Projected pH reductions by 2100 might put deep North Atlantic biodiversity at risk

. This study aims to evaluate the potential for impacts of ocean acidiﬁcation on North Atlantic deep-sea ecosystems in response to IPCC AR5 Representative Concentration Pathways (RCPs). Deep-sea biota is likely highly vulnerable to changes in seawater chemistry and sensitive to moderate excursions in pH. Here we show, from seven fully coupled Earth system models, that for three out of four RCPs over 17 % of the seaﬂoor area below 500 m depth in the North Atlantic sector will experience pH reductions exceeding − 0.2 units by 2100. Increased stratiﬁcation in response to climate change partially alleviates the impact of ocean acidiﬁcation on deep benthic environments. We report on major pH reductions over the deep North Atlantic seaﬂoor (depth > 500 m) and at important deep-sea features, such as seamounts and canyons. By 2100, and under the high CO 2 scenario RCP8.5, pH reductions exceeding − 0.2 ( − 0.3) units are projected in close to 23 % ( ∼ 15 %) of North Atlantic deep-sea canyons and ∼ 8 % (3 %) of seamounts – including seamounts proposed as sites of marine protected areas. The spatial pattern of impacts reﬂects the depth of the pH perturbation and does not scale linearly with atmospheric CO 2 concentration. Impacts may cause negative changes of the same magnitude or exceeding the current target of 10 % of preservation of marine biomes set by the convention on biological diversity, implying that ocean acidiﬁcation may offset beneﬁts from conservation/management strategies relying on the regulation of resource exploitation.


Introduction
Global ocean anthropogenic carbon inventories suggest that the ocean took up a cumulative ∼ 155 ± 31 Pg C (10 15 g of carbon) in 2010 (Khatiwala et al., 2013). This uptake of CO 2 is causing profound changes in seawater chemistry resulting from increased hydrogen ion concentration (decrease in pH, pH = −log10[H + ]), referred to as ocean acidification (IPCC, 2011). Experimental and modelling studies provide compelling evidence that ocean acidification will put marine ecosystems at risk (e.g. Orr et al., 2005;Kroeker et al., 2013). However, with the exception of assessments focusing on cold-water coral systems (Barry et al., 2005(Barry et al., , 2013Fleeger et al., 2006;Guinotte et al., 2006;Tittensor et al., 2010), quantifications of biological consequences of ocean acidification mostly targeted surface ocean or coastal environments (Kroeker et al., 2010).The aim of this study is to extend our understanding of broad scale impacts of ocean acidification from the existing shallow water studies to focus specifically on deep-sea ecosystems. The deep sea is under increasing anthropogenic pressure as technological advances allow for exploitation of formerly inaccessible regions (Clauss and Hoog, 2002). While waste disposal, fishing and, in the future, mineral extraction are well recognized as dominant human pressures (Ramirez-Llodra et al., 2011), expert assessments urge consideration of climate change and ocean acidification impacts in future ecosystem conservation/management strategies (Taranto et al., 2012;Billé et al., 2013).
While previous studies quantified changes in carbonate mineral saturation state as a measure for potential detrimental impacts on deep calcifying communities (Guinotte et al., 2005(Guinotte et al., , 2006Turley et al., 2007;Fautin et al., 2009), this model-based assessment uses pH. The tight control of pH at the cellular scale is an important prerequisite of proper cell functioning, and mechanisms of pH control are ubiquitous across many taxa (Seibel and Walsh, 2003, and references therein). Deep-sea organisms might be particularly vulnerable to changes in seawater chemistry, at least in part owing to limitations on rate processes, caused by low temperature (Childress, 1995;Seibel and Walsh, 2001) and possibly food availability (Ramirez-Llodra, 2002), as well as the environmental stability of their habitat in the past (Barry et al., 2011;Seibel and Walsh, 2003). A recent review by Somero (2012) highlights the link between environmental stability and the capacity to acclimate to future changes in environmental variables such as pH. According to this study, environmental stability might impair the potential for acclimation. This stands in sharp contrast to shallow water or intertidal organisms, which are adapted to a dynamic environment with large changes in temperature and seawater chemistry (Hofmann et al., 2011;Duarte et al., 2013).
A model sensitivity study by Gehlen et al. (2008) suggested the potential for large pH reductions (up to −0.6 pH units) in the deep North Atlantic. Regions of large pH reductions coincided with areas of deep-water formation. Deep-water formation drives the rapid propagation of surfacederived changes in carbonate chemistry to depth as underlined by high vertically integrated water column inventories of anthropogenic carbon (Sabine et al., 2004), as well as tritium and chlorofluorocarbon distributions (Doney and Jenkins, 1994). Gehlen et al. (2008) used output from a single model and for a scenario following an atmospheric CO 2 increase of 1 % per year over 140 years starting from an atmospheric CO 2 level of 286 ppm. This rate of increase is about twice as large as the rate typical for a high-end IPCC concentration pathway. The study did not include circulation changes in response to climate change.
Here we extend the study by Gehlen et al. (2008) by analysing pH projections from seven Earth system models that contributed to the Coupled Model Intercomparison Project Phase 5 (CMIP5) and for four different Representative Concentration Pathways (RCPs;Van Vuuren et al., 2011) ranging from a strong emission mitigation scenario (RCP2.6) to the high-CO 2 scenario RCP8.5. We assess the magnitude of deep-water pH reductions in the North Atlantic (35-75 • N, 0-90 • W) over this century in response to atmospheric CO 2 increase and climate change. The North Atlantic is a wellventilated region of the world ocean and, despite a projected increase in stratification, will remain well oxygenated in the future (Bopp et al., 2013). The study complements assessments by Bopp et al. (2013) and Mora et al. (2013), which evaluated large-scale average pH reductions in response to the same RCP pathways, but without a detailed discussion of spatial patterns and their link to circulation. We define a critical threshold for pH reductions based on evidence from palaeo-oceanographic studies, contemporary observations and model results. Future multi-model projections of pH changes over the seafloor are analysed with reference to this threshold and without discrimination of particular habitats first. Next, model results are put into the perspective of ecosystem conservation by evaluating changes in pH against the distribution of seamounts and deep-sea canyons. These features are known as sites of high-biodiversity deep-sea ecosystems, such as cold-water corals and sponge communities (ICES, 2007;Clark et al., 2010;De Leo et al., 2010) and are selected as representative examples of deep-sea environments.

Earth system models
Our study draws on results from two types of Earth system models: (1) the Bern3-D-LPJ carbon-cycle/climate model (Steinacher et al., 2013;Roth and Joos, 2013) and (2) seven fully coupled three-dimensional atmosphere ocean climate models that participated in the Coupled Model Intercomparison Project Phase 5 (CMIP5; Taylor et al., 2011) and contributed to the Fifth Assessment Report of the Inter-governmental Panel on Climate Change (IPCC AR5). The Bern3-D-LPJ is a model of intermediate complexity featuring a 3-D geostrophic-balance ocean and 2-D atmospheric energy and moisture-balance model. The cycle of carbon and related tracers is represented including prognostic formulations for marine production, a seafloor sediment and a dynamic global vegetation model. This model is relatively costefficient compared to CMIP5 models. It is used to evaluate the order of magnitude of pH reductions associated with past abrupt climate change by analysing results from freshwater hosing experiments (Bryan, 1986;Marchal et al., 1999;Matsumoto and Yokoyama, 2013).
Concerning the subset of CMIP5 models, we selected models for which 3-D pH fields were available and that had been part of a published multi-model evaluation (Bopp et al., 2013). We analyse output for four future atmospheric CO2 concentration scenarios (Representative Concentration Pathways, RCPs), along with the corresponding pre-industrial control simulations, piControl. The nomenclature follows CMIP5 recommendations. Historical simulations cover the period between 1870 and 2005 and are followed by climate change scenarios according to RCP8.5, RCP6.0, RCP4.5 and RCP2.6 from 2006 to 2100 (Van Vuuren et al., 2011;Moss et al., 2010). RCP identifiers refer to the additional radiative forcing in 2100 relative to pre-industrial (or before 2100 in case of the peak-and-decline scenario RCP2.6). These additional radiative forcings correspond to atmospheric CO 2 levels in 2100 of 421, 538, 670 and 936 ppm for RCP2.6, RCP4.5, RCP6.0 and RCP8.5. Individual RCPs differ with respect to the temporal evolution of atmospheric CO 2 and range from a stringent emission mitigation RCP2.6 to the high-CO 2 scenario RCP8.5. The complete set of RCPs was not available for all models. Please refer to Table S1 (Supplement) for model name, scenario and references.

Deep-sea ecosystems
This study uses data sets of seamounts (Yesson et al., 2011) and canyons (Harris and Whiteway, 2011). For seamounts, these data include location, height and surface assuming a conical shape. For canyons, the data consist of a highresolution vector database of canyon centre lines that was converted into a raster data set of canyon presence (using ArcGIS v10) for analysis. Data were projected on a 1 • × 1 • regular grid.

Post-treatment of CMIP5 model output
Model output is interpolated on a regular grid of 1 • × 1 • resolution. Anomalies are computed as the difference between the decade 2090-2099 and the long-term mean of the preindustrial state. As the focus of this study is on impacts on benthic communities, we quantify pH changes in the deepest model box over a topography range from 500 to > 4500 m water depth.

Computation of the area of seamounts for impact assessment
The area of North Atlantic seafloor impacted by ocean acidification is estimated on the basis of individual grid cells for which the reduction in pH exceeded ≥ 0.2 or 0.3 units. The impacted area follows as the integral of the area of these 1 • × 1 • grid cells. The area of seamounts with a pH reduction ≥ 0.2 or 0.3 units is computed based on distribution and height assuming a conical shape (Danovaro et al., 2008, Yesson et al., 2011. The database provides height above seafloor and base area. The area of the seamount (A) is given by where r is the base radius of the seamount and h + h is the height. The height impacted by a pH reduction exceeding the threshold (h ) is diagnosed from the depth of the pH anomaly corresponding to the threshold. The radius of the seamount at the depth of the anomaly (r ) is obtained from the Thales theorem: The final expression of r is the positive analytical solution of the fourth-order polynomial The impacted area of the seamount (A * ) follows from the depth of pH anomaly as a function of seamount height: where A is the total surface area of the seamount.

Environmental stability and critical threshold for pH reduction
Considering that environmental stability might impair the potential for acclimation, we assessed pH changes over glacialinterglacial timescales and for past events of rapid climate changes recognized for having driven major reorganizations in North Atlantic circulation and carbonate chemistry. The pH is defined as the negative logarithm of the hydrogen ion concentration ([H + ]). From the basic properties of logarithms it follows that the difference in pH equals the logarithm of the ratio of hydrogen ion concentrations.  Fig. S1). Contrasting shallow and deep environments highlights that absolute changes in [H + ] are amplified at depth for any threshold, that is for environments of low natural variability.

Glacial-interglacial timescales
The palaeo-record permits evaluation of environmental variability of the deep ocean over the past million years. Avail-able evidence indicates a low variability over this time interval (Elderfield et al., 2012;Yu et al., 2010Yu et al., , 2013. Changes in carbonate chemistry were small in the deep ocean compared to surface layers (Yu et al., 2010). Recent studies re-evaluated deep-water pH changes between glacial and present (Sanyal et al., 1995), arguing that carbonate compensation kept deepwater pH close to constant (Hönisch et al., 2008). We use data available in Yu et al. (2010) (and associated supplement) and follow their reasoning to infer dissolved inorganic carbon (DIC) changes from [CO 2− 3 ] and hence alkalinity in order to compute associated changes in pH for sediment core BOFS 8K (52.5 • N, 22.1 • W, 4045 m). This pH change is computed using CO 2 sys (http://cdiac.ornl.gov/ oceans/co2rprt.html) with alkalinity and DIC as input variables, along with temperature, depth, phosphate and silicate (Yu et al., 2010). We estimate a pH reduction of ∼ 0.1 pH units for North Atlantic deep water over the early deglacial (17 500 to 14 500 years before present).

Rapid events associated with freshwater release: Heinrich and Dansgaard Oeschger events
Model experiments yield maximum pH reductions in North Atlantic deep water below 0.15 pH units in response to a shutdown of the North Atlantic Meridional Overturning Circulation (AMOC, Fig. 1). To realize an abrupt shutdown of the AMOC, different durations of freshwater perturbations in the North Atlantic on top of a pre-industrial steady state have been tested releasing a total of 3 × 10 15 m 3 freshwater (∼ 9 m sea level equivalent). In terms of pH changes in the North Atlantic region, the experiment with a 300-year freshwater forcing of 0.33 Sv results in the strongest response (Fig. 1a).
In these experiments, the cause of the pH decrease is not high atmospheric CO 2 (CO 2 only increases a few ppm during the freshwater experiment), but is mainly a result of the decrease in deep-ocean ventilation. This leads to the additional accumulation of DIC by the respiration of organic matter. Although alkalinity is also increased in the deep by the dissolution of carbonate particles settling through the water column, it does not compensate for the increase in DIC leading to more acidic waters in the deep. The most extreme negative excursion of the pH averaged over the deep (below 2000 m) North Atlantic (45-65 • N) occurs ∼ 150 years after the end of the freshwater forcing with a decrease of ∼ 0.13 pH units relative to the unperturbed pre-industrial state (Fig. 1b). The pH decrease does not exceed −0.18 pH units in any of the individual grid boxes. In Fig. 1c and d the spatial distribution of the pH-reduction averaged over years 400-450 (i.e. during the maximum of the pH decrease) is shown in terms of pH anomalies at the seafloor and in a section through the Atlantic at 38.5 • W.

Critical threshold for pH reductions
For the purpose of evaluating the potential for negative impacts on deep-sea benthic environments, a critical threshold for pH reduction needs to be identified. Reductions of pH exceeding the envelope set by past and present natural variability are considered as critical. Palaeo-evidence suggests that the deep-sea fauna has evolved under conditions of environmental variability confined to a narrow range over the past million years (Yu et al., 2010;Elderfield et al., 2012). Many past episodes of climate change occurred over significantly longer timescales than the current anthropogenic perturbation of the climate system, allowing carbonate compensation to keep deep-water pH close to constant (Hönisch et al., 2008). This is corroborated by computing pH reduction over glacial-interglacial cycles for a North Atlantic site. Decadal-to-centennial changes are addressed by freshwater hosing model experiments to simulate effects of circulation changes associated with rapid Heinrich and Dansgaard Oeschger events. In both cases, pH reductions are below 0.15 pH units. Similarly, small amplitude natural temporal pH variability at depth emerges from a multi-annual time series station (González-Dávila et al., 2010) and the analysis of the long pre-industrial simulation "piControl" (Fig. S2).
In summary, natural pH variations on multi-annual, decadalto-century, and longer timescales were likely smaller than 0.2 pH units on the regional-to-basin scale in the deep Atlantic and at least for the past million years.
This suggests that pH variations of up to 0.1 to 0.2 pH units do not present a risk for marine life. This leads us to define two thresholds for pH reduction between pre-industrial and the end of the 21st century: −0.2 and −0.3 pH units. Both stand for pH reductions exceeding palaeo-record-based estimates of changes in North Atlantic deep-water chemistry over the past 10 000 years, as well as being much larger than the amplitude of natural temporal variability of pH in the deep North Atlantic (González-Dávila et al., 2010). The first threshold (−0.2) is in line with recommendations by environmental agencies (Schubert et al., 2006) following the precautionary principle, and is reported to increase mortality of deep-sea benthic organisms during in situ exposure experiments (Barry et al., 2005). The second threshold (−0.3) allows for bracketing of a range of changes spanning from an ∼ 58 % increase in hydrogen ion concentration up to ∼ 100 %.

Projections of pH reductions over the 21st century
Time series of atmospheric CO 2 (ppm) for three out of four IPCC RCP scenarios between 2006 and 2100 show an increase in CO 2 over this century; only RCP2.6 does not show a general increase with time (Fig. 2a). The corresponding simulated pH reductions for surface and deep North Atlantic waters are presented in Fig. 2b and c. Projected pH changes are indicated as the multi-model mean along with the between-model spread. Monitoring at time series stations reveals that the observed surface ocean pH decrease tracks increasing atmospheric pCO 2 (Orr, 2011). This trend is confirmed by the decline in simulated surface ocean pH (Fig. 2b) with a small between-model spread. In the surface ocean the extent of ocean acidification is set by the atmospheric CO 2 trajectory, along with physical climate change, namely warming and associated changes in ocean circulation and CO 2 thermodynamic properties. Surface waters, with high levels of dissolved anthropogenic CO 2 and characterized by low pH values, are entrained to the interior ocean during seasonal mixed layer deepening and deep convection episodes. As a result, deep pH changes (Fig. 2c) reflect atmospheric CO 2 to a lesser extent. Because the deep-water formation differs between models, the inter-model spread is significantly larger in deep waters than for the surface ocean.
The spatial pattern of pH reductions is exemplified for RCP4.5 and RCP8.5 in Fig. 3 (Fig. S3 for RCP2.6 and RCP6.0). Under RCP4.5 (Fig. 3a) and RCP8.5 (Fig. 3b), pH reductions crossing the −0.2 threshold are projected for continental slopes and a latitudinal band extending from 55 to 65 • N. Since the pH perturbation originates at the sea surface, the continental slope and topographic heights (e.g. mid-Atlantic ridge) experience the largest pH reductions. Increasing impact on the sea floor between RCP4.5 and RCP8.5 for a threshold of −0.2 reflects the depth exposure to the pH perturbation of continental slopes and the mid-Atlantic ridge. In summary, the spatial pattern is set by a combination of topography and North Atlantic circulation pathways. It reflects the transfer of the surface born anomaly of pH to the ocean interior during deep-water formation and downstream transport away from convection sites by the deep western boundary current.
By the end of the 21st century, projected pH reductions (Table 1) (Table 1). We expect, however, an increase in impacted area for all scenarios and pH thresholds beyond 2100 in response to legacy effects of CO 2 emissions and ongoing downward propagation of the pH perturbation (Frölicher and Joos, 2010).

Opposing effects of climate change and ocean acidification
The progression from RCP2.6 to RCP8.5 corresponds to a series of increasing geochemical (atmospheric pCO 2 ) and physical (climate change, defined here as changes in ocean dynamics in response to atmospheric warming) forcing with opposing effects on deep-ocean acidification. In order to distinguish between the physical and geochemical drivers of North Atlantic deep-water acidification, we assessed two contrasting simulations available for two Earth system models (GFDL-ESM2M and IPSL-CM5A-LR) for RCP4.5. The first simulation (Fig. 4a) includes climate change effects on ocean circulation and geochemical effects on the seawater CO 2 system in response to atmospheric pCO 2 increase (RCP4.5). In the second experiment (Fig. 4b), the circulation and ocean physics are kept at pre-industrial conditions, but atmospheric CO 2 levels following RCP4.5 are used to force ocean acidification (RCP4.5/fixclim). The difference in pH between RCP4.5 and RCP4.5/fixclim (Fig. 4c) allows, at first order and within the limits of non-linearities (Schwinger et al., 2014), for isolation of the effect of climate change on pH changes. The negative differences in pH on panel c indicate stronger acidification in RCP4.5/fixclim, and suggest a slight alleviation of ocean acidification at depth and over the timescale of this study by climate change. In the experiment where ocean circulation was held at pre-industrial conditions (RCP4.5/fixclim), there was a small increase in the area impacted by pH re-ductions for all thresholds (Table 1). Largest differences in projected pH values between RCP4.5 and RCP4.5/fixclim co-occur with large negative anomalies in winter mixed layer depth maxima in the Labrador Sea and negative pH anomalies downstream of convection sites following the deep western boundary current (Doney and Jenkins, 1994). This is in line with the projected enhancement of stratification across the North Atlantic in response to increasing temperatures and freshening. It will result in changes in winter mixed layer depth and deep convection and a decrease in the Atlantic Meridional Overturning Circulation (Meehl et al., 2007;Cheng et al., 2013). While increasing atmospheric CO 2 reduces pH, increasing climate change reduces surfaceto-deep water exchange. In addition, topography modulates the extent of deep-water acidification. The combination of climate change, the non-linearities of the carbonate system, and topography explains the levelling-off of impacts in Table  1 for pH reductions exceeding −0.2.

Projected impacts on ecosystems
In order to evaluate the risk for specific benthic ecosystems to be affected by pH reductions, we co-located seamounts (Fig. 3, black dots) and deep-sea canyons (Fig. 3, red dots) -both of which are key habitats of high biodiversity -and pH changes for RCP4.5 and RCP8.5 separately computed from the multi-model mean (see Supplement for RCP2.6 and RCP6.0). To further the evaluation of potential impacts of pH reductions beyond pH thresholds, we computed the area of seamounts for which a corresponding decrease is projected. A significant proportion of these habitats will be impacted by pH reductions exceeding −0.2 units by the end of the 21st century under moderate to high emission scenarios (Fig. 5). The geographic pattern results in close to 22.5 ± 5.3 % (14.7 ± 4.1 %) of North Atlantic deep-sea canyons and 7.7 ± 3.6 % (2.7 ± 0.9 %) of seamount ecosystems being exposed to pH reductions exceeding −0.2 (−0.3) units under RCP8.5. Under the moderate scenario, RCP4.5, model projections indicate that 19.0 ± 5.7 % of deep-sea canyons and 3.5 ± 1.6 % of seamounts still will experience pH reductions exceeding the −0.2 threshold. The close to constant impact reflects the use of a diagnostic that is based on counts of features being impacted, in addition to the depth distribution and propagation of the pH anomaly.
Seamounts and deep-sea canyons are known as hotspots of biodiversity and harbour a variety of distinct communities including reef-building cold-water corals, soft coral gardens and deep-sea sponge aggregations (Buhl-Mortensen et al., 2010, 2012Clark et al., 2010;De Leo et al., 2010;ICES, 2007). Recent assessments reveal a high level of anthropogenic pressures on these ecosystems Ramirez-Llodra et al., 2011). While fishing and resource extraction are recognized as the dominant human pressures at present and in the near future, expert assessments highlight the need for an appropriate quantification of the impacts of climate change and ocean acidification (Taranto et al., 2012). Present international conservation targets aim at preserving 10 % of marine biomes by 2020 (Convention of Biodiversity, 2011   communities are difficult to evaluate owing to the lack of experimental and observational data. Rapid changes in pH will likely lead to disruption of extracellular acid-base balance, impedance of calcification and other physiological effects in deep-water organisms, and whatever acclimation is required may have increased energetic costs (Widdicombe and Spicer, 2008) -e.g. for metabolism/maintenance, growth, reproduction -and could extend to increases in mortality of both 3. Impact is computed as the fraction of the surface area affected by a reduction exceeding the threshold for seamounts, as well as the number of canyons surrounded by waters for which the reduction in pH exceeding the threshold is projected. Model pH is the decadal mean (2090-2100). Note that the seamount and canyon multi-model averages for the RCP2.6 scenario overlie each other. Light coloured circles: values obtained for each Earth system model; dark coloured circles: multi-model average for each scenario. Vertical and horizontal bars: 2.5-97.5 % confidence interval of multi-model averages.
adults and juveniles. Changes at the individual and population level will inevitably lead to more widespread ecosystem and community level changes and potential shifts in biodiversity (Hendriks et al., 2010) and ecosystem functioning (Danovaro et al., 2008). Biodiversity reductions could arise from a loss of species, or even functional or taxonomic groups, sensitive to pH change. The ecological implications of pH change could be more severe if keystone or habitat-forming species are impacted (Widdicombe and Spicer, 2008), which seems likely (Guinotte et al., 2006). These effects may be likely exacerbated in the presence of other stressors (Walther et al., 2009), such as global warming and projected reductions in deep-sea food supply (Bopp et al., 2013), as well as elevated resource exploitation and pollution. In particular, reductions in food supply to deep benthic communities are projected to result in a decrease in biomass and a shift towards smaller sized organisms . These changes will modify energy transfer rates through benthic food webs and may leave communities more susceptible to pH reductions. We propose these and future model projections be taken into account when defining longterm preservation and management approaches to deep-sea ecosystems.

Conclusions
This study assesses the potential for detrimental pH reduction to occur across the deep North Atlantic by the end of the 21st century. It evaluates results from seven fully coupled Earth system models and for four Representative Concentration Pathways ranging from RCP2.6 to RCP8.5. In three out of the four scenarios, the multi-model analysis suggests that by 2100 over 17 % of the seafloor area below 500 m depth in the North Atlantic sector will experience pH reductions exceeding −0.2 units. Enhanced stratification in response to warming and freshening of surface waters slightly counteracts deep-water acidification. pH reductions co-occur with sites of high deep-sea biodiversity such as seamounts and canyons. Model projections indicate that by the end of this century and for the high CO 2 scenario RCP8.5, close to 23 % (∼ 15 %) of North Atlantic deep-sea canyons and ∼ 8 % (3 %) of seamounts will experience pH reductions exceeding −0.2 (−0.3) units. Seamounts proposed as sites of marine protected areas are potentially threatened by these pH reductions. The spatial pattern of impacts reflects the depth of the pH perturbation and does not scale linearly with atmospheric CO 2 concentration. Impacts may cause negative changes of the same magnitude or exceeding the current biodiversity target of 10 % of preservation of marine biomes, implying that ocean acidification may offset benefits from conservation/management strategies relying on the regulation of resource exploitation.
The Supplement related to this article is available online at doi:10.5194/bg-11-6955-2014-supplement.