Cryptic diversity of Eviota (Teleostei: Gobiidae) and their habitat use in the shallow waters of Okinawa Island

Eviota, known as dwarf gobies, is a diverse genus of marine fishes comprising more than 120 species found throughout the Indo-Pacific Ocean. Previous studies have suggested that morphologically similar cryptic species of Eviota may exist and that the species diversity may be significantly underestimated. This study evaluated the diversity of Eviota species in three habitat types (tidal flats, reef flats, and moats) in the shallow waters (0–1 m in depth at low tide) of Okinawa Island using morphological characteristics and molecular phylogenetic analysis. Eleven morphospecies were classified into 17 molecular operational units (MOTUs), and cryptic species for which speciation may have occurred relatively recently were detected in the morphospecies Eviota abax, Eviota japonica, Eviota queenslandica, and Eviota prasina. Most MOTUs exhibited only one habitat use, whereas three MOTUs were found in multiple habitats. Among each cryptic Eviota species, both habitat segregation and habitat overlaps were found. Therefore, habitat segregation among cryptic Eviota species may occur at spatial scales such as in this study, or at finer spatial scales, and multiple factors should be considered regarding speciation in Eviota.


Introduction
Eviota (Teleostei: Gobiidae), known as dwarf gobies, is a diverse genus of marine fish (Greenfield 2017).Eviota was originally proposed by Jenkins (1903), and Lachner and Karnella (1980) reviewed the diagnostic morphological characteristics of the genus (e.g., cephalic sensory-canal pore patterns, the shape of the male genital papilla, and body coloration) and summarized the geographical distribution.Subsequently, the description of new species continues to progress mainly based on morphology (e.g.Shibukawa and Suzuki 2005;Fujiwara et al. 2020), and 127 valid species are currently recognized throughout the Indo-Pacific Ocean except in the eastern Pacific region (Tornabene et al. 2021).In addition to its small body size and a large number of species, Eviota is characterized by a short life span (< 100 days) and, consequently, rapid generational turnover (Depczynski and Bellwood 2006).
Molecular phylogenetic analysis has been used to delimit species, leading to the discovery of many cryptic species, which are morphologically very similar but distinguishable via molecular analysis (Bickford et al. 2007).This analysis has greatly advanced the understanding of biodiversity and provides a standard method of species delimitation for many organisms (Fišer et al. 2018).Notably, Tornabene et al. (2015) identified the presence of multiple cryptic species of Eviota via molecular phylogenetic analysis, suggesting that the number of Eviota species may be significantly underestimated.Several other studies have used molecular data to detect cryptic species, with subsequent comparisons of morphological or coloration patterns.Tornabene et al. (2016) found cryptic species within Eviota sebreei (Eviota punyit); Greenfield and Tornabene (2014) recognized a cryptic lineage within Eviota nigriventris (Eviota brahmi); and Greenfield et al. (2019) recognized cryptic diversity within Eviota zebrina, many species of which were later described in Tornabene et al. (2021).However, studies performing molecular phylogenetic analysis of Eviota diversity have been insufficient for its geographic area and habitat.Three speciation modes (allopatric, sympatric, and parapatric) have been proposed for coral reef fish (Rocha and Bowen 2008).Previous studies have shown allopatric speciation by geographic barrier and sympatric speciation by niche partitioning in cryptobenthic reef fishes, including gobies (Brandl et al. 2018).Allopatric speciation is thought to be more likely to occur in cryptobenthic reef fishes with low dispersal capacity and limited adult mobility (Brandl et al. 2018).Taylor and Hellberg (2005) showed that allopatric speciation on a small geographic scale (within the Caribbean) occurred in parts of the goby genus Elacatinus.Sympatric speciation has been studied with a focus on habitat segregation, and Munday et al. (2004) demonstrated speciation by host shift in the coral-dwelling goby genus Gobiodon.Tornabene et al. (2015) suggested that local-scale speciation in Eviota within the Coral Triangle may be associated with niche partitioning (sympatric speciation) and geographic barriers to gene flow (allopatric speciation).In the early stages of sympatric speciation by habitat segregation, the species should have divided their habitats among themselves.However, it is unclear whether habitat segregation occurs among species that would have speciated relatively recently, such as in cryptic species.In the representative study on habitat segregation of cryptobenthic reef fishes, Herler (2007) showed that 21 species of seven genera of coral-and coral rock-associated gobiid fish, including Eviota, had clear spatial habitat segregation.A few Eviota species are associated with a single type of microhabitat (Tornabene et al. 2013a;Doll et al. 2021).Habitats overlap among some Eviota species, but little habitat overlap is observed among sister species, with few exceptions (Doll et al. 2021).Ahmadia et al. (2018) and Tornabene et al. (2013a) showed a few sister species that share habitat, such as E. sigillata/shimadai, as well as E. prasites/spilota.
This study investigated the diversity of Eviota species that were collected from three habitat types (tidal flats, reef flats, and moats) in the shallow waters of Okinawa Island using morphological characteristics and molecular phylogenetic analysis.I have tested the following hypotheses: i) cryptic species are recognized in Eviota in the shallow waters of Okinawa Island, and ii) habitat segregation and overlap occur among closely related species.

Field sampling
Eviota species were collected using hand nets during low tide between 2020 and 2022 in the shallow waters (Nakijin, Nago, Onna) of Okinawa Island (Fig. 1; Table 1).These sampling sites were 0-1 m in water depth at low tide.Three habitat types (tidal flats, reef flats, and moats) were identified at these sampling sites.Tidal flat: a coastal topography with a mud, sand, and rubble bottom that periodically emerges from below the surface of the sea with the ebb and flow of the tide.Reef flat: a coral reef topography with a flat bottom that periodically emerges from below the surface of the sea with the ebb and flow of the tide.Moat (shallow lagoon): a coral reef topography that develops between the shore and a reef flat with a sandy and coral rubble bottom; the water depth is shallow; it is often separated from the open sea at low tide.The Nakijin site (Fig. 1a; Table 1) is a coastal lagoon with patch reefs on sand and coral rubble bottoms, and contains moats and reef flats.Site Nago (Fig. 1b; Table 1) is a tidal flat adjacent to the mouth of a stream where the bottom is dominated by mud, sand, and coral rubble.Onna (Fig. 1c; Table 1) is a tidal flat adjacent to the mouth of a stream and also a coastal lagoon, and contains tidal flats, reef flats, and moats.Data on the habitat type for each Eviota specimen were recorded.

Morphological observation
The Eviota specimens were euthanized in an ice water bath, photographed, fixed in 10% formalin, and preserved in 70% ethanol.Before fixation in formalin, the right pectoral fin of each specimen was taken and stored in 99.5% ethanol until molecular analysis.All specimens were cataloged and deposited at the Okinawa Churashima Foundation.
Morphological species identification was performed based on the studies by Greenfield and Winterbottom (2016a), and subsequent studies describing Eviota species (Greenfield andRandall 2016, 2017;Greenfield and Winterbottom 2016b;Tornabene and Greenfield 2016;Greenfield and Suzuki 2016;Greenfield and Erdmann 2017;Greenfield et al. 2017Greenfield et al. , 2018Greenfield et al. , 2019;;Greenfield and Gordon 2019;Fujiwara et al. 2020;Greenfield and Erdmann 2020a, b;Winterbottom and Greenfield 2020;Tornabene et al. 2021) were based on fin counts, cephalic sensorycanal pore patterns, urogenital papillae, and pigment patterns.Fin counts and morphological measurements were performed using an ocular micrometer under a dissecting microscope (Nikon SMZ645).Cephalic sensory-canal pore patterns were observed as described by Lachner and Karnella (1980).Cyanine Blue staining was used to highlight the sensory pores and urogenital papillae (Saruwatari et al. 1997).

Molecular species delimitation
Total DNA was extracted using the HotSHOT method (Truett et al. 2000).A segment of the mitochondrial DNA cytochrome oxidase subunit 1 gene (CO1) was amplified Page 3 of 8 61 by polymerase chain reaction (PCR) in a thermal cycler (Gene Atlas 322, Astec, Fukuoka, Japan) using the primer pairs FISHCO1LBC (5ʹ-TCA ACY AAT CAY AAA GAT ATY GGC AC) and FISHCO1HBC (5ʹ-ACT TCY GGG TGR CCR AAR AATCA) (Ward 2012).For Eviota japonica, the newly designed primer pair CO1_EJF (5ʹ-TCA ACT AAC CAT AGG ACA TTG GCA C) and CO1_EJR (5ʹ-GAC CTC AGG GTG TCC GAA GAA TCA AAA), based on the FISHCO1LBC primer, FR1d primer (Ivanova et al. 2007), and complete mitochondrial sequence (Accession number AP019334), was used.PCR was performed using the KOD FX Neo DNA polymerase kit (Toyobo, Osaka, Japan) in a total reaction volume of 12.5 μl: 6.25 μl of 2 × PCR buffer, 0.5 μl of 10 μM of each primer, 2.5 μl of 2 mM dNTPs, 0.25 μl of KOD FX Neo, 0.25 μl of template DNA, and 2.25 μl dH 2 O.The PCR cycling conditions were as follows: 94 °C for 2 min; 10 cycles of denaturation at 98 °C for 10 s, annealing at 55 °C for 15 s, and extension at 68 °C for 20 s; 25 cycles of denaturation at 98 °C for 10 s, annealing at 53 °C for 15 s, and extension at 68 °C for 20 s; and a final extension at 68 °C for 2 min.After purification of amplified PCR samples, sequencing was performed using a BigDye™ Terminator v3.1 Cycle Sequencing Kit (Thermo Fisher Scientific, Waltham, MA, USA), followed by sequencing on a 3730xl DNA Analyzer (Thermo Fisher Scientific).The sequences were deposited in DDBJ/EMBL/GenBank (accession numbers LC753213-LC753261).
Three species delimitation methods were used for the CO1 dataset to delimit Molecular Operational Taxonomic Units (MOTUs).The first method was the General Mixed Yule Coalescent (GMYC) model (Fujisawa and Barraclough 2013).For the GMYC method, a Bayesian Inference (BI)  analysis was performed using BEAST2 (Bouckaert et al. 2019) with the GTR + I + G substitution model, which was inferred as the best fitting model using ModelTest-NG (Darriba et al. 2020), estimated base frequencies with four gamma categories, an optimized relaxed clock, and the Yule model.The length of the MCMC chain was 100 million, with sampling every 5,000 and the first 10% was discarded as burn-in.The effective sample size was confirmed to be greater than 200 using Tracer v1.7.1.Single-threshold GMYC analyses were performed on the webserver (https:// speci es.h-its.org/ gmyc/).The second method implemented included the multi-rate Poisson tree process (mPTP) model (Kapli et al. 2017).For the PTP method, the Maximum Likelihood (ML) tree was reconstructed using RAxML v8.2.10 (Stamatakis 2014) based on the GTR + I + G model with 1000 nonparametric bootstrap replicates.PTP analysis was performed using the web server (https:// mptp.h-its.org).The last method implemented was Assemble Species by Automatic Partitioning (ASAP) (Puillandre et al. 2021).ASAP analysis was performed on a web server (https:// bioin fo.mnhn.fr/ abi/ public/ asap/ asapw eb.html) with default settings.

Morphological identification
A total of 49 Eviota specimens were collected from the shallow waters of Okinawa Island and 11 morphospecies were identified (Fig. 2; Online Resource 1).Among these, Eviota hinanoae and Eviota bipunctata were recorded for the first time in Japanese waters.

Taxonomic issues
Additional sampling and detailed taxonomic studies are needed to confirm whether E. hinanoae and E. bipunctata in this study are truly their respective species.In particular, E. hinanoae of this study is likely not a true E. hinanoae because it is far from the type locality (Society Islands, French Polynesia: Tornabene et al. 2013b).Similarly, among multiple MOTUs of E. abax (1 and 2), E. japonica (1 and 2), E. prasina (1, 2, and 3), and E. queenslandica (1, 2, and 3), any one of which may be the true respective species, but none of which may be the true respective species.

Habitat use
Most MOTUs exhibited only one habitat use, whereas three MOTUs were found in multiple habitats (Fig. 2 E. queenslandica were classified into multiple MOTUs and showed various similarities and differences in habitat use (Fig. 2; Online Resource 2).Eviota abax 1 and 2 were found in tidal flats and moats, respectively.Eviota queenslandica 1 was found only in moats, whereas E. queenslandica 2 and 3 were found only in tidal flats.Eviota japonica 1 and 2 were found only in tidal flats, and were found to coexist in the same tidal flats at the Nago and Onna sampling sites.Eviota prasina 1 and 2 were found in the reef flat, whereas E. prasina 1 and E. prasina 3 were found in tidal flats and moats.Eviota prasina 1 and 2 coexisted in the same reef flat at the Onna sampling site, and E. prasina 1 and 3 coexisted in the same tidal flat at the Nago sampling site.

Discussion
This study shows how far we are from understanding the diversity of Eviota species.Molecular phylogenetic analysis of the mitochondrial DNA of the CO1 gene revealed that Eviota species collected from the shallow waters of Okinawa Island were divided into 17 MOTUs, and the number was more than expected from the morphological identification.Tornabene et al. (2015) also demonstrated the cryptic diversity of Eviota species within the Coral Triangle, which is known as the world's richest marine biodiversity hotspot.The fact that the cryptic diversity of Eviota species was observed in multiple regions suggests the possibility of finding cryptic diversity in other areas of the Indo-Pacific where Eviota occurs.
Previous studies have shown that niche partitioning, including habitat segregation, is associated with speciation in cryptobenthic reef fish, and that habitat segregation occurs across a differential spatial scale (Brandl et al. 2018).The most fine-scale habitat segregation is found in the coral-dwelling genus Gobiodon, with several species appearing as one or two species of host acroporid corals significantly more often than expected from the availability of these corals on the reef (Munday et al. 1997(Munday et al. , 2004)).Fine-scale habitat segregation also occurs within other Indo-Pacific gobiid genera Bryaninops and Pleurosicya (Herler 2007).Habitat segregation found in Eviota is coarser in scale, with closely related species found on different substrate types such as hard corals, soft corals, sponges, macroalgae, rubble, and sand (Tornabene et al. 2013a;Ahmadia et al. 2018;Doll et al. 2021).In the present study, most MOTUs exhibited only one habitat use, and habitat segregation was observed among multiple cryptic species (between E. abax 1 and E. abax 2, and between E. queenslandica 1 and E. queenslandica 2 and 3).This may indicate that Eviota speciated by habitat segregation.Nevertheless, habitat overlap was also observed between multiple cryptic species (between E. japonica 1 and 2, E. queenslandica 2 and 3, E. prasina 1 and 2, and E. prasina 1 and 3).Habitat overlap was also found between closely related species of Eviota in previous studies (Tornabene et al. 2013a;Ahmadia et al. 2018;Doll et al. 2021).This may indicate that habitat segregation of Eviota is occurring at a finer scale.In fact, Tornabene et al. (2016) showed that when two largely allopatric sister species (E.sebreei and E. punyit) co-occur at the same site, they will segregate into groups and avoid each other, rarely being found on the same coral head, even though they prefer the same morphotypes and species of coral.This suggests some competition and habitat segregation.Overall, it is important to consider that habitat segregation occurs at multiple spatial scales, especially for these very small fish.From another perspective, closely related species in this study might actually be part of a more complex, and therefore appear to have habitat overlap between closely related species.; e Eviota abax 2 (OCF-P10564, female, 14.7 mm SL); f Eviota afelei (OCF-P10467, female, 11.9 mm SL); g Eviota japonica 1 (OCF-P10501, male, 17.5 mm SL); h Eviota japonica 2 (OCF-P10569, male, 16.8 mm SL); i Eviota hinanoae (OCF-P4280, male, 12.0 mm SL); j Eviota rubrimaculata (OCF-P10440, male, 15.1 mm SL); k Eviota queenslandica 1 (OCF-P10455, male, 16.6 mm SL); l Eviota queenslandica 1 (OCF-P10441, sex unknown, 13.5 mm SL); m Eviota queenslandica 2 (OCF-P10537, female, 15.3 mm SL); n Eviota queenslandica 3 (OCF-P10549, male, 15.4 mm SL); o Eviota queenslandica 3 (OCF-P10550, female, 14.9 mm SL); p Eviota prasina 1 (OCF-P10567, male, 17.3 mm SL); q Eviota prasina 1 (OCF-P10456, female, 15.2 mm SL); r Eviota prasina 2 (OCF-P10470, male, 17.1 mm SL); s Eviota prasina 3 (OCF-P10559, female, 14.3 mm SL); t Eviota nigramembrana (OCF-P10463, male, 15.6 mm SL) Geographic barriers are another factor in the speciation of cryptobenthic reef fishes, including gobies (Brandl et al. 2018).Tornabene et al. (2015) indicated that speciation in Eviota occurred at small scales due to geographic barriers within the Coral Triangle.Tornabene et al. (2016) also found support for allopatric speciation in E. punyit and E. sebreei from the Coral Triangle.Moreover, Trimma gobies exhibit geographically distinct cryptic lineages in the Indo-Pacific region (Winterbottom et al. 2014).Additionally, Lachner and Karnella (1980) identified morphological differences among geographic populations of E. prasina.In the present study, an extensive survey was not performed; therefore, the possibility of geographic variation among the cryptic Eviota species identified herein cannot be excluded.
In conclusion, the cryptic diversity of Eviota species in the shallow waters of Okinawa Island was identified based on morphological observation and molecular phylogenetic analysis.Eviota is widely distributed in the Indo-Pacific Ocean (Greenfield 2017), suggesting that undiscovered cryptic species of Eviota may remain.In this study, both habitat segregation and habitat overlap among cryptic Eviota species were observed, which indicated that habitat segregation of Eviota may occur at spatial scales such as in this study, or at finer spatial scales.This highlights that multiple factors must be considered in association with speciation in Eviota.

Fig. 1
Fig. 1 Map showing the locations of the sampling sites near Okinawa Island in the northwestern region including photographs of the sampling sites (a, Nakijin; b, Nago; c, Onna)

Fig. 2
Fig. 2 Maximum likelihood (ML) tree based on mitochondrial DNA sequences from the CO1 gene with multiple species delimitation and habitat use of Eviota species collected in the shallow waters of Okinawa Island.Numbers on branches indicates the ML bootstrap

Table 1
Substrate and habitat of each sampling site of Okinawa Island ; Online Resource 2).Eviota japonica 1 and 2, E. abax 1, E. queenslandica 1 and 2, and E. hinanoae were found only in the tidal flats.Eviota afelei, E. bipunctata, E. teresae, E. smargdus, E. abax 2, E. queenslandica 1, and E. nigrimaculata were found only in the moat.Eviota prasina 2 was found only on the reef flat.Eviota rubrimaculata and E. prasina 3 were found in the tidal flats and moats.Eviota prasina 1 was found in tidal flats, moat, and reef flat.