Fig. 1 in Fig. 1 in Disappearance and Return of an Outbreak of the Coral-killing Cyanobacteriosponge in Southern Japan.

Disappearance and Return of an Outbreak of the Coral-killing Cyanobacteriosponge Terpios hoshinota in Southern Japan Masashi Yomogida1, Masaru Mizuyama2, Toshiki Kubomura3, and James Davis Reimer1,2,3,4,* 1Molecular Invertebrate Systematics and Ecology Laboratory, Department of Biology, Chemistry and Marine Sciences, Faculty of Science, University of the Ryukyus, 1 Senbaru, Nishihara, Okinawa 903-0213, Japan. E-mail: [email protected] 2Molecular Invertebrate Systematics and Ecology Laboratory, Graduate School of Science and Engineering, University of the Ryukyus, 1 Senbaru, Nishihara, Okinawa 903-0213, Japan. E-mail: [email protected] 3Molecular Invertebrate Systematics and Ecology Laboratory, Graduate School of Science and Engineering, University of the Ryukyus, 1 Senbaru, Nishihara, Okinawa 903-0213, Japan. E-mail: [email protected] 4Tropical Biosphere Research Center, University of the Ryukyus, 1 Senbaru, Nishihara, Okinawa 903-0213, Japan (Received 6 October 2016; Accepted 21 March 2017; Published 19 April 2017; Communicated by Yoko Nozawa) Masashi Yomogida, Masaru Mizuyama, Toshiki Kubomura, and James Davis Reimer (2017) Terpios hoshinota is cyanobacteriosponge that can cause serious damage to coral reef ecosystems by undergoing rapid breakouts in which it smothers and encrusts hard substrates, killing living sessile benthic organisms and reducing biodiversity of the affected area. The reasons for these outbreaks are still unclear, as are long-term prognoses of affected reefs. Some reports have suggested outbreaks may not be permanent, but very little long-term monitoring information exists. In this study, we report on a T. hoshinota outbreak (~24% coverage) at Yakomo, Okinoerabu-jima Island, Kagoshima, Japan between 2010 to 2014. In this period, the existing outbreak was seen to almost completely disappear (~0%) after unusually severe Typhoon Songda passed by in early June 2011. After this, macroalgae and a cyanobacterial bloom became the dominant benthos, but by September 2014, T. hoshinota coverage had recovered to approximately half of its pre-typhoon coverage, suggesting the conditions that had caused the outbreak still persisted at Yakomo. While the conditions promoting T. hoshinota growth at this site remain uncertain, it appears that subtropical typhoons could play an important role in the dynamics of T. hoshinota outbreaks and disappearances. Key words: Terpios hoshinota, Outbreak, Phase shift, Typhoon, Subtropical coral reefs. *Correspondence: Tel: +81-98-895-8542. Fax: +81-98-895-8577. E-mail: [email protected] BACKGROUND Sponges are filtration-feeding organisms, and provide habitat for a variety of marine life including prokaryotes, shrimp, worms, hydroids, zoantharians, and fish (van Soest et al. 2012; Montenegro et al. 2015). In addition, sponges contain symbiotic bacteria, and produce organic matter that contributes to the nitrogen cycle, and thus play an important role in marine ecosystems (van Soest et al. 2012). On the other hand, there are sponge species that can cause serious damage to marine ecosystems. The sponge species Terpios hoshinota Rützler and Muzik, 1993 (phylum Porifera, class Demospongiae, order Suberitida, family Suberitidae) is one such species. This species is in symbiosis with numerous cyanobacteria and is a photoautotrophic organism (Hirose and Murakami 2011; Liao et al. 2011). Distributed in the coral reefs of the Indo-Pacific, it often grows over other benthos, and has been known to undergo outbreaks in shallow coral reefs, causing death to hard corals and other benthos (Rützler and Muzik 1993; Reimer et al. 2011b). By growing approximately 1-2 mm a day (Liao et al. 2011), T. Zoological Studies 56: 7 (2017) doi:10.6620/ZS.2017.56-07 1 hoshinota has been reported to overgrow and kill numerous hard coral including genera Lobophylia, Montipora, Acropora, Merulina, and Goniastrea, resulting in the death of wide areas of coral reefs (Rützler and Muzik 1993). Terpios hoshinota was first discovered in Guam (Bryan 1973), and later reported from the Northern Mariana Islands, Western Caroline Islands, the Philippines, and American Samoa (Plucer-Rosaria 1987). In more recent years, the species has been reported from southern Taiwan (Liao et al. 2007), Australia’s Great Barrier Reef (Fujii et al. 2011), Sulawesi and Java (de Voogd et al. 2013), and most recently it has been recorded in the Indian Ocean, in the Maldives (Montano et al. 2015) and Mauritius (Elliott et al. 2016). Terpios hoshinota has also been confirmed from the Ryukyu Islands in southern Japan. In 1985-1986 there was an outbreak on the reefs of Tokunoshima Island, Kagoshima, in which many coral colonies were killed, particularly at Yonama coast (Rützler and Muzik 1993), where the species covered up to 87.9% of substrate (Marine Park Center Foundation 1986). From the thin, black, ‘coating’ this species makes over other benthos, it was reported as the “Black Death” in local media at that time. In subsequent investigations, another outbreak was confirmed at Yakomo site on Okinoerabu-jima Island, Kagoshima, in 2010 (Reimer et al. 2011a). However, these outbreaks may not necessarily be stable. In a re-investigation of the Yonama, Tokunoshima Island site approximately 25 years after the initial T. hoshinota outbreak, results showed that the species had disappeared (= 0%), and the dominant benthos had become Acropora hard coral colonies (Reimer et al. 2011a). These results suggest T. hoshinota outbreaks are not necessarily ‘dead-ends’ and coral reefs may recover from outbreaks. However, much remains to be understood regarding the ecology of T. hoshinota and the mechanisms of outbreaks. In this study, in order to understand the progression of outbreaks, we conducted field surveys between 2010 and 2014 to follow changes in T. hoshinota coverage at an outbreak site, Yakomo, on Okinoerabu-jima Island in the Ryukyu Islands. Over the same period we also recorded changes in other benthic categories. We then compared the results of this study to previous research and briefly discuss possible factors that may lead to increases or decreases in T. hoshinota cover. MATERIALS AND METHODS Study area From March 2010 to September 2014, transect surveys were performed at Yakomo (27°20'N, 128°32'E), China Town, on Okinoerabujima Island, Kagoshima Prefecture, Japan, in the Ryukyu Islands (Fig. 1). On subtropical Okinoerabu-jima Island, the average sea surface temperature is 23°C (China Town homepage: www.town.china.lg.jp/ modules/kikakushinkou/index.php?content_id = 166). Yakomo, located on the southwest coast of Okinoerabu-jima Island, has a white sandy beach approximately 3 km in length. A coral reef runs parallel to the sandy beach, with a moat of approximately 180 m width from crest to shore. Between the sandy beach and reef margin, the moat has a depth of up to 5 m (at high tide). The moat terrain generally consists of scattered macroalgae on sand with a few stony corals. In some areas there is carbonate substrate extending for tens to hundreds of square meters. Terpios hoshinota identification In this study, we surveyed the same location within Yakomo as previously reported (Reimer et al. 2011b). This previous report also identified the black encrusting cyanobacteriosponge at this site as T. hoshinota based on both the external form and internal spicules, with further confirmation of species identity via molecular analyses. In the field, Terpios hoshinota can be identified easily as it forms a < 1 mm thin coating that is slightly thicker at areas sheltered from waves and current. The color is bright gray in areas that are exposed to sunlight, and generally darker gray, black, or dark brown in shaded areas. There are osculae (2~3 mm in diameter) visible on the body with peripheral radially arranged astrorhizae (see Fig. 1 in Reimer et al. 2011b). Transect surveys Survey of Terpios hoshinota We utilized the line-intercept transect (LIT) method to investigate benthos. Surveys were carried out ten times (March, August, September, October, December in 2010; June, October in 2011; February and May in 2012; and September 2014). Transects were 10 m length and 1 m page 2 of 10Zoological Studies 56: 7 (2017) Fig. 1. (A) Location of Okinaerabu-jima Island, Kagoshima, Japan in the northwestern Pacific and, (B) map of Yakomo coast on Okinoerabu-jima Island. Red dotted box shows the Terpios hoshinota survey area, white dotted lines show the approximate area of T. hoshinota along the coast, and red solid lines approximate locations of permanent transects. Images from Google Earth. Okinoerabu-jima Island Pacific Ocean reef moat reef crest reef slope Okinoerabu-jima I. Pacific Ocean East China Sea Japan China Taiwan (A) (B) in width (Beenaerts and Berghe 2005). Five transects were conducted for each survey, except for in September 2014 when ten transects were conducted. The surveys from August 2010 to May 2012 were permanent transects (marked with pegs nailed into the carbonate substrate), with four placed to perpendicular to shore, and one parallel, and all at 2-3 m depths. However, initial transects from March 2010 were not permanent but placed within the Yakomo survey area, while in September 2014 the pegs marking transects had been lost, likely from typhoons, and again we performed nonpermanent transects within the survey area (Fig. 1). All transects (permanent and non-permanent) were separated by 10 to 30 m intervals from neighboring transects on the reef, within an area of approximately 150 × 100 m (Fig. 1). Digital camera images for each transect were taken. Each transect consisted of 25-40 images. We used Adobe Illustrator CS4 to create a continuous connected transect image for each transect that was utilized in subsequent analyses. Benthic survey image analyses Benthos captured in survey images were divided into the following categories: 1) T. hoshinota, forming a thin coating on coral or hard sediment, black or light gray in color; 2) Macroalgae, including both red and green algae, not including Ceratodictyon spongiosum (see #3 below); 3) Sponge algae (C. spongiosum), easily distinguishable in the field by dark green or black branching form; 4) Cyanobacteria, visible brown mat-like colonies on the seafloor; 5) Hermatypic corals, including scleractinians and the hydrozoan genus Millepora; 6) Dead coral, completely white or bleached page 3 of 10Zoological Studies 56: 7 (2017) coral, may be covered in algae; 7) Other benthos, including giant clams, zoantharians, soft corals, and all other organisms not covered by categories 1 to 6 above, including mobile sea cucumbers and sea urchins; 8) Sand/carbonate, generally sand, unconsolidated small rubble, or hard carbonate; 9) Unknown, portions of images hard to see due to shading or lack of focus. The above benthic categories were measured to the nearest mm. Values of each benthic category were then displayed as a percentage the total length (1000 cm). In cases when length of categories could not be confirmed due to twisting of the tape measure or shading, we used ImageJ (National Institutes of Health, USA) to obtain measurements. Extent of T. hoshinota outbreak To determine the extent of T. hoshinota on the Yakomo coast, we conducted a simple snorkel survey in September 2014. Teams of snorkelers swam in both directions (roughly southeast and northwest; Fig. 1) along the coast from the entry point at the survey area at Yakomo. The survey stopped when no T. hoshinota was noticed for 100 m since the previous last sighting of T. hoshinota. GPS coordinates were taken at the west and east ends of the T. hoshinota outbreak, and we then prepared a simple map based on coordinates. Typhoons Typhoon data (windspeed, wave height) were obtained from the Japan Meteorological Agency homepage (http://www.jma.go.jp/jma/index.html). As there was no locally available wave data for Okinoerabu-jima Island, we utilized data from the closest available point off Naha, Okinawajima Island, Okinawa Prefecture (26°12'44.4"N, 127°40'44.8"E; Ministry of Land, Infrastructure and Transport Port Authority, National Harbor Ocean Wave Information Network (Naufasu)). Statistical analyses The percentage data of transects were analyzed in R version 3.3.2 (R Core Team 2016) with the Bray-Curtis similarity index calculated using the percentage data. Subsequently, we also performed a cluster analysis using the Bray-Curtis similarity index. Non-parametric multi-dimensional scaling (NMDS) was utilized for examining the similarity of each survey’s data utilizing all nine benthic categories. RESULTS Transect surveys The results of the LIT surveys from March 2010 until September 2014 (n = 55 transects) are shown in figure 2 and table S1. Sand and carbonate was the dominant category (Fig. 2), accounting from 50.2% to 89.4% of the total benthic coverage in surveys. Among biological categories, Terpios hoshinota was the dominant category from March 2010 to December 2010, and again in September 2014. Macroalgae was dominant in June 2011 and May 2012, while cyanobacteria mats were dominant in February 2012 (Fig. 2). From March 2010 until October 2010, T. hoshinota cover was observed to be 24% or higher (Fig. 2). In December 2010, coverage dropped to 17.6 ± 7.0%, and it then decreased again to 0.02 ± 0.04% in June 2011. Between June 2011 and May 2012, T. hoshinota coverage was very low, between 0.02 to 0.3%. However, T. hoshinota did not disappear completely, with a few patches of diameters of 5 to 15 cm still present. By September 2014, the coverage had recovered to 11.4 ± 6.8%. Macroalgae also showed some differences between surveys. Coverage in May 2012 was the highest observed (13.6 ± 6.6%), and was less than 10% in all other surveys (Fig. 2). The large majority of macroalage (~98%) was Actinotrichia sp. and the remaining 2% was grass-like seaweed (turf algae). Cyanobacteria mat coverage was 0% in nine of ten surveys, with the only exception in October 2011, when this was the dominant biological benthic category, covering 39.9 ± 7.5% (Fig. 2). NMDS and similarity analyses both showed three major groupings: 1) March 2010 to December 2010 plus September 2014 (when Terpios was the dominant biological benthos); 2) June 2011 + February 2012 + May 2012, when Terpios was absent; and October 2011, when cyanobacteria was the dominant biological benthos (Fig. 3). No large differences were seen between surveys for sponge algae, hard corals, dead coral, other benthos, or unknown categories, and these values were generally very low, and most often not above 5% in any given survey (Fig. 2). Hard corals were exclusively encrusting morphologies, with no page 4 of 10Zoological Studies 56: 7 (2017) Fig. 2. Coverage of different benthic categories at Yakomo from March 2010 to September 2014. Note the large time gap between the surveys in May 2012 and Sept 2014. 0 5 10 15 20 25 30 35 40 Mar. 2010 Aug. 2010 Sep. 2010 Oct. 2010 Dec. 2010 June 2011 Oct. 2011 Feb. 2012 May 2012 Sep. 2014 Terpios hoshinota 0 5 10 15 20 25 Macroalgae 0 1 2 3 4 5 6 7 8 9 10 Ceratodictyon spongiosum 0 5 10 15 20 25 30 35 40 45 50 Cyanobacteria 0 2 4 6 8 10 12 14 16 18 20 Hard corals 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 Dead coral 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Other benthos 0 10 20 30 40 50 60 70 80 90 100 Sand/carbonate 0 2 4 6 8 10 12 Unknown Mar. 2010 Aug. 2010 Sep. 2010 Oct. 2010 Dec. 2010 June 2011 Oct. 2011 Feb. 2012 May 2012 Sep. 2014 Mar. 2010 Aug. 2010 Sep. 2010 Oct. 2010 Dec. 2010 June 2011 Oct. 2011 Feb. 2012 May 2012 Sep. 2014 Mar. 2010 Aug. 2010 Sep. 2010 Oct. 2010 Dec. 2010 June 2011 Oct. 2011 Feb. 2012 May 2012 Sep. 2014 Mar. 2010 Aug. 2010 Sep. 2010 Oct. 2010 Dec. 2010 June 2011 Oct. 2011 Feb. 2012 May 2012 Sep. 2014 Mar. 2010 Aug. 2010 Sep. 2010 Oct. 2010 Dec. 2010 June 2011 Oct. 2011 Feb. 2012 May 2012 Sep. 2014 Mar. 2010 Aug. 2010 Sep. 2010 Oct. 2010 Dec. 2010 June 2011 Oct. 2011 Feb. 2012 May 2012 Sep. 2014 Mar. 2010 Aug. 2010 Sep. 2010 Oct. 2010 Dec. 2010 June 2011 Oct. 2011 Feb. 2012 May 2012 Sep. 2014 Mar. 2010 Aug. 2010 Sep. 2010 Oct. 2010 Dec. 2010 June 2011 Oct. 2011 Feb. 2012 May 2012 Sep. 2014 % cover % cover % cover % cover % cover % cover % cover % cover % cover page 5 of 10Zoological Studies 56: 7 (2017) Fig. 3. Non-parametric multidimensional scaling analyses of nine categories of transect data showing three major groupings of survey data (red dotted lines = 75% Bray-Curtis similarity; red letters = relative effect of each benthic category, with abbreviations T = Terpios hoshinota, C = Ceratodictyon spongiosum, Sa = sand/carbonate, He = hermatypic corals, Ot = other, Un = unknown, Ma = macroalgae, De = dead coral, Cy = cyanobacteria). −0.5 0.0 0.5 −1.0 −0.5 0.0 0.5 NMDS1 NMDS2 2010/3 2010/8 2010/9 2010/10 2010/12 2011/6 2011/10 2012/2 2012/5 2014/9 T Ma C Cy He De Ot Sa Un page 6 of 10Zoological Studies 56: 7 (2017) branching species observed. Extent of Terpios hoshinota outbreak By snorkeling, the limits of the occurrence range of T. hoshinota were determined to be 27°20'8"N, 128°32'43"E and 27°19'55.3"N, 128°33'01.2"E (Fig. 1). The distance between the east and west edges of the outbreak was approximately 620 m, roughly centered on the area where our monitoring was performed. From the monitoring area, the frequency of T. hoshinota gradually reduced towards the east and west edges (data not shown). DISCUSSION The non-parametric multidimensional scaling results showed three clear groupings both with non-biological benthic categories included in the analyses (Fig. 3, Table S1) and without the inclusion of these categories (e.g. biological data only; data not shown). For the Yakomo site during our survey period, there was clearly dominant benthos for nine out of 10 surveys (Terpios or cyanobacteria), and the dominant benthos therefore obviously strongly shaped the community. In other words, the benthic community at Yakomo was clearly different after Terpios disappeared between June 2011 and September 2014, but then returned to a relatively similar community with the reappearance of T. hoshinota in September 2014. Understanding the causes of the reduction in T. hoshinota and the subsequent increase are therefore key to understanding the benthic community at Yakomo. Causes of decreases in T. hoshinota There was an obvious reduction of T. hoshinota some time during the period between December 6, 2010 and June 17, 2011. There are two factors that possibly caused the reduction of T. hoshinota, both related to typhoons. In the vicinity of Okinoerabu-jima Island, maximum wave height of Typhoon Songda was 10.22 m on May 28, 2011. When category 3 or 4 Typhoon Songda was positioned at 27°4'N, 128°0'E (near Okinoerabu-jima Island), the maximum wind speed of 38.58 m/s (139 km/h). The waves and wind caused by Typhoon Songda might have peeled off T. hoshinota colonies from the substrate. Waves approaching 5 or 6 m are experienced every year in the Ryukyu Islands, both in summer from typhoons and in winter from storms, but wave heights of >8 m were only seen this one time during the survey period, suggesting that these waves were abnormally large for the region. It is also possible that the decrease was due to sand deposition from Typhoon Songda. In the transect photographs of June 2011 after Typhoon Songda T. hoshinota appears white compared to as in images from other dates, and we observed sand deposited upon sponges and algae in the June 2011 survey. Although there was a time lapse of 19 days between Typhoon Songda and our survey on June 17, 2011, our observations show that the influences of this typhoon may have remained at this time. Regardless of whether the cause of T. hoshinota decreases was from wave damage or sand deposition, Typhoon Songda is the strongest candidate for observed decreases. Similar results have been suggested in a previous study at Green Island and Lanyu Island (Orchid Island) in Taiwan, where the main cause of reduction of T. hoshinota in 2009 was suggested to be Typhoon 8 (Morakot) (Chen 2011). Between 2013 until the last survey date on September 21, 2014, four additional large typhoons approached Okinoerabu-jima Island, however T. hoshinota cover had increased by the September 2014 survey. None of the typhoons approached wave size as seen in Typhoon Songda (all typhoons with wave sizes <8 m). Wind and wave strength may play a role in whether or not T. hoshinota cover decreases during typhoons, combined with the length of exposure to these typhoon conditions. In order to more conclusively demonstrate the influence of typhoons on T. hoshinota coverage there is a need to examine coverage more immediately before and after a direct hit by a very strong typhoon with >8 m waves. In summary, it appears that the close approach and large waves of Typhoon Songda in June 2011 caused the almost complete reduction of Terpios at Yakomo, although the exact mechanism (waves, burial by sand) remains to be elucidated. Causes of outbreak and re-appearance of T. hoshinota Terpios hoshinota has a high density of endosymbiotic cyanobacteria (Hirose and Murakami 2011), and in some cyanobacterial page 7 of 10Zoological Studies 56: 7 (2017) sponge species the internal interface between the host and cyanobacteria has been shown to provide both nitrogen-fixing products and photosynthetic products (Cheshire et al. 1997). Around Pagan Island in the Mariana Islands, volcanic ash due to Pagan Volcano activity was deposited on coral reefs in 2009-2010, and at the same time there was a sudden outbreak of cyanobacteria and the appearance of a large patch of T. hoshinota (Schils 2012). Subsequently, along with a cessation of the volcanic activity in 2011, outbreaks of cyanobacteria also ceased. When volcanic ash is dissolved in seawater, trace metals that affect organisms such as iron are immediately made available to primary producers (Frogner et al. 2001). From this, Schils (2012) hypothesized that on the coral reefs of Pagan Island an increase in iron concentration may have caused cyanobacteria and T. hoshinota outbreaks. In addition, Schils (2012) suggested that outbreaks may occur from not only the chemical components of volcanic ash, but also from sedimentation, via human origin such as shipwrecks, by sediments associated with natural phenomena such as forest fires and wildfires in flue dust, or via desert sandstorms. Similarly, it has been proposed that increased concentrations of iron, phosphates, and nitrites may be responsible for cyanobacterial blooms (Ng et al. 2012). Future research focusing on potential causes listed in Schils (2012) may help clarify environmental mechanism(s) behind Terpios outbreaks. Another possible reason Terpios outbreaks were not observed at any location asides from the Yakomo coast may be due to the low dispersion ability of T. hoshinota. Many sponge species occur in localized regions due to the poor swimming ability of larvae and/or a predominance of asexual reproduction (van Soest et al. 2012). Terpios hoshinota is no exception, as larvae have negative buoyancy, and it has been observed in tank experiments that cyanobacteria-containing larvae settle at very close distances to parents on dead coral substrates, suggesting the species has short larval dispersion distances in the field (Wang et al. 2012; Hsu et al. 2013; Nozawa et al. 2016). Therefore, even if environmental conditions susceptible to outbreaks in other regions are present, T. hoshinota may not be able to effectively disperse (Nozawa et al. 2016). However, typhoons may increase long-range dispersal of this species (Nozawa et al. 2016), and the subtropical Ryukyu Islands have the highest incidence of typhoons in the world. Furthermore, T. hoshinota can extend thin tissue threads to spread locally across unsuitable habitats (Soong et al. 2009). Given this and the fact that T. hoshinota is widespread in the central and southern Ryukyu Islands (Reimer et al. 2011b), we do not consider dispersal limitations to be a significant factor in causing the outbreak to be locally limited to Yakomo on Okinoerabu-jima Island. In summary, both the original cause of the T. hoshinota outbreak at Yakomo and the cause(s) of its partial recovery by September 2014 remain unclear, although it is likely due to a combination of local recruitment, fast growth rates (approximately 1-2 mm per day; Liao et al. 2011), and underlying environmental factor(s). CONCLUSIONS From previous research in Taiwan (Chen 2011) and this study, it appears that in subtropical regions typhoons may play an important role in the dynamics of T. hoshinota outbreaks. Other previous work from Tokunoshima Island in Kagoshima has shown that outbreaks may not necessarily be a ‘dead-end’ phase shift (Reimer et al. 2011b), and it can be concluded that subtropical T. hoshinota outbreaks have the potential to change rapidly. It has been theorized that T. hoshinota is expanding its range in the Indo-Pacific (van der Ent et al. 2016), including into tropical regions such as Indonesia (Madduppa et al. 2017) and the Maldives (Montano et al. 2015). Tropical regions generally do not experience typhoon events, and thus care should be taken to not extrapolate monitoring results from tropical regions to the subtropics: it may be that the dynamics of T. hoshinota are very different in the absence of large storm events. For example, outbreaks in Indonesia have been reported to persist for at least four years (Madduppa et al. 2017). Comparisons of long-term monitoring data between tropical and subtropical T. hoshinota outbreaks are clearly needed to examine this potential issue. Additionally, transect surveys at various sites should be carried out more regularly to confirm the trends of increases or decreases of T. hoshinota with more accuracy. Moreover, in order to identify the cause(s) of changes in coverage, it is necessary to further examine temporal changes of coastal environmental conditions (typhoons, water quality, seawater temperature, etc.). page 8 of 10Zoological Studies 56: 7 (2017) Acknowledgments: On Okinoerabu-jima Island, Shinichi Hiasashi is thanked for his logistical support; this study is dedicated to his memory. M.Y. thanks Prof. A. Takemura of the University of the Ryukyus (UR) and members of the MISE Laboratory, particularly I. Kawamura, Dr. J. Montenegro, and H. Kise for advice. Prof. E. Hirose, Prof. S. Suda, Dr. H. Kurihara, and Dr. J. Yasumoto (all UR) are acknowledged for helpful comments, and Dr. V. Denis (National Taiwan University), Dr. J. Parkinson (Oregon State University), and H.B. Wee (UR) are thanked for statistical advice. J.D.R. was supported by the International Research Hub Project for Climate Change and Coral Reef/Island Dynamics at UR. Comments from three anonymous reviewers greatly improved the manuscript. REFERENCES Beenaerts N, Berghe EV. 2005. Comparative study of three transect methods to assess coral cover, richness and diversity. West Ind Ocean J Mar Sci 4:29-38. Bryan P. 1973. Growth rate, toxicity, and distribution of the encrusting sponge, Terpios sp. (Hadromerida: Suberitidae) in Guam, Mariana Islands. Micronesica 9:237-242. Chen KY. 2011. The population dynamics of a coral-killing black sponge, Terpios hoshinota at Green Island and Orchid Island (Master thesis). National Sun Yat-sen University, Kaohsiung. (in Chinese) Cheshire AC, Wilkinson CR, Seddon S, Westphalen G. 1997. Bathymetric and seasonal changes in photosynthesis and respiration of the phototrophic sponge Phyllospongia lamellosa in comparison with respiration by the heterotrophic sponge Ianthella basta on Davies Reef, Great Barrier Reef. Mar Freshw Res 48:589-599. Elliott J, Patterson M, Vitry E, Summers N, Miternique C. 2016. Morphological plasticity allows coral to actively overgrow the aggressive sponge Terpios hoshinota (Mauritius, Southwestern Indian Ocean). Mar Biodiv 46:1-5. Frogner P, Gíslason SR, Óskarsson N. 2001. Fertilizing potential of volcanic ash in ocean surface water. Geology 29:487-490. Fujii T, Keshavmurthy S, Zhou W, Hirose E, Chen CA, Reimer JD. 2011. Coral-killing cyanobacteriosponge (Terpios hoshinota) on the Great Barrier Reef. Coral Reefs 30:483. Hirose E, Murakami A. 2011. Microscopic anatomy and pigment characterization of coral-encrusting black sponge with cyanobacterial symbiont, Terpios hoshinota. Zool Sci 28:199-205. Hsu C-M, Wang J-T, Chen CA. 2013. Larval release and rapid settlement of the coral-killing sponge, Terpios hoshinota, at Green Island, Taiwan. Mar Biodiv 43:259-260. Liao MH, Tang SL, Hsu CM, Wen KC, Wu H, Chen WM, Chen CA. 2007. The “black disease” of reef-building corals at Green Island, Taiwan - outbreak of a cyanobacteriosponge, Terpios hoshinota (Suberitidae; Hadromerida). Zool Stud 46:520. Liao MH, Tang SL, Hong MJ, Jane WN, Chiang PW, Chen CB, Chen CA. 2011. Bacteria associated with an encrusting sponge (Terpios hoshinota) and the corals partially covered by the sponge. Environ Microbiol 13:1179-1191. Madduppa H, Schupp PJ, Faisal MR, Sastria MY, Thoms C. 2017. Persistent outbreaks of the black disease sponge Terpios hoshinota in Indonesian coral reefs. Mar Biodiv 47:14-151. Marine Park Center Foundation. 1986. Urgent survey report of an abnormal phenomenon of marine ecosystem in the Amami Islands. Environmental Agency of Japan, Tokyo. (in Japanese) Montano S, Chou WH, Chen CA, Galli P, Reimer JD. 2015. First record of the coral-killing sponge Terpios hoshinota in the Maldives and Indian Ocean. Bull Mar Sci 91:97-98. Montenegro J, Sinniger F, Reimer JD. 2015. Unexpected diversity and new species in the sponge-Parazoanthidae association in southern Japan. Mol Phylogenet Evol 89:73-90. Ng CSL, Lin Jin Ong LJ, Chou LM. 2012. Lyngbya majuscula blooms in an enclosed marine environment. Environment Asia 5:93-98. Nozawa Y, Huang Y-S, Hirose E. 2016. Seasonality and lunar periodicity in the sexual reproduction of the coral-killing sponge, Terpios hoshinota. Coral Reefs 35:1071-1081. Plucer-Rosario G. 1987. The effect of substratum on the growth of Terpios, an encrusting sponge which kills corals. Coral Reefs 5:197-200. R Core Team. 2016. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/. Reimer JD, Mizuyama M, Nakano M, Fujii T, Hirose E. 2011a. Current status of the distribution of the coral-encrusting cyanobacteriosponge Terpios hoshinota in southern Japan. Galaxea 13:35-44. Reimer JD, Nozawa Y, Hirose E. 2011b. Domination and disappearance of the black sponge: A quarter century after the initial Terpios outbreak in southern Japan. Zool Stud 50:394. Rützler K, Muzik K. 1993. Terpios hoshinota, a new cyanobacteriosponge threatening Pacific reefs. Sci Mar 57:395-403. Schils T. 2012. Episodic eruptions of volcanic ash trigger a reversible cascade of nuisance species outbreaks in pristine coral habitats. PloS One 7:e46639. Soong K, Yang S-L, Chen CA. 2009. A novel dispersal mechanism of a coral-threatening sponge, Terpios hoshinota (Suberitidae, Porifera). Zool Stud 48:596. van der Ent E, Hoeksema BW, de Voogd NJ. 2016. Abundance and genetic variation of the coral-killing cyanobacteriosponge Terpios hoshinota in the Spermonde Archipelago, SW Sulawesi, Indonesia. J Mar Biol Assoc UK 96:453-463. van Soest RW, Boury-Esnault N, Vacelet J, Dohrmann M, Erpenbeck D, De Voogd NJ, Santodomingo N, Vanhoorne B, Kelly M, Hooper JN. 2012. Global diversity of sponges (Porifera). PLoS One 7:e35105. de Voogd NJ, Cleary DFR, Dekker F. 2013. The coral-killing sponge Terpios hoshinota invades Indonesia. Coral Reefs 32:755. Wang JT, Hirose E, Hsu CM, Chen YY, Meng PJ, Chen CA. 2012. A coral-killing sponge, Terpios hoshinota, releases larvae harboring cyanobacterial symbionts: An implication of dispersal. Zool Stud 51:314-320. page 9 of 10Zoological Studies 56: 7 (2017)