Eliciting demographic drivers of population response to environment stochasticity is a central issue in population ecology. Recent studies show that one of the major weakness in assessing the ecological response to climate change is in neglecting spatial and temporal autocorrelation. However, temporal auto-correlation can have sizable effect on stochastic population growth rate. We simulated the demographic effect of serial auto-correlations (-0.5 to +0.9) on the stochastic growth rate of twelve populations of a tropical tree (Khaya senegalensis) in West Africa. We tested how such effects change with additional effects of non-lethal but demographically damaging harvest of foliage (defoliation) in two contrasting climatic contexts. We used stochastic life table response experiment (SLTRE) to disentangle the contributions of the mean and variability in the demographic processes (stasis, growth, shrinkage, and fertility) to deviation of the stochastic population growth rate from the average trend.
Results/Conclusions
We found positive association between stochastic growth rate (λs) and serial auto-correlation for some populations, and quite the opposite trend was observed for other populations with no clear additional effect of climate or harvesting. λs was more sensitive to perturbation of the mean and variance of stasis than to growth, shrinkage and fertility. Furthermore, elasticity of λs to mean stasis was negatively associated with serial auto-correlation in the drier region while opposite trend was observed in the moist region. Interestingly, both negative and positive value of the elasticities to perturbation of the variability of stasis were positively associated with serial auto-correlation. Differences in mean growth followed by differences in mean stasis were the most contributors to differences in stochastic population growth rates. In contrast, contribution of variability in growth outweighed that of other life cycle components. Our study showed population-specific demographic response to serial auto-correlation perhaps due to the high heterogeneity in local biotic and abiotic conditions. Altogether, these findings suggest that serial-autocorrelation can shift population stochastic dynamics from growing to declining.