近期发表在《Ecological Applications》期刊上的一项研究，由加拿大麦克马斯特大学Madison L. McCaig等人完成，题为“Response of stream habitat and microbiomes to spruce budworm defoliation: New considerations for outbreak management”，深入探讨了东部云杉芽虫的剥叶行为对溪流栖息地和微生物群落结构及功能的影响。在本研究中，rdacca.hp包被用来评估累积剥叶对溪流栖息地变量（如营养物浓度、水流速率、水温）和溶解有机物（DOM）质量的影响。通过层次分割（hierarchical partitioning）建模方法，研究者能够量化累积剥叶与其它景观驱动因素（例如海拔、森林组成）对溪流微生物群落结构变化的相对重要性。文章中关于共线性的描述和rdacca.hp的使用非常到位，供大家描述方法时参考：

Given the intercorrelation between defoliation, elevation, and forest composition that was previously observed in these watersheds (Sidhu et al., 2024), we employed a hierarchical partitioning (HP) modeling approach for variables with significant relationships to defoliation (flow, temperature, and SUVA) to compare the relative importance of cumulative defoliation to other potential landscape drivers. Briefly, HP involves the calculation and partitioning of goodness-of-fit for all potential models (2k for k predictors) within a multiple regression, enabling the interpretation of relative explanatory contributions of colinear variables (whereas traditional modeling cannot) (Lai et al., 2022; MacNally, 2002). HP was conducted using the rdacca.hp (1.0-8) R package, and determined both the shared and unique variance explained for cumulative defoliation and the landscape variables in all possible model combinations (Lai et al., 2022).