Ocean warming and acidification alter the behavioral response to flow of the sea urchin Paracentrotus lividus

Abstract Ocean warming (OW) and acidification (OA) are intensively investigated as they pose major threats to marine organism. However, little effort is dedicated to another collateral climate change stressor, the increased frequency, and intensity of storm events, here referred to as intensified hydrodynamics. A 2‐month experiment was performed to identify how OW and OA (temperature: 21°C; pHT: 7.7, 7.4; control: 17°C‐pHT7.9) affect the resistance to hydrodynamics in the sea urchin Paracentrotus lividus using an integrative approach that includes physiology, biomechanics, and behavior. Biomechanics was studied under both no‐flow condition at the tube foot (TF) scale and flow condition at the individual scale. For the former, TF disk adhesive properties (attachment strength, tenacity) and TF stem mechanical properties (breaking force, extensibility, tensile strength, stiffness, toughness) were evaluated. For the latter, resistance to flow was addressed as the flow velocity at which individuals detached. Under near‐ and far‐future OW and OA, individuals fully balanced their acid‐base status, but skeletal growth was halved. TF adhesive properties were not affected by treatments. Compared to the control, mechanical properties were in general improved under pHT7.7 while in the extreme treatment (21°C‐pHT7.4) breaking force was diminished. Three behavioral strategies were implemented by sea urchins and acted together to cope with flow: improving TF attachment, streamlining, and escaping. Behavioral responses varied according to treatment and flow velocity. For instance, individuals at 21°C‐pHT7.4 increased the density of attached TF at slow flows or controlled TF detachment at fast flows to compensate for weakened TF mechanical properties. They also showed an absence of streamlining favoring an escaping behavior as they ventured in a riskier faster movement at slow flows. At faster flows, the effects of OW and OA were detrimental causing earlier dislodgment. These plastic behaviors reflect a potential scope for acclimation in the field, where this species already experiences diel temperature and pH fluctuations.

verity of wind-driven waves. A rise of the yearly mean wave height in the North East Atlantic by 20% to 40% has been already observed during the 20th century (Bacon & Carter, 1991;Bertin, Prouteau, & Letetrel, 2013).
As wave-induced water motion can potentially dislodge organisms from the substratum, hydrodynamics is considered as a major driver shaping the benthic intertidal and upper infralittoral communities (Denny, 1988). To resist dislodgement caused by hydrodynamic forces, benthic organisms rely on both the mechanics of their adhesive organs and their behavior (Hofmann & Todgham, 2010).
If OW and OA affect the ability of benthic organisms to withstand the hydrodynamic stress, the structure and dynamics of the ecosystems where they play key roles can be significantly affected (Agüera, Koppel, Jansen, Smaal, & Bouma, 2015;Britton-Simmons, Foley, & Okamoto, 2009;Duggins, 1981).
However, a longer exposure to these stressors can follow acclimation as observed in Sterechinus neumayeri (Morley, Suckling, Clark, Cross, & Peck, 2016;Suckling et al., 2015). Concerning behavior, simultaneous OW and OA increased grazing activity in Amblypneustes pallidus, while in H. erythrogramma, feeding rate increased with warming (Burnell, Russell, Irving, & Connell, 2013;Carey et al., 2016). After long exposure to OW and OA, Loxechinus albus feeding preference disappeared in response to acidification and not warming, while the vertical foraging speed and tenacity were not affected (Manríquez et al., 2017).
The impact of hydrodynamics on mechanical and behavioral responses of echinoids have been extensively studied. High flow velocity or wave exposure reduced movement distance and speed, feeding rate and particle capture efficiency, influenced righting and spine streamlining behavior, and enhanced TF mechanical properties (Cohen-Rengifo et al., 2018;Cohen-Rengifo, Moureaux, Dubois, & Flammang, 2017;Dance, 1987;Denny & Gaylord, 1996;George & Carrington, 2014;Jacinto & Cruz, 2012;Kawamata, 1998;Lauzon-Guay, Scheibling, & Barbeau, 2006;Lissner, 1980;Morse & Hunt, 2013;Stewart & Britton-Simmons, 2011;Tuya, Cisneros-Aguirre, Ortega-Borges, & Haroun, 2007). However, mechanical resistance and behavioral responses of sea urchins under the simultaneous impact of OW and OA coupled to an additional hydrodynamic stress have never been investigated. As echinoids play a major structuring role in many coastal ecosystems through their grazing activity results in a reference (Steneck, 2013), the impact of OW and OA in a more hydrodynamic ocean could be of tremendous importance.
The present study seeks to understand how OW and OA influence the resistance to an increasing flow regime in the echinoid P. lividus (Figure 1a). During a two-month experiment, echinoids were exposed to 6 fully crossed treatments including two temperatures (17 and 21°C) and three pH T (7.9, 7.7, and 7.4) according to the RCP8.5 scenario (IPCC, 2014). Potential functional effects on TF adhesive and mechanical properties under no-flow conditions, as well as behavioral responses associated with movement and shape modification (through spine reorientation) under flow conditions, were evaluated to identify their role in the resistance to dislodgment.

| Sea urchin collection and experimental setup
Aquaria were held within climate rooms to allow constant seawater temperature (°C). Between 10 and 20% of the water volume within each tank was renewed each day. Once a day, temperature, salinity, pH in NIST scale (pH NIST ), and electromotive force (mV) were measured as in Cohen-Rengifo et al. (2018). Daily electromotive force measurements were converted to pH in total scale (pH T ) using Tris/AMP buffer calibration (DelValls & Dickson, 1998). At day 5 of acclimation, temperature and pH were progressively modified from control values (17°C-pH T 7.9, +0.5°C/day, −0.05 pH per day). Target treatments (17°C-pH T 7.7, 17°C-pH T 7.4, 21°C-pH T 7.9, 21°C-pH T 7.7, 21°C-pH T 7.4) were reached after 12 days and were maintained for 12 more weeks. A computer-controlled IKS system was employed to manage experimental pH by bubbling CO 2 independently into each aquarium. IKS-pH measurements were calibrated against Metrohm pH meter daily measurements. Treatments were triplicated independently (1 aquarium = 1 replicate, 6 treatments = 18 aquaria).
To determine seawater total alkalinity (A T-SW ), a 50 ml sample was collected once a week from each aquarium and immediately filtered (0.22 μm MilliPore) and fixed with HgCl 2 7% (w/vol). Potentiometric titrations according to Gran (1950) were carried out using a Titrino 718 STAT (Metrohm AG). Quality control was performed using reference material supplied by Dickson laboratory (University of California, Batch 135), with measures being always within ±2% of the reference value. pCO 2 and the concentration of the carbonate system components (CO 2 , HCO − 3 , CO 2− 3 ) and calcite and aragonite saturation states (Ω) were calculated using the software CO 2 SYS (Pierrot, Lewis, & Wallace, 2006) and the dissociation constants for carbonate from Mehrbach, Culberson, Hawley, and Pytkowicz (1973) refitted by Dickson and Millero (1987).
In each of the 18 aquaria, 12 sea urchins were separated into three compartments (plastic-mesh cages). One compartment contained three individuals that were employed to monitor physiological state. Another compartment housed three individuals employed to evaluate biomechanics under no-flow conditions, and the last compartment housed six individuals employed to evaluate skeletal growth as well as biomechanics and behavior under hydrodynamic conditions. Samplings were carried out at week 1 (w1, just after the 12-day acclimation), 8 (w8), and 12 (w12). Individuals were fed ad libitum with Zeigler Bros., Inc. (USA) food pellets.

| Physiological state
Physiological state was assessed by evaluation of growth, respiration rate (µmol O 2 hr −1 g −1 ), coelomic fluid (CF) pH (pH NIST-CF ), total alkalinity (A T-CF , mmol/kg sw ), and buffer capacity. CF buffer capacity was calculated as the difference between total alkalinity of the CF and that of the seawater (ΔA T-CF = A T-CF -A T-seawater , mmol/kg sw ).
Measurements were taken at w1 and w8 on three individuals per aquarium (n = 3). Only growth was measured at w1 and w12 in 6 individuals per aquarium (n = 3), considering the increment (%) in ambital sea urchin test diameter with spines (d urchin ), test diameter (d test ) and height (h test ) without spines, ambital spine length (l spine ), and the increment (mm) in jaw size. See S2 in Appendix S1 for details.

| Microhabitat conditions
At Aber beach, sea urchins inhabiting intertidal pools are mainly wedged in self-burrowed pits where water movement and gas exchange are reduced during low tide. We hypothesized that sea urchins could be locally preadapted to low pH at the scale of the microhabitat. Therefore, pH T and A T over, below, and inside (i.e., the CF) an echinoid within its pit ( Figure S1) were measured in two tide pools at different tidal periods. See S3 in Appendix S1 for details.

| No-flow biomechanics
Biomechanical variables were measured at w1 and w8 following Cohen-Rengifo et al. (2018). Tenacity tests were performed on a whole individual or on a single tube foot disk. An individual was clamped with a metal grab and let to attach to a glass aquarium, while a single tube foot was directly let to attach to a glass piece.
The metal grab or the glass piece was connected to an Instron 5543 force stand (© Illinois Tool Works Inc). A force perpendicular to the substratum was applied at a constant speed of 25 mm/min. Force and displacement were recorded at a frequency of 10 Hz until detachment. Sea urchin detachment force (F urchin , N) or tube foot disk detachment force (F disk , N) was documented. The adhesive surface area of a single tube foot disk (S disk , mm 2 ) was estimated from stained footprints (Cohen-Rengifo et al., 2017;Santos & Flammang, 2007) while that of the sea urchin (S urchin ) was calculated by multiplying were prepared to measure the cross-sectional surface area of the stem connective tissue layer (S CT , μm 2 ), as this tissue bears all the external load exerted on a tube foot (Santos & Flammang, 2005).

| Flume tank setup and sea urchin dislodgement
At week 10, the whole experimental setup together with 107 alive sea urchins (18 aquaria, each containing 6 individuals; 1 died) was transported (duration 90 min) to the Royal Netherlands Institute for Sea Research (NIOZ-Yerseke). Hydrodynamic trials were performed in a recirculating flume tank (maximal speed: 90 cm/s; working section: 0.6 m wide × 2 m length) that was calibrated with a Vectrino Acoustic Doppler Velocimeter (Nortek Group). Before the trials, individuals were acclimated to their respective treatments for one week. In the aquaria, temperature, pH T-SW , and salinity were measured and controlled daily as described above. Hydrodynamic trials lasted 14 days (n = 14). In the flume tank, seawater parameters (mean ± SD) were salinity = 31.6 ± 0.5 (n = 14) and pH T-SW = 7.90 ± 0.01 (n = 14), while temperature was measured during the first 7 days for the control = 17.0 ± 0.4°C (n = 7) and during the last 7 days for the high-temperature treatment = 20.9 ± 0.5°C (n = 7). Because individuals were maintained for less than 30 min in the flume tank, pH T-SW was not manipulated.
A unidirectional flow parallel to the substratum was generated (Bouma et al., 2005). A transparent polymethyl methacrylate plate was employed as attachment substratum. Flow velocity (V F , in cm/s) in the flume tank was set at 30 cm/s to create an abrupt transition of water motion conditions. Each individual (6 per aquarium, n = 3) was placed alone, oral-side down in the middle of the working section under a plastic basket to limit displacement and allow attachment. After 5 min, the basket was removed, and V F was increased by 5 cm/s every 2 min (flow2′ regime) until reaching 90 cm/s or until detachment. During a hydrodynamic trial, detachment velocity (V Det , cm/s) and behavioral variables (see below) were measured for each individual. Afterward, skeletons were dried at 50°C for 48 hr and cleaned with NaOCl 2.5% for 2 hr. Four additional hours were needed to clean Aristotle's lanterns.

| Spine orientation and shape analysis
Sea urchin planform silhouettes were photographed with a Canon Powershot SX260HS camera at every V F to measure spine angle (Spine°, degrees, circular variable). A single spine situated between 180 ± 10° (considering that 0° corresponds to the positive Y mathematical axis; Figure 1b) was chosen for each individual. Using ImageJ v1.50i software, the angle formed by the tip of the spine with respect to 0° was measured in a clockwise direction (Cohen-Rengifo et al., 2018).
To determine whether the streamlined shape under climate change conditions was adopted earlier than in the control (i.e., 55 cm/s), planform shape was analyzed using pictures taken between 30-55 cm/s and at V Det . Shape analyses were performed with MATLAB v2015 software Image Processing Toolbox™, based on shape indices (S1 in Appendix S1) and elliptic Fourier coefficients (Agüera & Brophy, 2011;Cohen-Rengifo et al., 2018). Briefly, twenty elliptic Fourier harmonics were calculated per individual and V F . Shape of individuals from the same V F was averaged, so that a reconstructed overall shape can be visualized per V F according to treatment.

| Attached tube feet
A waterproof Canon Powershot d10 camera was placed under the transparent working section and took 10 frames/min of individuals' oral side. Individuals moved very fast reaching areas where photographs could not be taken. Therefore, a flow velocity gradient in which velocity increased every minute (Flow1′ regime) was implemented. Sea urchin photographs were used to count the number of attached TF was counted. The density of total attached TF relative to oral test surface area (TF att , mm −2 ) and the percentage of total attached TF relative to the number of adoral TF (TF att %, %) were calculated. The number of adoral TF (from the ambitus toward the oral peristome) was estimated on cleaned tests, considering that a pair of ambulacral pores corresponds to a single tube foot (Santos & Flammang, 2007;Smith, 1978).

| Statistics
Statistical tests are explained in detail in S4 in Appendix S1. A first general linear model (model 1)

| RE SULTS
Target treatments were maintained stable during the 81 days of experiment. Seawater and carbonate system parameters per treatment, averaged over the whole experiment, are available in Table 1. TA B L E 1 Seawater and carbonate system parameters (mean ± SD, n = 81 except for A T : n = 12) averaged over 12 weeks of experiment Values for seawater parameters and all linear variables are throughout expressed as mean ± SD.

| Physiological state
Increment in echinoid diameter, with and without spines, showed the highest values in control conditions (21 ± 6% and 19 ± 3%, respectively) and the lowest ones in the extreme treatment (10 ± 5% and 2 ± 3%, respectively). However, these differences were only significant for sea urchin diameter with spines (p = .001, Table 2).
Jaw size increment of control individuals significantly doubled that of individuals in all other treatments (p < .001).
Respiration rate and pH NIST-CF did not vary between treatments at any time (p > .13) and ranged between 0.72 and 1.28 µmol O 2 hr −1 g −1 or between 7.54 and 7.69 pH units, respectively (Table 2).
Buffer capacity of the CF (i.e., ΔA T-CF = A T-CF -A T-seawater ) was significantly higher at pH T 7.4 at w1 (p < .001) but it was not affected by treatment at w8 (Table 2).

| Microhabitat conditions
Coastal pH T-SW was 8.10 ± 0.05 (n = 4) while tide pool pH T-SW ranged from 7.4 to 8.8 and varied with time and compartment (S5 in Appendix S1). pH T-SW below the sea urchins was significantly lower than that over them during the day low tides (p ≤ .013), but not during the night low tides (S5 in Appendix S1, Figure S3). Seawater A T-SW was not affected by time nor by compartment (S5 in Appendix S1).

| Individual and tube foot disk adhesive properties
None of the disk properties varied with treatment at any time (Table S4). Sea urchin detachment force significantly differed with treatment only at w1 (p = .013), but not compared to the control.
Interestingly, whereas F urchin declined with time, T disk remained stable.

| Tube foot stem mechanical properties
Breaking force varied significantly at w1 (p = .018) and at w8 (p = .020), but only between the two extreme treatments (17°C-pH T 7.9 vs. 21°C-pH T 7.4, Figure 2a). S CT differed significantly only at w1 (p < .001), being the largest at 17°C-pH T 7.7 and the smallest at 21°C-pH T 7.4 compared to the control (Figure 2b). Extensibility did not differ between treatments at any time ( Figure 2c). Tensile strength, stiffness, and toughness (Figure 2d- .001 Jaw size increment (mm) Significant differences between means of treatments are indicated by letters in superscript; means sharing the same superscript are not significantly different (p-Tukey ≥ .05).

| In-flow biomechanics
In (V Det = 51.7 ± 6.6 cm/s, p-Tukey = .046). V Det in every treatment was modulated by the density of attached TF (p = .007) and by the interaction of both shape variables (spine° circularity: .003 ≤ p ≤ .007, estimate = 1.13; circularity is a shape index, see S1 in Appendix S1) and movement variables (V Mov -Dir Mov : .044 ≤ p ≤ .045, estimate = −0.005; Table S2, Figure S4). The probability of dislodgment increased with V F for every treatment (Figure 2b, Table S1), being the lowest in control conditions while the highest at 21°C-pH T 7.7 ( Figure 3b); this probability was higher for wider and taller individuals presenting longer spines (Table S1).

| Active movement
Active movement velocity (V Mov ) was affected by the interaction between treatment and V F (p < .047; Table S3, Figures S5 and S6).
Sea urchins reduced their V Mov with increasing V F in all treatments.

| Attached tube feet
The density of total attached TF (TF att ) decreased with V F (p reg ≤ 0.001) in all treatments ( Figure 5), though V F only accounted for 11%-34% of the variation. TF att significantly differed between treatments (F 5,754 = 10.8, p < .001), being the lowest at 21°C-pH T 7.7.
The percentage of detached TF at V F 90 was the highest (83%) in the control and the lowest (42%) at 17°C-pH T 7.4 (Table S6).

| Shape and spine analyses
The shape outline of the sea urchins differed according to V F and treatments (F two-way_MANOVA(30,210) = 1.33, p ≤ .001). Variation in shape according to V F was not significant at pH T 7.7 for both temperatures (p pairwise_MANOVA ≥ .37). In the control, shape significantly started to change at V F 60 (p pairwise_MANOVA ≤ .001) and was mainly modulated by homogeneous movements of upstream spines toward the downstream, that is, symmetrically to the right and left of the flow (Figure 6). At 17°C-pH T 7.4, shape started to change at V F 55 (p pairwise_MANOVA = .016), but the final streamlined shape was poor (Figure 6). At 21°C-pH T 7.9, shape variation occurred at V F 50 (p pairwise_MANOVA = .011), and the streamlined shape was more pronounced than in the control ( Figure 6). Finally, at 21°C-pH T 7.4, shape started to change at V F 50 (p pairwise_MANOVA = .024), but as spines situated upstream moved asymmetrically ( Figure 6) and spines situated perpendicular to the flow moved very little (p Moore's_Test > 0.2), the resulting shape was not streamlined.

| D ISCUSS I ON
The integrative approach of this study revealed that resistance to hydrodynamism in Paracentrotus lividus resulted from a complex array of behavioral and mechanical strategies performed to trade-off for negative effects in physiology, TF biomechanics, and behavior that occurred under a mid-term exposure to simultaneous OW and OA.
Furthermore, our experiments revealed plastic responses that were highly variable according to treatments and increasing flows.

| Physiological state and microhabitat preadaptation
Within pits, P. lividus experiences dual pH T-SW values at the same temporal scale, being significantly lower (−0.2 units) below the individual than over it. This difference in pH T-SW is probably linked to (a) the absence of water motion at low tide which restrains water renewal and gas exchange and to (b) the respiratory activity of the rock biofilm and the sea urchin. So, the natural pH and temperature diel fluctuations (Kwiatkowski et al., 2016;Moulin, Catarino, Claessens, & Dubois, 2011;Truchot & Duhamel-Jouve, 1980), together with the dual pH conditions, can explain P. lividus plastic physiological responses observed under a broad pH spectrum, and the tolerance to experimental chronic low pH that allows to maintain stable both coelomic fluid pH and respiration rate (Catarino et al., 2012;Collard et al., 2013;Collard, Dery, Dehairs, & Dubois, 2014).
Jaw and sea urchin (test diameter with spines) growths were 2.6fold and 2.5-fold, respectively, lower in the treatments compared to the control. Coping with warming and extracellular acidosis over the course of 8 weeks could have led to a lower resource allocation to growth (Hofmann & Todgham, 2010;Stumpp, Trübenbach, Brennecke, Hu, & Melzner, 2012). Although individuals were fed equally in every treatment, a possible modified digestive efficiency due to OW and/or OA (not measured) could account for the observed growth differences.

| No-flow tube foot mechanical performance
At w1, control breaking force is comparable with values previously reported for P. lividus (Cohen-Rengifo et al., 2017). Overall, mechanical properties showed no clear pattern at w1, which could reflect the different individual rates of acclimation to seawater changes. At w8, stems at 17°C-pH T 7.7 can absorb more energy during deformation (toughest stems) and therefore are more resistant to an external load It is important to highlight that a sevenfold higher force is needed to break the stem than to detach the disk from the substratum, meaning that, under a given external load, tube foot detachment can occur before tube foot breakage. It is probably more cost-effective to produce new adhesive compound than to lose an entire adhesive organ. Support for this statement comes additionally from nonquan-

| Improving tube foot attachment strategy
Intuitively, the first reaction of an organism that depends on its adhesive appendages to resist flow is by using them. According to our results, a first behavioral strategy, when climate change had mechanically weakened these organs, was to compensate by increasing the number of attached TF or by controlling the percentage of detached TF. This either amplifies the total adhesive force or avoids its decrease. The mean percentage of detached TF at V F 90 was lower in every climate change treatment, being even halved at 17°C-pH T 7.4 (−42%) with respect to the control (−83%). In the extreme condition, animals initially improved the density of attached TF, but the persistent hydrodynamic stress combined with OW and OA was detrimental (TF detachment up to 69%) and lead to earlier detachment.

| Escaping strategy
V Mov increased with temperature at pH T 7.9 and pH T 7.7. Enhanced locomotion and activity have been observed in response to warming (Kidawa, Potocka, & Janecki, 2010;Pewsey, 2004;Young, Peck, & Matheson, 2006) and acidification (Cripps et al., 2011;Manríquez et al., 2013;Nilsson et al., 2012;Spady, Watson, Chase, & Munday, 2014), but there is a poor understanding about the effects of combined stressors. As with the echinoid Loxechinus albus (Manríquez et al., 2017), our results revealed positive synergistic effects of OW and OA, since sea urchins were driven to adopt a riskier behavior by moving faster despite the hydrodynamic stress. On the contrary, negative synergistic effects of OW and OA were reported on locomotion of jumbo squid (Rosa & Seibel, 2008) and a decapod (Dissanayake & Ishimatsu, 2011) as a result of reduced metabolic scope.
At slow flow velocities, sea urchins reared under climate change conditions displayed an escaping behavior by moving fast, looking for less hydrodynamically stressful zones. Yet, their movement stopped sooner, probably because they cannot move safely anymore. In the control, at V F 50, 31% of sea urchins moved at the slowest V Mov and stopped moving at V F 65 (Cohen-Rengifo et al., 2018). This proportion increased to 50%-69% in individuals reared under climate F I G U R E 5 Regression slopes with R 2 and p-values for the density of total attached TF relative to ambital test surface (TF att ) per treatment and flow velocity. Data were transformed with X � = log 10 X + 1 . Significant differences between treatments are indicated by letters; means sharing the same letter are not significantly different (p-Tukey ≥ .05). *Data from Cohen-Rengifo et al. (2018) F I G U R E 6 Initial and final shape of Paracentrotus lividus for treatments in which shape varied significantly with flow velocity. Reconstructed outlines in white and planform shape in gray. Arrow represents flow provenance change conditions, in which a complete interruption of movement occurred mainly at V F 55.
Modified behavior resulted from three main pathways: elevated metabolic load, avoidance behavior away from the localized source of change, and information disruption (Briffa, Haye, & Munday, 2012;Lürling & Scheffer, 2007). The observed negative effects on growth and on behavior were probably due to an overall increase in metabolic load to cope acidosis and a shift in energy allocation. In addition, the implementation of an escaping behavior suggests that detection of the hydrodynamic stressor was well accomplished but as sea urchins adopted a riskier behavior, their interpretation of the hazardous hydrodynamic conditions may have been altered. Indeed, in marine vertebrates and crustaceans, behavioral changes often take place through info-disruption that occurs when their ability to gather and assess information and consequently their decision-making are impaired (Briffa et al., 2012;de la Haye, Spicer, Widdicombe, & Briffa, 2011).
Control P. lividus favored a mean downstream displacement at fast flows while in the other treatments, movement was random and characterized by constant back and forth. A change in movement patterns in acidified waters has been observed in a teleost probably because CO 2 affected its neurophysiology or because CO 2 was detected as a constant stressor (Green & Jutfelt, 2014).

| Streamlining strategy
Whereas spine movements led to a change in the overall shape, the direction and amplitude of these movements determined whether the final shape was streamlined or not. For instance, at 21°C-pH T 7.4 even though shape significantly changed, the final shape was not streamlined ( Figure 6). The lack in coordination between spines movements and shape modification demonstrated that individuals displayed streamlining behavior in an atypical way. Although atypical or absent streamlining is detrimental for detachment, it can also favor feeding behavior as spines in a "up position" can capture drifting algae.
Control P. lividus displayed a first reaction to increasing flows at V F 35 as observed in Strongylocentrotus franciscanus that perceives flow variations at flow velocities as slow as <10 cm/s, a behavior that hitches always leads to a streamlining behavior (Stewart & Britton-Simmons, 2011). In the climate change treatments, no early reaction to flow was observed in P. lividus, attesting for a possible disruption in information processing. In various organisms, behavioral abnormalities have been attributed to changes in seawater chemistry that leads malfunctions in neurological mechanisms (such as information processing) involving type A γ-aminobutyric acid (GABA A ) receptors (Chivers et al., 2014;Domenici et al., 2012;de la Haye et al., 2011;Nilsson et al., 2012;Tuomainen & Candolin, 2011). These ionotropic receptors present a high conductivity for Cl − and for HCO − 3 (Bormann, Hamill, & Sakmann, 1987;Nilsson et al., 2012). Echinoids  Spicer, & Hall-Spencer, 2007;Stumpp et al., 2012). The excitatory action of GABA resulting in increased Cl − has been already reported in echinoid tube feet (Florey, Cahill, & Rathmayer, 1975). We hypothesize that high CO 2 could alter information processing through the GABA A pathway, leading to the behavioral modifications observed in P. lividus.

| Integration of responses and ecological implications
Responses facing simultaneous OW and OA did not follow an intuitive pathway and were sometimes conflicting, making difficult to identify causalities and to discriminate the drivers governing one behavior or another. For instance, animals held at 21°C-pH7.7 showed improved TF mechanical properties but worst behavioral performance than animal held at 21°C-pH7.4. This might indicate that the biochemical paths and neurological mechanisms operate better in acidic environments. Likewise, sea urchins that attached with the lowest density of TF (21°C-pH T 7.7) did not move consequently at the lowest velocity, but on the contrary, showed the same median maximal velocity as individuals that attached with the highest number of TF (21°C-pH T 7.4). Therefore, TF att did not modulate the velocity at which sea urchins moved, at least not directly or not only by itself. When animals experience an environmental change, their earliest response is often a plastic modification of their behavior (Tuomainen & Candolin, 2011). The success of the behavioral change will depend on the rapidity of their reaction (Price, Qvarnstro, & Irwin, 2003) and whether populations have or not encountered similar conditions during their recent evolutionary history (Ghalambor, McKAY, Carroll, & Reznick, 2007). Adjustments in behavior could be beneficial if they improve fitness by increasing survival or reproductive success (Tuomainen & Candolin, 2011). Moving fast to escape the flow can be considered as an adaptive response as it reduces the probability of dislodgment and therefore improves survival. On the contrary, atypical or impaired streamlining can be maladaptive as it causes earlier dislodgment which reduces survival. Individuals showing plasticity are often selected to constitute populations that can survive rapid environmental changes and ensure population persistence (Kinnison & Hairston, 2007), which might be the case for the P. lividus Aber population.
In intertidal pits, sea urchins experience fluctuating pH that is often close to 7.7 or even lower below the animal. Whereas this reality induced adaptive physiological and mechanical responses when only two stressors were evaluated (OW and OA), maladaptive behavioral responses appeared when another factor (hydrodynamics) was included. In the near-future, P. lividus inhabiting pits will experience more severe pH/pCO 2 conditions. However, this could be partially compensate by diel natural fluctuations since animals will be exposed progressively to reduce pH (Jarrold, Humphrey, McCormick, & Munday, 2017). It is probable that P. lividus would successfully face pH stress, yet the outcome of the interaction with a fluctuating temperature remains unknown.

ACK N OWLED G M ENTS
We thank four anonymous reviewers for their fruitful and constructive comments. We thank P. Pernet, M. Bauwens, G. Bilodeau, and G. Martellacci for their support during fieldwork; N. Roevros and L.

None declared.
F I G U R E 7 Conceptual framework showing physiological, mechanical, and behavioral responses favoring (+), impairing (−), or not affecting (=) sea urchin attachment under ocean warming (OW) and acidification (OA). Sea urchins were subjected to 6 treatments including two temperatures (17°C and 21°C) and three pH T (7.9, 7.7, and 7.4). Compared to the control (C), responses increased (↑), decreased (↓), did not change (↔), were atypical (X), or not achieved (XX). Dashed lines stand for deficiency. Physiology and tube foot biomechanics were measured under no-flow, while behavior was evaluated under increasing flow velocity (V F : 30-90 cm/s). To avoid dislodgement, three behavioral strategies were undertaken: improving tube foot attachment, escaping the flow, and streamlining. See list of abbreviations in Appendix S2

AUTH O R S CO NTR I B UTI O N
MC-R, PF, and PD conceived the project. MC-R, AA, TB, PF, and PD provided ideas and initiatives. MC-R, AA, TB, PF, and PD designed the experimental design and setup. MC-R, AA, and SM collected and treated the data. MC-R, AA, and PD conceived the statistical treatments. MC-R led the writing of the manuscript. All authors contributed significantly to the drafts and gave final approval for publication.

DATA AVA I L A B I L I T Y S TAT E M E N T
Datasets are available in https ://doi.org/10.5061/dryad.123t3gr.