The role of spatial pattern in controlling forest dynamics.

摘要

Stand structure is a key factor in the growth, function, and disturbance regimes of forests. Many conventional descriptions of stand structure focus on stand-level collective attributes such as average tree size and density. Spatially explicit mapping and modeling of trees allow forests to be seen as compositions of individuals, with unique spatial arrangements and patterns. Analysis of these patterns provides insight into the interactions between trees, at the scales of neighboring individuals to neighboring clumps, which create structural heterogeneity and drive stand development. My dissertation explores the spatial and temporal relationships between resource availability, stand structure, and tree processes, in a natural stand of ponderosa pine in Colorado and a mixed-species plantation in Hawaii.; At coarse spatial scales, we found that patterns of ponderosa pine trees strongly influenced processes of mortality and regeneration. In turn, processes of mortality and regeneration influenced spatial patterning---creating strong clumping in young trees, regular spacing in old trees, and strong segregation of different size classes. Long term patterns in climate controlled the temporal dynamics of these pattern and process relationships.; At finer spatial scales we saw that tree pattern had enormous implications for tree growth and survival. The size and spacing of neighbors explained from 69 to 85% of the variability in individual tree performance in both the tropical plantation and the old-growth conifer forest. Soil nutrient supply significantly modified both the nature and strength of neighbor interactions in both of these systems. In Hawaii, we saw that the effect of a Eucalyptus or Falcataria neighbor could be either competitive or facilitative depending on the supply of soil nitrogen and phosphorus. In Colorado, we saw that the strength of competitive interactions, and the importance of soil nutrient supply, are highly dependent upon the scale of the neighborhood analysis.; This dissertation underscores the need to study forest dynamics at multiple spatial and temporal scales, and along environmental gradients in soil resource supplies. Neighborhood models and spatially explicit methods revealed much greater complexity of stand structure and tree interactions would have been seen with traditional tools.