The metabolic response of thecosome pteropods from the North 1 Atlantic and North Pacific Oceans to high CO 2 and low O 2 2 3

As anthropogenic activities, notably the burning of fossil fuels, increase carbon 32 dioxide (CO2) and result in a decrease in oxygen (O2) concentrations in the ocean system, it 33 becomes important to understand how different populations of marine animals will respond. 34 Water that is naturally lower in pH, with a high concentration of carbon dioxide (hypercapnia) 35 and a low concentration of oxygen, occurs at shallow depths (200-500 m) in the North Pacific 36 Ocean, whereas similar conditions are absent throughout the upper water column in the North 37 Atlantic. This contrasting hydrography provides a natural experiment to explore whether 38 differences in environment cause populations of cosmopolitan pelagic calcifiers, specifically the 39 aragonitic-shelled pteropods, to have a different physiological response when exposed to 40 hypercapnia and low O2. Using closed-chamber end-point respiration experiments, eight species 41 of pteropods from the two ocean basins were exposed to high CO2 (~800 μatm) while six species 42 were also exposed to moderately low O2 (10%, or ~130 μmol kg-1) and a combined treatment of 43 low O2/high CO2. None of the species tested showed a change in metabolic rate in response to 44 high CO2 alone. Of those species tested for an effect of O2, only Limacina retroversa from the 45 Atlantic showed a response to the combined treatment, resulting in a reduction in metabolic rate. 46 Our results suggest that pteropods have mechanisms for coping with short-term CO2 exposure 47 and suggest that there can be interactive effects between stressors on the physiology of these 48 open ocean organisms that correlate with natural exposure to low O2 and high CO2; these are 49 considerations that should be taken into account in projections of organismal sensitivity to future 50 ocean conditions. 51 52 53

Abstract.As anthropogenic activities directly and indirectly increase carbon dioxide (CO 2 ) and decrease oxygen (O 2 ) concentrations in the ocean system, it becomes important to understand how different populations of marine animals will respond.Water that is naturally low in pH, with a high concentration of carbon dioxide (hypercapnia) and a low concentration of oxygen, occurs at shallow depths (200-500 m) in the North Pacific Ocean, whereas similar conditions are absent throughout the upper water column in the North Atlantic.This contrasting hydrography provides a natural experiment to explore whether differences in environment cause populations of cosmopolitan pelagic calcifiers, specifically the aragonitic-shelled pteropods, to have a different physiological response when exposed to hypercapnia and low O 2 .Using closed-chamber end-point respiration experiments, eight species of pteropods from the two ocean basins were exposed to high CO 2 (∼ 800 µatm) while six species were also exposed to moderately low O 2 (48 % saturated, or ∼ 130 µmol kg −1 ) and a combined treatment of low O 2 /high CO 2 .None of the species tested showed a change in metabolic rate in response to high CO 2 alone.Of those species tested for an effect of O 2 , only Limacina retroversa from the Atlantic showed a response to the combined treatment, resulting in a reduction in metabolic rate.Our results suggest that pteropods have mechanisms for coping with short-term CO 2 exposure and that there can be interactive effects between stressors on the physiology of these open ocean organisms that correlate with natural exposure to low O 2 and high CO 2 .These are considerations that should be taken into account in projections of organismal sensitivity to future ocean conditions.

Introduction
Ocean acidification, a result of the dissolution of anthropogenically produced carbon dioxide (CO 2 ) into seawater, is increasingly considered to be one of the most pervasive human changes to the marine system (Halpern et al., 2008;Doney et al., 2009;Gruber, 2011).The pH of the ocean surface has already dropped by ∼ 0.1 units relative to preindustrial levels and is predicted to drop another 0.3-0.4pH units in the next one hundred years (Haugan and Drange, 1996;Bopp et al., 2013;IPCC, 2013).As CO 2 dissolves in the ocean, it causes changes in seawater carbonate chemistry, notably increasing hydrogen ion concentration and decreasing the concentration of carbonate ions.As a consequence of the changing equilibria, there is a reduction in pH and the saturation state of calcium carbonate (CaCO 3 ), including the biogenic forms of calcite and aragonite.In some regions, as ocean acidification continues, the water becomes undersaturated and corrosive, meaning that in the absence of compensative biological action, conditions will favor the dissolution of the CaCO 3 found in the shells and skeletons of calcifying organisms, with aragonite being more sensitive than calcite (Millero, 2007).
Therefore, ocean acidification impacts calcifying species on multiple fronts.Changes in environmental pH can modify the acid-base balance of intra-and extracellular fluids of marine organisms, which may result in reduced fitness or outright mortality (Seibel and Walsh, 2001;Seibel and Fabry, 2003;Widdicombe and Spicer, 2008).Perturbations of seawater carbonate chemistry can also affect the ability of some calcifying animals to create and maintain calcium carbon-Published by Copernicus Publications on behalf of the European Geosciences Union.
ate structures with implications for energetics, survival, competition, and biogeochemical export (Riebesell et al., 2000;Fabry et al., 2008;Ries et al., 2009).Understanding the longterm effects of this increase in ocean acidity on both organisms and ecosystems has, therefore, become of great concern.Important and outstanding research goals are to understand how changing CO 2 impacts current populations and to predict whether these populations will be able to adapt to the rate and severity of the rising anthropogenic CO 2 inputs (e.g., Sunday et al., 2011;Dam, 2013;Kelly and Hofmann, 2013).
One approach to understanding the response of marine animals to acidification is to examine places where animals already experience conditions of elevated CO 2 (hypercapnia).By comparing individuals that inhabit regions of high CO 2 with those that never experience high levels naturally, insight can be gained into the potential for adaptation of species to high CO 2 over evolutionary timescales.The oceanic chemistry of the northwestern Atlantic and the northeastern Pacific oceans provides such a natural experiment.High CO 2 concentrations are generally absent from the upper water column in the Atlantic (Wanninkhof et al., 2010).In contrast there are currently hypercapnic conditions, where the water is undersaturated with respect to aragonite, in the upper water column in parts of the Pacific.
The source of hypercapnia in the Pacific Ocean is a combined result of ocean circulation coupled with biological processes, leading the old deep waters of the Pacific to be some of the most CO 2 rich in the ocean (Broecker et al., 1982).In addition to this natural process, ocean acidification also plays a role: the pH of the upper water column in the North Pacific is decreasing by ∼ 0.002 pH units per year (Byrne et al., 2010;Chu et al., 2016), similar to the global average of 0.0022 pH units per year (Williams et al., 2015).Such a change corresponds to a total CO 2 , or dissolved inorganic carbon (DIC), increase of 1-2 µmol kg −1 yr −1 (Peng et al., 2003;Sabine et al., 2008;Sabine and Tanhua, 2010;Chu et al., 2016).Although the surface waters in these regions are typically well oxygenated and with a pH > 8, animals that live at or migrate to depth experience increasingly low oxygen (O 2 ), pH, undersaturation with respect to calcium carbonate, and elevated CO 2 (Seibel, 2011).Historically these regions, which occur in many ocean basins, were in fact known more for their low O 2 than for their high CO 2 and were termed oxygen minimum zones (OMZs).These carbon maximum/oxygen minimum zones are extensive in the North Pacific Ocean, whereas similar conditions are rare in much of the Atlantic (Paulmier et al., 2011).Closely related taxa and cosmopolitan species in these two regions therefore experience very different pH levels as well as CO 2 and O 2 concentrations in their normal distribution.Independent of high CO 2 , the reduced O 2 at depth in these OMZs has a profound impact on zooplankton distribution (i.e., Wishner et al., 2008;Escribano et al., 2009;Maas et al., 2014) and can have important implications for the physiology of zooplankton (Childress and Seibel, 1998;Rosa and Seibel, 2008;Seibel, 2011).
Thecosome pteropods are an interesting group for investigating planktonic exposure and response to hypercapnia and low O 2 .Broadly distributed throughout the open ocean, species of thecosomes found in shallow waters of temperate and polar seas can become a numerically dominant member of the zooplankton community (van der Spoel, 1967;Hunt et al., 2008;Bednaršek et al., 2012a).As such, they can be an important part of the food chain (Armstrong et al., 2005;Hunt et al., 2008;Karnovsky et al., 2008), and contribute substantially to carbon flux (Fabry and Deuser, 1991;Noji et al., 1997;Bauerfeind et al., 2009;Manno et al., 2010).Bearing thin shells of aragonite, one of the less stable forms of biogenic calcium carbonate, the calcification of thecosomes has been shown to be impacted by exposure to conditions replicating the projected changes in surface water pH and saturation state of the ocean in the next 100 years (Comeau et al., 2009;Lischka et al., 2011;Manno et al., 2012).Furthermore, recent assessments have shown that their shells are degraded in upwelling and polar regions characterized by undersaturated conditions with respect to aragonite (Bednaršek et al., 2012b(Bednaršek et al., , 2014;;Bednaršek and Ohman, 2015).Studies of metabolism and behavior, however, reveal a complex sensitivity to pH, which is dependent upon natural preexposure and the presence of interactive stressors (Comeau et al., 2010;Maas et al., 2012b;Manno et al., 2012;Seibel et al., 2012).
Previous work has shown that some tropical and subtropical thecosome species undergo diel vertical migrations into persistent and pronounced regions of low O 2 and hypercapnia in the eastern tropical North Pacific.These species showed no change in metabolic rate (O 2 consumption) when exposed to high CO 2 (1000 µatm), revealing the ability of some groups of thecosomes to maintain aerobic metabolism in acidified waters for short periods of time.The one species in the region that does not migrate, however, responded with a suppression of metabolism when exposed to high CO 2 (Maas et al., 2012b).This work in the eastern tropical North Pacific provides evidence that there may be the potential for environmental adaptation of thecosomes to high CO 2 , but it provides no insight into the combined effects of CO 2 and low O 2 .Although research into this topic is underway for other calcifying organisms in coastal habitats (Melzner et al., 2013;Gobler et al., 2014), in the open ocean our understanding remains limited.
The objective of this study, therefore, was to compare the effect of high CO 2 and low O 2 on thecosome pteropods from the northwestern Atlantic and the northeastern Pacific oceans.One of the benefits of this comparison is that there are a number of species of thecosomes that have cosmopolitan distributions occupying both basins and that are known to be diel vertical migrators (Table 1;van der Spoel, 1967;Bé and Gilmer, 1977).Thus, populations in the Pacific would naturally experience hypercapnia and low O 2 in their daytime deep habitat in the Pacific, while in contrast, those from the Atlantic would rarely experience the same environmental  (Bé and Gilmer, 1977;Lalli and Gilmer, 1989).Data include published full ranges at which organisms have been found, as well as previous authors' estimates of the preferred (optimal) ranges of each species, when available.Note that these are based on relatively sparse observations of broadly distributed species, many of which may be cryptic congeners, and thus should be treated as estimates.

Species
(Optimal) temp (Optimal) depth Migrator?stressors.The taxonomy of thecosomes has recently begun to be revisited using molecular and paleontological tools (i.e., Hunt et al., 2010;Jennings et al., 2010;Janssen, 2012;Maas et al., 2013) and there is growing evidence of cryptic speciation for some pteropod groups (Gasca and Janssen, 2014;Burridge et al., 2015).It thus should be noted that the interbasin comparisons performed here may be of cryptic congeners rather than conspecific populations.Using these organisms, which are presumably adapted to their local conditions, we can test whether species or congeners exhibit a population-specific physiological response to these environmental conditions indicative of different sensitivities.

Methods
Thecosome pteropods caught during cruises to the northwestern Atlantic and northeastern Pacific were exposed aboard ship to manipulated conditions of moderately high CO 2 and/or low O 2 for short durations (< 18 h).After this exposure their metabolic rates were measured and then compared to determine whether there were species-or regionspecific responses to the treatments.

Sampling
Animals were collected on two cruises, the first on 7 August to 1 September 2011 in the northwestern Atlantic aboard the R/V Oceanus, and the second in the northeastern Pacific from 9 August to 18 September 2012 aboard the R/V New Horizon.The majority of the sampling in the Atlantic took place along a three-part "z"-shaped transect running between 35 • N, 52 • W and 50 • N, 42 • W, as well as at sites during transit to and from port (Fig. 1).
The first portion of this cruise track corresponded to a segment of the World Ocean Circulation Experiment, Climate and Ocean: Variability, Predictability and Change project (WOCE/CLIVAR) line A20.In the North Pacific the main sampling took place along a two-part transect running between 50 • N, 150 • W and 33.5 • N, 135 • W, corresponding to a portion of WOCE/CLIVAR line P17N, as well as at sites during transit to and from port (Fig. 1).
Sampling was part of a larger interdisciplinary project employing a suite of tools to explore the natural distribution and hydrographic environment of the thecosomes.The sampling design included underway measurements of hydrography, carbonate chemistry, and multi-frequency acoustic backscattering.Comprehensive sampling of the water column was conducted at predetermined stations using a depthstratified 1 m 2 Multiple Opening/Closing Net and Environmental Sensing System with 150 µm mesh nets (MOCNESS; Wiebe et al., 1985), a towed broadband echosounder, video plankton recorder casts, and profiles with a 24-place 10 L Niskin bottle rosette and associated conductivity, temperature and depth (CTD) package.This CTD was equipped with dual temperature and conductivity sensors, a Digiquartz pressure sensor, a SBE43 dissolved oxygen sensor, a biospherical underwater photosynthetically active radiation (PAR) sensor with surface reference, a WET Labs C-Star transmissometer (660 nm wavelength), and a WET Labs ECO-AFL fluorometer.
Hydrographic profiles of temperature, O 2 , and salinity associated with this study were collected using the CTD rosette Niskin bottle package at stations along the main survey transects (Fig. 1).Where CTD casts were unavailable, at stations conducted during the transits to and from port, an expendable bathythermograph (XBT) was deployed to determine the temperature of the water column.Bottle samples of carbonate parameters, nutrients, and other parameters were collected at selected water depths using the CTD rosette package.
Table 2.The hydrography and location for each station where animals for experiments were collected.Each basin was characterized by multiple hydrographic regimes (see text and Fig. 2); transitions between regimes are denoted by horizontal lines between stations.At stations along the main transect the depth (m) at which O 2 decreased below 130 µmol O 2 kg −1 (∼ 48 % saturated), the average temperature from 25 to 100 m ( • C) and the average salinity from 25 to 100 m were derived from CTD casts.At a few stations (denoted via a ) in the Atlantic there was warm water at the surface and fresher cold water below.The only species in this region, Limacina retroversa, has an optimum temperature between 7 and 12 • C (Bigelow, 1924) and was generally found above 50 m (Lawson, unpublished data).At these sites the average temperature and salinity is reported first for 25-100 m and then also for 25-50 m to reflect the conditions likely experienced by the pteropods.pCO 2 and Ar were calculated from measured pH T and DIC bottle samples.We linearly interpolated the depths (m) at which the pH T decreased below 7.7, pCO 2 reached 800 µatm, and aragonite saturation ( Ar ) reached 1 from the discrete measurements at adjacent depths (NA if these values were never reached in the top 500 m).At stations conducted while in transit to the main study transects (denoted by prefix T) the average temperature from 25 to 100 m ( • C) was documented from XBT casts.At these transit stations no O 2 or carbonate chemistry data were available (noted with a dash).The species caught at each station and used in this study are demarcated with a star ( * ).2) and used in the experiments with animals from that station.

Environmental carbonate chemistry
Discrete pH samples were directly collected from the 10 L Niskin bottles into 10 cm cylindrical optical cells and measured within 4 h of collection (Clayton and Byrne, 1993;Dickson et al., 2007).These pH samples were analyzed spectrophotometrically on an Agilent 8453 spectrophotometer at a control temperature (25.0 ± 0.1 • C) following the method detailed in Dickson (2007) and Clayton and Byrne (1993) using m-cresol purple as the indicator.The pH results in total scale (pH T ) have been corrected for indicator impurity (Liu et al., 2011) and indicator perturbation to seawater samples.The pH measurements have a precision better than 0.001 and an accuracy of ∼ 0.002.
Nutrient samples (nitrate/nitrite, phosphate, silicate, and ammonia) were collected in 20 mL plastic bottles after filtration through a 0.22 µm Pall capsule filter and kept frozen until analysis.Nutrient samples were analyzed either at the WHOI Nutrient Analytical Facility or the University of California, Santa Barbara, using a Lachat Instruments QuickChem 8000 four-channel continuous flow injection system, following standard colorimetric methods approved by the US Environmental Protection Agency.
Discrete samples were also taken for dissolved inorganic carbon (DIC) and total alkalinity (TA).These were collected in 250 mL Pyrex borosilicate glass bottles after being filtered with a 0.45 µm in-line capsule filter and poisoned with saturated mercuric chloride (Dickson et al., 2007).DIC samples were analyzed on a DIC auto-analyzer (AS-C3; Apollo SciTech, Bogart, USA) via sample acidification, followed by nondispersive infrared CO 2 detection (LiCOR 7000: Wang and Cai, 2004;Wang et al., 2013).The instru-ment was calibrated with certified reference material (CRM) from Dr. A.G. Dickson at the Scripps Institution of Oceanography.The DIC measurements have a precision and accuracy of ±2.0 µmol kg −1 .TA measurements were made with an Apollo SciTech alkalinity auto-titrator, a ROSS combination pH electrode, and a pH meter (ORION 3 Star) based on a modified Gran titration method with a precision and accuracy of ±2.0 µmol kg −1 (Wang and Cai, 2004).
The remaining water column carbonate system parameters, including aragonite saturation state and pCO 2 , were calculated from DIC-pH T pairs at in situ nutrient, temperature, salinity, and pressure using the software CO2SYS (Pierrot et al., 2006) and the dissociation constants of Mehrbach et al. (1973), refitted by Dickson and Millero (1987), and the KHSO 4 dissociation constant from Dickson (1990).Depths for pH T = 7.7, pCO 2 = 800 µatm, and aragonite saturation state of 1 were then linearly interpolated using the closest available measurements.
Surface water pCO 2 was continuously measured throughout both cruises using an automated underway system (Model 8050; General Oceanics Inc., USA) based on headspace air-seawater equilibration followed by infrared detection (LiCOR 7000).This system was calibrated every 1-2 h with three CO 2 gas standards traceable to World Meteorological Organization CO 2 Mole Fraction Scale.These underway pCO 2 measurements have a precision and accuracy of ∼ ±1 µatm.Measurements made by the underway system provide insight into the surface carbonate chemistry parameters at stations made in transit where bottle samples were not collected.
A. E. Maas et al.: The metabolic response of thecosome pteropods

Specimen capture
Thecosome species were chosen for physiological study opportunistically as they appeared in net samples at successive stations.Species were targeted specifically for their abundance and the likelihood of their presence in both ocean basins, and only adult individuals were used.Most individuals were collected with a 1 m diameter, 150 µm mesh Reeve net with a ∼ 25 L cod-end in the Atlantic and a similar 1 m diameter, Reeve net equipped with 330 µm mesh in the Pacific.Use of the Reeve net with its large and heavy cod-end in combination with slow haul rates (typically 5-10 m min −1 ) allowed for gentle collection of the delicate thecosomes and consistently supplied animals in good condition with undamaged shells and external mantle appendages.Net tows were made at night when animals were expected to congregate at shallow depths, were ∼ 1 h in duration in an effort to minimize the handling time of the organisms, and reached a maximal depth between 100 and 150 m.Depths were targeted that had a high chlorophyll a peak during CTD casts, high acoustic backscattering on the echosounder, and/or where thecosomes had been abundantly sampled at the same station using the MOCNESS.Occasionally, individuals of less abundant species were collected from the nets of the MOCNESS for physiological study, but only if their shells were undamaged and they were swimming normally.
Post-capture, individuals were transferred to filtered water in densities of < 15 ind.L −1 and kept for at least 8 h in temperature-controlled water baths to allow for gut clearance.Temperatures for experimentation (20, 15, or 10 • C) were chosen to be generally representative of the waters from which the animals were sampled, based on the vertical distributions and hydrographic conditions documented with the stratified MOCNESS sampling.Chosen temperatures were typically the average of the water temperature between 25 and 100 m, although in the middle section of the Atlantic cruise experimental temperatures were reflective of the 25-50 m average due to the particularly shallow vertical distribution of the dominant species (Limacina retroversa) sampled in this region.This was to ensure that experiments were occurring at physiologically relevant and, presumably, natural temperatures for each species.After gut clearance, individuals that were in good condition (i.e., swimming and with shell intact) were used for oxygen consumption experiments.

Experimental exposures and oxygen consumption rate
Post-gut clearance, healthy animals were put into separate glass syringe respiration chambers, one individual per chamber, with a known volume of 0.2 µm filtered seawater and 25 mg L −1 each of streptomycin and ampicillin.This minimized the microbial respiration effects on the measurements of carbonate chemistry and O 2 consumption rates by pteropods during the experiments.The inclusion of antibi-otics, a method which has previously been used with thecosomes to prevent bacterial growth in respiration experiments (Maas et al., 2012a), was shown during the Pacific cruise to have no effect on the O 2 consumption of at least Limacina helicina for the exposure durations associated with these experiments (Howes et al., 2014).The volume of water in the treatments was chosen to complement the size of the organism and temperature of the experiment and ranged between 15 and 50 mL in 2011 and 8 and 20 mL in 2012.For every 3 to 5 treatment chambers, a "control" respiration chamber (experimental seawater with antibiotics and without pteropods) was set up to monitor microbial activity and to provide water for characterization of the starting conditions.Filtered seawater for experimental exposures was collected during both cruises in batches at approximately weekly intervals from the surface; experimental water thus began with chemical properties (notably including TA, DIC, pH, and salinity) reflective of the local environment and was then manipulated to modify CO 2 and/or O 2 concentrations.Manipulations were achieved by bubbling 1 L batches of collected seawater with gas mixes (certified accurate to ±2 %) for 45-60 min with one of two oxygen (100 and 48 % of O 2 saturation) levels crossed with two CO 2 (nominally 380 and 800 µatm) levels.At the time of the experiment, surface air pCO 2 conditions were on average ca.380 ppm, dictating our ambient (i.e., low carbon, LC) conditions.In 2011 the ambient condition (∼ 100 % of O 2 saturation and 380 µatm CO 2 ) was achieved by bubbling with an ambient clean air line, while in 2012 it was achieved by a certified 380 ppm gas mix.
The experimentally modified concentrations mimic the CO 2 and O 2 levels that would be experienced by thecosomes within the top 400 m of the Pacific Ocean, and they reflect the average atmospheric CO 2 level projected for the open ocean in the year 2100 (A2 emissions scenario, IPCC, 2007).This resulted in four total treatments: low (i.e., ambient) CO 2 , high oxygen (LC/HO), which is representative of current ambient surface ocean conditions; high carbon, high oxygen (HC/HO), which replicates what we expect the average future surface ocean to resemble; low CO 2 , low oxygen (LC/LO); and high carbon, low oxygen (HC/LO), which is similar to what organisms in the Pacific would experience during a diel vertical migration into the local oxygen minimum zone.The goal of this design was to allow us to compare directly the Atlantic and Pacific thecosomes to see whether exposure to 800 µatm pCO 2 and/or 48 % of O 2 saturation resulted in different outcomes.The level of low O 2 chosen for this study was well above the threshold that has been designated as stressful for non-specialized metazoan life (< 2 mg O 2 L −1 or 60 µmol O 2 kg −1 ; Vaquer-Sunyer and Duarte, 2008) in order to test the non-lethal effect of moderately low O 2 on individuals from the two ocean basins.Calculations based on the salinity and temperature of the water indicated that bubbling with 48 % of O 2 saturation achieved conditions of 48-62 % of O 2 saturation by the start of experiments.Subsequent analyses (see below) also confirmed that intended CO 2 concentrations were achieved for all treatments within reasonable ranges, with the exception of the LC/LO Atlantic treatment.In this case, the gas cylinder was evidently improperly mixed by the manufacturer and analyses suggested a ca. 100 ppm CO 2 concentration.The results for this treatment are still presented but should be interpreted as a distinct treatment.
Oxygen consumption was measured following similar techniques as described in Marsh and Manahan (1999).Briefly, at the conclusion of the experiment water was withdrawn from treatment or control chambers using an airtight 500 µL Hamilton syringe and injected past a Clarke-type microcathode (part #1302; Strathkelvin Instruments, North Lanarkshire, United Kingdom) attached to an O 2 meter (part #782) in a water-jacketed injection port (part #MC100).This was done three times, allowing the reading to stabilize for at least 30 s before a measurement was taken.Generally, the change in oxygen consumption was between 3 and 25 % of the control value.In high oxygen experiments, if the oxygen level fell below 70 % of air saturation they were excluded from the analysis.
Following exposure, animals were removed from the chamber, blotted dry, and frozen in liquid nitrogen.These individuals were later weighed using a microbalance (±0.0001 g) and the resulting mass-specific O 2 consumption rates are reported in µmoles (g wet weight) −1 h −1 .Wet weights are used here because they are more relevant for physiological understanding of animal function (Childress et al., 2008) but dry weights can be estimated from these using the wet weight-dry weight relationships developed previously for pteropods (Ikeda, 2014).To replicate the duration of exposure that would be experienced by most thecosomes in the Pacific undergoing a daily migration to depth, the experiments were targeted to last 6 to 12 h.In practice, experiments ranged from 6 to 18 h for normoxic and 3 to 10 h for low O 2 trials.This variation in duration resulted from balancing the need to elicit a measurable change in O 2 concentration with preventing extreme O 2 depletion of the chambers (< 6 % oxygen saturation) and accounting for multiple species of variable size and metabolic rate.

Experimental carbonate chemistry
Carbonate chemistry of the treatments was characterized in most cases via measurements of DIC and TA of experimental seawater, unless indicated otherwise.The process of measuring the O 2 in the treatments used up a large portion of the water and then the chamber was unsealed and disturbed to remove the animal, rendering it impractical to measure the carbonate chemistry directly from the respiration chambers.DIC measurements were thus taken from the control syringes within 18 h of the end of each experiment and used to represent the starting point of the carbonate chemistry conditions the animals experienced.Water samples were allowed to come to room temperature (> 6 h) before analysis.DIC was measured using the same system as that used for the hydrographic characterization (see above).Estimates of the effect of CO 2 production via respiration in treatment chambers on DIC were made using a respiratory quotient of 0.8 mole of CO 2 per 1 mole of O 2 consumed (calculated using Sagitta elegans; Mayzaud, 1976) to characterize the ending conditions of the experiments.
Due to the small volumes of water in the experimental chambers, it was not possible to measure both DIC and TA from the control syringes.Instead, TA samples intended to be representative of the starting experimental conditions were collected via siphoning from each batch of collected surface water.These samples were subsequently measured based on the analytical method described above (Wang and Cai, 2004).TA of experimental water was assumed to have been constant over the course of each experiment as water was filtered (0.2 µm) and antibiotic treated (thus microbial activities were kept at minimum).Although pteropod aerobic respiration, excretion, and calcification within a respiration chamber could influence TA, these are presumed to have not had a significant influence over the time scales in question.
In some instances, however, measured TA from the batches of experimental water was substantially dissimilar to that of the surface measurements made from nearby in situ surface bottle samples collected with the CTD (> 20 µmol kg −1 ; see Sect.3.3).Calculated pCO 2 values in these cases were also significantly different from batches of experimental water collected from other locations, but bubbled with the same CO 2 gas tank.These differences are more than 10 times the measurement precision/accuracy and 5 times the uncertainty of duplicate sampling and measurements during the cruises.They are also beyond the likely level of TA variation due to differences in sampling location (geographic and in depth) between the in situ bottle samples and experimental water batches and are more likely a consequence of the difficulties associated with cleanly siphoning the experimental water batches (i.e., contamination during sampling).For completeness, the carbonate chemistry system parameters for the experimental water, including aragonite saturation state and pCO 2 , are reported based on calculations using DIC-TA pairs using both the experimental TA and the in situ measurements from the CTD bottle samples; in those cases where the TA measurements diverged substantially (> 20 µmol kg −1 ), however, we base our interpretations on the in situ measured TA at nearby CTD stations instead of the values of experimental water.In those circumstances where batch water was taken from test stations during transit to and from the main study regions and CTD bottle data were unavailable, the experimental TA was checked using calculated TA values using DIC from the LC/HO treatments and pCO 2 from the underway measurements.
A. E. Maas et al.: The metabolic response of thecosome pteropods

Statistics
Oxygen consumption rates were tested for significant differences between groups using SPSS software.Univariate general linear models (GLMs) were conducted to determine the effect of CO 2 level, O 2 level, and their interactive effect using the log-transformed oxygen consumption with log-transformed wet mass as a covariate separately for each species (2 factor design; "CO 2 × O 2 ").In the Atlantic this full factorial design was confounded by the incorrect gas mixture so each treatment was tested independently (1 factor design; "treatment").Species that were collected during both years and basins, and experiments conducted on species at multiple temperatures, were analyzed separately so that the effect of variations in mass between seasons and the changes in metabolic rate at different temperatures would not confound the analysis.The datasets were tested for normality and homoscedasticity and, in cases where significance was found in the GLM, they were explored with Bonferroni pairwise post hoc comparisons.
For some species the temperature of experimentation was different among stations within a basin.For analyses with these species when comparing species between ocean basins, we applied a standard temperature coefficient (Q 10 ) to compare across temperatures.The adjusted rates (R f ) were calculated at 15 • C using a Q 10 of 2 according to the equation , where R i is the original metabolic rate measured at the original temperature (T i ).Although previous work with thecosomes has shown that Q 10 is species specific (Seibel et al., 2007;Maas et al., 2011Maas et al., , 2012a)), for many of the species used in this study there are no published estimates of Q 10 .Thus, this coefficient value was chosen as it is mid-range (1.3-2.7) for the published Q 10 of nonpolar thecosome species as recently compiled by Ikeda (2014) and is consistent with estimates of average Q 10 for marine ectotherms, which typically fall between 2 and 3 (Hochachka and Somero, 2002;Seibel and Drazen, 2007).

Specimen capture
Following currently accepted morphology-based taxonomy, adult individuals from a total of eight species of pteropods were collected over the course of the two cruises for physiological studies.Only relatively large adult specimens were used in respiration trials, in part to avoid any confounding effects of ontogeny and in part to ensure a measurable change in oxygen levels.We collected two species of thecosome pteropods exclusively from the Atlantic, Limacina retroversa (Fleming, 1823), a subpolar species, which is absent from the North Pacific, and Diacria trispinosa (Blainville, 1821), which can be found in temperate and tropical regions of the Atlantic, Pacific, and Indian oceans.Although present in both the North Atlantic and Pacific, the polar to sub-polar species Limacina helicina (Phipps, 1774) was only sampled in the Pacific transect.Collections of this species consisted of intermixed formae, the high spiraled Limacina helicina acuta (van der Spoel, 1967), the lower spiraled Limacina helicina pacifica (van der Spoel, 1967), and a forma that bore resemblance to both in a mixed morphology.Since both the assemblage and morphology of these formae were mixed they were tested as one population/species.In both ocean basins we collected Styliola subula (Quoy and Gaimard, 1827), Cavolinia inflexa (Lesueur, 1813), and Clio pyramidata (Linnaeus, 1767).There is some morphological and molecular evidence that Cuvierina columnella (Rang, 1827) is actually multiple distinct species, now including Cuvierina atlantica and Cuvierina pacifica (Janssen, 2005;Burridge et al., 2015), and we tested individuals of these species from their respective ocean basins.

Hydrography
Two hydrographic regimes were evident along the North Pacific study transect (Table 2; Fig. 2).The northernmost stations (50  W; stations 15-34, T9-T10; Fig. 1), temperatures at depths between 25 and 100 m were higher, ranging between 10 and 17 • C, representative of the transition zone into the North Pacific Gyre.Along this portion of the transect O 2 concentration consistently fell below 48 % saturated by depths of 340 and 400 m.The depth at which pH T fell below 7.7 increased gradually from ∼ 150 to 230 m as latitude decreased.Correspondingly, the depth at which pCO 2 in this area reached 800 µatm was 330 to 440 m, and the aragonite saturation horizon was 330 m to 430 m depth.The depth at which species would experience a pH T below 7.7 was within the inhabited depth range known from the literature for all of the species tested in this portion of the study region but only the species Clio pyramidata, with a typical vertical range of 0-500 m (Table 2), would be likely to experience 48 % of O 2 saturation, 800 µatm pCO 2 , and aragonite undersaturation in its typical distribution (Table 1).
In contrast to the Pacific, along the entire Atlantic transect O 2 concentration was above ∼ 200 µmol kg −1 (∼ 72 % saturation) in the top 500 m, while pCO 2 never reached Figure 2. Hydrography of sampling regions.Hydrographic profiles of stations representative of the specific water mass types from the northern (P-T5, P-6, A-26), middle (P-18, A-19), and southern (P-32, A-8) portions of the Pacific (P) and Atlantic (A) study transects (station locations: Fig. 1).At station P-T5, the temperature profile (grey) was from an XBT cast because no CTDs were conducted during transits.For all stations along the main transects, left-hand plots show temperature (grey), salinity (black), and oxygen (black dotted) measured via sensors on the CTD and binned to 1 m depth intervals.Middle plots show TA (black) and DIC (grey) from discrete bottle samples (dots show depths of bottle samples).Right-hand plots show pCO 2 (black) and aragonite saturation state ( Ar ; grey) calculated based on TA and DIC measurements.
800 µatm, and aragonite undersaturation never occurred throughout the top 1000 m.There were three dominant hydrographic regimes in the Atlantic (Table 2; Fig. 2).In the northeastern part of the sampling region (50 typified by a temperature and salinity anomaly with temperatures below 5 • C from 25 to 100 m and a salinity signature < 33, contrasting significantly with the surface salinities of the northern portion (∼ 34) and southern portion (∼ 36) of the Atlantic transect.As a consequence, these stations contained water of the lowest pH, with surface waters reaching 7.7 at depths shallower than 200 m.

Carbonate chemistry of experiments
Bubbling with CO 2 levels of ∼ 380 and ∼ 800 ppm resulted in a distinct separation of carbonate chemistry between treatments during the experiments in both oceans (Table 3).Due to pre-existing differences in the carbonate chemistry of the seawater collected in each ocean, TA differed between the two basin treatments.In the Atlantic the DIC of the ambient CO 2 treatments ranged from 2030 to 2090 µmol kg −1 and the high CO 2 treatments ranged from 2140 to 2220 µmol kg −1 , with an average difference between treatments of similar temperature and salinity of 132 µmol kg −1 .Surface TA in the region decreased from ∼ 2370 µmol kg −1 in the southern part of the transect to 2300 µmol kg −1 in the northern latitudes.
In the Pacific the DIC of the ambient CO 2 treatment ranged from 1930 to 2020 µmol kg −1 and the high CO 2 treatment ranged from 2030 to 2110 µmol kg −1 , with an average difference of 90.7 µmol kg −1 between the treatments.Surface TA in this basin was 2150 µmol kg −1 in the most northern collection and had increased to 2200 µmol kg −1 by the transect mid-point.Calculations of pCO 2 based on these measurements of DIC and TA suggested that target pCO 2 levels were generally attained and were consistent between the two cruises, with the exception of the LC/LO treatment in the Atlantic.In this case, there was a substantial deviation from the intended pCO 2 , suggesting values ranging from 99 to 139 µatm in contrast to a range of 311-391 µatm for the LC/HO in the Atlantic and 283-409 µatm for LC/HO and 295-397 µatm in the LC/LO in the Pacific.Evidently, this indicates improper mixing of the gas concentration in the Atlantic LC/LO gas cylinder by the manufacturer.The calculations for the high CO 2 treatments were more consistent between cruises, with pCO 2 for the HC/HO being 585-868 µatm and the HC/LO being 755-783 µatm in the Atlantic, while in the Pacific the HC/HO treatment was between 520 and 740 µatm and the HC/LO was 546-766 µatm.The variability in calculated pCO 2 values likely represents variations in bubbling time, temperature, and the degree to which the water reached saturation relative to the gas mixtures.
As a consequence of the natural differences in seawater carbonate chemistry, in particular the TA differences between the two ocean basins, there were inherent differences in the aragonite saturation state between the Pacific and Atlantic treatments (Table 3).In the Atlantic, Ar of the ambient CO 2 treatment ranged from 2.4 to 3.5, except for the LC/LO treatment ( Ar 4.0-5.5),which was bubbled with an incorrect gas mixture as discussed above.In comparison, in the Pacific the ambient CO 2 condition had a lower range of Ar (2.2-2.4) for both the LC/HO and the LC/LO treatments.The experimental conditions of the high CO 2 treatments reached their lowest value in the middle part of the transect ( Ar = 1.2 at mid-latitudes; Table 3), where cold northern waters of low salinity were encountered.Experimental Ar had a range of 1.5-2.0 for the rest of the transect in the Atlantic.The values of experimental Ar were lower overall in the Pacific, although the high CO 2 treatments also never reached undersaturation ( Ar 1.3-1.8).In general, the manipulation of carbonate chemistry in this study successfully created two distinct ranges for both pCO 2 and aragonite saturation state ( Ar ).
It is important to acknowledge that the production of CO 2 via organism respiration within the chambers would modify the carbonate chemistry of the treatments over the duration of the experiments.Based on the average respiration rate, we estimate an average DIC production of ∼ 18.0 µmol kg −1 by the end of an experiment.Applying such a change to the experimental conditions in the northeastern Pacific, where seawater is more sensitive to changes in DIC due to a lower buffering capacity compared to the Atlantic (i.e., a worst case scenario), Ar would only change by < 0.1 in both the LC and HC experimental chambers over the course of the respiration experiments.Although this is an appreciable effect, we nonetheless retain a wide separation between the ambient and high-CO 2 treatments and in no cases would the treatments reach undersaturation as a consequence of this biological activity.As such, for simplicity the results reported in Table 3 do not include this correction for respiration.

Effect of CO 2
Varying availability and abundances of the different thecosome pteropod species in the net samples precluded all species being exposed to the full factorial design, but individuals of all species were tested under the low CO 2 , high oxygen (LC/HO) and high carbon, high oxygen (HC/HO) treatments (Fig. 3, Table 4).To explore differences in metabolism attributable to a response to CO 2 , the log-transformed wet mass was used in a GLM as a covariate comparing the logtransformed oxygen consumption (response variable) under low and high-CO 2 conditions; each population within a species that was sampled in both basins or run at multiple experimental temperatures was examined separately.There was no significant effect of CO 2 for any species in either basin.

Effect of basin
Following this assessment, we were interested in determining whether there were differences in metabolic rate between basins.As such we ran a GLM using log-transformed Table 3. Carbonate chemistry during manipulation experiments.The manipulation experiments were conducted at multiple temperatures (T ) and salinities (S) based on the conditions the organisms were caught in.As described in more detail in the text, DIC measurements were made of water drawn from the control chambers while TA was measured for batches of experimental water (denoted as xpt.TA).In situ TA (i.s.TA), based on nearby CTD bottle sampling at the surface, is also shown.At test stations conducted while in transit to/from the main study regions, where bottle samples of in situ TA were unavailable, underway pCO 2 values and the LC/HO DIC were used to calculate in situ TA (denoted with * ).In some instances, measurements of experimental TA differed by > 20 µmol kg −1 from nearby in situ measurements of surface TA.This difference greatly exceeds expected variability based on measurement uncertainty and spatial (geographic and vertical) offsets in the locations of experimental water collection relative to the nearest CTD cast; in these circumstances, the experimental TA was likely erroneous due to sampling issues (e.g., contamination).For completeness, and to aid in identification of erroneous experimental TA values, calculations of carbonate chemistry parameters, including aragonite saturation state ( Ar ) and pCO 2 were made based on DIC and both experimental TA and in situ TA.In further data analysis and interpretation, calculations based on experimental TA are given preference except those few instances where experimental TA differed from in situ by > 20 µmol kg −1 (bold denotes preferred calculations).Calculated saturation state and pCO 2 are reported as the average and standard deviation per batch of water.Note that the LC/LO gas tank in 2011 (in italics) appears to have been improperly mixed by the manufacturer as calculations suggested it contained a much lower CO 2 level than the intended 380 µatm; it should consequently be considered an entirely separate treatment from the 2011 LC/HO (where CO 2 levels were based on bubbling with an ambient air line).metabolic rates for the three species that were found in both basins, normalized to 15 • C to account for differences in experimental temperature by applying a standard temperature coefficient.With the log-transformed wet mass as a covariate, we tested for an effect of basin, CO 2 , and an interactive term.Clio pyramidata had a similar metabolic rate between basins.In contrast, Cavolinia inflexa (F 1,20 = 10.358,p = 0.004) and Styliola subula (F 1,23 = 11.817,p = 0.002) both had a significantly lower metabolic rate in the Pacific, although there was no interactive effect of CO 2 .

Effect of O 2
For the species where enough individuals were collected to provide experimental replicates to explore the interactive effects of CO 2 and O 2 we also ran a species-and basin-specific GLM exploring the effect of treatment (Fig. 3, Table 5).Clio pyramidata, the only species we were able to test in both basins showed no significant effect of high CO 2 , low O 2 or the interactive treatment in either basin.In the Pacific, L. helicina and C. inflexa similarly showed no significant change in metabolic rate as a consequence of any of the treatments.In contrast, in the Atlantic there was a significant effect of treatment for L. retroversa and a Bonferroni post hoc analysis comparing the treatments found that the high CO 2 , low O 2 (HC/LO) treatment was significantly lower than all other treatments (Fig. 4; F 3,38 = 17.836, p < 0.001; a ∼ 60 % reduction in the average mass-specific metabolic rate in comparison with the LC/HO treatment; Table 4).Cuvierina atlantica was tested at both 15 and 20 • C in the Atlantic, so to make comparisons among these experiments a temperature coefficient was applied and rates were normalized to 15 • C, after which no significant effect of any treatment was found for this species.

Discussion
This study reveals that short-term exposure to low O 2 and high CO 2 , similar to what would be experienced by individuals in the Pacific during diel vertical migration, does not influence the oxygen consumption of adult individuals of most of the thecosome pteropod species examined from either the Atlantic or Pacific.The only species that had a significant change in respiration in response to any of the treatments was Limacina retroversa from the Atlantic, which responded to the combined effect of low O 2 and high CO 2 with a reduction in oxygen consumption rate.
Table 5. Statistical results of the univariate general linear models (GLM) for each species were analyzed separately by year and are listed relative to the temperature of the experiment (Temp.; • C).For species studied at multiple temperatures (denoted by * ), the metabolic rates were adjusted to 15 • C using a Q 10 = 2 to allow for direct comparison.The effect of the independent factors of CO 2 level (CO 2 ), O 2 level (O 2 ), their interactive effect (Int.), and the covariate of mass were analyzed in regards to the metabolic rate and reported as p values for the Pacific (mean mass-specific metabolic rate values found in Table 4).For the Atlantic, each treatment was tested as independent (Treat.)due to the accidentally low CO 2 condition in the LC/LO gas mixture.We report whether the data met the assumption of normality of the residuals with Shapiro-Wilk (norm.; for p under 0.05 the assumption is not met) and heterogeneity of variance (var.; for p under 0.05 the assumption is not met) and denote in bold where the dataset did not fully meet these statistical assumptions.Note that for the sole case where the treatment or CO 2 effect was significant (L.retroversa) all assumptions were met.

Experimental design
A factor that should be considered when interpreting our results is the dynamic hydrographic conditions that the animals experience naturally between and within the ocean basins.
Thecosomes of multiple species were found at a range of temperatures, salinities, and carbonate chemistries, meaning that they experienced a range of pH and aragonite saturation states in their natural habitat.When comparing animals from multiple locations, we chose to use local water in order to replicate these natural conditions and to exclusively manipulate the CO 2 concentration, as this is the factor that is changing due to anthropogenic activity.This approach, however, does not control for the other parameters of the carbonate chemistry system, which will vary between regions.Despite this fact, there was a clean distinction between treatments, notably in terms of aragonite saturation state as well as CO 2 concentration, which provides insight into the effect of moderate short-duration exposure to CO 2 .It is also important to note that the individuals of L. helicina from the Pacific experiments did occasionally have very high mortality during the period prior to experimentation (> 80 % at transit station T2 and T5, decreasing substantially to the northwest and along the main Pacific transect).These individuals, which are polar-subpolar organisms and are typically found between −2 and 10 • C (Lalli and Gilmer, 1989), were collected from water that was likely near the upper limit of their optimal temperatures, although alternate possibilities are that these were a population reaching senescence, or that they were collected in a hydrographic regime with low food availability.Animals collected from these sites that were used in subsequent respiration experiments may therefore have been taken from an already stressed population and should be recognized as such.

Carbon dioxide effect
Hydrographic profiles collected in the Pacific coincident to sampling of thecosomes indicate that organisms in the northern portion of the study region would experience conditions of high CO 2 and low O 2 in the upper ∼ 450 m of their distribution (Chu et al., 2016).Based on previous knowledge of the vertical distributions of the thecosomes used in this study, only the species Clio pyramidata would ever experience a pH T below 7.7 and none of the thecosomes studied would experience 800 µatm pCO 2 or undersaturation within their vertical range in the Atlantic study region (Table 1).Despite these environmental differences, we found no significant effect of increasing CO 2 alone on the respiration rates of any of the species from either ocean basin.These results increase the published evidence that short-term (6-18 h) exposure to enhanced CO 2 without synergistic stressors has no significant effect on the metabolic rate of many species of thecosome pteropods.Thus far, there are only two species that have been documented to show a change in metabolism based on exposure to manipulated CO 2 alone: Limacina antarctica (789-1000 µatm, 24 h: Seibel et al., 2012) and Diacria quadridentata (1000 µatm, 6-18 h: Maas et al., 2012b).The metabolic rates of all other species yet studied, including Hyalocylis striata, Clio pyramidata, Diacavolinia lon-  The species and temperature of the experiment are reported below the x axis.Significance is reported based on a basin, species, and temperature-specific GLM, which tested for the effect of treatment on O 2 consumption with a Bonferroni post hoc analysis (Table 5).In the Atlantic analysis each treatment was tested independently, while in the Pacific CO 2 and O 2 were treated as factors.For each species and temperature, treatments are reported as non-significant (N.S.) or, in the case of significance, by letters that indicate which treatments are statistically similar (same letter) or different (different letter) at a p value < 0.05.Note that for C. atlantica the metabolic rates of individuals respired at 20 • C were converted to 15 • C using a temperature coefficient of 2 (see methods) for this GLM analysis.
girostris, Creseis virgula (6-18 h: Maas et al., 2012b), and Limacina helicina (24 h: Comeau et al., 2010), were not significantly affected by short-term exposure to high CO 2 , although the latter species showed an increase in metabolic rate when high CO 2 was combined with high temperatures.Our results, which increase the geographic coverage for L. helicina and C. pyramidata and provide the first data about the species C. pacifica, C. atlantica, L.retroversa, D. trispinosa, C.inflexa, and S. subula, corroborate these earlier findings.One interpretation of these results is that physiological responses may have occurred, but involved the reallocation of resources to different tissues or metabolic pathways; this redistribution could serve to maintain the thecosome total energy budget, and subsequently would not significantly change the metabolic rate of the individuals.A transcriptomic study done with individuals of Clio pyramidata as a companion project to the present work in fact suggested that expression of some genes was influenced by CO 2 exposure even though metabolic rate was not (Maas et al., 2015), perhaps suggesting some reallocation among energetic demands.If this is the case it indicates that, to some degree, the short-term exposure to high CO 2 concentration is within the physiological tolerance of the tested species.Alternative hypotheses are that the duration of exposure was too short or the severity of the CO 2 treatment too minimal to elicit a measurable response.It is possible, for example, that some processes, like biomineralization, may be influenced by high CO 2 , but only after a longer exposure duration.Finally, it may be that changes in respiration rate were subtle, requiring a much greater sample size to identify them in light of biological variability, but exploration of this hypothesis would require a dedicated experiment to collect more individuals and likely a smaller number of species.This possible tolerance to short-term CO 2 exposure may be due to the fact that within their distribution or diel migrational range there are conditions, or perhaps seasons, where the natural hydrography causes many species of thecosome to experience conditions of high CO 2 /low pH, and the species are therefore adapted to this range of exposure.The Arctic species L. helicina and subarctic species L. retroversa, for instance, are thought to inhabit waters which have been shown to reach a concentration of > 950 µatm CO 2 and to be undersaturated with respect to aragonite during the winter season in Kongsfjorden, Svalbard (Lischka and Riebesell, 2012).These conditions are pervasive throughout the upper water column, meaning that L. helicina and L. retroversa, which are not strong diel migrators, would experience seasonal undersaturation in these polar regions.The more temperate and tropical open ocean thecosomes, including C. pyramidata, C. inflexa, and S. subula are all currently believed to be circumglobal and most, to varying degrees, diel migratory (Table 1; van der Spoel, 1967;Bé and Gilmer, 1977).Populations are therefore likely to encounter high CO 2 in subsurface waters in regions associated with OMZs, including much of the North Pacific and off the coast of Northern Africa.The ability to cope with high CO 2 for short durations may have been selected over time as a natural consequence of the types of unavoidable environmental variability experienced by these planktonic populations.

Low O 2 and combined effects
In the Pacific Ocean, none of the species for which we had enough individuals to perform the low O 2 study (L.helicina, C. pyramidata, and C. inflexa) had a significant change in metabolic rate under low (48 % saturated) O 2 , even when combined with enhanced CO 2 .These results indicate that the O 2 levels were above the concentration below which these species can no longer sustain their routine metabolic activity (critical partial pressure; Pcrit; Hochachka and Somero, 2002) and that any changes in physiology associated with the treatments required no increased energetic expenditure or metabolic reduction.As subsurface waters throughout the cruise were frequently below 48 % of O 2 saturation (< ∼ 130 µmol kg −1 ), this indicates that these species may be naturally adapted to coping with low-O 2 conditions.
In the Atlantic, examination of the effects of low O 2 is confounded by an unfortunate and accidentally low level of CO 2 (∼ 130 µatm) in the LC/HO treatment (Table 3).Tests of the effect of high CO 2 (HC/HO) and the interactive (HC/LO) treatments nonetheless remain valid, and for L. retroversa, exposure to HC/LO caused a large and significant reduction in metabolic rate.Suppression in metabolic rate is a common tactic for surviving unfavorable conditions (Guppy and Withers, 1999;Seibel, 2011).Although metabolic depression is generally survivable in the short term, over longer time scales there are often implications for growth, reproduction, and survival (reviewed in : Pörtner, 2010;Seibel, 2011).In the Atlantic, our measured in situ O 2 levels were never below 15 % (∼ 200 µmol kg −1 ).In contrast with the other species studied, which in at least some portions of their geographic range are occasionally found in association with subsurface low O 2 combined with hypercapnia, L. retroversa lives exclusively in the subpolar North Atlantic Ocean and the Antarctic Circumpolar Current.As such this is the only species in this study in which no population is likely to experience conditions of low O 2 and high CO 2 together naturally anywhere in its distribution.Its inability to maintain metabolic rate during this interactive exposure may be a short-term metabolic response to environmental conditions that are unsustainable over longer time periods.As a consequence of the very low CO 2 in the LC/LO treatment, it is impossible to determine whether the metabolic suppression for L. retroversa in the HC/LO was in response to reduced O 2 availability alone or to the interactive effect of low O 2 with high CO 2 .In the LC/LO treatment any change in respiration due to low O 2 could have been masked by a change in the energy budget as a response to the low (equivalent to pre-industrial atmospheric conditions) levels of CO 2 .The results suggest that further work in the Atlantic is warranted to disentangle these stressors and to determine whether the observed change in metabolic rate was solely a consequence of O 2 availability or truly a synergistic effect.
Interestingly, although the temperature coefficients were not species specific and may not, therefore, perfectly normalize the dataset, one trend revealed by their use was a www.biogeosciences.net/13/6191/2016/Biogeosciences, 13, 6191-6210, 2016 significant difference in the normalized metabolic rates between individuals of the species S. subula and C. inflexa from the Atlantic and Pacific oceans.The comparatively lower metabolic rates from the Pacific may be a real response to the lower availability of O 2 for aerobic metabolism.Having a slower routine rate of O 2 consumption may be the result of a more efficient respiratory mechanism or an adaptation for coping with occasional exposures to the relatively high-CO 2 and low-O 2 conditions found in the northeastern Pacific Ocean.

Conclusions
Of the oceanic zooplankton species, thecosomes pteropods are thought to be some of the most sensitive to acidification.The responses we documented in the face of short-term CO 2 exposure and low O 2 reveal interesting patterns about basin scale differences in sensitivity, possibly associated with adaptation to local environmental conditions.Importantly, our results indicate that short-term exposure to high CO 2 does not have an effect on the respiration rate of multiple species of temperate and subpolar thecosome species from both the North Atlantic and Pacific oceans, irrespective of recent likely environmental exposure.The lack of effect of CO 2 as a single stressor on metabolic rate in adult organisms of various species has been seen in a number of studies (reviewed in Dupont et al., 2010;Kroeker et al., 2013), although there are many other metrics that have been shown to be more consistently affected.As such, thecosomes may have physiological coping mechanisms that allow them to maintain their energy budget for short periods of time in the face of high CO 2 via the reallocation of their energetic resources.Over longer time periods, however, this could reduce their scope for growth and reproduction, negatively impacting the fitness of the population as has been demonstrated with other marine calcifiers (i.e., Stumpp et al., 2011;Dupont et al., 2013;Melzner et al., 2013).Testing these hypotheses remains difficult as thecosomes are hard to maintain in captivity and there are no published studies of individuals kept fed and exposed to CO 2 in laboratory conditions for long durations (reviewed in: Howes et al., 2014;Thabet et al., 2015).Keeping individuals well fed is a critical factor since high food availability has been suggested to modulate the effect of high CO 2 exposure in both thecosomes (Seibel et al., 2012) and other calcifying species (Thomsen et al., 2013).Comparative shortterm studies of wild caught animals such as the present experiments, therefore, currently give us the best insight into the sensitivity of these open-ocean populations and the ability to predict how they will respond to the expected changes in the ocean environment.
Furthermore, although adult individuals may show no change in metabolic rate, there is evidence that juvenile stages of many calcifying species are typically more sensitive to CO 2 exposure (i.e., Connell et al., 2013;Waldbusser et al., 2015) and emerging evidence supports the idea that eggs, veligers, and juveniles of L. retroversa and L. helicina are more vulnerable to acidification than adults (Lischka et al., 2011;Thabet et al., 2015;Manno et al., 2016).Thus, although adults may be capable of surviving short-term exposure, as acidity in surface waters increases there may be population level stress due to ontogenetic sensitivity.
These findings also draw attention to the consequences of the high degree of vertical variability in the open ocean environment, with animals in the Pacific found migrating between deep waters, which are undersaturated with respect to aragonite, and the surface (Lawson, unpublished data; Maas et al., 2012b;Chu et al., 2016).Recent studies in the California Current system indicate that thecosome shells show signs of in situ dissolution when associated with waters that are undersaturated with respect to aragonite (Bednaršek et al., 2014;Bednaršek and Ohman, 2015).Although our short duration experiments do not directly address the effect of longer-term exposure to high CO 2 , it does remind us that as open ocean environments respond to anthropogenic change there may be vertical refugia from ocean acidification stress as well as regions where animals may already experience high CO 2 .As surface waters acidify, the ability to endure short-duration exposure and to migrate in both the Atlantic and Pacific populations may provide mechanisms for mitigating detrimental effects of acidification exposure.The potential compression of vertical habitat associated with the shoaling of the aragonite compensation depth, however, may have implications for predator-prey interactions, carbon pumping, and other ecosystem functions (Seibel, 2011;Bednaršek and Ohman, 2015).Furthermore, it is clear that thecosome shells are highly sensitive to dissolution (Comeau et al., 2012;Lischka and Riebesell, 2012;Manno et al., 2012) and there could be fitness and ecological consequences of dissolution in regions with vertical variation in carbonate chemistry.
Finally, as concerns about increasing CO 2 drive further explorations of comparative organismal physiology in the marine system, it is important to recognize that often the exposure of animals to increased CO 2 will occur in concert with expanding regions of low O 2 .This has been explored in the coastal environment where the interaction of acidification with eutrophication and associated low O 2 is comparatively well studied (Cai et al., 2011;Melzner et al., 2013) and is present in theoretical frameworks (Pörtner, 2010;Gruber, 2011;Sokolova, 2013).Experiments in the open ocean environment, however, are only beginning to be conducted and their implications explored.This study suggests that to make accurate predictions about how populations will respond to climate change and adequately understand the factors affecting organismal response, further investigations of the interactive effects of low O 2 and hypercapnia should consider nat-ural environmental variability, population biogeography, and phylogenetic sensitivity.

Data availability
Cruise data for the project are available via the National Science Foundation's Biological and Chemical Oceanography Data Management Office (BCO-DMO) under the project "Horizontal and Vertical Distribution of Thecosome Pteropods in Relation to Carbonate Chemistry in the Northwest Atlantic and Northeast Pacific" (http: //www.bco-dmo.org/project/2154).The raw data for the respiration experiments are included in this deposition (doi:10.1575/1912/6421).The raw data for the carbonate chemistry of the manipulations are included as a Supplement.
The Supplement related to this article is available online at doi:10.5194/bg-13-6191-2016-supplement.
Author contributions.Amy E. Maas and Gareth L. Lawson designed the experiments.All co-authors participated in oceanographic cruises and collection of samples.Amy E. Maas conducted all of the experiments and statistical analyses.Zhaohui Aleck Wang advised on the manipulation of carbonate chemistry and provided the measurements of both the hydrographic and experimental conditions.Amy E. Maas prepared the manuscript with contributions from both co-authors.

(Figure 1 .
Figure 1.Cruise tracks and animal sampling.Thecosomes were collected at night at stations along the main survey transect (solid line) and at stations during transit (dashed line) during cruises to the northwestern Atlantic in 2011 and northeastern Pacific in 2012.The shapes correspond to the species caught at each station and used in this study.Blue (10 • C), grey (15 • C), and red (20 • C) boxes around the station numbers (#) correspond to the temperature that was representative of 25-100 m at each station (Table2) and used in the experiments with animals from that station.

N
e li c in a C .p y r a m id a ta D .tr is p in o s a C . in fl e x a S .s u b u la L .r e tr o v e r s a C .p y r a m id a ta C .p y r a m id a ta C .p a c ifi c a C . in fl e x a S .s u b u la C . a tl a n ti c a

Figure 3 .
Figure3.Thecosome respirometry.Mean metabolic rate and standard error (µmol O 2 g −1 h −1 ) of thecosomes exposed to low (i.e., ambient) CO 2 and normal O 2 (light blue; LC/HO), high CO 2 and normal O 2 (dark blue; HC/HO), low CO 2 and low O 2 (light red; LC/LO), or high CO 2 and low O 2 (dark red; HC/LO).The species and temperature of the experiment are reported below the x axis.Significance is reported based on a basin, species, and temperature-specific GLM, which tested for the effect of treatment on O 2 consumption with a Bonferroni post hoc analysis (Table5).In the Atlantic analysis each treatment was tested independently, while in the Pacific CO 2 and O 2 were treated as factors.For each species and temperature, treatments are reported as non-significant (N.S.) or, in the case of significance, by letters that indicate which treatments are statistically similar (same letter) or different (different letter) at a p value < 0.05.Note that for C. atlantica the metabolic rates of individuals respired at 20 • C were converted to 15 • C using a temperature coefficient of 2 (see methods) for this GLM analysis.

Table 1 .
Environmental preferences and diel vertical migratory patterns for the species used in this study based on previously published data

Table 4 .
The average wet mass (mass; g) and mass-specific oxygen consumption rate (MO 2 ; µmol O 2 g −1 h −1 ) ± the standard error (SE) for each treatment (Treat.)and species.The numbers of replicates (N) per treatment are reported and the species are arranged by temperature (Temp; • C) as well as the year and basin of collection.
Figure 4. Log-transformed metabolic rates (µmol O 2 h −1 ) for L. retroversa at 10 • C, not normalized to mass, plotted against the log-transformed wet mass (mg) of individuals exposed to low CO 2 and normal O 2 (black circles; LC/HO), high CO 2 and normal O 2 (dark grey diamonds; HC/HO), low CO 2 and low O 2 (white circles; LC/LO), or high CO 2 and low O 2 (light grey diamonds; HC/LO).