UNITED
STATES
ENVIRONMENTAL
PROTECTION
AGENCY
WASHINGTON,
D.
C.
20460
OFFICE
OF
PREVENTION,
PESTICIDES
AND
TOXIC
SUBSTANCES
MEMORANDUM
SUBJECT:
BPPD
review
of
data
and
support
materials
submitted
by
Monsanto
to
amend
the
insect
resistance
management
requirements
for
Bollgard
II
Bt
cotton.
EPA
Reg.
No.
524­
522.
DP
Barcode:
327258
Decision:
363974
MRID#:
467172­
01,
­
02,
­
03.

TO:
Leonard
Cole,
Regulatory
Action
Leader
Microbial
Pesticides
Branch
Biopesticides
and
Pollution
Prevention
Division
(
7511C)

FROM:
Alan
Reynolds,
Entomologist
Microbial
Pesticides
Branch
Biopesticides
and
Pollution
Prevention
Division
(
7511C)

and
Sharlene
Matten,
Ph.
D.,
Biologist
Microbial
Pesticides
Branch
Biopesticides
and
Pollution
Prevention
Division
(
7511C)

Action
Requested
BPPD
has
been
asked
to
review
data
and
support
materials
submitted
by
Monsanto
to
justify
a
request
to
amend
the
Insect
Resistance
Management
(
IRM)
requirements
for
Bollgard
II
(
Cry1Ac
and
Cry2Ab2)
Bt
cotton
(
EPA
Reg
No.
524­
522).
Specifically,
Monsanto
is
requesting
to
remove
the
requirement
for
structured,
non­
Bt
cotton
refuges
in
the
southeast
U.
S.
cotton
belt
(
East
Coast
to
Texas).
The
submitted
materials
include
a
description
of
tobacco
budworm
(
TBW)
biology
as
it
pertains
to
IRM,
results
of
a
2
year
sampling
program
designed
to
determine
­
2
­
the
plant
host
origin
of
TBW
throughout
the
southeast,
an
assessment
of
TBW
natural
refuge
(
i.
e.
production
from
alternative
host
plants
and
crops),
a
calculation
of
"
effective"
refuge
(
i.
e.
the
proportion
of
TBW
not
exposed
to
Bt
proteins),
revised
simulation
models
based
on
the
data
collected,
and
an
overall
assessment
of
Bollgard
II
IRM
for
TBW
and
cotton
bollworm
(
CBW)
in
the
southeast
U.
S.
cotton­
growing
region.
Information
obtained
from
two
follow­
up
letter
from
Monsanto
to
the
Agency
(
dated
April
3
and
April
7,
2006)
is
also
included
in
this
review.
Bollgard
(
Cry1Ac)
Bt
cotton
(
524­
478)
is
not
addressed
in
a
separate
amendment,
though
is
discussed
in
the
context
of
the
Bollgard
II
amendment
request
and
will
be
considered
in
this
review.

Conclusions
1)
BPPD
agrees
with
Monsanto
that
the
TBW
sampling,
gossypol
analyses,
and
effective
refuge
determinations
clearly
demonstrate
that
a
significant
portion
of
the
TBW
population
is
derived
from
non­
cotton
hosts.
However,
questions
remain
on
the
larger
issue
of
whether
these
alternate
hosts
can
serve
as
unstructured
refuge
for
TBW
for
Bollgard
II
Bt
cotton.
Within
the
Cotton
Belt,
the
gossypol
data
differed
greatly
by
state/
region,
although
resistance
modeling
scenarios
with
TBW
showed
that
Bollgard
II
was
predicted
to
retain
durability
for
at
least
30
years
for
all
regions.
On
the
other
hand,
there
are
a
number
of
unresolved
issues
with
the
data,
modeling,
and
interpretation
of
the
results
(
detailed
below).
Should
these
issues
be
resolved,
BPPD
could
support
the
use
of
natural
refuge
alone
for
Bollgard
II
in
the
southeastern
U.
S.
cotton
region.

2)
The
strongest
case
for
the
use
of
natural
refuge
for
TBW
resistance
management
can
be
made
in
North
Carolina
and
Georgia.
In
this
region,
the
proportion
of
non­
cotton­
origin
TBW
was
consistently
high
(>
90%)
throughout
the
cotton
growing
season.
The
trends
were
similar
for
both
2004
and
2005.
These
states
are
also
known
to
contain
substantial
acreage
of
alternate
crops
(
peanut,
tobacco,
and
soybean)
that
are
known
to
be
preferred
TBW
hosts.
Resistance
modeling
incorporated
with
the
estimations
of
effective
refuge
using
worst
case
counties
predicted
that
Bollgard
II
retained
its
durability
for
at
least
30
years.

3)
In
other
regions
(
i.
e.
the
Mississippi
Delta,
eastern
Texas)
the
use
of
natural
refuge
alone
for
Bollgard
II
cotton
is
less
certain.
In
these
areas,
the
data
were
variable
and
difficult
to
interpret.
The
specific
proportions
of
cotton­
origin
TBW
varied
by
state
and
county,
though
non­
cotton­
origin
TBW
usually
were
significant
(
at
least
20%
of
the
population)
except
in
a
few
cases.
In
some
counties,
the
proportion
of
non­
cotton­
origin
moths
was
below
20%
while
in
other
counties
the
percentage
of
non­
cotton
moths
was
high
(
over
90%)
during
the
same
time
periods.
The
general
trends
were
high
proportions
of
non­
cotton
TBW
early
in
the
season,
followed
by
a
decline
after
June.
These
results
seem
to
indicate
that
cotton
is
a
much
more
important
host
for
TBW
in
these
states
then
in
the
east
coast
states
and
that
fewer
alternate
hosts
may
be
available
during
the
cotton
season.
However,
like
North
Carolina/
Georgia,
resistance
modeling
for
TBW
and
CBW
with
the
estimations
of
effective
refuge
using
worst
case
counties
showed
that
Bollgard
II
was
predicted
to
retain
durability
for
at
least
30
years
(
though
one
scenario
in
Texas
with
CBW
showed
slightly
less
than
30
­
3
­
years
durability).

4)
No
statistical
correlation
comparisons
were
made
between
the
two
growing
seasons
in
which
samples
were
taken
(
2004
and
2005).
Because
of
this,
comparisons
of
the
trends
between
growing
seasons
can
be
conducted
only
on
a
qualitative
basis.

5)
BPPD
identified
a
number
of
questions
regarding
the
TBW
sampling
methodology.
First,
the
use
of
pheromone
traps
restricted
the
collections
to
male
TBW
(
females
are
not
attracted
to
the
pheromone).
It
is
unclear
whether
inferences
about
female
behavior
(
i.
e.
host
utilization)
can
be
made
from
data
obtained
solely
from
males.
Second,
BPPD
also
questions
whether
the
two
year
sampling
period
(
2004
­
2005)
is
adequate
for
an
experiment
of
this
magnitude.
Though
relatively
consistent,
cropping
and
landscape
patterns
can
change
over
time
­­
a
factor
that
may
not
be
reflected
in
a
study
of
limited
duration.
Given
the
inherent
variability
in
natural
and
agronomic
ecosystems,
a
two
year
study
may
present
only
a
"
snapshot"
of
the
host
availability
and
productivity
for
TBW
that
may
not
be
representative
of
future
conditions.

6)
It
was
noted
that
for
a
number
of
counties,
trapping
locations,
and
trapping
periods
there
were
low
trap
captures
of
TBW.
For
many
traps,
few
(
or
no)
moths
were
collected.
Low
numbers
of
TBW
have
also
hampered
recent
Bt
cotton
resistance
monitoring
efforts
(
see
BPPD
2005),
possibly
due
to
a
suppressive
effect
from
the
widespread
adoption
of
transgenic
Bt
cotton.
Data
from
collection
sites
with
few
captures
(
i.
e.
<
10)
were
included
in
the
effective
refuge
calculations,
but
not
in
the
modeling
(
where
worst
case
scenarios
were
used).
It
is
unclear
if
the
low
trap
captures
in
some
areas
may
have
affected
the
reliability
of
the
results.
Also,
low
numbers
of
TBW
could
present
a
concern
if
insufficient
susceptible
insects
were
available
to
mitigate
a
resistant
population
emerging
locally
from
a
Bt
cotton
field.

7)
Overall,
the
CBW
and
TBW
modeling
predicts
that
Bollgard
II
cotton
should
have
more
than
25
years
of
durability
in
all
regions
using
natural
refuge
as
the
only
source
of
susceptible
insects.
It
should
be
noted
that
the
"
number
of
years"
is
an
imprecise
designation
of
relative
value,
not
an
absolute
value.
On
a
relative
basis,
Bollgard
II
cotton
has
much
higher
durability
than
Bollgard
cotton.
Monsanto's
modeling
does
not
account
for
any
prior
selection
for
Cry1Ac
resistance.
Uncertainties
in
the
modeling
parameters
and
assumptions
will
impact
the
modeling
output.
The
extent
to
which
the
modeling
output
will
be
impacted
by
these
uncertainties
is
unknown.
The
presence
of
larger
amounts
of
Bollgard
II
cotton
in
the
marketplace
is
predicted
to
increase
the
durability
of
both
Bollgard
II
and
WideStrike
cotton
products
that
are
present
in
the
landscape.
On
the
other
hand,
larger
amounts
of
Bollgard
cotton
in
the
marketplace
are
predicted
to
decrease
the
durability
of
Bollgard
II
and
WideStrike
cotton.

8)
Bollgard
II
(
Cry1Ac
and
Cry2Ab2)
currently
exists
in
a
mosaic
with
single
gene
Bollgard
(
Cry1Ac),
with
Bollgard
accounting
for
the
vast
majority
of
Bt
cotton
acreage.
Modeling
work
has
shown
that
in
such
a
scenario,
the
pyramided
product
can
be
at
risk
for
resistance
if
­
4
­
there
is
significant
acreage
of
a
single
gene
variety.
The
single
gene
variety
may
in
a
sense
act
as
a
"
stepping
stone"
for
resistance
to
the
pyramided
product:
if
resistance
develops
to
the
Bollgard
then
Bollgard
II
would
effectively
become
a
single
gene
product
(
Cry2Ab2).
Monsanto's
natural
refuge
proposal
contained
no
details
on
how
this
mosaic
might
be
managed.
There
was
no
discussion
as
to
whether
Monsanto
will
"
phase
out"
Bollgard
to
facilitate
the
adoption
of
Bollgard
II.
Rather,
it
was
suggested
in
the
submission
that
if
there
is
no
requirement
to
plant
structured
refuge
a
strong
incentive
will
exist
for
growers
to
switch
to
Bollgard
II.
However,
considering
the
large
acreage
of
Bollgard
currently
in
the
marketplace,
it
is
highly
likely
that
a
significant
percentage
of
cotton
acreage
will
remain
planted
with
Bollgard
varieties.

9)
Field
resistance
to
Cry1Ac
places
additional
selection
pressure
on
the
Cry2Ab2
component
of
Bollgard
II
cotton.
Encouraging
the
adoption
of
Bollgard
II
will
increase
the
overall
durability
of
all
three
Bt
cotton
products.
From
an
insect
management
point
of
view,
removal
of
Bollgard
cotton
from
the
marketplace
should
be
encouraged
as
quickly
as
possible.

10)
Monsanto's
proposal
to
use
natural
refuge
for
Bollgard
II
represents
a
switch
from
structured
refuge
(
i.
e.
refuge
planted
and
managed
by
growers)
to
an
unstructured
(
i.
e.
unmanaged)
refuge.
In
developing
IRM
strategies,
there
have
been
three
major
considerations
for
evaluating
structured
refuge
that
could
also
be
applied
to
natural
(
unstructured
refuge):
i)
production
of
a
sufficient
number
of
susceptible
insects
relative
to
any
resistant
survivors
of
the
Bt
crop,
ii)
proximity
of
the
refuge
to
the
transgenic
crop
to
facilitate
random
mating
between
susceptible
(
from
the
refuge)
and
resistant
(
from
the
Bt
crop)
insects,
and
iii)
developmental
synchrony
of
the
refuge
with
the
transgenic
crop
to
promote
random
mating.
For
the
most
part,
Monsanto's
sampling
and
gossypol
assays
have
addressed
these
three
criteria.
Proximity
and
synchrony
were
addressed
by
the
sampling
strategy
(
traps
were
placed
next
to
cotton
fields
and
monitored
throughout
the
growing
season).
Production
of
susceptible
insects
was
not
evaluated
from
a
numerical
perspective
per
se;
rather
the
proportion
of
moths
originating
on
cotton/
non­
cotton
sources
was
assessed.
However,
BPPD
noted
(
as
described
in
#
6
above)
that
low
numbers
of
TBW
trap
captures
were
reported
for
many
areas.
Low
number
of
TBW
could
be
a
concern
if
a
resistant
population
were
to
emerge
locally
from
a
Bt
cotton
field
without
a
sufficient
number
of
susceptible
TBW
available
to
mitigate
the
threat.

11)
BPPD
generally
agrees
with
Monsanto
that
a
natural
refuge
strategy
(
allowing
planting
of
100%
Bollgard
II)
will
result
in
economic
and
environmental
benefits.
Bt
cotton
refuges
are
currently
planted
as
non­
Bt
(
conventional)
cotton
on
either
5
or
20%
of
growers'
total
cotton
acreage
and
may
be
treated
with
pesticides
under
the
20%
option.
Benefits
to
growers
will
result
largely
from
yield
increases
and
reduced
insecticide
use
on
cotton
acreage
previously
planted
as
refuge.
However,
these
benefits
will
likely
be
compromised
should
the
lepidopteran
target
pests
develop
resistance
to
the
Bollgard
II
toxins.

12)
If
the
natural
refuge
proposal
is
ultimately
accepted
for
Bollgard
II,
it
is
recommended
that
Monsanto
continue
to
conduct
resistance
monitoring
for
the
Cry2Ab2
toxin
with
both
TBW
­
5
­
and
CBW.
Also,
considering
that
replacement
of
structured
refuge
with
natural
refuge
may
increase
the
Bollgard
II
acreage
(
and
selection
pressure
for
resistance),
it
may
be
justified
to
increase
the
monitoring
efforts
for
Cry2Ab2
(
i.
e.
increased
sampling
sites
and
collections).

Background
[
NOTE:
This
review
will
focus
primarily
on
submitted
data
for
TBW
and
modeling
for
TBW
and
CBW.
Previously
reviewed
data
for
CBW
are
summarized
below
in
this
section.]

Bollgard
Bt
cotton
(
EPA
Reg.
No.
524­
478),
expressing
the
Cry1Ac
toxin,
was
registered
for
commercial
use
beginning
with
the
1996
growing
season.
The
three
major
target
pests
of
Bollgard
are,
tobacco
budworm
(
TBW),
cotton
bollworm
(
CBW),
and
pink
bollworm
(
PBW).
Bollgard
demonstrates
high
toxicity
against
TBW
and
PBW
(
considered
to
be
a
"
high
dose")
and
has
moderate,
but
still
effective,
activity
against
CBW
(
a
non­"
high
dose").

Since
Bollgard
(
and
other
Bt
crops)
express
insecticidal
toxins
at
high
levels
throughout
the
growing
season,
there
is
likely
to
be
high
selection
pressure
for
the
development
of
resistance
among
the
target
pests.
Therefore,
as
part
of
the
terms
and
conditions
of
the
Bollgard
registration,
EPA
imposed
insect
resistance
management
(
IRM)
requirements
including
the
use
of
structured,
non­
Bt
cotton
refuges.
These
refuges
will
theoretically
produce
susceptible
insects
not
exposed
to
the
Bt
toxin
that
should
be
available
to
mate
with
rare
resistant
insects
emerging
from
the
Bt
fields,
thus
diluting
the
resistance
genes
in
the
pest
population.
In
1995,
the
initial
refuge
strategy
for
Bt
cotton
included
the
use
of
either
a
4%
untreated
refuge
or
a
20%
insecticide­
treatable
refuge.
Other
IRM
requirements
included
resistance
monitoring,
remedial
action
plans
(
in
the
event
of
resistance),
grower
education
initiatives,
and
compliance
reporting.
In
2001,
the
Agency
reassessed
the
existing
Bt
crop
registrations
and
revised
the
refuge
requirements
for
Bt
cotton.
Based
on
the
reassessment,
the
following
three
refuge
options
were
established
for
Bollgard
cotton
(
TBW
and
CBW
resistance
management):

1)
A
5%
external,
untreated
structured
refuge
(
must
be
within
½
mile
of
Bollgard
fields,
preferably
within
1/
4
mile
or
adjacent);
2)
A
20%
external,
treatable
structured
refuge
(
must
be
within
1
mile
of
Bollgard
fields,
preferably
within
½
mile);
3)
A
5%
embedded
refuge
(
must
be
planted
as
a
continuous
block
within
the
Bollgard
field
and
must
be
must
be
at
least
150
feet
wide,
but
preferably
300
feet
wide).

"
Structured"
refers
to
the
management
of
the
refuge,
in
which
the
non­
Bt
cotton
acreage
is
planted
using
similar
hybrids
and
agronomic
practices
as
the
Bt
cotton
portion.
Similar
management
of
the
refuge
and
Bt
crop
is
important
to
ensure
that
susceptible
moths
from
the
refuge
emerge
at
the
same
time
and
in
close
proximity
to
any
potentially
resistant
moths
from
the
Bt
field,
thus
increasing
the
probability
of
random
mating.

Shortly
after
the
2001
reassessment,
registration
was
approved
for
Bollgard
II
Bt
cotton
(
524­
­
6
­
522).
Bollgard
II
is
a
"
pyramided"
event,
expressing
both
the
Cry1Ac
and
Cry2Ab2
Bt
toxins.
A
pyramided
transgenic
crop
can
offer
advantages
for
IRM
in
that
multiple
toxins
are
expressed
simultaneously.
This
can
decrease
the
likelihood
of
pest
resistance
development,
since
an
insect
resistant
to
one
toxin
will
still
be
susceptible
to
the
other
(
provided
there
is
no
cross
resistance
potential
between
the
toxins).
Like
Bollgard,
Bollgard
II
is
targeted
primarily
at
TBW,
CBW,
and
PBW,
though
with
the
combination
of
lepidopteran­
toxic
proteins,
Bollgard
II
is
considered
to
be
a
"
high
dose"
for
all
three
pests.
Since,
Bollgard
II
exceeds
the
dose
and
toxin
profile
for
Bollgard,
the
same
refuge
options
detailed
above
were
also
applied
as
terms
and
conditions
of
the
Bollgard
II
registration.

CBW
Alternate
Host
Analysis
As
an
additional
term
and
condition
of
the
Bollgard
II
registration,
Monsanto
was
required
to
investigate
whether
alternate
(
non­
cotton)
hosts
can
serve
as
refuge
for
CBW.
There
were
concerns
that
the
5%
unsprayed
refuge
option
might
not
be
sufficient
for
CBW,
considering
that
the
pest
is
typically
less
susceptible
to
Bt
toxins.
It
is
well
known
that
CBW
is
highly
polyphagous
and
utilizes
a
wide
range
of
plant
hosts.
However,
it
was
unknown
whether
these
hosts
might
serve
as
an
"
unstructured"
refuge
for
large
scale
planting
of
Bollgard
and
Bollgard
II
cotton.
In
response
to
the
term
and
condition,
Monsanto
submitted
an
interim
and
final
report
on
non­
cotton
CBW
refuge
to
the
Agency
that
have
been
reviewed
by
BPPD
(
see
BPPD
2004a).

To
assess
the
suitability
of
non­
cotton
hosts,
Monsanto
conducted
a
series
of
experiments
during
2002
and
2003
focusing
primarily
on
four
alternate
CBW
hosts:
peanut,
sorghum,
field
corn
and
soybeans.
The
experiments
consisted
of
a
CBW
sampling
program
on
the
various
host
crops,
a
"
C3/
C4"
host
determination
analysis
on
adult
CBW,
an
analysis
of
cropping
patterns,
and
satellite
mapping
of
alternate
crop
hosts.
From
these
data,
Monsanto
was
able
to
determine
CBW
production
in
various
crop
hosts
at
different
points
in
time
during
the
growing
season.

CBW
adults
were
collected
in
pheromone
traps
positioned
around
the
various
host
crop
fields
in
Arkansas,
Louisiana,
Mississippi,
Georgia
and
North
Carolina.
Larvae
were
also
sampled
to
determine
which
crop
hosts
supported
larval
CBW
populations
at
any
given
time.
The
"
C3/
C4"
analysis
was
based
on
the
ratio
of
13C
to
12C
found
in
the
wings
of
sampled
CBW,
which
can
be
used
to
determine
which
host
plants
were
utilized
by
the
insect
during
development.
The
ratios
differ
for
"
C3"
and
"
C4"
plants:
C3
plants
include
cotton,
peanuts,
clover,
and
soybeans;
while
C4
hosts
include
corn
and
sorghum.
Wings
from
sampled
moths
were
sampled
using
the
"
C3/
C4"
assays
and
compared
to
known
standards.
To
supplement
the
sampling
and
host
origin
experiments,
satellite
mapping
of
the
regions
(
1
or
10
miles)
around
the
traps
was
also
conducted.

Monsanto's
analysis
showed
that
CBW
larvae
were
typically
found
on
corn
early
in
the
growing
season
(
early
June)
and
on
other
crops
starting
several
weeks
later.
CBW
presence
on
sorghum
and
cotton
was
detected
mid­
June/
early
July,
while
on
sorghum
and
peanuts
populations
were
observed
in
mid­
July/
August.
Overall,
CBW
populations
from
the
alternate
crop
hosts
overlapped
occurrence
of
all
CBW
generations
on
cotton
and
that
larval
population
numbers
­
7­
were
comparable
between
the
crop
types.
From
the
CBW
adult
trap
captures,
similar
numbers
of
adult
CBW
were
observed
at
all
the
different
crop
interfaces
in
a
given
state.
The
comparable
numbers
captured
at
each
crop
interface
suggest
that
the
size
of
the
adult
CBW
populations
captured
in
the
traps
is
not
a
function
of
the
local
host
crops.
This
is
because
the
timing
and
the
number
of
adults
captured
is
greater
than
what
would
be
predicted
based
on
the
adult
emergence
numbers
from
these
host
crop
fields.
Monsanto
concluded
that
the
CBW
adults
must
be
moving
broadly
across
the
landscape
in
search
of
alternative
host
crops,
corn
in
the
early
season
and
other
crop
hosts
later
in
the
season.

The
C3/
C4
assays
confirm
the
literature
on
CBW
host
use
and
the
qualitative
predictions
made
based
on
the
larval
populations.
Early
in
the
season,
the
CBW
moths
collected
come
from
a
mixture
of
C3
and
C4
plants
(
20­
80%
from
C4
hosts).
By
late
June,
the
CBW
moths
collected
come
nearly
all
from
C4
sources,
and
remain
constant
until
early
to
mid
August
at
which
point
the
percentage
of
CBW
moths
coming
from
C4
hosts
declines
to
around
20­
50%
by
late
August.
The
percentage
of
moths
from
C4
hosts
then
rose
again
in
the
final
month
of
the
season
to
a
maximum
of
50­
80%.
These
data
demonstrate
that
C4
alternative
hosts
(
primarily
corn
and
sorghum)
are
making
a
significant
contribution
to
the
CBW
adult
population
throughout
the
season.
The
C3/
C4
data
could
not
be
well
correlated
with
predictions
based
on
the
spatial
data
obtained
from
satellite
imagery.
This
was
likely
because
the
scale
of
CBW
adult
movement
exceeds
the
10
mile
range
used
and
that
CBW
may
be
migrating
from
Mexico/
south
Texas
and/
or
originating
from
weedy
hosts.

Based
on
the
CBW
studies,
it
was
concluded
that
both
C3
and
C4
alternative
hosts
serve
as
an
unstructured
refuge
that
exceeds
the
currently
required
5%
external,
unsprayed
structured
non­
Bt
cotton
refuge
option
(
the
smallest
refuge
choice
available).
CBW
moths
are
produced
on
alternative
hosts
in
sufficient
numbers
throughout
the
cotton
growing
season
to
mate
with
any
putative
resistant
CBW
moths
emerging
in
Bollgard
or
Bollgard
II
cotton
fields
and
dilute
resistance.
For
a
full
discussion
of
these
studies
and
the
BPPD's
scientific
conclusions,
refer
to
the
full
review
(
BPPD
2004a).

Summary
of
Monsanto's
Submission
The
data
and
support
materials
submitted
by
Monsanto
in
their
amendment
request
for
Bollgard
II
are
contained
in
three
volumes:

1)
"
Production
of
Tobacco
Budworm
from
Alternate
Host
Plants
and
the
Role
of
These
Host
Plants
as
Natural
Refuge
for
Bollgard
II
Cotton"
(
MRID#
467172­
01)
2)
"
Modeling
the
Impact
of
Natural
Refuge
on
the
Evolution
of
Tobacco
Budworm
and
Cotton
Bollworm
Resistance
to
Bollgard
II
Cotton"
(
MRID#
467172­
02)
3)
"
Scientific
and
Economic
Justification
for
Not
Requiring
Structured
Cotton
Refuges
for
Bollgard
II
Cotton
in
the
U.
S.
Cotton
Belt
from
Texas
to
the
East
Coast"
(
MRID#
467172­
03)

Overall,
the
TBW
studies
and
analyses
follow
the
general
framework
that
was
utilized
for
CBW
­
8­
alternate
host
assessment
(
summarized
in
the
background
section).
The
substance
of
the
TBW
submission
can
be
broken
down
into
six
categories:
1)
pest
biology
of
TBW
as
it
relates
to
IRM
and
refuges;
2)
sampling
and
experiments
conducted
by
Monsanto
to
determine
the
origin
(
host
plants)
of
TBW;
3)
a
spatial
analysis
of
cropping
patterns;
4)
a
calculation
of
TBW
effective
refuge
(
cotton
and
alternate
hosts)
and
the
use
of
resistance
models
incorporating
the
effective
refuge
data;
5)
the
proposed
revision
to
the
Bollgard
II
IRM
plan
and
a
summarization
of
the
scientific
and
economic
rationales
for
the
changes.
Each
separate
category
will
be
reviewed
independently
in
this
document.

In
response
to
comments
from
BPPD,
Monsanto
submitted
supplemental
information
in
two
letters
to
Dennis
Szuhay
(
branch
chief,
Microbial
Pesticides
Branch,
BPPD)
dated
April
3,
April
7,
and
May
10,
2006.
Relevant
information
from
these
letters
is
also
included
in
this
review.

I.
Pest
Biology
of
TBW
Monsanto's
submission
included
a
literature
review
of
TBW
and
CBW
biology
compiled
by
Dr.
John
H.
Benedict
(
Texas
A&
M
University).
The
white
paper
covers
life
history,
utilization
of
host
plants,
dispersal/
migration,
and
genetic
diversity
of
TBW
and
CBW
biology
(
references
are
summarized
in
the
paper).
It
is
noted
that
BPPD
previously
reviewed
the
biology
of
the
major
cotton
pests
in
the
2001
Bt
crops
reassessment
(
see
EPA
2001);
therefore,
only
the
topics
directly
relevant
to
IRM,
Bt
cotton
refuge,
and
the
utilization
of
alternate
host
plants
will
be
discussed
in
this
review.

Regarding
TBW
and
Bt
cotton
refuge,
Monsanto
has
identified
two
primary
biological
topics
of
importance:
1)
the
range
of
host
plants
utilized
by
TBW,
and
2)
adult
movement
(
i.
e.
dispersal
and
migration).
Both
are
important
variables
in
terms
of
determining
whether
alternate
host
plants
(
non­
cotton)
can
provide
adequate
refuge
for
Bollgard
II
cotton.
Assessing
the
TBW
host
range
can
provide
an
indication
of
whether
non­
cotton
hosts
will
be
able
to
produce
sufficient
numbers
of
Bt­
susceptible
adults
to
dilute
any
resistance
genes
arising
from
Bollgard
II
cotton
fields.
Movement
is
critical
for
proximity
issues
(
i.
e.
the
likelihood
that
Bt­
susceptible
TBW
originating
from
non­
cotton
sources
will
be
available
and
proximate
to
mate
with
potentially
resistant
individuals
emerging
from
Bt
cotton
fields).
Both
topics
were
addressed
in
detail
by
Dr.
Benedict
in
Monsanto's
submission
and
are
summarized
below.

a)
Host
Plant
Range
of
TBW
It
is
well
known
from
the
literature
that
TBW
is
a
polyphagous
insect,
utilizing
a
wide
range
of
crop
hosts,
vegetables
and
weeds.
Its
preferred
host
has
been
tobacco,
though
as
an
opportunistic
pest
it
readily
infests
other
host
plants.
The
extent
to
which
TBW
utilizes
plant
hosts
depends
on
a
number
of
factors,
including
agricultural/
cropping
practices
(
i.
e.
landscape
mosaics),
land
use
patterns,
natural
plant
fauna,
climate,
and
seasonal
variability
(
e.
g.
weather,
changing
cropping
patterns,
etc.).
In
addition,
as
a
polyvoltine
insect
(
4­
6
generations
per
growing
season),
TBW
will
utilize
a
succession
of
crop
and
plant
hosts
within
each
growing
season.
­
9­
The
sequence
of
host
utilization
by
TBW
is
dependent
on
geography
and
available
plants.
Early
season
TBW
(
first
generation,
March
to
June)
are
often
found
on
weeds
or
sometimes
on
the
seedlings
of
cultivated
crops.
The
specific
hosts
vary
by
state;
for
example,
clover
serves
as
a
effective
TBW
host
in
the
Mississippi
Delta
region,
while
alfalfa
may
also
play
a
role
in
South
Carolina
and
Georgia.
Second
and
third
generation
TBW
(
occurring
from
June
through
August)
frequently
move
to
cultivated
crop
hosts.
Wild
and
cultivated
tobacco
is
the
preferred
host
for
TBW
(
common
in
the
Carolinas),
though
in
the
Delta
region
cotton
is
thought
to
be
a
primary
host
in
the
absence
of
tobacco.
Other
crops
(
including
soybean,
peanut,
and
tomato)
or
weeds
(
velvetleaf,
deergrass,
and
others)
also
provide
suitable
hosts.
The
overwintering
generations
(
fourth
or
later,
occurring
after
August)
also
utilize
a
range
of
available
hosts
including
tobacco,
cotton,
soybeans,
peanuts,
alfalfa,
and
some
weeds.
Table
1
below
is
a
general
summarization
(
though
not
complete)
of
known
patterns
of
host
utilization
by
region
through
the
growing
season.

Table
1.
TBW
Host
Range
Succession
by
Selected
Regions
(
Created
from
information
in
the
white
paper
"
Biology
and
Dispersal
of
the
Bollworm
and
Tobacco
Budworm
in
North
America"
by
Dr.
John
Benedict;
contained
in
MRID#
467172­
01)

1st
generation
(
March
­
early
June)
2nd/
3rd
generation
(
June
­
August)
4th
+
generation
(
August
­
October)
Carolinas
Wild
hosts:
deergrass,
beggarweed,
morning
glory,
prickly
sida,
bicolor
lespedeza,
toadflax,
Desmodium
spp.,
Ipomoea
spp.
Crops:
tobacco,
cotton,
peanut,
soybean
Wild
hosts:
deergrass
Crops:
regrowth
cotton,
peanut,
soybean,
tobacco
Wild
hosts:
deergrass,
beggarweed,
morning
glory,
prickly
sida,
others
Mississippi
Delta
Wild
hosts:
clover,
velvetleaf,
vetch,
geranium
Crops:
cotton,
soybean,
some
vegetables
Crops:
cotton,
soybean
Wild
hosts:
beggarweed,
burcucumber,
empress
tree,
spider
flower,
prickly
sida,
Royal
Paulownia,
others
Texas
Wild
hosts:
paintbrush,
bluebonnet,
vetch,
wild
tobacco
Crops:
alfalfa
Crops:
cotton,
tomato
Crops:
alfalfa,
tomato
Wild
hosts:
Amantillo,
hillside
verbena,
wild
tobacco
It
is
important
to
note
that
data
on
TBW
host
utilization
throughout
the
growing
season
are
limited,
often
lacking
definitive
quantitative
analysis.
Much
of
the
host
range
information
contained
in
Dr.
Benedict's
report
was
based
on
dated
or
anecdotal
observations.
There
is
likely
to
be
high
variation
to
the
hosts
and
use
patterns
listed
above
and
many
additional
wild
hosts
may
exist.
More
thorough
analyses
are
needed
to
fully
assess
the
extent
to
which
TBW
uses
alternate
and
wild
hosts
for
development.

There
is
also
significant
uncertainty
regarding
the
suitability
of
the
various
plant
host
types
to
­
10­
produce
significant
numbers
of
TBW
adults.
Though
TBW
infests
a
wide
range
of
plants,
some
hosts
are
likely
to
be
better
than
others
in
terms
of
adult
production
since
TBW
population
densities
are
known
to
vary
between
production
areas
and
host
plant
species.
Clearly
both
tobacco
and
cotton
are
productive
hosts
(
and
preferred
by
TBW),
considering
the
management
needs
for
the
pest
on
these
crops.
However,
for
cotton
and
other
cultivated
crops,
insecticide
use
may
play
a
major
role
in
reducing
TBW
production.
In
terms
of
other
plant
hosts,
such
as
noncultivated
plants
and
weeds,
there
is
far
less
information
on
their
contributions
to
the
overall
TBW
population.

Some
comparative
analyses
have
been
preformed
in
some
states
regarding
TBW
adult
production
(
references
summarized
in
the
Benedict
white
paper).
In
North
Carolina,
for
example,
tobacco
was
shown
to
be
by
far
the
dominant
producer
of
TBW
larvae
per
acre,
with
cotton,
peanut
and
soybean
contributing
far
fewer
insects
(
insecticide
use
is
thought
to
curtail
production
in
non­
tobacco
crops).
On
the
other
hand,
in
South
Carolina
(
with
similar
cropping
patterns),
cotton
produced
more
TBW
per
acre
than
tobacco
and
other
cultivated
crops.
A
study
of
wild
host
plants
in
Georgia
and
Florida
found
high
production
of
TBW
on
Japanese
honeysuckle,
clover
and
wild
geranium.
Other
studies
in
Mississippi
found
that
geranium
is
an
important
early
season
producer
of
TBW
(
before
the
second
generation
moves
to
cotton)
and
produces
more
larvae
per
acre
than
clover.
Royal
Paulownia
and
garbanzo
beans
have
also
been
identified
as
potentially
major
producers
of
heliothine
larvae
in
Mississippi.
In
Texas,
cotton
is
thought
to
be
the
greatest
TBW
producer,
though
some
may
are
also
produced
by
tomato,
wild
tobacco,
and
other
weedy
hosts.

There
are
also
some
indications
that
widespread
adoption
of
Bt
cotton
has
had
an
impact
on
TBW
numbers
across
planting
landscapes.
One
of
the
studies
noted
in
the
white
paper
demonstrated
that
in
one
Mississippi
county,
overall
abundance
of
TBW
dropped
(
from
2.2
males/
acre
to
0.77
males/
acre)
over
the
seven
year
period
after
Bollgard
cotton
was
first
commercialized.

b)
TBW
Movement
and
Dispersal
TBW,
like
other
heliothines,
is
capable
of
significant
movement
during
adulthood.
Moth
movement
has
been
characterized
in
three
different
ways:
1)
short
range
movement
(
most
frequent
movement,
usually
300
­
3000
feet,
typically
within
or
near
plant
canopy
for
mating,
oviposition,
feeding,
or
sheltering);
2)
long
range
movement
(
less
frequent,
1
to
10
miles,
occurring
above
the
plant
canopy
for
purposes
of
finding
host
plants,
movements
between
emergence
and
egg
laying
sites);
3)
migratory
movement
(
least
frequent
movement,
wind­
based,
10
to
300
miles,
flights
occurring
to
a
mile
above
the
plant
canopy).
Movement
can
result
from
a
number
of
factors,
including:
1)
loss
of
habitat
(
due
to
crop
and
host
plant
cycles
as
the
growing
season
progresses);
2)
population
density
(
increases
in
density
that
might
challenge
the
limits
of
existing
resources);
3)
environmental
effects
impacting
TBW
survival
or
reproduction.

As
with
TBW
host
plant
utilization,
there
is
still
significant
uncertainty
regarding
movement.
For
example,
as
Dr.
Benedict
points
out,
the
proportion
of
TBW
engaging
in
short
range,
long
­
11­
range,
or
migratory
movement
is
unknown.
However,
some
studies
on
the
three
types
of
movement
(
mainly
using
mark
and
recapture
techniques)
have
been
conducted
and
are
described
in
the
white
paper.

The
mark­
recapture
experiments
certainly
demonstrate
the
abilities
of
TBW
as
strong
fliers,
capable
of
flights
spanning
many
miles.
Such
flights
can
range
over
20
miles,
last
over
5
hours
in
duration,
and
occur
at
altitudes
of
up
to
3,000
feet
at
wind­
aided
speeds
of
30
mph
or
more.
The
typical
TBW
may
not
travel
such
distances,
though,
as
one
study
found
average
lifetime
movement
to
be
6
to
12
miles.
In
fact,
there
is
some
speculation
that
migratory
movements
of
TBW
may
occur
less
frequently
than
with
CBW.
Overall,
TBW
movement
is
greatest
early
in
the
season,
as
they
move
through
the
landscape
from
overwintering
sites
to
available
plant
hosts.
There
is
also
likely
a
significant
early
season
influx
of
TBW
from
southern
sites
in
Mexico
to
the
United
States
based
on
weather
and
air
movement
patterns.
Later
in
the
season
(
towards
fall),
TBW
have
been
observed
to
migrate
southwards
towards
Texas
and
Mexico,
again
based
on
favorable
weather
conditions.

Long
distance
TBW
movement
is
a
critical
component
of
gene
flow
between
populations.
TBW
have
fairly
genetically
homozygous
populations
from
Mexico
through
the
U.
S.,
largely
due
to
migratory
movements
and
interbreeding.
Understanding
gene
flow
is
important
for
IRM,
particularly
how
genes
for
Bt
susceptibility
and
resistance
move
through
populations.
For
example,
with
TBW
pyrethroid
resistance,
a
seasonal
pattern
has
been
observed.
Early
in
the
season,
resistance
is
low
(
a
period
of
high
gene
flow),
while
during
the
cotton
season
(
period
of
low
gene
flow),
resistance
increased.

BPPD
Review
(
Pest
Biology)

Dr.
Benedict's
white
paper
provides
an
excellent
summary
of
the
current
knowledge
base
for
TBW
biology.
However,
the
paper
also
identifies
major
gaps
in
the
understanding
of
how
TBW
utilizes
available
resources
and
moves
through
its
environment.
As
Dr.
Benedict
points
out,
much
of
the
information
is
anecdotal
or
was
derived
from
a
limited
geographical
region
over
a
short
time
period
(
i.
e.
a
"
snapshot"
of
local
TBW
populations).

Despite
the
range
of
information
on
TBW
host
plants
and
potential
production
of
moths,
there
has
been
no
assessment
of
these
hosts
might
function
specifically
as
refuge
(
i.
e.
a
source
of
Bt
susceptible
moths)
for
Bt
cotton.
The
goal
of
a
refuge
in
IRM
is
to
provide
a
source
of
susceptible
insects
to
be
available
to
mate
with
any
resistant
individuals
emerging
from
Bt
fields,
thus
diluting
the
frequency
of
resistance
genes
in
the
population.
There
are
three
major
considerations
for
a
refuge
to
function
properly:
1)
susceptible
insect
production
(
i.
e.
the
numbers
of
moths
produced
by
the
non­
Bt
refuge
relative
to
resistant
survivors
in
the
Bt
field)
 
a
500
(
susceptible)
to
1
(
resistant)
ratio
has
been
used
as
a
target
production
goal
for
a
non­
Bt
refuge
(
as
recommended
by
the
1998
SAP);
2)
proximity
(
i.
e.
the
distance
between
the
refuge
or
source
of
susceptible
insects
and
the
Bt
field)
 
susceptible
insects
must
be
close
enough
(
or
mobile
enough)
to
emerging
resistant
insects
to
have
a
reasonable
probability
of
finding
and
mating
with
them;
3)
synchrony
(
i.
e.
emergence
"
windows"
of
resistant
insects
from
Bt
fields
­
12­
and
susceptible
insects
from
refuge
sources)
 
there
must
be
sufficient
overlap
between
susceptible
and
resistant
emergence
to
ensure
a
high
probability
of
random
mating.
A
structured
refuge
(
i.
e.
a
non­
Bt
portion
of
the
crop)
is
specifically
designed
to
address
these
three
criteria
to
minimize
the
likelihood
of
resistance
development.
If
an
unstructured
refuge
(
e.
g.
alternate
and/
or
natural
plant
hosts)
function
in
a
similar
manner,
then
these
three
topics
should
also
be
considered.

In
terms
of
the
current
knowledge
of
TBW
(
as
provided
by
the
white
paper),
it
appears
that
alternate
plant
hosts
may
provide
sufficient
production
(
criteria
#
1
above)
to
serve
as
a
source
of
Bt
cotton
refuge.
This
follows
from
the
highly
polyphagous
nature
of
TBW
and
wide
range
of
host
plants
utilized
at
different
points
in
the
growing
season.
However,
additional
information
is
needed
to
confirm
this
hypothesis.
For
example,
in
some
regions
(
e.
g.
the
Mississippi
Delta)
TBW
are
"
funneled"
through
one
preferred
host
(
cotton)
with
seemingly
little
use
of
alternate
hosts.
As
for
the
other
refuge
criteria
(
proximity
and
synchrony),
far
less
is
known
for
the
potential
natural
refuge.
Information
on
TBW
movement
is
limited,
particularly
non­
migratory
movement
between
hosts
that
could
quantify
the
availability
of
Bt­
susceptible
TBW
near
Bt
cotton
fields.
Also,
the
timing
of
TBW
development
on
non­
cotton
hosts
compared
with
cotton
hosts
will
have
to
be
addressed
to
ensure
that
there
is
sufficient
overlap.

Monsanto
has
also
considered
the
data
gaps
for
TBW
and
has
designed
experiments
to
address
these
questions.
The
knowledge
gained
from
these
experiments
and
other
information
on
TBW
pest
biology
will
be
discussed
and
integrated
into
the
subsequent
sections
of
this
review
that
describe
the
analysis
of
alternate
hosts
as
potential
refuge.

It
should
also
be
noted
that
one
of
the
more
interesting
studies
noted
by
Dr.
Benedict
was
the
work
from
Monroe
county,
Mississippi
by
Dr.
John
Schneider.
This
study,
conducted
over
seven
years,
observed
landscape
production
of
male
TBW
before
and
after
the
introduction
of
Bollgard
cotton.
After
Bollgard
was
introduced,
TBW
production
dropped
from
2.2
per
acre
to
0.77
per
acre
(
Schneider
2003).
Also,
in
recent
resistance
monitoring
reports
for
Bt
cotton,
smaller
TBW
populations
have
been
sampled,
possibly
due
to
a
suppressive
effect
of
Bt
cotton
(
see
BPPD
2005).
Such
a
suppressive
effect
due
to
large
acreage
of
Bollgard
cotton
could
conceivably
reduce
the
overall
numbers
of
susceptible
TBW
available
(
as
part
of
a
natural
refuge
strategy)
and
should
be
addressed
as
part
of
proposed
IRM
revisions.

II.
Experiments
to
Assess
TBW
Alternate
Hosts
and
Natural
Refuge
In
consideration
of
the
limited
TBW
biological
information
available,
Monsanto
designed
a
large­
scale
experiment
to
investigate
the
potential
of
alternate
host
plants
to
severe
as
refuge
for
Bt
cotton.
The
two
year
experiment,
conducted
in
2004
and
2005,
consisted
of
two
major
components:
1)
a
TBW
sampling
program
in
cotton
growing
regions
with
large
cotton
acreage,
and
2)
bioassays
based
on
detection
of
gossypol
in
sampled
TBW
to
determine
the
plant
host
origin
(
i.
e.
cotton
vs.
non­
cotton).
Gossypol
is
a
natural
toxin
found
specifically
in
cotton
that
is
bound
and
metabolized
in
cotton­
feeding
moths.
Data
obtained
from
this
experiment
were
then
­
13­
used
to
calculate
"
effective
refuge"
from
alternate
hosts
and
subsequently
incorporated
into
IRM
models.
The
results
were
also
compared
with
known
cropping
patterns
near
sampling
sites
using
spatial
analyses.
The
use
and
interpretation
of
the
experimental
data
will
be
described
later
in
this
review.

a)
TBW
Sampling
Monsanto's
TBW
sampling
program
focused
on
seven
cotton
growing
states:
Arkansas,
Georgia,
Louisiana,
Mississippi,
North
Carolina,
Tennessee,
and
eastern
Texas
(
2005
only).
All
of
these
states
are
characterized
by
large
cotton
acreage,
with
at
least
480,000
acres
of
cotton
in
each
state.
Texas
(
5,710,800
acres),
Georgia
(
1,256,050),
Mississippi
(
1,026,500),
and
Arkansas
(
900,200)
are
the
four
largest
cotton
producing
states
in
the
country
(
2004
acreage
figures,
obtained
from
Monsanto's
report).
These
7
states
are
also
well
within
the
known
geographical
range
of
TBW.

For
each
state,
three
to
eighteen
counties
were
selected
each
season
for
TBW
sampling.
The
counties
were
chosen
based
on
density
of
cotton
acreage,
with
each
county
having
15­
24%
of
land
use
devoted
to
cotton
planting.
Bt
cotton
acreage
was
also
high
in
the
sampled
counties,
with
an
average
adoption
of
more
than
70%
Bollgard
or
Bollgard
II
(
based
on
2004
sales
data).
Sampling
sites
in
each
state
are
detailed
in
the
following
table.

Table
2.
TBW
Sampling
by
State
for
Gossypol
Analysis
(
reprinted
from
Table
4
in
Monsanto's
submission,
MRID#
467172­
01).

Sampling
locations
(
by
county)
State
2004
2005
Arkansas
3
(
Drew,
Little
River,
Mississippi)
3
(
Craighead,
Drew,
Mississippi)
Georgia
5
(
Decatur,
Dooly,
Mitchell,
Seminole,
Terrell)
6
(
Appling,
Decatur,
Dooly,
Mitchell,
Terrell,
Tift)
Louisiana
5
(
Bossier,
Catahoula,
Franklin,
Rapides,
Tensas)
5
(
Bossier,
Catahoula,
Franklin,
Rapides,
Tensas)
note
 
TBW
collected
in
only
3
of
the
counties
Mississippi
18
(
Bolivar,
Carroll,
Chickasaw,
Clay,
Coahoma,
Grenada,
Humphreys,
Itawamba,
Lee,
Leflore,
Lowndes,
Madison,
Monroe,
Noxubee,
Prentiss,
Tunica,
Washington,
Yazoo)
16
(
Bolivar,
Carroll,
Chickasaw,
Clay,
Coahoma,
Humphreys,
Itawamba,
Lee,
Leflore,
Lowndes,
Monroe,
Noxubee,
Prentiss,
Tunica,
Washington,
Yazoo)

North
Carolina
3
(
Lenoir,
Pitt,
Wilson)
5
(
Edgecombe,
Halifax,
Lenoir,
Pitt,
Wilson)
Tennessee
10
(
Carroll,
Crockett,
Dyer,
Fayette,
Gibson,
Haywood,
Lake,
Lauderdale,
Madison,
Tipton)
note
 
too
few
TBW
were
collected
to
be
analyzed
during
2004
10
(
Carroll,
Crockett,
Dyer,
Fayette,
Gibson,
Haywood,
Lake,
Lauderdale,
Madison,
Tipton)

Texas
None
4
(
Austin,
Burleson,
Fort
Bend,
San
Patricio)
­
14­
The
sampling
within
each
state
was
conducted
by
pheromone
traps,
which
capture
adult
TBW.
The
traps
were
usually
placed
next
to
cotton
fields
and
were
monitored
weekly
throughout
the
cotton
growing
season
(
April
through
October,
depending
on
the
state).
Within
each
state,
4
to
42
traps
were
deployed
across
the
sampled
counties
(
generally,
states
with
more
sampled
counties
had
a
greater
number
of
traps).
A
total
of
73
traps
were
used
in
2004
(
traps
in
Tennessee
were
not
included
due
to
low
numbers
collected)
and
157
in
2005.
From
these
traps,
17,282
moths
were
collected
in
2004,
of
which
3,385
were
used
in
the
gossypol
analysis.
In
2005,
a
total
of
17,204
adults
were
sampled
including
5,243
that
were
used
in
the
analysis.
Trap
collections
were
high
(
exceeding
50
per
trap)
in
Georgia
and
North
Carolina
in
2004
and
in
all
states
except
Tennessee
in
2005.
Lower
collection
numbers
were
observed
in
Arkansas,
Louisiana,
and
Mississippi
in
2004
and
Tennessee
during
2005.

BPPD
Review
(
TBW
Sampling)

Monsanto's
sampling
strategy
has
addressed
the
major
areas
of
TBW
habitat
as
it
relates
to
cotton
production.
Some
cotton­
growing
states
in
the
southeast,
were
not
sampled,
including
Alabama,
South
Carolina
and
Florida.
However,
these
states
have
either
little
TBW
pest
pressure
(
Alabama)
or
relatively
little
cotton
acreage
(
South
Carolina
and
Florida).
Generally,
the
states
with
large
cotton
acreage
(
e.
g.
Mississippi)
received
greater
numbers
of
sampling
sites
(
counties)
and
traps.
On
the
other
hand,
some
states
with
large
cotton
acreage
(
e.
g.
Georgia)
were
given
fewer
traps
(<
10)
over
the
two
year
sampling
period.
In
addition,
Texas
was
only
sampled
during
the
2005
growing
season.
Monsanto
targeted
major
cotton­
growing
counties
for
sampling
within
each
state,
though
the
submission
did
not
provide
a
specific
breakdown
of
the
sampling
locations
within
each
county
(
a
total
number
of
traps
per
state
was
provided
instead).
From
a
reading
of
the
statistical
analyses
of
the
data
it
appears
that
most
counties
had
one
trapping
location,
although
some
of
the
sampled
counties
(
e.
g.
Drew,
Arkansas)
had
up
to
24
traps.
It
is
unclear
if
the
number
of
traps
used
per
county
was
correlated
to
the
density
of
cotton
acreage
within
those
counties.

Traps
were
placed
in
proximity
to
cotton
fields,
though
the
report
did
not
specify
whether
these
were
Bollgard
fields,
conventional
(
non­
Bt)
cotton
fields,
refuge
fields,
or
a
combination
of
the
three.
If
traps
were
placed
exclusively
next
to
Bollgard
fields,
it
is
possible
that
trap
captures
(
and
the
proportion
of
gossypol
positive
moths)
could
have
been
lower
due
to
the
suppressive
effects
of
high
dose
Bt
cotton.
However,
since
refuges
are
required
to
be
planted
within
½
mile
of
Bt
fields,
it
is
likely
that
susceptible
TBW
produced
by
the
refuges
would
have
been
included
in
the
sampling.
This
could
also
have
created
a
"
worst
case"
scenario
of
a
landscape
with
high
adoption
of
Bt
cotton
in
which
the
only
trapped
gossypol
positive
TBW
originated
from
planted
refuges.
On
the
other
hand,
if
trap
deployment
was
near
conventional
cotton
fields
only
(
with
little
Bt
cotton
nearby),
the
sampling
could
have
produced
larger
numbers
of
gossypol
positive
TBW
that
might
not
have
occurred
with
larger
Bollgard
acreage.

The
submitted
report
provided
only
a
cursory
description
of
the
trapping
techniques,
although
some
supplemental
information
on
the
TBW
trapping
was
provided
by
Monsanto
in
response
to
­
15­
questions
from
BPPD
(
letter
to
Dennis
Szuhay,
dated
April
3,
2006).
All
TBW
were
collected
as
adults
using
traps
baited
with
a
sex
pheromone.
The
lure
used
in
the
traps
is
a
commercial
product,
"
Hercon
Luretrap
Tobacco
Budworm,"
that
is
highly
attractive
and
specific
to
TBW
males
(
no
females
were
sampled
or
tested
for
gossypol).
The
range
of
attraction
of
the
pheromone
lures
used
with
the
traps
was
not
discussed
(
i.
e.
the
ability
of
the
attractant
to
pull
moths
in
from
some
distance
away),
though
pheromone
trapping
is
typically
used
to
assess
pest
populations
on
a
local
scale.
Also,
given
the
proclivity
of
TBW
moths
to
move
in
the
environment
(
see
Pest
Biology
section),
it
may
be
reasonable
to
assume
a
high
degree
of
mobility
in
and
around
cotton
fields
(
a
valuable
resource
for
TBW).
The
area
mapping/
spatial
analysis
project
(
described
later
in
the
review)
also
provided
some
information
on
the
host
plant
fauna
occurring
around
the
trapping
locations.

BPPD
also
notes
that
the
sampling
project
encompassed
only
two
growing
seasons
(
2004
and
2005).
It
is
uncertain
whether
this
is
a
sufficient
sampling
period
to
be
able
to
fully
assess
and
quantify
the
productivity
of
alternate
plant
hosts
over
the
large
regions
of
TBW
habitat
included
in
the
survey.
As
Dr.
Benedict
notes
in
the
provided
white
paper
(
referenced
from
Kennedy
and
Storer
2000),
"
annual
cropping
systems
are
rapidly
changing
mosaics
of
plant
species,
plant
phenologies
and
physical
environments
through
the
year
 
with
no
two
years
being
identical."
Given
the
natural
variability
in
the
natural
and
agronomic
ecosystems,
a
two
year
study
may
present
only
a
"
snapshot"
of
the
host
availability
and
productivity
for
TBW
that
may
not
be
representative
of
future
conditions.
However,
it
is
noted
that
a
similar
sampling
approach
previously
taken
for
CBW
(
described
in
the
Background
section)
was
also
conducted
over
a
two
year
period
and
found
to
be
acceptable.

b)
Gossypol
Bioassays
and
Analysis
(
Methodology)

To
determine
the
origin
(
i.
e.
host
plant)
of
TBW
adults
captured
in
pheromone
traps,
Monsanto
utilized
a
bioassay
to
detect
the
presence
of
gossypol.
Gossypol
is
a
naturally
occurring
toxin
that
is
found
exclusively
in
cotton
plants
that
functions
as
a
plant
protectant
with
insecticidal
activity.
Therefore,
a
bioassay
designed
to
detect
gossypol
in
captured
TBW
should
distinguish
adults
that
arose
on
cotton
from
those
that
originated
on
other
plant
hosts.
Once
ingested
by
cotton­
feeding
TBW,
gossypol
is
metabolized
and
bound.
The
bound
form
can
then
be
extracted
and
detected
using
a
High
Performance
Liquid
Chromatography
(
HPLC)
technique.

Monsanto
employed
a
sequential
sampling
scheme
for
testing
the
collected
moths
based
on
the
numbers
obtained
in
each
trap.
For
trapping
locations
in
which
large
numbers
of
moths
were
collected,
subsets
of
moths
(
sets
of
10
per
trap
sampling
date)
were
analyzed
by
HPLC.
For
smaller
samples,
up
to
100%
of
the
collected
moths
were
tested.
If
the
gossypol
testing
results
from
individual
trap
locations
determined
that
TBW
from
non­
cotton
sources
were
less
than
10%,
additional
moths
were
tested.

The
gossypol
bioassay
system
was
first
tested
with
laboratory­
reared
TBW
that
were
raised
on
host
plants
of
known
origin.
These
hosts
included
cotton,
tobacco,
soybean,
pea,
and
velvetleaf.
Assays
were
run
to
determine
the
level
of
accuracy
with
false
positives
and
negatives.
The
­
16­
calibration
runs
detected
100%
of
moths
that
originated
on
cotton
(
i.
e.
positive
for
gossypol),
while
of
the
TBW
from
non­
cotton
hosts,
81%
were
found
to
be
negative
for
gossypol.
These
results
indicate
that
there
may
be
some
degree
of
false
positives,
which
could
inflate
the
numbers
of
sampled
moths
estimated
to
have
originated
from
cotton.

The
gossypol
HPLC
bioassays
were
conducted
in
96­
well
plates
using
a
number
of
positive
and
negative
controls.
Positive
controls
consisted
of
a
calibration
standard
(
known
gossypol
concentrations)
and
TBW
known
to
have
developed
on
cotton
while
negative
controls
included
blank
wells
(
with
buffer
solution
and
gossypol­
negative
extractions)
and
TBW
from
non­
cotton
hosts.
Criteria
were
established
for
rejecting
test
runs
based
on
abnormal
response
ranges
for
the
control
standards.
Gossypol
was
extracted
by
forming
a
Schiff's
base
with
aniline
to
create
a
dianilino­
gossypol
complex
which
can
be
detected
by
HPLC.

While
the
gossypol
analysis
can
determine
whether
TBW
developed
on
cotton,
it
is
unable
to
differentiate
between
non­
cotton
hosts.
Therefore,
Monsanto
developed
a
separate
analytical
method
to
ascertain
whether
non­
cotton
TBW
developed
on
tobacco,
a
highly
preferred
host
in
some
parts
of
the
insect's
range.
This
test
was
designed
for
use
with
moths
collected
in
North
Carolina,
a
state
with
high
tobacco
acreage
and
large
TBW
numbers.
The
assays
are
based
upon
the
detection
of
cotinine
(
a
nicotine
derivative)
using
gas
chromatography­
mass
spectrometry
(
GC­
MS)
techniques.
Moths
that
test
positive
for
cotinine
can
be
assumed
to
have
developed
on
tobacco
plants
based
on
uptake
of
nicotine
during
feeding.
As
with
the
gossypol
tests,
the
procedures
were
evaluated
for
false
positives
and
negatives,
and
none
were
observed.
The
cotinine
assays
were
conducted
using
the
same
general
procedures
as
the
gossypol
tests
(
i.
e.
the
same
type
of
control
groups)
with
a
subset
of
moths
trapped
at
locations
in
North
Carolina
during
the
2004
growing
season.
Moths
subjected
to
cotinine
testing
could
also
be
subsequently
tested
for
gossypol
(
the
two
processes
did
not
interfere
with
each
other).

BPPD
Review
(
Gossypol
Bioassay
Methodology)

Monsanto
has
provided
an
adequate
description
of
the
techniques
used
for
the
gossypol
and
cotinine
analyses,
though
several
details
were
omitted.
For
example,
details
of
the
gossypol
extraction
were
not
provided
(
e.
g.
how
the
moths
were
handled
and
shipped,
where
the
testing
was
conducted,
and
how
the
extractions
were
accomplished).
Monsanto
also
did
not
supply
any
information
on
the
number
of
test
runs
that
were
rejected
due
to
abnormal
performances
of
the
control
groups.
Such
information
could
give
an
indication
of
the
precision
of
the
testing
techniques,
but
is
not
critical
for
the
overall
analysis.
It
is
also
noted
that
a
more
complete
description
of
the
gossypol
methodology
is
contained
in
a
draft
journal
article
(
Orth
et
al.
in
draft)
that
was
included
in
a
supplemental
submission
(
Monsanto
letter
to
EPA
dated
April
3,
2006).

Since
the
pheromone
traps
used
to
sample
TBW
adults
were
attractive
to
males
only,
no
females
were
included
in
the
gossypol
bioassays.
Therefore,
it
was
not
possible
to
determine
if
response
trends
vary
by
sex.
Since
host
utilization
is
a
function
of
oviposition
(
i.
e.
the
location
the
female
selects
for
egg
laying),
there
would
not
likely
be
variation
between
sexes
in
terms
of
the
­
17­
gossypol
results
had
females
been
included.
On
the
other
hand,
as
described
in
Dr.
Benedict's
white
paper,
female
longevity,
fecundity,
and
reproductive
fitness
can
be
influenced
by
the
quality
of
the
host
plant.
In
addition,
patterns
of
movement
may
be
different
for
females,
particularly
during
and
after
mating.
Prior
to
mating,
females
find
suitable
host
plants
and
release
sex
pheromone
to
which
the
males
respond.
After
mating,
females
are
known
to
retain
spermatophores
throughout
their
lives
and
may
potentially
move
significant
distances
for
egg
laying.
The
trapping
system
employed
by
Monsanto
essentially
mimics
the
male
component
of
TBW
reproductive
behavior,
while
not
fully
accounting
for
the
behavior
of
females.
The
effect
could
be
a
bias
in
the
trap
captures
 
a
scenario
in
which
males
are
present
in
the
trapping
locations
at
greater
numbers
than
females.
This
could
be
an
important
consideration,
since
the
traps
are
designed
to
provide
a
"
snapshot"
of
the
TBW
population
in
and
around
cotton
fields.
If
there
is
a
difference
between
the
numbers
of
males
and
females,
it
is
conceivable
that
the
level
of
random
mating
between
any
resistant
survivors
of
Bt
cotton
and
susceptible
moths
from
alternate
host
plant
refuges
could
be
affected.
BPPD
also
notes
that
the
June,
2004
Scientific
Advisory
Panel
(
SAP)
that
assessed
IRM
considerations
for
Bt
cotton
expressed
concern
regarding
the
use
of
pheromone
traps
that
sampled
only
male
CBW
(
SAP
2004).
The
SAP
stated
that
the
"
traps
do
not
provide
a
meaningful
measurement
of
adult
productivity."
Based
on
these
concerns,
it
is
recommended
that
Monsanto
address
any
potential
bias
in
the
results
that
may
arise
from
sampling
only
male
TBW.

The
cotinine
analysis
is
a
useful
tool
that
should
further
help
characterize
the
host
plant
origins
of
TBW
in
areas
where
tobacco
and
cotton
(
both
are
preferred
hosts)
are
grown
in
proximity.
It
is
noted
that
nicotine
is
also
produced
at
lower
levels
in
other
nightshade
plants
that
may
serve
as
TBW
hosts
including
tomatoes
and
green
peppers.
However,
considering
the
lack
of
false
positives
and
negatives
observed
in
the
preliminary
assays,
it
is
unlikely
that
those
plants
biased
the
results.

c)
Gossypol
Bioassays
and
Analysis
(
Results)

The
results
from
the
2004
and
2005
gossypol
TBW
host
origin
testing
were
analyzed
and
reported
independently,
although
the
data
patterns
were
generally
consistent
in
both
years.
Within
each
growing
season,
Monsanto
presented
the
data
at
the
county
level
(
i.
e.
incorporating
all
trap
locations
within
a
county)
as
well
as
by
state
(
pooled
from
all
sampled
counties).
A
further
break
down
of
the
county
level
data
by
trapping
location
was
provided
in
the
statistical
analyses
(
see
description
later
in
this
section).

Overall,
cotton
and
non­
cotton
host
utilization
in
TBW
was
observed
to
follow
seasonal
trends
in
the
states
tested
during
the
two
year
experimental
period.
Early
in
the
growing
season
(
April
to
June),
TBW
were
found
to
develop
almost
exclusively
on
non­
cotton
hosts
(
i.
e.
close
to
100%
of
tested
TBW
were
negative
for
gossypol).
These
results
were
not
unexpected,
considering
the
early
season
wild
host
usage
that
has
been
previously
observed
with
TBW
(
see
the
Pest
Biology
section).
As
the
growing
seasons
progressed
(
July
and
August),
the
proportion
of
TBW
developing
on
cotton
(
i.
e.
those
testing
positive
for
gossypol)
increased
before
generally
leveling
off
later
in
the
season
(
September).
The
exact
number
of
TBW
developing
on
cotton
was
­
18­
variable,
depending
on
the
collection
region.
In
the
eastern
states
(
Georgia
and
North
Carolina),
production
on
non­
cotton
hosts
was
much
higher
than
in
Delta
states
(
Mississippi,
Louisiana,
Arkansas)
(
see
the
summary
by
state
in
table
3
below).

Table
3.
Gossypol
Bioassay
Results
by
Year
and
State
(
Compiled
from
Figures
2A
­
2E
and
3A
­
3E
in
Monsanto's
submission,
MRID#
467172­
01).

Percentage
of
TBW
originating
on
non­
cotton
hosts
1,2
State
Year
#
TBW
tested
(
total)
April/
May
June3
July
August
Sept/
Oct
Georgia
2004
2005
896
544
­­­
98.0
­
100
(
n<
10)
2
100
81.3
­
100
100
78.3
­
100
100
100
­­­
North
Carolina
2004
2005
684
1392
­­­
84.0
­
88.9
­­­
(
n<
10)
2
98.2
­
100
100
91.6
­
97.2
96.2
­
100
72.9
­
88.5
84.6
­
96.0
Arkansas
2004
2005
399
1216
­­­
25.0
­
93.1
95.9
­
98.0
99.3
50.0
­
100
41.4
­
76.4
20.0
­
92.3
25.7
­
44.8
86.0
22.2
­
25.0
Louisiana
2004
2005
417
245
­­­
(
n<
10)
2
100
100
69.2
­
85.7
38.5
­
90.0
7.0
­
54.8
24.1
­
51.0
46.7
­
63.4
35.0
­
56.0
Mississippi
2004
2005
989
1076
­­­
100
­­­
94.2
­
100
18.8
­
90.9
54.0
­
100
36.4
­
93.5
20.5
­
100
22.2
­
70.9
15.4
­
80.0
Tennessee
2004
2005
0
82
­­­
(
n<
10)
2
­­­
(
n<
10)
2
­­­
(
n<
10)
2
­­­
50.0
­
81.3
­­­
­­­
Texas
2004
2005
NT
688
­­­
93.8
­
100
­­­
18.2
­
95.0
­­­
14.3
­
61.5
­­­
17.6
­
59.7
­­­
24.0
­
63.3
1
Results
are
given
as
a
range
of
the
surveyed
counties
for
each
monthly
trapping
period.
2
Counties
with
sample
sizes
of
less
than
ten
were
not
included
in
the
ranges
(
no
confidence
intervals
were
created
for
samples
of
less
than
ten).
3
Data
from
2004
are
for
May/
June
trapping
period.

The
complete
host
origin
data
for
each
state
and
year
surveyed
(
pooled
by
state
or
for
individual
counties)
are
found
in
figures
1,
2,
and
3
from
Monsanto's
submission
(
MRID#
467172­
01).
Copies
of
these
figures
are
included
in
appendix
1
attached
to
end
of
this
review.

As
can
be
seen
from
the
ranges
included
in
table
3
above
and
in
the
data
summarized
in
Monsanto's
figures,
there
was
frequent
variability
in
the
results
by
state,
county,
year,
and
month
(
county
level
variability
is
reflected
in
the
ranges
for
each
collection
month).
As
previously
discussed,
the
proportion
of
TBW
originating
on
non­
cotton
hosts
was
consistently
high
in
the
eastern
states
of
Georgia
and
North
Carolina.
Non­
cotton
TBW
typically
accounted
for
greater
than
90%
of
those
sampled
throughout
the
growing
season.
These
states
typically
have
large
tobacco
(
a
highly
preferred
TBW
host),
peanut,
and
soybean
acreage
which
may
help
account
for
the
large
non­
cotton
contribution.
In
the
other
surveyed
states,
the
non­
cotton
contribution
was
much
more
variable,
particularly
in
the
mid
to
late
season
when
cotton
may
be
available
as
a
preferential
host.
For
the
Mississippi
Delta
(
Arkansas,
Louisiana,
and
Mississippi),
the
proportion
of
non­
cotton
TBW
is
sampled
counties
was
frequently
below
50%
after
June,
though
was
almost
always
at
least
20%.
In
Tennessee
the
testing
was
confounded
by
low
trap
captures
which
limited
the
number
of
moths
that
could
be
tested.
In
the
one
month
(
August,
2005)
in
­
19­
which
significant
numbers
from
Tennessee
could
be
tested,
there
was
high
variability
among
the
sampled
counties.
Like
Tennessee,
TBW
populations
from
Texas
(
tested
in
2005
only)
also
displayed
wide
and
variable
ranges
of
non­
cotton­
origin
moths,
though
the
overall
non­
cotton
contribution
was
lower
than
in
most
of
the
other
tested
states.
Monsanto
noted
that
the
variability
in
Tennessee,
Texas,
and
the
Delta
states
may
be
due
to
the
lack
of
alternate
host
crops
preferred
by
TBW
for
development.
In
Texas,
the
results
may
have
also
been
affected
by
differing
cotton
phenology
and
TBW
population
dynamics.

The
trends
between
seasons
(
2004
and
2005)
were
generally
similar,
although
there
were
some
notable
differences
(
figure
1
tracks
the
state
level
data
for
each
season).
In
both
years,
the
results
from
Georgia
and
North
Carolina
were
the
most
consistent,
with
average
non­
cotton
contribution
of
90%
or
more
throughout
the
growing
seasons.
Likewise,
in
the
other
states,
early
season
(
April/
May)
non­
cotton
TBW
proportions
were
consistently
high
in
both
years
(>
85%),
with
one
exception
in
Arkansas
during
2005.
Monsanto
suggested
that
there
may
have
been
some
TBW
in
Arkansas
that
developed
on
cotton
the
previously
year
and
then
emerged
locally
from
diapause.
Towards
the
middle
of
the
season
(
July),
the
percentage
of
non­
cotton
TBW
noticeably
dropped
in
both
years.
This
drop
continued
into
August,
before
leveling
off
in
September
and
October,
though
during
2004,
the
non­
cotton
percentages
actually
increased
substantially
in
two
Delta
states
(
Arkansas
and
Louisiana).
This
dramatic
late
season
increase
was
not
observed
during
2005,
however,
as
the
non­
cotton
percentages
remained
relatively
flat
after
August.

In
addition
to
the
gossypol
testing,
Monsanto
conducted
cotinine
testing
on
118
TBW
collected
from
three
counties
in
North
Carolina
(
2004
season
only)
to
distinguish
individual
originating
on
tobacco.
The
moths,
which
were
first
tested
for
gossypol,
were
trapped
during
July
and
August
in
the
midst
of
the
North
Carolina
cotton
season.
Of
the
assayed
moths,
less
than
5%
had
tested
positive
for
gossypol
(
development
on
cotton),
none
of
which
also
tested
positive
for
cotinine.
A
total
of
30%
tested
positive
for
cotinine,
indicating
larval
development
on
tobacco.
The
remainder
(>
65%)
are
likely
to
have
developed
on
plants
other
than
tobacco
or
cotton,
indicating
that
alternate
hosts
play
a
large
role
in
the
production
of
TBW.
Monsanto
suggests
that
this
may
also
be
the
case
in
other
states
without
significant
tobacco
acreage.

d)
Gossypol
Bioassays
and
Analysis
(
Statistical
Analyses)

For
the
statistical
analyses
of
the
gossypol
data,
Monsanto
analyzed
the
data
from
2004
and
2005
independently.
Within
each
year,
the
percentage
of
cotton­
origin
TBW
was
tabulated
for
each
trap
by
collection
date
(
within
each
month).
As
could
be
expected
in
such
large
scale
testing,
there
was
variability
in
the
numbers
of
TBW
captured
in
individual
traps.
Some
traps
collected
hundreds
of
moths
(
of
which
a
subset
was
tested)
while
others
captured
a
single
moth
(
traps
with
no
captures
were
not
included).

To
analyze
the
variation
in
the
trapping
and
gossypol
data,
Monsanto
utilized
the
Fisher's
Exact
Test
(
FET).
This
type
of
analysis
is
typically
done
to
determine
statistical
significance
with
categorical
data
and
small
sample
sizes,
as
is
the
case
with
much
of
the
gossypol
sampling
data.
­
20­
The
FET
analyses
were
used
for
two
purposes:
1)
to
determine
whether
the
gossypol
data
could
be
pooled
by
month
(
i.
e.
combining
all
trap
dates
within
the
month),
and
2)
whether
the
data
could
be
pooled
by
county
(
i.
e.
combining
all
of
the
trap
data
by
month
within
a
county).
For
the
pooling
by
date
determination,
Monsanto
employed
a
"
trap
by
month"
categorization
for
the
data
that
included
all
of
the
collection
dates
within
a
month
for
individual
traps.
The
Fisher's
Exact
Test
could
not
be
run
on
trap
by
month
samples
in
which
there
was
only
one
collection
date
(
degrees
of
freedom
would
equal
zero)
or
the
percent
of
cotton­
origin
moths
was
zero
(
marginal
total
would
equal
zero).
The
county
pooling
determination
involved
"
county
by
month"
figures,
which
included
all
of
the
traps
in
the
county
pooled
together
for
the
month.
As
with
the
date
pooling,
county
by
month
samples
with
only
one
trap
or
0%
cotton­
origin
moths
could
not
be
analyzed
with
a
FET.

2004
Analysis
For
the
2004
data,
there
were
a
total
of
102
trap
by
month
data
sets
that
could
be
analyzed
by
FET.
An
additional
96
trap
by
month
sets
could
not
be
analyzed
by
FET,
due
to
having
only
one
collection
date
or
0%
cotton­
origin
TBW.
Of
the
102
sets
that
were
tested
with
the
FET,
five
were
found
to
have
a
p­
value
of
less
than
the
0.05
level
of
significance,
indicating
that
date
was
a
significant
factor
in
the
gossypol
results.
For
the
other
97
sets,
the
gossypol
results
did
not
significantly
vary
by
date.
It
is
noted
however,
that
there
were
eight
other
trap
by
month
sets
with
p­
values
between
0.05
and
0.10
(
i.
e.
just
below
the
level
of
significance).
Among
the
96
sets
that
were
not
analyzed
by
FET,
there
were
33
that
had
more
than
one
week
of
collection
data
and
0%
cotton­
origin
TBW
(
the
remainder
had
only
one
collection
date).
Since
these
sets
had
no
variance
(
all
tested
TBW
were
negative
for
gossypol)
over
multiple
sampling
dates,
it
can
be
assumed
that
trap
date
was
not
significant
for
these
cases.
The
33
sets
not
tested
by
FEW
were
grouped
with
the
ones
tested
by
FET
to
produce
a
total
of
135
trap
by
month
sets
in
which
there
were
multiple
trap
dates.
Therefore,
in
130
of
these
135
sets,
the
gossypol
results
were
not
shown
to
be
significantly
different
by
trap
date.
Based
on
the
FET
outcomes,
Monsanto
pooled
the
gossypol
data
for
each
trap
by
month
set,
including
the
five
cases
in
which
trap
date
was
found
to
be
a
significant
factor.
The
pooled
data
from
individual
traps
was
then
analyzed
by
FET
at
the
county
level.
For
2004,
there
were
38
county/
month
combinations
in
which
data
were
collected
from
multiple
traps
and
could
be
analyzed
by
FET.
An
additional
49
counties
were
not
tested
with
FET
because
only
one
trap
in
the
county
yielded
data
or
all
traps
had
0%
cottonorigin
moths.
The
FET
analyses
showed
that
three
of
the
38
tested
counties
had
p­
values
below
0.05,
showing
significant
differences
between
individual
traps
in
those
counties.
For
the
other
35
counties,
the
gossypol
results
did
not
significantly
differ
between
individual
traps
within
each
month,
though
three
of
these
counties
had
p­
values
between
0.05
and
0.08
(
close
to
the
level
of
significance).
Based
on
the
FET
analyses,
Monsanto
pooled
the
individual
trap
data
by
county
for
each
month
in
the
experiment
(
including
the
three
counties
with
p­
values
below
0.05).
­
21­
2005
Analysis
The
2005
analysis
was
carried
out
in
much
the
same
manner
as
for
the
2004
data.
In
2005,
there
were
a
total
of
170
trap
by
month
data
sets
that
could
be
analyzed
by
FET,
with
an
additional
304
trap
by
month
sets
could
not
be
analyzed
by
FET,
due
to
having
only
one
collection
date
or
0%
cotton­
origin
TBW.
Of
the
170
cases
analyzed
by
FET,
there
were
nine
data
sets
in
which
date
was
a
significant
factor
(
i.
e.
p­
value
<
0.05).
An
additional
10
other
trap
by
month
sets
with
p­
values
between
0.05
and
0.10
(
i.
e.
just
below
the
level
of
significance).
Of
the
remaining
305
data
sets,
those
with
more
than
one
week
of
data
and
0%
cotton­
origin
TBW
(
107
total)
were
combined
with
the
170
that
were
analyzed
by
FET.
This
created
a
total
of
277
data
sets,
of
which
268
did
not
significantly
vary
by
date.
Based
on
these
results,
the
data
were
pooled
for
each
trap
within
each
month
(
including
the
nine
cases
with
significant
variance).

At
the
county
level,
there
were
57
county/
month
combinations
in
which
data
were
collected
from
multiple
traps
within
the
county
that
could
be
analyzed
by
FET.
The
other
116
sampled
counties
could
not
be
tested
by
FET
because
only
one
trap
in
the
county
yielded
data
or
all
traps
had
0%
cotton­
origin
moths.
For
seven
of
the
57
counties
tested,
there
was
a
significant
difference
in
gossypol
results
between
individual
traps
(
p­
value
<
0.05).
Of
the
counties
that
were
not
analyzed
by
FET,
44
county/
month
sets
had
multiple
traps
all
of
which
trapped
no
cotton­
origin
moths.
These
44
sets
were
included
with
the
57
tested
by
FET,
for
a
total
of
101
data
combinations.
Since
only
seven
of
the
101
sets
differed
significantly,
Monsanto
pooled
the
data
from
individual
traps
for
each
county
and
month.

Based
on
the
pooled
data,
Monsanto
calculated
upper
and
lower
95%
confidence
intervals
for
each
county
and
sampling
month.
Confidence
intervals
were
not
calculated
for
counties
in
which
less
than
10
moths
were
tested
for
gossypol.
The
gossypol
results
and
confidence
intervals
for
each
county
and
month
were
plotted
on
figures
2
(
2004)
and
3
(
2005)
in
Monsanto's
report
(
copies
of
the
figures
are
attached
to
the
end
of
this
review).

An
additional
analysis
was
conducted
with
data
pooled
at
the
state
level
to
determine
any
significant
differences
between
states
during
the
sampling
months.
In
2004,
comparisons
were
made
in
July,
August,
and
September
(
the
other
months
had
too
few
data
points
for
analysis)
using
FET.
The
results
revealed
that
most
of
the
state/
month
comparisons
were
significantly
different,
though
several
"
regional"
state
comparisons
did
not
differ
significantly
at
the
0.05
level.
For
example,
Louisiana
and
Mississippi
were
not
significantly
different
in
both
July
and
September.
For
2005,
FET
analysis
was
not
run
because
of
the
data
structure,
although
95%
confidence
intervals
were
calculated
for
each
state
during
each
sampling
month.

BPPD
Review
(
Gossypol
Bioassay
Results
and
Analyses)

BPPD
agrees
with
Monsanto's
overall
conclusion
that
the
gossypol
bioassay
results
from
the
two
year
sampling
period
clearly
show
that
there
is
a
significant
contribution
of
non­
cotton
hosts
to
­
22­
the
overall
TBW
population.
The
results
followed
a
seasonal
pattern
in
both
years
that
can
be
explained
by
the
availability
of
cultivated
and
non­
cultivated
host
crops
in
each
region
tested.
There
was
variability,
however,
between
different
regions
(
states)
in
terms
of
the
proportion
of
TBW
originating
from
non­
cotton
sources.

In
North
Carolina
and
Georgia,
the
proportion
of
non­
cotton­
origin
TBW
was
consistently
the
highest
among
all
of
the
tested
states.
The
percent
non­
cotton
TBW
for
these
two
states
was
typically
greater
than
90%
(
often
close
to
100%)
throughout
all
of
the
testing
months
in
both
2004
and
2005.
These
results
seemingly
indicate
that
cotton
plays
a
relatively
minor
role
in
the
TBW
population
as
a
whole,
with
alternate
plant
hosts
responsible
for
the
production
of
the
vast
majority
of
TBW.
Both
states
contain
significant
acreage
of
non­
cotton
cultivated
crops
(
including
soybean,
tobacco,
and
peanuts)
that
are
known
to
serve
as
effective
TBW
hosts.
Tobacco,
a
highly
preferred
host,
may
account
for
an
appreciable
percentage
of
the
non­
cottonorigin
TBW,
especially
in
North
Carolina.
The
cotinine
analysis,
conducted
by
Monsanto
to
test
for
a
nicotine
derivative,
found
that
30%
of
the
tested
individuals
(
collected
from
North
Carolina)
were
likely
to
have
developed
on
tobacco.
In
addition
to
non­
cotton
crops,
it
is
likely
that
wild
hosts
(
weeds)
also
play
a
large
role
in
supporting
TBW
populations
in
these
states.

For
the
other
tested
states,
particularly
those
in
the
Delta
region
(
Mississippi,
Louisiana,
and
Arkansas),
the
gossypol
results
were
more
variable,
though
a
general
seasonal
pattern
could
be
discerned.
Like
North
Carolina
and
Georgia,
non­
cotton­
origin
TBW
were
prevalent
early
in
the
season
(
i.
e.
April
­
June).
However,
once
the
cotton
growing
season
was
underway,
the
proportion
of
the
sampled
population
originating
on
cotton
increased,
in
some
cases
dramatically.
The
specific
amount
of
cotton­
origin
TBW
varied
by
state
and
county,
though
non­
cotton­
origin
individuals
usually
were
a
significant
factor
(
at
least
20%
of
the
population)
except
in
a
few
cases.
Some
counties,
such
as
Madison,
MS
(
18.8%
non­
cotton­
origin
TBW
in
July,
2004),
Franklin,
LA
(
7.0%
in
August,
2004),
and
Lee,
MS
(
15.4%
in
Sept./
Oct.,
2005)
had
particularly
low
proportions
of
non­
cotton­
origin
moths.
Alternatively,
in
other
Delta
counties
the
percentage
of
non­
cotton
moths
was
high
(
over
90%)
during
the
same
time
periods.
Regardless
of
the
variability,
these
trends
clearly
show
that
cotton
plays
a
much
bigger
role
as
a
TBW
host
in
the
Delta
states,
probably
due
to
fewer
available
alternate
cultivated
crops
or
wild
hosts
during
the
cotton
season.

In
Texas
and
Tennessee,
the
results
were
also
variable.
Data
obtained
in
Tennessee
were
confounded
by
low
trapping
numbers
and
gossypol
testing
was
conducted
solely
in
2005.
From
the
2005
collections,
only
collections
in
August,
2005
produced
significant
numbers
of
TBW
for
testing,
which
showed
a
fairly
wide
range
of
cotton
and
non­
cotton­
origin
moths.
Likewise
in
Texas,
sampling
was
only
conducted
in
2005.
The
gossypol
results
during
the
cotton
growing
season
(
July
­
October)
revealed
that
the
majority
of
sampled
TBW
originated
from
cotton.
In
three
counties,
the
cotton
contribution
exceeded
80%
of
the
total.
As
with
the
Delta
states,
it
is
likely
that
cotton
is
an
important
host
for
TBW
for
much
of
the
season.

In
relating
the
TBW
sampling
and
gossypol
testing
to
potential
natural
sources
of
(
unstructured)
refuge
for
Bt
cotton,
it
is
important
to
consider
the
three
major
aspects
of
structured
refuge.
­
23­
These
are
1)
production
of
sufficient
numbers
of
susceptible
insects
to
dilute
any
potential
resistance
genes,
2)
proximity
of
the
refuge
to
the
Bt
field
to
ensure
random
mating
between
susceptible
and
resistant
insects,
and
3)
synchrony
of
the
refuge
with
the
Bt
fields
to
ensure
overlapping
emergence
between
susceptible
moths
and
any
resistant
survivors
of
the
Bt
crop.
For
the
most
part,
Monsanto's
experiments
have
addressed
these
three
criteria.

In
terms
of
susceptible
production,
the
goal
of
structured
refuge
for
Bt
crops
has
been
a
ratio
of
500
susceptible
insects
for
every
resistant
insect
that
could
emerge
from
the
Bt
field
(
see
EPA
2001).
With
a
high
ratio
of
susceptible
to
resistant
insects,
any
surviving
resistance
alleles
can
be
effectively
diluted
through
random
mating
to
mitigate
the
potential
development
of
a
widespread
resistant
population.
Monsanto's
TBW
sampling/
gossypol
experiments
did
not
look
at
susceptible
insect
production
from
a
numerical
perspective
per
se;
rather
the
proportion
of
moths
originating
on
cotton/
non­
cotton
sources
was
assessed
(
all
trapped
moths
can
be
assumed
to
be
Bt
susceptible).
The
experiments
clearly
showed
that
a
portion
of
the
TBW
develops
on
cotton,
with
another
portion
developing
on
non­
cotton
crops
and
wild
hosts.
However,
the
numbers
collected
at
individual
traps
were
highly
variable,
with
some
traps
collecting
few
or
no
moths
and
others
trapping
over
one
hundred.
In
some
states
(
e.
g.
Tennessee)
moth
captures
were
erratic
and
frequently
at
numbers
too
low
for
testing.
Low
numbers
of
TBW
have
been
observed
elsewhere
as
well:
recent
Bt
cotton
resistance
monitoring
efforts
have
been
hampered
by
low
collections
of
TBW,
possibly
due
to
a
suppressive
effect
from
widespread
adoption
of
Bt
cotton
(
see
Blanco
2005
and
BPPD
2005).
It
is
noted
that
the
sampling
was
conducted
under
the
current
IRM
strategy
for
Bt
cotton
that
mandates
a
non­
Bt
cotton
structured
refuge
that
should
(
theoretically)
have
supplied
a
source
of
susceptible
moths.
On
the
other
hand,
should
the
IRM
strategy
for
Bollgard
II
cotton
be
revised
to
replace
structured
refuge
(
i.
e.
non­
Bt
cotton)
with
natural
refuge,
the
acreage
of
Bollgard
II
cotton
may
significantly
increase
while
overall
non­
Bt
cotton
acreage
may
decrease.
Given
this
scenario,
it
is
conceivable
that
the
portion
of
TBW
developing
on
cotton
(
as
determined
from
Monsanto's
experiments)
could
be
reduced,
leaving
mainly
the
non­
cotton­
origin
moths.
It
is
not
clear
whether
reducing
the
cotton­
origin
portion
of
the
susceptible
TBW
population
would
have
a
measurable
effect
on
the
ability
to
dilute
emerging
resistance
genes
(
the
modeling
reviewed
in
section
IV
will
more
fully
discuss
resistance
development
scenarios
using
Monsanto's
data).

Both
refuge
proximity
and
developmental
synchrony
criteria
were
directly
addressed
by
Monsanto's
experimental
design.
For
proximity,
the
TBW
traps
were
placed
adjacent
to
cotton
fields
(
both
Bt
and
non­
Bt
cotton).
Though
the
traps
were
baited
with
a
pheromone
attractant
that
could
attract
moths
from
surrounding
areas,
the
sampling
likely
presented
an
accurate
portrayal
of
the
TBW
population
around
the
cotton
field
trap
sites.
Developmental
synchrony
was
also
examined
by
collecting
samples
throughout
the
growing
season.
Data
were
assessed
for
each
month
of
the
cotton
season
and
a
statistical
analysis
showed
that
in
the
majority
of
cases
the
cotton/
non­
cotton
proportions
did
not
significantly
differ
within
the
individual
sampling
months.
At
variable
proportions,
the
results
showed
that
TBW
developing
on
non­
cotton
hosts
overlapped
TBW
from
cotton
during
the
major
part
of
the
cotton
growing
season
(
July­
September)
(
see
figure
1
attached
to
this
review).
These
results
were
not
surprising:
though
TBW
may
develop
on
different
plant
hosts
at
slightly
different
rates,
there
are
four
to
six
generations
per
year
that
­
24­
likely
create
a
natural
overlap
(
discussion
in
Benedict
2004).
Based
on
these
considerations
and
Monsanto's
data,
it
can
be
assumed
that
TBW
from
non­
cotton
hosts
will
develop
in
the
same
time
frames
as
TBW
developing
on
cotton
hosts.

For
the
statistical
analyses,
BPPD
agrees
with
Monsanto's
criteria
for
pooling
data
for
across
dates
for
individual
traps
and
across
traps
for
individual
counties.
The
statistical
tests
(
Fisher's
Exact
Test)
showed
that
in
the
vast
majority
of
cases,
the
gossypol
results
did
not
significantly
differ
by
trap
date
for
single
traps
or
by
trap
within
counties.
Given
the
lack
of
significant
effects
of
date
and
individual
trap,
Monsanto
pooled
all
of
the
data,
including
the
few
cases
that
were
found
to
be
significantly
different.
It
is
unclear
whether
the
pooling
of
the
significantly
different
cases
had
any
bias
on
the
overall
data,
though
given
the
few
examples,
it
is
unlikely
that
there
would
be
much
of
an
effect.

III.
Spatial
Analysis
of
Cropping
Patterns
in
Cotton
Growing
States
To
support
the
gossypol
testing
data,
Monsanto
conducted
an
analysis
of
the
alternate
host
distribution
around
the
trapping
sites.
This
spatial
analysis
was
then
compared
with
the
observed
proportions
of
TBW
originating
from
cotton
and
non­
cotton
hosts
for
various
counties
in
the
sampling
project.
In
addition,
Monsanto
analyzed
USDA/
NASS
cropping
data
for
a
ten
year
period
in
the
sampled
counties
to
further
assess
the
presence
of
alternate
non­
cotton
hosts.

The
cropping
patterns
around
trap
locations
were
determined
by
utilizing
aerial
photographs
that
encompassed
a
one
mile
radius
around
the
traps.
Each
trap
site
was
photographed
at
least
once
during
the
two
year
experimental
period.
From
the
photographs,
various
crop
and
plant
types
were
identified
by
ground­
truthing
and
spatial
analysis
software.
Of
the
known
TBW
crop
hosts,
cotton
(
non­
Bollgard),
soybean,
peanut,
and
tobacco
were
included
in
the
compilation
of
cover
crops
(
wild
hosts
and
weeds
were
not
included).
Within
individual
counties,
the
crop
pattern
data
was
averaged
together
from
multiple
traps
when
necessary.
The
cropping
patterns
tabulated
from
these
photographs
and
the
non­
cotton
TBW
contributions
(
from
the
trapping/
gossypol
data)
were
then
tested
by
regression
analysis
to
check
for
correlations
between
the
plant
hosts
observed
around
traps
and
the
cotton/
non­
cotton­
origins
of
the
trapped
TBW.

The
results
of
the
spatial
analysis/
land
cover
estimates
as
well
as
the
NASS
data
(
2004
season)
from
a
subset
of
the
sampled
counties
were
summarized
in
Table
9
and
10
of
Monsanto's
submission
(
MRID#
467172­
01).
Overall,
non­
Bollgard
cotton
accounted
for
a
sizable
portion
of
the
crop
acreage
in
the
one­
mile
trap
radius,
averaging
9.6%
of
the
cropped
area
across
all
counties.
In
eight
of
the
20
counties
included
in
the
results
summary,
cotton
had
the
greatest
acreage
(
relative
to
peanuts,
soybean,
and
tobacco)
within
the
one­
mile
area.
However,
the
specific
cropping
patterns
varied
by
county
and
region.
The
percentage
of
cotton
was
generally
higher
in
Arkansas
and
Texas,
though
individual
counties
in
Georgia,
Louisiana,
and
North
Carolina
were
shown
to
have
large
cotton
acreage
(>
10%
cotton).
In
Georgia,
peanut
acreage
was
abundant
relative
to
the
other
cataloged
crops,
while
in
North
Carolina,
soybean
was
planted
on
the
largest
portion
of
the
trapping
area.
Tobacco
acreage
was
recorded
only
in
North
Carolina
­
25­
at
low
levels
(<
5%
of
the
total
acreage
in
all
counties
reported).

The
2004
NASS
data,
compiled
at
the
county
level,
generally
mirrored
the
crop
pattern
trends
obtained
from
the
one
mile
trap
radius
photography.
Soybean
acreage
was
significant
at
the
county
level,
with
seven
of
20
counties
having
greater
than
10%
of
the
cropped
area
devoted
to
soybean
production.
Non­
Bt
cotton
was
grown
in
all
of
the
counties
(
except
two),
though
typically
at
lower
levels
than
soybean.
Tobacco
was
found
only
in
North
Carolina
at
low
acreage
(<
5%
of
the
crop
area)
and
peanuts
were
farmed
in
both
Georgia
and
North
Carolina.

As
part
of
the
analysis,
Monsanto
paired
the
gossypol
results
(
i.
e.
the
proportion
on
non­
cottonorigin
TBW)
with
the
crop
pattern
trends
for
individual
counties.
In
general,
counties
with
higher
percentages
of
cotton
(
i.
e.
>
10%)
within
the
one
mile
radius
(
e.
g.
counties
in
Arkansas
and
Texas)
had
higher
proportions
of
cotton­
origin
TBW.
On
the
other
hand,
counties
with
larger
plantings
of
soybean
and
peanuts
(
e.
g.
counties
in
Georgia
and
North
Carolina)
had
greater
proportions
of
non­
cotton­
origin
TBW.
To
test
for
any
statistical
correlations
between
the
gossypol
results
and
the
observed
crop
patterns,
Monsanto
conducted
a
regression
analysis
for
both
the
one
mile
trap
radius
and
county
level
data
(
see
Table
10
in
Monsanto's
submission,
MRID#
467172­
01).
The
analysis
was
conducted
with
the
gossypol
data
from
the
months
of
July,
August,
and
September
(
the
major
portion
of
the
cotton
growing
season).
At
the
trap
level
(
i.
e.
one
mile
radius),
there
was
no
significant
correlation
between
crop
type
and
the
non­
cotton
contribution
to
TBW,
except
for
tobacco
acreage
in
August.
However,
at
the
county
level
there
was
a
positive
correlation
between
the
non­
cotton
contribution
to
TBW
and
tobacco
and
peanut
acreage
in
all
three
months
tested.
This
correlation
is
not
surprising,
since
in
both
Georgia
and
North
Carolina
(
where
the
tobacco
and
peanut
acreage
is
located),
non­
cotton­
origin
TBW
accounted
for
greater
than
90%
of
the
trapped
population.
Cotton
acreage
could
not
be
positively
correlated
with
non­
cotton
TBW
contribution
in
any
of
the
tested
months,
either
at
the
trap
or
county
level.

In
addition
to
the
crop
pattern
analysis
described
above,
Monsanto
also
compiled
USDA/
NASS
data
for
the
seven
sampled
states
over
a
ten
year
period
(
1995­
2004).
These
data
(
presented
in
Figure
4
in
Monsanto's
submission,
MRID#
467172­
01)
included
acreage
figures
for
cotton,
soybean,
peanut,
and
tobacco.
Despite
some
year­
to­
year
fluctuations,
the
acreage
for
each
crop
generally
remained
stable
relative
to
the
other
crops
in
the
state.
For
example,
although
cotton
acreage
dipped
in
Louisiana
during
2002,
soybean
acreage
also
fell
that
season.
However,
in
other
states,
some
shifts
in
cropping
patterns
appear
to
be
visible.
In
North
Carolina,
tobacco
acreage
dropped
over
the
ten
year
period,
while
soybean
acreage
appears
to
have
increased.

Overall,
Monsanto
concluded
that
the
spatial
analysis
data
could
not
fully
account
for
the
variability
in
non­
cotton­
origin
TBW
observed
throughout
the
gossypol
testing
(
i.
e.
the
captures
of
cotton­
origin
TBW
in
non­
cotton
growing
areas
and
vice
versa).
It
was
suggested
that
high
TBW
mobility
may
explain
some
of
the
results;
migratory
TBW
may
have
moved
great
distances
during
the
seasons
to
be
trapped
at
locations
far
from
their
point
of
origin.
However,
Monsanto
indicated
that
stable
cropping
patterns
(
as
evidenced
by
the
ten
year
NASS
data)
should
provide
a
steady
source
of
non­
cotton
refuge
in
the
future.
­
26­
BPPD
Review
(
Spatial
Analysis
of
Cropping
Patterns)

The
spatial
and
cropping
pattern
analysis
by
Monsanto
provides
some
support
for
the
trends
observed
in
the
gossypol
data,
though
the
overall
conclusions
that
can
be
derived
from
these
data
are
somewhat
limited.
For
example,
crop
patterns
observed
in
Georgia
and
North
Carolina
(
large
peanut
and
tobacco
acreage)
seem
to
support
the
large
percentage
of
non­
cotton
TBW
trapped
during
the
experiments.
In
Texas,
relatively
large
amounts
of
(
non­
Bt)
cotton
may
help
explain
the
low
proportion
of
non­
cotton
TBW
observed
in
the
state.
On
the
other
hand,
despite
significant
cotton
acreage
in
some
states
(
Georgia
and
North
Carolina),
few
cotton­
origin
TBW
were
captured.

Monsanto
cataloged
four
crop
types
as
part
of
the
spatial
analysis:
non­
Bt
cotton,
soybean,
peanut,
and
tobacco.
All
four
are
known
to
be
favored
plant
hosts
for
TBW
and
are
cultivated
on
a
large
number
of
acres
in
the
pest's
geographical
range.
Bt
cotton
(
i.
e.
Bollgard
or
Bollgard
II
varieties)
was
not
included,
presumably
because
TBW
are
not
expected
to
survive
the
high
dose
of
Bt
toxin(
s)
expressed
by
the
plants.
However,
by
excluding
Bt
cotton
acreage,
the
overall
amount
(
and
importance)
of
cotton
in
the
landscape
is
reduced.
Wild
hosts
(
i.
e.
weeds)
were
also
not
included,
likely
due
to
the
extreme
difficultly
in
tabulating
such
acreage
(
NASS
data
do
not
address
wild
hosts).
Given
the
probable
high
importance
of
wild
hosts
for
TBW
populations,
the
lack
of
statistical
correlations
seen
in
the
cropping
data
may
be
explained
by
the
lack
of
wild
hosts
in
the
analysis.
Other
cultivated
crops
such
as
vegetables
were
not
included,
possibly
due
to
relatively
small
acreage.

Another
point
to
consider
is
that
all
four
of
the
crop
types
compiled
by
Monsanto
are
typically
managed
agronomically.
This
management
could
include
the
use
of
insecticide
treatments
(
targeted
at
TBW
or
other
pests)
that
could
reduce
TBW
populations.
Since
the
gossypol
experiments
did
not
distinguish
the
type
of
non­
cotton
host,
it
is
not
possible
to
ascertain
how
many
trapped
TBW
originated
from
these
managed
crop
acres
relative
to
wild
hosts.
To
better
understand
the
potential
of
alternate
hosts
to
serve
as
Bt
cotton
refuge,
it
would
be
of
interest
to
know
the
relative
importance
of
each
host
type
(
cultivated
non­
cotton
crops
and
wild/
weedy
hosts)
for
the
TBW
population
Though
the
submitted
report
indicated
that
all
traps
were
photographed
aerially
at
least
once
during
the
two
year
test
period,
only
a
subset
of
the
sampled
counties
was
included
with
the
data.
Monsanto
described
this
set
as
a
"
representative"
sample,
but
no
criteria
for
picking
the
subset
were
provided.
For
Mississippi,
a
state
with
high
cotton
acreage,
data
were
provided
for
only
one
county.
It
is
unclear
how
the
inclusion
of
results
from
all
counties
would
affect
the
overall
analysis.

The
ten
year
NASS
data
included
in
the
report
show
that
cropping
patterns
have
been
relatively
stable
over
the
decade
preceding
the
gossypol
experiment.
However,
it
is
unclear
whether
these
past
trends
can
serve
as
a
predictor
for
future
years.
One
crop
in
particular,
tobacco,
presents
an
interesting
case.
The
acreage
of
tobacco,
a
highly
preferred
TBW
host,
in
North
Carolina
was
­
27­
shown
to
decline
during
the
ten
year
period
captured
in
the
NASS
data.
Though
still
planted
on
a
sizable
area
of
the
state,
it
is
unknown
how
tobacco
will
be
farmed
in
future
years.
It
is
conceivable
that
tobacco
acreage
could
continue
to
decline
due
to
social
and
marketing
reasons.
If
that
scenario
proves
to
be
the
case,
it
will
be
unclear
how
TBW
is
affected.
In
addition,
the
NASS
data
do
not
compile
wild
and
weedy
TBW
hosts.
These
types
of
plant
hosts,
which
are
also
important
TBW
hosts,
can
be
affected
by
non­
agronomic
factors
(
e.
g.
development,
land
use,
etc.),
which
could
also
impact
the
makeup
of
TBW
populations.
As
previously
discussed,
it
is
impossible
to
determine
from
the
submitted
data
the
proportions
each
host
type
(
non­
cotton
crops
or
wild
plants)
contributes
to
the
overall
TBW
population.

IV.
Calculation
of
Effective
Refuge
and
Resistance
Modeling
for
Bt
Cotton
[
Note:
All
figures
and
tables
referenced
are
in
Appendix
2.]

The
effective
refuge
calculations
and
modeling
work
are
contained
in
MRID#
467172­
02
(
hereafter
referred
to
as
Gustafson
&
Head
2005).
Monsanto's
computer
model
used
for
this
study
is
executed
in
Microsoft
®
Excel
2000
(
version
9.9.4402
SR­
1),
and
is
a
modification
of
Caprio's
deterministic,
two­
compartment,
random­
mating,
random­
oviposition
model
used
previously
(
Caprio
1998a).
The
modification
involved
removing
the
assumption
of
constant
effective
refuge
size
in
response
to
the
June
2004'
s
Federal
Insecticide,
Fungicide,
and
Rodenticide
Act
(
FIFRA)
Scientific
Advisory
Panel's
(
SAP)
recommendations
(
SAP
2004).
Instead,
a
regionally­
specific
annual
cycle
of
effective
refuge
size
was
assumed,
according
to
data
collected
in
alternative
host
studies
of
CBW
(
Head
&
Voth
2004)
and
TBW
(
Head
&
Gustafson
2005).
These
data
were
combined
with
Bt
corn
planting
estimates
on
either
the
regional­
scale
for
CBW,
or
county­
scale
for
TBW,
to
construct
effective
refuge
sizes
for
each
of
what
were
conservatively
assumed
to
be
six
annual
generations
for
each
pest.

The
derivations
of
effective
refuge
size
for
both
CBW
and
TBW
are
provided
below.
Monsanto
has
corrected
their
calculation
of
effective
refuge
size
presented
in
Gustafson
&
Head
(
2004)
based
on
the
June
2004
SAP's
recommendations
(
SAP
2004).
The
following
equations
explicitly
account
for
the
lower
production
of
CBW
and
TBW
in
Bt
cotton
where
survival
of
these
insects
is
reduced.
All
figures
and
tables
regarding
effective
refuge
and
modeling
can
be
found
in
Appendix
2
as
well
as
in
Gustafson
&
Head
(
2005).

a)
CBW
effective
and
natural
refuge
calculations
The
following
description
of
the
CBW
effective
and
natural
refuge
calculations
is
taken
directly
from
Gustafson
&
Head
(
2005).

"
As
shown
in
Figure
1,
for
the
purpose
of
estimating
effective
refuge,
both
compartments
of
the
original
two­
compartment
model
were
subdivided
into
different
crop
types.
The
B.
t.
compartment
was
subdivided
into
the
two
B.
t.
crops
grown
commercially
in
the
southern
U.
S.
that
express
the
closely
related
Cry1Ac
and
Cry1Ab
proteins,
B.
t.
cotton
­
28­
and
B.
t.
corn
respectively,
and
the
refuge
compartment
was
subdivided
into
conventional
(
non­
B.
t.)
corn
and
other
C4
hosts,
conventional
(
non­
B.
t.)
cotton,
and
alternative
C3
host
crops
(
other
than
cotton).
The
effective
refuge
is
defined
as
the
proportion
of
the
overall
insect
population
not
exposed
to
the
relevant
B.
t.
protein
or
proteins
(
see
Figure
1).
The
relative
number
of
adult
moths
produced
by
each
of
these
five
sub­
compartments
is
given
by
the
following
equation:

Mij
=
Aij
Eij
LBij
LSij
[
1]

where
M
is
the
number
of
adult
moths
produced
per
unit
area
of
the
region;
A
is
the
proportion
of
the
region
occupied
by
the
crop
type
of
interest;
E
is
the
relative
(
to
cotton,
i.
e.
Ecotton=
1)
number
of
effective
eggs
(
eggs
that
would
produce
adults
in
the
absence
of
B.
t.
or
pyrethroid
sprays)
laid
in
the
crop
type;
LB
is
the
fraction
of
larvae
surviving
in
the
presence
of
the
B.
t.
crop;
LS
is
the
fraction
of
larvae
surviving
a
pyrethroid
insecticide
spray
on
the
crop;
the
subscript
i
refers
to
the
compartment
(
B
for
B.
t.
or
R
for
refuge);
and
the
subscript
j
refers
to
the
particular
crop
type
within
the
compartment
(
1
=
cotton,
2
=
corn,
3
=
other
C3
host
crop).
Definitions
for
all
model
input
parameters
are
given
in
Table
1.

The
fractional
land
areas
occupied
by
the
various
crop
types
were
calculated
as
regional
averages
from
a
combination
of
2004
data
from
Monsanto,
Doane
Agricultural
Services,
and
USDA
NASS
(
2005)
(
see
Table
2).
The
three
regions
are
defined
as
shown
in
Figure
2.
The
regions
include
all
contiguous
counties
in
which
B.
t.
cotton
was
planted
in
2003­
2005,
and
any
adjacent
counties
in
which
one
of
the
following
CBW
host
crops
was
planted
in
2004:
corn,
cotton,
peanuts
or
soybeans.

The
regional
values
for
the
proportion
of
B.
t.
corn
were
taken
from
the
most
recent
two
years
of
market
research
data
(
Doane
Agricultural
Services
2004).
The
Mississippi
regional
value
of
15%
B.
t.
corn
combines
the
totals
for
Arkansas,
Louisiana,
and
Mississippi.
For
Georgia,
the
state
totals
for
Alabama
and
Florida
were
included,
giving
a
regional
value
of
3%
B.
t.
corn.
The
North
Carolina
value
included
South
Carolina
and
Virginia
for
a
total
of
10%
B.
t.
corn.
The
eastern
Texas
value
is
based
on
the
state­
wide
average
of
12%.
Combining
values
from
the
additional
states
for
the
regional
averages
did
not
significantly
change
the
values
for
each
region,
but
it
increases
the
level
of
confidence
in
the
estimated
regional
values
and
is
consistent
with
the
widespread
nature
of
CBW
movement
indicated
by
field
study
results
(
Head
et
al.
in
draft;
Jackson
et
al.
2005b).

As
shown
in
Table
2,
available
market
research
data
(
Doane
Agricultural
Services
2004)
show
that
differing
proportions,
T,
of
B.
t.
and
non­
B.
t.
cotton
were
treated
with
pyrethroids.
When
treatment
occurred,
most
cotton,
regardless
of
whether
it
contained
B.
t.,
was
sprayed
twice.
For
the
purpose
of
estimating
effective
refuge,
it
is
necessary
to
estimate
the
total
proportion
of
adults
emerging
from
these
two
cotton
compartments.
This
estimation
was
made
using
the
following
equation,
which
assumes
there
are
three
­
29­
generations
of
CBW
on
cotton,
with
the
last
two
generations
getting
sprayed
and
a
proportion,
K,
of
larvae
killed
by
each
spray:

LSi1
=
[
Ti1
((
1/
3)
+
(
2(
1­
Ki1)/
3))]
+
(
1
­
Ti1)
[
2]

The
average
mortality
of
larval
CBW
resulting
from
pyrethroid
sprays
on
non­
B.
t.
and
Bollgard
cotton
has
been
reported
as
64.6%
and
82.8%,
respectively
(
Greenplate
2004).
Thus,
LS
for
each
region
and
cotton
type
can
be
obtained
using
equation
[
2]
after
inserting
the
value
for
the
percentage
of
the
crop
that
is
treated
in
the
appropriate
region.
For
the
purposes
of
estimating
the
size
of
the
effective
refuge,
the
fractional
survival
of
CBW
in
B.
t.
cotton
and
B.
t.
corn
were
set
at
15%
and
50%,
respectively
(
Kurtz
et
al.
2004).

The
final
terms
in
equation
[
1]
for
which
estimates
are
required
are
related
to
the
CBW
larval
productivity
of
the
various
crop
types.
Caprio
et
al.
(
2004)
showed
that
there
is
no
statistical
difference
in
the
number
of
CBW
eggs
laid
in
B.
t.
cotton
and
non­
B.
t.
cotton,
thus
EB1
and
ER1
may
be
assumed
equal
to
1.
Similarly,
we
assume
that
the
number
of
eggs
laid
in
B.
t.
corn
and
non­
B.
t.
corn
also
are
identical,
i.
e.
EB2
=
ER2.
Thus,
there
are
only
two
unknown
parameters
remaining:
the
relative
number
of
CBW
eggs
laid
in
corn
and
the
relative
number
laid
in
non­
cotton
C3
crops.

The
relative
larval
productivity
of
other
C3
crops
was
directly
measured
in
a
recent
set
of
field
studies
(
Jackson
et
al.
2005b).
The
relative
larval
productivity
of
corn
(
a
C4
crop)
also
was
determined
in
a
recent
study
using
13C/
12C
ratios
in
captured
CBW
moths
(
Gould
et
al.
2002;
Head
et
al.
in
draft).
These
C4
host
data
are
summarized
in
Table
3.
A
mathematical
expression
can
be
derived
relating
this
observed
C4/
C3
distribution
of
moths
to
the
relative
productivity
of
the
C4
crop,
conservatively
assumed
to
be
only
corn
(
sorghum
is
ignored).
The
fraction
of
observed
moths
coming
from
C4
crops
is:

fC4
=
3
2
1
2
1
2
2
R
R
R
B
B
R
B
M
M
M
M
M
M
M
+
+
+
+
+
[
3]

Inserting
the
definition
for
M
from
equation
[
1]
into
[
3]
and
rearranging
for
the
only
unknown
parameter,
EB2,
(
assumed
equal
to
ER2)
gives
the
following
equation:

EB2
=
ER2
=
(
)
(
)
[
]
2
2
2
4
3
1
1
4
)
1
(
R
B
B
C
R
R
B
C
A
LB
A
f
M
M
M
f
+
 
+
+
[
4]

The
final
remaining
unknown
in
the
calculation
of
M
is
the
relative
CBW
larval
productivity
of
non­
cotton
C3
crops.
Side­
by­
side
strip
trials
were
carried
out
by
Jackson
et
al.
(
2005)
for
several
non­
cotton
C3
crops
in
five
states:
Arkansas,
Georgia,
Louisiana,
Mississippi,
and
North
Carolina.
The
data
for
each
state
are
summarized
in
Table
4.
These
productivity
numbers
use
only
the
data
covering
the
period
when
CBW
larvae
­
30­
were
observed
in
cotton.
A
weighted
overall
average
value
for
ER3
for
each
of
the
three
regions
is
also
shown
in
Table
4.
In
the
case
of
Georgia
and
North
Carolina,
the
weighting
was
based
on
the
acreage
of
peanuts
and
soybeans
in
these
two
regions,
as
shown
in
Table
2.
In
the
case
of
the
Mississippi
region,
the
weighting
was
based
on
the
conservative
assumption
that
any
soybeans
of
maturity
group
6
have
the
same
productivity
as
that
of
maturity
group
5.
The
calculation
also
assumed
that
the
proportion
of
maturity
group
5
(
and
higher)
soybeans
in
the
three
states
was
as
follows:
Arkansas
(
50%),
Louisiana
(
40%)
and
Mississippi
(
20%)
(
Doane
Agricultural
Services
2004).

For
the
eastern
Texas
region,
which
was
not
studied
by
Jackson
et
al.
(
2005b),
the
relative
productivity
of
non­
cotton
C3
crops
conservatively
was
assumed
to
have
the
lower
values
found
in
the
Mississippi
region.

The
effective
refuge,
Reff,
is
defined
as
the
proportion
of
adult
moths
that
would
have
been
produced
in
the
refuge
compartment
in
the
absence
of
any
B.
t.­
induced
larval
mortality:

Reff
=
2
1
3
2
1
3
2
1
B
B
R
R
R
R
R
R
M
M
M
M
M
M
M
M
+
+
+
+
+
+
[
5]

In
the
calculation
of
adult
moths
from
the
two
B.
t.
compartments
in
equation
[
5],
LBB1
and
LBB2
are
both
set
to
1
to
ensure
that
this
calculation
is
made
prior
to
selection
by
B.
t.
crops.
The
assumed
and
fitted
values
used
in
the
calculations
are
shown
in
Table
5.
The
resulting
effective
refuge
sizes
for
each
of
the
six
annual
generations
of
CBW
are
shown
in
Table
6.
The
full
version
of
equation
[
5]
was
required
for
estimating
effective
refuge
only
when
CBW
populations
were
actively
feeding
in
cotton,
which
corresponded
to
generations
3­
5
(
Head
et
al.
in
draft;
Jackson
et
al.
2005b).
According
to
these
same
data,
essentially
all
moths
were
coming
from
C4
hosts
(
conservatively
assumed
to
be
corn)
during
generation
2.
The
C3
host
contribution
to
adult
moths
in
generations
1
and
6
was
assumed
to
be
coming
entirely
from
non­
cotton
crops
in
all
three
regions
based
on
published
literature
(
Benedict
2004).
The
simpler
version
of
equation
[
5]
for
all
of
these
"
non­
cotton"
generations
is
therefore
(
with
the
C3
contribution,
MR3,
vanishing
in
generation
2):

NC
eff
R
=

2
3
2
3
2
B
R
R
R
R
M
M
M
M
M
+
+
+
[
6]

The
natural
refuge
component
of
the
total
effective
refuge
is
found
by
eliminating
the
structured
refuge
from
[
5]
as
follows:

CBW
nat
R
=

2
1
3
2
3
2
B
B
R
R
R
R
M
M
M
M
M
M
+
+
+
+
[
7]
­
31­
For
those
periods
when
cotton
is
not
grown,
the
natural
refuge
is
equivalent
to
the
effective
refuge
and
is
still
defined
by
[
7].

The
modeling
of
CBW
resistance
development
against
Bollgard
cotton
is
described
in
detail
by
Gustafson
et
al.
(
2005)."

b)
TBW
effective
and
natural
refuge
calculations
The
following
description
of
the
TBW
effective
and
natural
refuge
calculations
is
taken
directly
from
Gustafson
&
Head
(
2005).

"
The
pooled,
county­
level
estimates
of
the
percent
cotton­
reared
TBW
moths
were
combined
with
county­
level
landcover
information
to
estimate
the
current
effective
refuge
and
the
natural
refuge
for
each
county
by
month
(
Head
&
Gustafson,
2005).
For
the
purposes
of
this
calculation,
the
following
assumptions
were
made:
 
None
of
the
observed
TBW
moths
came
from
cotton
fields
containing
the
cry1Ac
gene
(
in
2004,
this
corresponded
to
the
Bollgard
and
Bollgard
II
brands
alone
and
combined
with
Roundup­
Ready
®
)
,
therefore
all
observed
cotton­
reared
moths
came
from
non­
B.
t.
cotton;
 
The
county
is
the
appropriate
scale
to
relate
observed
proportions
of
cotton­
reared
TBW
moths
to
the
frequency
of
other
landcover
types;
and
 
The
non­
cotton
areas
produce
TBW
moths
with
a
relative
productivity,
ENC,
that
varies
by
month.

Under
these
assumptions,
the
relative
TBW
productivity
of
non­
cotton
areas
can
be
related
to
other
known
or
measured
factors:

ENC
=

NC
NBTC
NBTC
NBTC
A
A
P
A
 
)
/
(
[
8]

where
ENC
is
the
relative
TBW
productivity
of
non­
cotton
areas
within
a
county
for
a
specific
month;
ANBTC
and
ANC
are
the
relative
areas
of
non­
B.
t.
cotton
and
non­
cotton
landcover
types,
respectively;
and
PNBTC
is
the
observed
proportion
of
cotton­
reared
TBW
moths
for
the
month
of
interest
(
the
subscript
is
a
reminder
that
all
of
these
moths
are
from
non­
B.
t.
cotton).

The
current
effective
refuge
for
TBW
then
is
defined
as
the
proportion
of
TBW
moths
actually
produced
in
the
effective
refuge
compartment
(
prior
to
selection
by
B.
t.
cotton):

®
Roundup
Ready
is
a
registered
trademark
of
Monsanto
Technology
LLC.
­
32­
TBW
eff
R
=
)
(
)
(

NC
NC
NBTC
BTC
NC
NC
NBTC
E
A
A
A
E
A
A
+
+
+
[
9]

in
which
the
only
new
term
on
the
right
hand
side
of
the
equation
is
ABTC,
the
relative
area
of
cotton
in
the
county
containing
the
cry1Ac
gene.

The
potential
natural
refuge
for
TBW
is
given
by
a
similar
equation,
except
that
it
is
now
conservatively
assumed
that
all
current
non­
B.
t.
cotton
acres
would
contain
cry1Ac,
thereby
removing
the
first
term
from
the
numerator
of
[
9]:

TBW
nat
R
=
)
(
NC
NC
NBTC
BTC
NC
NC
E
A
A
A
E
A
+
+
[
10]

The
effective
and
natural
refuge
values
that
result
from
the
use
of
equations
[
8]­[
10]
are
summarized
in
Tables
7
and
8.
For
certain
counties
(
indicated
by
a
subscript
in
Tables
7
and
8),
the
non­
B.
t.
cotton
estimates
were
erroneously
low
(<
5%
of
all
cotton)
due
to
some
combination
of
inaccuracies
in
either
the
sales
figures
or
the
total
cotton
acres
or
both.
These
non­
B.
t.
cotton
values
were
adjusted
to
5%
of
all
cotton
for
the
purpose
of
estimating
current
effective
refuge.
For
cases
where
no
cotton­
reared
moths
were
observed,
a
lower
bound
on
the
productivity
of
non­
cotton
hosts
and
the
resulting
refuge
sizes
was
estimated
by
using
the
upper
95%
confidence
limit
for
the
percentage
of
cottonderived
moths
(
based
on
the
number
of
moths
collected).
However,
as
a
practical
matter,
this
uncertainty
is
of
little
concern
because
it
corresponds
to
cases
in
which
the
natural
refuge
is
relatively
large.

Table
9
contains
the
values
used
for
TBW
modeling,
which
were
the
averages
of
all
available
2004
and
2005
values
for
the
selected
counties
where
the
number
of
moths
collected
during
that
month
was
at
least
10.
Confidence
intervals
for
any
values
associated
with
fewer
than
10
moths
were
deemed
to
be
too
unreliable
for
the
modeling
work."

c)
Description
of
the
Three­
Gene
Model
with
Cross­
Resistance
The
three­
gene
model
for
insect
resistance
evolution
used
in
this
study
is
based
on
a
conceptual
model
similar
to
that
proposed
by
Dow
AgroSciences
(
DAS)
for
its
product,
WideStrike
®
cotton,
and
was
reviewed
by
a
recent
U.
S.
EPA
Scientific
Advisory
Panel
(
SAP)
(
SAP
2004).
However,
the
SAP
questioned
some
of
the
mathematical
details
of
the
DAS
model
and
Monsanto
has
made
some
changes
to
address
the
SAP's
concerns.
The
following
description
of
the
Monsanto
model
is
taken
directly
from
Gustafson
&
Head
(
2005).
­
33­
"
As
shown
in
schematic
form
in
Figure
3,
the
three­
gene
model
is
based
on
the
following
assumptions
concerning
the
mechanism
of
activity
of
the
three
commercial
B.
t.
cotton
products
(
Bollgard,
Bollgard
II,
and
WideStrike
cotton):

 
The
Cry1Ac
toxin,
present
in
all
three
products,
binds
to
two
receptors,
60%
to
receptor
A
and
40%
to
receptor
B.
 
The
Cry1F
toxin,
present
only
in
WideStrike
cotton,
binds
exclusively
to
receptor
A.
 
The
Cry2Ab2
toxin,
present
only
in
Bollgard
II
cotton,
binds
exclusively
to
receptor
C.

We
assume
the
mortality
due
to
each
of
these
toxins
to
be
given
by
a
standard
logistic
dose
response
curve
with
a
slope
factor
of
1:

Srk
=
))
50
/
(
1
(
1
rk
rk
LD
T
+
[
11]

where
Srk
is
the
survival
probability
for
a
pest
that
has
ingested
toxin
k,
which
is
bound
to
receptor
r;
Trk
is
the
concentration
of
toxin­
receptor
complex;
and
LD50k
is
the
toxinreceptor
dose
that
kills
half
the
pest
population.
The
concentration
of
the
toxin­
receptor
complex
is
directly
proportional
to
the
expression
level,
Ck,
of
the
toxin,
but
also
may
be
impacted
by
any
competitive
binding.
In
the
case
of
Cry1Ac,
60%
is
bound
to
receptor
A
and
40%
to
receptor
B.
In
the
particular
case
of
WideStrike
cotton,
there
is
competition
between
Cry1F
and
Cry1Ac
for
receptor
A.
This
competition
is
modeled
through
the
following
equation:

Trk
=

 
 

 
 

 
 

 
 

+
 
)
(
)
(
)
(
1
z
rz
k
rk
z
rz
k
rk
C
f
C
f
C
f
C
f
[
12]

where
frk
is
the
affinity
(
ranging
between
0
and
1)
of
toxin
k
for
receptor
r;
and
the
subscript,
z,
refers
to
a
toxin
that
competes
with
toxin
k
for
receptor
r.
In
this
model,
the
concentration
of
the
toxin­
receptor
complex
is
simply
the
expression
level
for
a
toxin,
such
as
Cry2Ab2,
for
which
there
is
no
competitive
binding
involved.

The
combined
toxic
effect
of
the
toxins
present
in
any
particular
crop
product
is
then
determined
as
the
product
of
all
Srk
that
are
relevant
to
that
crop:

Sp
=
 =

=
+
n
r
m
k
rk
rk
LD
T
,
1
,
1
))
50
/
(
1
(
1
[
13]

For
example,
in
the
case
of
Bollgard
II
cotton,
there
are
three
factors
in
[
13]:
two
for
Cry1Ac
and
one
for
Cry2Ab2.
The
results
of
applying
equations
[
11]­[
13]
to
the
three
­
34­
current
cotton
products
are
shown
in
Table
10
for
CBW
and
Table
11
for
TBW.
Most
of
the
expression
levels
and
intrinsic
toxicities
shown
in
this
table
are
taken
directly
from
published
literature
reports
(
Greenplate
et
al.
2003;
Perlak
et
al.
2001).
The
exceptions
are
the
expression
levels
for
WideStrike
cotton,
which
were
not
published
in
these
reports.
These
were
set
to
give
predicted
overall
survival
probabilities
consistent
with
the
levels
reported
by
DAS
for
WideStrike
cotton
and
its
two
corresponding
one­
gene
precursor
products
(
SAP
2004).

As
shown
in
Tables
10
and
11,
a
conservative
assumption
was
made
with
respect
to
the
relative
intrinsic
toxicities
of
the
two
toxic
complexes
that
are
formed
by
the
binding
of
Cry1Ac;
the
complex
formed
by
binding
to
Receptor
A
(
the
one
shared
with
Cry1F)
was
assumed
to
have
the
same
LD50
as
that
observed
for
Cry1F
alone.
In
sensitivity
analyses
not
included
here,
we
have
found
that
a
model
which
assumes
equivalent
toxicity
for
the
two
bound
forms
of
Cry1Ac
results
in
extremely
long
durability
(>>
30
years)
for
all
three
B.
t.
cotton
products
against
CBW
and
TBW,
regardless
of
market
share
assumptions.
This
is
because
such
a
model
requires
the
co­
occurrence
of
two
rare
alleles
for
a
genotype
that
is
functionally
resistant
to
Cry1Ac
to
exist.

The
survival
probabilities
shown
in
Tables
10
and
11
are
for
the
fully
susceptible,
homozygotic
genotype,
AABBCC.
To
model
resistance,
we
assume
that
each
receptor
has
an
alternate,
fully
resistant
form
whose
structure
is
determined
by
a
single,
diallelic
gene
that
assorts
independently
from
the
other
two
receptors.
We
further
assume
that
expression
in
a
heterozygote
is
additive
such
that
a
heterozygote
has
50%
susceptible
and
50%
resistant
receptors
present
in
its
gut.
The
concentration
of
the
toxin­
receptor
complex
in
such
a
heterozygote
is
then
half
the
level
given
by
[
12].
This
reduced
concentration
is
entered
into
[
11]
to
determine
the
quantitative
impact
on
the
probability
of
survival
due
to
that
particular
toxin­
receptor
combination.

The
net
result
of
applying
this
methodology
to
the
three
current
cotton
products
gives
the
CBW
and
TBW
survival
probabilities
shown
in
Table
12.
These
values
differ
somewhat
from
the
values
recently
reported
by
DAS
for
the
same
three
products
(
SAP
2004),
but
they
are
more
consistent
with
the
relative
activities
of
these
three
products
in
the
field:
Bollgard
II
>
WideStrike
>
Bollgard
(
Jackson
et
al.
2005a)."

d)
Other
modeling
assumptions
Monsanto
has
made
several
other
modeling
assumptions.
These
are
provided
below.

1.
No
fitness
costs.
The
modeling
reported
in
this
study
made
the
conservative
assumption
that
there
is
no
fitness
costs
associated
with
the
resistant
form
of
any
of
the
three
receptors.
This
assumption
is
made
as
a
worse
case
scenario
even
though
there
is
some
empirical
evidence
that
Cry1Ac
resistance
in
common
bollworm
(
a
close
relative
of
CBW),
Helicoverpa
armigera
(
Hubner),
and
the
pink
bollworm,
Pectinophora
­
35­
gossypiella
(
Saunders),
has
a
substantial
associated
fitness
cost.
In
H.
armigera,
Cry1Acresistant
individuals
had
significantly
slower
larval
development
on
non­
Bt
cotton
and
lower
adult
overwintering
survival
than
susceptible
individuals
(
Bird
&
Akhurst
2004).
In
P.
gossypiella,
Cry1Ac­
resistant
individuals
were
found
to
have
lower
overwintering
survival
(
Carrière
et
al.
2001b)
and
less
tolerance
of
the
cotton
defensive
compound
gossypol
(
Carrière
et
al.,
2005;
Carrière
et
al.,
2004)
than
susceptible
individuals,
resulting
in
fitness
being
decreased
by
over
50%
(
Carrière
et
al.
2001b;
Carrière
et
al.
2001b).

2.
Resistance
is
complete.
Resistance
to
each
of
the
three
Cry
proteins
was
assumed
to
be
complete.
This
is
a
conservative
assumption.

3.
Initial
frequency
of
resistant
allele.
The
initial
frequency
of
the
resistant
allele
for
each
of
the
three
receptors,
in
both
CBW
and
TBW,
was
set
equal
to
0.002,
based
on
data
reported
by
Gould
et
al.
(
1997).

4.
Proportion
of
Bt
cotton
is
varied
using
different
refuge
scenarios.
The
proportion
of
Bt
cotton
planted
to
the
three
available
Bt
products
was
varied
in
several
ways,
as
shown
in
Table
13.
As
listed
in
this
table,
there
were
two
assumptions
concerning
refuge,
designated
"
current"
and
"
natural".
These
refuge
scenarios
were
defined
as
follows:

Current
Refuge
Scenario:
The
effective
refuge
(
consisting
of
conventional
cotton
plus
noncotton
hosts)
was
maintained
at
current
levels,
while
the
relative
marketsares
of
the
three
available
Bt
products
were
allowed
to
vary.
The
amounts
of
non­
Bt
and
Bt
cotton
remained
at
current
levels.

Natural
Refuge
Scenario:
The
refuge
was
assumed
to
comprise
the
natural
refuge
plus
a
5%
structured
refuge
requirement
for
Bollgard
and
WideStrike
cotton.
Bollgard
II
cotton
was
assumed
to
have
no
structured
refuge.
The
total
amount
of
Bt
cotton
was
higher
than
currently
present
in
the
landscape,
but
total
cotton
acres
remained
at
current
levels.

5.
Four
regions
for
CBW.
For
CBW,
the
model
was
run
for
each
of
the
four
regions
shown
in
Figure
2.

6.
Worst­
case
county
for
TBW.
For
TBW,
the
model
was
run
for
the
"
worst­
case"
county
in
each
state
(
indicated
by
shading
in
Tables
7
and
8).

7.
Number
of
generations
of
CBW
and
TBW
per
year.
For
both
CBW
and
TBW,
it
was
conservatively
assumed
there
were
six
generations
of
the
pest
each
year.

8.
Resistance
definition.
The
resistance
model
was
run
in
a
simple,
deterministic
manner,
one
region
at
a
time.
The
number
of
years
until
the
frequency
of
each
resistance
allele
exceeded
0.5
was
recorded,
up
to
a
maximum
of
30
years.
The
final
efficacy
of
each
­
36­
product
was
recorded.
If
Bollgard
cotton
was
in
the
scenario
(
all
but
scenario
4),
this
product
was
(
conservatively)
left
in
the
landscape
even
after
its
efficacy
had
been
lost.

e)
Monsanto's
Modeling
Results
and
Discussion
All
results
(
tables
and
figures)
are
found
in
Appendix
2
and
taken
from
Gustafson
&
Head
(
2005).

Cotton
bollworm
Monsanto
has
examined
the
durability
of
each
of
the
three
Bt
cotton
products
that
are
available
commercially.
The
Bt
protein,
Cry1Ac,
is
common
to
all
three
products.
The
presence
of
each
of
these
products
in
the
marketplace
selects
for
potential
resistance
to
Bollgard
cotton,
expressing
only
the
Cry1Ac
protein,
and
also
selects
for
resistance
to
the
other
two
products
through
the
common
selection
for
Cry1Ac
resistance.
The
products
vary
greatly
in
the
rate
at
which
they
select
for
resistance
to
Cry1Ac
because
of
the
presence
of
additional
insecticidal
proteins
in
Bollgard
II,
Cry2Ab2,
and
in
WideStrike,
Cry1F.
Monsanto
examined
the
durability
of
each
of
the
three
Bt
cotton
products
individually
(
Figure
4)
and
together
(
Table
14)
in
the
marketplace.

The
rate
of
selection
for
resistance
or
intrinsic
durability
is
demonstrated
by
looking
at
the
durability
of
each
of
the
products
individually
in
the
marketplace,
as
if
they
were
occupying
the
entire
marketplace
and
with
no
structured
refuge
in
place
(
i.
e.,
all
natural
refuge)
(
see
Figure
4).
Based
on
the
modeling
results
shown
in
Figure
4,
Monsanto
concludes
that
resistance
in
CBW
is
predicted
to
take
more
than
30
years
to
evolve
to
Bollgard
II
cotton
in
three
of
the
four
modeling
regions,
and
more
than
22
years
to
evolve
in
eastern
Texas
reflecting
the
protective
impact
of
two
unique
insecticidal
modes
of
action
for
Cry1Ac
and
Cry2Ab2.
If
all
cotton
is
grown
as
Bollgard
cotton,
CBW
resistance
is
predicted
to
evolve
in
less
than
ten
years
to
Bollgard
cotton
in
Mississippi
and
eastern
Texas,
and
takes
only
12
years
to
evolve
in
Georgia,
reflecting
the
presence
of
only
the
single
insecticidal
mode
of
action.
The
predictions
for
WideStrike
cotton
fall
between
those
for
Bollgard
and
Bollgard
II
cotton
because
of
the
shared
receptor
for
Cry1Ac
and
Cry1F;
CBW
resistance
is
predicted
to
evolve
in
12
years
in
eastern
Texas
and
approximately
20
years
in
Mississippi.

Monsanto
also
modeled
the
impact
of
all
three
products
in
the
marketplace
on
the
selection
for
Bt
resistance
(
i.
e.,
Cry1Ac,
Cry1F,
Cry2Ab2)
using
the
full
three­
gene
model.
This
modeling
effort
is
far
more
complex
than
that
of
modeling
the
selection
for
resistance
of
each
of
these
products
individually.
The
results
for
CBW
with
the
full
model
are
summarized
in
Table
14.
Each
of
the
scenarios
included
in
the
resistance
modeling
are
shown
in
Table
13.

There
were
major
differences
in
the
predicted
durability
of
Bollgard
expressing
only
the
Cry1Ac
and
WideStrike
cotton
product
expressing
the
Cry1Ac
and
Cry1F
proteins
and
Bollgard
II
cotton
which
expresses
both
the
Cry1Ac
and
Cry2Ab2
proteins
due
to
regional
and
marketshare­
related
factors.
Bollgard
II
cotton
retained
its
efficacy
against
CBW
for
more
than
30
years
(
the
maximum
run
length
of
the
model)
in
all,
but
one
of
the
56
cases
modeled
(
25.5
years
when
R
=
­
37­
0.5
and
8%
efficacy
remaining
after
30
years)
in
Scenario
2­
N
(
Table
14)
in
east
Texas).
The
remaining
efficacy
after
30
years
ranged
from
96%
to
99%
in
all
scenarios
except
for
Scenario
2­
N.
Monsanto
states
that
the
greater
durability
of
Bollgard
II
is
due
to
the
higher
mortality
of
individuals
heterozygous
for
Cry1Ac
resistance
in
the
presence
of
Cry2Ab2.

The
presence
of
greater
amounts
Bollgard
II
cotton
(
Scenario
2,
Table
14)
in
the
marketplace
also
increased
the
durability
of
both
Bollgard
and
WideStrike
cotton.
Conversely,
the
presence
of
substantial
amounts
of
Bollgard
cotton
(
Scenario
1,
Table
14)
in
the
marketplace
tended
to
decrease
the
longevity
of
Bollgard
II
and
WideStrike
cotton.
These
results
suggest
that
the
overall
durability
of
Bt
cotton
will
be
enhanced
by
encouraging
increased
adoption
of
Bollgard
II
cotton.

For
the
region
centered
on
North
Carolina,
resistance
always
took
at
least
30
years
to
evolve
for
all
three
products,
even
for
all
of
the
natural
refuge
scenarios.
The
relatively
slow
rate
of
resistance
evolution
in
this
region
reflects
the
high
percentage
of
natural
refuge
(
and
therefore
also
effective
refuge)
present
throughout
the
season
(
Abney
et
al.
2004).
The
predicted
time
to
resistance
was
shorter
for
the
regions
centered
on
Mississippi
and
E.
Texas,
where
the
percentage
of
effective
refuge
was
smaller
than
in
the
regions
centered
on
North
Carolina
and
Georgia
during
the
third,
fourth,
and
fifth
CBW
generations
(
Table
6),
particularly
for
Bollgard
and
WideStrike
cotton.

Monsanto
indicates
that
based
on
the
conservative
assumptions
of
the
model,
a
natural
refuge
is
sufficient
to
delay
the
evolution
of
CBW
resistance
to
Bollgard
II
cotton
for
more
than
25
years
under
all
of
the
scenarios
run
in
the
model.
The
modeling
predictions
for
CBW
resistance
evolution
should
be
regarded
as
highly
conservative,
both
because
of
the
nature
of
the
resistance
assumed
(
including
complete
resistance
and
no
fitness
costs
for
resistance
to
Cry1Ac,
Cry1F,
and
Cry2Ab2)
and
the
values
used
for
alternative
host
contribution
to
CBW
populations
(
Table
6).
The
estimated
contributions
of
non­
cotton
alternative
hosts
used
in
the
model
were
derived
from
Head
et
al.
(
in
draft)
and
Jackson
et
al.
(
2005b).
Those
studies
focused
only
on
a
subset
of
CBW
hosts.

Monsanto
performed
additional
sampling
of
CBW
populations
in
Arkansas
and
east
Texas
in
July
through
September
of
2005,
followed
by
analysis
of
these
moths
for
gossypol
using
methods
described
in
Head
&
Gustafson
(
2005).
These
data
were
not
included
in
the
current
modeling
simulations.
In
these
subsequent
studies,
the
non­
cotton
contribution
to
CBW
populations
in
these
regions
was
many
times
higher
than
predicted
by
the
earlier
studies
presented
in
Head
et
al.
(
in
draft)
and
Jackson
et
al.
(
2005b).
For
Arkansas,
only
3.2%
of
993
moths
analyzed
came
from
cotton
(
tested
positive
for
gossypol),
while
4.7%
of
409
moths
collected
in
east
Texas
were
derived
from
cotton.
The
effective
and
natural
refuge
sizes
assumed
in
the
modeling
were
highly
conservative
because
>
40%
of
the
moths
from
Arkansas
and
>
70%
of
the
moths
from
Texas
were
assumed
to
be
from
cotton.
Overall,
these
results
suggest
that
there
are
greater
effective
and
natural
refuge
sizes
for
CBW
moths
in
the
E.
Texas
region
and
Arkansas
(
Mississippi
Region)
than
was
assumed
in
the
modeling.
­
38­
Tobacco
budworm
Monsanto
examined
the
durability
of
each
of
the
three
Bt
cotton
products
individually
and
together
in
the
marketplace.
The
intrinsic
durability
of
all
three
Bt
cotton
products
is
much
greater
for
TBW
than
CBW
because
of
the
"
high­
dose"
of
Cry1Ac
for
TBW
expressed
in
all
three
products
(
see
EPA
2001;
SAP
2004).
Based
on
the
modeling
simulations,
all
three
products
are
predicted
to
retain
their
efficacy
for
more
than
30
years
even
if
all
cotton
in
a
region
is
planted
to
that
product
and
no
structured
refuge
is
required.
The
only
exceptions
occur
for
Bollgard
cotton;
for
Mississippi
and
Tennessee.
In
these
cases,
Bollgard
cotton
is
predicted
to
lose
some
efficacy
within
the
30
years
modeled
[
Figure
5].

With
the
full
three­
gene
model,
the
durability
of
Bollgard
and
WideStrike
cotton
is
increased
in
cases
where
Bollgard
II
cotton
also
is
present.
In
all
cases,
the
three
products
were
predicted
to
retain
their
efficacy
against
TBW
for
more
than
30
years
in
all
cases
modeled.
This
is
a
direct
result
of
the
extremely
high
efficacy
of
Cry1Ac
against
this
pest,
and
the
fact
that
Cry1Ac
is
present
in
all
three
Bt
cotton
products.
In
the
state
with
the
lowest
nature
refuge
for
TBW
(
Mississippi,
Table
13),
resistance
to
Cry1Ac
and
Cry1F
evolved
after
21
years
in
Scenario
1­
N
if
the
structured
refuge
requirements
for
Bollgard
and
WideStrike
cotton
also
were
removed
[
Table
15].

BPPD
Review
(
Effective
Refuge
Calculations
and
Modeling
Efforts
for
CBW
and
TBW)

[
Note:
All
figures
and
tables
referenced
are
in
Appendix
2.]

CBW
In
2004,
BPPD
did
an
assessment
of
Monsanto's
analysis
of
the
impact
of
effective
refuge
size
and
typical
insecticide
use
practices
on
predicting
of
years
to
resistance
of
CBW
(
and
TBW)
to
Bollgard
cotton
(
BPPD
2004b).
In
this
analysis,
BPPD
agreed
with
Monsanto's
conclusion
that
the
model
output
is
very
sensitive
to
effective
refuge
size
and
use
of
insecticide
sprays
on
Bollgard
cotton
for
CBW
control.
With
this
understanding,
the
5%
external,
unsprayed
structured
refuge
option
was
considered
adequately
protective
to
delay
TBW
and
CBW
resistance
if
effective
refuge
size
and
typical
use
practices
(
i.
e.
pyrethroid
oversprays
of
Bollgard
fields)
were
included
as
parameters
in
the
model
(
BPPD
2004b;
SAP
2004).
BPPD
concluded
that
the
Gustafson
et
al.
(
2004)
model
was
limited
because
it
did
not
consider
the
spatial
and
temporal
dynamics
of
CBW
alternate
hosts
by
generation.
BPPD
also
disagreed
with
Monsanto's
calculation
of
estimated
effective
refuge
size.
Estimated
effective
refuge
was
calculated
as
the
sum
of
the
total
acres
of
conventional
cotton,
soybean,
and
other
alternate
hosts
(
assumed
to
be
10%)
per
county.
Rather
the
estimation
of
effective
refuge
size
must
be
a
weighted
average
on
each
host
for
each
generation.
The
June
2004
SAP
agreed
with
the
Agency's
analysis
regarding
the
calculation
of
the
effective
refuge
size,
the
spatial
limitations
of
the
Gustafson
et
al.
(
2004)
model
and
the
need
to
incorporate
more
explicitly
the
alternate
host
data
(
SAP
2004).
The
SAP
also
noted
that
"
the
estimation
of
the
total
refuge
proportion
also
requires
an
estimate
of
the
emergence
from
transgenic
crops
(
e.
g.
the
proportion
of
refuge
is
relative
to
the
total
number
of
adults,
including
those
emerging
from
transgenics)."
They
further
­
39­
indicated
that
the
estimate
of
total
adult
emergence
should
be
corrected
for
losses
due
to
selection
for
resistance.

In
Monsanto's
current
modeling
study,
alternative
hosts
are
more
explicitly
examined
in
the
model,
i.
e.,
five
compartments:
Bt
cotton,
Bt
corn,
non­
Bt
cotton,
non­
Bt
corn,
and
non­
cotton
C3
crops
vs.
the
original
two­
compartment
model
(
Gustafson
&
Head
2004)
This
is
shown
in
Figure
1.
Spatiality
is
now
included
in
the
model;
there
are
four
cotton
production
regions
that
differ
in
the
amount
of
alternate
hosts:
Mississippi,
Georgia,
North
Carolina,
E.
Texas.
The
modeling
analysis
provides
a
regional
snapshot
of
the
population
dynamics
of
CBW
and
thus
is
a
limited
picture
of
what
might
be
happening
in
the
landscape
(
although
these
snapshots
may
represent
the
worst
case
scenario).
Making
an
extrapolation
over
many
years
based
on
just
a
few
snapshots
in
time,
given
the
variability
of
the
data,
is
uncertain.

The
effective
refuge
size
is
defined
as
the
proportion
of
the
overall
insect
population
not
exposed
to
the
relevant
Bt
protein
or
proteins.
Natural
refuge
is
considered
to
be
the
proportion
of
adults
on
non­
Bt
corn
and
non­
cotton
C3
crops.
The
relative
number
of
adult
moths
produced
in
each
of
the
five
sub­
compartments
per
unit
area
is
given
by
Equation
1
above.
In
this
equation,
corrections
are
made
for
losses
due
to
selection
for
Bt
and
pyrethroids
for
the
appropriate
subcompartments
CBW
larval
productivity
is
calculated
on
the
various
crop
types
using
a
combination
of
direct
field
data
for
non­
cotton
C3
crops
(
Jackson
et
al.
2005b),
C13/
C12
isotope
analysis
data
for
the
productivity
on
corn
(
Gould
et
al.
2002;
Head
et
al.
in
draft),
egg
production
on
Bt
and
non­
Bt
cotton
(
Caprio
et
al.
2004
indicate
no
difference),
and
egg
production
on
Bt
cotton
and
Bt
corn
is
assumed
to
be
equal.
The
fractional
land
areas
occupied
by
the
various
crop
types
were
calculated
as
regional
averages
from
a
combination
of
2004
data
from
Monsanto,
Doane
Agricultural
Services,
and
USDA/
NASS.
The
average
mortality
of
larval
CBW
resulting
from
pyrethroid
sprays
on
non­
Bt
and
Bollgard
cotton
has
been
reported
as
64.6%
and
82.8%,
respectively
(
Greenplate
2004).
It
was
recommended
by
both
BPPD
(
BPPD
2004b)
and
the
SAP
(
SAP
2004)
that
pyrethroid
mortality
be
incorporated
into
the
model.
The
fractional
survival
of
CBW
in
Bt
cotton
and
Bt
corn
were
set
at
15%
and
50%,
respectively
(
Kurtz
et
al.
2004).

In
Gustafson
&
Head
(
2005),
Monsanto
addressed
BPPD's
and
the
SAP's
previous
concerns
regarding
the
calculation
of
the
effective
refuge
size
(
weighted
per
host
per
generation),
inclusion
of
alternate
hosts
(
five
sub­
compartments
vs.
two­
compartments),
increased
spatiality
(
four
regions),
impact
of
multiple
Bt
cotton
products
(
three
gene
model
and
scenarios
varying
the
level
of
adoption
of
each
product)
and
larval
mortality
caused
by
pyrethroid
sprays
and
B
.
t.
(
no
corrections
were
made
in
the
original
model).
This
means
that
Monsanto's
current
CBW
modeling
analysis
(
Gustafson
&
Head
2005)
is
more
robust
than
the
one
previously
reviewed
by
the
Agency
(
BPPD
2004b)
and
the
SAP
(
2004).

There
are
limitations
to
the
carbon
isotope
analysis.
There
is
uncertainty
as
to
the
relative
noncotton
C3
crop
contribution
vs.
the
C4
crop
contribution.
Non­
cotton
C3
crop
cannot
be
further
subdivided
because
of
the
limitations
of
carbon
C13/
C12
isotope
analysis
data.
This
was
­
40­
previously
noted
in
the
BPPD
review
(
BPPD
2004a)
and
by
the
SAP
(
2004)
of
the
CBW
alternate
host
data
(
Head
&
Voth
2004).

There
is
agreement,
however,
that
the
C4
crop
contribution
(
primarily
corn)
is
quite
high
throughout
most
of
the
cotton
growing
season
and
this
contribution
can
be
much
greater
than
that
of
cotton.
There
is
large
variability
in
the
number
of
moth
captures
in
the
pheromone
traps
due
to
the
impact
of
immigration
and
reverse
migration,
among
other
factors
(
Gould
et
al.
2002).
There
are
also
limitations
in
extrapolating
data
based
on
just
male
moth
captures
to
be
representative
of
the
entire
population
(
see
previous
discussion
in
Section
II
of
this
review
and
in
SAP
2004).

The
June
2004
SAP
concluded
that
"
without
more
definitive
data
quantifying
temporal
and
spatial
production
of
susceptible
CBW
moths
from
each
of
the
C3
and
C4
hosts,
and
confirmed
mating
behavior
of
subsequent
adults,
the
current
refuge
requirement(
s)
should
continue."
It
should
be
noted
that
this
comment
was
made
in
the
context
of
the
question
the
Agency
posed
to
the
SAP
on
the
relative
contribution
of
the
C3
and
C4
hosts
as
unstructured
(
natural)
refuge.
The
Agency
had
to
make
a
decision
as
to
whether
to
continue
the
5%
external,
unsprayed
refuge
requirement
for
Bollgard
cotton
based
on
the
scientific
assessment
of
the
CBW
alternative
host
data
as
effective
refuge.
No
field
data
was
collected
for
the
eastern
Texas
region
in
which
to
estimate
the
relative
productivity
of
CBW
on
non­
cotton
C3
crops.
For
this
region,
the
relative
productivity
on
non­
cotton
C3
crops
was
assumed
to
lower
than
in
Mississippi.
In
Georgia
and
North
Carolina,
the
non­
cotton
C3
crop
production
is
assumed
to
be
a
weighting
of
the
acreage
of
peanuts
and
soybeans
in
these
two
regions
and
in
the
Mississippi
region,
the
soybeans
of
maturity
group
6
are
assumed,
conservatively,
to
have
the
same
productivity
as
that
of
maturity
group
5.
The
effective
refuge
calculations
are
affected
by
the
uncertainties
in
the
quantity
and
spatial
production
of
susceptible
CBW
for
each
alternate.
On
a
qualitative
basis,
the
estimation
of
current
and
natural
effective
refuge
is
probably
adequate.

BPPD
agrees
with
the
following
assumptions
Monsanto
made
for
the
three­
gene
model
based
on
the
mechanisms
of
activity
of
Bollgard,
Bollgard
II,
and
WideStrike
cotton
(
i.
e.
expected
receptor
binding
patterns
of
Cry1Ac,
Cry1F,
and
Cry2Ab2
to
CBW
midgut
membranes).
It
is
assumed
that
the
primary
mechanism
of
Bt
resistance
is
associated
with
modifications
to
the
receptor
binding
site.
Some
level
of
cross­
resistance
between
Cry1Ac
and
Cry1F
associated
with
a
common
resistance
mechanism
of
reduction
in
receptor
binding
would
be
expected.
This
issue
is
of
particular
concern
for
WideStrike
cotton
because
it
expresses
both
Cry1F
and
Cry1Ac
and
there
is
competition
between
Cry1F
and
Cry1Ac
for
receptor
A.
These
general
assumptions
were
evaluated
by
the
June
2004
SAP
for
a
spatially­
explicit,
stochastic
CBW
resistance
model
used
by
DAS
for
its
WideStrike
cotton
product
and
found
to
be
acceptable
(
SAP
2004).

 
The
Cry1Ac
toxin,
present
in
all
three
products,
binds
to
two
receptors,
60%
to
receptor
A
and
40%
to
receptor
B.
 
The
Cry1F
toxin,
present
only
in
WideStrike
cotton,
binds
exclusively
to
receptor
A.
 
The
Cry2Ab2
toxin,
present
only
in
Bollgard
II
cotton,
binds
exclusively
to
receptor
C.
­
41­
Resistance
associated
with
modification
of
the
binding
site
receptor
has
been
the
primary
Bt
resistance
mechanism
reported
to
date
(
reviewed
in
Ferré
&
Van
Rie
2002).
Other
Bt
resistance
mechanisms
have
been
reported
that
are
based
on
alterations
in
the
proteases
that
cleave
the
protoxin,
processing
it
into
a
smaller
active
toxin
(
Candas
et
al.
2003)
and
most
recently,
the
discovery
that
esterases
can
bind
and
detoxify
Bt
toxins
(
Gunning
et
al.
2005).
Monsanto's
assumptions
regarding
the
binding
of
Cry1Ac,
Cry1F,
and
Cry2Ab2
to
receptors
A,
B,
and
C
are
acceptable
given
the
current
understanding
of
the
receptor
binding
patterns
of
Cry1Ac,
Cry1F,
and
Cry2Ab2
to
CBW
midgut
membranes.

The
June
2004
SAP
raised
an
issue
as
to
how
DAS
calculated
the
expected
mortality
of
different
genotypes
as
a
result
of
exposure
to
Cry1Ac,
the
factor
"
Z"
(
SAP
2004).
Monsanto
has
modified
the
mathematical
expression
of
the
competition
between
Cry1F
and
Cry1Ac
for
receptor
A
and
the
calculation
of
the
expected
mortality
of
different
genotypes
as
a
result
of
exposure
to
Cry1Ac.
The
survival
probabilities
of
all
27
CBW
genotypes
are
found
in
Tables
10.
To
model
for
resistance,
Monsanto
assumed
that
each
receptor
has
an
alternate,
fully
resistant
form
whose
structure
is
determined
by
a
single,
diallelic
expression
in
a
heterozygote
is
additive
such
that
a
heterozygote
has
50%
susceptible
and
50%
resistant
receptors
present
in
the
gut.
The
concentration
of
the
toxin­
receptor
complex
is
such
a
heterozygote
is
then
half
(
given
by
Equation
12)
and
then
this
concentration
is
entered
in
Equation
11
to
determine
the
quantitative
impact
on
the
probability
of
survival
due
to
that
particular
toxin­
receptor
combination.
BPPD
agrees
with
Monsanto's
analysis
of
the
expected
mortality
of
the
different
CBW
genotypes
as
a
result
of
exposure
to
Cry1Ac,
Cry1F,
and
Cry2Ab2
(
see
Table
12).
These
values
differ
from
those
reported
by
DAS
for
the
same
three
products
(
SAP
2004)
because
of
differences
in
the
mathematical
calculations.

BPPD
agrees
with
the
following
other
modeling
assumptions:
 
There
are
no
fitness
costs
associated
with
the
resistant
forms
of
any
of
the
three
receptors.
As
Monsanto
notes,
there
are
recent
studies
in
which
Cry1Ac
resistance
is
associated
with
fitness
costs
(
e.
g.
Bird
&
Akhurst
2004;
Carrière
et
al.
2001a,
b).
 
Resistance
to
each
of
the
three
Cry
proteins
was
assumed
to
be
complete.
 
The
CBW
model
was
run
for
each
of
the
four
regions
shown
in
Figure
2.
This
provides
a
snapshot
in
time
in
each
region.
 
The
proportion
of
Bt
cotton
was
varied
for
each
of
the
three
Bt
cotton
products
across
the
four
regions
by
"
current"
and
"
natural"
refuge.
 
For
CBW
there
were
six
generations
of
the
pest
per
year.
 
The
number
of
years
until
the
frequency
of
each
resistance
allele
exceeded
0.5
was
recorded,
up
to
a
maximum
of
30
years.
The
number
of
years
to
resistance
and
the
relative
efficacy
remaining
should
only
be
interpreted
on
a
qualitative
basis.
 
The
model
was
run
in
a
simple,
deterministic
manner,
one
region
at
a
time.
Spatiallyexplicit
stochastic
models
would,
however,
offer
a
more
nuanced
and
dynamic
picture
of
the
interactions
of
the
three
Bt
cotton
products
in
the
landscape
over
time.

BPPD
has
some
concerns
with
the
following
assumptions:
­
42­
 
Assumption
that
the
initial
resistance
allele
frequency
for
each
of
the
three
receptors
is
0.002.
Given
that
resistance
management
is
more
effective
when
the
initial
frequency
of
resistance
is
lower,
an
assumption
of
0.002
should
be
a
conservative
assumption.
Given
the
length
of
prior
use
of
Bollgard
cotton,
10
years,
this
estimation
may
or
may
not
be
optimistic.
Monsanto's
modeling
does
not
account
for
any
prior
selection
to
Cry1Ac.
Monitoring
efforts
have
not
detected
any
significant
change
in
susceptibilities
to
the
Bt
proteins
(
BPPD
2005;
Tabashnik
et
al.
2003).
Given
the
limitations
of
the
sampling,
probability
of
detection
based
on
the
sensitivity
of
the
discriminating
dose
(
LC99)
bioassay
(
0.01),
the
current
resistance
allele
frequency
for
Cry1Ac
is
estimated
to
be
lower
than
0.01,
based
onn
the
limit
of
detection
of
the
discriminating
dose
(
LC99)
bioassay
(
EPA
2001).
The
probability
of
detection
of
field
resistance,
therefore,
would
be
relatively
low
unless
resistance
allele
frequencies
were
high
and
the
resistance
was
dominant.
Given
minimal
adoption
of
WideStrike
and
Bollgard
II
and
therefore
minimal
selection
additional
selection
pressure
for
resistance
to
Cry1Ac,
Cry1F,
or
Cry2Ab,
it
is
likely
that
the
initial
resistance
allele
frequency
of
0.002
is
still
a
conservative
assumption.
Because
the
modeling
simulations
assume
that
all
three
Bt
products
are
introduced
as
a
mosaic,
then
an
initial
resistance
allele
frequency
of
0.002
for
Cry1Ac
would
be
a
conservative
assumption.
However,
as
noted
above,
given
the
adoption
of
Bollgard
cotton
and
the
selection
pressure
for
CBW
resistance
to
Cry1Ac,
it
is
more
likely
that
the
resistance
allele
frequencies
have
increased
and
that
a
more
relevant
set
of
modeling
scenarios
would
have
examined
prior
selection
(
see
discussion
below).
 
Assumption
that
resistance
is
recessive.
In
North
Carolina,
Burd
et
al.
(
2001)
provide
results
from
a
reciprocal
cross
study
that
suggests
CBW
resistance
to
Cry1Ac
and
Cry2Ab
is
dominantly
or
incompletely
inherited.
These
results
suggest
that
the
assumption
that
CBW
resistance
is
recessive
is
perhaps
incorrect.
Burd
et
al.
(
2001)
isolated
nonrecessive
Bt
resistance
genes
present
in
CBW
field
populations
in
North
Carolina
and
estimated
the
Cry1Ac
resistance
allele
frequency
for
CBW
as
0.0043
and
the
Cry2Aa
resistance
allele
frequency
for
CBW
as
0.00039.
 
Assumption
that
there
is
no
prior
selection
for
resistance
to
Cry1Ac.
Monsanto
has
presumed
no
prior
selection
for
CBW
resistance
to
Cry1Ac
even
though
Bollgard
has
been
commercially
available
for
10
years.
Monsanto
has
modeled
a
mosaic
of
all
three
Bt
cotton
products
rather
than
a
sequential
introduction
of
Bollgard
II
and
WideStrike
following
introduction
of
Bollgard.
Zhao
et
al.
(
2005)
have
shown
that
the
concurrent
use
of
one­
and
two­
gene
Bt
broccoli
plants
resulted
in
control
failure
of
both
types
of
Bt
plants
for
control
of
Plutella
xylostella
(
diamondback
moth)
rather
than
the
use
of
the
two­
gene
plants
by
themselves.
In
addition,
these
researchers
illustrated
through
their
modeling
efforts
that
resistance
will
evolve
first
to
the
toxin
that
is
being
used
singly.

Given
the
assumptions
of
the
three­
gene
model
and
its
limitations,
there
is
likely
enough
effective
natural
refuge
to
be
sufficient
to
delay
the
evolution
of
resistance
to
Bollgard
II
cotton
for
more
than
25
years
(
not
a
precise
number
of
years)
under
all
plausible
scenarios
in
all
four
regions
(
Table
14).
Based
both
on
the
intrinsic
durability
of
each
of
the
three
Bt
cotton
products
(
Figure
4)
and
the
three­
gene
modeling
analyses
for
all
three
Bt
cotton
products
together
in
the
marketplace
(
Table
14),
Bollgard
II
retained
the
highest
level
of
efficacy
against
CBW
in
all
­
43­
scenarios
(
all
regions).
This
is
because
of
the
relatively
high
mortality
of
individuals
heterozygous
to
Cry1Ac
resistance
in
the
presence
of
Cry2A2,
as
compared
to
WideStrike;
intermediate
in
many
scenarios
because
of
the
shared
binding
receptor
between
Cry1F
and
CryAc
and
likelihood
of
cross­
resistance,
and
Bollgard,
weakest
in
all
scenarios,
no
high
dose
for
CBW
and
a
single­
gene
product.
Monsanto's
models
predict
that
CBW
resistance
to
Bollgard
cotton
will
evolve
in
less
than
the
30
year
horizon
in
the
Georgia,
Mississippi,
and
E.
Texas
regions
in
most
scenarios
except
for
2­
C
(
Bollgard
=
0.1;
Bollgard
II
=
0.8;
WideStrike
=
0.1).
Resistance
always
took
at
least
30
years
to
evolve
for
all
three
Bt
cotton
products
in
the
North
Carolina
region
in
all
scenarios,
even
the
natural
refuge
scenarios.
Monsanto's
modeling
does
not
presume
any
prior
selection
for
Cry1Ac
resistance.

When
Bollgard
cotton
acreage
is
minimized,
Bollgard
II
and
WideStrike
longevity
is
maximized
(
Table
14).
Selection
pressure
for
resistance
to
Cry1Ac
is
reduced
and
the
relative
value
of
the
unique
proteins,
Cry1F
and
Cry2Ab2
is
enhanced.
Large
amounts
of
Bollgard
II
cotton
in
the
marketplace
increased
the
durability
of
both
Bollgard
and
WideStrike
(
Table
14).

TBW
In
Gustafson
and
Head
(
2005),
Monsanto
addressed
BPPD's
(
2004b)
and
the
SAP's
(
2004)
previous
concerns
regarding
the
calculation
of
the
effective
refuge
size
(
weighted
per
host
per
generation),
inclusion
of
alternate
hosts
(
three
sub­
compartments
for
TBW
vs.
twocompartments
increased
spatiality
(
four
regions),
impact
of
multiple
Bt
cotton
products
(
three
gene
model
and
scenarios
varying
the
level
of
adoption
of
each
product)
and
larval
mortality.
This
means
that
Monsanto's
current
TBW
modeling
analysis
(
Gustafson
&
Head
2005)
is
more
robust
than
the
one
previously
reviewed
by
the
Agency
(
BPPD
2004b)
and
the
SAP
(
2004).
As
noted
above,
the
modeling
analysis
provides
a
regional
snapshot
of
the
population
dynamics
of
TBW
and
thus
is
a
limited
picture
of
what
might
be
happening
in
the
landscape
(
although
these
snapshots
may
represent
the
worst
case
scenario).
Making
an
extrapolation
over
many
years
based
on
just
a
few
snapshots
in
time,
given
the
variability
of
the
data,
is
uncertain.

Monsanto
combined
pooled,
county­
level
estimates
of
the
percent
cotton­
reared
TBW
moths
with
county­
level
landcover
information
to
estimate
the
current
effective
refuge
and
the
natural
refuge
for
each
county
by
month
(
see
Head
&
Gustafson
2005,
discussed
in
Section
II
and
III
above).
The
landscape
is
divided
into
three
sub­
compartments
to
estimate
TBW
current
and
natural
effective
refuge:
Bt
cotton,
non­
Bt
cotton,
non­
cotton
area
(
Figure
1).
The
current
effective
refuge
for
TBW
is
defined
as
the
proportion
of
TBW
moths
actually
produced
in
the
effective
refuge
compartment
prior
to
selection
by
Bt
cotton
(
Equation
9).
The
potential
natural
refuge
for
TBW
is
given
by
Equation
10;
it
is
conservatively
assumed
that
all
current
non­
Bt
cotton
acres
would
contain
cry1Ac
so
the
term,
ANBTC
,
is
removed.
Table
9
contains
the
current
and
natural
effective
refuge
values
used
in
the
TBW
modeling.
These
were
the
averages
of
all
available
2004
and
2005
values
for
the
selected
counties
where
the
number
of
moths
collected
during
that
month
were
at
least
10.
BPPD
agrees
with
Monsanto's
method
for
calculating
the
relative
TBW
productivity
on
each
prospective
alternate
host.
However,
uncertainties
in
the
pheromone
sampling,
gossypol
analyses,
and
spatial
landscape
mapping
affect
the
estimation
of
­
44­
the
current
and
natural
effective
refuge.
These
issues
were
discussed
in
Sections
II
and
III
of
the
review.

BPPD
agrees
with
the
following
assumptions
Monsanto
made
for
the
three­
gene
model
based
on
the
mechanisms
of
activity
of
Bollgard,
Bollgard
II,
and
WideStrike
cotton
(
i.
e.
expected
receptor
binding
patterns
of
Cry1Ac,
Cry1F,
and
Cry2Ab2
to
TBW
midgut
membranes).
It
is
assumed
that
the
primary
mechanism
of
Bt
resistance
is
associated
with
modifications
to
the
receptor
binding
site.

 
The
Cry1Ac
toxin,
present
in
all
three
products,
binds
to
two
receptors,
60%
to
receptor
A
and
40%
to
receptor
B.
 
The
Cry1F
toxin,
present
only
in
WideStrike
cotton,
binds
exclusively
to
receptor
A.
 
The
Cry2Ab2
toxin,
present
only
in
Bollgard
II
cotton,
binds
exclusively
to
receptor
C.

As
noted
earlier,
these
general
assumptions
were
evaluated
by
the
June
2004
SAP
for
a
spatiallyexplicit
stochastic
CBW
resistance
model
used
by
DAS
for
its
WideStrike
cotton
product
and
found
to
be
acceptable
(
SAP
2004).
See
discussion
above
regarding
Bt
resistance
mechanisms.

BPPD
has
some
concern
about
whether
these
assumptions
are
still
valid
for
TBW.
BPPD
reviewed
the
relevant
literature
and
determined
that
three­
site
model
proposed
by
Jurat­
Fuentes
&
Adang
(
2001)
for
binding
of
Bt
Cry1
toxins
to
sites
in
the
TBW
midgut
membrane
is
relevant
to
this
discussion.
Based
on
the
ligand­
blot
analyses,
the
Cry1Ac
toxin
was
able
to
recognize
three
receptors:
A,
B,
and
C;
while
the
Cry1Fa
toxin
was
able
to
recognize
receptor
A
and
perhaps
a
second
predicted,
but
not
determined
site.
Receptor
C
is
only
recognized
by
the
Cry1Ac
toxin.
Neither
Cry1Ea
nor
Cry2a
competed
with
any
of
the
labeled
Cry1A
toxins.
Cry1Ac
competed
with
high
affinity
for
Cry1Ab
binding
sites.
Cry1Aa,
Cry1Fa,
and
Cry1Ja
also
competed
for
Cry1Ab
binding
sites.
Cry1Ab
competed
for
labeled
Cry1Ac
binding
sites
with
a
higher
affinity
than
did
Cry1Aa,
Cry1Fa,
and
Cry1Ja.
It
was
noted
by
the
study
authors
that
it
was
inherently
difficult
to
label
Cry1Fa.
One
can
conclude
from
Fuentes
and
Adang
(
2001)
that
Cry1F
and
Cry1Ac
have
shared
binding
affinity
for
Receptor
B
(
in
their
model)
which
is
analogous
to
what
was
proposed
by
Monsanto
in
its
three­
gene
model
(
competition
is
for
receptor
Ai
in
this
model)
(
Gustafson
&
Head
2005).
Some
level
of
cross­
resistance
between
Cry1Ac
and
Cry1F
associated
with
a
common
resistance
mechanism
of
reduction
in
receptor
binding
would
be
expected.
This
issue
is
of
particular
concern
for
WideStrike
cotton
because
it
expresses
both
Cry1F
and
Cry1Ac
and
there
is
competition
between
Cry1F
and
Cry1Ac
for
a
shared
receptor.
Competition
binding
experiments
indicate
that
perhaps
40%
of
the
labeled­
Cry1Ac
binding
to
brush
border
membrane
vesicles
was
reduced
by
competition
with
Cry1Fa,
although
a
saturation
concentration
of
Cry1Fa
was
not
reached.
Based
on
the
data
provided
in
Jurat­
Fuentes
&
Adang
(
2001),
Monsanto's
assumptions
regarding
the
binding
patterns
of
Cry1Ac,
Cry1F,
and
Cry2Ab2
to
receptors
A,
B,
and
C
on
TBW
midgut
membranes
are
acceptable
with
one
caveat.
Additional
competition
binding
studies
using
TBW
brush
border
membrane
vesicles
at
saturable
concentrations
of
Cry1Fa
are
recommended.
Based
on
the
results
of
additional
competition
binding
studies,
the
survival
probabilities
of
all
27
genotypes
may
need
to
be
recalculated
and
the
modeling
redone.
Uncertainty
in
the
shared
binding
affinity
­
45­
of
the
Cry1Ac
protein
to
receptor
A
and
B
(
in
Monsanto's
model)
would
impact
the
modeling
predictions
for
all
three
Bt
cotton
products
because
the
Cry1Ac
protein
is
common
to
all
of
them.
Monsanto's
modeling
does
not
assume
any
prior
selection
for
Cry1Ac
resistance.
The
degree
to
which
the
modeling
predictions
would
be
impacted
is
unknown
The
June
2004
SAP
raised
an
issue
as
to
how
DAS
calculated
the
expected
mortality
of
different
genotypes
as
a
result
of
exposure
to
Cry1Ac,
the
factor
"
Z"
(
SAP
2004).
Monsanto
has
modified
the
mathematical
expression
of
the
competition
between
Cry1F
and
Cry1Ac
for
receptor
A
and
the
calculation
of
the
expected
mortality
of
different
genotypes
as
a
result
of
exposure
to
Cry1Ac.
The
survival
probabilities
of
all
27
TBW
genotypes
are
found
in
Tables
11.
To
model
for
resistance,
Monsanto
assumed
that
each
receptor
has
an
alternate,
fully
resistant
form
whose
structure
is
determined
by
a
single,
diallelic
expression
in
a
heterozygote
is
additive
such
that
a
heterozygote
has
50%
susceptible
and
50%
resistant
receptors
present
in
the
gut.
The
concentration
of
the
toxin­
receptor
complex
is
such
a
heterozygote
is
then
half
(
given
by
Equation
12)
and
then
this
concentration
is
entered
in
Equation
11
to
determine
the
quantitative
impact
on
the
probability
of
survival
due
to
that
particular
toxin­
receptor
combination.
These
values
differ
from
those
reported
by
DAS
for
the
same
three
products
(
SAP
2004)
because
of
differences
in
the
mathematical
calculations.

BPPD
agrees
with
the
following
other
modeling
assumptions:
 
There
are
no
fitness
costs
associated
with
the
resistant
forms
of
any
of
the
three
receptors.
As
Monsanto
notes,
there
are
recent
studies
in
which
Cry1Ac
resistance
is
associated
with
fitness
costs
(
e.
g.
Bird
&
Akhurst
2004;
Carrière
et
al.,
2001a,
b).
 
Resistance
to
each
of
the
three
Cry
proteins
was
assumed
to
be
complete.
 
The
proportion
of
Bt
cotton
was
varied
for
each
of
the
three
Bt
cotton
products
across
the
four
regions
by
"
current"
and
"
natural"
refuge.
 
For
TBW,
the
model
was
run
for
the
"
worst­
case"
county
in
each
state
(
indicated
by
shading
in
Tables
7
and
8).
 
For
TBW
were
six
generations
of
the
pest
per
year.
 
The
number
of
years
until
the
frequency
of
each
resistance
allele
exceeded
0.5
was
recorded,
up
to
a
maximum
of
30
years.
The
number
of
years
to
resistance
and
the
relative
efficacy
remaining
should
only
be
interpreted
on
a
qualitative
basis.
 
The
model
was
run
in
a
simple,
deterministic
manner,
one
region
at
a
time.
Spatiallyexplicit
stochastic
models,
however,
would
offer
a
more
nuanced
and
dynamic
picture
of
the
interactions
of
the
three
Bt
cotton
products
in
the
landscape
over
time.

BPPD
has
additional
concerns
regarding
the
assumptions
for:
initial
resistance
allele
frequency
for
each
of
the
three
receptors
and
prior
selection
for
Cry1Ac
resistance.
These
were
discussed
above
in
the
context
of
CBW
modeling
but,
in
general,
are
applicable
also
to
the
TBW
modeling
with
the
addition
of
the
following
comments.
 
Assumption
that
the
initial
resistance
allele
frequency
for
each
of
the
three
receptors
is
0.002.
Gould
et
al.
(
1997)
estimated
the
TBW
resistance
allele
frequency
as
0.001.
Given
the
length
of
prior
use
of
Bollgard
cotton,
10
years,
this
estimation
may
or
may
not
­
46­
be
optimistic.
Monitoring
efforts
have
not
detected
any
significant
change
in
susceptibilities
to
the
Bt
proteins
(
BPPD
2005;
Tabashnik
et
al.
2003).
As
noted
above,
given
the
limitations
of
sampling
and
sensitivity
of
detection
(
0.01),
the
probability
of
detection
of
field
resistance,
therefore,
would
be
relatively
low
unless
resistance
allele
frequencies
were
high
and
the
resistance
was
dominant.
 
Assumption
that
resistance
is
recessive.
Because
Cry1Ac,
Cry2Ab2,
and
Cry1F
are
all
expressed
at
a
"
high
dose"
in
all
three
Bt
cotton
products,
it
is
expected
that
resistance
will
be
conferred
by
rare,
recessive
alleles
and
most
resistant
adults
from
Bt
crops
will
mate
with
susceptible
adults
from
refuges.
In
a
recent
review,
Tabashnik
et
al.
(
2003)
comment
that
" 
inheritance
of
resistance
to
transgenic
plants
with
high
concentrations
of
Bt
toxins
is
recessive
in
all
cases
examined
to
date
involving
pests
that
have
high
inherent
susceptibility
to
the
toxins
produced ".
 
Assumption
of
no
prior
selection
to
Cry1Ac.
See
earlier
BPPD
comments
on
CBW
modeling.

From
a
qualitative
perspective,
modeling
indicates
the
intrinsic
durability
of
Bollgard
II
is
greater
than
WideStrike
which
is
greater
than
Bollgard.
The
intrinsic
durability
of
all
three
Bt
cotton
products
is
much
greater
for
TBW
than
for
CBW
because
of
the
"
high
dose"
of
Cry1Ac
for
TBW
expressed
in
all
three
products.
In
virtually
all
cases,
all
three
products
retained
their
efficacy
(
i.
e.
no
resistance)
for
more
than
30
years
(
maximum
time
for
the
simulation)
even
if
all
cotton
in
a
region
is
planted
to
that
product
and
no
structured
refuge
is
required
(
i.
e.
all
natural
refuge)
(
see
Figure
5).
The
only
exceptions
occur
for
Bollgard
cotton
in
Tennessee
and
Mississippi.
Given
the
assumptions
of
the
three­
gene
model
and
its
limitations,
there
is
likely
enough
effective
natural
refuge
to
be
sufficient
to
delay
the
evolution
of
resistance
to
Bollgard
II
cotton
for
more
than
30
years
(
i.
e.
the
time
horizon
of
the
model,
not
to
be
interpreted
as
a
precise
number
of
years)
under
all
plausible
scenarios
in
all
four
regions
(
see
Table
15).
This
is
due
to
the
extremely
high
efficacy
of
Cry1Ac
against
TBW,
and
the
fact
that
Cry1Ac
is
present
in
all
three
Bt
cotton
products.
In
the
state
with
the
lowest
natural
refuge
for
TBW,
Mississippi
(
see
Table
13),
resistance
to
Cry1Ac
and
Cry1F
evolved
after
21
years
in
scenario
1­
N
if
the
structured
refuge
requirements
for
Bollgard
and
WideStrike
cotton
were
removed.
Uncertainties
in
the
pheromone
captures,
gossypol
analyses,
spatial
analysis,
effective
refuge
calculation,
degree
of
shared
binding
affinity
of
Cry1Ac
to
receptor
A
and
B,
effect
of
prior
selection
for
Cry1Ac
resistance,
and
other
modeling
assumptions
affect
the
precision
and
accuracy
of
the
modeling
predictions.

Caprio
(
2006)
used
a
spatially­
explicit,
stochastic,
two­
locus
resistance
model
to
investigate
the
likelihood
that
TBW
would
not
evolve
resistance
to
Bollgard
II
within
a
15­
year
time
horizon.
He
used
two
different
scenarios,
one
with
a
4%
structured
refuge
and
1%
wild
hosts
(
5%
untreated
refuge),
and
the
second
with
no
structured
refuge
and
a
1%
wild
host
refuge.
The
model
simulated
1024
cotton
patches,
each
ca.
100
acres
in
size.
The
model
assumed
that
only
Bollgard
cotton
was
used
initially
and
simulated
a
gradual
transition
to
Bollgard
II
over
five
years.
Mortality
rates
for
this
cotton
with
nine
different
genotypes
were
calculated.
The
initial
gene
frequency
was
assumed
to
be
0.001.
The
model
assumes
four
generations
of
TBW
develop
on
cotton
per
year.
Generations
outside
of
cotton
were
not
incorporated
into
the
model.
A
­
47­
carrying
capacity
of
80,000
larvae
(
eggs/
patch)
for
early
pre­
bloom
cotton,
800,000
for
peak
bloom
cotton,
and
500,000
larvae
(
eggs/
field)
for
flowering
cotton
were
used
in
the
modeling.
Fitness
costs
were
not
included
in
the
model.
The
model
parameters
were
not
varied.
Thirty
runs
of
the
simulation
for
each
of
the
two
scenarios
were
made
and
changes
in
resistance
allele
frequencies
at
both
resistance
loci
were
evaluated.
In
30
runs
of
the
simulation
with
a
structured
5%
untreated
refuge,
resistance
was
not
observed
at
either
locus.
The
mean
rate
of
increase
in
resistance
allele
frequency
for
the
Cry2A2
locus
was
1.032
(
SD
=
0.151),
while
the
mean
rate
of
increase
for
the
Cry1Ac
locus
was
1.0567
(
SD=
0.169).
In
30
runs
of
the
simulation
with
no
structured
refuge
and
a
1%
unstructured
refuge,
resistance
was
observed
to
occur
in
three
cases
to
the
Cry1Ac
trait
in
Bollgard
II.
The
mean
rate
of
increase
in
resistance
allele
frequency
for
the
Cry2Ab2
locus
was
0.8984
(
SD
=
0.391),
while
the
mean
rate
of
increase
for
the
Cry1Ac
locus
was
85.88
(
SD=
262.53;
three
outlier
populations).
These
results
suggest
that
the
risk
of
TBW
resistance
in
a
15­
year
time
horizon
with
or
without
a
structured
refuge
is
relatively
small
if
the
estimates
of
mortality
of
TBW
larvae
and
other
components
of
the
model
are
reasonably
accurate.
The
results
do
suggest
much
higher
risk
of
TBW
resistance
to
the
Cry1Ac
trait
if
the
structured
refuge
is
removed.
The
predictions
of
Caprio's
spatially­
explicit,
stochastic,
twolocus
TBW
resistance
model
are
similar
to
those
predicted
by
Monsanto's
three­
gene,
deterministic
model
(
Gustafson
and
Head,
2005).
That
is,
structured
and
natural
refuge
delay
TBW
resistance
effectively
to
the
two
proteins,
Cry1Ac
and
Cry2Ab2,
expressed
in
Bollgard
II.

Overall
Conclusions
and
Discussion
Overall,
the
modeling
suggests
that
Bollgard
II
cotton
should
have
more
than
25
years
of
durability
for
the
control
of
CBW
and
TBW
in
all
regions
with
natural
refuge
as
the
only
source
of
susceptible
insects.
The
presence
of
larger
amounts
of
Bollgard
II
in
the
marketplace
also
increases
the
durability
of
other
Bt
cotton
products
that
are
present
but
the
intrinsic
durability
of
Bollgard
II
cotton
is
much
greater
than
Bollgard
or
WideStrike
cotton.
However,
there
are
uncertainties
in
the
modeling
parameters
and
assumptions
which
could
impact
the
modeling
output.

Modeling
suggests
that
the
overall
durability
of
Bollgard
II
cotton
can
be
enhanced
if
Bollgard
cotton
is
removed
from
the
marketplace.
This
conclusion
is
supported
by
other
researchers
who
examined
the
benefit
of
managing
resistance
evolution
to
two
toxins
with
dissimilar
modes
of
action
using
a
pyramided
approach
(
Zhao
et
al.
2005;
Roush
1998;
Livingston
et
al.
2004;
Hurley
2000;
Caprio
2006).
On
the
other
hand,
the
concurrent
use
of
single­
and
two­
gene
Bt
plants
can
offer
exposed
populations
a
"
stepping
stone"
to
develop
resistance
to
both
proteins.
That
is
why
it
is
important
to
consider
the
removal
of
single,
less
durable,
Bt
cotton
products
from
the
marketplace
in
favor
of
more
durable,
two­
gene
(
assuming
independent
modes
of
action)
products.
In
Australia,
the
use
of
the
single­
gene
Bt
cotton
product
(
i.
e.,
Bollgard)
and
the
pyramided
Bt
cotton
product
(
i.
e.
Bollgard
II)
was
permitted
only
for
the
first
two
years
after
the
introduction
of
Bollgard
II,
but
now
only
Bollgard
II
is
permitted.
In
the
U.
S.,
Bollgard
II
cotton
has
been
in
the
marketplace
for
four
years
(
registered
in
December,
2002)
and
WideStrike
cotton
has
been
in
the
marketplace
for
two
years
(
registered
in
September,
2004).
In
2004,
Bollgard
cotton
acreage
accounted
for
>
95%
of
all
Bt
cotton
acreage
in
the
U.
S.
(
see
Head
et
al.
­
48­
2005,
MRID#
467172­
03).
Ten
years
of
selection
pressure
for
resistance
to
Cry1Ac
has
already
occurred.
Field
resistance
to
Cry1Ac
places
additional
selection
pressure
on
the
Cry2Ab2
component
of
Bollgard
II
cotton.
Monsanto's
modeling
does
not
assume
any
prior
selection
for
Cry1Ac
resistance
either
by
TBW
or
CBW.
Encouraging
the
adoption
of
Bollgard
II
will
increase
the
overall
durability
of
all
three
Bt
cotton
products.
From
an
insect
management
point
of
view,
removal
of
Bollgard
cotton
from
the
marketplace
should
be
encouraged
as
quickly
as
possible.

Researchers
from
Australia
(
Mahon
et
al.
2004)
have
reported
that
one
population
of
Helicoverpa
armigera
(
a
close
relative
of
Helicovera
zea,
CBW,
in
the
U.
S.)
collected
from
maize
plants
exhibited
resistance
to
Cry2Ab.
These
researchers
note
that
there
is
no
evidence
of
cross­
resistance
to
Cry1Ac
and
that
inheritance
of
resistance
is
recessive.
An
explanation
as
to
why
H.
armigera
had
unexpected
high
levels
of
Cry2Ab2
resistance
levels
is
unknown,
but
research
is
ongoing
to
evaluate
this
resistance.
A
better
understanding
of
background
levels
of
resistance
to
Cry2Ab
toxins
in
H.
armigera
populations
and
opportunities
for
survival
of
resistant
genotypes
on
Bollgard
II
is
needed.
What
this
means
for
either
TBW
or
CBW
resistance
to
Cry2Ab
toxins
in
Bollgard
II
in
the
U.
S.
is
not
known,
but
we
should
be
aware
of
the
possibility
that
Cry2Ab
resistance
may
be
more
common
than
expected.

V.
Proposed
Revisions
to
the
Bollgard
II
IRM
Program
Monsanto's
proposed
revisions
to
the
IRM
program
to
use
natural
refuge
for
Bollgard
II
cotton
are
contained
in
the
volume
titled
"
Scientific
and
Economic
Justification
for
Not
Requiring
Structured
Cotton
Refuges
for
Bollgard
II
Cotton
in
the
U.
S.
Cotton
Belt
from
Texas
to
the
East
Coast"
(
MRID#
467172­
03).
This
report
summarizes
the
alternate
TBW
host
studies,
TBW
effective
refuge,
and
resistance
modeling
previously
discussed
in
this
review.
In
addition,
Monsanto
has
included
sections
describing
cross
resistance
potential,
CBW
considerations,
economic
and
environmental
benefits
of
the
proposed
IRM
revision.

a)
Natural
Refuge
Proposal
As
described
in
the
Background
section,
the
terms
and
conditions
of
registration
for
Bt
cotton
varieties
(
including
Bollgard
and
Bollgard
II
cotton)
required
the
planting
of
a
structured,
non­
Bt
cotton
refuge.
For
TBW,
the
refuge
strategy
is
supported
by
"
high
dose"
expression
of
Bt
toxin(
s)
by
the
registered
hybrids.
For
CBW,
although
there
is
not
high
dose
expression,
the
pest
is
highly
polyphagous
and
experimental
data
have
shown
that
it
makes
use
of
alternate,
noncotton
hosts
(
see
discussion
in
the
Background
section).
The
IRM
plan
for
Bt
cotton
also
includes
resistance
monitoring
requirements
for
both
TBW
and
CBW
on
all
registered
toxins.
As
of
the
2004
growing
season,
no
documented
cases
of
resistance
of
significant
shifts
in
pest
susceptibility
to
Cry1Ac
(
the
toxin
in
Bollgard)
have
been
detected
(
see
BPPD
2005).

Monsanto
has
proposed
to
replace
the
structured
refuge
requirements
for
Bollgard
II
cotton
with
a
natural
refuge
approach
(
a
similar
proposal
for
Bollgard
cotton
has
not
been
proposed).
­
49­
Growers
would
no
longer
be
required
to
plant
a
non­
Bt
cotton
refuge
and
could
plant
up
to
100%
Bollgard
II
cotton.
The
natural
refuge
strategy
would
be
applicable
to
the
southeastern
cotton
growing
regions
including
Texas
(
western
cotton
regions,
in
which
pink
bollworm
is
the
primary
target
pest,
are
not
included).
This
proposal
is
based
on
a
number
of
supporting
factors:
1)
Bollgard
II
expresses
two
distinct
Bt
toxins
which
offers
an
IRM
advantage
over
single
toxin
products;
2)
Bollgard
II
offers
better
control
of
the
target
pests
(
TBW
and
CBW)
than
Bollgard
cotton;
3)
The
alternate
host
data
for
TBW
(
reviewed
in
this
memorandum)
and
CBW
(
previously
reviewed
­
see
Background
section)
indicate
that
adequate
natural
refuge
exists
independent
of
non­
Bt
cotton
refuges.

Monsanto's
submission
described
the
potential
resistance
management
benefits
of
using
multiple
toxins
(
also
known
as
"
pyramiding"
toxins).
A
number
of
models
(
Roush
1994,
1998
and
Caprio
1998b)
indicate
that
the
deployment
of
multiple
toxins
in
a
transgenic
crop
can
delay
the
onset
of
pest
resistance
longer
than
a
single
toxin
crop.
The
advantages
of
pyramided
toxins
are
contingent
on
several
factors:
high
efficacy
of
each
toxin
against
the
target
pest
and
distinct
modes
of
action
for
each
toxin
(
i.
e.
lack
of
cross
resistance).
In
the
case
of
Bollgard
II,
the
Cry1Ac
and
Cry2Ab2
Bt
toxins
are
expressed
(
the
event
was
created
by
inserting
the
Cry2Ab2
gene
into
existing
Bollgard
cotton).
Both
Cry1Ac
and
Cry2Ab2
each
are
known
to
have
high
activity
against
TBW
and
together
the
toxins
were
shown
to
increase
the
overall
efficacy
against
TBW
by
3.5
fold.
For
CBW,
Cry2Ab2
was
shown
to
have
high
activity
(
close
to
"
high
dose"),
while
Cry1Ac
has
less
activity
and
is
not
considered
high
dose.

b)
Cross
Resistance
The
IRM
benefits
of
a
multiple
gene
product
could
be
reduced
if
there
is
cross
resistance
between
the
expressed
toxins.
Cross
resistance
can
result
if
a
pest
develops
resistance
to
one
toxin,
which
then
confers
a
degree
of
resistance
to
the
second
toxin.
Monsanto's
submission
summarizes
information
to
support
the
contention
that
there
is
no
cross
resistance
potential
between
Cry1Ac
and
Cry2Ab2.
These
summarized
data
to
discuss
cross
resistance
were
originally
submitted
by
Monsanto
with
the
initial
registration
application
for
Bollgard
II
(
MRID#
455457­
01).

Structurally,
the
Cry1A
and
Cry2A
groups
of
Bt
toxins
are
divergent,
with
only
20%
similarity
between
amino
acid
sequences.
In
addition,
polyclonal
antibody
assays
showing
an
absence
of
cross­
reactivity
of
anti­
Cry2Ab2
antibodies
with
Cry1Ac
protein
(
and
vice­
versa)
demonstrate
that
the
tertiary
structure
of
Cry2A
proteins
differs
significantly
from
Cry1A
proteins.
Given
the
lack
of
structural
similarities,
Monsanto
concluded
that
the
insecticidal
mechanisms
of
Cry1Ac
and
Cry2Ab2
are
distinct
and
unlikely
to
result
in
cross
resistance.

Once
ingested
by
the
target
pests,
the
Cry1Ac
and
Cry2Ab2
toxins
behave
differently.
During
trypsin
digestion,
Cry1Ac
forms
a
stable
core
protein
that
is
not
produced
by
Cry2Ab2
under
similar
conditions.
Also,
in
the
insect
midgut,
the
two
toxins
bind
to
the
gut
membranes
in
different
manners:
Cry1Ac
interacts
with
brush
border
membrane
proteins
to
open
ion
channels
whereas
Cry2Ab2
interact
with
other
proteins
to
open
unique
ion
channels.
The
ion
channels
­
50­
created
by
Cry2Ab2
are
larger
and
have
reduced
ion
selectivity
than
those
created
Cry1Ac,
which
may
explain
the
increased
potency
observed
with
the
toxin.
As
with
the
structural
differences
between
the
two
toxins,
these
differences
in
membrane
interaction
suggest
low
cross
resistance
potential
between
for
the
Bollgard
II
toxins.

Further
evidence
to
document
the
lack
of
cross
resistance
potential
was
shown
in
studies
utilizing
Cry1Ac­
resistant
colonies
of
TBW
and
CBW.
When
challenged
on
Cry2Ab2
(
Bollgard
II
plant
material),
Cry1Ac­
resistant
TBW
remained
susceptible
to
the
protein
with
few
survivors
(
none
of
the
survivors
progressed
beyond
the
first
larval
instar).
A
similar
experiment
with
CBW
resulted
in
less
than
5%
survival
on
Bollgard
II
cotton
plants.

BPPD
Review
(
Cross
Resistance)

Cross
resistance
data
were
previously
submitted
to
BPPD
by
Monsanto
for
the
initial
registration
of
Bollgard
II.
These
data
(
the
same
data
set
described
above)
were
reviewed
in
detail
by
BPPD
(
see
BPPD
2002a)
prior
to
the
issuance
of
the
registration.
In
this
review,
BPPD
agreed
with
Monsanto's
overall
conclusions
that
the
cross
resistance
potential
between
Cry1Ac
and
Cry2Ab2
in
Bollgard
II
should
be
low
for
both
TBW
and
CBW
based
on
structural
and
binding
characteristics.

However,
since
Bollgard
II
has
been
registered,
additional
research
has
been
conducted
into
the
cross
resistance
potential
of
Cry1
and
Cry2
proteins.
In
one
study,
Jurat­
Fuentes
et
al.
(
2003)
investigated
resistance
to
Cry1Ac
and
Cry2Aa
in
TBW.
Several
strains
of
TBW
that
had
been
selected
on
and
developed
resistance
to
Cry1Ac
were
found
to
also
be
resistant
to
Cry2Aa.
Because
these
two
toxins
are
not
known
to
share
binding
sites
in
TBW,
the
authors
suggested
that
another
mechanism
may
be
involved
in
the
observed
resistance.
This
alternate
resistance
mechanism
could
be
one
that
affects
a
shared
process
in
the
mode
of
action
of
the
two
toxins.
As
was
described
by
Monsanto,
the
structure
and
binding
properties
of
Cry1A
and
Cry2A
toxins
are
different
at
the
membrane
level.
However,
both
toxins
have
similar
activation
processes
in
the
midgut
before
membrane
interaction;
an
alteration
of
the
toxin
activation
process
(
i.
e.
differential
midgut
protease
activity)
could
lead
to
common
resistance
to
both
proteins.
The
authors
also
suggested
"
midgut
epithelium
regeneration"
as
a
potential
mechanism
(
in
which
the
midgut
recovers
from
exposure
to
cry
toxins),
although
this
area
requires
additional
research.

Though
Cry2Aa
is
not
a
toxin
expressed
in
Bollgard
II,
the
work
of
Jurat­
Fuentes
et
al.
(
2003)
illustrate
the
need
to
consider
alternate
resistance
mechanisms
other
than
structure
or
binding
when
evaluating
cross
resistance
potential.
Other
potential
resistance
mechanisms
could
include
metabolic
changes
(
e.
g.
protease
inhibition,
gut
recovery)
or
behavioral
adaptations.
Such
alternate
mechanisms
may
be
complex
and
difficult
to
evaluate
in
the
context
of
transgenic
crops
(
most
cross
resistance
evaluations
have
been
conducted
in
the
laboratory).
BPPD
previously
discussed
some
of
the
theoretical
considerations
for
cross
resistance
in
transgenic
crops
as
part
of
the
2001
Bt
crops
reassessment
(
see
EPA
2001),
concluding
that
additional
study
is
needed
to
full
understand
the
ramifications
for
IRM.
­
51­
For
Bollgard
II,
the
data
evaluated
to
date
on
protein
structure
and
midgut
binging
seem
to
indicate
that
the
cross
resistance
potential
may
be
relatively
low
for
Cry1Ac
and
Cry2Ab2.
However,
based
on
the
work
of
Jurat­
Fuentes
et
al.
and
the
limited
overall
knowledge
base
on
the
subject,
cross
resistance
cannot
be
entirely
eliminated
as
a
possibility
with
Bollgard
II.
Another
important
consideration
is
the
presence
of
Bollgard
cotton
(
or
other
single
gene
varieties
expressing
Cry1Ac)
in
the
marketplace
with
Bollgard
II.
If
resistance
to
Cry1Ac
were
to
develop
on
Bollgard,
dual
gene
Bollgard
II
would
essentially
become
a
single
toxin
(
Cry2Ab2)
product.
In
this
case,
TBW
would
only
require
a
single
resistance
mechanism
to
Cry2Ab2,
negating
the
benefits
of
Bollgard
II
as
a
pyramided
crop.
Monsanto's
natural
refuge
proposal
made
no
specific
commitment
to
withdraw
Bollgard
from
the
market
should
the
proposal
be
accepted,
so
it
is
likely
significant
acreage
of
Bollgard
will
remain
in
place.

c)
Alternative
Hosts/
Natural
Refuge
­
CBW
CBW
is
commonly
known
as
a
highly
polyphagous
pest
that
feed
on
a
wide
range
of
plant
hosts,
including
cultivated
crops,
wild
host
plants,
and
weeds
in
addition
to
cotton.
To
evaluate
and
quantify
these
alternate
hosts
as
potential
natural
refuge,
Monsanto
conducted
a
large
scale
experiment
during
2002
and
2003.
This
study
involved
an
aerial
mapping
project
of
cropping
patterns
in
cotton
growing
states,
a
survey
of
CBW
production
on
alternate
hosts,
and
a
bioassay
designed
to
detect
the
type
of
host
plant
used
for
development
of
sampled
CBW
(
a
"
C3/
C4"
analysis
of
carbon
isotopes).
The
C3/
C4
bioassay
is
described
in
the
Background
section
of
this
review.

Overall,
Monsanto
concluded
that
these
studies
showed
that
there
are
substantial
areas
of
alternate
host
crops
(
corn,
peanut,
sorghum,
and
soybean)
available
in
cotton­
growing
areas
and
that
CBW
production
is
high
on
these
crops.
The
C3/
C4
assays
revealed
that
C4
alternative
hosts
(
primarily
corn
and
sorghum)
make
a
significant
contribution
to
the
CBW
adult
population
throughout
the
season.
These
data
have
been
previously
reviewed
by
BPPD
(
see
BPPD
2004a)
and
are
fully
discussed
in
the
Background
section.
BPPD
previous
reviewed
Monsanto's
model
as
part
of
the
June
2004
SAP
meeting
on
the
effective
of
natural
refuge
for
CBW
resistance
management
in
Bollgard
and
Bollgard
II
cotton
(
SAP
2004).
Monsanto
revised
its
current
and
natural
effective
refuge
calculations
and
model
based
on
the
June
2004
SAP
recommendations
(
SAP
2004).
Based
on
Monsanto's
modeling
of
intrinsic
durability
and
three­
gene
multiple
product
durability
modeling,
there
is
likely
enough
effective
natural
refuge
to
be
sufficient
to
delay
the
evolution
of
CBW
resistance
to
Bollgard
II
cotton
for
more
than
25
years
(
not
a
precise
number
of
years)
under
all
plausible
scenarios
in
all
four
regions.
Bollgard
II
retained
the
highest
level
of
efficacy
against
CBW
in
all
scenarios
(
all
regions)
because
of
the
relatively
high
mortality
of
individuals
heterozygous
to
Cry1Ac
resistance
in
the
presence
of
Cry2A2,
as
compared
to
WideStrike;
intermediate
in
many
scenarios
because
of
the
shared
binding
receptor
between
Cry1F
and
CryAc
and
likelihood
of
cross­
resistance,
and
Bollgard,
weakest
in
all
scenarios,
no
high
dose
for
CBW
and
a
single­
gene
product.
The
presence
of
greater
amounts
Bollgard
II
cotton
in
the
marketplace
increased
the
durability
of
both
Bollgard
and
WideStrike
cotton.
Conversely,
the
presence
of
substantial
amounts
of
Bollgard
cotton
in
the
marketplace
tended
to
decrease
the
longevity
of
Bollgard
II
and
WideStrike
cotton.
Monsanto's
modeling
did
­
52­
not
assume
prior
selection
for
Cry1Ac
resistance.
A
detailed
discussion
of
the
CBW
modeling
is
found
in
Section
IV
of
this
review.

d)
Alternative
Hosts/
Natural
Refuge
­
TBW
Like
CBW,
TBW
is
known
as
a
polyphagous
pest
that
utilizes
a
number
of
crops
and
wild
hosts
for
development.
To
further
analyze
the
potential
of
alternate
hosts
to
serve
as
refuge,
Monsanto
provided
a
literature
review
of
TBW
biology,
a
description
of
a
two
year
TBW
sampling
effort/
gossypol
bioassay
to
determine
host
plant
development,
a
calculation
of
TBW
"
effective"
refuge,
and
the
results
from
an
aerial
mapping
project
to
investigate
cropping
patterns.
From
these
efforts,
Monsanto
has
concluded
that
a
large
portion
of
the
TBW
population
in
cottongrowing
regions
originates
from
alternate
(
non­
cotton)
hosts,
resulting
in
an
effective
refuge
that
exceeds
20%.
Monsanto's
submitted
materials
and
conclusions
are
reviewed
and
discussed
in
detail
in
sections
I,
II,
III,
and
IV
of
this
document.

e)
Modeling
 
TBW
and
CBW
As
part
of
the
analysis
of
alternate
host
utilization
and
the
impact
on
IRM/
refuge,
Monsanto
ran
model
simulations
using
the
alternative
host
data
obtained
for
CBW
and
TBW.
Simulations
were
run
for
distinct
cotton­
growing
regions
(
CBW)
or
states
(
TBW)
with
scenarios
accounting
for
both
natural
refuge
only
and
effective
refuge
(
natural
refuge
plus
Bt
cotton
refuges
and
other
non­
Bt
cotton
acreage).
For
TBW,
"
worst
case"
counties
(
those
with
the
lowest
amount
of
natural
refuge)
were
modeled.
For
CBW,
the
four
regions
were
modeled.
Based
on
the
model
results,
Monsanto
concluded
that
Bollgard
II
cotton
would
have
more
than
25
years
without
CBW
or
TBW
resistance
using
the
natural
refuge
strategy.
None
of
three
Bt
cotton
products
lost
efficacy
against
TBW
during
the
30
year
time
horizon
of
the
modeling
using
any
current
or
natural
refuge
scenario
except
scenario
1­
N,
if
the
structured
refuge
requirements
for
Bollgard
and
WideStrike
cotton
were
removed.
CBW
modeling
predicted
that
higher
amounts
of
Bollgard
II
enhanced
the
durability
of
both
Bollgard
and
WideStrike.
Conversely,
higher
amounts
of
Bollgard
reduced
the
durability
of
both
Bollgard
and
WideStrike.
The
modeling
and
results
are
discussed
in
detail
in
section
IV
of
this
review.

f)
Economic
Benefits
Monsanto's
submission
concluded
that
replacing
the
structured
refuge
requirements
for
Bollgard
II
will
result
in
economic
benefits
to
growers.
Overall,
Bollgard
II
cotton
provides
higher
yields
and
less
input
costs
compared
with
non­
Bt
(
conventional)
cotton
treated
with
insecticides.
Under
the
current
IRM
requirements
(
5%
or
20%
non­
Bt
cotton
refuges),
growers
experience
refugerelated
costs
including
reduced
productivity
in
refuge
acres
in
addition
to
refuge
maintenance
requirements
(
time
and
labor).

For
the
20%
refuge
requirement,
growers
must
plant
the
required
non­
Bt
cotton
acreage
within
one
mile
of
the
Bollgard
II
acreage
and
may
treat
the
refuge
as
needed
with
insecticides
to
control
lepidopteran
pests.
Comparative
studies
summarized
by
Monsanto
have
shown
that
the
­
53­
pest
control
costs
are
smaller
(
under
moderate
to
high
pest
pressure)
and
crop
yields
are
higher
with
Bt
cotton
relative
to
conventional
cotton.
For
example,
data
from
1995­
2002
showed
an
average
increase
in
yield
(
lint)
of
6%
with
Bollgard
cotton
(
another
set
of
data
showed
a
10%
average
lint
increase
during
1995­
1999).
Since
refuge
acres
are
essentially
conventional
cotton
that
is
managed
with
insecticides,
growers
do
not
experience
these
benefits
of
Bt
cotton
on
at
least
20%
of
their
cotton
acres.
One
study
cited
by
Monsanto
(
Banerjee
and
Martin
2005)
demonstrated
that
Bt
cotton
had
a
$
38
per
acre
benefit
over
conventional
(
refuge)
cotton
plantings.
Other
data
sets
tabulated
by
Monsanto
(
1995­
2002)
and
a
third
party
study
(
1995­
1999)
indicated
that
growers
received
an
average
net
dollar
return
of
$
39.35/
acre
and
$
49.80/
acre
respectively
with
Bollgard.
Based
on
these
results,
Monsanto
estimated
that
Bt
cotton
provides
a
$
40/
acre
benefit
over
the
20%
treatable
refuge
acreage,
not
including
other
refuge
management
costs
(
e.
g.
scouting,
labor,
time,
etc.).
Therefore,
grower
using
this
refuge
option
will
lose
$
40
per
acre
on
20%
of
their
cotton
acreage;
when
averaged
across
all
cotton
acres
(
Bt
and
refuge),
the
cost
amounts
to
$
8
per
acre.
For
a
typical
grower
planting
1,800
acres
(
the
average
size
farm
in
the
Delta
region)
the
20%
refuge
option
results
in
a
total
cost
of
about
$
14,400
(
1,800
acres
x
$
8
per
acre
cost).

With
the
5%
refuge
options
(
embedded
or
external
unsprayed),
Monsanto
estimated
that
growers
can
lose
up
to
$
150
per
acre
relative
to
their
Bollgard
acreage.
Another
estimate
from
a
cotton
researcher
estimated
losses
of
$
67
per
acre
with
the
5%
options
(
Martin
et
al.
2006).
The
losses
can
be
attributed
primarily
to
reduced
yields
and
are
particularly
acute
for
the
5%
unsprayed
option,
where
as
much
as
100%
of
the
crop
can
be
lost
due
to
pest
infestation.
Monsanto
calculated
that
growers
planted
an
unsprayed
refuge
will
yield
almost
250
pounds
per
acre
less
than
conventionally
grown
(
insecticide
treated)
cotton.
For
the
5%
refuge,
the
total
cost
when
averaged
across
all
cotton
acres
(
5%
refuge
and
95%
Bt
cotton)
is
$
7.50/
acre
(
similar
to
the
$
8/
acre
cost
for
the
20%
refuge).
Considering
a
typical
1,800
acre
cotton
farm,
the
overall
cost
is
approximately
$
13,500
(
1,800
acres
x
$
7.50
per
acre
cost).

Monsanto's
submission
also
discussed
other
non­
quantified
refuge­
related
costs
to
growers
(
i.
e.
non­
yield
and
treatment
related).
In
much
of
the
cotton­
growing
area,
farms
are
becoming
larger
and
more
complex.
Cotton
plantings
may
include
many
fields
over
large
areas,
with
some
acreage
falling
in
multiple
counties.
The
need
to
plant
refuges
can
limit
the
flexibility
of
these
operations
by
requiring
in­
depth
land
use
planning.
During
planting,
refuges
add
logistical
impediments,
such
as
the
use
of
additional
labor
and
the
need
to
clean
planter
boxes
between
fields.

BPPD
Review
(
Economic
Benefits)

As
part
of
the
evaluation
of
the
Bollgard
II
registration,
BPPD
reviewed
benefits
information
related
to
the
product
(
BPPD
2002b),
including
an
assessment
of
the
economic
benefits
of
the
crop.
BPPD
estimated
the
per
acre
benefit
of
Bollgard
II
to
be
$
27.63
per
acre,
considering
the
(
projected)
technology
fee,
adoption
rate,
improved
efficacy/
pest
spectrum,
and
reduced
insecticide
costs
for
secondary
pests
(
e.
g.
beet
armyworm).
This
figure
differs
from
Monsanto's
calculations,
which
were
$
40
to
$
150
per
acre,
depending
on
the
refuge
option
employed.
It
is
­
54­
noted
that
BPPD's
assessment
was
conducted
prior
to
registration
using
projections
for
many
of
the
inputs,
while
Monsanto
used
or
cited
data
obtained
from
farm
surveys
of
growers
planting
Bollgard.
Regardless
of
the
specific
per
acre
benefit,
BPPD
agrees
that
Bollgard
and
Bollgard
II
provide
significant
economic
benefits
to
growers
relative
to
conventional
and
refuge
non­
Bt
cotton.

On
the
other
hand,
short
term
economic
benefits
to
growers
may
be
compromised
should
resistance
to
Bt
cotton
occur.
The
primary
objective
of
IRM
and
structured
refuges
has
been
to
preserve
the
effectiveness
of
Bt
toxins
and
maintain
long
term
benefits
to
growers.
If
resistance
were
to
occur,
growers
will
likely
experience
costs
related
to
reverting
to
other
(
still
efficacious)
Bt
varieties
or
conventional
(
insecticide­
treated)
cotton.
Additional
costs
could
result
from
implementing
the
remedial
action
plan,
which
requires
(
among
other
measures)
the
use
of
conventional
insecticides
to
control
outbreak
pest
populations.
Monsanto's
submission
did
not
address
the
potential
effects
of
resistance
on
the
cost/
benefit
analysis
for
Bollgard
II,
though
quantifying
such
costs
would
likely
be
difficult
and
involve
imprecise
estimates.

g)
Environmental
Benefits
In
addition
to
direct
economic
benefits
to
growers,
Monsanto's
submission
described
a
number
of
environmental
benefits
resulting
from
adoption
of
Bollgard
II
cotton.
These
benefits
include
improved
resistance
management,
reduced
use
of
conventional
insecticides,
and
prolonged
efficacy
of
conventional
insecticides.

In
terms
of
IRM
benefits,
Bollgard
II
provides
a
two
toxin
approach
to
controlling
the
target
pests.
Models
have
shown
that
a
dual
dose
product
can
delay
resistance
relative
to
single
dose
products,
although
this
advantage
can
be
curtailed
if
significant
acreage
of
single
toxin
products
also
coexists
with
the
dual
toxin
product
in
the
landscape.
Single
gene
products
(
e.
g.
Bollgard
cotton)
can
provide
a
"
stepping
stone"
to
resistance,
increasing
the
frequency
in
which
resistance
develops
in
the
dual
gene
transgenic
crop.
To
date,
Bollgard
cotton
has
much
greater
adoption
among
cotton
growers
than
Bollgard
II,
largely
due
to
economic
reasons
(
i.
e.
Bollgard
II
has
no
clear
advantage
over
Bollgard
for
most
growers).
However,
Monsanto
has
contended
that
should
the
natural
refuge
proposal
be
accepted,
growers
will
have
a
clear
incentive
to
adopt
Bollgard
II
cotton
on
a
wider
scale
(
i.
e.
no
required
structured
refuge,
increased
economic
benefits).
If
this
proves
to
be
the
case,
the
transition
from
Bollgard
to
Bollgard
II
across
the
landscape
will
improve
IRM
by
reducing
selection
pressure
to
Cry1Ac
found
in
single
toxin
Bollgard
hybrids.

Further
insecticide
use
reduction
benefits
could
be
obtained
through
the
adoption
of
the
natural
refuge
proposal
(
allowing
Bollgard
II
on
100%
of
growers'
acreage).
A
survey
cited
by
Monsanto
showed
that
22%
of
cotton
growers
utilized
the
20%
refuge
option,
which
allows
for
the
use
of
conventional
insecticides
to
treat
pests
in
the
refuge.
By
planting
100%
Bollgard
II,
growers
would
no
longer
need
to
treat
those
20%
refuge
acres
for
lepidopteran
pests.
Based
on
the
22%
adoption
figure,
Monsanto
estimated
that
385,000
total
refuge
acres
could
be
saved
from
insecticide
treatment.
Monsanto
also
noted
that
Bollgard
II
cotton
has
the
potential
to
reduce
up
to
three
insecticide
treatments
per
season
in
comparison
to
conventional
cotton.
The
­
55­
resulting
decrease
in
pesticide
use
has
potential
benefits
for
non­
target
species,
including
predatory
arthropods
that
would
avoid
exposure
to
non­
specific
insecticides.
Additional
benefits
could
result
from
reduced
fuel
use
and
the
need
to
dispose
of
fewer
pesticide
containers.

In
conjunction
with
overall
reduced
pesticide
usage,
Monsanto
has
proposed
that
the
efficacy
of
conventional
insecticides
could
be
preserved
for
longer
periods
of
time.
Less
use
of
pesticides
would
mean
less
exposure
to
pest
populations
and
ultimately
less
selection
pressure
for
resistance.
Pest
resistance
has
been
problematic
for
a
number
of
chemicals
used
in
cotton,
most
notably
the
pyrethroid
insecticides
and
TBW.

BPPD
Review
(
Environmental
Benefits)

BPPD's
assessment
of
the
benefits
for
Bollgard
II
identified
both
resistance
management
and
reduced
insecticide
use
as
positive
environmental
advantages
of
the
product.
For
IRM,
it
is
clear
that
a
two
gene
product
with
functional
high
doses
for
the
target
pests
is
a
better
option
than
a
single
gene
product
for
the
same
pests.
However,
as
Monsanto
has
stated
in
their
report,
a
landscape
containing
a
mixture
of
single
gene
and
dual
gene
products
can
compromise
the
IRM
benefits
of
the
two
gene
crop.
This
can
occur
if
resistance
develops
to
the
single
gene
crop
(
Bollgard/
Cry1Ac
in
this
case);
the
dual
gene
product
(
Bollgard
II)
essentially
becomes
a
single
gene
product
expressing
Cry2Ab2.
Monsanto's
natural
refuge
proposal
provides
no
details
or
plan
for
increasing
adoption
of
Bollgard
II
(
and
phasing
out
Bollgard)
other
than
the
incentive
to
forgo
structured
refuge
with
Bollgard
II.
It
is
not
clear
how
quickly
Bollgard
II
acreage
will
increase
and
whether
sufficient
hybrids
will
be
available
with
the
Bollgard
II
genes.
Given
the
wide
discrepancy
between
Bollgard
and
Bollgard
II
acreage,
it
is
highly
probable
that
significant
acreage
of
Bollgard
will
remain
in
place
even
if
the
natural
refuge
proposal
is
implemented.
Since
large
acreage
of
Bollgard
could
present
a
resistance
risk
for
Bollgard
II,
it
is
recommended
that
Monsanto
provide
details
as
to
how
the
adoption
of
Bollgard
II/
phase
out
of
Bollgard
will
be
facilitated.

BPPD
agrees
with
Monsanto
regarding
the
benefits
of
reduced
pesticide
use.
Bt
cotton
has
clearly
decreased
the
need
for
insecticide
treatments
for
lepidopteran
pests
(
refer
to
the
benefits
assessment
in
EPA
2001).
The
use
of
100%
Bollgard
II,
with
high
efficacy
against
the
major
target
pests
and
some
secondary
pests,
presumably
would
eliminate
the
need
for
treatment
of
the
20%
structured
refuge
option.
The
exact
amount
of
insecticide
reduction
would
be
dependent
on
the
scale
of
adoption
of
Bollgard
II,
an
issue
that
remains
unclear.
It
should
be
noted
that,
like
the
economic
and
IRM
benefits,
is
pest
resistance
occurs
to
Bollgard
II,
the
pesticide
use
reduction
may
be
compromised.
In
fact,
resistance
would
likely
result
in
a
substantial
increase
in
the
amount
of
pesticide
used,
as
growers
would
either
need
to
treat
Bollgard
acreage
(
as
part
of
remedial
action)
or
resume
planting
conventional
cotton
varieties
and
regular
pest
treatments.

BPPD
Review
(
Overall
Natural
Refuge
Proposal
for
TBW)

Taken
together,
the
TBW
sampling,
gossypol
analyses,
and
effective
refuge
determinations
clearly
demonstrate
that
a
significant
portion
of
the
TBW
population
is
derived
from
non­
cotton
­
56­
hosts.
This
conclusion
supports
what
is
commonly
known
in
the
literature:
that
TBW
is
a
polyphagous
pest
that
utilizes
a
number
of
cultivated
crops,
wild
hosts,
and
weeds
for
larval
development.
However,
questions
remain
on
the
larger
issue
of
whether
these
alternate
hosts
can
serve
as
unstructured
refuge
for
TBW
for
Bollgard
II
Bt
cotton.
Within
the
Cotton
Belt,
the
gossypol
data
differed
greatly
by
state/
region.
For
some
areas
(
i.
e.
North
Carolina/
Georgia),
the
proportion
of
non­
cotton­
origin
TBW
was
consistently
high
throughout
the
season
and
may
support
the
natural
refuge
proposal.
In
other
regions
(
i.
e.
the
Mississippi
Delta,
eastern
Texas)
the
data
were
variable
and
more
difficult
to
interpret.
Given
the
assumptions
of
the
Monsanto
model,
TBW
resistance
to
Bollgard
II
is
predicted
not
to
evolve
over
the
30­
year
time
horizon
of
the
model
for
any
scenario
(
current
or
natural
effective
refuge)
in
any
region.
On
the
other
hand,
there
are
a
number
of
unresolved
issues
with
the
data,
effective
refuge,
modeling,
and
interpretation
of
the
results
(
detailed
below).
Should
these
issues
be
resolved,
BPPD
could
support
the
use
of
natural
refuge
alone
for
Bollgard
II
in
the
southeastern
U.
S.
cotton
region.
The
strongest
case
for
support
of
a
natural
refuge
for
TBW
resistance
management
exists
for
the
North
Carolina
and
Georgia
regions.

While
Monsanto
utilized
the
same
gossypol
methodology
in
both
2004
and
2005,
there
was
no
statistical
analysis
of
the
gossypol
results
between
the
two
years,
likely
because
of
differences
in
sampling
for
each
season.
Therefore,
it
is
not
possible
to
compare
the
results
between
seasons
except
on
a
qualitative
basis.
Though
many
of
the
state
trends
were
similar
in
both
seasons,
it
is
not
known
if
there
was
a
definitive
correlation
between
the
data
sets.

As
previously
stated,
the
results
from
the
gossypol
bioassays
varied
by
region.
Perhaps
the
strongest
case
for
the
use
of
natural
refuge
can
be
made
in
North
Carolina
and
Georgia.
In
these
states,
the
data
were
more
consistent
and
clearly
showed
that
the
large
majority
(>
90%)
of
TBW
originated
from
non­
cotton
sources.
This
trend
was
clear
throughout
the
entire
cotton
growing
season
in
both
2004
and
2005
(
see
table
3
in
this
review
or
Monsanto
figure
1
in
appendix
1).
These
states
are
known
to
contain
substantial
acreage
of
alternate
crops
(
peanut,
tobacco,
and
soybean)
that
are
known
to
be
preferred
TBW
hosts.
Modeling
predicted
that
resistance
to
Bollgard
II
would
not
evolve
for
any
natural
or
current
effective
scenarios
for
either
the
North
Carolina
or
Georgia
region.

For
the
other
sampled
states,
the
data
more
variable
both
within
each
growing
season
and
between
seasons.
In
the
Delta
states
(
Mississippi,
Louisiana,
and
Arkansas),
the
general
trend
was
high
proportions
of
non­
cotton
TBW
early
in
the
season,
followed
by
a
decline
after
June
(
when
cotton
became
available
as
a
host).
During
this
period,
the
percentage
of
cotton­
origin
frequently
dipped
below
50%,
though
it
was
almost
always
at
least
20%
(
see
table
3).
In
Texas,
during
the
cotton
season
(
July­
October)
a
majority
of
TBW
were
found
to
be
cotton­
origin,
with
the
proportion
exceeding
80%
in
some
individual
counties.
These
results
seem
to
indicate
that
cotton
is
a
much
more
important
host
for
TBW
in
these
states
then
in
the
east
coast
states
(
North
Carolina
and
Georgia)
and
that
fewer
alternate
hosts
may
be
available
during
the
cotton
season.

BPPD
has
identified
several
issues
with
the
TBW
sampling
and
gossypol
methodology.
First,
the
use
of
pheromone
traps
for
sampling
resulted
in
the
collection
of
only
males.
As
such,
no
­
57­
females
were
included
in
the
bioassays.
It
is
unclear
whether
inferences
about
female
behavior
(
i.
e.
host
utilization)
can
be
made
from
data
obtained
solely
from
males.
BPPD
also
questions
whether
the
two
year
testing
period
is
adequate
for
an
experiment
of
this
magnitude.
For
both
Tennessee
and
Texas,
only
one
year
of
data
were
compiled.
Though
relatively
consistent,
cropping
and
landscape
patterns
can
change
over
time
­­
a
factor
that
may
not
be
reflected
in
a
study
of
limited
duration.
Given
the
inherent
variability
in
natural
and
agronomic
ecosystems,
a
two
year
study
may
present
only
a
"
snapshot"
of
the
host
availability
and
productivity
for
TBW
that
may
not
be
representative
of
future
conditions.
It
is
recommended
that
Monsanto
address
both
of
these
issues
in
a
future
submission.

In
addition
to
the
sampling
questions,
BPPD
noted
that
for
a
number
of
counties,
trapping
locations,
and
trapping
periods
there
were
low
trap
captures
of
TBW.
For
many
traps,
few
(
or
no)
moths
were
collected.
For
example,
for
Mississippi
during
July,
2004
(
the
height
of
the
cotton
season),
in
half
of
the
sampled
counties
(
6
of
12)
less
than
10
moths
were
collected.
In
Tennessee,
TBW
trap
captures
were
so
few
that
no
gossypol
analyses
were
conducted
during
2004.
Low
TBW
numbers
have
also
been
observed
with
Bt
cotton
resistance
monitoring
efforts
(
see
BPPD
2005),
possibly
due
to
a
suppressive
effect
from
the
widespread
adoption
of
transgenic
cotton.
Data
from
collection
sites
with
few
captures
(
i.
e.
<
10)
were
included
in
the
effective
refuge
calculations,
but
not
in
the
modeling
(
where
worst
case
scenarios
were
used).
It
is
unclear
if
the
low
trap
captures
in
some
areas
may
have
affected
the
reliability
of
the
results.
There
is
also
concern
that
should
a
resistant
population
emerge
locally
from
a
Bt
cotton
field,
there
will
be
insufficient
susceptible
TBW
available
to
mitigate
the
threat.

Overall,
the
modeling
suggests
that
Bollgard
II
cotton
should
have
more
than
30
years
of
durability
for
the
control
of
TBW
in
all
regions
with
natural
refuge
as
the
only
source
of
susceptible
insects.
The
predictions
of
the
Caprio's
spatially­
explicit,
stochastic,
two­
locus
TBW
resistance
model
(
Caprio
2006)
are
similar
to
those
predicted
by
the
Monsanto's
three­
gene,
deterministic
model
(
Gustafson
&
Head
2005).
From
a
qualitative
perspective,
modeling
indicates
the
intrinsic
durability
of
Bollgard
II
is
greater
than
WideStrike
which
is
greater
than
Bollgard.
The
presence
of
larger
amounts
of
Bollgard
II
in
the
marketplace
also
increases
the
durability
of
other
Bt
cotton
products
that
are
present
but
the
intrinsic
durability
of
Bollgard
II
cotton
is
much
greater
than
Bollgard
or
WideStrike
cotton.
However,
there
are
uncertainties
in
the
modeling
parameters
and
assumptions
which
could
impact
the
modeling
output.
Modeling
suggests
that
the
overall
durability
of
Bollgard
II
cotton
can
be
enhanced
if
Bollgard
cotton
is
removed
from
the
marketplace.
Uncertainties
in
the
pheromone
captures,
gossypol
analyses,
spatial
analysis,
estimation
of
effective
refuge
calculation,
degree
of
shared
binding
affinity
of
Cry1Ac
to
receptor
A
and
B,
genetics
of
resistance,
resistance
mechanism,
initial
resistance
allele
frequency,
and
other
modeling
assumptions
affect
the
precision
and
accuracy
of
the
modeling
predictions.
Additional
competition
binding
studies
using
TBW
brush
border
membrane
vesicles
at
saturable
concentrations
of
Cry1Fa
are
recommended.
Uncertainty
in
the
shared
binding
affinity
of
the
Cry1Ac
protein
to
receptor
A
and
B
(
in
Monsanto's
model)
would
impact
the
modeling
predictions
for
all
three
Bt
cotton
products
because
the
Cry1Ac
protein
is
common
to
all
of
them.
Monsanto's
modeling
also
assumes
no
prior
selection
to
Cry1Ac
even
­
58­
though
Bollgard
cotton
has
been
on
the
market
since
1996.
The
degree
to
which
the
modeling
predictions
would
be
impacted
is
unknown.

It
is
important
to
consider
that
Bollgard
II
(
Cry1Ac
and
Cry2Ab2)
exists
in
a
mosaic
with
single
gene
Bollgard
(
Cry1Ac).
Modeling
work
has
shown
that
in
such
a
scenario,
the
pyramided
product
can
be
at
risk
for
resistance
if
there
is
significant
acreage
of
a
single
gene
variety
(
Zhao
et
al.
2003,
2005;
Hurley
2000).
If
resistance
develops
to
the
single
gene
product
(
Cry1Ac
in
this
case)
the
dual
gene
product
would
effectively
become
a
single
gene
product
(
Cry2Ab2).
The
single
gene
variety
may
in
a
sense
act
as
a
"
stepping
stone"
for
resistance
to
the
pyramided
product.
Zhao
et
al.
(
2005)
have
shown
that
the
concurrent
use
of
one­
and
two­
gene
Bt
broccoli
plants
resulted
in
control
failure
of
both
types
of
Bt
plants
for
control
of
Plutella
xylostella
(
diamondback
moth)
rather
than
the
use
of
the
two­
gene
plants
by
themselves.
In
addition,
these
researchers
illustrated
through
their
modeling
efforts
that
resistance
will
evolve
first
to
the
toxin
that
is
being
used
singly.
Monsanto's
natural
refuge
proposal
contained
no
details
on
how
this
mosaic
might
be
managed.
There
was
no
discussion
as
to
whether
Monsanto
will
"
phase
out"
Bollgard
to
facilitate
the
adoption
of
Bollgard
II.
Rather,
it
was
suggested
in
the
submission
that
if
there
is
no
requirement
to
plant
structured
refuge
a
strong
incentive
will
exist
for
growers
to
switch
to
Bollgard
II.
However,
considering
the
large
acreage
of
Bollgard
currently
in
the
marketplace
(
relative
to
Bollgard
II),
it
is
highly
likely
that
a
significant
percentage
of
cotton
acreage
will
remain
planted
with
Bollgard
varieties
for
at
least
several
years.

The
potential
use
of
natural
refuge
for
Bollgard
II
represents
a
switch
from
structured
refuge
(
i.
e.
refuge
planted
and
managed
by
growers)
to
an
unstructured
(
i.
e.
unmanaged)
refuge.
There
have
been
three
major
considerations
to
designing
and
deploying
structured
refuge
for
Bt
crops.
These
are
1)
production
of
sufficient
numbers
of
susceptible
insects
to
dilute
any
potential
resistance
genes,
2)
proximity
of
the
refuge
to
the
Bt
field
to
ensure
random
mating
between
susceptible
and
resistant
insects,
and
3)
synchrony
of
the
refuge
with
the
Bt
fields
to
ensure
overlapping
emergence
between
susceptible
moths
and
any
resistant
survivors
of
the
Bt
crop.
When
evaluating
the
functionality
of
natural
(
unstructured)
refuge,
it
is
useful
to
consider
these
topics.

In
terms
of
susceptible
production,
the
goal
of
structured
refuge
for
Bt
crops
has
historically
been
a
ratio
of
500
susceptible
insects
for
every
resistant
insect
that
could
emerge
from
the
Bt
field
(
see
EPA
2001).
Monsanto's
TBW
sampling/
gossypol
experiments
did
not
look
at
susceptible
insect
production
from
a
numerical
perspective
per
se;
rather
the
proportion
of
moths
originating
on
cotton/
non­
cotton
sources
was
assessed
(
all
trapped
moths
can
be
assumed
to
be
Bt
susceptible).
The
experiments
clearly
showed
that
a
portion
of
the
TBW
develops
on
cotton,
while
another
portion
develops
on
non­
cotton
crops
and
wild
hosts.
However,
as
discussed
above,
the
numbers
collected
at
individual
traps
were
highly
variable,
with
some
traps
collecting
few
or
no
moths
and
others
trapping
over
one
hundred.
It
is
noted
that
the
sampling
was
conducted
under
the
current
IRM
strategy
that
mandates
a
non­
Bt
cotton
structured
refuge
which
should
(
theoretically)
have
supplied
a
source
of
susceptible
moths.
On
the
other
hand,
should
the
natural
refuge
strategy
be
employed,
the
acreage
of
Bollgard
II
cotton
could
significantly
increase
while
overall
non­
Bt
cotton
acreage
may
decrease.
Given
this
scenario,
it
is
conceivable
­
59­
that
the
portion
of
TBW
developing
on
cotton
(
as
determined
from
Monsanto's
experiments)
could
be
reduced,
leaving
mostly
non­
cotton­
origin
moths.
It
is
not
clear
whether
reducing
the
cotton­
origin
portion
of
the
susceptible
TBW
population
would
have
a
measurable
effect
on
the
ability
to
dilute
emerging
resistance
genes.

Both
refuge
proximity
and
developmental
synchrony
criteria
were
directly
addressed
by
Monsanto's
experimental
design.
For
proximity,
the
TBW
traps
were
placed
adjacent
to
cotton
fields
(
both
Bt
and
non­
Bt
cotton).
Though
the
traps
were
baited
with
a
pheromone
attractant
that
could
attract
moths
from
surrounding
areas,
the
sampling
likely
presented
an
accurate
portrayal
of
the
TBW
population
around
the
cotton
field
trap
sites.
Developmental
synchrony
was
also
examined
by
collecting
samples
throughout
the
growing
season.
Data
were
assessed
for
each
month
of
the
cotton
season
and
a
statistical
analysis
showed
that
in
the
majority
of
cases
the
cotton/
non­
cotton
proportions
did
not
significantly
differ
within
the
individual
sampling
months.
At
variable
proportions,
the
results
showed
that
TBW
developing
on
non­
cotton
hosts
overlapped
TBW
from
cotton
during
the
major
part
of
the
cotton
growing
season
(
July­
September)
(
see
figure
1
attached
to
this
review).
These
results
were
not
surprising:
though
TBW
may
develop
on
different
plant
hosts
at
slightly
different
rates,
there
are
four
to
six
generations
per
year
that
likely
create
a
natural
overlap
(
discussion
in
Benedict
2004).

BPPD
generally
agrees
with
Monsanto
that
a
natural
refuge
strategy
(
allowing
100%
Bollgard
II)
will
result
in
economic
and
environmental
benefits.
These
benefits
will
result
largely
from
yield
increases
and
reduced
insecticide
use
on
cotton
acreage
previously
planted
as
refuge
(
i.
e.
conventional
cotton).
However,
these
benefits
will
likely
be
compromised
should
the
lepidopteran
target
pests
develop
resistance
to
the
Bollgard
II
toxins.
In
such
a
case,
there
could
be
significant
costs
to
growers
that
might
have
to
switch
to
conventional
(
insecticide
treated)
cotton.

Under
the
current
terms
and
conditions
of
registration
for
approved
Bt
cotton
products,
registrants
are
required
to
monitor
for
pest
resistance
to
the
expressed
toxins.
To
date,
BPPD
has
reviewed
the
annual
monitoring
reports
through
the
2004
growing
season
and
found
no
evidence
of
resistance
to
Cry1Ac
or
Cry2Ab2
in
either
TBW
or
CBW
(
see
BPPD
2005).
Should
the
natural
refuge
proposal
ultimately
be
accepted,
it
is
highly
recommended
that
these
monitoring
efforts
continue
for
both
Cry1Ac
and
Cry2Ab2.
Monsanto's
submission
did
not
discuss
any
proposed
changes
to
the
monitoring
requirements
for
either
Bollgard
product.
However,
considering
that
replacement
of
structured
refuge
with
natural
refuge
may
increase
the
Bollgard
II
acreage
(
and
selection
pressure
for
resistance),
it
may
be
justified
to
increase
the
monitoring
efforts
for
Cry2Ab2
(
i.
e.
increased
sampling
sites
and
collections).

References
Monsanto
Submissions
(
others
referenced
by
author
in
the
next
section)

MRID#
455457­
01:
Insect
Resistance
Management
Plan
for
Bollgard
II
Cotton.
Submitted
in
2002.
­
60­
MRID#
467172­
01:
"
Production
of
Tobacco
Budworm
from
Alternate
Host
Plants
and
the
Role
of
These
Host
Plants
as
Natural
Refuge
for
Bollgard
II
Cotton."
Submitted
December
20,
2005.

MRID#
467172­
02:
"
Modeling
the
Impact
of
Natural
Refuge
on
the
Evolution
of
Tobacco
Budworm
and
Cotton
Bollworm
Resistance
to
Bollgard
II
Cotton."
Submitted
December
20,
2005.

MRID#
467172­
03:
"
Scientific
and
Economic
Justification
for
Not
Requiring
Structured
Cotton
Refuges
for
Bollgard
II
Cotton
in
the
U.
S.
Cotton
Belt
from
Texas
to
the
East
Coast."
Submitted
December
20,
2005.

April
3,
2006:
Letter
to
Dennis
Szuhay
containing
supplemental
information
requested
in
support
of
natural
refuge
amendment
request
for
Bollgard
II
cotton.
Supplemental
to
MRID#
467172­
01.

April
7,
2006:
Letter
to
Dennis
Szuhay
containing
supplemental
information
requested
in
support
of
natural
refuge
amendment
request
for
Bollgard
II
cotton.
Supplemental
to
MRID#
467172­
02.

May
10,
2006:
Letter
to
Dennis
Szuhay
containing
supplemental
information
requested
in
support
of
natural
refuge
amendment
request
for
Bollgard
II
cotton.
Supplemental
to
MRID#
467172­
02.

Author
References
Abney,
M.
R.,
C.
E.
Sorenson
and
J.
J.
R.
Bradley,
2004.
Alternate
crop
hosts
as
resistance
management
refuges
for
tobacco
budworm
in
NC.
Pages
1413­
1416
in
Proceedings
of
the
Beltwide
Cotton
Conferences,
San
Antonio,
Texas.
National
Cotton
Council.

Banerjee,
S.
and
S.
W.
Martin,
2005.
Farm
profits
from
observed
yields
of
Bt
and
non­
Bt
cotton,
with
and
without
spray
applications,
in
the
Mississippi
Delta.
Proceeding
of
the
Beltwide
Cotton
Conference,
New
Orleans,
Louisiana,
p.
297­
303.

Benedict,
J.
H.,
2004.
Biology
and
Dispersal
of
the
Cotton
Bollworm
and
Tobacco
Budworm
in
North
America
(
white
paper).
Submitted
in
Monsanto's
unpublished
study
"
Production
of
Tobacco
Budworm
from
Alternate
Host
Plants
and
the
Role
of
These
Host
Plants
as
Natural
Refuge
for
Bollgard
II
Cotton."
MRID#
467172­
01.

Bird,
L.
J.
and
R.
J.
Akhurst,
2004.
Relative
fitness
of
Cry1A­
resistant
and
­
susceptible
Helicoverpa
armigera
(
Lepidoptera:
Noctuidae)
on
conventional
and
transgenic
cotton.
J
Econ
Entomol
97(
5):
1699­
1709.

Blanco,
C.
A.,
2005.
Bacillus
thuringiensis
Cry1Ac/
Cry2Ab2
resistance
monitoring
program
for
­
61­
tobacco
budworm
and
bollworm
in
2004.
Unpublished
study
submitted
to
EPA.
MRID
#
465476­
01.

Blanco,
C.
A.,
L.
C.
Adams,
J.
Gore,
D.
Hardee,
M.
Mullen,
J.
R.
Bradley,
J.
W.
Van
Duyn,
P.
C.
Ellsworth,
J.
K.
Greene,
D.
R.
Johnson,
R.
Luttrell,
G.
Studebaker,
A.
Herbert,
M.
Karner,
R.
Leonard,
B.
Lewis,
J.
J.
D.
Lopez,
D.
Parker,
M.
L.
Williams,
R.
Parker,
M.
E.
Roof,
R.
Sprenkel,
S.
D.
Stewart,
J.
R.
Weeks,
S.
Carroll,
M.
Parajulee,
P.
Roberts
and
J.
Ruberson,
2004.
Bacillus
thuringiensis
monitoring
program
for
tobacco
budworm
and
bollworm
in
2003.
Pages
1327­
1331
in
Proceedings
of
the
Beltwide
Cotton
Conferences,
San
Antonio,
Texas.
National
Cotton
Council.

Burd,
A.
D.,
2001.
The
influence
of
environmental
factors
on
susceptibility
of
B.
t.
cottons
to
bollworm,
Helicoverpa
zea,
and
factors
affecting
resistance
to
B.
t.
toxins
for
bollworm.
Raleigh,
North
Carolina
State
University.

Burd,
A.
D.,
J.
R.
Bradley,
Jr.,
J.
W.
Van
Duyn,
and
F.
Gould,
2001.
Estimated
frequency
of
nonrecessive
B.
t.
resistance
genes
in
bollworm,
Helicoverpa
zea.
Proceedings
of
the
Beltwide
Cotton
Conferences.
Vol
2:
820­
822.

Candas,
M.,
O.
Loseva,
B.
Oppert,
P.
Kosaraju,
and
L.
A.
Bulla,
2003.
Insect
resistance
to
Bacillus
thuringiensis
­
alterations
in
the
Indianmeal
moth
larval
gut
proteome.
Mol.
Cell.
Proteomics
2:
19­
28.

Caprio,
M.
A.,
1998a.
Two­
compartment
model
for
insect
resistance.
http://
www.
msstate.
edu/
Entomology/
PGjava/
ILSImodel.
html
[
December
1,
2005].

Caprio,
M.
A.,
1998b.
Evaluating
resistance
management
strategies
for
multiple
toxins
in
the
presence
of
external
refuges.
J
Econ
Entomol
91(
5):
1021­
1031.

Caprio,
M.
A.,
2006.
Draft
Report:
Modeling
the
risk
of
resistance
in
Heliothis
virescens
to
Bollgard
II
cotton.
Provided
by
Monsanto
Company,
April
7,
2006.

Caprio,
M.
A.,
M.
K.
Faver
and
G.
Hankins,
2004.
Evaluating
the
impacts
of
refuge
width
on
source­
sink
dynamics
between
transgenic
and
non­
transgenic
cotton.
J
Insect
Sci
4(
3):
1­
5.

Carrière,
Y.,
C.
Ellers­
Kirk,
Y.­
B.
Liu,
M.
A.
Sims,
A.
L.
Patin,
T.
J.
Dennehy
and
B.
E.
Tabashnik,
2001a.
Fitness
costs
and
maternal
effects
associated
with
resistance
to
transgenic
cotton
in
the
pink
bollworm
(
Lepidoptera:
Gelechiidae).
J
Econ
Entomol
94(
6):
1571­
1576.

Carrière,
Y.,
C.
Ellers­
Kirk,
A.
L.
Patin,
M.
A.
Sims,
S.
Meyer,
Y.­
b.
Liu,
T.
J.
Dennehy
and
B.
E.
Tabashnik,
2001b.
Overwintering
cost
associated
with
resistance
to
transgenic
cotton
in
the
pink
bollworm
(
Lepidoptera:
Gelechiidae).
J
Econ
Entomol
94(
4):
935­
941.
­
62­
Carrière,
Y.,
C.
Ellers­
Kirk,
R.
Biggs,
D.
M.
Higginson,
T.
J.
Dennehy
and
B.
E.
Tabashnik,
2004.
Effects
of
gossypol
on
fitness
costs
associated
with
resistance
to
Bt
cotton
in
pink
bollworm.
J
Econ
Entomol
97(
5):
1710­
1718.

Carrière,
Y.,
C.
Ellers­
Kirk,
R.
Biggs,
B.
Degain,
D.
Holley,
C.
Yafuso,
P.
Evans,
T.
J.
Dennehy
and
B.
E.
Tabashnik,
2005.
Effects
of
cotton
cultivar
on
fitness
costs
associated
with
resistance
of
pink
bollworm
(
Lepidoptera:
Gelechiidae)
to
Bt
cotton.
J
Econ
Entomol
98(
3):
947­
954.

Constable,
G.
A.,
D.
J.
Llewellyn
and
P.
E.
Reid,
1998.
Biotechnology
risks
and
benefits:
the
Ingard
®
cotton
example.
http://
www.
csu.
edu.
au/
special/
agronomy/
papers
[
December
1,
2005].

Doane
Agricultural
Services,
2004.
U.
S.
farm
seed
corn
study
report.

Ferré,
J.
and
J.
VanRie,
2002.
Biochemistry
and
genetics
of
insect
resistance
to
Bacillus
thuringiensis.
Annu.
Rev.
Entomol
47:
501­
533.

Gould,
F.,
1998.
Sustainability
of
transgenic
insecticidal
cultivars:
integrating
pest
genetics
and
ecology.
Annu
Rev
Entomol
43(
1):
701­
726.

Gould,
F.,
A.
Anderson,
A.
Jones,
D.
Sumerford,
D.
G.
Heckel,
J.
Lopez,
S.
Micinski,
R.
Leonard
and
M.
Laster,
1997.
Initial
frequency
of
alleles
for
resistance
to
Bacillus
thuringiensis
toxins
in
field
populations
of
Heliothis
virescens.
PNAS
94(
8):
3519­
3523.

Gould,
F.,
N.
Blair,
M.
Reid,
T.
L.
Rennie,
J.
Lopez
and
S.
Micinski,
2002.
Bacillus
thuringiensis­
toxin
resistance
management:
stable
isotope
assessment
of
alternate
host
use
by
Helicoverpa
zea.
PNAS
99(
26):
16581­
16586.

Greenplate,
J.
T.,
2004.
Report
on
studies
to
assess
supplemental
pyrethroid
spray
effects
on
Helicoverpa
zea
populations
in
Bollgard
®
cotton.
Monsanto
Company,
Report
number
MSL­
19252,
MRID
46222402.
St.
Louis,
Missouri.

Greenplate,
J.
T.,
J.
W.
Mullins,
S.
R.
Penn,
A.
Dahm,
B.
J.
Reich,
J.
A.
Osborn,
P.
R.
Rahn,
L.
Ruschke
and
Z.
W.
Shappley,
2003.
Partial
characterization
of
cotton
plants
expressing
two
toxin
proteins
from
Bacillus
thuringiensis:
relative
toxin
contribution,
toxin
interaction,
and
resistance
management.
J
Appl
Entomol
127:
340­
347.

Gunning,
R.
V.,
H.
T.
Dang,
F.
C.
Kemp,
I.
C.
Nicholson,
and
G.
D.
Moores,
2005.
New
resistance
mechanism
in
Helicoverpa
armigera
threatens
transgenic
crops
expressing
Bacillus
thuringiensis.
Appl.
and
Env.
Microbio.
71:
2558­
2563.

Gustafson,
D.
and
G.
Head,
2004.
Impact
of
updated
alternative
host
and
pyrethroid
effectivness
data
on
model
predictions
of
cotton
bollworm
resistance
development
to
Bollgard
®
cotton.
Monsanto
Company,
Report
number
MSL­
19326.
St.
Louis,
Missiouri.
­
63­
Gustafson,
D.,
G.
Head
and
M.
A.
Caprio,
2005.
Modeling
the
impact
of
alternative
hosts
and
pyrethroid
sprays
on
Helicoverpa
zea
(
Boddie)
adaptation
to
Bollgard
®
cotton.
J
Econ
Entomol.
(
submitted).

Gustafson,
D.,
G.
Head
and
H.
K.
Reding,
2001.
Impact
of
effective
refuge
size
and
typical
insecticide
use
practices
on
model
predictions
of
years
to
resistance
of
tobacco
budworm
and
cotton
bollworm
to
Bollgard
®
cotton.
Monsanto
Company,
Report
number
MSL­
19229,
MRID#
462224­
03.
St.
Louis,
Missouri.

Head,
G.,
R.
E.
Jackson,
J.
R.
Bradley,
J.
W.
Van
Duyn,
B.
R.
Leonard,
R.
Luttrell,
J.
Ruberson,
J.
J.
Adamczyk,
J.
Gore,
D.
Hardee,
R.
Voth,
S.
Sivasupramaniam
and
J.
W.
Mullins,
in
draft.
Regional
host
utilization
by
Helicoverpa
zea
as
measured
by
stable
carbon
isotope
analyses.

Head,
G.
and
R.
Voth,
2004.
A
final
report
on
studies
to
assess
production
of
Helicoverpa
zea
from
alternate
host
plants
and
from
the
external
unsprayed
non­
Bt
cotton
refuge
for
Bollgard
®
cotton.
Monsanto
Company,
Report
number
MSL­
19238,
MRID
46222401.
St.
Louis,
Missouri.

Hurley,
T.
M.
2000.
Research
report:
A
Bioceconomic
evluation
of
the
gradual
introduction
of
multiple
toxin
Bt
corn.
Prepared
for
Monsanto
Company.
February
2000.

Jackson,
R.
E.,
J.
R.
Bradley,
F.
Gould
and
J.
W.
Van
Duyn,
2003.
B.
t.
resistance
evolution
in
the
Helicoverpa
zea
population
in
eastern
North
Carolina.
Pages
1168­
1176
in
Proceedings
of
the
Beltwide
Cotton
Conferences,
Nashville,
Tennessee.
National
Cotton
Council.

Jackson,
R.
E.,
J.
R.
Bradley
and
J.
W.
Van
Duyn.
2005a.
Comparative
efficacy
of
Bt
technologies
against
bollworm
in
North
Carolina.
Pages
1373­
1378
in
Proceedings
of
the
Beltwide
Cotton
Conferences,
New
Orleans,
Louisiana.
National
Cotton
Council.

Jackson,
R.
E.,
J.
R.
Bradley,
J.
W.
Van
Duyn,
B.
R.
Leonard,
R.
Luttrell,
J.
Ruberson,
J.
J.
Adamczyk,
J.
Gore,
D.
Hardee,
R.
Voth,
S.
Sivasupramaniam,
J.
W.
Mullins
and
G.
Head,
2005b.
Regional
assessment
of
Helicoverpa
zea
(
Lepidoptera:
Noctuidae)
populations
on
cotton
and
non­
cotton
crop
hosts.
J
Econ
Entomol
(
submitted).

Jurat­
Fuentes,
J.
L.
and
M.
L.
Adang,
2001.
Importance
of
Cry1
 ­
endotoxin
domain
II
loops
for
binding
specificity
in
Heliothis
virescens
(
L.).
Appl.
and
Env.
Entomol.
67:
323­
329.

Jurat­
Fuentes,
J.
L.,
F.
L.
Gould,
and
M.
J.
Adang,
2003.
Dual
resistance
to
Bacillus
thuringiensis
Cry1Ac
and
Cry2Aa
toxins
in
Heliothis
virescens
suggests
multiple
mechanisms
of
resistance.
Appl
Environ
Microbiol
69
(
10):
5898­
5906.

Koziel,
M.
G.,
G.
L.
Beland,
C.
Bowman,
N.
B.
Carozzi,
R.
Crenshaw,
L.
Crossland,
J.
Dawson,
N.
Desai,
M.
Hill,
S.
Kadwell,
K.
Lewis,
D.
Maddox,
K.
McPherson,
M.
R.
Meghji,
E.
Merlink,
R.
Rhodes,
G.
W.
Warren,
M.
Wright
and
S.
V.
Evola,
1993.
Field
performace
of
an
elite
­
64­
transgenic
maize
plants
expressing
an
insecticidal
protein
derived
from
Bacillus
thuringiensis.
Biotechnology
11:
194­
200.

Kurtz,
R.
W.,
F.
Gould
and
J.
J.
R.
Bradley,
2004.
Helicoverpa
zea
fitness
on
Bt
corn
and
cotton
in
eastern
North
Carolina:
potential
effects
of
alternate
host
crops
and
pyramided
Bt
plants.
Pages
1430­
1434
in
Proceedings
of
the
Beltwide
Cotton
Conferences,
San
Antonio,
Texas.
National
Cotton
Council.

Livingston,
M.
J.,
G.
A.
Carlson
and
P.
L.
Fackler,
2004.
Managing
resistance
evolution
in
two
pests
to
two
toxins
with
refugia.
Am
J
Agr
Econ
86:
1­
13.

Luttrell,
R.,
L.
Wan
and
K.
Knighten,
1999.
Variation
in
susceptibility
of
noctuid
(
Lepidoptera)
larvae
attacking
cotton
and
soybean
to
purified
endotoxin
proteins
and
commercial
formulations
of
Bacillus
thuringiensis.
J
Econ
Entomol
92:
21­
32.

Mahon,
R.,
K.
Olsen,
S.
Young,
K.
Garsia,
L.
Lawrence,
2004.
Resistance
to
Bt
toxins
in
Helicoverpa
armigera.
The
Australian
Cotton
Grower.
December
2003­
January
2004.
p.
8­
10.

Martin,
S.
W.,
K.
Bryant,
K.
Paxton,
and
S.
Banerjee,
2006.
Preliminary
evaluation
of
Bt
refuges
based
on
on­
farm
observations.
Submitted
as
poster
for
2006
Beltwide
Cotton
Conferences,
Memphis,
Tennessee.

Onstad,
D.
W.,
D.
W.
Crowder,
P.
D.
Mitchell,
C.
A.
Guse,
J.
L.
Spencer,
E.
Levine
and
M.
E.
Gray,
2003.
Economics
versus
alleles:
balancing
integrated
pest
management
and
insect
resistance
management
for
rotation­
resistant
western
corn
rootworm
(
Coleoptera:
Chrysomelidae).
J
Econ
Entomol
96(
6):
1872­
1885.

Orth,
R.
G.,
G.
Head,
and
M.
Mierkowski,
in
draft.
Determining
larval
host
plant
use
by
a
polyphagous
lepidopteran
through
analysis
of
adult
moths
for
plant
secondary
metabolites.

Perlak,
F.
J.,
M.
Oppenhuizen,
K.
Gustafson,
R.
Voth,
S.
Sivasupramaniam,
D.
Heering,
B.
Carey,
R.
A.
Ihrig
and
J.
K.
Roberts,
2001.
Development
and
commercial
use
of
Bollgard
®
cotton
in
the
USA
­
early
promises
versus
today's
reality.
The
Plant
Journal
27(
6):
489­
501.

Rausher,
M.
D.,
2001.
Co­
evolution
and
plant
resistance
to
natural
enemies.
Nature
411(
6839):
857­
864.

Roush,
R.
T.,
1994.
Managing
pests
and
their
resistance
to
Bacillus
thuringiensis:
can
transgenic
crops
be
better
than
sprays?
Biocontrol
Sci
Techn
4:
501­
516.

Roush,
R.
T.,
1998.
Two­
toxin
strategies
for
management
of
insecticidal
transgenic
crops:
can
pyramiding
succeed
where
pesticide
mixtures
have
not?
Philos
T
Roy
Soc
353:
1777­
1786.

Schneider,
J.
C.,
2003.
Overwintering
of
Heliothis
virescens
(
F.)
and
Helicoverpa
zea
(
Boddie)
­
65­
(
Lepidoptera:
Noctuidae)
in
Cotton
Fields
of
Northeast
Mississippi.
J
Econ
Entomol
96
(
5):
1433­
1447.

Tabashnik,
B.
E.,
Y.
Carrière,
T.
J.
Dennehy,
S.
Morin,
M.
S.
Sisterson,
R.
T.
Roush,
A.
M.
Shelton
and
J.­
Z.
Zhao,
2003.
Insect
resistance
to
transgenic
Bt
crops:
lessons
from
the
laboratory
and
field.
J
Econ
Entomol
96(
4):
1031­
1038.

Tabashnik,
B.
E.,
F.
Gould
and
Y.
Carriere,
2004.
Delaying
evolution
of
insect
resistance
to
transgenic
crops
by
decreasing
dominance
and
heritability.
J
Evolution
Biol
17(
4):
904­
912.

USDA
NASS,
2005.
Quick
Stats.
United
States
Department
of
Agriculture.
http://
www.
nass.
usda.
gov/
QuickStats/
[
December
1,
2005].

Whitten,
M.
J.,
R.
A.
Jefferson
and
D.
Dall,
1996.
Biotechnology
and
integrated
pest
management
in
Biotechnology
in
Agriculture
Series.
15.
CAB
International,
Washington,
D.
C.

Zhao,
J.
Z.,
J.
Cao,
Y.
Li,
H.
L.
Collins,
R.
T.
Roush,
E.
D.
Earle,
and
A.
M.
Shelton,
2003.
Transgenic
plants
expressing
two
Bacillus
thuringiensis
toxins
delay
insect
resistance
evolution.
Nature
Biotechnology
Online:
http://
www.
nature.
com/
naturebiotechnoloy
.

Zhao,
J.­
Z.,
J.
Cao,
H.
L.
Collins,
S.
L.
Bates,
R.
T.
Roush,
E.
D.
Earle
and
A.
M.
Shelton,
2005.
Concurrent
use
of
transgenic
plants
expressing
a
single
and
two
Bacillus
thuringiensis
genes
speeds
insect
adaptation
to
pyramided
plants.
PNAS.
102:
8426­
8430.

EPA
References
BPPD,
2002a.
EPA
Review
of
Monsanto
Company's
Bollgard
II
Cotton
Insect
Resistance
Management
Plan
for
Section
3
Full
Commercial
Registration.
S.
Matten
memorandum
to
L.
Cole,
October
24,
2002.

BPPD,
2002b.
EPA
Review
of
Benefits
for
Monsanto
Company's
Bollgard
II
Cotton
For
Section
3
Full
Commercial
Registration.
E.
Brandt/
S.
Matten/
A.
Reynolds
memorandum
to
L.
Cole,
November
6,
2002.

BPPD,
2004a.
Technical
review
of
Monsanto's
submission:
"
A
Final
Report
on
Studies
to
Assess
Production
of
Helicoverpa
zea
from
Alternate
Host
Plants
and
from
the
External
Unsprayed
Non­
Bt
Cotton
Refuge
for
Bollgard
®
Cotton."
S.
Matten
memorandum
to
L.
Cole,
April
22,
2004.

BPPD,
2004b.
Technical
review
of
Monsanto's
submission:
"
Impact
of
Effective
refuge
size
and
typical
insecticide
use
practices
on
model
predictions
of
years
to
resistance
of
tobacco
budworm
and
cotton
bollworm
to
Bollgard
cotton."
S.
Matten
memorandum
to
L.
Cole
dated
April
22,
2004.
­
66­
BPPD,
2005.
Review
of
2004
monitoring
data
submitted
by
Monsanto
for
Bt
cotton
(
Bollgard
and
Bollgard
II).
A.
Reynolds
memorandum
to
L.
Cole,
December
1,
2005.

EPA,
1998.
Bacillus
thuringiensis
(
B.
t.)
plant­
pesticides
and
resistance
managment.
United
States
Environmental
Protection
Agency,
Report
number
EPA
738­
F­
98­
001.

EPA,
2001.
Bt
plant­
incorporated
protectants
October
15,
2001
biopesticides
registration
action
document.
Available
at
http://
www.
epa.
gov/
pesticides/
biopesticides.

SAP,
2004.
Meeting
Minutes
of
the
FIFRA
Scientific
Advisory
Panel
Meeting,
June
8­
10,
2004.
A
Set
of
Scientific
Issues
Being
Considered
by
the
U.
S.
Environmental
Protection
Agency
Regarding:
Product
Characterization,
Human
Health
Risk,
Ecological
Risk,
and
Insect
Resistance
Management
for
Bacillus
thuringiensis
(
Bt)
Cotton
Products.
United
States
Environmental
Protection
Agency,
Washington,
D.
C.
http://
www.
epa.
gov/
scipoly/
sap/
2004/
june/
final1a.
pdf.
