Sensitivity of population growth to biotic interactions varies systematically with abiotic stress: Drivers of small mammal population dynamics in a Kenyan savanna

Background/Question/Methods

A long-standing ecological theory, first proposed by Darwin, suggests that the relative importance of abiotic stress v. biotic interactions (e.g., predation, competition) varies systematically across stress gradients. Darwin’s hypothesis suggests that abiotic stress strongly influences population dynamics in relatively stressful areas, whereas biotic interactions are more important in relatively unstressful areas. We used a large-scale herbivore exclosure experiment arrayed across a pronounced rainfall gradient in East Africa (UHURU) to test this prediction for three small mammal species: Aethomys hindei, Gerbilliscus robustus, and Saccostomus mearnsi. We trapped small mammals in a Robust Design framework at 2-month intervals for 58 months. We estimated survival and movement of each species with Program MARK, using AIC to select among models containing rainfall, large mammal density, and small-mammal competitor density as predictor variables. We identified competitors by calculating overlap of species pairs in stable isotope d13C/ d15N space. We synthesized these vital rates into estimates of population growth rate (lambda) for populations spanning this rainfall gradient, and compared elasticities to small- and large-mammal density as a function of rainfall. 

Results/Conclusions

Abiotic stress (rainfall) was included in the best-fit model for vital rates in all three species; however, species responded differently to rainfall. Gerbilliscus robustus lambda was higher in arid areas than in mesic areas (p<0.05), whereas lambda values for A. hindei and S. mearnsi were higher in mesic areas. Large mammal density depressed population growth of G. robustus vital rates. For S. mearnsi and A. hindei, small mammal competitor density, but not large mammal density, influenced lambda. Competitor density decreased lambda, though the strength of competitors’ per-capita effects differed across species. Consistent with Darwin’s hypothesis, the elasticities of populations to biotic interactions (large mammal density for G. robustus and competitor density for S. mearnsi and A. hindei) were higher in more stressful areas (mesic areas for G. robustus and arid areas for S. mearnsi and A. hindei). These results suggest that the strength of biotic interactions for population performance changes predictably with abiotic stress.