The past, present, and future of predator-prey interactions in a warming world: using species distribution modeling to forecast ectotherm-endotherm niche overlap

Climate change has the potential to disrupt species interactions across global ecosystems. Ectotherm-endotherm interactions may be especially prone to this risk due to the possible mismatch between the species in physiological response and performance. However, few studies have examined how changing temperatures might differentially impact species’ niches or available suitable habitat when they have very different modes of thermoregulation. An ideal system for studying this interaction is the predator-prey system. In this study, we used ecological niche modeling to characterize the niche overlap and examine biogeography in past and future climate conditions of prairie rattlesnakes (Crotalus viridis) and Ord’s kangaroo rats (Dipodomys ordii), an endotherm-ectotherm pair typifying a predator-prey species interaction. 

Methods
We downloaded occurrence records for both prairie rattlesnakes and Ord’s kangaroo rats from GBIF and VertNet using R (R Core Team, 2023). Point databases were cleaned by removing all incomplete records (missing key information, such as latitude or longitude), removing all records that were of subspecies, and removing duplicates. Records that were in the NAD27 projection were converted into the WGS84 projection. We initially defined 22 environmental predictors for use in building our ENMs. This included the 19 bioclimatic predictors (1970-2000) in the WorldClim 1.4 database (Hijmans et al. 2005) as well as elevation, terrain ruggedness index (TRI), and topsoil sand content. We included elevation due to the large variation of this metric throughout both species’ ranges. The elevation layers for both species were GTOPO30 tiles which we downloaded from the USGS Earth Explorer (USGS, 2000). We then merged tiles together using QGIS. We also included terrain ruggedness index as increased ruggedness has been shown to increase microhabitat variation and therefore areas available for snake refuge (Kirk et al., 2021). TRI was calculated from the previously downloaded elevation layer using the terrain ruggedness raster analysis function in QGIS. Lastly, we included topsoil sand content (percent sand) because Ord’s kangaroo rats are more abundant in areas with sparsely vegetated, sandy soils where they can construct their burrow systems (Gummer, 1997; Kissner et al., 2009). Topsoil sand content was downloaded from the Unified North American Soil Map (Liu et al. 2014). All environmental layers were at a 30 arc second (1 km) resolution, projected to WGS84, and were masked to the buffered range of the appropriate species for each model.