The Last Glacial Maximum distribution of South African subtropical thicket inferred from community distribution modelling
Article first published online: 2 NOV 2012
© 2012 Blackwell Publishing Ltd
Journal of Biogeography
Volume 40, Issue 2, pages 310–322, February 2013
How to Cite
Potts, A. J., Hedderson, T. A., Franklin, J., Cowling, R. M. (2013), The Last Glacial Maximum distribution of South African subtropical thicket inferred from community distribution modelling. Journal of Biogeography, 40: 310–322. doi: 10.1111/j.1365-2699.2012.02788.x
- Issue published online: 18 JAN 2013
- Article first published online: 2 NOV 2012
- South African National Research Foundation (NRF)
- University of Cape Town (UCT)
- Nelson Mandela Metropolitan University and the US National Science Foundation. Grant Number: 1138073
- Climate envelope modelling;
- ensemble modelling;
- Last Glacial Maximum;
- Pleistocene refugia;
- range shifts;
- South Africa;
- subtropical thicket;
- vegetation modelling
Southern Africa lacks the palaeo-archives needed to infer regional vegetation history. The megadiverse subtropical thicket of South Africa, locally termed the Albany subtropical thicket (AST), is postulated to have experienced severe reductions in range during the Pleistocene glacial periods. Here we test this glacial contraction hypothesis using spatial predictions from climate envelope models to explore the distributions of the AST vegetation subtypes at the Last Glacial Maximum (LGM).
We used an ensemble approach to assess the variability of predictions due to the many methodological decisions made during the modelling process. Five sources of uncertainty were included: initial locality data sets, selected climate parameters, statistical methods, LGM global climate models and threshold criteria.
The last three sources of uncertainty contributed significantly to the variation in projected LGM distributions. Nonetheless, estimates of gain, loss and turnover in range between present and past scenarios indicate dramatic range reduction and turnover during the LGM for all AST subtypes. Arid and valley thickets experienced the most severe reductions with an overall decline in elevation and concomitant fragmentation, whereas mesic thicket had a reduced but fairly continuous range.
Our results suggest that the Pleistocene glacial periods saw dramatically reduced distributions of AST subtypes, supporting the glacial contraction hypothesis. Potential LGM refugia were identified and extreme bottlenecks were predicted for species from all AST subtypes during the LGM. Watersheds were also more likely to be strong barriers to gene flow during the LGM due to the retraction of thicket subtypes into basin valleys.