Using unclassified remote-sensing imagery to produce species distribution models for addressing management-relevant challenges associated with landscape change

We have developed and are applying an alternative and cost-effective approach that uses bird monitoring data to evaluate landscape condition and change.  Using large datasets made available through the Avian Knowledge Network we assess the influence of land cover in species distribution modelling, addressing the fact that such assessments have thus far been limited by the availability of fine-resolution land-cover data appropriate for most species responses. We demonstrate that unclassified remote-sensing imagery can be used to produce species distribution models with high prediction success, without using the typical land cover data now available (e.g., gap), which are course-scaled and provide little or no information about habitat condition.  We then use this modeling approach with oak woodland species to estimate local- and landscape-scale habitat amounts across five HUC 8 watersheds. We conclude that raw Landsat Thematic Mapper data (made freely available every year) will be particularly useful in species distribution models when high-resolution predictions are required (e.g., habitat change detection studies, identification of fine-scale biodiversity hotspots, and reserve design).  We are now looking to apply this approach to several management relevant issues, including: evaluation of landscape conditions before and after natural and anthropogenic disturbances such as fire and timber harvest, setting objectives for and evaluating effectiveness of accelerated forest restoration intended to increase successional rates over years and decades, evaluating the distribution of early and mid-seral habitats, the lack of which (resulting from historic and current forest and fire management practices) is associated with landbird population declines in the west.