Factors affecting the distribution of two species of African Charaxes: the role of hostplant distribution, environmental covariates and biogeography

2014-08-25T14:23:00Z (GMT) by José R. Ferrer-Paris
<p>The genus Charaxes Ochsenheimer, 1816 is known for its high species richness and the diversity of hostplant associations. Here, I analyse the role of hostplant distribution, environmental variables and biogeographic regions in limiting the distribution of two african species of Charaxes: Charaxes cithaeron (C. & R. Felder, 1859) and Ch. xiphares (Stoll, 1781). I use species distribution records from available maps, museums and collections, and environmental variables from worldwide climatic datasets, and apply a logistic regression based on presence records and available habitat represented by localities with records for other species of Charaxes and fit models with several environmental variables (abiotic component, A), potential distribution of known hostplants (biotic component, B) and biogeographic region (geographic component, G). We then evaluate the models according to AIC and calculate AIC-weights to contrast the different hypothesis. For Ch. cithaeron the model with biogeographic region and environmental variables (A+G) was clearly better than alternative models, with AIC weights of 93.9%, but for Ch. xiphares the full model (A+B+G) with hostplant distribution as a covariate had lower AIC and AIC weights of 53.9%, but the simpler model without hostplant distribution (A+G) was still a good candidate with 46.1% of the weights. These results suggest that for some species the distribution is positively correlated with hostplant distribution, and adding this covariate might improve model performance. This might be the consequence of different environmental variables influencing the biology of the hostplants and butterflies. Several gaps in distribution data for butterfly and their hostplants are evident, but resulting predictions are informative and agree with patterns suggested in the literature. Extensive testing of distribution model requires more standardized sampling effort in regions with poor representation in collections.</p> <p>This poster was presented at the <strong><em>7th International Conference on the Biology of Butterflies</em></strong>, in Turku, Finland, 11-14 August 2014.</p> <p>This work was supported by a <strong><em>2013 EOL Rubenstein Research Fellowship Award</em></strong>.</p> <p> </p>