Mangrove and Seagrass Beds Provide Different Biogeochemical Services for Corals Threatened by Climate Change

Rapidly rising atmospheric CO2 concentrations are driving acidification in parallel with warming of the oceans. Future ocean acidification scenarios have the potential to impact coral growth and associated reef function, although reports suggest such affects could be reduced in adjacent seagrass habitats as a result of physio-chemical buffering. To-date, it remains unknown whether these habitats can actually support the metabolic function of a diverse range of corals. Similarly, whether mangroves provide the same ecological buffering service remains unclear. We examine whether reef-associated habitat sites (seagrass and mangroves) can act as potential refugia to future climate change by maintaining favorable chemical conditions (elevated pH and aragonite saturation state relative to the open-ocean), but by also assessing whether the metabolic function (photosynthesis, respiration and calcification) of important reef-building corals are sustained. We investigated three sites in the Atlantic, Indian and Pacific Oceans and consistently observed that seagrass beds experience an overall elevation in mean pH (8.15 ± 0.01) relative to the adjacent outer-reef (8.12 ± 0.03), but with periods of high and low pH. Corals in the seagrass habitats either sustained calcification or experienced an average reduction of 17.0 ± 6.1 % relative to the outer-reef. In contrast, mangrove habitats were characterized by a low mean pH (8.04 ± 0.01) and a relatively moderate pH range. Corals within mangrove-dominated habitats were thus pre-conditioned to low pH but with significant suppression to calcification (70.0 ± 7.3 % reduction relative to the outer-reef). Both habitats also experienced more variable temperatures (diel range up to 2.5°C) relative to the outer-reef (diel range less than 0.7°C), which did not correspond with changes in calcification rates. Here we report, for the first time, the biological costs for corals living in reef-associated habitats and characterize the environmental services these habitats may play in potentially mitigating the local effects of future ocean acidification.

We examine whether reef-associated habitats (seagrass, mangrove) can act as refugia to within each region and habitat. In doing so we provide novel data demonstrating that sites across 153 bioregions for both seagrass beds and mangroves consistently provide important, but very 154 different ecological services, driven by inherent differences in biogeochemical characteristics. 155 We define for the first time the different roles reef-associated habitats of seagrass and mangroves 156 will potentially play towards local mitigation of climate change, and clarify their potential as 55°44.05) and a mangrove (dominant species: Rhizophora mucronata, Lumnitzera racemose, 191 Brugueira gymnorhiza and Avicennia marina) dominated habitat (04°17.29, 55°43.89) located 192 within a bay known locally as Baie La Raie. The mangrove site was not directly under the 193 mangrove canopy (no influence from mangrove canopy shading) but in close proximity on the 194 seaward side. All sites were subjected to a semi-diurnal tidal cycle and currents at the mangrove 195 sites within Baie La Raie ran in an anti-clockwise direction during sampling. 196 The PO study sites were situated around Hoga and Kaledupa islands, located in the Indonesia (Tomascik et al., 1997). The outer-reef site (05°28.38, 123°43.73) was situated 201 adjacent to the fringing reef crest on the reef flat at a site locally known as Pak Kasims, off the 202 south coast of Hoga island. One of the reef-associated habitat sites was an adjacent inshore 203 seagrass habitat also off the south coast of Hoga island (05°28.38, 123°43.74) which was 204 dominated by Thalassia hemprichii. The second reef-associated habitat was immediately 205 adjacent to the "Langeria" mangroves located off the northern coast of Kaledupa island (05° 206 28.42, 123° 43.64). This site was situated outside of the mangrove canopy (again negating the 207 impact of canopy shading) on the seaward side, as for the IO site. The mangroves adjacent to the 208 site were primarily Rhizophora stylosa. The carbonate reef systems here experience good water 209 quality with minimal impact from sediment load (Bell and Smith, 2004) and light attenuation 210 (Hennige et al., 2010). During sampling currents ran in a southeast direction but were driven by 211 tides, with sites exposed to a semi-diurnal tidal cycle.

214
Environmental conditions and in situ metabolic activity were measured over five days within a 215 two week period during the annual dry seasons of each region. The mean and variance 216 (coefficient of variation ( CV )) of environmental conditions for this period did not significantly differ from values determined for a longer-term study across a full neap-spring cycle within the 218 same season (AO, Figure S1). As expected (Albright et al., 2013  Net P and R rates were determined for several time points (t) throughout the day and night, 323 respectively, and rates were normalized (to give mmol O 2 m 2 h -1 ) as described for calcification 324 rates to give: Daily P N and R (mmol O 2 m 2 d -1 ) were calculated by integrating all photosynthesis and 327 respiration measurements: 329 P G was calculated by the addition of P N and R. values, G to P:R, G to pH mean, G to pH CV (pH variability) , and percent cover of calcifying and 335 non-calcifying species to pH CV . Parametric test assumptions were met, with the Bartlett test used 336 to check homogeneity of variance and qq-plots to assess normality of the data.

337
Mixed Effects (LME) models were applied, with coral species as a random effect, to 338 examine effect of habitat on daily net P and R. Cleveland dot-plots were used to determine 339 outliers and boxplots and scatterplots were used to check for co-linearity within the dataset (Zurr

362
Across bioregions and habitats there were significant differences (see Table 1  AO and IO sites, pH peaks and troughs did not correspond with the tidal cycles ( Figure S2).

384
On average, the calcification-to-dissolution threshold (G-D) never fell below the Mg-

385
calcite Ω threshold levels of 3.0-3.2 for any of the habitats ( Across all bioregions, the outer-reef sites showed strongest co-variability between nA T 393 and nC T via calcification-carbonate dissolution (Figure 1). In contrast, the reef-associated 394 habitats exhibited co-variability between nA T and nC T more strongly influenced by cover of calcifying benthic photoautotrophs in the reef-associated habitats (8.6 ± 0.1%, Figure   420 3a), a number of coral species were present (7-15 species, Table 3).

421
Coral species found within the reef-associated habitats accounted for 28 -86 % of coral 422 cover on the main outer-reef (Table 3). Across regional locations the coral species found within 423 the reef-associated habitats of the AO collectively accounted for the highest percent coral cover 424 on the outer-reef (back-reef= 86 % and seagrass= 48 %). In the higher diversity regions of the IO 425 and PO, the reef-associated habitat coral species contributed between 28 -40 % to the coral 426 cover of the outer-reef habitats (Table 3). Coral cover in the AO (13.5 ± 0.5 %) outer-reef site 427 was ca. 60 % lower than the same habitat type in the IO (34.5 ± 1.4 %) and PO (32.3 ± 0.9 %).

430
Calcification rates per coral species were highest at the outer-reef sites (257.0 ± 15.9 431 mmol m 2 d -1 ), in particular for the fast growing Acropora spp. (340.0 ± 2.9 mmol m 2 d -1 ). Here, 432 environmental conditions were less variable than the reef-associated habitats (Table 1). Very 433 different patterns for coral calcification were observed between the reef-associated habitats. 38, P< 0.001, Figure 4b), but to a lesser extent with increasing pH CV (r 2 = 0.268, n= 38, P< 0.001, 444 Figure 4a). This potential regulatory function of mean pH is consistent with the change of 445 NEC:NEP across habitats ( Table 2). The similarity in mean and cv of the abiotic factors (light, 446 temperature, NO 3 -, see Table S1) between the reef-associated habitats suggests that differences in 447 carbonate chemistry are significant in structuring coral biomass and growth between mangroves 448 and seagrass systems. There were no significant relationships between calcification rates and 449 temperature or light (mean or cv).

450
Across all bioregions, an increase in the gross photosynthesis-to-respiration ratio (P:R) 451 corresponded with a positive increase in calcification (r 2 = 0.501, n= 38, P< 0.001, Figure 5). In 452 the outer-reef, P:R remained above one, however, in the reef-associated habitats P:R decreased, 453 largely due to a decrease in P (P< 0.05,

461
Within this study we demonstrate that both seagrass and mangrove reef-associated habitats (see Figure S2). The magnitude of influence of seagrass species on the carbonate budget is still 520 unresolved, with some species capable of direct carbonate production (Enríquez and Schubert, 521 2014).Ultimately this issue will need to be resolved through targeted investigation in order to 522 fully understand their potential role in carbonate loss relative to photosynthetic and respiration 523 activity, and hence their net contribution to the local carbonate system.

524
Mangrove habitats within this study had carbonate chemistry conditions in part 525 influenced by the local benthic composition, but they also appeared to be largely affected by  (Table 1).  (Table S1). It is possible that other nutrients may influence 555 coral metabolic activity within associated-reef habitats (Langdon and Atkinson, 2005).

556
Collectively however our results suggest that photosynthesis and calcification were most likely  The mean (± standard error, SE) and coefficient of variation ( CV ) in bio-physiochemical parameters for all habitats (outer-reef, seagrass, back-reef and mangrove) and bioregion sites (Atlantic, Indian and Pacific Ocean). n= 5 days and 40 discrete water samples.  Table 3. Coral species list for associated-reef habitat sites in the Atlantic, Indian and Pacific Oceans.

Figure Legends
Figure 1. Salinity-normalized total alkalinity (nA T ) and total carbon (nC T ) plots with best-fit linear regression for three sites and habitats in the Atlantic (AO), Indian (IO) and Pacific Oceans (PO). Data is from five days over a two week period during the dry seasons for each region between 2013-2014. The AO site consisted of a seagrass, back-reef and outer-reef control, whilst the IO and PO sites had a seagrass, mangrove and outer-reef habitat. Black lines represent the theoretical impact of calcification (C), carbonate sediment dissolution (D), photosynthesis (P), and respiration (R) on nA T and nC T . Average nA T and nC T is indicated by a yellow dot. C and D are dominant processes when a linear regression slope approaches 2.