Ocean Acidification and Increased Temperature Have Both Positive and Negative Effects on Early Ontogenetic Traits of a Rocky Shore Keystone Predator Species

Conditioned FSW with the two different pCO₂ levels for the treatment chambers was generated in 4 polyethylene (230 L) reservoir tanks (see [34]). Each chamber was filled with conditioned FSW, and a continuous stream of either air (ca. 400 μatm CO₂) or enriched CO₂ air (ca.1200 μatm CO₂) was bubbled through the water. Enriched CO₂ air was produced by blending air and pure CO₂ using mass flow controllers (see a full description of this type of seawater pCO₂ control system in Torres et al. 2013 [34]. During the first rearing phase the seawater parameters (i.e., pH, temperature, salinity and total alkalinity) were measured twice a week in the seawater used to replace the water in the rearing aquarium (Table 1). However, in the treatment phase the parameters were measured twice a week in water samples taken from the rearing chambers before the water change took place. Therefore, these measurements represent the values achieved with the continuous stream of either air or enriched CO₂ air plus the cumulative effect on the carbonate system of the 24 h of respiration, calcification and ammonium excretion associated with the metabolism of both trial snails and the prey in the 1.5 L chambers. The pH measurements were made in a closed 25 ml cell, thermostatically controlled at 25.0°C, with a Metrohm 713 pH meter (input resistance >10¹³ Ohm, 0.1mV sensitivity and nominal resolution 0.001 pH units) and a glass fixed ground-joint diaphragm electrode with Pt1000 (Metrohm 6.0257.000, Aquatrode plus) calibrated with 8.089 Tris buffer at 25.0°C [35]; pH values are therefore reported on the total hydrogen ion scale [35]. Total Alkalinity (A_T) was determined by potentiometric titration in an open cell, according to Haraldsson et al. [36]. The accuracy was controlled against a certified reference material supplied by Andrew Dickson (Scripps Institution of Oceanography). The correction factor was approximately 1.002, corresponding to a difference of < 5μmol kg^-1. Each sample was analysed using 2 or 3 replicates. Temperature and salinity were measured using a CTD (Ocean Seven 305). The pH, A_T and hydrographic data were used to calculate the rest of the carbonate system parameters (pCO₂, total alkalinity, pH in situ, pCO₂ in situ and [CO₃^-2]) and the saturation state of Ω aragonite and calcite, using CO₂SYS software [37] set with Mehrbach solubility constants [38] refitted by Dickson & Millero [39].

To estimate the effects of OA and warming on growth; shell size (the maximum length at the peristomal margin of the shell aperture or peristomal length), live wet weight and calcification rate via change in buoyant weight were measured at the beginning and end of the acclimatisation phase (2.1 months), and then at the end of the treatment phase (5.8 months). The differences in these variables were then converted in to percentage change over the course of the experimental period (specific growth rate). To estimate the effects on shell thickness we measured thickness at the base of the labral tooth that in C. concholepas forms a projection on the very end of the outer shell lip. This shell projection was selected as it can be measured without the need to euthanize the experimental individual and because in C. concholepas it is easily distinguishable and representative of the shell thickness at the growing edge (PHM personal observations). To standardise for shell size we computed a thickness index by dividing labral tooth thickness by the body size. The potential corrosive effect of the experimental conditions was evaluated by exposing 12 empty shells of small juveniles (ca. 3.0 cm) to each of the treatments (3 shells per treatment). Shell weight as a function of pCO₂ and temperature levels was measured at the beginning and end of the exposure period (1 month). Since different rearing methods were used in each phase, in terms of seawater conditioning, changes in size and weight during the acclimatisation phase are reported but not included in the statistical analyses.

After 3 months of exposure in the treatment phase 3 behavioural traits were measured: (1) predation avoidance (2) self-righting time; the time span from the moment that an individuals is placed upside down to when it returns to the normal upright position. Finally, after an additional 1 month of exposure (3) the minimum force (tenacity) required to dislodge the individuals from the substratum was measured.

After 3 month in the treatment phase the individuals were removed from the chambers and allowed to acclimatise for 2 h to the experimental chambers. The self-righting measurements were conducted in a plexiglass chamber (19 × 10 × 6 cm; length, width, height) divided in to two equal parts by plastic mesh and filled with 0.5 L of 1 μm FSW equilibrated to the same temperature and pCO₂ levels used during the treatment phase [see [15]. To accelerate the self-righting process a predatory crab Acanthocyclus hassleri [15,22,40] was placed on one side of the chamber and the experimental individuals on the other. The crab was introduced to the chamber 20 min before the test individual and a small air stone was used to circulate the crab-odour cues throughout the chambers. The air stone was then removed and the test individual carefully placed in the middle of the other side of the chamber from the crab and immobilised inside a piece of PVC tubing. After 10 min of acclimatisation the pipe was removed and the test individual was placed upside down. Self-righting time was considered as the total time span from the moment that the individuals were placed upside down to their return to the normal upright position.

The predation avoidance behaviour was observed in the same chambers described above. The snails were removed from the treatment conditions, and placed in the experimental chamber adjacent to the dividing mesh and immobilised inside a piece of PVC tubing. After 20 min of acclimatisation in the chamber a crab was introduced into the other side of the chamber and the PVC tubing removed to allow the snail to move freely. The final position in the chamber of the C. concholepas after 15 min of observation was recorded, with those snails moving to the wall of the chamber furthest away from the crab being considered to have displayed predator avoidance behaviour.

The measurements were conducted after 4 months of exposure in the treatment phase, in a specially designed plexiglass chamber. The chamber (Fig 1b) consisted of a rectangular box (16 × 8 × 7.6 cm; length, width, height), along with two plexiglass flaps joined at 90° to each other and suspended from the top of the chamber from the axis or fulcrum created by the intersection of the two flaps. The longer of the two flaps (11.8 cm) hangs vertically in the chamber and the shorter (5.2 cm) parallel to the base of the chamber. A vertical force applied to the horizontal flap results in the rotation of the vertical flap applying a horizontal force across the base of the chamber. The chambers were filled with 0.3 L of 1 μm FSW equilibrated to the same temperature and pCO₂ levels used during the treatment phase. The test individual was positioned in the base of the chamber with the left-hand edge of the shell in contact with the vertical flap. Temporary plastic barriers were used to prevent the snail from moving away and the snail it was given 20 min to attach itself to the base of the chamber. After the 20 min period the barriers were removed and the dislodgement force applied to the shell. To measure the minimum force required to dislodge the test individuals a vertical force was carefully applied with a digital push-dynamometer (PCE FM50) on the left side of the horizontal flap (Fig 1b) generating a horizontal force at the base of the vertical flap, which pushes against the shell of the test individuals. The push-dynamometer is able to measure a maximum force of 49 N (ca. 7.5 k), and the force applied can be held constant throughout the trial. As a general rule the snails with a larger foot area require more force to dislodge [41], therefore dislodgement force was standardised by pedal surface area. This was measured by tracing the outline of each attached foot onto a piece of glass, the outline was then transferred to paper, and a digital planimeter was then used to calculate the area of the foot.

The metabolic rate measurements were made using small juveniles of C. concholepas after 4.8 (first series, 6–7 individual per treatment) and 5.8 months (second series, 5 individuals per treatment) in the treatment phase. All metabolic rates were measured individually in respirometric chambers (100 ml Schott Duran^® glass bottles) after incubation in each pCO₂/temperature treatment. In both series, the seawater used for metabolism measurements was 0.45 μm FSW fully O₂ saturated. The first series used standardised pCO₂ (current-day pCO₂ levels) and temperature (15°C) levels during the oxygen measurements. Individuals were starved for 24 h in 1 μm FSW prior to the experimental measurements, sufficient time to ensure that the metabolic rate was not affected by the last meal [15]. The first series was used to determine the shock response of the rearing conditions on metabolic rate. For the second series, individuals were again starved for 24 h and oxygen consumption was measured at the same four temperature and pCO₂ combinations used for the rearing treatments. The second series was used to determine the long-term metabolic consequences of acclimatization to temperature and pCO₂. A fibre optic oxygen-meter (Fibox 3, PreSens, Germany) was used to measure oxygen consumption (mg O₂ g⁻¹ h⁻¹), zero calibration was performed using a Na₂O₃S solution (0% saturation) and 100% was calibrated using air-bubbled seawater. The temperature was stabilised using a temperature-controlled circulation bath. Individual measurements lasted for at least 60 min per individual, with the first ten minutes of data being eliminated to reduce the effects of manipulation stress on the final analyses. Special care was taken to prevent the oxygen levels from dropping below 70% of saturation to minimize stress effects [42]. Background respiration was determined from control experiments using chambers with treatment seawater but no organism. Oxygen consumption rates and background respiration were calculated using the linear decrease in oxygen concentration measured over intervals of between 30 and 90 min.

Two-way ANOVA were used to evaluate whether temperature and pCO₂ and the combination of these factors had an effect on the studied variables [43]. Where statistically significant effects were identified the ANOVA was followed by a Tukey's a posteriori HSD test to identify the significance of differences between pairs of conditions [44]. Differences in the proportion of individuals displaying avoidance behaviour after rearing at current-day and elevate pCO₂ levels for each rearing temperature were analysed with a χ² test. All statistical analyses were conducted using R version 3.0.2 [45]. Tenacity index and metabolic rates data (standard conditions) were log transformed, and metabolic rates data (treatment conditions) were rank transformed [46], to meet the assumptions of normality.

During the 2 months acclimatisation phase the survival rate was 97.5%. Ten individuals were exposed to each combination of pCO₂ and temperature, except for the current pCO₂/high temperature treatment where only 9 individuals were exposed, due to the loss of one test organism during the acclimatisation phase. During the treatment phase survival was high for all treatments, with the lowest survival rate of 88.9% being recorded in the current pCO₂/high temperature treatment.

Average shell growth estimated from changes in body size was significantly higher during the acclimatisation phase than during the treatment phase at current-day pCO₂ levels and at both temperatures (1-way ANOVA, F_2,27 = 17.518, p < 0.05; Fig 2a). During the treatment phase pCO₂ level did have a significant effect on shell growth (2-way ANOVA p < 0.05; Fig 2a; Table 2). At 15°C, elevated pCO₂ levels caused significantly reduced growth as measured by changes in body size (Tukey's HSD test p < 0.05; Fig 2a). However, no significant differences were found at 19°C (Tukey's HSD test; p > 0.05; Fig 2a; Table 2). During the treatment phase there was no significant effect on shell growth due to temperature and pCO₂ levels in combination, nor by temperature alone (Fig 2a; Table 2).

Average growth estimated from changes in total wet weight was significantly higher during the acclimatisation phase than during the treatment phase at current-day pCO₂ levels and at both temperatures (1-way ANOVA, F_2,26 = 2.212; p < 0.05; Fig 2b). Again, during the treatment phase, pCO₂ level did have a significant effect on total wet weight (2-way ANOVA p < 0.05; Fig 2b; Table 2; Fig 2b). However, only the post hoc comparison between current-day pCO₂ at 15°C and future pCO₂ at 19°C was significant (p < 0.05); the differences in total wet weight between the pCO₂ levels within the two temperature treatments were not significant (Fig 2b). During the treatment phase there was no significant effect on growth due to temperature and pCO₂ level in combination nor by temperature alone (Fig 2b; Table 2).

Shell calcification of live individuals as measured by changes in buoyant weight during the acclimatisation phase was not significantly different to that recorded during the treatment phase at current-day pCO₂ levels and at both temperatures (1-way ANOVA, F_2,17 = 1.207; p > 0.05; Fig 2c). However, during the treatment phase at both temperatures shell calcification was significantly lower under future pCO₂ levels (Tukey's HSD test p < 0.05; Fig 2c). During this phase temperature and pCO₂ in combination had no significant effect on total shell calcification (Fig 2c; see Table 2).

There was no significant effect of temperature, pCO₂ levels nor the interaction between these two variables on the shell thickness index of live individuals (2-way ANOVA p > 0.05; Fig 2d; see Table 2). No significant differences were found in the dry weights of the empty shells at the beginning of the rearing period (1-way ANOVA; F_3,8 = 0.04; p > 0.05; Table 3). However, after 1 month of exposure to the experimental treatments empty shell weight loss was significantly increased by temperature (2-way ANOVA; F_1,8 = 6.014; p < 0.05), pCO₂ (2-way ANOVA; F_1,8 = 22.752; p < 0.05) and the interaction between pCO₂ level and temperature (2-way ANOVA; F_1,8 = 11.347; p < 0.05; Table 3). At both rearing temperatures more shell material was lost in those empty shells exposed to elevated pCO₂ levels (1-way ANOVA, F_1,4 = 18.516 p < 0.05; 15°C; 1-way ANOVA, F_1,4 = 8.221; p < 0.05; Table 3).

Irrespective of the temperature, predation avoidance behaviour was observed in all individuals exposed to current-day pCO₂ levels. Individuals exposed to elevated pCO₂ levels exhibited a reduced frequency of predation avoidance at 15°C (χ²₍₁₎ = 10.77; p < 0.05) and 19°C (χ²₍₁₎ = 4.56; p < 0.05) (Table 4). During the observation period (15 min) the individuals reared at elevated pCO₂ levels tended to remain motionless or displayed short and random movements.

Self-righting was observed in individuals exposed to all of the experimental treatments (Table 4). There was no interaction between temperature and pCO₂ for self-righting nor for temperature alone (Table 4). At elevated pCO₂ levels self-righting times were significantly faster in individuals reared at high pCO₂ levels at both 15°C (Tukey's HSD test; p < 0.05) and 19°C (Tukey's HSD test; p < 0.05; Fig 3a). Temperature alone and temperature and pCO₂ in combination had no effect on self-righting times (Table 4).

Between 190 and 1200 grams per cm² of pedal area were required to dislodge the individuals (Fig 3b). At 15°C significantly more force was required to dislodge individuals at high pCO₂ levels compared to low pCO₂ levels, the same result was observed at 19°C. At high pCO₂ levels, significantly more force was required to dislodge individuals at 19°C compared to 15°C. However, at low pCO₂ levels the force required to dislodge individuals was not significantly different between temperatures. The combination of temperature and pCO₂ had no significant effect on tenacity (Table 5).

The results of the first series of measurements indicate a reduced metabolic rate in those individuals reared under high pCO₂ conditions at both temperatures. Temperature and pCO₂ level individually had a significant effect on metabolic rate (2-way ANOVA; p > 0.05; Table 6). At both rearing temperatures the metabolic rate was significantly reduced at elevated pCO₂ levels. (Tukey's HSD test; p < 0.05; Fig 4a). The combination of temperature and pCO₂ level had no significant effect on metabolic rate (Table 6). As expected, given the thermodynamic principles for ectothermic organisms, the second series of metabolic rates were higher at the higher temperature (Fig 4b). Only temperature had a significant effect on the metabolic rate (2-way ANOVA; p < 0.05; Table 6). The pCO₂ level alone and the combination of pCO₂ and temperature had not a significant effect on the metabolic rate (2-way ANOVA; p > 0.05; Table 6).

The observed higher average body size change during the acclimatisation phase compared to the treatment phase at 15°C in the present study agrees with previous observations and presumably reflects the characteristic decrease in growth rate after the fast exponential growth normally observed during the early ontogeny of this species [47]. Total weight change, calcification and shell thickness were not significantly affected by temperature. This suggests that within the investigated thermal range (15 to 19°C) this species can cope with the temperature increase. This study has also demonstrated that irrespective of the treatment temperature calcification in C. concholepas was negatively affected by high pCO₂ levels. However, the combination of high pCO₂ and temperature had neither a synergistic nor antagonistic effect. Thus the observations of previous studies [48] in which the negative effects of elevated pCO₂ levels were reduced by increases in temperature, were not observed in the present study. The presence of significantly reduced shell calcification at high pCO₂ levels and the absence of a significant combined effect of pCO₂ and temperature on shell calcification has also been reported in the mussel Mytilus chilensis [17]. The lack of interactive effects of pCO₂ and temperature levels in C. concholepas could mean that the temperature range used in the present study was not sufficient to induce negative or synergistic effects.

Live shells of molluscs do not behave as empty shells do because many calcifying species are able to upregulate their metabolism to compensate for the change in conditions [49], thus the results of the exposures of the empty shells must be interpreted taking this in to account. The shell in juveniles of C. concholepas is comprised mainly of aragonite which makes them more vulnerable to dissolution [50]. This is supported by the observations of this study where the loss of empty shell material was higher under high pCO₂ conditions. A previous study reported no significant differences in the loss of material from empty shells of C. concholepas maintained under acidified conditions (716 and 1036 μatm pCO₂, [15]). Although in both this and the previous study the shells were exposed for 30 days to the experimental conditions, the previous study was conducted with shells ca. 2.5 cm in size and at an average temperature of 12 ± 1°C (15) whereas the present study used shells ca. 3.0 cm in size at temperatures of 15 ± 1°C and 19 ± 1°C (Table 1). At both the experimental temperatures used in the present study, more shell material was lost from empty shells maintained under high pCO₂ levels (Table 3) and lower values of shell calcification in live individuals were detected under high pCO₂ conditions (Fig 4). Although shells of this species have a periostracum it does not remain attached to the shell surface after death. In the present study this external and protective organic layer was not present at the beginning of the experiment with the empty shells. This suggests that shells deprived of a periostracum may be more vulnerable to shell corrosion than the shells of live individuals [24]. The fact that more surface of the shell is exposed to seawater when the shell is empty than when it contains a living individual should be also considered beyond the lack of perisotracum. Under high pCO₂ conditions the seawater was saturated for calcite, in the case of live individuals, and close to saturation for empty shells, but it was not saturated for argonite in either case and thus the seawater was corrosive for the shells. This suggests that live individuals of C. concholepas at high pCO₂ levels are more vulnerable to shell corrosion. This might in part explain the high vulnerability of C. concholepas shells in southern Chile to shell-boring invertebrates [51], as observed around the site were the test individuals were collected (39 °S) and in central northern Chile (29–32°S, Manríquez PH pers. obs.). This susceptibility could be higher in southern Chile due to the natural and higher variability in seawater pH in response to significantly higher freshwater inputs to the coastal system (e.g. [52]) and higher pCO₂ sequestration (e.g. [53]). Thus the main structure associated with protection from predators and unfavourable conditions in C. concholepas and other shelled molluscs could be severely threatened under near future OA conditions.

Although in the present study we recorded the effects of elevated pCO₂ levels on shell calcification (live individuals) and shell corrosion (empty shells) measured as change in buoyant weight, no significant effects on the shell thickness of live individuals were found. This suggests that the low calcification recorded in live individuals of C. concholepas at high pCO₂ levels was not enough to produce detectable differences in shell thickness, that this part of the shell is not susceptible to changes in shell calcification or that those changes take places in other areas of the shell and not at the growing edge. We do not know why significant reductions in shell calcification did not take place along with reductions in shell thickness. Differences in the weight and morphology of shells of small C. concholepas collected from sites along the Chilean coast with different SST and carbonate system parameters together with an ontogenetic shift in shell mineralization of those individuals has been reported [50]. Likewise, increases in temperature at current day pCO₂ levels negatively affected calcium content in the barnacle Semibalanus balanoides but not the growth rate [54]. Therefore, it is possible that the low calcification recorded in the present study was occurring along with changes in calcium carbonate mineralization that do not alter shell thickness in C. concholepas.

With respect to shell corrosion measured as change in buoyant weight it should be borne in mind that under natural conditions only the exterior surface of the shell is exposed to the environment and that the interior surface is in contact with body fluids, which have been demonstrated in the case of bivalves to be more corrosive than the surrounding seawater during low-tide periods [49]. This study [49] indicated that well-fed and healthy molluscs are able to maintain the inner nacreous layer of their shells, even under corrosive seawater conditions. This suggests that it is difficult to infer shell corrosion based on empty shells exposed to contrasting conditions of pCO₂ alone or in combination with other stressors such as temperature. Therefore, measures based on empty shells exposed to different experimental conditions of pCO₂ and temperature may overestimate true levels of shell corrosion.

The positive effect of high pCO₂ levels alone on self-righting confirms the findings of a previous study on the same species [15]. This could be an important response in the early ontogeny for C. concholepas, or indeed any other species with a similar life history inhabiting the rocky intertidal habitats, where dislodgement is common and predators are abundant. The absence of predator-avoidance behaviour under high pCO₂ levels also agrees with a previous study [22]. Likewise, the absence of avoidance behaviour, with individuals remaining motionless in the experimental chamber when reared under high pCO₂ conditions, confirms the observations of a previous study in which C. concholepas were reared in the presence of the shell-crushing predatory crab A. hassleri, a visual predator [15,22]. This suggests that motionlessness and high tenacity might confer less vulnerability and may also be an energy saving strategy. As with other studies, the absence of avoidance behaviour may also suggest the presence of a negative effect of elevated pCO₂ levels on the chemosensory detection of predators [54–57].

Gastropod molluscs depend on mucus for locomotion and for attaching themselves to the substratum [58]. However, in C. concholepas, as in other similar gastropods, foot suction also plays an important role in mechanical adhesion [59]. Individuals of C. concholepas routinely attach themselves to rocks with notable force. In fact, rocky intertidal shellfish gatherers and divers use a tool called a "chope", a short steel spike, to pry them from the substratum [60]. More force per unit of foot area is needed to dislodge C. concholepas reared at high pCO₂ levels at both rearing temperatures, which suggests a positive effect of an increase in pCO₂ levels on tenacity. This highlights that increased and sustained near future conditions of pCO₂ and temperature could favour positive behavioural responses in small juveniles of C. concholepas; speedy self-righting and high tenacity. Both behavioural responses, self-righting and tenacity, involve mucus production, which in gastropods is described as a highly demanding activity in terms of energy use [61,62]. Thus, our results suggest that individuals of C. concholepas maintained under stressful conditions (high pCO₂ levels) might reduce their metabolic activity and growth to meet the high energy demands associated with important behavioural responses such as self-righting and adhesion without reducing their metabolic activity. This capacity is of particular importance in intertidal environments where individuals are exposed during low tides to environmental hypoxia [63], and considerable variability in pH [28] and temperature [29]. Therefore, as has been suggested recently in the literature, we conclude that the effects of OA and warming on animal behaviour are significant and might play an important role in determining the persistence of species and their success in ecological communities [64].

The average metabolic rates reported in the present study are within the range previously described for small juveniles of C. concholepas [15,23]. However, irrespective of the treatment temperature, when measurements were conducted under standard conditions (15°C and current-day pCO₂ levels) the present study found the existence of a significant reduction in the metabolic rates of individuals reared at high temperature and pCO₂ levels. This suggests the existence of a stress response in individuals moved from the treatment conditions to the measurement conditions at high temperatures. Metabolic depression in response to environmental stress has been recorded for virtually all the major animal phyla (see [65] for a review). In marine gastropods decreased metabolic rates have been linked with a strategy of matching demand to the availability of oxygen, and an increase in tissue concentrations of end-product metabolites indicating an increased reliance on anaerobic metabolism [66]. Individuals of C. concholepas inhabiting rocky intertidal habitats are commonly exposed to long periods (ca. 3 h) of air exposure and desiccation stress during low tides. In this environment invertebrates have evolved metabolic adaptations to environmental hypoxia and anoxia [63]. The metabolic depression in response to high pCO₂ levels could be interpreted as an effort to conserve energy for future energy demanding activities in individuals inhabiting intertidal pools where they are exposed to a succession of pH fluctuations [67]. This agrees with previous studies that suggest that the primary metabolic mechanism for responding to OA involves energy budget reallocation [68–70]. On the other hand, when measurements were done at the treatment levels of temperature and pCO₂ this reduction was not detectable. The metabolic rate was significantly affected by temperature, but not by pCO₂ level nor by the interaction between both stressors. Marine organisms within a certain range of temperature have the capability of acclimating their metabolic rate [71,72]. This suggests that under global change scenarios no significant effects of OA on the metabolism of C. concholepas will occur. However, significant effects of ocean warming on the metabolism of this species are expected.

The present study found that temperature alone did not significantly affect most of the investigated traits. The only investigated traits that were significantly affected by temperature were tenacity and the metabolic rate. This suggests that the isolated effect of temperature is variable and depends on the response trait being measured. Moreover, the present study found a lack of interactive effects of pCO₂ levels and temperature on all the investigated physiological, morphological and behavioural traits. This may be because the experimental temperature range used in the present study (i.e. 15–19°C) was not wide enough to include temperatures close to the upper thermal limit of C. concholepas and thus induce additive negative (sum of their individual effect of the stressors), synergistic negative (greater than the sum of their individual effects of the stressors) or antagonistic (contrasting actions of the stressors) effects of temperature and pCO₂ that may occur when the individuals are under higher levels of stress [73]. The effects of temperature on several traits in marine invertebrates have been shown to depend on the thermal variability experienced by the species in the field [73,74]. Sea-surface temperature along the coast of Valdivia, near the rocky intertidal zone where the individuals were collected (39°S), ranges between 10 and 15°C with an average (± SD) of 12°C (± 1) (Manríquez PH unpublished data). Therefore the control temperature conditions used in the present study (15°C) represent a temperature near the maximum recorded for the collection site. At this site settlers and small juveniles of C. concholepas inhabiting rocky intertidal habitats may face higher temperatures during the short (ca. 3 h) exposure to air temperatures during low tide. As a consequence, the population used in the present study is rarely exposed to steady high temperatures, such as the 19°C used in this study. In consequence our results suggest that the combined effects of warming and acidification on the early benthic ontogeny of C. concholepas is complex (negative or positive depending on the analysed trait) and without significant interaction between both stressors. A potential explanation for the different effects of elevated pCO₂ and temperature on some of the investigated traits could be that the underlying mechanism by which each individual stressor drives the investigated traits is different. Future studies are needed to go beyond this relatively moderate warming and describe the underlying mechanisms in order to fully understand the combined effect of warming and acidification on C. concholepas.

Despite the fact that intertidal organisms are presumably adapted to daily fluctuations in pCO₂ and temperature levels we conclude that sustained exposure to the expected increase in pCO₂ might have negative consequences for some important traits in the early life-history of this keystone species inhabiting the rocky intertidal zone. This is reinforced by the metabolic depression found in individuals moved from rearing conditions at high temperature and pCO₂ levels to standard conditions similar to current day conditions. This might be to the particular importance of acute changes in temperature and pH that organism in intertidal rock pools face [67]. We also conclude that in the early ontogeny of C. concholepas the near future increase in pCO₂ levels will not affect their metabolic rate but will have sub-lethal and negative consequences on shell growth and integrity (e.g. calcification and corrosion). However, the negative consequences of elevated pCO₂ levels on predator avoidance behaviour in this keystone species might have lethal consequences with knock on effects on population dynamics and community structure. Though it needs to be established whether the negative effect of reduced avoidance behaviour and possible weaker shells are compensated for by the increased dislodgement force and self-righting times. In agreement with a recent review [64] we conclude that without considering the effects of OA and warming on animal behaviour we will have a limited capacity to predict the ecological effects of ocean change.

Our results suggests that the consequences of elevated temperatures and pCO₂ level on a southern population of C. concholepas will be less severe than expected because of the absence of negative effects of increased temperature alone or in combination with acidification. However, the consequences of elevated pCO₂ levels described in this study might be worse than the potential ecological impacts described here if the investigated thermal range includes more stressful temperatures. Future studies need to go further than the relatively moderate warming scenario investigated here. Furthermore, populations naturally exposed to warmer thermal limits such as those found along the northern Chilean coast also need to be investigated in order to fully understand the combined effects of warming and acidification on this species and others with similar life histories.

All data files are available from the FigShare database: http://dx.doi.org/10.6084/m9.figshare.1591018↗.