Prioritizing land for oak savanna conservation using species distribution models (SDMs) and two umbrella species: Wild lupine (Lupinus perennis) and the Karner Blue butterfly (Lycaeides melissa samuelis)

Background/Question/Methods

Oak savannas are rare ecosystems that were once relatively abundant in central North America.  In recent years there has been a push to conserve or restore as much oak savanna as possible.  Because resources are often limited in such efforts, it is important for land managers and conservationists to focus their efforts.  One way to do this is by targeting 'umbrella' species for conservation, with the idea that their conservation will benefit a suite of different organisms.  The Karner Blue butterfly (KBB; Lycaeides melissa samuelis) and its larval host plant, wild lupine (Lupinus perennis), are often used as umbrella species for oak savannas in the North Midwestern U.S.  In this study, our goals were to: (1) evaluate the habitat requirements of the KBB and wild lupine, and (2) map potential habitat to prioritize land for oak savanna conservation or restoration. 

In 2012, we collected data from 485 plots in 36 sites to evaluate the habitat requirements of wild lupine.  We used four years (2008-2012) of butterfly survey data across 70 sites to evaluate the habitat requirements of the KBB.  Environmental variables found to be important for each species were then considered in the development of spatially-explicit species distribution models (SDMs).  The SDMs were constructed using datasets related to vegetation, soil, and physiography combined with species location data from the field.  Data were standardized to a spatial resolution of 30m².  SDMs were constructed using boosted regression trees, and tested by randomly assigning the data to either a training or validation dataset (70%:30%).

Results/Conclusions

The probability of wild lupine presence was driven primarily by soil taxonomic subgroup and dominant vegetation type.  Field data indicated that leaf litter type and depth are both important predictors of lupine presence, which may help explain the biological importance of vegetation type.  The primary driver of KBB presence was dominant vegetation, followed by elevation and distance to the next occupied savanna site.  The importance of other nearby savannas suggests metapopulation dynamics may be important in determining whether KBB are present at a site.  Both the receiver operating characteristic area under curve (AUC) and kappa statistics indicated that the SDMs performed reasonably well on the validation datasets.  Additional lab- and fieldwork will be conducted in 2013 to determine the importance of soil chemistry to lupine, and to measure microclimatic variation and it’s relation to both lupine and the KBB. These results will be used to further refine the SDMs.