Efficacy, selectivity, and herbicide concentrations following a whole-lake 2,4-D application targeting Eurasian watermilfoil in two adjacent northern Wisconsin lakes

Abstract

The herbicide 2,4-D (2,4-dichlorophenoxy acetic acid) has been used to control the nonnative aquatic plant Eurasian watermilfoil (Myriophyllum spicatum; EWM) since the 1950s. Although published research evaluates the herbicide's predicted and observed concentration and exposure times in both laboratory and field settings, few data are available evaluating selectivity and long-term efficacy as well as herbicide concentration behavior following large-scale, whole-lake applications. A controlled study was conducted on 2 adjacent oligo-mesotrophic northern Wisconsin lakes to determine the potential efficacy and selectivity of large-scale and low-dose 2,4-D applications. Initial 2,4-D concentrations in both treated lakes were approximately 100 μg/L higher than the nominal lakewide targets of 500 and 275 μg/L, respectively, and the herbicide dissipated and degraded more slowly than predicted. A lakewide regression model relating 2,4-D concentration at monitoring sites to days after treatment (DAT) found the mean half-life of 2,4-D to be 34–41 DAT, and the threshold for irrigation of plants not labeled for direct treatment with 2,4-D (<100 μg/L) was not met until 50–93 DAT. In the lake treated at the higher 2,4-D rate, EWM was not detected for 3 consecutive years posttreatment. Additionally, several native monocotyledon and dicotyledon species also showed sustained significant declines posttreatment. This study is the first to link field-collected 2,4-D concentration measurements to selectivity and long-term efficacy in EWM control following whole-lake management efforts. Although multiyear EWM control was achieved with these single low-dose applications, longer than expected herbicide persistence and impacts to native plants demonstrate the challenges facing aquatic plant managers and the need for additional field studies.

Key words:

• 2,4-dichlorophenoxy acetic acid
• aquatic plant management
• chemical control
• herbicide
• invasive species
• Myriophyllum spicatum
• native macrophytes

The herbicide 2,4-D (2,4-dichlorophenoxy acetic acid) has been used to control the invasive aquatic plant Eurasian watermilfoil (Myriophyllum spicatum; EWM) since the 1950s (Gallagher and Haller 1990). Both a granular ester and several liquid amine 2,4-D formulations are currently registered for aquatic use. The herbicide acts by mimicking the natural plant hormone auxin, resulting in epinastic bending and twisting of stems and petioles followed by chlorosis at growing points, growth inhibition, and necrosis of sensitive species (WSSA 2007). One of the key properties featured in using 2,4-D for EWM control programs has been an inherent selectivity that allows managers to target this invasive dicotyledon with a general lack of activity on numerous submersed monocotyledons (Bates et al. 1985).

Variability in achieving efficacy with 2,4-D is likely a function of concentration and exposure time (CET) relationships; Green and Westerdahl (1990) describe numerous short-term CET scenarios (up to 72 h of exposure) that can occur following various operational treatments. Prior field research has demonstrated efficacy and selectivity of 2,4-D following treatment of plots in larger waterbodies (Lim and Lozoway 1976, Carpentier et al. 1988, Parsons et al. 2001, Wersal et al. 2010). These studies included collection of herbicide concentration data, although the results were likely influenced by rapid dispersion from treatment sites. Moreover, evaluation and reporting on efficacy and selectivity following small field plot treatments have generally only been conducted over the short term. One long-term monitoring effort (5 yr) was undertaken following a 36 ha 2,4-D treatment on Lake Cayuga, New York, but this evaluation did not include any effort to link herbicide concentrations to the treatment outcomes (Miller and Trout 1985). The value in relating measured herbicide concentrations to long-term efficacy and selectivity following large-scale management efforts has been demonstrated in research with the aquatic herbicide fluridone (Getsinger et al. 2002, Madsen et al. 2002, Netherland et al. 2002, Valley et al. 2006, Wagner et al. 2007, Parsons et al. 2009).

EWM continues to be highly invasive in the northern United States and has the ability to grow to nuisance levels in some waterbodies. The problems caused by EWM and the various growth strategies the plant utilizes to support rapid spread and canopy formation are well reviewed (Grace and Wetzel 1978, Smith and Barko 1990, Madsen et al. 1991, Boylen et al. 1999). Herbarium records suggest that EWM became established in Wisconsin in the 1960s and 1970s (Couch and Nelson 1985), and various efforts have been employed to manage nuisance populations throughout the state. Historical management efforts include use of mechanical harvesters, hand pulling, benthic barriers, native weevils, and registered herbicides (Painter 1988, Smith and Barko 1990, Eichler et al. 1995, Newman 2004, Kelting and Laxson 2010).

A review of EWM management by the Wisconsin Department of Natural Resources (WI DNR) suggested that strategies for addressing EWM at a lakewide scale merited further attention. Recent field monitoring efforts by WI DNR staff and collaborators led to the observation that some larger-scale, early-season treatments with 2,4-D were resulting in a much larger area of control than the actual target area. These observations in combination with ongoing collection of posttreatment water samples throughout the state and mesocosm studies evaluating extended exposures to reduced concentrations of 2,4-D suggested that large-scale and low-dose applications of 2,4-D could potentially impact EWM at a lakewide level (Glomski and Netherland 2010). Based on these recent observations, a target lakewide dose of 500 μg/L acid equivalent (ae) was suggested, which is notably lower than recommended label doses of 2000–4000 μg/L ae.

The specific objectives of this study were to examine (1) the efficacy of a low-dose, whole-lake 2,4-D treatment on reducing EWM percent frequency of occurrence and biomass, (2) the selectivity of the treatment for EWM while minimizing damage to native macrophytes, and (3) the longevity of EWM suppression as well as any nontarget native macrophyte impacts. To achieve these objectives, we conducted a field trial in 2 oligo-mesotrophic northern Wisconsin lakes to determine the short- and long-term effects of large-scale and low-dose liquid 2,4-D applications.

Methods

Study sites

Tomahawk and Sandbar lakes (Bayfield County, WI) are oligo-mesotrophic seepage lakes of similar size and depth, separated by a narrow sandbar (∼40 m wide) that prevents the exchange of surface water. EWM was first observed in both lakes in August 2004. The lakes have similar physical features and plant community characteristics (Table 1). We employed a controlled design by applying a whole-lake, low-dose (500 μg/L ae) 2,4-D treatment to one lake (Tomahawk), while the other was retained for comparison as an untreated control (Sandbar). After serving as a untreated control for the initial 4 years of the study, Sandbar Lake also received a whole-lake, low-dose (275 μg/L ae) 2,4-D treatment. Both lakes were intensively monitored over a 6 yr period to evaluate both the short- and long-term responses of EWM and the native plant community.

Table 1 Pretreatment summary statistics for Tomahawk and Sandbar lakes, Bayfield County, WI. Aquatic plant surveys were conducted 9–13 July 2007.

Lake and plant community variables	Tomahawk	Sandbar
Lake area (ha)	53	51
Maximum depth (m)	12.8	14.9
% Littoral	84	52
pH^1	8.7	7.6
Conductivity (μS/cm)^1	65	59
Secchi transparency depth (m) ^2	4.5	5.3
Maximum depth of plant colonization (m)	6.8	7.0
% Vegetated^3	86	87
Simpson's Diversity Index^3	0.90	0.89
Number of native species	21	17
Mean number of natives per site^3	2.2	2.5
% EWM occurrence^3	40	26
^1Single measurement collected July 2011.
^2Mean calculated for biweekly readings from May through September 2000–2007.
^3Calculated for points within the littoral zone (shallower than maximum depth of plant colonization).

Herbicide application and concentration monitoring

Tomahawk lake treatment

To determine the quantity of 2,4-D herbicide required to achieve a lakewide target concentration of 500 μg/L ae, we recorded water depths in Tomahawk Lake at each point of a 50 × 50 m grid and calculated the lakewide volume. We assumed the lake was thermally mixed and herbicide would dissipate throughout; therefore, the entire lake volume was used in calculations for dosing. On 20 May 2008, when water temperatures were <15 C, 2,4-D dimethylamine salt was applied lakewide as the liquid formulation DMA 4 IVM using a powered injection system with weighted drop hoses to inject the herbicide ∼1 to 2 m below the water surface. Three years later, on 27 September 2011, a 1.9 ha isolated area of new EWM growth was treated with diquat (6,7-dihydrodipyrido [1,2-a:2′,1′-c]pyrazinediium ion) applied as the liquid formulation Reward at 0.37 mg/L cation.

Partnering with lake residents, we collected mid-depth water samples at 5 locations on Tomahawk Lake (T1, T2, T3, T4, T5; Fig. 1) to quantify 2,4-D degradation rates and verify herbicide mixing within the lake. To monitor vertical mixing at site T4 (maximum depth of 11 m), we collected samples at one-fourth, one-half, and three-fourths of the maximum depth (2.7, 5.5, and 8.2 m, respectively). We also collected water samples at 4 locations on the untreated reference lake (Sandbar) to monitor potential herbicide migration from Tomahawk Lake. We collected water samples at each location prior to treatment and at 1, 2, 3, 4, 7, 14, 21, 28, 42, 49, 56, 91, 145, and 166 days after treatment (DAT). We preserved samples with muriatic acid to prevent microbial degradation of 2,4-D after collection and stored samples at 5 C until concentration analysis at the University of Florida Center for Aquatic and Invasive Plants using an enzyme-linked immunosorbant assay (ELISA; Netherland et al. 2002, Graziano et al. 2006, Byer et al. 2008). Herbicide concentrations are reported here as 2,4-D ae.

Figure 1 Location of Tomahawk and Sandbar lakes, Bayfield County, WI. Aquatic plant sampling grid points are shown in gray with point spacing of 35 and 40 m, respectively. 2,4-D herbicide concentration sampling locations on both lakes are labeled.

Sandbar lake treatment

After serving as an untreated reference lake for the initial 4 years of the study, Sandbar Lake also underwent a whole-lake, low-dose 2,4-D treatment to control EWM. To determine the quantity of 2,4-D required to achieve a lakewide epilimnetic target concentration of 275 μg/L ae, we assessed thermal stratification at the deepest portion of Sandbar Lake. At the time of treatment, the thermocline occurred at 7.6 m, and only waters above this thermocline were included in volumetric calculations for dosing. On 24 May 2011, 2,4-D was applied to Sandbar Lake as the liquid formulation DMA 4 IVM using a powered injection system with weighted drop hoses to inject the herbicide ∼1 to 2 m below the water surface.

Partnering with lake residents, we collected mid-depth water samples at 4 locations on Sandbar Lake (S1, S2, S3, S4; Fig. 1) to quantify 2,4-D degradation rates and verify herbicide mixing within the lake. To monitor vertical mixing at site S2 (maximum depth of 13.5 m) we collected samples at approximately one-eighth, one-fourth, one-half, and three-fourths of the maximum depth (1.5, 3.0, 6.1, and 9.1 m, respectively). During this time, we also collected additional water samples at 2 locations on Tomahawk Lake to monitor potential herbicide migration from Sandbar Lake. Water samples were collected at each location prior to treatment and at 1, 3, 7, 10, 14, 22, 32, 60, and 71 DAT and were similarly processed and analyzed as described for Tomahawk Lake.

Statistical analysis was performed in the R environment for statistical computing (version 2.14.1, R Development Core Team 2011). We used analysis of covariance to examine patterns in the rate of 2,4-D degradation over time across sampling sites. To account for pseudoreplication introduced by resampling fixed monitoring locations over time, we tested for the influence of sampling site on the parameters estimated by the model. The simplest model with sufficient explanatory power included DAT as a fixed effect on 2,4-D concentration, but monitoring site was not an important interacting factor and did not require parameterization. Thus, we averaged the herbicide concentrations across sample sites for each sampling date. Linear models were fit to the concentration data over time for each application, and the response variable was log-transformed where necessary to satisfy the assumptions of linearity. The parameters of the model were used to estimate the initial concentration, the overall rate of degradation, and estimated half-life of 2,4-D.

Macrophyte community monitoring

We conducted lakewide pretreatment and posttreatment aquatic plant surveys following a grid-based, point-intercept approach (Madsen 1999, Hauxwell et al. 2010), which has been shown to be appropriate for comparative studies (Mikulyuk et al. 2010). We recorded species presence/absence and water depth at each site on a georeferenced sampling grid (Fig. 1). Grid spacing resolutions for Tomahawk Lake and Sandbar Lake were 35 m (427 sampling points) and 40 m (324 sampling points), respectively. At sites shallower than 5 m, a double-headed rake (0.34 m wide, 14 tines per head) was lowered on an adjustable pole vertically through the water column to the sediment surface, rotated twice, and then pulled straight out of the water. At sites deeper than 5 m, we used a similar rake head attached to a rope to collect plants. Plants retrieved on the rake as well as plant fragments detached from the bottom were identified to species level following Crow and Hellquist (2000a, 2000b). Surveys were conducted annually in mid-July from 2006 to 2012 on Tomahawk Lake and from 2007 to 2012 on Sandbar Lake. Species with at least a 10% littoral frequency of occurrence at any sampling event were tested for significant differences in percent frequency of occurrence between the year immediately before treatment and all posttreatment years using Pearson's chi-square test. We also conducted supplemental surveys on Tomahawk Lake in fall 2008, 2009, and 2010, recording only whether EWM was present at each survey point.

In 2007, we collected plant biomass samples at 10–15% of the littoral point-intercept sites (randomly selected) on both lakes using the vertical rake method described by Johnson and Newman 2011. Biomass samples were processed in the laboratory to remove below-ground biomass and rinsed to remove any remaining sediment or debris. All remaining plant material was separated into native and nonnative species and oven-dried at 65 C for 48 h. From 2008 to 2012, we roughly sorted and visually estimated the percent contribution of native versus nonnative biomass of total wet weight in the field. Of these samples, 10% (n = 5/yr; ∼25 total) were given estimated percent contributions and then individually separated into native and nonnative species, dried, and weighed to assess the comparability of our in-field percent weight estimations to actual dry weights. Comparison of in-field percent estimates of wet weights to actual lab-weighed dried and sorted samples were similar for both Tomahawk and Sandbar lakes (R² = 1 and R² = 0.99, respectively); therefore, all remaining samples were dried and weighed without sorting, and native versus nonnative biomass was assigned by applying field-estimated percentages to the total biomass dry weight.

Results and discussion

Initial concentrations of 2,4-D in Tomahawk Lake were approximately 100 μg/L higher than the nominal target of 500 μg/L, and the herbicide mixed vertically more slowly than expected (Fig. 2). Simple linear regression models relating log-transformed herbicide concentration to DAT were statistically significant for all stations on the treatment lake except for the deepest sample at T4 (Table 2). The adjusted R² for significant models ranged from 0.89 to 0.99. The estimated initial 2,4-D concentration for sites T1, T2, T3, T4, and T5 ranged from 593 to 636 μg/L. For the deeper samples collected at T4, the initial concentration was 482 μg/L for mid-depth and 14 μg/L near the bottom. Herbicide concentration samples analyzed at the Sandbar Lake reference sites were below detection limits throughout the entire treatment. Low 2,4-D concentrations in the near-bottom samples suggest that thermal stratification reduced vertical mixing of the herbicide and may have impeded movement of 2,4-D into deeper waters. This phenomenon has been observed during other chemical applications and indicates that application rate calculation may be more accurate if based on the volume of water above the thermocline rather than the volume of the whole lake (Netherland et al. 2002). The lack of complete mixing into the deeper water of Tomahawk Lake likely contributed to the actual concentrations exceeding the nominal concentration of 500 μg/L. Although initial observed concentrations slightly exceeded the 500 μg/L target, concentrations were one-sixth of the maximum label use rate of 4000 μg/L.

Table 2 Linear model parameters describing regression of 2,4-D concentration (y) at individual sampling stations and lakewide means of all surface samples (“mid” and “bottom” samples excluded) vs. days after treatment (x). Log-transformed parameters were back-transformed; the slope parameter describes a multiplicative change in herbicide concentration.

Lake	Site	Intercept	Slope	Adjusted R^2	F	P-value
Tomahawk^1	T1	636	−0.98	0.98	544	<0.001
	T2	637	−0.98	0.97	327	<0.001
	T3	614	−0.98	0.99	906	<0.001
	T4	596	−0.98	0.91	113	<0.001
	T4 (mid)	482	−0.98	0.89	93	<0.001
	T4 (bottom)	14	−1.00	−0.10	0	0.853
	T5	593	−0.98	0.95	209	<0.001
	Lakewide	623	−0.98	0.99	1658	<0.001
Sandbar^2	S1	398	−5.74	0.91	91	<0.001
	S2	394	−5.99	0.93	127	<0.001
	S2 (midT)	364	−4.83	0.72	24	0.001
	S2 (midB)	178	−1.84	0.25	1	0.299
	S2 (bottom)	27	−0.22	0.01	1	0.321
	S3	388	−5.89	0.91	96	<0.001
	S4	359	−5.37	0.95	171	<0.001
	Lakewide	386	−5.75	0.94	144	<0.001
^1 Samples collected May–November 2008. Concentrations log-transformed for regression.
^2 Samples collected May–July 2011. No transformation required for regression.

Figure 2 2,4-D concentration data from each monitoring site showing change over time in Tomahawk and Sandbar lakes. The respective lakewide target concentrations of 500 and 275 μg/L are indicated by the horizontal dotted lines (color figure available online).

We generated a linear model to relate lakewide mean 2,4-D concentrations in Tomahawk Lake across all monitoring sites (bottom sample at T4 excluded) to DAT (Fig. 3). The adjusted R² for this model was 0.99 and the initial concentration was estimated to be 623 μg/L. The 95% confidence interval around this estimate (576–675 μg/L) did not include the target concentration of 500 μg/L. The half-life of 2,4-D calculated using the lakewide regression equation was 41 d. The herbicide specimen label for 2,4-D applied as DMA 4 IVM indicates that treated water may not be used for irrigation purposes until concentrations are <100 μg/L based on an approved assay. The lakewide regression equation for Tomahawk Lake indicated that the 2,4-D concentration did not drop below the irrigation safety threshold of 100 μg/L until 93 DAT. Such an extended exposure period to 2,4-D has not been previously reported. Although the impacts on efficacy against EWM could be extrapolated from prior CET studies, information on the expected response of native plants to a prolonged exposure to 2,4-D was not readily available. The slow degradation of 2,4-D in Tomahawk Lake suggests that microbial breakdown was much lower in this oligo-mesotrophic seepage system.

Figure 3 Linear models describing relationship between mean [2,4-D] across sites to days after treatment in Tomahawk Lake and Sandbar Lake; Tomahawk Lake data has been back-transformed. Dashed lines indicate 95% confidence intervals, and the irrigation threshold of 100 μg/L is also indicated.

Pretreatment surveys conducted in 2007 revealed that aquatic plant community composition was similar in both lakes (Table 1; Fig. 4). After application of 2,4-D to Tomahawk Lake in 2008, the littoral frequency of EWM significantly decreased (p < 0.001) from 40% pretreatment to 0% and was undetected in subsequent surveys conducted in summer and fall 2009 and 2010. In spring 2011, a small (∼2 ha) population of EWM was discovered in Tomahawk Lake near a public access point. This isolated population was subsequently treated with diquat in fall 2011 to limit expansion to other areas of the lake. During this same timeframe (2007–2010), EWM in the untreated lake (Sandbar) significantly increased from 26 to 42% (p = 0.003). We observed no significant changes in littoral frequency between 2007 and 2008 in the aquatic plant community of the untreated reference Sandbar Lake; however, 8 species (including EWM) in the chemically treated Tomahawk Lake declined significantly pretreatment and 1 year posttreatment (Fig. 4).

Figure 4 Littoral frequency (% occurrence) for the most common aquatic plant species (>10% occurrence) in Tomahawk Lake (squares) and Sandbar Lake (circles). 2,4-D was applied to Tomahawk Lake in May 2008 and to Sandbar Lake in May 2011 (dashed vertical lines). Differences in littoral frequency relative to pretreatment condition in each lake were evaluated using Pearson's chi-squared test. Significant differences are indicated by solid-filled shapes; nonsignificant differences are indicated by open shapes. *Variable-leaved pondweed (Potamogeton gramineus) and Illinois pondweed (P. illinoensis) were combined for analysis.

Figure 5 Mean aquatic plant biomass (grams dry weight) from randomly selected sites on both Tomahawk and Sandbar Lakes. Year of herbicide treatments is indicated by the dotted vertical lines. Error bars indicate 95% confidence intervals.

Five native monocot species (Potamogeton robbinsii, Elodea canadensis, P. pusillus, P. amplifolius, and P. gramineus) demonstrated sustained reductions in littoral frequency after treatment with 2,4-D in Tomahawk Lake (p < 0.001; Fig. 4). Two native species (Vallisneria americana and Najas flexilis) exhibited initial significant declines after treatment and then showed signs of recovery in subsequent years, eventually exceeding pretreatment frequencies by the end of the study. The macroalgae Chara spp. exhibited a significant increase after treatment, which has also been observed in previously conducted studies (Miller and Trout 1985). Overall species richness was reduced by half after treatment, with 21 native species recorded in the pretreatment survey year and only 11 native species recorded the first year posttreatment.

Due to the high degree of nontarget native impacts observed after the treatment of Tomahawk Lake, the nominal target rate for Sandbar was set at approximately half that used in Tomahawk Lake. Initial concentrations of 2,4-D in Sandbar Lake were approximately 100 μg/L higher than the nominal epilimnetic target of 275 μg/L (Fig. 2). Linear models relating herbicide concentration to DAT were statistically significant for all stations on the treatment lake except for the deepest 2 samples at S2 (midB and bottom; Table 2). The adjusted R² for significant models ranged from 0.72 to 0.95. The intercept representing initial 2,4-D concentration for sites S1, S2, S3, and S4 ranged from 359 to 398 μg/L. The initial concentration from deeper samples at S2 was 364 μg/L (midT), 176 μg/L (midB), and 27 μg/L (bottom). Herbicide concentrations observed at the Tomahawk Lake reference sites were below detection limits throughout the entire treatment of Sandbar Lake. Low 2,4-D concentrations in near-bottom samples at S2 indicated that the herbicide did not mix into deeper waters due to thermal stratification. Although initial observed concentrations slightly exceeded the 275 μg/L target, concentrations were one-tenth of the maximum label use rate of 4000 μg/L.

We generated a lakewide linear model to relate mean 2,4-D concentration across all monitoring sites (excluding deeper samples at S2) to DAT (Fig. 3). The adjusted R² for this model was 0.94 and the initial concentration was 386 μg/L. The 95% confidence interval around the y-intercept (351–422 μg/L) did not include the target concentration of 275 μg/L. The half-life of 2,4-D calculated using the lakewide regression equation for Sandbar Lake was 34 d. Furthermore, the regression indicated that the 2,4-D concentration did not drop below the irrigation safety threshold of 100 μg/L until 50 DAT.

Littoral frequency of EWM in Sandbar Lake decreased from 42% in 2010 to 4% after treatment in 2011 (p < 0.001) and remained lower than pretreatment levels in 2012 (8%; p < 0.001). Two native species (P. pusillus and Najas flexilis) demonstrated sustained reduction in littoral frequency after treatment with 2,4-D in 2011. Two native species (Vallisneria americana and P. gramineus) exhibited initial declines after treatment and then showed signs of recovery, returning to pretreatment levels by 2012. Overall species richness remained relatively stable throughout the study, with 13–17 native species recorded during the pretreatment survey years and 16–17 native species recorded during the posttreatment surveys.

We observed a significant effect of treatment on EWM and native plant biomass over time (Fig. 5). In Tomahawk Lake, mean EWM biomass decreased by 100% following 2,4-D treatment and remained low (98% reduction) throughout the remainder of the study period. By contrast, EWM biomass in the reference Sandbar Lake showed a slight fluctuation from year to year but remained significantly higher than in Tomahawk Lake. Mean native plant biomass in Tomahawk Lake decreased by 87% following 2,4-D treatment and was still significantly reduced (62% reduction) by the conclusion of the study. Treatment of Sandbar Lake resulted in mean EWM biomass decreasing by 100% and remaining low (98% reduction) through the second year posttreatment. Mean native biomass on Sandbar Lake increased by 63% following 2,4-D treatment and was relatively constant through the second year posttreatment. In contrast to the high degree of native plant damage observed in Tomahawk Lake, the lower 2,4-D dosage used in Sandbar Lake was able to minimize damage to both native plant littoral frequency and biomass while still providing good control of EWM.

It is rare in aquatic plant management to have 2 lakes so similar in location, morphometry, size, and vegetative composition for comparison. The treatments of Tomahawk Lake and Sandbar Lake allowed a multiyear comparison of the efficacy of whole-lake 2,4-D treatments on EWM, the potential for sustained reductions of EWM, and impacts to native submersed aquatic vegetation. The unexpected persistence of 2,4-D in Tomahawk and Sandbar lakes (41 and 34 d half-lives, respectively) resulted in a much longer exposure time than originally predicted. Prior concentration and exposure time research suggests that exposures of approximately 3 d at 500 μg/L would result in EWM control (Green and Westerdahl 1990). Subsequent research by Glomski and Netherland (2010) demonstrated that long-term exposures (>14 d) to concentrations as low as 100 μg/L can result in EWM control. Information on the response of native submersed species to extended exposure at low concentrations of 2,4-D has not been published. In our study, plants in Tomahawk and Sandbar lakes were exposed to concentrations >250 μg/L for more than 46 and 24 d, respectively, eliminating EWM vegetative growth for 3 years in Tomahawk Lake and significantly reducing EWM in Sandbar Lake. Although these impacts on EWM were viewed as positive, the unexpected reduction in both frequency and biomass of some native monocots suggests that 2,4-D persistence in large-scale treatments may reduce selectivity. While we did not observe native species rapidly recolonizing areas vacated by EWM in Tomahawk Lake, we did observe some native recolonization in Sandbar Lake, which was treated at the lower 2,4-D rate.

The benefits and drawbacks associated with the responses of EWM and native vegetation to whole-lake, low-dose 2,4-D treatments demonstrate the challenges facing aquatic plant managers. Ongoing studies in Wisconsin following 2,4-D applications occurring on lakes across the state have yielded a wide range of outcomes, with exposures to 2,4-D ranging from hours to several weeks depending on the scale of treatment. Preliminary data from case studies of large-scale 2,4-D treatments have provided some insight on EWM control and native species impacts across multiple lakes and management scenarios (Nault et al. 2012). Use patterns of auxin mimic herbicides (2,4-D and triclopyr) that result in longer-term exposures to low concentrations across an entire lake (or embayment) have some resemblance to management strategies associated with the use of the herbicide fluridone (Getsinger et al. 2002). While photodegradation of fluridone is relatively predictable, the variables that affect the rate of 2,4-D microbial degradation in large-scale applications are currently not well known. Future studies of 2,4-D decay rates across lakes of varying trophic status and physiology are needed. As managers learn more about use patterns of herbicides that allow several weeks of exposure, initial treatment concentrations and tolerance for injury to selected native species must be carefully considered.

Acknowledgments

The authors thank Frank Koshere, Martha Barton, Meghan Porzky, Kelly Wagner, Scott Van Egeren, Alex Smith, Paul Frater, Kari Soltau, Jesse Schwingle, Michael Fell, Jeremy Bates, and Stefania Strzalkowska for assistance in field and laboratory. We appreciate the partnership of the Town of Barnes Aquatic Invasive Species Committee, specifically Ingemar Ekstrom who coordinated project management, as well as Ron Clark, Gus Gustafson, Jim Joswick, and Glenda Mattila who collected herbicide concentration samples. We also thank the editor and reviewers for valuable input and comments that improved the manuscript. This study was completed thanks to financial support from aquatic invasive species grants and technical assistance funds from the Wisconsin Department of Natural Resources as provided under s. 23.22 (2)(c).