Genetic and morphological evidence for cryptic species in Macrobrachium australe and resurrection of M . ustulatum ( Crustacea , Palaemonidae )

Abstract. Macrobrachium australe is an amphidromous prawn living in the insular freshwater systems of the Indo-Pacific. Because it possesses few informative morphological characters, that often vary from one habitat to another, M. australe has produced much taxonomic confusion and has historically been described under eight synonyms. Here, 53 specimens collected throughout the Indo-Pacific under the name M. australe were phylogenetically and morphologically examined. Results revealed that what has been called M. australe belongs to at least two distinct species: M. australe, distributed from the Southwest Indian Ocean to the Central Pacific Ocean, and a cryptic species potentially restricted to the Northwest Pacific Ocean, here identified as M. ustulatum, which until now was considered as a junior synonym. Although they are not quite found in the same habitat (lentic-lotic), the presence of these distinct, and reciprocally monophyletic entities in the same rivers on the islands of Palau and Santo strongly favors the hypothesis of two reproductively isolated entities. Six morphological characters, including the proportions of the joints of the male second pereiopod, the shape of the epistome lobe and the armature of the fourth thoracic sternite, are evidenced as diagnostic. A neotype of M. australe is designated and deposited in the Muséum national d’Histoire naturelle in Paris.


Introduction
Macrobrachium australe (Guérin-Méneville, 1838 in Guérin-Méneville 1829-1838) is a large-sized prawn species living in the insular freshwater systems of the tropical Indo-Pacifi c.The species possesses an amphidromous life cycle in which the larvae develop in the marine plankton before returning as juveniles to rivers to grow and reproduce.This specifi c life cycle is considered a critical element to the persistence of the populations and the viability of the species in insular freshwater habitats (Keith 2003), which are usually sparse and of an unstable nature.Macrobrachium australe, which is one of the most widespread amphidromous prawns, distributed from the Seychelles and Madagascar eastwards to French Polynesia, is considered as a "Least Concern" species in conservation priority (Jayachandran 2001).However, an unpublished primary molecular sequencing of specimens collected by the Muséum national d'Histoire naturelle (MNHN) over a range that spans the Indo-Pacifi c suggested the possible existence of distinct evolutionary units (species) within what was morphologically assigned to M. australe.This raised questions about the exact distribution of M. australe and its actual conservation status.
The genus Macrobrachium Spence Bate, 1868 is considered as one of the most challenging decapod crustacean groups for species delineation and alpha taxonomy (Holthuis 1950;Chace & Bruce 1993;Short 2004).While taxonomically informative morphological traits (e.g., on the rostrum and the second pereiopods) are mostly present only in fully developed males, they also often vary within species (Holthuis 1950;Short 2004).Indeed, different environmental conditions have been shown to induce carapace shape variation in the South American prawn M. borellii (Nobili, 1896) (Torres et al. 2014) and the Asian prawn M. nipponense (De Haan, 1849in De Haan 1833-1850) (Chen et al. 2015).Similarly, in populations of M. australe from Reunion Island, the rostral shape varies with water velocity, being long, thin and orientated upward in lentic populations and short, robust and straight in lotic populations (Zimmermann et al. 2012).Taking into account these data, it is likely that the carapace plasticity in Macrobrachium has contributed to taxonomic confusion.In this context, species delimitation in the genus should not be solely based on carapace features but on the integration of complementary sources of evidence (e.g., morphometric or molecular data).
Addressing these issues, however, requires a reasonably complete sampling size, ideally covering the whole distribution range of the target species.Indeed, the broad geographical distribution of M. australe, combined with the diffi culty of accessing tropical insular freshwater systems, has led to geographic sampling gaps blurring the distinction between geographical and/or ecological variations and specieslevel divergences.As a consequence, M. australe has historically been described under eight synonyms (De Grave & Fransen 2011), each more or less corresponding to an isolated location on the edge of the species range: 2 synonyms were described from the Central and Southwestern Pacifi c (Palaemon australis Guérin-Méneville, 1838 in Guérin-Méneville 1829-1838 from Tahiti and P. danae Heller, 1865 from Sydney, Australia (but probably erroneous, see Short 2004)); 3 synonyms were described from the Central Indo-Pacifi c (P. dispar von Martens, 1868 from Adonara Island, Indonesia; P. (Eupalaemon) ustulatus Nobili, 1899 from Rigo, Papua New Guinea; Leander lepidus de Man, 1915 from Jos Sudarso Bay, West Indonesian Papua); and 3 synonyms were described from the Western Indian Ocean (P.alphonsianus Hoffman, 1874 from Reunion Island; P. parvus Hoffman, 1874 from Nossy-Faly Island, Madagascar; and P. malliardi Richters, 1880 from Mauritius).
This study aims to clarify the evolutionary independence of lineages within the complex "M.australe", using both molecular and morphological characters.A molecular phylogeny was performed based on the DNA sequencing of two unlinked molecular markers (16S and 28S rRNA) in 53 specimens sampled over an area that covers the range of "M.australe", and includes localities close to the type localities of the synonymous species.To construct a robust molecular taxonomic framework, all published sequences of Macrobrachium were included in the phylogenetic analyses.This allowed the assessment of the status of "M.australe" lineages relative to one another and relative to other species of the genus Macrobrachium.Next, in order to assign molecular clusters to correct species names, morphological characters within and among the divergent genetic clusters identifi ed as "M.australe" were examined and compared to all museum data (from collections of voucher specimens and from original descriptions or taxonomic revisions) available on M. australe and its seven associated synonyms.Diagnostic morphological characters of identifi ed species were compiled to facilitate future identifi cations within the "M.australe" complex.

Collection, molecular data and analyses
Specimens were collected in the Indo-West Pacifi c (Fig. 1, Table 1) by electrofi shing (portable Dekka 3000 electric device, Germany).In the fi eld, most of the specimens were released back into the river after a piece of pereiopod was clipped off and fi xed in 95% ethanol for molecular analyses.Only some specimens (mostly corresponding to mature males presenting morphological features enabling specieslevel taxonomic identifi cation) were kept intact and fi xed in 75% ethanol for morphological analyses.
Sequences were used to search the NCBI database for related sequences using the Sequence Search (MegaBlast algorithm) tool implemented in Geneious software (Biomatters Ltd) for highly similar sequences (Zhang et al. 2000).Related sequences (Appendix) were downloaded and aligned with our sequences using the MUSCLE Server (Edgar 2004).Sequences from P. debilis Dana, 1852 were used as outgroup.Poorly aligned sites in the 28S dataset were identifi ed using Gblocks Server v. 0.91b (Castresana 2000) and removed from analyses.Parameters used in Gblocks allowed for smaller fi nal blocks and gap positions within the fi nal blocks.Identical sequences were identifi ed using DnaSP v. 5 (Librado & Rozas 2009) and one representative sequence retained for each haplotype.Nucleotide substitution models were determined using BIC in jModelTest v. 2.1.6(Darriba et al. 2012).
A phylogeny was inferred for each gene using the Bayesian inferences (BI) implemented in BEAST v. 1.8.2 (Drummond et al. 2012).BEAST input fi les were generated using BEAUti v. 1.8.0 (Drummond et al. 2012).The Yule model was used as a prior for the speciation process.When an optimal nucleotide substitution model was not available in BEAUti we selected a similar but more complex near-optimal model (Huelsenbeck & Rannala 2004).The heterogeneity of the mutation rate across lineages was set under an uncorrelated, relaxed log-normal clock and mean rate was fi xed to one.Each analysis ran for 100 000 000 generations with sample frequency of 1000.The fi nal trees were calculated based on 99 000 trees (after burning 1001 generations) with maximum clade credibility and median node heights.
Length of burning was determined by examination of traces in Tracer v. 1.6 (Rambaut et al. 2014).
In addition, to better assess the robustness of nodes, a Maximum Likelihood (ML) tree using the GTR + Γ model of rate substitution (the most inclusive model) was inferred for each gene using RAxML HPC2 (Stamatakis 2006) on Teragrid v. 7.2.7,implemented in the Cyber Infrastructure for Phylogenetic Research (CIPRES) portal v. 3.1 (Miller et al. 2011).The best-scoring ML tree was estimated from 100 independent searches, each starting from distinct random trees.Robustness of nodes was assessed using the rapid bootstrapping algorithm (1000 replicates) (Felsenstein 1985;Stamatakis et al. 2008).
After checking for congruency between 16S and 28S tree topologies, a concatenated dataset was built using BI and ML analyses similar to those described above.Sequence variation was partitioned between genes and gene-specifi c nucleotide substitution model parameters were used, with each gene allowed to evolve at a different rate.Nucleotide substitution models were similar to those used in the single-gene phylogenies.

Morphological study
Morphological observations were made on the specimens sampled in this study, as well as on specimens from various European museum collections corresponding to M. australe and its synonyms.The rostrum, the general cephalon, the second pereiopods and the abdomen were observed using a stereoscopic microscope.The carapace length and the proportions of the various joints of the second chelipeds were measured using callipers (under stereoscopic microscope when needed).Drawings were made using the "Digital Inking" method (Coleman 2003;Coleman 2006) by tracing vectorial paths on high resolution photographs with Adobe Illustrator (CS6) equipped with a WACOM PTZ-1230 graphic tablet.

Molecular analyses
Fifty-three specimens were successfully sequenced for their 16S or 28S gene, or both (Table 1).The 19 amplifi ed 16S sequences included 520 bp containing 36 variable sites.The 49 amplifi ed 28S sequences included 743 bp containing 33 variable sites.After alignment with 53 sequences from Genbank (Appendix), the 16S dataset of 520 bp of the ingroup included 299 variable sites, of which 167 were phylogenetically informative.The 28S dataset of 774 bp of the ingroup included 305 variable sites, of which 245 were phylogenetically informative.After alignment with 62 sequences from Genbank (Appendix) and removal of ambiguous blocks, 504 bp of sequence remained to be used in phylogenetic analyses (65% of the original 774 positions), of which 144 bp were variable and 123 bp phylogenetically informative.
In each of the 16S and 28S trees, specimens originally assigned to "M. australe" were segregated into two groups (hereafter named group 1 and group 2, Fig. 2) that appeared as sister taxa, and were genetically distinct from one another and from all other species included in the tree.Indeed, group 1 and group 2 formed a monophyletic group, highly supported by posterior probabilities (PP) and bootstrap (B) analyses (16S: PP = 1, B = 96; 28S: PP = 1, B = 95) (Fig. 2), and the monophyly of each group was supported by both PP and B with both genes (16S: group 1: PP = 1, B = 98; group 2: PP = 1, B = 100; 28S: group 1: PP = 1, B = 85; group 2: PP = 1; B = 100).Analysis based on the concatenation of 16S and 28S could only be performed on the M. australe complex and M. lar (Fabricius, 1798), for which the 2 genes were available.In this phylogeny, although the monophyly of the two groups was supported (group 1: PP = 1, B = 100; group 2: PP = 1, B = 100), the relationship of the two groups as sister taxa was not highly supported (PP = 0.56, B = 100) (data not shown).
Based on the 16S dataset, pairwise genetic distances between group 1 and group 2 varied from 5.2 to 5.9%.Pairwise genetic distances within group 1 varied from 0 to 1.4%, while no genetic variation was observed within group 2. Pairwise genetic distances between group 1 and palaemonid sequences ranged from 3.7 to 9.2%, while genetic distances between group 2 and other palaemonid sequences ranged from 5.9 to 8.5%.Based on the 28S dataset, pairwise genetic distances between group 1 and group 2 varied from 2.2 to 2.4% (variation of 13 nucleotide sites out of 504 bp).Pairwise genetic distances within group 1 varied from 0 to 0.6%, while no genetic variation was observed within group 2. Pairwise genetic distances between group 1 and sequences of Macrobrachium ranged from 2.2 to 12.2%, while genetic distances between group 2 and other sequences of Macrobrachium ranged from 2.2 to 11.9%.

Specimens morphologically examined from museum collections or from the literature
For Palaemon danae from the NHMW, we used data from De Man (1892), and the comments of Holthuis (1950) and Short (2004).For P. dispar _ENREF_42from the RMNH, we used data from De Man (1887, 1892) and Roux (1933).For P. malliardi from the ZMB, we used the original description by Richters (1880) and the comments of Holthuis (1950).For Leander lepidus from the RMNH, we used the comments of Holthuis (1950).The type specimens of P. australis, deposited in an unknown collection, P. alphonsianus and P. parvus (both originally deposited at the RMNH) could not be examined, as the specimens are lost.For P. ustulatus from the MSNG, two syntypes from Rigo (Papua New Guinea) were examined.These corresponded to a male (specimen VII-106) and a female (specimen VII-105).In ustulatum (Nobili, 1899).For these, sampling locality are given at nodes.See

Morphological study
In order to assign molecular clusters to correct species name, morphological characters within and among the genetic clusters, identifi ed in phylogenetic analyses as "M.australe", were examined and compared to all available museum data on "M.australe" and its eight associated synonyms (i.e., collections of voucher specimens and descriptions or taxonomic revisions).Specimens of group 1 were assigned to Macrobrachium ustulatum (Nobili, 1899) and specimens of group 2 to M. australe (Table 2).
The holotype of Palaemon ustulatus from MSNG is represented by one male only with the minor second cheliped (Table 2).The proportions of the various joints of the cheliped do not match with the morphology of the specimens belonging to M. australe: CAL / CHL = 0.69 (vs 0.79-1.04);CAL / PL = 1.15 (vs 1.19-1.60)and FL / PL = 0.66 (vs 0.46-0.53)(Table 2).Moreover, these characteristics fi t well with those observed on specimens identifi ed as M. ustulatum: CAL / CHL = 0.67-0.75;CAL / PL = 1.01-1.25 and FL / PL = 0.49/0.66(Table 2).The holotype shows an epistome with lobes strongly produced anteroventrally (vs not strongly diverging anteriorly) and a fourth thoracic sternite armed with a rounded median process (vs unarmed).Overall, six characters were identifi ed as diagnostic to distinguish the two species (Table 3).
Out of the seven synonyms of M. australe, six were confi dently identifi ed as junior synonyms based on the rostrum shape, the position of the hepatic spine toward the antennal spine and the proportions of the male second pereiopod (Table 4).The seventh synonym, P. ustulatus is no longer considered as such.

Description
ROSTRUM.Moderately long, reaching to or slightly beyond end of scaphocerite (Fig. 3A); tip slightly upcurved; dorsal margin with 10-11 teeth, including 2-3 teeth on carapace; ventral margin with 3-4 teeth; teeth large, evenly spaced; sometimes distance between ultimate and penultimate teeth much larger than distances between other teeth; sometimes distance between 1 st and 2 nd teeth larger than between 2 nd and 3 rd teeth.

Remarks
Since the type specimen of P. australis is lost, and because of the taxonomic problems discussed above, a specimen from the Avana River on Rarotonga Island (Cook Islands, Central Pacifi c), which is geographically close to the type locality of P. australis (Tahiti), is herein designated as the neotype of M. australe in order to stabilize the taxonomy of the species.One specimen in our sample was collected from Moorea, which is closer to Tahiti.However, the specimen was in a less good condition than specimens from Rarotonga; thus, diagnostic morphological characters defi ned in this study were less obvious.Specimens from Rarotonga and from Moorea, however, showed no genetic variation in both the 28S and the 16S genes.

Distribution
From the Indian Ocean (Madagascar, Comoros and Mascarenes Islands) to the Central Pacifi c Ocean (French Polynesia and Cook Islands).Syntopic with M. ustulatum in the Northwest Pacifi c Ocean (Taiwan, Philippines, Indonesia, Papua New Guinea and Palau) and the Southwest Pacifi c Ocean (Vanuatu) (Fig. 1).

Habitat
Lentic habitats in the lower courses of rivers and at the edge of rivers, in water plant communities (Keith et al. 2013).

Color patterns
Body colors (Fig. 4A) tend to match the substrate, varying from grey to brown, rather translucent, with three reddish stripes on each side of the cephalothorax.

Description
ROSTRUM.Moderately long, reaching end of scaphocerite (Fig. 5A); tip horizontal; dorsal margin nearly straight, faintly convex, with 10-12 teeth, including 3 teeth on carapace; ventral margin with 3-4 teeth; teeth large, generally evenly spaced, with gap near anterior end of dorsal series; sometimes distance between 1 st and 2 nd teeth larger than between 2 nd and 3 rd teeth.

Distribution
Syntopic with M. australe in the Northwest Pacifi c Ocean (Taiwan, Philippines, Papua New Guinea, Indonesia and Palau) and in the Southwest Pacifi c Ocean (Vanuatu) (Fig. 1).

Habitat
Lotic habitats in the middle course of rivers, clear and oxygenated waters, sometimes above waterfalls.Rocky substrates, boulders, pebbles and pools in countercurrents.

Color patterns
Body colors (Fig. 4B) tend to match the substrate, varying from black to brownish red, with three reddish stripes on each side of the cephalothorax.

Discussion
The DNA sequencing of two unlinked molecular markers in specimens collected from a large geographic range throughout the Indo-Pacifi c, covering the entire known distribution range of M. australe and including localities close to most of the type localities of its synonyms, suggested the existence of at least two distinct species under what is commonly assigned to "M. australe": M. australe (Guérin-Méneville, 1838 in Guérin-Méneville 1829-1838), with a large distribution range from the Southwest Indian Ocean to the Central Pacifi c Ocean, and a cryptic species found syntopically at some islands of the Western Pacifi c Ocean (Taiwan, Philippines, Indonesia, Papua New Guinea, Palau and Vanuatu).Although they are not quite found in the same habitat (i.e., lentic vs lotic), the presence of these distinct genetic entities in the same rivers in Palau and the Santo Islands strongly favors the hypothesis that these genetic entities are reproductively isolated from one another.In addition, the two single-gene phylogenetic reconstructions, which included several species of Macrobrachium available from GenBank, suggested that the two species are closely related and could even be sister species.However, this hypothesis remains to be tested using a more comprehensive molecular dataset.
One of the main diffi culties encountered in this study was to assign the genetic groups to the correct scientifi c name.Indeed, probably because of its large geographic range and morphological plasticity (Zimmermann et al. 2012), M. australe is currently also known under seven associated synonyms.Moreover, the lack of consensus on the diagnostic characters to use in the discrimination of species within this genus has led to descriptions that are diffi cult to compare.For instance, P. australis was described based on the shape of the rostrum; P. danae on the proportions of the various joints of the minor second pereiopod (see De Man 1892); P. dispar on the proportions of the major second pereiopod; P. alphonsianus on the second pereiopods; P. parvus on the rostrum, the position of the hepatic spine and the proportions of the minor second pereiopod; P. malliardi on the rostrum and the minor second pereiopod; P. ustulatus on the minor second cheliped; and L. lepidus on the rostrum, the cephalic region and the pereiopods I, II, III and IV (in very young specimens as well).In this context, to assign a scientifi c name to the cryptic species, we fi rst sequenced specimens collected from or close to the type localities of all these synonymous species.For example, we studied samples from Moorea (French Polynesia) and Rarotonga (Cook Islands), which are close to the type locality of P. australe, Palau and Papua (close to the type localities of P. dispar, P. ustulatus and L. lepidus) and the Mascarene Islands (close to the type localities of P. alphonsianus, P. parvus and P. malliardi).We then examined the morphological characters of sequenced specimens, with particular attention toward mature male specimens, and compared these characters to museum samples and/or taxonomic descriptions of available synonyms.
Overall, molecular and morphological results both suggested that among the eight synonyms of M. australe, one (M.ustulatum) corresponds to a valid species, which therefore needs to be resurrected, one (P.australis) corresponds to the original species name, and six are junior synonyms of M. australe (P.danae, P. dispar, P. alphonsianus, P. parvus, P. malliardi, and L. lepidus).Indeed, based on rostrum shape, the position of the hepatic spine toward the antennal spine and the proportions of the male second pereiopod, P. danae, P. dispar, P. alphonsianus, P. parvus, P. malliardi and L. lepidus can be considered as junior synonyms of M. australe (Holthuis 1950;Chace & Bruce 1993;de Grave & Fransen 2011).On the contrary, the proportions of the various joints of the minor second cheliped (i.e., carpus/chela, carpus/palm and fi nger/palm) of the holotype of P. ustulatus from MCSN do not match with the morphology of specimens belonging to M. australe, but fi t well with those corresponding to the cryptic species identifi ed in this study, M. ustulatum.We also emphasized our morphological analyses on the epistome and fourth thoracic sternite, two features known to be very consistent between sexes and among individuals at different development stages (Short 2004).
M. ustulatum lives in lotic habitats in the middle course of rivers.Indeed, P. dispar of de Man (1902) (most likely M. ustulatum) was from Halmahera, at an altitude of 600 m.
Macrobrachium ustulatum was reported as P. dispar by de Man in 1892, 1893, 1902and 1915. In 1915 de Man examined M. ustulatum from Papua New Guinea and gave a good illustration of its rostrum (de Man 1915: pl.29, fi g. 12).However he continued to report specimens from Papua New Guinea as P. dispar although they had the typical chelipeds of M. ustulatum (de Man 1915: pl.29, fi gs 14, 14a).Later, P. dispar was always cited as one among several synonyms of M. australe, notably by Holthuis (1950), Chace & Bruce (1993) and de Grave & Fransen (2011).In addition, M. ustulatum has until now been confused with M. australe because of the similar shape of their rostrum, the numerous teeth on the cutting edges of the fi ngers of the male second pereiopods and the similar life color pattern.

Conclusion
Overall, genetic and morphological results show that M. australe and M. ustulatum belong to two distinct species.Macrobrachium australe (Guérin-Méneville, 1838 in Guérin-Méneville 1829-1838) has a widespread distribution, from Mascarene to French Polynesia, whereas M. ustulatum (Nobili, 1899) could be more restricted, from Taiwan to Vanuatu.Further studies should be conducted on the known geographic range of M. ustulatum, in order to assess more precisely its distribution.While the biology and ecology of M. australe is relatively well known (e.g., r-reproductive strategy; Ito et al. 2002;Wowor et al. 2009), the life history traits of M. ustulalum are totally unknown, and it is not unlikely that the life history traits of these two species have been mixed in previous studies.This raises questions about the exact distribution of M. ustulatum and its actual species conservation status.Unlike M. lar, the most widespread amphidromous prawn, which so far has revealed no cryptic species (Castelin et al. 2013), M. australe with a similar widespread distribution shows one.Chace & Bruce (1993) and Wowor et al. (2009) all voiced their suspicions about the potential existence of many undescribed cryptic species in Macrobrachium.Furthermore, Liu et al. (2007)

Fig. 1 .
Fig. 1.Map of the Indo-Pacifi c showing localities where Macrobrachium australe (Guérin-Méneville, 1838 in Guérin-Méneville 1829-1838) (black area) and M. ustulatum (Nobili, 1899) (red area) were collected and/or recorded.Capitalized locality names correspond to the 7 localities sampled for this study.Non-capitalized locality names correspond to the localities reported from the literature.Stars shows the type localities of the synonyms of M. australe (black stars) and M. ustulatum (red star).
Abbreviations: CAL = carpus length CED = cutting edge of dactylus CEFF = cutting edge of fi xed fi nger CHL = chela length

Fig. 2 .
Fig. 2. Single-gene molecular phylogenies for Macrobrachium Spence Bate, 1868 based on the mitochondrial 16S and the nuclear 28S genes with Palaemon debilis Dana, 1852 as outgroup.Maximum likelihood (ML) and Bayesian (B) analyses, produced with BEAST and RAxML, respectively.Support values from Maximum likelihood and Bayesian analyses are posterior probabilities (PP) and bootstraps (B), respectively.GenBank accession numbers and working ID numbers are given at nodes.Species names are given on the right-hand side.The name of the sequences produced in the present study are colored in red for M. australe (Guérin-Méneville, 1838 in Guérin-Méneville 1829-1838) and in green for M. ustulatum(Nobili, 1899).For these, sampling locality are given at nodes.See Table1and Appendix for details.

Table 4 .
Measurements of the various joints of the second chelipeds (mm) in Macrobrachium australe (Guérin-Méneville, 1838 in Guérin-Méneville 1829-1838), as Palaemon australis, and in some of its synonyms.ss = slightly shorter.Missing data are represented by dashes.