3.4 Stable isotope analysis
Mice showed considerable variation in values of all three of the stable isotopes examined (Figure 5a-c, Table 2). Patterns included isotopic variation both among sites (indicating an elevational effect) and within sites (indicating individual variation in foraging habits). In the total dataset, δ13C values of P. vaccarum (Figure 5a, Table 2) varied between -22.8 and -12.7 ‰, with more13C-enriched values being recorded at lower elevations (<4000 m). However, mice with relatively13C-depleted values were captured at both sites 1 and 2. There was strong statistical support for inter-site differences in δ13C (PERMANOVA: PseudoF5,34 = 36.6, P9999 perms = 0.0001). Post-hoc comparisons indicated significant differences (P < 0.05) between δ13C values for all sites apart from sites 3 and 4 (t = 1.42, P = 0.13), sites 3 and 6 (t = 1.92,P = 0.05), and sites 4 and 5 (t = 1.4, P = 0.14).Phyllotis vaccarum liver δ15N ranged between 6.2 and 22.8 ‰ (Figure 5b, Table 2), with considerable variation among sites (PseudoF5,34 = 39.3, P9999 perms = 0.0001). Values were notably 15N-enriched at site 1 but included one individual with relatively low δ15N.Post-hoc comparisons showed overlap in δ15N values at sites 2 and 6 (t = 0.10, P = 0.92), sites 3 and 4 (t = 0.05, P = 0.97), sites 3 and 6 (t = 2.27,P = 0.05), and sites 4 and 5 (t = 0.97, P = 0.34). In the cases of both δ13C and δ15N,P. vaccarum showed a similar pattern of 13C- and 15N-enriched values at sites from lower elevations (< 4000 m) relative to individuals captured between 4000 and 5000 m (Figure 6, Table 2). This contrasted with the pattern in δ34S (Figure 5c, Table 2) where P. vaccarum showed less variation in general, with values ranging between -2.5 and 2.4 ‰, but exhibited a positive shift between34S-depleted values at sites <4000 m and34S-enriched values >4000 m. Values of δ34S for mouse livers varied among sites (PseudoF5,34 = 25.3, P9999 perms = 0.0001), although variation was lower than that observed for C and N.Post-hoc comparisons indicated that δ34S values were similar for mice captured from sites 1 and 2 (t = 0.93,P = 0.70), sites 3 and 5 (t = 1.39, P = 0.18), sites 3 and 6 (t = 0.67, P = 0.52), sites 4 and 5 (t =1.35, P = 0.21), sites 4 and 6 (t = 1.8,P = 0.09, and sites 5 and 6 (t = 0.84, P = 0.45).
Mice from the highest elevations did not exhibit negative δ15N values, suggesting that lichenivory is not especially common.
3.4.1 Assignment to capture elevations
As abiotic and biotic conditions (e.g. temperature, aridity, UV concentrations, plant nutrient availability, soil organic content) change with elevation, so too do the biomass and community composition of the primary producers (Díaz et al. , 2019), with consequent changes in the availability of food for consumers and isotopic shifts at the base of the food web (Díaz et al. 2016). Given the known elevational gradient in stable isotope values, we can expect that stable isotope values will provide a means of identifying variation in habitat use among mice captured at different elevations.
Analysis of the combined stable isotope dataset using CAP showed that mice could be reliably assigned to broad elevational zones using individual δ13C, δ15N and δ34S values (Figure 5d) (CAP: Trace = 1.71, P = 0.0001). The leave-one-out classification cross validation (Table 3) indicated that by using the three isotope ratios we could assign an individual mouse to a 1000 m interval with ~85 ‰ success. The CAP model predicted that the 6739 m summit mouse from Volcán Llullaillaco was isotopically most similar to mice captured from the 4000-5000 m interval (Figure 5d). The CAP model also suggested that 4 other individuals had stable isotope values characteristic of elevational zones distinct from where they were captured (Table 3). Including the summit mouse, this indicates that ca. 15 ‰ of the P. vaccarum in the study area had stable isotope values suggestive of upslope or downslope dispersal.
3.4.2 Trophic Position
Modal estimates of P . vaccarum trophic position (TP) varied between capture sites (Table 4) and ranged from 1.9 at site 2 to 4.3 at site 1. The latter estimate is extremely high and reflects the very high δ15N values from mice collected at site 1 (mean δ15N = 19.2 ‰). Discounting the results from site 1, mouse trophic position estimates were generally similar across sites and were indicative of omnivory with modal values between 1.9 and 2.3 at sites 2 to 5. The modal estimate for mice at site 6 was slightly higher (TP = 3.4), but the credibility limits overlapped with those from all sites apart from site 2. TP values between 2-3 are indicative of an herbivorous diet that includes some animal prey. TP values prey. The TP of the summit mouse from Volcán Llullaillaco was estimated as 2.2, quite close to that of mice from sites 3, 4 and 5.