Accounting for regional niche variation in habitat suitability models
Article first published online: 4 OCT 2006
Volume 116, Issue 1, pages 99–110, January 2007
How to Cite
Murphy, H. T. and Lovett-Doust, J. (2007), Accounting for regional niche variation in habitat suitability models. Oikos, 116: 99–110. doi: 10.1111/j.2006.0030-1299.15050.x
- Issue published online: 4 OCT 2006
- Article first published online: 4 OCT 2006
- Manuscript Accepted 1 September 2006
Ecological niche modeling has become an increasingly important tool to address issues in many fields of basic and applied ecology. The ecological niche space occupied across the geographic range, particularly for wide-ranging species, may vary for a variety of evolutionary and non-evolutionary reasons. However, ecological niche models are often applied over large geographic areas without regard for the potential effects of regional variation in adaptation, environmental conditions and their interactions, and species responses, thus significantly reducing their accuracy and utility. We develop regionally partitioned ecological niche models, using GARP, for the wide-ranging North American tree Gleditsia triacanthos (Fabaceae). Models were constructed based on known tree occurrences at peripheral and range-centre locations, as well as across the geographic range as a whole.
Our results suggest that the niche space occupied by G. triacanthos varies regionally and that between some regions in particular there may be a complete absence of niche overlap. In particular, while there is some overlap between the niche space occupied by trees in the western and central regions of the range, there appears to be virtually no overlap in the niche space occupied by trees in the south of the range and that occupied by western and central trees. This lack of overlap appears to be driven primarily by regional differences in abiotic conditions, rather than regional adaptation per se. The results of our study have several important implications for the future development of habitat suitability models over large geographic areas. Spatial partitioning of data is clearly necessary to improve predictions of models where regional niche variation occurs. For wide-ranging species in particular, regional differences in ecological characteristics may cause apparent niche variation.