Anthropogenic refugia ameliorate the severe climate-related decline of a montane mammal along its trailing edge

Abstract

We conducted detailed resurveys of a montane mammal, Urocitellus beldingi, to examine the effects of climate change on persistence along the trailing edge of its range. Of 74 California sites where U. beldingi were historically recorded (1902-1966), 42 per cent were extirpated, with no evidence for colonization of previously unoccupied sites. Increases in both precipitation and temperature predicted site extirpations, potentially owing to snowcover loss. Surprisingly, human land-use change buffered climate change impacts, leading to increased persistence and abundance. Excluding human-modified sites, U. beldingi has shown an upslope range retraction of 255 m. Generalized additive models of past distribution were predictive of modern range contractions (AUC = 0.76) and projected extreme reductions (52% and 99%, respectively) of U. beldingi's southwestern range to 2080 climates (Hadley and CCCMA A2). Our study suggests the strong impacts of climate change on montane species at their trailing edge and how anthropogenic refugia may mitigate these effects.

Figure 1.

Surveyed sites in California showing where U. beldingi has persisted in green (n = 43) and been extirpated in black (n = 31). Triangles indicate anthropogenic sites (e.g. campgrounds or agricultural fields). Red crosses (n = 47) mark resurvey sites where U. beldingi were absent in the historical and modern time periods; there are no known instances of colonization.

Figure 2.

Classification tree showing the effect of climate factors and human modification on U. beldingi persistence. Grey shading denotes the proportion of persistent sites modelled in each category. Increased probability of persistence is associated with low (average less than or equal to 4.4°C) coldest quarter temperatures and ‘anthropogenic refugia’ created by a human-mediated increase in food and water availability. Although modern wettest quarter precipitation and change in warmest quarter temperature were included in the best model, they were not significant at p < 0.05 and thus do not appear in the figure.

Figure 3.

Probability of U. beldingi persistence increases with elevation and is lower in natural sites (grey dots and line) than in human-modified sites (black dots and line). Such ‘anthropogenic refugia’ are especially important at low elevations (trailing edge).

Figure 4.

Urocitellus beldingi species distribution model (GAM) results projecting (a) current distribution from historical presences and absences, (b) future distribution from 2080 Hadley, A2 scenario and (c) future distribution from 2080 CCCMA, A2 scenario. Probability of persistence is marked by quantile (0–0.2 not shown, 0.2–0.4 in beige, 0.4–0.6 in light blue, 0.6–0.8 in blue and 0.8–1 in dark blue).