Page
1
of
85
August
20,
2004
MEMORANDUM
SUBJECT:
Transmittal
of
Minutes
of
the
FIFRA
Scientific
Advisory
Panel
Meeting
Held
June
8­
10,
2004:
Product
Characterization,
Human
Health
Risk,
Ecological
Risk,
And
Insect
Resistance
Management
For
Bacillus
thuringiensis
(
Bt)
Cotton
Products
TO:
James
J.
Jones,
Director
Office
of
Pesticide
Programs
FROM:
Paul
I.
Lewis,
Designated
Federal
Official
FIFRA
Scientific
Advisory
Panel
Office
of
Science
Coordination
and
Policy
THRU:
Larry
C.
Dorsey,
Executive
Secretary
FIFRA
Scientific
Advisory
Panel
Office
of
Science
Coordination
and
Policy
Joseph
J.
Merenda,
Jr.,
Director
Office
of
Science
Coordination
and
Policy
Please
find
attached
the
minutes
of
the
FIFRA
Scientific
Advisory
Panel
open
meeting
held
in
Arlington,
Virginia
from
June
8­
10,
2004.
These
meeting
minutes
address
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
Attachment
Page
2
of
85
cc:

Susan
Hazen
Margaret
Schneider
Adam
Sharp
Anne
Lindsay
Janet
Andersen
Debbie
Edwards
Steven
Bradbury
William
Diamond
Arnold
Layne
Tina
Levine
Lois
Rossi
Frank
Sanders
Margaret
Stasikowski
William
Jordan
Douglas
Parsons
Karen
Chu
Dayton
Eckerson
Enesta
Jones
Vanessa
Vu
(
SAB)
Leonard
Cole
John
Kough
Zigfridas
Vaituzis
Sharlene
Matten
OPP
Docket
Page
3
of
85
FIFRA
SAP
Members
Gary
Isom,
Ph.
D.
Janice
Elaine
Chambers,
Ph.
D.
H.
Christopher
Frey,
Ph.
D.
Kenneth
Portier,
Ph.
D.
Mary
Anna
Thrall,
D.
V.
M.

FQPA
Science
Review
Board
Members
Jonathan
A.
Arias,
Ph.
D.
Bruce
Chassy,
Ph.
D.
Steven
Gendel,
Ph.
D.
Dr.
Alexander
Haslberger
Barrie
Kitto,
Ph.
D.
Luke
Masson,
Ph.
D.
George
Cobb,
Ph.
D.
Paul
Jepson,
Ph.
D.
Thomas
La
Point,
Ph.
D.
Steven
Naranjo,
Ph.
D.
Michael
C.
Newman,
Ph.
D.
Sean
Richards,
Ph.
D.
Janice
Thies,
Ph.
D.
Michael
Caprio,
Ph.
D.
Steven
Castle,
Ph.
D.
Jesus
Esquivel,
Ph.
D.
Steven
L.
Peck,
Ph.
D.
Richard
Roush,
Ph.
D.
John
Schneider,
Ph.
D.
Douglas
Sumerford,
Ph.
D.
Mark
Whalon,
Ph.
D.
Page
4
of
85
SAP
Report
No.
2004­
05
MEETING
MINUTES
FIFRA
Scientific
Advisory
Panel
Meeting,
June
8­
10,
2004,
held
at
the
Holiday
Inn
Arlington
,
Arlington,
Virginia
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
Page
5
of
85
NOTICE
These
meeting
minutes
have
been
written
as
part
of
the
activities
of
the
Federal
Insecticide,
Fungicide,
and
Rodenticide
Act
(
FIFRA),
Scientific
Advisory
Panel
(
SAP).
This
report
has
not
been
reviewed
for
approval
by
the
United
States
Environmental
Protection
Agency
(
Agency)
and,
hence,
the
contents
of
this
report
do
not
necessarily
represent
the
views
and
policies
of
the
Agency,
nor
of
other
agencies
in
the
Executive
Branch
of
the
Federal
government,
nor
does
mention
of
trade
names
or
commercial
products
constitute
a
recommendation
for
use.

The
FIFRA
SAP
was
established
under
the
provisions
of
FIFRA,
as
amended
by
the
Food
Quality
Protection
Act
(
FQPA)
of
1996,
to
provide
advice,
information,
and
recommendations
to
the
Agency
Administrator
on
pesticides
and
pesticide­
related
issues
regarding
the
impact
of
regulatory
actions
on
health
and
the
environment.
The
Panel
serves
as
the
primary
scientific
peer
review
mechanism
of
the
EPA,
Office
of
Pesticide
Programs
(
OPP)
and
is
structured
to
provide
balanced
expert
assessment
of
pesticide
and
pesticide­
related
matters
facing
the
Agency.
Food
Quality
Protection
Act
Science
Review
Board
members
serve
the
FIFRA
SAP
on
an
ad
hoc
basis
to
assist
in
reviews
conducted
by
the
FIFRA
SAP.
Further
information
about
FIFRA
SAP
reports
and
activities
can
be
obtained
from
its
website
at
http://
www.
epa.
gov/
scipoly/
sap/
or
the
OPP
Docket
at
(
703)
305­
5805.
Interested
persons
are
invited
to
contact
Paul
Lewis,
Designated
Federal
Official,
via
e­
mail
at
lewis.
paul@
epa.
gov.

In
preparing
these
meeting
minutes,
the
Panel
carefully
considered
all
information
provided
and
presented
by
the
Agency
presenters,
as
well
as
information
presented
by
public
commenters.
This
document
addresses
the
information
provided
and
presented
within
the
structure
of
the
charge
by
the
Agency.
Page
6
of
85
SAP
Report
No.
2004­
05
MEETING
MINUTES:

FIFRA
Scientific
Advisory
Panel
Meeting,
June
8­
10,
2004,
held
at
the
Holiday
Inn
Arlington,
Arlington,
Virginia
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
Mr.
Paul
Lewis
Gary
Isom,
Ph.
D.
Designated
Federal
Official
FIFRA
SAP
Session
Chair
FIFRA
Scientific
Advisory
Panel
FIFRA
Scientific
Advisory
Panel
Date:
August
17,
2004
Date:
August
17,
2004
Page
7
of
85
Federal
Insecticide,
Fungicide,
and
Rodenticide
Act
Scientific
Advisory
Panel
Meeting
June
8­
10,
2004
Product
Characterization,
Human
Health
Risk,
Ecological
Risk,
And
Insect
Resistance
Management
For
Bacillus
thuringiensis
(
Bt)
Cotton
Products
PARTICIPANTS
FIFRA
SAP
Session
Chair
Gary
Isom,
Ph.
D.,
Professor
of
Toxicology,
School
of
Pharmacy
&
Pharmacal
Sciences,
Purdue
University,
West
Lafayette,
IN
FIFRA
Scientific
Advisory
Panel
Members
Janice
Elaine
Chambers,
Ph.
D.,
William
L.
Giles
Distinguished
Professor
and
Director,
Center
for
Environmental
Health
Sciences,
College
of
Veterinary
Medicine,
Mississippi
State
University,
Mississippi
State,
MS
H.
Christopher
Frey,
Ph.
D.,
Associate
Professor,
Civil
Engineering,
North
Carolina
State
University,
Raleigh,
NC
Kenneth
Portier,
Ph.
D.,
Associate
Professor,
Statistics,
Institute
of
Food
and
Agricultural
Sciences,
University
of
Florida,
Gainesville,
FL
Mary
Anna
Thrall,
D.
V.
M.,
Professor,
Microbiology,
Immunology
&
Pathology,
Colorado
State
University,
College
of
Veterinary
Medicine
and
Biomedical
Sciences,
Fort
Collins,
CO
FQPA
Science
Review
Board
Members
Jonathan
A.
Arias,
Ph.
D.,
Associate
Research
Scientist,
Biological
Sciences,
University
of
Maryland,
Baltimore
County,
Baltimore,
MD
Michael
Caprio,
Ph.
D.,
Associate
Professor,
Insect
Genetics,
Entomology
&
Plant
Pathology,
Mississippi
State
University,
Mississippi
State,
MS
Steven
Castle,
Ph.
D.,
Research
Entomologist,
USDA/
ARS,
Western
Cotton
Research
Laboratory,
Phoenix,
AZ
Bruce
Chassy,
Ph.
D.,
Associate
Executive
Director,
Campus
Biotechnology
Center
and
Professor,
Department
of
Food
Science
and
Human
Nutrition,
University
of
Illinois,
Urbana,
IL
George
Cobb,
Ph.
D.,
Associate
Professor,
Environmental
Toxicology,
Texas
Tech
University,
Lubbock,
TX
Page
8
of
85
Jesus
Esquivel,
Ph.
D.,
Research
Entomologist,
Areawide
Pest
Management
Research
Unit,
USDA/
ARS­
SPARC,
College
Station,
TX
Steven
Gendel,
Ph.
D.,
Branch
Chief,
Food
and
Drug
Administration,
National
Center
for
Food
Safety
and
Technology,
Summit­
Argo,
IL
Paul
Jepson,
Ph.
D.,
Director,
Integrated
Plant
Protection
Center,
Oregon
State
University,
Corvallis,
OR
Dr.
Alexander
Haslberger,
Professor
of
Microbiology,
Institute
of
Microbiology
and
Genetics,
University
of
Vienna,
Vienna,
Austria
Barrie
Kitto,
Ph.
D.,
Professor
of
Chemistry
and
Biochemistry,
The
University
of
Texas
at
Austin,
Austin,
TX
Thomas
La
Point,
Ph.
D.,
Professor
and
Director,
Biological
Sciences
and
Institute
of
Applied
Sciences,
University
of
North
Texas,
Denton,
TX
Luke
Masson,
Ph.
D.,
Senior
Research
Scientist,
National
Research
Council
of
Canada,
Biotechnology
Research
Institute,
Montreal,
Quebec,
Canada
Steven
Naranjo,
Ph.
D.,
Research
Entomologist,
USDA/
ARS,
Western
Cotton
Research
Laboratory,
Phoenix,
AZ
Michael
C.
Newman,
Ph.
D.,
Professor
of
Marine
Science,
Virginia
Institute
of
Marine
Science,
College
of
William
&
Mary,
Gloucester
Point,
VA
Steven
L.
Peck,
Ph.
D.,
Assistant
Professor,
Department
of
Integrative
Biology,
Brigham
Young
University,
Provo,
UT
Sean
Richards,
Ph.
D.,
UC
Foundation
Assistant
Professor,
Department
of
Biological
and
Environmental
Sciences,
University
of
Tennessee
at
Chattanooga,
Chattanooga,
TN
Richard
Roush,
Ph.
D.,
Director,
Statewide
IPM
Program,
University
of
California,
Davis,
CA
John
Schneider,
Ph.
D.,
Professor
and
Research
Entomologist,
Department
of
Entomology
and
Plant
Pathology,
Mississippi
State
University,
Mississippi
State,
MS
Douglas
Sumerford,
Ph.
D.,
Research
Entomologist,
Corn
Insects
and
Crop
Genetics
Research
Unit,
Genetics
Laboratory
c/
o
Insectary,
USDA/
ARS,
Iowa
State
University,
Ames,
IA
Janice
Thies,
Ph.
D.,
Associate
Professor,
Soil
Biology,
Department
of
Crop
and
Soil
Science,
Cornell
University,
Ithaca,
NY
Page
9
of
85
Mark
Whalon,
Ph.
D.,
Professor,
Center
for
Integrated
Plant
Systems,
Michigan
State
University,
East
Lansing,
MI
PUBLIC
COMMENTERS
Oral
statements
were
made
by:
Rod
Herman,
Ph.
D.,
Dow
Agrosciences
Monte
Mayes,
Ph.
D.,
Dow
Agrosciences
Larry
Sernyk,
Ph.
D.,
Dow
Agrosciences
Nick
Storer,
Ph.
D.,
Dow
Agrosciences
Laura
Tagliani,
Ph.
D.,
Dow
Agrosciences
Ray
Layton,
Ph.
D.,
Dupont
Company,
on
behalf
of
the
Agricultural
Biotechnology
Stewardship
Technical
Committee,
Non­
target
Organism
Subcommittee
Jane
Rissler,
Ph.
D.,
Union
of
Concerned
Scientists
Graham
Head,
Ph.
D.,
Monsanto
Written
statements
were
received
from:
National
Cotton
Council
Monsanto
JR
Bradley,
North
Carolina
State
University
INTRODUCTION
The
Federal
Insecticide,
Fungicide,
and
Rodenticide
Act
(
FIFRA),
Scientific
Advisory
Panel
(
SAP)
has
completed
its
review
of
the
set
of
scientific
issues
being
considered
by
the
Agency
pertaining
to
its
review
of
product
characterization,
human
health
risk,
ecological
risk,
and
insect
resistance
management
for
Bacillus
thuringiensis
(
Bt)
cotton
products.
Advance
notice
of
the
meeting
was
published
in
the
Federal
Register
on
May
12,
2004.
The
review
was
conducted
in
an
open
Panel
meeting
held
in
Arlington,
Virginia,
from
June
8­
10,
2004.
The
meeting
was
chaired
by
Gary
Isom,
Ph.
D.
Mr.
Paul
Lewis
served
as
the
Designated
Federal
Official.
Mr.
Joseph
J.
Merenda,
Jr.
(
Director,
Office
of
Science
Coordination
and
Policy,
EPA)
and
Janet
Andersen,
Ph.
D.
(
Director,
Biopesticides
and
Pollution
Prevention
Division,
Office
of
Pesticide
Programs,
EPA)
offered
opening
remarks
at
the
meeting.
Mr.
Leonard
Cole
(
Office
of
Pesticide
Programs,
EPA)
provided
an
introduction
and
highlighted
the
goals
and
objectives
of
the
meeting.
The
Agency's
product
characterization
and
human
health
safety
assessment
for
stacked
plant­
incorporated
protectants,
environmental
effects
assessment
for
WideStrike
cotton,
and
issues
related
to
establishing
an
insect
resistance
management
plan
for
WideStrike
cotton
were
presented
by
John
Kough,
Ph.
D.
(
Office
of
Pesticide
Programs,
EPA),
Zigfridas
Vaituzis,
Ph.
D.
(
Office
of
Pesticide
Programs,
EPA)
and
Sharlene
Matten,
Ph.
D.
(
Office
of
Pesticide
Programs,
EPA),
respectively.
Sharlene
Matten,
Ph.
D.
(
Office
of
Pesticide
Programs,
EPA)
concluded
the
Agency's
presentations
by
discussing
Bollgard
and
Bollgard
II
cotton
bollworm
insect
resistance
management.
Page
10
of
85
SUMMARY
OF
PANEL
DISCUSSION
AND
RECOMMENDATIONS
Human
Health
Risk
The
Panel
concluded
that
toxicity
studies
showed
no
adverse
effects
at
the
highest
levels
tested,
either
for
the
individual
Cry
proteins
or
for
the
two
proteins
together.
The
Panel
agreed
with
the
Agency
that
combined
oral
toxicity
studies
for
multiple
proteins
is
not
necessary
if
each
protein
had
been
previously
independently
tested,
unless
such
testing
would
be
intended
to
detect
any
effects
produced
by
interactions
between
the
proteins.

Based
on
the
evidence
presented
in
the
safety
assessment
of
the
single
Bt
containing
strain
and
the
more
recent
extensive
testing
of
the
stacked
Cry1F/
Cry1Ac
material,
the
Panel
concluded
that
there
is
little
to
suggest
that
the
stacked
variety
poses
significantly
more
risk
or
unanticipated
consequences
to
agronomic
performance
than
do
varieties
with
either
the
Cry1F
or
Cry1Ac
genes
introduced
singly.

Ecological
Risk
Direct
hazards
to
vertebrates
due
to
exposure
to
the
Cry1F
and
Cry1Ac
proteins
should
be
minimal
or
non
existent,
making
consideration
of
synergy
in
vertebrates
unwarranted.
Thus,
the
Panel
concluded
that
toxicity
testing
on
the
combinations
of
Cry
proteins
is
not
necessary.
While
synergism
is
not
considered
to
be
important
for
the
Cry1
proteins
under
consideration,
the
Panel
believed
that
future
testing
of
non­
target
effects
of
these
types
of
toxins
should
proceed,
with
both
toxins
to
be
expressed
by
genetically
engineered
plants
rather
than
with
individual
purified
toxins.
Testing
mixtures
of
compounds
would
provide
data
that
are
more
relevant
to
the
proposed
use
scenarios
and
likely
exposure
scenarios.

The
Panel
recognized
improvements
in
the
quality
of
the
Agency's
analysis
and
the
degree
to
which
this
reflected
a
positive
response
by
the
Agency
to
recommendations
made
by
previous
SAP
reviews
of
plant­
incorporated
protectants
(
PIPs).
The
Panel
also
noted
that
the
Tier
I
tests
were
generally
of
a
higher
quality
than
the
tests
that
other
SAPs
have
reviewed.
Overall,
however,
detailed
analysis
of
the
field­
based
evaluations
revealed
that
these
did
not
meet
current
standards,
and
the
Panel
repeated
requests
made
by
other
SAPs,
that
the
Agency
issue
detailed
guidelines
for
semi­
field,
and
field­
based
procedures.
The
Panel
recommended
that
the
exposure
pathway
and
diet
used
within
the
Chrysoperla
test
be
examined
further,
to
determine
the
rigor
and
repeatability
of
the
test,
and
also
suggested
that
Orius
spp.
be
considered
as
a
more
appropriate
test
subject.
The
Panel
also
considered
that
a
more
appropriate
parasitoid
than
Nasonia
should
be
selected.

The
Panel
agreed
that
multi­
year
field
studies
to
assess
potential
longer­
term
effects
of
WideStrike
cultivation
on
persistence
of
toxins
in
the
soil
and
on
populations
of
non­
target
organisms
are
applicable
to
Bt
cotton
and
should
be
considered.
Overall,
the
likely
reduction
in
the
use
of
broader­
spectrum
insecticides
afforded
by
the
cultivation
of
Bt
cotton
in
general
is
likely
to
have
positive
effects
on
the
ecosystems.
Nonetheless,
longer­
term
and
broader­
scale
Page
11
of
85
evaluation
of
PIP
crops
will
be
necessary
for
improved
ecological
understanding.
The
Panel
believed
strongly
that
long­
term
field
studies
should
not
proceed
without
some
guidance
relative
to
experimental
protocols
and
clearly
defined
endpoints.
Without
such
guidance,
we
are
unlikely
to
resolve
any
unexpected
detriments
or
benefits
associated
with
the
use
of
transgenic
crops.

WideStrike
Cotton
Insect
Resistance
Management
The
combined
expression
of
Bt
proteins
in
WideStrike
cotton
meets
the
Agency's
definitions
of
high
dose
for
pink
bollworm
(
PBW)
and
tobacco
budworm
(
TBW).
In
addition,
reasonable
doses
of
the
combined
protein
were
evident
for
control
of
cotton
bollworm
(
CBW).
Based
on
the
high
dose
evidence,
the
Panel
concluded
that
it
is
valid
to
assume
that
resistance
occurring
in
PBW
or
TBW
will
likely
be
inherited
as
a
recessive
trait.
However,
CBW
is
more
tolerant
of
both
proteins
and
it
seems
possible
that
resistance
will
be
less
recessive.
WideStrike
cotton
does
appear
to
offer
a
high
dose
for
TBW,
a
high
dose
of
Cry1Ac
for
PBW,
and
reasonable
doses
of
Cry1F
and
Cry1Ac
for
CBW.
The
same
high
dose/
refuge
strategy
practiced
thus
far
as
a
resistance
management
approach
for
Bollgard
cotton
should
be
applied
for
WideStrike.

While
the
Panel
supported
the
Agency's
conclusion
that
incomplete
shared
binding
of
Cry1Ac
and
Cry1F
receptors
in
TBW
and
CBW
is
expected
to
lead
to
incomplete
crossresistance
differences
were
expressed
on
the
molecular
mechanism
involved
in
the
process.
In
addition
the
Panel
raised
the
issue
that
another
as
yet
unidentified
major
resistance
mechanism
may
not
occur.
The
Panel
agreed
with
the
Agency
that
there
is
no
basis
to
believe
that
the
occurrence
of
resistance
in
the
field
will
be
due
to
a
mechanism
other
than
binding
site
modification.

The
Panel
identified
several
areas
of
concern
with
the
Dow
Agrosciences
CBW
model
that
make
its
use
problematic.
These
problems
must
be
addressed
if
this
model
is
to
be
used
to
assess
the
durability
of
WideStrike
cotton.
The
Panel
believed
that
use
of
the
current
model,
once
corrected
of
the
identified
errors,
would
be
an
appropriate
vehicle
to
explore
the
parameter
space
with
the
goal
of
finding
areas
where
resistance
does
occur
in
the
15
year
time
horizon
and
assessing
whether
it
occurs
within
biologically
plausible
initial
conditions
and
parameter
values.

Since
the
dose
of
the
Cry1Ac
and
Cry1F
in
WideStrike
Cotton
was
demonstrated
to
be
high
against
populations
of
TBW,
the
Panel
believed
that
WideStrike
will
be
more
durable
than
that
predicted
by
Peck
(
1999)
for
single
Cry1Ac
cotton.

The
Panel
agreed
that
the
HOSTS
database
is
insufficient
to
address
the
issue
of
CBW
alternate
hosts
as
natural
refugia.
The
Panel
agreed
that
there
are
insufficient
empirical
data
in
the
registrant
report
to
demonstrate
that
alternative
hosts
are
producing
susceptible,
fit
individuals
in
sufficient
quantity,
at
the
correct
time
and
proximity
to
maximize
the
probability
of
matings
between
homozygous­
susceptible
individuals
and
individuals
heterozygous
for
resistant
traits.

Even
though
the
Panel
raised
limitations
with
the
model
as
described,
the
Panel
was
in
Page
12
of
85
strong
agreement
that
the
proposed
IRM
plan
by
the
registrant
is
sufficient
for
WideStrike
cotton
and
supported
the
prediction
of
a
delay
in
resistance
of
TBW,
CBW
and
PBW
to
WideStrike
cotton
for
15
years.
The
overall
consensus
was
that
the
existing
IRM
options
that
have
been
applied
to
the
single­
toxin
Bollgard
cotton
will
be
equally
or
even
more
effective
in
protecting
against
resistance
in
the
double­
toxin
WideStrike
cotton.

Bollgard
Insect
Resistance
Management
The
Panel
could
not
determine
whether
CBW
reverse
migration
is
expected
to
have
any
impact
on
CBW
adaptation
to
Bt
cotton
or
Bt
corn.

While
the
Panel
agreed
that
pyrethroid
oversprays
in
Bollgard
cotton
improve
the
control
of
susceptible
corn
earworm
(
CEW),
the
effect
of
pyrethroid
oversprays
in
delaying
resistance
in
CEW
is
probably
overstated.
The
Panel
agreed
that
there
is
little
need
to
include
pyrethroid
oversprays
in
Bollgard
II
plots
in
the
models
of
Gustafson
et
al.
There
is
some
evidence
of
greater
tolerance
in
larvae
originating
from
Bollgard
fields
relative
to
those
coming
from
non­
Bollgard
fields.

The
Panel
agreed
that
sufficient
data
were
provided
to
establish
that
C3
and
C4
alternate
hosts
function
to
some
degree
as
unstructured
refugia.
However,
the
Panel
expressed
concern
on
the
methodologies
used
to
assess
adult
productivity
in
the
alternate
hosts.
The
Panel
agreed
that
CBW
production
should
be
measured
at
a
larger
scale
than
the
local
farm,
or
field
level
because
of
the
high
mobility
of
adult
CBW.
In
response
to
the
request
for
methods
on
quantitatively
calculating
"
effective
refuge
size,"
the
Panel
provided
techniques
for
quantifying
CBW
populations
in
the
identified
alternate
hosts
that
were
identified
as
natural
refugia.

The
Panel
agreed
with
the
Agency
that
a
weighted
average
is
an
appropriate
choice
for
determining
the
contribution
of
alternate
hosts
to
the
refuge
size.
The
Panel
believed,
however,
that
exploring
detailed
questions
about
time
to
resistance
and
the
effect
of
alternate
hosts
on
resistance
would
benefit
from
the
development
of
a
more
detailed
model.

PANEL
DELIBERATIONS
AND
RESPONSE
TO
THE
CHARGE
The
specific
issues
to
be
addressed
by
the
Panel
are
keyed
to
the
Agency's
background
documents,
references
and
Agency's
charge
questions.

Agency
Charge
Human
Health
Risk
1.
The
Agency
examines
the
safety
of
proteins
based
on
the
source
of
the
protein,
the
protein's
pesticidal
mode
of
action,
comparisons
of
the
amino
acid
sequence
to
toxins
and
allergens
and
the
results
of
acute
oral
toxicity
testing.
The
company
provided
numerous
mammalian
oral
toxicity
studies
to
demonstrate
the
safety
of
the
introduced
Cry
1Ac
and
Page
13
of
85
Cry1F
protein
insecticidal
toxins.
The
toxins
were
tested
both
separately
and
in
combination.
The
Agency
believes
tests
with
combinations
of
pure
proteins
may
address
possible
synergistic
interactions
between
introduced
proteins.
However,
the
Agency
believes
that
unless
there
is
an
indication
that
the
two
proteins
would
interact,
such
as
being
parts
of
a
binary
toxin
or
attaching
to
the
same
receptor,
there
is
little
to
justify
testing
the
two
proteins
together
when
separate
oral
toxicity
tests
indicate
a
lack
of
toxicity
for
the
individual
proteins.

Does
the
Panel
have
additional
comments
on
this
position
including
identifying
instances
where
it
would
be
justified
to
require
the
toxicity
testing
of
two
proteins
in
combination?

Panel
Response
The
Panel's
comments
to
this
question
are
specific
to
expressed
proteins.
Comments
on
unintended
effects
are
presented
by
the
Panel
in
their
response
to
question
2.
The
relevant
human
health
issue
is
related
to
consumption
of
these
proteins.
A
search
of
the
literature
regarding
these
substances
failed
to
find
any
indication
that
there
are
any
other
significant
issues
directly
related
to
human
health.

The
Agency
has
been
asked
to
evaluate
a
line
of
cotton
containing
two
Bt
Cry
proteins
intended
for
insect
control.
This
line
was
produced
by
cross
breeding
two
lines
containing
independent
transformation
events.
The
pesticide
registrant
provided
data
characterizing
the
two
transformation
events,
showing
that
the
structures
of
the
inserts
were
not
altered
during
crossing,
and
demonstrating
stable
Mendelian
inheritance
of
each
insert.
In
terms
of
human
health,
the
pesticide
registrant
provided
sequence
analysis
data
for
all
the
proteins
involved
(
the
Cry
proteins
and
the
markers),
stability
data,
and
the
results
of
oral
toxicity
studies
using
mice.
The
Panel
concluded
that
these
toxicity
studies
showed
no
adverse
effects
at
the
highest
levels
tested,
either
for
the
individual
Cry
proteins
or
for
the
two
proteins
together.
It
should
be
noted
that
the
testing
was
done
with
material
that
was
highly
equivalent
(
but
not
identical)
to
the
proteins
expressed
in
plants.
In
addition,
because
cotton
proteins
make
only
a
very
low
contribution
to
the
human
diet,
exposure
to
these
proteins
is
expected
to
be
very
low.

The
fact
that
the
cotton
line
involved
expressed
two
insect
control
proteins
raised
the
question
of
whether,
or
when,
it
is
necessary
to
carry
out
combined
oral
toxicity
studies
for
multiple
proteins
if
each
protein
has
been
independently
tested
previously.
This
testing
would
be
intended
to
detect
any
effects
produced
by
interactions
between
the
proteins.
The
Agency
has
stated
the
belief
that
such
testing
is
not
justified
unless
there
is
specific
evidence
for
such
interactions.
In
general,
the
Panel
agreed
with
this
statement.

The
basic
principles
for
assessing
the
safety
of
transferred
proteins
in
bioengineered
plants
have
been
well
developed.
A
good
summary
of
these
principles
can
be
found
in
the
CODEX
document
"
Guideline
for
the
Conduct
of
Food
Safety
Assessment
of
Foods
Derived
from
Recombinant­
DNA
Plants."
In
general,
these
principles
suggest
that,
for
proteins,
safety
assessment
should
focus
on
source,
function,
similarity
to
known
toxins,
anti­
nutrients
and
Page
14
of
85
allergens,
and
stability.
This
guidance
suggested
that
"
appropriate
oral
toxicity
studies
may
need
to
be
carried
out
in
cases
where
the
protein
 
is
not
similar
to
proteins
have
that
previously
been
consumed
safely."
These
are
essentially
the
same
criteria
specified
by
the
Agency.

First,
previous
evaluations
of
potential
human
health
risks
associated
with
(
PIPs)
plantincorporated
protectant
proteins
may
be
considered
as
having
set
a
precedent
for
this
approach.
In
earlier
cases
where
only
a
single
PIP
protein
was
expressed
in
a
transformed
plant,
it
was
in
fact
accompanied
by
a
second
protein
that
was
not
active
insecticidally,
but
instead
served
as
a
marker
to
identify
a
transformation
event.
In
the
present
consideration
of
WideStrike
cotton,
synthetic
genes
coding
for
the
insecticidal
proteins
Cry1F
and
Cry1Ac
were
integrated
independently
into
separate
cotton
breeding
lines
and
accompanied,
in
both
cases,
by
the
pat
gene
encoding
for
the
marker
protein
phosphinothricin
acetyltransferase
(
PAT).
In
effect,
the
individual
transformation
events
for
each
PIP
protein
involved
expression
of
two
new
proteins
in
a
target
plant
 
the
marker
and
the
pesticidal
protein.
In
these
cases,
each
protein
was
evaluated
independently.
The
classical
breeding
of
the
two
transformed
cotton
lines
eventually
produced
WideStrike
cotton
that
expresses
the
pyramided
PIP
proteins
Cry1Ac
and
Cry1F
along
with
the
intact
marker
protein
PAT
and
an
incomplete
version
of
PAT.
Whereas
the
insert
from
the
Cry1Ac
event
(
3006­
210­
23)
contains
one
intact
copy
of
the
insect
resistance
gene
cry1Ac
and
one
intact
copy
of
the
plant
selectable
marker
gene
pat,
the
insert
from
the
Cry1F
event
(
281­
24­
236)
contains
one
intact
copy
of
the
insect
resistance
gene
Cry1Ac,
one
intact
copy
of
the
marker
gene
pat,
plus
an
additional
hybridizing
fragment
of
the
marker
gene
pat
(
Dow
AgroSciences
Study
010075.01).
The
presence
of
the
partial
pat
fragment
results
in
a
4th
product
expressed
in
WideStrike
cotton
that
has
a
potential
amino
acid
sequence
90%
identical
to
PAT
and
generates
the
same
homology
profile
as
the
full
length
PAT
sequence
(
Dow
AgroSciences
Study
ID:
GH­
C
5573).
Despite
its
truncated
resemblance
to
the
intact
PAT
protein
and
reduced
expression,
the
4th
product
must
still
be
considered
along
with
Cry1Ac,
Cry1F
and
PAT
for
potential
effects
on
human
health.
The
question
remains
whether
testing
of
two
or
more
proteins
together
is
necessary
when
separate
toxicity
tests
for
individual
proteins
indicate
a
lack
of
toxicity.
Since
the
Panel
agreed
with
the
Agency
that
testing
for
interactions
between
the
proteins
is
not
justified
unless
specific
evidence
points
to
the
contrary,
the
Panel
sees
no
basis
why,
if
independent
evaluation
is
considered
appropriate
for
two
proteins,
it
should
not
also
be
considered
appropriate
for
four
or
more
proteins.

Second,
although
there
are
many
examples
of
binary
(
or
multimeric)
protein
toxins,
the
interactions
between
the
proteins
involved
are
specific
among
members
of
a
toxin
family.
The
Panel
is
not
aware
of
any
instances
where
a
"
new"
toxin
has
been
created
by
unexpected
interaction
between
two
known
proteins.
Conversely,
it
seems
unlikely
that
two
proteins
that
interact
to
produce
a
biological
effect
would
independently
have
unexpected
activity,
except
in
well­
defined
situations
(
for
example
as
blockers
of
transport
or
digestion
pathways).
Again,
it
appears
unlikely
that
unexpected
interactions
will
occur
between
two
unrelated
proteins.
The
only
examples
would
be
if
physical
interactions
between
two
unstable
proteins
inhibited
gastric
digestion
or
increased
stability
during
processing.
Therefore,
it
is
reasonable
to
treat
each
protein
independently
in
situations
such
as
this.
It
is
possible
that
simple
biochemical
tests
could
be
used
to
determine
whether
such
physical
interactions
occur
between
independent
proteins,
in
which
Page
15
of
85
case
the
combination
can
be
targeted
for
digestion
studies.

Third,
the
principle
of
independent
assessment
has
been
used
for
many
years
for
food
additives
and
is
implicit
in
microbial
risk
assessments.
There
is
no
reason
to
believe
that
the
issues
involved
with
these
proteins
should
affect
the
use
of
this
principle.

Fourth,
the
relatively
low
expression
levels
for
the
proteins
involved
make
it
unlikely
that
they
would
be
able
to
interact
extensively
and
maintain
their
biological
function.
Unanticipated
interactions
seem
more
likely
to
result
in
sequestration
and
reduced
activity
for
the
proteins
involved.

Fifth,
given
that
each
of
these
proteins
has
previously
been
used
independently
in
food
plants,
the
Panel
concluded
that
they
would
fit
the
criteria
of
being
similar
to
proteins
that
have
previously
been
consumed
safely.
Therefore,
they
do
not
rise
to
the
level
of
concern
that
would
suggest
the
need
for
combined
toxicity
testing.

In
the
charge
to
the
Panel,
the
Agency
also
raised
the
possibility
that
having
two
proteins
attached
to
the
same
receptor
could
trigger
a
need
for
combined
toxicity
testing.
It
appears
that,
in
this
context,
this
means
binding
to
the
same
receptor
in
humans.
Any
proteins
that
have
known
biological
effects
in
humans
need
to
be
evaluated,
regardless
of
mechanism.

There
is
one
point
of
terminology
that
needs
clarification
by
the
Agency
­
the
definition
of
the
term
"
stacked"
and
"
pyramidal"
such
as
when
used
for
stacked
traits
or
stacked
proteins.
There
are
different
ways
to
"
stack"
or
"
pyramid"
traits
(
multiple
transformation
or
cross
breeding)
and
it
is
unclear
to
the
Panel
whether
the
Agency
considers
these
differences
to
be
significant.

Agency
Charge
2.
When
traits
are
introduced
into
crop
plants
using
the
transformation
techniques
of
modern
biotechnology
or
even
traditional
breeding,
one
of
the
areas
of
concern
is
the
possibility
of
unintentional
changes.
There
is
a
general
difficulty
in
screening
for
these
unforeseen
changes
since
it
is
a
conceptual
leap
to
anticipate
the
unexpected.
However,
in
general
the
approach
has
been
to
examine
general
performance
of
the
new
cultivars
like
agronomic
performance
and
compositional
analysis
to
detect
unintentional
effects.
PIP
products
can
be
both
transformed
lines
and
the
result
of
traditional
breeding
of
two
transformed
lines
to
yield
a
new
product
with
combined
traits
like
WideStrike
cotton.
In
both
cases,
the
new
PIP
product
must
be
registered
just
as
other
new
combinations
of
pesticide
active
ingredients
must
be
registered.

For
PIP
products
resulting
from
traditionally
bred
transformed
lines,
under
what
circumstances,
if
any,
would
it
be
appropriate
to
examine
agronomic
performance
and
compositional
analysis
to
provide
a
screen
for
unintentional
changes
in
the
crop?
Please
describe
other
ways
EPA
might
consider
screening
for
potential
unintentional
changes
in
a
crop.
Page
16
of
85
Panel
Response
Agronomic
Performance
and
Composition
Analysis
Based
on
the
evidence
presented
in
the
safety
assessment
of
the
single
Bt
containing
strain
and
the
more
recent
extensive
testing
of
the
stacked
Cry1F/
Cry1Ac
material,
the
Panel
concluded
that
there
is
little
to
suggest
that
the
stacked
variety
poses
significantly
more
risk
or
unanticipated
consequences
to
agronomic
performance
than
do
varieties
with
either
the
Cry1F
or
Cry1Ac
genes
introduced
singly.
The
Panel
was
unaware
of
any
reason
to
believe
that
previously
undetected
and
unanticipated
changes
would
have
occurred
in
the
plant
breeding
beyond
the
uncertainties
normally
associated
with
plant
breeding
and
the
original
transformation
process
in
parental
GM
lines.

The
food
safety
of
the
two
individual
Cry
proteins
present
in
WideStrike
has
already
been
established.
It
is
worth
noting
that
cotton
fiber
is
not
a
food
and
does
not
contain
residual
Cry
proteins.
Cotton
meal,
which
may
contain
small
amounts
of
Cry
proteins,
is
usually
only
fed
at
low
levels
to
ruminants
because
they
cannot
tolerate
larger
amounts
of
gossypol
which
is
relatively
toxic
to
non­
ruminant
animals.
Cottonseed
oil
is
a
bleached
and
purified
product
that
contains
only
trace
amounts
of
protein.
Cottonseed
oil
is
one
of
several
vegetable
oils
consumed
by
humans
and
thus
is
a
relatively
minor
dietary
component.

Composition,
phenotypic
traits
and
agronomic
performance
can
be
powerful
tools
for
revealing
unintended
effects
and
can
be
viewed
as
screening
methods
since
they
are
the
end
result
of
the
balanced
and
coordinated
expression
of
numerous
multigenic
traits.
Plant
breeders
normally
eliminate
from
further
development
plants
with
visible
or
compositional
defects
or
undesirable
alterations.
Compositional
or
agronomic
studies
are
normally
not
required
on
plants
resulting
from
traditional
breeding.

Suggested
Additional
Procedures
To
Screen
For
Potential
Unintentional
Changes
In
the
Crop
For
the
Panel's
response
to
the
question,
the
Panel
considered
the
screening
for
unintended
changes
to
WideStrike
and,
on
a
broader
perspective,
other
plant­
incorporated
protectants
(
PIP)
and
genetically
modified
crops.
Screening
for
unintended
effects
is
encouraged.
Such
a
screening
could
use
profiling
methods
for
unintended
effects,
e.
g.
expression
microarrays
or
metabolomic
studies
(
e.
g.,
as
discussed
in
SAFOTEST
(
http://
www.
entransfood.
nl/
RTDprojects/
SAFOTEST/
safotest.
html).
However,
as
these
tests
are
still
under
development,
toxicology
studies
in
rodents,
looking
for
effects
of
unintended
effects,
although
unlikely,
could
be
appropriate.
While
the
targeted
safety
assessment
system
in
place
today
has
apparently
been
effective
in
risk
characterization,
it
cannot
eliminate
the
chance
that
unintended
or
unexpected
consequences
of
plant
breeding
will
escape
detection.
It
would
be
extremely
useful
to
have
comparative
data
regarding
the
frequency
and
magnitude
of
unintended
effects
associated
with
conventional
plant
breeding
with
or
without
the
use
of
genetic
engineering
techniques
upon
which
a
quantitative
risk
assessment
could
be
based.
The
Agency
could
play
an
Page
17
of
85
active
role
in
advancing
comparative
risk
assessment
by
supporting
research,
perhaps
in
cooperation
with
FDA
and
USDA,
on
unintended
effects
associated
with
plant
breeding.

Technologies
such
as
metabolic
profiling,
gene
expression
systems
and
proteomics
might
be
useful
for
evaluating
unintended
effects.
These
methods
need
to
be
more
fully
developed
and
their
utility
documented
before
they
can
be
used
to
inform
the
scientific
review.
Previous
reports
such
as
The
US
General
Accounting
Office
(
GAO),
in
its
assessment
of
the
FDA's
safety
testing
of
GM
foods
(
GAO
1992)
and
several
European
entities
(
http://
www.
entransfood.
com)
have
concluded
that
non­
targeted
profiling
methods
might
be
useful
in
assessing
the
risks
associated
with
unintended
effects.
The
Panel
recommended
that
the
EPA,
in
partnership
with
FDA
and
USDA,
support
research
on
the
development
and
validation
of
profiling
methods
and
provide
support
for
the
further
development
of
site­
specific
gene
insertion
techniques.
The
Panel's
detailed
comments
are
presented
below.

Potential
unintended
effects
of
GM
foods
on
human
health
Unintended
effects
of
traditional
breeding
methods
on
levels
of
anti­
nutritional
or
toxic
constituents
in
food
organisms
have
been
characterized
in
conventional
organisms.
Organisms
derived
from
conventional
breeding
methods
including
tissue
cultures
may
have
a
somewhat
enhanced
possibility
for
genetic
(
and
epigenetic)
instabilities,
such
as
the
activity
of
mobile
elements
and
gene­
silencing
effects
(
Bhat
and
Srinivasan
2002).
These
effects
could
result
in
an
increased
possibility
of
pleiotropic
unintended
effects,
e.
g.,
increased
or
decreased
expression
of
constituents
or
possibly
modifications
in
expressed
proteins
as
well
as
epistasis,
(
the
interaction
of
the
inserted
gene
with
other
genes).

In
general,
the
compositional
analysis
should
be
performed
on
the
basis
of
validated
scientific
methods.
Strategies
for
the
compositional
analysis
in
food
products
derived
from
GM
plants
have
been
established,
where
key
substances
are
identified
and
analyzed
per
species.
Furthermore,
in
order
to
be
able
to
interpret
the
data
from
the
compositional
analysis
of
individual
animal
products
adequately,
insight
into
the
natural
variation
in
the
relevant
macro­
and
micronutrients
and
antinutrients,
if
present,
will
be
required.

In
the
future,
compositional
analysis
may
also
be
based
on
unbiased
profiling
of
the
GM
food
product
and
the
conventional
counterpart.
Techniques
for
the
profiling
approach
are
now
under
development
and
can
be
divided
into
three
subsections:
genomics,
proteomics
and
metabolomics
to
screen
for
differences
in
the
GM
plants
in
relation
to
the
gene
transcription
products,
proteins
and
metabolites,
respectively.
At
the
moment,
however,
none
of
these
techniques
is
yet
validated
and
ready
for
routine
use
in
risk
assessment
(
WHO/
FAO
expert
consultation,
2003).

The
Molecular
Biology
of
Unintended
Effects
The
problem
of
assessing
unintended
effects
has
been
a
matter
for
scientific
discussion
for
a
long
time
and
the
principles
of
assessment
by
international
scientific
organizations
have
changed
considerably.
For
example,
the
use
of
the
concept
of
the
substantial
equivalent
has
been
criticised
Page
18
of
85
extensively,
and
the
idea
of
the
principle
has
been
reduced
to
its
use
as
a
starting
point
of
a
risk
assessment.

In
principle,
traditionally
bred
transformed
lines
should
have
the
same
characteristics
as
the
parent
organisms.
However,
analyses
of
events
have
pointed
to
potential
different
outcomes
in
some
cases
where
the
basis
of
these
events
is
not
presently
fully
understood.

Two
specific
issues
for
consideration
are:

(
1)
A
molecular
characterization
of
the
new
product
should
show
that
the
recombinant
traits/
sequences
in
the
new
product
are
identical
to
the
insertion
/
traits
in
their
parental
lines.
In
principle,
the
method
of
breeding
the
two
parental
lines
could
affect
the
genetic
characteristics
of
the
inserted
traits.
Therefore,
evidence
is
needed
that
no
such
effect
has
occurred.
For
these
products,
a
risk
assessment
combining
evidence
from
parental
lines
and
the
product
should
be
requested.

(
2)
Stability
is
another
concern.
For
other
similar
products,
it
is
uncertain
how
many
generations
need
to
be
observed
to
establish
stability
of
inserted
traits.

Sites
of
Insertion
of
Foreign
Genes
and
Potential
Effects
One
of
the
concerns
with
any
genetically
modified
organism
is
the
precise
location
point(
s)
of
the
inserted
gene
or
genes.
Single
copy
insertions
of
the
genes
at
locations
other
than
in
functional
gene
regions
or
in
regulatory
areas
is
the
goal
but
with
current
technologies
insertion
is
typically
relatively
random.
If
significant
anomalies
occur
they
will
surely
be
picked
out
and
discarded
through
breeding
programs
but
more
subtle
effects
might
possibly
slip
through.
In
the
specific
case
at
hand,
the
utilization
of
Cry1F
and
Cry1Ac
genes
incorporated
into
the
cotton
genome,
it
is
noteworthy
that
both
for
the
single
insertions
via
typical
recombinant
DNA
techniques
and
then
the
construction
of
the
Cry1F/
CRY1Ac
stacked
strain
by
traditional
breeding
involved
only
single
insertion
events
in
each
case.

Potential
health
or
environmental
risks
of
genetically
combining
T­
DNA
transgenes
in
WideStrike
cotton
through
traditional
breeding
are
unlikely
to
differ
from
those
of
the
parental
transgenic
lines.
However,
our
present
knowledge
concerning
unique
or
special
risks
associated
with
GM
crops
is
somewhat
limited
and
presently
under
study
(
NAS,
2002).
One
ongoing
concern
is
that
T­
DNA
insertion
into
the
genome
is
a
mutagenic
event,
which
could
trigger
unintended
changes
in
nutritional
or
toxicological
properties
of
a
transformed
crop
(
Schubert,
2002).
T­
DNA
insertions
that
disrupt
agronomic
performance
or
gross
chemical
composition
can
be
detected
by
current
methods
of
screening,
whereas
other
insertions
that
alter
biosynthesis
of
molecules
which
are
toxic,
allergenic
or
carcinogenic
would
likely
be
overlooked.
Given
that
there
are
no
a
priori
means
to
predict
these
different
outcomes,
additional
information
should
be
sought
on
potential
genetic
effects
of
T­
DNA
insertion,
as
outlined
below.

T­
DNA
inserts
and
effect
on
genome
organization
Page
19
of
85
Assessments
of
risks
posed
by
GM
crops
rely
on
comparisons
with
conventionally
bred
(
non­
transformed)
counterparts
having
a
history
of
safe
use.
Apart
from
comparing
agricultural
characteristics
and
chemical
composition,
GM
plants
are
also
routinely
analyzed
for
the
number
of
T­
DNA
insertion
sites,
the
number
of
gene
copies
at
these
sites,
and
the
organization
of
DNA
within
inserts.
In
addition,
the
effect
of
T­
DNA
on
proximal
open
reading
frames
of
cellular
proteins
is
also
considered.
However,
additional
measures
outlined
below
would
better
inform
us
of
potential
downstream
genetic
effects
associated
with
transgenic
insertion
of
foreign
genes.

For
instance,
routine
Southern
hybridization
analysis
of
WideStrike
cotton
and
its
parental
transformed
lines
suggest
that
T­
DNA
inserts
coding
for
individual
insecticidal
proteins
are
present
as
single
copies
in
the
cotton
genome.
The
importance
of
this
observation
is
that
single
transgene
inserts
breed
true
and
show
fewer
unwanted
effects
(
i.
e.,
transgene
silencing
or
mutations)
than
multi­
copy
transgene
lines.
However,
single­
copy
T­
DNA
inserts
are
known
to
trigger
large­
scale
chromosomal
rearrangements,
including
translocations
(
Forsbach
et
al,
2003).
In
such
cases,
large
portions
of
flanking
DNA
at
the
T­
DNA
insert
site
may
also
be
duplicated
and
translocated
to
multiple
chromosome
locations
(
Nacry
et
al,
1998,
Tax
and
Vernon,
2001).
These
types
of
chromosomal
rearrangements
would
likely
complicate
interpretations
of
Southern
hybridization
data,
by
suggesting
that
a
single­
copy
of
the
transgene
was
present,
when
in
fact
duplication
and
translocation
of
the
T­
DNA
resulted
in
an
additional
chromosomal
copy
with
identical
flanking
sequences.
Therefore
to
complement
and
extend
current
strategies
for
assigning
transgene
copy
number,
the
chromosome
location
of
each
T­
DNA
insert
should
be
routinely
mapped
through
use
of
Mendelian
segregation
results
or
molecular
markers
for
cotton
(
Tomkins
et
al,
2001;
www.
genome.
clemson.
edu/
projects/
cotton/
bac
and
www.
cottoninc.
com/
Agriculture/
homepage.
cfm?
page=
3157).

A
related
concern
is
whether
T­
DNA­
induced
mutations
are
qualitatively
or
quantitatively
similar
to
those
which
occur
as
a
consequence
of
traditional
crop
breeding.
It
is
known
that
crossbreeding
of
genetically
different
cultivars
can
result
in
major
chromosomal
rearrangements.
In
addition,
mobile
DNA
elements
(
transposons)
also
induce
genetic
changes
in
plants.
At
present,
it
is
far
from
clear
whether
plant
transformation
differs
from
these
processes
and
poses
unique
or
greater
genetic
risks.

T­
DNA
inserts
and
effect
on
gene
expression
Several
hundred
base
pairs
of
genomic
sequence
were
identified
by
sequencing
of
regions
flanking
T­
DNAs
bearing
the
Cry1Ac
and
Cry1F
transgenes
in
WideStrike
and
parental
cotton
lines.
When
used
as
in
silico
probes,
BLAST
searches
failed
to
reveal
significant
homology
between
these
sequences
and
those
previously
deposited
in
the
GenBank
database.
Moreover,
analysis
of
1032
bp
(
534
bp
from
5'
border
and
deleted
16
bp
+
482
bp
from
the
3'
border)
of
cotton
sequences
flanking
the
Bt
Cry1Ac
cotton
3006­
210­
23
insertion
site
revealed
no
significant
(<
450
bp,
150
aa)
open
reading
frames
at
this
cloned
locus.
Similarly,
analysis
of
5028
bp
(
2073
bp
from
5'
border
and
deleted
53
bp
+
2902
bp
from
the
3'
border)
of
cotton
sequences
flanking
the
Bt
Cry1F
cotton
281­
24­
236
insertion
site
revealed
no
significant
(>
450
bp,
150
aa)
open
Page
20
of
85
reading
frames
at
this
cloned
locus.
Although
cotton
genes
were
thus
not
computationally
identified
at
the
T­
DNA
insertion
sites,
it
is
unclear
why
attempts
were
not
made
to
determine
whether
these
flanking
sequences
correspond
to
one
or
more
transcriptionally­
expressed
cellular
genes.

In
this
regard,
it
is
noteworthy
to
consider
that
T­
DNA
is
widely
used
by
plant
biologists
as
a
mutagenic
agent
to
study
the
biological
roles
of
specific
genes
(
Forsbach,
2003).
Moreover,
T­
DNAs
in
the
present
application
contain
strong
agrobacterial
enhancer
sequences
(
4ocs)
which
are
known
to
drive
expression
of
genes
that
are
proximal
to
the
insertion
site
[
this
process,
known
as
"
activation
tagging"
(
Tani
H,
2004),
is
widely
used
in
modern
plant
biology
research
to
identify
novel
genes.]
Thus
to
more
fully
investigate
the
possibility
that
T­
DNA
inserts
are
in
active
genes,
Northern
blot
hybridization
to
detect
cognate
mRNA
transcripts
should
be
done
using
5'
and
3'
sequences
that
flank
each
T­
DNA
as
probes.
Comparing
the
size
of
transcripts
detected
by
Northern
blot
in
fractionated
RNA
from
nontransformed
and
transformed
lines
will
assess
whether
the
T­
DNA
insertion
physically
disrupted
or
significantly
(
2
SD
from
the
norm)
affected
expression
of
the
cellular
gene.
It
is
also
recommended
that
due
to
potential
"
enhancer
effects"
by
multiple
ocs
promoter
sequences,
additional
flanking
sequences
5'
and
3'
from
the
T­
DNA
should
be
identified
to
characterize
genes
that
are
distal
to
the
insert
site.
Expression
of
these
distal
genes
would
be
monitored
by
Northern
blot
analysis
or
through
more
systematic
investigations
using
DNA
microarrays.

Unintended
effects
that
may
result
from
the
insertion
of
DNA
into
the
plant
genome
represent
hypothetical
hazards
that,
as
noted
previously,
have
yet
to
be
demonstrated
to
occur
in
products
presented
to
the
Agency
for
review.
The
challenge
is
to
define
specific
biochemical
or
metabolic
changes
that
represent
risks
that
warrant
rejection
or
management.
This
requires
specific
evidence
of
a
molecular
or
compositional
change
that
would
trigger
further
investigation.

It
is
impossible
to
quantify
and
characterize
the
risk
associated
with
unintended
effects
without
defining
it
in
terms
of
molecular
or
metabolic
changes.
It
bears
repeating
that
while
unintended
changes
may
be
possible,
it
is
both
not
clear
that
they
present
any
new
or
different
risks
that
are
significant
(
meaningful
to
human
and
animal
health).
To
date
no
evidence
has
been
presented
that
significant
risks
occur
(
Beachy,
2002,
Kuiper
2001).

Conventional
plant
breeding
itself
produces
unintended
effects.
There
is
a
long
history
of
safety
associated
with
plant
breeding.
Historically,
plant
breeders
have
depended
on
phenotype
and
performance
to
detect
unintended
effects
and
reject
plants
harboring
undesired
characteristics.
In
some
cases,
compositional
studies
may
inform
the
selection
of
individuals
worthy
of
further
development.
One
is
reminded
of
the
fact
that
plant
breeders
have
successfully
employed
strong
mutagenic
techniques
such
as
chemical
mutagens,
UV
light,
and
radiation
to
produce
genetic
diversity
and
altered
traits.

New
Possibilities
for
Gene
Insertion
Major
advances
in
techniques
for
precise
placement
of
foreign
genes
into
host
organisms
Page
21
of
85
should
very
soon
render
these
particular
concerns
about
unintended
consequences
of
foreign
gene
insertion
relatively
insignificant.
Recent
research
has
shown
that
highly
precise
implanting
of
foreign
genes
into
specific
sites
in
a
host
genome
is
not
only
possible,
but
can
be
accomplished
fairly
readily.
The
work
of
Lambowitz
and
his
colleagues
at
the
Institute
for
Cellular
and
Molecular
Biology
at
the
University
of
Texas
at
Austin
on
site­
specific
DNA
insertion
through
the
use
of
autocatalytic
group
II
introns
is
notable
in
this
regard
(
Karlberg
et
al.,
2001,
Mohr
and
Lambowitz,
2003).
While
such
positional
gene
insertion
will
certainly
serve
to
greatly
ameliorate
unintended
effects,
it
will
not
eliminate
them
entirely,
As
Schubert
(
2001)
and
others
have
pointed
out,
genes
can
act
at
a
distance
in
unexpected
ways.
New
compounds
or
pathways
might
be
created
or
activated
and
unexpected
associations
between
molecules
may
occur.

Animal
Feeding
Studies
for
Unintended
Effects
It
is
often
suggested
that
animal
studies
should
be
used
as
a
non­
targeted
screen
for
undetected
changes
for
PIPs.
There
are
several
limitations
to
the
use
of
animal
studies
on
whole
foods.
The
first
challenge
is
that
it
is
often
difficult
to
formulate
a
diet
that
is
nutritionally
adequate
and
well­
accepted
by
the
test
animal
which
also
contains
a
large
percentage
of
a
whole
food
(
FAO/
WHO,
2000;
Chassy,
2004).
Animal
studies
have
proven
useful
for
the
evaluation
of
pharmaceuticals,
food
additives,
industrial
and
agricultural
chemicals,
and
environmental
contaminants
because
these
compounds
have
biological
activity
at
fairly
low
levels
of
exposure.
Carefully
performed
toxicology
studies
employing
various
concentrations
of
pure
chemicals
can
be
used
to
determine
a
lowest
observed
adverse
effect
level
(
LOAEL)
in
a
particular
animal
species,
typically
a
90
day
rodent
study.
Such
studies
allow
the
setting
of
maximum
permissible
exposure
levels
that
provide
an
acceptable
safety
margin.
It
is
often
impossible
to
devise
an
animal
diet
with
a
whole
food
such
that
it
contains
a
sufficient
quantity
of
a
toxicant
that
will
elicit
a
biological
response.
Chassy
et
al.
(
2004)
indicated:

"
A
review
(
Munro
and
others,
1996b)
of
120
rat
bioassays
(
each
of
90
day
duration)
of
chemicals
of
diverse
structure
including
food
additives,
pesticides,
and
industrial
chemicals
found
LOAELs
to
range
from
0.2
to
5000
mg/
kg
body
weight
with
a
median
of
100
mg/
kg
and
a
5th
percentile
of
2
mg/
kg.
To
achieve
the
5th
percentile
of
exposure
from
a
toxic
constituent
present
in,
say,
a
food
crop
in
a
rodent
bioassay
(
at
a
food
incorporation
rate
of
30%)
the
toxin
would
have
to
be
present
at
a
level
of
80
ppm.
To
achieve
the
median
exposure
of
100
mg/
kg
it
would
have
to
be
present
at
5000
ppm.
These
concentrations
fall
well
within
the
range
of
existing
analytical
techniques
for
detection
of
inherent
toxicants
in
food.
The
concentrations
should
also
be
readily
detected
during
compositional
analysis
of
the
known
toxicants
in
the
host
organism
used
to
generate
the
improved
nutrition
crop."

It
is
clear
that
analytical
techniques
have
far
greater
power
than
animal
studies
for
the
detection
of
individual
compounds.
Moreover,
animal
studies
on
whole
foods
suffer
from
the
lack
of
a
specific
targeted
hypothesis
as
well
as
the
possible
presence
of
a
variety
of
confounders.
It
is
much
easier
to
design
a
study
to
investigate
an
effect
on
a
specific
target
organ,
enzyme
level
or
serum
metabolite
concentration
than
it
is
to
compare
the
health
outcomes
of
two
diets.
Although
several
whole
food
studies
have
been
reported
(
Chassy
2004),
it
may
be
more
appropriate
to
use
Page
22
of
85
animal
studies
to
probe
for
suspected
metabolic
or
toxic
effects
on
a
case­
by­
case
basis
than
it
would
be
to
require
them
in
every
case.
It
is
also
not
clear
that
whole
food
animal
studies
would
extrapolate
very
well
to
human
populations
that
have
great
genetic
and
dietary
diversity.
Animal
studies
are
therefore
not
likely
to
be
an
effective
screen
for
unintended
effects.

Targeted
Compositional
Analysis
Phenotype
and
functionality
are
highly
complex
traits.
Expression
of
literally
thousands
of
genes
must
be
coordinated
in
both
extent
and
timing.
Expression
is
also
modulated
by
extrinsic
environmental
signals,
availability
of
nutrients,
pests,
diseases
and
a
host
of
other
factors.
Phenotype
and
function
are
therefore
highly
sensitive
indicators
that
should
not
be
regarded
lightly.

Compositional
studies
are
also
powerful
indicators
that
no
significant
unintended
effects
have
occurred
in
the
breeding
and
development
process.
There
are
three
important
points
to
make
about
composition.
The
first
is
that
"
we
are
what
we
eat."
It
follows
that
all
we
need
to
know
is
the
safety
of
the
components
of
the
food
we
eat.
The
only
thing
that
really
matters
in
a
safety
assessment
of
a
food
or
ingredient
derived
from
a
transgenic
crop
is
the
composition
of
the
product
that
consumers
will
eat.
Levels
of
transcription
and
translation
as
well
as
varying
rates
of
turnover
of
protein
and
mRNA
normally
occur
in
all
organisms.
If
these
natural
oscillations
do
not
result
in
compositional
changes
that
have
health
or
safety
implications,
such
changes
are
of
little
consequence.

The
second
important
point
to
be
made
about
composition
that
is
often
misunderstood
is
that
there
is
a
natural
range
in
concentrations
observed
for
many
metabolites
in
a
plant.
It
is
misleading
to
look
in
a
food
composition
database
and
conclude
that
maize
is
composed
of
9.5%
protein.
The
number
in
the
database
represents
an
average
value
that
may
be
reasonably
used
for
calculations
of,
for
example,
protein
content
of
a
diet.
It
does
not,
however,
reveal
the
wide
range
of
concentration
of
proteins
(
and
all
other
metabolites)
that
may
be
found
in
individual
samples
of
maize.
It
is
not
possible
to
draw
conclusions
about
unintended
effects
related
to
changes
in
composition
without
understanding
the
natural
variability
in
natural
product
composition.
Until
recently,
very
little
information
was
available
about
natural
variability
in
composition.
ILSI
has
recently
placed
a
free
and
easily
accessible
crop
composition
database
online
(
http://
www.
cropcomposition.
org/).
The
utility
of
the
database
has
been
described
in
a
recent
publication
(
Ridley
et
al.,
2004).

In
many
instances,
the
content
of
secondary
metabolites
that
are
of
interest
for
their
nutritional,
health
beneficial
or
health
protective
effects
may
be
even
more
variable
than
macronutrients.
A
health
claim
has
recently
been
approved
for
soy
protein
consumption
by
the
FDA.
Many
consumers
are
interested
in
increasing
the
soy
and
isoflavone
contents
of
their
diets
in
response
to
reports
of
various
potential
health
benefits.
The
data
presented
in
Table
1
demonstrate
that
soybeans
vary
greatly
in
isoflavone
content.
The
same
cultivar
grown
at
4
different
sites
gave
rise
to
a
nearly
4­
fold
difference
in
isoflavone
content.
Almost
3­
fold
differences
were
observed
between
four
commercial
varieties
of
soybean.
Page
23
of
85
The
third
point
to
be
made
about
the
value
of
composition
studies
is
that
they
are
powerful
sensing
probes
for
the
concentration
of
almost
all
the
remaining
cellular
metabolites.
This
is
because
many
of
the
analytes
that
are
evaluated
in
composition
studies
(
i.
e.,
amino
acids,
fatty
acids,
vitamins,
toxicants
such
as
gossypol,
etc.)
are
products
or
intermediates
in
one
or
more
metabolic
pathways.
Cell
metabolism
is
a
highly
integrated
and
interconnected
set
of
highly
regulated
enzymatically
catalyzed
pathways.
Knowledge
that
the
concentration
of
dozens
if
not
hundreds
of
metabolites
is
within
normal
range
is
strong
evidence
that
most
all
metabolites
are
within
normal
ranges.
It
is
therefore
not
necessary
to
measure
each
individual
metabolite.
Perhaps
more
importantly,
the
currently
used
targeted
composition
studies
often
account
for
over
99%
of
the
total
composition
and
it
specifically
targets
micro­
components
that
are
of
health
or
safety
significance
(
i.
e.
toxicants,
vitamins,
bioactive
phytochemicals,
allergens,
antinutrients,
etc.).
It
is
worth
repeating
that
there
is
no
scientific
evidence
that
targeted
composition
studies
that
evaluate
over
100
analytes
from
diverse
metabolic
pathways
have
failed
to
detect
significant
risks.
Moreover,
there
is
also
no
evidence
that
the
kinds
of
hypothetical
unintended
effects
that
might
arise
from
gene
insertion
do
not
also
occur
in
conventional
breeding.
Several
unintended
effects
of
this
kind
have
been
reported
to
occur
as
a
result
of
conventional
breeding
programs
(
Kuiper
et
al.,
2001).

It
is
important
to
stress
that
not
only
does
content
of
specific
metabolites
vary
over
a
broad
range
in
plants,
but
also
that
changes
in
the
composition
of
a
plant
may
be
beneficial.
In
fact,
numerous
development
projects
are
in
progress
that
seek
to
alter
the
content
of
macro­
or
micronutrients,
eliminate
allergens
or
toxicants,
or
introduce
other
compositional
changes
that
may
be
health
beneficial.
Golden
Rice
is
a
well­
known
example
of
this
kind
of
strategy
(
Potrykus
2001).
A
new
risk
assessment
paradigm
may
need
to
be
created
in
order
to
assess
the
unintended
changes
that
might
accompany
large
changes
in
composition.

Almost
all
of
the
first
wave
of
biotechnology­
derived
crops
that
have
been
approved
and
commercialized
were
designed
to
be
no
different
than
conventional
varieties
of
that
crop
except
for
the
addition
of
one
or
two
additional
traits
 
often
intended
to
enhance
the
agronomic
properties.
While
the
Panel
indicated
that
GMO
crops
were
considered
as
safe
since
being
shown
to
be
"
substantially
equivalent"
to
their
conventional
counterparts,
but
this
is
a
misunderstanding
of
the
concept
of
substantial
equivalence
(
FAO/
WHO,
2000).
Substantial
equivalence
uses
compositional
analysis
to
identify
differences
that
merit
further
risk
characterization
so
it
is
in
effect
a
comparative
assessment
paradigm.
The
finding
of
differences
per
se,
does
not
demonstrate
an
increased
risk.
As
a
consequence
of
misunderstanding
of
the
substantial
equivalence
paradigm,
it
has
been
suggested
that
it
should
be
evolved
into
the
Comparative
Assessment
paradigm
(
Kok
and
Kuiper
2003).
The
use
of
the
comparative
assessment
concept
also
lends
itself
to
safety
assessment
of
compositionally­
modified
novel
foods
(
Chassy,
2004).
The
Agency
can
reasonably
expect
to
be
asked
to
review
crop
plants
that
have
PIP(
s)
and
nutritional
enhancements
so
it
is
essentially
to
accept
that
composition
difference
is
now
and
will
increasingly
be
in
the
future
a
fact.

It
should
also
be
noted
that
human
diets
vary
by
an
even
greater
range
than
the
Page
24
of
85
composition
of
the
individual
components
in
the
diet.
This
has
enormous
consequences
for
public
health.
The
comparative
safety
of
any
unintended
changes
must
always
be
related
to
exposure
in
the
context
of
the
whole
diet
and
the
range
of
dietary
intakes
represented
in
the
particular
subject
population
(
Chassy
et
al.,
2004).

Metabolome
An
excellent
recent
review
by
Goodacre
(
2004)
described
the
current
state
of
art
of
technology
and
understanding
of
metabolism
and
metabolomics.

The
Panel
commented
that
translation
(
proteome)
and
transcription
(
transcriptome)
may
provide
useful
insights
into
cell
function
and
metabolism,
and
the
composition
of
the
food
derived
from
a
plant
that
determines
fitness
for
human
or
animal
consumption.
It
is
instructive
to
investigate
methods
that
might
shed
additional
insight
into
what
has
now
become
referred
to
as
the
metabolome.
One
approach
would
be
to
conduct
a
comprehensive
analysis
of
the
concentration
of
each
cellular
metabolite.
One
could
envision
a
rapid
parallel
analyzer
that
would
provide
a
complete
blueprint
of
the
metabolome.
The
short­
coming
of
this
approach
is
that
databases
that
establish
norms
for
each
metabolite
would
need
to
be
constructed,
the
normally
observed
ranges
in
concentration
documented,
and
concentrations
of
each
metabolite
that
has
safety
implications
determined.
There
is
at
present
little
data
that
would
relate
nutrition,
development,
performance
and
health
of
animals
to
more
than
a
few
dozen
compounds,
primarily
the
nutrients.
The
development
of
non­
targeted
metabolic
screening
methodology
as
well
as
paradigms
for
understanding
the
implications
of
specific
concentrations
is
a
subject
of
research.
We
may
be
some
years
away
from
this
kind
of
metabolic
profiling,
although
in
the
near
term
some
metabolite
screening
technologies
may
prove
useful
in
identifying
changes
that
could
then
be
evaluated
by
more
targeted
methods
in
order
to
understand
their
significance.

In
the
meanwhile,
there
is
another
approach
to
metabolome
screening
that
might
be
called
metabolic­
fingerprinting.
It
is
a
logical
extension
of
the
composition
analysis
that
is
now
used
to
assess
the
safety
of
transgenic
crops.
It
should
be
possible
to
look
at
the
metabolic
pathways
in
a
plant
cell
and
select
a
dozen
or
two
dozen
key
metabolic
intermediates
that
strongly
correlate
with
overall
cell
composition.
The
information
required
to
do
this
is
in
the
literature
and
expert
physiologists
and
biochemists
could
probably
reach
a
consensus
on
which
molecules
to
select
for
studies
aimed
at
validating
this
intelligent
targeted
fingerprinting.

There
is,
however,
no
way
that
such
studies
can
cope
with
totally
novel
metabolites
that
arise
from
newly
formed
pathways
or
activation
of
silent
genes.
This
would
be
in
effect,
"
knowing
the
unknowable."
Fortunately,
the
multi­
pronged
safety
assessment
process
employs
several
distinct
types
of
tests
(
as
noted
above)
in
order
to
screen
for
significant
unintended
effects
that
would
be
undetected
by
compositional
analysis.

Metabolic
Profiling
Plants
produce
an
extraordinary
array
of
small
compounds
and
these
can
affect
all
stages
Page
25
of
85
of
development
as
well
as
the
properties
of
the
final
materials
from
these
plants
that
are
used
in
commerce.
In
the
past,
analysis
of
such
metabolites
has
essentially
been
on
a
case
by
case
basis
and
has
been
limited
to
only
a
few
chemical
entities.
Nutritional
and
compositional
assays
of
GM
plants
examine
such
properties
as
the
levels
of
proteins,
sugars,
fats
and
other
materials.
These
targeted
assays
are
certainly
extremely
useful.
However,
when
screening
for
potential
unanticipated
consequences
of
foreign
gene
introduction,
it
would
be
very
desirable
to
have
very
broad
yet
precise
indicators
of
what
is
going
on
in
the
modified
and
non­
modified
plants
as
far
as
metabolites
are
concerned.
New
tools
for
looking
at
issues
such
as
metabolic
profiles
are
coming
into
widespread
use
and
these
could
provide
a
much
more
broad­
band
examination
of
modified
and
non­
modified
plant
materials
in
a
cost­
effective
manner.

The
recent
techniques
that
have
become
available
can
detect
and
quantify
several
hundreds
of
plant
metabolites
in
a
single
analysis
(
thus
the
term
metabolic
profiling),
using
techniques
such
as
gas
chromatography/
mass
spectrometry
(
GC/
MS)
and
liquid
chromatography/
mass
spectrometry
(
LC/
MS).
Such
procedures
lend
themselves
well
to
providing
a
more
detailed
understanding
of
the
potential
effects
of
foreign
gene
introduction
into
plants.
From
the
evidence
available
to
date,
it
is
very
likely
that
few
differences
will
be
seen
between
the
already
heavily
studied
GM
crops
and
their
non­
GM
counterparts.
It
would
be
very
reassuring
that
at
a
basic
level,
little
metabolic
disturbance
was
occurring
because
of
the
transgenic
alterations.
Finally,
testing
of
new
genetically
altered
materials
in
this
manner,
early
on
in
the
development
process,
would
serve
to
provide
an
early
warning
of
additional
concern.

Proteomics
Future
studies
of
genetically
engineered
plants
using
proteomic
approaches
look
very
favorable
for
comparison
of
transformed
and
non­
transformed
varieties
as
well
as
for
traditionally
bred
variants
of
transformed
species.
As
has
been
expressed
by
Voet
and
Voet
(
2004)
proteomics
investigates
"
all
proteins
expressed
by
an
organism,
with
an
emphasis
on
their
quantification,
localization,
modifications,
interactions
and
activities.
"
Tools
used
to
gather
such
information
include
2­
D
gel
electrophoresis
combined
with
mass
spectrometry,
amino
acid
composition
and
sequence
analysis,
peptide
mass
fingerprinting
and
matrix­
assisted
laser
desorption
ionization
(
MALDI)­
time
of
flight
(
TOF)
mass
spectrometry
for
identification
of
post
translational
modifications.
While
these
processes
can
be
highly
precise
and
definitive
in
ascertaining
differences
between
samples,
for
full
utilization
they
do
depend
upon
the
availability
of
adequate
databases,
which
are
still
being
assembled
for
many
plant
species.

Gene
Microarrays
The
use
of
gene
microarrays
to
assess,
for
example,
which
genes
are
being
transcribed
into
messenger
RNA
at
a
particular
time,
in
a
particular
organ,
tissue
type
or
even
sub­
cellular
organelle
has
been
transforming
molecular
biology
in
recent
years.
This
same
technology
can
potentially
be
employed
to
answer
questions
about
how
plant
metabolism
might
be
altered
by
the
incorporation
of
foreign
genes,
for
example
those
of
the
Bt
toxins.
Of
course
the
use
of
such
microarrays
in
a
meaningful
manner
requires
knowledge
about
the
genome
of
the
plant
in
Page
26
of
85
question.
In
this
connection,
it
is
encouraging
to
note
that
sequencing
of
the
cotton
genome
is
proceeding
apace,
that
approximately
50,000
EST
sequences
are
available
and
that
microarrays
covering
about
30%
of
the
cotton
genome
are
already
available
(
http://
cottongenomecenter.
ucdavis.
edu/
microarrays.
asp).

These
newer
techniques
of
metabolic
profiling,
proteomics
and
gene
chips
are
indeed
in
their
infancy
as
far
as
application
to
plant
systems
is
concerned
and
clearly
need
database
collection,
validation,
etc.
before
they
are
put
into
regulatory
use.
Additionally,
when
this
is
accomplished,
they
should
be
used
as
adjuncts
to
present
test
methodologies,
which
have
served
well.
What
these
new
tools,
with
their
broad
span
analysis
do
offer
is
great
potential
in
the
future
for
allowing
recognition
of
unintended
consequences
of
genetic
manipulation
early
in
the
process.
This
is
true
whether
the
genetic
changes
are
made
by
traditional
or
recombinant
methodologies.

Agency
Charge
Ecological
Risk
1.
WideStrike
cotton
is
a
product
expressing
pyramided
Cry1F
and
Cry1Ac
Bt
proteins.
The
submitted
non­
target
effects
studies
examined
the
effects
of
the
Cry1F
and
Cry1Ac
proteins
separately
and
in
combination
to
detect
any
synergistic
effects
on
non­
target
wildlife.
No
synergistic
effects
or
increase
in
non­
target
host
range
were
seen
as
a
result
of
combining
these
two
proteins
in
the
same
product.

The
Panel
is
requested
to
comment
on
the
need
for
non­
target
hazard
data
development
on
the
combinations
of
Cry
proteins
being
considered
for
registration
when
data
on
the
effects
of
the
individual
Cry
proteins
are
readily
available
and
show
no
adverse
effects.

Panel
Response
The
Agency
has
done
a
good
job
of
preparing
documents
presenting
the
available
data
regarding
control
of
TBW,
CBW
and
PBW
in
cotton.
Cry1Ac
and
Cry1F
seem
to
be
rather
specific.
Thus
direct
hazards
to
vertebrates
due
to
exposure
to
the
Cry1F
and
Cry1Ac
proteins
should
be
minimal
or
non
existent,
making
consideration
of
synergy
in
vertebrates
unwarranted.
Hazards
to
invertebrates
have
a
higher
likelihood
of
showing
effect.
In
fact,
synergy
has
been
observed
for
Bt
toxins
and
spores
in
various
lepidoptera
(
Moar
et
al.
1989,
Dubois
and
Dean
1995,
Johnson
and
McGaughey
1996),
and
between
different
CryIV
and
CytA
toxins
in
Aedes
aegypti
(
Chilcutt
and
Ellars
1988).
In
contrast,
other
studies
have
failed
to
demonstrate
any
synergistic
interactions
between
Cry1A
toxins
in
various
lepidopteran
insects
(
Moar
et
al.
2002,
by
Tabashnik
1992).
One
citation
(
Chakrabarti
et
al.
1998)
indicated
a
synergistic
interaction
between
Cry1Ac
and
Cry1F
in
Helicoverpa
armigera.
Access
to
the
article
would
have
assisted
the
Panel
to
fully
examine
the
experimental
methods
and
analytical
techniques.
Although
interactions
between
toxins
and
spores
would
not
be
an
issue
with
transgenic
plants
that
express
only
the
toxins,
potential
interactions
among
different
Cry
toxins
may
be
important
in
resistance
management
(
e.
g.,
Tabashnik
et
al.
1997).
Considering
all
data
presented
and
information
from
Page
27
of
85
one
literature
review,
the
predominance
of
data
demonstrated
no
synergy
for
Cry1
proteins.
Therefore,
the
Panel
concluded
that
synergistic
effects
are
unlikely
to
occur
when
plants
produce
Cry1Ac
and
Cry1F
at
the
concentrations
that
are
currently
needed
to
control
TBW,
CBW
and
PBW.

When
each
toxin
was
tested
independently
and
in
combination
at
environmentally
relevant
concentrations,
none
of
the
data
indicated
any
synergistic
or
additive
effects
on
the
target
or
nontarget
organisms
tested.
Evaluations
of
non­
target
organisms
were
generally
done
at
rates
far
in
excess
of
what
these
organisms
would
be
exposed
to
in
the
field.

Based
on
the
written
and
verbal
information
provided
by
the
Agency,
the
Panel
believed
that
non­
target
hazard
data
can
be
treated
similarly
to
those
data
in
the
Human
Health
Risk
Assessment,
for
which
a
similar
question,
as
noted
previously
was
raised.
Indeed,
this
is
a
very
similar
issue,
since
mammals
are
also
non­
target
organisms.
Therefore,
the
Panel
concurs
with
the
opinions
the
Panel
expressed
in
the
answer
to
Question
1
from
the
Product
Characterization/
Human
Health
Risk
Assessment,
that
toxicity
testing
on
the
combinations
of
Cry
proteins
is
not
necessary.
The
Panel's
opinion
that
Cry
proteins
do
not
require
further
synergistic
testing
will
not
necessarily
extend
to
other
types
of
Bt
toxins.

While
synergism
is
not
considered
to
be
important
for
the
Cry1
proteins
under
consideration,
the
Panel
believed
that
future
testing
of
non­
target
effects
of
these
types
of
toxins
should
proceed,
with
both
toxins
to
be
expressed
by
genetically
engineered
plants
rather
than
with
individual
purified
toxins.
Testing
mixtures
of
compounds
would
provide
data
that
are
more
relevant
to
the
proposed
use
scenarios
and
likely
exposure
scenarios.
Testing
the
toxicity
of
mixtures
for
several
non­
target
insect
species
would
also
provide
data
that
could
refine
field
evaluations
to
determine
any
effects
on
insect
species
diversity.
This
could
be
a
critical
parameter
controlling
food
resources
for,
and
thus
impacts
on
avian
and
mammalian
species.
Such
an
evaluation
will
ultimately
provide
data
that
could
be
used
in
ecological
risk
assessments
where
the
question
that
should
be
asked
is:
What
is
the
combined
effect
of
Cyr1Ac
and
Cry1F
on
species
elimination
from
ecosystems?

Some
concern
was
raised
regarding
the
small
amount
of
Cry1F
data
found
through
a
CAB
search
of
the
literature
since
1973.
A
single
non­
target
study
of
monarch
butterflies
(
Helmich
et
al.
2001)
and
11
studies
involving
target
species
were
found.
The
Panel
suggested:
a)
USDA
has
compiled
more
Cry1F
data
than
is
available
in
the
materials
prepared
for
this
review
or
in
the
peer
reviewed
literature;
b)
it
would
be
useful
if
all
non­
target
data
from
the
Agency
presentation,
the
USDA
evaluations,
and
peer
reviewed
literature
be
compiled
into
one
summary
table
or
a
series
of
tables
for
reference
purposes
in
future
evaluations
of
this
and
similar
products.

Agency
Charge
2.
The
weight
of
evidence
from
the
reviewed
data
indicates
that
there
will
not
be
a
hazard
to
wildlife
from
the
commercialization
of
WideStrike
cotton.
Although
the
Bt
proteins
expressed
by
WideStrike
are
known
to
affect
only
lepidopteran
insect
species,
the
Agency
Page
28
of
85
evaluated
studies
of
potential
effects
on
a
wide
variety
of
non­
target
organisms
that
might
be
exposed
to
the
Cry1F
and
Cry1Ac
protein,
i.
e.,
wild
mammals,
birds,
invertebrates,
and
aquatic
species.
EPA
concluded
that
aquatic
and
terrestrial
wildlife
was
not
likely
to
be
harmed
and
that
WideStrike
cotton
was
not
likely
to
threaten
the
long­
term
survival
of
any
non­
target
wildlife
populations.

The
Panel
is
requested
to
comment
on
the
Agency's
analysis
of
the
currently
available
data
on
the
potential
impacts
of
WideStrike
cotton
on
non­
target
species.

Panel
Response
The
Panel
recognized
improvements
in
the
quality
of
the
Agency's
analysis
and
the
degree
to
which
this
reflected
a
positive
response
by
the
Agency
to
recommendations
made
by
previous
SAP
reviews
of
PIPs.
The
Panel
also
noted
that
the
Tier
I
tests
were
generally
of
a
higher
quality
than
the
tests
that
other
SAPs
have
reviewed.
Overall,
however,
detailed
analysis
of
the
fieldbased
evaluations
revealed
that
these
did
not
meet
current
standards,
and
the
Panel
repeated
requests
made
by
other
SAPs,
that
the
Agency
issue
detailed
guidelines
for
semi­
field,
and
fieldbased
procedures.
The
Panel
recommended
that
the
exposure
pathway
and
diet
used
within
the
Chrysoperla
test
be
examined
further
to
determine
the
rigor
and
repeatability
of
the
test
and
also
suggested
that
Orius
spp.
be
considered
as
a
more
appropriate
test
subject.
The
Panel
also
considered
that
a
more
appropriate
parasitoid
than
Nasonia
should
be
selected.
Finally,
the
Panel
expanded
the
table
of
suggested
arthropod
test
organisms,
first
developed
by
the
August
2002
SAP,
to
include
aquatic
species.
A
detailed
response
to
the
question
by
the
Panel
is
provided
below.

The
Panel
noted
improvements
in
the
quality
of
the
Agency
analysis
and
assessment
since
the
Panel
last
met
on
this
subject
approximately
two
years
ago.
Given
that
some
of
these
improvements
appear
to
have
been
in
direct
response
to
recommendations
made
by
previous
SAPs,
the
current
Panel
argued
that
this
highlighted
the
value
of
the
public
review
process
and
that
it
also
provided
evidence
of
a
positive
response
to
SAP
comments
by
the
Agency.

With
respect
to
the
Tier
I
screening
procedures
reported
for
WideStrike
cotton,
the
general
quality
of
the
submissions
was
considered
to
be
better
than
in
previous
cases,
particularly
with
regard
to
characterization
of
test
material
in
the
bioassay
procedures,
the
selection
of
test
organisms
and
the
quality
and
clarity
of
the
reporting.
The
Agency
review
was
found
to
be
more
informative
of
its
logic
and
reasoning,
and
more
explicit
about
the
limitations
of
the
tests.
This
made
it
easier
for
the
Panel
to
be
constructively
critical.

The
Panel
had
general
comments
concerning
maximum
hazard
dose
evaluation.
The
Panel
did
not
find
the
protocols
for
Tier
1
testing
appropriate.
The
duration
of
many
tests
were
determined
by
control
mortality
exceeding
20%.
The
Panel
suggested
that
control
mortalities
as
high
as
20%
raise
a
concern
of
an
error
with
the
protocol
and
that
tests
either
need
to
be
repeated
or
that
some
adjustments
in
protocols
are
required.
Sample
sizes
also
seem
small,
providing
little
power
to
discern
differences
that
may
exist.
Page
29
of
85
The
greatest
challenge
raised
by
the
Panel
mainly
arose
through
the
lack
of
progress
with
development
of
EPA­
approved
protocols
for
field­
based
evaluations,
clarity
regarding
their
role
relative
to
laboratory­
based
evaluations,
and
the
availability
of
intermediate
options
that
might
mitigate
the
need
for
long­
term,
open­
ended,
field
studies,
that
are
not
designed
to
test
clearly
stated
hypotheses.
The
Agency
acknowledged
that
specific
protocols
are
lacking
and
that
science
needs
to
advance
in
order
for
protocols
to
be
developed.
Thus,
field
data
are
treated
as
supplementary
by
the
EPA,
and
procedures
are
not
subjected
to
the
critical
scrutiny
that
is
the
norm
for
required
tests.

With
regard
to
the
maximum
hazard
approach,
the
effectiveness
of
this
approach
is
dependent
upon
species
selection
(
where
a
range
of
relevant
physiologies
should
be
evaluated,
relative
to
the
mode
of
action
of
the
test
material),
and
rigor
in
the
conduct
of
tests,
which
should
follow
detailed
protocols.
The
Panel
noted
that
the
tests
reported
meet
many
of
the
criteria
listed
by
previous
Panels,
but,
it
argued
that
in
order
to
affect
permanent
advances
in
the
standards
of
testing,
criteria
for
the
conduct
of
tests
need
to
be
built
into
new
written
requirements
or
test
guidelines.

Of
particular
importance
for
inclusion
in
these
new
guidelines
are
criteria
noted
by
two
previous
SAPs
(
in
1999
and
Aug
2002).
These
include:
1)
verification
of
exposure
levels
of
test
organisms
to
proteins
throughout
the
bioassay;
2)
detailed
quantification
of
EEC
in
the
field;
3)
clearly
stated
endpoints;
4)
a
clear
statement
that
tests
which
fail
to
reach
the
designated
endpoint
are
not
eligible
for
consideration;
5)
use,
where
possible,
of
foods
used
by
test
species
in
their
relevant
habitat;
6)
verification
that
the
food
offered
to
the
species
actually
contained
the
administered
material,
at
the
intended
dose,
throughout
the
investigation;
7)
verification
that
all
life
stages
of
the
species
are
exposed
appropriately
to
the
transgene
product
(
i.
e.
actually
contact
the
toxin
in
relevant
ways);
and
8)
that
there
is
sufficient
replication
and
that
sufficient
numbers
of
insects
were
screened
based
on
statistical
power.

The
Agency
stated
that
field
testing
was
"
recommended
by
the
August
2002
SAP".
Ultimately,
scientific
confidence
in
Tier
I
screening
should
be
sufficient
to
limit
further
tiers
of
testing
when
no
effects
are
detected
at
the
maximum
hazard
dose.
This
would
mean
that
no
open
field
investigations
would
be
undertaken
with
some
products.
Confidence
in
this
approach
would
increase
if:

 
The
armory
of
guidelines
were
to
be
expanded
to
include
extended
laboratory
and
semifield
approaches
as
an
intermediate
step,
before
full
field
testing
was
requested.
No
effects
or
failure
to
exceed
a
trigger
or
threshold
value
in
these
tests
would
cause
the
testing
to
cease
at
that
point.
Widely
used
intermediate
testing
methods
include
extended
laboratory
tests
(
use
of
more
realistic
substrates
and
exposure
pathways
within
the
laboratory)
or
semi­
field
tests
(
confinement
of
individual
or
multiple
species
of
test
organisms
within
microcosms,
mesocosms,
field
cages
or
barriered
arenas).
The
Agency
is
encouraged
to
avail
itself
of
opportunities
to
discuss
method
development
in
the
international
arena,
for
example
in
relevant
working
groups
of
such
organizations
as
the
International
Standards
Page
30
of
85
Organization,
OECD,
International
Organization
of
Biological
Control,
and
SETAC,
among
others.

 
Bioassessment
protocols
were
developed
for
surveys
in
whole
fields
to
help
answer
concerns
about
the
possible
detection
of
rare
events,
effects
resulting
from
combinations
of
treatments
that
do
not
occur
within
conventional
experimental
designs
or
concerns
about
long­
term,
unforeseen
effects.
Such
protocols
could
then
be
employed
on
the
health
of
agroecosystems,
or
as
a
component
of
post­
release
product
stewardship.

 
The
field
evaluation
guidelines
presented
were
greatly
improved
with
respect
to
experimental
design,
sampling
procedures,
taxonomic
focus,
statistics
and
interpretation.

The
Panel
reiterated
its
recommendations
from
the
August
2002
FIFRA
SAP
meeting
regarding:

 
The
requirement
that
registrants
evaluate
and
select
test
sites
from
a
number
of
candidates,
in
the
season
before
testing
begins,
to
determine
whether
the
organisms
of
concern
are
present
and
sufficiently
abundant
to
provide
a
basis
for
statistical
discrimination
of
small
but
significant
effects.

 
Use
of
sampling
methods
of
known
efficiency
and
precision
with
consideration
to
within­
plot
variability
when
determining
the
intensity
and
frequency
of
sampling.

 
Use
of
a
scale
and
layout
of
the
experiment
that
minimizes
the
risks
of
edge
effects
and
reinvasion
from
untreated
control
plots,
and
which
takes
into
account
the
dispersal
rates
and
phenology
of
the
organisms
of
concern.

The
Panel
recognized
the
limitations
of
laboratory­
based
testing,
and
it
may
be
argued
that
the
requirement
for
field
testing
will
continue
until
a
dataset
has
been
generated
that
effectively
validates
the
maximum
hazard
method.
The
most
important
limitations
to
laboratory
testing,
that
can
be
compensated
for
by
field
tests,
were
noted
by
the
August
2002
FIFRA
SAP
and
included
the
statements
that:

 
Levels
and
routes
of
test
material
exposure
that
may
not
be
realistic
in
the
laboratory.

 
Effects
assessments
are
made
after
short­
term
exposure
of
organisms,
not
lifetime
exposures
as
might
occur
in
the
field.

 
Organisms
in
the
field
are
subject
to
supplementary
stresses
that
have
additive
effects,
including
sub­
optimal
temperatures
and
humidity,
and
starvation
and
parasitism,
that
amplify
impacts
that
occur
under
the
optimal
physical
and
biological
conditions
of
laboratory
tests.
Page
31
of
85
 
Laboratory
tests
may
evaluate
an
appropriate
category
of
organism,
but
they
inevitably
fail
to
evaluate
species
that
are
actually
exposed
to
test
substances
in
the
field.

Despite
these
limitations,
the
Panel
argued
that
the
Agency
could
already
be
striving
for
a
more
structured
approach
to
risk
assessment
for
PIPs,
guided
by
more
detailed
protocols,
that
expand
the
availability
of
replicated
intermediate
field
test
data,
which
increase
the
potential
for
long­
term,
multi­
field
surveys
after
release,
and
which
probably
decrease
the
expectation
for
open­
ended
field
census
studies
with
their
various
methodological
challenges
(
as
noted
in
the
August
2002
meeting
minutes).

With
respect
to
the
avian
hazard
assessment,
the
Panel
was
asked
to
consider
the
effects
of
a
reduced
food
base
for
birds
utilizing
corn
fields
that
have
a
high
non­
target
insect
mortality.
Are
nestlings
any
more
susceptible
to
Cry
proteins
after
they
have
been
biotransformed
by
the
insect
and
fed
to
the
nestling?
Do
birds
lack
the
bioactivating
enzyme
or
the
actual
receptor
site
for
the
toxin?
Could
songbird
or
game
bird
nesting
success
change
with
WideStrike
cultivation
relative
to
conventional
cotton
cultivation?
Perhaps
insect
prey
availability
would
be
sufficiently
different
to
produce
a
difference
in
nesting
success.
This
question
emerged
from
UK
work
with
grey
partridge
and
other
species
impacted
indirectly
by
agrochemicals.

With
respect
to
aquatic
species,
the
Panel
noted
that
it
was
correctly
stated
by
the
Agency
that
the
August
2002
SAP
recommended
a
list
consisting
only
of
terrestrial
taxa
for
Tier
1
testing.
The
Panel
noted
however,
that
this
does
not
mean
that
aquatic
taxa
were
excluded
as
being
potentially
affected
in
some
circumstances,
even
if
not
in
the
present
case
for
cotton.
A
proposal
was
made
to
extend
the
table
of
proposed
test
organisms
in
the
2002
FIFRA
SAP
meeting
minutes
to
indicate
where
aquatic
taxa
may
be
incorporated.
It
was
also
suggested
from
the
August
2002
FIFRA
SAP
meeting
minutes
that
Daphnia
might
be
substituted
by
a
member
of
the
shredder
functional
feeding
group
(
e.
g.,
some
Diptera
or
Trichoptera
species
might
be
more
appropriate)
to
increase
the
opportunity
for
uptake
of
the
toxin.
Precipitate
formation
in
the
reported
Daphnia
test
was
difficult
to
interpret,
but
may
imply
that
the
test
organisms
were
not
exposed
to
the
test
substance.
Page
32
of
85
The
table
of
recommended
test
taxa
from
the
August
2002
FIFRA
SAP
meeting
minutes
for
selection
of
test
taxa
has
been
edited
as
follows
(
underlined):

Functional
Group
Examples
Anthropocentric
Functions
Secondary
pests
­
Sporadic
pests,
induced
pests
Natural
enemies
­
Predators,
parasitoids,
parasites,
competitors,
ants,
and
weed­
eating
herbivores
Rare
or
endangered
species
­
Red
list
species
or
species
of
value
for
biodiversity
conservation
Species
that
generate
income
­
Honey
bees,
silk
moths
Species
of
social
or
cultural
value
­
e.
g.
Monarch
butterflies,
honey
bees,
organisms
conferring
health
of
lotic
or
lentic
aquatic
systems
associated
with
crop
or
near
crop
habitats
Ecological
Functional
Groups
Non­
target
herbivores
­
Plant
eating
species
that
are
not
the
target
of
the
transgene
(
including
aquatic
species
if
appropriate)
Secondary
consumers
­
Species
that
eat
herbivores;
predators,
parasitoids,
parasites
Pollinators
­
Bees,
selected
Diptera
(
e.
g.
Syrphidae)
and
Coleoptera,
etc.
Decomposers
­
Scavengers,
ants,
Collembola,
micro­
organisms,
earthworms,
mites,
nematodes,
and
aquatic
taxa,
particularly
those
within
the
shredder
functional
feeding
group
Seed
dispersers
­
Birds,
small
mammals,
ants
Concerning
possible
long­
term
impacts
on
soil
invertebrates,
the
hazardous
outcome
of
impacts
on
soil
taxa
was
narrowly
defined
as
crop
residue
build
up
in
the
field.
This
should
be
expanded
to
acknowledge
the
need
to
protect
soil
health
and
ecological
function.
Residue
build
up
may
be
one
unintended
negative
consequence
of
impact
on
soil
organisms
that
could
reveal
more
significant
harm
to
biogeochemical
processes.

Previous
Panels
have
recommended
changes
to
the
list
of
non­
target
insects
to
be
tested
and
those
same
concerns
are
raised
again
here.
Nasonia
is
a
poor
model
as
a
parasitoid
in
the
cotton
system
and
there
would
be
many
other
more
suitable
models
(
Cotesia,
Trichogramma,
etc.).
Chrysoperla
is
a
good
model
for
the
cotton
system,
but
others
are
just
as
good
and
maybe
better
from
the
standpoint
of
greater
exposure
to
toxins
via
plant
feeding.
Orius
insidiosus
has
been
suggested
in
past
Panels
and
would
be
a
good
model
because
it
is
very
common
and
abundant
in
cotton
and
also
feeds
on
pollen
and
plant
sap.
(
The
Panel
noted
that
the
minute
pirate
bug
is
O.
tristicolor
not
insidiosus).
Geocoris
punctipes
would
be
another
very
good
model
for
the
cotton
system.
This
species
occurs
in
high
abundance
throughout
the
cotton
belt.
It
is
also
a
Page
33
of
85
well­
known
plant
sap
feeder
and
feeds
on
small
caterpillar
larvae.
Both
Orius
and
Geocoris
are
available
commercially
and
easy
to
rear
in
the
laboratory.

The
registrant
was
congratulated
for
addressing
the
issue
of
toxin
decay
in
the
presented
diets
via
ELISA
and
SDS­
PAGE.
However,
the
Panel
commented
that
with
a
small
amount
of
additional
testing,
the
detection
of
toxins
in
the
non­
targets
themselves
would
be
possible
and
prudent.

Concerns
were
raised
by
the
August
2002
FIFRA
SAP
on
Bt
corn
relative
to
diet
exposure
in
Chrysoperla.
It
is
not
clear
that
this
issue
has
been
satisfactorily
addressed
here.
The
toxin
is
still
present
only
on
the
surface
of
the
moth
eggs.
Are
the
insects
ingesting
the
toxin?
Could
this
be
tested
by
ELISA?
Is
the
registrant
confident
that
the
other
insect
models
tested
are
actually
ingesting
the
toxins?
Could
this
be
verified
with
ELISA?
Previous
panels
have
suggested
artificial
diets
for
testing
purposes.
This
would
seem
to
alleviate
many
of
the
problems
associated
with
definitive
exposure.
Artificial
diets
have
been
developed
for
Chrysoperla,
Orius
and
Geocoris.
Could
these
be
explored
as
better
models
for
toxin
delivery?
One
Panel
member
suggested
that
while
this
may
be
a
useful
method
for
exposure,
some
care
needs
to
be
taken.
Many
diets
contain
enzymes,
pro­
oxidants
and
other
substances
that
may
affect
incorporated
toxins.
The
high
level
of
nutrition
may
also
mask
more
subtle
effects
of
the
toxin
or
may
even
enhance
their
effects.
The
Panel
member
suggested
that
special
precautions
may
be
needed
to
avoid
these
issues.
Several
companies
hold
ARS
patents
for
these
diets
including
Buena
Biosystems
and
Beneficial
Insectary
and
would
likely
be
able
to
supply
the
diets.
Even
if
these
diets
don't
permit
insect
reproduction
they
may
be
sufficient
for
shorter­
term
longevity/
survival
studies.

The
Panel
noted
that
the
Hymenopteran
parasitoid
test
was
incorrectly
described
as
a
larval
test
in
the
Agency's
review.
It
also
noted
enhanced
toxicity
in
the
treatment
with
both
proteins
at
the
maximum
hazard
dose.

The
post­
emergence
mortality
in
all
bee
treatments
detracted
slightly
from
the
confidence
one
has
in
the
honey
bee
test
conclusions.
More
confidence
would
have
been
warranted
if
postemergence
mortality
of
controls
had
not
indicated
that
something
other
than
the
dosed
toxins
was
influencing
bee
viability.

Regarding
non­
target
organism
susceptibility
in
general,
one
Panel
member
considered
an
analysis
for
Cry
1f
and/
or
Cry
1Ac
receptors
in
non­
target
organisms.
Perhaps
the
standard
nontarget
representatives
could
be
characterized
for
presence
of
Cry1F
and
Cry1Ac
receptors?
The
scientific
literature
indicated
that
lepidopterans
are
the
ONLY
group
of
organisms
that
are
capable
of
being
affected
by
Bt
toxins.
However,
the
Panel
member
had
seen
no
publications
which
actually
attempted
to
detect
Bt
receptors
in
the
spectrum
of
representative
non­
target
organisms
 
other
than
mammals.
Physical
confirmation
of
receptor
absence
in
non­
target
gut
epithelial
cells
would
reduce
the
level
of
uncertainty
and
reinforce
the
indication
that
Cry
proteins,
alone
and
in
combination,
exhibit
little
to
no
toxicity.
In
addition,
the
absence
of
receptors
would
negate
any
Page
34
of
85
concerns
for
food­
chain
transfer
of
bioactivated
Bt
toxins
by
lepidopterans
which
may
be
taken
as
a
food
source.

With
regard
to
the
field
studies
that
were
submitted,
the
Panel
noted
EUP
restrictions
in
the
first
year
that
limit
the
area
that
can
be
planted.
Thus,
a
series
of
concerns
arise
that
could
reasonably
have
been
considered
by
the
Agency
and
the
registrant
in
planning,
execution,
analysis
and
interpretation
of
these
investigations.
Although
guidelines
are
not
available,
the
Panel
expected
standards
of
current
science
to
be
met.

The
Panel
noted
that
page
28
of
the
registrant
review
stated
that
plot
sizes
in
these
studies
were
1,000
square
meters.
In
fact
the
scale
was
far
less
than
this,
namely
2,667
square
feet
in
the
Arizona
experiment
and
6,400
square
feet
in
the
Louisiana
experiment,
which
were
only
50
feet
and
40
feet
long
respectively,
with
mustard
and
pig
weed,
or
bare
earth
barriers
to
separate
plots.
Invertebrate
sampling
was
undertaken
from
the
central
40
feet
and
60
feet,
respectively,
in
each
study,
but
this
still
means
that
samples
in
neighboring
plots
include
points
that
are
within
a
few
feet
of
each
other.
The
consequences
of
this
are
that
predatory
invertebrates
traverse
plots
to
find
prey
within
the
scale
of
the
whole
experiment.
If
these
animals
survive
pesticide
or
PIP
treatments,
then
the
pattern
of
invertebrate
abundance
across
the
study
reflects
treatments
to
a
degree,
but
also
reflects
this
redistribution
of
animals
between
plots.
The
data
therefore
contribute
very
little
towards
answering
the
study
objectives,
which
require
treatments
to
be
independent
of
each
other.

The
Panel
noted
that
if
toxic
chemicals
are
used
in
a
non­
Bt
treatment,
organisms
may
move
between
plots
and
succumb
to
these
chemicals
within
the
period
of
toxicity,
if
they
are
susceptible.
Reinvasion
from
neighboring
plots
then
occurs
and,
once
toxicity
has
declined,
effects
that
would
persist
on
an
agriculturally
relevant
scale
are
obliterated.
These
scale
and
reinvasion
effects
reduce
abundance
across
the
study
as
a
whole
and
limit
radically
the
ability
to
discriminate
between
treatments.
It
is
impossible
to
determine
from
small­
scale
field
tests,
how
toxic
pesticides
are
to
beneficials,
and
it
is
also
not
possible
to
detect
impacts
or
even
positive
attributes
of
unsprayed
Bt
plots.

To
confirm
that
scale
and
reinvasion
effects
may
have
compromised
the
WideStrike
field
test
data
package,
at
least
in
Arizona,
the
Panel
noted
that
chlorpyrifos
was
used
in
the
nontransgenic
control.
This
material
is
very
toxic
to
natural
enemies,
including
spiders,
and
the
Agency
should
reasonably
have
expected
significant
reductions
in
the
non­
Bt
plots
as
a
check
that
the
experiment
had
sufficient
sensitivity
to
discriminate
treatment
effects.

The
Panel
noted
that
the
numbers
of
organisms
in
the
Louisiana
study
seemed
to
be
extremely
low,
and
this
requires
explanation
if
data
are
submitted
for
review
in
the
form
of
a
final
report.
The
Panel
suggested
that
the
numbers
were
too
low
to
discriminate
treatment
effects.

The
Panel
also
noted
that
the
current
codification
of
the
risk
assessment
approach
does
not
allow
some
important
positive
issues
to
be
highlighted.
An
important
advantage
of
the
use
of
WideStrike
cotton
is
much
lower
application
rates
of
chemical
pesticides.
The
early
field
trial
Page
35
of
85
showed,
as
might
have
been
predicted,
that
the
number
of
species
was
higher
in
the
WideStrike
field
than
in
the
conventional
(
non­
transgenic)
cotton
culture
involving
heavier
application
of
chemical
pesticides.

With
respect
to
soil
degradation,
the
Panel
noted
that,
strictly
speaking,
the
protein
half
life
study
quantifies
the
temporal
change
in
activity,
not
of
intact
protein.
If
there
was
direct
evidence,
then
the
protein
concentration
was
decreasing.
Otherwise,
a
short
sentence
qualifying
the
results
would
be
helpful.
Although
unlikely,
some
protein
might
still
be
intact
but
bound
or
sequestered
in
some
manner
such
that
its
activity
is
not
expressed.

Agency
Charge
3.
The
Agency
has
sufficient
information
to
conclude
that
there
is
no
hazard
from
the
proposed
uses
of
WideStrike
cotton
to
non­
target
wildlife,
aquatic
and
soil
organisms.
However,
the
Agency
is
requesting
additional,
primarily
long
term
effects
data
that
were
recommended
by
previous
Panels
for
PIP
corn.
The
supplementary
studies
would
provide
additional
weight
to
support
the
Agency's
conclusions.

The
Panel
is
asked
to
comment
on
(
a)
the
scientific
value
of
the
proposed
additional
studies
that
are
identified
at
the
end
of
the
Environmental
Assessment
section,
including
avian
chronic
exposure
testing
and
multi­
year
field
and
soil
persistence/
terrestrial
expression
studies,
and
(
b)
the
applicability
of
these
data
to
PIP
cotton.

Panel
Response
The
Panel
was
asked
to
comment
on
the
scientific
value
of
additional
studies
to
assess
longer­
term
effects
associated
with
the
commercial
cultivation
of
WideStrike
cotton.
Three
aspects
were
considered:
(
1)
additional
avian
chronic
exposure
testing;
(
2)
multi­
year
field
studies
to
assess
potential
longer­
term
effects
of
WideStrike
cultivation
on
persistence
of
toxins
in
the
soil;
and
(
3)
populations
of
non­
target
organisms.
The
Panel
agreed
that
such
studies
are
of
considerable
scientific
value
and
finds
that
with
the
exception
of
avian
chronic
exposure
testing,
that
such
studies
are
applicable
to
Bt
cotton
and
should
be
considered.
While
the
Panel
is
aware
of
limiting
its
comments
to
risk
assessment,
and
not
risk
management
issues,
the
Panel
did
not
find
that
such
long­
term
studies
should
necessarily
be
required
as
a
condition
for
registration
of
WideStrike
cotton.
Overall,
the
likely
reduction
in
the
use
of
broader­
spectrum
insecticides
afforded
by
the
cultivation
of
Bt
cotton
in
general
is
likely
to
have
positive
effects
on
the
ecosystems.
Nonetheless,
longer­
term
and
broader­
scale
evaluation
of
PIP
crops
will
be
necessary
for
improved
ecological
understanding.
The
Panel
believed
strongly
that
long­
term
field
studies
should
not
proceed
without
some
guidance
relative
to
experimental
protocols
and
clearly
defined
endpoints.
Without
such
guidance,
we
are
unlikely
to
resolve
any
unexpected
detriments
or
benefits
associated
with
the
use
of
transgenic
crops.

Avian
Chronic
Exposure
Testing
Page
36
of
85
The
Panel
found
that
testing
of
multi­
year
field
effects
on
non­
target
vertebrates
is
generally
essential.
It
can
readily
be
seen
from
the
numerous
publications
Brewer
et
al.
(
1989,
1989a,
1990,
1992;
Tank
et
al.
1992,
1992a)
that
significant
annual
differences
are
apparent
with
the
same
pesticide
treatments
by
the
same
applicators
in
the
same
fields.
However,
Tier
1
testing
of
representative
vertebrate
models
shows
that
the
Bt
toxins
found
in
WideStrike
cotton
have
no
measurable
short­
term
adverse
effects.
Furthermore,
direct
exposure
to
Bt
toxins
through
direct
ingestion
of
seed
cotton
or
through
ingestion
of
intoxicated
arthropod
prey
by
avian
species
in
the
field
is
likely
to
be
minimal
because
birds
rarely
forage
in
cotton
fields
(
September
21­
24,
1999
FIFRA
SAP
Meeting).
Thus,
the
probability
of
direct
chronic
exposure
to
Bt
toxins
in
cotton
fields
is
unlikely.
Indirect
effects
on
avian
populations
are
possible
through
the
elimination
of
caterpillar
prey.
However,
since
foraging
in
cotton
is
minimal,
this
effect
is
unlikely
to
be
important.
As
a
result,
the
Panel
finds
little
scientific
merit
in
conducting
additional
chronic
exposure
testing.

Soil
Persistence
Bt
toxins
find
their
way
into
the
soil
either
through
root
exudates
or
the
breakdown
of
crop
residues
(
shoots
and
roots).
The
Panel
agreed
that
data
submitted
relative
to
soil
persistence
suggest
that
Cry1
proteins
are
quickly
degraded.
However,
persistence
of
the
Cry
proteins
goes
hand
in
hand
with
persistence
of
crop
residues
in
which
they
reside.
The
registrant
conducted
tests
on
toxin
persistence
in
two
ways;
(
1)
by
incorporating
purified
Cry1Ac
and
Cry1F
alone
and
in
combination
into
soil
and
(
2)
by
incorporating
finely
ground
WideStrike
plant
material
into
a
cotton
soil.
In
both
cases,
the
ability
to
detect
the
toxin
was
monitored
as
was
the
decline
over
time
of
the
toxicity
to
a
target
insect.

The
Panel
had
several
comments
about
this
approach.
First,
the
Panel
found
that
incorporation
of
purified
toxins
serves
only
to
demonstrate
that
the
toxins
are
biodegradable.
It
does
not
reflect
what
is
likely
to
happen
in
the
field,
nor
are
half­
life
values
derived
from
these
data
meaningful.
The
Panel
agreed
that,
provided
it
is
accessible,
the
toxin
will
be
broken
down.
Second,
use
of
finely
ground
plant
material
does
not
reflect
actual
field
situations
where
residues
are
incorporated
in
large
`
chunks'
that
require
shredding
to
initiate
decomposition.
This
discrepancy
may
be
mitigated
somewhat
by
the
fact
that
defoliant
use
and
mechanical
harvesting
in
cotton
production
does
in
fact
generate
reasonably
pulverized
material
after
mechanical
harvesting
is
complete.
Finely
ground
plant
material
obviates
the
role
and
possible
effects
on
shredders
and
also
provides
a
vast
surface
area
for
microbial
attack
thus
leading
to
enhanced
rates
of
decomposition.
Access
to
the
toxin
by
proteases
is
key
to
its
decomposition.
If
the
toxin
is
bound
in
residues,
sequestered
in
soil
aggregates
or
bound
to
clays
or
humic
material,
it
is
not
accessible
and
hence
will
accumulate
or
at
least
enjoy
longer
term
persistence
in
the
soil.
This
is
likely
to
be
ecologically
unimportant
as
a
toxin
or
toxicant
bound
in
such
a
manner
to
resist
biodegradation
is
unlikely
to
be
bioavailable
to
most
soil
organisms
(
Anhault
et
al.
2000;
Awata
et
al.,
2000;
Gevano
et
al.
2000,
2001;
Morrison
et
al.
2000).

A
number
of
studies
have
shown
that
Cry
toxins
from
Bt
resist
microbial
degradation
by
binding
to
clay
and
humic
acid
fractions
in
soil,
but
retain
insecticidal
activity
despite
being
bound
Page
37
of
85
(
Venkateswerlu
and
Stotzky,
1992;
Tapp
and
Stotzky,
1995;
Koskella
and
Stotzky,
1997;
Crecchio
and
Stotzky,
1998;
Tapp
and
Stotzky,
1998;
and
Saxena
et
al.,
1999).
Further,
Saxena
and
Stotzky
(
2000)
showed
that
larvicidal
activity
increased
with
the
length
of
time
Bt
corn
plants
were
grown,
suggesting
that
microbial
degradation
may
not
have
kept
pace
with
the
increased
root
exudation
of
the
toxin
into
soil
with
time.
An
alternative
explanation
may
be
that
the
toxin
was
unavailable
for
degradation
as
it
was
bound,
and
concentrated
with
time
on
the
surface
of
clays
and
humic
materials.

The
Panel
found
that
the
fate
of
the
toxin
in
different
soil
types
with
different
charge
characteristics
is
needed.
Studies
to
understand
the
mechanism
of
toxin
binding
in
relation
to
access
of
proteases
is
also
important.
There
is
little
information
on
the
fate
and
persistence
of
Bt
toxins
in
the
field,
with
the
exception
of
one
field
study
reporting
that
while
the
levels
of
Cry1Ab
toxin
in
no­
till
corn
decreased
to
0.3%
of
the
initial
concentration
after
200
days,
degradation
was
initially
delayed
and
lower
in
a
conventionally
tilled
system
(
Zwahlen
et
al.,
2003a,
b).
However,
Head
et
al.
(
2002)
did
not
detect
Cry1Ac
toxin
in
soil
samples
taken
about
90
days
after
the
last
planting
from
fields
planted
with
Bt
cotton
for
several
consecutive
years.
The
Panel
agreed
that
more
data
are
clearly
needed
here
as
this
is
also
a
point
of
ecological
as
well
as
public
concern.
These
findings
are
consistent
with
and
largely
echo
the
concerns
expressed
by
the
August
27,
2002
SAP
and
the
current
Panel
fully
agree
with
the
recommendations
forwarded
by
this
prior
SAP
for
future
multi­
year
testing.
The
current
Panel
further
suggested
that
additional
studies
should
consider
the
role
of
residue
degradation,
residue
dry
matter
partitioning
and
major
pools
of
structural
carbon,
and
measurement
of
effects
on
residue
shredding
organisms.

Non­
Target
Effects
Not
withstanding
some
of
the
flaws
outlined
under
Question
2
to
this
Panel,
Tier
1
testing
under
current
guidelines
has
demonstrated
no
measurable
short­
term
hazard
from
the
two
Cry
toxins
expressed
in
WideStrike
cotton
to
representative
organisms
from
terrestrial,
soil
and
aquatic
systems.
In
addition,
Bt
Cry
toxin
proteins
are
susceptible
to
metabolic,
microbial,
and
abiotic
degradation
once
they
are
ingested
or
excreted
into
the
environment.
As
a
result,
unlike
many
insecticides
that
persist
and
accumulate
through
the
food
chain,
Cry
proteins
are
not
expected
to
have
the
potential
for
bioaccumulation.

Thus,
the
Panel
agreed
with
the
Agency's
conclusion
that
the
Cry
toxins
in
WideStrike
cotton
pose
no
short­
term
hazards.
However,
the
Panel
also
agreed
with
the
Agency
that
Tier
1
testing
is
insufficient
for
assessing
the
potential
for
long­
term
environmental
effects.
Assessing
this
level
of
hazard
will
require
carefully
designed
field
studies
conducted
over
longer
periods
of
time
and
at
relevant
spatial
scales.
The
bulk
of
evidence
(
Groot
and
Dicke
2002,
Conner
et
al.
2003)
suggested
that
most
of
the
effects,
if
any,
from
Bt
toxins
in
the
field
will
be
either
sublethal
and/
or
result
largely
from
indirect
effects
via
complex
community
interactions.
Tier
1
is
limited
to
direct
exposures
and
largely
measures
only
lethal
effects.
Sublethal
effects
are
usually
subtle
but
may
have
significant
effects
on
longer­
term
population
dynamics
and
community
level
interactions
(
Elzen
1989;
Croft
1990).
Examples
include
reductions
in
adult
longevity,
reductions
in
reproductive
output,
increased
susceptibility
to
other
environmental
factors,
and
altered
searching
Page
38
of
85
behavior
in
the
case
of
predators
or
parasitoids.
Although
some
of
these
factors
could
be
examined
in
laboratory
studies,
their
implications
and
impacts
can
only
be
realistically
evaluated
with
longer­
term
field
studies.
Indirect
effects
are
more
complex
and
likely
to
be
more
pervasive
due
to
the
large
number
of
interactions.
As
an
example,
the
soil
food
web
represents
an
intricate
set
of
feeding
relationships.

To
evaluate
the
risk
to
these
organisms
of
adverse
effects
of
the
stacked
Cry1Ac
­
Cry1F
genes
requires
a
consideration
of
the
hazard
these
toxins
present
and
the
level
of
exposure
the
organisms
are
likely
to
have.
Extant
research
has
shown
that
there
is
no
direct
toxicity
to
soil
bacteria
and
fungi,
rather,
these
are
the
organisms
responsible
for
active
degradation
of
the
Cry
toxins
in
soil.
Higher
order
feeders
such
as
protozoa,
nematodes,
collembolans,
mites
and
earthworms
have
no
receptors
for
toxin
binding
and
hence
are
unlikely
to
be
adversely
affected
by
the
presence
of
the
toxin
in
the
plant
tissues
being
broken
down
and
thus
no
hazard.
Field
evidence
with
Bt
corn
supports
this
point
(
Devare
et
al.
2004).

In
general,
the
evidence
presented
by
the
registrant
and
evaluated
by
the
Agency
for
collembolan
and
earthworms
also
indicated
a
lack
of
direct
toxicity.
However,
the
Panel
questioned
the
veracity
of
the
collembola
data,
because
the
first
two
tests
on
the
Cry1Ac
pure
toxin
indicate
an
effect
on
collembola
fecundity,
which
was
reduced
by
close
to
50%.
When
the
two
purified
toxins,
Cry1Ac
and
Cry1F,
were
tested
together
or
plant
tissue
expressing
the
Cry1Ac
toxin
were
fed
to
collembolans,
no
adverse
effects
were
detected.
The
report
submitted
to
the
Agency
for
review
suggested
the
Cry1Ac
toxin
may
have
been
contaminated
in
some
way.
If
the
Cry1Ac
preparation
used
was
compromised,
why
was
the
reduced
fecundity
not
also
observed
in
the
combined
test?
Long­
term
adverse
effects
are
not
precluded
by
these
results.
Changes
in
the
biological
quality
of
soil
accrue
slowly.
Even
in
conservation
agriculture,
no­
till
systems
and
legume
rotations,
benefits
may
not
be
detectable
for
5­
10
years.
In
order
to
have
confidence
that
WideStrike
does
not
have
adverse
ecological
effects,
long­
term,
post­
release
monitoring
is
needed.
As
already
noted,
effects
of
the
toxin
may
be
sublethal
or
cumulative
and
the
possibility
of
this
needs
to
be
ascertained.

One
aspect
of
crop
assessment
that
has
not
been
dealt
with
adequately
in
the
current
analysis
was
whether
there
are
changes
in
dry
matter
partitioning
or
carbon
allocation
within
the
WideStrike
cotton
plant
tissues
that
could
influence
the
decomposition
dynamics
of
the
crop
residues
after
harvest.
The
soil
microbial
community
relies
on
fixed
C
in
plant
material
for
energy
and
nutrients
for
metabolism,
growth
and
reproduction.
Different
forms
of
C
are
more
or
less
readily
decomposed.
Simple
sugars
and
amino
acids
represent
very
labile
C
pools,
whereas
cellulose
and
hemi­
cellulose
are
more
stable
pools
and
lignin
the
more
recalcitrant
pool.
Should
the
incorporation
of
the
Cry
genes
lead
to
unintended
changes
in
the
way
that
structural
C
is
allocated
within
the
plant,
then
decomposition
dynamics
of
the
material
could
be
altered.
An
increase
in
lignin,
for
example,
would
result
in
a
slower
rate
of
decomposition.
This
was
a
result
reported
by
Stotzky
(
2000).

Stotzky
(
2000)
reported
that
CO2
evolution
from
soil
amended
with
ground
Bt
corn
biomass
expressing
the
Cry1Ab
gene
was
significantly
lower
than
that
from
soil
containing
unmodified,
Page
39
of
85
isogenic
corn
biomass.
While
the
mechanism
by
which
Cry1Ab
might
depress
microbial
activity
is
unclear,
unintended
changes
in
structural
C
allocation
within
the
plant
is
a
possibility
and
Saxena
and
Stotzky
(
2001)
went
on
to
conclude
that
increased
lignin
content
in
the
corn
was
responsible
for
their
observations.
Stotzky
has
also
reported
that
finely­
ground
Bt
rice,
potato,
and
tobacco
residues
degrade
more
slowly
than
their
non­
Bt
counterparts
in
laboratory
studies
for
reasons
that
are
unclear
(
Stotzky
2002),
as
no
differences
in
lignin
content
were
detected
for
these
other
crops.
Analyzing
transgenic
crop
residues
for
proportions
of
structural
C
in
different
forms
may
be
useful
for
addressing
concerns
for
potential
differences
in
the
rates
of
residue
decomposition
between
Bt
crops
and
their
non­
Bt
isolines.
The
relevant
comparisons
would
be:
(
1)
insect
affected
nontransgenic
cotton,
as
C
partitioning
is
grossly
affected
by
insect
feeding;
and
(
2)
insecticide
protected
non­
transgenic
cotton,
the
commercial
standard.
Once
structural
C
forms
have
been
quantified,
a
soil
organic
matter
dynamics
model
such
as
Roth
C
or
the
Century
model
could
be
used
to
predict
if
any
change
in
decomposition
dynamics
would
be
expected.
The
Panel
noted
that
finely
ground
plant
material
is
not
an
appropriate
means
to
assess
decomposition
dynamics
that
might
be
operating
in
the
field,
nor
is
it
the
best
means
by
which
to
evaluate
toxin
persistence
(
as
discussed
above).
This
is
another
reason
why
long­
term
post­
release
monitoring
is
important
and
why
it
should
include
an
analysis
of
the
fate
of
the
cotton
residues
under
commercial
field
conditions.

In
and
of
itself,
slower
decomposition
of
residues
is
not
a
bad
thing.
Conservation
agriculture
strives
to
increase
soil
organic
matter
content
and
slower
decomposition
might
serve
to
be
beneficial
for
overall
soil
quality.
However,
a
build­
up
of
crop
residues
may
also
be
indicative
of
non­
target
toxicity
to
a
key
group
of
soil
decomposers
called
the
shredders.
These
are
microarthropods
for
the
most
part,
mites
being
a
prime
example.
Adverse
effects
on
mites
could
be
quite
destabilizing
to
the
soil
food
web,
and
an
effect
such
as
this
might
not
be
picked
up
without
longerterm
field
monitoring.
Some
consideration
should
be
given
to
using
mites
as
a
target
group
for
toxicity
testing
due
to
their
important
role
in
communinution.

Overall,
long­
term
testing
on
relevant
scales
will
only
be
possible
once
registration
is
granted
and
WideStrike
can
be
grown
commercially.
The
Panel
recommendations
for
long­
term
evaluation
are
consistent
with
the
spirit
of
recommendations
made
by
NRC
(
2002)
and
several
previous
FIFRA
SAPs
(
e.
g.,
December
1999,
August
2002).
The
NRC,
for
example,
envisioned
long­
term
field
testing
of
non­
target
effects
and
other
environmental
and
pest
management
concerns
as
part
of
what
they
call
"
postcommercialization
validation
testing".
The
main
thrust
of
such
studies
would
be
to
conduct
appropriate,
hypothesis­
driven,
and
adequately
replicated
field
studies
in
large
commercial­
scale
plots
for
the
purpose
of
determining
the
accuracy
and
adequacy
of
pre­
commercialization
testing.
This
same
general
reasoning
was
expressed
by
the
August
2002
SAP.
A
critical,
and
still
unresolved
issue
that
has
been
put
forward
by
both
the
NRC
(
2002)
and
prior
SAPs
is
the
lack
of
guidelines
for
conducting
long­
term
field
studies
such
that
these
studies
will
provide
meaningful
and
robust
results.
For
example,
(
1)
which
methods
and
types
of
data
should
be
collected;
(
2)
what
spatial
scale
should
be
examined;
(
3)
which
experimental
designs
are
most
appropriate;
and
(
4)
most
importantly,
what
criteria
will
be
used
to
determine
impact
or
the
lack
thereof?
The
Panel
believes
that
long­
term
field
studies
should
not
proceed
without
some
guidance
relative
to
protocols
and
clearly
defined
endpoints.
Page
40
of
85
The
Panel
recognized
that
field
studies
have
been
and
will
continue
to
be
conducted
to
examine
non­
target
effects.
A
number
of
field­
level
studies
have
been
published
on
transgenic
corn
and
cotton
expressing
various
Cry
proteins
in
the
last
few
years
(
e.
g.,
Candolfi
et
al.
2004;
Musser
and
Shelton
2003;
Men
et
al.
2003;
Al­
Deeb
et
al.
2003;
Al­
Deeb
and
Wilde
2003;
Dively
and
Rose
2002;
Moar
et
al.
2002;
Naranjo
2002
and
Devare
et
al.
2004).
In
general,
these
studies
largely
support
Tier
I
hazard
testing
results
reported
to
the
Panel
and
demonstrate
no
measurable
adverse
effects.
Several
of
these
studies
have
been
relatively
long
term
(>
3
years).
However,
in
light
of
the
issues
raised
by
Jepson
(
2002)
and
previous
FIFRA
SAPs
regarding
scale,
design
and
sampling
issues,
it
remains
unclear
whether
these
studies
would
be
considered
sufficiently
robust
to
definitively
address
population
and
community
level
effects
if
such
effects
exist.
Nonetheless,
scientists
are
using
the
growing
body
of
data
to
address
some
of
the
issue
of
concern
to
the
Agency
and
many
of
these
are
being
conducted
independent
of
registrant
needs.
It
seems
likely
that
these
types
of
studies
will
continue
to
be
conducted
with
or
without
the
support
of
industry
and
it
would
be
prudent
if
the
scientific
community
could
provide
guidelines
that
would
ensure
and
perhaps
improve
the
quality
and
value
of
the
results
for
both
risk
assessment
and
broader
ecological
understanding.
Issues
include
plot
sizes,
replication,
sampling
methods,
sampling
intensity,
appropriate
positive
and
negative
controls,
taxonomic
coverage
and
clearly
specified
endpoints.
The
Panel
suggested
that
some
of
this
guidance
may
be
extractable
from
existing
and
ongoing
studies.

One
of
the
more
significant
aspects
of
field
testing
that
needs
to
be
addressed
is
spatial
scale.
Jepson
and
colleagues
(
Jepson
and
Thacker
1990;
Sherratt
and
Jepson
1993;
and
Jepson
2002)
provide
compelling
evidence
that
spatial
and
temporal
components
must
be
considered
in
evaluating
toxicological
effects
of
insecticides
and
that
this
is
largely
driven
by
the
mobility
of
the
species
under
consideration
and
their
phenology
and
ecological
requirements.
While
the
Panel
believed
that
it
would
be
desirable
to
conduct
studies
at
spatial
scales
mimicking
commercial
production
practices,
it
may
not
always
be
feasible
to
do
so
based
on
economic
and
logistical
considerations.
Directly
working
in
commercial
fields
often
removes
an
important
level
of
control
necessary
to
implement
a
consistent
experimental
design.
Insecticides
needed
for
other
pests
and
other
production
constraints
may
interfere.
On
the
other
hand,
establishing
independent
commercial­
scale
research
plots
may
be
cost­
prohibitive,
especially
if
studies
are
to
be
conducted
with
sufficient
replication.
Very
large
plots
may
provide
the
necessary
independence,
but
also
introduce
significant
within­
plot
variation
thereby
requiring
additional
sampling
for
precise
estimation
(
Jepson
1994).
Although
more
commercial­
scale
study
plots
may
be
desirable,
the
issue
of
independence
versus
heterogeneity
will
need
to
be
assessed
relative
to
plot
size.
How
large
do
plots
really
need
to
be?
This
will
likely
be
system
specific
and
depend
on
the
taxa
being
assessed.
For
example,
relatively
small
plots
(<
0.5
acres)
may
be
sufficient
for
studying
some
predatory
arthropods.
Naranjo
(
2002)
found
that
0.25
acre
plots
were
sufficient
to
determine
differences
in
a
positive
control
of
broad­
spectrum
insecticides
for
a
complex
of
20
arthropod
predator
taxa,
indicating
that
there
is
a
fairly
high
degree
of
fidelity
within
plots
of
this
size.
Prasifka
et
al.
(
2004)
has
shown
that
Hippodamia
convergens
readily
move
from
sorghum
to
cotton
on
a
commercial
scale,
but
they
also
showed
fidelity
of
this
beetle
in
cotton
even
in
the
absence
of
its
preferred
aphid
prey.
Many
of
the
predatory
arthropods
found
in
cotton
are
generalist
feeders
and
are
relatively
insensitive
to
the
densities
of
any
one
prey.
In
the
western
US,
pink
bollworm,
the
main
target
of
Bt
cotton,
is
essentially
invulnerable
to
natural
enemies.
Page
41
of
85
Thus,
the
removal
of
this
insect
from
the
system
has
little
effect
on
natural
enemy
populations.
Other
Lepidoptera
in
this
area
are
relatively
rare
and
their
presence
is
inconsistent.
Abundant
prey
in
the
form
of
thrips,
whiteflies
and
Lygus
bugs
are
unaffected
by
Bt
cotton.
Thus,
predators
may
not
be
affected
by
changes
in
prey
density
that
may
cause
them
to
disperse.

Replication
is
another
critical
component
of
the
experimental
design
as
it
bears
directly
on
the
issue
of
statistical
power,
or
the
ability
to
detect
differences
that
may
be
relatively
small
between
Bt
and
non­
Bt
plots.
This
aspect
will
clearly
be
system­
specific.
Perry
et
al.
(
2003)
estimated
that
a
sample
size
of
60
may
be
necessary
to
measure
environmental
effects
of
herbicide­
tolerant
crops
in
the
UK,
and
higher
sample
sizes
were
suggested
by
NCR
and
the
August
2002
SAP.
In
a
recent
meta­
analysis
of
a
5­
year
study,
a
sample
size
of
40
was
only
sufficient
to
discern
a
20%
significant
difference
between
densities
of
arthropod
predators
in
non­
Bt
versus
Bt
cotton
expressing
Cry1A.
In
the
long
run,
there
will
likely
be
a
trade­
off
relative
to
plot
size
and
replication
and
this
will
need
to
be
determined
on
a
case
by
case
basis
depending
on
the
taxa
to
be
examined
and
the
ecological
setting.
In
addition,
it
may
be
possible
to
perform
retrospective
power
analyses
on
many
of
the
past
and
ongoing
non­
target
studies
that
would
be
useful
in
developing
guidelines
for
future
studies.

Sampling
methods
are
another
important
consideration.
For
example,
sweep
nets
and
beat
clothes
are
a
common
method
of
measuring
relative
abundance
of
foliar­
dwelling
arthropods
in
row
crops
such
as
cotton,
and
these
methods
are
probably
sufficient
for
comparative
studies.
Absolute
sampling
methods
such
as
whole
plant
counts
would
be
desirable,
but
such
methods
are
likely
to
be
much
more
costly
due
to
aggregated
arthropod
distributions.
In
the
supplemental
multi­
year
field
studies
being
requested
by
the
Agency,
protocol
suggests
the
use
of
sweep
nets,
sticky
traps
and
pitfall
traps
to
assess
effects
on
non­
target
arthropods.
These
methods
will
likely
produce
useful
data,
but
careful
consideration
will
need
to
be
given
to
the
number
of
samples
collected
in
each
experimental
unit.
The
interplay
of
plot
size,
replication,
sampling
method,
and
sample
size
for
these
methods
will
ultimately
determine
the
statistical
power
and
the
ability
of
the
assessment
to
discern
relevant
differences.

Taxonomic
resolution
is
another
aspect
that
is
critical
to
address.
Many
of
the
existing
non­
target
field
studies
have
focused
on
selected
taxa.
Such
an
approach
is
generally
necessary
as
there
are
well
over
500
species
of
arthropods
alone
inhabiting
cotton
across
the
southern
tier
of
the
US,
and
the
soil
invertebrate
community
is
large
and
diverse
as
well.
Predaceous
and
parasitic
natural
enemies
have
been
the
focus
in
many
studies
of
Bt
crops.
Such
species
are
consistently
present
in
the
cotton
system;
many
occur
at
moderate
to
high
densities,
and
further,
their
standing
in
higher
trophic
levels
may
provide
insight
into
effects
in
lower
trophic
levels
and
may
act
as
sensitive
indicators
of
community
change.
The
Panel
suggested
that
assessment
of
such
taxa
may
be
useful
for
multi­
year
testing.
The
supplemental
studies
by
the
registrant
provided
relatively
broad
taxonomic
coverage.
In
particular,
the
study
in
Maricopa,
Arizona
attempted
to
examine
all
above­
ground
taxa.
Such
studies
complicate
analyses,
and
careful
thought
will
need
to
be
applied
in
interpreting
the
meaning
of
any
observed
differences.
The
resolution
of
a
large
percentage
of
the
taxa
will
likely
be
too
poor
to
enable
definitive
conclusions.
However,
the
Panel
recognized
that
studies
like
this
may
be
helpful
in
narrowing
the
taxonomic
coverage
of
future,
more­
targeted
evaluations.
The
choice
of
taxa
to
examine
may
be
system
specific
and
should
include
consideration
of
likely
exposure
to
Bt
toxins
through
either
direct
or
indirect
Page
42
of
85
means.

Final
considerations
include
the
inter­
related
issues
of
study
length
and
desired
end­
points
for
assessment.
With
clear
focus
on
short­
term
hazard,
Tier
1
testing
generally
uses
direct
mortality
within
a
short
period
of
time
as
the
designated
endpoint.
Such
simple
endpoints
are
not
likely
to
be
useful,
or
even
measurable,
in
long­
term
field
assessments
for
environmental
effects.
Instead,
effects
(
either
positive
or
negative)
are
likely
to
be
manifested
as
reductions
or
increases
in
population
densities
of
specific
taxa,
changes
in
diversity,
or
changes
in
community
organization
and
function.
As
indicated
above
for
the
soil
system,
such
changes
may
be
slow
and
take
many
years
to
shape
observable
changes.
However,
even
when
changes
do
occur,
what
criteria
should
be
used
to
judge
a
positive,
negative
or
neutral
effect?
For
example,
a
five­
year
study
on
the
non­
target
effects
of
Cry1A­
expressing
cotton
on
20
arthropod
predator
taxa
(
Naranjo
2002,
in
part)
showed
a
general
pattern
of
population
reductions
in
Bt
cotton.
Changes
in
this
select
community
in
Bt
cotton
varied
from
a
negative
change
of
19%
in
one
year
to
a
positive
change
of
15%
in
another
year
when
compared
to
non­
Bt
cotton.
Irrespective
of
statistical
significance,
the
predator
community
declined
by
8%
averaged
over
the
entire
5
years.
This
average
value
would
have
been
different
after
2,
3
or
4
years
and
would
likely
be
different
if
the
study
proceeds
another
5
years.
In
contrast,
the
positive
control
treatments
based
on
use
of
broad­
spectrum
insecticides
caused
an
average
reduction
of
62%.
During
this
same
period,
functional
studies
on
the
natural
enemy
community
showed
no
statistically
significant
differences
in
rates
of
predation
and
parasitism
between
unsprayed
Bt
and
non­
Bt
cotton
on
two
different
pests.
Irrespective
of
statistical
significance,
rates
of
predation
on
pink
bollworm
eggs
and
pupae
declined
4­
6%
on
average
in
Bt
cotton;
parasitism
and
predation
of
whitefly
nymphs
increased
5­
8%
on
average
in
Bt
cotton.
Has
this
system
been
impacted
either
positively
or
negatively
by
the
use
of
Bt
cotton?

Horizontal
Gene
Transfer
Some
concern
has
been
expressed
over
the
potential
for
transgenes
to
move
from
Bt
crops
to
soil
microorganisms.
In
this
regard,
we
should
consider
the
likelihood
of
this
occurring
in
the
`
bestcase
scenario
 
that
is
between
closely
related
bacteria
in
the
soil
and
contrast
this
with
the
probability
presented
by
genes
located
in
the
plant
genome.
There
is
sufficient
genetic
similarity
within
this
clade
to
question
whether
B.
cereus
and
B.
thuringiensis
are
distinct
species
because
what
defines
B.
thuringiensis
as
distinct
from
B.
cereus
is
the
presence
of
the
plasmid
in
Bt
that
contains
the
Cry
toxin
and
associated
cytotoxic
genes.
If
the
plasmid
is
transferred
to
B.
cereus,
it
becomes
indistinguishable
from
Bt.
How
frequently
this
plasmid
is
exchanged
among
members
of
this
clade
in
soil
is
unknown.
B.
anthrasis
is
closely
related,
but
distinguished
by
carrying
a
different
plasmid
with
a
different
set
of
toxin
genes.
Recombination
or
lateral
gene
transfer
between
Bt
and
B.
anthracis
has
not
been
reported.
Thus
in
nature
such
gene
flow
appears
to
be
restricted
even
between
very
closely
related
species.
Given
this,
what
is
the
likelihood
that
a
full
length
toxin
gene
will
be
transferred
intact
to
a
recipient
bacterium?
First,
the
needed
flanking
insertion
regions
would
need
to
be
present
in
the
recipient,
which
in
itself,
is
highly
unlikely.
Second,
there
would
need
to
be
sufficient
selective
pressure
operating
to
drive
the
acquisition
of
the
genes
by
the
recipient
bacterium.
Such
selective
pressure
is
highly
unlikely
in
the
soil
environment
is.
Whilst
it
cannot
be
completely
ruled
out,
the
likelihood
of
Bt
genes
transfer
from
Page
43
of
85
a
transgenic
crop
to
soil
is
exceedingly
small.
Page
44
of
85
WideStrike
Cotton
Insect
Resistance
Management
Agency
Charge
1.
Dose.
Three
methods
(
two
laboratory
and
one
field)
outlined
by
USEPA's
Scientific
Advisory
Panel
(
1998)
were
used
to
demonstrate
that
WideStrike
cotton
expresses
a
high
dose
of
Cry1Ac
and
Cry1F
against
tobacco
budworm
(
Heliothis
virescens,
TBW).
Dow
AgroSciences
(
Dow)
employed
one
laboratory­
based
and
one
field­
based
method
to
demonstrate
that
WideStrike
cotton
has
a
high
dose
(
Cry1Ac
only,
Cry1F
is
non­
toxic)
against
pink
bollworm
(
Pectinophora
gossypiella,
PBW)
Results
of
two
field
studies
indicate
that
WideStrike
cotton
produces
a
moderate
dose
against
cotton
bollworm
(
Helicoverpa
zea,
CBW),
but
a
very
high
level
of
control
(
94%).
The
Agency
concluded
that
WideStrike
cotton
expresses
a
high
dose
of
Cry1F
and
Cry1Ac
against
TBW
(
Cry1Ac
alone
expresses
a
high
dose
and
Cry1F
a
nearly
high
dose);
a
moderate
dose
of
Cry1F
and
Cry1Ac
against
CBW,
and
a
high
dose
of
Cry1Ac
against
pink
bollworm.

The
Agency
asks
the
SAP
to
comment
on
the
Agency's
analysis
of
dose
for
TBW,
CBW,
and
PBW,
the
likelihood
that
resistance
will
be
inherited
as
a
recessive
trait,
and
its
impact
on
insect
resistance
management
for
WideStrike
cotton.

Panel
Response
The
Panel
addressed
this
question
with
an
understanding
that
a
combination
of
field
and
laboratory
studies
performed
by
the
registrant
had
evaluated
the
toxicities
of
the
Cry1Ac
and
Cry1F
expressed
in
WideStrike
cotton
against
TBW,
CBW,
and
PBW.
The
Panel
also
provided
a
detailed
response
of
field
efficacy
and
high
dose
considerations.

The
combination
of
the
laboratory
and
field
approaches
clearly
demonstrated
that
the
combined
expression
of
Bt
proteins
in
WideStrike
cotton
meet
the
Agency's
definitions
of
high
dose
for
PBW
and
TBW.
Based
on
existing
knowledge
of
the
reduced
susceptibility
of
CBW
to
Cry1Ac,
the
registrant
did
not
pursue
the
full
spectrum
of
high
dose
studies
on
CBW,
but
still
managed
to
show
good
field
control
relative
to
non­
Bt
cotton.
In
general,
there
was
good
replication
within
trials,
and
with
some
tests,
replication
across
states,
altogether
building
a
robust
data
set
in
support
of
the
contention
by
the
registrant
that
toxin
expression
in
WideStrike
cotton
constitutes
a
high
dose
for
PBW
and
TBW.

One
caveat
with
the
field
efficacy
trials
was
that
9
of
19
trials
were
conducted
using
artificial
infestation
techniques
in
the
trial
plots.
While
the
success
of
this
approach
was
gauged
in
some
trials
by
substantial
infestations
that
resulted
in
the
unsprayed
control
plots,
and
is
often
required
to
obtain
a
sufficient
infestation
in
the
field,
the
general
practice
of
using
laboratory
colonies
in
field
studies
should
be
avoided
when
possible
due
to
concerns
about
reduced
vigor
and
fitness
in
frequently
inbred
lab
strains.
The
potential
risk
is
that
genetically
compromised
lab
strains
may
not
provide
the
full
challenge
to
a
breeding
line
as
do
natural
field
populations,
and
therefore
may
not
represent
a
full
test
of
the
resistance
in
the
breeding
line.
In
future
Page
45
of
85
considerations
of
such
data,
the
Agency
may
want
to
consider
guidelines
for
use
of
laboratory
colonies
to
better
guarantee
quality
control.
Poorly
managed
colonies
have
a
greater
probability
of
producing
results
consistent
with
a
high
dose.

Having
established
the
high
dose
trait
of
WideStrike
cotton
against
PBW
and
TBW,
and
given
the
broad
history
to
date
of
resistance
being
recessive
wherever
resistant
larvae
can
survive
on
Bt
transgenic
hosts
(
Tabashnik
et
al.
2003),
the
Panel
concluded
that
it
is
valid
to
assume
that
resistance
will
likely
be
inherited
as
a
recessive
trait.
CBW
is
more
tolerant
of
both
proteins
and
it
seems
possible
that
resistance
will
be
less
recessive.
Without
abundant
alternate
host
plants
that
do
not
express
these
same
Bt
genes,
CBW
would
be
more
prone
to
resistance.

Based
on
these
assumptions
and
on
the
criteria
to
determine
functional
dominance
established
by
previous
FIFRA
SAPs,
the
combined
expression
of
Bt
proteins
in
WideStrike
cotton
meets
the
Agency's
definitions
of
high
dose
for
PBW
and
TBW.
In
addition,
reasonable
doses
of
the
combined
protein
were
evident
for
control
of
CBW.
Based
on
the
high
dose
evidence,
the
Panel
concluded
that
it
is
valid
to
assume
that
resistance
occurring
in
PBW
or
TBW
will
likely
be
inherited
as
a
recessive
trait.
However,
CBW
is
more
tolerant
of
both
proteins
and
it
seems
possible
that
resistance
will
be
less
recessive.
These
studies
support
the
position
for
durable
sprayable
Bts
in
organic
cotton
production
and
other
crops,
such
as
tomatoes.
The
same
high
dose/
refuge
strategy
practiced
thus
far
as
a
resistance
management
approach
for
Bollgard
cotton
should
be
applied
for
WideStrike.

WideStrike
cotton
does
appear
to
offer
a
high
dose
for
TBW,
a
high
dose
of
Cry1Ac
for
PBW,
and
reasonable
doses
of
Cry1F
and
Cry1Ac
for
CBW.
The
same
high
dose/
refuge
strategy
practiced
thus
far
as
a
resistance
management
approach
for
Bollgard
cotton
should
be
applied
for
WideStrike.

Field
Efficacy
Five
field
efficacy
trials
each
were
completed
for
TBW
and
CBW
while
a
single
trial
was
run
for
PBW
in
Arizona.
In
addition
to
these
three
principal
pests,
field
efficacy
data
was
also
collected
for
5
other
cotton
pests
(
Dow
Agrosciences
submission:
"
Efficacy
of
Cry1F/
Cry1Ac
Cotton
Against
a
Wide
Range
of
Lepidopterous
Pests",
2002).
By
measuring
cotton
boll
damage
and
infestations,
the
collective
field
data
were
conclusive
in
establishing
that
WideStrike
cotton
provided
superior
control
of
TBW
and
PBW
relative
to
either
sprayed
or
untreated
plots
of
the
conventional
cotton
line
PSC355.
In
the
case
of
CBW,
boll
damage/
infestation
levels
in
MXB­
13
were
equal
to
or
less
than
the
unsprayed
control
in
96%
of
the
comparisons
(
53%
significantly
less
damaged)
while
58%
of
the
comparisons
showed
equal
or
less
damage
than
the
sprayed
control.
Although
some
trials
were
conducted
in
2001
of
two
and
one
gene
plants
that
showed
lower
efficacy,
these
trials
were
believed
to
be
from
early
selections
that
likely
had
some
nonexpressing
plants.

High
Dose
Page
46
of
85
To
determine
if
the
two
Bt
proteins
expressed
in
WideStrike
cotton
meet
one
of
the
Agency's
definitions
of
high
dose,
i.
e.
25x
the
dose
required
to
kill
99%
of
susceptibles,
laboratory
bioassays
that
incorporated
a
25­
fold
dilution
of
carpal
wall
tissue
or
leaf
tissue
were
used
against
neonates
of
PBW
and
TBW,
respectively.
Extremely
high
mortalities
and
growth
inhibition
were
observed
against
both
species.
An
additional
test
using
2­
day
old
TBW
larvae
determined
to
be
approximately
25­
fold
more
tolerant
than
neonate
larvae
was
conducted
using
fresh
WideStrike
leaf
tissue
grown
in
Mississippi
or
California,
as
well
as
2
different
laboratory
colonies
of
TBW.
The
leaves
grown
in
Mississippi
proved
to
be
more
toxic
than
those
grown
in
California,
but
across
both
studies
mortality
was
considered
to
be
extremely
high.
The
field
component
of
the
high
dose
testing
was
performed
for
both
PBW
and
TBW
to
determine
if
field
mortality
on
WideStrike
met
the
other
Agency
definition
of
high
dose,
i.
e.
mortality
>
99.99%.
For
PBW,
only
one
third
instar
was
found
in
a
cotton
boll
out
of
>
12,000
infested.
Inspections
of
squares
and
bolls
using
three
different
methods
revealed
three
live
TBW
neonates
out
of
more
than
>
270,000
fruit
infested,
thus
demonstrating
field
mortality
meeting
the
high
dose
definition.
No
specific
studies
were
attempted
in
order
to
satisfy
the
high
dose
criterion
for
WideStrike
against
CBW.

Also
of
interest
for
the
future
are
the
doses
for
other
species
of
Lepidoptera,
which
have
not
been
of
concern
to
date
for
resistance
management
due
to
their
broad
host
range:

Spodoptera
frugiperda
(
Fall
armyworm):
Average
survival
was
8.1%
in
field
test.
Exposed
neonate
larvae
did
not
develop.
Therefore
a
high
level
of
control
of
this
species,
but
not
a
high
dose.

S.
exigua
(
Beet
armyworm):
Controlled
in
the
field.
However,
the
Panel
questioned
whether
this
was
a
moderate
dose?

S.
eridania
(
Southern
Armyworm):
Excellent
control
with
less
than
0.8%
defoliation
observed
in
field
studies.

Pseudoplusia
includens
(
Soybean
looper):
91­
98%
controlled
in
the
field.
A
high
level
of
control
but
not
a
high
dose.

While
not
of
concern
unless
and
until
more
crops
are
transformed
with
Bt
genes,
these
are
important
background
data
for
considering
resistance
management
strategies
for
these
species.

Agency
Charge
2.
Cross­
resistance.
Resistance
to
Bt
proteins
can
occur
through
several
different
mechanisms.
Alteration
of
binding
receptors
has
been
the
most
common
mechanism
observed.
The
binding
patterns
of
the
Cry1F
and
Cry1Ac
proteins
in
CBW
and
TBW
indicate
there
are
shared
and
unique
binding
sites.
In
TBW,
Cry1Ac
binds
to
at
least
three
receptors,
while
Cry1F
binds
to
at
least
two,
only
one
of
which
binds
Cry1Ac.
In
CBW,
Cry1Ac
and
Cry1F
each
bind
to
at
least
four
receptors,
of
which
two
are
shared.
Page
47
of
85
For
CBW,
approximately
60%
of
Cry1Ac
binding
is
to
receptors
that
also
bind
Cry1F,
and
the
remaining
40%
of
Cry1Ac
binding
is
to
receptors
that
do
not
bind
Cry1F.
Incomplete
shared
binding
is
expected
to
lead
to
incomplete
cross­
resistance
when
resistance
is
mediated
by
receptor
changes.
Thus,
a
mutation
in
a
gene
that
codes
for
a
receptor
that
binds
both
insecticidal
control
proteins
(
ICPs)
will
not
prevent
all
binding
of
either
ICP
and
thus
alone
will
not
allow
high
survival
of
the
insect
bearing
even
two
copies
of
it,
on
WideStrike
(
Cry1F/
Cry1Ac)
cotton
plants.

The
Agency
asks
the
SAP
to
comment
on
EPA's
conclusion
that
incomplete
shared
binding
of
Cry1Ac
and
Cry1F
receptors,
in
TBW
and
CBW,
is
expected
to
lead
to
incomplete
cross­
resistance
and
thus
the
likelihood
of
enhanced
survival
on
WideStrike
cotton
is
expected
to
be
small.
Please
comment
on
EPA's
conclusion
that
resistance
is
more
likely
to
be
associated
with
receptor
binding
modifications
rather
than
other
mechanisms
of
resistance
such
as
detoxification
in
the
midgut
lumen
by
proteases
that
cleave
the
insecticidal
control
protein(
s),
metabolic
adaptations,
protease
inhibition,
gut
recovery,
and
behavioral
adaptations.

Panel
Response
Apart
from
the
known
protoxin
solubility
and
proteolytic
processing
factors
governing
Bt
toxin
specificity,
the
insect
host
range
of
Bt
toxins
is
due
to
the
recognition
of
target
receptors
on
larval
gut
epithelial
cells.
As
resistance
management
plans
are
based
on
data
characterizing
these
molecules,
it
is
relevant
to
determine
which
toxin
classes
compete
for
larval
gut
receptor
binding
sites.
Intimate
knowledge
of
the
target
insect's
Bt
receptors
are
crucial
to
predict
and
avoid
the
possibility
of
cross­
resistance
occurring.
This
forms
the
subject
of
the
first
part
of
the
question
dealing
with
the
Agency's
conclusion
that
`
incomplete
shared
binding
of
Cry1Ac
and
Cry1F
receptors,
in
TBW
and
CBW,
is
expected
to
lead
to
incomplete
cross­
resistance
and
thus
the
likelihood
of
enhanced
survival
on
WideStrike
cotton
is
expected
to
be
small.'

While
the
Panel
supported
the
Agency's
conclusion
that
incomplete
shared
binding
of
Cry1Ac
and
Cry1F
receptors
in
TBW
and
CBW
is
expected
to
lead
to
incomplete
crossresistance
differences
were
expressed
on
the
molecular
mechanism
involved
in
the
process.
In
addition
the
Panel
raised
the
issue
that
another
as
yet
unidentified
major
resistance
mechanism
may
not
occur.
The
results
presented
by
the
registrant
on
incomplete
shared
binding
of
Cry1Ac
and
Cry1F
receptors
are
encouraging,
but
given
our
still
modest
understanding
of
the
relationship
between
membrane
binding
assays
and
actual
toxicity
to
the
target
insects,
it
would
be
premature
to
conclude
that
there
will
be
no
significant
cross­
resistance
between
the
two
toxins.
Crossresistance
does
not
seem
likely,
but
cannot
be
completely
dismissed
as
a
possibility.
The
Panel
cited,
as
the
basis
for
this
cautionary,
the
altered
cadherin
genes
described
for
the
YHD2
strain
of
H.
virescens
and
a
strain
of
pink
bollworm
Pectinophora
gossypiella
from
Arizona.
These
alleles
seem
to
be
characterized
by
high
fitness
costs
and
relatively
poor
survival
on
Bt
plants.
In
contrast,
resistant
strains
of
diamondback
moth,
for
which
efforts
to
find
either
altered
cadherins
or
altered
aminopeptidases
have
been
unsuccessful,
prosper
on
Bt
plants
and
seem
to
have
relatively
low
fitness
costs.
This
suggests
that
a
major
mechanism
of
resistance,
already
Page
48
of
85
widespread
in
field
populations
of
diamondback
moth,
remains
to
be
discovered.
That
being
said,
given
the
limited
or
incomplete
complementarity
between
shared
Cry
toxin
binding
sites
in
the
principle
lepidopterans
selected
so
far
(
TBW
and
CBW),
it
is
unlikely
in
the
absence
of
another
mechanism
(
as
described
above)
imparting
broad
cross
resistance
that
either
of
these
pests
will
develop
resistance
to
either
Bollgard
II
or
WideStrike
pyramided
cotton.

The
Panel
was
in
agreement
that
the
issue
of
insect
binding
proteins
is
a
complex
one
as
it
generally
involves
more
than
a
simple
`
one
toxin­
one
binding
protein'
scenario.
Current
data
indicated
that
any
one
toxin
may
potentially
have
one
to
six
insect
binding
molecules
which
in
turn
may
share
a
binding
site
with
a
different
toxin.

With
regards
to
CBW,
the
Agency
review
based
its
assessment
on
competition
binding
data
on
larval
brush
border
membrane
vesicles
(
BBMVs)
demonstrating
that
Cry1F
possessed
a
low
affinity
for
a
receptor
subset
shared
with
Cry1Ac.
The
data
also
showed
that
in
the
presence
of
high
saturating
levels
of
Cry1F,
approximately
40%
of
the
Cry1Ac
total
binding
capacity
was
still
available,
suggesting
that
40%
of
the
total
Cry1Ac­
specific
sites
are
unique
to
Cry1Ac
and
not
shared
with
Cry1F.
Since
CBW
is
not
sensitive
to
Cry1F,
the
Panel
was
uncertain
if
the
idea
of
Cry1F/
Cry1Ac
cross­
resistance
in
the
conventional
sense
was
applicable
in
this
case
but
concluded
that
the
presence
of
Cry1F
would
not
affect
Cry1Ac
toxicity.
If
the
shared
receptor
lost
its
ability
to
bind
to
Cry1F
(
or
Cry1Ac),
survival
on
WideStrike
cotton
would
not
be
expected
since
an
additional
mutation
would
also
have
to
occur
in
order
to
produce
a
homozygous
resistant
insect.
However,
for
an
insect
that
is
sensitive
to
both
Cry1F
and
Cry1Ac
toxins,
which
is
the
case
for
the
TBW,
the
molecular
details
are
less
certain.

A
general
comment
was
made
about
toxin­
receptor
binding
methodology
field
in
that
it
contained
a
number
of
inherent
weaknesses.
The
Panel
believed
it
was
important
to
outline
these
binding
weaknesses
as
IRM
decisions
are
based
on
models
generated
by
these
data.
The
Panel
then
noted
the
differences
between
receptor
binding
and
functional
receptor
binding.
To
illustrate
this
point,
the
Panel
mentioned
that
toxicity
to
insect
larvae
have
been
correlated
with
the
existence
of
high
affinity
binding
sites
on
BBMVs
of
susceptible
insects.
However
the
opposite
may
not
be
true.
There
are
examples
of
toxins
that
bind
to
BBMVs
with
high
affinity
but
are
nontoxic
to
the
insect.
The
most
relevant
were
data
showing
that
Cry1J,
a
protein
nontoxic
to
TBW,
possessed
a
five­
fold
higher
affinity
for
the
H.
virescens
receptor
A
than
Cry1F
which
is
toxic
to
CBW
(
Jurat­
Fuentes
and
Adang
2001).
A
similar
example
was
also
mentioned
for
the
non­
toxic
Cry1Ac
for
Spodoptera
frugiperda
(
Luo
et
al.
1999).
The
key
point
the
Panel
wished
to
make
was
that
some
predicted
high
affinity
binding
sites
may
not
directly
play
a
role
in
toxicity
i.
e.,
they
are
non­
functional.
In
reference
to
earlier
sessions
on
Ecological
Risk
Assessments
where
the
possibility
of
looking
for
receptors
in
non­
target
animals
was
discussed,
the
Panel
stressed
their
concern
about
the
potential
to
generate
spurious
binding
results
with
toxins
like
Cry1Ac
which
is
known
to
have
lectin­
like
activity
(
Cry1Ac
binds
to
N­
acetylgalactosamine
residues).
This
could
result
in
interactions
with
binding
potential
molecules
in
other
non­
target
species
but
yet
would
represent
non­
functional
binding
and
consequently
have
no
toxicological
impact.
So,
unless
the
receptors
are
individually
purified
and
tested,
competition
binding
patterns
and
what
constitutes
functional
binding,
are
subject
to
interpretation
of
indirect
data
or
deductive
reasoning.
Page
49
of
85
It
was
originally
shown
over
ten
years
ago
(
Van
Rie
et
al.
1989)
that
CBW
had
three
receptor
populations:
receptor
A
which
binds
all
three
Cry1A
proteins,
receptor
B
which
binds
only
two
(
Cry1Ab
and
Cry1Ac)
and
receptor
C
which
is
a
Cry1Ac
only
receptor.
In
the
case
of
CBW,
the
EPA's
assessment
is
based
on
newer
data
that
confirmed
the
earlier
work
and
expanded
it
to
include
Cry1F
and
Cry1J.
This
inclusion
helped
the
Panel
to
underline
a
second
weakness
in
binding
assays
which
is,
not
all
toxins
can
be
labeled.
In
this
specific
case,
radioiodination
of
Cry1F
resulted
in
loss
of
binding
capacity
and
subsequently,
toxicity.
Although
with
unlabelled
Cry1F
one
can
demonstrate
a
shared
receptor
site
with
Cry1Ac,
one
cannot,
however,
assess
whether
Cry1F
itself
has
its
own
unique
(
i.
e.,
unshared
with
Cry1Ac)
high
affinity
receptor
binding
site.
This
is
where
the
concept
of
incomplete
cross­
resistance
as
initially
put
forth
by
the
Agency
becomes
uncertain
as
it
is
based
on
the
existence
of
three
receptor
populations.
One
population
is
shared
between
Cry1F
and
Cry1Ac
and
two
unshared
populations,
one
for
Cry1Ac
and
one
for
Cry1F.
However,
based
on
current
methodology
limitations,
the
Panel
could
not
conclude
if
Cry1F
has
such
an
unshared
population.

A
final
weakness
in
current
binding
assays
discussed
by
the
Panel
involved
the
use
of
ligand
blot
assays.
Looking
at
the
model
presented
for
CBW
showing
four
receptor
groups
(
Jurat­
Fuents
and
Adang
2001),
there
was
a
separate
binding
protein
for
Cry1F
independent
of
Cry1Ac.
The
existence
of
this
binding
protein
was
deduced
from
ligand
blot
experiments.
Ligand
binding
assays
are
assays
where
BBMV
binding
proteins
are
denatured
by
boiling
in
anionic
detergent
(
SDS)
and
separated
based
on
their
molecular
weights
in
SDS­
PAGE
gels.
These
gels
are
then
blotted
onto
a
membrane
and
probed
with
a
labeled
toxin.
Unfortunately,
this
is
a
difficult,
highly
variable
technique,
subject
to
labeling
artifacts
and
differential
receptor
processing.
There
are
numerous
instances
in
the
literature
where
these
ligand
blots
vary
substantially
from
lab
to
lab.
However,
in
this
case,
the
Panel
was
reassured
to
see
that
the
data
also
included
biotinylated
toxins
and
anti­
Cry
toxin
antibodies
to
ensure
the
data
was
as
robust
as
it
could
be.
If
one
considered
only
the
competition
binding
data,
only
three
receptor
populations
would
remain
of
which
one
would
contain
a
shared
Cry1F
/
Cry1Ac
receptor.
The
Panel
noted
that
ligand
blots
indicated
the
presence
of
two
Cry1F
binding
proteins
from
one
study
(
Fuents
and
Adang
2001)
but
another
study
showed
the
existence
of
four
Cry1F
binding
proteins
(
Banks
et
al.
2001)
and
that
they
are
of
similar
size
as
the
Cry1Ac
binding
molecules.
The
Panel
recommended
the
Agency
exercise
caution
as
ligand
blot
data
can
occasionally
give
spurious
results
and
to
take
that
into
consideration
in
any
IRM
decision.

The
Panel
raised
the
issue
concerning
the
lab
derived
Cry1Ac
resistant
strain
of
H.
irescens
(
YHD2)
which
is
cross­
resistant
to
Cry1Aa
and
Cry1Ab
but
can
bind
Cry1Ab
and
Cry1Ac
in
a
fashion
similar
to
wildtype
(
Lee
at
al.
1995).
The
conclusion
from
that
study
showed
that
the
mutation
occurred
in
Receptor
A.
As
the
current
model
for
TBW
shows
that
Receptor
A
binds
all
three
Cry1A
toxins
as
well
as
Cry1F,
the
Panel
believed
that
it
was
important
to
note
that
this
strain
was
also
cross­
resistant
to
Cry1F,
thus
illustrating
two
points:
1)
the
presence
of
high
affinity
binding
sites
in
H.
virescens
BBMVs
that
do
not
appear
to
be
involved
in
toxicity;
2)
the
possibility
that
resistance
to
Cry1Ac
could
potentially
result
in
cross­
resistance
to
Cry1F
as
Page
50
of
85
shown
in
a
lab­
derived
resistant
strain
(
Gould
et
al.
1995).

In
presenting
the
question
of
incomplete
cross­
resistance,
the
Agency
indicated
that
the
mutation
of
a
shared
receptor,
when
resulting
in
the
loss
of
one
Cry
toxin,
causes
the
cessation
of
binding
of
the
second
toxin.
The
Panel
noted
exceptions
to
this
assumption
by
stating
various
examples
in
the
literature
(
e.
g.,
P.
xylostella
and
H.
virescens)
where
the
loss
of
binding
to
one
toxin
in
a
shared
receptor
did
not
result
in
the
loss
of
binding
of
the
second
toxin
(
Ferre
and
Van
Rie
2002).
This
simply
means
that
the
toxins
bound
to
different
epitopes
on
the
binding
molecules.
The
Panel
further
noted
that
in
the
case
of
the
120
kDa
aminopeptidase
(
APN)
purified
from
Manduca
sexta,
these
epitopes
are
situated
far
enough
apart
so
that
two
Cry1Ac
toxin
molecules
could
actually
bind
to
the
APN
at
the
same
time
(
Masson
et
al.
1995).

In
regards
to
the
original
hypothesis
that
incomplete
shared
binding
may
lead
to
incomplete
cross
resistance,
the
Panel
was
not
convinced
by
the
newer
four
receptor
site
model
for
TBW
since
it
was
unclear
that
Cry1F
had
its
own
unique
binding
site
(
unshared
with
Cry1Ac).
The
Panel
did
recognize
that
the
literature
indicated
a
plethora
of
shared
Cry1F/
Cry1Ac
binding
sites
in
TBW
which
was
presumably
a
result
of
similarities
in
both
toxin's
domain
II
loops
(
Jurat­
Fuentes
and
Adang
2001).
These
shared
sites
presumably
represent
a
mixture
of
different
functional
genes
(
i.
e.
aminopeptidases,
cadherins
or
possibly
large
glycoconjugates).
Therefore,
under
the
assumption
of
functional
binding,
the
same
rules
for
incomplete
cross
resistance
would
still
apply,
that
a
mutation
occurring
in
one
gene
causing
resistance
through
loss
of
binding
to
one
of
the
stacked
toxins
will
result
in
incomplete
cross­
resistance
as
the
second
toxin
can
bind
to
another
site.

The
Panel
believed
that
in
addition
to
incomplete
receptor
sharing
hypothesis,
it
is
important
to
consider
the
cost
in
fitness
that
would
occur
with
receptor
mutation/
decreased
toxin
binding.
Resistance
via
the
altered
toxin
binding
mechanism
is
accompanied
by
loss
of
binding
function,
and
therefore
presumably
imparts
a
"
cost"
to
resistant
individuals.
This
view
is
strengthened
when
both
principle
binding
sites
in
Lepidoptera
involve
cadherin
and
aminopeptidase
receptors
because
each
of
these
receptors
are
important
cellular
components
required
by
the
insect
having
significant
physiological
cell
adhesion
or
membrane
binding
and
transport
functions.
Apart
from
having
a
second
different
active
toxin
to
bypass
the
inactivated
receptor,
the
added
fitness
cost
would
also
contribute
to
the
prevention
of
cross­
resistance
under
this
scenario.

The
Panel
maintained
that
broad
receptor­
based
cross
resistance
is
an
improbable
outcome,
especially
with
refugia,
and
the
emerging
information
documenting
cotton
recruitment
of
unselected
CBW
from
surrounding
alternate
hosts
and
regions.
Therefore,
even
dramatic
selection
with
Cry1Ac
alone
in
the
absence
of
refugia
is
not
likely
to
result
in
a
reduced
field
efficacy
of
either
Bollgard
II
or
WideStrike
cotton
against
TBW
and
CBW.
The
hypothesis
of
incomplete
cross­
resistance
as
stated
should
only
be
considered
on
a
case
by
case
basis
as
what
applies
to
one
species
may
be
inappropriate
for
a
second.

The
Panel
also
recommended
that
the
Agency
be
aware
of,
and
take
into
consideration,
Page
51
of
85
the
problems
generated
with
current
receptor
binding
methodologies
and
urged
them
to
investigate
ways
to
circumvent
these
problems
such
as
alternate
labeling
strategies
for
recalcitrant
toxins
(
S35,
C14
methylation,
etc.)
or
non­
destructive
real­
time
optical
measurements
of
protein
binding.

Part
2.
Resistance
is
more
likely
to
be
associated
with
receptor
binding
modifications
rather
than
other
mechanisms
of
resistance.

Alternate
resistant
mechanisms
do
have
the
potential
to
occur
but
to
date
these
have
only
been
demonstrated
in
the
laboratory.
Throughout
the
history
of
resistance
management,
lab
selection
experiments
have
given
results
that
were
inconsistent
with
those
observed
in
the
field,
both
because
laboratory
populations
are
less
likely
to
include
rare
single
major
genes
(
due
to
an
inevitably
limited
sample
size
in
the
initial
field
collections)
and
because
laboratory
selection
is
more
likely
to
"
save"
resistance
genes
that
have
very
large
fitness
costs
in
the
field.
Selection
in
the
field,
in
contrast,
will
screen
far
more
rare
alleles,
and
alleles
with
large
fitness
costs
are
unlikely
to
increase
very
much.
The
Panel
agreed
with
the
Agency
that
there
is
no
basis
to
believe
that
the
occurrence
of
resistance
in
the
field
will
be
due
to
a
mechanism
other
than
binding
site
modification.

To
put
the
question
into
perspective,
a
summary
of
the
known
mode
of
toxin
action
occurring
in
the
insect
after
toxin
ingestion
is
needed.
After
ingestion
in
a
crystallized
protoxin
form,
solubilization
must
first
occur
in
the
gut
of
a
susceptible
insect
where
gut
proteases
then
activate
the
protoxin
producing
a
protease
resistant
core.
This
activated
toxin
then
attaches
to
a
specific
docking
protein
or
receptor
on
the
surface
of
gut
epithelial
cells.
Functional
binding
to
the
receptor
is
followed
by
oligomerization
into
a
tetrameric
structure
and
subsequent
insertion
into
the
membrane
causing
a
disruption
in
the
cellular
ion
balance,
and
eventual
cell
death
through
a
colloid
osmotic
lysis
process.
In
theory,
toxin
resistance
can
occur
at
any
of
these
steps.

It
was
stipulated
that
the
most
important
type
of
resistance
would
be
that
found
under
actual
field
conditions.
To
date,
only
one
major
Lepidopteran
resistance
mechanism
has
been
reported
which
was
a
reduction
in
toxin
binding
on
gut
brush
border
target
sites
in
the
pest
P.
xylostella,
the
diamondback
moth
(
Ferre
and
Van
Rie
2002).
Despite
the
possibility
of
resistance
through
other
mechanisms
described
below,
the
history
of
resistance
evolution
to
Bt
sprays
in
P.
xylostella,
strongly
implies
that
reduced
binding
to
receptors
is
likely
to
be
the
most
common
and
significant
mechanism
of
resistance,
even
if
the
specific
details
of
the
reduced
binding
mechanism
varies
between
species.
In
principle,
all
of
the
other
possible
mechanisms
of
resistance
are
also
available
in
P.
xylostella,
but
resistance
has
repeatedly
evolved
in
the
field
through
receptor
binding
as
at
least
the
main
if
not
sole
mechanism
of
resistance.

Switching
from
field­
derived
resistance
cases
to
those
produced
in
the
laboratory
setting,
the
Panel
pointed
out
that
for
the
majority
of
different
insect
species
showing
an
altered
binding
resistance
phenotype,
there
also
exist
resistance
strains
from
the
same
species
showing
three
major
resistance
mechanisms
other
than
reduced
binding.
For
example
it
has
been
shown
that
Cry
proteins
can
be
detoxified
in
the
midgut
through
a
decreased
rate
of
protoxin
activation
and
an
Page
52
of
85
increased
rate
of
toxin
degradation
in
the
resistant
CP­
73
line
of
TBW
(
Forcada
et
al.
1996).
In
contrast
to
this,
it
has
been
shown
in
Choristoneura
fumiferana,
even
though
this
insect
is
highly
susceptible
to
Cry1A
toxins,
increasing
concentrations
of
gut
juice
increases
the
proteolytic
degradation
of
purified
activated
toxin
(
Bah
et
al.
2004).
Decreased
susceptibility
to
Cry1Ac
in
Plodia
interpunctella
was
shown
to
be
due
to
the
absence
of
a
major
gut
protease
(
Oppert
et
al.
1997).

A
second
mechanism,
enhanced
epithelial
recovery,
was
evolved
after
selection
with
sublethal
doses
of
Cry1Ac
resulted
in
moderate
resistance
in
TBW
while
similar
selection
using
Cry3A
toxin
in
Colorado
potato
beetle
(
CPB)
resulted
in
high
(>
500fold)
resistance
via
this
putative
mechanism.
Selection
with
one
toxin,
Cry1Ac,
has
also
been
demonstrated
to
result
in
more
than
one
resistance
mode
of
action,
strong
cross
resistance
to
Cry1Fa
and
low
resistance
to
Cry2A
toxin
in
TBW.
In
that
Cry1Ac
resistant
strain
CP73­
3,
midgut
epithelial
cell
damage
had
occurred
in
resistant
TBW
leading
to
the
conclusion
that
enhanced
gut
recovery
had
occurred.
In
terms
of
cost,
resistance
via
altered
proteolytic
processing
may
or
may
not
impart
a
cost
to
the
resistant
individuals
while
putative
enhanced
epithelial
recovery,
at
least
in
CPB,
was
accompanied
by
reproductive
and
developmental
costs.
Therefore,
if
these
resistance
mechanisms
surfaced
and
they
were
recessive,
as
in
lab
strains
and
in
the
absence
of
Bt
cotton
selection
in
the
field,
resistance
reversion
may
occur
in
field
populations.

A
third
mechanism,
behavioral
resistance,
was
also
considered
by
the
Panel
to
be
theoretically
possible,
but
most
scenarios
would
depend
on
detection
of
the
Cry
toxin
in
the
transgenic
plant.
Although
anecdotal
accounts
of
these
phenomena
have
been
reported
from
some
conventional
Bt
spray
formulations,
to
date,
no
credible
data
supporting
this
phenomenon
in
Lepidoptera
has
been
reported.
In
principle,
insects
could
evolve
to
selectively
feed
on
those
parts
of
transgenic
plants
that
have
lower
expression.
However,
feeding
on
more
hidden
or
lower
parts
of
plants
has
always
been
a
potential
method
of
avoiding
pesticide
sprays
and
therefore
of
broad
cross­
resistance
across
the
last
50
years
but
has
yet
to
be
described
for
any
pest.
Although
there
are
examples
of
behavioral
resistance
in
insects
like
house
flies,
they
have
generally
been
rare.
A
probable
explanation
is
that
the
impacts
of
environmental
influences
are
often
high
on
the
expression
of
a
wide
range
of
behavioral
traits;
this
in
turn
lowers
the
heritability
of
such
traits
and
the
rate
at
which
they
can
be
selected.
In
contrast,
physiological
traits
(
such
as
reduced
binding)
are
less
influenced
by
the
environment
and
are
thus
more
likely
to
be
selected
rapidly.
The
general
consensus
of
the
Panel
was
that
behavioral
resistance
seems
unlikely
to
become
a
major
mechanism
of
resistance
on
transgenic
plants.

The
ability
of
aminopeptidases,
cadherins
as
well
as
some
relatively
uncharacterized
glycoconjugates
capacity
to
bind
Bt
toxins
with
high
affinity
has
been
known
for
many
years,
yet
determination
of
the
functional
nature
of
the
genes
genetically
linked
to
resistance
has
been
difficult.
To
date,
no
linkage
of
resistance
to
aminopeptidases
or
glycoconjugates
has
been
published.
Recently
one
publication
demonstrated
a
tight
linkage
between
a
cadherin
gene
known
to
bind
Cry
genes
and
resistance
in
the
YHD2
strain
of
TBW
(
Gahan
et
al.
2001).
Although
this
resistant
allele
has
not
been
found
in
TBW
in
the
field,
another
publication
recently
came
out
showing
that
field
derived
strains
of
the
P.
gossypiella
did
possess
mutated
alleles
of
the
cadherin
Page
53
of
85
gene
associated
with
resistance
to
Cry1Ac
(
Morin
et
al.
2003).
All
resistant
P.
gossypiella
screened
for
three
different
mutations
in
this
gene
were
homozygous
for
the
resistance
genes
and
all
the
susceptible
larvae
were
heterozygous.
This
makes
cadherin
the
leading
target
for
DNA
based
resistance
screening.
Although
the
evidence
for
altered
Cry
receptors
(
cadherin)
being
the
major
mechanism
of
Cry
toxin
resistance
in
the
field
is
compelling,
the
Panel
cautioned
that
due
to
fitness
costs
observed
with
this
mutation,
it
is
highly
likely
that
another
major
resistance
mechanism
existing
remains
high.

Agency
Charge
3.
CBW
modeling.
Dow's
CBW
modeling
efforts
show
that
EPA
can
have
high
confidence
that
there
will
not
be
a
significant
change
in
population
fitness
of
CBW
on
WideStrike
cotton
in
a
15­
year
time
horizon
even
without
a
high
dose
for
either
Cry1Ac
or
Cry1F
and
incomplete
cross­
resistance
(
20
to
60%
maximum
shared
binding).
Market
share
analysis
of
WideStrike
cotton
versus
other
Bt
cotton
products
had
little
effect
on
the
rate
at
which
CBW
may
adapt
in
either
the
North
Carolina
or
Mississippi
Delta
agroecosystem.
Refuge
size,
whether
sprayed
or
unsprayed,
had
no
significant
impact
on
CBW
population
fitness
on
WideStrike
cotton
after
15
years.
In
the
Delta,
the
immigrating
non­
selected
population
from
alternate
hosts
further
reduces
the
local
rate
of
adaptation.
The
local
structured
refuge
only
supplies
a
small
proportion
of
the
non­
selected
insects
in
the
Delta.
The
availability
of
CBW
alternate
hosts,
coupled
with
a
non­
Bt
cotton
refuge
are
additional
levels
of
assurance
for
WideStrike
cotton
product
durability.
Additional
empirical
information
is
needed
on
the
function
and
effectiveness
of
alternate
hosts
on
the
rate
of
CBW
adaptation.

The
Agency
asks
the
SAP
to
comment
on
the
predictions
made
by
the
DAS
CBW
model,
i.
e.,
the
likelihood
that
the
population
fitness
of
CBW
on
WideStrike
cotton
in
a
15­
year
time
horizon
will
remain
unchanged,
even
without
a
high
dose
for
either
Cry1Ac
or
Cry1F
and
incomplete
cross­
resistance
(
60%
of
Cry1Ac
binds
to
the
Cry1F
receptor).

Panel
Response
The
model
used
by
the
registrant
is
a
spatially
explicit
simulation
model
based
on
Storer
(
2003).
The
model
was
extended
to
explore
a
system
of
three
transgenic
products,
two
stacks
(
partially
sharing
one
receptor)
and
a
single
gene
product.
Together,
the
three
products
shared
a
total
of
three
protein
receptors.
It
explores
scenarios
using
crop
mixtures
from
two
agroecosystems,
North
Carolina
and
the
Mississippi
Delta.

The
Panel
identified
several
areas
of
concern
with
the
registrant's
CBW
model
that
make
its
use
problematic.
These
problems
must
be
addressed
if
this
model
is
to
be
used
to
assess
the
durability
of
WideStrike
cotton
in
these
areas.
The
Panel
believed
that
use
of
the
current
model,
once
corrected
of
the
identified
errors,
would
be
an
appropriate
vehicle
to
explore
the
parameter
space
with
the
goal
of
finding
areas
where
resistance
does
occur
in
the
15­
year
time
horizon
and
assessing
whether
it
occurs
within
biologically
plausible
initial
conditions
and
parameter
values.
Page
54
of
85
The
results
from
the
model
were
presented
as
population
fitness
(
or
changes
thereof).
Population
fitness,
measured
as
the
frequency
weighted
average
of
the
genotype
fitness,
is
a
nonlinear
function
of
both
time
and
log
gene
frequency.
In
short,
population
fitness
changes
little
until
the
resistance
allele
is
close
to
one,
thus
masking
many
of
the
important
gene
frequency
changes
that
occur
early
in
the
evolution
of
resistance.
The
Panel
agreed
that
this
property
of
population
fitness
makes
it
difficult
to
understand
the
results
presented.
The
Panel
recommended
that
resistance
allele
frequencies
to
the
individual
receptors
be
presented
instead,
as
this
would
be
much
more
informative
and
enable
Panel
members
to
better
understand
differences
between
runs
of
the
model
with
different
parameters.

The
Panel
identified
a
potentially
serious
problem
in
the
manner
in
which
the
fitnesses
of
the
receptor
genotypes
were
estimated.
The
fitness
of
the
nine
receptor
genotypes
for
binding
by
Cry1Ac
to
receptors
A
and
B
are
based
on
partitioning
W2,
the
fitness
of
the
doubly
homozygous
susceptible
genotype
on
Cry1Ac
cotton
into
W2A
=
1­
x

Z,
fitness
in
the
face
of
x

100%
binding
to
receptor
A,
and
W2B
=
[
1­(
1­
x)

Z],
fitness
in
the
face
of
(
1­
x)

100%
binding
to
receptor
B,
as
follows:
W2
=
W2A

W2B.
This
can
be
expressed
in
terms
of
survival
rates
[
keeping
in
mind
that
these
rates
are
relative
to
survival
of
the
corresponding
homozygous
resistant
genotype(
s)]
as
follows:
S2
=
S2A

S2B
for
S2A
=
(
1­
x

Z)
and
S2B
=
[
1­(
1­
x)

Z].
Thus,
Z
can
be
interpreted
as
the
probability
of
mortality
if
all
of
the
binding
is
to
receptor
A
or
if
all
of
the
binding
is
to
receptor
B,
but
this
leads
to
a
contradiction.
If
the
effect
of
a
toxin
molecule
binding
to
receptor
A
is
identical
to
that
for
binding
to
receptor
B,
then
the
resulting
survival
will
be
independent
of
x:
S2
=
(
1­
Z);
however,
the
formulation
of
the
DAS
model
results
in
a
dependence
on
x:
S2
=
(
1­
Z)+
x

(
1­
x)

Z2.
Moreover,
Z
can
exceed
a
value
of
1.0
 
clearly
an
invalid
result
for
a
probability:
e.
g.
for
S2
=
0.2
and
x
=
0.6
(
the
best
empirical
estimates
for
these
variables),
Z
=
1.08.
Expressing
the
problem
in
another
way,
survival
in
the
face
of
x

100%
binding
to
receptor
A
and
survival
in
the
face
of
(
1­
x)

100%
binding
to
receptor
B
are
not
independent
events
so
that
the
joint
survival
rate
can
not
be
set
equal
to
the
product
of
individual
survival
rates.
An
alternative
formulation
is
as
follows:
S2
=
{
1­[
x

ZA+(
1­
x)

ZB]}
for
ZA,
the
probability
of
mortality
if
all
binding
is
to
receptor
A,
and
ZB,
the
corresponding
probability
for
receptor
B.
For
this
model,
mortality
due
to
Cry1Ac
ranges
from
ZA
to
ZB
as
a
linear
function
of
x;
and,
for
ZA
=
ZB
=
Z,
S2
=
(
1­
Z)
as
expected.
The
fitnesses
of
the
nine
receptor
genotypes
(
p.
30)
can
all
be
expressed
in
terms
of
ZA
and
ZB,
but
it
is
not
possible
to
estimate
ZA
and
ZB
given
the
available
data.
With
one
empirically­
derived
datum
 
W2
 
one
cannot
estimate
two,
empirically
independent
quantities.
In
other
words,
one
cannot
solve
for
more
unknowns
than
the
number
of
equations
relating
those
unknowns.
The
survival
of
one
of
the
other
genotypes
on
Cry1Ac
cotton
is
also
needed.

Following
this
FIFRA
SAP
meeting,
Panel
member
John
Schneider
provided
additional
comments
on
the
biological
interpretation
of
Z.
Such
comments
were
not
considered
or
reviewed
by
the
Panel
and
are
being
provided
as
an
appendix
to
these
meeting
minutes
(
Appendix
A).

The
sensitivity
analysis
of
the
model
showed
that
the
model
was
sensitive
to
several
biological
parameters
about
which
we
have
limited
information
undermining
the
confidence
of
using
the
model
to
determine
the
durability
of
WideStrike.
For
example,
the
model
was
sensitive
Page
55
of
85
to
immigration,
initial
gene
frequency,
fitness
of
the
R­
alleles,
flowering
dates
and
the
use
of
alternate
hosts.
We
know
little
about
these
parameters
in
the
field
(
except
flowering
dates)
and
so
the
model
outcomes,
as
run,
are
hard
to
interpret.
In
particular
the
model's
claim
that
resistance
will
not
occur
in
the
15­
year
period
cannot
be
verified
without
understanding
the
biological
parameters
in
the
model.

The
parameters
used,
and
initial
starting
conditions,
were
biologically
appropriate.
They
were,
however,
uninformative
as
to
under
what
conditions
the
model
may
predict
the
appearance
of
resistance.
Simpler,
deterministic
models
often
predict
that
evolution
of
resistance
will
occur
in
a
shorter
time
frame
than
suggested
by
the
model
presented.
In
such
cases,
it
should
be
incumbent
upon
the
users
of
spatially
explicit
models
to
provide
an
explanation
for
the
differences.
The
discrepancy
could
be
caused
by
subtle
differences
in
model
formulation
that
may
or
may
not
be
realistic
(
e.
g.,
many
of
us
essentially
left
refuges
in
one
place
until
Peck
et
al.
(
1999)
identified
this
as
a
problem),
or
it
could
be
an
important
phenomenon
that
needs
further
exploration
(
as
in
the
Peck
model).
An
alternative
approach
to
the
problem
would
be
to
explore
under
what
conditions
of
initial
conditions
and
parameter
values
resistance
does
appear.
Looking
at
worse­
case
scenarios
for
resistant
development
allows
the
user
of
the
model
to
assess
how
likely
the
parameter
combinations,
in
which
resistance
occurs,
will
be
biologically
plausible.
Just
because
there
are
model
outcomes
using
biological
realistic
parameters
where
resistance
is
delayed
for
15
years,
it
does
not
follow
that
there
might
not
yet
be
cases
where
the
model
suggests
that
resistance
will
occur
within
the
15­
year
window
that
also
uses
biologically
realistic
parameters.

The
Panel
also
suggested
that,
as
the
product
was
not
a
high
dose
product
for
cotton
bollworm,
the
possibility
of
a
single,
additive,
resistance
mechanism
that
provided
low
levels
of
resistance
(
5­
50
fold)
to
all
receptors
should
be
explored.
While
such
a
mechanism
is
unlikely
to
provide
complete
resistance,
it
could
compromise
the
product
if
it
increased
population
fitness
(
survivorship
in
the
field)
to
unacceptable
levels.

The
Panel
also
noted
that
while
it
is
clear
that
immigration
of
bollworm
adults
into
the
Delta
agroecosystem
does
occur,
it
is
a
subject
of
debate
how
important
those
moths
are
relative
to
populations
that
overwinter
locally.
The
model
also
appeared
to
be
constructed
as
a
continentisland
model,
with
no
impact
of
reverse
migration
of
selected
individuals
into
the
southern
overwintering
population.
This
may
or
may
not
be
a
reasonable
assumption,
but
no
data
was
given
to
support
it.
We
know
that
in
a
high­
dose
refuge
system
that
resistance
evolves
by
the
slow
contamination
of
the
refuges,
it
is
unclear
how
much
impact
reverse
migration
and
contamination
of
the
overwintering
refuges
might
have
on
the
rate
of
the
evolution
of
resistance.

Agency
Charge
4.
TBW
modeling.
For
TBW,
durability
is
expected
to
be
greater
than
that
predicted
using
the
TBW
model
by
Peck
et
al.
(
1999)
where
the
worst
case
(
structured
refuge
is
moved
each
year)
is
17
years.
TBW
exhibits
similar
patterns
in
binding
studies
as
does
CBW
and
WideStrike
cotton
expresses
a
high
dose
against
TBW.
The
Cry1Ac
component
alone
is
a
Page
56
of
85
high
dose
and
the
Cry1F
component
alone
is
not
quite
a
high
dose.

The
Agency
asks
the
SAP
to
comment
on
the
relative
WideStrike
cotton
durability
against
TBW
using
the
Peck
et
al.
(
1999)
model.

Panel
Response
Since
the
dose
of
the
Cry1Ac
and
Cry1F
in
WideStrike
cotton
was
demonstrated
to
be
high
against
populations
of
TBW,
the
Panel
believed
that
WideStrike
will
be
more
durable
than
that
predicted
by
Peck
(
1999)
for
single
Cry1Ac
cotton.
In
most
models
with
which
the
Panel
is
aware,
two
stacked
products
outperform
single
toxin
products
in
delaying
resistance.
This
will
be
true
for
WideStrike
cotton
as
well.

Much
weight
is
given
to
the
1993
result
of
a
frequency
of
1.5x10­
3
Gould
et
al.
(
1997).
Since
it
has
been
11
years
since
that
original
work
was
done,
assumptions
about
initial
frequency
should
be
conservative.
It
does
not
seem
unreasonable
that
under
continued
selection
over
the
last
several
years
for
Bt
resistance,
the
potential
for
higher
frequencies
should
be
considered.
One
feature
of
the
Peck
(
1999)
model
that
seems
relevant
is
that
resistance
first
appeared
in
small
foci.
Unless
one
was
fortunate
to
be
monitoring
that
particular
focus
of
resistance
development
when
control
failures
had
begun,
it
may
be
too
late
to
stop
resistance
development.

While
looking
at
the
parameter
space
of
the
model
is
challenging,
it
is
important
to
understand
under
what
conditions
we
can
expect
resistance.
An
important
aspect
of
modeling
is
sensitivity
modeling,
but
it's
also
important
to
look
at
worst
case
scenarios,
and
to
explore
under
what
conditions
the
models
predict
shortened
times
to
resistance.

Agency
Charge
5.
Alternate
hosts.
Dow
utilizes
its
CBW
model
that
simulates
two
agroecosystems
that
consist
of
CBW
crops
hosts
soybean,
maize,
and
cotton
in
varying
amounts,
three
insecticidal
control
proteins
(
Cry1Ac,
Cry1F,
and
Cry2Ab),
and
three
protein
receptors.
Dow
also
uses
the
HOSTS
data
base,
and
carbon
isotope
work
by
Gould
et
al.
(
2002)
to
support
the
use
of
CBW
alternate
hosts
as
an
effective
means
of
reducing
the
populationwide
selection
pressure
to
the
two
ICPs
expressed
in
WideStrike
cotton
(
metapopulation
dynamics
effects).
To
support
the
effectiveness
of
alternate
hosts
as
natural
refugia,
data
are
needed
on
the
larval
and
adult
production
of
CBW
on
each
alternate
host
for
each
generation
relative
to
cotton
and
WideStrike
cotton
and
the
spatial
scale
and
source
of
moth
production.

The
Agency
asks
the
SAP
to
comment
on:
a)
the
sufficiency
of
the
WideStrike
cotton
database
to
address
the
issue
of
CBW
alternate
hosts
as
natural
refugia,
and,

b)
whether
additional
data
are
needed
on
the
larval
and
adult
production
of
CBW
on
each
Page
57
of
85
alternate
host
for
each
generation
relative
to
cotton
and
WideStrike
cotton
and
the
spatial
scale
and
source
of
moth
production
to
confirm
the
effectiveness
of
CBW
alternate
hosts
as
natural
refugia.

Panel
Response
5a)
The
Panel
agreed
that
the
HOSTS
database
is
insufficient
to
address
the
issue
of
CBW
alternate
hosts
as
natural
refugia.
The
registrant
references
an
online
data
base
for
Lepidoptera
where
researchers
can
search
for
host
plants
by
family,
genera,
and/
or
species.
The
database
is
a
good
research
tool,
but
in
the
context
of
the
current
discussions,
it
should
be
considered
as
a
starting
point
for
empirical
research
to
better
understand
how
CBW
utilizes
its
hosts.
The
database
may
contain
errors
and
does
not
provide
empirical
data
about
the
numbers
of
larvae
feeding
on
each
host
or
the
timing
of
host
utilization.

5b)
There
is
potential
for
the
utilization
of
alternative
hosts,
in
combination
with
the
long­
range
dispersal
ability
of
CBW,
to
diminish
the
community­
wide
responses
to
selection
for
adaptation
to
WideStrike
cotton.
However,
the
Panel
agreed
that
there
are
insufficient
empirical
data
in
the
registrant
report
to
demonstrate
that
alternative
hosts
are
producing
susceptible,
fit
individuals
in
sufficient
quantity,
at
the
correct
time
and
proximity
to
maximize
the
probability
of
matings
between
homozygous­
susceptible
individuals
and
individuals
heterozygous
for
resistant
traits.

In
order
to
develop
a
rational
IRM
plan
for
CBW
on
WideStrike
cotton,
the
effect
of
alternate
host
plants
on
the
proportion
of
each
CBW
population
that
is
subjected
to
selection
pressure
and
the
effect
of
migration
by
CBW
among
these
populations
on
resistance
frequencies
must
be
considered.

The
registrant
neither
presents
nor
cites
any
empirical
data
concerning
the
relative
numbers
of
CBW
larvae
on
various
host
plants
during
the
generations
that
selection
is
occurring
(
except
Gould
et
al.
2002,
as
discussed
below).
To
the
Panel's
understanding,
no
such
data
exist
for
wild
host
plants
due
to
the
difficulty
of
spatially
quantifying
the
abundances
of
these
hosts
and
due
to
the
absence
of
any
unique,
host
plant­
specific
tags
in
the
adults.
Due
to
this
lack
of
data
and
given
the
fact
that
these
host
plants
do
not
exert
selection
pressure
for
resistance,
the
conservative
approach
would
be
to
ignore
them.
Unfortunately,
the
model
presented
by
the
registrant
includes
wild
host
plants,
and
the
importance
of
these
hosts
relative
to
other
host
plants
is
unspecified.
For
cultivated
host
plants,
the
occurrence
of
CBW
on
Bt
and
non­
Bt
hosts
is
modeled
as
resulting
from
host
plant
preference
coefficients
and
dispersal
characteristics
of
CBW
and
from
the
overall
abundances
and
spatial
distributions
of
the
hosts.
However,
as
has
been
presented
earlier,
some
of
these
key
assumptions
in
the
model
are
unspecified;
and
no
model
output
showing
the
densities
of
CBW
on
the
various
cultivated
host
plants
is
presented.
The
registrant
refers
to
data
on
the
relative
importance
of
C3
and
C4
host
plants
for
CBW
(
published
in
Gould
et
al.
2002)
to
support
their
general
position
that
alternate
hosts
are
important.
Gould
et
al
(
2002)
report
that
>
40
%
of
the
CBW
collected
from
pheromone
traps
in
Texas
and
Louisiana
in
August
developed
on
C4
larval
host
plants
rather
than
on
cotton
or
other
C3
larval
host
plants.
Because
the
crop
phenologies
are
such
that
these
individuals
did
not
develop
on
local
corn,
they
Page
58
of
85
may
have
developed
on
other,
as
yet
unidentified,
local,
non­
Bt
host
plants.
This
interpretation
of
these
data
does
provide
qualitative
support
for
the
conclusion
that
resistance
to
WideStrike
cotton
is
likely
to
develop
slowly.

Gould
et
al.
(
2002)
made
a
different
interpretation
of
their
late
season
data
(>
1
August).
They
suggested
that
most
of
the
observed
late­
season
moths
from
C4
plants
developed
on
corn
at
higher
latitudes
and
migrated
into
Texas
and
Louisiana.
Given
the
presence
of
Bt
corn
in
the
Midwest,
an
understanding
of
the
development
of
resistance
to
Bt
crops
in
the
Midsouth
would
require
an
understanding
of
the
effect
of
the
contribution
of
these
immigrants
to
local
resistance
frequencies.
The
registrant
CBW
model
did
not
include
late
season
immigration,
but
it
did
incorporate
early
season
immigration
(
presumably
from
lower
latitudes)
into
the
Midsouth
location
of
the
Mississippi
Delta
agricultural
area.
Gould
et
al.
(
2002)
reported
that
>
40%
of
the
CBW
moths
from
14/
15
samples
collected
in
May
or
earlier
in
Texas
and
Louisiana
developed
on
C4
larval
hosts.
Assuming
that
there
were
no
important,
unidentified,
late­
season,
Midsouth
C4
hosts,
these
data
also
suggest
that
early
season
immigration
occurs.
However,
the
percentage
contribution
of
the
immigrants
to
the
local,
reproducing
population,
how
this
percentage
varied
among
years,
and
the
source
of
the
immigrants,
which
determined
their
history
of
exposure
to
Bt
crops,
would
need
to
be
known
in
order
to
understand
the
potential
for
resistance
in
CBW
in
the
Midsouth.

Further
evidence
of
the
lacking
of
empirical
data/
studies
in
the
registrant
report
concerns
the
long
distance
dispersal
of
CBW
adults
in
a
short
period
of
time.
This
has
been
shown
by
previous
migration
studies
from
southern
Texas
to
Central
Texas
and
into
Oklahoma.
Thus,
it
is
obvious
that
we
must
examine
moth
production
on
a
regional
and
possibly
even
state­
by­
state
basis.
Although
migrants
do
not
meet
the
registrant's
definition
of
adults
produced
in
refugia,
the
Panel
does
not
believe
that
discounting
the
influence
of
migrants
into
a
production
system
is
the
issue.
Instead,
the
problem
will
be
quantifying
the
migrants
and
their
effect
on
the
system.

Agency
Charge
6.
IRM
Plan.
The
WideStrike
cotton
IRM
plan
has
the
following
proposed
refuge
requirements:

a.
5%
external
unsprayed
refuge
option.
Five
percent
of
the
cotton
fields
must
be
planted
to
non­
Bt
cotton
and
not
be
treated
with
any
lepidopteran­
control
technology.
The
refuge
must
be
at
least
150
ft.
wide
(
preferably
300
ft.)
and
within
½
mile
(
preferably
adjacent
or
within
1/
4
mile
or
closer)
of
the
Bt
cotton.

b.
20%
external
sprayable
refuge
option.
Twenty
percent
of
the
cotton
fields
must
be
planted
to
non­
Bt
cotton
and
may
be
treated
with
lepidopteran­
active
insecticides
(
or
other
control
technology)
except
for
microbial
Bt
formulations.
The
refuge
must
be
within
1
mile
(
preferably
within
½
mile
or
closer)
of
the
Bt
cotton
fields.

c.
5%
embedded
refuge
option
for
TBW
and
CBW.
Five
percent
of
a
cotton
field
(
or
Page
59
of
85
fields)
must
be
planted
with
non­
Bt
cotton
as
a
block
within
a
single
field,
at
least
150
ft.
wide
(
preferably
300
ft.
wide)
or
single­
field
blocks
within
a
one
mile
squared
field
unit.
The
refuge
may
be
treated
with
lepidopteran­
active
insecticides
(
or
other
control
technology)
only
if
the
entire
field
or
field
unit
is
treated
at
the
same
time.

d.
Embedded
(
in­
field
strip)
refuge
option
for
PBW.
One
single
row
of
a
non­
Bt
cotton
variety
must
be
planted
for
every
6
to
10
rows
of
Bt
cotton.
This
can
be
treated
with
lepidopteran­
active
insecticides
(
or
other
control
technology)
only
if
the
entire
field
is
treated
at
the
same
time.

e.
Community
refuge
option.
Farmers
can
combine
neighboring
fields
within
a
one­
mile
squared
field
unit
that
act
as
a
20%
sprayable
refuge
or
the
5%
unsprayed
refuge.
Participants
in
the
community
refuge
option
must
have
a
community
refuge
coordinator,
and
appropriate
documentation
is
required.
It
also
includes
the
requirements
for
annual
resistance
monitoring,
annual
compliance
assurance
program,
grower
education,
remedial
action
plans,
and
annual
reporting.
Any
plan
that
focuses
on
TBW,
CBW,
and
PBW
should
be
adequate
to
maintain
susceptibility
in
secondary
pests,
such
as
fall
armyworm,
beet
armyworm,
southern
armyworm,
cabbage
looper,
and
soybean
looper.
A
market
mix
of
different
Bt
cottons
and
other
control
technologies
further
reduces
the
expected
selection
pressure
for
resistance
from
the
Cry1F
and
Cry1Ac
proteins
expressed
in
WideStrike
cotton.

The
Agency
asks
the
SAP
to
comment
on
the
scientific
data
available
to
support
the
proposed
IRM
plan
and
whether
that
data
support
a
delay
in
resistance
of
TBW,
CBW,
and
PBW
resistance
to
the
Cry1F
and
Cry1Ac
proteins
expressed
in
WideStrike
cotton
for
at
least
15
years.

Panel
Response
The
refuge
requirements
proposed
in
the
WideStrike
cotton
IRM
plan
represent
a
continuation
of
refuge
options
that
have
been
available
to
growers
of
Bollgard
cotton.
The
apparent
success
of
the
high
dose/
refuge
strategy
in
avoiding
resistance
to
Cry1Ac
is
perhaps
the
strongest
argument
for
maintaining
the
status
quo,
even
if
uncertainty
exists
about
whether
the
IRM
plan
is
indeed
responsible
(
Fox
2003)
for
the
lack
of
identifiable
resistance
in
the
field.
While
earlier
arguments
for
>
16%
untreated
refuges
were
not
adopted
as
regulatory
policy
at
a
time
when
no
commercial
outcomes
were
known,
the
accumulated
data
8
years
later
suggest
that
the
gap
between
theory
and
practice
(
Denholm
and
Rowland
1992)
remains
substantial
in
terms
of
our
ability
to
make
fine­
scale
predictions.
On
the
other
hand,
theory
has
provided
the
basis
for
adopting
the
current
IRM
strategy
by
predicting
that
resistance
at
a
single
locus
would
be
significantly
delayed
if
there
was
a
low
initial
frequency
of
the
resistance
allele,
mating
between
resistant
and
susceptible
adults
occurred
extensively,
and
that
R
was
inherited
recessively
(
Comins
1977,
Curtis
1985,
Carriere
and
Tabashnik
2001).
The
outcome
thus
far
of
8
years
of
commercial
use
suggests
that
the
basic
theory
is
correct,
but
that
our
ability
to
precisely
define
the
parameters
within
the
theory
remains
limited.
Page
60
of
85
The
Panel
was
in
strong
agreement
that
the
proposed
IRM
plan
by
the
registrant
is
sufficient
for
WideStrike
cotton
and
supported
the
prediction
of
a
delay
in
resistance
of
TBW,
CBW
and
PBW
to
WideStrike
cotton
for
15
years.
The
proposed
theory
is
supportive
of
the
principle
of
delaying
resistance
by
confronting
adaptable
genomes
with
two
or
more
toxins.
In
particular,
model
simulations
performed
by
Curtis
(
1985)
of
the
time
to
resistance
when
target
populations
are
confronted
with
two
toxins
instead
of
one
show
that
resistance
will
be
substantially
delayed
so
long
as
certain
conditions
are
met.
The
goal
of
each
of
the
proposed
refuge
options
is
to
ensure
that
these
theoretical
conditions
are
met
in
the
real
world
of
IRM
for
Bt
cotton.
The
overall
consensus
was
that
the
existing
IRM
options
that
have
been
applied
to
the
single­
toxin
Bollgard
cotton
will
be
equally
or
even
more
effective
in
protecting
against
resistance
in
the
double­
toxin
WideStrike
cotton.

Their
strategy
seems
sufficiently
robust
for
the
CBW,
the
pest
with
the
perceived
greatest
threat
for
the
development
of
resistance.
Previous
SAPs
determined
the
placement
of
refugia
described
in
the
registrant's
document
were
sufficient
for
cotton
cultivars
expressing
single
Cry1
toxins.
Because
WideStrike
expresses
two
Cry1
proteins
at
reasonable
levels
during
the
entire
season,
the
registrant's
recommendations
for
refugia
placement
are
relatively
more
conservative
than
those
recommended
for
cottons
expressing
single
Cry1
toxins.
In
addition,
the
consistency
in
refugia
requirements
across
different
Cry
products
will
reinforce
grower
education
relative
to
IRM
and
refugia
placement
in
cotton,
increasing
the
probability
of
grower
compliance.

The
Panel
recommended
that
there
should
be
an
inspection
of
a
random
sample
of
the
5%
external
refuges
to
compare
the
quality
of
the
plants
(
nodes,
fruiting
structure)
in
the
refuge
and
the
Bt
crops
for
which
they
are
matched.
Plant
quality
alone
should
be
a
fairly
easy
and
low
cost
measure
of
the
overall
quality
of
the
refuges.
It
should
also
be
easy
to
check
spray
records
for
the
site
and
make
cursory
inspections
of
whether
insects
and
spiders
are
roughly
as
common
as
should
be
expected
in
an
unsprayed
field.
In
addition,
for
a
sample
of
some
of
the
crops
and
refuges,
there
should
also
be
a
comparison
of
egg
densities
of
the
key
targeted
pests,
at
least
for
TBW
and
CBW,
on
the
refuge
and
neighboring
Bt
crops.
Some
states,
such
as
Mississippi,
already
have
a
field
sampling
system
for
so­
called
"
index
fields".
It
may
be
possible
to
extend
this
system
to
include
sampling
targeted
and
recorded
for
Bt
and
refuge
fields.

The
registrant
made
certain
assumptions
that
have
been
built­
in
to
each
of
the
refuge
options,
and
the
role
of
the
refuges
in
supplying
susceptible
genotypes
in
the
management
of
resistance
was
too
critical
to
allow
the
assumptions
to
go
untested.
Empirical
testing
of
the
performance
of
the
various
refuge
options
should
be
pursued
with
the
goal
of
determining
if
susceptible
moths
are
being
generated
and
dispersed
into
Bt
fields
at
a
rate
high
enough
to
satisfy
the
theoretical
ratio
of
500:
1
susceptible
moths
to
resistant
moths.
The
following
questions
were
posed
as
a
heuristic
device
for
critically
examining
how
refuges
are
contributing
to
managing
resistance
of
Bt
cotton:

 
Are
embedded
refuges
as
efficient
as
external
refuges
at
recruiting
moths
relative
to
the
rate
of
recruitment
that
occurs
in
the
parent
block
of
Bt?
Page
61
of
85
 
Will
a
greater
relative
proportion
of
moths
visit
the
Bt
cotton
before
treatment
thresholds
are
reached
in
the
non­
Bt
embedded
refuge?

 
Do
embedded
refuges
have
the
potential
of
reducing
regional
densities
of
target
Lepidoptera
by
avoiding
the
sometimes
large
buildups
that
occur
in
refuges?

Empirical
answers
to
these
questions
and
others
could
help
to
evaluate
the
veracity
of
assumptions
underlying
each
of
the
refuge
options
and
would
also
help
with
the
parameterization
of
simulation
models.

Implementation
of
the
5%
external
refuges
should
include
measures
of
their
effectiveness
in
attracting
egg
deposition
and
generating
susceptible
moths.
For
some
years,
there
has
been
concern
that
many
of
these
external
refuges
are
not
grown
in
the
same
manner
as
the
Bt
crops
they
are
intended
to
serve
(
i.
e.,
since
growers
believe
that
these
refuges
will
likely
suffer
yield
losses,
the
untreated
external
refuges
tend
to
be
planted
late
on
the
poorer
land
and
with
less
attention
to
irrigation,
fertilizer,
weed
management,
etc.,
and
some
may
in
fact
be
sprayed
through
oversight
or
perhaps
intentionally).

The
Agency
could
not
address
how
often
the
5%
external
refuge
is
used
or
what
have
been
the
results
of
the
compliance
assessments
to
date.
There
are
some
indications
that
this
is
not
a
popular
refuge
strategy
and
may
be
declining
in
use.
However,
external
comments,
including
written
comments
from
the
National
Cotton
Council,
have
supported
its
continued
availability.

The
20%
external
sprayable
refuge
option
may
often
be
the
best
from
the
standpoint
of
resistance
management
for
Bt
crops.
A
5%
refuge
can
never
be
more
than
5%.
On
the
other
hand,
although
a
20%
refuge
may
be
sprayed
(
perhaps
reducing
it
to
a
4%
effective
refuge
when
spray
mortality
is
80%,
which
has
historically
been
close
to
the
average),
whenever
the
refuge
is
not
sprayed
(
as
it
often
won't
be
at
various
times
during
the
season
due
to
low
pest
density),
it
is
a
20%
refuge.

Implementation
of
embedded
refuges
began
some
years
after
the
initial
commercialization
of
Bt
cotton
in
1996.
A
point
was
made
that
one
of
the
motivations
for
embedded
refuges
was
to
have
the
source
of
susceptible
moths
within
the
Bt
cotton
field
to
promote
random
mating
between
susceptible
moths
generated
in
the
refuge
cotton
with
any
resistant
genotypes
that
might
arise
in
the
Bt
cotton.
The
within­
Bt
field
proximity
of
the
refuge
cotton
would
help
to
overcome
potential
weaknesses
of
the
external
refuges
that
require
dispersal
over
a
longer
distance.
One
Panel
member
commented
that
the
principal
reason
for
the
embedded
refuges
was
to
generate
more
susceptible
moths
by
having
a
wide
enough
refuge
blocks
within
the
Bt
cotton
to
retain
susceptible
moths
to
produce
even
higher
numbers
within
the
refuge.
There
are
contrasting
requirements
of
embedded
refuges
for
PBW
and
CBW,
and
that
concern
in
part
about
adequate
dispersal
of
the
weak
flyer
PBW
from
external
refuges
into
Bt
cotton
prompted
the
advent
of
embedded
refuges
in
Arizona
Bt
cotton
production.

Bollgard
and
Bollgard
II
Insect
Resistance
Management
Agency
Charge
Page
62
of
85
As
a
condition
of
the
Bollgard
and
Bollgard
II
registrations,
EPA
required
that
the
Monsanto
Company
conduct
CBW
alternate
host
research
studies
and
pyrethroid
overspray
studies
to
support
the
adequacy
of
the
5%
external,
unsprayed,
structured
refuge.
In
addition,
EPA
required
that
the
Monsanto
Company
conduct
research
on
the
north­
south
movement,
i.
e.,
reverse
migration,
of
CBW
and
its
impact
on
Bt
corn
and
cotton
insect
resistance
management.

1.
North­
south
movement.
Based
on
the
modeling
studies
submitted
using
the
data
in
Gould
et
al.
(
2002),
CBW
(
also
called
corn
earworm
in
corn)
reverse
migration
has
no
significant
impact
(
0.05<
P)
on
CBW
adaptation
to
Bt
corn
and
cotton.

The
Agency
requests
that
the
SAP
comment
on
whether
CBW
reverse
migration
is
expected
to
have
any
significant
impact
on
CBW
adaptation
to
Bt
crops.

Panel
Response
The
Panel
believed
that
the
spatially
explicit,
two
patch
model
presented
by
Agricultural
Biotechnology
Stewardship
Technical
Committee
(
ABSTC)
provided
conclusions
that
are
overly
optimistic
when
compared
to
more
simple
models
that
are
in
the
literature.
However,
there
is
too
little
detail
in
the
ABSTC
report
to
examine
what
factors
are
favoring
the
slow
build
up
of
resistance
alleles.
Storer
et
al.
(
2003)
does
provide
more
detail
relative
to
the
variables
and
their
effect
on
resistance
evolution.
The
Panel
believed
that
if
spatially
explicit
models
will
be
used,
more
detailed
modeling
is
necessary,
especially
with
regard
to
the
seven
prerequisites
the
ABSTC
report
states
are
necessary
for
reverse
migration
to
influence
the
development
of
resistance
in
Bt
corn
and
Bt
cotton.
Thus,
the
Panel
could
not
determine
whether
CBW
reverse
migration
is
expected
to
have
any
impact
on
CBW
adaption
to
Bt
cotton
or
Bt
corn.

Agency
Charge
2.
Pyrethroid
oversprays.
Pyrethroid
oversprays
in
Bollgard
cotton
fields
will
increase
the
level
of
control
of
CBW,
delay
the
evolution
of
resistance,
and
increase
the
relative
effectiveness
of
the
5%
external,
unsprayed,
structured
refuge.
These
findings
support
the
general
predictions
of
the
Gustafson
et
al.
(
2001/
2004)
model.
Pyrethroid
sprays
on
Bollgard
II
plots
do
not
provide
a
statistically
significant
difference
in
reduction
of
CBW
infestation
or
damage
from
untreated
Bollgard
II
cotton
fields
or
from
treated
Bollgard
cotton
fields,
and
should
not
be
included
as
a
parameter
in
the
Gustafson
et
al.
(
2004)
model.

a.
The
Agency
requests
that
the
SAP
comment
on
whether
pyrethroid
oversprays
in
Bollgard
cotton
fields
are
likely
to
increase
the
level
of
control
of
CBW,
delay
the
evolution
of
resistance,
and
increase
the
relative
effectiveness
of
the
5%
external,
unsprayed,
structured
refuge.
Page
63
of
85
b.
The
Agency
also
requests
that
the
SAP
comment
on
EPA's
recommendation
that
pyrethroid
oversprays
not
be
included
as
a
parameter
in
the
Gustafson
et
al.
(
2004)
model
for
Bollgard
II.

c.
Marcus
et
al.
(
2004)
found
that
CBW
larvae
(
late
instars)
in
North
Carolina
Bollgard
plots
were
half
as
susceptible
to
Cry1Ac
(
i.
e.,
more
tolerant)
as
were
populations
from
non­
Bollgard
cotton
survivors
in
the
F1
generation.

The
Agency
requests
the
SAP
comment
on
whether
the
cotton
bollworm
larvae
coming
from
Bollgard
fields
are
more
tolerant
to
the
Cry1Ac
protein
than
those
larvae
coming
from
the
non­
Bollgard
fields.
What,
if
any,
additional
genetic
work
should
be
conducted
to
better
understand
the
nature
of
this
Cry1Ac
tolerance.

d.
The
Agency
requests
the
SAP
to
comment
on
the
value
of
using
a
Cry1Ac­
resistant
CBW
colony
to
investigate
the
genetic
basis
for
CBW
survival
on
Bollgard
cotton.

Panel
Response
2a)
The
Panel
agreed
that
the
data
provided
by
the
registrant
show
significantly
fewer
larvae
and
less
damage
to
the
cotton
plant
in
pyrethroid­
treated
Bollgard
cotton
relative
to
Bollgard
cotton
that
was
not
sprayed,
and
this
a
consistent
pattern
at
all
four
locations.
Based
on
these
data,
it
appears
sound
that
pyrethroid
oversprays
in
Bollgard
cotton
improve
the
control
of
susceptible
CEW.

The
Panel
agreed
that
the
effect
of
pyrethroid
oversprays
in
delaying
resistance
in
CEW
is
probably
overstated.
In
their
document,
the
registrant
stated
that
"
Pyrethroids
may
be
relatively
more
effective
in
Bollgard
cotton
because
Bollgard
cotton
survivors
may
be
compromised
in
some
way,
or
there
may
be
an
increased
probability
of
Bollgard
cotton
survivors
contacting
pyrethroid
residues
on
Bollgard
cotton".
The
field
studies
relied
on
natural
infestations
of
CEW,
presumably
susceptible
individuals.
Gustafson
et
al.
(
2004)
assume
that
RR
genotypes
on
Bollgard
cotton
experience
the
same
17%
survival
in
the
face
of
insecticide
applications
as
observed
for
SS
genotypes
on
Bollgard
cotton.
If
instead
one
assumes
that
RR
genotypes
on
Bollgard
cotton
experience
the
same
35%
survival
as
observed
for
SS
genotypes
on
conventional
cotton
[
averages
reported
by
Greenplate
(
2004)],
then
RR
genotypes
would
enjoy
a
survival
twice
and
RS
genotypes
a
survival
up
to
twice
that
for
SS
genotypes
on
sprayed
Bollgard
cotton.
Consequently,
resistance
should
develop
faster
(
but
perhaps
only
slightly)
when
Bollgard
cotton
is
sprayed
than
estimated
by
Gustafson
et
al.
(
2004b).
Especially
because
the
Bollgard
product
is
not
a
high
dose
for
CEW,
it
will
be
important
to
consider
differing
susceptibilities
to
pyrethroids
as
it
relates
to
the
general
differences
among
resistant
genotypes
in
the
larval
stress
they
encounter
when
feeding
on
Bollgard
cotton.

The
magnitude
of
the
delay
to
resistance
in
CBW
given
insecticidal
sprays
on
Bollgard
Page
64
of
85
cotton
will
depend
on
whether
the
sprays
are
performed
by
producers
as
effectively
as
in
the
reported
studies.
For
example,
Jackson
et
al.
(
2001,
2002,
2003)
reported
making
two
insecticide
applications
6­
19
days
apart
during
mid­
July
to
mid­
August.
Is
this
typical
for
NC
or
for
other
regions
of
the
cotton
belt?
The
insecticidal
effect
on
differential
survival
of
RR
and
SS
genotypes
could
be
greater
if
two
generations
of
CBW
were
treated.

2b)
The
Panel
discussed
two
points
of
view
concerning
the
inclusion
of
pyrethroid
oversprays
of
Bollgard
II
fields
in
the
Gustafson
et
al.
(
2004)
model.
The
first
was
the
use
of
oversprays
within
a
resistance
management
context.
If
resistance
begins
to
evolve,
oversprays
could
potentially
help
to
control
resistant
individuals.
Reductions
in
the
numbers
of
resistant
adults
emerging
from
pyrethroid
oversprays
targeting
larvae
feeding
on
Bollgard
II
cotton
would
have
the
potential
to
diminish
the
numbers
of
resistant
adults
emerging
from
Bollgard
II
fields
relative
to
those
coming
from
refuges.
As
a
consequence,
the
probability
of
matings
between
resistant
individuals
from
the
Bollgard
II
cotton
and
the
refuge
would
be
enhanced.
Inclusion
of
oversprays
into
the
Gustafson
model
would
evaluate
the
benefits
of
pyrethroids
as
a
resistance
management
tool.

The
second
point,
and
the
focus
of
the
question
from
the
perspective
of
the
Agency,
refers
to
the
likelihood
that
growers
will
use
pyrethroid
oversprays
in
Bollgard
II
plots.
The
data
supplied
by
the
registrant
indicated
that
there
were
no
significant
differences
in
the
control
of
CBW
in
Bollgard
II
and
Bollgard
II
oversprayed
with
a
pyrethroid.
As
a
consequence,
growers
are
less
likely
to
use
pyrethroids
in
Bollgard
II
fields
due
to
the
excellent
control
of
CBW
provided
by
the
product.
However,
pyrethroids
and
other
insecticides
could
be
used
in
Bollgard
II
fields
to
control
pests
such
as
stink
bugs
and
plant
bugs
that
would
incidentally
impose
mortality
on
any
CBW
in
the
fields;
and
Bt
cotton
fields
are
generally
treated
mid
to
late
season
more
frequently
for
such
pests
than
are
conventional
cotton
fields.
Nevertheless,
at
least
in
the
midsouth,
the
incidence
of
insecticide
applications
to
Bollgard
II
cotton
that
would
impose
significant
mortality
on
CBW
is
likely
to
be
low.
The
Panel
agreed
that
there
is
little
need
to
include
pyrethroid
oversprays
in
Bollgard
II
plots
in
the
models
of
Gustafson
et
al.
(
2004)
from
this
perspective.
However,
the
Agency
may
want
to
consider
that
as
resistance
to
Bollgard
II
evolves,
more
larvae
will
be
present
in
Bollgard
II
fields
and
the
management
practices
used
by
growers
may
include
oversprays.

2c)
Based
on
the
data
submitted
by
the
registrant,
the
Panel
concluded
that
there
is
some
evidence
of
greater
tolerance
in
larvae
originating
from
Bollgard
fields
relative
to
those
coming
from
non­
Bollgard
fields.
Non­
overlap
of
fiducial
limits
is
a
conservative
measure
of
a
statistically
significant
difference
between
population
means:
i.
e.,
fiducial
limits
may
overlap
and
the
population
means
still
be
significantly
different.
Given
that
the
overlap
in
their
fiducial
limits
was
small,
the
LC50s
for
the
offspring
of
the
CBW
strain
collected
from
Bollgard
cotton
(
BGF1)
and
from
non­
Bollgard
cotton
(
NBTF1)
observed
by
Marcus
et
al.
(
2004)
might
well
be
significantly
different
(
P<
0.05).
Also,
inadvertent
selection
for
genetic
change
of
insect
populations
under
laboratory
conditions
can
occur
very
rapidly
 
especially
for
small
populations
and
when
mating
is
communal.
Of
course,
maternal
effects
may
also
be
operating.

Although
it
is
most
likely
that
these
survivors
are
significantly
resistant,
these
populations
Page
65
of
85
need
to
be
further
characterized
and
an
understanding
of
the
nature
of
resistance
developed
beyond
the
F2.
For
instance,
vigor
tolerance
needs
to
be
eliminated.
If
the
R­
trend
holds
up,
then
pyrethroid
oversprays
may
be
warranted
and
the
High
dose/
refugia
strategy
may
need
to
be
modified
to
combat
the
situation.
After
the
registrant
answers
these
questions,
they
may
develop
a
thorough
follow
up
study
thereafter.
In
this
way,
actual
resistance
evolution
in
a
high
dose/
refugia
system
could
be
documented
and
the
factors
contributing
to
this
evolution
understood.
If
the
individuals
derived
from
this
field
situation
were
found
to
be
resistant
after
scientific
scrutiny,
then
appropriate
culture,
selection
and
mechanism
determination
studies
should
be
pursued.
A
good
method
to
accomplish
this
is
via
paired
matings,
family
analyses
resulting
in
isolines
with
varying
tolerances
to
Cry1Ac.
Follow­
up
studies
should
include
the
mechanism
of
resistance
inheritance
as
well
as
comparison
with
other
lab
selected
strains.
Use
of
Bollgard
plant
material
in
assays
would
also
greatly
enhance
the
research.

2d)
Greater
knowledge
of
how
RR
individuals
respond
to
pyrethroid
treatments
would
improve
the
pyrethroid­
survival
parameter
of
Cry1Ac­
resistant
genotypes
in
the
Gustafson
et
al.
(
2004)
model
mentioned
above.
The
Panel
discussed
two
methods
to
conduct
such
experiments:
(
1)
long­
term
selection
in
the
laboratory
with
field­
collected
individuals
and;
(
2)
screening
a
great
number
families
from
the
field.
This
would
lead
to
characterizing
their
abilities
to
perform
on
Cry1Ac
media,
with
the
subsequent
creation
and
maintenance
of
isolines
with
varying
levels
of
resistance
to
Cry1Ac.

Mass
selections
in
the
laboratory
require
the
maintenance
of
large
colonies
over
long
periods
of
time.
If
resistance
alleles
are
rare,
there
is
a
reasonable
likelihood
that
the
resistant
individuals
will
not
be
"
seen"
in
the
assays
until
after
many
generations
due
to
their
representation
only
in
RS
individuals.
To
maintain
any
resistant
alleles,
it
will
be
necessary
to
use
unrealistically
low
concentrations
of
Cry1Ac
to
allow
the
survival
of
some
RS
individuals.
In
the
mass
mated
arenas
it
will
take
many
generations
before
sufficient
numbers
of
RS
x
RS
matings
occur.
The
concern
of
using
such
low
doses
is
that
resistance
mechanisms
created
via
mass
selections
will
be
artificial
with
respect
to
the
type
of
resistance
that
would
be
selected
for
Bollgard
plants
in
the
field.
The
laboratory
artifact
of
using
low
concentrations
of
Cry1Ac
would
not
provide
useful
data
for
inclusion
in
the
Gustafson
et
al.
model.

Screens
of
isolines
from
field­
collected
individuals
using
plant
tissue
or
concentrations
of
Cry1Ac
comparable
to
expression
in
Bollgard
cotton
will
provide
better
information.
The
initial
investment
in
research
time
when
resistance
is
rare
will
be
great
since
thousands
of
families
need
to
be
evaluated
in
at
least
the
F2
generation.
If
any
isolines
exhibit
great
survival
on
a
medium
containing
Cry1Ac
then
these
isolines
may
be
maintained
and
used
for
further
characterization
of
resistant
genotypes.
If
none
of
the
initial
families
exhibit
resistance,
they
can
be
discarded
and
therefore
eliminate
the
need
for
long­
term
rearing
for
a
major
resistant
trait
that
is
not
present.

Research
of
this
nature
would
make
an
excellent
addition
to
the
model,
but
finding
a
single
major
gene
will
be
difficult
and
unlikely.

Agency
Charge
Page
66
of
85
3.
Alternate
hosts.
Based
on
the
two­
year,
studies
in
five
states,
both
C3
and
C4
alternate
hosts
serve
as
unstructured
refugia.
Data
show
that
CBW
moths
are
produced
on
alternate
hosts
throughout
the
landscape
(
spatial
scale
is
greater
than
10
miles)
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.
That
is,
the
susceptible
CBW
moths
coming
from
alternate
hosts
will
reduce
the
intensity
of
Cry1Ac
and
Cry2Ab2
resistance
selection
in
CBW
and
lower
the
likelihood
of
resistance
evolution.
The
contribution
of
susceptible
CBW
adults
from
alternate
hosts
is
greater
than
that
from
the
5%
external,
unsprayed,
structured
non­
Bt
cotton
refuge.
Despite
the
limitations
EPA
has
identified
associated
with
the
Gustafson
et
al.
(
2001/
2004)
model,
the
CBW
alternate
host
data
support
the
model's
predictions
that
alternate
hosts
will
substantially
delay
resistance.

a.
Based
on
the
larval
productivity
analyses,
adult
productivity
analyses,
and
satellite
imaging
analysis,
the
Agency
asks
the
SAP
to
comment
on
the
relative
contribution
of
the
C3
and
C4
alternate
hosts
as
unstructured
refugia
to
dilute
CBW
resistance.

b.
Based
on
the
data,
the
Agency
also
asks
the
SAP
to
comment
on
the
spatial
and
temporal
scale
across
the
landscape,
e.
g.,
1
mile,
10
mile
etc.,
in
which
CBW
adult
production
should
be
evaluated.

c.
EPA
concludes
that
"
effective
refuge
size"
should
be
a
weighted
average
of
the
proportion
of
moths
coming
from
each
alternate
host
for
each
CBW
generation
(
5
to
6
generations)
in
each
cotton
production
system
(
geography).

The
Agency
asks
the
SAP
to
comment
on
how
to
quantitatively
or
semi­
quantitatively
calculate
"
effective
refuge
size"
locally
and
regionally
using
available
data
(
see
above).

Panel
Response
3a)
The
Panel
agreed
that
sufficient
data
were
provided
to
establish
that
C3
and
C4
alternate
hosts
function
to
some
degree
as
unstructured
refugia.
However,
the
Panel
expressed
concern
on
the
methodologies
used
to
assess
adult
productivity
in
the
alternate
hosts.
The
primary
concern
was
the
use
of
pheromone
traps,
and
to
a
lesser
extent
the
use
of
larval
counts,
to
quantify
adult
CBW
productivity
in
alternate
hosts.
The
Panel
indicated
that
pheromone
traps
typically
only
indicate
adult
male
CBW
activity
in
a
given
area.
Thus,
traps
do
not
provide
a
meaningful
measurement
of
adult
productivity.
The
Panel
also
addressed
the
need
to
examine
the
temporal
and
spatial
availability
of
alternate
hosts
in
relation
to
CBW
populations
produced
in
Bt
cotton.
One
Panel
member
indicated
that
behavioral
and
mating
data
are
necessary
to
confirm
the
dilution
of
resistance;
that
is,
susceptible
insects
from
alternate
hosts
are
in
fact
mating
with
adults
from
Bt
crops.

Corn
has
long
been
recognized
as
a
primary
producer
of
CBW
moths
and
undoubtedly
Page
67
of
85
contributes
susceptible
moths
to
the
system.
Also,
it
has
long
been
accepted
that
a
CBW
population
emerging
from
corn
will
infest
cotton
and
other
alternate
hosts,
as
described
by
the
registrant.
Thus,
the
temporal
occurrence
of
the
adult
populations
between
these
hosts
and
cotton
are
crucial
for
mating
to
occur
between
susceptible
adults
reared
in
alternate
hosts
and
resistant
adults
surviving
transgenic
cotton.
However,
the
larval
and
adult
productivity
data
provided,
in
addition
to
C3
and
C4
analyses,
only
indicates
the
occurrence
of
the
production
of
bollworm
moths
in
the
C3
and
C4
alternate
hosts.
Head
and
Voth
(
2004)
provide
data
on
alternate
hosts
defined
as
soybeans,
corn,
sorghum,
peanuts,
and
non­
Bt
cotton.
The
methodologies
for
estimating
larval
productivity
and
adult
productivity
are
probably
overestimates
and
underestimates,
respectively,
of
adult
CBW
production.
The
larval
productivity
measurements
based
on
counting
of
late
instars
assumes
that
all
counted
larvae
will
reach
adulthood.
Pheromone
traps
only
indicate
adult
male
activity
in
a
given
area,
not
a
measurement
of
adult
productivity.
Further,
the
traps
do
not
discriminate
between
immigrants
and
insects
produced
locally,
and
must
compete
with
other
insect
behaviors
(
i.
e.,
mating/
calling).
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.

The
number
of
adults
emerging
from
alternate
hosts
may
also
be
an
incomplete
description
of
the
impact
of
those
hosts
on
the
evolution
of
resistance.
As
the
Peck
et
al.
(
1999)
model
demonstrated
for
between
years,
moving
refuges
over
time
can
reduce
their
effectiveness.
Data
was
presented
demonstrating
in
several
versions
and
revisions
of
the
Caprio
et
al.
(
1998)
model,
that
refuges
that
are
temporally
unstable
over
time
during
the
season
(
such
as
early
and
late
soybean
fields,
different
wild
hosts)
will
be
less
effective
per
unit
area
than
are
refuges
that
are
temporally
stable
over
time
(
e.
g.,
persist
for
at
least
2
generations).
The
model
suggested
that
an
individual
adult
moving
from
corn
into
a
cotton
refuge
(
a
refuge
that
would
persist
for
the
next
two
generations)
would
have
a
realized
fecundity
3­
fold
greater
than
a
similar
adult
moving
into
early
soybeans
(
a
refuge
that
only
persisted
for
the
ensuing
generation).
In
this
case,
the
Panel
defines
realized
fecundity
as
the
number
of
offspring
that
are
descendants
of
the
adult
that
enter
diapause.
Those
offspring
would
be
G2
insects
(
second
generation
from
the
initial
adults).

Following
this
FIFRA
SAP
meeting,
Panel
member
Michael
Caprio
provided
additional
comments
on
C3
and
C4
alternate
hosts.
Such
comments
were
not
considered
or
reviewed
by
the
Panel
and
are
being
provided
as
an
appendix
to
these
meeting
minutes
(
Appendix
B).

3b)
The
Panel
agreed
that
CBW
production
should
be
measured
at
a
larger
scale
than
the
local
farm,
or
field
level
because
of
the
high
mobility
of
adult
CBW.
The
Panel
expressed
concern
that
even
the
10
mile
range
may
be
relatively
short
in
some
cases
based
on
previously
observed
migratory
movement
within
a
very
short
time
frame,
but
local
phenomena
may
also
be
important.
The
Panel
also
provided
one
example
(
Raulston
et
al.
1992)
of
previous
large­
scale
field
studies
for
measuring
CBW
production
over
a
large
production
region.

Jackson
et
al.
(
2003)
indicated
"
the
average
cotton
field
in
the
area
has
been
estimated
at
15
acres "
in
North
Carolina.
These
are
relatively
small
field
sizes
for
cotton
and
it
is
not
Page
68
of
85
surprising
that
CBW
moth
movement
across
the
landscape
between
C3
and
C4
plants
would
be
observed.
After
all,
we
know
that
adult
CBW
can
move
approximately
400
km
in
just
under
8
hours
of
migratory
flight.
In
these
situations,
it
would
seem
that
spatial
and
temporal
scales
for
assessing
CBW
should
likely
be
evaluated
in
an
area
that
encompasses
a
representative
sample
of
all
suspected
C3
and
C4
hosts
within
an
agroecosystem
and
perhaps
other
cotton
production
regions.
Again,
there
is
the
problem
of
quantifying
the
migrants
and
their
effect
on
the
system.

As
previously
mentioned,
field
studies
for
assessing
CBW
development
are
not
simple
tasks
but
can
be
done.
For
example,
in
a
large­
scale
study
for
assessing
CBW
population
development
over
a
large
production
area,
Raulston
et
al.
(
1992)
used
pupal
digs
to
determine
the
number
of
moths
produced
from
fruiting
corn
in
northeastern
Mexico
and
South
Texas.
Nighttime
observations
have
also
been
used
to
assess
adult
abundance
in
a
region.
These
types
of
studies
provide
more
accuracy
in
terms
of
spatial
and
temporal
production
of
CBW
adults
but
labor
requirements
and
logistics
will
probably
be
the
limiting
factor
in
determining
the
extent
of
future
spatial
and
temporal
assessments.

3c)
The
Panel
agreed
that
"
effective
refuge
size"
should
be
a
weighted
average
of
the
proportion
of
moths
produced
from
each
alternate
host.
The
Panel
commented
that
the
term
"
effective
refuge
size"
implies
evaluation
of
varying
plot
sizes
of
all
alternate
hosts
and
suggested
that
some
clarification
by
the
Agency
may
be
in
order.
The
Panel
provided
additional
input
pertinent
to
this
issue
in
response
to
Question
4a
posed
by
the
Agency
for
Bollgard
and
Bollgard
II.
In
response
to
the
request
for
methods
on
quantitatively
calculating
"
effective
refuge
size",
the
Panel
provided
techniques
for
quantifying
CBW
populations
in
the
identified
alternate
hosts
that
were
identified
as
natural
refugia.

The
term
"
effective
refuge
size"
implies
the
evaluation
of
various
"
refuge
sizes"
and
sources
(
i.
e.
C3
and
C4)
to
identify
that
which
is
best
suited
for
providing
the
necessary
numbers
of
susceptible
insects.
Therefore,
the
techniques
described
below
for
measuring
adult
production
would
need
to
be
replicated
over
a
range
of
"
refuge
sizes"
and
suspected
sources.
These
data,
in
addition
to
the
described
biological
data,
would
also
assist
in
refining
the
Gustafson
et
al.
(
2001)
model
especially
since
it
would
include
corn,
the
preferred
CBW
host,
as
a
potential
source
of
susceptible
insects.
Gustafson
et
al.
(
2001)
did
not
include
corn
as
a
source
of
susceptible
insects.

One
Panel
member
was
of
the
opinion
that
an
"
effective
refuge
size"
cannot
be
calculated,
based
on
the
data
provided,
for
two
reasons:
1)
the
methodologies
with
which
larval
and
adult
productivity
estimates
were
obtained;
and
2)
lack
of
definitive
biological
data
(
i.
e.
temporal
and
spatial
adult
emergence
from
various
C3
and
C4
plants,
and
behavioral
or
mating
observations).

Thus
how
does
one
obtain
these
biological
data?
The
C3/
C4
data
presented
here
provides
a
starting
point
in
terms
of
relative
measurements
of
the
population,
but
as
previously
mentioned
these
are
probably
inadequate
for
providing
precise
population
numbers.
Other
possibilities
for
obtaining
quantitative
data
have
been
previously
implemented
(
Gore
et
al.
2004;
Jackson
et
al.
2002,
2003;
Raulston
et
al.
1992).
Some
of
these
methods
included:
Page
69
of
85
1.
Deployment
of
emergence
cages
(
possibly
a
square
meter
in
size
but
this
does
not
preclude
other
sizes)
throughout
all
suspected
C3
and
C4
sources,
under
varying
soil
types,
irrigation,
etc.,
to
provide
a
more
accurate
picture
of
spatial
adult
productivity
in
each
of
the
suspected
sources.
This
would
address
questions
regarding
whether
late
instars
reached
adulthood,
whether
males
captured
in
pheromone
traps
were
produced
locally
or
moved
in
from
adjacent
sources,
and
would
provide
a
measure
of
sex
ratios.
Additionally,
moths
emerging
from
these
sources
could
undergo
genetic
analyses
to
provide
data
on
the
genetic
composition
of
the
emerging
adult
population.
This
latter
would
be
very
informative
given
the
assumption
that
susceptible
moths
mate
with
resistant
(
homozygous
and
heterozygous)
moths.
Gore
et
al.
(
2004)
used
similar
caging
techniques
to
estimate
temporal
emergence
of
CBW
adults
from
field
corn.
2.
Harvesting
of
late
instars
on
developing
fruit
and
subsequent
monitoring
of
insect
development
to
adulthood
on
the
same
larval
food
source.
Jackson
et
al.
(
2002,
2003)
estimated
production
of
CBW
in
Bollgard
and
Bollgard
II
cotton
under
differing
insecticide
treatment
regimes.
3.
Digging
for
pupae
and
holding
pupae
in
surrogate
cells
in
the
soil
to
estimate
emergence
based
on
Raulston
et
al.
(
1992).
4.
Night­
time
observations
to
assess
insect
activity
in
an
area
and
obtain
information
such
as
occurrence
and
timing
of
mating
behavior,
mating
frequency
based
on
dissections
of
adults,
and
a
general
idea
of
the
effect
of
the
host
crop
on
the
insect
behaviors.

Based
on
previous
studies
(
Gore
et
al.
2004;
Jackson
et
al.
2002,
2003;
Raulston
et
al.
1992),
there
are
some
biological
data
on
corn
and
cotton,
two
of
the
identified
"
alternative
hosts,"
but
additional
data
are
needed
for
the
remaining
hosts.

Agency
Charge
4.
Gustafson
et
al.
CBW
model.
Monsanto
modified
Caprio's
(
1998a)
two­
patch,
deterministic,
non­
random,
population
genetics
model
(
publically
available)
to
create
a
new
CBW
model,
Gustafson
et
al.
(
2004,
originally
submitted
to
the
Agency
in
September
2001
as
part
of
the
Bt
Crops
Reassessment)
that
included
alternate
hosts
and
synthetic
pyrethroid
oversprays
as
parameters.
Sensitivity
analyses
showed
that
the
model
output
(
years
to
resistance)
was
sensitive
to
both
of
these
parameters.
Gustafson
et
al.
(
2004)
have
calculated
"
effective
refuge
size"
as
the
sum
of
the
total
acres
by
county
represented
by
the
four
alternate
crop
hosts
 
corn,
sorghum,
peanuts,
and
soybeans,
and
wild
hosts
(
defaulted
as
10%
of
the
cotton
acreage)
as
a
percent
of
cotton
acres.
This
model
predicts
that
the
5%
external,
unsprayed,
structured
refuge
option
is
adequately
protective
to
delay
CBW
resistance
if
effective
refuge
size
(
alternate
hosts)
and
typical
use
practices
for
Bollgard
cotton,
i.
e.,
synthetic
pyrethroid
oversprays,
are
included.
When
this
model
was
submitted
to
the
Agency
in
2001,
empirical
data
to
support
the
use
of
alternate
hosts
and
synthetic
pyrethroid
were
lacking.

a.
The
Agency
asks
the
SAP
to
comment
on
the
"
effective
refuge
size"
calculation.
Does
the
SAP
agree
with
the
Agency's
conclusion
that
"
effective
refuge
size"
is
a
weighted
average
Page
70
of
85
of
the
proportion
of
moths
coming
from
each
alternate
host
for
each
CBW
generation
(
5
to
6
generations)
in
each
cotton
production
system
(
geography)?

b.
The
Agency
requests
the
SAP
to
comment
on
the
strengths
and
weaknesses
of
the
Gustafson
et
al.
(
2004)
model
and
its
utility
with
regard
to
the
effective
contribution
of
alternate
hosts
as
natural
refuge
per
generation.
How
would
the
model
output
be
altered
if
the
calculation
of
"
effective
refuge
size"
is
changed
(
see
a.
above).
What
are
the
SAP's
recommendations
for
refining
the
Gustafson
et
al.
(
2004)
CBW
resistance
management
model
or
using
a
different
CBW
resistance
management
model
to
more
appropriately
consider
the
spatial
and
temporal
dynamics
of
CBW
utilization
of
alternative
hosts
by
generation
based
on
the
data
in
Head
and
Voth
(
2004)?

c.
The
Agency
requests
the
SAP
to
comment
on
validity
of
using
the
average
pyrethroid
efficacy
value
against
CBW
based
on
all
the
field
studies
conducted
in
all
four
states
(
North
Carolina,
Louisiana,
Mississippi,
and
South
Carolina)
as
the
parameter
value
in
the
Gustafson
et
al.
(
2004)
model
rather
than
just
the
Brickle
et
al.
(
2001)
data
from
South
Carolina.

Panel
Response
a.)
The
Panel
agreed
with
the
Agency
that
a
weighted
average
is
an
appropriate
choice
for
determining
the
contribution
of
alternate
hosts
to
the
refuge
size.
The
Panel
noted
that
the
estimation
of
the
total
refuge
proportion
also
requires
an
estimate
of
the
emergence
from
transgenic
crops
(
e.
g.,
the
proportion
in
refuge
is
relative
to
the
total
number
of
adults,
including
those
emerging
from
transgenics).
When
this
estimate
of
total
adult
emergence
is
made,
the
numbers
emerging
from
transgenic
crops
(
where
the
gene(
s)
of
interest
are
utilized)
should
be
corrected
for
losses
due
to
selection
(
see
Appendix
B).
Other
transgenics,
assuming
there
is
not
an
interest
in
resistance
to
transgenetics
and
other
pesticides,
should
be
assumed
to
be
additional
mortality
factors
and
no
correction
applied.

b.)
The
Gustafson
et
al.
model
is
a
deterministic,
two
patch,
generational
model
which
will
be
useful
for
exploring
broad
general
questions
about
the
use
of
alternate
hosts
and
pyrethroid
oversprays
in
Bollgard
and
Bollgard
II
transgenic
cotton.
Continued
use
of
the
model
would
certainly
require
incorporation
in
a
detailed
fashion
of
the
data
acquired
through
the
alternate
host
plant
study.
The
Panel
believed,
however,
that
exploring
detailed
questions
about
time
to
resistance
and
the
effect
of
alternate
hosts
on
resistance
would
benefit
from
the
development
of
a
more
detailed
model.
Specifically,
a
spatially
explicit
model
that
can
pick
up
the
more
nuanced
structure
and
timing
of
insect
emergence
in
alternate
hosts
will
be
necessary
to
address
the
questions
posed
by
the
Agency.

c.)
The
Panel
concluded
that
it
would
have
been
prudent
to
explore
the
impact
of
the
data
developed
as
a
result
of
the
Agency's
request.
The
additional
data
should
impact
the
mean
or
mid­
value
used,
and
perhaps
also
have
some
impact
on
the
extreme
values.
It
would
seem
that
the
additional
data
might
alter
the
relative
position
of
the
extreme
values
(
e.
g.,
are
they
symmetric
Page
71
of
85
about
the
mid­
value),
as
well
as
the
breadth
of
those
values
(
e.
g.,
as
additional
data
has
been
collected,
should
we
not
have
more
confidence
in
the
mid­
values?)
As
noted
by
the
Panel's
response
to
question
2(
a),
the
Panel
believed
it
is
likely
that
the
impact
of
pyrethroid
oversprays
might
vary
with
genotype,
and
this
must
be
incorporated
into
the
model.
Page
72
of
85
REFERENCES
Al­
Deeb
MA
and
GE
Wilde.
2003.
Effect
of
Bt
corn
expressing
the
Cry3Bb1
toxin
for
corn
rootworm
control
on
above
ground
non­
target
arthropods.
Environ.
Entomol.
32:
1164­
1170.

Al­
Deeb
MA,
GE
Wilde,
JM
Blair
and
TC
Todd.
2003.
Effect
of
Bt
corn
for
corn
rootworm
control
on
non­
target
soil
microarthropods
and
nematodes.
Environ.
Entomol.
32:
859­
865.

Arriola
P.
1998.
Benefits
and
risks
of
genetic
modification
in
agriculture.
Are
we
too
late?
http://
ezproxy.
library.
cornell.
edu:
2195/
nature/
debates/
gmfoods/
gmfoods_
frameset.
html
Bah
A,
FK
Van,
R
Brousseau,
and
L
Masson.
2004.
The
Bacillus
thuringiensis
Cry1Aa
toxin:
effects
of
trypsin
and
chymotrypsin
site
mutations
on
toxicity
and
stability.
J.
Invert.
Pathol.
85.

Banks
DJ,
FJL
Jurat,
DH
Dean
and
MJ
Adang.
2001.
Bacillus
thuringiensis
Cry1Ac
and
Cry1Fa
delta­
endotoxin
binding
to
a
novel
110
kDa
aminopeptidase
in
Heliothis
virescens
is
not
Nacetylgalactosamine
mediated.
Insect
Biochem.
Mol.
Biol.
31:
909­
918.

Bhat
and
Srinivasan.
2002.
Molecular
and
genetic
analyses
of
transgenic
plants:
Considerations
and
approaches.
Plant
Sci.
163:
673­
681.

Brewer
LW,
GP
Cobb,
MJ
Hooper,
RJ
Kendall
and
CM
Bens.
1990.
Third
Year
Investigation
of
the
Response
of
Selected
Wildlife
Populations
to
Planting
Time
Application
of
COUNTER
15G
Systemic
Insecticide­
Nematicide
in
an
Iowa
Corn
Agroecosystem.
Submitted
to
American
Cyanamid
Company.

Brewer
LW,
GP
Cobb,
MJ
Hooper,
RJ
Kendall
and
CP
Weisskopf.
1992.
An
Assessment
of
Chemical
Exposure
and
Survival
of
Avian
and
Mammalian
Species
Following
Planting­
Time
Application
in
a
Typical
Midwestern
Corn
Agroecosystem.
Submitted
to
E.
I.
duPont
de
Nemours
and
Company,
Inc.

Brewer
LW,
DE
Dix,
GP
Cobb,
MJ
Hooper,
SL
Tank,
RA
Craft,
JA
Buck,
WM
Buerger,
RJ
Kendall,
CM
Bens.
1989.
Second
year
investigation
of
the
response
of
selected
wildlife
populations
to
planting
time
application
of
COUNTER
15­
G
systemic
insecticide­
nematicide
in
an
Iowa
corn
agroecosystem.
152
pp.

Candolfi
MP,
K
Brown,
C
Grimm,
B
Reber
and
H
Schmidli.
2004.
A
faunistic
approach
to
assess
potential
side­
effects
of
genetically
modified
Bt­
corn
on
non­
target
arthropods
under
field
conditions.
Biocontrol
Sci.
Technol.
14:
129­
170.

Carriere
Y
and
BE
Tabashnik.
2001.
Reversing
insect
adaptation
to
transgenic
insecticidal
plants.
Proceedings
of
the
Royal
Society
of
London
Series
B
268,
1475­
1480.

Chakrabarti
SK,
Mandaokar
A,
Kumar
PA,
Sharma
RP.
Efficacy
of
lepidopteran
specific
delta­
Page
73
of
85
endotoxins
of
Bacillus
thuringiensis
against
Helicoverpa
armigera
J
Invertebr
Pathol.
1998
Nov;
72(
3):
336­
7.

Chassy
BM,
JJ
Hlywka,
GA
Kleter,
EJ
Kok,
HA
Kuiper,
M
McGloughlin,
IC
Munro,
RH
Phipps,
JE
Reid
(
2004).
Nutritional
and
Safety
Assessments
of
Foods
and
Feeds
Nutritionally
Improved
through
Biotechnology.
Comp.
Rev.
Food
Sci.
Food
Safety,
3:
35­
104.
http://
www.
ift.
org/
pdfs/
crfsfs/
crfsfsv3n2p0035­
0104ms20040106.
pdf
CODEX
guidelines
http://
www.
codexalimentarius.
net/
ccfbt4/
bt03_
01e.
htm
Comins
HN.
1977.
The
development
of
insecticide
resistance
in
the
presence
of
migration.
Journal
of
Theoretical
Biology
64,
177­
197.

Conner
AJ,
TR
Glare
and
JP
Nap.
2003.
The
release
of
genetically
modified
crops
into
the
environment.
Part
II:
Overview
of
ecological
risk
assessment.
Plant
J.
33:
19­
46.

Crecchio
C
and
G
Stotzky.
1998.
Insecticidal
activity
and
biodegradation
of
the
toxin
from
Bacillus
thuringiensis
subsp.
kurstaki
bound
to
humic
acids
from
soil.
Soil
Biol.
Biochem.
30:
463­
470.

Croft
BA.
1990.
Arthropod
biological
control
agents
and
pesticides.
John
Wiley
and
Sons,
New
York,
723
pp.

Courtial
B,
F
Feuerbach,
S
Eberhard,
L
Rohmer,
H
Chiapello,
C
Camilleri
and
H
Lucas.
2001.
Tnt1
transposition
events
are
induced
by
in
vitro
transformation
of
Arabidopsis
thaliana,
and
transposed
copies
integrate
into
genes.
Mol
Genet
Genomics
265:
32­
42.

Curtis
CF.
1985.
Theoretical
models
of
the
use
of
insecticide
mixtures
for
the
management
of
resistance.
Bulletin
of
Entomological
Research
75:
259­
265.

Denholm
I
&
MW
Rowland.
1992.
Tactics
for
managing
pesticide
resistance
in
arthropods:
theory
and
practice.
Annual
Review
of
Entomology
37:
91­
112.

Devare
MH,
CM
Jones
and
JE
Thies.
2004.
Effect
of
Cry3Bb
transgenic
corn
and
tefluthrin
on
the
soil
microbial
community:
biomass,
activity,
and
diversity.
J
Environ.
Qual.
33:
837­
843.

Dively
G
and
R
Rose.
2002.
Effects
of
Bt
transgenic
and
conventional
insecticides
on
the
natural
enemy
community
in
sweet
corn.
pp.
9­
22.
In
Hoddle
M,
Proceedings
3rd
California
Conference
on
Biological
Control,
15­
16
August
2002,
Davis,
CA.

Elzen
GW.
1989.
Sublethal
effects
of
pesticides
on
beneficial
parasitoids.
pp.
129­
150.
Pesticides
and
Non­
Target
Invertebrates.
Intercept
Limited,
UK.

Ervin
DE,
R
Welsh,
SS
Batie
and
CL
Carpentier.
2003.
Towards
an
ecological
systems
approach
Page
74
of
85
in
public
research
for
environmental
regulation
of
transgenic
crops.
Agric.
Ecosyst.
Environ.
99:
1­
14.

Ferre
J
and
J
Van
Rie.
2002.
Biochemistry
and
genetics
of
insect
resistance
to
Bacillus
thuringiensis.
Ann.
Rev.
Entomol.
47:
501­
533.

Forcada
C,
E
Alcacer,
MD
Garcera
and
R
Martinez.
1996.
Differences
in
the
midgut
proteolytic
activity
of
two
Heliothis
virescens
strains,
one
susceptible
and
one
resistant
to
Bacillus
thuringiensis
toxins.
Arch.
Insect
Biochem.
Physiol.
31:
257­
272.

Forsbach
A,
Schubert
D,
Lechtenberg
B,
Gils
M,
and
Schmidt
R.
A
comprehensive
characterization
of
single­
copy
T­
DNA
insertions
in
the
Arabidopsis
thaliana
genome.
Plant
Mol
Biol.
2003
May;
52(
1):
161­
76.

Fox
JL.
2003.
Resistance
to
Bt
toxin
surprisingly
absent
from
pests.
Nature
Biotechnology
21:
958­
959.

GAO.
1992.
United
States
General
Accounting
Office
Report
to
Congressional
Requesters.
GENETICALLY
MODIFIED
FOODS,
Experts
View
Regimen
of
Safety
Tests
As
Adequate,
But
FDA's
Evaluation
Process
Could
Be
Enhanced.
GAO­
02­
566,
May
2002.

Gahan
LJ,
F
Gould
and
DG
Heckel.
2001.
Identification
of
a
gene
associated
with
bit
resistance
in
Heliothis
virescens.
Science
293:
857­
860.

Gould
F,
A
Anderson,
A
Reynolds,
L
Bumgarner
and
W
Moar.
1995.
Selection
and
genetic
analysis
of
a
Heliothis
virescens
(
Lepidoptera:
Noctuidae)
strain
with
high
levels
of
resistance
to
Bacillus
thuringiensis
toxins.
J.
Econ.
Entomol.
88:
1545­
1559.

Gould
F,
N
Blair,
M
Reid,
TL
Rennie,
J
Lopez,
and
S
Micinski.
2002.
Bacillus
thuringinsis­
toxin
resistance
management:
Stable
isotope
assessment
of
alternate
host
use
by
Helicoverpa
zea.
Proc.
Nat.
Acad.
Sci.
USA
99:
16581­
16586.

Gould
F,
S
Micinski,
J
Lopez,
M
Laster,
R
Leonard,
A
Jones,
A
Anderson,
DG
Heckel,
D
Sumerford.
1997.
Initial
frequency
of
alleles
for
resistance
to
Bacillus
thuringiensis
toxins
in
field
populations
of
Heliothis
virescens.
Proceedings
of
the
National
Academy
of
Sciences
of
the
United
States
of
America.
94
(
8):
3519­
3523.

Gore
J,
DD
Hardee,
JJ
Adamczyk
and
CA
Abel.
2004.
Population
dynamics
of
bollworm
larvae
and
adults
in
Mississippi.
Proceedings
Beltwide
Cotton
Conferences
(
Poster).

Groot
AT
and
M
Dicke.
2002.
Insect­
resistant
transgenic
plants
in
a
multi­
trophic
context.
Plant
J.
31:
387­
406.

Gustafson
et
al.
2001.
Impact
of
refuge
size
and
typical
insecticide
use
practices
on
model
Page
75
of
85
predictions
of
years
to
resistance
of
tobacco
budworm
and
cotton
bollworm
to
Bollgard
cotton.
Monsanto
Technical
Reports.
MSL­
19229.

Haslberger
AG.
2003.
Codex
guidelines
for
GM
food
include
the
analysis
of
unintended
effects.
Nature
Biotechnology
21:
739­
741.

Head
GH
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
Report
MSL­
19238.

Head
G,
JB
Surber,
JA
Watson,
JW
Martin
and
JJ
Duan.
2002.
No
detection
of
Cry1Ac
protein
in
soil
after
multiple
years
of
transgenic
Bt
cotton
(
Bollgard)
use.
Environ.
Entom.
31:
30­
36.

Helmich
BK,
Dutton
RW.
The
role
of
adoptively
transferred
CD8
T
cells
and
host
cells
in
the
control
of
the
growth
of
the
EG7
thymoma:
factors
that
determine
the
relative
effectiveness
and
homing
properties
of
Tc1
and
Tc2
effectors.
J
Immunol.
2001
Jun
1;
166(
11):
6500­
8.

Ho,
Cummings,
http://
www.
i­
sis.
org.
uk/
index.
php
Kuiper
HA,
EJ
Kok,
KH
Engel.
Exploitation
of
molecular
profiling
techniques
for
GM
food
safety
assessment,
Curr.
Opin.
Biotechnol.
2003
Apr.
14(
2):
238­
43.

Kuiper
HA,
GA
Kleter,
PJM
Noteboom,
EJ
Kok.
2001.
Assessment
of
the
food
safety
issues
related
to
genetically
modified
foods.
The
Plant
Journal,
27:
503­
528.

Jackson
RE,
JR
Bradley,
Jr.
and
JW
Van
Duyn.
2003.
Bollworm
population
production
and
associated
damage
in
Bollgard
and
Bollgard
II
cottons
under
insecticide­
treated
and
non­
treated
conditions.
Proceedings
Beltwide
Cotton
Conferences.
CD_
ROM.
pp.
1022­
1025.

Jackson
RE,
JR
Bradley,
Jr.
and
JW
Van
Duyn.
2002.
Estimated
production
of
Helicoverpa
zea
adults
from
Bollgard
and
Bollgard
II
cottons
and
implications
for
resistance
management.
Proceedings
Beltwide
Cotton
Conferences.
CD­
ROM.

Jepson
PC.
1993.
Insects,
spiders
and
mites,
pp.
299­
325.
In
P
Calow
(
Ed.),
Handbook
of
Ecotoxicology.
Blackwell
Scientific
Publications,
UK.

Jepson
PC.
2002.
Scale
dependency
in
the
impact
of
pesticides
on
invertebrate
natural
enemies.
P.
100­
106.
In
M
Hoddle,
Proceedings
3rd
California
Conference
on
Biological
Control,
15­
16
August
2002,
Davis,
CA.

Jepson
PC
and
JRM
Thacker.
1990.
Analysis
of
the
spatial
component
of
pesticides
side­
effects
on
non­
target
invertebrate
populations
and
its
relevance
to
hazard
analysis.
Functional
Ecol.
4:
349­
355.
Page
76
of
85
Jurat­
Fuentes
JL
and
MJ
Adang.
2001.
Importance
of
Cry1
delta­
endotoxin
domain
II
loops
for
binding
specificity
in
Heliothis
virescens
(
L.).
Appl.
Env.
Microbiol.
67:
323­
329.

Kok
E
and
H
Kuiper.
2003.
Comparative
Safety
Assessment
For
Biotech
Crops.
Trends
In
Biotechnology.
21(
10):
439­
444.

Koprek
T,
S
Rangel,
D
McElroy,
JD
Louwerse,
RE
Williams­
Carrier
and
PG
Lemaux.
2001.
Transposon­
mediated
single­
copy
gene
delivery
leads
to
increased
transgene
expression
stability
in
barley.
Plant
Physiol.
125:
1354­
62.

Koskella
J
and
G
Stotzky.
1997.
Microbial
utilization
of
free
and
clay­
bound
insecticidal
toxins
from
Bacillus
thuringiensis
and
their
retention
of
insecticidal
activity
after
incubation
with
microbes.
Appl.
Environ.
Microb.
63:
3561­
3568.

Lee
MK,
F
Rajamohan,
F
Gould
and
DH
Dean.
1995.
Resistance
to
Bacillus
thuringiensis
CryIA
delta­
endotoxins
in
a
laboratory­
selected
Heliothis
virescens
strain
is
related
to
receptor
alteration.
Appl.
Env.
Microbiol.
61:
3836­
3842.

Luo
K,
D
Banks
and
MJ
Adang.
1999.
Toxicity,
binding,
and
permeability
analyses
of
four
Bacillus
thuringiensis
Cry1
delta­
endotoxins
using
brush
border
membrane
vesicles
of
Spodoptera
exigua
and
Spodoptera
frugiperda.
Appl.
Env.
Microbiol.
65:
457­
464.

Men
XY,
F
Ge,
XH
Liu,
and
EN
Yardim.
2003.
Diversity
of
arthropod
communities
in
transgenic
Bt
cotton
and
nontransgenic
cotton
agroecosystems.
Environ.
Entomol.
32:
270­
275.

Moar
WJ,
M
Eubanks,
B
Freeman,
S
Turnipseed,
J
Ruberson
and
G
Head.
2002.
Effects
of
Bt
cotton
on
natural
enemies
in
the
Southeast
USA.
pp.
31­
32.
In
Hoddle
M,
Proceedings
3rd
California
Conference
on
Biological
Control,
15­
16
August
2002,
Davis,
CA.

Musser
FR
and
AM
Shelton.
2003.
Bt
sweet
corn
and
selective
insecticides:
Impacts
on
pests
and
predators.
J.
Econ.
Entomol.
96:
71­
80.

Naranjo
SE.
Arthropod
communities
and
transgenic
cotton
in
the
Western
US.
Pp.
33­
38.
In
Hoddle
M,
Proceedings
3rd
California
Conference
on
Biological
Control,
15­
16
August
2002,
Davis,
CA.

Masson
L,
YJ
Lu,
A
Mazza,
R
Brousseau
and
MJ
Adang.
1995.
The
CryIA(
c)
receptor
purified
from
Manduca
sexta
displays
multiple
specificities.
J.
Biol.
Chem.
270:
20309­
20315.

Morin
S,
RW
Biggs,
MS
Sisterson,
L
Shriver,
C
Ellers
Kirk,
D
Higginson,
D
Holley,
LJ
Gahan,
DG
Heckel,
Y
Carriere,
TJ
Dennehy,
JK
Brown
and
BE
Tabashnik.
2003.
Three
cadherin
alleles
associated
with
resistance
to
Bacillus
thuringiensis
in
pink
bollworm.
PNAS
100:
5004­
5009.

Nacry
P,
Camilleri
C,
Courtial
B,
Caboche
M,
Bouchez
D.
Major
chromosomal
rearrangements
Page
77
of
85
induced
by
T­
DNA
transformation
in
Arabidopsis.
Genetics.
1998
Jun;
149(
2):
641­
50.

National
Research
Council.
2002.
"
Environmental
Effects
of
Transgenic
Plants:
the
Scope
and
Adequacy
of
Regulation."
Committee
on
Environmental
Impacts
Associated
with
Commercialization
of
Transgenic
Plants
and
Board
on
Agriculture
and
Natural
Resources
Division
on
Earth
and
Life
Studies,
Natural
Research
Council.
National
Academy
Press,
Washington,
DC.

Norton
RA.
2000.
Chapter
1,
Figure
2.
In:
Ingham
ER,
AR
Moldenke
and
CA
Edwards.
Soil
Biology
Primer.
Soil
and
Water
Conservation
Society.

Novak,
WK
and
AG
Haslberger.
2000.
Substantial
equivalence
of
antinutrients
and
inherent
plant
toxins
in
genetically
modified
novel
foods.
Food
Chem
Toxicol
38:
473­
83.

Oppert
B,
KJ
Kramer,
RW
Beeman,
D
Johnson
and
WH
McGaughey.
1997.
Proteinase­
mediated
insect
resistance
to
Bacillus
thuringiensis
toxins.
J.
Biol.
Chem.
272:
23473­
23476.

Peck
SL,
F
Gould,
S
Ellner.
1999.
The
spread
of
resistance
in
spatially
extended
systems
of
transgenic
cotton:
Implications
for
the
management
of
Heliothis
virescens
(
Lipidoptera:
Noctuidae).
Economic
Entomology
92:
1­
16.

Perry
JN,
P
Rothery,
SJ
Clark,
MS
Heard
and
C
Hawes.
2003.
Design,
analysis
and
statistical
power
of
the
Farm­
Scale
Evaluations
of
genetically
modified
herbicide­
tolerant
crops.
J.
Appl.
Ecol.
40:
17­
31.

Potrykus
I.
2001.
Golden
rice
and
beyond.
Plant
Physiol.
125:
1157­
1161.

Prasifka
JR,
KM
Heinz
and
KO
Winemiller.
2004.
Crop
colonisation,
feeding,
and
reproduction
by
the
predatory
beetle,
Hippodamia
convergens,
as
indicated
by
stable
carbon
isotope
analysis.
Ecol.
Entomol.
29:
226­
233.

Raulston
JR,
SD
Pair,
J
Loera,
AN
Sparks,
WW
Wolf,
JK
Westbrook,
GP
Fitt
and
CE
Rogers.
1992.
Helicoverpa
zea
(
Lepidoptera:
Noctuidae)
pupa
production
in
fruiting
corn
in
northeast
Mexico
and
South
Texas.
Environ.
Entomol.
21:
1393­
1397.

Ridley,
WP,
Shillito,
RD,
Coats,
I.,
Steiner,
HY,
Shawgo,
M.,
Phillips,
A.,
Dussold,
P.
and
Kurtyka,
L.
2004.
Development
of
the
International
Life
Sciences
Institute
Crop
Composition
Database.
Journal
of
Food
Composition
and
Analysis
17:
423­
438.

Rikilt:
http://
www.
rikilt.
nl/
Publications/
Publications/
Tekstrapport2000%
20004.
htm
Saxena
D,
S
Flores
and
G
Stotzky.
1999.
Transgenic
plants
­
Insecticidal
toxin
in
root
exudates
from
Bt
corn.
Nature
402:
480­
480.
Page
78
of
85
Saxena
D
and
G
Stotzky.
2000.
Insecticidal
toxin
from
Bacillus
thuringiensis
is
released
from
roots
of
transgenic
Bt
corn
in
vitro
and
in
situ.
FEMS
Microbiol.
Ecol.
33:
35­
39.

Saxena
D
and
G
Stotzky.
2001.
Bt
corn
has
a
higher
lignin
content
than
non­
Bt
corn.
Am.
J.
Bot.
88:
1704­
1706.

Schubert,
D.
2002.
"
A
different
perspective
on
GM
food."
Nature
Biotechnology
20(
10):
969.

Sherratt
TN
and
PC
Jepson.
1993.
A
metapopulation
approach
to
modelling
the
long­
term
impact
of
pesticides
on
invertebrates.
J.
Appl.
Ecol.
30:
696­
705.

Stotzky
G.
2000.
Persistence
and
biological
activity
in
soil
of
insecticidal
proteins
from
Bacillus
thuringiensis
and
of
bacterial
DNA
bound
on
clays
and
humic
acids.
J.
Environ.
Qual.
29:
691­
705.

Stotzky
G.
2002.
Release,
persistence
and
effects
of
larvicidal
proteins
from
Bacillus
thuringiensis
in
soil.
Agronomy
Abstracts,
p.
343.
American
Society
for
Agronomy
Annual
Meeting.
Indianapolis,
IN.

Tabashnik
BE,
Carriere
Y,
Dennehy
TJ,
Morin
S,
Sisterson
MS,
Roush
RT,
Shelton
AM,
Zhao
JZ.
Insect
resistance
to
transgenic
Bt
crops:
lessons
from
the
laboratory
and
field.
J
Econ
Entomol.
2003
Aug96(
4):
1031­
8.

Tank
SL,
LW
Brewer,
GP
Cobb,
RJ
Kendall
and
MJ
Hooper.
1992.
COUNTER(
r)
5th
year
Investigation
of
the
Response
of
Selected
Wildlife
Populations
to
Planting­
Time
Application
of
COUNTER(
r)
15G
Systemic
Insecticide
Nematicide
in
an
Iowa
Corn
Field.
Submitted
to
American
Cyanamid
Company.

Tani
H,
Chen
X,
Nurmberg
P,
Grant
JJ,
SantaMaria
M,
Chini
A,
Gilroy
E,
Birch
PR,
Loake
GJ.
Activation
tagging
in
plants:
a
tool
for
gene
discovery.
Funct
Integr
Genomics.
2004
May
20
Tapp
H
and
G
Stotzky.
1995.
Insecticidal
activity
of
the
toxins
from
Bacillus­
thuringiensis
subspecies
kurstaki
and
tenebrionis
adsorbed
and
bound
on
pure
and
soil
clays.
Appl.
Environ.
Microb.
61:
1786­
1790.

Tapp
H
and
G
Stotzky.
1998.
Persistence
of
the
insecticidal
toxin
from
Bacillus
thuringiensis
subsp.
kurstaki
in
soil.
Soil
Biol.
Biochem.
30:
471­
476.

Tax
FE,
Vernon
DM.
T­
DNA­
associated
duplication/
translocations
in
Arabidopsis.
Implications
for
mutant
analysis
and
functional
genomics.
Plant
Physiol.
2001
Aug
126(
4):
1527­
38.

Tomkins
et
al.
2001.
www.
genome.
clemson.
edu/
projects/
cotton/
bac
www.
cottoninc.
com/
Agriculture/
homepage.
cfm?
page=
3157).
Page
79
of
85
Van
Frankenhuyzen
K,
Gringorten
JL.
Frass
failure
and
pupation
failure
as
quantal
measurements
of
Bacillus
thuringiensis
toxicity
to
Lepidoptera.
J.
Invertebr.
Pathol.
1991
Nov
58(
3):
465­
7.

Van
Rie
J,
S
Jansens,
H
Hsfte,
D
Degheele
and
H
Van
Mellaert.
1989.
Specificity
of
Bacillus
thuringiensis
delta­
endotoxins.
Importance
of
specific
receptors
on
the
brush
border
membrane
of
the
mid­
gut
of
target
insects.
Eur.
J.
Biochem.
186:
239­
247.

Venkateswerlu
G
and
G
Stotzky.
1992.
Binding
of
the
protoxin
and
toxin
proteins
of
Bacillusthuringiensis
subsp
kurstaki
on
clay­
minerals.
Curr.
Microbiol,
25:
225­
233.

Voet
D
and
JG
Voet,
Biochemistry
3rd
Edition,
John
Wiley
and
Sons.
2004.

Zwahlen
C,
A
Hilbeck,
R
Howald
and
W
Nentwig.
2003a.
Effects
of
transgenic
Bt
corn
litter
on
the
earthworm
Lumbricus
terrestris.
Mol.
Ecol.
12:
1077­
1086.

Zwahlen
C,
A
Hilbeck,
R
Howald
and
W
Nentwig.
2003b.
Effects
of
transgenic
Bt
corn
litter
on
the
earthworm
Lumbricus
terrestris,
12:
P.
1077.
2003.
Mol.
Ecol.
12:
2279­
2279.
Page
80
of
85
APPENDIX
A
Public
commenter
Nicholas
Storer,
Ph.
D.
(
Dow
Agrosciences)
provided
supplemental
comments
on
the
biological
interpretation
of
Z:
"
a
pool
of
Cry1Ac
activity
(
which
in
turn
is
a
function
of
concentration
and
specific
activity)."
Based
on
this
information
and
further
analysis,
Panel
member
John
Schneider
provided
additional
comments.
Such
comments
were
provided
by
Dr.
Schneider
after
the
meeting,
thus
they
do
not
reflect
the
Panel's
position.

The
formulation
of
Dow's
CBW
model
is
consistent
with
the
following
interpretation
of
Z:
Z
=
M2A+
M2B,
for
MA
 
"
mortality
due
to
binding
receptor
A
sites"
=
(
1­
S2A)
and
M2B
 
"
mortality
due
to
binding
receptor
B
sites"
=
(
1­
S2B).
In
addition,
as
Storer
points
out,
this
formulation
assumes
that
the
survival
rates
S2A
and
S2B
are
independent
and
that
the
effect
of
a
molecule
of
Cry1Ac
binding
a
receptor
A
site
is
identical
to
the
effect
of
a
molecule
binding
a
receptor
B
site.
The
characteristics
of
this
model
can
be
discerned
by
considering
the
following
results.
For
W2
=
0
(
i.
e.
0%
mortality),
at
x
=
0
(
i.
e.
no
binding
to
receptor
A
sites),
M2A
=
0
and
M2B
=
1
as
expected;
but
at
x
=
0.5
(
i.
e.
50%
of
binding
by
Cry1Ac
is
to
receptor
A
sites
and
50%
to
receptor
B
sites),
M2A
=
M2B
=
1.
In
this
case,
one
observes
that
the
binding
capacity
available
for
each
of
the
two
receptors
is
two­
fold
that
necessary
to
cause
100%
larval
mortality.
For
W2
=
0.2
(
the
value
assumed
in
the
model),
at
x
=
0,
M2A
=
0
and
M2B
=
0.8
as
expected;
but
at
x
=
0.5
(
close
to
the
best
estimate
of
0.6),
M2A
=
M2B
=
0.55.
As
in
the
first
case,
one
observes
that
excess
binding
capacity
is
present
 
although
considerably
less
than
that
observed
for
W2
=
0.
In
addition,
the
behavior
of
the
model
with
the
latter
set
of
parameter
values
is
perhaps
more
revealingly
expressed
by
noting
that
mortality
due
to
binding
at
a
given
receptor
increases
more
rapidly
per
unit
increase
in
binding
at
low
levels
of
binding
than
at
high
levels.
This
would
appear
to
be
biologically
unrealistic
given
the
possibility
of
threshold
effects
in
the
consequences
of
toxin
binding.
The
ramifications
of
the
characteristics
of
this
particular
formulation
on
the
sensitivity
analysis
of
variation
in
binding
parameter
x
presented
by
Dow
are
unclear.
Page
81
of
85
APPENDIX
B
Panel
member
Michael
Caprio
provided
additional
comments
concerning
interpretation
of
C3
and
C4
ratios
as
a
response
to
the
Bollgard
cotton
insect
resistance
management
question
on
alternate
hosts
(
question
#
3).
Such
comments
were
provided
by
Dr.
Caprio
after
the
meeting,
thus
they
do
not
reflect
the
Panel's
position.

The
data
from
the
C3/
C4
isotope
studies
have
raised
questions
since
they
suggest
that
there
is
an
excess
of
C4
individuals
over
what
one
might
expect
from
our
knowledge
of
the
distribution
of
potential
host
plants.
The
data
have
been
interpreted
to
mean
that
there
is
a
source
of
C4
individuals
that
has
not
been
identified,
whether
this
be
wild
hosts
or
southward
migration
of
CEW
from
the
corn
belt.
The
question
might
be
rephrased
to
ask
not
where
the
excess
of
C4
individuals
come
from,
but
rather
why
there
is
a
paucity
of
C3
individuals.

The
interpretation
of
the
C3/
C4
isotope
data
is
complicated
because
it
cannot
be
immediately
utilized
to
estimate
the
proportion
of
the
population
exposed
to
a
C3
transgenic
crop
(
in
other
words,
what
is
the
contribution
of
C4
plants
as
a
refuge
to
C3
transgenic
crops).
A
second,
related
question,
is
what
is
the
overall
proportion
of
C4
plants
in
the
environment?
The
difficulty
with
the
first
question
is
that
the
C3/
C4
ratios
represent
post­
selection
numbers,
that
is,
the
numbers
emerging
from
habitats
after
selection
has
occurred
in
the
transgenic
habitat.
As
such,
the
preselection
number
of
individuals,
the
number
of
individuals
actually
exposed
to
selection,
must
be
corrected
for
the
individuals
that
were
lost
in
the
transgenic
fields
as
a
result
of
selection.

As
an
example,
Gould
et
al.
(
2002)
found
on
one
date
that
the
ratio
of
C3
to
C4
adults
was
60:
40.
These
are
valid
data
and
a
good
estimate
of
the
production
of
adults
from
C3
and
C4
plants.
The
problem
is
that
selection
by
transgenic
cotton
has
already
removed
individuals
from
the
C3
pool,
so
we
cannot
directly
translate
these
numbers
into
refuge
size,
at
least
as
we
currently
use
the
term:
the
proportion
of
oviposition
on
hosts
without
selection
for
Bt
counteradaptation.
In
order
to
estimate
the
actual
proportion
of
individuals
that
were
exposed
to
selection,
we
must
ask
what
habitat
distribution,
with
the
inclusion
of
selection,
would
have
produced
the
observed
post­
selection
numbers.
This
is
most
easily
done
by
correcting
the
number
of
C3
individuals
observed
by
the
expected
number
lost
to
selection
on
the
transgenic
crop.
We
assume
for
the
sake
of
simplicity
that
the
data
above
came
from
a
trap
in
which
one
captured
60
moths
identified
as
C3
and
40
as
coming
from
C4.
We
must
take
the
C3
portion
and
replace
the
numbers
removed
by
selection.
In
order
to
do
this,
we
need
to
know
the
proportions
of
C3
transgenic
and
nontransgenic
habitats
as
provided
by
the
registrant.
If
we
assume
that
the
overall
habitat
consisted
of
95%
transgenic
habitat
and
5%
unsprayed
refuge,
we
know
that
60
=
0.05x
+
.95*
M*
x
where
x
is
the
initial
number
of
eggs
laid,
and
M
is
the
survivorship
of
bollworm
on
Bt­
cotton
Assuming
M
=
0.1,
we
can
estimate
that
x=
414
(
i.
e.,
354
C3
adults
are
missing
due
to
Page
82
of
85
selection
in
the
transgenic
cotton).
Thus,
the
actual
refuge
size
due
to
C4
plants
is
closer
to
40/(
413+
40)*
100
=
9.1%.

This
correction
is
incomplete
depending
upon
the
selection
model
used.
If
the
Gustafson
model
incorporates
sprays
of
refuges
in
the
80­
20
option,
then
these
individuals
should
be
added
back
in
by
our
calculations
because
they
will
subsequently
be
removed
by
the
model.
Similarly,
if
sprays
in
the
Bollgard
fields
are
incorporated
into
the
model,
then
those
individuals
should
be
added
back
into
the
equation
(
otherwise
they
would
be
removed
twice).
The
same
goes
for
productivity
of
alternate
hosts,
though
in
that
case
we
should
not
add
them
now
because
they
will
be
added
by
the
model
later.
All
this
suggests
that
we
should
perhaps
give
more
thought
to
what
exactly
we
mean
by
"
refuge
%"
and
how
it
must
be
calculated.
Of
course,
some
of
the
refuge
is
also
C3,
so
the
calculation
that
follows
only
estimates
the
ratio
of
C3:
C4
assuming
all
the
C3
are
unselected.

This
addresses
the
second
general
question
­
where
are
all
these
C4
hosts,
when
in
fact
many
of
the
expected
C3
individuals
are
removed
by
sprays
or
transgenic
toxins.
If
we
assume
that
the
C4
hosts
are
unsprayed,
then
the
following
calculations
will
tell
us
what
proportion
of
eggs
were
laid
on
C3
versus
C4
hosts
prior
to
any
sprays
or
mortality
due
to
transgenics.
The
other
question,
what
is
the
actual
refuge
size,
would
incorporate
both
C3
and
C4
refuges
(
easily
done)
and
depend
on
specifics
of
the
selection
model
utilized.

The
correction
may
get
more
complicated
as
additional
habitats
are
included,
and
a
more
general
solution
is
required.
For
example,
we
now
assume
that
there
are
sprayed
refuges,
unsprayed
refuges
as
well
as
transgenic
cotton.
We
must
now
add
in
a
correction
factor
to
account
for
the
fact
that
the
production
of
moths
from
the
sprayed
refuges
is
less
than
the
production
from
the
unsprayed
refuges.
This
can
probably
be
estimated
reasonably
accurately
from
the
registrant's
late
larval
sampling,
assuming
that
there
are
no
large
differences
between
habitats
in
mortality
after
this
sample
date.
We
can
standardize
on
unsprayed
cotton
as
a
relative
one
(
i.
e.,
production
in
all
other
habitats
is
expressed
as
relative
to
the
number
produced
per
area
of
unsprayed
cotton).
This
factor
was
implicitly
included
above,
but
since
Bt
is
unsprayed
(
in
our
imaginary
case),
the
factor
in
both
cases
was
1.
In
the
case
of
sprayed
cotton,
sprays
might
reduce
production
of
adults
by
perhaps
a
specific
mortality
rate
(
assuming
80%
here).
This
could
also
be
directly
estimated
from
the
Monsanto
larval
data
if
data
exist
on
untreated
refuges,
or
less
preferably
by
estimation
from
the
number
that
one
would
expect
to
emerge
from
unsprayed
Bollgard
plots
(
sprayed
plots)/(
M*
observed
BG
larvae).
We
can
call
this
factor
Pi
(
productivity
of
habitat
i
relative
to
unsprayed
cotton).
We
can
now
write
a
more
generalized
equation:
Page
83
of
85
nhabitats
C3
=
 
Si
Mi
Pi
x
1
Si
is
proportion
of
the
ith
habitat
of
the
total
C3
producing
habitat
Mi
is
the
survivorship
in
the
transgenic
crop
Pi
is
the
productivity
of
the
ith
habitat
relative
to
unsprayed
non­
Bt
cotton
x
is
the
corrected
C3
component
C3
is
the
observed
number
of
C3
moths
Let
us
assume
now
that
the
habitat
producing
C3
individuals
can
be
assigned
into
90%
of
the
area
that
uses
the
80­
20
option
(
sprayed
refuge),
while
the
remaining
cotton
acreage
uses
a
95­
5
option
(
unsprayed
refuge).
We
can
calculate
the
total
portion
of
each
habitat
as
follows
Bollgard
Unsprayed
refuge
Sprayed
refuge
95
5
0
x
10%
80
0
20
x
90%

or
9.5
.5
72.0
0
18
81.5
0.5
18
=
100%

These
would
be
the
Si
in
the
equation
above
(
once
divided
by
100
to
represent
proportions),
and
we
can
now
expand
our
equation
60
=
.815*
.1
*
1
*
x
[
Bollgard
component]
+
.005
*
1
*
1*
x
[
unsprayed
refuge
option]
+
.18
*
1
*
0.2
*
x
[
sprayed
refuge
component]

60
=
.1225x
or
x
=
489.8
and
the
initial
C4
contribution
would
be
estimated
as
40/(
489+
40)
=
7.4%

This
is
the
proportion
of
eggs
oviposited
on
C4
hosts.
It
is
not
an
estimate
of
refuge
size
which
would
include
oviposition
on
non­
transgenic
C3
hosts
as
well.

We
could
include
highly
productive
refuges
(
as
an
example,
chickpea
produces
many
more
larvae/
unit
area),
and
in
such
cases
the
Pi
factor
might
exceed
1.
We
can
also
accommodate
sprayed
Bt­
cotton
by
changing
Pi
for
the
Bt
cotton
by
an
appropriate
amount.
If
Bt
cotton
is
sprayed
an
average
of
once
per
generation,
we
could
set
P1
in
the
previous
equation
to
0.2.
This
would
account
for
the
fact
that
some
of
the
survivors
on
the
Bt­
cotton
were
then
removed
by
Page
84
of
85
pyrethroid
sprays.
This
value
may
underestimate
the
number
produced,
as
multiple
sprays
are
used
to
reduce
populations
in
conventional
cotton,
but
only
one
is
used
in
Bt­
cotton.
Of
course,
data
exist
that
the
pyrethroid
sprays
are
more
effective
in
Bt­
cotton.
The
result
would
change
to:

60
=
.815*
.1
*
0.2
*
x
[
bollgard
component]
+
.005
*
1
*
1*
x
[
unsprayed
refuge
option]
+
.18
*
1
*
0.2
*
x
[
sprayed
refuge
component]
.
60
=
.0736x
or
x
=
815.2
and
the
initial
C4
contribution
would
be
estimated
as
40/(
815.2+
40)
=
4.7%

We
should
be
able
to
work
forward
from
a
specified
configuration
of
fields
to
generate
a
C3:
C4
ratio,
and
then
using
that
figure,
the
known
field
distributions,
and
productivity,
to
work
backwards
and
regenerate
the
initial
field
distribution.
We
assume
that
60%
of
the
farms
in
an
area
use
the
80­
20
option,
38%
use
the
95­
5
option,
and
2%
of
the
area
is
planted
to
chickpeas
(
to
demonstrate
incorporating
of
highly
productive
hosts).
Assuming
chickpeas
provide
a
3­
fold
greater
production
factor,
it
would
be
easiest
if
we
assume
an
arbitrary
number
of
moths
(
say
100)
are
produced
in
unsprayed
cotton
(
this
is
actually
the
number
of
C3
moths
we
would
catch
if
all
the
C3
producing
habitats
were
unsprayed
cotton).
We
can
now
calculate
the
number
of
moths
we
would
collect
in
our
traps,
assuming
trap
efficiency
is
density
independent
.
A
question
remains
whether
this
matters
as
C4
capture
would
also
be
reduced.
Thus
100
*
0.6
*
0.8
*
0.1
*
1
=
4.8
unsprayed
Bt
cotton
in
80­
20
100
*
0.6
*
0.2
*
1
*
0.2
=
2.4
sprayed
refuge
in
80­
20
100
*
0.38
*
0.95
*
0.1
*
1
=
3.6
unsprayed
Bt
cotton
in
95­
5
100
*
0.38
*
0.05
*
1
*
1
=
1.9
unsprayed
refuge
in
95­
5
100
*
0.02
*
1
*
1
*
3
=
6.0
area
planted
to
chickpeas.

This
is
a
good
example
of
where
moths
come
from
in
a
low
dose
situation.
The
refuges
only
account
for
slightly
more
than
half
of
the
adults,
but
they
do
it
on
16%
of
the
area.

The
predicted
total
C3
moths
sampled
would
then
be
X=
18.7.

To
work
backwards,
we
must
know
X,
the
Si
(
which
is
the
product
of
columns
2
and
3),
the
Mi
(
column
4)
and
Pi
(
the
relative
productivity
given
in
column
5).
Estimates
of
all
these
parameters
could
be
derived
from
the
data
collected
by
the
registrant.

We
then
apply
equation
1
from
above:
X
=
0.6*
0.8*
0.1*
x+
0.6*
0.2*
0.2*
x+
0.38*
0.95*
0.1*
x+
0.38*
0.05*
x+
0.02*
3*
x
18.7
=
0.1851x
x=
101
a
reasonable
estimate
given
rounding
error.
Page
85
of
85
Again,
this
is
not
an
estimate
of
refuge
size,
but
it
does
give
an
estimate
of
the
initial
proportion
of
eggs
that
were
laid
on
C3
crops.
If
our
selection
model
does
not
account
for
sprays
in
the
20%
refuges
nor
the
oversprays,
then
it
is
correct
to
add
these
numbers
back
into
the
C3
number.
If
the
model
does
not
include
differences
in
productivity,
those
individuals
should
be
accounted
for
here.
It
would
be
more
interesting
to
isolate
the
effects
of
the
selection
for
Bt
resistance
from
all
these
other
components,
based
on
an
interest
in
refuge
size.
If
you
want
to
account
for
all
these
C4
individuals,
then
the
complete
calculation
is
most
appropriate
(
and
would
make
your
job
easier).

The
question
of
most
general
biological
importance
would
be
to
estimate
the
actual
proportion
of
C3
and
C4
hosts,
and
this
would
require
adjustments
of
all
C3
hosts,
as
well
as
any
C4
hosts
that
might
be
contributing
and
have
their
productivity
impaired
by
anthropogenic
means
(
Bt­
corn,
perhaps
sorghum).
Although
Gould
et
al.
(
2001)
suggested
that
whatever
the
source,
something
must
be
producing
many
C4
moths,
it
may
be
instead
that
something
is
removing
C3
moths.
The
actual
ratio
of
eggs
laid
on
the
two
types
of
hosts
may
be
quite
different
than
the
ratio
of
adults
produced,
perhaps
significantly
so.
In
this
specific
case,
the
ratio
might
be
less
than
the
observed
40%,
perhaps
as
little
as
4­
5%.
Given
that
the
registrant
data
indicates
observed
ratios
as
low
as
10%,
the
corrected
ratios
could
be
as
low
as
1­
2%,
and
it
is
this
number
which
should
be
incorporated
into
most
models.
One
might
then
begin
to
seriously
suggest
that
wild
hosts
or
corn
regrowth
could
account
for
a
significant
portion
of
the
C4
moths,
and
it
might
also
make
us
rethink
the
importance
of
North­
South
migration.
It
should
be
the
job
of
the
selection
model
to
properly
account
for
points
such
as
sprays
and
productivity
differences
between
crops
(
as
the
both
the
Gustafson
and
Storer
models
attempt
to).
Of
course,
as
long
as
the
data
is
there,
it
would
make
sense
to
take
the
data
for
all
the
crops
as
is,
adjust
the
Bt­
cotton
data
as
we
are
interested
in
selection
for
resistance
to
it,
and
then
properly
allocate
all
the
habitats,
both
C3
and
C4,
to
refuge
that
are
not
selected.
This
would
give
us
a
reasonably
good
estimate
of
the
actual
refuge
size,
rather
than
just
mentioning
C4
hosts
as
a
sort
of
minimal
refuge
size.
Such
an
approach
would
better
utilize
the
registrant
data
and
may
be
workable
since
the
data
had
already
been
collected
by
the
registrant.
