Intraspecific variation reshapes coral assemblages under elevated temperature and acidity

Abstract Insights into assemblages that can persist in extreme environments are still emerging. Ocean warming and acidification select against species with low physiological tolerance (trait‐based ‘filtering’). However, intraspecific trait variation can promote species adaptation and persistence, with potentially large effects on assemblage structure. By sampling nine coral traits (four morphological, four tissue and one skeletal) along an offshore–inshore gradient in temperature and pH, we show that distantly related coral species undergo consistent intraspecific changes as they cross into warm, acidic environments. Intraspecific variation and species turnover each favoured colonies with greater tissue biomass, higher symbiont densities and reduced skeletal investments, indicating strong filtering on colony physiology within and across species. Physiological tissue traits were highly variable within species and were independent of morphology, enabling morphologically diverse species to cross into sites of elevated temperature and acidity. Widespread intraspecific change can therefore counter the loss of biodiversity and morphological structure across a steep environmental gradient.

(accuracy corresponds to a linearity of 0.99 R 2 ). Replicate readings were taken in October 2019, November 2019, January 2019 and April 2020. The instrument was deployed by hand from a boat, logging profile readings at the desired depth after a one minute pause to allow for sensor stabilization. All sensors are calibrated prior to use on the same day with high quality standard solutions. pH sensors were calibrated using pH-7 and pH-10 buffers, and are on an NBS scale, which prior analysis suggests is likely to be consistently higher than measurements on a total pH scale by approximately 0.1 using this sensor. Long-term temperature averages (between 2011 and 2021) were quantified using the National Oceanic and Atmospheric Administration's coastal watch program to visualise the thermal gradient ( Figure 1a). Yet localised readings (Figure 1b) are likely to reflect more accurate temperature information specific to our sites and depths (see also, Colin 2018).
Physiological and morphological analysis: For each sample, physiological analysis was conducted by removing the tissue from the intact skeleton using compressed air, mixing with 15 ml of 0.45 μm-filtered seawater, and homogenising (IKA T10 basic, Ultra Turrax Homogenizer) tissues. Algal symbionts (family: Symbiodiniaceae), or zooxanthellae density (ZD) was measured by adding 0.1 ml of formaldehyde to an 0.9 ml aliquot of tissue suspension, counting the cells in six replicate 0.1l subsamples using a Neubauer Haemocytometer at x40 magnification, and averaging across replicates (Stimson & Kinzie 1991). Chlorophyll was extracted from an 6 ml aliquot of tissue suspension after pelleting symbionts via centrifugation and addition to 4 ml of acetone to extract pigments, allowing them to soak for 12 hours under refrigerated conditions. Chlorophyll concentrations (CC) were measured using spectrophotometry on a SpectraMax Plus 384 Microplate reader (Molecular Devices). Chlorophyll-a was calculated as 11.43(A663 -A750) -0.64 (A630 -A750) and Chlorophyll-c as -3.63(A663 -A750) + 27.09(A630 -A750) where An signifies absorbance at wavelength n (Jeffrey & Humphrey 1975). Protein biomass (PB) was quantified from 1ml of tissue using spectrophotometry and the Red 660 protein assay, using bovine serum albumen protein (BSA) to construct a standard curve (Palmer et al. 2009). Tissue biomass (TB) measuring the total organic weight of the tissue (ash-free dry weight) was quantified by placing 5ml of tissue suspension into a freeze dryer (Christ, Alpaa 1-1 LO plus) for 48 hours to calculate dry weight, and then subtracting the ash weight measured after incinerating the sample in a muffle furnace at 550°C (Leuven & Brock 1985).
Tissue surface area, skeletal volume, and surface area to volume ratio (SV) of each fragment was measured to 0.1mm resolution using a 3D laser scanner (CREAFORM HandySCAN 3D and VXelements software). Physiological traits were scaled to coral tissue area by calculating total content per fragment accounting for the initial 15 ml dilution of water, and dividing by fragment area. The skeletal density (SD) of fragments was found by dividing the dry fragment weight (to the nearest 0.1g) by its volume. Morphological parameters such as planar surface area, corallite width, and branching dimensions were quantified from images using ImageJ (version 1.53a). Projections of the colony greater than 3cm in length were considered a new branch, excluding individual corallites in bottlebrush species, and classifying individual lobes as branches for mound-shaped species. Branch density (BD) was measured by dividing the total number branches per colony by planar area. Branch width (BW) was measured as the average diameter of three randomly selected branches measured halfway between the branch tip and the branch base. Branch height (BH) was measured as the average perpendicular length of three randomly selected branches. Finally, we used the R-package 'colordistance' (Weller & Westneat, 2019) to generate a rudimentary visualisation of colony colours across the gradient (shown in Figure 3C only). We took a planar image of the colony in broad daylight using a Canon G7X, took a 3 cm 2 snapshot of the centre of each colony, and used 'colordistance' to identify the dominant colour (most frequent RGB bin across pixels) for each image.

Appendix S1
Supplementary figure 1 Figure S1: Summary of relationships among traits of sampled colonies. Data is shown in lower panels.
Correlation coefficients in upper panels. The plot was produced using the R package "psych".  (r 2 c), indicating the variance explained by fixed and random factors.