Limitations of population models in predicting climate change effects: a simulation study of sociable weavers in southern Africa


  • Monika Schwager,

  • Rita Covas,

  • Niels Blaum,

  • Florian Jeltsch

M. Schwager (, Dept of Ecology, Inst. for Zoology, Univ. of Mainz, Becherweg 13, DE–55099 Mainz, Germany. – MS, N. Blaum and F. Jeltsch, Dept of Plant Ecology and Nature Conservation, Inst. for Biochemistry and Biology, Univ. of Potsdam, Maulbeerallee 2, DE–14469 Potsdam, Germany. – R. Covas, Percy FitzPatrick Institute (DST NRF Centre of Excellence), Univ. of Cape Town, Rondebosch 7701, South Africa. Present address for RC: CEFE-CNRS, 1919 Rte de Mende, FR–34293 Montpellier, France.


Current approaches for predicting climate change effects on populations comprise static models based on the geographical distribution of species, and dynamic population models based on the relationship between population processes and the recent variation in climate. Population models have the inherent advantage of considering a species’ response to climate as resulting from distinct mechanisms. However, they may have the disadvantage of considering only short-term processes as they occur under the current climate, disregarding slowly adapting mechanisms. It would be important, however, to know whether slowly adapting processes occur, and whether they will respond to climate change. A way of testing for a discrepancy between short-term and long-term mechanisms in the response to climate is contrasting a population model with the current distribution and abundance in different climates. We demonstrate this idea for the sociable weaver Philetairussocius, a passerine bird of semiarid southern Africa. We develop a dynamic population model that predicts the species’ response to mean annual precipitation (MAP) and tests whether the model can reproduce the occurrence of sociable weavers in their current range of MAP as well as realistic abundances in two study locations with different MAP. The model predicts a decreased abundance under a scenario of decreased MAP and it can reproduce realistic populations in a single location. However it cannot explain the occurrence of sociable weavers across their entire distribution nor does it produce realistic abundances in the two areas when they were tested simultaneously. The results imply that the sensitive short-term response to MAP is buffered by long-term processes such as adaptation or plasticity in life history, shifts in interspecific interactions or changes in habitat structure. We suggest that this result is common for animals in semi-arid and arid regions and discuss implications for static and dynamic modelling approaches in climate change research.