Article Preview
TopIntroduction
Over the last three decades, it has become clear that Earth’s climate is warming, and precipitation is becoming more variable (IPCC 2014). During this present period of climate change, desert mountain ranges may be critical in allowing vegetation to adapt to stress as species migrate upslope where temperatures are cooler and precipitation is greater (Colwell et al., 2008; Nogues-Bravo et al., 2007). Although, in some cases, species can shift downward in elevation with changing precipitation regimes that result in more water being distributed as rain instead of snowpack runoff (Crimmins et al., 2011).
Within the Mojave Desert in southern Nevada, there have been multiple cycles of drought throughout the last century, and while these vary on interannual and decadal scales, oscillations can occur every five years (Hereford et al., 2006). During periods when El Niño dominates, or the Pacific Decadal Oscillation is positive, precipitation is above average (Blainey et al., 2007; Hereford et al., 2006). However, the Mojave has experienced below average precipitation for much longer periods than above average precipitation over the last three decades. Compounding this drying, studies from the Mojave, including sites in southern Nevada, demonstrate rapid warming over the last several decades (Bai et al., 2014; Rapacciuolo et al., 2014). This warming is most drastic when considering minimum average temperatures (Miller, 2011), and these combined precipitation and temperature changes that correspond to less water availability are reducing areal habitat of climate-sensitive species (Guida et al., 2014).
To model how these climatic changes could influence species distributions, ecological niche modeling is often used. Maxent, one of the most prevalent models used in the biogeographical literature, has been used to develop over 2,000 species distribution models (Bradie and Leung, 2017). When using Maxent, temperature and precipitation data usually have the largest influence on habitat models when compared to other environmental and climate variables that have been used as independent variables (Bradie & Leung, 2017). Franklin et al. (2013) demonstrated that models using downscaled climate data as fine as 90-m resolution can be used from projections, whereas studies documented by Elith et al. (2011) used coarser resolution climate grids (i.e., 4 km or greater) to successfully demonstrate temporal and spatial changes in species likely habitat. However, the spatial resolution at which species should be modeled is not definitively settled, though when available, incorporating microclimatic data improves models (Lembrechts et al., 2019; Strachan and Daly, 2017; Fridley, 2009).
Climate grid resolution becomes more important for study areas with high topographic variation, since finer resolution data are necessary to represent microclimates that may provide refuge to species (Franklin et al., 2013). However, in areas such as southern Nevada, weather stations from which downscaled climate datasets are derived are sparsely distributed, with more observations concentrated around Las Vegas and larger cities rather than in remote areas that are often the focus of biogeographical research (Hereford et al., 2006; Miller, 2011). Additionally, finer-scale models that may be able to more accurately represent potential microclimatic habitats (e.g., areas with cold air drainage) could also overfit models by underpredicting the potential suitable habitat where species could be found. This is especially likely if there are not field samples of presence locations that correspond with the values in these finer-resolution climate grid cells within a raster dataset. This was the case in Joshua Tree National Park where a finer-scale analysis of Joshua trees was more optimistic than previous climate changed-influenced habitat models that showed declines at a coarser scale (Barrows and Murphy-Mariscal, 2012).