Diversity of coral-associated pit crabs (Crustacea: Decapoda: Cryptochiridae) from Hong Kong, with description of two new species of Lithoscaptus A. Milne-Edwards, 1862

Highly specialized cryptochirid crabs are obligate symbionts of scleractinian corals in tropical and subtropical seas. General morphologies of cryptochirid crabs remain poorly described due to their small size and difficulties in collection; thus, the current inventory is probably an underestimation. In the present study, we sampled cryptochirid crabs from coral communities in Hong Kong. In the literature, only Cryptochirus hongkongensis (now Neotroglocarcinus hongkongensis) with unknown hosts had been recorded in Hong Kong since 1936. In addition to morphological examination, identification in the present study is further supported by sequence divergence of mitochondrial cytochrome c oxidase I (COI) and 16S ribosomal DNA markers. Six operative taxonomic units (OTUs), representing four species and one species complex with two species, were revealed among our material: Cryptochirus coralliodytes, Lithoscaptus paradoxus, Lithoscaptus doughnut sp. nov., Lithoscaptus scottae sp. nov., and Xynomaia sheni species complex. Morphological description of these species is provided, including description of the two new pseudocryptic species. The hosts of the genus Lithoscaptus belong largely to the Merulinidae, while L. doughnut sp. nov. inhabits the Plesiastreidae.

Highly specialized cryptochirid crabs are obligate symbionts of scleractinian corals in tropical and subtropical seas. General morphologies of cryptochirid crabs remain poorly described due to their small size and difficulties in collection; thus, the current inventory is probably an underestimation. In the present study, we sampled cryptochirid crabs from coral communities in Hong Kong. In the literature, only Cryptochirus hongkongensis (now Neotroglocarcinus hongkongensis) with unknown hosts had been recorded in Hong Kong since 1936. In addition to morphological examination, identification in the present study is further supported by sequence divergence of mitochondrial cytochrome c oxidase I (COI) and 16S ribosomal DNA markers. Six operative taxonomic units (OTUs), representing four species and one species complex with two species, were revealed among our material: Cryptochirus coralliodytes, Lithoscaptus paradoxus, Lithoscaptus doughnut sp. nov., Lithoscaptus scottae sp. nov., and Xynomaia sheni species complex. Morphological description of these species is provided, including description of the two new pseudocryptic species. The hosts of the genus Lithoscaptus belong largely to the Merulinidae, while L. doughnut sp. nov. inhabits the Plesiastreidae.

Introduction
Coral reef ecosystems are highly productive, harboring remarkable species diversity (Crossland et al., 1991;Reaka-Kudla, 1997). Recent estimates on species associated with coral reefs range from 550,000 to 1,330,000 (Fisher et al., 2015), and over 91% of such species remain to be described (Mora et al., 2011). Decapod crustaceans are an important component of the tropical reef fauna (Castro, 1976;Ross, 1983), among which several brachyuran lineages, including the Cryptochiroidea and Trapezoidea (Domeciidae, Tetraliidae and Trapeziidae), and numerous xanthioid (e.g., Cymo) and pilumnoid (Tanaocheles) species (Castro, 2015), are symbiotic with scleractinian corals. While some of these lineages are "facultative symbionts" (Castro, 1976), none are as specialized as the Cryptochiridae with mature females have pleon modified as an inflated, egg-carrying pouch, with a size comparable to the rest of the individual, often sedentary in domiciles on the surface of scleractinian corals, and sacrificing mobility for physical protection and reproductive success (Vehof et al., 2014). This highly specialized niche is comparable to those of pea crabs of the family Pinnotheridae, which were found to show obligate associations with hosts including edible bivalves and gastropods, and ascidians, holothurians, and echinoids (see De Gier and Becker, 2020).
Members of Cryptochiridae are often referred to as "gall crabs." However, the form of domicile differs substantially within the family. One form inhabits branching corals of the Pocilliporidae, which induces development of an enclosed chamber of two hemispheres of host tissue (e.g., Potts, 1915;Hiro, 1937). Others settle on the surface of massive corals during the megalopa stage, inhibiting the growth of coral polyp at that spot, and from there, they excavate pits or channels of various forms, often leaving a shallow depression around the opening (Hiro, 1937;Simon-Blencher and Achituv, 1997). These two forms exhibit distinct feeding mechanisms (Abelson et al., 1991). Following the definition of galls in plants (Fernandes et al., 2011), typical domiciles induced by Hapalocarcinus can be recognized as true galls (thus "gall crabs"). As elaborated by Abelson et al. (1991), the term "pit crabs" might be more appropriate for those pit excavators living especially in massive corals. The several species herein reported from Hong Kong can be referred to as "pit crabs," and we refrain from referring cryptochirids exclusively as "gall crabs." However, given the considerable diversity of cryptochirid domiciles, such as those lodged between septa of mushroom coral of the Fungiidae (Hoeksema et al., 2012;van der Meij et al., 2015), and some forming a canopy-like structure, partially sheltering the opening in hosts of the Agariciidae Garcıá-Hernańdez et al., 2020), further definition of common names of cryptochirid crabs based on their domicile morphology may require further investigations.
Cryptic species (morphologically indistinguishable) and pseudocryptic species (minor morphological difference) are biologically distinct species that are erroneously classified (thus hidden) under one species name (Bickford et al., 2007;Lajus et al., 2015), and various cryptochirid lineages may contain previously unrecognized, cryptic diversity. In an alternative understanding, cryptochirids can be "cryptic" for being small in size, well camouflaged, and inhabiting poorly surveyed habitats, thus difficult to sample (see Hoeksema, 2017). In the immensely species-rich region of the Indo-West Pacific, investigation of cryptochirid diversity remains fragmented, despite the discovery of numerous new taxa on hosts previously unreported in the past decade (e.g., van der Meij, 2014;van der Meij, 2015a;van der Meij, 2015b;van der Meij, 2017). In Hong Kong, Cryptochirus hongkongensis (now Neotroglocarcinus hongkongensis) had been the only species of the Cryptochiridae known prior to this study, then described without reporting on its host (Shen, 1936). Van der Meij (2012) added a tentative record of Pseudocryptochirus viridis based on an image showing a domicile opening in a guidebook on the corals of Hong Kong (Scott, 1984). In this paper, we describe the cryptochirid fauna of Hong Kong, comprising of at least five species of pit crabs unrecorded in the literature, including two new pseudocryptic species of Lithoscaptus. This study is part of a study aiming to understand the diversity and biogeography of cryptochirids and host relations of coral-associated fauna.

Materials and methods
Surveyed sites, specimen collection, and morphological examination Hong Kong is located along the northern limit of the Tropics in the Northern Hemisphere, east of the Pearl River outlet (Zhujiang). Given the massive freshwater runoff of some 300 billion m 3 discharged seasonally, territorial seas of Hong Kong comprises of a west-to-east decreasing gradient of fluvial influences, reaching full oceanic conditions in the eastern seas (see Morton et al., 1996). This heterogeneity contributes to diversity of marine habitats and thus inhabited species. Under these conditions, scleractinian corals occur in eastern and northeastern waters as communities on substrates, and a total of 84 species is found in the territorial seas (Chan et al., 2005). Since rehabilitation from severe coastal pollution and disastrous habitat degradation (Morton, 1989;Scott and Cope, 1990), for the past two decades, natural recovery appears limited and difficult (KT Wong et al., 2018;Yeung et al., 2021a), while the process is anticipated to persist in extended time periods (Goodkin et al., 2011).
Six shallow-water sites, all of considerable scleractinian coverage (Yeung et al., 2021a), were surveyed from 2012 to 2019, during the implementation of coral bioerosion and coral bleaching projects (Xie et al., 2016;Yeung et al., 2021b;Zhang et al., 2022). Five of these sites were near Sai Kung, and one in Mirs Bay, northeast of the territory, and locations of these sites are shown in Figure 1. In general, coral coverage of these sites was distributed from the intertidal zone to depths <5 m, dominated by stress-tolerant scleractinian species, such as those belonging to the genera Psammocora, Pavona, Favites, and Platygyra, while the below the deeper reef zone, there were sandy or gritty bottoms of poor visibility.
Domicile openings of cryptochirids, as shown in Figures 2-4, were visually searched underwater during SCUBA diving. These openings were not analyzed in several previous studies of coral borer, such as dumbbell-shaped openings created by Lithophaga mussels and the circular ones created by Spirobranchus tetraceros (Xie et al., 2016), and colony surfaces immediately around openings were observed to suffer from lesions and prone to diseases ( KT Wong et al., 2015). Domicile openings and host corals were photographed in situ. The inhabiting crabs, along with the domicile and small fragments of the host, were retrieved and preserved in 95% ethanol. Sampled crabs ( Figure 5) were examined under a stereomicroscope (Olympus SZX7) and preliminarily sorted based on morphological characters presented in taxonomic works (Fize and Serène, 1957;Kropp, 1988a;Kropp, 1990). Line drawings were drawn based on structures photographed under a digital camera (Panasonic DM C-GH4). For collected crabs, taxonomic schemes, measurements, and morphological terminology follow those of Kropp (1990) and Davie et al. (2015). Abbreviations CW, CL, Mxl, Mxp, P, Plp, and G, respectively, represents carapace width and length, maxilla (1 and 2), maxilliped (1-3), pereiopod (1-5), female pleopod (1-3), and male gonopod (1 and 2). Host corals were identified from in situ photographs and retrieved fragments based on works of Scott (1984); Chan et al. (2005), and Dai and Cheng (2020). Images of preserved material are printed in monochrome. The material examined in this study was deposited into the collections of the Biodiversity Research Museum, Academia Sinica (ASIZCR) and Coastal Ecology Laboratory (CEL), Biodiversity Center, Academia Sinica, Taipei, and Swire Institute of Marine Science, the University of Hong Kong, Hong Kong (SWIMS). For all taxa mentioned in the text, the authority and year of original publication are enumerated in Appendix 1, and full references are not provided for simplicity.

Molecular analysis
Total genomic DNA was extracted from eggs of females, or pereiopod 5 of male crab specimens by using DNeasy ® Blood and Tissue Kit (Qiagen, CA, USA) according to instructions provided by the manufacturer. Partial sequences of two mitochondrial DNA markers (COI and 16S rDNA) were amplified following the protocol from previous studies: those of COI using primers LCO1490 and HC02198 (Folmer et al., 1994;Feller et al., 2013) and of 16S rDNA using 1471 and 1472 (Crandall and Fitzpatrick, 1996). Polymerase chain reactions (PCRs) were conducted in DNA Engine Thermal Cycler (Bio-Rad, Richmond, CA, USA), and the products were checked by electrophoresis on 1.5% agarose gel in 1× TAE buffer. DNA purification and Sanger DNA sequencing were performed by Genomics BioSci and Tech Ltd. (New Taipei City, Taiwan). The sequences were assembled and edited in Geneious 7.0.6 (https:// www.geneious.com).
The sequences were aligned with MUSCLE implemented in MEGA XI (Ver. 11.0.13; https://www.megasoftware.net/; Tamura et al., 2021), and species identification and delimitation by molecular evidence are addressed by phylogenetic affinities and genetic distances. Given close phylogenetic proximity of the Dotillidae and Cryptochiridae under the Thoracotremata , the sand bubbler crab Scopimera globosa (Family Dotillidae) was included as outgroup (accession number LC535358.1). Neighbor-joining (NJ) analysis was performed using Kimura 2-parameter (K2P) distance model, with gaps or missing data treated using "pairwise deletion," and bootstraps values estimated from 1,000 pseudoreplicates implemented were identified for both COI and 16S rDNA markers in MEGA XI.
Levels of K2P genetic distances were also calculated by MEGA XI. Values of genetic distances are indicated as mean ± standard deviation. In terms of genetic distances, interspecific discrepancies published on various thoracotreme crabs are taken into consideration as thresholds for species delimitation. This figure varies from 1.49% between Parasesarma liho and Parasesarma paucitorum (Shih et al., 2019; both now Leptarma), 2.79% between Paraleptuca crassipes and Paraleptuca splendida , 4.39% between Mictyris brevidactylus and Mictyris guinotae (Davie et al., 2010), to 6.25% between Ocypode stimpsoni and Ocypode mortoni (KJH Wong et al., 2012). These values serve as references in considering thresholds in species delimitation.

Species identification by COI and 16S rDNA sequencing
In total, 27 16S rDNA sequences and 38 COI mtDNA sequences were extracted from our specimens, and another 42 16S rDNA and 49 COI mtDNA sequences downloaded from Genbank were respectively added to both alignments as reference sequences. Accession numbers of query and reference sequences are provided in Table 1. Alignments of 615 and 625 bp were constructed for markers 16s rDNA and COI, respectively. NJ trees that resulted from analyses using the two markers are shown as Figure 6A (16S rDNA) and Figure 6B (COI mtDNA). Based on query and reference sequences, intraspecific K2P distances of cryptochirids have a mean of 1.06 ± 0.76% and interspecific (intrageneric) distances at 7.16 ± 3.27%. The distribution of these values is shown in Figure 7. Among Lithoscaptus species, including forms listed as tentative genetic identifications (Lithoscaptus sp. A, C, D, Z), this value ranges from 2.80% to 14.97% (mean 9.22 ± 2.60%), and among described species, the lowest pairwise distance was observed between Lithoscaptus tuerkayi (KU745732.1) and Lithoscaptus hellerii (KU041819.1) at 3.57%. We do not calculate the frequencies of pairwise distances in 16S rDNA sequences due to its poor resolution in performed analyses for identification (as in NJ tree in Figure 6A; see below).
For phylogenetic affinities between query and reference sequences, only one sequence (CEL-Hapa-022) clustered with available 16S rDNA reference sequence of C. coralliodytes collected from New Caledonia (KM114587.1; K2P distance, 0.36%). None of the other 26 sequences clustered with any reference sequences ( Figure 6A; Table 2). Analyses on COI sequences provided results of better resolution. Sequences from the local material clustered with four reference sequences with high bootstrap values (>80) and can be differentiated into at least six distinct operative taxonomic units (OTUs  Figure 6B) and genetic distances, at least two distinct OTUs can be recognized.
See below for elaborations on the identification of "X. sheni species complex." The remaining 14 COI sequences (CEL-Hapa-015, 017-019, 023-029, and 035-037) form a single clade (within group K2P distance 1.20 ± 0.55%), a sister group to that containing L. paradoxus, the latter of which included reference sequences from Guam (KU041820.1) and Indonesia (KU041825.1) (between group K2P distance, 4.05 ± 0.87%). This clade does not cluster or show specific affinities with available reference sequences of Lithoscaptus, including four from undescribed species sequenced by van der Meij and Nieman (2016), which were all from northern Sulawesi (for the accession number, see Table 1) and is referred as L. scottae sp. nov. Two specimens (CEL-Hapa-006 and 040) retrieved from hosts of the Plesiastreidae, distinct from the rest, differ from each other by a K2P distance of 2.73%, marginal for interspecific divergences. CEL-Hapa-040 is genetically distinct from both L. paradoxus and L. scottae sp. nov. by 6.00 ± 0.93% and 5.80 ± 0.73%, respectively, whereas CEL-Hapa-006 differs from these two taxa by 3.35 ± 0.73% and 3.31 ± 0.87%. With both only represented by only one specimen each, we tentatively refer the two as L. doughnut sp. nov. (CEL-Hapa-040) and L. cf. d o u g h n u t ( C E L -H a p a -0 0 6 ) . T h e s e t a x a c a n b e morphologically differentiated by subtle characters and separately addressed under systematic account below, including diagnoses and descriptions of the two new pseudocryptic species of Lithoscaptus.   van der Meij, 2014;b, van der Meij, 2015a;c, van der Meij, 2015b;d, van der Meij, 2017;e, van der Meij and Nieman, 2016;f, van der Meij and Reijnen, 2014;g, van der Meij and Schubart, 2014.
Description (based on CEL-Hapa-022, female 5.3 × 7.5 mm). Carapace 1.4 times longer than broad, anteriorly ovate in outline, posteriorly subquadrate, overall pronouncedly sculptured ( Figure 8A); anterior 2/5 depressed, strongly Scopimera globosa (OUTGROUP) LC535358.1 Kobayashi et al., 2021 deflexed, broadest, and most elevated at approximately half of CL ( Figure 8B); front broadly concave, inner orbital lobe convex, inflated, armed with numerous slender spines; mesogastric region scattered with small conical spines, each well-spaced from one another, followed by metagastric region as a circular cluster of densely aligned rounded tubercles, posteriorly separated from cardiac-intestine region by shallow but distinct transverse groove, posterior of which covered with numerous isolated, rounded tubercles; exorbital angle crested, confluent with anterolateral margin, which raised, mildly cristate, lined with series of acute spines, extending to 1/3 of CL; hepatic region depressed, regions at near base of eyestalk and behind inner orbital lobe sunken, giving an eroded texture; proto-and mesobranchial regions lined with three sets of short, deeply incised, oblique grooves, between which each furnished with dense cluster of rounded tubercles: first set lateral to mesogastric region, second lateral to metagastric region, third anterolaterally delineating cardiac-intestine region ( Figure 8A). Pterygostomial region mildly granular, completely fused dorsally with carapace ( Figure 8B). Basal plate of antennular peduncle longitudinally ovate, anteriorly armed with series of stout teeth, dorsally depressed, sunken medially, slightly inflated rim lined with flattened granules, ventrally densely granular, nearly flat ( Figures 8A,  B). Eyestalk short and stout, cylindrical, slightly concave along mesial margin, basal of cornea lined with several small spines on dorsal surface ( Figure 8A). Epistome medially elevated, faintly crested, extending to anterior apex along midline, laterally each of a well-defined longitudinal crest ( Figure 8C). Mxp3 ischium depressed, covered with low rounded granules, merus distalexternal lobe triangular, moderately produced; carpus dilated along internal margin; exopod elongated ovate ( Figure 8D). Mxp1 endopod elongated-triangular, mesial margin strongly convex ( Figure 8F).

FIGURE 6
Neighbor-joining (NJ) tree of 16S rDNA (A) and COI sequences (B). Branch length represents Kimura 2-parameter (K2P) distances, and bootstrap values are shown on the nodes when >80. Details of specimens from the present study (clades highlighted in gray), and reference sequences acquired from Genbank are listed in Appendix 1. Resolution of molecular analyses based on 16S rDNA sequences (A) do not allow identification of clades containing Lithoscaptus (*) and Xynomaia species complex (**). Depictions of both clades are inferred from reconstruction based on COI sequences (B).
As noted above, the only specimen that we examined (CEL-Hapa-022) was identified as the present species by both genetic markers, which matched the Cryptochirus coralliodytes from Malaysia or Indonesia (16S rDNA: KM114587.1) and New Caledonia (COI: KU041822.1) with K2P distance between both, respectively, at 0.36% for and 1.13% (Figures 6A, B; Table 2). For the NJ tree that resulted from analyses on the COI sequences, the clade containing sequences of C. coralliodytes from Hong Kong and New Caledonia, despite low support values, shows affiliation, or nested within a lineage inclusive of Lithoscaptus, Xynomaia, Fungicola, and Pelycomaia ( Figure 6B). Lithoscaptus has been demonstrated to be composite by van der Meij and Nieman (2016) (see below). Genetic (K2P) distance at 1.13% for COI falls within the range of intraspecific divergence among cryptochirids (Figure 7). This species can be morphologically identified from the sympatric Lithoscaptus spp. by dorsal ornamentations of carapace, characterized by having more pronounced grooves, and gastric region furnished with dense clump of rounded tubercles (Kropp, 1988a). In this treatment, Cryptochirus rugosus Edmondson, 1933 was placed under synonymy of C. coralliodytes. As enumerated above, the host range of this species appears to be much broader than other Indo-West Pacific taxa. Past host records would require verification. See Remarks under L. paradoxus below. 3.1 mm, cw 6.0 mm-5.4 × 7.5 mm; CEL-Hapa-010-012), Tai  Diagnosis. Carapace longitudinally ovate, anterior half depressed, deflexed, regions moderately defined, anteriorly by depression lateral to gastric region, scattered with small conical spines, posterior half of low, well-separated rounded tubercles, cardiac region anteriorly and laterally defined by two arc-shaped grooves, mesially notyt connected. Epistome medially elevated but nor crested, laterally each of one longitudinal crest. Female thoracic sternum relatively narrow, medially depressed, anterior plate spade shaped, approximately as long as broad, surface mildly granular; sutures 4/5, 5/6, and 7/8 medially interrupted, suture 6/7 nearly confluent, sternite 7 medially of well-defined median line; gonopore on sterniten 6 as a narrow, oblique slit, sheltered by a narrow eave-like structure. Plp2 and Plp3 uniramous.
Cheliped symmetrical, merus to chela strongly compressed, carpus and chela dorsally lined with small spines; chela palm longer than fingers, externally smooth, fingers slender, mildly deflexed, tapering into fine tips ( Figure 9I). P2-P4 short and robust, meri compressed, carpi and propodi armed with series of spines or stout nodules, dactyli slender, claw-shaped, shorter than respective dactylus ( Figures 9J-L). P2 merus subovate, approximately 2.1 times as long as broad, distally of small stout spines, posteriorly armed with one small spine; carpus and propodus of acute spines; dactylus proximally armed with small spinules on extensor margin ( Figure 9J). P3 merus approximately 1.4 times as long as broad, distally of numerous stout spines, posteriorly lined with one small spine; carpus and propodus externally lined with series of acute teeth; dactylus proximally armed with small spinules on extensor margin ( Figure 9K). P4 merus approximately 1.4 times as long as broad, distally of patch of stout nodules, posteriorly lined with small acute spinule; carpus and propodus externally lined with stout nodules ( Figure 9L). P5 elongated, segments cylindrical, merus 1.3 times as long as broad, nearly smooth, unarmed; carpus and propodus laterally furnished with numerous flattened, round tubercles; dactylus slender and elongated, tapering into an acute tip, curving and articulating ventrally ( Figures 9M, N).
Two characteristics, however, namely, fusion of pterygostomian plate and surface texture of anterior extension of female thoracic sternites, would require elaboration. In Lithoscaptus, the structured was defined as "fused to carapace" (Kropp, 1988a;Kropp, 1990). In our material identified as L. paradoxus, L. doughnut sp. nov., and L. scottae sp. nov., however, we find the pterygostomial plate functionally fused with carapace, but an inconspicuous but discernable suture can be observed. This fine suture can be made visible with application of ethanol-soluble dye and indicated in the line drawings herein provided (L. paradoxus: Figure 9B; L. doughnut: Figure 10B; L. cf. doughnut: Figure 11B; L. scottae: Figure 12B). In contrast, this suture is completely absent in C. coralliodytes ( Figure 8B) but more visible in our Xynomaia species ( Figure 14B). In describing L. prionotus, a somewhat aberrant member under the genus, Kropp (1994) casted doubt on whether the form of pterygostomial region can be considered one of the diagnostic features of the genus. As for the surface texture of anterior plate of female thoracic sternum, the structure was reported to be furnished with transverse band of granules in C. coralliodytes while smooth in L. paradoxus (Kropp, 1990). Among our material, that of C. coralliodytes is indeed granular ( Figure 8H), while those of L. paradoxus, L. doughnut, L. cf. doughnut, and L. scottae are finely granular and/or punctate ( Figures 9H, 10H, 11H, and 12H, respectively). Among past records of this genus, the latter had generally been neglected as a potential diagnostic feature, and the structure was illustrated for only 3 out of 10 previously described species [L. paradoxus: fig. 4c in Kropp (1988a); L. prionotus: fig. 4c in Kropp (1994); L. tuerkayi: fig. 4b  Diagnosis. Carapace longitudinally ovate, broader posteriorly, anterior half depressed, regions mildly defined, surface of scattered, isolated rounded tubercles interspaced with short conical spines, metagastric region laterally each defined by brachet-shaped groove. Epistome of two longitudinal crests. Female thoracic sternum relatively narrow, medially depressed, anterior plate transversely octagonal, 1.6 times broader than long, surface mildly granular, strongly sculptured; sutures 4/5, 5/6, and 7/8 medially interrupted, suture 6/7 nearly confluent; sternite 7 medially of well-defined median line; gonopore on sternite 6 as an oblique slit, laterally sheltered by an eave-like structure. Plp2 and Plp3 uniramous.
Cheliped symmetrical, merus to chela compressed; dorsal margins of carpus and palm lined with fine spinules; chela palm longer than fingers, external surface smooth; fingers slender, moderately deflexed, tapering into fine chitinous tips ( Figure 10I). P2-P4 short and robust, overall decreasing in size and acuteness of armature, meri of all compressed, each carpi and propodi externally lined with acute spines, dactyli shorter than respective propodus ( Figures 10J-M). P2 merus longitudinally ovate, 1.6 times as long as broad, distally armed with series of similar-sized acute teeth, posteriorly lined with inconspicuous granules, dactylus slender, shorter than propodus, externally armed with small spine ( Figure 10J). P3 merus 1.2 times as long as broad, distally of small spines and nodules, posteriorly of one stout spine, dactylus externally armed with small spine ( Figure 10K). P4 merus 1.4 times as long as broad, posteriorly of one small spine, distally of blunt nodules ( Figure 10L). P5 merus robust, nearly cylindrical, anteriorly of small tubercles, carpus elongated, externally of small tubercles ( Figure 10M).
Host coral. Plesiastrea peroni of the Plesiastreidae. Etymology. Specific epithet alludes to the loose resemblance between corallites of the host coral with the sugary treat, in the eyes of a snack-indulged graduate student. The name is used here as a noun in apposition.
Type locality. Basalt Island, Sai Kung, Hong Kong. Geographical distribution. So far only from type locality.
Remarks. The genus Lithoscaptus now contains 12 described species (updated from Ng et al., 2008), and recent descriptions include L. semperi van der Meij, 2015b, L. tuerkayi van der Meij, 2017, and in the present study, L. doughnut sp. nov. and L. scottae sp. nov. Several Indo-Pacific forms, being genetically distinct, are yet formally described (van der Meij and Nieman, 2016). As demonstrated by van der Meij and Nieman (2016), this genus is clearly heterogeneous, with genera such as Cryptochirus, Pelycomaia, and Xynomaia nested within (see below).
For the host of L. doughnut sp. nov., the coral genus Plesiastrea currently comprises of two species, including P. versipora, which were once considered to be the single species distributing across the Indo-Pacific but now shown to confined to temperate waters of southern Australia and a recently resurrected P. peroni, a tropical species (Juszkiewicz et al., 2022). The family Plesiastreidae is now first reported as a host species of cryptochirid crabs. This host species had long been placed under the Faviidae sensu lato, but recent molecular data showed which represents a distinct lineage under the "robust group," basal to multiple genera of the Merulinidae and Montastraeidae (Fukami et al., 2008;Benzoni et al., 2011;Arrigoni et al., 2012). The association is also surprising among Lithoscaptus species, which are all largely symbionts of the Merulinidae (sensu Huang et al., 2014).
For another resembling form found infesting also P. peroni, with likewise one female specimen examined, despite considerable morphological distinctions and moderate genetic distance, we prefer to stay conservative in reporting which is L. cf. doughnut as below, at least for the time being.
Description (based on CEL-Hapa-006, female 4.8 × 6.8 mm). Carapace longitudinally ovate, 1.4 times as long as broad, broadest, and most elevated at approximately half-CL ( Figures 11A, B); anterior half depressed, markedly deflexed, hepatic region sunken, strongly sculptured, more invaginated lateral to inner orbital lobes; posterior half inflated, roundish in outline, overall covered by rounded, bead-like tubercles of similar sizes, well isolated from each other, diminishing near posterior margin; front nearly transverse, lined with minute spines; inner orbital lobe broad-triangular, inflated, anteriorly armed with small slender spines, exorbital angle projecting, cristate, confluent with anterolateral margin, lined with series of slender, conical teeth, extending about 2/5 of CL; gastric region broad-triangular, mildly inflated, sparsely furnished by small, indistinct tubercles, mesogastric region lined by two broad, short longitudinal groove, laterally each defined by a much deeper, well-incised short longitudinal groove; cardiacintestine region well defined anterolaterally by deeply incised arc-shaped grooves (╭ and ╮), both not connect medially ( Figure 11A). Pterygostomian region finely granular, rhomboid plate of which fused dorsally with carapace, suture discernible ( Figure 11B).
Cheliped symmetrical, merus to chela compressed; carpus and palm dorsally lined with fine spinules; chela palm longer than fingers, external surface smooth, fingers slender, moderately deflexed, tapering into fine chitinous tips ( Figure 11I). P2-P4 short and robust, overall decreasing in size and acuteness of armature, meri of all compressed (Figures 11J-L). P2 merus elongated sub-pentagonal, 1.4 times as long as broad, distally armed with series of blunt conical nodules, similar-sized acute teeth, posteriorly lined with inconspicuous teeth; carpus and propodus externally lined with series of acute spine, dactylus slender, shorter than propodus, externally unarmed ( Figure 11J). P3 merus 1.2 times as long as broad, distally of low-rounded tubercles, posteriorly of one blunt tooth, carpus and propodus of series of stout nodules, dactylus externally armed with small spine ( Figure 11K). P4 merus 1.4 times as long as broad, distally of low tubercles, posteriorly of inconspicuous tooth, carpus and propodus lined with rounded granules ( Figure 11L). P5 robust, segments nearly cylindrical, merus anteriorly of small tubercles, carpus and propodus elongated, subequal in length, externally of small rounded tubercles, dactylus claw-shaped, slender, tapering into a fine tip, curving and articulating ventrally ( Figures 11M, N).
Remarks. Both the present species and L. doughnut sp. nov. described above shared host P. peroni, a scleractinian coral common in southern and eastern waters of Hong Kong (Chan et al., 2005).
The only acquired female specimen (CEL-Hapa-006) was damaged, with thoracic sternites laterally and posteriorly detached. Various diagnostic morphological features can still be examined and illustrated herein. The present form shares much resemblance with L. doughnut sp. nov. but nevertheless identifiable from which and other congeners by the following aspects: (1) metagastric region furnished with two shallow longitudinal grooves, laterally defined by two deeper grooves ( Figures 5E, 11A); (2) female thoracic sternum anterior plate octagonal, strongly sculptured, medially markedly grooved along midline ( Figure 11H); and (3) P2 merus 1.4 times as long as broad, sub-pentagonal ( Figure 11J). See Remarks under L. doughnut sp. nov. above for distinctions against other congeners. COI sequence obtained from the only specimen of this form was shorter, with length only 567 bp. Comparing against L. doughnut sp. nov. (n = 1) from the same host species, K2P distance was measured to be 2.73%, a value marginal for species recognition (see Figure 7). Surprisingly, among local material examined, L. cf. doughnut shows affiliation with both L. paradoxus (n = 17) and L. scottae sp. nov. (n = 14): differing from L. paradoxus on average by 3.35 ± 0.73% and L. scottae sp. nov. by 3.31 ± 0.87%. As such, genetic distance between L. cf. doughnut and L. doughnut sp. nov. (2.73%) overlaps with divergence range between L. cf. doughnut and both L. paradoxus and L. scottae sp. nov. In this consideration, available molecular evidence remains inconclusive on whether L. cf. doughnut being distinct. As in L. doughnut sp. nov., genetic distances between L. cf. doughnut and both C. coralliodytes and members of the "Xynomaia sheni" complex are substantial and beyond suggested inter-specific thresholds as addressed above: from C. coralliodytes (n = 2) by 9.25 ± 0.14% and from "Xynomaia species" (n = 10) on average by 9.40 ± 2.60%. Given only one damaged specimen obtained and examined, we remain hesitant in drawing conclusion on identities of this form and pending further investigation on local cryptichirids, particularly those infesting the host coral P. peroni.
Description (based on holotype CEL-Hapa-019, female 4.8 × 6.0 mm). Carapace longitudinally ovate, 1.25 times longer than broad, broadest breadth, and most elevated at about 3/5 of CL; anterior 3/5 depressed, markedly deflexed and sculptured, sunken on hepatic region, which each of an oblique, broad groove, furnished with isolated, small rounded tubercles, giving an eroded texture, mesogastric region sparsely scattered with small conical spines, metagastric region laterally defined by two short deeply incised oblique grooves on each side, both grooves separated by raised granular cluster; posterior 1/3 inflated, outline arched, cardiac-intestine region broad of nearly 1/2 of cw, defined by the latter pair of abovementioned oblique groove, medially not connected, surface covered with numerous small, well-spaced, slightly elongated nodules for most of the length ( Figure 12A). Front broad, inconspicuously convex, inner orbital angle of broad, convex, inflated lobe, dorsally lined with several small, rounded tubercles; exorbital angle crested, cristate, projecting beyond inner orbital lobes, confluent with anterolateral margin, which armed by series of small acute spines extending to approximately 1/3 of carapace length ( Figures 12A, B). Pterygostmial region granular, rhomboid plate of which fused dorsally with carapace, suture inconspicuous but discernible ( Figure 12B).
Description of male (based on paratype CEL-Hapa-015, male 2.9 × 4.7 mm). Carapace 1.6 times as long as broad, longitudinally ovate, anterior 2/5 depressed, moderately deflexed, broadest breadth and most elevated at approximately 2/5 CL, overall granulation and delineation of regions far weaker than conspecific females; anterior part mildly sculptured, bearing a depression of inverted "V" shape, delineating a triangular mesogastric region, laterally subparallel with anterolateral region, floor of which finely granulated, sparsely scattered with isolated, slightly larger tubercles; frontal margin broad, nearly straight, laterally flanked by an elevated inner orbital lobe, which anteriorly armed with several slender, acute spines; exorbital angle triangular, anteriorly extending beyond inner orbital lobe, confluent with anterolateral margin, lined by a series of weak serrations, extending for 1/3 of carapace length; cardiac-intestine region vaguely defined by fine grooves, which barely discernible; posterior half of carapace inflated, scattered with small, low tubercles, roughly equally distant from one another ( Figure 13A).
Basal plate of antennular peduncle outline narrow-triangular in dorsal view, anteriorly armed with several acute teeth, mesial margin basally of small raised longitudinal lobe and scattered with small spinules on dorsal surface ( Figure 13A). Eyestalks short and stout, oriented anterolaterally, cornea spheroidal, slightly expanded ( Figure 13A).
Host coral. So far, local material was retrieved from C. aspera and Favites pentagona of the Merulinidae. Host sharing was observed on C. aspera with X. sheni (see below).
Etymology. The present species is named after Dr. Paula J. B. Scott, author of The Corals of Hong Kong (1984). This researcher, an expert of coral-associated invertebrates, witnessed the abrupt decline of coral communities in Hong Kong during the 1980s (Scott and Cope, 1982;Scott and Cope, 1990). Her pioneer, thorough and reader-friendly book laid foundation for the current understanding of the local scleractinian fauna.
Type locality. Pak Lap Tsai (白鱲仔), Sai Kung, Hong Kong. Geographical distribution. So far recorded from only two sites, both coral communities in eastern waters of Hong Kong: Pak Lap Tsai, and Long Ke Tsai (浪茄仔) (see Figure 1). Considering the broad, Indo-Pacific distributional range of both host species (see Dai and Horng, 2009), potentially broader distribution of Lithoscaptus scottae sp. nov. is anticipated.
Remarks. The present species is recognized as closely related to, but distinct species from, L. paradoxus based on the three lines of evidence, namely, host records, morphological examination, and molecular analyses.
Females of L. scottae sp. nov. can be distinguished from those of L. paradoxus by the following features: (1) contours of carapace, L. scottae sp. nov. most elevated at approximately 2/3 of CL ( Figure 12B), whereas that of L. paradoxus at near midlength ( Figure 9B); (2) sculpturing of dorsal surface of carapace, that of L. scottae sp. nov. more markedly sculptured metagastric region, which laterally defined by two pairs deep, broad oblique grooves, whereas cardiac-intestine region relatively poorly defined laterally (Figures 5F, G, 12A), in comparison with L. paradoxus, which metagastric region not delineated by deep grooves, and cardiac-intestine region delimited by rather deep and narrow grooves ( Figures 5B, C, 9A); and (3) anterior extension of thoracic sternite, that of L. scottae sp. nov. markedly broader than long ( Figure 12H) compared with L. paradoxus merely as long as broad ( Figure 9H).

Xynomaia sheni Fize and Serène, 1956 species complex
Description (based on CEL-Hapa-002, female 3.5 × 4.8 mm). Carapace longitudinally ovate, 1.4 times longer than broad, mildly convex longitudinally and transversely, broadest and most elevated at approximately half of CL; anteriorly half densely furnished with stout, conical spines, roughly evenlyspaced, replaced by rounded granules posteriorly; front broadly concave, anteriorly lined with three slender spines, inner orbital lobe broad, inflated, anteriorly armed with numerous strong, elongated spines; exorbital angle spinose, confluent with anterolateral margin, which raised, marginally stout, armed with series of strong spines, lateral to eyestalks furnished with short, oblique deep groove; hepatic region behind each inner orbital lobe markedly depressed, region of which inverttriangular, floor likewise of stout conical spines, anteriorly defining mesogastric region; mesogastric region broadly triangular, moderately inflated, likewise of conical spines; cardiac-intestine region delineated by two shallow longitudinal depressions anteriorly, laterally barely defined; mesobranchial region lined with shallow longitudinal groove ( Figures 14A, B). Pterygostomial region granular, rhomboid plate of which incompletely fused with carapace, suture visible ( Figure 14B).
Remarks. Currently, three species have been placed under Xynomaia Kropp, 1990, namely, X. boissoni (Fize andSerène, 1956a), X. sheni, and X. verrilli (Fize and Serène, 1957); the genus is characterized by carapace not anteriorly deflexed, dorsally furnished with a W-shaped depression on the anterior portion, pterygostomian region not fused dorsally with carapace, antennal segment 2 distally armed with lateral spine, and P2 merus lacking disto-mesial projection (Kropp, 1990). All of these three species were described based on material from Nhatrang, southern Vietnam, all under Troglocarcinus [in Fize and Serène (1957) as T. (Troglocarcinus) boissoni, T. (T.) sheni, and T. (Favicola) verrilli]. Apart from host differences, the three species can be differentiated by outline and spinulation on carapace (see Fize and Serène, 1957). Unfortunately, relevant type lots were not re-examined or further reported so far. Xynomaia sheni was further reported from Guam by Kropp (1990) and selected as type species of Xynomaia. Several later reports under this name from the West Pacific contributed gene sequences (see below), but associated morphology remains yet illustrated in detail. Xynomaia boissoni, symbionts of the Merulinidae and Lobophylliidae, was later reported from (1) Honshu, Japan (Takeda and Tamura, 1983;Takeda and Tamura, 1985), yet as admitted by these authors, "lateral border of carapace in the specimens in hand is reasonably less convex than in the original figure" (p. 8, 1983) and (2) New Caledonia by Richer de Forges and Ng (2006), who provided no elaborations or illustrations. Xynomaia verrilli, symbiont of Oulophyllia crispa (as Platygyra gigantea), Dipsastraea speciosa (as Favia speciosa), both Merulinidae, a rare species at the time of description, was never reported elsewhere since.
Our local material consists of six female specimens identifiable as Xynomaia, among which only two were relatively intact: CEL-Hapa-002 (illustrated as Figure 14) and 004 (not dissected), differing slightly by overall shape of carapace (Figures 5I,J). Compared with illustrations of the original material provided by Fize and Serène (1957), we tentatively identify the local form as X. sheni, given the following morphological features: (1) relative broad width between bases of both orbits in dorsal view and (2) depression on anterior half of carapace and longitudinal groove laterally defining cardiacintestine region shallow. However, several features differ, namely, (1) convexity of lateral margin or carapace-mildly arched in our material while much convex in the type material; (2) carapace dorsally furnished with larger rounded tubercles on posterior half while less as tubercular as in the types; (3) mesobranchial region furnished with shallow longitudinal groove, while such groove is absent in types ( Figures 5I, 14A vs. fig. 16 in Fize and Serène, 1957); and (4) outline of ambulatory meri, in comparison with the type material, that of P2 more elongated in CEL-Hapa-002, while all short and stout as in P3 to P5 (Figures 14K-N vs. fig. 11G in Fize and Serène, 1957).
Molecular evidence as shown in the NJ tree of COI sequences ( Figure 6B) concerning this poorly known genus is perplexing. First, COI reference sequences from GenBank labeled as "X. sheni" (KJ923679.1, KJ923680.1, and KU041817.1) are clearly not conspecific: KJ923679.1 and KJ923680.1 both from Semporna, Borneo, Malaysia share a same haplotype but of different host families (Pectinia lactuca and Mycedium elephantotus of the Merulinidae), but these two sequences differ from KU041817.1 (Kropp's material from Guam) by a K2P distance of 14.97%. Sequences extracted from the Guam specimen as X. sheni (KU041817.1; shorter length of 396 bp) and another as X. sp. from Lembeh Strait, Indonesia (KU041835.1; 625 bp), after alignment, are identical throughout the 396 bp; thus, K2P distance is calculated to be 0 and very probably share the same COI haplotype. These two specimens were from hosts of different families (Pectinia paeonia, Merulinidae, and Oxypora lacera, Lobophylliidae, respectively). The Malaysian haplotype (KJ923679.1, KJ923680.1) was shown to nest among various Lithoscaptus species in reconstruction presented by van der Meij & Reijnen (2014;also Figure 6B). In this aspect, whether or not either of these two haplotypes represents the true X. sheni cannot be ascertained.
Difficulties in the morphological identification of this genus have been notorious. Small body sizes, and morphological features previously considered to be diagnostic (see Kropp, 1990) being probable results of convergent evolution, attribute to a "composite" Xynomaia in the phylogenetic reconstruction presented by van der Meij and Nieman (2016). At specific level, evidence of genetic differentiation suggests specific differentiation, while morphologies remain conservative, which might indicate occurrences of cryptic and sympatric speciation events. Reluctantly, we remain to identify the present genetically distinct forms as X. sheni species complex, noting its unusual genetic divergence and plasticity in host preferences. Deciphering true identities of X. sheni, however, might require examination of type or topographic material, which currently remains unresolved. Genetic and species diversity of this "X. sheni complex," anyhow, can be further elucidated with accession to fresh material, hopefully made possible in the near future.

Discussion
Sequence divergence and species identifications of cryptochirid crabs Molecular approaches adopted in species identification in the present study comprises two aspects, namely, phylogenetic relationships and pairwise distance between query and reference sequences. Results of both approaches are respectively indicated as Figures 6, 7. In contrast, as shown in Figures 6A, B, precise identification was reached in one species using 16S rDNA sequences, whereas precise identification was reached in all five (of six OTUs) by COI sequences (see also elaborations above). This differing resolution in species identification of the two adopted gene markers can at least be partially attributed to the 16S rDNA marker being slightly conserved than that of COI. This had been demonstrated in studies on various groups of decapod crustaceans, particularly among "young" species in rapidly evolving lineages, such as freshwater (e.g., Shih et al., 2007;Yeo et al., 2007) and intertidal crabs (Shih and Suzuki, 2008;Shih et al., 2010;Ragionieri et al., 2012). The discrepancy is likewise observable among cryptochirids as herein reported. Moreover, this is compounded by the current number of available sequences in the depository of GenBank. As of September 2022, for 16S rDNA, over 150 reference sequences representing 18 species (including two of tentative generic identification) were available, in comparison to over 380 of 43 species (including nine of tentative generic identification) for COI. In this consideration, COI remains to be an informative source as a genetic marker for species identification and barcoding analyses. In the present study, while the 16S rDNA marker shows inadequate resolution in species delimitation, that of COI corresponds well with morphological delineations.
Despite the lack of resolution in revealing phylogenetic relationships between deeper lineages, as demonstrated by van der Meij and Nieman (2016) on cryptochirids, the variability of the mitochondrial COI marker makes it useful for species identification and frequently used in barcoding analyses, and genetic (K2P) distances (maximum intraspecific and minimum interspecific genetic distances) may be useful for considerations in species delimitation (see e.g., Bucklin et al., 2011;Chu et al., 2015 for reviews). Based on query and reference sequences, we report mean intraspecific K2P distances of cryptochirids being 1.06 ± 0.76% and interspecific (intrageneric) distance at 7.16 ± 3.27%. Genetic distances among described Lithoscaptus species vary from 3.57% to 14.97%. Among described species of Lithoscaptus, lowest pairwise distance was observed between L. tuerkayi (KU745732.1) and L. hellerii (KU041819.1) at 3.57%. These values have been taken into consideration in species delimitations. These values are comparable with threshold values previously reported for thoracotreme crabs (1.49%-6.25%), from studies of groups in which taxonomy is relatively well resolved, and the examined taxa show discrete morphologies (see Davie et al., 2010;Shih et al., 2012Shih et al., , 2019KJH Wong et al., 2012). However, cryptochirid taxonomy is nowhere well resolved in two sense: numerous previously unrecognized species-level taxa (Wei et al., 2016;Bähr et al., 2021;Xu et al., 2021; see below) and previous morphology-based generic placement insufficient in encapsulate true species richness, thus rendered composite (van der Meij and Nieman, 2016; see Remarks under L. doughnut above). As such, both our intra-and interspecific distance values show considerable range of variation and are probably inflated. Adopting relevant threshold values for species delimitation might require alterative lines of supporting evidence. Nevertheless, the NJ tree constructed based on COI sequences ( Figure 6B) resulted in similar topology with that by van der Meij and Nieman (2016), likewise showing morphologically resembling Cryptochirus, Fungicola, Pelycomaia, and Xynomaia nested within. Lithoscaptus helleri and L. tuerkayi are probably aberrant members of the genus. This genetic marker does not foster resolution in resolving phylogenies of deeper lineages (see Chu et al., 2015), thus pend for further sampling of molecular data, and taxonomic revisions of, especially, Lithoscaptus and related groups.
Neglected symbionts of the scleractinian fauna: Case study of Hong Kong The scleractinian fauna of Hong Kong, as a historical parallel of local records of brachyuran crabs (see historical account under preparation by KJHW, BKKC, and colleagues), has been recorded since the mid-nineteenth century, both as a subsets of findings resulted from the North Pacific Exploring Expedition (1853-1856). The scleractinian fauna recorded during this expedition was reported by Verrill (1866), including 10 species from seas in the vicinity of Hong Kong (see Tam and Ang, 2008). The current understanding of the local scleractinian fauna had its foundations in the 1980s by Veron (1982) and Scott (1984), the latter a well-illustrated guide book mentioned above. By 2005, a revised guide book by Chan et al. (2005) reported 84 scleractinian species. During the recent decades, local shallowwater coral communities of Hong Kong have been routinely surveyed by researchers (e.g., KT Wong et al., 2018;Yeung et al., 2021a) and amateur divers as an environmental education program (Cheang et al., 2017), and the scleractinian inventory has been reasonably well established.
In comparison with the rather detailed reports of corals, cryptochirid crabs have been poorly documented in the literature. The first record of a cryptochirid crab from Hong Kong was published by Shen (1936), in which C. hongkongensis Shen, 1936[now Neotroglocarcinus hongkongensis (Shen, 1936] was described, while its host, precise locality, and deposition of types remained unspecified. Shen (1940), in compiling a checklist of brachyuran fauna of Hong Kong, included records of only two males and two females with no indication of their type status, listed as "No. 12799," probably the catalog number of the material deposited at the Fan Memorial Institute of Biology, Peiping, the whereabouts and status of which are currently unknown. This species was subsequently considered synonymous with P. viridis Hiro, 1938by Utinomi (1944. In deciphering identities of C. hongkongensis, Kropp (1988b) showed that both species are distinct and, based on the available material in various institutes, deduced that the host of N. hongkongensis should probably be Duncanopsammia peltata. This species remained the single host of cryptochirids recorded from Hong Kong, until van der Meij (2012), based on image of a crescent-shaped domicile opening on Turbinaria mesenterina presented by Scott (1984) (Fig. 42B), reported P. viridis. The latter record might not be as affirmative, as the p h o t o g r a p h e d d o m i c i l e m i g h t i n s t e a d b e l o n g t o Neotroglocarcinus dawydoffi (Fize and Serène, 1956a) instead (SET van der Meij, pers. comm.). We have not yet acquired a local material of any these two (or three?) recorded species. This bold asymmetry in the understanding of the local scleractinian and cryptochirid fauna is highlighted by the surprising addition of five gall crab species in this study, based on a limited material that was amassed during several days of intensive samplings. This asymmetry is likewise apparent among faunal records in the vicinity of South China: a scleractinian fauna of 315 species (Zou et al., 2008; excluding records from only Taiwan and East China Sea) while only three cryptochirid species reported (Yang et al., 2008). This can be partly attributed to the group being intuitively "cryptic" symbionts of scleractinians, easily overlooked among coral communities, a habitat generally poorly surveyed, difficult in sampling, and lack of taxonomic expertise. In this sense, the number of cryptochirids so far recorded from Hong Kong and the South China appears to be an underestimation.
[local hosts: Coelastrea aspera and Favites pentagona] 6b. Carapace elevated at about half mid-length ( Figure 10B), gastric region laterally each defined with short concave groove ( Figures 5D, 10A); anterior plate of thoracic sternum distinctly octagonal, broader than long ( Figure 10H) Is the form of female thoracic sternum a species-level diagnostic character?
Morphologies such as dorsal surface of carapace, and lateral surfaces of ambulatory legs, are often subjected to intense abrasion and wear, thus the substantial variation. Alternatively, the taxonomic value of the form of thoracic sternites, being a "nonexternal" morphological feature, has been demonstrated for female cryptochirids. While the form of sutures 4/5 to 7/8 (whether medially confluent or separated) is comprised of three patterns, warranting delineation of respective subfamilial grouping (Guinot et al., 2013), the form of the anterior plate (sternite 3, anterior to base of P1), in terms of outline and surface ornamentation, has been used by Kropp in the delineation of Opecarcinus species (Kropp, 1989) and the distinction between Pacific genera (Kropp, 1990). In this consideration, we attempted to employ this character among Lithoscaptus species. Among previously described species (n = 10), however, thoracic sternites of only three were illustrated: L. paradoxus [ fig. 4c in Kropp (1988a)], L. prionotus [ fig. 4c in Kropp (1994)], and L. tuerkayi [ fig. 4B in van der Meij (2017)]. Both of the two new species described in this study have a structure that differs markedly from those illustrated earlier (see Remarks under each species). As such, we propose the inclusion of the form of female thoracic sternites, particularly the shape and surface texture of the anterior plate, as diagnostic features of Lithoscaptus-related taxa. Further documentation of this structure from the relevant material would test its viability in species-level identifications.

Broad host range, or unrevealed diversity?
Based on our material from Hong Kong of at least five species, each was found symbiotic with one to two host species.
Among the over 40 Indo-Pacific species, most (n = 37) show a narrow range of one to two host genera. Given the above observations, can we reach the conclusion that, in terms of host range, C. coralliodytes and L. paradoxus are "generalist" along a generalist-specialist gradient? Patterns of host specificity can be confounded by unrecognized diversity. Recent investigations on H. marsupialis, a species distributed from western Indian Ocean to the Central Pacific (Wei et al., 2016;Bähr et al., 2021), showed important implications. Despite highly conserved morphologies, genetic data based on materials amassed across the distribution range recognize multiple putative species, each biologically distinct (COI divergences range from 3.2% to 15.7%; Bähr et al., 2021), and putative species differ in recorded host (of Pocillopora, Seriatopora, and Stylophora; Wei et al., 2016;Bähr et al., 2021) and early larval characters (Gore et al., 1983). A similar pattern is also observed among Opecarcinus species by Xu et al. (2021), who showed the current taxonomic understanding (of nine species; Kropp and Manning, 1987;Kropp, 1989) being a gross underestimation of true diversity (at least 25 OTUs). Both cases illustrate another aspect of "crypticity," revealing a discrepancy between the current species inventory and unrecognized true diversity of species or OTUs. Previously presumed "generalist" niches (as assessed by host range) are confounded by poor taxonomic resolution, where biologically distinct OTUs are examined to be undifferentiated. Additionally, for past records of Cryptochirus and Lithoscaptus, the ambiguity had been compounded by a confusing taxonomic history. These two genera, for many decades, had been placed as synonyms and collectively referred as symbionts of massive "Faviidae" corals. This was satisfactorily resolved merely 30 years ago (see Kropp, 1988a). For the 12 host genera of C. coralliodytes and L. paradoxus enumerated above, also Plesiastrea, the host of L. doughnut sp. nov., shows very broad Indo-Pacific distribution ranges (see e.g. Veron, 2000;Huang et al., 2014), and the presence of previously unrecognized cryptochirid taxa is certainly anticipated. As obligate symbionts, the biology of cryptochirids is highly dependent on that of their hosts (e.g., Hiro, 1937;Utinomi, 1944). Assessing any particular cryptochirid taxa along the specialistgeneralist continuum, however, remains much obscure until precise species-level diversity can be recognized. This highlights the imperativeness of investigations in species diversity in a refined, integrated approach for some twofold "cryptic" cryptochirids.

Data availability statement
The data presented in the study are deposited in the NCBI Genbank, with accession numbers indicated in Table 1.

Author contributions
KJHW, Y-FT, J-WQ, and BKKC conducted field sampling. J-WQ provided logistic support. Y-FT and KJHW performed molecular analyses. KJHW wrote early drafts of the MS. All authors contributed to the article and approved the submitted version.
Technology Council, Taiwan (109-2621-B-001 -003 -MY3). The cryptochirid crabs were collected while implementing projects supported by Environment and Conservation Fund, Hong Kong (2011-08, 2015-84, 2017. KJHW is supported by postgraduate studentship in Institute of Ecology and Evolutionary Biology, National Taiwan University, Taiwan. We are grateful to reviewers for their time and effort in providing much helpful advice on the manuscript.