The European Hare (Lepus europaeus): A Picky Herbivore Searching for Plant Parts Rich in Fat

European hares of both sexes rely on fat reserves, particularly during the reproduc-tive season. Therefore, hares should select dietary plants rich in fat and energy. However, hares also require essential polyunsaturated fatty acids (PUFA) such as linoleic acid (LA) and alpha-linolenic acid (ALA) to reproduce and survive. Although hares are able to absorb PUFA selectively in their gastrointestinal tract, it is unknown whether this mechanism is sufficient to guarantee PUFA supply. Thus, diet selection may involve a trade-off between a preference for energy versus a preference for crucial nutrients, namely PUFA. We compared plant and nutrient availability and use by hares in an arable landscape in Austria over three years. We found that European hares selected their diet for high energy content (crude fat and crude protein), and avoided crude fibre. There was no evidence of a preference for plants rich in LA and ALA. We conclude that fat is the limiting resource for this herbivorous mammal, whereas levels of LA and ALA in forage are sufficiently high to meet daily requirements, especially since their uptake is enhanced by physiological mechanisms. Animals selected several plant taxa all year round, and preferences did not simply correlate with crude fat content. Hence, European hares might not only select for plant taxa rich in fat, but also for high-fat parts of preferred plant taxa. As hares preferred weeds/grasses and various crop types while avoiding cereals, we suggest that promoting heterogeneous habitats with high crop diversity and set-asides may help stop the decline of European hares throughout Europe.

Introduction nutrients and FA) differs between different plant taxa, plant parts, and developmental stages of plants in the course of the year [23][24][25][26] and because the nutritional needs of female hares may change throughout the long breeding season.
We asked whether hares select food in order to meet their needs for dietary fat or, specifically, their needs for PUFA. Our reasoning was as follows: (1) if selective uptake of PUFA in the gastrointestinal tract can supply them with sufficient essential FA, we hypothesize that European hares will choose a diet with high energy content, preferring plants and plant parts rich in fat. This may even lead to an apparent avoidance of plants rich in certain PUFA, unless they also have high energy content. (2) Alternatively, if digestive mechanisms that enhance PUFA uptake are insufficient, we hypothesize that hares should prefer plant taxa rich in PUFA, namely LA and ALA. (3) Under both the above scenarios, we hypothesize that dietary preferences vary according to season, individual age and sex. We tested the hypotheses by analysing both the botanical and chemical composition of hare stomach contents, and comparing the values with those of the food plants available in four seasons for three years in an arable area in Austria. We aim to use knowledge of dietary preferences to make recommendations for the conservation of the hare through targeted habitat management.

Data collection
During the three years of the study, 570 European hares were shot during the day in February ("winter", n = 72), May ("spring", n = 127), August ("summer", n = 97), and November ("autumn", n = 274). All animals were sexed according to secondary sexual characteristics. Age was determined by the weight of the dried eye lenses [28]. Hares younger than one year of age (eye lens weight 276 mg) were classified as subadults. Stomach contents were analysed chemically and botanically (for methods see [7] (DM composition), [5] (FA composition), and [22] (botany)). We calculated gross energy (kJ/g) using energetic values given by Valencak and Ruf [7]. Stomach contents represented the food items taken up by the European hares during the previous nocturnal activity period [3]. Sample sizes vary because not all stomach contents could be analysed for both botanical and composition of nutrients.
During each season a semi-quantitative botanical inventory of the complete study area was conducted (for methods see [22]). Furthermore, a plant sample for each crop type and for fallow land was collected once per season, for which the DM and FA composition were determined (for analyses of DM composition see [7], for FA composition [5]). We assumed that the circular plot of 10 ha around the location where each hare was shot was that hare's potential feeding area during the last 12 hours, as GPS collared hares in this study site had 24-hour home ranges of up to 10 ha [29]. Only European hares with at least 75% of their circular home ranges falling within the study area were considered representative and used for further analyses (i.e. hares shot close to the boundaries of the study area were excluded, n = 171). Thus, the botanical inventory within each circular plot represented the available forage for a hare. Estimates of the average nutrient and energy content of available forage in each plot were calculated on the basis of this inventory.

Data analysis
Dietary preferences were measured by using Chesson's Electivity Index ε [30], an index based on Manly's alpha [31], which can be used to analyse dietary preferences [32].  Table 1). The reliability of each electivity index was tested by bootstrapping [33]. The original ε i values (ε i = Chesson's Electivity Index for the food type i) were resampled 1000 times with replacement and an accelerated bootstrap confidence interval was calculated. The accelerated bootstrap adjusted the confidence interval for bias and skewness [34]. If the lower and upper 95% CI featured the same algebraic sign, the selection for this food type was significant. We only bootstrapped ε i values for forage components if they were selected by 7 or more hares, as smaller sample sizes provide unreliable results. For this reason, it was not possible to bootstrap the electivity indices for seasonal differences between age classes or sexes.
The specific FA proportions are typically autocorrelated and, hence, not independent. This does not, however, prohibit statistically analysing preferences for each FA separately, it merely follows that resulting electivity indices can be expected to be similar for FA whose proportions are positively correlated.

Statistical analyses
All analyses were computed with the software R 3.0.2 [35]. We analysed the gross energy data by linear mixed-effects modelling using the package lme4 [36]. P-values and parameter estimates (β) were extracted by Markov chain Monte Carlo sampling based on 10000 simulation runs [37] using restricted maximum likelihood. We visually checked the normality of the model residuals by means of a normal probability plot. Homogeneity of variances and goodness-of-fit were examined by plotting residuals versus fitted values [38]. The model included hare identity as a random factor in order to allow paired testing for the available gross energy (in the plants in the circular plot) and used gross energy (in the diet), and a specific code for the month and year as a second random factor in order to account for the different seasons and years of the study. We then tested whether there was a difference between available and used gross energy.
DM and FA components (response variables) were also analysed using linear mixed-effects models with the package lme4 [36]. The full model for the DM response variables included the covariates season (4 levels), sex, age (adult vs. subadult) and their two-way interaction terms. Moreover, the models included year as a random factor in order to account for the different years of the study. The full models were used to create a set of models with all combinations of the independent variables using the package MuMln [39]. P-values and estimates (β) were extracted by model averaging (including all models with delta AIC<10). The residuals of the full models were checked for normal distribution by viewing QQ-plots and histograms. Posthoc tests with the best model were computed for the covariate season using the Tukey's all-pair comparisons method in the package multcomp [40].

Ethics Statement
Hares were shot by local hunters using rifles in winter, spring and summer. In autumn, hares were collected on regular hunts using shotguns. Outside the legal hunting time (autumn), local hunters were given permission to shoot hares by the local administration of Gänserndorf, Lower Austria. The method of collecting hares was not specified, but shooting was conducted according to hunting law (using rifles on single hunts and shotguns on drive hunts). Collecting hares outside the legal hunting period was permitted on the basis of the Lower Austrian Hunting Law § § 75 and 76, which allow exceptions to regular hunting periods for scientific research. Notification by the local administration of Gänserndorf was given on the 20th of January 2003. The land accessed was not protected and no protected species were sampled. The study complies with the current laws of Austria.

Results
Chemical preferences for DM and gross energy were assessed in 263 hares (122 males, 141 females), whereas the FA preferences of 269 hares (124 males, 145 females) were evaluated. The stomach contents of 382 hares (141 males, 241 females) were analysed botanically.

Selection for energy and DM
We found a significant difference between the gross energy of available (mean 12.450 kJ/g; SE ± 0.057) and used (mean 14.356 kJ/g; SE ± 0.072) forage plants (p MCMC = 0.0001; β = 0.126), showing that our study animals selected their food for high energy content. Hares preferred plants rich in crude fat and crude protein; all other DM components were avoided throughout the year (see S1 Table). The animals mostly avoided crude fibre in their diet, whereas crude fat was preferred the most. Our results show a distinct difference between the availability of crude fat in the environment and its use in the diet of European hares (Fig 1b). The variance between available and used crude protein and crude fibre was less pronounced, and it was almost nonexistent for carbohydrates (Fig 1a, 1c and 1d). Furthermore, we found significant differences in the selection of all DM components between the seasons ( Table 2). Although seasonal differences were recorded, the general preference for crude fat and the general avoidance of carbohydrates were maintained throughout the year (see S1 Table). However, crude protein was significantly preferred in autumn and winter, while it was not significantly selected in spring and summer. We found no significant effects of sex (p > 0.05) or age class (p > 0.05) on any DM component (see S2 Table).

Selection for certain FA
Hares avoided the FA 14:0, 16:0, 16:1, LA and ALA, whereas 18:0 and 18:1 were preferred throughout the year (Fig 2). Selection for every FA was influenced by the season (Table 3), but general avoidance or preference was maintained in the course of the year (see S1 Table). Furthermore, sex and age had a significant influence on ALA selection (sex: p = 0.016; β male = -0.138, age: p = 0.077; β subadult = 0.102, Table 4), i.e. females and subadults had higher ALA intake. Intake of none of the other FA was significantly influenced by sex (p > 0.10) or age (p > 0.10, see Table 5 for LA).

Selection for plant taxa
Hares selected several plant taxa, but the pattern of selectivity changed over the course of the year (Fig 3). Favoured plant taxa in winter were sugar beet roots (Beta vulgaris, ε = 0.48,   When grouping plant taxa to cereals, intertillage, other field crops, and weeds/grasses we determined a preference for weeds/grasses and other field crops (Fig 4). In the course of the year, other field crops were preferred in winter (ε = 0.87, n = 32), weeds/grasses in spring (ε = 0.64, n = 72) and summer (ε = 0.34, n = 68), and both plant groups were positively selected in autumn (other field crops: ε = 0.62, n = 58; weeds/grasses: ε = 0.26, n = 133).

Discussion
In previous studies, we showed that the reproductive performance of captive female European hares [2] and leveret growth and survival [1] are enhanced when the females are fed a diet rich in energy. The results of the present study are consistent with previous studies, and support our hypothesis (1): that European hares in the wild choose their diet for high energy intake. Hares preferred a diet rich in crude fat, irrespective of its content in specific FA. Hares did not show random dietary uptake. Rather, they are highly selective feeders, so we would not expect random sampling of plant nutrient composition in an area to reflect adequately the diet of foraging wild hares. This selectivity may explain why previous researchers [17][18] were unable to show an effect of mean food quality in hare habitats on life history traits and density in European hares. As well as female and subadult hares, males also preferred a high energy diet. Although the energy costs of lactation in females and growth of young are comparably high, we demonstrate that all hares, irrespective of age and sex, prefer plants rich in crude fat. Selecting an energy-rich diet may reduce gut content weight, thus enabling hares to run faster and escape from predators while meeting their daily energy demands. Moreover, in our continental study site in Lower Austria, where precipitation is less than 500 mm per year (see also [41]), a shortage of water in the diet might be compensated for by oxygenizing fat and producing metabolic water [42]. Our hypothesis (3), that dietary preferences vary according to season, individual age and sex, can be rejected with respect to age and sex effects. However, we did find seasonal effects: the preference for a diet rich in crude fat was highest in summer, when the hares' mating season in Central Europe is gradually coming to an end [43]. Furthermore, our data indicate that hares build up their fat reserves in winter: when fat stores were fully built up in winter, our European hares showed the least selection for crude fat in their diet (ε crude fat: winter = 0.13, spring = 0.28, summer = 0.36, autumn = 0.26, see S1 Table). This is in accordance with Zörner, Flux and Popescu et al. [3][4][5].
We did not find support for our hypothesis (2), that European hares preferred plants particularly rich in LA and ALA. Both FA were not preferred across seasons and study years. This may not indicate an active avoidance, but may result instead from preferences for other plant properties.
The high PUFA content in hares' body tissues [15] seems to be solely obtained by their targeted utilisation of the available PUFA in the gastrointestinal tract [5], and not, as suggested in the current literature, by selective uptake through the diet (see [44] for a review). Consequently, we suggest that the European hare's diet selection is adapted to maximise energy intake by taking a diet rich in fat. Simultaneously, requirements of specific FA are covered by physiological mechanisms in the gastrointestinal tract that allow the selective absorption of PUFA. Only The European Hare: A Picky Herbivore Searching for Fat during lactation, when the supply of certain PUFA such as LA is crucial [9], European hares positively select plants rich in LA as indicated by an increased content of LA in the gastrointestinal tract contents of lactating females [5].
Our results support the view that European hares prefer certain plant taxa. Of the 349 different plant taxa identified in the study area, 47 were found in the hares' stomachs, and only 10 of those were positively selected for by the European hares throughout the year or in the whole study period (Beta vulgaris, Glycine max, Hordeum vulgare, Medicago sativa, Polygonum The European Hare: A Picky Herbivore Searching for Fat aviculare, Stellaria media, Trifolium pratense, Zea mays, unidentified weed, and unidentified grass; see S1 Table). The preferred plant taxa in all our study years (2003)(2004)(2005) were only partly in line with previous findings for 2003 only at the same study site [22]. Beta vulgaris roots in winter, Glycine max in spring and Beta vulgaris roots in autumn were preferred consistently, regardless of whether the analysis was conducted for the year 2003 only or for the whole study period. However, whereas Triticum aestivum was positively selected in autumn in 2003 [22], the entire data set (2003)(2004)(2005) revealed that the hares avoided this field crop during all the four seasons. We believe that these inconsistencies are either due to the comparably small sample size for 2003 which did not allow rigorous statistical analysis [22] (2003: n = 117; 2003-2005: n = 382), or due to the different circular plot sizes used to represent the available food items for each hare. We used small potential feeding plots of 10 ha for the analysis of the 2003-2005 data, but Reichlin et al. calculated electivity indices on the basis of much larger potential feeding sites (50 ha) for the 2003 data [22]. When plant taxa were pooled into groups for analysis, we found that hares preferred weeds/grasses and other field crops over the cereals which characterize arable land in our study site (28% cover).
We expected the plant taxa selected by hares to be rich in crude fat. Glycine max was rich in crude fat, but Beta vulgaris and Trifolium pratense were poor in crude fat (see S3 Table). Similarly inconsistent results were recorded for the plant groups: the category other field crops was high in crude fat but weeds/grasses was low in crude fat. If hares select for crude fat, but the plant taxa they select are not per se rich in crude fat, then this suggests that hares are selecting high-fat plant parts which greatly exceed the average crude fat value of the plant. This is in concordance with the finding that hares select specific plants, although the average energy content of the plants in a habitat might be low (see above).
The selective feeding of European hares on energy sources which include non-crop plants supports the call to increase habitat heterogeneity, crop diversity and unused fields such as setasides and fallow land [45,2,[46][47]. Our findings suggest that, in order to increase the supply of preferred forage plants for the European hare in arable land, landscape structures such as field margins, fallow land, and buffer strips containing significant fractions of weeds and grasses should be created in preference to hedges, trees and afforested areas. To enhance arable biodiversity and provide a variety of the European hare's preferred food plants, different types of non-farmed features should be promoted. Suitable target areas for habitat improvement are: (1) areas of reduced arable productivity because of shadow effects, such as strips adjacent to woods or shelter belts, (2) areas that are already out of production, such as strips adjacent to water, and (3) strips between conventionally and organically farmed fields. Optimal set-asides for European hares should include weeds such as Medicago sativa, Polygonum aviculare, Stellaria media, Glycine max and Trifolium pratense as well as grasses. Moreover, the vegetation of the non-farmed features should be kept low and sparse to enhance its suitability for foraging hares [48]. Our results show that European hares are highly selective in their food choice during winter. Improved food availability for European hares in this season might be achieved by encouraging farmers to sow set-aside and field margin strips in early autumn, which will provide fresh fodder in late autumn and early winter. The proportion of land planted with wildlife-friendly seed mixtures or set-asides needs to be high (at least 14%), in order to be beneficial [49], and in order to avoid concentration of hares in a few areas, as this leads to intra-specific stress [50] and higher disease transmission (e.g. [51]). In addition, predators may search these areas systematically, especially if they are arranged as linear structures, which may reverse the positive effects of habitat improvements [52]. In the light of recent declines in populations of European hares throughout Europe [45], these recommendations for agri-environment schemes are important steps towards the evidence-based conservation of the species in intensively farmed arable land.
Supporting Information S1