Skip to main content
NCBI home page
ACS AuthorChoice logoLink to ACS AuthorChoice
. 2025 Jun 7;59(23):11853–11861. doi: 10.1021/acs.est.5c04555

Lasting Impacts of Wildfire Retardant on Phosphorus Storage and Downstream Transport in a Rocky Mountain Stream

Madelyn M Lux †, Uldis Silins †,*, Michael J Wagner ‡, Eamon R Turner †, J Jeremy Fitzpatrick †, Erin A Cherlet †, Kirk Hawthorn §, Craig Harriott §, Micheal Stone ∥, Kathleen M Beamish †, Monica B Emelko ⊥
PMCID: PMC12177916  PMID: 40481771

Abstract

Aerial application of fire retardants is a critical tool in wildland fire suppression but can impact aquatic ecosystems if accidentally misapplied into watercourses. However, exceedingly few studies have documented actual water quality impacts of such misapplications. Here, we explore the short- and longer-term effects of an operational misapplication of PHOS-CHEK LC95A into a Rocky Mountain stream on (a) water quality, (b) streambed storage and release potential of soluble reactive phosphorus (SRP), (c) its downstream transport, and (d) effects on periphyton productivity. Two weeks after the retardant drop, streambed SRP and its potential release into the stream were 2.6–3.0 times greater (p < 0.014) at the drop site than at a reference site above the drop, yet the greatest increases in aqueous stream SRP occurred 1 km downstream of the drop. One year later, while streambed SRP at the drop site declined by 34%, downstream transport of retardant residues and streambed-bound SRP strongly increased streambed and stream SRP 1.5–6.1 km below the retardant drop. The lasting effects of the retardant observed in this study reinforce the importance of continued refinement of air tanker operational guidelines and enhanced operational attention to protecting critical watersheds during fire suppression operations.

Keywords: PHOS-CHEK LC95A, streambed, soluble reactive phosphorus (SRP), equilibrium phosphorus concentration (EPC0)

graphic file with name es5c04555_0007.jpg

graphic file with name es5c04555_0005.jpg

1. Introduction

Climate-driven increases in active wildfire season length and occurrence-duration of severe burning conditions − are well understood as causal factors driving increased area burned and extreme fire behavior. , This produces more challenging and costly wildfire suppression efforts in fire-prone regions worldwide. , For example, while a growing series of record-breaking fire seasons in western Canada over the past 20 years have increasingly challenged fire suppression capabilities, the extreme 2023 fire season burned over 2× greater area than the previous national record, stretching suppression capabilities across the country. Similarly, federal wildfire suppression expenditures in the United States have increased by 6.3-fold between 1985-1994 and 2013-2022 periods. Aerial application of chemical fire retardants by fixed-wing air tankers or rotary-wing aircraft is a critical wildfire suppression tool, particularly where rugged terrain or intense burning conditions can threaten the safety of ground-based fire suppression personnel. Fire retardants are often used in locations of notable human or ecosystem values at risk in front of fire perimeters to constrain rapid fire expansion in areas of high fire intensity to enable ground-based suppression efforts to be more effective. Increased contemporary occurrence of intense fire behavior has led to increased retardant use. Between 2000-2010 and 2012-2018, the average number of aerial retardant drops and total retardant volume applied have both increased by 66% across the United States, with proportionally greater increases in the number of retardant drops (159%) and total retardant volume applied (74%) in more fire-prone western regions over this same period.

Firefighting chemicals commonly used in wildfire suppression include both long-term retardants, typically fertilizer salts (ammonium phosphates or ammonium sulfates) often containing thickeners, coloring agents, corrosion inhibitors, and stabilizers that form a lasting combustion barrier on fuels after water has evaporated, and low-viscosity short-term retardants (foams containing proprietary wetting agents or surfactants) with diverse formulations that penetrate fuels to limit combustion for several hours after application until the water carrier evaporates , (see refs − for additional background on retardant formulations). Environmental concerns surrounding retardant use include potential adverse impacts on vegetation and soils, with a particular focus on potential impacts to aquatic biota where such concerns have led to litigation over fire management agencies use of retardants in the past two decades. Furthermore, the incremental or compound impacts of retardants on drinking water supplies over the effects of wildfire alone , remain uncertain. While most wildfire agencies closely avoid aerial retardant application near watercourses, accidental misapplications to watercourses periodically occur. , Small streams situated on steep terrain or in canyons may be more susceptible to misapplication because retardants may be dropped from greater altitudes, making precise drop zone control more challenging and increasing the potential for wind drift of retardants.

Accidental retardant misapplication into water bodies can produce acute and sublethal toxic effects in aquatic organisms where primary concerns have focused on dissolution of retardant constituents to ammonia (NH3) or other associated compounds. While numerous artificial stream and tank studies have documented initial toxicity of both short- and long-term retardants on fish and other aquatic organisms, − longer-term impacts on aquatic organisms remain uncertain. Occurrence of toxic concentrations and their downstream transport depends on stream size in relation to the volume of the retardant applied and downstream mixing dynamics. Experimental aerial retardant application across five variable-sized streams in western U.S.A. showed greatest NH3 concentrations immediately downstream of the point of application, diminishing rapidly within several hours because of dilution with upstream water and dispersion at increasing downstream distance from the retardant application, though initial short-term changes in water quality were detectable in one stream 800 m downstream. Several authors concluded that retardants were likely to produce relatively short-lived (<8–12 h) impacts on water quality with shorter potential toxic exposure durations for organisms as retardant plumes travel and disperse downstream.

Despite growing concerns over retardant use with increased fire activity, surprisingly, few studies have directly documented the short- or longer-term water quality or ecological impacts from fire retardants at stream scales after either controlled field-scale experimental application , or accidental misapplication during fire suppression operations. As a consequence, fire management agencies resort to estimating the downstream fate and toxicology of watercourse retardant misapplications using mixing models of variable resolution. Phosphorus (P) is a key constituent of fire retardants; however, there is a paucity of research on impacts from this component of retardants despite its key role in regulating aquatic productivity, , particularly in nutrient-poor (oligotrophic) Rocky Mountain streams. Because fine grained sediments (<2 mm) are the primary factor governing storage, transport, and geochemical cycling of P in rivers, , it is likely that phosphate in long-term retardants would be stored over longer periods of time and transported over longer distances with seasonally variable streamflow conditions regulating streambed erosion and subsequent downstream deposition processes. This would presumably extend the potential longevity of retardant impacts in lotic systems but has not been explored.

On August 15, 2020, an accidental misapplication of a long-term fire retardant (PHOS-CHEK LC95A; ammonium polyphosphate) directly into North Racehorse Creek (Alberta, Canada) occurred during fire suppression operations on a small (0.034 ha; ∼20 m diameter), low-intensity, surface fire near the creek caused by an improperly extinguished, backcountry campfire during a period of high wildfire hazard. Suppression operations consisted of two fixed-wing air tanker groups each loaded with water or retardant and a rotary-wing helitack (HAC) crew, where one air tanker delivered two water drops (3,978 L each) blanketing the fire, and the second tanker delivered three retardant drops (2838 L each) to the flanks of the fire situated in a steeply incised section of the valley. An unknown portion of one retardant drop landed directly in the stream along a 70–90 m reach (Figure ); the specific location the other two retardant drops were not recorded. Because this small, low-intensity surface fire (situated on a flat bench 60 m above the stream) only partially burned the forest floor, it would be unlikely to cause detectable fire impacts to streamwater quality. This provided a unique opportunity to document the short- and longer-term impacts of a fertilizer-based long-term retardant on this oligotrophic Rocky Mountain stream in the absence of meaningful compound retardant and wildfire effects on water quality.

1.

1

Open in a new tab

Photograph of North Racehorse Creek immediately below the drop site shortly after the accidental retardant drop, August 15, 2020 (photo credit Government of Alberta).

Accordingly, the primary objectives of this study were to document both the short- and longer-term impacts of accidental misapplication of fertilizer-based retardants on (a) aqueous water quality, (b) sediment storage and aqueous exchange of soluble reactive phosphorus (SRP) both at the site of accidental application and at increasing downstream distances including below its confluence with another unimpacted stream, and (c) to evaluate if initial and longer-term changes in water quality produced lasting effects on periphyton productivity at the application site. The study was designed to examine potential phosphorus storage and exchange mechanisms that would regulate the longevity of retardant impacts based on the notion that if impacts at the misapplication (drop) site and further downstream were substantial, they would be detectable with modest replication.

2. Materials and Methods

2.1. Study Area and Sampling Approach

The accidental retardant drop occurred in North Racehorse Creek, Alberta, Canada (49.82731°N, −114.52949°W). The study area included a 1.36 km section of North Racehorse Creek above and below the drop site, extending 4.94 km further downstream on Racehorse Creek below the confluence of South and North Racehorse Creeks (Figure ), which both originate on Alberta’s southwest Rocky Mountain continental divide. The region experiences large winter snowpacks where the flow regime is strongly regulated by seasonal snowmelt. The study area is in the upper Montane natural subregion, where forests are dominated by Lodgepole pine (Pinus contorta) with Engelmann spruce (Picea engelmannii), and Subalpine fir (Abies lasiocarpa) at higher elevations. Mean monthly air temperatures range from −6.6 °C (Jan) to 13.6 °C (Jan) with a mean annual precipitation of 643 mm (40% as snow).

2.

2

Open in a new tab

Map of study area, sampling sites, Meteorological Service of Canada (MSC) and Water Survey of Canada (WSC) climate, and hydrometric stations. Flow direction is from west (left) to east. The retardant was dropped into N. Racehorse Ck. at site 2. A small black dot south of site 2 indicates the fire CWF-114 location and size.

A series of six sample sites were established in September 2020 on North Racehorse Creek from 180 m upstream of the retardant drop site, downstream to its confluence with South Racehorse Creek, and 4.94 km further downstream on Racehorse Creek below the confluence (Figure and Table ). Sampling included three sites on North Racehorse Creek (a reference site 180 m above the dropsite 1, the retardant dropsite 2, and 1 km below the dropsite 3) and two additional retardant-affected sites further downstream on Racehorse Creek (1.5 km below the drop sitesite 5 and 6.1 km below the drop sitesite 6). An additional sampling site (site 4) was established on South Racehorse Creek 200 m upstream of the confluence of North and South Racehorse Creeks to serve as a reference for downstream retardant-affected sites on Racehorse Creek (sites 5 and 6). All sites were sampled on September 1–2, 2020 (2 weeks after the retardant drop) and September 21–22, 2021 (1 year after the retardant drop).

1. Sample Site Characteristics in North Racehorse Creek above and on Racehorse Creek below the Confluence of North and South Racehorse Creeks (Site Numbers Correspond to the Locations Shown in Figure ).

site elevation (m.a.s.l.) watershed area (km2)
(1) N. Racehorse Ck. reference 1572 45.1
(2) retardant drop 1565 45.4
(3) 1 km below the drop 1536 46.0
(4) Racehorse Ck. reference 1535 70.5
(5) 1.5 km below the drop 1530 116.7
(6) 6.1 km below the drop 1490 127.9
Open in a new tab

2.2. Streamwater Quality and Discharge

Depth-integrated streamwater samples were collected manually at each site in 2020 and 2021 in the centroid of flow in single-use, acid-washed (10% HCL), triple-rinsed, high-density polyethylene bottles. Samples were transported in a cooler and stored at 4 °C before transport (within 2 days) to the University of Alberta Biogeochemical Analytical Laboratory for analysis, including total suspended solids (TSS), soluble reactive phosphorus (SRP), and total phosphorus (TP). Turbidity, pH, magnesium (Mg2+), calcium (Ca2+), dissolved organic carbon (DOC), nitrate–nitrite (NO2 – + NO3 –), ammonia–ammonium (NH3 + NH4 +), and total nitrogen (TN) were also measured as other parameters potentially affected by the retardant. Analytical methods are outlined in Table S1. Stream discharge was measured concurrently with water quality sampling to calculate SRP export as the product of the SRP concentration and stream discharge. Discharge was measured using standard area–velocity techniques with a SonTek FlowTracker 2 (YSI inc.) current meter.

2.3. Streambed Sediments

Streambed sediments were sampled at each site in 2020 and 2021 with freeze-core techniques to collect undisturbed streambed samples while avoiding elutriation of fine sediments during sampling. Three replicate freeze-core samples were collected near each sampling site from pool or stream margin backwater locations where slower flow conditions would enable retardant intrusion into streambeds. The sampler was driven 15 cm into the streambed, where compressed gas was passed through the cooling unit for 2 min to freeze a 1–2.5 cm thick layer of interstitial water and sediment on the outside of the sampler. The frozen sediment plug was withdrawn from the streambed, thawed into sealed plastic bags in a cooler for transport, and then refrozen for transport within 6–10 h (Figure S1). Frozen sediment plugs were thawed, air-dried, and sieved (Ro-Tap, Model Rx-29, ASTM #140 sieve, <106 μm diameter) to separate the fine, most geochemically active sediment fractions for subsequent analysis. Previous characterization of fine streambed sediments from South Racehorse Creek ∼3 km upstream of site 4 showed a median particle size of 33 μm (D50) with 75% of sieved bed sediments <65 μm.

2.4. Sediment–Phosphorus Exchange

The equilibrium phosphorus concentration (EPC0) is used to describe the potential of streambed sediments to adsorb or release SRP to streams depending on the ambient stream SRP concentrations and is a useful indicator of SRP storage in bed sediments. ,, EPC0 is determined experimentally by plotting mass of P sorbed per mass of sediment compared to the initial concentration of SRP prior to contact with the sediment where negative values indicate P release from sediments to the water column, while positive values indicate sorption of P from the water column to sediments. EPC0 is the aqueous SRP concentration at which no sorption or release with bed sediments occurs. It is also an indicator of the potential for P exchange between the streambed and the water column boundary under variable ambient SRP conditions. ,− Batch equilibrium experiments were conducted to determine the EPC0 of streambed sediments from replicate freeze-core samples from the six sites. For each subsample, a five-point isotherm was established by mixing 0.14 g of the <106 μm diameter dried sediment fraction with 14 mL of deionized water and potassium dihydrogen phosphate to produce five increasing P concentrations (0, 25, 50, 100, and 125 μg L–1 KH2PO4) in 15 mL polypropylene centrifuge tubes. Sorption experiments were conducted over 24 h at room temperature on an orbital shaker table (50 rpm) and then centrifuged (5000 rpm for 10 min), after which an aliquot of the supernatant was analyzed for SRP concentration using a Thermo Gallery Plus Beermaster auto analyzer using the molybdenum blue method. The phosphorus release potential (PRP) of streambed sediments (y-axis of linear sorption isotherms) was estimated using ambient stream SRP concentrations (x-axis) in 2020 and 2021. It should be noted that while sorption isotherms were measured using deionized water, EPC0 and the streambed PRP estimates may be variably overestimated from those under field conditions due to seasonal variation in both stream temperature and potential influence of varying competitor ions in native streamwater; our estimates of these variables under standard conditions serve as indicators of the relative effects of the retardant on these variables across the six sites in 2020 and 2021.

2.5. Periphyton Productivity

Floating or benthic periphyton growth was measured by deploying artificial substrate tiles (Plexiglas or porcelain tiles for floating and benthic periphyton, respectively) where periphyton was scraped from tiles into opaque plastic scintillation vials after 4–6 weeks of colonization and growth. Total periphyton growth was determined as ash-free dry mass (AFDM) and chlorophyll a. Triplicate floating and benthic tiles were deployed and sampled in midstream riffles at the retardant drop (site 2) and the North Racehorse Creek reference (site 1) in September–November 2020 and again in May–October 2021.

2.6. Statistical Analysis

Retardant impacts on streambed sediment EPC0 and PRP were evaluated separately for North Racehorse and Racehorse Creeks each year (2020 and 2021) using one-way ANOVA to compare the two retardant-affected sites in each creek against their respective reference sites using Tukey’s Honestly Significant Difference test performed using R (version 4.2.2). Given the modest replication (n = 3) of streambed sediment sampling at each site, (a) we considered α = 0.075 as statistically meaningful and (b) carefully considered the variable strength of retardant impacts based on their departure from reference conditions in interpreting our findings.

3. Results and Discussion

3.1. Stream Water Quality

Two weeks after the retardant drop, stream SRP was 33% greater at the drop site (site 2), whereas a larger 133% increase (along with TP) was observed 1 km downstream (site 3) compared to the reference site above the drop (site 1). No clear effects of the retardant on SRP were evident further downstream below the confluence with Racehorse Creek (sites 4–6, Tables and S2). One year later, while SRP declined at and below the drop in North Racehorse Creek (sites 2 and 3), SRP was elevated by 33 and 167% further downstream below the confluence with Racehorse Creek at sites 5 and 6 (1.5 and 6.1 km downstream of the retardant drop, respectively) compared to the Racehorse Creek reference (site 4). Stronger retardant impacts were observed in discharge-weighted SRP export, where a gradient of increasing downstream SRP export (0.039 to 0.104 kg d–1) was observed from sites 2, 3, and 5 shortly after the retardant drop in 2020 with a similar but stronger gradient (0.019 to 0.192 kg d–1) 1 year later extending further downstream to site 6 (Table ). Both TN and NO2 – + NO3 – concentrations were generally very low at all study sites, where no clear effects of the retardant were evident in 2020 or 2021 (Table S2). Moreover, it is particularly noteworthy that NH3 + NH4 + was below detection limits (<3 μg L–1) at all sites in both years.

2. Stream Discharge (Q), Total Suspended Ssolids (TSS), Total Phosphorus (TP), Soluble Reactive Phosphorus (SRP), Flow Weighed SRP Export, and Mean Equilibrium SRP Concentration (EPC0) Sampled on (a) September 1–2, 2020 and (b) September 21–22, 2021 .

  stream
 streambed
site Q (m3 s–1) TSS (mg L–1) TP (μg L–1) SRP (μg L–1) SRP export (kg d–1) EPC0 (μg L–1)
(a) Sept 1–2, 2020, 2 weeks after the retardant drop      
(1) N. Racehorse Ck. reference 0.112 1.2 4 3 0.029 31.9 (2.6)
(2) retardant drop 0.113 1.2 5 4 0.039 88.3 (12.4) **
(3) 1 km below the drop 0.123 0.4 8 7 0.074 34.9 (6.4)
(4) Racehorse Ck. reference 0.151 4.0 6 4 0.052 26.8 (2.6)
(5) 1.5 km below the drop 0.240 2.8 6 5 0.104 30.2 (7.8)
(6) 6.1 km below the drop 0.239 3.2 5 3 0.062 25.0 (6.3)
(b) Sept 21–22, 2021, 1 year later            
(1) N. Racehorse Ck. reference 0.116 3.1 23 4 0.040 25.8 (1.3)
(2) retardant drop 0.112 3.1 30 2 0.019 58.4 (20.2)
(3) 1 km below the drop 0.114 3.5 19 5 0.049 44.7 (17.9)
(4) Racehorse Ck. reference 0.126 6.0 32 3 0.033 16.3 (2.5)
(5) 1.5 km below the drop 0.240 4.6 8 4 0.083 51.8 (9.6) *
(6) 6.1 km below the drop 0.278 3.5 29 8 0.192 41.8 (11.0)
Open in a new tab
a

**<0.05, *p < 0.075, values in brackets = 1 std. error.

These findings show larger initial and longer-term retardant impacts on stream SRP than reported in the only previous study of an operational retardant misapplication during fire suppression operations (to our knowledge) where SRP increased at the drop site by only 4% in a weakly flowing Australian stream affected by both wildfire and retardant (suppressant foams), though no effects were evident in a more strongly flowing burned stream 2 weeks after the retardant drop. Indeed, SRP concentrations observed at our downstream retardant-affected sites 3 (2020) and 6 (2021) were broadly comparable to those observed after a severe wildfire 20 km to the south in this same region. Ammonia was strongly elevated at drop zones immediately after experimental aerial retardant application into several western U.S. streams, however elevated ammonia dissipated with stream flushing within 12–24 h. Consistent with those findings, we did not observe any effects of the retardant on stream NH3 + NH4 +, NO2 – + NO3 –, or TN 2 weeks after the retardant drop or 1 year later, supporting the suggestion that effects of long-term retardants on nitrogen may be short-lived. In contrast, 2 weeks after the accidental retardant drop in North Racehorse Creek, incomplete flushing of retardant residues from the stream and stream margins enabled the detection of elevated SRP at both the drop site and more strongly 1 km downstream. More importantly, 1 year after the retardant drop, elevated SRP was most strongly evident further downstream (6.1 km), suggesting that the P component of fertilizer-based long-term retardants may not be flushed as rapidly as previously suggested for other retardant components.

3.2. Streambed SRP Exchange

Intrusion of the retardant into streambed sediments elevated streambed P storage at the drop zone and further downstream. Sorption isotherms for sediments from North Racehorse Creek shortly after the retardant drop in 2020 showed similar rates of SRP sorption to streambed sediments with increasing aqueous SRP concentrations above, at, and 1 km below the retardant-impacted stream site (Figure a). However, while the EPC0 of sediments from both site 1 above the retardant drop and site 3 (1 km below the drop site) were similar (31.9 and 34.9 μg L–1, respectively), mean EPC0 at the retardant drop (site 2) was over 2.6 times greater (88.3 μg L–1, p = 0.007), indicating large increases in streambed SRP storage shortly after the retardant drop (Table ). Furthermore, while variability of bed sediment SRP sorption among replicate freeze-core samples was relatively low in the reference reach above the drop (std. error = 0.19 μg g–1), SRP sorption was 6.2 and 3.4 times more variable in bed sediments at the retardant drop (site 2) and 1 km below the drop (site 3), respectively, presumably reflecting high microsite variability in depositional dynamics of retardant-affected sediments. Further downstream below the confluence of North Racehorse with Racehorse Creek, SRP sorption was similarly more variable in retardant-affected bed sediments 1.5 and 6.1 km below the drop compared to the Racehorse Creek reference (site 4) where retardant effects on the EPC0 of streambed sediments were not evident in either of these latter two downstream sites (Figure b, p > 0.915 at sites 5 and 6).

3.

3

Open in a new tab

Streambed sediment soluble reactive phosphorus (SRP) sorption isotherms in (a) North Racehorse Creek, September 2020, (b) downstream in Racehorse Creek, September 2020, (c) North Racehorse Creek, September 2021 (1 year later), and (d) downstream in Racehorse Creek, September 2021 (1 year later).

Repeat freeze-core sampling 1 year later showed that (a) there was some recovery of elevated streambed storage of SRP and EPC0 at the retardant drop site but also (b) retardant-affected streambed sediments had moved further downstream in North Racehorse Creek and further below its confluence with Racehorse Creek (Figure c,d). The mean EPC0 of bed sediments at the retardant drop (site 2) had declined to 58.4 μg L–1 (34% lower than 1 year earlier, Table ). However, the mean EPC0 of downstream sites were notably elevated compared to the previous year (by 28, 72, and 67% in sites 3, 5, and 6, respectively), suggesting that elevated SRP from retardant residues or SRP sorbed to streambed sediments at the retardant drop site had been transported downstream over the previous year. Variation in EPC0 of downstream retardant-affected bed sediments was similar to or greater than that observed 1 year earlier. As a consequence, downstream effects on SRP storage and EPC0 were not strong, where only site 5 (1.5 km downstream of the drop site) differed from its respective reference (site 4, p = 0.059, Table ). Precipitation-driven stormflows or snowmelt events capable of mobilizing and transporting sediment-bound retardants in Racehorse Creek (gauged 4.8 km downstream of site 6) between the initial and final sampling occurred mid-May to early June, 2021 and included three larger 30–50 mm stormflow events (Figure S2). Indeed, 22 significant stormflows caused by precipitation or melt events up to 84 mm, likely capable of mobilizing intruded retardant residues or P-enriched streambed sediments, occurred between the successive streambed sampling 2 weeks after the retardant drop and 1 year later (Table S3).

In contrast to aqueous streamwater quality where dilution, mixing, and downstream dispersion processes can potentially reduce point source water quality impacts from some retardant constituents, downstream transport and deposition of retardant residues or P-enriched sediments into streambeds can produce differential impacts for parameters with an affinity for storage and exchange with sediments (such as SRP). Initial retardant impacts on streambed SRP storage (EPC0) in 2020 were apparent at the drop site only. Downstream transport of retardant-affected sediments by streambed erosion and depositional processes clearly showed both upstream flushing at the drop site and sequential downstream deposition in North Racehorse Creek and below its confluence with Racehorse Creek 1 year later. While increased downstream EPC0 1 year after the drop was significant only at site 5 (due to higher variability in retardant-affected bed samples compared to reference sites, Figure ), despite only modest replication (n = 3), a distinct pattern of downstream sediment-retardant transport on EPC0 was evident up to 6.1 km downstream. This is important for several reasons: (1) this clearly illustrates that misapplication of retardants into streams can have much longer-lasting impacts than previously reported, (2) these impacts can propagate substantial distances downstream over time, and (3) study of aqueous water quality alone without considering the role of sediment in transport and fate of some retardant constituents may not fully capture the scope of potential retardant impacts. Indeed, while our findings showed downstream retardant effects on aqueous P (Table ), measurement of streambed SRP storage and phosphorus release potential (PRP) revealed clear patterns of temporal and downstream spatial impacts after retardant misapplication.

3.3. Streambed Sources of SRP

Streambed sediments in all 6 study sites represented a source of aqueous SRP in both North Racehorse and Racehorse Creeks throughout the study, reflecting the oligotrophic conditions of the region. Similar to findings for EPC0, streambed phosphorus release potential (PRP) 2 weeks after the retardant drop was notably greater at the retardant drop (Figure ), where the PRP increased 3-fold from 2.53 μg g–1 in the upper reference (site 1) to 7.65 μg g–1 at site 2 (p = 0.014), but no effects of the retardant were observed further downstream (p > 0.992, sites 3, 5, and 6). However, 1 year later, while PRP at the drop site (site 2) declined by 37% compared to the previous year, PRP increased by 93%, 2.7 times, and 1.5 times over that observed the previous year at sites 3, 5, and 6, respectively (Figure ), indicating that downstream transport of sediment-bound SRP had increased the strength of streambed SRP sources in these downstream retardant-affected sites. Despite the larger relative increases in mean PRP in 2021, these increases were not strong except at site 5 (p = 0.031).

4.

4

Open in a new tab

Streambed phosphorus release potential (PRP) and stream SRP above, at, and downstream of the retardant drop in September 2020 and September 2021 (1 year later). Bars indicate mean PRP (+1 std. error), and symbols indicate stream SRP.

These findings are important in illustrating that downstream transport/deposition of retardant-affected sediments can subsequently increase downstream streambed SRP sources, thereby producing potentially long-lasting effects on aqueous water quality. It is particularly noteworthy that while retardant impacts on both SRP storage (EPC0) and streambed PRP were strongest at the drop site (site 2) in 2020 and 1.5 km downstream of the drop (site 5) in 2021, peak aqueous stream SRP was observed at the next-most subsequent downstream sampling sites during each of these years (i.e., peak stream SRP occurred at site 3 in 2020, and site 6 in 2021). These data suggest that spatial patterns in downstream water quality may be spatially separated or lagged below peak streambed sources, potentially reflecting slower streambed boundary diffusion and mixing processes with streams.

3.4. Periphyton Response

No clear effects of the retardant on periphyton productivity were observed for either ash-free dry mass (AFDM) or chlorophyll a in either floating tiles (p > 0.164) or benthic-streambed tiles (p > 0.865) at the retardant drop compared to the reference site above the drop in 2020 or 2021 (Table S4). While greater mean AFDM and chlorophyll a appeared weakly evident in late fall 2020 shortly after the retardant drop and in May 2021 after the first major postretardant snowmelt freshet (chlorophyll a only, Table S4), even these initial retardant effects of periphyton production were weak (p > 0.092). Despite the fact that early 2020 effects on periphyton could not be meaningfully assessed due to lost growth tiles, no coherent effects of diminished but still elevated streambed EPC0 and streambed SRP sources were evident at the drop site.

While periphyton productivity was measured only at the upstream reference (site 1) and the retardant drop (site 2) based on the hypothesis that retardant effects on stream SRP would be greatest at the drop site, retardant effects on SRP were only moderate at this site in 2020 (25% increase relative to the upstream reference site) and completely diminished the following year (Table ), whereas much stronger retardant effects were noted 1 and 6.1 km downstream in 2020 and 2021, respectively. Indeed, because we did not anticipate the extent to which P-rich sediments would move downstream in our periphyton sampling design, we did not capture those downstream effects on periphyton productivity. However, our previous research on periphyton response to elevated P after wildfire in the same region showed very large shorter-term (4 year) and longer-term (∼10 year) increases in both AFDM and chlorophyll a after similar or smaller increases in SRP than reported here after retardant inputs. , Thus, it is reasonable to speculate that the downstream retardant effects on SRP we observed at sites 3 (2020) and 5 (2021) would likely have produced periphyton growth responses had we sampled those sites.

3.5. Implications

In some respects, the accidental application of a fertilizer-based long-term retardant (PHOS-CHEK LC95A) into North Racehorse Creek represents a worst-case example of aerial retardant misapplication, while in other respects, this case study may not reflect the most serious conditions possible. For example, while most fire suppression agencies (Canada, and others) employ strict guidelines to avoid aerial retardant application within ∼100 m of watercourses , to protect water bodies and adjacent riparian habitats from retardants, the misapplication of a substantial volume of retardant directly into North Racehorse Creek likely produced much stronger impacts than would be expected from misapplication to adjacent riparian margins or areas more distant from watercourses where foliar interception and less efficient runoff pathways have been previously reported to limit stream impacts. Similarly, the seasonal timing of retardant misapplication relative to the timing of regional runoff could either abate or amplify impacts to streams. In the present case, misapplication of retardants occurred in late summer–early fall during the period of lowest seasonal streamflow (Figure S2) when the potential for downstream flushing was lowest and the potential for retardant intrusion into streambeds would be greatest. In contrast, retardant misapplication during higher flow periods in the spring or early summer snowmelt freshet may result in more efficient downstream flushing and dilution of retardants. Similarly, while North Racehorse Creek is a highly turbulent, upper-elevation Rocky Mountain stream, comparable retardant misapplication into less-turbulent, lower-gradient streams could produce larger impacts due to less efficient flushing with extended contact time and longer mixing lengths in stream reaches than reported here.

To our knowledge, this is the first study to document the short- and longer-term impacts of an operational aerial misapplication of a large (unquantified) volume of a fertilizer-based long-term fire retardant directly into a stream and the first to focus on the P component of retardants where contaminant–sediment interactions may prolong effects over that of other aqueous components that may be more efficiently diluted and dispersed. The study showed clear spatial and temporal patterns of retardant impacts where the strongest impacts were observed at the drop site and 1 km downstream shortly after the misapplication, whereas those impacts moved downstream below the confluence with another stream, where they were detectable at least ∼6 km below the retardant drop site 1 year later.

Given contemporary increases in the extent and severity of wildfire behavior in fire-prone regions of western North America , and similar regions worldwide, the use of an aerial fire retardant as a critical fire suppression tool is likely to increase in the future. , While wildfire suppression agencies will likely advance efforts to avoid accidental retardant misapplication into watercourses during wildfire suppression, wildfire managers will also likely face increasingly difficult emergency tactical decisions balancing the need to concurrently protect life, key infrastructure, forest, and watershed values at risk during wildfires.

Supplementary Material

es5c04555_si_001.pdf (1.7MB, pdf)

Acknowledgments

The authors gratefully acknowledge the helpful comments of three reviewers in improving this manuscript and are particularly grateful to Ben Williamson (Foothills Research Institute) for concept development and creation of the graphic abstract. The authors thank Sydney Enns and Brooke Hehr for assistance with field sampling. The authors also gratefully acknowledge the assistance of the Alberta Ministry of Forestry and Parks in completing this study and are especially grateful to the government agencies and industrial partners for their financial support of this work.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.5c04555.

  • Analytical methods for water quality analysis (Table S1); results for allied water quality parameters sampled on September 1–2, 2020 and September 21–22, 2021 (Table S2); photos of freeze-core sampling (Figure S1); analysis methods and summary of significant stormflow events on Racehorse Creek, September 2020–September 2021 (Figure S2 and Table S3); and results of floating and benthic periphyton growth in fall 2020 and June–October, 2021 (Table S4) (PDF)

This work was supported by grants to U.S. from the National Science and Engineering Research Council of Canada (05497 and 494312), Alberta Innovates (222301747), Canadian Forest Products-FRIAA (FFI-15–010), Parks Canada; Waterton Lakes National Park (GC-1159), Alberta Ministry of Forestry and Parks, West Fraser, Weyerhaeuser, Forest Resources Improvement Association of Alberta, and the Foothills Research Institute (U24001).

The authors declare no competing financial interest.

References

  1. Abatzoglou J. T., Juang C. S., Williams A. P., Kolden C. A., Westerling A. L.. Increasing synchronous fire danger in forests of the western United States. Geophys. Res. Lett. 2021;48(2):e2020GL091377. doi: 10.1029/2020GL091377. [DOI] [Google Scholar]
  2. Wang X., Thompson D. K., Marshall G. A., Tymstra C., Carr R., Flannigan M. D.. Increasing frequency of extreme fire weather in Canada with climate change. Clim. Change. 2015;130:573–586. doi: 10.1007/s10584-015-1375-5. [DOI] [Google Scholar]
  3. Jolly W. M., Cochrane M. A., Freeborn P. H., Holden Z. A., Brown T. J., Williamson G. J., Bowman D. M. J. S.. Climate-induced variations in global wildfire danger from 1979 to 2013. Nat. Commun. 2015;6:7537. doi: 10.1038/ncomms8537. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Hoffman K. M., Christianson A. C., Gray R. W., Daniels L.. Western Canada’s new wildfire reality needs a new approach to fire management. Environ. Res. Lett. 2022;17(6):061001. doi: 10.1088/1748-9326/ac7345. [DOI] [Google Scholar]
  5. Westerling A. L.. Increasing western US forest wildfire activity: sensitivity to changes in the timing of spring. Philos. Trans. R. Soc., B. 2016;371(1696):20150178. doi: 10.1098/rstb.2015.0178. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Hope E. S., McKenney D. W., Pedlar J. H., Stocks B. J., Gauthier S.. Wildfire suppression costs for Canada under a changing climate. PLoS One. 2016;11(8):e0157425. doi: 10.1371/journal.pone.0157425. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. OMB, Climate financial risk: The federal government’s budget exposure to financial risk due to climate change Office of Management and Budget 2024, p 131.
  8. Jain P., Barber Q. E., Taylor S. W., Whitman E., Acuna D. C., Boulanger Y., Chavardès R. D., Chen J., Englefield P., Flannigan M., Girardin M. P., Hanes C. C., Little J., Morrison K., Skakun R. S., Thompson D. K., Wang X., Parisien M. A.. Drivers and Impacts of the Record-Breaking 2023 Wildfire Season in Canada. Nature Commun. 2024;15:6764. doi: 10.1038/s41467-024-51154-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. National Interagency Fire Centre (U.S.A) . Wildland fire summaries, Suppression costs 2024. https://www.nifc.gov/ (accessed June 1).
  10. USFS . Nationwide Aerial Application of Fire Retardants on National Forest System Land: Final Environmental Impact Statement, United States Department of Agriculture; 2011. [Google Scholar]
  11. Adams R., Simmons D.. Ecological effects of fire fighting foams and retardants: a summary. Aust. For. 1999;62(4):307–314. doi: 10.1080/00049158.1999.10674797. [DOI] [Google Scholar]
  12. Kalabokidis K. D.. Effects of wildfire suppression chemicals on people and the environment – A review. Global NEST J. 2000;2(2):129–137. doi: 10.30955/gnj.000144. [DOI] [Google Scholar]
  13. Giménez A., Pastor E., Zarate L., Planas E., Arnaldos J.. Long-term forest fire retardants: a review of quality, effectiveness, application and environmental considerations. Int. J. Wildland Fire. 2004;13:1–15. doi: 10.1071/WF03001. [DOI] [Google Scholar]
  14. Schammel M. H., Hold S. J., McCurry D. L.. Metals in wildfire suppressants. Environ. Sci. Technol. Lett. 2024;11:1247–1253. doi: 10.1021/acs.estlett.4c00727. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Parks N.. Use of wildfire retardants under scrutiny. Front. Ecol. Environ. 2011;9(8):426 [Google Scholar]
  16. Emelko M. B., Silins U., Bladon K. D., Stone M.. Implications of land disturbance on drinking water treatability in a changing climate: Demonstrating the need for “source water supply and protection” strategies. Water Res. 2011;45:461–472. doi: 10.1016/j.watres.2010.08.051. [DOI] [PubMed] [Google Scholar]
  17. Bladon K. D., Emelko M. B., Silins U., Stone M.. Wildfire and the Future of Water Supply. Environ. Sci. Technol. 2014;48(16):8936–8943. doi: 10.1021/es500130g. [DOI] [PubMed] [Google Scholar]
  18. USFS . Nationwide Aerial Application of Fire Retardant on National Forest System Lands: Final Supplemental Environmental Impact Statement, United States Department of Agriculture, Forest Service; 2023. [Google Scholar]
  19. Norris, L. A. ; Webb, W. L. . Effects of Fire Retardant on Water Quality, United States Department of Agriculture, Forest Service, Gen. Tech. Rep.. [Google Scholar]
  20. Dietrich J. P., Van Gaest A. L., Myers M. S., Strickland A., Hutchinson G. P., Van Gaest A., Krupkin A. B., Arkoosh M. R.. Toxicity of PHOS-CHEK LC-95A and 259F fire retardants to ocean- and stream-type Chinook salmon and their potential to recover before seawater entry. Sci. Total Environ. 2014;490:610–621. doi: 10.1016/j.scitotenv.2014.05.038. [DOI] [PubMed] [Google Scholar]
  21. Dietrich J. P., Myers M. S., Strickland S. A., Van Gaest A. L., Arkoosh M. R.. Toxicity of forest fire retardant chemicals to stream-type Chinook salmon undergoing parr-smolt transformation. Environ. Toxicol. Chem. 2013;32:236–247. doi: 10.1002/etc.2052. [DOI] [PubMed] [Google Scholar]
  22. Gaikowski M. P., Hamilton S. J., Buhl K. J., McDonald S. F., Summers C. H.. Acute toxicity of three fire-retardants and two fire-suppressant foam formulations to the early life stages of rainbow trout (Oncorhynchus mykiss) Environ. Toxicol. Chem. 1996;15:1365–1374. doi: 10.1897/1551-5028(1996)0152.3.co;2. [DOI] [Google Scholar]
  23. Buhl K. J., Hamilton S. J.. Acute toxicity of fire-control chemicals, nitrogenous chemicals, and surfactants to rainbow trout. Trans. Am. Fish. Soc. 2000;129:408–418. doi: 10.1577/1548-8659(2000)129&#x0003c;0408:ATOFCC&#x0003e;2.0.CO;2. [DOI] [Google Scholar]
  24. Buhl K. J., Hamilton S. J.. Acute toxicity of fire-retardant and foam suppressant chemicals to early life stages of Chinook salmon (Oncorhynchus tshawytscha) Environ. Toxicol. Chem. 1998;17:1589–1599. doi: 10.1002/etc.5620170821. [DOI] [Google Scholar]
  25. Calfee R. D., Little E. E.. The effects of ultraviolet-b radiation on the toxicity of fire-fighting chemicals. Environ. Toxicol. Chem. 2003;22:1525–1531. doi: 10.1002/etc.5620220714. [DOI] [PubMed] [Google Scholar]
  26. McDonald S. F., Hamilton S. J., Buhl K. J., Heisinger J. F.. Acute toxicity of fire control chemicals to Daphnia magna (Straus) and Selenastrum capricornutum (Printz) Ecotoxicol. Environ. Saf. 1996;33:62–72. doi: 10.1006/eesa.1996.0007. [DOI] [PubMed] [Google Scholar]
  27. McDonald S. F., Hamilton S. J., Buhl K. J., Heisinger J. F.. Acute toxicology of fire-retardant and foam-suppressant chemicals to Hyalella Azteca (Saussure) Environ. Toxicol. Chem. 1997;16:1370–1376. doi: 10.1002/etc.5620160706. [DOI] [Google Scholar]
  28. Rehmann C. R., Jackson P. R., Puglis H. J.. Predicting the spatiotemporal exposure of aquatic species to intrusions of fire retardant in streams with limited data. Sci. Total Environ. 2021;782:146879. doi: 10.1016/j.scitotenv.2021.146879. [DOI] [Google Scholar]
  29. Angeler D. G., Moreno J. M.. Impact-recovery patterns of water quality in temporary wetlands after fire retardant pollution. Can. J. Fish Aquat. Sci. 2006;63:1617–1626. doi: 10.1139/f06-062. [DOI] [Google Scholar]
  30. Boulton A. J., Moss G. L., Smithyman D.. Short-term effects of aerially-applied fire suppressant foams on water chemistry and macroinvertebrates in streams after natural wildfire on Kangaroo Island, South Australia. Hydrobiologia. 2003;498:177–189. doi: 10.1023/A:1026213301871. [DOI] [Google Scholar]
  31. Mainstone C. P., Parr W.. Phosphorus in riversecology and management. Sci. Total Environ. 2002;282–283:25–47. doi: 10.1016/S0048-9697(01)00937-8. [DOI] [PubMed] [Google Scholar]
  32. Schindler D. W.. Recent advances in the understanding and management of eutrophication. Limnol. Oceanogr. 2006;51:356–363. doi: 10.4319/lo.2006.51.1_part_2.0356. [DOI] [Google Scholar]
  33. Silins U., Bladon K. D., Kelly E. N., Esch E., Spence J. R., Stone M., Emelko M. B., Boon S., Wagner M. J., Williams C. H. S., Tichkowsky I.. Five-year legacy of wildfire and salvage logging impacts on nutrient runoff and aquatic plant, invertebrate, and fish productivity. Ecohydrology. 2014;7(6):1508–1523. doi: 10.1002/eco.1474. [DOI] [Google Scholar]
  34. House W. A.. Geochemical cycling of phosphorus in rivers. Appl. Geochem. 2003;18:739–748. doi: 10.1016/S0883-2927(02)00158-0. [DOI] [Google Scholar]
  35. Jarvie H. P., Jurgens M. D., Williams R. J., Neal C., Davies J. J. L., Barrett C., White J.. Role of river bed sediments as sources and sinks of phosphorus across two major eutrophic UK river basins: the Hampshire Avon and Herefordshire Wye. J. Hydrol. 2005;304:51–74. doi: 10.1016/j.jhydrol.2004.10.002. [DOI] [Google Scholar]
  36. Hawthorn, K. The role of fine sediment in phosphorus dynamics and stream productivity in Rocky Mountain headwater streams: Possible long-term effects of extensive logging, Thesis; University of Alberta: Edmonton, AB, 2014. [Google Scholar]
  37. House W. A., Denison F. H.. Factors influencing the measurement of equilibrium phosphate concentrations in river sediments. Water Res. 2000;34:1187–1200. doi: 10.1016/S0043-1354(99)00249-3. [DOI] [Google Scholar]
  38. Froelich P. N.. Kinetic control of dissolved phosphate in natural rivers and estuaries; A primer on phosphate buffer mechanism. Limnol. Oceanogr. 1988;33:649–668. doi: 10.4319/lo.1988.33.4_part_2.0649. [DOI] [Google Scholar]
  39. Simpson Z. P., McDowell R. W., Condron L. M., McDaniel M. D., Jarvie H. P., Abell J. M.. Sediment phosphorus buffering in streams at baseflow: A meta-analysis. J. Environ. Qual. 2021;50(2):287–311. doi: 10.1002/jeq2.20202. [DOI] [PubMed] [Google Scholar]
  40. Methods for the De-termination of Inorganic Substances in Environmental Samples, Determination of Phosphorous by Semi-Automated Colorimetry. 1993.
  41. Aloi J. E.. A critical review of recent freshwater periphyton field methods. Can. J. Fish. Aquat. Sci. 1990;47:566–670. doi: 10.1139/f90-073. [DOI] [Google Scholar]
  42. Sartory D. P., Grobbelar J. U.. Extraction of chlorophyll a from freshwater phytoplankton for spectrophotometric analysis. Hydrobiologia. 1984;114:117–187. doi: 10.1007/BF00031869. [DOI] [Google Scholar]
  43. Martens A. M., Silins U., Proctor H. C., Williams C. H. S., Wagner M. J., Emelko M. B., Stone M.. Long-term impact of severe wildfire and post-wildfire salvage logging on macroinvertebrate assemblage structure in Alberta’s Rocky Mountains. Int. J. Wildland Fire. 2019;28:738–749. doi: 10.1071/WF18177. [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

es5c04555_si_001.pdf (1.7MB, pdf)

Articles from Environmental Science & Technology are provided here courtesy of American Chemical Society

RESOURCES