Study animals comprised two species of Australian spiny lobsters. Cold-temperate southern rock lobsters (Jasus edwardsii, body mass 1074.6 g [976.0-1173.2], 12 males, 7 females) were caught from the Taroona marine research reserve at Crayfish point, Tasmania, using baited lobster pots from 5–9th February 2018. Subtropical eastern rock lobster (Sagmariasus verreauxi, body mass 1138 g [1054.1-1222.6], 12 males, 7 females) were obtained from commercial rock lobster fishermen (from Triabunna marina, Dover Seafoods and Leale Fishing, St Helens, Tasmania) from August 2017 till February 2018, who caught S. verreauxi as bycatch in coastal waters of eastern Tasmania. Lobsters were transported in ice-chilled polystyrene boxes to the IMAS Taroona research facilities, and each species kept separately in two 4000-L fiberglass outdoor tanks, supplied with flow-through seawater till the beginning of acclimation experiments.
Their distribution ranges from southern Victoria, around Tasmania and across South Australia into Western Australia as well as New Zealand waters for J. edwardsii, and along the east Australian coastline between Brisbane and the North-East coast of Tasmania, including the northern waters of New Zealand for S. verreauxi (Fig. 1). Both species live at depths ranging from 5–200 m at the Australian continental shelf 37. Natural temperature ranges are ~ 10.8–17.5°C for J. edwardsii (Ion Pot, Tasmania 2,88) and ~ 12–28°C for S. verreauxi 89. Animal ethics were not required at the time of experimentation.
Animals were acclimated to three temperatures for at least eight weeks. Acclimation temperatures were i) the annual average- (14.0°C, [14.1–14.2], n = 46,764), ii) the average summer- (17.5°C, [17.4–17.5], n = 7,988, for the two warmest months February-March) and iii) the average summer sea water temperature predicted for the East-Coast of Tasmania in 2070 (21.5°C, predicted temperature increase of 4°C based on IPCC AR5 scenario RCP8.5 following Hobday 2018 personal communication and the A1FI scenario in Pecl et al. 2009 46). Temperature means were calculated from temperature records at 9 m depth, Ion Pot, Tasmania from 2006–2016 2,88 (data and R script in Supplementary File S1). As only two lobsters could be examined within 48 hours, lobsters were acclimated, sequentially in pairs of one lobster for each species. This assured similar acclimation times at the time of the first measurement and further reduced variability introduced by parallel respirometry setups and handling by multiple experimenters. Lobsters were added randomly to acclimation tanks while accounting for balanced body mass distribution between acclimation groups (ANOVA, F(1, 38) = 0.78, P = 0.384) and species (ANOVA, F(1, 38) = 1.16, P = 0.289). Although we were unable to obtain equal numbers of females and males, we assured even distribution of sex across species and acclimation groups (Supplementary File S3).
Lobsters were housed in identical coated glass-fibre tanks (W × L × H, in cm, 100 × 100 × 75) for each acclimation temperature, and prior to each addition, transitioned to the new acclimation temperature overnight in a separate 300 L tank at a temperature change rate of 0.5°C per hour. Both species shared one tank but were physically separated by an oyster mesh barrier (Supplementary Figure S2). Tanks were equipped with two-level oyster mesh dwellings (W × L × H, in cm, 40 × 50 × 50), one for each species, to provide shelter and additional vertical space (Supplementary Figure S2). Each acclimation tank was filled up to 0.65 m³ and supplied with 14.0°C ([14.0–14.0], n = 784,797) cold sea water, temperature controlled via an external heat-chill unit. Additional submersible 2000 W titanium heaters maintained the water temperature in the summer temperature acclimation tanks at 17.5°C ([17.5–17.5], n = 750,404) and 21.4°C on average ([21.4–21.4], n = 775,440) respectively. The temperature for each tank was continuously monitored and logged using open-hardware components 90, i.e. an open-source microcontroller (Arduino Uno R3, Italy), three waterproof digital temperature probes (DS18B20, China) and a SD card logger module (Adafruit data logger shield, USA), fitted in a custom designed 3D printed plastic enclosure. Two RGB flood lights supplied weak blue light to illuminate acclimation tanks under a 12:12 light:dark photoperiod, controlled remotely via infra-red LEDs and a custom programmed open source microcontroller (Arduino Uno R3, Italy).
Acclimation vessels received filtered and disinfected (sand filtration, foam fractionation, ozonation and activated carbon filtration) flow through sea water at an exchange rate of the tank volume approximately every two hours (flow rate ~ 300 L h-1) and regular removal of debris and left-over mussel shells every 2–3 days. Previous testing showed that water quality parameters such as nitrate, ammonia or heavy metals were below critical levels at those water exchange rates.
Animals were fed on Monday, Wednesday, and Friday with live blue mussel (Mytilus galloprovincialis) or frozen, chopped sardines (Sardina pilchardus). Lobsters that moulted were not used for experiments and allowed to recover for at least two weeks. Three southern rock lobsters and two eastern rock lobsters had to be replaced due to excess moulting stress and cannibalism following moulting. At the end of all experiments, we took final measurements of body mass, body volume, carapace length, carapace width, blood pH and Brix index, and returned lobsters to IMAS aquaculture holding tanks.
Metabolic rate measurements
Once each lobster pair completed its respective acclimation period and was starved for at least 48 hrs, we measured oxygen consumption rates using intermittent respirometry over a 48–72 hr period. Measurements followed a repeated measures design, starting at the respective acclimation temperature for each lobster pair, followed by measurements at the other two acclimation temperatures. This cross-over design allowed to compare acclimation effects at identical measurement temperatures. To avoid carry-over effects by repeated experiments, animals were given at least two weeks’ time between experiments to recover in their original acclimation tanks. In addition, experiments with a large difference between acclimation- and experimental temperature (e.g. 14.0°C ◊ 21.5°C) were performed last and included a short, stepwise acclimatization period to the final experimental temperature for at least three hours prior to measurements to reduce acute temperature stress. To minimise effects by capture and transfer stress immediately before experiments, each lobster was placed overnight in a covered bucket floating in the same acclimation tank. The next morning, lobsters could be moved directly from the bucket to the chase tank with minimal visual or physical disturbance by the experimenter, to assure lobsters were fully rested before chasing.
In the 300 L chase tank (W × L × H, in cm, 120 × 60 × 50), lobsters were exhausted to fatigue, to allow subsequent measurements of maximum metabolic rate (MMR) and aerobic scope in the respirometer. Due to variable effectiveness of single chasing methods among individuals we applied the following multi-step chase protocol to assure that lobsters were fully exhausted: i) touching of antennae and ii) gentle pressing of the ventral soft tissue, located between the last pair of pereiopods and the abdomen, to trigger tail flips and iii) lastly turning of the lobster on its back till it failed to turn back within 60 secs three times. Steps i) and ii) were considered completed till lobsters failed to respond five or more times to the respective procedure. Each chase procedure was video recorded using a camera (GoPro Hero5, GoPro Inc., USA) mounted on top of the chase arena with a flexible gooseneck clamp. A waterproof RGB multi-colour LED strip (5 m 5050 RGB, 60 SMD LEDs / m, Brightness: 900 LM, China) layered 20 cm below the translucent bottom of the chase tank, illuminated the chase arena with yellow light and provided a sharp contrast for the subsequent video analysis. The temperature of the chase tank was set to the experimental temperature using a 2000 W titanium heater modified with a programmable PID (proportional–integral–derivative) controller (SmartPID, Arzaman, Italy). The heating was performed via an external buffer tank, connected to a recirculation pump, to allow an obstruction free chase arena.
Immediately after chasing, we measured the lobster’s oxygen consumption rates (MO2) as a measure of aerobic metabolism using intermittent flow respirometry. For this, we placed lobsters into two cylindrical 10 L custom-made Perspex respirometry chambers (L×D in cm, 66 × 15), tail first to prevent blocking and injury due to the lobster’s guarding posture 91. Lobsters could gain traction to oyster mesh added to the bottom of the respirometer, held in place with an open cut piece of plastic pipe (see Fig. 7A in Oellermann et al. 2020 50). Chambers were sealed within two minutes after addition of lobsters.
Oxygen concentration was measured every 10 seconds using a fibre optic two-channel oxygen meter (HQ40d, Hach, USA) and oxygen probes positioned into an external recirculation loop. Re-circulation pumps provided continuous mixing of water via Tygon® tubing within respirometers at a flow of 1200 L min− 1. Following a six-minute respiration cycle, flush pumps re- oxygenated chambers at a flow of ca. 1500 L min− 1 for eight minutes, using a time controlled digital recycling timer (DRT-1, Sentinel, USA). The two adjacent respirometers were housed in a buffer tank (W×L×H in cm, 102 × 52 × 50), filled with 190 L filtered and filtered flow-through sea water at a flow rate of 130–150 L h− 1. The water temperature of the buffer tank was maintained at the respective experimental temperature using a 2000 W titanium heater. An air stone ensured homogenous mixing and aeration. Black building foil covering the experimental setup and corrugated plastic sheet between respirometers prevented the lobsters from being visually disturbed during the trials. A yellow LED flood light illuminated the setup permanently to reduce dark-induced activity of the nocturnal lobsters 92. After completion of respirometry, we returned lobsters to their original acclimation tanks. After each experiment, the chambers and buffer tank were cleaned and flushed with fresh water. Preliminary tests confirmed appropriate mixing of water in the chamber, the lack of leaks (i.e. dye test) and appropriate flush/respiration cycles.
Oxygen consumptions rates were calculated as in Svendsen et al. 93. Oxygen saturation in respiration chambers decreased to a minimum of 86.6% ([85.8–87.3], n = 108) on average, and never fell below 77%. Individual body volume of lobsters was accounted for in MO2 calculations and measured as the volume overflow in ml after adding lobsters to a water levelled container. Body density did not differ significantly between species (ANOVA, F(1, 45) = 2.31, P = 0.136) and averaged 1.12 g ml− 1 ([1.10–1.14], n = 46). Background respiration was recorded before and after each experiment and accounted on average for 4.5 % ([3.9–5.1], n = 105) or 5.0 % ([4.3–5.6], n = 105) respectively of standard metabolic rate. Maximum metabolic rate (MMR) was the single highest MO2 measured over the entire experiment, and routine metabolic rate the mean MO2 following a 16 hr recovery period till the end of the trial (16 hrs was observed to be the maximum time all lobsters required to recover). Standard metabolic rate was calculated as the mean of the 10% lowest MO2 values 33,94. The time at which MO2 fell three times below routine metabolic rate + one standard deviation after exercise, was marked as recovery time. Excess post-exercise oxygen consumption (EPOC) was calculated as the area under the MO2 curve from the start of MO2 measurements till recovery time. Aerobic scope was calculated as maximum metabolic rate – standard metabolic rate 33,94. Recovery rate was calculated as EPOC divided by recovery time and expressed as mg O2 per kg body mass and hour.
Data and statistical analysis
We analysed the chase videos using a customised Python script (Python 3.8, Supplementary File S1) to record total number of escapes and escape speed. Here we marked individual lobster positions (by mouse clicks) before and after each tail flip and saved the x and y pixel positions and the respective video frame number in a csv file. All further data processing and statistical data analysis was performed using R statistical software 95 and RStudio 96. In R we calculated the Euclidian distance Eq. (1) and divided this by the number of video frames for each corresponding escape event, times the recording frame rate (24 frames sec− 1) and a conversion factor to scale pixels to cm to obtain the escape speed (cm sec− 1, R scripts in Supplementary Files S2).
For the statistical analysis we employed linear mixed effect models to test the effects of acclimation- and experimental temperatures on each of the measured variables (MO2Standard, MO2Routine, MO2Max, aerobic scope, fractional aerobic scope, EPOC, recovery time, escape speed and total escapes) using the lme4 package 97. For all models, we included acclimation temperature, experimental temperatures and species and their three-way interaction as fixed factors. As random effects, we had lobster ID as intercepts, to account for individual variation between lobsters. Initially we constructed a more complex model for each response variable using maximum likelihood estimation (ML), which included sex, body mass, respirometer chamber, experimental duration, and acclimation time as additional fixed factors. We then identified the simplest model using a stepwise backward elimination process of fixed factors, via Sattethwaite's approximation of p-values, using the lmerTest package 98. We then re-fitted the final model using restricted maximum likelihood (REML) estimation and performed post hoc pairwise comparisons with Bonferroni correction between levels of acclimation temperature and experimental temperature using the emmeans package 99. We assessed the linearity, homoskedasticity and normality of residuals using residual plots and the Shapiro-Wilk normality test respectively, to test if the data meet the linear mixed effect model assumptions. If data failed to meet the model assumptions, data were log (SMR) or square transformed (aerobic scope). Relations between escape metrics and EPOC or maximum metabolic rate were tested separately for each species using Pearson correlation, if data passed the Shapiro-Wilk normality test. All values in the manuscript are expressed as mean and the 95% confidence interval in squared brackets. Full model details and formulae are available as R Markdown files in the Supplementary Files S2. The complete data set can be found in Supplementary Table S1.