Cold water and harmful algal blooms linked to coral reef collapse in the Eastern Tropical Pacific

Study system

The region of Papagayo is under the influence of wind-driven coastal upwelling particularly during the dry season (December–April). However, off-season upwellings are common and seawater conditions are similar to those during the upwelling season at least for a few days (Rixen, Jiménez & Cortés, 2012). The strong seasonality is clearly manifested in the mean (±SD) values during the seasons (dry and wet season). For seawater parameters in the study area the following has been recorded: temperature (22.2 ± 1.7 and 30.08 ± 0.27 °C, 1.5 m depth), pH (7.90 ± 0.12 and 8.02 ± 0.03), salinity (34.9 ± 0.8 and 32.7 ± 1.5 psu), and chlorophyll-a (1.19 ± 0.66 and 0.43 ± 2.88 mg m⁻³), (Alfaro et al., 2012; Jiménez et al., 2010; Sánchez-Noguera et al., 2018; Saravia-Arguedas et al., 2021). Physiological processes, such as calcification and respiration, of corals and other calcifying organisms are affected by these seasonal changes (Rixen, Jiménez & Cortés, 2012; Sánchez-Noguera et al., 2018).

Papagayo is the coastal region in the north-western Pacific of Costa Rica that until recently exhibited profuse coral communities and numerous small and discontinuous reefs in sheltered areas of islands, islets and coves (Cortés & Jimenez, 2003; Glynn et al., 2017; Jiménez et al., 2010). Most of the reef building was done by massive colonies of Pavona spp., Gardineroseris planulata and branching species of Pocillopora. The latter were responsible for significant framework accretion in a few shallow (<8 m depth) places along the Papagayo coast (Cortés & Jimenez, 2003; Glynn, Druffel & Dunbar, 1983; Toth, Macintyre & Aronson, 2017).

The strong seasonality of the area and the absence of major rivers, and, hence, year-round fresh water input producing low salinity conditions and high sediment loads, offered more suitable habitats for coral reef and communities’ development than other localities in the Pacific of Costa Rica (Jiménez et al., 2010) and partially explain the prevalence of two coral genera (Pocillopora and Porites) as the main reef builders in southwestern Pacific Costa Rica (Glynn et al., 2017). The biodiversity of corals in the northern Papagayo area, 22 zooxanthellate species, is the highest of mainland Costa Rica (last tally in Glynn et al. (2017)). Seventeen of those species are found in the study area of Islas Murciélago, making it one of the most important areas of the country in terms of biodiversity of coral and coral reef-associated species, and coral habitats.

Coral reef monitoring

Coral reef surveys were conducted at Islas Murciélago, Área de Conservación Guanacaste’s (ACG) Marine Protected Area, Papagayo, Costa Rica (10°N 86°W; Fig. 1) between 1994 and 2019. Ten-metre line intercept transects were used to measure live coral cover (to the nearest 5 cm) and record coral genera at six sites; Isla San Pedrito, Cocinera, Lara, Golondrina, Dos Emes and Pelada (Fig. 1). In total, 732 transects were conducted between three and 12 m deep in replicates of three to five (Table S1). Monthly sea surface temperatures (SST) from 1982 to 2019 were obtained from NOAA OI.v2 (Reynolds et al., 2002) for Papagayo (1 × 1 grid centred 10°N 86°W), and upper and lower coral bleaching thresholds were defined as 1 °C above or below the mean monthly SST (Donner et al., 2005). Harmful algal bloom (HABs) observations in the Papagayo region were recorded by CJ and MMCH.

Fieldwork were approved by the Área de Conservación Guanacaste (project number FOI-004-001).

Data analysis

Statistical analysis was conducted in R v3.6.3 (Team, 2020) (Supplemental Material). We used linear mixed models (LMMs) in R package lme4 (Bates et al., 2015) to model coral cover dynamics over time at each site. Logit-transformed proportion coral cover was fitted as a response variable with a Gaussian error structure (Warton & Hui, 2011), and year as a predictor. We included depth and transect ID as random intercept terms, allowing us to marginalise the effects of each when estimating the average effect of year for each site. We evaluated support for both linear and quadratic effects of year by comparing models with the AIC. Models containing the quadratic term for year had highest support in the data for all sites (Table S2A). We generated posterior mean predictions and 95% credible intervals for each site-specific best-supported model in the R package brms (Bürkner, 2017, 2018) (Table S2B).

To examine the drivers of coral declines, we fitted bivariate response mixed effects models in brms. Each model contained a 2-column response of a SST variable (minimum or maximum) and logit-transformed mean coral cover per year-site combination, and linear year as a predictor variable. These models allow us to control for temporal trends in each response whilst simultaneously estimating the posterior correlation among the residuals from each model. Therefore, if there is a true association relationship between SST and coral cover, positive residuals in one model should drive a consistent directional relationship in the residuals of the other, inducing a correlation (see Houslay & Wilson, 2017). If maximum SST drives coral declines, a negative correlation among residuals would be expected (positive residuals representing above average SST drive negative residuals in the coral model representing greater than average loss of coral cover). The converse would be the case if minimum SST drives coral loss, with a positive correlation expected. We tested for this relationship between coral cover and current year SST_max and SST_min, as well as lagged models incorporating SST data from the year prior to coral cover surveys, referred to as lag+1 model. We evaluated the significance of the posterior residual correlations based on whether the credible intervals for the estimate crossed zero.

Drivers of mass coral die-off

The cold-water upwelling during La Niña of 2008 to 2009 and a widespread HAB’s event drove a crash in coral cover at Islas Murciélago in ACG. Such an extreme upwelling had not occurred since 2000. During upwellings, rapid temperature drops of up to 10 °C (Rixen, Jiménez & Cortés, 2012) potentially induces coral bleaching and mortality through cold-water shock (Muscatine, Grossman & Doino, 1991), as well as hypoxia, increased acidity and low aragonite saturation state (Sánchez-Noguera et al., 2018). While Islas Murciélago’s coral communities are locally adapted to seasonal upwellings (Stuhldreier et al., 2015a, 2015b), this did not confer reef resilience to the 2009 extreme event, which appeared to breach tolerance thresholds for the majority of corals (Palmer, 2018a). HABs are no longer an infrequent natural occurrence fuelled by the increased primary productivity of cold water upwellings, but a direct result of poor wastewater management, with devastating impacts to marine ecosystems and the fishing communities that depend on them (Lapointe et al., 2019; Paerl & Scott, 2010). The Costa Rican coastal current predominantly brings water from the highly developed southern region to the undeveloped, national park coastline of ACG’s MPA. Likely, the synergy between persistant HABs and the 2009 extreme upwelling caused a devastating impact on Islas Murciélago’s coral communities (Lapointe et al., 2019; Paerl & Scott, 2010).

The lack of relationship between coral declines and SST maxima at Islas Murciélago, suggests corals are able to physiologically tolerate warmer water (Palmer, 2018b). Such physiological tolerance would likely confer ecosystem resilience to the increasing threat of elevated SST (Romero-Torres et al., 2020), provide refugia (Freeman, 2015) and is highly relevant for research into increasing coral warmer water tolerance (e.g., (Anthony et al., 2017; Oppen et al., 2017; Camp, Schoepf & Suggett, 2018)). While levels of coral immunity relate directly to tolerance (Palmer, 2018a; Palmer, Bythell & Willis, 2010), the mechanisms of Islas Murciélago coral physiological tolerance have not been explored. A type II error is also a possibility, with Papagayo mean monthly SST breaching upper thermal bleaching thresholds in 2012 and 2015, when surveys were not conducted. Notably, however, Papagayo did not experience a breach in upper thermal bleaching threshold in several years reported as global SST maxima, including 2005 to 2006, 2009 to 2010 and 2016 (Hughes et al., 2018a), suggesting that local oceanographic and/or atmospheric buffering has protected reefs from marine heatwaves (Jiménez, 2002).

Coral decline and recovery rates among sites

The rate of coral declines at Islas Murciélago varied among the six reef sites, with those dominated by the highly susceptible coral genus, Pocillopora (Palmer, Bythell & Willis, 2010) declining faster and recovering less. This primary loss of the most susceptible coral genera (Côté & Darling, 2010), drove a composition shift to massive and encrusting Pavona and Psammocora spp., at San Pedrito, Lara and Dos Emes, consistent with other reefs during extreme events (Knowlton, 2001). Previous studies have documented significantly lower levels of constituent immunity of fast growing (e.g. Pocillopora spp) compared to slower growing massive corals such as Porites spp (Palmer, Bythell & Willis, 2010), which tend to invest more in immunity and seem to be more resistant to disease and bleaching given that they are longer lived (Palmer, 2018a). The reduction of benthic structural complexity provided by coral habitats (e.g., Pocillopora dominated communities and reefs), translates into a loss of potential habitats for associated species and changes in ecological processes affecting the entire ecosystem. Important ecological changes have been documented in the area of Papagayo (Bahia Culebra) after drastic deterioration of coral habitats during the last decade: a reduction of ca. 40% of reef fish orders (Arias-Godínez et al., 2019) changes in the trophic structure of reef fish assemblages (decline of fish mesopredators; Arias-Godínez et al., 2019), and drastic changes in the composition of the biodiversity associated with Pocillopora colonies (obligate commensal species replaced by boring and opportunistic or facultative species Salas-Moya et al., 2021).

Such high and non-random coral loss (Côté & Darling, 2010) from all sites is consistent with the low ecosystem resilience reported for low diversity reefs (McClanahan et al., 2012). Similarly, the lack of recovery at five sites indicates that the 2009 die-off event pushed ACG’s coral communities beyond their tipping point, preventing significant recovery and causing system collapse (Nyström et al., 2008).

Coral recovery was hampered by local Pocillopora extirpations, which likely reduced larval and fragmented propagule supply required for recruitment (Ayre & Miller, 2004). Furthermore, with acidic upwelled water promoting rapid skeletal dissolution and sea urchin dominance causing intensive bio-erosion (Alvarado et al., 2016), dead coral is rapidly reduced to unconsolidated rubble beds, unsuitable for coral recruitment (Fordyce et al., 2019). In contrast, Pelada regained pre-decline coral cover by 2017, including a low proportion of Pocillopora, which may have been enabled by the higher proportion of more tolerant massive corals and site exposure to larvae-carrying ocean currents. Such recovery suggests that, with careful management, ACG’s reefs could recover and continue to support ecological and societal ecosystem services through escalating climatic changes.

Resilience-based management for coral reef restoration

The exploration of ACG’s coral reef dynamics provides locally-relevant research outcomes, beneficial for implementing RBM (Mcleod et al., 2019) and which refute the generic narrative that marine heatwaves are the primary driver of declines. Conservation priorities of ACG’s marginal reef system should target preventing further coral loss, aiding its recovery and promoting ecological and social resilience (Mcleod et al., 2019). To prevent further coral loss, improving the regions wastewater management and coastal land use is vital for reducing the frequency and magnitude of HABs. ACG’s limited coral recovery highlights a clear need for coral restoration, as well as substrate consolidation. While restoring reefs with fast-growing Pocillopora would facilitate rapid coral cover recovery, the impact of cold events on Pocillopora-dominated reefs demonstrates their limited resilience. Slower growing, more tolerant massive coral species (Palmer, Bythell & Willis, 2010) should therefore be included in restoration efforts.

The inevitability of increasing climate-associated extreme events, and the uncertainty of whether these may include warming, cooling, increased hypoxia and acidity (Bakun et al., 2015), supports the need to explore experimental conservation approaches—with any risk dwarfed by the greater risk of not acting (Mcleod et al., 2019). With unknown, multiple and/or variable perturbations affecting corals, beyond marine heatwaves, determining how coral stays healthy, maintains or re-establishes homeostasis and survives a variety of challenges is key for effective conservation (Palmer, 2018a). As such, understanding coral immunology is vital, with the potential recovery, such as re-establishing homeostasis of the meta-organism (holobiont) under new, or perturbed, conditions (Palmer, 2018a, 2018b) dependent on investment in and efficacy of immune mechanisms (Palmer, 2018a; Palmer, Bythell & Willis, 2010). Understanding such fundamental biological knowledge, beyond tolerance of marine heatwaves, is key for developing optimised conservation strategies that can be adapted across all reef systems.

Like many reefs globally, ACG’s coral communities have been pushed beyond their tipping point and are unlikely to recover without proactive, locally targeted conservation interventions. The importance of understanding coral tolerance biology to a range of impacts is highlighted by cold water as the primary driver of ACG’s coral loss and underscored by the uncertainty of future climate perturbations. Locally targeted, evidence-driven RBM approaches that incorporate immunology into coral restoration presents a more viable and globally scalable strategy for conserving coral reefs.