UNITED
STATES
ENVIRONMENTAL
PROTECTION
AGENCY
WASHINGTON
D.
C.,
20460
OFFICE
OF
PREVENTION,
PESTICIDES
AND
TOXIC
SUBSTANCES
MEMORANDUM
DATE:
March
29,
2006
DP
Barcode:
327264
PC
Code:
128821
SUBJECT:
EFED
responses
to
Imazapyr
Phase
3
Comments
FROM:
Stephen
Carey,
Biologist
Lucy
Shanaman,
Chemist
Pamela
Hurley,
Toxicologist
Environmental
Fate
and
Effects
Division
Environmental
Risk
Branch
III
(
7507C)

THRU:
Daniel
Rieder,
Branch
Chief
Environmental
Fate
and
Effects
Division
Environmental
Risk
Branch
III
(
7507C)

TO:
Eric
Olson,
Team
Reviewer
Special
Review
Registration
Division
The
Environmental
Fate
and
Effects
Division
(
EFED)
has
completed
its
review
of
the
Imazapyr
Phase
3
Comments.
In
this
document,
EFED
provides
its
response
to
comments
from
BASF,
California
Indian
Basketweavers
Association,
&
Nebraska
Department
of
Agriculture.

1)
Response
to
BASF
comments
BASF
is
concerned
about
the
procedures
that
EPA
has
used
in
performing
the
assessment
for
nontarget
terrestrial
and
semi­
aquatic
plants.
Specifically,
EPA
has
used
BANKS
(
1988)
MRID
No.
40811801
for
the
effects
endpoints.
EPA
considers
the
study
supplemental
or
invalid
(
Table
IIIC­
14,
p.
56).
Elsewhere
in
the
text
(
p.
10)
,
EPA
refers
to
the
study
as
supplemental
and
highlights
that
the
plants
were
stressed
and
that
wet
weight
was
measured
instead
of
dry
weight.

EPA
response:
The
study
was
categorized
as
supplemental
due
to
stress
and
EPA
test
guidelines
require
dry
weight
and
height
measurements.
However,
the
results
were
considered
useful
for
risk
assessment
despite
the
deficiencies.

BASF
comment:
There
are
several
other
points
that
should
be
mentioned
regarding
the
study.
First,
the
study
was
done
with
technical
a.
i.
diluted
in
acetone.
EPA
typically
requires
that
nontarget
plant
tests
be
done
with
a
representative
formulation.
BASF
points
out
the
Feutz
and
Canez
study
(
MRID
No.
43889101)
was
done
with
the
2ASU
formulation.

EPA
response:
The
Agency
determined
that
the
Banks
study
(
MRID
40811801)
was
tested
with
TEP,
which
included
a
surfactant,
Tween
20,
and
the
Feutz
and
Canez
study
(
MRID
43889101)
was
tested
without
any
surfactant(
s).
The
Banks
study
was
tested
with
an
adjuvant
thus
making
it
more
representative
of
a
TEP
than
the
Feutz
and
Canez
study
with
salt
alone.
Another
reason
the
Agency
did
not
use
the
Feutz
and
Canez
study
was
that
it
was
only
conducted
with
3
species
as
opposed
to
the
required
10
species.

BASF
comment:
Second,
the
report
calculates
EC25
and
EC50
values
using
simple
linear
regression.
It
appears
that
EPA
has
calculated
the
EC05
values
used
in
the
assessment.
However,
there
are
no
details
provided
as
to
the
model
used
or
the
fit
of
that
model
to
the
data.
BASF
is
unaware
of
a
DER
for
the
Banks
study.

EPA
response:
Based
on
the
October
21,
1994
memorandum;
the
lack
of
a
No
Observed
Adverse
Effect
Concentration
(
NOAEC)
will
not
invalidate
a
terrestrial
plant
study
as
long
as
the
slope
is
adequate
for
calculation
of
valid
EC50,
EC25,
and
EC05
values
from
the
most
sensitive
endpoints.
Studies
that
lack
a
NOAEC
but
meet
the
above
criteria
will
be
classified
supplemental.

However,
most
of
the
invalid
NOAECs
were
calculated
to
be
above
the
EC25,
the
EC05
was
obtained
from
the
Bruce
and
Versteeg
method.
Therefore,
the
Agency
calculated
EC05s
with
which
to
assess
potential
effects
to
endangered
species.

Statistical
Method:
Data
for
plant
length,
and
shoot
weight
for
the
seedling
emergence
and
vegetative
vigor
tests
were
analyzed
to
determine
the
NOAEC,
EC05,
EC25,
and
slope
values.
Continuous
data
(
height
and
weight)
were
assessed
for
normality
and
homogeneity
of
variance
prior
to
all
analyses.
Treatment
effects
were
assessed
using
Williams'
tests.
The
ECx
estimates
were
done
using
the
Bruce
and
Versteeg
method
via
Nuthatch
software.
The
Seedling
Emergence
and
Vegetative
Vigor
DERs
for
the
Banks
and
Feutz/
Canez
studies
are
attached
2)
Response
to
comments
on
Imazapyr
from
the
California
Indian
Basketweavers
Association
Issue
1
pertained
to
tolerance
setting,
EFED
is
only
responding
to
issues
2­
7.

Response
to
Issue
2:
The
information
available
to
the
Agency
on
usage
of
imazapyr
on
forests
indicated
it
would
only
be
applied
1­
2
times
in
multiple
years,
ie
10
years,
or
20­
30
years.
However,
our
exposure
and
risk
modeling
is
not
affected
(
reduced)
by
that
assumption.
We
modeled
a
single
application
and
drew
conclusions
of
risk
based
on
that
application.
For
risk
to
terrestrial
plants,
repeat
applications
in
different
parts
of
a
watershed
would
not
result
in
higher,
local
exposure
or
risk
because
the
terrestrial
plant
scenario
is
at
the
field
level.
Other
treatments
in
other
fields,
whether
in
the
same
watershed
or
not,
would
still
be
represented
at
the
field
level
by
the
model.
For
aquatic
risk,
a
single
application
resulted
in
a
conclusion
of
significant
risk
for
plants.
The
scenario
modeled
was
actually
a
pond,
with
no
inflow
or
outflow,
so
it
would
not
take
into
account
the
dilution
that
would
occur
in
a
watershed
stream.
The
scenario
modeled
also
assumed
the
entire
watershed
was
treated
at
one
time.
So
the
multiple
applications
that
might
be
possible
in
different
areas
of
a
given
watershed
would
be
unlikely
to
result
in
higher
exposures
than
calculated
using
the
tier
one
surface
exposure
model
GENEEC2.

Response
to
Issue
2
cont'd:
The
application
rates
were
those
available
to
Agency
based
on
label
analysis.
Applications
at
rates
higher
than
the
label
were
not
assessed.
The
model
used
by
the
Agency
at
this
screening
level
is
not
capable
of
modeling
multiple
applications
to
different
parts
of
a
given
watershed.
See
the
comment
above
discussing
the
nature
of
the
scenario
modeled
by
the
Agency
that
assumes
100%
of
a
watershed
is
treated
simultaneously.
The
EECs
presented
are
considered
to
be
representative
of
the
exposure
from
forestry
use
under
conditions
that
maximize
exposure.
Response
to
Issue
3:
The
Agency
risk
assessment
did
not
model
the
exposure
from
forestry
use
or
any
other
use
pattern
in
any
way
assuming
setbacks
or
buffers
from
waterbodies.
Our
model
assumed
treatment
occurred
adjacent
to
the
waterbody
in
which
EECs
were
estimated.

Response
to
Issue
3
cont'd:
In
response
to
the
concern
for
downstream
indirect
effects,
this
is
taken
into
account
in
the
Agency
risk
assessment
where
it
discusses
indirect
affects,
it
mentions
effects
due
to
loss
of
habitat.
The
outcome
from
loss
of
habitat
would
be
potential
loss
of
vegetation
and
possibly
increased
sedimentation.
At
this
stage
of
the
Agency's
assessment
for
endangered
species,
the
specific
possible
impacts
from
indirect
effects
have
not
been
fully
explored,
and
remain
a
possibility.

Response
to
Issue
3
cont'd:
The
degree
to
which
the
100
yard
buffer
would
reduce
risk
is
uncertain
in
that
exposure
is
expected
to
occur
both
from
drift
and
runoff.
Buffers
that
are
effective
to
reduce
spray
drift
may
not
be
effective
to
reduce
runoff.
However,
it
is
assumed,
as
a
general
principle,
that
any
setback
from
water
bodies
would
reduce
the
probability
and
possibly
the
level
of
exposure
in
that
waterbody.

As
noted
in
the
science
chapter
discussion
of
this
incident,
because
of
the
mode
of
action
of
imazapyr,
and
the
way
it
affects
the
ALS
enzyme,
which
is
only
found
in
plants,
it
is
considered
unlikely
that
imazapyr
itself
directly
killed
birds
and
fish.
It
is
likely
this
incident
reflects
consequences
of
exposure
to
other
pesticides
also
applied.
However,
this
does
not
diminish
the
possibility
of
adverse
effects
to
plant
communities
that
might
adversely
affect
animals
living
in,
or
depending
on
these
plants.
Response
to
Issue
4:
This
comment
is
reflected
in
the
science
chapter's
risk
conclusions.
The
Agency
is
not
aware
that
imazapyr
is
used
in
the
practice
of
converting
native
forest
to
tree
farms.
However,
labeled
uses
were
modeled,
and
the
exposure
from
such
a
treatment
would
not
be
different
than
what
was
modeled
in
the
Agency's
risk
assessment
assuming
it
is
used
according
to
the
label.

Response
to
Issue
5:
The
information
obtained
by
the
Agency
either
indicated
single
applications
of
imazapyr,
or
did
not
specify
the
number
of
applications.
Since
even
with
a
single
application,
risk
was
concluded
for
aquatic
plants
from
the
roadside
or
rights
of
way
use.
If
multiple
applications
were
permitted,
the
risk
numbers
might
be
higher,
but
the
pattern
of
LOC
exceedances
would
not
change.
Multiple
applications
may
also
extend
the
duration
of
risk;
however,
the
model
would
not
show
the
effect
of
this
increased
duration.
It
is
also
important
to
note
that
the
modeling
used
to
assess
the
roadside
use
pattern
assumed
a
large
contiguous
treated
area
that
would
take
into
account
repeated
sequential
applications.

Response
to
Issue
6:
As
indicated
earlier,
assessing
risk
to
terrestrial
organisms
from
multiple
applications
in
different
parts
of
a
watershed
would
not
result
in
higher
exposure
to
local
habitats
since
exposure
to
terrestrial
plants
is
not
linked
to
a
watershed
scenario.
As
for
assessing
risk
to
plants
from
formulations,
currently
the
Agency
assesses
risk
based
on
toxicity
from
formulations
or
of
mixtures
of
ai
with
adjuvants
to
account
for
the
influence
of
the
formulations.
So
the
potential
effects
of
formulations
in
the
environment
is
accounted
for
in
the
toxicity
data.
Exposure
to
plants
is
expressed
in
ai
or
ae,
but
is
estimated
assuming
a
formulation
was
applied.
As
for
assessing
risk
from
mixtures,
the
Agency
does
not
currently
have
a
process
to
assess
risk
from
and
regulate
based
on
mixtures.

Response
to
Issue
7:
The
comment
presents
statements
from
the
Agency
risk
assessment.
It
should
be
clarified
that
the
phrase
"
It
is
accepted"
does
not
connotate
that
the
risk
represented
by
the
toxicity
is
"
accepted"
from
a
regulatory
standpoint.
The
phrase
means
the
information
is
"
accepted"
from
a
scientific
standpoint.

3)
Comments
from
Pesticide
Program,
Nebraska
Dept
of
Agriculture
The
Agency
acknowledges
this
error,
and
has
modified
the
endangered
species
list,
Appendix
H.
4)
EFED
deferred
response
to
error­
only
comments
concerning
aquatic
exposure
estimations
(
pp.
4­
6)

BASF
responded
that
the
Agency
chose
to
use
GENEEC
as
the
aquatic
exposure
estimation
model
when
a
suitable
Forestry/
Non­
crop
Tier
II
scenario
and
a
corn
scenarios
exists,
and
then
requested
that
a
Tier
II
exposure
assessment
be
conducted.

EFED
Response:
The
corn
scenario
used
by
BASF
is
only
one
of
10
standard
scenarios
available
for
Tier
II
aquatic
modeling,
and
would
not
be
representative
of
all
use
sites
nationally.
Additionally,
the
application
rate
for
use
of
Imazapyr
on
corn
is
much
lower
than
labeled
use
rates
for
non­
food
uses
that
there
is
little
benefit
in
investing
the
degree
of
effort
necessary
to
conduct
a
Tier
II
evaluation
for
corn.

The
forestry
scenario
cited
by
BASF
as
a
suitable
non­
crop
scenario
was
developed
for
the
OP
cumulative
assessment,
and
was
not
intended
to
be
representative
enough
of
typical
use
sites
to
be
used
in
a
national
assessment.
An
informal
range
finding
experiment
was
conducted
by
running
the
Agency
Oregon
Christmas
tree
scenario
using
weather
data
files
for
other
geographic
locations.
Varying
only
the
meteorological
data
resulted
in
a
great
deal
of
variation
in
estimated
concentrations.
These
results
indicated
that
a
suitable
Forestry/
Non­
crop
scenario
is
not
currently
available,
and
new
scenarios
would
need
to
be
developed
in
order
to
conduct
a
Tier
II
aquatic
assessment
for
Imazapyr.

BASF
then
requested
that
a
"
correction
factor"
be
applied
to
the
estimated
concentrations,
and
cited
an
article
by
Jackon
et
al
as
justification
for
applying
the
factor.

EFED
Response:
In
a
prepublication
peer
review
for
the
Journal
of
Agricultural
and
Food
Chemistry,
Elizabeth
Behl
and
Nelson
Thurman
submitted
comments
reflecting
Agency
analysis
of
an
article
titled
Comparison
of
Regulatory
Method
Estimated
Drinking
Water
Exposure
Concentrations
with
Monitoring
Results
from
Surface
Water
Supplies
written
by
Scott
Jackson,
BASF
Corporation
et
al.
A
summary
of
those
comments,
outlining
Agency
concerns
about
suggestions
regarding
the
application
of
a
correction
factor
for
modeled
estimated
surface
water
concentrations
proposed
in
that
article
appears
below.

The
basis
of
the
comparison
made
in
this
article
is
fundamentally
flawed.
The
endpoints
derived
from
Agency
modeling
and
the
reported
monitoring
results
are
wholly
different.
For
the
conclusions
presented
in
this
article
to
be
valid,
the
authors
would
have
needed
to
demonstrate
that
the
monitoring
data
used
for
the
comparisons
represented
the
full
range
of
use
conditions
for
each
of
the
pesticides
used
in
the
comparison.
The
paper
did
not
provide
adequate
detail
about
(
1)
the
differences
in
the
endpoints
for
modeling
and
monitoring,
(
2)
the
assumptions
and
inputs
used
in
modeling,
or
(
3)
the
design
of
the
monitoring
study.
It
did
not
describe
the
nature
of
the
EPA
models,
or
how
these
models
are
used.
These
details
are
important
in
interpretation
of
results,
in
supporting
both
the
author's
recommendations
to
develop
adjustment
factors,
and
the
author's
judgment
on
the
validity
of
the
conceptual
model.
Additionally,
the
author's
description
of
how
EPA's
endpoints
are
derived
was
inadequate.
Application
rates,
application
frequencies,
and
percentage
cropped
area
adjustment
factors
used
in
modeling
should
have
closely
represented
actual
pesticide
use
and
usage
in
each
of
the
12
watersheds
monitored.
This
could
not
be
determined
because
the
authors
had
not
provided
the
information
necessary
to
interpret
monitoring
study
results.
The
study
authors
failed
to
compare
a
representative
monitoring
dataset
to
the
screening
model
estimates,
did
not
design
the
monitoring
to
capture
a
maximum
concentration
from
a
large
rainfall
event,
and
did
not
provide
details
of
pesticide
usage
in
the
watershed.
As
a
result,
statistical
analysis
of
the
data
cannot
provide
meaningful
results.
Given
the
flaws
in
the
overall
methodology
used
for
comparison,
the
adjustment
factor
development
has
no
basis,
and
conclusions
about
the
adequacy
of
EPA's
conceptual
model
for
water
modeling
is
not
supported
by
the
published
analysis.

In
summary,
the
authors
of
the
report
merely
ran
the
Agency
aquatic
screening
models
for
every
chemical
included
on
USGS's
analytical
list,
whether
the
pesticides
were
actually
used
in
any
of
the
watersheds
or
not.
Then
they
compared
those
screening
level
estimates,
which
are
intended
to
represent
a
1­
in­
10­
year
concentration
in
an
intensive
use
area,
to
whatever
concentrations
where
reported
in
an
18­
month
study
covering
12
reservoirs.
These
12
reservoirs
weren't
necessarily
representative
of
high
pesticide
use
areas
(
it
definitely
underrepresented
midwestern
reservoirs,
which
are
generally
more
vulnerable
because
of
the
intensity
of
agriculture),
and
were
certainly
not
representative
of
particularly
runoff­
prone
rainfall
years.
Indeed,
many
of
the
reservoirs
were
sampled
during
drought
periods
(
from
the
USGS
report).
From
that
analysis,
they
derived
a
"
correction"
factor
that,
at
best,
is
only
applicable
to
those
12
reservoirs
during
lower­
than
normal
rainfall
years.
That's
not
particularly
conducive
to
leading
to
a
reasonable
certainty
of
no
harm.
