2.3.1. Great Barrier Reef

Gnathiids, all belonging to the species *Gnathia aureamaculosa*, were collected from the culture in the morning and afternoon, and placed together into 75 mL holding containers filled with seawater.

They were collected by moving a black tray (35 × 25 × 5 cm) up the side of the gnathiid culture tank and were removed using a pipette. From the holding containers, gnathiids were individually transferred into 5 mL vials that were half-filled with seawater. These vials were then individually labelled. Collecting and processing took approximately 2 to 4 h, depending on the catch size of the day (ranging from 9 to 226 gnathiids). The daily number varied as a result of fluctuations in the number that were active, most likely due to normal high variation in their population dynamics [102]. A mixture of fed and unfed gnathiids was used, and it was not known how much time elapsed since the last feeding. Gnathiids were not fed for practical reasons. After processing the gnathiids, the vials were randomly allocated, in a balanced way (approximate equal number), to a temperature treatment and aquarium replicate combination; there were three aquarium replicates per temperature treatment. Vials were labelled with a unique number across all replicates. Only the lids of the vials were labelled, reducing any potential bias when viewing them under the microscope. It also made it easier to monitor and return them to their respective treatment and aquarium daily. Vials were held underwater in plastic baskets (17 × 17 × 10 cm), one for each treatment and replicate (*n* = 9 baskets). Baskets had four mesh (1 mm2) windows (12 <sup>×</sup> 5 cm) on the sides and one on the lid (12 <sup>×</sup> 12 cm) to allow for flow of water. A dive weight was used to submerge the baskets. Aquaria were supplied with flow-through seawater, with seawater that was either chilled or heated in a sump under the aquarium benches and pumped up to the aquaria. Each bench had a different temperature treatment and held 10 aquaria (previously used for another experiment, see Graba-Landry et al. 2020 [106]). Three aquaria were randomly selected per bench and allocated to replicates.

We estimated the predicted ambient seawater temperature (29.25 ◦C ± 0.013 SE) based on the Australian Institute of Marine Science long-term average water temperature for February [107] (Figure S1). Actual average daily seawater temperature during the experiment was 29.0 ◦C ± 0.67 SE (February 1 to 23, 2018 available only). The temperature of the water that gnathiids had been maintained in throughout their lifetimes was not available. However, the temperature of incoming water from the station's holding tanks was on average 1.4 ◦C warmer than the ocean, when sampled at two sources at three times of the day (09:00, 15:00 and 21:00 h from 15 to 20 October, 2018) relative to the same period in the ocean [108].

Temperature was manipulated in an outdoor seawater flow-through system at LIRS using purpose built 1KW steel bar heaters and chillers (Teco®) in a header or sump tank. Each sump, one per temperature treatment, fed replicate 40 L tanks with the appropriate experimental flow-through seawater using 1000 L hr−<sup>1</sup> pumps (Eheim®) at a rate of approximately 1 L minute−1. Tanks were wrapped in Insulbreak® insulation to stabilize water temperatures. Temperature (±0.1 ◦C) was also measured at 12:00 h daily from each of the nine tanks housing the gnathid cultures/vials using a portable temperature probe (Comark®) calibrated to 26 ◦C, 28 ◦C and 30 ◦C (National Association of Testing Authorities certified) to ensure temperature remained stable across treatments. Experimental temperatures at 12:00 h per treatment, averaged across the means of the three replicate aquaria, were: 29 ◦C: 29.5 ◦C ± 0.07 SE, 31 ◦C: 31.4 ◦C ± 0.02 SE, and 32 ◦C: 32.6 ◦C ± 0.02 SE. Temperatures were very similar among replicates within a treatment, (Figure S2). One calibrated temperature logger (HOBO Pendant temperature/light logger, UA-002-08) per treatment also recorded the temperature every 2 h throughout the course of the study to account for diurnal fluctuations in temperature [mean (SE) per treatment: 29 ◦C: 28.9(0.042); 31 ◦C: 31.2(0.035): 32 ◦C: 31.9(0.06); Figure S3].

Each day, one random basket from each treatment was removed (to reduce time exposed to air temperature). The vials from each basket were rinsed in freshwater and placed in a large tub (all treatments were examined together to avoid bias). Each vial was then examined under a dissection microscope to check for gnathiid mortality. Vials with alive gnathiids were sorted back into their respective treatments/replicates and placed back into the aquaria. This was repeated for each of the remaining baskets from each treatment-replicate combination. Gnathiids were monitored until all had died (except for four survivors, see Results for details). Vials with dead gnathiids were preserved for later to undertake headwidth measurements, by adding a few drops of formalin into the seawater.
