We used 6000 specimen records to examine % species loss at 30 sampled fixed sites across Illinois. A total 193 USEPA StreamCat variables were reduced to an 11 variable data set through correlation, relative weight, and dominance analysis in R software. The MuMin package with dredge function and generalized linear models was used to examine all possible variable combinations, each model compared using Akaike information criterion (AIC). Models with ?AIC?2 were deemed "good" models. Importance in the MuMin package was used to assign variable relative importance. The % loss was high across sites. The best model contained % catchment as forest, catchment mean depth to bedrock, watershed mean population density, and catchment % barren land as variables. "Good" models contained % catchment as forest and catchment mean depth to bedrock. Variables above 0.4 importance were: deciduous forest, depth to bedrock, population density, barren land, % impervious cover in catchment, and catchment area. Greater % forest cover, shallower soils, lower population densities, less barren land, and smaller catchment areas led to lower % loss of stoneflies. Our study is the first to link loss of historical stonefly diversity to various factors.
|Original language||English (US)|
|Title of host publication||Society for Freshwater Science 2017 Annual Meeting, Raleigh, North Carolina|
|State||Published - 2017|