Long-term fire retardants are used to prevent the spread of wildland fires. These products are normally applied by aircraft to create containment perimeters or long-term combustion barriers around the fire and are intended specifically for terrestrial application. Between the years 2000 and 2010, 36,148 aerial loads were dropped in the United States to combat wildfires, averaging over 31 million liters of fire retardant per year 1. The U.S. Forest Service attempts to avoid introduction of fire retardants into aquatic systems by using a 91.4-m buffer 1. Nevertheless, fire retardant chemicals may enter aquatic ecosystems that provide habitat for endangered and threatened species either directly through intentional applications within the buffer or unintentionally via accidental drop or surface runoff 2. In fact, between the years 2008 and 2010, there were 49 accidental fire retardant drops within the buffer, and another 54 fire retardant drops were granted exceptions to the buffer guideline by the U.S. Forest Service for fires that posed greater threat to life, property, or natural resources 1. Although data are limited, fire retardants have caused fish kills; for example, significant trout mortality occurred in the Fall River, Oregon, USA, as a result of misapplication while fighting a fire in 2002 2, 3.
During application, fire retardants are essentially 85% water and 10% inorganic salts (fertilizers), with 5% additives such as gum thickeners, coloring agents, corrosion inhibitors, stabilizers, and bactericidal agents 2. Older product formulations contained sodium ferrocyanide as a corrosion inhibitor, but were found to be severely and acutely toxic when exposed to ultraviolet radiation 3. The use of fire retardants containing sodium ferrocyanide was discontinued in 2007 1. Approved fire retardants contain diammonium phosphates (DAP; [NH4]2HPO4) and ammonium polyphosphate salts. The primary risk associated with their use is the potential for ammonia toxicity 3–6, which is also influenced by pH and temperature 7.
Twenty-eight Pacific salmonid Evolutionarily Significant Units (ESUs) in Washington, Oregon, Idaho, and California, USA, are currently listed as either threatened or endangered (www.nwr.noaa.gov/ESA-Salmon-Listings/upload/1-pgr-8-11.pdf). A stock of Pacific salmon is considered an ESU under the Endangered Species Act if it is reproductively isolated from other stocks and “represent[s] an important component in the evolutionary legacy of the species” 8. Each ESU is considered a distinct population segment, or species, under the Endangered Species Act 8. Studies have examined the toxicity of fire retardants on various life stages of salmonids. Among early life stages, swim-up fry and salmonids that are 60 to 90 d posthatch are more sensitive than eyed eggs 4, 6. However, no data are available regarding the sensitivity of fish to fire retardants undergoing the parr–smolt transformation. The parr–smolt transformation is an important period for salmonids wherein they undergo physiological and behavior changes in advance of transitioning to a marine environment 9, 10. Several of the physiological and behavior changes during the transformation (e.g., decreased total body lipids, increased oxygen consumption, changes in endocrine activity, increased ammonia production, and increased growth rate 11) may also increase fish stress. Not only have salmonids been found to be more sensitive to contaminants during the parr–smolt transformation relative to other life stages 12, 13 but the success of parr–smolt transformation can also be affected by exposure to contaminants at earlier life stages 14–16.
The objectives of the present study were to determine the acute toxicity and sublethal effects of two fire retardants, PHOS-CHEK 259F and LC-95A, on yearling chinook salmon (Oncorhynchus tshawytscha) at the smolt stage. These formulations are currently approved by the U. S. Forest Service for fighting wildland fires. Acute toxicity was determined in 96-h static exposures. Separately, histopathology examinations and analyses were completed on tissues collected from salmon exposed to fire retardants for up to 96-h under static conditions. Sublethal effects on seawater tolerance and disease susceptibility were assessed with separate challenges conducted on survivors of previous fire retardant exposures.
MATERIALS AND METHODS
Both PHOS-CHEK LC-95A and 259F (ICL Performance Products LP) are proprietary compounds provided by the U.S. Department of Agriculture Forest Service Wildland Fire Chemical Systems Program of the Missoula Technology Development Center (Missoula, MT, USA). Product information for both retardants is provided by the Forest Service and the manufacturer. In brief, PHOS-CHEK 259F is a diammonium phosphate-based dry powder formulation with a manufacturer-recommended mix ratio of 136.6 g per liter of water (1.14 lb per U.S. gallon). At the approved mix ratio, 259F has a specific weight of 1.066 kg/L (8.90 lb/gal), has a low viscosity of 75-259 cP, and is 10.9% (w/w) DAP, (NH4)2HPO4. The formulation contains a red fugitive dye agent that fades over time in sunlight. PHOS-CHEK LC-95A is an ammonium polyphosphate-based liquid concentrate formulation with a specific gravity of 1.473 kg/L (12.29 lbs/gal) and a manufacturer-recommended mix ratio of 5.5 parts concentrate to 1 part water. At the approved mix ratio, LC-95A has a specific weight of 1.074 kg/L (8.97 lb/gal), has a low viscosity of 75 to 225 cP, and is 7.6% (w/w) P2O5 equivalent.
Experimental animals and facility
Yearling stream-type chinook salmon (O. tshawytscha) were raised from embryos in the Fish Disease Laboratory at the National Oceanic and Atmospheric Administration's Newport Research Station (Newport, OR, USA). The Fish Disease Laboratory is supplied with two water sources: municipal freshwater that is dechlorinated by carbon filtration prior to use and seawater pumped from Yaquina River bay at high tide that is sand-filtered and UV-irradiated prior to use. The experimental fish represent the 2007 brood year at Rapid River Hatchery (Riggins, ID, USA), which would have been released to begin their outmigration to the ocean in February and March 2009 and would characteristically enter the ocean in May through June 2009. The 96-h PHOS-CHEK exposures and subsequent seawater and disease challenges commenced once the experimental stock fish had transitioned sufficiently from parr to smolts to ensure high survival when seawater was introduced. At the time of the fire retardant exposures, the mean fish weight was 12.2 g, and mean fork length was 105.8 mm.
The status of the physiological transition from parr to smolt of the laboratory-raised fish was monitored weekly from February 11 to April 16, 2009. Monitoring involved determining the Na+-K+-ATPase (hereafter ATPase) activity in the gills of 10 stock fish per week held in freshwater. From February 26 to April 16, 2009, an additional 10 stock fish per week were placed in full-strength seawater for 24 h. At the conclusion of the 24-h challenge, the numbers of survivors were tallied and compared with the previous week's challenge.
To determine ATPase activity, the first gill arch on the left side of each fish was collected and placed in a 0.5-ml microcentrifuge tube filled with SEI buffer (250 mM sucrose, 10 mM ethylenediaminetetraacetic acid [EDTA]-Na2, 50 mM imidazole, pH 7.3 17) and flash frozen in the gas phase of liquid nitrogen. Samples were then stored at −80°C and analyzed within six to eight months. Fish-gill ATPase activities were determined as the difference in the linear rate of ATP hydrolysis in the absence and presence of ouabain in micrometers ADP/mg protein/h as per McCormick 17. ATPase activities in the gill homogenates were corrected for protein content following the bicinchoninic acid (BCA) protein assay (Pierce; part of Thermo Fisher Scientific). Mean ATPase activities among sample groups (exposure concentration or weekly samples) were compared by one-way analysis of variance (ANOVA), followed by Tukey's honestly significant difference post hoc test in SYSTAT 12 (Systat Software), with significance level (α) set at 0.05.
Acute PHOS-CHEK exposures
Acute toxicity of either PHOS-CHEK 259F or LC-95A was evaluated by exposing yearling chinook salmon to eight concentrations of each formulation together with an unexposed control group (PHOS-CHEK control) for 96 h under static freshwater holding conditions (Table 1). The salmon were not fed during the exposure period. The exposures were initiated on April 23, 2009. Concentrations of the fire retardant were prepared on a PHOS-CHEK formulation mass-to-exposure volume basis (i.e., milligrams of PHOS-CHEK 259F dry powder or milligrams of PHOS-CHEK LC-95A liquid concentrate). For each PHOS-CHEK treatment, volumes of concentrated stock solutions, prepared with a magnetic mixer, were added to 30-L, polyethylene exposure vessels and thoroughly mixed with 25.5 L freshwater to achieve the desired exposure concentration immediately prior to the addition of fish. The concentrated fire retardant was manually mixed within the exposure vessel with a stirrer rod for roughly 10 to 20 s in each direction. Three replicate exposure vessels were used for each PHOS-CHEK treatment. The exposure vessels were placed within larger tanks with a continuously refreshed water jacket surrounding the tanks to maintain a uniform water temperature (∼11–12°C). Each of the tanks contained a porous air-stone with constant air supply. Once the exposure vessels had been thoroughly mixed, 25 fish were added to each of the vessels, with a mean loading density of approximately 12.0 g fish/L. During the 96-h exposure period, dead fish were removed from the exposure vessels at approximately 5, 24, 30, 48, 52, 72, and 78 h, with the exact time of collection recorded. The lengths and weights of all fish were obtained at the time of collection (mortalities) or conclusion of the saltwater exposure (survivors). The lethal concentration (LC) curves and estimates of the concentration at which 50% of the fish would experience mortality (LC50) when exposed to PHOS-CHEK 295F and LC-95A were determined by logistic regression in SYSTAT 12 at the conclusion of the 96-h acute exposure.
Table 1. Concentrations of PHOS-CHEK fire retardants used during the acute 96-h toxicity test and mean cumulative mortalities
Mean cumulative mortality observed after 96 h of exposure to PHOS-CHEK LC-95A in three replicate tanks per fire retardant dose. The median lethal concentration (LC50) calculated from logistic regression was 339.8 mg/L.
Mean cumulative mortality observed after 96 h of exposure to PHOS-CHEK 259F in three replicate tanks per fire retardant dose. The LC50 calculated from logistic regression was 140.5 mg/L.
Water quality measurements
Water quality was assessed during each of the 96-h static exposures from one randomly selected tank per replicate concentration at the onset of exposure and at 24-h intervals during the 96-h exposure. Water quality measurements included dissolved oxygen, temperature, combined nitrate and nitrite, pH, ammonia, and water hardness. Temperature and dissolved oxygen were determined with a multiprobe (YSI) suspended in the water column of individual tanks or exposure vessels. Individual 125-ml grab samples were used for the remaining parameters.
Nitrate and nitrite
Combined nitrate and nitrite (nitrate–nitrite) concentration in water samples was determined by first reducing all nitrate to nitrite. The combined nitrite concentration was then determined with the Griess reaction 18. All water samples were preserved with 250 µl concentrated sulfuric acid (95–98% H2SO4) and stored at 4°C prior to analysis 19. The samples were then serially filtered through glass fiber prefilters (Millipore and Corning), intermediate cellulose fiber filters (Whatman), and 0.45-µm HAWP final filters (Millipore) to remove colored PHOS-CHEK particles before reducing the sample nitrate to nitrite. As per method 4500-NO3-E of Eaton et al. 19, 25 ml filtered sample was mixed with 75 ml ammonium chloride-EDTA solution at pH 8.5 19 and passed through a column filled with copper–cadmium granules. The first 25 ml of the reduced sample to exit the column was discarded. The reduced nitrate plus the nitrite originally present in the sample (nitrate–nitrite) was then determined for each water sample following method 4500-NO3-E 19. In brief, a 50-ml aliquot of the reduced sample was incubated with 2 ml color reagent (sulfanilamide/N-[1–naphthyl]ethylenediamine dihydrochloride solution; Ricca Chemical). Samples were incubated at room temperature for 10 min to allow color development, followed by an absorbance reading on a Genesys 10 spectrophotometer at 543 nm (Thermo Fisher Scientific). The nitrate–nitrite concentration was calculated from a standard curve generated by the reduction of a serial dilution of a 0.01 M nitrate calibration standard (Sigma-Aldrich). To determine the maximum nitrite contribution of PHOS-CHEK formulations in the nitrate–nitrite measurements, nitrite was determined by method 4500-NO3-E of Eaton et al. 19, without nitrate reduction. The nitrite contribution analyses were conducted only for the highest PHOS-CHEK concentrations (1,132 mg LC-95A/L and 685 mg 259F/L) and controls and in the absence of fish.
The pH of water samples was determined by using a Ross Sure-Flow pH electrode 8172 (Thermo Fisher Scientific). Three standards (pH 4.0, 7.0, and 10.0; Fisher Scientific) were used to calibrate the electrode each day prior to reading samples. Samples were stored at 4°C for less than one week before analysis. Samples were brought to room temperature and mixed well, and 100 ml was aliquoted into a glass beaker on a stirrer plate.
Water samples were analyzed for ammonia by method 4500-NH3-D 19 using an Orion ion-selective ammonia electrode 9512HP (Thermo Fisher Scientific). Samples were preserved with 250 µl concentrated sulfuric acid (95–98% H2SO4) and stored at 4°C for less than 30 d before analysis. Five ammonium chloride standards (0.5, 5.0, 50.0, 100.0, and 200.0 mg/L) were made from a 0.1 M ammonium chloride standard solution (Thermo Fisher Scientific) and were used to calibrate the electrode each day prior to reading samples. Samples were brought to room temperature and mixed well, and a 100-ml aliquot was added to a glass beaker on a stirrer plate. Direct readings were taken after 2 ml of ionic strength adjustor (5 N sodium hydroxide; Thermo Fisher Scientific) had been added to the sample. The electrode was rinsed, and a duplicate reading was taken of each sample; the average of the two readings was recorded as the result. Un-ionized ammonia concentrations were then determined from tables presented by Thurston et al. 20 based on the calculations of Emerson et al. 21 using temperature, pH, and total ammonia measurements obtained from each tank.
Measurements of water hardness were attempted with a water hardness test kit (model AG-3, code 4766; LaMotte). However, the red dye in the fugitive PHOS-CHEK fire retardant formulations interfered with the colorimetric assay's ability to measure hardness. Consequently, hardness values could not be determined. Water hardness of the laboratory freshwater was 42 mg/L prior to any PHOS-CHEK addition.
Seawater challenge tests were conducted as a measure of smolt readiness for seawater entry 22. Immediately after the 96-h acute toxicity test, fish in exposure vessels that had 10 or more survivors were transferred to tanks (∼300 L) with full-strength, flow-through seawater (34 ppt) for 24 h. In addition, three replicates of 25 fish were transferred from a freshwater holding tank to individual flow-through seawater tanks to be included as naive fish that had not experienced 96 h of static holding. Immediately after the 24-h seawater challenge, the mortalities were recorded to assess survival to the challenge. Chi-square tests of association (Systat Software) were used to compare the mean survival among the controls and the PHOS-CHEK concentrations. In addition, up to five survivors per seawater exposure tank were necropsied for an evaluation of their parr–smolt development status by measuring the gill ATPase levels.
The impact of fire retardant exposure on chinook salmon disease susceptibility was evaluated with a disease challenge assay. On May 4, 2009, fish were exposed to sublethal concentrations of PHOS-CHEK 259F (34.3 and 13.7 mg/L), LC-95A (113.3 and 22.7 mg/L), and a nonchemical control (PHOS-CHEK control) in static, 96-h exposures. The two sublethal concentrations of fire retardant represented the lowest concentrations with no observed mortality during the 96-h acute exposure and a less than 1% lethal concentration of PHOS-CHEK LC-95A and 259F (LC0.01 and LC0.70, respectively) based on the logistic regression of the acute exposure data. For each PHOS-CHEK treatment, volumes of concentrated stock solutions were added to six replicate 100-L, polyethylene exposure vessels and thoroughly mixed with 60 L freshwater to achieve the desired exposure concentration immediately prior to the addition of fish. The 100-L tanks were held within larger tanks and surrounded by a continuously refreshed water jacket, as described above. Fifty fish were held in each of the exposure vessels for 96 h at a mean loading density of approximately 9.8 g fish/L. After the exposure, fish from each of the tanks were transferred to larger tanks with approximately 300 L flow-through freshwater for a 24-h recovery period. In addition, six replicates of 50 fish were transferred from a freshwater holding tank to individual flow-through freshwater tanks to be included as naive fish that had not experienced 96 h of static holding. Each tank was then transitioned to full-strength seawater (34 ppt) over 6 d. During this period, the proportion of seawater was incrementally increased. The fish remained on full-strength seawater for 1 d prior to the disease challenge.
On May 15, 2009, a disease challenge was performed as per Arkoosh et al. 23. Briefly, three of the six replicate tanks per PHOS-CHEK concentration, PHOS-CHEK control, and naive fish were exposed to 2.9 × 104 cfu/ml of Listonella anguillarum in trypticase soy broth (TSB) supplemented with 1.5% NaCl. The fish were exposed for 1 h in 38-L containers with 18.75 L static seawater and aeration at an approximate density of 50 g fish/L. The remaining three replicate tanks per treatment were used as no-pathogen controls and were exposed to sterile TSB supplemented with 1.5% NaCl for 1 h under identical challenge conditions. After the 1-h bath exposure, all the fish were returned to their original flow-through seawater tanks (∼300 L) for a 9-d observational period. Mortalities were collected twice daily (at ∼9:00 and ∼16:00) during the observation period. Approximately one fish was necropsied for every three dead fish (39.9% of mortalities necropsied) collected, to confirm that the deaths were associated with L. anguillarum infection. Immediately after the dead fish had been removed from the tank, they were sprayed with 75% ethanol, and the kidneys were surgically exposed and aseptically struck onto a trypticase soy agar (TSA) plate supplemented with 0.5% NaCl. If colonies formed on the TSA plates after 48 h at 26°C, L. anguillarum growth was confirmed with a Mono-Va agglutination test kit (Bionor, A/S). Of the mortalities that occurred in the pathogen-exposed tanks, 276 (or 92.3% of the fish struck) had growth on TSA, with 98.9% that agglutinated. Four of the mortalities that occurred in the no-pathogen controls (or 5.8%) had growth on TSA, with only one colony that agglutinated.
Survival curves were generated using the nonparametric Kaplan–Meier method across all replicate tanks for fish in each pathogen-exposed treatment. Survival curves were assessed for the potential impact of both fire retardant exposures on disease susceptibility and tank effects using the Mantel method for the log-rank chi-square test with the null hypothesis of a common survival curve. Any replicate tanks identified as outliers, by log-rank chi-square tests, within a common treatment were removed from later analysis. SYSTAT 12 was used to generate all survival curves and to perform log-rank chi-square tests, with the significance level (α) set at 0.05 in all comparisons.
To investigate pathological effects on specific tissues, additional stream-type chinook were exposed to PHOS-CHEK formulations. Fish were exposed to five different concentrations of PHOS-CHEK 259F and LC-95A and a nonchemical control (PHOS-CHEK control) in static exposures for periods up to 96 h (Table 2). The durations of exposures were selected based on acute toxicity results, with fish exposed to the three highest doses of 259F and LC-95A for no longer than 24 h because of the expected rapid mortality (data not shown). At 24-h intervals (and one 6-h sample point), 10 fish were sacrificed, and the gills and viscera were dissected from each fish and placed into separate tissue cassettes with Davidson's fixative (at a minimum volume of 1 part tissue to 20 parts Davidson's; 24). Tissues were fixed for a minimum of 48 h and then stored in 70% ethanol. Subsequent histological processing included paraffin infiltration and embedding, sectioning at 5 µm, and staining with hematoxylin and eosin reagents 25, 26. Gill, liver, endocrine and exocrine pancreas, esophagus, stomach, pyloric caeca, upper intestine, lower intestine, heart, anterior and posterior kidney, spleen, gonad, skin, and lateral line system from each fish were all evaluated for histopathological lesions and conditions that could result from fire retardant exposure. Differences in diagnosis prevalence across treatments for the same exposure duration were determined with the Fisher's exact test. Lesion severity was classified on an ordinal scale (0–7), and statistical comparisons were made by ANOVA and Fisher's protected least significant difference post hoc test, and α was set at 0.05 in all statistical comparisons of pathology results.
Table 2. Prevalence of gill lesions and conditions detected by histopathology in chinook salmon smolts exposed to PHOS-CHEK LC-95A and 259F compared with controls, for up to 96 h
Monitoring of gill ATPase and seawater survival indicated that the stock fish were smolting and thus were prepared for seawater entry at the time of PHOS-CHEK exposure (April 23). During the 10-week monitoring period, ATPase activity levels in fish gills significantly increased (Fig. 1). Each of the first four weeks of monitoring had significantly less mean ATPase activity than each of the last four weeks (p ≤ 0.001). Weeks 5 and 6 indicated a potential transitional period (Fig. 1). In weeks 5 and 6, ATPase activity levels were not significantly different from each other (p = 1.0), but the levels in each week were significantly greater than in each of the first four weeks (p ≤ 0.05) and significantly less than in each of the last four weeks (p ≤ 0.038). In addition, 70 to 100% of the fish survived weekly seawater challenges after week 3. The results are consistent with a stock population that is undergoing the physiological changes necessary for seawater entry. There is no fixed ATPase activity level that equates to seawater preparedness; rather, ATPase activity is generally low at the hatcheries, increases at the time when fish are released into the river system, and continues to increase as outmigration progresses 11.
Acute PHOS-CHEK exposures
Yearling chinook salmon had different sensitivities to acute exposures of the two PHOS-CHEK formulations. The lowest no-effect concentrations (no mortalities) were observed in tanks exposed to 34.3 and 115.5 mg/L 259F and LC-95A, respectively, with 100% mortality observed in tanks exposed to 274.0 and 566.0 mg/L 259F and LC-95A, respectively (Table 1). The logistic regression estimates of the LC50s for 259F and LC-95A were 140.5 mg/L and 339.8 mg/L, respectively. The PHOS-CHEK control and eight PHOS-CHEK doses for each of the formulations were well described by logistic regression (LOGIT) models. The predicted LOGIT models of survival probabilities following 96 h of exposure to 259F and LC-95A are presented in Figure 2.
High significance and classification strength of the models were indicated from various assessment metrics accompanying the logistic regression. The estimates of the parameters in both LOGIT models (constant and PHOS-CHEK concentration) were significant (p < 0.001). Two indicators of model fit (Cox and Snell R2 and Naglekerke's r2; both mimic r2 values for linear regression and are based on log likelihoods and sample size) had maximum values of 1.0 for each of the models (SYSTAT Software). A similar metric, McFadden's RHO2, had high values of 0.620 and 0.828 for the LOGIT models of the 259F and LC-95A data, respectively. Finally, prediction-success analysis indicated that, overall, 82.9% of the model predictions were correct for the 259F LOGIT model and 92.7% correct for the LC-95A model.
Water quality measurements
Exposure conditions were designed to limit variation of temperature and dissolved oxygen levels in different treatment tanks. The minimum optimal dissolved oxygen level for maintaining salmon health is 8.8 mg/L at 11.3°C 27. Each tank was supplied with constant aeration throughout the 96-h exposure period and resulted in an average of 10.6 mg/L dissolved oxygen. Dissolved oxygen ranged from 7.2 to 11.2 mg/L during individual measurements (data not shown). Average daily measurements of dissolved oxygen were lowest after the first 24 h (7.2–9.4 mg/L), but increased in the subsequent days. The temperature ranged from 11.2 to 11.6°C in most tanks, with two exceptions in which the water jacket temporarily malfunctioned. During the malfunction, one replicate of LC-95A (1132.1 mg/L) had a temperature reading of 12.2°C immediately before fish were added; and one replicate of LC-95A (340.8 mg/L) had a temperature reading of 13.6°C at the first sampling point (24 h). Neither of the temperature anomalies resulted in tank survivals that were different from their treatment replicates.
The mean concentration of nitrate–nitrite was 0.89 mg/L (range, 0.76–1.04 mg/L) in all PHOS-CHEK treatments prior to fish addition compared with 0.83 mg/L in the PHOS-CHEK control. The nitrite contribution was not detectable in laboratory freshwater or the highest concentration of 259F (685 mg/L). The nitrite contribution in the highest concentration of LC-95A (1132 mg/L) was just detectable at 0.01 mg/L. After fish addition and during exposure, the mean concentration of nitrate–nitrite increased slightly to 0.93 mg/L (ranging from 0.84 to 1.02 mg/L) in all PHOS-CHEK treatments and to 0.86 mg/L in the PHOS-CHEK control. Nitrite concentration less than 0.1 mg/L and nitrate levels less than 1.0 mg/L are considered safe for rearing salmon 27.
The pH of the water in tanks containing LC-95A ranged from 6.9 to 7.3, with no observable trend in dose or exposure time (Fig. 3A). In contrast, the pH of the water in tanks containing 259F ranged from 6.9 to 7.6 and seemed to increase slightly with increasing 259F concentration (Fig. 3B). The PHOS-CHEK control had an initial pH of 7.6 prior to the addition of fish but ranged from 6.9 to 7.2 thereafter.
In contrast to pH measurements, total ammonia in the water increased with increasing concentrations of both PHOS-CHEK formulations (Fig. 3C and D). Total ammonia also increased with exposure time in lower doses of 259F and LC-95A and in the PHOS-CHEK control. However, the majority of total ammonia was detected immediately after PHOS-CHEK addition. Given that total ammonia was not detected in the PHOS-CHEK control prior to fish addition, the steady increase most likely is due to accumulation of fish excreta during the static holding conditions. This increase with time was not observed with higher doses because of the high mortalities in these tanks, presumably reducing the source of ammonia.
Un-ionized ammonia concentrations were calculated based on the tank temperature, pH, and total ammonia detected in each of the tanks. Given the relatively constant temperature range for all tanks, variation in pH and total ammonia had the greatest impact on un-ionized ammonia concentrations. In the LC-95A treatments, the pH was relatively constant, whereas the total ammonia increased with increasing dose (Fig. 3A and C). In contrast, in the 259F treatments, pH and total ammonia increased with increasing dose (Fig. 3B and D). The result was a fairly uniform range of un-ionized ammonia concentrations across all LC-95A treatments (0.14–0.49 mg/L; Fig. 3E) and increasing un-ionized ammonia concentrations with increasing 259F treatments (0.10–2.84 mg/L; Fig. 3F).
Sublethal exposures to PHOS-CHEK 259F and LC-95A for 96 h affected the ability of yearling spring chinook salmon to survive a subsequent 24-h seawater challenge (Fig. 4). The 96-h static experimental design significantly reduced the seawater survival of salmon in the PHOS-CHEK control (31%; p < 0.001) relative to the naive salmon that had not experienced 96 h of static holding. High survival (97%) among the naive fish did indicate that the yearling chinook were prepared for seawater entry at the time of the challenge.
Given the difference in survival among the naive fish and fish in the PHOS-CHEK control group, logistic regression of PHOS-CHEK concentration and survival was not possible. However, chi-square statistical analysis of the seawater survival did indicate significant differences between the PHOS-CHEK control and chinook salmon with sublethal exposure to the PHOS-CHEK formulations. All three concentrations of PHOS-CHEK LC-95A with sufficient survivors from the 96-h exposure to participate in the subsequent seawater challenge exhibited reduced survival after challenge in seawater (Fig. 4A). Fish from the lowest LC-95A treatment (115.5 mg/L) experienced significantly reduced survival (35%; p < 0.001) relative to PHOS-CHEK control (p < 0.001). The 340.8 mg/L treatment was nearly equal to the estimated LC50 value for LC-95A exposure, and fish in this treatment had significantly lower survival (61%; p < 0.001) than the PHOS-CHEK control during the seawater challenge.
The 68.5 mg/L treatment of 259F was the lowest concentration tested that was significantly different from the PHOS-CHEK control (p < 0.001), and fish experienced a 42% decrease in survival relative to the PHOS-CHEK control (Fig. 4B). The mean survival of fish during the seawater challenge that had previously been exposed to 102.8 mg/L 259F was not significantly different from that of the PHOS-CHEK control. Fish in treatments bracketing the estimated 259F LC50 (i.e., 137.0 and 164.4 mg/L) had 54 and 63% lower survival than the PHOS-CHEK control during the seawater challenge.
Impact of fire retardant exposure on parr–smolt development indicators
Previous PHOS-CHEK exposure did not have any detectable effect on the gill ATPase activity levels measured in the survivors of the acute 96-h exposure test and seawater challenge (data not shown). Although not statistically significant, ATPase activities were greatest in naive fish and fish from the PHOS-CHEK controls as well as from the lowest PHOS-CHEK 259F dose (34.25 mg/L). Small sample size and large standard errors might have contributed to our inability to determine statistically significant trends or differences.
Yearling chinook salmon exposed to sublethal doses of PHOS-CHEK fire retardants were not found to be more susceptible to disease from L. anguillarum. The fish were exposed to PHOS-CHEK 259F doses with LC values of 1.5% (34.3 mg/L) and 0.7% (13.7 mg/L) and LC-95A doses with LC values of 0.16% (113.3 mg/L) and 0.01% (22.7 mg/L) based on their logistic models. No mortalities were observed in any of the treatments during the 96-h sublethal fire retardant exposures or the gradual transition to full-strength seawater. The survival rates of fish in the PHOS-CHEK controls and fish with the lowest dose of LC-95A (22.7 mg/L), as well as the naive fish, were not significantly different from each other following a bath challenge with L. anguillarum (p ≥ 0.900); but fish from these treatments had significantly reduced survival relative to fish previously exposed to 113.3 mg/L of LC-95A (p ≤ 0.003; Fig. 5A). The survival rates of salmon exposed to the two PHOS-CHEK 259F doses were not statistically different from each other (p = 0.428; Fig. 5B). However, the survival rates of naive and PHOS-CHEK control fish were significantly reduced relative to fish exposed to either dose of 259F (13.7 and 34.3 mg/L; p ≤ 0.013).
Histopathological examination of chinook salmon indicated that gills were the principal tissues affected by increasing concentrations of PHOS-CHEK LC-95A and 259F. Full examination of other tissues (liver, endocrine and exocrine pancreas, esophagus, stomach, pyloric caeca, upper intestine, lower intestine, heart, anterior and posterior kidney, spleen, gonad, skin, and lateral line system) revealed no significant lesions that could be related to exposure to either of the fire retardants (data not shown). Forms of gill pathology included (1) frank respiratory epithelial necrosis; (2) respiratory epithelial hyperplasia, exfoliation, and hypertrophy; lamellar microaneurysms; and (3) phagosomes containing basophilic material in both respiratory epithelium and macrophages. Respiratory epithelial hyperplasia and lamellar microaneurysms were not attributed to PHOS-CHEK exposure because they were sporadically detected or equally detected in controls. Gill tissues from the PHOS-CHEK controls were healthy (Fig. 6A), with no other lesions or conditions observed at any sampling time (Table 2).
Respiratory epithelial necrosis, epithelial exfoliation, and hypertrophy
Frank respiratory epithelial necrosis was exhibited in one fish with the 206 mg/L dose of 259F. However, the concomitant occurrence of respiratory epithelial exfoliation and epithelial hypertrophy prior to mortality at 6 h suggests that these conditions are potential precursors of necrosis (259F-206 mg/L; Table 2). Respiratory lifting and exfoliation (Fig. 6B) occurred at 30 and 20% prevalence in the 206 mg/L dose, and respiratory epithelial hypertrophy was 40 and 30% after 6 and 24 h of exposure, respectively (Table 2). Similarly, the next lower dose of 259F (137 mg/L) showed a 40% prevalence of respiratory epithelial exfoliation and 20% prevalence of respiratory epithelial hypertrophy after 24 h of exposure. After 96 h of exposure, the two lowest doses of 259F (14.0 and 41.0 mg/L) also exhibited respiratory epithelial exfoliation and epithelial hypertrophy at 10 and 40% prevalence, respectively.
In contrast, the prevalence of respiratory epithelial exfoliation and epithelial hypertrophy was low in fish exposed to PHOS-CHEK LC-95A and occurred only after 72 h of exposure (Table 2). These lesions were prevalent (10% and 20%, respectively) with the 23.0 mg/L dose of LC-95A after 96 h of exposure. In addition, both lesions were prevalent in 10% of the fish exposed to 113.0 mg/L of LC-95A at individual sampling times prior to 96-h. No fish receiving the three highest doses of LC-95A (227.0–453.0 mg/L) were diagnosed with respiratory epithelial exfoliation or hypertrophy after 24 h of exposure (Table 2).
Phagosomes in respiratory epithelium and macrophages
Fish exposed to both PHOS-CHEK formulations exhibited phagosomes containing basophilic material (potentially PHOS-CHECK formulation) in the respiratory epithelium and in subepithelial macrophages after 24 h of exposure. The variably sized phagosomes or cytoplasmic lysosomes were present in both respiratory epithelium and subepithelial macrophages in the gill filaments and gill lamellae, most commonly between adjacent gill filaments, indicating intracellular uptake, phagocytosis, and digestion of the fire retardant compounds or dyes (illustrated in Fig. 6C). The prevalence of this condition showed a dose–response behavior, with 100% of the fish diagnosed at the highest doses of LC-95A (453.0 mg/L) and 259F (206.0 mg/L), progressively decreasing to 10% and 0% at 227.0 mg/L and 103.0 mg/L of LC-95A and 259F, respectively (Table 2). No fish among the controls or receiving lower doses of 259F or LC-95A were diagnosed with this condition during any of the exposure periods.
Our results indicate that two current-use fire retardants, PHOS-CHEK 259F and LC-95A, are acutely toxic to smolting chinook salmon at concentrations below suggested mix ratios. According to the manufacturer's mix ratios, the suggested working concentrations of LC-95A and 259F are roughly 226,590 and 136,600 mg/L, respectively. These are the concentrations that would initially enter a water body during an accidental drop, prior to any dilution. The estimated 96-h LC50 concentrations for yearling chinook salmon were 339.8 mg/L and 140.5 mg/L for LC-95A and 259F, respectively, or roughly 0.15 and 0.10% of the suggested working concentrations for LC-95A and 259F, respectively. In other words, misapplied LC-95A or 259F fire retardant must be diluted roughly 670 or 1,000 times, respectively, from its working concentration to reach the observed LC50 concentration and roughly 2,000 or 4,000 times, respectively, to reach the no-mortality concentrations observed in the present study. The level of actual mortality and dilution that could occur in the environment would depend on the physical characteristics (e.g., water volume and flow rate) and biological characteristics (e.g., species density, movement) of the water body that received the accidental drop, as well as on the volume of fire retardant received. Consider a scenario in which 10% of a full single-engine tanker (i.e., 300 L) was accidentally dropped into a small rectangular river channel with a depth of 1.5 m and width of 3 m. In this simplified example, 65 m of river channel would be needed to dilute 259F to its LC50 concentration, or 270 meters of river channel would be needed to dilute 259F to the no-mortality concentration observed in the present study, assuming an instant and homogeneous dilution. The LC50 of LC-95A was roughly 2.5 times greater than 259F and requires diluting with roughly two-thirds the water volume of 259F to achieve LC50 concentrations if applied at the manufacturer-suggested mix ratios.
Water quality measurements
Greater un-ionized ammonia concentrations in the 259F treatments may have resulted in more ammonia-associated mortalities with the 259F PHOS-CHEK formulation than with the LC-95A formulation. Once fire retardants are introduced to water, the ammonia within the formulation equilibrates between ionized and un-ionized forms depending on water pH and temperature 2. The un-ionized ammonia form is more toxic to fish 28. Increased pH with increasing 259F concentrations resulted in a greater ratio of un-ionized ammonia to total ammonia than in the LC-95A treatments. Acute toxicity tests of ammonia on juvenile chinook salmon performed by other researchers determined LC50 values ranging from 0.399 to 0.476 for water pHs between 7.8 and 8.0 29, 30. The range of these LC50 values is indicated in Figure 3E and F with a hatched box. The hatched box clearly indicates that the majority of LC-95A treatments had un-ionized ammonia concentrations below the ammonia LC50 value, whereas half of the 259F treatments had un-ionized ammonia concentrations at or above the ammonia LC50 values. Two exposure concentrations (164.4 and 205.5 mg/L) of 259F had un-ionized ammonia concentrations that overlapped the unionized ammonia LC50 values during the 96-h exposure and had 66 and 98% mortality, respectively. This close relationship between reported ammonia toxicity and observed 259F toxicity suggests that ammonia significantly contributed to the observed mortality.
In contrast, an additional factor contributed to mortality of fish exposed to PHOS-CHEK LC-95A. Un-ionized ammonia was below the LC50 value for all but the highest LC-95A dose (1,132 mg/L), but 96% mortality was observed during the 96-h exposure period at doses greater than or equal to 453 mg/L. At the estimated LC50 value of LC-95A, the maximum un-ionized ammonia concentration was 0.28 mg/L.
Gill tissues were the only tissues identified with histopathological lesions and conditions that could be attributed to fire retardant exposure. Frank respiratory epithelial necrosis was observed in one fish at the highest 259F concentration, but the prevalence pattern of respiratory epithelial hypertrophy and exfoliation suggests that they are precursor lesions to necrosis. These diagnoses may be due to ammonia present in the PHOS-CHEK formulations and waste produced during the exposure. Ammonia has been shown to cause gill damage such as hypertrophy of gill epithelium as well as gill exfoliation in rainbow trout 31, 32. Respiratory epithelial exfoliation and epithelial hypertrophy were associated primarily with 259F toxicity, because of their rapid occurrence at the high doses, where 50 to 100% mortality was observed during acute exposure, as well as their occurrence within 96 h in fish exposed to low doses. These conditions may also be related to exposure to LC-95A, but only at later exposure periods. Epithelial lifting is considered to be an initial gill response to exposure to a number of contaminants 33 and indicative of the edema component of the acute inflammatory response 34.
Exposure to either 259F or LC-95A at concentrations equal to or greater than 137 and 227 mg/L, respectively, resulted in the appearance of variably sized phagosomes or cytoplasmic lysosomes in the respiratory epithelium and in subepithelial macrophages. The presence of phagosomes indicated intracellular uptake, phagocytosis, and digestion of the fire retardant compounds or dyes. No fish among the controls or with lower doses of 259F or LC-95A were diagnosed with phagosomes during any of the exposure periods. The presence of phagosomes suggests that an additional factor contributed, or was a precursor, to PHOS-CHEK toxicity. At their LC50 concentrations, 90% of the fish exposed to LC-95A were diagnosed with phagosomes, compared with 20% of the fish exposed to 259F. The increased occurrence of phagosomes suggests a differential source of toxicity in LC-95A relative to 259F.
In addition to acute mortality, 96-h of PHOS-CHEK LC-95A and 259F exposure resulted in delayed effects to smolting chinook salmon. Chinook salmon smolts that experienced less than 5% mortality during the 96-h PHOS-CHEK exposure had significantly reduced survival (35–40%) during the subsequent seawater challenge, compared with controls. Seawater challenge results indicated a general sensitivity to subsequent stress and a specific sensitivity to seawater. No disruptions to specific parr–smolt indicators were found, suggesting that the seawater sensitivity might have been due to physical damage during PHOS-CHEK exposure. The histopathological changes observed in the gills after exposure to fire retardants might have the potential to cause reduced oxygen diffusion capacity of the gills, ultimately resulting in reduced survival 32.
Both PHOS-CHEK LC-95A and 259F exposure did not adversely affect the juvenile chinook salmon's disease susceptibility during challenge with L. anguillarum. In fact, previous exposure to 259F resulted in significantly increased survival relative to the PHOS-CHEK control. Gills are a critical portal of entry for the L. anguillarum pathogen 35. Damaged gill tissue, possibly from high ammonia levels in the fire retardant exposures, might have contributed to reduced L. anguillarum infections in fish exposed to the fire retardants. Morris et al. 36 demonstrated that survival of juvenile Lost River suckers (Deltistes luxatus) challenged with Flavobacterium columnare increased with prior exposure to increasing concentrations of un-ionized ammonia (0.125, 0.25, and 0.5 mg NH3/L). However, ammonia can be an immunosuppressant for pathogens that do not rely on gill entry for pathogenesis or if the gills are bypassed during exposure. For example, Saprolengia parasitica is a pathogen that infects and breaks down the superficial tissue of the fish 37, 38. Rainbow trout exposed to S. parasitica, following a 1- or 10-d exposure to un-ionized ammonia (0.50 or 0.05 mg/L, respectively), were more susceptible 37, 38. Finally, L. anguillarum mortality was significantly higher in chinook salmon pre-exposed to 10 mg/L total ammonia nitrogen when the pathogen was injected 39.
This is the first study to determine the sensitivity of chinook salmon to fire retardant chemicals at the peak of the parr–smolt transition. In summary, PHOS-CHEK 259F and LC-95A formulations were acutely toxic to chinook salmon physiologically prepared for smoltification at doses significantly diluted from manufacturer-suggested mix ratios and likely to occur during misapplication in salmon habitat. Full strength PHOS-CHEK 259F or LC-95A would have to be diluted 4,000 or 2,000 times, respectively, in order to reach the concentrations at which no mortalities were observed in the present study. Chinook salmon engaged in the parr–smolt transition were more sensitive to PHOS-CHEK 259F than LC-95A. The sensitivity of earlier chinook salmon life stages to 259F and LC-95A is not known. In LC50 comparisons with other salmonids, chinook salmon exposed at the smolt stage were similarly or slightly more sensitive to 259F than swim-up rainbow trout (148 mg/L 1, 168 mg/L 5, 94–165 mg/L 5) and coho salmon (170–250 mg/L 5). Chinook salmon do appear to be more sensitive to LC-95A exposure at the smolt stage than rainbow trout (435 mg/L 1) at 60 d posthatch.
Exposed salmon were diagnosed with histopathological lesions or conditions in gill tissues that were attributed to PHOS-CHEK exposure. Gills are the respiratory organs of fish and also represent the primary site of osmoregulation and ammonia excretion in anadromous salmon 40. Ammonia may be significantly contributing to the toxicity of 259F, based on un-ionized ammonia concentrations at doses near the 259F LC50 and the histopathology typically associated with ammonia exposure. However, some additional source of mortality likely exists in LC-95A. The increased presence of phagosomes in the respiratory epithelium and subepithelial macrophages in the gill filaments and lamellae of fish exposed to LC-95A relative to 259F did suggest a differential mode of toxicity. A further possibility is an adverse reaction to unknown (proprietary) compounds within the fire retardants. The PHOS-CHEK exposure concentrations that did not result in acute mortality were found to impair seawater survival significantly but not susceptibility to L. anguillarum. Although L. anguillarum infection may have been reduced by damaged gill tissues, ammonia toxicity can increase susceptibility to L. anguillarum when the gills are not the portal of entry 39. Ammonia exposure can also increase the susceptibility of fish to other pathogens 37, 38. The same gill damage that might have prevented pathogen entry may have also impaired chloride cells in the gills. Chloride cells have been shown to be sites of critical molecular functions during changes in salinity 14, 16. Reduced osmoregulation by the gills may explain the observed inability of fire-retardant-exposed salmon to cope with immediate seawater entry.
The cumulative adverse impact of fire retardants on chinook salmon abundance includes not only the acute mortality immediately following a misapplication but also the delayed mortality once the exposed salmon enter seawater. Movement from freshwater rearing environments to seawater is critical in the life cycle of Pacific salmon. However, this active outmigration to the sea takes time. Delayed mortality in seawater was observed immediately after exposure to the PHOS-CHEK formulations. The permanence of this seawater sensitivity or time to recovery from previous PHOS-CHEK exposure is currently unknown.
We appreciate the technical efforts of D. Boylen, J. Osborn, and G. Hutchinson. We are also grateful for the insights provided by E. Little, U.S. Geological Survey, Columbia, Missouri, and the thoughtful comments of N. Scholz, K. Peck-Miller, and K. MacNeal of the National Marine Fisheries Service during manuscript review. The present study was financially supported by the U.S. Department of Agriculture-Forest Service Wildland Fire Chemical Systems Program.