Photo‐Reactivity of Surfactants in the Sea‐Surface Microlayer and Subsurface Water of the Tyne Estuary, UK

We report the first estimates of total surfactant photo‐reactivity in the sea‐surface microlayer (SML) and in subsurface water (SSW) (Tyne estuary, UK; salinity 0.3–32.0). In addition to temperature, a known driver of surfactant adsorption kinetics, we show that irradiation contributes independently to enhanced interfacial surfactant activity (SA), a notion supported by coincident CDOM photodegradation. We estimate a mean SA production via irradiation of 0.064 ± 0.062 mg l−1 T‐X‐100 equivalents h−1 in the SML and 0.031 ± 0.025 mg l−1 T‐X‐100 equivalents h−1 in the SSW. Using these data, we derive first‐order estimates of the potential suppression of the gas transfer velocity (kw) by photo‐derived surfactants ∼12.9%–22.2% in coastal North Sea water. Given the ubiquitous distribution of natural surfactants in the oceans, we contend that surfactant photochemistry could be a hitherto unrecognized additional driver of air‐sea gas exchange, with potential implications for global trace gas budgets and climate models.

• Irradiation results in increased surfactant activity in the sea-surface microlayer and in subsurface water in the Tyne estuary (UK) • Surfactant activity increased in parallel to photodegradation of chromophoric dissolved organic matter • Insolation driven increases in sea-surface microlayer surfactant activity may have global implications for air-sea trace gas exchange

Supporting Information:
Supporting Information may be found in the online version of this article.
of trace gases (e.g., CO 2 , CH 4 , N 2 O, and DMS) is of global significance (Cunliffe et al., 2013;Engel et al., 2017;Frka et al., 2009;Wurl et al., 2011) and critical to global climate change research (Donelan & Wanninkhof, 2002). Hence, our hypothesis is that insolation modifies SML SA, which will likely modify k w . We therefore irradiated (solar simulator) samples from the River Tyne estuary (UK), generating the first direct evidence for photochemical changes in SML SA. We compared these data with simultaneous changes in subsurface water (SSW) SA, and with corresponding changes in spectral CDOM characteristics, to evaluate the potential for air-sea gas exchange control by photochemically derived surfactants in the SML.
All statistical procedures used SPSS. Data were screened for normality (Shapiro-Wilk tests), and where appropriate, comparison of means were assessed using Independent t-tests (d = Cohen's d), and correlations assessed using Kendall's Tau-b correlation coefficient (τ b ; strong monotonous correlation: −0.5 ≥ τ b ≥ 0.5); significance at p < 0.05.
SA changes during irradiations indicated both photochemical and temperature effects in the SML and in SSW ( Figure 2). For 64 of 67 time-points SA IS exceeded SA DC , for all time-points (39) SA IS exceeded SA TC , and for 38 of 39 time-points SA DC exceeded SA TC . The data thus confirm a photochemical SA source in the Tyne estuary. The largest changes in SA IS consistently occurred during the initial 2 hr of irradiation and changes in both SA IS and SA DC were generally greater in the SML (SA IS : 0.236 ± 0.108 mg L −1 T-X-100 eq.; SA DC : 0.108 ± 0.100 mg L −1 T-X-100 eq.) than in SSW (SA IS : 0.130 ± 0.042 mg L −1 T-X-100 eq.; SA DC : 0.068 ± 0.062 mg L −1 T-X-100 eq.). Overall SA IS increase during the initial 2 hr was significantly greater in the SML than SSW (t (8) = 2.374, p = 0.045, d = 1.242), while no significant difference was found for that of SA DC (t (11) = 0.854, p = 0.411, d = 0.475). In general, SA IS increased over 24 hr in both SML and SSW (Figures 2a,  2b, 2e, 2f, 2h, 2j-2n), although some experiments showed overall decreases (Figures 2c and 2d) or no discernible change (Figure 2g and 2i). Comparison of initial SA in unfiltered and 0.22 μm filtered SML subsamples from TE1 (Figure 2g and 2h) indicated a significant particle contribution (40%). Importantly, SA IS increased in both subsamples during irradiation and remained higher than both SA DC and SA TC , consistent with photochemical SA production.
As our experimental design precluded SML interaction with SSW or air, the variable changes in SA we observed (Figure 2) must reflect a dynamic balance between production and removal. To clarify the overall extent of SA change we subsequently consider only those production rates due to irradiation (SA irr ) and temperature (SA temp ) estimated over 0-2 hr (Table 1), the interval for which the greatest SA changes were consistently observed. These estimates are reasonable for our study area, for which total daylight ranged from ∼7.5 hr (2 December 2016) to ∼17.3 hr (27 June 2016) (Supporting Information S1). Although we also calculated SA irr and SA temp over 0-24 hr (Table 1), these do not represent conditions in situ.
Mean SA irr (Table 1) was greater in the unfiltered SML than unfiltered SSW (0.064 ± 0.062 vs. 0.031 ± 0.027 mg L −1 T-X-100 eq. h −1 respectively), whereas mean SA temp was greater in SSW than in the SML (0.056 ± 0.031 vs. 0.024 ± 0.054 mg L −1 T-X-100 eq. h −1 respectively). A strong correlation between CDOM a 300 and SA in initial (T 0 ) samples (τ b (11) = 0.745, p = 0.001; Supporting Information S1), corroborates previous SA and CDOM data from estuaries and the open ocean, where SA and CDOM negatively correlate with salinity (e.g., Pereira et al., 2016Pereira et al., , 2018Uher et al., 2001). In many estuaries photochemical SA production could be masked by strong lateral SA gradients from the mixing of high SA river water with low SA coastal water (Pereira et al., 2016).
CDOM photodegradation (SML and SSW) coincided with SA photoproduction across the salinity range sampled (0.3-32.0). CDOM is an important seawater surfactant component (e.g., Tilstone et al., 2010) whose photodegradation in coastal and oceanic waters is widely documented (Mopper et al., 2014). Eight of 12 irradiations where CDOM was quantified showed significant positive correlations between SA and S 275−295 (τ b (10-15) = 0.529-0.740, p = 0.003-0.027), implying increased SA during irradiation to be consistent with decreasing CDOM molecular weight. We therefore contend that relatively low molecular weight surfactants are a likely by-product of CDOM photodegradation in marine waters.
We found moderately strong positive correlations between SA irr and initial S 350−400 (p = 0.018, τ b = (11) 0.561, Supporting Information S1) but not for SA temp , or for a 300 , S 275−295 or S R at T 0 (p = 0.176-1.000 and τ b (7-11) = −0.50-0.429 for all; data not shown). This suggests that the initial chemical composition and hence reactivity of the CDOM pool, rather than CDOM abundance, impacts rates of SA production during irradiation.

Discussion and Implications
We have shown the first evidence of coincident SA photoproduction and CDOM photodegradation in marine (estuarine) waters, although photoreactions implicating specific components of the marine surfactant pool are well established (e.g., Grzybowski, 2009;Kieber et al., 1997;Ortega-Retuerta et al., 2009). Our irradiations showed typical CDOM photobleaching reflected in decreasing a 300 and increasing S R with time, indicative of decreases inDOM molecular weight.
Our irradiation data inevitably include a temperature related component due to warming that could cause increases in microbial production (e.g., Kurata et al., 2016) or the interfacial adsorption of surfactants due to entropic effects in the hydration shell (e.g., Gosálvez et al., 2009;Mohajeri & Dehghan Noudeh, 2012;Southall et al., 2002;Tielrooij et al., 2010), or an aggregate of both. At higher temperatures, the hydrogen bond network in the hydration shell is more dynamic (Tielrooij et al., 2010). Hence, an increase in temperature increases hydration shell entropy by breaking hydrogen bonds (Southall et al., 2002). Consequently, the size of the hydration shell diminishes, and surfactant adsorption density increases (Gosálvez et al., 2009). We contend that changes in surfactant adsorption behavior are a plausible driver of temperature-related SA changes because the SA temp data showed no concomitant changes in CDOM a 300 or S R . Nonetheless, microbial processing, adsorption and photodissolution cannot be excluded in these unfiltered water samples. Changes in CDOM spectral characteristics may be used to diagnose CDOM processing: increasing S 275−295 and S R and decreasing S 350−400 , indicate photobleaching, while opposite trends indicate microbial alteration (Helms et al., 2008). However, S 350−400 changes during irradiations were negligible between sample treatments, suggesting that microbial activity followed the same trend in each. Note. SA production due to irradiation (SA irr ) is the difference in SA (mg l −1 T-X-100 eq.) between irradiated samples and dark controls (DC) at each timepoint (divided by the appropriate time) and SA production due to temperature (SA temp ) is the corresponding difference between DC and temperature controls. a Mean and one standard deviation calculated using all available data. b Mean and one standard deviation calculated using only experimental data where temperature controls were included as a sample treatment.

Table 1 Surfactant Activity (SA) Production Rates (mg l −1 T-X-100 eq. h −1 ) Estimated Over 0-2 hr and 0-24 hr of Irradiation for All Tyne Estuary Samples
A noteworthy feature was that irradiation per se was an independent driver of SA production, where SA irr in the unfiltered SML (0.064 ± 0.062 mg L −1 T-X-100 equivalents h −1 ) generally exceeded that in unfiltered SSW (0.031 ± 0.027 mg L −1 T-X-100 equivalents h −1 ). Overall SML enrichments in relatively labile DOM compounds are an established feature of coastal systems (e.g., Galgani & Engel, 2016); these compounds transfer to the SML via bubble scavenging (Hardy, 1982;Robinson et al., 2019) and can be produced in situ by microbial processing. Our data support the notion of SA photoproduction, either via the formation of new surface-active substances, or by photochemical transformations of existing surfactants allowing adsorption to the air-sea interface in greater numbers. CDOM photodegradation in parallel with SA photo-production strongly supports this concept.
Our data imply potential contributions of SML photochemistry to k w suppression by surfactants (e.g., Brockmann et al., 1982;Calleja et al., 2009;Frew et al., 1990;Mustaffa et al., 2020;Pereira et al., 2016Pereira et al., , 2018Ribas-Ribas, Helleis, et al., 2018;Salter et al., 2011) and to marine boundary layer aerosol and trace gas photochemistry Bernard et al., 2016;Brüggemann et al., 2017;Ciuraru et al., 2015aCiuraru et al., , 2015bClifford et al., 2008;Fu et al., 2015;Reeser et al., 2009;Rossignol et al., 2016) that demand further scrutiny. Pereira et al. (2018) applied a positive relationship between sea surface temperature (SST) and k w suppression at the ocean basin scale, implicating daily insolation as a driver of surfactant production via primary productivity. Our results indicate that irradiation of the SML is a likely important independent driver of SA production in addition to skin layer temperature, and consequently is an important independent control on k w .
It is instructive to estimate the potential scale of such control, by re-examining k 660 (k w for CO 2 in seawater at 20°C) estimates for the coastal North Sea (B1-B5; Figure 1), made by Pereira et al. (2016) in a gas exchange tank, that showed strong inverse relationships with SA. We applied these to our T 0 irradiation data assuming them to represent in situ SA (Table 2). This resulted in k 660 values of 0.6-13.4 cm hr −1 spanning TE1-TE4 (salinity 0.3-32.0) that are typical of other coastal sites (e.g., Kremer et al., 2003;Ribas-Ribas, Kilcher, & Wurl, 2018). These are toward the lower range found to be mediated in situ by surfactants in oceanic regimes (e.g., Calleja et al., 2009;Mustaffa et al., 2020), notwithstanding any difference in CDOM and surfactant properties.
Due to the proximity of our samples to those of Pereira et al. (2016) (Figure 1), differences in organic composition between them, even when accounting for potential temporal variability, are likely to be smaller than contrasts with other geographical regions, and we note that SML surfactant photochemistry is yet to be explored at different insolation intensities, in either oceanic waters or indeed in freshwater systems. Given that SML surfactant pool composition is likely to be important in addition to SA in controlling the magnitude of k w (Pereira et al., 2016), regional to global differences in the composition of the SML surfactant pool and the attendant temporal variability will likely be reflected in a variable photochemical contribution to k w control that demands further scrutiny.

Conclusions
Adequate parameterization of the factors controlling air-sea gas exchange is a long-standing scientific goal deemed essential to predicting global climate change. An increasing scientific focus is now on SML surfactant control of k w (e.g., Brockmann et al., 1982;Frew et al., 1990 Salter et al., 2011). Temperature is a known control of surfactant adsorption kinetics, but we have shown irradiation to be an additional, independent driver, in parallel with CDOM photodegradation. We contend that photoinduced increases in SA will likely impede k w at the global scale, with implications for the global budgets of climate-active gases. Consequently, studies of surfactant photoreactivity in a range of estuarine, coastal, and oceanic waters will be important, specifically those that examine how differences in total surfactant pool composition might differentially affect photochemistry and hence k w .

Data Availability Statement
Supporting data are available in the in-text data citation references: Rickard et al. (2021), findable and accessible via the Newcastle University data repository http://doi.org/10.25405/data.ncl.17006176; Copernicus Atmosphere Monitoring Service (CAMS) (2021), generated using CAMS information accessible via ads.atmosphere.copernicus.eu/cdsapp#!/dataset/cams-solar-radiation-timeseries?tab=form, using dates and locations listed in Table S1 in Supporting Information S1 for both cloud-free and actual weather conditions in 1-hr time steps. Neither the European Commission nor European Centre for Medium-Range Weather Forecasts (ECMWF) is responsible for any use that may be made of the CAMS information or data it contains.