Recovery of Freshwater Invertebrates in Alpine Lakes and Streams Following Eradication of Nonnative Trout with Rotenone

Invasive species have caused global declines in abundance and biodiversity of native fauna across aquatic and terrestrial landscapes (Simberloff et al. 2013). Freshwater organisms are among the most imperiled animals worldwide (World Wildlife Fund [WWF] 2016), and fish taxa are particularly vulnerable to invasive species through predation, competition, and hybridization leading to extinctions and biotic homogenization (Muhlfeld et al. 2014). Given the threat to ecosystems and economies that human-induced invasions of exotic species present, fish and wildlife managers are increasingly implementing remedies to control the spread of invasive species (Vander Zanden and Olden 2008; Vitule et al. 2009). To control the threat of invasive fish species to sensitive and at-risk taxa in aquatic systems of high conservation value, chemical eradication with the piscicide rotenone is often used due to its effectiveness in lentic and lotic systems (Vinson et al. 2010).

Since 1934, North American fisheries managers have used rotenone as a tool for quantification of fish populations, disease control, eradication of undesirable fish populations, and conservation of native species (McClay 2000; Finlayson et al. 2018). A botanical extract found in many tropical plants in the Leguminosae family (Ling 2003), rotenone is the key component of commercially available piscicide formulations. Indigenous peoples along the Pacific Ring of Fire first discovered rotenone's lethal effect on aquatic organisms and used it to harvest fish for human consumption (Krumholz 1948). Rotenone affects gill-breathing organisms by passing into the bloodstream via osmoregulation in the lamellae and is transported to cells, thereby inhibiting cellular respiration (Lindahl and Öberg 1961).

The effectiveness of rotenone to control fish is well documented, but it is also lethal at piscicidal concentrations to ecologically important nontarget aquatic organisms including benthic invertebrates (Skorupski 2011; Lan Phan et al. 2018; Bellingan et al. 2019), zooplankton (Dalu et al. 2015), and amphibians (Fried et al. 2018). Vinson et al. (2010) reviewed the impacts of rotenone on freshwater aquatic invertebrate assemblages and found effects ranging from minor to substantial as well as variation in recovery time among taxa—from only a few months to several years. Indeed, recovery of aquatic invertebrate assemblages to rotenone exposure tends to be highly variable due to differences in taxa-specific sensitivity to water quality, differing environmental conditions among studies, and low statistical power resulting from small sample sizes (Vinson et al. 2010). Few studies have implemented the broad temporal and spatial study design necessary to assess recovery time of invertebrate abundance and diversity beyond individual systems at short time scales.

The South Fork Flathead River drainage (Montana, USA; Figure 1) is a region of high conservation value for native Westslope Cutthroat Trout Oncorhynchus clarkii lewisi and Bull Trout Salvelinus confluentus. In addition, the Bull Trout has threatened status (U.S. Fish and Wildlife Service [USFWS] 1998) under the Federal Endangered Species Act (ESA 1973, as amended). Both species are highly vulnerable to invasive species through hybridization, competition, and predation (Donald and Alger 1993; Leary et al. 1993; Shepard et al. 2005)—trends exacerbated by climate change (Muhlfeld et al. 2014; Kovach et al. 2017). In the South Fork Flathead River drainage, the primary threat to Westslope Cutthroat Trout is hybridization with nonnative Rainbow Trout Oncorhynchus mykiss and Yellowstone Cutthroat Trout Oncorhynchus clarkii bouveri stocked in historically fishless alpine lakes during the early 20th century. This lake stocking resulted in naturally reproducing populations of nonnative trout, and their subsequent movement down the outlet stream over barrier waterfalls resulted in the spread of hybridization with downstream populations of native Westslope Cutthroat Trout (Bonneville Power Administration 2005). Undertaking a watershed-scale effort to secure the South Fork Flathead drainage as a range-wide stronghold for native trout populations, the Montana Department of Fish, Wildlife and Parks proposed eradication of nonnative and hybridized fish with the piscicide rotenone and restocking of Westslope Cutthroat Trout to create local genetic reserves for this species and provide continued opportunities for recreational fishing. Although rotenone is a commonly used fisheries management tool for native fish restoration (Rayner and Creese 2006; Clancey et al. 2019), social and biological concerns for the potential collateral effects on nontarget taxa required rigorous study, monitoring, and evaluation.

From 2002 to 2020, we quantified and evaluated taxonomic diversity and density of freshwater benthic invertebrate and zooplankton assemblages in 13 alpine lakes and their outlet streams from 1 y prerotenone treatment up to 8 y postrotenone treatment. Our objective was to quantify the relative recovery of the benthic invertebrate and zooplankton assemblages following rotenone exposure within the context of a long-term, landscape-scale native trout restoration project. Based on the variability of results in previous studies, we hypothesized that aquatic invertebrates would experience no substantial long-term decreases in density and diversity after rotenone exposure.

The South Fork Flathead River watershed encompasses approximately 4,320 km² in northwestern Montana. The drainage is a predominantly pristine aquatic system in the Flathead National Forest, much of which is protected as wilderness in the Bob Marshall Wilderness complex (Figure 1). The South Fork Flathead River is free flowing for 89 river kilometers before reaching the 53-km-long Hungry Horse Reservoir constructed in 1956. The South Fork watershed comprises more than half of the remaining genetically pure, interconnected populations of Westslope Cutthroat Trout in Montana, making this watershed a regional stronghold for the species (Bonneville Power Administration 2005). To protect this unique and highly prized watershed, the Montana Department of Fish, Wildlife and Parks partnered with the U.S. Forest Service and Bonneville Power Administration to implement an 11-y project (from 2007 to 2017) to treat 13 headwater lakes and associated outlet creeks with rotenone and restock the lakes the following summer with genetically pure Westslope Cutthroat Trout. We treated all the lakes with a 5% liquid formulation of rotenone under trade names Prenfish or CFT Legumine at the concentration of 50 ppb of rotenone (Table S1, Supplemental Material).

We collected benthic invertebrate samples from randomly selected riffles in lake outlet streams within the first 100 m downstream of the lake. We chose riffles in the lake outlet streams as monitoring sites for two reasons: 1) flowing water habitats frequently possess more structural heterogeneity and increased taxonomic diversity of benthic invertebrate assemblages (Thorp and Covich 1991); and 2) invertebrates at these sites are primarily composed of sensitive taxa and expected to experience the most severe effects from rotenone, both in terms of concentration and exposure duration. We collected samples by using a 0.09-m² Surber sampler with 500-μm mesh, selecting riffle areas with appropriate width, depth, and substrate size to allow for successive Surber plots in adjacent and/or upstream diagonal movement to cover the left, center, and right of the riffle. Each sample consisted of two Surber plots (0.18 m²), and total sample area comprised one to three individual samples (sample area, 0.18–0.55 m²). We sampled lakes located in wilderness 1 mo/y due to logistical constraints. We sampled all other lakes monthly (in June, July, August, and September) 1 y before rotenone treatment, and we collected repeat samples both seasonally and annually for up to 8 y posttreatment. Each sample was kept separate and preserved in 95% ethanol for future analysis. We enumerated sample contents and identified to the lowest practical taxonomic unit, typically genus. For practicality, we subsampled samples that contained an exceedingly large number of dipteran larvae by portioning the contents to estimate the density of dipteran taxa. In total, we collected 63 pretreatment samples and 144 posttreatment samples in the outlet streams from the 13 lakes (Table S2, Supplemental Material).

We collected zooplankton samples from lake limnetic zones by using a Wisconsin-style plankton net with 80-μm mesh, equipped with a removable 200-mL mesh bucket. Each sample consisted of two 15-m vertical tows taken from an inflatable pack raft. If the lake depth was less than 15 m, we took the tows from the deepest part of the lake, determined from bathymetric survey maps and a battery-operated, handheld sonar device. We collected one to two samples at each sampling event across all lakes. We sampled zooplankton at the same temporal interval as we sampled the benthic invertebrates, at a given lake. We kept each sample separate and preserved in 95% ethanol for future analysis. We transferred samples to a Ward counting wheel for enumeration and identified to the lowest practical taxonomic unit, typically genus. With a few exceptions, we identified the entire contents of the samples. When necessary, we subsampled contents by using a 1-mL Hensen Stempel pipette and a mean derived from three to five subsamples. In total, we collected 100 pretreatment samples and 127 posttreatment samples from the 13 lakes (Table S2).

To evaluate the effects of rotenone on benthic invertebrate and zooplankton assemblages, we compared pre- and posttreatment densities and two measures of diversity: taxa richness and Shannon's diversity index (H) (Shannon 1948). Taxa richness was the only diversity measure calculated for zooplankton due to the low number of unique taxa observed. We analyzed differences between years for individual lakes by using a Kruskal–Wallis nonparametric analysis of variance (ANOVA). Sampling effort across all the lakes was inconsistent in the number of samples collected and number of years of pre- and posttreatment sampling (Table S2). The unbalanced design produced heteroscedastic results and prevented any robust statistical assessment for each lake.

Next, we pooled and compared pre- and posttreatment mean density and diversity values for each year using a 1-way ANOVA and post hoc Tukey's test. To meet parametric assumptions of normality and independence, we pooled mean values of density and diversity from all lakes within each sample year, which standardized sample sizes across lakes and years. We also pooled posttreatment years 4–6 and 7–8 because sample sizes were small. Benthic invertebrate mean density values were log transformed to normalize extreme outliers. We determined the annual change from pretreatment mean density and diversity estimates for each lake by subtracting the mean pretreatment values from the mean posttreatment values for each year. A P value less than or equal to 0.05 was used to evaluate significant change between years. We generated all statistics and figures by using R version 2.6.2 (R Core Team 2018) and ‘tidyverse' packages (Wickham et al. 2019). The complete benthic invertebrate (Data S1, Supplemental Material) and zooplankton (Data S2, Supplemental Material) datasets can be found in the Supplemental Material.

The Kruskal–Wallis ANOVA for individual lakes revealed changes in posttreatment density and diversity, but the magnitude and direction of change varied lake by lake (Table 1). There were consistent decreases in Ephemeroptera, Plecoptera, and Trichoptera (EPT) density at 8 lakes and in EPT diversity at 10 lakes 1–2 y posttreatment and decreases in zooplankton density at nine lakes and in genera richness at 5 lakes 1 y posttreatment. We observed no consistent changes in density and diversity in other benthic invertebrate taxa.

Compared with pretreatment levels, mean benthic invertebrate density significantly increased after rotenone treatment (ANOVA: P = 0.03). Although the Tukey's honestly significant difference (HSD) test did not statistically differentiate between years, Figure 2A displays a trend in annual increases in benthic invertebrate density posttreatment. Mean EPT density significantly decreased and then increased after treatment (ANOVA: P = 0.01). Although the Tukey's HSD test did not statistically differentiate between years, Figure 2A shows mean EPT density decreased 1–2 y after treatment, recovered to baseline by 3 y posttreatment, and increased above baseline 4–8 y posttreatment. Mean decreases in EPT density 1–2 y posttreatment were largely offset by increases in mean dipteran density such that mean benthic invertebrate density increased above baseline level (Figure 2A). Mean Shannon's diversity index decreased 1–2 y posttreatment, but was not significant (ANOVA: P = 0.19), and recovered to baseline by year 3 (Figure 2B). Mean family richness significantly decreased after treatment (ANOVA: P = 0.03). Although the Tukey's HSD test did not statistically differentiate between years, Figure 2C shows mean family richness remained below baseline 3 y posttreatment and returned to baseline 4–8 y posttreatment.

Compared with pretreatment levels, mean zooplankton density (Figure 3A) did not significantly change after rotenone treatment (ANOVA: P = 0.93). Similarly, mean zooplankton genera richness (Figure 3B) was similar to baseline across all years following rotenone treatment (ANOVA: P = 0.50).

Surveys conducted 1–2 y after rotenone treatment found mean benthic invertebrate diversity and richness were lower than pretreatment levels, largely due to declines of EPT taxa (Figures 2B and 2C). Despite this short-term decrease in diversity and richness, overall benthic invertebrate densities remained high posttreatment, initially due to increased numbers of larval Diptera and, in subsequent years, from the recovery of EPT taxa (Figure 2A). The EPT numbers eventually exceeded pretreatment levels, a pattern similar to that observed in a montane stream treated with rotenone for a fish removal project in the Madison River basin of Montana (Clancey et al. 2019). A similar pattern in community response following rotenone treatment also was observed by Lam Pham et al. (2018), who documented rapid community recovery within 4–12 mo postrotenone treatment. Mean zooplankton genera richness decreased 1 y posttreatment and recovered in subsequent years (Figure 3B). We noted largely unchanged zooplankton density in any posttreatment year (Figure 3A), corroborating results found by Eilers et al. (2011). Other studies reported recovery time of zooplankton in rotenone-treated lakes ranging from 3 mo (Kiser et al. 1963) to 3 y (Anderson 1970).

Patterns in freshwater benthic invertebrate recovery rates are likely attributable to taxon-specific differences in sensitives to rotenone (Kjaerstad et al. 2016; Lam Pham et al. 2018). Although dipterans are relatively tolerant of degraded water quality, EPT taxa are sensitive to aquatic pollutants and habitat perturbations, making the EPT index a widely used and sensitive biological indicator of water quality (Rosenburg and Resh 1993). Sensitivity to rotenone exposure is one explanation for the stability in dipteran density compared with the decrease in EPT density posttreatment. Another possible explanation for variable taxon-specific recovery rates is the effect of terrestrial adult dispersal during and after the rotenone treatment (Dalu et al. 2015). Although zooplankton have shown high mortality to rotenone exposure (Melaas et al. 2001), their rapid recovery in our study might be due to their ability to produce resting or dormant eggs that can persist in the substrate of lakes during periods of unfavorable environmental conditions (Hairston et al. 1995).

Our study examined potential long-term impacts of rotenone treatment on benthic invertebrate and zooplankton assemblages across 13 remote alpine lakes. Benthic invertebrate and zooplankton samples taken 4–8 y posttreatment indicated no long-term impacts to density or taxa richness. Increases in benthic invertebrate density in years following treatment may be explained by initially lower numbers of insectivorous fishes poststocking (Clancey et al. 2019) and increased niche space related to periodic disturbance and biomass differences between early-recovering and late-recovering species. Another explanation could be the role of downstream drift from untreated water in the catchment. Our study lakes were in high-elevation headwater systems with varying degrees of freshwater inputs.

Given the rugged and remote terrain of our study area, treated lakes were resource intensive to sample, which meant consistent sampling of control waterbodies and standardized sampling effort was not always possible across all years of the study. In general, pretreatment means consist of fewer individual samples than posttreatment means. Both Shannon's diversity index and taxa richness can be sensitive to sample size (Soetaert and Heip 1990), which is a potential concern for certain waterbodies and years in our study. Nonetheless, the consistent lack of long-term change of diversity in response to rotenone treatment across lakes increases our confidence that calculated diversity indices adequately reflect actual invertebrate diversity and is consistent with reported results from other studies (e.g., Lam Pham et al. 2018; Clancey et al. 2019).

Rotenone is an effective tool to eliminating invasive fish populations and their associated negative impacts on indigenous fauna (Vinson et al. 2010). To date, numerous studies have documented the rapid recovery of nontarget aquatic invertebrates (e.g., Kjaerstad and Arnekleiv 2003; Hamilton et al. 2009; Eilers et al. 2011; Skorupski 2011). Although multiple studies in Utah have demonstrated much longer recovery times or extirpation of certain taxa (Binns 1967; Mangum and Madrigal 1999; Whelan 2002), these projects used multiple treatments and higher rotenone concentrations compared with our study. It is also possible that the arid Utah region offered far fewer sources for invertebrate recolonization than the South Fork Flathead River basin. Therefore, invertebrate recovery time is likely related to rotenone application frequency and concentration and may be biome specific. Our results are likely applicable to most alpine lakes in nonarid regions where a single rotenone treatment was conducted within the range of manufacturer-prescribed concentrations to control salmonid fishes.

Fisheries professionals recognize the purpose and need for biological sampling and monitoring of nontarget organisms and have adopted specific monitoring requirements for rotenone treatments to inform future treatments and minimize impacts (Finlayson et al. 2018). Based on our results, a multisample effort, both spatially and temporally, up to 3 y posttreatment was adequate to determine whether benthic invertebrate density and diversity in lake outlets recovered to approximate pretreatment levels. It does not appear that zooplankton assemblages similar to those in the alpine lakes of the Flathead Basin require extensive posttreatment monitoring. However, before rotenone treatment, sampling could determine whether rare or at-risk invertebrates are present and facilitate treatments tailored to minimize impacts to these taxa. Our study adds to the growing body of literature documenting a lack of substantial long-term impacts of rotenone treatments to nontarget organisms, including benthic invertebrates, zooplankton, and amphibians (Woodford et al. 2013; Fried et al. 2018; Lan Pham et al. 2018). Implicit in the long-term success of most rotenone treatments is the establishment of a healthy native fish community that requires a healthy invertebrate assemblage to persist. Our research suggests that application of 50 ppb of rotenone has no long-term effects on freshwater invertebrate density or diversity in headwater lakes and outlet streams. Future studies aimed at understanding modes of macroinvertebrate recolonization will be useful for predicting community recolonization rates in other systems exposed to piscicide treatments.

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

Table S1. Physical characteristics, rotenone application totals, fish stocking rates, and walking distance for 13 South Fork Flathead alpine lakes treated with rotenone from 2007 to 2017.

Available: http://doi.org/10.3996/JFWM-20-040.S1 (17 KB DOCX)

Table S2. Benthic invertebrate and zooplankton sample dates (month/year) pre- and postrotenone treatment and number of samples per date for 13 South Fork Flathead alpine lakes treated with rotenone from 2007 to 2017.

Available: http://doi.org/10.3996/JFWM-20-040.S2 (24 KB DOCX)

Data S1. Dataset of benthic invertebrate samples pre- and postrotenone treatment for 13 South Fork Flathead alpine lakes treated with rotenone from 2007 to 2017 (MS Excel).

Available: http://doi.org/10.3996/JFWM-20-040.S3 (11 KB CSV)

Data S2. Dataset of zooplankton samples pre- and postrotenone treatment for 13 South Fork Flathead alpine lakes treated with rotenone from 2007 to 2017 (MS Excel).

Available: http://doi.org/10.3996/JFWM-20-040.S4 (6 KB CSV)

Reference S1.Bonneville Power Administration. 2005. South Fork Flathead Watershed Westslope Cutthroat Trout Conservation Program final environmental impact statement. Portland, Oregon: Department of Energy. DOE/EIS-0353.

Available: http://doi.org/10.3996/JFWM-20-040.S5 (293 KB PDF)

Reference S2. Ling N. New Zealand Department of Conservation. 2003. Rotenone—a review of its toxicity and use for fisheries management. Wellington, New Zealand: Department of Conservation. Science For Conservation 211.

Available: http://doi.org/10.3996/JFWM-20-040.S6 (156 KB PDF)

Reference S3. Skorupski JA. 2011. Effects of CFT Legumine rotenone on macroinvertebrates in four drainages of Montana and New Mexico. Master's thesis. Denton: University of North Texas.

Available: http://doi.org/10.3996/JFWM-20-040.S7 (4.87 MB PDF)

Reference S4. Whelan JE. Utah Division of Wildlife Resources. 2002. Aquatic macroinvertebrate monitoring results of the 1995 and 1996 rotenone treatments of Manning Creek, Utah. Salt Lake City: Utah Division of Wildlife Resources Publication 02-04.

Available: http://doi.org/10.3996/JFWM-20-040.S8 (450 KB PDF)

This work was funded by Bonneville Power Administration grant 199101903 to Department of Montana Fish, Wildlife and Parks. We thank Jim Deraleau, Gary Michael, and Lynda Fried for assisting with sample collection and lab analysis and all the Journal reviewers and Associate Editor for their expert analysis and editorial contributions.

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.