IJA_2026v16n5

International Journal of Aquaculture, 2026, Vol.16, No.5, 293-314 http://www.aquapublisher.com/index.php/ija 296 formalin solution, and the bigger ones in 10% aqueous formalin solution (v/v), following a small incision in the abdominal wall to ensure the preservative adequately penetrated the internal organs of the fish. Taxonomic identification followed standard ichthyological references (Talwar and Jhingran, 1991; Daniels, 2002), and current scientific nomenclature was verified using Eschmeyer’s Catalog of Fishes (Fricke et al., 2024). Species abundance data (N = 140) were used to estimate the following diversity metrics: Shannon-Wiener index (H′ = −∑pᵢ ln pᵢ), maximum diversity (Hmax = ln S), and Pielou’s evenness (J′ = H′/Hmax), where pᵢ is the proportional abundance of each species, and S is total species richness (Shannon and Weaver, 1949; Pielou, 1966). Relative contributions of species were assigned as the total abundance (%) of functional guilds based on their cross-flow, habitat use, and substrate preferences (Root, 1967). The resulting assemblage structure was assessed across flow regime, substrate complexity, and anthropogenic disturbance along the headwater reach (Naiman et al., 2005). 2.7 Water quality analysis and WQI calculation Physico-chemical parameters, including temperature, pH, Dissolved Oxygen (DO), Biochemical Oxygen Demand (BOD), Chemical Oxygen Demand (COD), Total Dissolved Solids (TDS), total hardness, total alkalinity, chloride, nitrate, and phosphate, were analysed following standard methods prescribed by the American Public Health Association (APHA), 2017. All measurements were conducted seasonally (summer, monsoon, and winter) during 2023-2024. Water Quality Index was computed using the weighted arithmetic index method (Horton, 1965; Brown et al., 1973): WQI= =1 =1 where =100 × − − , and = 1 1 calculated based on observed values of each physico-chemical factor ( ), the standard BIS/WHO permissible values for each of the corresponding ( ) and ideal values ( ). 2.8 Statistical methodology Seasonal water quality patterns were evaluated using a stepwise multivariate statistical approach. Principal Component Analysis (PCA) was first used to reduce dimensionality and identify major environmental gradients (Jolliffe, 2002; Hair et al., 2019). In the first step, correlated physical and chemical parameters were transformed into orthogonal Principal Components (PCs), isolating key variance drivers and filtering out extra noise (Legendre and Legendre, 2012). Subsequently, Multivariate Analysis of Variance (MANOVA) was conducted on scores for the first two PCs as dependent variables. This methodology was specifically employed to overcome any potential rank deficiency concerns due to small sample sizes (n=9), where the number of parameters outnumbers the observations and statistical confidence in differentiating seasonal groups was required, ensuring a sound statistical identification of meaningful differences across all three seasons (Rencher and Christensen, 2012; Hair et al., 2019). The relationships between specific parameters (loadings) and seasonal clusters (scores) were visualized by a PCA Biplot, presenting a spatial representation of the hydrochemical niche for each season (Gabriel, 1971; Legendre and Legendre, 2012). Then, a Pearson correlation heatmap was used to evaluate the linear dependencies between variables, including the inverse dependency that exists between temperature and dissolved oxygen (Benesty et al., 2009), in order to help support the internal validity of the data. Finally, a Hierarchical Cluster Analysis (HCA) generated a dendrogram, using the Ward’s method to determine intraspecific similarities and detect potential sampling outliers (Ward, 1963; Murtagh and Legendre, 2014). Execution of PCA followed by MANOVA allows for assessing the overall covariability on a global level, which provides insights into the more general patterns of environmental relevance; specific component pairwise interactions can be assessed through correlation analysis and, in turn, treated at the sample level via HCA: from

RkJQdWJsaXNoZXIy MjQ4ODYzNA==