Stock Structure Analysis of Declining Shovel Nosed Lobster Thenus Unimaculatus (Burton and Davie, 2007)for Effective Management and Conservation Along the Indian Coast

Thenusunimaculatus (Burton and Davie 2007)is one of the most important commercially exploited lobster species in India distributed throughout the coast. The declining trends and collapse of the sand lobster �shery has been reported from the northwest coast of India. A number of hypotheses have been proposed to explain the decline, but a lack of basic information on population demographics inhibits hypothesis testing. In this study, morphometric stock structure analysis of Thenusunimaculatus was attempted to discriminate the spawning population collected from �ve locations (673 specimens) during 2017-2019 along the Indian coast. Data was subjected to allometric transformation to remove size effect. The transformed data was subjected to multivariate analysis, such as Principal Component Analysis (PCA) followed by linear discriminant analysis (LDA) using R software. The �rst eight PCs cumulatively explained 73.97% of the total variance,wherein the �rst two PCs together explained 45.50% of the total variance. The accuracy (ACC) and misclassi�cation rate (MR) of the linear discriminant analysis (LDA) model performed to optimize separation among different sampling locations were 0.70 and 0.30, respectively, which indicates clear overlapping of the stocks among different sampling locations.The accuracy (ACC) and misclassi�cation rate (MR) of the linear LDA model developed to discriminatebetween the coasts were 0.74 and 0.26,respectively, indicating no apparent differentiation in the samples between the coasts. The results of this study revealed the presence of asingle spawning stock of shovel nosed lobster, T. unimaculatus along the Indian coast.The information will aid in the development of better holistic management strategy for the conservation of this declining resource.


Introduction
Lobsters form one of the most valuable crustacean resources, which have been exploited commercially for the past four decades.They have a good market and great demand around the world and thereby fetch foreign exchange for the country (Radhakrishnan and Thangaraja 2008).The diverse habitats and the physical environment of the southern coast (both the west and east coasts) of India favor the settlement of a wide variety of lobster species (Radkhakrishnanet al. 2019).Marine lobsters are members of the suborder MacruraReptantia with 4 infraorders, 6 families, 54 genera, and 260 extant species (including 4 subspecies) (Chan2019).The lobster fauna of India is so diverse that 38 species belonging to ve families and three infraorders (Astacidea, Achelata, and Polychelida) are distributed in the seas surrounding the Indian subcontinent.Thenusis the only genus in 7 scyllarid genera that is economically signi cant (Jones 1990).The scyllarid lobster, Thenusunimaculatus, described earlier as Thenusorientalis (Chhapgar and Deshmukh 1964) is commonly known as sand lobster, slipper lobster, or shovel-nosed lobster.The species of shovel nosed lobsters available in Indian waters has been con rmed as Thenus unimaculatus (Jeena et al. 2011).It is one of the important commercially exploited lobster species in India.It is distributed almost throughout the Indian coast and forms an important shery along the north-west and south-east coasts (Holthuis1991).The declining trend in catches, as well as the collapse of the shery in some localities (Deshmukh 2001;Jeena et al. 2015;Subramanian 2004) necessitated an assessment of the stock structure to manage this resource sustainably.
From sheries management perspective, a stock is described as an intraspeci c assemblage of n shes or shell shes with spatial or temporal stability, that randomly mate with members of each group having identical growth, mortality, and reproductive rates (Hilborn and Walters 1992).Identi cation of stocks with variable life history attributes such as growth, mortality and reproductive potentials are considered being important for understanding population dynamics and developing sustainable resource management strategy.Therefore, stock structure analysis is an essential technique to identify and understand the difference among the individuals of the same species exhibiting different growth, mortality and maturation characteristics so that they can be grouped into separate management units for the effective and e cient implementation of management interventions (Begg et al. 1999; Turan et al. 2005).Stock identi cation is an interdisciplinary eld that concerns the identi cation of self-sustaining components within natural populations (Cadrin et al. 2005).There are several techniques for identi cation of stocks like meristic, morphometric, parasite as natural tags, otolith study and several molecular markers (Awasthiet al. 2015).It has been observed that variations in morphometric traits do not always necessarily caused due to genetic variations, as variations in environmental parameters during critical developmental stages of the animal can also signi cantly in uence the morphometric traits (Cadrin 2000 Morphometric and meristic methods remain the simplest and most direct way among methods of species identi cation.Recently, truss networking system has evolved as a new system for morphometric measurement.Truss network systems constructed with the help of landmark points are powerful tools for stock identi cation.The methodology predicts the measurement of across-body distances connecting two morphological landmarks from a sequential series of connected polygons.This type of landmarkbased technique using geometrical morphometrics imposes no restrictions on the direction of variation and localization of shape changes and therefore is highly effective in capturing information regarding the shape of an organism (Cavalcanti et al. 1999).Subsequently, several multivariate techniques, such as principal component analysis and discriminant analysis,are used to discriminate the stocks.In the present study, attempts have been made to describe the stock structure of scyllarid lobster, Thenusunimaculatusalong Indian subcontinent through morphometric approach so that appropriate management strategy could be developed to rebuild this declining crustacean resource.

Sampling
A total of 673 specimens of Thenusunimaculatusconsisting of 305 males and 368 females were collected from ve major shing harbours along the Indian coast.The specimens were identi ed as Thenusunimaculatus according to Burton and Davie (2007).The sampling locations were, Sakthikulangara (SWS) along the south-west coast, Veraval (SWV) along the north-west coast, Chennai (SEC) and Nagapattianam (SEN) along the south-east coast, and Vishakapatnam (SEZ) along the northeast coast of India (Fig. 1).The descriptive locality, sex ratio and sample size from each location are shown in Table 1.All the specimens procured were in mature condition representing the spawning stock.Breeding season of Thenusunimaculatus was selected for the collection of samples to ensure that they belonged to their parent population.The specimens from the west and east coasts were obtained from August to May and June to March, respectively, for a period of 2 years from 2017 to 2019.Male and female specimens were separated according to their sexual dimorphic features.In mature females, the leaf-like endopods of the abdominal pleopods bear long ovigerous setae which are used for attaching spawned eggs until they are hatched, whilst in juvenile females, the pleopods are devoid of setae.Mature males possess a small club-shaped process midway along the inner margin of the endopod of the rst pair of pleopods (Kizhakudan 2014).The total length (TL) is measured as the distance between the notch in the carapace (anterior region) and the posterior margin of the telson and carapace length (CL) is measured as the distance between the notch and the posterior margin of the carapace.Immature and egg-bearing (berried) females were totally excluded while only matured specimens (carapace length ≥6.0 cm) were included in the study.

Digitization of samples and measurement of truss distances
The specimens of T. unimaculatus were collected and preserved on site in an insulated ice box and brought to the laboratory.The specimens were thoroughly cleaned with running tap water, drained, and wiped with absorbent paper.The specimens were placed on a at platform pinned with graph paper for calibrating the coordinates of the digital images.The distances between the vertical and horizontal grids of graph paper were used in calibrating the coordinates covering an area of 1cm 2 .The lobster specimens were then digitally captured with a camera (Canon G-15, Tokyo, Japan) xed on a tripod.Further, all the specimens were labelled with a speci c code marking the identity of the specimen.The extraction of truss distances from the digital images of specimens was conducted using a linear combination of three software platforms: tpsutil, tpsDig2 v2.1 (Rohlf, 2006), and Paleontological Statistics (PAST) (Hammer et al. 2001).
Each image was acquired by placing the scale beside it, to have uniformity in all the measurements which were scaled up with tpsDig2 software by employing the millimeter grid on the graph paper.A truss box of 43 landmark points was selected to provide a homogeneous coverage of the basic shape of sand lobster (Strauss and Bookstein 1982) (Fig. 2).All measurements were transferred to a spreadsheet le (Excel 2007), and the X-Y coordinate data was transformed into linear distances by computer for subsequent analysis (Turan 1999).Sampling locations, coasts and sexes were used as the class variables to test for signi cant differences in morphometric characters, if any, between male and female T. unimaculatus.

Data analysis and statistical study
All the morphometric truss measurements were log transformed and were tested for normality assumption and outliers were removed before further analysis.Multivariate analysis of covariances (MANCOVA) was performed to examine any signi cant differences between the stocks using sampling locations, coasts and sexes as factors, log transformed carapace length (CL) as a covariate and other log-transformed morphometric data as dependent variables.This statistical model was used to the test the effect covariate (CL) on other dependent morphometric variables and to test interaction effect between factors (sampling locations, coasts and sexes) with covariate (CL).Any size-dependent variation was corrected by adapting an allometric method suggested by Elliott et al. (1995) using the formula Where, M trans is thetransformed morphometric measurement, log M is the log-transformed original measurement, log CL is thelog-transformed standard length (i.e., carapace length) of each specimen, log CL mean is the arithmetic mean of the carapace length of the population and parameter β was estimated as the slope by regressing the values of log M against log CL.The correlation coe cient between the transformedvariables and the carapace length of the lobsterbefore and after transformation wasestimated to examine whether the transformed data were successfulin eliminating the size effect (Khan et al. 2012).A univariate ANOVAwas performed consequent to allometric correction to testwhether there was any statistically signi cant variance for eachmorphometric measurement among the sampling sites, between coasts and sexes.
Principal component analysis (PCA) was used for morphometric data reduction (Veasey et al. 2001) Bartlett's test of sphericityto ascertain the requisite correlation among the variables, determinant test to check multicollinearity and Kaiser-Meyer-Olkin tests to measure the sample adequacy were performed to ascertain that the data follow the assumptions for PCA.The signi cantly contributing principal components were selected based on Kaiser selectioncriteria by scrutinizing the Eigen values using a scree plot (Kaiser 1960).The variables with signi cant loadings to rst two principal components (PC1 and PC2) were selected to examine and describe any possible variations in the morphometric variables among locations and between coasts and sexes.
The coordinate scores of the signi cant principal components (PCs) were used forDiscriminant function analysis (DFA) to discriminate the effects of variables into known groups and to explore the effectiveness of variables in predicting factor-wise different groups for locations, coasts and sexes (Tomovic and The ndings of the univariate test (ANOVA) performed to compare the morphometric measurement among the sampling locations, between the coasts (East vs. West), and between the sexes (Males vs. Females) are presented in Table 3.All the morphometric measurements showed a signi cantdifference (p<0.05)among the sampling locations.Most of the morphometric measurements except for TL, CW, T26, T78, T1920, T2122, T3132, T3233, T3743, and T3839 showed signi cant difference (p<0.05) when compared between the coasts (East vs. West).Morphometric measurements except for T16, T34, T45, T712, T812, T910, T1011, T1920, T2122, T2324, T2526, T2728, T2930, T3132, and T3437 were found to be signi cantly different when compared between the sexes.The correlation of individual morphometric measurements with covariate (CL) before transformation and after transformation of morphometric measurements is shown in Fig. 3.All the morphometric measurements showed high to very high level of signi cant correlation (p<0.01) with CL.The correlation coe cient (r) varies from a minimum of 0.53 (for T3437) to a maximum of 0.95 (for CW and T410).However, after allometric correction, the correlation coe cient was drastically reduced from a minimum of 0.001 (for T3334) to a maximum of 23 (for T46).After allometric transformation, most of the morphometric measurements except for T26, T34, T35, T45, T46, T1011, T1112, T1920, T2122, T2324, T2526, T2728, T2930 were not found to be signi cantly (p>0.01)correlated with CL.
The data was found suitable for PCA as it passed the Bartlett's test of sphericity test (χ 2 = 18938.07,df=496, p<0.01), multicollinearity test (determinant=3.5x10−13 ) and Kaiser-Meyer-Olkin tests (overall MSA=0.81) to measure the sample adequacy.Principal component analysis (PCA) indicated that rst eight principal components (PCs)individually explained the variance greater than average variance based on Kaiser Selectioncriteria (Eigenvalue >1.0).The rst eight PCs cumulatively explained 73.97% of the total variance,wherein the rst two PCs together explained 45.50% of the total variance (Fig. 4, Table 4).
The respective loadings of the morphometric measurements to the rst eight principal components along with the eigenvalues, explained variance (%) and cumulative explained variance (%) is shown in Table 4.
Morphometric measurements having signi cant loadings for PC1 and PC2 are shown in variables PCA biplot (Fig. 5).
Top ten morphometric measurements with the highest signi cant loadings on PC1 and PC2 are T2122, TL, T1920, T2324, T410, T2526, T712, CW, T35 and T911.The corresponding morphometric measurements have been highlighted and presented in Fig. 6.Bivariate PCA biplots of PC1 against PC2 describing the variations in morphometric measurements between locations, coasts and sexes areshown in Figure 7, 8 and 9 respectively.No clear separation was observed among the locations, between the coasts or between the sexes, as the overlapping in PCA scores was apparent.
The confusion matrix and statistics to measure the performance of linear discriminant analysis (LDA) models developed consequent to principal component analysis to optimize the separation among the groups (locations, coasts, and sexes) is given Table 5 and the pictorial separation has been illustrated as LDA plots in Fig. 10, 11 and 12.
The accuracy (ACC) and misclassi cation rate (MR) of the LDA model for discriminating among different sampling locations were 0.70 and 0.30,respectively.Location-wise sensitivity (True Positive Rate, TPR) and precision (Positive Prediction Value, PPV) were less than 0.90and 0.80, respectively for all the sampling locations with the lowest values estimated for SEZ (TPR=0.20 and PPV=0.39).Furthermore, the kappa coe cient that measures the agreement between classi cation and truth valueswas found to be low (Kappa= 0.62).Therefore, no clear separation was observed among any of the sampling location (Fig. 8).The accuracy (ACC), misclassi cation rate (MR), sensitivity (TPR) and precision (PPV) of the LDA model for discriminating between the coasts were 0.74, 0.26, 0.84 and 0.75, respectively.The Kappa coe cient for the model was also low (Kappa=0.45)and therefore, clear horizontal overlapping was observed between the coasts,which indicate that the samples were not different between the coasts (Fig. 9).The lowest accuracy (ACC), highest misclassi cation rate (MR), lowest sensitivity (TPR), precision (PPV) and Kappa of 0.58, 0.42, 0.65, 0.61 and 0.15, respectively were observed for the LDA model developed to discriminate between the sexes which was also apparent from the horizontal overlapping between the sexes (Fig. 10).Moreover, the Mcnemar's test p values for the LDAs of locations and coasts were signi cant (p ≤ 0.001) which indicates that there are signi cant differences in the frequencies of false positive and false negative during classi cation ).Therefore, it is essential to identify and understand the differences among the stocks so that they can be grouped into separate management units for the effective and e cient management interventions (Turan et al. 2005).
In the present study, multivariate analysis (MANCOVA) performed to determine the signi cant difference among the factors,i.e., sampling locations, between the coasts and between the sexes,showed signi cant variations (p<0.001) in morphometric measurements for all the factors.MANCOVA is an essential multivariate statistical tool which reduces the error by controlling the effect covariate(s) (here, carapace length) on the relationship between the categorical independent variables or factors (here, locations, coasts and sexes) and continuous dependent variables (here, morphometric measurements).In the present study, signi cant interaction effects (p<0.001) of covariate (CL) with factors,viz., locations, coasts and sexes, were observed, which indicates that the factors are related tocovariate.This interaction effect is due to signi cant variations (p<0.001) in covariate i.e., carapace length (CL) among and along with the factors, i.e., the sampling locations, between the coasts and between the sexes.The above results evidently indicate that there is a size related variation in the morphometric traits for the species and therefore, an allometric correction to remove size effect is essentially required before further analysis.The allometric transformation suggested by Elliott et al. (1995) to address the size effect successfully reduced the size effect as the correlation coe cients of individual morphometric measurements with covariate (CL) before transformation weresigni cantly reduced for most of the morphometric measurements after transformation.Interestingly, after allometric correction, all the morphometric measurements for sampling locations, most of the measurements between the coasts and even some of the measurements between the sexes,exhibited signi cant variation as obtained from univariate analysis, i.e., ANOVA.
From principal component analysis (PCA), rst eight principal components (PCs)with cumulatively explained variance of 73.97% were selected based on Kaiser Selection criteria (Kaiser 1960) as each one of them individually explained the variance greater than average variance.Out of eight, rst two PCs were further investigated to evaluate the loadings of the variables which elucidated that 10 morphometric measurements i.e., T2122, TL, T1920, T2324, T410, T2526, T712, CW, T35 and T911 loaded signi cantly highest on PC1 and PC2.These morphometric measurements were concentrated more along the abdominal region viz., second to fth abdominal pleuron characters and anterior portion of cephalothoracic region viz. the antennal region.Most of these measurements (6 out of 10) were distributed perpendicular to longitudinal axis (i.e., TL) of the sand lobster and therefore, the variation explained by the analysis is concentrated more along the width of the lobster, probably for the dorsoventral compressed shape of the sand lobster.Nevertheless, there was no apparent separation of population, neither among the locations nor between the coasts,which is evident from the PCA biplot using PC1and PC2.It must be noted that with all the morphometric measurements taken into account, both PC1 and PC2 cumulatively explained only 45.50% of the total variance which appears to be low and also gives hint to the fact that the signi cant variations in the morphometric measurements obtained by univariate analysis (ANOVA) might not be adequate enough to segregate the population of sand lobster into separate stocks.Like-wise no clear separation was observed between the sexes.
Discriminant analysis performed to optimize the separation of the sand lobster population based on different factors, i.e., sampling locations and between coasts and also to discriminate any variations in morphometric traits between sexes failed to elucidate any clear separation.The linear discriminant analysis (LDA) model developed to separate between the sexes showed a lower sensitivity with low precision and highest misclassi cation rate of 42% which clearly indicates that the signi cant variations in some of the morphometric measurements between the sexes is not adequate enough tosegregate them into two separate groups.Among the locations,though sensitivity and precision of the model was relatively high for four of the locations, it was very low for SEZ location,resulting in an overall misclassi cation rate of 30% for the model.
Similarly, despite a relatively high sensitivity and precision, a misclassi cation rate of 26% was experienced forthe model while attempting to segregate the population between the coasts.These results clearly indicate that there is not adequate evidence based on which the populations can be segregated into separate stocks using the observable variations in morphometric measurements in sand lobster, T. unimaculatus.The misclassi cation rates obtained from DFA clearly indicatethe similarity between the populations,which could be attributed to the common environment, genetic origin at an earlier period, and associated withgenetic mixing of the stocks,particularly those in the transition zones.The mixing of populations from east coast and west coast could be attributed to the presence of panmictic stocks almost on the same latitudes lying on either side of the coast and due to absence of any natural geographical impediment.The information derived from the present morphometric study further supports the inference drawn by an earlier molecular study where low levels of genetic differentiation among shovel-nosed lobsterpopulations was observed using molecular techniques like RAPD and mitochondrial DNA genes (COI &Cyt b) probably due to higher connectivity among the populations (Jeena et al. 2015).
Lobster have longer pelagic larval period, allowing them longer dispersal and population connectivity to remote locations, which can mask the genetically distinct sub-population (Kough et al., 2013).The present study, based on the morphometric assessment, found no structuring in the population along the Indian coastline when sampling sites were used as factors.However, a closer insight in to the specimens grouping in a 2D space created by the linear discriminant 1 and 2 (LD 1 and 2), the samples from southern (SWS, SEC and SEN) and northern localities(SWV and SEZ) were found to be grouped together.But the present assessment cannot conclude that the latitudinal grouping is for genetic structuring or driven by phenotypic responses to ambient environment (physical and biological) which invariably is a function of latitude.Though previous study (Jeena et al., 2015) demonstrated poor genetic structuring of Indian shovel-nosed lobster populations, a more detailed study using more matrilineal and nuclear gene coupled with tagging experiments is required to have robust information on the population structure (Corrigan et al., 2018) of shovel-nosed lobsters along Indian coastline of over 8000 km.

Conclusion
The ndings of the study supplement the present scienti c understanding about the sand lobster population in India, which would greatly aid in the better management and conservationof shovel-nosed lobster resource in India.Being a single population distributed along the Indian sub-continent, a common uni ed management strategy with a speci c focus on implementation of minimum legal size of capture (MLS) by strictly preventing the capture of sand lobster below the size at rst maturity and release of berried female lobsters back to sea should be adopted to rebuild the declining stock and ensure sustainable management of the resource.

Declarations
Ethicsstatement Shrimp samples were directly obtained from the commercial trawl catches and were purchased from the local shermen.Permission from authorities was not required for the studied sites from where the shrimps were collected as it was outside the protected areas (PAs).Linear discriminant analysis (LDA) plot illustrating the separation between the coasts Linear discriminant analysis (LDA) plot illustrating the separation between the sexes ; Pinheiro et al. 2005 and AnvariFar et al. 2011).Therefore, morphometric traits are more promising for studying short-term environmentally induced variation, which is essential for sheries management (Begget al. 1999; Swain and Foote 1999).Several earlier studies have emphasized the utility of morphometric and meristic data in separating the stocks of the sh living in the same or different environments (Krumholz and Cavanah 1968; Anyanwu and Ugwumba 2003; Turan et al. 2005).
, in order to decrease redundancy among the 32 morphometric variables (Samaeeet al. 2006) and to extract a number of dependent variables for population differentiation (Anvarifaret al. 2011; Kuberanet al. 2020).

Figures
Figures

Figure 1 Sampling locations along Indian coast Figure 2
Figure 1

Figure 5 Variables
Figure 5

Table 1
Sampling locality, coast, geographical coordinates, sex and number of specimens collected stocks.A scatter plot analysis based on linear discriminant analysis (LDA) scores was used for visual observation of groups.All the statistical analysis was performed usingR statistical software package,Boston, MA,RStudio, Version 1.4.1106,Release name: Tiger Daylily (RStudio Team, 2021).

Table 3
Summary of ANOVA performed on 32 morphometric measurements of T.unimaculatus collected along India to compare the difference between locations, coasts and sexes

Table 4
Morphometric measurements showing their respective loadings and eigen values to the rst eight principal components * Top ten morphometric measurements with the highest signi cant loadings on PC1 and PC2 * Top ten morphometric measurements with the highest signi cant loadings on PC1 and PC2

Table 5
(Panikkar and Jayaraman 1966;Iles and Sinclair 1982e performance of linear discriminant analysis (LDA) models developed to investigate the separation among the sampling locations and between the coasts and sexes This present study is the rst time investigation in which an attempt has been made to identify any possible variations in sand lobster, T. unimaculatus population from Indian waters using truss morphometry techniques.Animals in a population, apart from genetic factors, experiencing different physical, biological and ecological factors such as geographic variation, salinity, temperature, photoperiodicity, essential food availability and shing intensity may differentiate into separate stocks with identical growth, mortality and maturation characteristics(Panikkar and Jayaraman 1966;Iles and Sinclair 1982