Hierarchical partitioning of predictor variables explaining fish and macroinvertebrate richness.

Table shows the independent (I), joint (J) and total effect of predictors on taxonomic richness. The percent contribution of the predictor to the explained variance of the response variable is shown as I%. Z-scores are based on the distribution of randomized Is and are calculated as [observed−mean (500 randomizations)]/SD (500 randomizations), and statistical significance (*) is based on the upper 0.95 confidence limit (Z≥1.65). The simple correlation coefficient is also shown to clarify the nature of predictors' relationship with taxonomic richness, but it is not part of the hierarchical partitioning analysis. Acronyms as in Table 1.