Mountain hares Lepus timidus follow the green-up wave in the pursuit of high-quality food

Spatial patterns in animal behavior can provide insight into habitat quality and the distribution of resources. Understanding how, when, and why animals use certain areas is critical to their conservation and management. We investigated the distribution of pellets of mountain hares Lepus timidus in the Swiss Alps and compared differences between spring and autumn. 1515 pellet locations from 119 individuals (70 males, 49 females) were used. Pellets were collected from 2014 to 2019; individuals were determined using an established, non-invasive genetic technique. We found evidence of an altitudinal shift in the occurrence of pellets from lower elevations in spring to higher elevations in autumn. This seasonal pattern is also supported by altitudinal shifts of three individual core activity areas (2 males, 1 females) and by higher hare activities at the high plateau in autumn. We conclude that the annual ‘wave’ of greening from lower to higher elevations, which correlates with forage availability and quality, explains this shift.

Spatial patterns in habitat use by animals can provide insight into habitat productivity and resource distribution (Phillips et al. 1998, Bennetts andKitchens 2000). For example, in landscapes where high-quality habitats are available and resources are predictable, individuals can develop a high site fidelity and range over relatively small areas (Switzer 1993). Understanding how animals use different habitats due to seasonal changes in food quantity and quality helps to explain how these resources influence individual life-history trajectories (Larter and Nagy 2001).
Diet composition in terrestrial mammals like leporids differs greatly from season to season because they eat many types of grasses, shrubs and forbs (Chapman and Flux 1990). This flexibility allows arctic-alpine species like mountain hares Lepus timidus to cope with changing environmental conditions and the corresponding changes in food quantity and quality (Thulin and Flux 2003). For example, a study by Rehnus (2013) showed that fewer graminoids and forbs are consumed during the snow season as compared to the rest of the year. At the same time, hares consume more gymno-sperms during the winter months than in snow-free times. Many studies on the dietary requirements of leporids during the snow-free season have observed that animals tend to select inflorescences and forbs which are known to have a higher nutritive value (Chapuis 1990, Johannessen and Samset 1994, Paupério and Alves 2008, Seccombe-Hett and Turkington 2008. However, during flowering and seed production, plants produce fewer leaves and more stems. This increases the fiber content and reduces the protein content, thereby reducing the forage quality (Albon and Langvatn 1992, Duru 1997, Bumb et al. 2016). To take advantage of plants' high nutritive state, herbivore species should follow the green-up wave along an altitudinal gradient. For example, red deer Cervus elaphus are found at high altitudes in the Alps during summer due to food quality (Luccarini et al. 2006, Zweifel-Schielly et al. 2009). However, it is still under discussion as to whether elusive and difficult-to-spot leporids from arctic-alpine environments also undertake seasonal shifts in mountainous habitats. According to Bisi et al. (2011) mountain hares do not completely shift their home ranges in relation to resource distribution from one season to the next in the Alps (but see Slotta-Bachmayr 1998), whereas the same species used distinct seasonal areas depending on resource distribution in the boreal forest (Dahl 2005).
Mountain hare populations of the Alps are highly suitable for analysing altitudinal shifts because this genuine species lives at high elevations in isolated glacial relict populations (Thulin andFlux 2003, Rehnus 2013). Climate habitat model, however, show a decreasing population trend in the Alps (Rehnus et al. 2018).
In this study, we investigated the occurrence of altitudinal shifts in pellet locations from spring to autumn based on non-invasive genetic population monitoring data (Rehnus and Bollmann 2016) from six years (2014-2019). We also used multiple locations of the same individuals in both seasons from the same year to investigate altitudinal shifts in core activity areas of mountain hares as further support of seasonal distribution patterns.

Study area
The study area comprises 3.5 km 2 and is situated along the Ofenpass in the Swiss National Park in southeastern Switzerland (46°39′N, 10°11′E; Fig. 1). The study area was selected to represent the ecological range occupied by the mountain hare in the Swiss National Park and is accessible under different snow conditions with a minimal risk of avalanches. The Swiss National Park is designated by the International Union for the Conservation of Nature (IUCN 2016) as a Category Ia nature reserve (strict nature reserve/wilderness area). It is closed to the public in spring; and open in summer and autumn when hiking is limited to pre-defined walking routes. There is no grazing by domestic animals in the National Park.
The study's elevational range was from 1693 to 2587 m a.s.l. Its topography is characterised by mountain slopes with a typical vegetation zoning consisting of subalpine coniferous forests, alpine meadows and rock and scree in the uppermost part of the mountain, which is shaped as a high plateau. Habitats within the study area were classified according to the habitat categories of HABITALP, a standardized habitat classification system for protected areas in the Alps (Lotz 2006). The study area encompasses seven main habitat types: meadows (29%; with diverse grasses, including Nardus stricta, Festuca sp., Poa sp., Agrostis sp., Luzula sp. and sedges), timber stands (24%), scree slopes (16%), storeyed stands (12%; mixed Larix decidua, Pinus cembra, P. sylvestris, P. mugo spp., Picea abies), sapling stands (6%; dominated by P. mugo spp.), pole timber (5%) and mature stands (5%). Residual habitats cover 3% of the area. The climate in the Swiss National Park is continental, with mean January and July temperatures of −9°C and 11°C, respectively (Haller et al. 2013). The mean monthly precipitation measured at 1970 m a.s.l. is 34 mm in January and 108 mm in July (Haller et al. 2013).

Species data
We collected fresh pellets over six consecutive years (2014-2019) in spring (end of March until first half of April) and autumn (October). Samples were collected both systematically and opportunistically, as described in detail by Rehnus and Bollmann (2016). Systematic sampling was conducted on 91 plots that were pre-selected on a 200-m square grid; all hare pellets within each trial plot were collected during each visit. For the opportunistic sampling, we collected pellets as we moved from one systematic plot to the next. Only fresh fecal pellets were collected because amplification success rates were significantly lower for pellets older than five days (Rehnus and Bollmann 2016). Samples were collected and stored in separate plastic Figure 1. Location of the study area (3.5 km 2 ) with sampling locations (circles) in the Swiss National Park (grey region) in Switzerland. Background colours illustrate forests (dark grey), meadows (light grey) and scree and rocks (black stipples). To demonstrate the sampling strategy, we present sampling results from spring 2014. White circles indicate the absence of mountain hare pellets during systematic sampling; black (grey) circles indicate the presence of faeces during systematic (opportunistic) sampling (Rehnus and Bollmann 2016, adapted). The uppermost part of the mountain (≥ 2500 m a.s.l.) is shaped as a high plateau (dashed line among sampling locations) and is composed by 39% of alpine meadows and 61% of rock and scree. tubes without skin contact to minimize DNA contamination (Sloan et al. 2000). After collection in the field, samples were frozen and stored until they were analysed in the lab.
We used nine nuclear microsatellites (Lsa1, Lsa3 (Kryger 2002), Sat 2, Sat5, Sat8, Sat12 (Mougel et al. 1997), Sol30, Sol8 (Rico et al. 1994), Sol33 (Surridge et al. 1997)) and a sex marker (SRY; Wallner et al. 2001) to identify individuals and assign them to the collected feces samples. DNA samples were genotyped in three independent replicates and consensus homozygote genotypes were accepted if all three replicates were consistent. Consensus heterozygote genotypes were accepted if at least two replicates were consistent and no more than two alleles were found across all three replicates (Schenkerz et al. 2020).

Statistical analysis
We evaluated altitudinal shifts from spring to autumn as indicated by individual pellet locations and individual core activity areas. Pellet locations are good indicators of feeding sites and core activity areas of mountain hare (Angerbjörn 1983, Krebs et al. 2001, Hiltunen 2003. Our dataset contains 1001 pellet locations during the early breeding season and 514 pellet locations during the post-reproductive period from 70 males and 49 females. We tested altitudinal shifts for systematic and opportunistic sampling using linear models with altitude as the response variable, season as the predictor variables and individual ID and year as a random factors. We also analysed the number of used systematic plots at the high plateau and samples found opportunistically in alpine meadows at the high plateau (≥ 2500 m a.s.l.) in our study area and compared them between spring and autumn with a Mann-Whitney U-test.
To support the hypothesis of a potential seasonal shifts, we calculated the area of minimum convex polygons 95% (MCP95) for seasonal locations of unique individuals in the R-package 'reproducible core activity area analysis' (Signer and Balkenhol 2015). Core activity areas are part of the individual home ranges that depend on seasonal ecological conditions (Hiltunen et al. 2004, Bisi et al. 2011). We used a dataset of six activity core areas from three unique individuals (2 males, 1 females) which fulfilled the criteria of a minimum of 10 observations at unique locations per season within the same year. We assume that this minimum adequately represents the core activity areas of mountain hares in our study area because of the species' site fidelity (Bisi et al. 2011). Altitudinal shifts of core activity areas were calculated as the difference of the centroids of seasonal MCP95-polygons. Further, we estimated the number of unique individuals at the high plateau in our study area and tested for seasonal differences.
All statistical tests were conducted using R ver. 3.6.3 (<www.r-project.org>) and full-model table is supplemented (Supplementary material Appendix 1 Table A1). Analyses were not separated by sex because between male and female mountain hares is a weak habitat segregation in both seasons (Rehnus and Bollmann 2020).
On average, mountain hares had core activity areas that were 2-to 4-times larger in spring than in autumn (Table 1). The centers of these area are found at higher altitudes  in autumn compared to spring while seasonal overlaps were below 1.8% between areas (Table 1). The number of unique individuals at the high plateau was slightly higher in autumn (8.0 ± 1.2) than in spring (4.5 ± 1.0; F 1,5 = 4.9, p = 0.079; Fig. 3).

Discussion
We found clear evidence of an altitudinal shift in pellet locations from lower elevations in spring to higher elevations in autumn. This finding is supported by a shift of individual core activity areas. The use of higher amounts of alpine meadow habitat later in the vegetation period is in line with these results.
Upward altitudinal shifts of mountain hares from spring to autumn have also been recorded by Slotta-Bachmayr (1998) in the Austrian Alps. He noted that altitudinal shifts may be driven by snow cover, which influences food availability. Accordingly, upward shifts start in spring and follow the receding snow line until early winter, when snow cover begins to reduce food availability again (Slotta-Bachmayr 1998). The length of this vegetation pattern depends on the inter-annual variability of winter conditions and duration and could influence the between year difference in the altitudinal shift of activity areas. Altitudinal shifts to higher altitudes after snow melting have also been shown in other herbivores living in mountain habitats, such as red deer, Alpine ibex (Apra ibex and chamois Rupicapra rupicapra (Grignolio et al. 2003, Zweifel-Schielly et al. 2009, Darmon et al. 2012. Mason et al. (2017) showed that the shifting pattern of Alpine ibex is caused by a tradeoff between behavioral thermoregulation and foraging. During warmer periods, ibex utilize areas that are less productive and therefore consume lower quality forage (Mason et al. 2017). We cannot rule out such a mechanisms for the mountain hare as the species has been shown to be sensitive to warmer temperatures (Rehnus et al. 2018). The average ambient temperature over the study period is 3.0°C in autumn and as such 4°C higher than in spring, but 8°C lower than in summer (Rehus and Bollman unpubl.). Accordingly, mountain hares would be expected to occur much higher in the summer season. However, this is not possible due to topographic limitations. As mountain hares also stay in the area in summer, we assume that they move upward to co-benefit from more comfortable ambient temperatures and mainly form higher quality food which is linked to the higher proportions of meadows above the timber line. Similarly, it has been shown that altitudinal shifts by red deer are driven by food quality and the maximal protein concentrations of graminoids and herbs (Albon andLangvatn 1992, Zweifel-Schielly et al. 2009). If high-quality food is not available, mountain hares are likely to use an intermediate feeding strategy in which they switch from grazing to browsing (Hulbert et al. 2001). Only minor altitudinal shifts of mountain hares have been observed in the Italian Alps (Bisi et al. 2011). This is likely due to differences in resource distribution and availability between study areas.
The use of higher-elevation habitat increases the risk of predation (e.g. eagle owl Bubo bubo) because meadows offer limited cover during hares' nocturnal activities (Rehnus 2013). However, it has been observed that large herbivores often favor improved forage options over predator avoidance (Fryxell and Sinclair 1988). Similarly, the high site fidelity of mountain hares (Bisi et al. 2011) is likely due to the increased benefits of familiarity: the more familiar an animal is with an area's distribution of its conspecifics, predators and resources, the greater advantage it has (Nathan et al. 2008, van Moorter et al. 2009).
Our study presents details about altitudinal shifts in the feeding sites and activity areas of mountain hares in the Alps. However, further research is needed on how these shifts are affected by climate change (Rehnus et al. 2018) and how they interact with nutritional needs and thermoregulatory mechanisms. Table 1. Altitudinal shifts in individual core activity areas from spring to autumn in the Swiss National Park. Altitudinal shifts were calculated as the difference in the altitude of the core activity area in autumn and spring for unique individuals (2 males, 1 female) with a minimum of 10 observations at unique locations per season within the same year from 2014 to 2019.