The importance of inter‐individual variation in predicting species' responses to global change drivers

Abstract Inter‐individual variation in phenotypic traits has long been considered as “noise” rather than meaningful phenotypic variation, with biological studies almost exclusively generating and reporting average responses for populations and species’ average responses. Here, we compare the use of an individual approach in the investigation of extracellular acid–base regulation by the purple sea urchin Paracentrotus lividus challenged with elevated pCO2 and temperature conditions, with a more traditional approach which generates and formally compares mean values. We detected a high level of inter‐individual variation in acid–base regulation parameters both within and between treatments. Comparing individual and mean values for the first (apparent) dissociation constant of the coelomic fluid for individual sea urchins resulted in substantially different (calculated) acid–base parameters, and models with stronger statistical support. While the approach using means showed that coelomic pCO2 was influenced by seawater pCO2 and temperature combined, the individual approach indicated that it was in fact seawater temperature in isolation that had a significant effect on coelomic pCO2. On the other hand, coelomic [HCO3 −] appeared to be primarily affected by seawater pCO2, and less by seawater temperature, irrespective of the approach adopted. As a consequence, we suggest that individual variation in physiological traits needs to be considered, and where appropriate taken into account, in global change biology studies. It could be argued that an approach reliant on mean values is a “procedural error.” It produces an artefact, that is, a population's mean phenotype. While this may allow us to conduct relatively simple statistical analyses, it will not in all cases reflect, or take into account, the degree of (physiological) diversity present in natural populations.


| INTRODUC TI ON
Real individuals are unique combination of traits, some above and some below average. It is time to recognize the uniqueness of the individual and to turn it to our advantage as biologists. (Bennett, 1987) Darwin (1859) introduced the notion that variation is a prerequisite for natural selection to operate. More than a century later, Bennett (1987) suggested that the variability in organismal responses, due to the inter-individual variation in natural populations or species, should not be written off as "noise," but instead could provide new powerful insights into how organisms function and have evolved. And yet historically, and even currently, biological studies have largely focused on mean organismal responses, a method that can significantly underestimate the importance of the individual and its genetic and phenotypic characteristics (Aldrich, 1975;Bennett, 1987;Spicer & Gaston, 1999).
The relative paucity of studies that take an inter-individual approach most likely biases our perception and understanding of organisms' physiological processes. The mean values we commonly generate and use in our analyses can be considered useful abstractions or even artefacts constructed to aid analysis or interpret (i.e., simplifying) our perception of nature. While convenient and powerful, a mean does not capture the level of phenotypic variation that exists in natural populations. This emphasis on mean values has been termed "the tyranny of the Golden Mean" (Bennett, 1987). In turns, the broad acceptance of the utilization of this approach has led us to consider values of central tendency being more important and meaningful than the data used to generate it (Spicer & Gaston, 1999).
It is common practice to use constants derived from one species or population, and expressed as a mean value, in the calculation of physicochemical characteristics of an organism. This could be because it has not been possible to derive that constant for the individual or species of interest, there was a desire to minimize "nuisance" inter-individual variation, or because the statistical analysis is more straightforward if mean data were used. However, it is possible that an individual approach could produce different, and potentially more accurate, outcomes of environmental challenge, compared with more traditional mean-based approaches.
On a different but related note, it is often assumed that constants derived from one system or species are equally applicable to another. For example, the mean dissociation constant (pKʹ 1 ) for hemolymph from the shore crab Carcinus maenas has often been used to calculate acid-base parameters (e.g., pCO 2 and [HCO 3 − ]) for other species when empirical values are not available (e.g., Calosi, Rastrick, et al., 2013;Miles, Widdicombe, Spicer, & Hall-Spencer, 2007;Spicer, Raffo, & Widdicombe, 2007). However, the efficacy of doing so has never been formally assessed.
The importance of maintaining a healthy acid-base status is critical as it ensures successful intracellular enzymatic processes, such as protein synthesis and ATP production (Grainger, Winkler, Shen, & Steinhardt, 1979;Walsh & Milligan, 1989). This results in positive growth and reproductive investment, ultimately determining the fitness of an individual. This said, the significance of inter-individual variation on acid-base responses to perturbation has rarely been considered while investigating the impact of OW and OA on marine ectotherms (although cf. Pistevos et al., 2011;Schlegel et al., 2012;Calosi, Turner, et al., 2013). Furthermore, to our knowledge, there is no published work where individual-level variation in the acid-base balance of ectotherms has been investigated within the context of the ongoing global change.
Consequently, the aims of this study are threefold. First, to characterize the level of inter-individuals variation in acid-base parameters of an ectotherm exposed to OW and OA conditions. Secondly, to assess the difference and the implications of using an individual-based approach, compared with an approach using mean values for pK 1 ʹ from the same species, when investigating the acid-base responses of a species exposed to global change challenges. And finally to assess the difference, and the implications, of using an individual-based approach or an approach using mean values for pK 1 ʹ from the same species, compared with an approach using mean values for pK 1 ʹ from a different species. The second and third aims will provide insight into understanding the significance of inter-individual variation in physiological responses and processes, ultimately used to predict species responses to global change drivers.
To achieve our aims, we exposed adult individuals of P. lividus to combinations of three seawater temperatures and two seawater pCO 2 in an orthogonal experimental design. First, we measured the pH and total carbon dioxide content (TCO 2 ) of coelomic fluid from each individual tested. We then calculated the first apparent dissociation constant (pKʹ 1 ) for coelomic fluid from that individual.
We then calculated the first apparent dissociation constant (pK 1 ʹ) (a) for coelomic fluid for each individual urchin, (b) as a mean value derived from values for a number of urchins. These pK 1 ʹ values were then used to generate individual and mean acid-base parameters of the coelomic fluid. These same parameters were also calculated for urchin coelomic fluid using pK 1 ʹ values derived from crab hemolymph. We then qualitatively compared the outputs of the three approaches, in order to test whether mean approaches are representative of the outcome of the analyses considering inter-individual variation in acid-base ability shown by individuals of P. lividus. Immediately upon arrival, sea urchins were transferred to aquaria (vol. = 300 L, approx. 40 indiv. per aquarium), supplied with water (S = 33, T = 15°C) from a recirculating seawater system connected to a filter (2213 External Filter, Eheim GmbH & Co., Deizisau, Germany) all of which was housed within a controlled-light and room temperature at L:D 12 hr:12 hr. Here, urchins were maintained for 3 days before use in any experiment, in order to recover and fed seaweed Laminaria digitata ((Huds.) Lamouroux, 1813) ad libitum.
Approx. 23 individuals were individually placed in labeled mesh hand-made cages (mesh size 1 × 1 cm, Cage vol. = 0.5 L) and each cage assigned haphazardly to each of the six treatments. Urchins were fed ad libitium during the exposure period but were starved 24 hr prior to coelomic fluid sampling to avoid postprandial increases in metabolic activity.
Finally, urchins were observed every day before and after a water change, as well as following sampling, to visually determine their health conditions and survival.

| Experimental setup, environmental monitoring, and carbonate system characterization
Urchins were held in large trays (vol. = 300 L) filled with seawater aspirated with either untreated air (pCO 2 ≈ 380 µatm) or CO 2 -enriched air (pCO 2 ≈ 1,000 µatm). Inside the trays, the desired seawater temperature was maintained using chillers (L-350, Guangdong Boyu Group & CO., Guangdong, China) and aquarium stick heaters (3614, Eheim, Deizisau, Germany) in combination. To aid water mixing, each tray was fitted with a submersible pump (Koralia Nano Evolution 900, HYDOR USA Inc., Sacramento, CA USA). Water changes were performed daily to maintain good water quality. For details on the environmental monitoring and carbonate system characterization, see the dedicated section and the Figure S1 in the Appendix S1.

| Perivisceral coelomic fluid sampling and analyses
To avoid the negative effects of multiple intrusive (i.e., with a needle) sampling, a single coelomic fluid was taken at the end of the exposure period (7 days). While obtaining multiple samples from the same individual is good practice to avoid bias from intra-individual variation on the characterization of inter-individual variation (Bennett, 1987), intrusive sampling of the coelom activates an immune response (Smith et al., 2010) by introducing bacteria and other pathogens: Changes in cellular immune condition co-occurred with changes in extracellular acid-base balance of the green sea urchin Strongylocentrotus droebachiensis were reported by Dupont and Thorndyke (2012). Therefore, we had to assume that such acid-base responses were repeatable and did so on the basis of the acid-base individual-level responses of S. droebachiensis exposed to stable seawater conditions in a previous study (see Appendix S1: Figure   S4). Here, sea urchins displayed relatively stable and "consistent" coelomic fluid pH and bicarbonate concentrations when these parameters were measured repeatedly over time (see Appendix S1: Figure S4).
Perivisceral coelomic fluid (vol. = 500 µl) was extracted anaerobically at day 7 from each individual using a gas-tight syringe (500 µl, 1750 RN, Hamilton Bonaduz, Switzerland) while positioning the urchin ventral side uppermost, submerged just below to the water surface. The needle of the syringe was carefully inserted at an angle of approx. 45° and to a depth of 10 mm through the soft membrane surrounding the Aristotle's lantern directly into the urchins' perivisceral coelom. A second sample was obtained from the main coelomic cavity by positioning the needle at about 90° relative to the oral surface and deeply inserting it into the individuals' main extracellular space (as described in Calosi, Rastrick, et al., 2013). Great care was taken to avoid damaging the gut and gonads, and thereby contaminating coelomic fluid.
Measuring coelomic fluid carbon dioxide and pH was carried out using well-established methods (Donohue et al., 2012;Marchant, Calosi, & Spicer, 2010;Miles et al., 2007;Rastrick et al., 2014;Small, Calosi, White, Spicer, & Widdicombe, 2010;. To determine coelomic fluid TCO 2 , 50 µl of fluid was introduced anaerobically into a previously calibrated CO 2 analyzer (965D, Ciba Corning Diagnostic Cor., Cambridge, MA, USA) less than 30 s after sampling. The pH of the coelomic fluid (pH cf ) was measured, within 60 s of extraction. The sample was placed in a microcentrifuge tube (1.5 ml, Fisherbrand, Thermo Fisher Scientific Inc.) and a micro-pH electrode (MI-413, Microelectrodes, Bedford, MA, USA) immersed in the fluid. The electrode was coupled to a calibrated pH meter (Five Easy, Mettler Toledo). TCO 2 and pH cf measures were performed at the respective temperature of incubation according to the treatment at which the urchin was exposed in order to maintain constant environmental conditions and avoid animals stress.
Unused samples of coelomic fluid were frozen at T = −20°C in a microcentrifuge tube (1.5 ml) and subsequently used to determine individuals' non-bicarbonate buffer (NBB) line (see below).

| Determination of individuals' pK 1cF values, pCO 2cf , and [HCO 3 − ] cf
The first (apparent) dissociation constants for the coelomic fluid UK) against a range of CO 2 tensions (0.04 to 1.01 kPa roughly equivalent to 0.3 to 7.6 mmHg) supplied by precision gas mixing pumps (Wösthoff, Bochum, FRG). Using tonometers coelomic fluid samples were maintained at the environmental temperature the urchin was exposed to (T = 10, 15 and 20°C). Both TCO 2cf and pH cf were measured, as described above, at each CO 2 tension. At high pH and low pCO 2 , carbamate concentration cannot be ignored (Truchot, 1976). Acidifying the environment resulted in an acidification of extracellular body fluids. Consequently, we estimated that any carbamate present would be in negligible quantities and so has not been calculated. Also, our calculated values for [HCO 3 − ] may also include very small amounts of CO 2 in other chemical forms.

| Determination of mean pCO 2cf and [HCO 3 − ] cf
To investigate the effect of adopting a "mean approach," and to compare its outcome to the that for an individual approach, pCO 2cf and [HCO 3 − ] cf were calculated using the average value of individual sea urchins' pKʹ 1cf , as well as calculated using mean pKʹ 1cf values for C. maenas: 6.057, 6.029, and 6.000 at 10, 15, and 20°C (Truchot, 1976)

| Determination of key morphometric parameters
After the coelomic fluid was sampled, the height and diameter of the tests for each sea urchin were measured using a calliper (PD-151, Pro's kit Industries Co., Ltd., Taiwan) and used to calculate the spheroidal volume (as in Calosi, Rastrick, et al., 2013). Finally, sea urchins were weighed with a digital high-precision scale (PS-200, Fisher Scientific Ltd., Corby, UK-0.1 mg accuracy).

| Survival
No mortality was recorded in either the "elevated temperature" or the "extreme temperature" treatments, while among the other treatments between one and three urchins died during the experiment.
No significant relationships between mortalities and seawater pCO 2 , temperature or their combination were detected (χ 2 = 2.25, df = 3, p = 0.522). In addition, survivors appeared in excellent health conditions, that is, none showed any noticeable loss in spines, and all responding actively by moving their spines and extruding their tube feet when coelomic fluid was sampled.

| Statistical analysis
To investigate the effect of elevated pCO 2 , temperature and their interaction on urchins mortality and mean pH cf , TCO 2cf , pCO 2cf , and [HCO 3 − ] cf , determined in the three different ways (a) individual pKʹ 1cf for P. lividus (from this study), (b) mean pKʹ 1cf for P. lividus (from this study), and (c) using mean pKʹ 1cf for C. maenas (Truchot, 1976). We then used a two-way analysis of covariance (ANCOVA) with individuals' body volume (cm 3 ), wet mass (g) as covariates. Finally, we compared the patterns of significance obtained from the three approached employed to explore whether substantial differences exist between these different approaches. The same analysis was conducted on the individual pKʹ 1cf to assess OW and OA combined effect on it. As preliminary analyses showed that covariates never exerted a significant effect on any of the traits investigated (maximum F 1,129 = 1.021, p > 0.05), they were removed from further analyses and so we were able to use an ANOVA test. All data met the F I G U R E 1 The effects of elevated seawater pCO 2 and temperature on coelomic fluid (a) pH (pH cf ) and (b) Log 10 TCO 2 (TCO 2cf ) of the sea urchin Paracentrotus lividus. Temperature treatments are indicated by blue, orange, and red colors for 10, 15, and 20°C, respectively. Ambient seawater pCO 2 (≈300 µatm) and elevated seawater pCO 2 (≈1,000 µatm) are indicated by plain and hatched box plot, respectively. Lower case letters identify significant differences (p < 0.05) between treatments. Capital letters identify significant differences (p < 0.05) between temperature treatments. Asterisks identify significant differences (p < 0.05) between pCO 2 treatments at the same temperature treatment assumption for normality of distribution and homogeneity of the variance as untreated or Log 10 transformed data, with the exception of pH cf . However, as log 10 transformation was not beneficial for pH cf , and considering that our experimental design included six treatments with a minimum of 19 replicates per treatment per measurement, we assumed that the experimental design employed should be tolerant to deviation from the assumptions of normality and heteroscedasticity (Sokal & Rohlf, 1995;Underwood, 1997). In addition, for no variable investigated, a significant relationship between its unstandardized residuals and the factors investigated was found, indicating that as our experimental design was solid we could use row data for pH cf . Finally, pairwise comparisons were conducted using the estimated marginal means test with Fisher least significant difference (LSD) correction.
All analyses were conducted using IBM SPSS Statistic 21.
The effects of exposure to seawater pCO 2 , temperature, and their combination on pH cf and TCO 2cf are presented in Figure 1a,b, respectively.
Mean pH cf was positively affected by elevated seawater pCO 2 at 15°C, while there was no significant effect of this factor on mean pH cf at 10 and 20°C, as indicated by a significant interaction between seawater pCO 2 and temperature (see Table 1). In addition, seawater temperature in isolation had a significant effect on mean pH cf : mean pH cf was greater in urchins kept at 20°C than those kept at 10°C. However, mean pH cf for urchins kept at 15°C was comparable to that of the other two temperatures tested (see Figure 1a).
Mean TCO 2cf was significant greater at the higher seawater pCO 2 conditions for all temperatures tested (see Table 1), while it was greater at 15°C and lower at 20°C, with seawater pCO 2 and temperature having a significant positive and negative effect on mean TCO 2cf , respectively (see Figure 1b and Table 1). Nonetheless, seawater pCO 2 and temperature together had no significant effect on mean TCO 2cf (see Table 1).
Results for the analysis of individual pKʹ 1cf for P. lividus are presented in Figure 2a. Individual pKʹ 1cf values ranged between 5.50 and 7.51 (see Figure 2b).
Individual pKʹ 1cf values decreased significantly with increasing seawater temperatures (see Figure 2a and Table 1). There were no significant effects of seawater pCO 2 on its own, or in combination with temperature (see Table 1).

| Calculation of coelomic fluid pCO 2cf and [HCO 3 − ] cf determined using individual data for P. lividus
Results for the analysis of pCO 2cf and [HCO 3 − ] cf determined using individual data for P. lividus are presented in Figures 3a and 4a, respectively.
Seawater temperature had a negative effect on mean pCO 2cf for urchins kept at the highest temperature tested compared to those that experienced 10 and 15°C, and mean pCO 2cf values were significantly lower at 20°C (see Figure 3a and Table 1).
Furthermore, neither seawater pCO 2 on its own nor in combination with temperature had any significant effect on mean pCO 2cf (see Table 1).
There was a positive effect of seawater pCO 2 on mean [HCO 3 − ] cf at all temperatures tested (see Table 1). Conversely, seawater temperature had a negative effect on mean [HCO 3 − ] cf between 15 and 20°C (see Figure 4a and Table 1), lower values being characteristic of the highest temperature tested (see Figure 4a). Additionally, seawater pCO 2 effect was stronger than seawater temperature effect. (see Fs in Table 1), while no effect of seawater pCO 2 and temperature together was recorded (see Table 1). ] cf values ranged between 1.03 and 8.23 mmol/L (see Appendix S1: Figure S3b).

| Calculation of coelomic fluid pCO
Mean pCO 2cf was negatively affected at the highest seawater temperature, for the low seawater pCO 2 treatment, compared with the other two seawater temperature tested (see Figure 3b), while there was a progressive negative effect on mean pCO 2cf from 10 to 20°C for the high seawater pCO 2 treatment, as indicated by the presence of a significant interaction between seawater pCO 2 and temperature (see Figure 3b and Table 1). Temperature in isolation had a negative stronger effect on mean pCO 2cf as there was a significant difference in mean pCO 2cf within all seawater temperature treatments (see Table 1), and mean values decreased with increasing seawater temperature (see Figure 3b).
There was a significant positive effect of seawater pCO 2 on mean individual  ] cf at all temperature tested (see Table 1). On the contrary, seawater temperature in isolation had a negative effect on mean individual [HCO 3 − ] cf between 15 and 20°C (see Table 1), lower mean values being found at the highest seawater temperature tested (see Figure 4b). However, seawater temperature effect in isolation was weaker than the effect of seawater pCO 2 on its own (compare Fs Table 1). Additionally, no interaction between seawater pCO 2 and temperature interaction was detected (see Table 1).
TA B L E 1 Results of ANOVAs investigating the effects of elevated pCO 2 and temperature and their interaction on the coelomic fluid parameters of the sea urchin Paracentrotus lividus

| Calculation of coelomic fluid pCO 2cf and [HCO 3 − ] cf using mean pKʹ 1 for C. maenas
Results for the analysis of pCO 2cf and [HCO 3 − ] cf determined using C.
Mean pCO 2cf was reduced by increasing in temperature alone from 15 to 20°C, and only for the low seawater pCO 2 treatments.
In contrast, there was no significant effect of temperature while no effect of temperature on urchins exposed to high seawater pCO 2 conditions, as shown by the significant interaction between seawater pCO 2 and temperature (see Figure 3c and Table 1). Seawater ] cf were calculated using: (a) individual pKʹ 1 determined for P. lividus, (b) mean pKʹ 1 for P. lividus, and (c) mean pKʹ 1 for C. maenas. Significant p-values are given in bold.

TA B L E 1 (Continued)
F I G U R E 2 The effects of seawater pCO 2 and temperature (a) on coelomic fluid pKʹ 1 (pKʹ 1cf ) and (b) on inter-individual variation of coelomic fluid pKʹ 1 (pKʹ 1cf ) in the sea urchin P. lividus. Temperature treatments are indicated by blue, orange, and red colors for 10, 15, and 20°C, respectively. Ambient seawater pCO 2 (≈300 µatm) and elevated seawater pCO 2 (≈1,000 µatm) are indicated by (a) plain and hatched box plot, respectively and (b) by clear and darker colors, respectively. (a) Capital letters identify significant differences (p < 0.05) between temperature treatments and (b) dots identify individual measurements temperature had a negative effect on mean pCO 2cf values between 15 and 20°C, being significantly lower at 20°C when compared to 10 and 15°C (see Figure 3c and Table 1).
No significant effect of the combination of altering seawater pCO 2 and temperature on [HCO 3 − ] cf was detected (see Table 1), although seawater pCO 2 on its own had a positive effect on mean [HCO 3 − ] cf at all seawater temperatures tested (see Table 1). That said seawater temperature had a negative effect on mean [HCO 3 − ] cf between 15 and 20°C, being significantly lower at 20°C (see Figure 4c and Table 1). Additionally, seawater pCO 2 appeared to have a stronger effect on mean [HCO 3 − ] cf than seawater temperature (see Fs in Table 1).

| D ISCUSS I ON
The analyses and interpretation of average responses in populations and species have proved a powerful and useful tool in advancing our understanding of biological systems and their responses to environmental changes. However, the success of this approach has in some ways eclipsed the ecological and evolutionary significance of individual responses and individual variation (Darwin, 1859). Here, we show that integrating information on inter-individual variation in the first (apparent) dissociation constant for sea urchin coelomic fluid is key to understanding the nature of and pronounced variability in the acid-base responses of sea urchins P. lividus to changes in seawater temperature and pCO 2 . Furthermore, we demonstrate that using inter-individual values for the first (apparent) dissociation constant of the coelomic fluid paint, a significantly different picture in predicting how the acid-base status of urchins will be impacted by future ocean warming (OW) and acidification (OA), compared with the common practice of using mean values for the first (apparent) dissociation constant. Below we discuss these findings are greater detail and discuss the importance of using an individual-based approach in F I G U R E 3 The effects of elevated seawater pCO 2 and temperature on coelomic fluid Log 10 pCO 2 of the sea urchin P. lividus. pCO 2cf determined (a) using individual values of pKʹ 1 for P. lividus, (b) using the mean individual pKʹ 1 for P. lividus, and (c) using pKʹ 1 for Carcinus maenas. Temperature treatments are indicated by blue, orange, and red colors for 10, 15, and 20°C, respectively. Ambient seawater pCO 2 (≈300 µatm) and elevated seawater pCO 2 (≈1,000 µatm) are indicated by plain and hatched box plot, respectively. Lower case letters identify significant differences (p < 0.05) between treatments. Capital letters identify significant differences (p < 0.05) between temperature treatments F I G U R E 4 The effects of elevated seawater pCO 2 and temperature on coelomic fluid Log 10 [HCO 3 − ] of the sea urchin P. lividus. [HCO 3 − ] cf determined using (a) individual values of pKʹ 1 for P. lividus, (b) the mean individual pKʹ 1 for P. lividus, and (c) pKʹ 1 for C. maenas. Temperature treatments are identified by blue, orange, and red colors for 10, 15, and 20°C, respectively. Ambient seawater pCO 2 (≈300 µatm) and elevated seawater pCO 2 (≈1,000 µatm) are identified by plain and hatched box plot, respectively. Capital letters indicate significant differences (p < 0.05) between temperature treatments. Asterisks indicate significant differences (p < 0.05) between pCO 2 treatments at the same temperature treatment interpreting species' acid-base responses to the future environmental challenges of the global change.
Here, we detected a high level of phenotypic variation in the pKʹ 1cf , values of individual sea urchins. Values ranged between 5.50 at 15°C and 7.51 at 10°C, varying by 36.55% overall, and by 23.32%, 23.10%, and 32.55% at 10, 15, and 20°C, respectively. Not taking such variation into account alters the values of acid-base variables calculated using this "constant" and so changes the outcome of any study of the effects of altered environmental factors on extracellular acid-base balance. Using a mean value for the first (apparent) dissociation constant to calculate coelomic fluid pCO 2 in individuals exposed to combined seawater pCO 2 and temperature leads to an underestimate of the values. When individual variation in the first (apparent) dissociation constant is taken into account when calculating coelomic fluid pCO 2 , the outcome of the experiment was different. No significant interaction between seawater pCO 2 and temperature was detected, and seawater temperature had a strong negative effect on urchins' coelomic fluid pCO 2 , especially for the urchins kept at 20°C.  (Truchot, 1976) and P. lividus: 6.455, 6.256, and 6.158 at 10, 15, and 20°C, respectively. The result of this similarity is that using mean pKʹ 1cf for the decapod crab Carcinus maenas to calculate acid-base parameters for urchins was not considerably different from using the mean pKʹ 1cf for the sea urchin P. lividus.
As hypometabolic and osmoconformers echinoids are considered to be particularly vulnerable to climate and global change drivers (see Dupont & Thorndyke, 2013), as also evidenced by a growing number of studies investigating the effects of OW and OA on urchin's acid-base status (e.g., Miles et al., 2007Spicer & Widdicombe, 2012), calcification (e.g., Stumpp, Trübenbach, et al., 2012), growth (e.g., Albright et al., 2012), fecundity and development (e.g., Dupont, Ortega-Martinez, & Thorndyke, 2010;Byrne, 2011;Byrne et al., 2011), energy budget (e.g. Stumpp, Trübenbach, et al., 2012) and distribution (e.g., Calosi, Rastrick, et al., 2013). Despite this growing attention, still little is known about combined effects of OA and OW on their physiology. Catarino, Bauwens, and Dubois (2012) suggested that seawater pCO 2 had a greater effect than seawater temperature, on P. lividus coelomic acid-base status showing that this species is eurythermal. Consequently, this species inhabits the thermally labile intertidal (Ulbricht and Pritchard, 1972;Lawrence, 1990) Farmanfarmaian, 1966;Miles et al., 2007;Stumpp, Trübenbach, et al., 2012;Collard et al., 2013;Collard, Ridder, David, Dehairs, & Dubois, 2015). However, the inter-individuals approach we adopted revealed a pronounced effect of temperature on the acid-base status of P. lividus as individuals' physiological parameters were all positively affected in urchins exposed at 20°C. Coelomic fluid pH was greater in urchins exposed to elevated seawater pCO 2 and temperatures, suggesting that urchins exposed to future OW and OA may be able to trigger buffering mechanisms to compensate for body fluid acidosis caused by an increase in seawater pCO 2 , future studies which investigate longer term exposure and further improving our experimental design will help to further validate our findings, hopefully overcoming current limitations. In our study, increasing the temperature led to an overcompensation of the acidosis incurred. This may be in part explained by the fact that equilibrium constants (pK) of chemical reactions are generally temperature-dependent, including those for the protonation of imidazole groups (pKIm), imidazole being largely responsible for intracellular and extracellular non-bicarbonate buffering in ectotherms (Burton, 2002). However, it is important to note that the "imidazole alphastat hypothesis" assumes a single temperature-dependent pK value for all non-bicarbonate buffers (Burton, 2002), while we clearly show here that pK values vary considerably among individuals of the same species maintained at the same environmental temperature.
Although the importance of using an individual approach in a global change context has been recently recognized (e.g., Pistevos et al., 2011;Schlegel et al., 2012), studies explicitly addressing this issue at the physiological level are scarce (e.g., Calosi, Turner, et al., 2013;Melatunan et al., 2013). Nonetheless, our study shows that using an individual-based approach results in a reduced variation in the calculation of sea urchins' coelomic fluid pCO 2 and [HCO 3 − ] exposed at 10°C at both ambient and elevated pCO 2 seawater conditions, but a greater variation in urchins' coelomic fluid pCO 2 for urchins that experienced ambient seawater pCO 2 at 20°C. This finding supports the idea that our understanding of sea urchins' acid-base regulation, both in general and specifically in responses under global change challenges, is shaped and dependent on the experimental approach used. As a consequence, considering individuals' variation is fundamental. This is particularly true, as up to now physiological ecologists have largely ignored physiological inter-individual variation, as they did not perceive it as biologically significant (Bennett, 1987). Trying to raise awareness on the importance of inter-individual variation, Forsman and Wennersten This study is the first, to our knowledge, that uses an individualbased approach to determine the acid-base responses of an echinoid species facing a multiple factorial global change challenge. As previous studies have taken a mean approach to investigate the same or similar questions, it is difficult to assess whether the interpretation of organisms' acid-base responses in a changing environment to date is fully accurate. It is possible that previous work determining echinoids sensitivity to the global change depicts at worst an erroneous at best incomplete, picture, particularly if we consider the high complexity of physiological responses. This means that currently, without considering individual responses, it may be very difficult to predict how populations, species, communities, and ecosystems will actually respond to a changing ocean.

ACK N OWLED G M ENTS
We thank M. Hawkins for valuable help with the transport and culturing of sea urchins, and for her help with the experimental system.

CO N FLI C T O F I NTE R E S T S
The authors declare no competing or financial interests.

AUTH O R CO NTR I B UTI O N S
EG and PC conceived the study. EG conducted the experiment, carried out the biological measurements, calculations, and statistical analyses, with training from PC and JIS. EG wrote the first draft of the manuscript with input from PC. All authors contributed to the final version of the manuscript.