Carbonate production of Micronesian reefs suppressed by thermal anomalies and Acanthaster as sea-level rises

Coral reefs are essential to millions of island inhabitants. Yet, coral reefs are threatened by thermal anomalies associated with climate change and by local disturbances that include land-use change, pollution, and the coral-eating sea star Acanthaster solaris. In combination, these disturbances cause coral mortality that reduce the capacity of reefs to produce enough carbonate to keep up with sea-level rise. This study compared the reef-building capacity of shallow-water inner, patch, and outer reefs in the two islands of Pohnpei and Kosrae, Federated States of Micronesia. We identified which reefs were likely to keep up with sea-level rise under different climate-change scenarios, and estimated whether there were differences across habitats in the threshold of percentage coral cover at which net carbonate production becomes negative. We also quantified the influence of A. solaris on carbonate production. Whereas the northwestern outer reefs of Pohnpei and Kosrae had the highest net rates of carbonate production (18.5 and 16.4 kg CaCO3 m-2 yr-1, respectively), the southeastern outer reefs had the lowest rates of carbonate production (1.2–1.3 and 0.7 kg CaCO3 m-2 yr-1, respectively). The patch reefs of Pohnpei had on average higher net carbonate production rates (9.5 kg CaCO3 m-2 yr-1) than the inner reefs of both Pohnpei and Kosrae (7.0 and 7.8 kg CaCO3 m-2 yr-1, respectively). A. solaris were common on Kosrae and caused an average reduction in carbonate production of 0.6 kg CaCO3 m-2 yr-1 on Kosraean reefs. Northern outer reefs are the most likely habitats to keep up with sea-level rise in both Pohnpei and Kosrae. Overall, the inner reefs of Pohnpei and Kosrae need ~ 5.5% more coral cover to generate the same amount of carbonate as outer reefs. Therefore, inner reefs need special protection from land-use change and local pollution to keep pace with sea-level rise under all climate-change scenarios.


Introduction
Coral reefs are an integral component of global marine ecosystems and are essential to millions of people that benefit from the goods and services that coral reefs provide. For example, coral reefs reduce storm-wave energy by up to 97% [1], reducing the threat of coastal inundation a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 Kosrae, we only stratified the reefs as either inner reefs or outer reefs (because of the lack of patch reefs). Sample size of each strata was determined by calculating the geographic area of each reef type, using the area function from the R package raster [31], and allocating the number of sites in accordance with the area estimates. Reef surveys focused on the 2-5 meters depth contour to estimate shallow-water carbonate production.
Six, 10 m transects, using a modified line-intercept technique that followed the reef substrate, were used to measure the benthic composition for every centimeter, at each site of the 48 sites [32,17]. A few meters gap was allocated between the ends of the transects to ensure no overlap of substrate between transects. Corals were recorded to species level, except massive Porites and encrusting Montipora, which were recorded in the field as growth forms. All other organisms along each transect were identified to the highest possible taxonomic resolution. Rugosity was recorded using the planar length of a second transect that spanned across the reef horizontally. Echinoids were recorded within 30 cm on either side of the 10 m tape. The urchins were recorded as Echinometra, Diadema, and 'Other', and the diameter of each echinoid test was measured to the nearest 0.5 cm. The abundance of Acanthaster solaris (crown-of-thorns sea star) were recorded within 5 m along each of the six 10 m transects. Herbivorous parrotfishes were videoed and identified to species and their estimated length was recorded to the nearest cm along six transects, each of which was 30 m long by 4 m wide. Care was taken to record the fish-transect videos ahead of the other transects to avoid any disturbance to the fishes.

Carbonate production
Net carbonate production (kg CaCO 3 m -2 yr -1 ) was estimated using the following equation: where calcification is the gross carbonate production by reef building organisms at site i [33]; sedimentation is the contribution of sediment to the reef, where it increases carbonate production rates if sedimentation is low (< 0.05 kg m -2 d -1 ) [33,34] and then sgn (x) is positive, whereas if terrestrial sedimentation is high sgn (x) is negative because the sediment smothers corals; erosion is the rate of erosion, estimated following van Woesik (2013) [33]; and Acanthaster is amount of carbonate potentially lost by Acanthaster solaris (i.e., the crown-of-thorns sea star) eating corals. High densities of A. solaris reduce live coral cover [35], which in turn reduces a reef's capacity to produce carbonate. Calcification of organisms was calculated as follows: where r is the rugosity of site i averaged across six transects; m is the adjustment coefficient for the morphology of species j at site i (following van Woesik and Cacciapaglia 2018) [17]; x is the planar percent cover averaged across site i for species j, d is the density (g cm -3 ) of species j (following [17]) in site i; g is the vertical growth rate of coral species j (cm yr -1 ) (after van Woesik and Cacciapaglia 2018 [17]); 10 is an adjustment constant to convert units back to kg CaCO 3 m -2 yr -1 ; and ca is the contribution of coralline algae to carbonate production, calculated following Perry et al. (2012) [14] as: where pca is the planar coralline algae cover averaged across six transects at site i, 0.018 is the gross carbonate production (g cm -2 yr -1 ) estimated using averages from Perry et al. (2012) [14]; and 10 is an adjustment constant used to convert units from g cm -2 yr -1 to kg m -2 yr -1 .

Sedimentation
The accretion of reefs can be supplemented by calcareous sedimentation [9,10], or compromised by excessive amounts of terrestrial sedimentation (when > 0.05 kg m -2 d -1 ) [34], which causes coral smothering and reduces the rate of carbonate production [36]. The sedimentation rate that was used in Pohnpei and Kosrae was 0.4 kg CaCO 3 m -2 yr -1 following estimates from Montaggioni (2005) [10] and Hubbard (1997) [36]. We witnessed some terrestrial runoff and a high deposition of fine sediment in the southern bay of Kosrae (Utwe Bay), and we therefore introduced a negative sedimentation component to Eq 1, using -0.4 kg CaCO 3 m -2 yr -1 [37], at sites that were downstream of river runoff at that location.

Erosion
Reef erosion was comprised of three biological components, echinoids or sea urchins, herbivorous fishes, and macroboring organisms. Gross erosional rates were calculated as: where parrotfish is the rate of erosion by herbivorous fish species j at site i; urchin is the rate of biological erosion by sea urchins species j at site i; and macroboring is the erosional forces of macroboring organisms in site i. Erosion by parrotfish was estimated after [14] using the equation: where vol is the estimated volume of the bites of individual parrotfish n for species j at site i; sp is the scar proportion, or the proportion of bites that leave scars on corals for individual n, of species j at site i; br is the bite rate (bites day -1 ) of individual n, of species j, at site i; the average density D of corals was calculated at site i based on coral composition; the constant 365 was to convert days into years; and 0.001 was a constant to convert grams into kilograms. Bite volume vol was further defined using the following equation: where length is the length (cm) of each parrotfish n, of species j at site i; the constants were gained using a linear regression of data collected by Ong and Holland (2010) [38]; the constant one thousand was used to convert cubic millimeters to cubic centimeters. Scar proportion, sp, from Eq 5 was further defined as follows: where length is the length of fish n of species j at site i. The equation was based on a regression using data from Bonaldo and Bellwood (2008) [39] and Ong and Holland (2010) [38]. Bite rate, br, from Eq 5 was defined as: br i;j;n ¼ 60 f½ð4:31 þ brc i;j À 0:36Þ À ð0:045 � reeftime � length i;j;n Þ�g ð8Þ where brc is a bite rate constant, reeftime is the amount of time fishes spend grazing on reefs, estimated to be 9 hours per day. These constants were estimated by Peter Mumby (personal communication). Length is length of fish n of species j at site i.
Erosion by sea urchins (kg CaCO 3 m -2 ) was estimated after [14] using the following equation: where Diadema is the erosion caused by a Diadema individual n at site i; Echinometra is the erosion caused by an Echinometra individual n at site i, and Other urchins is the erosion caused by sea urchins that were not Echinometra or Diadema. Diadema was defined as: where diameter is the test size of individual n outside the genus of Echinometra or Diadema at site i. Macroboring organisms were included into Eq 4 to incorporate the erosional forces of boring sponges following the equation: where plamc is the planar cover of the macroboring organisms averaged over at site i; and mec is the constant used to define macroboring erosion, which was set as a conservative 10 kg CaCO 3 m -2 yr-1 following Glynn (1997) [40].

Acanthaster
While Glynn (1973) [41] estimated the densities of Acanthaster that would overwhelm the ability for corals to persist, there have been no studies aimed at quantifying the influence of Acanthaster on carbonate production rates. We used field estimates to evaluate the effect of Acanthaster solaris on carbonate production as follows: where Acanthaster i is the reduction in gross production caused by observed A. solaris at site i; RI i is the rate of coral ingestion at site i (see Eq 16); 50 is a constant to convert observational transect size to m 2 ; and tc.transect i is the per transect consumption rate: where A.sp i is the number of A. solaris observed in site i divided by the number of transects; con is the average consumption rate (0.01 m 2 d -1 ) estimated from Keesing and Lucas (1992) [42]; Density i is the average density of corals in site i (g cm 3 ); 10 is the constant used to convert the unit to kg m 2 , and 365 converts days to years.
where RI i is the rate of coral ingestion in site i; R i is the resource density or live coral cover (%) at site i. The handling rate, or how long it takes for a single A. solaris to eat a coral colony, h, was estimated using the average size of coral colonies and the average rate of consumption, which was conservatively estimated to be around 3.5 days. The 4.53 constant is used to rescale RI i , which is resource dependent, to match average coral density with estimated consumption rate; a i is the attack rate, estimated using the speed at which A. solaris can move and the density of corals in the transect following Eq 17: where 12 is a constant for active predatory hours; R i is the resource density or live coral cover in site i as a percentage, subtracted from 100; and 1 E 10 was added to convert the value to a non-zero area where corals are present. These values are divided by 10 for the transect length, to determine average distance between corals, and it was assumed the A. solaris had to search the area of a circle with this average distance between corals equaling the diameter of that circle. The area was then divided by the speed at which A. solaris can move, speed, (504 m d -1 ; Muller et al. 2011 [43]). It was assumed that A. solaris could only reduce or negate carbonate production in this model, so the effects were subtracted from gross production to a maximum erosional force netting zero gross production.
To convert net carbonate production, from Eq 1, to vertical reef growth (in mm) we used: where Cp is carbonate production (from Eq 1) and alpha is a coefficient estimated as -0.01949 (after van Woesik and Cacciapaglia 2018 [17]).

Carbonate thresholds
We used an additive mixed effects model in a Bayesian framework [44] to estimate the value of coral cover, for the different habitats, at which net carbonate production became negative, using the following: where G is the net carbonate production at site i; f(coral cover) uses an O'Sullivan spline smoothing function [45]; Habitat is the fixed effect of interest; a is a random intercept for site; and error is the error term for the residuals. We assumed that no prior information was known and therefore used multivariate normal diffuse and normal diffuse priors [44]. All models were run in R and coded in JAGS [46] (all the R code is available in S1 Data and at https:// github.com/rvanwoesik). We would also like to thank Eugene Joseph the Director of the Conservation Society of Pohnpei and Andy George the Director of the Kosrae Conservation Society for granting us permission to conduct research on Pohnpei and Kosrae respectively.
Gross carbonate production on Kosrae averaged 7.4 kg CaCO 3 m -2 yr -1 , although on average the outer and inner reefs did not vary greatly (7.6 and 7.0 kg CaCO 3 m -2 yr -1 , respectively). Similar to Pohnpei, carbonate production on the outer reefs on Kosrae was variable and was highest on the northern outer reefs (16.4 kg CaCO 3 m -2 yr -1 ), and lowest on southeastern outer reefs (0.7 kg CaCO 3 m -2 yr -1 ) ( Table 1 and Fig 3). Net production differed from gross Table 1

. Carbonate production and erosion rates across shallow-water coral reef habitats (2-5 m depth) on Pohnpei and Kosrae, Federated States of Micronesia.
All values are in kg CaCO 3 m -2 yr -1 . Gross production is the total rate of carbonate production across habitats, excluding erosion rates. Erosion includes parrotfish and urchin erosional forces combined (which does not include Acanthaster solaris erosion). Net production is gross carbonate production minus the erosional estimates and sedimentation inputs. Acanthaster erosion is the erosion caused by A. solaris. Combined reef strata were only available for inner and outer reefs, due to the lack of patch reefs in Kosrae.

Coral-reef habitat
Gross production (kg CaCO 3 m -2 yr -1 )  Reef growth in Micronesia production at sites where sedimentation and erosion were much higher than background rates. This occurred in Utwe Bay in southern Kosrae, where terrestrially derived sediment was much higher than elsewhere (personal observations). Terrestrially derived sediment smothers coral colonies and thereby reduces carbonate production. The reefs of Pohnpei and Kosrae supported similar coral assemblages, although there were some differences in species dominance. The reefs of Pohnpei, particularly the patch and inner reefs, were dominated by Porites rus, Porites cylindrica, and Porites lobata. The outer reefs were dominated by encrusting Montipora and Acropora hyacinthus (Fig 4). These five species contributed 82% of the gross carbonate production on Pohnpei. The reefs of Kosrae were dominated by encrusting Montipora, Porites rus, Goniastrea retiformis, Porites lobata, and Porites lichen (Fig 4). These five species contributed 78% of the total gross carbonate production on Kosrae. Importantly, the inner reefs of both Pohnpei and Kosrae, and the patch reefs of Pohnpei had a higher live-coral-cover threshold than the outer reefs of both islands, although there was considerable uncertainty in the thresholds for the inner reefs (i.e., high 95% credible intervals) (Fig 5).
Acanthaster solaris were observed on reefs of both islands although in 2018 populations indicative of an outbreak (>30 hectare -1 ) were only observed on some of the shallow reefs of Kosrae. Since outer reefs tended to have the highest densities of A. solaris, carbonate production on these outer reefs were most affected (Table 1 and Figs 6 and 7). We re-ran the carbonate production model for both islands to incorporate A. solaris and found that carbonate production was reduced on Kosrae by an average 0.6 kg CaCO 3 m -2 yr -1 and on Pohnpei by 0.04 kg CaCO 3 m -2 yr -1 , across all habitats (Table 1 and Figs 6 and 7). A. solaris densities did not reduce carbonate production to negative values, although at two northwestern sites on Kosrae gross carbonate production was reduced by 80% and 62%, where A. solaris densities were 17 and 14 (per 300 m 2 ) and where coral cover was low. For mitigation purposes, and to sustain a productive reef, we found that A. solaris densities should be kept below a density threshold that is proportional to 7.3% of the relative coral densities ( Figure B in S1 File). For example, if a 100 m 2 site supports 30% live coral cover, any more than two Acanthaster in that site for one year will likely reduce gross carbonate production to zero.
Other echinoids also reduce carbonate production, particularly Echinometra mathaii, and related species in high densities [38]. We noticed that large populations of E. mathaii were common on southeastern outer reefs and caused considerable erosion (Figures C and D in S1 File) and are less common on inner and patch reefs. Carbonate erosion by parrotfishes was also high on the southeastern reefs and along western reefs, and was lower elsewhere (Figures E and F in S1 File).

Discussion
This study aimed to identify the spatial variation of shallow-water carbonate production in Pohnpei and Kosrae, Federated States of Micronesia, to assess which reefs are likely to keep up with sea-level rise, and to determine what role Acanthaster solaris plays in carbonate Reef growth in Micronesia production. While the leeward, northern and northwestern facing reefs had the highest rates of net carbonate production (16.5-18.5 kg CaCO 3 m -2 yr -1 ), the windward, southeastern facing reefs showed the lowest rates of net carbonate production (0.7-1.3 kg CaCO 3 m -2 yr -1 ). Such high variation in carbonate production along the outer shallow-water reefs is important, especially since habitats with low rates may not have the capacity to keep up with predicted sealevel rise. Based on different greenhouse-gas-emission scenarios, most frequently conveyed as Representative Carbon Pathways (RCPs) 2.6. 4.5, 6.0, and 8.5 Wm -2 , the predicted rates of sealevel rise by the year 2100 have been conservatively estimated at 5, 6.5, 6.7, and 9 mm yr -1 , respectively [25]. Converting the field estimated rates of carbonate production to vertical rates of reef accretion (following Eq 18) the northwestern shallow-water outer reefs of Pohnpei and the northern shallow-water outer reefs of Kosrae are estimated to vertically accrete at 11.8 mm yr -1 and 11.2 mm yr -1 , respectively. These rates of vertical accretion are relatively high for contemporary reefs and exceed the rates of sea-level rise under RCP 8.5. Therefore, if effectively managed, the northern shallow-water outer reefs of Pohnpei and Kosrae will likely have the capacity to keep up with sea-level rise and maintain their essential ecosystem functions.
By contrast, the southeastern shallow-water reefs of Pohnpei and Kosrae, have estimated vertical accretion rates of only 1.3 mm yr -1 . Although we did not directly measure sedimentation rates nor did we measure micro-bioerosion rates, our data suggest that the southeastern reefs are not likely to keep up with sea-level rise by the year 2100. Indeed, the projections of our model show that without considering spatial variation, on average, Pohnpei's and Kosrae's shallow-water outer reefs fall short of the moderate rates of sea-level rise projected under RCP 4.5 (i.e., 6.2 mm yr -1 accretion and 6.5 mm yr -1 of sea-level rise under RCP 4.5, Figure G in S1 File). Most concerning is that the shallow-water inner reefs of both islands have estimated Reef growth in Micronesia rates of vertical accretion averaging 5.9 mm yr -1 , which is lower than most predicted rates of sea-level rise, even the conservative rates associated with RCP 4.5 by the year 2100 (Figure G in S1 File). Although there is some uncertainty in the live-coral-cover thresholds for the inner reefs of Pohnpei and Kosrae, these inner reefs on average require around 5.5% higher live coral cover than outer reefs to produce the same amount of carbonate (Fig 4). These results provide a strong conservation message that in order for nearshore reefs to have a chance to keep up with sea-level rise, it is critical to mitigate land-use discharge and pollution to nearshore shallow-water reefs. Without conserving these relatively sensitive, nearshore reefs our projections suggest that they would likely drown in the near future. Additionally, mitigating terrestrial runoff may also prevent large, persistent outbreaks of Acanthaster [21,22].
Increases in the survival of Acanthaster brachiolaria-stage larvae have been associated with river discharge and elevated nutrient concentrations [21,22]. Therefore, implementing management strategies on small Pacific islands that mitigate terrestrial discharge will not only reduce sedimentation stress on corals but will also effectively suppress chronically dense Acanthaster populations that reduce carbonate production. In Kosrae, A. solaris reduced carbonate production by on average 0.6 kg CaCO 3 m -2 yr -1 , with a maximum reduction of 5.1 kg

Reef growth in Micronesia
CaCO 3 m -2 yr -1 when A. solaris were at high densities (> 15 individuals per 300 m 2 ). Although our estimates of the impact of A. solaris on carbonate production are novel, they still comprise a degree of uncertainty because of the assumptions underlying Eqs 14-17. Therefore, future improvements in these estimates can be made by examining these assumptions, which could include an adjustment for coral composition.
The coral species that were the most important contributors to carbonate production on both islands were Porites rus and Porites lobata, particularly on the shallow-water inner reefs. The most important contributors to carbonate production on the shallow-water outer reefs were encrusting Montipora, merulinids, and acroporids. There was a lack of patch reefs and large lagoons in Kosrae, therefore Porites cylindrica was less common on Kosrae than on Pohnpei. Yet, if the shallow-water outer reefs are unable to keep up with sea-level rise and are breached by offshore waves, the patch reefs of Pohnpei and the inner reefs of both islands will likely become more similar in coral composition to that of the outer reefs [47]. Still, whether these reefs, even altered in composition, will be able to produce enough carbonate to keep up with sea-level rise is an open question.
The shallow-water coral-reef carbonate production rates measured on both Pohnpei and Kosrae are lower than the field estimates recorded farther west on Palau and Yap (~2.2 kg CaCO 3 m -2 yr -1 less, when averaged among habitat types) [17]. The lower carbonate production rates are most likely a result of reduced Acropora cover caused by the recent thermal-stress events on both Kosrae and Pohnpei in 2016 and 2017 (Peter Houk, pers. comm.). At the same time, similar thermal-stress events were not recorded in Palau and Yap. Thermal-stress events are known to significantly reduce a reefs' capacity to produce calcium carbonate [19], and under extreme events can temporarily reduce net accretion to negative values [18]. Although there are some studies on the net ecosystem calcification of coral reefs and the influence of coral bleaching on that process [47][48][49][50][51], more field studies are needed that examine (i) thermal-stress events and the dynamics of carbonate production through those events, and (ii) the rates of recovery of carbonate production from thermal-stress events.
The capacity of coral reefs to keep up with rising sea level is important for coastal residents and is particularly relevant to residents of low-lying islands who cannot move to higher elevations. Historically, healthy coral reefs have kept up with dynamic shifts in sea level through glacial-interglacial periods [52], yet disturbances to modern reefs are suppressing the capacity of coral reefs to produce enough carbonate [53] and protect island residents from storm-wave damage. In addition, drowned reefs will not be able to provide goods and services or support fisheries. If the coral species contributing to reef complexity and carbonate production are unable to persist under the stress of climate change then the coral reefs will not keep up with sea-level rise and drown.
Supporting information S1 File. Seven supporting figures. These figures include (A) a location map, (B) limit on the number of Acanthaster solaris sea stars, as a proportion of live coral cover (LCC) that a 100 m 2 shallow-water coral-reef habitat, (C and D) erosional kriged maps for both islands, and (E and F) erosional maps for parrotfishes for both islands, and (G) net vertical accretion split by strata compared to the rates of sea-level rise under representative concentration pathways 2.6, 4.5, 6.0, and 8.5 W m -2 . (PDF) S1 Data. Data and R code. Spreadsheet data for each site of the 48 sites on Kosrae and Pohnpei and all the R scripts that produced the manuscript figures. (RAR) environment and ensured our every need was met. We would also like to thank Andy George and his marine monitoring team at the Kosrae Conservation Society for a collaborative opportunity. Special thanks go to Sandra J van Woesik, Kelly J van Woesik, Liana J van Woesik, and Marina L Fleming for their invaluable voluntary field and lab assistance. We would also like to extend our thanks to Sandra J van Woesik and the anonymous reviewers for their editorial comments on the manuscript. This is contribution number 217 from the Institute for Global Ecology at the Florida Institute of Technology.