Shifts in spawning patterns in three species of reef-building corals in the Mexican Caribbean

Most Caribbean reef-building corals reproduce by broadcast spawning numerous egg-and-sperm bundles simultaneously into the water column prior to external fertilization. This is a pivotal bottleneck in the lifecycle of broadcast-spawning organisms because it relies on the synchronization of gamete release to precise species-specific nights and times. However, changes in spawning patterns have been documented in a wide variety of species from different oceans. In recent years we have also detected variations in spawning dates of three reef-building species relative to historical windows, which led us to analyze the Mexican Caribbean spawning database. Our aim was to assess whether a temporal shift in spawning phenology has occurred in Acropora palmata, Diploria labyrinthiformis and Orbicella faveolata and to determine if these reproductive patterns vary along a coastal development gradient. We detected statistically significant shifts in the spawning pattern and loss of synchrony in A. palmata but not in D. labyrinthiformis or O. faveolata over the last 19 years. Acropora palmata colonies located in reefs near urban areas lost spawning synchrony over time, whereas corals located in reefs near low coastal development, such as Punta Allen, based on three spawning nights over seven years, maintained their spawning window. The consequences of the loss in synchrony and shifts in coral spawning patterns in A. palmata are of major concern, particularly because, in combination with the loss in coral cover, they reduce the probability that natural fertilization takes place due to the Allee Effect. Therefore, human-assisted fertilization may be a valuable tool needed to help restore this reef-building species and the coral reef ecosystem.

Most Caribbean reef-building coral species reproduce through broadcast spawning events restricted to one or two nights a year, releasing numerous eggs and sperm simultaneously into the water column for external fertilization (Szmant, 1986) that relies on the precise synchronization of gamete release by conspecifics. External fertilization is a pivotal bottleneck in the lifecycle of broadcast-spawning organisms (Bishop, 1998). Behaviors such as coordinated gamete release and spawning aggregations are widely accepted as evolutionary strategies to maximize fertilization success in the oceans (Babcock et al., 1986). Synchronization offers two major advantages: it increases the probability of gametes encountering one another and it generates a “predator satiation” effect, which reduces mortality rates of gametes, embryos and larvae, to ensure that a significant proportion survives (Babcock et al., 1986; Harrison et al., 1984; Mendes and Woodley, 2002; Lin and Nozawa, 2017; Lin et al., 2021).

Coral spawning synchronization is regulated on four temporal scales: month, day, hour, and minute (Fogarty and Marhaver, 2019). The month-scale regulation relies on the increase in water temperature and daylength during spring as cues for initiating gametogenesis (Szmant, 1986; Lai et al., 2022). To synchronize spawning to an exact date and time, corals are thought to track lunar cycles, the onset of darkness via circadian clock mechanisms and light-sensing proteins (Brady et al., 2009; Sweeney et al., 2011). Experiments have demonstrated that corals may respond to the presence or absence of moonlight at a certain time of the day to trigger spawning (Lin et al., 2021). Other cues include seasonal fluctuation in water temperature, wind speed and tidal movements (Monfared et al., 2023; Sheridan et al., 2025).

However, changes in spawning patterns have been documented in various regions around the world. In an analysis of the historical coral reproductive phenology in the Red Sea, Shlesinger and Loya (2019) reported that spawning patterns in five coral species had shifted dramatically. Warmer seawater temperatures have been shown to advance the spawning window in Acropora in reefs of Australia and Japan (Sakai et al., 2020). Using over 2,000 coral spawning observations from 156 species across 19 ecoregions around the world (2000-2019), Davies et al. (2023) found that broadcast spawning corals liberated gametes closer to the full moon in recent years attributing it to chronic light pollution. Alteration in spawning timing has also been reported in the Hawaiian mushroom coral Lobactis scutaria when exposed to warmer temperatures and filtered ultraviolet radiation (Bouwmeester et al., 2023). Shifts in spawning timing closer to the full moon, have also been described for Acropora species in the Red Sea (Osman et al., 2024) and Maldives (Sheridan et al., 2025). Osman et al. (2024) showed that earlier spawning timing recorded in Acropora species in the Red Sea, relative to the full moon, was correlated with higher sea surface temperatures and warming rates during the six weeks prior to spawning whereas Sheridan et al. (2025) reported that spawning events of Acropora closer to the nights of the full moon were significantly correlated with lower tidal depth in reefs of the Maldives.

Since 2007, spawning data have been collected for reef-building species as part of a study on the biology and ecology of coral reproduction that has led to the development of a restoration program based on the production and outplanting of sexual recruits of reef-building species in the Mexican Caribbean (Banaszak et al., 2025). Caribbean reef-building species spawn from July to September during species-specific windows related to the full moon (Banaszak et al., 2025). Over 19 years, spawning has been monitored in hundreds of colonies in the Puerto Morelos Reef National Park located in the northern part of the Mexican Caribbean. In recent years we have detected variations in species spawning dates, which led us to analyze the data in the Mexican Caribbean spawning database.

Seawater temperature has been increasing in the region, and in 2023–2024 corals were exposed to an extensive and long-term heat wave, which resulted in mass coral bleaching and high mortality rates (Birkart and Álvarez-Filip, 2025). Moreover, poor water quality, due to nutrient pollution (Velázquez-Ochoa and Enríquez, 2023) has become an increasingly important stressor in this region. Elevated nutrient pollution caused by inadequate sewage treatment threatens coral reefs in the Mexican Caribbean due to tourism driven coastal development (Banaszak, 2021). The Riviera Maya corridor, which spans from Cancun to Tulum in the Mexican Caribbean, attracts 20.2 million tourists annually (SEDETUR, 2026). To accommodate this influx, the region has experienced an exponential growth in hotel rooms and associated infrastructure for numerous support staff over the past two decades, placing a huge strain on nearby coastal ecosystems. As an example of the growth of the tourism support network, Cancun (Benito Juárez Municipality) had a population of approximately 120,000 inhabitants in the late 1980s. By 2015, the population grew to 740,000 and by 2020 reached 911,500 (INEGI, 2023). Playa del Carmen (located 68 km south of Cancun) experienced an annual population growth rate of 6.3% in 2020, nearly five times the national average of 1.2% (INEGI, 2023). This large number of tourists, the permeable karst nature of the Yucatan peninsula, and the inconsistent quality of wastewater treatment result in discharges of various pollutants, including medications such as naproxen, ibuprofen, illicit drugs (Metcalfe et al., 2011) as well as endocrine-disrupting chemicals (EDCs) from contraceptive methods that could reach the ocean (Sumpter, 2005). Local coastal development is not the only factor affecting water quality in the region. The ever-increasing annual influx of the floating seaweed, Sargassum, which began in 2014 (Rodríguez-Martínez et al., 2016) has resulted in mortality of seagrass meadows (van Tussenbroek et al., 2017) and marine fauna (Rodríguez-Martínez et al., 2019) alongside accelerated coastal erosion (Chávez et al., 2020). The decomposition of beach-stranded Sargassum biomass releases particulate organic matter, triggering “brown-tides” along the entire coastline. These events compromise water quality within the associated reef lagoon by inducing hypoxia and attenuating light penetration due to the increased turbidity. These deleterious effects persist long after the macroalgae have been removed from the system (van Tussenbroek et al., 2017). The turbidity and the concomitant lower light penetration could affect the perception of environmental cues by corals. In addition, contaminants such as estrogen could alter gamete production and reproductive phenology. Therefore, our aim was to determine whether a temporal shift in spawning synchrony has occurred in three reef-building coral species (Acropora palmata, Diploria labyrinthiformis and Orbicella faveolata) in the Mexican Caribbean and whether it is related to coastal development in the region.

Study sites

The reef sites where spawning was monitored are part of an extended fringing reef (Jordán-Dahlgren and Rodríguez-Martínez, 2003) located within the Mexican section of the Mesoamerican Reef System (Figure 1A). A total of 19 sites were monitored. Eight sites are located within the Puerto Morelos Reef National Park (Figure 1B), one is in Punta Venado (Figure 1C), eight are located near Akumal in the Mexican Caribbean Biosphere Reserve (Figure 1C) and two sites are found close to Punta Allen in the Sian Ka´an Reef National Park (Figure 1D).

The reefs in the Puerto Morelos Reef National Park run parallel to the coastline from 500 m to 1.5 km offshore forming a shallow (3–8 m deep) lagoon. The reef crest is dominated by Acropora palmata, whereas the lagoon and back reef are dominated by Orbicella faveolata. The Punta Venado site has a stand of Acropora palmata. The reefs in the Mexican Caribbean Biosphere Reserve and the Sian Ka´an Reef National Park are mostly extended fringing reefs with similar lagoon characteristics to that of the Puerto Morelos Reef National Park (Jordán-Dahlgren and Rodríguez-Martínez, 2003).

Spawning monitoring

Annually, before summer, the Mexican Caribbean Coral Spawning Database is used to build spawning prediction calendars (Banaszak et al., 2025). The database contains 908 records collected between 2007 and 2025 from reef sites distributed along the Mexican portion of the Mesoamerican Reef System. Spawning dates are predicted relative to the number of days after the full moon and vary for each species. Based on the prediction calendar, field work is conducted starting one day prior to the expected start of spawning and continues until most or all colonies spawn. If no spawning is detected, monitoring continues till the end of the expected spawning window. The number of colonies and area monitored are logged along with the number of colonies that did or did not spawn, start and end of monitoring, start and end of setting and spawning as well as weather conditions. Field effort (defined here as the number of field trips per species per spawning event) fluctuates annually based on the number of nights that monitoring is undertaken per year. As spawning windows have been less predictable in recent years, field monitoring continues until spawning is confirmed for each species. It is halted once the historical spawning window has passed.

Analysis of spawning patterns

We analyzed coral spawning of three reef-building species (A. palmata, D. labyrinthiformis and O. faveolata) relative to the full moon. For the purposes of this study, full moon night (Night 0) is defined as the date when the moonrise occurs after sunset, the moon passes the meridian, and the illumination percentage reported by Time and Date A.S (2026) exceeds 99%. The data were collected between 2007 and 2025 for A. palmata and O. faveolata and from 2017 to 2025 for D. labyrinthiformis. Based on a prior analysis of the earliest and the latest dates of spawning events registered for these three species in our database (Banaszak et al., 2025), A. palmata can spawn between July and September; whereas spawning of D. labyrinthiformis occurs in July and August, and in August and September for O. faveolata.

We analyzed the coral spawning patterns at two distinct spatial scales along a coastal development gradient: A) A. palmata spawning events along 150 km of the Mexican Caribbean and B) A. palmata, O. faveolata and D. labyrinthiformis spawning events within the 21 km long Puerto Morelos Reef National Park (Table 1). Thermal-style plots (R Core Team, 2025) were used to visualize spawning patterns relative to the nights after full moon (NAFM).

To investigate variation in coral spawning timing relative to the lunar cycle, all multivariate analysis of variance (PERMANOVA) were performed using PAST 4.17 software (Hammer et al., 2001), based on Jaccard distance matrices, which is designed specifically for dichotomous variables (Anderson et al., 2011), with 9999 permutations. For all PERMANOVA analysis, the variable was presence or absence of spawning per night after full moon, where each row represents a single spawning event per night after full moon per year.

A spatial-temporal two-way PERMANOVA was performed to determine if spawning timing shifted in A. palmata along the three Mexican Caribbean sites (Puerto Morelos, Punta Venado and Akumal), over time (2007 to 2025), or due to their interaction (when the effect of one factor depends on the level of the other factor). Given the limited data from Punta Allen (3 spawning events), that site was excluded from the two-way PERMANOVA.

Following this, separate one-way PERMANOVAs were executed for each of the three species independently to evaluate specific interannual fluctuations at the Puerto Morelos Reef National Park scale. For these analyses, each row in each matrix was designated with the presence or absence of a single spawning event per night after full moon per year. Spawning data were available for A. palmata (2007-2025), O. faveolata (2007-2025), and for D. labyrinthiformis (2017 to 2025).

Finally, a global two-way PERMANOVA was conducted to assess whether spawning patterns, at the Puerto Morelos Reef National Park spatial scale, varied between the three species and across years in which spawning was monitored, as well as to test potential interaction effects between these two factors.

For each PERMANOVA test that revealed significant factors, a Test for Multivariate Dispersion (PERMDISP) was performed using Jaccard distance matrices in PAST (version 4.17). The primary purpose of these analyses was to determine whether any significant differences detected by PERMANOVA were driven by true shifts (site or year effects) or by variation in within-group dispersion (dispersion effects that we interpret as a loss of spawning synchrony), thereby ensuring a robust interpretation of the multivariate data (Anderson et al., 2008).

Analysis of changes in coastal development

The area of urban expansion closest to each of the 19 reef sites was calculated using data from the Global Human Settlement Layer (GHSL) processed in the Google Earth Engine (Pesaresi and Politis, 2023; Pesaresi et al., 2024). A 5-km diameter zone of influence (buffer zone) was used for each site. The imagery covered the years 1975, 1990, 2000, 2015, 2020, and 2025. Pixels representing built-up areas were classified as urban and the “pixelArea” and “reduceRegions” tools were used to calculate the area. For each site and year, the urban area (km 2) and the percentage of urbanized surface relative to the total buffer area were estimated. This information was then exported for further analysis. To complete the annual time series from 2007 to 2025, values were estimated via linear interpolation between years for which GHSL data were available.

Analysis using generalized additive models

To evaluate the influence of years (time), the lunar cycle, and the urban area expansion on Acropora palmata spawning events at the northern Mexican Caribbean spatial scale, Generalized Additive Models (GAMs) were used (R Core Team, 2025; Wood, 2011, 2017). Explanatory variables included the observation year (2007-2025), nights after full moon (NAFM) and the annual urban area (km 2) in a 2.5 km radius from each monitored site. The GAM was fitted using a negative binomial distribution and a log link function. The model included smoothing effects for NAFM and Urban expansion, while Years and Site were incorporated as random effects:

The smoothing terms were estimated using penalized splines, with Restricted Maximum Likelihood (REML) as the smoothing parameter selection criterion (Marra and Wood, 2011). Model fit was evaluated using a simulation-based diagnostic via the DHARMa package, checking for residual uniformity, dispersion, and outliers (Hartig et al., 2026). Model interpretation was based on the significance of the smoothing terms, effective degrees of freedom (edf) and explained deviance. The probability, p< 0.05, was considered statistically significant.

Sampling effort

Spawning sampling effort varied over time (Figure 2), which stems from multiple factors including weather conditions that prevent monitoring at sea, high colony mortality following the 2018 Stony Coral Tissue Loss Disease (SCLTD) event and the heat waves during 2023 and 2024. The prominent peaks in 2018 and 2023–2024 reflect the extended field work required to detect spawning following these major coral mortality events. In 2020, the strict COVID lockdown restrictions prevented field surveys in Puerto Morelos during the A. palmata spawning-window. However, sampling was permitted at other locations (e.g., Punta Venado). During the spawning windows for Diploria labyrinthiformis and Orbicella faveolata in Puerto Morelos monitoring was permitted under a special license.

Northern Mexican Caribbean spatial scale

From 2007 to 2012, Acropora palmata colonies were characterized by narrow spawning windows spanning two to three nights (Figure 3A). However, this pattern began to change starting in 2013, progressively shifting toward nights closer to and further away from the full moon such that by 2025 spawning has been logged from full moon night to eight nights after full moon. In 2023 and 2025, spawning events in Limones (site 1, Figure 1B) were registered on the night of the full moon (Figure 3B, Limones). At Bocana (site 7, Figure 1B), A. palmata historically spawned four to six nights after the full moon (Figure 3B, Bocana), but more recently has spawned from one to six nights after full moon. In Punta Venado (Figure 3B, Punta Venado-site 9) the spawning window has advanced by one day and in Punta Allen-sites 18 and 19 (Figure 3B, Punta Allen) it does not appear to have changed. However, in Punta Allen, there are only three spawning observations, two in 2018 and one in 2025. There are also limited data for Akumal, because fieldwork could not be conducted there in all years (Figure 3B, Akumal), however spawning does appear to occur closer to the full moon in 2022 and 2023 than it did in 2017.

The two-way PERMANOVA across the three Caribbean sites (Puerto Morelos, Punta Venado, and Akumal) for Acropora palmata spawning events over multiple years (2007-2025) revealed no significant interaction effect between years and sites on A. palmata spawning patterns (Pseudo-F (36, 53) = 0.232, p (perm) = 0.4516). By contrast, highly significant main effects were detected for both factors independently (Table 2). Spawning timing differed significantly among the monitored years (Pseudo-F (18, 53) = 1.052, p (perm) = 0.0018) and exhibited significant variation across sites along the Mexican Caribbean (Pseudo-F (2, 53) = 2.44, p (perm) = 0.0003, Table 2). The subsequent test for homogeneity of multivariate dispersion (PERMDISP) also detected a highly significant difference in dispersion among years (F (18, 91) = 4.597, p (perm) = 0.0001; and among sites in the Mexican Caribbean (F (2, 91) = 6.96, p (perm) = 0.0016).

These widespread dispersion effects indicate that the significant differences previously identified by the PERMANOVA are influenced, at least in part, by variation in within-group variances, such as shifting levels of predictability or synchronization in A. palmata spawning events rather than solely reflecting directional shifts in behavioral centroids. PERMDISP pairwise comparisons revealed significant differences in dispersion between Puerto Morelos and Akumal (p = 0.029), Puerto Morelos and Punta Venado (p = 0.0004). In contrast, no significant differences were detected between Akumal and Punta Venado (p = 0.1941).

Coastal development within the areas of influence for each of the 19 spawning sites between 1975 and 2025 has increased for all study sites except for Nichehabin and Pajaritos (Figures 4A, B). The Generalized Additive Model (GAM) did not detect a significant association between urban area and the nights after full moon when spawning events occurred (edf = 1.00, χ 2 = 0.00, p = 0.996). Similarly, days after the full moon (NAFM; edf = 1.00, χ 2 = 0.112, p = 0.738) and year (edf = 1.00, χ 2 = 0.174, p = 0.677) showed no significant effects. The random effect of site was also not significant (edf ≈ 0, p = 0.799). Overall, the model explained only 4.71% of the deviance, indicating that the included variables explained a limited proportion of the observed variation in spawning event occurrence.

Puerto Morelos reef national park spatial scale

Analysis of the three species at the Puerto Morelos spatial scale shows a clear broadening of spawning events in Acropora palmata (Figure 5A). The nights when no spawning was recorded, prior to the expected spawning window, are confirmed to be zero. These are nights when fieldwork is conducted even though spawning is not expected but is dedicated to training new divers, tagging colonies, and other logistical preparations in the field, while also monitoring spawning activity. In 2017 and 2018, Diploria labyrinthiformis exhibited a two-night spawning window on the 10 th and 11 th night after full moon. However, from 2020 onwards, spawning events broadened to four nights with spawning logged on nights 9 and 12 after full moon, indicating a loosening of the species´ precise synchronization (Figure 5B). Orbicella faveolata colonies historically spawned on nights five through seven after the full moon. However, this pattern started to shift, such that by 2024 spawning was also extended, starting earlier on night four, with spawning events documented up to nights eight and nine after the full moon (Figure 5C).

To isolate the effect of “year” on individual taxa, one-way PERMANOVAs were performed for each coral species independently at the Puerto Morelos Reef National Park spatial scale (Table 3). The analysis revealed that spawning patterns relative to the full moon night have not changed significantly across years for Diploria labyrinthiformis (One-way PERMANOVA, Pseudo-F (7,38) = 1.39, p (perm) = 0.1629) and Orbicella faveolata (Pseudo-F (18,49) = 1.08, p (perm) = 0.3163) as can be seen in Figure 5. By contrast a significant interannual variation in spawning was detected for Acropora palmata (Pseudo-F (16,57) = 1.40, p (perm) = 0.0171; Table 3), indicating that its spawning window relative to the lunar cycle has changed significantly over time, as suggested by the trend shown in Figure 5A.

The subsequent test for equal multivariate dispersion (PERMDISP) demonstrated a significant difference in multivariate dispersion among year groups (F (16,57) = 5.78, p (perm) = 0.0007). This indicates that while the overall distribution of spawning centroids remains homogenous across Puerto Morelos sites, the temporal predictability and synchronization of spawning events undergo significant changes from year to year due to strong within-group variance effects.

Statistical analysis confirmed that spawning patterns relative to the full moon varied consistently across “species” and “years” as factors (Table 4), and the interaction between them was not significant (Two-way PERMANOVA, Pseudo-F (36,131) = 0.47, p (perm) = 0.9821). As expected, a strong and highly significant effect was observed for “species” (Pseudo-F (2,131) = 16.92, p (perm) = 0.0001), which accounted for approximately 15.9% of the total variance. Additionally, the “year” factor also showed a highly significant effect on spawning timing relative to the full moon (Pseudo-F (18,131) = 1.71, p (perm) = 0.0001).

Concurrently, the test for equal multivariate dispersion (PERMDISP) demonstrated a significant difference in multivariate dispersion among years (F (18,169) = 3.42, p (perm) = 0.0001) and among species (F (2,185) = 17.94, p (perm) = 0.0001). These indicate that the variation in within-group dispersion contributes significantly to the observed multivariate patterns, rather than the difference being driven purely by shifts in group centroids (Anderson et al., 2008). Pairwise comparisons using permutations indicated significant differences in dispersion between A. palmata and D. labyrinthiformis (p = 0.026) and between A. palmata and O. faveolata (p = 0.012), whereas no significant differences were observed between D. labyrinthiformis and O. faveolata (p = 0.360).

Overall, our findings show that there has been a distinct shift in the spawning phenology of three Caribbean reef-building corals. Recent spawning events associated with the three species span more consecutive nights when compared to the historical spawning window at the same sites, alongside a shift in temporal patterns, suggest that synchronization has become less precise.

When the temporal trends are isolated by individual species, we found that the variation in spawning windows is species specific. There is a much less evident disruption in the spawning synchrony of Diploria labyrinthiformis and Orbicella faveolata when compared to Acropora palmata. Diploria labyrinthiformis and O. faveolata have wider reproductive windows compared to their respective historical windows indicating a loss of the tight intraspecific synchronization in both. However, a drift towards nights closer to full moon is not evident. It is important to note that for the Diploria data, the timeframe is more recent (starting in 2017), therefore we lack the longer-term baseline that we have for the other two species. While D. labyrinthiformis and O. faveolata expanded their windows by one or two days over time, A. palmata exhibited a significant interannual shift and can spawn within a nine-day window (0 to 8 nights after full moon) versus the historical three-day window (4 to 6 nights after full moon). The differences that have been observed between species could be related to species phylogeny, life history and gamete biology. When comparing both years and species using the two-way PERMANOVA, both factors were highly significant and more importantly the interaction was not significant. On the other hand, the highly significant PERMDISP results for both year and species exhibit higher variability and lower predictability in its spawning patterns relative to historical records, suggesting a potential disruption in reproductive synchrony by some factor (e.g., environmental, genetic, etc.). Although such dispersion could be interpreted as a result of increased field effort, we interpret such high dispersion as a loss of synchrony rather than due to increased sampling effort. An increased sampling effort does not necessarily equate to more spawning events being detected. Corals tend to spawn on one to a maximum of three consecutive nights. The results show that spawning activity has extended over more nights, though this is not likely to be an artifact of increased field effort. For instance, in 2018 (Figure 3), field effort was significantly higher following heavy coral mortality, yet A. palmata spawning remained concentrated mostly on the 5th night after the full moon within a four-night window. If increased sampling effort alone accounted for detecting wider spawning activity windows, we would expect window breadth to scale proportionally with field effort, which is not the case.

The spawning window of Acropora palmata has not only changed significantly over time, but it also demonstrated the highest dispersion in spawning dates relative to full moon. This shallow-water dwelling species has historically dominated the reef crest in the Caribbean, with spawning in Limones, Bocana, and Akumal sites, occurring as early as the night of the full moon in 2023 and 2025, at four to five nights prior to its historical spawning window. This species also exhibited marked delays, spawning on nights seven and eight after the full moon in Punta Venado, relative to patterns registered prior to 2014. The broadening of the spawning window in A. palmata is further supported by our spatial-temporal analysis along the Mexican Caribbean sites, which confirmed that A. palmata spawning timing varies independently across both monitored years (19) and sites (4). The absence of a significant interaction between species and year in the global model suggests fluctuations in spawning patterns. Rather than a coordinated regional shift, the spawning behavior of A. palmata appears to be undergoing localized desynchronization over time.

Although we have only three data points for spawning in Punta Allen, A. palmata has maintained its historical spawning pattern. At Pajaritos (site 19) near Punta Allen, A. palmata spawned on night six after the full moon in 2018 whereas in 2025, this species was recorded spawning on nights six and seven at the nearby Nichehabin (site 18). Due to the lack of continuous data for this site, it was not included in the two-way PERMANOVA test at the Caribbean scale. While more data are needed to confirm that the A. palmata colonies at sites 18 and 19 maintain their spawning window, it is important to note that coastal development has been virtually nil in Punta Allen, which is the closest area of influence to these two sites (Figure 2).

Environmental pressures may be fracturing the historical predictability and synchronization of coral spawning in the Mexican Caribbean. In fact, A. palmata colonies in Cozumel have exhibited an almost complete lack of spawning from 2018 to 2025. These colonies form a shallow (< 2.5 m depth) reef crest located across from a frequented recreational zone (Calle-Triviño et al., 2025). Taken together the data indicate a loss of precision and a low predictability of spawning by this species. The loss of precision could lead to a decreased probability of successful natural fertilization, whereas low predictability of spawning events increases field work related costs associated with research on coral reproductive ecology and restoration efforts that use sexual recruits. Shifting and variable spawning windows could have severe implications for fertilization success and larval supply, which could help to explain why natural A. palmata recruits are so rare in the Caribbean (Williams et al., 2008).

Coastal development in the Mexican Caribbean has generated important impacts that potentially could be related to the spawning shifts. We considered a 5-km diameter zone of influence (buffer zone) for each site as an adequate parameter to test its relationship with shifts in coral spawning windows. However, the GAM only explains 4.7% and dispersion is high. Regarding the non-significance of the model, it is important to note that the GAM is heavily influenced by the zero-inflated nature of the spawning dataset, because non-spawning nights vastly outnumber positive spawning observations, thus standard conditional variance structures struggle to isolate the signal due to the abundance of zeros. This makes it particularly challenging to disentangle the relationship between a continuously increasing linear variable over time and changing spawning patterns relative to nights after full moon.

Synchronous spawning is likely regulated by successive environmental cues across multiple timescales (Oliver et al., 1988). There are many factors that could be impacting coral spawning patterns either individually or synergistically, that are associated with global climate change (e.g., thermal stress due to increased sea water temperature), coastal development (e.g., light contamination, inadequate wastewater treatment and disposal practices causing eutrophication of the water column), or other stressors (e.g., brown tides due to Sargassum decomposition).

In the spring, water temperature and daylength increase are responsible for initiating gametogenesis (Lai et al., 2022). However, above-normal seawater temperature and wind speed within one to two months prior to the full moon of the spawning month have been shown to strongly correlate with spawning day deviations in the genus Acropora (Sakai et al., 2020) and in reducing coral spawning (Levitan et al., 2014). Sea temperature has increased in the Puerto Morelos Reef National Park at sites such as Bocana (SAMMO, 2026). Following the fourth heatwave (2023-2024), the number of spawning events in all three coral species was low compared to historical events and almost no spawning was seen in A. palmata after the 2023–2024 heatwave, presumably due to metabolic energy being invested in recovery from bleaching rather than using it in gametogenesis (Ward et al., 2000). In fact, A. palmata was recorded spawning only in Punta Venado on nights five and six after the full moon in 2024 (Figure 3B) and at no other sites in the Mexican Caribbean. However, increased seawater temperatures do not explain the shifts in spawning times (Davies et al., 2023). Thermal stress affects gamete maturation and increases the percentage of developmental abnormalities (Randall and Szmant, 2009) suggesting that this factor could be more related to spawning presence or absence than to shifts in spawning windows.

For years, the prevailing view has been that multiple environmental triggers work in tandem with endogenous biological clocks to synchronize gamete release (Kaniewska et al., 2015). In a study of 12 coral genera analyzed over 20 years, light pollution was found to induce a shift in coral spawning events, one to three nights closer to the full moon than those found on pristine reefs, because chronic light pollution can mask natural moonlight cycles (Davies et al., 2023). Although we did report a shift in the spawning pattern back towards full moon night, we also see a broadening of the spawning window in the opposite direction in Acropora palmata. Pristine sites such as Punta Allen apparently retain the spawning window in time, however it is only based on three data points (two in 2018 and one in 2025) so more data points are needed to confirm if this pattern holds. The shift in spawning phenology on both sides of the spawning window cannot therefore be fully explained by chronic light contamination. Blue and white LEDs can induce reproductive asynchrony because they impair gamete maturation and a reduced number of oocytes per polyp (Ayalon et al., 2021) analogous to that observed in firefly larvae (Chen et al., 2021). That would explain the reduced amount of gamete release but not the shifts we have documented.

Gametes can also be affected by the synergistic effects of elevated temperature and nutrient enrichment (Humanes et al., 2016). The chemical stress from degraded water quality could interfere with coral reproduction, in combination with the warmer sea surface temperatures in the Caribbean during recent years. The major sources resulting in water quality degradation are inadequate treatment and disposal of sewage effluents by the urban and hotel sectors, exacerbated by the massive Sargassum blooms that strand on the coast and decompose resulting in brown tides. The normal disposal practice is to inject waste into the groundwater, which due to karst topography of the region (Banaszak, 2021) is hydrologically connected to the sea resulting in nutrient enrichment within the water column and altering its optical properties and reducing water transparency (Enríquez and Pantoja-Reyes, 2005; Velázquez-Ochoa and Enríquez, 2023). Water transparency is further reduced by the turbidity induced by brown tides caused by decomposing Sargassum (van Tussenbroek et al., 2017) and may affect perception of cues associated with moonlight. Corals lack specialized visual structures and detect moonlight in the blue spectral region using photosensitive proteins such as opsins or cryptochromes (Gorbunov and Falkowski, 2002; Levy et al., 2007; Reitzel et al., 2013), which allow them to synchronize (Panda et al., 2002) at the days and hours scale (Brady et al., 2009). Deeper dwelling species such as Diploria labyrinthiformis and Orbicella faveolata which are typically found at 9–10 meters could be exposed to lower levels of moonlight during the spring and summer months when Sargassum influx is high. However, our results showed synchronization shifted in Acropora palmata, which tends to be distributed close to the surface.

Another possible factor that could be affecting coral reproduction and spawning patterns are endocrine disruptors (Azulay et al., 2026) that interfere with natural hormones such as estrogen. Estrogens naturally occur in corals and regulate reproduction varying in concentration over the course of their reproductive cycle. In the coral Acropora eurystoma from the Red Sea, estrogen levels reach a peak, months prior to spawning during vitellogenesis, when oocytes are formed. As the oocytes mature estrogen concentrations decrease and then increase again closer to the spawning window (Azulay et al., 2026). Exposure to exogenous or environmental estrogens may reduce fecundity in corals as was shown when estradiol reduced the number of bundles spawned by colonies of the coral Montipora capitata relative to control colonies (Tarrant et al., 2004). Estradiol is considered a potent endocrine-disrupting chemical (EDC) when it enters the environment affecting non-target organisms (Dorabawila and Gupta, 2005). Endocrine disruptors such as the natural and synthetic estrogens estrone (E1) and 17 β-estradiol (E2), as well as the synthetic contraceptive 17 α-ethinylestradiol (EE2) have been reported in surface water, wastewater, and groundwater in Mexico potentially carried by inadequately treated sewage effluents (Pastorok and Bilyard, 1985; Armoza-Zvuloni et al., 2012; Vázquez-Tapia et al., 2022).

The highly significant dispersion in spawning patterns in Acropora palmata of moving both closer to full moon night and drifting away from it, could be explained by unpredictable changes in concentrations of contaminants such as endocrine disruptors injected into groundwater and carried by it to the ocean. The concentrations of these contaminants would vary depending on a number of factors, such as population fluctuations due to tourism that could be highly variable throughout the year, rainfall levels that would affect dilution rates of contaminants reaching the ocean along with currents and winds that would affect dispersion rates especially within the reef lagoon. More research is needed to determine whether endocrine disruptors are the main drivers in shifting spawning patterns in corals.

The consequences of losing spawning synchronization in corals could be devastating. Healthy colonies are increasingly scarce due to diseases such as Scleractinian Coral Tissue Loss Disease (SCTLD) and mass bleaching events, such as that which occurred in 2023-2024. Even if the few surviving colonies spawn but fail to do so simultaneously, natural fertilization is highly unlikely. When spawning desynchronizes or is variable across nights, gametes from distinct colonies, which each become diluted rapidly by currents, are exponentially less likely to encounter each other. Consequently, human-assisted fertilization could become an increasingly important tool to save these species from extinction. Under these conditions, traditional passive conservation and fragmentation-based restoration strategies are no longer sufficient to guarantee reef resilience (Banaszak et al., 2025). Human-assisted fertilization and larval propagation emerge as pivotal intervention strategies required to avoid the Allee effect, maximize genetic diversity, and restore declining coral populations. Alongside these biological interventions, improving regional and local wastewater treatment infrastructure offers a direct and far more achievable mitigation strategy compared to global threats such as global warming or Sargassum influx. An important knowledge gap exists about the presence and variation in concentrations of endocrine disruptors in coral reef environments in the Mexican Caribbean due to the large volumes of inconsistently treated wastewater that reaches these delicate ecosystems.