Volume applied in the U.S
The United States has the world’s most complete, publicly accessible data on glyphosate use. The combination of NASS, EPA, and USGS glyphosate use data provides a solid foundation to track trends in agricultural, non-agricultural, and total glyphosate use from commercial introduction through 2014. A report issued by the National Center for Food and Agricultural Policy [40] provides useful, detailed information on glyphosate use by state and crop for 1995, drawing on NASS, EPA, and information from land grant university weed management specialists.
Annual agricultural glyphosate use volumes in the nine EPA pesticide use reports issued between 1997 and 2007 exceed NASS annual totals for the same years by 20–70 %, largely because EPA had access to multiple data sources that made it possible to estimate the volume of glyphosate applied on all crops, as well as non-crop use patterns (e.g., pasture and range uses). NASS estimates, on the other hand, were limited in any given year to the crops surveyed in a particular year, and NASS never or rarely surveys pesticide use on crops grown on limited acreage. The differences are largest in the first two decades of glyphosate use (through 1995), and reflect the array of glyphosate uses not covered in NASS, crop-by-crop pesticide use surveys. But as total agricultural use rises sharply post-1996 in the wake of the introduction of GE-HT crops, glyphosate use on the major GE crops (maize, soybeans, cotton) is fully captured in NASS, EPA, and USGS data. Differences in agricultural use estimates between the datasets all but disappear by 2007 (NASS, 184.2 million pounds glyphosate use; EPA mid-range, 182.5; USGS, 183.2; [27], Additional file 1: Table S18).
Factors driving use upward
Several factors have driven the increase in glyphosate use since commercial introduction in 1974. In terms of area treated, the dominant factor has been the commercialization of GE-HT crops. Not only has glyphosate been sprayed on more hectares planted to HT crops, it has also been applied more intensively—i.e., more applications per hectare in a given crop year, and higher one-time rates of application [13, 28].
In the U.S. soybean sector, the average number of glyphosate applications rose from 1.1 per crop year in 1996 to 1.52 in 2014, while the one-time rate of application rose from 0.7 kg/hectare (0.63 pound/acre) to 1.1 kg/hectare (0.98 pound/acre) in the same period ([27], Additional file 1: Table S2). Shifts in weed communities favoring species less susceptible to glyphosate, coupled with the emergence of first, less sensitive, and eventually glyphosate-resistant weeds drove the incremental rise in the intensity of glyphosate applications on GE-HT crops [13, 10]. Rising reliance on glyphosate by soybean producers in the U.S. is graphically portrayed in Fig. 1a, while Fig. 1b shows modest change during the GE era in soybean yield/acre or production per soybean seed planted. Steady increases in the number of applications of glyphosate, rate per crop year, and glyphosate’s share of overall soybean herbicide use are shown in Fig. 1c.
Other factors contributed to rising glyphosate use. These include steady expansion in the number of crops registered for use on glyphosate product labels, the adoption of no-tillage and conservation tillage systems, the declining price per pound of active ingredient (see Fig. 2b), new application method and timing options, and new agricultural use patterns (e.g., as a desiccant to accelerate the harvest of small grains, edible beans, and other crops).
The one-time average rate of glyphosate application on Kansas wheat has incrementally risen threefold, from 0.33 kg/hectare in 1993 to 0.95 kg/hectare in 2012 ([27], Additional file 1: Table S5). The trend toward no-till and conservation tillage systems has increased wheat farmer reliance on herbicides, including glyphosate. The average two applications in recent years on winter wheat could include a pre- or at-plant spray, an application during a summer fallow period, and/or a late-season application intended to speed up harvest operations (a so-called “harvest aid” or “green burndown” use) [41]. The average rate per crop year—the single most important indicator of the intensity of glyphosate use—rose even more dramatically, from 0.47 kg/hectare in 1993 to 2.08 kg/hectare in 2012 (4.4-fold).
Harvest-aid uses of glyphosate have become increasingly common since the mid-2000s in U.S. northern-tier states on wheat, barley, edible beans, and a few other crops, as well as in much of northern Europe [41–43]. Because such applications occur within days of harvest, they result in much higher residues in the harvested foodstuffs [42]. To cover such residues, Monsanto and other glyphosate registrants have requested, and generally been granted, substantial increases in glyphosate tolerance levels in several crops, as well as in the animal forages derived from such crops. Table 7 provides an overview of key crops on which regulatory authorities have granted large increases in glyphosate tolerances to accommodate GE-HT crop uses, as well as harvest aid, green burndown applications. Note the 2,000-fold increase in the glyphosate tolerance on dry alfalfa hay and silage from 1993 to 2014, an increase made necessary by the approval and planting of GE-HT alfalfa. In response to the large increase in expected residues from such uses, some European countries now prohibit harvest-aid applications on food crops (e.g., Germany, since May 2014).
Table 7 Changes in selected U.S. EPA glyphosate tolerance levels (ppm)
Global use of glyphosate
Farmers worldwide applied about 51.3 million kgs (113 million pounds) of glyphosate in 1995 ([27], Additional file 1: Table S23). To place this volume of global glyphosate use in perspective, in just one country (the U.S.) that year, farmers applied ~60 million kgs (132 million pounds) of two herbicides (atrazine and metolachlor) on mostly one crop (maize) ([27], Additional file 1: Table S19).
But the scope and intensity of glyphosate use worldwide rapidly changed as GE-HT crops gained market share. There were about 1.4 billion hectares of actively farmed, arable cropland worldwide in 2014 [44]. Across this landmass, there were an estimated 747 million kg of agricultural applications of glyphosate. Accordingly, if this volume of glyphosate had been applied evenly, about 0.53 kg of glyphosate could have been sprayed on every hectare of cropland on the planet (0.47 lbs/acre).
Glyphosate was, of course, not applied evenly on every hectare of cropland. The average rate of glyphosate applications per hectare per crop year during 2014 fell in the range of 1.5–2.0 kg/hectare [27]. At these rates of application, the total volume of glyphosate applied in 2014 was sufficient to treat between 22 and 30 % of globally cultivated cropland. No pesticide in history has been sprayed so widely.
Since losing global patent protection around 2000, dozens of companies began manufacturing technical glyphosate, and/or formulating glyphosate products. Some two-dozen Chinese firms now supply 40 % of the glyphosate used worldwide, and export most of their annual production [45].
The loss of patent protection and increased generic manufacturing of glyphosate has placed downward pressure on prices since 2000 [30, 45, 46]. The major manufacturer, Monsanto, has typically not competed directly or solely on price, and instead has been successful in holding or expanding market share by bundling purchase of higher-price, Monsanto brand, Roundup herbicides with the purchase of Monsanto herbicide-tolerant seeds [45–47]. Especially in the U.S., this bundling strategy has been augmented by various volume incentives and discounts, special financing, rebates for purchase of other herbicides working through a mode of action other than glyphosate’s (to delay the spread of resistant weeds), and other non-price benefits tailored to appeal to large volume customers [46–48].
The diversity of global uses in agriculture and other sectors has grown over the past 40 years [9], making it more difficult to compile accurate global data across all glyphosate uses, especially by sector and specific use. As a result, global glyphosate use projections can only be based on industry-wide glyphosate production figures, as done from 1997–2014 in Table 4 and Additional file 1: Table S24 [27].
Impact of GE-HT technology
The development and marketing of GE, Roundup Ready crops fundamentally changed how crop farmers could apply glyphosate. Before RR technology, farmers could spray glyphosate prior to crop emergence, for early-season weed control, or after harvest to clean up late-season weeds. But with RR crops, glyphosate could also be sprayed 1–3 times or more after the crop had emerged, leaving the crop unharmed but controlling all actively growing weeds. This historically significant technological advance set the stage for unprecedented and rapid growth in the area planted to RR crops and sprayed with glyphosate (from usually less than 10 % of cotton, maize, and soybean acres pre-1996, to 90 % or more today) [47, 49, 50].
The interplay of various factors leading to increased glyphosate use is apparent in Fig. 2a, which shows the trend in overall glyphosate use on the key GE-HT crops in the U.S., the correlation between reductions in average price per pound and use (Fig. 2b), and rising use and the emergence of resistant weeds (Fig. 2c).
Use of glyphosate on some GE-HT crops may have declined, or may soon begin declining in some regions because (a) adoption of GE-HT soybeans, cotton, and canola has peaked in most of the countries that have embraced GE technology [9], and (b) farmer willingness to pay for repeat applications of glyphosate, or further increase application rates, typically declines as glyphosate-resistant weeds become well established, as they have in much of the U.S. [13] and in Brazil and Argentina [10]. On the other hand, GE-HT crops may move into some regions not previously planting them (e.g., China), and reductions in the price of generic glyphosate herbicides could lead to more intensive use in some countries.
In the countries that have planted the largest shares of GE-HT crops (the U.S., Argentina, and Brazil), glyphosate use rates per hectare per crop year have risen sharply since around 2000 [20, Additional file 1: Tables S2, S3, S22]. Worldwide on GE soybean and cotton, average total herbicide use per crop year per hectare has approximately doubled from 1996 to 2014, with the increase in glyphosate volumes applied per hectare accounting for nearly all of the per hectare increase. Maize herbicide use per hectare has risen modestly, if at all, in large part because adoption of GE-HT maize hybrids allowed farmers to reduce reliance on a half-dozen other widely used maize herbicides applied at relatively high rates (e.g., ~1 kg/hectare per crop year) [11].
Because GE-HT soybeans account for two-thirds of the total hectares planted to GE-HT crops worldwide, the doubling of average herbicide use per hectare of HT soybeans drives the sizable increase in overall herbicide on all GE crop hectares. There is, as well, a clear connection throughout South America in the adoption of GE-HT technology and no-tillage systems [17, 38]. No-till farming in South America lowers machinery and labor costs, and reduces soil erosion, but at the expense of heightened reliance on herbicides for weed control, and other pesticides to control insects and fungal pathogens.
Despite gaps in publicly accessible data, the dramatically upward trajectories in glyphosate use in the U.S. and globally are unmistakable. In the pre-GE era (1974–1995) in the U.S., non-agricultural glyphosate uses accounted for ~34 to 42 % of total use. The share of total glyphosate use accounted for by the agricultural sector shifted markedly upward post-1996, starting at 66 % in 1996 and reaching 81 % 5 years later (2001) and 92 % by 2014 ([27], Additional file 1: Table S18).
The total volume of use and the split between agricultural and non-agricultural uses in the pre-GE era period are subject to greater uncertainty than in the 1996–2014 period. However, pre-1995 glyphosate use is minor compared to the post-GE period, when both data quantity and quality improved, especially covering applications in the U.S. and on global GE-HT hectares planted.
Figure 3 arrays milestones in the history of glyphosate discovery, commercialization, and regulation, while Fig. 4 displays key events in the history of glyphosate use and impacts.
Rising use triggers new concerns
Driven by the growing diversity of uses and dramatic increases in volumes applied, levels of glyphosate and its primary metabolite aminomethylphosphonic acid (AMPA) have been detected in the air [51], soil [52], and water [49, 53]. With few exceptions though, contemporary levels of glyphosate in the air, water, and food result in typical human exposure estimates that remain well below the “levels of concern” or “Acceptable Daily Intakes” established by regulatory bodies around the world.
Still, a growing body of literature points to possible, adverse environmental, ecological, and human health consequences following exposure to glyphosate and/or AMPA, both alone [54] and in combination with ingestion of GE proteins (e.g., EPSPS, Bt endotoxins) [55]. Environmental studies encompass possible glyphosate impacts on soil microbial communities and earthworms [56–58], monarch butterflies [59], crustaceans [60], and honeybees [61].
Studies assessing possible risks to vertebrates and humans include evidence of rising residue levels in soybeans [62, 63], cancer risk [64], and risk of a variety of other potential adverse impacts on development, the liver or kidney, or metabolic processes [54, 55, 65–80].
Relative toxicity and impacts
For years, glyphosate has been regarded as among the least chronically toxic herbicides for mammals, and indeed only three EPA-registered synthetic pesticides in current agricultural use have a higher chronic Reference Dose (the imidazolinone herbicides imazamox, imazethapyr, and imazapyr).
For human exposures, the U.S. EPA has set glyphosate’s daily chronic Reference Dose (cRfD) at 1.75 milligrams per kilogram of bodyweight (mg/kg bodyweight/day). The EU-set cRfD for glyphosate was recently raised from 0.3 to 0.5 mg/kg/day, 3.5-fold lower than EPA’s. A team of scientists has compiled evidence supporting the need for a fivefold reduction in the EU cRfD to 0.1 mg/kg/day [81], a level 17-times lower than EPA’s.
Glyphosate is a moderate dose herbicide with relatively low acute and chronic mammalian toxicity, to the extent mammalian risk is accurately reflected in required EPA toxicology studies. After an exhaustive review, however, glyphosate was classified in 2015 as a “probable human carcinogen” by the International Agency for Research on Cancer [64], based on increased prevalence of rare liver and kidney tumors in chronic animal feeding studies, epidemiological studies reporting positive associations with non-Hodgkin lymphoma, and strong mechanistic evidence of genotoxicity and ability to trigger oxidative stress [64].
The body of toxicological studies supporting glyphosate’s current EPA and EU cRfD, and hence all contemporary uses of this herbicide, dates back to the early 1970s through mid-1980s [82]. Recent studies suggest that glyphosate in its pure form, and some formulated glyphosate end-use products, may be triggering epigenetic changes through endocrine-mediated mechanisms [54, 73, 75, 76, 79, 81, 83].
Evidence from multiple studies suggests that the kidney, and secondarily the liver, is at risk of glyphosate-triggered, or glyphosate-enhanced chronic degeneration [55, 71, 72, 84, 85]. Industry metabolism studies in farm animals, rats and mice, and rabbits were conducted in the 1970s and 1980s, and show that in animal feeding studies, glyphosate levels in the kidney usually exceed those in the liver by three- to tenfold, and those in the liver exceed levels in other tissues by a wide margin [86].
The apparent tendency of glyphosate to concentrate in the kidneys, coupled with glyphosate’s action as a chelating agent, has led some scientists to hypothesize that glyphosate can bind to metals in hard drinking water, creating metallic-glyphosate complexes that may not pass normally through kidneys [71, 72]. For this, or other as yet unrecognized reasons, the risk of chronic kidney disease may be heightened in human and animal populations with heavy glyphosate exposure.
The IARC classification and emerging evidence relative to kidney damage and endocrine effects heightens the need for, and will complicate ongoing and future glyphosate worker and dietary-risk assessments. Annual residue tests are carried out by the U.K. Food Standards Agency (FSA). Residues of glyphosate were found in 10–30 % of grain-based samples from 2007–2013, at generally rising levels [87]. Glyphosate and AMPA residues are present at relatively high, and rising levels (over 1 ppm) in a high percentage of the soybeans grown in the U.S., Canada, Brazil, Argentina, Paraguay, countries which account for 86.6 % of the 11.6 billion bushels of soybeans produced globally in 2014, and nearly all global trade in soybeans and soybean-based animal feeds [34, 62].