Differential effects of ocean acidification and warming on biological functioning of a predator and prey species may alter future trophic interactions

9 Independently, ocean warming (OW) and acidification (OA) from increased anthropogenic 10 atmospheric carbon dioxide are argued to be two of the greatest threats to marine organisms. 11 Increasingly, their interaction (ocean acidification and warming, OAW) is shown to have wide-12 ranging consequences to biological functioning, population and community structure, species 13 interactions and ecosystem service provision. Here, using a multi-trophic experiment, we tested the 14 effects of future OAW scenarios on two widespread intertidal species, the blue mussel Mytilus edulis 15 and its predator Nucella lapillus . Results indicate negative consequences of OAW on the growth, 16 feeding and metabolic rate of M. edulis and heightened predation risk. In contrast, Nucella growth 17 and metabolism was unaffected and feeding increased under OAW but declined under OW 18 suggesting OA may offset warming consequences. Should this differential response between the two 19 species to OAW, and specifically greater physiological costs to the prey than its predator come to 20 fruition in the nature, fundamental change in ecosystem structure and functioning could be 21 expected as trophic interactions become disrupted.


INTRODUCTION
Ocean warming (OW) and acidification (OA) are arguably two of the greatest threats facing marine organisms as a result of increasing atmospheric carbon dioxide (CO 2 ) from anthropogenic sources (Shukla et al., 2019).Increasingly, their interaction (i.e.ocean acidification and warming (OAW)) has been shown to have wide-ranging consequences to the biological functioning of organisms including changes to physiology (Lemasson et al., 2018;Li et al., 2015), morphology (Knights et al., 2020), and behaviour (Manríquez et al., 2021) resulting in changes to population and community structure (Lemasson et al., 2018;Manríquez et al., 2021), inter-and intra-species interactions (Sadler et al., 2018), and the provision of ecosystem services (Listiawati and Kurihara, 2021).
Exposure to multiple stressors like OAW has been shown to be more biologically costly than a single stressor (e.g., temperature or pH; Gunderson et al., 2016); a scenario arguably more reflective of expected change in marine environments (Wernberg et al., 2012) than independent fluctuations in these metrics.While organisms can adapt to changes in the abiotic and biotic conditions where they occur (Alley, 1982;Jupe et al. 2020), this can come at a biological cost such as change in metabolic performance and fitness (e.g., Braby and Somero, 2006;Lemasson et al., 2018;Breitberg et al., 2015;Clements and Comeau, 2019).These costs may be detectable as an upregulation of metabolism (Lemasson et al., 2018;Matoo et al., 2013) or through increased O 2 consumption rates as individuals attempt to maintain homeostasis through physiological (e.g.cardio-circulation and the 'oxygen and capacity dependent thermal tolerance' concept, Pörtner 2012) or behavioural compensation (Giomi et al., 2016;Lemasson et al., 2018).
Increasing feeding may be one behavioural mechanism available to an organism to negate the negative effects of OAW (Clements and Darrow, 2018).But what remains unclear is the extent to which organisms can 'upregulate' feeding in response to associated increases in the metabolism, and whether this regulation can be maintained (Harvey and Moore, 2017;Lord et al., 2017).Indeed, in some cases, increasing energy intake may not be a viable option, such as when animals reduce feeding as an anti-predator response (Naddafi and Rudstam, 2013) which itself can indirectly result in modified biomineralization processes (Bibby et al., 2007), or changes in body size and reproductive output (Harvey and Moore, 2017;Lemasson and Knights, 2021).Predation is well known to be an essential driver of ecosystem dynamics (e.g., Sherker et al. 2017;Sadler et al. 2018) influencing prey population dynamics via both consumptive effects (CEs) and non-consumptive effects (NCEs) (Orrock et al., 2008).
Under OAW, changes in the magnitude of NCEs (Bibby et al., 2007;Clements and Comeau, 2019) and CEs (Sadler et al., 2018) during predator-prey interactions are predicted (see reviews: Briffa et al., 2012;Clements and Hunt, 2015), affecting physiological, morphological and behavioural mechanisms, J o u r n a l P r e -p r o o f iii as well as feeding strategies and induced defences (Lemasson and Knights, 2021;Manríquez et al., 2021;Sadler et al., 2018).However, the biological cost of reduced physiological performance in metrics like O 2 consumption and acid-base regulation may ultimately affect the extent to which organisms can respond to an external input and individuals may autonomously prioritise maintenance of internal homeostasis over a behavioural response (Bibby et al., 2007;Briffa et al., 2012;Harvey and Moore, 2017;Lord et al., 2017).
Susceptibility to OAW may be dependent on species and taxa (Briffa et al., 2012;Clements and Comeau, 2019) although calcifying species are shown to be particularly sensitive (Lemasson & Knights, 2021;Knights et al., 2020;Li et al., 2015;Sadler et al., 2018).Mytilus edulis (blue mussel) is a marine intertidal and subtidal bivalve most commonly distributed in the Atlantic Ocean in temperate regions (Knights, 2012), forming highly complex reef structures which support a multitude of other organisms.
In the UK, Mytilus spp. is an integral part of UK aquaculture and the national fisheries economy.The species is resilient to environmental perturbations but stressors like OAW may reduce their nutritional quality and fitness (Lemasson et al., 2019;Li et al., 2015).A major predator of M. edulis is the gastropod whelk, Nucella lapillus, (Hunt and Scheibling, 1998).Unlike bivalves, gastropods have been shown to be physiologically more resilient to OAW (Clements and Comeau, 2019) and mis-match in response to environmental change could lead to predator-prey relationships becoming unbalanced (Harvey and Moore, 2017;Sadler et al., 2018).
Given ocean acidification and temperature conditions are predicted to drastically change by end-ofcentury (Shukla et al., 2019) with potential consequences for the physiology and morphology of species and trophic interactions, here we evaluate the effects of elevated temperature and acidification scenarios on the performance and trophic interaction of Mytilus edulis and Nucella lapillus.Specifically, we test the effect of future climate scenarios on (1) individual physiological responses of M. edulis and N. lapillus including standard metabolic rate, feeding rate, changes in shell and somatic growth, and condition index of mussels (CI); and (2) the strength of trophic interactions between M. edulis and N. lapillus to assess potential changes in ecosystem functioning under future climate scenarios.

Animal collection and husbandry
Adult blue mussels (M.edulis) and adult dog whelks (Nucella lapillus) of a similar size were collected from a mid-shore intertidal site in Sidmouth, UK (50°40'41.1"N, 3°14'05.1"W) in April 2021.All animals were cleaned of epibiota and individually marked using a water-based non-toxic nail varnish J o u r n a l P r e -p r o o f iv (Acquarella (USA) which excludes toluene, formaldehyde, Dibutyl phthalate (DBP) and other solvents) and a permanent marker to allow for identification.Animals were acclimated for 2-weeks under standard laboratory conditions (12:12 h Light:Dark cycle, 15 °C, Salinity = 34 -36, pH 8 (with natural variation due to fluctuations in atmospheric pressure; see Lemasson et al. 2018 andKnights et al. 2020 for a full description)).Throughout acclimation and treatment, not including experimental starvation periods, mussels and whelks were fed twice weekly.Full water changes were conducted post feeding to maintain water quality (NH 3 < 0.5 mg L -1 ).Each N. lapillus was fed one opened mussel (M.edulis) (< 10 mm) and given 12 h to feed.Each M. edulis was given 1 h to feed on cultured Isochrysis galbana at a cell density of 24000 to 30000 cells mL -1 .

OA and Temperature Design
For the controls and OW treatments, air stones gently bubbling ambient air under atmospheric pressure were present in each tank.For OA, pure CO 2 was slowly released into a Buchner flask mixed with dry air (≈ 500 ppm pCO 2 ) using multistage CO 2 regulators (EN ISO 7291; GCE, Worksop, UK).pCO 2 levels were monitored using a CO 2 analyser (LI-820; LI-COR, Lincoln, NE, USA).pH was measured twice a week using a microelectrode (InLab® Expert Pro-ISM; Mettler-Toledo Ltd, Beaumont Leys, UK) J o u r n a l P r e -p r o o f v attached to a pH meter (S400 Seven Excellence; Mettler-Toledo Ltd, Beaumont Leys, UK), calibrated with Mettler Toledo buffers.
The experiment took place in a 15°C controlled temperature laboratory.Tanks under elevated temperature were kept in a water bath, with the temperature kept constant using aquarium heaters (thermocontrol e 200, EHEIM Jager GmbH and Co. KG, Stuttgart, Germany).
For predator cue treatments, two N. lapillus were placed in an individual perforated polypropylene plastic container to prevent predation of mussels and submerged in each tank for the duration of the experiment.Whelk density per tank is representative of Nucella lapillus densities on U.K. intertidal shores (Knights, unpublished data) and similar to densities found elsewhere (e.g.Hunt and Scheibling, 1998).

Carbonate chemistry
Total alkalinity (TA) was measured weekly using a calibrated potentiometric titrator (TitraLab AT1000© series HACH Company, USA).Weekly, a 50 mL sample was taken from each tank and tested to calculate TA.Temperature and salinity were taken in situ using a temperature probe (HH806AU, Omega, U.K.) and a handheld refractometer (S/Mill, Atago, Tokyo, Japan) respectively.TA, salinity, and temperature data were recorded to calculate calcite and aragonite saturation, and pCO 2 concentration in each treatment tank using CO2SYS software (Lewis and Wallace, 1998) using Mehrbach solubility constants (Mehrbach et al., 1973), refitted by Dickson and Millero (1987).
Seawater chemistry data are shown in Appendices Tables 1 and 2. 2.5.Morphological and physiological metrics 2.5.1.Body measurements and dry mass equation Body metrics and mass were recorded at three time points: (1) prior to experimental treatment exposure; (2) week 4; and (3) week 8.For M. edulis, length, width, and height were recorded.Wet weight was recorded using an analytical balance (Mettler Toledo, ML, Germany) after placing animals on paper towel for 15 min.Dry mass of M. edulis was estimated for each time point using the equation (eq. 1) from Knights (2012) as follows: .(1)  = 0.0508 0.9441 where x is shell length and y is total dry mass.
After 8-weeks, M. edulis were dissected and biometrics (length, width, height, total wet weight, wet tissue weight, shell weight, and dry tissue weight) were measured (see condition index).
J o u r n a l P r e -p r o o f vi For N. lapillus, wet weight was recorded by leaving animals out of water for 5 min and drying, then recording total weight to the nearest 1/100 th g using an analytical balance (Mettler Toledo, ML, Germany).Length from apex to siphonal canal was recorded using callipers.

M. edulis condition index
Body condition index (CI) of M. edulis was calculated using the following equation (eq.2) after Davenport and Chen (1987;BCI, eq. (1): ℎ ℎ ℎ × 100 The shell length of M. edulis was measured to the nearest 0.05 mm using callipers.Animals were dissected to remove all tissue from shell, which was placed into a pre-weighted plastic weighing boat to dry in an oven at 60°C.Tissue was weighted at 48 h and 72 h to ensure a constant mass (dry tissue) had been achieved and CI calculated from eq. 2.

O 2 consumption rate
Respiration rate was used as a proxy for Standard Metabolic Rate (SMR).Respiration rate was recorded using microfibre optic oxygen sensors (Fibox 4, PreSens Germany).Temperature and salinity were recorded prior to each set of data collection and barometric pressure was obtained from the Plymouth Live weather Station (http://www.bearsbythesea.co.uk).Each was input into the PreSens to allow O 2 measurements to be corrected for fluctuations in temperature, salinity, and pressure.
All M. edulis (n = 72; 9 per treatment) and N. lapillus (n = 36, 9 per treatment) were placed in 250 mL and 120 mL sealed jars, respectively.For the first respiration data point, sea water (salinity = 34 -36) was filtered to 2 μm and then autoclaved and aerated at 15°C.For time point 4 and 8, water was preequilibrated to the appropriate treatment conditions.To maintain stable temperature, during data collection, jars were kept in a water bath at 15°C or 20°C.All animals were starved for ~8 days prior to data collection to eliminate any change in respiration due to digestion and alter respiration rates (Sejr et al. 2004, Ansell & Sividas 1973).Within the jar, water was mixed using a magnetic stir bar for the duration of the experiment (400 rpm).Data collection started when jars were closed.For M. edulis, data points were only counted if the animal was visibly open.All data points before 15 min were discounted for both animals to allow for acclimation.O 2 (mg L -1 ) was recorded every 5 min for 40 min or until O 2 saturation reached 75 % to avoid exposure to hypoxic conditions.O 2 measurements were corrected for background bacterial respiration or primary productivity by offsetting respiration rate with O 2 changes in jars without an animal in them.Respiration rate was also normalised to 1 g of calculated dry weight (Knights, 2012).SMR was calculated using the following equation (eq.3).
J o u r n a l P r e -p r o o f where v is volume of jar (L), r is change in O 2 in jar (mg L -1 ), t is time (min), and DM is dry mass (g) calculated using the relationship defined in Knights (2012).

M. edulis clearance rate
The same individuals used for respirometry were also used for clearance rate (CR).M. edulis were starved for 24 to 72 h.The CR assay followed methodology in Lemasson et al. (2018).Individuals were placed in 300 mL of UV treated and filter sea water (15 °C, 500 ppm pCO 2 , salinity = 34 -36) and subsequent data points were recorded in water pre-equilibrated to treatment conditions.A dilution of 1:100 mL shellfish diet (Shellfish diet 1800, Reed Mariculture, USA) was used as feed.M. edulis were given up to 20 min to open and algae added once opened.Any animals which closed during the assay were discounted and re-done the following day.Once open, 700 μL of stock solution was used per beaker at a concentration of 24,000 to 30,000 cell mL -1 .In each beaker, a magnetic stirrer bar (400 rpm) was used to keep the water well-mixed.A 20 mL sample (t 0 ) was taken 2 min after stock solution was added to allow for adequate mixing of algae.Another 20 mL sample (t 1 ) was taken after 20 min of filtering.Counts of the algae in the water were done in triplicate by a Coulter Counter (Beckman Coulter, Z2).CR was calculated using the following equation (eq.4).where CR is clearance rate (L h -1 ), v is volume (L), t 0 is the initial sample (cell L -1 ) and t 1 is the sample (cell L -1 ) taken after 20 min.CR was then normalised to 1 g by dividing by calculated dry mass of individual (Knights, 2012).

Feeding behaviour
Feeding behaviour of N. lapillus was assessed under treatment conditions to look at both predator risk of M. edulis and feeding rate of N. lapillus.After 8-weeks of exposure to the experimental treatments, N. lapillus were starved for 7 to 9 d.M. edulis used in the experiment were pre-acclimated for 8 weeks in each of the experimental treatments.Five pre-acclimated M. edulis (20 to 45 mm length) were placed in each tank, with 11-12 tanks in each of the four treatments.There were control tanks (n = 3) included in each treatment which contained only M. edulis.N. lapillus were placed into tanks and mortality was measured every 24 h over 8 d.Mussels were considered dead when they gaped open and did not respond when physically disturbed (Lupo et al., 2021).In tanks without predators there was one mortality in the elevated temperature and pCO 2 treatment (6.7 % mortality).
J o u r n a l P r e -p r o o f viii 2.7.Statistics Data were tested for assumptions of normality, bias and homoscedasticity of residuals.Data were logtransformed or square-root transformed if data did not meet assumptions.All data were analysed using R (version 4.1.1,R Core Team, 2021) and all graphs were produced using the 'ggplot2' package (Wickham, 2016).Where significance was identified Tukey HSD post-hoc pairwise comparison was used to find differences between groups.'Tank' was included as a random factor in all analyses. 2.7.
M. edulis mortality was significantly higher in the OAW treatment over all other treatments (p < 0.01, F 4,67 = 4.382).

J o u r n a l P r e -p r o o f
ix There was a significant interaction between OW, OA and predator presence on shell length (p < 0.05, F 1,61 = 5.635) and a significant reduction in growth when predators were present (Fig 1).Shell length increased by 159 %, from an average increase of 0.24 mm, in the presence of predators, to 0.63 mm, when predator cues were absent (Tukey HSD; p < 0.001).There was no effect of OW alone (Tukey HSD; p = 0.438) or predator presence alone (Tukey HSD; p = 0.300) on growth in length.
There was a significant interaction between OW and OA on height (p < 0.01, F 1,53 = 7.420).There was a 113 % increase in shell height compared to the control under OA increasing by 0.40 mm under OA, versus just 0.19 mm under ambient conditions (Tukey HSD; p < 0.01).There was no effect of OAW on height (Tukey HSD; p = 0.415).There was also an interaction between cue presence and OA (p < 0.05, F 1,53 = 5.420) on mussel height with a 175 % increase in height under OA compared to the control (Tukey HSD; p < 0.01).There was no effect of OA on height when predators were present (Tukey HSD; p = 0.574).

M. edulis
There was a significant interaction between OW and OA on SMR in M. edulis (p < 0.05, F 1, 194 = 4.44)(Fig.2).O 2 consumption rates increased in an additive fashion by 18.8 % under OA.OW increased the SMR J o u r n a l P r e -p r o o f x of M. edulis by 33 %.However, there was no effect of OA on SMR under OW.There was also an interaction between time in treatment and cue presence (p < 0.05, F (2,194) = 3.399).In the absence of cues, SMR decreased by 16 % from week 0 to 4 and remained the same from week 4 to 8. SMR in response to predator presence was maintained until week 4, and from week 4 to week 8, SMR reduced by 31 %.

DISCUSSION
OAW impacts are being documented ubiquitously across marine taxa and marine ecosystems with wide ranging variable effects and complex interactions between pH and temperature stressors (e.g., Clements and Hunt, 2015;Knights et al., 2020;Kroeker et al., 2013).In this study, the impacts of future J o u r n a l P r e -p r o o f xi predicted OAW on growth and physiology have been highlighted in two major marine invertebrate taxa, mussels and dog whelks.Further investigation elucidated the impacts to the predator response of M. edulis and effects of OAW on the predator-prey relationship between these species.Results indicate significant effects of OA, OW, OAW, and predator presence on growth, CR and SMR in M.
edulis.Less pronounced effects on growth and SMR were seen in N. lapillus, alongside an increase in predation rate under OAW, indicating increased predation risk to M. edulis.

Growth and condition
The effect of OAW on shell and somatic growth in marine invertebrates appears highly species dependent (Gazeau et al., 2013;Kroeker et al., 2013;Lemasson et al., 2018;Lemasson and Knights, 2021).For M. edulis, OA was found to increase shell growth (length and height) alongside an increase in SMR.In N. lapillus, there was no effect of OAW or individual effects of OA or OW on growth in shell length or growth in wet weight (but see Mayk et al. 2022 where shell growth was shown to increase under OA).Increased shell growth in M. edulis may be explained in terms of carbonate chemistry.For example, M. edulis biomineralize using two different forms of calcium carbonate, a mixture of calcite (~17 %) and aragonite (~83 %) (Hubbard et al., 1981).Aragonite has a greater dissolution rate to calcite under OA conditions (Feely et al., 2004).Therefore, dissolution of the shell under lower pH may lead to mineralogical plasticity in biomineralization, as seen in this study, despite some evidence for a net decrease in calcification rate under OA (Leung et al., 2017;Li et al., 2015).There are variable effects of OAW on shell growth in the literature with the majority of the literature reporting negative impacts on growth (e.g., Fitzer et al., 2015;Lemasson and Knights, 2021).Despite this, we observed an increase in length of M. edulis under OA.However, the literature shows that animals calcifying under OA may prioritise investment in lower quality shell structure (i.e.greater size, weaker shell; Leung et al. 2022), which consequently may increase predation risk (Gazeau et al., 2013;Li et al., 2015;Sadler et al., 2018).
Environmental stressors can interact to influence the overall effect of a stressor on an organismal trait (Kroeker et al., 2017).The increase in growth under OA was counteracted under elevated temperature or in the presence of predator cues indicating an antagonistic relationship between these variables and biomineralization traits.In the presence of predators, mussels can induce calcification to increase shell thickness as an anti-predator response.This upregulation of calcification is a common nonconsumptive effect (NCE) of predators within a prey population (Freeman, 2007).However, under environmental stress the cost of upregulating calcification increases, particularly under OA as shell dissolution increases (Nienhuis et al., 2010).Mussels may be calcifying at the same rate but reallocating the energy used to prioritise shell thickness over shell size as an anti-predator defence J o u r n a l P r e -p r o o f xii strategy.Shell thickness has been shown to decrease under OA conditions over a relatively short time scale (8 weeks) in M. edulis (Fitzer et al., 2015;Sadler et al., 2018).Alongside this, net calcification rate has been reported to decrease in mussels under elevated pCO 2 (Li et al., 2015).Within this study, shell thickness and net calcification rate were not recorded, however, based on the literature, we predict there was likely a trade-off between structural integrity and size of shell mussels exposed to OA (Fitzer et al., 2015;Knights et al., 2020;Sadler et al., 2018).

Condition index
CI is used to comparatively assess the reproductive condition of mussels between treatments (Knights, 2012).In this study, perhaps surprisingly, CI increased under predator presence.Given the CI calculation uses shell length and dry tissue weight, this suggests the animals are investing more in somatic tissue than length as length change did not differ between OAW treatments when cues were present.OAW had no effect on CI in mussels despite evidence suggesting otherwise in the literature.
For example, temperature increase (Sunila, 1981) and enhanced food availability (Hatcher et al., 1997) both led to an increased metabolism which resulted in a greater CI of Mytilus sp.. Low pH was also found to increase condition index in M. californianus (Rose et al., 2020).On the other hand, Lemasson and Knights (2021) found effects of OAW on CI to be species-specific and found no effect of OAW on CI in European flat oysters (Ostrea edulis).The results suggest that M. edulis may be prioritising reproduction and fecundity over long term survival.A similar finding was shown in Daphnia magna, which displayed greater investment in fecundity under size selective predation pressure (Zhang et al., 2016) and also in M. edulis, where gonad development was accelerated when exposed to starfish cues (Reimer, 1999).

Metabolism
Maintaining metabolic rate in response to energetic demand is essential for survival and basic functions like growth and feeding (Gazeau et al., 2013).Metabolism is closely linked to temperature, particularly in marine ectotherms (Seibel and Walsh, 2003).O 2 consumption rates (SMR), increased under OAW by 33 % for M. edulis, in concordance with previously reported increases in SMR in bivalves under OAW (Lemasson et al. 2018).The upregulating effect of low pH on SMR in mussels at 15 °C was, however, masked by elevated temperature.pH had no additional effect on SMR alongside elevated temperature.Similar results have been found in M. edulis, where temperature is the dominant factor in influencing SMR and addition of low pH stress does not affect the SMR response (Lemasson et al., 2018;Matoo et al., 2021).However, the increased SMR of M. edulis in response to OA at 15 °C is not well documented.OA exposure puts physiological stress on the internal homeostasis on an organism; energetic demand for acid-base regulation increases as pH of internal fluids lowers (Gazeau et al., J o u r n a l P r e -p r o o f xiii 2013).The metabolic upregulation seen in M. edulis was not evident in N. lapillus, indicating a greater resilience of N. lapillus to OA and supports the suggestion that some species of gastropod are more resilient to OA than bivalves (Clements and Comeau, 2019).
Change in physiology in response to an external stimulus (i.e.predator presence or OAW) can result in metabolic depression in animals over time (Gazeau et al., 2013;Seibel and Drazen, 2007).In this study, time in treatment and predator presence interacted to induce metabolic depression in M. edulis after just 4 weeks of exposure and may be explained by anti-predator response strategies.Animals respond in different ways to predators depending on their mobility.Mobile animals may upregulate the metabolism to escape a predator, immobile animals, such M. edulis, may downregulate the metabolism to reduce predator contact through processes such as feeding (Gazeau et al., 2013;Seibel and Drazen, 2007).Alongside these findings, metabolic depression resulting from predator exposure under OA has been observed in mussels (Brachidontes pharaonis) exposed to crab predator cues (Eriphia verrucosa) (Dupont et al., 2015).The metabolic depression of M. edulis seen in this study may have resulted from a reduced feeding rate when predator cues were added (i.e.reduced energy acquisition), coupled with increased physiological stress of exposure conditions (i.e.offsetting shell dissolution and maintaining acid-base homeostasis) (Gazeau et al., 2013;Seibel and Drazen, 2007).M.
edulis may have the capacity over short time scales (< 4 weeks) to maintain physiological performance under climate change stressors thereby compensating using trade-offs.However, over a longer time scale (> 4 weeks), in the presence of predators, M. edulis have a reduced metabolic performance which may be unsustainable and fitness-reducing as less energy is available for other physiological processes (Gazeau et al., 2013).

Clearance rate
Clearance rate (CR) is a semi-quantitative measurement and can be used as a measure of physiological or behavioural performance (Lemasson et al., 2018) and is closely linked to metabolic processes so that it can be used to balance energy acquisition and expenditure (Giomi et al., 2016).Increase in metabolic rate, from OAW, can be an issue if energy acquisition does not also increase.Here, a complex interaction was found between temperature, pCO 2 , predator presence, and time in treatment.Despite increases seen in SMR as a result of OA and OW exposure, food intake (CR) did not increase under the same scenarios.Food availability or intake is a known limiting factor of animal resilience to OAW stressors (Clements and Darrow, 2018).Therefore, energetic requirements may not have been met, resulting in the decreasing trends seen over time in SMR.
From a behavioural perspective, feeding is a behaviour that increases the predation risk of an animal and can be downregulated by the animal accordingly (Křivan and Eisner, 2003).Here, cue presence J o u r n a l P r e -p r o o f xiv led to much greater change in CR than OAW scenario, resulting in reduced feeding of the animals when cues were present in all treatments but elevated temperature.Under elevated temperature, with predators present, feeding rate of M. edulis did not decrease as expected, suggesting a potential trade-off or 'decision' to prioritise physiological demand over predator behavioural response (Briffa et al., 2012).While this may reduce fitness in relation to predation risk, it has potential to work as a compensatory mechanism for OW as more energy is acquired to offset negative impacts of OW (Giomi et al., 2016).Animals may upregulate feeding to maintain physiological processes despite greater predation risk.This removes or alleviates food intake as a limiting factor for animal wellbeing under OAW scenarios (Clements and Darrow, 2018).
In M. edulis, reduced feeding rate as an anti-predator response in conjunction with the increased energy requirement observed in individuals exposed to OAW illustrates a clear juxtaposition between behavioural and physiological responses when relating to fitness maintenance.When exposed to OAW and predation, energy intake falls short of energy expenditure as seen in eventual metabolic depression.On the one hand, reduced feeding rate when exposed to cues, under OAW, indicates the mussels behavioural response (e.g., cue perception) is not impaired (Clements and Comeau, 2019).
On the other, this may be detrimental to future adaption as it shows the animal is prioritising behavioural rather than physiological mechanisms of survival.

Species-specific differences and interactions
Molluscs, as a taxa, have been shown to be particularly sensitive to OAW in terms of survival, calcification, growth, and development, compared to crustaceans, fish, and algae (see review : Kroeker et al., 2013).Within the taxa, the effects of OAW exposure are often species-specific as illustrated in the growth and mortality differences found here.N. lapillus had no significant response to OAW scenarios either in length or wet weight whereas growth rate increased for M. edulis, in length and height, and decreased for wet weight.Mortality was also significantly greater in M. edulis under OAW conditions, whereas N. lapillus mortality was unaffected.The disparity between OAW responses in bivalves and gastropods has been documented in behavioural defences (see review by Clements and Comeau, 2019) with bivalves more sensitive than gastropods to OAW illustrated by predator avoidance behaviour (e.g bivalves: Clements et al., 2017;gastropods: Queirós et al., 2015).
Behavioural responses to OAW can be indicative of physiological underlying effects of OAW, such as impacts to metabolism or growth (Clements and Comeau, 2019;Gazeau et al., 2013).An increased susceptibility of bivalves to OAW over their predators may lead to incongruity in their biotic relationships.That is, if there is greater biological cost to the bivalve and no change in cost to the J o u r n a l P r e -p r o o f xv gastropod predator as shown here, this may suggest potential for modification of predator-prey dynamics and wider trophic impacts.
Predation rate is a key driver in ecosystem dynamics (Holling, 1959).Change in a predator's feeding rate has potential to destabilise lower trophic levels (Kroeker et al., 2017).In this study, elevated temperature increased N. lapillus feeding rates by 82 %.In the literature, Quieros et al. (2015) reported that N. lapillus foraging distance and foraging time increased under OA suggesting an increased feeding rate.However, this may increase their own susceptibility to predation from higher trophic levels (Křivan and Eisner, 2003).Nevertheless, increased feeding of N. lapillus as a result of elevated temperature coupled with the negative impacts of OAW on M. edulis, and increased SMR but reduced feeding rate, could negatively affect M. edulis populations.Predation risk of prey animals is reported to increase in bivalves under OA regardless of predator exposure to OAW stress (Sadler et al., 2018;Sanford et al., 2014).Increased predation leading to greater consumptive effects of N.
lapillus on M. edulis may have knock-on consequences to ecosystem services and wild mussel fisheries (Lemasson and Knights, 2021;Sadler et al., 2018).However, local ecosystem effects may vary depending on functional redundancy (i.e.biodiversity) within a community and plasticity of the populations affected (Kroeker et al., 2017).Investigating OAW with predation as a stressor adds ecological relevance to a study and help elucidate the interacting effects of OAW in an ecologically relevant setting (Kroeker et al., 2013).

CONCLUSION
The relative biological cost of OAW impacts individual animal fitness and will reflect into the population.The two species, M. edulis and N. lapillus, had contrasting responses in terms of growth, metabolism, and feeding to OAW exposure.In addition to this, the species chosen are ecologically linked in marine ecosystems, therefore impacts to one will affect trophic relationships (Holling, 1959).
Here, M. edulis demonstrated greater effect sizes from OAW exposure than N. lapillus.The interaction between the two species also changed under OAW exposure, exhibited though change in both NCEs (e.g., reduced CR in M. edulis) and CEs (e.g., increased predation rate of N. lapillus).The differential responses of the two species and the increased feeding rate seen in N. lapillus indicates that under future climate change scenarios, M. edulis may experience greater predation risk alongside physiological implications whereas in contrast, N. lapillus may largely be unaffected if food is not limited.This could lead to shifts in ecosystem functioning and services depending on the functional redundancy within the ecosystem and susceptibility of different species to OAW (Kroeker et al., 2017).