CENTRAL-PLACE FORAGING IN AN URBAN LANDSCAPE: BODY ...

POLISH JOURNAL OF ECOLOGY
(Pol. J. Ecol.)
58 2 387–392 2010
Short research contribution
Justyna KUBACKA 1 *, Michał ŻMIHORSKI 2 , Paweł MIRSKI 3 , Łukasz REJT 4
1
Institute of Environmental Studies, Jagiellonian University, Gronostajowa 7, 30-387 Kraków, Poland
*e-mail: justyna.kubacka@uj.edu.pl (corresponding author)
2
Museum and Institute of Zoology, Polish Academy of Sciences, Wilcza 64, 00-679 Warszawa, Poland
3
Institute of Biology, University of Białystok, Świerkowa 20B, 15-950 Białystok, Poland
4
Museum and Institute of Zoology, Polish Academy of Sciences, Wilcza 64, 00-679 Warszawa, Poland
CENTRAL-PLACE FORAGING IN AN URBAN LANDSCAPE:
BODY MASS OF COMMON VOLES (MICROTUS ARVALIS PALL.)
CAUGHT BY BREEDING KESTRELS (FALCO TINNUNCULUS L.) IS
POSITIVELY CORRELATED WITH AVAILABILITY OF HUNTING SITES
ABSTRACT: In the reproduction period
a male Kestrel (Falco tinnunculus) is a centralplace
forager, i.e. it transports food from hunting
grounds to a central location – the nest. A centralplace
forager is predicted to take larger or more
prey when distance to a foraging site is longer.
We studied kestrels breeding in a large Central
European city (population 1.7 million), whose
main prey are common voles (Microtus arvalis).
Kestrel nests are located in the centre and the outskirts,
although common voles are scarce in the
former. The aim of our study was to analyse the
body mass of common voles found in pellets under
kestrel nests and relate it to the availability of
common vole habitats within 1 km from the nests,
controlled for vole frequency in the pellets. We
assumed that the greater availability of common
vole habitats, the shorter the distance to a foraging
site. We found that the body mass of common
voles found in pellets was significantly positively
correlated with the availability of their habitats,
but was not affected by their frequency in pellets.
Our results may indicate that, contrary to the central-place
foraging rule, and irrespectively of the
amount of other prey taken, the kestrels hunted
smaller voles when foraging grounds were further
away. This might stem from decreased selectivity
caused by competition, either in the native territory
(due to the high density of kestrels in the centre)
or in territories of outskirt kestrels, invaded
by city centre kestrels. On the other hand, due to
lack of data on the body size of common voles in
our study area, the results may suggest that common
voles were on average smaller in the centre
than in the outskirts. Although the published data
do not support the second explanation, more research
is needed to verify this.
KEY WORDS: prey size selection, competition,
urban environment
In the breeding season, birds make repeated
excursions from and back to a central location,
such as a nest, to which they carry food.
This is an instance of central-place foraging
(Orians and Pearson 1979). It has been
shown that the choice of prey type and size by
central-place foragers is affected by distance
to foraging grounds, competition, demand at
the delivery point (determined by e.g. brood
size) and variance-sensitivity (Ydenberg
2007). For example, many studies found that
when distance to foraging grounds increased,
so did food load transported by a centralplace
forager (e.g. Kacelnik 1984, Krebs
and Aver y 1985, Gil and Pleguezuelos
2001). Also, in Eastern Chipmunks Tamias
stratus load size decreased in the presence
of competitors at resource collection points
(Giraldeau et al. 1994).
We studied central-place foraging patterns
in the urban population of European
journal 22 v02.indb 387 2010-06-18 20:20:40

388
Justyna Kubacka et al.
Kestrels Falco tinnunculus (L.) in Warsaw,
Poland. These raptors specialise in microtine
voles, especially in Central and Northern Europe
(e.g. Korpimäki 1985, Village 1990).
The Warsaw kestrels, although urban dwellers,
have also been confirmed to be vole specialists,
feeding predominantly on Common
Voles Microtus arvalis (Pallas), which comprise
approximately 70–80% of their prey as
estimated from pellets (Romanowski 1996,
Rejt 2001, Ż mihorski and Rejt 2007). This
prey occurs in the outskirts but is thought to
be rare or absent in the city centre, where
its habitat is scarce and patchy, and which is
dominated by synurbic rodent species, such
as the Striped Field Mouse Apodemus agrarius
(Andrzejewski et al. 1978, Białożej 2003,
Gryz et al. 2008, Gliwicz unpubl.). Kestrels
build nests both in the centre and outskirts of
Warsaw, but they breed especially densely in
the former, where chances for catching a vole
close to the nest are very low.
The aim of our study was to examine the
body mass of common voles caught by Warsaw
kestrels in the breeding season (March
and April), when they are central-place foragers,
and relate it to the percentage of common
vole habitats within 1 km from the nest. We
used this percentage as an indicator of probability
that a kestrel hunts further away from
its central place (the nest), i.e. the more vole
habitats within 1 km from a nest, the closer to
it a kestrel is predicted to hunt. We expected
the body mass of common voles preyed by
kestrels to be negatively correlated with the
amount of common vole habitats within 1 km
from a kestrel’s nest.
We analysed kestrel pellets gathered as
part of a larger study. The pellets were collected
in 1998, 1999 and 2001, under 11 different
kestrel nests situated within the administrative
boundaries of Warsaw, Central Poland
(52ºN, 21ºE, area 517 km 2 , population 1.7 million).
We narrowed our interest to the breeding
period, when the male kestrel becomes a
central-place forage r and is virtually the only
food supplier for the female and later also for
the young (Village 1983, Riddle 1993).
It starts feeding his partner during courtship,
which in Warsaw takes place mainly in
March, and continues over the incubation period
(3 rd decade of March – end of April) and
the young rearing period, when he supplies
food for the incubating female and later for
the whole family (Śliwa and Rejt 2006). Because
in the region of our study common voles
are born between April and May (e.g. Adamczewska-Andrzejewska
and Nabagło
1977), we decided to limit our analysis to
March and April. This was done in order to
avoid the confounding effect of change in the
age structure of common vole populations.
The nest surroundings varied in terms of
the amount of habitat suitable for the Common
Vole, such as lawns, fallow land, roadsides,
weed strips, meadows and pastures
(Stein 1958). We calculated the proportion
of these common vole habitats within a 1-km
radius from each of the nests. We assumed
that this proportion negatively correlates with
the distance that Warsaw kestrels need to cover
to find a vole: the more common vole habitats
within 1 km, the shorter the distance. The
proportion of common vole habitats ranged
from 0.7 to 19.6%. For the computations we
used GIS software and 1:10 000 maps (as of
2004), scanned with the resolution of 1 pixel
to 70 cm in the field.
Out of the 11 nest locations some were
sampled in more than one year, whereas others
only in one year. It should be noted that
our replicates are in fact males (as decisionmakers)
and not nest locations, and so we
assumed the pellet sets collected under the
same nests to be independent between years.
This is justified by the fact that kestrels typically
change partners between years (e.g.
Palokangas et al. 1992). Thus, it is unlikely
that from 1998 to 2001 a pair breeding in a
given location comprised the same male.
From the pellets we dissected all skeletal
elements, according to a standard method
(Raczyński and Ruprecht 1974). Bones of
common voles were identified to the species
level with the identification key by Pucek
(1984). Next, mandibular first molars (M1s)
were separated from common vole jaws and
their length was measured. Most pellets
contained remainings of 1 vole individual,
in which case we randomly selected one of
the two M1s for measurement. If there were
more (2 or 3) individuals in a pellet, either
left or right M1s were measured. Teeth that
showed digestion traces (i.e. were rounded)
or mechanical damage (i.e. were broken or
cracked) were discarded. Measurements were
journal 22 v02.indb 388 2010-06-18 20:20:40

Size of common vole prey in breeding urban Kestrels
389
taken with a vernier calliper, to the nearest
0.02 mm, using a 2.3 × Zeiss eyepiece. We
also used data on percentage frequency of
Microtus voles in the pellets relative to other
prey (calculated for different purposes). From
other studies on Warsaw kestrels (e.g. Rejt
2004, Ż mihorski and Rejt 2007) we know
that Microtus arvalis is dominant in their diet
and very rarely M. oeconomus and M. agrestis
are found. We therefore used percentage
of Microtus voles as an indicator of common
vole frequency in the pellets that we analysed.
It was calculated separately for each nest and
year, and was based on the average of 67.0 ±
49 pellets (min. 16, max. 177).
In the Common Vole, the length of M1 is
significantly positively correlated with body
mass (r = 0.63, P< 0.001, Balčiauskienė
2007). Because the Kestrel easily digests
bones of rodents, usually only their skulls or
jaws are present in pellets (e.g. Yalden 1980).
Consequently, M1 appears to be one of the
best indicator of body mass of common voles
identified in kestrel pellets. We converted all
the M1 measurements into body mass values,
using the linear regression formula by
Balčiauskienė (2006), and body mass was
used for further analysis.
In total, we measured 562 M1s of common
voles (Table 1). We excluded the nests
for which the number of M1 measurements
in a given year was smaller than 4. For each
of the remaining nests all M1 lengths were
averaged, separately for each year. The distribution
of individual measurements, categorised
by year, was slightly skewed to the
left, but divergence from normality was significant
only for 1999. The dependent variable
– mean body mass per nest in a given
year – did not diverge from normality. We
then applied a general linear mixed model to
test for the effect of year (random), % common
vole habitats (covariate), % voles in pellets
(covariate) and the interactions: year x %
common vole habitats, year x % voles in pellets
and % common vole habitats x % voles
in pellets on the body mass of the common
voles. The percentage variables were not
transformed, but their arc-sine transformation
yielded the same final model and results
(not shown). The dependent variable, mean
body mass of common voles per nest in a
given year, was weighted by frequencies on
which the means were based. The random
effects hypotheses in the model were tested
with the Wald Z test and the fixed effects
hypotheses with the F-test. We applied backward
selection and the AIC value (Akaike
information criterion) to reduce the number
of predictors and find the most parsimonious
model explaining variation in body mass
(Quinn and Keough 2002). The lowest
AIC model is considered to explain the most
variation in the response variable with the
lowest number of independent variables. We
used the corrected AIC value (AIC c
), which
should be applied when sample sizes are
small, as in our case (N = 19). For statistical
analysis we used SAS 9.1 (with Enterprise
Guide 4.1) and Statistica 8.0.
The ΔAIC C value of the initial model
was 21.6 and all the factors and interactions
were non-significant. Backward elimination
of the least significant predictors (beginning
from interactions) always resulted in decreasing
the AIC C . After removal of all the three
interactions from the model (P> 0.150 in all
cases, not shown) the common vole frequency
remained non-significant (P = 0.550, not
shown), and consequently was dropped from
the model as well. The most parsimonious
model (with ΔAIC C = 0) retained ‘year’ and
‘% common vole habitats’ (Table 2).
The amount of common vole habitats
within 1 km from the nest of the kestrels was
significantly positively correlated with the
body mass of common voles caught by the
kestrels (F (0.05;1,15)
= 6.98, P = 0.018). The model
predicts that the kestrels with the highest
percentage of common vole hunting grounds
Table 1. Number (No) of kestrel nests sampled in each year of the study out of the total of 11 nests, and
the mean number of measurements of mandibular first molars of the common voles found in the pellets
collected under the nests (±SD).
year No of nests M1s/nest
1998 7 23.3 ± 14.2
1999 5 28.0 ± 21.3
2001 7 37.0 ± 29.6
journal 22 v02.indb 389 2010-06-18 20:20:40

390
Justyna Kubacka et al.
Table 2. Results from the most parsimonious general linear mixed model, testing for the effect of year
and % common vole habitats within 1 km radius from the nests of kestrels Falco tinnunculus on the
body mass of common voles Microtus arvalis caught by the kestrels at the beginning of the breeding
season (March–April), in the years 1998–1999 and 2001 in Warsaw.
Source of variation statistic p-value
Random effects
Z-value
year 0.93 0.176
residual 2.74 0.003
Fixed effect
F-value
% common voles habitats 6.98 0.018
Fig. 1. Relationship between percent common vole (Microtus arvalis) habitats within 1-km radius from
the nests of urban kestrels (Falco tinnunculus) and the body mass of common voles found in pellets collected
under the nests in 1998–1999 and 2001, in Warsaw, Poland. Each data point corresponds to one
nest in a given year. Symbol size denotes number of measurements per each nest in a given year.
(19.6%) on average catch voles that are 1.8
g heavier than in the case of the birds with
the lowest ranking hunting grounds (0.7%)
(Fig. 1). The body mass of common vole prey
captured by the kestrels did not differ significantly
between the years (P = 0.176). The mean
frequency of Microtus voles in the pellets was
85 ± SD 14%, and the range was 58 to 100%.
Contrary to the expectations, our results
showed that the body mass of common voles
caught by urban kestrels increased with the
amount of common vole habitats within 1 km
from the nest (Table 2, Fig.1). Importantly,
both the percent frequency of voles in the pellets
(of which the Common Vole was dominant)
and its interaction with the percent
common vole habitats were non-significant.
These results may suggest that kestrels, when
central-place foragers, are less selective when
opportunities to find a vole close to their
central location are worse, irrespectively of
the relative number of alternative prey they
consume. One of the causes of decreased selectivity
in central foragers may be intraspecific
competition, as observed in the Eastern
Chipmunk (Giraldeau et al. 1994). In Warsaw,
kestrels reach high densities (approx. 1.2
pairs per 10 km 2 ), and nest especially densely
in the city centre (approx. 25.7 pairs per 10
km 2 ) (Luniak et al. 2001), which offers extremely
limited common vole resources. It
follows that kestrels occupying territories
journal 22 v02.indb 390 2010-06-18 20:20:40

Size of common vole prey in breeding urban Kestrels
391
with small amount of common vole habitats
must suffer increased intraspecific competition
for their main prey and consequently
may be less selective towards the size of
this prey. It is also possible that the kestrels
which nest in territories poorer in common
vole habitats visit more distant common vole
patches. This is consistent with observations
of Riegert et al. (2007) in the Czech Republic,
who found that territories of city-centre
kestrel males, with scarcer vole habitats, were
larger than and overlapped with territories of
outskirts kestrels. This suggests that city centre
kestrels may forage in more distant hunting
grounds, which are occupied by other
kestrels. In our study kestrels with fewer common
vole habitats in their territories may have
invaded richer territories of other kestrels,
which eventually may have lead to enhanced
competition between ‘intruders’ and ‘owners’
of a territory. It should be emphasized that
since in the breeding season the male kestrel
is a central-place forager, it should carry
larger prey when it covers longer distance to
hunting grounds (e.g. Ydenberg 2007). Our
results suggest an opposite pattern, pointing
out that other factors, such as competition,
may interact with this optimal foraging rule.
This result, to our knowledge, has only been
documented in the case of mammalian foragers,
Eastern Chipmunks.
On the other hand, our results might be
confounded by variation in body mass in the
common vole populations in our study area,
for which we did not control. First, it is probable
that in the different kestrel hunting territories
the population dynamics of the Common
Vole was asynchronous (e.g. in some
locations voles started breeding earlier than in
others, and the mean body mass in their populations
was lower). It has been shown, however,
that common vole populations inhabiting
areas that are relatively close to each other
may be expected to breed at the same time in
a given year (e.g. Adamczewska-Andrzejewska
and Nabagło 1977, Mackin-Rogalska
and Nabagło 1990). Second, since
the potential vole habitats located in the centre
are smaller and patchier, which implies
that they are poorer and more vulnerable to
disturbance, common voles may be expected
to be smaller in the centre and larger in the
outskirts. However, Jacob (2003) reported
that in more disturbed agricultural land (crop
field, pasture) body weight of common voles
was higher than in less disturbed areas. This
suggests that common voles may not necessarily
be smaller in the city centre. Third, it is
also possible that common vole densities vary
between the centre and the outskirts, and that
there are density-dependent effects on body
mass in these populations (Chitty 1952).
Finally, due to the low amount of suitable
habitats the city centre common voles may be
subordinate, and hence smaller individuals
forced to the sub-optimal area of the city. In
general, we cannot conclusively say that the
pattern that we found stems from selectivity
of the kestrels and not differences in common
vole populations between the hunting sites.
In sum, we observed that the amount of
vole habitats within 1 km from the nests of
kestrels is positively correlated with the size
of their main prey – the Common Vole. We
propose that our result may suggest a relationship
between availability of hunting sites
and selectivity in the kestrels, which may be
decreased by competition either in a bird’s
territory (given the very high kestrel density
in Warsaw) and/or in territories of other individuals,
visited by the kestrels whose territories
are scarcer in voles. In the latter case,
it is plausible that competition may alter the
effect of distance to foraging grounds on prey
size selection by kestrels. It should be stressed
that we did not control for the abundance or
body mass structure of common vole populations
at the nest locations which we studied,
and hence it is also possible that our results
stem from these factors and not selectiveness
of the kestrels. Although literature data does
not lend support for this explanation, more
research is needed to clarify this.
ACKNOWLEDGMENTS: We thank R.
Martyka, T. Mazgajski and anonymous reviewers
for valuable comments on a draft of this paper. M.
Cichoń and P. Chylarecki provided statistical advice.
REFERENCES
Adamczewska-Andrzejewska K.A.,
Nabagło L. 1977 – Demographic Parameters
and Variations in Numbers of the Common
Vole – Acta Theriol. 22: 431–457.
journal 22 v02.indb 391 2010-06-18 20:20:41

392
Justyna Kubacka et al.
Andrzejewski R., Babińska-Werka J.,
Gliwicz J., Goszczyński J. 1978 – Synurbization
process in population of Apodemus
agrarius. I. Characteristics of populations in
an urbanization gradient – Acta Theriol. 23:
341–358.
Balčiauskienė L. 2006 – Feeding ecology of
Tawny Owl (Strix aluco) and Long-eared Owl
(Asio otus) based on craniometry of prey –
Ph.D. thesis, Institute of Ecology, Vilnius University.
Balčiauskienė L. 2007 – Cranial growth of
captive bred Common Voles Microtus arvalis
– Acta Zool. Lith. 17: 220–227.
Białożej M. 2003 – Wpływ struktury przestrzennej
terenów kolejowych na populację
małych ssaków [Influence of spatial structure
of railway areas on the small mammal community]
– MSc thesis, Department of Ecology,
University of Warsaw (in Polish).
Chitty D. 1952 – Mortality among voles (Microtus
agrestis) at Lake Vymwy, Montgomeryshire
in 1936-9 – Phil. Trans. R. Soc. Lond.
B, 36:505–552.
Gil J.M., Pleguezuelos J. M. 2001 – Prey
and prey-size selection by the short-toad eagle
(Circaetus gallicus) during the breeding season
in Granada (South-Eastern Spain) – J. Zool.
255: 131–137.
Giraldeau L.A., Kramer D.L., Deslandes
I., Lair H. 1994 – The effect of competitors
and distance on central place foraging in Eastern
Chipmunks, Tamias striatus – Anim. Behav.
47: 621–632.
Gryz J., Krauze D., Goszczyński J. 2008
– The small mammals of Warsaw as inferred
from Tawny Owl (Strix aluco) pellet analyses –
Ann. Zool. Fenn. 45: 281–285.
Jacob J. 2003 – Body weight dynamics of common
voles in agro-ecosystems – Mammalia,
67: 559–566.
Kacelnik A. 1984 – Central place foraging in
Starlings (Sturnus vulgaris). I. Patch residence
time – J. Anim. Ecol. 53: 283–99.
Korpimäki E. 1985 – Diet of the Kestrel Falco
tinnunculus in the breeding season – Ornis
Fennica, 62: 130–137.
Krebs J.R., Avery M.I. 1985 – Central place
foraging in the European Bee-Eater (Merops
apiaster) – J. Avian Ecol. 54: 459–472.
Luniak M., Kozłowski P., Nowicki W.,
Plit J. 2001 – Ptaki Warszawy 1962–2000
[Birds of Warsaw 1962–2000] - Atlas Warszawy
8 – IGP i P PAN.
Mackin-Rogalska R., Nabagło L. 1990 –
Geographical variation in cyclic periodicity
and synchrony in the Common Vole, Microtus
arvalis – Oikos, 59: 343–348.
Orians G.H., Pearson N.E. 1979 – On the
theory of central place foraging (In: Analysis
of Ecological Systems, Ed: D.J. Horn) – Ohio
State University Press: Columbus, pp.154–177.
Palokangas P., Alatalo R.V., Korpimäki
E. 1992 – Female choice in the Kestrel under
different availability of mating options –
Anim. Behav. 43: 659–665.
Pucek Z. 1984 – Klucz do oznaczania ssaków
Polski [Key to identification of mammals of
Poland] – PWN, Warszawa.
Quinn G.P., Keough M.J. 2002 – Experimental
Design and Data Analysis for Biologists
– Cambridge University Press.
Raczyński J., Ruprecht A.L. 1974 – The effect
of digestion on the osteological composition
of owl pellets – Acta Ornithol. 14: 25–38.
Rejt Ł. 2001 – Peregrine falcon and Kestrel in
urban environment - the case study of Warsaw
(In: Naturschutz und Verhalten, Eds: E. Gottschalk,
A. Barkow, M. Muehlenberg, J. Settele)
– UFZ-Bericht 2: 81–85.
Riddle G.S. 1993 – A 20 year study of the Kestrels
in Ayrshire – Scottish Birds, 17: 9–13.
Riegert J., Fainová D., Mikeš V., Fuchs
R . 2007 – How urban Kestrels Falco tinnunculus
divide their hunting grounds: partitioning
or cohabitation? – Acta Ornithol. 42: 69–76.
Romanowski J. 1996 – On the diet of urban
Kestrels (Falco tinnunculus) in Warsaw – Buteo,
8: 123–130.
Stein G.H.W. 1958 – Die Feldmaus (Microtus
arvalis Pallas) (In: Die Neue Brehm-bücherei)
H. 225 – Wittenberg-Lutherstadt, A. Ziemsen-
Verl.
Śliwa P., Rejt Ł. 2006 – Pustułka [The Kestrel]
– Monografie przyrodnicze, vol. 15, Klub
Przyrodników Press, Świebodzin (in Polish
with English and German summary).
Village A. 1983 – Seasonal changes in the hunting
behaviour of Kestrel – Ardea, 71: 117–123.
Village A. 1990 – The Kestrel – T and A D Poyser,
London.
Yalden D.W. 1980 – Notes on the diet of urban
Kestrels – Bird Study, 27: 235–238.
Ydenberg R.C. 2007 – Provisioning (In: Foraging
– Behavior and Ecology, Eds: D.W. Stephens,
J. Brown, R. Ydenberg) – The University
of Chicago Press, Chicago and London,
pp. 273–304.
Żmihorski M., Rejt Ł. 2007 – Weather-dependent
variation in the cold-season diet of
urban Kestrels Falco tinnunculus – Acta Ornithol.
42: 107–113.
Received after revision August 2009
journal 22 v02.indb 392 2010-06-18 20:20:41