In this post Julia Barthold discusses the recent paper “Bayesian estimates of male and female African lion mortality for future use in population management” by her and co-authors A. J. Loveridge, D.W. Macdonald, C. Packer, and F. Colchero. The article is part of the BES cross-journal special feature “Demography Beyond the Population”.
In 2013, I wanted to project the population dynamics of African lions. This seemingly straightforward task turned out to be deceptively simple when a literature search came up empty for a necessary quantity: estimates of male mortality by year of age. How could this be? It was, after all, lions we aimed to study, not some rare species from the deep sea. It subsequently took the help of three lion experts, two years of hard work, and one Bayesian statistician to fill this knowledge gap.
So what was the problem? Maturing lion males leave their birth pride to seek mating opportunities with unrelated females. In the process, they commonly migrate out of areas covered by field studies. Since neither migration events nor deaths are regularly observed, the fate of males that go missing after the age at maturity remains uncertain: they may have died or dispersed.
To cope with this kind of missing age-at-death data, we developed a model to estimate mortality in a Bayesian hierarchical framework. The model simultaneously estimates the coefficients of a parametric mortality model and a dispersal model.
Using the model, we compared mortality for two populations that experienced different levels of human impact. For an undisturbed population, mortality showed the typical trajectory with age-related changes in risks of death.
In contrast to this, for a population that was exposed to trophy hunting and killings in accidents and human-wildlife conflict, mortality had a different profile. Here, the mortality model detected that about 70 out of 100 male deaths can be attributed to a risk of death that strikes regardless of age. This age-independent risk is usually interpreted as reflecting deaths due to non-natural causes such as hunting.
The result did not surprise us. For the past 15 years Andrew Loveridge and David Macdonald from the Wildlife Conservation Research Unit at Oxford University have been fighting to protect this population from human impact. The population’s plight recently gained tragic fame when Cecil, one of the population’s cherished males, was killed by an American dentist.
Where demography-informed population management measures are needed, mortality estimates are crucial. Our model now widens the range of species for which conservation scientists can estimate mortality because it overcomes an old and pervasive problem in the estimation of mortality: missing death data due to migration. Given the near ubiquity of migration due to dispersal of at least one of the sexes, we hope our model will help to fill similar knowledge gaps for other species in need of population management.