We have provided evidence that supports our mechanistic explanation of linkage in forest communities. Linkage among forest strata appears to arise from similarities among strata of forest vegetation in the responses of their respective species to environmental gradients. In central Appalachian hardwood stands of West Virginia, these responses may change through secondary forest succession and thus may be a function of stand age. Early in succession, spatial variation in species composition of the overstory appears largely related to the density of the stand, whereas variation in herb composition is related more to soil fertility. This changes later in succession when variation in herb composition responds more to stand structure (i.e., tree density), a time when overstory variability is also related to density. In successional aspen forests of northern lower Michigan, where correlations between strata were significant for both young and mature stands, these responses may be much less related to stand age. The degree to which linkage between strata is the result of higher beta diversity in the Michigan sample is not clear. We believe, however, that it is not an artifact of simply sampling across two extremes of environmental conditions (see McCune and Antos 1981), given the continuum of points along axis 1 for both vegetation strata. as depicted in figure 8.2. Although the concept of linkage among vegetation strata of forest communities likely will continue to be debated among vegetation scientists, we believe that it is a concept with a high degree of importance and application. It furthers our understanding of and appreciation for the complexities underlying the structure and function of forest ecosystems (e.g.. responses to disturbance [chapters 11 and 13. this volume] and mechanisms of secondary succession [chapter 9, this volume]). It may also be applied toward landscape level investigations of forest cover types and remote sensing. The generality and utility of this explanation, however, will only be seen after it is tested in a more formal and structured way. We invite futher testing of this mechanism as a predictive hypothesis in other forest types and over a wider range of stand ages. Studies that examine linkage across different breadths of beta diversity and spatial scales would be particularly useful. Given the widespread popularity and use of CCA among vegetation scientists (Barbour et al. 1999), this might be done with reasonable ease.