deep-sea research ii - max planck society · d department of biological and environmental sciences,...

The potential impact of ocean acidification upon eggs and larvaeof yellowfin tuna (Thunnus albacares)

Don Bromhead a,n, Vernon Scholey b, Simon Nicol a, Daniel Margulies b, Jeanne Wexler b,Maria Stein b, Simon Hoyle a, Cleridy Lennert-Cody b, Jane Williamson c,Jonathan Havenhand d, Tatiana Ilyina e, Patrick Lehodey f

a Secretariat of the Pacific Community, BP D5 Noumea, New Caledoniab Inter-American Tropical Tuna Commission, 8901 La Jolla Shores Drive, La Jolla, CA 92037-1509, USAc Department of Biological Sciences, Macquarie University, NSW 2109, Australiad Department of Biological and Environmental Sciences, University of Gothenburg, SE-452 96 Stromstad, Swedene Max Planck Institute for Meteorology, Bundesstr. 53, D-20146 Hamburg, Germanyf Collecte Localisation Satellites, Space Oceanography Division, 8-10 rue Hermes, 31520 Ramonville Saint-Agne, France

a r t i c l e i n f o

Keywords:Ocean acidificationCarbon dioxideTuna fisheriesMortalityGrowth

a b s t r a c t

Anthropogenic carbon dioxide (CO2) emissions are resulting in increasing absorption of CO2 by theearth's oceans, which has led to a decline in ocean pH, a process known as ocean acidification (OA).Evidence suggests that OA may have the potential to affect the distribution and population dynamics ofmany marine organisms. Early life history processes (e.g. fertilization) and stages (eggs, larvae, juveniles)may be relatively more vulnerable to potential OA impacts, with implications for recruitment in marinepopulations. The potential impact of OA upon tuna populations has not been investigated, although tunaare key components of pelagic ecosystems and, in the Pacific Ocean, form the basis of one of the largestand most valuable fisheries in the world. This paper reviews current knowledge of potential OA impactson fish and presents results from a pilot study investigating how OA may affect eggs and larvae ofyellowfin tuna, Thunnus albacares. Two separate trials were conducted to test the impact of pCO2 onyellowfin egg stage duration, larval growth and survival. The pCO2 levels tested ranged from present day(�400 μatm) to levels predicted to occur in some areas of the spawning habitat within the next 100years (o2500 μatm) to 300 years (�o5000 μatm) to much more extreme levels (�10,000 μatm).In trial 1, there was evidence for significantly reduced larval survival (at mean pCO2 levelsZ4730 μatm) andgrowth (at mean pCO2 levelsZ2108 μatm), while egg hatch time was increased at extreme pCO2

levelsZ10,000 μatm (nintermediate levels were not tested). In trial 2, egg hatch times were increased atmean pCO2 levelsZ1573 μatm, but growth was only impacted at higher pCO2 (Z8800 μatm) and there wasno relationship with survival. Unstable ambient conditions during trial 2 are likely to have contributed to thedifference in results between trials. Despite the technical challenges with these experiments, there is a needfor future empirical work which can in turn support modeling-based approaches to assess how OAwill affectthe ecologically and economically important tropical tuna resources.

& 2014 Elsevier Ltd. All rights reserved.

1. Introduction

1.1. Background

Anthropogenic (man-made) carbon dioxide (CO2) emissions areresulting in increasing concentrations of CO2 in the earth’s atmo-sphere (IPCC, 2007). This buildup in atmospheric CO2 is, in turn,causing a gradual warming and acidification of the earth’s oceans

(e.g. Caldeira and Wickett, 2003; Feely et al., 2004; Barnett et al.,2005). Both warming and acidification have the potential to affectthe distribution and population dynamics of many marine organ-isms (Raven et al., 2005; Fabry et al., 2008). Tuna populations arekey components of pelagic ecosystems and, in the Pacific Ocean,form the basis of one of the largest and most valuable fisheries inthe world (Williams and Terawasi, 2009). The sustainable manage-ment of this fishery requires an understanding of not only fisheryimpacts, but impacts of other factors upon population biomass andstructure over time. While fishery scientists are now attempting topredict how ocean warming will affect Pacific tuna populations(Lehodey et al., 2010, 2013), no one has previously investigated

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/dsr2

Deep-Sea Research II

http://dx.doi.org/10.1016/j.dsr2.2014.03.0190967-0645/& 2014 Elsevier Ltd. All rights reserved.

n Corresponding author. Mob.: þ61479119142.E-mail address: [email protected] (D. Bromhead).

Please cite this article as: Bromhead, D., et al., The potential impact of ocean acidification upon eggs and larvae of yellowfin tuna(Thunnus albacares). Deep-Sea Res. II (2014), http://dx.doi.org/10.1016/j.dsr2.2014.03.019i

Deep-Sea Research II ∎ (∎∎∎∎) ∎∎∎–∎∎∎

www.sciencedirect.com/science/journal/09670645

www.elsevier.com/locate/dsr2

http://dx.doi.org/10.1016/j.dsr2.2014.03.019



mailto:[email protected]





how ocean acidification (OA) may affect these species and asso-ciated fisheries.

1.2. The process of ocean acidification

Concentrations of CO2 in the ocean tend towards equilibriumwithCO2 in the atmosphere. To date, the world’s oceans have absorbedabout 30–50% of global man-made CO2 emissions (Feely et al., 2004;Sabine et al., 2004), substantially changing ocean water chemistry byincreasing concentrations of dissolved CO2 (aq), H2CO3 (carbonicacid), HCO3� (bicarbonate ions), and Hþ (hydrogen ions), anddecreasing concentrations of CO3

2� carbonate ions (Fabry et al.,2008). Increased concentrations of oceanic CO2 (i.e. increased pCO2)have lowered the average sea-surface pH by 0.1 units (i.e., making theocean more acidic and less alkaline) since the start of the industrialrevolution. It is projected that uptake of atmospheric CO2 by globaloceans will further reduce average sea-surface pH by0.3–0.4 units in 2100 and up to 0.7 units by the year 2300(Caldeira and Wickett, 2003, 2005). These represent larger and fastershifts in oceanic pH than is believed to have occurred in millions ofyears (Feely et al., 2004). These model based projections of increasingacidification are supported by recent time series of ocean pH andpCO2 off Hawaii in the Pacific Ocean and Bermuda in the AtlanticOcean (Doney et al., 2009; Bates et al., 2012).

These predictions describe globalmean changes in surface waters,but pH and pCO2 concentrations show spatial heterogeneity bothbetween oceanic regions (surface waters) (Fig. 1) and throughout thewater column (Fig. 2). Currently, surface layer pH values are lowest inhigher latitudes and areas where regular upwelling brings subsurfacewaters with lower pH to the surface, including in the eastern andcentral tropical Pacific (Fig. 2). Although seawater pH is expected todecrease globally (and pCO2 to increase), the rate of this change ispredicted to be greater in high latitudes, and lower in tropical andsubtropical waters (Fig. 1). The degree of change is also dependent onfuture anthropogenic CO2 emissions and whether these are higher orlower than “business as normal” IPCC projections, and may beimpacted by other predicted consequences of climate change, suchas changed ocean circulation patterns (IPCC, 2011).

1.3. Impacts on fish

OA is recognized as having the potential to impact marineorganisms via direct physical or physiological impacts upon indivi-duals which then propagate to the population level, or indirect

ecological impacts through impacts upon habitat and interactingspecies (e.g. predators, prey, competitors, symbionts, parasites).Evidence for direct impacts of OA upon key population processes(e.g. calcification, growth, development, mortality, reproduction,activity levels) has been presented for many marine organismsincluding phytoplankton and zooplankton, marine calcifiers such ascorals, molluscs, crustaceans, echinoderms, and cephalopods as wellas fish (see reviews by Raven et al., 2005; Fabry et al., 2008; Wittmanand Portner, 2013).

The direct influence of pH upon fish has been studied fordecades (e.g. see Brauner and Baker, 2009), although early studiestypically focused on adults and extreme pH levels. It is only morerecently that research has focused on early life stages of fish andtheir physiological responses to pH and pCO2 levels that areecologically relevant to those forecast to occur within the next100–300 years. While many of the original OA studies focused onreef species, a range of non-reef species have now been considered(e.g. Frommel et al., 2010, 2011; Bignami et al., 2013a, 2013b;Sundin et al., 2013; Tseng et al., 2013), but not tuna species.

Adult fish within the species studied to date generally show astrong capacity to compensate for prolonged exposure to elevatedpCO2 due to their ability to control acid–base balance by bicarbonatebuffering across the gills and to a more limited degree, via thekidneys (Brauner and Baker, 2009; Esbaugh et al., 2012). The early lifestages of fish may be more vulnerable to OA impacts than are adultfish, due to different modes of respiration and ion exchange (Jonzand Nurse, 2006; Pelster, 2008); a situation possibly exacerbated bytheir higher surface to volume ratio (Kikkawa et al., 2003; Ishimatsuet al., 2004). Exposure to elevated pCO2 has been found to adverselyaffect embryonic development (Tseng et al., 2013), larval and juvenilegrowth (Baumann et al., 2011), tissue/organ health (Frommel et al.,2011), and survival (Baumann et al., 2011) in some species of fish.However a number of studies have not detected any effect uponembryogenesis (Franke and Clemmesen, 2011), hatching (Frommel etal., 2012), growth and development (Munday et al., 2011a; Frommelet al., 2011, 2012; Hurst et al., 2012, 2013; Bignami et al., 2013a) orswimming ability (Bignami et al., 2013a; Munday et al., 2009b).

Otolith growth appears to increase in size and/or density in somespecies (Checkley et al., 2009; Munday et al., 2011b; Hurst et al.,2012; Bignami et al., 2013a, 2013b) possibly due to increased internalHCO2� concentrations that result from pH regulation (Esbaugh et al.,2012) with possible implications for auditory sensitivity (Bignami etal., 2013b). However, some species of fish have not shown increasedotolith growth (Franke and Clemmesen, 2011; Munday et al., 2011a;

Fig. 1. Projections of OA in CMIP5 simulations, calculated with the model MPI-ESM (Source: Ilyina et al., 2013; see also Bopp et al., 2013) for changes in pH in surface watersfrom (A) 1850 to 2010 and (B) 1850 to 2100 (following the “business as usual” scenario RCP8.5 as described in Van Vuuren et al. (2011) in which radiative forcing increases upto 8.5 W/m2 in the year 2100).

D. Bromhead et al. / Deep-Sea Research II ∎ (∎∎∎∎) ∎∎∎–∎∎∎2





Frommel et al., 2012). Further investigation is required given theimportant role of otoliths in detecting sound, acceleration and bodyposition (Bignami et al., 2013b).

Many studies show no direct relationship between “near future”levels of elevated pCO2 and survival (Munday et al., 2011a; Frommel etal., 2012) and it is believed that for many species, sub-lethal effects ofOA may present the largest risk to individuals and populations (Briffa etal., 2012). Somatic impacts such as reduced growth and size at age havebeen linked to increased mortality in natural fish populations due toincreased risks of predation and reduced ability to find food(e.g. Houde, 1989; Leggett and Deblois, 1994). Neurological and beha-vioral effects associated with elevated CO2 levels have also beenobserved in coral reef fish (and other organisms; Briffa et al., 2012).These include impaired mating propensity (Sundin et al., 2013),impaired ability to make settlement choices (Munday et al., 2009c),altered timing of settlement (Devine et al., 2012), impaired response topredator and prey olfactory cues (Allen et al., 2013; Cripps et al., 2011;Dixson et al., 2010; Nilsson et al., 2012), reduced escape distances (Allenet al., 2013), altered responses to visual threats (Ferrari et al., 2012a) andauditory signals (Simpson et al., 2011), behavioral lateralization(Domenici et al., 2012; Nilsson et al., 2012) and a reduced capacity tolearn (Ferrari et al., 2012b). Many of these studies used larval or juvenilefish. Such behavioral changes were recently linked in reef species to theimpact of elevated pCO2 upon a key brain neurotransmitter GABA-A(Nilsson et al., 2012), with potential implications for other species giventhe highly conserved nature of GABA-A across species.

Evidence suggests that the effects of OA will be species specificand many key uncertainties remain regarding its implications forfish populations, including:

� How impacts on predators will affect prey mortality rates (e.g.Allen et al., 2013),

� Potential parental environment effects (e.g. Miller et al., 2012);� The effect of natural variability in pH (pCO2) in a species

environment upon its past evolution and resilience to futureacidification (e.g. Bignami et al., 2013a, 2013b; Franke andClemmesen, 2011; Frommel et al., 2012, 2011);

� Individual variability in responses and species capacity torapidly adapt through natural selection (Munday et al., 2012);

� Indirect ecological impacts on fish populations through impactsupon habitat and interacting species populations (e.g. preda-tors and prey). While logistically challenging, such studies havebeen shown to be important in understanding pCO2 impactsupon marine populations (e.g. Alsterberg et al., 2013) andshould be a focus of future fish research.

The potential impacts of OA upon commercially valuable tro-pical tunas are currently unknown. The only experiment known tohave tested impacts upon a tuna species (Eastern little tuna) testedpCO2 levels that far exceeded those predicted using IPCC scenarios(Kikkawa et al., 2003). In the remainder of this paper, we describethe likely vulnerability of yellowfin tuna, and report the first set ofexperiments exploring the impact of OA on the early life stages ofyellowfin tuna.

1.4. Potential vulnerability of tunas: Yellowfin tuna (Thunnusalbacares)

Yellowfin tuna (T. albacares) are epipelagic, serial batch spawnerswhich inhabit all tropical and subtropical oceans of the world(Schaefer, 2001, 2009). The distribution and occurrence of yellowfintuna eggs and larvae may be largely determined by encounters withfavorable biological and physical conditions for growth and survival(Boehlert and Mundy, 1994; Lang et al., 1994; Wexler et al., 2007,

Fig. 2. Average present day pH in the global ocean surface waters and at depths of 50 m, 100 m and 500 m .(Source: Ilyina et al., 2013)

D. Bromhead et al. / Deep-Sea Research II ∎ (∎∎∎∎) ∎∎∎–∎∎∎ 3





2011), the physical structure of the water column and advectivewater mass movements (Boehlert and Mundy, 1994; Sabatés et al.,2007), and spawning location (Davis et al., 1990). Yellowfin larvaeinhabit subsurface waters within the mixed layer at depthso50 mand exhibit diel vertical migration (Boehlert and Mundy, 1994; Lauthand Olson, 1996; Owen, 1997). Suboptimal physical conditions (e.g.low dissolved oxygen levels, low water temperatures) restrict thevertical and horizontal distribution utilized by early life stages oftunas and thereby influence the spatial and temporal overlap withtheir food and predators, which could potentially affect subsequentlarval fish survival (Blaxter, 1992; Breitburg et al., 1999).

Pelagic species, such as tuna, are considered in general to haveevolved in a relatively more stable pH environment than coastaland reef species, and this, in addition to low blood pCO2 levels thatcan be expected to result from their high rates of gas exchange,may make them more vulnerable to changes in ocean pCO2 and pH(e.g. Nilsson et al., 2012). That said, yellowfin tuna spawn inequatorial (tropical) and subtropical waters from the far westernPacific Ocean to the far eastern Pacific Ocean, and in both nearcoastal waters and oceanic waters (Schaefer, 2001). Across thisvery broad region there is considerable seasonal and vertical/horizontal spatial variation in pH (Fig. 2) and pCO2.

The 5th IPCC assessment (IPCC, 2013) estimates a global averagedecline in surface pH of 0.3–0.32 by 2100 compared to today underthe high-CO2 climate change scenario RCP8.5, but there will be spatialvariation around global mean values, both between different oceanregions and vertically through the water column (Bopp et al., 2013).In the eastern tropical Pacific Ocean, surface water pH is on averagelower than in the western tropical Pacific and the pH at 50 m depth(in the eastern Pacific) is on average 0.54pH units less than at thesurface. Across the region of yellowfin spawning habitat, mean surfacewater pH is predicted to decrease between 0.26 and 0.49pH units by2100 relative to pre-industrial levels (under the high CO2 scenario inFig. 1) (Ilyina et al., 2013). In the eastern Pacific Ocean the pH in surfacewaters is predicted to decrease from 8.05 to about 7.73 while in thewestern Pacific Ocean, the mean decline in pH is projected to be0.40pH units (with a maximum predicted decline of 0.46) (Ilyina et al.,2013). Decreases in pH at depths accessible to yellowfin larvae may beeven higher and could result in further “compression” of suitablehabitat area. It is unknown whether the level of variability across thetropical Pacific has been sufficient to have conferred some evolvedresilience in tropical tunas to predicted future levels of OA.

To advance our knowledge of the impacts of OA upon tunapopulations and fisheries, a pilot study was undertaken at the Inter-American Tropical Tuna Commission’s (IATTC’s) Achotines Laboratorylocated in the Republic of Panama. The objectives of the pilot studywere to develop and test experimental protocols to examine thepotential effects of OA on yellowfin tuna egg fertilization, egg and larvaldevelopment, growth, and survival, and on the rapid selection ofresistant genotypes. The study also aimed to use results from thesetrials to assess levels of variability in tank specific survival responses tohelp design future experimental trials with sufficient statistical sam-pling power. This paper describes each of these elements, with theexception of egg fertilization and genetic components. It also discussesthe results in the context of current implications for potential impactson yellowfin populations and implications for future empirical andmodeling based approaches that will be required to fully assess the riskthat OA places on tropical tuna populations and associated fisheries.

2. Methods

2.1. Physical system design

Two separate experimental trials were conducted at the Acho-tines Laboratory (Fig. 3), in October and November of 2011, where

a broodstock population of yellowfin tuna has spawned on a near-daily basis since 1996 (Wexler et al., 2003; Margulies et al., 2007a).

The experimental system was housed in a semi-enclosedlaboratory and consisted of fifteen 840-L experimental tanks, eachnested inside a larger 1100-L tank filled with seawater to stabilizethe water temperature (although this still varied slightly with localambient conditions). Each experimental tank was an individualopen system with seawater inflow of 1.7 L min�1. Appropriatelevels of turbulence (to facilitate aeration of eggs and larvalfeeding) were achieved utilizing air diffusers delivering the CO2–

air gas mix into each experimental tank. Bleed valves were used asnecessary to ensure the turbulence levels in all 15 tanks weresimilar. Key environmental parameters (water flow, lighting, aera-tion and turbulence levels) were set to standard levels used atAchotines Laboratory and continuously fine-tuned to ensure theseparameters remained as uniform as possible.

Four pH treatment levels (6.9, 7.3, 7.7 and 8.1) were targeted ineach trial with three replicate tanks per treatment level. Theselevels were chosen based on results from the ocean-carbon-cyclemodels using the IPCC A1B Scenario (e.g., the Hamburg OceanCarbon Cycle model, as in Ilyina et al., 2009) and reference topublished studies on predicted pH levels (e.g. Fig. 1, Ilyina et al.,2013; Caldeira and Wickett, 2003) and take into account spatialvariation in predicted declines, not just the global average pre-dicted declines. Target pH levels 7.3–8.1 encompass potentialmean ocean pH levels estimated for the current oceans andpredicted for future oceans (to the year 2300).

The coastal waters supplying the Achotines Laboratory sea-water system were close to pH 8.2 during the trial period soambient seawater was used for the high pH level with only airused for aeration. The other three pH levels were maintained ineach set of replicate experimental tanks by varying the amount of

Fig. 3. The location of the Achotines Laboratory.






CO2 mixed with the air delivered through a mixing manifold foreach replicate distributing the gas mixture to air diffusers in eachtank. A Porter Cable C2002 air compressor delivered air through apressure regulator set at 2.1 bar (30 psi) to a precision electronicgas-flow controller (Sierra 810 Series Mass-Trak) and then to themixing manifold. The air flow rate to each of the pH levels variedfrom 2.3 to 6.9 lpm. The CO2 flowed from CO2 cylinders to apressure regulator set at 2.1 bar (30 psi) through an electronic gas-flow controller and to the mixing manifold. The CO2 flow ratevaried from 0.04 to 0.34 lpm. The use of CO2 was critical tomodifying water chemistry (i.e., increasing carbonic acid, increas-ing Hþ , lowering pH) in a manner consistent with CO2-inducedocean acidification (Gattuso et al., 2010).

2.2. Water chemistry measurements and estimates

The pH was measured in each tank every 2–3 h during thetrials (Table 1). Measurements of pH (NBS scale) were taken usinga handheld Royce Model 503pH/CO2 Analyzer with matchingRoyce pH probe, after checking initial readings for consistencyagainst two other pH meters (a handheld Hach HQ40d PortablepH, Conductivity, Dissolved Oxygen Multi-Meter and a tabletopAccumet XL25 Dual Channel pH/Ion Meter). New pH sensors wereused and calibrated using Hach pH calibration solutions. Tempera-ture, salinity, and dissolved oxygen levels were recorded approxi-mately every 8 h, and alkalinity every 2–3 days. Alkalinity wasmeasured with a LaMotte Alkalinity chemical test kit Model WAT-DR. The pCO2 (and a range of other carbonate system parameters)was estimated by inputting measured levels of pH, temperature,salinity and alkalinity into the Excel based CO2SYS macro (http://cdiac.ornl.gov/oceans/co2rprt.html; Lewis and Wallace, 1998).

Means and 95% confidence intervals for pH (measured directlyin tanks) and pCO2 (estimated) are provided in Table 1. Means and95% confidence intervals for water quality variables (temperature,dissolved oxygen, % oxygen saturation, salinity and alkalinity) andestimated carbonate chemistry variables are provided in theAppendix. Variation within and between tanks for non-carbonatevariables (e.g. temperature, salinity and oxygen) was low acrossboth trials, and temporal variation in these parameters occurredsimultaneously across tanks. Mean pH and pCO2 varied betweentanks within target pH treatments (i.e. tanks targeted at pH7.7 ultimately settled around slightly differing mean pHs) resultingin a spread of mean tank pCO2 values ranging from 356 μatm (pH8.233) to 10,644 μatm (pH 6.87) in trial 1 and from 471 μatm (pH8.11) to 9691 μatm (pH 6.91) in trial 2.

Temporal variation in pH was highest in tanks associated withtarget pH treatment level 7.3 (both trials) and in lower pH tanks

from day 5 of feeding in trial 2 (see Appendix). However, with pHbeing a log scale, variation in pCO2 tended to increase at lower pHtreatment levels and there is a clear positive correlation betweenstandard deviation of pCO2 and mean pCO2, with SD being higherat higher mean pCO2 (Appendix). A comparison of mean pCO2 withCV of pCO2 shows the CV of tanks in target treatment pH 7.3 to behigher than for other treatments (Appendix).

2.3. Egg and larval rearing

Each of the two trials was continuous and spanned threedevelopmental phases: egg phase, yolk-sac larval phase, andfirst-feeding larval phase (to 7 days after first feeding). Fertilizedeggs were collected two hours after spawning from the broodstocktank and randomly stocked at a density of 177 eggs/L (14,000 eggstotal per tank) in each of 15 cylindrical egg-incubation nets (79-Lvolume) nested one per experimental tank. Spawning time wasvisually determined by broodstock behavior and the presence ofeggs at the tank surface. Stocking density was made consistent inthe following manner: After determining hatch tank volume, eggswere mixed evenly through the tank (via gentle upwelling) priorto removing three 300-mL samples. The eggs were counted ineach of the three samples and the mean was used to calculate thenumber of eggs per litre in the hatch tank volume. The sameamount of eggs were stocked randomly in each of the 15 tanks.Yolk-sac larvae were then dispersed from the egg-incubation netsinto their respective experimental tanks 2 h after 50% of the eggshad hatched. The yolk-sac phase in yellowfin tuna larvae continuesuntil approximately 50–70 h after hatching depending on watertemperature. Yolk-sac-stage and feeding larvae were maintainedin the same experimental tanks until early on the seventh day offeeding (approximately 8.5 days post-hatching; 9.38 days totalfrom egg transfer), when the trials were terminated.

Larvae were fed cultured Brachionus plicatilis (rotifers) at densitiesof 3–5 rotifers/mL three times daily. Dense blooms of unicellular algae(500,000–750,000 cells/mL; “green water”) were maintained in eachtank to facilitate rearing (Margulies et al., 2007b).

2.4. Sampling protocols

During each phase (egg, yolk-sac, and first-feeding), 15–17 freshsamples were taken at various intervals from each tank to bemeasured (total length, notochord length, body depth at pectoral,body depth at vent) and processed for dry-weight determina-tion. Samples collected during each developmental phase werealso fixed and stored for later analyses of tissue histology andorgan development, feeding success, genetic variability and otolith

Table 1Mean pH and mean estimated pCO2 (μatm) for each tank and each trial (trials 1 and 2) with 95% CI indicated in parentheses. Target pH and experimental tank number arealso indicated.

pH targetted Trial 1 Trial 2

Tank Mean pH achieved (95% CI) Mean pCO2 (95% CI) Tank Mean pH achieved (95% CI) Mean pCO2 (95% CI)

8.1 13 8.23 (8.22–8.25) 348 (339–357) 15 8.11 (8.1–8.12) 460 (444–476)8.1 3 8.19 (8.18–8.2) 391 (384–398) 5 8.09 (8.08–8.11) 467 (450–483)8.1 8 8.21 (8.2–8.22) 366 (355–376) 6 8.1 (8.09–8.11) 464 (448–480)7.7 15 7.51 (7.49–7.52) 2413 (2338–2487) 13 7.63 (7.61–7.66) 1835 (1725–1946)7.7 5 7.62 (7.6–7.64) 1767 (1722–1811) 3 7.63 (7.62–7.65) 1599 (1529–1668)7.7 6 7.54 (7.53–7.55) 2145 (2099–2191) 8 7.63 (7.61–7.65) 1723 (1647–1799)7.3 12 7.36 (7.33–7.39) 3738 (3509–3966) 12 7.24 (7.21–7.26) 4731 (4473–4990)7.3 2 7.35 (7.31–7.39) 3958 (3644–4272) 2 7.26 (7.23–7.28) 4536 (4288–4784)7.3 9 7.33 (7.3–7.35) 3763 (3525–4001) 9 7.19 (7.17–7.21) 4931 (4701–5160)6.9 1 6.93 (6.91–6.94) 9200 (8970–9430) 1 6.96 (6.94–6.98) 8983 (8611–9355)6.9 10 6.87 (6.85–6.89) 10,467 (10153–10780) 10 6.91 (6.9–6.92) 9523 (9244–9802)6.9 11 6.9 (6.88–6.92) 9806 (9502–10111) 11 6.99 (6.98–7.01) 8036 (7747–8325)



http://cdiac.ornl.gov/oceans/co2rprt.html

http://cdiac.ornl.gov/oceans/co2rprt.html




development, however the results from these response variablesare not presented in this paper.

2.5. Data analyses

By current conventions, pCO2 is the preferred variable used toexamine ocean acidification impacts in these types of experiments(Riebesell et al., 2010). Our analyses assessed three key questions:

(1) Is there a significant relationship between pCO2 and each ofthe key response variables (time to hatching completion, finalday larval survival and growth), over the range of pCO2

explored? If so, what is the functional form (shape) of thatrelationship?

(2) For response variables showing a statistically significant over-all relationship with pCO2, at what level of pCO2 does theresponse first become statistically significant?

(3) What is the level of individual variability in larval responsesand how many tanks and experiments would be needed to be80% sure of detecting an effect of the identified size.

The overall relationship between each response variable andmean pCO2 was evaluated using either generalized linear models(GLM) or generalized additive mixed models (GAMMs), treatingmean pCO2 as a continuous numerical quantity rather than atreatment effect, due to variance in mean pCO2 between tankswithin target treatments. Both linear and non-linear relationshipswere explored. Where GLMs were used, the sample-size adjustedAkaike Information Criterion (AICc; Akaike, 1973; Burnham andAnderson, 2002) was used to select between linear and non-linearmodels for each response variable. Due to differences in ambientseawater conditions in Achotines Bay (i.e. the seawater intake)between trials, data collected from each trial were modeledseparately to explicitly recognize and compare treatment effectswithin each of the two trials. All statistical analyses were carriedout in R version 2.12.1 (R Development Core Team, 2010) using thebase and mgcv (Wood, 2004, 2006) packages.

In addition, for those response variables for which there was asignificant trend with increasing mean pCO2 (see Section 3), aclassical test of all pair-wise comparisons was done using the pHtreatment level (comprising the set of tanks targeted at specificpHs) as the independent factor. Results of the all pair-wisecomparison tests indicates which treatments were most influen-tial with respect to the overall trends. The multiple comparisonstesting was conducted using Tukey contrasts with a 95% family-wise confidence level (Hsu, 1996) and implemented in R with thebase and multcomp (Hothorn et al., 2008) packages.

2.5.1. Egg stage durationEgg-stage duration in hours was determined from the time of

egg transfer until 2 h following the time at which 50% of the eggshad hatched. Based on hatch rate estimates that have beenconducted 2 hours following 50% hatching, very few (o5%) liveunhatched eggs remain in the tanks at this time (Margulies et al.,2007a). To determine the time at 50% hatching, we collected asample of eggs from the incubation tank at 15-min intervals,beginning about 12–15 h (depending on water temperature) afterthe estimated time of spawning. GLMs were used to assess therelationship between mean pCO2 and egg stage duration.

2.5.2. SurvivalEach trial was terminated on the seventh and final day of

growth (six full days of feeding), and each tank was slowly drainedof water. All surviving larvae were removed by beaker and countedand the expected survival was estimated for each tank after

adjusting for previous sample removals (Margulies, 1989; Wexleret al., 2011). A series of nested cubic spline regression models werefitted to expected survival values with mean pCO2 as the inde-pendent variable, and the final model for each trial was selectedusing AICc. Nested models examined were: a constant only model;a linear model; and a non-linear model with 2 degrees of freedom.The final survival count in Tank “15” of trial 2 was excluded from theanalyses because, while larval densities were observed to be veryhigh up until the final night of the trial, an unexplained sudden massmortality event occurred in this tank in the hours prior to finalsurvival counts (contrasting sharply to gradual mortality rates in allother tanks).

2.5.3. GrowthGeneralized Additive Mixed Models (GAMMs) were used to

explore the relationship between pCO2 and size on the seventhand final day of growth, modeling “tank” as a random effect (toaccount for multiple measurements from the same tank), with thetwo trials modeled separately. Two size indicators were analyzed(dry weight and standard length) in separate models. Smoothterms were based on thin plate regression splines and the degreeof smoothing was selected by cross-validation (Wood, 2006).

2.5.4. Power analysisOne of the key objectives for this pilot study was to gain an

understanding of individual variability in larval survival responsesto better design future experimental trials and ensure greaterstatistical power/sensitivity. Firstly, the size of the effect that wewish to be able to detect was identified. Secondly, we simulatedfrom the observed errors to identify how many tanks and experi-ments would be needed to be 80% sure of detecting this size of aneffect. For survival we would need to identify a 20% reduction insurvival between egg transfer and 6 days given a change in pHfrom 8.2 to 7.8. This translates into 300 fewer survivors given anaverage survival of 1380 larvae at pH of 8.2. For the power analysiswe simulated 10,000 trials comparing survival at 8.2 and 7.8, foreach count of tank pairs between 3 and 15. Survival in each tankwas sampled from a normal distribution with standard deviationestimated for the tank effect in the trial, and a difference of 300between the expected values in pH 8.2 vs 7.8. The data wereanalyzed using a generalized linear model with normal errors, andstatistical power was estimated from the proportion of trialsshowing statistical significance at the 5% level. The coefficient ofvariation of the estimated treatment effect was also estimated.

3. Results

3.1. Egg stage duration

In Trial 2, spawning occurred one hour earlier in the night(compared to trial 1) and the final transfer of eggs also occurredone hour earlier. Hatching occurred within 17–18 h in Trial 1 andwithin 19–20 h in Trial 2 from the times that eggs were transferredto the treatment tanks (Fig. 4). AICc based GLM model selectionidentified a significant positive linear relationship between meanpCO2 and hours until complete hatching in both trials (Fig. 4 andTable 2). For trial 1, hatch times at mean pCO2 values less than�3000 μatm were very similar but hatch times at pCO249500 -μatm were increased (relative to those o3000 μatm). Pairwiseanalyses were not conducted due to the lack of intermediatetreatments. For trial 2, pair-wise comparisons (Table 3) indicatethat the mean hatch time at pH 8.06 (pCO2¼533) was significantlyless than that at each of mean pHs 7.69 (pCO2¼1573), 7.25(pCO2¼5251) and 6.99 (pCO2¼9602).






3.2. Survival

Survival by day 7 of feeding was found to decrease withincreasing pCO2 in trial 1 (Fig. 5 and Table 4) with AICc selectinga significant non-linear relationship between mean pCO2 andlarval survival. In contrast to trial 1, we did not detect a significantrelationship between survival and mean pCO2 in trial 2. The pair-wise comparisons of mean survival among treatments in trial 1(Table 3) indicates that mean survival at the control level (pH¼8.23, pCO2¼368) was not significantly different to survival at pH7.56 (pCO2�2108) but was significantly greater than survival at pH7.35 (pCO2�4732) and pH 6.90 (pCO2�8847). No pair-wise com-parisons were conducted for trial 2 survival because the overalltrend was not significant (Table 4).

3.3. Growth

3.3.1. Dry weightsGAMMs applied to larval dry weight data indicated a significant

negative linear relationship between size (dry weight) of larvaeand mean pCO2, in both trials 1 and 2, with the effect of pCO2 onsize (slope of relationship) greater for larvae from trial 1 (Fig. 6 andTable 5). Larvae reared at the highest pCO2 levels (48000 μatm)

were smaller on average than larvae reared at pCO2 close topresent day levels (356–477 μatm), in both trials. Pair-wise com-parisons for both trials (Table 3) indicated a significant differencein dry weights between larvae reared in the control treatment(mean pCO2�368 in Trial 1 and �464 in Trial 2) and the highestpCO2 treatments (mean pCO2�9824 in Trial 1 and �8847 inTrial 2) but not between larvae reared at the control and inter-mediate pCO2 levels. The significant overall trends (from theGAMM) are therefore mainly due to the highest pCO2 treatmenteffects.

3.3.2. Standard lengthGAMMs applied to larval standard length data indicated a

significant negative non-linear relationship between size (standardlength) of larvae and mean pCO2, in trial 1 and a significantnegative linear relationship between size (standard length) oflarvae and mean pCO2 in trial 2 (Fig. 7 and Table 5). Larvae fromhigher pCO2 tanks were on average larger in trial 2 than trial 1.Pair-wise comparisons for trial 2 (Table 3) indicated that thesignificant overall decreasing trend (from GAMM analyses) instandard length was influenced by the difference in mean lengthsbetween the control pCO2�464 and pCO2�8847 (the highest andlowest pCO2 treatments). However, in trial 1, larvae reared in thetreatment tanks with mean pCO2�2108 and mean pCO2�4732were significantly smaller than those reared in the control treat-ment (pCO2�368) (Table 3).

3.4. Power analyses

The power analysis of larval survival from trial 1 determinedthat the standard deviation of the survival data residuals for larva(on day 7 of feeding) was �200. Given a single experiment withtanks at 8.2 and 7.4, an 80% confidence of detecting a size effect of300 between the 2 treatments at the po0.05 level would require10 tanks per trial, which would also give an expected cv of about33%. Simulations using between 3 and 15 tanks per treatmentestimated the power levels provided in Table 6.

4. Discussion

Our study tested the effect of a range of pH (�6.9 to 8.2) andpCO2 (330–10,467 μatm) conditions upon yellowfin tuna eggs andlarvae. This range is broad enough to take into account current andfuture (to the year 2300) spatial variability in water chemistryacross the yellowfin tuna spawning habitat range in the Pacific.Experimental tanks with mean pCO2 of less than 2500 μatmprovided conditions that are relevant to the assessment of “nearfuture” (i.e. year 2100) impacts on yellowfin tuna. Over longer timeframes (e.g. to 2300) or with further increases in CO2 emissions,pCO2 may increase further and pH decrease further (e.g. by0.7 pH units according to Caldeira and Wickett, 2003), in whichcase our experimental tanks with mean pCO2 between 2500 and5000 μatm are relevant. In the current study, it was clear that thestrongest and most consistent impacts across larval yellowfin earlylife history processes occurred at the highest pCO2 levels tested(48500 μatm) and that these levels are outside those predicted tooccur in the next 300 years. However, there was evidence ofsignificantly reduced survival at mean pCO2 levels ofZ4730 μatm(trial 1), significantly reduced larval size at mean pCO2Z2108(trial 1) and prolonged egg hatch time (trial 2) at mean pCO2

levelsZ1573 μatm, that are relevant to near future predictedlevels. Significant effects at near future pCO2 levels were not,however, consistently predicted in both trials and the reasons forthis inconsistency are discussed below. The results will assist inthe future design of controlled experiments that will estimate

0 2000 4000 6000 8000 10000 12000 14000

15

16

17

18

19

20

21

mean pCO2

hatc

h ho

urs

Fig. 4. Predicted linear relationship between mean pCO2 and hours to completehatching (Trial 1, black line) (Trial 2, red line). Dashed lines represent 95%confidence intervals; open circles indicate the data used to fit the models. (Forinterpretation of the references to color in this figure legend, the reader is referredto the web version of this article.)

Table 2Summary model statistics for the egg stage duration analysis assessing therelationship between mean pCO2 and time (h) until complete hatching in each oftwo separate experimental trials.

Model set pCO2 range (μatm) Model term tested Pr(4 |t|) AICc

Trial 1 330–13480 pCO2 (linear) 0.0000 (þ) �3.28pCO2 (cs: df¼2) 0.0121 (þ);

0.0000 (þ)1.24

Constant (�1) 19.6

Trial 2 525–10993 pCO2 (linear) 0.0426 (þ) 7.94Constant (�1) 9.45pCO2 (cs: df¼2) o0.135; o0.190 12.5






interaction effects (discussed below) and the functional forms ofthe relationships. This information should permit populationdynamics models to include acidification effects when forecastingspecies distribution and abundance.

Inconsistent results between trials are likely to result fromdiffering experimental conditions. We observed reduced survivalof yellowfin larvae after 7 days of feeding in the first trial withincreasing mean pCO2. Compared to control (present day pCO2)conditions, significantly lower survival became apparent at pCO2

treatment levels of 4732 μatm or higher. There was no statisticalrelationship between survival and mean pCO2 in the second trial,due in large part to the high level of intra-treatment variability in

Table 3Results of all pair-wise comparisons for each quantity and experiment. The differences in mean levels and adjusted p-values (in parentheses) are provided. Note that themean pH and pCO2 conditions up until hatching in Trial 2 differed to the mean conditions to the end of Trial 2 (when weights and lengths were measured) and so hatch timesand size measures for Trial 2 are reported in separate tables.

Comparison (pH targets) pH Comparison (actual means) pCO2 Comparison (actual means) Survival Standard length Dry weight

Trial 1 – Survival, length and weight8.1–7.7 8.23–7.56 368–2108 363.7(0.320) 0.42(0.004) 19.6(0.367)8.1–7.3 8.23–7.35 368–4732 870.0(0.010) 0.36(0.017) 11.5(0.777)8.1–6.9 8.23–6.90 368–8847 955.7(0.006) 0.68(o0.001) 40.4(0.005)7.7–7.3 7.56–7.35 2108–4732 506.3(0.121) �0.06(0.970) �8.1(0.908)7.7–6.9 7.56–6.90 2108–8847 592.0(0.066) 0.26(0.147) 20.7(0.319)7.3–6.9 7.35–6.90 4732–8847 85.6(0.971) 0.32(0.052) 28.8(0.081)

Comparison (pH targets) pH comparison (actual means) pCO2 comparison (actual means) Hatch time

Trial 2 – Hatch time8.1–7.7 8.06–7.69 533–1573 �0.61(0.001)8.1–7.3 8.06–7.25 533–5251 �0.36 (0.026)8.1–6.9 8.06–6.99 533–9602 �0.69 (o0.001)7.7–7.3 7.69–7.25 1573–5251 0.25 (0.126)7.7–6.9 7.69–6.99 1573–9602 �0.08 (0.830)7.3–6.9 7.25–6.99 5251–9602 �0.33 (0.038)

Comparison (pH targets) pH Comparison (actual means) pCO2 Comparison (actual means) Standard length Dry weight

Trial 2 – Length and weight8.1–7.7 8.10–7.63 464–1719 0.16 (0.536) 4.4 (0.946)8.1–7.3 8.10–7.23 464–4732 0.18 (0.430) 7.1 (0.817)8.1–6.9 8.10–6.95 464–8847 0.49 (o0.001) 25.5 (0.016)7.7–7.3 7.63–7.23 1719–4732 0.02 (0.998) 2.7 (0.989)7.7–6.9 7.63–6.95 1719–8847 0.33 (0.059) 21.1 (0.082)7.3–6.9 7.23 - 6.95 4732–8847 0.31 (0.099) 18.4 (0.173)

0 2000 4000 6000 8000 10000

0

500

1000

1500

2000

2500

mean pCO2

surv

ival

Fig. 5. Predicted relationship between mean pCO2 and larval survival after 7 daysof growth (Trial 1, black line) (Trial 2, red line). Dashed lines represent 95%confidence intervals; points indicate the data used to fit the models. (Forinterpretation of the references to color in this figure legend, the reader is referredto the web version of this article.)

Table 4Summary model statistics for survival analyses by generalized linear model (GLM).

ModelSet

Mean pCO2 range (μatm) Model term tested Pr(4 |t|) AICc

Trial 1 348–10467 pCO2 (cs: df¼2) 0.0011 (þ);0.0258 (þ)

176.8

pCO2 (linear) 0.0055 (þ) 179.2constant (�1) 185.2

Trial 2 460–9563 Constant (�1) 181.8pCO2 (linear) 0.579 185.4pCO2 (cs: df¼2) 0.263;

0.818189.1

0 2000 4000 6000 8000 10000

0

20

40

60

80

100

120

140

Dry

wei

ght (

ug)

Mean pCO2

Fig. 6. Predicted relationship between mean pCO2 and dry weight for Trial 1 (blacklines and points) and Trial 2 (red lines and points). Dashed lines represent 95%confidence intervals; circles indicate the data used to fit the models. (For interpretationof the references to color in this figure legend, the reader is referred to the web versionof this article.)






survival. The two trials were conducted in different months andunder conditions that differed in a number of ways. Firstly, whiledissolved oxygen in trial 1 was relatively stable (�8.2 to 8.5 mg/L),there was a drop in dissolved oxygen from �8.5 to 7.5 mg/L in themiddle of trial 2. While these levels are above those previouslyreported to impact yellowfin growth and survival (Wexler et al.,2011), the possibility of an interaction with pCO2 (or pH) cannot bediscounted. Secondly, during trial 2 a phytoplankton bloom (ofunknown species) occurred near the experimental water intake inAchotines Bay and the impact of this upon the chemistry of thewater sourced from Achotines Bay is unknown. Thirdly, broodstockspawning (eggs produced) dropped sharply in the days just priorto the start of trial 2 and the relative contributions of individualbroodstock to spawning in each trial may have differed, as maybroodstock condition, and affected the mean viability of offspringin each trial. Fourthly, technical problems led to a loss of pHcontrol across tanks in the lowest pH treatment (pH 6.9) towardsthe end of trial 2. Finally, there was a sudden mortality event(crash) in tank 15 (pH 8.1) on the last night of trial 2, associatedwith a very high density of larvae throughout the trial in the sametank, which may have caused an ammonia spike. Such events are

uncommon but have been observed previously in high larvaldensity tanks at Achotines after 4–5 days of feeding. High survivalin that tank, if it had been included in the analysis, would haveprovided a declining survival trend with increasing pCO2 in trial 2.We therefore believe that the results from Trial 1 were obtainedunder more stable conditions and are more likely to representaverage larval responses to increasing pCO2 levels, than the resultsfrom Trial 2. It will be important to undertake further trials to testthis conclusion.

The egg stage and growth results were somewhat more consis-tent between the two trials. Mean larval sizes (dry weights andstandard lengths) after 7 days of feeding declined with increasingmean pCO2 in both trials, with significant effects upon larval lengthsapparent at lower pCO2 levels (i.e. mean pCO2�2108 or above) intrial 1 than in trial 2 (effects significant for pCO2�8847 only). Slowerlarval growth during the first week of feeding may subsequentlyimpact survival by affecting foraging success and by prolonging stagedurations, increasing larval size-dependent susceptibility to preda-tion (review in Leggett and Deblois, 1994).

Egg stage duration increased with increasing pCO2, indicating aone hour delay until hatching at the highest pCO2 levels tested.Similar delays in hatching have also been observed during adversephysical conditions of very low dissolved oxygen levels andextremely high water temperatures (Wexler et al., 2011). Whilethe relationship was not, according to GLM, highly significant intrial 2, pairwise testing indicated that hatching was delayed in alltreatments relative to the control (present day). In Trial 1 it wasnot possible to achieve the desired spread of treatment pCO2s(after 2 days) and it was not possible to test pCO2 levels between3000 and 9000 μatm, meaning the shape of the relationship overthe range of pCO2 tested is poorly defined. Note that target pHlevels (and associated pCO2 levels) were not closely attained in anyof the tanks in the first 24 h of trial 1 (Table A1 in Appendix), andthe range and distribution of mean pCO2 across tanks at thecompletion of egg hatching differed between the two trials. Itshould be noted that it was not possible to determine hatchingrates in the current study due to uneven distribution of eggs in thehatching tanks, but the influence of this on final survival estimatesshould be looked at in the future.

The impact of increased pCO2 (or decreased pH) upon larvalyellowfin, particularly as observed in trial 1, needs to be placed in thecontext of the impact of other natural and anthropogenic factors thatinfluence larval yellowfin growth and survival. Some of the mostpotent biological and physical factors controlling the survival andgrowth of yellowfin larvae during the first two weeks of life includefood abundance and quality (foremost), as well as physical factorssuch as microturbulence (wind induced) and light intensity thatinfluence the distribution and physical characteristics of larval prey(Margulies et al., 2007a). These biological and physical factors canresult in order-of-magnitude (2–5� ) effects on yellowfin larvalsurvival and significant effects on growth in experimental studies(Margulies et al., 2007a). It is notable that the potential effects ofocean acidification on survival and growth of larvae indicated in Trial1 of this study are of roughly the same magnitude.

There are also physical factors that impact yellowfin larvalgrowth and survival that are predicted to change in the future as aresult of climate change and whose impacts may interact withocean acidification. Water temperature and dissolved oxygen

Table 5Summary model statistics for dry weights and standard length analyses. “edf”indicates the estimated degrees of freedom.

Growth measure Trial edf – s (mean pCO2) F p value

Dry weight 1 1 11.65 0.00082 1 13.00 0.0004

Standard length 1 2.15 13.35 o0.000012 1 20.69 0.00001

0 2000 4000 6000 8000 10000

3.0

3.5

4.0

4.5

5.0

Sta

ndar

d le

ngth

(mm

)

Mean pCO2

Fig. 7. Predicted relationship between standard length (mm) and mean pCO2 intrial 1 (black lines and points) and trial 2 (red lines and points) with dashed linesindicating 95% CIs. Open triangles indicate the data used to fit the models. (Forinterpretation of the references to color in this figure legend, the reader is referredto the web version of this article.)

Table 6Power levels based on simulations with between 3 and 15 tanks.

Tanks/treatment 3 4 5 6 7 8 9 10 11 12 13 14 15

Power 0.24 0.36 0.45 0.54 0.63 0.7 0.75 0.8 0.84 0.87 0.9 0.92 0.94Mean CV 0.57 0.51 0.46 0.43 0.4 0.37 0.35 0.33 0.32 0.3 0.29 0.28 0.27






levels are recognized as key factors influencing early life historystages in fish and have been studied in yellowfin tuna. Wexleret al. (2011) determined that the optimal range of temperatures forrapid growth and moderate to high survival of larval yellowfintuna (at 2 days of feeding) was 26–31 1C, with survival signifi-cantly reduced outside this range (despite growth rates continuingto increase above 31 1C). This temperature range encompassessurface temperatures observed across much of the yellowfinspawning habitat in the Pacific Ocean (indeed it should not besurprising that yellowfin might choose to spawn in conditions thatincrease the survival likelihood of their larvae). The same studyobserved that yellowfin larval survival is reduced when dissolvedoxygen levels drop below 2.65 mg O2/L, but these levels aretypically well below the levels found in the upper mixed layerwhere yellowfin larvae reside (Wexler et al., 2011). Events such asupwellings (with lower temperatures and dissolved oxygen) canhowever reduce the vertical habitat of the larvae (Wexler et al.,2011). Thus yellowfin larvae appear well adapted to cope with therange of temperatures and oxygen that typically occur within theirnormal habitat. The current experiments provided no evidencethat survival would vary within the current seasonal range of pHin surface waters off Achotines (�pH 7.9–8.1), noting that effectson survival in trial 1 were only apparent at pH 7.56 (pCO2

4732 μatm), a level that might only be reached after 300 years.However, consideration will need to be given in future yellowfin

tuna experiments to the potential interactive effects of ocean acid-ification with temperature and oxygen, which will also vary underfuture climate change (Gruber, 2011). Ocean acidification and otherparameters such as temperature have already been shown to interactfor some species (Nowicki et al., 2012; Munday et al., 2009a; Enzoret al., 2013). Hypoxic zones in the ocean may increase in the future asa result of climate driven changes in temperatures and deep oceanmixing (Hofman and Shellhuber, 2009). Survival during the onset offeeding in yellowfin tuna larvae is greatly affected by short-termoxygen deficits at water temperatures Z26 1C (Wexler et al., 2011).The mean pCO2 was confounded with both dO2 and variance in pCO2

across tanks. The level of variance in dO2 was small and the lowestobserved level of dO2 was substantially higher than levels required todepress larval yellowfin growth or survival (Wexler et al., 2011).However, larval sensitivity to dO2 might in fact increase underelevated pCO2 conditions and this needs to be tested in futureexperiments.

This study has highlighted the need for further research into thepotential impacts of ocean acidification upon yellowfin tuna andtropical tunas in general. In particular, the current results can be usedto help design more intensive experimental trials. Power analyses ofvariability in survival responses within and between tanks/treatmentsindicated that more replicate tanks per treatment are needed toincrease the statistical power and sensitivity of the experiments, sothat survival relationships can be identified and described. This wasless of an issue for the growth analyses for which sample numberswere an order of magnitude higher. It will also be important to seekfull randomization (across rows and columns of the block design) infuture experiments to ensure greater confidence in the statisticaloutputs, and to include within future experimental designs a lowerpCO2 treatment (�1000 μatm) to test yellowfin larval responses tolower “near future” pCO2 levels. Aside from improving experimentaldesign, research on other fish species has indicated that a number ofadditional key factors will need to be considered when assessing thepotential for ocean acidification to impact yellowfin tuna populations.

Recent research has demonstrated that exposure of broodstock toelevated pCO2 prior to spawning may reduce or remove negativeimpacts of elevated pCO2 upon behavior of larvae derived fromsubsequent spawnings (Miller et al., 2012), a mechanism that shouldalso be assessed in yellowfin. Similarly, investigations into the capacityof yellowfin tuna to adapt (through selection for more resistant

individuals) should also continue. The very high fecundity andrelatively short generation time of yellowfin tuna may permit themto adapt more rapidly than less fecund, longer lived species. It will beimportant to determine the extent to which inter-individual variationmediates different selection responses (Schlegel et al., 2012). Somespecies show significant individual variability in somatic and beha-vioral responses to elevated pCO2 (e.g. Munday et al., 2010; Munday etal., 2012; Ferrari et al., 2011), and in one study resistant individualssuffered lower predation-based mortality rates when released in thefield (Munday et al., 2012) suggesting potential for rapid selection andadaptation and higher robustness of populations to ocean acidification(Frommell et al., 2012). However selection processes may weaken athigher pCO2 levels, if fewer individuals are resistant (Munday et al.,2010; Ferrari et al., 2011). It is also unknown whether resistantindividuals traits are heritable (Munday et al., 2012). Genetic analysesof samples taken from the current trials are in progress to assesswhether some genotypes are more robust to changes in pCO2, andthese results should provide insights on the designs needed to assessthis important question. Results of our supplemental sampling on thefeeding patterns, otolith morphology and histological condition ofinternal tissues of larvae will also permit us to further investigate howacidification affects early life stages. Analyses of samples taken fromyolk-sac and early feeding larvae will allow us to assess at what pointin development any effects on growth first became apparent.

Sub-lethal effects on behavior also warrant attention. Behavioralchanges have been observed in larval reef fish (including responses topredator cues and capacity to find and consume prey – see Section 1)due to the altered functioning of a key brain neurotransmitter receptorGamma-Amino Butyric Acid (GABA)-A, that results from exposure toelevated CO2 (Nilsson et al., 2012). GABA-A is highly conserved acrossmarine fish species, raising the possibility that behavioral changescould be common across many fish species. Noting that suchbehavioral effects have already been demonstrated for one reef speciesto lead to increased predation in the field (Munday et al., 2010), weconsider that behavioral impacts of elevated CO2 may have significantimplications for the recruitment of affected species. Similarly, investi-gating the impacts of increasing pCO2 on sperm motility and fertiliza-tion success is warranted (e.g. Havenhand et al., 2008; Frommelet al., 2010).

Empirical research such as presented here can allow modelssuch as SEAPODYM (Lehodey et al., 2008) to be parameterised toinclude acidification effects and subsequently enable scientists andtuna fishery managers to better understand how these changes inocean chemistry will alter the distribution and abundance ofyellowfin tuna (Hobday et al., 2013). Environmental data are usedin SEAPODYM to functionally characterize the habitat of thepopulation depending on its thermal, biogeochemical, and foragepreferences (Lehodey et al., 2008, 2010). To this end, it will becritical that further empirical trials are conducted to more clearlyidentify the functional form of the relationship between pH (orpCO2) and larval survival, and in particular, identify any interac-tions between pH (or pCO2) and other key physical oceanographicfactors such as temperature and oxygen, so as to provide relevantinformation for appropriately altering the spawning-habitat indexand adjust local mortality rates in SEAPODYM. Subsequent popu-lation level predictions of ocean acidification impacts will enhancethe capacity of regional fisheries management organizations tomake more-informed decisions regarding the management of thehighly valuable tropical tuna resources, particularly with regard toattaining key sustainability-related objectives (Bell et al., 2013).

Acknowledgments

We would like to thank the laboratory staff of the IATTCAchotines Laboratory (L. Tejada, S. Cusatti, D. Ballesteros, D. Solis,






C. Vergara, L. Castillo, D. Dominguez, D. Ramires, A. Ortega, D.Perez, H. Espinoza, Y. Ballesteros, W. Reluz, D. Mancilla) for directsupport during the experimental trials as well as assembly andmaintenance of the experimental system, three anonymous reviewersfor their insightful comments and suggestions that helped improvethis manuscript, P. Munday from James Cook University for commentson the introduction and the IATTC’s Director, G. Compeán, and ChiefScientist, R. Deriso, for their support. The research was funded underCooperative Agreement NA09OAR4320075, Project 661553 from theNational Oceanic and Atmospheric Administration with the JointInstitute for Marine and Atmospheric Research (JIMAR), University ofHawaii (UH).

Appendix A. Supporting information

Supplementary data associated with this article can be found inthe online version at: http://dx.doi.org/10.1016/j.dsr2.2014.03.019.

References

Akaike, H., 1973. Information theory and an extension of the maximum likelihoodprinciple. In: Second International Symposium on Information Theory, vol. 1.Springer Verlag, Akademia Kiado, Budapest, pp. 267–281.

Allen, B., Domenici, P., McCormick, M.I., Watson, S., Munday, P., 2013. Elevated CO2

affects predator–prey interactions through altered performance. PLoS One 8 (3),e58520, http://dx.doi.org/10.1371/journal.pone.0058520.

Alsterberg, C., Eklof, J.S., Gamfeldt, L., Havenhand, J.N., Sundback, K., 2013.Consumers mediate the effects of experimental ocean acidification and warm-ing on primary producers. PNAS 110, 8603–8608, http://dx.doi.org/10.1073/pnas.1303797110.

Barnett, T.P., Pierce, D.W., Achuta Rao, K.M., Gleckler, P.J., Santer, B.D., Gregory, J.M.,Washington, W.M., 2005. Penetration of human-induced warming into theworld‘s oceans. Science 309, 284–287.

Bates, N.R., Best, M.H. P., Neely, K., Garley, R., Dickson, A.G., Johnson, R.J., 2012.Detecting anthropogenic carbon dioxide uptake and ocean acidification in theNorth Atlantic ocean. Biogeosci. Discuss. 9, 989–1019.

Baumann, H., Talmage, S.C., Gobler, C.J., 2011. Reduced early life growth andsurvival in a fish in direct response to increased carbon dioxide. Nat. Clim.Change 2, 38–41, http://dx.doi.org/10.1038/NCLIMATE1291.

Bell, J.D., Ganachaud, A., Gehrke, P.C., Griffiths, S.P., Hobday, A.J., Hoegh-Guldberg, O.,Johnson, J.E., Le Borgne, R., Lehodey, P., Lough, J.M., Matear, R.J., Pickering, T.D.,Pratchett, M.S., Sen Gupta, A., Senina, I., Waycott, M., 2013. Tropical Pacific fisheriesand aquaculture will respond differently to climate change. Nat. Clim. Change 3,591–599.

Bignami, S., Sponaugle, S., Cowen, R.K., 2013a. Response to ocean acidification inlarvae of a large tropical marine fish, Rachycentron canadum. Global ChangeBiol. 19, 996–1006.

Bignami, S., Enochs, I.C., Manzello, D.P., Sponaugle, S., Cowen, R.C., 2013b. Oceanacidification alters the otoliths of a pantropical fish species with implicationsfor sensory function. PNAS.

Blaxter, J.H.S., 1992. The effect of temperature on larval fishes. Neth. J. Zool. 42,336–357.

Boehlert, G.W., Mundy, B.C., 1994. Vertical and onshore–offshore distributionalpatterns of tuna larvae in relation to physical habitat features. Mar. Ecol. Prog.Ser. 107, 1–13.

Bopp, L., Resplandy, L., Orr, J.C., Doney, S.C., Dunne, J.P., Gehlen, M., Halloran, P.,Heinze, C., Ilyina, T., Seferian, R., Tjiputra, J., Vichi, M., 2013. Multiple stressors ofocean ecosystems in the 21st century: projections with CMIP5 models.Biogeosciences 10, 6225–6245, http://dx.doi.org/10.5194/bg-10-6225-2013(2013).

Brauner, C.J., Baker, D.W., 2009. Patterns of acid–base regulation during exposure tohypercarbia in fishes. In: Glass, M.L., Woods, S.C. (Eds.), Cardio-RespiratoryControl in Vertebrates. Springer-Verlag, Berlin, pp. 43–63.

Breitburg, D.L., Rose, K.A., Cowan Jr., J.H., 1999. Linking water quality to larvalsurvival: predation mortality of fish larvae in an oxygen-stratified watercolumn. Mar. Ecol. Prog. Ser. 178, 39–54.

Briffa, M., de la Haye, K., Munday, P.L., 2012. High CO2 and marine animalbehaviour: potential mechanisms and ecological consequences. Mar. Pollut.Bull. 64, 1519–1528.

Burnham, K.P., Anderson, D.R., 2002. Model Selection and Multimodel Inference:A Practical Information-theoretic Approach. Springer Verlag, New York.

Caldeira, K., Wickett, M.E., 2003. Anthropogenic carbon and ocean pH. Nature 425,365.

Caldeira, K., Wickett, M.E., 2005. Ocean model predictions of chemistry changesfrom carbon dioxide emissions to the atmosphere and ocean. J. Geophys. Res.110, C09S04, http://dx.doi.org/10.1029/2004JC002671.

Cripps, I.C., Munday, P.L., McCormick, M.I., 2011. Ocean acidification affects preydetection by a predatory reef fish. PLoS One 7, e22736, http://dx.doi.org/10.1371/journal.pone.0022736.

Checkley, D.M., Dickson, A.G., Takahashi, M., Radich, J.A., Eisenkolf, N., Asch, R.,2009. Elevated CO2 enhances otolith growth in young fish. Science 324, 1683.

Davis, T.L.O., Jenkins, G.P., Young, J.W., 1990. Patterns of horizontal distribution ofthe larvae of southern bluefin (Thunnus maccoyii) and other tuna in the IndianOcean. J. Plank. Res. 12, 1295–1314.

Devine, B.M., Munday, P.L., Jones, G.P., 2012. Rising CO2 concentrations affectsettlement behaviour of larval damselfishes. Coral Reefs 31 (1), 229–238.

Dixson, D.L., Munday, P.L., Jones, G.P., 2010. Ocean acidification disrupts the innateability of fish to detect predator olfactory cues. Ecol. Lett. 13, 68–75.

Domenici, P., Allan, B., McCormick, M.I., Munday, P.L., 2012. Elevated CO2 affectsbehavioral lateralization in a coral reef fish. Biol. Lett. 8, 78–81, http://dx.doi.org/10.1098/rsbl.2011.0591.

Doney, S.C., Balch, W.M., Fabry, V.J., Feely, R.A., 2009. Ocean acidification: a criticalemerging problem for the ocean sciences. Oceanograhy 22 (4), 16–25.

Esbaugh, A.J., Heuer, R., Grosell, M., 2012. Impacts of ocean acidification onrespiratory gas exchange and acid–base balance in a marine teleost, Opsanusbeta. J. Comp. Physiol. B 182, 921–934.

Enzor, L., Zippay, M.L., Place, S.P., 2013. High latitude Fish in a high CO2 world:synergistic effects of elevated temperature and carbón dioxide on the metabolicrates of Antarctic notothenioids. Comp. Biochem. Physiol. Part A 164, 154–161.

Fabry, V.J., Seibel, B.A., Feely, R.A., Orr, J.C., 2008. Impacts of ocean acidification onmarine fauna and ecosystem processes—ICES. J. Mar. Sci. 65, 414–432.

Feely, R.A., Sabine, C.L., Lee, K., Berelson, W., Kleypas, J., Fabry, V.J., Millero, F.J., 2004.Impact of anthropogenic CO2 on the CaCO3 system in the oceans. Science 305,362–366.

Ferrari, M.C.O., Dixson, D.L., Munday, P.L., McCormick, M.I., Meekan, M.G., Sih, A.,Chivers, D.P., 2011. Intrageneric variation in antipredator responses of coral reeffishes affected by ocean acidification: implications for climate change projec-tions on marine communities. Global Change Biol. 17, 2980–2986.

Ferrari, M.C.O., McCormick, M.I., Munday, P.L., Meekan, M.G., Dixson, D.L., Lonn-stedt, O., Chivers, D.P., 2012a. Effects of ocean acidification on visual riskassessment in coral reef fishes. Funct. Ecol. 26, 553–558, http://dx.doi.org/10.1111/j.1365-2435.2011.01951.x.

Ferrari, M.C.O., Manassa, R., Dixson, D.L., Munday, P.L., McCormick, M.I., Meekan, M.G.,Sih, A., Chivers, D.P., 2012b. Effects of ocean acidification on learning in coral reeffishes. PLoS One 7 (2), e31478, http://dx.doi.org/10.1371/journal.pone.0031478.

Franke, A., Clemmesen, C., 2011. Effect of ocean acidification on early life stages ofAtlantic herring (Clupea harengus L.). Biogeosci. Discuss. 8, 7097–7126.

Frommel, A.Y., Stiebens, V., Clemmesen, C., Havenhand, J., 2010. Effect of oceanacidification on marine fish sperm (Baltic cod: Gadus morhua). Biogeosciences7, 3915–3919.

Frommel, A.Y., Maneja, R., Lowe, D., Malzahn, A.M., Geffen, A.J., Folkvord, A.,Piatkowski, U., Reusch, T.B. H., Clemmesen, C., 2011. Severe tissue damage inAtlantic cod larvae under increasing ocean acidification. Nat. Clim. Change 2,42–46, http://dx.doi.org/10.1038/NCLIMATE1324.

Frommel, A.Y., Schubert, A., Piatkowski, U., Clemmesen, C., 2012. Egg and earlylarval stages of Baltic cod, Gadus morhua, are robust to high levels of oceanacidification. Mar. Biol. 160, 1825–1834, http://dx.doi.org/10.1007/s00227-011-1876-3.

Gattuso, J-P., Gao, K., Lee, K., Rost, B., Schulz, K.G., 2010. Approaches and tools tomanipulate the carbonate chemistry. In: Riebesell, U., Fabry, V.J., Hansson, L.,Gattuso, J.-P. (Eds.), Guide to Best Practices for Ocean Acidifi Cation Researchand Data Reporting. Publications Office of the European Union, Luxembourg,pp. 41–52 (2010).

Gruber, N, 2011. Warming up, turning sour, losing breath: ocean biogeochemistryunder global change. Philos. Trans. R. Soc. a—Math. Phys. Eng. Sci. 369,1980–1996, http://dx.doi.org/10.1098/rsta.2011.0003.

Havenhand, J.N., Buttler, F.R., Thorndyke, M.C., Williamson, J.E., 2008. Near-futurelevels of ocean acidification reduce fertilization success in a sea urchin. Curr.Biol. 18 (15), R651–R652.

Hobday, A.J., Young, J.W., Abe, O., Costa, D.P., Cowen, R.K., Evans, K., Gasalla, M.A.,Kloser, R., Maury, O., Weng, K.C., 2013. Climate impacts and oceanic toppredators: moving from impacts to adaptation in oceanic systems. Rev. FishBiol. Fish. 23 (4), 537–546, http://dx.doi.org/10.1007/s11160-013-9311-0.

Hofman, M., Shellhuber, H., 2009. Oceanic acidification affects marine carbon pumpand triggers extended marine oxygen holes. PNAS 106 (9), 3017–3022.

Hothorn, T., Bretz, F., Westfall, P, 2008. Simultaneous inference in generalparametric models. Biom. J. 50 (3), 346–363.

Hurst, T.P., Fernandez, E.R., Mathis, J.T., Miller, J.A., Stinson, C.M., Ahgeak, E.F., 2012.Resiliency of junvenile walleye pollock to projected levels of ocean acidification.Aquat. Biol. 17, 247–259.

Hurst, T.P., Fernandez, E.R., Mathis, J.T., 2013. Effects of ocean acidification on hatchsize and larval growth of walleye pollock (Theragra chalcogramma). ICES J. Mar.Sci. 70 (4), 812–822, http://dx.doi.org/10.1093/icesjms/fst053.

Houde, E., 1989. Subtleties and episodes in the early life of fishes. J. Fish Biol. 35,29–38.

Hsu, J.C., 1996. Multiple Comparisons: Theory and Methods. Chapman and Hall, CRCPress, Florida, USA.

IPCC, 2013. Climate Change 2013: the physical science basis. Contribution of WorkingGroup I to the Fifth Assessment Report of the Intergovernmental Panel on ClimateChange [Stocker, T.F., Qin, D., Plattner, G.-K., Tignor, M., Allen, S.K., Boschung, J.,Nauels, A., Xia, Y., Bex, V., Midgley, P.M. (Eds.)]. Cambridge University Press,Cambridge, United Kingdom; New York, NY, USA, 1535 pp.




http://refhub.elsevier.com/S0967-0645(14)00102-7/sbref1



http://dx.doi.org/10.1371/journal.pone.0058520



http://dx.doi.org/10.1073/pnas.1303797110










http://dx.doi.org/10.1038/NCLIMATE1291



















http://dx.doi.org/10.5194/bg-10-6225-2013

http://dx.doi.org/10.5194/bg-10-6225-2013

http://dx.doi.org/10.5194/bg-10-6225-2013















http://dx.doi.org/10.1029/2004JC002671

http://dx.doi.org/10.1029/2004JC002671

http://dx.doi.org/10.1029/2004JC002671
















http://dx.doi.org/10.1098/rsbl.2011.0591
























http://dx.doi.org/10.1111/j.1365-2435.2011.01951.x

http://dx.doi.org/10.1111/j.1365-2435.2011.01951.x

http://dx.doi.org/10.1111/j.1365-2435.2011.01951.x

http://dx.doi.org/10.1111/j.1365-2435.2011.01951.x












http://dx.doi.org/10.1007/s00227-011-1876-3

http://dx.doi.org/10.1007/s00227-011-1876-3

http://dx.doi.org/10.1007/s00227-011-1876-3

http://dx.doi.org/10.1007/s00227-011-1876-3






http://dx.doi.org/10.1098/rsta.2011.0003






http://dx.doi.org/10.1007/s11160-013-9311-0

http://dx.doi.org/10.1007/s11160-013-9311-0

http://dx.doi.org/10.1007/s11160-013-9311-0








http://dx.doi.org/10.1093/icesjms/fst053










Ishimatsu, A., Kikkawa, T., Hayashi, M., Lee, K-S., Kita, J., 2004. Effects of CO2 onmarine fish: larvae and adults. J. Oceanogr. 60, 731–741.

Ilyina, T., Zeebe, R.E., Maier-Reimer, E., Heinze, C., 2009. Early detection of oceanacidification effects on marine calcification. Global Biogeochem. Cycles 23,GB1008, http://dx.doi.org/10.1029/2008GB003278.

Ilyina, T., Six, K.D., Segschneider, J., Maier-Reimer, E., Li, H., Núñez-Riboni, I., 2013.The global ocean biogeochemistry model HAMOCC: model architecture andperformance as component of the MPI-Earth System Model in different CMIP5experimental realizations. J. Adv. Model. Earth Syst., 2013, http://dx.doi.org/10.1002/jame.20017.

IPCC. 2007. Climate Change 2007: synthesis report. Contribution of Working GroupsI, II, and III to the Fourth Assessment Report of the Intergovernmental Panel onClimate Change [Core Writing Team: Pachauri, R.K., Reisinger, A. (Eds.)]. IPCC,Geneva, Switzerland..

IPCC, 2011: Workshop Report of the Intergovernmental Panel on Climate ChangeWorkshop on Impacts of Ocean Acidification on Marine Biology and Ecosystems[Field, C.B., Barros, V., Stocker, T.F., Qin, D., Mach, K.J., Plattner, G.-K., Mastran-drea, M.D., Tignor, M., Ebi, K.L. (Eds.)]. IPCC Working Group II Technical SupportUnit. Carnegie Institution, Stanford, CA, USA, 164 pp.

Jonz, M.G., Nurse, C.A., 2006. Ontogenesis of oxygen chemoreception in aquaticvertebrates. Respir. Physiol. Neurobiol. 154, 139–152.

Kikkawa, T., Ishimatsu, A., Kita, J., 2003. Acute CO2 tolerance during the earlydevelopmental stages of four marine teleosts. Environ. Toxicol. 18, 375–382.

Lang, K.L., Grimes, C.B., Shaw, R.F., 1994. Variations in the age and growth of larvalyellowfin tuna, Thunnus albacares, collected about the Mississippi River plume.Environ. Biol. Fish. 39, 259–270.

Leggett, W.C., Deblois, E., 1994. Recruitment in marine fishes – is it regulated bystarvation and predation in the egg and larval stages. Neth. J. Sea Res. 32,119–134.

Lehodey, P., Senina, I., Murtugudde, R., 2008. A Spatial Ecosystem and PopulationsDynamics Model (SEAPODYM) modeling of tuna and tuna-like populations.Prog. Oceanogr. 78, 304–318.

Lehodey, P., Senina, I., Sibert, J., Bopp, L., Calmettes, B., Hampton, J., Murtugudde, R.,2010. Preliminary forecasts of population trends for Pacific Bigeye Tuna underthe A2 IPCC scenario. Prog. Oceanogr. 86, 302–315.

Lehodey, P., Senina, I., Calmettes, B, Hampton, J, Nicol, S., 2013. Modelling theimpact of climate change on Pacific skipjack tuna population and fisheries.Clim. Change 119 (1), 95–109, http://dx.doi.org/10.1007/s10584-012-0595-1.

Lewis, E., Wallace, D.W. R., 1998. Program developed for CO2 system calculations.ORNL/CDIAC-105. Carbon Dioxide Information Analysis Center, Oak RidgeNational Laboratory, U.S. Department of Energy, Oak Ridge, TN.

Lauth, R.R., Olson, R.J., 1996. Distribution and abundance of larval scombridae inrelation to the physical environment in the northwestern Panama Bight.Interam. Trop. Tuna Commun. Bull. 21, 125–167.

Margulies, D., 1989. Effects of food concentration and temperature on development,growth and survival of white perch, Morone americana, eggs and larvae. Fish.Bull. 87, 63–72.

Margulies, D., Scholey, V.P., Wexler, J.B., Olson, R.J., Suter, J.M., Hunt, S.L., 2007a.A Review of IATTC Research on the Early Life History and Reproductive Biologyof Scombrids Conducted at the Achotines Laboratory from 1985 to 2005. IATTCSpecial Report 16.

Margulies, D., Suter, J.M., Hunt, S.L., Olson, R.J., Scholey, V.P., Wexler, J.B.,Nakazawa, A., 2007b. Spawning and early development of captive yellowfintuna (Thunnus albacares). Fish. Bull. 105, 249–265.

Miller, G.M., Watson, S.A., Donelson, J.M., McCormick, M.I., Munday, P.L., 2012.Parental environment mediates impacts of increased carbondioxide on a coralreef fish. Nat. Clim. Change 2, 858–861, http://dx.doi.org/10.1038/NCLIMATE1599.

Munday, P.L., Crawley, N., Nilsson, G.E., 2009a. Interacting effects of elevatedtemperature and ocean acidification on the aerobic performance of coral reeffishes. Mar. Ecol. Prog. Ser. 388, 235–242.

Munday, P.L., Donelson, J.M., Dixson, D.L., Endo, G.G.K., 2009b. Effects of oceanacidification on the early life history of a tropical marine fish. Proc. R. Soc. Lond.B 276, 3275–3283.

Munday, P.L., Dixson, D.L., Donelson, J.M., Jones, G.P., Pratchett, M.S., Devitsina, G.V.,Døving, K.B., 2009c. Ocean acidification impairs olfactory discrimination andhoming ability of a marine fish. PNAS 106, 1848–1852.

Munday, P.L., Dixson, D.L., Mccormick, M.I., Meekan, M., Ferrari, M.C.O., Chivers, D.P.,2010. Replenishment of fish populations is threatened by ocean acidification. PNAS107, 12930–12934, http://dx.doi.org/10.1073/pnas.1004519107.

Munday, P.L., Gagliano, M., Donelson, J.M., Dixson, D.L., Thorrold, S., 2011a. Oceanacidification does not affect the early life history development of a tropicalmarine fish. Mar. Ecol. Prog. Ser. 423, 211–221, http://dx.doi.org/10.3354/meps08990.

Munday, P.L., Hernaman, V., Dixson, D.L., Thorrold, S.R., 2011b. Effect of oceanacidification on otolith development in larvae of a tropical marine fish.Biogeosciences 8, 1631–1641, http://dx.doi.org/10.5194/bg-8-1631-2011.

Munday, P.L., Mccormick, M.I., Meekan, M., Dixson, D.L., Watson, S., Chivers, D.P.,Ferrari, M.C.O., 2012. Selective mortality associated with variation in CO2

tolerance in a marine fish. Ocean Acidif. 1, 1–6, http://dx.doi.org/10.2478/oac-2012-0001.

Nilsson, G., Dixson, D.L., Domenici, P., McCormick, M.I., Sorensen, C. Watson, S., andMunday, P.L. 2012. Near-future carbondioxide levels alter fish behaviour byinterfering with neurotransmitter function. Nat. Clim. Change 2, 201–204,10.1038/NCLIMATE1352.

Nowicki, J.P., Miller, G.M., Munday, P.L., 2012. Interactive effects of elevatedtemperature and CO2 on foraging behavior of juvenile coral reef fish. J. Exp.Mar. Biol. Ecol. 412, 46–51, http://dx.doi.org/10.1016/j.jembe.2011.10.020.

Owen, R. 1997. Oceanographic atlas of habitats of larval tunas in the Pacific Oceanoff the Azuero Peninsula, Panama, 32 p. Inter-American Tropical Tuna Commis-sion Data Report 9. Inter-American Tropical Tuna Commission, 8604 La JollaShores Drive, La Jolla, CA 92037.

Pelster, B., 2008. Gas exchange. In: Finn, R.N., Kapoor, B.G. (Eds.), Fish LarvalPhysiology, Part 2, Respiration & Homeostasis. Science Publishers, Einfield, NH,pp. 102–103.

Raven, J., Caldeira, K., Elderfield, H., Hoegh-Guldberg, O., Liss, P., Riebesell, U.,Shepherd, J., Turley, C., Watson, A., 2005. “Ocean acidification due to increasingatmospheric carbon dioxide.” Policy Document 12/05. Royal Society London,London.

R Development Core Team, 2010. R: A Language and Environment for StatisticalComputing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0. URL ⟨http://www.R-project.org/⟩.

Riebesell, U., Fabry, V.J., Hansson, L., Gattuso, J.-P. (Eds.), 2010. Guide to BestPractices for Ocean Acidification Research and Data Reporting. PublicationsOffice of the European Union, Luxembourg.

Sabine, C.L., Feely, R.A., Gruber, N., Key, R.M., Lee, K., Bullister, J.L., et al., 2004.The oceanic sink for CO2. Science 305, 367–371.

Sabatés, A., Olivar, M.P., Salat, J., Palomera, I., Alemany, F., 2007. Physical andbiological processes controlling the distribution of fish larvae in the NWMediterranean. Prog. Oceanogr. 74, 355–376.

Schaefer, K.M., 2001. Reproductive biology of tunas. In: Block, B.A., Stevens, E.D.(Eds.), Fish Physiology,vol. 19, Tuna: Physiology, Ecology, and Evolution. Aca-demic Press, San Diego, CA, pp. 225–270.

Schaefer, K.M., 2009. Stock structure of bigeye, yellowfin, and skipjack tunas in theeastern Pacific Ocean. In: Status of the Tuna and Billfish Stocks in 2007. StockAssessment Report 9, pp. 201–218.

Schlegel, P., Havenhand, J.N., Gillings, M.R., Williamson, J.E., 2012. Individualvariability in reproductive success determines winners and losers under oceanacidification: a case study with sea urchins. PLoS One 7, e53118.

Simpson, S.D., Munday, P.L., Wittenrich, M.L., Manassa, R., Dixson, D.L., Gagliano, M.,Yan, H.Y., 2011. Ocean acidification erodes crucial auditory behaviour in amarine fish. Biol. Lett. 7, 917–920, http://dx.doi.org/10.1098/rsbl.2011.0293.

Sundin, J., Rosenqvist, G., Berglund, A., 2013. Altered oceanic pH impairs matingpropensity in a pipefish. Ethology 119, 86–93.

Tseng, Y-C., Yu, M.H., Stumpp, M., Yin, L-Y., Melzner, F., Hwang, P-P., 2013. CO2-driven seawater acidification differentially affects development and molecularplasticity along life history of fish (Oryzias latipes). Comp. Biochem. Physiol. PartA 165, 119–130.

Van Vuuren, D.P., Edmonds, J., Kainuma, M., Riahi, K., Thomson, A., Hibbard, K.,Hurtt, T., Kram, T., Krey, V., Lamarque, J-F., Masui, T., Meinhausen, M.,Nakicenovic, N., Smith, S.J., Rose, S.K., 2011. The representative concentrationpathways: an overview. Clim. Change 2011 (109), 5–31, http://dx.doi.org/10.1007/s10584-011-0148-z.

Wexler, J.B., Margulies, D., Scholey, V.P., 2011. Temperature and dissolved oxygenrequirements for survival of yellowfin tuna, Thunnus albacares, larvae. J. Exp.Mar. Biol. Ecol. 404, 63–72.

Wexler, J.B., Chow, S., Wakabayashi, T., Nohara, K., Margulies, D., 2007. Temporalvariation in growth of yellowfin tuna (Thunnus albacares) larvae in the PanamaBight, 1990–97. Fish. Bull. U.S. 105, 1–18.

Wexler, J.B., Scholey, V.P., Olson, R.J., Margulies, D., Nakazawa, A., Suter, J.M., 2003.Tank culture of yellowfin tuna, Thunnus albacares: developing a spawningpopulation for research purposes. Aquaculture 220, 327–353.

Wittman, A.C., Portner, H-O., 2013. Sensitivities of extant animal taxa toocean acidification. Nat. Clim. Change 3, 995–1001, http://dx.doi.org/10.1038/nclimate1982.

Williams, P., Terawasi, P., 2009. Overview of Tuna Fisheries in the Western andCentral Pacific Ocean, Including Economic Conditions—2008. WCPFC-SC5-2009-GN-WP-01. Western and Central Pacific Fisheries Commission ScientificCommittee Fifth Regular Session, 10–21 August 2009, Port Vila, Vanuatu.

Wood, S.N., 2004. Stable and efficient multiple smoothing parameter estimation forgeneralized additive models. J. Am. Stat. Assoc. 99, 673–686.

Wood, S.N. (2006) Generalized Additive Models: An Introduction with R. Chapmanand Hall/CRC, Florida, USA.






http://dx.doi.org/10.1029/2008GB003278

http://dx.doi.org/10.1029/2008GB003278

http://dx.doi.org/10.1029/2008GB003278

http://dx.doi.org/10.1002/jame.20017





















http://dx.doi.org/10.1007/s10584-012-0595-1

http://dx.doi.org/10.1007/s10584-012-0595-1

http://dx.doi.org/10.1007/s10584-012-0595-1














dx.doi.org/10.1038/NCLIMATE1599
















http://dx.doi.org/10.3354/meps08990




http://dx.doi.org/10.5194/bg-8-1631-2011

http://dx.doi.org/10.5194/bg-8-1631-2011

http://dx.doi.org/10.5194/bg-8-1631-2011

http://dx.doi.org/10.2478/oac-2012-0001




http://dx.doi.org/10.1016/j.jembe.2011.10.020











http://www.R-project.org/


























http://dx.doi.org/10.1007/s10584-011-0148-z

http://dx.doi.org/10.1007/s10584-011-0148-z

http://dx.doi.org/10.1007/s10584-011-0148-z

http://dx.doi.org/10.1007/s10584-011-0148-z










http://dx.doi.org/10.1038/nclimate1982









deep-sea research ii - max planck society · d department of biological and environmental sciences,...

Documents