Morphological and molecular characterization of freshwater prawn of genus Macrobrachium in the coastal area of Cameroon

Abstract Macrobrachium (Bate, 1868) is a large and cosmopolitan crustacean genus of high economic importance worldwide. We investigated the morphological and molecular identification of freshwater prawns of the genus Macrobrachium in South, South West, and Littoral regions of Cameroon. A total of 1,566 specimens were examined morphologically using a key described by Konan (Diversité morphologique et génétique des crevettes des genres Atya Leach, 1816 et Macrobrachium Bate, 1868 de Côte d'Ivoire, 2009, Université d'Abobo Adjamé, Côte d'Ivoire), leading to the identification of seven species of Macrobrachium: M. vollenhovenii (Herklots, 1857); M. macrobrachion (Herklots, 1851); M. sollaudii (De Man, 1912); M. dux (Lenz, 1910); M. chevalieri (Roux, 1935); M. felicinum (Holthuis, 1949); and an undescribed Macrobrachium species M. sp. To validate the genetic basis of the identified species, 94 individuals representing the species were selected and subjected to genetic characterization using 1,814 DArT markers. The admixture analysis revealed four groups: M. vollenhovenii and M. macrobrachion; M. chevalieri; M. felicinum and M. sp; and M. dux and M. sollaudii. But, the principal component analysis (PCA) separated M. sp and M. felicinum to create additional group (i.e., five groups). Based on these findings, M. vollenhovenii and M. macrobrachion may be conspecific, as well as M. dux and M. sollaudii, while M. felicinum and M. sp seems to be different species, suggesting a potential conflict between the morphological identification key and the genetic basis underlying speciation and species allocation for Macrobrachium. These results are valuable in informing breeding design and genetic resource conservation programs for Macrobrachium in Africa.

Taking into consideration both sex and size of the prawn, Konan (2009)

developed a new key for identification of West Africa
Macrobrachium. Using the newly developed key, Makombu et al. (2015) described a tentative range of the biodiversity of Macrobrachium in the South region and increased the number of known species in Cameroon from four (Monod, 1980) to six (Makombu et al., 2015).
Other studies also pointed out the higher species richness of Cameroon Macrobrachium (Doume, Toguyeni, & Yao, 2013;Tchakonté et al., 2014). However, these recent studies in Cameroon have not covered the whole coastal area, which encompasses three regions namely South, South West, and Littoral regions. With the increasing threat of the quality of fresh and brackish water of the coastal area of Cameroon (E & D, 2009;Folack, 1995) that can affect species integrity, information on the genetic diversity of Macrobrachium in the whole coastal region is urgently needed to implement a management plan.
Application of species identification keys relies heavily on distinct morphological features unique to each species. However, due to a restricted number of characters available for identification, with many features common to all known species of Macrobrachium, morphological identification of species of this genus is quite difficult (Qing-Yi, Qi-qun, & Wei-bing, 2009). Characterization based only on morphological examination could lead to under-or overestimation of biodiversity (Lefébure, Douady, Gouy, & Gilbert, 2006). Given the current scenario, unbiased taxonomic classification through both morphological characterization and molecular characterization could shed more light into the diversity of this genus in the region.
The emergence of next-generation sequencing tools has revolutionized taxonomic classification studies, as cost per sequencing output is continuously decreasing (Kilian et al., 2012). This has resulted in a shift of focus from molecular identification studies using universal genetic markers to high-throughput genotyping using single nucleotide polymorphisms (SNPs). One of the emerging new genotyping technologies is Diversity Arrays Technology (DArT) (Imelfort, Batley, Grimmond, & Edwards, 2009;Kilian et al., 2012), which allows for simultaneous detection of several thousand of DNA polymorphisms (depending on the species) by scoring the presence or absence of DNA fragments in genomic representations generated from genomic DNA through a process of complexity reduction (Kilian et al., 2012).
The efficacy of DArT markers in the analysis of genetic diversity, findings, M. vollenhovenii and M. macrobrachion may be conspecific, as well as M. dux and M. sollaudii, while M. felicinum and M. sp seems to be different species, suggesting a potential conflict between the morphological identification key and the genetic basis underlying speciation and species allocation for Macrobrachium. These results are valuable in informing breeding design and genetic resource conservation programs for Macrobrachium in Africa.

K E Y W O R D S
Cameroon, DArT markers, freshwater prawn, Konan key, Macrobrachium, morphological and molecular characterization population structure, association mapping, and construction of linkage maps has been demonstrated for a variety of species (Appleby, Edwards, & Batley, 2009). DArT does not rely on previous sequence information for initial marker development, and this makes it the chosen platform for genetic characterization of species with little sequence information like African Macrobrachium (Sánchez-Sevilla et al., 2015).
This study sought to determine the morphological and genetic diversity of Macrobrachium species in the main rivers of the South, South West, and Littoral regions of Cameroon using Konan (2009) key and DArT technology. It will serve to validate the current morphological-based classification of West Africa Macrobrachium and contribute to the design of Macrobrachium breeding in Africa.

| Ethics statement
In Cameroon, freshwater prawn fishing is artisanal and an authorized activity. We bought fresh specimens from fishermen who chill and market wild prawn immediately after capture.

| Morphological identification of prawns
Before measurements, specimens were weighed individually using an electronic balance, coded, and preserved in 95% ethanol.
Morphometric variables were recorded according to the measurement technique described by Kuris, Ra'anan, Sagi, and Cohen (1987)  All dimensions of the two legs of the second pair of the pereiopods and their joints were taken along the external lateral line. For each of the specimens collected, a total of 33 morphometric and six meristic characters were recorded (Appendix 1). After measurement, the specimens were identified to species level using the key described by Konan (2009) (Appendix 2). The Monod (1980) key was used when the species description was not found in Konan key. Samples were then stored in 95% ethanol for further molecular analysis.

| Measurements of physicochemical parameters
Measurements of water physicochemical parameters of the rivers were done according to APHA (1998) and Rodier, Legube, and Brunet (2009) standards to see whether they have an influence in the distribution of Macrobrachium species in the three regions. Water temperature and dissolved oxygen were monitored monthly using oxygen meter (HI 9146, Hanna, Italy), while pH was measured using a pH meter (HI 98129, Hanna, USA).

| Morphometric analysis
The Hierarchical Ascending Classification (AHC) based on Euclidean distance and Ward's algorithm was carried out to cluster species identified according to their morphometric similarities.

| Data filtering
Genotypic data quality control and checks were undertaken using PLINK v 1.9 (Purcell et al., 2007) entailing removal of SNPs with <80% call rate and <5% minor allele frequency (MAF). Consequently, a total of 1,814 SNPs were remained for further analysis.

| Genetic diversity
Minor allele frequencies (MAF) were estimated using PLINK v 1.9 (Purcell et al., 2007). Observed and expected heterozygosities were calculated using ARLEQUIN software, version 3.5 (Excoffier & Lischer, 2010). The expected heterozygosity per locus was calculated as follows: where n is the number of gene copies in the sample, k is the number of haplotypes, and p i is the sample frequency of the ith haplotype.

| Population structure
Principal component analysis (PCA) was performed using PLINK (Purcell et al., 2007) and results were visualized using the GENESIS package (Buchmann & Hazelhurst, 2015) A model-based unsupervised clustering method implemented in the program ADMIXTURE v. 1.3.0 (Alexander, Novembre, & Lange, 2009) was used to estimate the genetic composition of individual prawns using the 1,814 markers. The analysis was run with K (number of distinct species) independent runs ranging from 2 to 20. A 10-fold cross-validation (CV = 10) was specified, with the resultant error profile used to explore the most probable number of clusters (K), as described by Alexander et al. (2009). The optimal K was confirmed using discriminate principal component analysis (DPCA) and the Evanno ΔK methods. Graphical display of the admixture analysis was done using the Microsoft Excel package.

| Analysis of genetic relationships
Pairwise F ST was computed with 1,000 permutations using ARLEQUIN software, version 3.5 (Excoffier & Lischer, 2010). A phylogenetic tree was then generated from a matrix of pairwise F ST estimates using Splits Tree software, version 4.13.1 (Huson & Bryant, 2006).

| Physicochemical parameters of the rivers
The physicochemical parameters of the eight rivers sampled are shown in Table 1. The mean pH of all the rivers was between 7.08 and 7.70.
Temperature varied from 23.66 to 29.28°C. Lokoundje River recorded the highest mean temperature (26.64°C). Dissolved oxygen was highly variable in the rivers of the Littoral region (Nkam River: 2-8 mg/L), whereas in the South West region, it was high in all the rivers with the lowest value recorded in Batoke River (5-6.63 mg/L).

| Morphological analysis and distribution of the species in the three regions
Of the 1,566 specimens examined morphologically using Konan (2009) and Monod (1980) keys (Table 2) These species were not found in all the three regions (

| Morphometric similarities between species identified
A dendrogram of hierarchical cluster analysis showing morphological similarities between Macrobrachium species is shown in Figure 3.

| Genetic diversity
Diversity Arrays Technology markers presented an average genotype call rate of 40.8% and an average scoring reproducibility of  Table 4 and Figure 4, respectively. Approximately 85% of all loci had minor allele frequencies <0.1.

| Population structure
The admixture analysis revealed four main clusters (K = 4)

| Population differentiation
The genetic distance of the species based on F st ranged from −0.0105 to 0.9461 (Table 5)

| D ISCUSS I ON
Major systematic treatments of freshwater prawns have been based on morphological characteristics alone (Murphy & Austin, 2003;  Cameroon, further studies on its biology and ecology are highly recommended. The relationship between the species identified in this study based on morphological features is the same as that observed in the Makombu et al. (2015) and Konan (2009) Raman et al., 2010;Sánchez-Sevilla et al., 2015;Wenzl et al., 2004). The use of DArT in characterization of animals (and particularly aquatic animals) has been limited (Melville et al., 2017). In this study, even though up to 50,000 SNP markers were available after genotyping, only 1,814 met the criterion for further analysis, which limits the extent of the genetic diversity that can be captured. A larger study with more robust markers is paramount to completely characterize the genetic structure and relationships among the target species. we do not have conclusive evidence to suggest they are the same species, at the genetic level they seem to be highly similar. The phenotypic differences seen between them may be due to differential expression of genes that control the morphometric features used for classification (Dimmock et al., 2004). The phenotypic differences observed may also be the possibility of morphotypes within a species. M. sollaudii male has 2nd pereiopods (chelipeds) more developed than M. dux male, this could be two male morphotypes of a same species. Moreover, looking at the sex ratio, male highly dominate female in M. sollaudii collected in the three regions (>90% male).
Cases of morphotypes within the genus Macrobrachium have been documented. Kuris et al. (1987)  colocation of these species in the same rivers and habitat, as well as their amphidromous behavior patterns characterized by female migration from rivers to estuaries following hatching, larval development in saltwater, and a return upriver migration by postlarvae (Bauer and Delahoussaye, 2008), possibly provides ample opportunity for mating and hence gene flow between these two species.
The lack of genetic differentiation between these species has been previously observed. J. G. Makombu et al. (unpublished results) observed similar results using mitochondrial DNA, increasing the possibility that they are descended from the same maternal genetic stock. Konan (2009) also found no genetic differentiation between M. vollenhovenii and M. macrobrachion using enzymatic polymorphism. However, given the quite divergent marker profiles observed, there is evidence to suggest that these species are different. The large differences in number of polymorphic markers and genetic heterozygosity measures observed point to significant differentiation driven by differential speciation. The fact that they are colocated in the same rivers and habitats rules our differential manifestation of environmental influences. Perhaps a study of differential gene expression may shed more light as to the genetic basis of the huge phenotypic differences. It is instructive to note that during morphological analyses, specimens having characteristics of both of M. vollenhovenii and M. macrobrachion were found. These "hybrid" individuals may represent the admixed individuals borne out of the two species. This could not be further investigated owing to limited resources available for this study. Given the relatively small marker set used, a deeper characterization of these species using dense genetic markers (both organellar and autosomal) would be necessary to remove any doubts as to the genetic relationship between them.
In contrast to the admixture results, both the PCA and the ge- Despite their close similarity in terms of phenotypic and morphometric features, they are not located in the same habitat; hence, there is reasonable chance to conclude that they are different species.
According to Dimmock et al. (2004), Macrobrachium is a notoriously difficult genus taxonomically, as the morphological plasticity of important traits changes extensively and gradually during the growth and is influenced by environmental parameters. Morphologically similar species are often quite genetically distinct. Analogous situations have been reported for some marine crustaceans (Knowlton, 2000) and freshwater macroinvertebrates (Baker, Hughes, Dean, & Bunn, 2004;Shih, Ng, & Chang, 2004).

The perils of morphological taxonomy of species of the genus
Macrobrachium have been recorded in previous studies (Murphy & Austin, 2005;Vergamini et al., 2011). So far, many studies have called into question morphological classification of members of this group (Boulton & Knott, 1984;Fincham, 1987;Holthuis, 1952). Additionally, Qing-Yi et al. (2009, 2004, and Short

I D E NTI FI C ATI O N K E YS O F M ACRO B R ACH I U M
a. Key to family and genus (Monod, 1980) Legs of the first two pairs of pereiopods different and finished by pincers; legs of the second pair are well developed and end with strong pincers; rostrum is developed and has teeth on both sides; medium to large or very large size (45-182 mm) …… Palaemonidae (Macrobrachium).
2. Key to species of Macrobrachium according to Konan, 2009 1. The carpus of the second pereiopod is shorter or same length with the merus, the ratio of the length of carpus/length of merus is <1.08 …… 2 The carpus of the second pereiopod is longer than the merus; the ratio of the length of carpus/length of merus is >1.08 …… 3 2. The merus is 0.20-0.27 times longer than large; the length of the carpus is 0.70-0.95 times the length of the palm; the rostrum tip is at the same level or longer than the antenna scale …… M. macrobrachion The merus is 0.27-0.47 times longer than large; the length of the carpus is 0.45-0.75 times the length of the palm; the rostrum tip is shorter than the extremity of the antenna scale …… M. vollenhovenii 3. The carpus is longer than the palm (length of carpus/length of palm = 1.17-1.39); the merus is as long or longer than the palm (M/P1 = 0.98-1.14) …… 4 The carpus is shorter or the same length with the palm (length of carpus/length of palm = 0.61-1); the merus is shorter than the palm (M/P1 = 0.58-