Interacting effects of unobserved heterogeneity and individual stochasticity in the life history of the southern fulmar
Résumé
1. Individuals are heterogeneous in many ways. Some of these differences are incorporated
as individual states (e.g. age, size, breeding status) in population models.
However, substantial amounts of heterogeneity may remain unaccounted for, due
to unmeasurable genetic, maternal or environmental factors.
2. Such unobserved heterogeneity (UH) affects the behaviour of heterogeneous cohorts
via intra-cohort selection and contributes to inter-individual variance in demographic
outcomes such as longevity and lifetime reproduction. Variance is also
produced by individual stochasticity, due to random events in the life cycle of wild
organisms, yet no study thus far has attempted to decompose the variance in demographic
outcomes into contributions from UH and individual stochasticity for an
animal population in the wild.
3. We developed a stage-classified matrix population model for the southern fulmar
breeding on Ile des Pétrels, Antarctica. We applied multievent, multistate mark–recapture
methods to estimate a finite mixture model accounting for UH in all vital
rates and Markov chain methods to calculate demographic outcomes. Finally, we
partitioned the variance in demographic outcomes into contributions from UH and
individual stochasticity.
4. We identify three UH groups, differing substantially in longevity, lifetime reproductive
output, age at first reproduction and in the proportion of the life spent in each
reproductive state.
– 14% of individuals at fledging have a delayed but high probability of recruitment
and extended reproductive life span.
– 67% of individuals are less likely to reach adulthood, recruit late and skip breeding
often but have the highest adult survival rate.
– 19% of individuals recruit early and attempt to breed often. They are likely to raise
their offspring successfully, but experience a relatively short life span.
Unobserved heterogeneity only explains a small fraction of the variances in longevity
(5.9%), age at first reproduction (3.7%) and lifetime reproduction (22%).
5. UH can affect the entire life cycle, including survival, development and reproductive
rates, with consequences over the lifetime of individuals and impacts on cohort
dynamics. The respective role of UH vs. individual stochasticity varies greatly among
demographic outcomes. We discuss the implication of our finding for the gradient of
life-history strategies observed among species and argue that individual differences
should be accounted for in demographic studies of wild populations.