This map shows modeled habitat suitability for Bombus franklini across the Pacific Northwest and northern California, based on occurrence records prior to the species’ last confirmed sighting in 2006. The model integrates climate data from 1970–2000 and 1992 land cover data from the National Land Cover Database (NLCD).
Habitat suitability is displayed as a color gradient from light yellow (low suitability) to dark purple (high suitability), with county boundaries shown for spatial reference. The model predicts the highest suitability in the Klamath-Siskiyou region, particularly around Mt. Ashland in southwestern Oregon. This area is especially significant as it represents the last known location of Bombus franklini and has been the focus of intensive survey efforts in recent years.
These two habitat suitability maps for Bombus franklini show remarkably consistent patterns between historical and current predictions. Both maps identify the same core region of highest suitability (dark purple to black) in the Klamath-Siskiyou area of northern California and southern Oregon, with moderate suitability extending along the Cascade and Coast Ranges.
The overall spatial distribution remains largely unchanged, suggesting stable environmental conditions in the species' preferred mountainous habitat. However, subtle differences are apparent in the extent and intensity of the highest suitability areas, with the historical prediction appearing to show slightly more extensive core habitat. The areas of low to moderate suitability (yellow to orange zones) display consistent boundaries in both scenarios, indicating that the fundamental environmental drivers of Bombus franklini habitat requirements have remained relatively stable over time, though minor reductions in optimal habitat extent may be ecologically significant for this rare species.
Effect of Multicollinearity Filtering on Predicted Habitat Suitability for Bombus franklini
These two maps use the same modeling methods as described above, but with land cover data removed. In the first map (left), we applied the same VIF (Variance Inflation Factor) filtering as before, so only a reduced set of climate variables was used. The second map (right) uses all available climate variables without any VIF filtering or land cover data. This comparison highlights the variability in species distribution model (SDM) results when multicollinearity is not addressed through VIF filtering.
Effect of Land Cover Inclusion on Predicted Habitat Suitability
for Bombus franklini
These two maps demonstrate the difference between land cover and no land cover comparisons for franklins bubmlebee. Although land cover is often considered an important predictor in species distribution models (Franklin, 2010), the comparison between models with and without land cover for Bombus franklini showed minimal differences in output. Both the Schoener’s D index (0.9) and Pearson correlation (0.982) suggest that the spatial predictions are nearly identical. This may reflect a strong climate-driven niche for B. franklini, or that the land cover data used did not add unique information beyond what the climate variables already captured.
The United States Fish and Wildlife Service (Service) shall not be held liable for improper or incorrect use of the data described and/or contained herein. While the Service makes every reasonable effort to ensure the accuracy and completeness of data provided for distribution, it may not have the necessary accuracy or completeness required for every possible intended use. The Service recommends that data users consult the associated metadata record to understand the quality and possible limitations of the data. The Service creates metadata records in accordance with the standards endorsed by the Federal Geographic Data Committee.
As a result of the above considerations, the Service gives no warranty, expressed or implied, as to the accuracy, reliability, or completeness of the data. It is the responsibility of the data user to use the data in a manner consistent with the limitations of geospatial data in general and these data in particular. Although these data have been processed successfully on a computer system at the Service, no warranty, expressed or implied, is made regarding the utility of the data on another system or for general or scientific purposes, nor shall the act of distribution constitute any such warranty. This applies to the use of the data both alone and in aggregate with other data and information.