International Journal of Aquaculture, 2026, Vol.16, No.5, 293-314 http://www.aquapublisher.com/index.php/ija 304 3.5.2 PCA dimensionality reduction and variance apportionment The PCA reduced the detailed water-quality dataset into two orthogonal axes that together contained 91.65% of the total environmental variance. PC1 becomes the first dimension of change extracted, accounting for 60.05% of the observed variation within the data, as noted in the loadings part of Table 3. This component is marked by a clear divide between mineral-based properties and elements associated with organic enrichment. The dominant positive loadings for Total Alkalinity (0.371), Total Hardness (0.367), and Chemical Oxygen Demand (0.354) suggest a defining axis of ionic concentration and chemical oxidation potential. Total Dissolved Solids (-0.371), Sulphate (-0.365), and Chloride (-0.362) have much higher negative loadings than all other variables, which suggests that the first principal component (PC1) captured an important, large gradient in dissolved mineral inputs to Lake Merritt, which may vary with hydrological conditions during sampling periods. PC2 captures an additional 31.6% of the variance in the system and acts as a primary indicator of physicochemical and thermal seasonality. Additionally, as indicated in Table 3, this component is positively affected by Temperature (0.479) and Nitrate-Nitrogen (0.279); however, it exhibits negative loadings at a notable extent for pH (-0.478), Dissolved Oxygen (-0.466), and Phosphate (-0.420). The significant inverse relationship between Temperature and Dissolved Oxygen on this axis reflects a known theory that higher thermal energy decreases gas solubility. Defaulting Nitrate and Temperature against Dissolved Oxygen and pH extremes, with respect to PC2, suggests that it expresses the biological and physical "pulse" of the waterbody correlated with the variations in nutrient availability as well as lower levels of oxygen in hotter seasons. Table 3 First two principal components loadings Parameters PC1 Loadings PC2 Loadings Temperature 0.124 0.479 pH 0.096 -0.478 DO -0.125 -0.466 BOD -0.243 0.182 COD 0.354 -0.146 TDS -0.371 0.003 TH 0.367 -0.068 TA 0.371 0.018 Chloride -0.362 0.035 Sulphate -0.365 -0.094 NN -0.238 0.279 Phosphate -0.21 -0.42 3.5.3 Multivariate significance of seasonal variation (MANOVA) To determine if the visual patterns seen in Figure 9 represent statistically significant differences rather than stochastic noise, a Multivariate Analysis of Variance (MANOVA) was performed on the PC scores in Table 4. The analysis showed a Pillai's Trace of 1.9631 with an approximate F-statistic of 159.8 and a p-value of 2.697e-10. These results indicate significant multivariate differences among the sampled seasons. The near-zero p-value gives us the confidence to reject our null hypothesis, which proposes that each season would share the same hydrochemical "fingerprint". The use of PC scores as the dependent variables in this model was a critical methodological step to ensure the analysis remained statistically identified, given the small sample size (n=9) relative to the original number of parameters (d=12). 3.5.4 Interpretation of seasonal clustering and biplot analysis The findings are visually synthesized in Figure 8, which provides a seasonal division of the sampling units into three domains, which is further corroborated by the coordinate values found in the Scores section (Table 4). Samples from the Summer (Points 1-3) appear in the upper-right quadrant of Figure 1 with high positive scores on both PC1 and PC2, coinciding with vectors describing Temperature, Total Alkalinity, and COD. This trend
RkJQdWJsaXNoZXIy MjQ4ODYzNA==