Heat stress can disrupt acid-base homeostasis in reef-building corals and other tropical cnidarians, often leading to cellular acidosis that can undermine organismal function. Temperate cnidarians experience a high degree of seasonal temperature variability, leading us to hypothesize that temperate taxa have more thermally robust pH homeostasis than their tropical relatives. To test this, we investigated how elevated temperature affects intracellular pH and calcification in the temperate coral Astrangia poculata. Clonal pairs were exposed to elevated (30°C) or control (22°C) temperatures for 17 days. Despite causing damage to host tissues and symbiont cells, elevated temperature did not affect intracellular pH or inhibit calcification in A. poculata. These responses contrast with those of tropical cnidarians, which experience cellular acidification and decreased growth during heat stress. A. poculata therefore appears to have thermally resilient cellular acid-base homeostasis mechanisms, possibly due to adaptation to large seasonal temperature variations. However, we also observed tissue damage and lower egg densities in heat-treated individuals, suggesting that increasingly severe marine heatwaves can still threaten temperate coral fitness. These results provide insight into corals’ nuanced adaptive capacity across latitudes and biological scales.

(a)  Animal populations

Adult Astrangia poculata corals were obtained from Narragansett Bay, RI, USA (41.49231 N, -71.41883 W) in August 2022 and transported in seawater first to the University of Rhode Island (Kingston, RI, USA) and then to XXXXXXXXXXXX (XXXXXXXXXXXX, USA)

where they were maintained in a tank of artificial seawater (31–33 ppt; Instant Ocean Reef Crystals, Spectrum Brands, Blacksburg, VA, USA) at room temperature (~18°C) for 5 months prior to the experiment. Corals received ~200 µmol m-2 sec-1 photosynthetically active radiation as measured at colony depth below the surface using a LI-192 Underwater Quantum Sensor (LI-COR Biosciences, Lincoln, NE, USA) from NICREW HyperReef Dimmable Aquarium Lights (Shenzhen NiCai Technology Co., Shenzhen, China) on the white + actinic setting, on a 12h:12h light:dark schedule. These light conditions were chosen to be well below the known photoinhibition point for this species [Jacques et al. 1983] and remained consistent through the acclimation and experimental periods. In late January 2023, 35 colonies ≥10 cm in diameter with relatively homogeneous coloration within each colony were selected, assigned colony numbers, and divided in half with a chisel (Fig. 1A-B). These genetically identical pairs (ramets) were affixed to tagged ceramic plugs (Frag Station Coral Frag Disks, The Alternative Reef, Green Bay, WI, USA) using cyanoacrylate reef glue on the face(s) of the colony lacking live tissue (CorAffix, Two Little Fishies, Miami Gardens, FL, USA). A total of N = 70 coral fragments (35 total colony pairs) were then placed into a tank that was gradually acclimated to 22°C by increasing the temperature from 18°C at a rate of ~0.5°C per day using an APEX Temperature Probe and 832 Power Strip energy bar (APEX, Neptune Systems, San Jose, CA, USA) powering a 50 W aquarium heater (Aqueon, Aqueon Products, Franklin, WI, USA). The temperature was monitored using a HOBO temperature logger (UA-001-64 64K Pendant, Onset Computer Corporation, Bourne, MA, USA) that recorded the temperature every 15 minutes. Corals were held at 22°C for 12 days prior to the experiment. As the optimal respiration temperature for Rhode Island A. poculata is between 22-26°C [Aichelman et al. 2019 JEB], 22°C was chosen as a control temperature based on the Narragansett Bay, RI summer mean monthly maximum (NOAA) to represent normal non-heatwave heat exposure. Corals were fed 48-hour-old Artemia (1 full hatchery for the population; Brine Shrimp Direct, Ogden, UT, USA) weekly until the start of the experiment. 50-75% water changes were performed 4 hours after feeding, with the last pre-experimental water change five days before temperature treatments began. Water pH was maintained between 8.15-8.22 for the duration of the experiment and monitored using APEX pH probes.

(b)  Experimental treatment

A total of 6 plastic tubs (7.57 L each) were randomized into one of two treatments, 22°C (control) and 30°C (heated) for a total of 3 tubs per treatment (Fig. 1C). Each tub was fitted with an egg crate, an air stone, a HOBO temperature logger, APEX pH and temperature probes, and an Aqueon 50 W aquarium heater. The corals were randomized into each tub, with one colony ramet placed into a control tub and its corresponding ramet placed into a heated tub. Control tubs were held at 22°C (22.53 ± 0.02°C; range = 18.14°C-24.16°C) (Fig. 2A). Heated tubs increased by ~0.5°C every 12 hours (1°C per day) until the 30°C target was reached, after which temperatures were maintained at 30°C (30.11 ± 0.01°C; range = 27.21°C-31.88°C) for another 10 days (Fig. 2A) and recorded every 15 minutes using the HOBO temperature logger in the bottom of each tub. Water level was maintained over the course of the experiment with top-offs of deionized water to compensate for evaporation. Corals received a complete water change of each tub and a gentle removal of algae from each ramet on day 7 of the 18-day experiment. After the water change, corals were fed 48-hour-old Artemia as described above. Each day for the first 17 days of treatment, between 12:00–14:30 (6-8.5 hours into the 12-hour light cycle), tubs were photographed, and every individual ramet was evaluated for mortality (scored as a binary), as well as polyp extension and ramet pigmentation. Aquarium lights were covered with black fabric during photography to prevent glare on the water surface, and photographs were taken using an iPhone 11 held ~40 cm above colonies. Both extension and color were scored qualitatively on a per-ramet basis in increments of 20% from 0%-100%. Color was assessed based on the coloration of live polyps per colony using a color standard made of white waterproof paper marked with red, green, and black electrical tape with color scale as follows: 0 = fully white, 20 = mostly pale, 40 = mix of brown and pale, 60 = most polyps brown, 80 = all polyps brown, 100 = all polyps dark brown. The extension was estimated based on the percentage of live polyps per colony extended from the skeleton.

(c)  Sampling

Sampling was spread over the final two days of temperature treatment (days 17 and 18) so that in vivo intracellular pH measurements could be taken concurrently with physiological and histological sampling. Each sampling day, a haphazardly selected subset of approximately half of the corals (N = 17-18 ramet pairs day-1) from each treatment were removed for sampling. A subset of 4-6 ramet pairs day-1 (10 pairs total) was further broken into 3 pieces: 1) cell isolation for in vivo measurements of intracellular pH (see below), 2) freezing for later physiological assessments, and 3) preservation in 4% paraformaldehyde for histology. Nubbins sampled for physiological assessments, along with all other unbroken ramets, were placed in labeled sample bags (Whirl-Pak, Pleasant Prairie, WI, USA) and immediately stored at -80°C until processing.

(d)  Physiological assessments

Frozen ramets (N = 70) were thawed on ice and airbrushed (Vivohome 60 Hz air Compressor, City of Industry, CA, USA) using 10 mL chilled phosphate-buffered saline (PBS) as described [Innis et al. 2021]. As much filamentous algae as possible was excluded from the resulting tissue slurry. Tissue slurry was centrifuged at 7000 x g for 5 minutes at 4°C to separate host contents from algal cells. Host tissue supernatant was removed and stored at -80°C for protein quantification. Pelleted symbiont cells were immediately resuspended in PBS, homogenized at 25,000 rpm for 10 sec (Fisherbrand 850 Homogenizer, Thermo Fisher Scientific, Waltham, MA, USA), and kept on ice. Symbiont density was measured from these homogenates using existing protocols [Innis et al. 2021, Veal et al. 2010]. Briefly, each homogenized sample was diluted at a ratio of 1:3, 1:9, and 1:27 in 0.1% w/v sodium dodecyl sulfate (SDS) in PBS. Each dilution was loaded in triplicate into a round-bottom 96-well plate, and cell concentrations were determined using a Luminex Millipore Guava flow-cytometer (Guava Incyte Software, Austin, TX, USA). Relative chlorophyll fluorescence per symbiont was also measured on a per-ramet basis from the same flow cytometry run and was calculated as the average red autofluorescence of all events designated as symbionts based on existing gates [Innis et al. 2010]. Host protein was measured from the host fraction using the Bradford method in triplicate in a 96-well plate using a spectrophotometer (BioTek PowerWave XS2, Agilent, Santa Clara, CA, USA). Host protein and symbiont densities were normalized to colony skeletal surface area, which was determined using the single wax dipping method on airbrushed skeletons as described [Veal et al. 2010]. Care was taken to avoid dipping portions of each skeleton covered by glue or the ceramic base. As standards, eleven unique wooden dowels of known surface area spanning expected fragment sizes (1.5-114.0 cm²) were also wax dipped to generate a linear fit relating surface area to dipped wax weight (R2=0.99).

(e)  Calcification

The net change in dry skeletal weight over the course of the experiment was determined using the buoyant weight method. Buoyant weight protocols were derived from previous studies that used this technique to measure calcification in A. poculata [Holcomb et al. 2012, Holcomb et al. 2010]. Briefly, each coral was placed in a glass beaker filled with seawater that remained submerged in a water bath to maintain room temperature (18.3-20.5°C) and suspended from a lab balance (Model AX223, OHAUS, Parsippany, NJ, USA). A DeWalt metal drill bit (Black & Decker, Towson, MD, USA) was weighed at the start of measurements as a standard, and reweighed if the temperature changed by at least 0.1°C to account for changes in seawater density at different temperatures. Seawater density at each temperature was calculated based on the buoyant weight, dry weight, and density of the metal standard. The density of the standard was calculated from its dry weight and volume, which was measured by submerging the standard in water and measuring displacement. The initial buoyant weight of all individuals (N = 70) was taken before placing each in their experimental tubs. After the experimental period (day 18), each coral was cleaned to remove any algal growth on its plug and re-weighed using the same procedure. Change in skeletal mass was calculated as described in [Davies 1989].

(f)  Intracellular pH (pHi) measurements

Intracellular pH (pHi) was measured for 10 colony pairs (N = 10 fragments per temperature) using an existing confocal imaging method [Innis et al. 2021, Allen-Waller & Barott 2023, Allen-Waller et al. 2024, Venn et al. 2009]. After a 15-minute dark-acclimation period, cells were incubated with 10 µM of SNARF-1AM (Thermo Fisher Scientific) in filtered artificial seawater (FASW) for 20 minutes. FASW was sourced from the ASW used for coral husbandry (pH 8.15-8.22). Cells were then spun down to remove supernatant containing excess dye and resuspended in FASW. Cells were imaged on a confocal microscope (Leica SP8 DMi8, Leica Camera, Wetzlar, Germany) at 63x magnification (HC PL Apochromat C52 Oil objective, numerical aperture = 1.4). Cells were excited using 514 nm light from an argon laser (10% emission, 1.5% power, 458/514/561 nm beamsplitter). SNARF-1 fluorescent emission was simultaneously acquired at a frame rate of 400 Hz (frame rate = 0.77/s, pixel dwell time = 3.16 µs) in two channels (585 ± 15 and 640 ± 15 nm) using HyD detectors (gain = 75, pinhole = 1.00 airy unit). For each sample, 15-40 images were collected, about half of which pertained to each cell type (non-symbiocytes and symbiocytes). Images were analyzed using ImageJ, and coral cytoplasmic yellow:red fluorescence ratios were converted to pHi values using a standard curve generated using cells from non-experimental A. poculata incubated in solutions of known pH and osmolarity with 30 µM nigericin [Allen-Waller et al. 2024, Venn et al. 2009] (Fig. S1). Cell quality was checked to ensure only live cells were measured as described previously [Allen-Waller et al. 2024]. Each colony measurement per cell type represents an average from N ≥ 8 cells.

(g)  Histology

Paraformaldehyde-fixed samples were decalcified in 0.5 M EDTA + 0.5% paraformaldehyde in calcium-free S22 buffer (450 mM NaCl, 10 mM KCl, 58 mM MgCl₂, 10 mM CaCl₂, 100 mM HEPES, pH 7.8), changed daily until the skeleton was gone. Decalcified tissue was then held in 70% ethanol at 4°C until samples were dehydrated for histology. Samples were dehydrated using a series of incubations (3 x 20 minutes each at 90% and then 100% ethanol), clarified in xylene substitute (SafeClear, Fisher HealthCare, Fisher Scientific), and sent to Pacific Pathology (San Diego, CA, USA) for paraffin wax embedding, slicing, and staining with hematoxylin and eosin. The resulting slides were treatment-blinded and assessed for tissue health by scoring (1) symbiont quality (roundness, cell integrity), (2) tissue layer definition/structural integrity, (3) cellular integrity (lack of necrosis/granulation), and (4) epidermal clarity. Each variable was qualitatively assessed on a scale of 1-5, adapting existing methods developed for tropical corals [Kruse et al. 2025]. A score of 5 was assigned to healthier tissue (e.g. clear definition between tissue layers, no necrosis, well-defined epidermis) while a score of 1 corresponded to more tissue damage (e.g. disordered tissue layers, granular/necrotic cells, degraded/no epidermis). Epidermal thickness was measured in ImageJ [Abramoff et al. 2025] wherever possible by drawing segments across the epidermis at regular intervals, wherever the epidermis was clearly defined (3-27 measurements per slide, depending on epidermal integrity of the ramet). Segment lengths were converted from pixel units to mm using a line drawn along a stage micrometer (0.01mm Microscope Camera Calibration Slide, OMAX Microscope), and average thickness per ramet was calculated using all measurements from each ramet. Egg production was also assessed as a binary by the presence/absence of eggs in all slides. Total eggs per area were calculated by counting the number of eggs per slide image and dividing it by the area of the tissue slice within the image. Sperm or spermatocytes were not seen in any individuals and were therefore not quantified. Slice area was measured by freehand outlining in ImageJ to obtain pixel area, which was then converted to mm² using the micrometer.

(h)  Statistical methods

All data were analyzed in RStudio version 2022.07.2 (https://www.rstudio.com/) and plots were generated using the package ggplot2 [Wickham 2016]. Daily average treatment temperatures were compared between treatments using a linear model with treatment and date as fixed effects. To calculate the effect of temperature on qualitative metrics (colony polyp extension and pigmentation) over time, linear mixed effects models were constructed using the lme4 package [Bates et al. 2015] with temperature treatment and date fixed effects, and colony and tub as random intercepts. Post hoc pairwise comparisons by date were analyzed by Tukey’s HSD using the emmeans package [Lenth & Lenth 2017]. To calculate the impact of temperature treatment on each physiological metric, the value of each measured physiological parameter (symbiont cell density, chlorophyll fluorescence per symbiont, epidermal thickness, total daily calcification rate, nonsymbiocyte intracellular pH, and symbiocyte intracellular pH) for each 22°C-treated ramet was subtracted from the 30°C-treated ramet of the same colony. To test the effect of initial colony symbiont density on colonies’ temperature response, each resulting set of ∆ values was regressed against 22°C-treated ramets’ symbiont cell densities.

Effects of temperature on final symbiont cell density, chlorophyll, colony color, and histological metrics were further tested via linear mixed effects models with temperature treatment as a fixed effect and colony as a random intercept. Q-Q and residual plots were checked to ensure each model met normality and homogeneity of variance assumptions. To test how colony bleaching response impacted organismal physiological metrics, colonies were divided between those that lost symbionts (∆ symbiont density < 0) and those that gained symbionts (∆ symbiont density > 0). Linear models for each subpopulation were constructed, testing the fixed effect of temperature treatment on every other physiological metric.