Diet Selection by the Italian Hare (Lepus corsicanus de Winton, 1898) in Two Protected Coastal Areas of Latium

Simple Summary In order to better understand the ecological niche of the Italian hare, we evaluated the diet selection of the species in two protected areas of the Latium coastal environment. The main results emerging from our study were: the wide feeding spectrum of the Italian hare; the high incidence of grasses in dry and in wet season diets; the low number of plant species ingested at relatively high rates; the plastic feeding behaviour of this hare, as diet preferences changed with the variety and abundance of food species. These results highlighted the great adaptability of the species to different niches and the influence of the floristic composition on its feeding habits. In the Italian hare, the assessment of habitat suitability is of strategic importance for its conservation. In particular, feeding preferences of the species may lead to defining some food items as key plant species for identifying its elective habitat and, hence, planning effective re-introduction initiatives. Abstract This study was focused on the diet and feeding behaviour of Lepus corsicanus in two protected coastal areas of Latium, Castelporziano Presidential Estate (CPE) and Circeo National Park (CNP). Plant frequency was assessed by the quadrat method, while diet composition was determined by microhistological analysis of faecal samples. Over the year, the Italian hare fed on 185 of the 229 plant species identified in vegetation, with most of them ingested in low percentages (≤1%). During the dry season (DS), in both areas, Brachypodium sylvaticum, Cynodon dactylon, and Avena fatua were among the most consumed species. In the wet season (WS) the most common plant species in diet were B. sylvaticum, Poa trivialis, and Carex distachya in CPE and Dactylis glomerata, Cynosurus echinatus, and Spartium junceum in CNP. In both sites, considering the annual selection of life forms, grasses and leguminous forbs were preferred, while non-leguminous forbs and shrubs were used less than expected according to their availability. ANOSIM analysis showed significant differences between sites in DS and WS diets. Our study evidenced that the Italian hare behaved as generalist, revealing its capability for exploiting several plant species and to adapt its diet preferences to space-time variation of food availability.


Introduction
The areal extent of the Italian hare (Lepus corsicanus de Winton, 1898) covers central and southern Italy, Sicily, and Corsica but different size, density, and range of the Italian hare populations characterise each of these subareas [1]. In the peninsular area, the distribution of this endemic species has been subjected in the last decades to a substantial contraction accompanied by a significant reduction in the consistence of populations. The taxon Both areas contain several land-cover types representative for the Mediterranean area: natural oak woods with evergreen (Quercus ilex and Quercus suber) and deciduous (Quercus cerris and Quercus frainetto) species, broad-leaved mixed oaks forest, pasture, Mediterranean maquis, pseudo steppe, and mixed or pure forest of domestic pine (Pinus pinea) [15].
To cover different types of vegetation, five different sampling sites in CPE (site 1, 2, 3) and CNP (site 4, 5) have been chosen.
Site 1 (Casa del Pastore)-This site, located on the southwestern side of the Estate, is covered by a pine forest of P. pinea with trees up to 30 m high. The undergrowth is made up of sparse bushes of Asparagus acutifolius, Laurus nobilis, Phillyrea latifolia and Rubus spp. The herbaceous layer is very scarce and mainly formed by C. distachya, Carex flacca and Poa trivialis. A tree pasture with scattered specimens of Q. suber also characterises the site. The prevailing herbaceous species are annual-growing grasses, such as Anthoxanthum odoratum, Briza maxima, Bromus mollis, and C. echinatus. In addition, there are nitrophilous spiny species (Cirsium strictum and Galactites tomentosa) whose presence is due to the grazing of cattle [16]. A fallow area, characterised by annual growing grasses (in prevalence, A. fatua, C. dactylon, Dasypyrum villosum, Lagurus ovatus, and P. trivialis), completes the vegetation mosaic of the site [14].
Site 2 (Coltivi nord)-Situated in the North of the Estate, this site features a mosaic of vegetation characterised by low forest cover of P. pinea. Along the margins and clearings of this forest, in contact with pastures and crops, there are bushes of deciduous species (e.g., Crataegus monogyna, Prunus spinosa, Cornus sanguinea, Clematis vitalba, Tamus communis, and Rubus ulmifolius) mixed with evergreens, such as P. latifolia, Rhamnus alaternus, Myrtus communis, and A. acutifolius.
Site 3 (Santola)-This wooded site, centrally located in the Estate, is mainly characterised by forest vegetation, with a prevalence of Q. suber, due to reforestation carried out after 1970 with native cork oak; lying on acidic sandy substrates, it is characterised by the presence in the underwood of evergreen shrubs (e.g., P. latifolia, Ramnus alaternus, Cistus creticus) and lianose shrubs, such as Smilax aspera and Rubia peregrina.

Sampling and Analysis Procedures
To assess the relative frequencies of plant species, 25 permanent transects were utilised (five from each site). Sampling took place in the dry season (DS, May-August) and in the wet season (WS, November-February). Transects 50 m long were located to cover all the types of vegetation present in the study areas and were spaced by at least 100 m from each other. The quadrat method was used to assess plant frequency [20]: twenty-five samplings were carried out per transect, analysing 1 m 2 of vegetation and skipping the following. Plant species were grouped into four vegetation forms: grasses (G), including in this form also graminoids; leguminous forbs (L); non-leguminous forbs (NLF); shrubs (S). The taxonomic nomenclature of the identified taxa followed Bartolucci et al. [21].
A plant from each observed species in the transepts was collected and processed according to the method described in Maia et al. [22]. In order to create a reference collection, histological fragments of each anatomical part were photographed by light microscopy and catalogued in a database using the image analyser Leica Q500IW (Leica Imaging System Ltd., Cambridge, UK).
Faecal sampling took place monthly in the aforementioned periods along eight transepts (2 × 30 m) randomly distributed throughout each study site and distant at least 100 m from each other in order to reduce the probability to collect pellets from the same animal. All the collected pellets were fresh (bright brown faeces) and, for each collection, a minimum of six pellets, of various sizes and formats, were mixed to form a single composite sample. A total of 40 composites samples were analysed for L. corsicanus (8 months × 5 sites). Our consolidated experience in the microhistological technique made us prefer this method to others, perhaps faster (e.g., DNA metabarcoding) but also not without drawbacks [4,14].
Faecal pellets were processed according to the method described in Freschi et al. [11,14]. For each composite sample, 10 microscope slides were mounted. The slides were examined by light microscopy using the image analyser Leica Q500 IW, obtaining 200 readings for each sample, counting non-overlapping plant fragments in systematic transepts across a slide along alternate rows. Identification of plant species was performed by comparing the different characteristics of the epidermal cells and other structures (e.g., stomates and trichomes) with those of the plant reference collection built by collecting monthly the plants found in the study site. Microphotographs from all taxon/structures were made with the same magnification to facilitate a fast comparison between the reference collection and the faecal material.
This reference material is available at the Laboratory of Environmental and Applied Botany, University of Basilicata. Not identified fragments (6.7%) were classified as 'unidentified' and excluded from the analysis.

Statistical Analysis
Relative frequencies (rf ) of plant species, families, and life forms were calculated by dividing the total number of fragments attributed to a given taxon by the total number of identified fragments. Data of the plant species identified in the study site were used to calculate the relative frequencies of each taxon, family, and vegetation form. Similarly, we calculated the relative frequencies of the plant species identified in the faeces by dividing the total number of fragments attributed to a given taxon by the total number of identified fragments [9][10][11]23,24].
Data of identified plant species composing the diet were also used to compute the following alpha diversity indices:
For each of the above indices, differences, between DS and WS were tested by Student's t-test (p < 0.05).
To compare dietary similarity between DS and WS the Sørensen similarity index (C S ) [29] was computed. C S index varies between 0 (no similarity) and 1 (complete similarity).
Diet composition was analysed by multivariate analysis. Similarity matrices were constructed by using averages of the Bray-Curtis similarity coefficient [30]. Analysis of similarities (ANOSIM) was performed to test diet differences among sites using 999 permutations [31].
Diet selection was estimated for life forms and for shared plant families in vegetation and diet by Resource selection ratio (w i ) [32]: where o i is the proportion of the botanical family (or life form) in the diet and p i is its available proportion (w i > 1, preference; w i = 1, indifference; w i < 1, avoidance). Differences were tested by χ 2 test [33]. Data were analysed by R software (R Core Team, Wien, Austria) [34].
For each of the above indices, differences, between DS and WS were tested by a Student's t-test (p < 0.05).
To compare dietary similarity between DS and WS the Sørensen similarity index (CS) [29] was computed. CS index varies between 0 (no similarity) and 1 (complete similarity).
Diet composition was analysed by multivariate analysis. Similarity matrices were constructed by using averages of the Bray-Curtis similarity coefficient [30]. Analysis of similarities (ANOSIM) was performed to test diet differences among sites using 999 permutations [31].
Diet selection was estimated for life forms and for shared plant families in vegetation and diet by Resource selection ratio (wi) [32]: Where oi is the proportion of the botanical family (or life form) in the diet and pi is its available proportion (wi > 1, preference; wi = 1, indifference; wi < 1, avoidance). Differences were tested by χ 2 test [33]. Data were analysed by R software (R Core Team, Wien, Austria)[34].

Diet Composition in the Dry Season
In both sites, grasses were the most utilised life form in CPE (73.95%) and CNP (53.71%), followed by non-leguminous forbs (14.23% in CPE and 25.83% in CNP), shrubs (9.23% in CPE and 12.62 % in CNP), and leguminous forbs (3.71% in CPE and 7.74% in CNP) ( Figure 2). In the diet of L. corsicanus from the Latium coast, 133 taxa belonging to 36 families were found (Tables A1 and A2). Poaceae was the most representative family in the diet (63.7 % in CPE and 43.68% in CNP), followed by Asteraceae (7.13%), and Cyperaceae (5.61%) in CPE, Fabaceae (8.83%) and Asteraceae (6.82%) in CNP ( Figure 2). The number of determined species was higher in CPE (103) than in CNP (96). In both sites, most of the taxa (71 in CPE and 68 in CNP) were ingested in low percentages (≤1%). Conversely, B. sylvaticum, C. dactylon, and A. fatua were among the most consumed species, together representing 17.01% and 16,26% of the diet in CPE and CNP, respectively (Tables  A1 and A2). Figure 3 shows, similarly to the dry period, that grasses was the most representative life form in the diet (74.77% in CPE and 59.69% in CNP), followed by non-leguminous forbs (

Dietary Diversity and Similarity
Differences in DS vs. WS diet richness were observed only in CNP (D, 7.624 vs. 5.570, p = 0.029; E, 0.598 vs. 0.674, p = 0.021) ( Table 1). In both sites, Cs similarity index showed a medium overlap among seasonal diets (0.677 in CPE and 0.569 in CNP). ANOSIM analysis revealed that there were significant differences between sites in both DS and WS diets. Moreover, seasonal diets were significantly different in CNP (R = 0.515; p ≤ 0.001) ( Figure A1).

Dietary Selectivity
Among the most abundant species in diets, those characterised by particularly high selectivity indices (W i > 2) are highlighted: P. trivialis, C. distachya, Brachypodium retusum, and Allium triquetrum (WS) and P. trivialis, C. dactylon, and B. sylvaticum (WS) in CPE; Spartium junceum, D. glomerata, C. echinatus, and C. dactylon (WS) in CNP ( Figure 4). During DS, in CPE, only the Poaceae family has been used more than expected according to its availability (Table 2). Conversely, Apiaceae, Asparagaceae, Asteraceae, Fagaceae, Geraniaceae, Malvaceae, Rhamnaceae, Rosaceae, and Rubiaceae were negatively selected. Instead, Amaryllidaceae, Asteraceae, Cyperaceae, and Poaceae, were positively selected in WS, and Apiaceae, Asteraceae, Brassicaceae, Caryophyllaeae, Fagaceae, Geraniaceae, Oleaceae, Rhamnaceae, Rosaceae, and Rubiaceae were avoided. In CNP a positive selection was observed only in the wet period in Fabaceae and Poaceae (Table 3). Avoided families in both periods were Apiaceae, Asteraceae, Lamiaceae, and Oleaceae, while Cistaceae, Cyperaceae, and Rubiaceae were avoided only in WS. Considering the annual selection of life forms, in both sites ( Figure A2) grasses and leguminous forbs were preferred; conversely, non-leguminous forbs and shrubs were avoided.

Discussion
The main results emerging from our study were: (a) the wide feeding spectrum of the species, since it fed annually on 185 of the 229 plant species identified in vegetation; (b) the prevalence of grasses in CPE and in CNP, in DS and in WS diets, with the predominance of Poaceae, followed Cyperaceae, Amaryllidaceae, Asparagaceae, and Juncaeae, as other families of this life form; (c) the low number of plant species ingested at relatively high rates; (d) the plastic feeding behaviour of the Italian hare, as diet selectivity changed with the variety and abundance of food species. In the study sites, the most observed taxa were C. dactylon, A. fatua and B. sylvaticum. In particular, this last species is confirmed as an important constituent of the diet. High incidence in the diet of Brachypodium spp. was observed in studies conducted in the Basilicata region and in Corsica [4,[9][10][11].
The preference for Brachypodium spp., also observed in ruminants [35], is probably linked to its wide distribution in various vegetation covers all year round. Considering Poaceae as a whole, their high contribution to the Italian hare's diet could be motivated by their good palatability and high cellulose content, which can provide a useful reserve of energy [36].
Poaceae, Asteraceae and Fabaceae families constituted the bulk of the diet throughout the dry season. Similar preferences in diet were observed in the Italian hare in south Italy [12] and in Haute-Corse [4]. Castellaro et al. [36] underline the great importance of this group of plant species in the nutrition of herbivores with cecal fermentation, given the characteristics of their digestive system and the way in which nutrients are used. The increased palatability of forbs in the dry period could be attributed to their higher water and lower fibre contents in tissues in comparison with grasses [37,38]. Palatability was defined by Greenhalg and Reid [39] as the dietary characteristics that stimulate a selective response by the animal. Vallentine [40] cites, among the morphological and chemical factors that positively influence the palatability of a plant: the presence of succulent leaves, the absence of thorns, poor flowering, the accessibility to edible parts, the presence of young vegetative parts, the high content of protein and sugars, the low content of tannic substances that confer bitter taste, and the absence of alkaloids and glucosides with toxic action. On the other hand, the species which were normally avoided could be grazed on under compulsion due to the scarcity of food in the area. Concerning this observation, Asparagaceae, Amarillidaceae, and Cyperaceae which were avoided in CPE during DS were instead positively selected during WS. Moreover, CPE hares excluded H. helix and Smilax aspera from their diet, conversely CNP hares fed on these species even showing selectivity for S. aspera in WS. The lower availability of food herbaceous species determined by the dense canopy of CNP sampling sites could explain this feeding behaviour. Rubiaceae was used in small quantities and not selected. Conversely, in Corsica, this family was used more than expected according to its availability [4]. Overall, with the exception of Poaceae which was always preferred, we observe that feeding preferences of the Italian hare vary across different niches. Plants from this family represent the bulk of the diet also in L. europaeus [41][42][43][44][45][46][47][48], Lepus timidus hibernicus Bell, 1837 [49][50][51][52], Lepus arcticus Ross, 1819 [53], Lepus californicus Gray, 1837 [54][55][56][57], Lepus flavigularis Wagner, 1844 [58], L. granatensis [37], Lepus starcki Petter, 1963 [59]. In the present study grasses and non-leguminous forbs constituted a large portion of the diet of the Italian hare, while shrubs and leguminous forbs appeared to not be consumed in large quantities. Nevertheless, an underestimation of the incidence of these life forms in the diet could be related to their high digestibility. In Mediterranean environments, this underestimation could be lower in DS, when herbivores show a reduced digestibility of the dry matter of the selected plants [24].
Feeding preferences are very difficult to interpret and to understand as the factors involved vary spatially and in time, as well as to the availability and to relative abundance to associated species.
In herbivores, several food strategies influence the rank-order selection of plants and their ingestion level at any given site in order to maximise energy intake, reduce energy expenditure or predation risks, or attenuate the toxic effects of plant secondary metabolites [60]. According to Shipley et al. [61], mammalian herbivores are considered generalists or specialists if the incidence of a family plant on diet is over or under 60%, respectively. These authors consider as facultative generalists to be the species in which the broad fundamental niche allows them to consume a wide variety of foods and that, occasionally, demonstrate a narrow realised niche, focused on less difficult plants than is the case with specialists. According to this definition, we can consider the Italian hare as a facultative generalist in its feeding strategy. Studies on feeding preferences of Brown hare [45,48] and Snowshoe hare [62] classified these species as predominantly generalist. Nevertheless, in these species, as in L. corsicanus, grasses are the main diet item even if with a declined importance in the dry season, when fibre content increases from early to late summer. In this period, in particular, the species could select some plant species that even if ingested in low quantities, would fulfil a nutritional role of production and a functional role as diet improvers [36].

Conclusions
Our study demonstrated that the diet of the Italian hare was characterised by a wide diversity of plant species in the dry and wet seasons. Nonetheless, the bulk of the diet consisted of a few species, among which the most abundant were C. distachya, B. Sylvaticum, and C. dactylon. Probably, the high selectivity toward these plants was also favoured by their high availability throughout the year. The significant differences in the composition of the diet-highlighted in the diversity indices-confirmed the great adaptability of the Italian hare to different niches and the influence of the vegetation on the feeding habits of the species. On the other hand, the wide spectrum of diet, besides reflecting the adaptation of the species to its habitat may be more beneficial to maintain the richness of species more so in environments characterised by high plant richness, such as our study sites. The Italian hare revealed its ecological plasticity highlighted by its capability for exploiting food resources, exhibiting an opportunistic behaviour in response to changes in their spatial and temporal availability.