Behavioural Processes 108 (2014) 142–150
Contents lists available at ScienceDirect
Behavioural Processes
journal homepage: www.elsevier.com/locate/behavproc
Skylarks trade size and energy content in weed seeds to maximize
total ingested lipid biomass
Sabrina Gaba a,∗,1 , Claire Collas a,b,1 , Thibaut Polwony c,d , François Bretagnolle e,f ,
Vincent Bretagnolle c,g
a
INRA, UMR1347 Agroécologie, 17 rue Sully, F-21065 Dijon Cedex, France
INRA, UMR1213 Herbivores, Centre de Clermont-Ferrand/Theix, 63122 Saint-Genès-Champanelle, France
c
CEBC, UMR 7372 CNRS – Université de La Rochelle, F-79360 Villiers enBois, France
d
Office National de la Chasse et de la Faune Sauvage, Carrefour de la Canauderie, 79360 Villiers en bois, France
e
Université de Bourgogne, UMR6282 Biogéosciences, 6 Boulevard Gabriel, 21000 Dijon, France
f
UMR 5175 CEFE-CNRS, 1919 Route de Mende, 34293 Montpellier 5, France
g
LTER « ZA Plaine & Val de Sèvre », CNRS-CEBC, F-79360, Beauvoir sur Niort, France
b
a r t i c l e
i n f o
Article history:
Received 25 March 2014
Received in revised form 8 September 2014
Accepted 17 October 2014
Available online 28 October 2014
Keywords:
Optimal foraging
Weed
Seed size
Lipid content
Trade-offs
Skylark
a b s t r a c t
The trade-off between forage quality and quantity has been particularly studied in herbivore organisms,
but much less for seed eating animals, in particular seed-eating birds which constitute the bulk of wintering passerines in European farmlands. The skylark is one of the commonest farmland birds in winter,
mainly feeding on seeds. We focus on weed seeds for conservation and management purposes. Weed
seeds form the bulk of the diet of skylarks during winter period, and although this is still a matter for
discussion, weed seed predation by granivorous has been suggested as an alternative to herbicides used
to regulate weed populations in arable crops. Our objectives were to identify whether weed seed traits
govern foraging decisions of skylarks, and to characterize key seed traits with respect to size, which is
related to searching and handling time, and lipid content, which is essential for migratory birds. We
combined a single-offer experiment and a multiple-offer one to test for feeding preferences of the birds
by estimating seed intake on weed seed species differing in their seed size and seed lipid content. Our
results showed (1) a selective preference for smaller seeds above a threshold of seed size or seed size
difference in the pair and, (2) a significant effect of seed lipid biomass suggesting a trade-off between
foraging for smaller seeds and selecting seeds rich in lipids. Skylarks foraging decision thus seems to be
mainly based on seed size, that is presumably a ‘proxy’ for weed seed energy content. However, there
are clearly many possible combinations of morphological and physiological traits that must play crucial
role in the plant–bird interaction such as toxic compound or seed coat.
© 2014 Elsevier B.V. All rights reserved.
1. Introduction
Following the optimal foraging theory, predators should maximize the consumption of prey with highest energetic content while
minimizing searching and handling times to reduce both predation
risk and the cost of prospecting (Krebs, 1980). Hence, consumers
should selectively forage for preys according to their size and
energy in order to maximize energy intake rate (Sutherland, 1996).
Seed-eating birds are among the most pervasive seed predators
and numerous factors are involved in seed selection by birds. Some
∗ Corresponding author. Tel.: +33 0 3 80 69 31 87; fax: +33 0 3 80 69 32 62.
E-mail address: [email protected] (S. Gaba).
1
These authors contributed equally to this article.
http://dx.doi.org/10.1016/j.beproc.2014.10.004
0376-6357/© 2014 Elsevier B.V. All rights reserved.
studies report positive associations between the proportion of
seeds eaten by birds and the seeds’ content of energy (Carrillo et al.,
2007; Glück, 1985), lipids (Greig-Smith and Wilson, 1985), protein
(Valera et al., 2005), carbohydrates (Kelrick et al., 1986; Ríos et al.,
2012a), and water (Carrillo et al., 2007). However, the role of seed
size and nutritional quality on foraging intensity and seed selection
by birds remains to be established. In particular, whether seed traits
govern, or not, the foraging decision by granivorous birds is not
known, though this is a prerequisite for any attempt to characterize
the relationships between granivorous birds and seeds given that
their effects on the dynamics and distribution of plant populations
are poorly documented (Marone et al., 2008). On one hand, small
seed might be selected due to ease of handling and high encounter
probability. On the other, large seeds can also be attractive for granivores due to their high nutritive and energetic values which may
S. Gaba et al. / Behavioural Processes 108 (2014) 142–150
compensate for the cost of transport and consumption time (CelisDiez et al., 2004; Martínez et al., 2006). Seed selection in birds can be
heavily conditioned by the seeds’ chemical composition combined
with the animal’s physiological capacities to process nutrients and
toxins (e.g. Ríos et al., 2012a). Among the nutritional compounds,
lipids are a fundamental dietary component for granivorous birds in
winter because they constitute fatty supplies that are mobilized to
resist cold temperatures (Blem, 1990; Krams et al., 2010). Moreover, seeds with high oil content may provide a more valuable
source of metabolic water for birds (Frank, 1988; Carrillo et al.,
2007). Lipids are also essential for migratory birds since they can
rapidly mobilize and oxidize them during physical efforts (McCue
et al., 2009). However, other constraints may act on seed selection
by granivores, such as secondary metabolites (Díaz, 1996; Kelrick
et al., 1986; Ríos et al., 2012b). Plants store secondary metabolites
in their seeds, some of which can be toxic or reduce digestibility
for granivores (Anhalzer et al., 2010; Kerley and Erasmus, 1991;
Ramírez and Traveset, 2010). Overall therefore, the traits under
selection, as well as the proxy granivorous birds may use to select
seeds with regard to energetic inputs, remain a puzzle.
The aim of the present study was to analyze seed preferences
of the skylark (Alauda arvensis), a farmland specialist passerine
bird, in relation to seed trait variation. This species was selected
because it is one of the commonest farmland birds in winter, feeding mainly on seeds (Robinson, 2004), and therefore skylarks can be
regarded as one of the main seed predators in European farmland
landscapes in winter (Boatman et al., 2010; Donald, 2004). Here
we use an experimental approach to identify seed trait candidates
that may be used by skylarks to select their seed food. Our experiments particularly focused on the possible trade-off between seed
oil content and seed size which was quantified by seed mass as
they are positively correlated. We selected weed seeds which form
the bulk of the diet of skylarks during the critical wintering period
(Geiger et al., 2014) and because weed seed predation is suggested
as an alternative to herbicides used to regulate weed populations
in arable crops (Westerman et al., 2008), although this is still a matter for discussion. Available weed seeds (more than 200 species in
typical winter wheat stubbles in, e.g., our study site: Gaba et al.,
2010; Meiss et al., 2010) are highly variable with regard to size and
lipid content, both traits strongly influencing seed nutritive value.
Our experimental design consisted in disentangling the two traits,
and choosing weed species in trials so that these two traits were
uncorrelated. First, we test whether skylarks seed use is influenced
by seed size. We predicted that skylarks will prefer small seeds
since they require less energy to be eaten and digested than larger
seeds (Thompson et al., 1987). Second, we test whether skylarks
seed selection is influenced by seed lipid content. We predicted
that skylarks will prefer high lipid content seeds since they would
have a higher caloric content than high protein or high carbohydrates seeds (Celis-Diez et al., 2004; Anhalzer et al., 2010). Third,
by using multiple-offer tests with pairs of seeds, we test whether
skylarks trade mass against lipid content, and which trait, if any, is
preferred in the trade-off.
143
2. Materials and methods
were maintained from their capture to the next morning in opaque
aviaries (60 cm × 50 cm × 50 cm) on the back of our field truck, protected from weather or light. During the capture session, nobody
was left in the truck to limit loud noises or disturbance. They were
then transported to the laboratory which was between 5 and 8 km.
Thus, the transport times were approximately 10 min. The 105 birds
were randomly assigned to 10 groups of 10 individuals and acclimatized for 3 months in 4 m × 3 m × 2 m outdoor aviaries, before
the start of the experiment. Each cage contained a maximum of
five birds. Each aviary contains two different shelters of 2 m2 each.
Birds were fed ad libitum with a commercial seed mix, grit,
oilseed rape and tap water (Powolny et al., 2012). Food and water
were respectively dispensed on three synthetic green turfs (height:
1 cm; density: 12 blades/cm2 ) and three cup of water (1l each)
scattered at different spots in the outdoors aviaries changed daily.
Aviaries were placed c. 100 m from any building, in the lab centre, in remote grassland. They were not disturbed since nobody
had access to the aviaries. Despite that skylark is a terrestrial
species (feeding only on the ground), perches were placed in each
aviaries. Only three dead birds were observed (from the 105) during
the period in which birds were maintained under captive conditions (from October 2010 to early March 2011, i.e. around the
start of spring migration). Therefore the observed mortality rate
of c. 3% was much lower than natural winter mortality in this
species, which is about 40% (Daunicht, 1998). During the captive
period, veterinary checks were performed twice and each bird was
weighed weekly. No birds lost weight during the winter, indicating that subordinate birds (if any) did not suffer from competition.
After several observations and body mass measures, seven birds
showed signs of weakness, i.e. a decrease in their body mass.
These birds were isolated from others and each bird was individually treated with Baycox (Bayer Company) against coccidiose by
a veterinarian. After a controlling period of two weeks following
treatment, birds were weighed and released safe into the wild,
under clear weather, in the same place where they were caught.
Only birds that did not show any sign of weakness were used for the
experiments.
Birds were colour ringed, weighed (±0.1 g), measured (maximum wing chord, tarsus and beak) and the sex of each individual
was determined by genetic analysis using blood samples (200 ␮l)
from the brachial vein (Eraud et al., 2006). According to the
licence/permit, all birds were returned to the wild after the
study period in March 2011 during the pre-nuptial migration.
Before being release, skylarks were checked by a veterinarian
and the body mass of 90% of the birds was higher after the
period in captivity. This work was performed with governmental authorizations from the Préfecture des Deux-Sèvres (Niort,
France, No. 79-219). All experiments were carried out in compliance with French legal requirements and with the permission
of the National Conservation Authority (no. 79349). Bird capture and blood sampling were performed under permit from the
National Hunting and Wildlife Agency to TP (No. 2009-014). Animal experiments were in agreement with the guidelines for the
treatment and use of animals in behavioural research and teaching as published by the Association of Animal Behaviour (ASAB
2012).
2.1. Capture and housing conditions of skylarks
2.2. General experimental design
A total of 105 (55 males and 50 females), wild skylarks
were trapped by mist-nets along the French Atlantic coast during post-nuptial nocturnal migration, in five trapping sessions
(October–November 2010). During the nights of capture, nets were
deployed from 6 pm to 5 am the next morning and, each net was
checked every 15 min. Accordingly, birds did not stay more than
15 min (maximum) before they were removed from the net. Birds
Experiments started on February 7th and ended on March 9th
2011 at the Centre d’Etudes Biologiques de Chizé (West of France).
Birds were weighed the evening before each trial and placed into
an outdoor aviary without any food until the beginning of the
trial the next morning to avoid satiety. We used both multipleand simple-offer feeding experiments to detect food preferences
by skylarks as recommended by Cueto et al. (2006). Trials were
144
S. Gaba et al. / Behavioural Processes 108 (2014) 142–150
Table 1
List of the 10 weed species used for the multiple-offer experiments including the four species used for the simple-offer experiments which are indicated by *.
Pairs
Effects between seed
mass and lipid content
Species
Family
Seed mass1
(in mg)
1
Crossed
Chenopodium
album*
Chenopodium
polyspermum*
Fumaria
officinalis
Papaver rhoeas
Geranium molle
Geranium
robertianum
Polygonum
aviculare*
Polygonum
lapathifolium*
Stellaria
holostea
Stellaria media
Papaver rhoeas
Stellaria media
Chenopodiaceae
0.6
9.44
22,138
Chenopodiaceae
0.3
15.223
24,011
Papaveraceae
3.17
29.563
Papaveraceae
Geraniaceae
Geraniaceae
0.2
1.09
1.5
1
Polygonaceae
Polygonaceae
2
Crossed
3
Additive
4
Additive
5
Crossed
6
Crossed
Lipid content
(in %)
Energy3 (in
J/g)
Seed mass
difference (in log)
Seed Lipid content
difference (in log)
0.53
0.27
26,882
0.32
0.24
43.47
19.533
24.824
28,289
27,560
19,487
2.76
0.39
1.3
4.24
20,651
1.73
0.84
2.2
5.54
Caryophyllaceae
2.7
Caryophyllaceae
Papaveraceae
Caryophyllaceae
0.48
0.2
0.48
18,711
2
2.29
18,487
0.69
0.48
5.283
43.471
5.283
18,202
28,289
18,202
0.88
2.11
1. Royal Botanic Gardens Kew (2008). Seed Information Database (SID). Version 7.1. Available from: http://data.kew.org/sid/.
2. J. Am. Oil Chem. Soc. (2004) 81:559–561.
3. UMR1347 Agroecology, Weed-Data – a data base of weed traits.
4. Gardarin, A., Daurr, C., Colbach, N., 2011. Prediction of germination rates of weed species: relationships between germination speed parameters and species traits. Ecological
Modelling 222, 626–636.
performed from 9 am to 12 pm. All trials were carried out in a separate outdoor aviary, showing identical proportions and layout to
the housing aviaries. Each focal bird was placed in an observational
wire mesh cage (50 cm × 50 cm × 50 cm; mesh size: 1 cm × 1 cm)
enabling visual contact among birds during the trials. Theses cages
contained no perches, food or water (during the trials) and were
placed on a shaven green fitted carpet to facilitate the location of
remaining seeds without eliminating foraging time. During winter,
skylarks adopt particular aggregative strategies and sharply restrict
its foraging behaviour in the absence of conspecifics (Powolny et al.,
2012). Hence, one non-focal bird was randomly assigned to each
trial, and individually kept in identical wire mesh cages at a distance
of 15 cm from the focal individual. To avoid synchrony in behaviour
(Fernández-Juricica et al., 2004), non-focal birds were not provided
with food during tests. Group composition, i.e. the focal individual and one conspecific (non-focal bird) varied systematically from
test to test to avoid stable association between partner birds and
experimental treatments or conditions.
In multiple-offer experiments two food options were presented
simultaneously. In each trial, 100 seeds (50 per food option) were
randomly scattered on 0.25 m2 of the carpet floor, in order to reach
a density of 400 seeds/m2 which lies within the range of seed densities recorded in arable fields during winter (Moorcroft et al., 2002).
Each trial lasted 5 min. It starts either by when the first peck was
recorded, or 2 min after the introduction of the focal bird into the
cage when any pecking behaviour was recorded. At the end of each
trial, all remaining seeds were removed and the number of seeds
consumed was recorded by only one person. A total of 120 trials (20
replications for each pair) were performed. In simple-offer experiments we used the same procedure, but in each trial 100 seeds
of a single species were offered to a bird during 5 min. With this
protocol, we tested the four seeds separately using birds once or
twice in each trial but with no bird duplication for the same seed
species. Seed species presentation was randomly assigned for each
bird. A total of 80 trials (20 trials × 4 species) were performed in a
random order. 69 individuals (12 females and 57 males) of which
51 individuals were tested twice with different seed pairs were
used in multiple-offer experiments and 61 individuals (12 females
and 49 males) of which 19 birds were used twice in simple-offer
experiments.
2.3. Selection and origin of weed seed species
We chose 10 weed species based on their seed mass and lipid
content from five botanical families (see Table 1). Seed colour was
not tested, and kept constant: all seeds were dark. There was no
significant correlation between mass (mg) and lipid content (%)
of the 10 species (rs = −0.21, P = 0.56, N = 10) nor mass (mg) and
energy content (J/g) (rs = −0.32, P = 0.37, N = 10). All the seeds used
were obtained from the Herbiseed company, Twyford, Berkshire,
UK. Among the 10 species, we chose four which were found in
skylard gizzard (Eraud et al. in revision): from the smallest to the
largest, Chenopodium, Polygonum (0.3 mg and 15.22% oil content),
Chenopodium album (0.6 mg and 9.4%), Polygonum aviculare (1.3 mg
and 4.2%) and Polygonum lapathifolium (2.2 mg and 5.5%). In the
multiple-offer experiment, weed species were paired into six pairs.
In four pairs, one seed was smaller but richer (i.e. high lipid content) that the other seed (Table 1; pairs 1, 2, 5 and 6). In pairs 3
and 4, the larger seed was also the richest one (i.e. high lipid content). Therefore, pairs 1, 2, 5 and 6 showed crossed effects between
mass and lipid content while pairs 3 and 4 displayed additive
effects.
2.4. Statistical analyses
In the single-choice experiment, we compared seed consumption in the single choice trials between the four species with a
Kruskal–Wallis non-parametric test. A post hoc procedure was
applied to detect differences between pairs of species. Four
analyses were conducted with each of the four metrics. Similarly,
we compared the intake of seed biomass (i.e. the product of the
number of eaten seeds and of seed mass), lipid biomass (i.e. the
product of seed biomass and seed lipid content) and total energy
ingested (see Table 1 for seed individual values). The total energy
ingested referred to the caloric value of the seeds which determined
using an IKA VERKE C200 adiabatic oxygen bomb calorimeter (Bretagnolle F., pers. comm.).
In the multiple-offer experiment, we first explored the effects
of sex and body condition of the bird on the total seed intake (i.e.
number of seeds) per species over the trials, using generalized linear mixed model (GLMM) with Poisson error and the individual
S. Gaba et al. / Behavioural Processes 108 (2014) 142–150
bird identity as a random effect. Body condition was estimated
using the residuals of a skylark mass-size linear regression (Green,
2001; Schulte-Hostedde et al., 2005); the highest the value of
the residual the better the condition of the bird. Bird sizes were
estimated with tarsus measurements prior the experiments (see
Section 2.1).
Data of multiple-offer experiments were analyzed by Manly’s
˛ preference index (Manly, 1974) to quantify, for each replicate
of each pair, the differential intake rates between the two seed
species, and therefore test for a selection for seed number or
seed lipid biomass, allowing to explore selection guided by energy
inputs.
Following Seppäla et al. (2004), we used the equation:
˛in
or b
=
ln((ai − ci )/ai )
ln((ai − ci )/ai ) + ln((aj − cj )/aj )
where ai and aj are respectively the available quantity of seeds
of the two species (i and j) at the beginning of the experiment
(i.e. 50 seeds per species), and ci and cj the intake quantity of
seeds of these two species at the end of the experiment, i.e. after
5 min. Manly’s ˛ was tabulated either for the number of seeds (n)
or the lipid biomass intake (b). Manly’s ˛ preference index can
vary from 0 (total preference for seeds of species j) to 1 (total
preference for seeds of species i). A Manly’s ˛ preference index
of 0.5 indicates no preference. For testing the effect of seed size
on seed intake (number), Manly’s ˛ was computed with n, the
number of seeds and i, the smallest seeds (˛Sn ); while i was the
richest seeds for testing the effect of seed lipid content on seed
intake (˛Rn ; Fig. 1). Manly’s ˛ were also computed to investigate
selection for seed lipid biomass (Fig. 1). In that case, the quantity
of seeds of interest (n) was the seed lipid biomass (b) instead of
the number of seeds. Similarly to seed number, i was the smallest
seeds when testing for the effect of seed size (˛Sb ) and the richest
seeds (˛Sb ) for testing the effect of seed lipid content, respectively
(Fig. 1).
Two statistical analyses were performed with Manly’s ˛· The
first one (“overall intake”) was carried out over the 120 trials to investigate the relationship between seed consumption
(either based on number of seeds or seed lipid biomass) and
seed traits (seed size, i.e. seed mass, lipid content and energy
content). We applied four GLMM (individual bird as a random
effect) with normal error distribution to analyze bird preference
on seed intake using Manly’s ˛ computed with seed intake for
preference for seed size (˛Sn ) and for seed lipid content (˛Rn ) and
on seed lipid biomass intake using Manly’s ˛ computed with
seed lipid biomass for preference for seed size (˛Sb ) and for seed
lipid content (˛Rn ), respectively. The main effects were the differences in seed size (estimated by the differences in seed mass),
lipid and energy content between the two species of a pair, the
interaction between seed size and seed lipid content, the interaction between seed size and seed energy content and the sex
and body condition of the bird. We could not explore all the
interactions in the models because the design was unbalanced;
hence we focused on those of biological interest for this study, i.e.
interaction between seed size and nutrients (lipids and energy).
Differences between seeds within a pair were calculated as the
difference between the log values to allow for comparisons
between pairs.
The second analysis (“selective intake within a pair”) tested for
a selective predation of one of the two seeds by comparing, within
a pair, the observed values of Manly’s ˛ to a situation with no preference on seed intake (˛ = 0.5) using a two-tailed nonparametric
sign test.
All data analyses were performed using R software (2.13.0).
145
3. Results
3.1. Simple-offer experiment
Simple-offer experiment showed significant differences in the
total number of seeds consumed (hereafter seed intake) between
seed species (Kruskal–Wallis tests, X32 = 19.6, P < 0.001). Seed
intake of the lightest seeds (C. album and Chenopodium polyspermum) was significantly higher than of the larger one (Fig. 2a) with
the second lightest seed (C. album) being the most preferred. We
observed no linear relationship between seed intake and three seed
traits, i.e. mass, lipid and energy content, suggesting that no seed
character was maximized by skylarks (as revealed by the total number of seeds consumed) (Fig. 2a–c). Conversely, we found quadratic
relationships between seed intake and seed mass, and seed intake
and seed energy content.
Significant positive relationships were found between seed
intake and the intake of biomass, lipid biomass and energy biomass
(Fig. 2d–f). The higher correlation was observed between seed
intake and the total lipid biomass intake (Fig. 2e).
3.2. Multiple-offer experiment
3.2.1. Effect of the bird sex and condition
A total of 1926 seeds were consumed by 69 skylarks in the
multiple-offer experiment with in average 16.05 seeds per trial.
There was a strong effect of seed pair identity on total number of consumed seeds (GLMM, with bird identity as a random
factor, X2 = 211.38, P < 0.0001), but in addition, the sex and body
condition of the skylarks tested had significant effects (but not
their interaction: z = 0.442, P > 0.05). Seed intake was higher for
males (estimate ± SE = 0.330 ± 0.07, z = 2.017, P = 0.04) and lower
when birds were in good condition (estimate ± SE = −0.271 ± 0.07,
z = −3.971, P < 0.0001). As a consequence, we kept these two factors
in the further analysis.
3.2.2. Overall intake
Highest intake rates were observed on seeds of the pair 1 (C.
album and C. polyspermum: 31.85 ± 1.58 seeds per trial, N = 20) that
involved two small seeds, and pair 2 (Fumaria officinalis and Papaver
rhoeas: 22.1 ± 0.97 seeds per trial, N = 20) which involved the smallest seed. The lowest intake rates were noticed for the pair 4 (P.
lapathifolium and P. aviculare: 7.45 ± 0.33 seeds per trial, N = 20)
which involved two large seeds. These results were therefore in
agreement with the results of the simple-choice test which concluded that smaller seeds were overall more selected than large
seeds.
None of the Manly’s ˛ preference indices were significantly correlated with seed intake (i.e. Spearman correlation test between ˛Sn
and seed intake, rs = 0.0987, P > 0.05), hence the use of this index
was relevant for analysing the data collected in our experiment
when the consumer reduces the amount of food available after each
selection (Chesson, 1983).
The results of the four GLMM models are summarized in Table 2.
The analysis of the GLMM models showed a systematic opposite
signs between the simple fixed effects of the difference in seed size
(i.e. seed mass), lipid and energy (always positive for ˛Sn and ˛Sb , and
negative for ˛Rn and ˛Rb ) and the interaction coefficients (always negative for ˛Sn and ˛Sb , and positive for ˛Rn and ˛Rb ) suggesting a trade-off
in the seed choice of skylarks. The major effects (higher coefficient
values) on selective predation were the interaction between seed
size (i.e. seed mass) and seed energy content (both measured as differences (in log) between the two seeds of the pair) for the fourth
Manly’s ˛ indices. Our results did not show any significant effects
of bird sex and body conditions on Manly’s ˛ indices.
146
S. Gaba et al. / Behavioural Processes 108 (2014) 142–150
Fig. 1. Description of the statistical analysis of the multi-choice offer experiment according to the hypothesis on the trait (seed size or seed lipid content) governing the
predation choice by skylarks and the variable that reflects selection, i.e. seed (number) intake or lipid biomass intake. Two analyses are performed. The first one (on the
bottom right) tested for a preferential intake within a pair. The second one (on the bottom left) investigated the effect of seed traits on the intake. DM, DL and DE refer to the
difference of mass, lipid content and energy content between the two seed of a pair, respectively.
3.2.3. Selective intake within pairs
Manly’s ˛ indices are shown in Fig. 3. Overall, we detected selection for the smallest seed (Fig. 3a and c) rather than for the richest
one (Fig. 3b and d), and the selection was better explained when
the intake quantity was expressed by the total lipid biomass (Fig. 3c
and d) than by the total number of seeds consumed (Fig. 3a and b).
In particular, Manly’s ˛ index ratio based on total lipid biomass (˛Sb )
increased linearly with the difference in seed mass (Fig. 3c; linear
2 = 0.294, P < 0.0001). Conversely, no trend
model: F1,118 = 50.57, radj
between energy difference and seed selection was observed.
Selective predation on the number of seeds eaten according to
the difference in mass between the two seeds of the pair was significant only when the difference was large and one of the seeds
was small (pairs 2 and 5) hence selection ratio was highest for the
smallest seed (Fig. 3a). When the mass difference was small and
the pair consisted of small seeds (pairs 1 and 6), skylarks showed
either a preference towards the smallest seed (although not significant) or none at all. When the mass difference was minimal but
the two seeds were large (pairs 3 and 4), skylarks selected for the
larger seed (pair 4) or had no preference (pair 3; Fig. 3a). In pair 3,
our analysis showed a significant selection in terms of numbers of
seeds consumed for the richest seed (˛Rn ; Fig. 3b) and no preference
towards the smallest seed (˛Sn ; Fig. 3a). The opposite was shown for
pair 4 where we observed a significant selection of the larger seed
(˛Sn ; Fig. 3a) and no preference for the richest one (˛Rn ; Fig. 3b). This
would suggest that skylarks would select for lipid rich seeds when
the difference in seed size is small and seeds are large (pair 3) and
would optimize the seed intake when seed lipid content is small
(pair 4).
3.2.4. Single-offer experiment versus multiple-offer experiment
The results between the single-offer and the multiple-offer
experiments choice could be compared for two pairs (pairs 1 and 4)
and showed contrasted results. In pair 4 (two large seeds, both poor
in lipid content), the smallest seed (which was also the poorest in
lipid) was strongly preferred in single choice test, to the extent that
the total amount of lipid ingested (total biomass) was higher for
the smallest seed than for the largest one (Fig. 2). In multiple-offer
experiment, the largest seed was significantly selected when the
quantity ingested was expressed in number of seeds, but not in total
lipid biomass (average ˛Rb of pair 4 ≈ 0.5, Fig. 3d). In other words, as
expected on the basis of selection towards total lipid biomass, skylarks seemed to equal the total reward between the two seeds in
adjusting the relative intake of the two seeds species. In the pair 1, in
which the two seeds were small seeds and were rather rich in lipid
content, we should not expect therefore strong relative selection.
Indeed skylarks did not show significant preference in the singleoffer experiment (though they ate more of the largest seed that
was poorest in lipid content), and maximized the intake in terms
of total lipid biomass with the highest value obtained for the relatively larger seed (see Fig. 2b and d). Similarly in multiple-offer
experiment, they did not show any preference when we consider
the number of seeds consumed (average ˛Sn of pair 1 ≈ 0.5; Fig. 3a),
which maximized the total amount of lipid obtained.
4. Discussion
The aim of the present study was to analyze seed preferences
of skylark in relation to seed traits, with a particularly focus on
the possible trade-off between seed lipid content and seed size.
We predicted firstly that skylarks would prefer small seeds since
they require less energy to be eaten and digested than large seeds
(Thompson et al., 1987) and secondly that skylarks would prefer
high lipid content seeds since they would have higher caloric content than high protein or high carbohydrates seeds (Celis-Diez et al.,
2004; Martínez et al., 2006). No clear trend for a preference of high
S. Gaba et al. / Behavioural Processes 108 (2014) 142–150
147
Fig. 2. Seed intake according to (a) seed mass (i.e. seed size), (b) seed lipid, (c) seed energy, (d) seed biomass intake, (e) seed lipid biomass intake and (f) seed energy intake
for the four seed species used during the simple-offer experiments. Black dots indicate average values and arrows the standard errors. We used mean values of mass, lipid
and energy content per seed for each species; hence we could not compute any standard errors for the x-axis on (a), (b) and (c). Numbers indicate the species: 1 for C. album,
2 for C. polyspermum, 3 for P. aviculare and 4 for P. lapathifolium. Results of the post hoc tests are presented by letters.
lipid content seeds was found, but in both types of experiments,
skylarks selected for total lipid biomass. Since total lipid biomass is
the product of lipid content per seed and seed intake, this suggests
that there is a potential trade-off between eating a lot of seeds poor
in lipids and fewer seeds rich in lipids. Our results revealed that skylarks indeed trade mass against lipid content, with a preference for
seed mass (i.e. seed size) when both traits were competing. In both
types of tests, skylarks showed preference for the smallest seeds.
Results of simple-offer experiments were consistent with gizzard
analysis (Eraud et al. in revision) where Chenopodium sp. (in particular C. album) and Polygonum sp. (in particular P. aviculare) are
generally preponderant in skylark gizzards analyzed to date (see
also Cramp, 1988). Following our prediction, we thus suggest that
seed size is a putative trait for selection by the skylark which seems
to prefer small seeds in general. However, when seeds are both large
in size and poor in lipid content (Polygonum sp.–pair 4), i.e. there is
an additive (negative) effect of size and lipid content, bigger seeds
were always preferred than smaller seeds. These large and low oil
content seeds actually combine two potentially prejudicial features
for bird predation since they impose high handling costs not offset
by their nutritional provision. When the size difference between
seeds is important (pairs 2 and 5), small seeds are strongly selected
even though the difference in lipid content is small (pair 5). However, when total lipid biomass was traded (Fig. 3c), a selection (in
number of seed intake) towards the smallest seeds was observed
as soon as the mass difference increases (Fig. 3c). Thus seed size
seems to be a preponderant choice criterion for seed selection by
these farmland birds.
The question thus remains whether seed size is a proxy of seed
energy content, i.e. whether there is a relationship between size
and lipid content, in which case seed selection could be indirectly
driven by lipid content. The influence of seed size as a “proxy” for
other traits has previously been suggested in seed predation (e.g.
seed predation by rodents, Wang and Chen (2009). Several studies
demonstrated that the amount of calories in seeds was positively
correlated with seed mass (Kelrick et al., 1986), suggesting that
seed nutrient and/or seed energetic global value increases with size
(Celis-Diez et al., 2004; Martínez et al., 2006). However, large seeds
can contain carbohydrates or proteins levels higher than lipids level
and be less attractive for birds in winter. Moreover, large seeds are
more likely to contain toxic secondary metabolites (Díaz, 1996). In
the case of farmland birds, since smaller seeds are potentially richer
in lipids (Bretagnolle et al., in revision), a preferential predation of
small seeds would ensure for maximizing net energy intake while
reducing handling time and avoiding toxic secondary metabolites,
as predicted by the optimal foraging theory (Krebs, 1980). Our
results do not permit to determine whether and how skylarks have
the ability to detect seed nutrient content, though it is likely since
Manly’s preference index computed with lipid biomass (˛Sb or ˛Rb ) in
average never revealed a preference for the poorest seed. Moreover,
some evidence on the ability to determine seed nutrient content
has been shown in other bird species. The red-winged blackbird
148
S. Gaba et al. / Behavioural Processes 108 (2014) 142–150
Table 2
Outputs of the GLMM models. (a) and (b) give the results of the GLMM with Manly’s ˛ index computed with seed intake on an assumption of selection towards smallest seeds
(˛Sn ) or richest seeds (˛Rn ). (c) and (d) give the results of the GLMM analysis with Manly’s ˛ index computed with seed lipid biomass intake on an assumption of selection
towards smallest seeds (˛Sb ) or richest seeds (˛Rb ).
(a) ˛Sn
Intercept
Size difference (log)
Lipid difference (log)
Energy difference (log)
Sex (M)
Body condition
Size difference × lipid difference
Size difference × energy difference
(b) ˛Rn
Intercept
Size difference (log)
Lipid difference (log)
Energy difference (log)
Sex (M)
Body condition
Size difference × lipid difference
Size difference × energy difference
(c) ˛Sb
Intercept
Size difference (log)
Lipid difference (log)
Energy difference (log)
Sex (M)
Body condition
Size difference × lipid difference
Size difference × energy difference
(d) ˛Rb
Intercept
Size difference (log)
Lipid difference (log)
Energy difference (log)
Sex (M)
Body condition
Size difference × lipid difference
Size difference × energy difference
Value
Std. error
df
t-Value
p-Value
−0.6542
1.0116
1.7626
3.7841
0.0789
−0.0022
−1.1564
−7.6992
0.2839
0.296
0.6608
0.9187
0.0727
0.0104
0.5128
2.1404
67
45
45
45
67
45
45
45
−2.3046
3.4181
2.6672
4.1188
1.0852
−0.2149
−2.2549
−3.5971
0.0243
0.0013
0.0106
0.0002
0.2817
0.8308
0.0291
0.0008
1.9035
−1.3114
−3.0803
−4.2775
−0.0266
−0.014
2.4085
9.6529
0.2903
0.3047
0.6816
0.9472
0.0664
0.0099
0.5284
2.2127
67
45
45
45
67
45
45
45
6.5579
−4.3034
−4.5193
−4.5159
−0.4002
−1.1057
4.5579
4.3626
<0.0001
0.0001
<0.0001
<0.0001
0.6903
0.1667
<0.0001
0.0001
0.2632
0.2731
0.2208
0.9532
−0.0503
−0.0012
−0.1163
−1.4756
0.2096
0.2192
0.4899
0.681
0.5103
0.0075
0.3799
1.5881
67
45
45
45
67
45
45
45
1.2557
1.2454
0.4507
1.3998
−1.0389
−0.1647
−0.3061
−0.9291
0.2136
0.2194
0.6543
0.1684
0.3026
0.8699
0.761
0.3578
0.9426
−0.3849
−1.1457
−1.4381
−0.0287
−0.0023
0.891
3.5323
0.2111
0.2208
0.04935
0.6859
0.5106
0.0075
0.3828
1.6002
67
45
45
45
67
45
45
45
4.4658
−1.7429
−2.3216
−2.0963
−0.5594
−0.3021
2.3278
2.2075
<0.0001
0.8820
0.0248
0.0417
0.5778
0.7640
0.0245
0.0324
P-values <0.05 are indicated in bold.
(Agelaius phoeniceus) is able to discriminate between sunflower
seeds of various oil contents when difference was at least of 15% (or
5% in one case; Mason and Wilson, 2006). Birds were also shown
to prefer the consumption of seed with high contents of starch (see
Cueto et al., 2006; Ríos et al., 2012a,b). Seeds with high content of
starch are generally poor in lipids. These discrepancies about seed
preference by granivorous birds might be explained by variation
in bird’s nutritional requirements during the year. In our experiment, birds were captured during post-nuptial migration, a time
where birds feed on weed seeds in farmlands and are accumulating fat reserve. Further research should be conducted to determine
whether and how the selection of food with different nutrient
contents is affected during the year.
While body condition and sex had significant influence on seed
intake when analyzed alone, their influence was negligible influence on seed trait selection. Body size differences, particularly beak
length, between male and female skylarks (Scebba, 2001) might
be too small to significantly affect their seed preferences. Females
showed shorter beaks than males in our sample test (data not
shown). A sexual difference in beak length may lead to sex differences in seed size preferences (Avery, 1996; Willson, 1972).
In fact, beak size may influence foraging strategies because beak
depth is related to the maximal compression force that mandibles
can apply without risks of skull injuries (Diaz, 1994). Mason et al.
(1991) suggested that other factors may influence seed discrimination such as training, sensory or visual cues. Indeed, seed colour
is a potential factor influencing seed selection by birds since it
can favour seed detection contrasting with substrate or creating
signal of toxicity. In our experiment, we controlled for seed colour
by choosing only dark seeds. However, we did not control for
other seed traits such as seed coat thickness or seed strength that
can influence seed selection. For instance, the relative strengths
of the seeds was found to be negatively correlated with seed size
(Lundgren and Rosentrater, 2007), suggesting that smaller seeds
require a greater force and energy to crush them. It would be interesting in further studies to take into account other morphological
seed traits even though the relationship between seed traits such
as size, nutrient composition, secondary compounds or digestibility is often complex and nonlinear (Gardarin et al., 2010; Blate
et al., 1998; Shimada, 2001). Artificial seeds could enable to tease
apart the relative influence of different seed traits (see for example, Wang and Chen, 2012). Finally, seed selection by granivores is
likely to depend on seed availability, which varies with season and
weed species (Molokwu et al., 2011). Therefore, plant species spatial and temporal distributions need to be incorporated in a seed
selection analysis in order to quantify the impact of seed predation
on plant communities. Such spatial and temporal distributions of
weeds through crop spatio-temporal arrangement in agricultural
landscape could provide a management option for skylark conservation. For instance, in our study, preferred species tend to be
associated with spring crops (Chenopodiaceae and Polygonaceae),
introducing spring crops in the landscape may by a sustainable way
to maintain skylard populations. However this should be thought
in a holistic way integrating other weed functions (i.e. competition with the crop). To conclude, our experiments provide evidence
for a preferential seed predation of skylarks which may offer
S. Gaba et al. / Behavioural Processes 108 (2014) 142–150
149
Fig. 3. Variation of Manly’s ˛ preference index computed with seed intake (˛n ) and seed lipid biomass intake (˛b ) according to the difference in seed mass (a, c) and in
seed lipid content (b, d), respectively. Species i was the smallest seed on a and c, and the richest in lipid on b and d. “*” indicates a significant selective predation following
non-parametric Sign test. Tests with species for which seed mass and lipid content were additive are represented by black dots. Numbers indicate the pair of seeds: 1 for C.
album and C. polyspermum; 2 for P. rhoeas and F. officinalis; 3 for G. molle and G. robertianum; 4 for P. aviculare and P. lapathifolium; 5 for S. media and S. holostea and 6 for P.
rhoeas and S. media.
potential management perspectives through weed seed bank regulation. Further studies should explore the potential role of skylarks
in mediating changes in the composition and structure of seed
banks in farmlands by using for example the response-effect framework initially proposed by Lavorel and Garnier (2002) and extended
by Lavorel et al. (2013) to link plant traits to higher trophic levels.
Acknowledgements
Annick Matejicek, Emilie Cadet and Cyril Eraud are warmly
thanked for their help during field work. This work has been funded
by ANR Systerra (STRA-08-02 Advherb), INRA and CNRS. TP was
supported by a PhD grant from ONCFS. Two referees provided helpful comments on the manuscript.
References
Anhalzer, G., Fournier, M., O’Connor, T., Stevenson, L., Yglesias, M., 2010. Traits for
predator selection on Pentaclethra macroloba seeds. Am. J. Undergrad. Res. 8,
1–8.
Avery, M.L., 1996. Food avoidance by adult house finches, Carpodacus mexicanus,
affects seed preferences of offspring. Anim. Behav. 51, 1279–1283.
Blate, G., Peart, D., Leighton, M., 1998. Postdispersal predation on isolated seeds: a
comparative study of 40 tree species in a Southeast Asian rainforest. Oikos 82,
522–538.
Blem, C., 1990. Avian energy storage. In: Power, D. (Ed.), Current Ornithology.
Plenum, New York.
Boatman, P., Pietravalle, S., Parry, H., Crocker, J., Irving, P., Turley, D., Mills, J., Dwyer,
J., 2010. Agricultural land use and skylark Alauda arvensis: a case study linking a
habitat association model to spatially explicit change scenarios. Ibis 152, 63–76.
Bretagnolle, F., Matejicek, A., Grégoire, D., Reboud, X., Gaba, S., in revision. Determination of fatty acids content, global antioxidant activity and energy of selected
weed seeds from agricultural fields in France. Seed Sci Res.
Carrillo, C.M., Moreno, E., Valera, F., Barbosa, A., 2007. Seed selection by the trumpeter finch, Bucanetes githagineus. What currency does this arid-land species
value? Ann. Zool. Fenn. 44, 377–386.
Celis-Diez, J.L., Bustamante, R.O., Vásquez, R.A., 2004. Assessing frequencydependent seed size selection: a field experiment. Biol. J. Linn. Soc. 81, 307–312.
Chesson, J., 1983. The estimation and analysis of preference and its relationship to
foraging models. Ecology 64, 1297–1304.
Cramp, S., 1988. Handbook of the Birds of Europe, The Middle East and North Africa.
The Birds of the Western Palearctic. Oxford University Press, Oxford, UK.
Cueto, V.R., Marone, L., Lopez de Casenave, J., 2006. Seed preferences in sparrow
species of the Monte desert, Argentina: implications for seed–granivore interactions. Auk 123, 358–367.
150
S. Gaba et al. / Behavioural Processes 108 (2014) 142–150
Daunicht, W., PhD thesis 1998. Zum Einfluss derFeinstruktur in der Vegetation auf
die Habitatwahl, Habitatnut- zung, Siedlungsdichte und Populationsdynamik
von Feldlerchen (Alauda arvensis) in grossparzelligem Ackerland. University of
Bern.
Diaz, M., 1994. Variability in seed size selection by granivorous passerines: effects
of bird size variability, and ecological plasticity. Oecologia 99, 1–6.
Díaz, M., 1996. Food choice by seed-eating birds in relation to seed chemistry. Comp.
Biochem. Physiol. 113, 239–246.
Donald, P.F., 2004. The Skylark, London.
Eraud C., Cadet E., Powolny T., Gaba S., Bretagnolle F., Bretagnolle V., in revision.
Weed seeds, not grain, contribute to the diet of wintering Skylarks in arable
farmlands of western France. Europ J Wildlife Res.
Eraud, C., Lallemand, J., Lormee, H., 2006. Sex-ratio of skylark Alauda arvensis in
relation to timing of breeding: capsule earlier broods tend to be more male
biased than later broods. Bird Study 53, 319–322.
Fernández-Juricica, E., Vacab, R., Schroederb, N., 2004. Spatial and temporal
responses of forest birds to human approaches in a protected area and implications for two management strategies. Biol. Conserv. 117, 407–416.
Frank, C., 1988. Diet selection by a heteromyid rodent: role of net metabolic water
production. Ecology 69, 1943–1951.
Gaba, S., Chauvel, B., Dessaint, F., Bretagnolle, V., Petit, S., 2010. Weed species richness in winter wheat increases with landscape heterogeneity. Agric. Ecosyst.
Environ. 138, 318–323.
Gardarin, A., Durr, C., Mannino, M.R., Busset, H., Colbach, N., 2010. Seed mortality in
the soil is related to seed coat thickness. Seed Sci. Res. 20, 243–256.
Geiger, F., Hegemann, A., Gleichman, M., Flinks, H., de Snoo, G.R., Prinz, S., Tieleman,
I., Berendse, F., 2014. Habitat use and diet of skylarks (Alauda arvensis) wintering in an intensive agricultural landscape of the Netherlands. J. Ornithol. 155,
507–518.
Glück, E., 1985. Seed preference and energy intake of Goldfinches Carduelis carduelis
in the breeding season. Ibis 127, 421–429.
Green, A.J., 2001. Mass/length residuals: measures of body condition or generators
of spurious results? Ecology 82, 1473–1483.
Greig-Smith, P.W., Wilson, F.M., 1985. Influences of seed size, nutrient composition
and phenolic content on the preferences of Bullfinches feeding in ash trees. Oikos
44, 47–54.
Kelrick, M.I., MacMahon, J.A., Parameter, R.R., Sisson, D.V., 1986. Native seed preferences of shrubsteppe rodents, birds and ants: the relationship of seed attributes
and seed use. Oecologia 68, 327–337.
Kerley, G.I.H., Erasmus, T., 1991. What do mice select for in seeds. Oecologia 86,
261–267.
Krams, I., Cirule, D., Suraka, V., Krama, T., Tantala, M., Ramey, G., 2010. Fattening
strategies of wintering great tits support the optimal body mass hypothesis under conditions of extremely low ambient temperature. Funct. Ecol. 24,
172–177.
Krebs, J.R., 1980. Optimal foraging, predation risk and territory defence. Ardea 68,
83–90.
Lavorel, S., Garnier, E., 2002. Predicting the effects of environmental changes on
plant community composition and ecosystem functioning: revisiting the Holy
Grail. Funct. Ecol. 16, 545–556.
Lavorel, S., Storkey, J., Bardgett, R.D., de Bello, F., Berg, M.P., Le Roux, X., Moretti, M.,
Mulder, C., Pakeman, R.J., Diaz, S., Harrington, R., 2013. A novel framework for
linking functional diversity of plants with other trophic levels for the quantification of ecosystem services. J. Veg. Sci. 24, 942–948.
Lundgren, J.G., Rosentrater, K.A., 2007. The strength of seeds and their destruction
by granivorous insects. Arthropod-Plant Interact. 1, 93–99.
Manly, B.F.J., 1974. A model for certain types of selection experiments. Biometrics
30, 281–294.
Marone, L., de Casenave, J.L., Milesi, F.A., Cueto, V.R., 2008. Can seed-eating birds
exert top-down effects on grasses of the Monte desert? Oikos 117, 611–619.
Martínez, I., García, D., Obeso, J.R., 2006. Allometric allocation in fruit and seed packaging conditions the conflict among selective pressures on seed size. Evol. Ecol.
21, 517–533.
Mason, J.R., Clark, L., Shah, P.S., 1991. Taxonomic differences between birds and
mammals in their responses to chemical irritants. In: Doty, R., Muller-Schwarze,
D. (Eds.), Chemical Signals in Vertebrates. Plenum Press, New York.
Mason, N.W.H., Wilson, J.B., 2006. Mechanisms of coexistence in a lawn community:
mutual corroboration between two independent assembly rules. Commun. Ecol.
7, 109–116.
McCue, M.D., Amitai, O., Khozin-Goldberg, I., McWilliams, S.R., Pinshowa, B., 2009.
Effect of dietary fatty acid composition on fatty acid profiles of polar and neutral lipid tissue fractions in zebra finches, Taeniopygia guttata. Comp. Biochem.
Physiol. 154, 165–172.
Meiss, H., Médiéne, S., Waldhardt, R., Caneill, J., Bretagnolle, V., Reboud, X., MunierJolain, N., 2010. Perennial lucerne affects weed community trajectories in grain
crop rotations. Weed Res. 50, 331–340.
Molokwu, M.N., Nilsson, J.A., Olsson, O., 2011. Diet selection in birds: a trade-off
between energetic content and digestibility of seeds. Behav. Ecol. 2, 639–647.
Moorcroft, D., Whittingham, M.J., Bradbury, R.B., Wilson, J.D., 2002. The selection of
stubble fields by wintering granivorous birds reflects vegetation cover and food
abundance. J. Appl. Ecol. 39, 535–547.
Powolny, T., Eraud, C., Bretagnolle, V., 2012. Group size modulates time budget and
foraging efficiency in captive skylarks, Alauda arvensis. J. Ornithol. 153, 485–490.
Ramírez, N., Traveset, A., 2010. Predispersal seed-predation by insects in the
Venezuelan Central Plain: overall patterns and traits that influence its biology
and taxonomic groups. Perspect. Plant Ecol. Evol. Syst. 12, 193–209.
Ríos, J.M., Mangione, A., Marone, L., 2012a. Effects of nutritional and anti-nutritional
properties of seeds on the feeding ecology of seed eating birds of the Monte
desert, Argentina. Condor 114, 44–55.
Ríos, J.M., Mangione, A., Marone, L., 2012b. Tolerance to dietary phenolics and diet
breadth in three seed-eating birds: implications to graminivory. J. Exp. Zool. 317,
425–433.
Robinson, R.A., 2004. The diet of seed-eating birds on lowland farmland. Br. Birds
97, 464–467.
Scebba, S., 2001. Biometrics and sex ratios of Skylark Alauda arvensis during migration in southern Italy. Ringing Migr. 20, 364–370.
Schulte-Hostedde, A.I., Zinner, B., Millar, J.S., Hickling, G.J., 2005. Restitution of
masse-size residuals: validating body condition indices. Ecology 86, 155–163.
Seppäla, O., Karvonen, A., Tellervo Valtonen, E., 2004. Parasite-induced change in
host behaviour and susceptibility to predation in an eye fluke–fish interaction.
Anim. Behav. 68, 257–263.
Shimada, T., 2001. Nutrient compositions of acorns, and horse chestnuts in relation
to seed-hoarding. Ecol. Res. 16, 803–808.
Sutherland, W.J., 1996. From Individual Behaviour to Population Ecology. Oxford
University Press, Oxford.
Thompson, D.B., Tomback, D.F., Cunningham, M.A., Baker, M.C., 1987. Seed selection
by dark-eyed juncos (Junco hyemalis): optimal foraging with nutrient constraints? Oecologia 74, 106–111.
Valera, F., Wagner, R.H., Romero, M., Gutiérrez, J.E., Rey, P., 2005. Dietary specialization on high protein seeds by adult and nestling serins. Condor 105, 29–40.
Wang, B., Chen, J., 2009. Seed size, more than nutrient or tannin content, affects
seed caching behavior of a common genus of old world rodents. Ecology 90,
3023–3032.
Wang, B., Chen, J., 2012. Effects of fat and protein levels on foraging preferences
of tannin in scatter-hoarding rodents. PLoS ONE 7, e40640, http://dx.doi.org/
10.1371/journal.pone.0040640.
Westerman, P.R., Borza, J.K., Andjelkovic, J., Liebman, M., Danielson, B., 2008.
Density-dependent predation of weed seeds in maize fields. J. Appl. Ecol. 45,
1612–1620.
Willson, M.F., 1972. Seed size preference in finches. Wilson Bull. 84, 449–455.