Table 3

Pearson correlation, multiple regression, and random forest results for relating residual regional fish species richness to regional environmental data composited at the fine-grain ecoregion extent. Variables sorted in order of decreasing correlation strength.

Variable Pearson correlation coefficient Multiple regression coefficient Random forest % increase in MSE Random forest node purity
Road Density (km/km2)# 0.611***  
Mean Elevation (m)# –0.593***   11.7 8313
% Grass + Shrub Land Cover –0.588*** –1.00***
Population Density (#/km2)# 0.577***  
Mean Air Temperature (°C) 0.571*** 5.03*** 9.35 5949
Mean Precipitation (mm)# 0.549*** –51.5* 12.3 7620
% Developed Land Cover 0.403**   19.5 15690
% Agricultural Land Cover 0.324*   15.0 8666
% Bare Land Cover –0.305*   6.40 4535
% Wetland Land Cover 0.211   1.76 2367
Mine Density (#/km2)# –0.190   4.49 3450
% Forested Land Cover 0.188   7.06 3577
Surface Area (km2)# –0.138   4.35 7171
% Open Water Land Cover 0.117 –4.22*
% Historical Glaciation –0.037   6.42 1159
         
Multiple Regression Intercept   130.6    
R2   0.631 0.546  
Root Mean Square Error   21.8 23.2  
Sample Size 61 61 61 61
#

Variable was log10 transformed for multiple regression and Pearson correlation.

– redundant variable not included in random forest.

Significance of the correlation, or that the regression coefficient is unequal to zero; *p < 0.05, **p < 0.01, ***p < 0.001.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.