Life history and phylogeny explain the bias in sensitivity analysis for structured population models

A primary aim in conservation is to either bolster threatened populations or eradicate invasive ones. Classical elasticity analysis suggests the survival of long-lived adults will have the largest impact on changing population growth, but this result relies on a linear approximation. In reality, the population growth response to perturbing a single demographic rate is nonlinear. We applied exact perturbation analyses to a large database of plant matrix population models in a meta-analysis to determine when and under what conditions population growth may be ‘stabilized’. We found this relationship to be strongly nonlinear, making standard methodology misleading. The vital rate change required for population replacement (λ = 1) is more biased and more likely to fail when mitigating decline as opposed to controlling population expansion. However, these biases and conservation deficits are structured among vital rates and life history strategies. The patterns we describe provide guidance not only for when to expect decits and bias, but also for deciding which vital rates may be most responsive to conservation interventions.