Md Asduzzmn, Ismt Jhn, Aysh Rhi Noor, Md Moudud Islm,Md Moshiur Rhmn

aFaculty of Fisheries, Chattogram Veterinary and Animal Sciences University, Khulshi, 4225, Bangladesh

bFisheries and Marine Resource Technology Discipline, Khulna University, Khulna, 9208, Bangladesh

Keywords:

Mangrove crabs

Morphometric traits

Species delineation

Discriminant function analysis

Principal component analysis

Univariate ANOVA

ABSTRACT

Despite the importance of mud crab species (genus Scylla) in both coastal aquaculture and artisanal fisheries,knowledge of the species diversity and stock structure of mud crab population for their sustainable management across the coastal region of Bangladesh remains obscure. Therefore, the present study was conducted to delineate the diversity and stock structure of mud crab species based on the 24 morphometric length measurements and 27 morphometric ratios data collected from the three coastal regions (Chittagong, Cox's Bazar and Bagerhat) of Bangladesh. Based on the taxonomic keys, majority of the species (about 82.4%) were identified as S. olivacea,while the remaining (about 17.6%) was S. serrata. In consistent to our preliminary analysis, both the univariate proportion and multivariate analysis of the collected morphometric lengths and ratios data also distinctly divided the mud crabs population into two multivariate spaces as S. serrata and S. olivacea. For the further analysis of the stock structure of these two species, data were subjected to the univariate ANOVA, multivariate ANOVA, Canonical Variate Analysis (CVA), and Principal Component Analysis (PCA) according to their collection sites.Scatter plots of CVA scores between CV1 and CV2 and PCA scores between PC1 and PC2 showed that the individuals of both species clearly made clusters into three multivariate spaces with minimum overlaps according to their collection sites. In morphometric traits Linear Discriminant Function Analysis (LDFA), the average percentage of correctly classified (PCC) into the original groups of the S. olivacea and S. serrata was 100%. For stock discrimination, the average PCC into their original groups of collection sites were 100% and 99% for the S. olivacea and S. serrata, respectively. Finally, dendrogram based on the Euclidean distances consistently separated S. olivacea and S. serrata and stock structures similar to other multivariate analyses. In conclusion, our study shows that S. olivacea is the dominant mud crab species and S. serrata is the minor one, and the stock discrimination of each species should be considered in the management and conservation policy of these species along the coastal region of Bangladesh.

1.Introduction

The mud crabs belong to genus

Scylla

, are fast growing species that attains larger size among portunids and are widely distributed throughout the estuarine environment between the lower intertidal zones and offshore, such as mangrove forest and creeks of the Indo-Pacific region (Keenan, Davie, & Mann, 1998; Knusten, Jorde, Andre,& Stenseth, 2003; Macintosh, 1988). They represent a valuable component of traditional, small scale coastal fisheries in several tropical and subtropical Southeast Asian countries including Bangladesh(Overton, Macintosh, & Thorpe, 1997). Besides the importance of small-scale coastal fisheries, mud crabs are now considering an attractive alternative to shrimp farming along the coastal areas of Bangladesh due to their rapid growth, less susceptible to disease, more resistant to adverse environmental conditions and high commercial value in both local and international markets (Hussain, 2014; Uddin, Shah, Khanom,& Nesha, 2013). At present, almost all forms of mud crab farming is based on the stocking of juvenile crabs caught in the wild, and such dependence on the natural stocks is deemed unsustainable in the medium- and long-term farming (Ikhwanuddin, Azra, Siti-Aimuni, &Abol-Muna fi, 2012; Lindner, 2005; Shelley, 2008; Shelley & Lovatelli,2011). Moreover, the degradation of natural habitats, increasing water pollution, and overexploitation result in declining the wild population of mud crabs (Jirapunpipat, Yokota, & Watanabe, 2009). Therefore,continuous reliable supply of hatchery-produced juveniles and exploring stock assessment are utmost necessary for the long-term sustainability of mud crab farming as well as its conservation in the wild. In order to attain these objectives, a clear understanding of the species diversity and stock structure of the local mud crabs of a country is prerequisite.However, taxonomic identification of the mud crab species is very contentious because of their morphological plasticity, overlapping morphological and morphometric traits (Joel & Raj, 1983).For several years, the taxonomy of mud crabs has long been controversial worldwide due to their high morphological resemblance and the absence of distinct morphological diagnostic characters. Such controversy of taxonomic identification of mud crab species has been remarkably resolved by the revision of morphological traits by Keenan et al. (1998). They recognized four species of

Scylla

:

S. serrata

(former

S. oceanica

),

S. tranquebarica

,

S. olivacea

(former

S. serrata

) and

S paramamosain

(former

S. serrata

var.

paramamosain

). Subsequently, many other researchers (Macintosh, Overton, & Thu, 2002; Ma, Zhang, Ma, &Qiao, 2006; Sangthong & Jondeung, 2006; Jirapunpipat, Aungtonya, &Watanabe, 2008; Ogawa, Hamasaki, Dan, Obata, & Kitada, 2012) in various parts of the world further revised the taxonomic status of local mud crabs fundamentally based on morphological keys and/or molecular techniques based on Keenan et al.'s (1998) study. Unfortunately,the taxonomic studies of mud crabs along the Indian coasts, including the Bay of Bengal were remained obscure for long time. Therefore, a significant number of studies suggested the occurrence of three species

S. serrata

,

S. olivacea

, and

S. tranquebarica

along the coastal regions of India (Radhakrishnan & Smuel, 1982; Joel & Raj, 1983; Kathirval &Srinivasagam, 1992; Padate, Rivonker, & Anil, 2013; Trivedi & Vachhrajani, 2013). Afterwards, recent studies conclusively reported that only two species of mud crabs,

S. serrata

is the most abundant followed by the

S. olivacea

, are commonly present in Indian coastal waters (Mandal et al., 2014a,b; Balasubramanian et al., 2016). Although extensive studies were conducted in the Indian coast including the Bay of Bengal,the knowledge about the species diversity and population stock structure of mud crabs remains inadequate in case of Bangladesh. For the long time, the mud crab species found in the coastal region of Bangladesh has always been considered as

S. serrata

in all grey and scientific literatures without proper taxonomic identification of the species. Two recent studies reported that the common mud crab species which is being caught from the coastal areas of Bangladesh is only

S. olivacea

, and S

.serrata

is not prevailing (Rouf, Shahriar, Sarower, & Ahsan, 2016; Sarower, Shahriar, Nakamura, Rouf, & Okade, 2017). Since some previous studies (e.g. Mandal et al., 2014a,b; Balasubramanian et al., 2016) reported more than one mud crab species along the Bay of Bengal areas in Indian coastal regions, further intensive and reliable studies are necessary to confirm their diversity along the coastal areas of Bangladesh.The study of morphological characters with the objective of defining and characterizing populations, has a long tradition in ichthyology and is considered one of the simplest, most cost-effective and commonly used tools (Almeida, Almodovar, Nicola, & Elivera, 2008). Although genetics is the major underlying cause, recent studies have also proved that morphological characters can be varied with environmental factors including physico-chemical parameters of the water, habitat and substrate types (Cabral et al., 2003; Nahar, Siddik, Alam, & Chaklader,2015; Sharker, Siddik, Nahar, Shahjahan, & Faroque, 2015). Therefore,morphological studies can potentially contribute to the stock assessment of mud crab population for their better management and conservation along the coastal region of Bangladesh. Although many studies reported morphological differences among different species of mud crabs worldwide, information on population variability and differentiation of the morphological traits due to the local environmental adaption of a particular mud crab species is still rather limited. Therefore, the major aims of this study were the followings: (1) taxonomic identification of mud crabs species available along the coastal regions of Bangladesh, (2)obtaining information about the population differentiation of the mud crab species among the different coastal regions of Bangladesh, (3)identifying the best set of characters to delineate species identification and stock structure of mud crabs. For this purposes, multivariable analyses were performed using the twenty-four morphometric lengths and twenty seven morphometric ratios of two hundreds mud crab individuals from the three major coastal regions (i.e. Chittagong, Cox's Bazar and Bagerhat) of Bangladesh. The results of the present study would be considered as a preliminary step towards exploring the stock structure of

Scylla

genus for its sustainable development of aquaculture farming and management of the wild population across the coastal regions of Bangladesh.

2.Materials and methods

2.1.Sample collection

About two hundred fifty wild mud crab samples were collected from each of the three coastal districts along the eastern (Cox's Bazar and Chittagong) and western (Bagerhat) regions of the coast in the Bay of Bengal by the help of local fishermen from June 2017 to April 2018(Fig. 1). For Bagerhat region (adjacent to the Sundarbans mangrove forest), crabs were collected from the Mongla (22.48°N 89.61°E), while crabs were collected from the Maheshkahli Channel (21.55°N 91.95°E)for Cox's Bazar region's samples and from the Sandip (22.47°N 91.54°E)for Chittagong region's crab samples. After sample collection, the intensive morphological and morphometric studies were carried out at the Fisheries Oceanography Laboratory, Department of Marine Bioresource Science, Chattogram Veterinary and Animal Sciences University, Bangladesh.

Fig. 1.Map showing the collection sites of Scylla sp. from the three coastal areas of Bangladesh.

2.2.Initial classification

Before proceeding for intensive morphometric analysis, specimens were broadly assigned to

S. olivacea

and

S. serrata

based on the morphological characters provided by Estampador (1949), Kathirval and Srinivasagam (1992), Keenan et al. (1998) and Jirapunpipat et al.(2008). The shape and number of carpus spine was considered for primary identification between

S. serrata

and

S. olivacea

. Primarily,

S. olivacea

was identified by blunted outer and absent inner carpus spine(Fig. 2E) and

S. serrata

by prominent outer and blunted inner carpus spine (Fig. 2e). Subsequently,

S. olivacea

were more precisely identified by distinct morphological features of blunted frontal lobe (Fig. 2B), thick and slightly curved dactylus (Fig. 2C), blunted outer and inner propodus spine (Fig. 2D), absence of polygonal patterns on body and legs (Fig. 2F).In contrast,

S. serrata

were identified by distinct morphological features of bluntly pointed frontal lobe (Fig. 2b), elongated dactylus (Fig. 2c),prominent outer and inner propodus spine (Fig. 2d), presence of polygonal patterns on body and legs (Fig. 2f). For further taxonomic identification of crabs, the 1st and 2nd male gonopod were removed carefully using clean forceps, and studied under microscope (Optika microscope Italy, B-190 Series) to observe their structure. Afterwards,

S. olivacea

were further confirmed by a narrow mouth tip of 1st male gonopod (Fig. 2G) and wider biloped structure of 2nd male gonopod(Fig. 2H) and

S. serrata

was confirmed by wide and thick mouth tip of 1st male gonopod (Fig. 2g), and less wide biloped structure of 2nd male gonopod (Fig. 2h).

Fig. 2.Morphological comparison between Scylla olivacea (A—H) and Scylla serrata (a—h). S. olivacea were identified by distinct morphological features of full body shape (A), blunted frontal lobe (B), thick and slightly curved dactylus (C), blunted outer and inner propodus spine (D), blunted outer and absent inner carpus spine(E), absence of polygonal patterns on body and legs (F), narrow mouth tip of 1st male gonopod (G) and wider biloped structure of 2nd male gonopod (H). Scylla serrata were identified by distinct morphological features of full body shape (a), bluntly pointed frontal lobe (b), elongated dactylus (c), prominent outer and inner propodus spine (d), prominent outer and blunted inner carpus spine (e), presence of polygonal patterns on body and legs (f), wide mouth tip of 1st male gonopod (g),and wide biloped structure of 2nd male gonopod (h).

2.3.Morphometric analysis

About hundred similar sized (112.54 ±5.26) default specimen with no missing appendages or broken parts for each of

S. olivacaea

(Cox's Bazar =30, Chittagong =35; Bagerhat =35) and

S. serrata

(Cox's Bazar=30, Chittagong =30; Bagerhat =40) were selected for detail morphometric analysis by using multivariate approach. The male and female ratio were kept equal (1:1) to reduce the sexual dimorphism effects on the morphometric analysis. Twenty four morphometric characteristics include carapace width (CW), internal carapace width(ICW), carapace width at spine 8 (8CW), carapace length (CL), posterior width of carapace (PWC), 9th lateral spine height (LSH), frontal width(FW), frontal median spine height (FMSH), distance between frontal median spines (DFMS), distance between frontal lateral spines (DFLS),sternum width (SW), abdomen width (AW), dactylus length (DL), propodus length (PL), propodus width (PW), propodus depth (PD), inner propodus spine (ICS), outer propodus spine (IPS), inner carpus spine(ICS), outer carpus spine (OCS), merus length (ML), 5th pereiopod dactyl length (5 PL), 5th pereiopod dactyl width (5 PW), 3rd pereiopod merus length (3PML) were measured by using digital calipers to the nearest 0.01 mm. These characters were measured according to the diagram by Devi, Joseph, Mandal & Kor (2017) as well as 27 morphometric ratios of these length were also calculated as described by Keenan et al. (1998). All measurements were made by the same person.

2.4.Statistical analysis

Prior to the analysis, size effects were eliminated from the dataset for ensuring the morphological variations were attributed to body shape differences, and not to the relative sizes of the mud crabs (Hoq and Alam, 2018). Among the 24 morphometric measurements, 11 morphometric lengths (ICW, 8CW, CL, PWC, SW, DL, PL, PW, ML, 5PL and 5 PW)were found to have significant linear correlations with the CW of the collected mud crabs. Therefore, size-dependent variations from these 11 morphological measurements were eliminated by adapting an algometric method as suggested by Elliot, Haskard, and Koslow (1995):

where M is the original measurement, Mis the size adjusted measurement, CWis the carapace width of the mud crabs, CWis the overall mean of the mud crabs for all samples in each analysis, and b is estimated for each character from the observed data as the slope of the regression of log M on log CWusing all mud crabs from both groups.The results derived from the allometric method were confirmed by testing the significance of the correlation between transformed variables and carapace width (Turan, 1999).

The univariate analysis of variance (ANOVA) model was performed for each morphometric character to evaluate the significant differences among the two species (Supp. Table 1) and stock discrimination of

S. olivacaea

(Supp. Table 2) and

S. serrata

(Supp. Table 3) (Zar, 1984).Wilks' lambda was used to compare the differences among all individuals of the two species and three collection sites of each species.The morphometric characters which showed significant variation (P

＜

0.05) were only used for obtaining the stable outcome from the multivariate analysis. In the present study, Principal Component Analysis(PCA), Canonical Variates Analysis (CVA), and cluster analysis using Euclidean distance method (CA) were employed to confirm the existence of two species and stock discrimination of the three population for each species. PCA helps in morphometric data reduction, in decreasing the redundancy among the variables (Anvarifer, Farahmand, Silva, Bastop,& AnvariFar, 2013; Mousavi-Sabet & Anvarifar, 2013; Veasey, Schammas, Vencovsky, Martins, & Bandel, 2001), and in extracting a number of independent variables for population differentiation. Therefore, we used this PCA to reduce the number of selected morphological characteristics to a few composite measure of morphological traits to explore the variation between the species according to selected regions and measured characteristics. The PCAs were performed using the ‘Facto-MineR' package (Sebastien, Josse, & Husson, 2008) of R, version 3.5.2(R development core team, 2018) to extrapolate the specimens' distribution patterns among the three selected regions. We used only the first and second PCAs in all cases as they explained most of the variability.The CVA was performed by using ‘MASS' package of R (Venables &Ripley, 2002). All graphs were made by using ‘ggplot2' package(Wickham, 2009). The LDFA was also used to calculate the percentage of correctly classified (PCC) mud crabs according to their species and the three regions of each species. A cross validation using PCC was done to estimate the expected actual error rates of the classification functions. As a complement, morphometric distances among the individuals of the two species and three regions of each species were inferred to cluster analysis (Veasey et al., 2001) by adopting the Euclidean distance as a measure of dissimilarity and the UPGMA (unweighted pair group method with arithmetical average) as the clustering algorithm. Cluster analysis was done using the ‘dendextend' package of R (Galili, 2015).

Table 1Factors loading of different morphometric measurements and their ratios by principal component analysis between two species of mud crabs Scylla olivacea and Scylla serrata populations collected from the coastal regions of Bangladesh.

Table 2Classification of Scylla olivacea and Scylla serrata into their original species group using classification matrix of the DFA based on the different morphometric lengths and their ratios data.

Table 3Factors loading of different morphometric measurements and their ratios by principal component analysis among different populations of Scylla olivacea collected from the three coastal regions of Bangladesh.

3.Results

3.1.Species diversity of mud crabs along coastal areas of Bangladesh

Based on the taxonomic keys of Estampador (1949), Kathirval and Srinivasagam (1992), Keenan et al. (1998) and Jirapunpipat et al.(2008), the collected mud crabs were broadly assigned to

S. olivacea

and

S. serrata

as described in Fig. 2. The results showed that

S. olivacea

is the dominant (82.4%) mud crab species, while a meagre number of

S. serrata

(17.6%) is also found along the coastal regions of Bangladesh.Among the collected mud crabs specimens, 81.5% was identified as

S. olivacea

and 18.5% as

S. serrata

in Chittagong region, 79.6% was identified as

S. olivacea

and 20.4% as

S. serrata

in Cox's Bazar region,and 86.1% was identified as

S. olivacea

and 13.9% as

S. serrata

in Bagerhat region (Fig. 3). However, we did not observe the presence of

S. tranquebarica

and

S paramamosain

in our collected mud crab specimens from the coastal regions of Bangladesh. Although it is well known that the female and male specimens of the crustaceans have some morphological differences, but morphometric measurements used in this study by sexes were not significant (P

＞

0.05) for the both mud crab species, demonstrating a negligible effect of sex on observed variations.Therefore, data for both sexes were pooled for all subsequent analyses.The ANOVA model showed that 8 measurements (FMSH, DFLS, AW, ICS,OCS, IPS, OPS, 3PML) out of 24 morphometric lengths, and 13 data(FW/ICW, PWC/FW, FMSH/FW, FMSH/DFMS, AW/SW, PL/ICW,IPS/PL, OPS/PL, IPS/OPS, ICS/PL, OCS/PL, ICS/OCS, 3PML/ICW) out of 27 morphometric ratios were significantly differed to varying degree between the

S. olivacea

and

S. serrata

(Supp. Table 1). These significant traits were used for further multivariate (CVA, PCA, LDFA and CA) analyses. Under these circumstances, the N:P ratio was 9.52 (200/21) that revealed mud crabs samples size were adequate. The value of Kaiser-Meyer-Olkin (KMO) for overall matrix is 0.87, and the Bartlett's Test of sphericity is significant (P ≤0.01). The results of KMO and Bartlett's suggest that the sampled data is appropriate to proceed with a factor analysis procedure.

Fig. 3.Bar graph showing the species composition (%) of mud crab identified based on the morphological features as described in Fig. 1 from the three coastal regions of Bangladesh. Scylla olivacea is the dominant species representing over 82% of the mud crabs along the coastal region of Bangladesh.

In order to determine which morphometric lengths and ratios made most effectively differentiates between the two species of the mud crabs,the contributions of variables to principal components (PC) were examined by the R statistical software (R development core team, 2018).The PCA of 8 morphometric lengths and 13 ratios data extracted six factors with eigenvalues

＞

1, explaining 86.82% of the variance between two species of mud crabs. The first principal component (PC1) accounted for 37.75% of the variation and the second principal component(PC2) for 17.52% (Table 1). The most significant loadings variables on PC1 were OPS/PL, OPS, ICS/PL, ICS, OCS, OCS/PL, IPS/PL, IPS, AW/SW and ICS/OCS, and on PC2 were FMSH, 3PML/ICW, FW/ICW,FMSH/DFMS, FMSH/FW, IPS/PL, OCS/PL and ICS/OCS (Table 1). PCA variables also confirmed that these loading variables most effectively differentiated the

S. olivacea

population from the

S. serrata

(Fig. 4A).Visual investigation of plotted PC1 and PC2 scores showed that the

S. olivacea

and

S. serrata

samples were grouped into two clear multivariate spaces (Fig. 4A). The existence of two mud crab species along the coastal region of Bangladesh was further confirmed by the CVA analysis(Fig. 4B). In this analysis, coordinates of the canonical variates function 1 showed a degree of complete separation between

S. olivacea

and

S. serrata

samples (Fig. 4B). The linear discriminant analysis revealed that 100 of 100

S. olivacea

(100%) and 100 of 100

S. serrata

(100%) were original correctly classified in their respective grouped that indicate a perfect correct classification of the two species of mud crabs into their original group (Table 2). For the both

S. olivacea

and

S. serrata

, the cross-validation testing procedure was exactly given the same results of the PCC results. Finally, our results demonstrated that morphometrically there are two species of mud crabs in which S

. olivacea

is the dominant ones and

S. serrata

is comparatively less abundant along the coastal regions of Bangladesh.

Fig. 4.Principal component analysis (PCA) and canonical variates analysis (CVA) of the morphometric data depicting the existence of two mud crab species Scylla olivacea and Scylla serrata along the coastal regions of Bangladesh. A: Biplot of PCA is shown based on the significantly differed variables (8 morphometric lengths and 13 ratios) of the S. olivacea and S. serrata. B: Coordinate plot of the canonical variate 1 of the significantly differed variables (8 morphometric lengths and 13 ratios) of the S. olivacea and S. serrata.

3.2.Stock structure of mud crabs species along coastal region of Bangladesh

The ANOVA model showed that most of 22 out of 24 morphometric lengths and 24 out of 27 morphometric ratios in

S. olivacea

(Supp.Table 2), and 23 out of 24 morphometric lengths and 21 out of 27 morphometric ratios in

S. serrata

(Supp. Table 3) were significantly differed to varying degree among the three population (Chittagong,Cox's Bazar and Bagerhat). These significant traits were used for further multivariate (PCA, CVA, LDFA and CA) analysis. The PCs of the most significant morphometric lengths and ratios data of the both mud crab species extracted ten factors with eigenvalues

＞

1, explaining 88.2% of the variance among three population of

S. olivacea

and 88.7% of the variance among three population

S. serrata

. The PC1 accounted for 33.41% and 32.61% of the variation, while the PC2 for 16.21% and 15.98% for the

S. olivacea

(Table 3) and

S. serrata

(Table 4), respectively.In

S. olivacea

, the most significant loadings of variables on PC1 were PD,PL, PW ×PD/PL, PL/ICW, PD/PL, PW, ICW, OCS/PL, DL and OPS/PL and on PC2 were FMSH/FW, FMSH/DFMS, FMSH, 3PML, IPS/OPS, 5PL,PWC and 3PML/ICW (Table 3). Whereas, in

S. serrata

, the most significant loadings of variables on PC1 were OCS/PL, PW ×PD/PL, PD, OPS/PL, ML, PL, AW, IPS/PL, OPS and OCS and on PC2 were FW, 3PML, FW/ICW, 3PML/ICW, LSH and LSH/ICW (Table 4). The contribution of these loading variables also confirmed through variable PCA analysis for the population differentiation in

S. olivacea

and

S. serrata

(Fig. 5). Biplot of the plotted PC1 and PC2 scores showed three populations of the both

S. olivacea

and

S. serrata

were grouped into three multivariate spaces with a moderate overlapping between Cox's Bazar and Chittagong population and a minor overlapping between Chittagong and Bagerhat population (Fig. 5). The scores of the two canonical variates for each population of

S. olivacea

and

S. serrata

demonstrated that mud crabs individuals of each species grouped into three distinct areas with almost no overlapping among the three populations (Fig. 6). In morphometric traits linear discriminant function analysis, the average of PCC of the three locations was 100% for

S. olivacea

(Table 5) and 99% for

S. serrata

populations (Table 6), only one individual from Cox's Bazar were mistakenly grouped in Chittagong population for

S. serrata

(Table 6). In cross validated group, LDFA revealed that average PCC of the three locations was 97% for

S. olivacea

and 95% for

S. serrata

populations,indicating a highly correct classification of the specimen into their original populations.

Table 4Factors loading of different morphometric measurements and their ratios by principal component analysis among different populations of Scylla serrata collected from the three coastal regions of Bangladesh.

Fig. 5.Principal component analysis of morphometric data of Scylla olivacea and Scylla serrata populations obtained from the three coastal regions of Bangladesh.Biplots of principal component analysis are shown based on the different morphometric lengths and ratios of the three population of S. olivacea (A—B) and S. serrata(C—D). In left side image panels, ratios are written by putting dot between two lengths.

Fig. 6.Canonical Variates Analysis (CVA) of morphometric data of Scylla olivacea and Scylla serrata populations obtained from the three coastal regions of Bangladesh. Biplots of sample centroids of the canonical variates scores are shown based on the morphometric length and ratios of the three population of S. olivacea(A—B) and S. serrata (C—D).

Table 5Classification of Scylla olivacea into their original population using classification matrix of the DFA based on morphometric measurements of different lengths and ratios data.

Table 6Classification of Scylla serrata into their original population using classification matrix of the DFA based on morphometric measurements of different lengths and ratios data.

In consistent with PCA, CVA and LDFA results, clustering analysis based on Euclidean distances between the groups of centroids using an UPGMA also displayed two main clusters, one for

S. olivacea

and another for

S. serrata

(Fig. 7). Three population of the each clusters of

S. olivacea

and

S. serrata

further segregated from each other into two distinct clusters, mud crabs collected from the Bagerhat area in one cluster while the mud crabs collected from the Chittagong and Cox's Bazar region in another clusters (Fig. 7). Based on the PCA, CVA and cluster analysis, our results demonstrated that each sampling sites of Chittagong, Cox's Bazar and Bagerhat represents an independent stocks for the both species of

S. olivacea

and

S. serrata.

Fig. 7.Dendrogram derived from cluster analyses of morphometric measurements on the basis of Euclidean distance for Scylla olivacea and Scylla serrata population collected from the coastal regions of Chittagong, Cox's Bazar and Bagerhat districts of Bangladesh. In the figure, collection areas are abbreviated as Ctg for Chittagong, Cox for Cox's Bazar and Bag for Bagerhat.

4.Discussion

4.1.Taxonomic confirmation of mud crab species along the coastal areas of Bangladesh

Due to the overlapping morphological and morphometric traits among the species, identification of

Scylla

species have led to much confusion. There have been only two main and widely referred groups working on the mud crab species classification. Estampador (1949)classified mud crabs from the Philippines, based on gametogenesis and external morphology, into three species

S. serrata, S. oceanica, S. tranquebarica,

and a variety of

S. serrata

;

S. serrata

var.

paramamosain.

On the other hand, Keenan et al. (1998) classified mud crabs based on the external morphology, genetic variations and multivariate analysis of morphometric characters and suggested that crabs in the genus

Scylla

should be classified up to four species, such as

S. olivacea, S. paramamosain

,

S. tranquebarica,

and

S. serrata.

Among these four mud crab species, we primarily identified two mud crab species,

S. olivacea

and

S. serrata

, along the coastal region of Bangladesh. In our study,

S. olivacea

was identified by blunted frontal lobe, thick and slightly curved dactylus, blunted outer and inner propodus spine, blunted outer and absent inner carpus spine, absence of polygonal patterns on body and legs, narrow mouth tip of 1st male gonopod, and wider biloped structure of 2nd male gonopod (Estampador, 1949; Kathirval & Srinivasagam, 1992; Keenan et al., 1998; Jirapunpipat et al., 2008). In contrast,

S. serrata

was identified by bluntly pointed frontal lobe, elongated dactylus, prominent outer and inner propodus spine, prominent outer and blunted inner carpus spine, presence of polygonal patterns on body and legs, wide mouth tip of 1st male gonopod, and wide biloped structure of 2nd male gonopod (Estampador, 1949, Kathirval & Srinivasagam, 1992; Keenan et al., 1998; Jirapunpipat et al., 2008). Recent studies reported that common mud crab species which is being caught from the coastal areas of Bangladesh is

S. olivacea

instead of S

. serrata

(Rouf et al., 2016; Sarowar et al., 2017). In our study, we also observed that

S. olivacea

is the dominant species (82.4%) along the coastal region of Bangladesh. However, S

. serrata

was also exist comparatively at less proportion (17.6%). In consistent to our findings,

S. serrata

and

S. olivacea

have been reported from the neighboring countries along the Bay of Bengal in India (Trivedi & Vachhrajani, 2013; Mandal et al.,2014a,b) and also in Thailand (Jirapunpipat et al., 2008). In another morphometric study, three mud crabs species

S. serrata

,

S. olivacea

and

S. paramamosain

were identified from the coastal water of Bangladesh(Hoq & Alam, 2018). However, we did not observe the presence of

S paramamosain

in our collected mud crab specimens from the coastal regions of Bangladesh. Although,

S. tranquebarica

was reported to occur,

S. paramamosain

was not reported to occur along the Indian coast including the Bay of Bengal region (Joel & Raj, 1983; Kathirval & Srinivasagam, 1992; Padate et al., 2013; Radhakrishnan & Smuel, 1982;Trivedi & Vachhrajani, 2013).Multivariate analysis of morphometric characters has been shown to be a rapid and effective technique in providing an insight into the discrimination of many fish and crustacean species including mud crabs(Overton et al., 1997). To further validate our primary classification, we have conducted multivariate analysis by using PCA, CVA and LDFA statistical tools. In consistent to our preliminary classification, PCA analysis by R statistical software showed that the

S. olivacea

and

S. serrata

samples were grouped into two clear multivariate spaces,confirming the presence of two mud crab species along the coastal regions of Bangladesh. The multivariate analysis of morphometric characters of mud crabs of the genus

Scylla

also revealed significant differences among the two species. Moreover, PCA analysis gave information to know which morphometric lengths and ratios were most effectively differentiates among the two species of the mud crabs. In our study, we observed that among the morphometric length ICS, OPS, OCS,IPS, FMSH and DFLS, and morphometric ratios OPS/PL, ICS/PL,OCS/PL, AW/SW, FW/ICW, FMSH/FW, PL/ICW, FMSH/DFMS,ICS/OCS and 3PML/CW most significantly contributed to the differentiation between two species (See Table 1). Many of these morphometric lengths and ratios have been extensively used for the taxonomic identification of the mud crabs. Keenan et al. (1998) observed five morphometric length including ICS, OCS, FMSH, FW and ICW and seven morphometric ratios including ICS/OCS, FMSH/FW, FW/ICW, ML/PL,AW/SW, PL/ICW and IPS/PL that could distinguish between the

Scylla

species. Moreover, mean values of the morphometric ratios of ICS/OCS,FMSH/FW and FW/ICW are mostly used to differentiate the

Scylla

species (Mandal et al., 2014a,b; Rouf et al., 2016; Sarowar et al., 2017).Besides PCA analysis, CVA analysis also demonstrated that the coordinates of the two species of mud crabs also shown complete separation between

S. olivacea

and

S. serrata

samples (See Fig. 4B). The LDFA analysis also revealed that 100% of the

S. olivacea

and

S. serrata

were correctly classified into their respective groups. Therefore, from the multivariate morphometric investigation study, it can be concluded that two species of mud crabs, S

. olivacea

and

S. serrata

, are available along the coastal waters of Bangladesh.

4.2.Stock discrimination of the mud crab species along coastal areas of Bangladesh

Morphometric differences among the different stocks of a particular mud crab species are expected, because they are geographically separated and may have originated from different ancestors. However, information on population variability and differentiation of the morphological traits due to local environmental adaption of a particular species of mud crab species is still rather limited. Therefore, other objectives of our study were obtaining information about the population differentiation of the mud crab species, and identifying the best set of characters to delineate species stock structure of particular species of mud crab. In order to attain our objectives, we used multivariate analysis using PCA, CVA and LDFA statistical tool as they could be suitable methods to differentiate different stocks of the same species (Karakousis,Triantaphyllidis, & Economidis, 1991). Our study results demonstrated that there is significant phenotypic variation among the three studied populations (Chittagong, Cox's Bazar and Bagerhat) of

S. olivacea

and

S. serrata

. Previous studies showed population morphometric discrimination changes according to location, environment and ecological changes (Cabral et al., 2003; Chaklader et al., 2016; Díaz de Astarloa et al., 2011). Environmental parameters influences morphometric characters which were considered by scientists in fish population segregation already (e.g. Cardin, 2000; Swain & Foote, 1999; Turan,2000). At first, the discrimination was carried out by PCA, where pictorial analysis of plotted PC1 and PC2 scores for every specimen showed that three population of the both

S. olivacea

and

S. serrata

were grouped into three multivariate spaces with a significant overlapping between Cox's Bazar and Chittagong population and a minor overlapping between Chittagong and Bagerhat population (See Fig. 5). In consistent with the PCA analysis, clustering analysis based on Euclidean distances using an UPGMA also displayed that mud crabs collected from the Bagerhat area in one cluster while the mud crabs collected from the Chittagong and Cox's Bazar region in another clusters with significant distances between them (See Fig. 7). This type of discrimination of the both species of

S. olivacea

and

S. serrata

is might be due to geographic location as Chittagong is geographically closer to the Cox's Bazar, and also located in the middle between the Cox's Bazar and Bagerhat (See Fig. 1). In consistent with our results, Vatandoust, Mousavi-Sabet,Mansour, AnvariFar, and Heidari (2015) noted some ranges of overlap in multivariate analysis during comparing the morphometric characteristics between two groups of Caspian lamprey,

Caspiomyzon wagneri

.Similarly, Siddik, Hanif, Chaklader, Nahar, and Fotedar (2016) also noted significant overlapping during multivariate morphometric investigation to delineate stock structure of gangetic whiting,

Sillaginopsis panijus

. In our study, we observed that morphometric lengths PD, PL,PW, ICW, DL, FMSH, 3PML, 5PL and PWC in

S. olivacea

and PD, ML, PL,AW, OPS, OCS, FW, 3PML and LSH in

S. serrata

were contributed most to delineate species stock structure among the three locations of these two species. For the morphometric ratios, PW ×PD/PL, PL/ICW, PD/PL,OCS/PL, OPS/PL, FMSH/FW, FMSH/DFMS, IPS/OPS, and 3PML/ICW in

S. olivacea

and OCS/PL, PW ×PD/PL, OPS/PL, IPS/PL, FW/ICW,3PML/ICW, and LSH/ICW in

S. serrata

were contributed most to delineate species stock structure among the three locations.In morphometric traits LDFA analysis, the average PCC of the original classification and cross validated groups were 100% and 97% for

S. olivacea

and 99% and 95% for

S. serrata

populations, indicating a highly correct classification of the specimen into their original populations. The scores of the two canonical variables for each population of

S. olivacea

and

S. serrata

demonstrated that mud crabs individuals of each species grouped into three distinct areas with almost no overlapping among the three populations of Chittagong, Cox's Bazar and Bagerhat (Fig. 6). Consistently, morphometric analysis showed a significant difference between the crabs' morphometric characteristic and their populations from the three locations of sampling of the male swimming crab (

Portunus pelagicus

) from the East Sahul Shelf, Indonesia(Hidayani, Trijuno, Fujaya, Alimuddin, & Umar, 2018). In other crustaceans, Konana, Adépo-Gourèneb, Ouattaraa, Nyingyc, and Gourènea(2010) applied PCA on the populations of freshwater shrimp

Macrobrachium vollenhovenii

collecting from Côte d'Ivoire Rivers and reported notable morphometric variation due to distance and geographical location of rivers. Paugy and Lévêque (1999) also showed that populations of

Macrobrachium vollenhovenii

originating from different geographical areas were morphologically different. Apart from curstaceans, Siddik et al. (2016) reported that the inter-population variation gangetic whiting,

Sillaginopsis panijus

can be attributed due to distance between rivers, separate geographical locations and environmental constrains habituated by each population. However, it is very difficult to explain the actual causes of morphological difference among the different population of mud crabs and other fish species (Poulet,Berrebi, Crivelli, Lek, & Argillier, 2004). In general, it has been suggested that morphological characteristics of aquatic organisms are determined by genetic, environment, and the interaction between them(Pinheiro, Teixeira, Rego, Marques, & Cabral, 2005; Poulet et al., 2004;Swain & Foote, 1999). There is a probability that the observed morphological discrimination among the three population of

S. olivacea

and

S. serrata

might be due to the genetic difference as they are collected from the geographically isolated area and admixture may not be happened due to their slow movement patterns (Alberts-Hubatsch et al.,2016). In general, crabs that live in enclosed habitats normally do not move more than 1 km (Hill, 1975; Perrine, 1978; Hyland, Hill, & Lee,1984; Bonine, Bjorkstedt, Ewel, & Palik, 2008) and those are found in open environments like intertidal flats in open bays show larger movement (average 3.7 km) (Hyland et al., 1984). The influences of local environmental parameters and habitat types on the variation of morphometric characters are very important for mud crabs, and are already well discussed by several authors in the course of fish population segregation (Ferrito, Mannino, Pappalardo, & Tigano, 2007; Swain &Foote, 1999). Therefore, the variation among the three stocks of

S. olivacea

and

S. serrata

might be due to isolation by distance and different environmental conditions such as variation in food abundance and abiotic characteristics like tidal fluctuation, salinity, turbidity, dissolved oxygen and other factors.

5.Conclusion

The present study showed that morphometrically there are two species of mud crabs in which S

. olivacea

is the dominant ones, whereas

S. serrata

is comparatively a meagre species along the coastal regions of Bangladesh. Furthermore, multivariable analysis also demonstrated that each sampling sites of Chittagong, Cox's Bazar and Bagerhat represents an independent stocks for the both species of

S. olivacea

and

S. serrata

. In order to develop appropriate conservation plans and sustainable management of mud crab fishery, the findings of the present study would serve as a baseline information for the stock management of mud crab population along the coastal regions of Bangladesh. Obviously morphometric data cannot be expected to give all the answers alone.Besides genetical factors, the morphometric variations among different stocks of S

. olivacea

and

S. serrata

are controlled by many external factors, such as diet, habitat and other environmental factors. Therefore,further studies are needed to understand the role of these external factors in the morphometric variation of mud crab species in near future. A detailed study involving the molecular genetics may further confirm the present findings unambiguously. These ambiguities are, therefore, the next target of our research.

CRediT authorship contribution statement

Md Asaduzzaman: Conceptualization, Formal analysis, Writing -original draft, Funding acquisition. Ismat Jahan: Methodology. Aysha Rahi Noor: Writing - original draft. Md Moudud Islam: Writing -original draft. Md Moshiur Rahman: Formal analysis, Writing - review& editing.

Declaration of competing interest

The authors declare that there is no conflicts of interest.

Acknowledgement

This study was funded by the Chittagong Veterinary and Animal Sciences University (CVASU) research budget and University Grants Commission (UGC) of Bangladesh. Supports from the masters' students and supporting staffs of the Department of Marine Bioresource Science(MBS) of CVASU are gratefully appreciated.

Appendix A.Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.aaf.2020.03.010.