Genetic Diversity and Conservation of an Endemic Taiwanese Species, Platyeriocheir formosa

Mitten crabs (Brachyura, Varunidae, Varuninae) are native to east Asia and currently classified into eight species belonging to four genera: Eriocheir De Haan, 1835, with E. japonica (De Haan, 1835) and E. ogasawaraensis Komai et al., 2006; Paraeriocheir gen. nov., with P. hepuensis (Dai, 1991) and P. sinensis (H. Milne Edwards, 1853); Platyeriocheir Ng et al., 1999, with P. formosa (Chan et al., 1995) and P. guangdonga sp. nov.; and Neoeriocheir T. Sakai, 1983, with N. leptog‐ natha (Rathbun, 1913); plus an eighth, currently nameless species with the status of a species inquirenda [1].

Catadromous mitten crabs have the unusual life history of spawning in the sea and growing up in rivers. Fertilized eggs hatch into zoea, which leave the female and begin life in the sea as plankton. After passing through five ecdysis cycles over a period of several weeks, they metamorphose into megalopa (post-larval stage) that live in estuaries and migrate upstream to freshwaters where a second metamorphosis, into juvenile crabs, occurs [2]. The larval stage drifts passively with coastal currents, providing high potential for gene flow within coastal waters. Larvae mainly drift in proximity to the coastline rather than the open sea, so long distance dispersal across open seas is restricted [3]. Juvenile crabs move into rivers and dwell in their middle or upstream reaches where they grow until adulthood. They generally inhabit clear rushing waters as well as hiding in rock crevices by day and coming out at night to feed, their main food being periphyton growing on rocks and aquatic vegetation. They spend most of their lifetime (1-3 years) in freshwater and migrate downstream to coastal waters when mature to mate and spawn [3].
Of this group, only two genera and species, Paraeriocheir hepuensis and Platyeriocheir formosa, are native to Taiwan. The distribution of P. hepuensis in Taiwan extends from Dasi, Yilan County to Wu Stream, Taichung City. P. formosa, a Taiwan endemic, is mainly distributed in rivers of eastern Taiwan (Figure 1). Both species are of economic importance; nevertheless, neither have been artificially cultured from egg to maturity, and as a result, people catch large numbers of them from the wild for sale. Therefore, these wild populations must be able to tolerate intense exploitation. In addition, they also suffer other adverse impacts, such as habitat destruction, natural population declines, climatic oscillations, and so on. Too, many rivers were severely impacted when typhoon Morakot hit the eastern and southern portions of Taiwan in 2009 by mudflows that covered riverbeds, resulting in a significant decrease in the P. formosa population size. In addition, wild crab catches are insufficient to meet consumer demand. Farmers directly import juvenile crabs of P. sinensis from China for culture until they are grown to adults for sale. However, when P. sinensis escape from farms and invade Taiwan rivers, hybrids with native species can occur to degrade the genetic structure of native species. At present, populations of these two native species are dwindling in population size due to habitat destruction by natural disasters and unrestricted over-harvesting. Moreover, P. sinensis has also been introduced into Taiwan for short-term aquaculture, disregarding the consequences of potential threats to native species. It is clear that conservation of P. formosa should be taken more seriously. P. formosa, P. hepuensis, and P. sinensis have similar morphological characters. Consequently, species identification was clarified using a molecular marker, mitochondrial DNA cytochrome oxidase subunit I (COI), which is often used in taxonomy, biodiversity assessments, phylogenetics, and phylogeographic studies [4][5][6][7][8]. Here we discriminate these three species using mt COI sequences. Because microsatellites have large mutation rates of 10 −5~1 0 −2 per generation [9][10], they are widely used as markers for studying genetics, population structure, kinship, and mating system [11][12][13][14]. In a previous study, the intraspecific genetic diversity of P. formosa was analyzed using COI sequences, with results showing insignificant genetic differences among samples from different streams [15]. To conserve P. formosa population diversity and ensure the sustainable use of this natural resource, its genetic diversity needed to be determined using polymorphic microsatellite loci.
This study attempted to distinguish P. hepuensis, P. sinensis, and P. formosa, and explored interspecies and intraspecific genetic diversity, using mt COI gene sequences. In addition, the genetic diversity of an endemic Taiwanese species, P. formosa, was examined by microsatellite loci. Effective management and conservative strategies also are proposed herein.

Sample collection
A total of 40 Platyeriocheir formosa specimens were collected in November 2010 from Jin-Luen, Taitung (120°55′ E, 22°32′ N), southeastern Taiwan. All Paraeriocheir hepuensis (n = 20) were collected from Lau-Mei Stream in New Taipei City and 2 P. sinensis individuals were collected from an aquaculture farm in Pingtung County.

Genomic DNA isolation
Muscle tissues from all specimens were preserved in 95% ethanol until DNA extraction. Genomic DNA was isolated and purified from the muscle tissue of all individuals. Five hundred milligrams of tissue with 1 mL lysis buffer was digested with 55 µL proteinase K solution. Small amounts of DNA were extracted for polymerase chain reactions (PCR) using a Puregene core kit A (Qiagen, Valencia, CA, USA).

COI subcloning and analysis
The complete COI gene was amplified using the specific forward primer 5'-CTCTAACR-GATTCCCCATCTTCTC-3' and reverse primer 5'-ATCCTACACATCTGTCTGCC-3' designed by the authors. A PCR consisted of approximately 50 ng genomic DNA, 50 pmol each of the forward and reverse primers, 25 mM dNTP, 0.05 ~ 0.1 mM MgCl 2 , 10× buffer, and 5 U Taq polymerase (Takara Shuzo, Shiga, Japan), and brought up to 100 µL with Milli-Q water (Millipore, Billerica, MA, USA). The PCR program included one cycle of 4 min at 95 °C, 38 cycles of 1 min at 94 °C, 50 s at 50 °C, and 1 min at 72 °C, followed by a single further extension of 10 min at 72 °C. We evaluated 8 µL of each product on 0.8% agarose gel to check PCR success and confirm product sizes. The remaining PCR products were run on 0.8% agarose gels and purified using a DNA Clean/Extraction kit (GeneMark, Taichung, Taiwan). Purified DNA was subcloned into the pGEM-T easy vector (Promega, Madison, WI, USA) and transformed into Escherichia coli JM109. Plasmid DNA was isolated using a mini plasmid kit (Geneaid, Taichung, Taiwan). Clones from all individuals were sequenced on an Applied Biosystems (ABI, Foster City, CA, USA) automated DNA sequencer 377 (ver. 3.3) using a Bigdye sequencing kit (Perkin-Elmer, Wellesley, MA, USA).
In total, 42 COI sequences were subcloned. All sequences were aligned using Clustal W [16] and then checked with the naked eye. Intraspecific and interspecific genetic distances and numbers of different nucleotides were calculated using MEGA software [17]. The interspecific variable site numbers and intraspecific nucleotide diversities were computed by DnaSP v5 [18]. The phylogenetic trees for COI sequences were constructed using neighbor-joining [19] and maximum parsimony methods [20]. Cluster confidence was assessed using a bootstrap analysis with 1000 replications [21]. The minimum spanning tree (MST) was computed from the matrix of pairwise distances between all pairs of haplotypes in each sample using a modification of the algorithm [22]. We evaluated whether sequences had evolved under strict neutrality. Fu's Fs [23] and Tajima's neutrality tests [24] were performed in Arlequin 3.1 [25]. The significance of the statistics was tested by generating random samples under the hypothesis of selective neutrality and population equilibrium, using a coalescent simulation algorithm [26]. Tajima's test is based on an infinite-site model without recombinations. A significant D value can be due to factors other than selective effects, like population expansion, a bottleneck, or heterogeneity of mutation rates [27]. The possible occurrence of historical demographic expansions was examined using the mismatch distribution [28] implemented in Arlequin [25]. The distribution is unimodal in samples following a population demographic expansion [29].

Genotyping and data analysis
All 18 microsatellite loci [30] were amplified in this study. A PCR was performed in a volume of 25 µL that included ~10 ng genomic DNA, 10 pmol reverse primer, 10 pmol forward primer, 25 mM dNTP, 0.05-0.1 mM MgCl 2 , 10× buffer, and 0.5 U Taq polymerase (Takara Shuzo, Tokyo, Japan) with Milli-Q water. The PCR products were subjected to a 1.5% agarose gel and allele sizes were checked by comparison with a DNA ladder and the length of the original sequence. Forward primers were labeled with FAM, TAMRA, or HEX fluorescence markers. PCR amplifications were carried out in a Px2 Thermal Cycler (Thermo Fisher Scientific, Waltham, MA, USA) with the following temperature profile: 1 cycle of 95 °C for 4 min, followed by 38 cycles of 94 °C for 30 s, and annealing at 50-60 °C for 30 s and 72 °C for 30 s. Each 5 µL of PCR product from three loci labeled with different fluorescence tags was mixed and precipitated with 95% alcohol. Semi-automated genotyping was performed using a capillary ABI 3730XL DNA Analyzer (ABI). Genotypes were scored with GeneMapper 4.0 (ABI).
The total number of alleles (na) and effective allele numbers were estimated for each locus using Popgene [31]. Observed (H O ) and expected (H E ) heterozygosities were independently calculated for each locus. Deviations from Hardy-Weinberg expectations (HWEs) were examined by an exact test using GENEPOP [32]. Linkage disequilibrium among all pairs of loci was determined using Burrow's composite measure [33] and χ2 values.

Interspecific diversity
Molecular systematics and historical population dynamics were also analyzed using mt COI gene sequences. The full-length of all 42 COI sequences in Platyeriocheir formosa (n = 20), Paraeriocheir hepuensis (n = 20), and P. sinensis (n = 2) are consistent with 1534 bp in length. The average percentage of nucleotide components consist of 64% A + T for P. sinensis, 63% for P. hepuensis, and 62.6% for P. formosa, with slight differences among these three species.
The numbers of interspecific nucleotide differences ranged 196 -221 with an average of 205.08 ± 4.14 between P. formosa and P. hepuensis, 199 -214 with an average of 204.82 ± 3.30 between P. formosa and P. sinensis, and 60 -79 with an average of 66.55 ± 3.87 between P. hepuensis and P. sinensis. Some interspecific nucleotide variable sites are shown in Figure 2.
The interspecific genetic distances ranged 0.142 -0.163 with an average 0.150 ± 0.006 for P. formosa vs. P. hepuensis, 0.144 -0.157 with an average 0.148 ± 0.002 for P. formosa vs. P. sinensis, and 0.041 -0.054 with an average 0.045 ± 0.003 for P. hepuensis vs. P. sinensis. All 42 sequences in the study and one outgroup sequence from Xenograpsus testudinatus (NCBI accession number NC013480) were used to construct a phylogenetic tree by neighbor-joining (NJ) and maximum-parsimony (MP) methods. Phylogenetic trees presented significant clustering among the three species indicated P. formosa, P. hepuensis, and P. sinensis are an individually monophyletic group and share a common ancestor in two genealogical trees (Figure 3a, b). P. hepuensis and P. sinensis have closer relationships than P. formosa, indicating that the former two share a common recent ancestor. In addition, we also concluded that the COI gene is an effective genetic marker for distinguishing these mitten crabs having similar morphological characteristics.

Intraspecific diversity and historical population dynamics
In total, 86 and 84 variable sites were respectively observed within intraspecific sequences of P. formosa and P. hepuensis. All 20 COI sequences from P. formosa contain seven highly variable sites (Figure 5a). The intraspecific number of nucleotide differences ranged 2 -20. Haplotype diversity (H d ), the mean number of nucleotide differences (k), and mean nucleotide diversity (π) are 1, 9.73 ± 3.63, and 0.006 ± 0.003, respectively. A total of 86 substitutions containing 75 transitions and 11 transversions occur within these 20 sequences. There are 23 highly variable sites observed within 20 COI sequences of P. hepuensis (Figure 5b). Intraspecies nucleotide differences ranged 1 -27. Haplotype diversity (H d ), the mean number of nucleotide differences (k), and mean nucleotide diversity (π) are 1, 11.36 ± 4.83, and 0.007 ± 0.004, respectively. The 85 substitutions include 68 transitions and 17 transversions. Intraspecific genetic distances of P. formosa ranged from 0.001 -0.013, with an average of 0.006 ± 0.002. In contrast to P. formosa, P. hepuensis had similar intraspecific genetic distances that ranged from 0.001 to 0.018, with an average of 0.007 ± 0.003. Extremely high levels of H d and low to moderate levels of π were discovered in these two species.
A similar genetic pattern is observed in many marine species [34]. The most likely explanation is that the accumulation of mutations over time in a rapidly growing population leads to an increase in the number of haplotypes; even so, population sizes suffer seriously when there is   low genetic diversity. Except for wild mitten crabs, which must counteract overharvesting, probable natural causes include climate oscillations that result in temperature and water quality changes, mudflows covering riverbeds that can block migratory pathways, and any other environmental factor that affects adult reproduction and larval survival in estuaries. A second explanation for low levels of genetic diversity in these species could be due to their high dispersal potential during the planktonic egg and larval stages, resulting in strong gene flow among populations. Intraspecific genetic diversity in P. formosa analyzed by COI sequences indicate insignificant genetic differences among different populations [15]. It is interesting to note that P. formosa spawns in the sea and planktonic stage dispersal trends northward due to seasonal currents along the eastern Taiwan coast. This easily explains why P. formosa has low genetic variability among samples from different streams.
Four groups of marine fishes were defined based on the haplotype diversity (h d ) and nucleotide diversity (π) of various mtDNA coding regions [35]. The most widespread group possesses a high number of haplotypes (H d > 0.5) and moderate to low levels of sequence divergences (0.4% < π < 0.8%). P. formosa and P. hepuensis were found in this study to have high haplotype diversity (H d = 1) and moderate nucleotide diversities (0.6% & 0.7%), which fit the most common pattern  observed in marine fishes. High haplotype diversity within regional populations can be maintained through historically rapid population increases, resulting in the accumulation of mutations in populations [36]. Nevertheless, the shallow mtDNA branch structure of the NJ tree in these two mitten crabs might have resulted from catastrophic reductions in population size, which would produce low values of π.
The D values of Tajima's D neutral tests were analyzed to test this. P. formosa and P. hepuensis were -2.451 (p <0.001) and -2.129 (p <0.003), respectively. Negative Fu's Fs values of -12.646 (p = 0) in P. formosa and -11.333 (p = 0) in P. hepuensis suggested that these two species experienced a recent population expansion event. The mismatched distribution analysis presented average intraspecific nucleotide differences among COI sequences of 9.726 ± 7.731 in P. formosa and 11.358 ± 6.344 in P. hepuensis, which were unimodally distributed ( Figure 6) and indicated that the population experienced a historical expansion event. One and two central haplotypes were found in the minimum spanning tree (MST) of P. formosa and P. hepuensis, and most of the haplotypes were located at the tips (Figure 7), implying that adaptive radiation occurred. sinensis [37]. The mean allele number per locus (na) in the catadromous P. formosa was estimated to be 9.61, which was higher than that of P. sinensis on average (na = 4.94), but lower than that of the marine P. trituberculatus (na = 22) [37][38]. These results agree with the observation that catadromous P. formosa have lower genetic diversity than marine P. trituberculatus species. P. formosa and P. hepuensis have a peculiar migratory history. Juvenile crabs migrate from the sea into rivers where they grow until adulthood. Mature adults move from their habitats in middle and upstream river reaches down to coastal waters for reproduction. Consequently, the most important conservation considerations are high water quality, freely flowing channels, reduced harvesting by humans, and preventing the invasion of the exotic P. sinensis. Focus must start on the catadromous journey of juvenile crabs from estuaries, and good water quality is the key to their survival and sustainable populations. Secondly, uninterrupted river flows are necessary, as mitten crabs must migrate throughout rivers in order to complete their full life cycle regardless of whether they are upstream-swimming juveniles or downstream-swimming adults. When river bottoms are buried under mudflows, previously established aquatic organisms are lost. However, temporarily created waterways will allow aquatic organisms to survive and complete their life cycles. Furthermore, overfishing results in decreasing crab resources and increases in their selling prices which results in additional overharvesting. However, capture can be banned for temporary periods of time to allow populations to recover. Finally, P. sinensis must be prevented from invading Taiwan's rivers. Because there are insufficient harvests of P. formosa and P. hepuensis to meet human demand, aquaculturists import large numbers of juvenile P. sinensis crabs and raise them to adulthood for sale. However, there are some problems with the culturing process. For one thing, P. sinensis effortlessly escapes from aquaculture even though ponds are equipped with anti-slipping nets. What is more, market prices may collapse if supply exceeds demand, and the industry might give up raising crabs and dump them. It is, therefore, certainly possible that these non-native crabs might colonize all of Taiwan's streams, causing an ecological catastrophe to native Taiwanese crab populations. Preventing the invasion of non-native mitten crabs should be of universal concern to Taiwan crab management and conservation.

Microsatellite locus
The genetic structure and population dynamics of Taiwan's P. formosa population diversity must be continually monitored to ensure the sustainable use of this valuable natural resource.

Conclusions
The interspecific and intraspecific genetic diversity of two native Taiwanese mitten crabs, P. formosa and P. hepuensis, were determined in this study using the mtDNA COI gene and microsatellite loci. These two species possess similar genetic patterns with extremely high haplotype diversity and low to moderate nucleotide diversity. These results suggest that their population sizes historically underwent expansions but are currently undergoing serious decreases. Consequently, a conservation policy is proposed here that includes maintaining free-flowing stream channels and good water quality, preventing overharvesting by limiting harvesting to specific seasons, preventing the establishment of non-native mitten crabs, and conducting research on improved methods for the aquaculture of native mitten crabs.