Caribbean massive corals not recovering from repeated thermal stress events during 2005–2013

Abstract Massive coral bleaching events associated with high sea surface temperatures are forecast to become more frequent and severe in the future due to climate change. Monitoring colony recovery from bleaching disturbances over multiyear time frames is important for improving predictions of future coral community changes. However, there are currently few multiyear studies describing long‐term outcomes for coral colonies following acute bleaching events. We recorded colony pigmentation and size for bleached and unbleached groups of co‐located conspecifics of three major reef‐building scleractinian corals (Orbicella franksi, Siderastrea siderea, and Stephanocoenia michelini; n = 198 total) in Bocas del Toro, Panama, during the major 2005 bleaching event and then monitored pigmentation status and changes live tissue colony size for 8 years (2005–2013). Corals that were bleached in 2005 demonstrated markedly different response trajectories compared to unbleached colony groups, with extensive live tissue loss for bleached corals of all species following bleaching, with mean live tissue losses per colony 9 months postbleaching of 26.2% (±5.4 SE) for O. franksi, 35.7% (±4.7 SE) for S. michelini, and 11.2% (±3.9 SE) for S. siderea. Two species, O. franksi and S. michelini, later recovered to net positive growth, which continued until a second thermal stress event in 2010. Following this event, all species again lost tissue, with previously unbleached colony species groups experiencing greater declines than conspecific sample groups, which were previously bleached, indicating a possible positive acclimative response. However, despite this beneficial effect for previously bleached corals, all groups experienced substantial net tissue loss between 2005 and 2013, indicating that many important Caribbean reef‐building corals will likely suffer continued tissue loss and may be unable to maintain current benthic coverage when faced with future thermal stress forecast for the region, even with potential benefits from bleaching‐related acclimation.

The long-term effects of bleaching events on coral colony growth and coral reef community structure are still not clear (Baker et al., 2008;Pandolfi et al., 2011). Bleached colonies can suffer relatively immediate bleaching-related mortality (Brandt, 2009;Carilli, Norris, Black, Walsh, & Mcfield, 2009), but also can potentially develop enhanced resilience as a result of acclimation to thermal stress (Buddemeier & Fautin, 1993). The adaptive bleaching hypothesis (ABH) suggests that changes in the coral photosymbiont populations following a bleaching event may allow some corals to re-establish a symbiosis with different strains of endosymbiotic dinoflagellates, resulting in a holobiont better suited to the altered environmental conditions (Brown, Dunne, Phongsuwan, Patchim, & Hawkridge, 2014). This hypothesis does not explicitly state that this will result in enhanced growth or improved survival rates, but opens the possibility of such, as well as the possibility of enhanced resistance to bleaching when faced with repeated thermal stress.
Recent laboratory studies indicate that repopulation of the coral endosymbionts with a different symbiont community following a bleaching and recovery response is necessary to increase heat tolerance for individual colonies (Silverstein, Cunning, & Baker, 2015). In this study, coral colonies that previously bleached were less affected by subsequent thermal stress. In contrast, the same adaptive response was not seen from past exposure or acclimation to warmer temperatures prior to subsequent thermal stress sufficient to potentially cause bleaching. Given that individual coral colonies have the potential for highly stochastic responses (Baird, Bhagooli, Ralph, & Takahashi, 2009;Mydlarz, Mcginty, & Harvell, 2010), even for conspecifics exposed to the same environmental stimuli, the identification of these differences in response, recovery, and future resistance related to bleaching is necessary for estimates of larger population or community aggregate responses.
Our research investigates the different long-term responses of individual coral colonies stemming from thermal stress exposure alone vs. those following expressions of acute visible bleaching, and differs from Silverstein et al. (2015) in that it extends that time frame for response observation to 89 months, and importantly provides validation of this phenomenon in a field setting following a natural bleaching event ( Figure 1). These three species are all are massive, hermatypic, relatively slow-growing corals, and all have similar physiological energetics (i.e., zooxanthellate) and are broadcast reproduction (Szmant, 1986), leading to relatively high genetic connectivity across the Caribbean basin (Nunes, Norris, & Knowlton, 2011). Skeletal extension rates are also roughly similar, with range of 1.4-10.0 mm/year for O. franksi (Huston, 1985), 1.8-3.8 mm/year for S. michelini (Hubbard & Scaturo, 1985), and 2.7-9.3 mm/year for S. siderea (Huston, 1985).
We sought to investigate the long-term massive coral species colony-level responses to the 2005 Caribbean mass bleaching disturbance over a sufficiently long period to demonstrate both initial and delayed mortality from bleaching and/or thermal stress, to allow for recovery of colony function to stable predisturbance rates, and to potentially expose differences in outcome stemming from organismal or symbiotic adaptive responses. We focused on massive type, hermatypic (or reef-building) species, as these are the key ecosystem architects in these systems, creating essential habitat for the reef community. An expanded understanding of the long-term response of corals to specific disturbances, in this case bleaching, is essential F I G U R E 1 Bleaching in 2005 in Bocas del Toro, Panama. This image shows typical bleaching in October 2005 on the fringing reef inside Bahia Almirante, Bocas del Toro. A number of species are seen in this landscape, and also both the manner in which conspecifics (in this image, Orbacella franksi) can exhibit very different individual colony responses to identical thermal stress, as well as how partial bleaching can be seen over single colonies. Both of these phenomena were significant for our analysis for guiding how we structure management and conservation goals.
Our quantification of recovery responses is also central for gauging expectations for recovery and persistence of coral reef ecosystems, which is particularly relevant as the frequency and intensity of some disturbances, including elevated sea surface temperature, are currently changing in response to global climate change (Hoegh-Guldberg & Bruno, 2010), and massive coral bleaching events may soon become much more frequent or effectively continuous (Pandolfi et al., 2011).

| MATERIALS AND METHODS
Making extended repeat in situ time series measurements of coral condition over time for massive-type scleractinian colonies presents several practical and logistical challenges. First, coral colonies have relatively slow growth rates (1-10 cm/year) and require long observation times to record detectable changes (Lough & Cantin, 2014), and second the predominant method for determining growth commonly utilizes potentially destructive methods such as radiographic sclerochronology (DeLong et al., 2013). Radiographic sclerochronology determines coral colony skeletal extension rates by either coring or slicing the coral skeleton, and measuring growth bands either visually or with X-ray methods, utilizing either simple planar or computed tomographic analysis. These techniques can be at a minimum partially destructive, but nonlethal, to larger live colonies but can be wholly destructive of smaller colonies. As our goal was to measure postdisturbance mortality/growth for an extended number of years repeated coring was simply not practical on an annual basis for the small to medium-sized colonies in our sample group, due to the potentially deleterious effect on the coral subjects. Furthermore, we hoped to relate colony growth with observed recovery and status of the surface tissue condition for each of those years, and skeletal analysis does not provide information on surface tissue condition. We therefore recorded planar area of live tissue and extent of bleaching and partial bleaching within that live tissue area through the noninvasive analysis of underwater photographs, taken approximately annually.

| Study site
This study was conducted on a protected fringing reef near Punta Caracol on the western (leeward) side of Isla Colon, Bocas del Toro, Panama (9.363°N, 82.282°W; Figure 2). The reef is located within the Bahia Almirante embayment, an area of extensive but patchy coral cover, with coral development along a slope from the surface to 20 m (Guzman, PaG, Lovelock, & Feller, 2005). Coral species diversity and cover are high for Panamanian reefs, and are typical for well-developed coral reefs in the western Caribbean (Guzman & Guevara, 2001). This site was chosen to represent as much as possible an unaffected inshore reef site, to assess as nearly as possible the impact primarily from the thermal disturbances. However, the embayment does have high environmental variability, as it is heavily influenced by both oceanic water input as well as high freshwater input (Collin, Huber, Macintyre, Ruetzler, & Ruiz, 2009), and water clarity is lower than many reef sites due to higher nutrient and chlorophyll concentrations, contributing to shallower reef development (Kaufmann & Thompson, 2005

| Thermal stress determination
In 2005, the Caribbean basin was subjected to the highest water temperatures recorded to date for this area (Eakin et al., 2010), resulting in extensive bleaching across the basin (Lajeunesse, Smith, Finney, & Oxenford, 2009;Miller et al., 2006;Whelan, Miller, Sanchez, & Patterson, 2007). Thermal stress conditions of lower magnitude were also experienced in 2010 (Guest et al., 2012), as was widespread bleaching. A much more detailed analysis of this methodology and the calculations of thermal stress conditions for Bocas del Toro for both the 2005 and 2010 bleaching events can be found in Neal et al. (2013); for the purpose of this manuscript, we updated this previous record through late 2014, but did not alter the methodology or algorithms.
Local temperature conditions and estimations of coral thermal stress in the study area were assessed using an ongoing local in situ depth-stratified sea temperature time series recorded at three depths Thermal stress conditions are defined as HotSpot values 2.0°C or greater, and the onset of potentially damaging coral bleaching was defined to begin at DHW values of 4°C weeks or greater.

| Measurements of bleaching response and recovery
Conspecific groups of bleached and unbleached individual colonies of the three target species (Orbicella franksi, Siderastrea siderea, and Stephanocoenia michelini; n = 198 total) were selected, tagged, and monitored mortality and growth outcomes on approximate annual intervals for nearly 8 years after the major 2005 thermal stress event.
This time span of observations also fortuitously included a second major thermal stress event in late 2010, allowing for evaluation of response of previously exposed colonies with known bleaching history to repeat bleaching events. Metrics of colony response aimed to elucidate: (1) species-specific rates of partial or total colony mortality following the 2005 thermal anomaly; (2) recovery time for colonies to return to steady state or positive growth of tissue area; (3) differences between species and between bleached and unbleached conspecific groups in postdisturbance tissue growth rates; and (4) tissue loss following the second 2010 thermal stress event. All of these metrics have implication for predicting decadal-scale changes in coral reef community composition.
They were chosen to be characteristic of structure-building species contributing to both the creation of living habitat as well as long-term carbonate reef accretion.
Photographic monitoring was begun in early October 2005 during the mass bleaching event. Colonies were selected opportunistically in the same general area along three depth transects (<4, 7-10, and 10-13 m) while on SCUBA. Colony selection was not entirely random, due to limited numbers of suitable subjects along the depths transects, but effort was made to find a comparably similar number of representative samples of bleached and unbleached colonies for the three species in each transect, and it should be noted that the S. michelini population in situ was predominantly bleached, and the S. siderea population in situ was predominantly unbleached, requiring somewhat wider examination of the area to obtain representative populations for the two bleaching status groups. Effort was made to find proximally located pairs (bleached and unbleached colonies), to minimize the impact of different flow patterns, source water, and other localized environmental variables, with the entire sample located along an approximately 150-m transect on the same reef section. Selected colonies were permanently tagged with white PVC plastic tags with stamped numbers, fastened to dead areas of the substrate with stainless steel nails. These tags were removed and replaced three times with flexible plastic, laser engraved, numbered cattle tags (Allflex, Dallas, TX, USA) over the course of the time series to ensure that the numbers were legible and that tags were not broken or lost. Colonies were photographed immediately after selection and tagging, as the initial time point for the time series. These photographs were not all taken on the same day, due to the practicality of locating, tagging, and imaging numerous colonies, but were all marked and sampled during the acute bleaching period over approximately 2 weeks, near the peak of water temperatures during the 2005 thermal stress event, and are thus considered a single time point for analysis.
All colonies were revisited for additional photography at 6,9,22,34,46,58,71, and 89 months after the initial observations in October 2005. Images were taken over the period with a variety of consumergrade underwater cameras, with later years using a Canon 5D MkII 23 megapixel DSLR, with a Canon EF 17-40 mm lens with dome port with extender fitted to minimize radial distortion (Treibitz, Schechner, Kunz, & Singh, 2012). Underwater lighting was provided as needed with a variety of equipment, in later years using dual Sea&Sea YS250 strobes, and images were color-corrected for spectral water column effects.
Colony photographs were analyzed for projected planar area (in units of cm 2 ) and pigmentation condition of live coral tissue, expressed as a percent of the total live area (with any patches of dead exposed skeleton surrounded by live tissue removed from the live area analysis), following Neal et al. (2015). All photographs included a color and size reference in the image and were analyzed with a custom MATLABbased image segmentation tool (http://vision.ucsd.edu/content/coral-colony-segmentation-and-area-measurement-tools), resulting in planar area measurements of live coral tissue, dead area on the colony, and bleached and partially bleached tissue (as subsets of the live area measurement) . Bleached tissue was defined as tissue exhibiting nearly completely white areas, and partially bleached tissue as that with residual pigmentation remaining in the bleached areas (also referred to as paling). These two designations could be, and often were, present on the same colony in complex patterns ( Figure 3).

| Colony sample sets
Two coral colony size classes were defined postanalysis for each species (large and small), with the class division based on the mean size for each species in the sample set. A total of 52 O. franksi colonies were tagged and photographed in 2005, with 37 visibly bleached, and 15 unbleached. Twelve colonies were at a water depth of 4 m or less, 16 at 7-9 m depth, and 24 at 10-12 m. Colonies ranged in planar area from the smallest at 108 cm 2 to the largest at over 3,400 cm 2 , with a mean size of 457.5 cm 2 (±63.4 SE). Thirty-two colonies were classified as large, and 20 as small. In the case of this species only, some large individual colonies (over 1,500 cm 2 ) were spatially subsampled for bleaching state, as measurement of these large colonies in their entirety was not practical with these photographic methods; bleaching extent and growth measurements for these colonies were all included in the large colony group dataset. Twenty-one colonies were in 4 m of water or less, 21 in 7-10 m depth, and 16 in 10-13 m. Colonies ranged in planar area from 39 to 1133 cm 2 , with a mean size of 383.9 cm 2 (±37.5 SE). Twenty-four S. siderea colonies were classified as large and 34 as small.
F I G U R E 3 Segmented live coral tissue areas for a single example colony from 2005 to 2013. (A) Healthy pigmented portions of live tissue area are screened from the color-corrected background images and outlined in green, with partially bleached tissue in blue and bleached areas in red. Some interannual differences in image color are due to water conditions, lighting, or camera sensitivities. (B) Bar plot shows changes in bleaching state across the time frame of the study for this colony. This example colony (Tag #249) did not demonstrate massive tissue loss, recovering quickly from significant bleaching in 2005, but did experience long-term patchy tissue mortality, largely localized to previously bleached areas

| Comparisons of bleaching severity, mortality and recovery
Analysis and discussion of the time series is divided into four time points/sections, defined as: (1) Table 2.
The initial bleaching period is when bleaching was acutely and visibly manifested; the response period is when direct bleaching effects may no longer be visible but are manifested in disturbance-associated tissue loss (whole or partial mortality), whole-colony mortality, and pigment changes. Delineation of the response period was determined post hoc from observed recovery trajectories in this study (i.e., when tissue began to increase in size again following the losses from the disturbance event, and largely recovered visible pigmentation), along with estimations from the literature, including observations of post bleaching zooxanthellate densities recovering within 12 months (Fitt, Spero, Halas, White, & Porter, 1993), significant reduction of visual Bleaching Index for many taxa of massive corals within 9 months (Mcclanahan, 2004), and near complete recovery of pigmentation in Caribbean Montastrea ssp., (now genus Orbacella) within 6 months (Goreau, Mcclanahan, Hayes, & Strong, 2000). The recovery period was thus defined in this study as the first 9 months following the disturbance, and trajectories calculated for this period include three sets of observations and size data, the first being the initial set of images, and two taken 6 and 9 months later. It must be recognized that recovery is not a uniform process, nor is the definition of a "recovered coral", so this temporal delineation for this analysis must be taken as a heu- These final two observations, beginning some 11 months after a second thermal stress event in late 2010, represent the response to this second elevated temperature event, but not the acute short-term response. This event took place approximately 60 months after the 2005 event and was of lesser disturbance magnitude than the 2005 event, but still severe enough to cause widespread bleaching warnings and coral bleaching (Guest et al., 2012). The peak of this disturbance event took place approximately 4-6 weeks after our 2010 imaging trip. There is consequently (and unfortunately) no direct maximum acute bleaching response record for this second event, as we had with the first, but there are before and after photographs for the event, and remotely and locally sensed records of the magnitude of thermal stress. Given the primary desire to observe and quantify long-term colony mortality, the lack of direct observation of maximal bleaching is not a critical issue, but does assume that mortality seen following the thermal stress event is associated with stress (and consequent bleaching). We have made this assumption, but given the multitude of environmental stressors affecting coastal coral reef systems, assigning attribution of effect to a given single environmental cause must be done cautiously.

| Statistical analyses
For comparisons of initial bleaching reaction in 2005, within-species differences in both bleaching and partial bleaching extent between the large and small size groups were evaluated with Student's t tests (α = .05). As there was an a priori division in the selection of bleached and unbleached individuals, no statistical evaluation was made of differences between bleached and partially bleached areas for these groupings, although these means are reported as confirmation that the three unbleached sample sets had little visible effect from the thermal stress, and thus do in fact represent a distinct group of colonies in terms of visible physiological response to thermal stress. Analysis and discussion of the study time frame was divided into these periods, relative to both coral state and acute thermal stress periods. The recovery period was selected post hoc for this analysis and may vary for different regions and species.
Annual change rates for live tissue area for both response and recovery periods are calculated in two ways: First, an annualized % change calculated from linear monthly regression coefficients across multiple observations, and second as annualized % mean change in individual colony planar size calculated from difference between beginning and ending observations for the time periods. There was close agreement between these two methods, and we primarily use the results from the regression analysis for discussion. Recovery time for colonies to return to positive tissue growth was determined by fitting a second-order polynomial regression and calculating the time when slope = 0, and periods for recovery to initial live areas were determined through basic compound interest calculations.  Small 34 11.2 ± 1.9 10.3 ± 1.9

| Thermal stress conditions
Bleaching (BL) and partial bleaching (PB) extent for large and small colonies of each species in 2005. Large colonies were defined as those greater than the mean size for each species and small as those less than the mean. Only S. siderea showed significant differences between the two size classes (in bold), with larger colonies less likely to bleach or partially bleach after a thermal stress event.
changes were more gradual and subtle than for the other two spe-  (Table 1).
It should be noted that for S. siderea our estimate of bleaching extent also explicitly applied only to the selected individuals in the visibly bleached group, and the extent of bleaching across the entire population would be considerably less than the amount reported here as we preferentially searched for affected colonies to sample both bleached and unbleached colonies, and thus our bleaching rates possibly overestimate the extent of bleaching in the natural population.

| Tissue loss and mortality following bleaching
The Siderastrea siderea showed the least amount of tissue loss per colony, with the bleached colonies (n = 25) losing an average of 11.2% (±3.9 SE) of live tissue area per colony in the first nine months following the bleaching event (Table 2). However, unlike the other two species, live tissue area continued a slow but even decline across the remaining period, and did not return to positive growth. The unbleached group (n = 33) did not have significant initial tissue loss in response to the thermal stress event, losing <1% in the first 6 months, but this group subsequently showed a steady loss of tissue area over the remaining time, similar to the bleached coral group. S. siderea had the lowest total colony mortality, with no colonies dying in the 9 months following the bleaching, and only one colony having documented total mortality  (Table 3).
The mean bleached tissue extent of bleached S. siderea colonies was less than the other species and also declined more slowly, from nearly 13.1% (±2.6 SE) in the initial observation to 3.0% (±0.6 SE) by 9 months.
Bleached area in subsequent observations remained low (~1%) and stayed at that level through the second thermal event. Colonies in the unbleached group maintained low areas of both bleached and partially bleached area throughout the time series (Table 3).

| Colony growth following initial bleaching/ stress event
The period from month 22 to month 56 was unaffected by major water temperature anomalies, and it is assumed that there were little or no direct physiological impacts on the corals from thermal stress during this time (Stephenson et al., 2014). This lack of physiological impact is inferred from the temperature record and was not measured in situ. This time period is taken to be indicative of normal environmental conditions and thus potentially represents colony tissue expansion and growth unaffected by thermal stress and is refereed to below as the recovery period.  Includes the total study sample, from both response (between 0 and 9 months after bleaching) and recovery (between 22 and 58 months after bleaching) periods. Totals shown for each species include both the unbleached and bleached colony groups for those species.
growth trajectories for individuals vary, but was not large in the initial years, indicating consistent initial growth and mortality response

| Response to thermal stress varies by species
The extent of bleaching in the initially bleached group for each of the three species varied widely. Recognizing our aim to quantify speciesspecific bleaching recovery dynamics (and not overall taxon-specific bleaching susceptibility-i.e., the extent of bleaching within a representative population), our findings indicate that S. michelini and O. franksi were highly susceptible to severe bleaching, while S. siderea.
appeared highly bleaching resistant. This notable difference in visible response, along with the inherent subjectivity of field identification of "bleached" individuals, brings into question how the widely used designation of "bleached" should be defined for different species and raises the need for more sensitive field methods for the diagnosis of onset of coral stress, such as in situ fluorescence , measurements of cellular apoptosis (Ainsworth, Hoegh-Guldberg, Heron, Skirving, & Leggat, 2008), or concentrations of heat shock proteins (HSP) (Rosic, Pernice, Dove, Dunn, & Hoegh-Guldberg, 2011).
Given that S. siderea continued to show a decline in live tissue extent across the remainder of the study period, the lack of visible bleaching symptoms during the initial stress event may not indicate a resistance to thermal stress-related impacts, but simply a lack of immediate visual response to thermal stress. Similar taxon-specific variable response by corals to thermal stress has been previously demonstrated (Anthony, Kline, Diaz-Pulido, Dove, & Hoegh-Guldberg, 2008;Marshall & Baird, 2000). Many factors other than coral species can confer significant within-and between-species variation in response to thermal stress (Brandt, 2009), including symbiont phylotype (Van Oppen, Baker, Coffroth, & Willis, 2009), size-structured or agerelated differences in a population (Brown et al., 2014), differences in thermal stress with depth (Neal et al., 2013), local stressors such as sedimentation (Carilli et al., 2009), CO 2 exposure (Anthony et al., 2008), or historical exposure to water temperature variation (Carilli, Donner, & Hartmann, 2012). Further species-specific effects may differ between reef sites, as individual colony history, environmental exposure, and benthic and community structure vary.

| Area of partial mortality over time following bleaching was closely related to total initial bleached area
Partial mortality in the 9 months following the 2005 bleaching disturbance also varied widely by species and was closely correlated with the initial amount of bleaching in that species. This suggests that areal bleaching extent in an affected coral colony is a strong predictive indicator for magnitude of subsequent eventual tissue loss. Tissue loss was greatest for S. michelini, followed by O. franksi, and was nearly negligible for S. siderea colonies. Maximum partial mortality was not fully reached in the first 6 months of the study, indicating the importance of following recovery of bleached corals for longer than the time span needed for symbiont repopulation, and suggests a persistent reduction in colony function may be incurred through thermal stress. This delayed response was first reported by Glynn in 1990(Glynn & D'croz, 1990, noting that following the 1982-1983 bleaching event in the eastern Pacific several massive-type corals (Porites panamensis) regained normal visible appearance and coloration within 2-3 months after bleaching, but then experienced total colony mortality between 7 and 10 months.
Our findings indicate the this postbleaching mortality window could be as long as a year for S. micheleni and O. franksi, which occupy similar niches in the Caribbean to Porites spp. in the Pacific, and that the time frame for reduced growth following thermal exposure could be up to 3 years.

| Recovery from bleaching possible in thermally stable conditions
Growth rates during the 4-year recovery period are assumed to represent growth rates for these species under nonstressful temperature conditions. This assumption appears to be potentially valid for S. michelini and O. franksi, but not for S. siderea, which demonstrated a slow but regular decline across the study period. Furthermore, the relatively small growth rates we report here, and the large variance in growth across colonies, point out the difficulty in determining true growth rates for massive-type corals, which must be measured over many years, i.e., longer than the 7 years of this study. Given inherent methodological variance in measuring planar area from photographs  and the absence in the record of some colonies in some years due to tag shedding or missed photographs, the narrowly positive rates we report may not be maximum individual growth rates for these corals, but could alternatively indicate net maintenance of tissue (i.e., a flat growth rate) which may persist for many years. In either case, these even or positive rates for the recovery period do provide a good comparison to the marked losses of the initial and response periods.
The consistent tissue loss for S. siderea remains unexplained and may be related to other factors than temperature. It may be due to senescence from age, competitive interactions or predation, or effects from unmeasured environmental factors such as ocean acidification, sedimentation, pollution, disease, or corallivory preferentially affecting this species. However, senescence or targeted predation appears unlikely given the relatively small sizes of our colonies (39-1133 cm 2 ; mean = 383.9 cm 2 (±37.5 SE)), compared to recorded larger sizes for this species (Lewis, 1997). Furthermore, the regularity of tissue loss across our samples and the rarity of total mortality across the 89 months (only six colonies dead or lost, from n = 58 in 2005) indicate a possible chronic force at work. The steady decline of only S. siderea in this location, for all years, and similar in both bleached and unbleached groups, suggests that it may be unrelated to the temperature anomalies or bleaching event, or may be a synergistic effect from bleaching stress combined with other factors. In particular, ocean acidification (OA) may manifest with such slow but persistent impacts (Shaw, Phinn, Tilbrook, & Steven, 2015). Furthermore, the effect of altered growth or persistence on specific members of the reef community may affect overall net reef deposition (Shaw, Hamylton, & Phinn, 2016); one indication of this is that areas of the northern Florida Reef Tract may already be in a condition of negative net community calcification (Muehllehner et al., 2016). The species-specific differences in postdisturbance recovery documented here increase capacity for improved modeling of growth and calcification dynamics for reef systems.
Stephanocoenia record, but the other three groups showed steady growth in live area.

| Reduced thermotolerance seen in second stress event for previously unbleached corals
The and it is compelling to find this effect also seemingly demonstrated 5 years after the initial exposure. This acclimative effect on previously bleached corals has been described as a possible "nugget of hope" for coral reefs in times of climate change, and our results do provide possible in situ support for the concept that previous bleaching can confer a possible benefit to the short-term survival of individual colonies of some scleractinian species, and thus to the long-term survival of coral reefs as growing, structural ecosystems (Berkelmans & van Oppen, 2006). However, this acclimation must be viewed within a temporal context that allows this benefit to be fully expressed, which may not be the case for our study period, or for the projected future.

| Previously bleached and unbleached corals had similar cumulative mortality when exposed to repeat thermal stress
Another surprising finding supporting the idea that acclimative benefits were limited in their overall impact on coral survival and growth was that the total live area for each of the conspecific groups of previously bleached and previously unbleached colonies were not significantly different after the 89 months of the study. This was true for all three species, despite notable differences in growth response and total area between some of the bleached and unbleached groups throughout this study. This similar tissue decline in tissue area seen by the endpoint of the study is likely attributable in two of the species (S. michelini and O. franksi) to the effects of the second thermal stress event on the previously unbleached groups. This is most clearly seen in O. franksi, where the previously unbleached group had no tissue loss following the initial thermal disturbance and demonstrated steady growth for a number of years immediately following this event, but the tissue area gains accumulated during this period were subsequently completely lost following the 2010 stress event, and this group was not larger than the previously bleached group, which was still making up area lost to direct mortality in 2005. In other words, while the survival of colonies previously exposed to severe high-stress thermal exposure without visible bleaching does incur an immediate clear benefit, as there is less immediate partial colony mortality, but if there is further stress this may be an ephemeral advantage, as they appear at greater risk for larger losses in subsequent bleaching events.
With thermal stress conditions occurring in this ecosystem in 1998 (Glynn, Maté, Baker, & Calderón, 2001(Eakin et al., 2010, 2010 (Levitan, Boudreau, Jara, & Knowlton, 2014), and potentially (at the time of writing) in late 2015, the time between bleaching events could likely be less in the future than the needed recovery periods indicated by these results, meaning that cumulative mortality from multiple events could possibly overwhelm any benefit conferred by either "escape" from bleaching or from individual acclimation from prior bleaching.
The concept that bleaching is exceeding the inherent necessary recovery period for these corals is supported by this study as all three species (in aggregate) ended the time series having experienced net loss of live tissue, regardless of prior bleaching history. This outcome did vary by species, but the living tissue decline for all of these three critical ecosystem-structuring species when faced with repetitive thermal stress suggests that Caribbean reefs may face massive challenges in a warmer future. If thermal stress and bleaching events increase in frequency and severity, and there is no acclimative increase in colony resilience, then the future persistence of massive-type corals in the western Caribbean appears possibly uncertain. While the rates of chronic decline shown in this study are annually still low, they indicate the ecologically chatastrophic possibility of reduction in live coral cover in the Caribbean to functionally collapsed levels within time frames of a few decades. Coral holobiont communities (specifically including associated symbiont communities) do appear to have rapidly formed specific associations conferring greater thermal tolerance in limited areas subjected to rapid and extreme environmental change, such as in the Persian/Arabian Gulf in the Holocene (Hume et al., 2016), but in that case the time frame was ~1-6 thousand years.
While this example is rapid by evolutionary standards, it is not nearly as rapid as current environmental change. Given the possible reduction in both diversity and community extent suggested in this work as the result of repeated thermal stress, existing coral host and symbiont biodiversity and reproductive success may simply not be maintained in these systems over a sufficiently long time frame to allow for stresstolerant associations to evolve and persist.