Impact of plot size and spatial pattern of forest attributes on sampling efficacy

We aimed to investigate how the spatial pattern of a forest attribute is connected with design-based sampling efficacy in relation to the selection of plot size and quantity, expressed together by the cruise percent (CP). Spatially random or aggregated patterns of stem volume (m³/ha) with various hierarchical strengths were generated as model plantations. Two design strategies using rectangular plots were examined under conditions of increasing CP: one featured a similar number of sample plots with varying sizes and the other an increasing number of sample plots of a fixed size. Each design was evaluated for respective populations by sampling distribution simulation. The results show that the strength of the spatial patterns was negatively correlated with the sampling precision and that aggregated patterns weakened the sampling precision much more severely than did random patterns. As a result, a small plot size should be preferred, because for a given CP, the smaller the plot is, the more it will improve the sampling precision and convergent speed. We confirmed by comparing the two design strategies that brand-new plots should be established at brand-new sites rather than simply enlarging the size of old plots. We conclude that a spatially dispersed population favors a spatially nondispersed design and a spatially nondispersed population favors a spatially dispersed design. Because these results were derived on a pan-population basis, the knowledge acquired here may be useful as a rule of thumb for planning design-based forest inventories in the future.