Sea surface temperatures that were conducive to coral bleaching was experienced at the ARNP, an offshore and well-managed coral reef ecosystem in the Philippines. The impact of this large-scale disturbance at the ARNP varied among taxonomic communities. Benthic communities were more strongly dissimilar among years than among sites, with variation among years attributed to the increase in dead coral and the decrease in coral cover. While the influence of site in benthic communities was comparatively minor, two (i.e., Ego Wall and Mabuti) of the five sites experienced more drastic declines (41–48%) in coral cover over others. The higher coral loss at these sites correlates with their comparatively higher coral cover pre-bleaching. This suggests that small-scale (≤ 3km) variability in the composition of benthic communities influence the impact of bleaching disturbance. In contrast, fish communities were more strongly dissimilar among sites than years. This suggests that despite the large-scale disturbance, there were little changes in the fish communities. The site variation in the fish communities was likely associated with the predominant benthos in each site, which supports a different suite of fishes. Indeed, at two sites (i.e., Aladdin and Shark Airport) with different predominant benthos (i.e., coral in Aladdin and other organisms and turf algae at Shark Airport), the associated fishes were different. Coral-associated damselfishes, cardinalfishes and butterflyfishes were associated with Aladdin, while benthic invertivores and herbivores/detritivores were associated with Shark Airport. Although, the influence of year in fish communities was small, one site in particular (Mabuti) had dissimilar communities in 2019 when compared to 2016 and 2018. This suggests that fish communities at the ARNP were minimally affected by the bleaching disturbance. This minimal effect was perhaps influenced by the availability of alternative and deeper habitats that provided shelter and food for reef fishes, maintaining taxonomic composition. These results suggests that localized protection does not safeguard against bleaching disturbance, but small-scale variation in benthic composition and availability of alternative and/or deeper habitats may alleviate the negative consequences.
Benthic communities at the ARNP varied more strongly among years than among sites. These yearly differences were attributed predominantly to changes in the benthic composition at Ego Wall and Mabuti. In these two sites, there were pronounced differences in coral and dead coral cover among years. Coral cover decreased by 41–48% at both sites in 2019, while dead coral cover increased by 44–50%. This loss in coral cover may be attributed to 3 coral genera (Acropora, Porites and Seriatopora) of branching morphology. Previous studies have shown the vulnerability of branching Acropora, Porites and Seriatopora to bleaching disturbance (Loya et al. 2001; van Woesik et al. 2011). Furthermore, coral cover prior to the bleaching disturbance may have been associated with the severity of coral loss as Ego Wall and Mabuti had higher coral cover compared to the other sites. Higher coral cover is associated with larger sized colonies or number of individual colonies thereby increasing the likelihood of colonies being affected by bleaching disturbance when compared to sites with smaller sized colonies or lower number of individuals (Bena and van Woesik 2004; Lafratta et al. 2017; Kim et al. 2019; Quimpo et al. 2020). Indeed, at South Corner and Shark Airport, where coral cover was comparatively lower, there were no changes in coral cover among years. Interestingly, Aladdin had high Acropora cover that remained relatively unchanged throughout the years. This was perhaps associated with water flow as the direction of currents pushes water from Aladdin southward (Cabaitan et al. 2019). The potentially high flow environment in Aladdin likely minimized the effects of bleaching disturbance (Fabricius 2006; Bayraktarov et al. 2013).
Fish community dissimilarity at the ARNP was more strongly influenced by sites than by years. This site variation was attributed to the predominant benthos at each site, perhaps supporting different suites of reef fishes. Indeed, at Aladdin, where corals were abundant, majority of the fishes were coral-associated damselfishes (Amblyglyphidodon aureus, Pomacentrus moluccensis, Chromis viridis, C. retrofasciata and Dascyllus reticulatus), cardinalfishes (Cheilodipterus quinquelineatus and Neoniphon samara) and butterflyfishes (Chaetodon baronessa and C. auriga) (Coker et al. 2014). In contrast, in Shark Airport, where other organisms and turf algae were abundant, the associated fishes were benthic invertivores (Oxycheilinus unifasciatus, Sufflamen bursa, Scolopsis bilineata and Parupeneus multifasciatus) and herbivores/detritivores (Acanthurus japonicus and Ctenochaetus striatus). These site differences in fish communities likely reflect the differential dietary and habitat requirements of fish species. While the influence of year on fish communities at the ARNP was minor, communities in 2019 at Mabuti were dissimilar in the multidimensional space when compared to communities in 2016 and 2018. It is currently unknown why only Mabuti experienced shifts in fish communities at the ARNP.
The relative stability of the fish communities among years despite changes in the benthos, particularly coral cover (upwards of 40% loss at two sites), may be attributed to the availability of alternative habitats that were unaffected by the bleaching disturbance (Graham et al. 2008). These habitats perhaps sustained the dietary and shelter requirements of most fishes at the ARNP. While this study only examined depths of 8-10m, previous works have revealed that coral reefs at the ARNP are expansive spanning multiple depths from 0 to 150 m (Cabaitan et al. 2019), with broadly comparable coral cover at depths of 10 to 30 m (Dumalagan et al. 2019; Albelda et al. 2020). Depths of 30m have been documented to reduce the impacts of bleaching disturbance (Bridge et al. 2013; Crosbie et al. 2019). Furthermore, environmental variables until 30m are not dissimilar enough from shallow reefs perhaps allowing movement of some fishes among depth gradients (Slattery et al. 2011). Studies in Hawaii (Walsh 1983) and Australia (Lewis 1997; Crosbie et al. 2019) similarly documented no changes in fish communities despite substantial (> 30%) coral loss. In contrast, other studies have documented declines in fish species richness and shifts towards alternative fish communities following coral loss of greater than 10% cover (Graham et al. 2006; Wilson et al. 2006; Richardson et al. 2018). These contrasting results reinforce the spatially variable impacts of bleaching disturbance on benthic and fish communities.
There were differences in the abundance of reef fishes at the ARNP attributed to the interaction between sites, years, and functional groups. Contrary to our expectations, we did not detect changes in corallivores at the ARNP. Previous studies have shown the sensitivity of corallivores to coral loss due to their reliance on corals as food (Graham et al. 2008; Pratchett et al. 2008). This lack of change was attributed to (1) three sites (i.e., Aladdin, Shark Airport and South Corner) not experiencing drastic changes in coral cover and (2) the potential alternative food sources at the two other sites (i.e., Ego Wall and Mabuti) that experienced heavy coral loss. At Ego Wall and Mabuti, other potential food sources such as Pocillopora and Montipora (to a lesser degree) (Pratchett 2007) showed rather stable percentage cover at the ARNP. In contrast, herbivores are presumed to positively respond to coral loss as the demise of corals frees up substrate for algae to occupy (Russ et al. 2015a, b, 2018; Taylor et al. 2020). At the ARNP, however, the response of herbivores was site-dependent, where abundance at Ego Wall and Mabuti declined, while it increased at the other three sites. The cause of this decline at Ego Wall and Mabuti was probably the loss of corals at these sites because even if herbivores are not entirely dependent on corals, some species are still highly associated (Coker et al. 2014). At the other three sites, the reason for the increase in herbivores is unknown as there were no noticeable changes in turf and macroalgae. This suggests that other variables aside from food availability may be shaping the herbivores at the ARNP. Planktivores increased in abundance at the ARNP in all sites surveyed. Interestingly, this increase was also recorded in Ego Wall and Mabuti where coral cover declined, even though coral loss has been associated with reduction in planktivore abundance (Russ et al. 2017). The persistence and increase of planktivores despite coral loss at the ARNP may be attributed to the presence of deeper habitats (Quimpo et al. 2018b; Dumalagan et al. 2019; Albelda et al. 2020) and/or the documented upwelling events (Cabaitan et al. 2019) that move nutrient-rich water to the surface.
The coral loss experienced at the ARNP for some sites were broadly comparable to other offshore reefs in the country that experienced bleaching disturbance, but much lower than offshore reefs at other geographic locations. In 1998, thermal stress and mass coral bleaching was documented in the Philippines, with ~ 45% decrease in coral cover recorded at one offshore location (Tubbataha; Arceo et al. 2001). This large decline in coral cover following bleaching was associated with high (~ 60%) coral cover pre-bleaching (Arceo et al. 2001). In contrast, more recent studies from the 2014–2017 bleaching event showed minimal effects of thermal stress on corals at 3 fringing reef sites in the Philippines because of cold water influx (Keith et al. 2018), turbidity (Valino et al. 2021) and preponderance of heat tolerant corals (Quimpo et al. 2020). This suggests that within the country, offshore reefs may be more vulnerable to increases in sea surface temperatures than fringing reefs. When contrasted with coral loss of other offshore reefs in the Seychelles (Graham et al. 2006) and northwestern Australia (Gilmour et al. 2013), coral loss at the ARNP was 30–50% lower. The presence of deeper habitats and proximity to cooler waters at the ARNP (Cabaitan et al. 2019; Dumalagan et al. 2019; Albelda et al. 2020) may have reduced the impacts of the bleaching disturbance as similarly documented in offshore reefs at Chagos (Graham et al. 2008) and Australia (Harrison et al. 2019).
In summary, this study shows that offshore and well-managed reefs are equally vulnerable to bleaching disturbance and, in the Philippine context, are more susceptible than fringing reefs of similar or less management status. While bleaching disturbance was experienced at the ARNP, the effects varied with taxonomic communities and sites. Benthic communities were altered by the bleaching disturbance, but the magnitude varied with sites. Noticeable declines in coral cover were documented for some sites where upwards of 40% coral cover was lost, while at other sites, there were minor changes in cover. The severity of coral loss was perhaps associated with the cover of coral pre-bleaching as the two most affected sites had the highest cover before the disturbance. This suggests the importance of small-scale differences in benthic composition at alleviating the impacts of bleaching disturbance. Fish communities were structured more strongly by sites due to differences in benthos and the suite of fishes they support and showed minimal change across years. This suggests that fish communities were comparatively stable after the bleaching disturbance. This stability despite high coral loss at some sites was probably influenced by the presence of alternative and deeper habitats that provided shelter and/or food (Graham et al. 2008; Quimpo et al. 2018b; Dumalagan et al. 2019; Albelda et al. 2020). While offshore and well-managed reefs may experience lower localized stressors (Sandin et al. 2008), they are still affected by large-scale global disturbances, including climate change and coral bleaching. However, reef topography, particularly the presence of alternative and/or deeper habitats, seems to minimize the consequences of bleaching disturbance (Graham et al. 2008; Harrison et al. 2019).