Questions concerning the potential impact of glyphosate-based herbicides on amphibians

Authors


Abstract

Use of glyphosate-based herbicides is increasing worldwide. The authors review the available data related to potential impacts of these herbicides on amphibians and conduct a qualitative meta-analysis. Because little is known about environmental concentrations of glyphosate in amphibian habitats and virtually nothing is known about environmental concentrations of the substances added to the herbicide formulations that mainly contribute to adverse effects, glyphosate levels can only be seen as approximations for contamination with glyphosate-based herbicides. The impact on amphibians depends on the herbicide formulation, with different sensitivity of taxa and life stages. Effects on development of larvae apparently are the most sensitive endpoints to study. As with other contaminants, costressors mainly increase adverse effects. If and how glyphosate-based herbicides and other pesticides contribute to amphibian decline is not answerable yet due to missing data on how natural populations are affected. Amphibian risk assessment can only be conducted case-specifically, with consideration of the particular herbicide formulation. The authors recommend better monitoring of both amphibian populations and contamination of habitats with glyphosate-based herbicides, not just glyphosate, and suggest including amphibians in standardized test batteries to study at least dermal administration. Environ Toxicol Chem 2013;32:1688–1700. © 2013 SETAC

INTRODUCTION

Glyphosate is the active ingredient in many nonselective (or broad-spectrum) herbicide formulations. It has mortal effects on most species of green plants by inhibiting the monomeric enzyme 5-enolpyruvylshikimate-3-phosphate synthase, the key enzyme of the shikimate pathway for the biosynthesis of aromatic amino acids [1-3]. Also, glyphosate is known to be a strong systemic metal chelator [4], which can impair micronutrient uptake in plants and reduce their growth [5, 6]. Glyphosate-based herbicides—for instance, different formulations under the brand name Roundup—are applied in no-tillage farming in conventional agriculture, winegrowing, forest management, noncultivated areas, and other areas; likewise, glyphosate-based herbicides are applied in private gardening [2, 3]. However, the main use of glyphosate-based herbicides is in the cultivation of genetically modified crops that are made herbicide-resistant. That is, they resist nonselective herbicides at a degree lethal to weeds. Because of the growing importance of herbicide-resistant crops, glyphosate-based herbicides are dominating the worldwide herbicide market [7]. Taking the United States as an example, the glyphosate sales volume has significantly increased since 1996, the year when herbicide-resistant crop cultivation began, from approximately 25 million pounds to 180 million pounds of active ingredient in 2007 [8, 9]. Against the background that the worldwide use of glyphosate-based herbicides is increasing and unparalleled in modern weed management, a comparison of ecotoxicological endpoints (like median lethal concentration [LC50] values) of glyphosate and glyphosate-based herbicides with those of other pesticides and contaminants [10] is a necessary step; however, it is by far insufficient to assess and compare the impacts of different herbicides on amphibians [11].

Herbicide-resistant systems consist of an herbicide-resistant crop and its associated complementary herbicide, which is usually nonselective (glyphosate-based herbicides or others). Herbicide-resistant systems are propagated to result in a net reduction of herbicide use compared with conventional cropping systems. While a large 2-yr field experiment supports this view [12], agricultural practice in the first years since adoption in the United States rather indicates no significant decrease in amounts of active ingredients [13]. A study that covered the whole adoption period of genetically modified crops in the United States concluded that herbicide-resistant systems actually increased herbicide use between 1996 and 2011 [14]. According to the study, which was based on National Agricultural Statistics Service survey data, herbicide use increased by 7%. Herbicide increase seems to be the case in unsustainable monocultures, where rotation of crops and traits is disregarded and increasing resistance of weeds creates a demand for higher amounts of nonselective herbicides as well as the reuse of selective herbicides [14]. The number of herbicide-resistant weeds has increased since the 1980s (http://www.weedscience.org/chronIncrease.gif). In the case of glyphosate, a total of 23 glyphosate-resistant weeds have been identified worldwide as of July 2012 (http://www.weedscience.org/summary/MOASummary.asp).

The environmental safety of glyphosate, compared to the herbicides it replaces, is frequently highlighted [13]. Based on official ecotoxicological standard tests, glyphosate is considered to be “practically nontoxic” to “slightly toxic” to animals [3, 15, 16]. However, this information cannot be simply transferred to glyphosate-based herbicides because some added substances are occasionally more toxic than the active ingredient itself, glyphosate [3, 17]. Higher toxicity of glyphosate-based herbicides compared to glyphosate, especially to organisms in the aquatic environment, is mainly caused by surfactants that are added for better entry of the active ingredient into the plant tissue [3].

Amphibians are not standard test organisms in official studies, and usually results from fish and aquatic invertebrates (for aquatic amphibian life stages), birds, and mammals (for terrestrial amphibian life stages) are taken as proxies [11]. Both the absence of amphibians from such tests and the transferring of results from other animal groups to amphibians are questionable for 3 reasons. First, worldwide, amphibians are dramatically declining, with more than one-third of all known (∼7000) species threatened with extinction. Within the last 3 decades, we have witnessed species and population declines and extinctions at alarming rates, and environmental contaminants are supposed to play a role [18, 19]. Second, most amphibian species are ecologically unrivaled among vertebrates by spending part of their early life in water, while later (adult) stages are on land. Consequently, these organisms face an unparalleled risk of contamination with glyphosate-based herbicides (and other pesticides) in both the aquatic and the terrestrial milieus. Several methods of exposure are possible such as overspraying [20, 21], runoff after downpour [22, 23], airborne drift [24-26], and contact with residues on soil and plant material [27]. Moreover, amphibians may incorporate contaminants via their larval plant and adult insect diets [28]. Finally, amphibians are biologically unique among all vertebrates. They are the only tetrapods with an embryo not protected by an amnion cavity and with a free-swimming larva. In addition, they are the only Anamnia (i.e., fishes and amphibians) which breathe through well-developed lungs and have a highly permeable skin used for relevant uptake from the environment [29]. For these reasons, it may be grossly misleading when conclusions gained in tests with warm-blooded tetrapods (mammals and birds) are transferred to amphibians. Test substances are administered orally to mammals and birds [11]; but for amphibians, dermal uptake seems more important. Xenobiotics can diffuse into amphibians 1 or 2 orders of magnitude faster (depending on the species' hydrophobicity) than into mammals [30]. Although most tests with teleost fish were conducted with larval forms that also have a permeable skin, transferring those results to amphibian larvae may also be misleading because amphibian species differ considerably in their response to herbicides. The present study deals, among other things, with the question of which cases data from standard test organisms may be transferred to amphibians.

When it comes to glyphosate-based herbicide concentrations in the environment that are relevant to amphibians, experimental studies draw inconsistent conclusions. Numerous studies are in agreement that at least some glyphosate-based herbicides—in particular those with the surfactant polyethoxylated tallowamine or polyethoxylated alkylamine—can have substantial lethal or sublethal effects on amphibians in the environment. Other studies have found deleterious effects on amphibians as well, but concluded no risk from the use of polyethoxylated alkylamine–containing glyphosate-based herbicides under realistic environmental exposures and normal-use scenarios. Results mainly refer to frogs and toads (Anura) since the other 2 amphibian groups (Caudata, Gymnophiona) remain underrepresented in studies. Up to now, amphibian experiments have shown considerable variation in terms of their aims, applied designs (glyphosate-based herbicide formulation, concentration, renewal of test media, etc.), and the species and life stages examined, which makes it a challenge to directly compare them. The purpose of the present study is to review the available literature and conduct a qualitative meta-analysis concerning the potential effects of glyphosate and glyphosate-based herbicides on specific endpoints. We consider such a synthesis important against the background of the increasing use of glyphosate-based herbicides, the different biology of amphibians and test organisms, and the ongoing amphibian decline. Furthermore, we understand amphibians as wild animals that persist in agricultural landscapes because, in many countries, primary amphibian habitats such as natural wetlands and floodplains have been mainly destroyed to expand the arable land [31].

Environmental fate and concentrations

First of all, it is most important to realize that we know next to nothing about the environmental concentrations of surfactants and adjuvants used in glyphosate-based herbicides. Therefore, glyphosate concentrations can only be used as approximations for contamination with glyphosate-based herbicides.

In both soil and water, glyphosate is almost exclusively degraded by microorganisms to its main metabolite aminomethylphosphonic acid and eventually to carbon dioxide [1]. Glyphosate and aminomethylphosphonic acid are rapidly immobilized by adsorption to soil and sediment. Because this is highly dependent on soil type, temperature, and so on, half-life values of glyphosate in soil can range from a few days to over 200 d [1, 32-34]. Aminomethylphosphonic acid shows a significantly stronger adsorption behavior [35], making it less available to microorganisms [36] and leading to commonly longer half-life values (∼80–900 d [1]) than for glyphosate. In water, half-life values of both glyphosate and aminomethylphosphonic acid are estimated to range from 7 d to 14 d [1]. Glyphosate competes with inorganic phosphate for soil binding sites [37-39], and adsorbed glyphosate can be remobilized by phosphate fertilization [40]. For aminomethylphosphonic acid, long-term leaching from fields that were treated with glyphosate-based herbicide years before has been reported [41]. In general, it should be understood that although glyphosate is the main source of aminomethylphosphonic acid, it is not the only source because other phosphonate compounds are degraded to aminomethylphosphonic acid as well [42]. Since many amphibian species choose nonflowing, small, and shallow water bodies for reproduction, data on glyphosate and aminomethylphosphonic acid concentrations in these kinds of water bodies are more relevant than in large, deep, or flowing waters [20, 31, 43]. The rate of glyphosate dissipation from such small water bodies is mainly a function of the local conditions and should be considered site-specific [1].

Although data about the concentration of glyphosate and aminomethylphosphonic acid in soil and sediments are available, risks to amphibians cannot be well assessed because specific studies are missing. Negative effects on amphibians by simple contact with contaminated soil have been suggested for pesticides other than glyphosate [44]. Although both glyphosate and aminomethylphosphonic acid should be strongly adsorbed to the soil for most of their time (compared with other substances), it should not be dismissed that terrestrial life stages of amphibians may suffer from contact with soil (or plant material) contaminated with glyphosate [27]. Furthermore, many amphibian larvae incorporate sediment during feeding [29] (e.g., by grazing algae); how much soil they take up and whether adsorbed pesticides become remobilized in the larvae has not been investigated so far.

Herbicides can be wind-drifted from the field into aquatic and terrestrial habitats during application [45], but the scale is largely unknown [24-26]. Maximum drift rates of 9.1 ng/m3 for glyphosate and 0.97 ng/m3 for aminomethylphosphonic acid were reported [46]. It is notable that both glyphosate and aminomethylphosphonic acid were also found in rainfall [46-48] and even in minipools of rainforest plants (phytotelmata) [49]. The concentrations in rain, reported for the first time, were generally low (glyphosate maximum of 2.5 µg acid equivalent [a.e.]/L, aminomethylphosphonic acid maximum of 0.48 µg/L [46]) but demonstrate that environmental contamination of glyphosate by drift and exposure to amphibians in this way is a realistic scenario [49].

Concentrations in water

It is notable that in water aminomethylphosphonic acid was detected more frequently than glyphosate but usually at much lower concentrations [23, 47, 50]. This is because aminomethylphosphonic acid stays longer in soil and sediment than glyphosate and gradually leaches from it into the water body [41]. Only 1 study reported similar concentrations for both glyphosate and aminomethylphosphonic acid (maximum of ∼0.4 mg/L [51]) in a subsurface drain, which is not an amphibian habitat. Due to its relatively rapid dissipation, it can be assumed that most glyphosate values measured directly in the environment are probably below the peak concentrations present directly after herbicide applications or after downpour. However, knowing maximum concentrations—even if they prevail for short periods only—is crucial because most studies observed acute toxic effects of glyphosate-based herbicides on tadpoles, that is, immediately within the first 24 h after tadpoles were exposed to the glyphosate-based herbicide [11, 52-54].

Because analysis of glyphosate is more expensive and challenging than for many other pesticides—which can be determined simultaneously by gas chromatography–mass spectrometry [55]—the data on environmental concentrations of glyphosate are generally sparse, especially for small, shallow water bodies that are often inhabited by amphibians [50, 56]. Water bodies are directly oversprayed when aquatic weeds are combated [56], and only certain formulations, which are less toxic, are labeled for use against aquatic weeds such as Accord. In some countries, aerial application of glyphosate-based herbicides is practiced in agricultural herbicide-resistant systems and in forest management to kill broadleaf trees and favor the more marketable conifer trees (e.g., in Canada the formulation Vision, which is equivalent to Roundup Original) [57]. It can be postulated that aerial application in agriculture and forestry is responsible for the highest glyphosate concentrations in aquatic habitats.

Most available data on surface water concentrations are from pesticide-monitoring programs in the United States and relate to streams or large nonflowing waters like lakes [47], which are not commonly used by amphibians [56]. For glyphosate, concentrations in surface waters can be divided into the following 3 groups according to the applied method: 1) those directly measured in the environment without intervention (available for the Americas and Europe); 2) those measured in field experiments shortly after application or in runoff during the first heavy rains (available for the Americas and Europe); and 3) those estimated in worst-case scenarios (available for North America and Germany). The 5 maximum glyphosate concentrations range from 0.17 mg a.e./L to 0.70 mg a.e./L for group 1, 0.27 mg a.e./L to 3.10 mg a.e./L for group 2, and 1.43 mg a.e./L to 7.60 mg a.e./L for group 3 (see Table 1). In group 1, some high concentrations derive from direct overspraying with glyphosate-based herbicides approved for aquatic use (e.g., Rodeo), but the highest concentration of 0.70 mg a.e./L is from waters near an herbicide-resistant soybean cultivation area [23]. In group 3, the highest expected environmental concentration in surface water of 7.6 mg a.e./L relates to the scenario of a direct application of Roundup (360 g active ingredient [a.i.]/L) at the maximum rate to a lentic water body of 5 cm in depth, like a flooded field [58]. The worst-case expected environmental concentrations for glyphosate calculated by national agencies are usually lower (e.g., 1.4 mg a.e./L for Canada [59] and 0.9 mg a.e./L for Germany [60]) than those calculated by some researchers (see group 3 in Table 1).

Table 1. Top 5 maximum glyphosate concentrations in surface waters as reported in various studiesa
Environmental sampleField shortly after applicationWorst-case expected environmental concentrationReferences
  1. a Measured in 1) environmental samples (degradation state unknown); 2) field studies (shortly after application); and 3) estimated worst-case scenarios (expected environmental concentration or predicted environmental concentration). Amounts given in milligrams acid equivalent per liter (mg a.e./L).
0.703.107.60[23, 52, 58]
0.431.952.90[47, 51, 57, 74]
0.331.702.80[1, 56]
0.291.242.70[51, 148, 149]
0.170.271.43[150-152]

In the European Union the Water Framework Directive (Directive 2000/60/EC) provides a procedure to set Environmental Quality Standards. These are 100 µg a.e./L for glyphosate and 450 µg/L for aminomethylphosphonic acid (http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=CELEX:32000L0060:EN:HTML). Taking Germany as an example, most glyphosate-based herbicides require one to maintain a distance of 5 m to 20 m from adjacent water bodies during application. A 5-m buffer strip would reduce the expected environmental concentration of 0.9 mg a.e./L to approximately 0.005 mg a.e./L [60] (note that only pesticide drift is calculated, not runoff or leaching). There are also some formulations such as Roundup UltraMax and Roundup Turbo where no distance has to be regarded, although these 2 formulations have tallowamine surfactant systems (http://www.roundup.de); but regular buffer strips between 5 m and 50 m are required in most of the different German states (http://www.umweltdaten.de/wasser/laender.pdf). This means that in most water bodies adjacent to fields the concentration of glyphosate should not legally exceed 5 µg a.e./L, if herbicides are applied according to the provisions. However, small, ephemeral ones like puddles, which are also used by some amphibian species for reproduction, are usually not regarded as water bodies and, therefore, are not covered by application provisions [56, 61].

METHODS

We searched the Web of Knowledge and Google Scholar for published scientific articles with the following keywords (latest by 15 March 2012): glyphosate*, glyphosate* + amphibia*, Roundup*, Roundup* + amphibia*. Furthermore, we employed the database ECOTOX (US Environmental Protection Agency [USEPA]). We examined the references of the retrieved publications for further information. In addition, a limited amount of miscellaneous gray literature, largely distributed via national authorities, was used when relevant. The debate as to whether amphibians could be affected in their natural environment is sometimes perhaps steered by motivation other than scientific research because some authors question each other's vested interests [11, 43, 62]. Nevertheless, or precisely because of this, we included all and did not exclude information based on its source. To conduct a qualitative meta-analysis, we used the vote-counting method [63, 64] and quantified how many studies found or did not find significant effects (α = 0.05) of glyphosate and glyphosate-based herbicides on the following response variables: unnatural deformity rates of larvae, endocrine disruption in larvae, effects on development, inhibition of specific enzymes, genotoxic effects, effects on embryos, behavioral effects, abiotic interactions, and biotic interactions. We used Fisher's exact test to test against the null hypothesis that glyphosate and glyphosate-based herbicides do not have effects on the considered response variable. All studies on acute toxicity found dose–response relationships, and we solely descriptively show published LC50 values to compare them with environmentally relevant concentrations. Analyses were performed with the software R (R Developmental Core Team, 2009).

RESULTS

In total, 51 toxicological studies concerning the impact of glyphosate or glyphosate-based herbicides on amphibians were found; important details are summarized in Supplemental Data, Table S1. Most studies (n = 36) investigated effects on tadpoles (i.e., anuran larvae); a few investigated embryos (n = 5), terrestrial life stages of anurans (n = 8), or newts and salamanders (all life stages, n = 6); and overall, no studies at all are known on amphibians of the order Gymnophiona (caecilians). Only in an (anecdotal) field observation, terrestrial life stages of a caecilian species suffered burnlike wounds 5 h to 7 h after Roundup applications in a tea plantation in Sri Lanka [65].

Acute toxicity

Toxicological studies focused mainly on larval anurans (tadpoles) from North America. It is challenging to directly compare the various experiments because of differences in design, length, and exposure conditions. Nevertheless, some general conclusions can be drawn.

Terrestrial life stages

Generally, absorption of pesticides is faster in amphibians than other vertebrates [30]. However, effects of glyphosate and glyphosate-based herbicides on terrestrial life stages of amphibians have rarely been investigated. Direct overspraying at recommended application rates resulted in 79% of the tested juvenile frogs and toads from the United States dying within 24 h in 1 experiment [20] and up to 30% in another carried out on Colombian anurans [21]. Not all glyphosate-based herbicides seem to pose immediate risks for amphibians under field conditions because mortality can vary between 0% and approximately 80%, depending on the formulation [66]. The presence of soil or litter remarkably reduced adverse effects on oversprayed terrestrial life stages in 2 cases [21, 66]. When compared with direct overspraying, exposure of adults and metamorphs of an Australian frog species to dilute glyphosate-based herbicides and glyphosate solutions revealed slightly different 48-h LC50 values (∼50 mg a.e./L and 80 mg a.e./L, respectively) [58].

Aquatic larvae

Studies have shown that both the salt and acid forms of glyphosate, which are used as active ingredients, are slightly toxic (10 mg/L < LC50 < 100 mg/L) or even practically nontoxic (LC50 > 100 mg/L) to tadpoles [58, 67]. One study stated a 96-h LC50 value as low as 6.5 mg a.e./L for glyphosate, but this value should be treated with caution because it is based on tests with a glyphosate-based herbicide formulation and a surfactant, rather than with glyphosate directly [52]. Some glyphosate-based herbicides are classed as slightly toxic to practically nontoxic (e.g., Roundup Biactive) [58, 67]. Conversely, other glyphosate-based herbicides (e.g., Roundup Original) have been found to be moderately toxic (1 mg/L < LC50 < 10 mg/L) or even highly toxic (0.1 mg/L < LC50 < 1 mg/L) to tadpoles [68, 69]. Categories are those defined by the USEPA. When LC50 values of different glyphosate-based herbicides for 37 amphibian larvae species were plotted (Figure 1), the median was approximately 2 mg a.e./L regardless of whether the less toxic glyphosate-based herbicides, which are also mainly approved for aquatic weed management, were considered (Figure 1A) or just moderately toxic glyphosate-based herbicides (Figure 1B).

Figure 1.

Median lethal concentration (LC50) values of glyphosate-based herbicide for amphibian larvae of 37 species. All values are in milligrams acid equivalent per liter (mg a.e./L) [21, 52, 58, 59, 68, 69, 73, 76, 80, 84, 85, 103, 114, 153]. (A) With practically nontoxic (PNT) and slightly toxic (ST) glyphosate-based herbicide (tested glyphosate-based herbicide = 18, minimum = 0.20, 1st quartile = 1.70, median = 2.15, mean = 21.70, 3rd quartile = 3.45, maximum = 494.00). (B) Without PNT and ST glyphosate-based herbicide (tested glyphosate-based herbicide = 12, minimum = 0.20, 1st quartile = 1.60, median = 2.05, mean = 2.53, 3rd quartile = 2.80, maximum = 9.00).

The relatively wide range of toxicity seems to be associated with the different surfactants present in the glyphosate-based herbicides. Surfactants are mainly from the alkoxylated alkyl amine family [55]. In concert with several other studies on aquatic organisms [70-72], glyphosate-based herbicides with tallowamine surfactants, especially polyethoxylated alkylamine, appear to be among the most dangerous formulations to tadpoles [20, 55, 58, 73]. Polyethoxylated alkylamine appears to contribute most to the acute toxicity since it has been shown that polyethoxylated alkylamine is more toxic than the glyphosate-based herbicide itself (and the glyphosate-based herbicide is more toxic than only glyphosate) [73, 74]. Polyethoxylated alkylamine is supposed to take effect by increasing the membrane permeability of amphibian skin melanophores, which was observed in the African clawed frog (Xenopus laevis) [75]. Also, polyethoxylated alkylamine has been shown to disrupt respiratory membranes in aquatic organisms [66] and is considered responsible for gill damage [55, 68, 76]. In concert with these findings, for instance, Roundup Original (including polyethoxylated alkylamine) was found to be highly toxic to larval Pacific chorus frog (Pseudacris regilla) [69]. The 24-h LC50 (0.23 mg a.e./L) was below drinking water standards defined by the USEPA [55]. However, also other surfactants can pose risks to tadpoles; for example, glyphosate-based herbicides with nonylphenol surfactants are also more toxic than glyphosate alone [52], and new glyphosate-based herbicide formulations such as Roundup OriginalMAX and Roundup WeatherMAX (with unknown surfactant systems) are more or less equally or even more highly toxic to tadpoles than polyethoxylated alkylamine–containing glyphosate-based herbicides [11, 68]. Roundup OriginalMAX was highly toxic to tadpoles of the American bullfrog (Lithobates catesbeianus) and the spring peeper (Pseudacris crucifer) [68]. In the most extreme case, exposure to Roundup WeatherMAX at levels as low as drinking water standards (0.57 mg a.e./L) resulted in 80% mortality in tadpoles of the gray tree frog (Hyla versicolor) [55]. Overall, glyphosate-based herbicides including surfactants are known to induce oxidative stress in larvae of the neotropical toad Rhinella arenarum [77] and American bullfrogs [78], which could be another mechanism for toxicity next to the disruption of skin melanophores or respiratory membranes [77, 78].

Variation in sensitivity

Some light may be shed on the intraspecific (within a species) and interspecific (between species) variation in sensitivity to glyphosate and glyphosate-based herbicides to better understand some varying or seemingly inconsistent outcomes. Intraspecific variation was observed for the different life stages of amphibians. Tadpoles form a rather sensitive life stage compared to terrestrial life stages [58, 67] and embryos of the same species, apparently because the latter miss target organs, such as functional gills, or they have become insensitive [59, 73]. Tadpoles of later developmental stages (commonly these are expressed in Gosner stages [79]) were sometimes more sensitive than earlier conspecifics, which may be explained in some cases with space competition in larger tadpoles [53]. In several cases, it could be shown that reactions were species-specific. Differences in sensitivity over anuran families have been found. The general sensitivity order to glyphosate-based herbicides of common amphibian families is: Hylidae > Ranidae > Bufonidae [55, 69]. Note that this order is not universal for all pesticides as, for instance, ranid frogs are supposed to be more sensitive to endosulfan than hylid frogs [80].

Salamander larvae seem to be more resistant than those of anurans [68, 69]. For example, different from tadpoles, no mortality of larval salamanders was observed after 24 h in 1 study [69]. This could be related to the fact that carnivorous salamander larvae filtrate much less water by time and have a lower activity rate than sometimes obligately filtrating anuran larvae with higher activity during foraging [29].

With regard to variation at the population level, acute toxicity tests on 2 spadefoot species larvae (Spea bombifrons, S. multiplicata), from cropland and natural grassland, indicated no significant differences when exposed to Roundup WeatherMAX [76]. Pesticide tolerance because of regular applications in cropland has not been observed. A related study showed varying sensitivities among 10 populations of southern leopard frogs (Lithobates sphenocephalus) obtained from different localities which were all exposed to the same concentrations of the insecticide carbaryl [81]. However, since previous pesticide exposure or land use around the populations was not considered, it remains unclear if the observed variations were in fact caused by local adaption to the pesticide or just because of random effects of geographic variation [81]. In addition, significant carbaryl tolerance among individuals of the same population has been found for this species and for Hyla versicolor and linked to genetic variations [82, 83]. Likewise, slightly different 96-h LC50 values (both studies with a nonrenewal design) of 2.0 mg a.e./L [73] and 4.22 mg a.e./L for Roundup Original [84] were observed for green frog (Lithobates clamitans) larvae in the same developmental stage but sampled from different localities. However, many other factors, such as variation between the laboratories, may be responsible for the difference.

Effects of sublethal concentrations

Unnatural deformity rates in larvae

Just 2 studies investigated deformity rates after glyphosate-based herbicide exposure and found effects, but because of the small number, the null hypothesis that glyphosate-based herbicides have no effects on deformity rates cannot be rejected (p > 0.05) and further studies are required. Internal and external damages (craniofacial and mouth deformities, eye abnormalities, and bent, curved tails) were observed directly after 96-h LC50 acute toxicity testing in surviving tree frog (Scinax nasicus) tadpoles [85] and in northern leopard frog (Lithobates pipiens) larvae when they were raised until metamorphosis [73].

Endocrine disruption in larvae

One study reported endocrine effects of polyethoxylated alkylamine and glyphosate-based herbicides, while another reported no effects of glyphosate (p > 0.05). In northern leopard frogs the thyroid axis was disrupted because of increased thyroid hormone receptor β mRNA expression when tadpoles were exposed to Roundup Original at 1.8 mg a.e./L and Roundup Transorb at both 0.6 mg a.e./L and 1.8 mg a.e./L [73]. Whether some glyphosate-based herbicides may affect the sexual development of frogs and thereby affect the reproductive portion of a population remains unclear. While about 20% of northern leopard frogs developed abnormal gonads (intersex) after exposure to polyethoxylated alkylamine, Roundup Original, or Roundup Transorb, the sex ratio was unaffected when compared to the controls [73]. Another study, which tested glyphosate but not glyphosate-based herbicides, found that gonadal steroidogenesis in Pelophylax kl. esculentus was not affected by glyphosate but was by another herbicide (paraquat) [86].

Effects on development

Here, 6 out of 6 studies found effects (p < 0.01) at least on some of the studied species. Precipitated metamorphosis (Rana cascadae) [87] and delayed metamorphosis (L. pipiens, Anaxyrus americanus) [55, 73] have been observed, apparently depending on the studied species. Delayed time to metamorphosis can be caused by disruption of the thyroid axis (see Endocrine disruption in larvae) or energy consumptive detoxification processes in tadpoles. Also, tadpoles of some species are able to metamorphose earlier under stress, for example, when predators are present or when the pond is drying out [88, 89]. In a field experiment which studied the effect of the glyphosate-based herbicide formulation Accord, with nonylphenol ethoxylate as surfactant, larvae of northern leopard frogs increased in size, while larval tiger salamanders (Ambystoma tigrinum) decreased in a density-dependent manner [90]. Another study found that Roundup OriginalMAX induced morphological changes in tadpoles similar to the adaptations induced by predator cues [92], suggesting that the herbicide activates developmental pathways used for antipredator responses [92].

Inhibition of specific enzymes

Only 1 study investigated inhibition of certain enzymes and observed effects after short exposure (24–48 h). Some of the tested glyphosate-based herbicides inhibited enzymes involved in neurotransmission and detoxification (esterases) at sublethal concentrations in the test species, Rhinella arenarum [77]. Inhibition of these enzymes put the animals under general stress and reduced their individual fitness. Similar to acute toxicity, the responses of the test species remarkably differed with the tested glyphosate-based herbicide [77]. It is notable that Samsel and Seneff [93] compiled evidence that glyphosate inhibits cytochrome P450 enzymes in human liver cells and rats. Like esterases, cytochrome P450 enzymes are involved in detoxification of xenobiotics; the authors conclude that glyphosate enhances the damaging effect of environmental toxins. Since cytochrome P450 enzymes are widespread in nature, their possible inhibition by glyphosate should be tested in amphibians as well because that could explain the synergistic effects of glyphosate with other pesticides.

Genotoxic effects

Both retrieved studies concerning this endpoint found genotoxic effects (but p > 0.05, see Unnatural deformity rates in larvae). Using the standard methods comet assay and micronucleus test, it was found that some glyphosate-based herbicides have a genotoxic and mutagenic potential on tadpoles of the American bullfrog [94] and on adults of the neotropical Odontophrynus cordobae and Rhinella arenarum [95]. Damage of this kind could negatively affect individuals as well as their offspring. Out of 5 tested herbicides, Roundup Original (together with a metalochlor formulation) had the greatest impact on tadpoles of the American bullfrog [94].

Effects on embryos

Four out of 5 studies found effects on embryos (p < 0.05), but only 1 directly stated teratogenicity—increased mortality and malformation rates [91]. One study observed increased malformation rates only at concentrations above the 96-h LC50 [74], and similarly, 2 other studies reported effects on body indices [96, 97]. Applying the frog embryo teratogenesis assay–Xenopus, no significant increase of malformation in tadpoles was observed at concentrations of Roundup Original and Rodeo that were below 96-h LC50 for embryos. The surfactant polyethoxylated alkylamine alone was more toxic (96-h LC50 6.8 mg a.e./L) than Roundup Original containing polyethoxylated alkylamine, which was 700 times more toxic than Rodeo, which lacks a tallowamine surfactant (96-h LC50 9.3 mg a.e./L and over 7000 mg a.e./L, respectively) [74]. Using a different study design from the standardized frog embryo teratogenesis assay–Xenopus, teratogenic effects of both glyphosate and glyphosate-based herbicides on X. laevis have been reported [91]. In that study, embryos showed an increase of head defects and craniofacial malformations when they were exposed to a one–five-thousandth dilution of Roundup Classic equaling 72 mg a.e./L (i.e., nearly the 8-fold LC50 found in Perkins et al. [74]) or when glyphosate was directly injected (360 pg and 500 pg). That study [91] was followed by strong replies about the pertinence of study designs (especially high doses and injection) and conflicting interests [60]. In another study, late embryos of X. laevis that were in the period of organ morphogenesis were exposed to different concentrations of Roundup Original (0.25 mg, 0.5 mg, 1 mg, and 5 mg of formulation per liter) for only 2 d. No significant mortality or malformation rate was observed [98]. When glyphosate-based herbicide nonylphenol surfactants were tested, they did not produce significant teratogenic effects in anuran embryos but inhibited growth [97]. Roundup (exact formulation unknown) had no impact on survival or malformation rate in an experiment with embryonic stages of the gold-striped salamander (Chioglossa lusitanica), but exposed embryos had a larger mean size at hatching than the control group [96]. One may postulate that the observed increase in body size is related to an impact of the glyphosate-based herbicide on the thyroid hormonal balance as observed in other studies [99].

Behavioral effects

One study stated effects and one did not (p > 0.05). Gray tree frogs similarly avoided artificial breeding ponds containing either predatory fish or glyphosate-based herbicide [100]. It is intriguing to suggest that the tree frogs may both perceive external stimuli and trigger similar reactions as already suggested for morphological adaptations in tadpoles (see Effects on development). However, the sample size was rather small. In another study, no effects on site selection in European frog and newt species were observed, for either glyphosate or glyphosate-based herbicides [101].

Interactions with abiotic and biotic stressors

Some authors have tried to simulate aquatic communities in which amphibian larvae (mainly tadpoles) occur [20, 102-105]. In particular, the findings derived from such mesocosm studies have provided information on how applied glyphosate-based herbicides can interact with other factors, including stressors, that can be present in larval amphibian habitats. Others have tested the effect of glyphosate-based herbicide treatment and 1 additional stressor under laboratory conditions [59, 106, 107].

Abiotic interactions

Seven out of 8 studies stated effects of abiotic costressors on considered endpoints (p < 0.01). Several mesocosm studies which tested glyphosate or glyphosate-based herbicides included sediments in their setting. It is a matter of discussion whether sediments interfere with the outcome due to adsorption of the test substances. While 1 study suggests just that [21], another denies it and assumes that the primary, acute toxic effect of the substances takes place before they adsorb to the sediment [20]. In 1 mesocosm study, mixtures of Roundup Original and other pesticides (4 other herbicides and 5 insecticides) at sublethal concentrations (10 µg a.e./L of each pesticide) affected larval survival because of indirect effects [105]. Each pesticide was tested singly and in combination with other pesticides on tadpoles of 2 North American species (L. pipiens, Hyla versicolor). Insecticide-induced decline of zooplankton, which is the assumed primary acute effect, led to an increase of phytoplankton and—because of increased shadowing—a decrease of periphyton, which is consumed by tadpoles. Hence, insecticides (particularly endosulfan and diazinon) had the largest impact on tadpole survival and single glyphosate-based herbicide effects did not stand out [105]. Another study investigated the impact of pesticide mixtures on tadpoles of North American anurans (L. pipiens, L. clamitans, L. catesbeianus, A. americanus, Hyla versicolor) under laboratory conditions [106]. Each pesticide was tested alone (i.e., Roundup Original and 3 insecticides) at 1 mg a.i./L and 2 mg a.i./L, respectively, and in pairwise combinations (1 mg a.i./L of each pesticide). Larval growth was affected in nearly all cases. At low concentrations, pesticides alone had no negative impact on tadpole survival. At high concentrations, all pesticides tested—except for carbaryl—caused significant tadpole mortality. Pesticide mixtures occasionally affected larvae more than single pesticide treatments, but the impact of pesticide mixtures on survival never exceeded that of the single pesticides at 2 mg a.i./L. Therefore, growth and survival can be predicted from the total concentration of the tested pesticides [106]. With regard to Caudata, mixtures of Roundup Plus and a fertilizer had no effect on salamander embryos [96]. Other studies showed that mortal effects of glyphosate-based herbicides including the surfactant polyethoxylated alkylamine increased at higher pH levels. This may be related to the circumstance that the nonionized form of polyethoxylated alkylamine, which is present at higher pH levels, readily accumulates on the gills of tadpoles [59, 107]. Additional sublethal ultraviolet-B radiation to glyphosate or Roundup exposure significantly reduced survival of L. clamitans tadpoles [108].

Biotic interactions

Eight out of 12 studies stated effects of biotic costressors on considered endpoints (p < 0.01). An increase of trematode infection in glyphosate-treated (3.7 mg a.e./L) tadpoles has been found [109], suggesting immunosuppressive effects; but no synergistic effects of glyphosate-based herbicide exposure and zoospores of the amphibian chytrid fungus (Batrachochytrium dendrobatidis) have been observed [110-112]. This is the agent of the dangerous emerging amphibian disease chytridiomycosis, hypothetically responsible for population declines in many species [18, 19]. There is also a study reporting that a glyphosate-based herbicide, here Roundup WeatherMAX at 2.89 mg a.e./L, reduced the mortality of wood frog (Lithobates sylvaticus) tadpoles caused by these zoospores [113]. A plausible explanation is that treatment with Roundup WeatherMAX affected the pathogen more than the frogs [113], which is in line with reports about the inhibitory effect of glyphosate on fungi [3]. It is surprising that in that study, the glyphosate-based herbicide on its own had no effect on the survival of tadpoles (the experiment lasted until metamorphosis), which is not in line with 96-h LC50 values (nonrenewal) determined for Roundup WeatherMAX (1.33–3.26 mg a.e./L) in 6 other North American anuran species [84]. However, this could be related to the aforementioned variation in sensitivity or differences between the laboratories.

Predator cues and competition can enhance the toxicity of contaminants including pesticides [90, 103, 114]. Predator cues can make pesticides significantly more toxic to tadpoles (up to 46 times for carbaryl) [115, 116]. However, only some species were affected by the presence of predator cues when exposed to a glyphosate-based herbicide [114]. Experiments have demonstrated that intraspecific competition among individual American bullfrogs not only reduced their growth but also could make individual tadpoles more susceptible to glyphosate-based herbicides [103]. In another study, North American tree frogs (Hyla versicolor, H. chrysoscelis) avoided breeding ponds that contained either predatory fish or glyphosate-based herbicide [100].

DISCUSSION

Acute toxicity

The literature survey has revealed that the sensitivity of amphibians to glyphosate-based herbicides is formulation-, species-, and life stage–specific (surfactants > glyphosate-based herbicide > glyphosate; Anura > Urodela; Hylidae > Ranidae > Bufonidae; larvae > embryos > juveniles and adults). Glyphosate-based herbicides with polyethoxylated alkylamine or other tallowamine surfactants are moderately toxic or even highly toxic to larvae, but so are some new glyphosate-based herbicide formulations that contain unspecified surfactants. Direct overspraying of terrestrial life stages with glyphosate-based herbicides can pose risk, the actual harm depending on other factors including formulation, vegetation buffers, etc. Because the potential impact of most pesticides on terrestrial life stages is still unknown [27], a comparison of the application of glyphosate-based herbicides with other herbicides is not possible yet, and further research is needed. Overall, glyphosate-based herbicides can be classified as moderately toxic (Figure 1) and glyphosate as slightly toxic to practically nontoxic to amphibian larvae. A general classification of the toxicity of glyphosate-based herbicides or glyphosate for embryos and terrestrial life stages seems not to be reasonable due to a lack of studies.

Effects of sublethal concentrations

Results of different studies indicate that also chronic and delayed effects are formulation- and species-specific. For example, glyphosate-based herbicide exposure precipitated metamorphosis in R. cascadae [87], delayed it in L. pipiens and A. americanus, and hardly affected it in other anuran species [55, 73]. Although there has been more research on the toxicological effects of glyphosate and glyphosate-based herbicides on amphibians compared to other pesticides, studies on the effects of sublethal exposure do not allow conclusions about most considered endpoints due to a lack of data. Apparently, developmental parameters of larvae (e.g., body indices at metamorphosis, time to metamorphosis) are the most sensitive endpoints to study, in so far as this is known. This corresponds with the finding that the aquatic larval stage is also the most sensitive life stage.

The occurrence of abnormal gonads (testicular oocytes) in the study by Howe et al. [73] has recently been attributed to normal-development hermaphroditism in leopard frogs [117]. There could also be variation in response to endocrine disrupters based on the life history of the species exposed [118], and no definitive answer could be given because of the small number of studies.

It remains also unclear if glyphosate and glyphosate-based herbicides have teratogenic effects, and further studies are desirable. In this context, the extent to which X. laevis represents the response of other taxa will be briefly discussed. Some authors question if X. laevis is sensitive enough to be a sentinel species. Although husbandry conditions can affect the sensitivity of this species [119], it is a good sentinel species for, among other things, endocrine responses [63, 120]. Several studies on acute and developmental toxicity compared responses of X. laevis and other, phylogenetically distinct anuran species to different contaminants. The results showed that X. laevis is among the more sensitive and often the most sensitive of the species tested [98, 121-123]. When responses of embryos and early larvae of several amphibian species including X. laevis to the glyphosate-based herbicide Vision were compared at varying pH values, there was no clear trend [59]; but in a different study, agricultural surfactants narcotized (median effective concentration) tadpoles of X. laevis more than larvae of other species [124]. In general, X. laevis is among the most sensitive amphibian species and serves as a good sentinel species in amphibian risk assessments. However, it cannot be denied that some amphibian species are more sensitive to certain contaminants. They can be taken into account with sufficient safety factors (see Comparison between standard test organisms and amphibians).

Interactions with abiotic and biotic stressors

Costressors mainly enhance the impact of glyphosate-based herbicides; therefore, glyphosate-based herbicide concentrations that have no effects when applied singly in laboratory studies may become the opposite in the wild where costressors such as predators are commonly present [125]. However, most of the studies with costressors used relatively high concentrations of glyphosate-based herbicide. It is interesting that none of 4 studies found adverse interactions between glyphosate-based herbicide exposure and chytrid infection.

Comparison between standard test organisms and amphibians

At this point, it seems most important to compare the responses of standard test organisms used in risk assessment with those of amphibians. The main question is if amphibians are more sensitive to glyphosate and glyphosate-based herbicides than standard test organisms. With regard to acute toxic effects, LC50 values can be compared, but chronic effect testing of glyphosate and glyphosate-based herbicide exposure on amphibians mainly considered endpoints such as time to metamorphosis or size at metamorphosis. If no-observed-effect concentrations for amphibians were available (e.g., http://cfpub.epa.gov/ecotox/), results were calculated from differing study designs such as mesocosm experiments with costressors. Hence, a comparison of chronic effects even among amphibians is not possible yet; in terms of the comparison between amphibians and standard test organisms, we are restricted to acute toxicity.

Laboratory studies on official aquatic test organisms have revealed that glyphosate isopropylamine salt was practically nontoxic, technical-grade glyphosate acid was practically nontoxic to slightly toxic, and Roundup Original was moderately toxic to them [3]. Pesticide regulation practice in Europe and the United States considers safety factors of 10 to 100, depending on the studied species and endpoints (http://www.epa.gov/oppfead1/trac/science/fqpa_resp.pdf, http://ec.europa.eu/enterprise/epaa/3_events/3_3_workshops/full-report-tox-workshop-september-2012.pdf) and derived toxicity to exposure ratio values (meaning endpoint/exposition; http://www.oecd.org/chemicalsafety/agriculturalpesticidesandbiocides/1944146.pdf). With regard to their allowed environmental concentration (e.g., 5 µg a.e./L with 5 m buffer strip for Germany [60]), risk assessment of pesticides that is based on ecotoxicological endpoints gained from tests with larval teleost fish (which also have a permeable skin) and aquatic invertebrates will most likely cover acute toxic effects on amphibian larvae including differences in the sensitivity of amphibian species (one exception would be the North American Pseudacris regilla: 24-h LC50 = 0.23 mg a.e./L [69], but whether similarly sensitive species occur in Germany is unknown). These findings conform to the results of a study that compared several pesticide endpoints of amphibians with those of other aquatic organisms [126]. Even the presence of costressors will be covered in most cases; but, for example, the lowest LC50 value in a study on the effects of Roundup in combination with predatory stress was 0.41 mg a.e./L for wood frogs [114]. This value is just outside the 100-fold safety factor (5 µg acid equivalent/L → 4 µg acid equivalent/L).

With regard to terrestrial life stages of amphibians, in only 1 amphibian study [28] was glyphosate isopropylamine salt administered intraperitoneally. The lowest 96-h median lethal dose (LD50) value was 1170 mg a.e./L. The lowest 192-h LD50 value for the same substance to the bird species Colinus virginianus was >3851 mg a.e/kg body weight (http://cfpub.epa.gov/ecotox/), while the 120-h LD50 value of glyphosate acid to the same species was >4971 mg a.e./kg diet [3]. Likewise, a single dose of glyphosate acid to Rattus norvegicus resulted in a LD50 value of 4275 mg a.e./kg body weight [3]. Since all values are in the same order of magnitude, mammals and birds apparently serve as proxies for amphibians' risk with regard to oral administration of the active ingredient; however, studies on the toxicity of orally administered glyphosate-based herbicides to amphibians are missing.

However, as already mentioned, amphibians seem to be at higher risk due to dermal uptake that, in the case of glyphosate, occurs 26 times faster for amphibians than for mammals [30, 127]. Therefore, studies are required that test this exposure route, for example, by directly applying glyphosate-based herbicides or glyphosate on amphibians (or their skin [30]) or by incubating them in contaminated water. For this purpose, mammals and birds do not seem to suit, and we suggest including at least 2 amphibian species from different families in standardized test batteries to study dermal administration.

Contribution to population/species decline

The main question is if glyphosate and glyphosate-based herbicide use affects natural amphibian populations. Herbicide use is 1 among many partial causes of ongoing global amphibian declines [128]. However, its exact role remains difficult to assess as field data remain sparse [24-26] and abnormal population changes have been suggested to often result from multiple interacting causes [18, 19, 129]. Hence, no definitive answer can be given with regard to the past, current, or future contribution of glyphosate-based herbicides as well as other environmental contaminants. In the following, some aspects are discussed regarding whether glyphosate-based herbicide use may be a relevant regional contributor to amphibian decline.

With regard to temporality, the increasing use of glyphosate-based herbicides cannot be made responsible for most amphibian population and species losses during the last decades. The majority of declines in developed countries occurred in the 1960s to 1970s [130], partly due to the intensification of agriculture in general. Although glyphosate-based herbicides are currently distributed in these countries, glyphosate marketing only started in 1974 [3]. A new wave of dramatic declines and extinctions of amphibians has been witnessed over the last 3 decades [18], but these have largely taken place in pristine and remote areas in the tropics. Glyphosate-based herbicide use as a reason can be ruled out here because other potential causes have been identified (e.g., habitat destruction and emerging infectious diseases [19, 131]). Thus, there is no temporal evidence of any association between glyphosate-based herbicide use and observed amphibian declines.

With regard to the concentration–response relation, most laboratory and mesocosm studies state effects in a dose–response manner (see Results). There is consistency in studies that reported that adverse effects of some glyphosate-based herbicides (especially those with tallowamine surfactants like polyethoxylated alkylamine) on amphibians at concentrations that can occur in the environment (see Table 1), but also that other less harmful glyphosate-based herbicides (e.g., which are labeled for aquatic use) will probably not affect individuals in normal-use scenarios. The formulation used and the site-specific application practice make the difference.

It is important to consider each life stage of amphibians separately in risk assessment. For adults, it may have fatal consequences when applications of certain glyphosate-based herbicides coincide with migration activities, for example, when the main reproductive part of a population crosses fields during seasonal migrations to breeding sites and would be directly oversprayed [20], but this holds true for other pesticides as well [127]. That can also happen during the daytime when pesticides are applied (as is the case in explosive breeding European amphibians [132]). Such mass mortalities of adults due to tilling have already been observed and studied in the field [133], and one can postulate that herbicide applications, which replace mechanical weed management in no-tillage farming, can have similar effects. Nondirected migration of newly metamorphosed juveniles that emerged from water bodies near agricultural land has also been observed, leading them onto fields [133] (threat of direct overspraying, e.g., in winter grain cultivation). However, it will not matter for an amphibian population that persists in an agrarian landscape if migrating individuals suffer mechanical or chemical death. Most studies used tadpoles, and some described adverse (mainly chronic) effects at environmentally relevant concentrations. Additional effects of costressors in the natural environment can enhance the impacts of glyphosate-based herbicide concentrations. It is important to note that the majority of anuran species follow a reproductive strategy of overproduction, enduring high mortality rates of their offspring [134], and that, therefore, (subjectively) high mortality or failure rates in larvae may not substantially affect population viability [135, 136]. However, early metamorphosis is accompanied by limited body condition (size, mass), resulting in reduced individual fitness (e.g., decreased overwinter survival and reproductive potential and prolonged time to first reproduction) [53, 137-140]. Delayed metamorphosis lengthens the time larvae are vulnerable to aquatic predators and may increase mortality rates in species using ephemeral ponds (common in anurans), as these habitats are more likely to dry out toward the summer or in dry periods [128]. Effects of both precipitated and delayed metamorphosis on population viability are poorly understood [133].

Many of the above-named potential threats to different amphibian life stages do not apply exclusively to glyphosate-based herbicide but also to most other pesticides. However, unique to glyphosate-based herbicides—and other nonselective herbicides—is their application during the whole growing season. Glyphosate-based herbicides can be used from the time before sowing in no-tillage farming until harvest (desiccation of several conventional crops [3]). Furthermore, glyphosate-based herbicides are applied later in herbicide-resistant systems compared to conventional cultivations because they allow elimination of weeds postemergence of the herbicide-resistant crops [141]. Hence, some amphibian species may be multiply affected by glyphosate-based herbicide applications during the year. Although comprehensive official data are missing, it can be inferred that genetically modified soybeans have driven agricultural expansion and land-use change in Argentina and other South American countries [142], meaning that the complementary glyphosate-based herbicide was applied in former forestland and other natural habitats. With the advent and adoption of herbicide-resistant crops that additionally tolerate unfavorable conditions (e.g., soil pH and drought), glyphosate-based herbicides might extend even further into natural habitats.

CONCLUSION

To study the contribution of glyphosate-based herbicide (and any other pesticide) used to observed amphibian population declines, a species- and site-specific evaluation has to be conducted. We see the need to include population viability analyses into amphibian risk assessment. These have been successfully used with regard to similar questions, for instance, impacts of elevated ultraviolet-B radiation on amphibian populations [143, 144]. Therefore, field data on both amphibian populations in the agrarian landscape and contamination with glyphosate-based herbicides and other pesticides are needed. While this involves cost and time-intensive fieldwork, it cannot always guarantee a general answer, as demonstrated in the case of the European tree frog (Hyla arborea). This species is disappearing from many agrarian landscapes where it was previously common, although terrestrial and aquatic habitats are still present [145]. Tree frogs are the anuran family most sensitive to glyphosate-based herbicides, but their disappearance cannot be causatively linked to increasing glyphosate-based herbicide or other pesticide use. The same holds for the observed disappearance of other sensitive species in agrarian landscapes. While glyphosate-based herbicide use is increasing, farming has been intensified in general and tree frog populations are affected by a multitude of other factors (such as landscape fragmentation) [145]. Another case is the Argentinean anuran fauna of agricultural sites with intensive cultivation of herbicide-resistant soybeans. Individuals show significant enzymatic alterations and reduced body size [146], but it remains unclear how pesticides are applied locally and to what proportion glyphosate-based herbicides or other pesticides (e.g., insecticides) can be causatively linked to poor body conditions.

In general, more fieldwork and modeling are required on how failure rates of different amphibian life stages contribute to the survival of populations in the agrarian landscape. Without such work, amphibian population declines cannot causatively be linked to glyphosate-based herbicide or other pesticide use. However, this should not distract from the possibility that glyphosate-based herbicide overapplications or missing buffer strips may indeed contribute to current or future damage to local amphibian populations. Illegal overapplications of pesticides in general and ignorance of buffer strips are well known, for instance, in Germany (http://www.umweltdaten.de/publikationen/fpdf-l/3566.pdf). Stricter controls for farmers may be recommended.

As is the case in the European Union, animal species listed in Annex IV(a) of the Habitats Directive 92/43/EEC are strictly protected within their natural range and any deterioration or destruction of their breeding sites or resting places is prohibited by Article 12(1)(d). Many amphibians that inhabit agricultural landscapes are listed in Annex IV(a). Only this requires further monitoring action of both glyphosate-based herbicide contamination and amphibian populations in agrarian landscapes. Amphibian monitoring, as proposed for herbicide-resistant crop cultivation in some European countries [147], could help in recognizing population changes related to glyphosate-based herbicides.

We see the need to expand future research in 3 main areas: 1) filling basic knowledge gaps through studies with comparable design and modeling, addressing especially chronic and delayed effects of glyphosate-based herbicides and added substances; 2) long-term monitoring of glyphosate-based herbicides in the environment and of free-living amphibians, aiming at the detection of effects on individuals and especially abnormal local population changes; and 3) an ongoing analysis of information and risk assessment (especially of new substances to come and costressors) in the context of worldwide amphibian decline and extinction. Research in these areas is well in line with the recommendations formulated under the International Union for Conservation of Nature's “Amphibian Conservation Action Plan” [128], which in general points toward further research on environmental contaminants.

SUPPLEMENTAL DATA

Table S1. (42 KB XLS).

Acknowledgment

This review was written within the framework of an expert opinion on the possible correlation of the increasing use of glyphosate and the worldwide amphibian decline, funded by the Federal Agency for Nature Conservation (Bundesamt für Naturschutz), Bonn, Germany. For support of the first author, we are also indebted to the Graduiertenkolleg of Trier University, funded by the German Science Foundation (Deutsche Forschungsgemeinschaft). We thank M.D. Boone and 3 anonymous reviewers for useful comments on the manuscript and K. Pond for help with the English language.

Ancillary