A state-dependent model for assessing the population consequences of disturbance on income-breeding mammals

Abstract

Human activities continue to expand in marine and terrestrial environments, leading to increased interactions with wildlife that can have negative impacts on population dynamics. Approaches for quantifying how these interactions translate to population-level effects are therefore crucial for effective management practices and balancing human-wildlife tradeoffs. We developed a method using state-dependent behavioral theory implemented via Stochastic Dynamic Programming (SDP) for predicting the population consequences of disturbance on the physiology and reproductive behavior of an income-breeding mammal, using California sea lions (Zalophus californianus) as a motivating species. Emergent properties of the model included reproductive characteristics associated with long-lived species, such as variation in the age at first reproduction, early termination of pregnancy, and skipped breeding. In undisturbed model simulations, reproductive rates and the average wean date were consistent with empirically-derived estimates from sea lions and other marine mammals, highlighting the utility of this model for quantifying fecundity estimates of data-deficient species and addressing fundamental ecological processes. In disturbed model simulations, exposure to prolonged, repetitive disturbances negatively impacted population growth; in addition, short, infrequent disturbances had the potential for adverse effects depending on the behavioral response of sea lions and the probability of being disturbed. The adverse effect of disturbance on population dynamics was due to a combination of reduced pup recruitment (survival to age one) resulting from a lower wean mass and increased abortion rates that led to skipped reproductive years, both of which have been documented for marine mammal populations experiencing natural fluctuations in prey availability. The derivation of state- and time-dependent reproductive decisions using an SDP model is an effective approach that links behavioral and energetic effects at the individual level to changes at the population level, and one that serves a dual purpose in the ability to quantify basic biological parameters and address ecological questions irrespective of disturbance.