Abstract:
This paper describes methods that can be used to
evaluate stand and tree growth response to a single
application of fertilization and/or thinning with data
collected from multiple installations. Two kinds of
methods were proposed: (1) structure analysis which
applies covariance analysis in a blocked design with and
without sampling units, and (2) multi-step analysis which
first fits a control model to control data and then uses
it to evaluate treatment response. Ideally, the former
method is preferred to the latter method for evaluating
the treatment response. However, when the experimental
data are large in sampling size and/or complex in their
designed structure, structure analysis often can not be
performed on most statistical packages, and, therefore,
the multi-step analysis is a viable alternative.
The methods were applied in modeling fertilization
and thinning effects on Douglas-fir [Pseudotsuga
menziesii (Mirb.) Franco] dominant height growth and
diameter growth of single trees. Both responses due to
improved nutrition from a single application of nitrogen
fertilizer were significant (a=O.05) in the first 5-year
growth period, and not significant after two 5-year
cycles. The response in diameter growth was stronger
than that in dominant height growth. Thinning neither
increased nor decreased the dominant height growth. It
accelerated diameter growth, but the interaction of
fertilization with thinning was not significant.
As a comparison, both structure analysis and two-step
analysis were used to model the direct effect of
fertilization on the gross basal area growth of plots.
The results showed that the appropriate variance
estimates to test the significance of the parameters in
fertilizer response equation were larger when using two-step
analysis than when using structure analysis.
