Seeking the flowers for the bees: Integrating biotic interactions into niche models to assess the distribution of the exotic bee species Lithurgus huberi in South America

• Published in: Ecological modelling 2014-02, Vol.273, p.200-209
• Main Authors: Silva, Daniel P., Gonzalez, Victor H., Melo, Gabriel A.R., Lucia, Mariano, Alvarez, Leopoldo J., De Marco, Paulo
• Format: Article
• Language: English
• Subjects: Algorithms; Americas; Animal, plant and microbial ecology; Apidae; Biological and medical sciences; Climate; Convolvulaceae; Exotic species; Flowers; Fundamental and applied biological sciences. Psychology; General aspects. Techniques; Lithurgini; Mathematical models; Megachilidae; Methods and techniques (sampling, tagging, trapping, modelling...); Plants (organisms); Pollen; Pollination; Species distribution models; Species interactions
• ISSN: 0304-3800; 1872-7026
• DOI: 10.1016/j.ecolmodel.2013.11.016

•Distributions with and without the host plants were not different.•SVM had low performance when we considered all host plants distributions.•MaxEnt and SVM performed similarly while predicting the new L. huberi occurrences.•Northeastern, eastern and southeastern South America are indicated for surveys. The wood-boring bee Lithurgus huberi Ducke (Apidae: Megachilinae: Lithurgini) is arguably an exotic species to South America. This solitary bee is the only representative in the Western Hemisphere of the Old World genus Lithurgus, and likely a conspecific with the Indo-Australian species Lithurgus atratus. L. huberi appears to have reached the continent at least 100 years ago, when it was discovered and described. Because this species seems to be oligolectic on pollen of Convolvulaceae flowers in South America, we attempted to integrate this biotic interaction (plant–bee relationships) to our species distribution model (SDM) procedures to predict its potential distribution in South America. The modeled distribution of seven L. huberi's host plant species did not improve the algorithms’ ability to predict its distribution, but it produced constrained ranges. These results suggest that our biotic variables are not independent of the abiotic variables used (mostly related to climate). We employed five modeling algorithms, Envelope Score, GARP, Mahalanobis Distance, Support Vector Machines, and MaxEnt, but only the former two showed a good performance when predicting the occurrence of both, the host plant species and L. huberi. Our results indicate that this exotic pollinator is mainly distributed in eastern, northeastern, central, and southwestern South America, with few suitable areas in the Amazon region. We also highlight suitable areas for future surveys and present new occurrence records.