We designed a customized video-microscopy imaging system to non-invasively visualize Pacific oyster spat heart contractions, while controlling oyster chamber seawater flow and temperature (Figure 1). The system combined off-the-shelf components coupled with customized components employing the principles of Peltier elements (Peltier, 1834; Kubes et al., 1989; Correges et al., 1998). The latter were designed to provide a feedback control system between the chamber temperature and the Peltier element. In our protocols the system provided 2°C step changes in temperature in less than 80 s, although slower and more rapid changes are possible.

The system provided clear images of the contracting oyster heart (Figure 2 and Video 1). Contraction was routinely identified in real-time by the automated, commercial edge detection system, while non-detection (due to valve motion, feeding, waste clearance etc.) was easily identified by the experimenter for exclusion. Coupled with the data acquisition software, the edge detection system provided inter-beat interval (IBI; time between consecutive cardiac contractions) and HR for subsequent calculation of HRV responses of spat during acute temperature change. The system also provided visual (video) and quantitative means to assess the incidence of asystole (cessation of cardiac contraction). The oyster spat cardiac contraction detection system and its components (Figure 1) can be obtained and assembled by investigators with moderate effort to provide robust detection for the calculation of IBI and HR.

The experimental set-up consisted of three components: an experimental chamber, a customized electronic temperature-controller, and video microscopy coupled with commercial edge-detection technology (Figure 1).

The experimental chamber was contained within a custom-fabricated 8 × 4.5 × 5 cm anodized aluminum block with a 2.5 cm-diameter bore drilled lengthwise through its center. At a depth of 2.5 cm into the bore (measured from upper block surface) a clear watch-glass was secured and sealed in place. Three small (~1–1.5 mm diameter) ports were drilled into three of the block sides (depth of ca. 2 cm from upper block surface) to allow for the insertion of a thermocouple and seawater inflow/outflow tubes (catheter tubing inserted via permanently-affixed syringe tips).

Seawater was maintained at a depth of ca. 1 cm within the experimental chamber (bounded by the watch-glass and walls of the aluminum bore), with inflow and outflow rates of ~2.5 mL/min (inflow: gravity drip; outflow: peristaltic pump and Pressure Servo Control; Living Systems Instrumentation, Burlington). The oyster spat, which were individually placed atop the watch-glass within the water, were trans-illuminated via a Microlight 150 fiber optic light source (Fibreoptic Lightguides Australia; AIS Optical) contained within the aluminum bore below the watch-glass chamber. A custom servo temperature controller maintained chamber water temperature within ±1°C of each desired experimental temperature (T_exp) set-point by means of a heat sink (fan)-coupled Peltier element; feedback was obtained via the thermocouple inserted into the center port of experimental chamber.

The trans-illuminated oyster was visualized from above using a microscope (Zeiss Stemi SV6, Zeiss, Germany) coupled with a digital video camera (SONY Hyper HAD Color Video Camera, Japan). The camera video signal was transmitted in real-time to the computer via a video capture device (Roxio Video Capture – Roxio Central Fx/Roxio Central4; 60 fps) connected to edge detection hardware/software (V94 Video Dimension Analyzer, Living Systems Instrumentation, Burlington, Vermont). Edge detection was used to track cardiac motion in real-time by comparison of pixel color at a reference point in the video image (stationary area with no color change over time; see white line on right of Figure 2 and Video 1) with pixel color at an active area of interest (area of cardiac motion as indicated by significant color change over time; see left white line in Figure 2 and Video 1).

Analog edge-detection and thermocouple data were acquired for analysis via a data acquisition system (PowerLab 4/20, LabChart®7 Pro; ADInstruments, Bella Vista, NSW, Australia). High-fidelity periods of detection were routinely obtained and used for analysis, while occasional segments in which edge detection was compromised were excluded from analysis (for example: short periods of active valve movement, excretion of waste). Detection of IBI (see Figure 2, square waveforms) was standardized to the leading edge of the raw edge detection waveforms, and analysis was performed using LabChart's HRV software module with manual inspection of all analyzed data to confirm accurate detection. We adopted a working definition for cardiac asystole as fewer than 45 cardiac contractions within a particular 10 min protocol segment. This definition reflected the observation that periods of asystole presented as a series of increasingly slow, sporadic contractions at a cooler T_exp, after which complete asystole occurred and was sustained until a certain T_exp was reached during the warming arm of the protocol.

The juvenile spat stage follows the 15–30 day larval stage in the oyster life cycle, with few studies available compared with mature oysters (Albentosa et al., 1994; Beiras et al., 1995). During the spat stage oysters develop the morphology typical of bivalves, comprising a flat upper (right) valve and domed lower (left) valve joined at an anterior hinge by an adductor muscle. The spat is covered by two mantles, which participate in shell formation and sensation, and possesses two sets of gills, the vascular sites of gas exchange, in the mantle cavity (Quayle, 1969). As closure of the valves requires activation of the adductor muscle, bivalves preferentially keep their valves open unless faced with marked environmental changes, including acutely lowered temperature (Bernard, 1983). The myogenic oyster heart consists of two auricles and a ventricle, which supply a poorly-defined circulatory network that bathes tissues in blood that is returned to the heart via numerous veins; pulsating accessory hearts are also present, supplying circulation to the kidneys (Quayle, 1969). Oyster hearts are regulated extrinsically and intrinsically by neural, biochemical, and environmental/physical (e.g., temperature, pressure, salinity) mediators (Koester et al., 1979).

Triploid Crassostrea gigas spat (age: 3.5–4 months, length: 1.1–1.7 mm) were obtained from a commercial hatchery (Shellfish Culture Ltd., TAS; hatchery temperature ~22°C) and acclimated for a minimum of 2 weeks to either 10°C (T_a10, n = 8) or 22°C (T_a22, n = 8). Spat were maintained in aerated housing containers and supplied daily with hatchery-sourced, filtered seawater containing 2–3 million cells/mL of Chaetoceros I (TAS) and Skeletonema pseudocostatum (SA), then individually transferred into the experimental chamber (placement of right valve surface toward the video-camera unit).

The experiment consisted of a temperature ramp (TR) protocol, where the chamber was set to T_exp = 22°C at the onset of the protocol, cooled to T_exp = 10°C, and re-warmed to T_exp = 22°C. This was done in 2°C intervals, with 10 min at each T_exp (Figure 3). The design of the chamber ensured oxygenation and food supply throughout the protocol (as above).

Heart rate, Q10, and time- and frequency-domain HRV analyses were calculated for the data using Excel. During the TR protocol we noted that the spat routinely displayed asystole as a component of the physiologic response to cool T_exp and, therefore, quantified the occurrence of asystole for each acclimation temperature (T_a) group.

Preliminary HRV analysis was applied in the time and frequency domains using scaling factors similar to human HRV, since spat HR was similar to human values. As a result, the scaled time-domain analysis parameter, pNN50, was used across the tested T_exp. The four time-domain parameters studied were: mean IBI, SDNN (standard deviation of normal IBIs), RMSSD (the root mean square of successive differences of successive IBIs), and pNN50 (the percentage of normal IBIs differing from their preceding IBI by more than 50 ms). Only oysters with active cardiac contraction (i.e., not in asystole) were considered in HRV analyses.

Frequency-domain HRV analysis of IBIs was performed using Kubios software (Kubios HRV version 2.1, University of Eastern Finland, Kuopio, Finland; Tarvainen et al., 2014). Our analysis revealed that oyster spat power spectra failed to display clear peaks aligned with the low- and high-frequency domain peaks used to justify frequency domain analysis in mammalian species (Thireau et al., 2008). As a result, we restricted our HRV analyses to the time domain.

Statistical analyses were performed using SAS V9.4 (SAS Institute Inc., Cary, NC, USA). In order to assess the acute cardiac response to temperature change and the impact of T_a, which reflects the transition from hatchery to open aquaculture environment, we compared rates of asystole within acclimation groups. Rates of asystole were compared between T_exp = 10°C and T_exp = 22°C by McNemar's exact test. The difference in asystole rates at 10°C between acclimation groups was tested by Fisher's exact test. Continuous outcomes (HR and HRV parameters) were modeled by the linear mixed effect model estimated by restricted maximum likelihood as implemented by the MIXED procedure of SAS V9.4. The model included fixed effects for acclimation group, temperature, phase (warming vs. cooling), and an acclimation group by temperature interaction to allow comparison of temperature slopes between acclimation groups. The model accounted for the repeated serial measures on spat by including a random effect for spat and allowing for within-spat first order auto-correlated residuals. Q10 values between and within groups were assessed via 2-way ANOVA on ranked data with Shapiro-Wilk and Brown-Forsythe tests. Mean values were reported as mean ± SEM. SEM was not presented in cases where, due to asystole, n < 4 for a given group. All tests were two-sided without correction for multiplicity.

Oyster spat routinely adopted cardiac asystole during cooling, and the occurrence of asystole was inversely related to T_exp (Figure 4). During cooling from 22°C to 10° C the percentage of spat in asystole increased from 0 to 100% in the T_a22 group (P = 0.008) and from 0 to 38% (P = 0.25) in the T_a10 group. The 62% difference in asystole at 10° C was statistically significant (P = 0.026). Figure 4 shows that cardiac activity ceased at higher T_exp during the cooling phase than the T_exp at which it resumed during warming in T_a22 spat.

Mean spat HR was positively associated with T_exp in both acclimation groups (Figures 5, 6). In T_a22 spat the average initial HR at 22° was 83.2 beats/min (95% CI, 70.8–95.7) and this decreased by an average of 7.3 beats/min (5.9–8.7, P < 0.001) per degree Celsius during cooling (when asystolic spat, as defined in Methods and Materials, were included in calculations as HR = 0). In T_a10 spat the mean initial HR at 22° was 78.6 beats/min (66.2–91.1) and this decreased by an average of 5.1 beats/min (3.7–6.5, P < 0.001) per degree Celsius during cooling (including asystolic spat as HR = 0). The HR of T_a10 spat at the initial 22° acute temperature was not different from the T_a22 spat (difference in HR between T_a10 and T_a22 = 4.6; −11.7 to 20.9; P = 0.57), however, they did exhibit a lower rate of decrease during cooling (difference = 2.2; 0.3–4.2; P = 0.026; this included asystolic spat as HR = 0). Excluding spat in asystole, the average decrease in heart rate per degree Celsius while cooling was 5.5 beats/min (4.6–6.3, P < 0.001) within T_a22 compared to 4.3 beats/min (3.7–4.9, P < 0.001) within T_a10 with a significant difference in slopes between acclimation groups (being 1.1; 0.1–2.2, P = 0.032).

Figure 5 shows that in the T_a10 group the T_exp specific HR was similar during cooling and warming regardless of inclusion of asystolic spat. However, in the Ta₂₂ group, the HR appears to have dropped more quickly during cooling due to the increased proportion of spat in asystole. The cooling and warming patterns in T_a22 spat did not vary substantially when restricting analysis to spat that maintained active cardiac contraction (i.e., excluding asystolic spat from analysis). Figure 6 (left panel) shows that among spat with active cardiac contraction, T_a22 and T_a10 had a similar HR response to T_exp.

Q10 values for mean HR were calculated in the cooling and warming phases for spat in each acclimation group (mean ± SEM). There were no significant differences in Q10 between T_a groups, between the warming vs. cooling phases within T_a groups, or interactions between T_a and phase (Q10 values: T_a22 – cooling phase: 2.6 ± 0.3, warming phase: 2.4 ± 0.2; T_a10 – cooling phase: 2.4 ± 0.2, warming phase: 2.7 ± 0.3; all P > 0.05, 2-way ANOVA).

There was an inverse relationship between each of IBI, RMSSD, SDNN, and pNN50 with T_exp in each acclimation group (after excluding spat in asystole, all P < 0.01 for association with T_exp; Figure 7). At any given T_exp RMSSD and pNN50 were significantly higher in the T_a10 group than the T_a22 group (both P < 0.001), but there were no differences between T_a10 and T_a22 groups for IBI (P = 0.11) or SDNN (P = 0.13; both Figure 7).

Our system takes advantage of the relatively stationary nature of bivalves and the translucence of spat shells. The experimental chamber enabled rapid changes in T_exp during which cardiac contraction and HR could be assessed relative to T_a. The system can provide rapid as well as slow temperature changes, which contrasts with existing reports, where experimental chambers are placed or immersed within large, temperature-controlled water baths to make cardiorespiratory or metabolic measurements (Frappell and Mortola, 2000; Polymeropoulos et al., 2014).

Adult oyster hearts have previously been studied in vitro and in vivo (Greenberg, 1965; Greenberg et al., 1980; Andrewartha et al., 2016), largely through invasive means that disrupt valve integrity to view/record from the exposed heart and negatively influence outcomes (Veitch, 1974; Koester et al., 1979). For example, HR has been measured by incident laser beam reflection from the heart through a several mm diameter hole (Ritto et al., 2001, 2003; Hellicar et al., 2014). Ultrasound stethoscopes (Dieringer et al., 1978) or infrared diodes/phototransistor detectors (Depledge and Andersen, 1990; Depledge et al., 1996; Curtis et al., 2000) have been reported to measure bivalve HR non-invasively. However, the large hardware components make this incompatible with spat (< 2 mm) and these approaches have limited means for verifying that only HR is being detected. By contrast, the thin valves of young spat are amenable to trans-illumination, eliminating physical stressors and attachments, while video microscopy provides an easily verifiable HR signal for the experimenter. Our automated, non-invasive method could be applied to younger or older spat once the limits of microscopic magnification and shell calcification (i.e., animal maturity and size) are defined, or in other species amenable to trans-illumination (e.g., other invertebrates, alevin, Polymeropoulos et al., 2014; eggs, Mortola et al., 2010, 2012). We used the minimum light intensity necessary for cardiac visualization; however, near red light sources could be used to determine/minimize the potential impact of light exposure on spat physiology. The system could also be adapted for use “in the field” in open, aquaculture environments.

Oyster growth is most rapid at warm temperatures with high food and oxygen availability (Walne and Spencer, 1974); however, the cardiac control phenotype during temperature challenge was not known. As cultured oysters are initially reared in hatcheries providing constant, ideal conditions, our data on the impact of acute changes in variables such as T_exp, such as those encountered upon initial ocean exposure, suggest ways to measure/improve species health and production.

The asystole induced during cooling was inversely related to T_exp, and was more prevalent in T_a22 spat (8/8 asystolic at T_exp = 10°C; Figure 4). Residual beats were generally observed at the onset of a cooler T_exp, after which complete asystole occurred and was sustained until T_exp increased during the warming arm of the TR protocol. This is consistent with a report for the adult blue mussel (Mytilus edulis), where asystole occurred at < 8°C if T_a = 18°C (Kittner and Riisgard, 2005), although spat achieved asystole at much higher temperatures (all >10°C; Figure 4). Similar to spat, cool-acclimated mussels maintained cardiac activity at lower temperatures than warm-acclimated ones (Pickens, 1965; Braby and Somero, 2006). Thus, lowering the hatchery water temperature in anticipation of placement in ocean oyster beds would reduce periods of asystole and minimize potential reductions in survival or growth. T_a22 spat became asystolic at warmer T_exp during cooling than the T_exp at which they regained HR during warming, which may represent a compensatory HR response to clear accumulated metabolites/provide tissues with oxygen and nutrients (Koester et al., 1979; Bernard, 1983). Some conflicting reports exist, with asystole in adult clams (Noertia ponderosa) reported to occur at < 5°C but with cardiac activity (HR) only resuming when >15°C (DeFur and Mangum, 1979); this may be attributable to inter-species variation or differences in protocol and acclimation timing. Examining the time-course of acclimation (i.e., resumption of HR in asystolic T_a22 spat during sustained cool temperature), the ontogeny of the responses in spat of varying ages, and outcomes of growth and time to maturity would be valuable.

Asystole was accompanied by valve closure, which is typically a period of quiescence that may be influenced by food availability, decreased temperature (Koester et al., 1979; Higgins, 1980; Curtis et al., 2000), decreased metabolism with reduced growth (Bernard, 1983), or environmental stressors (Kittner and Riisgard, 2005; Braby and Somero, 2006). Our observations of sporadic residual beats just prior to valve closure due to cooling are similar to reports of the impact of the stressor/pollutant, copper, where increasingly sporadic HR in mussels is present prior to asystole (Curtis et al., 2000). Changes in HR preceding valve closure suggest neural mediation for this phenotype, rather than one in which valve closure, decreased ventilation, and eventual hypoxia trigger decreased HR (Trueman and Lowe, 1971; Koester et al., 1979). Since we maintained equal food availability and water oxygenation, our data for oyster spat likely reflect control mechanisms governed by temperatures lower than the temperature of acclimation.

Consistent with being ectothermic (Widdows, 1973; Braby and Somero, 2006), HR was related to T_exp in all spat (Figures 5, 6). The magnitude of the impact of T_a on HR depended on whether asystolic spat were included. In contrast to asystole, there was no statistically significant impact on HR between T_a22 and T_a10 spat. Interestingly, this was despite what appeared to be a modest tendency toward decreased HR in T_a10 when asystole was excluded (Figure 6, left panel).

Time- and frequency-domain HRV analyses are used to investigate autonomic cardiovascular control (ESC/NASPE Task Force, 1996; Tarkiainen et al., 2005; Thireau et al., 2008). Myogenic molluscan hearts have diffuse pacemaker properties, receive modulatory input from extrinsic cardiac nerves, and are regulated by a combination of neural, biochemical, and environmental/physical (e.g., temperature, pressure, salinity…) mediators (Divaris and Krijgsman, 1953; Trueman and Lowe, 1971; Trueman et al., 1973; Lowe, 1974; Irisawa, 1978; Koester et al., 1979; Ritto et al., 2005). HRV analysis assumes that beat-to-beat variations in HR (i.e., inter-beat intervals, IBIs) are effected by neural modulation (Sandercock and Brodie, 2006; Chattipakorn et al., 2007; Domnik et al., 2012). There is debate on whether to correct for non-cardiac physiologic rhythms (e.g., respiratory sinus arrhythmia) when performing mammalian HRV analyses; however, many studies employ raw HR in the calculation of HRV (Tzeng et al., 2003; Thireau et al., 2008). HRV application to invertebrates is rare, and the relative neural, physical and/or hormonal modulation of HR in oysters has not been fully elucidated (Greenberg, 1965; Quayle, 1969; Mayeri et al., 1974; Liebeswar et al., 1975; Hill and Yantorno, 1979; Koester et al., 1979; Greenberg et al., 1980). Thus, we adopted an approach that performed HRV calculations based on raw HR data, analogous to that employed in mammals.

Few studies have explored HRV in bivalves (Curtis et al., 2000; Ritto et al., 2005). Our analyses revealed none of the defined spikes in post-spectral density (PSD) at specific frequencies as seen in mammals, which is an interesting biologic observation that precluded the application of frequency domain HRV analysis (ESC/NASPE Task Force, 1996; Curtis, 1998; Curtis et al., 2000; Thireau et al., 2008). Mussels have also been reported to have no spikes in PSD (Curtis, 1998; Curtis et al., 2000). The hypothesis of independent control of HR, distinct from ventilation or other physiological functions, is one that deserves further attention. Time-domain HRV analysis (IBI; SDNN; RMSSD; pNN50) applied to oysters not in asystole revealed that all HRV variables were inversely related to T_exp (Figure 7). This is similar to findings in mammals, where HRV is inversely related to HR (Billman, 2013; Sacha et al., 2013a,b). It is also reminiscent of the response reported for toxin exposure (e.g., copper), where an inverse relationship between coefficient of variation (variability) and HR was present in mussels (Curtis et al., 2000). Both acclimation groups had low HRV at T_exp of 22°C, consistent with low HRV in mussels at control conditions (Depledge et al., 1996; Curtis et al., 2000). This varies from the marked HRV in resting mammals (ESC/NASPE Task Force, 1996; Thireau et al., 2008; Domnik et al., 2012), and may reflect a greater scope for beat-to-beat neural regulation of HR in mammals. T_a10 spat had increased HRV compared to the T_a22 group (RMSSD and pN50).

Our findings raise the question of how best to compare and interpret HRV in invertebrates and mammals. Baseline HRV differs between invertebrates and mammals; however, the inverse relationship between HR and HRV remains (Depledge et al., 1996; Kazuma et al., 1997; Curtis et al., 2000; Gong et al., 2004). In mammals, HRV reflects overall autonomic balance and decreased HRV is a negative prognostic indicator (Sandercock and Brodie, 2006; Chattipakorn et al., 2007); one possible interpretation of our findings is that increasing HRV at decreased T_exp may suggest the inverse for invertebrate species. We speculate that the adaptive thermal compensatory response to cold T_exp and T_a may increase HRV in aquatic invertebrates, similar to copper (Curtis et al., 2000) or starvation (Depledge et al., 1996). Further work is required to elucidate the details of this relationship, since T_a10 spat were less likely to become asystolic at cool temperatures despite exhibiting higher HRV than T_a22 spat with active HR. The potential predictive power of HRV could be tested using the impact of known stressors (including low food availability and Pacific Oyster Mortality Syndrome (POMS)), which negatively impact on the physiology, growth, and viability of oysters (Widdows, 1973; Duthie, 2014; Andrewartha et al., 2016). The impact of environment on oyster outcomes has been approached through the development and application of habitat suitability indices (HSI); however, outcome measures typically rely on mortality/viability as the primary outcome (Brown and Hartwick, 1988; Theuerkauf and Lipcius, 2016). Evaluation of oyster HR or HRV is the physiologic complement to HSI assessments and has the potential to provide insight into animal health prior to overall mortality measures, as well as to allow the detection of acute environmental changes. Indeed a recent report in the adult oyster suggests that biosensors could monitor HR and health in “sentinel” animals (Andrewartha et al., 2016). This concept is similar to those proposed for other species, such as zebrafish, as biological models or pharmacologic screens (Liu et al., 2016; Rahbar et al., 2016). HRV has been used to explore cardiac regulatory mechanisms, which could be employed to assess cardiac phenotype between triploid and diploid spat or other marine invertebrates. Acclimating oyster spat to lower temperatures prior to release in ocean oyster beds may enhance adaptability to temperature-variable environments, and provide the means to test for early signs of adaptive or compensatory responses to negative environmental changes.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.