Glyphosate (EHC 159, 1994)

INTERNATIONAL PROGRAMME ON CHEMICAL SAFETY

ENVIRONMENTAL HEALTH CRITERIA 159

GLYPHOSATE

This report contains the collective views of an international group of
experts and does not necessarily represent the decisions or the stated
policy of the United Nations Environment Programme, the International
Labour Organisation, or the World Health Organization.

First draft prepared by Dr H. Mensink and
Dr. P. Janssen, National Institute of Public
Health and Environmental Hygiene,
Bilthoven, The Netherlands

Published under the joint sponsorship of
the United Nations Environment Programme,
the International Labour Organisation,
and the World Health Organization

World Health Orgnization
Geneva, 1994

The International Programme on Chemical Safety (IPCS) is a
joint venture of the United Nations Environment Programme, the
International Labour Organisation, and the World Health
Organization. The main objective of the IPCS is to carry out and
disseminate evaluations of the effects of chemicals on human health
and the quality of the environment. Supporting activities include
the development of epidemiological, experimental laboratory, and
risk-assessment methods that could produce internationally
comparable results, and the development of manpower in the field of
toxicology. Other activities carried out by the IPCS include the
development of know-how for coping with chemical accidents,
coordination of laboratory testing and epidemiological studies, and
promotion of research on the mechanisms of the biological action of
chemicals.

WHO Library Cataloguing in Publication Data

Glyphosate.

(Environmental health criteria ; 159)

1.Glycine - analogs and derivatives 2.Herbicides
3.Environmental exposure I.Series

ISBN 92 4 157159 4 (NLM Classification: WA 240)
ISSN 0250-863X

The World Health Organization welcomes requests for permission
to reproduce or translate its publications, in part or in full.
Applications and enquiries should be addressed to the Office of
Publications, World Health Organization, Geneva, Switzerland, which
will be glad to provide the latest information on any changes made
to the text, plans for new editions, and reprints and translations
already available.

^(c) World Health Organization 1994

Publications of the World Health Organization enjoy copyright
protection in accordance with the provisions of Protocol 2 of the
Universal Copyright Convention. All rights reserved.

The designations employed and the presentation of the material
in this publication do not imply the expression of any opinion
whatsoever on the part of the Secretariat of the World Health
Organization concerning the legal status of any country, territory,
city or area or of its authorities, or concerning the delimitation
of its frontiers or boundaries.

The mention of specific companies or of certain manufacturers'
products does not imply that they are endorsed or recommended by the
World Health Organization in preference to others of a similar
nature that are not mentioned. Errors and omissions excepted, the
names of proprietary products are distinguished by initial capital
letters.

CONTENTS

ENVIRONMENTAL HEALTH CRITERIA FOR GLYPHOSATE

1. SUMMARY

1.1. Identity, physical and chemical properties,
and analytical methods
1.2. Sources of human and environmental exposure
1.3. Environmental transport, distribution and
transformation
1.4. Environmental levels and human exposure
1.5. Kinetics and metabolism in laboratory animals
and humans
1.6. Effects on laboratory mammals, and in vitro
test systems
1.7. Effects on humans
1.8. Effects on other organisms in the laboratory
and field

2. IDENTITY, PHYSICAL AND CHEMICAL PROPERTIES, AND ANALYTICAL
METHODS

2.1. Identity
2.2. Physical and chemical properties
2.3. Formulations
2.4. Conversion factors
2.5. Analytical methods
2.5.1. Sample handling and preparation
2.5.2. Analytical methods

3. SOURCES OF HUMAN AND ENVIRONMENTAL EXPOSURE

3.1. Anthropogenic sources
3.1.1. Production levels and processes
3.1.2. Uses
3.1.3. Drinking-water

4. ENVIRONMENTAL TRANSPORT, DISTRIBUTION AND TRANSFORMATION

4.1. Transport and distribution between media
4.1.1. Water
4.1.2. Soil sorption
4.1.3. Mobility in soils
4.1.4. Dissipation from the soil in the field
4.1.5. Uptake and dissipation from plants
4.1.6. Ingestion by animals
4.2. Abiotic degradation
4.2.1. Hydrolytic cleavage
4.2.2. Photodegradation

4.3. Biodegradation
4.4. Bioaccumulation
4.5. Waste disposal

5. ENVIRONMENTAL LEVELS AND HUMAN EXPOSURE

5.1. Environmental levels
5.2. General population exposure
5.3. Occupational exposure during manufacture,
formulation, or use

6. KINETICS AND METABOLISM IN LABORATORY ANIMALS AND HUMANS

6.1. Absorption
6.2. Distribution
6.3. Metabolic transformation
6.4. Elimination and excretion
6.5. Retention and turnover

7. EFFECTS ON LABORATORY ANIMALS AND IN VITRO TEST SYSTEMS

7.1. Single exposure
7.2. Short-term exposure
7.2.1. Oral studies
7.2.2. Dermal studies
7.2.3. Inhalational studies
7.3. Long-term toxicity and carcinogenicity
7.4. Skin and eye irritation; sensitization
7.5. Reproductive toxicity, embryotoxicity and
teratogenicity
7.5.1. Teratogenicity studies
7.5.2. Reproduction studies
7.6. Mutagenicity and related end-points

8. EFFECTS ON HUMANS

8.1. Cases of intentional and accidental exposure
8.2. Occupational exposure
8.3. Subpopulations at special risk

9. EFFECTS ON OTHER ORGANISMS IN THE LABORATORY AND FIELD

9.1. Laboratory experiments
9.1.1. Microorganisms
9.1.1.1 Water
9.1.1.2 Soil
9.1.2. Aquatic organisms
9.1.2.1 Plants
9.1.2.2 Invertebrates
9.1.2.3 Vertebrates

9.1.3. Terrestrial organisms
9.1.3.1 Plants
9.1.3.2 Invertebrates
9.1.3.3 Vertebrates
9.2. Field observations
9.2.1. Microorganisms
9.2.1.1 Water
9.2.1.2 Soil
9.2.2. Aquatic organisms
9.2.2.1 Plants
9.2.2.2 Invertebrates
9.2.2.3 Vertebrates
9.2.3. Terrestrial organisms
9.2.3.1 Plants
9.2.3.2 Invertebrates
9.2.3.3 Vertebrates

10. EVALUATION OF HUMAN HEALTH HAZARDS AND EFFECTS ON THE
ENVIRONMENT

10.1. Human health hazards
10.2. Evaluation of effects on the environment
10.2.1. Exposure levels and toxic effects
10.2.2. Hazard evaluation for aquatic organisms
10.2.3. Hazard evaluation for terrestrial organisms

11. RECOMMENDATIONS FOR PROTECTION OF HUMAN HEALTH

12. FURTHER RESEARCH

REFERENCES

RESUME

RESUMEN

WHO TASK GROUP ON ENVIRONMENTAL HEALTH CRITERIA FOR GLYPHOSATE

Members

Dr S. Dobson, Institute of Terrestrial Ecology, Monks Wood
Experimental Station, Huntingdon, United Kingdom
(Chairman)

Dr A.H. El-Sebae, College of Agriculture, Alexandria University,
El Shatby, Alexandria, Egypt

Dr P. Janssen, National Institute of Public Health and
Environmental Hygiene, Bilthoven, The Netherlands

Dr H. Mensink, National Institute of Public Health and
Environmental Hygiene, Bilthoven, The Netherlands

Dr M.S. Morrow, Health Effects Division, Office of Pesticide
Programs, US Environmental Protection Agency, Washington, DC,
USA

Professor R. Nilsson, Department of Scientific Documentation and
Research, National Chemicals Inspectorate, Solna, Sweden^a

Dr R. Ye, National Environmental Protection Agency, Beijing,
People's Republic of China

Observers

Dr C. Hastings, Agricultural Group, Monsanto, Missouri, St.
Louis, USA

Secretariat

Dr M. Gilbert, International Programme on Chemical Safety, World
Health Organization, Geneva, Switzerland (Secretary)

___________

^a Invited but unable to attend.

NOTE TO READERS OF THE CRITERIA MONOGRAPHS

Every effort has been made to present information in the
criteria monographs as accurately as possible without unduly
delaying their publication. In the interest of all users of the
Environmental Health Criteria monographs, readers are kindly
requested to communicate any errors that may have occurred to the
Director of the International Programme on Chemical Safety, World
Health Organization, Geneva, Switzerland, in order that they may be
included in corrigenda.

* * *

A detailed data profile and a legal file can be obtained from
the International Register of Potentially Toxic Chemicals, Case
postale 356, 1219 Châtelaine, Geneva, Switzerland (Telephone No.
9799111).

* * *

This publication was made possible by grant number 5 U01
ES02617-14 from the National Institute of Environmental Health
Sciences, National Institutes of Health, USA.

ENVIRONMENTAL HEALTH CRITERIA FOR GLYPHOSATE

A Task Group on Environmental Health Criteria for Glyphosate
met at the Institute of Terrestrial Ecology, Monks Wood, United
Kingdom, from 23 to 27 August 1993. Dr S. Dobson welcomed the
participants on behalf of the host institution, and Dr M. Gilbert
opened the Meeting on behalf of the three cooperating organizations
of the IPCS (UNEP/ILO/WHO). The Task Group reviewed and revised the
draft monograph and made an evaluation of the risks for human health
and the environment from exposure to glyphosate.

The first draft of this monograph was prepared by Dr H. Mensink
and Dr P. Janssen, National Institute of Public Health and
Environmental Hygiene, Bilthoven, The Netherlands.

Dr M. Gilbert was responsible for the overall scientific
content of the monograph and for the organization of the meeting,
and Dr P.G. Jenkins, IPCS, for the technical editing of the
monograph.

The efforts of all who helped in the preparation and
finalization of the monograph are gratefully acknowledged.

ABBREVIATIONS

a.i. active ingredient

ALAT alanine aminotransferase

AMPA aminomethylphosphonic acid

AP alkaline phosphatase

CHO Chinese hamster ovary

CNS central nervous system

HPLC high-performance liquid chromatography

i.p. intraperitoneal

IPA isopropylamine

MATC maximum acceptable toxicant concentration

NOAEL no-observed-adverse-effect level

NOEC no-observed-effect concentration

1. SUMMARY

1.1 Identity, physical and chemical properties, and analytical
methods

Glyphosate is a weak organic acid consisting of a glycine and a
phosphonomethyl moiety. The empirical formula is C₃H₈NO₅P.
Glyphosate is usually formulated as a salt of the deprotonated acid
of glyphosate and a cation, e.g., isopropylamine or
trimethylsulfonium. The purity of technical grade glyphosate is
generally above 90%. Technical grade glyphosate is an odourless
white crystalline powder with a specific gravity of 1.704, a very
low vapour pressure, and a high solubility in water. The
octanol-water partition coefficient (log K_ow) is -2.8. Glyphosate
is amphoteric and may exist as different ionic species, dependent on
the actual pH.

Determination of glyphosate is in general laborious, complex,
and costly. Derivatization with fluorogenic substances is the most
common method and may be applied pre- or post-column. Determination
is usually carried out with high performance liquid chromatography
or gas liquid chromatography. Limits of determination for glyphosate
in water, plants, soil and human urine, are 0.02-3.2 µg/litre,
0.01-0.3 mg/kg, 0.05-1 mg/kg and 0.1 mg/litre, respectively.

1.2 Sources of human and environmental exposure

Glyphosate is a post-emergent, systemic and non-selective
herbicide that is used in both agricultural and non-agricultural
areas all over the world. Glyphosate is applied to many crops and in
various commercial formulations. The major formulation is Roundup in
which glyphosate is formulated as the isopropylamine salt.
Recommended application rates do not exceed 5.8 kg a.i./ha and are
dependent on the type of use. Environmental exposure may occur
because of deposition due to drift and accidental releases.

1.3 Environmental transport, distribution and transformation

The most important processes of dissipation that may be
involved after application of glyphosate are complexation in water
with ions, e.g., Ca²⁺ and Mg²⁺, sorption to sediment, suspended
particles in water, and soil, photodegradation in water, uptake by
plants, and biodegradation.

Glyphosate dissipates from the water with DT₅₀ values
(dissipation) ranging from a few days to more than 91 days. Sediment
or suspended particles are shown to be the major sink.

The adsorption coefficients (K_s/l) of glyphosate in
laboratory experiments vary between 8 and 377 dm³/kg for various
soils and clay minerals. No data on the sorption of
aminomethylphosphonic acid (AMPA), the major metabolite, under
laboratory conditions are available.

R_f values of glyphosate do not exceed 0.2 in soil thin-layer
chromatography experiments. Between less than 0.1% and 11% of the
applied activity is recovered in the eluate of soil columns under
leaching conditions simulating an extremely high rainfall. From
field experiments it appears that AMPA is not likely to leach.

Glyphosate dissipates in field experiments from the soil with
DT₅₀ values between 3 and 174 days, mainly depending on edaphic
and climatic conditions. Up to 1.8% of the applied dose dissipated
from the soil due to run-off in some field experiments.

Under laboratory conditions, up to 45% of the applied activity
may be absorbed by treated leaves, and this is followed by a
substantial translocation.

Hydrolysis of glyphosate in sterile buffers is very slow with
DT₅₀ values >> 35 days. Photodegradation in water under natural
conditions occurs with DT₅₀ values < 28 days. No substantial
photodegradation in soil was recorded in a study lasting 31 days.

The time needed for 50% biodegradation of glyphosate in the
whole system of a test with water and sediment is > 14 days under
aerobic conditions and 14-22 days under anaerobic conditions in the
laboratory. The time needed for 50% biodegradation of glyphosate in
the soil is 2-3 days under aerobic conditions.

The major metabolite in soil and water is AMPA. Maximum amounts
of AMPA in soils are approximately 20% of the applied activity under
aerobic conditions and 0.5% under anaerobic conditions. Maximum
amounts of AMPA in sediments are 25% under both aerobic and
anaerobic conditions.

Bioconcentration factors are low in laboratory tests with
invertebrates and fish. Bluegill sunfish in a flow-through test
showed a depuration half-life of 35 days, after being exposed for 35
days. AMPA is recovered in bluegill sunfish up to 21 days after
continuous exposure to glyphosate. Glyphosate has not been detected
in fish living in directly sprayed water in field experiments. In
one experiment, AMPA was detectable in carp up to 90 days after
application. No biomagnification of glyphosate in litter by
herbivorous and omnivorous small mammals in a forest brush ecosystem
was indicated in a field experiment. Concentrations of up to 5 mg
a.i./kg were measured in deermice immediately after spraying in this
experiment.

A range of bacterial strains can degrade glyphosate. Bacteria
capable of using the compound as sole phosphorus, sole carbon or
sole nitrogen source have been identified. Growth is slow compared
to growth on inorganic sources of P, C and N. There is evidence from
the field that bacterial populations adapted to metabolise
glyphosate. The presence of inorganic phosphate inhibits degradation
of glyphosate with some, but not all, bacteria. Biodegradation of
glyphosate may involve co-metabolism with other energy sources.

1.4 Environmental levels and human exposure

Data on the occurrence of glyphosate in environmental biota and
abiota as part of regular monitoring programmes are very scarce.
Data from field experiments in which common agricultural practice is
simulated are used to indicate maximum environmental concentrations:
< 1-1700 µg/litre surface water, 0.07-40 mg/kg dry weight soil,
< 0.05-19 mg/kg dry weight sediment, 261-1300 mg/kg foliage, 5 mg/kg
the viscera of deermice, 1.6-19 mg/kg wild berries, and 45 mg/kg
lichens. The corresponding maximum concentrations of AMPA are:
< 1-35 µg/litre (surface water), 0.1-9 mg/kg dry weight (soil),
< 0.05-1.8 mg/kg dry weight (sediment), 1.7-< 9 mg/kg (foliage),
0.02-0.1 mg/kg (wild berries), and 2.1 mg/kg (lichens). The
above-mentioned concentrations of glyphosate are generally found
immediately after application. The concentration in lichens was
found 270 days after application.

Measurements of daily human intake of glyphosate via food and
drinking-water (total diet studies) are not available. The few data
on occupational exposure indicate that exposure levels for workers
applying glyphosate as the herbicide formulation Roundup are low.

1.5 Kinetics and metabolism in laboratory animals and humans

Technical glyphosate is only partially absorbed from the
gastrointestinal tract. In studies with ¹⁴C-labelled glyphosate,
absorption percentages of 30-36% were found in several species.
Dermal absorption is low. From the herbicide formulation Roundup,
< 5.5% of the glyphosate present is absorbed through the skin
(contact time about 24 h). In body tissues, the highest
concentrations, approximately 1% of the oral dose, are found in
bone. Following a single oral dose, 62-69% is eliminated in the
faeces without absorption. Of the absorbed glyphosate, 14-29% is
excreted in urine and 0.2% or less in expired air. Biliary excretion
following intravenous application was only 5-8%. In lactating goats,
excretion in milk was shown to occur to a minor extent only
(concentration < 0.1 mg/kg whole milk at a dose level of
120 mg/kg diet). Biotransformation of glyphosate occurs to a very
low degree only. The only metabolite, AMPA, accounts for 0.3% of the
dose or less; the rest is unchanged glyphosate. Whole body clearance
(99% of an oral dose) occurs in approximately 168 h.

1.6 Effects on laboratory mammals, and in vitro test systems

In experimental animals, technical glyphosate has very low
acute toxicity by the oral and dermal administration routes; it is
markedly more toxic by the intraperitoneal route than by other
routes. Short-term feeding studies have been conducted in several
species, but few effects were seen in most of these tests. In one
13-week study in mice with technical glyphosate, increased weights
of several organs and growth retardation were observed at
50 000 mg/kg diet. In a 13-week study in rats no effect occurred
(technical glyphosate dose levels up to 20 000 mg/kg diet). In
another 13-week study, lesions of the salivary glands were found in
rats and mice. In mice, the NOAEL was 3125 mg/kg diet; in rats, it
was < 3125 mg/kg diet. These findings were not present in any other
short-term or long-term studies conducted in different strains and
species. The salivary lesions suggest that glyphosate may be acting
as a weak adrenergic agonist.

Long-term toxicity was studied in mice and rats. Few effects
were observed and, in almost all cases, at relatively high dose
levels only. In mice, technical glyphosate produced growth
retardation, hepatocyte hypertrophy or necrosis and urinary bladder
epithelial hyperplasia at 30 000 mg/kg. In rats, the same test
compound produced decreased growth, increased liver weights,
degenerative lens changes and gastric inflammation at 20 000 mg/kg
diet.

The available studies do not indicate that technical glyphosate
is mutagenic, carcinogenic or teratogenic. Two multigeneration
studies were carried out in rats. The main effects of technical
glyphosate were decreased body weights of parent animals and pups
and decreased litter size at 30 000 mg/kg diet. In one reproduction
study, an increase in the incidence of unilateral renal tubular
dilation in F_3b male pups at 30 mg/kg body weight was reported.
The absence of a renal effect in pups at a higher dose level in the
other reproduction study indicates that the reproducibility of this
lesion is uncertain.

1.7 Effects on humans

The available controlled studies are limited to three
irritation/sensitization studies in human volunteers, the results of
which indicated no effect. Several cases of (mostly intentional)
intoxications with technical glyphosate herbicide formulation
Roundup have been reported. In a study on health effects in workers
applying Roundup herbicide formulation, no adverse effects were
found. Available data on occupational exposure for workers applying
Roundup indicate exposure levels far below the NOAELs from the
relevant animal experiments.

1.8 Effects on other organisms in the laboratory and field

Technical grade glyphosate is moderately to slightly toxic to
aquatic microorganisms, with EC₅₀ (3-4 days) values of
1.2-7.8 mg/litre, and 7-day NOEC values of 0.3-34 mg/litre.
Formulations of glyphosate are slightly to highly toxic to aquatic
microorganisms with 3-day EC₅₀ values of 1.0 to > 55 mg product
per litre. Cyanophyta (blue-green algae) are more sensitive to
Roundup than true algae. Physiological processes that are affected
include the greening process, respiration, photosynthesis, and the
synthesis of aromatic amino acids.

Soil bacteria in culture have shown effects of glyphosate on
nitrogen fixation, denitrification and nitrification. However, field
studies after application of formulations have not shown significant
effects. Closely related species of bacteria have been shown capable
of degrading glyphosate.

Mycelial growth of ectomycorrhizal fungi in pure cultures is
inhibited at concentrations of > 29 µg Roundup/litre. Sensitive
genera are Cenococcum, Hebeloma and Laccaria.

Glyphosate is slightly toxic to aquatic macrophytes with a
14-day NOEC value of 9 mg/litre, when dissolved in water. Roundup is
also slightly toxic with 14-day NOEC values of 2.4-56 mg
Roundup/litre, when dissolved in water. No data on acute toxicity
are available. Phytotoxicity is much higher when sprayed deposits
are not washed off.

Technical grade glyphosate is slightly to very slightly toxic
to aquatic invertebrates with 2- to 4-day LC₅₀ or EC₅₀ values of
> 55 mg/litre, and a 21-day NOEC value of 100 mg/litre.
Formulations of glyphosate are moderately to very slightly toxic to
aquatic invertebrates with 2-day EC₅₀ values of 5.3-5600 mg
product/litre and 21-day MATC values of 1.4-4.9 mg product per
litre. The higher toxicity of Roundup is mainly due to the presence
of surfactants.

Technical grade glyphosate is moderately to very slightly toxic
to fish, with 4-day LC₅₀ values of 10 to > 1000 mg/litre, a
21-day NOEC value of 52 mg/litre, and an MATC value of >
26 mg/litre. Formulations of glyphosate are also moderately to very
slightly toxic to fish with 4-day LC₅₀ values of 2.4 to > 1000 mg
product per litre, and 21-day NOEC values of 0.8-2.4 mg
product/litre. The most sensitive species is the carp, when exposed
to the formulation Sting. No treatment-related effects of Roundup on
fish have been found under field conditions, with the exception of
stress immediately after application of a recommended rate and
avoidance of concentrations of > 40 mg Roundup/litre.

Nodulation of sub-clover inoculated with Rhizobium is inhibited
in a dose-related way in soil-free systems with nutrient solutions
at concentrations of > 2 mg a.i./litre. Seed germination of
various forest species is not affected by glyphosate at the
recommended application rates. The root length of red pine seedlings
is decreased under laboratory conditions in a dose-related way at
application rates of > 0.54 kg a.i./ha. This decrease was not
confirmed in a comparable field experiment.

Technical grade glyphosate and Roundup are slightly toxic to
bees when applied either orally or topically. The 2-day LD₅₀
values are > 100 µg (a.i. or product) per bee. The oral 2-day
LD₅₀ of Sting to bees is > 100 µg/bee. Roundup and Roundup D-pak
are slightly toxic to earthworms with 14-day NOEC values of 500 and
158 mg product per kg dry weight, respectively. No adverse effects
of Roundup were found on the fecundity and fertility of green
lacewings, and there were no effects of Sting on the food uptake and
mortality of the beetle Poecilus.

Technical grade glyphosate is slightly toxic to birds, with an
LD₅₀ of >3851 mg/kg body weight, an 8-day LC₅₀ of >4640 mg/kg
feed, and 112- to 119-day NOEC values of > 1000 mg/kg feed.
Roundup and an unknown formulation are also slightly toxic to birds,
with an LD₅₀ of > 2686 mg product/kg body weight and an 8-day
LC₅₀ of > 5620 mg product/kg feed. Generally no treatment-related
effects of technical grade glyphosate or Roundup on mammals are
found under laboratory conditions, except at very high application
rates. Treatment-related effects on birds and mammals under field
conditions appear to be primarily due to habitat changes after
treatment with Roundup.

2. IDENTITY, PHYSICAL AND CHEMICAL PROPERTIES, AND ANALYTICAL
METHODS

2.1 Identity

Glyphosate is the primary name of a weak organic acid that
consists of a glycine moiety and a phosphonomethyl moiety. The
chemical name is N-(phosphonomethyl)glycine according to IUPAC
nomenclature. The CAS name is glycine, N-(phosphonomethyl)-, and
its CAS registry number is 1071-83-6. The empirical formula is
C₃H₈NO₅P, and the structural formula is as follows:

The relative molecular mass of glyphosate is 169.07. Technical
grade glyphosate has a purity of > 80%, but the purity generally
exceeds 90%. Glyphosate usually is formulated as a salt of the
deprotonated acid of glyphosate and a cation, e.g., isopropylamine.
The CAS registry number of the salt of glyphosate and
isopropyl-amine is 38641-94-0.

Surfactants and inerts may be added to formulations of
glyphosate. The type of surfactant and its concentration may differ
per formulation. A common surfactant in the major formulation
Roundup is polyoxyethylene amine. Other known surfactants are ortho
X-77 (Mitchell et al., 1987), LI-700, R-11 and Widespread (Monsanto,
1990a). Other additives in formulations may be sulfuric and
phosphoric acids.

2.2 Physical and chemical properties

The physical and chemical properties of glyphosate are
tabulated in Table 1. Glyphosate is an amphoteric compound of which
the ionic species and their pKa values are presented in Fig. 1. Due
to its high polarity glyphosate is practically insoluble in, for
instance, ethanol, acetone and benzene.

Table 1. Physical and chemical properties of glyphosate^a

Remarks

Physical state crystalline powder

Colour white

Odour none

Melting point^b 184.5 °C decomposition at 187 °C

Boiling point n.a.

Specific gravity (density)^c 1.704 20 °C

Vapour pressured < 1 x 10-5 Pa 25 °C

Solubility in water^b,e 10 100 mg/litre 20 °C

Henry's law constant < 7 x 10-11

Octanol-water partition
coefficient (log K_ow)^d -2.8

Surface tension^d 0.072 N/m 0.5% (w/v) at approx. 25 °C

pKa values^d,f < 2, 2.6, 5.6, 10.6 Sprankle et al. (1975)

Molar absorptivity^c 0.086 litre/mol per cm at 295 nm

Flammability^d not flammable

Explosiveness^d not explosive

pH^d 2.5 1% solution

^a data provided by Monsanto Ltd
^b purity 96%
^c purity 100%
^d purity not reported
^e pure glyphosate had been reported to have a water solubility of
11 600 mg/litre at 25 °C
^f free acid
n.a. = not applicable

2.3 Formulations

Glyphosate can be applied in various formulations. A synopsis
of these formulations, their concentrations of active ingredient,
and the countries in which the use is permitted is presented in
Table 2. This synopsis is not complete. Formulations may contain
specific surfactants. The major formulation of glyphosate is Roundup
containing 480 g/litre of the isopropylamine salt, which is
equivalent to 360 g/litre of the free acid. Some other Roundup
formulations that are characterized by other a.i. concentrations or
other surfactants have been developed for specific applications.
Other formulations that have been developed for special equipment
are Roundup Ultrabax for CDA equipment, Glyphosate Nomix for Nomix
equipment, and EZ-JECT for tree injections. In Canada, Roundup was
re-labelled as Vision in 1987 for use in forestry.

Table 2. Composition of various commercial formulations with glyphosate^a

Name Synonyms Concentration Country
a.i. (%)

Roundup Spasor, 48.0 (w/v); Most countries

Sting Vision, 41.0 (w/w)^b

Swing, 21.7 (w/w) Belgium, Cameroon, France,
Arcade, Holland, Kenya, Malawi,
Tomahawk Portugal, South Africa,
United Kingdom

Armada Frontier 16.6 (w/w) Belgium, Cameroon, Ivory
Coast, Gabon, Greece, Zaire
Dardo Ricochet^g, 12.2 (w/w) Cameroon, Egypt, France,
Rival, Greece, Israel, Italy,
Ultrasonic Portugal, Spain, United
Kingdom

Squadron 20.2 (w/w) Argentina, Australia, Columbia
Stirrup Nomix, Expedite 18.3 (w/w) France, United Kingdom
Wallop 20.8 (w/w) Malaysia
Deploy Dry^c 94.0 (w/w) USA
Quotamaker^d 75.0 (w/w) USA
Landmaster II^e 13.3 (w/w) USA
Landmaster BW,
Campaign^f 12.9 (w/w) USA
Roundup D-Pak 62.0 (w/w) USA
Rodeo 53.8 (w/w) USA
Ranger 28.6 (w/w) USA
Roundup Lawn and
Garden Conc. 18.0 (w/w) USA
Roundup-Ready-
To-Use 0.96 (w/w) USA
Fusta 22.5 (w/w) Spain

Table 2 (continued)
^a all formulations produced by Monsanto Ltd; data provided by Monsanto Ltd
^b based on the isopropylamine salt; equivalent to 36.0% (w/v) and 30.5% (w/w)
of the free acid
^c dry formulation of the monoammonium salt
^d dry formulation of the sodium sesqui salt
^e also contains 11.1% 2,4-D (isopropylamine salt)
^f also contains 20.6% 2,4-D (isopropylamine salt)
^g also contains simazine

Formulations may contain other active ingredients, e.g.,
simazine in Ricochet, 2,4-D in Landmaster, and MCPA in Fusta.

2.4 Conversion factors

1 ppm = 6.91 mg/m³ at 25 °C and 101.3 kPa

1 mg/m³ = 0.145 ppm

2.5 Analytical methods

2.5.1 Sample handling and preparation

The first preparative step before detection and measurement of
glyphosate is generally extraction. As both glyphosate and its main
metabolite aminomethylphosphonic acid (AMPA, see Fig. 2) show high
polarity, and are therefore highly water soluble, they are difficult
to extract with organic solvents. However, various methods have been
developed. Some recently developed extraction methods for different
media are summarized in Table 3.

Table 3. Sampling, preparation, and analysis of glyphosate

Medium Sampling Preparations Derivatization Analytical Limit of Recovery Reference
volume or reagent method determination^a
weight

Air n.r. collected onto an trifluoroacetic GC-MS and approx. 0.3 94% Jauhiainen
absorption liquid; anhydride and GC-EC µg/m³ et al. (1991)
evaporation to trifluoroethanol
dryness

Cyano-bacteria 100 ml dry, resuspend in PITC HPLC with a n.r.^b 78% Powell et al.
methanol/sodium - radically (1990)
acetate/ column compressed
triethylamine

Plants 5 g extraction with trifluoroacetic GC-NPD 0.03 mg/kg 72-92% Konar & Roy,
water/chloroform; anhydride and (1990)
preconcentration trifluoroethanol
and clean-up on
cation-exchange
resin

Plants 25-50 g extraction with TLC with 0.01 mg/kg n.r. Bunyathyan &
water/chloroform; ninhydrin Gevorgyan
preconcentration detection (1984)
and clean-up on
anion-exchange
and cation-exchange
resin; evaporation
to dryness

Table 3. cont'd (2)

Medium Sampling Preparations Derivatization Analytical Limit of Recovery Reference
volume or reagent method determination^a
weight

Water 250 ml extraction with o-phthalaldehyde LC 3.2 µg/litre 89% Wigfield &
dichloromethane; Lanouette
adsorption on (1990)
anion-exchange
resin

Water 25 ml extraction with FMOCCl HPLC and TLC 0.02 µg/litre 80% Gauch et al.
dichloromethane/ (1989)
2-propanol;
acidification
with H₂SO₄;
evaporation
to dryness

Water 1-1.5 litre no extraction; TLC with - 0.05 mg/litre n.r. Bunyathyan &
preconcentration ninhydrin Gevorgyan
and clean-up detectionc (1984)
with anion-exchange
and cation-exchange
resin

Soil 5 g extraction with trifluoroacetic GC-NPD 0.05 mg/kg 75% Roy & Konar
deionized water/ anhydride/ (1989)
H₃PO₄; addition trifluoroethanol
of Darco
charcoal

Table 3. cont'd (3)

Medium Sampling Preparations Derivatization Analytical Limit of Recovery Reference
volume or reagent method determination^a
weight

Soil 2 g (sandy extraction with FMOCCl HPLC 1 mg/kg 80-119% Miles & Moye,
soil); 25 g KH₂PO₄ (sandy soil), (1988b)
(clayish KOH (clayish soil);
soil) no clean-up

Soil, 5 g (soil); extraction with NH₄OH; ninhydrin LC 0.05-0.1 73-79% Thompson
sediment, 20 g (sed); adsorption on mg/kg (soil); (soil) et al. (1989)
foliage 5 g (fol) anion-exchange resin; 0.1 mg/kg 65-84%
further clean-up with (sed); 0.3 (sed)
Dowex cation-exchange mg/kg (fol)^d 81-84%
resin (fol)

Urine and 5-6 g extraction with H₂O HPLC (ion n.r. 81-99% Monsanto
faeces of (only faeces); protein pair, strong (1988a)
the rat precipitation and anion and
lyophilization cation-exchange),
(only urine); LSC, ¹H NMR, ³¹P
clean-up with C₁₈ NMR, GC/MS
column

Urine n.r. adsorption on trifluoroacetic GC-MS and 0.1 mg/litre n.r. Jauhiainen
(human anion-exchange resin anhydride/ GC-EC et al. (1991)
male) (SAX); elution of the trifluoroethanol
resin with HCl;
evaporation to
dryness

Table 3. cont'd (4)

Medium Sampling Preparations Derivatization Analytical Limit of Recovery Reference
volume or reagent method determination^a
weight

Serum 0.5 ml extraction with p-toluene HPLC with UV n.r.^e n.r Tomita et
(human) trichloroacetic sulfonyl chloride detection al. (1991)
acid; adsorption
on anion-exchange
resin; elution
with HCl; evaporation
to dryness

^a In no study with a non-liquid medium was it reported whether the limit of determination was based on dry or fresh weight,
except in the study of Thompson et al. (1989).
^b The order of magnitude was reported to be picomol.
^c The use of TLC with ninhydrin, copper nitrate and rhodamine B detection is reported for glyphosate in distilled water in
Ragab (1978).
^d The limits of determination in soil, sediment, and foliage are expressed per kg dry weight.
^e Only the limit of detection was reported: 0.3 mg/litre (approximately 75% recovery).

PITC = phenylisothiocyanate; FMOCCl = 9-fluorenyl-methyl chloroformate; GC =gas chromatography;
(HP)LC = (high-performance) liquid chromatography; TLC = thin layer chromatography; MS = mass spectroscopy;
EC = electron capture detector; NPD = nitrogen-phosphorus detector; n.r. = not reported;
LSC = liquid scintillation counting; NMR = nuclear magnetic resonance; sed = sediment; fol = foliage

The second preparative step is the clean-up, which may include
extraction, preconcentration by evaporation, ion-exchange
chromatography or gel chromatography. Clean-up procedures may
involve different combinations of chromatographic techniques. In a
validation study in which plant tissues and water were analysed, a
Chelex column was combined with anion-exchange clean-up (Cowell
et al., 1986). No chromatography was included in the clean-up
procedures for analysing glyphosate and AMPA in natural waters
(Miles et al., 1986). In this procedure samples were successively
filtrated, supplied with phosphate buffer, concentrated by
evaporation, and filtrated, prior to derivatization.
Samples with urine and faeces of the rat were subjected to clean-up
with a C₁₈ column (Monsanto, 1988a). Prior to this extraction,
proteins were precipitated and the samples were lyophilized; samples
of faeces were, however, only extracted with water.

The third preparative step is derivatization. Derivatization
with a fluorogenic reagent is common. Burns (1983), however,
developed a preparation technique without derivatization.
Derivatization prior to detection and measurement with HPLC can be
pre-column (Miles et al., 1986; Lundgren, 1986; Miles & Moye,
1988a) or post-column (Moye et al., 1983; Tuinstra & Kienhuis,
1987). 9-Fluorenylmethyl chloroformate, phenylisothiocyanate and
1-fluoro-2,4-dinitrobenzene may be used as pre-column reagents,
whereas ortho-phthalaldehyde-mercaptoethanol and ninhydrin may be
used as post-column fluorogenic reagents. With post-column
techniques, derivatives can be formed on-line, but it requires more
equipment and experience. On the other hand, pre-column techniques
are often more rapid and require less equipment and experience. In
general the facilities required for derivatization with fluorogenic
substances are very specific, and therefore not available in many
laboratories (Konar & Roy, 1990). These authors proposed
derivatization with a mixture of trifluoroacetic anhydride and
trifluoroethanol prior to analysis with gas chromatography as a
simpler, less laborious and more economical method. This proposal
referred to the determination of glyphosate and AMPA in plant
tissues. This and other recently developed techniques of clean-up
and derivatization are summarized in Table 2. These techniques are
intended to simplify and improve preparative techniques, which in
general used to be complex and costly (Marcotte et al., 1977;
Guinivan et al., 1982; Roseboom & Berkhoff, 1982; Moye et al.,
1983; Moye & Deyrup, 1984; Deyrup et al., 1985; Miles et al.,
1986; Lundgren, 1986; Miles & Moye, 1988b).

Sample preparation and derivatization, as developed by Powell
et al. (1990) for cyanobacteria without deproteinization (see
Table 3), should also be usable for plant and animal tissue. In this
case, a simple maceration step prior to ethanol extraction should be
included. Bunyathyan & Gevorgyan (1984) developed preparative
techniques for different media prior to analysis with TLC. Only

their procedures for plants and water are summarized in Table 3. The
preparative technique for soil samples was comparable with that of
Thompson et al. (1989), although samples of 25-50 g were required.
Bunyathyan & Gevorgyan (1984) also developed a method for preparing
20-litre air samples prior to TLC. They extracted the residues
collected on a filter with water before clean-up on a
cation-exchange resin.

2.5.2 Analytical methods

Various analytical methods for the determination of glyphosate
have been described, including thin-layer chromatography (Young
et al., 1977; Ragab, 1978; Bunyathyan & Gevorgyan, 1984),
colorimetry (Glass, 1981), differential pulse polarography (Friestad
& Bronstad, 1985), gas chromatography (Guinivan et al., 1982; Moye
& Deyrup, 1984; Deyrup et al., 1985), high-performance liquid
chromatography (Miles & Moye, 1988a; Powell et al., 1990), and
³¹P NMR (Dickson et al., 1988). Some of these techniques, their
analytical recoveries and limits of determination are listed in
Table 3. The corresponding determination limits for AMPA, i.e.
analysed with the same techniques, are listed in Table 4. Recoveries
in the different media appear to be higher for glyphosate than for
AMPA. This is probably due to optimization of the systems for
glyphosate, as was done by Thompson et al.(1989).

Table 4. Limits of determination of AMPA

Medium Limit of Recovery Reference
determination

Plants 0.01 mg/kg 61-73% Konar & Roy (1990)

Water 1.2 µg/litre 86% Wigfield & Lanouette (1990)

Soil 0.01 mg/kg 66% Roy & Konar (1989)

Soil 0.03-0.05 mg/kg 58-68% Thompson et al. (1989)

Sediment 0.03 mg/kg 54-67% Thompson et al. (1989)

Foliage 0.008 mg/kg 55-70% Thompson et al. (1989)

Urine (human) 0.05 mg/litre n.r. Jauhiainen et al. (1991)

Serum (human) n.r.^a n.r. Tomita et al. (1991)

^a n.r. = not reported; only the limit of detection was reported: 0.2 mg/litre
(approximately 88% recovery)

TLC techniques are generally based on silica gel or cellulose
plates; cellulose plates give a better separation (Dubelman, 1988).
Ninhydrin and phosphate sensitive reagents may be used for
detection, although interference from co-extractives may occur.
According to Dubelman (1988), fluorogenic reagents may be preferable
in case of interference.

Fluorogenic derivatives can be determined in HPLC analysis with
fluorescence detectors (Wigfield & Lanouette, 1990) and also with a
spectrophotometer (Powell et al., 1990). In a GC analysis a
nitrogen-phosphorus, electron capture or a flame photometric
detector can be used.

3. SOURCES OF HUMAN AND ENVIRONMENTAL EXPOSURE

3.1 Anthropogenic sources

3.1.1 Production levels and processes

No data on the world production of glyphosate and its
formulations are available. In addition, no data on losses to the
environment during normal production and formulation or accidental
losses have been reported.

The first phase of the production of glyphosate consists of
refluxing a mixture of glycine (50 parts), chloromethylphosphonic
acid (92 parts), an aqueous solution with 50% sodium hydroxide (150
parts), and water (100 parts) in a suitable reaction vessel. Another
50 parts of an aqueous solution with 50% sodium hydroxide are added
to maintain the pH between 10 and 12, whereafter the reaction
mixture is refluxed for another 20 h. The mixture is then cooled to
room temperature and filtered. After adding 160 parts of
concentrated hydrochloric acid, this mixture is again filtered.
Glyphosate will slowly precipitate in the filtrate (IRPTC, 1991).

3.1.2 Uses

Glyphosate is a post-emergent, systemic and non-selective
herbicide intended for use against deep-rooted perennial species,
and also biennial and annual broad-leaved, grass and sedge species
(WSSA, 1983; Monsanto, personal communication to the IPCS, 1991).
Glyphosate is used in both agriculture and forestry. Fields of
agricultural use include grassland renovation, horticulture,
fructiculture, arable cultivation, and rice cultivation. Use in
forestry includes the killing of fast growing competitors in conifer
plantations or conservation areas, and the treatment of tree stumps.
Glyphosate may also be used for weed killing in non-agricultural
areas such as water systems, including irrigation and temporarily
drained waters, parks, road verges and gardens.

The uses of glyphosate indicate that it can be applied in
various crops for specific purposes. The major formulation Roundup
may, for instance, be used in pre-plant treatments for seed bed
preparations, and also against bracken infestations in forestry,
against couchgrass (Elytrigia repens) infestations on pastures, in
direct treatments between rows of crops, or by direct wiping of the
leaves of the weed, assuming the weeds are taller than the existing
crop.

Glyphosate is used worldwide. In 1987, 35 160 ha of the area in
British Columbia where vegetation management activities took place
had been treated with Roundup. This was 94% of the total area where
there were such activities (Ackurst, 1989).

The application rates of glyphosate are dependent on the
formulation and type of use. In the Netherlands, recommended rates
for the application of Roundup are 0.3-2.9 kg a.i./ha. In Canada the
recommended application rates of Roundup are 1.1-1.7 kg a.i./ha for
annual weeds and 1.2-5.8 kg a.i./ha for perennial weeds. The
recommended application rates for Vision in Canadian forestry are
1.1-2.1 kg a.i./ha (Task Force on Water Quality Guidelines, 1991).
Glyphosate is generally applied as a 0.5-5% solution in water by
spraying, and as a 10-50% solution in water by wiping with, for
instance, a rope-wick (Monsanto, personal communication to the IPCS,
1991).

The timing of application is dependent on the use. Application
in late summer or autumn is recommended for use in forestry in
Canada (Hildebrand et al., 1982). Application in agriculture may
be pre- or post-harvest. In the Netherlands, for instance,
glyphosate may be applied to cereals, potatoes and asparagus
immediately (up to 7 days) before harvest, but only when the
ripening is complete. Treatment of immature crops would result in
higher residue levels, early crop desiccation and reduced yields.

Glyphosate may be applied in different ways. For large-scale
treatments aerial application can be appropriate, small-scale
treatments can be done with spraying equipment on the back or behind
vehicles, or by wiping equipment.

Aerial applications will lead to losses due to wind-drift.
Exposure of flora and fauna due to off-target deposits may take
place. These downwind deposits depend on the meteorological
conditions, the plant canopy structure and the application method,
including the release height (Payne et al., 1989; Feng et al.,
1990; Payne, 1992; Payne & Thompson, 1992). The non-volatile
tank-mix fraction and the speed of the aircraft may influence the
drop-size spectrum, and it can be expected that dispersal systems
causing relatively small droplets and having a relatively low
non-volatile fraction will cause the highest off-target deposits.
Payne (1992) assumed that the large differences in deposits in two
comparable experiments were due more to different aircraft airspeeds
than to different wind speeds. In these experiments the maximum
deposits at a downwind distance of 50 m were 19 and 3 mg a.i./m²
at aircraft airspeeds of 45 and 11-20 m/second, respectively. The
application rate in both experiments was 2.1 kg a.i./ha. In other
experiments with the same application rate, Payne & Thompson (1992)
found that the meteorological conditions had a considerable impact
on the off-target deposition up to 400 m downwind when spraying at
different wind speeds (2.2-5.7 m/second) and turbulences. The
deposits at a downwind distance of 400 m varied between 0.001 and
0.06 mg a.i./m², whereas they varied between 0.6 and 4 mg
a.i./m² at a downwind distance of 50 m. Remarkably, the deposition
was highest with an intermediate wind speed and intensity of

turbulence. Payne et al.(1989) investigated the deposits for
aerial applications of Roundup with different dispersal systems.
When 2.1 kg a.i./ha was applied with a helicopter in a single
crosswind swath over 100 ha, up to 13.4 mg a.i./m² was deposited
on a downwind distance of 50 m. This maximum deposition was caused
by a D8-46 hydraulic nozzle, whereas the highest depositions with a
Thru Valve Boom and a Microfoil Boom were 2 and 0.4 mg a.i./m²,
respectively. These depositions were also found at a downwind
distance of 50 m. At the time of application the windspeed 13 m
above ground level was 0.4-0.5 m/second. Riley et al.(1991)
modelled spray deposition of glyphosate using results from
helicopter applications under semi-operational conditions. The study
was designed to test the appropriateness of a New Brunswick "buffer
zone" of 65 m to minimize the effects of spray drift. At a distance
of 65 m, it was estimated that between 3.7% and 5.6% of the nominal
spray rate was deposited.

3.1.3 Drinking-water

Appraisal

The low mobility of glyphosate in soil would indicate a
minimal potential for the contamination of drinking-water from
groundwater aquifers. The only possible source of drinking-water
contamination is, therefore, surface waters. There have been no
reported incidences of drinking-water contamination with
glyphosate.

Conventional plants for processing of drinking-water would not
remove glyphosate, but this could be achieved by coprecipitation
after adding iron salts (AMA van der Linden, personal communication
to the IPCS, 1991). Ozone, increasingly used as an alternative to
chlorine in drinking-water treatment, does effectively remove
glyphosate through the hydroxyl radical (HO^. ) chain processes that
occur in most ozonated waters (Yao & Haag, 1991; Haag & Yao, 1992).

4. ENVIRONMENTAL TRANSPORT, DISTRIBUTION AND TRANSFORMATION

4.1 Transport and distribution between media

Appraisal

Following application, glyphosate selectively partitions to
particulate matter suspended in surface water or to the soil
substrate. This partitioning is usually rapid and occurred within 14
days in reported studies. The mechanism of sorption to soil is only
partially understood. Glyphosate can adsorb to soils through
phosphate binding sites. Competition with inorganic phosphate has
been demonstrated in the laboratory but not measured in the field.
Specific ions (Fe²⁺, Fe³⁺ and Al³⁺ ) complex glyphosate; metal
complexes with humic acids in soil may be a main binding mechanism
for glyphosate in soil. There is little reported information on
desorption from soil; the data available suggest "strong" binding.
This is supported by mobility studies which show little leaching of
glyphosate below the upper few centimetres of the soil profile. The
major metabolite, AMPA, is also retained in the upper soil layers.

There is very little information on the bioavailability of
sediment-bound glyphosate to either aquatic or terrestrial
organisms. Bioaccumulation and ecotoxicity studies have not,
generally, been performed with added sediment.

Applied glyphosate can be translocated in plants. Glyphosate
in plant foliage or leaf litter does not seem to represent a source
of contamination of aquatic systems. Animals can ingest the
herbicide residues in or on plants.

Dissipation of glyphosate from soil has been widely studied
with very variable results (DT₅₀ between 3 and 174 days).
Biodegradation appears to be the major source of dissipation.

Run-off was minimal in experimental studies, but field results
suggest that aquatic systems may be receiving glyphosate bound to
soil particles following rainfall.

In this chapter the terms biodegradation and dissipation are
used to distinguish between the decrease of the concentration in,
for instance, the soil that is due to microbes transforming the
molecule to a smaller size (biodegradation) and the decrease of the
concentration that might be due to microbial activity but also to
other processes, e.g., sorption, leaching and run-off (dissipation).

4.1.1 Water

Glyphosate dissipates from the water with 50% dissipation times
ranging from a few days to 2 weeks (Newton et al., 1984; Monsanto
1990a; see also Table 5). These DT₅₀ values were deduced from both
laboratory and field experiments in which sediment or suspended
particles were shown to be the major sink.

In water with a near-neutral pH, the formation of an insoluble
complex of Ca²⁺ with glyphosate was demonstrated in a laboratory
experiment (Subramaniam & Hoggard, 1988). It was confirmed with
X-ray powder diffraction and infrared spectra that this complex was
not an ionic salt. At a near-neutral pH, the dianionic species of
glyphosate is dominant. Insoluble complexes have also been found
with Mg²⁺, Fe³⁺ and Cu²⁺.

In a field experiment in a temperate coastal rainforest in
British Columbia, Canada, the highest concentration of glyphosate in
water was 162 µg/litre (Feng et al., 1990). This maximum was found
in a directly sprayed tributary 2 h after an aerial application of
Roundup at a rate of 2 kg a.i./ha. Concentrations in oversprayed
tributaries without a high cover of overhanging riparian vegetation
increased after the first rainfall. In oversprayed tributaries with
a high cover of riparian vegetation almost no residues were found.
Within 96 h after application the residues in all waters had
declined below detection limits, indicating rapid dissipation. After
rainstorms, peak concentrations of glyphosate were found in the
sediments and on suspended particles of the oversprayed tributaries,
with maximum concentrations of 7 mg a.i./kg dry weight and 0.06 µg
a.i./litre unfiltered water, respectively. The amounts in the
sediments of these waters were variable but declined over time. As
0.1-2 mg residue/kg dry weight sediment was found between 196 and
364 days after application, the residues appear to be persistent in
sediments of oversprayed waters. Feng et al.(1990) concluded,
therefore, that after rainstorms sediments appear to be the major
sink.

In another field experiment in the same forest ecosystem,
glyphosate dissipated rapidly from a small perennial, very slow
flowing stream, in a site of 8 ha aerially sprayed with Roundup at a
rate of 3.3 kg a.i./ha (Newton et al., 1984). In water, 50% of the
initial concentration had dissipated in 2 days. In sediment, maximum
concentrations of approximately 0.6 mg a.i./kg were found 14 days
after application. These were reduced to approximately 0.3 mg
a.i./kg in 28 days, and to < 0.2 mg a.i./kg in 55 days after
application. A comparable rapid dissipation from the water column
was found for small forest ponds in a boreal forest in Manitoba,
Canada, after applying Roundup at a rate of 0.9 kg a.i./ha
(Goldsborough & Beck, 1989). The highest concentration in filtered
water was 141 µg a.i./litre, within 6 h after application. The main

mechanism of dissipation was probably sorption to the sediment. This
was confirmed by additional experiments with polyethylene basins
filled with unfiltered water and sediment that were placed in the
spray zone. Without sediment, only a very small amount of the dose
actually applied had dissipated after 30 days, whereas with sediment
the initial concentrations in the water had decreased by 50%,
approximately 6 days after application.

Comparable dissipation patterns were found in a field
experiment (Monsanto, 1990a) in which Accord (30.5% a.i. w/w) was
applied at a rate of 4.2 kg Accord/ha on three forestry sites with
non-flowing pond water and flowing water. Concentrations of up to
1700 µg/litre filtered water were found in the pond water
immediately after spraying. The initial concentrations in both pond
and flowing water were reduced by 50% within 7 days. Concomitantly
initial AMPA concentrations (maximally 35 µg/litre) were reduced by
50% within the same period. In flowing water the dissipation of both
glyphosate and AMPA was even more rapid. Concentrations of
glyphosate increased up to 19 mg/kg dry weight in the sediment of
one pond 28 days after application. Concentrations of up to 1 mg/kg
of both compounds were measured in the sediments of non-flowing
ponds up to 400 days after application.

In field experiments in turbid Australian irrigation water,
glyphosate adsorbed to suspended particles at different rates,
apparently mainly depending on the application rate (Bowmer et al.,
1986). At an initial concentration of 5 mg a.i./litre, 10-16% of the
load adsorbed to suspended matter, whereas at an initial
concentration of 0.05 mg a.i./litre, 53-71% adsorbed. In more saline
water the degree of sorption was less, probably due to rapid
flocculation. Maximum adsorbed amounts were approximately 7000 mg
a.i./kg in less saline supply water, and approximately 2500 mg
a.i./kg in more saline drainage water. When a supply channel was
emptied before spraying with 3.6 kg a.i./ha for control of aquatic
weeds, and filled again with water 4 days after the treatment, the
amount in the unfiltered water used for irrigation was 7% of the
applied dose.

4.1.2 Soil sorption

Glyphosate is readily bound to many soils and clay minerals
(Sprankle et al., 1975; Hance, 1976; Glass, 1987; Miles & Moye,
1988b). In laboratory experiments in which glyphosate was added to
aqueous soil suspensions, the adsorption coefficient K_s/l was
18-377 dm³/kg in nine soils ranging from sandy loam to peat
(Hance, 1976), and 33-76 dm³/kg in three soils ranging from sandy
loam to clay loam (Glass, 1987). These K_s/l values indicate strong
sorption. In both experiments the sorption could be described by the
Freundlich equation. Glass (1987) found K_s/l values for the clay
minerals montmorillonite, illite and kaolinite of 138, 115 and 8
dm³/kg, respectively.

Table 5. Biodegradationa of technical grade glyphosate in water and sediments in the laboratory

Water type Sediment Test Sediment Organic Temperature pH of Experimental Parameter Time Reference
type type (%) matter in (°C) water time (days)
sediment (days)
(%)

Pond water silty A 17 0.9 23-25 5.9-7.0 30 DT₅₀ 14^b PTRL East
clay Inc. (1990a)
loam

Pond water silty An 16 0.9 20-27 5.7-6.5 365 DT₅₀ 14^c,e PTRL East
clay Inc. (1990b)
loam

Surface waterd n.r. A 9 n.r. 30 8.2-8.6 14 DT50 < 14 Monsanto
(1972a)

Lake water sandy An 33 1.4 30 6.6 42 DT50 22^e Monsanto
clay (1978a)
loam

^a Biodegradation in the whole system
^b The biodegradation stopped after approximately 15 days
^c The biodegradation stopped after approximately 150 days
^d Three rivers and one lake in the USA
^e Approximate value derived from data of the author(s)
A = aerobic; An = anaerobic; n.r. = not reported.

The mechanism of sorption of glyphosate to soil is only
partially understood. Several factors may be involved. The
phosphonic moiety adsorbs weakly to unoccupied phosphate binding
sites and can be displaced by phosphate (Hance, 1976). In laboratory
experiments with nine soils the author showed that sorption was
positively correlated with the unoccupied phosphate sorption
capacity, and not correlated with the total phosphate sorption
capacity, organic matter, clay, iron or aluminium content. No data
are available that confirm competition of glyphosate and phosphate
under field conditions, e.g., after application of artificial
fertiliser. Miles & Moye (1988b) suggested that the main mechanism
was probably by H-bonding and ion-exchange, as the degree of
sorption in their experiments was not correlated with cation
exchange capacity (CEC) values or surface areas. Contrary to the
results of Miles & Moye (1988b) and of Hance (1976), sorption
appeared to be correlated with CEC values and clay content in a
sorption study with clay loam, silt loam and sandy loam (Glass,
1987).

The binding is also influenced by the presence of specific
cations. Hensley et al.(1978) demonstrated that Fe²⁺, Fe³⁺
and Al³⁺ inactivated glyphosate much more than Ca²⁺, K⁺ and
Na⁺. This was confirmed by Glass (1987) and Sprankle et al.
(1975). Glass (1987) suggested that glyphosate was complexed by
cations, released from cation-saturated clays via a cation-exchange
with solution protons.

According to Heinonen-Tanski (1989), most of the soil-bound
residues of glyphosate were recovered in the fulvic acid fraction
(21-33%). Sorption of glyphosate to fulvic acids was also reported
by Madhun et al.(1986), who added ¹⁴C-glyphosate to an aqueous
soil extract (ASE) of peat. In this study sorption was mainly on ASE
fractions with a relative molecular mass ¾ 1000. Piccolo et al.
(1992) studied the interaction of glyphosate with a pure iron-humic
acid complex. Maximum adsorption values indicated that adsorption to
the complex occurred to as great an extent as to whole soils. This
suggested that humic acid complexes with polyvalent cations might
represent a main binding substrate for glyphosate in soils. There
was no desorption of bound residues of glyphosate following shaking
with 0.01 mol CaCl₂/litre solution for 48 h, the maximum shaking
time for the adsorption studies.

Desorption of glyphosate with ionized water from
montmorillonite and illite needed three days before reaching an
equilibrium in a study of Miles & Moye (1988b).

It can be concluded that sorption of glyphosate can be expected
in the presence of available phosphate binding sites, the presence
of iron and aluminium (oxides or hydroxides), and appropriate
combinations of clay and organic matter.

4.1.3 Mobility in soils

In view of its K_s/l, glyphosate can be expected to be
immobile or slightly mobile in many soils. This was confirmed by
several experiments, both in the laboratory and in the field. In
thin-layer chromatography studies with sandy loam, clay loam and
sandy clay loam, the Rf values of ¹⁴C-glyphosate were 0.14-0.20
(Sprankle et al., 1975). In comparable studies with silt loam,
silty clay loam, and sandy loam Rf values were < 0.2 (Monsanto,
1972c). In a leaching study with columns of 30 cm and a high water
flux of 51 cm over less than 2 days, < 0.1-6.6% of the applied
activity was leached (Monsanto, 1978b). This experiment was
performed with eight soils, ranging from sandy loam (organic matter
content 0.7%) to volcanic ash (organic matter content 9.5%). More
than 90% of the applied activity was recovered in the upper 0-14 cm
layer.

Only one leaching study under laboratory conditions with
respect to the mobility of AMPA has been reported. In this
experiment with 30-day-old residues, < 0.1-1.6% of the applied
activity was leached over 45 days (Monsanto, 1978b). The columns
were 30 cm and the water flux over 45 days was low (17 cm). The
amount of AMPA that was recovered after 45 days in the upper 0-2 cm
layer was low (0.5-12% of the applied activity), due to a high rate
of mineralisation.

4.1.4 Dissipation from the soil in the field

Many field experiments on the dissipation of glyphosate from
the soil have been performed. Some relevant studies are summarized
in Table 6. They indicate DT₅₀ values based on dissipation that
range from 3 to 174 days depending on edaphic and climatic
conditions. In a forest brush ecosystem in Oregon, USA, the DT₅₀
value in loam was 29 days with and 40 days without litter (Newton
et al., 1984). In field experiments in Sweden, Roundup was sprayed
over reforestated sites (Torstensson et al., 1989). In the soils
of these sites the DT₅₀ values were < 50 days, apparently
depending on the soil respiration rate. The dissipation consisted of
a fast first, and a much slower second phase, especially in sites in
northern Sweden, which was possibly due to a longer frost period. In
these sites 1-2% of the actually applied dose was recovered 1080
days after application. A comparable dissipation pattern was found
in a field experiment on Finnish agricultural soils (Heinonen-Tanski
et al., 1985). In this experiment 25% of the concentration in a
sandy loam 2 days after the treatment was recovered one year after
application. The application rate was 1.4 kg a.i./ha.

A study in a temperate coastal rain forest in British Columbia,
Canada, showed that, 360 days after application, 6-18% of the
initial levels was recovered (Feng et al., 1990). In this
experiment Roundup was applied at a rate of 2 kg a.i./ha. The soils

were alluvial sandy loam or sandy clay loam with highly organic
surface horizons. Some of these soils were well drained, others were
seasonally flooded. At each sampling time more than 90% of the
recovered residues was in the upper 0-15 cm layer. Under all
conditions the amount of glyphosate declined over time, whereas
there was a transient increase of AMPA.

In other field experiments on boreal forest soils, comparable
dissipation patterns were found. Stark (1983) reported DT₉₀ values
of 30-720 days, and Roy et al.(1989b) found a DT₅₀ value of
approximately 20 days on a sandy soil planted with jackpines (Pinus
banksiana). In the field experiments of Roy et al.(1989b),
glyphosate was detectable up to 335 days after application; almost
all residues in the sandy soil were recovered in the organic top
layer. In field experiments of Monsanto (1990a) in three forest
locations in the USA, the concentration course of glyphosate
appeared to be rather irregular, especially during the first four
months. However, 50% of the initial concentrations in the soil had
mostly dissipated within 120 days. One clear exception was a site in
Corvallis in which glyphosate increased up to 0.15 mg/kg dry weight,
180 days after application. On the same site AMPA increased up to
0.32 mg/kg, 346 days after application. The application rate in
these experiments was 4.2 kg Accord/ha.

On a clay soil of a clear-cut boreal forest area, Roy et al.
(1989b) found no dissipation of glyphosate due to run-off on a 8°
slope. In a field experiment on agricultural soils without
conventional tillage, the dissipation of glyphosate due to run-off
on 6-16° slopes was < 1% of the applied dose when 1.1-3.4 kg
a.i./ha was applied (Edwards et al., 1980). However, when 9.0 kg
a.i./ha was applied, 1.8% of the applied dose dissipated due to
run-off, mainly because of a rainstorm shortly after application.

4.1.5 Uptake and dissipation from plants

Uptake of ¹⁴C-glyphosate by leaves of trembling aspen
seedlings (Populus tremuloides) was initially rapid, after which
it slowed down (Sundaram, 1990). The seedlings were exposed to
Roundup that was dripped with a micro-applicator on some central
leaves. The application rate was 0.35 kg a.i./ha leaf surface area.
Most activity was washable from the leaves (61-77%), and 22-28% was
recovered in the treated leaves within 48 h. As only 1-10% was
recovered in the other parts of the seedlings, this indicated a
rather low translocation after absorption. A rapid uptake of
¹⁴C-glyphosate within a few hours was indicated for sugar beets
(Beta vulgaris), when applied to a mature leaf (Gougler & Geiger,
1981). ¹⁴C-glyphosate probably entered the phloem in a
non-facilitated way. The subsequent transport through the phloem

Table 6. Biodegradation and dissipation of glyphosate in soils

Soil type Compound Test Moisture Temperature pH Organic Experimental DT₅₀ Reference
type content (°C) matter duration (days)
(%) (%) (days)

Biodegradation

Sandy loam Tgg L,A 14-16 25 7.3 2.8 360 2^b PTRL East Inc. (1991)
Silt loam Tgg L,A 12-14 25 7.5 1.0 360 2^b PTRL East Inc. (1991)

Dissipation

Sandy loam Tgg G 11 32 5.7 1.0 112 130^b Monsanto (1972b); Rueppel
et al. (1977)
Silt loam Tgg G 11 32 6.5 1.0 112 3^b Monsanto (1972b); Rueppel
et al. (1977)
Silty clay loam Tgg G 11 32 7.0 6.0 112 25-27^b Monsanto (1972b); Rueppel
et al. (1977)
Sand Ru F n.r. n.r. 3.5-3.7 40 762 approx Roy et al. (1989b)
(humoferric 20^a
podsol)
Sandy loam, Ru F n.r. n.r. 4.2-4.9 15-31 360 45-60^b Feng & Thompson (1990)
sandy clay loam
Loam Ru F n.r. n.r. 4.0-4.7 3.8-5.2 55 29-40^b Newton et al. (1984)
Loamy sand Ru F n.r. n.r. n.r. 0.8 370 3-4^b Monsanto (1983a)
Sandy clay loam Ru F n.r. n.r. n.r. 7.0 370 122-174^b Monsanto (1983a)

^a Based on data of the author(s) ^b Data reported by the author(s)
L = laboratory study; F = field study; G = greenhouse study; A = aerobic; An = anaerobic;
Tgg = technical grade glyphosate; Ru = Roundup; n.r. = not reported

appeared to be according to an "intermediate permeability
mechanism". When exposed for a longer time, plants may show
substantial translocation of absorbed ¹⁴C-glyphosate, as was shown
for potatoes (Solanum tuberosum) by Smid & Hiller (1981). In the
treated leaves of the potatoes 45% of the absorbed activity was
recovered, whereas the rest was mainly translocated to the apical
meristem and the roots. Up to 5% was recovered in the mother tuber.
The degree of translocation was age-dependent, as older plants
showed less translocation than younger plants.

The uptake of glyphosate by red raspberries (Rubus strigosus)
was 9% of the amount that was deposited on the leaves after spraying
Roundup at a rate of 2 kg a.i./ha (Roy et al., 1989a). In the same
field experiment the uptake was 14% by wild blueberries (Vaccinium
myrilloides). Most glyphosate was recovered in the washings, which
was also found under laboratory conditions. The initial absorbed
amounts were 0.92-2.0 mg a.i./kg dry weight. The absorbed and
washable amounts together were reduced by 50% within 13 days in the
raspberries and within 20 days in the blueberries. AMPA was
detectable up to 33 days after application. Metabolism occurred to
only a minor extent as AMPA concentrations were less than 1.5% of
the concurrent concentrations of glyphosate (similar results were
reported by FAO/WHO, 1986b). In a field experiment by Feng &
Thompson (1990) in a temperate coastal rainforest in British
Columbia, Canada, the main target species for treatment with Roundup
were red alder (Alnus rubra) and salmonberry (Rubus spectabilis).
Immediately after spraying, the concentrations in leaf tissue were
up to 448 mg a.i./kg dry weight. Glyphosate dissipated rapidly from
the leaf litter with a DT₅₀ value of 8-9 days. The leaf litter
included leaves directly exposed on the trees and existing leaf
litter from natural defoliation before treatment with Roundup. The
authors assumed that leaf litter of these major brush species is an
insignificant source of glyphosate input into streams or onto forest
floor, because of the fast dissipation. A rapid dissipation of
glyphosate from fresh foliage was also found in a field study
(Monsanto, 1990a) in which initial concentrations of up to 1300 mg
a.i./kg and 2.6 mg AMPA/kg decreased rapidly. A transient
accumulation of glyphosate and AMPA was found in the leaf litter on
some sites, but these amounts were reduced by approximately 90%
within 100 days.

Glyphosate dissipated completely from wild berries (Vaccinium
vitis-idaea, Vaccinium myrtilus) within one year in a field
experiment in Finland in which Roundup was applied at a rate of
0.25-2.2 kg a.i./ha with a knapsack sprayer (Siltanen et al.,
1981). Contrary to this dissipation pattern was that of glyphosate
in reindeer lichens (Cladonia rangiferina) that were sampled in
the same experimental plots. Around 270 days after application,
dose-related concentrations of glyphosate and AMPA were recovered in

lichens with maxima of 45 and 2.1 mg/kg for glyphosate and AMPA,
respectively. Approximately 390 days after application of 0.8 kg
a.i./ha, 6.4 and 0.3 mg/kg of glyphosate and AMPA were still
detectable.

4.1.6 Ingestion by animals

As the concentration in the foliage may increase up to high
amounts immediately after application, this implies the possibility
of entry into the food chain through ingestion by herbivorous or
omnivorous animals. This was confirmed by Sullivan & Sullivan (1979)
who investigated the effects of glyphosate on the feed preference
and daily chow consumption of black-tailed deer (Odocoileus
hemionus columbianus). These herbivores did not avoid eating
browse of alder (Alnus rubra) and alfalfa (Medicago sativa) that
was treated with glyphosate at a rate of 2.2 kg/ha. Sometimes the
treated alder browse was even preferred. Reindeer may be exposed to
glyphosate, since reindeer lichens, which are an important food
source, can take up a substantial amount of glyphosate (see above).

4.2 Abiotic degradation

Appraisal

Hydrolysis of glyphosate is very slow. Photodegradation in the
field may occur.

4.2.1 Hydrolytic cleavage

Hydrolysis of glyphosate in sterile buffers is very slow. After
32 days < 6.3% of the applied activity was recovered as AMPA,
after applying ¹⁴C-glyphosate at rates of 25 and 250 mg/litre to
aqueous buffer solutions of pH 3, 6 and 9 (Monsanto, 1978b). These
tests were performed at both 5 and 35 °C.

4.2.2 Photodegradation

Photochemical degradation in water may occur under both
laboratory and field conditions, mainly depending on the type of
light source. In sterile aqueous buffers of pH 5, 7, and 9, less
than 1% of the applied dose was degraded (photodecomposition of
¹⁴C-phosphonomethyl-labelled glyphosate) during 29-31 days, when
exposed to sunlight (PTRL Inc., 1990).

Lund-Hoie & Friestad (1986) exposed Roundup to several light
sources under different conditions. When exposed to UV light (lambda
= 254 nm) under laboratory conditions, concentrations of 1 and
2000 mg a.i./litre in deionized water showed DT50 values of 4 and 14
days, respectively. When exposed to sunlight under field conditions
1 mg a.i./litre in polluted water without sediment showed a much

slower decomposition (DT50 > 63 days), possibly due to pollution
preventing adequate UV penetration in the water. Polluted water with
sediments showed a rapid dissipation from water, probably due to
adsorption onto the sediments. In another field experiment 2 and
100 mg a.i./litre in deionized or polluted water without sediment
showed DT₅₀ values of < 28 days, when exposed to sunlight. At
the low concentration the dissipation in polluted water was more
rapid than in deionized water. In the dark no dissipation occurred.

In laboratory experiments 1 mg/litre of glyphosate in
sterilized natural and deionized water showed DT50 values of 4 to >
14 days when exposed to artificial light (350-450 nm) in
photoreactors without sediment (Monsanto, 1978a). In these
experiments Ca2+ acted as a photosensitizing agent.

Photodegradation by sunlight of glyphosate applied to a soil
appeared to be an insignificant route of dissipation (PTRL Inc.,
1989). In this study, ¹⁴C-glyphosate mixed with unlabelled
glyphosate was exposed for 31 days to natural sunlight, after
application to a sandy loam at a rate of 4.5 kg a.i./ha.
Extrapolated DT₅₀ values that were based on first-order kinetics
were 90 days in the sunlight and 96 days in the dark, indicating no
substantial degradation due to photolysis. The temperature of the
soil surface was 22-23 °C. Under unnatural light conditions
glyphosate appeared not to be photodegraded substantially (Monsanto,
1972c; Rueppel et al., 1977; Monsanto, 1978a).

4.3 Biodegradation

Appraisal

Selected studies of the biodegradation of glyphosate have been
considered; selection was on the basis of test conditions and modern
methodologies. There is considerable variation in rate of breakdown
in water, aquatic sediment and soil. Degradation occurs more rapidly
in aerobic than anaerobic conditions. Half-times for biodegradation
in the three media under laboratory conditions range between a few
days and approximately 20 days. No data on biodegradation under
anaerobic conditions are available.

The main route of biodegradation of glyphosate appears to be
by splitting the C-N bond to produce AMPA. However, a second route
with splitting of the C-P bond can also occur.

the field that bacterial populations adapt to the metabolism of
glyphosate. Presence of inorganic phosphate inhibits degradation of
glyphosate with some, but not all, bacteria. Biodegradation of
glyphosate may involve co-metabolism.

The most relevant laboratory experiments in which the
biodegradation in systems with water and sediment have been studied
are summarized in Table 5. These studies indicate that the rate of
biodegradation may vary substantially, depending on experimental
conditions, e.g., the availability of oxygen, temperature and type
of sediment. The time needed for 50% biodegradation of glyphosate in
the whole system of a test with water and sediment is < 14 days
under aerobic and 14-22 days under anaerobic conditions in the
laboratory.

In the experiments of PTRL East Inc. (1990a,b), less then 10%
of the applied activity was recovered in the pond water over a
period of 30 days under aerobic condition and 365 days under
anaerobic conditions. During all experiments more than 50% of the
applied activity was recovered in the sediment.

In experiments with water and their associated sediments the
amount of a.i. declines over time with a generally transient
increase of ¹⁴C-AMPA, an increase of ¹⁴CO₂, and an increase of
sediment-bound residues. An exception to this pattern of
biodegradation can be observed in some aerobic and anaerobic
experiments that were performed with pond water and a silty clay
loam sediment (PTRL East Inc, 1990a,b). In this water/sediment
system the biodegradation stopped after approximately 15 days under
aerobic conditions and after approximately 150 days under anaerobic
conditions. The glyphosate residues (a.i. plus AMPA) at both time
points remained approximately 40% of the applied dose, which
indicated substantial persistence in spite of the rapid initial
degradation.

AMPA is the main metabolite of glyphosate found in both the
water column and the sediment. Maximum amounts of AMPA under both
aerobic and anaerobic conditions in the sediment were 25% of the
applied activity (PTRL East Inc., 1990a,b). These maxima were found
at 7-20 days after application. In the same experiments maximum
amounts of sediment-bound residue were 9% of the applied activity
under aerobic conditions and 4% under anaerobic conditions. These
maxima were found at the end of the experiments. The amounts of
evolved ¹⁴CO₂ in these studies gradually increased in most cases
up to 24 and 35% of the applied activity after 30 days (aerobic),
and 365 days (anaerobic), respectively. This indicates substantial
differences in the mineralization rate. These differences are partly
due to the availability of oxygen, since under anaerobic conditions
the mineralization rate was slower than under aerobic conditions.

This was also found by Monsanto (1972a, 1978a). In the aerobic
experiments of Monsanto (1972a), four sediments that differed by up
to two orders of magnitude in the total number of micro-organisms
did not show substantial differences in mineralization rate.

Biodegradation studies with glyphosate in the soil under
conditions where unequivocal interpretation is justified are scarce.
Table 6 summarizes some relevant studies, indicating that the
biodegradation rate may differ substantially, depending on the
experimental conditions. The laboratory and greenhouse experiments
in Table 6 were performed with moisture contents (> 75% of the
field capacity) that were adequate for optimal biodegradation.

In most laboratory experiments the biodegradation rate of
glyphosate in soils appears to be rapid (see Table 6). Mostly
biodegradation can be described with linear first-order kinetics.

Sometimes a non-linear first-order model taking into account
spatial variability better describes the results observed (PTRL East
Inc., 1991):

C = C₀ (1 + ßt)-^alpha

C in this equation is the concentration in the soil at time t,
C₀ the initial concentration, and alpha and ß are rate constants
reflecting spatial variability.

The main metabolite under aerobic conditions of glyphosate in
soil is AMPA. In aerobic laboratory experiments the maximum amounts
in sandy loam and silt loam were 27 and 29%, respectively, of the
applied activity. These maxima were reached 14 days after
application (PTRL East Inc., 1991). From the data of PTRL East Inc.
(1991), DT₅₀ values for AMPA of approximately 50 days in sandy and
silty loam can be derived. That AMPA is more persistent than
glyphosate was also shown in a laboratory experiment with sandy loam
(Monsanto, 1972b). The amounts of AMPA after 111 days were 10-17% of
the applied activity. In this study, the temperature (32 °C) was
higher than in the other studies discussed above.

Some minor unidentified metabolites were quantified in an
aerobic laboratory experiment lasting 364 days with sandy loam and
silt loam (PTRL East Inc., 1991). Two unknown metabolites did not
exceed 3.5% of the applied activity, whereas some other unknown
metabolites did not exceed 1.5% each. Rueppel et al.(1977)
quantified some minor metabolites that did not exceed 1% of the
applied activity. These metabolites were
N-methylamino-methylphosphonic acid, glycine,
N,N-dimethylaminomethylphos-phonic acid, hydroxymethylphosphonic
acid, and two unknown metabolites.

In aerobic laboratory experiments, the amounts of soil-bound
residues immediately after application were 9-35% of the applied
dose, after which they showed an irregular time-course during these
experiments of approximately 112 days (Monsanto, 1972b). In general,
the initial amounts were also the maximum amounts. In other
laboratory experiments however, maximum amounts of soil-bound
residues appeared to be reached after 14 days, whereafter they
remained more or less constant or even decreased (PTRL East Inc.,
1991). These maximum amounts were 7-9% of the applied activity, and
were probably lower compared with other studies due to better
extraction procedures.

Mineralization in the soil occurs under both aerobic and
anaerobic conditions in the laboratory, although the rates may
differ greatly, apparently mainly depending on the soil respiration
rate and the temperature. When ¹⁴C-phosphonomethyl-labelled
glyphosate was applied to sandy loam and silt loam, 70-78% ¹⁴CO₂
evolved during an aerobic laboratory experiment of 360 days (PTRL
East Inc., 1991). In this study the application rate was 4 mg
a.i./kg dry weight. In an aerobic laboratory study with 15 Swedish
forest soils, DT₅₀ values based on ¹⁴CO₂ evolution varied
between 6 and 200 days. Mineralization was highly correlated with
the soil respiration rate, but not with pH or organic matter content
(Torstensson & Stark, 1981). This was confirmed by Torstensson &
Stenström (1986) and Heinonen-Tanski (1989). Torstensson & Stenström
(1986) reported that glyphosate was co-metabolized. In this case,
co-metabolizing microorganisms are not supplied with energy by
biodegrading glyphosate.

Establishing the correlation between soil respiration and
mineralization requires both a standardized measurement of the
respiration rate and an accurate measurement of the actual dose that
reaches the soil (Torstensson & Stenström, 1986). In a laboratory
experiment simulating temperatures under arctic conditions in forest
soils, 51-71% of the applied activity was recovered as ¹⁴CO₂ 217
days after application of ¹⁴C-glyphosate. In this study the
mineralization rate was reduced 10-15 times during a temperature
decrease of 10 °C over the first part of the study. The rate
increased only 3.7-4 times with a temperature increase of 10 °C
during the second part (Heinonen-Tanski, 1989).

Glyphosate in the soil appears to be degradable by
micro-organisms in two ways (Jacob et al., 1988), as shown in
Fig. 3. One route is via the formation of AMPA and a C₂ fragment
which might be glyoxylate. This scheme for degradation was proposed
by many researchers (Monsanto, 1972b; PTRL East Inc., 1991). In this
route the splitting of the C-N bond is the first step. There is,
however, another route of biodegradation via sarcosine
(N-methyl-glycine) and orthophosphate, after which sarcosine is

degraded to glycine and a one-carbon unit that eventually might form
CO₂ via formaldehyde (Kishore & Jacob, 1987; Jacob et al., 1988).
In this route the splitting of the C-P bond is the first step. In
experiments with ¹⁴C-glyphosate, isolated cultures of Pseudomonas
sp. strain LBr were able to degrade glyphosate according to both
routes (Jacob et al., 1988). Approximately 5% of the applied
¹⁴C-glyphosate was not degraded via AMPA, but via sarcosine.

The growth rate of bacteria isolated from a sandy loam garden
soil that was sprayed with Tumbleweed (a garden product) was less
inhibited by technical grade glyphosate than the growth rate of
bacteria from an unsprayed control (Quin et al., 1988). This
indicated adaptation of the bacterial populations of the sprayed
site. As addition of aromatic amino acids prevented growth
inhibition in the population of the unsprayed site to a greater
extent than in the population of the sprayed site, different
mechanisms of biochemical interference were indicated. The
composition of the bacterial population on the unsprayed site was
also different from the sprayed one. Pseudomonas sp. and
lactose-fermenting bacteria could be identified in an inoculum from
the sprayed soil able to use glyphosate as a sole source of
phosphorus (Quinn et al., 1988). A different regulatory mechanism
for biodegradation in unsprayed and sprayed sites was assumed: in
the latter the aromatic amino acid pathway might be regulated by
direct control of 3-deoxy-D-arabino-heptulosonate-7-phosphate (DAHP)
by the end-products, whereas in the unsprayed site DAHP synthase
might be indirectly regulated by prephenate. Also in other
experiments bacteria were shown to use glyphosate as a sole P source
(Kishore & Jacob, 1987; Pipke & Amrhein, 1988; Weidhase et al.,
1990), thereby primarily degrading glyphosate to orthophosphate and
sarcosine, by splitting the C-P bond. In the study of Weidhase et
al. (1990), 18.2% of the applied activity was recovered as sarcosine
8 h after application of ¹⁴C-1-methyl-labelled glyphosate to a
pure culture of Pseudomonas sp. GS. This biodegradation route of
glyphosate via sarcosine was also demonstrated by Kishore & Jacob
(1987). In their experiments with glyphosate as sole P source for
Pseudomonas sp. PG2982, one hour after application of
¹⁴C-labelled glyphosate, glycine, phosphate, and a one-carbon
unit, possibly formaldehyde, were identified as metabolites. After
one hour, the ¹⁴CO₂ evolution when the phosphonomethyl moiety
was labelled was substantially higher, as compared with the 1- or
2-glycine-labelled moieties. The authors suggested that the
so-called phosphate-starvation-inducible proteins, as identified by
others, might be responsible for cleaving the C-P bond. In an
experiment with pure cultures of a mutant of Arthrobacter sp.
GLP-1 able to use glyphosate as a sole P source, 90% of the applied
activity was released as orthophosphate at 240 h after application
of ¹⁴C-1-methyl-labelled glyphosate (Pipke & Amrhein, 1988).
Orthophosphate inhibited further biodegradation of glyphosate.
Flavobacterium sp. was found by Balthazor & Hallas (1986) to be

able to degrade glyphosate in spite of the presence of
orthophosphate. Liu et al. (1991) showed that 12 strains of bacteria
from the family Rhizobiaceae could degrade glyphosate present in the
medium as the sole phosphorus source; although growth of the
bacteria was slower than with inorganic phosphate. Sarcosine was the
intermediate breakdown product, indicating initial cleavage of the
C-P bond, in Rhizobium meliloti, the strain used for detailed
metabolic studies.

Carlisle & Trevors (1986a) deduced from their experiments that
nitrate-reducing bacteria are involved in metabolizing glyphosate.
Involvement of nitrifying bacteria in the biodegradation of
glyphosate was also demonstrated by Murthy et al. (1989), when they
investigated the treatment of waste water from a Roundup formulating
factory.

Pseudomonas sp. may use glyphosate as a sole P or C source,
as demonstrated by Weidhase et al. (1990). Only slight growth of the
wild-type strain of the bacterium Pseudomonas fluorescens was
observed with glyphosate as sole carbon or nitrogen source. The
herbicide was metabolized to aminomethylphosphonate (Zboinska
et al., 1992). Murthy et al. (1989) isolated a denitrifying
bacterial species that was also able to use glyphosate as a C
source. This species was isolated from activated sludge in a
waste-water treatment plant. A mutant of Arthrobacter sp. strain
GLP-1 was able to utilize glyphosate as a sole N source, whereas
this was not possible for the normal strain (Pipke & Amrhrein,
1988), probably due to the uptake of inorganic P released during
biodegradation.

As the Biological Oxygen Demand and the Chemical Oxygen Demand
of glyphosate are < 2 mg/g and 0.53 g/g, respectively, glyphosate
cannot be considered as readily biodegradable (LISEC, 1990a,b). In
suitable systems, however, glyphosate is biodegradable, as shown by
Murthy et al. (1989), who investigated the biodegradation of
glyphosate in waste-water treatment plants under different
conditions in sequencing batch reactors on a laboratory scale. These
reactors were fed with waste water from a Roundup manufacturing
facility. Glyphosate was degraded completely within one cycle of
24 h, independent of whether there was an initial aerated or anoxic
phase of 4 h. However, more glyphosate could be processed with an
anoxic initial phase, probably due to better conditions for
denitrification. Not only denitrifiers but also ammonifiers and
nitrifiers appeared to be involved in the biodegradation of
glyphosate. Only at the very high concentration of approximately
5000 mg a.i./litre was biodegradation repressed by non-glyphosate
COD and inhibited by excess ammonia production.

Pseudomonas sp. strain LBr, Flavobacterium sp. and a
denitrifying bacterial species were isolated from activated sludge
as species with the ability to use glyphosate as a P source
(Balthazor & Hallas, 1986; Jacob et al., 1988; Murthy et al., 1989).
The denitrifier was also able to use glyphosate as a sole C source.
Flavobacterium sp. degraded glyphosate to AMPA in both the
presence and absence of PO₄^3- (Balthazor & Hallas, 1986). In
this experiment the further degradation of AMPA appeared to be
hampered in the presence of PO₄^3-.

Pseudomonas sp. strain LBr was capable of completely
eliminating amounts of glyphosate up to 3212 mg/litre from a growth
medium (Jacob et al., 1988).

Continuous exposure of an activated sludge treatment system in
a pilot plant increased the ability of the sludge to metabolize
glyphosate to AMPA (Hallas et al., 1992). In this trial an influent
concentration of 50 mg a.i./litre was reduced to less than 5 mg
a.i./litre under continuous-flow conditions with an average
residence time of 10 min. The sludge was inoculated with immobilized
bacteria capable of degrading glyphosate. The effectiveness of the
treatment was dependent on the presence of a nitrogen source and a
non-glyphosate carbon source, and required a pH range of 6.0 to 8.0.

No data are available on the amounts of glyphosate that can be
eliminated in conventional waste-water treatment plants under
practical conditions. In waste water from glyphosate-producing
plants, 28-45% is reported to be eliminated through biological
treatment (Task Force on Water Quality Guidelines, 1991).

No data are available on the biodegradability of the
surfactants in formulations. It is, however, probable that
polyoxyethylene amine is biodegraded fairly rapidly in view of the
biodegradability of structurally related compounds (Swisher, 1987).

4.4 Bioaccumulation

Appraisal

Glyphosate is not expected to bioaccumulate in view of its
high water solubility and its ionic character. This was confirmed by
several laboratory experiments with fish, crustaceans and molluscs
and by field experiments.

In a static test, channel catfish (Ictalurus punctatus) were
exposed to 0.94-0.99 mg ¹⁴C-labelled a.i./litre (actual
concentrations) for 10 days (ABC Inc, 1981d; Monsanto, 1981a). Of
the absorbed amount, 76% was recovered in the viscera. More than 90%
of the extractable residues in the viscera and the fillet was
identified as glyphosate, whereas less than 2% was identified as
AMPA. After 10 days of depuration 80% of the absorbed activity was
eliminated. For exposed channel catfish the calculated
bioconcentration factor based on the activity absorbed by the whole
fish was 0.27. For depurated channel catfish the calculated
bioconcentration factor was 0.052.

The marsh clam (Rangia cuneata) and crayfish (Procambarus
simulans) were exposed in static tests lasting 28 days to
synthetic uncontaminated sea water and a sandy loam sediment that
was incorporated with 3 mg ¹⁴C-labelled a.i./kg (ABC Inc.,
1982d,e). These experiments were set up to assess the degree of
bioconcentration of glyphosate when used in flooded rice levees and
tidal water. The calculated bioconcentration factor for the edible

parts of the clam increased during exposure up to 4.8, whereas for
the whole crayfish it increased up to 12. The highest concentrations
in the edible parts of the clam and the whole crayfish were 0.3 mg
¹⁴C-labelled residues/kg for both. After 28 days of depuration 48%
of the accumulated residues were eliminated from the edible parts of
the clam. The concentration in these parts was then 0.16 mg
¹⁴C-residues/kg. The crayfish finally had eliminated 91% after 14
days of depuration. The concentration in the whole crayfish was then
0.02 mg ¹⁴C-residues/kg. It must be stated that this test refers
to the accumulation of ¹⁴C and not glyphosate.

In a static test without sediment, in which rainbow trout
(Salmo gairdnerii) were exposed to 2 mg a.i./litre (nominal
concentration) for 12 h, the fillets of the fish contained 80 µg
a.i./kg (in the original article the erroneous figure of 80 mg/kg
was reported), and the eggs 60 µg a.i./kg (Folmar et al., 1979).
This indicates a bioconcentration factor of 0.04 for the edible
parts. Roundup was applied in this test.

In a flow-through test in which bluegill sunfish (Lepomis
macrochirus) were exposed to 11-13 mg ¹⁴C-labelled a.i./litre
(actual concentrations) for 35 days, calculated daily
bioconcentration factors based on the whole fish increased from <
0.1, 0.2 days after the start of the test, to 0.4-0.5 at the end
(ABC Inc., 1989f). Maximum concentrations in the whole fish, viscera
and fillet were 13, 7.6 and 4.8 mg ¹⁴C-residues/kg, respectively.
The time required to reach 90% of the steady state and the uptake
rate constant were calculated to be 120 days and 0.02 mg/kg fish x
(mg/litre water)-1 x day-1, respectively. During 21 days of
depuration, the half-life of depuration was calculated to be 35. A
slow decrease in tissue concentration during depuration was
indicated. After the period of depuration 2.2 mg ¹⁴C-residues/kg
whole fish was still present. In an additional study to characterize
the ¹⁴C-residues, 95-97% of the residues in the water was
glyphosate, whereas in the whole fish and tissues 28-91% of the
recovered activity was glyphosate (ABC Inc., 1989g). In a whole fish
sample 21 days after starting the test, 49% of the recovered
activity was found to be AMPA. By treating homogenates with
proteinase K it was indicated that a substantial amount of the
absorbed residues was tightly associated with, or incorporated into,
protein.

In a field experiment in a forest ecosystem in Oregon, USA,
neither glyphosate nor AMPA were recovered in salmon fingerlings
(Oncorhynchus kisutch) after aerial application of Roundup at a
rate of 3.3 kg a.i./ha (Newton et al., 1984). The fingerlings were
released at the downstream edge of the sprayed site and analysed up
to 55 days after treatment. Glyphosate was not recovered in carp
(Cyprinus carpio) in a field experiment in which ponds were
sprayed with Roundup at rates of 1.3-1.4 kg a.i./ha (Monsanto,
1980). In this experiment of approximately 90 days, AMPA was not

recovered until 30 days after application. It then increased up to
0.21 mg/kg whole fish, remained constant for another 30 days, and
then decreased to around the limit of determination (0.1 mg/kg) at
the end of the experiment.

In a forest ecosystem in Oregon, USA, Roundup was aerially
applied at a rate of 3.3 kg a.i./ha (Newton et al., 1984).
Concentrations in mammals were of the same order of magnitude as the
concentrations in litter and ground cover. The concentrations of
glyphosate in the viscera of herbivorous small mammals decreased
more slowly than in omnivorous and carnivorous small mammals, which
was probably due to a higher ingestion of contaminated litter. The
highest concentration was found in the viscera of omnivorous
deermice (Peromysces maniculatus) immediately after spraying: 5 mg
a.i./kg. Only small traces of AMPA were found in mammalian viscera.

4.5 Waste disposal

Small amounts of glyphosate can be disposed of by mixing with
alkali and soil prior to burial in a pit or trench, whereas large
amounts should be incinerated in units equipped with effluent gas
scrubbing (IRPTC, 1991).

5. ENVIRONMENTAL LEVELS AND HUMAN EXPOSURE

Appraisal

The low toxicity, low volatility and low body absorption of
glyphosate makes its application by backpack sprayer safe under
field condition provided that the worker wears full protective
clothing.

5.1 Environmental levels

A synopsis of concentrations of glyphosate is tabulated in
Table 7. Measurements as part of regular monitoring programmes are
very scarce; measurements in field experiments with recommended
application rates simulating common agricultural practice are
therefore included in Table 7. Only maximum amounts are tabulated as
indicative values, since the rate at which they dissipate is not
included here (see sections 4.1 and 4.3). Data on the occurrence of
glyphosate and AMPA in sewage sludge are not available.

In biota the highest concentrations of glyphosate and AMPA were
found in fresh foliage and reindeer lichen (Cladonia rangiferina).
In abiota the highest concentrations of both compounds were found in
the soil (see Table 7). The occurrence of glyphosate in the
groundwater of Texas, USA, was reported by Hallberg (1989), but the
measured concentration and the year of measurement were not
specified.

Use of glyphosate as a herbicide may result in the presence of
residues in crops and animal tissues destined for human consumption.
Application as a herbicide may also be responsible for the presence
of glyphosate in drinking-water. Direct measurements of glyphosate
in foodstuffs (as part of food surveillance), drinking-water or
total diets have not been carried out. The only information
available comes from controlled residue studies. With technical
glyphosate formulated as the isopropylamine salt in aqueous
solution, numerous residue studies have been carried out in
vegetables, grasses, oil seeds, mammalian products, poultry products
and primary feed commodities. The results are summarized in the
various reports of the FAO/WHO Joint Meeting on Pesticide Residues
(FAO/WHO, 1986a, 1987, 1988). For detailed information on these
studies the reader is referred to these reports. The appraisals made
by the JMPR included the following more general statements.
Pre-harvest (5-14 days) application of glyphosate (isopropylamine
salt) in the cultivation of cereals results in significant residues
in the grain and plant materials. Studies are available to show the
fate of glyphosate in milling, baking and brewing. Residue levels in
white flour were approximately 10-20% of the levels in wheat, while
the bran residue levels were 2 to 4 times as high as those in the
wheat. Glyphosate residues were not lost during baking, but residue
levels decreased when bread was made from flour because of dilution.

Glyphosate residue levels in malt and beer derived from
field-treated barley were, respectively, about 25% and 4% of the
original level in the barley. Some glyphosate is lost during
washing, but most of the decrea