Banner-tailed kangaroo rats (Dipodomys spectabilis) have been used as models to explore competition, examine mating systems, and assess interactions among biotic and abiotic forces shaping population dynamics. However, forces shaping variation in individual survival, a key element for modeling population dynamics and assessing risk of population extinction, have received little attention. By unraveling the sources of variation in survival it is possible to gain a more thorough understanding of the evolution of life-history traits and the potential conservation needs of this keystone species. Our objective was to use population-level and environmental drivers to explain the variation in apparent annual survival among distinct populations of D. spectabilis. We collected mark-recapture data (2004-2007) from 11 populations in southern New Mexico and estimated annual survival using a model selection framework. A two-step approach was employed: 1) survival was modeled using categorical sources of variation and 2) a priori hypotheses divided categorical variation into specific population-level and environmental drivers. Models were ranked according to QAIC_c and estimates were drawn from multiple models to reduce the effect of model uncertainty. The importance of categorical sources of variation was ranked using the sum of QAIC_c weights (w).

Results/Conclusions

Annual survival was higher for females (Φ = 0.35 – 0.39 [SE = 0.04 – 0.06]) than for males (Φ = 0.28 – 0.31 [0.04 – 0.05]) and higher for adults (Φ = 0.35 – 0.37 [0.03 – 0.06]) than for subadults (Φ = 0.13 – 0.23 [0.06 – 0.09]). Average annual survival at each site increased through time, from a low of 0.17 (0.07) to a high of 0.52 (0.09). Recapture probability was high (0.92 [0.05]) and constant.  Age (w = 0.91) had overwhelming support as an important source of variation. Further decomposition of age revealed important interactions with time (trend effect: w = 0.39) and space (w = 0.35). Differences in survival between the sexes may be related to mate seeking activities of males whereas the importance of age likely reflects subadult naïveté and dispersal, which are predicted to increase mortality.  Temporal trend effect may have been related to the cessation of a 3-year drought (2001-2003) that was followed by 4 years of successively increasing rainfall (2004-2007). A bimodal distribution of precipitation drives primary production in the Chihuahuan Desert and topography yields heterogeneous patterns of vegetation; hypotheses based on these environmental drivers are currently being examined.