A connectivity and wildlife management conflict in isolated desert waters

In deserts, amphibian reproduction is restricted to a dynamic network of small, isolated, ephemeral water sites, which support gene flow between isolated populations. Establishment of artificial catchments to augment natural water sources for target species other than amphibians (e.g., larger-bodied and more-vagile game species) can increase connectivity between potential breeding sites and populations of amphibians. These anthropogenic waters, however, may differ in quality from natural waters, with ammonia concentrations high enough to potentially affect amphibian health, reproduction, and population persistence. Thus, water supplementation has set up a potential conflict for managing landscape connectivity for sympatric species differing in dispersal abilities. To explore this possibility, we used graph theory to examine potential connectivity under current and potential future scenarios among natural, modified natural, and human-made waters on the United States Air Force's Barry M. Goldwater Range East (BMGR-E) and adjacent Bureau of Land Management (BLM) lands in Arizona. The network of 117 waterbodies on the study site (82 natural and unmodified, 35 anthropogenic or modified) coalesced at 15.7 km, meaning that wildlife must be capable of traveling ≥15.7 km from any water to the next to traverse the network, assuming all the waterbodies are in fact wet. At this threshold dispersal distance, we identified those waters that played important linkage roles in supporting connectivity through the network, 23% (5 of 22) of which were found to be either anthropogenic or modified waters, indicating that human activities play an outsized role in countering natural isolation of the waters. Two of these important waters were hubs (i.e., waters linked to a high number of nearby wetlands), meaning that the potential negative factors in a catchment could affect relatively more wildlife than in waters not associated with a hub. Because the coalescence distance exceeded the dispersal range of many smaller-bodied species, we also examined 2 smaller dispersal distances more appropriate for smaller-bodied and less-vagile species like amphibians (0.6 km and 6 km). We identified 2 wetland clusters that contained anthropogenic or modified waters. When we performed simulated removal of catchments (representing potential management action), the network showed surprisingly few effects: coalescence distance did not increase, and although the number of wetland clusters changed slightly, wetland density in the majority of clusters was not reduced. These results suggest that catchments could be removed without much negative impact on larger, high-vagility (i.e., those capable of dispersing ≥15.7 km; large mammals) or smaller, low-vagility (e.g., amphibians) species. Simulated removal of waters allowed us to generate a prioritized list of waters found to be consistently important for connectivity conservation for wildlife (large and small) on BMGR-E and adjacent BLM lands. Such an approach could be adopted in any situation where a quantitative assessment of connectivity among habitat patches and management options is needed.