3.2 Species differences in adjusted morphometric characters

Following size adjustment, all eleven morphometric characters differed significantly among the three species (p < 0.001). The largest differences were observed for adjusted total length, pre-anal length, pre-dorsal length and dorsal fin length. Games-Howell comparisons showed significant differences between every species pair for all adjusted characters.

Pseudotolithus senegalensis recorded the highest adjusted values for TL, HD, CFL, PDL, PAL and DFL. Lutjanus agennes recorded the highest adjusted values for BD, HL, ED, PFL and AFL, whereas P. quadrifilis recorded the lowest mean values for all adjusted morphometric characters. The mean values and standard deviations of the size-adjusted morphometric characters, together with the Welch's ANOVA results and pairwise comparisons, are summarized in Table 2.

3.3 Principal component analysis

The ten size-adjusted morphometric characteristics were analysed through principal component analysis (PCA) for each species separately: Lutjanus agennes, Pseudotolithus senegalensis and Polydactylus quadrifilis. Sampling adequacy varied among species. The Kaiser-Meyer-Olkin (KMO) measure for L. agennes was 0.513 and Bartlett's test of sphericity was significant, χ²(45) = 343.138, p < 0.001, indicating that the correlation matrix differed from an identity matrix; however, the KMO value indicated marginal sampling adequacy. The eigenvalues of the five components found were greater than 1 and accounted for 62.719% of the total variance. PC1 and PC2 explained 18.368% and 12.309% of the variance respectively. Following Varimax rotation, PC1 had high loadings with HD_ADJ (0.914) and HL_ADJ (0.908); and PC2 had high loadings with PAL_ADJ (0.665) and AFL_ADJ (0.602). 

For P. senegalensis, the KMO value was 0.789 and Bartlett's test was significant, χ²(45) = 395.974, p < 0.001. Two components with eigenvalues greater than 1 were retained, together explaining 37.033% of the total variance. The eigenvalues of PC1 and PC2 were 2.595 and 1.108, respectively, and the percentages of the variance accounted for were 25.955% and 11.078%, respectively. Following Varimax rotation, PC1 showed stronger loadings for PFL_ADJ (0.632), CFL_ADJ (0.606), DFL_ADJ (0.600), and AFL_ADJ (0.492), whereas PC2 was mainly associated with ED_ADJ (0.626), HD_ADJ (0.593), and PAL_ADJ (0.553).

The KMO measure for P. quadrifilis was 0.638 and Bartlett's test of sphericity was significant, χ²(45) = 126.805, p < 0.001. Four components with eigenvalues greater than 1 were retained, explaining 49.134% of the total variance. The eigenvalues of the PC1 and PC2 were 1.745 and 1.092, respectively, with 17.453% and 10.918% of the variance explained, respectively. Following Varimax rotation, PC1 had high loadings for HD_ADJ (0.697), BD_ADJ (0.677) and HL_ADJ (0.649), while PC2 had high loadings for ED_ADJ (0.748), AFL_ADJ (0.606), DFL_ADJ (0.586) and PAL_ADJ (0.503). The species-specific rotated component loadings and communalities are summarized in Table 3. The corresponding species-specific PCA scatter plots are shown in Figure 1.

3.4 Canonical discriminant analysis

The group centroids were clearly separated. P. senegalensis had the highest Function 1 centroid (12.210), P. quadrifilis had the lowest (-14.773), and L. agennes occupied an intermediate position (2.563). On Function 2, L. agennes had the highest centroid (8.014).

The original discriminant classification correctly classified all 1,215 specimens (100%), with all individuals assigned to their respective species. The LOO procedure also correctly classified all 1,215 specimens in the pre-standardised dataset (100%). Because the allometric coefficients and species-specific reference mean standard lengths were estimated from the complete species datasets before cross-validation and were not re-estimated within each LOO iteration, the LOO classification rate should not be interpreted as a fully independent estimate of predictive performance for new specimens. The structure coefficients of the canonical discriminant functions are presented in Table 4. The group centroids of the three fish species on the two canonical discriminant functions are summarized in Table 5. The distribution of individual specimens in the canonical discriminant space is illustrated in Figure 2.

3.5 Meristic differentiation

Counts of dorsal soft rays, anal soft rays, lateral-line scales, pectoral rays, caudal rays, and gill rakers differed significantly among species (Kruskal-Wallis test, p < 0.001). No within-species variation was found in dorsal-spine, anal-spine, or pelvic-ray counts; these were therefore reported as constant within each species. Dunn's post hoc procedure with Bonferroni adjustment was used for pairwise comparisons after significant Kruskal-Wallis tests. The three possible pairwise comparisons among species were performed for each meristic character using the Bonferroni correction. Values reported in Table 5 are presented as mean ± standard deviation (SD); the Kruskal-Wallis test was based on ranked observations rather than the means reported. Superscript letters represent the adjusted pairwise comparisons: values sharing the same superscript letter were not significantly different, whereas values with different letters were significantly different after Bonferroni adjustment. L. agennes and P. quadrifilis did not differ significantly in pectoral-ray, caudal-ray, or gill-raker counts, as indicated by the shared superscript letters.

4 Discussion

4.1 Allometric relationships and size-adjusted morphometric characters

The species-specific relationships in Table 1 show that the magnitude of scaling varied among morphometric characters and species. The majority of characters had high coefficients of determination values meaning that standard length accounted for a substantial proportion of the variation in the majority of characters. For P. senegalensis, eye diameter had a relatively weak scaling whereas total length, pre-dorsal length, pre-anal length and some fin measurements had a strong scaling with the standard length. The variation in the allometric exponents suggests that the bodies of Lutjanus agennes, P. senegalensis and Polydactylus quadrifilis were not similarly scaled. Famoofo and Abdul (2020) reported similar differences in morphometric relationships among 16 species of fish in the freshwater ecotype of Lekki lagoon. Kumari et al. (2020) also observed significant morphometric and meristic differences between the congeneric sciaenids (O. cuvieri and O. ruber). Furthermore, Mahe et al. (2024) found variations in growth pattern of Pseudotolithus species in the West African region. These findings justify the use of species-specific coefficients in the present study, instead of using a common scaling coefficient for all three species.

The differences in size-adjusted morphometric characters shown in Table 2 indicate that the observed variation was not attributable to body size alone. The 11 adjusted morphometric characters were all found to be significantly different among the three species, and the Games-Howell comparisons also revealed that all species pairs were found to be significantly different. There was a higher adjusted value for several characters taken along the length of the body and for the fins for Pseudotolithus senegalensis, whereas Polydactylus quadrifilis had a lower value for most of the measured characters. This pattern is similar to those found by Kumari et al. (2020) who were able to separate congeneric sciaenid species using morphometric characters. Morphological discreteness was also observed between the estuarine croaker Pseudotolithus elongatus by Asuquo and Ifon (2022) which showed that measurable body form can distinguish closely related estuarine fishes. Damora et al. (2025) also found that geometric morphometric and meristic characters were able to show the body shape variation among populations of Anabas testudineus. The present results thus suggest that species identity was still highly correlated with body proportions even when the effects of standard length were taken into account.

The peculiar status of Lutjanus agennes when the new morphometric comparisons were implemented is also corroborated by the study of Ewhuwhe Ezo et al. (2024) which described the L. agennes of the Escravos and Qua Iboe estuaries based on morphological characters and DNA barcoding. Their study illustrated the usefulness of morphological characters in distinguishing and characterizing the African red snapper in Nigerian estuarine environments. Eriegha et al. (2024) also noted that, when viewed from different estuaries along the Gulf of Guinea, there were some morphological differences among samples of the Ethmalosa fimbriata, indicating the potential of comparative morphometric evaluation for detecting spatial morphological variation in estuarine fishes. The morphometric and meristic characters were also used to characterize Bagrus bayad from northeastern Nigeria by Nyaku et al. (2026). The fact that there were significant differences after size adjustment in the present study offers a quantitative basis for distinguishing L. agennes from P. senegalensis and P. quadrifilis.

4.2 Species differences in size-adjusted morphometric characters

The size-adjusted morphometric measurements gave good separation between Lutjanus agennes, Pseudotolithus senegalensis and Polydactylus quadrifilis. Welch’s ANOVA procedure revealed significant differences for each of the 11 adjusted morphometric characters (p < 0.001); Games-Howell procedure revealed that there was significant difference among each species pair (Table 2). The magnitude of the differences was particularly pronounced for adjusted total length, pre-anal length, pre-dorsal length and dorsal fin length. Among the adjusted measurements, P. senegalensis scored the highest values for all measurements except body depth, head length, eye diameter, pectoral fin length and anal fin length, which were higher in L. agennes. P. quadrifilis had the lowest value for all the adjusted measurements (Table 2). These differences were apparent after the adjustment for standard length, and, therefore, suggest differences in body proportions rather than variations due to size alone.

This is in line with previous morphometric evidence of estuarine fishes. It was reported by Eriegha et al. (2024) that morphometric characteristics like standard length, dorsal fin base and caudal peduncle length varied significantly between populations of E. fimbriata from Qua Iboe, Escravos and Lagos Lagoon, suggesting that these characteristics can differ between populations of fish from different estuarine systems. Their finding focused on geographic differences in body size, whereas the present analysis shows that body size could be statistically accounted for and that there was still some differentiation between the three species. This distinction is important as the significant differences in Table 2 cannot be explained by simply the difference in the overall size of individuals. Rather, the uniform separation of all the adjusted characters suggests species-specific differences in body configuration.

A similar discriminatory value of morphometric characters has been reported for closely related fishes. Kumari et al. (2020) concluded that congeneric sciaenids Otolithes cuvieri and O. ruber could be distinguished by morphometric and meristic characters; however, only a few characters are necessary for good separation. In the present results, the same is true for three estuarine species of different taxonomic groups, with all 11 size-adjusted morphometric variables giving significant differences between species. The significant pairwise differences identified by the Games-Howell procedure provide more specific information about which species pairs differed than the overall omnibus test alone.

The present findings are comparable with the report of morphological differentiation in the Cross River estuary for P. elongatus. Asuquo and Ifon (2022) found that fish sampled from estuary islands and estuary mouth can be divided into two main groups, based on the first two principal components, which explained 85.6% of the total variation and correctly classified 66.0% of the specimens using the discriminant analysis. They found that size-adjusted morphometric measurements were able to identify structural variations between estuarine groups. In the present study, all adjusted morphometric characters differed significantly among species, and the post hoc comparisons separated all species pairs (Table 2).

4.3 Principal component analysis

Species-specific principal component analyses described the major sources of within-species morphometric variation. For L. agennes, five components with eigenvalues > 1 explained 62.719% of the variance; PC1 was mainly related to adjusted head depth and head length, whereas PC2 was mainly related to adjusted pre-anal length and anal fin length. For P. senegalensis, two components explained 37.033% of the variance; pectoral fin length, caudal fin length and dorsal fin length contributed most to PC1, whereas eye diameter, head depth and pre-anal length contributed most to PC2. For P. quadrifilis, four components explained 49.134% of the variance; head depth, body depth and head length had the strongest loadings on PC1, whereas eye diameter, anal fin length, dorsal fin length and pre-anal length contributed most to PC2 (Table 3).

The species-specific PCA plots illustrate within-species morphometric variation (Figure 1). Because PCA was conducted separately for each species, the three plots should be interpreted as separate descriptions of variation rather than as a common ordination showing interspecific separation. The results are consistent with the use of PCA to describe morphometric variation in estuarine fish populations reported by Asuquo and Ifon (2022).

Taken together, the univariate comparisons, species-specific PCA results and canonical discriminant analysis show that the three species differ in body and fin proportions. The PCA results describe how the relative contribution of these traits varies within each species, whereas the discriminant analysis provides the direct multivariate comparison among species (Tables 2-5; Figures 1;  Figure 2).

4.4 Canonical discriminant analysis

Canonical discriminant analysis provided an additional multivariate assessment of separation among the three species. Table 4 shows that PAL_ADJ, PDL_ADJ, DFL_ADJ and HD_ADJ contributed most strongly to Function 1, whereas PFL_ADJ, AFL_ADJ, ED_ADJ, BD_ADJ and HL_ADJ contributed more strongly to Function 2. These variables therefore contributed to the observed interspecific discrimination.

The group centroids in Table 5 show clear separation among the three species. P. senegalensis had the most positive centroid on Function 1, P. quadrifilis had the most negative centroid on Function 1, and L. agennes was separated primarily on Function 2. This pattern is consistent with the canonical discriminant plot (Figure 2) and indicates that the sampled species occupied distinct positions in the discriminant space. Similar applications of multivariate morphometric measurements have been reported for fish groups by Ivanova et al. (2024) and Gonzalez-Martinez et al. (2020).

The classification results support the discriminatory ability of the measured characters within the sampled dataset. Both the original classification and leave-one-out classification on the pre-standardised dataset correctly classified all sampled specimens. However, because the species-specific allometric coefficients and reference mean standard lengths were estimated from the complete datasets before the LOO procedure, the 100% classification rate represents classification consistency within the present pre-standardised dataset rather than independent predictive performance for unknown specimens. These results indicate that the selected morphometric variables distinguished the sampled individuals of the three species; however, the classification accuracy should be interpreted within the context of the present samples and should not be generalised to all populations of the three species.

4.5 Canonical discriminant plot

Figure 2 graphically presents the canonical discriminant scores of the three species. The groups showed minimal overlap, consistent with the separation of the group centroids in Table 5. The plot provides a visual summary of the multivariate differences among the sampled species.

4.6 Meristic differentiation

Meristic characters provided an additional basis for distinguishing the three species. Dorsal soft rays, anal soft rays, lateral-line scales, pectoral rays, caudal rays and gill rakers differed significantly among species (Table 6). These results complement the morphometric findings because meristic characters are countable anatomical traits. Similar use of morphometric and meristic characters for fish discrimination has been reported for sciaenids and other fish groups (Kumari et al., 2020; Scott et al., 2020; Nyaku et al., 2026).

The meristic profile of L. agennes can also be considered alongside the characterisation reported by Ewhuwhe Ezo et al. (2024), who combined morphological identification with DNA barcoding for specimens from the Escravos and Qua Iboe estuaries. In the present study, L. agennes differed from the other two species in several meristic characters. Together, these observations support the usefulness of comparative morphological characters for describing Nigerian estuarine fishes.

5 Conclusion

This study provides quantitative evidence of the difference in morphometric and meristic characteristics of the three species, Lutjanus agennes, Pseudotolithus senegalensis and Polydactylus quadrifilis sampled from the Benin, Brass, and Escravos estuaries of the Niger Delta. Allometric analysis was done for each species to obtain the appropriate size adjustment relationship, and then all 11 morphometric characters exhibited statistically significant interspecific differences. The Games-Howell comparisons revealed significant differences between all three species for all the adjusted morphometric characters with notable differences occurring in the pairwise comparisons of all three species, thus the observed interspecific differences were expressed consistently in all the body and fin proportions measured. L. agennes recorded higher adjusted values for body depth, head length, eye diameter, pectoral-fin length, and anal-fin length, whereas P. senegalensis recorded higher values for total length, head depth, caudal-fin length, pre-dorsal length, pre-anal length, and dorsal-fin length. P. quadrifilis generally showed the lowest adjusted values.

The study also shows that the characters contributing to the variation were not identical in the three species. Within-species variation was determined by PCA, which resulted in different combinations of characters being important for each species, and by CDA, which demonstrated clear multivariate separation among the three species based on combined size-adjusted morphometric characteristics. The pre-anal length and pre-dorsal length, dorsal fin length and head depth were the most important for the first discriminant function, while pectoral fin length, anal fin length, eye diameter, body depth and head length were more important for the second function. Each of the original and LOO classification procedures resulted in a 100% correct classification in the present pre-standardised data set, but this was in the context of correctly classifying data points from the same data set rather than predicting which data points from an unknown data set they would classify correctly.

Species-level evidence also came from meristic data which showed significant interspecific differences in dorsal soft rays, anal soft rays, lateral-line scales, pectoral rays, caudal rays, and gill rakers, with dorsal-spine, anal-spine, and pelvic-ray counts constant within each species in the sampled specimens. The present study integrated species-specific allometric adjustment, univariate morphometric comparison, species-specific PCA, multivariate discrimination and meristic characters to describe the morphological variation in these three estuarine fishes.

Combined, these data provide a quantitative morphological baseline for L. agennes, P. senegalensis and P. quadrifilis in the estuarine systems sampled in the Niger Delta and identify specific morphometric and meristic characters that can be used for comparative species characterisation and identification. Because the analyses were carried out using fish from the sampling sites described here, the results are applicable only to the sampled populations and should not be extrapolated to other populations without broader geographic sampling and further validation.

Author Contributions

Akpomughe E. designed and conceived the study, sampled and collected the specimens in the field, assisted in data collection, identified the species, coordinated the statistical analysis, interpreted the results and prepared the original manuscript. O.F. Nwachi participated in study design, field sampling and collection of the specimens, morphometric and meristic measurements, data interpretation and critical revision of the manuscript. J.K. Ekelemu participated in the design of the study and supervision, sampling in the field and species identification, interpretation of results and critical review and editing of the manuscript. The manuscript has been read and approved by all authors and each author agrees to be accountable for all aspects of the work.

Acknowledgements

The authors would like to express their appreciation to the field assistants and persons who helped in the collection, handling, measurement and identification of fish specimens during the study. Their cooperation and assistance in sampling operations in the estuaries of Benin, Brass and Escravos are gratefully acknowledged. The authors appreciate the support of all who helped to complete the field and laboratory activities of this study.

References

Asuquo P.E., and Ifon H.T., 2022, Morphological discreteness of the estuarine croaker Pseudotolithus elongatus (Teleostei: Sciaenidae), Journal of Fish Biology, 100(3): 619-624.

da Costa L.M., Maia H.A., and Almeida A.J., 2022, The fishes of the Gulf of Guinea oceanic islands, Biodiversity of the Gulf of Guinea Oceanic Island, Science and Conservation, 2022: 431-478.

Damora A., Afrisal M., Wibowo K., Nur F.M., and Haryono H., 2025, Landmark based geometric morphometric and meristic analysis of body shape variation among Indonesian climbing perch Anabas testudineus (Bloch, 1792) populations, The European Zoological Journal, 92(1): 454-466.

Dolo O., Kondiano S.G., Camara R.H., Diallo I., Sangare A., and Traore L., 2026, Assessment of the nutritional value of fresh Polydactylus quadrifilis and Galeoides decadactylus sold at the Bonfi Market in the Republic of Guinea, Food and Nutrition, 17(1): 14-26.

Eriegha O.J., Ekelemu J.K., and Nwachi O.F., 2024, Comparative morphology of Ethmalosa fimbriata (Bowdich, 1825) from three estuaries adjoining the Gulf of Guinea, Nigeria, Ilmu Kelautan: Indonesian Journal of Marine Sciences, 29(2): 211.

Ewhuwhe Ezo J., Eriegha O.J., and Eyo V.O., 2024, Morphological characterization and DNA barcoding of Lutjanus agennes (African red snapper) from Escravos and Qua Iboe estuaries, International Journal of Maritime and Interdisciplinary Research, 6(2): 18-31.

Famoofo O.O., and Abdul W.O., 2020, Biometry, condition factors and length weight relationships of sixteen fish species in Iwopin freshwater ecotype of Lekki Lagoon, Ogun State, Southwest Nigeria, Heliyon, 6(1).

Gonzalez-Martinez A., Lopez M., Molero H.M., Rodriguez J., González M., Barba C., and García A., 2020, Morphometric and meristic characterization of native chame fish (Dormitator latifrons) in Ecuador using multivariate analysis, Animals, 10(10): 1805.

Ivanova P.P., Zlateva I.Y., Raykov V.S., Yankova M.H., Dzhembekova N.S., Slabakova V.H., and Raev Y.R., 2024, Comparative analysis of morphometric and meristic characters of turbot Scophthalmus maximus (Linnaeus, 1758) from four sites along the Bulgarian Black Sea coast, Acta Zoologica Bulgarica, 76(1): 41-49.

Kumari S., Jaiswar A.K., Jahageerdar S., Chakraborty S.K., and Kumar T., 2020, Morphometric and meristic variation of congeneric sciaenid fishes Otolithes cuvieri Trewavas, 1974 and Otolithes ruber (Schneider, 1801) from Maharashtra, west coast of India, Indian Journal of Geo-Marine Sciences, 49(1): 80-86.

Mahe K., Ekperusi A.O., Diouf K., Munschy C., Aminot Y., Andrialovanirina N., Dussuel A., Diop M., Diop M., Ekperusi A.O., 2024, Growth assessment of croakers (Pseudotolithus species) in West African waters, Frontiers in Fish Science, 2: 1498784.

Nyaku R.E., Garba T.H., Bello A.S., Yahaya S.M., Mohammed Z., Haruna U.N., and Mohammed N., 2026, Assessment of morphometric and meristic characterization of silver catfish (Bagrus bayad) from Northeast Nigeria, Sahel Journal of Life Sciences FUDMA, 4(1): 463-472.

Scott S., Rojas P., and Vila I., 2020, Meristic and morphological differentiation of Orestias species (Teleostei; Cyprinodontiformes) from the southern Altiplano, Environmental biology of fishes, 103(8): 939-951.