Actual precipitation, predicted precipitation, and large herbivore condition in arid and semi-arid southern New Mexico

Abstract

Abstract Precipitation affects large herbivore performance, but data is often available from only a limited number of weather stations that may not correspond with home ranges. Remotely-sensed estimates could potentially address this by providing estimates of precipitation at any point on the landscape. We tested whether 4-km resolution precipitation estimates derived from the Parameter-elevation Relationships on Independent Slopes Model (PRISM) were related to nutritional condition of 3 species of large herbivores and accurately predicted precipitation as measured at weather stations in arid New Mexico. Canonical correlation analyses indicated that condition of elk, mule deer, and pronghorn was related to both PRISM estimates and actual precipitation, but lactation status always had a stronger effect on condition with the exception of pronghorn. Both PRISM and actual precipitation showed comparable influence of precipitation, and the same general ranking of importance of seasonal precipitation, on condition. PRISM estimates were generally positively related to actual precipitation, but deviance was often high and PRISM tended to underestimate actual precipitation. For our study populations, PRISM provided no advantage over the use of dispersed weather stations for assessing nutritional condition. Performance of PRISM may have been hindered because of lack of spatial autocorrelation in precipitation within our sites.