U.S. Environmental Protection Agency Office of Pesticide Programs Details of the Consultation for Department of State Use of Pesticide for Coca Eradication Program in Colombia, August 2002

Table of Contents

Executive Summary

1. Section 1. Description of Glyphosate Use in the U.S. For Comparison To Use in Colombia For Coca Eradication

   1. Use of Glyphosate in the United States Agricultural Use Sites
   2. Non-Agricultural Uses Including Forestry
   3. Properties of Glyphosate
   4. Formulation of Glyphosate
   5. Glyphosate Used With Surfactant For Foliar Absorption

2. Section 2. Human Health Risk Assessment For the Use of Glyphosate Herbicide As Part of the Aerial Eradication Program of Coca in Colombia, S.A.

   1. Hazard Identification
   2. Dose Response Assessment
   3. Exposure Assessment

3. Section 3. Review of Glyphosate Incident Reports With Special Reference To the Aerial Spraying Program in Colombia

   1. California Data1982 Through 2000
   2. Literature Review
   3. A Study of Health Complaints Related To Aerial Eradication in Colombia
   4. Conclusions

4. Section 4. Ecological Risk Assessment For the Use of Glyphosate Herbicide As Part of the U.S. Supported Aerial Eradication Program of Coca in Colombia

   1. Background
   2. Ecological Risk Assessment
   3. Potential Spray Drift of Glyphosate
   4. Environmental Fate And Transport Assessment of Glyphosate
   5. Risk Characterization

EXECUTIVE SUMMARY

CONSULTATION REVIEW OF THE USE OF PESTICIDE FOR COCA ERADICATION IN COLOMBIA

The Department of State currently assists the Government of Colombia with training, contractor support, financial assistance, and technical and scientific advice for an aerial pesticide spraying program designed to eradicate illicit crops (coca and poppy). In accordance with a provision in its 2002 Appropriations Bill, Department of State has consulted with the U.S. Environmental Protection Agency (EPA) on two specific questions: that fumigation is "...being carried out in accordance with regulatory controls required by the Environmental Protection Agency as labeled for use in the United States" and, that the chemicals being used "...in the manner they are being applied, do not pose unreasonable risks or adverse effects to humans or the environment."

FINDINGS

EPA has reviewed information provided by Department of State concerning the pesticide formulation applied (combination of the pesticide active ingredient, glyphosate, and inert ingredients), and application rates and methods. According to the most recent figures (1999 sales and usage) glyphosate is the most widely used conventional pesticide in the United States. The Agency evaluates all pesticides used in the U.S. to determine whether they meet the U.S. safety standard of no unreasonable adverse effects. Consequently, EPA has a significant amount of information about glyphosate from a health and environmental standpoint because of our reviews of use conditions in the U.S. In the U.S., the Agency can assure significant controls on use and potential health and environmental impacts through the pesticide label, and through a state infrastructure which governs label compliance to address issues such as drift and worker and bystander exposure.

Based on a comparison of the glyphosate use pattern in Colombia, as described by the Department of State, and use in the U.S., EPA determined that the most equivalent U.S. uses of glyphosate would be forestry or rights-of-way. The glyphosate product which has been identified to us as used in Colombia is registered in both the U.S. and Colombia, although it has never been marketed in the U.S. The Agency found application rates described as used in Colombia to be within the parameters listed on U.S. labels. The addition of a spray adjuvant (to facilitate the formulation adhering to and penetrating the coca plant) is also in keeping with U.S. practice. While the specific spray adjuvant product identified as that used in Colombia is not sold in the U.S., similar substances and products are commonly used. Most U.S. labels for forestry and right-of-way use of glyphosate suggest application by helicopter. Since application in Colombia is done by fixed-wing aircraft, it is likely conducted at a higher speed and from a greater altitude, than would be typical in the U.S. Department of State has assured the Agency that mixers/loaders and applicators of the glyphosate formulation receive training comparable to U.S. label requirements for glyphosate products including the use of personal protective equipment such as gloves and goggles.

As for potential human health effects of the coca eradication program, there are no risks of concern for glyphosate, per se, from dermal or inhalation routes of exposure, since toxicity is very low. There is concern for acute eye toxicity due to an inert ingredient in the glyphosate formulated product used to treat coca. The potential eye effects are related to an inert ingredient, not the glyphosate itself, and greatest potential for exposure is expected for workers mixing and loading the concentrated glyphosate product. The components of the spray adjuvant, Cosmoflux 411F, are substances with low oral and dermal toxicity that have been approved for use in/on food by EPA and the Food and Drug Administration. There are no expected toxicological effects of concern for acute (short-term) or chronic (long-term) dietary exposure through food and water from the coca eradication program. Incident data from Colombia involving humans, livestock, mammals and birds, are based on potential exposure to glyphosate from fumigation of poppy fields, which may differ from use of and exposure to glyphosate from coca eradication, so conclusions should be made cautiously.

Relative to the potential environmental effects of the spraying program based on U.S. data, phytotoxicity to non-target plants outside of the application zone would be expected, since glyphosate is a broad spectrum herbicide. Given the application method described by Department of State, offsite exposure from spray drift is probable, as it would be under similar uses in the U.S. This proposed use of glyphosate itself does not appear to pose a significant direct risk to terrestrial or aquatic animals, although secondary adverse effects from the temporary loss of habitat in the spray area could occur. EPA would not expect any risk to birds and mammals, including livestock, based on dietary exposure to the active ingredient glyphosate. Anticipated effects to animals are based on an extrapolation of data related to North American species. Glyphosate does not have a high potential to leach to ground water or reach surface water as dissolved runoff but does have potential to contaminate surface water as a result of residues suspended in runoff water. A more refined assessment is difficult due to uncertainty regarding the exact formulation of the spray solution.

As part of its consultation, EPA reviewed available scientific studies and information on the human health and environmental effects of glyphosate and the inert ingredients and on exposure pathways; conducted a literature search for human health incidents related to glyphosate use in the U.S.; and summarized use patterns for glyphosate in the U.S., including use sites, methods and rates of application, and differing formulations. In addition, the Agency considered information, provided by non-governmental sources, concerning adverse effects reportedly connected to the eradication program.

Details of the Agencys findings are provided in the attached document.

SECTION 1. Description of Glyphosate Use in the U.S. for Comparison to Use in Colombia for Coca Eradication

INTRODUCTION: The Biological and Economic Analysis Division (BEAD) within the Office of Pesticide Programs, Environmental Protection Agency, has been asked to describe the use of glyphosate within the United States with a more detailed description of its use in forestry sites so that methods of use in the United States may be used as a basis for comparison for coca eradication in Colombia (1).

SUMMARY: Glyphosate is the herbicide most widely used in the United States¹ (2). In agriculture this popularity is due, in large part, to the development of crops that are highly tolerant to broadcast applications of glyphosate which allows growers to use this non-selective herbicide as their principal method of weed control in certain crops. Growers have rapidly adopted glyphosate-based weed control programs with glyphosate tolerant crops because they are simple and economical (3). In addition, due to its unique properties, glyphosate is also widely used for non-agricultural weed control situations including home lawns and gardens, forestry and other non-crop sites where total vegetation control is desired.

Glyphosate must be applied to the target plants foliage to be effective. Glyphosate is non-selective in action, controlling a wide variety of plants. Once absorbed, it circulates to untreated portions of the plant; and it has no residual activity after contact with soil. Glyphosate may be applied using hand-held, ground-driven or aerial equipment; the choice of equipment is determined by the site to be treated. Although higher rates of application are allowed, actual rates per application in agricultural sites average less than 0.75 pounds of the active ingredient glyphosate per acre (Table One). For the non-agricultural site, forestry, use is allowed at rates per application ranging from 2 to 10 pounds per acre (2.2 to 11.2 kilograms of active ingredient/ hectare) of glyphosate in the form of the isopropylamine salt (6). This rate of use may also be expressed as 1.5 to 7.5 pounds per acre of acid equivalent (pure glyphosate; not a salt). A more detailed discussion of the use of glyphosate in the U.S. follows.

USE OF GLYPHOSATE IN THE UNITED STATES: Products containing glyphosate are registered with the U.S. Environmental Protection Agency, which means they may legally be used within the United States if used in accordance with label instructions.

Glyphosate may be used on over 400 crop and non-crop sites. The largest agricultural use sites include soybeans, cotton and field corn. The following table summarizes estimates the use of glyphosate in three primary agricultural use sites.

Table One. 2000 Glyphosate Use in U.S. Agriculture: In Total and for the Three Major Use Sites

Site	Acres Grown (million acres)	Base Acres Treated (million acres)1	Percent of Crop Treated2	Total Acres Treated (million acres)3	Pounds Applied (million lbs ai)	Avg. Number of Apps	Avg. Application Rate (lbs ai/acre/app)
All Ag. Sites				102.7	73.5
Corn	73.8	6.6	9%	7.3	4.4	1.1	0.59
Cotton	14.4	8.1	56%	13.8	9.5	1.7	0.67
Soybeans	71.0	44.0	62%	57.2	41.8	1.3	0.68

Source: USDA, National Agricultural Statistics Service (NASS) 2000 field crop chemical use (May, 2001), and EPA proprietary data.
¹Base acres treated = acres treated 1 or more times with glyphosate.
²Percent of crop treated = base acres treated with glyphosate divided by acres grown.
³Total acres treated = based acres treated with glyphosate multiplied by the average number of applications of glyphosate.

In addition to agricultural use, EPA estimates that 16-22 million pounds of the technical grade active ingredient were applied to non-agricultural sites in 1999 (this is the most recent year for which adequate data are available). The estimate includes both home owner and professional applications as well as use on forested lands (11). Based on EPA data for 1999, an estimated 1-2 million pounds of glyphosate was applied to forest acres, with more than 650,000 forest acres treated.

AGRICULTURAL SITES: In certain annual crops, glyphosate may be applied before planting (preplant) to control existing weeds; often replacing tillage as a weed control measure in "no-till" crop culture systems designed for reducing soil erosion. However, most of the glyphosate currently used in agricultural sites is used in a cropping system employing crop varieties that have been developed to be resistant to glyphosate so that glyphosate may be applied "over-the-top" of the resistant crop to kill susceptible weeds. The most prevalent of these systems is the Round-up Ready Soybean system. These soybeans, which are highly tolerant to glyphosate, were marketed starting in 1996. Since then this system has been widely and rapidly adopted; in 1990-1991, glyphosate ranked 11^th among conventional pesticides used in the U.S. with annual use estimated to be 18.7 million pounds (4). In 2000, glyphosate was the most widely used herbicide in soybeans; nearly 42 million pounds of glyphosate were applied to soybeans alone (2) with over 60 percent of soybeans surveyed by USDAs National Agricultural Statistical Service treated with glyphosate (5). Round-up, a glyphosate product marketed for agricultural use is formulated with a surfactant during manufacture to facilitate foliar absorption. The following table (Table Two) summarizes the use rates specified in label instructions for Roundup Original^TM product herbicide concentrate (12) which contains the isopropylamine salt of glyphosate for use in glyphosate-tolerant soybeans.

Table Two. Allowed Rate of the Isopropylamine Salt of Glyphosate Per Application in Tolerant Soybeans

NON-AGRICULTURAL USES INCLUDING FORESTRY: BEAD has been asked to describe the use of glyphosate in U.S. forestry sites since that use most closely corresponds to the use of glyphosate in Colombia for coca control. For simplicity this document only refers to Accord which is intended specifically for forestry use (6). This product contains the isopropylamine salt of glyphosate (41.5%), and is labeled for non-agricultural uses including Forestry Site Preparation and Utility Rights-of Way, Forestry Site Conifer and Hardwood Release, and Wetland Sites in the U.S. Table Three describes the rates at which it may be used. It is recommended for use in site preparation prior to planting any tree species, including Christmas tree and silvicultural nursery sites (7). Specific methods of application for forestry uses include: aerial spraying; spraying from a truck, backpack or hand-held sprayer; wipe application; frill treatment; cut stump treatment (7).

For forestry site preparation it may be applied using either ground or aerial equipment at rates from 2 to10 pounds glyphosate active ingredient per acre which is equivalent to 2.2-11.2 kilograms of active ingredient per hectare (Table Three). It may also be applied using hand-held equipment. Product instructions specify that a non-ionic surfactant be added to the spray mixture for all forestry uses at a rate of 0.5 to 1.5 percent by spray volume (2 to 6 quarts of surfactant per 100 gallons of spray solution). It may also be combined with certain residual herbicides to extend the period of weed control beyond that obtained with glyphosate alone.

The isopropylamine salt of glyphosate may also be used in forestry conifer and hardwood release as a directed spray or by using selective equipment. This product may also be used in or around wetland sites generally at no more than 5 quarts of product per acre (5 lbs isopropylamine salt of glyphosate per acre which is equal to 5.6 kg/ha) using over-water broadcast application (5).

Table Three. Rate of Isopropylamine Salt of Glyphosate Per Application For Certain Use Sites

PROPERTIES OF GLYPHOSATE: Glyphosate is a foliar-active herbicide; to exert herbicidal properties it must enter the plant through foliage (or in some cases, the stem). Glyphosate applied to foliage is absorbed by leaves and rapidly moves through the plant. It acts by preventing the plant from producing an essential amino acid. This reduces the production of protein in the plant, and inhibits plant growth.

Glyphosate has systemic activity, meaning that it circulates through the plants vascular system; affecting the entire plant, not just the treated foliage. Other foliar-active herbicides, like paraquat for example, are contact herbicides; affecting only the portion of the plant onto which they are applied. After treatment with a contact herbicide, a plant may then regrow from untreated portions, often necessitating re-treatment for complete control. The advantage to a systemic herbicide is that if applied at an appropriate dose, it can kill an entire plant, thus preventing regrowth from an untreated plant part such as a root.

Glyphosate has no residual activity, once adsorbed to soil it quickly becomes unavailable to plants and no longer has herbicidal activity. This means that a plant that would ordinarily be susceptible to glyphosate can be planted shortly after an application of glyphosate; this is common practice in U.S. agriculture. In contrast, some herbicides have month-long or even year-long residual activity which limits the plants that may be grown following their use.

Glyphosate is non-selective. Some herbicides are selective in their action, controlling only grassy weeds in a broadleaf crop like soybeans, for example. However, glyphosate exerts herbicidal action on a variety of plants; it is active on grasses, herbaceous plants including deep rooted perennial weeds, brush, some broadleaf trees and shrubs, and some conifers. However, glyphosate does not control all broadleaf woody plants. Plants vary in their susceptibility to glyphosate, so the treatment of dose in important. Plants of certain species and older plants are less susceptible to glyphosate Timing is critical for effectiveness on some broadleaf woody plants and conifers.

FORMULATIONS OF GLYPHOSATE: Glyphosate and four salts of the parent glyphosate molecule are currently used as active ingredients in registered pesticide products in the U.S.(9). These products are registered with the U.S. EPA for use in the U.S. in many different crop, non-crop, industrial and residential sites.

Table Four. Number of Products and Sites for Different Formulations of Glyphosate

GLYPHOSATE USED WITH SURFACTANT FOR FOLIAR ABSORPTION: Since glyphosate is only effective if absorbed by plant foliage, glyphosate is combined with a surfactant to facilitate its absorption. Many herbicide concentrates, including glyphosate, are mixed with water before application. Without a surfactant, the aqueous spray mixture is repelled by the plants waxy cuticle layer ("beads up"), and quickly runs off the plants surface, preventing absorption.

Surfactants are commonly used as wetting agents with herbicides and in other products such as laundry and dishwashing detergent. Non-ionic surfactants, which are comprised of alcohols or fatty acids and considered an all-purpose surfactant are commonly used with glyphosate-containing products. Surfactants are frequently added during manufacture of the herbicide concentrate. If not, a non-ionic surfactant is generally mixed with the herbicide and water before spraying to enable the liquid to make better contact with the waxy cuticle of the plant. These glyphosate products, which are formulated without a surfactant, are considered "non-loaded" (10).

A short description of international usage of glyphosate, including Colombia, appears after the following references to the above discussion.

(1) Description of Use of Glyphosate in Coca Eradication in Colombia in attachment to a letter from Secretary of State Colin Powell to Environmental Protection Agency Administrator Governor Christine Whitman.

(2) Donaldson, D., T. Kiely, and A. Grube. Pesticide Industry Sales and Usage, 1998 and 1999 Market Estimates. June 2002. Biological and Economic Analysis Division, Office of Pesticide Programs, U.S. Environmental Protection Agency.

(3) Gianessi, L.P., Silvers, C., Sankula, S., and Carpenter, J. Plant Biotechnology: Current and Potential Impact for Improving Pest Management in U.S. Agriculture - An Analysis of 40 Case Studies. June 2002. National Center for Food and Agricultural Policy.

(4) Glyphosate - Reregistration Eligibility Decision. United States Environmental Protection Agency. September 1993.

(5) Agricultural Chemical Usage - 2000 Field Crops Summary. May 2001. United States Department of Agriculture. National Agricultural Statistics Service.

(6) Accord Herbicide Specimen Label. Available on-line from CDMS.

(7) Pesticide Fact Sheet. Prepared for the U.S. Department of Agriculture, Forest Service.

(8) D. Lantagne., M. Koelling, and D. Dickman. Effective Herbicide Use in Christmas Tree Plantations. Michigan State University Extension.

(9) U.S. Environmental Protection Agency, Office of Pesticide Programs. Search of the Reference Files System dated June 11, 2002.

(10)Miller, P., and P. Westra. Crop Series: Production. Colorado State University.

(11) EPA Proprietary Information

(12) Roundup Original Herbicide^TM Supplemental Labeling for Postemergence Applications to Soybeans with Roundup Ready Gene. Available online from CDMS.

BEAD was asked to report on the use of glyphosate in forestry sites since it seemed similar to the use pattern for coca eradication. However, it is not clear how closely this use approximates that for coca eradication. Glyphosate is typically applied to forestry sites using helicopters at air speeds of 50-70 knots (about 60-80 miles per hour). Application to forestry sites by fixed wing aircraft, if practiced at all, is extremely rare (1). The recommended rate of application for pine release (conifer release) is 1.5 to 2 pounds active ingredient per acre.

Aerial application to other sites comprises less than one percent of the total amount of glyphosate applied in the United States (3).

In addition to surfactants, drift control agents may be added to the spray mixture for forestry uses in an effort to prevent drift to off target sites. BEAD has not investigated the prevalence of use or the effectiveness of these products.

BEAD estimates total global use of glyphosate to be between 350 and 360 million pounds of glyphosate per year. Annual use in the United States is approaches 100 million pounds of active ingredient and an estimated 250 to 260 million pounds of glyphosate is used outside of the United States. Use of glyphosate in Colombia accounts for between four and five million pounds of this use. Primary sites in Colombia include coffee, bananas, pasture-land and rice (3).

(1) Personal communication between Virginia Werling, United States Environmental Protection Agency and John Taylor, United States Forest Service on August 9, 2002.

(2) Hamilton, R.A. "2002 North Carolina Agricultural Chemicals Manual - Chemical Weed Control In Forest Stands" Extension Forest Resources Department, North Carolina State University. Available on-line at http://ipm.ncsu.edu/agchem/chptr8/821.pdf

(3) United States Environmental Protection Agency Proprietary Data.

The Department of State has requested that the US Environmental Protection Agency (EPA), Office of Pesticide Programs (OPP) provide a human health risk assessment for the aerial coca eradication in Colombia. The Department of State (DoS) has requested that the risk assessment address whether or not the aerial eradication program in Colombia is being carried out in accordance with regulatory controls required by the EPA as labeled for use in the United States, and the chemicals used, in the manner in which they are being applied, do not pose unreasonable risks or adverse effects to humans and or the environment. According to information provided by DoS, the eradication program, includes the use of a spray mixture of a glyphosate containing pesticide product, an adjuvant (Cosmo-Flux 411F) and water. The glyphosate tank mixture is applied in an over the top aerial foliar application in certain provinces within Colombia. To facilitate the request, the DoS met with members of OPP on April 18 and sent a written request, dated May 8, 2002, with documentation on the coca eradication program, including a description of the pesticide spray mixture components, application methods, target site identification, and potential exposures. DoS also supplied EPA with incident reports for aerial eradication of illicit poppy in Colombia.

The Field and External Affairs Division (FEAD) of OPP, which has the responsibility of managing the EPAs role of providing technical information and assistance to DoS for this program, forwarded the DoS request to Health Effects Division (HED), the Environmental Fate and Effects Division (EFED), and the Biological and Economics Assessment Division (BEAD) for scientific assessments. The HED of OPP is charged with estimating the risk to human health from exposure to pesticides. Registration Action Branch 1 (RAB1), Reregistration Branch 1 (RRB1) and the Chemistry and Exposure Branch (CEB) of HED as a team have performed the assessment requested by the Department of State and have evaluated the potential hazard, exposure, and risk to human health from the U.S. supported coca eradication program Colombia.

A summary of the findings and an assessment of human risk resulting from the use of glyphosate in Colombia to eradicate coca is provided in this document.

Unless otherwise specified, all information pertaining to the U.S. supported coca eradication program in Colombia was provided to the Agency from two sources: (1) Department of State (DoS) Presentation, DoS Coca Eradication Program, 4/18/02, (2) DoS document entitled Chemicals Used for the Aerial Eradication of Illicit Coca in Colombia and Conditions of Application.

According to the State Department, the glyphosate tank mixture is applied as an over the top aerial foliar application to coca in certain provinces within Colombia. The tank mixture sprayed for eradication of coca in Colombia contains 55% water, 44% of glyphosate herbicide product, and 1% adjuvant (Cosmo-Flux 411F). Up to two applications of the glyphosate tank mixture are sprayed over coca crops at a maximum of 1.25 gallons of product/Acre. In order to assess the hazard of what was sprayed in Colombia, components of the mixture were evaluated separately.

The Cosmo-Flux 411F adjuvant used in the glyphosate tank mix is produced by a Colombian company and is not sold in the U.S. All ingredients identified as contained in this product are substances that are not highly toxic by oral or dermal routes. They may cause mild eye and skin irritation. Cosmo-Flux 411F consists mainly of (information not included as it may be entitled to confidential treatment) with a nonionic surfactant blend primarily composed of (information not included as it may be entitled to confidential treatment).

The available hazard data base on experimental animals indicates that the Glyphosate technical grade active ingredient (TGAI) has low acute toxicity via the oral and dermal routes. It is a mild eye irritant and a slight dermal irritant. It is not a dermal sensitizer. The requirement for an acute inhalation study was waived since no respiratory or systemic toxicity was seen following subchronic inhalation exposure in rats. In the subchronic and chronic oral toxicity studies (1-year dog, 24-month mouse, 2-year chronic/carcinogenicity rat, and 2-generation rat reproduction), systemic toxicity manifested most commonly as clinical signs, decreases in body weight and/or body weight gain, decreased food consumption, and/or liver and kidney toxicity at doses equal to or above the limit dose (1000 mg/kg/day). No dermal or systemic toxicity was seen following repeated dermal exposures. There was no quantitative or qualitative evidence for increased susceptibility in fetuses following in utero exposure to rats and rabbits in developmental toxicity studies or following pre/post-natal exposure to rats in the 2 generation reproductive toxicity study in rats. Effects in the offspring were observed only at or above treatment levels which resulted in evidence of appreciable parental toxicity.

The Food Quality Protection Act (FQPA) Safety Factor Committee (SFC) concluded that the safety factor, to protect infants and children, of 10x be removed (reduced to 1x). The Hazard Identification Assessment Review Committee (HIARC) met on March 26, 1998 and, again, on November 20, 2001. The most recent report of the HIARC for glyphosate has the complete assessment of the endpoints selected for dietary exposure and residential/occupational exposure. No endpoints were selected for the acute Reference Dose (RfD) since no hazard attributed to a single dose was identified from the oral toxicity studies, and there are no concerns for developmental or reproductive toxicity. In addition, the HIARC did not identify endpoints of concern for dermal and inhalation exposures for any exposure period (short term 1-30 days, intermediate term- 1 to 6 months, or long term- 6 months to lifetime) since no hazard was identified due to the low toxicity of glyphosate. HIARC did identify an incidental oral endpoint for short- and intermediate-term exposure. The chronic dietary RfD of 1.75 mg/kg/day was based on diarrhea, nasal discharge, and mortality in a rabbit developmental toxicity study. Glyphosate was not mutagenic in a full battery of assays. Based on the lack of evidence for carcinogenicity in two acceptable studies in mice and rats, glyphosate is classified as a "Group E" chemical (no evidence of carcinogenicity to humans).

An exposure and risk assessment is required for an active ingredient if: (1) certain toxicological criteria are triggered and (2) there is potential for exposure. Upon review and analysis of the hazard database in total, the Agencys HIARC did not identify a hazard of concern for acute dietary, dermal, or inhalation exposures. Therefore, quantitative estimates of risk for these exposure durations have not been conducted (TXR No. 0050428, W. Dykstra, 22-JAN-2002).

Acute dietary exposure is possible for persons consuming livestock or food crops which have been inadvertently sprayed as a result of the aerial eradication program in Colombia. However, since glyphosate is a contact herbicide that systemically kills plants after absorption through leaves, dietary exposure due to consumption of treated crops is expected to be limited. In addition, since an acute dietary endpoint was not identified in the hazard database, no significant risk due to acute dietary food exposure to glyphosate residues is expected. Based on the fact that a coca field is sprayed no more than twice to eradicate the crop, no chronic food exposure is expected.

Handler (e.g., individuals mixing the concentrated formulated product to prepare the tank mix and loading the tank mix in the aircraft) exposure is anticipated for short-term (1-30 days) and, possibly intermediate-term (1-6 months) durations based on the frequency of application and duration of the spray program.

Based on the use pattern described by the DoS, short-term dermal post-application exposures are expected for persons pruning, or leaf pulling treated coca plants immediately after spray events. In cases such as glyphosate, where the vapor pressure is negligible, OPP experience with post-application data suggests that inhalation exposure is minimal and OPP does not quantitatively assess post-application inhalation exposure. Intermediate- and long-term post-application exposures are not expected due in part to the fact that a coca field is sprayed twice to eradicate the crop. Additionally, glyphosate is a translocated herbicide which is rainfast within 48 hours after spraying. Therefore, potential exposure to dislodgeable residues of glyphosate after 48 hours is expected to be minimal.

DoS states that pilots are instructed not to spray fields where people are present. Therefore, incidental oral exposure (hand-to-mouth) resulting from individuals being directly sprayed by glyphosate was not assessed. Non-dietary incidental oral exposure was not quantitatively assessed for the use of glyphosate in Colombia.

There is potential for exposure to bystanders in areas near those targeted for spraying. However, the technology and other safeguards described by DoS as being used in this program are consistent with common approaches in the United States for reducing spray drift. Therefore, it is likely that drift is minimized in this program if all procedures are adhered to and operational equipment is in working order.

From the review of Colombian glyphosate product human incident reports for poppy eradication, it should be emphasized that the overwhelming majority (95%) of the illnesses reported are likely background incidents unrelated to the spraying of herbicide to poppy. The remaining 5% increase could be due to a variety of causes and do not support a conclusion that the spraying of the glyphosate tank mixture was responsible for these complaints. Furthermore, the individual with the highest potential for exposure would be the mixer loader. They are handling the concentrated glyphosate product and the tank mix. The incidence data that has been submitted to the Agency by DoS, does not include any incident reports for those individuals. There are data to suggest that the poppy spray eradication program could have resulted in minor skin, eye, or respiratory irritation, and perhaps headache or other minor symptoms. However, the detailed information on timing of application, history of exposure, and medical documentation of symptoms related to exposure to glyphosate tank mix were not available. Given the limited amount of documentation, none of the data in the report from Colombia provide a compelling case that the spraying of the glyphosate mixture has been a significant cause of illness in the region studied. Prospective tracking of reports of health complaints, documenting times of exposure and onset of symptoms, are recommended during future spray operations to evaluate any potential health effects and ameliorate or prevent their occurrence.

A direct comparison of the epidemiological data in Colombia (which is from aerial application to poppy) to the conditions of use, (as presented at the April 18, 2002 briefing for aerial application to coca by DoS to OPP risk assessors), would be limited. The briefing and the materials provided did not address the conditions of use for poppy. Nor was the Agency provided any human incident data for the coca eradication program. Subsequent to the April 18 briefing HED received an e-mail communication from OPP/Field and External Affairs Division, stating that DoS informed EPA that the application rate for poppy was lower than that for coca. According to the DoS, the use pattern of the glyphosate mixture on poppy differs from the use on coca. Other details of the differences between the two spray programs have not been supplied to the Agency. Specifically, the Agency has no information as to the exact makeup of the tank mixture sprayed on poppy, or whether the same glyphosate product and adjuvants used in the coca eradication program were used in the poppy eradication program. The Agency also has questions as to the geographical area differences, the frequency of repeated applications, and the size of the area treated on each spray mission. Therefore, generalized conclusions drawn from human incident data as a result of application to opium poppy, in comparison to conditions of use for the coca eradication program should be made with caution.

The glyphosate formulated product used in the coca eradication program in Colombia contains the active ingredient glyphosate, a surfactant blend, and water. The acute toxicity test of the glyphosate technical is classified as toxicity category III for primary eye irritation and toxicity category IV for acute dermal and oral toxicity, and skin irritation. It is not a dermal sensitizer. However, the surfactant used in the formulated product reportedly can cause severe skin irritation and be corrosive to the eyes, as would be expected for many surfactants. The label for the formulated product used in the coca eradication program in Colombia includes the "Danger" signal word. These findings suggest that any of the reports of toxicity to the eye may be due to the surfactant, not glyphosate per se. The product has been determined to be toxicity category I for eye irritation, causing irreversible eye damage.

There may be a correlation between the ocular toxicity caused by the surfactant and reported incidents of ocular effects. This is supported by data obtained from the California Pesticide Illness Surveillance Program (1982-2000). In 1992 the glyphosate product was reformulated in the US to reduce the amount of surfactant which posed a hazard to the eye. From 1982 through 1991, there were 221 illnesses involving the eye or 22.1 cases per year. From 1994 (allowing 2 years for the product to be introduced into trade and widespread use) through 2000, there were 65 illnesses involving the eye or 9.3 cases per year, a decline of 58%. Therefore, these data support the finding that use of the reformulated glyphosate product since 1992, has resulted in a significant drop in illnesses. Overall, the total illnesses due to glyphosate declined by 39% from the 1982-1991 time period to the 1994-2000 time period, largely due to reduction in eye injuries.

The greatest potential for eye exposure is expected for workers mixing and loading the concentrated glyphosate product. There is also the potential for eye exposure as a result of entering treated fields after treatment to perform pruning or harvesting activities.

During an April 18 briefing, the Department of State agreed to supply the Agency with a full battery of the six acute toxicity tests on the tank mix. To date this information has not been received. Until such information is supplied to the Agency, EPA cannot evaluate the potential ocular or dermal toxicity resulting from direct contact with the tank mixture. Therefore, due to the acute eye irritation caused by the concentrated glyphosate formulated product and the lack of acute toxicity data on the tank mixture, the Agency recommends that DoS consider using an alternative glyphosate product (with lower potential for acute toxicity) in future coca and/or poppy aerial eradication programs.

EPA regulates pesticides under two statutes, the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) and the Federal Food, Drug, and Cosmetic Act (FFDCA). FIFRA provides the authority to register and review pesticides as well as the authority to suspend and cancel if use poses unreasonable risks. FFDCA provides authority to set maximum residue levels (tolerances) for pesticides used in or on foods or animal feeds.

Section 3 of FIFRA provides authority to register (license for sale and distribution) pesticide products. The label of the pesticide product specifies the use (pest and crop/site), amount of product to be applied, frequency, timing of use, restrictions, storage and disposal practices and precautionary statements. The active ingredient in a pesticide product is the "ingredient which will prevent, destroy, repel, or mitigate any pest." The inert or other ingredient(s) in a pesticide product is "an ingredient which is not active." The registrant must provide data for the Agency to assess potential environmental and human health risks. The data required to make a safety finding are dependent on the intended use, e.g., food use vs non-food use. The data requirements for pesticides may be found in 40 CFR Part 158. For human health risk assessment, data is required to permit characterization of hazard and exposure.

Data requirements on the chemical identity and composition of the formulated pesticide product, may be found in 40 CFR 158.150. The list of ingredients for a pesticide product and the percent of each ingredient in the formulation are contained in the confidential statement of formula (CSF). The CSF is FIFRA confidential business information (CBI) and is entitled to treatment as trade secret or proprietary information. Agency risk assessments do not typically contain this information. In finalizing the current document, FEAD and HED consulted with OPPs Information Resources and Services Division/Public Information and Records Integrity Branch regarding CBI. It was determined that the document did contain some CBI and therefore, some sections have been adapted.

Residue chemistry data required as per 40 CFR 158.240 support the ability of the Agency to estimate the amount of pesticide that will result in food as a result of application of the pesticide according to the product labels directions for use. The magnitude of the residue studies for crop field trials use the typical end use product as the test material. The livestock feeding studies are required whenever a pesticide residue will be present in livestock feed. The livestock feeding studies evaluate the magnitude of the resulting pesticide residue in meat, milk, poultry, and eggs. The studies are conducted with the technical grade of the active ingredient or the plant metabolites Residue chemistry data are also required to identify any potential metabolites of concern. These data are used to determine the tolerances for the parent and or metabolites. Additional data is required on environmental fate, degradation, metabolism, and dissipation.

Hazard data required for human health risk assessment are provided in 40 CFR 158.340. The use of the active ingredient (i.e., food use or non-food use) will determine what studies are required. The acute toxicity data on the technical grade of the active ingredient are used for classification and precautionary labeling for protective clothing requirements, and worker reentry intervals. The only studies that are required to be conducted on the manufacturing use product or end use product are the acute toxicity studies. The remaining toxicology studies (e.g., developmental toxicity, reproduction, subchronic, chronic feeding, or carcinogenicity studies) require that the test substance is the technical grade of the active ingredient. Subchronic toxicity studies provide data on potential target organ toxicity and are also used to select dose levels for long term or chronic toxicity studies. Chronic toxicity or carcinogenicity studies are conducted for food use chemicals to determine potential effects following prolonged or repeated exposure that may have a latency period for expression. The test animals are exposed orally for a significant portion of their life span. Developmental toxicity studies are required in two species (usually the rat and rabbit) for food use chemicals. They are conducted to detect alterations in the normal development of fetuses following in utero exposure. The 2-generation rat reproductive toxicity study is required to assess potential alterations in gonadal function, estrus cycles, mating, conception, birth, lactation, weaning, as well as growth and development of offspring. The Agency also requires a battery of mutagenicity studies to assess the potential induction of changes in the genetic material of cells. The above studies are required for food use active ingredients. In general, less data is required for non-food use active ingredients and inerts unless a concern has triggered additional testing.

The Agency conducts separate risk assessments for all pesticide active ingredients and has conducted risk assessments for some inerts. The remaining inerts are cleared by the Agency. It should be understood that whenever the inert ingredient was cleared, whenever the tolerance exemption was established, the inert met the standards of the time. Inert ingredients, also known as "other ingredients," are the carrier for the active ingredients which allow the product to deliver the active ingredient at a specific rate and ensure proper distribution during application. Currently there are over 3200 inert ingredients cleared by EPA for use in various domestic pesticides products. There are two major classifications: non-food use (such as lawn care products and bathroom cleaners), and food-use, which require an exemption from the requirement of a tolerance and can also be used in non-food products.

The Agency has a newly developed methodology for evaluating low or low/moderate toxicity chemical substances by way of a screening process that incorporates elements of a tiered approach (US EPA, May 2002). Use of this process will permit the Agency to clear more chemicals of low to moderate toxicity for use in pesticide products. The Agency is aware that some chemicals may be used as inert ingredients in some formulations and as active ingredients in other formulations. EPA believes this methodology is appropriate for evaluating some low toxicity chemicals regardless of whether they are categorized as active or inert ingredients. The new process will permit the Agency to be able to conduct more in-depth evaluations of other ingredients that are of potentially higher toxicity. Chemicals of higher toxicity that can not be appropriately addressed in the lower tiers would be evaluated in a manner substantially similar to that of an active ingredient. Later as the Agency begins to review chemical-specific or surrogate information in the open literature, the preliminary tier determination may be revised (US EPA, May 2002).

Inert ingredients that are exempt from tolerance are listed in 40 CFR 180.1001 (c). The inert ingredients in the glyphosate formulation have been approved by the Agency. The components of the adjuvant (Cosmo-Flux 411F), that DoS indicates have been sprayed on coca plants in Colombia have also been determined to be approved for use on food by the Agency.

The two federal statutes for regulating pesticides in the US give EPA limited authority to regulate the sale, or use of adjuvants in the US. EPA only has authority to regulate the pesticide product itself. For example, if a chemical in an adjuvant was intentionally included in the formulation of a pesticide product, the chemical would be regarded as an inert ingredient. In the US as with all countries, adjuvants are commonly used and added to pesticides as wetting agents, spreaders. emulsifiers, antifoamers, penetrants, or for other purposes. These may contain surfactants, solvents, or other types of chemicals to achieve the desired purpose.

An adjuvant is a subsidiary ingredient or additive product added to a pesticide in a mixture that aids the effectiveness of the primary or active ingredient. Adjuvants are most commonly added to tank mixes of pesticide products prior to application to the site to be treated. Adjuvants are not directly subject to FIFRA registration if no pesticidal claims are made. Pesticide manufacturers choose whether or not to address on their product labels the use of adjuvants with their product(s). However, when added to a tank mix for application to a food or feed crop/site, the individual components must be cleared under FFDCA. While adjuvant products are not registered on the federal level, they are subject to registration under some state laws. The states of Washington and California are two states that register adjuvants. The adjuvant (Cosmo-Flux 411F) used in the glyphosate tank mix is produced by a Colombian company and is not sold in the U.S. The Department of State has agreed to provide the Agency with acute toxicity data performed on the actual tank mix that has been sprayed in Colombia.

The glyphosate product used in Colombia according to the Department of State was registered in the US in April 1974. From 1974-1992 the product was registered for use on a number of agricultural and nonagricultural sites. The product had a "Warning" signal word for eye irritation. In 1992, the registrant submitted an eye irritation study that was categorized as Toxicity Category I and required a "Danger" signal word. The registrant decided that they did not want to market a glyphosate product with a "Danger" signal word. Around July 1992, the registrant registered a reformulated glyphosate product for use in the United States that had the percentage of surfactant reduced to a level that produced Toxicity Category II eye irritation. The original product was re-labeled - "Not for use in the United States". Because the Agency never rescinded the registrations for the use sites that were on the initial label before it was changed, the product technically remains registered for use on numerous agricultural and nonagricultural sites although it is not currently labeled for these uses. In November 2001, the Company submitted a label for the original product for ground and aerial application to control undesirable vegetation in nonagricultural sites. This registration was intended to register a glyphosate product that corresponds to the product being used in Colombia. In February 2002 the product name was changed and the maximum application rate was reduced as per the request of the registrant.

Hazard identification is the first step in the risk assessment process. The objective is to qualitatively characterize the inherent toxicity of a chemical. Scientific data are evaluated to establish a causal relationship between the occurrence of adverse health effects and exposure to a chemical. Because high quality controlled toxicology studies on humans are frequently unavailable, regulatory scientists rely on animal data to estimate hazard to support regulatory decision making. Prior to and subsequent to initial registration, the Agency has required the registrants of glyphosate products to submit appropriate studies according to contemporary study requirements and testing protocol requirements.

The available hazard data base on experimental animals indicates that glyphosate has low acute toxicity via the oral and dermal routes with LD₅₀s > 5000 mg/kg. It is a mild eye irritant and a slight dermal irritant. It is not a dermal sensitizer. The requirement for an acute inhalation study was waived since no respiratory or systemic toxicity was seen following subchronic inhalation exposure in rats. In the subchronic and chronic oral toxicity studies (1-year dog, 24-month mouse, 2-year chronic/carcinogenicity rat, and 2-generation rat reproduction), systemic toxicity manifested most commonly as clinical signs, decreases in body weight and/or body weight gain, decreased food consumption, and/or liver and kidney toxicity at doses equal to or above the limit dose (1000 mg/kg/day). No dermal or systemic toxicity was seen following repeated dermal exposures. There was no quantitative or qualitative evidence for increased susceptibility in fetuses following in utero exposure to rats and rabbits in developmental toxicity studies or following pre/post-natal exposure to rats in the 2-generation reproductive toxicity study in rats. Effects in the offspring were observed only at or above treatment levels which resulted in evidence of appreciable parental toxicity. Glyphosate was not mutagenic in a full battery of assays. Based on the lack of evidence for carcinogenicity in two acceptable studies in mice and rats, glyphosate is classified as a "Group E" chemical (no evidence of carcinogenicity to humans).

1. Polyoxyethylene alkylamine (POEA). POEA is a compound that is used as a surfactant with many glyphosate formulations. In a safety evaluation and risk assessment of glyphosate, the Roundup formulation and the surfactant POEA, Williams et al. (2000) reported that POEA can cause severe skin irritation and be corrosive to the eyes. In subchronic oral studies, POEA was mainly a gastrointestinal irritant in rats at high doses (~ 100 mg/kg/day) and in dogs at lower doses (30 mg/kg/day). In a developmental toxicity study in rats, POEA did not cause any developmental effects up to 300 mg/kg/day, but did induce maternal toxicity at 100 and 300 mg/kg/day (Farmer et al., 2000). The concentrated formulated Roundup product can also be strongly irritating to the eyes and slightly irritating to the skin (Williams et al., 2000).

2. (information not included as it may be entitled to confidential treatment) are substances that are not highly toxic by oral or dermal routes and are not irritating to the skin. They may cause mild, transient eye irritation. Many (information not included as it may be entitled to confidential treatment) are known not to be sensitizers (study citation not included as it may be entitled to confidential treatment). The molecular weight of a (information not included as it may be entitled to confidential treatment) determines its biological properties, and, thus, its toxicity. The lower molecular weight (information not included as it may be entitled to confidential treatment) tend to be more toxic than the higher-weighted (information not included as it may be entitled to confidential treatment) and are absorbed by the digestive tract and excreted in the urine and feces, while the higher molecular weight (information not included as it may be entitled to confidential treatment) are absorbed more slowly or not at all (study citation not included as it may be entitled to confidential treatment). (information not included as it may be entitled to confidential treatment) have low acute and chronic toxicity in animal studies. No significant adverse effects have been noted in inhalation toxicology studies, carcinogen testing, or mutagen assays. High oral doses have resulted in toxic effects to the kidneys and loose feces (study citation not included as it may be entitled to confidential treatment). Topical dermal application of (information not included as it may be entitled to confidential treatment) to burn patients with injured skin has resulted in toxicity (study citation not included as it may be entitled to confidential treatment).

Cosmo - Flux 411F (Adjuvant)

The Cosmo-Flux 411F adjuvant product used in the glyphosate tank mix is produced by a Colombian company and is not sold in the U.S. The Agency is not in possession of toxicity data from direct dosing of test animals with Cosmo-Flux 411F. However, the Agency has made a hazard assessment based on the toxicity of the individual components. As stated above, sale or use of spray adjuvant products in the U.S. are generally not regulated by EPA. However, the DoS has provided the EPA with a copy of this products label and a description of the product ingredients. To be able to provide an opinion on hazard characterization of the Cosmo-flux ingredients, the EPA relied on available technical information from various sources. Cosmo-Flux 411F consists mainly of (information not included as it may be entitled to confidential treatment) with a nonionic surfactant blend primarily composed of (information not included as it may be entitled to confidential treatment). All ingredients of this product are substances that are not highly toxic by oral or dermal routes. They may cause mild eye and skin irritation. All components of the adjuvant have been approved for use in/on food by EPA (40 CFR 180.1001, Letter from R.Forrest/EPA, to R.Woolfolk/DoS, 7/30/2001).

1. (information not included as it may be entitled to confidential treatment). The (information not included as it may be entitled to confidential treatment) can cause dermal and ocular irritation and, in high doses orally, can cause significant toxicity. However, small amounts are not a concern and these substances have been approved as food additives by the FDA and are exempt from tolerances by EPA on certain commodities.

2. (information not included as it may be entitled to confidential treatment). The other major component of Cosmo-Flux 411F, (information not included as it may be entitled to confidential treatment), is not considered highly toxic. It may cause mild eye and skin irritation. The corresponding (information not included as it may be entitled to confidential treatment), has low subacute, subchronic and chronic oral toxicity and is used as a direct food additive and a component in cosmetics. The higher molecular weight (information not included as it may be entitled to confidential treatment) is less likely to be absorbed orally or dermally and most likely of less toxicological concern. The other minor components, are not known to be highly toxic compounds and would not be of toxicological concern at the concentrations and conditions in which they are used.

Dose response analysis is the second step in the risk assessment process i.e.; characterization of the quantitative relationship between exposure (dose) and response based on studies in which adverse health effects have been observed. The objective is to identify endpoints of concern which correspond to the route and duration of exposure based on the exposure patterns.

HED selects doses and endpoints (effects of concern) for risk assessment via an internal peer review process. HED uses a standing Committee - the Hazard Identification Assessment Review Committee (HIARC), to consider the available hazard data (studies required to be submitted by registrants in 40 CFR part 158 and open peer reviewed literature) to identify endpoints for use in risk assessment.

Ideally, each safety study identifies a dose level that does not produce a biologically or statistically significant increased incidence of an adverse effect or no observable adverse effect level (NOAEL). The threshold dose is the smallest dose required to produce a detectable effect. Below this dose, there is no detectable response. Glyphosate

On March 26, 1998 and, again, on November 20, 2001 the HED HIARC met to examine the hazard data base and identify dietary endpoints for Females 13-50 years old, as well as the General Population, the chronic reference dose. The HIARC also considered toxicological endpoints for incidental oral exposure (on 20-NOV-01) appropriate in residential exposure risk assessments.

The most recent report of the HIARC for glyphosate has the complete assessment of the endpoints selected for dietary and residential/occupational exposures (W. Dykstra, 01/22/02; TXR# 0050428). OPP calculates acute (24 hour or single day) and chronic (continuous lifetime exposure) RfDs for the purposes of calculating dietary risk for food and drinking water. The RfD is calculated by dividing the appropriate no observed adverse effect level by a ten fold factor for interspecies variability ("average" human sensitivities might be up to 10 times that of lab animals) and a ten fold factor for intraspecies variability (i.e., some individuals within a population might be 10 times more sensitive than the "average" person).

For glyphosate, no endpoints were selected for the acute RfD since no hazard attributed to a single dose was identified from the oral toxicity studies, and there are no specific concerns for toxic effects on the developing fetus or infants and children. In addition, the HIARC did not identify endpoints of concern for dermal and inhalation exposures for any exposure period (short term- 1 to30 days, intermediate term- 1 to 6 months, or long term- 6 months to lifetime) since no hazard was identified due to the low toxicity of glyphosate (TXR No. 0050428, W. Dykstra, 22-JAN-2002). The chronic dietary RfD of 1.75 mg/kg/day was based on diarrhea, nasal discharge, and mortality in a rabbit developmental toxicity study. A summary of doses and toxicological endpoints selected for various relevant exposure scenarios are summarized in Table 1.

Table 1. Glyphosate Endpoint Selection Table

EXPOSURE SCENARIO	DOSE (mg/kg/day)	ENDPOINT	STUDY
Acute Dietary (24 hour or single exposure)	An effect of concern attributable to a single dose was not identified from the oral toxicity studies; there are no concerns for developmental or reproductive toxicity.
Chronic Dietary (continuous lifetime exposure)	NOAEL = 175 uncertainty factor (UF) = 100	Maternal toxicity based on clinical signs (diarrhea and nasal discharge)resulting in mortality of some dams at 350 mg/kg/day	Developmental toxicity -Rabbit
		Chronic RfD = 2.0 mg/kg/day
Incidental Oral, Short- (1-30 days), and Intermediate-(1-6 months) Term	NOAEL= 175	Maternal toxicity based on clinical signs (diarrhea and nasal discharge)resulting in mortality of some dams at 350 mg/kg/day	Developmental toxicity -Rabbit
Dermal, Short-, Intermediate-and Long-Term	No hazard was identified, therefore quantification of dermal risk is not required. No systemic toxicity was seen at the Limit Dose (1000 mg/kg/day) following repeated dermal applications to New Zealand White rabbits.
Inhalation, Short-, Intermediate-, and Long-Term	Quantification of inhalation risk is not required because 1) no hazard was identified in the 28 day inhalation toxicity study in rats - NOAEL = 0.36 mg/L (highest dose tested (HDT)); lowest observable adverse effect level (LOAEL) not established based on 6 hours/day, 5 days/week for 4 weeks and 2) due to the physical characteristics of the technical (wetcake), exposure to high levels of the active ingredient is unlikely via the inhalation route, so there was no purpose to test at higher doses.

On August 3, 1996 the FQPA amended FIFRA and FFDCA. Section 408(b)(2)(C) of the Federal Food, Drug, and Cosmetic Act addresses exposure of infants and children. Under this provision EPA must apply the default 10X safety factor when establishing, modifying, leaving in effect or revoking a tolerance or exemption for a pesticide chemical residue, unless the EPA concludes, based on reliable data, that a different safety factor would protect the safety of infants and children. Risk assessors, therefore presume that the default 10X safety factor applies and should only recommend a different factor, based on an individualized assessment, when reliable data shows that such different factor is safe for infants and children that it does not rely on a default value or presumption in making decisions under Section 408 where reliable data are available that support an individualized determination. The OPP FQPA Safety Factor Committee (SFC) makes specific case-by-case determinations as to the need and size of the additional factor if reliable data permit. Determination of the magnitude of the overall safety factor or margin of safety involves evaluating the completeness of the toxicology and exposure databases and the potential for pre- or post-natal toxicity. Individualized assessments may result in the use of additional factors greater or less than, or equal to 10X, or no additional factor at all. (OPP Guidance Document on Determination of the Appropriate FQPA Safety Factor(s) in Tolerance Assessment, 2002)

The HIARC addressed the potential enhanced sensitivity of infants and children from exposure to glyphosate as required by the FQPA of 1996 at the March 26, 1998 meeting and reaffirmed the decision at the November 20, 2001 meeting. The HIARC concluded the following:

• Based on the available data, there was no evidence of quantitative and qualitative increased susceptibility to in utero and/or postnatal exposure to glyphosate in rats or rabbits.
•  Based on a weight of evidence consideration, the HIARC decided not to require the conduct of a developmental neurotoxicity study with glyphosate to evaluate the potential for developmental neurotoxic effects because there was no evidence of neurotoxicity and neuropathology in adult animals.

The FQPA SFC met on April 6, 1998 to evaluate the hazard and exposure data for glyphosate. The FQPA SFC concluded that the safety factor of 10x be removed (reduced to 1x) since there is no evidence of quantitative or qualitative increased susceptibility of the young demonstrated in the prenatal developmental studies in rats and rabbits and pre/post natal reproduction study in rats. In addition the toxicology data base is complete, a developmental neurotoxicity study is not required, and the dietary (food and drinking water) exposure assessments will not underestimate the potential exposures for infants and children.

The exposure assessment is the third step in the risk assessment process. The objective is to determine the source, type, frequency, magnitude, and duration of actual or hypothetical contact by humans with the agent of interest. To conduct this assessment EPA relied upon the information provided by DoS from two sources: (1) Department of State (DoS) Presentation, DoS Coca Eradication Program, 4/18/02, (2) DoS document entitled Chemicals Used for the Aerial Eradication of Illicit Coca in Colombia and Conditions of Application. These data were used in accordance with standard policies and procedures used by the Agency in conducting pesticide exposure assessments.

Acute dietary exposure is possible for persons consuming livestock or food crops which have been inadvertently sprayed as a result of the aerial eradication program in Colombia. However, since glyphosate is a contact herbicide that systemically kills plants after absorption through leaves, dietary exposure due to consumption of treated crops is expected to be limited. Since a coca field is sprayed no more than twice to eradicate the crop, no chronic food exposure is expected. Based on an evaluation of the hazard database, the Agency did not identify a toxic effect attributed to a single oral dose. Therefore, an acute dietary risk assessment was not performed (TXR No. 0050428, W. Dykstra, 22-JAN-2002). No significant risk due to dietary exposure to glyphosate residues is expected.

The tank mixture sprayed for eradication of coca in Colombia contains 55% water, 44% of glyphosate herbicide product, and 1% adjuvant (Cosmo-Flux 411F). No more than two applications of the glyphosate tank mixture are sprayed over coca crops at a maximum of 1.25 gallons/Acre (equivalent to 1.1 gallons/Acre of glyphosate product, 0.03 gal/Acre of Cosmo-Flux 411F, and 0.12 gal/Acre of water). DoS also stated that the average field size for coca in Colombia is 3-5 hectares (approximately 7-12 acres). The program for aerial eradication of coca treats a maximum of 1000 Acres/day, during 3-5 missions/day.

Exposure is expected for workers mixing and loading the glyphosate formulated product and tank mix, and applicators applying the pesticidal mixture via fixed-wing aircraft. Mixers, loaders, and applicators (handlers) have the potential for dermal exposure to the concentrate glyphosate formulated product or tank mix from droplets contacting the skin. There is also the potential for inhalation exposure to the concentrated glyphosate formulated product or mixed formulation from breathing in aerosolized spray droplets.

According to the DoS, the mixer/loaders are trained on the label requirements for handling the chemicals in the spray mixture, first aid, and use of personal protective equipment (PPE). The required PPE according to the label includes long-sleeved shirts and long pants, waterproof gloves, shoes and socks, and protective eyewear. PPE is expected to mitigate potential exposure to handlers. Exposure to handlers is anticipated for short-term (1-30 days) durations. There also may be the possibility for intermediate-term(1-6 months) handler exposure for individuals mixing, loading, and applying the glyphosate mixture to multiple fields for more than 30 days. However, the Agency does not have information pertaining to the duration of coca spray programs or number of days spent mixing, loading, and applying the glyphosate mixture.

An occupational handler exposure and risk assessment is required for an active ingredient if: (1) certain toxicological criteria are triggered and (2) there is potential exposure to handlers (i.e., mixers, loaders, applicators, etc.) during use. Upon review and analysis of the hazard database in total, the Agencys HIARC did not identify a hazard of concern for dermal or inhalation short- and intermediate-term exposures. Therefore, quantitative estimates of risk for short-term dermal and inhalation have not been conducted (TXR No. 0050428, W. Dykstra, 22-JAN-2002). No significant handler risk is expected.

Post-application Exposure

According to the DoS, Colombian coca plants (Erythroxylum species) are woody perennial shrubs native to the Andean region. Coca plants have leaves with waxy cuticles which retard herbicide uptake in the plant. The coca bushes grow to approximately chest level and are harvested mainly by leaf pulling, 4 to 5 times per year. Coca plants grow from seedlings to a harvestable plant in 12 to 18 months. Representatives from DoS indicated that, growers will prune the coca plants, immediately after spraying, in order to salvage the coca crop. Specifically, since glyphosate is a contact herbicide that works systemically to kill the plant after absorption through the leaves, workers may enter fields immediately after spraying in order to prune or pull off the coca leaves in order to prevent the coca plant from dying. In the US, most uses of glyphosate are applied to kill weeds - which are the target. In general, glyphosate is not applied in the US to destroy or kill the raw agricultural commodity. The intended US uses are for undesired vegetation in and around crop fields, forests, industrial areas and residential areas (for more detailed information, please refer to the June 28, 2002 memorandum entitled Description of Glyphosate Use in the U.S. for Comparison to Use in Colombia for Coca Eradication from Virginia Werling and Timothy Kiely to Jay Ellenberger).

DoS states that pilots are instructed not to spray fields where people are present. Therefore, based on the use pattern described by the DoS, potential short-term dermal exposures are expected for persons pruning, or leaf pulling treated coca plants immediately after spray events. These activities are expected to result in dermal exposure from treated foliage contacting the skin. In cases such as glyphosate, where the vapor pressure is negligible, HED experience with post-application data suggests that inhalation exposure is minimal and therefore, HED does not quantitatively assess post-application inhalation exposure. Intermediate- and long-term post-application exposures are not expected due in part to the fact that a coca field is sprayed no more than twice.

Additionally, glyphosate is a translocated herbicide which is rainfast (unable to be rinsed off by water) within 48 hours after spraying. Therefore, potential exposure to dislodgeable residues of glyphosate after 48 hours is expected to be minimal. Glyphosate has no residual soil activity. Results from the first 12 months of bareground field dissipation trials from eight sites show that the median half-life (DT50) for glyphosate (Roundup) applied at maximum annual use rates (7.95 lb a.i./acre, 10.7 lb a.i./acre) was 13.9 days with a range of 2.6 (Texas) to 140.6 (Iowa) days. Acceptable aerobic soil, aerobic aquatic and anaerobic aquatic metabolism studies demonstrate that under those conditions at 25^oC in the laboratory glyphosate degrades rapidly with half-lives of approximately 2, 7 and 8 days respectively. The reported half-lives (DT50) from the field studies conducted in the coldest climates, i.e. Minnesota, New York. and Iowa, were the longest at 28.7, 127.8, and 140.6 days respectively indicating that glyphosate residues in the field are somewhat more persistent in cooler climates as opposed to milder ones (Georgia, California, Arizona, Ohio, and Texas) (Memo, J.Carleton, 10/26/98, D238931). The climate in Colombia would favor a shorter half life than the colder regions of the US. Thereby, HED believes glyphosate would not be persistent or be available for intermediate-term or long-term post-application exposures in the Colombian climate.

A post-application exposure and risk assessment is required for an active ingredient if: (1) certain toxicological criteria are triggered and (2) there is potential exposure. Upon review and analysis of the hazard database in total, the Agencys HIARC did not identify a hazard of concern for these durations or routes of exposure. Therefore, quantitative estimates of risk for short-term dermal and inhalation have not been conducted (TXR No. 0050428, W. Dykstra, 22-JAN-2002). No significant post-application risk due to glyphosate exposure is expected as a result of this use.

Since DoS states that pilots are instructed not to spray fields where people are present, incidental oral exposure (hand-to-mouth) resulting from being directly sprayed by glyphosate was not assessed. Non-dietary incidental oral exposure was not quantitatively assessed for the use of glyphosate in Colombia.

As a point of comparison, screening level risk estimates for toddler incidental oral exposures (hand-to-mouth) to the U.S. for registered residential turf uses of glyphosate have been calculated (D280831, Memo, W.Donovan, 20-FEB-2002). All resulting risks for toddler incidental oral exposure do not exceed HEDs level of concern. The assumptions for toddler incidental oral exposures, (based on the maximum application rate of 1.62 lbs acid equivalent (ae)/Acre), are expected to be conservative. For example, it is assumed that there is no dissipation of transferable residues, so that toddlers are exposed to day of treatment residues for each day of exposure. Even though the application rate for the coca eradication program is higher (3.3 lbs ae/Acre), using the same standard screening level assumptions as used in the residential assessment for the U.S. registered turf use and taking the higher application rate into account, the potential risk would not exceed HEDs level of concern.

As indicated in the turf assessment, glyphosate was directly applied to residential lawns and did not result in exposures of concern to HED. Although spray drift is always a potential source of exposure to residents nearby aerial spraying operations, AgDrift (a spray drift model) consistently predicts drift from applications is only a fraction of the applied rate (lb ai/acre). Based on this assessment, HED believes that it is unlikely that there is a higher potential for risk of exposure to spray drift from agricultural operations.

The greatest potential for eye exposure is expected for workers mixing and loading the concentrated glyphosate product. Potential exposure is expected through 2 main pathways: (1) exposed hands transferring the glyphosate product to the eye(s), (2) splashing of the liquid concentrated glyphosate product into the workerseye(s). However, the label requires mixer/loaders to wear protective eyewear and this level of PPE is expected to mitigate the potential for eye exposure.

There is also the potential for eye exposure as a result of entering treated fields immediately after treatment to perform pruning or harvesting activities. Specifically, persons dermally contacting treated foliage may transfer residues from the hand to the eye. However, the Agency currently does not have a defined method or model to assess quantitative eye exposures resulting from occupational or residential exposures to pesticides. For products registered for use in the United States which have high acute toxicity to the eye, mitigation of exposure to potential eye effects for post-application workers is done by lengthening restricted entry intervals (REI).

Due to spray drift, there is potential exposure for persons in areas near those targeted for spraying. Exposure through drift is not expected to exceed that which is identified in the exposure characterization provided above. The coca eradication program described by the Department of State has incorporated several features designed to minimize the potential for off-target drift, provide quality assurance on a mission-by-mission basis, and evaluate the performance of the program to the extent possible given current conditions. Three types of aircraft are used in the program including the Ayres Corporation T65 Thrush, modified OV10D Bronco aircraft converted from military observation use to spray aircraft, and the Air Tractor AT802. The T65 and AT802 are common to the agricultural sector in the United States. The nozzles are Accu-Flow as described at the April 18, 2002 briefing to the Agency. The droplet spectra characteristics, under use conditions for these nozzles, produce a very large droplet which has a volume median diameter (VMD) between 300 and 1500 microns. Use of droplets this size is consistent with minimizing spray drift in agriculture in the United States. A surfactant (Cosmo-Flux 411F) is also used in the spray solution along with water and the glyphosate formulated product. The use of spray adjuvants (in this case Cosmo-Flux 411F) in pesticide product formulations and/or the spray solution is also consistent with common agricultural practices in the United States.

The quality assurance standard operating procedures identified as incorporated into the program are also consistent with standard agricultural practices. These include reconnaissance of the spray sites, use of global positioning satellite technology (GPS), and criteria for aborting missions (e.g., based on climatological conditions or presence of persons or livestock in the treatment areas). Reconnaissance of spray sites is intended to define the treatment zones through the use of sophisticated GPS mapping which is then overlaid with GPS spray records from missions to evaluate performance. GPS technology is used for planning, assessments of mission performance, and for archival purposes to evaluate potential claims against the program.

Finally, to a limited extent where feasible, DoS reports that on-site ground inspections for spray efficacy and potential adverse effects are performed. Reports suggest approximately 90 percent efficacy in the spray swath and minimal collateral damage to surrounding vegetation (e.g., aerial photos of treated areas) based on information supplied by the DoS at the April 18, 2002 briefing.

The Agency did not complete a quantitative risk analysis of the drift potential of glyphosate in the water/surfactant solution used in this program. However, the technology and other safeguards used in this program are consistent with common approaches in the United States for reducing spray drift. Therefore, it is likely that drift is minimized in this program if all procedures are adhered to and operational equipment is in working order. At the April 2002 briefing, it was indicated to the Agency that quantitative spray drift studies had been completed by the DoS in conjunction with the University of Georgia. These were not supplied to the Agency nor were they considered in this evaluation. Additionally, it should be noted that the Agency did not review the primary source of information provided (e.g., the method by which the VMD was determined was not described, written application protocols describing target site conditions when applications would be aborted were not provided, and methods for scoring or measuring off-target damage were not provided).

Dr. Jerry Blondell is a health statistician and the point of contact for human incident data in the EPA Pesticide Program. He has reviewed the poppy incident data from Colombia and compared these data to the glyphosate incident data reported from California and the Poison Control Center. The entire review can be found beginning on page 38 of this document.

The report, prepared by the Department of Narino, Municipality of El Tablon De Gomez, makes a concerted effort to identify any health problems that might be related to use of the glyphosate tank mix in aerial eradication programs. The study was commissioned by the U.S. Embassy in Bogota and conducted independently by Dr. Camilo Uribe, Director of Clinica Uribe Cualla, the national poison control center. Sections of this report are summarized below with the sections numbered in bold corresponding to the original report.

An exact comparison of the epidemiological data in Colombia (which is from aerial application to poppy) relative to the conditions of use, presented at the April 18, 2002 briefing (for aerial application to coca) by DoS to OPP risk assessors, would have limitations and uncertainties. The briefing did not address the conditions of use for poppy. DoS also did not provide human incident data for the coca eradication program. Subsequent to the April 18 briefing HED received an e-mail communication from OPP/ Field and External Affairs Division, stating that the application rate for poppy was lower than that for coca. According to the DoS, the use pattern of the glyphosate mixture on poppy also differs from the use on coca. Other details of the differences between the two spray programs have not been supplied to the Agency. Specifically, the Agency has no information as to the exact makeup of the tank mixture sprayed on poppy, or whether the same glyphosate product and adjuvants used in the coca eradication program were used in the poppy eradication program. Therefore, generalized conclusions drawn from human incident data as a result of application to opium poppy, in comparison to conditions of use for the coca eradication program should be made with caution.

This report primarily concerns the area around the municipality of El Tablon in southern Colombia. The total population is given as 16,770, of which 89% is categorized as rural. The main crops in this area include coffee, corn, wheat, oats, potatoes, and illicit opium poppy. It is known that a variety of other pesticides, more toxic than glyphosate, are used on these crops. The municipality has three health centers, including Aponte, which is the focus of this report. The Aponte health center is staffed by a medical doctor, a nurse, and a nurses aide. Aerial eradication of the illicit opium poppy reportedly occurred in this region in June, July, and November of 2000.

The Narino Departmental Health Institute provided summary morbidity and mortality information for the El Tablon De Gomez area and the Aponte settlement for the year 1999. Data for the year 2000 had not yet been officially released, but estimates are provided. These data are reported here to provide an approximate description of glyphosate tank mix exposure upon use on coca fields in Colombia. However, no quantitative conclusions can be drawn from these data. Six illnesses likely to be related to pesticide exposure were identified and tabulated. They include, acute diarrhea, acute respiratory infection, dermatitis, intoxication, conjunctivitis and headache. The authors note that the first three illnesses listed (diarrhea, respiratory infection, and dermatitis) are likely to be related to problems with inadequate nutrition, housing, and lack of health services. The basis for this listing of symptoms is not specified, but it does agree with the list of symptoms likely to result from exposure to glyphosate products based on Poison Control Center data, California surveillance reports, and the world literature. Total morbidity for 1999 and estimated morbidity for 2000 are given in the Table below for El Tablon De Gomez and the Aponte Settlement below. Note, however, that the overwhelming majority of these illnesses did not occur at the time of spraying and, therefore, could not be related to spray exposure.

The Aponte settlement is contained within the El Tablon De Gomez area, where there has been a concern for herbicide spraying-related health effects. Figures in the report are listed by five separate age groups. This reveals that the majority of the cases of diarrhea and respiratory infection occurred in children less than five years old, as would be expected given known demographics of those health effects. Nationwide data show that 53% of intoxications are suicides or suicide attempts, but it is not clear how many of the four poisonings listed above might be suicidal or, more importantly, are due to other products such as medications. In both Tables 2 and 3 there is an increase of 5% from 1999 to the estimate for 2000 for the total of the six suspected illnesses. Given that spraying is reported to have occurred in 2000 and not 1999, this suggests that the overwhelming majority (95%) of illnesses reported would be background incidence unrelated to the spraying of herbicide. The remaining 5% increase could be due to a variety of causes and do not support a conclusion that the glyphosate tank mixture was responsible for these complaints.

In addition to the summary of general morbidity in the population, there is a mandatory health reporting system in Colombia for 34 illnesses including pesticide poisonings. The review of these records found no reports of pesticide poisoning for the municipality of El Tablon in the year 2000 or the first 9 weeks of 2001. Weekly reports were examined to determine how many pesticide poisonings were reported each month. It did not appear that the times of spraying correlated with reports of pesticide intoxication.

Out of a total of 125 reported pesticide poisonings in 61 weeks, 15 occurred during 5 weeks when spraying eradication occurred. Given the variation in the data, this could easily be due to chance and be unrelated to exposure from the spraying of the glyphosate tank mixture. More work is required to determine whether locations of the 15 suspect poisoning matched the location and timing of spraying.

In 2000, the Narino Department of Health requested all municipalities to report the human health effects of pesticide spraying. Ten municipalities supplied the reports. They are:

Three municipalities including Tablon de Gomez, Barbacoas, and Magui reported no cases. However, the reports were completed prior to the November spraying in Barbacoas and Magui and prior to (or perhaps during) the July and before the November spraying in Tablon de Gomez. Buesaco reported one patient with sore throat, numbness in limbs, and conjunctivitis in June. In Tumaco, six case of patients with conjunctivitis and dermatitis were reported as of October 6, 2000. In San Pablo, 50 cases of dermatitis, conjunctivitis, respiratory conditions, and digestive problems were reported after as of October 6, 2000.

In La Cruz, two cases of allergic rhinitis, two cases of dermatitis, and five cases of conjunctivitis were reported as of October 6, 2000. San Jose de Alban did not report any specific cases, but the scientific coordinator and chief nurse noted an increase in gastrointestinal, dermatological and respiratory conditions. The exact quantity of these conditions in relation to spray times was not given. El Rosario reported five cases of conjunctivitis and rhinitis that might have been related to spraying carried out on July 31. San Pedro de Cartago reported an increase in gastrointestinal symptoms but no quantitative relationship between illnesses and spray times was provided.

The absence of any reports of pesticide poisoning combined with the information from the ten municipalities is difficult to interpret. The glyphosate formulated product is known to cause irritation to the skin, eyes, mucous membranes which may account for some of the reports of sore throat, conjunctivitis, dermatitis and other conditions described above. However, it is not possible to evaluate these reports in any detail due to the lack of any information on how many of these cases experienced exposure immediately prior to their illness and lack of information on investigation of potential alternative causes. This anecdotal information does not provide any substantial evidence of health effects due to the spraying of the glyphosate tank mixture in Colombia. Many of the reports are consistent with exposure to glyphosate products by the dermal route, as reported in California and the literature. So, it is possible that some cases could be related to the aerial eradication program.

To provide context for comparison, the California Pesticide Illness Surveillance Program (1982-2000) data for glyphosate were reviewed for this risk assessment. Starting in 1992, the glyphosate product was reformulated in the US to reduce the amount of surfactant which posed a hazard to the eye. From 1982 through 1991, there were 221 illnesses involving the eye or 22.1 cases per year. From 1994 (allowing 2 years for the product to be introduced into trade and widespread use) through 2000, there were 65 illnesses involving the eye or 9.3 cases per year, a decline of 58%. Therefore, these data support the finding that the reformulated glyphosate product used since 1992, have resulted in a significant drop in illnesses. Overall, the total illnesses due to glyphosate declined by 39% from the 1982-1991 time period to the 1994-2000 time period, largely due to the reduction in eye injuries.

A commission visited the municipality of El Tablon on January 22, 2001 and spoke with Dr. Tordecilla and reviewed health records of his patients. A number of records of skin conditions were noted for the months of October, December 2000, and January 2001. The exact number of cases, selection criteria, and method of analysis was not specified in the summary report. Nevertheless, the commission concluded "that the information available permitted the commission to consider only the possibility of an association between exposure to pesticides and the effects". The commission noted that it lacked the technical expertise, the data on dates and locations of spraying, and therefore could not conclude whether the observed conditions were related to pesticide exposure.

Employees of the Narino Department Health Institute were interviewed by Colombian authorities. A Fatima Health Promoter thought children were most affected, suffering gastrointestinal problems and eye irritation. One possible route of exposure was the village water fountains which supply some of the drinking water. The most common symptoms in children, according to the Health Promoter, were stomach aches and vomiting, which were different from the most common symptoms of glyphosate exposure reported by Lee et al. (2000), sore throat and nausea. This inconsistency suggests that some cause other than glyphosate products was responsible for the childrens complaints. The Health Promoter reported one case of a boy with skin lesions like sores after the spraying. The Health Promoter was particularly concerned that peasants receive more health care from the government.

A nurses aide reported that three or four patients with burning eyes, headache, and dizziness were seen at her health center. One boy with a respiratory infection was sent to another health center, later died. Medical records were sought to substantiate this report but there was no clinical history, autopsy or other information to support glyphosate spraying as a factor. She referred a patient with urinary problems to the hospital. Subsequent review of the medical records of this case did not find reference to glyphosate tank mix exposure and suggested an infectious origin. There were also cases of dermatitis, headache, abdominal pain and gastrointestinal symptoms, but she could not say whether the symptoms were related to exposure to the spraying of glyphosate tank mixture.

Another nurses aide reported by telephone that her impression was that the number of dermatological consultations had increased. However, there was no clear association with glyphosate tank mix exposure and many of the reasons for the consultations were the same as in previous years when glyphosate was not used, so no clear relationship between the spraying and these dermatological conditions was identified.

Reports of anecdotal evidence by nurses aides and the health promoter have not established a link between the spraying of glyphosate tank mix and health effects. Follow-up to determine the timing and evidence of exposure and examination of other potential causes of these effects was not performed. These interviews do not add significant evidence about the health risks from the use of glyphosate tank mixture in Colombia.

There were 29 cases reported by Dr. Tordecelli and clinical records were obtained for 21 of them. Two other reports of skin lesions were sought but could not be confirmed. After careful review of the 21 records, it was determined that all but four cases were likely due to other causes. Most had skin conditions known to be related to bacteria or parasites, not chemical exposures and the onset of their symptoms did not correspond with the times of spraying. There were seven patients whose symptoms started after spraying and three of these were conditions known to be caused by bacteria or parasites. For the remaining four cases possibly related to the spraying of glyphosate tank mixture, one was an allergic reaction that had been seen in this patient before when there was no spraying. A second and third case were contact eczema that is endemic in this region and thought to be more likely due to an infectious origin. One of these two cases did not initiate until 52 days after the last spraying. The fourth case was dermatitis on the thigh which would typically be protected by clothing and thereby protected from aerial spray applications. This reviewer agrees with the conclusion that "the twenty-one clinical histories . . . reveals that any relationship between aerial eradication with the herbicide glyphosate (tank mixture) and the skin conditions treated in Aponte is unlikely".

In summary, the evidence collected and presented in this report cannot confirm that the glyphosate tank mixture used in Colombia as the likely cause of illness in the surrounding community. There is suggestive evidence in the form of reported increases of morbidity and reports from municipalities that some cases of relatively mild complaints could have occurred in relation to the spraying eradication program. Some of the reports appear to be similar to those reported in the literature and by California. These cases report irritation to skin, eyes, and respiratory passages and suggest that the Cosmo-Flux 411F added to the glyphosate product in Colombia has little or no effect on the overall toxicity of the formulated product.

Rather than review incomplete medical records, it would be better to collect information prospectively. For example, if pesticide poisoning is a mandatory reporting condition, a form documenting the exposure, health effects and medical data on each case could be designed and used to establish whether any particular conditions might be related to spraying the glyphosate tank mixture. Without prospective collection of data and follow up, it is difficult to evaluate potential health effects of the glyphosate tank mixture sprayed in Colombia. Better records of the time of exposure relative to the onset of symptoms would also enhance interpretation of the incidence data.

Risk characterization combines the assessments of the first three steps to develop a qualitative or quantitative estimate of the probability, that under the assumed conditions or variables of the exposure scenario, that harm will result to an exposed individual. Risk is equal to hazard multiplied by exposure. For the scenarios that are relevant to the subject use, the Agency has not identified toxic effects attributable to a single oral exposure, short- or intermediate-term dermal, or short- or intermediate-term inhalation exposures (TXR No. 0050428, W. Dykstra, 22-JAN-2002). Therefore, no quantification of exposure or risk was performed. Nonetheless, it is appropriate to qualitatively characterize the potential for risk concerns for this use.

From the review of glyphosate product incident reports for the use on poppy, it should be emphasized that the spraying reported to have occurred in 2000 and not in 1999 suggests, that the overwhelming majority (95%) of the illnesses reported would be background incidents unrelated to the spraying of herbicide. The remaining 5% increase could be due to a variety of causes and do not support a conclusion that the spraying of the glyphosate tank mixture was responsible for these complaints. Furthermore, the individual with the highest potential for exposure would be the mixer loader. They are handling the concentrated glyphosate product and the tank mix. The incident data that has been submitted to the Agency by DoS, does not include any incident reports for those individuals. There is some data to suggest that the poppy eradication program could have resulted in minor skin, eye, or respiratory irritation, and perhaps headache or other minor symptoms. However, the detailed information on the use, timing of application, history of exposure, and medical documentation of symptoms related to exposure to glyphosate tank mix were not available. The evidence collected and presented in the epidemiology report cannot confirm that the glyphosate tank mixture used in Colombia as the likely cause of a single illness. There is suggestive evidence in the form of reported increases of morbidity and reports from municipalities that some cases of relatively mild complaints could have occurred in relation to the spraying eradication program. Some of the reports appear to be similar to those reported in the literature and by California. These cases report irritation to skin, eyes, and respiratory passages and suggest that the Cosmo-Flux 411F added to the glyphosate product in Colombia has little or no effect on the overall toxicity of the formulated product. The information so far collected indicates that any increase in health problems is likely to be relatively small at most and the severity of those symptoms is likely to be minor to moderate at most. The Amazon Alliance and Earth Justice submission provided little, if any, information on the number of persons affected, age and sex, symptoms of illness, or diagnosis or treatment received. Without such information EPA cannot even begin to characterize the extent and pattern of the health effects claimed to result from glyphosate application. Given the limited amount of documentation, none of the data in the report from Colombia provide a compelling case that the spraying of the glyphosate mixture has been a significant cause of illness in the region studied. Prospective tracking of reports of health complaints, documenting times of exposure and onset of symptoms, are recommended during future spray operations to evaluate any potential health effects and ameliorate or prevent their occurrence.

The glyphosate formulated product used in the coca eradication program in Colombia contains the active ingredient glyphosate, a surfactant blend, and water. The acute toxicity test of the glyphosate technical is classified as category III for primary eye irritation and category IV for acute dermal and oral toxicity, and skin irritation. It not a dermal sensitizer. However, the surfactant used in the formulated product reportedly can cause severe skin irritation and be corrosive to the eyes, as would be expected for many surfactants. The label for the formulated product used in the coca eradication program in Colombia includes the "Danger" signal word. The product has been determined to be toxicity category I for eye irritation, causing irreversible eye damage. Some of the findings reported in the incident data are in alignment with that, reports of toxicity to the eye due to the surfactant, not glyphosate per se. This is supported by data obtained from the California Pesticide Illness Surveillance Program (1982-2000). As stated previously, in 1992 the glyphosate product was reformulated in the US to reduce the amount of surfactant which posed a hazard to the eye. From 1982 through 1991, there were 221 illnesses involving the eye or 22.1 cases per year. From 1994 (allowing 2 years for the product to be introduced into trade and widespread use) through 2000, there were 65 illnesses involving the eye or 9.3 cases per year, a decline of 58%. Therefore, these data support the finding that the use of the reformulated glyphosate product since 1992, has resulted in a significant drop in illnesses. Overall, the total illnesses due to glyphosate declined by 39% from the 1982-1991 time period to the 1994-2000 time period, largely due to the reduction in eye injuries.

The acute toxicity of the undiluted glyphosate product is most pertinent to mixers and loaders, who are potentially exposed to that form of the glyphosate product. On April 18, 2002, during a consultation with the DoS, in preparation for the current risk assessment, the DoS agreed to supply the Agency with a full battery of the six acute toxicity tests on the tank mix. To date, the Pesticide Program has not received this data. Until such information is supplied to the Agency, EPA cannot evaluate any potential acute toxicity effects resulting from direct contact with the tank mixture. Therefore, due to the acute eye irritation caused by the concentrated glyphosate formulated product and the lack of acute toxicity data on the tank mixture, the Agency recommends that an alternative glyphosate product (with lower potential for acute toxicity) be used in future coca and/or poppy aerial eradication programs.

A direct comparison of the epidemiological data in Colombia (which is from aerial application to poppy) to the conditions of use, (as presented at the April 18, 2002 briefing for aerial application to coca by DoS to OPP risk assessors), would be limited. The briefing did not address the conditions of use for poppy. Subsequent to the April 18 briefing HED received an e-mail communication from OPP/ Field and External Affairs Division, stating that the application rate for poppy was lower than that for coca. According to the DoS, the use pattern of the glyphosate mixture on poppy differs from the use on coca. Other details of the differences between the two spray programs have not been supplied to the Agency. Specifically, the Agency has no information as to the exact makeup of the tank mixture sprayed on poppy, or whether the same glyphosate product and adjuvants used in the coca eradication program were used in the poppy eradication program. The Agency also has questions as to the geographical area differences, the frequency of repeated applications, and the size of the area treated on each spray mission. Therefore, generalized conclusions drawn from human incident data as a result of application to opium poppy, in comparison to conditions of use for the coca eradication program should be made with caution.

In summary, HED concludes that:

• There are no risks of concern for glyphosate, per se, from the dermal or inhalation routes of exposure, since toxicity is very low.
• The identified components of the adjuvant Cosmoflux 411F are not highly toxic by the oral and dermal routes; they have been approved for use in/on food by the Agency.
• Glyphosate is not highly toxic. Based on the conditions of glyphosate use described by DoS, there is likely minimal exposure or concern for acute and chronic dietary or incidental oral risks.
• The incident data from Colombia based on the poppy use may differ from use of glyphosate as part of the coca eradication program, so conclusions should be made with caution.
• There is concern for acute eye toxicity because of an inert ingredient present in the glyphosate formulated product used to treat coca. The potential for eye effects is primarily for mixers/loaders of the concentrated glyphosate product, which should be mitigated by protective eye wear which DoS states is being used.
• Due to the acute eye irritation caused by the concentrated glyphosate product and the lack of acute toxicity data on the tank mixture, the Agency recommends that DoS consider using an alternate glyphosate product in future coca and/or poppy aerial eradication efforts.

REFERENCES

HIARC Report for Glyphosate (TXR No. 0050428, W. Dykstra, 22-JAN-2002)

Glyphosate in/on Pasture and Rangeland Grasses, Roundup Ready Wheat, and Nongrass Animal Feeds. (DP Barcode: D280831, 20-FEB-2002)

Farmer, D.R., T.A. Kaempfe, W.F. Heydens and W.R. Kelce. 2000. Developmental toxicity studies with glyphosate and selected surfactants in rats. Teratology 61(6): 446.

US Environmental Protection Agency, Office of Pesticide Programs May 9, 2002: Guidance Document on Methodology for Determining the Data Needed and the Types of Assessments Necessary to Make FFDCA Section 408 Safety Determinations for Lower Toxicity Pesticide Chemicals.

Williams, G.M., R. Kroes and I.C. Munro. 2000. Safety evaluation and risk assessment of the herbicide Roundup and its active ingredient, glyphosate, for humans. Reg. Toxic. Pharm. 31: 117-165.

Certain references deleted for protection of CBI.

SECTION 3. Review of Glyphosate Incident Reports with special reference to aerial spraying in Colombia

On May 8, 2002 the U. S. Department of State requested that the U.S. Environmental Protection Agency provide consultation on the U.S.-supported aerial coca eradication program in Colombia. Specifically, the Department of State requests advice on whether the aerial application program may pose unreasonable risks or adverse effects to humans or the environment. This review is part of a health risk assessment performed by the Health Effects Division of the glyphosate product used in Colombia. This review will focus on reports of human health effects reported from the leading pesticide poisoning surveillance data sources in the United States, which include Poison Control Centers and the California Pesticide Illness Surveillance Program. The California data source is especially useful for this review because of its high quality, documentation going back to 1982, and because glyphosate is the second most widely used pesticide in California affording ample opportunity for unintentional exposures. The world scientific literature on glyphosate and a report from Colombia examining reports of the adverse health effects are also reviewed.

The aerial spray mixture used in Colombia consists of water, glyphosate formulation, and 1 percent Cosmo-Flux 411F. According to documentation supplied by the Department of State, this diluted mixture is applied to coca at the rate of 2.53 gallons per acre (U. S. Department of State 2002). "The commercial glyphosate formulation used in the spray mixture is registered with U. S. Environmental Protection Agency (EPA) for sale in the United States for non-agricultural use and contains 41 percent glyphosate salt and 59 percent inert ingredients. Approximately three fourths of the inert ingredient content are water and the remainder is a surfactant blend. A surfactant is essentially a soap that enhances the ability of the herbicide to penetrate the waxy cuticle of the leaf surface."(U. S. Department of State 2002).

This review will not be able to fully assess the formulation containing 1 percent Cosmo-Flux 411F because that particular surfactant has not been used in the United States. Nevertheless, it will consider the summary of the investigation in Colombia of the formulation which does contain this additional surfactant. Though all aspects of glyphosate human poisoning data will be considered, this review will focus on one particular scenario, namely the effects of dermal and inhalation exposure from spray drift or residues, that result from aerial application.

The following data bases have been consulted for the poisoning incident data on the active ingredient glyphosate (PC Code: 103601):

1) Poison Control Centers - as the result of a data purchase by EPA, the Office of Pesticide Programs (OPP) received Poison Control Center data covering the years 1993 through 1998 for all pesticides. Most of the national Poison Control Centers (PCCs) participate in a national data collection system, the Toxic Exposure Surveillance System which obtains data from about 65-70 centers at hospitals and universities. PCCs provide telephone consultation for individuals and health care providers on suspected poisonings, involving drugs, household products, pesticides, etc. Note that Poison Control Center data does not have information on the type of application. So it is not possible to limit the review to the aerial application scenario or to limit it to only those persons secondarily exposed to drift or residue. However, it will be possible to exclude oral exposures which are inconsistent with the focus of the present review.

2) California Department of Pesticide Regulation - California has collected uniform data on suspected pesticide poisonings since 1982. Physicians are required, by statute, to report to their local health officer all occurrences of illness suspected of being related to exposure to pesticides. The majority of the incidents involve workers. Information on exposure (worker activity), type of illness (systemic, eye, skin, eye/skin and respiratory), likelihood of a causal relationship, and number of days off work and in the hospital are provided. The California data permits assessing the risk of exposure both to handlers and to bystanders. The exposure of bystanders and others to drift and residue will be a primary focus of this review.

3) Scientific Literature - A search was performed on Medline for scientific literature related to the human health effects of glyphosate. All articles were retrieved and reviewed for relevance. Articles involving dermal or ocular exposure are given priority because this fits with the primary scenario of concern for this review.

4) A report from the Department of Narino, Municipality of El Tablon De Gomez "A Study of Health Complaints Related to Aerial Eradication in Colombia", Final Report dated September 2001 is reviewed. This document addresses the specific exposure of interest and therefore will be given special attention.

I. Poison Control Center Data - 1993 through 1998

Results for the years 1993 through 1998 are presented below for occupational cases, non-occupational involving adults and older children, and for children under age six. Reports of intentional exposures (e.g., suicide attempts) and exposures to multiple products are excluded. Cases where the outcome was determined to be unrelated to the exposure were also excluded. Tables 1-4 present the hazard information for glyphosate compared with all other pesticides on six measures: percent with symptoms, percent with moderate, major (includes life-threatening or residual disability) outcome, percent with major outcome, percent of exposed cases seen in a health care facility, and percent hospitalized and percent seen in a critical care facility. There were no cases with a fatal outcome between 1993 and 1998. Table 1 reports the number of cases on which the data derived in Tables 2-4 are based. Table 2 presents this information for occupational cases, Table 3 for non-occupational cases involving adults and older children (six years or older), and Table 4 for children under age six. Note that Table 2, involving occupational exposure, is of less relevance to this review because it is inconsistent with the exposure scenario of interest in Colombia.

Table 1. Number of glyphosate exposures reported to the Toxic Exposure Surveillance System (AAPCC), number with determined outcome, number seen in a health care facility for occupational and non-occupational cases (adults and children six years and older) and for children under six years of age only, 1993-1998 .

Subgroup	Exposures	Outcome determined	Seen in Health Care Facility
Occupational: adults and older children	875	663	263
Non-occupational: adults and older children	7491	5177	940
Children under age six	4897	3589	207

Table 2. Comparison between glyphosate and all pesticides for percent cases with symptomatic outcome (SYM), moderate or more severe outcome (MOD), life-threatening or residual disability (LIFE-TH), seen in a health care facility (HCF), hospitalized (HOSP), or seen in an intensive care unit (ICU) reported to Poison Control Centers, 1993-1998 for occupational cases only.

Pesticide	SYM*	MOD*	LIFE-TH*	HCF*	HOSP*	ICU*
Glyphosate	77.8%	7.54%	0.15%	30.0%	2.28%	0.76%
All Pesticides	86.0%	18.8%	0.62%	47.0%	6.08%	2.36%
Ratio	0.90	0.40	0.24	0.64	0.38	0.32

* Symptomatic cases based on those cases with a minor, moderate, major, or fatal medical outcome. Denominator for SYM, MOD, and LIFE-TH is the total cases where medical outcome was determined. Denominator for HCF is all exposures. Denominator for HOSP and ICU is all cases seen in a health care facility.

Table 3. Comparison between glyphosate and all pesticides for percent cases with symptomatic outcome (SYM), moderate or more severe outcome (MOD), life-threatening or residual disability (LIFE-TH), seen in a health care facility (HCF), hospitalized (HOSP), or seen in an intensive care unit (ICU) reported to Poison Control Centers, 1993-1998 for non-occupational cases involving adults and older children.

Pesticide	SYM*	MOD*	LIFE-TH*	HCF*	HOSP*	ICU*
Glyphosate	61.9%	4.27%	0.15%	12.5%	2.87%	0.85%
All Pesticides	68.5%	10.5%	0.36%	16.4%	6.24%	2.67%
Ratio	0.90	0.41	0.43	0.76	0.46	0.32

* Symptomatic cases based on those cases with a minor, moderate, major, or fatal medical outcome. Denominator for SYM, MOD, and LIFE-TH is the total cases where medical outcome was determined. Denominator for HCF is all exposures. Denominator for HOSP and ICU is all cases seen in a health care facility.

Table 4. Comparison between glyphosate and all pesticides for percent cases with symptomatic outcome (SYM), moderate or more severe outcome (MOD), life-threatening or residual disability (LIFE-TH), seen in a health care facility (HCF), hospitalized (HOSP), or seen in an intensive care unit (ICU) for adults and children six years and older reported to Poison Control Centers, 1993-1998 for children under six years old.

Pesticide	SYM*	MOD*	LIFE-TH*	HCF*	HOSP*	ICU*
Glyphosate	23.5%	0.45%	0.056%	4.23%	3.38%	0.48%
All Pesticides	21.8%	1.40%	0.12%	16.4%	4.78%	1.36%
Ratio	1.08	0.32	0.47	0.26	0.71	0.35

* Symptomatic cases based on those cases with a minor, moderate, major, or fatal medical outcome. Denominator for SYM, MOD, and LIFE-TH is the total cases where medical outcome was determined. Denominator for HCF is all exposures. Denominator for HOSP and ICU is all cases seen in a health care facility.

Glyphosate clearly has a pattern of much lower toxicity than other pesticides. When cases with symptoms, moderate, and major medical outcome are evaluated, glyphosate not only has a lower ratio than other pesticides, but the ratio tends to decrease markedly with the more severe measure of outcome. Children under six years of age appear to differ from this finding somewhat, but this appears to be due to relatively small numbers. They had slightly more symptoms than children exposed to other pesticides and the ratio for life-threatening or residual disability was somewhat higher than the ratio for moderate outcome. This finding was based on two cases. One of these case reported severe burns and the other experienced multiple seizures which were considered to be unknown if related to the exposure to glyphosate. If this latter case were not included, then the pattern of decreasing ratio with more severe effect would be maintained. Overall, glyphosate was less than half as likely to result in serious effects (moderate or major outcome) as all pesticides combined based on over 9,000 exposures.

A similar pattern was seen for cases receiving health care. For occupational cases (Table 2), the ratio of cases receiving health care, hospitalization, and treatment in a critical care unit were 0.66, 0.37, and 0.32 respectively, decreasing with increasing level of medical care. A similar pattern was found for non-occupational adults and older children whose respective ratios were 0.76 for health care, 0.46 for hospitalization, and 0.32 for critical care. For children under six years of age, there was one cases requiring critical care and only 7 cases hospitalized. These relatively low numbers resulted in a ratio of 0.26 for health care, 0.71 for hospitalization, and 0.35 for critical care. When oral exposures, common among children under age six, are excluded, there were only 2 cases hospitalized and none required critical care treatment. Both of the children that were hospitalized experienced seizures that were considered to be unknown if related to their exposure to glyphosate. Thus, even in the most sensitive population, young infants, there was no strong evidence of serious effects from glyphosate.

II. California Data - 1982 through 2000

Detailed descriptions of 911 cases involving glyphosate, submitted to the California Pesticide Illness Surveillance Program (1982-2000), were reviewed. In 675 of these cases, glyphosate was used alone or was judged to be responsible for the health effects. These 675 cases include only those with a definite, probable or possible relationship. Table 4 presents the types of illnesses reported by year. Table 5 gives the total number of workers that took time off work as a result of their illness and how many were hospitalized and for how long.

Table 4. Cases Due to glyphosate in California Reported by Type of Illness and Year, 1982-2000.

Year	Illness Type
	Systemica	Eye	Skin	Respiratoryb	Combinationc	Total
1982	7	27	12	-	-	46
1983	4	24	22	-	2	52
1984	3	24	11	-	-	38
1985	7	23	11	-	1	42
1986	6	20	6	-	1	33
1987	5	18	8	-	-	31
1988	5	18	13	-	1	37
1989	7	18	12	-	-	37
1990	6	21	18	1	3	49
1991	13	28	16	1	5	63
1992	11	18	12	-	4	45
1993	6	12	11	-	1	30
1994	5	12	6	-	2	25
1995	4	17	14	-	2	37
1996	6	8	7	-	5	26
1997	3	7	10	-	1	21
1998	4	8	6	2	3	23
1999	6	7	5	3	0	21
2000	4	6	6	1	2	29
Total	112	316	206	8	33	675

a Category includes cases where skin, eye, or respiratory effects were also reported.
^b Category not used until 1990. Prior respiratory cases classified as systemic.
^c Category includes combined irritative effects to eye, skin, and respiratory system.

Table 5. Number of Persons Disabled (taking time off work) or Hospitalized for Indicated Number of Days After Glyphosate Exposure in California, 1982-1999.

Time period	Number of Persons Disabled	Number of Persons Hospitalized
One day	47	-
Two days	28	1
3-5 days	27	-
6-10 days	2	-
more than 10 days	19	-
Unknown	43	6

Starting in 1992, glyphosate was reformulated to reduce the amount of surfactant which posed a hazard to the eye. From 1982 through 1991, there were 221 illnesses involving the eye or 22.1 cases per year. From 1994 (allowing 2 years for the product to be introduced into trade and widespread use) through 2000, there were 65 illnesses involving the eye or 9.3 cases per year, a decline of 58%. Therefore, these data support the finding that the reformulated glyphosate used since 1992, have resulted in a significant drop in illnesses. Overall, the total illnesses due to glyphosate declined by 39% from the 1982-1991 time period to the 1994-2000 time period, largely due to the reduction in eye injuries. More important for the purpose of this review are those illnesses involving bystanders or other workers exposed to drift or residue, rather than handlers exposed directly to the product during application, mixing/loading, maintenance, cleaning, repair, transport or disposal. A variety of worker activities were associated with exposure to methyl bromide as illustrated in Table 6 below.

Table 6. Illnesses by Activity Categories for Glyphosate Exposure in California, 1982-1999

Activity Category	Illness Category
	Systemica	Eye	Skin	Respiratoryb	Combinationc	Total
Applicator	80	214	173	5	25	497
Mixer/Loader	4	61	15	-	1	81
Clean/Prepare/Repair	3	16	9	-	2	30
Transport/Disposal	2	10	3	-	-	15
(Handler-subtotal)	(89)	(301)	(200)	(3)	(28)	(623)
Direct Spray/Spill	1	6	-	1	-	8
Drift	12	4	2	2	3	23
Residue	2	3	2	-	2	9
Other and Unknown	8	2	2	-	-	12
Total	112	316	206	8	33	675

a Category includes cases where skin, eye, or respiratory effects were also reported.
^b Category not used until 1990. Prior respiratory cases classified as systemic.
^c Category includes combined irritative effects to eye, skin, and respiratory system.

Table 6 shows that activities that involve direct handling of glyphosate account for over 90% of the illnesses. Only 43 illnesses, 6% of the total, could be ascribed to direct spray/spill, drift, or residue, scenarios that could conceivably occur in Colombia as a result of the spray eradication program. Note that the 43 cases include 3 cases listed as unknown because the exposure could have been either direct spray, drift, or residue which could not be determined. Among the 43 cases, 30 had a causal relationship classified as possible or 70% of the total. Among all 675 cases, only 21% were classified as possible. Another 6 of the 43 cases were classified as having a probable relationship between exposure and health effects and 7 had a definite relationship. Therefore, it should be noted that the evidence for a causative relationship for the majority of cases involving drift or residue is often lacking.

Of the 43 cases, 7 took one day off of work as a result of their illness and 2 people took off two days, and another 2 people took off three days. A possible case picking plums did not know whether pesticides were applied prior to picking, took five days off work. A possible case occurred in a teacher who reported headache, nausea, fatigue, and vomiting after the glyphosate odor was sucked into her classroom by an air-conditioning unit. The final case, a gardener, took 13 days off after pulling weeds and possibly exposed to residue which got in his eyes, resulting in pain, burning sensations, and impaired vision.

Most of the symptoms reported in this group of 43 cases were relatively minor. Symptoms reported in four or more individuals included eye irritation (includes itching, pain, burning), conjunctivitis, rash, swelling, skin irritation (includes itching, blisters, pain, or numbness), throat irritation or burning, nasal congestion, headache, nausea, shortness of breath or breathing difficulty, and asthmatic reactions. Note that these symptoms are consistent with those specified in the fifth edition of Recognition and Management of Pesticide Poisonings (Reigart and Roberts 1999) which states that glyphosate is "irritating to the eyes, skin, and upper respiratory tract".

Many of the 43 cases described in the California report involved heavier exposures than are likely to occur as a result of aerial application. For example, most of the direct spray cases resulted when a bystander was inadvertently sprayed by an applicator on the ground and was often drenched. Illnesses due to residue were often the result of extensive contact with soil or foliage recently treated with glyphosate. Reports of illness from drift from an aerial application were relatively rare, accounting for four of the 23 cases reported above.

Out of 1,384 incidents related to drift reported to the California Pesticide Illness Surveillance Program from 1982 through 1997, only 8 cases were related to glyphosate and suffered mild to moderate effects such as headache, dizziness, coughing, sore throat and chest pain. Two individuals reported an allergic reaction which included hives and rash. Glyphosate is second most widely applied pesticide in California (see Wilhoit et al. California EPA web site: http://www.cdpr.ca.gov/docs/pur/purmain.htm) and unlike many other pesticides has never been responsible for a large number of illnesses due to drift from a nearby application. California reported the number of applications of glyphosate and all pesticides from 1991 through 1996 (see web site for data). There were 5,576 systemic poisonings (possible, probable, and definite) reported in this time period for all pesticides and 45 systemic poisoning reported for glyphosate. The number of poisonings per 1,000 applications was 0.6204 for all pesticides and 0.0781, thus glyphosate had an estimated rate of systemic poisoning that was only 12% that of all pesticides.

The review of California illness reports suggest that even diluted glyphosate can be a cause of skin, eye, or throat irritation. These effects are almost always self limiting and do not require hospital admission for treatment. Only one of the 675 California cases required hospitalization. This 1999 case occurred in an individual who had severe respiratory problems after applying glyphosate on three successive days, handled moldy grass clippings, and had pre-existing medical conditions, including asthma.

Systemic illness (as opposed to skin, eye, and respiratory illness), such as headache, nausea, and difficulty breathing have been reported, in sensitive individuals, such as persons with asthma. However, in most such cases, there was insufficient documentation to say that these systemic effects were definitely due to their exposure to glyphosate. There were only 3 systemic cases (primarily headache) due to spray drift that were considered to have a definite or probable relationship between the exposure and the illness.

III. Literature review

The literature review is arranged according to principle route of exposure. Dermal and ocular routes are of most interest because these are the routes of exposure that persons in Colombia might experience as a result of aerial applications.

Hindson and Diffey (1984a) describe a case of a 64 year-old untanned Caucasian male who developed acute dermatitis after spraying weeds with a glyphosate product. Patch testing with a 1% and 5% aqueous solution of the product were negative. However testing patches followed by exposure to ultraviolet radiation did reveal a marked papulovesicular reaction, indicating a phototoxic reaction. However, subsequent questions were raised about the benzisothiazolone preservative used in this weedkiller (Hindson and Diffey 1984b). Separate testing of the both the glyphosate and the benzisothiazolone, revealed that the benzisothiazolone was the phototoxic agent and not the glyphosate.

Maibach (1986) performed extensive testing of 346 volunteers for evidence of dermal effects of glyphosate including acute irritation, cumulative irritation, photoirritation, and allergic and photoallergic contact potential. The test compound was 98.4% pure (made up of 41% glyphosate) and contained isopropyamine salt of glyphosate, water and surfactant. The test compound was used at full strength and diluted in distilled water. All test subjects were adults and exposed by covering skin with non-woven fiber patches. Baby shampoo, all purpose cleaner, and dishwashing liquid was used for comparison. "Compared to the baby shampoo, the herbicide was statistically indistinguishable in its irritant potential. The tests also show that it did not induce sensitization, photoirritation or photosensitization." The author noted, however, that he could not rule out the possibility that product contaminants might cause sensitization in unusual circumstances. Mild irritation was observed in a few individuals who had concentrated product applied to the skin for 24 hours.

Wester et al. (1991) examined the potential of glyphosate for skin binding, skin absorption and residual tissue distribution. In vitro percutaneous absorption through human skin into human plasma was reported to be no more than 2% over a concentration range of 0.5-154 ug/cm². Other testing in rhesus monkeys will not be discussed here. The important finding from this study is that relatively small amounts of glyphosate are absorbed across the skin and therefore, absent moderate to high toxicity, dermal exposure is likely to result in only dermal effects.

Talbot et al. (1991) reviewed 93 cases of glyphosate exposure reported to the emergency room in Taiwan from 1974 through September 30, 1989. Cases involving exposures to other products were excluded. The majority of these cases were suicidal and involved oral exposures. There were two dermal exposures reported both of which were asymptomatic.

Temple and Smith (1992) reviewed a series of cases reported to the New Zealand National Poisons Information Centre. The majority of these cases were unintentional exposures, mostly while spraying glyphosate containing herbicides. "In general, these cases exhibited minor local irritant effects which were self limiting and responded well to symptomatic and supportive care." The authors give three examples of such cases. In the first case, the male adult accidently rubbed concentrate in his eye and developed edema around the eye and the conjunctiva around the cornea. These symptoms were associated with fast pulse, palpitations, elevated blood pressure, headache, and slight nausea. His symptoms resolved with treatment and he resumed work the next day. In the second example, a male adult sprayed a double strength solution which contacted his hand due to a faulty hand grip on the spray unit. He wiped his face which became swollen with paraesthesia. These symptoms resolved over 48 hours and did not require specific treatment. The third case was accidently drenched with diluted glyphosate. He developed a vesicular skin eruption especially on his arms and hands associated with burning and itching. This condition required treatment at two monthly intervals.

In a review of skin reactions to pesticides, OMalley (1997) provided a brief review of glyphosate. His summary is quoted below:

OMalleys finding is supported by the California Pesticide Illness Surveillance Program, which is regarded as the best, most comprehensive source of information on human pesticide exposure in the United States (U.S. General Accounting Office 1993). This information and the earlier review of data from California, strongly support the conclusion that the dermal risk of glyphosate, as formulated in the United States, is primarily to pesticide handlers with very little or minor risk to others (e.g., bystanders) who may be exposed to glyphosate drift or residue.

There was one recent case report in the literature of a 54 year-old man in Brazil who unintentionally sprayed himself and developed skin lesions six hours later (Barbosa et al. 2001). He developed severe conjunctival hyperemia (excess blood flow) and a rash which became blisters and persisted for 15 days. One month after the exposure he developed symmetrical parkinsonian syndrome. The authors acknowledge that "it is not possible to exclude the coincidence [idiopathic Parkinsons disease] with exposure to glyphosate" and add that no other report of parkinsonism induced by glyphosate has been reported. The authors propose a possible mechanism for excitatory mechanisms but characterize their finding as a hypothesis. Other more detailed studies are underway to determine whether pesticides might be related to Parkinsons disease and any conclusions about the potential involvement of glyphosate will have to await the results of those studies.

Williams et al. (2000) prepared an extensive risk assessment and safety evaluation of glyphosate, partly supported by scientists with the manufacturer. They cite a study by Jauhiainen et al. (1991) which evaluated short-term effects among five forest applicators. and compared results with pre-exposure baseline as well as to data from a group of five controls. "There were no effects on hematology, clinical chemistry, ECG, pulmonary function, blood pressure, or heart rate 1 week after application." They also cited California data as reviewed by Pease et al. (1993) and noted that irritation of skin and eye effects were common, but not exceptional taking into account the widespread product use. Reviewing the Temple and Smith (1992) report (reviewed above), Williams et al. suggest that the systemic symptoms reported (e.g., headache, fast pulse, slight nausea) "probably represent a nonspecific response related to pain associated with eye and/or skin irritation." Other studies cited by Williams related to dermal effects have already been reviewed above.

Acquavella et al. (1999) reviewed ocular effects reported to the American Association of Poison Control Centers in the United States from 1993 through 1997. They identified 1513 records involving ocular or dermal/ocular exposure. Information from patient notes kept by at least one Poison Center were also reviewed. More than 80% of the exposures were residential and about 15% were occupational. Only 5% of the calls involved concentrated product. "Approximately 70% of callers had minor effects, primarily transient irritations, attributed to exposure. Ninety-nine percent of those with minor effects complained of eye pain, 3% complained of lacrimation (watery eyes), and 3% complained of blurred vision." Those exposed to more concentrated formulations (>2% glyphosate, >1%Polyethoxylated tallow amine) were more likely to report lacrimation but not blurred vision, however, there was little evidence of a trend between concentration categories and lacrimation. A total of 30 callers (2%) were classified as having a moderate medical outcome, such as persistent irritations, low grade corneal burns or abrasions. There was one caller (0.1%) with a major effect - scarring of the upper palpebral conjunctiva. This patient was wearing extended wear contact lenses that were rinsed and replaced right after the exposure. Over the next 17 days the patient had a corneal abrasion and conjunctivitis which resolved as the vision returned to its pre-exposure state. The additional information on this case suggests the case should be reclassified as moderate because the scarring of the tear duct system, which was thought to be permanent, did heal. There were 95 calls lost to follow-up where medical outcome remained undetermined. In summary, there was some temporary injury in about 2% of the reported cases, but no case of permanent damage.

Jamison et al. (1986) conducted a study of pulmonary function in workers handling flax which had previously had the fibers softened and separated and either wetted or treated with glyphosate 6 weeks prior to harvest. The authors concluded that workers had a significant decrease in pulmonary function which was likely due to exposure to the dust. Though there was very little residue of glyphosate at the end of the six weeks, the authors stated that glyphosate could not "be excluded as a cause of the increase pulmonary function impairment observed." However, Williams et al. (2000) took issue with this view, noting that the levels of glyphosate would be very low, "if present at all, and could not be responsible for the altered pulmonary function observed." They felt the production of dust particles and/or different microorganisms during the process were a more likely explanation.

Pushnoy et al. (1998) reported on a 42 year-old mechanic who cleaned and repaired a spray rig in a confined space. He reported to the emergency department complaining of shortness of breath, irritative cough, dizziness, discomfort in the throat, and coughing up blood. He was admitted to the hospital and diagnosed with acute chemical pneumonitis. The authors suggested that the polyoxyethylene amine surfactant was largely responsible for irritant effects on the mucosal lining and lung tissue, and therefore, the likely cause of the pneumonitis. Goldstein et al. (1999, 5 authors with industry and one with Yale University) took issue with the findings of Pushnoy et al. They argued that neither glyphosate nor any compound in the finished product could vaporize sufficiently even in a poorly ventilated space to cause such an exposure. They added that occupational pneumonitis had never been reported in connection with glyphosate products. The original authors replied that a longer presentation of the circumstances surrounding this case would have permitted Goldstein et al. to reject the alternative exposures (e.g., chlorinated solvent, diesel fuel, welding) they suggested. They further state that even though the vapor pressure of glyphosate was low "we have concluded that the patients clinical symptomatology resulted from exposure and inhalation of a mixture of vapor and air-borne droplets containing glyphosate . . . that part of the parenchymal reaction was due to the effect of a surfactant (such as polyoxyethylene amine) on the alveolar lining". However, the authors acknowledge that the effect of the surfactant, though it seems plausible, is "based just on clinical evidence".

Inadvertent oral exposure to glyphosate sprayed on coca plants in Colombia is extremely unlikely. Therefore, the review of the scientific literature on oral exposure will be cursory. In a letter to Lancet Sawada et al. (1988) reported on 56 ingestions of glyphosate product. They found that the average dose among fatal cases was 104 ml and 206 ml among fatal cases. They describe the clinical picture as one of hypovolaemic shock likely due to the 15% polyoxyethylene amine surfactant. Of the 56 cases reviewed, 48 cases were attempted suicides, 3 unintentional (all infants), and 5 with unknown intent. Jackson (1988 with Monsanto) responded to the Sawada et al. report and stated that there were no reports of deaths following accidental ingestion.

Kageura et al. (1988) reported on the death of a 26 year old woman who ingested glyphosate in a suicide. They attributed the death to inhalation of vomitus into the lungs causing asphyxiation. Talbot et al. (1991) reviewed 93 cases of glyphosate exposure reported to the emergency room in Taiwan from 1974 through September 30, 1989. Cases involving exposures to other products were excluded. The majority of these cases were suicidal and involved oral exposures. Those cases where the amount ingested was not recorded were also excluded. They noted that some cases had only moderate effects even after ingestion of up to 500 ml and death had resulted from ingestion of concentrate in amounts above 85 ml. Oral ingestions by mistake in seven cases was usually of a small amount and "resulted in only minor mouth discomfort". The authors concluded that "the data suggest that those over 40 years of age, who ingest more than 100 ml, are at the highest risk of a fatal outcome."

Tominack et al. (1991) reported on 97 telephone consultations with the Taiwan National Poison Center involving ingestion of glyphosate-surfactant herbicide concentrate from January 1986 through September 1988. Eighty-eight cases were suicidal, five unintentional, and four with uncertain intention. Non-fatal cases ingested an average of 120 ml (range 5-500 ml) and fatal cases averaged 263 ml (range 150-500 ml). Of these 97, 12 were asymptomatic, 28 had mild, 33 had moderate, and 16 had severe symptoms. Increasing dose and increasing age were significant predictors of fatality. It should be noted that 10 of the 97 cases ingested another substance in addition to glyphosate. They found that ingestion of a mouthful of concentrate or more was capable of producing symptoms including gastrointestinal mucosal injury, pulmonary edema, decreased or absent urine output, metabolic acidosis, leukocytosis, fever, and hypotension that possibly could develop into shock. Similar to the paper cited above, a dose of 150 ml or more and age of 40 years or more were found to be at highest risk of fatal outcome.

Menkes et al. (1991) reported on four cases of suicidal ingestion of glyphosate, one of them fatal. Two of the cases experienced massive fluid and electrolyte loss, probably due to tubular necrosis. After considering all of the evidence concerning the glyphosate and the surfactant the authors stated "it seems unlikely that toxicity can be ascribed solely to the surfactant."

Temple and Smith (1992) reviewed a series of cases reported to the New Zealand National Poisons Information Centre. The majority of these cases were unintentional. Three ingestions are described, two of them fatal. The authors concluded "Small ingestions (less than 5 ml of the concentrate in adults) pose little problem and simple dilution to minimize gastrointestinal irritation should suffice."

Hung et al. (1997) reviewed 53 cases reported between 1992 and 1996 in Taiwan to assess laryngeal injury. Of these, 36 reported significant laryngeal injury was strongly correlated with aspiration pneumonitis (reported in 8 cases). The average amount ingested in such cases was 300 ml.

Lin et al. (1999) reported on a suicide case who drank 150 ml of concentrate (41% isopropylamine salt of glyphosate, 15% polyoxyethylene amine). This 26 year-old man experienced cardiogenic shock which may have been due to transient suppression of the cardiac conduction system and contractility, rather than intravascular hypovolemia.

Chang et al. (1999) reported on 50 patients with suicidal glyphosate-surfactant ingestions and evaluated their upper gastrointestinal tract injuries. Esophageal injury was seen in 68% of patients, gastric injury in 72%, and duodenal injury in 16%. The authors considered these injuries "minor in comparison with those by other strong acids."

Lee et al. (2000) reviewed 131 cases of glyphosate ingestion seen in their emergency department in Taiwan over a seven year period. There were 11 fatalities (mortality rate 8.4%). The most common presentations included sore throat (80%), nausea (74%), vomiting, and fever (41%). The most common laboratory abnormalities were leukocytosis (68%), low bicarbonate (48%), acidosis (36%), elevated AST, hypoxemia (28%), and elevated BUN. Of the 81 cases receiving an electrocardiogram, 15 were abnormal, mainly sinus tachycardia and nonspecific ST-T changes. Twenty-two of 105 patients who had chest x-rays had abnormal infiltrates or patches. Three patients with renal failure all died. Poor outcome was predicted by respiratory distress, renal dysfunction, abnormal CXR, shock, and ingestion of 200 ml or more, altered consciousness, hyperkalemia, and pulmonary edema. The 11 cases that died ingested an average of 330 ml which was higher than the previous reports by Sawada et al. (1988) and Tominack et al. (1991). The authors propose that direct damage to the airway passage is an important factor in severe poisoning.

There were two studies located that evaluated reproductive outcome in farmers handling pesticides that specifically analyzed for the effects of glyphosate. However, these studies are both retrospective, subject to numerous biases and confounders, and only suggest associations rather than causative relationships. Thus any of these finding would require replication and further evaluation before they could become established. These studies are summarized below.

Savitz et al. (1997) examined male pesticide exposure three months before conception and through conception in relation to pregnancy outcome in an Ontario farm population. The risk for miscarriage was not statistically significant, though somewhat elevated for glyphosate users. This finding was true for both use of glyphosate on crops (17 cases) and in yards (13 cases). A similar result was found for preterm delivery based on 5 cases involving crop use of glyphosate. On the other hand, there was no statistical significance or elevation of risk for small for gestational age infants. The authors acknowledge that the lengthy recall interval may have reduced the quality of information collected on exposure and health outcome. They advise "Replication of these findings in other geographic settings in a study of similar quality would be of value; however, to improve on our strategy, the availability of unusually detailed source of historical exposure data would be necessary."

Arbuckle and Mery (2001) evaluated the risk of spontaneous abortion in the same Ontario farm population examined by Savitz et al. (1997). In this refined analysis they found that late abortions were statistically associated with preconception exposure to glyphosate (odds ratio = 1.7, 95% confidence interval 1.0-2.9). This finding is only just marginally significant. The authors state their findings have "several limitations . . . Because dose information was not available, misclassification of exposure is likely." Finally they state their "analyses were designed to generate, not to test, hypotheses". Due to the fact that multiple comparisons were conducted some findings may be due to chance.

IV. A Study of Health Complaints Related to Aerial Eradication in Colombia

This report, prepared by the Department of Narino, Municipality of El Tablon De Gomez, makes a concerted effort to identify any health problems that might be related to use of glyphosate in aerial eradication programs. The study was commissioned by the U.S. Embassy in Bogota and conducted independently by Dr. Camillo Uribe, Director of Clinica Uribe Cualla, the national poison control center. Sections of this report are summarized below with the sections numbed in bold as in the original report.

This report primarily concerns the area around the municipality of El Tablon in southern Colombia. The total population is given as 16,770, of which 89% is categorized as rural. The main crops in this area include coffee, corn, wheat, oats, potatoes, and illicit opium poppy. It is known that a variety of other more toxic pesticides are used on these crops. The municipality has three health centers, including Aponte, which is the focus of this report. The Aponte health center is staffed by a medical doctor, a nurse, and a nurses aide. From July 2000 to February 2001, the primary time period of this report, Dr. Tordecilla was the medical doctor. Aerial eradication of the illicit opium poppy reportedly occurred in this region in June, July, and November of 2000.

The Narino Departmental Health Institute provided summary morbidity and mortality information for the El Tablon De Gomez area and the Aponte settlement for the year 1999. Data for the year 2000 had not yet been officially released, but estimates are provided by a method not specified. Six illnesses likely to be related to pesticide exposure were identified and tabulated, including acute diarrhea, acute respiratory infection, dermatitis, intoxication, conjunctivitis and headache. The author notes that the first three illnesses listed (diarrhea, respiratory infection, and dermatitis) are likely to be related to problems with inadequate nutrition, housing, and lack of health services, rather than due to pesticide exposure. The basis for this listing of symptoms is not specified, but does agree fairly well with the list of symptoms likely to result from glyphosate exposure based on Poison Control Center data, California surveillance reports, and the world literature. Total morbidity for 1999 and estimated morbidity for 2000 are given in the Table below for El Tablon De Gomez and the Aponte Settlement below:

Table 7. Morbidity reported in the El Tablon De Gomez of Colombia in 1999 and estimated for 2000.

Table 8. Morbidity reported in the Aponte Settlement of Colombia in 1999 and estimated for 2000.

Pathology	1999	2000 Estimated
Acute diarrhea	181	190
Acute respiratory infection	199	222
Dermatitis	210	180
Poisoning/Intoxication	4	4
Conjunctivitis	87	104
Headaches	78	95
Total for 6 suspected illnesses	759	795

It appears the Aponte settlement is contained within the El Tablon De Gomez area, though this is not entirely clear. The figures in the report are listed by five separate age groups. This reveals, that the majority of the cases of diarrhea and respiratory infection occurred in children less than five years old, as would be expected given the known demographics of those diseases. Nationwide data show that 53% of intoxications are suicides or suicide attempts, but it is not clear how many of the 13 poisonings listed above might be suicidal or, more importantly, are due to other products such as medications. In both Tables 7 and 8 there is an increase of 5% from 1999 to the estimate for 2000. Given that spraying is reported to have occurred in 2000 and not in 1999, this suggests that the overwhelming majority (95%) of illnesses reported would be background incidence unrelated to the spraying of herbicide. The remaining 5% increase could be due to a variety of causes and do not support a conclusion that glyphosate was responsible for these complaints.

In addition to the summary of morbidity, there is a mandatory health reporting system in Colombia for 34 illnesses including pesticide poisonings. The review of these records found no reports of pesticide poisoning for the municipality of El Tablon in the year 2000 or the first 9 weeks of 2001. Weekly reports from Attachment 5 were examined to determine how many pesticide poisonings were reported each month. It was not clear whether the dates on each report represented the starting date or ending date for a reporting period. Regardless of which is correct, it did not appear that the times of spraying correlated with reports of pesticide intoxication.

Table 9. Reports of Pesticide Intoxication provided to the Narino Department of Health Institute, Epidemiology Section January 12, 2000 through March 7, 2001.

Month in 2000	Number of Poisonings	Month in 2000 or 2002	Number of poisonings	Poisonings occurring at time of spraying
January	0	July	11	9
February	0	August	6
March	8	September	12
April	13	October	8
May	7	November	13*	6*
June	15	December	2
--	-	Jan. 2001	7
--	-	Feb. 2001	19
--	-	Mar. 2001	0

* Reports for weeks number 43 and 44 in the first half of November were missing.

Out of a total of 121 reported pesticide poisonings in 61 weeks, only 15 occurred during 5 weeks when spraying eradication occurred. This given the variation in the data, this could easily be due to chance and be unrelated to glyphosate exposure. More work would be required to determine whether locations of the 15 suspect poisoning matched the location and timing of spraying.

In 2000, the Narino Department of Health requested all municipalities to report on the effects of spraying on human health. Ten municipalities supplied the following reports:

Three municipalities including Tablon de Gomez, Barbacoas, and Magui reported no cases. However, the reports were completed prior to the November spraying in Barbacoas and Magui and prior to (or perhaps during) the July and before the November spraying in Tablon de Gomez.

Buesaco reported one patient with sore throat, numbness in limbs, and conjunctivitis in June.

In Tumaco, six case of patients with conjunctivitis and dermatitis were reported.

In San Pablo, 50 cases of dermatitis, conjunctivitis, respiratory conditions, and digestive problems were reported after spraying.

In La Cruz, two cases of allergic rhinitis, two cases of dermatitis, and five cases of conjunctivitis were reported.

San Jose de Alban did not report any specific cases, but the scientific coordinator and chief nurse noted increase in gastrointestinal, dermatological and respiratory conditions. The exact quantity of these conditions in relation to spray times was not given.

El Rosario reported five cases of conjunctivitis and rhinitis that might have been related to spraying carried out on July 31.

San Pedro de Cartago reported an increase in gastrointestinal symptoms but no quantitative relationship between illnesses and spray times was provided.

The absence of any reports of pesticide poisoning combined with the information from the ten municipalities is difficult to interpret. Glyphosate is known to cause irritation to the skin, eyes, mucous membranes which may account for some of the reports of sore throat, conjunctivitis, dermatitis and other conditions described above. However, it is not possible to evaluate these reports in any detail due to the lack of any information on how many of these cases experienced exposure immediately prior to their illness and lack of information on investigation of potential alternative causes. This anecdotal information does not provide any substantial evidence of health effects due to glyphosate spraying in Colombia. Many of the reports are consistent with exposure to glyphosate by the dermal route, as reported in California and the literature. So, some number of cases (impossible to estimate) could be related to the aerial eradication program.

A commission visited the municipality of El Tablon on January 22, 2001 and spoke with Dr. Tordecilla and reviewed health records of his patients. A number of records of skin conditions were noted for the months of October, December 2000, and January 2001. The exact number of cases, selection criteria, and method of analysis was not specified in the summary report. Nevertheless, the commission concluded "that the information available permitted the commission to consider only the possibility of an association between exposure to pesticides and the effects". The commission noted that it lacked the technical expertise, the data on dates and locations of spraying, and therefore could not conclude whether the observed conditions were related to pesticide exposure.

Employees of the Narino Department Health Institute were interviewed. According to a Fatima Health Promoter, he/she thought the children were most affected, suffering gastrointestinal problems and eye irritation. One possible route of exposure was the village water fountains that were reportedly sprayed and that water could reach the water that supplies the settlement. The most common symptoms in children, according to the Health Promoter, were stomach aches and vomiting, which were different from the most common symptoms reported by Lee et al. (2000) which were sore throat and nausea. This suggests that some cause other than glyphosate was responsible for the childrens complaints. He/she reported one case of a boy with skin lesions like sores after the spraying. He/she was particularly concerned that peasants receive more health care from the government. He/she also noted that he lost a considerable sum of money when most of his peas were affected by the spraying in June. He/she claimed to have a tape made during the spraying by pilots where they were heard to say they would dump the remainder of their herbicide in a field because they had too much quantity.

A nurses aide reported that three or four patients with burning eyes, headache, and dizziness were seen at her health center. One boy with a respiratory infection was sent to another health center where he arrived dead. Medical records were sought to substantiate this report but there was no clinical history, autopsy or other information to support it. She referred a patient with urinary problems to the hospital. Subsequent review of the medical records of this case did not find reference to glyphosate exposure and suggested an infectious origin. She said there had been cases of dermatitis, headache, abdominal pain and gastrointestinal symptoms, but could not say whether the symptoms were related to glyphosate exposure. She also was concerned that the government supply aid to this needy population, especially better health services.

Another nurses aide reported by telephone that her impression was that the number of dermatological consultations had increased. She admitted her impression was subjective and that the reasons for the consultations were the same as in previous years, so she would not commit to there being a relationship between the spraying and these dermatological conditions.

Reports of anecdotal evidence by nurses aides and the health promoter have little value for establishing any link between the spraying of glyphosate and health effects. Only with follow-up to substantiate the suspicions could a more substantial case be made. These interviews do no add significant evidence about the health risks of glyphosate used in Colombia.

Some videos made by reporters were reported at the Aponte Health Center. One of these videos was located and reviewed. It purported to show spraying on November 3, 2000 which caused unspecified "calamities". However, specific evidence of health complaints in humans was, apparently, not provided.

This video does provides opinion but does not add substantive information about the potential health effects of glyphosate used in Colombia.

There were 29 cases reported by Dr. Tordecelli and clinical records were obtained for 21 of them. Two other reports of skin lesions were sought but could not be confirmed. After careful review of the 21 records, it was determined that all but four cases were likely due to other causes. Most had skin conditions known to be related to bacteria or parasites, not chemical exposures and the onset of their symptoms did not correspond with the times of spraying. There were seven patients whose symptoms started after spraying and three of these were conditions known to be caused by bacteria or parasites. For the remaining four cases possibly related to glyphosate spraying, one was an allergic reaction that had been seen in this patient before when there was no spraying. A second and third case were contact eczema that is endemic in this region and thought to be more likely due to an infectious origin. One of these two cases did not initiate until 52 days after the last spraying. The fourth case was dermatitis on the thigh which would typically be protected by clothing and thereby protected from aerial spray applications. This reviewer agrees with the conclusion that "the twenty-one clinical histories . . . reveals that any relationship between aerial eradication with the herbicide glyphosate and the skin conditions treated in Aponte is unlikely".

In summary, the evidence collected and presented in this report cannot confirm the glyphosate used in Colombia as the likely cause of a single illness. There is suggestive evidence in the form of reported increases of morbidity and reports from municipalities that some cases of relatively mild complaints could have occurred in relation to the spraying eradication program. Some of the reports appear to be similar to those reported in the literature and by California. These cases report irritation to skin, eyes, a respiratory passages and suggesting that the Cosmo-Flux 411F added to the glyphosate in Colombia has little or no effect on the overall toxicity of the formulated product. If true, this would mean that the evaluation of glyphosate, as used in the United States and elsewhere, would be expected to have the same toxicologic properties and effects as glyphosate formulated in Colombia.

Rather than review incomplete medical records, it would be better to collect information prospectively. For example, if pesticide poisoning is a mandatory reporting condition, a form documenting the exposure, health effects and medical data on each case could be designed and used to establish whether any particular conditions might be related to spraying glyphosate. Without prospective collection of data and follow up it is difficult to evaluate potential health effects of glyphosate.

V. Conclusions

There is some data to suggest that the spray eradication program could have resulted in minor skin, eye, or respiratory irritation, and perhaps headache or other minor symptoms. However, the detailed information on timing of application, history of exposure, and medical documentation of symptoms related to glyphosate exposure were not available. Thus, not a single case of the reported symptoms can be confirmed as caused by the spray applications. The information so far collected gives the impression that any increase in health problems is likely to be relatively small at most and the severity of those symptoms is likely to be minor to moderate at most. Given the limited amount of documentation, none of the data in the report from Colombia provide a compelling case that glyphosate spraying has been a significant cause of illness in the region studied. Some of the reports in Colombia, potentially related to glyphosate, are similar to those reported in the literature and by California. These cases report irritation to skin, eyes, a respiratory passages and suggesting that the Cosmo-Flux 411F added to the glyphosate in Colombia has little or no effect on the overall toxicity of the formulated product. Colombia. Prospective tracking of reports of health complaints, documenting times of exposure and onset of symptoms, are recommended during future spray operations to evaluate any potential health effects and ameliorate or prevent their occurrence.

References

Acquavella JF, Weber JA, Cullen MR, Cruz OA, Martens MA, Holden LR, Riordan S, Thompson M, Farmer D. 1999. Human ocular effects from self-reported exposures to Roundup herbicides. Hum Exp Toxicol.18(8):479-86.

Arbuckle TE, Lin Z, Mery LS. 2001. An exploratory analysis of the effect of pesticide exposure on the risk of spontaneous abortion in an Ontario farm population. Environ Health Perspect 109(8):851-7.

Barbosa ER, Leiros da Costa MD, Bacheschi LA, Scaff M, Leite CC. 2001. Parkinsonism after glycine-derivate exposure. Mov Disord. 16(3):565-8.

Chang CY, Peng YC, Hung DZ, Hu WH, Yang DY, Lin TJ. 1999. Clinical impact of upper gastro-intestinal tract injuries in glyphosate-surfactant oral intoxication. Hum Exp Toxicol. 18(8):475-8.

Goldstein DA, Johnson G, Farmer DR, Martens MA, Ford JE, Cullen MR. 1999. Pneumonitis and herbicide exposure. Chest. 116(4):1139-40.

Hindson TC, Diffey B. 1984a. Phototoxicity of glyphosate in a weedkiller. Contact Dermatitis. 10(1):51-2.

Hindson TC, Diffey BL. 1984b. Toxicity of a weedkiller: a correction. Contact Dermatitis. 11(4):260.

Hung DZ, Deng JF, Wu TC. 1997. Laryngeal survey in glyphosate intoxication: a pathophysiological investigation. Hum Exp Toxicol. 16(10):596-9.

Jackson JR. 1988. Toxicity of herbicide containing glyphosate. Lancet. Feb 20;1(8582):414.

Jamison JP, Langlands JH, Lowry RC. 1986. Ventilatory impairment from pre-harvest retted flax. Br J Ind Med. 43(12):809-13.

Jauhiainen A, Rasanen K, Sarantila R, Nuutinen J, Kangas J. 1991. Occupational exposure of forest workers to glyphosate during brush saw spraying work. Am Ind Hyg Assoc J 52:61-64.

Kageuka M, Hieda Y, Hara K, Takamoto M, Fukuma Y, Kashimura S. 1988. Analysis of glyphosate and (aminomethyl) phosphonic acid in a suspected poisoning case. Nippon Hoigaku Zasshi. 42(2):128-32.

Lee HL, Chen KW, Chi CH, Huang JJ, Tsai LM. 2000. Clinical presentations and prognostic factors of a glyphosate-surfactant herbicide intoxication: a review of 131 cases. Acad Emerg Med. 7:906-10.

Lin CM, Lai CP, Fang TC, Lin CL. 1999. Cardiogenic shock in a patient with glyphosate-surfactant poisoning. J Formos Med Assoc. 98(10):698-700.

Maibach HI. 1986. Irritation, sensitization, photoirritation and photosensitization assays with a

glyphosate herbicide. Contact Dermatitis. 15(3):152-6.

Menkes DB, Temple WA, Edwards IR. 1991. Intentional self-poisoning with glyphosate-containing herbicides. Hum Exp Toxicol. 10(2):103-7.

OMalley MA. 1997. Skin reactions to pesticides. In: Occupational Medicine: State of the Art Reviews 12:209-220. Philadelphia, Hanley & Belfus, Inc.

Pease WS Morello-Frosch RA, Albright DS, Kyle AD, Robinson JC. 1993. Preventing Pesticide-Related Illness in California Agriculture. California Policy Seminar, University of California, Berkeley, CA.

Pushnoy LA, Avnon LS, Carel RS. 1998. Herbicide (Roundup) pneumonitis. Chest. 114(6):1769-71.

Reigart JR, Roberts JR. 1999. Recognition and Management of Pesticide Poisonings, Fifth Edition. EPA 735-R-98-003. U.S. Environmental Protection Agency, Washington, D.C.

Savitz DA, Arbuckle T, Kaczor D, Curtis KM. 1997. Male pesticide exposure and pregnancy outcome. Am J Epidemiol. 146(12):1025-36.

Sawada Y, Nagai Y, Ueyama M, Yamamoto I. 1988. Probable toxicity of surface-active agent in commercial herbicide containing glyphosate. Lancet. Feb 6;1(8580):299.

Talbot AR, Shiaw MH, Huang JS, Yang SF, Goo TS, Wang SH, Chen CL, Sanford TR. 1991. Acute poisoning with a glyphosate-surfactant herbicide ('Roundup'): a review of 93 cases. Hum Exp Toxicol. 10(1):1-8.

Temple WA, Smith NA. 1992. Glyphosate herbicide poisoning experience in New Zealand. N Z Med J. 105(933):173-4.

Tominack RL, Yang GY, Tsai WJ, Chung HM, Deng JF. 1991. Taiwan National Poison Center survey of glyphosate--surfactant herbicide ingestions. J Toxicol Clin Toxicol. 29(1):91-109.

U. S. Department of State. 2002. Letter to Christine Todd Whitman, Administrator, Environmental Protection Agency from Colin L. Powell. and attachment "Chemicals Used for the Eradication of Illicit Coca in Colombia and Conditions of Application.

U.S. General Accounting Office. 1993. Pesticides on Farms: Limited capability exists to monitor occupational illnesses and injuries. Report to the Chairman, Committee on Agriculture, Nutrition, and Forestry, U.S. Senate, December 1993. GAO/PEMB-94-6.

Wester RC, Melendres J, Sarason R, McMaster J, Maibach HI. 1991. Glyphosate skin binding, absorption, residual tissue distribution, and skin decontamination. Fundam Appl Toxicol. 16(4):725-32.

Williams GM, Kroes R, Munro IC. 2000. Safety evaluation and risk assessment of the herbicide Roundup and its active ingredient, glyphosate, for humans. Regul Toxicol Pharmacol. 31(2 Pt 1):117-65.

I Introduction

At the request of the Department of State (DoS), the Office of Pesticide Programs (OPP) of the U.S. Environmental Protection Agency has developed an ecological risk assessment for the aerial coca eradication program in Colombia. The DoS met with members of OPP on April 18, 2002 to provide information on exposure and use of the glyphosate tank mixture for aerial eradication of illicit coca in Colombia. This assessment is based on the information provided in that meeting and in the appendix included in the formal request from the Secretary of State. The eradication program includes the use of a spray mixture of a glyphosate formulation, an adjuvant (Cosmo-Flux 411F) and water. The glyphosate tank mixture is applied aerially as a foliar application in certain provinces within Colombia.

II Background

Before a pesticide can be sold in the United States, the Agency requires pesticide companies which request product registrations in the U.S. to perform certain required environmental fate and ecological effects studies and to submit the resulting data. OPP uses the environmental fate studies to assess potential environmental exposure; data requirements are listed in the regulations (40 CFR 158.290). The ecological effects studies are used to assess potential toxicity to non-target organisms; data requirements are established in 40 CFR 158.490 (Terrestrial and aquatic organisms data requirements), 158.540 (Plant protection data requirements), and 158.590 (Nontarget insect data requirements).

All non-target terrestrial and aquatic animal toxicity studies, and aquatic plant studies, are performed using the technical grade active ingredient (TGAI). Non-target terrestrial plant toxicity tests are performed with pesticide in a formulated product (as sold to users). Aquatic fish and invertebrate toxicity studies using formulated product are also required if the use of the formulation is expected to lead to transport to water bodies, either directly or through runoff. The potential exposure and toxicity of each pesticide are considered to characterize the potential of ecological risk.

The present environmental fate assessment is based on regulatory environmental fate studies submitted to the Agency to support the registration of glyphosate salts and their formulated pesticide products. These studies were conducted under Good Laboratory Practices (GLP), as required under the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA). The studies used in the assessment have been deemed acceptable and have served to generate previous environmental and ecological risk assessments for glyphosate.

The number and types of environmental fate studies required for each pesticide depends on its proposed use pattern (terrestrial, aquatic, forestry, indoors, greenhouse). The required studies consist of a series of core laboratory studies for all pesticides and field dissipation studies for pesticides used on outdoor crops (terrestrial/aquatic), non-crops (terrestrial/aquatic), and forestry. Each study provides specific data that, together with the physical and chemical properties of the pesticide, are then combined to produce an integrated environmental fate assessment and to identify the potential of the pesticide to leach to groundwater, and/or reach surface water, and/or bioaccumulate in aquatic organisms. The data are also used as input parameters in models to estimate exposure concentrations in the environment. Monitoring data, if available, are also incorporated into the assessment.

The limited number of species and environmental systems tested can introduce a degree of uncertainty when attempting to extrapolate the data outside the experimental conditions of the studies, such as different soils, geographical regions, and ecosystems. As part of the overall risk characterization of a pesticide, the Agency also identifies uncertainties associated with the available data and those introduced by the assumptions needed to estimate concentrations using models.

III Ecological Risk Assessment

The Agency does not expect any risk to birds and mammals based on dietary exposure to active ingredient glyphosate. Acute avian dietary studies using bobwhite quail and mallard ducks resulted in no mortality at concentrations up to 5200 ppm, and no reproductive effects were seen up to 1000 ppm, the highest levels tested. Toxicity tests involving feeding or direct contact of honey bees to glyphosate also resulted in no mortality at the highest rate tested (100 micrograms/bee). Acute LD₅₀ values could not be established in oral and dermal mammalian studies at concentrations up to and including 5000 mg/kg, and chronic mammalian effects were only seen in a series of studies at 1000 mg/kg/day or higher. The Agency waived the requirement for an acute inhalation study for mammals with active ingredient glyphosate since no respiratory or systemic toxicity was seen following subchronic inhalation exposure in rats.

Risk to non-target terrestrial plants is likely from exposure to glyphosate as a result of its use in the coca eradication program. Glyphosate is a foliarly applied, broad spectrum herbicide effective at very low exposure rates. Vegetative vigor studies for North American crops reviewed by the Agency in 1999 indicate that 25% of exposed plants can be damaged by exposure to glyphosate applied at rates as low as 0.07 lb ai/A.

Table 1. Vegetable Vigor Toxicity of Glyphosate Wettable Powder to US Crops

^*Determined by linear interpolation.

The application rate of glyphosate recommended by the State Department for the coca eradication program is 3.34 lb ai/A in acid equivalents. This is well above the rates listed in the table above. AgDrift modeling of potential spray drift from the use in Colombia (detailed below) simulates that non-target plants hundreds of feet away may be exposed to a fraction of this glyphosate application. Based on the toxicity data for North American crops, AgDrift indicates the possibility that 50% of young crop plants would be expected to show measurable reductions in dry weight from150 to nearly 600 feet downwind (depending on spray and wind conditions). Some affected plants would likely recover while more sensitive plants may die, have reduced reproductive success, or reduced yields (crop plants).

Laboratory studies indicate glyphosate is slightly toxic to fish, invertebrates and aquatic plants. The lowest resulting acute LC₅₀ values (concentrations at which half the test animals died) were in parts-per-million (ppm) for active ingredient glyphosate. For instance, the most sensitive freshwater fish (fathead minnow) had an LC₅₀ of 85 ppm, while chronic effects were not seen in another study at the highest test concentration of 26 ppm. The EC₅₀ (level at which adverse effects are seen in half the test animals) for the freshwater invertebrate Daphnia magna was 134 ppm, and the chronic NOEL 50 ppm.

OPP exposure models indicate that surface-water exposure in the parts-per-billion could be expected from the use on coca. OPP also considered a more conservative exposure scenario of the direct application of 3.75 lb acid eq./acre of glyphosate to a 1-acre, 6-foot deep pond². The calculated maximum concentration of 230 ppb is well below the glyphosate toxicity values measured for aquatic organisms in the laboratory. Therefore, aquatic organisms should not be at risk from exposure to glyphosate. The environmental fate assessment which is the basis of this exposure calculation is described in the following section.

²Salts of glyphosate (isopropylamine, diamine, and trimesium) are registered in the United States for a wide variety of agricultural and non-agricultural uses. All of the salts of glyphosate generate the "free acid of glyphosate" (glyphosate acid), the actual herbicide-active chemical. The glyphosate acid equivalents vary from salt to salt to salt, as it depends on the ratio of the molecular weight of the glyphosate free acid to that of the salt. For this reason, application rates are generally expressed in terms of "glyphosate acid equivalents" when estimating exposure concentrations of glyphosate in water and soil. The ratio of glyphosate acid to the glyphosate isopropylamine salt is 0.75. Thus, each pound of this salt is equivalent to 0.75 pounds of glyphosate acid (or 1 g of the salt is equivalent to 0.75 g of the acid)

It is possible that much greater exposure could occur from direct overspray of water bodies much smaller than a 1-hectare, 6-foot deep pond, but such simulation is not a standard component of Agency risk assessments. The product label of the specific glyphosate product DoS indicates is being used against coca, and the DoS application guidelines, prohibit direct overspray of water bodies. It is possible that some ecologically important water bodies too small to appear on maps could be sprayed directly in a project as large as the coca eradication program. EPA has registered other glyphosate products for direct application to aquatic sites to kill undesirable vegetation.

Freshwater aquatic plants also seem unlikely to be at risk from exposure to active ingredient glyphosate. Submitted studies resulted in EC₅₀ values of 12.5 ppm for green algae (Selenastrum capricornatum), 21.5 ppm for duckweed (Lemna gibba) and 38.6 ppm for the freshwater diatom Navicula pelliculosa. These values are well beyond the 230 ppb exposure calculated for direct overspray in the pond simulation described above.

Ecological toxicity studies submitted to EPA for some of the formulations of glyphosate products that EPA has registered have shown them to be more toxic than glyphosate alone. The results of these studies indicate that the formulations will pose a risk primarily to non-target plants, as described above. For instance, the minimum bluegill sunfish LC₅₀ of 5.8 ppm reported for a 41.8% glyphosate formulation in EPAs glyphosate reregistration eligibility document (RED, 1993) is 20 times more toxic than the bluegill sunfish LC₅₀ observed for technical glyphosate, but is still much higher than exposure levels expected in the environment. The bluegill sunfish LC₅₀ for a test with surfactant MONO818 by itself was 1.0 ppm.

The risk to non-target terrestrial and aquatic animals from formulated glyphosate used for coca eradication is uncertain because the Agency does not have relevant toxicity data for the Colombian formulation, nor for the adjuvant Cosmo-Flux 411F. An adjuvant is a subsidiary ingredient or additive in a mixture that adds to the effectiveness of the primary or active ingredient. Adjuvants are most commonly added to tank mixes of pesticide products before they are applied. Further discussion is provided in the Health Effects Divisions assessment of the coca eradication program.

Potential Spray Drift of Glyphosate

The AgDrift model (version 2.01) was used to estimate downwind deposition of aerial applications of herbicide sprays during coca eradication efforts. The aerial part of the AgDrift model, which was used in this assessment, was developed from USDA Forest Service models designed to estimate deposition of forestry applications. The model has been the subject of a Scientific Advisory Panel (SAP) meeting³ and showed a good correlation with field trial data of downwind deposition. Reviews and descriptions of AgDrift have been published.⁴

AgDrift uses a number of input parameters associated with the application equipment and the meteorology during application in calculating deposition levels. An attempt was made to enter important input parameters appropriate for coca eradication applications in Colombia as described by the Department of State (DoS) in their presentation⁵ to the Office of Pesticide Programs (OPP) or in documents provided by DoS to OPP. Many input parameters in AgDrift do not greatly affect deposition levels and a number of default inputs were used for these parameters. The inputs considered to be more important in determining drift levels that were used to model coca eradication spraying are listed in Table 1 below.

³http://www.epa.gov/scipoly/sap/1997/december/spraydrift.htm
⁴Hewitt AJ, DR Johnson, JD Fish, CG Hermansky, and DL Valcore. 2002. Development of the Spray Drift Task Force database for aerial applications. Environmental Toxicology and Chemistry. 21(3) pp. 648-658.Teske ME, SL Bird, DM Esterly, TB Curbishley, SL Ray, and SG Perry. 2002. AgDRIFT: A model for estimating near-field spray drift from aerial applications. Environmental Toxicology and Chemistry. 21(3) pp. 659-671. Bird SL, SG Perry, SL Ray, and ME Teske. 2002. Evaluation of the AGDISP aerial spray algorithms in the AgDRIFT model. 2002. Environmental Toxicology and Chemistry. 21(3) pp. 672-681.
⁵April 18, 2002. Crystal City, Arlington,VA.

There are a number of general uncertainties associated with AgDrift modeling to estimate spray drift from coca spraying in Colombia. The AgDrift model is intended to represent a flat area with uniform vegetation while coca cultivation is reported to occur in some instances on irregular topography with scattered trees and shrubs. The AgDrift model is not intended to model spray drift under very stable atmospheric conditions (e.g. temperature inversions). Although coca eradication efforts attempt to avoid these conditions they can be difficult to detect.

Table 1. AgDrift model (version 2.01) inputs used to model spray drift deposition from coca eradication spraying in Colombia.

Parameter	Input	Comment
Aircraft	Air Tractor AT-802A	The spray program is reportedly buying AT-802A aircraft for spraying operations. The AT-802A is heavier than the currently used Ayers T-65 Thrush which may result in slightly higher drift levels.
Spray volume	2.53 gal/acre	Reported in (1)
Nonvolatile rate	5.49 lbs/acre	Calculated from values reported on p. 4 in (1): Water = 0.74 Surfactant and a.i. = 0.26 2.53 gal/acre * (0.26 Surfactant and a.i.) = 0.66 gal/acre 0.66 gal/acre * (8.35 lbs/gal) = 5.49 lbs/acre
Droplet spectrum	ASAE medium or ASAE very coarse to extremely coarse	In (1) the droplet size spectrum is reported to have a volume median diameter of 300 to 1,500 microns. This is a large range for one of the most important factors in estimating off-target drift. Two categories of droplet size spectrum were chosen to represent the range. ASAE medium sprays have a VMD of approximately 300 microns. The ASAE very coarse to extremely coarse is the coarsest ASAE spray available in AgDrift 2.01with a VMD of 520 microns. It was not stated if droplet size was measured under application conditions. In the presentation at OPP offered by the DoS the VMD during application was said to be 200 to 300 microns.
Wind speed	3 & 10 mph	Reference (1) p.6 states missions are canceled if wind speed measured at the airport is above 10 mph. Wind speed at the target site may vary but 10 mph was used as the best available input for modeling.
Relative humidity	75%	Reference (1) p.6 states missions are canceled if relative humidity measured at the airport is above 75%. Relative humidity at the target site may vary but 75% was used as the best available input for modeling.
Temperature	90 degrees F	Reference (1) p.6 states missions are canceled if temperature measured at the airport is above 90 degree F. Temperature at the target site may vary but 90 degree F was used as the best available input for modeling.
Release height	100 ft	Reference (1) p. 6 states the altitude above spray targets is normally less than 100 feet. This value was used as the best available input for modeling.
Spray lines	4	Based on video of spraying operations with multiple aircraft, the number of spray lines used in modeling was 4.

(1) Chemicals Used for the Aerial Eradication of Illicit Coca in Colombia and Conditions of Application. An undated, unsigned, 9-page document provided to OPP by the State Department.

In addition to the general uncertainties above, there are also uncertainties associated with the inputs used for modeling the spray applications. Droplet size is one of the most important parameters affecting drift of pesticides. There is uncertainty as to the droplet size spectrum used in aerial coca spraying. Documentation supplied by the DoS⁶ describes the droplet size used in terms of the volume median diameter (VMD) which is the droplet size for which half of the volume of spray is contained in droplets with smaller diameter and half of the spray is contained in droplets of larger diameter. The VMD was stated to range from 300 to 1500 microns which is a wide range. In the DoS presentation the VMD was stated to be 200 to 300 microns during application conditions. In addition to the wide range of VMD values presented, VMD is not a good descriptor of droplet size spectra for estimating spray drift. Spray drift is predominately associated with finer sprays and VMD does not define the amount of small droplets contained in spray. Although specific data on droplet size under application conditions was not provided, it is unlikely that very coarse sprays would be achievable due to shearing effects of releasing droplets at high airspeeds. Large droplets released into the turbulence created by an aircraft traveling in excess of 120 mph tend to break into smaller more driftable droplets.

⁶Chemicals Used for the Aerial Eradication of Illicit Coca in Colombia and Conditions of Application. An undated, unsigned, 9-page document provided to OPP by the State Department.

Other uncertainties associated with inputs include inputs for meteorology and release height. AgDrift modeling requires site-specific inputs for meteorology. In coca eradication efforts (as well as agricultural applications in the US) wind speed, temperature and humidity are measured at the airport which may not be representative of these parameters at the application site. The applicator is ultimately given the responsibility of determining if conditions at the target site are acceptable. DoS reports that the coca eradication program selects experienced applicators for spray missions with the expectation they will better be able to identify unacceptable conditions and make applications within specified parameters.

In order to capture the range of deposition values expected during coca eradication applications, AgDrift was run with two droplet size spectra and at two wind speeds. The droplet size spectra were extremely coarse to very coarse and medium. The definitions refer to the American Society of Agricultural Engineering (ASAE) Standard 572 definition of droplet size spectra. The wind speeds used were 3 mph and 10 mph. AgDrift was run in tier 3 to estimate downwind depositions shown in Figure 1 below.

Figure 1 shows the lowest levels of drift are associated with applications using the extremely coarse to very coarse sprays at a 3 mph wind speed. The highest levels of drift are associated medium sprays at wind speeds of 10 mph. Downwind deposition levels from coca eradication spraying is likely to be bounded by these estimates. The effect level for 50% of young plants[at] is based on glyphosate toxicity studies on ten crop plants. At the level corresponding to approximately 11% of the application rate, 50% of plants species would be expected to show measurable reductions in dry weight. Of the affected plants some would likely recover while more sensitive plants may die, have reduced reproductive success, or reduced yields (crop plants).

V. Environmental Fate and Transport Assessment of Glyphosate

The major route of transformation of glyphosate identified in laboratory studies is microbial degradation. In the field, glyphosate dissipation appears to correlate with climate, being more persistent in cold than in warm climates. Dissipation of glyphosate in Colombia may therefore be more rapid than in the U.S. Glyphosate was not observed in laboratory studies to break down by abiotic processes such as hydrolysis and direct photolysis.

Glyphosate is very soluble, and has a low potential to volatilize, but adsorbs strongly to soils and sediments. Therefore, glyphosate does not have a high potential to leach to ground water or reach surface water as dissolved runoff. However, glyphosate has the potential to contaminate surface water as a result of residues adsorbed to soil particulates suspended in runoff water. Offsite exposure is also possible due to spray drift or inadvertent direct overspray.

Glyphosate belongs to the glycine family of herbicides. Glyphosate is a phosphono derivative of glycine, the simplest of all of the amino acids. It works as an herbicide by inhibiting the enzyme A5- enolpyryl-shikimate-3-phosphate[at] synthase (i.e., it is an ESPS inhibitor). Glyphosate is not an organophosphate and it is not an inhibitor of cholinesterase activity.

Chemical name: N-(phosphonomethyl)glycine

Chemical Abstracts Registry Number: 1071-83-6

Chemical structure:

The major route of transformation of glyphosate in soils is microbial degradation. In laboratory studies in soils incubated under aerobic conditions, ¹⁴C-labeled glyphosate degraded with half-lives ranging from 1.85 to 5.4 days in two sandy loam soils, and 2.06 days in a silt loam. These studies were conducted in the absence of light and at 25E C. The major degradate that formed in these soils was aminomethyl phosphonic acid (AMPA), which reached a maximum of ca. 29% at 40 days, but declined afterwards. After 1 year, $70% of the applied radioactivity was found as ¹⁴CO₂, indicating that the ultimate fate of glyphosate and AMPA is mineralization (i.e., formation of CO₂ and inorganic carbonates and bicarbonates). Photolysis on soil is not a degradation route for glyphosate, as the half-lives under both irradiated and dark conditions were 6.6 days and any degradation that occurred during the studies was likely to be microbial.

Batch-equilibrium adsorption studies in a wide range of soils and sediments from the US and United Kingdom have shown that glyphosate and AMPA adsorbed strongly to soils, with adsorption coefficients (K_ads) ranging from 9.4 to 700 mL/g. Therefore, glyphosate has a low potential to leach to groundwater or reach surface water by runoff, but may enter surface water through soil erosion.

Terrestrial field dissipation studies conducted with a formulation of the non-radiolabeled isopropylamine salt at an application rate of 10.7 lb of salt/acre (7.95 lb acid equivalent/acre) showed that dissipation (i.e., transformation plus transport) was slower in colder than in warmer climates. The reported half-lives at each site were 2.9 days in Texas, 13 to 20 days in Georgia, California and Arizona, 127 days in New York, and 140 days in Iowa. Glyphosate and AMPA were found predominantly in the 0 to 6 inch layers, indicating that they are not potential leachers. All of these studies were conducted with the formulation applied directly to soil in bare ground plots. Glyphosate is a foliar herbicide that is not applied directly to soils and would only reach soil by wash-off from foliage. Therefore, direct application to soils in bare ground plots represent a worse case of glyphosate use.

The low vapor pressure of glyphosate (4.3 x 10^-10, mmHg at 25E C and 1.8 x 10^-10, mmHg at 45E C) suggests that it has low potential to volatilize from soils.

Laboratory studies suggest that abiotic hydrolysis (i.e., hydrolysis in the absence of microorganisms) is not a primary degradation pathway for glyphosate. Glyphosate remained stable for at least 30 days in sterile aqueous buffered solutions of pH 5, 7 and 9 that were kept in the dark at 25E C. Buffered solutions of glyphosate were stable to sunlight, suggesting that direct photolysis is not a likely degradation pathway for glyphosate. EPA has no data at this time to assess any contribution of indirect photolysis in natural waters.

The half-life of glyphosate in a silty clay loam sediment incubated under anaerobic conditions (flooded plus a nitrogen atmosphere) was estimated as 8.1 days and 199 days in a water-clay loam sediment system. Most of applied radioactivity was found in the sediment phase. The major metabolite was AMPA, which also remained associated with the sediment. The observed half-live in an aerobically incubated silty clay loam sediment was 7 days.

Aquatic field dissipation data showed that the half-life of glyphosate in water used as irrigation source was 7.5 days (farm pond in Missouri, 408816-01). In Michigan, Georgia, and Oregon pond and stream water systems, glyphosate dissipated rapidly immediately after treatment. Accumulation was higher in the pond than in the stream sediments (415528-01).

The low Henrys Law constant of glyphosate suggests that glyphosate is not likely to volatilize from water. Given the strong adsorption to sediments, glyphosate is most likely to be associated with the sediment. The very low n-octanol -water partition coefficient is indicative that glyphosate is not likely to bioaccumulate in aquatic organisms

Residues of an aerially applied glyphosate product at a rate of 3.75 lb of acid equivalents/acre declined rapidly from tree foliage in less than 1 day at the Michigan and Georgia sites and less than 14 days at a site in Oregon. The rate of dissipation in the foliage (wash-off) correlated with the amount of rainfall and leaf drop, which also determines the movement of glyphosate and AMPA through the forest ecosystem. The average half-life for overall dissipation from the forest ecosystem was 100 days for glyphosate (35 to 158 days) and 118 days for AMPA (71 to 165 days). In all cases, the maximum combined residue of glyphosate and AMPA in soil was less than 5 ppm , but the amount of residues declined with time (MRID 415528-01).

VI. Risk Characterization

The ongoing use of a glyphosate spray for coca eradication is likely to pose a risk to non-target plants. Vegetative vigor toxicity laboratory tests performed using a formulated glyphosate product (glyphosate acid WP 48.3%) on North American crops indicated toxicity to terrestrial plants with applications of less than 1.0 lb of active ingredient per acre (lb ai/acre) (Table I). The State Department proposes to use a rate of 3.34 lb acid equivalents/acre for direct, aerial application to coca. A second application is possible if fields are replanted, or the first is determined after 3 to 6 months to have been inadequate. The product reported by DoS as used in Colombia has a formulation which matches the formulation of a product that is registered, but not used, in the United States.

AgDrift modeling of potential spray drift indicates that non-target plants hundreds of feet away may be exposed to a fraction of this glyphosate application. Based on the toxicity data for North American crops, AgDrift indicates the possibility that 50% of young plant crops would be expected to show measurable reductions in dry weight from 150 to nearly 600 feet downwind (depending on spray droplet size and wind conditions). As detailed below, there are several hundred non-target terrestrial plant incident reports in the Agencys Ecological Incident Information System (EIIS) database connected with the use of glyphosate products.

There is uncertainty whether crops or other plants in Colombia, whether similar to crops tested in the United States or not, would be affected similarly at the same exposure levels. However, since glyphosate is an effective, broad spectrum herbicide, risk to non-target plants outside of the application zone would be expected. The Agencys EIIS database includes several hundred reports of possible non-target plant incidents in the United States attributed to use of glyphosate.

This use of the active ingredient glyphosate itself would not pose a significant direct risk to terrestrial or aquatic animals, although temporary secondary adverse effects from the loss of habitat in the spray area may occur. Neither acute nor chronic adverse effects were observed in mammalian and avian laboratory toxicity tests using the active ingredient alone. Mortality was observed in fish and aquatic invertebrate studies. However, the resulting acute LC₅₀ values (concentrations at which half the test animals died), and lowest effect levels for chronic effects, were in parts-per-million. Toxicity endpoints for aquatic plants also ranged from 0.85 to 39.9 ppm. Considerably lower surface-water exposure, in the parts-per-billion, could be expected from the use on coca using runoff simulations from Agency exposure models. The Agency considered an even more conservative scenario, estimating the concentration that would result from the direct application of 3.75 lb acid eq./acre of glyphosate to a 1-acre, 6-foot deep pond. The calculated maximum concentration of 230 ppb is well below the toxicity values measured for aquatic organisms in the laboratory.

It is possible that much greater exposure could occur from direct overspray of water bodies much smaller than a 1-acre, 6-foot deep pond, but such simulation is not a standard component of Agency risk assessments. It is possible that some ecologically important water bodies too small or ephemeral to appear on maps could be sprayed directly in a project as large as the coca eradication program.

There are several aspects of an ecological risk assessment which, while included as a regular part of the U.S. registration process, are not relevant to the use of glyphosate on coca. In considering the risk of a chemical to terrestrial or aquatic animals in the United States, the Agency has set levels of concern (LOCs) at between 5 to 20% of the acute toxicological endpoints for further consideration of risk to endangered species, or eligibility of a chemical for Restricted Use (application permitted only by Certified Pesticide Applicators.) Our document does not include endangered species because the Agency lacks information on the species which might be present in areas of spraying. In addition, the Health Effects Division determined that results of mammalian toxicology studies did not warrant the establishment of a chronic toxicological endpoint for the calculation of a Reference Dose (RfD, a reference endpoint for human health risk assessment ) for glyphosate. Since there is not a chronic endpoint, a drinking water exposure assessment is not necessary for the use of glyphosate on coca.

Although the measured toxicity and estimated exposure indicate that only non-target plants are likely to be adversely affected by the use on coca, there are important uncertainties that should be considered. One of these, which was emphasized by the Amazon Alliance in a memo to the Agency, is the extrapolation of North American data to the conditions and wildlife found in Colombia. The toxicity of a pesticide to different classes of animals and plants can vary widely among species within an individual ecosystem. The Agency uses the test species as surrogates for other North American species not tested, but has little experience with tropical flora and fauna. Similarly, laboratory and field estimates of the environmental fate of pesticides, including potential surface- water contamination, are performed with North American soils, hydrology and climate data.

The potentially most important uncertainty in this risk assessment concerns differences in the formulation and tank mix for use in Colombia from those used in the United States. Toxicity studies indicate that U.S. formulations of glyphosate are more toxic to non-target animals than the technical product alone, but not toxic at levels of expected exposure. However, none of the ecological effects studies submitted to or encountered by the Agency for glyphosate were performed with the formulation that the DoS has indicated is used in Colombia, which may contain different types of cationic surfactants than those in formulations for which the Agency has data. Consultant Jeremy Bigwood presented a literature search of over 200 citations to the Ecuadorian Minister of the Environment in March 2002, stating that A(t)here have been NO scientific investigations on the past or present formulations being used in Colombia.[at]

In addition, the Agency does not have ecological toxicity information on adjuvant Cosmo-Flux 411F, which is neither manufactured nor sold in the United States. There is some inconsistency in the description of Cosmo-Flux in the two available labels, in Spanish and in English. However, all of the individual ingredients (surfactants) which comprise the adjuvant are substances with low oral and dermal mammalian toxicity. The toxicity of the blend of these surfactants is not known; although the Agency often requires formulation toxicity data for non-target plants and aquatic organisms, tank-mix adjuvants are not required to be included in these studies.

Reports From External Sources

The effect of Cosmo-Flux 411F or its individual ingredients on non-target organisms is unknown, although Mr. Bigwood suggests some possible effects in his report to the government of Ecuador. Mr. Bigwood cites studies from the Western Australia Department of Environmental Protection (WADEP) as indicating that a formulation equivalent to that which the Department of State has indicated is used in Colombia Acan be acutely toxic to adult frogs and tadpoles at the recommended application rates (1.8 to 5.4 kg/ha).[at] The Agencys tox database cites LC₅₀ values for two Australian frog species (Crinia insignifera and Litoria moorei) of 40 and 8 ppm, respectively, presumably from the same studies. Such concentrations are greater than those likely to occur from transport of glyphosate in runoff to ponds. These concentrations might be possible for frogs exposed by direct overspray, such as tree frogs. However, the Agency does not have a method for estimating the possible dietary intake of pesticides for tree frogs; the Agencys model for dietary exposure of terrestrial animals is based on agricultural field data collected in the United States. Extrapolation of toxicity to Australian frog species to Colombian species includes significant uncertainty.

Suggestions of risk to other organisms in Mr. Bigwoods report are less specific. This report states that (t)oxicity of glyphosate formulations in riverine systems is not merely limited to fish, but also to amphibians, insects, crawfish and water fleas, and undoubtedly to other species found in rivers and other bodies of water.[at] This is followed by a discussion of toxicity values for a surfactant other than Cosmo-Flux 411F . As described above, while aquatic exposure to glyphosate itself (or U.S. formulations) is not likely to pose a risk to aquatic animals and plants, data on the toxicity of tank- mix adjuvant Cosmo-Flux 411F to these organisms would be required to assess formulation and tank-mix risks. Tank-mix ecological toxicity data are not routinely provided in the U.S. pesticide registration process.

The Agency would need to obtain and review literature studies cited in Mr. Bigwoods report in order to comment on other suggested risks. Mr. Bigwood suggests that glyphosate enhances the growth of pathogenic fungi according to several research papers.[at] Based on the titles of the papers, most of the studies cited concern the effect of fusarium fungi on glyphosate efficacy. The Agency cannot comment on the magnitude of this effect without reviewing the data, nor on the potential for risk from this effect. The effects of pesticide applications on microbiota is not a standard component of the Agencys risk assessments.

There are several hundred non-target terrestrial plant incident reports in the Agencys Ecological Incident Information System database connected with the use of glyphosate products. This is consistent with the risk assessment above, which suggests that the efficacy of glyphosate is such that non-target plants hundreds of feet away could be at risk from glyphosate in spray drift. The variety of crops, ornamentals and trees included in the EIIS reflect the wide spectrum of glyphosate efficacy.

There are a small, limited number of reported incidences to fowl, fish and a dog allegedly resulting from label use of glyphosate products. Further analysis of the data on the actual incidence report forms indicates that other factors or other pesticides may explain the adverse effects. For instance, one incident was apparently due to overstock and improper oxygen levels in a catfish pond, and another to a spill of several barrels of Roundup directly into a creek. Another fish kill was more likely attributable to diuron runoff into a pond than to glyphosate exposure. Although glyphosate was associated with these incidents, the fact that other conditions or pesticides were likely responsible is consistent with the fact that glyphosate has very low mammalian, avian and aquatic acute and chronic toxicity.

Only one incident appears to be attributable to glyphosate, involving 2 iguanas that ate dandelions apparently sprayed with a Roundup product. However, the reptiles exhibited signs of neurological effects (shock, depression and tremor) which are not generally associated with glyphosate toxicity and which may be due to the inert ingredients in the formulation. The Agency does not have any test protocols nor does it require toxicity testing on reptiles for any pesticide. Thus, information on this class of animals in general is an uncertainty.

Mr. Bigwoods report for the Government of Ecuador states that exposure to a dried formulation containing both glyphosate and a surfactant in Roundup (the commercial product being used) caused the death of over 50%[at] of several beneficial insect species. The report does not indicate the level of exposure or the nature of the dried formulation[at] that caused these effects. Therefore, the Agency cannot comment on the relevance of these data to aerial spray of the formulation. Mr. Bigwood cites a Los Angeles Times article in which Guatemalan farmers contend that the discontinued poppy eradication program has devastated the areas traditional agricultural base, particularly tomatoes and bees. Honeybee oral and contact toxicity tests provided to the Agency using technical glyphosate (active ingredient only, not the formulated product) could not establish an LC₅₀ at concentrations up to 100 micrograms per bee. However, these studies cannot be used to rule out the possible toxicity of the Colombian formulation and tank mix to beneficial insects.