Morphological Variation between Hatchery Bred and Wild Caught Anabas testudineus in Malaysia

This study was designed to characterise the cultured and wild populations of Anabas testudineus in Malaysia using morphological parameters. Fish samples from the East and West coast of the country were obtained from fishermen (wild samples) and well-recognised climbing perch farmers in Kedah, Kelantan, Johor and Selangor. The Truss network method was applied to obtain necessary data and analysed to examine phenotypic variation between the cultured and wild stocks. Results obtained suggest that each hatchery population belonged to a distinct stock as revealed by their separate clustering into individual unique groups. However, an extensive overlap was observed in the wild population suggesting similarity of origin. The most important morphological parameters for the discrimination of the two populations are the homologous landmark B (i.e., snout to insertion of the pelvic fin) and C (i.e., above the eye to insertion of the pelvic fin). Genetic characterisation of the A. testudineus is needed to complement the findings of this study and establish a baseline for the development of a selective breeding programme for the fish species in Malaysia.


INTRODUCTION
The climbing perch Anabas testudineus is one of the economically important and highly valued fish species around the world (Ambak et al. 2010;Chaturvedi et al. 2015). It is widely distributed in freshwaters, brackish waters, and estuaries of several countries in Asia (Zalina et al. 2012;Bungas et al. 2013). The popularity of this species is predicated on its hardiness; hence, it can tolerate unfavourable water conditions, like high turbidity and variations in salinity up to 12 ppt (Hitchcock 2008;Chotipuntu & Avakul 2011). This is because the fish is an obligatory airbreathing species that is capable of surviving for prolonged hours outside the water (Hughes et al. 1986). Consequently, A. testudineus has prolonged freshness even when out of water for a long time and is traditionally esteemed for its fine flavour (Rahman et al. 2013).
The non-pigmented and transparent nature of the egg has also made this fish an important animal model for diverse biological studies due to the ease of observation of early life development . Several fundamental studies earlier conducted have led to the domestication, breeding, and aquaculture production of the A. testudineus (Atal et al. 2009;Cacot et al. 2009;Kohinoor et al. 2013). However, there is a paucity of information about morphological variation between domesticated hatchery-bred stocks and their wild-caught counterparts. Such studies in conjunction with genetic characterisation can be the basis upon which a selective breeding programme could be conducted aimed at improving the performance characteristics of fish through the cross-breeding of strains (Solomon et al. 2015).
According to Normala et al. (2017), conventional morphological methods continue to have an important role in stock identification even despite the development of advanced techniques that can directly examine biochemical or genetic variations in the fish. The criticism about traditional morphometric measurements being contradictory and generating ambiguous results (Garrido-Ramos et al. 1997;Doukakis et al. 2000) has given rise to the development of geometric landmark-based techniques. Hence, this is currently the most rigorous morphometric technique used around the world ( Barriga-Sosa et al. 2004;203 Pinheiro et al. 2005;Bagherian & Rahmani 2009). Therefore, the current study which seeks to investigate the landmark-based morphological variations between hatchery bred and the wild populations of A. testudineus in Malaysia represents the first step towards developing a selective breeding programme for the species.

Study Area and Sample Collection
The samples of A. testudineus were collected at 12 different sites in four states around Peninsular Malaysia (Kedah, Kelantan, Johor and Selangor) as shown in Table 1. The wild samples were collected from contracted fishermen from the three states and were caught using gillnet and angling. For the hatchery samples, collections were made from hatcheries listed by the Department of Fisheries (DOF) directory as the main captive producer of A. testudineus in Malaysia. It should be noted that the standard operating procedures of the different hatcheries differ in terms of culture and care for the fish under captive conditions. However, this was not thought to be a source of bias for delineating the morphometric parameters of the fish. All the collected samples were placed on ice and transported to the bioscience laboratory at the Faculty of Food Science and Fisheries for further morphological analyses. The fish sample's identity was confirmed using the identification key given by Ambak et al. (2010). Thereafter, they were tagged for morphological analysis. The experimental protocols for this study were approved by the Universiti Malaysia Terengganu committee on research. All methods used in this study involving the care and use of animals were following international, national and institutional guidelines (Protocol QL638.99.N6).

Morphological Discrimination of A. testudineus using Truss Network
The morphological measurements were conducted based on a truss network anchored on 10 homologous landmarks as shown in Fig. 1. This resulted in 19 linear measurements as detailed in Table 2. Pictures of the A. testudineus were captured using a Sony camera (Cyber-shot 16.2MP Model number: DSC-TX10 50i, Japan). The photos were then measured using the NIS element Basic Research software.

Morphological Data Analysis
The original measurements of the Truss network were firstly standardised to remove the size effect from the data set (Murta et al. 2008;Jaferian et al. 2010). Hence, to normalise the individuals in a sample to a single arbitrary size, common to all samples, but maintains the individual variation (Sen et al. 2011), the allometric formula by Elliott et al. (1995) was used as shown: Where M = Observed character measurement; M adj = size-adjusted measurement; L o = Standard length of the fish; Ls = Overall mean of the TL for all the progenies; b = Estimated for each character from the collected data as the slope of the regression of log M on log L o , using all fish of all the progenies.
The data obtained from the Truss network technique were analysed by the multivariate analysis of discriminant function analysis (DFA) using Statistical Package for Social Science (SPSS) version 16.0 software for windows. The relative importance of discriminant variables (functions) is determined on some basic measures namely: the relative percentage of the Eigenvalue/the percent of variance existing in the discriminating values and the associated canonical correlation. The sample centroids graph was also done using the SPSS software to see the separation on the biplot and determine the morphological character that can be used to separate the fish samples into distinct groups.

RESULTS
The result for the DFA for the transformed homologous landmark distances of A. testudineus is presented in Table 3. The Eigenvalue of the first four principal components was selected, as they were more than 1, following the recommendations of Kaiser (1961). Results obtained revealed that the first principal component (PC1) accounted for 91.6% of the total variance with only positive coefficients. Principal component 2 (PC2) had positive and negative coefficients and counted for 4.3% of the total variance in the samples. Similarly, PC3 and PC4 accounted for 2.6% and 1.4% of the total variation, respectively. In total, all four principal components accounted for 99.8% of the variance observed for A. testudineus from cultured and wild populations. Canonical correlations in the four principal components were also high ranging from 0.996 (PC1) to 0.815 (PC 4). In general, the most influential variables of the principal components include the homologous landmark B (i.e., the snout to insertion of the pelvic fin) and C (i.e., above the eye to insertion of the pelvic fin). The means of canonical variances scores and the scattered plot are presented in Table 4 and Fig. 2. The result obtained showed a cluster of all wild populations into a single group. However, the hatchery-bred population was distinctly separated from one another without any noticeable overlap between them or the wild population.

DISCUSSION
Stock discrimination is fundamental to fisheries and hatchery management (Begg & Waldman 1999). This is because of its importance in designing and managing a productive breeding programme (Wedemeyer 2001). Fish stocks are identified based on differences in the characteristic among different groups with similar life histories (Hossain et al. 2010). Generally, fishes are known to be the most susceptible to environmentally induced morphological variations among vertebrates (Allendorf et al. 1987;Wimberger 1992;Solomon et al. 2015).
Just like many other studies, the morphological characteristic has proven useful in the discrimination of hatchery bred and the wild population of A. testudineus from different parts of Malaysia. This is based on the observation of a clear separation among the hatchery population and distinct from the wild stock. Hence, suggesting that each population in the cultured stock belongs to different stocks while all the wild population belongs to a single distinct stock. Similar to the finding of this study, Swain et al. (1991) had shown that wild and hatchery Coho salmon could be distinguished using a Truss network system. Phenotypic differences in Clarias gariepinus and endangered Tor putitora have also been reported to be high for different environments (i.e., culture and wild) in the study reported by Solomon et al. (2015) and Patiyal et al. (2014), respectively. The variation observed between wild and hatchery populations has been suggested to be due to different rearing conditions in the environment rather than genetic differences between the stocks (Swain et al. 1991). Husbandry conditions in the wild and captive environments are far apart in many ways. The hatchery provides almost everything to ensure better survival of the reared fish in a restricted environment while the natural environment conditions are uncertain and survival of the fittest is championed by natural selection (Wedemeyer 2001;Sedwick 1995). Many researchers have highlighted the role of environmental conditions such as food abundance and temperature as an important cause of fish morphological plasticity. These parameters greatly differ for wild and hatchery environments. Hence, despite the hatchery population being a fragment of the wild population, the various process of domestication, artificial selection adopted in hatchery over the years, and husbandry methods used may have impacted the morphological parameters significantly (Blanchet et al. 2008). Ahmad (2015) had also opined that domestication and adaptation of fish to the various hatchery conditions are potential courses of observed morphological deviation of hatchery and wild stock. It could then be right to hypothesise that the different husbandry practices used in the different hatcheries in this study may have also resulted in the emergence of distinct stocks for the different stocks of A. testudineus studied. Another school of thought has also suggested that morphological variation in culture fishes could have resulted from gene pollution of the original domesticated stock occasioned by several years of inbreeding and accidental hybridisation with closely related species (Olufeagba et al. 2002;El-Serafy et al. 2007;Solomon et al. 2015;Okomoda et al. 2018;Okomoda 2018). However, this can only be correctly inferred by genetic characterisation. This study observed that the most important morphological parameters for the discrimination of both cultured and wild populations were related to the head. An earlier study by Leslie and Grant (1990), Schweigert (1990), and Haddon and Willis (1995) have revealed that morphological measurement of the head and body depth are the most important characters of discrimination in samples of Lophius vomerinus, Clupea pallasi, and Hoplostethus atlanticus, respectively. In African catfish too, Turan et al. (2005) and Solomon et al. (2015) had reported that C. gariepinus samples from the wild and the cultured environment can be discriminated against using head-related morphometric parameters.
Although morphometric differences among stocks have been linked to differences in geographical and ancestral origins (Hossain et al. 2010), our findings for the wild population suggest they all fit into a single distinct group. Contrary to this finding, the reports of Turan et al. (2004) for wild populations of Liza abu from the Orontes, Euphrates, and Tigris rivers in Turkey show the fish were differentiated into distinct groups using morphometric parameters. Wedemeyer (2001) had stated that natural environmental influences cause a random process of mutation, genetic drift, and gene flow resulting in changes in wild population morphology. The similarity of wild stock in the current study may be suggestive of similarity in the environmental factors of the wild in the different states sampled or the ancestral origins of the A. testudineus.
Since each hatchery stock described in this study is individually distinct morphometrically, mating broodstock from the different hatchery stock or with wild lineages could theoretically help to produce offspring with better performance (Aung et al. 2010). The selection of wild population as a supplement to hatchery population has been used in some previous selective breeding programmes aimed at improving the fitness of hatchery bred fishes especially in the area of disease-resistant (Kapuscinski et al. 1996;Uraiwan et al. 2007). This could be investigated in future research for the A. testudineus samples reported in this study. However, it should be noted that mating broodstock from a distinct population might not translate into an improvement of the performance of cultured fishes; but, it can increase variation within the hatchery population which could be important for future breeding programmes (Tallmon et al. 2004).

CONCLUSION
Through the morphological data collected, the study concluded that each hatchery population belonged to a distinct stock, however, the wild population is likely from a similar origin as indicated by the extensive overlap noticeable in the biplot. Despite the findings of this study, the genetic characterisation of the fish (i.e., A. testudineus) must be conducted to validate the reports of this study.