Species diversity of freshwater shrimp in Henan Province, China, based on morphological characters and COI mitochondrial gene

Abstract Freshwater shrimp are a rich species group, with a long and problematic taxonomic history attributed to their wide distribution and similar morphological characteristics. Shrimp diversity and species identification are important cornerstones for fisheries management. However, identification based on morphological characteristics is a difficult task for a nonspecialist. Abundant freshwater shrimp species are distributed in the waters of Henan Province, but investigations of freshwater shrimp are limited in this region, especially concerning molecular features. Here, we combined morphology and DNA barcodes to reveal the species diversity of freshwater shrimp in Henan province. A total of 1,200 freshwater shrimp samples were collected from 46 sampling sites, and 222 samples were chosen for further microscopic examination and molecular delimitation. We used tree‐based methods (NJ, ML, and bPTP) and distance‐based methods (estimation of the paired genetic distances and ABGD) to delimit species. The results showed that there were nine morphospecies based on morphological characteristics; all could effectively be defined by molecular methods, among which bPTP and ABGD defined 13 and 8 MOTUs, respectively. The estimation of the paired genetic distances of K2P and the p‐distances had similar results. Mean K2P distances and p‐distances within species were both equal to 1.2%. The maximum intraspecific genetic distances of all species were less than 2%, with the exception of Palaemon modestus and M. maculatum. Various analyses have shown that P. modestus and M. maculatum have a large genetic differentiation, which may indicate the existence of cryptic species. By contrast, DNA barcoding could unambiguously discriminate 13 species and detect cryptic diversity. Our results demonstrate the high efficiency of DNA barcoding to delimit freshwater shrimp diversity and detect the presence of cryptic species.


| INTRODUC TI ON
Freshwater shrimp (Decapoda: Caridea: Caridean) are a highly species-rich group with a long taxonomic history. However, the taxonomic status of these shellfish is controversial (De Grave et al., 2014;Martin & Davis, 2001). There are about 770-800 Caridea species in freshwater habitats, accounting for about one-fifth of the described shrimp species (De Grave et al., 2015). At present, freshwater shrimp exist in seven Caridea families (De Grave et al., 2014). The two families Atyidae and Palaemonidae dominate, comprising 443 and 300 species, respectively, and accounting for 97.4% of freshwater shrimp species (De Grave et al., 2015). Shrimp are an important component of biodiversity, as they provide a source of animal protein for people. In addition, freshwater shrimp have significant economic and nutritional value and research significance (Holthuis, 1980;New & Nair, 2012). At present, Jamaica (Hunte, 1978), Japan (Suzuki et al., 1993), Myanmar (Cai & Ng, 2002), China (Li et al., 2007;Liang, 2004), and many Chinese provinces (Deng & Wu, 1997;Zheng, 1989;Zhu & Miao, 1990) have carried out studies on the species diversity of freshwater shrimp, but most of the early studies were based on traditional morphological characteristics. The molecular methods have been gradually applied to research on the diversity of freshwater shrimp in recent years (De Grave et al., 2008;Makombu et al., 2019;Mar et al., 2018;New & Nair, 2012).
Studying species diversity is basic to biological research, but it is also a huge challenge and a harsh burden (Hebert, Cywinska, et al., 2003). As the main method of species diversity research, traditional morphological identification has high requirements and restrictions on samples and researchers, and the identification results are affected by both subjective and objective factors (Carvalho et al., 2011;Hebert, Ratnasingham, 2003;Shen et al., 2016). Since the early 2000s, DNA barcode technology has rapidly developed and has gradually become one of the main methods for biological identification (Hebert, Ratnasingham, et al., 2003). Compared with traditional morphological identification, barcode technology has many advantages. First, DNA is more stable than morphological characteristics, because DNA characters are constant throughout development. However, morphological characteristics vary with age, developmental stage, environment, and other factors. For example, molecular identification of deformed and underdeveloped shrimp larvae has absolute advantages over morphological identification (Burghart et al., 2014;Lee & Kim, 2014). Second, one can obtain sample DNA through some small parts of tissues, secretions, and even an organism's living environment (Pont et al., 2018), which reduces the requirements of sampling (Chang et al., 2016). More importantly, DNA barcoding is easy to operate, fast, and efficient. Samples can be identified in batches, and the method requires less professional knowledge (Takahara et al., 2013;Tinacci et al., 2018). With the implementation of the Barcoding of Life project, DNA barcodes have been widely recognized as a basic tool for species identification, and the mitochondrial gene cytochrome c oxidase I (COI) serves as the core of the global animal biometric system could effectively distinguish species of Crustacea (Costa et al., 2007;Hebert, Cywinska, et al., 2003;Hebert, Ratnasingham, et al., 2003).
In the era of high-throughput sequencing, there is the probability of tentative, incorrect, or low-quality sequences being submitted to databases (Wong et al., 2011). Compared with the commonly used barcode databases GenBank (National Center for Biotechnology Information, NCBI), DDBJ (DNA Data Bank of Japan), and EMBL-EBI (The European Molecular Biology Laboratory-European Bioinformatics Institute), the BOLD (the Barcode of Life Database) database conducts strict review and screening of submitted data, and thus, it is relatively more accurate and applicable (Macher et al., 2017;Wang et al., 2009). In addition, with the acquisition of a large number of barcodes, there has been growing use of molecular approaches for species delimitation; this has improved the accuracy of species identification (Hebert & Gregory, 2005;Luo et al., 2018).
At present, tree-based methods, distance-based methods, and character-based methods are commonly used in DNA-barcoding studies (Birch et al., 2017). The combined use of multiple methods will make the results of species delimitation more objective and comprehensive (Schlick-Steiner et al., 2010). Therefore, as many different types of molecular methods as possible should be used for comprehensive species identification.
Henan province is located inland and harbors four major river systems, the Yellow River, the Yangtze River, the Huaihe River, and the Haihe River. Our investigation of fisheries in Henan Province has shown that there are abundant fishery resources, but research on the province's freshwater shrimp is relatively scarce, and thus, the status of freshwater shrimp species diversity is relatively unknown.
To date, eight species of shrimp have been reported; surveys have used traditional morphological recognition methods to identify 352 samples and describe eight species from 15 sampling points (Wang, 1989). In view of the above, it is important to enrich shrimprelated research in Henan province in order to append the list of shrimp species and to assess the biodiversity in this area.
Combining molecular and morphological evidence in taxonomy is advocated (DeSalle et al., 2005;Miralles & Vences, 2013), so both morphological identification and molecular definitions have been used for species identification of freshwater shrimp that covered most of rivers in Henan Province, China, in our study. In order to obtain more objective species identification, multiple methods were employed. The main aims of this study were (a) to assess the shrimp diversity based on morphological features; (b) to build a reference DNA-barcoding library for these morphological species, and (c) to detect whether cryptic diversity occurred in shrimp in the province.
Our study will provide helpful information for future conservation and fisheries management of the shrimp in Henan province.

K E Y W O R D S
COI, freshwater shrimps, species delimitation, species diversity 2 | MATERIAL S AND ME THODS

| Ethics statement
The study conformed to the National Institutes of Health Guide for the Care and Use of Laboratory Animals (NIH Publication No. 85-23, 1996) (2011).

| Sample collection
A total of 46 sampling sites were covered in this survey for collecting freshwater shrimp (Figure 1). The sampling sites covered the main streams and tributaries of the four major rivers (i.e., the Yangtze River, the Huaihe River, the Yellow River, and the Haihe River) of the province (Table S1). In this study, about 1,200 samples representing nine species, six genera, and four families were collected. Most of the shrimp were collected by shrimp traps, but many individuals were obtained from markets. The samples were preserved in 95% ethanol for subsequent morphological observation and molecular identification. All voucher specimens were stored in the Fisheries College of Henan Normal University.

| Morphological identification
Morphological identification was mainly classified in situ by visual inspection in the field, and then detailed morphological identification and classification were conducted in the laboratory by stereomicroscope microscopic examination. All samples were taxonomically classified based on the distinguishing morphological characters of the male collected specimens according to Liu (1955), Liang (2004, and Li et al. (2007).

| DNA extraction, amplification, and sequencing
According to the results of morphological identification, multiple representative individuals of each taxonomic group were selected for abdominal muscle sampling. The obtained tissue samples were immediately stored in 95% ethanol and numbered for DNA extraction. To ensure the coverage of each species, individuals with moderate body size were selected as far as possible for EP tube preservation and numbering, and the larger individuals were marked with winding coils.
The amplification of the COI gene was carried out by polymerase chain reaction (PCR). A 632 bp fragment was amplified using the forward primer (LCO1490: 5′-GGTCAACAAATCA TAAAGATATTGG-3′) and reverse primer (HCO2198: 5′-TAAACTTCAGGGTGACCAAAAAA TCA-3′) (Folmer et al., 1994). PCRs were performed in a total volume of 50 μl containing 50-100 ng DNA template, 5 µl of 10× PCR buffer, 1.5 mmol/L of MgCl 2 , 0.2 mmol/L of each dNTP, 2 unit (U) of Taq polymerase, and 0.2 µmol/L of each primer. Thermal cycling began with one cycle of pre-denaturation at 94°C for 5 min, 35 cycles of denaturation at 94°C for 30 s, annealing at 50°C for 45 s, extension F I G U R E 1 Sampling sites of freshwater shrimp in Henan province CHINA 117°30'0" 116°40'0" at 72°C for 45 s, and a final extension holding at 72°C for 7 min (Feng et al., 2008). The PCR products were separated by electrophoresis on 1.0% agarose gels.
Primer synthesis and DNA sequencing were conducted by commercial companies. Among the 222 specimens, 141 were sequenced in one direction (63.51%), and the other specimens were two-way sequenced. Except for the sequences obtained from the genomic DNA in this study, the other COI sequences were obtained from GenBank for comparative analyses (Table S2).
In this study, traditional morphological identification and a variety of different molecular methods were used for comprehensive analysis and species delimitation. Due to the uneven sampling and the differences in effective population sizes of species (Blair & Bryson, 2017), we chose Automatic Barcode Gap Discovery (ABGD) and Poisson tree processes (PTP) for quantifying and delimiting taxonomic diversity. The specific analysis is described below.

| Distance-based approaches
Given that previous studies showed that the use of the Kimura-2parameter (K2P) model in DNA-barcoding studies is poorly justified, but no more suitable model has been derived at present; therefore, in order to obviate the requirement for model correction in DNA barcoding, a p-distance model was used in our analysis and calculations, while the K2P model was also used (Srivathsan & Meier, 2012;Collins et al., 2012). The K2P and p-distance models were used to construct a neighbor-joining tree and to calculate the pairwise genetic distances using MEGA 7.0 (Kumar et al., 2016). The haplotype diversity and nucleotide diversity of COI sequences were calculated using DnaSP 5.0 (Librado & Rozas, 2009). Then, ML tree analysis was implemented using RaxmlGUI (Stamatakis, 2014) with the default parameters and 1,000 replications. In all trees, bootstrap values below 70% are not shown.
Each sequence was selected for further species confirmation by the IDENTIFICATION of BOLD and the BLAST of NCBI to evaluate the accuracy of the morphological identification and to obtain reference sequences with high relative similarity. In the selection of similar sequences, we have defined 97% as a relatively loose standard to indicate potential species identification (Wong & Hanner, 2008).
In this study, a total 42 COI sequences with high similarity were obtained by aligning from GenBank. Gammarus pisinnus (GenBank accession number: KF824592) was selected as outgroup. All novel sequences obtained in this study were submitted to GenBank, and their accession numbers are provided in the Electronic Appendix (Table S2).

| Tree-based approach
A large number of tests have shown that PTP is superior to GMYC (Generalized mixed Yule-coalescent) on simulated data, and the results are comparable to GMYC on real datasets. Meanwhile, PTP requires less data and only a simple phylogenetic tree (Luo et al., 2018;Zhang et al., 2013). Therefore, in this study, we chose PTP analysis to assist in species definition. PTP can delimit species based on the Phylogenetic Species Concept. Therefore, the entities output by PTP are in theory species. Bayesian Poisson tree process (bPTP) analysis was run on the web server (https://speci es.h-its.org/ptp) with 100,000 MCMC generations, and other parameters as default values (Stamatakis 2006). showed that some individuals of the above species had varying degrees of differences and changes in the length, shape, and tooth form of their rostrums. Those morphological changes were at times inconsistent with the descriptions in the literature (Li et al., 2007;Liang, 2004;Liu, 1955), even exceeding the range of variation of those described species. In addition, consistent with the general distribution trend of freshwater shrimp, specimens in this province mainly belonged to Macrobrachium and Neocaridina. Among these species, M. nipponense, N. denticulate, and N. davidi were dominant species in Henan Province.

| Database search
In general, our morphological identification results matched the BLASTN annotations of the NCBI and BOLD databases, with at least 97% identities (Wong & Hanner, 2008 suggesting effective identification of the species. The identities of Macrobrachium sp. "qilianensis" and P. modestus were all greater than 98%, but the search results showed that the sequence identities between P. modestus and the three unpublished M. sp. "qilianensis" in the library were also high (at times having the highest identities). At the same time, in the retrieval of M. sp. "qilianensis," the identities of this and two unpublished P. modestus were also relatively high. After verification, the above M. sp. "qilianensis" (Accession: FJ958200, FJ958201) was sourced from GenBank and was found to be a direct and unpublished submission by Cheng (2009). However, there was no corresponding morphological description, and species identification of M. sp. "qilianensis" was found in his study (Zhang et al., 2009  were divided into 91 haplotypes, with widely distributed species such as M. nipponensis, Palaemon modestus, and M. maculatum having greater genetic differentiation (Figure 3).
In view of the differences between the morphological and mo-

| Barcoding success
It is well known that taxonomic identification of organisms is the most fundamental and important task of all biological research (Luo et al., 2018). The early classification identification was mainly based on detailed morphological characteristics observation and anatomical structure verification by professional taxonomists; however, this task needs significant time and has high requirements for researchers and experimental specimens (Carvalho et al., 2011;Hebert, Ratnasingham, et al., 2003;Shen et al., 2016). In addition, there is always the demise of existing species and the emergence of new species; with the rapid development of science and technology, increasing numbers of new species have been discovered, so that the number of specialists in alpha taxonomy is not sufficient to carry out extensive and complex morphological identification (Oliver, 2015).
Our traditional morphological identification results showed that there are nine species of freshwater shrimp in Henan Province. On the whole, there was more obvious morphological variation in the widespread taxa such as Macrobrachium, Palaemon, and Neocaridina.
The rostrum variation in shape, length, and number of serrations of some individuals of N. denticulate, Neocaridina davidi, M. maculatum, and M. nipponense was obvious, even exceeding the definition range of those species' descriptions, and this may be caused by their wide distributions and geographical separation (Li et al., 2007;Liang, 2004). In addition, due to the severe morphological damage, samples S0400 and M0301 could not be identified. Therefore, traditional taxonomic recognition is not only complicated and difficult, but also not conducive to widespread implementation.
With the development of modern technology and the arrival of the molecular era, molecular identification has gradually become popular and has been widely used in biological identification. Since the first use of COI for species identification, it has been shown that this gene fragment can be used in "DNA barcoding" for biological authentication in many invertebrate species (Barrett & Hebert, 2005;Clare et al., 2007;Hebert, Ratnasingham, et al., 2003;Hendrich et al., 2014). The research of Costa and Mar and colleagues further demonstrated that barcode technology is efficient and accurate in the species identification of the freshwater shrimp (Costa et al., 2007;Mar et al., 2018). Our study showed that both the identification results of the NJ phylogenetic analysis and the bPTP analysis identified at least 13 MOTUs among the freshwater shrimp in Henan Province. There was a close evolutionary relationship between M. were calculated as a whole, the genetic distance within species was 0.025, beyond the intraspecific threshold. Our molecular identification results show that COI DNA barcode technology can not only effectively identify species identified by morphology but also identify species that are nearly identical in terms of morphology.
The results of the study show that all nine species identified by traditional morphology could be further divided and confirmed by molecular methods. The molecular analysis identified N. ikiensis (M0301), N. palmata (S0400), and M. sp. "qilianensis," three additional species. N. ikiensis (M0301) and N. palmata (S0400) were morphologically identified as Neocaridina due to severe morphological damage. Our study has shown that the number of species identified by molecular biological identification is usually higher than that using traditional morphology, and it also demonstrated that the COI DNA barcode technology is efficient in the species identification of freshwater shrimp.

| Species diversity
The morphological identification results showed that there are nine species of freshwater shrimp in Henan Province. Compared with the study of Wang (1989), our sampling points covered his 15 sampling points plus the main river systems and tributaries in Henan Province.
Unfortunately, we have not collected and identified Macrobrachium superbum, Macrobrachium asperulum, or Macrobrachium iusulare. In order to avoid the single sampling error, we repeatedly went to the collection sites where the distributions were recorded, and the collection range was further expanded. Even so, we have not collected these species. The records indicate that the above three freshwater shrimp are mainly distributed in some provinces and waters of southern China (Li et al., 2007), and the morphological characteristics of Macrobrachium are similar, making the species difficult to identify. Therefore, we hypothesize that these species may have existed in Henan Province before, but the environmental changes of the sample sites may have proven unsuitable for these species and that they have migrated or disappeared from the province. In addition, they may never have been distributed in Henan Province, and similar morphological characteristics may have led to their incorrect identification. All in all, more samples and more direct evidence are needed to support the existence of these species in Henan Province.
At present, the classification status of a variety of freshwater shrimp has changed, indirectly hindering the effective identification of their species and the estimation of biodiversity. First, the taxonomic status of Caridina denticulata sinensis (Kemp, 1918) and Palaemon (Exopalaemon) modestus (Heller, 1862) collected by Wang has been controversial and has changed to some extent (Wang, 1989 (Cai, 1996). In this revision, Cai considered that C. davidi (Bouvier, 1904) was a subspecies of N. denticulata (N. denticulata davidi) and transferred it to the genus Neocaridina. However, Liang considered C. davidi (Bouvier, 1904), N. denticula davidi (Kubo, 1938), and N. denticula sinensis (Kemp, 1913) as synonyms of Neocaridina heteropoda heteropoda (Liang, 2002). Our molecular and morphological identification results also confirmed this point (Klotz et al., 2013;Liang, 2004). Klotz pointed out that N. denticulata sinensis reported by Englund and Cai (1999) and N. davidi reported here are conspecifics (Klotz et al., 2013). Here, we followed Klotz et al. (2013) and considered that C. davidi (Bouvier, 1904) as the senior synonym has clear priority (article 23 of the ICZN), and we continue to name it N. davidi (Klotz et al., 2013). In addition, Palaemonetes, Exopalaemon, and Coutierella have been transferred to Palaemon, and this is widely accepted (Ashelby et al., 2012). Due to the genus classification status changes, Palaemon (Exopalaemon) modestus should be renamed Palaemon modestus, and Palaemonetes sinensis should also be renamed as Palaemon sinensis.
Second, due to the failure to identify enough morphological differentiation in M. sp. "qilianensis," and the lack of a sufficient description in the relevant references and original literature, we tentatively inferred that M. sp. "qilianensis" may be an invalid species and that it may be a synonym of Palaemon modestus. In addition, given that only one sample was obtained, and N. ikiensis and N. palmata were damaged, they cannot be effectively identified by morphology. Thus, N. ikiensis and N. palmata need to be further collected and confirmed.

TA B L E 2
The genetic distances of the four-water system freshwater shrimp populations in Henan Province Note: The list of K2P is K2P genetic distance within populations; diagonal bold is P-distance genetic distances within populations; below diagonal is K2P genetic distance among populations; above diagonal is P-distance genetic distance among populations.
In conclusion, the comprehensive results of morphological characteristics and molecular delimitation indicated that there are at least nine species of freshwater shrimp that have been morphologically identified in Henan Province.

| Cryptic species
The aims of DNA barcoding are identification of unknown specimens via DNA barcodes of a priori defined taxonomic entities in databases (Merckelbach & Borges, 2020). The method is being increasingly utilized to tackle many issues, including illegal species exploitation, food fraud, biological invasions, and biodiversity monitoring (Bohmann et al., 2014;Gonçalves et al., 2015;Hubert et al., 2015;Khaksar et al., 2015). The DNA barcode solves the problem of molecular delimitation of species to a certain extent, but to rely on it exclusively is far from sufficient to solve the delimitation of species and the dis- Both molecular and morphological characteristics showed that there were significant genetic differentiation and morphological differences between the above species, but there is no definitive criterion for whether these differences are sufficient to indicate the emergence of a new species or the existence of an underlying species.
In the process of speciation, the boundaries of new species become clearer over time. However, before the completion of this process (known as gray zone sense), the boundaries between species are often fuzzy and difficult to recognize. Cryptic species are the intermediate products or even final products of this process (De Queiroz, 2007). Species delimitation studies are dedicated to defining the species that are unknown or problematic by compiling molecular, morphological, and karyotype data (Kekkonen & Hebert, 2014).
This analysis is usually applicable to the groups for which there has been substantial research, but its ability to define many taxonomic species with less basic knowledge and description is limited (Common, 1990;Raven & Yeates, 2014). In fact, even though there is sufficient evidence to support the species hypothesis and species delimiting, there are still many newly discovered species that have not been described (Pante et al., 2015), a situation that hinders taxonomic progress, species identification, and biodiversity estimation (Schlick-Steiner et al., 2007). Thus, if a species is marked as merely presumed rather than formally described and therefore fully established, the taxonomy is still incomplete; so, the transition from species delimitation to species description is still a major task to be accomplished (Merckelbach & Borges, 2020;Miralles & Vences, 2013).
In our results, the delimitation of almost all species of freshwater shrimp was in accordance with the genetic and morphological definitions, and most of the molecular delimitation analyses showed a higher species number than those indicated by morphological identification. This suggests that there are likely to be cryptic species that have yet to be identified and described, even if they are not sufficiently differentiated to support the formation of a single new species. The analysis also shows that the ability of DNA barcodes to identify the undescribed species from recent speciation events is limited, although it can be widely used to identify new taxa in complex groups, identify unknown species, and find cryptic species (Iyiola et al., 2018). Further studies and descriptions of species are needed to determine whether the intermediate process of a species' differentiation is sufficient to form a new species, and whether there are cryptic species.

ACK N OWLED G M ENTS
We are grateful to the anonymous reviewers for their constructive comments. We thank Xi Wang, Hui-hui Wu, Ru-yao Liu, and Xuemeng Yan for assistance with fieldwork and their help. In addition, this work was supported by the following funding: the National Natural Science Foundation of China (31872199, U2004146).

Support for this study was provided by The High Performance
Computing Center of Henan Normal University. We thank these funding agencies for their support.

CO N FLI C T O F I NTE R E S T S
All authors declare that they have no competing interests.  Table S2.