A landscape index approach to evaluating the small mammal habitat quality of designed scenarios for agricultural watersheds.

A common goal of landscape planning and design is to improve ecological functions, yet planners and designers lack good tools to objectively judge the ecological consequences of alternatives. Landscape pattern indices could be efficient and useful tools for comparing the ecological consequences of alternative plans and designs if indices can estimate ecological functions. I hypothesized that landscape pattern indices could discriminate the habitat consequences of designed scenarios for two Corn Belt (Iowa) agricultural watersheds. Literature suggests that heterogeneous patterns of clustered good habitat and small landscape patches provide better habitat quality than homogeneous patterns that are less clustered and dominated by few landcover types in large landscape patches. I selected and applied a few landscape pattern indices to measure the heterogeneity, clustering, patch size, and proportion of habitat for small mammals in each landscape scenario. I tested the reliability and validity of landscape pattern indices and analyzed results by (a) comparing index values to published research in similar settings; (b) comparing the rank-order of index values to the intent of each designed landscape scenario; and (c) correlating the index values to an independent evaluation (a spatially-explicit population model) for small mammals. Results indicate that a few landscape indices (proportion, Simpson's evenness index, mean patch size, aggregation index) demonstrated limited utility for judging the habitat quality of alternative scenarios. Other indices, especially measures of landscape configuration, did not perform well, including interspersion and juxtaposition index, relative patch richness, contagion, and mean nearest neighbor distance. Landscape context and spatial data characteristics (like fragmentation and linear features, fine-scale resolution and incomplete data representation) dramatically influenced index values, particularly indices measuring landscape configuration. I report how spatial data representation, resolution, and classification affect indices. Findings suggest that indices are of limited utility for judging the ecological consequences of alternative plans and designs for highly fragmented landscapes. Planners and designers are advised to use caution when applying landscape pattern indices, especially those measuring landscape configuration, in highly fragmented contexts. Planners and designers should scrutinize spatial data for accuracy and select only those indices that have demonstrated at least some utility for judging specific ecological functions.PhDBiological SciencesCommunication and the ArtsEcologyLandscape architectureUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/123145/2/3068841.pd