Abstract
Resource acquisition is a key factor governing patterns of animal movement. While the link between spatio-temporal resource patterns and movement behaviour has been widely documented for migration, our understanding of home ranging behaviour in relation to resource dynamics has been limited by challenges of quantifying resource heterogeneity at fine spatio-temporal scales. In this study, we addressed this issue by analysing the movement responses of a resident large herbivore in response to an in situ manipulation of a high-quality, concentrated food resource. Specifically, we fitted roe deer (Capreolus capreolus) with GPS collars in the Eastern Italian Alps and recorded their fine-scale movement responses to an alteration of feeding site accessibility, for a total of 25 animal-years. The experiment involved identifying, for each animal, a familiar feeding site, whose accessibility was transitorily restricted (i.e., closed) while maintaining the availability of forage at alternative feeding sites. First, we tested whether individuals altered their spatial behaviour to track dynamical changes in resource availability. Experimental closure led to larger, spatially-shifted home ranges, resulting from more exploratory movements. Individual roe deer compensated the loss of their familiar feeding site by using alternative ones, and by doing so maintained their overall use of these concentrated resources. Our results demonstrate that roe deer actively track resource dynamics, and rapidly adjust their space-use and movement behaviour in response to changes in resource availability. Second, we showed that individual resource preferences mediate these behavioural adjustments: individuals characterized by a high preference for feeding sites exhibited larger space-use shifts and stronger changes in resource use in response to the resource manipulation. In contrast, sex had relatively minor influence on the observed patterns. Third, we investigated the role of site familiarity in roe deer foraging decisions. We found that space-use and resource use during the post-closure period were restored to pre-closure patterns, thereby providing evidence for the inherent benefits of site familiarity to animals maintaining a home range. Together, our results establish the connections between movement, space-use, individual preference, and the spatio-temporal pattern of resources in deer home ranging behaviour.
Introduction
Animals move to change the environmental context they experience (Van Moorter, Rolandsen, Basille, & Gaillard, 2016), including abiotic conditions, the presence of predators and competitors, and the availability of resources. Because foraging efficiency can be linked to individual fitness (Stephens & Krebs, 1986), food acquisition is thought to be a primary driver underlying animal movements (e.g., Mueller & Fagan, 2008). Consequently, space-use represents the geographic realization of optimizing fitness as a function of resource availability and acquisition costs (Mitchell & Powell, 2004).
Food resources are usually dynamic in both space and time (Wiens, 1976). In the case of herbivores, animals typically feed on resources distributed in patches, which are characterized by important temporal variations in quantity and quality (Owen-Smith, Fryxell, & Merrill, 2010). In this context, Mueller et al. (2011) have shown that strong spatio-temporal gradients in resource availability at either landscape or regional scales appear to drive migration and nomadism tactics. In many herbivore populations, however, individuals show a high year-round fidelity to a spatially-localized home range. It has been suggested that the foraging benefits of site familiarity, where resources are constant or predictable, are responsible for the formation of a stable home range (see Fagan et al., 2013 for a review). While the home range has traditionally been perceived as a relatively static space-use tactic, recent evidence suggests that animals have sub-seasonal home ranges (Benhamou 2014), i.e. focus their movements into particular areas in response to seasonal variation in local resource availability. For example, Couriot et al. (2018) have shown that in roe deer (Capreolus capreolus) sub-seasonal home ranges are a ubiquitous behavioural tactic across a wide ecological gradient.
The link between movement behaviour and resource dynamics is less clear when observing home ranging behaviour than migration or nomadism (Mueller & Fagan, 2008), because of the difficulty to quantify spatio-temporal variability in resource heterogeneity at small spatial scales (Couriot et al., 2018). In this study, we address this issue by experimentally manipulating the spatio-temporal patterns of food availability within home ranges. In situ food manipulation experiments have a long history in the study of population dynamics, with a primary focus on understanding the numerical response to food supplementation (e.g., Krebs, 1971), and of animal communities (e.g., Brown & Munger, 1985). Although these field experiments have provided fundamental insights in animal ecology, they have seldom been combined with the emerging technological capabilities of animal tracking (Cagnacci, Boitani, Powell, & Boyce, 2010) to investigate the implications of food availability on individual movements and space-use. In a rare example of field experiment in large herbivores, Kilpatrick and Stober (2002) showed that white-tailed deer (Odocoileus virginianus) shifted their core-area i.e., familiar areas of use, in response to novel food supplementation. In a second experiment, van Overveld and Matthysen (2010) demonstrated that the individual variability in responses to an alteration of resource distribution are personality-dependent in great tits (Parus major). Our research builds upon these two studies by investigating the spatial responses of a large herbivore, roe deer, to an experimental in situ manipulation of a high-quality, concentrated food resource in relation to both individual resource preferences and site familiarity.
Observational studies have suggested that, as browsers with limited fat reserves (Andersen, Gaillard, Linnell, & Duncan, 2000), roe deer exhibit a tight association between movement and resource dynamics (Ossi et al. 2017) with a strong plasticity to adapt its resource acquisition at different spatio-temporal scales (De Groeve et al., 2019; Morellet et al., 2013; Peters et al., 2017). In contrast to group-living ungulates, the foraging decisions of roe deer are expected to be clearly expressed at the level of individuals. The experiment was conducted during winter, when food scarcity limits roe deer foraging performance, and individuals are most inclined to adjust their spatial behaviour to continue meeting their energy requirements (Ossi, Gaillard, Hebblewhite, & Cagnacci, 2015).
We tagged roe deer in the Eastern Italian Alps with GPS units and followed their movements during transitory alterations of food availability at supplemental feeding sites (FS), i.e. discrete resource patches with an identifiable resource value distinguishable from the vegetation matrix (Mitchell & Powell, 2004; Wiens, 1976). The experiment therefore mimics – on free-ranging animals – the variation in the availability of concentrated, high-reward resources akin to watering holes for savannah ungulates (Owen-Smith, 2004) and feral horses (Equus ferus; Rozen-Rechels et al. 2015), or fruit trees for hornbills (Ceratogymna atrata and C. cylindricus; Holbrook et al. 2002), forest chimpanzees (Pan troglodytes; Normand et al. 2009) and other frugivorous primates.
In large herbivores, and roe deer in particular, our initial hypothesis states that individuals alter their movement behaviours and consequently space-use patterns to track dynamics in resource availability (H1; Table 1). We predicted that the loss of a key foraging resource should lead to larger (P1.1), and spatially-shifted (P1.2) home ranges, resulting from more explorative movements (P1.3). Furthermore, we predicted that roe deer reduced the intensity of use of the familiar FS when food accessibility was prevented (P1.4a) and compensated for this loss by using other accessible FS (P1.4b).
We further hypothesized that the behavioural adjustments to changes in resource availability would vary between individuals (H2; Table 1). In particular, because roe deer males have been shown to maintain a high year-round fidelity to their summer territory (Linnell & Andersen, 1998), we predicted that they would respond less markedly to the experiment than females (P2.1). We also predicted the responsiveness of roe deer to be positively influenced by the individual’s prior preference for FS (P2.2).
If the spatial patterns of roe deer home ranges result from the foraging benefits of site familiarity (Riotte-lambert, Benhamou, & Chamaillé-Jammes, 2015; Van Moorter et al., 2009), animals should strive to use familiar areas and resources when accessible (H3; Table 1). Accordingly, we predicted that when initial conditions of food accessibility are re-established after perturbation, the initial space-use patterns would be restored (P3.1), following a return to high use of the familiar FS (P3.2).
Materials and methods
Study area
Roe deer is the primary large herbivore in the study area (7-8 individuals km-2), located in the north-eastern Italian Alps (Argentario range, Autonomous Province of Trento). The area is characterized by a continental climate (mean temperature of January: 1.0 °C; July: 21.0 °C; mean annual rainfall: 966 m) with occasional snow cover, and is largely forested (80%). It encompasses four hunting reserves in which selective hunting occurs between September and December (see Appendix S1 for further details).
Supplemental feeding management of roe deer is conducted year-round at > 50 distinct feeding sites (FS; Appendix S2: Fig. S2.1). FS are typically wooden hopper dispensers that provide a continuous supply of corn accessible through a tray (Fig. 1a). They are managed by private hunters for roe deer but are also attended sporadically by red deer (Cervus elaphus).
Experimental design
We took advantage of roe deer use of a focal, identifiable resource – the FS – to design an in situ experimental manipulation of resource availability. We created three successive experimental phases of the availability of this resource – pre-closure, closure and post-closure – by physically managing the accessibility of food at the FS. During the closure phase, access to forage at FS was transitorily restricted by placing wooden boards obstructing the tray; these were then removed again in the post-closure phase (Fig. 1).
The experiment was conducted between January and April, when roe deer use of supplemental feeding is the most intense (Ossi et al., 2017), for three consecutive winters (2017, 2018 and 2019). We implemented the experiment on 18 individuals, including five recaptures and two deployments spanning two winters, leading to a total of 25 animal-years (21 adults: 15 females, 6 males; 4 yearlings: 2 females, 2 males; sample size n=4, 11 and 10 in 2017, 2018 and 2019 respectively; see Appendix S2 for details). The animal-year was our sampling unit, on the assumption that the same individual may respond independently to manipulations in different years. Roe deer were captured using baited box traps (n=16) or net drives (n=3), and were fitted with GPS-GSM radio collars programmed to acquire hourly GPS locations for a year, after which they were released via a drop-off mechanism. Radio-collared roe deer moved an average of 61.2 m per hour. This value of the average hourly movement distance (l) was subsequently utilized in the analyses described below.
For all captured animals, we assumed a post-capture response in ranging behaviour. We therefore considered the first re-visitation of the capture location as the sign of resettlement in the original range and we used this time as the earliest possible onset of the experimental pre-closure phase. Although not all the individuals were manipulated at the same time, we avoided interference between capture operations and FS manipulations and between co-occurring different manipulation phases (i.e. ensuring that co-occurring manipulations occurred in separate areas).
During the pre-closure phase, we assessed the use of FS by radio-collared roe deer. We identified the “main” FS (M) for each individual as the site with the largest number of locations within a radius 𝑙 during this initial phase, and considered it as the FS to which an individual is most familiar. During the closure phase, corn was made inaccessible at M for a duration of about 15 days, depending on personnel availabilities (min=14.0 days, max=18.1, mean=15.5). M was then re-opened, thereby initiating the post-closure phase. During both pre- and post-closure phases, corn was available ad libitum at M. All “other” (O) managed FS – i.e., that were provisioned at least once in the month prior to the experiment – had corn available ad libitum throughout the experiment.
Data preparation
To ensure meaningful comparisons between animal-years, we homogenized the durations of each experimental phase to the minimum length of the closure phase (i.e., 14 days). Specifically, we truncated the movement data by removing initial excess positions for the pre-closure and closure phases, and terminal excess positions for the post-closure phase. GPS acquisition success was extremely high (99.57 % during the experiment) and we did not interpolate missing fixes in the collected data.
The analyses of space-use and movement behaviour were based on spatially-explicit, raw movement trajectories. The analyses of resource use relied on spatially-implicit, state time series derived from the underlying movement data. To this end, we created an initial time series, for each animal, by intersecting the relocations with three spatial domains: vegetation (the matrix; V), main FS (M) and other FS (O). We converted FS locations (M and O) into areas by buffering them. To investigate the sensitivity of buffer choice we considered six buffer sizes: 𝑙 (i.e., 61.2 m) multiplied by 0.5, 1, 1.5, 2, 3 and 4. We associated all locations falling outside M and O to the state V. The three-state time series was then converted into three single-state presence/absence time series.
Preference for feeding sites
We calculated each individual’s preference for FS (hFS) as the relative use of FS over natural vegetation during the pre-closure phase (i.e., the proportion of GPS fixes classified as either M or O). Because preference is considered to be temporally dynamic (Beyer et al., 2010), we chose to evaluate hFS for each year separately in case individuals were manipulated in two separate years. This reasoning allowed for the influence of individual condition and of the relative quality and quantity of vegetation resources on hFS. We included hFS in all space-use, movement, and resource use analyses described below.
The variability of hFS across animal-years was maximal when FS attendance was defined as a GPS location within a 61.2 m distance (i.e., the population mean hourly step length) from the FS (interquartile range=0.278, mean=0.343; Appendix S3: Table S3.1). Accordingly, the results described below are based on this definition (see Appendix S7 for sensitivity analysis). At this scale, hFS did not differ consistently between sex (mean for females=0.346; mean for males=0.336; t-test: p-value=0.901).
Analysis
We analysed how the experimental manipulation, and its interaction with both preference for FS and sex, affected roe deer space-use, movement behaviour, and resource use.
Space-use
We assessed the changes of home range and core area sizes (P1.1), and of space-use overlap (P1.2, P3.1) between experimental phases. We calculated utilization distributions, (UD; sensu Worton 1989) for each animal-year and experimental phase using a Gaussian kernel density estimation. After visual inspection, we chose to compute the UDs at a spatial resolution of 10 m and with a fixed bandwidth, set to half the average hourly movement distance (i.e. 𝑙/2=30.6 m).
For home range and core area sizes, we calculated the area (in hectares) corresponding to the 95% and 50% UD contours, respectively, for each experimental phase (Phase; three levels; reference level: Pre-closure). We then analysed the log-transformed areas using a linear mixed-effect model (LMM) with five fixed effects: Phase, hFS, Sex (categorical predictor; reference level: Female), and two interaction terms (Phase:hFS and Phase:Sex). We included animal-year (ind) as random effect (intercept). In all analyses, interaction terms were dropped when statistically non-significant (p-value>0.05).
For space-use patterns, we estimated the overlaps for three pairs of UDs – pre- and post-closure, pre-closure and closure, and closure and post-closure (Contrast; three levels; reference level: Pre-/Closure) – using the volume of intersection statistic (VI; Fieberg and Kochany 2005). VI ranges between 0 (no overlap) and 1 (complete overlap). We then analysed the logit-transformed overlaps using an LMM with Contrast, hFS, Sex, Contrast:hFS and Contrast:Sex as fixed effects, and ind as random intercept.
Movement behaviour
We investigated the movement responses of roe deer to the experiment (P1.3) by analysing the changes in hourly step length (Euclidean distance between two successive relocations) and turning angle 𝜃t (angle between two successive movement steps). Turning angles range between −𝜋 and 𝜋, and were symmetric around 0. We analysed the log-transformed step length, 𝑠t, and the logit-transformed absolute turning angle, using LMMs with Phase, hFS, Sex, Phase:hFS and Phase:Sex as fixed effects, and ind as random intercept. Because step length was characterized by strong serial autocorrelation at short lags and at circadian periodicities (a common pattern in movement trajectories; Fieberg, Matthiopoulos, Hebblewhite, Boyce, & Frair, 2010), we also included step length measured at lags 1, 2 and 24 h (i.e., st-1, st-2, st-24:) as fixed effects to reduce the autocorrelation of the model residuals.
Resource use
To test whether the experiment led to a transitory change in resource use (P1.4a-b, P3.2), we fitted separate mixed-effect logistic regression models to the three single-state presence/absence time series (uM,t, uO,t and uV, t) using Phase, hFS, Sex, Phase:hFS and Phase
Sex as fixed effects, and ind as random intercept. The pre-closure level for Phase was dropped for uV to avoid circularity . We also included the response variables measured at lags 1, 2 and 24 h (e.g., uM, t-1, uM, t-2, uM, t-24) as fixed effects to reduce the autocorrelation of the model residuals. However, for the sake of conciseness and clarity, we omit these response lags when visualizing resource use predictions. Because the model results were consistent regardless of the inclusion of the response lags (Appendix S6: Table S6.2), this decision had no impact on the interpretation. Two animal-years were excluded from the analyses of resource use due to the absence of suitable O-state: F4-2017 did not seem to have visited any other FS (O) prior to the experiment; and F16-2016 had two distinct, highly-used FS during pre-closure, but only the second most visited FS could be manipulated (due to stakeholder acceptance). While the use of O was more variable when including these two outliers, the general patterns remained unchanged (Appendix S6: Table S6.3).
Software
All analyses were conducted in the R environment (R Development Core Team, 2016). We used the packages adehabitatLT and adehabitatHR (Calenge 2006) for the spatial analyses, fitted all mixed-effect models via Maximum Likelihood with lme4 (Bates, Mächler, Bolker, & Walker, 2015), and obtained the coefficients of determination using MuMin (Bartón, 2018).
Results
Space-use and movement responses to alteration of resource availability
Roe deer space-use changed significantly during the experiment: the size of both home ranges (95% UD isopleth; Appendix S4: Table S4.1) and core areas (50% isopleth; Table 2) increased significantly during the experimental closure (Fig. 2; P1.1). On average, home range size increased from 27.99 ha (𝜎=11.02) during pre-closure to 34.97 ha (𝜎=10.17) during closure, and settled to 29.40 ha (𝜎=9.27) during post-closure. Core area size followed a similar trend with averages of 4.23 ha (𝜎=2.34), 5.85 ha (𝜎=2.33) and 4.98 ha (𝜎=2.09), respectively.
Home range and core area sizes were influenced by individual preference for FS (hFS) and there was an interaction between hFS and experimental phase: individuals with a high hFS had smaller home ranges during the pre-closure and, overall, smaller core areas than those with lower hFS, but stronger increases following the experimental closure (Fig. 2; Table 2; Appendix S4: Table S4.1; P2.2). There was no significant effect of sex or interactions between sex and experimental phase on home range size (Appendix S4: Table S4.2; P2.1), but a marginally significant interaction between sex and experimental phase on core area size (Appendix S4: Table S4.3) with responses to closure tending to be slightly larger for males. Overall, the models quantifying the changes in observed home range and core area sizes accounted for a high proportion of the total variance (conditional coefficient of determination, cR2: 0.62 and 0.51, respectively).
Alongside home range size, the spatial pattern of roe deer home ranges shifted dramatically following the experimental closure (Fig. 3): the degree of space-use overlap between pre-closure and closure phases was significantly lower (mean=0.370, CI=0.301-0.405; P1.2) than the overlap between the temporally-separated pre- and post-closure phases (mean=0.535, CI=0.475-0.594; P3.1). Space-use overlap was significantly affected by hFS (Fig. 3; Table 3; P2.2), with higher hFS being associated to larger space-use shifts. However, there was no apparent influence of sex in the space-use patterns (Appendix S4: Table S4.4; P2.1). The model predicting space-use overlap accounted for an important proportion of the variance (cR2=0.50).
Underpinning these changes in home range size and space-use patterns were significant changes in roe deer movement behaviour during the experiment. Average hourly step length during the pre-closure phase was 60.32 m (𝜎=85.79); during closure it increased to 74.26 m (𝜎=108.11); and during post-closure it decreased to 68.18 m (𝜎=96.61, P1.3). In general, males (Appendix S5: Fig. S5.1, right-hand panels, Table S5.1; P2.1), and individuals associated with high hFS values (Appendix S5: Fig. S5.1, top panels, Table S5.1; P2.2) were characterized by stronger increases in step length during the closure phase. In addition, roe deer movements were more persistent during the closure phase, as shown by a significant decrease in the mean absolute turning angle for males with a high hFS (Appendix S5: Fig. S5.2, top-right panel, Tables S5.2, 5.3; P1.3, P2.1, P2.2).
Resource use responses to alteration of availability
The spatio-temporal dynamics of resources availability during the experiment led to important shifts in FS resource use (Fig. 4; Table 4). On average, the proportion of use of the main FS (M) dropped from 31% during the pre-closure phase to 4% during closure (P1.4a), and then rebounded to 19% in the post-closure phase (P3.2). This decrease in the use of M during the closure phase was partially compensated by elevated use of the other FS (O) – which increased from 3% to 16% following closure (P1.4b), and an increase of the use of vegetation (V) from 66% to 80% following closure. During the post closure, use of O and V declined to 9% and 72%, respectively. The shifts in resource use were very consistent among animal-years for M and O but were more variable for V (Fig. 4, top panels).
Roe deer preference for FS significantly influenced how animals used the three resource types and, in particular, interacted with experimental phase for M and O (Fig. 4, bottom panels; Table 4). Roe deer characterized by a high hFS had significantly higher use of M during pre-closure (by definition) and post-closure, as well as consistently lower use of V. High hFS animals were associated with stronger decreases in use of M and larger increases in the use of O during closure (P2.2). This compensation for O during closure was stronger for females (Table 4; Appendix S6: Fig. S6.1; P2.1). However, sex did not influence the use of M or V (Appendix S6: Table S6.1; P2.1). Overall, the fitted models accounted for a high proportion of the variance in resource use (cR2: 0.35, 0.21 and 0.31 for M, O and V, respectively).
Discussion
The results of this field resource manipulation experiment provide direct evidence for the tight coupling between the spatio-temporal distribution of resources and consequently spatially-restricted movements of a large herbivore. Specifically, we show that roe deer track resource dynamics (Fig. 4; H1), which leads to changes in their space-use (Figs. 2, 3) and underpinning movements (Appendix S5: Figs S5.1, S5.2), and that individual traits, especially resource preference, mediate these behavioural adjustments (H2). In additional, we show that roe deer exhibit a high attraction to familiar locations, a process which leads to site fidelity (H3). As far as we are aware, this is the first experimental demonstration of these interdependencies in a large mammalian herbivore.
Roe deer alter space-use and movement behaviour to track resources
The experimental alterations of food availability led to larger (Fig. 2), spatially-shifted home ranges (Fig. 3), and more explorative movements by roe deer (Appendix S5: Figs S5.1, S5.2), thereby directly establishing the connections between movement, space-use and the spatio-temporal patterns of resources.
In a previous observational study, Fryxell et al. (2008) showed that elk alternated between two movement modes: a low speed and high sinuosity mode thought to be within-patch area-restricted search, and a high speed and low sinuosity mode between resource patches. In our experimental study, we can directly link these movement modes to changing resources: the exploratory movements of roe deer (high velocity and low sinuosity) observed during the closure phase (P1.3) suggested that the animals were motivated to find alternative resource patches when their familiar feeding site (FS) became inaccessible, thereby increasing (P1.1), but mainly shifting, their home range (P1.2). While changes in home range size and location following resource manipulation have been found in studies of lizards (Eifler 1996), birds (van Overveld and Matthysen 2010) and voles (Ims 1987), to date, there have been few experimental investigations of the connections between space-use and the spatio-temporal distribution of resources in large herbivores.
In an earlier study, Kilpatrick and Stober (2002) showed that white-tailed deer shifted their home range core towards the vicinity of newly deployed FS. Our study builds upon these results by demonstrating multiple, successive responses to resource manipulation, linking measured changes in underlying fine-scale movement behaviour of individuals to resulting patterns of space-use that indicate dynamic resource tracking behaviour by roe deer (H1). Although roe deer increased their use of the vegetation matrix during the closure phase, individuals compensated the loss of their main FS (M), to a large degree, by shifts in their movements and space-use towards alternative FS (O; Figs 4; P1.4a-b). Consequently, individuals maintained a high overall use of FS throughout the experiment.
While resource tracking behaviour may be expected under the optimal foraging theory, other individual- and population-based factors, such as social fences due to territoriality or density-dependent resource competition, can constrain the movement responses of individuals to changes in the spatial distribution of resources (e.g., Mysterud et al. 2011). The marked responses shown here are likely to be explained by three factors. First, with the exception of adult males during spring and summer (Liberg, Johansson, Andersen, & Linnell, 1998), roe deer do not generally defend territories, and consequently their spatial distribution can approximate that of an ideal free distribution (Walhström & Kjellander, 1995). Territorial tenure (Linnell & Andersen, 1998) may explain the marginally different response of males (P2.1), specifically their tendency to have larger core areas (Appendix S4: Table S4.3), more explorative movements (Appendix S5: Table S5.1, S5.2) and lower resource compensation than females following closure (Table 4; Appendix S6: Fig. S6.1). In fact, except for one individual (M11) showing high spatial instability during its two-year monitoring, the space-use patterns of males tended to be less influenced by the experimental closure than females.
Second, the rapidity and magnitude of the observed responses are likely linked to the inability of roe deer to buffer transitory shortages of food availability: as income breeders with little fat reserves, roe deer need a constant, high-nutritional intake (Andersen et al., 2000). Third, while intra- and inter-specific competition in herbivores is largely linked to resource depletion (Owen-Smith et al., 2010), this density-dependent constraint of food availability was prevented by providing ad libitum forage at the FS.
Resource tracking is mediated by individual preference
This study moves beyond the findings of Kilpatrick and Stober (2002) by demonstrating that inter-individual variation in preference for FS strongly mediated the responses of roe deer movement patterns, space-use and resource use to changes in the spatio-temporal distribution of resources (H2). During the closure phase, the changes in all measured variables were of larger magnitude for individuals associated with a high FS preference (P2.1). The influence of FS preference was particularly striking in the shifts of space-use (Fig. 3) and in the compensating use of alternative FS following loss of the familiar resource (Fig. 4).
In this study, FS preference varied across individuals (Appendix S3: Table S3.1). These inter-individual differences may be linked to either the environment the individuals were exposed to, or a property of the individuals themselves. In our experimental setting, all roe deer had access to at least one FS provided with ad libitum food where use was not prevented by inter-individual competition (Ossi et al. 2017; Ossi et al. in prep). Hence, we could quantitatively approximate FS preference as the relative use of FS over natural vegetation. Preference for FS was derived from the relative use of FS, and so our definition implies a conditionality on both the physiological state of the individual and on the relative nutritional value of the provisioned food (corn) over natural vegetation in a given season. For example, the body condition at the onset of winter is likely to vary between years. In addition, the dynamics in the quality and quantity of natural browse – either spatial (e.g., between home ranges) and/or temporal (e.g., between years) – could lead to variations in FS preference. Indeed, FS preference was higher in 2017 than in 2018 for all roe deer manipulated in these consecutive years (Appendix S3: Table S3.1). Preference can therefore be considered a dynamic variable (Beyer et al., 2010) that we evaluated at the individual level over a short period of relative stability (pre-closure phase in each winter). We considered the temporal extent of our experiment (ca 6 weeks) short enough to consider FS preference for each animal-year to be relatively constant, because in this time period the physiological conditions and vegetation nutritional value would not vary substantially or consistently.
Individuals attending a FS benefit from exploiting a forage-rich location, but risk elevated intra- and inter-specific contacts (Ossi et al. in prep), high anthropogenic disturbance, and an increased susceptibility to predation by humans. Because FS are associated with such risks, individuals in poorer condition may be more willing to trade on predation risk to access a forage-rich patch (McNamara & Houston, 1987). Alternatively, FS preference could be linked to differences in personality. In roe deer, Bonnot et al. (2015) investigated the association between individual behavioural profile (e.g., body temperature at capture) and the use of risky but profitable habitats (open areas), and suggested that variations in personality could lead to individual differences in habitat use. Similarly, FS preference could be associated to bold or risk-taking personalities. Interestingly, preference for FS tended to correlate with individual body temperature at capture (Pearson’s r = - 0.37 p-value = 0.084; Appendix S8), with bolder individuals (lower temperature) using FS more intensely.
The underlying drivers of preference listed above may not only lead individuals to use resources to a different extent, but also condition their tendency to track spatio-temporal resource dynamics (Spiegel, Leu, Bull, & Sih, 2017). For instance, risk-taking personalities tend to be associated with explorations of novel environments (Cote, Clobert, Brodin, Fogarty, & Sih, 2010), as shown experimentally in great tits following the loss of a familiar foraging area (van Overveld & Matthysen, 2010).
The role of site familiarity
During the post-closure phase of the experiment, roe deer increased their use of familiar FS (M), whose food accessibility had been restored after a transitory restriction (Fig. 4, left-hand panels; P3.1), and home ranges shifted back to pre-closure patterns, as suggested by the high overlap between temporally-disjointed pre- and post-closure space-use (Fig. 3; P3.2). The restoration of these pre-manipulation patterns supports the hypothesis that site familiarity provides inherent benefits to animals maintaining a home range (H3; Piper 2011). These results are coherent with published literature demonstrating that ungulates tend to select for previously visited locations i.e., site familiarity (e.g., Dalziel, Morales, & Fryxell, 2008; Merkle, Fortin, & Morales, 2014).
Unlike most observational studies, our experimental approach allowed us to contrast two alternative, concentrated resources (M and O) of equal nutritional value but of distinct familiarity, and hence to separate the effects of resource tracking from those of familiarity. Resource tracking can explain roe deer use of other FS (O) to compensate the inaccessibility of the familiar, main FS (M) during closure; however, the systematic return of roe deer to M during the post-closure phase, while both alternative resources were accessible, cannot be explained by resource tracking alone. Our experiment suggests that roe deer were actively selecting for familiar areas and that site familiarity has an inherent value.
In observational studies of animal movement, a spurious familiarity effect (sensu Van Moorter et al. 2013) can occur when an important factor influencing animal behaviour is not considered, and the re-visitation of particular locations is interpreted as an evidence for site familiarity selection. However, this confounding effect is unlikely to affect the results of this experiment. First, corn was delivered ad libitum across all FS (M or O) i.e., homogeneous foraging value. Second, the FS were located in comparable environments with close proximity to cover, a factor that largely influence roe deer movements and space-use (e.g., Tufto et al. 1996, Bongi et al. 2008). Third, and most importantly, the specific identities of M and O varied interchangeably between individuals. Hence, we conclude that the return to pre-closure patterns of foraging behaviour and space-use are unlikely to be result of variations in the characteristics of specific FS, but rather of an inherent familiarity effect. In roe deer, site familiarity could allow a more profitable exploitation of natural forage, as seen in bison (Bison bison; Merkle et al., 2014) or reduce intraspecific competition for such resource (see Riotte-lambert et al. 2015 for a theoretical argument). Alternatively, the attraction to familiar areas could be related to a predator (natural or human) avoidance tactic (Gehr et al., in prep; Stamps, 1995).
Our results imply that when resource patterns are changing, individual behavioural decisions probably reflect a trade-off between the advantages of site familiarity and resource tracking. In some cases, finding an alternative resource patch of high nutritional value may cause an individual not to return to familiar sites (e.g., Ims 1987, Eifler 1996, van Overveld and Matthysen 2010), as seen in this study during the closure phase. The relative influence of resource tracking and site familiarity likely depends on species traits and ecology. For example, it is possible that the greater capacity of capital breeding species to buffer temporal dynamics in resource compared to income breeders such as roe deer alters the trade-off between tracking resources and site fidelity.
Ultimately, site familiarity is the manifestation of an animal’s ability to acquire spatial information, in particular by means of spatial memory (Fagan et al., 2013). Large herbivores are capable of memorizing resource location (e.g., Merkle et al., 2014). In this study, it is likely that the variations in roe deer responses to resource changes that are not explained by preference for FS may be the result of individual prior experience and knowledge of the status and distribution of alternative FS. An interesting avenue for further studies will be to evaluate the role of these cognitive processes on individual foraging decisions.
Author’s contributions
NR, FC and PM conceived the ideas and designed the methodology; FC supervised the field study; NR, FO, WH and TS collected the data; NR analysed the data with support from WH and TS; NR led the writing of the manuscript, together with FC and PM. All authors contributed critically to the drafts and gave final approval for publication.
Data availability statement
Data will be made available from the Dryad Digital Repository upon acceptance.
Acknowledgements
N. Ranc was supported by Harvard University Graduate Fellowship and Fondazione Edmund Mach International Doctoral Programme Fellowship. F. Cagnacci was supported by the Sarah and Daniel Hrdy Fellowship 2015-2016 at Harvard University OEB during part of the development of this ms. We are grateful to all the people involved in the roe deer captures and data collection, and especially M.B. Almeida, P. Bonanni, A. Corradini, J. De Groeve, G. Lilli, S. Mumme, S. Nicoloso, D. Righetti, M. Rocca, M. Salvatori, M. Sanchez and P. Semenzato. We thank J.W. Cain, B. Ölveczky, N. Pierce, all the members of the Cagnacci lab and of the Moorcroft lab, and the graduate students of the Harvard Statistics Consulting Service for their valuable suggestions. We also thank the Wildlife and Forestry Service of the Autonomous Province of Trento and the Trentino Hunting Association (ACT) for their support. Finally, we express our gratitude to the hunters of Albiano, Civezzano, Fornace and Trento Nord without whose help this study could not be conducted.
Footnotes
↵a Environmental Studies Department, University of California Santa Cruz, USA