COS 42-3
Effect of connectivity and habitat loss in remnant prairie plant communities

Tuesday, August 6, 2013: 2:10 PM
L100F, Minneapolis Convention Center
Amy O. Alstad, Zoology, University of Wisconsin - Madison, Madison, WI
Ellen I. Damschen, Zoology, University of Wisconsin-Madison, Madison, WI
John A. Harrington, Landscape Architecture, University of Wisconsin Madison, Madison, WI
Background/Question/Methods

Biotic communities increasingly occupy small, fragmented patches of habitat surrounded by a human-modified matrix. Understanding the impacts of this landscape modification on biodiversity has become an important research focus. However, despite the proliferation of studies on the impacts of fragmentation on biodiversity, the results remain unclear, with reports of positive, negative and no significant effects of fragmentation from a wide variety of systems. One explanation for these conflicting results is that the impacts of fragmentation can be slow to manifest, especially in perennial plant systems. Species loss may lag well behind disruption of ecosystem structure. The limitations associated with detecting lagged impacts of fragmentation illustrate the value of long-term datasets. Here, we present an analysis on a long-term data set investigating the relative impacts of habitat loss and fragmentation on prairie remnants in Wisconsin. Sites first visited in 1950 were resampled in 2012. Replicating the methodology of the original researchers, we created a list of all plant species present. Patch size, isolation and landscape context were characterized for both the historic (1950) and present day landscapes using historical aerial photos and current satellite imagery. Past and present landscape response variables were related via linear regression to plant community composition.

Results/Conclusions

Forty-seven prairies were resampled during the 2012 season, 62 years after the first sampling. Mean patch size decreased from 1.6 ha in 1950 to 0.6 ha in 2012. Mean patch proximity, a unit-less area-weighted measure of connectivity, decreased from 57.2 in 1950 to 1.8 in 2013. Spearman’s rank correlation indicates that mean coefficient of conservatism, a measure of species’s fidelity to a particular habitat, was significantly related to current patch area (rho=0.56, p<0.0001), but not historical landscape variables or current patch connectivity. Forward model selection using AIC resulted in a model in which mean coefficient of conservatism was best explained by current patch area alone. Historic patch connectivity, historic patch area, and current patch connectivity were not included in the model as significant predictors. Together, these results suggest that there are no lagged effects operating in this system, and that plant communities respond relatively quickly to habitat loss and fragmentation. Furthermore, the analyses indicate that current patch area is more important in determining the character of the plant community than is current patch isolation. These results corroborate a large number of studies that report that habitat loss poses a more serious threat to biodiversity than does fragmentation per se.