Coral reefs are highly dynamic ecosystems. Our findings show that the effects of disturbances largely depend on the condition of the system before the impact, disturbance severity, the cumulative effect of previous impacts, and biogeographic region, with the latter reflecting differences among life-history traits of geographically distinct coral assemblages, such as those between the Tropical Atlantic and Indo-Pacific reefs (e.g., 24). Reefs with higher initial coral cover underwent the most severe losses in coral cover, regardless of the disturbance type. Despite the numerous sources of stress, both natural and anthropogenic, our findings confirm that thermal stress is the leading driver of change in coral reefs, as it causes high coral cover declines that do not depend on the previous condition of a reef and because bleaching can occur over very large spatial scales (25; Table S1). However, we have also shown that losses in coral cover after a bleaching event are modulated by the cumulative stress of previous impacts, regardless of the bleaching intensity or initial coral cover, which suggests the presence of an ecological memory of thermal stress in the entire system. On the contrary, the effect of a cyclone (and likely other physical impacts) is primarily modulated by the initial condition of a reef. Moreover, the system does not seem to develop a predisposition to better adapt to future disturbance events in the short-term. Lastly, our findings show that in the absence of disturbance, almost 80% of coral reefs were capable of recovery.
Pre-disturbance coral cover is a good predictor of the ability of a reef to resist disturbance, with reefs with lower levels of coral cover tending to be more resistant than reefs with higher levels of coral cover (10, 35; Fig. S3). This was confirmed by the observed differences in coral cover trends between the Indo-Pacific and tropical Atlantic. In the Indo-Pacific, initial coral cover was considerably higher (32.4% vs 16.9% in the tropical Atlantic; Fig. S3B) and consequently was considerably more affected by bleaching events and cyclones, although not by disease, which are more pervasive in the Caribbean region (Fig. 2; 14). Furthermore, the cumulative effect analysis revealed consecutive losses in coral cover in the Indo-Pacific after the first and second impacts of climate-driven disturbances, which were not observed in the reefs of the tropical Atlantic (Fig. 4C). This trend is likely due to two circumstances. First, most of our data for the tropical Atlantic were collected beginning in the late 1990s (Fig. S2) when most reefs were already severely degraded (27). Therefore, there was already little left in this region to be affected by disturbance. Second, regimes of chronic and acute disturbances in many sites have resulted in non-random losses of coral species, and many of these impaired systems are now dominated by opportunistic species (28, 29). Although these species are more likely to resist stressful conditions, they also tend to lock ecosystems into low functional states (20, 21, 24). Overall, our findings suggests that Caribbean reefs are not more resistant than those in the Indo-Pacific but rather are more degraded and thus have less coral cover to lose.
In addition to reef condition, we have shown that cumulative effects and the disturbance type determine the trajectory of coral cover following future impacts. For thermal stress, we found that the greatest losses in coral cover occurred during the first impact; furthermore, no significant changes in coral cover were observed after the second impact regardless of the intensity of the subsequent events (Fig. 5). Our findings are consistent with recent reports that show that coral communities can acclimatize to increasing levels of thermal stress (17, 23), which can be explained by the potential ability of corals to enhance their thermal tolerance through multiple mechanisms (30, 31). For example, shifts in gene expression enable corals to acclimate to thermal stress (22, 32). Such acclimatization processes can be anticipated at community scales and may vary among reef zones (33) or even regions such as the reefs of the eastern tropical Pacific, which have shown to be highly resilient after encountering acute thermal stress (34). Moreover, some studies have found that reefs that experience more infrequent bleaching events are more susceptible to coral loss than reefs that experience more frequent disturbances (23, 35). This suggests that the ecological memory of past disturbances in ecological systems fades over time (16, 23). Thus, we must consider that the probability of another disturbance of the same type occurring diminishes as the disturbance intensity increases (16, 36).
Despite the ecological memory generated after extreme environmental events, an increase in the frequency and intensity of thermal stress events may disrupt future ecological processes in coral reefs. Although coral reefs did not show significant losses after the second or third impacts of thermal stress (Fig. 4), this loss pattern may be driven by some species being capable of resisting and adapting to more extreme disturbances while an elevated frequency and intensity of thermal stress events will compromise species with limited abilities to escape warming, which is likely to increase the intrinsic vulnerability of reefs and possibly lead to local extirpations (37). Thus, shifts in the composition of coral assemblages may jeopardize ecosystem functioning (6, 38). Furthermore, coral species that are vulnerable due to thermal stress can exhibit altered metabolic processes and physiological functioning that affect their ability to accumulate calcium carbonate (39) as well as their reproductive output (40), which can dampen coral recruitment and compromise recovery processes in reef systems (36, 41).
Contrary to what was observed with multiple thermal stress events, changes in coral cover following the cumulative impact of tropical cyclones (and potentially other physical disturbances) hardly reflected any potential mechanisms of acclimatization or adaptation within reefs. This might be explained by the fact that the effects of cyclones on coral communities largely depend on the life-history traits, biophysical features, and initial cover of the coral assemblages prior to disturbance (42, 35; Figs. 3 & S3). Although our database did not allow us to consider the identity of species and their susceptibility to disturbance, ample evidence has shown that morpho-functional traits and life-history strategies are related to the ways in which coral species respond to disturbance (21). These factors are particularly relevant for our findings in the tropical Atlantic, in which the effects of cyclones were relatively small (Figs. 2 & 4). Most of our data from this region were representative of times in which most reefs had already undergone several ecological and structural changes (43), and some of the most fragile species, such as the branching acroporids, already exhibited very low abundance values across the region (44, 45). Thus, the tropical Atlantic reefs included in our analysis are most likely largely represented by shifted coral assemblages that are defined by low-coral cover and dominated by low-relief species (21, 46), which are less susceptible to physical damage. On the contrary, the Indo-Pacific reefs were composed of a wide variety of branching, tabular, and digitate morphologies, which are prone to breakage (47). Consequently, the high coral cover of those reefs tends to be in greater danger of decline due to breakage after the passing of a tropical cyclone when compared to the danger present in Caribbean corals (Fig. 2).
Our findings show that coral reefs are still able to recover in the absence of disturbance (Fig S5). Although we did not compare recovery processes in different regions (given the low sample size), the evidence suggests that the coral reefs of the Indo-Pacific have a much greater potential to recover than those in the Caribbean (48), which can be explained by various processes. For example, marked recovery following disturbances, even to pre-disturbance levels of coral cover, has been observed in the Indo-Pacific (49–51). In contrast, Caribbean coral recovery has been only reported in a few areas and is influenced by management strategies (52). In part, this is due to the higher recruitment rates of the Indo-Pacific compared to those of the tropical Atlantic, which promote recovery and are driven by the diversity of coral morphologies in the region that promote recovery of key functions (53). In contrast, the lack of recruitment of key reef framework corals in the Caribbean may partially explain the functional lockdown that has been observed in the coral reefs of the region (21, 53). Reef recovery can also drive species reconfigurations within reefs, which may either positively or negatively affect reef functionality. For example, in many reefs worldwide, reef recovery has been driven by low-relief opportunistic species and other foliose or digitate corals, which can contribute to the three-dimensional complexity of the reef at fine-scales (21, 54). However, these species are susceptible to breakage due to physical disturbances, which may compromise reef functionality after future disturbance events (42). Furthermore, as marine heatwaves become more frequent and intense, larger colonies and the most fecund corals are likely to become increasingly vulnerable, which may reduce the capacity of coral reefs to recover after stressful events (54).
Reversing reef degradation will require a reduction in human pressure, yet the lack of action to reduce anthropogenic impacts and the omission of immediate global action regarding greenhouse gas emissions will exacerbate reef degradation and erode resilience. In light of this, understanding the relative importance of various risks is necessary to properly guide reef management efforts. Our findings contribute to the broader understanding of how disturbance shapes coral reef ecosystems and allow for management actions to be adequately prioritized to mitigate severely stressful events. If the signs or precursors of rapid ecological change can be reliably detected, managers may be able to take early preventative action (55). Furthermore, we expect our findings to aid reef managers in making financial assessments and taking decisions regarding where and how to invest scarce resources to ensure that coral reefs are restored and protected (56). This is particularly relevant given the growing spatial mismatch between the scales of potential threats and planned responses (57). Risk financing and insurance are becoming increasingly relevant (58), as these tools are critical when absorbing financial losses in the wake of disasters and natural catastrophes, thus significant opportunities for restoration may lie in risk financing associated with risk reduction services for reefs (59). In this context, a clear message from our study is that special attention should be allocated to reefs with the highest live coral cover, as these are likely the be more vulnerable to disturbance. Furthermore, integrating multiple properties of contemporary disturbances, such as their spatial extension, frequency, intensity, and cumulative effects, into our perceptions of coral reef dynamics is the key to improving our understanding of how legacy impacts transform coral reefs worldwide.