Heatwaves during low tide are critical for the physiological performance of intertidal macroalgae under global warming scenarios

The abundance and distribution of intertidal canopy-forming macroalgae are threatened by the increase in sea surface temperature and in the frequency and intensity of heatwaves caused by global warming. This study evaluated the physiological response of predominant intertidal macroalgae in the NW Iberian Peninsula (Bifurcaria bifurcata, Cystoseira tamariscifolia and Codium tomentosum) to increased seawater temperature during immersion and increased air temperatures during consecutive emersion cycles. We combined field mensuration and laboratory experiments in which we measured mortality, growth, maximum quantum yield and C:N content of the macroalgae. Air temperature was a critical factor in determining physiological responses and survivorship of all species, whereas high seawater temperature had sublethal effects. Cystoseira tamariscifolia suffered the greatest decreases in Fv/Fm, growth and the highest mortality under higher air temperatures, whereas C. tomentosum was the most resistant and resilient species. Two consecutive cycles of emersion under atmospheric heatwaves caused cumulative stress in all three macroalgae, affecting the physiological performance and increasing the mortality. The potential expansion of the warm-temperate species B. bifurcata, C. tamariscifolia and C. tomentosum in the NW Iberian Peninsula in response to increasing seawater temperature may be affected by the impact of increased air temperature, especially in a region where the incidence of atmospheric heatwaves is expected to increase.


Results
Field measurements. Environmental conditions. The average seawater temperature recorded by a data logger deployed at the study site in July and August was 17.02 ± 1.05 °C. The day of the field measurements the average seawater temperature was 18.18 ± 0.61 °C, the average amount of light measured on bare rock was 2956 ± 516 μmol photon m −2 s −1 and the air temperature reached a maximum of 38 °C (Supplementary Fig. S1). The mean temperature below the canopies was 25.68 ± 0.12 °C for B. bifurcata, 25.76 ± 0.16 °C for C. tamariscifolia and 25.79 ± 0.24 °C for C. tomentosum. Humidity (%) below the canopies was 86.13 ± 0.21 for B. bifurcata, 86.41 ± 0.29 for C. tamariscifolia and 88.27 ± 0.31 for C. tomentosum. Temperature below the canopies did not differ between the macroalgae (GLM for temperature: χ 2 = 0.15, df = 2, p = 0.928, n = 3), whereas humidity was significantly higher below C. tomentosum (GLM for humidity: χ 2 = 37.49, df = 2, p < 0.001, n = 3) ( Supplementary  Fig. S2).
∆Weight and ∆F v /F m . The weight of the fronds of all three species decreased with increasing air temperature during emersion, although the fronds were not completely dry. After immersion and recovery for 30 min, the weight increased significantly (GEEs: B. bifurcata: χ 2 = 120, df = 1, p < 0.001; C. tamariscifolia: χ 2 = 245, df = 1,
During the four hours of emersion in the tanks, the air temperature remained constant in the control treatment, whereas it increased gradually in the atmospheric heatwave and extreme atmospheric heat wave treatments. After the fourth hour in the two emersion cycles, when target temperatures were reached, the temperature under the macroalgal canopies was 22.93 ± 0.08 °C in the control treatment, 28.40 ± 0.19 °C in the heatwave treatment and 30.60 ± 0.14 °C in the extreme heatwave treatment.
Mortality. After the first emersion cycle, dead fronds of all species were observed in the atmospheric heatwave and extreme atmospheric heatwave treatments, although the mortality rates differed between species. The probability curves for mortality in B. bifurcata and C. tomentosum were similar and showed a lower probability of mortality than those of C. tamariscifolia due to the increase in air temperature (Fig. 2a-c). Seawater temperature also had a significant additive effect on the mortality of C. tamariscifolia (Table 1) with a significant increase in the probability of mortality in the extreme marine heatwave treatment.
After the second emersion cycle under the control air temperature, dead fronds of C. tamariscifolia, but not of B. bifurcata or C. tomentosum, were observed ( Fig. 2d-f). Under the atmospheric heatwave and extreme atmospheric heatwave treatments, the increase in air temperature had a significant effect on B. bifurcata mortality, which was much higher in the second than in the first emersion cycle, indicating cumulative effects (Fig. 2d). As in the first emersion cycle, seawater and air temperatures had a significant additive effect on C. tamariscifolia mortality ( Table 1). The probability of mortality increased under the extreme marine heatwave conditions compared to the control, and it also increased when air temperature increased in the atmospheric heat wave treatment, indicating cumulative effects in the second emersion cycle (Fig. 2e). Air temperature had a marginal effect on C. tomentosum mortality (Table 1, Fig. 2f). No dead fronds of any of the species were recorded after the recovery period.
After the first emersion cycle, the increase in air temperature significantly affected the growth of B. bifurcata and C. tomentosum (Table 1). At the control air temperature, the weight of C. tamariscifolia decreased, whereas the weight of B. bifurcata and C. tomentosum remained unchanged. In the atmospheric heatwave and extreme atmospheric heatwave treatments, the weight of all three species decreased, with C. tamariscifolia showing the greatest decrease, followed by C. tomentosum and B. bifurcata (Fig. 3d-f). Because of the high mortality rate, www.nature.com/scientificreports/ which resulted in an insufficient number of replicates, the effect of air temperature on the growth of C. tamariscifolia could not be modelled, although the trend was for a marked decrease in growth. After the second emersion cycle, the growth of B. bifurcata and C. tomentosum decreased significantly as the air temperature increased ( Table 1). Growth of the three species at the control air temperature was similar to that observed after the first emersion cycle. In the atmospheric heatwave and extreme atmospheric heatwave treatments, the greatest decrease in growth occurred in C. tamariscifolia, followed by C. tomentosum and B. bifurcata ( Fig. 3g-i). The increase in air temperature during consecutive emersion cycles had a cumulative negative effect on the growth of C. tamariscifolia. This trend was clearly evidenced in those fronds at the control seawater temperature (18.3 °C), which presented a steeper negative slope after the second emersion (Fig. 3e,h).
After the recovery period, the fronds of B. bifurcata and C. tomentosum subjected to an increase in water temperature and previously exposed to an increase in air temperature still showed a significant additive effect that led to a decrease in growth (Table 1). Although the fronds of C. tamariscifolia and C. tomentosum in the control seawater temperature treatment grew, the growth did not compensate for the previous weight loss during the emersion stress. The weight of C. tamariscifolia and C. tomentosum exposed to the extreme marine heatwave was still lower after the recovery period ( Fig. 3j-l). www.nature.com/scientificreports/ ∆F v /F m . The photosynthetic performance of the three species decreased when the fronds were subjected to atmospheric heatwave and extreme atmospheric heatwave treatments, as shown by the change in F v /F m , with values well below 1, especially in C. tamariscifolia ( Fig. 4a-f). After the first emersion cycle, under control air temperature, the change in F v /F m was similar in B. bifurcata and C. tomentosum and was slightly higher in C. tamariscifolia, with a value below 1. Under atmospheric heatwave and extreme atmospheric heatwave treatments, the change in F v /F m was minimal in C. tomentosum (ΔF v /F m ~ 1), and the ΔF v /F m changed very little in B. bifurcata, to values slightly below 1. Cystoseira tamariscifolia showed the most marked variation in ΔF v / F m, (Fig. 4b), with the F v /F m being markedly lower after emersion than prior to emersion. The effect of extreme atmospheric heatwave on the ΔF v /F m in C. tamariscifolia could not be modelled, because of the insufficient number of replicates caused by the high mortality. However, the increase in seawater temperature had a marginally significant effect on the variation in ΔF v /F m in C. tamariscifolia, for which values were below 1 in the marine heatwave and extreme marine heatwave treatments (Table 1, Fig. 4b).
After the second emersion cycle at control air temperature, the values of ΔF v /F m in C. tomentosum and B. bifurcata were higher than 1, indicating an increase relative to the pre-emersion values, whereas the ΔF v /F m in Table 1. The summarised results of GLMs used to test the effects of seawater (ST) and air temperature (AT) on the mortality, growth and ΔF v /F m of fronds after the emersion cycles and during the recovery period, in the laboratory experiment. Significant effects (p < 0.01) are indicated in bold. * indicates marginal significance. "n.a. " indicates not analysed as no mortality was recorded in the recovery period. Note that due to an insufficient number of replicates, the effect of the extreme atmospheric heatwave (35.7 °C) on the growth and the ΔF v /F m of B. bifurcata after the second emersion cycle was not modelled. Likewise, the effect of air temperature on the growth and the ΔF v /F m of C. tamariscifolia after the cycles of emersion were not modelled. www.nature.com/scientificreports/ fronds of C. tamariscifolia was ~ 1, indicating no variation in F v /F m . Under atmospheric heatwave and extreme atmospheric heatwave treatments, the ΔF v /F m tended to change in the three species, to values below 1, when air temperature increased ( Fig. 4d-f), although this factor only significantly affected the ΔF v /F m of C. tomentosum ( Table 1). The consecutive exposure to high air temperature had a cumulative effect on the decrease in photosynthetic performance in C. tomentosum (Fig. 4f). After recovery in immersion tanks, at control seawater temperature, ∆F v /F m varied notably in fronds of C. tamariscifolia, indicating that after recovery, the F v /F m values exceeded those values before this period, whereas ∆F v /F m in B. bifurcata and C. tomentosum remained around 1, indicating no recovery of the F v /F m values ( Fig. 4 g-i). In the marine heatwave treatment (~ 19.9 °C), the ∆F v /F m in C. tamariscifolia fronds differed markedly from that in fronds in control seawater, implying the absence of recovery and additional loss of F v /F m . By contrast, in marine heatwave and extreme marine heatwave conditions, ∆F v /F m scarcely varied in C. tomentosum and increased only slightly in B. bifurcata (Fig. 4g,i). The overall change in F v /F m after recovery was slight, and the greatest increase (1.5-fold) appeared in C. tamariscifolia previously subjected to atmospheric heatwave conditions (air temperature = 32.7 °C).
C:N ratio. The C:N ratio was highest in fronds of B. bifurcata, followed by those of C. tamariscifolia and C. tomentosum (Fig. 5). In general, the C:N ratio increased as the seawater temperature increased (Table 2, Fig. 5), whereas air temperature had only a marginally significant effect on the C:N ratio in B. bifurcata. More specifically, the increase in seawater temperature led to a significant decrease in %N in all three macroalgae ( Table 2, Fig S4). By contrast, the increase in air temperature tended to decrease the %C in B. bifurcata and C. tomentosum ( Table 2, Supplementary Fig. S4). Air and seawater temperature interacted in the C:N ratio in C. tomentosum as 3 °C) after acclimatization to seawater temperature (a-c), after emersion cycles (d-i) and after recovery (j-l) in the laboratory experiment. The points represent replicates and the smoothed curves are the data fitted by GLMs with a confidence interval of 95%. Note that due to an insufficient number of replicates, the effect of the highest air temperature (35.7 °C) on the growth of B. bifurcata after the second emersion cycle was not formally modelled. Likewise, the effect of air temperature on the growth of C. tamariscifolia after the two emersion cycles was not formally modelled.

Scientific Reports
| (2020) 10:21408 | https://doi.org/10.1038/s41598-020-78526-5 www.nature.com/scientificreports/ the C:N ratio tended to increase as the seawater temperature increased, although the trend was less marked in fronds exposed to higher air temperature during emersion cycles (note the less steep slopes for air temperatures of 32.7 and 35.7 °C in Fig. 5c).

Discussion
This study showed that the stressful conditions created particularly by atmospheric heatwaves and, to less extent, by marine heatwaves negatively affected the survival and physiology of intertidal macroalgae after exposure to two consecutive emersion cycles, and that these stressors had an additive effect. Increased air temperature caused important negative short-and intermediate-term effects on the physiological performance of the macroalgae in the field and the laboratory experiment, effects that were cumulative after exposure to two consecutive emersion cycles in the laboratory experiment. This pattern is consistent with the findings of other studies in which air temperature was found to be a relevant factor modulating physiological stress in intertidal macroalgae 6,15,20,26,37 and of others that related contractions in the distribution of intertidal macroalgae to large increases in air temperature 37,52,53 . The seawater temperatures applied in the laboratory experiment (which reflected realistic present and future climatic scenarios according to historic databases, scientific literature and IPCC predictions) were below the physiological threshold of 23 °C for B. bifurcata 54 . The highest temperature was above   55 . Thus, the increased seawater temperatures negatively affected the physiological performance of fronds, but led to sub-lethal effects. The different indicators measured in macroalgae after the emersion cycles showed a consistent decline of the physiological performance caused by atmospheric and marine heatwaves, which was greater under the most extreme conditions, i.e. global warming scenario. The F v /F m ratio reflected short term regulatory mechanisms such as photoprotection 56 , whereas the changes in growth and the C:N ratio integrated physiological processes over longer periods 48 . Values of ΔF v /F m much lower than 1, caused by the high air temperature, suggest short-term regulatory effects on the physiology of macroalgae and were probably a consequence of desiccation and thermal stress. Tissue desiccation tends to decrease the F v /F m, due to a diminution of osmotic potential 18,49,52,57,58 , affects photosynthesis, C fixation and generates overproduction of reactive oxygen species (ROS) 23,59,60 with negative consequences on physiological processes 23,57,59,61 , and the high temperatures also cause enzyme denaturalization 59,62 . The macroalgae resist this disruptive stress and prevent lethal damage by reducing all metabolic activities 63,64 , thereby inhibiting growth 49 . This capacity was evidenced by the survival of the large proportion of fronds that suffered severe desiccation and also the increased weight and F v /F m observed after recovery. The increasing seawater temperature also increased significantly the C:N ratio, as previously observed in diverse red and brown seaweeds 50 . This increase occurred as a consequence of significant decreases in %N (Table 2, Supplementary Figure S4) coincident with patterns observed in laminarians 65 . Higher seawater temperatures have often been linked to a decrease in the activity of nitrate reductase in intertidal and subtidal macroalgae 66,67 , implying a lower assimilation capacity of NO 3 − and lower incorporation rates of N. The abovementioned effects, together with the cumulative stress caused by two consecutive cycles of emersion under  www.nature.com/scientificreports/ atmospheric heatwaves, probably overwhelmed the shelf-shading protective mechanisms of the canopies against desiccation 6,13,14,52,68 , the production of heat shock proteins and the repair processes 10,23,48,59 , and prevented the fronds from recovering the pre-stress F v /F m and weight, leading to death in some cases as evidenced by the cumulative mortality in C. tamariscifolia and B. bifurcata. This may result in a decline of the abundance of these habitat-forming macroalgae in the intertidal, as reported in 37 . The species under study showed different capacities to tolerate increase in air temperature, with C. tamariscifolia being the most vulnerable and C. tomentosum the most resistant and resilient. The high resilience of C. tomentosum was indicated by the recovery of F v /F m to pre-stress values, although the second emersion weakened this capacity suggesting cumulative stress. The decrease in F v /F m in B. bifurcata after emersion in the field was followed by a large increase in F v /F m after the recovery period, revealing the resilience of this alga after moderate desiccation. Moreover, in C. tomentosum and B. bifurcata, the F v /F m changed little, with values slightly higher than 1 obtained after the second laboratory emersion experiment at control air temperature. This suggests that after moderate desiccation, the water loss was not high enough to alter the efficiency of charge separation during photosynthesis 18,58 , indicating the absence of a high level of physiological stress. The morphological traits also influence resistance to desiccation 68 . The desiccation rate is lower in the thick and voluminous fronds of C. tomentosum, which have a smaller specific surface than the fronds of C. tamariscifolia 10,69 . In comparison to C. tamariscifolia, the less pronounced decrease in growth observed in B. bifurcata after the two emersion cycles was mainly a consequence of a slower growth rate 51 . The different resistance and resilience to increasing air temperatures might imply a change in the structure of macroalgal assemblages on intertidal rocky shores of NW Spain under global warming scenarios, with an increase in the relative abundance of C. tomentosum and B. bifurcata and a decrease in C. tamariscifolia.
Although experimental and field responses showed similar patterns, extrapolation of laboratory results to the field must be done with caution. In the laboratory experiment, the humidity levels were higher than in the intertidal, which implies that emersions under the same conditions at the intertidal could lead to stronger negative effects such as higher desiccation, lower physiological performance and greater mortality. By contrast, the irradiance levels in the laboratory were much lower than those experienced in the intertidal, and might have been below the saturation points (I k ) of the species, as for example the I k of the intertidal fucoid Fucus spiralis is above 300 μmol photon m −2 s −1 under emersion with air temperatures of 20°C 22 . This light limitation might result in lower photosynthetic efficiency and smaller growth rates during the laboratory experiment with potentially more intense negative effects of the desiccation during emersions 22 and inhibition of the recovery responses during immersions (see 70 ). Nevertheless, our results might have broader consequences for the future geographical distribution of macroalgae. The NW Iberian Peninsula represents the range center of the species studied (www. marin espec ies.org; www.obis.org) 71 . In this area, previous studies have predicted an expansion and increase in the abundance of B. bifurcata, C. tamariscifolia and Codium spp. in response to the increased sea surface temperature expected under global warming scenarios (see Introduction). However, the present findings show that atmospheric heatwaves during low tide may be detrimental to the performance and survival of these species, probably inhibiting the predicted shifts. The importance of air temperature for the habitat suitability of intertidal macroalgae is evident; however, species distribution models (SDMs) and databases that take air temperature into account [72][73][74] are scarcer than models including only seawater temperature. To improve the forecasting of potential range shifts in intertidal macroalgae due to global warming, air temperature should be included as a factor in SDMs. Air and seawater temperature acted additively increasing the mortality of C. tamariscifolia after emersion cycles and decreasing the growth of B. bifurcata and C. tomentosum after a period of recovery. These additive effects coincide with patterns observed in other intertidal macroalgae 20 , probably because both factors affect similar physiological mechanisms 75 .
The fact that the macroalgae exposed to atmospheric heatwaves during two consecutive cycles of emersion suffered cumulative effects on their physiological performance and mortality is important, as air temperature and the frequency of heatwaves are expected to increase under global warming scenarios predicted for the region 8,9 . A decline in these bio-engineering macroalgae may remove habitat and shelter for several species in higher trophic levels, degrading the benthic community and having negative consequences on diversity and fisheries 33 . Future experiments could apply higher-resolution air temperature gradients to intertidal macroalgae to obtain data for constructing thermal physiological performance curves, especially under consecutive periods of exposure at low tide. Despite the limitations of laboratory experiments, their contribution to assessing the cumulative effects of chronic exposure to atmospheric heatwaves during low tide is crucial, as the findings can help to improve predictive models that forecast the effects of global warming on the physiology and distribution of ecologically important intertidal macroalgae.

Materials and methods
Field measurements. The field measurements were carried out on 4 August 2019 at noon during low spring tide to examine the effects of air temperature on the photosynthetic performance and wet weight of the co-occurring species B. bifurcata, C. tamariscifolia and C. tomentosum in situ. These species coexist at the middle intertidal zone on the Isle of Monteagudo (area surrounding coordinates 42.23551°N, 8.89956°W), which belongs to the Cíes Islands archipelago, which forms part of the Atlantic Islands of Galicia National Park (NW Iberian Peninsula; Fig. 6). The study site is located in the Ría de Vigo, which experiences mesotidal semi-diurnal tides. At this site, the target species cover wide areas of the mid intertidal zone, forming dense, bushy canopies that emerge during an average of 4 h under low spring tides under favourable conditions (high atmospheric pressure and low wave height).
On the day that the field measurements were conducted, the insolation was 76.8%, the average wind speed was 6 km/h, and the average air temperature was 19 °C (www.meteo galic ia.gal).
Scientific Reports | (2020) 10:21408 | https://doi.org/10.1038/s41598-020-78526-5 www.nature.com/scientificreports/ A HOBO Pendant temp/light data logger (Onset Computer Corp., Bourne, MA, USA) with an accuracy range of ± 0.53 °C and resolution of 0.14 °C placed on the bare rock recorded light and temperature data every 5 min. In addition, three temperature and humidity iButton Hygrochron data loggers (Maxim Integrated Products, Dallas Semiconductor, USA) with an accuracy range of ± 0.50 °C and resolution of 0.0625 °C placed below the canopy of each species (total N = 9) recorded data every minute.
The physiological responses of the macroalgae were assessed after emersion during low tide and subsequent recovery by immersion in seawater. Once the tide ebbed, 10 fronds of each species (N = 30) were cut from the canopies, numbered and weighed to determine the pre-stress wet weight. The fronds were then dark adapted with two layers of thick opaque plastic for 30 min to allow the complete oxidation of reaction centers for determining basal fluorescence (F o ). The maximum quantum yield of photosynthesis (F v /F m ) was then measured in the central part of the fronds with a Pulse-Amplitude-Modulation (PAM) chlorophyll fluorometer. The fronds were held under direct sunlight for two hours and then were dark-adapted for 30 min, and the F v /F m and wet weight were again determined. The recovery treatment was applied by placing the fronds in a bucket filled with seawater and covered with opaque plastic for 30 min, before repeating wet weight and F v /F m measurements.
The variation in photosynthetic performance (ΔF v /F m ) was calculated using Eq. (1): This ratio was used to evaluate the physiological condition of macroalgae, where values > 1 indicate an increase in the F v /F m , i.e. improved performance, and values < 1 indicate a decrease in F v /F m , i.e. poorer performance after the treatments. The variation in weight (ΔWeight) indicated the potential for desiccative water loss under emersion in the field, which is linked to the different physiological responses of each species under desiccation stress. ΔWeight was calculated in the same way as ΔF v /F m .  . The seawater and air temperature treatments used in this experiment were selected to simulate realistic present and future situations according to baseline scenarios in which the current trends in greenhouse gas emissions remain unaltered 7 . Control seawater temperature was the average ambient summer temperature in July and August between 2009 and 2018 at the nearby Baiona coastal station in the Ría de Vigo (http://www.intec mar.gal/Multi Param /Defau lt.aspx); the marine heatwave temperature was the average of values higher than the 95th percentile of the records; and the extreme marine heatwave temperature was the marine heatwave temperature under a future scenario of global warming, with a 2.4 °C increase for the Iberian upwelling in the 2070-2100 period 41 . The emersion air temperatures corresponded to average air temperatures at noon on the Cíes Islands in July and August between 2009 and 2018 (www.meteo galic ia.gal). Control air temperature was the average summer temperature at noon; atmospheric heatwave temperature was the average temperature at noon above the 95th percentile, recorded on more than three days (www.aemet .es); and the extreme atmospheric heatwave temperature represented a heatwave under a global warming scenario of a 3.0 °C increase. This increase is the average calculated from the predictions of baseline scenarios for the period 2081-2100 7 .
A total of 270 vegetative fronds (90 fronds per each species) were collected in the middle intertidal zone (Fig. 6) and transported to the Estación de Ciencias Mariñas de Toralla (ECIMAT) facilities, which form part of the Marine Research Centre (CIM) of the University of Vigo (Spain). The fronds were acclimated for 6 days to laboratory conditions to enable a steady growth response to be reached. All fronds were numbered, attached to nylon strings inside two 100 L tanks and immersed in running filtered (50 μm) seawater at 16 °C under the ambient 15:9 photoperiod typical of June. To simulate the average seawater temperatures experienced in the region in summer (~ 18 °C), the temperature of the seawater in the tanks was increased gradually over three days, by decreasing the 16 °C water flux and increasing the 18 °C water flux.
After acclimatization at 18 °C, the fronds were placed in 35L tanks with an open circulating system within an isothermal walk-in chamber at ~ 20 °C. Each seawater temperature was assigned to two tanks (total N = 6 tanks). The desired seawater temperatures were achieved by heating the 18 °C seawater with titanium aquarium heaters (100 W) inside head tanks, which supplied the heated water to other tanks containing the fronds. Heated water was mixed with water at 18 °C, and both fluxes were gradually regulated to increase the temperature at a rate of 1 °C per day. The fronds were acclimatized to the seawater temperatures for 12 days. In each tank, 15 fronds of each species were randomly placed and exposed to summer photoperiod conditions of 15:9 h light: darkness ( Supplementary Fig. S5). Light was supplied from above by LED lamps. To prevent any effects due to differences in light within the tanks, the fronds were moved to different positions in each tank every 2 days. To minimize any effects of the tank, fronds were moved every 5 days between tanks containing seawater at the same temperature.
After the acclimatization period, the fronds were subjected to a daily emersion period of 4 h (from 10:00 to 14:00) simulating low spring tides, during 3 consecutive days. The fronds were then subjected to an immersion period (8 days) simulating neap tides and then to another period of 3 consecutive days with a daily emersion of 4 h. After the second emersion period, fronds were again submerged in seawater for 6 days at the respective seawater temperatures to check their recovery. The laboratory experiment lasted for 38 days in total (Supplementary Fig. S6).
During emersion, fronds were transferred from the seawater tanks to tanks containing moistened granite tiles (natural rocky substrata of the algae in the field) placed on shelves inside the same chamber for 4 h, i.e. average emersion time experienced by macroalgae in the mid intertidal zone during spring tides at the sampling site. The fronds were distributed in layers (fronds covering other fronds) simulating their natural distribution in the field. The emersion tanks were lit from above by LED lamps, with the same light intensity and photoperiod as in the seawater tanks. The air temperature was increased gradually during emersion by using infrared ceramic heaters positioned over the tanks, in which the air was re-circulated with small battery-operated fans. The air temperature was regulated with digital temperature controllers positioned above the canopies. Temperature and humidity during emersion were monitored by data loggers positioned below the canopies.
The response variables mortality, wet weight and F v /F m were measured in fronds before each emersion period and 20 h after each (days 18,22,29,33) and after six days of recovery (day 38).
The numbers of dead fronds were recorded. Mortality was considered a binary response variable (1 = dead, 0 = alive). Wet weight was obtained in fronds previously blotted with paper to remove excess water from the surface. Growth was determined as the difference in wet weight 20 h after each treatment and the wet weight before each treatment (emersion cycles 1 and 2/recovery), divided by the number of days, according to 20 . This calculation enabled to directly assess if there was an increase in weight and net growth (positive values), no increase (zero), or decrease in fronds weight due to losses of tissue (negative values) after the emersions and the recovery. The F v /F m was measured before emersion cycles and/or recovery and 20 h later, and the variation in photosynthetic performance (ΔF v /F m ) was calculated using Eq. (1). The C:N ratio was determined in fronds removed from the tanks at the end of the experiment. The fronds were rinsed repeatedly to remove salts and then ground to produce a fine, homogenous powder. The C:N ratio was determined by CNHS elemental microanalysis (in a Fisons Carlo Erba EA1108 elemental analyser) at the analytical facilities of the University of Vigo (CACTI-UVIGO).
To determine whether the Δweight and ΔF v /F m in species differed between treatments in the field, generalized estimating equation models (GEEs) were used, with Treatment (2 levels: emersion vs. recovery) as a fixed factor. GEEs estimate the regression parameters considering the correlation between observations in clustered data 76 . The GEEs had an autoregressive correlation structure and a Gaussian error distribution.
Between-species differences in temperature and humidity in the field were tested using a general linear model (GLM) including Species (3 levels: B. bifurcata, C. tamariscifolia and C. tomentosum) as a fixed factor. In the laboratory experiment, GLMs were also used to test for any differences in seawater temperature and light between the immersion tanks, as well as for differences in air temperature and humidity between the emersion tanks.
To determine whether mortality, growth, ΔF v /F m and C:N ratio differed between treatments in the laboratory experiment, the GLMs included Air temperature (3 levels: control ~ 22.3 °C, atmospheric heatwave ~ 32.7 °C and extreme atmospheric heatwave ~ 35.7 °C) and Seawater temperature (3 levels: control ~ 18.3 °C, marine heatwave ~ 19.9 °C and extreme marine heatwave ~ 22.3 °C) as fixed orthogonal factors. Mortality data were fitted by GLMs with a binomial error structure, whereas growth, ΔF v /F m and C:N ratio data were fitted by GLMs with a Gaussian error structure. The GLMs were only used when treatment groups (Air temperature x Seawater temperature) included at least 4 observations, because in some cases there was an insufficient number of replicates due to the high mortality rate. Therefore, the models did not include the effect of the highest air temperature (35.7 °C) on the growth and the ΔF v /F m of B. bifurcata after the second emersion cycles or the effect of the air temperature factor on the growth or the ΔF v /F m of C. tamariscifolia after the two emersion cycles.
When response variables did not fulfil the parametric assumptions, or when the model was unbalanced, a more conservative significance level of p = 0.01 was established. The goodness of fit of the models was assessed by graphically checking the error distribution, and the statistical significance of the models was calculated using Chisquare tests 77 . When models were significant, post hoc multiple comparison tests were conducted and Bonferroni p value corrections were applied. All data are reported as means ± S.E. Statistical analyses were performed with R 3.5.3 software, by using the "geepack" package 76 for the GEEs and default system packages for the other tests.