Assessment of biodiversity in alternative future landscapes of the Willamette Valley, Oregon

Background/Question/Methods

Metrics based on change in habitat area are often used to assess potential impacts of landscape change on native species and to identify gaps in conservation protection. Here, we asked: How robust are metrics based on change in habitat area for evaluating the potential impact of restoration or habitat loss on species diversity? We compared metrics based on change in habitat area with population viability analysis (PVA) for four species selected from groups defined by their species-habitat associations. Multivariate statistical analyses of the species-habitat association matrix for non-fish vertebrates in the Willamette River Basin were used to group over 250 species by habitat affinity, resulting in identification of five major species groups. A decision-support tool was used to generate six future landscapes for a sub-watershed in the basin that represented three different sets of user-defined priorities, with two levels of restoration (10% and 25% of watershed area). The percent change in habitat area was computed for each species in each habitat weighted by habitat suitability for the species for the restoration landscapes. This metric was compared to results of PVA for four species selected to represent major habitat associations (grassland, conifer forest, hardwood forest, and wetlands).

Results/Conclusions

Landscapes in which restoration targeted achievement of multiple goals (water quality improvement, habitat enhancement for multiple species groups) were characterized by increased area of wetlands and shrub cover, whereas landscapes that focused on habitat restoration only (or habitat restoration for riparian-associated species only) had lower levels of wetland restoration, more shrubland, natural grassland, and oak savanna. One advantage of PVA over metrics comparing change in habitat area relative to the present was immediately apparent; the ability of PVA to illustrate the potential effects of lack of certain habitat types in the present (for example wetlands, which occupy less than 1% of the existing landscape). Model runs (average of 50 runs for a 200-year time period) conducted using the PATCH model for the Marsh Wren (a wetland associated species) indicated high potential for extirpation of this species in the existing landscape, as well as in those future landscapes with low levels of wetland and shrubland restoration. The PVA results for species associated with habitat types that were more abundant in the watershed (such as forest and grassland) tended to be proportionate to the total area restored, and were more similar to the results from the habitat area metric.