Juveniles reveal natural recovery potential of Caribbean coral species after a widespread disease die-off

The post-disturbance recovery of coral populations and communities relies heavily on the survival, recruitment, and growth of newly settled individuals^1,2. Corals often exhibit high mortality during early life stages³, which can create bottlenecks in recovering populations^4,5. This is particularly relevant following widespread events of coral mortality often associated with acute thermal stress and disease outbreaks, which result in the reshuffling of coral communities and compromise the ecological integrity of the system^6,7. The recovery of coral communities following these events depends on the survival of remnant coral colonies and on coral recruitment. Therefore, understanding the early life stages of corals can provide valuable insights into potential post-disturbance shifts in coral community composition, the long-term effects of disturbances, and the potential of coral recovery^8,9,10.

Stony Coral Tissue Loss Disease (SCTLD) is the most lethal disease ever recorded in the Caribbean, having threatened the populations of over 20 coral species within a remarkably short time frame^7,11. Species such as Dendrogyra cylindrus, Dichocoenia stokesii, Eusmilia fastigiata, and Meandrina meandrites experienced staggering mortality rates (>90%), while other species, such as brain corals, the Orbicella species complex, and Montastraea cavernosa, experienced notable population losses ranging from 20% to 60%^7,12. Furthermore, on afflicted colonies that survived, the reduction in colony living area and size can compromise reproductive potential, ultimately hindering offspring numbers and impeding the potential recovery of affected species¹³. The severity of the situation and the high number of affected species underscore the need to assess the natural processes that facilitate population replenishment, such as coral reproduction, larval recruitment, and the early growth and survival of juvenile corals.

Recruitment and early life stage dynamics are the primary drivers of coral recovery, yet multiple biotic and abiotic factors intricately influence the incorporation of new individuals and the survival of coral juveniles. The settlement and recruitment of coral propagules rely heavily on the substrate, with success rates notably higher on topographically complex surfaces^14,15 and those covered with crustose coralline algae (CCA¹⁶). Conversely, high coral coverage or adverse benthic conditions, such as macroalgal or sediment cover, tend to impede juvenile recruitment and survival due to competition for space, allelopathy, or abrasion^15,17,18. Depth also seems to play a relevant role in shaping juvenile communities, as positive trends in juvenile density and diversity have been observed with increasing depth^19,20,21. However, the relationship between coral juveniles and these environmental variables can be altered by mass mortality events triggered by diseases²² or disturbances²³, prompting questions about the persistence of these patterns following SCTLD outbreaks.

Coral reef recovery varies worldwide; however, a decrease in coral recruitment has been observed in recent decades, raising doubts about their recovery potential^3,15,24. A notable example of this can be observed in Caribbean reefs, where there have been major declines in coral cover and the reef’s functionality⁷. This, along with low recruitment rates²⁴, appears to have hindered their capacity for recovery^25,26. Another crucial concern stems from the species or functional groups that successfully establish and their ecological strategies²⁷. In recent decades, Caribbean coral communities have shifted from being dominated by massive, slow-growing, spawning corals to being increasingly represented by small, rapidly growing, brooding corals with elevated population turnover rates²⁸. Given that SCTLD primarily affects important reef-building species, most sites affected by this disease are likely to be dominated by opportunistic corals⁷. This change could further alter community dynamics, given that clear changes in coral recruitment have already been observed in post-disease communities²⁹.

Wide-spread coral mortality has resulted from SCTLD in the Caribbean, causing non-random changes in community structure that have affected the functional integrity of these communities. In this study, we used extensive post-SCTLD data along a 450-km reef track in the Mesoamerican Reef, including coral communities across sites with different geomorphological and environmental factors, to examine whether the patterns and predictors that drive juvenile corals’ abundance and diversity still govern the post-outbreak composition of early-stage corals. Then, to gain insights into the impacts of SCTLD on coral recruitment, we estimated the establishment date for juveniles of highly susceptible species based on growth rates obtained from the literature and observed colony size to determine whether they survived the outbreak or settled after the event. Lastly, we investigated the broad effects of the outbreak on the population size structure of diseased-affected corals by analyzing the colony size frequency (including juveniles and adults) of affected species in both affected and unaffected sites.

Density and composition of juvenile corals in SCTLD-affected sites

After the SCTLD outbreak, we recorded 4209 juvenile corals (33 species, mean density of 4.18 ± 1.86 individuals·m^–2) in the affected sites. Additionally, we observed 22,254 adult colonies belonging to 43 species (average density of 5.34 individuals·m^–2 ± 2.13) across all sampling sites. For some rare species, such as Dendrogyra cylindrus and Mussa angulosa, we found no juveniles and few adults (we only observed adult D. cylindrus in the non-affected sites of Banco Chinchorro). Post-outbreak communities of juvenile and adult corals were dominated by three species: Agaricia agaricites, Porites astreoides, and Siderastrea siderea (Fig. 1a). Collectively, these species accounted for nearly 70% of all recorded individuals (Fig. 1). It is noteworthy that in the case of specific species, such as S. siderea, E. fastigiata, and D. stokesii, the density of juvenile individuals surpassed that of adults (Fig. 1a). In general, recruit density was positively correlated with adult density (Fig. 1b).

Post-outbreak predictors of juvenile coral communities

The density of conspecific adults stands out as a key determinant of juvenile presence across various susceptible species and the density of the most abundant species (Fig. 2). This observation persisted regardless of the reproductive strategy of the species and susceptibility to SCTLD. Interestingly, while a positive relationship between conspecific juveniles and adults was evident at the species level (Fig. 2), a negative relationship was observed when analyzing the broader community. At the community level, diversity indices showed declines in sites with higher adult coral density: q1 decreased by more than 20%, while q2 declined by over 50% (Figure. S3d, f).

Depth was strongly correlated with the presence of juveniles of nearly half of the species susceptible to SCTLD (Fig. 2a, b). Moreover, depth was positively associated with higher diversity values across all Hill numbers, with deeper sites nearly doubling diversity indices compared to the shallower sites (Figure. S3a, c, e). Similarly, higher CCA coverage increased species richness (q₀) by >30% (Figure. S3b), although slightly less pronounced than depth (Figure. S3b), as this effect was only observed in two species: A. agaricites (Fig. 2a) and Pseudodiploria strigosa (Fig. 2b). Notably, for S. siderea and A. agaricites, latitude had a significant, yet opposite effect. Specifically, S. siderea exhibited an increase in juvenile density in southern reefs, whereas A. agaricites juvenile density increased towards the north (Fig. 2a). It is worth mentioning that, while these variables were significant, their impacts were subtle (Table S4). It is important to acknowledge the challenges in drawing definitive conclusions for certain species, such as Diploria labyrinthiformis, M. meandrites, and Colpophyllia natans, given the limited data points and resulting uncertainty in the calculated odds ratios.

Recruitment among dominant and highly susceptible species

Overall, we found that almost all species affected by SCTLD recruited before and after the outbreak. However, it is important to note that the majority of individuals for most species were recorded before the outbreak (Fig. 3). This makes the behavior of certain species, such as E. fastigiata and D. stokesii, particularly notable, as they showed higher recruitment after the outbreak (92% and 62% of individuals recruited post-outbreak, respectively). Although not as pronounced as in the previously mentioned species, it is also noteworthy that nearly half of the recruits of P. strigosa established after the outbreak. The pattern exhibited by this species hints at a potential recovery response to the substantial mortality experienced by the adult colonies.

We also found a clear difference in the recruitment of A. agaricites (n = 1659), P. astreoides (n = 1121), and S. siderea (n = 1002), which exhibited greater numbers of recruits compared to the other species that were present in almost all sites. This difference became even more pronounced when we considered the low representation of juveniles from highly affected species, such as M. meandrites (n = 6), C. natans (n = 11), and D. labyrinthiformis (n = 10), as well as moderately affected species like Orbicella (n = 20).

Comparison of size structure in SCTLD-affected and healthy sites

A comparison among the most susceptible species in affected and unaffected (i.e., Banco Chinchorro) sites revealed differences. Average colony size values were notably higher in Banco Chinchorro compared to the affected sites (Table S6). Additionally, skewness was higher in the sites affected by the disease (Table S6). Highly affected species, such as C. natans, D. labyrinthiformis, E. fastigiata, and M. meandrites, showed significant shifts in size structure following SCTLD. Notably, larger size classes were almost entirely lost in affected sites (Fig. 4). The most striking example is M. meandrites, where large colonies accounted for 66% of individuals in unaffected sites but less than 1% in affected sites. This pattern is also evident in C. natans (Fig. 4), where the mean log-transformed colony size in affected sites (3.66 cm²) was nearly half that in unaffected sites (5.48 cm²), and the skewness was the highest recorded among all species (Table S6). This illustrates a change in the size structure of certain species due to SCTLD, with populations transitioning toward domination by smaller colonies.

Other species showed similar but subtler trends, including moderately affected species such as P. strigosa, M. cavernosa, and S. siderea (Fig. 4). While differences in average sizes and skewness were significant (Table S5), these differences were smaller than those observed in the previously mentioned species. Among these species, the most significant reduction in large size classes was observed in P. strigosa, where the proportion of large colonies decreased from 65% in unaffected sites to 43% in affected sites. It is worth highlighting that the size distribution of S. siderea in both affected and unaffected sites was skewed toward smaller colonies, with colony size decreasing further following the onset of SCTLD (Fig. 4).

In contrast, Orbicella was the only species with nearly identical log-transformed average sizes (6.66 cm² in affected sites vs. 6.97 cm² in unaffected sites) and distribution values (Fig. 4). Similarly, this species showed the lowest skewness, indicating a significant inclination towards larger sizes. Although the average size of D. stokesii remained comparable in affected and unaffected sites (Fig. 4), the skewness indicated a reduction in population size.

The ecology and functionality of coral assemblages in the Caribbean were undergoing severe ecological changes prior to the SCTLD outbreak. Chronic and acute disturbances had progressively driven a decline in the abundance of the main reef-building corals that was accompanied by a concomitant increase in the relative or absolute abundance of opportunistic species^7,30. However, given the magnitude of the resulting coral mortality associated with SCTLD, understanding the roles of recruitment and juvenile dynamics is crucial to identifying the natural recovery potential of the affected species. Our findings confirm that the populations of several susceptible species underwent severe changes, not only in overall number but also in terms of the skewness of their size distributions, resulting in disproportionately smaller colonies. We recorded juveniles of the most affected species during our surveys and identified consistent ecological patterns regarding the abundance and diversity of species despite the mass mortality event. Overall, we found evidence suggesting that some species may naturally recover through recruitment, including some species that are highly susceptible to SCTLD, such as P. strigosa and E. fastigiata. In these species, we found that a high proportion of juveniles had been recruited to the population following the peak of the SCTLD outbreak, revealing that the surviving adults were still capable of sexually reproducing despite drastic decreases in their abundance³¹. However, the prognosis is not as encouraging for the species that were most afflicted by SCTLD, such as D. cylindrus and M. meandrites. Indeed, we did not observe any living colony (either juvenile or adult) of D. cylindrus in the affected sites, and we only observed a small number of juvenile M. meandrites, which did not allow us to explore any trends for this species.

The juvenile community post-SCTLD was dominated by three species: A. agaricites, P. astreoides, and S. siderea. The dominance of a few coral species within juvenile communities is not surprising, as similar patterns have been documented since the early 1980s (e.g.,^19,32) and more recently, following the SCTLD outbreak^29,33. While A. agaricites and P. astreoides are brooder species and only mildly affected by the disease, S. siderea is a spawner and was moderately affected by SCTLD (40%)^7,34. Despite this, S. siderea shares similar traits with brooder species, such as high rates of population turnover and sexual maturity at small sizes that enable the larvae of this species to extensively settle near adult colonies, even under marginal conditions^35,36,37. These life history traits allow these three species to colonize space and dominate juvenile communities rapidly. This dynamic may create a positive feedback loop between adults and juveniles in the community³⁸, accounting for the positive relationship observed between conspecific adults and juvenile density in these three species.

The widespread mortality of corals caused by SCTLD led to changes in species composition and a decrease in coral diversity. Despite these impacts, it is noteworthy that our study still identified certain consistent patterns in juvenile community structure; notably, we observed a positive relationship between depth and diversity, as shown in previous studies^39,40. This trend persisted despite SCTLD affecting all depths similarly⁷. The continued presence of these patterns may indicate a form of ecosystem resilience and may suggest a hierarchy of environmental filters for the impacts caused by the disease⁴¹. Similar to depth, benthic composition played a crucial role in structuring juvenile communities. We found that CCA coverage and adult density played significant yet contrasting roles in determining diversity and juvenile density patterns in different species. Higher CCA coverage may facilitate the establishment of various coral species^4,16, thereby increasing site richness. On the other hand, we observed that the diversity of juveniles tends to decrease as adult density increases (Figures. S3D, F), particularly at sites with high adult densities of A. agaricites and P. astreoides. This suggests that adult density of this abundant species may reduce diversity by excluding juveniles of other species attempting to settle, indicating a mechanism of preemptive competition^21,42. Furthermore, competition in these saturated sites likely extends beyond corals, potentially involving other benthic organisms such as macroalgae.

However, it is also important to note that at the species level, we observed a positive effect of conspecific adult density on the density and presence of several species, including those affected by SCTLD. These findings suggest that recruitment in most coral species, regardless of their reproductive strategy, relies more on the local group of adults within the same site than on larval input from colonies at other sites. Such a pattern suggests that Caribbean coral populations may function as relatively closed systems, a view consistent with previous studies that highlight the predominance of local over external recruitment^43,44,45. However, these dynamics may differ between reproductive strategies. Broadcast spawners, in particular, often display greater interannual variability in recruitment than brooders, with outcomes influenced not only by larval dispersal but also by environmental conditions affecting fertilization, transport, and settlement success^15,33,46. Taken together, these findings raise the question of whether the recruitment restriction we observed reflects a general feature of Caribbean coral populations or instead results from historical disturbances and changes in environmental conditions^47,48. At the same time, it is essential to recognize that our observations may not fully capture the finer interannual variability or other ecological processes that influence recruitment.

Juveniles from most species were observed in our post-SCTLD outbreak surveys, indicating that either they survived the outbreak or recruited during or after it. We aimed to elucidate this by predicting the time of recruitment based on growth rates obtained from the literature. We, however, must acknowledge that results might not be conclusive, as our dates of recruitment are crude estimations based on published growth rates obtained under different environmental conditions and in an overall different ecological context^49,50,51. Coral growth is a highly variable process, influenced by a wide range of biotic and abiotic factors, resulting in substantial variation across both temporal and spatial scales⁵¹. We, however, and although growth rates derived from the literature may not fully capture local conditions, consider the values used here to be conservative, as we prioritized the most recent and typically more moderate estimates (see Methods and⁵¹), which would likely bias our results toward underestimating the actual number of recruits established after the SCTLD outbreak. Our estimation of SCTLD spread also assumed a uniform rate, an approach that likely oversimplifies the complex dynamics of disease transmission, which can be influenced by factors such as hydrodynamics, host density, and stochastic events^52,53. Each of these assumptions, whether related to growth rates or disease spread, introduces its own uncertainty, and their combined effect may substantially influence the accuracy of recruitment dates. Nevertheless, we consider that applying a consistent approach to estimate recruitment dates enables robust comparisons across reef sites and species, even in the absence of raw recruitment data, and allows for the retrospective assessment of pre- and post-outbreak patterns, providing valuable insights into the recovery potential of coral populations affected by SCTLD.

First, although the proportion of corals that settled after the SCTLD outbreak is significantly lower compared to pre-outbreak levels, for many species, there were small recruits that very likely recruited within months of our surveys (Fig. 3). Second, the recruitment of new individuals following the outbreak implies that despite substantial declines in population size and colony damage, adults retained some capacity for sexual reproduction, which has been previously shown³¹. Third, it is likely that some juveniles of the most afflicted species survived through the outbreak (e.g., S. siderea and M. cavernosa in Fig. 3). This inference is supported by the fact that virtually no juveniles exhibited signs of the disease during our post-outbreak surveys. This can be interpreted as a differential resistance to the disease based on the life history stage, suggesting that juveniles are less affected than adults^54,55. However, this interpretation does not agree with the available evidence regarding the susceptibility of different life stages to SCTLD. Williamson et al.⁵⁶ found that recruits and adults of C. natans and D. labyrinthiformis were similarly susceptible to SCTLD in controlled conditions. Additionally, Hayes et al.²⁹ and Croquer et al.⁵⁷ reported that small colonies and recruits were equally prone to infection. Considering this evidence, it is likely that our observations reflect only the surviving recruits without extending to the survival and resistance of all recruits. This suggests that the abundance of juvenile corals from afflicted species may have been higher before the outbreak, and the severity of the disease may have been comparable to that experienced by adults.

One key question is whether the affected species will be able to recover naturally after the SCTLD outbreak. Overall, we observed corals in the early life stages of the most affected species, which either resisted the outbreak or recruited after the mortality event. In both cases, our results are positive and show some level of resilience of the affected populations at the regional scale. Furthermore, we identified different population mechanisms that could lead to the potential recovery of some of the affected species, at least in the short-term. A response marked by a considerable focus on recruitment amidst mass mortality events could suggest a Boom or Bust strategy. Such a strategy would entail prioritizing resource allocation towards reproduction rather than individual maintenance^37,58. This could be the case for species like E. fastigiata, in which we observed that most individuals recruited after the outbreak (Fig. 4). Similar patterns have also been reported for S. siderea, which experienced a mass-recruitment boom (~70-fold increase) within a single year across the Florida Reef Tract, enabling this species to dominate recruit cohorts over hundreds of kilometers³³. Comparable strategies have been described in Indo-Pacific corals^59,60, suggesting that Boom-or-Bust recruitment pulses may represent a broader mechanism of population recovery and persistence in corals. Another population mechanism that could promote recovery is the ability to tolerate disturbances and chronic stressors in different life stages. Species like P. strigosa show remarkable longevity in large colonies, resistance to fission, lower annual mortality rates for juveniles compared to other Caribbean corals^19,61, and great adaptability to adverse conditions^62,63. These characteristics could explain the high survivorship of their juveniles and the capacity of this species to sexually reproduce following the outbreak.

The remarkably low values and lack of juveniles for some other species are concerning and noteworthy and may be attributed to two reasons that are not necessarily mutually exclusive. The low juvenile values may be intrinsic to these species, as their life histories may not heavily rely on a consistent supply of juveniles for population turnover. Species that are crucial for the structure of the reef, such as the Orbicella species complex, could exhibit a storage effect^64,65 and recruit through infrequent “masting” events that unfold at broad temporal scales that ecological studies have not yet detected. However, even if this recruitment strategy is at work, some researchers question its ability to restore populations to previous levels due to the extensive damage they have suffered in recent decades⁶⁵.

The chronic stressors and frequent disturbances in the Caribbean compromise the recovery of species that were more prominent in previous decades, such as C. natans, M. meandrites, D. cylindrus, and D. labyrinthiformis. The absence of juveniles of these species seems to be linked to the decline in their populations, colony sizes, and even the mortality of recruits rather than reflecting a specific ecological strategy that results in these notably low values⁵⁶. This is evident in the disparity in juvenile density for these species from the 1980s to the present^19,32. D. labyrinthiformis serves as a prime example to further illustrate this point. Although this species reproduces by spawning, it exhibits multiple reproductive events per year, a short planktonic phase, and rapid settlement, which have been associated with brooder reproduction and high recruitment rates⁶⁶. The exceptionally low abundance of individuals of this species cannot be solely explained by its reproductive strategy but also by the extensive population damage that it has suffered.

Our findings show a reduction in the size distribution of the affected species compared to those in sites unaffected by SCTLD. This might have population-level implications. First, a decrease in the size distribution of a population might limit their reproductive capacity¹³. This is particularly important as colony size plays a pivotal role in determining gamete production capabilities³⁵. Furthermore, this mass mortality event unfolded in the context of a pre-existing population decline of several species affected by the disease⁶⁷. This historical trend, coupled with the high mortality attributed to the disease, exacerbates the already notable decrease in population size⁷.

More specifically, the observed size distributions and skewness in species like D. stokesii, E. fastigiata, and S. siderea illustrates that large colonies are relatively rare in the populations of affected and unaffected sites. However, their sizes in affected sites continued to diminish further. This aligns with the findings reported by Lewis⁶⁸ and suggests that skewness could be a consequence of biotic and abiotic disturbances. For species like P. strigosa and M. cavernosa, the survival of large-sized colonies is crucial for maintaining healthy populations⁶¹. Our study also revealed a decline in these specific size classes within sites affected by SCTLD. This observation underscores that despite the relatively high recruitment, a notable segment of the population characterized by larger sizes was adversely affected, posing a considerable challenge for this species to recover.

While this study offers only a snapshot of the juvenile coral community following a mass mortality event, it provides valuable insights into the early stages of recovery. Ultimately, the recovery of affected species will rely on the survival into adulthood of some of the juveniles and their ability to reproduce successfully. Consequently, conducting long-term studies that follow the trajectories of these populations are crucial for drawing more robust conclusions and generating a deeper understanding of post-disturbance dynamics in coral juvenile communities. Despite its temporal limitations, our study highlights the potential to identify priority conservation sites to conserve diversity and protect the populations of species most impacted by SCTLD. In particular, we suggest prioritizing sites with depths ≥10 m that contain adults of the affected species. These sites could serve as crucial refuges to safeguard the diversity and populations of the affected species⁶⁹.

The study area included 75 sites across the Mexican Caribbean (Figure. S1): 65 sites along the mainland and Cozumel Island were affected by SCTLD; the 10 sites of Banco Chinchorro were unaffected by SCTLD (Figure. S1⁷). In total, we recorded 43 coral species, of which nine species were considered to be highly susceptible to SCTLD (Table S7, given the high disease prevalence (>20%) and mortality observed during the SCTLD outbreak in the same geographical region⁷. The data set included information from 63 fore-reefs and 12 back-reefs (1–24 m⁷⁰). Some of these locations were sampled twice, once in 2021 and again in 2022. The data from repeat sampling sites in 2021 were excluded from multivariate, diversity, and density analyses to prevent the sampling effort and year from inflating the data and introducing noise. However, these sites were included in the estimations of settlement dates and size structure. The decision to include these sites was motivated by the goal of identifying the maximum number of species susceptible to SCTLD. The Banco Chinchorro data were used exclusively to compare affected and healthy sites in the size-structure analysis.

Coral community census

We focus on surveying juvenile (defined as any individual coral polyp or colony with a maximum diameter ≤4 cm⁵¹) and adult coral colonies (colonies > 4 cm). In this context, first, it is relevant to distinguish between a recruit and a juvenile, as this distinction is crucial for interpreting patterns of population dynamics and early life-stage survival⁷¹. Recruit refers to the initial establishment of individuals, often immediately after larval settlement, while juvenile relates to individuals that have already survived the earliest and most vulnerable stages and have begun to grow within the habitat⁷¹. Although recruitment was not directly measured in this study, we use the term recruit when referring to individuals for whom we inferred the approximate time of settlement into the community.

In each site, we established 4–8 transects for juvenile surveys (10 m × 0.25 m, 15 m²·site^–1 ≈ 1125 m² total area) and 4–9 transects for adult surveys (10 m × 1 m, 60 m²·site^–1 ≈ 4500 m² total area) (average of 6 ± 0.76 transects). The sampling effort for juveniles was considerably greater than in other studies and monitoring protocols because we aimed to maximize the possibility of including rare species, many of which have been affected by SCTLD.

We recorded the maximum diameter, perpendicular diameter from the maximum diameter, and height of all corals. Additionally, we checked for any signs of partial mortality, bleaching, or disease. All juvenile transects were conducted by experienced surveyors. Juvenile and adult densities were calculated for the entire sampling area. Because of the difficulties in identifying juvenile corals in field surveys and the non-destructive nature of our surveys, certain individuals were only classified at the genus level (Orbicella species complex, Porites digitata complex, Scolymia spp., Madracis spp., and Mycetophyllia spp.). We were unable to identify corals in only 15 cases.

Diversity of juvenile corals

We used Hill numbers to represent the diversity patterns of coral juveniles after the SCTLD mass mortality event. The exponent “q” serves as an indicator of diversity. A “q” value of 0 makes the Hill number equivalent to species richness; as “q” increases, greater emphasis is placed on common species over rare ones. When utilizing Hill numbers, diversity should be presented in terms of richness (q = 0), common species (q = 1), and dominant species (q = 2). Given that sampling efforts varied across sites, indices were standardized based on sampling coverage. Following the recommendations of Chao et al.⁷², a coverage of 1 was used to extrapolate the cases of q1 and q2, while a coverage of 0.95 was employed for q0. Extrapolation and rarefaction analyses were conducted using the ‘iNext’ package in R⁷².

Predictors of juvenile density and diversity after SCTLD

We examined the factors influencing juvenile coral density patterns following SCTLD mortality in the affected sites. Our focus encompassed a range of biotic and abiotic variables, and we identified several critical factors affecting early life stages (Table S1). To obtain information on benthic characteristics, we evaluated coral density and the coverage of CCA (including Peyssonnelids), macroalgae, hydrocorals, and sediments. These variables were selected based on their negative or beneficial impacts on coral recruitment (Table S1). Additionally, we evaluated the influence of anthropogenic pressures, which have been shown to adversely impact coral survival and recruitment, such as human population density and thermal stress (i.e., maximum degree heating week). We also assessed the effects of several environmental conditions, such as depth, latitude, and reef complexity.

First, we conducted a redundancy analysis (RDA) with all the covariates. Then, we performed an analysis of variance (ANOVA) to evaluate the impact of each predictor on the response matrix through sequential hypothesis tests. Based on these results, we selected the variables that demonstrated a significant effect; these were adult coral density, depth, sediments, CCA, population density, and latitude (Figure. S2). Using these pre-selected variables from the RDA, we constructed linear models (Table S3 and S4) and a binomial logistic generalized linear model (Table S5) to assess their effects on the density and presence of the most affected and the most abundant species, as well as species diversity. Only the effects of the significant variables are shown in Fig. 2 and Figure. S3.

Recruitment among susceptible species before and after the SCTLD outbreak

We used colony size and species to estimate when every juvenile coral established itself in the affected sites. First, we conducted an extensive literature review to obtain the growth rates for juvenile coral species. We prioritized those containing the most recent information (Table S2) following Edmunds⁵¹, who suggested that current juvenile growth rates could be considerably lower than those observed several decades ago. When species-specific information was unavailable, we used the growth rate of the genus; if no information was available for the juvenile stage, we used data from adults. We estimated the number of days that had passed since recruitment by multiplying the maximum observed diameter by the daily growth rate of the species. We determined whether an individual was recruited before or after the onset of SCTLD. For this, we assumed the outbreak started in Puerto Morelos in July 2018 and subsequently spread from north to south^7,73. We also assumed a spread rate of 1 km·day^–1, as this rate was previously estimated for our study region⁷⁴. With this information, we were able to define a specific date for the onset of the SCTLD outbreak in all sampling sites. If the number of days since recruitment was greater than the estimated outbreak date for a coral, then the recruitment of that coral occurred prior to the SCTLD outbreak. This standardization allowed us to compare the recruitment dates of all the observed corals.

Population size structure after SCTLD

To investigate the impacts of SCTLD on the size structure of the populations of the most affected species, we calculated the total living area for all juvenile and adult corals in all sites, including those impacted by SCTLD and those in Banco Chinchorro, which served as a control site as SCTLD was not present. To ensure consistency despite variations in sampling size, we extrapolated juvenile records to align with the adult sampling area. We assumed that all colonies had a hemispheric shape and calculated the living surface area of each following the methods of González-Barrios and Álvarez-Filip⁷⁵. We log-transformed the living surface area of each colony and assigned a size category⁷⁶. The log transformation increased the number of smaller size classes while reducing the number of larger size classes⁷⁷, resulting in a relatively normal distribution of the data.

We then use the skewness, the coefficient of variation, mean colony size, and standard deviation to describe distributions across various species susceptible species in affected and unaffected sites. In particular, skewness indicates the shape of the data distribution around the mean, where a value different from 0 indicates that the distribution is asymmetric. This asymmetry can be interpreted in two ways. If the skewness value is negative, the tail is skewed to the left, and most of the size classes are large; in contrast, if the value is positive, most of the size classes are small⁷⁶. Skewness is also related to the input of new individuals into populations and coral longevity⁷⁷. We tested for differences in population size structure between sites using the Kolmogorov- Smirnov test.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.