Caridina variabilirostris (Crustacea: Decapoda: Atyidae), a new species of freshwater shrimp from Pohnpei (Micronesia)

Recently, the status of a new species of atyid shrimp from Pohnpei (Micronesia) was discussed in relation to C. brachydactyla De Man, 1908 and C. mertoni J. Roux, 1911. By combining morphological data with a phylogenetic analysis with closely related species, this species is here described as Caridina variabilirostris sp. nov. Notes on its ecological distribution are also provided. The new species is characterized by a highly variable rostrum and is present in rivers all over Pohnpei Island. The status of this new species is clarifi ed and it is shown that neither C. brachydactyla De Man 1908 nor C. mertoni J. Roux, 1911 occur on Pohnpei Island.


Introduction
Micronesia is a vast expanse of more than 2000 Pacific islands and atolls stretching from the Caroline and Mariana Islands in the West, to the Marshall, Nauru and Kiribati Islands in the East.The Caroline Islands consist of a chain of seamounts, atolls and high islands extended southeastward to the Marshall Islands.The high island of Pohnpei (formely known as Ponape) is situated at 6°54′ N latitude and 158°14′ E longitude (Fig. 1).It belongs to Pohnpei State, one of the four states in the Federated State of Micronesia.The island of Pohnpei covers 345 km 2 .The highest elevation point is 791 m.Aged of 8.7 Ma, this volcanic island was not only the remnant of a hotspot trace, but was also produced in a fracture-induced subduction-related tectonic environment (Rehman et al. 2013).
The freshwater shrimp genus Caridina Milne Edwards, 1837, comprising 298 species (WoRMS database, as of May 2018) and mostly present in the Indo-Pacific region, is the most diversified genus of the Atyidae (De Grave et al. 2015) and an important ecological component in the tropical streams (Covich et al. 1999;Pringle et al. 1993).Its high diversity combined with the lack of informative morphological characters has led to a confused taxonomy (Richard & Clark 2009).Indeed, until recently, the taxonomy of the genus was mainly based on morphological characters.Some of those have been proven highly variable within a species (e.g., rostrum shape and indentation or coloration) and so taxonomically non-informative, making it difficult to establish boundaries between species (Rintelen & Cai 2009;Mazancourt et al. 2017).Thus, there is a need for an integrative and standardized approach to improve the group's systematics, focusing on informative morphological features and using molecular characters (Page et al. 2005;Page & Hughes 2011).Maciolek & Ford (1987) recorded Caridina brachydactyla De Man, 1908 among seven species of Atyidae in rivers on Pohnpei.Later, this species was not collected either by Nelson et al. (1996) or by Buden et al. (2001).Keith et al. (2012), reviewing the decapods of Ponhpei, conducted a freshwater survey from 7 to 17 March 2012, identified the material of Caridina belonging to the complex of C. nilotica as C. brachydactyla or C. mertoni Roux, 1911 (following the previous identifications made from Guam, another Micronesian island, by Leberer & Cai 2003).
One of the aims of the Muséum national d'Histoire naturelle (MNHN) in Paris is to carry out faunistic inventories of rivers in tropical islands in order to establish a better protection of these fragile ecosystems and, in this context, to clarify taxonomy of poorly known organisms.As we examined more and more specimens from Pohnpei, we gradually started to question the validity of some species from this island.Consequently, we here re-examined our specimens collected in 2012 in combining morphological data with a 16S mtDNA analysis and found that neither C. brachydactyla nor C. mertoni occur in Pohnpei, but have been confused until now with one undescribed species.Recently, Mazancourt et al. (2017) highlighted the "Pinocchio-shrimp effect", in this species, which has a variable length of the rostrum depending on the altitude.Detailed description of this new species is given as well as its ecological distribution.The position of this species is clarified by comparing it with C. brachydactyla and C. mertoni.1).

Collection of specimens
Rivers and sites surveyed in March 2012 are indicated in Figure 1 and Table 1.Specimens from Pohnpei were collected by electrofishing (portable Dekka 3000 electric device, Germany).All material preserved in 75%-95% alcohol has been deposited in the collections of the Muséum national d'Histoire naturelle in Paris (MNHN, specimens n° MNHN-IU-2018-231 to MNHN-IU-2018-256).

DNA extraction, amplification and sequencing
For recent specimens, DNA was extracted from abdominal tissues using the semi-automatic Eppendorf ep-Motion 5075 robot.Fragments of the mitochondrial 16S rRNA (~ 520 bp) were amplified using the primers 16Sa-L (CGCCTGTTTATCAAAAACAT) and 16Sb-H2 (CTCCGGTTTGAACTCAGATCA) (Palumbi 1996).DNA amplification was performed in 25 µl PCR reactions, containing approximately 3 ng of template DNA, 2.5 mM MgCl 2 , 0.26 mM of each nucleotide, 0.3 µM of each primer, 5% DMSO, 1 ng of BSA and 1.5 units of QBIOTAQ polymerase (MPBiomedicals).Amplification products were generated by an initial denaturation step of 4 min at 94°C followed by 35 cycles of denaturation at 94°C for 30 s, annealing at 52°C for 40 s, extension at 72°C for 60 s and a final extension step at 72°C for 7 min.
For old collection specimens (syntypes of C. mertoni), a CTAB protocol was used to extract DNA from pleopods.A shorter fragment of the 16S rRNA (332 bp) was amplified using two newly designed primers: 16S-Car-81F (AGGTAGCATAATAAATAGTC) and 16S-Car-413R (CTGTTATCCCTAAAGTAAC).DNA amplification was performed in 25µl PCR reactions, containing 2.5 mM MgCl 2 , 0.26 mM of each nucleotide, 0.3 µM of each primer, 1 ng of BSA and 1.5 units of QBIOTAQ polymerase (MPBiomedicals).Amplification products were generated by an initial denaturation step of 4 min at 94°C followed by 45 cycles of denaturation at 94°C for 30 s, annealing at 55°C for 30 s, extension at 72°C for 40 s and a final extension step at 72°C for 7 min.
PCR products were sequenced using the same primers and in both directions to insure the accuracy of base calls.Chromatograms were edited using Geneious v. 8 software (Kearse et al. 2012).All sequences were deposited in GenBank (Numbers MH476222 to MH476227).A sequence retrieved from GenBank for Paratya australiensis published by Page et al. (2007) was included in our analysis, used as outgroup.
A total of 17 recent specimens were sequenced (Table 2): six of the new species, of which two firstly identified as C. brachydactyla with a long rostrum and four as C. mertoni with a short rostrum; five of C. mertoni from Kolombangara Island; three type specimens of C. variabilis Mazancourt, Rogers & Keith, 2018 and three of C. brachydactyla from the type locality, Sulawesi.In addition, three old collection specimens, syntypes of C. mertoni from Kai Island were included (Mazancourt et al. 2018).

Molecular analyses
DNA sequences were aligned using MEGA7 software (Kumar et al. 2016)  distinct random trees, were conducted.Robustness of the nodes was assessed using non-parametric bootstrapping (Felsenstein 1985) with 1000 bootstrap replicates.Best-scoring Bayesian Inference trees were estimated using MrBayes v. 3.2.6 (Ronquist & Huelsenbeck 2003) also implemented in CIPRES with the previously determined model, running for 10 000 000 generations, a sampling frequency of 2000 and a burn in of 25%.Support for nodes was determined using posterior probabilities calculated by MrBayes.

Morphological comparison
The rostrum, general cephalon, pereopods 1, 2, 3 and 5 and abdomen were observed using a stereoscopic microscope.The proportions of the various joints of the appendages were measured using microphotographs and AnalySIS Works software (Olympus).Drawings were made using the "Digital Inking" method (Coleman 2003(Coleman , 2006) by tracing vectorial paths on stacks of high-resolution photographs using Adobe Illustrator (CS6) and a WACOM MPTZ-1230 graphic tablet.

Collection of specimens
About 50 specimens were collected by electro-fishing in the different rivers prospected in Pohnpei.Of these, 26 were sequenced, 18 with a short rostrum and eight with a long one.A representative sample of six specimens is shown here in the genetic study.

Phylogenetic analyses
After checking that consensus trees obtained from the two different methods were congruent, we decided to show the Bayesian consensus tree with both the ML bootstrap values and Bayesian posterior probabilities on branches.The specimens are grouped in four distinct and moderately supported clades.
The most basal comprises the specimens of C. brachydactyla from Sulawesi (type locality), then, the type specimens of C. variabilis and, finally, two sister clades, one comprising recent and old specimens of C. mertoni and the other the specimens from Pohnpei.Due to the low support values, we cannot discuss the affinities of the different species, but we can confidently assert that the terminal clades are distinct species and consider the last one to be a new species that we describe below.
head.Eyes well developed, anterior end reaching to 0.7 length of antennular peduncle basal antennomere.Antennular peduncle 0.7 times as long as carapace.Anterolateral angle reaching 0.25 length of second antennomere; second antennomere distinctly longer than third.Stylocerite reaching to 0.7 length of antennular peduncle basal antennomere.
Mouthparts.Mandibles dimorphic; left mandible (Fig. 4b) more developed, corpus large, robust with five strong sharp teeth separated by ridged gap; incisor and molar processes separated by patch of long simple setae.Right mandible (Fig. 4c) with five sharp incisor teeth, medially with group of long setae; molar process narrow, elongate, ridged.First maxilla (Fig. 4d) having a lower lacinia with margin broadly rounded, bearing several rows of plumose setae.Upper lacinia elongate, with medial margin bearing a number of distinct teeth and simple setae, palp bearing long simple setae.Second maxilla (Fig. 4e) upper and middle endite with marginal and submarginal simple or slightly plumose setae.Lower endite with simple setae; palp narrow, shorter than upper endite cleft with few setae.Scaphognathite fringed with long simple setae, tapering posteriorly with some long, curved simple setae at posterior end.First maxilliped (Fig. 4h) endopodite ultimate segment medial margin with long plumose setae.Palp elongate, setose.Exopod flagellum long and narrow distally with marginal plumose setae.Caridean lobe large, with marginal setae.Second maxilliped (Fig. 4g) endopodite ultimate and penultimate antennomeres fused, reflected against basal antennomeres.Ultimate, penultimate and basal antennomeres medial margins with long setae of various types; flagellum very long, slender with marginal plumose setae distally.Third maxilliped (Fig. 4f) with terminal article reaching third antennular peduncle antennomere apex; distal antennomere about 10 times as long as wide, slightly shorter than penultimate, ending in large hamulate apical spine surrounded by simple setae.Penultimate antennomere about eight times as long as wide, with group of transverse rows of simple setae.Exopod flagellum well developed, about a third the length of endopodite second article, distal margin with long plumose setae.
abdoMen.Third abdominal (Fig. 4a) somite with moderately convex dorsal profile.Sixth abdominal somite about 0.68 of carapace, 1.76 times as long as fifth somite, slightly shorter than telson.Telson (Fig. 3k) with four pairs of dorsal spines and one pair of dorsolateral spines; posterior margin, with or without median process, exhibits variations, triangular or rounded with four to five intermediate simple setae longer or equal than lateral ones.pl1 (Fig. 3m).Endopod foliform with a developed appendix interna in males.Pl2 (Fig. 3n): appendix masculina on second pleopod reaching 0.52 times length of endopod; appendix interna reaching about 0.85 times appendix masculina length.

Habitat
This species is found among macrophytes in flowing fresh water of the rivers of Pohnpei all along the course, showing good adaptability to different temperatures and hydrological conditions.It is more abundant in higher elevations due to reduced predator pressure.
Colour pattern (Fig. 5) The colour of the body is hyaline with many reds dots.An oblique red band on the cephalothorax is very characteristic.&Yan, 1977 andC. peninsularis Kemp, 1918.The new species displays a variable rostrum length among specimens.When the rostrum is short the general appearance is like C. mertoni, whereas when the rostrum is long, the general appearance is of C. brachydactyla.In C. variabilirostris spec.nov. the antennal spine is placed below the orbital angle, the P5 dactylus with 18-29 spinules whereas in C. mertoni, the antennal spine is somewhat fused with the orbital angle and P5 dactylus with 24-43 spinules.C. variabilirostris spec.nov.differs from C. brachydactyla by the absence of spine on the preanal carina, a shorter distal unarmed portion of the rostrum 0.0-0.4(vs 0.4-1.6)times that of armed portion, by a longer P5 propodus, which is 5.3-8.6 time as long as the dactylus (vs 3.9-5.7)and a lower number of spinules on the P5 dactylus (32-42 in C. brachydactyla).
The new species looks very much like C. variabilis from Guam and Palau with the preanal always nonarmed, the number of teeth and their placement on the rostrum, the proportions between the joints of pereiopods, and the egg size (Mazancourt et al. 2018).But the P5 dactylus has a single terminal spine and a short distal propodus seta (vs two strong distal spines and a very long distal propodus seta), and the posterior margin of the telson with intermediate setae longer or equal than lateral ones (vs shorter or equal than lateral ones).

Discussion
Recently Mazancourt et al. (2017) highligted the "Pinocchio shrimp effect" on this new species.Indeed some specimens with a long rostrum were attributed morphologically to C. brachydactyla whereas others, with a short rostrum, were identified as C. mertoni.In fact all specimens belong to this new species.Indeed the rostrum length widely used in the taxonomy of Caridina might not be as reliable as it was thought.It is highly plastic and varies with environmental parameters.
Despite an increasing use of integrative taxonomy for the systematics of Caridina (and shrimps in general), some authors keep describing new species based not only on morphology alone, but even on a single character.For example, different species described by Richard & Clark (2009, 2014) based on rostrum morphology proved to be synonyms, and the differences observed were after all recognized as intra-specific variation (see Wood et al. 2018;Mazancourt et al. unpublished).
We thus advocate to always perform some molecular work prior to studying a new species in order to unveil this kind of variation within a species, focus on the morphology and expose the variation when describing it.Alternatively, if retrieving molecular data is not possible, using a combination of reliable characters seems more sound than a single one.This would allow to provide characters to identify confidently the species when encountered in the field.It is likely that different species described in this genus will prove to be the same one, exhibiting the same kind of polymorphism as we showed here.

Table 2 .
Specimens used in the genetic analysis.