Northeastern Naturalist NENHC-26 C. Picone 22001155 NORTHEASTERN NATURALIST 22(1):NENHC-26V–oNl.E 2N2H, NCo-3. 91 Effects of Aquatic Herbicides and Housing Density on Abundance of Pond-Breeding Frogs Chris Picone* Abstract - Aquatic herbicides are applied to control nuisance vegetation in lakes and ponds, and are often re-applied over many years. This study compared the abundance of 5 frog species in treated and untreated lakes in Ashburnham, MA. At each lake, I assessed the density of human housing and lake area to determine their relationship with frog abundance. I employed a standard calling index to estimate the abundance of each frog species and the sum of calling indices from all species served as a measure of total frog abundance at each survey site. The sum of calling indices declined with increasing density of human housing; herbicide treatment was not an important main effect. However, there was marginal interaction between herbicides and housing density: lakes with moderate–high housing densities seemed to have lower frog abundance with herbicide treatments. Although the data set was limited, my results suggest that frogs may benefit from permanently untreated refuge zones on lakes. When I assessed frog species individually, only Rana clamitans (Green Frog) was less common as housing density increased. None of the 5 frog species were less abundant in herbicide-treated lakes. In my study, long-term use of aquatic herbicides did not generally reduce abundance of adult frogs, but more studies may be needed to determine whether some herbicides may impair frog populations that are already stressed by housing development. Introduction Global declines in amphibian populations are among the most important conservation issues identified in the last several decades (Stuart et al. 2004). The loss of frogs has resulted from habitat destruction, emerging diseases, climate change, agro-chemicals, and synergistic interactions among these factors (Blaustein et al. 2011). The most commonly used agricultural herbicides are especially implicated. Roundup® (active ingredient glyphosate; Monsanto, Cambridge, MA) drastically reduced larval survival of many amphibian species in mesocosm studies (Relyea 2005a, 2005b; Relyea and Jones 2008) and atrazine reduced the immune response and altered gonadal development, among other developmental effects in a metaanalysis study (Rohr and McCoy 2010). The effects of aquatic herbicides have received much less attention than those of agricultural herbicides. Aquatic herbicides are applied directly into ponds and lakes to control nuisance vegetation. For example, diquat (in Reward®, Syngenta Crop Protection, Inc., Geensboro, NC) is commonly used to control submerged weeds like Myriophyllum spp. (Milfoil) and Utricularia spp. (Bladderwort), while glyphosate (in Rodeo®; Dow Agrosciences, Indianapolis, IN) is used for floating and emergent pondweeds (Mattson et al. 2004). These herbicides may be applied in *Department of Biology, Fitchburg State University, Fitchburg, MA 01420; cpicone@ fitchburgstate.edu. Manuscript Editor: Catherine Bevier Northeastern Naturalist Vol. 22, No. 1 C. Picone 2015 NENHC-27 conjunction with copper sulfate or chelated copper complexes to control filamentous algae. Aquatic herbicides are thought to have little or no impact on non-target organisms when used at recommended doses (Mattson et al. 2004, Wagner 2004), but very few studies have examined effects on amphibians in the field. Why are field studies needed when federal and state regulators have already determined what are considered safe and legal limits? First, permissible concentrations of pesticides are typically determined from a few lab studies, often confounded by conflicts of interest with the manufacturers (Boone et al. 2014). In the case of agricultural pesticides, some caused no significant effects in simple laboratory conditions while the same chemicals were found to reduce frog survival in more natural experimental conditions (Blaustein et al. 2011, Boone et al. 2014). For example, some agricultural pesticides have synergistic effects with chemical cues issued by predators such as dragonfly nymphs or newts that tadpoles find stressful (Relyea 2003, 2005c). Other pesticides change trophic webs in ways that reduce food sources for tadpoles in mesocosms (Relyea and Diecks 2008) or increase pathogens that attack frogs in natural habitats (Rohr et al. 2008). Secondly, effects of pesticides on non-target species are usually assessed with each chemical applied alone (e.g., Cooke 1977) while in practice multiple herbicides may be combined to control different types of weeds. When agricultural pesticides are tested in mixtures that mimic those found in the field, they can reduce tadpole development and impair their immune systems (Hayes et al. 2006). Finally, non-target effects of pesticides are often assessed in short-term studies, yet in practice, aquatic herbicides may be applied repeatedly over decades (C. Picone, pers. observ.). Therefore, this study sought to test whether long-term use of aquatic herbicides reduces populations of adult breeding frogs when assessed under field conditions. This question is especially relevant for agencies that grant permits for chemical control of aquatic vegetation, such as environmental agencies at state and local levels. In particular, each town in Massachusetts has a conservation commission staffed by volunteers who have the authority to permit, modify, or reject applications to use aquatic herbicides (Dawson and Zielinski 2006). Some commissions have sought evidence from practical field studies to verify that aquatic herbicides are safe for non-target species, especially given concerns about frog declines and suspected links to agro-chemicals (C. Picone, pers. observ.). This study was intended to provide such evidence by comparing the abundance of breeding frogs in treated and untreated lakes in north-central Massachusetts. Methods I conducted surveys of pond-breeding frogs on 12 lakes in Ashburnham, MA (Fig. 1, Table 1). Seven of the lakes served as untreated controls, and 5 lakes had been treated annually or biannually with herbicides for ~20 years. Treatments were carried out by Aquatic Control Technologies, Inc. of Sutton, MA and by Lycott Environmental Inc. of Southbridge, MA. Both applicators are licensed by the MA Department of Agricultural Resources (Pesticide Bureau), and they followed Northeastern Naturalist NENHC-28 C. Picone 2015 Vol. 22, No. 1 Figure 1. Ashburnham is located on the northern edge of Worcester County, MA (see inset). Study areas were the 12 lakes and ponds marked in this figure. Surveys were conducted at potential frog breeding sites located along different shorelines or coves marked with the symbols labeled in the legend. Survey sites on 3 treated lakes were not directly sprayed in most years (triangles with dark centers), but most of each lake was treated and so those samples were considered as treated replicates. Northeastern Naturalist Vol. 22, No. 1 C. Picone 2015 NENHC-29 manufacturers’ label instructions and state guidelines (e.g., Mattson et al. 2004, Wagner 2004). All treated lakes received diquat dibromide (as Reward®) to treat submerged weeds in June or July. According to the permits filed by the herbicide applicators, Reward® was applied at a rate of 1–1.5 gallons/acre (7.57 L–9.46 L/ ha), which is below the maximum dose of 2 gallons/acre (18.9 L/ha) recommended by the manufacturer. Most lakes were also treated with glyphosate (as Rodeo®) in August or September to reduce floating-leaf pondweeds. Glyphosate was applied at recommended doses of 3 quarts/acre (7.1 L/ha). One lake (Sunset Lake) was treated with both diquat and an algaecide that contains chelated copper complexes (Captain ®). Copper doses should rarely have exceeded 0.3 mg/L if used as recommended (Wagner 2004). Survey sites were shorelines and coves that provided likely breeding habitat for frogs. These were areas with shallow, gently sloped shores with some vegetation. The average survey site consisted of approximately 430 m of shoreline, with a few ranging as low as 150–250 m long and some large sites spanning 900 m of shoreline (Table 1). I did not include sites with extensive human development or sandy beaches unless I heard frogs calling there. Ten of the 12 lakes contained multiple survey sites (Fig. 1, Table 1), so I averaged data for all survey sites/lake to get a composite value for each lake; thus, each lake or pond served as an independent statistical sample (n =12). Table 1. Summary of the lakes used in this study. The survey sites on 3 treated lakes were in untreated refuge zones but most of the lake shoreline was treated. The active ingredient in Reward® is diquat, Rodeo® contains glyphosate, and Captain® is a copper-based algaecide. Treated and untreated lakes varied in lake area, density of housing, and amount of shoreline with viable breeding habitat (= total shoreline surveyed). At each lake, 1–5 sites were surveyed for frog calls. If more than 1 site was surveyed, results from multiple sites were averaged for that lake so each lake = 1 replicate. Estimates of housing density only apply to survey sites that presented habitat for frog breeding; other parts of a lake may have had higher housing densities. Note that Stodge Meadow Pond had only 1 small cove with likely breeding habitat; the rest of the shoreline was too steep, sandy, and/or developed to offer sites for frogs. Housing density given in # houses w/in 30 m per 30 m shoreline. * indicates lakes where most survey sites were in untreated refuge zones. Total length Herbicide Housing Lake area # of of shoreline Lake or pond treatment density (km2) survey sites surveyed (m) Marble Pond none 0.02 0.07 2 980 Lincoln Pond none 0.00 0.11 1 400 Wallace Pond none 0.00 0.12 3 1220 Lake Winnekeag none 0.40 0.42 3 1300 Stodge Meadow Pond none 0.40 0.47 1 150 Lake Wampanoag none 0.00 0.91 5 1620 Upper Naukeag Lake none 0.07 1.22 5 2600 Little Watatic Pond* Reward® and Rodeo® 0.07 0.07 2 1070 Billy Ward Pond* Reward® and Rodeo® 0.23 0.19 3 1400 Watatic Lake Reward® and Rodeo® 0.24 0.52 3 670 Lower Naukeag Lake* Reward® and Rodeo® 0.05 1.09 2 2350 Sunset Lake Reward® and Captain® 0.38 1.02 3 730 Northeastern Naturalist NENHC-30 C. Picone 2015 Vol. 22, No. 1 The survey sites on 3 treated lakes overlapped with shorelines that were not directly sprayed with herbicides and I considered them to be permanent refuge zones (Fig. 1). Untreated zones are recommended to provide habitat for non-target organisms, especially fish (Mattson et al. 2004, Wagner 2004). Although the survey sites were mostly untreated refuge zones, I considered those 3 lake replicates as treated because the whole lake was the statistical sample. The refuge zones were only 100 m–300 m long and herbicides like diquat can drift over 160 m (Serdar 1997); therefore, most of these untreated shorelines likely had some indirect herbicide exposure. Even if some refuge zones were unaffected by herbicide exposure, the question of this study was whether treated lakes differed from untreated lakes, especially for permitting purposes. I targeted 5 frog species that can breed along the edges of lakes and ponds. Study taxa included Anaxyrus americanus Holbrook (American Toad) and Lithobates palustris LeConte (Pickerel Frog), which start breeding in the spring, and Hyla versicolor LeConte (Gray Treefrog), Lithobates clamitans melanota Rafinesque (Green Frog), and Lithobates catesbieanus Shaw (American Bullfrog), which breed from mid- to late spring into summer. I conducted surveys of breeding frogs by listening from a kayak just after sunset (20:00–23:00). I estimated population sizes of the 5 frog species at each survey site with a standard calling index established by the North American Amphibian Monitoring Program (NAAMP 2012). Population sizes were categorized based on the male calls: 0 = absent; 1 = a few frogs with non-overlapping calls; 2 = males with overlapping calls, but individuals could be distinguished; and 3 = a full chorus in which individuals could not be easily distinguished. I completed the surveys during April–June in 2009, 2010, and 2012. Over the course of the study, I assessed each lake 4–5 times during different parts of the breeding season to reduce the likelihood of failing to detect a species. Each survey from a kayak lasted 30–90 minutes, a period that provided ample time to hear a species if it was calling. I counted a species as present and breeding if I heard it at any time during the survey. I also checked several survey sites on both treated and untreated lakes from nearby roads on additional dates to verify that I was detecting all frog species during the kayak surveys. Very often, I heard the same species calling in the same coves from year to year, but in some years these haphazard surveys still could have missed some breeding species. Therefore, for each survey site I used only the single highest calling index heard for each species over the 3 sample years. I added together the estimates for each species at each survey site to generate the sum of calling indices, a measure of total frog abundance. Treated and untreated lakes varied in surface area and density of human housing, so I quantified those factors to assess potential effects on frog abundance (Table 1). I estimated both factors from the Massachusettes state GIS database OLIVER (2014). To measure housing density, I counted the number of houses within 30 m of the shoreline along each survey site. Similar to my procedure for frog-calling indices, I averaged housing densities from different survey sites for each lake. Across Northeastern Naturalist Vol. 22, No. 1 C. Picone 2015 NENHC-31 the 12 lakes, housing density was not correlated with lake area (Pearson correlation P = 0.7), nor was housing density related to herbicide treatment (t-test P = 0.5). I used Minitab 17 (Minitab.com) to develop a general linear model (GLM) to assess whether frog calling was affected by herbicide treatment, lake area, housing density, or the interaction of housing density*herbicide treatment. This GLM included the terms that were of most interest for this study a priori, and this model had the strongest support from Akaike information criterion (AIC) scores when assessed with R (version 3.0.1; R Core Team, 2013). The same factors included in this GLM were used to assess the sum of calling indices as well as the calling index from each frog species assessed separately. Results Total breeding frog abundance was controlled mainly by density of human housing, but not herbicide treatment (Fig. 2, Table 2). The sum of calling indices generally declined with increased housing density (F1,7 = 6.5; P = 0.04). Herbicide Figure 2. The sum of calling indices for each lake plotted against density of human housing and herbicide use. This measure of total frog abundance was calculated at each survey site by adding the highest calling indices from each of the 5 frog species. If there was more than 1 survey site at a lake (Table 1), the sum of calling indices for that lake was averaged from its survey sites. Total frog abundance generally declined as human housing increased (P = 0.04), despite one outlier (Stodge Meadow Pond). Although the herbicide treatment had no consistent effect as a main factor (P > 0.7), there was a marginal interaction between herbicide treatment and housing density (P = 0.06). Northeastern Naturalist NENHC-32 C. Picone 2015 Vol. 22, No. 1 use was not a significant main factor (F1,7 = 0.09; P > 0.7), but there was a marginal interaction between herbicide use and housing density (F1,7 = 5.15; P = 0.06). That is, although herbicides had no effect on frog abundance near lower housing densities, breeding frogs near higher housing densities were more abundant in untreated lakes compared to treated lakes (Fig 2). However, this statistical interaction was mostly driven by the data from one outlier, Stodge Meadow Pond. That pond had an exceptionally high residual and leverage value in the analysis, but contained only a single survey site that was much smaller than the shorelines sampled in the other lakes (Table 1). When I excluded Stodge Meadow Pond from the analysis, the negative correlation between frog calling and housing density was much stronger (F1,6 = 24.9; P = 0.002), while the interaction between housing density and herbicides appeared weaker (F1,6 =4.01; P = 0.09). In the original GLM summarized in Table 2, lake area had no effect on the sum of calling indices (Fig. 3). However, when the outlier Stodge Meadow Pond was excluded, the sum of calling indices appeared to decline as lake area increased (F1,6 = 6.30; P = 0.046). The sum of calling indices was a crude measure of frog abundance at each sample site because that measure could reach high values (e.g., ≥3) due to either a small number of several frog species or a large number of a single species. Therefore, to estimate community diversity I also calculated a version of the Shannon diversity index based on the species-calling indices. Here, I calculated the Shannon index as follows (Magurran 2003): Shannon index = –Σ[Pi*ln(Pi)], where Pi = the proportion that each species contributed to the sum of calling indices at each site. The results from the Shannon index were qualitatively identical to the results from the sum of calling indices (see Supplemental File 1, available online at https://www.eaglehill.us/NENAonline/suppl-files/n22-1-N1322a-Picone-s1, and, for BioOne subscribers, at http://dx.doi.org/10.1656/N1322a.s1). None of the calling indices of individual frog species were significantly affected by herbicide treatment (P > 0.5 for all species; Fig. 4; for full results from GLM, Table 2. Results from the general linear model (GLM) used to assess which factors correlated with the sum of calling indices presented in Figure 2. Qualitatively identical results were obtained when lake area was excluded from this model. The same GLM was used to analyze the Shannon diversity index and the abundance of each frog species; complete results are presented in Supplemental File 1, available online at https://www.eaglehill.us/NENAonline/suppl-files/n22-1-N1322a-Picone-s1, and, for BioOne subscribers, at http://dx.doi.org/10.1656/N1322a.s1. Source DF F-value P-value Herbicide treatment 1 0.09 0.77 Housing density 1 6.50 0.04 Lake area 1 2.75 0.14 Herbicide*housing 1 5.15 0.06 Error 7 Total 11 Northeastern Naturalist Vol. 22, No. 1 C. Picone 2015 NENHC-33 Figure 3. The sum of calling indices for each lake plotted against lake-surface area. According to the general linear model used here (see Methods), lake area had a negligible impact on abundance of breeding frogs (P = 0.14) when accounting for herbicide use and housing density. see Supplemental File 1, available online at https://www.eaglehill.us/NENAonline/ suppl-files/n22-1-N1322a-Picone-s1, and, for BioOne subscribers, at http://dx.doi. org/10.1656/N1322a.s1). It is worth noting that I heard Grey Treefrogs calling from the shores of 4 of the 6 untreated lakes, but I heard none at treated lakes; however, the herbicide effect on Grey Treefrogs was not significant according to the GLM used here (F1,7 = 0.47; P = 0.5). Furthermore, no interaction was detected between herbicides and housing density for any species (P > 0.2). Housing density was correlated with a decline in calling index for Green Frogs (F1,7 = 5.75; P = 0.05), but housing density was not a significant factor for any other species (P ≥ 0.3). Even when I omitted the outlier site Stodge Meadow Pond, housing density had no significant effect on any individual species, though there was a marginal effect for Green Frogs (F1,6 = 4.04; P = 0.09). Discussion The total abundance of breeding frogs was negatively correlated with density of human housing rather than herbicide treatments (Fig. 2, Table 2). There are many reasons why frog density would be negatively correlated with human development. While people tend to focus housing development on drier lots rather than the mucky shorelines and wetlands that provide optimal frog breeding sites, suburbanization Northeastern Naturalist NENHC-34 C. Picone 2015 Vol. 22, No. 1 of natural habitats typically reduces amphibian species richness and abundance (Hamer and McDonnell 2008). Habitat loss, fragmentation, and road density deter amphibian migration and reduce their abundance at broad geographic scales (Cosentino et al. 2014, Cushman 2006, Eigenbrod et al. 2008). However, these geographic factors are probably less important mechanisms for the effect of housing Figure 4. Relationship between housing density, herbicide treatment, and the calling index for each frog species. The same GLM shown in Table 2 was used for statistical analyses. Herbicide use was not associated with a decline in the calling of any particular frog species, nor were there any interactions between herbicide and housing density. Only Green Frogs declined with an increase in housing density (P = 0.05). Full results from the GLM for each species are presented in Supplemental File 1, available online at https://www.eaglehill.us/ NENAonline/suppl-files/n22-1-N1322a-Picone-s1, and, for BioOne subscribers, at http:// dx.doi.org/10.1656/N1322a.s1. Northeastern Naturalist Vol. 22, No. 1 C. Picone 2015 NENHC-35 density in this study because all lakes and most survey sites were adjacent to forests and wetlands that would permit easy dispersal (Fig. 1). More likely, the effects of housing density at my survey sites were from factors that impacted immediate shorelines. For example, artificial light can reduce the breeding calls of Green Frogs by half (Baker and Richardson 2006), anthropogenic noises like engines deter breeding calls of some species (Sun and Narins 2005), and amphibian survival and abundance is reduced by pollutants associated with increased suburbanization, such as heavy metals (e.g., zinc, lead, copper), pesticides from lawns, and nutrient enrichment (Hamer and McDonnell 2008). Finally, housing developments would make the shoreline less attractive to frogs by removing edge vegetation, adding sandy beaches, and/or building retaining walls (C. Picone, pers. observ.). My results also indicate that herbicide treatments did not impact abundance of breeding frogs at sites with low housing densities (less than 0.1 houses/30 m of shoreline), but herbicides may have reduced frog abundance at sites with moderate–high levels of development (>0.2 houses/30 m). This interaction was only marginal (P = 0.06) and was partially caused by a statistical outlier (Stodge Meadow Pond) in a small data set. Therefore, more lakes must be surveyed before concluding whether the herbicide*housing interaction identified in my study is a real phenomenon or simply an artifact of small sample size. Even the marginal effect of herbicides that I detected is worth noting because the methods used in this study provide a conservative assessment of the impacts of herbicides on frogs prone to Type II error. Herbicides were applied in mid-late summer, but frogs were surveyed each subsequent spring. Although unlikely (Mattson et al. 2004, Wagner 2004), if herbicide treatments impaired frog adults or tadpoles, then treated lakes could be recolonized the following spring from adjacent, untreated wetlands, and subsequent surveys might not detect any treatment effects. The frog species in this study are not specialists to lake edges: they also breed in shallow wetlands and small pools. A typical frog population disperses up to 100–800 m to find breeding sites (Semlitsch 2008), so in-migration of frogs from untreated wetlands could have masked potential negative effects from the previous year’s herbicide application. This caveat could be resolved by studies that focus on herbicide impacts especially on tadpoles soon after application in the fie ld. Although such short-term studies have not been done in the field, results from lab and mesocosm studies indicate that 2 common aquatic herbicides, diquat and glyphosate, are unlikely to impair amphibian larvae or adults in the short term. If applied at recommended rates, diquat should only reach temporary concentrations of 0.1 mg/L–0.2 mg/L, and may rarely approach 0.7 mg/L (Mattson et al. 2004, Riley and Finlayson 2004). In lab tests that employed these concentrations, diquat had no detrimental effects on frog embryos or tadpoles (Dial and Buehler-Dial 1987), and even higher concentrations did not increase tadpole mortality in artificial ponds (Cooke 1977). Furthermore, lake sediments rapidly absorb diquat so its half-life in water is generally less than 1 day (Emmett 2002, Mattson et al. 2004, Ritter et al. 2000). Likewise, when the aquatic form of glyphosate (Rodeo®) is applied according to state guidelines (Mattson et al. 2004), and without any toxic surfactants, Northeastern Naturalist NENHC-36 C. Picone 2015 Vol. 22, No. 1 then concentrations are orders of magnitude below the LC50 for tadpoles (Perkins et al. 2000, Riley and Finlayson 2004). In this study, copper-based algaecides were used only at Sunset Lake, and it is concerning that this lake had the lowest total frog abundance and that I heard only 1 species calling there (Green Frogs). When applied as recommended, temporary copper concentrations should rarely reach above 0.3 mg/L (Wagner 2004), but in practice they can reach up to 0.5–1.0 mg/L (Mattson et al. 2004, Riley and Finlayson 2004). All of these concentrations are well above the LC50 of copper sulfate for some species of tadpoles (0.06–0.1 mg/L), and concentrations above 0.01 mg/L can impair tadpole development and escape behavior (Garcia-Muñoz et al. 2011, Gürkan and Hayretdag 2012). Fortunately, concentrations in the water column do not remain high for long periods because copper-based herbicides are absorbed by lake sediments within 1–7 days (Liu et al. 2006, Mattson et al. 2004). Furthermore, short-term (3-year) studies suggest that copper bound in sediments is non-toxic to amphipods and plants (Han et al. 2001). However, effects of long-term copper application and its accumulation have not been tested on amphibians or other nontarget species. Permanent, untreated refuge zones may help maintain frog populations on treated lakes. I observed the lowest frog abundances at lakes with survey sites that were directly treated (Watatic and Sunset Lakes; Fig. 2). Herbicides such as diquat and copper-based algaecides are rapidly bound to lake sediments and will accumulate after years of repeated treatments (Emmett 2002, Hullebusch et al. 2003). Even if they are non-toxic, herbicide residues in sediments might deter adult frogs from breeding in sites that have been directly treated. This question could be addressed with a study comparing treated and untreated survey sites on the same lake. In conclusion, the results of this field study do not provide evidence that herbicide applications are consistently associated with declines of breeding frogs. There may be negative effects of herbicides at higher housing densities, but I observed no effects at lower housing densities. One strategy to help mitigate potential negative effects of herbicide spraying, especially in lakes with higher housing densities, could be to provide permanently unsprayed refuge areas. Given that lakes have been treated with aquatic herbicides for decades, we need to explore more of the long-term impacts on non-target organisms in field settings. Acknowledgments I am deeply grateful to John Ludlam (Fitchburg State University, Fitchburg, MA) for statistical analyses and valuable advice, and to Nick Beauregard, a student volunteer who helped record frog-calling data in 2009. Two anonymous reviewers and the manuscript editor (C. Bevier) greatly improved the writing, analysis, and conclusions in this paper. Several Ashburnham landowners generously granted access for us to survey their lakes and ponds. Literature Cited Baker, B.J., and J.M.L Richardson. 2006. The effect of artificial light on male breedingseason behavior in Green Frogs, Rana clamitans melanota. Canadian Journal of Zoology 84:1528–1532. Northeastern Naturalist Vol. 22, No. 1 C. Picone 2015 NENHC-37 Blaustein, A.R., B.A. Han, R.A. Relyea, P.T.J. Johnson, J.C. Buck, S.S. Gervasi, and L.B. Kats. 2011. The complexity of amphibian population declines: Understanding the role of cofactors in driving amphibian loss. Annals of the New York Academy of Sciences 1223:108–119. Boone, M.D., C.A. Bishop, L.A. Boswell, R.D. Brodman, J.Burger, C. Davidson, M. Gochfeld, J.T. Hoverman, L.A. Neuman-Lee, R.A. Relyea, J.R. Rohr, C. Salice, R.D. Semlitsch, D. Sparling, and S. Weir. 2014. Pesticide regulation amid the influence of industry. Bioscience 64:917–922. Cooke, A.S. 1977. Effects of field applications of the herbicides diquat and dichlobenil on amphibians. Environmental Pollution 12:43–50. Cosentino, B.J., D.M. Marsh, K.S. Jones, J.J. Apodaca, C. Bates, J. Beach, K.H. Beard, K. Becklin, J.M. Bell, C. Crockett, G. Fawson, J. Fjelsted, E.A. Forys, K.S. Genet, M. Grover, J. Holmes, K. Indeck, N.E. Karraker, E. Kilpatrick, T.A. Langen, S.G. Mugel, A. Molina, J.R. Vonesh, R. Weaver, and A. Willey. 2014. Citizen science reveals widespread negative effects of roads on amphibian distributions. Biological Conservation 180:31–38. Cushman, S.A. 2006. Effects of habitat loss and fragmentation on amphibians: A review and prospectus. Biological Conservation 128:231–240. Dawson, A.D., and S.A. Zielinski. 2006. Environmental Handbook for Massachusetts Conservation Commissioners. 9th Edition. Massachusetts Association of Conservation Commissions, Belmont, MA. 656 pp. Dial, N.A., and C.A. Bauer Dial. 1987. Lethal effects of diquat and paraquat on developing frog embryos and 15-day-old tadpoles, Rana pipiens. Bulletin of Environmental Contamination and Toxicology 38:1006–1011. Eigenbrod, F., S.J. Hecnar, and L. Fahrig. 2008. The relative effects of road traffic and forest cover on anuran populations. Biological Conservation 141:35–46. Emmett, K. 2002. Final risk assessment for diquat bromide. Special technical report, Washington State Department of Ecology: The Water Quality Program Publication Number 02-10-046. Available online at https://fortress.wa.gov/ecy/publications/publications/ 0210046.pdf. Accessed 15 June 2014. García-Muñoz, E., F. Guerrero, and G. Parra. 2011. Larval escape behavior in anuran amphibians as a wetland rapid pollution biomarker. Marine and Freshwater Behavior and Physiology 44:109–123. Gürkan, M., and S. Hayretdag. 2012. Morphological and histological effects of copper sulfate on the larval development of Green Toad, Bufo viridis. Turkish Journal of Zoology 36:231–240. Hamer, A.J., and M.J. McDonnell. Amphibian ecology and conservation in the urbanizing world: A review. Biological Conservation 141:2432–2449. Han, F.X., J.A. Hargreaves, W.L. Kingery, D.B. Huggett, and D.K. Schleck. 2001. Accumulation, distribution, and toxicity of copper in sediments of catfish ponds receiving periodic copper sulfate applications. Journal of Environmental Quality 30:912–919. Hayes, T.B., P. Case, S. Chui, D. Chung, C. Haeffele, K. Haston, M. Lee, V.P. Mai, Y. Marjuoa, J. Parker, and M. Tsui. 2006. Pesticide mixtures, endocrine disruption, and amphibian declines: Are we underestimating the impact? Environmental Health Perspectives 114:40–50. Hullebusch, E.V., P. Chatenet, V. Deluchat, P.M. Chazal, D. Froissard, M. Botineau, A. Ghestem, and M. Baudu. 2003. Copper accumulation in a reservoir ecosystem following copper sulfate treatment (St. Germain Les Belles, France). Water, Air, and Soil Pollution 150:3–22. Northeastern Naturalist NENHC-38 C. Picone 2015 Vol. 22, No. 1 Liu, R., D. Zhao, and M. Barnett. 2006. Fate and transport of copper applied in Channel Catfish ponds. Water, Air, and Soil Pollution 176:139–162. Magurran, A.E. 2003. Measuring Biological Diversity. Wiley-Blackwell Publishing, Malden, MA. 264 pp. Mattson, M.D., P.J. Godfrey, R.A. Barletta, and A. Aiello. 2004. Eutrophication and aquatic-plant management in Massachusetts. Final generic environmental impact report. Edited by Kenneth J. Wagner. Special technical report, Department of Environmental Protection and Department of Conservation and Recreation, Executive Office of Environmental Affairs, Commonwealth of Massachusetts, Boston, MA. 620 pp. North American Amphibian Monitoring Program (NAAMP). 2012. Patuxent Wildlife Research Center and US Geological Survey. Available online at www.pwrc.usgs.gov/ naamp/. Accessed 15 June 2014. OLIVER. 2014. MassGIS’s online mapping tool. Available online at http://maps.massgis. state.ma.us/map_ol/oliver.php. Accessed 7 May 2014. Perkins, P.J., H.J. Boermans, and G.R. Stephenson. 2000. Toxicity of glyphosate and triclopyr using the frog-embryo teratogenesis assay: Xenopus. Environmental Toxicology and Chemistry 19:949–945. R Core Team. 2013. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Available online at http://www.R-project. org/. Accessed 30 June 2014. Relyea, R. 2003. Predator cues and pesticides: A double dose of danger for amphibians. Ecological Applications 13:1515–1521. Relyea, R.A. 2005a. The impact of insecticides and herbicides on the biodiversity and productivity of aquatic communities. Ecological Applications 15:618–627. Relyea, R.A. 2005b. The lethal impact of Roundup® on aquatic and terrestrial amphibians. Ecological Applications 15:1118–1124. Relyea, R.A. 2005c. The lethal impact of Roundup® and predatory stress on six species of North American tadpoles. Archives of Environmental Contamination and Toxicology 48:351–357. Relyea, R.A., and N. Diecks. 2008. An unforeseen chain of events: Lethal effects of pesticides at sublethal concentrations. Ecological Applications 18:1728–1742. Relyea, R.A., and D.K. Jones. 2008. Toxicity of Roundup Original Max® to 13 species of larval amphibians. Environmental Toxicology and Chemistry 28:2004–2008. Riley, F., and S. Finlayson. 2004. Acute toxicities of herbicides used to control Water Hyacinth and Brazilian Elodea in larval Delta Smelt and Sacramento Splittail. Special technical report, California Department of Fish and Game, Sacramento, CA. Office of Spill Prevention and Response Administrative Report 04-003. Available online at http://www. cdpr.ca.gov/docs/emon/surfwtr/hazasm/hazasm04_03.pdf. Accessed 15 June 2014. Ritter, A.M., J.L. Shaw, W.M. Williams, and K.Z. Travis. 2000. Characterizing aquatic ecological risks from pesticides using a diquat dibromide case study. I. Probabilistic exposure estimates. Environmental Toxicology and Chemistry 19:749–759. Rohr J.R., and K.A. McCoy. 2010. A quantitative meta-analysis reveals consistent effects of atrazine on freshwater fish and amphibians. Environmental Health Perspectives 118:20–32. Rohr, J.R., A.M. Schotthoefer, T.R. Raffel, H.J. Carrick, N. Halstead, J.T. Hoverman, C.M. Johnson, L.B. Johnson, C. Lieske, M.D. Piwoni, P.K. Schoff, and V.R. Beasley. 2008. Agrochemicals increase trematode infections in a declining amphibian species. Nature 455:1235–1239. Northeastern Naturalist Vol. 22, No. 1 C. Picone 2015 NENHC-39 Semlitsch, R.D. 2008. Differentiating migration and dispersal processes for pond-breeding amphibians. Journal of Wildlife Management 72:260–267. Serdar, D. 1997. Persistance and drift of the aquatic herbicide diquat following application at Steilacoom and Gravelly Lakes. Special technical report, Washington State Department of Ecology, Olympia, WA. Publication No. 97-301. Available online at https:// fortress.wa.gov/ecy/publications/publications/97301.pdf. Accessed 15 June 2014. Sun, J.W.C., and P.M. Narins. 2005. Anthropogenic sounds differentially affect amphibiancall rate. Biological Conservation 121:419–427. Stuart, S.N., J.S. Chanson, N.A. Cox, B.E. Young, A.S.L. Rodrigues, D.L. Fischman, and R.W. Waller. 2004. Status and trends of amphibian declines and extinctions worldwide. Nature 306:1783–1786. Wagner, K.J. 2004. The practical guide to lake management in Massachusetts. A companion to the final generic environmental impact report on eutrophication in aquatic-plant management in Massachusetts. Special technical report, MA Department of Environmental Protection, Department of Conservation and Recreation, and Executive Office of Environmental Affairs. 160 pp. Available online at http://www.mass.gov/eea/docs/dcr/ watersupply/lakepond/downloads/practical-guide-no-pics.pdf. Accessed 30 June 2014.