Designing plots for precise estimation of forest attributes in landscapes and forests of varying heterogeneity

Models of relationships among forest inventory sampling efficiency and cluster plot configuration variables inform decisions by inventory planners. However, relationships vary under different spatial heterogeneity scenarios. To improve understanding of how spatial patterns of forests affect these relationships, we implemented a factorial experiment by simulating forest pattern at both the landscape and stand scales. We sampled these simulated forests with a variety of cluster plot configurations, calculated coefficient of variation (CV) of trees per hectare for each replicate, and tested the relationships among CV and the heterogeneity and cluster plot configuration factors within a linear mixed model framework. Both landscape- and stand-scale pattern aggregation had a significant relationship with CV. Changing cluster plot configuration factors did little to change the overall CV when using larger subplots but had some important effects when using smaller subplots. These impacts were stronger in the more uniform landscapes. Results were opposite for stand-scale heterogeneity; changing plot configuration in areas with aggregated patterns had a stronger impact than it did in areas with more uniform patterns. Results of this study reveal the importance of accounting for spatial pattern at multiple scales when making cluster configuration choices if the goal is statistical efficiency.

Automated summary

The study reveals that the spatial pattern of forests significantly affects sampling efficiency, and different pattern configurations have varying impacts at different scales. In more uniform landscapes, changes in cluster plot configurations are more important for CV; whereas in stand with aggregated patterns, altering plot configurations has a stronger impact on CV.