Coral reefs exist in a delicate balance between calcium carbonate (CaCO3) production and CaCO3 loss. Ocean acidification (OA), the CO2-driven decline in seawater pH and CaCO3 saturation state (Ω), threatens to tip this balance by decreasing calcification and increasing erosion and dissolution. While multiple CO2 manipulation experiments show coral calcification declines under OA, the sensitivity of bioerosion to OA is less well understood. Previous work suggests that coral and coral-reef bioerosion increase with decreasing seawater Ω. However, in the surface ocean, Ω and nutrient concentrations often covary, making their relative influence difficult to resolve. Here, we exploit unique natural gradients in Ω and nutrients across the Pacific basin to quantify the impact of these factors, together and independently, on macrobioerosion rates of coral skeletons. Using an automated program to quantify macrobioerosion in three-dimensional computerized tomography (CT) scans of coral cores, we show that macrobioerosion rates of live Porites colonies in both low-nutrient (oligotrophic) and high-nutrient (>1 µM nitrate) waters increase significantly as Ω decreases. However, the sensitivity of macrobioerosion to Ω is ten times greater under high-nutrient conditions. Our results demonstrate that OA (decreased Ω) alone can increase coral macrobioerosion rates, but the interaction of OA with local stressors exacerbates its impact, accelerating a shift toward net CaCO3 removal from coral reefs.
Tropical coral reefs are oases of productivity that support some of the world’s most biologically diverse ecosystems and important fisheries. High productivity by sessile organisms on reefs requires formation of hard calcium carbonate (CaCO3) substrate in the euphotic zone, where photosynthesis can occur. This is achieved through biogenic calcification by reef organisms such as corals, coralline algae, echinoids, foraminifera, and mollusks, which, together with precipitation of abiogenic CaCO3, build and cement the reef framework. Coral reef frameworks are degraded through bioerosion, the biologically mediated breakdown and dissolution of CaCO3 skeletons, as well as natural dissolution and export of sand and rubble off the reef (Glynn, 1997). Today, net CaCO3 accretion typically exceeds, albeit barely, net erosion and dissolution, allowing reefs to remain near the sea surface (Stearn et al., 1977; Hubbard et al., 1990).
Of mounting concern is that ocean acidification (OA), the decrease in ocean pH caused by absorption of anthropogenic CO2, could shift this delicate balance toward a negative CaCO3 budget where CaCO3 loss exceeds CaCO3 production. Addition of CO2 to seawater decreases pH and lowers the CaCO3 saturation state (Ω), creating a less favorable environment for CaCO3 precipitation. Aragonite is the polymorph of CaCO3 that corals use to build skeletons, and the CaCO3 saturation state with respect to aragonite (ΩArag) is therefore a useful quantity in identifying how OA impacts the reef CaCO3 budget. CO2 laboratory manipulation experiments show that as ΩArag decreases, rates of calcification by corals and coralline algae generally decline (Kroeker et al., 2010; Chan and Connolly, 2013). Additionally, laboratory CO2 manipulation experiments show that rates of bioerosion of coral skeleton increase with decreasing pH (Tribollet et al., 2009; Wisshak et al., 2012; Reyes-Nivia et al., 2013). The combination of declining calcification and increasing bioerosion under low pH and ΩArag implies that OA alone could drive coral reefs toward a state of net CaCO3 loss. However, the impact of OA on coral reef bioerosion has not been unequivocally demonstrated outside of the laboratory because in the tropical oceans, low ΩArag generally covaries with elevated nutrients, and high nutrient concentrations can drive high rates of coral bioerosion in the absence of acidification (Risk et al., 1995; Edinger et al., 2000; Holmes et al., 2000; Tribollet and Golubic, 2005).
We exploited natural gradients in ΩArag and nutrient concentrations across the Pacific basin to investigate the independent and interactive effects of ocean acidification and nutrients on macrobioerosion rates of live colonies of the Indo-Pacific coral Porites spp. While macrobioerosion (>1 mm boring diameter including bivalves, worms, and sponges) of coral skeleton is a fraction of total CaCO3 bioerosion on a reef (Glynn, 1997), independent studies show that macrobioerosion occurs in proportion to total bioerosion of coral rubble (Holmes et al., 2000) and experimental blocks of coral skeleton (Chazottes et al., 2002), and can thus be linked to total reef bioerosion. Macrobioerosion also affects the longevity of individual coral colonies, increasing their susceptibility to breakage and dislodgment by waves and storms (Scott and Risk, 1988; Chen et al., 2013).
MATERIALS AND METHODS
A total of 103 skeletal cores (3–7 cm diameters) were collected using underwater pneumatic and/or hydraulic drills from live Porites spp. coral colonies (∼40–100 cm tall) that were visually healthy at 11 sampling locations within seven reef systems across the Pacific basin (Fig. 1; Table 1). Cores were drilled downwards along the axis of maximum growth from approximately the center of the colonies, to an average depth of ∼35 cm. Across the Pacific basin, strong natural gradients exist in ΩArag and nutrient concentrations (Fig. 1), and in general, this pattern is supported by in situ sampling of the carbonate chemistry and dissolved inorganic nutrients of reef seawater (Table 1). Two eastern Pacific reefs (Pearl Islands and Taboga) in the Gulf of Panama are exposed to local upwelling water of low ΩArag and high nutrient concentrations (D’Croz and O’Dea, 2007; Manzello et al., 2008). In the central Pacific, Jarvis Island, Palmyra Atoll, and Kingman Reef are located near the margin of the Pacific cold tongue, where wind-driven upwelling along the equator brings water to the surface that is relatively acidic and nutrient rich compared to surrounding water. Rose Atoll and Wake Atoll are not exposed to cold-tongue waters and are characterized by high-ΩArag, low-nutrient conditions. On Palau, in the tropical western Pacific, a strong natural gradient in ΩArag exists across the archipelago, at persistently low nutrient concentrations (Table 1) (Shamberger et al., 2014). This reef system provides a unique opportunity to investigate the effect of low ΩArag on coral macrobioerosion in the absence of the confounding effect of elevated nutrients.
To characterize ΩArag and nutrient concentrations in reef seawater, samples were collected during multiple years, seasons, and times of day at the majority of our 11 reef locations (Table 1). Nevertheless, some degree of uncertainty remains because accurate estimates of the average ΩArag and nutritional environment over the lifetime of the coral requires sampling on all relevant time scales, including diurnal, seasonal, inter-annual, and decadal. Comparison with other in situ data sets suggests that this uncertainty is small relative to the range captured by our study sites (details are provided in the GSA Data Repository1).
We developed an automated computer program to quantify calcification and macrobioerosion rates in coral skeleton cores scanned by computerized tomography (CT). The program quantifies coral extension rate following the methods of Cantin et al. (2010), with modification to automatically trace the three-dimensional (3-D) growth paths of individual corallites within the core. This enables growth information to be collected from the entire 3-D core. Bulk skeletal density was determined from CT scans by comparison to coral standards, cylinders of coral skeleton whose density is calculated from mass and volume. Annual coral calcification rate (g cm–2 yr–1) was calculated as the product of skeletal density (g cm–3) and extension rate (cm yr–1). The automated program is described in detail in the Data Repository.
Equation 1 is equivalent to the product of percent volume bioeroded (Fig. 2) and coral calcification rate. Converting percent volume bioeroded to a mean bioerosion rate corrects potential biases caused by differences in growth rates and density amongst corals.
The data for percent volume bioeroded were fit with ΩArag as the predictor variable using a generalized additive model for location, scale, and shape with a beta inflated distribution (GAMLSS-BID; Rigby and Stasinopoulos, 2005). GAMLSS allows both the mean percent volume bioeroded and the skewness toward zero values (i.e., cores without macrobioerosion) to depend on ΩArag and nutrients. Sensitivity of macrobioerosion to ΩArag between low-nutrient (<1 μM nitrate) and high-nutrient (>1 μM nitrate) reefs was evaluated by comparing slopes of ordinary least-squares regressions fit to the reef mean macrobioerosion rates. Heteroscedasticity of the data precluded significance tests using linear regression, but did not invalidate the regression coefficients.
RESULTS AND DISCUSSION
Using only those cores collected from low-nutrient reefs spanning a natural gradient in ΩArag, we first quantified the impact of ocean acidification on macrobioerosion without the confounding influence of nutrients (Fig. 3). Our results show a significant (p < 0.05) increase in macrobioerosion with decreasing seawater ΩArag. This result confirms that ocean acidification alone increases rates of coral macrobioerosion, consistent with laboratory experiments that show increased sponge (Wisshak et al., 2012) and micro- (Tribollet et al., 2009; Reyes-Nivia et al., 2013) bioerosion of coral skeleton under simulated OA, low-nutrient conditions. In our corals, macrobioerosion rates increase by 10 mg CaCO3 cm–2 yr–1 per unit decrease of ΩArag.
Other field studies have reported high rates of bioerosion where seawater ΩArag is relatively low. For example, in the eastern tropical Pacific, high bioerosion rates (Reaka-Kudla et al., 1996) were measured on coral reefs bathed with naturally low ΩArag upwelled water (Manzello et al., 2008). Similarly, the density of macrobioeroders observed at the surface of live Porites colonies increased along a natural acidification gradient caused by CO2 venting onto reefs in Papua New Guinea (Fabricius et al., 2011). Low-pH seawater caused by submarine discharge was also linked to higher incidence of bioerosion in Porites astreoides colonies in the Yucatan (Crook et al., 2013). In these studies, however, either low pH and low ΩArag covary with high nutrient concentrations (Manzello et al., 2008; Crook et al., 2013), or nutrient data were not reported (Fabricius et al., 2011), making it difficult to attribute increased bioerosion or bioeroder density solely to OA.
Using a second set of cores, collected from high-nutrient reefs spanning a natural gradient in ΩArag, we investigated the combined impact of ocean acidification and elevated nutrients on coral macrobioerosion rates (Fig. 3). Our results show that sensitivity of macrobioerosion rate to ΩArag increases by an order of magnitude (from 10 to 110 mg CaCO3 cm–2 yr–1 per unit decrease of ΩArag) from low-nutrient reefs to high-nutrient reefs. The GAMLSS-BID analysis showed a significant effect of ΩArag on macrobioerosion within high-nutrient reefs, and a significant effect of nutrients when all reefs were included with ΩArag as a continuous predictor and nutrients as a categorical predictor. Our observation that nutrients accelerate coral bioerosion rates is consistent with that reported for live corals (Sammarco and Risk, 1990; Risk et al., 1995; Edinger et al., 2000; Holmes et al., 2000; Chen et al., 2013), coral rubble (Holmes et al., 2000), and experimental blocks of coral skeleton exposed on high-nutrient reefs (Chazottes et al., 2002; Tribollet and Golubic, 2005).
There are several potential mechanisms for coral macrobioerosion rates to increase with decreasing ΩArag and with increasing nutrients. First, relatively acidic seawater may increase the efficiency with which coral skeleton is dissolved by bioeroding organisms. For example, boring algae that infest live coral colonies, and increase their susceptibility to macrobioerosion, drive dissolution along the most soluble crystal surfaces (Kobluk and Risk, 1977). Second, nutrient enrichment may stimulate primary productivity, elevating particulate food availability and turbidity, making nutrient-rich reefs favorable environments for filter-feeding bioeroders. The role of coral skeletal density in determining sensitivity to macrobioerosion has been considered previously, with mixed results (Highsmith, 1981; Sammarco and Risk, 1990). We found no significant effect of skeletal density on macrobioerosion in the GAMLSS-BID analyses, nor did we find a relationship to water depth or reef type (Table 1).
Bioerosion is a natural process on coral reefs that supplies carbonate sediments critical to the cementation of the reef (Glynn, 1997), and may contribute to propagation of certain coral species that reproduce by fragmentation (Tunnicliffe, 1981). However, calcification must exceed bioerosion in order for reefs to grow and persist in the euphotic zone. Ocean acidification will drive a decrease in rates of calcification by corals and coralline algae, and ocean warming will exacerbate these impacts by inducing coral bleaching and mortality (Hoegh-Guldberg et al., 2007). If decreased calcification co-occurs with increased bioerosion, the CaCO3 balance will shift more rapidly toward a negative CaCO3 budget.
The results of this study show that the combination of OA (low ΩArag) and nutrient loading is ten times more effective at driving coral macrobioerosion than OA alone. Over the next century, ΩArag of reef seawater will be governed by the ocean’s absorption of anthropogenic CO2 and local and regional variability in biogeochemical processes (e.g., net photosynthesis and net calcification). Anthropogenic nutrient loading is already a major threat to coral reef ecosystems, with at least one quarter of coral reefs impacted by coastal development and watershed pollution (Burke et al., 2011). Curtailing global CO2 emissions, the primary driver of ocean acidification, cannot be tackled at a local level. However, effective local management strategies can limit anthropogenic nutrient fluxes to coral reefs, and are urgently needed to slow the shift to net CaCO3 removal for corals, and potentially coral reef ecosystems, worldwide.
We are grateful to G.P. Lohmann (Woods Hole Oceanographic Institution, WHOI), Kathryn Rose (WHOI), Jay Andrew (Palau International Coral Reef Center), Danny Merritt (National Oceanic and Atmospheric Administration, NOAA), and Edguardo Ocho (Smithsonian Institution, SI) for field assistance, and Julie Arruda (WHOI) and Darlene Ketten (WHOI) for CT scanning. Juan Mate (SI), Oris Sanjur (SI) Amanda Meyer (U.S. Fish and Wildlife Service, USFWS), Susan White (USFWS), the staff of the Palau International Coral Reef Center, and Camilo Ponton (WHOI) assisted with permitting, access to the PRIA sites, and translation of permit applications. Elizabeth Drenkard (WHOI) collected and analyzed Fall 2012 seawater samples from Jarvis Island. We thank Aline Tribollet for insightful discussion, and three anonymous reviewers whose suggestions significantly improved the manuscript. This work was supported by National Science Foundation (NSF) grant OCE 1041106 to Cohen and Shamberger, NSF grant OCE 1220529 to Cohen, The Nature Conservancy award PNA/WHOI061810 to Cohen, NSF Graduate Research Fellowships to DeCarlo and Barkley, and a WHOI-Ocean Life Institute post-doctoral fellowship to Shamberger. The NOAA Coral Reef Conservation Program provided field and logistical support for Pacific Reef Assessment and Monitoring Program research cruises. NOAA’s Ocean Acidification Program provided scientific support to Brainard and Young. This paper is dedicated to the memory of Jay Andrew.