Biological control of the non-native, invasive saltcedar (Tamarix spp.) in southwestern states is controversial regarding the protection of the federally endangered southwestern willow flycatcher (Empidonax traillii extimus). The songbird nests primarily in native riparian vegetation, but it sometimes occupies saltcedar where floodplain-level invasion replaces native habitat. We could not find any detailed habitat-suitability modeling studies of saltcedar and flycatcher in the same landscape. Biological control of Tamarix spp. with the saltcedar leaf beetle (Diorhabda elongate) began along the Virgin River near St. George, Utah, in 2006. In subsequent years, area impacted and intensity of defoliation events have increased, enhancing the need for a comprehensive understanding of the saltcedar-flycatcher relationship. We used maximum entropy (Maxent) modeling to separately quantify the current invasion of Tamarix spp. and the potential habitat available for E. traillii extimus within the studied watersheds. The models used transformations of 2008 Landsat Thematic Mapper images from eight months of the growing season along with a digital elevation model to identify floodplain habitat. Transformations of Landsat images included tasseled cap, normalized difference vegetation index, and neighborhood statistics. Classification of thresholds and comparison of the two Maxent outputs revealed discrepancies between saltcedar invasion and suitable habitat for E. traillii extimus.
Results/Conclusions
Maxent models performed quite well for the flycatcher and saltcedar with Area Under the ROC Curve (AUC) values of 0.960 and 0.932, respectively. Comparison of multiple threshold classifications indicated that suitable habitat for E. traillii extimus and current invasion of Tamarix spp. were not equivalent. Dense saltcedar invasions (occupying >50% of habitat) comprised an average of 234 km2, of which 10% was also modeled as potential habitat for E. traillii extimus. Results of modeling potential habitat for E. traillii extimus revealed a similar trend. Potential habitat constituted an average of 180 km2, of which 13.1% was modeled as also containing dense saltcedar invasion. Results indicated that native vegetation exists along the riparian corridor in the modeled watersheds, and these areas contain the specific characteristics required by E. traillii extimus for breeding. In addition, models indicated that large saltcedar invasions exist within the study watershed, but these infestations may not contain characteristics specific enough to satisfy the habitat requirements of E. traillii extimus. Based on this study, effective biological control of Tamarix spp. should not considerably reduce the suitable habitat available to E. traillii extimus within the study area and alternatively may increase suitable habitat if native vegetation replaces bio-controlled areas.
