The fieldwork and BVOC-measurements were performed in May and August 2025, in Ireland and Finland, respectively. In Ireland, BVOC measurements were performed at Mace Head Atmospheric Research Station (MHD), that is situated on the west coast of Ireland on the shore of the North Atlantic Ocean (Fig. 1a). In Finland, measurements were conducted at Tvärminne Zoological Station (TZS) that is located on the southern tip of Finland next to the Baltic Sea (Fig. 1a). Due to the stations' excellent placement, all macrophytes were collected from the near vicinity of the respective stations.

The environment and local biology differ considerably between the studied marine regions. At Mace Head, salinity fluctuates around 30 psu, while in the brackish waters outside TZS salinity is quite stable around 5–6 psu. The Finnish coastal waters experience large annual temperature fluctuations, reaching > 20 °C in the summer and freezing conditions during the winter months, while on the Irish coast the seawater fluctuates less (6.5–18.5 °C at COMPASS Mace Head Buoy 2 km of the coast; data through https://erddap.marine.ie/, last access: 28 September 2026). Coincidentally, during both campaigns, the relevant sea areas were experiencing marine heatwaves, with seawater reaching over 20 °C in the shallows outside Mace Head in May (personal observation, 16 °C measured at Mace Head Buoy) and over 22 °C on the Finnish coast in August (Monitoring data from TZS, MONICOAST: http://www.helsinki.fi/monicoast, last access: 28 September 2026). The Baltic Sea is well known to be highly eutrophic, which has considerable effects on local biology e.g., high pelagic and filamentous algae production that affects benthic light availability. On the west coast of Ireland seawater is less nutrient-rich, translating to better visibility and lower biological activity in the water column. Tidal regime also differs significantly between regions. At Mace Head, low tide occurs twice a day (tidal difference during campaign ∼ 3 m), while the coastal areas of Finland are tideless, though some water-level fluctuations occur irregularly due to weather-related drivers (e.g. atmospheric pressure, wind).

At both study sites, the seawater used in experiments was pumped from a depth of approximately 5 m from the nearby sea (Ireland: Carna Research Station which is about 5 km from MHD; Finland: TZS). Prior to the experiments in Ireland, we evaluated whether background signal quality could be improved by filtering the seawater (5 µm pore filter) and autoclaving (121 °C for 20 min) the filtered seawater. Filtering the water improved the background signal, while autoclaving worsened the measurement quality by increasing certain BVOC emissions. Although the exact source of these increased emissions could not be identified, we assume that they originated either from the plastic canister in which the water was autoclaved or from microorganisms that burst due to the sterilisation process. Based on these findings, filtered but not autoclaved seawater was used at both study sites (5 µm filter in Ireland and 0.2 µm filter in Finland). In Finland, the finer filter was necessary to reach satisfactory background measurements due to high amounts of organic particles/microorganisms being present in the water. Additionally, before measurements, the filtered seawater was bubbled in 20 L containers for at least 12 h to further reduce BVOC concentrations in background measurements. This practice was added to the measurement protocol after test incubations revealed high DMS concentrations in the background measurements that did not stabilize during the incubation, thereby negatively affecting the quality of the results.

The three selected macrophyte species belong to distinct organismal groups. Bladderwrack (Fucus vesiculosus, hereafter Fucus) is a perennial brown macroalgae (Orchrophyta, Phaeophyceae) that usually grows attached to rocky substrates. Gut weed (Ulva intestinalis, hereafter Ulva) is a green macroalgae (Chlorophyta, Ulvophyceae), that unlike the two other species investigated has an annual lifecycle. Common eelgrass (Zostera marina, hereafter Zostera) is a perennial marine flowering plant (Zosteraceae, Angiospermae). It has roots that anchor it to soft-sedimentary seafloors where it can create dense underwater meadows.

Macrophytes were collected in the morning of emission measurements and kept in mesh bags submerged in the sea prior to incubations. During both campaigns, we collected and measured all replicates (n=5) per species during the same day and all three species during the same week. The different macrophyte species were measured on separate days, because it was not feasible to collect Zostera every day and we wanted to measure BVOC emissions from fresh specimens. Zostera was collected by snorkeling in Ireland (1 m depth during low tide) and by SCUBA in Finland (4 m depth). Instead of individual plants, Zostera replicates consisted of bundles of shoots (each shoot attached to an individual rhizome), averaging 22 shoots in Ireland and 15 in Finland. The shoot counts differed between countries, because plants were larger in Finland, and > 20 shoots would not have fit in the incubation flask. Each Ulva replicate consisted of algae thalli (fronds) collected from a specific rockpool. Hence, Ulva from 5 separate rockpools were measured at both study sites ensuring independent material. Replicates of Fucus consisted of individual intact thalli that were gently removed from the rocky substrate. The collection of Fucus replicates was semi-random, because the algae had to be small enough to fit the 5 L incubation-flask. At Mace Head, Fucus grows in the intertidal zone and falls dry twice a day, while Fucus in the northern Baltic Sea always remain submerged.

Before measuring their BVOC-emissions, the macrophytes were carefully cleaned and rinsed with filtered seawater to remove sediment and most visible epibiota. After cleaning, small amounts of filamentous algae remained attached to Zostera and Fucus replicates, but > 95 % of the biomass consisted of the target species. It was especially important to rinse Ulva-replicates, because during test incubations we noticed that unrinsed replicates released certain BVOCs at very high concentrations immediately when the algae were submerged in the flask, which was likely due to a priori accumulation of BVOCs on the surface of the algal matrix. This was unwanted as it obscured the measurement of real-time emission rates.

BVOCs were measured using a Vocus proton-transfer-reaction time-of-flight mass spectrometer (Vocus PTR-TOF, Tofwerk AG/Aerodyne Research, Inc.), a high-sensitivity instrument described by Krechmer et al. (2018). The instrument is designed to continuously quantify BVOC concentrations in the air but has rarely been used to measure BVOC concentrations in water. The system employs a low-pressure discharge reagent ion source to generate H₃O⁺ ions, which are then coupled to a Focusing Ion-Molecule Reactor (FIMR). The FIMR consists of a resistive glass tube with a radio frequency (RF) quadrupole field that collimates the ion beam, minimizing wall losses. For this study, the FIMR was operated at a reactor pressure of 2.5 mbar and an axial voltage gradient of 450 V, resulting in an E / N ratio (electric field to gas number density) of approximately 120 Td. The reactor temperature was maintained at 100 °C. Ions were detected by a time-of-flight mass spectrometer operating with a mass resolving power of 4500m/Δm at m/z 100 Th. Data were acquired at a time resolution of 6 s.

The ionization mechanism in the Vocus PTR-TOF-MS is proton transfer reaction, which is a relatively “soft” ionization technique that predominantly yields the protonated molecular ion (MH⁺) for most compounds of interest, though fragmentation becomes more common with larger and more functionalized molecules (Zhang et al., 2025; Li et al., 2022). The selectivity of the Vocus PTR-TOF-MS is primarily governed by the proton affinity (PA) of the target analytes relative to the reagent ions. Using H₃O⁺ as the reagent ion, the instrument selectively ionizes VOCs with a PA greater than that of water (691 kJ mol⁻¹), efficiently suppressing the detection of major atmospheric constituents such as N₂, CO, and others. The sensitivity of the Vocus PTR-TOF-MS is influenced by the proton affinity, fragmentation patterns, and transmission efficiencies of the analytes. Fragmentation can occur depending on the functional groups present and the energetic conditions within the reactor. In the Vocus FIMR, the collision energy is modulated by the E / N ratio. For this study, the reactor conditions were optimized to balance the suppression of water clusters and minimize collision-induced dissociation.

For any analyte molecule, knowledge of its proton transfer rate constant (kPTR) and the instrument's response to VOC standards under the applied operating conditions allows straightforward determination of the Vocus PTR-TOF-MS sensitivity for that molecule. During the campaign, the Vocus PTR-TOF-MS was calibrated weekly using a gas-phase standard (Apel-Riemer Environmental, Inc.), which contained 19 compounds. The sensitivity of these compounds can be found in Table S1 in the Supplement. Based on the calibration results of known compounds in the gas mixture, the transmission efficiency and the relationship between kPTR and sensitivity factors were determined, as shown in Figs. S1 and S2 in the Supplement. For detected species with known PA, the sensitivity factors can be calculated based on the established relationship between kPTR and sensitivity. For compounds with unknown PA, we used an averaged kPTR value of 2.5 × 10⁻⁹ cm³ molec.⁻¹ s⁻¹. The error introduced by this assumption is less than 50 %, assuming no significant fragmentation occurs during ionization.

Incubations were performed in a 5 L borosilicate glass filtration flask (Schott Duran), made airtight during incubations with custom-made Teflon corks. Before the start of each incubation, the airtightness of the flask and incubation setup was checked with a flowmeter. During the whole measurement period (30 min), the seawater was purged with clean air (produced by Vocus ZeroAir-generator) through a Teflon tube with a flow rate of 0.2 L min⁻¹ (Fig. 1b). The purging flow was pushed through a metallic bubbling stone that created small bubbles, increasing the area of the purged air and enhancing the effectiveness of capturing and transporting BVOCs into the headspace. From the headspace, the 0.2 L min⁻¹ flow containing purged BVOCs continued out from the flask through another Teflon-tube. Before entering the Vocus PTR-TOF-MS, the flow was mixed with a clean air dilution flow (0.2 L min⁻¹) to decrease the humidity and thereby risks for water condensation in the lines or in the filter (PTFE membrane filter, pore size 1.0 µm, Advantec MFS. Inc.) in front of the Vocus. Of the total 0.4 L min⁻¹ flow, 0.1 L min⁻¹ entered the PTR, while the rest was pumped out from the system or exhausted (Fig. 1b). The flows were controlled by mass flow controllers (Alicat Scientific, Inc. USA).

For the background measurements, the flask was filled with 4.5 L of pre-bubbled filtered seawater, leaving 1 L of headspace for BVOC measurements by PTR-TOF-MS (yielding a total of 5.5 L, as the neck volume is excluded from the nominal flask capacity). Immediately after finishing a background measurement, a macrophyte-replicate was added to the flask (without replacing the seawater) and its BVOC emissions were measured. The macrophytes were tied to Teflon sticks to keep them submerged. During all incubations, the flask was illuminated by two LED-lamps with light spectrums suitable for submerged plants (Fluval Plant LED). The light environment inside the flask corresponded to ∼ 200 PAR (Photosynthetically Active Radiation). The seawater temperature in all incubations was ∼ 20 °C, similar to ambient seawater temperatures during both campaigns (see Sect. 2.1). BVOC emissions from both background and macrophyte replicates were purged and measured continuously for 30 min. All replicates were measured in the same glass flask, which was rinsed twice with Milli-Q water between replicates and subsequently filled with fresh pre-bubbled filtered seawater. After measurements, the macrophyte-replicates were dried (60 °C, 48 h) to acquire their individual dry weight (hereafter DW).

Post-processing of Vocus PTR-TOF-MS data was performed using the data analysis package “Tofware” (version 4.0.2, http://www.tofwerk.com/tofware, last access: 28 September 2026) running in the Igor Pro (Wavemetrics, v.9.0, OR, USA) environment. High-resolution peak fitting was conducted to deconvolve overlapping isobaric ions, sharing the same nominal mass. The peak shape was determined empirically from isolated high-intensity ions and applied to fit the entire mass spectrum. Molecular formulas were assigned to fit peaks based on mass defects and isotopic pattern matching. BVOC emissions were identified by evaluating time-series data for all identified peaks (m/z 19–400 Th). A peak was classified as a macrophyte-derived BVOC and included in subsequent analyses, if a distinct change relative to the seawater background was observed in at least three out of the five replicates within an experimental group.

Despite filtering and pre-bubbling of seawater background measurements did not always properly stabilize during the incubation period, which also affected the quality of the subsequent measurement of macrophyte BVOC emissions. For instance, if the seawater contained high concentrations of a BVOC that continued to decrease during the entire background measurement, it could mask emissions from the macrophytes (BG1 and Sample1 in Fig. S3). Due to failed background measurements, two replicates (a Finnish Zostera- and Ulva-replicate) were removed from further analysis. Problems with background measurements also occurred in other replicates (mainly in Finland) for individual data points (e.g., first data point in Fig. S3) and if the results were deemed unreliable in the data preprocessing stage, the data points were classified as NA in the final data set.

For emission rate calculations (see Sect. 2.7), we only used data collected during the last 15 min of both background and macrophyte incubations. The emissions of most compounds reached a stable plateau in the first 15 min of the measurements (Figs. 1c, S3), but the concentration of some compounds kept increasing during the whole measurement period. The compounds that stabilized are assumed to be limited by the emission rate of the macrophyte, and their transfer from the water into the sampled air is rapid. This is the case for poorly soluble compounds, such as hydrocarbons. The compounds that kept increasing throughout the measurement are most likely the more water-soluble compounds (e.g., Br₂CH⁺ and DMS; Table S2) for which the concentrations in the sample air are more limited by water-to-air transfer. All data points (1 measurement/minute for each compound) from the last 15 min were averaged, after which the background averages were subtracted from the macrophyte averages to acquire cps-values (counts per second) for how much macrophytes were emitting each compound. The raw signal intensities (counts per second) were converted to mixing ratios (ppb) using reaction rate constants and the instrument-specific transmission curve as described in Sect. 2.4. Emission rates could then be accurately deduced for compounds whose concentrations stabilized during the incubations (see Sect. 2.7). For compounds that failed to reach a stable concentration, the emission rates could still be estimated, though these estimates are not considered accurate because they systematically underestimate the true emissions (all instances marked in Table S2). Despite underestimations, the comparisons between experimental groups for these compounds are valid due to the same sampling period across all replicates. Some BVOCs also clearly decreased following the addition of a macrophyte (Fig. 1c). We attribute this behaviour to uptake of BVOCs, but did not attempt to determine any uptake rates, as the observed decrease will depend on the initial concentrations available in the background water and possible balances between emission and uptake. All instances where BVOCs were taken up during macrophyte incubations are listed in the supplementary materials (Table S2).

The use of a Vocus PTR-TOF-MS allowed us to detect the emissions of a large range of compounds (C₁–C₁₅; Excel). However, when a Vocus PTR-TOF-MS is used as the sole analytical method the exact identity of the measured molecules cannot usually be identified, because the method is mass-selective, which means that isomers cannot be separately distinguished. Therefore, we used the GLOVOCS-database (Yáñez-Serrano et al., 2021) to assign potential identities to our measured compounds (Table S3). Most compounds have several potential identities and although the GLOVOCS database is extensive, many especially larger compounds were not found in the database. Certain well studied compounds could with good confidence be appointed their identity (e.g., C₂H₇S+= DMS).

Emission rates standardized to macrophyte mass (ERmass; ng g⁻¹ h⁻¹) for the emission-data collected with the Vocus PTR-TOF-MS were calculated using the following equations: 1ERmolmin-1=Cppb⋅Nppb⋅D⋅FflowNA2ERmass=ERmolmin-1⋅MmDW⋅th where ERmolmin-1 is the emission rate in moles min⁻¹, Cppb is the measured ppb-concentration of each emitted compound, Nppb is the number of molecules corresponding to 1 ppb (2.5×1010 molec. cm⁻³), D stands for the dilution (=2 in this study), Fflow is the purging flow (0.2 L min-1=200 cm³ min⁻¹), NA is the Avogadro constant, Mm is the molar mass of emitted compounds, DW is the dry weight of incubated macrophyte, th is for 60 min h⁻¹. To express emission rates in ng g⁻¹ h⁻¹, the calculated emission rates were multiplied by 10⁹ to convert from grams to nanograms.

All statistical analyses were performed in RStudio (R version 4.3.0; R Core Team, 2023). AI models were used to assist with writing parts of the R code, but AI was never allowed to work directly with the data. We used non-metric multidimensional scaling (NMDS) to visualize patterns and variability in BVOC profiles across replicates. As NMDS requires complete data, we removed compounds that contained NA-values from the ordination analysis. This resulted in the NMDS being performed on 117 compounds of the total 170 detected compounds (166 compounds showing emissions and 4 compounds that were only taken up in the dataset). Prior to analysis, BVOC variables were standardized (z-score transformation) to account for differences in scale and variance among compounds. The NMDS with Euclidean distance, was performed using the metaMDS function (vegan-package; Oksanen et al., 2022). Euclidean distance was specifically selected because the dataset included both positive (emission) and negative (uptake) values. To assess the relationships between individual BVOC compounds and the ordination structure, environmental vector fitting was performed using the envfit function (999 permutations; vegan package). Only variables showing significant correlations with the ordination (p ≤ 0.05) were retained for visualization.

To explore if BVOC composition differed significantly between groups, a permutational multivariate ANOVA (PERMANOVA, adonis2 function in vegan package) was performed on the reduced dataset used for NMDS. The PERMANOVA model (Euclidean distance, 999 permutations) included Macrophyte species (Zostera, Fucus, Ulva) and Region (Finland and Ireland) as fixed factors, and their interaction (Macrophyte ⋅ Region). Sequential (type I) sums of squares were used to evaluate the contribution of both factors and their interaction. Beforehand, a permutation test for homogeneity of multivariate dispersions (betadisper, vegan package) was used to confirm that the group dispersions were similar between groups. Afterwards, pairwise comparisons were performed to investigate which groups (Macrophyte ⋅ Region) emitted significantly different BVOC compositions from each other. Resulting p-values were adjusted for multiple comparisons using the Benjamini–Hochberg procedure to control for false discovery rate.

To examine whether experimental groups differed in their total BVOC emission rates and the rates of the most highly emitted individual compounds, we used generalized linear models (GLMs) with a Gamma distribution and log link. This statistical approach was chosen to appropriately account for right-skewed data and heteroscedasticity. Macrophyte species, Region, and their interaction were included as fixed effects in the models. Model significance was assessed using analysis of deviance (χ2 tests), and afterwards, Tukey-adjusted pairwise comparisons were conducted using estimated marginal means to explore significant differences between experimental groups.

Although BVOC emissions of the three macrophytes were more similar within-species than between species, substantial intraspecific variation was also observed. Within the groups (Macrophyte ⋅ Region), emission profiles were consistent across replicates (Fig. 3a), but significant differences were evident among marine regions for all three species. Emission rates of specific compounds showed intraspecific differences (see e.g., Fig. 4b and c; DMS and sesquiterpenes for Zostera), but many compounds were also emitted in only one of the studied regions (see Figs. 2b, 5 and S4–S6). Some differences in emission rates were expected (e.g., driven by differing seasons and temperatures), but the large differences in the identities of emitted compounds were more surprising (only ∼ 50 % of compounds shared within species for all three species between regions). Here, we cannot pinpoint the exact factors driving these differences within-species, but below we discuss the most likely contributing factors.

Macrophyte BVOC emissions respond to several abiotic factors such as temperature, light, seasonality and organic pollution (Saunier et al., 2025a and references within). Light and temperature are important triggers of BVOC emissions, and changes in these drivers can rapidly affect emissions (Broadgate et al., 2004; Zhao et al., 2023; Wang et al., 2025). Although light and temperature were kept comparable throughout all incubations and in situ seawater temperatures were broadly similar during the campaigns, differing seasons and the markedly different annual temperature regimes in the two regions likely contributed to the observed differences in emissions. Additionally, season-specific developmental processes likely also affect macrophyte BVOC emissions. For instance, in terrestrial plants, factors like flowering, leaf senescence, and dormancy affect BVOC emissions (Kesselmeier and Staudt, 1999), but how these processes affect BVOC emissions from macrophytes remains unknown. In addition, the growth environment likely also drives BVOC emissions by affecting macrophyte metabolism, physiology or stress (Bäck et al., 1992). For instance, the stark differences in salinity and nutrient availability between regions likely affected the emissions (as has been observed with phytoplankton; Zhao et al., 2023 and references within).

The macrophyte populations of Finland and Ireland have been separated for millennia, leaving populations ample time to adapt to their local environmental conditions (e.g., salinity, temperatures, tides, etc.). Therefore, within-species genetic diversity will differ significantly between regions, which in turn could affect the organisms BVOC emissions, e.g., through metabolic responses to local conditions. For example, permanent exposure to low salinity/high nutrients in the Baltic Sea has likely modified biochemical profiles, with direct influence on both emission profiles and rates. However, the effect of genetic diversity on BVOC emissions is completely unexplored for macrophytes, leaving the influence of population-specific genetics an important question to answer. The quantified BVOC emissions were likely also affected or directly released by the microbial communities associated with the macrophytes, as microbes are known to produce and consume many BVOCs (Lemfack et al., 2018). For instance, seagrass meadows host diverse microbial communities active in the many phases of DMS production (e.g., DMSP-demethylation, DMSP-cleaving and DMS-oxidization; Jonkers et al., 2000), and regional differences in these communities could potentially explain why BVOC emissions differed so greatly within the studied species. Additionally, the fact that we found region-specific compounds emitted by all three macrophytes, suggests that these were formed by site-specific circumstance such as local microbial communities.

The Baltic Sea is an outlier when compared to most marine regions globally, that despite very low salinities and high nutrient loads, harbours marine macrophytes. It is possible that if this study was performed at different locations along the eastern Atlantic coast, macrophyte BVOC emissions (especially compound identities) would have been more similar between sites than shown here, due to more similar environmental conditions. Regardless, when comparing our results with Zostera marina BVOC-data collected from the Mediterranean Sea (Coquin et al., 2024), it becomes apparent that large differences even occur between fully saline regions. Of the 51 BVOCs detected by Coquin et al. (2024) only 12 fit compounds emitted by the seagrass in this study (out of 126 in total). Differing methodologies between studies (GC-MS and destructive sampling in Coquin et al., 2024) partly explain this large difference, but overall, these findings demonstrate that relying on data from a single site or population is insufficient to accurately estimate macrophyte BVOC emission rates or characterize their emission profiles. Hence, future emission estimates need to be compiled from several locations with differing environmental conditions. Our understanding of the dynamics governing macrophyte BVOC emissions lags knowledge of emissions from terrestrial plants by decades (e.g., Peñuelas and Llusià, 2001, Kesselmeier and Staudt, 1999) and investigating these dynamics will be crucial so that large scale emission estimations can eventually be established.

Terrestrial deforestation has been shown to reduce BVOC emissions, with direct effects on atmospheric processes and local climate (Tripathi et al., 2025), and the large losses of forested areas during the last centuries have led to significant reductions in atmospheric BVOCs and SOAs (Vella et al., 2025). During the last century global macrophyte cover has also reduced greatly (e.g., 40 %–60 % decline of kelp forests and ∼ 19 % of seagrass; Wernberg et al., 2019; Dunic et al., 2021) and the negative trend is expected to continue (Manca et al., 2024). Such losses have already impacted rates of marine BVOC emissions to some degree and will continuously do so in the future, with unknown climate impacts as a result. However, this particular ecosystem service has largely remained unexplored for macrophytes. Similarly to the Blue Carbon boom of the last decade (Macreadie et al., 2021), if macrophyte BVOC emissions and their climate impact can be estimated, the information would bring much needed attention and value to these threatened ecosystems, which in turn could be used to aid their conservation.

In this study the seagrass Zostera showed the highest standardized total emission rates in both marine regions and very similar rates between regions. Globally, algae (both micro- and macro-) will still be larger BVOC-sources than seagrasses due to their substantially larger spatial coverage and biomass. Yet, our results (together with Saunier et al., 2025b) show that seagrasses emit many atmospherically relevant BVOCs at notable rates. For the seagrass genus Zostera, a global mean biomass dry weight per area has recently been estimated (226 g DW m⁻²; Foster et al., 2025), and by using this value, the total BVOC emission rates calculated in this study (Fig. 4a) translate to 12.0 and 12.3 µg BVOCs m⁻² h⁻¹ in Finland and Ireland, respectively. At present, the low amount of data and high unexplained variability in emissions make it unreliable to upscale these estimations even further (e.g., to global emissions with seagrass coverage data). Large-scale emission estimates are not only lacking for seagrasses, but all macrophytes and, in the future, more research efforts should be made to acquire quantitative and qualitative macrophyte emission data so that macrophytes emissions can eventually be added to marine BVOC budgets.