Dependency and independency among fish density and electivity indices in a stream fish assemblage

Habitat preferences can be identified from population density and electivity indices, which can be derived from either abundance or occurrence data. This study examined the preference detectability and independency of density and electivity indices from different data sources -- abundance and occurrence. Eight scenarios comprising four fish species along two environmental gradients (flow velocity and water depth) in an undisturbed mountain stream were used in this study. Fish density and seven electivity indices were calculated from abundance and occurrence data for the eight scenarios. The correlation analyses illustrated that abundance electivity indices were positively correlated with density in all eight scenarios while occurrence electivity indices were positively correlated with density in only three scenarios. Electivity indices were positively correlated with each other within the same data quality (abundance or occurrence) but not between. The Chesson's L index was less correlated with other indices and should be used cautiously. Natural historical traits, such as social interactions among organisms, may affect the preference performance of density and electivity indices. In conclusion, density is a good preference indicator. Abundance electivity indices can represent density while occurrence electivity density should be used cautiously for fast assessment. Three electivity indices, Ivlev's E and E′ and Jacobs' D, were recommended for their ease of understanding.