Accidental spills of chemicals and other pollutants can decimate populations of stream-dwelling species. Recovery from such accidents can be relatively fast and complete when the affected stream reaches can be recolonized from upstream and downstream sources. However, faunal recoveries from accidental spills that extirpate populations from entire headwater streams have not been extensively documented, and understanding resilience of headwater-stream biota is relevant for assessing threats to at-risk species. We assessed recovery of fish populations in a 5.7-km-long headwater stream in the southeastern United States following a complete, or nearly complete, fish-kill caused by a chemical spill near the source of the stream. We sampled for fishes at five stream locations, two downstream and three upstream from a perched, culverted road-crossing located 2.4 km upstream from the stream mouth, over a period of 18.5 mo following the poisoning event. We observed 11 fish species, representing ≤65% of the fish species expected based on occurrences in nearby tributary streams. In postpoisoning sampling, only three of these taxa were observed upstream of the culvert; all 11 species, including the federally threatened Cherokee Darter Etheostoma scotti, were found downstream of the culvert but were mostly represented by a few, large individuals. In contrast, dead individuals of at least eight taxa including the Cherokee Darter were observed upstream of the culvert at the time of the fish-kill. These observations provide evidence of slow recovery of a headwater fish fauna, and especially upstream of a barrier to fish movement, where the recolonization sources are primarily downstream. Additional case studies may reveal whether this result applies generally to headwater streams. Slow recovery could make species that primarily inhabit or maintain greatest abundances in headwaters, including multiple at-risk fishes, particularly vulnerable to the threat of accidental spills that result in local population extirpation.

Chemical spills into surface waters happen often. In the United States, the National Response Center (U.S. Coast Guard; https://nrc.uscg.mil) reports tens of thousands of emergency calls annually, many of which are spills of industrial chemicals, fuel or sewage in or near water, and many more incidents almost certainly go unreported. Such spills can result in catastrophic loss of aquatic life (Olmsted and Cloutman 1974; Lytle and Peckarsky 2001; Kubach et al. 2011; Erős et al. 2015). Accidental spills of toxic materials that affect long reaches of streams can thus reduce or extirpate populations of aquatic species and may pose a particular threat to populations of imperiled stream species that exist at low abundances and in restricted ranges.

Whether and how quickly a stream species can reestablish a population following catastrophic loss is a key component of population resilience and depends on multiple factors, including distances to occupied habitats that can provide colonizers and species-specific dispersal capabilities (Detenbeck et al. 1992). Studies have documented relatively quick recovery by stream fishes following catastrophic spills in cases where colonists come from unaffected tributary streams to the kill zone, or from upstream and downstream reaches. In these cases, fish densities and species richness can approach predisturbance levels within 1–2 y (Ensign et al. 1997; Meade 2004; Erős et al. 2015). However, we know less about how quickly fishes in headwater streams are likely to recover from spills that extirpate populations over the length of the stream. In these cases, the primary sources of colonizers would be downstream reaches of typically larger streams, which may have differing species assemblages than existed in the headwaters.

Barriers to fish movement have been shown to impede population recovery from catastrophic spills (Kubach et al. 2011) as well as from experimental fish removal (Albanese et al. 2009). Barriers thus factor into management considerations for imperiled fishes. Understanding the extent to which human-built structures (e.g., mill dams, culverted road-crossings) and natural barriers (e.g., waterfalls) isolate populations or impede recolonization can help clarify the risk to populations that are vulnerable to catastrophic loss. Fish populations that are isolated above potential movement-barriers in headwater streams could be particularly vulnerable to local extirpation events. Understanding the extent to which putative barriers slow recovery can help natural resource managers allocate resources for barrier removal or mitigation.

We examined fish recolonization following a catastrophic poisoning event in a headwater stream, where a culverted road-crossing forms a probable impediment to upstream fish movement. The study stream is in the southeastern United States, which is a global hotspot of temperate freshwater diversity, including high rates of aquatic species endemism and imperilment (Abell et al. 2000). Moreover, the region has relatively few protected lands, leaving imperiled species vulnerable to a wide range of human-caused disturbances (Elkins et al. 2019), including spills resulting in population loss. In the case presented here, conservation managers are concerned with population reestablishment by an imperiled fish species endemic to smaller streams in the study basin. More generally, understanding resilience in headwater fish populations could allow managers to assess more accurately the threat of catastrophic events to population viability.

Flat Creek is a small tributary stream in the Coosa River drainage in north Georgia, United States. Flat Creek runs for most of its approximate 5.7-km length through the city of Dawsonville, Dawson County, Georgia, before joining Shoal Creek (Figure 1), which is a tributary to the Etowah River. Flat Creek and Shoal Creek contain populations of the Cherokee Darter Etheostoma scotti (family Percidae), which is listed as threatened pursuant to the U.S. Endangered Species Act (ESA 1973, as amended; USFWS 1994) and is also protected as threatened under the State of Georgia's Endangered Wildlife Act (GDNR undated). The Cherokee Darter is endemic to the Etowah River system (which includes Shoal Creek and Flat Creek) and primarily occupies small to medium-sized streams in the Piedmont physiographic province (USFWS 1994; Storey et al. 2006; Barton and Powers 2010).

On March 20, 2018, Flat Creek experienced a poisoning event reportedly caused by an accidental release of ∼190 L of ferric chloride (FeCl3) into the stream's headwaters (Durniak 2018). Ferric chloride combines with water to form ferric hydroxide, releasing hydrogen ions and thereby lowering pH. Low pH can cause mortality in fishes by interrupting ion transport, damaging gills, and inducing suffocation (McDonald et al. 1991; Kwong et al. 2014). Ferric hydroxide forms a precipitate that can clog gills of fishes (Dalzell and McFarlane 1999) and smother streambed sediments and benthic organisms (Cadmus et al. 2018). Biologists from the Georgia Department of Natural Resources (GDNR) responded to the event and counted 988 dead fish (some identified to species; others to genus or family) at six sites along the length of the stream from its headwaters, downstream to the confluence with Shoal Creek (Durniak 2018). Measured stream pH values on March 23, 3 d after the initial chemical release, were as low as 4.5, and the stream bed sediments were blanketed by an orange precipitate. The GDNR biologists concluded that the event had resulted in a complete kill of fish (estimated at >8,000 individuals) and other aquatic life in the entire length of Flat Creek, but that the fish kill did not extend into Shoal Creek (Durniak 2018).

Flat Creek is nearly bisected by a perched (i.e., elevated above the downstream hydrologic control) concrete culvert installed in 2006 to accommodate a new road-crossing, ∼2.4 km upstream from the confluence with Shoal Creek (Figure 1). The culverted road-crossing rests on a concrete sill that is elevated ∼60 cm above the streambed and forms a likely barrier to upstream fish movement except possibly when streamflow is high enough to submerge the sill. Although Shoal Creek harbors a diverse fish assemblage and is expected to provide colonists to repopulate Flat Creek, conservation managers have expressed concern that this culvert may impede upstream recolonization by fishes, and perhaps other aquatic fauna, and particularly the imperiled Cherokee Darter.

Field methods

We sampled fishes and associated aquatic fauna in Flat Creek using a seine (2.4 m × 1.8 m, 3-mm mesh) on 6 dates, from July 2018 to October 2019 (Table 1), ∼4, 6, 12, 13, 16, and 18.5 mo following the poisoning event. We accessed five sample locations from road crossings or with permission from land-owners (Figure 1), and selected them to represent the length of Flat Creek from near the headwater origin of the spill to ∼1.4 km upstream from the Shoal Creek confluence. The length of stream sampled at each location varied depending on access and stream size, and ranged from ∼50 m (upstream-most location) to 700 m (location immediately downstream from the perched culvert; Table 1). Locations A and B, both upstream of the perched culvert, were ∼1.2 km and 2.4 km, respectively, downstream from the location of the spill, which occurred immediately upstream of the mapped origin of Flat Creek (Durniak 2018). The three downstream-most locations (C, D, E; Figure 1) were essentially contiguous, spanning from ∼280 m upstream of the perched culvert (and ∼600 m downstream from location B), to ∼1,000 m downstream from the perched culvert. We sampled most extensively and repeatedly (i.e., in each visit after July 2018; Table 1) downstream from the perched culvert to maximize our ability to observe recolonizing fish species (which we believed would move upstream from Shoal Creek), while also allocating effort to locations upstream of the culvert to observe fish species that either survived the poisoning event or were able to traverse the culvert during high flows.

Sampling events consisted of multiple seine-sets and seine hauls in all habitat types at a location (Table 1). We designed fish sampling methods to minimize harm to animals and habitat disturbance, while achieving a sufficient capture efficiency to meet project goals (following Use of Fishes in Research Committee 2014 guidelines). We counted and measured (standard length) all captured fish, and released fish at the point of capture (low capture numbers ensured individuals were unlikely to be counted more than once on a date). After the initial samples in July 2018, we recorded fishes and other larger organisms (including salamanders and crayfishes) captured in each seine-set or haul, and noted bed sediment type. For 2019 samples, we measured water depth with a wading rod (nearest 3 cm) and visually estimated (based on passage rate of floating objects, nearest 5 cm/s) or measured (in Oct 2019, using an electronic meter) water velocity.

Analyses

We estimated mean and range of depths sampled by location for the eight sampling events in 2019; and for six of these dates, we quantified the percentage of seine-sets or hauls that had water velocities >20 cm/s. We used these summary data to qualitatively compare the range of habitats sampled among locations, and to identify differences upstream compared with downstream of the culvert that could account for differences in species observed. For these comparisons, we combined data from March and April 2019 samples because flow levels were similar on those dates (i.e., ∼16.14 m3/s, U.S. Geological Survey [USGS] gage 0238915, Etowah River at GA 9 near Dawsonville, Georgia, on both sampling dates). We summarized fish data as species observed at each location, along with numbers of individuals and the range of individual standard lengths. We combined data across sampling events to describe species occurrences in Flat Creek, rather than explicitly estimating species-specific occupancy (i.e., using repeated samples at locations to estimate species-specific detection), because we assumed that each location was open to colonization or emigration during the periods between our samples. Thus, our analysis aimed to compare the total list of species observed over the 14-mo sampling period to species expected to occur based on fishes previously observed in Shoal Creek or in nearby tributaries.

Expected versus observed fish species recolonization.

We did not have fish assemblage data for Flat Creek prior to the poisoning event. Therefore, we used historical fish-collection data for the Shoal Creek system to compile a list of species conceivably available to recolonize Flat Creek. We obtained collection records from the Georgia Museum of Natural History, which archives complete lists of fishes encountered, counted, or both, along with dates and locality data for samples throughout Georgia. We extracted records that represented efforts to capture a variety of fish species, mostly by seining but sometimes also with electrofishing equipment or by snorkeling surveys, in the Shoal Creek system. Although these collections represented a variety of sampling efforts, stream conditions and time periods, we believe these records taken together provided the best available basis for inferring expected species occurrence in Flat Creek in the absence of a complete list of fishes present in the stream itself prior to the poisoning event or in Shoal Creek at the time of our study.

We used two sets of collection records to infer the fish species expected to recolonize Flat Creek. First, we compiled records of fishes captured in samples from the Shoal Creek mainstem to represent the pool of species most available to colonize Flat Creek (i.e., by moving directly from Shoal Creek into Flat Creek). These records comprised 41 distinct collections, made from 1948 to 2017, at 10 unique localities in Shoal Creek, Dawson County, Georgia. Second, we compiled collection records from samples in other nearby, tributary streams to Shoal Creek to represent the assemblage of fish species that commonly occurs in tributaries similar to Flat Creek. We expected that species that occur in the Shoal Creek mainstem would not be equally likely to occupy smaller tributary streams, in which case fishes observed in tributary samples could better predict species recolonization in Flat Creek compared with basing predictions on Shoal Creek samples. Tributary records comprised 16 distinct collections made from 1950 to 2011 at 6 unique locations across 4 tributary streams: Burt Creek (6 collections), Pigeon Creek (4), Sweetwater Creek (5), and an unnamed tributary near the headwaters of Shoal Creek (1), all in Dawson County and in the same ecoregion as Flat Creek. For comparison with Flat Creek, we used Geodata Crawler (Leasure 2014) to calculate watershed area and percent urban landcover at each collection site based on the 2016 National Land Cover Database (Yang et al. 2018). Watershed area at these 6 tributary collection sites averaged 8.0 km2 (range = 1.2–12.3 km2), similar to the 5.9-km2 drainage area for the Flat Creek watershed. As of 2016, the Flat Creek basin had more urban land cover (29%) compared with the mean for the tributary locations (14%, range = 4–31%).

We used four variables to evaluate whether fish species occurrences in Flat Creek postpoisoning were better predicted by the colonizer pool in Shoal Creek or by the assemblage composition of nearby tributaries. First, we estimated observation frequency by separately summing the number of Shoal Creek and tributary collections in which each potentially colonizing species occurred. Additionally, 18 of the Shoal Creek collections and 15 of the tributary collections reported the number of individuals captured for each species. Using these two subsets of collection records, we summed the number of individuals captured for each species. We thus used two predictor variables, species-specific observation frequency and summed capture abundance, computed using each set of collections (i.e., Shoal Creek samples and tributary samples). These four variables were correlated with each other (values for Pearson correlation coefficient, r, = 0.66–0.92). Therefore, we used each of the four variables in separate logistic regression models to predict the observed occurrence of a species in our combined postpoisoning samples in Flat Creek. The response variable was a binary value representing whether we observed each of the potentially colonizing species in Flat Creek during the 14-mo sampling period. We placed predictor variables on a common scale by subtracting the variable-specific mean from each value and dividing by the standard deviation. We compared model fit using Akaike's Information Criterion (AIC; smaller values represent better support by the data; Burnham and Anderson 2002). We fit models in Program R (R Core Team 2020, version 4.0.0) using the function glm. We used the best-supported models to identify species that could be expected to recolonize Flat Creek but were not observed in our samples.

Fish species observed

We captured 11 fish species in Flat Creek across the 6 sampling occasions (Table 2). Eight of these species were only captured downstream from the perched culvert, whereas three species—Creek Chub Semotilus atromaculatus, Largescale Stoneroller Campostoma oligolepis, and Bluegill Lepomis macrochirus—were observed at sites upstream and downstream from the culvert. We also captured a larval fish in the July 2018 sample at the upstream-most site that was identified in the field as a percid and gently released. We had no other evidence of species other than Creek Chub, Largescale Stoneroller, or Bluegill occurring upstream from the perched culvert. In contrast, we captured plethodontid salamanders (Desmognathus sp., Eurycea sp., or both) and crayfishes (Cambarus spp., Procambarus spiculifer, or both) in all collections except the initial qualitative sample in July 2018 at the upstream-most site (location A, Figure 1).

Fish species were mostly represented by five or fewer individuals during sampling events, even 18.5 mo following the poisoning event (i.e., October 2019; Table 2). The exceptions were Creek Chub and Largescale Stoneroller, for which young-of-year were numerous in July 2018 (4 mo postpoisoning) upstream of the perched culvert. Creek Chub captures outnumbered all other species at all sites except during March 2019, when we captured few fish of any species (Table 2). Creek Chub also was the only species observed consistently upstream and downstream from the perched culvert, captured in all except one sampling event (location C, March 2019) in which we captured no fish.

Eight of the 11 fish species observed downstream from the perched culvert were captured in September 2018, and thus apparently colonized Flat Creek within ∼6 mo of the poisoning event. Seven of these species (e.g., excepting Creek Chub) were only captured as large-sized adults (Table 2). Two taxa, Rainbow Shiner Notropis chrosomus and Banded Sculpin Cottus carolinae, were only captured during one or both of the final two samples (i.e., ≥16 months postspill) and were also represented by one or two large-sized adults (Table 2). Across all dates, we observed young-of-year individuals (<35 mm standard length) of only four taxa (Table 2). The federally protected Cherokee Darter was captured downstream from the perched culvert on three dates, 6, 12, and 18.5 mo postpoisoning (Table 2). The four observed individuals were adults and included a male and female captured together in September 2018.

Stream habitat conditions

Habitat appeared similar between most sites upstream compared with downstream of the perched culvert. The stream channel was smallest at the upstream-most site (location A, Figure 1), where a box culvert created a nearly stagnant pool upstream of the road crossing, with a narrow (2–4-m wetted width) rocky channel downstream. The channel varied from ∼2–6 m (wetted width) at the more downstream sites, with mostly wooded riparian areas, riffle and pool structure, and cobble, gravel and sand bed-sediments. The three downstream-most sites were characterized by extensive bedrock. Mean depths varied from 27 to 34 cm, except during low-flow conditions in July 2019 when the stream was shallower (Table 3). During higher flow conditions in March and April 2019, sites upstream and downstream of the perched culvert included deeper (80–85 cm) pools as well as riffles with estimated velocities exceeding 40 cm/s (Table 3). Stream temperatures measured on any given sampling date were the same or within 1°C upstream compared with downstream of the perched culvert. Among all sites, water temperatures during sampling varied from 11 to 14° C in March, and 16 to 17°C in April, to 20 to 23°C in July, September, and October. Stream pH measurements at locations A and B in July 2018 were 6.53 and 7.12, respectively, and we did not observe evidence of precipitated ferric hydroxide, which presumably was the orange precipitate observed during and immediately after the fish kill.

Expected versus observed fish species recolonization

Thirty three species were represented in archived collection records from Shoal Creek and nearby tributaries. We excluded from analyses two species (Striped Shiner Luxilus chrysocephalus, Brown Bullhead Ameiurus nebulosus) that were each represented in a single sample without count data, leaving 31 taxa represented in the data sets for both observation frequency and capture abundance (Table 4). All four predictors (observation frequency and capture abundance in Shoal Creek and in tributary streams) were positively associated with observed occurrence in Flat Creek (Table 5). The best-supported model was based on observation frequency in tributary samples, although observation frequency and capture abundance in tributaries were nearly equally supported as predictors of observed species occurrence in Flat Creek (Table 5).

Eight of the 11 species observed postpoisoning in Flat Creek were among the 9 taxa most-frequently observed or captured in greatest numbers (Figure 2) in samples from other Shoal Creek tributaries. The remaining three taxa observed in Flat Creek, Bluegill, Yellowfin Shiner N. lutipinnis and Redbreast Sunfish L. auritus, were observed in approximately one out of three or four tributary samples (Figure 2). Two taxa not observed in Flat Creek (Coosa Shiner N. xaenocephalus and Blackbanded Darter P. nigrofasciata) occurred in at least half of the tributary samples and a third species, Alabama Shiner Cyprinella callistia, had among the highest numbers of individuals captured in tributary samples (Figure 2). Finally, Bronze Darter Percina palmaris, Bluehead Chub Nocomis leptocephalus, and Redeye Bass Micropterus coosae occurred relatively frequently in tributary samples (Figure 2) and also were specifically identified among the mortalities recovered following the Flat Creek fish kill (Durniak 2018) but were not observed in our postpoisoning samples. We infer that ≥6 fish species (i.e., Coosa Shiner, Blackbanded Darter, Alabama Shiner, Bronze Darter, Bluehead Chub, Redeye Bass) would be expected to occur in Flat Creek in addition to the 11 taxa observed during this study.

Our collections in Flat Creek indicate slow recolonization by fishes following a catastrophic poisoning event. Samples taken over a 14-mo period, from 4 to 18.5 mo postpoisoning, recovered 11 fish species; however, 9 of these taxa were represented only by relatively few, large individuals. Eight of the 11 fish species found in Flat Creek since the poisoning event have only been observed downstream from a perched, culverted road-crossing. In contrast, GDNR biologists who responded to the poisoning event collected >980 dead individuals of ≥8 fish species, not including small-bodied minnows and sunfishes (Lepomis spp.) that were not identified to species, along ∼5 km of Flat Creek downstream from the spill. These observations during the poisoning event in Flat Creek included dead “darters, sculpins, chubs, suckers, sunfish, and salamanders” near our sites B and C, upstream from the perched culvert (Durniak 2018). Comparing our observations with data collected during the fish kill and with fish collection data from nearby tributaries provides evidence of slow and incomplete recovery of the fish fauna in Flat Creek even 18 mo following the poisoning event, and that recovery has been additionally impeded by a perched culvert that blocks upstream fish passage at least during low streamflow.

Slow recovery in Flat Creek contrasts with observed fish repopulation of defaunated stream reaches that are open to colonization from upstream and downstream sources. For example, fishes eliminated from a small prairie stream by a manure spill had recovered to similar abundances and species richness within 8 mo of the event (Meade 2004). Similarly, 31 of 42 possible fish species returned within 1 y to a Hungarian river after populations were decimated by an industrial spill (Erős et al. 2015). In contrast, fish recovery from a diesel oil spill in a section of river between two dams required >4 y (Kubach et al. 2011). In this case, tributaries to the kill zone were the primary sources of colonizers. We suspect that slow recovery in Flat Creek is at least partly a consequence of the absence of upstream or tributary sources for the majority of recolonizing fishes (Creek Chubs and Largescale Stoneroller being possible exceptions). Similarly, Albanese et al. (2009) reported slower fish recovery following experimental defaunation in a headwater compared with a downstream site.

Long-term effects of the chemical spill on water quality or habitat productivity are unknown; however, several observations suggest that slow recovery of the fish fauna in Flat Creek may not be a result of residual chemicals. The acid initially formed by the spill would have been diluted over time along the length of Flat Creek downstream from the spill. Measured pH values appeared normal when we sampled 4 mo following the event, and in fact had reached ∼6 in a lower reach of Flat Creek 3 d following the spill (Durniak 2018). Although it is possible that total iron remains elevated in Flat Creek, the harmful effects of ferric hydroxide (formed by solution of ferric chloride in water) appear to be mediated by physical clogging of animal gills and benthic habitats (Dalzell and MacFarlane 1999; Cadmus et al. 2018), and we have not observed ferric hydroxide precipitate when sampling in Flat Creek. It is plausible that seasonal high-flow events have flushed much of the precipitate from the system. Additionally, we have observed salamanders, crayfishes and a variety of aquatic insects during sampling, as well as abundant juvenile Creek Chubs, all of which we would expect to be vulnerable to smothering effects of ferric hydroxide. Thus, although there possibly are lingering effects of the spill on water quality, we conclude that slow recovery of the fish fauna in Flat Creek most probably reflects delayed fish recolonization.

The perched culvert on Flat Creek likely forms at least a partial barrier to fish recolonization from downstream. Counts of dead individuals recovered upstream of the culvert (Durniak 2018 and B. Albanese, GDNR, unpublished data) show that prior to the chemical spill, the upstream portion of Flat Creek supported populations of “shiners” and at least four other taxa that we have either not observed (Coosa Bass) or only observed downstream from the culvert postpoisoning (Cherokee Darter, Banded Sculpin, Alabama Hogsucker). The three taxa that have persisted upstream from the culvert (Creek Chub, Largescale Stoneroller, Bluegill) may have colonized from very small tributaries that enter Flat Creek near sites A and B. Possibly these two small streams, or others, provided local refugia during the spill. We also suspect that crayfishes and salamanders were able to exit the stream during the spill and thus avoid local extirpation. It is possible that other fish species will eventually pass upstream through the perched culvert. Norman et al. (2009) documented upstream and downstream movement past similarly perched box-culverts by three species of small-bodied minnows when flows had been high enough to submerge the concrete sill. Even so, fish passage at the perched box culverts in that study were notably lower than at a bottomless (i.e., not perched) culvert (Norman et al. 2009), and other researchers have similarly demonstrated that perched culverts impede passage or are associated with fewer species in upstream assemblages (Nislow et al. 2011).

The fish species that have been observed in Flat Creek since the poisoning event mostly are those expected to occur based on observation frequencies in other, nearby tributary streams. Three additional fish species expected given occurrences in other tributaries—Bluehead Chub, Redeye Bass, and Bronze Darter—were in fact recovered dead in Flat Creek by GDNR biologists who responded to the poisoning event. We do not know whether other fish species frequently observed in nearby tributaries—such as Blackbanded Darter, Coosa Shiner, and Alabama Shiner—also occurred in Flat Creek at the time of the chemical spill. However, including these taxa with those observed suggests that ≤∼65% of the expected fish species were observed 18 mo postpoisoning. Other studies have found evidence of delayed recovery by particular fishes, including darters, following fish kills (Olmsted and Cloutman 1974; Ensign et al. 1997).

Several factors may have slowed the recovery of fishes even downstream of the perched culvert in Flat Creek. One factor is low mobility of the fish taxa involved (Albanese et al. 2009). Sculpins (Cottus spp.), for example, were reported as abundant in the fish kill (and are ubiquitous and abundant in collections from nearby tributaries), and yet we observed only a single, large individual in Flat Creek postpoisoning. Sculpins are generally sedentary (Petty and Grossman 2004), although mark–recapture and genetics studies have shown that individuals do occasionally disperse long (e.g., >500 m) distances (Hudy and Shiflet 2009; Lamphere and Blum 2012; Wells et al. 2017). Mobility has not been measured for most of the other taxa found in Flat Creek, although our observations of individuals >1 km upstream from the mouth of the stream provide evidence that at least adults of these species undertake long movements. Another factor correlated with recolonization is abundance in the source populations (Albanese et al. 2009, Erős et al. 2015), which we do not know for Flat Creek. Shoal Creek, the immediate downstream source of colonists, supports diverse and abundant fishes, however sampling to assess current fish abundances near the confluence with Flat Creek could be informative. Additionally, Flat Creek flows over a series of bedrock cascades near the confluence with Shoal Creek. Possibly these cascades have impeded upstream fish movement. A number of studies have reported that stream fish are more likely to disperse during higher flows (Albanese et al. 2004; Petty and Grossman 2004; Peterson and Shea 2015), whereas others have observed upstream fish dispersal in mountain streams during low flows (Grossman et al. 2010). Periods of elevated flow and low flow are evident in gage data for nearby streams (Amicalola Cr, USGS gage 02390000; Russell Cr, USGS gage 02388985) during summer and autumn 2018 following the spill, and during spring and summer 2019. We do not know the passage conditions during high flows at the cascades in lower Flat Creek or through the perched culvert, but it appears unlikely that an absence of high flows (or of low flows) is the reason for slow recovery in Flat Creek.

The accidental spill of ferric chloride that resulted in an extensive fish kill in Flat Creek has provided evidence of slow recovery of a headwater fish fauna, even in the absence of human-built barriers to fish movement, when the population loss extends over multiple kilometers and recolonization sources are primarily downstream. Structures that impede fish passage are likely to slow recovery further. Additional case studies may reveal whether this result applies generally to headwater streams. Slow recovery could make species that primarily inhabit or maintain greatest abundances in headwaters, including multiple at-risk fishes, particularly vulnerable to the threat of accidental spills that result in local population extirpation.

Please note: The Journal of Fish and Wildlife Management is not responsible for the content of functionality of any archived material. Queries should be directed to the corresponding author for the article.

To cite this archived material, please cite both the journal article (formatting found in the Abstract section of this article) and the following recommended format for the archived material.

Freeman MC. 2021. Slow recovery of headwater-stream fishes following a catastrophic poisoning event. U.S. Geological Survey. Available: https://doi.org/10.5066/P92ASIYQ

Data A1. Location description, latitude and longitude for five localities sampled for fishes in Flat Creek (Dawson County, Georgia) on six dates following a catastrophic fish kill caused by a chemical spill near the headwater origin of Flat Creek.

Available: https://doi.org/10.5066/P92ASIYQ (Flat_Cr_Sample_Locales, 0.57 KB CSV)

Data A2. Point measurements of water depth (cm) and velocity (cm/s) made during fish sampling events on four dates in 2019 at five localities in Flat Creek (Dawson County, Georgia).

Available: https://doi.org/10.5066/P92ASIYQ (Flat_Cr_Depth_Velocity_Values, 36 KB CSV)

Data A3. Fish species captured during 13 sampling events in 2018 and 2019 at 5 localities in Flat Creek (Dawson County, Georgia). Water temperature (° Celsius) and number of kick-sets conducted during each sampling event are also shown, along with the number of individuals and range in standard lengths (mm) for each species captured.

Available: https://doi.org/10.5066/P92ASIYQ (Flat_Cr_Species_by_collection, 6 KB CSV)

Funding to track fish recolonization in the study stream was provided by Gold Creek Foods, LLC (June 2018, to University of Georgia) and by the U.S. Geological Survey Quick Response Program (March 2019). Robin Goodloe (U.S. Fish and Wildlife Service [USFWS]) and Brett Albanese (Georgia Department of Natural Resources) provided information on the fish kill. Gary Barr, City of Dawsonville, provided access to Flat Creek. We appreciate field help provided by Maggie England-Johns and Laura Rack (University of Georgia), Anthony Sowers (USFWS), and Brett Albanese. The Georgia Museum of Natural History generously provided historical collection records for the Shoal Creek system. David Smith, John Young, Corey Dunn, two anonymous reviewers, and the Associate Editor provided helpful comments on earlier drafts of this paper.

Any use of trade, product, website, or firm names in this publication is for descriptive purposes only and does not imply endorsement by the U.S. Government.

Abell
RA,
Olson
DM,
Dinerstein
E,
Hurley
PT,
Diggs
JT,
Eichbaum
W,
Walters
S,
Wettengel
W,
Allnutt
T,
Loucks
CJ,
Heao
P,
Taylor
CL.
2000
.
Freshwater ecoregions of North America: a conservation assessment
.
Washington, D.C
.:
World Wildlife Fund
.
Albanese
B,
Angermeier
PL,
Dorai-Raj
S.
2004
.
Ecological correlates of fish movement in a network of Virginia streams
.
Canadian Journal of Fisheries and Aquatic Sciences
61
:
857
869
.
Albanese
B,
Angermeier
PL,
Peterson
JT.
2009
.
Does mobility explain variation in colonisation and population recovery among stream fishes?
Freshwater Biology
54
:
1444
1460
.
Barton
SD,
Powers
SL.
2010
.
Life-history aspects of the Cherokee Darter, Etheostoma scotti Actinopterygii: Percidae), an imperiled species in northern Georgia
.
Southeastern Naturalist
9
:
687
698
.
Burnham
KP,
Anderson
DR.
2002
.
Model selection and multimodel inference: a practical information-theoretic approach. Second edition
.
New York
:
Springer-Verlag
.
Cadmus
P,
Brinkman
SF,
May
MK.
2018
.
Chronic toxicity of ferric iron for North American aquatic organisms: derivation of a chronic water quality criterion using single species and mesocosm data
.
Archives of Environmental Contamination and Toxicology
74
:
605
615
.
Dalzell
DJB,
MacFarlane
NAA.
1999
.
The toxicity of iron to Brown Trout and effects on the gills: a comparison of two grades of iron sulphate
.
Journal of Fish Biology
55
:
301
315
.
Detenbeck
NE,
DeVore
PW,
Niemi
GJ,
Lima
A.
1992
.
Recovery of temperate-stream fish communities from disturbance: a review of case studies and synthesis of theory
.
Environmental Management
16
:
33
53
.
Durniak
J.
2018
.
Fish kill investigation: Flat Creek, Dawson County, Etowah River Drainage
.
Social Circle
:
Georgia Department of Natural Resources, Wildlife Resources Division
.
Elkins
D,
Sweat
SC,
Kuhajda
BR,
George
AL,
Hill
KS,
Wenger
SJ.
2019
.
Illuminating hotspots of imperiled aquatic biodiversity in the southeastern US
.
Global Ecology and Conservation
19
:
e00654
.
Ensign
WE,
Leftwich
KN,
Angermeier
PL.
1997
.
Factors influencing stream fish recovery following a large-scale disturbance
.
Transactions of the American Fisheries Society
126
:
895
907
.
Erős
T,
Takács
P,
Czeglédi
I,
Sály
P,
Specziár
A.
2015
.
Taxonomic- and trait-based recolonization dynamics of a riverine fish assemblage following a large-scale human-mediated disturbance: the Red Mud disaster in Hungary
.
Hydrobiologia
758
:
31
45
.
[GDNR] Georgia Department of Natural Resources, Wildlife Resources Division.
Undated. Georgia rare natural element data portal.
Grossman
GD,
Ratajczak
RE
Jr,
Farr
MD,
Wagner
CM,
Petty
JT.
2010
.
Why there are fewer fish upstream
.
American Fisheries Society Symposium
73
:
63
81
.
Hudy
M,
Shiflet
J.
2009
.
Movement and recolonization of Potomac Sculpin in a Virginia stream
.
North American Journal of Fisheries Management
29
:
196
204
.
Kubach
KM,
Scott
MC,
Bulak
JS.
2011
.
Recovery of a temperate riverine fish assemblage from a major diesel oil spill
.
Freshwater Biology
56
:
503
518
.
Kwong
RWM,
Kumai
Y,
Perry
SF.
2014
.
The physiology of fish at low pH: the Zebrafish as a model system
.
The Journal of Experimental Biology
217
:
651
662
.
Lamphere
BA,
Blum
MJ.
2012
.
Genetic estimates of population structure and dispersal in a benthic stream fish
.
Ecology of Freshwater Fish
21
:
75
86
.
Leasure
DR.
2014
.
Applications of a new Geodata Crawler for landscape ecology: from mapping natural stream hydrology to monitoring endangered beetles. Doctoral dissertation
.
Fayetteville
:
University of Arkansas
.
Lytle
DA,
Peckarsky
BL.
2001
.
Spatial and temporal impacts of a diesel fuel spill on stream invertebrates
.
Freshwater Biology
46
:
693
704
.
McDonald
DG,
Freda
J,
Cavdek
V,
Gonzalez
R,
Zia
S.
1991
.
Interspecific differences in gill morphology of freshwater fish in relation to tolerance of low-pH environments
.
Physiological Zoology
64
:
124
144
.
Meade
R.
2004
.
Fish and invertebrate recolonization in a Missouri prairie stream after an acute pollution event
.
North American Journal of Fisheries Management
24
:
7
19
.
Nislow
KH,
Hudy
M,
Letcher
BH,
Smith
EP.
2011
.
Variation in local abundance and species richness of stream fishes in relation to dispersal barriers: implications for management and conservation
.
Freshwater Biology
56
:
2135
2144
.
Norman
JR,
Hagler
MM,
Freeman
MC,
Freeman
BJ.
2009
.
Application of a multistate model to estimate culvert effects on movement of small fishes
.
Transactions of the American Fisheries Society
138
:
826
838
.
Olmsted
LL,
Cloutman
DG.
1974
.
Repopulation after a fish kill in Mud Creek, Washington County, Arkansas following pesticide pollution
.
Transactions of the American Fisheries Society
103
:
79
87
.
Peterson
JT,
Shea
CP.
2015
.
An evaluation of the relations between flow regime components, stream characteristics, species traits, and meta-demographic rates of warm-water-stream fishes: implications for aquatic resource management
.
River Research and Applications
31
:
1227
1241
.
Petty
JT,
Grossman
GD.
2004
.
Restricted movement by Mottled Sculpin (Pisces: Cottidae) in a southern Appalachian stream
.
Freshwater Biology
49
:
631
645
.
R Core Team.
2020
.
R: a language and environment for statistical computing
.
Vienna
:
R Foundation for Statistical Computing
.
Storey
CM,
Porter
BA,
Freeman
MC,
Freeman
BJ.
2006
.
Analysis of spawning behavior, habitat, and season of the federally threatened Etheostoma scotti, Cherokee Darter (Osteichthyes:Percidae)
.
Southeastern Naturalist
5
:
413
424
.
[ESA] U.S. Endangered Species Act of 1973, as amended, Pub. L. No. 93-205, 87 Stat. 884 (Dec. 28, 1973)
.
[USFWS] U.S. Fish and Wildlife Service.
1994
.
Endangered and threatened wildlife and plants; determination of Threatened Status for the Cherokee Darter and Endangered Status for the Etowah Darter
.
Federal Register
59
:
65505
65512
.
Use of Fishes in Research Committee (joint committee of the American Fisheries Society, the American Institute of Fishery Research Biologists, and the American Society of Ichthyologists and Herpetologists).
2014
.
Guidelines for the use of fishes in research
.
Bethesda, Maryland
:
American Fisheries Society
.
Wells
WG,
Johnson
TC,
Gebhard
AE,
Paine
RTR,
Hix
LA,
Ferrell
HN,
Engle
AN,
Perkin
JS.
2017
.
March of the sculpin: measuring and predicting short-term movement of Banded Sculpin Cottus carolinae
.
Ecology of Freshwater Fish
26
:
280
291
.
Yang
L,
Jin
S,
Danielson
P,
Homer
C,
Gass
L,
Bender
SM,
Case
A,
Costello
C,
Dewitz
J,
Fry
J,
Funk
M,
Granneman
B,
Liknes
GC,
Rigge
M,
Xian
G.
2018
.
A new generation of the United States National Land Cover Database: requirements, research priorities, design, and implementation strategies
.
ISPRS Journal of Photogrammetry and Remote Sensing
146
:
108
123
.

The findings and conclusions in this article are those of the author(s) and do not necessarily represent the views of the U.S. Fish and Wildlife Service.

Author notes

Citation: Freeman M, Elkins D, Maholland P, Butler Z, Kleinhans M, Skaggs J, Stowe E, Straight C, Wenger S. 2021. Slow recovery of headwater-stream fishes following a catastrophic poisoning event. Journal of Fish and Wildlife Management 12(2):362–372; e1944-687X. https://doi.org/10.3996/JFWM-20-080