Pest Management Science Pest Manag Sci 64:417–421 (2008)

Managing the risk of glyphosate

resistance in Australian glyphosate-
resistant cotton production systems
Jeff A Werth,1,2,3∗ Christopher Preston,2 Ian N Taylor,4 Graham W Charles,3
Grant N Roberts5 and Jeanine Baker2
1 LeslieResearch Centre, Queensland Department of Primary Industries and Fisheries, Toowoomba, Australia
2 School of Agriculture, Food and Wine, University of Adelaide, and Cooperative Research Centre for Australian Weed Management,
Adelaide, Australia
3 NSW Department of Primary Industries, Cotton Catchment Communities Cooperative Research Centre, Narrabri, Australia
4 Cotton Research and Development Corporation, Narrabri, Australia
5 CSIRO Plant Industries, Cotton Catchment Communities Cooperative Research Centre, Narrabri, Australia

Abstract

BACKGROUND: Glyphosate-resistant cotton varieties are an important tool for weed control in Australian cotton
production systems. To increase the sustainability of this technology and to minimise the likelihood of resistance
evolving through its use, weed scientists, together with herbicide regulators, industry representatives and the
technology owners, have developed a framework that guides the use of the technology. Central to this framework
is a crop management plan (CMP) and grower accreditation course. A simulation model that takes into account
the characteristics of the weed species, initial gene frequencies and any associated fitness penalties was developed
to ensure that the CMP was sufficiently robust to minimise resistance risks.

RESULTS: The simulations showed that, when a combination of weed control options was employed in addition
to glyphosate, resistance did not evolve over the 30 year period of the simulation.

CONCLUSION: These simulations underline the importance of maintaining an integrated system for weed
management to prevent the evolution of glyphosate resistance, prolonging the use of glyphosate-resistant cotton.
 2007 Society of Chemical Industry

Keywords: transgenic glyphosate-resistant cotton; integrated weed management; resistance model; simulation
model

1 INTRODUCTION Ready and 11.5% to Roundup Ready Flex

Traditional weed management in Australian cotton
systems prior to the introduction of glyphosate-
resistant (Roundup Ready ) cotton was based on
sound integrated weed management principles.1 This
system, which evolved over time, consisted of the
widespread use of a diversity of herbicide inputs
and other alternative control methods including crop
rotations, cultivation and hand-hoeing of weeds.2
The adoption of an integrated approach to weed
management in the Australian production system
prevented the evolution of herbicide resistance in
weeds of cotton, including resistance to glyphosate.
Glyphosate-resistant cotton was introduced in the
Australian cotton industry in the 2000–2001 cotton
season. Since then it has been widely adopted,
with 75.5% of cotton planted being glyphosate
resistant in 2006–2007, 64% planted to Roundup
(enhanced glyphosate-resistant cotton, with adequate
gene expression in reproductive and vegetative parts
of the plant3 ). This rapid adoption was due to
the combination of (a) improved control of some
difficult-to-control weeds with glyphosate, which is
effective on a broad spectrum of weed species, and
(b) the ability to substitute glyphosate, a relatively
inexpensive and environmentally benign herbicide,
for some conventional inputs such as pre-emergence
residual herbicides which have often been associated
with early-season damage and poor growth of cotton
seedlings.4 In addition, the shift towards conservation
tillage practices and the decline in the use of
hand-hoeing has resulted in a greater reliance on
herbicides, particularly glyphosate, for the majority
of weed control in cotton.5 The increased use of
glyphosate associated with glyphosate-resistant cotton

∗
Correspondence to: Jeff A Werth, Leslie Research Centre, Queensland Department of Primary Industries and Fisheries, PO Box 2282, Toowoomba, Qld
4350, Australia
E-mail: jeff.werth@dpi.qld.gov.au
(Received 7 March 2007; revised version received 20 May 2007; accepted 7 June 2007)
Published online 13 December 2007; DOI: 10.1002/ps.1508

 2007 Society of Chemical Industry. Pest Manag Sci 1526–498X/2007/$30.00

Jeff A Werth et al.
has reduced the frequency of use of some alternative
weed control methods such as residual herbicides and
hand-hoeing, increasing the risk of weeds evolving
resistance to glyphosate.
2 MANAGING THE RISK
In order to manage the risk of glyphosate resistance
evolving, the Transgenic and Insecticide Management
Strategy Committee (TIMS) Herbicide Tolerant Crop
Technical Panel, consisting of a small group of
weed scientists and growers, in conjunction with
Monsanto, the Office of the Gene Technology
Regulator (OGTR) and the Australian Pesticides and
Veterinary Medicines Authority (APVMA), developed
a framework to guide the use of glyphosate-resistant
cotton.
2.1 Crop management plan
Central to this framework was the development of
a crop management plan (CMP) specific to the
particular technology as it was introduced into the
Australian production system. The APVMA required
that a CMP be developed prior to the commercial
release of any herbicide-resistant cotton variety.
Owing to the different glyphosate use patterns in
glyphosate-resistant cotton and enhanced glyphosate-
resistant cotton, a different CMP was developed
to accompany each of the technologies. The use
pattern of glyphosate-resistant cotton is restricted
to two over-the-top 1.035 kg AE ha−1 applications
before the four-leaf stage of the crop and a
directed/shielded 1.035 kg AE ha−1 prior to canopy
closure.6 Enhanced glyphosate-resistant cotton allows
three 1.035 kg AE ha−1 glyphosate applications up to
the 16-leaf stage and a 0.69 kg AE ha−1 application
between the 16- and 22-leaf stages. The CMP
encourages the use of integrated weed management
(IWM) practices in glyphosate-resistant cotton, with
the goal of minimising the likely risk of the evolution
of glyphosate-resistant weeds while protecting the
sustainability of the technology and herbicide.
Critical to the CMP is the requirement that any
weed that survives a glyphosate application must be
controlled by an alternative weed control tool before
it sets seed. With this in mind, the OGTR and
APVMA require detailed annual reports confirming
the successful management of glyphosate-resistant
cotton. This information is gathered from two weed
surveys conducted in glyphosate-resistant fields during
the cotton growing season. The results of these
surveys are compiled by Monsanto and form the basis
of the reports that are submitted to the regulatory
bodies. The surveys involve a detailed assessment of
surviving weeds present 10–14 days after a glyphosate
application, and reporting on the remedial action taken
to control these survivors. In the case of an adverse
event, such as a weed with suspected resistance, the
matter is referred to Monsanto, who investigate and
418
inform the TIMS Herbicide Tolerant Crop Technical
Panel.6
Monsanto, manufacturer of glyphosate and gly-
phosate-resistant cotton, also conducts random audits
of growers of glyphosate-resistant cotton to ensure
compliance with the conditions of the CMP. Those
who do not comply can be charged a compliance levy,
be denied access to seed and be referred to the relevant
regulatory authorities.6
A review of the CMP is conducted each year,
and results of the weed surveys are examined by
the TIMS Herbicide Tolerant Crop Technical Panel.
Grower concerns are also discussed, and changes to
the CMP are discussed and implemented as necessary.
This ensures that the CMP is constantly monitored
and flexible. A new CMP for enhanced glyphosate-
resistant cotton has been developed. This has been
modified to match the glyphosate-resistant cotton
system with the different glyphosate use patterns, and
to take into account lessons learnt from the initial
glyphosate-resistant cotton CMP.
2.2 Accreditation course
An important part of the CMP is a requirement that
an accreditation course be completed and passed
by those who sell the seed, provide advice on the
growth and management of the crop or grow and
make weed management decisions on glyphosate-
resistant cotton.6 The course educates participants
on how the glyphosate-resistant technology works,
how a glyphosate-resistant variety is developed, weed
management strategies in glyphosate-resistant cotton,
management of glyphosate-resistant cotton volunteers
and glyphosate application. The course also highlights
weed resistance management strategies for glyphosate-
resistant cotton, along with informing users of the legal
requirements.6
2.3 Simulation modelling
In order to determine whether the requirements of
the CMP are sufficient to prevent the evolution of
glyphosate-resistant weeds, a simulation model was
constructed to determine the potential for glyphosate
resistance evolution in glyphosate-resistant cotton.
The model was based on a weed management
system experiment that collected data over 3 years.
Two grass weeds, barnyardgrass [Echinochloa crus-galli
(L.) Beauv.] and liverseed grass (Urochloa panicoides
Beauv.) were chosen for the experiment as examples of
common weeds of Australian cotton. The emergence
characteristics of the two species under a range of
treatments (Table 1) and the survival of the species
after weed control actions were recorded. Additional
data on seed production and dose–mortality under
glyphosate were collected (default values are listed
in Table 2). These data formed the basis of the
population dynamics model which incorporated
resistance characteristics.
Resistance characteristics used in this model
were adopted from ‘example’ glyphosate-resistant
Pest Manag Sci 64:417–421 (2008)
DOI: 10.1002/ps
 Managing the risk of glyphosate resistance in glyphosate-resistant cotton

Table 1. Treatments used in systems experiment that formed the basis of the resistance model

Treatments

Action Glyphosate only Glyphosate + grass Glyphosate + residual Glyphosate + IWM

Post-plant pre-emergence herbicide Yes Yes Yes Yes

Pre-plant residual herbicide Yes
At-plant residual herbicide Yes Yes
3 × 1.5 kg Roundup Ready herbicide Yes Yes Yes Yes
Mid-season residual herbicide Yes Yes
Post-emergent herbicide Yes Yes
Interrow cultivation Yes

Residual herbicides used Herbicide Rate (g AE/AI ha−1 )

Post-plant pre-emergence herbicide Glyphosate 450

Pre-plant residual herbicides Trifluralin 1104
Diuron 1530
At plant residual herbicides Fluometuron 748
Prometryn 748
Pendimethalin 990
Mid-season residual herbicide Prometryn 1080

Table 2. Default parameter settings use in resistance model

Rfreq = 1 x 10-8
simulations 100
90
Description Default
Percent resistant alleles

80
Field size 50 ha 70
Initial seed bank density per metre 750 60
Initial resistant gene frequency9 1 × 10−8 or 10−6 50
Probability of resistant alleles surviving 1 40
glyphosate application
30
Probability of heterozygous alleles 0.8
20
surviving glyphosate application7,8
10
Probability of weeds surviving haloxyfop 0.5–0.15
application11 0
0 5 10 15 20 25 30
Probability of weeds surviving tillage 0.3
Season
(assuming 70% area covered by
interrow cultivator) Rfreq = 1 x 10-6
Maximum seed production 100
Barnyardgrass11 13 000 seeds m−2 90
Liverseed grass11 1950 seeds m−2 80
Percent resistant alleles

Fitness penalty of homozygous resistant 0.3 70

alleles9,12 60
Fitness penalty of heterozygous alleles11 0.15 50
Percent self-fertilising 40
Barnyard grass13 0.85
30
Liverseed grass11 0.6
20
10
0
weeds Lolium rigidum (Gaud.) and Conyza canaden- 0 5 10 15 20 25 30
sis (L.) Cronq. References indicate that glyphosate Season
resistance is inherited as an incompletely dominant Figure 1. Cumulative probabilities for predicted rates of glyphosate
nuclear trait.7,8 The model applied initial resistance resistance evolution for barnyardgrass for treatments in Table 1:

gene frequencies similar to those for L. rigidum.9
The model simulated resistance evolution with
glyphosate only; ♦ glyphosate + grass;
 glyphosate + IWM.
glyphosate + residual;
°
respect to the treatments over 30 years. The
simulations used initial resistance gene frequencies which the proportion of resistant alleles reached 50%)
of 1 × 10−8 and 1 × 10−6 (Figs 1 and 2). only when glyphosate was used alone. This point was
The weed management strategy chosen had a reached after 12 years. The simulation showed that, at
major influence on the rate of glyphosate resistance this initial gene frequency, resistance would not occur
evolution. At the lower initial resistant gene frequency, within 30 years if alternative weed control methods
resistance evolved (determined as the frequency at were used in addition to glyphosate.

Pest Manag Sci 64:417–421 (2008) 419

DOI: 10.1002/ps
Jeff A Werth et al.

Rfreq = 1 x 10-8 herbicides and hand-hoeing, is likely to be required to

100 minimise the potential for resistance evolution. This is
90 illustrated by the simulations with the full integrated
Percent resistant alleles

80 weed management strategy that included the use

70 of glyphosate. The simulation model demonstrated
60 that, in populations where the initial frequency for
50 resistance was relatively high, that is, 1 × 10−6 ,
40
resistance to glyphosate did not evolve within 30 years
30
if a fully integrated weed management strategy was
20
employed.
10
0
0 5 10 15 20 25 30
Season 3 CONCLUSION
These simulations gave an approximate timeframe
Rfreq = 1 x 10-6
100 of resistance evolution on the basis of mea-
90 sured population dynamics characteristics, with
unknown resistance characteristics being estimated by
Percent resistant alleles

80
70
60
50
40
30
20
10
0 The findings from the simulations support the
0 5 10 15 20
Season
Figure 2. Cumulative probabilities for predicted rates of glyphosate
resistance evolution for liverseed grass for treatments in Table 1:

glyphosate only; ♦ glyphosate + grass;
 glyphosate + IWM. °
glyphosate + residual;
When the initial gene frequency for resistance
was altered to 1 × 10−6 , the timeframe estimated
for resistance evolution decreased to 8 years when
glyphosate was used alone. Under this higher
initial gene frequency, differences between the weed
management treatments imposed were observed.
The addition of a post-emergent grass herbicide
option delayed the onset of resistance by 1 year.
Similarly, the use of a soil-applied residual herbicide at
planting and mid-season delayed resistance evolution
by 2 years. Although these treatments had additional
weed management options, there was still a heavy
reliance on glyphosate to provide the bulk of the weed
control. The addition of soil-applied herbicides had a
greater effect on liverseed grass than on barnyardgrass
at the higher initial gene frequency as a result of
differing emergence patterns. Liverseed grass emerged
in one major cohort, while barnyardgrass emergence
was more prolonged, with the soil-applied residual
herbicide applied at planting being therefore more
effective.10
In situations where weed pressure is low, or
the initial resistant gene frequency for the target
weed is 1 × 10−8 or less, the incorporation of a
single additional weed control option may prove
suitable for resistance prevention. However, in higher
weed pressure systems, a combination of weed
control options, such as residual and post-emergent
extrapolation.11 The simulations showed that employ-
ing an integrated strategy for weed control is the
best way to prevent resistance evolution. Weed con-
trol strategies must complement each other so that
plants surviving a herbicide treatment are controlled
by another treatment, be it mechanical or via a different
herbicide mode of action.11
25 30
resistance management strategies in the CMP.
Preventing seed set from survivors of glyphosate
applications by using alternative weed control options
minimises the chances of resistance evolution. One of
the important benefits of the CMP is the accreditation
course. Through this, growers are educated on
the principles of resistance management and the
importance of implementing an integrated weed
management strategy. The combined aspects of this
course, along with penalties for non-compliance with
the CMP, should lead to sustainable use of glyphosate-
resistant cotton in the Australian cotton industry. It
is also essential that the industry continues to practise
proactive resistance management, and that these issues
are revisited as other new herbicide resistance traits
become available.
ACKNOWLEDGEMENTS
This work was funded by the Cotton Catchment
Communities Cooperative Research Centre (CRC),
the CRC for Australian Weed Management and
Monsanto Australia Ltd.
REFERENCES
1 Charles GW, A grower survey of weeds and herbicide use in
the New South Wales cotton industry. Aust J Exp Agric
31:387–392 (1991).
2 Roberts GN, Developing integrated weed management systems
for cotton, in Proc 2nd World Cotton Conference, 6–12
September, Athens, Greece (1998).
3 May OL, Culpepper AS, Cerny RE, Coots CB, Corken CB,
Cothren JT, et al, Transgenic cotton with improved resistance
to glyphosate herbicide. Crop Sci 44:234–240 (2004).
4 Charles GW, Constable GA and Kennedy IR, Current and
future weed control practices in cotton: the potential use

420 Pest Manag Sci 64:417–421 (2008)

DOI: 10.1002/ps
 Managing the risk of glyphosate resistance in glyphosate-resistant cotton
of transgenic herbicide resistance, in Herbicide-resistant Crops
and Pastures in Australian Farming Systems, ed. by Mclean GD
and Evans G. Bureau of Resource Sciences, Parkes, ACT,
Australia, pp. 89–100 (1995).
5 Faircloth WH, Patterson MG, Dale Monks C and Good-
man WR, Weed management programs for glyphosate-
tolerant cotton. Weed Technol 15:544–551 (2001).
6 Roundup Ready Flex Cotton Technical Manual. Monsanto
Australia Ltd. http://www.monsanto.com.au/content/cotton/
rr flex/RRFlexQA.pdf (2006).
7 Lorraine-Colwill DF, Powles SB, Hawkes TR and Preston C,
Inheritance of evolved glyphosate resistance in Lolium rigidum
(Gaud.). Theor Appl Genet 102:545–550 (2001).
8 Zelaya IA, Owen MDK and VanGessel MJ, Inheritance of
evolved glyphosate resistance in Conyza canadensis (L.) Cronq.
Theor Appl Gen 110:58–70 (2004).
Pest Manag Sci 64:417–421 (2008)
DOI: 10.1002/ps
9 Neve P, Diggle AJ, Smith FP and Powles SB, Simulating
evolution of glyphosate resistance in Lolium rigidum 1:
population biology of a rare resistance trait. Weed Res
43:404–417 (2003).
10 Christoffers MJ, Genetic aspects of herbicide resistant weed
management. Weed Technol 13:647–652 (1999).
11 Werth JA, Weed resistance risk management in glyphosate-
resistant cotton. PhD Thesis, University of Adelaide, 201 pp.
(2007).
12 Jordan N, Fitness effects of the triazine resistance mutation in
Amaranthus hybridis: relative fitness in maize and soybean
crops. Weed Res 39:493–505 (1999).
13 Maun MA and Barrett SCH, The biology of Canadian weeds.
77. Echinochloa crus-galli (L.) Beauv. Can J Plant Sci
66:739–759 (1986).
421