Relationship between bird species richness and limnological variables
The redundancy analysis (RDA) identified two significant RDA axis and 34.4% of the variation in species richness of the reservoirs is explained by environmental variation (R2adj = 0.34375; P < 0.001). Four variables were selected as the best factors explaining the variation in bird species richness observed: elevation, water depth, pH and nutrients. A clear association between the occurrence of specific bird species and several of the measured environmental variables was detected. A strong relationship between species richness and size of reservoir is detected but not the depth. That is bigger reservoirs did support a higher number of bird species (see Appendix S4 in supporting information).
The RDA indicates that a significant portion of the variation in species composition is explained by both biological variables (BV) and environmental variables (ENV) (R2adj = 0.18, p<0.001). If only environmental and biological variables are used in the RDA, environmental variation explained 9.5% of the variation in species composition (R2adj = 0.0953, p<0.001), whereas the biological variables (BV) explains about 8.9% of the variation in species composition (R2adj = 0.089, p<0.001; Fig. 5A). Among the environmental variables, elevation, depth, pH, transparency and TP contributed significantly to explaining the variation in species composition in the studied reservoirs. In a partial RDA analysis, the complete model (ENV ∪ BV) accounted for about 18% of the variation in species composition (R2adj = 0.1799, p<0.001; Fig. 5A) with environmental variables (R2adj = 0.091, p<0.001) being most important. But significant portion of the variation in species composition was also explained by pure biological variables (R2adj = 0.085, p<0.01), while shared environmental and biological variables contributed less than 1% (Fig. 5A and see Appendix S5 in supporting information). A large part of the variation in species composition, however, remained unexplained (R2adj = 0.82; Fig. 5A).
Including age of reservoirs (AGE) as an independent variable in the partial RDA did change the percentage of variation explained by environmental variables (R2adj = 0.107, p< 0.001; Fig. 5B) and biological variables (R2adj = 0.074, p< 0.001) differently. It did, reduce the variation explained by biological descriptors (BV), and the pure effect of BV is down to 7.4% (R2adj = 0.074, p< 0.001; Fig. 5B). The combined effect of ENV, AGE and BV (R2adj = 0.203) was higher than that of the effect without age. The pure effect of AGE was small but statistically significant (R2adj = 0.023, p < 0.05; Fig. 5B) and its effect was also confounded with BV effects (Fig. 5B; see also appendix S5 in supporting information). Only a small fraction of the variation was shared among BV descriptors and environmental heterogeneity (R2adj = 0.005, p< 0.001; Fig. 5B).
The global model, using ENV, BV, AGE and geographic location of each reservoir (SPACE) as an independent variables in the partial RDA did not change the percentage of variation explained by environmental variables (R2adj = 0.107, p< 0.001; Fig. 5B) and biological variables (R2adj = 0.074, p< 0.001). However, the model indicated the marginal effect of SPACE on the bird species richness to be 5.6%. Out of these, 4% is pure effect of SPACE and 1.6% is confounded with effect of AGE and BV variables (see Appendix S5 in supporting information).