For several decades, the U.S. Department of Defense (DoD) widely used aqueous film-forming foam (AFFF) formulations for training and operations involving fire suppression. These AFFF formulations contained relative high quantities of perfluorooctane sulfonate (PFOS), as well a range of other of per- and polyfluoroalkyl substances (PFAS). This project is being conducted in three phases. The objective of Phase I of this project was to develop a framework for conducting scientifically sound risk assessments for PFAS in threatened and endangered (T&E) species at these sites. The Phase I study involved (i) a comprehensive literature review of physical-chemical properties, bioaccumulation metrics and environmental concentrations, (ii) development of a risk assessment framework for assessing PFAS bioaccumulation and exposure risks in T&E species at AFFF-impacted DoD sites and (iii) application of the proposed framework to several DoD site sfor which existing PFAS monitoring data were available. The study aimed to help guide future research efforts and risk assessment initiatives related to exposure of legacy PFAS in T&E species at AFFF-impacted DoD sites. Results are available in the Phase I Final Report.

To support this activity-based approach as well other approaches, there is a need for reliable information regarding the partitioning properties of PFAS in biological media and tissues of biota. The primary objective of the Phase II effort is to conduct laboratory-based investigations to determine the partitioning properties or “solubilities” of PFAS in biological media and organism tissues. The laboratory-generated data will then be used to parameterize mechanistic models representing PFAS bioaccumulation in aquatic and terrestrial food webs, including those comprising water-respiring and air-breathing T&E species. These models will be tested by comparing model predictions with observed data generated from a bioconcentration experiment using rainbow trout (Oncorhynchus mykiss), as well as other literature bioaccumulation data.

Technical Approach

During the Phase I study, the approach generally followed conventional methods employed for ecological risk assessment, including exposure characterization, effects characterization, and risk estimation. In particular, the approach utilized a combination of field-based measurements and bioaccumulation modeling to evaluate exposure in T&E species. Toxicity reference values (TRVs) were derived using the available toxicity data, along with species-sensitivity distributions (SSDs) and resulting fifth percentile of the hazardous concentration levels (HC5s) or application of uncertainty factors to the lowest observed toxicity values.

In the framework, a chemical activity-based risk assessment approach was used. The chemical activity of a chemical (a, unitless) in a given medium is the ratio of the concentration (C, mol/m3) of the chemical in the medium and the chemical’s apparent solubility (S, mol/m3) in the medium (i.e., C/S). The approach involved three key steps, including (i) determining chemical activities of PFAS in external environmental media (e.g., water, soil, prey items), (ii) determining internal chemical activities of PFAS in T&E species and (iii) comparing those estimated activities to activities related to biological effects observed in vivo and/or in vitro. The merits of this approach were that monitoring data from several or diverse environmental media and sampling devices could be included in a risk assessment and monitoring data of multiple PFAS can be interpreted in terms of toxicity. This approach increased the weight of evidence in risk evaluations and facilitates coordination of research efforts by different research groups by expressing available information (i.e., from monitoring, modelling and toxicity assays) in terms of a common metric.

The chemical activity-based risk assessment approach also incorporated ToxCast AC50 and other in vitro assay data in a risk assessment, which was particularly useful for T&E species that can often not be used in in-vivo toxicity studies. This approach was consistent with the goal of minimizing animal studies in toxicity testing, highlighted in the National Research Council’s vision and strategy for exposure and toxicity testing in the 21st Century. While the chemical activity-based approach can increase the information that is used in a risk assessment, the risk assessment remains primarily focused and reliant on ecologically relevant metrics (e.g., growth, development, reproduction) in wildlife.

In the Phase II effort, the research involves three tasks:

  • Task 1 involves conducting a series of laboratory-based experiments that will employ rapidly equilibrating thin-film solid-phase microextraction (TF-SPME) passive dosing method for determining the partitioning properties and “solubilities” of PFAS commonly detected at AFFF-impacted DoD sites, including PFOS, PFHxS, PFOA, PFNA, PFDA, FOSA and n:2 FtS in different biological media and organism tissues as well as biofluids and tissue homogenates from rainbow trout.
  • Task 2 involves conducting a laboratory-based bioconcentration experiment with juvenile rainbow trout to test and evaluate the ability of partitioning properties of PFAS in biological constituents (proteins, phospholipids, neutral lipids), measured as part of Task 1, to determine the partitioning properties of PFAS in in-vivo tissues of aquatic organisms.
  • Task 3 involves entering the measured partitioning properties into previously developed PFAS aquatic and terrestrial food web bioaccumulation and risk assessment models and then testing the models by comparing predictions of PFAS concentrations in organisms to previously compiled observed concentrations in various species including T&E species.

Based on the results of these tasks, the team will further evaluate the overall performance of the activity-based modelling and risk assessment approach for practical risk assessment purposes.

Interim Results

Phase I efforts showed that results of the available ToxCast data for individual PFAS indicate that commonly detected perfluoroalkyl acids (e.g., PFOS, PFHxS, PFOA, PFNA, PFDA) exhibit specific mode of toxic action, in the chemical activity range of between 10-6 to 10-3, generally below levels related to narcosis (= 0.01). Chemical activities of PFAA associated with effects in vitro (i.e., ToxCast AC50 values) are generally similar to those associated with toxic effects in vivo. ToxCast data for PFAA precursors (N-Et-FOSA and PFOSA) suggests these neutral hydrophobic compounds tend to exhibit baseline toxicity behavior, with effect levels occurring in the range known to be associated with nonpolar narcosis (= 0.01). As PFAAs exhibit toxic effects in the same chemical activity range, a simple additivity approach may be adopted to incorporate mixture effects. However, as PFOS is typically the predominant PFAA (> 95%), contribution of other PFAAs to the toxicity of PFAA is often negligible. Risk assessments based solely on PFOS may adequately represent the overall PFAS risk at a given site, especially if PFAAs are the main PFAS class of concern.

A preliminary mechanistic PFAS food web bioaccumulation model was developed to predict internal exposure levels (concentrations and activities) and external exposure (daily intake, μg/kg BW/d) of individual PFAS in various aquatic and terrestrial organisms that include T&E species and their prey items. The model was parameterized and applied to simulate PFAS bioaccumulation in T&E species at several DoD sites that have existing PFAS monitoring data. The model was shown to predict internal PFAS exposure levels in biota at DoD sites reasonably well, with model predicted values generally within a factor of three of the observed field data. The developed PFAS food web bioaccumulation model indicates this mechanistic modeling approach may be useful for future risk assessments of T&E species potentially exposed to PFAS at AFFF-impacted DoD sites. However, further development and testing of this modeling approach is still needed. In particular, information is needed on the partitioning properties of PFAS in biological media. This information is not only crucial to the development of a chemical activity-based risk assessment approach but also for other approaches.

T&E species with habitat ranges overlapping AFFF-impacted DoD sites included the bog turtle (Clemmys muhlenbergii), northern long-eared bat (Myotis septentrionalis), red-cockaded woodpecker (Picoides borealis) and eastern massasauga rattlesnake (Sistrurus catenatus). For sites with relatively high PFAS concentrations, risk quotients (RQs) related to PFOS exposure in T&E species often exceeded the level of concern (LOC) of 0.1. Omnivorous and carnivorous birds, mammals and reptiles are shown to exhibit a relatively high degree of PFAS bioaccumulation and hence exposure risk, compared to aquatic organisms at a given site. Model predictions indicate that at some sites with elevated PFAS concentrations in sediments, concentrations in benthic invertebrates can attain levels similar to those expected to induce acute effects in aquatic organisms. Biomagnification of PFAS in aquatic insectivorous bird species (feeding on benthos) cause very high exposure levels and associated risks.

PFAS concentrations in soils were found to be very important for exposure risks in numerous T&E species within terrestrial food webs, including terrestrial reptiles (eastern massasauga rattlesnake, Kirtland’ warblers). PFAS exposure risks to upper trophic terrestrial wildlife were in many cases high. Risk quotients often exceed the LOC of 0.1. Sites exhibiting high PFAS concentrations in soils, such as those at several active USAF sites, are expected to cause high levels of risks to terrestrial organisms. In some cases, the estimated dose in terrestrial wildlife exceeds the PFOS LD50 of 150 mg/kg BW/d. The initial findings show that risks of PFAS to T&E species of terrestrial food-webs are of particular concern. 

It is important to note that risk estimates for T&E species in the present Phase I study are based on scenarios that assume exposure occurs via concentrations at the studied DoD sites. The extent of interaction of T&E species and their prey with AFFF-impacted soils and surface waters is a major knowledge gap in the present assessment of PFAS exposure risks of these species at DoD sites. Other knowledge gaps include the frequency and duration of various T&E species at AFFF-impacted DoD sites. In particular, studies to determine PFAS concentrations in prey and relative prey consumption rates would be useful. Other important research needs include investigations to better understand the transfer of PFAS from to insect-consuming animals and upper trophic terrestrial wildlife.


The Phase I efforts demonstrated the potential and merit of a chemical activity-based approach for assessing bioaccumulation and exposure risks of PFAS to T&E species of concern. A limitation of this approach was that the apparent solubility values used to estimate chemical activities were based on numerous assumptions regarding physicochemical properties, phase partitioning, protein-binding and toxicokinetics. Currently, there is a need for further laboratory-based measurements of PFAS solubilities in different environmental and biological media, as well as media-water distribution coefficients for different transporter proteins and distribution coefficients for different transporter proteins, structural proteins, phospholipids, neutral lipids, carbohydrates and organic carbon. Accurate estimates of solubility and distribution coefficient values will undoubtedly strengthen the reliability of the activity-based risk assessment approach. This will also aid PFAS bioaccumulation modeling efforts, as the various distribution coefficients are key parameters within the proposed mechanistic food web bioaccumulation model.

The Phase II study will generate new data to improve our understanding of PFAS partitioning into different biological media and organism tissues in support of risk assessment and will improve bioaccumulation and risk assessment models for PFAS in T&E species by determining the input parameters for the partitioning properties or “solubilities” of PFAS in organism’s tissues. The study will provide testing results that will strengthen confidence in the application of the model’s results for practical bioaccumulation and exposure based risk assessment of PFAS at AFFF-impacted DoD sites, and will also characterize uncertainty of the bioaccumulation and risk assessment of PFAS. In short, the generated data will help to strengthen predictive models used in ecological risk assessment of PFAS, with a specific focus on T&E species for which empirical approaches are often not an option.