Content uploaded by Ales Vorel
Author content
All content in this area was uploaded by Ales Vorel on Oct 20, 2015
Content may be subject to copyright.
ORIGINAL PAPER
AlešVor el
1
&Lenka Válková
2
&Lenka Hamšíková
1
&Jaroslav Maloň
2
&Jana Korbelová
1
Received: 5 July 2014 /Revised: 5 May 2015 /Accepted: 5 May 2015/Published online: 14 May 2015
#Springer-Verlag Berlin Heidelberg 2015
Abstract Classical theory states that generalist foragers
should specialise to maximise their net energy intake on a
temporal basis. Within broad distributions, generalists select
optimal food sources; however, the availability and represen-
tation of selected food items in the diets of generalist con-
sumers vary geographically. Conversely, specialists, within
their entire range, concentrate on specific food resources,
and their diets remain constant in varying environments. The
distributions of highly palatable food resources might influ-
ence the foraging patterns of species with opportunistic forag-
ing behaviours. We used the Eurasian beaver (Castor fiber)as
a model for what is considered an opportunistic generalist
herbivore. We analysed whether the beavers’foraging pattern
could show specialisation. Five beaver populations living in
diverse environments were evaluated, and their diet choices
for overwintering were examined. Our data indicated low pro-
portions of generality with regard to the beavers’foraging
behaviours. Within 110 territories, we observed that beavers
did pursue choosy opportunistic strategies, though only to a
small extent. Although we confirmed the established knowl-
edge that beavers prefer to browse mainly poplars and
willows, we also found that, contrary to many published stud-
ies, their preferences occurred regardless of riparian vegeta-
tion composition. While the beavers examined in this study
preferred six genera of woody vegetation, we argue that they
may have been inclined primarily to use two genera with
regard to both food choice and spatial territory placement.
Thus, the opportunistic feeding behaviours of generalists ob-
served in this study may be influenced by evaluation under
varying conditions.
Keyword Castor .Foraging .Beaver .Herbivory .
Generalist .Opportunists
Introduction
The factors affecting animals’diet choices are commonly
interpreted in the context of optimal diet theory (Pulliam
1974), which proposes that a particular foraging strategy aims
to maximise certain components related to fitness (e.g. rate of
energy intake), which depends on the energy content of alter-
native prey, encounter rates,and how effective the forager is at
detecting, capturing, handling and digesting the prey.
Herbivorous foraging behaviour can be basically
categorised as generalist or specialist, depending upon the
given species’habitat and food niche (Herrera and Pellmyr
2002). Staddon (1983) postulated that the most important con-
sideration for foragers is their net rate of energy intake. This
intake has direct consequences for diet choice, allocation of
foraging time to patches, and spatial search patterns in hetero-
geneous environments. Thus, the generalists always select to
consume the most optimal food sources; however, the avail-
ability of food items in the environment and their representa-
tion in the consumers’diets vary markedly with geographic
distribution. Conversely, specialists concentrate on specific
food sources within their entire range, and their diets remain
constant even in a varying environment. Stephens and Krebs
(1986) proposed that the specialisation of generalists occurs
Communicated by K. E. Ruckstuhl
*AlešVo re l
vorel@fzp.czu.cz
1
Department of Ecology, Faculty of Environmental Sciences, Czech
University of Life Sciences Prague, Kamýcká 129, 165
21 Prague, Czech Republic
2
Department of Ecology, Faculty of Science, Palacký University,
tř. Svobody 26, 771 46 Olomouc, Czech Republic
Behav Ecol Sociobiol (2015) 69:1221–1235
DOI 10.1007/s00265-015-1936-7
Beaver foraging behaviour: Seasonal foraging specialization
by a choosy generalist herbivore
when a particular resource is abundant and its quality provides
a high energy yield. On the other hand, increasing handling
time decreases the net benefit of the selected resource. Thus,
past a certain threshold, selective diet choice becomes unprof-
itable and the foragers utilise alternative resources with lower
energy yield. The transposition from generalist to specialist
foraging behaviour has been studied along habitat and climat-
ic gradients (Panzacchi et al. 2008) and is alsoknown to occur
seasonally (Green and Flinders 1980). Individual differences
in behavioural patterns also have been described (Woo et al.
2008).
Although broadly confirmed transitions along the general-
ist–specialist continuum have been stated (reviewed by
Bolnick et al. 2003), it is not clear whether this is a common
pattern. It is uncertain whether species conventionally known
as generalists might adopt specialisation anywhere when the
food web contains a highly profitable diet item, and we lack
information on how foragers behave in heterogeneous habi-
tats. Theory suggests that conspecific individuals behave
similarly, but that they may use different resources or
niches (Hutchinson 1957;Belovsky1986). Bolnick
et al. (2003) noted individual specialisation of foraging
patterns under conditions of heterogeneous resource dis-
tribution and quality. Obviously, a generalist’sdietis
regionally moderated by the distribution of sufficient
resources, foraging behaviour and patch size. Hence,
foraging behaviour varies depending upon the possibili-
ty to maximise energy intake in a given area (Belovsky
1978;Crawley1983;Grossetal.1993).
We used the Eurasian beaver (Castor fiber) as a model
species for large, opportunistic and choosy generalist herbi-
vores and studied how foraging behaviour potentially varies
according to regional habitat heterogeneity. Both beaver spe-
cies (Eurasian as well as North American Castor canadensis)
are large territorial rodents with similar feeding behaviours.
Wilsson (1971) and Svendsen (1980) defined their food
niches to consist of three vegetation categories (aquatic plants,
grass and forbs, and woody plants), with each component’s
proportion in the diet fluctuating seasonally. The relative con-
sumption of these diet components varies geographically.
While Svendsen (1980) described aquatic plants as constitut-
ing a substantial component of the diet, Ganzhorn and
Harthun (2000) documented a regionally lower consumption
of this resource. Although few recent studies (e.g. Milligan
and Humphries 2010; Severud et al. 2013) have found aquatic
plants to comprise an important component of the diet either
during summer or winter, that source is rather more typical for
colder regions (i.e. boreal or subarctic). In contradiction to
other authors (e.g. Heidecke 1989; Law et al. 2014),
Krojerová-Prokešová et al. (2010) found for the Central Eu-
ropean region no indication of aquatic plants among foraged
food items. During autumn and winter, when green vegetation
is unavailable, beavers forage largely on the bark (and
cambium) of woody vegetation (Jenkins 1979; Nolet et al.
1994; Krojerová-Prokešová et al. 2010). Beavers’potential
diet is comprised of many woody plant species within their
wide distribution (Novak 1987), thus their feeding behaviour
corresponds to the commonly known Bopportunistic foraging
strategy^(Nolet et al. 1994; Haarberg and Rosell 2006).
When willows and poplars (including aspens) dominate in
occupied areas, then these genera are detected in the diet
(across the north–south gradient of the beavers’distribution)
(Hall 1960; Aleksiuk 1970; Basey et al. 1990; Fryxell and
Doucet 1993; Fryxell et al. 1994). By contrast, in other habi-
tats with diverse riparian compositions, beavers consume the
prevailing genera, such as alders, birches, rowans and others
(Wilsson 1971; Haarberg and Rosell 2006). To date, no com-
prehensive study of beavers’foraging behaviour across di-
verse habitat types has been conducted. The aforementioned
authors conducted foraging studies focused on single ecosys-
tem types, describing narrow ranges of food availability and
foraging behaviours.
Like other vertebrate herbivores, beavers forage on avail-
able resources to maximise long-term energy intake (Doucet
and Fryxell 1993). This foraging strategy also improves opti-
mal energy intake during the critical pre-winter and winter
phase of the year, which is the most important phase for bea-
vers’survival and reproduction. However, for beavers, nutri-
tionally rich forbs are available mainly in the summer sea-
son—mainly for temperate regions (Svendsen 1980). Never-
theless, Aleksiuk and Cowan (1969) showed how in cold
regions diet composition and extreme climatic factors might
directly impact the animals’overwintering, inasmuch as bea-
vers’daily activity and bodysize decrease with declining win-
ter temperature and low diversity of riparian stands (also
Smith et al. 1991).
Our objective was to evaluate the beavers’pre-winter and
winter forage in relation to regional variations in available
resources. Due to the negligible consumption of forbs and
aquatic plants during the winter, we assessed only the woody
vegetation diet component. We examined the collective forag-
ing behaviours of beaver families within various studied sites,
and then we tested which deciduous woody vegetation would
be selected at the overall population level in a given site. We
hypothesised that if beavers are opportunistic feeders, their
forage will favour the most dominant woody vegetation at
the various sites. Because our results revealed high selectivity
of just two woody genera, we decided to accept and further
test the alternative hypothesis: during overwintering beavers
are selective feeders. Here, we assumed that during winter
beavers might be highly selective only for a few food items
regardless of their availability at various sites. Thus, we sub-
sequently tested whether feeding pressure prevails in propor-
tion to the ratio of abundance of the preferred woody vegeta-
tion in the studied sites. We predicted that if narrow foraging
exists it would be evident when beavers’foraging of preferred
1222 Behav Ecol Sociobiol (2015) 69:1221–1235
woody vegetation increased, even as the proportional resource
availability declined.
Materials and methods
Study area
The fieldwork was carried out in temperate broadleaf and
mixed forests within Central Europe. We selected five differ-
ent environments occupied by beavers in the Czech Republic
(for details, see Table 1and Fig. 1). The free-living beaver
populations had originated through spontaneous migrations
from neighbouring countries (Germany and Austria), where
successful reintroduction programs have been implemented
(Halley and Rosell 2002). The selected habitats within
human-made continental landscapes offer various environ-
mental conditions with regard to characteristics such as bodies
of water, land use, level of urbanisation and landscape man-
agement (Table 1). Except for land use (level of human ex-
ploitation), vegetation composition is the most dominant en-
vironmental factor characterising diverse habitat conditions
for beavers. In Český les (CL), riparian vegetation is com-
prised of alders (Alnus spp.), spruces (Picea spp.) and birches
(Betula spp.). In Chropyňský luh (CH), willows (Salix spp.),
poplars (Populus spp.), alders and maples (Acer spp.) are most
abundant. Similarly, in Labe (LA), willows are dominant with
a small contribution from poplars. The riparian composition of
Niva Dyje (ND) is comprised primarily of willows, maples,
ashes (Fraxinus spp.) and poplars. Finally, in Soutok–Podluží
(SP), ashes dominate, followed by maples, oaks (Quercus
spp.) and willows (detailed species composition of the studied
sites is in Table 3).
None of the studied populations are either under hunting
pressure or predated by large carnivores (able to pursue adult
beavers). The five studied populations are well developed and
far beyond the initial phases of colonisation (Barták et al.
2013).
Although dam and lodge building behaviour is not com-
mon in four of the five populations, it is evident on a small
scale, as constructions occur here occasionally (dams occur in
<1 % of territories, lodges are constructed on at most 5–10 %
of all territories; AV et al. unpubl. data). There is well-
developed building behaviour at the Český les site (CL),
where beavers live in lodges and on beaver pond systems.
At the time of sampling, we observed that the construction
of dams and lodges had been completed on the majority of
territories; newly built or reconstructed buildings were record-
ed on up to 10–20 % of all territories (Vorel et al. unpubl.).
However, a substantial proportion (unquantified) of felled
woody vegetation was used here as construction material.
Territory utilisation analysis
During the winter of 2007/2008, we evaluated within the studied
sites the number and demarcation of sites recently occupied and
used by beavers (i.e. ‘territories’in our terminology, while the
utilisation of space actively occupied by one beaver family was
theoretically determined using a stochastic process). We
inspected all areas surrounding the water bodies in the selected
sites—as real or potential beaver habitat—to record signs of
beaver spatial activity (channels, dams, resting places, slides,
scent marking as well as cuttings of trees or shrubs). By survey-
ing the area on foot or by boat, we collected and spatially located
all fresh marks and recorded only those marks made during
autumn 2007 and winter 2007/2008. We focused on locating
and assessing shelters (i.e. lodges and bank burrows) or winter
stores in active use, as well as determining scent marks.
The shelters and other locations of beaver activities were
transposed in ESRI ArcGIS 9.3 software (ESRI, Redlands,
California, USA) to the central line of bodies of water. Then,
we used the KernSmooth package (Wand and Ripley 2013)to
apply a kerneldensity estimation (KDE; bandwidth selector is
implicitly included in the package) to the data. We determined
a territory to be where KDE spatially clustered beaver activity
marks around an active shelter, i.e. a territory patch included
Tabl e 1 Description of studied sites where beaver foraging was carried out
Name of site Abbreviation
in text
Character Coordinates Elevation
(m a.s.l.)
Woo dy vegeta tion
canopy
Majority
land use
Usual
flooding
Český les CL Small sub-mountain
watercourses
49° 40′16″N12°33′57″E550 30–50 % Pasture No
Chropyňský luh CH Agricultural landscape 49° 26′26″N17°18′24″E210 10–30 % Agriculture Yes
Labe LA Large river 50° 45′03″N14°11′20″E120 10–30 % Industry Yes
Niva Dyje ND Deciduous forests and
Agricultural mosaic
landscape
48° 47′54″N16°50′16″E150 40–100 % Timber production Yes
Soutok–Podluží SP Flood plain forests 48° 40′49″N16°56′37″E150 40–100 % Timber production Yes
All sites had been occupied by beavers at least 15 years. All of them are Sites of Community Importance under Natura 2000 in the Czech Republic
(Fig. 1). These differ by type of water environments, land use and levels of urbanization and landscape management
Behav Ecol Sociobiol (2015) 69:1221–1235 1223
all beaver activity marks in one unimodal kernel cluster (con-
taining active shelter). The borders of the territories were de-
fined as areas where kernel estimations reached zero density
or where the densities of other clusters began to increase in
neighbouring territories. Such borders were projected into the
GIS environment as territory borders (polygon layer).
Food resource availability
The available woody vegetation was described numerically
during the springs and summers of 2006, 2007 and 2008.
The entire river system of each studied site was divided into
belts 10 km long. Due to large differences in both size and
complexity of the studied sites, the proportions of mapped
belts differed among them. The general rule was that the larger
a site the lower was the proportion of belts used within the site.
We randomly selected belts to be mapped while using be-
tween 10 and 20 belts per site to cover at least 40 % of the
utilised territories with belts. In each such selected belt, the
vegetation was assessed within patches 50 m long and 10 m
wide (i.e. the nearest riparian belt projecting 10 m outward
from the shoreline) on each bank (in total, 400 patches per
belt). The width of 10 m was chosen in accordance with Nolet
et al. (1994), who have shown that most beaveractivityoccurs
within 10 m from the water line. Our observations corroborat-
ed those of Nolet et al., as we found that beavers were primar-
ily active in close proximity to the riparian zone. Woody veg-
etation was identified at the genus level (available species
composition per each site are provided in Table 2), and five
field researchers estimated the proportions of genera in each
patch. For each genus, we classified the proportion of treetop
at ground cover per patch (50× 10 m); proportions were
categorised as occurrences of <5, 5–40 and >40 %,and genera
were measured separately at the tree and the shrub level. Sub-
sequently, both levels of woody vegetation within a patch
were combined. For the foraging analysis, we used only the
proportion ofall clumped woody vegetation per patch (regard-
less of growth form).
Beaver diet
We summarised the use of woody vegetation during February
2008, while we only quantified the fresh cuts (as indicated by
the clean and light coloration of those cuttings) that had re-
sulted from pre-winter and winter consumption (cuts were
felled from early autumn woody feeding activity in
September/October 2007 up to February 2008). Foraging dur-
ing this period suggests effort to improve energy intake for
overwintering, as it exceeds the daily consumption require-
ments. During this period, beavers were building food caches
and developing body fat, both of which are necessary for their
overwintering survival. Thus, we noted all freshly cut trunks
and branches (when beavers were able to nibble the bark and
foliage of those plants from the ground or water). Although
we were not able to evaluate whether cut vegetation was eaten
or stored, we nevertheless propose that most of those cuts
made during the overwintering period are intended for con-
sumption. Standing woody genera only having peeled bark
and no bite marks in the wood, as well as incomplete felling
of stands, were omitted from consideration as forage. When
beavers felled the woody vegetation, we assumed that the
majority of the usable biomass from this cut vegetation was
LA
CL
CH
ND SP
Germany
Poland
Czechia
Austria
Slovakia
Fig. 1 Distribution of studied
sites in the Czech Republic,
locations are marked with solid
polygons: CL Český les, CH
Chropyňský luh, LA Labe, ND
Niva Dyje, SP Soutok–Podluží
1224 Behav Ecol Sociobiol (2015) 69:1221–1235
consumed. Little or no biomass was used as building material
(except at CL, where building activity was more common),
but the majority of branches and trunks used for construction
were barked (i.e. bark was consumed before use for construc-
tion). Occasionally, we found that beavers did not use all com-
pleted cuts (mainly trunks); the proportion of used trunks or
branches was higher than the proportion of those remaining.
The genus was determined for each grazed tree or shrub, and
its diameter was recorded in the appropriate category (0.5–2.5,
2.6–6, 6–12, 12–20, 20–30, 30–40, 40–50 and ≥50 cm).
Those cuts with the diameters smaller than 0.5 cm were not
evaluated because there was no clear evidence of grazing by
beavers. We determined foraging in all patches in which we
evaluated the food resource availability. GPS coordinates for
each grazing location were also recorded.
Data analyses
Resource availability
The woody vegetation at all (50×10 m) patches was quanti-
fied. The availability of a given genus was represented as its
ground cover in each patch (in m
2
), i.e. category of genus’s
canopy at ground cover multiplied by patch size. The data
were entered in ESRI ArcGIS 9.3 software (ESRI, Redlands,
California, USA), where each recorded genus per patch was
attributed tothe centroid of the patch (point layer) with a value
estimating its categorical canopy in the patch.
Comparison of vegetation composition among sites
We compared the composition of riparian forests among stud-
ied sites using the vegetation descriptions obtained from our
analyses described above. First, we clumped all previously
mapped patches (of size 50×10 m) into vegetation strips of
200×10 m. We then checked the spatial autocorrelation of the
data clustered within these strips. The results showed (auto-
correlation measured by Moran’s I index=0; Z score=−0.54
standard deviations) that strips were spatially random at dis-
tances of 15 km. Therefore, we compared vegetation only in
strips regularly spaced 15 km apart (strips were proportionally
distributed according to size of studied sites). For our compar-
isons of vegetation composition, we only used those genera
whose total representation (per all strips) exceeded 0.5 %. The
relationships among the sites were determined using canonical
correspondence analysis (CCA) in the CANOCO 4.5 package
(Microcomputer Power, Ithaca NY, USA).
Beaver consumption estimation
We summarised the beaver diet to estimate the actual amount
of biomass for a given genus utilised within a patch. First, the
asymptotic curve model of distribution of individual cuttings
Tabl e 2 Description of species of the woody vegetation detected in the
studied sites
CL CH LA LP ND SP
Acer pseudoplatanus xx xxx x
Acer campestre xx x
Acer negundo xx x
Acer platanoides xxxxx
Alnus glutinosa xx xxx x
Alnus incana xx
Betula pendula xx xxx x
Betula pubescens xx
Carpinus betulus xx xxx x
Cornus sanguinea xxxxx x
Corylus avellana xx xxx x
Crataegus laevigata xx xxx x
Crataegus monogyna x
Euonymus europaeus xx xx x
Fraxinus excelsior xx xxx x
Fraxinus angustifolia xxx
Juglans regia xxxxx
Pices abies xx xxx x
Populus nigra xx xxx x
Populus×canadensis xxx
Populus×canescens xxxx
Populus alba xx x
Populus tremula xxx x x
Populus spinosa xx xxx x
Prunus padus xxxx
Prunus domestica xx
Prunus cerasifera x
Quercus petraea xx xxx x
Quercus robur xxxxx
Robinia pseudoacacia xxxxx
Rosa canina xx xxx x
Salix alba xx xxx x
Salix caprea xx xxx x
Salix cinerea xxx xx x
Salix fragilis xxx xx x
Salix×rubens x
Salix triandra xx xxx x
Salix viminalis xx
Sambucus nigra xx xxx x
Sorbus aucuparia xx xx x
Sorbus torminalis xx
Tilia cordata xx xx x
Tilia platyphyllos xx
Ulmus glabra xx
Ulmus laevis xx xxx x
Ulmus minor xxxx
Viburnum opulus xxx
Behav Ecol Sociobiol (2015) 69:1221–1235 1225
(n=31,102 units) in diameter categories was fitted (nonlinear
least squares equation):
Y¼Asym−RO⋅e−lrc X
with regression parameters A
sym
=0.012704, R
0
=0.590678,
l
rc
=−1.160745, obtained using the nlme statistical package
(Pinheiro et al. 2007) in R (R Development Core Team
2009). All medians for each diameter category were acquired
from the fitted curve. The diameter category (i)median(dm
i
)
was used in the general allometric volume and biomass equa-
tion for hardwood species (alder, ash, aspen and willow) in
accordance with Jenkins et al. (2004), Muukkonen and
Mäkipää (2006) and Muukkonen (2007):
bmi¼exp‐2:2094þ2:3867 ln dmi
ðÞ
where bm
i
is total biomass (kg). Jenkins et al. (2004)have
stated a biomass ratio equation dividing aboveground biomass
into four main components: stem bark, stem wood, coarse
roots and foliage. Only the ratios of stem bark and foliage
were used as potential sources of beavers’diet. Finally, useful
biomass for beavers was expressed as:
bmsum ¼Xbminiexp −4:0813 þ5:8816
dmiþexp −2:0129 þ−1:6805
dmi
!
;
where bm
sum
is the summation of beavers’feeding activity at
one location for one genus, and n
i
is the number of cuts in the
diameter category for the given genus at a particular location.
Thus, we obtained a comparable volume measure of the bio-
mass utilised for each foraged genus in each analysed patch.
Preference indices
We used two formulae to express the relationship between
beavers’feeding behaviour and the riparian vegetation. First,
we compared the availability of stands for beaver food con-
sumption across entire studied sites; design-I of resource
utilisation (sensu Thomas and Taylor 2006) was evaluated
using the selectivity index (Jacobs 1974). A positive value
for index lnQ indicates a preference for a resource when it is
utilised by beavers proportionally more than its representation
in the riparian stands. Conversely, values below zero are ob-
tained when beavers use proportionally less than the resources
available. The test of significance was established according
to Jenkins (1979): one degree of freedom compared with the
χ
2
distribution.
Second, we used Manly’s selection ratios (Manly et al.
2002) to establish a model design-II (Thomas and Taylor
2006). The estimations of beavers’food consumption and
the previously mapped available resources in each utilised
territory were used. In a GIS environment, we combined the
estimated availability of woody vegetation per patch with
localised beaver consumption; both parameters were attribut-
ed to a polygon layer representing utilised territories by deter-
mining the windex (for equation and test of White–Garrott
statistics, refer to Manly et al. 2002).
Sequentially, principal component analyses
(eigenanalyses) (Calenge and Dufour 2006)wereperformed
on each sampled population. We used the adehabitat package
(Calenge 2006) in R (R Development Core Team 2009). The
analyses used the previously stated Manly’s selection ratios
per territory, where the feeding behaviour within a territory
was compared with its riparian stand composition. The final
ordination was projected on the basis of the two best factorial
axes. Their explanatory contribution was equal to the White–
Garrott statistic (White and Garrott 1990) of the axis per the
sum of all White–Garrott statistics in the analysis. The signif-
icance of an axis’sWhite–Garrott statistic was evaluated using
aχ
2
distribution, with degrees of freedom matched to the
number of resulting axes.
Feeding pressure on highly preferred poplars and willows
We summarised the relationship between the availability and
selection indices of the two main positively selected genera:
willows and poplars (all following analyses were carried out
separately for the two woody vegetation genera). First, we
tested whether the variability of nonzero values of Jacob’s
selectivity indices in territories varied among sites. Due to
unequal variances and non-normal distribution, we used the
nonparametric Kruskal–Wallis test. We used Bonferroni’scor-
rection for multiple comparisons. Further, we tested whether
there was any significant effect among the studied sites in
terms of availability of poplars and willows (each genus was
analysed separately). We used analysis of covariance, where
the response variable was the preference index (with log-
normal transformation) and the independent variable was the
availability of woody vegetation at the selected site as
covariate.
Results
Territory utilisation—population sample size
The number of beaver territories was evaluated in 875 km of
aquatic environments (total availability of water bodies). We
identified 207 recently used territories here, and the numbers
of territories in the estimated sites ranged from 11 to 79. The
density of estimated territories was similar (number of
territories/length of water bodies, mean=0.23; range= 0.21–
0.27), as was the rate of occupied water bodies from the total
available water ecosystems (sum of territory length/water
body length, mean= 0.39; range=0.31–0.52). The riparian
vegetation was estimated across a total of 599 km. Although
1226 Behav Ecol Sociobiol (2015) 69:1221–1235
the proportions of the estimated territories in the studied sites
were unbalanced, the ratios of the territories to the analysed
riparian vegetation in a site were always greater than 42 %
(sampled/total, mean=53 %; range=42–100 %), with a min-
imum of 11 territories at one site and a maximum of 33 terri-
tories at one site.
Comparison of vegetation composition among sites
At the studied sites, we selected 173 strips of riparian vegeta-
tion to compare the compositions of vegetation among areas.
In the established strips, we detected 22 genera of woody
vegetation, which allowed us to conduct the comparisons.
The correspondence analysis revealed significant differences
in the vegetation compositions of the surveyed sites (Monte
Carlo test with 999 permutations, F=16.50,P= 0.01). The first
axis, representing vegetation composition among studied
sites, explained 28.2 % of variability in the reduced model.
The vegetation patterns showed clear differences among ripar-
ian stands along the occupied water bodies (Fig. 2).
We identified three diverse vegetation clusters (similar hab-
itats) among the five studied sites. Similarity was detected
among productive forests in low-lying flood plains, i.e. sites
ND and SP (for abbreviations, see Table 1). Woody vegetation
composition was also similar among sites near rivers with
higher water level fluctuation, i.e. sites LA and CH. The com-
position of riparian vegetation in the site with small brooks at
higher elevation (CL) differed largely from those of all other
sites.
Beavers’consumption
We evaluated 11,980 patches of vegetation alongside water
bodies settled or actually unused by beavers; we determined
38 genera of woody vegetation to be available for beaver
consumption here. In the patches, we found 31,102 gnawed
trunks or branches, and cuts were found only on 22 of the 38
available genera, suggesting 16 genera were avoided. Ob-
served cuts were considered to have been consumed by bea-
vers (we were unable to evaluate the proportion of actually
consumed to unconsumed grazing). Available deciduous
woody vegetation comprised 96.1 % of the riparian stands
studied. Although coniferous trees represented 3.89 % of the
stands, their usage by beavers did not exceed 0.01 % of the
total gnawed trunks or branches (Table 3). The volume of
fallen woody vegetation determined by February 2008 repre-
sented 668,569 kg of usable biomass (bark, cambium, thin
branches and foliage).
The percentages of willowsand poplars selected among the
studied sites varied, ranging from 23.8 to 73.1 % for willows
and from 16.3 to 54.6 % for poplars (Table 3). The combined
usage of these two genera as a percentage of the totals in all
studied sites varied from 78.4 to 94.4 % (Fig. 3, Table 3). By
contrast, the combined proportions of both genera within the
available vegetation composition varied in the range of 13.3–
78.9 %. Willows predominated in three of the studied sites,
while poplars were the second most abundant woody vegeta-
tion at two sites (Table 4).
Preference indices
The detailed feeding behaviour comparisons of available and
used sources at the territorial level showed a similar foraging
pattern. When Jacob’s indices were used, willow and poplar
values never fell below zero. Hence, both genera were always
positivelyselected; the values of the indices were in the ranges
of 0.31 to 2.10 for willows and 0.82 to 4.58 for poplars. The
strong preferences for both genera occurred everywhere and
did not depend on the variations of both stands with regard to
available resources (Table 4). There was no evidence to indi-
cate that, in the studied sites, any other abundant genera were
positively preferred. In two studied sites, beavers preferred
common hazel (Corylus spp.) and common hawthorn
(Crataegus spp.). At one site, Prunus spp. and rowans (Sorbus
spp.) were positively selected (Table 4). Each of these four
-2 3
-1 4
Salix
Populus
Alnus
Acer
Fraxinus
Quercus
Carpinus
Ulmus
Tilia
Rosa
Crataegus
Corylus
Sambucus
Cornus
Picea
Prunus
Euonymus
Betula
Sorbus
SP
CL
LA
ND
CH
A2
A1
Fig. 2 Canonical correspondence analysis (CCA) of relationships of
riparian vegetation similarity or dissimilarity in the studied sites.
Ordination was acquired on the basis of CCA of regularly spaced
sampled patches (for more details see methods). Similarity of studied
sites (black circles) is based on trees’relative abundance (triangles), the
picture shows projection of the data with use of first two axes (axes A1
and A2 are evident within the picture) having highest explanatory values.
Evident three clusters defined by similar distribution at riparian
vegetation level: flood plain forests (ND Niva Dyje, SP Soutok–Podluží),
rivers with intense water level fluctuation (LA Labe, CH Chropyňský luh)
and sub-mountain watercourses (CL Český les)
Behav Ecol Sociobiol (2015) 69:1221–1235 1227
less preferred genera represented less than 1 % of the total
riparian vegetation composition.
Eigenanalysis showed that foraging in the majority of terri-
tories was ordered along the first two axes, which corresponded
with distribution of willows and poplars (Fig. 4a–e). The ex-
plained variability of the model based on the direction of willow
and poplar vectors among sites varied between 85.8 and
99.7 %. When we tested all territories together (Fig. 4f), their
distribution pattern was also explained by the first two axes,
which comprised 84.9 % of model variability. Only 7.2 % of the
territories (8 out of 110) were found in quadrants that did not
correspond with the distribution of willows and poplars.
Feeding pressure on highly preferred poplars and willows
Willow stands were not present as an available resource in
2.7 % (3 of 110 territories in total) of established territories,
while poplars were absent in 14.5 % (16 of 110). Only two
0%
10%
20%
95%
100%
CL CH LA ND SP ALL
Site
ssamoibtnecreP
Salix spp. & Populus spp.
Ulmus spp.
Quercus spp.
Prunus spp.
Fraxinus spp.
Betula spp.
Alnus spp.
Acer spp.
Fig. 3 Proportions of main
woody species used at study sites
as food (estimated biomass).
Proportions on bars represent the
most foraged woody species.
Shown are only those species the
proportions of biomass from
which relative to the totals for all
sites exceeded 1 %. The willows
and poplars are combined into a
single category. Proportions are
shown per each studied site (CL
Český les, CH Chropyňský luh,
LA Labe, ND Niva Dyje, SP
Soutok–Podluží) and for all study
sites together (ALL)
Tabl e 3 Comparison of woody vegetation genera available and used
Salicaceae Remaining genera Number of genera Number of preferred genera
Available Used Available Used Available Used Salicaceae Rem. Gen.
CL 13.34 % 78.40 % 84.80 % 21.59 % 15 12 2 (20+24) 4 (3+3+2+3)
CH 46.83 % 92.23 % 50.57 % 7.76 % 20 9 2 (14+8)
LA 78.99 % 94.41 % 22.87 % 3.60 % 16 5 2 (4+4) 1 (2)
ND 17.34 % 78.54 % 83.02 % 18.60 % 20 11 2 (17+16)
SP 28.40 % 78.71 % 69.95 % 21.27 % 20 13 2 (25+ 17) 1 (1)
Shown separately are percentages of willows and poplars (combined into one value) and all other woody vegetation (combined into a second value). Also
shown are numbers of genera other than willows and poplars available and used. Finally, there is summarized how many genera (within the utilised
territories) were preferred (for Salicaceae and other genera separately), with the values in brackets indicating the number of territories where particular
genera were preferred
1228 Behav Ecol Sociobiol (2015) 69:1221–1235
territories had neither willows nor poplars. A high proportion
of all available willow stands in the studied sites was included
in the established territories (in the range of 61.1–83.0 %).
Poplars covered a smaller extent of the established territories,
as beaver families included between 32.9 and 76.5 % of total
poplar stands into the studied sites.
Differences between the availability of poplars and willows
at the selected sites were determined in this study, and vari-
ability of its preference indices was also detected. These
values (particularly within each territory) were sequentially
used for interaction testing. For instance, the variability of
poplars’preference indices was the highest for CL of all sites
(significantly higher in comparison to other sites; Kruskal–
Wallis test, H
4
=36.72, P<0.001). By contrast, the variability
of poplar stands as available resource was the lowest at this
site (for CL significantly lower in comparison to other sites;
Kruskal–Wallis test H
4
=13.87,P=0.008).Theanalysesofthe
interaction between the volume of available poplar and
beavers’feeding behaviour (represented by the territorial
preference indices) revealed a positive interaction
(ANCOVA, F
5,62
=52.55; P<0.001). Poplar consumption sig-
nificantly depended on the resource availability; as the pre-
dominance of poplar decreased, its selection as a resource
increased (Fig. 5a). The availability of willow stands also
differed within studied sites (Kruskal–Wallis test; H
4
=38.11,
P<0.001). Additionally, the variability of willow preference
indices differed significantly among the sites (Kruskal–Wallis
test; H
4
=28.42, P<0.001). The foraging pattern showed on
poplars selection (inverted power of selection to predomi-
nance of available resource) was observed also in willows.
When we tested the interaction between the variation of wil-
low preference indices (per studied site) and the availability of
willow stands (both on territorial level), the intensity of willow
consumption (most apparently for CH and LA) significantly
decreased as the proportion ofwillow woody vegetation in the
riparian stands increased (Fig. 5b).
Tabl e 4 Predominance of woody vegetation according to genus in terms of availability (dom.), consumption by beavers (used), and their Jacob’s
selection indices (ind.) in the five studied sites
Genus CL CH LA ND SP
dom. used ind. dom. used ind. dom. used ind. dom. used ind. dom. used ind.
%% %% %% %% %%
Acer 1.52 0.07 −3.06 8.72 0.41 −3.14 2.67 0.79 −1.23 16.75 3.97 −1.58 13.70 3.76 −1.40
Alnus 42.90 7.97 −2.09 7.45 0.31 −3.25 5.11 0.02 −5.72 0.67 0.15 −1.53 4.87 0.49 −2.35
Betula 15.44 6.51 −0.90 0.66 0.04 −2.80 1.27 –0.08 0.02 −1.21 0.27 –
Carpinus 0.10 0.06 −0.54 0.09 –0.53 –2.66 0.70 −1.36 0.78 0.49 −0.47
Cornus 0.02 * −2.13 2.08 0.05 −3.80 0.11 –1.88 0.02 −4.73 6.77 0.39 −2.91
Corylus§ 0.28 0.38 0.37 0.21 0.07 −1.13 0.47 1.77 1.34 0.03 * −3.09 0.25 0.04 −1.86
Crataegus§ 0.02 0.04 0.50 0.45 –0.02 –3.67 0.24 −2.75 0.70 0.88 0.23
Euonymus 0.02 –1.08 –0.14 * −7.10 0.76 * −8.05
Fraxinus ** 2.44 *** 6.93 3.69 −0.66 1.84 * −5.14 27.49 5.22 −1.93 12.28 7.10 −0.61
Juglans 0.49 –0.01 –0.04 –0.51 0.01 −4.50
Picea 23.03 –0.19 0.04 −1.61 0.01 –0.01 –0.08 –
Populus§ 1.39 54.61 4.58 12.36 24.33 0.82 6.52 16.27 1.02 5.28 26.83 1.88 11.88 32.33 1.27
Prunus§ 0.14 0.26 0.69 2.81 2.58 −0.09 0.10 0.02 −1.86 5.66 1.70 −1.25 5.50 3.29 −0.54
Quercus ** 3.44 −0.31 2.12 0.57 −1.33 0.25 –12.82 5.43 −0.94 2.48 0.91 −1.02
Robinia 0.84 –1.98 –0.13 –1.33 0.07 −2.94
Rosa 0.01 –1.20 –2.18 –0.67 * −4.71 1.04 * −7.27
Salix§ 11.95 23.79 0.91 34.47 67.90 1.39 72.47 78.14 0.31 12.06 51.71 2.06 16.52 46.38 1.48
Sambucus 0.58 0.06 −2.20 9.53 –0.15 –0.61 0.00 −7.89 8.35 0.02 −6.29
Sorbus§ 0.31 0.35 0.19 0.03 –0.00 –0.02 –
Til ia 0.11 * −3.60 3.29 * −5.72 2.73 * −5.42 3.61 0.80 −1.53
Ulmus 0.30 0.02 −2.52 1.11 –4.00 1.01 −1.41 6.30 1.15 −1.75 5.53 3.02 −0.63
Vib ur num 0.12 –0.02 * −5.16 0.06 –
Indices ofpositively selected trees (index >0) are shown in italics, while an index ofa positively selected genuswith minority predominance (under 1%)
is marked in italics
Studied sites are CL Český les, CH Chropyňský luh, LA Labe, ND Niva Dyje, SP Soutok–Podluží
Symbols: § indicate genus at least once positively selected; −proportion of genus used by beavers not detected
*Proportion of genus used by beavers lower than in 0.01 %;
**
Predominance of genera under detectable level,
***
Unable to calculate Jacob’s selection
index
Behav Ecol Sociobiol (2015) 69:1221–1235 1229
Discussion
We have reported beavers’foraging sampled in dissimilar
habitats. Although the available resources in our sites
consisted of many types of woody vegetation commonly re-
ported as being preferred by beavers (Hall 1960;Novak1987;
Nolet et al. 1994; Haarberg and Rosell 2006), we focused on
only six genera in this study. Additionally, poplars and
willows comprised the majority of the total grazed woody
vegetation, regardless of their availability within the riparian
stands. The preference indices for both of these genera in-
creased as their availability decreased in the riparian vegeta-
tion. In our studied sites, willows and poplars were still highly
preferred regardless of the composition of the riparian stands,
and foraging of these genera increased proportionally with
their increasing rarity.
Grazing on willows and poplars has been broadly studied,
and beavers’selectivity of both genera has been asserted;
many authors have sampled the foraging behaviours of North
American beavers: Novakowski (1966), Brenner (1967),
Aleksiuk (1970), Jenkins (1975) and Jenkins (1980)aswell
as of Eurasian beavers: Wilsson (1971), Nolet et al. (1994)and
Haarberg and Rosell (2006). The results of these studies, how-
ever, describe the beavers’preferences within only one select-
ed ecosystem. Furthermore, their sample sizes usually repre-
sent small numbers of territories. A few more extensive stud-
ies on Eurasian beavers have been conducted. For instance,
Danilov and Kan’shiev (1983) based their foraging report on
hundreds of beavers in Russia; Erome and Broyer (1984)
studied a large-scale settlement in the Rhône Valley; and
Heidecke (1989) reported findings with regard to woody veg-
etation felling of 825 individuals on the River Elbe. Nonethe-
less, in these studies, it is not clear whether documented for-
aging comprised preference indices or only showed frequency
of the vegetation selection. Thus, a comprehensive study of
the plasticity of beaver foraging has not been conducted for
either Eurasian or North American beavers.
To test opportunistic feeding strategies in beavers, we se-
lected sites containing diverse habitats with markedly different
riparian forest compositions. Our goal was to study the inter-
actions among beaver foraging patterns and different resource
availabilities in habitats. While the relative proportions of gen-
era of woody vegetation varied among studied sites (Table 4
and Fig. 3), beaver consumption patterns did not proportion-
ately depend on the abundance of woody vegetation. Outside
of willows and poplars, only two genera were positively se-
lected in more than one site (hawthorns and hazels), though
their existence at both sites was minor. Although both of these
genera were also present in the other sites, they were negative-
ly selected in all except for the two aforementioned sites.
Thus, it is likely that their selection was occasional and did
not indicate any general foraging pattern (rowans were pre-
ferred in 5.5 % of all territories and hazels in 8.2 %). Similarly,
wecouldmakethesameassumptionwithregardtothe
3
7
11
17
22
26
Acer
Carpinus
Crataegus
Fraxinus
Populus
Prunus
Quercus
Salix
Cornus
Tilia
Ulmus
Acer
Alnus
Padus
Populus
Prunus
Quercus
Robinia
Rosa
Salix
Sambucus
Cornus
Tilia
Ulmus
1
6
Fraxinus
Acer
Alnus
Betula
Fraxinus
Rosa
Salix
Ulmus
1 2
5
9
Robinia
Populus
Acer Alnus
Euonymus
Fraxinus
Padus
Prunus
Quercus
Rosa
Salix
Sambucus
Cornus
Tilia
Ulmus
Populus
Acer
Alnus
Betula
Padus
Picea
Populus
7
19
Salix
Acer
Alnus
Betula
Carpinus
Crataegus
Populus
Prunus
Quercus
Robinia
Rosa
Salix
Sambucus
Cornus
Tilia Ulmus
1
2
3
4
6
7
8
9
12
13
22
23
24
26
27
28
31
32
33
34
36
37
39
40
41
43
47
54
58
59
64
84
99
Fraxinus
ab
cd
ef
0%
50%
100%
0%
50%
100%
0%
50%
100%
0%
50%
100%
0%
50%
100%
0%
50%
100%
A1 A2 A3 A4 A5 A6 A7
A1
A1 A2 A3 A4 A5 A6 A7
A1 A2 A3 A4 A5 A6 A7 A1 A2 A3 A4 A5 A6 A7
A1 A2 A3 A4 A5 A6 A7 A1 A2 A3 A4 A5 A6 A7
A2
A1
A2
A1
A2
A1
A2
A1
A2
A1
A2
Fig. 4 Eigenanalysis combining
ordination of territories foraging
with woody vegetation availability
in riparian forests. Feeding
behaviour of beavers (for each
utilized territory—represented by
empty rectangle) is arranged with
distribution of willows (Salix spp.)
and poplars (Populus spp.). Picture
shows projection of the data with
use of first two axes (axes A1 and
A2 are evident within the picture),
where the explanation of the
eigenanalysis by the first two axes
(marked as A1 and A2) is at highest
levels. Graph in the corner of each
ordination presents the explanation
level (White–Garrott statistics) for
each axis of the model (bars
labelled by A1–A7). All five
studied sites are shown individually
and in total (aCL Český les, bCH
Chropyňský luh, cLA Labe, dND
Niva Dyje, eSP Soutok–Podluží, f
all study sites)
1230 Behav Ecol Sociobiol (2015) 69:1221–1235
observed preferences for Prunus spp. and rowans (each one
preferred only at one site). Although Prunus spp. were much
more abundant in the CH, ND and SP sites, they were much
more frequently unselected than selected; beavers preferen-
tially foraged Prunus spp. only in CL (in only two territories),
even though they were rare here. It seems that occasionally
beavers exhibit a preference for sparsely occurring species.
Even in locations where Prunus spp. were more abundant,
for example, in the majority of territories, the genus was not
selected (in ND and SP); however, where it was less available
we evidenced selection exceptionally (e.g. in CL site). Inci-
dentally, this demonstrates the purpose of using preference
indices. While on the one hand the indices might show forag-
ing patterns on a broad level; on the other hand, they can
uncover infrequent deviations and atypical individual behav-
iour that is partly hidden. Otherwise, in general, only two
genera appeared to have important roles in beavers’foraging
patterns: poplars and willows were positively selected at all
sites, and the proportions of both genera in overall consump-
tion were major—in total, more than 80 % of the biomass
utilised by the individuals was represented by willows and
poplars (Fig. 3). Haarberg and Rosell (2006) also indicated a
preference for willow consumption, though alders were pre-
dominately consumed (although not positively selected). In
our site CL (environmentally similar to southern Norway hab-
itats), alders were dominant in riparian areas, but alder con-
sumption was low and the species appeared to be avoided by
beavers. In terms of estimated biomass consumption in CL,
we found willows and poplars to be the most foraged woody
vegetation (sum of consumption was almost 79 %), while
other genera (including alders) comprised 20 % of consumed
vegetation. High consumption and selectivity of ashes or ma-
ples have been mentioned by many authors (e.g. Belovsky
1984;Heidecke1989; Müller-Schwarze et al. 1994 (only for
the genus Acer,althoughforAcer rubrum they documented
strong avoidance)). At our sites, these species were noticeably
foraged, but their consumption was never substantial and both
genera were more often avoided than they were selected (see
Table 4). The reason for this discrepancy between this study
and previously published studies probably lies in the previous
small sample sizes and in the non-homogeneous distribution
of maples and ashes in riparian forests in comparison to the
broadly distributed Salicaceae (a taxon more adapted to and
evolved for riparian forests). Furthermore, the effect of less
preferred ashes and maples in our habitats is probably not so
evident due to the relatively sufficient availability of the two
dominant genera, willows and poplars.
The beavers in our study utilised the willow and poplar
stands in the studied territories as much as possible. The bea-
vers inhabited riparian stands that included either poplars or
willows or both (in 99.1 % of established territories). Territory
delimitation was determined by the distribution of both woody
vegetation and the allocation of beavers across the studied
sites. Thus, beavers also selected localities wherein willows
and poplars prevailed among riparian vegetation. In utilised
territories, however, there was no crossover in the sense of
significant reorientation of foraging behaviours towards more
abundant but less nutritious sources; no such effect was ob-
served at sites with either lower poplar or willowavailabilities.
In addition to these genera, beavers did not exhibit any global
pattern for allocation of the other genera in riparian forests.
Inclusion of the specific (ideal) riparian forests in the popula-
tion territorial system might help beavers to increase and
CL CH LA ND SP
01530456075
Site
1234
0
log (Jacobs' index)
dom (log {Populus})
CL CH LA ND SP
020406080100
Site
123456
dom (log {Salix})
log (Jacobs' index)
a
b
Fig. 5 Interaction of feeding pressure on species with its availability in
the riparian stands at the studied sites (shown separately for poplars (a)
and willows (b)). Feeding pressure (selectivity index) is low where the
species (poplars or willows) is abundant (highly dominant). With
decreasing dominance (species is rarer), the foraging of the resource
(selectivity index) is increasing. For each studied site (CL Český les,
CH Chropyňský luh, LA Labe, ND Niva Dyje, SP Soutok–Podluží),
two metrics (white and black boxplots) are always shown. Grey bars
indicate median value for each metric. White boxplots (shifted to
normalized logarithmic scale) indicate feeding pressure expressed as
Jacob’sindex.Black boxplots (shifted to normalized logarithmic scale)
indicate availability shown as proportional dominance of stand relative to
all riparian vegetation. Error bars representing variation of values within
95 % range out of total
Behav Ecol Sociobiol (2015) 69:1221–1235 1231
simplify the optimisation of their food choice, which would
contrast the findings of Holt and Kimberell (2007), who stated
that an optimal forager is more likely to generalise its feeding
behaviour if the supply of selected resources decreases with
usual variations in environmental conditions; this scenario
would likely be valid until population density markedly in-
creases. By contrast, Müller-Schwarze and Schulte (1999)
documented shifts in foraging patterns due to density-
dependent processes in the population.
The exclusive position of willows and poplars within
the beaver diet is probably based on their higher
digestibility, especially in the case of poplars. Doucet
and Fryxell (1993) documented their shorter retention
time when compared to other genera. Hall (1960)
showed how consumption of overgrazed poplars is con-
tinuously replaced by proportionally higher consumption
of willows. His explanation for this is based on slower
re-sprouting of poplars in comparison to willows. More-
over, the Salicaceae contain much greater volumes of
nutrients in comparison to other woody vegetation
(Nolet et al. 1994). Although in general beavers have
energetically poorer food when depending only on
woody vegetation (Aleksiuk and Cowan 1969;Wilsson
1971), additive consumption of other than woody vege-
tation (more common from spring to early autumn,
according to Wilsson 1971 and Kollar and Seiter
1990) probably provides more accessible nutritive com-
ponents (e.g. protein, sodium and iron (Nolet et al.
1994; Ganzhorn and Harthun 2000)). Thus, higher con-
sumption of Salicaceae (over other woody species)
could reduce the nutrient deficit (occurring mainly dur-
ing winter). There might be a strong positive association
between beavers and poplars and willows. This could be
considered, too, as a result of co-evolution within a
prey–predator interaction. In general, woody vegetation
reacts to intense grazing both by improving biomass
regrowth (i.e. increasing initiation of sprouts) and
further by releasing phytohormones that inhibit con-
sumption by grazers (reviewed by Bryant et al. 1991).
The described reactions probably evolved during coex-
istence of the prey and its predators. The digestibility of
Salicaceae is very favourable for beavers (Doucet and
Fryxell 1993), but the woody vegetation resists too-
intensive exploitation by a strong defence that involves
the growth of repulsive young sprouts (Boege and Mar-
quis 2005). Basey et al. (1988,1990) provided evidence
that beavers are selective in relation to the age of poplar
growth they consume due to the production of second-
ary metabolites by juvenile poplar re-sprouts.
The observed narrowing of the diet (strong selection of
poplars and willows) might also have an adaptive behav-
ioural explanation, based on Belovsky’stheory(1984)
that beavers are energy maximisers. First, there is a
trade-off between foraging yield (net energy intake per
time for allocation and acquisition) and energy losses
(handling time restricted by predation risk). As postulated
by Stephens and Krebs (1986), if the preferred food
source is highly abundant, the individuals focus on its
consumption and the animal emerges as a food specialist.
The more the primary food source varies in abundance
and fails to meet the energy requirements of the individ-
uals, the more the individual is obliged to use another and
less adequate food source, thereby adopting more gener-
alist foraging behaviours. The appearance of a transition
point at which feeding behaviour moves from generalist
to specialist strategy depends upon access to a high-
quality and abundant food item (as measured by the
benefit/handling time ratio). Regarding handling time,
the proximity or ease of access generally is crucial. This
might be expressed as distance of the food from water
(Belovsky 1984) but also as hardness of the cut trees
(Müller-Schwarze et al. 1994). If one regards poplars
and willows as high-quality food for beavers (documented
for North American beavers by Doucet and Fryxell 1993;
Tabl e 5 Feeding behaviour observed in studied sites in comparison to other published results across Europe where similar patterns of available woody
vegetation have been found
Referred article State Netherlands France Norway Germany
Region Biesbosch Rhona Valley Telemark County Steckby, Magdeburg
Cited from Nolet et al. (1994) Erome and Broyer (1984) Haarberg and Rosell (2006) Heidecke (1989)
Predominant woody veg. Willows Willows, poplars Alders, rowan,
willows, birches
Willows, poplars,
alders, oaks, elms
Preferred woody veg. Willows Dogwood, poplars, willows Willows, rowans, birches Poplars, willows
This study Similarity to our study site LA CH CL ND
Predominant woody veg. Willows Willows, poplars Alders, birches, willows Ashes, maples,
oaks, willows
Preferred woody veg. Poplars, willows Willows, poplars Poplars, willows Willows, poplars
All genera are given in descending order according either to their predominance or preference
Studied sites are CL Český les, CH Chropyňský luh, LA Labe, ND Niva Dyje
1232 Behav Ecol Sociobiol (2015) 69:1221–1235
Fryxell and Doucet 1993;Veraartetal.2006), then the
beavers at our sites demonstrated feeding behaviour strat-
egy that narrows diet to only these genera (and they also
select sites with sufficient availability of the Salicaceae).
We were not able to evaluate the proportion of the
winter diet that was represented by aquatic plants.
Svendsen (1980) has empirically quantified the propor-
tion of woody to nonwoody food as 25:6.2 kg per
month in the winter diets of North American beavers.
Although he and other American authors have con-
firmed that aquatic plants are an important source of
energy even during winter (supplementing woody vege-
tation), this notion has not been clearly demonstrated
for Eurasian beavers. Furthermore, our decision to as-
sess winter food diets consisting entirely of the woody
vegetation component was also based on regional evi-
dence. In analysing beaver faeces, Krojerová-Prokešová
et al. (2010) found evidence only of rare utilisation of
just one aquatic plant in a central European sample (the
sites sampled in that study and the present study are
nearly identical). Although several authors have sug-
gested that aquatic plants represent an important food
item for Eurasian beavers (e.g. Heidecke 1989), infor-
mation regarding its proportion in their diet during win-
ters is lacking. For North American beavers, however,
Northcott (1971) demonstrated that aquatic plants are an
important component of beavers’diets, though he also
showed that this source plays an important role in the
diet mainly from spring to autumn.
The foraging behaviours studied here were evaluated
only at the territorial level (110 territories), thereby ex-
cluding the possibility of identifying an individual’sdiet
choice (design-I of resource utilisation) (Thomas and
Taylor 2006). The assessed feeding behaviour for each
territorial unit (containing animals grouped into the
unique territory) represents the foraging pattern for all
members of the beaver family. Thus, the described for-
aging pattern represents the behaviours of approximately
400–500 animals (assuming mean abundance per territo-
ry is 4–5 individuals (Zurowski and Kasperczyk 1986;
Rosell and Parker 1995; Campbell et al. 2005;AVetal.
unpubl. data)). Although some local and individual dif-
ferences among the studied animals may exist (Bolnick
et al. 2003;Egasetal.2005), the common behaviour
pattern was unambiguous in our large sample. Our anal-
yses are supported by our use of identical methodolo-
gies among diverse habitat conditions, as we obtained
comparable results showing common beaver foraging
behaviours under varying environmental conditions. Be-
cause the majority of existing articles discussing the
foraging patterns of Eurasian beavers have examined
food selection in one region with a constant composi-
tion of riparian vegetation, we sought to strengthen our
results across a broader European range. We compared
the results observed at our studied sites with articles
describing studies carried out in habitats with similar
compositions. Our comparisons (Table 5) showed simi-
lar foraging patterns among comparable sites, which im-
ply that beavers found in regions comparable with our
sites (within diverse habitats across Europe) behave in a
manner similar to that which we observed.
Conclusions
As our data indicate a low proportion of generality in the
foraging behaviours of beavers, we conclude that during
overwintering, beavers preferentially selected willows and
poplars across all studied sites. Both genera also constituted
a major portion of the total volume of consumed biomass.
This observation resulted from the methodology applied in
all studied sites, wherein we derived equations for
summarising the biomass of cut woody vegetation. The com-
bined usage of biomass of poplars and willows as a percentage
of the totals at all studied sites varied between 78.4 and
94.4 %. The foraging pattern also indicated the inclusion of
poplar and willow stands in utilised and defended territories.
We have demonstrated a general pattern of feeding behaviour
that may occur under conditions found across a wide range of
habitats within continental Europe. While there were devia-
tions from the general feeding strategy described in our study
(e.g. due to different individual diet preferences or locally
diverse compositions of riparian stands), these deviations
were local and minor. Moreover, the comparisons of diet in
varying environments showed very similar patterns. We veri-
fied this foraging pattern in a sample of approximately 500
beavers in five diverse populations. In view of our large sam-
ple size, we contend that during overwintering, beavers exhib-
it consistent and narrow spatial and foraging for specific
woody species, such as willows and poplars. Contrary to the
view of beavers as opportunistic, such feeding behaviour
could indicate much greater specialisation with regard to these
two genera.
Acknowledgment We would like to thank to our main field colleague
J. Korbel. Also, our thanks go to K. Konečná for her English reviews and
to J. Douda for his ordination analysis. For assistance in the field, we
thank S. Baker, M. Cvoligová, V. Dvořák, K. Fuksová, J. Matrková, M.
Mazalová, M. Pihera, V. Spurná, H. Slezáková, J. Šimčík, and K.
Šimůnková. This research material has been produced with grant support
from the Ministry of Environment of the Czech Republic (SP/2d4/52/07)
and support of The Agency for Nature Conservation and Landscape Pro-
tection of the Czech Republic.
Ethical standards All experiments comply with the current laws of the
Czech Republic. We did not use to manipulation with animals demanding
special issues of Agency for Nature Conservation and Landscape
Behav Ecol Sociobiol (2015) 69:1221–1235 1233
Protection of the Czech Republic (state nature conservation authority of
the Czech Republic).
References
Aleksiuk M (1970) The seasonal food regime of arctic beavers. Ecology
51:264–270
Aleksiuk M, Cowan IMT (1969) The winter metabolic depression in
arctic beavers (castor canadensis kuhl) with comparisons to
California beavers. Can J Zool 47:965–979
Barták V, Vorel A, Šímová P, PušV (2013) Spatial spread of Eurasian
beavers in river networks: a comparison of range expansion rates. J
Anim Ecol 82:587–597
Basey JM, Jenkin SH, Busher PE (1988) Optimal central-place foraging
by beavers: tree-size selection in relation to defensive chemicals of
quaking aspen. Oecologia 76:278–282
Basey JM, Jenkins SH, Miller GC (1990) Food selection by beaver in
relation to inducible defenses of Populus tremuloides. Oikos 59:57–
62
Belovsky GE (1978) Diet optimization in a generalist herbivore: the
moose. Theor Popul Biol 14:105–134
Belovsky GE (1984) Summer diet optimization by beaver. Am Midl Nat
111:209–221
Belovsky GE (1986) Generalist herbivore foraging and its role in com-
petitive interactions. Am Zool 25:51–69
Boege K, Marquis RJ (2005) Facing herbivory as you grow up: the
ontogeny of resistance in plants. Trends Ecol Evol 20:441–448
Bolnick DI, Svanbäck R, Fordyce JA, Yang LH, Davis JM, Hulsey CD,
Forister ML (2003) The ecology of individuals: Incidence and im-
plications of individual specialization. Am Nat 161:1–28
Brenner FJ (1967) Spatial and energyrequirements of beavers. Ohio J Sci
67:242–246
Bryant JP, Danell K, Provenza F, Reichardt PB, Clausen TA, Werner RA
(1991) Effects of mammal browsing upon the chemistry of decidu-
ous woody plants. In: Tallamy DW, Raup MJ (eds) Phytochemical
induction by herbivores. Wiley, New York, pp 134–154
Calenge C (2006) The package adehabitat for the R software: a tool for
the analysis of space and habitat use by animals. Ecol Model 197:
516–519
Calenge C, Dufour AB (2006) Eigenanalysis of selection ratios from
animal radio-tracking data. Ecology 87:2349–2355
Campbell RD, Rosell F, Nolet BA, Dijkstra VAA (2005) Territory and
group sizes in Eurasian beavers (castor fiber): Echoes of settlement
and reproduction? Behav Ecol Sociobiol 58:597–607
Crawley MJ (1983) Herbivory: the dynamics of animal-plant interactions,
1st edn. Blackwell Scientific Publications, Oxford
Danilov PI, Kan’shiev VY (1983) The state of populations and ecological
characteristics of european (castor fiber L.) and Canadian (castor
canadensis kuhl.) beavers in the northwestern USSR. Acta Zool
Fenn 174:95–97
R Development Core Team (2009) R: A language and environment for
statistical computing, URL: http://www.R-project.org
Doucet CM, Fryxell JM (1993) The effect of nutritional quality on forage
preference by beavers. Oikos 67:201–208
Egas M, Sabelis MW, Dieckmann U (2005) Evolution of specialization
and ecological character displacement of herbivores along a gradient
of plant quality. Evolution 59:507–520
Erome G, Broyer J (1984) Analyse des relations castor-vegetation. Bièvre
6:15–63
Fryxell JM, Doucet CM (1993) Diet choice and the functional response of
beavers. Ecology 74:1298–1306
Fryxell JM, Vamosi SM, Walton RA, Doucet CM (1994) Retention time
and the functional response of beavers. Oikos 71:207–214
Ganzhorn JU, Harthun M (2000) Food selection by beavers (castor fiber
albicus)inrelationtoplantchemicals and possible effects of
flooding on food quality. J Zool 251:391–398
Green JS, Flinders JT (1980) Habitat and dietary relationships of the
pygmy rabbit. J Range Manag 33:136–142
Gross JE, Shipley LA, Thompson Hobbs N, Spalinger DE, Wunder BA
(1993) Functional response of herbivores in food-concentrated
patches: Tests of a mechanistic model. Ecology 74:778–791
Haarberg O, Rosell F (2006) Selective foraging on woody plant species
by the Eurasian beaver (castor fiber) in telemark, Norway. J Zool
270:201–208
Hall JG (1960) Willow and aspen in the ecology of beaver on sagehen
creek, California. Ecology 41:484–494
Halley DJ, Rosell F (2002) The beaver’s reconquest of Eurasia: Status,
population development and management of a conservation success.
Mammal Rev 32:153–178
Heidecke D (1989) Ökologische bewertung von biberhabitaten.
Saugetierkd Inf 3:13–28
Herrera CM, Pellmyr O (2002) Plant-animal interactions. Wiley,
Blackwell, Totnes Devon
Holt RD, Kimberell T (2007) Foraging and population dynamics. In:
Stephens DW, Brown J, Ydenberg R (eds) Foraging, behavior and
ecology. The University of Chicago Press, Chicago, pp 365–396
Hutchinson GE (1957) Concluding remarks. Cold Spring Harb Sym 22:
415–427
Jacobs J (1974) Quantitative measurement of food selection. A modifi-
cation of the forage ratio and Ivlev’s electivity index. Oecologica 14:
413–417
Jenkins SH (1975) Food selection by beavers. A multidimensional con-
tingency table analysis. Oecologia 21:157–173
Jenkins SH (1979) Seasonal and year-to-year differences in food selection
by beavers. Oecologia 44:112–116
Jenkins SH (1980) A size-distance relation in food selection by beavers.
Ecology 61:740–746
Jenkins JC, Chojnacky DC, Heath LS, Birdsey RA (2004)
Comprehensive database of diameter-based biomass regressions
for North American tree species. U.S. Department of Agriculture,
Forest Service, Northeastern Research Station,Newtown Square, PA
Kollar HP, Seiter M (1990) Biber in den Donau-Auen östlich vor Wien.
Eine erfolgreiche Wiederansiedlung. Verein für Ökologie und
Umweltforschung, Wien
Krojerová-Prokešová J, Barančeková M, Hamšíková L, Vorel A (2010)
Feeding habits of reintroduced Eurasian beaver: Spatial and seasonal
variation in the use of food resources. J Zool 281:183–193
Law A, Bunnefeld N, Willby NJ (2014) Beavers and lilies: Selective
herbivory and adaptive foraging behaviour. Freshw Biol 59:224–
232
Manly BFJ, McDonald LL, Thomas DL, McDonald TL, Erickson WP
(2002) Resource selection by animals: Statistical design and analysis
for field studies, 2nd edn. Kluwer Academic Publisher, Dordrecht
Milligan HE, Humphries MM (2010) The importance of aquatic vegeta-
tion in beaver diets and the seasonal and habitat specificity of
aquatic-terrestrial ecosystem linkages in a subarctic environment.
Oikos 119:1877–1886
Müller-Schwarze D, Schulte BA (1999) Behavioral and ecological char-
acteristics of a Bclimax^beaver population. In: Busher PE,
Dziecilowski RM (eds) Beaver protection, management, and utili-
zation in Europe and North America. Plenum Press, New York, pp
109–127
Müller-Schwarze D, Schulte BA, Sun L, Müller-Schwarze A, Müller-
Schwarze C (1994) Red maple (acer rubrum) inhibits feeding by
beaver (castor canadensis). J Chem Ecol 20:2021–2034
Muukkonen P (2007) Generalized allometric volume and biomass equa-
tions for some tree species in Europe. Eur J For Res 126:157–166
Muukkonen P, Mäkipää R (2006) Biomass equations for European trees:
Addendum. Silva Fenn 40:763–773
1234 Behav Ecol Sociobiol (2015) 69:1221–1235
Nolet BA, Hoekstra A, Ottenheim MM (1994) Selective foraging on
woody species by the beavercastor fiber, and its impact on a riparian
willow forest. Biol Conserv 70:117–128
Northcott TH (1971) Feeding habits of beaver in Newfoundland. Oikos
22:407–410
Novak M (1987) Beaver. In: Novak M, Baker JA, Obbard ME, Malloch
B (eds) Wild furbearer management and conservation in North
America. Ministry of Natural Resources, Ontario, pp 284–310
Novakowski NS (1966) The winter bioenergetics of a beaver population
in northern latitudes. Can J Zool 45:1107–1118
Panzacchi M, Linnell JDC,Odden J, Odden M, Andersen R (2008)When
a generalist becomes a specialist: Patterns of red fox predation on roe
deer fawns under contrasting conditions. Can J Zool 86:116–126
Pinheiro J, Bates D, DebRoy S, Sarkar D (2007) nlme: Linear and
Nonlinear Mixed Effects Models. R package version 3.1-102,
URL: http://www.R-project.org
Pulliam HR (1974) On the theory of optimal diets. Am Nat 108:59–74
Rosell F, Parker H (1995) Beaver management: present practice and
Norway’s future needs. Telemark University College, Bø i
Telemark, Norway (in Norwegian with English summary)
Severud WJ, Belant JL, Windels SK, Bruggink JG (2013) Seasonal variation
in assimilated diets of American beavers. Am Midl Nat 169:30–42
Smith DW, Peterson RO, Drummer TD, Sheputis DS (1991) Over-winter
activity and body temperature patterns in northern beavers. Can J
Zool 69:2178–2182
Staddon JER (1983) Adaptive behaviour and learning. Cambridge
University Press, Cambridge, UK
Stephens DW, Krebs JR (1986) Foraging theory. Princeton University
Press, Princeton
Svendsen GE (1980) Seasonal change in feeding patterns of beaver in
southern Ohio. J Wildl Manag 44:285–290
Thomas DL, Taylor EJ (2006) Study designs and tests for comparing
resource use and availability II. J Wildl Manag 70:324–336
Veraart AJ, Nolet BA, Rosell F, De Vries PP (2006) Simulated winter
browsingmay lead to induced susceptibility of willows to beavers in
spring. Can J Zool 84:1733–1742
Wand MP, Ripley BD (2013) Kern Smooth: functions for kernel smooth-
ing for Wand and Jones (1995). R package version 2.23-10, http://
CRAN.R-project.org/package=KernSmooth. Accessed 14
March 2014
White GC, Garrott RA (1990) Analysis of wildlife radio-tracking data.
Academic, London
Wilsson L (1971) Observations and experiments on the ethology of the
European beaver (Castor fiber L.). Viltrevy 8:115–266
Woo KJ, Elliott KH, Davidson M, Gaston AJ, Davoren GK (2008)
Individual specialization in diet by a generalist marine predator re-
flects specialization in foraging behaviour. J Anim Ecol 77:1082–91
Zurowski W, Kasperczyk B (1986) Characteristics of a European beaver
population in the Suwalki Lakeland. Acta Theriol 31:311–325
Behav Ecol Sociobiol (2015) 69:1221–1235 1235