Centre for Marine Biodiversity and Biotechnology School of Life Sciences, Heriot- Watt University, Edinburgh, EH14 4AS, United Kingdom

Transcription

1 Oceanography and Marine Biology: An Annual Review, 2012, 50, R. N. Gibson, R. J. A. Atkinson, J. D. M. Gordon, R. N. Hughes, D. J. Hughes and I. P. Smith, Editors Taylor & Francis Benthic invertebrates in a high- co 2 world Laura C. Wicks 1 & J. Murray Roberts 1,2,3 1 Centre for Marine Biodiversity and Biotechnology School of Life Sciences, Heriot- Watt University, Edinburgh, EH14 4AS, United Kingdom 2 Center for Marine Science, University of North Carolina Wilmington, 601 S. College Road, Wilmington, NC , United States of America 3 Scottish Association for Marine Science, Scottish Marine Institute, Oban, Argyll, PA37 IQA, United Kingdom E- mail: L.Wicks@hw.ac.uk; J.M.Roberts@hw.ac.uk Abstract Ocean acidification (OA), whereby increases in atmospheric carbon dioxide (CO 2 ) over the past 200 years have led to a decline in the ph and carbonate ion availability of the oceans, has emerged as one of the major drivers of twenty- first century marine scientific research. Here we describe the current understanding of OA effects on benthic marine invertebrates, in particular the calcifiers thought to be most sensitive to altered carbonate chemistry. We describe the responses of benthic invertebrates to OA conditions predicted up to the end of the century, examining individual organism response through to ecosystem- level impacts. Research over the past decade has found great variability in the physiological and functional response of different species and communities to OA, with further variability evident between life stages. Over both geological and recent timescales, the presence and calcification rates of marine calcifiers have been inextricably linked to the carbon chemistry of the oceans. Under short- term experimentally enhanced CO 2 conditions, many organisms have shown trade- offs in their physiological responses, such as reductions in calcification rate and reproductive output. In addition, carry- over effects from fertilization, larval and juvenile stages, such as enhanced development time and morphological changes, highlight the need for broad- scale studies over multiple life stages. These organism- level responses may propagate through to altered benthic communities under naturally enhanced CO 2 conditions, evident in studies of upwelling regions and at shallow- water volcanic CO 2 vents. Only by establishing which benthic invertebrates have the ability to acclimate or adapt, via natural selection, to changes from OA, in combination with other environmental stressors, can we begin to predict the consequences of future climate change for these communities. Introduction Historically, the oceans have been viewed as so vast that they are immune to anthropogenic interference, but this is no longer the case. Evidence of how human activities have irreversibly altered the chemistry and functioning of the oceans is mounting. For marine science, much of the research focus in recent years has been on the ramifications of oceanic warming, caused by increased atmospheric CO 2 from anthropogenic activities, such as the burning of fossil fuels. Increases in global temperature have already resulted in polewards shifts in species geographical distributions (Mieszkowska et al. 2006) and alterations in species interactions (Pörtner 2008). However, there is another potentially greater problem facing marine life, the so- called other CO 2 problem (Doney et al. 2009) of ocean acidification (OA). Increased CO 2 dissolving into the oceans has led to a decline in the ph and 127

2 Laura C. Wicks & J. Murray Roberts Year Atmospheric CO 2 (ppmv) Atmospheric CO 2 Historical IS92a S650 Carbonate (µmol kg 1 ) ph ph CO 3 2 Aragonite saturation Preind S 2100 I Preind S 2100 I 0 80 S 40 S Calcite saturation 0 40 N 80 N Latitude Figure 1 Predicted changes in future atmospheric pco 2, ocean ph and ocean carbonate concentrations over a latitudinal scale compared to preindustrial. I refers to the Intergovernmental Panel on Climate Change (IPCC) IS92a continually increasing scenario, S to the IPCC S650 stabilization scenario. (From Orr et al with permission from Macmillan Publishers.) carbonate ion concentration of the oceans, with profound effects predicted for marine bio diversity, trophic interactions and other ecosystem processes (Raven et al. 2005). Furthermore, as OA conditions continue into the future (Figure 1), its effects are potentially compounded by the warming of the oceans, creating a new environment for many, if not all, marine species. Although understanding the implications of such changes is in its infancy, concentrated efforts across the globe have seen the number of OA publications increase from less than 20 in the 1990s to more than 800 by 2010 (ISI Web of Knowledge, search conducted on 1 December 2010). The first deliberate OA experiment was conducted on the coralline alga Porolithon gardineri in the mid-1980s (Agegian 1985), with the detrimental effects of OA on coral calcification addressed in the late 1990s (Gattuso et al. 1998). However, it has only been in the last few years that we have 128

3 Benthic invertebrates in a high- co 2 world seen well- designed experiments conducted using realistic pco 2 scenarios on a variety of organisms across a range of habitats. This review addresses the effects of OA on the benthos. Globally, benthic invertebrates make up a significant component of life on Earth, with many species having a wide latitudinal distribution and inhabiting regions of variable ph. These invertebrates are potentially vulnerable to OA due to their sessile or slow- moving nature. The dispersal capability of the larvae of sessile organisms may act as a mechanism by which they can escape to environments more physiologically suitable. However, for long- lived organisms, this mechanism will be restricted. For example, 70% of deep- water corals, which form extensive reefs in the deep oceans around the world, are likely to be unable to form their carbonate skeletons by the end of the century, and their slow growth rates mean they will be unable to escape the shallowing corrosive waters (Guinotte et al. 2006). Conversely, many shallow- water invertebrates may be able to tolerate or adapt to OA conditions by virtue of the inherently variable chemistry of the intertidal (Segal 1961). The physiological variability between invertebrate species on geographical, temporal, depth and habitat scales means that responses to OA are complex and not easily summarized or meta- analyzed. Furthermore, many responses are likely to be chronic; rather than OA causing mortality, more subtle effects such as decreased reproductive effort are probable, with potentially devastating long- term consequences. To this end, this review presents a synopsis of research conducted on benthic invertebrates, examining the responses of their energy budgets to OA and what the rest of the century may hold for these animals and ecosystems. Following explanation of the basic chemistry underlying oceanic changes into the future and approaches currently used to assess the effect of such changes, gaps in the knowledge of OA effects are highlighted to emphasize the need for further studies on the subtle effects of our changing ocean. Carbonate chemistry of the ocean: past, present and future The carbon cycle is one of the fundamental processes in the functioning of the Earth s system, and through the production and decomposition of organic matter, the cycle is closely coupled with many other elemental cycles. To understand the impact on organisms from increased CO 2 in the world s oceans, we must first understand the dynamics, history and future of carbon- dependent chemical processes in the oceans. Carbonate chemistry Carbon, in many forms, continually cycles between the atmosphere and the oceans, with the rate and extent of diffusive input to the oceans governed by a complex suite of natural and anthropogenic processes. Atmospheric CO 2 is provided by volcanic outgassing, combustion of organic material, weathering of silicate rock and the respiration of living organisms. As atmospheric CO 2 equilibrates rapidly with surface water, the oceans are the largest carbon reservoir on the surface of the earth. Regions of downwelling transfer carbon from the surface to the deep ocean, where further carbon is also regenerated by the oxidative destruction of organic matter. Dissolved CO 2 is added back to the atmosphere from warm surface waters in regions of upwelling, with the whole system acting as a recirculating pump. On a latitudinal scale, the fluxes of CO 2 between the ocean and atmosphere vary due to the increased solubility of CO 2 at lower temperatures. In the polar surface waters, CO 2 is transported more readily to the deep ocean, whereas in tropical waters there is a net transport of CO 2 from the surface to the atmosphere. There is relatively large spatial variability in these fluxes, with areas of upwelling bringing CO 2 -rich intermediate waters to the surface (Feely et al. 2008). The inorganic carbon cycle is largely responsible for controlling the ph of seawater, with the specific form of carbon in the ocean controlled by a series of chemical reactions, which are locally in equilibrium (Figure 2A). Transitions between the different forms of carbon take place in the 129

4 Laura C. Wicks & J. Murray Roberts (A) CO 2 CO 2 + H 2 O H 2 CO 3 H + CO 3 2 (B) HCO 3 HCO 3 External seawater Z Mitochondria Zooxanthellae Oral tissue Anion exchanger H + /Ca 2+ exchanger Z Z CO 2 H + + OH Coelenteron HCO 3 H + Aboral tissue CO 2 + H 2 O HCO 3 + H + A + Skeleton HCO CaCO 3 + H 3 + Ca2+ Figure 2 Pathways of carbon from the atmosphere to the coral skeleton. (A) The chemical equilibria of carbon dioxide in seawater. (B) Model of inorganic carbon entering the coral tissue (solid arrows) and H + (broken arrows) fluxes associated with zooxanthellae photosynthesis and coral host calcification. (Adapted from Brownlee 2009 with permission.) It is proposed that zooxanthellae in the oral endodermal cells of the noncalcifying parts of the coral use CO 2 produced in the endodermal cells via the equilibration of HCO 3 - that diffuses from the external medium. In the calcifying aboral endodermal cells, it is proposed that respirationderived CO 2 and HCO 3 - uptake from the coelenteron provide the inorganic carbon source for calcification. The calcification reaction may be driven by HCO 3 - anion antiporters and ATP- driven H + /Ca 2+ antiporters in the calicoblastic epithelium adjacent to the extracellular site of calcite deposition. These fluxes will ensure an appropriately high ph and supply of Ca 2+ and HCO 3 - for the deposition of CaCO 3. Further details are given in Brownlee (2009). direction that tends to maintain a constant ph, with the equilibria strongly dependent on temperature, pressure, salinity, and the presence of other ions, particularly borate (Figure 3). At normal seawater ph (8.1), the favoured form of inorganic carbon in surface seawater is bicarbonate (HCO 3 ), making up 88.6% of inorganic carbon in the oceans, with 10.9% of the ocean s carbon present as carbonate (CO 3 2 ) and 0.5% as carbon dioxide (CO 2 ) in aqueous form. Speciation of carbon is also altered by biotic processes that consume or release the different carbon species, altering the 130

5 Benthic invertebrates in a high- co 2 world 2 pk 1 (ref.) pk 2 (ref.) 2.5 CO 2 HCO 3 CO 3 2 Log [concentration (mol kg 1 )] T = 25 C, S = 0 T = 0 C, S = 35 T = 25 C, S = 35 T = 0 C, S = 35, P = 300 atm ph Figure 3 Bjerrum plot illustrating the effect of temperature, pressure and salinity on carbon speciation in seawater. The reference case is T = 25 C, S = 35 and P = 1 atm (solid line), with DIC at 2 mmol kg 1 in all cases. pk 1 and pk 2 refer to the first and second dissociation constants of carbonic acid, respectively. (Reprinted from Zeebe & Wolf-Gladrow 2001 with permission from Elsevier.) equilibrium usually on short timescales at small spatial scales. For example, photosynthesis and calcification reduce dissolved inorganic carbon (DIC), whereas respiration increases DIC (Figure 4). The equilibrium of the carbonate system is also a function of the total alkalinity (TA), which is a measure of the acid/base balance of the fluid, defined conventionally as the excess of proton acceptors (bases formed from weak acids) over proton donors relative to a reference point and therefore its buffering capacity (discussed in detail by Zeebe & Wolf- Gladrow 2001). Alkalinity varies regionally, and as such, some regions will have a greater buffering capacity for alterations in the carbon cycle. On small spatial scales, TA is also altered by biotic processes; CaCO 3 formation reduces TA as well as DIC, shifting a system to higher CO 2 levels and reduced ph (Figure 4; Zeebe & Wolf- Gladrow 2001). Calcium carbonate and calcification The availability of carbonate ions in the ocean is important for calcifying species, which combine these ions with calcium to form their biogenic calcium carbonate skeletons or shells: co Ca 2+ CaCO 3 CaCO 3 crystals are nucleated and grown in an isolated or semi- isolated internal compartment, separate from ambient seawater (Tambutté et al. 2007). In benthic invertebrates, the process of calcification primarily occurs within compartments created between the tissue and the existing skeleton or shell, while some organisms, including gorgonian corals, produce their CaCO 3 intracellularly (Goldberg & Benayahu 1987). While calcification is one of the most fundamental processes regulating the ocean carbon cycle, the precise cellular and molecular mechanisms controlling 131

6 Laura C. Wicks & J. Murray Roberts Total alkalinity (mmol kg 1 ) ph [CO 2 ] Photosynthesis CO 2 release CaCO 3 dissolution CO 2 invasion Respiration CaCO 3 formation DIC (mmol kg 1 ) Figure 4 Effect of various processes on dissolved inorganic carbon (DIC) and total alkalinity (TA) (arrows). Solid and dashed lines indicate levels of constant dissolved CO 2 (in mmol kg 1 ) and ph, respectively, as a function of DIC and TA. (Reprinted from Zeebe & Wolf- Gladrow 2001 with permission from Elsevier.) biocalcification and internal ph regulation remain poorly understood (Wilt et al. 2003, Allemand et al. 2004, Cohen & Holcomb 2009), with increased knowledge of these complex processes becoming a high priority in light of OA. It is known that calcification is highly controlled and energetically costly, as the organism must modify and regulate the conditions of the calcifying fluid within the calcifying space. Seawater is the likely starting fluid for calcification in corals (Cohen et al. 2001, Gaetani & Cohen 2006, Cohen & Holcomb 2009, Holcomb et al. 2009). It is suggested that molluscs and crustaceans nucleate their shells from discrete calcifying fluids (Cameron 1985), known for the former as pallial fluid (e.g., Crenshaw 1972), although haemocytes have also been suggested to play a role in shell formation (Mount et al. 2004). In addition, echinoids are thought to initiate calcification on Ca 2+ -binding organic matrices within cellular vacuoles (Ameye et al. 1998). Two general models of coral calcification exist: the physiochemical model and organic matrix model. The former proposes that calcification takes place in an extracellular space beneath calicoblastic cells of the coral ectoderm (reviewed in Cohen & McConnaughey 2003), whilst the organic matrix model proposes that calcification is mediated via an organic matrix secreted by the coral (Clode & Marshall 2002, Tambutté et al. 2007). Once seawater is transported to the site of calcification, although the route remains unclear (Braun & Erez 2004), corals elevate the saturation state of this seawater by the influx of calcium ions and the efflux of protons to generate an extracellular microenvironment with elevated calcium carbonate saturation states that favour calcification (Cohen & McConnaughey 2003, Pomar & Hallock 2008). These processes are believed to involve Ca 2+ -ATPase (adenosine triphosphatase), proton pumps, CO 2 consumption via photosynthesis in zooxanthellate corals, and other enzymes (Borowitzka & Larkum 1976, Cohen & McConnaughey 2003, Zoccola et al. 2004, Cohen & Holcomb 2009). By increasing the concentration of calcium 132

7 Benthic invertebrates in a high- co 2 world ions at the calcifying site as well as the intracellular ph within the calcifying space, an increase in the aragonite saturation state of the calcifying fluid occurs (5 10 times that of seawater; Al- Horani 2003). This increase suggests that the elevation in fluid saturation state in the coral s calcifying compartment is driven largely by an increase in carbonate ion concentration and to a lesser extent by the change in calcium ion concentration. Although saturation state is a key factor in coral calcification, many other marine calcifiers can calcify in undersaturated waters (e.g., the mussel Mytilus edulis, Thomsen et al. 2010). Regardless of the processes involved, it is known that an organism s ability to control ph will be important in determining how it will respond to changes in external seawater ph. The flux of carbon species through an organism will clearly affect the ph of the intra- and extracellular fluids, indicating that complex mechanisms must come into play to enable regulation of calcification. Microelectrode measurements reveal that the ph inside coral cells containing the dinoflagellate symbiont Symbiodinium varies from 7.4 to 7.1 (Venn et al. 2009). However, the calcifying fluid of the temperate coral Astrangia poculata was found to be substantially elevated relative to external seawater ph under both control and acidified conditions, consistent with the assumption of proton pumping (Ries 2011a). Indeed, recent studies using ph- sensitive probes to read intracellular ph in corals (Ries 2011a) and their symbionts (Venn et al. 2009) will pave the way for future understanding of these mechanisms. It should be noted that although carbonate ions are used in the calcification reaction, crustaceans, molluscs, corals and echinoderms are known to use bicarbonate from the external seawater as their carbonate source as well as metabolically produced CO 2, which is actively converted to bicarbonate intracellularly (Wilbur & Saleuddin 1983, Cameron & Wood 1985, Dubois & Chen 1989, Herfort et al. 2008, Jury & Whitehead 2009, Holcomb et al. 2010). In zooxanthellate corals, it has been shown that the zooxanthellae can use the CO 2 produced by calcification for photosynthesis (Taylor 1983), which may also neutralize the protons produced by calcification (Cohen & McConnaughey 2003, Brownlee 2009), where the preferred carbon species is bicarbonate from external seawater (Goreau 1977, Gattuso et al. 1999). A complex organization of transport mechanisms is involved in these processes, with spatial separation of inorganic carbon and calcium transporters in different cell types (Tambutté et al. 1996, Brownlee 2009). Saturation states The two crystal forms of CaCO 3 commonly used by calcifiers are calcite and aragonite, which differ in their mineralogy and thus their physical and chemical properties. Aragonite is more soluble than calcite at the same temperature, salinity and pressure. However, high magnesium calcite (magnesium ions randomly substituted for calcium ions at mole fractions > 12%) can have a higher solubility than aragonite. An additional form used by marine calcifiers is low- magnesium calcite, which has mole fractions of magnesium of less than 4% and solubility lower than that of calcite. As Ca 2+ concentrations are typically times that of CO 3 2 and do not vary considerably, the saturation state Ω of CaCO 3 is largely determined by variations in CO 3 2. Saturation state is highly dependent on pressure, with increased solubility with pressure having a significant impact on the distribution of CaCO 3 in marine sediments. The effect of temperature on calcium carbonate solubility is small. Low temperatures cause decreased saturation states; thus, solubility is at its lowest in deep, cool, waters regions that can be devoid of calcifying organisms (Guinotte et al. 2006). At the boundary between supersaturation (Ω > 1) and undersaturation (Ω < 1) lies the saturation horizon (Ω = 1), where a natural horizontal boundary is formed as a result of temperature, pressure, and depth; CaCO 3 is neither dissolving nor forming. When dead calcified organisms sink below this saturation horizon, skeletal dissolution returns the carbon back to the water column. Since the solubility of aragonite and calcite differ, so do the depths of the corresponding aragonite and 133

8 Laura C. Wicks & J. Murray Roberts calcite saturation horizons. Surface seawater is approximately six and four times supersaturated with respect to calcite and aragonite, respectively. Over much of the Atlantic, the calcite saturation horizon is between about 4500 and 5000 m, in contrast to the aragonite saturation horizon (ASH), which is around 3000 m. On a geographical scale, these horizons also vary, which stems from interactions between ocean circulation and biological activity. The saturation state of the Pacific is lower than that of the Atlantic as Pacific deep water is older than Atlantic deep water; this means CO 2 was ventilated with the atmosphere a lot earlier, and more CO 2 has been taken up from remineralization of organic matter, lowering its carbonate ion content (Broecker et al. 2003). Historic carbon state and ecological consequences Although this review is concerned with future carbonate chemistry and its implications for benthic ecosystems, it is also helpful to look at information from the past, as seawater carbonate chemistry has altered significantly over geological time, with significant ramifications for calcifying organisms. Much of this evidence has been extrapolated from ice- core climate records and the geological fossil record, allowing highly debated estimates of the atmospheric and oceanic composition and the ecosystems present over the past 542 million years (my). Benthic invertebrates have existed in some form since the Proterozoic, becoming prolific and diverse in the Cambrian era, about 530 million years ago (mya), and have been subject to non- periodic variations in their environment over geological time (Ridgwell & Schmidt 2010). From the initial occurrence of marine calcifiers, about 550 mya, atmospheric carbon dioxide is thought to have fluctuated between about 200 and 8000 parts per million (ppm) on million- year and longer timescales (Tripati et al. 2009) due to the venting of volcanic CO 2, the weathering of silicate rocks on land, the deposition of carbonate sediments in the ocean, the weathering and oxidation of fossil organic matter, and the formation and burial of organic matter. The highest CO 2 levels occurred in the Early to Middle Paleozoic (~543 to 400 mya), Late Triassic/ Early Jurassic (~200 mya) and Cretaceous (~125 mya). From the Cretaceous up to the industrial revolution, CO 2 levels fell to between 200 and 280 ppm; therefore, the current level of 380 ppm has not been reached in the past 15 my (Raven et al. 2005, Tripati et al. 2009). Over the past 542 my, Earth s climate has also changed, and although a much- debated issue, episodes of cooler climate (but perhaps non- glacial) tend to have had atmospheric CO 2 less than 1000 ppm, whereas warmer periods generally exhibited atmospheric CO 2 greater than 1000 ppm (Royer et al. 2004, Royer 2006). Based on the isotopic record of boron assimilated in foraminiferal tests, past oceanic ph has also fluctuated over geological time and has been estimated to vary between 8.0 and 8.3 since the Cretaceous (Pearson & Palmer 2000, Tyrrell & Zeebe 2004). Records of atmospheric CO 2 levels over the past 300 my and geochemical modelling revealed there is no evidence that ocean ph was more than 0.6 units lower than today (Caldeira & Wickett 2003). In that same period, carbonate ions have nearly quadrupled in concentration, whilst calcium concentration has halved (Tyrrell & Zeebe 2004), with oceanic surface waters likely to have been supersaturated or saturated with respect to calcite and aragonite since early Precambrian time (Holland 1984). Surprisingly, the carbonate saturation horizons have not varied by more than 1500 m over the past 100 my (see Tyrrell & Zeebe 2004). In terms of more recent changes, surface water ph is estimated to have decreased by units in the tropical Atlantic and Pacific Oceans in the past 12,000 years (Sanyal et al. 2000). The more recent history of the calcium content of seawater shows that it has decreased during the last 20 my, but that the carbonate content increased in similar proportions; the carbonate saturation was therefore roughly the same during the last 5- to 20-my period (Broecker & Peng 1974). Using the mineralogy of inorganic (non- skeletal) carbonate cements and ooids (spherical CaCO 3 grains) in the geological record, changes in the dominance of calcitic and aragonitic calcifiers have 134

9 Benthic invertebrates in a high- co 2 world been used to illustrate how seawater chemistry varied over time (Sandberg 1983). During intervals of low seawater Mg/Ca (<2, e.g., Cambrian to Devonian), the predominant form of abiotically produced calcium carbonate (CaCO 3 ) derived in shallow seas was low- Mg calcite (i.e., calcite seas ). During intervals of high seawater Mg/Ca (>2, e.g., Permian to Triassic) through Modern time, aragonite and high- Mg calcite were the predominant polymorphs (i.e., aragonite seas, Stanley & Hardie 1998). It has been suggested that perturbations of the carbonate chemistry of the oceans in the past have led to five mass extinction events (Veron 2008). Specifically, the end- Ordovician mass extinction (434 mya) and end- Triassic mass extinction (205 mya) were linked to globally high temperatures and possible extreme levels of CO 2. Conversely, the Late Devonian mass extinction (364 mya) was linked to a precipitous drop in atmospheric CO 2 owing to uptake by vascular plants, low global temperatures and wildly fluctuating sea levels (Copper 2001). The cause of the end- Permian mass extinction (251 mya) is unknown, but the oceans are widely believed to have turned anoxic and to have contained free hydrogen sulphide (Wood 1999). Alternatively, the extinctions could have been caused by a sudden release of methane or volcanic CO 2 (Ryskin 2003, Knoll et al. 2007). It is suggested that 65 mya an extinction of nearly all organisms with calcium carbonate shells occurred as part of the end- cretaceous mass extinction (known as the K-T boundary between the Cretaceous and Tertiary periods), linked to the reaction of sulphate with water and oxygen to form corrosive sulphuric acid in the surface waters. Coral reefs disappeared from the fossil record for at least 2 million years (Stanley 2003), until the surface oceans had mixed sufficiently to again form a suitable environment. Although the absence of fossil records is strong evidence for such an extinction, some species may have survived such events without maintaining a skeleton (Fine & Tchernov 2007). Finally, the Paleocene- Eocene thermal maximum (PETM) 55 mya, which was characterized by rapid ( years) increase in temperature (5 9 C), drop in marine carbonate δ 13 C (3 4 ), and shoaling of the carbonate saturation horizon (>2000 m), probably caused a mass extinction of benthic Foraminifera, as well as a shift in marine plankton communities (Zachos et al. 2005). The PETM was followed by a more gradual relaxation over several hundred thousand years and probably represents the closest geological analogue to the current atmospheric pco 2 conditions (Zachos et al. 2003). ph proxies To fully appreciate past ph changes in the oceans and help us understand what will occur in the future, research has focused on the use of isotopes in biogenic carbonates (Royer 2001, Yu & Elderfield 2007, Pelejero et al. 2010). For example, boron isotopes (δ 11 B) from coral skeletons and Foraminifera tests are beginning to provide valuable information on ph fluctuations over multiple timescales from seasons to decades and even across millions of years. When calcifying organisms incorporate boron into their structures, they incorporate it as borate, which has a boron isotopic composition that is dependent on seawater ph. The relationship between boron elemental abundance and seawater ph is confounded by an array of other factors, such as temperature, salinity, biological controls and kinetic factors (Hemming & Hanson 1992, Sinclair et al. 1998, Fallon et al. 1999, Montagna et al. 2007, 2009). Thus, the use of δ 11 B relies on obtaining accurate calibrations between δ 11 B and ph, something that has proved problematic in the past (Hönisch et al. 2004, Pagani et al. 2005, Pelejero et al. 2005). However, with technique development in recent years, ph estimates are becoming more accurate, reaffirming the use of δ 11 B as a powerful tool (Ni et al. 2007, Foster 2008, Trotter et al. 2011). δ 11 B has successfully been used in Porites corals to provide evidence of intra- and interannual ph cycles at Flinders Reef in the Coral Sea, Australia (Pelejero et al. 2005), as well as an apparent trend of decreasing δ 11 B (and thus seawater ph) from the 1950s that is coincident with the pronounced downward trend in the δ 13 C record (Wei et al. 2009). An abrupt fall in ph during the 135

10 Laura C. Wicks & J. Murray Roberts twentieth century has also been demonstrated using δ 11 B in Porites corals from the subequatorial Pacific (Douville et al. 2010). On longer timescales, δ 11 B has been used to examine ph variability at the sea surface and in deeper waters over glacial and interglacial periods (Hönisch & Hemming 2005, Foster 2008), finding the ph of water on the ocean surface was lower during interglacials (high levels of atmospheric CO 2 ) and higher during glacial periods (low levels of atmospheric CO 2 ). Understanding past ocean ph using δ 11 B relies heavily on the quantification of the vital effects of the study organism during calcification, biological processes involved in the biomineralization of new skeleton, which should be a key focus of research for the future. With OA experiments being conducted globally, accurate estimates across a wide range of organisms should be possible once we have a better understanding of the underlying biological mechanisms involved in calcification. Skeletal records The decline in ph of 0.1 units in the past years is manifested in changes in the calcification rate of corals, which have been assessed using skeletal records from cores taken through massive corals. There have been mixed results to date, stemming from the difficulty in detecting an acidification signal within a naturally highly variable record and thus the need for multiple cores across multiple locations (Lough 2004). De ath et al. (2009) found an increase in extension and calcification of corals from 1900 to 1970 from the skeletal records of 328 Porites colonies on the Great Barrier Reef (GBR), followed by a 14.2% decrease in calcification since the mid-1990s. Comparable declines have been found in tropical corals from Thailand (23.5% decline in calcification from 1984 to 2003; Tanzil et al. 2009), the Caribbean (7 11% decline from 1971 to 2002; Bak et al. 2009) and the Red Sea (30% reduction in extension, 18% in calcification from 1998 to 2008; Cantin et al. 2010). A similar study on a Mesoamerican reef found declines in calcification from 1975 to 2007, although data were confounded by the effects of increases in human population in the local area over the same period (Carilli et al. 2010). Conversely, a longer- term study of coral proxy records in the Pacific ( ) found that calcification has actually increased over recent decades, although the study used only one coral core (Bessat & Buigues 2001). A multispecies study was recently conducted in Panama and found declines to have occurred in some, but not all, species of corals surveyed. Specifically, linear extension of the reef- builder Pocillopora damicornis declined by about 33% from 1974 to 2006, whereas extension of Pavona spp. increased by 2 5% in the same period (Manzello 2010). Interestingly, corals in Florida were found to have increased their extension rates, but with a concurrent decline in skeletal density, meaning that rates of calcification were stable between 1937 and 1996 (Helmle et al. 2011). Thus, in some regions corals have been able to maintain rates of extension and calcification over the past century in spite of ph declines, thermal increase and, in many cases, a combination of local environmental factors. Ongoing retrospective monitoring of coral growth is necessary to better identify changes in growth rates across the oceans; once analytical difficulties are overcome and synergistic responses accounted for, the differing calcification sensitivities of species will be a fundamental determinant of benthic communities of the future. Ocean acidification Today, the overwhelming cause of perturbation to the ocean carbon cycle is anthropogenic atmospheric CO 2, produced by the burning of fossil fuels, deforestation, industrialization, land- use changes and cement production. In recent decades, only 45% of anthropogenic CO 2 has remained in the atmosphere; the other half has been taken up by the terrestrial biosphere (~29%) and the oceans (~36%) (Sabine et al. 2004). In 2010, atmospheric CO 2 concentration was approximately 387 ppm, with half of the increase from preindustrial level (290 ppm) occurring since the 1980s (Luthi et al. 2008). Concurrently, the ph of the oceans has dropped by 0.1 units from preindustrial 136

11 Benthic invertebrates in a high- co 2 world levels, which represents an approximate 30% increase in hydrogen ions (Feely et al. 2004, Orr et al. 2005). Although OA is a global effect, the decline in ph and carbonate ion availability is not globally homogeneous and will vary on local and regional scales. For example, organisms inhabiting the intertidal zone currently experience high ph variability over the tidal cycle, with the extent of ph change likely to be amplified by OA. The deep oceans are also being affected by OA, with the deepest penetrations of anthropogenic carbon observed in areas of deep and intermediate water formation, such as the North Atlantic and Southern Ocean. In some areas, the anthropogenic CO 2 signal can be found in depths of up to 2500 m and is suspected to have already penetrated down to 5000 m (Feely et al. 2001, Tanhua et al. 2007). Since the preindustrial era, ASHs in the North Pacific have shoaled by about m (Feely et al. 2008) and will continue to shallow as pco 2 in the oceans increases. Anthropogenic pco 2 is projected to increase by 0.5% per year throughout the twenty- first century, a rate approximately 200 times faster than has occurred during the last eight glacial cycles (Siegenthaler et al. 2005) and 8 15 times faster than any change in the past 60 my (Zeebe et al. 2009). Model results from various sources indicate that atmospheric pco 2 will reach 780 ppm towards the end of the century under a business- as- usual scenario (Intergovernmental Panel on Climate Change [IPCC] IS92a), with some predictions as high as 1000 ppm (Figure 1). By 2300, values of about 2000 ppm are predicted (Caldeira & Wickett 2003). This pco 2 increase equates to a projected drop in ocean ph of another ph units by the end of the century, a magnitude and rate of change that has not occurred for more than 20 my of Earth s history (Feely et al. 2004). Increases in sea- surface temperature, an additional stressor for marine organisms, will accompany these ph changes, with a 0.74 C ± 0.18 C increase in global temperature over the past 100 years (Solomon et al. 2007) predicted to be followed by an additional warming of 2 C to 6.4 C by 2099 (Solomon et al. 2007, Sokolov et al. 2009). Projections of future pco 2, ph and saturation states vary on a latitudinal and oceanic scale. By 2100, the ASH in the Southern Ocean is predicted to shoal from the present average depth of 730 m to the surface. Elsewhere, saturation horizons are predicted to shallow from 2600 to 115 m in the North Atlantic and from 140 m to the surface in parts of the North Pacific (Orr 2005). The situation is exacerbated towards the poles; undersaturation of aragonite and calcite in surface waters will occur when atmospheric pco 2 reaches 560 and 900 ppm in the southern polar and subpolar regions, respectively. The largest changes in ocean ph will occur in the Arctic Ocean, with a 0.24-unit drop when pco 2 reaches 560 ppm. Complete undersaturation of the Arctic Ocean water column is predicted before the end of this century (Steinacher et al. 2009). In the subpolar and polar North Pacific, aragonite and calcite undersaturation will occur if (or when) pco 2 reaches 740 and 1040 ppm, respectively. With pco 2 levels at 800 ppm, the ph decrease will be about 0.4 units, dissolved carbonate ion concentrations will have decreased by more than 60%, and it is likely all coral reefs will be in erosional states (Hoegh- Guldberg 2009, Veron et al. 2009). It should be noted that other benthic calcifiers can survive and calcify in undersaturated waters, exemplified by the presence of mussel beds in the undersaturated (Ω arag < 1) waters of Kiel Fjord (Thomsen et al. 2010). In fact, in many of the shallow- water coastal environments where benthic calcifying organisms reside, seawater pco 2 is already at levels significantly higher than expected from equilibrium with atmospheric levels. In many cases, this locally enhanced pco 2 is due to net heterotrophic conditions, whereby the rate at which organic material is remineralized exceeds production. Ocean acidification is irreversible on short time frames and is affecting ecosystems on a global scale. Despite large knowledge gaps in the physiological, ecological and ecosystem impacts of OA, research over the past few decades has increased our basic understanding of short- term impacts on different species. Continuing scientific experimentation is creating a growing understanding of the wider ecosystem and longer- term implications. 137

12 Laura C. Wicks & J. Murray Roberts Methodology and approaches A better understanding of the impact of OA on benthic organisms is hampered by huge variability in experimental design, causing a lack of comparability among experiments. Disparities between experimental designs have recently been tackled by the publication of a best practices guide by the European Project on Ocean Acidification (EPOCA), providing guidelines and standards for carbonate chemistry manipulation, experimental design and interpretation (Riebesell et al. 2010). Launched in 2008, EPOCA was the first large- scale international research initiative on OA, with an overall goal of furthering our understanding of the biological, ecological, biogeochemical and societal implications of OA. In this review, we do not detail the strengths, weaknesses and complexities of all approaches (for this, see the EPOCA guide), but rather, we summarize the key points, setting out the experimental criteria used for studies included in this review. Firstly, it is critical to separate studies assessing the effect of extremes in CO 2 concentrations, linked with carbon sequestration, from those using realistic CO 2 values predicted for this century. Although extreme ph/pco 2 values are useful to constrain boundaries of an organism s performance, the shock effect on the organism from the sudden increase in pco 2 /decrease in ph will confound potential effects of the intended pco 2 treatment. As the plasticity or acclimatory capacity of an organism to an environmental change depends in part on the speed at which a change occurs, it is important to include a reasonable acclimation period, which is unlikely to be feasible at high CO 2 concentrations. In the context of this review, we limit ourselves to studies conducted with pco 2 levels up to and including 1000 ppm (i.e., up to a 0.5-unit reduction in ph; Caldeira & Wickett 2003, IPCC 2007) or where a study has used conditions predicted for the study region up to the end of the century (i.e., coastal upwelling zones with substantially higher pco 2 values currently and predicted for 2100). It should be noted that different authors use different units to express partial CO 2 pressure (e.g., ppmv, ppm and μatm. These units cannot be converted into a common unit without supportive environmental data; thus, we report the original units used. The exploratory nature and novelty of OA research brings with it many logistical challenges. Manipulation of seawater chemistry to mimic future conditions can be achieved in various ways (i.e., acid addition or mixed CO 2 and air bubbling), which, if not carefully planned and monitored, can give misleading responses in the organism being exposed (see Gattuso & Lavigne 2009 and Schulz et al for reviews). To replicate future OA conditions, the majority of experiments bubbled CO 2 -enriched air (77 of 92 studies meeting review criteria) due to its efficiency and close mimic of future carbon chemistry. A minority of studies combined the addition of acid to reduce ph with bicarbonate/carbonate addition, which accounts for the inherent change in TA from the addition of acid. As both OA manipulation approaches change the carbonate chemistry in the same way, it is suggested there will be no fundamental differences in biological responses (Schulz et al. 2009); therefore, both approaches are included in this review. Many early studies used acid addition alone (e.g., Kuffner et al. 2007), which may have led to some overestimation of impacts due to the absence of compensation for changes in TA; no studies included in this review used such methods. Many OA studies suffer from a lack of replication or a pseudoreplicate effect in the number of individuals of a species or community or the number of experimental systems or mesocosms. The natural plasticity within species and populations is a further challenge for OA researchers, requiring adequate replication of organisms and experimental systems to produce biologically meaningful differences. Although in this review we use all known benthic invertebrate studies that matched the pco 2 /ph requirements, we highlight studies that suffered from a lack of replication or used an experimental design that may not capture a true response and thus require cautious interpretation. Exposure time will be a key parameter governing the calcification response of organisms to OA, as long- lived species may have physiological mechanisms that allow them to cope with 138

13 Benthic invertebrates in a high- co 2 world relatively short perturbations (Widdicombe et al. 2010); however, OA conditions will not be an acute stressor. Problematically, the speed of change in carbonate conditions due to OA obviously cannot be replicated artificially. To date, the longest OA exposure using realistic pco 2 levels for benthic invertebrates is 1 year for the temperate coral Cladocora caespitosa (Rodolfo- Metalpa et al. 2010). However, the majority of exposures last less than a month (60 of 92 studies), which may overestimate the impact of OA due to the lack of acclimation of the organism. Although meta- analyses have been conducted on the responses of organisms to OA (e.g., Hendriks et al. 2010, Kroeker et al. 2010), we suggest that quantitative interpretation of these responses must be treated with some caution due to the variability in experimental duration. For example, a 1- to 2-hour incubation of the coral Acropora eurystoma at a ph drop of 0.4 units led to a 4-fold increase in calcification (Schneider & Erez 2006); however, exposing Acropora intermedia to a ph drop of 0.3 units for 8 weeks caused a 30% drop in calcification (Renegar & Riegl 2005). We do not limit this review to studies of a minimum duration, but if possible clarify experimental duration. To truly assess the impact of OA, multigeneration experiments are needed to understand the potential for adaptation. Thus, we may predict that natural selection will act on organisms with short generation cycles to select for genotypes that can cope with an ocean characterized by high pco 2 values. However, since many benthic invertebrates have long generation cycles (months, years), designing and running such studies remain challenging. Another problem inherent to laboratory or mesocosm experiments is the difficulty in mimicking the organism s natural environmental conditions and the future conditions under multiple synergistic stressors. For example, many short- term OA experiments starve the target organisms to avoid potentially confounding effects of variable feeding rates on measured parameters. In addition, although many reproduction studies have shown no alteration in fertilization success under OA conditions, some of these studies used high sperm concentrations, which were not representative of a natural situation and thus would not have allowed significant changes in fertilization success to be observed had they been present (e.g., Byrne et al. 2009; see Reuter et al. 2010). Furthermore, environmental conditions in the ocean are inherently variable, variations that are largely impossible to reproduce under artificial conditions. Finally, natural analogues of OA conditions exist in the form of CO 2 vent sites and upwelling regions, which allow insights into how ecosystems and organisms exist in environments naturally high in CO 2. For example, the underwater volcanic CO 2 vents abundant in the Mediterranean are at ambient seawater temperature, lack toxic sulphur compounds, and cause local acidification of seawater by as much as 1.5 ph units below the ocean average of (Hall- Spencer et al. 2008). However, these sites are not precise analogues of global- scale OA; they are temporally and spatially variable, with ever- shifting variations in ph likely to be stressful to some organisms owing to the extra physiological burden of acclimation. Furthermore, these sites are open systems, interconnected with populations existing under non- OA conditions with which ecological interactions will occur that are not representative of a future under OA conditions. However, they can provide useful information on ecosystem- level processes such as production, competition and predation, which are nearly impossible to re- create in artificial conditions. Similarly, upwelling waters in the eastern tropical Pacific have created areas of waters that are naturally high in CO 2 and low in ph and characterized by poorly cemented coral reefs, with the framework only held in place by a thin envelope of encrusting organisms (Manzello et al. 2008). Interpreting these sites is confounded by high nutrient levels in the area, but they do provide an example of the potential future for coral reefs in a high- co 2 world. It is important that further research on these upwelling regions develops, as species abundant in these variable high- pco 2, low- ph regions have physiologically adapted to such conditions, and research to date indicated some keystone species may be able to cope with future projected ocean conditions. 139

14 Laura C. Wicks & J. Murray Roberts Organism responses The impact of OA on organisms is highly variable even among closely related species, between mineralogical form, life- cycle stage and habitat. To date, the majority of OA studies focused on one or a few individual species, using experimental systems to assess the measurable effect on one or more energy budget parameters (Table 1). In this section, we review what is known so far for each energy budget parameter, focusing on experiments conducted using realistic ph scenarios predicted up to the end of the century (up to ~ 0.5 ph unit decline from present day). We also discuss what little is known of behavioural and immunological responses to OA. With the wealth of published data over the past few years, it is of course not feasible to describe every experiment, so we summarize the overall patterns that are emerging in this field, using a hypothesis- driven approach. It should be noted that there is a clear bias in the taxonomic groups studied to date, with molluscs and scleractinian corals the focus of much research; as such, caution should be applied in making overarching conclusions for the benthos. Energy budget responses Like any stressor, the impact of OA on benthic organisms may be shown in different components of their energy budget, from energy intake and metabolism, through to calcification and reproductive output. Knowledge of organism- level responses is essential for understanding how stressors cause adverse biological effects, how a population will respond, and the strategies adopted by organisms to tolerate stress. Growth and calcification Physiological responses to environmental changes such as OA are often exhibited as altered growth rates, perhaps due to energy being reallocated towards enhanced maintenance costs (i.e., Pörtner et al. 2005b). In calcifying organisms, OA brings with it the additional challenge of reduced carbonate ions for calcification and thus reduced aragonite and calcite saturation. We hypothesize that growth and calcification will be reduced in all benthic invertebrates, with deleterious effects amplified in calcifiers using more soluble polymorphs of CaCO 3. As shown in Table 1, the majority of calcification responses to OA are negative (37 of 52 experimental organisms), with changes in calcification rate from a ph drop of 0.3 to 0.5 units ranging from a 99% decline to a 400% increase (e.g., Ries et al. 2009). Research into the effect of OA has focused largely on corals, molluscs and echinoderms, with the adults and juveniles exhibiting predominantly reduced growth/calcification in response to reduced ph. Many experiments on scleractinian corals, which mainly precipitate aragonite skeletons, have shown a linear decline in calcification with increases in seawater pco 2 (see Table 1). However, recent experiments have also shown that some tropical and temperate species exhibit either no response (i.e., Stylophora pistillata, Reynaud et al. 2003) or a non- linear response to OA. For example, the non- linear response in the temperate coral Oculina arbuscula to OA indicated a tipping point in calcification between 0.8 and 1.6 Ω arag, below which calcification declined, above which there was no alteration in calcification rate from the control (60 days, Ries et al. 2010). A non- response may be an indication of the corals ability to control the carbonate chemistry of their calcifying fluid equally well under all treatment levels (see section on calcification). Alterations in skeletal morphology have also been observed in corals in response to enhanced pco 2 and associated declines in aragonite saturation (Cohen & Holcomb 2009, Holcomb 2010). Reductions in aragonite crystal growth rates with decreasing saturation states resulted in shorter, fatter crystals, which are disorganized and misaligned (Figure 5). Similarly, species- dependent differences in the crystallization patterns of fibres were observed in four tropical coral species subject to OA conditions ( 0.3 ph units for 8 days, Marubini et al. 2003).v 140

15 Benthic invertebrates in a high- co 2 world Table 1 Responses of benthic invertebrates to ocean acidification at pco 2 /ph levels predicted up to the end of the century Process Adult growth/ calcification Cnidaria Favia fragum a,68 Oculina arbuscula a,73,74 Porites astreoides a,68 Echinodermata Eucidaris tribuloides a,73 Foraminifera Marginopora kudakajimensis a,45 Mollusca Mytilus edulis a,10 Strombus alatus a,73 Polychaeta Hydroides crucigera a,73 Bryozoa Celleporella hyalina 67 Cnidaria Acropora cervicornis 70 Acropora eurystoma a,76 Acropora intermedia a,5 Acropora verweyi 54 Astrangia poculata 39 Coral community 46,47 Fungia spp. 40 Galaxea fascicularis 54 Lophelia pertusa Montipora capitata 33 Pavona cactus 54 Porites compressa 53 31, a,50 Cnidaria Cladocora caespitosa 75 Lophelia pertusa 31 Madracis auretenra 42 Stylophora pistillata 72 Echinodermata Amphiura filiformis 88 Ophiocten sericeum 90 Mollusca Laternula elliptica 20 Mytilus edulis 73 Venerupis decussata 7 Echinodermata Pisaster ochraceus 36 Cnidaria Acropora sp. a,33 Echinodermata Arbacia punctulata a,73 Foraminifera Baculogypsina sphaerulata a,32 Calcarina gaudichaudii a,32 Mollusca Crepidula fornicata a,73 Porites lobata a,5 Arthropoda Homarus americanus a,73 continued 141

16 Laura C. Wicks & J. Murray Roberts Table 1 (continued) Responses of benthic invertebrates to ocean acidification at pco 2 /ph levels predicted up to the end of the century Process Porites lutea 40,63,a Stylophora pistillata a,33,55 Turbinaria reniformis 54 Echinodermata Ophiura ophiura 90 Strongylocentrotus franciscanus 60 Foraminifera Amphisorus hemprichii a,32 Mollusca Argopecten irradians a,73 Crassostrea gigas a,26 Crassostrea virginica a,73 Littorina littorea a,73 Mercenaria spp. 84 Mya arenaria a,73 Mytilus edulis a,34,83 Nucella lamellosa a,61 Pinctada fucata a,88 Urosalpinx cinerea a,73 Venerupis decussata 7 142

17 Benthic invertebrates in a high- co 2 world Photosynthesis/ productivity Cnidaria Acropora intermedia a,5 Porites lobata a,5 Cnidaria Acropora formosa a,18 Stylophora pistillata 55,72 Feeding Echinodermata Asterias rubens 25 Mollusca Mytilus edulis larvae 9 Respiration Cnidaria Lophelia pertusa 31 Porites astreoides larvae a,3 Echinodermata Ophionereis schayeri 16 Cnidaria Acropora eurystoma a,76 Anthopleura aureoradiata 22 Cladocora caespitosa 75 Coral community 49 Stylophora pistillata 22 Echinodermata Pisaster ochraceus 36 Strongylocentrotus franciscanus 60 Strongylocentrotus purpuratus larvae 81 Mollusca Patella vulgata 51 Venerupis decussata 7 Arthropoda Hyas araneus 85 Cnidaria Acropora digitifera larvae 59 Acropora eurystoma a,76 Cladocora caespitosa 75 Stylophora pistillata 72 Echinodermata Ophiocten sericeum 91 Ophiura ophiura 90 Mollusca Crassostrea gigas 48 Patella vulgata 51 Cnidaria Coral community 46,47 Cnidaria Anthopleura aureoradiata 22 Stylophora pistillata 22 Echinodermata Amphiura filiformis 89 Strongylocentrotus purpuratus larvae 81 Mollusca Laternula elliptica 20 Mytilus edulis 83 continued 143

18 Laura C. Wicks & J. Murray Roberts Table 1 (continued) Responses of benthic invertebrates to ocean acidification at pco 2 /ph levels predicted up to the end of the century Process Reproduction Bryozoa Cnidaria Celleporella hyalina 67 Montipora capitata 41 Echinodermata Amphiura filiformis 89 Fertilization Cnidaria Cnidaria Acropora digitifera a,57 Acropora palmata a,2 Echinodermata Heliocidaris erythrogramma a,38 Holothuria spp. a,57 Paracentrotus lividus a,58 Strongylocentrotus franciscanus a,71 Mollusca Crassostrea gigas a,65 Placopecten magellanicus 21 Saccostrea glomerata a,64,65 Cnidaria Montipora capitata 41 Echinodermata Arbacia punctulata a,15 Centrostephanus rodgersii a,12 Echinometra mathaei a,44 Heliocidaris erythrogramma a,11 13 Heliocidaris tuberculata a,12 Hemicentrotus pulcherrimus a,44 Paracentrotus lividus a,52 Patiriella regularis a,12 Sterechinus neumayeri a,27 Strongylocentrotus purpuratus a,8 Tripneustes gratilla a,12 144

19 Benthic invertebrates in a high- co 2 world Embryonic development/ hatching Mollusca Arthropoda Mytilus edulis a,35 Semibalanus balanoides 28 Mollusca Haliotis coccoradiata 14 Littorina obtusata 26 Acropora tenuis 43 Mollusca Crassostrea gigas 37 Haliotis coccoradiata a,12 Mytilus edulis 9 Mytilus galloprovincialis 9 Nemertea Parborlasia corrugatus a,27 Cnidaria Echinodermata Arbacia punctulata a,15 Hemicentrotus pulcherrimus a,44 Paracentrotus lividus a,58 Sterechinus neumayeri a,27 Nemertea Parborlasia corrugatus a,27 continued 145

20 Laura C. Wicks & J. Murray Roberts Table 1 (continued) Responses of benthic invertebrates to ocean acidification at pco 2 /ph levels predicted up to the end of the century Process Larval calcification/ growth Echinodermata Paracentrotus lividus a,52 Arthropoda Homarus gammarus 6 Hyas araneus 85 Cnidaria Acropora digitifera 59 Porites astreoides a,1 Echinodermata Echinometra mathaei 86 Evechinus chloroticus a,17 Hemicentrotus pulcherrimus a,44 Lytechinus pictus a,62 Ophiothrix fragilis a,23 Sterechinus neumayeri a,17 Strongylocentrotus franciscanus 79 Strongylocentrotus purpuratus 81,a,92 Tripneustes gratilla a17,77 Arthropoda Homarus gammarus 6 Hyas araneus 85 Cnidaria Acropora tenuis 43 Echinodermata Paracentrotus lividus a,52,58 Pseudechinus huttoni a,17 Mollusca Crassostrea ariakensis a,56 Echinodermata Crossaster papposus 24 Chordata Ascidiella aspersa 25 Ciona intestinalis 25 Oikopleura dioica

21 Benthic invertebrates in a high- co 2 world Larval morphology Mollusca Argopecten irradians 82 Crassostrea gigas a,65 Crassostrea virginica 56a,83 Haliotis coccoradiata 14 Haliotis kamtschatkana 19 Mercenaria mercenaria a,82 Mytilus edulis a,9,35 Mytilus trossulus 79 Saccostrea glomerata a,64 66,87 Echinodermata Echinodermata Tripneustes gratilla a,77 Lytechinus pictus a,62 Ophiothrix fragilis a,23 Strongylocentrotus purpuratus 81 Mollusca Crassostrea gigas a,11,65 Haliotis kamtschatkana 19 Mytilus galloprovincialis 9 Saccostrea glomerata a,64,65,87 Echinodermata Evechinus chloroticus a,17 Heliocidaris erythrogramma a,11 Paracentrotus lividus a,52,58 Pseudechinus huttoni a,17 Sterechinus neumayeri a,17 Strongylocentrotus purpuratus a,81,92 Tripneustes gratilla a,17 Mollusca Mytilus edulis 9 continued 147

22 Laura C. Wicks & J. Murray Roberts Table 1 (continued) Responses of benthic invertebrates to ocean acidification at pco 2 /ph levels predicted up to the end of the century Process Larval survival Arthropoda Hyas araneus 85 Cnidaria Acropora tenuis 80 Echinodermata Ophiothrix fragilis a,23 Mollusca Argopecten irradians 82 Crassostrea gigas a,65 Haliotis kamtschatkana 19 Mercenaria mercenaria a,82 Saccostrea glomerata a,64,65,87 Arthropoda Homarus gammarus 6 Hyas araneus 84 Cnidaria Acropora digitifera 59,80 Porites panamensis 4 Echinodermata Evechinus chloroticus a,17 Paracentrotus lividus a,52 Pseudechinus huttoni a,17 Sterechinus neumayeri a,17 Tripneustes gratilla a,17 Mollusca Crassostrea virginica 82 Mytilus edulis 35 Chordata Ascidiella aspersa 25 Ciona intestinalis 25 Oikopleura dioica 25 Echinodermata Strongylocentrotus droebachiensis 25 Strongylocentrotus purpuratus 81 Settlement Cnidaria Cnidaria Acropora palmata a,2 Acropora tenuis 43 Pocillopora damicornis41 Porites astreoides a,3 Porites panamensis 4 Echinodermata Crossaster papposus 24 Mollusca Haliotis kamtschatkana 19 Mytilus edulis 9 148

23 Benthic invertebrates in a high- co 2 world Juvenile calcification/ growth Arthropoda Semibalanus balanoides 29,30 Cnidaria Acropora digitifera 80 Acropora palmata a,2 Acropora tenuis 43,80 Porites astreoides a,3 Porites panamensis 4 Echinodermata Echinometra mathaei 78 Hemicentrotus pulcherrimus 78 Mollusca Haliotis coccoradiata 14 Strombus luhuanus 78 Arthropoda Elminius modestus 29 Semibalanus balanoides 29 Mollusca Ruditapes decussatus 69 Echinodermata Crossaster papposus 24 Note: 1 Albright et al. (2008); 2 Albright et al. (2010); 3 Albright & Langdon (2011); 4 Anlauf et al. (2011); 5 Anthony et al. (2008); 6 Arnold et al. (2009); 7 Bamber (1987); 8 Bay et al. (1993); 9 Bechmann et al. (2011); 10 Berge et al. (2006); 11 Byrne et al. (2009); 12 Byrne et al. (2010c); 13 Byrne et al. (2010b); 14 Byrne et al. (2010a); 15 Carr et al. (2006); 16 Christensen et al. (2011); 17 Clark et al. (2009); 18 Crawley et al. (2010); 19 Crim et al. (2011); 20 Cummings et al. (2011); 21 Desrosiers et al. (1996); 22 Doherty (2010); 23 Dupont et al. (2008); 24 Dupont et al. (2011); 25 Dupont & Thorndyke (2008); 26 Ellis et al. (2009); 27 Ericson et al. (2010); 28 Findlay et al. (2009); 29 Findlay et al. (2010a); 30 Findlay et al. (2010c); 31 Form & Riebesell (2010); 32 Fujita et al. (2011); 33 Gattuso et al. (1998); 34 Gazeau et al. (2007); 35 Gazeau et al. (2010); 36 Gooding et al. (2009); 37 Havenhand & Schlegel (2009); 38 Havenhand et al. (2008); 39 Holcomb et al. (2010); 40 Hossain & Ohde (2006); 41 Jokiel et al. (2008); 42 Jury & Whitehead (2009); 43 Kurihara (2008); 44 Kurihara & Shirayama (2004); 45 Kuroyanagi et al. (2009); 46 Langdon & Atkinson (2005); 47 Langdon et al. (2003); 48 Lannig et al. (2010); 49 Leclercq et al. (2002); 50 Maier et al. (2009); 51 Marchant et al. (2010); 52 Martin et al. (2011); 53 Marubini et al. (2001); 54 Marubini et al. (2003); 55 Marubini et al. (2008); 56 Miller et al. (2009); 57 Morita et al. (2010); 58 Moulin et al. (2010); 59 Nakamura et al. (2011); 60 Nienhuis (2009); 61 Nienhuis et al. (2010); 62 O Donnell et al. (2010); 63 Ohde & Hossain (2004); 64 Parker et al. (2009); 65 Parker et al. (2010); 66 Parker et al. (2011); 67 Pistevos et al. (2011); 68 Putron et al. (2010); 69 Range et al. (2010); 70 Renegar & Riegl (2005); 71 Reuter et al. (2010); 72 Reynaud et al. (2003); 73 Ries et al. (2009); 74 Ries et al. (2010); 75 Rodolfo- Metalpa et al. (2010); 76 Schneider & Erez (2006); 77 Sheppard- Brennand et al. (2010); 78 Shirayama & Thornton (2005); 79 Sunday et al. (2011); 80 Suwa et al. (2010); 81 Stumpp et al. (2011b); 82 Talmage & Gobler (2009); 83 Thomsen et al. (2010); 84 Waldbusser et al. (2010); 85 Walther et al. (2009); 86 Walther et al. (2011); 87 Watson et al. (2009); 88 Welladsen et al. (2011); 89 Wood et al. (2008); 90 Wood et al.(2010); 91 Wood et al. ( 2011); 92 Yu et al. (2011). a More than one experimental pco 2 /ph tested. 149

24 Laura C. Wicks & J. Murray Roberts A E B F C G D H Figure 5 Progressive changes in the mesoscale skeletal development (A D), including distortion of basal plate and retardation of septal development, of 8-day- old corallites of Favia fragum with decreasing seawater saturation state. Progressive changes in the morphology and orientation of crystals within the corallites are documented by scanning electron microscopy imaging of broken faces of primary septa (E H). (continued) 150

25 Benthic invertebrates in a high- co 2 world Calcification in benthic Foraminifera reduces as ph declines in the four species examined to date (Kuroyanagi et al. 2009, Fujita et al. 2011), but for three of the four species a tipping point was apparent, below which calcification was reduced but above which there was no response or even enhanced calcification. For example, no change in shell diameter or weight was observed in Marginopora kudakajimensis when exposed to a ph drop of 0.2 units for 10 weeks, but when ph was reduced by 0.4 units, both parameters were significantly lower (Kuroyanagi et al. 2009). At intermediate pco 2 /ph levels, enhanced calcification was observed in two of the species examined (Baculogypsina sp. and Calcarina sp.), suggested to be linked to increased bicarbonate ion concentration in the culture medium, which may enhance diffusion of bicarbonate, thereby promoting calcification (Kuroyanagi et al. 2009). Conversely, the lack of enhanced calcification at intermediate pco 2 levels in the imperforate species Amphisorus spp. and M. kudakajimensis may be due to differences in calcification mechanisms. These species take up carbonate ions for calcification directly from seawater and thus will not be able to compensate for reductions in carbonate ion concentration. As the species examined all harboured algal symbionts, any increased energy demand for calcification due to OA may be met by the photosynthetic activity of the symbionts, which may be enhanced with increased bicarbonate ion concentration. Further studies are needed to measure photosynthetic rates of algal symbionts at various pco 2 levels to assess the effectiveness of this mechanism. The Echinodermata have also shown parabolic (positive under intermediate pco 2, negative under high pco 2 ) calcification responses, for instance in the urchin Arbacia punctulata (60 days, Ries et al. 2009). But, this response is not universal. Two other urchin species examined showed a negative linear calcification response (4 months, Nienhuis 2009; 60 days, Ries et al. 2009), and echinoderms such as ophiuroids (brittlestars) and asteroids (seastars) have not shown any negative response to OA. In fact, in three separate studies the brittlestar Amphiura filiformis, seastar Pisaster ochraceus and juveniles of the seastar Crossaster papposus increased their growth in response to OA conditions (40 days, Wood et al. 2009; 10 weeks, Gooding et al. 2009; and 38 days, Dupont et al. 2010, respectively). Crustaceans exhibited a variable calcification response to OA seemingly dependent on motility; calcification in the sessile barnacles Semibalanus balanoides and Elminius modestus declined by 79% and 20%, respectively, in response to a 0.3 ph unit drop (30 days, Findlay et al. 2009), 7% and 38% increases were observed, respectively, in the lobster Homarus americanus and crab Callinectes sapidus with a 0.31 ph unit drop (60 days, Ries et al. 2009). A negative linear decline in calcification has been observed in 11 of the 14 molluscan species examined (Table 1), with declines up to 98% in response to a 0.3 ph unit drop (Mya arenaria, 60 days, Ries et al. 2009). Malformation of new nacre tablets has been observed in oyster shells exposed to acidified conditions (ph 7.6) after 28 days at a ph 0.5 units below control (Welladsen et al. 2011). As different forms or polymorphs of CaCO 3 used by marine calcifiers have different solubilities in seawater, it has been suggested that organisms using the most soluble polymorphs would be more adversely affected by OA. However, a recent meta- analysis using 56 studies of various benthic organisms revealed calcification in organisms using aragonite and low- magnesium calcite was negatively affected by OA, whereas organisms that use high- magnesium calcite, the most soluble isomorph of CaCO 3, were not significantly affected (Kroeker et al. 2010). In calcifiers exhibiting no change in calcification rate under OA conditions, biogenic calcification processes are therefore compensating for the change in carbonate chemistry of seawater (Pörtner 2008), regardless of the polymorph of CaCO 3 used. In addition, most calcifying marine organisms cover their shells, skeletons, or tests with some type of protective organic layer, which serves to isolate their biomineral Figure 5 (continued) In A and E, saturation state Ω = 3.71 (control); in B and F, Ω = 2.40; in C and G, Ω = 1.03; in D and H, Ω = (A D) Scale bar = 200 mm. (Reproduced from Cohen et al with permission from A. Cohen and the American Geophysical Union.) 151

26 Laura C. Wicks & J. Murray Roberts from the surrounding seawater (Ries et al. 2009). Alternatively, shell/skeletal polymorph mineralogy may be a phenotypically plastic trait (Ries 2011b), whereby bimineralic organisms increase the incorporation of less- soluble polymorphs (Lee & Morse 2010, Ries 2011b). However, to date this has only been observed in one benthic species (the whelk Urosalpinx cinerea, Ries 2011b), and further study is needed to confirm this potential adaptive mechanism. Overall, the hypothesis that growth and calcification will be reduced in all benthic invertebrates can be rejected. Although negative responses were observed in many of the invertebrates studied, particularly corals and molluscs, enhanced calcification was observed in some echinoderm species in response to OA. In addition, many invertebrates exhibited no detectable growth response to OA. In terms of enhanced susceptibility to OA of invertebrates with soluble calcium carbonate polymorphs, this hypothesis can also be rejected, with no clear enhancement of negative effects for aragonitic species. Respiration and metabolism An organism may physiologically respond to stress in three ways: (1) increase metabolic rate to cover energy demand (only energetically feasible in the short term); (2) reduce metabolic rate below normal levels (metabolic depression) by downregulation of anabolic and catabolic processes to maintain homeostasis (Guppy & Withers 1999); or (3) acclimate to the new conditions by expressing plasticity and thus maintaining metabolism via alternative mechanisms. We hypothesized that benthic invertebrates will increase their metabolic rate to compensate for enhanced energy demand. However, as with growth and calcification, metabolic responses have been variable between species, with many species exhibiting no detectable change in respiration (Table 1). As such, this hypothesis of enhanced metabolism under OA conditions is rejected. To understand why metabolism is affected by OA, we must explore the cellular response to enhanced pco 2 and decreased ph. Intracellular hypercapnia and acidosis When pco 2 in seawater rises, diffusion across animal surfaces increases, so that pco 2 reaches equilibrium in both intra- and extracellular spaces (hypercapnia). The internal pco 2 levels of the animal then rise, and cause extracellular acidosis (internal fluid ph falls), until a new pco 2 level is reached that is sufficient to restore a diffusive gradient to allow CO 2 excretion. In addition to diffusion, the active process of ion exchange allows for complete or partial ph compensation, whereby H + or bicarbonate is exchanged in epithelial membranes between animal and water and in membranes separating intra- and extracellular compartments, leading to an accumulation of bicarbonate. Intracellular ph is units lower than the extracellular ph, which means a drop in intracellular ph is much less than in the extracellular spaces, attributable to high levels of bicarbonate buffering in the former. Mechanisms employed by organisms to mitigate acidosis and hypercapnia include metabolic production and consumption of protons, passive buffering of intra- and extracellular fluids, transport and exchange of relevant ions, and transport of CO 2 in the blood in those species that have respiratory pigments. Intracellular compensation was typically completed within hours in the animals tested to date, and some return towards control values was usually observed within 24 hours (Pörtner et al. 1998), often at the expense of extracellular ph. In some organisms, acid- base parameters and ionic concentrations may also reach new steady- state values rather than returning to preacidosis levels (Pörtner et al. 2004). Ion transport systems (i.e., H + /Na + and HCO 3 /Cl transport), located in the gills or renal systems, form the basis for efficient compensation of ph disturbances during exposure to elevated environmental pco 2, although much remains to be learned of such compensatory mechanisms in benthic invertebrates (see review by Pörtner et al. 2004). When uncompensated, acidosis can lead to metabolic suppression, compromised enzyme activity (Somero 1985), reduced protein synthesis (Kwast & Hand 1996), respiratory stress, reduced metabolic scope, loss of consciousness due to disruption of oxygen transport mechanisms, and ultimately death (Seibel & Walsh 2003). Such metabolic suppression has been directly observed in 152

27 Benthic invertebrates in a high- co 2 world the cold- water coral Lophelia pertusa, which exhibited reduced respiration rates when exposed to OA conditions for 6 months (ph drop of 0.3 units, Form & Riebesell 2011). Further, genomic tools have recently provided important new indirect evidence of metabolic suppression. In the oyster Crassostrea virginica, the transcriptome responds to OA conditions by reducing the adenosine triphosphate (ATP)-expensive protein synthesis and cell growth machinery and increasing ATP generation by elevating transcripts involved in the mitochondrial oxidative phosphorylation (Chapman et al. 2011). Metabolic suppression is, in many cases, an adaptive strategy for survival under transiently stressful conditions, such as hypoxia and anoxia. However, it can also be an indication that an organism cannot compensate for the internal hypercapnia via control of acid- base imbalances in the intra- and extracellular spaces (Guppy & Withers 1999). Metabolic suppression in the urchin Psammechinus miliaris was observed in response to a 0.5 ph unit drop over 8 days; the urchins could only regulate their coelomic fluid to compensate partly for acidosis, achieved by a series of increases in HCO 3 concentration via test dissolution (Miles et al. 2007). No compensation to maintain internal ph was observed in the seastar Asterias rubens after 1 week and 6 months at a ph 0.4 units below ambient (Hernroth et al. 2011). It was suggested that the absence of protein in the perivisceral fluid, and thus a low capacity for CO 2 storage, in addition to a poor capacity for ionic regulation and dependency on a magnesium calcite test make echinoderms particularly vulnerable to anthropogenic acidification. However, two other studies of ophiuroids at realistic ph levels for 2100 found no metabolic response (Ophiocten sericeum, 20 days, Wood et al. 2011) or an increase (Amphiura filiformis, 40 days, Wood et al. 2009; Ophiura ophiura, 40 days, Wood et al. 2010). Buffering of intra- and extracellular fluids is often manifested as a trade- off in skeletal/shell integrity, observed in the mussels Mytilus edulis (Lindinger et al. 1984) and M. galloprovincialis (Michaelidis et al. 2005); shell dissolution was found to be the main source of bicarbonate. Although the latter study used ph values outside our specified ph range (7.3), intertidal animals regularly experience hypercapnia and are thus interesting to consider. For example, long- term hypercapnia in M. galloprovincialis caused a permanent reduction in the haemolymph ph, despite the mussels increasing haemolymph bicarbonate levels with the dissolution of shell CaCO 3. Metabolic rate and growth rates also declined, possibly also related to net degradation of protein (Michaelidis et al. 2005) or a higher capacity for somatic growth versus shell accretion under hypercapnic conditions (Thomsen et al. 2010). Clearly, the extent to which metabolic processes can compensate for acidosis depends on the metabolic rate of the organism in question. Animals with enhanced metabolic rates (e.g., some corals, molluscs and crustaceans) may be less sensitive to elevated seawater pco 2 as they possess highly efficient mechanisms to maintain physiological homeostasis via increased metabolic production of CO 2 (Seibel & Walsh 2003, Pörtner et al. 2004, Pörtner 2008, Melzner et al. 2009). Alternatively, these animals may exhibit less flexibility in their metabolic responses to OA as they are already maintaining a high metabolism, which could lead to metabolic depression. In the deep sea, animals have metabolic rates typically an order of magnitude lower than their shallow- living counterparts (Childress et al. 1990), which confers a reduced ability to adapt to changes in pco 2 and ph. However, to date there have been few, if any, studies examining the metabolic response of deep- sea organisms to OA. Using a moderate ph drop (0.3 units), Seibel & Walsh (2003) showed that reduced ph causes acidosis of the blood of the deep- sea octopus Benthoctopus sp. This reduced oxygen binding by haemocyanin by 40% and thus indicated potential sensitivity to OA. The blood oxygen- binding systems of crustaceans and other benthic invertebrates may also leave them vulnerable to OA, as increased CO 2 in the bloodstream will affect the blood oxygenbinding process by reducing the oxygen affinity of haemoglobin. Known as the Bohr effect, this is dependent on the gradient in oxygen concentration from the environment to the blood and the blood to the metabolizing tissues, and thus changes in CO 2 levels in the bloodstream will alter these gradients. Reduced O 2 affinity in response to increasing pco 2 or decreasing ph has been demonstrated 153

28 Laura C. Wicks & J. Murray Roberts in invertebrate haemoglobins, as well as annelid pigments erythrocruorin and chlorocruorin, compromising O 2 loading at the gas exchange surface (Weber et al. 1978). In contrast, enhanced CO 2 increases the affinity of erythrocruorin for O 2 in the lugworm Arenicola marina, a species adapted to a hypoxic environment (Weber 1980). Productivity The vulnerability of zooxanthellate corals to OA will be determined by both the susceptibility of the coral hosts and their photosynthetic symbionts, which could possibly have differing responses. It was originally suggested that corals may flourish under OA conditions, as CO 2 enrichment has been found to increase the photosynthetic rate of microalgae (Riebesell et al. 1993) and macroalgae (Borowitzka & Larkum 1976, Gao et al. 1993). It is possible that under conditions of increased pco 2 the energy- costly carbon- concentrating mechanism in the symbionts may be deactivated, allowing them to use external CO 2 as it becomes more available relative to HCO 3, of which the latter is preferentially used for photosynthesis (Gao et al. 1993, Beardall et al. 1998). While most studies assessing OA effects on respiration and productivity of tropical and temperate corals found no change or a decline in photosynthetic rate, studies of coral communities rather than individual species do report an increase in productivity (1 2 months, Langdon et al. 2003; 7 12 days, Langdon & Atkinson 2005). A lack of altered photosynthetic rate is suggested to be due to nutrient limitation, typical of tropical coral reefs (Leclercq et al. 2002, Langdon et al. 2003), as nutrient enrichment has been shown to enhance photosynthesis (Ferrier- Pagès et al. 2001). Declines in photosynthesis have been linked to high CO 2 or lowered ph disrupting the photoprotective mechanisms of coral symbionts or algal chloroplasts by lowering rates of photorespiration and the capacity for thermal dissipation (Crawley et al. 2010). Interestingly, enhanced productivity was observed in Acropora spp. at intermediate CO 2 levels ( ppm, 4 days, Crawley et al. 2010), suggesting that the rate of photosynthesis was stimulated directly either by increased CO 2 supply or by an increase in excitation pressure driven by bleaching- induced increases in internal light fields (Enriquez et al. 2005), processes overridden at high CO 2 levels. Despite corals being the focus of many OA experiments, there remains much to be learnt as their responses have proved variable and complex. This is further complicated by the predominantly negative effect of increased temperatures and the potentially differential responses between coral and symbiont. Reproductive output Reproductive processes are among the first components of the energy budget to suffer in response to environmental stress as they are energetically expensive and can be delayed when the survival of the individual is in jeopardy (reviewed in Schneider 2004). Thus, we hypothesized that OA conditions would lead to reduced reproductive output. A reduction in energy invested into reproduction in response to OA was evident in the few organisms that have been tested; however, with only four studies to date, it is difficult to make judgments on how OA will affect reproduction, and thus our hypothesis can be neither accepted nor rejected (Table 1). The deficit of studies is linked to the longevity of many benthic organisms, which complicates, and in many cases prohibits, experimental testing. The brittlestar Amphiura filiformis is an example of such complications; egg size and structure were not affected by reduced ph ( 0.4 ph units), but the study was conducted during a latent period of egg growth, which suggests that any difference would be very hard to detect (Wood et al. 2008). Gonad development was delayed compared to controls in the urchin Hemicentrotus pulcherrimus reared for 10 months at a ph drop of 0.03 units (Shirayama & Thornton 2005). Interestingly, reproductive investment in the bryozoan Celleporella hyalina actually increased in response to enhanced pco 2 and temperature for 15 days (Pistevos et al. 2011), which would indicate a mechanism for immediate reproduction for colonies threatened with imminent death. The budding of additional male zooids is thought to be an adaptive strategy, promoting the rapid acquisition 154

29 Benthic invertebrates in a high- co 2 world of reproductive success through male function that avoids the necessity of prolonged investment through brooding in the female role. Clearly, the impact of OA on the reproductive investment in invertebrates is a key area that needs to be addressed; the impact of future OA on this critical developmental stage could have devastating consequences for species survival. Energy intake The physiological effect of reduced ph on benthic invertebrates may be a reduction in their feeding ability, either linked to morphological changes in feeding apparatus or as a trade- off between starvation and predation risk (Houston et al. 1993). Feeding is energetically costly, and foraging opportunities typically also increase predation risk; as such, trade- offs often exist between foraging opportunities and the risk of predation (Sih 1987, Houston et al. 1993). Under altered environmental conditions, which are potentially stressful for an organism, alterations in foraging behaviour may occur; individuals in poor body condition often accept higher risks to forage in higher- quality habitats (McNamara & Houston 1987). Alternatively, if an organism s protective structure (i.e., shell) is weakened by OA, it may be less likely to risk predation during foraging. We hypothesized that OA conditions would reduce feeding rates in benthic invertebrates due to altered feeding organs or reduced risk- taking behaviour. However, to date, near- future OA levels have caused no significant change in feeding rate in adults of the benthic species examined (Table 1), although only seven studies have been conducted, and the controlled conditions of such studies (absence of predators) confound potential responses. The hypothesis is tentatively rejected but highlights a key area of behavioural research for future OA studies. One study of note is that of the limpet Patella vulgata (Marchant et al. 2010), which, although exhibiting no change in feeding rate under OA conditions ( 0.6 units for 5 days), showed evidence of significant hypercapnia- related damage to the feeding organ (radula), that may in turn result in a reduction in feeding efficiency and related fitness. Further investigation of this response is needed over longer time periods to examine exact physiological mechanisms determining radula damage. As illustrated in this section of the review, in many benthic invertebrates OA is likely to cause trade- offs in how resources are allocated and increases in their energy demands to compensate for the enhanced energetic costs of growth and maintenance. The absence of enhanced feeding in response to the OA condition suggests that increased energy intake is not a compensatory mechanism employed by these organisms per se. However, interpretation of many OA studies is further confounded by the starvation of organisms in the experimental period to avoid potentially confounding effects of variable feeding rates on measured parameters (n > 6 of heterotrophic invertebrates, feeding state unknown in five studies). It can be suggested that when deleterious consequences of OA were recorded, an alteration in feeding rates may have played a role, but further examination is required. Energy budget trade- offs and consequences for fitness Many benthic invertebrates have been shown to exhibit reductions in one or more of their energy budget parameters in response to OA, including growth, calcification and reproduction. These reductions may be a factor of altered energy budget partitioning; energy is partitioned away from growth towards increased costs of metabolism, maintenance of intracellular ph and general cellular homeostasis (Deigweiher et al. 2010), as well as limitations in feeding ability. These reductions in observable energy budget parameters lead to the suggestion that the organism involved does not have enough inherent physiological flexibility in its energy budget to compensate for changing environmental conditions. However, to date no studies have assessed the full energy budget response of any adult benthic invertebrate species to OA. Potential energy budget responses can be extrapolated from studies examining a single energydriven parameter, such as calcification. Reduced calcification rates in response to OA indicate that the organism will be requiring more energy to pump protons out of the calcifying fluid to counter 155

30 Laura C. Wicks & J. Murray Roberts dissolution pressure, and maintain the protective skeleton or shell, and thus will be allocating more energy to maintenance than growth. Where enhanced growth and productivity have been observed in some animals, it can be suggested that these enhancements are at the cost of other energydependent processes. However, full energy budget responses to OA for benthic invertebrates are needed to unravel such responses. Potential trade- offs in energy allocation have been observed when more than one parameter was measured. For example, while enhanced metabolism was seen in the bivalve Laternula elliptica in response to OA, whole- organism functioning was negatively affected, indicating L. elliptica were working harder to calcify at this lower ph (Cummings et al. 2011). This upregulation of metabolism indicates the animals have the ability to internally compensate during periods when acidosis incurs physiological and energetic costs (Pörtner et al. 2005a, Arnold et al. 2009); however, mechanisms for compensation of short- term hypercapnia may not be possible during longer exposures. The field of OA research would benefit from scope- for- growth (SfG) studies; energy available for growth and reproduction is calculated for energy budget models. SfG is commonly used in pollution assays and provides a robust tool for measuring whole- animal performance under stress (Widdows & Johnson 1988, Widdows et al. 2002). When an animal is stressed, SfG decreases due to either an increased energy loss through respiration or excretion or decreased energy input. Early life stages Benthic invertebrates, like most animals, differ in their susceptibility to stress at different stages of their life cycle. These differences can relate to their lifestyle, whether they are in the pelagic or benthic zone, and how their past history affects their ability to adapt. The early life cycle is predicted to be a bottleneck; following fertilization, survival and settlement success are low, with few larvae making it to the juvenile stage (Gosselin & Qian 1997). Thus, when a stressor compromises performance at these early stages, the effects will be felt at the population and community level. The thermo- and ph/pco 2 tolerance of fertilization and development in marine invertebrates has been recently reviewed (Byrne 2011); thus, we only summarize the findings on early life- stage processes in the context of our hypothesis, including any subsequent published studies. As early development stages of marine invertebrates are generally the life phases most sensitive to environmental stresses (Pörtner & Farrell 2008, Melzner et al. 2009), we hypothesized that processes at these early life- cycle stages and transitions between stages would respond negatively to OA conditions, with increased development times and reduced growth and settlement. Robust fertilization Fertilization in the water column is challenging for benthic invertebrates under ambient conditions, let alone with the added stressor of OA. Consequently, many OA studies have recently focused on this critical stage but have found differing results between even closely related species. While most studies suggest that fertilization of benthic invertebrates is robust to near- future OA (Table 1), some organisms appear susceptible, with reductions in sperm motility, speed and fertilization success reported (Havenhand et al. 2008, Parker et al. 2009, Morita et al. 2010, Reuter et al. 2010). Thus, we reject the hypothesis that fertilization is negatively affected in all benthic invertebrates studied. However, it is clear that the differences in results to date are very much dependent on the methods employed, as fertilization success is highly dependent on sperm- egg contact time, egg size, gamete age, and sperm density, amongst other factors (see Byrne 2011 for review). Indeed, multiple studies have observed that as pco 2 increased, higher sperm concentrations were necessary to achieve high fertilization success (Reuter et al. 2010, Albright & Langdon 2011), a result that would not be observed if eggs were oversaturated with sperm. Through early studies of the effects of ph on fertilization, it is known that three mechanisms/ stages can be affected. Firstly, the diffusion of pco 2 across the egg membrane can decrease the 156

31 Benthic invertebrates in a high- co 2 world internal ph of an egg, which can prevent sperm entering the egg (Johnson & Epel 1981). Secondly, a decline in internal ph can delay the fast block, which would allow the egg to be fertilized by more than one sperm and lead to egg death (Eisen et al. 1984, Levitan et al. 2007). It should be noted that many pco 2 studies do not take into account polyspermy and thus may erroneously report higher levels of fertilization success than the actual values post-fertilization (see Reuter et al. 2010). An increase in polyspermy, where an egg has been fertilized by more than one sperm, was observed in the giant scallop Placopecten magellanicus, leading to decreased fertilization in seawater with a ph less than 7.5 (Desrosiers et al. 1996). Thirdly, hypercapnia has been shown to reduce sperm swimming speed and consequently fertilization in broadcast spawning corals, sea cucumbers and sea urchins (Havenhand et al. 2008, Morita et al. 2010). In many organisms, the increased oxygen tension of seawater can overcome hypercapnic effects when sperm are released into the water column, even if seawater is at a reduced ph (Chia & Bickell 1983). Furthermore, it has been shown in sea urchins that activation of sperm motility occurs when intracellular proton concentration is reduced, causing an internal ph (phi) of and the release of the sperm (Johnson et al. 1983). Therefore, the organism may still have the ability to alter phi despite reduced external ph. From an ecological standpoint, if fertilization success is reduced under future OA conditions, the reproductive and population viability of broadcast- spawning marine species will be affected. For example, if sperm motility is negatively affected by OA, the chances of successful fertilization in the water column diminishes before we take into account external environmental factors that reduce the likelihood of sperm and eggs meeting. However, to allow comparisons to be made between different species, and on a geographical scale, standard test protocols need to be employed, including the use of a realistic range of sperm concentrations and realistic exposure times. Smaller, delayed embryos and larvae Post-fertilization, benthic invertebrates undergo embryonic and larval development of speciesdependent duration before they can settle and develop into adults. In many invertebrate species, the embryonic and planktonic larval phases have proved vulnerable to experimental OA conditions (Table 1), evident in extended development times, altered morphologies and reduced growth and survival. However, positive responses have also been observed, and thus the hypothesis that embryonic and larval stages are all negatively impacted by OA in all benthic invertebrates can be rejected. Whilst many studies have examined the larval response to OA, the embryonic stage and transition to larvae (embryonic mitosis/cleavage, morulae, blastulae, gastrulae, hatching) has been largely ignored, with only 10 invertebrate species embryos examined in near- future OA conditions to date. Sea urchin embryos appear fairly resilient, with four species examined showing no significant effect at realistic 2100 ph levels (Table 1). Conversely, embryonic development of the gastropod Littorina obtusata was negatively affected by a 0.5 ph unit drop, with a decline in embryo viability observed (88% compared to > 96% in control). Additionally there was a reduction in embryonic movement, heart rate and spinning rate, which in turn led to reduced size of hatchlings (Ellis et al. 2009). Decreases in hatching and development rates were also observed in the mussel Mytilus edulis (Gazeau et al. 2010) and the barnacle Semibalanus balanoides; however, in the latter development rate still resembled natural rates seen in populations found in similar locations (Findlay et al. 2010a). Where embryonic development is affected by OA, this may indicate the mechanisms employed to cope with acidosis and hypercapnia are not yet functional at this early stage of development. For example, the gametes, zygote and early cleavage stages are more vulnerable than cells during later ontogenetic stages as they lack specialized ion- regulatory epithelia that could compensate for changes in internal pco 2 (Melzner et al., 2009). At the larval stage, calcifying benthic invertebrates produce their protective or supportive shells/skeletons, which at this early stage will be fragile and vulnerable to both predation and dissolution. The development of these structures will be highly dependent on availability of CaCO 3 for skeletogenesis, as well as sufficient energy availability. Some invertebrates, such as abalone, secrete 157

32 Laura C. Wicks & J. Murray Roberts more soluble polymorphs of CaCO 3 at the larval stage (Weiss et al. 2002) and thus may be more vulnerable to OA at this critical stage. As with energy budget parameters of adult invertebrates, there appears to be no clear generarelated response at the larval stage to OA. Calcifying larvae, such as crustaceans, corals and echinoderms, varied in their response, with some species exhibiting reduced growth rates, altered morphology and reduced survival rates (Table 1). Malformed skeletons have been observed in response to OA in the oyster Saccostrea glomerata (Parker et al. 2009) and abalone Haliotis kamtschatkana (Crim et al. 2011) as a result of decreased calcification rates and retarded development. In the larval urchin Strongylocentrotus purpuratus, hypercapnia has been shown to induce a development delay by increasing the cost of maintenance and thus decreasing scope for growth (Stumpp et al. 2011b). Negative larval responses are not ubiquitous; non- calcifying larvae of the seastar Crossaster papposus grew faster when reared at a ph 0.4 units below ambient, with no visible effects on survival or skeletogenesis (Dupont et al. 2010). Similarly at low ph, survival of urchin larvae Strongylocentrotus droebachiensis and three non- calcifying tunicate species was enhanced compared to ambient conditions (Dupont & Thorndyke 2008). When larval development had been unaffected by OA conditions, authors have suggested this may be due to predisposition to an already- stressful environment. For example, larvae of the Antarctic sea urchin Sterechinus neumayeri were the least affected by low ph ( 0.4 units) compared to tropical and temperate sea urchin species (Clark et al. 2009), as the former evolved in an environment with naturally high CO 2 levels. Predisposition was suggested to explain differences in larval response to OA between the oysters Crassostrea gigas and Saccostrea glomerata, with the former less sensitive to ph reductions than the latter (Parker et al. 2010). Crassostrea gigas has a large geographic distribution and temperature tolerance, and the temperate urchins regularly experience seasonal environmental changes. Interestingly, the negative effects of OA on the larval stages of the oysters C. gigas and S. glomerata were amplified when fertilization was also conducted in low- ph seawater, highlighting the importance of studying cumulative effects of OA. A ph reduction of 0.2 units from ambient caused a significant reduction in the percentage of embryos reaching the D- veliger stage, an increase in percentage of abnormal larvae and a reduction in larval size (Parker et al. 2010). Recent studies have begun to look at the impact of OA on gene expression in larval invertebrates. Todgham & Hofmann (2009) found the genes involved in biomineralization, cellular stress response and energy metabolism in the sea urchin Strongylocentrotus franciscanus were suppressed during the period that larvae were exposed to reduced ph (0.05 units) conditions (178 genes of 1057 showed altered expression in acidified treatment). Also affected were genes involved in translational control, ion regulation, acid- base balance, cell cycle and development. The authors suggested that in response to OA, larval urchins are not upregulating genes in various pathways to compensate for the effect of OA and defend cellular homeostasis. However, these alterations may simply be a product of delayed development, which is commonly observed under low- ph conditions (Pörtner et al., 2010, Martin et al. 2011, Stumpp et al. 2011a). Expression patterns of calcification genes in larval urchins Paracentrotus lividus were unaffected by a 0.4 ph unit drop (Martin et al. 2011), with substantial transcriptomic plasticity observed in response to the short- term (3-day) ph drop. Reduced ph (0.15-unit drop) caused an upregulation of genes involved in acid- base regulation and downregulation of genes related to energy metabolism and biomineralization in the larval sea urchin Lytechinus pictus, in correlation with depressed metabolic activity (O Donnell et al. 2010). This suggests a molecular strategy is in place to maintain biological processes. Stumpp et al. (2011a) observed a significant upregulation of metabolic genes in response to elevated seawater pco 2, suggested to be related to increased energy demand for cellular homeostasis and calcification. Interestingly, Kurihara (2008) found no effect of high pco 2 (1000 μatm) on the expression of the spicule elongation gene or crystal regulation gene in the sea urchin Hemicentrotus pulcherrimus, although spicule size and morphology of larvae were affected. 158

33 Benthic invertebrates in a high- co 2 world It is important to consider the ecological consequences of seemingly small effects of OA at the larval stage. Alterations in morphology, development rate and behaviour could have severe implications for the survival of the organism. Indeed, the cost of planktonic life is high; the longer the larva is planktonic, the higher the risk of predation, starvation, and offshore transport (Dupont et al. 2010). Planktotrophic larvae, which have a long larval duration and feed exogenously, are suggested to be more at risk to OA than non- feeding lecithotrophic larvae, although evidence is sparse (Byrne et al. 2010b, Dupont et al. 2010, Byrne 2011). In addition, many invertebrates synchronize their larval release into the plankton to coincide with the spring bloom and thus a high concentration of prey; any delay due to increased development time is likely to lead to high larval mortality. Normal settlement but prolonged juvenile stage Assuming they survive the larval phase in the water column, larvae must then settle on a suitable substratum, a process that involves the recognition of water- soluble and substratum- bound chemical cues, followed by physical attachment to the substratum and ensuing metamorphosis to the juvenile stage. Environmental factors have been shown to disrupt settlement (Rodriguez et al. 1993); however seven of eight invertebrates examined were resilient to near- future OA conditions. The only negative response was observed in the coral Acropora palmata, with settlement success reduced by 45 69% at pco 2 concentrations expected for the middle and the end of this century (Albright et al. 2010). Other molluscs and corals examined showed no significant change in settlement rates (Table 1). Like their adult counterparts, invertebrate juveniles showed variable responses to OA conditions, with predominantly negative calcification responses. Coral juveniles are considered to be vulnerable to OA as postsettlement skeletogenesis occurs as part of the early calcification stage (Cohen & Holcomb 2009). Under low- ph conditions (drop of 0.4 units), polyp size was reduced in Porites astreoides and Acropora digitifera (Suwa et al. 2010), and polyp malformation was recorded in A. tenuis (Kurihara 2008). Two barnacle species showed no significant reduction in juvenile growth under OA conditions ( 0.4 ph units), but the calcium content of shells was reduced, a response enhanced by an increase in temperature (Findlay et al. 2010b). The vulnerability of juveniles to OA is likely to be linked to their metabolic priorities. As adults, maintenance and reproduction are key energy- driven processes. However, juveniles use their energy primarily for growth to enable successful colonization and metamorphosis from larva to adult. These developmental stages have been shown to be vulnerable to OA, particularly for invertebrates that use highly soluble amorphous calcium carbonate (ACC) at the onset of calcification, which later stabilizes into less- soluble forms of CaCO 3 (Brecevic & Nielsen 1989, Politi et al. 2004). The acquisition of ACC has been demonstrated during sea urchin larval spine formation (Beniash et al. 1997, Politi et al. 2004) and mollusc shell formation (Weiss et al. 2002, Marxen et al. 2003), and it is thought to be used by corals (Meibom et al. 2004). The dominance of negative calcification responses at the juvenile stage, compared to adulthood, is likely due to the experimental conditions. Most OA studies involving juveniles have subjected the invertebrates to OA conditions from the embryonic or larval stage, and thus the responses measured at the juvenile stage are cumulative (whether positive or negative) and thus more ecologically relevant. Conversely, studies of adult invertebrates have tended to introduce already- mature adults and monitored their response. In future experiments, it is important to understand carry- over effects of OA between life- cycle stages, with even seemingly minor effects on the fitness of larvae and juveniles carrying over to the adults, reducing their fitness and possibly augmenting their vulnerability to competition and predation. Although this section only summarizes the key findings on OA impacts at the early life stages, it is clear that the impact of OA varies profoundly between species at each stage of their life cycle. The larval stage of echinoderms and molluscs appears most vulnerable, and the postsettlement stage is most severely affected in corals, although much more evidence is required. These differences are 159

34 Laura C. Wicks & J. Murray Roberts partially explained by the fact that most echinoderms and molluscs start shell and skeleton synthesis at their larval stage, whereas corals start at the settlement stage. In general, early life stages exhibit reduced growth, increased development time and altered morphology, but the overall pattern is far more complex; variability in responses between closely related species is coupled with variation across geographic ranges. Finally, it should be noted that experiments typically do not include predators or spatial competitors and do not consider the advective loss of eggs, embryos or larvae. The key issue therefore is to ensure experiments are ecologically relevant and as representative as possible of the natural world. Sensory capacity and behaviour The ability of an organism to detect and avoid predators is one of the most important behaviours to ensure survival. If any learned or innate behaviours are affected by OA, the prospects of survival of an organism are reduced. Only one study of OA effects on behaviour exists for benthic invertebrates. The gastropod Littorina littorea exposed to low- ph conditions for 15 days increased its avoidance behaviour in response to a predator cue, compared to snails in ambient ph (Bibby et al. 2007). Associated with this increased avoidance was a decline in shell thickening, suggesting compensation for a lack of morphological defence. With only this study to date, the behavioural response of benthic invertebrates to OA is clearly a significant gap in our knowledge. Immune response and disease susceptibility OA conditions could potentially make benthic invertebrates far more susceptible to other stressors in the environment. It is well known in coral reef research that anomalously high temperature and other environmental stresses can influence the severity and dynamics of infectious diseases (Bruno et al. 2007). Thus it is hypothesized that OA will increase the susceptibility of benthic invertebrates to disease. Our hypothesis can neither be accepted nor be rejected in light of the lack of evidence to date, with only six studies examining the effect of near- future OA on immune response and disease susceptibility in benthic invertebrates. Bibby et al. (2008) showed a 0.1-unit ph reduction from ambient was enough to suppress the immune response of the mussel Mytilus edulis after 32 days, a possible result of elevated calcium ions in the haemolymph inhibiting signal transmission. Reduced ph, although at a level lower than predicted for the end of the century (7.3), also suppressed immune response, measured as cytotoxic activity, in the mussel Mytilus galloprovincialis, making the animals more vulnerable to pathogen aggression (Malagoli & Ottaviani 2005). Similarly, Burnett & Burnett (2000) found that the oyster Crassostrea virginica produced significantly fewer reactive oxygen intermediaries when exposed to hypercapnia and hypoxia (ph unknown), which indicates a suppressed immune response. It is known that during stress, immune defence is affected by the transfer of energy allocation to maintain protein integrity via the chaperone Hsp70 (Mayer & Bukau, 2005), as well as transferring peptides through the cell (Moseley 2000). As such, upregulation of Hsp70 has been observed in both oysters (Crassostrea virginica, Chapman et al. 2011) and seastars (Asterias rubens, Hernroth et al 2011); however, after a 6-month exposure this upregulation was not maintained in the seastar. It was suggested that it is too energetically costly for seastars to over-express Hsp70 in the long term (Clark et al. 2007), and that lowered ph inhibits transcriptional or translational levels (Langenbuch et al. 2006, Todgham & Hofmann 2009). At the cellular level, elevated pco 2 and decreased ph can disrupt cellular pathways, compromise enzyme integrity, and reduce lysosomal health (Beesley et al. 2008, Bibby et al. 2008). Lysosomes digest the foreign bacteria or waste that invades cells and also help repair cell damage; thus, a leaky lysosome caused by OA can lead to increased permeability to substrates, activation 160

35 Benthic invertebrates in a high- co 2 world of previously latent hydrolytic enzymes (which can lead to cell death), disruption of normal lysosomal function and possible release of hydrolases into the cytoplasm, where they could cause cytolytic damage (Beesley et al. 2008). In addition, low ph (~0.5-unit drop) has been shown to lead to upregulation of peroxinectin, a gene implicated in phagocytosis and a process important in clearing bacteria from haemolymph (Chapman et al. 2011). Coelomocytes and phagocytic capacity were also reduced in Asterias rubens, which has potentially serious consequences for pathogen resistance (Hernroth et al. 2011). Despite the vast array of evidence that thermal stress increases disease susceptibility in corals (e.g., Jones et al. 2004, Bruno et al. 2007), little is known of the impact of reduced ph. Coral pathogens have different growth rates under different ph levels, and thus some pathogens may become more prevalent in a future of OA, whilst others may become scarce. Recently, it has been shown that corals maintained at a lower ph than ambient had an increase in disease- and stress- induced bacteria (Vega Thurber et al. 2008); however, the study used extreme ph values (1.2 unit ph drop, respectively). Studies are urgently needed to assess how projected ph levels up to the end of the century will affect disease susceptibility in both corals and other benthic invertebrates. Community- level responses Any reduction in wider health or loss of species due to their sensitivity to OA will affect the dynamics of their community, whether via their role in the food web, in creating habitat complexity, in the cycling of carbon or nutrients, or in the biological regulation of competition. However, assessing community- level impacts of OA is a complex challenge as it is not something easily replicated under artificial conditions. There has been some success in using large- scale mesocosms (e.g., Biosphere 2, Langdon et al. 2003); however, when mindful of caveats (see Methodology and Approaches section), naturally high- co 2 areas have proved useful as proxies for what the future may hold, as well as extrapolating potential responses from individual organisms to a community level. We examine the current understanding of community- level responses, addressing the hypothesis that OA leads to altered community composition and dynamics. Abundance and distribution Importantly, the community dynamics of an ecosystem can be altered by OA if even just one species is vulnerable to the changing water chemistry. Many early studies of the effect of OA used survival as a measured response; however, the majority of organisms examined at near- future pco 2 levels showed more subtle responses than mortality. Exceptions include the barnacle Semibalanus balanoides, which after 104 days at 0.4 ph units had 22% lower survival than controls (Findlay et al. 2010a), whilst the coral Acropora cervicornis experienced 81% mortality after 8 weeks at 0.2 ph units (Renegar & Riegl 2005). If early life- cycle stages are severely affected by OA, such as in the brittlestar Ophiothrix fragilis, which exhibited 100% larval mortality in response to a 0.2-unit ph drop, extinction is likely within the foreseeable future (Dupont et al. 2008, 2010). Recently, mesocosms have been successfully used to assess changes in benthic invertebrate community structure and biodiversity in response to OA and warming. Hale et al. (2011) observed alterations in community composition after a 60-day period at a ph reduced by 0.3 units, despite no significant difference in species diversity and number of individuals. The effects were amplified when combined with increased temperature (+4 C). Differential responses between species to OA conditions, with members of some phyla more vulnerable to the changes than others, indicate that the response of a community will depend on the general response of each phylum, as well as species- specific responses and alterations in their ecological interactions. Alterations in benthic community structure in response to enhanced pco 2 have been observed at the volcanic CO 2 vents near Ischia in the Mediterranean Sea, with a substantial shift in the 161

36 Laura C. Wicks & J. Murray Roberts community composition from areas of normal ( ) to low ph ( ) and a 30% reduction in species number (Hall- Spencer et al. 2008). A wide range of benthic macroorganisms was affected by the low ph. For example, aragonitic calcifiers such as the corals Caryophyllia and Cladocora were present outside vents but absent where mean Ω arag was 2.5 or less (with a minimum of recorded over a 1-month period). In fact, the only cnidarian not affected by the low ph was the anemone Anemonia viridis, which may have benefited from the increased pco 2 for photosynthesis by its dinoflagellate symbionts. Overall at these vents, there was no indication of adaptation or replacement of sensitive species by others capable of filling the same ecological niche. Reductions in coral diversity, recruitment and abundance were also observed on the shallow CO 2 vents of Papua New Guinea (PNG, Figure 6), with massive Porites corals dominating over branching, foliose and tabulate corals at lowered ph (0.3-unit drop, Fabricius et al. 2011). However, these Porites colonies at the lowered ph were visibly paler, were heavily colonized by macrobioeroders, and had growth rates 30% lower than expected given their latitude (Lough 2008, Fabricius et al. 2011). At both Ischia and PNG, resilient non- calcareous macroalgae dominated regions of low ph. Species such as these that are abundant in the variable high- pco 2, low- ph regions may well be inherently adapted to cope with the variation, suggesting some keystone species may be able to cope with projected ocean conditions for the future. Assessing depth distributions of organisms in relation to the ASH allows us to see how OA has affected their distribution to date and the changes that may occur in the future. For example, in examining the depth distribution of non- calcareous corallimorpharians and calcareous scleractinians, Fautin et al. (2009) found non- calcareous corals at significantly greater depths than calcareous forms, suggesting their distribution is regulated by the ASH. However, a lack of unbiased data means further information is needed to test this statistically. The prevalence of cold- water corals in the North Atlantic, whilst somewhat a function of sampling bias, may also be related to ASH depth (Guinotte et al. 2006). Despite extensive surveys in the North Pacific, cold- water scleractinian bioherms have not been documented to the extent of the North Atlantic, perhaps due to the shallow depth of the ASH throughout the former ( m compared to > 2000 m in the Atlantic). Instead, the North Pacific is dominated by octocorals and stylasterids, which use calcite, the less- soluble form of CaCO 3, to build their spicules and skeletons (Cairns & Macintyre 1992). Where scleractinians are present in the North Pacific, they are found in close proximity to, or slightly shallower than, the ASH, making them vulnerable to any future shallowing of the ASH. Worryingly, an apparent lack of vertical larval dispersal of the cold- water coral Desmophyllum dianthus between stratified water bodies suggests deep populations of the coral are unlikely to colonize shallow water as the ASH rises and deep waters become uninhabitable (Miller et al. 2011). On a larger scale, modelling can, to some extent, be used to predict the effect OA will have on the distribution patterns of benthic organisms. By using Markov chain models over a 9-year period, Wootton et al. (2008) found strong links between in situ benthic species dynamics and variation in ocean ph, with calcareous species generally performing more poorly than non- calcareous species in years with low ph. They suggested that OA will cause reductions in large dominant calcifying animals, little change in calcareous coralline algae, and increases in non- calcareous algae and subdominant calcareous acorn barnacles. Similarly, Findlay et al. (2010c) used models to assess the relative influence of sea- surface temperature and OA on a barnacle population and found that although temperature was the main driver in the population dynamics at the study site, small decreases in ph can have a significant impact on community structures by further lowering recruitment in some organisms. Competition and predation The competitive ability of a species will govern its abundance within an ecosystem, whether it is competition for food or space, and as a corollary any reduction in physiological performance will 162

37 Benthic invertebrates in a high- co 2 world HC cover (%) %, P = Massive porites (%) %, P = Juv HC m %, P = Juv porites m %, P = Juv HC richness %, P = Juv SC richness %, P = CCA cover (%) %, P = Non calc MA (%) ph predicted 14%, P = Figure 6 (See also colour figure in the insert) Progressive changes in reef biota along a ph gradient at Upa- Upasina Reef, Papua New Guinea. Red and blue points indicate high and low pco 2 transect sections, respectively, and mean ph was predicted from seawater measurements. Low pco 2 regions measured ph at total scale and ppm pco 2, high pco 2 regions measured ph , ppm pco 2. The black lines indicate the log- linear fits, and the grey bands indicate upper and lower 2 SE. Also presented are the percentage variance explained by ph and the significance of the relationships. HC, hard corals; SC, soft corals. (From Fabricius et al with permission from Macmillan Publishers.) 163

38 Laura C. Wicks & J. Murray Roberts reduce an organism s competitive ability, albeit dependent on the relative change compared to its competitors. For example, the persistence of scleractinian corals in an ecosystem depends in part on the factors that constrain excessive biomass of algae that compete with corals for space. Evidence suggests that OA is likely to reduce coral growth rates (e.g., Jury & Whitehead 2009) in parallel with increasing algal photosynthesis and growth rates (Connell & Russell 2010), leading to a coralalgal phase shift (Ostrander et al. 2000). This in turn will modify the ability of other invertebrates, which rely on the coral skeleton as habitat, to recruit to the reef (Booth & Beretta 2002), as well as increasing grazer pressure on the algae, which will lead to excessive bioerosion and potentially the loss of coral framework (Graham et al. 2006). Alterations in dominance between species in a community, such as with corals and algae seen in natural CO 2 vents (Fabricius et al. 2011), are likely to become more common with OA, including the potential for new invasive species to colonize rapidly and outcompete previously dominant species. Many organisms would be more vulnerable to predation under future OA conditions, either due to loss or weakening of CaCO 3 shells or skeletons (e.g., Talmage & Gobler 2009) or to extension of the vulnerable larval stage at which mortality is already high under ambient conditions (Rumrill 1990, Morgan 1995). Equally, larger calcifying organisms have been suggested to be more vulnerable to OA, which will have a knock- on effect in terms of density- dependent predation (Paine 1966). Smaller organisms that use calcifying invertebrates as shelter will also be more vulnerable to predation if their shelters reduce in size or abundance. Acidification- induced disruption of predator avoidance strategies could occur in some organisms, a response already reported in the snail Littorina littorea (Bibby et al. 2007). Recent reports of disruption of predator cues in fish are worrying (Dixson et al. 2010) and highlight a key area of research that needs to be addressed in benthic invertebrates. There is clearly a need for more detailed information on the impact of anticipated increases in pco 2 and its synergies on long- term processes such as competition, fecundity, recruitment and grazing (Hoegh- Guldberg et al. 2007). Mesocosm experiments and field manipulation will help achieve these goals by assessing community dynamics, building on the success of CO 2 -controlled mesocosms of plankton communities (Riebesell et al. 2008). Nutrient and carbon cycling Benthic invertebrates play an important role in the cycling of nutrients and carbon within the water column and underlying sediment. In particular, their presence and activity can set or modify the geochemical nature of the sediment, which governs the type and distribution of the microbes involved in the transformation of nutrients (Ford et al. 1999). In altering the composition of benthic infaunal communities, OA has been shown to alter the flux of key nutrients across the sedimentwater interface significantly, particularly in permeable sediments (Widdicombe & Needham 2007, Widdicombe et al. 2009). Specifically, their evidence strongly suggests that reduced sediment nitrification and silicate flux occurred in response to increased seawater acidification. Similarly, OA was shown to have a significant effect on the bioturbating role of the ophiuroid Amphiura filiformis, altering the relationships between its density and sediment fluxes of nitrate, ammonium, phosphate and silicate (Wood et al. 2009). Altered nutrient flux is attributable to an increased demand for oxygen or food by these ophiuroids or indirectly by muscle wastage that reduces their capacity to bioirrigate. Community- level assessments of OA effects on nutrient cycling found the relationship between species richness and ammonium concentrations was modified by OA, negatively impacting nutrient release (Bulling et al. 2010). Ocean acidification could also affect carbon cycling through bottom- up controls involving phdependent speciation of nutrients and metals (Huesemann et al. 2002), which in turn may alter species composition and rates of primary productivity. On a global scale, robust information on the flux of nutrients and biogeochemical cycles under the influence of enhanced pco 2 is scarce. Theoretical considerations of nutrient equilibria with ph (Zeebe & Wolf- Gladrow 2001) would 164

39 Benthic invertebrates in a high- co 2 world suggest significant reductions in phosphate concentration and alteration of the ammonia- ammonium equilibrium at a ph decrease of 0.3 units (although two independent studies have found little or no change in phosphate and nitrate under altered ph; Tanaka et al. 2007, Rees et al. 2009). For benthic invertebrates, much remains to be learnt of how OA- induced changes in nutrient cycles will directly and indirectly impact their lifestyle. Marine calcifiers play an important role in carbon cycling, with biologically mediated reactions (precipitation and dissolution) contributing significantly to the global carbon balance (Doney et al. 2009, Lebrato et al. 2010). If marine calcifiers experience physiological failure (survival and calcification), the global carbon cycle will also be affected; however, the extent of the effect is unknown. Impacts have been examined on a local scale on coral reefs, where photosynthetic CO 2 fixation and CO 2 release by calcification are tightly coupled, with net community calcification dependent on the aragonite saturation state of the seawater (e.g., Leclercq et al. 2002, Langdon & Atkinson 2005). A recent mesocosm study found a decline in net community calcification occurred at enhanced pco 2 due to increased dissolution and reduced calcification (Andersson et al. 2009). However, extrapolating such results up to a global level to estimate thresholds is problematic due to the inherent temporal and spatial variability in calcifying communities (Yates & Halley 2006, Andersson et al. 2009). Where net dissolution occurs, a decrease in the contribution of CaCO 3 to the ballasting of organic carbon to the deep sea may occur (Armstrong et al. 2001, Klaas & Archer 2002), causing more organic carbon to remineralize in shallow waters and decreasing the ocean s CO 2 uptake efficiency. It should be noted that although dissolution consumes CO 2 and thus can act as a negative feedback to buffer OA, the rate at which this occurs will not be sufficient on a global or regional scale (Andersson et al. 2005). Food webs and trophic dynamics Benthic marine invertebrates span all trophic levels in marine ecosystems and are important food sources at higher trophic levels. Many benthic invertebrates feed on small planktonic calcifiers, which are known to have variable sensitivities to OA (e.g., Comeau et al. 2010). The strength of trophic interactions can change when climate change differentially affects consumer and resource species (Phillippart et al. 2003, Fabry et al. 2008); however, experimental difficulties mean that the impacts of OA on trophic dynamics are not well understood. From a bottom- up perspective, losses of plankton and juvenile shellfish, for example, will alter or remove trophic pathways and intensify competition among predators for food all the way up the food chain (Richardson & Schoeman 2004). From a top- down perspective, changes in the predator field will have implications for feeding intensity on benthic invertebrates. The effect of OA on food web and trophic dynamics will be determined by the vulnerability of individual species to OA. For example, if keystone species are vulnerable to OA, their removal will have a disproportionate effect on the trophic dynamics of a system, as these species maintain the organization and structure of entire communities (Fischer et al. 2006). As such species often occur in small numbers or have small biomass, they are frequently vulnerable to local extinction. Functionally unique species and species from intermediate trophic levels are also crucial to maintaining food webs and thus benthic biodiversity. As discussed, OA may cause changes in the export of carbon from calcifying systems, which will in turn reduce sediment community oxygen consumption, bioturbation intensities, sediment mixed- layer depths, faunal biomass and body sizes of invertebrate taxa. Shifts in the quality of sinking particulate organic carbon caused by community changes at the planktonic level will alter the nutritional quality of this food material, favouring success of some benthic species and reducing success of others (e.g., Hudson et al. 2004). The complexities of the food web and potential effects of OA are not easily studied, but only once we understand this will we truly appreciate which communities will exhibit resilience to future OA. 165

40 Laura C. Wicks & J. Murray Roberts Ecosystem engineers Many of the calcifying organisms threatened by OA act as ecosystem engineers, providing habitat and other services for a rich diversity of organisms. For example, cold- water corals are arguably the most 3-dimensionally complex habitats in the deep ocean, providing niches for many species, with more than 1300 species associated with cold- water coral reefs in the north- eastern Atlantic (Roberts et al. 2006, 2009). Likewise, mollusc shells, whether dead or alive, introduce complexity and heterogeneity into benthic environments, ultimately changing the availability of resources to other organisms (Gutiérrez et al. 2003). Like their cold- water counterparts, tropical corals are important ecosystem engineers, but bleaching of these reef- building corals has been increasing in frequency and magnitude, with mass coral bleaching primarily attributed to a combination of light and abnormally high seawater temperatures (Hoegh- Guldberg 1999). Even 1 2 C above summer maximum has been shown to cause a breakdown in the symbiotic relationship between coral host and symbiotic algae (Hoegh- Guldberg 1999), an effect exacerbated by enhanced pco 2. Indeed, corals were found to bleach at lower temperatures in acidified water (Anthony et al. 2008). If OA reduces calcification of live corals and molluscs, available habitats for other organisms will be reduced. On coral reefs, coralline algae play an important structural role, reinforcing the skeletal structure of corals, filling cracks and cementing together much of the sand, dead coral and debris, creating a stable substratum and reducing reef erosion (reviewed by Nelson 2009). In addition, coralline algae have been shown to release compounds that have been implicated in the settlement and morphogenesis of a range of species, including tropical corals (Morse et al. 1988). The effect of OA on calcified macroalgae is not fully understood, but whether the impact is positive or negative, there will be a knock- on effect to the highly diverse species with which they associate. Dissolution of carbonate structures will also have ecosystem- level ramifications, with OA projected to cause dissolution of CaCO 3 skeletons and shells when saturation levels are less than 1. This level will be reached in deeper, cooler waters by the end of the century (Guinotte et al. 2006). Dissolution is an ongoing process in all coral reef environments, a result of the metabolic activity of microbes and endolithic microorganisms that generate corrosive conditions in microenvironments and sediment porewaters (Morse & Mackenzie 1990, Tribollet 2008). A decrease in seawater saturation state with respect to carbonate minerals will result in increasingly corrosive conditions and subsequently increased rates of dissolution, particularly in organisms with slow growth rates. Dissolution has already been seen in the barnacle Semibalanus balanoides when subject to a 0.4-unit drop (Ω arag < 1), measured as the difference between the proportion of calcium in dead and live barnacles (Findlay et al. 2010b). In contrast, when examined using scanning electron microscopy, skeletons from the coral Oculina arbuscula showed no dissolution when exposed to undersaturated conditions for 60 days (Ω arag < 1, Ries et al. 2010). Using natural CO 2 vents as a proxy for OA, dissolution of the bryozoan Myriapora truncata was found to be significantly higher in low- ph regions (7.4), but interestingly, this species increased its calcification rate under the acidified conditions, as gross calcification was higher in the low- ph regions (Hall- Spencer et al. 2008). Whether organisms can prevent dissolution at lowered ph is an urgent question to be addressed, particularly as it is predicted that almost all coral reefs will be in a state of net dissolution once atmospheric CO 2 concentrations reach 560 ppm (Silverman et al. 2009), a level that is projected to occur around the year Future direction and approaches The variable responses to OA between species, habitats and over time make this one of the most complex challenges facing twenty- first century research. It is clear from the literature that not only do we need to assess the response of multiple energy budget parameters of individual organisms to 166

41 Benthic invertebrates in a high- co 2 world enable future predictions, but we also need to undertake experiments that closely replicate natural conditions and variability. For example, OA is occurring concurrently with ocean warming, as well as an upsurge of anthropogenic impacts such as pollution and overfishing. Individually, these effects are partially understood, but we know worryingly little about synergistic effects. Furthermore, very few studies have assessed the ability of organisms to adapt or acclimate to near- future OA conditions. Although the rate of ocean carbonate chemistry change is more rapid than at any previous time in Earth s history, this rate is still considerably slower than could possibly be replicated in experimental settings, and thus significant experimental design compromises must be made. In this final section, we examine what little is known of additional stressors and acclimation/ adaptation potential, identifying how OA studies can be developed to examine these crucial factors and how we perceive OA research should be directed. Additional stressors In addition to the challenge benthic invertebrates face with respect to OA, there are multiple concurrent stressors that need to be examined in light of their cumulative or synergistic physiological and ecological impacts. Local- scale stressors include eutrophication, increased terrestrial run- off, pollution, storm events, overfishing, invasive species and reduced salinity in polar regions due to ice melt. To address the impact of these factors, experimental manipulation is possible, evident in studies that examined the effect of nutrient enrichment with OA on corals (Marubini & Atkinson 1999, Langdon & Atkinson 2005, Renegar & Riegl 2005, Holcomb et al. 2010), which could be used as a basis to design experiments and include the role of eutrophication. However, to date, we are not aware of any studies that have addressed these synergies. At a regional scale, alterations in ocean stratification and sea- level rise are predicted, with little known of their potential impacts on benthic communities. It is predicted that by the end of the century, there will be an increase in upper- ocean stratification, in part driven by upper- ocean warming (low- to- temperate latitudes) and upper- ocean freshening (high latitudes). This in turn will lead to reduced uptake of anthropogenic CO 2 and thus an increase in the concentration of atmospheric CO 2, thereby augmenting surface OA. Importantly, there will also be an expansion of the oxygen minimum zones, caused by a reduction in the transport of oxygen from the near- surface ocean to the interior (Keeling et al. 2010). The expansion of these zones will increase the area where O 2 levels are too low to support many macrofauna, and profound changes in biogeochemical cycling will occur. As thresholds for hypoxia typically depend not only on O 2 levels but also on levels of CO 2 and temperature, the consequences of deoxygenation in a warming world must be considered concurrently with the effects of warming and acidification (Pörtner & Farrell 2008, Keeling et al. 2010). Both expansion of the oxygen minimum zones and sea- level rise will be particularly detrimental for benthic invertebrates, as many are sessile and therefore limited in their ability to escape such pressures. However, those animals with pelagic larval stages may maintain their ability to shift distribution and thus might be especially likely to experience range shifts with global climate change (reviewed by Harley et al. 2006). Synergistic effect of temperature Few studies have assessed the synergistic effect that enhanced temperature with OA will have on benthic invertebrates (19 of the 92 studies in this review), despite the fact that enhanced temperature reduces the saturation state, which translates as fewer carbonate ions in the seawater available for calcification. As well as a synergistic negative effect of enhanced temperature on calcification with pco 2 being observed in multiple species (e.g., the corals Stylophora pistillata, Reynaud et al. 2003; Porites panamensis, Anlauf et al. 2011; and the barnacle Elminius modestus, Findlay et al. 2010c), other species have shown a tolerance to enhanced pco 2 but sensitivity to enhanced temperatures (e.g., Pistevos et al. 2011). For example, primary polyp growth of the coral Porites panamensis was 167

42 Laura C. Wicks & J. Murray Roberts reduced only marginally by a to 0.25-unit ph decrease; however, the combined effect of high temperature (+1 C) and lowered ph caused a significant reduction in growth of primary polyps by almost a third at 1 C (Anlauf et al. 2011). In contrast, while a ph drop of 0.35 units reduced calcification in the larval urchin Tripneustes gratilla, warming actually increased growth, although the net effect was negative (Sheppard- Brennand et al. 2010). In some organisms, temperature has been shown actually to enhance growth where pco 2 has no effect; for example, the brittlestar Ophiocten sericeum increased the growth of regenerated arms with a 3.5 C temperature increase, irrespective of ph when assessing realistic scenarios (Wood et al. 2011). Another species that may fare well with increased pco 2 and temperatures is the seastar Pisaster ochraceus (Gooding et al. 2009), which increased its relative growth by about 67% relative to control treatments over 10 weeks in high pco 2 alone (780 ppm, 0.06 ph unit drop), while a 3 C increase in temperature with high pco 2 increased relative growth by about 190% (no significant interaction between pco 2 and temperature). Further, urchins Strongylocentrotus droebachiensis exhibited improved responsiveness under warmer, high- co 2 conditions (H.O. Pörtner, personal communication, 11 May 2010). It is suggested that high CO 2 narrows the thermal window of an animal by enforcing limitations in the flexibility of the organism to respond to changing energy demand (Pörtner et al. 2005b). Warming leads to rising oxygen demand that can initially be met by oxygen supply through enhanced ventilation and heart rate (Zainal et al. 1992, Frederich & Pörtner 2000). However, when pco 2 and temperature rise, energy supply may not be sufficient to meet such demands, and the onset of anaerobic metabolism can occur (see Pörtner 2010 for a review). The sole study of the effect of pco 2 on thermal windows to date was in line with this projection, with thermal sensitivity in the crab Hyas araneus rising with increasing CO 2 concentration (Walther et al. 2009). As many benthic invertebrates already live on the edge of their thermotolerance limit, they may not survive even small changes in temperature when combined with OA. As temperature rises in the future, the distribution and abundance of benthic species will shift latitudinally and tidally according to their thermal tolerance and ability to adapt (Fields et al. 1993), to the extent that dispersal and resource availability will allow. Widespread biogeographical range shifts clearly occur in association with changing climatic conditions in marine environments, observed in abundant fossil records from the Pleistocene- Holocene transition (reviewed in Fields et al. 1993). On a shorter timescale, species distributional limits have also been observed to change, for example, during El Niño Southern Oscillation events (Keister et al. 2005). Predicting whether organisms have the ability to shift their distribution range requires additional attention to the environmental factors, in addition to temperature, which determine species range boundaries (Harley et al. 2006). Whether at a local or regional scale, when such stressors lead to shifted or degraded communities, they may be more vulnerable to OA. Conversely, the stress- tolerant species that remain following disturbance may prove more resistant to the effects of OA. In fact, the threshold level of pco 2 may actually increase in communities where only hardy species remain. This highlights the need for OA studies on a range of invertebrate species, from those seemingly most vulnerable to those most tolerant. The lack of studies stems from the difficulties associated with conducting empirically robust OA experiments, difficulties that are amplified when including additional environmental stressors. Many studies assessing the synergistic effects of OA and warming are still in their infancy, but there is now a global effort to address these impacts on a wide range of invertebrate responses. Adaptive response The ability of an organism to survive in a future of altered ocean chemistry will be a function of its acclimatory or adaptive potential, of which we know very little. Acclimation occurs at the 168

43 Benthic invertebrates in a high- co 2 world organismal level, where an organism will have short- or long- term physiological adjustments to an environmental perturbation. While short- term responses are usually reversible, long- term responses are not immediately reversible and often lead to the development of altered phenotypes, potentially improving performance in the modified environment. We have already seen examples of short- term acclimation in benthic invertebrates at the cellular level, including metabolic suppression and passive buffering of intra- and extracellular fluids. Furthermore, we have shown how some organismal processes are unaffected by altered pco 2 /ph, and thus these organisms may be acclimating to these conditions. However, use of these acclimatory mechanisms is usually at the energetic cost of another process, for example, internal acidosis is buffered via dissolution of the CaCO 3 skeleton or shell. Longer- term experiments on the effects of OA are urgently needed to establish whether organisms have the potential to acclimate to future altered ocean chemistry (see Methodology and Approaches section). This was demonstrated in a recent study on the cold- water coral Lophelia pertusa, which was able to compensate for adverse effects after a 6-month period, contrasting with the reduced growth and respiration experienced during 1-week incubations (Form & Riebesell 2011). Similarly, decreases in egg production in the urchin Strongylocentrotus droebachiensis after 4 months in OA conditions were shown to be short- term responses, as after 16 months no negative effects were observed (S. Dupont, personal communication, 1 July 2011). By combining long- term studies with investigations of the physiological mechanisms employed by an organism to maintain homeostasis at higher pco 2 and reduced ph levels, we should gain a better understanding of the varying sensitivities to OA. Adaptive responses occur at the population level over timescales covering multiple generations, where evolutionary changes in genotypes may occur (i.e., natural selection) to adapt a population to a modified environment. Adaptation to changing ocean chemistry depends critically on the extent to which populations can evolve in response to their changing environment, with a need for sufficient genetic diversity to express a range of tolerances for OA. Speciation events in the fossil record that coincide with changes in environmental CO 2 indicate adaptation has occurred in the past (Cornette et al. 2002), albeit across much longer timescales than current atmospheric changes would allow. Some benthic invertebrates are slow- growing, long- lived species, which means they are not likely to exhibit adaptation; in some cases, generation times are decades long, and thus there is only the potential for about 10 generations by Potential adaptations required to cope with OA include the ability to calcify at normal rates under OA conditions and changes in life- history strategy and immune/defence mechanisms. Although little is known of the adaptive potential of most benthic invertebrates, physiological variation between individuals in OA experiments indicates intraspecific variation in response traits, which is often regarded as noise in the response traits. Such variation among genetic individuals may be important in enabling future adaptation to reduced ph via natural selection, leading to a mosaic response. Variations between genetic individuals of a species in response to OA exist. Firstly, the clonal bryozoan Celleporella hyalina exhibited contrasting responses in growth rate and reproductive investment under a 0.4 ph unit drop for 15 days (Pistevos et al. 2011). Using a full factorial breeding design, Sunday et al. (2011) showed that both the urchin Strongylocentrotus franciscanus and mussel Mytilus trossulus exhibited phenotypic and genetic variation for development rates to future CO 2 conditions (1000 ppm, 0.4 ph unit drop), with the former exhibiting vastly greater phenotypic variation. Further studies of intraspecific variation in response to OA across a range of species will enable a greater understanding of the evolutionary potential for benthic invertebrates, a huge gap in our knowledge of OA effects. Habitat- induced acclimation and adaptation The ability of an organism to cope with changing environmental conditions is in many cases determined by the organism s past history; if an animal is in a habitat predisposed to variable environmental conditions, it may have mechanisms by which it can acclimate to future changes in OA. For example, growth of mussels (Mytilus edulis) from a naturally pco 2 -variable environment (375 to 169

44 Laura C. Wicks & J. Murray Roberts 2309 μatm pco 2, 7.49 to 8.23 ph over a 1-year period) was unaffected by a ph drop of 0.4 units (Thomsen et al. 2010). In contrast, estuarine M. edulis decreased their calcification by up to 62% in response to the same ph drop (Gazeau et al. 2007); thus, fjord mussels may possess alternative mechanisms to maintain physiological integrity under future OA conditions. This is further demonstrated by animals living in variably acidic and hypercapnic environments, such as deep- sea hydrothermal vents (Goffredi et al. 1997) or estuaries (Burnett 1997), which typically have a large capacity for acid- base regulation (Seibel & Walsh 2002). However, whilst organisms in variable environments may be able to tolerate current conditions, these habitats will be exposed to rates of change of ocean pco 2 that go well beyond the worst scenarios predicted for surface oceans (Caldeira & Wickett 2003) and thus may reach a tipping point beyond which an organism s processes suffer. Investigations of OA effects on single species across a wide latitudinal cline are urgently needed to exclude the influence of species- specific effects. Differences in fertilization responses to OA have been observed within the same species from differing habitats. For example, when individuals of the oyster Crassostrea gigas from New South Wales, Australia, were exposed to ph reductions of about 0.37 units, fertilization success was reduced by approximately 25% (Parker et al. 2010); however, those from western Sweden showed no significant effect with a 0.35-unit reduction (Havenhand & Schlegel 2009). The differences in results among studies, even when using the same species, may be linked to the aforementioned methodological differences, although it has also been suggested that habitat may play a role, as intertidal and shallow subtidal species may be inherently adapted to the fluctuating ph and hypercapnic conditions in their habitat (Byrne 2011). Similarly, the heightened calcification tolerance to OA of cold- adapted Antarctic urchin larvae, when compared to tropical and temperate species, led the authors to suggest that Antarctic urchins may possess greater mechanistic control over the chemical nature of the extracellular space that surrounds the site of calcification of the larval endoskeleton (Clark et al. 2009; Figure 7). A lower capacity for calcium incorporation in crab larvae (Hyas araneus) living at the cold end of their distribution range has also been observed, suggesting that they might be more sensitive to OA than those in temperate regions (Walther et al. 2011). Benthic invertebrates present in the cold waters of the Antarctic and Arctic, as well as those in the deep waters of the North Atlantic and Pacific, already naturally experience lower ph and carbonate ion concentrations than the global average and will be among the first affected by OA. Despite knowledge of the impending changes in ocean ph and temperature, there is a lack of studies in both these vulnerable regions and on a global scale (Figure 8). It is crucial that future research focuses on invertebrates from high- latitude areas as well as areas that exhibit naturally low ph, such as the northern and eastern Pacific and upwelling areas. These areas will be key to understanding the potential for adaptation, as may already have occurred, and will enable us to examine the speed and extent at which adaptation can occur and how gene flow and dispersal may affect future adaptation. Conclusions This review summarizes the vast variability in responses to OA between species, habitats, life- cycle stages and experimental systems. We know that OA is ongoing, affecting all of the oceans to varying degrees and causing physiological change for many organisms. To enable a better understanding of the future for benthic invertebrates in warmer, lower- ph seas, as well as understanding how an individual s entire functioning will be affected, we also need to look at the fundamental ecosystem principles, relationships, structure and roles of individual species and how these will be altered. We cannot go back to the past; changes have been made to the chemistry of the oceans that are irreversible on human timescales. Only by establishing whether benthic invertebrates have the ability to acclimate or adapt to these changes can we begin to predict the consequences of future climate change for marine ecosystems. 170

45 Benthic invertebrates in a high- co 2 world (A) Evechinus chloroticus Normal (ph 8.1) sm Lowered (ph 7.7) p/r a (B) Tripneustes gratilla Normal (ph 8.2) Lowered (ph 7.8) sm a a rs sm (C) Pseudechinus huttoni Normal (ph 8.1) Lowered (ph 7.7) sm sm p/r a (D) Sterechinus neumayeri Normal (ph 8.0) Lowered (ph 7.6) sm sm a Figure 7 Skeletal larval rods of urchins Evechinus chloroticus, Tripneustes gratilla, Pseudechinus huttoni, and Sterechinus neumayeri grown in normal or lowered ph seawater, imaged using scanning electron microscopy. The regions of the larval body from which rods were extracted are indicated on the representative larvae (sm, smooth surface; rs, raised surface; p/r, pitted/eroded surface; a, preparation artefact). Scale bars in whole larvae are 100 mm, and scale bars in micrographs represent 5 mm. (From Clark et al with permission from Springer Science + Business Media.) 171

46 Laura C. Wicks & J. Murray Roberts Figure 8 Locations of experimental OA simulations on benthic invertebrates, using realistic ph values up to the end of the century. Size of circle represents number of organisms studied. Acknowledgements We thank M. Carter, S. Dupont, F. Hopkins and S. Hennige for comments on the draft manuscript; J. Moreno- Navas for assistance compiling Figure 8; and the EPOCA community for valuable discussions. Support for this work was provided by Heriot- watt University s Environment and Climate Change theme. This chapter is a contribution to the U.K. Ocean Acidification Research Programme (Natural Environment Research Council grant NE/H017305/1) and the European Commission s Seventh Framework Programme (FP7/ ) projects EPOCA (grant agreement no ) and HERMIONE (grant agreement no ). EPOCA is endorsed by the International Programmes IMBER, LOICZ and SOLAS. References Agegian, C The biogeochemical ecology of Porolithon gardineri (Foslie). PhD thesis, University of Hawaii at Manoa. Albright, R. & Langdon, C Ocean acidification impacts multiple early life history processes of the Caribbean coral Porites astreoides. Global Change Biology 17, Albright, R., Mason, B. & Langdon, C Effect of aragonite saturation state on settlement and postsettlement growth of Porites astreoides larvae. Coral Reefs 27, Albright, R., Mason, B., Miller, M. & Langdon, C Ocean acidification compromises recruitment success of the threatened Caribbean coral Acropora palmata. Proceedings of National Academy of Sciences of the United States of America 107, Al- Horani, F.A Microsensor study of photosynthesis and calcification in the scleractinian coral, Galaxea fascicularis: active internal carbon cycle. Journal of Experimental Marine Biology and Ecology 288, Allemand, D., Ferrier- Pagès, C., Furla, P., Houlbrèque, F., Puverel, S., Reynaud, S., Tambutté, E., Tambutté, S. & Zoccola, D Biomineralisation in reef- building corals: from molecular mechanisms to environmental control. Comptes Rendus de l Academie des Sciences. Palevol 3, Ameye, L., Compère, P., Dille, J. & Dubois, P Ultrastructure and cytochemistry of the early calcification site and of its mineralization organic matrix in Paracentrotus lividus (Echinodermata: Echinoidea). Histochemistry and Cell Biology 110,