Toxicity of Wildland Fire Retardants to Rainbow Trout in Short Exposures

Long‐term wildland fire retardants are one important tool used to control and suppress wildfires. During suppression activities, these retardants may enter water bodies; thus, there is a need to understand their potential effects on aquatic biota. We investigated the effect of three current‐use wildland fire retardants to juvenile rainbow trout (Oncorhynchus mykiss) survival in short exposures more realistic to actual intrusion scenarios. Lethal effect concentrations decreased with time and varied among chemicals (LC95A‐R > 259‐Fx > MVP‐Fx). The lowest effect concentrations observed were 2 to 10 times above the threshold used by federal agencies to assess potential impacts to aquatic organisms following a retardant intrusion. These data can be used by resource managers to balance wildfire control with potential environmental impacts of retardant use. Environ Toxicol Chem 2024;43:398–404. Published 2023. This article is a U.S. Government work and is in the public domain in the USA. Environmental Toxicology and Chemistry published by Wiley Periodicals LLC on behalf of SETAC.


INTRODUCTION
Wildland fires in the United States are becoming more intense and more frequent in the 21st century (Dennison et al., 2014;Senande-Rivera et al., 2022).Fires on public lands result in economic and recreational damage and can impact surrounding communities if the spread of the fire is not contained (Bawa, 2017;Downing et al., 2022).Also, fires can be ignited on private lands and cross boundaries into public land, causing widespread impacts.In 2018, California alone estimates the economic damage caused by wildfires to be between $126.07 and $192.93 billion (Wang et al., 2021).Although wildfire has many ecological benefits, there are still damaging effects, especially in forest catchments where water quality will rapidly change when directly exposed to fire (Yu et al., 2019).Controlling the spread of wildfire helps protect local communities as well as natural and cultural resources.
Between 2017 and 2021, the US Forest Service (USFS) controlled 10,900 to 15,200 fires each year, leaving between 3.1 and 7.1 million acres of burned USFS land (Hoover & Hanson, 2022).The application of ammonia-based fire retardant chemicals on the landscape is one control method used by firefighting agencies, including the USFS, to suppress active wildland fires.From 2012 to 2020, over 123.7 million gallons of fire retardant were applied to control wildland fire by the USFS, with an average increase of over 950,000 gallons applied per year (US Department of Agriculture [USDA] Forest Service, 2022).According to US federal wildland fire management policy, a 91.4-m buffer is required around water bodies where fire retardant is aerially applied, to limit negative environmental effects to aquatic ecosystems (National Interagency Fire Center, 2023).Despite the policy of avoidance areas around water bodies, intrusions occur, which is when retardant is applied either within the buffer or directly to water (USDA Forest Service, 2022).Fish kills have been documented following intrusions of chemicals (USDA Forest Service, 2011).Thus, there is a need to understand the toxicity of these chemicals to sensitive aquatic species within North American freshwater systems under practical exposure conditions.
Wildland fire retardants are applied during active fire scenarios; thus, safety concerns preclude measurement of retardant in streams immediately following application.In addition, application in flowing water may result in a potentially short time span of measurable retardant concentrations.Finally, instrusions are unplanned, making it difficult to predict where measurements or samples should be collected.Therefore, safety concerns, flowing water, and the unintentional nature of intrusions have prevented measurement of retardant concentrations in the environment after an intrusion.As a result, tools have been developed to aid resource managers, including the USFS, in understanding the potential exposure conditions experienced by stream biota following an intrusion of fire retardant.One tool, a spill calculator, models the duration and concentration of exposure in the event of an intrusion, incorporating the processes of advection and dispersion (R Core Team, 2021;USDA Forest Service, 2022).Based on >1150 simulations, the model predicted a maximum exposure duration of 5.2 h to concentrations above a threshold (10% of 96-h lethal concentration to x% of test population, in this case 50% [LC50]) at any point downstream of the intrusion site.To date, the only effect concentrations available for most currentuse wildland firefighting chemicals are based on 96-h exposure endpoints, a standard duration for toxicity bioassays.Given that the maximum exposure duration predicted by the spill calculator is far shorter than 96 h, there is a need to determine the toxicity of fire chemicals to sensitive stream biota at shorter, more relevant time intervals.
The goal of the present study was to determine effect concentrations of current-use fire retardants at relevant time intervals.To accomplish this, we assessed the toxicity of three retardants to juvenile rainbow trout (Oncorhynchus mykiss) at 6, 12, and 24 h of exposure.

METHODS
Rainbow trout eyed eggs (Erwin/Arlee strain) were obtained from the US Fish and Wildlife Service, Ennis National Fish Hatchery (Ennis, MT).We chose rainbow trout because they are a commonly used salmonid in toxicological studies.Salmonids are an important cultural and economic resource in the western United States (Atlas et al., 2021;Criddle & Shimizu, 2014;Yoshiyama, 1999).Also, many US salmonids are imperiled, making them focal species for resource managers in North America (Ford, 2022).Eggs were held in incubators at 12 °C in flowing well water (pH 8.0, hardness 280-300 mg/L as CaCO 3 ) until hatch and swim-up of fry.Trout fry were transferred to a 425-L flow-through tank at 12 ± 1 °C on a 16:8-h light:dark cycle on swim-up and held until the start of experiments (30-60 days posthatch [dph]).Swim-up fry were fed 1-day-old brine shrimp (Artemia sp.) nauplii (Brine Shrimp Direct, Ogden, UT) and slowly transitioned to Rangen.The fish were acclimated to 12 ± 1 °C ASTM soft water (ASTM International, 2014) 3 days prior to the start of each assay.The present study was reviewed and approved by the Institutional Animal Care and Use Committee at the Columbia Environmental Research Center under IACUC 19-007.
Three current-use fire retardants (Phos-Chek ® MVP-Fx, Phos-Chek ® LC95A-R, and Phos-Chek ® 259-Fx) from the USFS Qualified Products List were selected for testing.These products are manufactured as either a liquid or powder concentrate comprised primarily of either ammonium polyphosphate (LC95A-R) or a combination of mono-and diammonium phosphate (MVP-Fx and 259-Fx;Perimeter Solutions, 2018, 2019a, 2019b).Additional ingredients include dye, clay, iron oxide, and trade secret performance additives (USDA Forest Service, 2022).Prior to application on the landscape, these concentrates are mixed with water to achieve a solution that is between 11% and 22% retardant by weight, depending on the product.Three static 24-h assays, one for each retardant tested, were conducted using a series of six concentrations plus a control treatment of ASTM soft water (Table 1; ASTM International, 2014).The concentration series for each chemical was based on previously reported 96-h LC50 values and the estimated initial concentrations from simulated intrusion scenarios (see Rehmann et al., 2021, Supporting Table S1).For each chemical, a 50% to 70% dilution series was prepared by mixing stock concentrations of fire retardant with ASTM soft water (ASTM International, 2014).Ten replicates per treatment were tested (7 concentrations [including control] × 10 replicates = 70 test chambers) in each assay.Each test chamber, a 3.7-L (1 gallon) glass jar, was filled with 1 L of the appropriate treatment solution (chemical or control) and stocked with one juvenile rainbow trout (30-60 dph).Because the primary constituent of each retardant used in the present study is a form of ammonia, as described above, total ammonia was measured in all treatments to confirm that the target retardant concentration series was achieved.
We recorded mortality in replicate chambers either 30 (LC95A-R and 259-Fx) or 60 (MVP-Fx) min after the initiation of each assay.Mortality was checked hourly until 12 h of exposure had elapsed.A final mortality observation was recorded at the conclusion of each assay, after 24 h of exposure.All surviving fish were counted and euthanized by immersion in neutral buffered tricaine methanesulfonate.Dead fish were removed from exposure chambers on discovery.
Water quality parameters were measured prior to the addition of fish.Composite water samples of all replicates per treatment were collected, and measurements of temperature, dissolved oxygen, specific conductance, pH, alkalinity, hardness, and total ammonia were taken.

Statistical analysis
Kaplan-Meier survival curves were generated for each concentration treatment in a given assay.Log-rank tests were used to identify differences in survival curves for each chemical.Pairwise log-rank tests, with a Bonferroni correction for multiple comparisons (α = 0.05/6), were used to compare the survival curve of the control to each concentration within a chemical treatment.Survival data were analyzed and plotted using the packages "survival" and "survminer" in R, Ver 4.2.2 (Kassambara et al., 2021;R Core Team, 2021;Therneau, 2023).Also, the curves were used to determine the time to death of 20% of the sample for each concentration of each chemical.
The LC50, LC20, LC10, and LC5 were calculated at 6, 12, and 24 h for each chemical.These time points were selected to account for the maximum exposure duration estimated in Rehmann et al. (2021) as well as more conservative estimates at 12 and 24 h.Effect concentrations were calculated using the packages "drc" and "multcomp" in R, Ver 4.2.2, for each of the three chemicals (Hothorn et al., 2008;Ritz et al., 2015).An asterisk in the legend identifies the lowest treatment concentration with a mortality curve statistically different from the control using log-rank tests with a Bonferroni correction for multiple comparisons (α = 0.05/6 = 0.008).A horizontal black dashed line is placed at 20% probability of mortality.There is an x-axis break after 12 h because no observations were made between 12 and 24 h after initiating exposure of fire retardant to fish.
One-way analysis of variance was used to compare differences in mortality among concentrations within a chemical at each time point of interest.Following each analysis of variance, lowest-observable-effect concentrations (LOECs) were determined using post hoc Tukey's honestly significant difference tests to identify the lowest concentration significantly different from control for mortality at 6, 12, and 24 h (α = 0.05).

RESULTS
Water quality measurements are provided in Table 1.Dissolved oxygen and temperature were relatively consistent across treatments within each chemical assay.Specific conductance, pH, alkalinity, and ammonia levels were relatively consistent in control water across all assays.Increased concentrations of fire retardant in solution increased the conductance, alkalinity, and ammonia and decreased the pH for all retardants.A 70% dilution series was targeted for MVP-Fx.Measured total ammonia concentrations confirm that an average of 75 ± 7% dilution series was achieved (Table 1).A 50% dilution series was targeted for LC95A-R and 259-Fx.Measured total ammonia concentrations confirm that an average of 54 ± 14% and 50 ± 4% dilution series was achieved for LC95A-R and 259-Fx, respectively.
Lethal effect concentrations for all chemicals are reported in Table 2.In general, as exposure time increased, effect concentration decreased.This pattern held across all chemicals with the exception of LC95A-R, where effect concentration was similar or increased between 12 and 24 h of exposure.Lethal concentrations of 5% of the test population, which may account for sublethal effects of chemical exposure, ranged from 4610.22, 429.74, and 711.59 mg/L at 6 h to 1960.40, 256.90, and 208.59 mg/L at 12 h for MVP-Fx, LC95A-R, and 259-Fx, respectively.
Mortality at hours 6, 12, and 24 differed for all chemicals among concentrations (Table 3).Table 2 shows LOECs for all time points of interest.Greater LOEC values were observed at Hour 6 than at Hour 12 or 24.

DISCUSSION
Two factors that determine exposure are concentration and duration, which ultimately influence toxicity and their resulting effects on organisms.This relationship is apparent in the general pattern of decreasing LC50s with increasing exposure duration (6 to 12 to 24 h) for all chemicals tested in the present study (Figure 2 and Table 2).There was a decrease between 30% and  50% in LC50s from 6 to 12 h across chemicals.This pattern continued for MVP-Fx with a 30% decrease in LC50 from 12 to 24 h, but for LC95A-R and 259-Fx, the percentage of change was less (7.5% decrease for LC95A-R and 5.5% increase from 259-Fx) between 12 and 24 h.Lethal effect concentrations at 96 h reported by the USFS (USDA Forest Service, 2022) are similar to effect concentrations found in the present study at 24 h for MVP-Fx and LC95A-R.The 24-h LC50 for 259-Fx was lower in the present study compared to the 96-h LC50 reported by the USFS.Mortality observed at 24 h in the present study was similar to mortality observed in Buhl (2023) at the same time point with comparable concentrations for MVP-Fx and 259-Fx.More variability was observed in LC95A-R comparisons with greater toxicity documented at lower concentrations in the present study.
To put our current data in context, we rely on a published model (spill calculator) that estimates exposure duration to an established threshold concentration (Rehmann et al., 2021).As previously discussed, this threshold concentration is 10% of available 96-h LC50s (202.4 mg/L for MVP-Fx, 38.6 mg/L for LC95A-R, and 86.0 mg/L for 259-Fx; USDA Forest Service 2022; orange dashed line in Figure 2).These thresholds were selected to minimize potential fish kill events and to account for sublethal effects.In >1150 simulations, the spill calculator predicted a maximum exposure duration of 5.2 h to concentrations above these thresholds downstream of an intrusion.In addition, the maximum exposure duration was ≤1 h in approximately 75% of the simulated intrusions.In the present study, the lowest-effect concentrations (from LC5s to LC50s and LOECs; Table 2) between 6 and 24 h were approximately 2 to 10 times above these thresholds (Figure 2).In addition, Wells et al. (2004) found that rainbow trout avoided retardant at all concentrations tested; thus, mobile organisms may move out of retardant plumes, further shortening exposure durations.Therefore, our data suggest that under typical firefighting operations, concentrations are unlikely to exist for the duration necessary to result in ≥20% mortality downstream of a retardant intrusion (Figure 1).
We documented the same pattern in toxicity among chemical formulations, LC95A-R > 259-Fx > MVP-FX, as previous studies with these retardants (Puglis et al., 2022;USDA Forest Service, 2022).This pattern in formulation toxicity may be explained in part by the differences in un-ionized ammonia, the primary toxic component of these fire retardants (Buhl & Hamilton, 1998), found in the three chemicals.The concentration of un-ionized ammonia in 7000 mg/L of MVP-Fx is similar to that found in 500 mg/L of 259-Fx and 312.5 mg/L of LC95A-R (Table 1), meaning lower-concentration treatments of 259-Fx and LC95A-R have a similar level of un-ionized ammonia than the second highest-concentration treatment of MVP-Fx.Therefore, it would take a much more concentrated intrusion of MVP-Fx to contain the same amount of ammonia as an intrusion of 259-Fx or LC95A-R.There may also be other, unmeasured components of the retardants that could be contributing to the toxicity of these products.The retardants are acquired as a powder or liquid concentrate and must be mixed into solution with water prior to application on the landscape.The mix ratio of product to water is greatest in LC95A-R and smallest in MVP-Fx (Perimeter Solutions, 2020).Thus, not only is LC95A-R the most toxic of the three retardants but intrusions of LC95A-R would result in a greater concentration of retardant in streams compared to 259-Fx and MVP-Fx at the same application rate.
In the present study, we examined the lethal effects of fire retardants to rainbow trout; however, retardants could have sublethal effects on fish such as altered behavior, changes in growth and development, disrupted chemosensory function, or degradation of habitat.High levels of ammonia can cause oxidative stress and an imbalance in ion regulation in fish (Eddy, 2005;Zhang et al., 2019).Acute effects can lead to health outcomes such as edema, loss of equilibrium, and hyperventilation (Eom & Wood, 2021;Smart, 1978;Zhang et al., 2019).Dietrich et al. (2013) documented damage to gill tissue and reduced survival after transitioning to seawater in salmon smolts following acute exposure to current-use fire retardants.Gill (purple), 12 (yellow), and 24 h (red) of exposure to wildland fire retardants MVP-Fx (A), LC95A-R (B), and 259-Fx (C).Curves were drawn using a loglogistic model in the "drc" package in R, Ver 4.2.2.The gray vertical dotted line is an intercept at the reported median lethal concentration (LC50) for the corresponding chemical in Rehmann et al. (2021).The orange vertical dotted line is an intercept at the spill calculator toxicity threshold (10% 96-h LC50) reported in Rehmann et al. (2021).Shading represents 95% confidence limits.
damage has also been documented in chronic ammonia exposures (Benli et al., 2008;Smart, 1976;Zhang et al., 2019).We did not investigate the potential effects of retardant on fish tissues, nor were we able to observe fish behavior in most of the retardant treatments due to reduced visibility in experimental chambers.
While our data provide effect concentrations for short, environmentally relevant exposure durations, aquatic organisms are likely to experience multiple stressors alongside retardant intrusion.Wildfire can impact streams by increasing sedimentation, debris flow, and ash and may also cause fluctuations in dissolved oxygen, pH, and temperature (Dunham et al., 2007;Earl & Blinn, 2003;Rhoades et al., 2011;Sanders et al., 2022).These stressors could interact to intensify the impact of a retardant intrusion on the aquatic environment.Incorporating multiple stressors in future studies will increase our understanding of the effects of wildland fire retardant intrusions to aquatic organisms under more realistic scenarios.This information would provide more relevant data to resource managers as they balance wildfire control with impacts to the environment.

FIGURE 1 :
FIGURE 1: Kaplan-Meier curves showing mortality probability for 30to 60-day posthatch rainbow trout (Oncorhynchus mykiss; n = 10/concentration) throughout 24-h exposures of MVP-Fx (A), LC95A-R (B), and 259-Fx (C).Each colored line represents the mortality curve at a specific test concentration.An asterisk in the legend identifies the lowest treatment concentration with a mortality curve statistically different from the control using log-rank tests with a Bonferroni correction for multiple comparisons (α = 0.05/6 = 0.008).A horizontal black dashed line is placed at 20% probability of mortality.There is an x-axis break after 12 h because no observations were made between 12 and 24 h after initiating exposure of fire retardant to fish.

TABLE 1 :
Water chemistry from 24-h assays exposing 30-to 60-day posthatch rainbow trout (Oncorhynchus mykiss) to wildland fire retardants (MVP-Fx, LC95A-R, and 259-Fx) Emerson et al. (1975)d temperature were measured in each concentration (n = 1) at the beginning of each assay.Un-ionized ammonia concentrations were calculated using equations fromEmerson et al. (1975).dup = a duplicate sample run to assess intersample variation in measurements.

TABLE 2 :
Lethal concentrations of wildland fire retardants to 30-to 60-day posthatch rainbow trout (Oncorhynchus mykiss; n = 10/concentration) 10%, and 5% of the sample at three selected time points (6, 12, and 24 h) for each fire retardant tested.Standard deviation is reported in parentheses.The lowest concentration different from the control treatment was determined using post hoc Tukey's honestly significant difference test (α = 0.05).The 96-h LC50 is reported from previously published data for comparison.LCx = x% lethal concentration; LOEC = lowest-observable-effect concentration.