Latitudinal Gradient of UV Attenuation Along the Highly Transparent Red Sea Basin

Abstract The tropical and subtropical oceans experience intense incident ultraviolet radiation (280–400 nm) while their water columns are thought to be highly transparent. This combination represents a high potential for harmful effects on organisms, yet only few reports on the UV penetration properties of oligotrophic tropical waters exist. Here, we present the pattern of UV attenuation over a wide latitudinal range of the oligotrophic Red Sea. We recorded spectroradiometer profiles of PAR and UV, together with chlorophyll‐a (Chl‐a) and light absorption by chromophoric dissolved organic matter (CDOM) to determine the contribution of phytoplankton and CDOM toward UV attenuation. Transparency to UV exhibited a distinct latitudinal gradient, with the lowest and highest diffuse attenuation coefficients at 313 nm (K d (313)) of 0.130 m−1 and 0.357 m−1 observed at the northern coast off Duba, and in the south close to the Farasan islands, respectively. Phytoplankton and CDOM both modulated UV attenuation, but CDOM was found to be the key driver despite the lack of riverine inputs. We confirm that ultraviolet radiation can reach deeper into the Red Sea than previously described, which means its potential to act as a stressor and selective driver for Red Sea organisms may have been underestimated to date.


INTRODUCTION
The intensity of ultraviolet (UV) radiation reaching the ocean surface is dependent on various factors, including latitude, season and atmospheric conditions like cloud cover, dust and tropospheric ozone (1). Upon reaching the water surface, the intensity and spectral characteristics of UV in the water column are determined by the incident irradiance, the conditions of the ambient light field (e.g. solar angle and wave action) and the concentration of UV-absorbing and UV-scattering substances in the water (2).
Two of the main components that modulate UV attenuation in the water column are phytoplankton and chromophoric dissolved organic matter (CDOM) (3). In the Red Sea, both components are generally found at very low concentrations. Due to the arid climatic conditions, no major river catchments flow into the Red Sea; hence, only minor amounts of nutrient-rich terrigenous material are added via watercourses (4). Consequently, the Red Sea is one of the most oligotrophic seas in the world with a very low phytoplankton biomass (<0.5 mg Chl-a m À3 ) (5-8) that is mostly dominated by picophytoplankton (9)(10)(11). The only exception to this is the southern Red Sea, where phytoplanktonic biomass can reach up to 4 mg Chl-a m À3 during blooms, and microphytoplankton such as diatoms may account for more than half of the total phytoplankton pool in terms of biomass (6,12). In the northern Red Sea, an intense and stable water stratification occurs during the summer that inhibits the replenishment of nutrients in the surface layers, resulting in chlorophyll-a (Chl-a) concentrations as low as 0.05 mg Chl-a m À3 (6).
Chromophoric dissolved organic matter (CDOM) is a chemically complex constituent of the total dissolved organic matter (DOM) that absorbs in the visible spectrum and represents the primary attenuator of UV radiation (13,14), thereby protecting organisms in the upper euphotic zone from harmful UV (15). The chemical nature and dynamics of CDOM are highly complex and, for the most part, not well understood (16)(17)(18). However, aromatic amino acids (19), lignin phenols, and humic and fulvic acids have been identified as contributors to the CDOM pool (20,21). Close to shore, the bulk of the CDOM pool is often allochthonous in nature and originates from terrestrial sources, rivers and coastal ecosystems such as salt marshes, seagrass meadows and mangrove forests (22)(23)(24)(25)(26)(27)(28)(29). In the open ocean, however, the primary source of CDOM is autochthonous material, which is generated by phytoplankton and heterotrophic microbial processes (13,(30)(31)(32).
Several studies have described the contribution of CDOM toward UV attenuation in the water columns of seas and oceans around the world (33)(34)(35)(36)(37)(38). However, the few existing studies on absorption by CDOM performed in the Red Sea mostly report on the role of DOM in attenuating the wavelengths of the photosynthetically active radiation (PAR, 400-700 nm) (39) or UV-A (320-400 nm) spectra (40). These studies agree that the concentration of CDOM in the Red Sea is relatively low in a global context, due to negligible inputs from fluvial and other allochthonous land sources as well as strong photobleaching.
The parameter commonly used to define water transparency is the downwelling diffuse attenuation coefficient (K d (k)), which describes the fractional rate of decay of downwelling irradiance with depth (3). Values of K d range widely across several orders of magnitude, from highly transparent oceanic waters to turbid coastal waters. For example, K d (305 nm) values as low as 0.08 m À1 have been reported for the South Pacific Gyre, which contains some of the most optically transparent water to UV-B globally (41,42), while Kuwahara et al. (43) calculated K d (305 nm) as high as 2.08 m À1 for coastal waters around the Malaysian peninsula. Other high K d (305 nm) values have been measured in coastal waters and lagoons around the world, including southern Spain (0.58 m À1 ), Japan (1.25 m À1 ), the Great Barrier Reef (0.65 m À1 ) and the Gulf of Mexico (0.77 m À1 ) (44)(45)(46)(47).
Though the Red Sea is not oceanic by definition, it shares several characteristics with open ocean waters such as consistently low concentrations of Chl-a and only minor amounts of CDOM in surface waters. Consequently, the Red Sea is thought to be among the optically transparent bodies of water worldwide (3,8,48). While most of the existing bio-optical studies investigated the attenuation coefficient of the Red Sea in relation to PAR (8,11,39), similarly high transparency has been reported for the UV spectrum. A comprehensive study by Dishon et al. (49) found that the 1% attenuation depth of UV-B (300-320 nm) varied between 19 m (K d (UV-B):~0.24 m À1 ) and 29 m (K d (UV-B):~0.16 m À1 ) in the Gulf of Aqaba, with the highest and lowest UV-B transparencies measured in summer (Aug-Sep) and spring (Mar-Apr), respectively.
However, studies reporting the optical properties (particularly UV transparency) of the Red Sea are scarce and have been predominantly carried out in the far north and the Gulf of Aqaba (8,(49)(50)(51). Hence, there is a great need to study if and how the UV-A and UV-B attenuation properties vary, particularly in the central and southern Red Sea, since those areas experience more intense incident UV and higher sea surface temperatures (SSTs) than the Gulf of Aqaba (52,53). High SST can exacerbate the damaging effects of UV on marine biota, including key species such as scleractinian corals and other reef organisms (54)(55)(56)(57)(58), which are of particular importance to the Red Sea (59,60).
The goal of this study was to quantify geographical UV attenuation variability in the Red Sea, extending existing data of UV-A and UV-B attenuation to cover a wider latitudinal range of the basin. A further aim was to determine the extent to which UV attenuation is modulated by Chl-a and absorption by CDOM (a CDOM ). The present study provides a comprehensive overview of the geographical variability in UV attenuation (UV-A and UV-B), which can be used to determine the biologically effective optical depth and allows comparing the UV properties of the Red Sea with those of other water bodies around the world.

MATERIALS AND METHODS
Study sites and spectroradiometer measurements. A total of 22 spectroradiometer profiles were recorded across the Red Sea during the Threats and Dust research cruises on board R/V Thuwal in October and November/December 2016, respectively, and as part of two Center Competitive Fund (CCF) cruises in August 2017 (CCF Summer cruise) and March 2018 (CCF Winter cruise) on board R/V Al Azizi (Table 1).
In 2016, downwelling irradiance (E d ) depth profiles at three monochromatic wavebands in the UV-B spectral domain (305, 313, 320 nm), as well as the integrated PAR spectrum (400-700 nm), were recorded with a BIC Compact Profiling Radiometer (Biospherical Instruments, San Diego, USA). For the cruises in 2017 and 2018, a PUV-2500 Profiling UV Radiometer (Biospherical Instruments, San Diego, USA) was used to measure downwelling irradiance (E d ) for individual UV-B (305, 313, 320 nm) and UV-A (340, 380, 395 nm) wavelengths as well as the integrated PAR spectrum (400-700 nm). Before each measurement, an instrument dark correction was performed for the depth and optical channels. The boat was maneuvered relative to the sun to insure the hull did not cast a shadow over the instrument, and the radiometer was deployed over a pulley~3 m astern using an A-frame at a speed of approximately 0.4 m s À1 . Because the radiometer was equipped with a pressure sensor, irradiances were recorded continuously during the profiles, generating five data points per second (5 Hz). All profiles were recorded around midday (11 am-1 pm).
Chlorophyll-a and CDOM analysis. As part of the Threats, Dust and CCF Winter research cruises, seawater samples were collected from 5 m depth in Niskin bottles attached to a CTD rosette system for quantitative analysis of Chl-a and CDOM. To measure the Chl-a concentration, 300 mL of seawater was filtered through a Whatman GF/F (0.7 lm); the filter was transferred into a 15 mL centrifuge tube and stored at À20°C until further processing. To extract Chl-a, 7 mL of 90% acetone was added to each 15 mL tube, and the samples were kept in the dark at 4°C. After 24 h, the samples were centrifuged (3000 rpm) at 18°C for 15 min. Chl-a in the extract was analyzed using the nonacidification method on a Trilogy Fluorometer equipped with a Chl NA module (Turner Designs, San Jose, USA).
For the CDOM analysis, 300 mL of seawater was collected from 2-5 m depth using a Niskin bottle and stored in amber bottles in darkness at 4°C until further processing. Within a few hours of collection, the water samples were filtered through a 0.22 lm polycarbonate filter prerinsed with Milli-Q water. Using a peristaltic pump (flow ratẽ 1 mL min À1 ), the samples were injected into a 2.5 m liquid waveguide capillary cell (LWCC) (61) coupled with a miniature fiber-optic spectrometer (USB2000 + , preconfigured to 200-850 nm; Ocean Optics Inc.) and a dual lamp (tungsten and deuterium) light source (DH-2000; Ocean Optics Inc.). Absorbance was determined for the wavelength range 250-750 nm at a resolution of~0.4 nm. The sample spectrum was compared against the spectrum of Milli-Q water after the refractive index differences were corrected by subtracting the mean value of OD CDOM (k) between 683 and 687 nm (OD null , CDOM ) from the whole spectrum, following Babin et al. (62). The CDOM absorption coefficients (a CDOM (k), m À1 ) were calculated using Eqn 1, where l is the optical path length (m) and 2.303 is the factor used to convert base e to base 10 logarithms. We report a CDOM for the individual wavelengths in the range 270-600 nm. Furthermore, we have determined the slopes in the regions 275-295 nm (S 275-295 ) and 350-400 nm (S 350-400 ), and the ratio of these slopes (S R = S 275-295 : S 350-400 ). Spectral slopes reported were calculated using linear regression of the log-transformed a CDOM spectra, following the method of Helms et al. (63). Downwelling diffuse attenuation coefficient (K d ). Initially, a visual quality check was performed on the spectroradiometer profile data to remove any erroneous irradiance values, such as obvious outliers. Additionally, for each spectral measurement at every depth, we approximated the instantaneous downwelling irradiance for the integrated UV-B (280-320 nm) and UV-A (320-400 nm) using trapezoidal integration. For the UV-B spectrum, the integration was performed using the instantaneous downwelling irradiance (E d ) of the 305, 313 and 320 nm channels, while for the UV-A spectrum, we used the E d of the 320, 340, 380 and 395 nm channels. The instrument detection limit for the most attenuated wavelength (i.e., 305 nm) was generally reached at a depth of 30-35 meters, so we calculated the spectral diffuse attenuation coefficient (K d (k)) for all UV channels (305, 313, 320, 340, 380 and 395 nm) down to 30 m depth and, for PAR, down to the deepest sampling depth of approximately 40 m. Irradiance versus depth profiles from the spectroradiometer casts were used to calculate K d (k) by performing a linear regression of the natural logarithm using Lambert-Beer's law, where E d (Z, k) is the downwelling irradiance at depth Z and E d (-0, k) is the irradiance just beneath the air-water interface. The slope of the linearegression informed about the value of K d (k). Generally, the fitting process was highly robust with correlation coefficients of Percent attenuation depth (Z % ) and mixed layer depth (MLD). For each UV wavelength (305, 313, 320, 340, 380 and 395 nm), as well as the integrated PAR spectrum (400-700 nm), we calculated the 10% and 1% attenuation depths (Z n% ), which are defined as the depths at which irradiance is reduced to n% of the value immediately below the water surface. The depths were calculated using the fraction of surface irradiance (E d (Z, k)/E d (-0, k)) and K d , as shown in Eqn 3.
The 10% and 1% attenuation depths are especially significant in terms of the PAR spectrum, as they are traditionally used to identify the midpoint and bottom of the euphotic zone, respectively (3,64). In the context of UV, Kuwahara et al. (43) defined the 10% penetration depth as the biologically effective optical depth and the 1% penetration depth as the implicit UV within the euphotic zone.
Additionally, the mixed layer depth (MLD), an index of the stability of the water column, was calculated from the CTD profiles as the shallowest depth at which the water density (r h ) differs from surface values by more than 0.05 kg m À3 (65). For stations where salinity data could not be recorded due to an instrument malfunction, we calculated the MLD using a threshold value in temperature (DT = 0.2°C), as described by de Boyer Montegut et al. (66).
Statistical analysis. Linear regression analyses were performed to explore the linear relationship between K d (k) and the two independent variables a CDOM (k) and Chl-a concentration. Data analyses and visualizations were performed using the software packages JMP Pro 13 (SAS Institute Inc., Cary, USA) and GraphPad Prism 7 (GraphPad Software, La Jolla, USA).   , with no distinct north-to-south or onshore-offshore gradient in water transparency. Across all cruises, however, we identified that the K d of UV-B wavelengths and PAR showed a negative and strongly significant linear relationship with latitude (P < 0.001).

Percent attenuation depth (Z % )
The depths to which 10% and 1% of the surface irradiance reached (i.e., Z 10% and Z 1% ) varied considerably depending on the wavelength, with the shortest UV wavelengths reaching the shallowest depths. For example, the average Z 10% values for the three UV-B wavelengths 305, 313 and 320 nm were 9.8, 12.1 and 13.9 m, respectively; for the UV-A wavelengths, the average  Across all the UV wavelengths, the highest a CDOM were measured close to the Farasan Islands (17°N) in late November 2016. The lowest a CDOM , on the other hand, were measured in early October 2016 off the coast of Yanbu (24°N) (Fig. 4).
The spectral slopes ranged from 0.019 to 0.046 nm À1 (S Chlorophyll-a  (Table 1). Surface Chl-a generally increased in concentration from south to north, exhibiting a latitudinal gradient (Fig. 5). Additionally, we could observe a clear seasonal pattern with the highest Chl-a concentrations being present in fall when the surface waters were deeply mixed (max. MLD: 72 m) ( Table 1). In contrast, minimal Chl-a levels were measured in summer when the upper layer of the water column was highly stratified and the MLD was found to be as shallow as 11 m.

Attenuators of UV
Based on data from three cruises (n = 14), we identified a positive linear relationship between K d (k) and the two parameters Chl-a concentration and a CDOM (Fig. 6). For each of the three UV-B wavelengths (305, 313 and 320 nm), we found that the slope of the regression line for a CDOM was greater than that of the Chl-a concentration. Furthermore, the slope of a CDOM (1.172, 1.128 and 1.157 for 305, 313 and 320 nm, respectively) increased with increasing wavelength, whereas the opposite pattern was identified for the slope of Chl-a (0.351, 0.269 and 0.245 for 305, 313 and 320 nm, respectively). For each of the UV-B wavelengths, the coefficient of determination (R 2 ) was higher between a CDOM and K d (k) than between the Chl-a concentration and K d (k) (Fig. 6). The strongest correlation (R 2 = 0.90, y = 1.157x + 0.020, P < 0.0001) was found between a CDOM (320 nm) and K d (320 nm), whereas the weakest correlation was identified between the concentration of Chl-a and K d (305 nm) (R 2 = 0.77, y = 0.351x + 0.167, P < 0.0001).

DISCUSSION
Our findings evidence that the degree to which ultraviolet radiation (UV) is attenuated in the Red Sea water column varies considerably across latitudes. The far northern Red Sea is the most optically transparent to UV, whereas the waters around the Farasan Islands in the south of the Red Sea basin strongly attenuate UV. Our results further indicate that the Red Sea might be more  optically transparent to both UV and PAR than previously described. For example, Dishon et al. (49) calculated a minimum K d for the integrated UV-B spectrum (300-320 nm) of 0.159 m À1 in the Gulf of Aqaba, which is located in the far north of the Red Sea. In comparison, we obtained an even lower value of 0.131 m À1 for the UV-B spectrum (280-320 nm) in the northern Red Sea off the coast of Duba (Fig. 2). Additionally, we identified a minimum attenuation coefficient for PAR (K d (PAR)) of 0.039 m À1 , which is lower than the K d (PAR) values reported for the Red Sea by both Dishon et al. The observed latitudinal pattern of UV attenuation is governed by the Red Sea's natural gradient in physical and biogeochemical properties. The northern Red Sea and the Gulf of Aqaba are both characterized by ultra-oligotrophic waters, with minimal concentrations of Chl-a and CDOM for the majority of the year (6,39,72,73). However, the Gulf of Aqaba becomes deeply mixed in spring, favoring phytoplankton growth, whereas the northern Red Sea is stratified during that time. This difference was reflected in our results by the attenuation coefficients we measured for UV-A, UV-B and PAR in the northern Red Sea, which were lower than those previously described for the Gulf of Aqaba (8,49). In the central and southern Red Sea, we found that UV attenuation gradually increased toward the south, where water turbidity and nutrient concentrations were highest, and Chl-a reached concentrations of 1-3 mg Chl-a m À3 (5,6,12,74) due to the influx of nutrient-rich water from the Gulf of Aden through the strait of Bab-el-Mandeb (74,75). While the southernmost area of the Red Sea was not sampled as part of the present study, we measured the strongest UV attenuation at our three southernmost sampling stations close to the Farasan Banks.
The main drivers determining UV attenuation are UV-absorbing and UV-scattering substances present in the water column. Of these, Chl-a, which is indicative of phytoplanktonic biomass, and CDOM are the most significant (3). Both of those water constituents exhibited a distinct seasonality in the Red Sea during our study. Furthermore, we measured the lowest attenuation for all UV wavelengths off Duba during summer when the water column was highly stratified (MLD: 22 m) and the concentration of Chl-a and the absorption by CDOM were lowest across the Red Sea basin. In contrast, the strongest UV attenuation was observed close to the Farasan Banks during winter and spring, coinciding with higher Chl-a and CDOM concentrations in the water column and intensified vertical mixing processes (MLD: 51 m).
We found that K d for the UV-B wavelengths (305, 313 and 320 nm) was significantly (P < 0.05) correlated with the Chl-a concentration and a CDOM . For example, K d increased more per unit of Chl-a for shorter UV wavelengths, based on the slope coefficients for 320 (y = 0.25x), 313 nm (y = 0.27x) and 305 nm (y = 0.35x). The coefficients of determination (R 2 ) between K d and the Chl-a concentration were generally high, but the highest R 2 was identified for 305 nm (R 2 = 0.77), confirming that Chl-a plays a larger role in the attenuation of short UV-B wavelengths (e.g. 305 nm) than those closer to the UV-A spectrum (i.e. 313 nm and 320 nm). In comparison, a study of the South Pacific Ocean and the Mediterranean Sea reported an average correlation coefficient between chlorophyll concentration and K d (320 nm) of R 2 = 0.694 (37). However, Dishon et al. (49) found an even stronger correlation (R 2 = 0.78) between the Chl-a concentration and light attenuation (i.e. K d (PAR)), which suggests that Chl-a predominantly contributes to the absorption of radiation in the PAR spectrum. In contrast, a study by Stambler (8) identified that Red Sea phytoplankton strongly absorb in the UV spectrum and that the absorption coefficients for UV wavelengths exceed the coefficient for 675 nm (the major absorption peak of Chl-a in the PAR spectrum). However, we note that Chl-a concentration is merely used as a proxy for phytoplankton biomass and that other UV-absorbing or UV-screening cell components besides Chl-a could be responsible for the strong UV attenuation by phytoplankton, such as mycosporine-like amino acids (MAAs) or scytonemin (76)(77)(78).
Regarding the overall contribution of phytoplankton and CDOM toward UV attenuation, we identified CDOM as the key modulator. Using the data displayed in Fig. 6, we calculated that a CDOM (k) contributed an estimated 73 AE 11.1%, 77 AE 10.2% and 77 AE 10.0% toward the downwelling K d values of the 305, 313 and 320 nm wavelengths, respectively. This finding is in agreement with other studies on natural marine waters that identified CDOM as the principal attenuator of UV (33,69,79), whereas the contribution of phytoplankton (<20%) and nonpigmented particles (<15%) toward the total absorption of UV wavelengths was reported to be comparatively low (21). Additionally, our values fall within the range reported for the southeastern Pacific Ocean by Tedetti et al. (80), who found that the relative contribution of CDOM to the total absorption of 305 nm in surface waters varied between 71% and 95% (compared to 73% in our study). In the present study, we only analyzed CDOM samples collected from 5 m depth, yet, it is known that the concentration and absorption of CDOM can vary considerably along the water column (31), including in the Red Sea (39), due to changes in the source material, microbial activity and photobleaching that often result in higher CDOM concentrations found below the upper mixed layer (22,81). A large proportion of our spectroradiometer profiles did not reach below the mixed layer, while for those that went beneath it, we did not notice an obvious change in K d at the boundary between the two layers, suggesting no marked change in the concentration of CDOM from 5 m to~40 m (our maximum sampling depth), possibly due to the absence of high amounts of source material. Large quantities of CDOM are usually transported from land into coastal waters by rivers (23,26,27,82), however, the Red Sea lacks riverine inputs (83,84), contributing toward its high transparency. Our finding that a CDOM and UV attenuation were strongest at the stations around the Farasan Islands could be due to CDOM sources from coastal marine vegetation, such as mangroves and seagrasses (29,85,86), which are abundant in this area characterized by islands, shallow sandbanks and reefs (60,74,87).
Previous studies suggested that both the northern Red Sea and the Gulf of Aqaba are Case-1 waters (8,88); that is, phytoplankton and their associated CDOM and detritus degradation products govern the optical properties (89,90). Our findings indicate that not just the northern Red Sea but rather the vast majority of the basin can be classified as Case-1 waters, since CDOM closely mirrors the optical properties of phytoplankton (see Figs 4 and 5) and the values of S 275-295 were larger than S 350-400 (i.e. S R > 1) (Table 1), which is typical for low molecular weight (LMW) CDOM in open ocean waters (63,91). This finding also suggests that the bulk of CDOM in the Red Sea is heavily photobleached and of marine rather than terrigenous origin (63,89,92). In fact, this closely coupled relationship between Chl-a and CDOM in the Red Sea, confirming the predominantly phytoplanktonic origin of CDOM, was recently described by Kheireddine et al. (39). One exception to the Case-1 classification could be the Farasan Islands in the far south of the Red Sea, an area previously categorized as Case-2 waters based on remote sensing data (93). At the Farasan Banks, however, we did not find an unusually low S R in this region to support this hypothesis. Therefore, a thorough study should be carried out to determine whether phytoplankton and its associated degradation products or benthic marine vegetation represent the primary source of CDOM and, therefore, the main driver of the high UV attenuation observed in this area.

CONCLUSION AND FUTURE OUTLOOK
Here, we confirm that the Red Sea is indeed highly transparent to both UV-A (maximum Z 10% (UV-A): 57 m) and UV-B (maximum Z 10% (UV-B): 18 m) and that its bio-optical properties are comparable to those of the clearest ocean waters. The high transparency of the Red Sea, in combination with the intense incident UV and minimal cloud cover (5) in this region, means that the biota in its surface waters is exposed to extreme doses of damaging UV. Therefore, UV is likely a key ecological factor governing the distribution, metabolism and survival of organisms in the upper euphotic zone of the Red Sea, as previously proposed for a wide variety of Red Sea taxa, including zooplankton (94), soft corals (95) and picophytoplankton (50,96). While the northern Red Sea was found to be more transparent to wavelengths of the UV spectrum than its southern counterpart, its higher latitude means that the incident solar radiation is less intense. Consequently, the maximum UV doses in the water column of the northern and southern Red Sea might be comparable in magnitude.
Ongoing and complex atmospheric and environmental changes (97) impose a high degree of uncertainty onto future projections regarding the intensity of incident UV and UV attenuation by Red Sea waters. For instance, current models predict that the total amount of stratospheric ozone over the tropics might decrease by the end of this century due to the emission of greenhouse gases and changes in the global circulation pattern of ozone (98,99). Similarly, while upper stratospheric ozone seems to show renewed growth, a recent study by Ball et al. (100) found that lower stratospheric ozone levels are being depleted, which may be linked to the persistent emission of ozone-reducing chlorofluorohydrocarbons (CFCs) despite the Montreal Protocol banning their use (101,102). Consequently, the reduction in stratospheric ozone could lead to elevated incident UV radiation in the Red Sea region over the next few decades. An increase in UV would also likely lead to intensified photobleaching of CDOM, which in turn would further increase the optical transparency of the water column (13,99). Additionally, SST of the Red Sea is increasing at a rate that exceeds the warming rate of the global ocean (53). This trend could affect the UV attenuation properties of the Red Sea; an increase in SST would result in stronger thermal stratification of the water column during summer and, potentially, deeper penetration of UV into the water column.
In summary, we found that UV attenuation in the Red Sea exhibits a distinct latitudinal gradient and that the basin contains some of the most optically transparent marine waters in the world. However, great uncertainty remains regarding future changes to UV attenuation and UV doses in the Red Sea as a result of imminent environmental change, and global warming and stratospheric ozone depletion in particular.