Ocean acidification impacts sperm swimming performance and pH_i in the New Zealand sea urchin Evechinus chloroticus

Reproduction is a central part of a species' life cycle, with an egg being fertilized by a single sperm to produce a viable zygote. In most broadcast-spawning and sperm-casting marine invertebrates, external fertilization depends on the flow environment into which the gametes are spawned and, at smaller spatial scales, biological traits such as sperm motility and chemotaxis to bring the gametes together (Bishop, 1998; Crimaldi and Zimmer, 2014). The process of spawning also shifts the sperm and eggs from the intracellular chemical environment of the adult reproductive organs into that of the surrounding seawater (Bishop, 1998; Crimaldi and Zimmer, 2014; Hurd et al., 2020).

Sea urchins have been used as a model system to study fertilization and early development in marine invertebrates since the mid-19th century (Monroy, 1986; Adams et al., 2019). Male and female sea urchins have an annual reproductive cycle, generally with mature gonads in the perivisceral coelom during the summer months (Chia and Bickell, 1983; Adams et al., 2019). In the male gonad, cellular respiration of the concentrated sperm is regulated through CO₂ tension, with high CO₂ (hypercapnia) and a low internal pH (pH_i) of <7.0 maintaining the sperm in a non-motile quiescent state (Chia and Bickell, 1983; Mohri and Yasumasu, 1963; Mita and Nakamura, 2001; Hurd et al., 2020).

Upon release to the surrounding seawater, the sperm cells experience a drastic environmental change: from K⁺-rich (>30 mmol l⁻¹), low oxygen tension and acidic conditions in the testes to lower K⁺ concentrations, oxygen-rich conditions and basic pH of around 8.2 in seawater (Christen et al., 1982; Johnson et al., 1983; Tosit, 1994; Bögner, 2016). Spawning and the subsequent dilution of sperm in seawater results in the alkalinization of pH_i through the release of CO₂ and H⁺ (Johnson et al., 1983; Trimmer and Vacquier, 1986). When pH_i is above 7.2–7.3, the dynein ATPase flagellar motors that produce swimming are activated, with the maximum rate of increase in motility and respiration at about pH 7.5 (Gibbons and Fronk, 1972; Christen et al., 1982; 1986; Neill and Vacquier, 2004; Nishigaki et al., 2014).

Because sperm are unicellular, they have a limited capacity to tolerate changes in P_CO2 and pH due to anthropogenic climate change (Melzner et al., 2009). In addition, hypercapnia, which has a narcotic effect within the gonad, may also impact sea urchin sperm released into ocean acidification (OA) conditions (increased P_CO2, decreased pH; Havenhand et al., 2008; Morita et al., 2010) through a failure to obtain the pH_i required for the activation of sperm swimming. The impaired ability to regulate pH_i may then have a narcotic effect on sea urchin sperm, resulting in decreased motility and swimming speed (Kurihara, 2008; Byrne, 2011; Reuter et al., 2011; Bögner, 2016; Hurd et al., 2020).

Here, we tested the relationship between pH_i and sperm motility in the New Zealand sea urchin Evechinus chloroticus under near-future (2100) and far-future (2150) atmospheric P_CO2 conditions under Representative Concentration Pathway (RCP) 8.5 (IPCC, 2014). We first developed a new methodology for measuring pH_i in sea urchin sperm using the fluorescent dye SNARF-1 and a nigericin calibration which required concurrent development of an appropriate ionic buffer. Then, as there is increasing recognition of the importance of individual responses to climate change (e.g. Pistevos et al., 2011; Schlegel et al., 2012; Guscelli et al., 2019), we measured pH_i and performance in sperm from the same male [using SNARF-1 and computer-assisted sperm analysis (CASA; Partyka et al., 2012), respectively]. Sperm was activated by dilution in control and lower pH seawater (OA treatment) to test whether there was a narcotic effect on sperm performance (e.g. decreased motility and swimming speed;, Kurihara, 2008; Byrne, 2011; Reuter et al., 2011; Bögner, 2016; Hurd et al., 2020) as a result of a failure to achieve the pH_i required for activation of sperm swimming under OA conditions.

Protocols for manipulating and analysing elevated P_CO2 experimental seawater were based on recommendations from Dickson et al. (2007) and Riebesell et al. (2010). In brief, seawater was bubbled with target CO₂ concentrations to produce three OA treatment seawaters: (1) Control: atmospheric CO₂ at the time of the experiments (2014, 380 µatm); (2) business as usual emission scenario RCP 8.5 at 2100 (range 851–1370 μatm CO₂); (3) business as usual emission scenario RCP 8.5 at 2150 (range 1371–2900 µatm CO₂; IPCC, 2014). Seawater for all experiments was obtained pre-filtered (from Sea Life Kelly Tarlton's, Auckland, New Zealand) and was re-filtered before P_CO2 manipulation through two nested, 1 µm filter bags (filtered seawater, FSW). Air was initially dried and stripped of CO₂ before being mixed with instrument grade CO₂ (99.98%) using a dial-a-gas mass flow controlled (Smart Trak2, Sierra Instruments, Monterey, CA, USA) mixing setup based on the experimental system outlined by Fangue et al. (2010) and calibrated with an infrared CO₂ analyser (Qubit S151, Qubit Systems Inc., Kingston, ON, Canada). To ensure the equilibrium of gas partial pressures between target gas mixtures and experimental seawaters, self-contained 20 l containers of FSW were bubbled for a minimum of 16 h. Bubbling used a temperature-controlled (20±1°C) circulation system where gas mixtures were added through a venturi injector (Mazzei, MK-384, Mazzei Injector Company, Bakersfield, CA, USA) to optimize gas exchange efficiency.

Sets of OA treatment waters (Control, RCP 8.5 at 2100 and RCP 8.5 at 2150) were handled as described in SOP1 (Dickson et al., 2007). OA seawaters were taken from the bubbling buckets with water samples siphoned through a silicone hose, over-filling 250 ml borosilicate sample bottles by at least 50% for experiments and seawater analysis. A separate set of OA waters was prepared for each motility and pH_i experiment and placed in a 20°C water bath [Grant W28, Grant Instruments (Cambridge) Ltd, Shepreth, UK] at least 2 h before experiments for temperature equalization. Each set of bottles was held in a custom Styrofoam insulator to minimize temperature and CO₂ changes during the motility and pH_i experiments. Experimental aliquots were rapidly taken from the mid-water column for analysis. Water temperature (Thermapalm, Thermoworks, American Fork, UT, USA; certified calibrated accuracy 0.05°C) and salinity (YSI 30, YSI Inc., Yellow Springs, OH, USA) were measured before experiments.

Total pH (pH_Total) and total alkalinity (TA) were determined from OA treatment waters for the calculation of seawater carbonate chemistry. All pH_Total analyses were performed within 2 h of sampling; TA samples (250 ml) were preserved with a 50 µl aliquot of saturated mercuric chloride and stored at 4°C in the fridge for later analysis. Spectrophotometric pH_Total determination using m-Cresol Purple (Sigma) was based on SOP6b (Dickson et al., 2007) and Fangue et al. (2010); calibration was performed using Tris standard (SOP3a) (Dickson et al., 2007) and accuracy was confirmed against certified pH reference standards to be 8.104 (Dickson et al., 2007). Open-cell potentiometric TA measurements were run using a Mettler-Toledo T50 automated titrator (Mettler-Toledo International Inc., Columbus, OH, USA) based on SOP3b (Dickson et al., 2007) and Fangue et al. (2010) with the accuracy of TA values confirmed using Scripps certified reference seawater (batch 108 at TA 2218 µmol kg⁻¹).

Carbonate chemistry for OA treatment waters was recalculated from measured parameters (salinity, temperature, pH_Total, TA) to experimental temperature points using the seacarb (v.3.0.8) package in R (v.3.1.2). The seacarb R package was selected based on comparisons of 10 ocean carbonate chemistry packages by Orr et al. (2015) using CO₂ constants as outlined in ‘Guide for best practices for ocean CO₂ measurements’ (Dickson et al., 2007). seacarb uses the recommended formulations for first and second dissociation constants K₁ and K₂ (Lueker et al., 2000), K_f (Perez and Fraga, 1987) and K_s (Dickson, 1990). Additionally, the recommended (Dickson et al., 2007) boron/chlorinity ratio from Uppström (1974) and default values for atmospheric and hydrostatic pressures, silicate and phosphorous content were used. All our experiments were within the specified salinity (19–43 ppt) and temperature (2–35°C) limitations for constants K₁ and K₂ (Orr et al., 2015).

During the austral summer of 2014–2015, adult Evechinus chloroticus Valenciennes 1846 were collected from the shallow subtidal zone (1–3 m depth) at Matheson's Bay (36°18′6.58″S; 174°48′0.71″E) within the Hauraki Gulf, New Zealand. Seawater pH near the collection site on a representative day during the spawning season (28 November) averaged 8.040±0.037 (mean±s.d.; range 7.920–8.109; N=1440) in the Ecklonia kelp forest, and 8.011±0.046 (range 7.822–8.081; N=4320) at the seafloor (C. Blain, SeaFET data).

Adults were transported to the seawater facilities at the University of Auckland within 2 h of collection and maintained in aerated aquaria (80 l) in environmentally controlled conditions (18°C, 12 h:12 h dark:light cycle). Urchins were fed to satiation on seaweed (Ecklonia radiata and Carpophyllum spp.) and spawned using standard procedures (1–3 ml injection of 0.5 mol l⁻¹ KCl into coelom) within 2 weeks of collection. Sperm was collected dry from the aboral surface and stored in 1.5 ml Eppendorf tubes in an iced water bath.

Paired sperm motility and pH_i measurements were made on 13 different males during a 2 week period, as shown in Fig. 1. Sperm quality was visually assessed for high levels of motility using a Sedgewick Rafter slide at 100× magnification. Sperm concentration for all experiments was standardized through haemocytometer counts from a standard dilution of 10 µl dry sperm mixed with 30 ml FSW. From these counts, the volume of dry sperm to give a target sperm concentration of 1 million cells ml⁻¹ for experiments was calculated for each male. Multiple males were run on a single day, with the order of OA seawaters randomly selected for each male. This order was repeated for the paired motility tests, pH_i measurements, nigericin calibration and incubation controls for that male (Fig. 1).

Optimized video capture for sperm motility analysis used an inverted microscope (Nikon Eclipse Ti at 5× magnification) adjusted for differential interference contrast (DIC) imagery, following the recommendations of Boryshpolets et al. (2013) and Castellini et al. (2011) for consistent within-experiment equipment setup and video capture frame rate. Sperm movement was captured in a 250 image Tiff-stack with an ANDOR iXon camera at room temperature (20°C) using proprietary IQ2 software with 2×2 binning to give a capture rate of 34 frames s⁻¹. To minimize the boundary (or wall) effect from glass surfaces (Gee and Zimmer-Faust, 1997; Elgeti et al., 2010) and maximize thermal inertia, we designed a set of custom, deep and relatively large-volume imaging chambers to the size of a standard microscope slide (25 mm×75 mm). There were two 12 mm diameter chambers per plate, constructed from 3 mm thick PVC plastic, with a glued coverslip on the base and a removable coverslip on top. The chamber volume was ca. 339 µl.

Motility recordings were performed as rapidly as logistically possible, with all video recordings started ∼10 s after initiating each sperm dilution. The pre-calculated volume of dry sperm was added to a 14 ml aliquot of OA seawater in a 15 ml Falcon tube and agitated gently for 4 s to make a relatively homogeneous sperm suspension of ∼1 million cells ml⁻¹. To minimize air bubbles when the chamber was closed with the top coverslip, a 350 µl aliquot of the sperm suspension was gently added to over-fill the imaging chamber, and the top coverslip lowered in place to close. The chamber was immediately put on the inverted microscope stage with the focus pre-set for midway between the two coverslips, and the 250 image Tiff-stack acquisition started. Sperm dilutions were mixed in rapid succession producing back-to-back recordings with technical replication of 10 for each CO₂ treatment (Control, RCP 8.5 at 2100, RCP 8.5 at 2150; Fig. 1). Preliminary analyses showed a high degree of technical replication was required to characterize variance within males.

CASA was performed in ImageJ (v.1.50a running on FIJI ‘Fiji Is Just ImageJ’, National Institutes of Health, Bethesda, MD, USA), which produces outputs similar to commercially available systems (Boryshpolets et al., 2013). A customized batch-processing macro (inspired by Purchase and Earle, 2012) incorporating the CASA plugin (Wilson-Leedy and Ingermann, 2007) was used to analyse a 2 s sub-stack (images 18–85) of the original Tiff-stacks. The CASA plugin was parameterized to identify E. chloroticus sperm cells and define motion characteristics for motile and non-motile sperm (see Supplementary Materials and Methods, ‘Motility experiment and CASA parameterization’, Table S1). For each sample, output data on sperm swimming performance included the proportion of motile sperm, and CASA parameters, as described in Table 1.

pH_i of sperm was measured using a single laser flow cytometer and the pH-sensitive dye carboxy-SNARF-1 [5-(and-6)-carboxy SNARF^TM-1, acetoxymethyl ester, acetate; C1272, lot no. 1151593, Invitrogen, Molecular Probes (ThermoFisher), Waltham, MA, USA]. SNARF-1 is ideal for analysing pH_i as the pK_a of ∼7.5 (ThermoFisher catalogue) is approximately the activation pH_i in sea urchin sperm (Christen et al., 1982; 1983; Johnson et al., 1983; Lee et al., 1983), and the dye can be used to measure pH changes between pH 7 and pH 8.

Calibration of SNARF-1 fluorescence measurements used the polyether antibiotic nigericin (Bond and Varley, 2005). Nigericin acts as an ionophore to equalize H⁺ concentrations (Thomas et al., 1979; Hamidinia et al., 2004), allowing the construction of a calibration curve for SNARF-1 fluorescence using pH buffers across the expected pH_i range. However, nigericin also equalizes other cations, including Na⁺ and K⁺ (Pressman, 1976; Riddell et al., 1988; Negulescu and Machen, 1990). Therefore, to limit any nigericin-induced shifts in SNARF-1 fluorescence, resulting in an over-estimation of pH_i (Negulescu and Machen, 1990), we needed to formulate a buffer appropriate for E. chloroticus sperm. Changes in [Na⁺]_i and [Ca²⁺]_i were minimized by replacing NaCl with choline chloride, as choline blocks nigericin-induced across-membrane movement of Na⁺ and Ca²⁺ (García-Soto et al., 1987; Pressman, 1976; Riddell et al., 1988). Nigericin also induces equalization of internal and external [K⁺]; therefore, [K⁺] of the buffers must equal [K⁺]_i of activated sperm (Thomas et al., 1979; Babcock, 1983; Negulescu and Machen, 1990). A review of existing literature for experimentally determined or referenced values of [K⁺]_i indicated a wide range (120–480 mmol l⁻¹ for marine invertebrate tissues) and a lack of certainty for an appropriate [K⁺]_i to use for sea urchin sperm (see Supplementary Materials and Methods, ‘Analysis of internal pH in sea urchin sperm’, and Table S3). Multiple analytical techniques were used to measure [K⁺]_i in E. chloroticus sperm (see Supplementary Materials and Methods, ‘Analysis of internal pH in sea urchin sperm’, Table S2), with calculations based on average sperm volume (1.29 fl; Hudson et al., 2015), average sperm count and average sample K⁺ content.

Full details of the optimization of SNARF-1 incubations, flow cytometer settings and appropriate buffers for determining pH_i in E. chloroticus sperm are provided in Supplementary Materials and Methods (‘Analysis of internal pH in sea urchin sperm’, Figs S1–S3, Table S2). In brief, to minimize activation and metabolism of sperm before experimental incubations, dry sperm were minimally diluted and kept on ice. Every second day, a new 1.25 mmol l⁻¹ SNARF-1 stock was prepared in anhydrous dimethyl sulfoxide (DMSO, Sigma-Aldrich, Merck KGaA, Darmstadt, Germany) and stored in the dark at −18°C between use. A standard 4 µl aliquot of dry sperm was added to 75.4 µl of FSW and 0.64 µl of stock SNARF-1 to give a final dye concentration of 10 µmol l⁻¹ in a brown 1.5 ml Eppendorf tube. The 10 µmol l⁻¹ final concentration for SNARF-1 was based on experiments in E. chloroticus sperm (see Supplementary Materials and Methods, ‘Analysis of internal pH in sea urchin sperm’, Fig. S2A) and recommendations for 1–10 µmol l⁻¹ concentrations by the dye manufacturer (ThermoFisher). A parallel incubation with the SNARF-1 replaced with FSW was used as a control (Fig. 1). Incubations were mixed well by hand and left in ice water for 90 min in the dark.

After the 90 min SNARF-1 incubation, a predetermined volume of sperm (calculated for each male from haemocytometer counts, n=2; Fig. 1) was diluted in 3 ml of OA treatment water to give a target sperm concentration of 1 million cells ml⁻¹. Immediately after mixing, samples were analysed in a flow cytometer (BD Biosciences FACS Calibur, Becton, Dickinson and Company, Franklin Lakes, NJ, USA) with a 488 nm argon laser, optimized for E. chloroticus sperm cells with a combination of forward and side scatter (FSC and SSC, respectively; see Supplementary Materials and Methods, ‘Analysis of internal pH in sea urchin sperm’). Stained sperm were identified using the SNARF-1 emission from detectors FL2 and FL3 (585/42 nm and 670 nm long-pass, respectively) with voltages adjusted for autofluorescence and position on the axes. An event rate of 350–150 events s⁻¹ resulted in a single consolidated population of sperm cells identifiable by FSC and SSC. Data from approximately 5000 events were collected for each sample. As with the analysis of sperm motility, consecutive replicate dilutions (n=10) were mixed for each OA seawater from the SNARF incubation mix (Fig. 1) and run in immediate succession followed by controls for (1) seawater alone, (2) sperm without SNARF-1 and (3) sperm viability confirmed with 1 mmol l⁻¹ propidium iodide (PI).

Flow cytometer output files (.fcs) were processed in FlowJo (v.10.0.8) by gating the single population of sperm cells using FSC and SSC. The SNARF-1 stained cells of this gated population were isolated by setting additional active gates on both FL2 and FL3 histograms based on the unstained control. A small proportion (average for FL2 and FL3 for each sample <5%) of the samples were unstained and excluded from further analysis. The SNARF-1 fluorescence ratio (R) was calculated from FL3/FL2 using the mean fluorescence intensities of the gated populations (e.g. Bond and Varley, 2005). R was calculated for all males in both control and OA treatment waters, as well as the calibration solutions.

pH_i was quantified from the fluorescence ratio R by calibrating pH buffers across the expected pH_i range using the ionophore nigericin, as described above. The use of nigericin calibration is well established (e.g. Weider et al., 1993; Bond and Varley, 2005; Chávez et al., 2020) and preferred for flow cytometry; however, the calibration solution is critical for pH_i determination (Bond and Varley, 2005). Nigericin calibration in E. chloroticus sperm required a buffer [K⁺] of >450 mmol l⁻¹ (see Supplementary Materials and Methods, ‘Analysis of internal pH in sea urchin sperm’, Fig. S3B). A final buffer formulation with [K⁺] of 472.5 mmol l⁻¹ consisted of 22.5 mmol l⁻¹ HEPES-KOH, 450 mmol l⁻¹ KCl and 120 mmol l⁻¹ CoCl (modified from Nakajima et al., 2005). The buffer series was prepared to target pH values (6.81, 7.00, 7.26, 7.51, 7.75, 8.00 and 8.11) with 472.5 mmol l⁻¹ KCl in 3 mol l⁻¹ HCl and 3.472 mol l⁻¹ CoCl to standardize the combined volume added to each buffer. Calibration buffers were stored in the fridge, and 2 h before experiments they were put in the same 20°C water bath as OA treatment waters. A calibration was performed for each male at the end of the pH_i experiment (Fig. 1). A global pH_i calibration curve for all individuals was calculated from the 3rd order polynomial regression using a linear model in the statistical software ‘R’.

Statistical analysis was performed using the software RStudio (v.0.99.486; https://www.rstudio.com/products/rstudio/) and ‘R’ (http://www.R-project.org/), unless specified otherwise, with basic plotting done using the ggplot2 package. Data from seawater pH_Total and pH_i calibrations were separately analysed with a one-way ANOVA. Assumptions of normality were checked with Q–Q plots and Shapiro–Wilk tests, and heteroscedasticity with Levene's test, with a post hoc Tukey–Kramer's test for comparisons between means (α=0.05).

The effect of elevated CO₂ seawater on sperm motility and pH_i was calculated on non-standardized data using the logarithmic response ratio [lnRR=ln(⁠treatment/ control); Nakagawa and Cuthill, 2007; Schlegel et al., 2012; Sewell et al., 2021). Technical replicates were averaged to the level of the individual male and lnRR calculated for all sperm performance measures (CASA outputs from Table 1 and pH_i). Each parameter's lnRR for all 13 males was then boot-strapped in the statistical software ‘R’ running the boot package, and the resulting mean and 95% normal confidence intervals (CIs) returned from 100,000 iterations. The CIs provide a measure of the variance between individuals, and results are interpreted as significant where CIs do not overlap with zero (Schlegel et al., 2015).

For multivariate analysis, all data were standardized to a mean of zero and a variance of 1 to prevent distortion of the analysis from 3 orders of magnitude difference between values for different parameters of sperm performance (Quinn and Keough, 2002). To reduce multi-collinearity, the seven CASA parameters (Table 1) for each technical replicate (N=9–10), OA treatment (N=3) and male (N=13) were condensed into principal components (PC) using a principal component analysis (PCA; PRIMER v.6, Plymouth Marine Laboratory). A Euclidean distance matrix was then calculated for each male/OA treatment based on the average of the technical replicates for PCs and the pH_i. An initial permutational multivariate analysis of variance (PERMANOVA) was performed with two factors (male, OA treatment) using 9999 permutations of residuals under a reduced model. A lack of replication at the lowest level (i.e. due to the averaging of technical replicates to a single value) resulted in the automatic exclusion of the highest level interaction term due to being confounded by the variance of the residuals (Anderson et al., 2008). Multivariate dispersion using PERMDISP tested for homogeneity in the cluster dispersion around the centroids for the three OA treatments. A final analysis used PERMANOVA with pairwise comparisons of the factor OA treatment.

There was a clear and consistent separation between control and OA treatments during the experiments (Table 2), with a significant difference in pH_Total (ANOVA: F_2,9=727.4, P<0.0001). There was a ΔpH of −0.32 pH units between the Control and RCP 8.5 at 2100 and a ΔpH of −0.58 pH units between the Control and RCP 8.5 at 2150 (Table 2). Seawater was undersaturated in aragonite only in the RCP 8.5 at 2150 treatment (Ω_Ar<1; Table 2).

Evechinus chloroticus sperm in all OA treatments swam with a high degree of directionality with a lack of circular paths, suggesting the edge effect on swimming behaviour was avoided. Boot-strap analysis of lnRR showed significant negative effects on sperm motility in both RCP 8.5 treatments (CI did not overlap with zero; Fig. 2). In control conditions, the proportion of sperm that were motile was 0.83 (grand mean for 13 males; Table 3). Sperm in RCP 8.5 at 2100 (pH 7.77) had a reduced proportion of motile sperm (0.74), a decrease in the proportion motile of 0.09 from the control (Table 3). Sperm in RCP 8.5 at 2150 (pH 7.51) showed a further decrease in the proportion motile to 0.64 (Table 3), a decrease in proportion motile of 0.19 from the control (Table 3).

CASA measures of sperm swimming performance were also altered under OA conditions (Table 3). Some measures showed greater variability between males, as shown by the size of the 95% CIs in the lnRR (Fig. 2). In general, sperm performance was significantly negatively impacted by elevated P_CO2 seawaters, except for the positive, but non-significant changes, in the velocity measures VCL and VAP (Fig. 2). The significantly lower LIN (pH 7.77=−0.2, pH 7.51=−0.26) and VSL (pH 7.77=−21.16, pH 7.51=−29.12 µm s⁻¹) suggest that the actual path taken by sperm was significantly more curved, with an overall significant decrease in the distance covered from the starting point (PROG; Fig. 2, Table 3).

All sperm parameters with a significant OA impact showed a maximum negative effect size at RCP 8.5 at 2100 (pH 7.77; Fig. 2). No significant differences in sperm performance effect size were seen between the pH 7.77 and pH 7.51 OA treatments (Fig. 2).

Procedural controls showed no contamination of the OA seawaters (no or very low events of FSC and SSC recorded) and high levels of sperm cell viability for all trials (minimal PI signal indicating intact cellular membranes; see Supplementary Materials and Methods, ‘Analysis of internal pH in sea urchin sperm’). The SNARF-1 fluorescence ratio (R) calibration curve, used for calculation of pH_i, showed a consistent relationship between males, with highly significant differences between R values for each buffer (Fig. 3; ANOVA: F_6,98=656.8, P<0.0001).

Sperm in control conditions had a pH_i of 7.52 (Table 3), which was −0.57 units lower than the seawater in which they were activated (control seawater pH 8.09; Table 2). Sperm in RCP 8.5 at 2100 and RCP 8.5 at 2150 treatments showed a smaller difference in pH_i relative to the pH of the OA seawater in which they were activated. Sperm in RCP 8.5 at 2100 had a pH_i of 7.35 (Table 3), −0.42 pH units below the OA seawater treatment (pH 7.77; Table 1). Sperm in RCP 8.5 at 2150, similarly, had a pH_i of 7.31 (Table 3), −0.2 pH units below the OA seawater treatment (pH 7.51; Table 2).

Sperm pH_i was significantly impacted by OA treatment, with all sperm exhibiting a similar effect size (lnRR) with relatively slight among-male variation (narrow CIs; Fig. 2, Table 3). A negative effect size for pH_i was observed at RCP 8.5 at 2100 (pH 7.77; Fig. 2). However, there was no significant difference in the effect size for pH_i between the pH 7.77 and pH 7.51 OA treatments (Fig. 2), the same pattern as seen in the sperm performance parameters described above.

PCA on the seven CASA output variables (Table 1) resulted in three PCs with eigenvalues >1 that together explained more than 75% of the variation. PC1 was primarily influenced by the directionality of sperm (VSL, LIN and PROG), PC2 by the negative contribution of non-linear measurements (VCL and VAP), and PC3 by the head movement and actual path taken (WOB and VCL, respectively); sperm motility contributed almost evenly to PC1 and PC3 (Table 4). pH_i had an eigenvalue <1 (0.958) and explained a smaller percentage of the variation (14.1%; Table 4).

A principal coordinate analysis (PCO) was then conducted based on PC1–PC3 from the PCA (above) and the independent measurement of pH_i. Most of the variation in sperm performance between OA treatments could be explained by PCO1 (63.4%) and PCO2 (21.3%, Fig. 4). There were no significant differences between males (pseudo F_12,24=0.0031, P=1). However, there were highly significant differences between the three OA treatments (PERMANOVA: pseudo F_2,24=15.329, P=0.0001), but with homogeneous multivariate dispersion [PERMDISP: F_2,36=0.001647; P(perm)=0.9843], suggesting that differences between OA treatments can be attributed to both sperm motility (PCs) and pH_i. Pairwise comparison showed differences in sperm performance in both RCP 8.5 OA treatments compared with control [RCP 8.5 at 2100: t=4.046, P(perm)=0.0001; RCP 8.5 at 2150: t=5.246, P(perm)=0.0001], but not between the RCP 8.5 OA treatments [t=1.319, P(perm)=0.1651].

There was a positive relationship between pH_i and the proportion of sperm attaining activation (proportion motile) in all males (Fig. 5). However, there were significant differences between males in the slope of this relationship (PERMANOVA: pseudo F_12,24=6.0362, P=0.0001; Fig. 5), reflecting male-to-male variation in the response of sperm to OA treatment.

Activation of Evechinus chloroticus sperm in OA-treated seawater had significant negative impacts on sperm motility, four of the six CASA sperm performance measures (VSL, LIN, WOB, PROG) and pH_i. In control conditions, sperm had an activated pH_i of 7.52; however, sperm in both RCP 8.5 OA seawater treatments could not attain a pH_i of this magnitude. Instead, there was a stepped decrease in pH_i from the Control to RCP 8.5 at 2100 (pH 7.77), with no significant difference in mean pH_i in the two OA seawater treatments (RCP 8.5 at 2100 and 2150, pH 7.7 and 7.51, respectively). Using paired data collected from the same individuals, we found a positive relationship between pH_i and sperm motility in E. chloroticus, but with a significant difference in the response between males.

In OA conditions, there was an overall decreased motility and reduced swimming performance of E. chloroticus sperm, but no significant change in through-the-water speed (VCL). OA experiments in other sea urchin species have shown a variable response in VCL: an increase in Paracentrotus lividus (in Ireland, Graham et al., 2015), Psammechinus miliaris (Caldwell et al., 2011) and Lytechinus pictus (Smith et al., 2019), a decrease in Heliocidaris erythrogramma (Smith et al., 2019) and P. lividus (in Italy, Munari et al., 2022), or no significant change here in E. chloroticus.

Sperm of E. chloroticus in OA conditions swam a more curved path, so the straight-line speed between two points was effectively slower (reflected in lower VSL and PROG). Lower directionality (PROG and VSL), combined with a lower proportion of motile sperm, is modelled to negatively impact fertilization success (Vogel et al., 1982; Levitan et al., 1991; Lewis et al., 2002). Specifically, a decrease in the proportion of motile sperm lowers the effective sperm concentration, and decreased swimming speed reduces the rate constant for sperm–egg collisions, β₀, which is estimated from sperm swimming velocity (ν) and the cross-sectional area of the egg σ₀ (Vogel et al., 1982).

In an OA context, the direct impacts of lower sperm motility and/or speed correspond with lower fertilization rates in H. erythrogramma (Havenhand et al., 2008; Smith et al., 2019), L. pictus (Smith et al., 2019) and P. lividus (Graham et al., 2015; Munari et al., 2022), or indirectly through the effect of sperm concentration on fertilization rates in Mesocentrotus franciscanus (Reuter et al., 2011), H. erythrogramma (Schlegel et al., 2012) and Sterechinus neumayeri (Ho et al., 2013; Sewell et al., 2014). The changing parameters of sperm performance may well have further implications for sperm limitation, polyspermy and the plasticity of egg size (Millar and Anderson, 2003; Luttikhuizen et al., 2011; Reuter et al., 2011; Okamoto, 2016). For example, the faster VCL observed for sperm of sea urchins under OA conditions (Graham et al., 2015; Caldwell et al., 2011; Smith et al., 2019) will be a disadvantage in conditions of sperm limitation, as sperm longevity will be limited by more rapid depletion of finite endogenous energy stores (Fitzpatrick et al., 2012).

Broadcast-spawning marine invertebrates in OA conditions will release both the egg and the sperm into a lower pH environment. OA can also impact unfertilized eggs, including changes in pH_i (Bögner et al., 2014) and a reduction in the size of the jelly coat that contains sperm-attracting chemicals (Foo et al., 2018a,b). Theoretical and experimental studies in sea urchins have suggested that larger jelly coats are a greater target for sperm and have a higher fertilization success (Levitan and Irvine, 2001; Podolsky, 2002; Deaker et al., 2019). As the jelly coat is the source of the chemical attractants used in chemotaxis in sea urchins (Miller, 1985; Wood et al., 2015; Ramírez-Gómez et al., 2020), a reduction in the size of the jelly coat, in combination with decreased sperm performance (motility, speed), may have a major impact on fertilization success in OA conditions.

In these experiments, we measured sperm motility after activation in control and OA seawater. As noted above, chemotaxis of sperm towards eggs has been shown in a number of sea urchin species (Miller, 1985; Wood et al., 2015; Ramírez-Gómez et al., 2020), but no information is available on sperm chemotaxis in E. chloroticus. Future work in this species might combine the impacts of OA on sperm performance, pH_i of both eggs and sperm, and the chemotactic behaviour of sperm to eggs in OA seawater to increase understanding of sperm–egg interactions in a future ocean.

Under control conditions, the pH_i of activated E. chloroticus sperm measured with SNARF-1 was 7.52. This pH_i value is consistent with the alkalinization found in other sea urchins using a range of experimental techniques. Fluorescent amine probes, for example, showed a pH_i increase at activation from pH_i 6.9–7.0 to ∼pH_i 7.4 in buffered artificial seawater (Christen et al., 1982; Lee et al., 1983). Nuclear magnetic resonance (NMR) gave a slightly broader alkalinization range, from about pH_i 7.0 to 7.4–7.6 in artificial seawater (Christen et al., 1983; Johnson et al., 1983). SNARF-1 in this study, therefore, provided high-resolution pH_i values that could be used to compare between OA treatments. However, cellular [K⁺]_i in E. chloroticus sperm (466 mmol l⁻¹) is approximately 3–4 times higher than that reported in mammals and at least double that previously determined in sea urchins (Lee et al., 1983; Guerrero and Darszon, 1989; Darszon et al., 2004; see Supplementary Materials and Methods, ‘Analysis of internal pH in sea urchin sperm’, and Table S3). Thus, the use of SNARF-1 in future studies of marine invertebrate sperm needs to use a validated value of [K⁺]_i.

pH_i is suggested to play a fundamental role in activation and swimming in sea urchin sperm. The involvement of CO₂ and pH_i is well described (e.g. Christen et al., 1982; Johnson et al., 1983) and has led to the suggestion that sea urchin sperm will be unable to compensate for OA changes in seawater carbonate chemistry, with consequences for activation and motility (Havenhand et al., 2008; Kurihara, 2008; Graham et al., 2015). The effect of OA on pH_i of sea urchin gametes has recently been demonstrated in the unfertilized eggs of Strongylocentrotus droebachiensis, with a stepped change in pH_i shown at 1000 µatm P_CO2, through changes in uncalibrated fluorescence ratio (R) from the pH-sensitive dye BCECF (Bögner et al., 2014). Under similar conditions in this study, a calibrated stepped decrease in the average pH_i of activated E. chloroticus sperm at pH 7.7 (846 µatm P_CO2) confirmed the suggested significant negative impact of OA.

Why might there be a non-linear decrease in pH_i with external pH (pH_e)? Although we cannot rule out limitations in the measurement of pH_i, the observation of a stepped change in the pH_i of unfertilized echinoid eggs using the dye BCECF (Bögner et al., 2014) and echinoid sperm using SNARF-1 (herein) suggests that there may be as yet undescribed physiological process(s) underpinning these OA-induced changes. For example, an OA-induced disruption of H⁺ extrusion from the cell may occur through the lower activity of the voltage and soluble adenylyl cyclase (sAC)-dependent Na⁺/H⁺ exchanger (Lee, 1985; Neill and Vacquier, 2004). Reduced exchanger activity might lead to an increased reliance on passive diffusion of H⁺ across the cellular membrane (Mohri and Yasumasu, 1963; Hamamah and Gatti, 1998; Missner and Pohl, 2009; Boron et al., 2011). The effects of OA on the maintenance of pH_i may also have knock-on impacts on other cellular processes, including possible involvement in the proposed on/off state of the sperm mitochondrion (Schlegel et al., 2015).

The suggested narcosis of sperm in an elevated P_CO2 world, through the pH_i inhibition of dynein-ATPase activity and bioenergetic pathways, is not supported in this study on E. chloroticus sperm. Although the pH sensitivity of dynein-ATPase has been clearly described (Gibbons and Fronk, 1972; Christen et al., 1986; Neill and Vacquier, 2004), this does not appear to be the primary point of regulation under the tested OA treatments, where there was a non-linear decrease in pH_i with experimental changes in pH_e. A similar stepped pattern is seen in other pH-dependent processes, such as the kinetics of the regulatory enzyme sAC, which is also involved in the initiation and maintenance of sperm motility (Nomura et al., 2005; Vacquier et al., 2014). Future research should also investigate the role of pH_i in E. chloroticus sperm swimming by using OA treatments that approach the pH_i of sperm within the gonads (pH <7.0). Alternatively, oxygen tension may play a more significant role in sperm activation than currently acknowledged in many OA studies (Cohn, 1918; Mohri and Yasumasu, 1963; Webster and Giese, 1975; Chia and Bickell, 1983; Byrne, 2011), or there may be a role of decreased mitochondrial membrane potential in lower swimming speed (Schlegel et al., 2015).

The relationship between OA, pH_i, sperm activation and swimming performance in E. chloroticus does not appear to be straightforward, with variable OA responses between males, as previously seen in other sea urchins (Schlegel et al., 2012; 2015; Smith et al., 2019). At the same pH_i, sperm from some males showed narcosis that led to a lower proportion of motile sperm (Fig. 5), while others were able to activate sperm with an increased swimming speed (VCL; Fig. 2).

There are two points to note about variability in E. chloroticus sperm performance in response to OA. First, all males used in this study were collected from a single location so these results reflect intra-population variability. Adult sea urchins collected from this shallow kelp habitat are exposed to daily pH variability (7.920–8.109; C. Blain, unpublished data; see details in Materials and Methods) but would be unlikely to experience the pH levels used in our OA treatments (7.77, 7.51). Evechinus chloroticus from this location may show potential for genetic adaptation to climate change (temperature, OA), as previously shown in a genotype-by-environment interaction to increased seawater temperatures during early development (cleavage, gastrulation; Delorme and Sewell, 2016).

Second, E. chloroticus is found from the Three Kings Islands (34°10′S) to the Snares Islands (47°60′S), and throughout mainland New Zealand (Barker, 2013). Kapsenberg et al. (2017) have shown that Strongylocentrotus purpuratus collected at sites in Oregon and California with different levels of pH variability due to upwelling show differing sensitivities of fertilization to experimental pH. Sea urchins from sites with frequent exposure to low pH required much lower pH treatments to significantly alter fertilization rates, with greater sensitivity in sperm than eggs (Kapsenberg et al., 2017). A similar broad-scale study in E. chloroticus across sites with different local pH may test the potential for local adaptation in response to OA.

The hypothesis that sea urchin sperm may exhibit a form of narcosis as a result of the elevated P_CO2 and low pH of the surrounding seawater was first proposed by Kurihara (2008). Paired analysis of sperm pH_i and sperm motility in E. chloroticus showed that sperm could not defend pH_i in future OA conditions and that sperm performance was not a linear pH_i-dependent narcosis at low pH_e. Differences in sperm performance, when activated in OA conditions, may also impact fertilization success in E. chloroticus in a future ocean.

We thank A. Turner for assistance with microscopy, T. Brittian for probes, A. Brookes for flow cytometry, J. Havenhand for CASA, R. Gallego and E. Frost for coding and image analysis, and, in the laboratory, M. Clark, D. Baker, I. Ruza and L. van Oosterom. C. Blain generously provided pH data from the collection location. Many thanks to the reviewers and Editor, whose comments improved the manuscript.