Short-term, sub-lethal response variables are increasingly used to provide rapid indications of whole organism responses to future climate conditions. Accumulating evidence suggests, however, that these response variables may not consistently reflect whole organism responses which manifest over longer time scales. Here, we consider the effect of moderate warming on longer-term whole organism fitness, as reflected by survival, as well as two shorter-term response variables, feeding rate and oxygen consumption, for two tropical gastropod species. We found a significant reduction in survival under warming, despite no significant effect of warming on feeding or oxygen consumption rates. This result demonstrates that the maintenance of physiological rates alone is not sufficient to sustain organism survival under elevated temperatures; rather an increase in physiological process rates is likely required. Consequently, isolated short-term physiological processes may not adequately reflect longer-term whole organism responses to altered climate. For improved understanding, both short- and long-term responses need to be considered.

Experimental design and set-up

To determine the effects of warming on tropical gastropods, we quantified the response of two species (Chlorostoma argyrostoma, Lunella coronata) to experimentally manipulate temperatures in a laboratory-based tank experiment. We considered adults of these species (maximum shell diameter C. argyrostoma 22.2 mm ± 0.4 mm, L. coronata 24.3 ± 0.5 mm; mean ± SE), which typically live to be between 5 and 10 years of age, although they can live for 30 years (Walsby 1977, Cooper & Shanks 2011). We exposed experimental tanks to current or future temperatures for six weeks from July to August 2019. The ‘current’ temperature (25.5 °C) was based on long-term measurements of the collection site in the summer experimental period (25.4 ± 0.2 °C was the average temperature over the experimental months, i.e. July – August, for the last 10 years of data published, i.e. 2009 – 2018, at the site nearest our collection location, i.e. Eastern Buffer station EM2; Hong Kong Environmental Protection Department 2019), while the ‘future’ temperature (+ 2.5 °C) simulates ocean temperature anticipated between the years 2081-2100 (Stocker et al. 2013). We used six replicate tanks for each treatment, with replicate specimens of each gastropod species in each tank (n = 3 of each species per tank).

Collection of gastropods and holding conditions

We collected adult gastropods (C. argyrostoma, L. coronata) and algae (Gelidium spp.) in early July at Joss House Bay in Tai Miu Wan, Hong Kong (22.27°S, 114.29°E). We then brought experimental organisms back to the Simon F.S. Li Marine Science Laboratory, Hong Kong, and cleaned all gastropod shells of epibiotic organisms.

To allow for acclimation, prior to the experimental period we kept all gastropods and algae in separate holding tanks (19 L seawater; n = 10 individuals per tank) filled with natural seawater under current conditions (temperature = 26.1 °C, salinity = 35, pH = 8.03, 12:12h light:dark regime) for seven days. During this period, we provided gastropods algae collected from the field to graze upon.

Experimental treatments

Following acclimation, we transferred individual gastropods into experimental tanks (n = 3 gastropods of each species, 6 total individuals per tank) where each were held isolated from the others in a cage, with the same conditions as in the holding tanks described above. We provided each gastropod a piece of algae for food which was replenished weekly throughout the experimental period. Half of the tanks we maintained at current temperatures (n = 6), while gradually increasing temperature of the other half to the future treatment (n = 6). The water temperature in the ‘future’ treatment tanks was warmed at a rate of 0.5 °C per hour to be 2.5 °C hotter than the ‘current’ treatment tanks on the day before the experimental period. This rate was used to minimise the chance of mortality to result from initial heat shock (the first gastropod mortality was recorded 9 days after establishment of treatment conditions). Bar heaters were used inside of the tanks to establish and maintain experimental temperatures. iButtons monitored temperature in each tank, where they took recordings every 10 minutes (Thermochron iButton logger, Embedded Data Systems). Mean water temperature was significantly different between the current and future tanks (mean ± SE, current 25.7 ± < 0.1, future 28.2 ± <0.1 °C, F_{1, 10} = 98.22, p = <0.0001). We maintained water quality by removing faecal matter and renewing one-third weekly with fresh marine water.

Experimental responses

Responses of gastropods to experimental treatments were quantified at the end of the experiment in terms of survival, algal consumption (feeding), and oxygen consumption.

We measured survival on the 39^th and 40^th days of the experiments by quantifying the number of living gastropods remaining (this was the same as the number used in concurrent oxygen measurements). During the experimental period we regularly checked the snails and removed any dead individuals which were identified by probing the gastropods to prompt a reaction. While we do believe the gastropods were physiologically dead, they were certainly dead from an ecological viewpoint (that is, unable to attach and retain their place on the substrate, move in response to external cues).

We quantified feeding rates by providing each grazer a piece of Gellidium spp. to feed on for 7 days with the feeding trial started on the 31^st day of the experiment and concluded on the 38^th. We quantified the change in algal mass (final – initial measurement) by gently patting the algae dry and then weighing to the nearest 0.0001 g at the beginning and end of the feeding period. The gastropods were provided sufficient algae that they were able to feed to satiation (0.1992 ± 0.0082 g; mean ± SE). For each tank, we also had control cages in which algal pieces were added without grazers; these cages indicated change driven by algal rather than grazer responses and were used to correct for any growth or tissue loss not due to the herbivores. We calculated consumed algae as: initial mass × correction factor – final mass of the piece of algae. The correction factor was: the final mass/initial mass for each control algal unit, which we then averaged to provide a single value per tank (following Renaud et al. 1990).

We measured oxygen consumption rates on the 39^th and 40^th days of the experiments (at the same time as survival was considered) by transferring each gastropod to an airtight chamber (55mL falcon tubes) filled with seawater from the treatment conditions, ensuring the absence of any gas bubbles. The chamber was kept within the tank to maintain treatment temperature. We also filled chambers without gastropods with water; measurements from these were used to account for any possible biological activity in the water. Each chamber had a pre-calibrated oxygen spot attached to the inner wall (Fibox4 trace, PreSens). Using these spots and a fibre optic oxygen sensor, we measured oxygen concentration twice in rapid succession every minute for a 10-minute period. We subtracted the background levels of oxygen consumption measured in paired controls from the measured rates of each snail before statistical analysis.

Statistical analysis

We used PERMANOVAs (with Euclidian distance matrices) to examine the effects of temperature and species on survival, feeding rates, and oxygen consumption. We corrected rates of oxygen consumption and feeding for gastropod size effects by dividing the oxygen and algal consumption rate measured for each gastropod by the gastropod wet mass. In all analyses, we treated temperature and species as orthogonal and fixed, with two levels (current and future or C. argyrostoma and L. coronata, respectively), and tank used as replicates. The PERMANOVAs were done using the PERMANOVA+ routines that are an ‘add-in’ to the PRIMER 6 software (Clarke & Gorley 2006, Anderson et al 2008). To evaluate the chances of committing type II errors, we used dissimilarity-based multivariate standard error (MultSE) to assess the effect of increased sample size on the precision of our parameter estimates using the SSP package in R (Anderson & Santana‐Garcon 2015, R Core Team 2020, Guerra-Castro et al. 2021).

R Script files walks user through transforming the raw data into usable data sets. Analyses are completed in Primer instead of R. There are simplified data sets that are prepared for that purpose, however the cleaning of the data and transforming raw data into rate data can be found in R.