A new application of change point analysis reveals extensive edge effects on a temperate mixed forest

Questions How do landscape changes along edges of protected areas affect forest interiors and stand development? What are the locations, spatial extents and magnitudes of these effects? Location The 8500-ha Sylvania Wilderness Area, Michigan's Upper Peninsula, USA. Methods We conducted vegetation surveys in 202 plots in ten transects crossing the Sylvania Wilderness border in 2013 and 2014. We recorded characteristics of forest structure, trees, shrubs, saplings, seedlings and herbaceous species. We constructed GLMM to estimate the location, spatial extent and magnitude of change of edge effects on Sylvania with a range of possible edge effect locations and widths of effect. We selected best-fit models that minimized the AIC and applied likelihood ratio tests to assess the statistical significance of each edge effect. Results Overall, evidence of edge effects occurred up to 625 m into the Sylvania Wilderness, with most significant changes occurring within 400 m of the wilderness border. Wide zones of change occurred across the wilderness border, while zones of change farther from the edge tended to be narrower, suggesting that distinct environments are established beyond the transitional habitats surrounding the border region. Canopy-level and understorey-level variables exhibited the largest magnitudes and steepest gradients of change, indicating these communities are strongly influenced by edge effects in this forest system. Canopy-level heterogeneity also increased approaching the internal core area of Sylvania. Conclusions In this case study, we applied a linear change point model and found a minimum buffer zone of 400 m to mitigate edge effects in an old-growth temperate mixed forest. Regionally, land managers could implement this buffer to existing edges of protected areas or negotiate this buffer zone in land acquisitions. A more stringent buffer zone of 625 m internal and 250 m external to old-growth forests would be ideal. This application of change point analysis provides a simple, efficient method to establish effective buffer zones and to identify functional groups or ecosystem attributes for which edge effects are of greatest conservation concern. We recommend modifying our open-source change point package to estimate local edge effects that take into account regional characteristics.