An in situ approach for measuring biogeochemical fluxes in structurally complex benthic communities

Handling Editor: Clive Trueman Abstract 1. The exchange of energy and nutrients are integral components of ecological functions of benthic shallow-water ecosystems and are directly dependent on in situ environmental conditions. Traditional laboratory experiments cannot account for the multidimensionality of interacting processes when assessing metabolic rates and biogeochemical fluxes of structurally complex benthic communities. Current in situ chamber systems are expensive, limited in their functionality and the deployment is often restricted to planar habitats (e.g. sediments or seagrass meadows) only. 2. To overcome these constraints, we describe a protocol to build and use non-invasive, cost-effective and easy to handle in situ incubation chambers that provide reproducible measurements of biogeochemical processes in simple and structurally complex benthic shallow-water communities. Photogrammetry tools account for the structural complexity of benthic communities, enabling to calculate accurate community fluxes. We tested the performance of the system in laboratory assays and various benthic habitats (i.e. algae growing on rock, coral assemblages, sediments and seagrass meadows). In addition, we estimated community budgets of photosynthesis and respiration by corals, rock with algae and carbonate sediments, which were subsequently compared to budgets extrapolated from conventional ex situ single-organism incubations. 3. The tests highlight the transparency (>90% light transmission) of the chambers and minimal water exchange with the surrounding medium on most substrates. Linear dissolved oxygen fluxes in dependence to incubation time showed sufficient mixing of the water by circulation pumps and no organismal stress response. The comparison to single-organism incubations showed that ex situ measurements might overestimate community-wide net primary production and underestimate respiration and gross photosynthesis by 20%–90%. 4. The proposed protocol overcomes the paucity of observational and manipulative studies that can be performed in in situ native habitats, thus producing widely applicable and realistic assessments on the community level. Importantly, the tool provides a standardized approach to compare community functions across a wide


| INTRODUC TI ON
Anthropogenic environmental change is rapidly transforming the community composition, structure and functioning of coastal and estuarine benthic shallow-water ecosystems on a global scale (Harley et al., 2006). Processes such as primary production, calcification, organic matter (OM) remineralization and nutrient cycling are important indicators of ecosystem status, health and alteration (Cloern et al., 2016;Halpern et al., 2008;Lotze et al., 2006).
However, the complexity of interactions underlying these processes requires a holistic assessment with accurate measurements of community metabolism and biogeochemical fluxes (Griffiths et al., 2017).
Consequently, community-wide and standardized measurements of biogeochemical properties of benthic communities are a prerequisite for ecosystem management (Brierley & Kingsford, 2009). However, obtaining such data has proven challenging.
Currently, metabolic rates and element cycling processes of various benthic communities are mainly derived from experimental studies conducted in aquaria or mesocosm systems (Althea & Duffy, 2016;Bellworthy & Fine, 2018;Comeau, Carpenter, Lantz, & Edmunds, 2015;Russell, Thompson, Falkenberg, & Connell, 2009;Wagenhoff, Lange, Townsend, & Matthaei, 2013). In such studies, benthic communities are reconstructed ex situ according to information gathered during benthic field surveys. Mesocosms (continuous flow or static) provide a unique way to measure biogeochemical exchange rates under controlled conditions. These traditional methods, however, are seldom capable of accommodating the complexity and variability in natural systems (Riebesell, Fabry, Hansson, & Gattuso, 2010). Moreover, substrates and individual organisms must be actively removed and incubated ex situ, making this process destructive and prone to experimental artifacts. Additionally, community measurements in ex situ flumes and mesocosms are extremely costly, and few laboratories are capable of accommodating the complexities of natural systems.
Most community-wide in situ measurements of metabolic and biogeochemical processes have generally quantified spatial geochemical changes in the water column using the Eulerian (e.g. Falter, Lowe, Atkinson, Monismith, & Schar, 2008) or Lagrangian (e.g. Gattuso, Pichon, Delesalle, Canon, & Frankignoulle, 1996) flow respirometry technique. These methods involve measuring the upstream-downstream changes of the chemical properties in a parcel of water in a unidirectional flow field. This approach has difficulties in accurately tracking the movement of water parcels due to the complex topography of many habitats, resulting in a significant margin of error (e.g. Shaw, Phinn, Tilbrook, & Steven, 2014). Moreover, water must have sufficient contact time with the substrate for its chemistry to be affected by processes of interest, limiting the resolution of geochemical measurements (Monsen, Cloern, Lucas, & Monismith, 2002).
These chambers effectively isolate a limited volume of water over individual organisms, enabling short-term incubation experiments.
While providing valuable measurements of processes associated with individual organisms, community-wide responses cannot be achieved. Nonetheless, single-organism incubations are the foremost approach to extrapolate responses of representative taxa of a known habitat to community-wide budgets (e.g. Cardini et al., 2016;Eidens et al., 2014;Naumann, Jantzen, Haas, Iglesias-Prieto, & Wild, 2013).
In recent years, larger in situ incubation chambers (5-120 L volume) have been used across different benthic shallow-water habitats (Hughes, Atkinson, & Ansell, 2000;L'Helguen et al., 2014;Tengberg et al., 1995). These chambers enclose both the underlying substrate and overlying water to assess areal rates of production and uptake by quantifying temporal changes in the overlying water chemistry.
They provide a non-invasive technique that better reflects the ambient conditions, nevertheless, these studies have often been restricted to plain two-dimensional (2D) habitats, such as sediments (e.g. L' Helguen et al., 2014;Rasheed, Wild, Jantzen, & Badran, 2006;Tengberg et al., 1995) or seagrass meadows (Silva, Santos, Calleja, & Duarte, 2005). Here, fluxes could simply be related to the covered seafloor area and the known water volume of the chamber.
The application of such systems has been notably absent for communities with structurally complex, three-dimensional (3D) structures (e.g. rocky beds, oyster banks or coral reef habitats). Photosynthesis, calcification and OM cycling, however, are all examples of processes affected by the volume and surface area of the organisms, yet no tools have been available to quantify these variables in situ non-intrusively.
Previous studies that measured community processes on intertidal rocks (Tait & Schiel, 2010) or coral reefs Yates & Halley, 2003) considered the 2D area of the seafloor alone; hence, results may have a large margin of uncertainty.
Given the above, a reliable, versatile and widely reproducible method for assessing the biogeochemical functioning of benthic range of benthic habitats. We identify multiple experimental strategies, including the manipulation of stressors/factors, and discuss how the method may be implemented in a variety of aquatic studies.

K E Y W O R D S
3D models, community budgets, field-based measurements, incubations, photogrammetry, respiratory chambers shallow-water communities in their natural environment is lacking.
To address these challenges, we developed an end-to-end in situ incubation protocol that aims at measuring biogeochemical fluxes associated with whole benthic communities. The technique can be applied across a wide range of benthic shallow-water habitats, from simple (e.g. sediments) to complex (e.g. corals or rocky beds).
We inform on material selection and construction processes to build cost-effective and easy to handle in situ chambers made from ridged polymethyl methacrylate cylinders. Combined with recent advances in computer-assisted photogrammetry, this approach offers a new way for accurately measuring all biogeochemical fluxes in a non-intrusive manner. We present data from laboratory tests, field incubations and, subsequently, compare our results to data from a commonly applied protocol of estimating community budgets from incubations of single organisms. Lastly, we discuss how the method may be implemented in a variety of marine and aquatic ecological studies that include standard and manipulative experimental approaches.

| Construction of benthic incubation chambers
Benthic chambers were constructed from 5-mm-thick polymethyl methacrylate (PMMA) cylinders (diameter: 0.50 m, height: 0.39 m) with a removable, gas-tight lid of the same material (0.40 m total height) (Figure 1; Information Table S1) sits in a gland to create a gas-tight seal between chamber and lid. Once mounted correctly, the chambers enclose a theoretical volume of 66.8-76.7 L of water (depending on the depth of insertion into the ground) and cover 0.2 m 2 of the seafloor.
Due to their flat surface, lids can be easily modified to accommodate measurement and sampling requirements for the desired experiment. In the setup presented here, all chambers were equipped with autonomous recording dissolved oxygen (DO) sensors (HOBO U26, Onset Computer Corporation, Cape Cod, USA; precision 0.02 mg/L, F I G U R E 1 Benthic incubation chamber deployed in situ over a natural coral reef community (a), and general layout of the chamber (b).
Furthermore, the chambers were equipped with two sampling ports: (a) One one-way stopcock female to male luer-lock (Supporting Information Table S1); (b) one stopcock ¾ inch with barbed hose adapter (Supporting Information Table S1). Among others, these sampling ports offer the possibility for controlled sampling of many biological and chemical variables, such dissolved organic carbon (DOC), dissolved inorganic nutrients, dissolved inorganic carbon (DIC), total alkalinity (TA), bacterial or viral samples.
All chambers were equipped with individual water circulation pumps to ensure that withdrawn water samples for later analysis are representative of the entire volume and to avoid the development of chemical gradients (Supporting Information Figure S4). An adjustable flow control further allowed to mimic hydrodynamic conditions under ambient conditions. For this, 6 V DC motors (300 mA, 2 W) with a magnetic impeller (Supporting Information when withdrawing a water sample to avoid a reduced pressure inside the chamber and to replace the withdrawn volume of water by water from outside the chamber rather than by sediment pore water.
Considering that 200 ml of water is sufficient for the concurrent measurement of inorganic nutrients, TA, DIC, DOC and fluorescent dissolved OM, a water exchange during sampling of 0.26% is negligible. Furthermore, the dilution of the water body inside the chambers with surrounding seawater can be easily corrected for mathematically. Additionally, the two sampling ports are positioned opposite of each other on the lid, reducing the risk of sampling water from the inlet valve given sufficient mixing by the pumps.

| Cleaning procedures
After field deployment, the incubation chambers can be rinsed with fresh water. Wing-head screws, valves and other attached parts can be removed and cleaned separately. All materials of the presented setup can be washed with 4% HCl solution if necessary (e.g. for reliable DOC measurements). We recommend rinsing with deionized water before long-term storage.

| Surface area and volume quantification of structurally complex habitats
To evaluate physiological and biogeochemical processes of complex habitats, it is crucial to determine the volume and surface area of the enclosed benthic communities. Here, we quantified the 3D surface area and volume of a community in a chamber by photogrammetric techniques involving computer modelling discussed in Lavy et al. (2015) and Gutierrez-Heredia, Benzoni, Murphy, and Reynaud (

| Calculating community fluxes from discrete water samples and continuous measurements
Community fluxes of water chemistry parameters can be calculated from discrete water samples or from continuous measurements by sensors (e.g. dissolved oxygen loggers). All fluxes should be normalized to incubation water volume (V; in L), incubation duration (I T ; in h) and the surface area (SA; in m 2 ) of the enclosed benthic community.
Community fluxes for a water chemistry variable (Var) can be calculated from discrete start-end point sampling as: For variables that can be continuously measured (e.g. dissolved oxygen) community fluxes of monotonic metabolic processes (e.g. photosynthesis and respiration) can be calculated from linear regressions. Here, the slope (m) of a linear regression is used to calculate flux rates as: As full time series are often nonlinear at times (e.g. the rate of O 2 initially decelerates as the organisms equilibrate after handling), calculations using the full time series can result in underestimating metabolic rates. Truncating the data to exclude these nonlinearities manually is subjective and difficult to reproduce. Therefore, we recommend using the "LoLinR" package for R (

| Laboratory assays for leakage testing and light transmission
To ensure that a watertight seal can be created around benthic communities, chambers were secured on different substrates in 200-L holding tanks in the laboratory according to the protocol above. Six replicate chambers were placed on coarse reef sand (1-3-mm-grain size) and six on coarse rock/rubble pieces (c. 5 cm) respectively. After closing the chambers, red food colour dye Allura Red (Red 40) was injected into the chambers via the luer-lock valve. All chambers were equipped with temperature, light and DO loggers and the circulation pumps were switched on to simulate field deployment conditions. An initial 50-ml aliquot of water was withdrawn from each chamber and from the surrounding tank water (background control) using a syringe for initial concentrations. Each chamber was then left in the holding tank for 6 hr before another water sample was collected from inside the chamber and the surrounding water. A Trilogy Laboratory Fluorometer with an absorbance module (Trilogy 7200-050) was used to measure the absorbance of light at 504 nm by each sample to determine if any dye had transferred from inside the chambers to the surrounding tank water. A series of volumetric dilutions made from a stock solution of known Allura Red concentrations (7-point calibration) was used to determine the concentration of Allura Red in a sample.

| Incubations of simple and structurally complex benthic substrates
To validate the versatility of the system on various substrates and conditions, incubations for DO production measurements were car-

| Case study: Coral reef community budgets of primary production and comparison to ex situ singleorganism incubations
To illustrate the applicability over simple (i.e. sediments) and structurally complex (i.e. coral and algae) communities, we conducted a case study at a coral reef in the Central Red Sea. Field tests were carried out at Abu Shosha reef located on the west coast of Saudi Arabia (22°18′16.3''N; 39°02′57.7''E) in July 2017. The reef is characterized by a patchwork of coral-and algae-dominated communities that are distributed on an area of coarse carbonate reef sand at a water depth of 4-7 m. For the incubations, four coral-dominated communities surrounded by sediment (>40% coral cover, but <10% algal cover), four rocky communities overgrown by heterogeneous assemblages of algae surrounded by sediment (hereafter "algaedominated"; >40% algae, but <10% coral cover) and carbonate sediments (hereafter "sediments") were selected haphazardly with a minimum distance of 3 m. All communities were incubated for measurements of net primary production (NPP) and respiration (R) according to the protocol above (detailed procedures for the case study can be found in the supplementary material). To estimate gross primary production (GPP), each R measurement was added to its corresponding NPP measurement (NPP + |R| = GPP), assuming that respiration rates in the light were equal to those derived from dark incubations (covered chambers) during the day (Bidwell, 1977).
For comparison with other benthic systems and methods, resulting contributions were expressed in mmol C m −2 h −1 .
Subsequently, in situ community measurements were compared to ex situ community budgets estimated from single-organism incubations-a commonly applied technique to measure community budgets of structurally complex habitats, such as coral reefs.
Community budgets from single-organism incubations were esti-

| Light
Spectral analysis showed a light transmission through the chamber material averaging 92% within PAR (photosynthetic active radiation; 400-700 nm) (Figure 4a), while light within the UV range was attenuated (16% and 1% transmission for UVA and UVB respectively).
During field deployment, on average 9.4% reduced light availability within PAR was measured (90.6% transmission) for incubation chambers compared to outside of the chambers at the same time and water depth (Figure 4b).

| Incubations of hard-bottom and soft-bottom communities
Incubations of both structurally complex (i.e. "rocks with algae" and "corals on sand") and simple (i.e. "carbonate reef sediments" and "seagrasses") habitats revealed a production of DO over time (Figure 5a).
Incubation chambers that were deployed on hard-bottom substrates (i.e. rocks with algae) had detectable water intrusions from outside the chamber, resulting in an average linearity of the DO production of In an intertidal habitat with high wave action, chambers without additional weights on the PVC skirts were shaken by the water movement, resulting in a nonlinear DO production (r 2 = 0.689 ± 0.211; Pearson correlation; n = 4; <20%-50% of the data points included in F I G U R E 3 Leakage test of incubation chambers on different substrates. Six replicate chambers each were placed in large holding tanks on rock/rubble pieces (c. 5 cm) and coarse reef sand (1-3 mm), and red food dye was injected into the chambers. The concentrations of Allura Red dye (determined by absorbance of light at 504 nm) was measured from samples in the chambers after adding the dye (referred to as "Before") and after 6 hr of incubation (referred to as "After 6 hr"). Boxplots showing the median (line across a box), quartiles (upper and lower bounds of each box) and extremes (upper and lower whisker). The ends of the whisker are set at 1.5*IQR above the third quartile (Q3) and 1.5*IQR below the first quartile (Q1). Scale bars are given in each picture

Chambers on rock
After 6

| Community-wide budgets for photosynthesis and respiration in coral reef habitats
In situ incubations revealed a significant difference between benthic coral reef habitats (i.e., coral-dominated vs. algae-dominated vs. sediments) for NPP, R and GPP (Figure 6a; detailed statistical test results in Supporting Information F I G U R E 5 Dissolved oxygen production during incubations of various substrates (a), with/without active circulation pumps (b) and under high wave action in an intertidal zone with/without additional weights (c). Shown is the production of dissolved oxygen over time (incubation periods ranging from 90 to 120 min), corrected for seawater background metabolism and normalized to the surface area of the incubated substrate. Points of different colours represent independent replicate incubations of the respective substrate. Opaque points with lines ( , , and ) represent data points of incubations that were included by the "LoLinR" package for R (Olito et al., 2017) in a local linear regression for dissolved oxygen calculations. The amount of points included in the local linear regression is given in per cent in each plot. As recommended by Olito et al. (2017), an asterisk indicates that <20% of the data points followed a linear trend, hence model assumptions were not met. Transparent points ( , , and ) represent all data point of the respective incubation that did not follow a linear trend

| Comparison of in situ and ex situ estimates of community-wide oxygen fluxes
The comparison of community oxygen fluxes between in situ and ex situ incubations revealed significant differences between these methods (Figure 6, A; detailed statistical test results in Supporting Information Table S3). Community fluxes derived from ex situ incubations showed significantly lower R rates for each of the three habitats compared to in situ incubations (25%, 62% and 92% lower for coral, algae and sediments respectively). Furthermore, ex situ incubations overestimated NPP by 35% in the case of coral-dominated communities, and by 82% in sediments. While GPP among the two methods was similar in coral-dominated communities, it was underestimated by ex situ incubations by >50% both in algae and sediments. As a consequence, ex situ extrapolations were not able to resolve differences in community metabolism between algae and coral-dominated communities.

| D ISCUSS I ON
Most marine and aquatic ecosystems are presently under threat due to the combined effects of anthropogenic perturbations (Halpern et al., 2008). Understanding and predicting the responses of these ecosystems to stress requires considering processes across all levels of biological organization (Brierley & Kingsford, 2009). While current knowledge is advancing rapidly on scales ranging from molecular to whole-organism levels, our understanding of responses of whole communities in their natural environment is limited due to logistical and methodological constraints. In particular, measuring responses of communities with complex, three-dimensional structures is challenging, as many biogeochemical variables are affected by the volume and surface area of the organisms, with no tools available to quantify, non-intrusively, these variables in situ. Thus, it is crucial to develop a tool to accurately assess biogeochemical processes central to the ecological success of structurally complex benthic communities in situ. We have demonstrated that reliable and reproducible metabolic data can be generated using a combination of freely accessible photogrammetry tools and low-cost, easily deployable incubations chambers.

| Advantages of the new in situ chamber incubation method
A major advantage of the here-described protocol is the ability to measure metabolic and biogeochemical processes in various benthic habitats, with both simple and complex structures. In the existing literature, a multitude of in situ incubation chambers are presented.
In addition, most chambers are purpose-built for one specific study and community type, limiting the ability to compare ecosystem functions across a wide range of benthic habitats.
Both the laboratory and field assays demonstrated that the herepresented system could be deployed successfully on sediments, seagrass meadows, rocky habitats and coral assemblages-some of the most conspicuous benthic shallow-water habitats (Diaz, Solan, & Valente, 2004). Thereby, a combination of hardware features and the post-processing of data renders the approach extremely versatile. The use of PMMA cylinders offers multiple advantages, such as that they are commercially available in large sizes and at low costs.
Particularly hard-bottom communities (e.g. branching coral colonies) require chambers that are large enough to be mounted over rigid features (e.g. coral branches). The size of 50 cm diameter and 40 cm height has, thereby, proven more versatile over systems using small chambers that cover less surface area (e.g. Migné, Trigui, Davoult, & Desroy, 2018;Silva et al., 2005), or dome-shaped chambers (e.g. Smith, Böhlke, Repert, & Hart, 2009) that are restricted in their height in the peripheries. The removable lid maintains the good manoeuvrability underwater that, usually, goes along with smaller devices only (but see Haas et al., 2013). As the chambers are constructed entirely from PMMA, incubated communities are not subjected to shading by opaque materials (e.g. as used for frames in Haas et al., 2013 for stabilization), and the light transmission (90%-92% within the PAR spectrum) exceeds values reported for flexible systems made from polyvinyl chloride (Camp et al., 2015). Thereby, the risks to underestimate processes related to photosynthesis are reduced.
In addition, PMMA is durable and can be acid-washed, as needed for precise measurements of DOC and related compounds. The large, rigid and level surfaces of the chambers allow easy customization, permitting multiple sensors and additional device or sampling port attachments that can promote to use the system for numerous experimental strategies. As such, the dual two-way valve mechanism supports additions and removals of small and large amounts of water to be made to and from the chambers without disturbance of the incubation.
A major consideration when deploying benthic chambers is the mixing of the enclosed water. Despite its importance, water circulation systems are often disregarded as the individual power supply and a watertight encapsulation is deemed challenging. We present custommade, rechargeable pumps that ensure constant mixing within the chambers. The water movement simulates as closely as possible the hydrodynamic conditions of the undisturbed benthic substrate and ensures that water phase measurements reflect the average conditions inside the chamber. The correct alignment of the pump's outflow and the adjustable flow control creates a unidirectional current flow in round chambers, resulting in a radial pressure gradient that simulates sediment percolation as generated by natural boundary flowtopography interactions (Glud, Forster, & Huettel, 1996;Huettel & Rusch, 2000;Rasheed, Wild, Franke, & Huettel, 2004).
Besides the hardware, one of the most important features of the here-presented approach is the use of cloud-based photogrammetry tools. The freely available software enables, for the first time, non-intrusive assessments of the composition, volume and surface area of benthic communities (Gutierrez-Heredia et al., 2016;Lavy et al., 2015). This step is essential to relate changes in the water chemistry to the benthic community of interest and to make results widely comparable across regions and ecosystems.
Lastly, an important advantage of the system is that all materials required to build the incubation chambers are commercially available online at relatively low cost (c. 250 USD per chamber-including pumps, excluding sensors) (Supporting Information Table S1). In combination with minimal construction efforts (i.e. simple drilling, gluing, etc.) and an available protocol, the system can be used with high replication and at high temporal and spatial resolution without the need for costly infrastructure during deployment and sampling. Thereby, the system can increase the sampling scale of in situ measurements and facilitates precise, standardized flux measurements across many different benthic habitats.

| Lessons learned from pilot deployments
The field deployment of the autonomous operating chambers provided in situ metabolic data from various benthic shallow-water habitats. The chambers performed consistently when deployed on soft-bottom substrates. On sediments and seagrass meadows, community-wide DO production rates were highly linear over time, and no significant water intrusions were detected. The large PVC skirts around the chambers maintained a good seal from the surrounding water on most substrates. Yet, the linearity of DO fluxes was slightly reduced as some water intrusions were detected on rocky substrates, despite that reliable DO production rates were calculated by local linear regressions. For future considerations, we recommend additional soft edges (e.g. thick closed-cell polystyrene foam) at the bottom of the chambers that, if compressed, fill irregularities in the substrate, forming a good seal between the rock and chamber (Tait & Schiel, 2010). DO production rates during field tests were not linear when the circulation pumps were either switched off (indicator for stratification within the chambers) or when chambers were not secured to the substrates with weights in an intertidal habitat with strong waves (indicator for water intrusions). These tests highlight that special attention should be paid to use and deploy the chambers correctly to achieve reliable measurements. All incubations had in common that the rate of O 2 production initially decelerates as the organisms equilibrate after handling and closing the chambers. Thereby, the use of the LoLinR package for local linear regressions offers the advantage to use nonlinear time-series data for calculating monotonic metabolic processes (Olito et al., 2017). We further highlight the importance of including continuous monitoring of variables (e.g. dissolved oxygen). Such measurements allow identifying seawater intrusions and improper mixing in the chambers, thereby inform on possible confounding factors for other non-continuously monitored variables. If sensors for continuous measurements cannot be used for a variable of interest, water samples should be taken after an initial equilibration period of 10-20 min.
We highlight the versatility of the chambers in a case study calculating community-wide photosynthesis and respiration budgets of different coral reef communities of the Central Red Sea. The standardized protocol enabled a direct comparison between structurally complex (i.e. coral-and algae-dominated reef communities) and more simple, planar (i.e. sediments) habitats. Thereby, oxygen fluxes of sediments were in the same order of magnitude to previous studies from the Red Sea (e.g. Wild et al., 2009) and elsewhere (e.g. Rasheed et al., 2004) that used small (5-6 L) stirred benthic chambers. A direct comparison to values derived from in situ incubations of complex habitats (i.e. coral-and algae-dominated communities) is not possible, as no comparable approach has been applied before. Consequently, we related resulting oxygen fluxes from our in situ approach to estimated community budgets from ex situ single-organism incubations from the same reef. While in the same order of magnitude, an apparent trend was detected: Community budgets from ex situ single-organism incubations, as a conventional technique, tended to overestimate NPP, while community-wide R and GPP rates were underestimated by up to 90%. This tendency is also evident when comparing our findings to previous coral reef community budgets from single-organism incubations (e.g. Cardini et al., 2016;Naumann et al., 2013). Using this traditional ex situ approach, communities may be disturbed, and cryptic habitats, and particularly cracks and crevices within the reef matrix are not considered. Hence, studies of benthic fluxes in the natural environment result in more realistic estimates, since cryptic habitats encompass about 60%-75% of the total surface area of the reef (Richter & Wunsch, 1999). The organisms inhabiting the cryptic spaces (e.g. sponges, bryozoans and tunicates) are generally not included in ex situ budgets, however, they metabolize organic matter in the order of 15%-30% of the gross production of a coral reef (reviewed in de Jongh & Van Duyl, 2004;Maldonado, Ribes, & Duyl, 2012).
During all experiments, all system components proved extremely reliable. Importantly, the "plug and play" design of system enabled two divers to deploy up to eight chambers and to start the incubations within 45 min of dive time. An additional dive time of around 5-10 min will allow taking a sufficient number of pictures from each community for a later 3D model generation.
Before the final experimental application of chambers, the users should identify the optimal incubation time, and biomass to water volume ratio, to avoid anoxic or hypoxic conditions during incubations of communities of interest. Hence, due to the large size of the selected coral reef communities of the case study, incubations were already terminated after 2 hr. At this time, the O 2 saturation in the chambers reached values of 135% during light, and 35% during dark incubations.
The black PVC sleeves for each chamber provided a way to measure dark respiration rates during the day, that can be considered similar to respiration rates during the night (Bidwell, 1977). Sleeves are not needed for respiration measurements during the night.

| Limitations of benthic in situ incubations
Overall, no direct limitations regarding deployment, replication and reproducibility of in situ incubations were observed during our field testing. However, some general limitations with benthic chambers include: (a) Depending on the size of the community relative to the water volume, measurements can only be conducted for a few hours out of the day, making daily rates challenging to obtain. Longer incubations would require replacing the seawater periodically to maintain natural conditions (e.g. as in Tait & Schiel, 2010); (b) Although the chambers are among the largest currently available, size constraints still limit incubations to relatively small assemblages and results may not necessarily reflect all aspects of fully sized communities; (c) PMMA effectively removes most UV radiation, hence, the chambers possibly affect measured rates of photosynthesis for very shallow waters, where UV may inhibit some photosynthetic processes; (d) Some environments may be too energetic for using enclosures or do not enable to achieve a good isolation of the water. Additional

| Potential use in marine and aquatic research
Although this study focused on tropical benthic shallow-water communities, the method can easily be applied to any benthic community of interest that is accessible from land or within the limits of a scientific diving operation. As a descriptive tool at the community level, the chambers offer a new way to achieve reliable data of biogeochemical functioning in response to natural and anthropogenic changes, as synergistic and antagonistic interactions between organisms are not considered when organisms are incubated separately. We highlight the possibility to measure multiple parameters simultaneously (through sensors and water sampling) that can help linking responses that have often been assessed independently only, or where high species diversity and biotic interactions shape ecosystem functions.
Importantly, the presented approach can easily be used in experimental and manipulative strategies to quantify community functional responses to relevant global and local disturbances. While not pursued in the initial construction phase, the large rigid surfaces and the individual power supply of each chamber can support additional accessories, such as peristaltic dosing pumps, immersion heaters or pH flow feed-back control systems as applied previously (Kline, Kuntz, Breitbart, Knowlton, & Rohwer, 2006;Marker et al., 2010).
These developments should not be considered as a limitation to the existing approach but rather highlight the advantages of such an adaptable system.

| CON CLUS ION
The here-presented incubation chambers offer an easy and costeffective sampling solution that overcomes some challenges of conventional respirometry chambers. By allowing non-destructive, in situ metabolic and biogeochemical analysis of both simple and structurally complex benthic communities, the presented chambers offer a versatile platform for novel applications and experimental strategies. Measuring the effects of community interactions underlying benthic-pelagic biogeochemical processes may rapidly advance our understanding of ecosystem functioning and complement findings from aquaria and mesocosm studies. In doing so, the platform can transcend the boundaries of current research in multiple directions: (a) Providing a standardized method that can be applied across multiple habitats; (b) Measuring the response of whole communities, rather than from isolated organisms; (c) Overcoming the shortage of observational and manipulative studies carried out directly in the field, thus producing widely applicable and realistic findings; and (d) Enhancing our mechanistic understanding of how environmental drivers modulate particular processes across different habitats.

ACK N OWLED G EM ENTS
We are grateful to the personnel from the King Abdullah University of Science and Technology (KAUST) Coastal and Marine Resources Core Lab for logistical support. The authors thank Luis Ribeiro da Silva, Rodrigo Villalobos and João Cúrdia who helped during fieldwork. We thank Fatima Mamhud, Vincent Saderne and Ute Langner for the help with seagrass incubations. This work was supported by KAUST baseline funding to B.H.J.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.
wrote original draft of the manuscript. All authors read and approved the final manuscript.

DATA ACCE SS I B I LIT Y
The data used in this study are available at Dryad Digital Repository https://doi.org/10.5061/dryad.80c1rq2 .