Reality Check: What can we expect from an Ecological Risk Assessment of UV filters on coral reefs?

In 2018, Hawaiʻi banned the sale or distribution of sunscreens containing the ultraviolet (UV) filters oxybenzone and octinoxate based on laboratory studies that indicated they cause adverse impacts on coral reefs (Downs et al. 2014). While this was not the first ban on sunscreen UV-filters, it was the most widely reported and controversial in the United States. Proponents of the ban highlighted the importance of coral reefs and the multitude of stressors contributing to their rapid global declines. Those who opposed it expressed concerns that it may reduce sunscreen options and lead to increasing incidents of skin cancers; this was succinctly summarized as “Essentially, … two ingredients that are both safe [for humans] and effective for use in sunscreen are being banned… on the basis of a single study…” (Hawaiʻi bans sunscreens that harm coral reefs, CNN July 3, 2018). While most can agree that the effectiveness of a chemical should not negate risks to the environment (Carson 1962), it is important to realize that chemicals are often regulated on the basis of a single study – or no studies at all. For example, new chemicals registered under the Toxic Substance Control Act (TSCA) may be regulated based solely on chemical structure. However, such reactions highlight that most stakeholders do not have a good understanding of how environmental risks are evaluated and will be disappointed in the data available to inform such decisions for UV-filters.

In 2020, US Congress passed an omnibus appropriations bill requiring the US Environmental Protection Agency (USEPA) to partner with the National Academy of Sciences (NAS) to conduct a review of potential impacts of currently marketed UV-filters on the environment. The mandate was to summarize the scientific literature, identify additional research needed to conduct an Ecological Risk Assessment (ERA), and identify potential public health implications of reduced sunscreen use. The NAS found that UV-filters are detected in water samples from around the world in concentrations that cause effects to organisms in laboratory tests and are found in the tissues of organisms ranging from crayfish to dolphins. The NAS recommended that the US Environmental Protection Agency (USEPA) should conduct an ERA for all currently marketed UV filters and any new ones that become available (NAS 2022). ERAs evaluate the likelihood that the environment might be adversely impacted by a chemical and are often conducted in a tiered process that begins with a more protective screening-level assessment and moves to more realistic assessments, as needed, to reduce uncertainties. ERAs are comprised of exposure and effects analyses that are integrated in a risk characterization. Each of these analyses contain their own uncertainty that provide fodder for criticism, even though the uncertainties typically stem from lack of or limited data that are beyond the control of the risk assessor. First, I review the uncertainties associated with the primary phases of the ERA as relevant to UV-filters in marine environments to provide a reality check of what we can expect from such efforts. From there, I provide a recommendation for next steps that will require a global network of collaboration to provide useful and impactful assessment to reduce the environmental impacts of UV-filters.

Uncertainty in ERAs

UV-filters applied as sunscreens enter aquatic environments from recreational activities and waste effluent (Mitchelmore et al. 2021). The exposure analysis of an ERA is a qualitative and quantitative characterization of where a chemical is found and its movement in the environment. Data risk assessors would like to have for an exposure analysis include chemical sources; where, when, and how much UV-filters are in different environmental compartments (air, water, sediment, biota); movement and transport within and across compartments; and bioaccumulation and degradation processes. Data that are available for UV-filters are chemical properties, monitoring data, and limited model output. At best, monitoring data would be adequate for model validation, but this is rarely the case for even data-rich chemicals. Monitoring data represent snapshots in time and space and are not robust representations of chemicals in environment. Chemical occurrence and concentration are highly variable across space and time and can span orders of magnitude. Detection and quantification are heavily influenced by sampling designs and analytical capabilities. Robust characterization of a chemical in the environment requires abundant monitoring data, which is resource intensive. Monitoring data may identify locations with high exposure potential, such as beaches at embayments with shallower water, long residence time, and high use by people (NAS 2022). Exposure models for UV-filters in marine environments are being developed to include UV-filter concentrations in sunscreen formulations, estimates of applied sunscreen, surface area of swimmer’s bodies, number of swimmers in an area of interest, and UV-filter solubility (Sharifan et al. 2016). These models estimate mass loading, but do not include hydrodynamics, environmental conditions, compartmentalization, bioaccumulation, degradation, or frequency/duration of peak concentrations. Including these factors for more comprehensive estimates of environmental concentrations may be achieved using models such as ChemicalDrift, which couples chemical processes with a Lagrangian particle tracking model, OpenDrift (Aghito et al. 2023). Unfortunately, as model complexity increases to allow for more realistic scenarios, so does the variability associated with its predictions. The Pesticide in Water Calculator, a model regularly used to estimate pesticide concentrations in surface and ground water, may predict concentrations that span >10 orders of magnitude using realistic estimates for over 50 parameters (Sinnathamby et al. 2020). As such, stakeholders should be prepared for modeled exposure estimates to span large ranges, and minimal monitoring data for validation.

Effects analyses of all chemicals are challenged by the availability of data, representativeness of tested species, interpretation of endpoints, laboratory to field extrapolation, relevance of exposure duration, individual-to-population extrapolation, and extrapolation across space and time. Data that risk assessors would like to have for an effects analysis include quantitative Adverse Outcome Pathways, interspecies linkages, and studies on diverse taxa exposed for acute and chronic durations. Effects data for ten UV filters compiled by the NAS (2022) are limited to 140 publications. These data include many studies that did not follow best practices of standard toxicity test protocols, reported non-standard endpoints, and generally resulted in inadequate comparability among studies. Additionally, data are grossly lacking for chronic toxicity, tests on benthic organisms and sediment exposure, degradate toxicity, impacts to communities and ecosystems, and linking downstream processes to individuals or populations. From these data, we can expect that some compounds will be evaluated based on data for fewer than 5 species (most < 10 species), the majority of which will be standard test species that are not relevant to coral reefs. We can expect that many endpoints will not be ecologically relevant (e.g., gonad histology) and will lack standardization that will make robust conclusions difficult.

Risk characterization is performed for both acute and chronic exposure durations and estimates the likelihood of adverse effects based on the co-occurrence of chemical and species in concentrations that cause harm. One of the most common metrics calculated for risk characterization is the Risk Quotient (RQ), which is the ratio of an Estimated Exposure Concentration (EEC; i.e., the 90^th percentile of a modeled exposure distribution, Young 2019) and an effects estimate. The latter may be the toxicity value of the most sensitive species or the 5^th percentile of a Species Sensitivity Distribution (SSD, i.e., cumulative probability distribution of species sensitivity). The RQ is an artifact of available data and does not inform or relate to probability or magnitude of adverse impacts. It is, in essence, a “yes” or “no” answer with limited interpretation.

The Role of ERAs in EPA

The USEPA conducts full scope ERAs under statutes aligned with the regulation or mitigation of chemicals. ERAs are conducted under the Federal Insecticide, Fungicide and Rodenticide Act and TSCA for the registration of pesticides and industrial chemicals, respectively, and to inform the mitigation of contaminated sites under the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA, “Superfund”). The practical challenge here is that while the USEPA has the expertise to conduct an ERA of sunscreens, they have no authority to use that ERA to regulate personal care products including sunscreens (which are regulated by the Food and Drug Administration under the Federal Food, Drug and Cosmetic Act). Rather than looking to EPA to conduct a full scope ERA using limited data rife with uncertainties, collaborative efforts could focus on collecting data to inform water quality benchmarks, which are limited to effects assessment and are recommendations to States, not regulatory decisions. States could use benchmarks to guide discharge permits and mitigate problem areas. For example, a recreation area that controls the admission of bathers in embayments containing coral reefs can implement their own monitoring program and use benchmarks to guide the volume of people permitted into the site. In this approach, long established methods using the best available science equips local municipalities with peer-reviewed guidance to manage their local reefs.

A Path Forward

The most significant challenge for establishing water quality benchmarks is obtaining the Minimum Data Requirements (MDRs) that ensure that benchmarks are derived from diverse taxa (Stephan et al. 1985). A list of MDRs for fresh and saltwater environments and the data currently available to meet them for oxybenzone is presented in Table 1. Oxybenzone is the most data rich UV-filter and only ~50% of the MDRs are met for acute data, and less than 25% are met for chronic data (see NAS 2022). For the international research community that is invested in reducing impacts to coral reefs from UV-filters, a network of global collaborations is needed to acquire data to meet the MDRs. In doing so, we must collaborate to establish and follow a set of best practices to ensure that standardized methods are followed, endpoints and exposure duration are relevant and comparable across species, and ensure data are not over-represented for some taxa (i.e., bony fish) while underrepresented for others. Only through such a collaboration can we produce guidance for resource managers based on the best available science to support state and local actions for management and mitigation of UV filters for the protection of coral reefs.

Table 1.

Minimum Data Requirements (MDR) for water quality benchmarks. Green highlights indicate MDRs that are currently met for oxybenzone acute studies.

Freshwater		Saltwater
A	The family salmonid	A	Family in the phylum Chordata
B	A second family of Osteichthyes1 preferably a commercially or recreationally important warmwater species	B	Family in the phylum Chordata
C	A third family in the phylum Chordata2	C	Either the Mysidae or Penaeidae family
D	A planktonic crustacean	D	Family in a phylum other than Arthropoda or Chordata
E	An insect	E	Family in a phylum other than Chordata
F	A family in a phylum other than Arthropoda3 or Chordata	F	Family in a phylum other than Chordata
H	A family in any order of insect or any phylum not already represented	H	Family in a phylum other than Chordata

¹Bony fish;

²Vertebrates and relatives;

³Invertebrates with exoskeleton

Data Availability Statement:

No data were generated for the development of this article.