| Gradient analysis as a study of community pattern has a long and distinguished history in montane systems. The elevation complex stands out from these studies as the most consistent environmental correlate with changes in forest community composition. As a result, major vegetation zones typically are described along elevation gradients. Trees are not affected by elevation, however, but rather by variables such as temperature and precipitation that covary with elevation. Because these variables are difficult to measure at large spatial scales, I propose an iterative approach of leveraging traditional gradient analysis studies to (1) develop new hypotheses regarding the dynamics of forest communities, (2) identify major data needs and shortcomings, and (3) target locations on the landscape that are best suited to test new hypotheses and fill data gaps.; I developed a working model of community pattern from a landscape sample of old-growth forest stands in the Lookout Creek Watershed of the H. J. Andrews Experimental Forest, Oregon. Analytic techniques in classification, ordination, and spatial regression were used in synthesis to identify major forest communities and the terrain and soil factors influencing their distribution. Elevation and spatial location were highlighted as the strongest explanatory variables. These results, though typical of a gradient analysis study, allow for considerable variability in the interpretation of the underlying mechanisms. To make educated predictions of how these forests may respond to environmental change, such as that predicted under greenhouse warming scenarios, requires a more detailed description of the interactions between forest communities and their environment.; To refine the working model, I developed a series of novel sampling and analytic approaches to study fine-grain environmental patterns over large geographic areas. These methods include: (1) new approaches to empirically interpolate relative differences in temperature, radiation and soil moisture across landscapes; and (2) a replicated study of plant demographics (growth, mortality and regeneration) at the dominant community ecotone.; These studies were intended to replace elevation and basal area from the working model with more plant-relevant explanatory variables and the demographic components that they affect. The landscape-scale models illustrate that using elevation to approximate environmental variability ignores the multi-scale structure of the physical template. Similarly, the focused study of the Tsuga heterophylla-Abies amabilis ecotone illustrates that a coarse-scale analysis of community distributions might not accurately reflect the dynamics within active areas of community transition. |
