Hierarchical partitioning as an interpretative tool in multivariate inference

摘要: Much of biogeography, conservation and evolutionary biology, ecology involves very large spatial temporal extents. Direct manipulation to test hypotheses is usually almost impossible at appropriate scales so that multivariate modelling especially regression are used draw causal inferences about which ‘independent’ variables influence the distribution abundances species. Such clearly crucial for successful management biological resources conserving threatened A succession approaches has arisen, many yield inconsistent implications. The main problem been quest one (the ‘best’ or ‘optimal’) model from impacts independent inferred. This note attention ecologists a relatively recent method, hierarchical partitioning, does not aim identify best as such but rather uses all models in hierarchy distinguish those have high correlations with dependent variable. likely be most influential controlling variation Hierarchical partitioning regarded substitute experimental when appropriate, it produce better deductions than common ecological situations doubtful value.