Effects of plot size and census interval on descriptors of forest structure and dynamics
This study was designed to explicitly formulate the effect of census interval and plot size on the variability of descriptors of tropical forest structure (stand density, basal area, aboveground biomass [AGB]) and dynamic (tree growth, mortality and recruitment rates, biomass fluxes). A unique dataset from a broad plot network (37.5 ha) surveyed every 2 yr over a 16-yr period was used to develop and parameterize a new statistical model predicting the coefficients of variation for each forest descriptor. More than 90 percent of the inherent variability of these coefficients was predicted using a simple model including plot size and census interval in a Bayesian modeling framework. All descriptors of forest structure varied by o10 percent for plot sizes 42 ha. Among the descriptors of forest dynamics, AGB loss was the most variable. The number of 6.25 ha plots required to estimate its mean, over a 16-yr period, within a 20 percent error of the mean remains above four. This contrasts with a relative constant flux of biomass entering the plot through tree growth and tree recruitment. Tree growth was remarkably well estimated with o15 percent variability for a 2-yr census in a plot of 2 ha. This study provides an easy method to assess dataset limitations in efforts to estimate descriptors of forest structure and dynamic, which is of primary importance to decipher any clear consequences of global change in tropical forests.
| Summary: | This study was designed to explicitly formulate the effect of census interval and plot size on the variability of descriptors of tropical forest structure (stand density, basal area, aboveground biomass [AGB]) and dynamic (tree growth, mortality and recruitment rates, biomass fluxes). A unique dataset from a broad plot network (37.5 ha) surveyed every 2 yr over a 16-yr period was used to develop and parameterize a new statistical model predicting the coefficients of variation for each forest descriptor. More than 90 percent of the inherent variability of these coefficients was predicted using a simple model including plot size and census interval in a Bayesian modeling framework. All descriptors of forest structure varied by o10 percent for plot sizes 42 ha. Among the descriptors of forest dynamics, AGB loss was the most variable. The number of 6.25 ha plots required to estimate its mean, over a 16-yr period, within a 20 percent error of the mean remains above four. This contrasts with a relative constant flux of biomass entering the plot through tree growth and tree recruitment. Tree growth was remarkably well estimated with o15 percent variability for a 2-yr census in a plot of 2 ha. This study provides an easy method to assess dataset limitations in efforts to estimate descriptors of forest structure and dynamic, which is of primary importance to decipher any clear consequences of global change in tropical forests. |
|---|