HOTFLOOR: A benthic chamber system to simulate warming on the seafloor

The frequency of abnormally warm water events is increasing not only in surface waters, but also in subsurface layers, with major impacts on benthic ecosystems. Previous insights on heatwave effects have been obtained through field observations or manipulative laboratory experiments. Here, we introduce a system capable of inducing elevated water temperatures in benthic habitats in situ over several days. The system consists of a commercially available electric boiler, usually applied in domestic underfloor heating, and custom‐designed benthic acrylic glass chambers connected to individual thermostats. Furthermore, the chambers are semi‐open, allowing constant water exchange, maintaining otherwise near‐natural conditions, including oxygen concentrations, while the temperature is elevated. The water exchange can be stopped to facilitate incubations measuring changes in benthic fluxes. We conducted a 15‐d trial study in July 2021 on a bare‐sediment habitat at 2.5 m depth, exposing five chambers to water temperatures 5°C above ambient temperatures for 6 d and comparing with five control chambers. In this assessment, we demonstrate that the temperature control and stability were reliable while maintaining natural oxygen conditions. The modular character of the system permits adaptations for various benthic habitats, facilitating the investigation of elevated temperatures in situ for future climate change scenarios.

The climate is rapidly changing with far-reaching ecological consequences.While controlled laboratory experiments are useful for investigating ecological mechanisms and interactions, they are limited to extrapolating insights into natural ecosystems.This is mostly because laboratory experiments keep the majority of environmental variables constant except for the variable to be investigated.Mesocosm experiments allow for some more natural variability, but still lack in their realism.Field experiments, on the other hand, mostly aim to keep the majority of environmental factors naturally varying and ideally keep one factor constant (Connell 1974).Compared to terrestrial systems, the marine environment is far more challenging for in situ experimentation, with limited accessibility to study sites and, for example, the challenges with using electricity in seawater environments.The overall challenge generally increases with depth, from the intertidal, which can be accessed relatively straightforward though the highly dynamic system involves different challenges, to the subtidal, where scuba diving is necessary, to the deeper waters where manipulations have to be done remotely (e.g., benthic landers or remotely operated vehicles).Nevertheless, it is imperative that researchers conduct field investigations in natural ecosystems, with their inherent complexity, to understand large-scale ecosystem responses to environmental drivers.
In situ benthic studies often apply benthic chambers to conduct incubations, typically focusing on benthos-water interactions, nutrient cycling, and/or community metabolism.In their simplest design, benthic chambers seal a certain water volume on the seafloor, provide ports for water sampling, and have a stirring mechanism.A variety of chamber systems have been used on bare-bottom sediment (Rasheed et al. 2006;Norkko et al. 2013;Helguen et al. 2014) and fewer in more complex habitats, for example, seagrass (Olivé et al. 2016), coral assemblages (Roth et al. 2019), or hard-bottom habitats (Haas et al. 2013).In standard chambers, it can be difficult to directly manipulate an environmental variable in a controlled and realistic scenario.Nevertheless, indirect manipulations of environmental factors are possible by, for example, inducing hypoxia and studying the community performance afterward (Villnäs et al. 2012).A crucial factor is the duration of the chamber placement.Studies exceeding a certain deployment time, usually a few hours, require a constant movement of the water inside the chamber to avoid the build-up of unnatural physical or chemical gradients, which ultimately requires a circulation system supported by a power source (Camillini et al. 2021).Once conducting longer-term experiments, exceeding a few hours, the exchange of water inside the chamber is required to allow for natural conditions and avoid an accumulation of metabolites and a depletion of oxygen, while at the same time maintaining control over the desired environmental factor.Research in, for example, ocean acidification has made great advances in applying the Free Ocean CO 2 Enrichment system in various habitats.These systems generally consist of multiple semi-open enclosures placed on the seafloor with a flowthrough of ambient seawater that can be manipulated in its pH via the addition of CO 2 .Nevertheless, these systems are highly complex and quite costly (Stark et al. 2019).
Temperature is a major driver of ecosystem processes, and there is an urgent need to understand the impacts of rapidly increasing ocean temperatures on natural ecosystems.Both mean temperature as well as frequency of extreme events such as marine heatwaves (MHWs) are increasing rapidly (Bindoff et al. 2019).For example, the ongoing trend of increasing mean temperatures and extremes has already caused a shift in the baseline temperature for the latest 30-yr period from 1991 to 2020 compared to earlier 30-yr periods in the Northern Baltic Sea (Goebeler et al. 2022).Devastating consequences related to MHWs have been observed and well documented (Garrabou et al. 2009;Smale et al. 2017;Wernberg 2021).Yet, experimental in situ studies that involve warming are rare and hence, our mechanistic insight on their effects on a scale of natural communities remains limited.While conducting experimental work in situ, especially in the subtidal, is accompanied by a number of logistical difficulties, conducting heating experiments is particularly demanding since heating the marine environment requires an exceptional amount of energy due to the high thermal conductivity of water.The most recent and notable advance has been achieved by placing small rigid cylinders with gas-permeable plastic bags and 100 W heaters connected to the battery of a boat for 110 h in a shallow seagrass meadow, resulting in a 3 C above ambient water temperature (Egea et al. 2023).The gas-permeable, flexible bags enabled an oxygen exchange to avoid unnatural concentrations, for example, oversaturation during light hours and anoxia during dark hours, and were claimed to prevent water stagnation through hydrodynamic forcing.It is yet unclear if the temperature increase by 3 C was due to the maximum capacity of the heater or if it was controlled for in the experimental design.
Here, we present our HOTFLOOR system, which we developed in close cooperation with Roth Finland Oy (https:// www.roth-finland.fi/), which provides domestic and industrial heating systems.In essence, it is an underfloor heating system for houses adapted to heat the water in benthic chambers with otherwise near-natural conditions for long-term studies in seafloor ecosystems.We tested this setup in an experiment in the summer of 2021, aiming to investigate the effect of an MHW on macrofauna community structure and associated oxygen and solute fluxes on a shallow, sandy seafloor habitat at 2.5 m depth.Here, we demonstrate that we succeeded in inducing a controlled MHW, with a mean intensity of about 5 C above ambient water temperature and with similar oxygen concentrations as in the surrounding water for 6 d.For the majority of the time, the chambers were semi-open, allowing for water exchange, and only during 4-h flux incubations were the chambers fully sealed.The experimental results regarding changes in the nutrient fluxes and the benthic community structure in response to the induced MHW will be published separately.

General design of the setup
The system was set up on the north shore of the island Furuskär (59 50 0 0.25 00 N, 23 15 0 45.54 00 E; Supporting Information Fig. S1) in the south-west Finnish Archipelago in July 2021.The northern side of the island has a bay-like structure with some skerries and is protected from the prevailing, southwesterly winds.The seafloor of the experimental area is mainly unvegetated, sandy sediment.Ten chambers, with a water volume of about 28.5 L and tubing (⌀ 10.5 mm) on the insides of the walls, were placed on the seafloor in treatment pairs (MHW/control), with each pair about 4 m apart.The pairs were about 10 m away from the heating source at a depth of about 2.5 m.The chambers were placed, with the tubing attached, by divers on 01 July 2021 by gently pushing the walls about 5 cm into the sediment and left for 1 d without a lid to acclimate.The warming of five chambers to about 26 C started on 03 July 2021 and lasted 6 d until 09 July 2021, with the ambient water temperature being, on average, about 21 C. Between 10 July 2021 and the end of the experiment on 15 July 2021, all chambers were at ambient water temperature.To maintain nearnatural conditions, a large hole (⌀ 50 mm) in the wall of the chamber enhanced the water exchange with the surrounding water, which was further facilitated through a constantly running circulation pump with an attached diffuser tube to avoid the build-up of metabolites or unnatural solute concentrations.

Chamber design
For ease of reading, technical specifications (e.g., dimensions, voltages) and a detailed technical drawing of the benthic chambers are presented in Supporting Information Table S1 and Supporting Information Fig. S2.
The circular chambers are 20 cm high, made of transparent, acrylic glass pipe pieces of 500 mm outer diameter (Fig. 1).The top of the chamber has an acrylic glass flange glued on the outside to host the lid of the chamber.The flange comes with a notch to fit a nitrile butadiene rubber O-ring for gastight sealing and holes suitable for wing-bolt screws to attach the lid or for rebars to fix the chambers in place on the seafloor.When placed 5 cm deep into the sediment, the chamber encloses a volume of about 28.5 L and covers an area of 0.1901 m 2 .In addition, we developed chamber extensions of 20 cm, with a 15 mm wide flange glued to the top to lock onto the bottom side of the previously described chamber segment, optionally increasing the chamber height to include more complex, vegetated benthic communities, such as seagrass beds.Yet, we did not apply these segments in this study.
The flat acrylic glass lid contains four air-tight ports that connect to the inside of the chamber.One port is in the center of the chamber equipped with a 15 cm PVC tube (reaching about 10 cm into the chamber) to attach a syringe for water sampling.The other three ports are positioned halfway between the central port and the inner side of the chamber.One port is equipped with a 5 cm PVC tube (reaching about 2.5 cm into the chamber) and left open during sampling.This is necessary to replace the sampling volume with surrounding water and to avoid low pressure possibly disturbing the sediment surface.When sampling with a 100 mL syringe, the change in solute concentration inside the chamber can be neglected (corresponds to 0.004% of the total chamber volume).The other two ports serve the sensor cable of the thermostat and the electric cable of the circulation pump.
The circulation pump was permanently running to avoid the build-up of chemical or physical gradients, particularly a strong thermal gradient.Pretrials, during the developmental phase of this project before this experiment, showed that the pump can be run at maximum power (max.performance 300 L h À1 ) without visibly disturbing the sediment surface when a silicone tube (40 cm length with small holes every 10 cm) is attached to the outlet to diffuse the stream.The pump was attached to the lid with the inlet, covered by a suction filter, pointing toward the sediment.Furthermore, the pump enhanced the exchange with the surrounding water when the 5 cm (diameter) hole on the chamber wall was opened.This was to avoid the accumulation of metabolites and ensure near-natural oxygen conditions.In addition, the hole was covered by an oversized 2 mm mesh to prevent the entry of larger organisms and could be closed with a rubber plug during incubation periods for estimating community metabolism rates.
The temperature and oxygen concentration in each chamber were recorded by placing a HOBO Dissolved Oxygen Logger (U26-001; Onset Computer) in the center of the chamber using strings attached to the clip rails of the tubes.A piece of foam was mounted onto the logger to make it buoyant and to reduce the load on the strings and ensure no disturbance of the sediment.We consciously chose to place devices inside and on the chamber using strings, cable ties, or glue as much as possible to allow for flexibility in future setups and reduce the strain, by drilling, on the structural integrity of the chamber.

Design of the heating system
The heating system is a closed-circuit, water heating system with a 9 kW mobile electric heater, coupled with a circulation pump (Fig. 2).The electric heater was powered by a 16 A power supply converted from a 32 A power source.It can run in two phases, providing 9 or 4.5 kW.The latter was mostly sufficient to maintain the temperature during the MHW treatment.Hence, there is potential for higher temperatures (see Heating potential in Assessment).The heating unit pumped the water into a 10-fold manifold and supplied each of the 10 chambers through a total of 40 m tubing.Ten meters of the tubing were coiled on the inside wall of the chamber fixed in clip rails and the other 30 m served as a connection to the heating pump.This means that 15 m of tubing were available for the placement of the chambers.The inflow and outflow tubing of each chamber in the MHW treatment were fully covered with nitrile-moisture-cured foam insulation (NMC insulation), whereas the tubing of the control chambers had three ca.30 cm insulation pieces to avoid the tubes from spreading apart in the water.The tubing of the control chambers did not require any additional insulation, as they did not provide any heated water.Furthermore, weights (in form of heavy chains) were added on top of the chambers, particularly on the side where the insulated tubing (strong buoyancy) reached the chamber.The manifold was equipped with individual flowmeters attached to the outflow valves, whereas the inflow valves were equipped with an actuator.Each actuator was controlled by a thermostat, with its sensor placed at the center of each chamber.The thermostats were intercalibrated shortly before the experimental start and one thermostat had an offset of 4 C, but measured temperature changes correctly.The 10 programmable thermostats were arranged in two water-proof electrical junction boxes.In conclusion, the heating unit provided the warm water where the flow was regulated by each chamber's thermostat.All pieces of the heating system are commercially available and were provided by Roth Finland Oy.

Sampling procedures
Here, we focus on the sampling procedures relevant to this study (see Discussion for sampling options).Changes in nutrient flux rates, as well as oxygen consumption rates, can only be accurately estimated during incubations when the water exchange is stopped and the enclosure is properly sealed.Nighttime and daytime incubations of solute fluxes lasting 4 h were performed at the start and end of each experimental period (MHW or control).Incubations were initiated by sealing the 5 cm water exchange hole on the side of the chamber with a rubber plug.Shortly after sealing and at the end of the incubation, water samples were withdrawn from the central sampling port by attaching a 100 mL syringe and opening a second port to compensate the sampling volume with surrounding water.After initial sampling, both ports were closed.Once the final sample of the incubation was withdrawn, the Fig. 2. Schematic overview of the HOTFLOOR system developed to investigate changing fluxes during elevated temperatures.The electric heater (1) provides warm water that feeds into the 10-fold manifold (2), which directs the water through polyethylene tubing (3) (thick, solid, black lines show bundles of tubing; red lines provide warmed water, blue lines provide water at ambient temperature) into the benthic chambers (4).Inside the chamber, the tubing forms an about 10 m coil for heat exchange, and the cooled water goes back to the manifold into the electric heater.Actuators attached to the valves (5) of the manifold control the supply of warm water.The opening and closing of the valves is controlled by thermostats (6) (electric boxes with thermostats inside, one for each chamber) with their sensors (7) (thick, dashed, black line show bundles of sensor cables; thin, dashed, black lines with dot at the end show individual sensors) placed inside the chamber.See Supporting Information Table S1 for details on the products used and Supporting Information Fig. S2 for a detailed technical drawing of the benthic chambers.
ports were left open and the rubber plug was removed to allow water exchange to resume.No sediment cores were taken or any other invasive technique applied during the experiment, as this would have caused disturbances to the sediment.

Data analyses
The data presented in this assessment, with the focus to demonstrate the functionality of the system and suitability for measuring benthic fluxes, stem from the HOBO Dissolved Oxygen Loggers (U26-001; Onset Computer).The logger was placed centrally inside the chamber about 5 cm above the seafloor and recorded dissolved oxygen and temperature at 15-min intervals.Oxygen rates were calculated by taking the last recorded oxygen concentration before opening the chamber at the end of the incubation and subtracting it from the last oxygen concentration recorded before the start of the incubation.The difference in concentration was divided by the incubation duration, multiplied by the water volume in the chamber and, considering the area of the chamber, expressed in [mmol] oxygen per square meter per day.In our analysis, we grouped oxygen rates by sampling, time of day (night/day) and treatment (control/MHW).We omitted the treatment grouping for the initial "acclimation" sampling, when all chambers (n = 10) were under identical conditions.A Shapiro-Wilk test indicated non-normal distribution for the "acclimation" night incubation (p = 0.031) and for the MHW group during the night of the "ambient MHW" sampling (p = 0.0474).Consequently, we applied boxplots to effectively visualize the distribution and gain the central tendency (median), which is robust to non-normally distributed data and outliers.

Other tests
Disturbance of surface sediment Pretrials in the development phase of this system indicated that a circulatory pump with a diffuser tube attached and running at 300 L h did not cause any resuspension of the sediment surface.To investigate if this stirring mechanism disturbs the sediment surface when running over 2 weeks, which could cause resuspension of fine particles and influence the biochemical fluxes, inert luminophore tracers (eco-trace©, environmental tracing systems, density 2.5 g cm À3 ) corresponding to the grain size of the sediment were spread inside the chambers in the morning of 02 July 2021.About 70 g of yellow sand and pink silt mixture (ratio 2.3 : 1, respectively) to mimic the sediment structure at the experimental site were evenly spread via scuba diving across the surface of the sediment inside the chamber before the lid was attached.The distribution of the luminophores on the sediment surface was assessed visually over the cause of the experiment.

Heating potential
In an additional setup of the HOTFLOOR system in September 2021 next to the Tvärminne Zoological Station (59 50 0 40.68 00 N, 23 14 0 59.29 00 E), we investigated the maximum heating potential and the power consumption of the system.Ten chambers with fully NMC-insulated tubing (equally long as previously described) were placed in 1 m water depth on sandy sediment with constantly running pumps and open water exchange.Temperature sensors were placed inside the chambers and one additional sensor about 10 m away to measure the ambient water temperature.To estimate the electric power consumption, we observed the electric heater for about 1.5 h and measured for how long it remained in either 9 or 4.5 kW phases (indicated on the display).

Assessment
The system was trialed in a study to investigate the effect of an MHW on the biodiversity and ecosystem functioning of a soft-sediment, benthic community.Here, we demonstrate the successful application of the HOTFLOOR system based on the achieved temperature and oxygen conditions.
We applied a category II ("strong") MHW (Hobday et al. 2018) based on the reference period 1931-2020 of the surface layer from Storfjärden (Goebeler et al. 2022), a monitoring site (59 51 0 33 00 N, 23 15 0 35 00 E) close to the experimental site (about 2.8 km).Five chambers were exposed to this MHW for 6 d, while the five control chambers were at ambient water temperature.However, a real MHW started in the field on 02 July 2021, exceeding the threshold for extreme temperatures (about 18 C) and increased to about 23 C on 15 July 2021.The temperature in the MHW chambers was, on average, about 5 C higher than in the controls (about 21 C) during the induced MHW (Fig. 3).Due to a mix-up in the sensor cables of the thermostats, MHW chamber #3 experienced about 28 C for the first 2 d after the experiment start whereas MHW chamber #5 was at ambient temperatures.The heating was initiated on 03 July 2021 at about 03:00 in the morning and achieved the desired daily average temperature of about 26 C on 04 July 2021 (MHW chamber #5 on 06 July 2021).The average daily temperature variation of the five chambers exposed to the induced MHW (when all chambers were at the desired temperature 06-09 July 2021) was 0.7 C (Supporting Information Table S2).
The key to conducting longer-term in situ studies is to ensure that all environmental parameters, except the one being manipulated, are as close to natural conditions as possible inside the chamber.This requires the water to be constantly exchanged, except for incubation periods, and circulated.The successful circulation and water exchange are demonstrated by using the changes in oxygen concentrations between incubation periods as a proxy.Oxygen levels were recorded inside the chamber every 15 min and compared to ambient oxygen levels recorded with a multiparameter YSI (average depth 2.03 m, 30 cm above the seafloor) placed in the vicinity of the experimental area, about 20 m away at the same depth.Due to a battery failure, the YSI only recorded until 10 July 2021.General oxygen levels in most chambers were slightly lower than ambient between incubations when the chambers were semi-open, and even lower for control 1 (Fig. 4).Control 1 was missing the suction filter for the circulation pump, which increased the intake area, probably lowering the suction pressure.Yet, this did not cause control 1 to be permanently below the critical hypoxia concentration of 2 mg L À1 O 2 or 0.0623 mmol L À1 (Rabalais et al. 2010).For estimating oxygen production and consumption as well as nutrient fluxes, it is essential that the chambers are sealed during incubation periods.We conducted 4-h incubations during night and day times on three different occasions: (1) shortly after installing the chambers when all chambers were exposed to ambient temperatures (02 July 2021; "Acclimation"), (2) after 6 d of exposure to the experimentally induced MHW (09 July 2021; "Induced MHW") and ( 3) after an additional 6 d with all chambers at ambient MHW conditions (15 July 2021; "Ambient MHW") (Fig. 5).
The median oxygen consumption during the acclimation at night was at 46.5 mmol m À2 d À1 (n = 10; interquartile  (n = 5; IQR = 39.9 mmol m À2 d À1 ), for MHW and control, respectively.See Supporting Information Fig. S3 for detailed oxygen concentrations and rates.

Heating potential
In a separate setup, where the chambers were placed at 1 m water depth, we investigated the heating potential and power consumption of the system.While the ambient temperature was at 13.8 C (SD = 0.13), the highest temperature achieved in all chambers was, on average, 26.2 C (SD = 1.75).Nine kilowatt cycles lasted on average about 18.3 min and 4.5 kW about 7 min.This results in an overall electric power consumption of 186.35 kW d À1 (about 7.76 kW h À1 ).Any further temperature increases caused an unbalanced heating across the chambers.

Discussion
We developed a new system suitable to conduct longerterm, warming experiments aiming to investigate in situ responses of benthic communities to ocean warming.At its core, it is a commercially available domestic underfloor heating system repurposed to fit custom-built benthic chambers.The system was trialed in a shallow coastal area with the chambers placed in a soft-sediment habitat.The temperature of the induced MHW was in accordance with the planned temperature and stable (Fig. 3), and warming was achieved even during the occurrence of a natural MHW.Nevertheless, it must be noted that the temperature in these thermostats can only be set in 0.5 C steps.This system can heat 10 chambers with fully insulated tubing up to about 26 C when the ambient temperature is about 13 C, consuming about 186.35 kW d À1 .
Realistic, longer-term studies require achieving and maintaining natural conditions within the experimental unit.Sufficient water exchange and circulation were realized by placing circulatory pumps inside the chamber, permanently running at maximum power (300 L h À1 ).A 40 cm long PVC tube with holes every 10 cm was attached to the pump outlet to diffuse the flow.While this setting successfully mixed the water inside the chamber and allowed for near-natural oxygen conditions without visibly causing increased turbidity, we found signs that the sediment surface had likely been disturbed by this setup.In some chambers, an accumulation of pink luminophores was noticed that was not due to natural conditions, for example, the moving of organisms, but possibly as a result of the water circulation setup (Fig. 6).It might have caused a short-term resuspension of particles, but no measurable or obvious disturbance over the longer term.The diffusion of the stream should be improved and further standardized, possibly by applying a rigid diffusing structure with more outlets lowering the pressure.Mimicking natural conditions inside the chambers in terms of oxygen supply worked well, though not identical, due to the constant water exchange.Slightly lower internal oxygen concentrations can be explained, as the inflow and outflow of water occur through the same hole where ambient and internal water are mixed.Yet, we believe that a limitation of the water exchange is necessary to create a well-mixed environment inside the chamber and to achieve consistent warming in its center.One chamber was lacking the suction filter to the circulation pump, likely reducing suction pressure and, as a consequence, this chamber experienced limited water movement.Nevertheless, the oxygen concentration in this particular chamber only experienced hypoxic conditions of less than 2 mg O 2 L À1 or 0.0623 mmol L À1 (Rabalais et al. 2010) during very short periods and it did not affect the community oxygen rates during incubation periods, as control 1 did not stand out as an outlier.Furthermore, the small variation of oxygen consumption/production rates during the incubations within the treatments highlight the gas-tight sealing of the chambers, as well as a sufficiently large area of the seafloor covered to include a representative benthic community.The oxygen consumption and/or production rates strongly differ between nighttime and daytime incubations, with seemingly lower net consumption rates during daytime, particularly during "Acclimation" and "Induced MHW" sampling.This indicates that during the daytime, the microphytobenthos and/or phytoplankton is receiving sufficient light through the walls to conduct photosynthesis, producing oxygen.The oxygen production is particularly noticeable during daytime incubation on the last sampling occasion, which is possibly related to the observed strong algal bloom that only started toward the end of the experiment.Coastal waters can be very dynamic and to improve the comparison between samplings, information needs to be gathered about what and who is metabolizing during an incubation.The constantly exchanging water, and the plankton communities with it, in the chambers might be the reason (1) that the control chambers during the induced MHW remained at a similar oxygen consumption level as during the acclimation, although the ambient temperature increased between samplings by 4.6 C and (2) that the algal bloom caused higher oxygen consumption rates during the last nighttime sampling, although the temperature was about 2.4 C less than during the induced MHW.It is, therefore, essential in future studies, to additionally measure key variables in the water column (e.g., chlorophyll a content, cell abundance, etc.) to identify and quantify the different drivers of the metabolism.Nevertheless, the treatment effect remains and underscores the crucial influence of temperature on benthic community metabolism and the suitability of this system to investigate these effects in more detail.

Potential modifications
This system allows the warming of an experimental unit in situ while maintaining near-natural conditions for a period long enough to mimic, for example, a MHW.It has now been tested on a bare-sediment seafloor in a short-term experiment with a land-based power supply.There are multiple options to further facilitate the heating potential or reduce the energy consumption, by, for example, insulating the electric heater itself or bundling the warm water tubing separatly from the returning cold water tubing, though reducing the placement flexibility, or separately insulating each tube increasing the material costs or insulating the chambers with neoprene covers at the cost of increasing buoyancy and reducing light availability, and so forth.Furthermore, extensions of the chambers increasing the height to include more complex habitats, for example, seagrass meadows, are possible through increasing the energy demand.In addition, a different mixing strategy to the one presented in this study, would likely be necessary to overcome the obstruction of seagrass to ensure gentle mixing also just above the seafloor, for example, with outlets placed closer to the seafloor.Attaching PVC skirts to minimize water exchange could be an option to investigate hard-bottom substrate communities (Roth et al. 2019).In order for the experimental design to be more spatially flexible and allow investigations of different habitats, it will be necessary to substitute the dependence of a land-based power supply with an independent power supply, possibly by a combination of wind and solar energy (Srednick et al. 2020) and a back-up diesel generator, and placing the heating system on a floating platform.Overall, the system can be adapted to the requirements of the user and the question of interest.

Fig. 1 .
Fig. 1.Benthic incubation chamber with the main ports, and so forth listed (a), as a general drawing (b), and deployed in situ (c).See Supporting Information Fig. S2 for additional details.

Fig. 3 .
Fig. 3. Daily average temperature of control and MHW treatment (n = 5) with daily SD.Linetype: solid-daily average temperature; (following linetypes and color in order from bottom to top:) solid + blue-climatological mean; black + dashed-threshold climatology; dotdash + black-2x_threshold climatology, dotted + black-3x_threshold climatology.MHWs of moderate intensity (category I) are of yellow color, strong MHWs (category II) of orange color and severe MHWs (category III) are of dark red color.The heating was terminated on 10 July 2021 at 03:30 in the morning allowing the MHW chambers to cool to ambient water temperature.

Fig. 4 .
Fig.4.Hourly averaged oxygen concentration between incubations when the chambers were semi-open (in order from top to bottom) by a YSI logger placed in direct vicinity of the experimental area (solid line; n = 1), of the MHW chambers (dashed; n = 5), of the control chambers (dotted; n = 4) and of control chamber 1 as an outlier (dot-dashed; n = 1).Control chamber 1 was missing a piece from the circulation pump, likely reducing water movement, and is shown as an outlier.

Fig. 5 .
Fig. 5. Oxygen consumption of 4-h incubation periods during daytime and nighttime in the acclimation phase (all chambers at ambient temperature summarized in gray, n = 10; 02 July 2021), after 6 d of the induced MHW (controls in light gray, n = 5; MHW in dark gray, n = 5) and after an additional 6 d at ambient MHW (controls in light gray, n = 5; MHW in dark gray, n = 5).

Fig. 6 .
Fig. 6.Luminophores spread in the morning of 02 July 2021 indicate a possible disturbance of the sediment surface inside the chambers.Panel (a) shows evenly distributed luminophores in the afternoon of 02 July 2021.No obvious luminophore accumulations occurred in control chamber 4 on 07 July 2021 (b), although we noted some indications in other chambers.On 15 July 2021 control chamber 4 (c) had clear accumulations of luminophores, not only below to the pump inlet.