Reconciling forest demography with gradient analysis in montane landscapes

Seedling distributions are determined by the distribution of seeds and modified by environmental constraints such as the availability of light, moisture and nutrients. Trees have different dispersal strategies which includes different scales over which they disperse their seed. Trees also differ in their life history strategies and form a continuum between pioneer species and stress tolerators. Environmental constraints can form gradient complexes and mosaics of variable habitat suitability.; I examined seedling spatial patterns in montane landscapes to ascertain how the spatial distribution of tree abundance and environment impact the development of species-environment relations. This study was conducted in parallel at the Coweeta Hydrologic Laboratory and Sequoia National Park. Seedlings were examined at two scales. Ordinations and partial Mantel tests were used to analyze seedling-tree, seedling-environment and tree-environment relations. Demographic analysis was incorporated at the landscape-scale by considering trees as relative seed source before considering seedling-environment correlations.; The spatial patterns of seedlings and trees differed across scales and across study sites. At the stand-scale in Coweeta, many species displayed seedling-tree correlations and were correlated with environmental variables. At Sequoia only responses to canopy openings were observed.; At the landscape-scale, Sequoian species exhibited a high degree of spatial sorting. All species exhibited tree-seedling correlations. In ordinations seedlings and trees of the same species occurred in close proximity in eight out of nine cases. In Coweeta, 14 out of 28 species had seedling-tree correlations and many seedling-tree pairs in ordinations were well separated.; The physical templates of the sites were compared using variograms of environment and species Mantel correlograms. Seedlings and trees had similar spatial patterns in Sequoia. In Coweeta, trees had a sinusoidal correlogram while seedlings displayed decreasing similarity with distance. This indicated seedling patterns may be responding to spatial variation in soil chemistry parameters and slope-aspect relationships.; A model of field-sampling for patterns resulting from multiple agents was used to examine differences in spatial extent, patterns of constraint, the effects of different dispersal kernels and the effect of tree spatial patterns. The detection rates for tree-seedling and seedling-environment correlations changed with the spatial extent of sampling and with tree arrangement.