1. INTRODUCTION
Extreme high-elevation surveys of small mammals in the Central Andes have yielded live captures of numerous specimens of the Andean leaf-eared mouse Phyllotis vaccarum at elevations at or above the elevational limits of vegetation (>5,000 m)(Storz et al., 2024). One specimen was captured at 6,739 m (22,100 feet) above sea level on the very summit of Volcán Llullaillaco, a stratovolcano in the Central Andes that straddles the Argentina-Chile border (Storz et al., 2020). This summit specimen far surpasses previous elevational records for wild mammals in the Andes and Himalaya. Documentation of active burrows of P. vaccarum at >6,100 m on the flanks of Llullaillaco and the discovery of desiccated cadavers (‘mummies’) ofPhyllotis on the summits of Llullaillaco and several neighboring elevations well above the apparent limits of vascular plants (Halloy, 1991; Steppan et al., 2022; Storz et al., 2023, 2024). Evidence that high-elevation mice are living in an apparently barren world of rock, ice, and snow prompts numerous questions, perhaps none more basic than: What are they eating? In the perennial winter conditions that prevail at elevations >6,000 m, the scarcity of food poses a special physiological challenge for small endotherms like mice because of the energetic demands of thermoregulation. Moreover, leaf-eared mice in the genus Phyllotis do not hibernate, so the energetic challenge of sustaining endothermy in cold, hypoxic conditions is especially acute (Storz & Scott 2019).
It has been suggested that windblown arthropods and/or vegetation could provide a source of food for animals living at elevations that exceed the limits at which green plants grow (the ‘aolian zone’) (Swan, 1961, 1992). According to this hypothesis, the transport of airborne nutrients from lower elevations sustains life on the upper reaches of a volcano like Llullaillaco similar to the way that the fallout of organic detritus from upper layers of the water column sustains life in the aphotic zone of the deep ocean. Aolian deposits of windblown arthropods along the lee edge of mountain summits and ridgelines can attract birds and other insectivorous animals that normally forage at much lower elevations (Antor, 1995; Spalding, 1979). If aolian deposits of windblown arthropods (‘arthropod fallout’) help sustain populations ofP. vaccarum at extreme elevations on Atacama volcanoes, we would expect arthropods to constitute a much larger part of their diet than at lower elevations where the species is mainly granivorous and frugivorous (Bozinovic and Rosenmann 1988; López-Cortés et al. 2007; Sassi et al. 2017). On the Quinghai-Tibetan Plateau, high elevation plateau pikas (Ochotona curzoniae ) exploit the feces of yak (Bos grunniens ) as a food source (Speakman et al., 2021). IfPhyllotis mice practice a similar form of interspecific coprophagy, feces from Andean camelids such as vicuña (Lama vicugna ) and guanaco (Lama guanicoe ) would provide the most readily available source. Although neither vicuña nor guanaco typically spend much time above the elevational limit of vegetation, both species are known to traverse mountain passes at elevations >5,500 m in the Central Andes (J. Storz, personal observation). It is also possible that the mice feed on lichen that grows on rock substrates (saxicolous lichen) or some cryptic form of vegetation that is not currently known to occur at such extreme elevations.
Here we test the above-mentioned hypotheses by conducting metagenomic and metabarcoding analyses of gut contents from the world-record specimen of P. vaccarum that was live-captured on the summit of Llullaillaco at 6,739 m. Since the metagenomic approach involves high-throughput sequencing of all DNA extracted from a sample without PCR enrichment of specific markers, it is not biased by a priori expectations about which taxonomic groups to expect and is therefore well-suited to dietary assessments of omnivorous species (Chua et al., 2020). DNA metabarcoding complements the metagenomic approach and can be used to estimate the diversity and relative abundance of different items in the diet (Deagle et al., 2019; Stapleton et al., 2022).
To complement the metagenomic and metabarcoding analysis of the summit specimen, we conducted a stable isotope analysis of liver samples from a larger sample of wild-caught mice from a broad range of elevations in the Chilean Altiplano and Puna de Atacama (2,370-6,739 m). We used stable isotope values of three key elements (carbon, nitrogen and sulfur) from liver tissue to characterize the diet of P. vaccarum over a timespan of weeks to months. In the livers of small mammals, isotopic half-lives are <1 week for both carbon (δ13C) and nitrogen (δ15N) and we expect a similar half-life for sulfur (δ34S). Examination of stable isotope values permits inferences about several key components of trophic ecology.
Carbon stable isotopes (δ13C) reflect the relative consumption of food derived from different sources of primary production ( Cerling et al. , 1997; DeNiro & Epstein, 1978). The stable isotope of nitrogen is typically used as an indicator of consumer trophic position (Vanderklift & Ponsard, 2003; Quezada-Romegialliet al. , 2018) but can also be used to discriminate between consumption of food from distinct habitats (Harrod et al., 2005). For example, it should be possible to assess the extent to which mice rely on lichen at high elevations. Lichens are typically very15N-depleted relative to terrestrial plants (Fogelet al., 2008; Lee, Lim & Yoon, 2009; Pinho et al., 2017), and this holds true for lichens from high elevations (Biazrov, 2012; Marris et al., 2019; Szpak et al. , 2013) and volcanic fumeroles (Tozer et al. 2005). If lichen forms an important part of the diet of P. vaccarum at high elevations, we would expect to observe negative δ15N values. Sulfur stable isotopes (δ34S) are also useful indicators of consumer habitat use as values measured from plants and their consumers often exhibit high levels of spatial variation across biogeochemical gradients (Krouse et al., 1991; Nielsen et al. , 1991), as might be expected along the flanks of an historically active volcano like Llullaillaco.
By combining metagenomics, metabarcoding, and stable-isotope analyses, we tested several hypotheses about the diet of mice living at extreme elevations. The arthropod fallout hypothesis would be supported by the presence of DNA from insects or other arthropods that could be blown upslope, and isotopic estimates of trophic position would be higher for mice living at especially high elevations on the flanks or summit of the volcano in comparison with those from lower elevations in the surrounding Altiplano. Interspecific coprophagy (or scavenging) would be supported by the presence of DNA from vicuña, guanaco, or other co-distributed mammals. Lichenivory would be supported by the presence of DNA from lichen-associated fungi or green algae and especially low15N values in mice from high elevations. The particular plants detected in the gut contents of the summit mouse may suggest that some plant species actually occur at much higher elevations than currently assumed, but they may be sparsely distributed in cryptic microhabitats (e.g., in rock crevices or under the snowpack). If that is the case, then the diet of mice living at >6,000 m may include a subset of the same plants that mice feed on at lower elevations.