Rapid onsets of warming events trigger mass mortality of coral reef fish

Edited by Nancy Knowlton, Smithsonian Institution, Washington, DC, and approved August 17, 2020 (received for review May 17, 2020)
September 21, 2020
117 (41) 25378-25385

Significance

Our study reveals a hitherto overlooked effect of warming on coral reefs. Traditionally, ecological studies of warming events focused on maximum temperatures and duration, rather than the rate of warming at the onset. Here, we show that onsets can trigger widespread mortality of reef fish. Hundreds of thermally stressed fish, belonging to dozens of species, became fatally infected with a common pathogen in the Red Sea. Differential susceptibility led to selective mortality, with disproportional death among predators and benthic feeders. A reassessment of past reports suggests that extreme onset might have been an overlooked trigger of fish kills elsewhere. Warming tropical and subtropical reefs may face an increasing frequency of extreme onsets, eliciting calamities far beyond coral bleaching.

Abstract

Our study reveals a hitherto overlooked ecological threat of climate change. Studies of warming events in the ocean have typically focused on the events’ maximum temperature and duration as the cause of devastating disturbances in coral reefs, kelp forests, and rocky shores. In this study, however, we found that the rate of onset (Ronset), rather than the peak, was the likely trigger of mass mortality of coral reef fishes in the Red Sea. Following a steep rise in water temperature (4.2 °C in 2.5 d), thermally stressed fish belonging to dozens of species became fatally infected by Streptococcus iniae. Piscivores and benthivores were disproportionately impacted whereas zooplanktivores were spared. Mortality rates peaked 2 wk later, coinciding with a second warming event with extreme Ronset. The epizootic lasted ∼2 mo, extending beyond the warming events through the consumption of pathogen-laden carcasses by uninfected fish. The warming was widespread, with an evident decline in wind speed, barometric pressure, and latent heat flux. A reassessment of past reports suggests that steep Ronset was also the probable trigger of mass mortalities of wild fish elsewhere. If the ongoing increase in the frequency and intensity of marine heat waves is associated with a corresponding increase in the frequency of extreme Ronset, calamities inflicted on coral reefs by the warming oceans may extend far beyond coral bleaching.
Marine heat waves (MHWs) and heat spikes—a prolonged discrete event with anomalously warm water (1)—have become more frequent as the global ocean warms (2, 3). Some of those events have triggered profound changes in marine communities, among them a catastrophic, world-wide degradation of coral reefs due to bleaching (4, 5), a tropicalization of benthic communities along the coast of Western Australia (6, 7), and a devastation of the gorgonian-dominated community along the northeast coast of the Mediterranean Sea (8, 9). In most cases, widespread mortalities that preceded those changes affected nonmotile organisms, such as stony corals, gorgonians, seaweeds, mollusks, and sponges (4, 5, 79). Unlike motile animals that can escape warming events by descending to deeper water (10, 11), sedentary organisms lack the ability to move away. In the few cases where the mortality of wild fish was attributed to MHWs, either the warming conditions were not substantiated with measurements (1214), the trigger of mortality remained unclear (6, 15), or other factors, such as toxic algal bloom (13, 14) or hypoxia (16), turned out to be the cause of death. As shown below, an anomalous rate of onset (Ronset), defined as the rate of warming at the start of a warming event (1), can be a trigger for widespread mortality of reef fish. Note that our definition of Ronset refers to the maximum rate of warming observed during the course of warming, not necessarily the one ending with the event’s maximum temperature, as defined by Hobday et al. (1). Mass mortalities of cultured fish following an abrupt rise in temperature are common among farmed fish as they are enclosed in cages or ponds from which they cannot escape (12, 13, 17, 18). Aside from cases of hypoxia, a ubiquitous cause of death of farmed fish following abrupt warming has been bacterial infection (12, 13, 19, 20). Evidently, crowding and enclosure facilitate pathogen transmission while preventing escape. In accordance with this dichotomy between farmed and wild fish, the common understanding is that the effect of global warming on the latter will be gradual, emerging through long-term physiological changes, alterations in habitat structures, and modified productivity and trophic pathways (7, 2125). A review of 727 events of mass mortalities of wild animals during the past ∼75 y (26) found that over half of the documented cases of mass mortality events in wild fish followed incidents of cooling whereas events related to hot thermal stress in fish started to appear only in the 1980s. Along that line, a recent compilation of opinions, in which 33 experts listed crucial knowledge gaps in our understanding of the impacts of climate change on coral reef fishes (27), did not even consider mass mortalities as a consequence of warming. Here, we challenge those views, showing how warming events with high Ronset threaten the survival and sustainable functioning of fish communities in coral reefs.

Results and Discussion

A warming event with an unprecedented Ronset (4.2 °C rise of sea surface temperature [SST] in 2.5 d) (Fig. 1A) occurred over the coral reefs of Eilat, northern Red Sea, in early July 2017. This outstanding Ronset was the steepest recorded since daily measurements started 32 y ago (Fig. 1 C and D). A second event of warming with extreme Ronset occurred 2 wk later (14 to 16 July) when the water warmed by 3.4 °C in 2.5 d (25.1 to 28.5 °C) (Fig. 1A). Temperatures remained high (>28 °C) for 2 or 3 d after the first and second onsets, respectively. As the warming event occurred in early summer, the maximum temperature (28.6 °C) reached was by no means exceptional for that season (Fig. 1B).
Fig. 1.
Water temperature at the coral reef of Eilat. (A) Time series of temperature recorded every 10 min at 2 m depth from 1 May to 13 September 2017. Rectangles indicate the extreme Ronsets of the two successive warming events. The first Ronset (red rectangle) was steeper, with the water warming up by 4.2 °C in 2.5 d (24.4 to 28.6 °C between predawn 2 July and late afternoon on 4 July). During the second Ronset (14 to 16 July; green rectangle), the water warmed by 3.4 °C in 2.5 d (25.1 to 28.5 °C). (B) Time series of daily means of water temperature during the period June to September in the years 2007 to 2016 and 2018 (blue lines) and 2017 (black line). The red and green rectangles are as in A. Note that, despite the event’s extreme Ronsets, the peak temperatures in July and August 2017 remained well within the normal range for summers in Eilat. (C) Mean daily increase in SST over three consecutive days since 1988 at the coral reef of Eilat (32 y). Only positive (warming) values are plotted. Note that the maximum warming rate in July 2017 shown in C was lower than that shown in A because the daily measurements, recorded every morning between 8–9 AM, missed the coldest (predawn) and warmest (late afternoon) hours of the day. (D) Frequency distribution of the daily increase in SST in the 32-y-long time series shown in C. The red and green arrows indicate the first and second Ronset, respectively.
Indications of an unusual occurrence of fish death emerged 2 d after the start of the first warming event, with the more extreme Ronset, when carcasses of two parrotfish and two groupers were observed lying on the seabed on 4 July (Fig. 2). Nine additional carcasses were found 2 d later, and 10 more on the following day (Fig. 3A). Normally, findings of fish carcasses were rare in the coral reefs of Eilat (SI Appendix, Fig. S1). Following the realization that this unusual finding of many carcasses might indicate a start of widespread mortality, a citizen-science campaign was initiated, through which divers and swimmers were asked to search for carcasses, record their findings, and remove the carcasses from the sea. Mortality rates peaked during the campaign’s third week, following the second extreme Ronset (Fig. 1A), when 104 carcasses were found (Fig. 3A). The epizootic lasted 10 wk, with declining carcass sightings during the second week of August (6 wk after the warming event), and a return to normal levels in early September (Fig. 3A). In total, 427 fish carcasses belonging to 28 families (at least 42 species) were collected during the event.
Fig. 2.
Fish carcasses photographed during the summer 2017 die-off event in Eilat. Photographs are numbered as follows: 1, Rhinecanthus assasi (Arabian picassofish); 2 and 5, Variola louti (grouper); 3, Arothron hispidus (white-spotted puffer); 4, Siganus rivulatus (marbled spinefoot); 6, Cheilinus lunulatus (broomtail wrasse); 7, Scarus fuscopurpureus (purple brown parrotfish); 8, partly eaten Scarus gibbus (heavybeak parrotfish); 9, Tetrosomus gibbosus (humpback turretfish); and 10, Chaetodon fasciatus (diagonal butterflyfish). (Yellow scale bars: ∼5 cm.)
Fig. 3.
Spatiotemporal characteristics of the fish epizootic in summer 2017 in Eilat. (A) Total number of fish carcasses recorded per week along the coast of Eilat. Dates on the horizontal axis indicate the first day of the corresponding week. (B) Total number of fish carcasses recorded at 21 sites along the entire coast of Eilat during the epizootic. Symbols on the map indicate the total number of carcasses found at each site as indicated in the legends shown in the upper left corner. Sites 1 to 7 had sandy bottom with scattered reef boulders; sites 12 to 18 are continuous coral reefs; all other sites are a mixture of sandy bottom and patch reefs. The red star indicates the location of the IUI.
Carcasses were found along the entire coast of Eilat (11.4 km), extending from the sandy beaches near the northern end of the Gulf (sites 1 to 4 in Fig. 3B) through the coral reefs in the south (site 21 in Fig. 3B). The highest number was recorded within the protected reef of the nature reserve (site 14 in Fig. 3B). Our records undoubtedly underestimated the true extent of the mortality since some carcasses had been missed by observers, and others might have been devoured by predators and scavengers (Movie S1).
Movie S1.
Fish feeding on a moribund striped eel catfish. (Recorded by A. Diamant on July 25th, 2017.)
Many carcasses (41%) were found floating, about one-third lying on the bottom, and the remainders washed ashore. The maximum depth at which carcasses were found on the seabed was 25 m. While searches were most intensive in the shallows, a handful of technical and scientific divers recurrently examined the deep reefs, down to 60 m, at sites 9, 10, and 15 to 17 (Fig. 3B), but found no carcasses at those depths.
The three dominant families among the carcasses were parrotfish (Scaridae), groupers (Serranidae), and triggerfish (Balistidae) (Fig. 4A). An analysis using Chesson’s Index of Selectivity α (28), comparing the proportions of different functional groups among the carcasses to those found in the local reefs prior to the die-off event, indicated that piscivores (mostly groupers) and benthic grazers (mostly parrotfish) were disproportionately affected whereas zooplanktivores (damselfish, reef anthias)—the most abundant guild of fish on local coral reefs (29)—were rarely affected (Fig. 4). Were zooplanktivores truly spared, or were their carcasses relatively hard to find? While most fish belonging to this guild were small (<10 cm in length), it was unlikely that their paucity among the collected carcasses was due to their obscurity. Dozens of research divers regularly operated in the Interuniversity Institute (IUI) coral reef (site 17 in Fig. 3B) during the die-off period but found no carcasses of those planktivorous fish. Moreover, in an unrelated study, in which 10 groups of coral-inhabiting damselfish (Dascyllus marginatus; n = 46) were individually tagged in situ, none disappeared during the epizootic.
Fig. 4.
Taxonomic composition of the carcasses and live fish in the coral reef of Eilat. (A) Frequency of occurrence of different fish taxa (genus or family) among the carcasses (black bars; n = 392 identified fish), before the die-off event (blue bars; summer 2016; n = 2,351) and in fall 2017 after the epizootic ended (red bars; n = 2,128). (B) Chesson α index of selectivity (28) comparing the relative occurrence among the carcasses to that in the living community prior to the die-off event, calculated for five different functional groups, based on their feeding mode. The expected value under random mortality is α = 0.2 (dashed line). Pisc, piscivores, consisting mostly (76%) of Serranidae (groupers); Graze, benthic herbivores, consisting mostly (90%) of Scaridae (parrotfish) and some (∼5% each) Siganidae (rabbitfish) and Acanthuridae (surgeonfish); Invert, benthic invertebrate feeders, consisting mostly (24%) of Balistidae (triggerfish), Chaetodontidae (butterflyfish, 17%), and Plotosiidae (eel-tail catfish, 17%); Omni, omnivores, consisting mostly (15%) of Ostraciidae (boxfish, 15%); Plank, planktivores, consisting mostly (80%) of Pomacentridae (damselfish) and the scalefin anthias, P. squamipinnis.
Most of the dead fish were large adults (>30 cm in length). Many had bright red gills, indicating that they died shortly before sighting. In some cases, moribund fish exhibited disoriented movements or were lying on the seabed, still ventilating (Movies S2 and S3). Moribund and freshly dead specimens were immediately taken to the pathology laboratory for examination. Necropsies were performed on 14 fish belonging to eight different species (SI Appendix, Table S1). Clear evidence of severe infection by Streptococcus iniae was found in 13 cases. The infections were found in the blood, spleen, liver, kidney, and brain. In all 13 individuals, S. iniae was the only pathogenic bacterium isolated. The single fish lacking S. iniae was heavily infected by Vibrio sp.
Movie S2.
Moribund parrotfish on the bottom at the coral reef. (Recorded by O. Legum on July 16th, 2017.)
Movie S3.
Moribund lionfish swirling in the water above the coral reef. (Recorded by A. Diamant on July 15th, 2017.)
S. iniae is a ubiquitous pathogen of fish, found throughout the world’s warm waters (19), including the northern Red Sea (30). While nonsymptomatic fish may be covert carriers of this bacterium (20), a healthy immune system normally prevents debilitating infections from developing. Our visual, microscopic, and molecular examination of the carcasses did not reveal unusual occurrences of parasites or infections by other pathogens. The identification of S. iniae was based on cultures isolated from infected organs, followed by molecular analysis. Fish infected by S. iniae typically show acute signs of bacteremia, meningo-encephalitis, and panophthalmitis (14, 19). Indeed, behavioral observations of some moribund fish (Movie S3) exhibited signs compatible with infection of the central nervous system. S. iniae appeared to be the direct cause of death of most fish that died during the epizootic.
Measurements of dissolved oxygen at the coastal and deep monitoring stations during the epizootic indicated surface concentrations around 208 µM, well within the range of 189 to 211 µM typical for that season (SI Appendix, Fig. S2). Similarly, no harmful algal bloom, a known cause of fish mortality (13, 14), coincided with the event as the biomass of phytoplankton in the water and the benthic algae on the reef were well within their normal ranges (SI Appendix, Fig. S3). Nor was air temperature unusual for that period (SI Appendix, Fig. S4).
The onset of the two warming events coincided with a declining barometric pressure, weakening winds, a moderate increase in relative humidity, and, consequently, a remarkable decrease in latent (evaporative) heat loss and corresponding increase of the net heat flux to the sea surface (Fig. 5). Latent heat flux is the major cooling mechanism of the sea surface in the Gulf of Aqaba (31, 32). Satellite records of SST indicated that the warming event happened simultaneously along most of the Gulf of Aqaba (∼180 km), showing the occurrence of the steep Ronset over its entire northern section (Fig. 6). The decline in wind speed appeared to be the outcome of a thermal low over the Sahara Desert, as evident from the large-scale mapping of geopotential height (SI Appendix, Fig. S5) and the corresponding decline of in situ barometric pressure during each of the two occurrences of extreme Ronset (Fig. 5D). These observations support the hypothesis that the warming event was the outcome of a broader atmospheric mechanism.
Fig. 5.
Time series of key environmental conditions related to the onset of the warming event in summer 2017 in Eilat. Shown are (A) water temperature, (B) wind speed, (C) barometric pressure, (D) air temperature, (E) relative humidity, and (F) net and latent (evaporative) heat fluxes. In all panels except F, colored lines are hourly means, and black lines are daily means. Lines in F indicate daily means of net heat flux (red) and latent heat flux (blue), with positive values indicating upward fluxes (sea surface cooling). Note the reversed direction of the axis of the net flux. The two vertical dashed lines (gray) in all panels indicate the first and second Ronsets (1 and 14 July, respectively). Note that the extreme Ronset during each of the two successive warming events coincided with a declining barometric pressure, subsiding winds, reduced latent cooling, and intensified negative net heat flux (i.e., sea surface warming), as well as a marginal increase in relative humidity (E). A coincidence with increasing air temperature was more obvious during the second event (D). The declines in barometric pressure and wind speed at the onset of the two warming events coincided with two separate occurrences of a thermal low over the Sahara Desert (SI Appendix, Fig. S5).
Fig. 6.
Remote sensing SST depicting daily changes over the Gulf of Aqaba from 25 June through 10 July 2017, showing the extreme Ronset of the first warming event (1 to 4 July). The satellite-derived SST values agreed remarkably well with the daily means of direct measurements of water temperature at the IUI pier (SI Appendix, Fig. S8A). The small red star at the top of the Upper Left panel indicates the location of the IUI. Note that the maximum temperature shown here for 4 July (∼27.5 °C) is lower than the maximum seen in Fig. 1A (28.6 °C) because the latter was obtained from measurements taken every 10 min, thereby recording the true maximum of the day, usually occurring in late afternoon. The spatial characteristics of the onset of the second warming event (14 to 16 July) were similar (SI Appendix, Fig. S8A).
Of the two occurrences of extreme Ronset, the first (2 to 4 July; +4.2 °C in 2.5 d) (Fig. 1A) appeared to be the trigger of the epizootic as it corresponded with the start of fish die-off (4 July). The later, less extreme Ronset (14 to 16 July; +3.4 °C in 2.5 d) probably exacerbated the fish death, leading to the corresponding peak in the number of recorded carcasses (Fig. 3A).
The virulence of Streptococcus infection in fish depends on both the state of the individual’s immune system and the dose introduced (17, 20). Therefore, a consumption of bacteria-laden carcasses (photograph 8 in Fig. 2 and Movie S1) likely contributed to the spread of the infection (33). Correspondingly, all but one of the 13 most-affected families (Fig. 4A) were benthic feeders, potentially consuming carcasses or their remains found on the seabed. Infections via feeding and fecal–oral routes (20) appeared to be the likely reason for the prolonged epizootic, lasting several weeks after the end of the warming event. Overall, the decline in the populations of the eight hard-hit taxa (Scaridae to Scorpaenidae excluding Pomacentridae in Fig. 4A) was 4.8%, indicating that the decline of the epizootic 2 mo after it began was unlikely an outcome of a sharp decline in population size. Note that our removal of carcasses from the sea might have lowered the spread of the bacterial infection.
Why did the motile fish not escape the warming waters (3, 10, 11)? Unlike sessile or sedentary animals (4, 5, 79), fish could have sought a nearby refuge in deep waters. For example, in the coral reef off the IUI, fish residing at 10 m depth could find ∼3 °C cooler waters at 50 m depth, a short distance (∼250 m) away. Yet they remained in the warming zone. A likely reason for this behavior is that most of the fish that died during the event (Fig. 4A), were species that maintain limited home ranges (34, 35), with some being strictly territorial (36).
An extensive study on the upper temperature threshold of tropical fishes (37) concluded that global warming should not yet affect coral reef fishes directly because the upper thermal tolerance of many species greatly exceeds the range to which they are currently exposed. Challenging this conclusion, our study shows that the critical parameter at stake is not necessarily the peak temperature, but the rate of warming. The maximum temperature during the event in Eilat (28.6 °C) was not exceptional as higher temperatures with no noticeable mortality of fish occurred in over half of the 32 summers on record (SI Appendix, Fig. S6).
A rapid change in temperature is a well-documented stressor of fish, suppressing immune response and increasing susceptibility to pathogens (19, 38, 39). An experimental study on sockeye salmon during its prespawning migration in the Fraser River in Canada (40) coincidentally simulated a similar warming rate to the one we measured in Eilat. There, the fish were immersed in water that was gradually warmed at a rate of 3 °C in 2 d (to 18 °C) and acoustically monitored after their release back to the river (15 °C). That warming treatment increased the mortality of the exposed fish by ∼50%, as compared with the mortality of control individuals immersed in ambient temperature. The reported reason for the lower survival in the warm-treatment fish was a rise in parasite infection (40).
Our reexamination of two documented events of mass mortality of wild fish that had been attributed to extreme temperature anomalies (6, 12, 13) indicated that also in those cases an extreme Ronset might have been the trigger. The first event occurred in August 2001 in Kuwait Bay (12, 13), allegedly due to an extreme MHW. However, a later compilation of monthly SST in Kuwait Bay (41) indicated that water temperature in August 2001 was not unusual as warmer summers with no reports on local fish mortalities occurred in preceding years (figure 5 in ref. 41). Our analysis of satellite-derived daily SST changes in Kuwait Bay during that period indicated the occurrence of a steep Ronset several days prior to the start of the fish die-off (SI Appendix, Fig. S7A). Similar to the warming event in Eilat, the high Ronset in Kuwait Bay coincided with weak winds and calm seas, and the cause of death was infection by Streptococcus agalactiae (13). The second reexamined event was the mass mortality of fish observed during the 2011 MHW along the coast of Western Australia (6). There, as well, our remote-sensing analysis showed that the fish die-off started a few days after a sharp Ronset (SI Appendix, Fig. S7B). The occurrence of a steep onset and its relationship to the mortality event was briefly stated, describing it as a warming spike with a 2 to 3 °C rise in water temperature over a few days (6, 42). As in Eilat and Kuwait Bay, the rise in temperature coincided with weak winds and calm seas (6). No identification of pathogens was possible because the carcasses were collected after being washed up on beaches (6) and therefore already decomposing. Our assessment of these two warming events suggest that an extreme Ronset could have been an overlooked trigger of past mortalities of wild fish elsewhere prior to the 2017 event in Eilat.
A selective loss of predators and benthic feeders (Fig. 4B) is expected to change the diversity and functioning of fish communities in affected reefs, especially if recurring. Will the frequency of extreme Ronsets increase alongside the predicted increase in frequency and magnitude of MHWs in the 21st century (43, 44)? To the best of our knowledge, the relationships between occurrences of extreme Ronsets and MHWs or thermal lows have not yet been explored. However, rationally, if extreme Ronsets occur at a certain percent of such warming events, regardless of how small that percent is (e.g., Fig. 1D), the outcome of more frequent warming events may be a proportional increase in the frequency of extreme Ronset. Calamities inflicted on coral reefs by the warming oceans may extend far beyond bleaching and its consequences (45).

Materials and Methods

Environmental Conditions: The National Monitoring Program.

As detailed below, much of the background information presented here uses data recorded by Israel’s National Monitoring Program (NMP) in the Gulf of Eilat (46). NMP started to operate in 2003 and is active to date. Some activities were added in later years. The access to the NMP database is publicly open at https://iui-eilat.huji.ac.il/Research/NMPMeteoData.aspx. Following are succinct descriptions of the NMP methods related to the parameters used in this study. Detailed descriptions are found at https://iui-eilat.huji.ac.il/Research/NMPMethodsEng.aspx.

Continuous measurements.

Water temperatures, air temperature, wind speed and direction, barometric pressure, solar radiation, and relative humidity (used in Figs. 1 A and B and 5) were measured at the pier of the Interuniversity Institute for Marine Sciences of Eilat (“IUI Pier”; 29.5033° N; 34.9178° E), located 30 m offshore, over the upper reef slope, where the bottom depth is ∼4 m. These measurements were recorded at 10-min intervals and later used to calculate hourly and daily means. A real-time display of the data is available at www.meteo-tech.co.il/eilat-yam/eilat_en.asp. Measurements started in September 2006 and continue to date. Water temperature was measured with a thermistor attached to a piling at ∼2 m below the surface (at low tide). All other parameters were measured at 6 to 10 m above sea surface.

Daily measurements of SST and Chlorophyll a.

These daily measurements were made in the morning (0800 to 0900) since 1988, at a fixed point, ∼10 m seaward of the reef flat on the pier of the Coral World Underwater Observatory, located at the Coral Reef Nature Reserve, ∼350 m northeast of the IUI pier. Being measured once a day in the morning, these measurements missed the coolest (predawn) and warmest (late afternoon) times of the day. Therefore, these measurements (used in Fig. 1 C and D and SI Appendix, Fig. S6) differed from the aforementioned daily means used in Fig. 1B that were calculated based on the continuous measurements.
Surface concentrations of extracted chlorophyll a were measured daily as described in ref. 47. Briefly, two samples (330 mL each) of surface water were collected concurrently with the SST measurements at the same location, prefiltered using a 100-µm plankton net (to remove zooplankton), and transferred to the laboratory where phytoplankton was retained by filtering the water through GF/F filters. The chlorophyll was extracted in buffered 90% acetone solution for 24 h under dark and cold (4 °C) conditions, followed by measurements using a calibrated fluorometer (TD700; Turner Design).

Measurements of dissolved oxygen.

Vertical profiles of dissolved oxygen (and other parameters) from the surface to 700 m depth have been carried out as part of NMP at the permanent “Station A” (29.5° N; 34.95° E) once a month since 2003. Station A is located in the open waters, ∼3 km southeast of the Nature Reserve Coral Reef of Eilat. In addition, we used NMP’s monthly measurements of dissolved oxygen concentration at 1 m below surface at another station (“OS”), located ∼2 km offshore, directly in front of the coral reef. The dominant southward winds in the Gulf generate Ekman transport that drives surface waters westward, from the offshore region toward the reef (48). Therefore, the concentrations of dissolved commodities in the open waters, around Stations A and OS, were similar to those near the reef (49).

Quantification of benthic algae.

Our aim in analyzing the aforementioned chlorophyll concentrations and the abundance of benthic algae was to assess the possibility that the fish die-off was due to a harmful algal bloom (50, 51). Since 2008, NMP has monitored the “potential” growth rate of macroalgae in the coral reef of Eilat on caged polyvinyl chloride (PVC) plates (10 × 10 cm in size) at three sites, three replicates per site: the reef off IUI (7 to 10 m depth), and the shallow (5 m) and deep (20 m) reefs at the Coral Reef Nature Reserve. The meshed cages (52) (1-cm mesh size) effectively exclude grazers, including herbivorous fish and sea urchins. The concentration of chlorophyll a per area was used as proxy for the algal biomass. New plates were deployed once a month and retrieved 2 mo afterward. Retrieved plates were transferred to the laboratory where the algae grown on each plate was scraped off, excess water was removed by filtering the material through GF/A filters, then the scraped algae was immersed in acetone–methanol solution (1:1 ratio) for 24 h of cold (4 °C) dark extraction, followed with spectrophotometric reading of the extracted chlorophyll a.

Census of coral reef fishes.

Annual surveys of coral reef fishes have been carried out as part of NMP since 2015. Fish were counted by divers along 25 to 30 cubic-shaped transects, with a 1 × 7-m rectangular base, extending from the seabed (5 to 8 m depth) through the water column, up to the sea surface. Counted fish were identified to the species or family levels. The data were sorted based on the fish diets to one of the following five functional groups: 1) piscivores (mostly groupers); 2) grazers (mostly parrotfish, surgeonfish, and rabbitfish); 3) feeders on benthic invertebrates (mostly triggerfish, butterflyfish, and marine catfish); 4) omnivores (mostly boxfish); and 5) planktivores (damselfish and the serranid Pseudanthias squamipinnis). Here, we used the results of the census carried out in summer 2016, a year before the fish die-off, and the one carried out in September 2017, after the end of the epizootic.

Long-term monitoring of fish carcasses.

Occasional findings of floating fish carcasses in 1990 to 2003, when fish farms operated near the northern end of the Gulf of Eilat (∼1 km east of site no. 1 in Fig. 3B), prompted NMP in 2004 to start daily transects to record fish carcasses on the coral reef. The transect was visually surveyed every morning by a snorkeler, swimming at the surface along a ∼1.5-km transect over the upper reef slope (5 to 15 m bottom depth) of the Coral Reef Nature Reserve (between sites 14 and 17 in Fig. 3B). The surveys have continued to date, long after the fish farms were removed from sea in 2008. Unfortunately, the routine transects were interrupted during the fish die-off period (July to September 2017) as the person in charge of that activity joined the extensive, public search for carcasses, resuming the routine surveys only in late 2017, after the die-off event was over. Therefore, annual totals presented in SI Appendix, Fig. S1 do not include 2017.

Calculations of Heat Fluxes.

The latent and sensible heat fluxes were calculated with the Coupled Ocean–Atmosphere Response Experiment (COARE) V. 3 algorithm (53), covering low wind conditions. The long-wave (infrared) cooling was calculated based on the algorithms of Bignami et al. (54). Calculations were based on the aforementioned measurements of solar radiation, wind speed, relative humidity, water and air temperatures, and barometric pressure at the IUI pier.

Satellite Remote Sensing Observations.

The Group for High Resolution Sea Surface Temperature (GHRSST) (version 1.0) product was acquired from the NASA Earth Observing System Data and Information System (EOSDIS) Physical Oceanography Distributed Active Archive Center (PO.DAAC) at the Jet Propulsion Laboratory (Pasadena, CA) (https://podaac.jpl.nasa.gov/). The datasets have been produced on an operational basis at the Naval Oceanographic Office (NAVOCEANO) on a global 0.1° grid since April 2008. The K10 L4 SST product consists of merged observations from the Advanced Very High-Resolution Radiometer (AVHRR), the Advanced Microwave Scanning Radiometer for EOS (AMSR-E), and the Geostationary Operational Environmental Satellite (GOES) Imager. The daily, level 4 SST data (tuned to represent SST at 1-m depth) were processed over the Gulf of Aqaba and analyzed to depict the spatiotemporal distribution of SST. The final choice of the GHRSST K10 L4 SST, among the few remote-sensing products available, followed our finding of an excellent agreement between the satellite-derived SST product (average over the pixel nearest to IUI) and the in situ measurements of water temperature at 2 m depth at the IUI pier (SI Appendix, Fig. S8). A 3-y-long cross-correlation analysis between the two daily time series indicated an extremely high, highly significant correlation coefficient (Pearson r = 0.98, P < 0.0001; n = 1,080) (SI Appendix, Fig. S8B).

Recording Carcasses–Citizen Science.

Following the realization in early July 2017 that the abnormal occurrence of dying fish on the coral reef might indicate a beginning of widespread mortality, we issued a call to the public through media and local dive centers. Divers and swimmers were asked to search for carcasses, record the date, time, location, and depth of finding, remove the carcasses from the sea, and, if possible, take a photograph of the carcass to assist with taxonomic identification and size estimates. The recording of carcasses by nonexpert citizens sometimes necessitated the pooling of different taxa in groups (genus, family), each consisting of several species that could not be reliably differentiated. Therefore, our report of the total number of species that died during the event underestimated the actual number. The objective of removing the dead fish from the sea was twofold: to help reduce the spread of the epizootic through the consumption of infected fish and to prevent repetitive counts of the same carcasses. The collected carcasses were taken to the local headquarters of the Nature and Parks Authority for incineration. Moribund or freshly dead fish (those with red gills) were rushed to the National Center of Mariculture in Eilat for necropsy and further analysis. The information received on the carcasses was compiled by one of the authors (L. L.). As the number of swimmers and divers was relatively higher during the weekends, the search effort was nonuniform during the week. Therefore, the temporal dynamics of the fish mortality in Fig. 3A is presented using weekly intervals.

Necropsy.

Primary bacterial cultures were isolated in a laminar flow hood from spleen, liver, head kidney, blood, and brain of moribund and freshly dead fish. Inoculations were made on tryptic soy agar (TSA) (Difco) prepared with 25% sterile seawater. Incubation was at 24 ± 1 °C. Following subculture, DNA was extracted from a pure bacterial colony. Briefly, DNA was extracted from a needle-touched bacterial colony (in three replicates), using grinding buffer (100 mM Tris⋅HCl, pH 9, 100 mM ethylenediaminetetraacetic acid [EDTA], and 1% sodium dodecyl sulfate [SDS]). The homogenate was incubated for 30 min at 70 °C. Forty-two microliters of 8 M potassium acetate were then added, and the mixture was placed on ice for 30 min, after which it was centrifuged at 12,000 × g for 15 min at 4 °C. To eliminate pellet traces, the supernatant was transferred to a fresh tube and centrifuged again for 5 min. DNA was precipitated with 1 volume of isopropanol and left for 15 min at room temperature. Pelleted DNA was washed twice in 70% ethanol, and the air-dried pellet was dissolved in 50 µL of double-distilled H2O. DNA quantity and quality (260:280 ratios) were measured in a Nanodrop One (Thermo Scientific). PCR analysis of the 16S gene using 27f (5′-AGA​GTT​TGA​TCC​TGG​CTC​AG-3′) and 1492R (5′ TAC​GGC​TAC​CTT​GTT​ACG​ACT​T-3′) primers was performed. PCR products were purified using QIAquick PCR Purification Kit 250 (QIAGEN, Hilden, Germany). DNA quantity and quality (260:280 ratios) were estimated for the second time in a microplate spectrophotometer (PowerWave XS; BioTek). Purified PCR products from template DNA were sequenced at Hy Laboratories Ltd. (Hylabs, Rehovot, Israel).

Statistical Analysis.

Pearson correlation and regression analyses were carried out using SYSTAT (V. 13).

Data Availability

All study data are included in the article and SI Appendix.

Acknowledgments

We thank the Israel NMP in the Gulf of Eilat for providing the invaluable environmental data. We thank members of the NMP, IUI, and Nature and Park Authority in Eilat; and Orie Legum and the many divers, snorkelers, and residents of Eilat for their contribution to the campaign of recording and collecting carcasses. We thank local dive centers, editors of local and national newspapers, and social media administrators for their contribution to the call for this citizen-science campaign. We thank Roberto Ehrlich for his help with the necropsy and its analysis. We thank the Israel Meteorological Service for wind data from their station “Elat” and the GHRSST from the NASA EOSDIS PO.DAAC at the Jet Propulsion Laboratory (Pasadena, CA) for the satellite-derived SST products. Parts of the study were supported by grants from the Israel Science Foundation (1211/14) and Israel Ministry of Science and Technology (3-16729) (to A.G.). The NMP is funded by the Israel Ministry of Environmental Protection. We thank Sophie Dove, Ove Hoegh-Guldberg, Margarita Zarubin, and Daniela Genin for insightful discussions on the ecology of the warming event, and L. G. C. Genevier and G. Krokos for helping with the analysis of remote sensing and atmospheric circulation. We thank Prof. Steve Brenner (Bar Ilan University) for the calculations and interpretation of the heat flux dynamics. Comments and suggestions made by two anonymous reviewers greatly improved the manuscript.

Supporting Information

Appendix (PDF)
Movie S1.
Fish feeding on a moribund striped eel catfish. (Recorded by A. Diamant on July 25th, 2017.)
Movie S2.
Moribund parrotfish on the bottom at the coral reef. (Recorded by O. Legum on July 16th, 2017.)
Movie S3.
Moribund lionfish swirling in the water above the coral reef. (Recorded by A. Diamant on July 15th, 2017.)

References

1
A. J. Hobday et al., A hierarchical approach to defining marine heatwaves. Prog. Oceanogr. 141, 227–238 (2016).
2
E. C. J. Oliver et al., Longer and more frequent marine heatwaves over the past century. Nat. Commun. 9, 1324 (2018).
3
D. A. Smale et al., Marine heatwaves threaten global biodiversity and the provision of ecosystem services. Nat. Clim. Chang. 9, 306–312 (2019).
4
O. Hoegh-Guldberg et al., Coral reefs under rapid climate change and ocean acidification. Science 318, 1737–1742 (2007).
5
T. P. Hughes et al., Coral reefs in the anthropocene. Nature 546, 82–90 (2017).
6
A. F. Pearce, M. Feng, The rise and fall of the “marine heat wave” off Western Australia during the summer of 2010/2011. J. Mar. Syst. 111–112, 139–156 (2013).
7
T. Wernberg et al., Climate-driven regime shift of a temperate marine ecosystem. Science 353, 169–172 (2016).
8
C. Cerrano et al., A catastrophic mass-mortality episode of gorgonians and other organisms in the Ligurian Sea (North-western Mediterranean), summer 1999. Ecol. Lett. 3, 284–293 (2000).
9
J. Garrabou et al., Mass mortality in Northwestern Mediterranean rocky benthic communities: Effects of the 2003 heat wave. Glob. Change Biol. 15, 1090–1103 (2009).
10
M. L. Pinsky, A. M. Eikeset, D. J. McCauley, J. L. Payne, J. M. Sunday, Greater vulnerability to warming of marine versus terrestrial ectotherms. Nature 569, 108–111 (2019).
11
A. L. Perry, P. J. Low, J. R. Ellis, J. D. Reynolds, Climate change and distribution shifts in marine fishes. Science 308, 1912–1915 (2005).
12
A. Al-Marzouk et al., “Fish kill of mullet Liza klunzingeri in Kuwait Bay: The role of Streptococcus agalactiae and the influence of temperature” in Diseases in Asian Aquaculture, P. Walker, R. Lester, M. G. Bondad-Reantaso, Eds. (Asian Fisheries Society, Manila, 2005), Vol. vol. 5, pp. 143–153.
13
P. M. Glibert et al., A fish kill of massive proportion in Kuwait Bay, Arabian Gulf, 2001: The roles of bacterial disease, harmful algae, and eutrophication. Harmful Algae 1, 215–231 (2002).
14
H. W. Ferguson et al., Caribbean reef fish mortality associated with Streptococcus iniae. Vet. Rec. 147, 662–664 (2000).
15
N. D. Keirstead et al., Fatal septicemia caused by the zoonotic bacterium Streptococcus iniae during an outbreak in Caribbean reef fish. Vet. Pathol. 51, 1035–1041 (2014).
16
J.-P. A. Hobbs, C. A. McDonald, Increased seawater temperature and decreased dissolved oxygen triggers fish kill at the Cocos (Keeling) Islands, Indian Ocean. J. Fish Biol. 77, 1219–1229 (2010).
17
K. Yuasa, N. Kitancharoen, Y. Kataoka, F. A. Al-Murbaty, Streptococcus iniae, the causative agent of mass mortality in rabbitfish Siganus canaliculatus in Bahrain. J. Aquat. Anim. Health 11, 87–93 (1999).
18
K. H. Kim, Y. J. Hwang, S. R. Kwon, Influence of daily water temperature changes on the chemiluminescent response and mortality of cultured rockfish (Sebastes schlegeli). Aquaculture 192, 93–99 (2001).
19
C. A. Shoemaker, D.-H. Xu, E. M. Soto, “Streptococcus iniae and S. agalactiae” in Fish Viruses and Bacteria: Pathobiology and Protection, P. T. K. Woo, R. C. Cipriano, Eds. (CAB Intl, 2017), pp. 298–313.
20
E. S. Bromage, L. Owens, Infection of barramundi Lates calcarifer with Streptococcus iniae: Effects of different routes of exposure. Dis. Aquat. Organ. 52, 199–205 (2002).
21
A. D. Rijnsdorp, M. A. Peck, G. H. Engelhard, C. Mollmann, J. K. Pinnegar, Resolving the effect of climate change on fish populations. ICES J. Mar. Sci. 66, 1570–1583 (2009).
22
S. K. Wilson, N. A. J. Graham, M. S. Pratchett, G. P. Jones, N. V. C. Polunin, Multiple disturbances and the global degradation of coral reefs: Are reef fishes at risk or resilient? Glob. Change Biol. 12, 2220–2234 (2006).
23
W. W. L. Cheung et al., Large-scale redistribution of maximum fisheries catch potential in the global ocean under climate change: Climate change impacts on catch potential. Glob. Change Biol. 16, 24–35 (2010).
24
W. W. L. Cheung et al., Shrinking of fishes exacerbates impacts of global ocean changes on marine ecosystems. Nat. Clim. Chang. 3, 254–258 (2013).
25
J. A. Gittings, D. E. Raitsos, G. Krokos, I. Hoteit, Impacts of warming on phytoplankton abundance and phenology in a typical tropical marine ecosystem. Sci. Rep. 8, 2240 (2018).
26
S. B. Fey et al., Recent shifts in the occurrence, cause, and magnitude of animal mass mortality events. Proc. Natl. Acad. Sci. U.S.A. 112, 1083–1088 (2015).
27
S. K. Wilson et al., Crucial knowledge gaps in current understanding of climate change impacts on coral reef fishes. J. Exp. Biol. 213, 894–900 (2010).
28
J. Chesson, The estimation and analysis of preference and its relationship to foraging models. Ecology 64, 1297–1304 (1983).
29
J. M. Megdadi, M. A. Khalaf, F. A. Al-Horani, R. S. Manasrah, Community structure of coral reef fishes in relation to habitat and depth in the northern Gulf of Aqaba, Red Sea. Fresenius Environ. Bull. 26, 1824–1834 (2017).
30
A. Colorn, A. Diamant, A. Eldar, H. Kvitt, A. Zlotkin, Streptococcus iniae infections in Red Sea cage-cultured and wild fishes. Dis. Aquat. Organ. 49, 165–170 (2002).
31
M. Ben-Sasson, S. Brenner, N. Paldor, Estimating air–sea Heat fluxes in semienclosed basins: The case of the Gulf of Elat (Aqaba). J. Phys. Oceanogr. 39, 185–202 (2009).
32
E. Biton, H. Gildor, The general circulation of the Gulf of Aqaba (Gulf of Eilat) revisited: The interplay between the exchange flow through the Straits of Tiran and surface fluxes. J. Geophys. Res. 116, C08020 (2011).
33
P. Klesius, Transmission of Edwardsiella ictaluri from infected, dead to noninfected channel catfish. J. Aquat. Anim. Health 6, 180–182 (1994).
34
M. Cowlishaw, “Determinants of home range and territory size in coral reef fishes”, PhD dissertation, James Cook University, Townsville, QLD, Australia (2014), p. 190.
35
J. Q. Welsh, C. H. R. Goatley, D. R. Bellwood, The ontogeny of home ranges: Evidence from coral reef fishes. Proc. Biol. Sci. 280, 20132066 (2013).
36
M. Shpigel, L. Fishelson, Habitat partitioning between species of the genus Cephalopholis (Pisces, Serranidae) across the fringing reef of the Gulf of Aqaba (Red Sea). Mar. Ecol. Prog. Ser. 58, 17–22 (1989).
37
C. Mora, A. F. Ospina, Tolerance to high temperatures and potential impact of sea warming on reef fishes of Gorgona Island (tropical eastern Pacific). Mar. Biol. 139, 765–769 (2001).
38
J. Dittmar, H. Janssen, A. Kuske, J. Kurtz, J. P. Scharsack, Heat and immunity: An experimental heat wave alters immune functions in three-spined sticklebacks (Gasterosteus aculeatus). J. Anim. Ecol. 83, 744–757 (2014).
39
L. Tort, Stress and immune modulation in fish. Dev. Comp. Immunol. 35, 1366–1375 (2011).
40
G. T. Crossin et al., Exposure to high temperature influences the behaviour, physiology, and survival of sockeye salmon during spawning migration. Can. J. Zool. 86, 127–140 (2008).
41
T. B. Al-Rashidi, H. I. El-Gamily, C. L. Amos, K. A. Rakha, Sea surface temperature trends in Kuwait Bay, Arabian Gulf. Nat. Hazards 50, 73–82 (2009).
42
B. Jones, “Dead fish—simmered or chilled?” in The Marine Heat Wave Off Western Australia During the Summer of 2010/2011, A. F. Pearce et al., Eds. (Department of Fisheries, Western Australia, 2011), Research Report No. 222, p. 8.
43
E. C. J. Oliver, Mean warming not variability drives marine heatwave trends. Clim. Dyn. 53, 1653–1659 (2019).
44
E. C. J. Oliver et al., Projected marine heatwaves in the 21st century and the potential for ecological impact. Front. Mar. Sci. 6, 734 (2019).
45
T. P. Hughes et al., Global warming impairs stock-recruitment dynamics of corals. Nature 568, 387–390 (2019).
46
Y. Shaked, A. Genin, Israel national monitoring program in the northern Gulf of Eilat/Aqaba–Scientific report for the year 2017. https://iui-eilat.huji.ac.il/uploaded/NMP/reports/NMP%20Report%202017.pdf. Accessed 18 June 2018.
47
A. Genin, S. G. Monismith, M. A. Reidenbach, G. Yahel, J. R. Koseff, Intense benthic grazing of phytoplankton in a coral reef. Limnol. Oceanogr. 54, 938–951 (2009).
48
R. G. Labiosa, K. R. Arrigo, A. Genin, S. G. Monismith, G. van Dijken, The interplay between upwelling and deep convective mixing in determining the seasonal phytoplankton dynamics in the Gulf of Aqaba: Evidence from SeaWiFS and MODIS. Limnol. Oceanogr. 48, 2355–2368 (2003).
49
M. Zarubin, Y. Lindemann, A. Genin, The dispersion-confinement mechanism: Phytoplankton dynamics and the spring bloom in a deeply-mixing subtropical sea. Prog. Oceanogr. 155, 13–27 (2017).
50
J. H. Landsberg, The effects of harmful algal blooms on aquatic organisms. Rev. Fish. Sci. 10, 113–390 (2002).
51
M. L. Richlen, S. L. Morton, E. A. Jamali, A. Rajan, D. M. Anderson, The catastrophic 2008–2009 red tide in the Arabian gulf region, with observations on the identification and phylogeny of the fish-killing dinoflagellate Cochlodinium polykrikoides. Harmful Algae 9, 163–172 (2010).
52
E. Brokovich, Grazing pressure on coral reefs decreases across a wide depth gradient in the Gulf of Aqaba, Red Sea. Mar. Ecol. Prog. Ser. 399, 69–80 (2010).
53
C. W. Fairall, E. F. Bradley, J. E. Hare, A. A. Grachev, J. B. Edson, Bulk parameterization of air-sea fluxes: Updates and verification for the COARE algorithm. J. Clim. 16, 571–591 (2003).
54
F. Bignami, S. Marullo, R. Santoleri, M. E. Schiano, Long-wave radiation budget in the Mediterranean Sea. J. Geophys. Res. 100, 2501–2514 (1995).

Information & Authors

Information

Published in

Go to Proceedings of the National Academy of Sciences
Proceedings of the National Academy of Sciences
Vol. 117 | No. 41
October 13, 2020
PubMed: 32958634

Classifications

Data Availability

All study data are included in the article and SI Appendix.

Submission history

Published online: September 21, 2020
Published in issue: October 13, 2020

Keywords

  1. epizootic
  2. warming rate
  3. Streptococcus
  4. heat flux
  5. Red Sea

Acknowledgments

We thank the Israel NMP in the Gulf of Eilat for providing the invaluable environmental data. We thank members of the NMP, IUI, and Nature and Park Authority in Eilat; and Orie Legum and the many divers, snorkelers, and residents of Eilat for their contribution to the campaign of recording and collecting carcasses. We thank local dive centers, editors of local and national newspapers, and social media administrators for their contribution to the call for this citizen-science campaign. We thank Roberto Ehrlich for his help with the necropsy and its analysis. We thank the Israel Meteorological Service for wind data from their station “Elat” and the GHRSST from the NASA EOSDIS PO.DAAC at the Jet Propulsion Laboratory (Pasadena, CA) for the satellite-derived SST products. Parts of the study were supported by grants from the Israel Science Foundation (1211/14) and Israel Ministry of Science and Technology (3-16729) (to A.G.). The NMP is funded by the Israel Ministry of Environmental Protection. We thank Sophie Dove, Ove Hoegh-Guldberg, Margarita Zarubin, and Daniela Genin for insightful discussions on the ecology of the warming event, and L. G. C. Genevier and G. Krokos for helping with the analysis of remote sensing and atmospheric circulation. We thank Prof. Steve Brenner (Bar Ilan University) for the calculations and interpretation of the heat flux dynamics. Comments and suggestions made by two anonymous reviewers greatly improved the manuscript.

Notes

This article is a PNAS Direct Submission.

Authors

Affiliations

The Interuniversity Institute of Marine Sciences in Eilat, 88103 Eilat, Israel;
Department of Ecology, Evolution and Behavior, Silberman Life Science Institute, The Hebrew University of Jerusalem, 91904 Jerusalem, Israel;
Present address: Ocean and Coasts Program, Global Change Institute, University of Queensland, St. Lucia, QLD 4072, Australia.
Liraz Levy
The Interuniversity Institute of Marine Sciences in Eilat, 88103 Eilat, Israel;
Nature and Parks Authority, 88000 Eilat, Israel;
Galit Sharon
National Center of Mariculture, Israel Oceanographic and Limnological Research Institute, 88112 Eilat, Israel;
Department of Biology, National and Kapodistrian University of Athens, 15784 Athens, Greece;
Morris Kahn Marine Research Station, Leon H. Charney School of Marine Science, University of Haifa, 3780400 Sdot Yam, Israel

Notes

2
To whom correspondence may be addressed. Email: [email protected].
Author contributions: A.G., L.L., G.S., and A.D. designed research; A.G., L.L., G.S., D.E.R., and A.D. performed research; G.S. and A.D. contributed new reagents/analytic tools; A.G., L.L., D.E.R., and A.D. analyzed data; A.G., L.L., G.S., D.E.R., and A.D. wrote the paper; and D.E.R. performed the remote-sensing analyses.

Competing Interests

The authors declare no competing interest.

Metrics & Citations

Metrics

Note: The article usage is presented with a three- to four-day delay and will update daily once available. Due to ths delay, usage data will not appear immediately following publication. Citation information is sourced from Crossref Cited-by service.


Citation statements




Altmetrics

Citations

If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.

Cited by

    Loading...

    View Options

    View options

    PDF format

    Download this article as a PDF file

    DOWNLOAD PDF

    Get Access

    Login options

    Check if you have access through your login credentials or your institution to get full access on this article.

    Personal login Institutional Login

    Recommend to a librarian

    Recommend PNAS to a Librarian

    Purchase options

    Purchase this article to access the full text.

    Single Article Purchase

    Rapid onsets of warming events trigger mass mortality of coral reef fish
    Proceedings of the National Academy of Sciences
    • Vol. 117
    • No. 41
    • pp. 25183-25947

    Media

    Figures

    Tables

    Other

    Share

    Share

    Share article link

    Share on social media