Nested species-area data sets reveal “breaks” in slope which may indicate scale of heterogeneity perceived by plant communities

Background/Question/Methods

Community ecologists estimating the degree to which different factors structure ecological communities find considerable scale dependence in results.  A typical approach tests the correspondence of species composition among sites with changes in habitat and the underlying spatial configuration of those sites.  Such studies rarely have an independent estimate of the appropriate grain of sampling that should be used; this will depend on the scale of underlying habitat heterogeneity. In the context of a study of beta diversity in Great Lakes shoreline plant communities, we used nested species-area plots to estimate the plant-relevant scale of habitat heterogeneity in order to determine the appropriate minimum size and spacing of plots.  At 20 sites on Beaver Island, in northern Lake Michigan, we recorded the number of species in a 10 x 10 cm plot, noting additional species encountered in an extended 20 x 20 cm plot sharing the same starting corner, then a 40 x 40 cm plot, etc., until we had accumulated species for a plot that measured 25.6 m on a side. We plotted a semi-log species-area relationship and looked for the scale at which a change in slope or a stair-step break in the pattern indicated the addition of new species responding to qualitatively different environmental factors.

Results/Conclusions

Using a general linear model that included a breakpoint parameter, we found that there were substantially more species than expected in the ~40 m² plot (6.4 x 6.4 m) compared to the next smaller plot.  An additional series of nested plots (n=314) had been sampled in geographically dispersed but otherwise similar shoreline habitats in the Great Lakes. Linear modeling showed little evidence for a best choice breakpoint parameter in that dataset, although the model R² was highest with a breakpoint between 1 and 2 m². The largest plot in this dataset was only 8 m²—smaller than the break point scale in our island dataset.  We conclude that from a plant’s eye view of this shoreline plant community, the character of the environment changes qualitatively at a scale of about 40 m².  To adequately sample diversity in this system, therefore, we will need to have multiple plots smaller than 40 m² that can be adequately replicated within that area. We suggest the implications of such methodological considerations for other ecological communities.