Ichthyofaunal diversity and vertical distribution patterns in the rockpools of the southwestern coast of Yakushima Island , southern Japan

The community composition of rockpool fish on the southwestern coast of Yaku-shima Island, southern Japan, in the northwest Pacific was investigated by sampling of 22 rockpools and recording the range of vertical heights (a total of 76 sampling events from May 2009 to February 2010). A total of 72 species belonging to 19 families were collected from the study site. This species richness is the highest recorded of similar studies undertaken worldwide, reflecting the highest diversity of coastal fishes in the western Pacific. Increases in species richness due to transient and accidental visitors increased the total number of species in the lower vertical zones. Variations in the vertical distribution pattern of the resident and transient species suggests habitat partitioning and/or physical preferences for a particular habitat of each species. Color images of rockpool fish recorded at the site and a list of all the voucher specimens are provided.


INTRODUCTION
Yaku-shima Island is located in the subtropical region of southern Japan, in the northwestern Pacific (30°20′ N, 130°32′ E, approximately 60 km south-southwest of the southernmost tip of Kyushu Island; Figure 1), surrounded by a strong, warm ocean current from the south (the Kuroshio Current) and multidirectional, irregular currents from Kyushu Island (Motomura et al. 2010).Despite the uniform rocky shore that surrounds almost the entire island, it is of interest biogeographically because of the mix of ocean currents.Ichthyofaunal surveys have been undertaken resulting in a total of 988 marine and estuarine fish species being listed from compiled literature sources, underwater photographs and voucher specimens (Motomura et al. 2010;Motomura and Aizawa 2011;Murase et al. 2011).
To identify the temporal dynamics of the coastal fish assemblage of Yaku-shima Island, Murase (2013) quantitatively sampled the intertidal rocky shore and investigated the community structure of rockpool fish on the southwestern coast of the island over four seasons.His results showed a statistically stable diversity throughout the year at the study site, although seasonal variability of species richness and fish density was detected.Furthermore, the species richness of the rockpool fish recorded in Murase's study (54 species) represented the highest number compared to other similar studies in the Indo-Pacific region (Moring 1986;Varas and Ojeda 1990;Willis and Roberts 1996;Beckly 2000;Griffiths 2003a;Castellanos-Galindo et al. 2005;Arakaki and Tokeshi 2006;

LISTS OF SPECIES
the lower and exposed zones shared the same elevation at 0.85-0.94m, the zones differed in terms of their exposure to waves, or specifically, the exposed zone was much nearer to the surge zone and was more affected by waves than the lower zone.Due to its location, it was not possible to enter the exposed zone to sample fish during autumn and winter (October and February in this study), as the tidal height at low tide was higher here in the daytime.Therefore sampling events were divided in two terms: Term I, in May and July 2009 included the exposed zone, and Term II, in October 2009 and February 2010 excluded the exposed zone.Fish sampling was performed once at each rockpool in each month (22 pools × 2 months + 16 pools × 2 months = 76 sampling events in total) by completely emptying the rockpool with a bucket and a mug as in Murase (2013).The rockpool volumes ranged from 0.05−1.79m 3 .The captured fish were immediately killed in an ice box, packed in plastic bags and transported to the laboratory.They were fixed with 10% formalin and conserved in 70% alcohol.Some were photographed when fresh to provide good color images.All the specimens collected were deposited in the ichthyological collection of the Kanagawa Prefectural Museum of Natural History, Japan (KPM-NI) and the numbers are listed in the Appendix.
The taxonomic arrangement of families followed Nelson (2006).Fish identification and standard lengths (SL) were according to Nakabo (2013), and the taxonomic-revised groups were described as follows: Murase (2013) considered species of Scorpaenodes as "Scorpaenodes spp." because of a taxonomic problem between S. guamensis and S. scaber following a personal communication from Dr. H. Motomura (Kagoshima University Museum), however, thereafter Motormura (2014) stated that S. scaber was an endemic species of Murase et al. 2010;Cox et al. 2011). But, Murase (2013) did not consider the spatial distribution pattern of the rockpool fish assemblage in relation to the vertical levels of each rockpool sampled.This study aims to describe the rockpool ichthyofauna of the entire intertidal zone on the southwestern coast of Yaku-shima Island and the variability of species in relation to their vertical height, by updating Murase's (2013) data as follows: sampled rockpools were divided into various vertical zones, and included rockpools at exposed zones which had been not included in Murase (2013).Color images of the rockpool fish and a catalogue of the voucher specimens collected in this study are also provided.

MATERIALS AND METHODS
The study site was located on a rocky shore of Kurio Beach, on the southwestern coast of Yaku-shima Island (30°16′ N, 130°24′ E; Figure 2), detailed information of which was provided in Murase (2013).To record the entire fish fauna inhabiting intertidal rockpools at the site, 22 rockpools were randomly selected that were located at different vertical heights from the nearly lowest low water level (NLLWL) and fish sampling was carried out during the daytime at spring tides over four seasons (May, July, October 2009 andFebruary 2010).Heights of each rockpool surface from the NLLWL were measured by the two-pole method (see Murase et al. 2010: Figure 1) along with the estimated tidal height around the Kurio Beach which was provided by the Hydrographic and Oceanographic Department of the 10th Regional Coast Guard Headquarters (Kagoshima, Japan).Rockpools were categorized into the following five vertical zones: higher zone (n = 4, 2.01-2.22m from NLLWL); mid-higher zone (4, 1.37-1.75m); mid-lower zone (4, 1.32-1.34m); lower zone (4, 0.85-1.02m); and the exposed zone (6, 0.16-0.94m).Of these, although Murase | Rockpool fish diversity of Yaku-shima Island, southern Japan Australia and was not distributed in Japanese waters.Following this, specimens of Scorpaenodes were identified as S. guamensis.Motomura et al. (2010) reported an unidentified species of Enneapterygius as E. sp. 5, and that species has recently been identified as the nuptial phase of E. leucopunctatus, although the distinctive characteristics of females of E. leucopunctatus and E. hemimelas still remain unclear (S.Tashiro, Kagoshima University, personal communication).Additionally, some Enneapterygius individuals were treated as "E.leucopunctatus or E. hemimelas".Enneapterygius sp. in Meguro (2013), which is the same species as Enneapterygius sp. 1 of Motomura et al. (2010), were listed here as "Enneapterygius sp.(sensu Meguro, 2013)".Murase (2013) listed Praealticus margaritarius, but the scientific name was revised as P. bilineatus (Nakabo 2013).Information on Palutrus cf.reticularis (Gobiidae), Istiblennius edentulus and I. enosimae (Blenniidae) was based on Murase (2013).All species recorded in this study were divided into three life-style status categories (residents, transients and accidental visitors) according to the definition of Murase (2013).
The occurrence rates of the fish species in each vertical zone were calculated in each term as the "frequency of occurrence in the species within a zone/ frequency of sampling event within the zone × 100 (%)".Four rockpools were sampled twice (2 months) within a single zone, making eight sampling events in total within a single zone for each term, with the exception of the exposed zone, where 12 sampling events were carried out during Term I only.Therefore, for example, the occurrence rate of a fish species collected in five sampling events within the higher zone during a single term was shown as, "5/8 × 100 = 62.5 (%)".The occurrence rates of each recorded species are listed in Table 1.
The relationships between the vertical levels and the number of species in each life history status are shown in Figure 10.In Term I, total species richness did not vary much from the higher to mid-lower zones (a range of 15-19 species), but this increased to 30 and 41 species in the lower and exposed zones, respectively.Similarly, in Term II, the total number of species remained stable between the higher and mid-lower zones (12-16 species) and increased ~twofold in the lower zone during Term II (26 species).Species richness of the resident species ranged from 8-10 species (Term I) and 7-10 species (Term II), and did not show any clear vertical distribution.Species richness of the transients varied from 7-8 (Term I) and 3-9 (Term II) from the higher to the lower zone, however, 17 species of transients were recorded in the exposed zone of Term I. Species richness of the accidental visitors increased in inverse proportion to vertical height.During Term I, 0−3 species of accidental visitors occurred in the higher to mid-lower zones, and 12 and 15 species were recorded in the lower and exposed zones, respectively.During Term II, no accidental visitors were recorded in the higher and mid-lower zones, however, seven species were recorded in the lower zone.
The six common species showed various occurrence patterns along the vertical zones (Table 1).P. bilineatus occurred more frequently between the mid-higher and lower zones during both terms.P. tanegasimae was generally found in the higher rockpools, and in particular, was always (100%) found in the higher zone.R. nitidus was less frequent at the higher zones, but more common at the lower zone.Ev. prasina was distributed similarly to R. nitidus, but always occurred in the two lowest zones.B. cocosensis was found equally throughout all vertical zones (62.5-75.0%).En. philippinus was found more frequently (75.0−100.0%)at the lower zone in both terms.The two gobiid residents (which were not common       species), B. coalitus and Priolepis semidoliata, showed inverse tendencies in their vertical distribution with the former occurring more frequently in the higher zone and the latter occurring only in the lower and exposed zones.The predominant transients also had different vertical distribution patterns.En. etheostoma, T. cupido and Plectroglyphidodon leucozonus were more frequent in the lower rockpools, and a tripterygiid, Enneapterygius sp.occurred only in the exposed zone.In contrast, Abudefduf sordidus, A. notatus and Girella leonina, were most frequently recorded at the higher zone.

DISCUSSION
With the addition of the exposed zone category, 72 species were recorded in this study, which was a greater number than the 54 species of rockpool fish recorded in Murase (2013).Comparisons between these results and those from other regions of the world are listed Murase | Rockpool fish diversity of Yaku-shima Island, southern Japan in Table 2. Gibson and Yoshiyama (1999) noted the potential effects of rockpool size on species richness in fish communities.In fact, species richness, numbers of individuals and biomass of each pool were positively associated with pool size (Mahon and Mahon 1994) and larger pools can permit permanency of schooling species (Macieira and Joyeux 2011).Therefore, studies targeting obviously large rockpools should be removed for regional comparison of species richness to avoid the bias arisen from large pool size.Some Taiwanese studies of rockpool fish communities recorded a large number of fish species (124-177) compared with other regions listed in Table 2 (Lee 1980a, b;Mok and Wen 1985).Lee (1980b) studied a large rockpool with a 400 m 2 surface area and 0.5-2.0m in depth, and a number of subtidal species, such as muraenids, apogonids and labrids, and adult size of those species were captured from sites across Taiwan.So, those studies should not be considered as true intertidal environment studies (Castellanos-Galindo et al. 2005;Murase 2013).Thomson and Lehner (1976) also investigated the fish assemblages of rockpools of a similar size to the Taiwanese study, in the Gulf of California.Lardner et al. (1993) recorded 99 fish species from a relatively large rockpool on the southeastern coast of Australia over 19 years, with the species richness not only reflecting the species diversity in that area of the world but also the rockpool size and long sampling duration (Griffiths 2003a, b).For these reasons, Table 2 does not contain the results of these studies mentioned above.Consequently, this study of the southwestern coast of Yaku-shima Island recorded the highest number rockpool fish species worldwide, to date (Table 2).These results agree with findings suggesting that the western Pacific region is generally considered a center of coastal fish diversity (Mora et al. 2003;Carpenter and Springer 2005;Tittensor et al. 2010), and even higher species richness of rockpool fish assemblages are likely to be recorded in future studies in the western Pacific region.
Although the total number of species did not differ between the higher and mid-lower zones, there was a

South Africa
Western South Africa 9-20 Bennett and Griffiths (1984), Prochazka and Griffiths (1992), Prochazka (1996) Southern South Africa 16-35 Christensen and Winterbottom (1981), Bennett and Griffiths (1984), Beckley (1985), Bennett (1987) Figure Murase | Rockpool fish diversity of Yaku-shima Island, southern Japan clear increase in species richness of accidental visitors, which led to an increase in the total number of species in the lower zone during both terms (Figure 10).At the exposed zone, the species richness of transients and accidental visitors increased maximally, resulting in the highest total number of species among the zones.A negative correlation between species richness of rockpool fish and vertical height has already been demonstrated (Castellanos-Galindo et al. 2005;Murase et al. 2010), and this could have been due to more occasional species entering the intertidal zone from the subtidal zone for feeding and/or to avoid predators (e.g.Black and Miller 1991;Rangeley and Kramer 1995a, b).
The results of the present study supported this finding.
The results of this study showed the different vertical distribution patterns of the six most common species found in rockpools.Of the common blenniid species, Praealticus tanegasimae occurred more frequently in the higher zone and Rhabdoblennius nitidus occurred more frequently in the lower and exposed zones, whereas P. bilineatus showed a median distribution tendency between these two species (Table 1).These blenniids have similar feeding habits, feeding on algae and detritus (Hundt et al. 2014), but their different vertical distribution tendencies could moderate interspecific competition and allow them to predominantly inhabit intertidal zones at the same site.Eviota prasina showed a similar distribution pattern to R. nitidus but differs in its diet; E. prasina feeds predominantly on detritus and some invertebrates, and R. nitidus feeds mainly on algae (Murase, unpublished data).This difference in feeding habits may enable the coexistence of these two species.In contrast, Bathygobius cocosensis showed a uniform vertical distribution tendency at the study site, and its size distribution tends to be smaller in higher zone through two terms (Murase, unpublished data) suggesting the intraspecific competition (Gibson and Yoshiyama 1999;Zander et al 1999).Enneapterygius philippinus was distributed throughout the lower zone, and this likely reflected its physical preference for a habitat that was closed and with a good water supply.The two resident gobies, B. coalitus and Priolepis semidoliata, occurred more frequently in the higher and lower pools, respectively.More detailed studies on the ecology of these species are needed to understand the mechanisms behind their distribution patterns in the intertidal zone.
Tidepool environments can act as temporal habitats for the different life-stages of fish in various regions (Gibson and Yoshiyama 1999;Murase et al. 2010;González-Murcia et al. 2012;Murase 2013), and a number of transient species were recorded in this study.These transients comprised two groups in terms of their vertical distribution pattern.First, Thalassoma cupido, Enneapterygius etheostoma, Enneapterygius sp. and Plectroglyphidodon leucozonus occurred more frequently in the lower pools, suggesting their temporal use of the intertidal environment without intertidal adaptation, as mentioned in Zander et al. (1999).The other group containing Abudefduf sordidus, A. notatus and Girella leonina was in the higher pools, even though their adult forms generally inhabited the subtidal rocky shore (Masuda and Kobayashi 1994;Araga 1997).Small individuals of A. sordidus were recorded in higher rockpools than the congeneric A. vaigiensis on the Pacific coast of Japan (Eto et al. 2005), and the girellid, G. elevata, in southeastern Australia was recorded in the highest rockpools (1.4-1.8 m vertical height) by Griffiths et al. (2003).Physiological and/or behavioral characteristics adapted to an intertidal life may allow the juveniles of these transient species to inhabit these rockpool environments (Griffiths et al. 2003) and may moderate interspecific competition.The five transient tripterygiids that were recorded only in the exposed zone (Enneapterygius sp., E. leucopunctatus, E. bahasa, Helcogramma fuscipectoris and E. hemimelas; Table 1) were found at 10 or 20 m maximum depth in previous ichthyofaunal surveys in southern Japan (Motomura et al. 2013;Motomura and Matsuura 2014).The results of this study suggested that these tripterygiids also inhabited rockpool environments that were affected by the open sea.
In addition to a series of ichthyofaunal surveys (Motomura et al. 2010;Motomura and Aizawa 2011;Murase et al. 2011), this study updates the fish fauna of Yaku-shima Island with the following voucher specimens: Specimens of two gobiid species, Palutrus cf.reticularis (Figure 5A, KPM-NI 30017) and Priolepis borea (Figure 5B, KPM-NI 29921), representing the first record of this species on Yaku-shima Island, and the specimen of P. borea was the southernmost record of this species in Japanese waters (Akihito et al. 2013), although Murase (2013) listed this species he did not provide the voucher species information.A juvenile labrid specimen (KPM-NI 24797) identified as Halichoeres margaritaceus in Motomura et al. (2010) was herein re-identified as Halichoeres nebulosus, and that specimen, along with another (KPM-NI 29908, 30027), represent the first record of H. nebulosus as a voucher specimen at Yakushima Island, although they were previously known only from underwater photographs in Motomura et al. (2010).

ACKNOWLEDGMENTS
I would like to express my sincere thanks to the following people who contributed to this study: Tomoki Sunobe (Tokyo University of Marine Science and Technology) for his help and advice during this study; Yusuke Miyazaki (Kanagawa Prefectural Museum), Masatoshi Meguro (Kagoshima University), and Mikio Murase | Rockpool fish diversity of Yaku-shima Island, southern Japan Watai (National Research Institute of Far Seas Fisheries) who helped with the field sampling; Hiroshi Senou (Kanagawa Prefectural Museum) managed the voucher specimens, provided taxonomic information to identify the Labridae and Gobiidae, and took the photographs of the preserved specimens; Hiroyuki Motomura (Kagoshima University Museum) and Satokuni Tashiro (Kagoshima University) who provided taxonomic information to identify the scorpaenid and tripterygiid species, respectively; Rie Takahashi (Kanagawa Prefectural Museum) who managed the voucher specimens deposited in the museum; the staff of the Hydrographic and Oceanographic Department of the 10th Regional Coast Guard Headquarters (Kagoshima, Japan) who provided tidal information about the study site.I also thank two anonymous reviewers for their valuable comments on the manuscript.This study was partly supported by the Japan Society for the Promotion of Science (Research Fellowship for Young Scientists No. 21•6886 to A. Murase).

Figure 1 .
Figure 1.Map showing the position of Yaku-shima Island.

Figure 2 .
Figure 2. Rocky shore environment of Kurio Beach, southwestern coast of Yaku-shima Island.Small window showing the rockpools at the study site.Photos taken May 2009.
Figure 10.Spatial variation of the species richness for each life-style category (red circles, total of all categories; blue diamonds, residents; green squares, transients; yellow triangles, accidental visitors) in relation to the vertical height of pools.Abbreviations of x-axis: H, higher zone; MH, midhigher zone; ML, mid-lower zone; L, lower zone; Ex, exposed zone.The upper and lower graphs show data from Terms I and II, respectively.

Table 1 .
List of rockpool fish species and their occurrence rates in each vertical zone at Kurio Beach, Yaku-shima Island.Families and species are listed in order of total abundance.

Table 2 .
Variation of regional species richness in rockpool fish assemblages worldwide.