Research Article

Breakdown in spawning synchrony: A silent threat to coral persistence

See allHide authors and affiliations

Science  06 Sep 2019:
Vol. 365, Issue 6457, pp. 1002-1007
DOI: 10.1126/science.aax0110

Invisible threat

Our changing climate is a threat to corals, causing disfiguring bleaching and mortality to reefs that once teemed with life. Shlesinger and Loya alert us to an equally dangerous yet nearly invisible hazard to coral: loss of breeding synchrony (see the Perspective by Fogarty and Marhaver). They found that environmental changes have resulted in shifts in the timing of gamete release in several species of broadcast-spawning corals in the Red Sea. Similar changes are likely occurring globally. Such a loss of spawning synchrony could result in reproductive failure, a much less obvious but no less insidious threat to coral reefs.

Science, this issue p. 1002; see also p. 987


The impacts of human and natural disturbances on coral reefs are typically quantified through visible damage (e.g., reduced coral coverage as a result of bleaching events), but changes in environmental conditions may also cause damage in less visible ways. Despite the current paradigm, which suggests consistent, highly synchronized spawning events, corals that reproduce by broadcast spawning are particularly vulnerable because their reproductive phenology is governed by environmental cues. Here, we quantify coral spawning intensity during four annual reproductive seasons, alongside laboratory analyses at the polyp, colony, and population levels, and we demonstrate that, compared with historical data, several species from the Red Sea have lost their reproductive synchrony. Ultimately, such a synchrony breakdown reduces the probability of successful fertilization, leading to a dearth of new recruits, which may drive aging populations to extinction.

Many of the coherent ecological impacts of climate change are reflected in shifts in animal and plant phenology (i.e., the timing of periodic life cycle events, such as reproduction, migration, flowering, etc., mediated by the environment) (14). These shifts have been shown to result in population declines, ecosystem alterations, and maladaptions (14). Studies of phenological responses to changing environments, however, have been heavily biased toward terrestrial ecosystems (36). In the marine domain, climate change–induced ecological effects, such as species range expansion, decline in sea ice extent, and mass bleaching of tropical coral reefs (68), are well studied and receive global attention, whereas phenological changes remain poorly understood. Additionally, most of the studies on phenological changes have related temporal shifts to mismatched interactions or mismatched synchrony between species or trophic levels (e.g., consumer-prey, hatchling–food source, plant-pollinator) (14). Here, we focus on mismatches of population-level reproductive-phenology of reef-building corals in the Gulf of Eilat (also referred to as the Gulf of Aqaba) in the Red Sea (fig. S1) and show that the once highly synchronized, iconic spawning events of certain corals have lost synchrony, and we examine the demographic consequences.

Coral spawning is often presented as a prominent example of synchronized phenomena in nature. Coral colonies spanning vast areas release their reproductive material simultaneously into the water column (fig. S2 and movie S1), where the gametes remain viable for only a few hours (9, 10). Successful fertilization, which only takes place within this narrow time frame and is further challenged by gamete dilution (912), has led to the evolution of precise spawning synchrony within populations. Such synchronicity relies on environmental cues operating on different scales (13, 14). For example, the exact month of spawning has been found to be correlated with temperature (1517), solar irradiance (18), and wind (19), while lunar cycles are believed to cue the exact night (2023) and sunset cues the exact hour (11, 20, 24). The widely accepted paradigm regarding coral reproduction suggests a precise within-population spawning synchrony, with a brief spawning period around specific lunar phases, which for most species occurs during the warmer months of the year. Given that global warming is intensifying mass bleaching and mortality events of coral reefs (6, 8) and that temperature strongly influences coral reproductive phenology (1517), several authors have raised concerns about the possible deleterious effects of ocean warming on coral spawning phenology and reproductive success (13, 14, 17, 25). However, the possibility of phenological mismatches affecting coral reefs has remained unexplored.

When we revisited the historical coral community reproductive phenology in the Gulf of Eilat (21, 26), we discovered that in some coral species the reproductive phenology has dramatically changed. Can spawning events, therefore, still be considered highly precise, synchronized, and stable through time? By combining intense nightly field observations with meticulous sampling and cascading analyses at the oocyte, polyp, colony, and population levels, we investigated the temporal patterns of reproductive traits at different levels over four annual reproductive seasons (27). To assess the possible demographic consequences, we established a long-term population and community dynamics study (27). We focused on five of the most abundant coral species in the Red Sea (fig. S2) that belong to four families with different life history traits and growth morphologies (27). In comparison with the historical data (21, 26), our observations indicated spawning synchrony breakdown in some of these species, and this led us to question whether high coral cover or abundance necessarily indicates healthy and functional communities.

Coral spawning phenology

The majority of reef-building corals are simultaneous hermaphrodites (i.e., individuals produce both sperm and oocytes) that reproduce once a year by broadcast spawning within one or several consecutive nights (13, 14). In some coral species, split or protracted spawning events may occur in different years or habitats (15, 23, 28). Such populations are still highly synchronized, but spawning takes place over two consecutive months and occurs during the same lunar phase. For the Gulf of Eilat, extensive studies from the early 1980s that incorporated histological studies of weekly or monthly samples, in conjunction with nightly in situ and ex situ spawning observations performed throughout May to September on the exact same reef (21, 26), showed that the major coral breeding season occurs between June and September, with many species having discrete spawning periods corresponding to different lunar phases. To determine whether the reproductive phenology of corals in the Gulf of Eilat has changed from what was described from the 1980s studies (Fig. 1A), we initiated long-term monitoring of coral spawning in 2015. We performed nightly field surveys during the major annual coral reproductive season (June to September) throughout 2015 to 2018 (total of 225 night surveys, lasting 2.5 to 5.5 hours per night) and recorded the number of spawning individuals of each species (Fig. 1B and data S1) (27). In contrast with the widely accepted paradigm of highly synchronous coral spawning, we found that, in some species, the within-population spawning synchrony had become “out of tune.”

Fig. 1 Breakdown in coral spawning synchrony in the Gulf of Eilat, Red Sea.

(A) During the 1980s, the studied coral species demonstrated distinct and concise spawning time frames (21, 26). (B) However, in recent years a marked breakdown in spawning synchrony has been evident in several species: multiple spawning events within one season were observed regardless of the lunar phase and with no consistency between years. (C) Mean daily temperature and wind speed data for our study area were acquired from the Israel National Monitoring Program (27). (D) To compare data among years, the exact dates of nightly spawning surveys were transformed to lunar age (with the new moon as lunar day 1). For the 1980s data (A), only the nights in which spawning was observed are indicated, i.e., “no spawning” was recorded for all other nights. *Spawning periodicity during the 1980s for this species was inferred by the complete disappearance of gametes in histological sections; it was not recorded in situ (26). Some data points indicating no spawning in (B) are not visible in the figure due to overlapping zero values in different years (for full details see Data S1).

During 2015 to 2018, three species displayed no consistent multiannual pattern (Fig. 1B) relative to the lunar phase, sea temperature, or wind speed (Fig. 1C). These species spawned annually, in an irregular way, along several weeks, with individual colonies spawning on different nights. For Acropora eurystoma, we observed only one spawning event in both 2015 and 2016, which in 2015 was closer to a full moon night and in 2016 was closer to a new moon. Furthermore, dissections of tissue samples collected during those years indicated that additional spawning events of this species had probably occurred (figs. S3 and S4). In both 2017 and 2018 we observed five to six spawning events of A. eurystoma, occurring at different lunar phases each season. For both Galaxea fascicularis and Platygyra lamellina, numerous spawning events occurred almost every night over periods of 1.5 to 2.5 months (Fig. 1B), whereas two other species displayed a precise spawning synchrony within a brief period, as expected (Fig. 1B). For example, Dipsastraea favus displayed a discrete and brief spawning period matching the lunar phase reported from the 1980s, while Acanthastrea echinata also had a discrete and brief spawning periodicity but with a shift along the lunar cycle compared to the periodicity reported from the 1980s.

Scale of breakdown in spawning synchrony

During the spawning periods of the four years of this study, we repeatedly sampled small fragments for laboratory dissection analyses, either from random or individually tagged colonies of A. eurystoma at different sites or times (n = 259 colonies) (data S2) (27). We quantified four parameters reflecting coral reproductive state and degree of synchrony on four different levels (Fig. 2): (i) at the cell level we quantified sizes of oocytes; (ii) at the single coral polyp level, fecundity (i.e., number of oocytes per polyp); (iii) at the colony level, colony fertility (i.e., proportion of gravid polyps within a colony); and (iv) at the population level, population fertility (i.e., proportion of gravid colonies). To examine possible local effects on reproductive traits and temporal patterns, in 2015 we performed collections at two sites in the Gulf of Eilat (fig. S1B) representing two contrasting levels of local disturbances to the reefs. One site, located in the northernmost part of the Gulf of Eilat and closer to urban areas, is subjected to stressors such as higher levels of nutrients, sedimentation, light pollution, etc. (2931). The second site, located ~7 to 8 km farther south, is a nature reserve with limited public access and fewer environmental and human perturbations (29, 30, 32). Our analyses indicated that coral populations at both sites lacked precise within-population synchrony (fig. S3). We concluded that local stressors alone cannot explain this breakdown in synchrony, and in the following years we concentrated our sampling in the southern, more protected site (fig. S1).

Fig. 2 Temporal dynamics in reproductive traits of A. eurystoma along the 2018 spawning season with multiple sporadic spawning events.

A breakdown in spawning synchrony is evident from the level of an individual polyp to the population level. While oocytes that are not spawned continue to grow (A), they may give the false impression that spawning did not occur. In contrast, other aspects such as fecundity (i.e., the number of oocytes per polyp) (B) and colony fertility (i.e., the proportion of gravid polyps within a colony) (C), as well as population fertility (i.e., the proportion of gravid colonies within the population) [which is presented as the shaded area in (C)], provide a better indication of whether and when spawning actually took place. Dashed red lines represent the dates on which spawning was observed in situ. Violin-shaped plots (the colored areas behind the boxplots) show the distribution of the data (i.e., probability density) and emphasize the changing variability in all levels, thus indicating a lack of synchrony. Letters above plots indicate significant grouping based on Tukey post hoc test after repeated-measures analysis of variance using a permutation approach (P < 0.001 for all).

The analyses of all samples between 2015 and 2018 revealed that multiple spawning events had occurred each reproductive season regardless of lunar phase and that the breakdown in spawning synchrony exists on all four levels noted above (Fig. 2 and figs. S3 to S5). To further understand the exact temporal trajectories and levels of breakdown, in 2018 we tagged 25 colonies of A. eurystoma and repeatedly sampled them (five times during the spawning period) for tissue dissection (Fig. 2). Traditional approaches, describing gametogenesis dynamics as a function of changes in gamete sizes (e.g., Fig. 2A), and simple observations scoring the presence or absence of gametes as a basis for predicting their spawning time, can easily overlook other spawning events that may have occurred. For example, in our study, from the presence alone of gametes in some samples, or their sizes (Fig. 2A), one might conclude that spawning had occurred only between the last two sampling dates (30 June and 10 July). However, this approach ignores the full ecological picture, which can be unveiled only by quantifying reproductive traits on other levels (Fig. 2, B and C). Our examination of polyp fecundity indicated a gradual decrease (Fig. 2B), suggesting that some spawning had taken place and that some polyps may have partially spawned. Additionally, the increasing variability found in fecundity along the spawning period also indicated a lack of synchrony between polyps (within colony), i.e., more polyps possessing either no or just a few oocytes, alongside other polyps that still revealed high fecundity at different times (Fig. 2B). The within-colony lack of synchrony became even clearer when we examined colony fertility (Fig. 2C), which also showed a gradual decrease. Thus, at different times, some colonies contained high proportions of gravid polyps alongside colonies without or with very few gravid polyps. Lastly, population fertility (Fig. 2C), reflecting the population-level (within-species) breakdown in synchrony, revealed a similar gradual decrease.

The overall evaluation of these different levels (Fig. 2, A to C) led us to conclude that a clear lack of synchrony was occurring at all the examined reproduction levels. In many spawning events, some colonies spawned all their reproductive products, some colonies did not spawn at all, and others spawned partially, either with only some polyps spawning or with polyps spawning incompletely. These patterns were repeatedly evident throughout the 4 years of study (Fig. 2 and figs. S3 to S5).

Connecting mismatched phenology to population trajectories

Does a breakdown in coral spawning synchrony translate into reproductive failure and gamete wastage? To answer this question, we examined coral community and population dynamics using two approaches (27). First, in 2015, we established long-term monitoring of permanent plots at a depth of 4 to 5 m (3 m2 each; n = 10). All corals within these plots were mapped, and monitoring of dynamics of coral recruitment and mortality was carried out annually by visual surveys and high-resolution photography (Fig. 3, figs. S6 and S7, and table S1). Second, in 2017, we measured the size frequency distributions of species that demonstrated a decline in synchrony (n = 589 for P. lamellina, n = 112 for G. fascicularis, and n = 59 for A. eurystoma) (Fig. 4 and table S2). Coral size can be related to processes such as elapsed time since larval settlement, in addition to many environmental factors that affect growth and survival. Healthy and stable coral populations generally display size distributions that are extremely skewed to the right (positive skewness), with a wide range of size classes represented (33). By contrast, a negatively skewed size distribution indicates a population with a lack of recruitment and the dominance of large-sized colonies, implying a population at risk of decline (33).

Fig. 3 Absent or minute recruitment of species with declining spawning synchrony, in contrast with high overall coral recruitment rates.

(A) Total number of recruits for the highest-recruiting genera and for the species that showed a decline in spawning synchrony. (B) Annual recruitment rate did not differ between the years (average 7.3 ± 4.35 newly recruited corals per m2 per year).

Fig. 4 Size frequency distributions.

Dominance of the larger size classes can be seen in all three species that exhibited a decline in spawning synchrony, together with an absence of recruits from recent years in A. eurystoma (A) and G. fascicularis (B) and minute recruitment in P. lamellina (C). These population structures suggest aging populations without sufficient juvenile replenishment. Size frequency distributions are presented as the mean (±SD) number of colonies per transect (n = 44; 10-m by 1-m belt transects). Numbers below bars indicate the total number of colonies for each group.

Over the course of this study we recorded a total of 657 newly recruited corals in the permanent plots, of which 526 (80%) were identifiable to at least the genus level (Fig. 3 and fig. S7). Despite a high annual recruitment rate of 7.3 ± 4.35 (mean ± SD) new recruits per square meter of the overall coral community (Fig. 3B), for both A. eurystoma and G. fascicularis not even one recruit (Fig. 3A) or juvenile (Fig. 4) from the previous few years could be detected. Notably, the two genera with the highest number of recruits [Stylophora and Leptastrea (Fig. 3)] include species that are not broadcast spawners but brooders (i.e., employ internal fertilization). The population of A. eurystoma had a strong negatively skewed distribution toward large (old) colonies, with a clear lack of juvenile (small) new generations. Similarly, G. fascicularis also suffered from a lack of new recruits. Finally, as external fertilization success in the sea is density dependent (912), the high abundance of P. lamellina at our study site apparently still resulted in limited reproductive success and a subsequent limited recruitment. However, since this species is also dominated by the largest-size class (>25 cm), the small proportion of recruits may prove insufficient to sustain a stable population in the longer term. Indeed, among the five focal species of this study, only D. favus and A. echinata, which demonstrated precise spawning synchrony (Fig. 1), exhibited high recruitment leading to population growth, while those species that exhibited a breakdown in synchrony and no recruitment were in decline (fig. S7).


The mass bleaching and mortality of coral reefs have become prominent manifestations of the destructive impacts of human activities and climate change on marine environments. These impacts are generally quantified using macroscopic attributes of the perceivable visible damage, such as declines in live coral coverage and abundance (6, 8, 29, 34). Nonetheless, changes in the environmental drivers that underpin the reproductive phenology and success of marine species may challenge these species’ viability in less visible ways. Here, we identify an overlooked threat to marine species that reproduce by broadcast spawning. We show that coral reproductive phenology has not merely shifted to a different time period relative to what was historically known but also has lost its pronounced synchrony, reducing the probability of successful fertilization. Ultimately, the breakdown in spawning synchrony leads to a dearth of new recruits, creating aging populations that lack vital juvenile replenishment.

Coral reefs in the Gulf of Eilat have undergone conspicuous decline for several decades, mostly due to anthropogenic disturbances (29). However, in the last decade they have shown encouraging signs of recovery, including increased live coral coverage and high recruitment rates (30, 32). Our present findings emphasize that the assessment of a coral reef’s state based solely on broad coral community attributes masks the trajectories of underlying species. Moreover, it appears that the current proportion of recruited brooding versus broadcast-spawning species (Fig. 3 and fig. S7) is higher than it was a few years earlier (32), providing further evidence that the breakdown in spawning synchrony may be altering reproductive success. Different species possess different tolerances to environmental conditions; for example, corals demonstrate species-specific responses to global warming and subsequent bleaching and mortality (25, 34). Throughout the years of this study, some species at the same area still demonstrated well-synchronized spawning, as expected (28, 35) (Fig. 1B). Therefore, the species studied may actually represent an array of coral species that are more sensitive in their phenological responses to changing environments and may be just the first to show early warning signs of disruptive phenological mismatch, with other species potentially following this trend. Given the long life span of corals, while some species may appear to be surviving under certain environmental conditions, creating an impression of thriving populations, certain vital processes for species persistence, such as sexual reproduction, might be undergoing silent compromises.

Several possible mechanisms may be driving the breakdown in spawning synchrony reported here. Because coral spawning is tightly related to lunar cycles and in most species occurs during or after sunset (13, 14), a growing concern regarding spawning phenology has to do with light pollution (22, 24). However, our results with populations from two sites (fig. S1B) with high versus low light pollution (31) indicated a clear breakdown in spawning synchrony in both, irrespective of such local effects (fig. S3). Moreover, as it appears that the precise hour of each species’ spawning was not affected but rather the month and night differed, the possible causes of variation in spawning may originate in the earlier stages of gametogenesis and gamete maturation. Temperature has a strong influence on coral gametogenic and spawning cycles (1317, 36). In our study region, sea temperatures are rising fast, at a rate of 0.31°C per decade (Fig. 5), and we therefore posit that the breakdown in spawning synchrony reported here may reflect a potential sublethal effect of ocean warming (which may result not only in higher temperatures but also in changes to the timing and pace of seasonal cooling and warming). Indeed, several studies have provided evidence that coral gametogenic cycles may be initiated asynchronously, presumably as a result of the temperature regimen (28, 36, 37). Lastly, another plausible driver that may potentially extend over larger areas, encompassing both our study sites, may be related to hormonal (endocrine-disrupting) pollutants. As in many other vertebrates and invertebrates, steroids (e.g., progesterone, testosterone, estradiol) play an important role in coral reproduction (38, 39). Anthropogenic chemicals that can disrupt the endocrine system accumulate in the marine environment as a result of ongoing fluxes of herbicides, pesticides, plastics, sewage, and many other contaminants (3840). Nonetheless, the adverse impacts of both human polluting activities and climate change on the marine environment are rapidly increasing worldwide (5, 6, 8). Therefore, regardless of the exact cause leading to declines in spawning synchrony, our findings here serve as a wake-up call to start considering these subtler challenges in coral survival, which likely also impact additional species in other regions but remain unnoticed. The major reasons for such oversights could be (i) the lack of historical baseline data against which current day findings can be compared and evaluated; (ii) the rarity of long-term or appropriate temporal resolution studies of coral phenology; and (iii) the widely accepted paradigm regarding coral reproductive phenology (i.e., discrete, brief, and fixed spawning periodicities with high within-population synchrony) that results in neglect of the possibility that it might change over time. Thus, although a recent study from the Philippines found unusual reproductive patterns (similar to those described here), in the absence of any historical data those authors concluded that the patterns might be a local characteristic of the studied species (41). Researchers in the Caribbean noted a marked multiannual inconsistency of Acropora spp. spawning periodicities (42) relative to those observed during the 1980s (37).

Fig. 5 Warming trend of sea temperature in the Gulf of Eilat.

The warming rate in the northern part of the Gulf of Eilat during the last 4 decades is 0.31°C per decade (calculated by multiplying the slope of the regression line by 365 days and then by 10 years). Mean daily sea surface temperatures from 1981 to 2018 were acquired from the NOAA OI SST V2 high-resolution dataset (27), which is based on data from NOAA Pathfinder V5.2 (AVHRR 4-km-resolution satellite data).

An important implication of our findings for future studies of marine broadcast-spawning species is that the common assumption or conclusion of well-synchronized spawning events should not be reached without long-term or adequate sampling. Identifying these early warning signs in species that show such reproductive mismatch will contribute to directing our future research and conservation efforts toward those species that are at risk of decline. Regardless of whether the within-population breakdown in reproductive synchrony reported here is a sublethal effect of global warming or a consequence of any other adverse human impact, there is a clear need for global quantification of the extent of this change and the identification of its underlying mechanisms. Without proper assessment of what constitutes reproductive success, we could easily overlook the possibility that species that appear to be abundant may actually be nearing extinction through reproductive failure. We call for further studies on the potential phenology and reproductive mismatches in marine species aimed at the early identification of populations that are at possible risk of collapse, long before they display any visual signs of impending mortality.

Supplementary Materials

Materials and Methods

Figs. S1 to S7

Tables S1 and S2

Movie S1

Data S1 and S2

References and Notes

  1. Materials and methods are available online as supplementary materials.
Acknowledgments: We thank D. Akkaynak, A. Heyward, P. Harrison, Y. Nozawa, and three anonymous referees for constructive comments on an earlier version of the manuscript. We are grateful to the Underwater Observatory Marine Park and the Interuniversity Institution for Marine Sciences (IUI) at Eilat for their ongoing support. We thank H. Rapuano, I. Kaufman, and Y. Ahouvi for their help in the lab work and N. Paz for further editorial assistance. Funding: This study was funded by Israel Science Foundation (ISF) grant no. 1191/16 to Y.L. and doctoral grants from the Israel Taxonomy Initiative, Rieger Foundation, PADI Foundation, and IUI to T.S. Author contributions: The study was conceived and designed by T.S. and Y.L. The field and laboratory work were conducted by T.S., and both authors analyzed and interpreted the data. T.S. wrote the first manuscript draft, and both authors contributed to writing the final version. Competing interests: The authors declare no competing interests. Data and materials availability: All data are available in the main text or the supplementary materials. Sea surface temperature data for 1981 to 2018 provided by the NOAA/OAR/ESRL PSD are available at Wind speed and sea temperature data for 2015 to 2018 provided by the Israel National Monitoring Program are available at
View Abstract

Stay Connected to Science

Navigate This Article