IJA_2026v16n5

International Journal of Aquaculture, 2026, Vol.16, No.5, 354-361 http://www.aquapublisher.com/index.php/ija 357 The size standardisation itself did not require extrapolation to an arbitrarily selected body size outside the sampled populations because the reference value for each species was its observed mean SL. The mean reference lengths of 28.8642 cm for L. agennes, 36.5891 cm for P. senegalensis, and 21.2689 cm for P. quadrifilis were all located within their respective observed size ranges. However, the three species did not have completely overlapping size ranges. L. agennes ranged from 12.90 to 45.60 cm, P. senegalensis from 26.00 to 48.50 cm, andP. quadrifilis from 11.00 to 68.00 cm. Consequently, the size-adjusted values should be interpreted as comparisons of body proportions after accounting for species-specific allometric scaling rather than as direct comparisons of fish measured at an identical observed body size across the entire size range. All regression coefficients used in the adjustment were estimated independently for each species from the 405 specimens available for that species. The resulting size-adjusted morphometric variables were subsequently used for the comparative multivariate analyses. 2.5 Statistical analysis of adjusted morphometric characters Morphometric and meristic differentiation between three fish species, Lutjanus agennes, Pseudotolithus senegalensis and Polydactylus quadrifilis sampled from three estuaries of Niger Delta (Benin, Brass and Escravos) were assessed by using a comparative cross-sectional sampling design. In each of the three estuaries all three species were collected and so a representation of all species at the sample locations was obtained. A total of 405 specimens per species were used for morphometric measurements, and 60 specimens per species were used for meristic characters. Allometric relationships were first found for each species and then used to size adjust the measurements for morphometric comparisons. Lastly, the species were compared with respect to the adjusted morphometric variables using Welch’s ANOVA because the homogeneity-of-variance assumption was not satisfied. Games-Howell between pair comparisons were conducted for significant overall differences. The within-species variation was analysed by the principal components analysis for each species and the contribution of morphometric characters in the within-species variation was evaluated. Canonical discriminant analysis (CDA) was used to assess the multivariate ability to discriminate among the three species. The Kruskal-Wallis test was used to compare the meristic characters among species and Dunn's post hoc test and Bonferroni corrections were used when significant differences were found. 2.6 Principal component analysis The adjusted morphometric characters were used for the principal component analysis. The suitability of the dataset for PCA was assessed using the Kaiser-Meyer-Olkin (KMO) measure and Bartlett’s test of sphericity. Components with eigenvalues greater than 1 were retained. To facilitate the interpretation of the loadings on the components, varimax rotation and Kaiser normalisation were used. Component scores were calculated for each species and used to visualise within-species morphometric variation. 2.7 Canonical discriminant analysis The species was used as the grouping variable, and ten size-adjusted morphometric characters: pre-anal length (PAL_ADJ), pre-dorsal length (PDL_ADJ), length of dorsal fin (DFL_ADJ), depth of the head (HD_ADJ), depth of body (BD_ADJ), length of the head (HL_ADJ), length of the pectoral fin (PFL_ADJ), length of the anal fin (AFL_ADJ), length of the caudal fin (CFL_ADJ) and diameter of the eye (ED_ADJ) were used as predictor variables. Eigenvalues, canonical correlations and structure coefficients were used to evaluate the discriminatory contribution of the canonical functions, group centroids were used to examine the relative positions of the three species in the discriminant space and the discriminatory contribution of the canonical functions was evaluated using structure coefficients. Prior to CDA each morphometric character was standardised separately for each species, using the species-specific allometric coefficient, and the actual mean standard length of the entire data set for each species. Size-adjusted measurements were used as predictors in the CDA. The classification performance was evaluated using the original classification and leave-one-out (LOO) cross-validation was used to evaluate the classification

RkJQdWJsaXNoZXIy MjQ4ODYzNA==