Assessing the impact of culling on population size in the presence of uncertain density dependence: lessons from a great cormorant population.

Published online
10 Dec 2008
Content type
Journal article
Journal title
Journal of Applied Ecology

Green, R. E.
Contact email(s)

Publication language
UK & England


In applied population ecology, simulation models are often used to predict the likely consequences of different management options and to inform policy. This study examines a recent population modelling study used to support a decision by the UK Government to increase the legal cull of great cormorants Phalacrocorax carbo in England. The main models used in the study rely heavily upon estimates of the strength of density dependence obtained solely from analysis of a time series of population counts. Previous literature has questioned the validity of estimates of density dependence based upon time-series population data. It is shown, by means of simulations, that an important component of the cormorant population models, the strength of density dependence, cannot be estimated reliably from the available time series without allowing for count error. Proper allowance for the effect of count error is not thought to be feasible in this case. It is further shown, by means of simulations, that projections from population models which assume density independence lead to considerably larger uncertainty in the future population trajectory than was recognized in the cormorant study. This is because of failure to allow for the possibility of density-independent trends over time in population growth rate. Synthesis and applications. The study examined did not provide a sound basis for decisions about population management. Negative density dependence tends to reduce the impact upon population size of a given culling regime, and because its strength is usually overestimated by the methods used, culled populations will tend to decline more rapidly than expected and possibly continue to decline without stabilizing at a lower level.

Key words