﻿An updated, illustrated inventory of the marine fishes of the US Virgin Islands

﻿Abstract The US Virgin Islands (USVI) include St. John and St. Thomas on the Puerto Rican Platform (PRP) and St. Croix, isolated by 2000 m deep water 45 km south of that platform. Previous inventories of the marine fishes of these islands include a comprehensive 2014 checklist of the fishes of St. Croix and a list of the fishes of the PRP produced in 2000. The latter list noted the locations of many records of the plateau’s fishes, allowing the construction of a combined inventory for St. John and St. Thomas. Those two islands are treated here as a single faunal unit because they are only 3.5 km apart on a shared shallow shelf with various islets and reefs in between. Here we provide updated information on those two USVI (St. Croix and St. John-Thomas) marine fish faunas. The additions to the St. Croix and St. John-Thomas inventories presented here are based on a combination of information from the two sources indicated above, more recent publications dealing with those faunas, a review of location records on various online sources of biogeographic data, and voucher photographs taken of fishes in the field by authors of this paper and other citizen scientists. This assessment increased the known fauna of St. Croix by 7.5% to 585 species. The inventory for St. John-Thomas increased by 39.9% from 401 species on the 2000 PRP list to 561 with the inclusion of records from other sources. On-site mtDNA (COI) barcodes are available for approximately one-third of the species of the St. John-Thomas fauna, but for only one species collected at St. Croix. A set of underwater photographs of 372 species (34 of them representing the sole record of a species) from St. John-Thomas and of 11 shallow-water species added to the St. Croix fauna is included. These represent occurrence vouchers and also are intended to facilitate future work that builds on the present compendium.


Introduction
The United States Virgin Islands (USVI) comprise a US territory adjacent to Puerto Rico, in the northeast Caribbean, that includes three large, inhabited islands, St. John, St. Thomas and St. Croix, and approximately 50 smaller islands and cays around them. The former two are situated only 3.5 km apart, in the center of the Puerto Rico Plateau (PRP), which has an area approximately twice the 9,100 km 2 of Puerto Rico Island and extends ~ 150 km eastwards from Puerto Rico. St. Croix is located south of St. John and St. Thomas, on its own insular platform, which is separated by 45 km of deep water from the southern edge of the PRP.
The fish fauna of St. Croix was comprehensively reviewed by Smith-Vaniz and Jelks (2014), who built upon an older list by Clavijo et al. (1980), using their own extensive collections of shallow fishes of the Buck Island Reef National Monument on the northern side of St. Croix (Smith-Vaniz et al. 2006), and a review of literature and examination of specimens of fishes collected at St. Croix that are lodged in various museums. In 2000, George Dennis produced an extensive (244 page; 500+ sources cited) U.S. Geological Survey report based on collections and observational records for marine and brackish-water fish from Puerto Rico, St. John and St. Thomas, and other islands on the PRP. Although never formally published in a scientific journal, and no longer available through the USGS source cited by Dennis et al. (2004), that compendium is available online (Dennis 2000).
Here we add new information to update the 2014 list for St. Croix and assemble an inventory for St. John and St. Thomas that includes and expands on data for those two islands contained in Dennis (2000). We extracted the additional information from museum records in online sources of biogeographic data, publications produced since Dennis (2000), digital images of live fishes obtained at the USVI, plus our recent collections and mtDNA barcode records obtained from the database BOLD. The great majority of the species in this compendium are marine, plus we include a small number of species found in fresh to brackish waters.

Study sites
St. Croix is a 215 km 2 island in the northeast corner of the Caribbean. It is isolated by ~ 45 km of deep water from the Puerto Rican Platform (PRP). Other islands of the Lesser Antilles chain lie within ~ 150 km to the east and southeast of St. Croix. The surrounding shallow (above ~ 150 m depth) shelf of St. Croix, extending almost 20 km eastward, has approximately the same area as the island. In addition to exposed and sheltered coral reefs and soft bottoms, the island has extensive areas of seagrasses and mangroves.
St. John (area 50 km 2 ) and St. Thomas (area 83 km 2 ) are situated in the center of the shallow (to ~ 150 m deep) tongue of the PRP that extends 150 km eastwards from Puerto Rico. St. Thomas is closest to and 64 km from the main island of Puerto Rico. St. John and St. Thomas are separated from each other by only 3.5 km of water shallower than 20 m deep, with scattered islets and shallow reefs in between them. They have a similar range of habitats as St. Croix, with large areas of both sheltered and deeper shelfedge coral reefs, rocky shores, seagrass beds and mangroves. Due to their proximity and similarity of habitats we treat them here as a single unit (hereafter St. John-Thomas). The shallow PRP associated with St. John-Thomas extends ~ 25 km north and ~ 15 km south of those islands and covers an area of ~ 2,100 km 2 (Rohmann et al. 2005).
Suppl. material 2: File S1 shows the bathymetry of bottom habitats on the above-150 m shelves of the USVI. The shelf area of the St. John-Thomas EEZ is not only much larger than that of St. Croix but also differs from the latter in containing a much greater diversity of areas of different depths. There are large expanses, in both absolute and relative terms, of habitat between 40-60 m deep to the north of St. Thomas and to the south of both islands. In contrast, most of the smaller shelf of St. Croix is shallower than 20 m deep.

Data sources
We reviewed and cited only publications from which we extracted information relating to the USVI fishes that were published after those cited by Dennis (2000) for St. John-Thomas, and after that by Smith-Vaniz and Jelks (2014) for St. Croix, plus a few earlier publications that contained additional relevant information.
Smith-Vaniz and Jelks (2014) published a comprehensive, annotated checklist of 544 fishes known from St. Croix. That checklist was based, in large part, on the yield of fishes from 106 rotenone stations obtained by Smith-Vaniz et al. (2006) and by later workers to document the shallow cryptobenthic fauna. That 2014 list identified questionable records, a few of which, as we show, have turned out to be valid. Smith-Vaniz and Jelks (2014). That checklist also excluded deep-water fishes not found above 200 m as well as Exocoetids and Myctophids. For completeness we have included any such species recorded by other sources among the additions noted here. We used the 2014 list of valid species and reviewed fishes listed by other surveys: a SCUBA study of the shallower parts (30-50 m depth) of a mesophotic coral ecosystem at the eastern end of the shelf (García-Sais et al. 2014); two JSL submersible dives off St. Croix to 30-600 m (Nelson and Appeldoorn (1985); and two ROV dives off St. Croix at depths greater than 800 m (Quattrini et al. 2017). In addition, we reviewed the records of fish species from St. Croix available from various online sources: the aggregators GBIF (https://www. gbif.org/), FishNet2 (http://www.fishnet2.net/), iDigBio (https://www.idigbio.org/ portal), OBIS (https://obis.org/) and Vertnet (http://vertnet.org/), and the American Museum of Natural History (AMNH; https://www.amnh.org/research/vertebrate-zoology/ichthyology). Those searches were made within a quadrat with latitudinal limits of 17.62°N to 17.85°N, and longitudinal limits of -64.4°W to -65.0°W, encompassing St. Croix and all of its platform. The sources of St. Croix records produced by those online searches were evaluated and museum records within the known geographic range of various species were accepted. Evaluation of individual records is necessary because aggregator information includes significant numbers of erroneous records.
Finally, the list includes shallow-reef fishes photographed by authors AME and CJE during a month spent at the island from 19 December 2020 to 13 January 2021. Suppl. material 3: File S2A presents a list, with georeferenced locations, of the 11 dive sites at which they together made 25 dives (total 47 hours duration per person) during that period (see also Fig. 1B and Suppl. material 4: File S3, a Google Earth © KMZ file that shows, for each of those sites, its location and georeferenced coordinates, and the number of dives and total dive time spent at that site). These photographs document a few species not previously recorded at the island, plus several not accepted by Smith-Vaniz and Jelks (2014) due to a lack of reliable information.
For St. John-Thomas we extracted a list of 401 species listed at those islands by Dennis (2000) and reviewed various publications dealing with fish records at and near those islands that were subsequently produced. Finally, we also used the same online data sources as for St. Croix (see above) to obtain records of fishes from the part of the Exclusive Economic Zone of the USVI that includes St. John-Thomas and extends between the northern and southern edges of the PRP. That irregularly shaped EEZ was obtained from Marineregions.org, which provides a standard set of global maps of EEZs (https://www.marineregions.org/eezsearch.php).
CJE and AME spent six months between 3 November 2020 and 29 May 2021 diving at both islands and photographing fishes to obtain voucher images of as many members of those islands' marine fish fauna as possible. File S2A presents a list, with georeferenced locations, of their dive sites at St. John (37) and St. Thomas (12), at which they made 113 joint dives (involving multiple dives at some sites) totaling 221 hours per person and 37 dives totaling 37 hours per person, respectively. Fig. 1A is a map with those 49 dive sites at St. John-Thomas indicated and File S2 provides additional information. Fig. 1A (and see File S2B) also indicates the location of sites from other sources at which additional species not recorded by CJE and AME were documented photographically by other divers.

Reef-associated bony fishes of the USVI
Greater Caribbean (GC) reef systems have reef-fish faunas that are dominated by members of typical, shallow-reef families of bony fishes extending down to depths of ~ 250-300 m (Baldwin et al. 2018). Here we focus on species belonging to those families, which have traditionally been viewed as reef fishes. We classed species living entirely or largely below 40 m depth as belonging to the deep-reef subset. Species classed here as shallow include both species restricted to depths shallower than 40 m and those with depth ranges that extend above and below that level. These reefassociated fishes include not only benthic and demersal species found on hard-reef substrata, but also pelagic fishes that facultatively associate with reefs and benthic and demersal species that live on soft bottoms within and immediately around the fringes of reefs. Benthic species (e.g., eels, flatfishes) are restricted to life on and in different types of substrata, while demersal species (e.g., snappers and grunts) use both substratum habitats and the water column. Cryptobenthic species are visually cryptic and typically small. We followed Brandl et al. (2018) in classifying families dominated by small cryptobenthic coral-reef species as Core Coral Reef Fish families (CCRFs).
We also evaluate the ecological and zoogeographic composition of the two USVI fish faunas (St. Croix and St. John-Thomas) compared to the complete checklist of the regional fauna of reef-associated bony fishes, which includes 992 species in 342 genera and 84 families (Robertson and Tornabene 2021). These aspects of the fauna of the USVI are also compared with results from another recent comprehensive survey of the fish fauna of nearby Sint Eustatius, which is 170 km from St. Croix (Robertson et al. 2020).

mtDNA-barcode coverage of fishes collected in the USVI and Puerto Rico
Relatively few small marine locations have been comprehensively sampled for fish DNA barcoding, i.e., tissues sequenced for the mtDNA COI marker as a standard for identifying fishes, as compiled in the Barcode of Life Database, BOLD (Ward et al. 2009). Notably, BOLD not only includes a wide variety of projects, most of which are publicly available, but also regularly harvests all available COI sequences from Gen-Bank. In contrast, GenBank does not harvest from BOLD, and BOLD sequences are generally submitted to GenBank only by request. As a result, only a fraction (~ 15% for GC fishes) of COI sequences on BOLD also are present on GenBank, despite its widespread use as the sole source for barcoding studies. BOLD further differs from GenBank by applying quality control to sequences and taxon identifications as data is entered, including sequences harvested from GenBank. It also has post-hoc quality control via a tagging and comment option on individual records. BOLD also includes a large number of private sequences, which can be assessed to a limited degree (with some metadata removed) via the BIN portal, which compiles all records, public and private, within a lineage, assigns a code, and presents some statistics, especially variance and nearest neighbor distances, as well as countries of origin.
The BOLD BIN code is a key advance enabling the compilation and comparison of mtDNA barcoding lists, since it supplies an independent identifier for a monophyletic genetic lineage, which is not the same as a species name. BOLD creates BINs (Barcode Index Numbers) by clustering barcode sequences algorithmically. The BIN often represents a particular species, but there are many exceptions to the "one-species, one BIN" concept: either multiple BINs per species, indicating genetically divergent populations within species (usually allopatric, but not always), a subset of which are putative new cryptic species awaiting morphological confirmation; or shared BINs by two or more species that retain shared or closely related haplotypes due to a short time since speciation, to incomplete lineage sorting, or to a small degree of hybridization.
Our broad assessment suggests that BOLD has a BIN that can be assigned (with widely varying degrees of confidence) to ~ 900 species of shallow-dwelling, reef-associated bony fishes from the GC. A list of sequences obtained in a particular area is obtained from BOLD by using a vector map in its search engine. The resulting list is from public projects (including all GenBank COI sequences), as well as whichever private projects the user has permission to access (often granted by an email request to the source of the sequence). In our case, we have been given access to all of the larger private projects in the region and barcodes for the vast majority (~ 90%) of sequence records in BOLD that could be evaluated in their respective BINs. The list of records from the geographic-area search on BOLD are individual sequences with metadata (including GenBank number if a sequence has one) and photographs of specimens (when available), together with a link to the BIN code to which it belongs. The species name originally submitted for each is preserved, and the accuracy of the assignment can be assessed by examining the BIN to which it belongs, which has details on the various names applied to sequences in the BIN and by whom and where they were obtained. Accuracy assessments are critical, especially for more obscure species, since a "majority rules" decision is often inaccurate due to multiple identifications by inexperienced contributors, the tendency to repeat the species-level identification made by others as a shortcut, and the practice of assigning species-level names to submitted records that are from eggs, larvae, isolated tissue, or fish-market specimens. GenBank records are harvested by BOLD with whatever name is assigned in GenBank, often a preliminary one from submission, rather than the one later corrected or published in the subsequent literature.

The island faunas
St. Croix: The checklist of Smith-Vaniz and Jelks (2014) included 544 species from 280 genera in 94 families. We obtained records of 41 species (belonging to 39 genera and 35 families; see Table 1) that were not included on that checklist, an increase of 7.5% in the number of species. Those new records included 19 deep-living species, six of them (11.1% of all deep species at St. Croix) resulting solely from observations by the JSL submersible (Nelson and Appeldoorn 1985;García-Sais et al. 2014) and an ROV (Remotely Operated Vehicle; Quattrini et al. 2017). It should be noted that almost all of that group belong to very deep taxa specifically excluded by Smith-Vaniz and Jelks (2014) from their list, which was focused primarily on shallower fishes. The remaining 22 species are shallow-water, reef-associated fishes. Ten of the latter group were photographed by AME and CJE (Table 1; Suppl. material 1: Plate S1). These additions include three species (Eucinostomus melanopterus, Coryphopterus glaucofrenum and Opistognathus macrognathus) that Smith-Vaniz and Jelks (2014) referred to but did not include in their checklist due to lack of confirmed records. Records of two mobulid rays consisted of identified photographs/videos provided by Mantatrust.org (https:// www.mantatrust.org/) that were inspected by DRR. The list (Table 1, which includes source information) also includes records from museum collections that provide online data directly or indirectly through aggregators, which were included if consistent with the known geographic range of each of those species.
St. John-Thomas: Table 2 presents a list of species recorded from those islands together with the source(s) of each record (images, publications, DNA barcodes, or online museum records) and which species have a voucher image in the supplementary plates (Suppl. material 1: Plates S2-S18). In addition, for uncommon species (those encountered by AME, CJE, LR, or third-party photographers at three or fewer dive sites) the names of the sites at which those uncommon species were found are included, to aid future investigations. Dennis (2000) also included information on species that were collected using the ichthyocide Rotenone (see Table 2). Smith-Vaniz and Jelks (2014) list for St. Croix also included some species recorded at these St. John-Thomas as a result of collections using  Victor, 2007 Kuna Goby S1 Oxyurichthys stigmalophius (Mead & Böhlke, 1958) Spotfin Goby S1 NOAA Kyphosidae Kyphosus cinerascens (Forsskål, 1775) Topsail Seachub S1 Macrouridae Nezumia aequalis (Günther, 1878) Atlantic Blacktip Grenadier yes 6 Malakichthyidae Verilus pseudomicrolepis (Schultz, 1940) False   ( . Uncommonspecies seen at 3 or less named sites by CJE and AME (see Suppl material 3: File S2a, b (for site codes) and Suppl. material 4: File S3). Ichthyocide -species collected by this method as noted in Dennis (2000); parentheses indicate ichthyocide was the only collection method noted by Dennis (2000). Gobiidae -we follow Thacker (2009) Dennis (2000) listed 401 species from 216 genera and 79 families from those two islands (Table 2). We found records of an additional 159 species, producing an increase of 39.7% in the number of species, 37.0% in the number of genera and 36.7% in the number of families known from there (Table 3). The additions include 34 species for which the only source is a voucher image, 50 species recorded in post-2000 publications, and 49 species recorded only by online sources of museum (and other) data (Table 3). Of the 561 in Table 2, 24.6% were uncommon. Although 30.1% were collected using rotenone, species accounts by Dennis (2000) mentioned no other collecting method for only 10.4% of that subgroup of species. The 561 include three non-natives to the area (Oreochromis niloticus, Poecilia reticulata and Pterois volitans), 11 freshwater/estuarine species (Anguilla rostratus, Dormitator maculatus, Eleotris perniger, Gobiomorus dormitor, Awaous banana, Sicydium plumieri, Sicydium punctatum, Dajaus monticola, Microphis lineatus and Pseudophallus mindii and 547 marine species native to the GC.  Table 2. Deep species are those exclusively or typically found below 40 m depth. Uncommon shallow species are those found at 1-3 sites by CJE, AME, LR, and thirdparty photographers as indicated in Table 2. Ichthyocide collection: recorded as being collected with rotenone by a source cited by Dennis (2000). Ichthyocide only: the only collection method listed for a species from St. John-Thomas by Dennis (2000 The species richness of the USVI marine fauna (i.e., the combined St. John-Thomas plus St. Croix faunas) was 15-20% greater than that of either of the two insular faunas (Table  4). Those two faunas had slightly higher relative rates of richness of genera and families than of species. The larger size of the USVI fauna of species derives from ~ 1/5 of species in each insular fauna not being present in the other, with lower proportions of genera and families also being recorded only at one of the two islands. Relative faunal richness at all three taxonomic levels and the relative abundance of taxa present at only one island were ~ 5% higher at St. Croix than St. John-Thomas. In both island faunas the relative representation of species, genera, and families in the entire USVI fauna was substantially greater among shallow species than deep species. The deep fauna was much smaller than the shallow fauna at each island and there was much less overlap in occurrence of species, genera, and families between the two insular deep faunas than between their shallow faunas (Table 4). Ecotypic structure of the USVI reef-fish faunas vs. the region ( We compared the ecotypic structure of the St. John-Thomas and St. Croix faunas of reefassociated fishes with that of the GC fauna (see Robertson and Tornabene 2021). Both St. Croix and St. John-Thomas have faunas that are almost half the size of the total regional   (Table 5). Compared to the GC fauna both islands have slightly higher percentages of pelagic species, distinctly higher percentages of demersal species, and correspondingly lower percentages of benthic, cryptobenthic, small cryptobenthic, and CCRF species. These differences for non-pelagic types apply to each entire USVI fauna, and to both shallowand deep-reef subgroups of those faunas. Both USVI sites also have markedly lower relative abundances (~ 1/3) of deep-reef species than the regional fauna. The relative abundances of different ecotypes are remarkably similar at both islands, except for the presence of a few deep cryptobenthic and CCRF species detected only at St. John-Thomas.
Zoogeographic structure of the USVI faunas (Table 6) The zoogeographic structures of the faunas of the two USVI sites and nearby Sint Eustatius are quite similar (   (Hanner et al. 2011;EBFSF), and the Museum and Art Gallery of the Northern Territory (GOBY) in Australia. The available DNA-barcode sequence records from specimens collected at St. John-Thomas represent coverage of 27.8% of the species, 31.4% of the genera and 38% of the families of fishes known from that site. Barcode records represent the sole source of information on the presence of one species known from those islands and are also available for another four species currently identifiable only to genus. Distinctly fewer species have been barcoded from fish taken at Puerto Rico, and there are almost no such data available from either St. Croix or the British Virgin Islands. Barcode records from Puerto Rico and the British Virgin Islands exist for 52 species occurring in St. John-Thomas but not sequenced from there, bringing the total PRP DNA-barcoded species to 36.5% of St. John-Thomas fauna. All but seven of the 200 barcoded species are reef-associated bony fishes. The vast majority (98.5%) of barcoded species are shallow forms. Deep-living species are especially under-represented among the barcoded forms: only three of 51 such species have barcode data (File S6).

St. Croix
The species records we have added increased the size of that island's fauna by 7.5%. Almost a third of the additions arise from voucher photographs of shallow-reef species photographed by CJE and AME (and provided by Mantatrust.org). Those include several not accepted by Smith-Vaniz and Jelks (2014) due to inadequate information available at that time. Cryptobenthic fishes, which, by definition, are generally difficult to observe, are a major component of Greater Caribbean reef-fish faunas, including that of St. Croix. Such species comprised all but one of those added by CJE and AME. The exception, Kyphosus cinerascens, may have been misidentified previously, since the taxonomic status and global distributions of members of the genus were only comprehensively reviewed by Knudsen and Clements (2016), after Smith-Vaniz and Jelks (2014) published their checklist. Almost half the additions were deep-living species, one third of which were recorded only by submersible or ROV, with the remainder coming from online and literature records.
The process of obtaining location records is an ongoing one for online aggregators, which have vastly increased the amounts of data they host during the last half decade. Although the aggregators offer such information, and are involved in collaborative data sharing, such sharing is sufficiently incomplete that it is necessary to examine records from multiple aggregators to obtain a comprehensive picture of all the data available for any particular site. Even "old" data becomes newly available on the aggregators from time to time and needs to be included in faunal inventories of well-studied sites. The increase in faunal size, although not large in percentage terms, demonstrates the utility of citizen-science efforts to provide photographic vouchers, of reviews of submersible and ROV studies of deep-living fishes, and of periodic evaluations of information available online from aggregators.

St. John-Thomas
Although the 401 species list for this site extracted from Dennis (2000) was substantial (74% the size of Smith-Vaniz and Jelks (2014) count for St. Croix), our use of the same methods as those that produced an increase in the St. Croix fauna produced a much larger increase in the St. John-Thomas fauna: 40% vs. 7.5% for St. Croix. Dennis (2000) was the sole source for 29% of species recorded in our expanded list of the St. John-Thomas fauna. Records from additional sources brought the size of the St. John-Thomas fauna to within 5% of the size of the St. Croix fauna. Citizen-scientists' photographic records accounted for 22% of the new additions and data only available from online databases for 33%, while other literature sources provided the sole records for 32% of the additional species. Multiple types of sources accounted for the remaining 13% of new records.
The size, and taxonomic-and ecotypic structure of the two USVI marine faunas Both insular marine faunas are over 80% the size of the combined USVI fauna in terms of species richness. Species found at only one of the two islands represent ~ 20% of each fauna. For shallow species the size of each insular fauna is 85-90% that of the combined fauna, with correspondingly lower rates of occurrence at only one island. Two factors may contribute to these differences between the island faunas: variation in ecological conditions between the islands and inadequate sampling. The possibility of differing ecological conditions seems small as both islands have the same range of large-scale habitat types, although those vary in abundance between the islands. The shelf area of St. John-Thomas is close to 10 times the size of the St. Croix shelf, yet the former has the smaller known fauna. At both islands the great majority of sampling has occurred in quite shallow water, often very close to shore in the case of St John-Thomas. Shelf habitats likely are under-sampled at both islands, strongly so at St. John-Thomas, where there are large areas of habitat between 40-60 m depth some distance from the islands on both the northern and southern parts of the PRP. At St. Croix most shallow sampling has occurred in and near the Buck Island Reef National Monument, rather than spread around different parts of the platform and different sides of the island. Hence both insular faunas likely are larger than indicated here, particularly in the case of St. John-Thomas.
Review of the two USVI marine species lists show that species not shared between the two islands are distributed through many genera and families (Suppl. material 5: File S4; Table 4). None are endemic to either USVI island and single-island endemics are rare amongst the Greater Caribbean fauna and limited to highly isolated islands such as Cayman. Most species in that region have geographic ranges much larger than the USVIs. The larger size of the St. Croix fauna, particularly of cryptobenthic species can be attributed to a greater effort to find such species. This was done using rotenone during two intensive sampling campaigns that occurred ~ 40 y after rotenone sampling at St. John-Thomas, plus some subsequent minor efforts in the shallow part of a Buck Island Reef National Monument that, in its entirety constitutes ~ 1/3 of the St. Croix insular platform: 46% (262) of the native marine species known from St. Croix are shallow species collected using rotenone (Smith-Vaniz and Jelks 2014), vs. 31.7% (173) of such species from St. John-Thomas. Later sampling by Pittman et al. (2008) at the same small, shallow St. Croix site as studied by Smith-Vaniz et al. (2006) added 10.9% more species to the tally of the first two series of collections. Smith-Vaniz and Jelks (2014) produced a list of 41 species from 22 families that, at that time, were known from St. John-Thomas but not St. Croix. Since then, five of the 35 shallow species on that table have been added to the St. Croix fauna (Table 1 here), together with two others that were listed as unconfirmed by those authors. Photographic sampling of shallow reef fishes at St. John-Thomas by CJE, AME and other citizen scientists, by itself increased the size of the fauna registered by Dennis (2000) by 8.5%. Finally, the species composition of local reef-fish faunas can change substantially through time at intensively sampled sites, for varying reasons (e.g., see changes registered by Starck et al. 2017 over a 50y period), highlighting the utility of temporally dispersed sampling. With further sampling many shallow species currently known from only one of the USVI should be expected to be found at the other, in which case the shallow faunas of each island would be 10-15% larger than the current figures.
The deep-species fauna represents only 13.1% of the entire (shallow plus deep) USVI fauna and deep species exhibit much lower rates of faunal overlap between the two islands than occurs among shallow species. The two islands have experienced low rates of exploration of deep habitats, particularly deep reefs, by submersibles and ROVs, which were limited to observational studies. The few ROV (Quattrini et al. 2017) and submersible dives (Nelson and Appeldoorn 1985;Garcia-Sais 2005) were the sole source of only 11.1% and 28% of records of deep fishes at St. Croix and St. John-Thomas, respectively. The edges of the insular platforms of the two USVIs are < 50 km apart and the suite of deep species involved have ranges much larger than the area occupied by the USVI. Low levels of sampling can account for the small size of both USVI deep faunas, particularly the deep-reef component, and to the low level of overlap between the deep faunas of the two islands.
At both USVI sites the deep-reef species represent only 4.2-5% of the entire local reef-fish fauna, i.e., ~ 1/3 of the percentage for the GC regional fauna . In contrast, when intensive submersible collecting and observations have been aimed specifically at assessing the diversity of deep-reef fish faunas, such as has occurred at other Caribbean islands (Curacao, Roatan and Sint Eustatius), the inventory of deepreef species at individual islands has increased ~ 9 fold, with such species representing 16% of the entire (shallow plus deep) reef-fish fauna at the most intensively sampled island , i.e., more than three times the level at each USVI. Similar sampling at both USVI undoubtedly will increase the absolute and relative sizes of their deep-reef faunas. Smith-Vaniz and Jelks (2014) concluded that there was no indication at the time of their study that the St. Croix fauna had reached asymptotic size. The additions reported here and patterns of variation in faunal composition between the two islands support that view for St. John-Thomas as well as St. Croix.
Reef-associated bony fishes comprised 84% and 91%, respectively, of the faunas of St. John-Thomas and St. Croix, and the St. John-Thomas reef-fish fauna was 94.3% the size of the equivalent fauna of St. Croix. The ecotypic structure of those two USVI reef-fish faunas was very similar, with both differing from the broad structure of the GC regional fauna by having larger proportions of pelagic and demersal species that are readily visible to observers and correspondingly smaller proportions of cryptic species. Similarities in the zoogeographic structures and sizes of the two USVI faunas support the view that both can be considered to be sampled with a similar level of efficiency, at least in terms of their shallow faunas.

mtDNA-barcode coverage
In terms of the availability of DNA-barcodes for marine fishes, the Greater Caribbean currently is the most well-sampled large marine biogeographic region in the tropics, with ~ 90% of the shore-fishes barcoded and up to 95% of the shallow reefassociated species (Victor et al. 2015). However, several specific locations account for the vast majority of sequences. Those include Florida, Yucatan (Mexico), Belize, Panama, and Curacao; with species lists published for Yucatan by Valdez-Moreno et al. (2010) and lists for additional locations in Weigt et al. (2012). The Puerto Rican Plateau has been only lightly sampled, with information derived mostly from older collections by author BV at St. John-Thomas and Puerto Rico, and from a set of lionfish stomach contents from La Parguera in Puerto Rico sequenced by Harms- Tuohy et al. (2016). The latter identified 39 species from 16 families. A few additional sequences come from open-ocean sampling for larvae around the USVI, by Lamkin et al. (2009). The older collections from St. Thomas and Puerto Rico were collected by BV for recruitment and otolith studies as well as some taxonomic reviews (e.g., the genera Coryphopterus and Emblemariopsis). The recent additions of 19 species from St. John were collected by CJE and AME mainly for DNA confirmation of the species identification of diagnostic underwater photographs that serve as vouchers here, mostly of cryptobenthic fishes. No collections at St. John-Thomas or elsewhere on the PRP that provided DNA barcodes were expressly made for assembling an inventory of fish species-hence the absence of some of the most abundant and widespread shallow reef fishes in the barcode list presented here (e.g., the Bluehead Wrasse, Thalassoma bifasciatum).
We cannot directly compare barcode coverage of fishes at St. John-Thomas with that at other intensively barcoded locations noted above because neither the number of barcoded species nor the local species inventory have been comparably evaluated at any of those sites. The results of the present assessment of DNAbarcode coverage for the USVI and the remainder of the PRP highlight the usefulness of the DNA-barcode database for ancillary projects. Accumulating sequences for unrelated purposes, such as taxonomic reviews, stomach-content studies, larval or e-DNA surveys (environmental DNA, where water is sampled for dissolved DNA sequences), augments the general DNA-barcode coverage for specific biogeographic regions and helps confirm species identifications for faunal surveys.

Permits
Collections from St. John in 2021 were made under National Park Service Collecting Permit VIIS-2021-SCI-0006: We especially thank Thomas Kelley (Virgin Islands NP/ Virgin Islands Coral Reef NM) for facilitating that work.