We evaluate the utility of univariate and multivariate analytical methods for comparing the composition between subgroups (two age classes) within a single community. Data were collected from twenty-six 1000 m² plots located across three distinct mountain ranges and geologies in the southern Appalachians of North Carolina. We inventoried vegetation using the Carolina Vegetation Survey protocol, assessing species presence at 0.01, 0.1, 1, 10, 100, and 1000 m², and percentage cover at 100 m².

Results/Conclusions

We found no difference between age classes using aggregate univariate methods, such as species diversity measures and species-area relationships. Similarly, multivariate methods (Nonmetric Multidimensional Scaling [NMS] and Multi-Response Permutation Procedures [MRPP]) revealed no age class differences (MRPP test: p = 0.85). However, multiple comparison procedures (MCPs) identified 13 of 79 species with either lower abundance or incidence of occurrence in second growth rich coves. The relative importance of Type I (erroneously reject the null hypothesis) and Type II (erroneously fail to reject the null hypothesis) statistical errors played a key role in comparing multivariate and univariate methods. We concluded that: (1) aggregate univariate methods facilitate comparison of unrelated communities, but have unnecessarily high Type II errors when comparing closely related communities, (2) multivariate methods support comparison of related communities but can mask significant differences among individual species, and (3) univariate MCPs readily detect individual species changes within a community, at the expense of high Type I errors.  Thus, the selection of analytical methods should consider the similarity of the communities and the need for evaluating communities versus species.