Ecotoxicological Evaluation of Sunscreens on Marine Plankton

In recent years, a large number of sunscreens have emerged to protect our skin. Most of them are made up of simple or compound aromatic structures, which can pose a threat to marine ecosystems. In order to understand their effects on the marine environment, different ecotoxicological bioassays were carried out using planktonic organisms from three phyla and two different trophic levels: larvae of the sea urchin Paracentrotus lividus, the copepod Acartia tonsa, and the microalga Tisochrysis lutea. The aim of these tests was to expose these organisms to leachates from eight sunscreen formulations. All of them showed a great variability in toxicity on the different plankton organisms. The highest toxicity level was found for cream number 4 when tested on sea urchin, exhibiting an EC50 = 122.4 mg/L. The toxicity of the UV filter 2-phenyl-5-benzimidazolesulfonic acid, exclusively present in that cream, was evaluated in sea urchin, where an EC10 = 699.6 mg/L was obtained under light exposure. According to our results, all tested creams become nontoxic to plankton upon 30,000-fold dilution in seawater; thus, only local effects are expected. This study highlights the need to understand the toxic effects generated by solar protection products, as well as their ingredients, on marine organisms.


Introduction
Sunscreens are defined as substances designed to protect our skin from the sun's harmful rays. They reflect, absorb, and scatter ultraviolet A and B radiation [1]. Sunscreen formulations are composed of a wide variety of ingredients [2]. Their active components are organic and inorganic ultraviolet (UV) filters [3,4], each with different structures and solubilities. Most used UV filters such as octocrylene (OC), avobenzone, homosalate, and oxybenzone are organic molecules with conjugated aromatic structures whose function is mainly to absorb UV rays [3,5]. Organic UV filters are also added to plastics and other materials to provide them with resistance to light exposure [6]. In addition, sunscreens contain other components such as chemical preservatives, fragrances, and antioxidants substances [7,8]. These compounds have been detected in different seas at levels of ng/L [5,[9][10][11]. UV filters reach the marine environment directly from the skin of bathers, which is the main source. In addition, there is an indirect route involving the effluents from wastewater treatment plants and atmospheric deposition [5,8].
The toxicity of sunscreens depends on their composition. Numerous effects on marine organisms have been observed; in fact, some UV filters pose a significant risk to marine invertebrates [6,12,13]. UV filters such as OC, titanium dioxide (TiO 2 ), and ethylhexyl methoxycinnamate (EHMC) cause bleaching of some coral species by inducing the lytic cycle of viruses present in symbiont zooxanthellae [14][15][16], thus reducing their photosynthetic efficiency [17]. In addition, other effects have been seen in corals such as the induction of acute stress after exposure or abnormal fatty-acid metabolism [15,18]. Moreover, they

Test Organisms and Bioassays
The algal growth inhibition test was carried out according to the standard method from [49] adapted to T. lutea by [50]. The initial strain of T. lutea was obtained from the ECIMAT-Universidade de Vigo collection. Tests started with a 10,000 cells/mL inoculum in exponential growth phase. Cell numbers were recorded by using a Multisizer™ 3 Coulter Counter ® , Beckman-Coulter. Eight replicates for the control and four for the test solutions were prepared. Test tubes were incubated in an isothermal room at 20 • C and light conditions for 72 h. After 3 days, growth was measured as Growth inhibition (I) was calculated as where GR c and GR i are the growth rate in the control and growth rate in each test tube, respectively. Responses were corrected by control and expressed as R = GR i /GR c . Adult sea urchins (P. lividus) in ripe conditions were supplied by ECIMAT, and the sea urchin embryo test (SET) was carried out following standard methods [51]. Briefly, mature oocytes were fertilized in a 50 mL graduated cylinder with ASW, and fertilized eggs were transferred into glass vials (four replicates per treatment and eight controls) with airtight Teflon including 4 mL of exposure medium (final density of 40 per mL). Then, fertilized eggs were incubated at 20 • C in dark conditions for 2 days. After 48 h, the vials were fixed with three drops of 36% formaldehyde, observed in a Leica DMI 4000B inverted microscope, and the length (maximum linear dimension) of 35 individuals per vial was recorded using Leica LAS image analysis software (Leica microsystems, Wetzlar, Germany). The mean length increase for each treatment was expressed as the net response (R) in relation to the control response according to the following expression: where L t is the mean length of a treatment, L e is the mean length of eggs, and L c is the mean size of control.
In the case of the toxicity test with the PBSA filter, four additional replicates of each dilution and eight control vials were incubated in a room with a 16:8 h light/darkness photoperiod at 20 • C. According to [51], the pH of the PBSA leachate was adjusted with a drop of NaOH to meet the requirement of pH >7.
The acute lethal toxicity test with copepods followed standard methods [52] adapted by [53] to use nauplius larvae of Acartia tonsa. Mature copepods were obtained from a laboratory stock maintained by ECIMAT from 48 to 72 h before the start of the test. From the initial stock, adults were collected by a 300 µm mesh and incubated in laboratory conditions to produce nauplius <24 h following [53]. A total of 10 nauplii were transferred to 20 mL glass vials using a binocular stereoscope. Four vials for each dilution and eight vials for control were used. Copepod nauplius survival was recorded after 48 h and expressed as the percentage net response according to the expression R = (S t · 100)/S c , where S t is the mean survival of a treatment, and Sc is the mean survival of the control.

Statistical Analysis
Statistical analyses were performed using IBM SPSS statistics (v.24) (provided from Universidade de Vigo, Vigo, Spain). Normal distribution of data and homoscedasticity was checked using the Shapiro-Wilk and Levene's tests, respectively. Dilutions significantly different from the control (p < 0.05) were identified using Dunnett's post hoc test or Dunnett's T3, when the variances were not homogeneous, in order to find the lowest observed adverse effect concentration (LOEC) and the highest concentration with no observed adverse effects (NOEC). The dilutions that produced a 50% and 10% decrease in the endpoint (EC 50 and EC 10 ) and their 95% confidence intervals (CIs) were also calculated by fitting the data to a Probit dose-response model. Toxic units (TU) were calculated as TU = 1/EC 50 [54].

Growth Inhibition in Tisochrysis lutea
The toxicity of all sunscreens tested on T. lutea is shown in Table 1. As we can see, cream 3 was the most toxic for this species, (9.60 TU), whereas creams CR7 and CR8 were the least toxic to this microalga. Looking at the EC 50 values, toxicity was ranked as CR3 > CR4 > CR5 > CR1 > CR6 > CR2 > CR8 > CR7. The growth rate per day compared with control (R) is shown in Figure 1. For CR1 and CR3, the growth rate was negative in the undiluted leachate.

ENSITIVE
Dunnett's T3, when the variances were not homogeneous, in order to find the lowest observed adverse effect concentration (LOEC) and the highest concentration with no observed adverse effects (NOEC). The dilutions that produced a 50% and 10% decrease in the endpoint (EC50 and EC10) and their 95% confidence intervals (CIs) were also calculated by fitting the data to a Probit dose-response model. Toxic units (TU) were calculated as TU = 1/EC50 [54].

Growth Inhibition in Tisochrysis lutea
The toxicity of all sunscreens tested on T. lutea is shown in Table 1. As we can see, cream 3 was the most toxic for this species, (9.60 TU), whereas creams CR7 and CR8 were the least toxic to this microalga. Looking at the EC50 values, toxicity was ranked as CR3 > CR4 > CR5 > CR1 > CR6 > CR2 > CR8 > CR7. The growth rate per day compared with control (R) is shown in Figure 1. For CR1 and CR3, the growth rate was negative in the undiluted leachate. (a)

Sea Urchin Embryo Test (SET)
Results for the different dilutions of the sunscreen creams are shown in Table 2. According to the toxicity units and the EC50, cream CR4 stood out as the most toxic; its 10 g/L leachate needed to be diluted 300 times to lose toxicity. It was followed by creams CR8, CR6, and CR1, whereas creams CR7, CR5, CR3, and CR2 were classified as the least toxic. The larval size increase compared with control data (R) is shown in Figure 2.

Sea Urchin Embryo Test (SET)
Results for the different dilutions of the sunscreen creams are shown in Table 2. According to the toxicity units and the EC 50 , cream CR4 stood out as the most toxic; its 10 g/L leachate needed to be diluted 300 times to lose toxicity. It was followed by creams CR8, CR6, and CR1, whereas creams CR7, CR5, CR3, and CR2 were classified as the least toxic. The larval size increase compared with control data (R) is shown in Figure 2. (b) for sunscreens 5 to 8. * p < 0.05, ** p < 0.01, *** p < 0.001; n.s., no significant difference with control.

Survival Test in Copepods
The effects of all sunscreens on copepods are reflected in Table 3. As we saw previously in sea urchin, cream CR4 was once again the most toxic with an EC50 of 354.4 mg/L. Creams CR5, CR6, and CR7 were the least toxic to copepods.

Survival Test in Copepods
The effects of all sunscreens on copepods are reflected in Table 3. As we saw previously in sea urchin, cream CR4 was once again the most toxic with an EC 50 of 354.4 mg/L. Creams CR5, CR6, and CR7 were the least toxic to copepods. The survival percentage compared to control data (R) is shown in Figure 3. Cream CR4 caused 100% mortality even after 10-fold dilution of the leachate. Cream CR6 was the least toxic, showing survival rates close to 100% after just threefold dilution of the leachate. The survival percentage compared to control data (R) is shown in Figure 3. Cream CR4 caused 100% mortality even after 10-fold dilution of the leachate. Cream CR6 was the least toxic, showing survival rates close to 100% after just threefold dilution of the leachate.

PBSA Sea Urchin Embryo Test (SET)
The SET was conducted with PBSA, the specific UV filter present only in the most toxic sunscreen for zooplankton, and the results are shown in Table 4. As we can see, light conditions affected the results, since the PBSA leachate had to be diluted just threefold to lose toxicity when tested in darkness, whereas it was necessary to dilute it 30-fold when incubations were conducted under light. However, in both cases, toxicity was moderate and could not explain the toxicity of this cream to the early life stages of sea urchins and copepods. The larval size increase compared with control (R) is shown in Figure 4.

PBSA Sea Urchin Embryo Test (SET)
The SET was conducted with PBSA, the specific UV filter present only in the most toxic sunscreen for zooplankton, and the results are shown in Table 4. As we can see, light conditions affected the results, since the PBSA leachate had to be diluted just threefold to lose toxicity when tested in darkness, whereas it was necessary to dilute it 30-fold when incubations were conducted under light. However, in both cases, toxicity was moderate and could not explain the toxicity of this cream to the early life stages of sea urchins and copepods. The larval size increase compared with control (R) is shown in Figure 4.

Discussion
According to our results, all creams were within the nontoxic category in all species tested according to the classification of chemical substances and mixtures for their labeling [55], since all EC50 values were above 100 mg/L. Creams CR4 and CR8 for Paracentrotus lividus were closest to the limit with EC50 = 122.4 mg/L and 191.6 mg/L, respectively. In the case of copepods, cream CR4 was again the closest to this limit with EC50 = 354.4 mg/L. Some of the UV filters present in creams such as EHMC and OC have been previously studied in marine planktonic species [9,20,30]. For all sunscreens, higher amounts are needed to produce a 50% growth inhibition when they are part of a cream than independently [32]. The same has been observed in sea urchin [9,30] and in other marine organisms [12]. This may be due to the presence of humectants, hydrophobic excipients, and

Discussion
According to our results, all creams were within the nontoxic category in all species tested according to the classification of chemical substances and mixtures for their labeling [55], since all EC 50 values were above 100 mg/L. Creams CR4 and CR8 for Paracentrotus lividus were closest to the limit with EC 50 = 122.4 mg/L and 191.6 mg/L, respectively. In the case of copepods, cream CR4 was again the closest to this limit with EC 50 = 354.4 mg/L. Some of the UV filters present in creams such as EHMC and OC have been previously studied in marine planktonic species [9,20,30]. For all sunscreens, higher amounts are needed to produce a 50% growth inhibition when they are part of a cream than independently [32]. The same has been observed in sea urchin [9,30] and in other marine organisms [12]. This may be due to the presence of humectants, hydrophobic excipients, and emulsifiers that have a high affinity for these filters and, therefore, can reduce their bioavailability. Thus, it is important to know the composition because the concentration of one single UV filter can differ between creams, as can be seen in Table A6 for undiluted leaching (10 g/L). In fact, from these values, it is possible to estimate at what concentration these filters present toxicity or not with the NOEC or LOEC values.
The sensitivity of plankton organisms to UV filters largely varies among different taxa [6,30]. In the case of TiO 2 , a relevant toxicity has been detected at a concentration of 1 mg/L in Tisochrysis lutea and in other species of microalgae [20]. However, for T. lutea, no toxicity was noted for this filter in creams CR1 and CR5 at 16 and 18 mg/L, respectively, although toxicity was noted in cream CR6 over 0.72 mg/L. For OC, toxicity has been detected at concentrations lower than 1 mg/L even at concentrations of 0.1 mg/L in marine plankton organisms [30]. Nonetheless, no toxicity was observed in creams CR2, CR4, and CR5 at 10, 33, and 27 mg/L, respectively. More toxicity of the EHMC filter has been detected in invertebrates than in microalgae [9].
The differences in the ranking of toxicity of the creams tested on the three plankton species point to the differential toxicity of some of the sunscreens' components. In general, the toxicity on microalgae was lower with respect to the two species of zooplankton. This test is usually less sensitive, as has been seen with some UV filters [9,30]. The authors of [38] found that organic filters were less toxic than inorganic ones for Corophium orientale (crustacean), whereas the opposite was observed in Phaeodactylum tricornutum (phytoplankton). According to our results, this trend was not seen, since creams CR5 and CR6, which contain TiO 2 , were the least toxic toward the crustacean used. In addition, in Tisochrysis lutea, there were similar effects of creams with and without this type of filter.
Cream CR4 showed a remarkable toxicity on the two species of zooplankton tested, and this was the only formulation that included PBSA. Therefore, this UV filter was individually tested using the SET. Results (see Table 4) did not support the hypothesis that PBSA was the main cause of the toxicity of this formulation. However, the chemical speciation of this substance is highly influenced by pH. At high pH values, the nonionized form of the compound predominates, which is more toxic because it can readily pass through the lipid bilayer of the cell membrane [56]. For this reason, a bioassay in which the toxicity of PBSA is tested at different pH values is recommended before being able to rule out the contribution of this substance to the toxicity of cream CR4.
Lastly, in this UV filter studied separately, a significantly higher toxicity was observed when sea urchin larvae were incubated with light. This may be due to the generation of ROS [43] and the formation of transformation products resulting from the degradation of PBSA in the presence of UV radiation. The main degradation pathway has been found to generate four stable phototransformation products, which are indicated to be more dangerous to organisms than the parental compound [46]. Regarding the effects on biota, this UV filter is not well studied; the most severe concentration at which toxicity has been found is at 5 mg/L on zebrafish [45]. According to our results, a concentration of 1 g/L is necessary to cause observable effects on sea urchin larvae.

Conclusions
None of the tested sunscreens can be considered totally innocuous for the tested plankton species, since deleterious effects of undiluted 10 g/L leachates were observed in all cases on both phytoplankton and zooplankton. However, taking into account the high dilution factors of coastal environments, these effects are expected to be very limited on a geographical scale since a 300-fold dilution of the leachate produced at a 1:100 creamseawater ratio (i.e., a 30,000-fold overall dilution) resulted in the absence of toxicity, even for the most toxic cream on the most sensitive test species (cream CR4 on sea urchin).
The toxicity of sunscreens on microalgae and copepods was lower than on sea urchin larvae, since a 2-28-fold higher concentration of sunscreen was needed to cause a level of effect of 50%. This highlights the need to use a battery of test species and not a single model when assessing the ecological risk of chemicals on the marine environment.
The toxicity of the UV filter PBSA on P. lividus larvae increased in the presence of light, likely due to formation of photo-oxidized metabolites upon exposure to UV radiation.
The results of this study suggest the importance of a better understanding of the toxicity of the components of sunscreens both alone and as part of mixtures, which is how they are most often found in the environment.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.

Acknowledgments:
The authors thank Clara Mendoza Segura for her technical support during the tests and the ECIMAT staff for supplying the organisms.

Conflicts of Interest:
The authors declare no conflict of interest.