Results
Our original models suggested that the elevational distributions of 19 of the 32 species shifted (Tables S1-S4). Simulations indicated high confidence that the shifts of 12 species were not an artifact of stochasticity (Table S5 and S6). Further discussion focuses on these 12 species.
In the western Great Basin, the elevational distributions of four species shifted. The distributions of three species shifted downslope along the full elevational gradient: House Wren (Troglodytes aedon ) by 99 m, Black-headed Grosbeak (Pheucticus melanocephalus ) by 65 m, and Lazuli Bunting (Passerina amoena ) by 194 m (Table 1). At the lower edge of the elevational gradient, the distributions of three species, including two that shifted along the full elevational gradient, changed. Warbling Vireo (Vireo gilvus ) shifted upslope, whereas House Wren and Lazuli Bunting shifted downslope (Table 1, Figure 2). At the upper edge of the elevational gradient, the distribution of House Wren shifted downslope.
In the central Great Basin, no species shifted along the full elevational gradient. However, movement at the edges of the elevational ranges was relatively common. At the lower edge of the elevational gradient, five species’ distributions changed. Brewer’s Sparrow (Spizella breweri ) and Lazuli Bunting shifted upslope, whereas Broad-tailed Hummingbird (Selasphorus platycercus ), Mountain Chickadee (Poecile gambeli ), and Northern Flicker (Colaptes auratus ) shifted downslope (Table 2, Figure 3). Lazuli Bunting was the only species that shifted within the lower edge of the elevational gradient in both the western and central Great Basin. However, the species shifted downslope in the western Great Basin and upslope in the central Great Basin. At the upper edge of the elevational gradient, the elevational distributions of five species changed. Mountain Chickadee, Rock Wren (Salpinctes obsoletus ), and Vesper Sparrow (Pooecetes gramineus ) shifted upslope, whereas Mountain Bluebird (Sialia currucoides ) and Spotted Towhee (Pipilo maculatus ) shifted downslope (Table 2, Figure 3).
The average distance moved was smaller at elevational edges than along the full gradient. In the western Great Basin, the absolute value of the average elevational shift across the full elevational gradient was 119 m, compared to 50 m at the lower edge and 59 m at the upper edge (Table 1). In the central Great Basin, the absolute value of the average elevational shift was 33 m at the lower edge and 48 m at the upper edge (Table 2).
Irrespective of distributional shifts, many of the associations between bird occupancy and temperature or precipitation were statistically significant, and this information is relevant to understanding species’ responses to climate variables. However, given that the effect of elevation on temperature or precipitation did not change through time (e.g., higher elevations do not seem to be warming faster than lower elevations, or receiving more or less precipitation through time), these associations should not necessarily be interpreted as drivers of the observed distributional shifts. Occupancy of 10 of the 12 species with elevational distributions that shifted was significantly associated with winter or spring precipitation (Tables 1 and 2). Spring temperature was associated with shifts of 2 of the 12 species, and NDVI was associated with shifts of 5 of those species. Only occupancy of Spotted Towhee in the central Great Basin and Lazuli Bunting in the western Great Basin was not significantly associated with any of those four variables. Both associations with spring temperature were positive. In all but two cases (House Wren in the western Great Basin and Lazuli Bunting in the central Great Basin), the association with precipitation was negative. NDVI was significantly related to occupancy of House Wren in the western Great Basin both across the full elevational gradient and within the lower edge (positive association; Table 1), and with occupancy of four species in the central Great Basin (two positively and two negatively; Table 2).
Over the study period, spring temperature, winter precipitation, and spring precipitation increased significantly in both the western and central Great Basin (Figure 4). NDVI also changed, but not in a uniform manner. In the western Great Basin, NDVI was negatively associated with the interaction of year and elevation, whereas in the central Great Basin, the association was the opposite (Figure 4); both effects were small. Over time, NDVI decreased as elevation increased in the western Great Basin, and slightly increased with elevation in the central Great Basin.
Discussion
Our results add to a growing body of evidence that many temperate bird species are not consistently shifting upslope as climate changes. Elevational distributions of bird species in the western and central Great Basin shifted in a variety of ways, with a greater number of distributional shifts occurring at the edges of the elevational gradient than along the full gradient. All three distributional shifts along the full elevational gradient were downslope. Our results also illustrate the importance of assessing population variability in conjunction with range shifts. In 7 of 19 cases, what initially appeared to be a deterministic elevational shift is likely attributable to stochasticity. Although bird occupancy was strongly associated with climate variables, there was little consistency between an elevational shift and temporal changes in temperature, precipitation, or primary productivity. The duration of our time-series data was relatively short, but comparable to other studies of elevational shifts (Campos-Cerqueira et al. 2017, DeLuca and King 2017), and indicated considerable plasticity in elevational distributions.
Our simulations explicitly tested whether the linear effect of year was driving significant interactions between year and elevation, which we interpreted as evidence of elevational shifts. Although we did not collect demographic data, we found that observed shifts of seven species could reflect stochastic changes in temporal and elevational occupancy. There are two main reasons why a significant interaction term in the original model might not be consistent with simulation results. First, mean occupancy at high or low elevations might differ between the latest and earliest years, in effect driving a linear trend. Second, annual mean occupancy of some species is highly variable, and the apparent linear trend may have been coincidental. Annual variability in abundance or occupancy of passerines can be caused by factors including conditions at overwintering grounds, cyclic weather events, or variable migration mortality. Population dynamics are rarely considered in studies of distributional shifts, but likely influence the accuracy of detected shifts (McCain et al. 2016).
Among the environmental variables we examined, only NDVI changed over both time and elevation. This may be due to the resolution at which the climate variables and NDVI were measured. Because the resolution of the climate variables was 4 km, some of our survey points within a canyon fell within the same pixel. By contrast, because the resolution of the NDVI data was 250 m, each survey point had a unique value, allowing for finer-resolution changes in NDVI to be captured in our analysis. Additionally, spatial variation in temperature and moisture availability in montane environments is much greater than in lowlands (Suggitt et al. 2011). For example, some narrow montane canyons are prone to temperature inversions (Curtis et al. 2014, Rupp et al. 2020). As a result, microclimate in areas with complex topography may be unpredictable, and short-distance movements may be sufficient for birds to access microclimates favorable for feeding, mating, or nesting.
The majority of observed elevational shifts occurred at range edges. Distributional shifts may be more common at elevational edges than along the full elevational gradient due to higher rates of climate change at higher elevations and reduced competition at lower elevations (Alexander et al. 2015). In our study system, plant phenology at higher elevations can be 21 days later than at lower elevations (Zillig, unpublished manuscript). As climate change results in increasing temperatures and earlier snowmelt, birds may shift upslope to take advantage of habitat that is becoming available earlier in the breeding season, resulting in changes in species-level occupancy. For example, Rock Wren and Vesper Sparrow, both of which nest on the ground, moved upslope at the upper edge of their elevational ranges. Individuals may have dispersed into nesting habitat that previously was covered in snow or did not green up until late in the breeding season.
At the lower range edges, elevational movement may be driven by downslope expansion of riparian vegetation. About 60–70% of the vertebrate species native to the Great Basin, including most of the birds that breed in the region, are associated with riparian areas (Brussard et al. 1998, Poff et al. 2011). NDVI significantly increased during the time span of our study, with NDVI at higher elevations increasing faster in the central Great Basin and NDVI at lower elevations increasing faster in the western Great Basin (Figure 4). The primary productivity and extent of riparian areas may be expanding in some parts of the Great Basin in response to greater water-use efficiency as concentrations of carbon dioxide increase, especially where the intensity of livestock grazing is decreasing, as it is in our study system (Dwire et al. 2018, Fesenmyer et al. 2018, Albano et al. 2020). In the western Great Basin, species that nest in riparian vegetation, such as Lazuli Bunting and House Wren, may be moving downslope in response to expansion of that nesting habitat at the lower edge of their elevational ranges.
In all but two of the 13 instances in which winter or spring precipitation was significantly associated with occupancy, the association was negative (Tables 1 and 2). In general, one would expect bird communities in arid ecosystems to respond positively to increased precipitation, as many species may be limited by water availability (Bolger et al. 2005, Riddell et al. 2019). Our counterintuitive result may be explained by climate change-driven changes in precipitation across the Great Basin. Increased winter and spring precipitation across much of the western United States, including the Great Basin (Chambers 2008), is driven in part by increasingly severe storms (Xue et al. 2017), which could affect birds adversely. Increases in precipitation also may have delayed the breeding season or decreased survival or reproduction, leading to a decrease in occupancy (Kozlovsky et al. 2018, Zuckerberg et al. 2018). Moreover, the proportion of precipitation that falls as rain rather than as snow is increasing, resulting in decreases in snow depth, earlier snowmelt, and water inputs to the soil becoming earlier and more sporadic (Abatzoglou and Kolden 2011, Petersky and Harpold 2018).
Our inferences might be biased if the elevational gradient we surveyed did not encompass species’ full elevational distributions in our study regions. However, our point-count locations appeared to capture the upper limits of each species’ elevational distribution, and the lower elevational limits of most species (Figures S1 and S2). Of the 32 bird species we examined, the elevational ranges of 12 appeared to shift. With the exception of a 194 m downslope shift by Lazuli Bunting in the western Great Basin, the shifts were less than 100 m. The breeding ranges of all species examined extend beyond our study system. We acknowledge that the bird populations we examined may be responding to climate change differently than populations in wetter, less topographically diverse systems, and do not suggest that population-level responses we observed are necessarily indicative of species-level responses.
Although temperature and precipitation changed considerably even over the 10-20 years of our study, few elevational shifts were significantly related to temperature. Diel temperature in our study canyons during the breeding season is highly variable: day and night differ by as much as 19°C (M. Zillig unpublished data). We suspect that Great Basin bird populations have relatively broad thermal tolerances, consistent with higher tolerances of temperate than tropical bird species to high and low temperatures after controlling for body mass and experiential humidity (Pollock et al. 2020). The elevational shifts that we observed were relatively rapid. The lack of consistent associations between elevational shifts and temperature or precipitation suggest that birds may be responding to elements of habitat that are indirectly associated with our measured variables or with those that we did not measure, such as competition or quantity and quality of food.
We are aware of no other studies that examine elevational range shifts in arid bird populations. Our results reinforce that not only are responses to climate species-specific, but birds respond to numerous and compounded types of environmental change. Great Basin bird populations may be responding to climate change through shifts within the edges of the elevational gradient, yet the lack of a strong overall climate-response signal suggests that these populations may be relatively resilient to climate change.