Response of foundation macrophytes to near‐natural simulated marine heatwaves

Marine heatwaves have been observed worldwide and are expected to increase in both frequency and intensity due to climate change. Such events may cause ecosystem reconfigurations arising from species range contraction or redistribution, with ecological, economic and social implications. Macrophytes such as the brown seaweed Fucus vesiculosus and the seagrass Zostera marina are foundation species in many coastal ecosystems of the temperate northern hemisphere. Hence, their response to extreme events can potentially determine the fate of associated ecosystems. Macrophyte functioning is intimately linked to the maintenance of photosynthesis, growth and reproduction, and resistance against pathogens, epibionts and grazers. We investigated morphological, physiological, pathological and chemical defence responses of western Baltic Sea F. vesiculosus and Z. marina populations to simulated near‐natural marine heatwaves. Along with (a) the control, which constituted no heatwave but natural stochastic temperature variability (0HW), two treatments were applied: (b) two late‐spring heatwaves (June, July) followed by a summer heatwave (August; 3HW) and (c) a summer heatwave only (1HW). The 3HW treatment was applied to test whether preconditioning events can modulate the potential sensitivity to the summer heatwave. Despite the variety of responses measured in both species, only Z. marina growth was impaired by the accumulative heat stress imposed by the 3HW treatment. Photosynthetic rate, however, remained high after the last heatwave indicating potential for recovery. Only epibacterial abundance was significantly affected in F. vesiculosus. Hence both macrophytes, and in particular F. vesiculosus, seem to be fairly tolerant to short‐term marine heatwaves at least at the intensities applied in this experiment (up to 5°C above mean temperature over a period of 9 days). This may partly be due to the fact that F. vesiculosus grows in a highly variable environment, and may have a high phenotypic plasticity.


| INTRODUC TI ON
Current climate models predict not only an increase in average global temperatures but also increased climate variability characterized by extreme events as heatwaves, floods and storms (IPCC, 2013). Over the past 90 years  both the northern and the southern hemispheres experienced marine heatwaves with increasing intensity (17%) and frequency (34%), resulting in an increase of annual marine heatwave days by 54% (Oliver et al., 2018).
Such extreme events, of sustained hot temperatures, cause noteworthy impacts on humans and local economies and affect terrestrial and marine ecosystems (Easterling et al., 2000;Walther et al., 2002) in various ways, such as by causing species range contractions and extirpations of marginal populations (Smale & Wernberg, 2013;Wernberg et al., 2016). For example, a marine heatwave in early 2011 in Western Australia caused a range contraction of the habitat forming macrophyte Scytothalia dorycarpa by ~100 km, which is ~5% of its global distribution (Smale & Wernberg, 2013).
In Central Europe, 3 weeks of extreme temperatures in summer 2003 (Schär & Jendritzky, 2004) led to sea surface temperatures (SST) of up to 28.8°C in the Mediterranean Sea, about 2.2°C above mean annual maximum temperature (Marbà & Duarte, 2010). Another heatwave hit the same region in 2006 resulting in temperatures of up to 28.5°C. Both heatwave events caused high mortality of the seagrass Posidonia oceanica, with long-lasting ecosystem-wide consequences (Marbà & Duarte, 2010).
Marine macrophytes, that is, seaweeds and seagrasses, represent the dominant flora in coastal ecosystems worldwide. They are ecosystem engineers (Miller et al., 2018;Teagle, Hawkins, Moore, & Smale, 2017) that provide a suite of ecologically valuable functions such as nutrient cycling, carbon sequestration and sediment stabilization (Krause-Jensen & Duarte, 2016;Smith, 1981). They provide habitat to a range of other dependent marine flora and fauna (Christie, Norderhaug, & Fredriksen, 2009;Sogard & Able, 1991;Thomaz & Cunha, 2010) and also act as nursery grounds to diverse juvenile fishes (Cheminée et al., 2013). Hence, macrophytes support one of the most productive and diverse coastal marine ecosystems (Kumar, Kuzhiumparambil, Pernice, Jiang, & Ralph, 2016;Thomson et al., 2015), which also provide a range of commercially important products such as nutraceuticals, pharmaceuticals, food for humans and animals, soil conditioner, biofuels and hydrocolloids (Holdt & Kraan, 2011). Seagrass meadows in particular are important as a nursery ground for a range of economically important fish (Heck, Hays, & Orth, 2003). The brown seaweed Fucus vesiculosus is a perennial species that inhabits cold and temperate regions on the Eastern and Western North Atlantic. The eelgrass Zostera marina is the most widely spread and often dominant marine angiosperm in the northern hemisphere (Den Hartog, 1970), providing many of the above-mentioned ecosystem services (Hughes & Stachowicz, 2009;Rönnbäck et al., 2007).
Thus, their response to climate change, including an increase in the occurrence of extreme events such as marine heatwaves can potentially determine the fate of the related ecosystems and the provided ecosystem functions and economic services.
Temperature can affect various physiological traits of macrophytes, while the magnitude and extent of thermal stress determines the severity of effects. Photosynthesis, the source of carbon and energy in plants, is highly responsive to temperature, following a typical optimum curve (with increasing temperature, the photosynthetic rate increases gradually until reaching its highest rate at optimal temperature, followed by a rapid decrease of photosynthetic rate at temperatures above the optimum, due to damage to the photosystem; Bulthuis, 1987;Graiff, Bartsch, Ruth, Wahl, & Karsten, 2015). The photosynthetically gained energy is allocated into growth, reproduction and cell maintenance, and defence mechanisms against grazers, epiphytes and pathogens. Extremely high temperatures, in turn, can cause stress responses in macrophytes, which may include an overall lower energy input and a reallocation of energy normally used for growth into mechanisms preventing cell damage (e.g. production of heat shock proteins and superoxide dismutase; Bergmann et al., 2010;Harvell, 1998;Ireland et al., 2004;Jueterbock et al., 2014;Winters, Nelle, Fricke, Rauch, & Reusch, 2011). Furthermore, increased temperatures can promote the abundance and/or activity of grazers, pathogens and epiphytes (Harvell et al., 2002;Wahl et al., 2010), which would require higher defence activity by macrophytes against these organisms (Harvell, 1998). Depending on a macrophyte's resilience capacity and fitness (e.g. stress tolerance, abundance of energy reserves, etc.) defence activity may be increased (Saha & Wahl, 2013), remain unaltered with increasing threat (Saha et al., 2014) or be decreased in cases of resource limitation (Sudatti, Fujii, Rodrigues, Turra, & Pereira, 2011). Specific components of the defence mechanism may also be impaired due to temperature stress .
The brown algae F. vesiculosus is generally prone to fouling by micro-and macroepibionts (Saha et al., 2012;Saha, Rempt, Grosser, Pohnert, & Weinberger, 2011;Wahl et al., 2010) and to grazing, for example, by the isopod Idotea balthica in the Baltic Sea (up to 95% loss of biomass, Kangas et al., 1982). At the same time, F. vesiculosus is known to be chemically well-defended against foulers (e.g. Brock, Nylund, & Pavia, 2007;Saha et al., 2011;Wahl et al., 2010) and grazers (Rohde, Molis, & Wahl, 2004). However, relatively little is known about its regulation of chemical defence upon temperature stress (but see laboratory experiments of Wahl et al., 2010;Weinberger et al., 2011). In contrast to F. vesiculosus, the chemical defence of Z. marina is poorly understood, despite being frequently found infected  (Muehlstein, 1989;Sullivan, Sherman, Damare, Lilje, & Gleason, 2013). Whether the infection of Z. marina with L. zosterae is more likely to increase under extreme temperature stress imposed by marine heatwaves is unknown.
There is a general understanding on how the physiology of macrophytes may respond to temperature stress, but there is a lack of understanding on how foundation species may react to temporally imposed stress from marine heatwaves in situ. Most of our knowledge to date comes from experiments conducted on single species under constant temperature stress, neglecting potentially important features of the natural habitat such as fluctuating environmental variables  along with a variety of potentially interacting abiotic and biotic factors. Abiotic factors include, for example, natural sunlight, diurnal and day-to-day fluctuations of temperature and natural nutrient concentrations. Biotic factors comprise the diverse community associated with the macrophyte ecosystems, ranging from microbes to grazers present on the macrophyte surface and in the water column, which interact with the macrophytes and respond to temperature themselves (Werner, Graiff, & Matthiessen, 2016).
Recently, Pansch et al. (2018) described a mesocosm experiment in which benthic in-and epi-fauna associated to macrophyte communities were subjected to different short-term heatwave regimes.
Such experiments, which aim to maintain and manipulate 'near natural' conditions, involve trade-offs between ecological realism and replication. As a result, they often lack statistical power (Kraufvelin, 1998(Kraufvelin, , 1999. Despite low replication, the results of Pansch et al. (2018) concluded that different species-level responses can drive changes at the community level, that the frequency of heatwaves can be an important factor determining changes in community structure and that species that do not respond to heatwaves (50% of all tested species) are rather tolerant to short-term heatwaves.
In the present study, we report the effect of short-term heatwaves (with near-natural fluctuations) on the structurally important macrophytes F. vesiculosus (adults and germlings) and Z. marina (adults).
We tested whether single and sequential heatwaves lead to an increase or decrease in sensitivity to thermal stress (when treatments were compared to the control and among each other) on F. vesiculosus and Z. marina populations of the Baltic Sea. F. vesiculosus germlings were included in the experiment, in order to determine whether early life-history stages (which are critical to maintain seaweed or seagrass beds; Coelho, Rijstenbil, & Brown, 2000) are more sensitive to environmental changes, as often proposed (Andrews, Bennett, & Wernberg, 2014;Capdevila et al., 2019;Coelho et al., 2000;Wernberg et al., 2010). We investigated the impacts of single  .
Three different temperature regimes were applied, each replicated in four independent tanks: (a) a control regime with no simulated heatwave, but natural stochastic temperature variability (0HW); (b) a three heatwaves treatment that simulated the occurrence of two spring/ early summer heatwaves in June and July followed by a more severe heatwave in August (3HW); and (c) a one heatwave treatment that simulated a single more severe heatwave event in August (1HW). A generalized additive mixed model (GAMM) fitted to a 15 year temperature time series recorded at 1.5 m depth in the Kiel Fjord (data provided by the Marine Meteorology research unit of GEOMAR by F Nevoigt, U Hecht and K Bumke) was used to identify long-term changes and within-year temperature variability. The temperature profile of the year 2009 was selected for the control tanks (0HW), since this year constituted the lowest deviation from the GAMM (at 95% confidence interval; for detailed description of the model and the treatment justification, see Pansch et al., 2018). The same GAMM was used to identify temperature anomalies (Pansch et al., 2018). A temperature increase of at least 0.7°C/day over a period of two or more days was defined as a common heatwave within the Kiel Fjord, while increases up to 1.2°C/ day occurred as well. A maximum temperature anomaly of 5.2°C above the average occurred twice in the 15 year time series. Hence, the first (June) and second (July) heatwaves of the 3HW treatment followed a temperature increase of 1.2°C/day over 3 days, remained at the target temperature (3.6°C above the average temperature of the year 2009) for 4 days and dropped to the control conditions over a period of 2 days. The heatwave in August, in the 3HW and 1HW treatments, increased by 1.7°C/day over 3 days, remained at the target temperature  To recreate similar communities to those observed in the field, 14 common organisms found in F. vesiculosus beds and Z. marina meadows were added to the tanks (Pansch et al., 2018) in near-natural densities (Werner et al., 2016). Those included, amongst others, the bivalves Cerastoderma edule and Mytilus edulis, the amphipods Gammarus salinus and Gammarus locusta, the isopod I. balthica, the snail Littorina littorea and a number of sediment dwelling worms. The response of these organisms to the heatwaves has been published by Pansch et al. (2018).

| Macrophyte collection and distribution to the experimental units
Responses below of Z. marina and F. vesiculosus were measured on two individuals from each treatment tank and two individuals from each control tank. Thus, eight individuals were used for each treatment and eight individuals for control.

| Photosynthesis and respiration rates
In each tank, one beaker with two Z. marina plants was placed in a custom-built 6 L cylindrical incubation chamber equipped with a battery-run stirrer and an oxygen sensor spot (PreSens Precision Sensing GmbH) on the inside of the lid allowing optical oxygen measurements through the wall of the chamber. In addition, one beaker with sediment only was placed in a second incubation chamber and served as a control. Both incubation chambers (with and without Z. marina) were filled with water from the respective tank and remained inside the tank to guarantee incubation under similar conditions as experienced by the plants inside the experimental units. Photosynthesis and respiration were measured as changes in oxygen concentration over a 5 hr period. Photosynthesis was measured from 10:00 to 15:00, and respiration was measured after darkening the chambers from 15:00 to 20:00. Incubations for the 12 experimental units were performed on two consecutive days (six on each day). After the incubations, the surface area of the seagrass leaves was determined and used to normalize net photosynthesis and respiration rates (μg O 2 cm −2 hr −1 ).

| Growth
Growth rate was measured on the same two plants (from each tank) used for the incubations described above. Growth marks, consisting of a small plastic ring open on one side, were carefully penetrated through each leaf 3 cm above the base (one ring per leaf) and growth was measured after 2 weeks. The same data were also used to calculate how long it takes to grow a new leaf (P L ). The hole in the seagrass leaves containing the plastic ring was similar in size to holes punched F I G U R E 1 Implemented temperature regime within the experimental units over the experimental period (May-August 2015). Control (0HW), one summer heatwave (1HW) and two late spring/early summer heatwaves followed by a summer heatwave (3HW). Figure has been adapted from Pansch et al. (2018). The dates on X-axis are given as dd-mm-yyyy following general recommendations (Short & Coles, 2001). The ring assisted in finding the growth mark.

| Wasting disease and L. zosterae (pathogen) abundance
Two plants were randomly chosen from the boxes in each tank and were used to quantify wasting disease symptoms by counting the number of leaves with lesions. Subsequently, their leaves were frozen (−20°C) for quantifying L. zosterae abundance and defence capacity of Z. marina against L. zosterae. L. zosterae was quantified using RTqPCR. For this, DNA was extracted using the Invisorb DNA plant kit (Stratek). The manufacturer's protocol was modified by adding 1 µl of untargeted salmon sperm DNA (Invitrogen, Life Technologies) at 500 ng/µl to saturate silica columns with DNA to increase the yield of target DNA . Cell numbers of L. zosterae were determined following a TaqMan-based RTqPCR assay (as described in Bockelmann, Tams, Ploog, Schubert, & Reusch, 2013), using a fluorescent labelled probe binding to the internal transcribed spacer region and standardized L. zosterae DNA solutions of known cell numbers.

| Z. marina defence capacity against L. zosterae
To quantify the defence capacity of Z. marina against L. zosterae, part of the frozen leaves mentioned above (Section 2.3.3) were freeze-dried, homogenized and extracted. Extraction was done at 1/16 of the natural concentration to facilitate good comparisons between treatments, since Z. marina extracts at natural concentrations strongly inhibit L. zosterae growth (Jakobsson-Thor, Toth, Brakel, Bockelmann, & Pavia, 2018). For this, 10 ml methanol/dichloromethane 1:1 was added to each sample for 1 hr, the resulting extract was filtered through a GF/F filter (pore size 0.45 µm) and the solvent was removed by evaporation (Speedvac, at 40°C). Each sample extract was redissolved in 0.5 ml serum seawater media (SSM) and 1% dimethyl sulphoxide. The SSM was prepared with 1 g glucose/L, 0.1 g pepton/L, 0.1 g yeast extract/L, 3 mg germanium dioxide/L, 10 ml horse serum/L, 25 ml streptomycin/penicillin (10 mg streptomycin/ml and 10,000 units penicillin) and dissolved in sterile-filtered sea water (35 psu). The inhibitory capacity of metabolites extracted from Z. marina on L. zosterae growth was investigated in a bioassay described by Martin, Boone, Caldwell, Major, and Boettcher (2009). Extracts were transferred to 6 well plates (Thermo Scientific), and an agar disc (Ø = 7 mm) containing L. zosterae was placed upside down in the centre of each well. The L. zosterae strain used in the assay was isolated from a seagrass meadow just outside the Kiel Fjord in 2015 (Bockelmann et al., 2013). SSM without Z. marina extract served as controls. The plates were sealed with parafilm and incubated in the dark at 25°C for 18 hr. Following incubation, the edge of each colony was outlined, photographed and ImageJ software was used to calculate the area of growth of L. zosterae.

| F. vesiculosus response variables
As for Z. marina, photosynthesis and respiration were determined for F. vesiculosus. However, sensor failure precluded reliable results.

| Growth in adults
In each tank, two of the five F. vesiculosus plants were chosen and two thallus tips per plant were marked by inserting a coloured thread 22 mm below the apical meristem 2 weeks prior to the end of the experiment. Growth rate was determined by comparing the distance between thread and thallus tip measured initially and at the end. To determine the growth rate of germlings, digital images of 10-15 randomly chosen individuals per tank were taken at 40× magnification (Steen & Scrosati, 2004). Individual germlings were chosen randomly during the measurement, since germlings were too small for labelling. The side-view area of the germlings was measured with the image analysis software ImageJ. The mean area of germlings of each experimental population was calculated.

| F. vesiculosus defence capacity against grazers
The defence capacity of F. vesiculosus against grazers was quantified following Rohde et al. (2004). For this, thalli of two individuals from each experimental unit were collected, freeze-dried for 48 hr, weighed and ground. Food pellets were made of each alga by mixing 0.5 g of powdered thallus with 2 ml of deionized water and adding it to a hot agar solution (0.18 g in 2.5 ml deionized water). The gelatinous mixture was quickly poured onto a piece of gauze (mesh size 1.5 mm) and squeezed between two sheets of paper. After cooling for 10 min, the hardened agar sheet with the gauze was cut into 1 × 1 cm food pellets. For the feeding experiment, Petri dishes were filled with seawater (16 psu) and one treatment pellet and one control pellet were added to each dish. I. balthica (isopods, common grazer of F. vesiculosus) was collected from the Kiel Fjord, and single individuals were placed into each Petri dish. Petri dishes were placed in a climate chamber (20°C) under dark conditions. The experiment was stopped for each Petri dish independently, when approximately 50% of one pellet in a dish was eaten. Using the mesh squares of the gauze as a reference, the amount of pellet eaten was quantified.
Five subreplicates were used per treatment level to account for the possible variation in feeding rates. Preference ratios were calculated based on the amount of treated and control pellets consumed.

| Statistics
Each experimental unit (or tank) served as a true replicate (n = 4 for each treatment, n = 4 for control). Each replicate was the mean of two individuals measured per experimental unit. Generalized linear mixed models (GLMM) and permutation-based ANOVAs (PERMANOVA) were used to test the effect of treatments on each variable measured.
The GLMM (Bolker et al., 2009;Zuur, Ieno, Walker, Saveliev, & Smith, 2009) were preformed using the lme4 package from the software R (Bates et al., 2017). The heatwave treatments were included as fixed effects, testing the hypotheses that 1HW is different from 0HW, 3HW is different from 0HW and 1HW is different from 3HW.
Since the KOB experimental units (tanks) are arranged in pairs (6 × 2 pairs) along a floating platform and the treatments were equally spaced between units, the identity of these pairs was considered as a random intercept adjusting potential sources of variability related to the spatial arrangement of tanks. The gamma distribution and a logarithmic link function were used for all models. Differences between treatments were expressed as logarithmic response ratios calculated using the GLMM estimates (see a detailed description of the calculations in Data S1). Diagnostic plots of residuals were visually inspected for every model (see Figure S1).
For the PERMANOVA approach, the PERMANOVA+ add-in (Anderson, Gorley, & Clarke, 2008) for PRIMER v7 (Clarke & Gorley, 2015) was used. Although the use of permutations in PERMANOVA avoids an assumption of normality for each of the variables, there is still an assumption of homogeneity of dispersions among treatments.
This was assessed using the PERMDISP routine and for the majority of variables there was no significant heteroscedasticity (p > .05). A log transformation was used to stabilize the variance of Z. marina photosynthesis measurements. PERMANOVA is known to be robust to small departures from homogeneity, so no transformation was used for F. vesiculosis germling survival (PERMDISP p = .050). The variance of counts of epibacteria on F. vesiculosis could not be stabilized, so an alternative nonparametric test (ANOSIM; Clarke & Green, 1988) was used to check the PERMANOVA results. For each variable a one-way PERMANOVA (999 unrestricted permutations of the raw data, Type III sum of squares) with subsequent pairwise tests was implemented.
Respiration rates changed among treatments; however, no significant differences were found due to the high variability between replicates (0HW: 0.27 ± 0.07 μg O 2 cm −2 hr −1 ; 1HW: 0.58 ± 0.51; 3HW: 0.18 ± 0.24). The abundance of wasting disease, the abundance of L. zosterae and the defence against L. zosterae were not affected by the applied treatments. Signs of wasting disease were low with 19.5 ± 5.3% (0HW), 12.0 ± 9.5% (1HW) and 15.0 ± 8.1% (3HW) of leaves with lesions. The defence capacity against L. zosterae was not affected by the heatwave treatments, with growth inhibition of L. zosterae being 58 ± 9% in 0HW, 52 ± 17.9% in 1HW and 46 ± 16.9% in 3HW. In contrast to the GLMM, PERMANOVA revealed a significant difference between treatments only in Z. marina leaf production, while Z. marina growth was (marginally) insignificant (p = .078). Furthermore, pairwise tests revealed that the differences were found between the 1HW and 3HW treatments, and that neither of these differed significantly from the control (0HW) treatment.

| Fucus vesiculosus
GLMM indicated that none of the response variables for F. vesiculosus changed significantly between treatment and control tanks, except for the abundance of epibacteria (Figure 2, Table 1). Epibacterial abundance was significantly lower in 1HW (1.5 × 10 3 ± 4,400 cells/ cm 2 ) compared to 0HW tanks (3.0 × 10 3 ± 1,500 cells/cm 2 ), while they were similar in the 3HW (3HW: 2.7 × 10 3 ± 8,050 cells/cm 2 ) and 0HW tanks. Growth rates of adult algae were 0.67 ± 0.08 mm/day in 0HW, 0.45 ± 0.40 in 1HW, and 0.57 ± 0.36 in 3HW. Growth rates of F. vesiculosus germlings were 0.74 ± 0.66%/day in 0HW, 0.90 ± 0.44 in 1HW and in 0.75 ± 0.52 in 3HW. Survival of germlings was generally high with 92 ± 5.1% in the control and 87 ± 4.35% in 1HW and 85 ± 17% in the 3HW treatment. The defence capacity of F. vesiculosus against bacterial foulers and grazers was not significantly affected by the heatwave treatments. PERMANOVA (Table 2) followed the GLMM results, revealing a significant change due to applied treatments in epibacterial abundance only. As dispersions could not be stabilized, a nonparametric ANOSIM test was used to confirm the significance of this difference (ANOSIM R = 0.523, p = .012). Pairwise tests showed that the difference was only between the 0HW and 1HW treatment.
F I G U R E 2 Mean logarithmic response ratios for 13 response variables measured in Zostera marina and Fucus vesiculosus, showing the proportional change in the means of one heatwave (orange, 1HW/0HW) and three heatwaves (red, 3HW/0HW) treatments in relation to the 0HW, and 3HW in relation to 1HW (green, 3HW/1HW). Response ratios and 95% confidence intervals were estimated using generalized linear mixed models. Asterisks show logarithmic response ratios significantly different from 0 (*p ≤ .05, **p < .01 and ***p < .001), meaning that the ratio between numerator and denominator was different from 1 (0 in logarithmic scale)

| D ISCUSS I ON
Although there is a growing body of literature on the possible effects of temperature stress on macrophytes and their related ecosystems (Duarte et al., 2018), very few studies attempted to conduct experiments applying realistic heatwave regimes (but see Ehlers, Worm, & Reusch, 2008;Winters et al., 2011), and even fewer studies mimicked near-natural conditions i.e. including natural seasonality, stochastic and daily variability (but see Pansch et al., 2018). In order to overcome the lack of near-natural experiments, we used an outdoor mesocosm approach. Our experiment: (a) was conducted in relatively large mesocosms to overcome the limits of small-scale laboratory investigations and to reduce potential impacts from wall effects; (b) included the natural variability in temperature (in contrast to a steady-state temperature regime); (c) had a constant flow-through system utilizing unfiltered seawater directly from the Fjord (no storage tank) and included a number of macrophyte-associated organisms, which potentially entailed indirect heatwave effects on macrophytes.
Both habitat-forming macrophytes, Z. marina and F. vesiculosus, showed little or no response to the occurrence of a single 9 day summer heatwave or successive 9 day heatwaves from late spring to summer (9 days = 3 days of temperature increase, 4 days of sustained high temperature and 2 days of temperature decrease). At the end of the experiment, the only significant effect from the applied heatwave treatments was found in the growth of Z. marina (linear extension rates and time required for leaf production) after three successive heatwaves, and in the abundance of F. vesiculosus epibacteria after one heatwave. Although the growth of Z. marina was reduced by the sum of three consecutive short-term heatwaves, the maintained photosynthetic rate can likely restore Z. marina's biomass after the heatwaves. Furthermore, potential indirect heatwave effects through changes in the abundance of the associated fauna could not be detected. Indeed, the fauna with potentially the strongest effect on macrophytes (the grazer community) did not change in abundance and biomass (Pansch et al., 2018). This fauna includes the snail L. littorea that grazes upon F. vesiculosus and Z. marina epiphytes, thereby sustaining high light availability for photosynthesis, and the grazer I. balthica feeding preferably on F. vesiculosus.
In contrast to the present study, Pansch et al. (2018) found mixed responses of the macrophyte-associated invertebrate community to the simulated heatwaves of the same experiment. About 50% of the benthic, free living and infaunal species showed either positive or negative responses in abundance and/or biomass demonstrating a large range of invertebrate susceptibilities to heatwaves leading to shifts in community structure (Pansch et al., 2018).
In contrast to the findings by Winters et al. (2011), showing that a continuous 3 week heatwave treatment impaired seagrass photosynthesis, our results showed that photosynthesis was not affected by a single or several 9 day heatwaves. This indicates that the length of heatwaves is likely critical with respect to the detrimental impact of increased temperatures on species and communities. Indeed, an optimum temperature range between 25 and 30°C was found during a short-term experiment with Z. marina of the temperate East coast of the USA (Marsh, Dennison, & Alberte, 1986), suggesting that the temperature initially increases the activity of photosynthesis-related enzymes. Elevated temperatures over longer time periods (weeks), however, damages the photosystems of Z. marina and leads to reduced photosynthetic rates (Bulthuis, 1987;Nejrup & Pedersen, 2008). Respiration rates likewise did not vary significantly in the current experiment between controls and treatments after the more severe summer heatwave, indicating that these extreme events had no lasting effects on the Z. marina metabolism.
We hypothesized that high temperatures would increase the abundance and activity of pathogens in Z. marina, resulting in a further increase in the energy invested into antimicrobial defence and repair of damage caused by pathogens. However, the assess- among the treatments. In contrast to studies on phenolic acids, considered as deterrents of L. zosterae and known to decrease with warming (Vergeer, Aarts, & De Groot, 1995), an abrupt increase of temperature for few days did not reduce the production of inhibitory compounds against L. zosterae in the present study. Altogether, little is known about how a changing environment affects secondary metabolites in seagrasses (Zidorn, 2016 (Wahl, Saderne, & Sawall, 2016;Walter, Jentsch, Beierkuhnlein, & Kreyling, 2013).
Reduced growth and prolonged leaf production rates may entail a decrease in biomass, in particular, if seagrass consumption is high. In the Baltic Sea, however, there is no evidence for significant Z. marina consumption. In the neighbouring North Sea, Z. marina consumption was estimated to be <10% of the annual production, where the main consumers are birds and the isopod Idotea chelipes (Nienhuis & Groenendijk, 1986). In the Baltic Sea, the isopod I. balthica (inhabiting a similar ecological niche as I. chelipes in the North Sea; Leidenberger, Harding, & Jonsson, 2012) was included in the mesocosms, but was not found to feed on Z. marina (no evident grazing marks, Y. Sawall, personal observation). Therefore, we conclude that even though Z. marina growth is impaired by three heatwaves, overall, short-term heatwaves may not necessarily be detrimental. Also, the capacity of Z. marina to store large amounts of carbon (energy) in rhizomes (up to 90%; Olsen et al., 2016), may allow for buffering short-term stress events such as heatwaves (Carruthers & Walker, 1997 (Pansch et al., 2018).
Given that growth and leaf production rate of Z. marina was negatively impacted by the three subsequent heatwaves, we may speculate that an increase in heatwave frequency and intensity, as predicted, may eventually reduce the abundance of Z. marina. This may only be mitigated by selection of more stress-resistant and phenotypically plastic Z. marina genotypes after heatwave events as suggested previously for the Baltic Sea (Bergmann et al., 2010;Franssen et al., 2011;Jueterbock et al., 2016;Winters et al., 2011).
In contrast to the overall low impact of simulated heatwaves on the macrophytes, the macrophyte-associated invertebrate community showed substantial community shifts, in particular after three consecutive heatwaves (Pansch et al., 2018). These community shifts did not show immediate effects on the macrophytes, but could eventully lead to a negative feedback on the macrophyte community.
Mesocosm approaches have the potential to provide a more realistic understanding of the impacts of natural temperature extremes on marine communities than small-scale laboratory experiments . They come, however, at the cost of considerably increased investment, which limits replication (in the present study n = 4 tanks for each temperature regime), and thereby statistical power (Kraufvelin, 1998(Kraufvelin, , 1999. At the same time, environmental variability is higher in outdoor mesocosms than in more controlled indoor conditions, which may be a reason for the relatively high response variability observed in our study. A further explanation for a high response variability may be a generally high genotypic diversity in a species-poor coastal ecosystem of the Baltic Sea (Reusch, Ehlers, Hammerli, & Worm, 2005). This, in turn, was previously hypothesized to enhance ecosystem resilience due to its ability to buffer against extreme climatic events (Reusch et al., 2005). Additionally, the Baltic Sea currently provides combinations of multiple stressors that mimic those expected for many coastal areas in the future , training organisms to tolerate abiotic stressors like acidification (Thomsen, Casties, Pansch, Körtzinger, & Melzner, 2013).
Our study aimed to understand the response of two foundation macrophytes to realistic heatwave scenarios of the Western Baltic Sea under near-natural conditions (e.g. presence of common associated faunal species and fluctuating abiotic conditions). Simulated short-term heatwaves, either as a single summer heatwave or as a series of two spring and one summer heatwaves, showed an overall low impact on the morphology, physiology and the chemical defence of Z. marina and F. vesiculosus. A negative impact was only evident in reduced growth of Z. marina under reoccurring heatwaves.
While we cannot rule out the possibility of longer term effects, maintained photosynthesis and respiration indicate that recovery from biomass loss is likely. The high variability of response variables may partly be explained by the low replication of this mesoscale experiment. However, it can also be well explained by the fact that in particular F. vesiculosus is adjusted to a highly variable environment.
The consequent high physiological and metabolic plasticity of F. vesiculosus may, therefore, be an indicator of its tolerance capacity . While we cannot conclude that the species is resilient based on the experimental results alone, additional lines of evidence such as its habitat preferences suggest that certain populations may be well adapted to short-term extreme temperature fluctuations as applied in our study.