Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

At the time, I had never heard of atrazine. Now, you google “tyrone” and “atrazine” and you get tens of thousands of hits. I guess you can say we are joined at the ­carbon bond.

Tyrone B. Hayes

Introduction

Atrazine is a triazine herbicide. Up until 1999, it was the number one selling ­pesticide in the world. Now it has been displaced by another broad spectrum ­herbicide, glyphosate (Roundup) (Relyea 2010). Atrazine was initially introduced by CIBA in 1958. Eventually, through various mergers and acquisitions, NOVARTIS and then SYNGENTA took over as the major manufacturer. Over the last couple of decades, approximately 80 million pounds of atrazine per year have been applied (primarily to corn) in the USA. Atrazine is a water soluble, persistent, and highly mobile herbicide. As much as 500,000 pounds of atrazine are transported in rainfall and 1.2 million pounds collect in rivers and flow into the Gulf of Mexico each year. Atrazine kills weeds by disrupting electron transport in photosynthesis, but monocot crops, such as corn, have the capacity to remove atrazine via the glutathione reaction and excretion (Solomon et al. 1996). Thus, atrazine can be applied to corn fields and kills weeds but not corn, which is naturally resistant. This natural ­resistance and the resulting ability to apply atrazine without harming the crop is what made atrazine the big seller (number one in fact), until Round up-Ready corn came along. It was about that time (1997) that I was introduced to atrazine, when I was a new assistant professor at the University of California Berkeley.

History and Background

I’m still just a little boy who likes frogs.

Tyrone B. Hayes

As long as I can remember, I have been fascinated by amphibians. As an undergraduate at Harvard, I explored the effects of temperature on growth and development rates and sexual differentiation in woodfrogs (Rana sylvatica). As I developed my undergraduate honor’s thesis, I became increasingly interested in endocrine mechanisms that regulate growth and development and how environmental changes regulate endocrine signals. How are environmental factors (changes in the external world) translated, through hormones, into changes in physiology, development and growth? These were the types of questions that I pursued in my doctoral work, examining interactions between temperature and hormone production and function during growth, metamorphosis, and sex differentiation in amphibian larvae.

Fulfilling another childhood dream, I began travelling to Africa during my ­doctoral studies. While working in Kenya, in the Arabuko Sokoke Forest, I was introduced to the species Hyperolius argus. The species caught my eye and drew my interest as a result of its unusual sexual dimorphism. The males and females were so differently colored, that initially I thought they were different species. As adults, the males are a bright lime-green color, whereas the females are reddish brown with large white spots. I brought the species back to Berkeley and developed a breeding colony. In our first studies, we followed the development and examined hormonal regulation of pigmentation (Hayes and Menendez 1999). Our studies revealed that the green coloration was the default pattern. Both males and females metamorphosed with the green (male-type) coloration and females changed color at sexual maturity. In addition, we examined hormonal regulation of color change and found that only estrogens induce the female color pattern. Exogenous estrogen not only prematurely induced color change in juvenile females, but induced the female color pattern in males as well. So males H. argus have the receptor to respond to estrogen and the ability to synthesize the pigments to produce the female color pattern, but in the absence of ovaries do not synthesis the estrogens necessary to initiate color change.

My laboratory screened dozens of steroids and other estrogenic chemicals, and as it turns out, the list of estrogens that induced color change in H. argus, correlated directly with estrogens that induce breast cancer cell growth (Hayes 2000b; Noriega and Hayes 2000). While giving a lecture on our findings, my wife, Katherine Kim, happened to be in the audience once. Having recently completed her MBA and MPH, she saw an opportunity. “You should patent that,” she said. That simple comment led to my relationship with NOVARTIS and my introduction to atrazine (Hayes 2000b).

From the Wild to the Regulatory World

SYNGENTA means bringing people together.

Syngenta

At first, I served as a consultant only. Along with Dr. Tim Gross (at the time a researcher at the University of Florida at Gainesville and a USGS employee) and Dr. James Carr (a professor at Texas Tech University) and a number of other scientists hired as consultants and NOVARTIS employees, we were tasked with evaluating whether atrazine was an endocrine disruptor in fish, amphibians, and reptiles. The panel met several times per year and was a joint effort between NOVARTIS and ECORISK Inc. ECORISK was a consulting firm run by Dr. Ron Kendall (then at Clemson University) and the CEO Robert Bruce, who occasionally joined the panel meetings. As a developmental endocrinologist (not an ecotoxicologist), I found myself learning a whole new vocabulary. Whereas I was used to talking about ­hormone concentrations measured in ng/ml and pg/ml, I now had to translate ­chemical concentrations into ppb and ppt.

After completing a report that amounted to a large multi-authored review paper, I assumed that I had served my duty to ECORISK and NOVARTIS and moved on. A few weeks later, however, I was asked to evaluate a proposal written by Jim Carr to examine atrazine effects in frogs using the African clawed frog (Xenopus laevis). I reviewed the proposal, offered my criticisms, my praises, and made suggestions for improvement. Eventually, I was asked by Alan Hosmer, representing NOVARTIS, to describe how I would approach this question. I responded and was offered a ­contract. I submitted several proposals that utilized a diversity of species (including my patent pending Hyperolius argus Endocrine Screen (Hayes 2000a)), but the company decided on Xenopus laevis because it was a well-known model.

Given the lack of data at the time, there was no real hypothesis, only a question, “Does atrazine affect thyroid hormone, androgen, or estrogen-dependent development in larval amphibians?” I designed a study that exposed larvae throughout development (from hatching to metamorphosis) with several atrazine concentrations ranging from 0.1 to 200 ppb in a static renewal system. Using positive controls that included thyroid hormone, estradiol, and dihydrotestosterone (DHT), we proposed to monitor growth, development, and metamorphosis as measures of thyroid hormone interference or activity by atrazine; gonadal development as a measure of estrogen-like activity; and laryngeal growth and size as a measure of androgen (DHT) activity or interference. In the absence of any preliminary or background data on which to base predictions, the best hypothesis was, “atrazine does not affect endocrine-regulated end points in amphibians.”

The first set of data suggested some effects on the larynx. We analyzed a few samples only, but even this small sample size revealed that atrazine-exposed ­animals had smaller larynges than controls. We increased the sample size slightly, and the problem got worse approaching statistical significance.

The ECORISK panel requested a few samples at a time, but after a couple of meetings and conference calls, I completed the work. I suggested to the panel that we repeat the experiment with a different population and that we consult Dr. Darcy Kelley (Columbia University), the world’s expert on development and hormonal regulation of the larynx in Xenopus laevis. I requested to present the data at the Western Regional Conference on Comparative Endocrinology. Although, I was told I could present the data, corporate review took so long (more than a month) that I missed the abstract deadline. I was told that I could repeat the experiment. The ­proposed experiment was exactly the same as the first, but compared frogs from two separate sources: our lab colony of mixed origin and animals from NASCO Inc. (Fort Atkinson, Wisconsin). A second part of the study proposed to examine the ultimate effects on reproductive potential after metamorphosis. By the time the committee reviewed the proposal, however, November was approaching. ECORISK and SYNGENTA did not provide a review, and eventually suggested that we should postpone starting the project to avoid it extending beyond the Christmas holidays. No funds came forward, but on Oct 28, 2000, I initiated the experiment without funding from SYNGENTA or ECORISK.

It was during this experiment, that I observed that several of the exposed animals had gonads with unusual morphologies. Histological analysis revealed that some animals from the atrazine treatments had a mix of testicular and ovarian ­tissues. We then re-examined all of the animals in more detail (including animals from the ­previous experiment) and discovered that three types of gonadal ­morphologies were observed only in the atrazine-exposed animals: multiple (or lobed) testes, nonpigmented ovaries, and hermaphrodites (animals with both testes and ovaries). We would later go on to show that these abnormalities could be induced by an androgen blocker (cyproterone acetate) and/or by brief periods of estrogen exposure, suggesting that some morphologies represented demasculinization and that others represented ­partial (incomplete) feminization (Hayes et al. 2006b).

On January 26, 2001, representatives of Ecorisk and Syngenta came to Berkley and I arranged everything as if it were a local conference. I organized the hotels, local ground transportation, seminar rooms, and arranged all of the meals and refreshments. I developed an agenda. I would present a summary of the consultation and research conducted under the direction of the ECORISK panel, SYNGENTA would present on their additional needs, and then I would present my laboratory’s new data. In addition to the students from my laboratory and the staff who worked on the project, I invited Beth Burnside (then, Vice Chancellor for Research), Paul Licht (then, Dean), Marvalee Wake (then, Chair of my department), David Wake (then, Director of the Museum of Vertebrate Zoology, where I had an appointment), and the University’s legal counsel (Michael Smith). Beth Burnside opted not to attend and Smith determined that he did not feel he needed to be there. Lastly, my wife, Katherine Kim, was also in attendance.

They arrived: The ECORISK panel (Ron Kendall, Keith Solomon, John Giesy, Earnest Smith, and Jim Carr) along with Alan Hosmer (SYNGENTA), with Charles Breckenridge arriving later that afternoon. They immediately requested to meet with me in private. I refused. I presented our introduction and they again asked for a private meeting. I refused. Then they introduced Robert Sielken (Sielken and Associates Consulting Inc.) who attended with them.

In his presentation, Sielken presented a critique of our previous experiments. He pointed to corrections made in the hand-written data, some as trivial as recording the time in standard instead of military time, but, in fact, corrected. In his opinion, our data on the larynx was of no significance. He had gone back to our initial data sets. After we realized that the data showed a decrease in the size of the male larynx, we focused on the males because there was no effect (in this regard) on females. Thus, we did not analyze any more than the first three estrogen-treated frogs because all estrogen-treated animals were female. Sielken felt that because the smallest ­sample size was three, then all of the data had to be simulated, using a Monte Carlo simulation, to approximate a sample size of three. Even after this alternate approach to analysis of the data, Sielken obtained a P value of 0.045. Sielken and his firm only accepted P values of <0.04 as significant (Parshley 2000).

Shortly after the meeting at Berkeley, ECORISK would drop out of the picture. Alan Hosmer joined by Peter Hertl and Janis McFarland at SYNGENTA sought to draw up a contract directly between myself and SYNGENTA. Hosmer worked with SYNGENTA Vice President, Gary Dickson. I would stay in touch with John Giesy for a while, who assured me that Gary Dickson would be helpful.

We never worked out a new contract, however. Just 1 day before I was to speak with SYNGENTA, National Geographic News, to my surprise, aired footage that they shot in my lab showing the hermaphrodites produced by atrazine. That was the end of my relationship with SYNGENTA. My laboratory subsequently prepared the manuscript. Although I had no idea it would cause the stir it did, I knew the data deserved to be in a widely read journal. I submitted it to the Proceedings of the National Academy of Sciences (PNAS).

I suggest you get yourself a working phone.

David Wake, after our first paper was accepted into PNAS

Even before our first paper came out in PNAS, I received phone calls from both the Natural Resources Defense Council and Tom Steeger from the US Environmental Protection Agency. Tom Steeger told me that ECORISK, assuming that I had reported data to the EPA, were submitting a critique of my work (Parshley 2000).

Tom Steeger forwarded SYNGENTA’s data and their unpublished manuscripts. He was similarly forwarding my unpublished data, submitted manuscripts, etc., to others even though he told me he was holding them in confidence and using them only in his internal review. The Natural Resources Defense Council (NRDC) would later obtain, through the freedom of information act, email exchanges between Steeger and Hosmer during this time (2000–2003), discussing my data, information that I gave Steeger during a visit to my laboratory and other confidential issues. The noise that SYNGENTA and ECORISK made would stimulate a debate, however, that would culminate in the 2003 EPA/FIFRA Scientific Advisory Panel meeting. With the exception of Carr’s paper, none of the SYNGENTA data were published at the time, but through Steeger, I had seen most of it.

When SYNGENTA’s first paper finally appeared (Carr et al. 2003), it was indeed everything I had seen previously from communications with Steeger. Although they would later claim that they “could not” reproduce my work, the fact is they “did not” reproduce my work. With the exception that we both used static renewal, no other efforts to replicate my study were made: the rearing medium, the food type and amount, the container size, the volume of rearing medium, the number of ­animals per tank, the amount of atrazine added, the temperature, the animal source, etc., virtually everything was different. Further, the health of their animals under these conditions was questionable. Greater than 50% of the animals in their study had suffered mortality (compared to 10% on average in my laboratory). The animals that survived showed an inverse relationship between time to metamorphosis and size at metamorphosis, an indication that they were in dire health. Atrazine not only affected gonadal differentiation (P  =  0.0003 for intersex and P  =  0.0042 for discontinuous or lobed testes), but the effects they reported were of greater statistical ­significance than in our studies. Furthermore, the metamorphs in Carr’s study were smaller than any I had ever seen, indicating their poor health. Carr also reported several other significant adverse effects on growth and development: Atrazine reduced the number of animals reaching metamorphosis (P  =  0.03 for foreleg emergence and P  =  0.04 for tail reabsorption). Atrazine also induced edema (P  =  0.02) and abnormal erratic swimming (P  =  0.004). Despite these very significant effects, however, they repeatedly referred to their results as “weak trends” even though the effects were well below the significance value of (P  <  0.04).

Carr’s work was followed by a study conducted in Giesy’s laboratory in Michigan State (Coady et al. 2005). Here, the authors suggested that hermaphroditism was normal in X. laevis. That conclusion was based on their finding that hermaphrodites were also observed in their “controls.” They conducted their experiments at Michigan State using uncovered tanks. In their report to the EPA (Hecker et al. 2003) and in their publication, they reported that frogs jumped from tank to tank (frogs would appear in previously empty tanks, etc.) and in some cases, even disappeared from the experiment (escaped from the tanks). In addition, due either to the aspiration of atrazine-contaminated water, or atrazine contamination in their water source, their “controls” had more than two times the biologically effective concentration of atrazine based on our studies (Hayes et al. 2002a) thus explaining why they found abnormalities in their “controls.”

Carr’s laboratory would go on to show that atrazine and especially atrazine  +   nitrates resulted in testicular oocytes in leopard frogs, Rana pipiens (Orton et al. 2006). Their finding mirrored our studies that showed testicular oocytes and even vitellogenic eggs produced in male gonads in Rana pipiens larvae exposed to atrazine (Hayes et al. 2002b, c).

SYNGENTA and ECORISK also followed with a series of field studies. In one of their studies in the USA, they showed that newly metamorphosed frogs collected from areas with atrazine contamination were more likely to have testicular oocytes (P  <  0.04) (Murphy et al. 2006), yet they concluded “given the lack of a consistent relationship between atrazine concentrations and testicular oocytes (TO) incidence, it is more likely the TOs observed in this study result from natural processes in development rather than atrazine exposure” (Murphy et al. 2006). In South Africa, they conducted another study, where they reported that (again) gonadal abnormalities were found, but were equally likely to occur in frogs from “non-corn-growing” areas as they were to occur in frogs collected from “corn-growing areas.” (Du Preez et al. 2005b). In the study in South Africa, they examined adult frogs and claimed that frogs were confined to the areas where they were captured and were not likely to have moved from elsewhere (e.g, from a pond in an agricultural area to a pond in a nonagricultural area). However, they reported losing all of their animals in one study, because catfish moved into the pond and ate all of the frogs. As pointed out in my review in Biosciences, “Thus, the authors’ claims rely on the unlikely assumption that fish (which are truly aquatic and have no lungs or legs) are capable of moving between ponds, but frogs (which have lungs and legs) are not”(Hayes 2004). What’s more, the atrazine levels reported from the ponds in their “non-corn-growing” area were as high as those reported from their corn-growing regions, and both exceeded ­biologically active levels.

The ECORISK group also conducted a microcosm study. They reported gonadal abnormalities in the controls (Jooste et al. 2005b). Specifically, they reported ­testicular oocytes in newly metamorphosed Xenopus laevis. This is an interesting finding, because my laboratory has never observed testicular oocytes in Xenopus laevis even in atrazine-exposed or estrogen-exposed frogs at metamorphosis. Females do not normally have oocytes at metamorphosis in Xenopus laevis in my laboratory’s experience. The figure, eventually published by the ECORISK group (Jooste et al. 2005a), in response to a letter to the editor from me (Hayes 2005), did not contain a scale bar, but the size of the gonad in the photomicrograph can be estimated based on the average oocyte diameter for the species. It is not likely that a gonad of this size with oocytes as advanced as those in the photomicrograph could have come from a newly metamorphosed animal. Considering that the population that they examined simply might develop differently from others, my laboratory travelled to Potchefstrum in 2006 and brought our own animals into the laboratory and established a breeding colony. Like other populations we worked with, even females did not have oocytes at metamorphosis from that population. The only possibilities were a contaminant in their rearing water, or that the unusual gonadal development was an artifact of the near lethal temperatures (Jooste et al. 2005b) used in their studies.

To date, 100% of the studies conducted by SYNGENTA claimed no affect of atrazine (Coady et al. 2005; Du Preez et al. 2002, 2005a, 2005b, 2008; Hecker et al. 2004, 2005; Jooste et al. 2005b; Kloas et al. 2009a, b, c; Preez et al. 2009), even SYNGENTA studies that showed effects (Carr et al. 2003; Murphy et al. 2006). On the other hand, studies not funded by Syngenta report significant effects of atrazine on amphibian reproductive development (Hayes et al. 2002a, b; 2006b; Langlois et al. 2009; McDaniel et al. 2008; Oka et al. 2008; Orton et al. 2006; Reeder et al. 1998; Tavera-Mendoza et al. 2002a; Tavera-Mendoza et al. 2002b). But, once again, SYNGENTA took the opportunity to say that they were unable to repeat our studies, and again, focused on critiquing my studies and ignoring the fact that several other independent laboratories have shown adverse effects of atrazine on gonadal development in frogs, in addition to my laboratory: Solomon (Solomon et al. 2008) ­concluded in his review that “with rare exceptions, the only studies that report adverse effects on amphibian development and reproduction are those from the Hayes laboratory.” That view contradicts his long-term coauthor, Carr’s statement, “The important issue is for everyone involved to come to grips with (and stop ­minimizing) the fact that independent laboratories have demonstrated an effect of atrazine on gonadal differentiation in frogs. There is no denying this” (Hayes 2004; 2009a). Furthermore, as Rohr and McKoy pointed out (Rohr and Mckoy 2009), the review by Solomon et al. “regularly dismissed significant effects of atrazine.”

Eventually, SYNGENTA published what they reported as “the definitive” ­atrazine studies in Xenopus laevis (Kloas et al. 2009a). Along with the EPA, SYNGENTA oversaw two atrazine studies in Xenopus laevis. One of the studies contracted by SYNGENTA was conducted in the laboratory of Werner Kloas, a member of the EPA’s 2003 FIFRA Scientific Advisory Panel (SAP). An SAP member (Kloas), who affected EPA decisions, benefitted financially from the EPA’s decision (at the recommendation of the SAP) that “there is not sufficient scientific evidence to indicate that atrazine consistently produces effects across the range of amphibian species examined” (Kloas et al. 2009a). This scenario is analogous to a jury coming back with a decision that there is not enough evidence to convict, with one of the jurors then receiving a lucrative contract to investigate further and bring the case back to court.

Kloas et al. concluded that “it seems likely that no further endocrine mechanism of atrazine affecting sexual differentiation in X. laevis exists” (Kloas et al. 2009b). As in the past, it was not that SYNGENTA was unable to repeat our studies, but rather that they did not repeat our studies. Not only were many variables different in their studies (Kloas et al. 2009a, b, c), including constant flow-through vs. static exposure, different food and different feeding patterns, different sized containers, and different animal densities, but in addition both of the new SYNGENTA studies used a new and unexamined population of frogs from a company called Xenopus 1 (Dexter, Michigan, 48130) that reports that their adults come from Chile.

It was, in fact, this single Syngenta-funded study that formed the basis for the EPA’s second Scientific Advisory Panel review of atrazine. Among others, the New York Attorney General’s Office raised concern prior to the EPA’s review: “EPA’s approach of evaluating the effects of atrazine only on frog gonads, and relying almost entirely on only one unpublished industry-sponsored study (Hosmer et al. 2007) is clearly not based on the available science and may lead to a biased outcome.” (http://www.regulations.gov/search/Regs/home.html#documentDetail?R=09000064802ff8e0), The EPA not only ignored this advice, but even ignored the concerns of their advisory panel. Similar to the concern of the New York State Attorney General’s office, the panel expressed “concern that the Agency did not utilize any of these studies” (from the open literature) and “noted that these studies may have provided some added value in evaluating the conclusions drawn from the data provided in response to the DCI.” The panel noted several concerns with the DCI (Data call in study, provided by Syngenta) including the use of a flow-through system, the appropriateness of the animal model, and concern that the population used in the study was a resistant strain: “The strain used in the DCI studies was apparently an insensitive strain. Panel members were concerned that this apparent insensitivity may have resulted in insensitivity of the apical endpoints to atrazine in general.” The panel recommended that additional statistical analyses be applied, that histopathalogical results be evaluated by an independent laboratory, that metabolites of atrazine be examined, and that “studies with X. laevis be followed up with comparable studies using a North American species as soon as possible.” Further, “Some panel members expressed concern that EPA completely rejected its own hypothesis based solely on the ­negative results of the DCI study” and concluded that “From the scientific perspective, the Panel agreed that the relevance of the uncertainties justifies the generation of ­additional data.” (http://www.epa.gov/oscpmont/sap/meetings/2007/100907_mtg.htm#transcripts). Despite these concerns, the EPA concluded, “At this time, EPA believes that no additional testing is warranted to address this issue.” (http://www.epa.gov/pesticides/reregistration/atrazine/atrazine_update.htm)

Moving the Science Forward

For the EPA, Hayes’ work is interesting but irrelevant to any decision to regulate the pesticide.

Stephen Bradbury, US EPA; Oakland Tribune, Dec 3, 2006

Rather than focus my immediate attention on EPA recommendations and answering SYNGENTA’s challenges at the time, I focused my research efforts on the ­remaining important scientific questions. In my mind, the issue of atrazine’s effects on gonadal development was no longer an open question, even if there were population ­variation in the response or with variability in environmental parameters or study conditions. Several important questions superseded this one, however. Are the hermaphrodites produced by atrazine exposure males with ovaries or females with testes? And what was the ultimate outcome? Do animals that display hermaphroditism at metamorphosis remain hermaphrodites throughout life or do they transform into one sex or the other at sexual maturity? Further, what are the long-term effects on reproduction in exposed animals? Though SYNGENTA previously funded a study where they claimed no long-term effects of atrazine on adult males, their studies did not examine morphology at all, did not examine competition between atrazine-exposed males and control males, or allow female choice. Finally, their animals were treated with hormones prior to examinations of fertility (Du Preez et al. 2008) effectively providing hormone replacement therapy for the exposed animals.

We hypothesized it was more likely that atrazine affected genetic (ZZ) males only. Xenopus laevis is a female heterogametic species. Females are ZW and males are ZZ, therefore females determine the sex of the offspring. Applying estrogen to the rearing water of larvae will produce 100% females (Chang and Witschi 1955; Gallien 1953; Hayes 1997a, b; 1998; Villapando and Merchant-Larios 1990), but adding androgens, or any other steroid for that matter, will not alter the sex ratio (Hayes 1998). In other words, the W chromosome is dominant and environmental factors or exogenous hormones do not override the genetics of being female. Genetic males (ZZ), however, can be manipulated. In addition, it has now been revealed that the W chromosome in females carries at least one unique gene, DMW (Okada et al. 2009; Yoshimoto et al. 2008). As it turns out DMW is a transcription factor that induces aromatase expression, which leads to estrogen production and subsequent ovarian differentiation (Yoshimoto et al. 2008). Atrazine induces aromatase in fish (Suzawa and Ingraham 2008), reptiles (Crain et al. 1997; Keller and McClellan-Green 2004), and mammals (Fan et al. 2007a, b; Heneweer et al. 2004; Sanderson et al. 2000, 2001, 2002; Suzawa and Ingraham 2008) and results in increases in estrogen in fish (Spano et al. 2004), amphibians (Hayes et al. 2010b), and mammals (Stoker et al. 2000; Wetzel et al. 1994). Thus, a plausible mechanism for sex reversal of males by atrazine was available. Further, other studies suggested that atrazine induces complete feminization in not only Xenopus laevis (Oka et al. 2008) but also in Rana pipiens (McDaniel et al. 2008) and in zebra fish (Danio rerio) (Suzawa and Ingraham 2008). These studies showed a dose-dependent decrease in males, shifting the sex ratio in favor of females, but it was not known if the shift in the sex ratio was truly the result of sex reversal of genetic males.

To answer these questions, I took advantage of a population of X. laevis generated in my laboratory 18 years ago. A feral population of X. laevis originally collected in San Diego was treated with estradiol, resulting in all females. Sex-reversed genetic males (ZZ-females) were identified by crossing each animal back to unexposed ZZ males from the same population. Females that produced only male ­offspring were isolated and had been maintained in my laboratory since 1992. By crossing ZZ-females with ZZ males, we produced a population that contained only genetic male (ZZ) larvae. After exposing these larvae to atrazine, any hermaphrodites that were produced would have to be genetic males with ovaries and any females produced would have to be truly sex-reversed males (ZZ females).

This study revealed that atrazine indeed completely sex-reversed genetic males (Hayes et al. 2010b). At sexual maturity, 10% of the animals had protruding cloacae typical of females. Dissection or laparatomy revealed that these animals also had fully developed ovaries and vitellogenic eggs. These animals expressed aromatase and produced estrogen. These neo-females were capable of copulating with males and producing viable eggs. The majority of the remaining males treated with ­atrazine, were also demasculinized. These males suffered from suppressed testosterone levels and their androgen-dependent breeding glands were reduced, and sperm production suppressed. Further, atrazine-exposed males were unable to compete with control males for females and had severely reduced fertility when paired with females in the absence of control males (Hayes et al. 2010c).

As to what these laboratory studies tell us about effects in the real world it is difficult to say. Again our work (Hayes et al. 2002b, c) showed a strong correlation between atrazine contamination and testicular oocytes in the field, as did Reeder et al.’s study (1998). Similarly, McKoy et al. found demasculinized and feminized frogs in areas where atrazine was used in Florida (Mckoy et al. 2002, 2008) and a mesocosm study found that atrazine exposure resulted in sex ratios skewed toward females (McDaniel et al. 2008). Though they state differently, SYNGENTA’s data showed an association between atrazine contamination and frogs with feminized gonads also (Murphy et al. 2006). Alone, these findings would still only represent correlations and could not establish a cause–effect relationship; however, together with controlled laboratory studies consistent with these field studies, the case for atrazine as a causative agent in the wild is strengthened.

…use of atrazine according to the label instructions will not likely result in harm to human health or the environment.

Office of Pesticide Programs, US EPA (2008)

In addition to the effects on reproductive development and function in frogs, atrazine has a number of other important effects that could impact wild amphibians and contribute to population declines. The numerous nonreproductive effects of atrazine are best summarized in a qualitative meta-analysis published recently by Rohr and McKoy (2009). Rohr and McKoy reported that atrazine reduced size at metamorphosis in 15/17 studies and in 14/14 amphibians examined. They reported that atrazine increased activity in fish and amphibians in 12/13 studies and decreased predator avoidance behaviors or defense in 6/7. Such behavioral effects can have dramatic effects on amphibian survival. For example, Rohr et al. (2004) reported that atrazine increased activity (and thus energy expenditure), reduced shelter use, decreased the larval period, and reduced size at metamorphosis in exposed salamanders. Interactions with food limitations and drying conditions decreased the chances of survival for exposed individuals in this study, demonstrating the importance of examining atrazine exposure in combination with other stressors (Rohr et al. 2004). The work of Rohr et al. (2004), also demonstrated that the multiple effects of ­atrazine are quite complex and there are significant interactions when atrazine is applied to species communities (Rohr and Crumrine 2005). One must also consider the immediate and long-term (“carry over”) effects of atrazine. Rohr et al. (2006), showed that not only did atrazine exposure induce mortality in the streamside salamander Ambystoma barbouri, but atrazine had a carry-over effect 14 months later, reducing the ability of the surviving salamanders to recover (Rohr et al. 2006). In a later study, Rohr and Palmer showed that atrazine resulted in a greater risk of mortality from water loss in salamander up to 8 months after exposure (Rohr and Palmer 2005). All of these well-documented effects have the potential to contribute to amphibian declines, but in addition likely impact other exposed wildlife as well.

In particular, atrazine’s negative effects on immune function raise concern, given the focus on disease-driven mortality as a driving factor in amphibian decline. Rohr and McKoy report that atrazine decreased 33/43 immune function parameters and 13/16 infection endpoints. Atrazine exposure increases Rana virus infection in exposed salamanders (Ambystoma tigrinum) (Forson and Storfer 2006a, 2006b). Atrazine also increases trematode infections that result in limb deformities in exposed amphibians (Kiesecker 2002). In fact, Rohr et al. (2008), showed that atrazine was the best predictor (out of 240 factors examined) for trematode infections and that atrazine and phosphate accounted for most (74%) of the variation in trematode abundance. Further, atrazine, in combination with other pesticides is associated with decreased immune function in a number of studies in several ­species and including a wide range of disease pathogens (Bishop et al. 2010, Hayes et al. 2010a).

With regards to mechanisms, atrazine exposure alters gene expression of a number of genes, including genes involved in growth, development and immune function (Langerveld et al. 2009). It is also possible that atrazine (and other pesticides) decrease immune function by increasing stress hormone (corticosterone) levels. At least one study from our laboratory showed that a mixture of pesticides containing atrazine increases corticosterone levels (Hayes et al. 2006a). Studies in fish, show that atrazine increases the related glucocortioid, cortisol (Cericato et al. 2009; Nieves-Puigdoller et al. 2007).

There is no direct scientific information to assess this hypothesis.

Anne Lindsay, US EPA, before the Agriculture and Rural Development Committee of the Minnesota House of Representatives February 16, 2005

In addition to the multitude of effects of atrazine on amphibians, the number of taxa affected by atrazine also provides strong evidence for the impact of atrazine in the environment. Glen Fox wrote, “In ecoepidemiology, the occurrence of an association in more than one species and species population is very strong evidence for causation” (Fox 1991). Perhaps, then, the most compelling case against atrazine is the fact that similar effects are produced across not just species, or species populations, but across vertebrate classes.

The evidence for demasculinization is available from several studies. Atrazine decreases androgen production in fish and mammals, in addition to amphibians (references below). In addition, atrazine exposure causes a decrease in androgen-dependent development, morphology, and behavior in fish, reptiles, and mammals, similar to its demasculinizing effects in amphibians (references below). In salmon, atrazine caused a dose-dependent decrease in androgens, resulting in the absence of male reproductive behaviors (response to the female pheromone) and a decline in milt (semen) (Moore and Waring 1998). These effects occurred at exposure levels in the low ppb range and were similar to our recent reports in Xenopus laevis. Atrazine caused a decline in androgens and sperm production in caiman (Caiman latirostris) (Stoker et al. 2008). Atrazine-exposed caiman developed testes with reduced sperm content and histological appearance nearly identical to our observations in adult X. laevis (Hayes et al. 2010b). Similarly, atrazine demasculinizes laboratory rodents, consistent with findings in wildlife: atrazine causes a decline in testosterone in exposed rats, Rattus norvegicus, (Friedmann 2002; Stoker et al. 2000; Trentacoste et al. 2001). Atrazine also suppresses sperm production and ­fertility in rats (Friedmann 2002; Stoker et al. 2000; Trentacoste et al. 2001). EPA laboratories also showed that atrazine delayed puberty in male rats (Stoker et al. 2000), and in fact the EPA laboratories were among the first to conclude that ­atrazine is an endocrine disruptor, “ATR tested positive in the pubertal male screen that the Endocrine-Disrupter Screening and Testing Advisory Committee (EDSTAC) is considering as an optional screen for endocrine disrupters” (Stoker et al. 2000). Cross-generational effects have also been demonstrated in rodents: exposure of pregnant dams to atrazine reduced androgen levels in male pups. Male pups also had a decreased anal-genital distance and delayed preputial separation, both androgen-dependent aspects of development (Rosenberg et al. 2008)

In addition to atrazine’s feminizing effects reported in amphibians, similar effects have been reported in every vertebrate examined, including fish, reptiles, birds, and mammals. Atrazine caused an increase in estrogens in goldfish (Spano et al. 2004). Suzawa and Ingraham showed that atrazine upregulated gonadal aromatase in zebrafish and that atrazine exposure resulted in a dose-dependent increase in females in exposed fish (Suzawa and Ingraham 2008). The latter effect was similarly reported in X. laevis (Oka et al. 2008) and Rana pipiens (McDaniel et al. 2008), but my laboratory was the first to show directly that atrazine completely feminizes truly genetic males (Hayes et al. 2010b).

Aromatase induction and estrogen production have also been shown in amniotes. The induction of aromatase by atrazine was first proposed in a study on alligators (Alligator mississippiensis) (Crain et al. 1997), but the effect reported here was not statistically significant. Atrazine induced aromatase in gonadal cells from turtles as well (Keller and McClellan-Green 2004). Fewer studies have been conducted in birds, but at least one study reported sex reversal effects on the gonads of chickens, possibly through aromatase induction (Matsushita et al. 2006).

There are supporting findings for the induction of estrogen synthesis and subsequent estrogenic effects in laboratory rodents. In addition to declines in circulating androgens, atrazine exposure through food results in an increase in circulating estrogens in exposed rats (Stoker et al. 2000). In female Sprague–Dawley rats, ­atrazine exposure results in increased mammary tumors (Eldridge et al. 1994; Stevens et al. 1994; Ueda et al. 2005), studies funded by SYNGENTA argue that the higher tumor incidence is rather an earlier onset (Eldridge et al. 1994). Nevertheless, the tumors are estrogen responsive (Ueda et al. 2005) and thus, likely caused by excess estrogen. This effect of atrazine does not occur in Fischer rats. Eldridge et al. (1994), argues that Sprague–Dawley (SD) rats have higher rates of mammary tumor anyway, and thus are abnormally sensitive to atrazine. However, this two taxon statement, does not negate the effects in the SD rat.

In addition, laboratory rodents show a number of other reproductive effects when exposed to atrazine that do not necessarily occur through alterations in aromatase expression and/or activity. Perhaps most concerning are a number of studies that have examined effects of atrazine on pregnant dams. Atrazine induced abortion in four strains of rats in a series of studies conducted in EPA laboratories (Cummings et al. 2000; Narotsky et al. 2001). The authors suggest that the effect is due to disturbance in the brain, hypothalamus and pituitary rather than direct effects on the gonads (Cooper et al. 1999, 2000). This same laboratory showed that male pups exposed in utero and perhaps through the dam’s milk, developed prostatitis (Stoker et al. 1999). Yet another EPA laboratory showed cross-generational effects as well. Female pups exposed in utero suffered from severe inhibition of mammary growth (Rayner et al. 2004, 2005). Follow-up studies showed that even a second generation (not exposed directly) were affected (Rayner et al. 2004). At sexual maturity, the females exposed in utero lacked sufficient mammary growth to provide nutrients for their offspring. The second generation, though never exposed directly, displayed retarded growth and development as result of the effect on their mother’s who were exposed in utero.

Syngenta assumes no obligation to update forward-looking statements to reflect actual results.

Sherry Ford, SYNGENTA Crop Protection, August, 2009, “Syngenta Cautionary Statement Regarding Forward-Looking Statements”

Although SYNGENTA’s scientists disagree that atrazine is an endocrine disruptor, it is difficult to imagine how so many studies, conducted in laboratories around the world, in every vertebrate class examined, can coincidentally and wrongfully come to similar conclusions. Although the EPA has held two SAPs to evaluate the effects of atrazine on amphibians, their questions asked have been fairly narrow, focused on variability in the response of different amphibian population under varying environmental conditions and have never asked the fundamental question: “Is atrazine an endocrine disruptor in vertebrates?” This question would draw all of the data available from all vertebrates examined.

I have not stated the case for human health effects of atrazine related to its demasculinizing and feminizing effects here; however, data do exist. Atrazine detection in the urine of men in Missouri (≥0.1 ppb) was associated with decreased sperm count, semen quality, and low fertility (Swan et al. 2003). Men who work in agricultural fields applying atrazine can have levels up to 2,400 ppb atrazine in their urine (Lucas et al. 1993). In addition, men who worked in SYNGENTA’s (NOVARTIS at the time) atrazine production facility in San Gabriel, Louisiana (so called “Cancer Alley”), experienced an eightfold increase in prostate cancer (Maclennan et al. 2002; Sass 2003). Though SYNGENTA has tried to downplay these findings excluding cases from the analysis that were identified after the study and claiming that the incidence of prostate cancer only appear higher because of their careful screening of their employees (Sass 2003), several features of their report are important and worth quoting here. The authors reported: “The increase in all cancers combined seen in the overall study group was concentrated in the company employee group.” (page 1052); “The increase in prostate cancer in male subjects was concentrated in company employees” (page 1052); “The prostate cancer increase was further concentrated in actively working company employees” (page 1053); “all but one of the cases occurred in men with 10 or more years since hire” (page 1052); and “analyses restricted to company employees also found that the prostate cancer increase was limited to men under 60 years of age” (page 1053). The induction of prostatitis (Stoker et al. 1999) and prostate cancer (Pintér and al 1980) in laboratory rodents supports the findings in humans.

Atrazine also upregulates aromatase expression and estrogen production in ­several human cell lines (Fan et al. 2007a, b; Heneweer et al. 2004; Sanderson et al. 2000, 2001, 2002; Suzawa and Ingraham 2008). In fact, the strongest case for the induction of aromatase comes from these studies in human cells. The mechanism was initially proposed by Sanderson et al. (2000), who showed that atrazine inhibits phosphodiesterase (PDE). PDE regulates cAMP levels in the cytoplasm. By blocking PDE, the exposed cell experiences increased cAMP levels. cAMP regulates a number of phosphorylation events in the cell cytoplasm as well as binds to a receptor protein (cyclic-AMP response element binding protein: CREBP) that is translocated to the nucleus. After binding the promoter for the gene cyp19-aromatase, aromatase expression is enhanced. Interestingly, this regulatory pattern for aromatase is observed for aromatase expression in all vertebrate gonads and in fibroblasts associated with breast cancer. An increase in local estrogen production, could explain the increase in breast cancer associated with atrazine contamination of drinking water in at least one study (Kettles et al. 1997). Interestingly, NOVARTIS and ASTRAZENECA both market aromatase blockers as first-line treatments for breast cancer (Hayes 2009b).

Immune-suppressive effects of atrazine (discussed earlier) are not restricted to amphibians. In rodents, atrazine has a number of detrimental effects that have been more fully characterized than effects in amphibians. In mice, atrazine causes a decrease in white blood cells (Pruett et al. 2003). Atrazine increased neutrophils and T lymphocytes, and decreased lymphocytes, Natural Killer cells and B cells. Interestingly, atrazine also elevated corticosterone in this study (Pruett et al. 2003). In other studies, atrazine increased spleenic T cells, decreased cytotoxic T cell function and decreased mixed leukocyte responses. The result was a decrease in resistance to melanoma (Karrow et al. 2005). The thymus and spleen weight, splenic cell number, and macrophage function were reduced in mice exposed to atrazine (Karrow et al. 2005). Atrazine also inhibited lytic granule release, thus blocking Natural Killer cell function in mice (Rowe et al. 2006, 2007, 2008). Many of these effects are worsened when rodents are exposed during early development. For example, prenatal and lactational exposure reduced humoral and cell-mediated immune function in mice (Rowe et al. 2008). Depression of immune function was pronounced in females (Rowe et al. 2008), which is interesting because similar sex-specific effects on immune-suppression have been reported in frogs (Langerveld et al. 2009). Some of the effects may also involve the inhibition of dendritic cell maturation, via inhibition of several signaling molecules (Filipov et al. 2005).

He’s taking his information to people who don’t have enough independent information to make a truly independent decision.

Tim Pastoor, SYNGENTA, Minnesota Star Tribune 2005

The manufacturer’s position, even if we accept the proposed mechanism of endocrine disruption, is that the effective doses in mammalian studies are far beyond what humans would be exposed to. The data, or a standard way to translate data from wildlife studies to human exposures, do not exist.

Environmental Relevance?

The effective doses used in amphibian and fish studies are in the low ppb range in most cases (0.1 ppb in our studies). Those levels are the concentrations applied to and measured in the rearing medium. For reptile studies (Crain et al. 1997; Stoker et al. 2008) and for birds (Matsushita et al. 2006), the atrazine levels represent the concentration of solutions painted onto or injected into the eggs, although Keller and McClellan-Green used cell lines exposed to atrazine in the incubation medium (Keller and McClellan-Green 2004). For the rodent studies, the effective levels reported are in the ppm range, but those are the levels added to the food, dissolved in the drinking water, or in some cases delivered by gavage to the test animals. There is no indication in the rodent studies of how much food or how much water was consumed by the test animals, so we have no idea how much atrazine was ­consumed by the test animals, or (even if we did) how to compare consumption rates with absorption from the rearing medium in the case of cell lines or aquatic organisms, or across an egg shell in the case of reptiles and birds. Perhaps constantly absorbing low ppb amounts across the skin, gills and digestive tract in fish and amphibians, is equivalent to periodic exposure to higher concentrations in consumed food and water or by gavage (Although it should be noted that atrazine is absorbed across the skin of humans and by inhalation but transfer is much more significant across amphibian skin). In human cell line studies, atrazine is also effective in the low ppb range (Fan et al. 2007b) similar to effects in fish and amphibians. The levels in those cell line studies reflect concentrations in the rearing medium, and thus might be most appropriately compared to amphibian and fish studies.

On the other hand, levels in humans are measured in the urine (not in the blood and not at the tissue level) (Barr et al. 2007; Lucas et al. 1993; Swan et al. 2003). Atrazine levels in human urine, can be in ppm and are probably underestimates due to the metabolite measured in those studies (Barr et al. 2007), but it is not clear how those exposures can be compared to levels in rearing medium or how urine levels can be used to estimate tissue level exposure. Further, though we know what ­concentrations were supplied in food and water in rodent studies, there are no blood or urine measurements from exposed rodents available to compare to urine levels in humans.

Two other commonly misunderstood and misrepresented issues regarding dose are at play, in addition to the fact that “doses” (or concentrations) used in aquatic organisms cannot be translated into doses used in rodent studies and concentrations to which wildlife and humans are exposed. First, the effective dose of atrazine in amphibians, fish and cell lines (low ppb), often referred to as “extremely low” is not low at all. Estrogens are also active at that dose range in amphibians. Depending on the population, estradiol is very effective below 1 ppb, feminizing male larvae in amphibians. Thus, although atrazine (and other pesticides) may not be toxic otherwise at such levels, they are potent endocrine disruptors. Secondly, it is a mistake to assume that rodents (smaller than humans) would respond to lower amounts of atrazine relative to humans: i.e., it is not true that an effect in the ppm range in a rodent would have to be scaled up to an enormous exposure to get the same effect in a larger animal, such as humans. The contrary is true. I am reminded of the tragic story of Tusko the elephant, given an overdose of LSD, because the researchers simply scaled up the amount of LSD based on human responses (West et al. 1962). In fact, within endotherms, smaller species, with higher metabolisms and variations in enzymes, would likely tolerate higher exposures not lower, relative to humans.

Summary and Conclusions

The ultimate decision of whether or not to ban atrazine is much bigger than science…It weighs in public opinion.

Stephen Bradbury (US EPA), Oakland Tribune, Dec 3, 2006

Regardless of the mechanism(s), the number of adverse effects of atrazine that have been shown across wildlife, in laboratory rodents, and associations in human epidemiological studies, demonstrate its significant impact on environmental health and public health. In addition to the many effects on wildlife and the effects of in utero exposure documented in rodents, increasing studies are examining the impacts of atrazine on the unborn human fetus. Atrazine and other agrichemicals are associated with birth defects (Winchester et al. 2009) and low birth weight and small for gestational age in humans (Ochoa-Acuna et al. 2009; Villanueva et al. 2005). Ultimately, for chemicals like atrazine, the question becomes a cost–benefit ­analysis. Are the health risks to the environment and humans worth the benefits? Some ­estimates suggest that atrazine increases corn yield by less than 1% (Ackerman 2007), others suggest no effect at all. Although it is commonly argued that ­chemicals in agriculture help produce food economically, less than 1.5% of the corn grown in the USA is directly consumed as food (http://usda.mannlib.cornell.edu/usda/­current/FDS/FDS-11-12-2010.pdf), while 15% of the world’s population is faced with starvation (http://www.fao.org/publications/sofi/en/), in a world where amphibian are declining globally (likely along with other vertebrates) with atrazine and other environmental contaminants likely key players (Bishop et al. 1999; Blaustein and Kiesecker 2002; Boone et al. 2007; Carey and Bryant 1995; Hayes et al. 2010a). It appears that the concerns of Sanderson et al (2000): “A logical concern would be that exposure of wildlife and humans to triazine herbicides, which are produced and used in large quantities, and are ubiquitous environmental contaminants, may similarly contribute to estrogen-mediated toxicities and inappropriate sexual differentiation.” have been borne out, as predicted. Furthermore, the impact on human health remains a concern: “The observed induction of aromatase, the rate-limiting enzyme in the conversion of androgens to estrogens, may be an underlying explanation for some of the reported hormonal disrupting and tumor promoting properties of these ­herbicides in vivo.” (Sanderson et al. 2000). Considering these concerns along with the many other mechanisms of action and effects produced by atrazine, has atrazine truly been used safely for over 50 years? Or have financial considerations masked a “no-brainer”?