Restoration stocks of a caribbean coral species contain low levels of inherent stony coral tissue loss disease resistance Authors: Sara D Williams, Michael S Studivan, Stephanie Sirotzke, Natalia Andrade Rodriguez, Zachary Craig, Sarah B Hamlyn, Daniel M Holstein, Lys M Isma, Nicholas J MacKnight, Joseph A Mandara, Stephanie M Rosales, Ashley M Rossin, Nikki Traylor-Knowles, Cory Walter, Benjamin D Young, Erinn M Muller Published: 2026 ABSTRACT AbstractAs coral diseases continue to devastate reefs worldwide, variation in disease resistance must be considered in conservation and management strategies. Stony coral tissue loss disease (SCTLD) has decimated Florida’s Coral Reef since 2014 and is now endemic within the region. It has affected many endangered coral species, including the major reef-building species Orbicella faveolata; however, SCTLD resistance has been documented within this species. Utilizing O. faveolata in propagation for restoration efforts, we exposed 154 multi-locus genotypes to SCTLD during four sequential ex situ experiments. Four resistance groups were identified using measured disease phenotype metrics, but only 12 genotypes with multiple replicates were consistently resistant to SCTLD. Additionally, several genotypes were both resistant to SCTLD in the ex situ experiments and had a lower risk of SCTLD when outplanted in a targeted study in the Lower Florida Keys, indicating congruency in disease resistance in field and lab settings. However, long-term outplant survival was more variable, likely due to the interaction of multiple stressors and confounding variables. Although we determined overall low levels of inherent SCTLD resistance, our results support integration of resistance testing to multiple stressors, including disease, within restoration strategies to enhance outplant survival and recovery of reefs.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-56396-7.Keywords: Coral disease-resistance, Stony coral tissue loss disease, Disease transmission experiment, Resilient restoration, Coral restoration and ecologySubject terms: Ecology, Ecology, Evolution, Genetics, Microbiology Introduction Florida’s coral populations have declined substantially over the past four decades, leading to habitat degradation and loss of economic and coastal resiliency benefits1–3. Stony coral tissue loss disease (SCTLD) has been a major contributor to the deterioration of Florida’s Coral Reef since 2014, when it was first observed near Miami, Florida4. It is now considered endemic within the region and has spread across much of the Caribbean5,6. Coral disease dynamics involve a complex interplay between the coral host, its microbiome, and the environment, all of which interact to influence individual susceptibility7. About half of the species observed in the Western Atlantic and Caribbean are susceptible to SCTLD; however, susceptibility varies by species and region8–11. Intraspecific variation in SCTLD susceptibility has been observed in both ex situ12 and in situ studies13, though no studies to date have identified congruence in resistance patterns between field and lab settings. SCTLD has likely killed millions of coral colonies across nearly half of the reef-building species in Florida8,14,15, making its impacts unprecedented relative to other coral diseases. Restoration within Florida is critical to recovering some semblance of ecosystem function, as protection of some reef habitats alone has not resulted in coral population recovery16. Many restoration groups are actively seeking to identify and propagate resilient genotypes as they present the highest rate of success following outplanting. However, the process of characterizing disease resistance in corals has posed difficulties. In situ studies of coral disease resistance are influenced by co-occurring and potentially confounding environmental factors known to affect disease dynamics7,17, local hydrodynamics affecting transmission18, and a potentially more variable pathogen load. Controlled, ex situ disease transmission experiments provide a direct comparison of disease-exposed and control outcomes. However, they are often limited in scope (i.e., few genotypes) based on resources, may underestimate in situ levels of disease resistance due to potentially larger pathogen loads, and do not mimic the complex interactions of abiotic and biotic variables in situ. Therefore, multiple approaches for assessing variation in disease resistance are needed to best inform conservation and restoration strategies. Orbicella faveolata is considered to be an intermediately susceptible species to SCTLD15, as it has relatively slower progression rates than other SCTLD-susceptible species10. Additionally, the high frequency and prevalence of SCTLD in O. faveolata8 may help sustain local outbreaks on Florida’s Coral Reef10, making it a priority species for disease interventions. Although O. faveolata is a major reef-building species in the Caribbean, its ability to naturally recover from disturbances is hindered by a relatively slow growth rate19. While microfragmentation can increase growth rates and colony fusion20, restoring this threatened species takes substantial time and effort, especially while SCTLD remains a threat to susceptible species. Therefore, alternative strategies to increase growth rates and outplant success to improve restoration outcomes are necessary. One such approach is to bolster genetic variation in restoration stocks of O. faveolata to include advantageous traits, such as resilience to various stressors. Given that intraspecific variation in O. faveolata SCTLD susceptibility has been observed12,13, this variation could be tested for within restoration stocks and then incorporated into restoration strategies. Adaptive management strategies that consider genetic diversity and its role in resistance to disease are paramount to ensuring the success of coral reef conservation and restoration initiatives, particularly as disease outbreaks continue to worsen. Utilizing Mote Marine Laboratory’s existing restoration inventory of O. faveolata in propagation, we exposed multiple replicates of 170 putative genotypes (i.e., based on provenance information and prior genetic testing) to SCTLD during four sequential ex situ experiments to determine levels of inherent disease resistance. We used whole-genome sequencing of one replicate from each of the 170 putative genotypes to confirm genetic identity and relatedness. Whole-genome sequencing identified 157 multi-locus genotypes, which were then classified into four resistance groups using disease phenotype metrics measured during the experiments. In addition to the monitoring conducted by Mote’s Coral Reef Restoration Program for all outplanting events, a subset of the genotypes tested in the transmission experiments was also included in a targeted outplant study of SCTLD resistance at the time that the disease front was moving through the Lower Keys. This allowed us to compare in situ and ex situ SCTLD resistance levels for Mote’s restoration stock of O. faveolata. Our results support the need for management strategies that consider both genetic and phenotypic diversity, while strategically integrating genotypes with known disease resistance, to improve coral restoration efficacy in the Caribbean – a known disease ‘hot-spot’7,21,22. Results Transmission experiment risk comparison Four transmission studies conducted in 2022 exposed 170 putative genotypes of Orbicella faveolata to SCTLD (Fig. 1; Table S1). Relative risk analysis determined that all experiments had statistically indistinguishable, significant disease risks, with all median relative risks greater than one and overlapping credible intervals (Fig. 1b). Disease signs were observed in some corals after one day of exposure in all experiments, with only a small portion, 12.8 ± 2.2 (mean ± S.E.), of individuals exposed during each study having no visible signs in the approximately two-week long experiments (Fig. 1c-f; Fig. S1). On average, exposed fragments developed disease signs within 4.3 ± 0.5 days after exposure, and tissue loss progressed to cover 10% or more of a fragment, at which point they were removed, within 2 ± 0.1 days after initial disease signs were observed, per experiment (Fig. S1). Although four of the twelve disease donor colonies had cellular signs more consistent with “healthy” colonies according to histological analysis (see supplemental material for details; Fig. S2), all diseased donor median relative risks were greater than one and had overlapping credible intervals (Fig. S3). This, in combination with visual observations and quantification of tissue loss progression (Fig. S3), confirmed that all exposed corals were at similar risk to SCTLD, regardless of the disease donor colony they were exposed to or the experiment conducted. Lastly, control corals exposed to healthy donor colonies in all experiments had significantly less risk of disease transmission relative to the fragments exposed to disease donor colonies (Fig. S3). General results of the four sequential direct-contact, SCTLD transmission experiments. (a). Photo of a disease-exposed aquaria in a transmission experiment with the two experimental fragments touching the diseased donor fragment at its lesion. (b) The Bayesian relative risk analysis for Orbicella faveolata fragments exposed to SCTLD in the four experiments. Points are median relative risk values, and lines denote the 95% credible intervals, where relative risk is on the log scale. Credible intervals entirely above a relative risk of 1 (vertical black line at 100) indicate a significant increase in disease risk after exposure to disease donor colonies. (c-f) Disease incidence over time for all four transmission experiments was visualized as the number of exposed individuals in one of five health categories: healthy, watch (stressed, but not clear disease signs), initial disease (first day of disease signs), diseased (clear disease signs), and removed (more than 10% tissue loss, sampled, and removed from the experiment). Whole-genome sequencing A total of 180 samples (174 putative genotypes from the transmission experiments and Mote’s restoration stock, including 6 sequenced duplicates) were processed for 10x coverage whole-genome sequencing (WGS), producing a total of 4.6 billion sequenced reads, with a mean read count of 26 million per sample. Following quality trimming and alignment to the O. faveolata genome23, an average of 16.3 million coral-aligned reads remained, which was an average alignment rate of 62.5%. Cleaned coral reads were then processed through the GATK pipeline24 according to best practices (described in associated GitHub Repository, see Data Availability), resulting in 24,097,804 raw SNP loci. Quality filtering resulted in 10,366,466 high-quality SNPs that were used for genotype assignments of experimental samples (Fig. S4). Based on three out of the six sequenced duplicates that had sufficient read counts and alignment rates, a clonal detection threshold of 0.015 (1.5%) was used on the genetic distance matrix, where any samples with a genetic dissimilarity below that value were considered true clonal genotypes (Fig. 2a). Genotype assignment, therefore, resulted in 157 multi-locus genotypes (Fig. 2b), which corresponded to a clone rate of 12.6%. None of the multi-locus genotype assignments clashed with Mote’s provenance metadata (e.g., such as if a sexual recruit matched with a wild-collected coral of opportunity as a clone-mate), and therefore analyses could proceed with the combined genotype and provenance metadata. Dendrograms of putative and confirmed genotypes based on genetic distance. (a) Dendrogram of putative genotypes based on provenance data, where gray squares indicate technical replicates sequenced in duplicate, and the horizontal dashed line indicates the clonal threshold used in clonal genotype detection. Any branches below the threshold are considered clonal genotypes. (b) Dendrogram of unique multi-locus genotypes based on genetic distance following removal of clonal genotype samples. Genotypic variation in disease resistance 154 multi-locus genotypes, hereafter referred to as “WGS-genotypes,” were identified from the 170 putative genotypes used in the transmission experiments. Three disease phenotype metrics, on a scale from 0 (resistant) to 1 (highly susceptible), were determined for each WGS-genotype: susceptibility, transmission, and progression. The susceptibility, or fraction of exposed individuals that developed disease signs, across all WGS-genotypes was 0.82 ± 0.31 (mean ± S.E.). Transmission, the standardized (0–1, slow to rapid) metric of days until disease signs, was 0.57 ± 0.26 (mean ± S.E.) across all WGS-genotypes. Progression, the standardized (0–1, slow to rapid) metric of days with disease signs until ≥ 10% tissue loss, was 0.71 ± 0.28 (mean ± S.E.) across all WGS-genotypes. A hierarchical clustering analysis using the three disease phenotypes resulted in five resistance groups – 1/1 resistant (only one replicate was exposed and it was resistant, thus all metrics were 0), resistant, intermediate, susceptible, and highly susceptible – for all 154 exposed WGS-genotypes (Fig. S5). Twenty-six WGS-genotypes were categorized as resistant, fourteen of which had only one replicate exposed (Fig. S5). All resistant genotypes had susceptibility, transmission, and progression values ≤ 0.5 (Fig. S5). Considering solely the 53 WGS-genotypes that had three or more exposed replicates yielded similar hierarchical clustering results, therefore we will focus on the results of this highly replicated subset (Fig. 3). Susceptibility (Kruskal-Wallis: df = 3, chi-squared = 49.90, P = 8.4e-11), transmission (ANOVA: df = 3, residuals df = 49, F = 51.63, P = 3.4e-15), and progression (Kruskal-Wallis: df = 3, chi-squared = 38.57, P = 2.14e-8) all significantly differed among the resistance groups determined by the hierarchical clustering analysis (statistical results for multiple comparisons provided in Table S3). Only three out of the 53 highly replicated WGS-genotypes were determined to be resistant (Fig. 3a). These three resistant WGS-genotypes had at least two out of three exposed replicates remain healthy during the transmission experiments (Susceptibility = 0.33; Fig. 3b), significantly lower transmission values than the other resistance groups (Fig. 3c; Table S3), and slower progression rates than the other resistance groups (Fig. 3d). Eighteen intermediately susceptible WGS-genotypes had at least one out of three exposed replicates remain healthy (Susceptibility = 0.66; Fig. 3b). However, the disease phenotype metrics of the intermediately susceptible WGS-genotypes were not significantly different from the four WGS-genotypes determined to be susceptible (Fig. 3; Table S3). Susceptible WGS-genotypes differed from the 28 highly susceptible WGS-genotypes by having significantly lower transmission values (Fig. 3c; Table S3). Highly susceptible WGS-genotypes had a significantly higher proportion of exposed fragments develop disease signs (Susceptibility = 1 ± 0.00, mean ± S.E.; Fig. 3b) than intermediate and resistant genotypes, developed disease signs significantly faster than all other resistance groups (Transmission = 0.75 ± 0.02; Fig. 3c), and lost tissue significantly faster than intermediate and resistant genotypes (Progression = 0.87 ± 0.02; Fig. 3d). Results of the hierarchical clustering analysis utilizing the disease phenotypes to determine resistance groups for the 53 WGS-genotypes with three or more replicates. (a) Dendrogram of the hierarchical clustering utilizing the three disease phenotype metrics (b-d). Colors denote the four resistance group clusters. Boxplots of the three disease phenotype metrics for the four resistance groups include: (b) the susceptibility (fraction of replicates that showed disease signs), (c) Transmission (standardized metric of days until disease signs), and (d) Progression (standardized metric of days with disease signs until ≥ 10% tissue loss). Values closer to 0 are considered more resistant. Outplant disease risk and survivorship Eight of the WGS-genotypes were represented in a targeted outplanting study that took place from September 2018 to November 2019 as SCTLD was actively spreading through the Lower Florida Keys. Of the 415 fragments outplanted at two offshore sites, 51 (12.3%) fragments from six of the eight genotypes developed signs of SCTLD. We compared the outplanted genotypes’ relative risk of developing SCTLD with their resistance group determined by the hierarchical clustering analyses (Fig. 4a). Two genotypes which were ‘1/1 Resistant’ in the transmission experiments had median relative risks less than one when outplanted, indicating a lower risk of developing SCTLD on the reef. The one intermediately susceptible WGS-genotype represented in the targeted outplant study, WGS-58, also had a median relative risk less than one, while its credible interval overlapped with several more at-risk WGS-genotypes, thus its resistance level was intermediate both in situ and ex situ. Three out of the five WGS-genotypes categorized as highly susceptible that were present in the targeted outplant study had median relative risks greater than one, indicating a higher risk of developing SCTLD on the reef. WGS-152 was significantly more at risk to developing SCTLD on the reef. Two WGS-genotypes had conflicting resistance between lab and field studies. WGS-154 outplants were significantly less at risk of developing SCTLD on the reef, but were determined to be highly susceptible in the transmission experiments. WGS-12 had a median relative risk less than one, but its credible interval extended greater than one, indicating a risk similar to the intermediate genet, WGS-58, on the reef. In general, these results suggest high levels of convergence between lab- (ex situ) and field-based (in situ) studies of O. faveolata SCTLD resistance. Mote Orbicella faveolata outplant SCTLD susceptibility as determined by relative risk of SCTLD in the targeted outplant study and general outplant percent survival in relation to the experimentally derived resistance groups. (a) In the caterpillar plot of the Bayesian relative risk analysis of the WGS-genotypes in the targeted field study, points are median relative risk values and lines denote the 95% credible intervals (log scale). Credible intervals entirely above a relative risk of 1 (vertical black line at 100) indicate a significant increase in disease risk after exposure to the risk. WGS-Genotypes are colored by their resistance group determined by the hierarchical clustering analysis of the three disease phenotype metrics. (b) Boxplots summarize the average survival of WGS-genotypes outplanted 07/11/2018-11/16/2022 and monitored 07/17/2019-12/16/2022, grouped by their identified experimental resistance group. The dashed line is the median percent survival of all outplants (80%) and the solid line is the mean percent survival of all outplants, 63.9%. (c) Image of a typical outplant cluster of five O. faveolata fragments, each on ~ 3 cm diameter plugs, attached to the reef substrate with cement. In addition to the aforementioned targeted study designed to assess outplant SCTLD risk as the disease front was invading the Lower Florida Keys, we also evaluated survival of O. faveolata outplanted as part of Mote’s science-based coral restoration program. There were 23 WGS-genotypes outplanted between July 2018 and November 2022 and last monitored between July 2019 and December 2022, the time period when SCTLD was prevalent and spreading throughout the Lower Florida Keys. Survival of these 23 WGS-genotypes was assessed in relation to the experimentally determined resistance groups (Fig. 4b). There was substantial variation in the survival of outplanted corals among WGS-genotypes, and in many cases, there was no congruence with resistance groupings determined by the transmission experiments (Fig. S6). For example, highly susceptible genotypes had 100% survival in the field in some cases, whereas resistant genotypes at times had some of the lowest survival rates, near 10% (Fig. S6). When these data were averaged by resistance group, there was no strong indication that SCTLD susceptibility was driving outplant survival rates (Fig. 4b). However, the median outplant survival of the intermediate and resistant groups were above the median overall outplant survival (80%, Fig. 4b), suggesting that SCTLD resistance may have had some impact on Mote’s outplant survival during July 2018 to December 2022 in the Lower Florida Keys, but further study is warranted. Discussion Conducting likely the largest stony coral tissue loss disease (SCTLD) ex situ transmission experiment to date, we quantitatively determined that there are disease-resistant genotypes of Orbicella faveolata within Mote Marine Laboratory’s restoration stock. However, the majority of the 154 multi-locus genotypes (“WGS-genotypes”) tested, 71.4%, were identified as susceptible or highly susceptible to the disease. Experimentally determined resistance levels were consistent with observations of SCTLD on outplants of the same genotypes in the Lower Florida Keys when the disease was first invading the region. SCTLD resistance likely had some impact on overall Mote outplant survival from 2018 to 2023, but survival was highly variable within the resistance group most represented in the dataset, highly susceptible, indicating that myriad factors beyond disease resistance alone contribute to outplant success. Based on our results, disease resistance traits have begun to be operationalized to target the propagation of known SCTLD-resistant genotypes for large-scale production and outplanting. We determined low levels of inherent SCTLD resistance in Mote’s stock of O. faveolata. A total of 12 WGS-genotypes out of the 154 tested were resistant to SCTLD, each having had two out of three exposed replicates remain healthy. An additional 14 WGS-genotypes which only had one replicate exposed, and remained healthy may also be considered resistant. This level of disease resistance within a population (~ 17%) is similar to outcomes associated with population-level resistance to other coral diseases25,26. Lower resistance of O. faveolata to SCTLD was observed in an ex situ experiment conducted by Eaton et al.12, which exposed multiple fragments of ten O. faveolata genotypes, also sourced from the Lower Florida Keys, to SCTLD, and only 10% did not have an increased risk of disease after exposure. However, there were an order of magnitude more genetic individuals tested for SCTLD-resistance in this study relative to Eaton et al.12. As disease dynamics and resultant host susceptibility are a complex web of interacting factors from the environment and pathogen load to the microbiome and host genetic variation7, there are many possible reasons for the much higher susceptibility level observed in our study. Further, our results support the need to consider intraspecific variation in disease transmission experiments to start disentangling that complex web. When comparing the susceptibility groupings from the experimental results to the same genotypes outplanted within the invasion and epidemic zone as SCTLD moved through the Lower Florida Keys, the susceptibility of these genotypes showed comparatively similar outcomes. These results suggest that the ex situ experimental results, which likely included a higher dose of SCTLD disease agents due to the enclosed tank conditions during the exposure, potentially mimicked field-based, ‘real-world’ scenarios during the invasion and epidemic periods of the disease outbreak. Although the prevalence of disease was approximately six times higher in the ex situ study (85.3%) than for the outplants (12.3%), the proportion of resistant genotypes when outplanted (37.5%) was only about two times higher than the proportion found resistant in the experiment (17%). Prevalence of SCTLD on the outplants monitored here was lower than previously observed for this species on Florida’s Coral Reef10,27. This could be due to the potential influence of colony size on SCTLD susceptibility, as larger colonies appear more susceptible to this disease10,28,29 and several other coral diseases17,30. Alternatively, when comparing the overall susceptibility of genotypes to SCTLD with long-term outplant survival from 2018 to 2023 in the Lower Florida Keys, there did not appear to be a strong association. Therefore, drivers of SCTLD-resistance may not be long-term or even short-term indicators of survival in outplanted O. faveolata, at least not within the Lower Florida Keys. Within this outplant survival dataset, there was no way to control the number of replicates among resistance groups, and the outplanting events occurred over several years and within numerous different habitat types, both of which are known variables contributing to outplant survival31. Additionally, almost half of the 23 represented WGS-genotypes were those determined to be highly susceptible to SCTLD. Thus, it is not surprising that there was field-based variability in outplant survival within lab-based susceptibility assignments. The median survival of outplants in the intermediate and resistant groups were higher than the overall outplant median survival, suggesting that SCTLD resistance may have played a role in long-term outplant success. However, the validity of this result is obscured by the combination of uneven observations among resistance groups and high variability in outplant conditions over the five-year period. Resilience to multiple stressors, not just disease, will likely drive long-term outplant survival, and should be considered in future studies. Perhaps the more important conclusion from these long-term outplant survival results is the need to better manage the balance of genetic diversity and advantageous traits present in outplanted populations, as higher diversity generally supports higher resilience by increasing the potential for resilient traits to be present within a population, while genotypes with selected traits may increase resistance to stressors such as disease31–33. The most comparable study to-date of O. faveolata SCTLD susceptibility was conducted on 90 in situ colonies in Southeast Florida and the Lower Florida Keys, from two distinct genetic lineages13. Sampled colonies were chosen based on their SCTLD susceptibility history such that resistant, moderate susceptible, and high susceptible corals would be equally represented13. Klein et al.13 determined that genetic differences did not explain the pattern of observed SCTLD susceptibility within the curated subset of observed wild colonies. If the same number of wild colonies had been randomly observed, we might have expected a higher proportion to be resistant to SCTLD than what we observed in ex situ experiments, as in situ transmission would be affected by additional environmental factors that may reduce susceptibility. The majority of the 154 WGS-genotypes screened here were originally sourced from the Lower Florida Keys or from a 2015 batch cross of Upper Florida Keys parents. Population genetics studies with O. faveolata, particularly in the Florida Keys, suggest low levels of genetic differentiation among populations, as well as sites with evidence of clonality13,34,35. This study also found clone rates to be relatively high (12.6%). While some of these clones were expected due to accidental duplicate sampling of individuals whose metadata were lost over time, we nonetheless predict that a large proportion of our observed clones were in fact true clones, corroborating field-based studies of genetic structure with this species13,34,35. Although we did not explore genetic differences in resistance here, this work is ongoing and will provide additional insight into the association, or lack thereof, of genetic differences and potential other drivers for SCTLD resistance36. Future studies utilizing the samples collected during the experiments reported here will examine the interplay of symbiosis and genotype, as well as the influence of host microbiome and gene expression, on disease susceptibility36. If genetic differences do not explain SCTLD resistance, but other traits do, like algal symbiont genera13, then corals with known SCTLD-resistant traits, not just those with high genetic diversity, should be prioritized for production and outplanting on SCTLD-affected reefs. Coral disease reduces the success of restoration efforts37–39. Thus, disease management by strategic integration of resistant genotypes in outplanting designs has the potential to notably impact restoration efficacy, for example, by reducing the number of disease-susceptible hosts in an outplant plot. Our results show congruence of particular O. faveolata genotypes showing high levels of resistance within lab-based (ex situ) experiments and targeted outplant (in situ) studies, highlighting their potential within strategic restoration planning. In addition to identifying disease-resistant individuals, maintaining both genotypic and phenotypic diversity in nursery stocks is a priority for maximizing the adaptive potential of restored coral populations facing multiple stressors beyond disease outbreaks alone32. There has been long-term recruitment failure of O. faveolata in the Caribbean40, likely exacerbated by disease-reduced fecundity in some regions41, so assisted sexual reproduction efforts may be necessary to increase genetic diversity of the remaining populations42,43. Not only should disease-resistant corals be included in outplanting strategies, they should also be utilized within a managed breeding program to promote disease resistance within restored populations, but only if the traits governing disease resistance are actually heritable. To date, not only is that outcome unknown, but the traits governing SCTLD resistance are still unresolved. Although the severity and wildfire-like spread of SCTLD has resulted in a considerable number of studies that have utilized various ‘omics tools to piece apart the complexity of this disease, results have yet to converge on a biomarker for SCTLD resistance44 or the causative pathogen11. Finding a biomarker of SCTLD resistance in massive coral species is the ‘holy grail’ for restoration applications in the Caribbean. Much more is known about biomarkers of white band disease resistance in the primary branching species used in restoration, Acropora cervicornis. Within A. cervicornis, the presence or absence of an intracellular parasite, Aquarickettsia rohweri, is a putative marker for white band disease susceptibility and resistance45. Disease resistance in this species has also been associated with specific gene regions46. A similar, yet more nuanced outcome is expected for O. faveolata resistance to SCTLD. In addition to the results presented here, our study generated over 2,500 samples for determining a biomarker for SCTLD resistance in O. faveolata, the largest sample collection dataset on record for this species. At this point, there could be genetic, algal symbiont13, bacterial47, or gene expression48,49, as well as other processes at work, that could influence SCTLD susceptibility within this species. While a single marker will be hard to decipher, we anticipate finding a pattern or suite of markers within the cohort of metrics taken for gene expression, microbial community, and genetics that could be used as a toolkit for assessing relative SCTLD resistance of individuals within this species36. Once a profile is identified, we can then screen additional corals within propagation or within the reef environment to identify other individuals with similar profiles, which can be used for more informed restoration activities. Our study provides a strong foundation for future research efforts into coral resilience, and supports the integration of resilience background into management strategies to maximize success of adaptive conservation and restoration efforts. Methods Transmission experiments Four consecutive SCTLD transmission experiments were conducted in Mote Marine Laboratory’s Coral Health and Disease wetlab, located in Sarasota, Florida, from January through April 2022. Each transmission experiment exposed up to 100 putative genotypes of Orbicella faveolata directly to either disease or healthy donor coral fragments (Table S1). Disease donor colonies of Montastraea cavernosa or O. faveolata (~ 10–20 cm in diameter) with ~ 30% SCTLD progression were collected from reefs in the Lower Florida Keys. Collections were conducted under Permit #FKNMS-2021-150, and additional details are provided in the Supplemental Material. Three apparently healthy (i.e., not displaying visible disease lesions or other signs of stress) donor M. cavernosa colonies of approximately 20–25 cm in diameter were collected from the Florida Keys National Marine Sanctuary dockside nursery (24.55101, −81.80846) on January 13, 2022, and were used for all four experiments (Permit: FKNMS-2021-150). Samples from each donor colony were collected prior to the start of each experiment for use in histological analyses (detailed in Supplementary Material) to confirm SCTLD following Meiling et al.9 and Studivan et al.50. Experimental fragments of O. faveolata (~ 3 cm diameter attached to ceramic plugs) were sourced from Mote Marine Laboratory’s ex situ nursery at the Elizabeth Moore International Center for Coral Reef Research and Restoration facility in Summerland Key, Florida. Corals were either originally sourced from the Lower Florida Keys or were fully grown sexual recruits from past batch crosses that had been integrated into production for restoration. Putative genotype identities were previously determined using genetic testing (2bRAD or microsatellites) or by tracking fragment source. Fifty 9.5 L aquaria held within three raceways were established as disease-exposed aquaria, and twenty 19 L aquaria in a separate raceway were established as control aquaria. Each disease-exposed aquaria held up to two experimental fragments, and each control aquaria held up to five control-exposed fragments. This design was utilized to maximize the number of genotypes that could be included in each experiment and ensure that the aquaria volume scaled with contained coral biomass, while working within available laboratory space. Flow-through water, sourced from Sarasota Bay, FL, and processed through a high-rate sand filter and ozonation, was provided to each aquaria at a rate of ~ 270 ml per minute through independent taps. Additional experimental aquaria and water quality monitoring details are provided in the Supplemental Material. At the beginning of each experiment, experimental fragments were randomly distributed among the 70 aquaria in four raceways to have one fragment per putative genotype in each treatment. Samples of the donor corals and experimental fragments were collected at several time points throughout each experiment, and the associated methodology is described further in the Supplemental Material. All coral fragments were allowed to acclimate to the experimental system for approximately two weeks prior to exposure to disease or apparently healthy donor colonies. On the day of disease exposure, the disease donor colonies were cut into 2–3 cm wide fragments using a tile saw, each containing an active lesion area. Apparently healthy donor colonies were initially cut into similarly sized fragments before the start of the first experiment. Once donor coral fragments were placed into the experimental aquaria, corals in the disease treatments were placed so that they were directly touching the lesion of the disease donor coral (Fig. 1a). If a disease donor colony lost 100% of its tissue during an experiment or if its lesion appeared to stop progressing, it was replaced with a new fragment. In addition to visual observations of coral health, overhead photographs of the corals were captured daily (Olympus Tough camera). Initial disease signs were noted when a fragment had ~ 5% or less of its surface area affected by tissue loss. Final disease signs were noted when apparent signs of disease, tissue loss, or “jelly-like,” necrotic tissue appeared on ≥ 10% of the fragment (Fig. S7), and it was removed from the experiment. Experiments ended after two consecutive days of no new disease observations. Tissue samples were collected throughout the experiments for additional analyses not presented here (Table S2). All data analyses and visualization were performed in R (version 4.3.3) as described in the associated GitHub repositories (See Data Availability). Whole-genome sequencing DNA extractions of experimental samples (one per putative genotype) and additional samples collected from Mote’s restoration O. faveolata stock were performed using a modified version of the manufacturer’s protocol for the Zymo QuickDNA MiniPrep Plus kit (detailed methods provided in Supplemental Material). One representative extraction of each putative genotype (n = 180; 174 putative genotypes from the transmission experiments and Mote’s restoration stock, including 6 duplicates) was normalized to 500 ng, then sent to Azenta/Genewiz (South Plainfield, NJ) for whole-genome library preparation and sequencing at 10x coverage. WGS sequences were processed using the bioinformatics scripts available in the associated GitHub repository (See Data Availability). Raw sequences were evaluated for read quality using FastQC51, and low-quality reads were removed using Trim-galore52, a wrapper of cutadapt53. Trimmed reads were then mapped to the O. faveolata reference genome23 using bowtie254. Coral-aligned reads were then processed using the Genome Analysis ToolKit (GATK, v4.0)24 following their best practices guide. In brief: (1) individual variant (SNP) calls were first conducted using the function HaplotypeCaller, (2) resulting individual variant call format (vcf) files were combined into a genomics database using the function GenomicsDBImport, (3) then joint genotyping (comparisons among individual variant calls) was conducted using the function GenotypeGVCFs. Raw global SNPs were then hard-filtered based on GATK best practices, as ‘truth’ and ‘training’ datasets were not available for this species to use Variant Quality Score Recalibration (VQSR). True SNPs (excluding insertions/deletions [indels]) were isolated for all samples using the function SelectVariants, then the following filtering steps were applied using the function VariantFiltration: QUAL < 30.0, QD < 2.0, FS > 60.0, SOR > 3.0, MQ < 40.0, MQRankSum<−12.5, ReadPosRankSum<−8.0. Quality scores (QUAL) measure the quality of each variant call. SNPs passing/failing these quality thresholds were visualized, and SNPs passing these quality filters were then exported for downstream analyses using the function SelectVariants. High-quality SNP calls from WGS samples were analyzed using custom R scripts to visualize genetic relatedness as a genetic distance matrix and dendrograms with the packages poppr55 and ggdendro56, respectively. Pairwise genetic distance values were compared for sequenced duplicates to determine an appropriate threshold for recognizing clonal genotypes. The custom clonal threshold was used to identify multi-locus genotype identities (‘true’ distinct genotypes) for all samples. Multi-locus genotype assignments were then compared to Mote’s provenance metadata for validation and QAQC purposes. Experimental susceptibility metrics All putative genotypes exposed in the transmission experiments were reclassified and grouped by their multi-locus genotype assignment (hereafter referred to as “WGS-genotypes”), or retained their original identities if they did not have a sample processed through whole-genome sequencing due to failed extractions or library preparation. Three disease phenotypes were recorded for each individual exposed to SCTLD during the four transmission experiments: the fragment’s final health state (healthy or diseased), the number of days the fragment was exposed before it developed visual disease signs (transmission rate), and the number of days from initial observation of disease signs on a fragment to when it had lost 10% or more surface area to disease and was removed from the experiment (progression rate approximation). These phenotype observations were then standardized and derived to be on a 0 (resistant) to 1 (highly susceptible) scale for the three disease phenotype metrics: susceptibility, transmission, & progression. Susceptibility was calculated as the number of diseased fragments divided by the total number of exposed fragments of a WGS-genotype. Transmission was first calculated individually as one minus the transmission rate divided by the length of the experiment, and was then averaged per WGS-genotype. Progression was first calculated individually as one minus the progression rate approximation divided by the length of the experiment, and then averaged per WGS-genotype. To categorize the WGS-genotypes into resistance groups, we used a hierarchical clustering analysis of the disease phenotype metrics for all 154 WGS-genotypes. The ‘mclust’ package57 was utilized to perform Ward’s hierarchical clustering with a Gower dissimilarity matrix58 to categorize the genotypes into five resistance groups: highly susceptible, susceptible, intermediate, resistant, and 1/1 resistant (any genotype that only had 1 exposed replicate and it remained healthy). The hierarchical clustering was repeated for the 53 WGS-genotypes that had three or more replicates exposed to SCTLD, so only four clusters were used: highly susceptible, susceptible, intermediate, & resistant. Statistical differences in disease phenotype metrics between resistance groups were assessed using either analysis of variance (ANOVA) tests or Kruskal-Wallis tests and the appropriate post-hoc multiple comparison tests, depending on compliance with test assumptions of normality. Outplant susceptibility and survivorship A targeted outplant study was conducted from September 2018 through November 2019 to quantify the impacts of SCTLD on outplanted, microfragmented corals that were previously raised within a land-based nursery (Permits: FKNMS-2017-136 & SAL-18-2048A-SCRP). Within this study, we considered the subset of 415 outplants of Mote O. faveolata genotypes matching eight of the identified WGS-genotypes that were each present at two offshore reefs in the Lower Florida Keys. Five clonal fragments (~ 3 cm diameter plugs) were attached onto coral skeletons or reef substrate in clusters approximately 10–15 cm in diameter by inserting the post of each coral plug into a hole drilled in the dead substrate (Fig. 4c). A small amount of cement was used to attach the coral fragment and its associated ceramic plug flush to the surface of the reef. Once secured, the coral fragments grow over the ceramic plug, forming a new colony permanently affixed to the reef. These corals were outplanted for the explicit task of monitoring for the presence of disease and determining if there was genetic variation in SCLTD susceptibility. Sites were monitored for active disease approximately monthly for the first year post-outplanting, with data collection including documentation of survival/mortality, adherence to the reef, predation, and active disease lesions. This outplant monitoring dataset included the time when SCTLD was progressing through the Lower Florida Keys, where the corals were outplanted, and so captured the susceptibility of genotypes during the invasion and epidemic period of SCTLD at these sites. Mote Marine Laboratory has been outplanting O. faveolata colonies in the Lower Florida Keys since 2013. This species is generally outplanted in clusters of 5–20 plug-sized (~ 3 cm diameter) clonal fragments. In situ monitoring is conducted one month and one year after outplanting, during which the survival of each fragment is recorded. To supplement the targeted outplant study, we assessed survival of the WGS-genotypes outplanted from July 2018 to November 2022 and last monitored from July 2019 to December 2022, e.g. from the start of the SCTLD epidemic in the Lower Florida Keys through when it was prevalent and spreading throughout the region (Permits: FKNMS-2017-136 & SAL-18-2048A-SCRP). The monitoring subset of just WGS-genotypes included 3,301 individual fragments outplanted in 362 clusters/arrays at sixteen reefs in the Lower Florida Keys. This subset was used to compare experimentally-derived resistance groups with overall outplant survival on the reef, where survival was calculated as the percent of fragments per array with surviving tissue when last monitored during the SCTLD epidemic time period. The relationship between resistance groups and outplant survival was not statistically evaluated here due to variability in genotypic representation at sites and time since outplanting, as well as a lack of information on known SCTLD presence. Relative risk analyses Relative risk analyses compare the probability of an event occurring between two groups: those exposed to a risk and those not exposed. We used a hierarchical Bayesian framework to quantify the relative risk of developing SCTLD signs in each of the four transmission experiments. Relative risk was calculated as: Where the risk in exposed individuals was calculated as the number of diseased at the end of an experiment divided by the total number of replicates exposed to SCTLD. The risk in the non-exposed fragments was the incidence of those not exposed to SCTLD, the control corals. This analysis closely followed Muller et al.25, using a Bayesian approach to calculate the posterior distribution of the relative risk, which was estimated using a binomial likelihood distribution and a uniform-Beta prior distribution. To assess the risk of developing SCTLD signs when exposed to the different donor colonies during the experiments, or when outplanted on the reef, we adapted the relative risk analysis conducted by van Woesik and Randall59, which assessed disease risk relative to a larger population – the case when there are no “non-exposed” individuals. We calculated the standardized expected ratio as: . Where i was a given subgroup – fragments exposed to each donor colony or WGS-genotypes in the targeted outplant study – and ni was the number of diseased individuals in that subgroup. We then used a negative binomial distribution for relative risk and a gamma distribution to account for variance. All Bayesian relative risk models were run with uninformative priors using 3000 Markov chain Monte Carlo simulations in the ‘R2OpenBugs’ package60 to obtain the posterior probability distributions. We then calculated 95% credible intervals for relative risk estimates. Credible intervals that did not overlap one were considered significant, and those that were higher than one indicated a higher disease risk. Three relative risk analyses were performed. Relative risks of developing SCTLD within each experiment were compared to determine if disease risk was comparable among the four experiments conducted in 2022. Relative risks of developing SCTLD when exposed to specific donor colonies were also compared among disease and control donor colonies, as well as with the histology results. Relative risks of SCTLD in the outplant study were compared with the resistance groups determined by the hierarchical clustering. This comparison was used to ground-truth levels of risk captured within the lab-based transmission studies with field-based, ‘real world’ outplant scenarios. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements We would like to thank Erich Bartels for his role in fieldwork management & monitoring efforts for the outplant data used in this study, Kari Imhof, Nick McMahon, Cody Engelsma, Dakotah Merck, Emmanuel Perry, Dick Rosenbloom, & Mike Williams for their assistance with coral care and experiments, and Joe Kuehl for his assistance with diseased coral collections. Funding was provided by Florida’s Department of Environmental Protection through the Coral Protection, Restoration Grant Award No. C2002 & Award No. C1FB43, and FWC Contract No. 17176. This research was carried out in part under the auspices of the Cooperative Institute for Marine and Atmospheric Studies, a cooperative institute of the University of Miami and the National Oceanic and Atmospheric Administration (NOAA), cooperative agreement NA 20OAR4320472. Author contributions SS, SW, & EM managed and conducted the four transmission experiments and sampling. AR & DH processed and analyzed histology samples. ZC, SH, JM, & CW conducted the targeted outplant study. SH, ZC, & CW managed outplanting and monitoring from 2018 to 2023. MS & BY processed the genetics samples and conducted the WGS analysis. SW conducted statistical analyses and visualization. SW & MS drafted the manuscript. All authors contributed to project development and manuscript edits. Data availability Disease resistance and susceptibility analysis scripts are available on GitHub ( [https://github.com/saradwms/OFAV\_SCTLD\_Resistance](https:/github.com/saradwms/OFAV_SCTLD_Resistance) ). WGS raw sequences and associated metadata are available at the NCBI under project number PRJNA1123826, and bioinformatics pipelines and analysis scripts are available on GitHub ( [https://github.com/mstudiva/SCTLD-resistance-genomics](https:/github.com/mstudiva/SCTLD-resistance-genomics) ). All metadata generated from this project are available on the SCTLD DataOne Portal ( [doi:10.48522/D34W2M](https:/doi.org/10.48522/D34W2M) ). Declarations Competing interests The authors declare that a coauthor on this study, Michael Studivan, is an editor for the Coral Reef Conservation and Restoration collection. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement Disease resistance and susceptibility analysis scripts are available on GitHub ( [https://github.com/saradwms/OFAV\_SCTLD\_Resistance](https:/github.com/saradwms/OFAV_SCTLD_Resistance) ). WGS raw sequences and associated metadata are available at the NCBI under project number PRJNA1123826, and bioinformatics pipelines and analysis scripts are available on GitHub ( [https://github.com/mstudiva/SCTLD-resistance-genomics](https:/github.com/mstudiva/SCTLD-resistance-genomics) ). All metadata generated from this project are available on the SCTLD DataOne Portal ( [doi:10.48522/D34W2M](https:/doi.org/10.48522/D34W2M) ).