Modelling the Influence of Physicochemical Variables on Fish Species Richness in the Kaduna River System

This study developed predictive models relating physicochemical water quality variables to fish species richness in the Kaduna River system, encompassing sites across Kaduna, Niger, and Kogi states in Nigeria's North Central zone. Understanding the environmental drivers of fish community composition is essential for designing effective habitat conservation and water quality management strategies. Fish sampling and concurrent physicochemical measurements were conducted at 14 stations distributed along the river over six sampling occasions spanning wet and dry seasons. Fish samples were collected using standardised multi-gear methods, and species richness and diversity indices were calculated for each station-occasion. Multiple linear regression and generalised additive models were employed to identify the physicochemical variables most strongly predictive of species richness. Results showed that dissolved oxygen, conductivity, and substrate type were the three most significant predictors of fish species richness in both wet and dry season models. Dissolved oxygen explained 38 percent of variance in species richness in the dry season model, reflecting the constraining influence of oxygen depletion during low-flow conditions. Elevated turbidity, linked to upstream erosion, was negatively associated with species richness. Stations below major urban discharge points showed significantly reduced richness compared to upstream reference sites. The generalised additive model outperformed linear regression, capturing nonlinear responses of species richness to conductivity and temperature. The study provides quantitative evidence for the primacy of dissolved oxygen management in Kaduna River conservation and recommends its adoption as a target indicator in state-level water quality standards applied to the Kaduna River system. Keywords: species richness modelling, physicochemical variables, Kaduna River, freshwater ecology, predictive modelling