Missing in action: Species competition is a neglected predictor variable in species distribution modelling

Kudzai Shaun Mpakairi, Henry Ndaimani, Paradzayi Tagwireyi*, Tawanda Winmore Gara, Mark Zvidzai, Daphine Madhlamoto

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review

20 Citations (Scopus)
122 Downloads (Pure)


The central role of species competition in shaping community structure in ecosystems is well appreciated amongst ecologists. However species competition is a consistently missing variable in Species Distribution Modelling (SDM). This study presents results of our attempt to incorporate species competition in SDMs. We used a suit of predictor variables including Soil Adjusted Vegetation Index (SAVI), as well as distance from roads, settlements and water, fire frequency and distance from the nearest herbivore sighting (of selected herbivores) to model individual habitat preferences of five grazer species (buffalo, warthog, waterbuck, wildebeest and zebra) with the Ensemble SDM algorithm for Gonarezhou National Park, Zimbabwe. Our results showed that distance from the nearest animal sighting (a proxy for competition among grazers) was the best predictor of the potential distribution of buffalo, wildebeest and zebra but the second best predictor for warthog and waterbuck. Our findings provide evidence to that competition is an important predictor of grazer species’ potential distribution. These findings suggest that species distribution modelling that neglects species competition may be inadequate in explaining the potential distribution of species. Therefore our findings encourage the inclusion of competition in SDM as well as potentially igniting discussions that may lead to improving the predictive power of future SDM efforts.

Original languageEnglish
Article numbere0181088
JournalPLoS ONE
Issue number7
Publication statusPublished - 1 Jul 2017


Dive into the research topics of 'Missing in action: Species competition is a neglected predictor variable in species distribution modelling'. Together they form a unique fingerprint.

Cite this