strong>and</strong> smallest <strong>in</strong>dividuals <strong>in</strong> a cohort (Folkvord et al.,<br />
1994b) . In a culture situation, this can easily be resolved by satiation<br />
feed<strong>in</strong>g <strong>with</strong> suitable feeds <strong>and</strong> size grad<strong>in</strong>g <strong>of</strong> the fish (Hecht <strong>and</strong><br />
Pienaar, 1993) .<br />
9 .3 CANNIBALISM AS A SELF''TIVE PROCESS<br />
Effects on size distribution<br />
Cannibalism is both a cause <strong>and</strong> an effect <strong>of</strong> size variation (Hecht <strong>and</strong><br />
Pienaar, 1993) . In fish it is generally a size-selective process, usually<br />
limited by the mouth size <strong>of</strong> the cannibal (Fig . 9 .2(B) ; Hecht <strong>and</strong><br />
Appelbaum, 1988 ; van Damme et al ., 1989 ; Parazo et al ., 1991 ; Bry et<br />
al., 1992 ; Sogard <strong>and</strong> Olla, 1994) . It follows that <strong>in</strong>tracohort <strong>cannibalism</strong><br />
will selectively remove the smallest <strong>in</strong>dividuals . The effect can be very<br />
dramatic when the size variation <strong>in</strong> the population is sufficiently large<br />
(Fig . 9 .4 ; Folkvord <strong>and</strong> Ottera, 1993) . The proportion <strong>of</strong> cannibals <strong>in</strong> the<br />
population need not be very high to cause high mortalities . Two large cod<br />
<strong>in</strong>dividuals were capable <strong>of</strong> consum<strong>in</strong>g 56 sibl<strong>in</strong>gs (56% <strong>of</strong> the population)<br />
<strong>with</strong><strong>in</strong> 4 weeks (Folkvord, 1991) .<br />
The importance <strong>of</strong> the largest <strong>in</strong>dividuals <strong>in</strong> the <strong>cannibalism</strong> process<br />
<strong>in</strong>dicates that the relative size difference between the largest <strong>and</strong> smallest<br />
<strong>in</strong>dividual <strong>in</strong> a co-occurr<strong>in</strong>g group <strong>of</strong> conspecifics may be a better<br />
measure <strong>of</strong> <strong>cannibalism</strong> risk than the coefficient <strong>of</strong> variation (CV) <strong>of</strong><br />
length or weight . It is important to note that large relative size differences<br />
are more likely to be present <strong>in</strong> a large group <strong>of</strong> fish than <strong>in</strong> a small group<br />
<strong>of</strong> fish . A simple simulation study illustrates this po<strong>in</strong>t . Theoretical populations<br />
generated r<strong>and</strong>omly from the same orig<strong>in</strong>al population (same CV)<br />
show a logarithmic <strong>in</strong>crease <strong>in</strong> max :m<strong>in</strong> length ratios <strong>with</strong> <strong>in</strong>creas<strong>in</strong>g<br />
population size (Fig . 9 .5) . Thus not only does the number <strong>of</strong> encounters<br />
<strong>in</strong>crease <strong>with</strong> <strong>in</strong>creas<strong>in</strong>g density, but also higher relative size differences<br />
between <strong>in</strong>dividuals will be present at higher densities . The <strong>in</strong>crease <strong>in</strong><br />
max:m<strong>in</strong> length ratios <strong>with</strong> <strong>in</strong>creas<strong>in</strong>g population size is more rapid when<br />
the population size variability is high (Fig . 9 .5(A,B)) .<br />
Size variation <strong>with</strong><strong>in</strong> a fish population also depends on previous growth<br />
259
260 <strong>Ontogeny</strong> <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> <strong>larval</strong> <strong>and</strong> <strong>juvenile</strong> <strong>fishes</strong><br />
Fig. 9 .4 Size-selective <strong>cannibalism</strong> mortality <strong>in</strong> <strong>juvenile</strong> cod (Folkvord <strong>and</strong> Otters,<br />
1993). Estimated relative loss <strong>of</strong> small cod <strong>in</strong> tanks <strong>with</strong> 2 (solid l<strong>in</strong>e) <strong>and</strong> 10<br />
(broken l<strong>in</strong>e) large sibl<strong>in</strong>gs added (out <strong>of</strong> 50 fish) relative to control tanks <strong>with</strong> no<br />
large sibl<strong>in</strong>gs added . The percentages miss<strong>in</strong>g are calculated at 0 .05 g <strong>in</strong>tervals (see<br />
po<strong>in</strong>ts) . The experiment lasted 16 days, <strong>and</strong> the <strong>in</strong>itial CVs (length) averaged 12%,<br />
15% <strong>and</strong> 20% respectively for the groups <strong>with</strong> 0, 2 <strong>and</strong> 10 large sibl<strong>in</strong>gs added .<br />
<strong>and</strong> mortality history (Pep<strong>in</strong>, 1989), <strong>and</strong> can rema<strong>in</strong> relatively constant or<br />
even be reduced depend<strong>in</strong>g on the extent <strong>of</strong> <strong>cannibalism</strong> (Folkvord <strong>and</strong><br />
Otters, 1993) . Individual-based models (IBMs) seem particularly useful <strong>in</strong><br />
evaluat<strong>in</strong>g the effect <strong>of</strong> size variation on <strong>cannibalism</strong> because these models<br />
are effective at deal<strong>in</strong>g <strong>with</strong> rare events or <strong>in</strong>dividual characteristics <strong>of</strong> the<br />
population members (Dong <strong>and</strong> Polis, 1992) . The appearance <strong>of</strong> size<br />
bimodality <strong>in</strong> the population is, however, not necessarily the result <strong>of</strong> <strong>in</strong>teraction<br />
between <strong>in</strong>dividuals . Huston <strong>and</strong> DeAngelis (1987) listed four<br />
possible factors <strong>in</strong>fluenc<strong>in</strong>g the changes <strong>in</strong> size distribution : (1) <strong>in</strong>itial<br />
sizes, (2) distribution <strong>of</strong> growth rates among <strong>in</strong>dividuals, (3) size- <strong>and</strong><br />
time-dependent growth rate, <strong>and</strong> (4) selective mortality . These factors may<br />
also act <strong>in</strong> concert . In the case <strong>of</strong> cod, a size bimodality is <strong>in</strong> part due to<br />
<strong>in</strong>creased growth rate <strong>of</strong> the cannibals (Folkvord <strong>and</strong> Otters, 1993) .<br />
Prolonged <strong>cannibalism</strong> can eventually lead to the removal <strong>of</strong> the smallest<br />
size mode, result<strong>in</strong>g <strong>in</strong> a unimodal distribution (DeAngelis et al ., 1980) .<br />
Modell<strong>in</strong>g studies have also predicted that the variation <strong>in</strong> growth rates<br />
would decrease <strong>with</strong> <strong>in</strong>creas<strong>in</strong>g predator abundance (Pep<strong>in</strong>, 1989), but the<br />
estimates <strong>of</strong> mean <strong>and</strong> variance <strong>in</strong> size-frequency distributions would have<br />
to be very precise to detect changes <strong>in</strong> predator abundance . In a similar<br />
50 100 200 500 1000<br />
Population size<br />
Fig . 9 .5 Simulated max :m<strong>in</strong> length ratios <strong>in</strong> r<strong>and</strong>omly generated populations .<br />
These populations were generated from underly<strong>in</strong>g populations <strong>with</strong> a CV (length)<br />
<strong>of</strong> 10% (A) <strong>and</strong> 15% (B) respectively . Fifty populations were generated for each<br />
population size <strong>in</strong> both (A) <strong>and</strong> (B) . Regression equations are (A) :<br />
y = 1 .071 + 0 .305*loglO(x) <strong>and</strong> (B) : y = 1 .066 + 0 .623*log10(x) . Note differ<strong>in</strong>g<br />
vertical scales .<br />
study (Rice et al ., 1993), high variance <strong>in</strong> growth rate <strong>with</strong><strong>in</strong> a cohort<br />
gave substantially higher survival when size-selective predation pressure<br />
was present . This simulation study was based on a piscivore predator<br />
(<strong>in</strong>tercohort predation), <strong>and</strong> the results cannot necessarily be transferred<br />
to cases <strong>of</strong> <strong>in</strong>tracohort <strong>cannibalism</strong>.<br />
Cannibalism as a selective process<br />
20<br />
261
262 <strong>Ontogeny</strong> <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> <strong>larval</strong> <strong>and</strong> <strong>juvenile</strong> <strong>fishes</strong><br />
Bias <strong>in</strong> growth estimation<br />
Growth <strong>in</strong> fish can be related to size <strong>in</strong> two ways . First, growth rate<br />
generally decl<strong>in</strong>es <strong>with</strong> <strong>in</strong>creas<strong>in</strong>g size (Brett, 1979), <strong>with</strong> the exception <strong>of</strong><br />
the early <strong>larval</strong> period <strong>and</strong> periods <strong>of</strong> compensatory growth after temporary<br />
food shortage (Blom et al ., 1991 ; van der Meeren <strong>and</strong> Nmss, 1993) .<br />
Secondly, the <strong>in</strong>dividual growth rate <strong>in</strong> a group may vary accord<strong>in</strong>g to<br />
relative size <strong>and</strong> social hierarchy (Brawn, 1969) . This can be caused by<br />
various size fractions <strong>of</strong> the population feed<strong>in</strong>g on different-sized food<br />
particles (Folkvord <strong>and</strong> Otters, 1993 ; Folkvord et al., 1994b), genetic differences<br />
or behavioural differences (Brawn, 1969) .<br />
When estimat<strong>in</strong>g growth <strong>in</strong> populations <strong>with</strong> size-selective mortality it is<br />
necessary to dist<strong>in</strong>guish between growth estimates based on <strong>in</strong>dividual<br />
growth trajectories <strong>and</strong> estimates based on average sizes <strong>of</strong> fish at various<br />
periods . In the follow<strong>in</strong>g I refer to average <strong>in</strong>dividual weight (or length)<br />
growth rates <strong>in</strong> cases where the estimates are based on separate <strong>in</strong>dividual<br />
growth rates :<br />
<strong>and</strong><br />
Average <strong>in</strong>dividual growth rate = Y [100 * (e5 - 1)] / n (9 .1)<br />
i=1<br />
gi = [ln(W12) - ln(W,1)] / ( t2 - t1)<br />
where W11 <strong>and</strong> Wit are the weights <strong>of</strong> <strong>in</strong>dividual i at times t 1 <strong>and</strong> t 2 respectively,<br />
<strong>and</strong> n is number <strong>of</strong> <strong>in</strong>dividuals <strong>in</strong> the population (or sample) .<br />
Population growth rates, on the other h<strong>and</strong>, are obta<strong>in</strong>ed by us<strong>in</strong>g<br />
average population weights (or lengths) as <strong>in</strong>put <strong>and</strong> are def<strong>in</strong>ed as :<br />
<strong>and</strong><br />
Population growth rate = 100 * (ea - 1) (9 .3)<br />
g = [ln(W2 ) - ln(W1)] / (t2 - t1) (9 .4)<br />
where W1 <strong>and</strong> W2 are the average weights <strong>in</strong> the population at t1 <strong>and</strong> t2 .<br />
When <strong>cannibalism</strong> rates are high, large differences between population<br />
growth rates <strong>and</strong> <strong>in</strong>dividual growth rates are observed (Patriqu<strong>in</strong>, 1967 ;<br />
Ricker, 1975) . Similar effects can also be observed when predation or<br />
fish<strong>in</strong>g rates are strongly size selective (Hanson <strong>and</strong> Chou<strong>in</strong>ard, 1992) .<br />
Thus one cannot <strong>in</strong>fer <strong>in</strong>dividual growth rates from population growth<br />
rates <strong>with</strong>out any measures <strong>of</strong> size-dependent mortality (Otters, 1992) .<br />
On the other h<strong>and</strong>, knowledge <strong>of</strong> size-dependent growth is essential<br />
because it may <strong>in</strong>fluence overall survival <strong>and</strong> recruitment (Tsukamoto et<br />
al ., 1989) .<br />
In <strong>juvenile</strong> cod it has been shown that population growth rate can be<br />
Cannibalism as a selective process<br />
Table 9 .3 Calculation <strong>of</strong> growth rates under three different mortality scenarios :<br />
(A), no mortality ; (B), selective mortality <strong>of</strong> the smallest <strong>in</strong>dividuals ; (C), nonselective<br />
mortality . Rates were calculated as 100*(ee-1), where g is the<br />
<strong>in</strong>stantaneous rate <strong>of</strong> weight <strong>in</strong>crease dur<strong>in</strong>g a 14 day growth period . The<br />
corrected average start weight was obta<strong>in</strong>ed by omitt<strong>in</strong>g the fraction <strong>of</strong> the<br />
smallest fish correspond<strong>in</strong>g to the mortality <strong>in</strong> the follow<strong>in</strong>g period (arbitrary<br />
weight units)<br />
Variable Start F<strong>in</strong>al A F<strong>in</strong>al B F<strong>in</strong>al C<br />
Weight fish 1 20 40 40 40<br />
Weight fish 2 20 40 40 -<br />
Weight fish 3 5 10 - 10<br />
Weight fish 4 5 10 10<br />
Weight fish 5 5 10<br />
Weight fish 6 5 10<br />
Average weight 10 20 40 20<br />
Corrected average start weight 10 20 15<br />
Individual growth rate (% day-1 1 5 5 5<br />
j )<br />
Population growth rate (% day 5 10 5<br />
Corrected growth rate (% day -') 5 5 2<br />
more than twice as high as the estimated average <strong>in</strong>dividual growth rate<br />
(Folkvord <strong>and</strong> Otters, 1993) . Dur<strong>in</strong>g the early <strong>juvenile</strong> stage, it is difficult<br />
to sample cod quantitatively <strong>in</strong> the enclosures . Thus it is common to only<br />
estimate average mortalities from metamorphosis to harvest (Blom et al .,<br />
1991) . Tak<strong>in</strong>g <strong>in</strong>to account the possibility <strong>of</strong> prom<strong>in</strong>ent size-selective<br />
mortality, any population growth estimates dur<strong>in</strong>g this period most likely<br />
overestimate the average <strong>in</strong>dividual growth rates <strong>and</strong> should be treated<br />
<strong>with</strong> caution . Us<strong>in</strong>g population growth rates is equivalent to assum<strong>in</strong>g no<br />
size-selective mortality at all (see also Miller, Chapter 7, this volume) .<br />
If the mortality rate is known, approximate <strong>in</strong>dividual growth rates can<br />
be estimated us<strong>in</strong>g a subpopulation concept (Rosenberg <strong>and</strong> Haugen,<br />
1982 ; Folkvord <strong>and</strong> Otters, 1993 ; van der Meeren <strong>and</strong> Nmss, 1993)<br />
(Table 9 .3) . This estimate will be an underestimate <strong>of</strong> the average <strong>in</strong>dividual<br />
growth rate if any <strong>of</strong> the larger <strong>and</strong> presumed surviv<strong>in</strong>g <strong>in</strong>dividuals<br />
died dur<strong>in</strong>g the growth period (Table 9 .3, scenario C) . In a population<br />
where <strong>cannibalism</strong> <strong>and</strong> removal <strong>of</strong> the smallest <strong>in</strong>dividuals is likely to<br />
occur, the corrected estimate will be a good approximation to the average<br />
<strong>in</strong>dividual growth rates (Folkvord <strong>and</strong> Otters, 1993) (Table 9 .3, scenario<br />
B) . The reliability <strong>of</strong> the method is, however, dependent on the accuracy<br />
<strong>of</strong> the mortality estimate <strong>and</strong> the obta<strong>in</strong>ed size-frequency distribution .<br />
In the field, <strong>in</strong>dividual <strong>larval</strong> <strong>and</strong> early <strong>juvenile</strong> growth rates can be<br />
263
264 . <strong>Ontogeny</strong> <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> <strong>larval</strong> <strong>and</strong> <strong>juvenile</strong> <strong>fishes</strong><br />
<strong>in</strong>ferred from otolith microstructure . These rates are dependent on accurate<br />
age determ<strong>in</strong>ation for size-at-age studies (Bolz <strong>and</strong> Lough, 1988) . Growth<br />
can also be back-calculated based on known otolith size :body size relations<br />
(Campana, 1990) . Although there are few reliable estimates <strong>of</strong> <strong>in</strong>dividual<br />
growth rates <strong>of</strong> cod at present (Suthers <strong>and</strong> Sundby, 1993), otolith microstructure<br />
analysis still rema<strong>in</strong>s as one <strong>of</strong> the few promis<strong>in</strong>g applicable<br />
techniques for obta<strong>in</strong><strong>in</strong>g <strong>in</strong>dividual growth estimates <strong>of</strong> <strong>larval</strong> <strong>and</strong> <strong>juvenile</strong><br />
cod <strong>in</strong> the field .<br />
9 .4 IMPORTANCE OF CANNIBALISM IN THE FIELD<br />
Intracohort <strong>cannibalism</strong><br />
The spawn<strong>in</strong>g season <strong>of</strong> the Arcto-Norwegian cod stock typically lasts 2-3<br />
months, <strong>and</strong> this should produce co-occurr<strong>in</strong>g larvae <strong>of</strong> sufficiently large<br />
size disparity for <strong>cannibalism</strong> to occur. Still, no accounts <strong>of</strong> <strong>in</strong>tracohort<br />
<strong>cannibalism</strong> on cod larvae <strong>in</strong> the field are documented <strong>in</strong> the literature<br />
(e.g . Ellertsen et al ., 1984) . Until recently, <strong>in</strong>tracohort <strong>cannibalism</strong> among<br />
0-group cod <strong>juvenile</strong>s had not been encountered either (Wiborg, 1960 ;<br />
Perry <strong>and</strong> Neilson, 1988) . The f<strong>in</strong>d<strong>in</strong>gs <strong>of</strong> two 0-group cod (7-14 cm) <strong>of</strong>f<br />
Icel<strong>and</strong> <strong>in</strong> 1990 <strong>with</strong> conspecific <strong>juvenile</strong>s <strong>in</strong> their stomachs is the first<br />
documentation <strong>of</strong> <strong>in</strong>tracohort <strong>cannibalism</strong> <strong>in</strong> cod <strong>in</strong> the field (Bogstad et<br />
al., 1993) .<br />
The low <strong>in</strong>cidences <strong>of</strong> <strong>in</strong>tracohort <strong>cannibalism</strong> <strong>in</strong> the field are to some<br />
extent due to density effects . First, the average abundance <strong>of</strong> <strong>larval</strong> <strong>and</strong><br />
<strong>juvenile</strong> cod compared <strong>with</strong> their most common prey organisms is low<br />
(Wiborg, 1960) . Secondly, the density <strong>of</strong> cod <strong>juvenile</strong>s itself tends to be<br />
low, although some exceptions have been observed . Olsen <strong>and</strong> Soldal<br />
(1989) observed over 3 million <strong>juvenile</strong> cod <strong>in</strong> northern Norway <strong>in</strong> large<br />
aggregations <strong>with</strong> average densities <strong>of</strong> 5-8 fish M-3, which is higher than<br />
the average density i "the <strong>juvenile</strong> production ponds (Blom et al ., 1991) .<br />
The highest local deities <strong>in</strong> the field may, therefore, be close to 100 fish<br />
M-3 , the lowest density used <strong>in</strong> the experiments by Otterlei et al . (1994) .<br />
Very low <strong>cannibalism</strong> rates were observed at this density when the<br />
<strong>juvenile</strong> cod were fed ad libitum .<br />
In the field, <strong>in</strong>tracohort <strong>cannibalism</strong> <strong>and</strong> competition may be reduced by<br />
spatial segregation <strong>of</strong> the <strong>of</strong>fspr<strong>in</strong>g due to advection (Economou, 1991) .<br />
The 0-group cod will also be vertically segregated as the settl<strong>in</strong>g process<br />
commences among the larger <strong>juvenile</strong>s (God® et al ., 1993) . In addition,<br />
the shoal<strong>in</strong>g behaviour <strong>of</strong> fish may reduce <strong>cannibalism</strong> <strong>in</strong> the field . It has<br />
been documented for several species that fish prefer to shoal <strong>with</strong> conspecifics<br />
<strong>of</strong> a similar size (Pitcher <strong>and</strong> Parrish, 1993), <strong>and</strong> the relatively low size<br />
variation <strong>in</strong> the shoal will reduce the probability <strong>of</strong> <strong>in</strong>tracohort canni-<br />
Importance <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> the field<br />
balism . In summary it therefore seems unlikely that <strong>in</strong>tracohort <strong>cannibalism</strong><br />
among young cod is <strong>of</strong> any importance <strong>in</strong> the field .<br />
Intracohort <strong>cannibalism</strong> <strong>and</strong> predation among other 0-group gadiforms<br />
have been observed <strong>in</strong> the field, but these <strong>in</strong>stances have usually been<br />
coupled <strong>with</strong> poor feed<strong>in</strong>g conditions (Perry <strong>and</strong> Neilson, 1988 ; Koeller et<br />
al ., 1989) . Intracohort <strong>cannibalism</strong> occurred among silver hake, Merluccius<br />
bil<strong>in</strong>earis, as small as 22-25 mm, <strong>and</strong> accounted for over 25% <strong>of</strong> the<br />
stomach content by weight <strong>in</strong> <strong>juvenile</strong>s larger than 46 mm (Koeller et al .,<br />
1989) . 0-Group cod occurred <strong>in</strong> the stomachs <strong>of</strong> 0-group haddock, Melanogrammus<br />
aeglef<strong>in</strong>us (<strong>in</strong>tracohort predation), at a site characterized by low<br />
zooplankton biomass (Perry <strong>and</strong> Neilson, 1988) .<br />
Young <strong>and</strong> Davies (1990) observed <strong>in</strong>tracohort <strong>cannibalism</strong> <strong>in</strong> 1 .5% <strong>of</strong><br />
the southern bluef<strong>in</strong> tuna, Thunnus maccoyii, larvae <strong>with</strong> food <strong>in</strong> their<br />
stomachs . The consumed larvae were smaller than 4 mm, <strong>and</strong> occurred<br />
<strong>in</strong> 3 out <strong>of</strong> 16 (19%) <strong>of</strong> the larvae between 8 <strong>and</strong> 9 .5 mm length . Larval<br />
<strong>and</strong> <strong>juvenile</strong> tunas generally have relatively high mouth size :body size<br />
ratios (Shirota, 1970 ; Kawai <strong>and</strong> Isibasi, 1983), <strong>and</strong> this most likely<br />
facilitates <strong>cannibalism</strong> to take place earlier <strong>in</strong> ontogeny compared <strong>with</strong><br />
other species .<br />
The degree <strong>of</strong> piscivory <strong>and</strong> <strong>in</strong>tracohort <strong>cannibalism</strong> <strong>in</strong> 0-group<br />
pikeperch, Stizostedion lucioperca, showed marked annual variations dur<strong>in</strong>g<br />
the period 1976-1983 (van Densen, 1985) . Cannibalism was highest <strong>in</strong><br />
1982, when the abundance <strong>of</strong> pikeperch <strong>in</strong>itially was more than 10 times<br />
higher than <strong>in</strong> the other years . Density effects were also found to be<br />
important dur<strong>in</strong>g a large-scale mark-recapture study . Tsukamoto <strong>and</strong> coworkers<br />
(1989) found seven 20mm newly released red sea bream, Pagrus<br />
major, <strong>in</strong> the stomachs <strong>of</strong> simultaneously released fish <strong>of</strong> 40 mm length .<br />
Cannibalism was, however, not considered to have a serious effect on<br />
mortality <strong>of</strong> red sea bream <strong>juvenile</strong>s <strong>in</strong> the field, because this phenomenon<br />
was limited to the stock<strong>in</strong>g area on the first 2 days after release . The high<br />
<strong>cannibalism</strong> rate dur<strong>in</strong>g this period was considered an artifact due to<br />
unnaturally high concentration <strong>of</strong> <strong>juvenile</strong>s follow<strong>in</strong>g the release<br />
(756 000 <strong>in</strong>dividuals on the same site) .<br />
In summary, there is little <strong>in</strong>formation from the field that po<strong>in</strong>ts to<br />
<strong>in</strong>tracohort <strong>cannibalism</strong> as be<strong>in</strong>g <strong>of</strong> importance <strong>in</strong> regulat<strong>in</strong>g overall<br />
survival or ultimately recruitment (Smith <strong>and</strong> Reay, 1991) . Intracohort<br />
<strong>cannibalism</strong> is only expected to be operat<strong>in</strong>g <strong>in</strong> some species under <strong>special</strong><br />
conditions <strong>with</strong> limited food availability . The possibility <strong>of</strong> detect<strong>in</strong>g<br />
<strong>cannibalism</strong> among 0-group fish <strong>in</strong> the field is higher <strong>in</strong> areas <strong>with</strong> low<br />
food availability, but rapid digestion <strong>of</strong> smaller conspecific prey will still<br />
require large numbers <strong>of</strong> potential predators to be <strong>in</strong>vestigated (Folkvord,<br />
1993) . Thus, the local importance <strong>of</strong> <strong>in</strong>tracohort <strong>cannibalism</strong> cannot be<br />
ruled out .<br />
265
266 . <strong>Ontogeny</strong> <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> <strong>larval</strong> <strong>and</strong> <strong>juvenile</strong> <strong>fishes</strong><br />
Intercohort <strong>cannibalism</strong><br />
Numerous accounts <strong>of</strong> <strong>in</strong>tercohort <strong>cannibalism</strong> <strong>in</strong> cod <strong>and</strong> other gadoids <strong>in</strong><br />
the field have been reported (Daan, 1973 ; Dwyer et al., 1987 ; Mehl, 1988 ;<br />
Bailey, 1989) . Large regional <strong>and</strong> temporal differences <strong>in</strong> the frequency <strong>of</strong><br />
<strong>cannibalism</strong> have been observed . These were usually coupled <strong>with</strong> the cooccurrence<br />
<strong>of</strong> 0-group fish <strong>and</strong> older conspecifics (Daan, 1973 ; Dwyer et<br />
al ., 1987) . It is conceivable that <strong>cannibalism</strong> <strong>in</strong> gadoids is <strong>of</strong> <strong>special</strong> importance<br />
at the time <strong>of</strong> settl<strong>in</strong>g <strong>of</strong> the 0-group fish, but horizontal <strong>and</strong> vertical<br />
separation <strong>of</strong> 0-group <strong>and</strong> older cod may to some extent reduce predation<br />
at this stage (Riley <strong>and</strong> Parnell, 1984 ; Godo et al ., 1993) . Intercohort<br />
<strong>cannibalism</strong> has, however, been shown to account for over 20% <strong>of</strong> the<br />
diet <strong>of</strong> older cod, <strong>and</strong> 40% <strong>of</strong> the annual mortality <strong>of</strong> 0-group cod <strong>in</strong> years<br />
when the abundances <strong>of</strong> 0-group cod have been relatively high compared<br />
<strong>with</strong> other prey items (Daan, 1973 ; Mehl, 1988) . Adult walleye pollock,<br />
Theragra chalcogramma, have been shown to consume larger-than-average<br />
0-group <strong>juvenile</strong>s, <strong>and</strong> this was related to the vertical distribution <strong>of</strong><br />
<strong>juvenile</strong>s (Bailey, 1989) . The smaller <strong>juvenile</strong>s found near the surface<br />
were not recovered <strong>in</strong> the stomachs <strong>of</strong> adults . Large <strong>and</strong> faster-grow<strong>in</strong>g 0group<br />
fish settl<strong>in</strong>g early may thus experience a higher mortality dur<strong>in</strong>g this<br />
period than their smaller conspecifics .<br />
In a study on Cape hake, Merluccius capensis, Macpherson <strong>and</strong> Gordoa<br />
(1994) found that large adult hake preferentially selected smaller hake<br />
irrespective <strong>of</strong> their density or <strong>of</strong> the occurrence <strong>of</strong> alternative prey . This<br />
lack <strong>of</strong> density-dependent regulation was possibly compensated for by the<br />
distributional pattern <strong>of</strong> the different size groups <strong>of</strong> hake . The majority <strong>of</strong><br />
the large adult hake were distributed <strong>in</strong> an area which only partially<br />
overlapped <strong>with</strong> the area occupied by the smaller conspecifics . Cannibalism<br />
presure by adult threesp<strong>in</strong>e sticklebacks, Gasterosteus aculeatus, has been<br />
suggested to be responsible for an ontogenetic shift <strong>in</strong> habitat use <strong>of</strong><br />
<strong>juvenile</strong>s <strong>in</strong> this species (Foster et al ., 1988), but similar mechanisms were<br />
not confirmed for hake . A shift from the pelagic habitat to the benthic<br />
habitat for coastal cod was modelled based on mortality rate/growth rate<br />
ratios, <strong>and</strong> the predictions from the model were consistent <strong>with</strong> field observations<br />
(Salvanes et al ., 1994) . Although <strong>in</strong>tercohort <strong>cannibalism</strong> was<br />
documented to be important <strong>in</strong> the benthic habitat, this was possibly<br />
compensated for by <strong>in</strong>creased prey availability <strong>in</strong> the same habitat .<br />
Egg <strong>cannibalism</strong> is a <strong>special</strong> case <strong>of</strong> <strong>cannibalism</strong> that has been confirmed<br />
for several clupeoid filter-feed<strong>in</strong>g species, <strong>and</strong> field estimates have shown<br />
that it can account for 6-70% <strong>of</strong> the daily mortality (Hunter <strong>and</strong><br />
Kimbrell, 1980b ; Valdes Sze<strong>in</strong>feld, 1991) . The overall consequences <strong>of</strong><br />
<strong>cannibalism</strong> are, however, strongly dependent upon the degree <strong>of</strong> overlap<br />
between adults <strong>and</strong> their spawn<strong>in</strong>g products (MacCall, 1981) . Interspecific<br />
Importance <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> the field<br />
predation by co-occurr<strong>in</strong>g species (<strong>in</strong>traguild predation) has been shown <strong>in</strong><br />
some cases to account for an even larger proportion <strong>of</strong> the mortality . Up to<br />
56% <strong>of</strong> the daily anchovy, Engraulis capensis, egg mortality was due to<br />
sard<strong>in</strong>e, Sard<strong>in</strong>ops ocellatus, predation, while 6% was due to <strong>cannibalism</strong><br />
(Valdes Sze<strong>in</strong>feld, 1991) . Egg <strong>cannibalism</strong> rates are expected to be lower<br />
<strong>in</strong> particulate-feed<strong>in</strong>g <strong>fishes</strong> such as most gadiforms, <strong>and</strong> egg <strong>cannibalism</strong><br />
<strong>in</strong> walleye pollock is estimated to account for less than 3% <strong>of</strong> the total egg<br />
mortality (Brodeur et al ., 1991). Several <strong>of</strong> the largest pelagic fish stocks <strong>in</strong><br />
Norwegian waters have demersal eggs (Clupea, Mallotus <strong>and</strong> Ammodytes),<br />
<strong>and</strong> this reproductive strategy effectively elim<strong>in</strong>ates egg <strong>cannibalism</strong> <strong>in</strong><br />
these species .<br />
Management implications<br />
Extensive <strong>cannibalism</strong> will have implications for both fish production <strong>and</strong><br />
stock assessment <strong>of</strong> the given species . The effects <strong>of</strong> <strong>cannibalism</strong> should<br />
therefore be modelled <strong>in</strong> fisheries models . Dur<strong>in</strong>g the mid 1980s the<br />
capel<strong>in</strong>, Mallotus villosus, stock <strong>in</strong> the Barents Sea was drastically reduced<br />
(Mehl, 1988) . As a consequence, the young year classes <strong>of</strong> cod were significantly<br />
reduced due to <strong>cannibalism</strong> . Up to 85% <strong>of</strong> the mortality <strong>of</strong> the igroup<br />
to <strong>in</strong>-group stage was due to <strong>cannibalism</strong> from older year classes<br />
(Mehl, 1988) . The failure to take this effect <strong>in</strong>to account resulted <strong>in</strong>itially<br />
<strong>in</strong> far too optimistic predictions <strong>of</strong> cod recruitment <strong>and</strong> projected total<br />
allowable catch <strong>in</strong> the region . The failure to anticipate this dramatic<br />
reduction <strong>of</strong> some <strong>of</strong> the year classes led to the <strong>in</strong>clusion <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong><br />
the multispecies models for the Barents Sea region . Through an extensive<br />
stomach-sampl<strong>in</strong>g programme undertaken by Norwegian <strong>and</strong> Russian<br />
researchers, the managers are now able to monitor the annual variations<br />
<strong>in</strong> <strong>cannibalism</strong> <strong>in</strong>tensity (Bogstad et al ., 1993) .<br />
An <strong>in</strong>crease <strong>in</strong> the occurrence <strong>of</strong> <strong>cannibalism</strong> <strong>with</strong> size/age is observed<br />
for several cod stocks (Bogstad et al ., 1993) . The age structure <strong>of</strong> the<br />
Arcto-Norwegian cod stock <strong>in</strong> the late 1940s <strong>and</strong> early 1950s was<br />
dom<strong>in</strong>ated by older <strong>in</strong>dividuals, partly due to reduced fish<strong>in</strong>g pressure<br />
dur<strong>in</strong>g the period after the Second World War . It is <strong>in</strong>terest<strong>in</strong>g to note<br />
that the overall occurrence <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> this period seemed to be<br />
higher than dur<strong>in</strong>g the 1980s, when the age distribution has been shifted<br />
towards younger <strong>in</strong>dividuals (Bogstad et al., 1993) . This example emphasizes<br />
the importance <strong>of</strong> underst<strong>and</strong><strong>in</strong>g the age- <strong>and</strong> size-related predation<br />
processes occurr<strong>in</strong>g <strong>in</strong> a stock .<br />
Cannibalism <strong>in</strong> hake was <strong>in</strong>cluded <strong>in</strong> a virtual population analysis (VPA)<br />
model developed by Lleonart <strong>and</strong> co-workers (1985) . Mortality due to<br />
<strong>cannibalism</strong> accounted for 48%, <strong>of</strong> natural mortality . As a consequence, it<br />
was shown that the st<strong>and</strong>ard VPA model systematically underestimated<br />
267
268 . <strong>Ontogeny</strong> <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> <strong>larval</strong> <strong>and</strong> <strong>juvenile</strong> <strong>fishes</strong><br />
the number <strong>and</strong> biomass <strong>of</strong> the youngest cohorts . Without the correction,<br />
the stock appeared older <strong>and</strong> the calculated age-specific mortalities <strong>of</strong> the<br />
youngest age classes were underestimated . The authors further suggested<br />
that <strong>in</strong> a stock where <strong>cannibalism</strong> by older year classes is common, a<br />
management regulation <strong>of</strong> mesh size will be more effective than a regulation<br />
<strong>of</strong> total effort . An <strong>in</strong>creased mesh size will selectively remove large<br />
potential cannibals, enhanc<strong>in</strong>g survival <strong>of</strong> younger year classes by<br />
reduc<strong>in</strong>g <strong>cannibalism</strong> (Lleonart et al., 1985) . These conclusions were<br />
questioned by Punt <strong>and</strong> Hilborn (1994), who concluded that little<br />
precision <strong>in</strong> the management models was lost by omitt<strong>in</strong>g <strong>cannibalism</strong><br />
<strong>in</strong>teractions <strong>in</strong> this species . This result was attributed to uncerta<strong>in</strong>ties <strong>in</strong><br />
other important aspects <strong>of</strong> the population regulation <strong>in</strong> hake . In addition,<br />
considerable effort would have to be made to estimate the parameters<br />
needed <strong>in</strong> the external model.<br />
MacCall (1981) <strong>in</strong>corporated <strong>cannibalism</strong> on eggs <strong>and</strong> larvae <strong>in</strong> a stock-<br />
recruitment model <strong>and</strong> concluded that <strong>cannibalism</strong> <strong>in</strong> northern anchovy,<br />
Engraulis mordax, is sufficiently <strong>in</strong>tense to be a regulatory mechanism . The<br />
densities <strong>of</strong> adult clupeoids are not generally proportional to stock size<br />
ow<strong>in</strong>g to the expansion <strong>and</strong> contraction <strong>of</strong> ranges <strong>with</strong> vary<strong>in</strong>g<br />
abundance . The harvest potential may thus depend on the spatial fish<strong>in</strong>g<br />
pattern <strong>of</strong> <strong>juvenile</strong> <strong>and</strong> adult clupeoids relative to the distribution <strong>of</strong> eggs<br />
<strong>and</strong> larvae (MacCall, 1981) . Usually the spawn<strong>in</strong>g migrations undertaken<br />
by most clupeoid species will reduce the potential for filial <strong>cannibalism</strong> . The<br />
migration pattern <strong>of</strong> the adults may, however, be <strong>in</strong>fluenced by stock size,<br />
as happened <strong>in</strong> the Norwegian spr<strong>in</strong>g-spawn<strong>in</strong>g herr<strong>in</strong>g, Clupea harengus,<br />
follow<strong>in</strong>g the collapse <strong>in</strong> the 1960s . The traditional migration pattern <strong>in</strong>to<br />
the Norwegian Sea after spawn<strong>in</strong>g was ab<strong>and</strong>oned, <strong>and</strong> the stock rema<strong>in</strong>ed<br />
near the Norwegian coast, <strong>in</strong> the drift route <strong>of</strong> their own <strong>of</strong>fspr<strong>in</strong>g<br />
(Rott<strong>in</strong>gen, 1990) . The presence <strong>of</strong> adult herr<strong>in</strong>g <strong>in</strong> the Norwegian<br />
coastal current may thus have delayed the recovery <strong>of</strong> the stock due to<br />
<strong>cannibalism</strong> on larvae .<br />
Special attention to the role <strong>of</strong> <strong>cannibalism</strong> <strong>and</strong> other density-dependent<br />
mechanisms is needed prior to the onset <strong>of</strong> large-scale enhancement enterprises<br />
(Peterman, 1991) . Theoretical considerations have shown that when<br />
<strong>cannibalism</strong> by older conspecifics is responsible for a major part <strong>of</strong> the<br />
<strong>juvenile</strong> mortality, the effect <strong>of</strong> the release will be higher at lower stock<br />
sizes or higher fish<strong>in</strong>g pressure (Ulltang, 1984) . The ga<strong>in</strong> from such a<br />
release may, however, be lost if the fish<strong>in</strong>g pressure exceeds that giv<strong>in</strong>g<br />
the maximum susta<strong>in</strong>able yield (MSY) <strong>of</strong> the natural population . Intercohort<br />
<strong>cannibalism</strong> <strong>of</strong> cod <strong>juvenile</strong>s has been documented by stomach<br />
analyses carried out <strong>in</strong> connection <strong>with</strong> the major cod enhancement<br />
programmes <strong>in</strong> Norway (Svas<strong>and</strong> <strong>and</strong> Kristiansen, 1990 ; Smestad et al .,<br />
1994) . Smestad <strong>and</strong> co-workers (1994) concluded that the large-scale<br />
Perspectives 269<br />
releases <strong>of</strong> cod <strong>juvenile</strong>s <strong>in</strong> a Norwegian fjord did not contribute signifi-<br />
cantly to recruitment <strong>in</strong> the area, <strong>and</strong> attributed this to competition <strong>and</strong><br />
predation from other gadids . The production <strong>of</strong> cod <strong>in</strong> this fjord depends<br />
to a large extent on advected zooplankton from outside the fjord, <strong>and</strong> the<br />
abundance <strong>of</strong> ii-group cod was not different <strong>in</strong> release areas compared<br />
<strong>with</strong> control areas . The effects <strong>of</strong> cod enhancement programmes will<br />
therefore most likely not be worthwhile if the predation pressure from cod<br />
<strong>and</strong> other species on young cod is high (Ulltang, .1984), <strong>and</strong> the possibilities<br />
<strong>of</strong> a successful ranch<strong>in</strong>g programme will be higher when the popula-<br />
tion <strong>in</strong>volved is already at a low level due to overfish<strong>in</strong>g .<br />
In summary, studies on <strong>in</strong>tercohort <strong>cannibalism</strong> <strong>in</strong> cod <strong>and</strong> other species<br />
have also shown that it can be a major source <strong>of</strong> mortality <strong>and</strong> an<br />
important density-dependent mechanism <strong>in</strong> natural fish populations<br />
(Hunter <strong>and</strong> Kimbrell, 1980b ; Ulltang, 1984 ; Valdes Sze<strong>in</strong>feld, 1991) . In<br />
the dome-shaped Ricker curves <strong>of</strong> stock aga<strong>in</strong>st recruitment, this is<br />
apparent as a drop <strong>in</strong> recruitment at high stock levels . It is likely,<br />
however, that the density effect <strong>of</strong> adults is <strong>of</strong>ten exerted via the density <strong>of</strong><br />
eggs <strong>and</strong> larvae they produce, <strong>and</strong> not necessarily through a direct impact<br />
<strong>of</strong> their own abundance (Ricker, 1975) .<br />
9 .5 PERSPECTIVES<br />
Cannibalism among <strong>fishes</strong> has <strong>in</strong> the past <strong>of</strong>ten been viewed as an artifact<br />
occurr<strong>in</strong>g under artificial circumstances . On the other h<strong>and</strong>, recent reviews<br />
<strong>in</strong>dicate that <strong>cannibalism</strong> is far too widespread <strong>in</strong> the animal k<strong>in</strong>gdom <strong>and</strong><br />
<strong>in</strong> <strong>fishes</strong> to be classified as an obscurity (Smith <strong>and</strong> Reay, 1991) .<br />
Genetic <strong>and</strong> evolutionary aspects<br />
The evolution <strong>of</strong> non-predatory <strong>in</strong>terference (e .g . territoriality) is unlikely <strong>in</strong><br />
an open environment such as the pelagic ecosystem where resource<br />
monopolization is impossible (Polls, 1988) . Cannibalism <strong>in</strong> fish is usually<br />
an unequal contest where the smaller victim presents no direct risk to the<br />
cannibal. The structural simplicity <strong>of</strong> the pelagic habitat, coupled <strong>with</strong> the<br />
tendency <strong>of</strong> conspecifics to co-<strong>in</strong>habit a common environment, will also<br />
promote multiple encounters between <strong>in</strong>dividuals <strong>of</strong> the same species . The<br />
school<strong>in</strong>g behaviour <strong>of</strong> many fish species will also further <strong>in</strong>crease the<br />
encounter rate between conspecifics . There are thus several sound ecological<br />
<strong>and</strong> evolutionary reasons for <strong>cannibalism</strong> be<strong>in</strong>g a part <strong>of</strong> the natural<br />
behavioural repertoire <strong>of</strong> many fish species (Polls, 1981 ; Edgar <strong>and</strong> Crespi,<br />
1992) . The selective advantage <strong>of</strong> <strong>in</strong>dividuals exhibit<strong>in</strong>g cannibalistic traits<br />
is evident <strong>in</strong> situations <strong>of</strong> food shortage . In addition to <strong>in</strong>creas<strong>in</strong>g the fitness
270 <strong>Ontogeny</strong> <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> <strong>larval</strong> <strong>and</strong> <strong>juvenile</strong> <strong>fishes</strong><br />
<strong>of</strong> the cannibal, the result<strong>in</strong>g reduced competition for food will possibly<br />
<strong>in</strong>crease the fitness <strong>of</strong> all other surviv<strong>in</strong>g <strong>juvenile</strong>s (Polis, 1981 ; Elgar <strong>and</strong><br />
Crespi, 1992) .<br />
The cannibal benefits directly from obta<strong>in</strong><strong>in</strong>g a meal <strong>of</strong> high nutritional<br />
value (Polis, 1981) . The proximal composition <strong>of</strong> the prey is also similar to<br />
the proximal composition <strong>of</strong> the predator . There are also some <strong>in</strong>dications<br />
that cod <strong>and</strong> other <strong>fishes</strong> <strong>and</strong> amphibians grow better on a diet <strong>of</strong> conspecifics<br />
(Crump, 1992 ; Folkvord <strong>and</strong> Otters, 1993) . Post<strong>larval</strong> mahi-mahi,<br />
Coryphaena hippurus, grew better on a diet <strong>of</strong> conspecific yolk-sac larvae<br />
than on live br<strong>in</strong>e shrimp, Artemia, <strong>and</strong> the growth rates were up to 34%<br />
day l (Kraul et al ., 1992) . The authors attributed this result to the<br />
proximal composition <strong>of</strong> yolk-sac larvae, which had relatively high levels<br />
<strong>of</strong> polyunsaturated fatty acids .<br />
One <strong>of</strong> the many strik<strong>in</strong>g differences between the terrestrial <strong>and</strong> aquatic<br />
ecosystems is the common size disparity between members <strong>of</strong> the lower <strong>and</strong><br />
higher trophic levels . In the mar<strong>in</strong>e environment this is also manifested <strong>in</strong><br />
the high biomass density <strong>of</strong> relatively small planktonic organisms<br />
(Boudreau <strong>and</strong> Dickie, 1992) . Intermediate-sized organisms are <strong>of</strong>ten<br />
needed <strong>in</strong> order to obta<strong>in</strong> an efficient energy transfer between these<br />
plankton resources <strong>and</strong> the higher trophic levels . Accord<strong>in</strong>g to Nellen<br />
(1986), <strong>cannibalism</strong> <strong>of</strong> younger planktivorous conspecifics represents such<br />
an <strong>in</strong>termediate trophic level . In their analysis <strong>of</strong> the life history <strong>of</strong> Japanese<br />
<strong>fishes</strong>, Kawai <strong>and</strong> Isibasi (1983) observed a between-species discont<strong>in</strong>uity<br />
<strong>in</strong> growth patterns dur<strong>in</strong>g the <strong>juvenile</strong> period . The authors suggested that<br />
this was due to differential adaptation <strong>of</strong> the various species to food acquisition<br />
dur<strong>in</strong>g the early <strong>juvenile</strong> stage . Species <strong>with</strong> relatively large mouths,<br />
<strong>and</strong> result<strong>in</strong>g high piscivory <strong>and</strong> <strong>cannibalism</strong> potential, were expected to<br />
outgrow the plankton-eat<strong>in</strong>g species dur<strong>in</strong>g this period .<br />
Whether fish preferentially cannibalize non-sibl<strong>in</strong>gs is unclear . There is,<br />
however, some evidence that certa<strong>in</strong> species <strong>of</strong> amphibians are able to<br />
recognize their own k<strong>in</strong> . The ability to recognize their own k<strong>in</strong> is<br />
necessary for k<strong>in</strong> selection to take place, <strong>and</strong> such mechanisms are<br />
documented for salam<strong>and</strong>ers <strong>and</strong> toads (Walls <strong>and</strong> Roudebush, 1991 ;<br />
Pfennig et al ., 1993) . Female poeciliids preferentially consumed <strong>in</strong>dividuals<br />
<strong>of</strong> other females rather than their own (Loekle et al., 1982), but further<br />
studies are needed to determ<strong>in</strong>e the mechanisms <strong>in</strong>volved . Lower canni-<br />
balism rates were also observed <strong>in</strong> full-sib groups <strong>of</strong> pike, Esox lucius,<br />
compared <strong>with</strong> mixed groups (Bry <strong>and</strong> Gillet, 1980) . This could possibly<br />
have been due to the lower <strong>in</strong>herent size variation <strong>of</strong> the full-sib groups<br />
<strong>and</strong> not directly to genetic effects .<br />
Many aquaculturists have noted the presence <strong>of</strong> unusually small <strong>and</strong><br />
slow-grow<strong>in</strong>g <strong>in</strong>dividuals dur<strong>in</strong>g rear<strong>in</strong>g <strong>of</strong> various fish species (own observations<br />
; Polis, 1981) . Although the size <strong>of</strong> these <strong>in</strong>dividuals may be a result<br />
Perspectives<br />
<strong>of</strong> <strong>in</strong>jury or disease, the existence <strong>of</strong> so-called runts may have evolutionary<br />
significance . In snails it is common for some <strong>of</strong> the <strong>of</strong>fspr<strong>in</strong>g to feed on<br />
trophic eggs (Polis, 1981) . It has been suggested that the production <strong>of</strong><br />
small <strong>in</strong>dividuals serves the same purpose <strong>in</strong> fish populations, where some<br />
<strong>of</strong> the <strong>of</strong>fspr<strong>in</strong>g are provided as suitable-sized prey for the largest <strong>in</strong>dividuals<br />
later dur<strong>in</strong>g ontogeny (Polis, 1981) . Whether this is an acceptable <strong>in</strong>terpre-<br />
tation for the pelagic environment, where the provided <strong>of</strong>fspr<strong>in</strong>g <strong>in</strong> no way<br />
can be 'reserved' for conspecifics, is questionable (Polis, 1988) .<br />
In some amphibians there are well-documented accounts <strong>of</strong> cannibalistic<br />
polyphenism, i .e . phenotypic differences <strong>in</strong> behaviour, morphology, growth<br />
rates or life history between cannibal <strong>and</strong> non-cannibal forms (Polis,<br />
1981) . In most cases the development <strong>of</strong> the cannibal morph seems to be<br />
environmentally <strong>in</strong>duced when <strong>larval</strong> densities are high or food levels are<br />
low (Crump, 1992) . The development <strong>of</strong> cannibalistic morphs is also<br />
dependent on the presence <strong>of</strong> close k<strong>in</strong> <strong>and</strong> alternate prey (Pfennig <strong>and</strong><br />
Coll<strong>in</strong>s, 1993) .<br />
Few examples <strong>of</strong> cannibalistic polymorphism are found among <strong>fishes</strong>, but<br />
<strong>in</strong> Arctic charr, Salvel<strong>in</strong>us alp<strong>in</strong>us, several coexist<strong>in</strong>g morphs have been<br />
identified (S<strong>and</strong>lund et al ., 1992) . The morph <strong>with</strong> the largest mouth<br />
dimensions was ma<strong>in</strong>ly piscivorous, <strong>and</strong> was the only morph documented<br />
to be cannibalistic . There are also polymorphic adaptations to reduce the<br />
effect <strong>of</strong> predation <strong>and</strong> possibly <strong>cannibalism</strong> . Crucian carp, Carassius<br />
carassius, liv<strong>in</strong>g <strong>in</strong> ponds <strong>with</strong> larger piscivore predators develop enlarged<br />
body heights compared <strong>with</strong> those exposed to a lower predation risk<br />
(Tonn et al ., 1994) . Although <strong>cannibalism</strong> is documented <strong>in</strong> carp, it is not<br />
clear if this morphological response is triggered <strong>in</strong> the presence <strong>of</strong> large<br />
sibl<strong>in</strong>gs .<br />
Not surpris<strong>in</strong>gly, a genetic component <strong>of</strong> cannibalistic behaviour has<br />
been demonstrated (Thibault, 1974 ; Hecht <strong>and</strong> Pienaar, 1993) . Canni-<br />
balism may also <strong>in</strong>directly be affected by genetic effects because <strong>in</strong>herent<br />
size variation <strong>with</strong><strong>in</strong> full-sibl<strong>in</strong>g groups tends to be lower than that<br />
between mixed-sibl<strong>in</strong>g groups (Knutsen <strong>and</strong> Tilseth, 1985 ; Folkvord et al .,<br />
1994b) . Dur<strong>in</strong>g the extensive rear<strong>in</strong>g process <strong>of</strong> <strong>juvenile</strong> cod, periods <strong>of</strong><br />
food limitation are common, <strong>and</strong> are expected to favour cannibalistic<br />
<strong>in</strong>dividuals (Blom et al ., 1994) . Caution is thus appropriate when select<strong>in</strong>g<br />
for rapid growth among broodstock <strong>in</strong> cannibalistic species such as cod,<br />
because the fast-grow<strong>in</strong>g survivors may also be the <strong>in</strong>dividuals <strong>with</strong> the<br />
highest cannibalistic propensity (Hecht <strong>and</strong> Pienaar, 1993) .<br />
There are also mathematical derivations which show that <strong>cannibalism</strong><br />
can function as a 'lifeboat' mechanism, prevent<strong>in</strong>g all specimens <strong>in</strong> a<br />
population from becom<strong>in</strong>g ext<strong>in</strong>ct (van den Bosch et al ., 1988) . Such<br />
mechanisms should be expla<strong>in</strong>ed <strong>in</strong> terms <strong>of</strong> selection at the <strong>in</strong>dividual<br />
level . It can, however, be concluded, regardless <strong>of</strong> whatever selective<br />
271
272 <strong>Ontogeny</strong> <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> <strong>larval</strong> <strong>and</strong> <strong>juvenile</strong> <strong>fishes</strong><br />
agent is responsible, that <strong>cannibalism</strong> has the potential <strong>of</strong> prevent<strong>in</strong>g a<br />
population from becom<strong>in</strong>g ext<strong>in</strong>ct by self-regulation (Polis, 1981) .<br />
Conclud<strong>in</strong>g remarks<br />
Through evolutionary processes, <strong>larval</strong> <strong>and</strong> <strong>juvenile</strong> fish are adapted to<br />
variable feed<strong>in</strong>g conditions . In a farm<strong>in</strong>g or experimental situation these<br />
adaptations represent <strong>in</strong> some cases undesirable features that have to be<br />
dealt <strong>with</strong> . In extensive <strong>juvenile</strong> rear<strong>in</strong>g, it is important to match the<br />
released numbers <strong>of</strong> fish larvae <strong>with</strong> the tim<strong>in</strong>g <strong>and</strong> production <strong>of</strong> suitable<br />
prey . The potential zooplankton production (<strong>and</strong> supply) <strong>in</strong> the ponds has<br />
been shown to impose a limit on the <strong>juvenile</strong> production <strong>of</strong> cod <strong>and</strong> other<br />
species (McIntyre et al ., 1987) . A common error has been to release<br />
relatively high numbers <strong>of</strong> larvae to be certa<strong>in</strong> that some will survive .<br />
Almost <strong>with</strong>out exception, this has led to a zooplankton collapse <strong>in</strong> the<br />
ponds before the fish are readily harvested or weaned . Future r studies on<br />
extensive rear<strong>in</strong>g <strong>with</strong> lower <strong>in</strong>itial <strong>larval</strong> densities <strong>and</strong>/or earlier harvest<br />
are therefore needed .<br />
In a culture situation, coeval <strong>cannibalism</strong> represents an undesirable<br />
trophic level reduc<strong>in</strong>g the potential output given a limited food resource,<br />
<strong>and</strong> should thus be avoided . On average, around 60% <strong>of</strong> the zooplankton<br />
energy <strong>in</strong>gested by the cannibal victims will be added heat loss <strong>in</strong> an<br />
extra trophic level (Blom et al ., 1991), <strong>and</strong> cont<strong>in</strong>ued <strong>cannibalism</strong> will thus<br />
quickly reduce the population biomass (Kawai <strong>and</strong> Isibasi, 1983) . An<br />
exception is the use <strong>of</strong> added fish larvae as a direct food source for the<br />
older conspecifics . Studies on post<strong>larval</strong> mahi-mahi have demonstrated<br />
that this can be a viable strategy if available broodstock can produce sufficient<br />
quantities <strong>of</strong> eggs . It was estimated that four females could produce<br />
enough eggs <strong>and</strong> yolk-sac larvae to raise a few hundred postlarvae<br />
through wean<strong>in</strong>g (Kraul et al ., 1992) .<br />
Cannibalism <strong>in</strong> the field is highly dependent on the co-occurence <strong>of</strong><br />
older conspecifics . The process <strong>of</strong> settl<strong>in</strong>g <strong>in</strong> cod <strong>and</strong> other <strong>fishes</strong> st<strong>and</strong>s<br />
out as an important event which is poorly described <strong>and</strong> documented .<br />
Spatial <strong>and</strong> temporal variations <strong>in</strong> the time <strong>of</strong> settl<strong>in</strong>g are expected to<br />
have an important impact on <strong>in</strong>tercohort <strong>cannibalism</strong> rates (Bailey, 1989) .<br />
Cannibal morphs are well documented <strong>in</strong> amphibians <strong>and</strong> future studies<br />
on fish should look for polymorphic traits <strong>in</strong> situations where <strong>cannibalism</strong><br />
is important . K<strong>in</strong> recognition <strong>in</strong> fish, if it exists, can have wide-rang<strong>in</strong>g<br />
implications <strong>in</strong> our culture strategies . At present there is no evidence <strong>of</strong><br />
k<strong>in</strong> recognition play<strong>in</strong>g an important role <strong>in</strong> reduc<strong>in</strong>g fish <strong>cannibalism</strong>,<br />
but this possibility needs to be addressed . Comparative studies on allometric<br />
mouth morphology can also yield new <strong>in</strong>sight to ontogenetic changes <strong>in</strong><br />
the <strong>in</strong>tracohort cannibalistic propensity .<br />
References<br />
A f<strong>in</strong>al comment regard<strong>in</strong>g the role <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> the field : although<br />
it undoubtedly does occur <strong>in</strong> a large range <strong>of</strong> species under captive conditions,<br />
<strong>special</strong> care should be taken to avoid extrapolation <strong>of</strong> laboratory data<br />
to the field (Nesbit <strong>and</strong> Meffe, 1993) . Controlled experiments <strong>in</strong> the laboratory<br />
are well suited for isolat<strong>in</strong>g factors <strong>of</strong> importance, but the rates <strong>of</strong><br />
<strong>cannibalism</strong> cannot be directly transferred to the field . The ultimate<br />
evidence <strong>of</strong> the role <strong>of</strong> <strong>cannibalism</strong> has to be found <strong>in</strong> the respective<br />
habitats <strong>of</strong> the species under <strong>in</strong>vestigation .<br />
ACKNOWLEDGEMENTS<br />
The constructive comments <strong>of</strong> G . Blom, C . Booman <strong>and</strong> A . Johannessen<br />
<strong>and</strong> three anonymous referees are greatly appreciated . The work has been<br />
funded by research fellowships from the Norwegian Research Council<br />
(former NFFR) <strong>and</strong> the University <strong>of</strong> Bergen .<br />
REFERENCES<br />
Bailey, K .M . (1989) Interaction between the vertical distribution <strong>of</strong> <strong>juvenile</strong> walleye<br />
pollock Theragra chalcogramma <strong>in</strong> the eastern Ber<strong>in</strong>g Sea, <strong>and</strong> <strong>cannibalism</strong> . Mar<strong>in</strong>e<br />
Ecology Progress Series, 53, 205-213 .<br />
Balon, E .K . (1984) Reflections on some decisive events <strong>in</strong> the early life <strong>of</strong> <strong>fishes</strong> .<br />
Transactions <strong>of</strong> the American Fisheries Society, 113, 178-185 .<br />
Blom, G ., Otters, H., Svss<strong>and</strong>, T . et al. (1991) The relationship between feed<strong>in</strong>g<br />
condition <strong>and</strong> production <strong>of</strong> cod fry (Gadus morhua L.) <strong>in</strong> a semi-enclosed mar<strong>in</strong>e<br />
ecosystem <strong>in</strong> western Norway, illustrated by use <strong>of</strong> a consumption model . ICES<br />
Mar<strong>in</strong>e Science Symposia, 192, 176-189 .<br />
Blom, G ., Svsstad, K .E ., Otters, H . et al . (1994) Comparative survival <strong>and</strong> growth<br />
<strong>of</strong> two stra<strong>in</strong>s <strong>of</strong> Atlantic cod (Gad us morhua) through the early life stages <strong>in</strong> a<br />
mar<strong>in</strong>e pond. Canadian Journal<br />
1023 .<br />
<strong>of</strong> Fisheries <strong>and</strong> Aquatic Sciences, 51, 1012-<br />
Bogstad, B., Lilly, G ., Mehl, .S . et al .<br />
273<br />
(1993) Cannibalism <strong>and</strong> year-class strength <strong>in</strong><br />
Atlantic cod (Gadus morhua L .) <strong>in</strong> the Arcto-boreal ecosystems, Barents Sea,<br />
Icel<strong>and</strong> <strong>and</strong> Eastern Newfoundl<strong>and</strong> . ICES Mar<strong>in</strong>e Science Symposia, 198, 576-<br />
599 .<br />
Bolz, G .R . <strong>and</strong> Lough, R .G . (1988) Growth through the first six months <strong>of</strong> Atlantic<br />
cod, Gadus morhua, <strong>and</strong> haddock, Melanogrammus aeglef<strong>in</strong>us, based on daily otolith<br />
<strong>in</strong>crements . Fishery Bullet<strong>in</strong>, U.S., 86, 223-235 .<br />
van den Bosch, F ., Deroos, A .M . <strong>and</strong> Gabriel, W. (1988) Cannibalism as a life boat<br />
mechanism . Journal <strong>of</strong> Mathematical Biology, 26, 619-633 .<br />
Boudreau, P .R. <strong>and</strong> Dickie, L .M. (1992) Biomass spectra <strong>of</strong> aquatic ecosystems <strong>in</strong><br />
relation <strong>in</strong> fisheries yield . Canadian Journal <strong>of</strong> Fisheries <strong>and</strong> Aquatic Sciences, 49,<br />
1528-1538 .<br />
Brawn, V .M . (1969) Feed<strong>in</strong>g behaviour <strong>of</strong> cod (Gadus morhua) . Journal <strong>of</strong> the<br />
Fisheries Research Board <strong>of</strong> Canada, 26, 583-596 .
274 , <strong>Ontogeny</strong> <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> <strong>larval</strong> <strong>and</strong> <strong>juvenile</strong> <strong>fishes</strong><br />
Brett, J .R . (1979) Environmental factors <strong>and</strong> growth, <strong>in</strong> Fish Physiology, Vol . 8<br />
(eds W .S . Hoar, D .J . R<strong>and</strong>al <strong>and</strong> J .R . Brett), London, Academic Press, pp . 599-<br />
675 .<br />
Brodeur, R .D ., Bailey, K .M . <strong>and</strong> Kim, S . (1991) Cannibalism on eggs by walleye<br />
pollock Theragra chalcogramma <strong>in</strong> Shelik<strong>of</strong> Strait, Gulf <strong>of</strong> Alaska . Mar<strong>in</strong>e Ecology<br />
Progress Series, 71, 207-218 .<br />
Bry, C. <strong>and</strong> Gillet, C . (1980) Reduction <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> pike (Esox lucius) fry by<br />
isolation <strong>of</strong> full-sib families . Reproduction, Nutrition et Developpement, 20, 173-<br />
182 .<br />
Bry, C., Basset, E ., Rognon, X . <strong>and</strong> Bonamy, F . (1992) Analysis <strong>of</strong> sibl<strong>in</strong>g <strong>cannibalism</strong><br />
among pike, Esox lucius, <strong>juvenile</strong>s reared under semi-natural conditions .<br />
Environmental Biology <strong>of</strong> Fishes, 35, 75-84 .<br />
Campana, S .E . (1990) How reliable are growth back-calculations based on otoliths?<br />
Canadian Journal <strong>of</strong> Fisheries <strong>and</strong> Aquatic Sciences, 47, 2219-2227.<br />
Crump, M .L . (1986) Cannibalism by younger tadpoles : another hazard to metamorphosis<br />
. Copeia, 1986, 1007-1009 .<br />
Crump, M.L. (1992) Cannibalism <strong>in</strong> amphibians, <strong>in</strong> Cannibalism . Ecology <strong>and</strong><br />
Evolution among Diverse Taxa (eds M.A . Elgar <strong>and</strong> B.J . Crespi), Oxford University<br />
Press, Oxford, pp . 256-276 .<br />
Daan, N . (1973) A quantitative analysis <strong>of</strong> the food <strong>in</strong>take <strong>of</strong> North Sea cod, Gadus<br />
morhua . Netherl<strong>and</strong>s Journal <strong>of</strong> Sea Research, 6, 479-517 .<br />
van Damme, P ., Appelbaum, S .A . <strong>and</strong> Hecht, T. (1989) Sibl<strong>in</strong>g <strong>cannibalism</strong> <strong>in</strong> Koi<br />
carp, Cypr<strong>in</strong>us carpio L ., larvae <strong>and</strong> <strong>juvenile</strong>s reared under controlled conditions .<br />
Journal <strong>of</strong> Fish Biology, 34, 855-863 .<br />
DeAngelis, D .L ., Cox, D.K. <strong>and</strong> Coutant, C .C . (1980) Cannibalism <strong>and</strong> size dispersal<br />
<strong>in</strong> young-<strong>of</strong>-the-year largemouth bass : experiment <strong>and</strong> a model . Ecological<br />
Modell<strong>in</strong>g, 8, 133-148 .<br />
van Densen, W.L.T . (1985) Piscivory <strong>and</strong> the development <strong>of</strong> bimodality <strong>in</strong> the size<br />
distribution <strong>of</strong> 0 + pikeperch (Stizostedion lucioperca L .) . Zeitschrift fur Angew<strong>and</strong>te<br />
Ichthyologie, 3, 119-131 .<br />
Dong, Q. <strong>and</strong> Polis, G .A . (1992) The dynamics <strong>of</strong> cannibalistic populations : a<br />
forag<strong>in</strong>g perspective, <strong>in</strong> Cannibalism . Ecology <strong>and</strong> Evolution among Diverse Taxa<br />
(eds M.A . Elgar <strong>and</strong> B.J . Crespi), Oxford University Press, Oxford, pp . 13-37 .<br />
Dwyer, D .A ., Bailey, K .M . <strong>and</strong> Liv<strong>in</strong>gston, P .A . (1987) Feed<strong>in</strong>g habits <strong>and</strong> daily<br />
ration <strong>of</strong> walleye pollock (Theragra chalcogramma) <strong>in</strong> the eastern Ber<strong>in</strong>g Sea, <strong>with</strong><br />
<strong>special</strong> reference to <strong>cannibalism</strong> . Canadian Journal <strong>of</strong> Fisheries <strong>and</strong> Aquatic Sciences,<br />
44,1972-1984.<br />
Economou, A .N. (1991) Is dispersal <strong>of</strong> fish eggs, embryos <strong>and</strong> larvae an <strong>in</strong>surance<br />
aga<strong>in</strong>st density dependence? Environmental Biology <strong>of</strong> Fishes, 31, 313-321 .<br />
Elgar, M.A . <strong>and</strong> Crespi, B .J. (eds) (1992) Cannibalism . Ecology <strong>and</strong> Evolution among<br />
Diverse Taxa, Oxford University Press, Oxford, 351 pp .<br />
Ellertsen, B ., Fossum, P ., Solemdal, P . <strong>and</strong> Tilseth, S . (1984) A case study on the<br />
distribution <strong>of</strong> cod larvae <strong>and</strong> availability <strong>of</strong> prey organisms <strong>in</strong> relation to<br />
physical processes <strong>in</strong> L<strong>of</strong>oten . Fledevigen Rapportserie, 1, 453-477 .<br />
Folkvord, A . (1991) Growth, survival <strong>and</strong> <strong>cannibalism</strong> <strong>of</strong> cod <strong>juvenile</strong>s (Gadus<br />
morhua L .) : effects <strong>of</strong> feed type, starvation <strong>and</strong> fish size . Aquaculture, 97, 41-59 .<br />
Folkvord, A . (1993) Prey recognition <strong>in</strong> stomachs <strong>of</strong> cannibalistic <strong>juvenile</strong> cod<br />
(Gad us morhua L.) . Sarsia, 78, 97-100 .<br />
Folkvord, A . <strong>and</strong> Otters, H . (1993) Effects <strong>of</strong> <strong>in</strong>itial size distribution, day length <strong>and</strong><br />
feed<strong>in</strong>g frequency on growth, survival <strong>and</strong> <strong>cannibalism</strong> <strong>in</strong> <strong>juvenile</strong> Atlantic cod<br />
(Gad us morhua L .) . Aquaculture, 114, 243-260 .<br />
References<br />
Folkvord, A ., Blom, G ., Dragesund, 0 . et al . (1994a) A conceptual framework for<br />
enhanc<strong>in</strong>g <strong>and</strong> stabiliz<strong>in</strong>g recruitment <strong>of</strong> mar<strong>in</strong>e fish stocks . Aquaculture <strong>and</strong><br />
Fisheries Management, 25, 245-258 .<br />
Folkvord, A., Oiestad, V . <strong>and</strong> Kvenseth, P .G . (1994b) Growth patterns <strong>of</strong> three<br />
cohorts <strong>of</strong> Atlantic cod larvae (Gadus morhua L.) studied <strong>in</strong> a macrocosm . ICES<br />
Journal <strong>of</strong> Mar<strong>in</strong>e Science, 51, 325-336 .<br />
Foster, S .A ., Garcia, V .B . <strong>and</strong> Town, M .Y . (1988) Cannibalism as the cause <strong>of</strong> an<br />
ontogenetic shift <strong>in</strong> habitat use by fry <strong>of</strong> the threesp<strong>in</strong>e stickleback .<br />
74,577-585 .<br />
Oecologia,<br />
Giles, N ., Wright, R .M . <strong>and</strong> Nord, M.E . (1986) Cannibalism <strong>in</strong> pike fry, Esox lucius<br />
L. : some experiments <strong>with</strong> fry densities . Journal <strong>of</strong> Fish Biology, 29, 107-113 .<br />
Gode, O .R., Valdemarsen, J .W . <strong>and</strong> Engss, A . (1993) Comparison <strong>of</strong> efficiency <strong>of</strong><br />
st<strong>and</strong>ard <strong>and</strong> experimental <strong>juvenile</strong> gadoid sampl<strong>in</strong>g trawls . ICES Mar<strong>in</strong>e Science<br />
Symposia, 196, 196-201 .<br />
Hanson, J.M . <strong>and</strong> Chou<strong>in</strong>ard, G .A . (1992) Evidence that size-selective mortality<br />
affects growth <strong>of</strong> Atlantic cod (Gadus morhua L .) <strong>in</strong> the southern Gulf <strong>of</strong> St<br />
Lawrence . Journal <strong>of</strong> Fish Biology, 41, 31-41 .<br />
Hausfater, G . <strong>and</strong> Hrdy, S .B . (eds) (1984) Infanticide: Comparative <strong>and</strong> Evolutionary<br />
Perspectives, Ald<strong>in</strong>e Publish<strong>in</strong>g Co ., New York, 598 pp .<br />
Hecht, T . <strong>and</strong> Appelbaum, S .A . (1988) Observations on <strong>in</strong>traspecific aggregation<br />
<strong>and</strong> coeval sibl<strong>in</strong>g <strong>cannibalism</strong> by <strong>larval</strong> <strong>and</strong> <strong>juvenile</strong> Clarias gariep<strong>in</strong>us<br />
(Clariidae : Pisces) under controlled conditions . Journal <strong>of</strong> Zoology, 214, 21-44 .<br />
Hecht, T . <strong>and</strong> Pienaar, A .G . (1993) A review <strong>of</strong> <strong>cannibalism</strong> <strong>and</strong> its implications <strong>in</strong><br />
fish larviculture . Journal <strong>of</strong> the World Aquaculture Society, 241, 246-261 .<br />
Howell, B .R . (1984) The <strong>in</strong>tensive rear<strong>in</strong>g <strong>of</strong> <strong>juvenile</strong> cod, Gadus morhua L . Fladevigen<br />
Rapportserie, 1, 657-675 .<br />
Hunter, J .R . <strong>and</strong> Kimbrell, C .A . (1980a) Early life history <strong>of</strong> Pacific mackerel,<br />
Scomber japonicus . Fishery Bullet<strong>in</strong>, U.S ., 78, 89-101 .<br />
Hunter, J .R. <strong>and</strong> Kimbrell, C.A . (1980b) Egg <strong>cannibalism</strong> <strong>in</strong> the northern anchovy,<br />
Engraulis mordax . Fishery Bullet<strong>in</strong>, U.S., 78, 811-816 .<br />
Huston, M.A . <strong>and</strong> DeAngelis, D .L . (1987) Size bimodality <strong>in</strong> monospecific populations:<br />
a critical review <strong>of</strong> potential mechanisms . American Naturalist, 129, 678-<br />
707 .<br />
Katavic, I ., Jug-Dujakovic, J . <strong>and</strong> Glamuz<strong>in</strong>a, B . (1989) Cannibalism as a factor<br />
affect<strong>in</strong>g the survival <strong>of</strong> <strong>in</strong>tensively cultured sea bass (Dicentrarchus labrax) f<strong>in</strong>gerl<strong>in</strong>gs<br />
. Aquaculture, 77, 135-143 .<br />
Kawai, T . <strong>and</strong> Isibasi, K . (1983) Change <strong>in</strong> abundance <strong>and</strong> species composition <strong>of</strong><br />
neritic pelagic fish stocks <strong>in</strong> connection <strong>with</strong> <strong>larval</strong> mortality caused by <strong>cannibalism</strong><br />
<strong>and</strong> predatory loss by carnivorous plankton . FAO Fisheries<br />
1082-1111 .<br />
Report, 3,<br />
Knutsen, G .M. <strong>and</strong> Tilseth, S . (1985) Growth, development <strong>and</strong> feed<strong>in</strong>g success <strong>of</strong><br />
Atlantic cod larvae Gadus morhua related to egg size . Transactions <strong>of</strong> the American<br />
Fisheries Society, 114, 507-511 .<br />
Koeller, P .A ., Coates-Markle, L . <strong>and</strong> Neilson, J .D . (1989) Feed<strong>in</strong>g ecology <strong>of</strong> <strong>juvenile</strong><br />
(age-0) silver hake (Merluccius bil<strong>in</strong>earis) on the Scotian Shelf . Canadian Journal <strong>of</strong><br />
Fisheries <strong>and</strong> Aquatic Sciences, 46, 1762-1768 .<br />
Kraul, S ., Nelson, A ., Britta<strong>in</strong>, K . et al . (1992) Evaluation <strong>of</strong> live feed for <strong>larval</strong> <strong>and</strong><br />
post<strong>larval</strong> Mahi mahi Coryphaena hippurus . Journal <strong>of</strong> the World Aquaculture<br />
Society, 23, 299-306 .<br />
Li, S . <strong>and</strong> Mathias, J .A . (1982) Causes <strong>of</strong> high mortality among cultured <strong>larval</strong><br />
walleyes. Transactions <strong>of</strong> the American Fisheries Society, 111, 710-721 .<br />
275
276 . <strong>Ontogeny</strong> <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> <strong>larval</strong> <strong>and</strong> <strong>juvenile</strong> <strong>fishes</strong><br />
Lleonart, J., Salat, J . <strong>and</strong> MacPherson, E . (1985) CVPA, an extended VPA <strong>with</strong><br />
<strong>cannibalism</strong> . Application to a hake population . Fisheries Research, 3, 61-79 .<br />
Loekle, D .M ., Christian, J.J . <strong>and</strong> Madison, D .M . (1982) Time dependency <strong>and</strong> k<strong>in</strong><br />
recognition <strong>of</strong> cannibalistic behavior among poeciliid <strong>fishes</strong> . Behavioral <strong>and</strong><br />
Neural Biology, 35, 315-318 .<br />
MacCall, A .D . (1981) The consequences <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> the stock-recruitment<br />
relationship <strong>of</strong> planktivorous pelagic <strong>fishes</strong> such as Engraulis, <strong>in</strong> Workshop on the<br />
effects <strong>of</strong> environmental variation on the survival <strong>of</strong> <strong>larval</strong> pelagic <strong>fishes</strong> (rapporteur G .<br />
Sharp) . UNESCO Intergovernmental Oceanographic Commission, Workshop<br />
Report, 28, pp . 201-220 .<br />
McIntyre, D .B ., Ward, F .J . <strong>and</strong> Swanson, G .M. (1987) Factors affect<strong>in</strong>g <strong>cannibalism</strong><br />
by pond reared <strong>juvenile</strong> walleyes . The Progressive Fish-Culturist, 49, 264-<br />
269 .<br />
Macpherson, E, <strong>and</strong> Gordoa, A . (1994) Effect <strong>of</strong> prey densities on <strong>cannibalism</strong> <strong>in</strong><br />
Cape hake (Merluccius capensis) <strong>of</strong>f Namibia . Mar<strong>in</strong>e Biology, 119, 145-149 .<br />
van der Meeren, T. <strong>and</strong> Na?ss, T . (1993) How does cod (Gadus morhua) cope <strong>with</strong><br />
variability <strong>in</strong> feed<strong>in</strong>g conditions dur<strong>in</strong>g early <strong>larval</strong> stages? Mar<strong>in</strong>e Biology, 112,<br />
637-647 .<br />
Mehl, S . (1988) The Northeast Arctic cod stock's consumption <strong>of</strong> commercially<br />
exploited prey species <strong>in</strong> 1984-1986 . Rapports et Proces-verbaux des Reunions,<br />
Conseil International Pour ]'Exploration de la Mer, 188, 185-205 .<br />
Nellen, W . (1986) A hypothesis on the fecundity <strong>of</strong> bony fish . Meeresforschung, 31,<br />
75-89 .<br />
Nesbit, D .H . <strong>and</strong> Meffe, G .K . (1993) Cannibalism frequencies <strong>in</strong> wild populations <strong>of</strong><br />
the eastern mosquit<strong>of</strong>ish (Gambusia holbrooki, Poeciliidae) <strong>in</strong> South-Carol<strong>in</strong>a .<br />
Copeia, 1993, 867-870 .<br />
Oiestad, V . (1985) Predation on fish larvae as a regulatory force, illustrated <strong>in</strong><br />
mesocosm studies <strong>with</strong> large groups <strong>of</strong> larvae . NAFO Science Council Studies, 8,<br />
25-32 .<br />
Olsen, S . <strong>and</strong> Soldal, A .V . (1989) Observations on <strong>in</strong>shore distribution <strong>and</strong><br />
behaviour <strong>of</strong> 0-group northeast Arctic cod . Rapports et Proces-verbaux des<br />
Reunions, Conseil International Pour ]'Exploration de la Mer, 191, 296-302 .<br />
Otters, H . (1992) Bias <strong>in</strong> calculat<strong>in</strong>g growth rates <strong>in</strong> cod (Gadus morhua L.) due to<br />
size selective growth <strong>and</strong> mortality . Journal <strong>of</strong> Fish Biology, 40, 465-467 .<br />
Otters, H . <strong>and</strong> Folkvord, A . (1993) Allometric growth <strong>in</strong> <strong>juvenile</strong> cod (Gadus morhua<br />
L .) <strong>and</strong> possible effects on <strong>cannibalism</strong> . Journal <strong>of</strong> Fish Biology, 43, 643-645 .<br />
Otters, H . <strong>and</strong> Lie, 0 . (1991) Wean<strong>in</strong>g trials <strong>with</strong> cod (Gadus morhua L.) fry on<br />
formulated diets . Fiskeridirektoratets Skrifter, Serie Ernar<strong>in</strong>g, IV, 85-94 .<br />
Otterlei,-E ., Folkvord, A . <strong>and</strong> Moller, D . (1994) Effects <strong>of</strong> temperature <strong>and</strong> density on<br />
growth, survival, <strong>and</strong> <strong>cannibalism</strong> <strong>of</strong> <strong>juvenile</strong> cod (Gadus morhua L .) . ICES Mar<strong>in</strong>e<br />
Science Symposia, 198, 632 .<br />
Parazo, M.M., Avila, E .M . <strong>and</strong> Reyes, D .M. Jun . (1991) Size- <strong>and</strong> weight dependent<br />
<strong>cannibalism</strong> <strong>in</strong> hatchery-bred sea bass (Lates calcarifer Bloch) . Journal <strong>of</strong> Applied<br />
Ichthyology, 7, 1-7 .<br />
Patriqu<strong>in</strong>, D.G . (1967) Biology <strong>of</strong> Gadus morhua <strong>in</strong> Ogac Lake, a l<strong>and</strong>locked fjord on<br />
Baff<strong>in</strong> Isl<strong>and</strong>. Journal <strong>of</strong> the Fisheries Research Board <strong>of</strong> Canada, 24, 2573-2594 .<br />
Pedersen, T . <strong>and</strong> Falk-Petersen, I .B . (1992) Morphological changes dur<strong>in</strong>g<br />
metamorphosis <strong>in</strong> cod (Gadus morhua L .), <strong>with</strong> <strong>special</strong> reference to the development<br />
<strong>of</strong> the stomach <strong>and</strong> pyloric caeca . Journal <strong>of</strong> Fish Biology, 41, 449-461 .<br />
Pep<strong>in</strong>, P . (1989) Predation <strong>and</strong> starvation <strong>of</strong> <strong>larval</strong> fish : a numerical experiment <strong>of</strong><br />
size- <strong>and</strong> growth-dependent survival . Biological Oceanography, 6, 23-44.<br />
References<br />
Perry, R.I . <strong>and</strong> Neilson, J .D. (1988) Vertical distributions <strong>and</strong> trophic <strong>in</strong>teractions <strong>of</strong><br />
age-0 Atlantic cod <strong>and</strong> haddock <strong>in</strong> mixed <strong>and</strong> stratified waters <strong>of</strong> Georges Bank .<br />
Mar<strong>in</strong>e Ecology Progress Series, 49, 199-214 .<br />
Peterman, R .M . (1991) Density-dependent mar<strong>in</strong>e processes <strong>in</strong> North-Pacific<br />
salmonids : lessons for experimental design <strong>of</strong> large-scale manipulations <strong>of</strong> fish<br />
stocks . ICES Mar<strong>in</strong>e Science Symposia, 192, 69-77 .<br />
Pfennig, D.W. <strong>and</strong> Coll<strong>in</strong>s, J .P . (1993) K<strong>in</strong>ship affects morphogenesis <strong>in</strong> cannibalistic<br />
salam<strong>and</strong>ers . Nature, 362, 836-838 .<br />
Pfenn<strong>in</strong>g, D.W ., Reeve, H .K . <strong>and</strong> Sherman, P.W. (1993) K<strong>in</strong> recognition <strong>and</strong> <strong>cannibalism</strong><br />
<strong>in</strong> spadefoot toad tadpoles . Animal Behaviour, 46, 87-94 .<br />
Pitcher, T .J. <strong>and</strong> Parrish, J .K . (1993) Functions <strong>of</strong> shoal<strong>in</strong>g behaviour <strong>in</strong> teleosts, <strong>in</strong><br />
Behaviour <strong>of</strong> Teleost Fishes, 2nd edn (ed . T .J . Pitcher), London, Chapman <strong>and</strong> Hall,<br />
pp . 363-439 .<br />
Polls, G .A . (1981) The evolution <strong>and</strong> dynamics <strong>of</strong> <strong>in</strong>traspecific predation .<br />
Review <strong>of</strong> Ecology <strong>and</strong> Systematics, 12, 225-251 .<br />
Annual<br />
Polis, G .A . (1988) Exploitation competition <strong>and</strong> the evolution <strong>of</strong> <strong>in</strong>terference, <strong>cannibalism</strong>,<br />
<strong>and</strong> <strong>in</strong>traguild predation <strong>in</strong> age/size-structured populations, <strong>in</strong> Size-structured<br />
Populations (eds B. Ebenman <strong>and</strong> L . Persson), Berl<strong>in</strong>, Spr<strong>in</strong>ger-Verlag, pp .<br />
185-202 .<br />
Polis, G .A . <strong>and</strong> Myers, C .A . (1985) A survey <strong>of</strong> <strong>in</strong>traspecific predation among<br />
reptiles <strong>and</strong> amphibians. Journal <strong>of</strong> Herpetology, 19, 99-107 .<br />
Post, J .R . (1990) Metabolic allometry <strong>of</strong> <strong>larval</strong> <strong>and</strong> <strong>juvenile</strong> yellow perch (Perca<br />
flavescens) : <strong>in</strong> situ estimates <strong>and</strong> bioenergetic models . Canadian Journal <strong>of</strong> Fisheries<br />
<strong>and</strong> Aquatic Sciences, 47, 554-560 .<br />
Punt, A.E. <strong>and</strong> Hilborn, R . (1994) A comparison <strong>of</strong> fishery models <strong>with</strong> <strong>and</strong> <strong>with</strong>out<br />
<strong>cannibalism</strong> <strong>with</strong> implications for the management <strong>of</strong> the Cape hake resource <strong>of</strong>f<br />
southern Africa . ICES Journal <strong>of</strong> Mar<strong>in</strong>e Science, 51, 19-29 .<br />
Rice, J .A ., Miller, T .J ., Rose, K .A . et al . (1993) Growth rate variation <strong>and</strong> <strong>larval</strong><br />
survival : <strong>in</strong>ferences from an <strong>in</strong>dividual based size-dependent predation model .<br />
Canadian Journal <strong>of</strong> Fisheries <strong>and</strong> Aquatic Sciences, 50, 133-142 .<br />
Ricker, W .E . (1975) Computation <strong>and</strong> <strong>in</strong>terpretation <strong>of</strong> biological statistics <strong>of</strong><br />
fish populations . Bullet<strong>in</strong> <strong>of</strong> the Fisheries Research Board <strong>of</strong> Canada, 191, 1-<br />
382 .<br />
Riley, J .D . <strong>and</strong> Parnell, W .G . (1984) The distribution <strong>of</strong> young cod . Flodevigen<br />
Rapportserie, 1, 563-580 .<br />
Rosenberg, A .A . <strong>and</strong> Haugen, A .S . (1982) Individual growth <strong>and</strong> size-selective<br />
mortality <strong>of</strong> <strong>larval</strong> turbot,<br />
Biology, 72, 73-77 .<br />
Scophthalmus maximus reared <strong>in</strong> enclosures . Mar<strong>in</strong>e<br />
Rett<strong>in</strong>gen, I . (1990) A review <strong>of</strong> variability <strong>in</strong> the distribution <strong>and</strong> abundance <strong>of</strong><br />
Norwegian spr<strong>in</strong>g spawn<strong>in</strong>g herr<strong>in</strong>g <strong>and</strong> Barents Sea capel<strong>in</strong> . Polar Research, 8,<br />
33-42 .<br />
Salvanes, A .G .V ., Giske, J. <strong>and</strong> Nordeide, J .T . (1994) Life-history approach to habitat<br />
shifts for coastal cod, Gadus morhua L . Aquaculture <strong>and</strong> Fisheries Management, 25,<br />
215-228 .<br />
S<strong>and</strong>lund, O .T ., Gunnarsson, K ., Jonasson, P.M. et al . (1992) The arctic char Salvel<strong>in</strong>us<br />
alp<strong>in</strong>us <strong>in</strong> Th<strong>in</strong>gvallavatn . Oikos, 64, 305-351 .<br />
Shirota, A . (1970) Studies on the mouth size <strong>of</strong> fish larvae . Bullet<strong>in</strong> <strong>of</strong> the Japanese<br />
Society for Scientific Fisheries, 36, 353-368 .<br />
Smestad, O.M., Foss6, J.H., Salvanes, A.G .V . <strong>and</strong> Nordeide, J.T. (1994) Enhancement<br />
<strong>of</strong> cod, Gadus morhua L ., <strong>in</strong> Masfjorden, western Norway: an overview .<br />
Aquaculture <strong>and</strong> Fisheries Management, 25, 117-128 .<br />
277
278 . <strong>Ontogeny</strong> <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> <strong>larval</strong> <strong>and</strong> <strong>juvenile</strong> <strong>fishes</strong><br />
Smith, C . <strong>and</strong> Reay, P . (1.991) Cannibalism <strong>in</strong> teleost fish . Reviews <strong>in</strong> Fish Biology<br />
<strong>and</strong> Fisheries, 1, 41-64.<br />
Sogard, S . <strong>and</strong> Olla, B .L . (1994) The potential for <strong>in</strong>tracohort <strong>cannibalism</strong> <strong>in</strong> age-0<br />
walleye pollock, Theragra chalcogramma as determ<strong>in</strong>ed under laboratory conditions<br />
. Environmental Biology <strong>of</strong> Fishes, 39, 183-190 .<br />
Suthers, I .M . <strong>and</strong> Sundby, S . (1993) Dispersal <strong>and</strong> growth <strong>of</strong> pelagic <strong>juvenile</strong> Arcto-<br />
Norwegian cod (Gadus morhua), <strong>in</strong>ferred from otolith microstructure <strong>and</strong> water<br />
temperature . ICES Journal <strong>of</strong> Mar<strong>in</strong>e Science, 50, 261-270 .<br />
Svas<strong>and</strong>, T . <strong>and</strong> Kristiansen, T .S . (1990) Enhancement studies <strong>of</strong> coastal cod <strong>in</strong><br />
western Norway . Part IV . Mortality <strong>of</strong> reared cod after release . Journal du<br />
Conseil, Conseil International Pour l'Exploration de la Mer, 47, 30-39 .<br />
Thibault, R .E . (1974) Genetics <strong>of</strong> <strong>cannibalism</strong> <strong>in</strong> a viviparous fish <strong>and</strong> its relationship<br />
to population density . Nature, 251, 130-140 .<br />
Tonn, W .M ., Holopa<strong>in</strong>en, I .J . <strong>and</strong> Paszkowski, C .A . (1994) Density-dependent effects<br />
<strong>and</strong> the regulation <strong>of</strong> crucian carp populations <strong>in</strong> s<strong>in</strong>gle-species ponds . Ecology,<br />
75,824-834 .<br />
Tsukamoto, K ., Kuwada, H ., Hirokawa, J . et al . (1989) Size-dependent mortality <strong>of</strong><br />
red sea bream, Pagrus major, <strong>juvenile</strong>s released <strong>with</strong> fluorescent otolith-tags <strong>in</strong><br />
News Bay, Japan . Journal <strong>of</strong> Fish Biology, 35, 59-69 .<br />
Ulltang, 0 . (1984) The management <strong>of</strong> cod stocks <strong>with</strong> <strong>special</strong> reference to growth<br />
<strong>and</strong> recruitment overfish<strong>in</strong>g <strong>and</strong> the question whether artificial propagation can<br />
help to solve management problems . Fledevigen Rapportserie, 1, 795-817 .<br />
Valdes Sze<strong>in</strong>feld, E .S . (1991) Cannibalism <strong>and</strong> <strong>in</strong>traguild predation <strong>in</strong> clupeoids .<br />
Mar<strong>in</strong>e Ecology Progress Series, 79, 17-26.<br />
Walls, S .C . <strong>and</strong> Roudebush, R .E . (1991) Reduced aggression toward sibl<strong>in</strong>gs as<br />
evidence <strong>of</strong> k<strong>in</strong> recognition <strong>in</strong> cannibalistic salam<strong>and</strong>ers . American Naturalist,<br />
138,1027-1038 .<br />
Wiborg, K .F . (1948) Experiments <strong>with</strong> Clarke-Bumpus plankton sampler <strong>and</strong> a<br />
plankton pump <strong>in</strong> the L<strong>of</strong>oten area <strong>in</strong> northern Norway . Fiskeridirektoratets<br />
Skrifter, Serie Havundersekelser, 9 1-23 .<br />
Wiborg, K .F . (1960) Investigations on pelagic fry <strong>of</strong> cod <strong>and</strong> haddock <strong>in</strong> coastal <strong>and</strong><br />
<strong>of</strong>fshore areas <strong>of</strong> northern Norway <strong>in</strong> July-August 1957 . Fiskeridirektoratets<br />
Skrifter, Serie Havundersekelser, 12, 1-18 .<br />
Young, J.W. <strong>and</strong> Davies, T .L .O . (1990) Feed<strong>in</strong>g ecology <strong>of</strong> larvae <strong>of</strong> southern<br />
bluef<strong>in</strong>, albacore <strong>and</strong> skipjack tunas (Pisces : Scombridae) <strong>in</strong> the eastern Indian<br />
Ocean . Mar<strong>in</strong>e Ecology Progress Series, 61, 17-29 .