Stochastic demography and population viability of the emperor penguin: A march to extinction

To address the effect of past and possible future climatic conditions on the population viability of the emperor penguin, we analyze stage-classified demographic models based on 43 years of data from Pointe Géologie, Antarctica. Distributed around Antarctica, they are unique in breeding during the Antarctic winter. Because emperor penguins only occur where there is sea ice, they are sensitive to changes in sea ice characteristics, and sea ice characteristics are sensitive to climate change. Penguin population growth was strongly affected by a regime shift in the 1970s, during which unusual warm anomalies caused a population crash. The population consequences of climate variability depend on how the vital rates and their fluctuations determine population growth rates. A set of deterministic models for the period before, during, and after the regime shift produced population growth rates of 1.02, 0.94 and 0.99 per year, respectively. Thus warmer conditions caused a serious reduction in population growth rate. We constructed stochastic models for the three periods by randomly selecting one of the annual matrices each year, with a range of values of temporal autocorrelation. Neither environmental variability nor autocorrelation caused appreciable changes in population growth rate. We then constructed a stochastic model based on a two-state model for the environment (warm versus regular conditions) and studied the effect of the frequency and duration of warm events. This model produced mostly negative population growth rates, except when the frequency and duration of warm events were both low. Because we hypothesize that the effects of warm conditions are mediated by sea ice, we related the transition probabilities to past and projected future sea ice conditions. Based on projected future trends in sea-ice, this model suggests a serious risk of both short-term and long-term decline of the emperor penguin population.