UNITED
STATES
ENVIRONMENTAL
PROTECTION
AGENCY
WASHINGTON,
D.
C.
20460
OFFICE
OF
PREVENTION,
PESTICIDES
AND
TOXIC
SUBSTANCES
December17,
2002
MEMORANDUM
SUBJECT:
EFED
response
to
the
RRTF's
errors­
only
comments
on
the
Agency
document
"
Comparative
Risks
of
Nine
Rodenticides
to
Birds
and
Nontarget
Mammals"

TO:
John
Pates,
Chemical
Review
Manager
Susan
Lewis,
Branch
Chief
FROM:
William
Erickson,
Biologist
Douglas
Urban,
Senior
Biologist
Environmental
Risk
Branch
III,
Environmental
Fate
and
Effects
Division
THRU:
Stephanie
Irene,
Acting
Chief
Environmental
Risk
Branch
III,
Environmental
Fate
and
Effects
Division
The
Environmental
Fate
and
Effects
Division
(
EFED)
has
reviewed
the
Rodenticide
Registrants
Task
Force's
(
RRTF)
"
errors­
only"
response
to
the
Agency
document
"
Comparative
Risks
of
Nine
Rodenticides
to
Birds
and
Nontarget
Mammals"
dated
October
3,
2001.
Their
comments
of
December
10,
2001
were
prepared
by
J.
F.
Hobson,
MorningStar
Consulting,
on
behalf
of
the
RRTF.
As
stated
in
the
Agency's
October
23,
2001
cover
letter
for
the
assessment,
the
registrants'
30­
day
response
should
address
only
mathematical,
computational,
typographic,
or
other
similar
errors.
Matters
of
policy,
interpretation,
or
applicability
of
data
will
be
addressed
after
the
public
comment
period
in
accordance
with
the
Agency's
reregistration
process
for
pesticides.

In
response
to
error
comments
by
the
RRTF
and
rodenticide
registrants,
EFED
has
made
necessary
computational
and/
or
typographical
corrections.
However,
EFED
notes
that
many
comments
relate
to
policy,
interpretation,
or
applicability
of
data,
and
those
comments
will
be
addressed
along
with
public
comments
after
the
60­
day
public­
comment
period.
Page
RRTF
comment
1
See
December
8­
9,
1998
http://
www.
epa.
gov/
scipoly/
sap/
1998/
index.
htm
2
i
Hazard,
not
risk.
In
the
Executive
Summary,
the
authors
of
the
Comparative
Risks
of
Nine
Rodenticides
to
Birds
and
Non­
target
Mammals
(
PRA)
state
the
risks
from
brodifacoum
and
bromadiolone
are
high
for
mammalian
predators
and
scavengers
that
feed
on
poisoned
target
species
based
on
laboratory
secondary
hazard
studies
and
field
data.
The
relationship
of
these
hazard
studies
to
the
potential
for
exposure,
and
thus
risk,
to
these
mammals
from
commensal
uses
has
not
been
characterized;
therefore,
this
assessment
cannot
be
called
a
"
risk"
assessment
and
it
is
inappropriate
to
say
that
the
"
risk"
is
"
high"...
Presentation
of
laboratory
secondary
toxicity
studies.
The
presentation
of
secondary
toxicity
in
the
laboratory
is
misleading
and
reflects
a
poor
understanding
of
the
concepts
of
hazard
versus
risk.
These
are
actually
a
type
of
dose­
response
studies
and
how
many
animals
die
is
related
to
the
dose
selection
and
not
necessarily
to
the
risk
of
the
compound.
Exposure
in
these
lab
studies
is
often
not
the
same
as
(
or
sometimes
even
close
to)
exposure
under
actual
field
conditions.
Furthermore,
the
protocols
and
test
conditions
(
e.
g.,
target
and
non­
target
species,
number
of
animals,
period
of
feeding)
used
for
these
studies
often
differed
significantly,
therefore
it
is
not
appropriate
to
compare
their
results
as
if
they
were
the
same
as
acute
LD50
studies
or
other
"
standard"
guideline
studies.

EFED
response:
This
has
been
addressed
in
the
revised
document.
As
the
RRTF
knows,
rodenticide
baits
are
formulated
to
be
lethal
to
rodents
and
a
few
other
small
mammals,
and
they
are
not
selective
to
the
target
species.
Although
many
factors
influence
which
nontarget
animals
might
be
exposed
to
baits,
many
nontarget
organisms
are
attracted
to
and
consume
grain­
based
baits.
Predators
and
scavengers
also
feed
on
rats
and
mice
or
other
target
species,
and
they
are
not
likely
to
avoid
feeding
on
those
that
have
eaten
rodenticide
bait.
Thus,
rodenticide
baits
also
pose
potential
secondary
risks.
EFED
believes
that
the
potential
for
risks
to
birds
and
nontarget
mammals
is
well
established
for
some
of
these
rodenticides.

The
risk
assessment
is
based
on
the
available
data.
Registrants
have
not
submitted
the
data
that
would
be
needed
to
assess
the
probability
of
exposure.
These
data
have
been
outlined
in
a
section
on
Uncertainty
and
Data
Needs
in
the
revised
assessment.
The
methodology
used
is
similar
to
that
used
in
the
Agency's
"
Comparative
Analysis
of
Acute
Risk
From
Granular
Pesticides"
(
EPA
1992)
and
"
A
Comparative
Analysis
of
Ecological
Risks
from
Pesticides
and
Their
Use:
Background,
Methodology,
Case
Study"
(
EPA
1998)
1;
both
were
reviewed
by
a
FIFRA
Scientific
Review
Panel.
Concerning
the
latter
analysis,
the
Panel
noted
the
many
scientific
uncertainties
in
the
method,
yet
agreed
that
it
was
a
useful
screening
tool
that
provides
a
rough
estimate
of
relative
risk.
The
Page
RRTF
comment
2
See
Guidelines
for
Ecological
Risk
Assessment
(
EPA/
630/
R­
95/
002F,
1998)
at
http://
cfpub.
epa.
gov/
ncea/
cfm/
recordisplay.
cfm?
deid=
12460
3
Panel
made
a
number
of
helpful
suggestions
to
improve
the
utility
of
the
method,
most
of
which
are
included
here.

Risk
conclusions
are
presented
in
tabular
and
graphical
form
based
on
two
analyses
of
the
available
data.
The
first
is
a
comparative
ranking
of
the
potential
risk
based
on
a
comparative­
analysis
model,
and
the
second
is
a
tabular
comparative
rating
of
potential
risk
based
on
a
qualitative
"
weight­
of­
evidence"
assessment.
Quantitative
estimates
of
risk
are
used
in
both;
however,
the
"
weight­
of
evidence"
assessment
includes
qualitative
assessments
of
secondary
risk
based
on
mortality
and
other
adverse
effects
reported
in
laboratory
and
field
studies,
operational
control
programs,
and
incident
reports,
as
well
as
toxicokinetic
data
and
residue
levels
reported
in
primary
consumers.
This
approach
is
in
concert
with
EPA's
risk­
assessment
guidelines2,
where
professional
judgement
or
other
qualitative
evaluation
techniques
may
be
used
to
rank
risks
using
categories
such
as
low,
medium,
and
high
when
exposure
and
effects
data
are
limited
or
are
not
easily
expressed
in
quantitative
terms.

i
Dietary
data
are
available
for
mammals
for
bromadiolone
and
should
be
referenced.

EFED
response:
Dietary
data
are
not
required
for
mammals,
and
none
are
present
for
bromadiolone
in
EFED's
toxicity
database
or
the
EPA/
OPP
Health
Effects
Division's
toxicity
database.
Statements
such
as
"
are
available"
can't
be
addressed
unless
a
citation
is
provided,
such
as
an
EPA
MRID
number
for
the
study.

ii
Incident
data.
The
discussion
of
rodenticide
wildlife
incidents
misrepresents
the
data
and
does
not
take
into
consideration
the
RRTF
review,
which
noted
that
approximately
one­
third
of
incidents
were
redundant
within
the
Environmental
Incident
Inventory
System
(
EIIS)
and
many
others
are
incorrectly
attributed
to
anticoagulants.
Summary
numbers
overstate
the
number
of
incidents
and
in
a
"
weight
of
the
evidence"
argument
the
number
of
incidents
should
not
be
overstated.
The
authors
must
adjust
the
numbers
appropriately.

EFED
response:
The
RRTF
has
not
identified
a
single
redundant
incident
in
the
risk
assessment.
Stating
that
there
are
redundancies
in
the
EIIS
is
misleading,
because
the
EIIS
is
a
database,
and
not
everything
listed
in
the
database
is
cited
in
the
assessment.
The
RRTF
should
address
the
incident
data
presented
in
the
assessment,
not
that
in
the
database.
Page
RRTF
comment
4
ii
Kit
fox
mortalities.
This
parenthetical
reference
is
misleading,
speculative,
and
inappropriate
in
the
Executive
Summary.
The
9
Kit
foxes
were
reported
as
mortalities.
Of
these,
7
mortalities
were
attributed
to
vehicular
impact,
and
2
were
attributed
to
unknown
causes
(
not
anticoagulants).
While
low­
level
(
trace)
residues
were
found
in
8
of
9
Kit
foxes,
attributing
these
mortalities
to
brodifacoum
is
unfounded
and
speculative
and
these
statements
must
be
removed
from
the
PRA.

EFED
response:
This
statement
has
been
removed
from
the
Executive
Summary.
However,
it
is
not
misleading
or
speculative
to
state
that
brodifacoum
residue
was
detected
in
the
liver
of
nine
kit
fox
carcasses;
that
is
fact,
not
speculation.

ii,
19,
150
Presence
of
liver
residues.
To
state
that
brodifacoum,
or
other
anticoagulants,
have
been
"
implicated"
or
"
involved"
in
rodenticide
incidents
does
not
"
affirm
causality"
(
Ecological
Risk
Assessment
(
ERA)
Guidelines
at
86).
Liver
residues
are
a
biomarker
of
exposure.
In
the
initial
phase
of
elimination,
liver
residues
are
independent
of
the
magnitude
of
exposure
and
a
poor
correlate
with
toxicity,
although
they
are
persistent.
Persistent
low­
level
residues
of
brodifacoum,
and
bromadiolone,
have
been
observed
by
the
State
of
California
in
numerous
feral
coyotes
and
other
animals
in
perfect
health.
This
fact
alone
supports
an
inconsistency
in
association
and
a
"
basis
for
rejecting
causality"
(
ERA
Guidelines
at
86).
The
statement
should
say,
"
residues
of
brodifacoum,
or
other
anticoagulants,
have
been
detected
in
wildlife
incidents."
The
role
of
low­
level
residues
(
the
case
for
the
majority
of
residues
reported
in
the
EIIS)
in
these
incidents
is
questionable.

EFED
response:
The
issue
of
potential
adverse
effects
to
nontarget
organisms
from
sublethal
exposure
to
rodenticides
is
discussed
in
the
assessment,
and
EFED
notes
that
the
lack
of
reproduction
studies
that
could
help
characterize
this
potential
adverse
effect
adds
to
the
uncertainty
of
the
analysis.
These
studies
will
be
required
through
a
data
call­
in.

ii
Gastro­
Intestinal
Tract
(
GIT)
should
be
established
as
an
abbreviation
and
used
consistently
throughout
the
document.

EFED
response:
That
correction
had
already
been
made
in
the
revised
risk
assessment.

ii,
iii
Define
numbers
in
parenthesis
(
9)
after
brodifacoum
and
(
3)
after
bromadiolone.
They
appear
to
have
no
relationship
to
anything.

EFED
response:
These
numbers
are
not
present
in
the
revised
risk
assessment.

iv,
89,
Table
47
Risk
presumptions
in
tables.
EPA
does
not
describe
how
the
risk
Page
RRTF
comment
5
presumptions
(
i.
e.,
low,
moderate,
high)
in
the
two
tables
were
determined.
They
appear
to
have
been
set
using
risk
summary
value
data;
however,
no
rating
scales
or
other
discrimination
criteria
have
been
described.

EFED
response:
That
has
been
addressed
in
the
revised
assessment.
See
also
EFED's
first
comment
above.

1
PCO
vs.
PCA.
Reference
to
a
Pest
Control
Operator
(
PCO)
is
incorrect.
Currently,
40
C.
F.
R.
Part
171
refers
to
a
Pest
Control
Applicator
(
PCA)
commercial
and
private.
See
40
C.
F.
R.
§
171.2.

EFED
response:
That
correction
has
been
made.

1,
2
Presumption
of
equal
exposure.
This
is
a
critical
error
in
the
PRA.
This
is
inappropriate
and
there
is
no
justification
made
for
this
assumption.
Exposure
is
a
key
factor
in
any
risk
assessment.
This
presumption
makes
the
entire
analysis
a
"
hazard
assessment"
and
not
a
"
risk
assessment."
It
is
inappropriate
to
compare
Section
24(
c)
registrations
for
field­
use
only
and
Section
18
island
restoration
uses
with
products
labeled
for
commensal
uses
only.
Besides
inappropriately
assuming
equal
exposure,
this
assumption
also
does
not
account
for
the
large
differential
in
market
share
among
the
products
registered
for
commensal
uses,
a
fact
clearly
stated
in
Table
1,
page
2....
Interchangeability
of
rodenticides.
EPA's
justification
for
a
presumption
of
equal
exposure
is
that
it
will
allow
for
an
evaluation
of
how
risks
(
but
actually
hazard)
might
increase
or
decrease
as
one
rodenticide
is
used
instead
of
another.
This
means
that
EPA
is
assuming
that
all
rodenticides
can
be
used
interchangeably
and
substituted
for
one
another.
This
may
be
true
for
rodenticide
active
ingredients,
but
is
clearly
not
the
case
for
rodenticide
end­
use
products
which
may
have
different
formulations,
bait
strengths,
target
species,
use
sites,
application
methods
and
rates,
use
restrictions,
and
so
on.
This
means
that
EPA's
entire
hazard
analysis
applies
only
to
rodenticide
active
ingredients
and
has
no
meaning
for
evaluating
the
potential
risks
of
end­
use
products
because
product­
specific
and
use­
pattern
specific
factors
have
not
been
accounted
for
through
exposure
assessments.
This
limits
the
usefulness
of
EPA's
analysis
from
a
risk
management
perspective
because
it
is
not
possible
to
propose
risk
mitigation
measures
for
active
ingredients
per
se
and
it
is
inappropriate
to
propose
them
for
end­
use
products
without
first
evaluating
product­
specific
risks.

EFED
response:
See
previous
EFED
comments
on
hazard
versus
risk
above.
A
section
titled
Use
and
Exposure
Considerations
has
been
added
to
the
assessment.
In
this
section,
EFED
explains
the
basis
for
its
exposure
calculations
and
its
assumptions.
In
addition,
the
Agency
does
not
know
the
quantity
of
rodenticides
sold
and
applied
in
the
U.
S.,
although
we
have
repeatedly
requested
this
information
from
rodenticide
registrants.
The
RRTF,
in
a
conference
proceedings
(
Kaukeinen
et
al.
2000),
cites
over­
Page
RRTF
comment
3
Ibid.

4
See
ECOFRAM
Terrestrial
Draft
Report,
1999
at
http://
www.
epa.
gov/
oppefed1/
ecorisk/

6
the­
counter
container
sales
for
four
of
the
nine
rodenticides,
but
provides
no
information
on
geographical
or
state
usage,
urban
versus
non­
urban
use,
quantity
of
active
ingredient
and
bait
sold,
or
any
information
on
use
by
Certified
Applicators.
Submission
of
this
information
will
help
EFED
refine
it's
risk
assessment.

2
For
Field
Uses,
include
control
of
rats
and
voles
under
Zinc
Phosphide.

EFED
response:
That
information
was
included
in
Table
2
in
the
revised
risk
assessment.

3
Correct
reference.
Table
2,
reference
to
EPA
1998
a,
b
should
be
referenced
as
EPA
1998
a,
b
Reregistration
Eligibility
Decision
(
RED).

EFED
response:
That
correction
has
been
made.

4
Spelling.
Fourth
line,
"
sties"
should
be
"
sites."

EFED
response:
That
correction
has
been
made.

7
SAP
review
of
the
Decision
Table
Analysis.
The
Scientific
Advisory
Panel
(
SAP)
reviewed
this
approach
and
strongly
recommended
that
the
term
Risk
Quotient
(
RQ)
as
used
here
should
be
called
a
"
hazard"
quotient
(
HQ).
The
RRTF
agrees
and
believes
that
the
terminology
should
be
changed
throughout
the
document
consistent
with
the
SAP's
comments
(
SAP
Report
No.
99­
01A,
Jan.
22,
1999).

EFED
response:
The
recommendations
presented
to
the
Agency
following
a
SAP
review
are
just
that,
recommendations.
The
Agency
must
consider
the
recommendations
in
light
of
extant
Agency
policies
and
guidance.
In
this
case,
EPA's
Guidelines
for
Ecological
Risk
Assessment3
uses
the
term
Risk
Quotient
to
describe
a
simple
comparison
of
a
measure
of
exposure
divided
by
a
measure
of
toxicity.
In
addition,
the
same
guidelines
notes
that
risk
quotients
provide
an
efficient,
inexpensive
means
of
identifying
high­
or
low­
risk
situations
that
can
allow
risk
management
decisions
to
be
made
without
the
need
for
further
information.
Further,
subsequent
to
the
aforementioned
SAP
review,
another
panel
of
scientists
and
risk
assessors
­
the
Ecological
Committee
on
FIFRA
Risk
Assessment
Methods
(
ECOFRAM),
stated
that
RQs
do
not
quantify
risk
but
are
useful
for
comparisons
among
alternative
compounds.
4
Thus,
no
change
in
the
terminology
is
needed.
Page
RRTF
comment
7
7
Definition
of
"
effect."
The
term
"
measures
of
effect"
as
used
in
the
Decision
Table
Analysis
is
in
error
because
several
of
the
"
effects"
discussed
are
not
truly
effects,
but
fate
properties
of
the
chemical.
For
example,
it
is
inappropriate
to
use
the
terms
"
blood
retention
time"
and
"
liver
retention
time"
as
measures
of
"
effect"
when
the
values
being
used
in
the
assessment
are
actually
elimination
half­
life
values.
The
elimination
and
excretion
of
second­
generation
anticoagulants
is
biphasic
and
the
initial
phase
is
primarily
from
the
liver.
Research
has
shown
that
the
residues
involved
in
this
terminal
phase
do
not
appear
to
contribute
to
coagulopathy.
Further,
at
non­
toxic
concentrations
the
initial
phase
of
elimination
appears
to
be
absent
(
Batten
and
Bratt,
1987).
If
this
is
true,
then
retention
time
in
the
liver,
at
low
levels,
is
not
an
effect,
but
a
marker
of
exposure.

EFED
response:
While
the
retention
time
is
not
a
direct
measure
of
effect
for
secondary
risk
to
birds
and
mammals,
it
is
an
important
contributing
factor.
The
combination
of
mean
%
mortality
from
secondary
laboratory
toxicity
studies
which
characterizes
the
secondary
toxicity
from
short­
term
exposures,
and
available
data
on
retention
time
in
both
blood
and
liver
which
indicates
how
long
toxic
levels
can
persist
in
target
animal
tissues,
can
characterize
the
secondary
risk
to
birds
and
mammals.
If,
however,
retention
time
in
blood
and
liver
were
removed
from
consideration
in
secondary
risk
for
birds
and
mammals,
the
ranking
of
the
rodenticides
providing
the
greatest
overall
risk
to
birds
and
mammals
would
not
change
(
As
seen
in
the
graphs
below,
brodifacouum,
zinc
phosphide
and
diafethialone
provide
the
greatest
overall
risk
in
both
cases.
Figure
1
shows
the
comparison
with
retention
time
included
in
secondary
risk.
Figure
2
show
s
the
comparison
with
retention
times
are
removed
from
consideration.
When
retention
times
are
removed
from
consideration,
the
sum
of
the
weighted
averages
of
measures
of
effect
for
brodifacoum
increases,
as
does
that
for
difethialone.
In
addition,
the
summary
values
for
zinc
phosphide
and
difethialone
are
almost
equal
­
4.63
and
4.60).
Page
RRTF
comment
8

































































































































































































































































































































































































































































































































































































































































































































































































Brodifacoum
50ppm
Bromadiolone
50ppm
Bromethalin
100ppm
Chlorophacinone
100ppm
Chlorophacinone
50ppm
Cholecalciferol
750ppm
Difethialone
25ppm
Diphacinone
100ppm
Diphacinone
50ppm
Warfarin
250ppm
Zinc
Phosphide
20,000ppm
Rodenticide
Baits
0
2
4
6
8
10
Sum
of
Weighted
Averages
(
0
to
10)












Primary
Risk
to
Birds












Primary
Risk
to
Mammals












Secondary
Risk
to
Birds












Secondary
Risk
to
Mammals
Graph
5.
Greatest
Overall
Risk
to
Birds
&
Mammals
Sum
of
Weighted
Averages
of
Measures
of
Effect
Figure
1
Page
RRTF
comment
9





























































































































































































































































































































































































































































































































































































































































































































































































































































Brodifacoum
50ppm
Bromadiolone
50ppm
Bromethalin
100ppm
Chlorophacinone
100ppm
Chlorophacinone
50ppm
Cholecalciferol
750ppm
Difethialone
25ppm
Diphacinone
100ppm
Diphacinone
50ppm
Warfarin
250ppm
Zinc
Phosphide
20,000ppm
Rodenticide
Baits
0
2
4
6
8
10
Sum
of
Weighted
Averages
(
0
to
10)












Primary
Risk
to
Birds












Primary
Risk
to
Mammals












Secondary
Risk
to
Birds












Secondary
Risk
to
Mammals
Graph
5.
Greatest
Overall
Risk
to
Birds
&
Mammals
Sum
of
Weighted
Averages
of
Measures
of
Effect
Figure
2
Page
RRTF
comment
10
Page
RRTF
comment
5
N.
B.
A
correlation
coefficient
is
a
number
between
­
1
and
1
which
measures
the
degree
to
which
two
variables
are
linearly
related.
If
there
is
perfect
linear
relationship
with
positive
slope
between
the
two
variables,
the
correlation
coefficient
is
equal
to
1;
if
there
is
positive
correlation,
whenever
one
variable
has
a
high
(
low)
value,
so
does
the
other.
If
there
is
a
perfect
linear
relationship
with
negative
slope
between
the
two
variables,
the
correlation
coefficient
is
equal
to
­
1;
this
is
a
negative
correlation,
that
is,
whenever
one
variable
has
a
high
(
low)
value,
the
other
has
a
low
(
high)
value.
A
correlation
coefficient
of
0
means
that
there
is
no
linear
relationship
between
the
variables.

11
7,
Table
28
Measures
of
effect
for
primary
risk
to
birds.
The
two
measures
of
effect
used
in
the
analysis
(
dietary
RQ
and
amount
of
bait
needed
to
produce
an
LD50)
are
not
truly
independent
measures
of
effect.
Both
are
based
on
the
inherent
toxicity
of
the
active
ingredient
and,
though
different,
are
highly
correlated.
This
amounts
to
"
double
counting"
of
the
same
measure
of
effect
which
skews
the
analysis.

EFED
response:
EFED
disagrees
that
these
measures
of
effect
are
correlated.
The
two
measures
of
effect
for
primary
risk
to
birds
were
tested
for
correlation
using
the
`
Correlation
and
Regression
Calculator'
at
http://
www.
ebook.
stat.
ucla.
edu/
cgibin
php.
cgi/
calculators/
correlation.
phtml,
and
the
correlation
coefficient
was
0.272307,
indicating
little
linear
correlation.
5
7,
Table
40,
Table
41
Use
of
two
retention
times
as
measures
of
effect.
Blood
retention
time
and
liver
retention
time
are
not
independent
measures
of
elimination
(
half­
lives).
The
values
for
the
two
retention
times
are
usually
not
the
same
for
any
given
species,
but
are
highly
correlated
because
of
similarities
in
metabolism
between
different
organs
and
tissues.
Because
the
measures
are
correlated,
it
is
inappropriate
for
both
measures
to
be
used
in
the
analysis
as
this
amounts
to
"
double
counting"
the
same
endpoint.
This
"
double
counting"
tends
to
exaggerate
the
magnitude
of
the
summary
values,
either
higher
or
lower,
for
all
of
the
rodenticides,
and
makes
those
that
are
more
persistent
look
worse
than
is
actually
the
case.
Furthermore,
retention
times
make
poor
measures
of
effect.
For
example,
they
cannot
distinguish
differences
in
hazard
between
different
bait
strengths,
as
is
apparent
from
the
data
presented
in
Tables
40
and
41.

EFED
response:
See
previous
EFED
response
on
definition
of
effect
on
pages
7
&
8.
In
addition,
the
values
are
not
"
double
counted";
each
is
given
a
weight
one­
half
that
of
other
measures
(
total
weight
of
blood
retention
time
=
5;
total
weight
of
liver
retention
time
=
5;
thus,
total
weight
for
retention
time
=
10),
so
that
the
two
together
have
a
weighting
equal
to
other
measures
(
i.
e.,
10).
Further,
EFED
disagrees
that
retention
times
are
correlated.
The
retention
times
for
blood
and
liver
were
tested
for
correlation
using
the
`
Correlation
and
Regression
Calculator'
at
http://
www.
ebook.
stat.
ucla.
edu/
cgi­
Page
RRTF
comment
6
Ibid.

12
bin/
php.
cgi/
calculators/
correlation.
phtml,
and
the
correlation
coefficient
was
0.105801,
indicating
little
linear
correlation.
6
7,
Table
40,
Table
41
Double
counting
of
retention
times
in
the
analysis.
In
addition
to
the
double
counting
issue
discussed
above,
another
problem
with
EPA's
methodology
is
that
it
uses
the
same
measures
of
effect
for
evaluating
secondary
risks
to
both
birds
and
non­
target
mammals.
Because
the
values
for
the
blood
and
liver
retention
times
are
identical
for
both
the
bird
and
non­
target
mammal
analyses,
this
leads
to
double
weighting
of
these
factors
when
the
overall
summary
values
are
calculated.
This
double
weighting
exaggerates
the
previously
described
problem
that
these
two
measures
of
effect
are
not
independent
and
further
compounds
their
weighting
in
the
analysis,
giving
them
the
equivalent
of
a
quadruple
weighting.

EFED
response:
See
previous
EFED
response
on
definition
of
effect
on
pages
7
&
8.
Again,
the
values
are
not
"
double
counted";
each
retention
time
(
blood
and
liver)
is
given
a
weight
of
2.5
when
used
to
evaluate
each
secondary
risk
(
birds
and
mammals).
Thus,
the
total
weight
of
blood
retention
time
and
the
total
weight
of
liver
retention
time
is
equal
to
a
weighting
equal
to
other
measures
(
i.
e.,
10).

7
Secondary
toxicity
(
hazard)
vs.
secondary
risk.
The
mean
%
mortality
from
secondary
toxicity
studies
are
measures
of
hazard,
not
risk,
because
exposure
in
these
lab
studies
is
often
not
the
same
(
or
sometimes
even
close)
to
exposure
under
actual
field
conditions.
Further,
there
is
no
consideration
of
the
probability
of
exposure,
a
key
aspect
of
any
risk
assessment.

EFED
response:
See
previous
EFED
response
on
the
potential
for
risk
from
rodenticides.
We
also
note
that
to
determine
the
probability
of
risk
would
require
additional
data
on
toxicity
and
exposure.
Additional
data
needed
to
refine
this
risk
assessment
is
presented
in
a
section
on
Uncertainty
and
Data
Needs
in
the
comparative
risk
assessment.

7,
Table
40,
Table
41
Inappropriate
use
of
data
from
secondary
toxicity
studies.
The
specific
end­
use
products
(
including
bait
strengths),
protocols,
and
test
conditions
(
e.
g.,
target
and
non­
target
species,
number
of
animals,
period
of
feeding)
used
in
these
studies
often
differed
significantly.
Therefore,
it
is
inappropriate
to
compare
the
results
of
mean
mortality
from
one
set
of
studies
with
those
from
another
set
of
studies
as
if
they
were
performed
under
identical
conditions.
Page
RRTF
comment
13
EFED
response:
A
number
of
laboratory
tests
using
avian
and
mammalian
predators
and
scavengers
to
test
for
mortality
due
to
secondary
exposure
were
available
and
used
in
this
assessment.
Their
design
and
methods
varied
considerably
adding
unknown
variability
to
their
results
and
to
the
analysis.
Pending
the
development
of
standard
methods
and
testing
requirements
for
these
tests
they
provide
the
best
data
available.
EFED
has
identified
additional
data
needed
to
refine
this
risk
assessment
in
a
section
on
Uncertainty
and
Data
Needs.

8
Assignment
of
importance
and
weights
for
importance.
All
measures
of
effect,
except
for
two,
were
assigned
a
"
high"
measure
of
importance
for
the
analysis.
The
two
that
were
assigned
a
"
medium"
importance
(
half
lives
in
blood
and
liver)
are
correlated
so
"
persistence"
was
also
indirectly
given
a
"
high"
weighting
due
to
double
counting.
There
is
no
explanation,
or
rationale,
given
by
EPA
for
the
selection
of
importance
(
high,
medium,
low)
for
the
different
measures
of
effect
or
the
weights
assigned
to
the
importance
values
(
i.
e.,
high
=
10,
medium
=
5,
low
=
3.33).

EFED
response:
See
previous
EFED
responses
on
weighting
on
page
9,
Use
of
two
retention
times
as
measures
of
effect,
and
on
page
10,
Double
counting
of
retention
times
in
the
analysis.
As
noted
on
page
6
of
the
document,
all
measures
of
effect,
except
two,
are
assigned
a
"
high"
(
10
out
of
10)
measure
of
importance
for
the
rodenticide
analysis.
The
half­
life
in
blood
and
liver
are
each
given
a
weight
of
"
low"
(
2.5
out
of
10)
for
analyzing
secondary
risks
to
birds
and
mammals,
so
that
the
overall
importance
of
the
persistence
data
(
2.5
x
4=
10)
equals
but
does
not
exceed
that
of
the
mortality
data.
The
intention
was
to
weigh
all
measures
of
effects
and
all
risks
equally
in
the
analysis.
This
would
eliminate
the
introduction
of
any
value
judgements
on
the
part
of
the
risk
assessors.

13
GIT.
See
GIT
comment
above
for
page
ii.

EFED
response:
Previously
addressed.

13
Categorizing
second­
generation
rodenticides.
Categorizing
rodenticide
active
ingredients
as
"
bad
actors"
is
to
use
non­
standard,
subjective,
and
qualitative
terminology
in
a
regulatory
document.
It
is
not
a
scientific
or
regulatory
term
and
therefore
difficult
to
interpret
in
the
regulatory
context.
The
Pesticide
Action
Network
(
PAN)
is
not
an
official
government
organization
and
should
not
be
used
as
a
reference
in
this
document
without
proper
qualification.
The
World
Health
Organization
(
WHO)
may
state
that
the
second­
generation
active
ingredients
are
"
extremely
hazardous"
(
not
representing
high
risk),
but
all
formulations
containing
these
active
ingredients
are
highly
diluted
(
20,000x)
in
formulation
and
as
formulated
products
are
Category
IV
(
label
word,
Caution)
for
all
five
acute
hazard
indicators.
This
should
be
a
key
factor
in
Page
RRTF
comment
14
any
risk
assessment
and
must
be
included
in
the
PRA.
Thus,
the
concepts
of
hazard
and
risk
are
again
blurred
and
poorly
delineated
by
the
authors
of
this
document.

EFED
response:
That
descriptor
was
removed
from
the
revised
risk
assessment.
The
RRTF
is
correct
in
stating
that
rodenticide
baits
are
highly
diluted
from
the
pure
active
ingredient.
Nonetheless,
registered
products
have
been
tested
and
proven
efficacious
in
killing
target
species
(
rats,
mice,
and
other
small
mammals);
even
larger
mammals,
including
humans,
have
died
after
ingesting
formulated
bait.
According
to
the
New
York
State
Department
of
Environmental
Conservation,
deer
died
after
consuming
bait,
and
HED's
toxicity
database
lists
an
incident
in
Indonesia
in
which
20
people
died
after
consuming
brodifacoum­
treated
rice
intended
and
labeled
for
use
as
a
rodenticide.

16
Correct
acute
toxicity
data.
Listing
for
Laboratory
Rat,
2.5
and
2.1
should
not
be,
as
EPA
found
this
study
deficient
and
therefore
unacceptable
with
a
new
study
being
required
(
EPA
letter
dated
Feb.
6,
1992).
In
the
replacement
study,
accepted
by
EPA,
the
laboratory
oral
LD50
for
rats
was
7.0
mg/
kg.
This
number
should
be
listed
and
used
in
later
references.

EFED
response:
The
RRTF
provides
no
supporting
documentation
that
this
study
is
"
unacceptable".
The
study
is
categorized
as
"
supplementary"
in
the
HED's
toxicity
database,
and
data
from
supplementary
studies
are
used
in
OPP
risk
assessments.

29
Target
species.
Table
15
data
citation
Riedel
et
al.,
1991
is
incorrect.
Target
species
is
listed
as
mouse
in
table;
in
Literature
Citations
target
is
listed
as
voles.
It
should
be
noted,
however,
that
there
are
no
registrations
for
brodifacoum
in
the
U.
S.
with
voles
as
a
target
species.

EFED
response:
The
RRTF
provides
no
supporting
documentation
that
this
citation
is
incorrect.
The
information
cited
in
the
risk
assessment
is
correct
according
to
Joermann
(
1998).

30
Correct
reference.
Table
15
data
citation
Riedel
et
al.,
1991
is
footnoted
with
reference
to
Joerman,
1998.
This
is
incorrect.

EFED
response:
The
information
is
cited
in
Joermann
(
1998).
The
RRTF
does
not
state
why
this
citation
is
supposedly
incorrect.

30
Correct
residue
data.
Footnote
a
must
be
corrected.
Data
on
residue
levels
in
target
species
have
been
submitted
to
EPA
(
MRIDs
43534601
and
43534602).
The
data
indicate
the
results
of
field
trials
conducted
with
diphacinone
baits
against
the
California
ground
squirrel
­­
the
principal
target
species
for
which
diphacinone
is
used
in
field
applications.
Genesis
Laboratories,
on
behalf
of
the
California
Department
of
Food
and
Page
RRTF
comment
15
Agriculture
(
CDFA),
conducted
these
trials.
As
part
of
these
trials,
dead
ground
squirrels
were
collected
and
analyzed
for
diphacinone
residues.
For
the
0.005%
treatment
(
10
samples),
mean
whole
body
residues
found
were
1.4
ppm
with
a
deviation
of
0.8
ppm.
For
the
0.01%
treatment
(
10
samples),
mean
whole
body
residues
found
were
1.4
ppm
with
a
deviation
of
0.7
ppm.

EFED
response:
Previously,
EFED
had
no
record
of
these
studies
but
has
since
obtained
copies
from
the
Agency's
microfiche
files.
The
residue
data
from
these
studies
are
now
included
the
residue
data
in
the
revised
risk
assessment.

35
Lower
number
of
secondary
studies.
Last
paragraph,
sentence
3:
In
the
Bullard,
Thompson,
and
Holguin
diphacinone
study,
accepted
and
cited
by
EPA
for
the
liver
retention
time
(
of
concern)
­­
90
days,
30
rats
were
fed
these
same
livers
for
14
days
and
there
were
"
0"
deaths
with
no
increase
in
the
rats'
prothrombin
times.
That
would
change
the
number
of
secondary
studies
on
diphacinone
to
4
studies
and
change
the
calculations
to
19
(
30%)
of
63
tested
mammals
dying.
To
be
scientifically
consistent,
mention
of
the
results
of
this
portion
of
the
study
should
be
made
by
EPA.

EFED
response:
The
Rodenticide
Cluster
Reregistration
Eligibility
Decision
(
RED)
issued
in
July,
1998,
required
secondary
toxicity
studies
with
a
mammalian
predator
and
an
avian
predator
to
support
reregistration
of
0.005%
ai
and
0.01%
ai
diphacinone
baits.
Four
years
have
passed
without
the
registrant
addressing
this
data
gap.
Because
the
rat
is
a
target
species
for
rodenticides,
citing
rat
data
will
not
fulfill
this
secondary­
toxicity
requirement.

35
Dietary,
not
secondary
studies.
In
general,
studies
with
captive
or
laboratory
animals
where
the
chemical
is
provided
to
carnivores/
omnivores
as
spiked
meat
or
dog
food
preparations
are
not
secondary
toxicity
studies.
These
are
dietary
exposure
studies
and
these
references
should
be
removed
from
this
section.

EFED
response:
EFED
considers
these
studies
as
indicative
of
secondary
toxicity.
Nontarget
predators
and
scavengers
can
be
exposed
to
and
adversely
affected
by
rodenticides
via
dietary
exposure
to
dead
or
moribund
target
organisms.

Table
37
Diphacinone
retention/
elimination.
Data
in
Diaz
and
Whitacre,
1976
(
which
were
discussed
on
the
previous
page)
indicate
that
elimination
of
diphacinone
in
the
rat
is
rapid
and
similar
to
chlorophacinone.
These
data
were
not
included
in
Table
37
or
in
EPA's
analysis,
which
relied
only
on
elimination
data
for
blood
and
liver.
Instead,
EPA
used
blood
data
from
cattle
and
liver
data
from
humans
that
indicated
much
higher
retention
times
and
produced
much
higher
measure
of
effect
values
in
Table
40.
It
is
inappropriate
to
directly
compare
elimination
data
generated
with
cattle
and
humans
with
those
generated
with
rats
or
other
species
because
of
interspecies
variations
in
Page
RRTF
comment
16
metabolism
and
study
dosing
regimens.
It
is
also
inappropriate
to
use
cattle
and
human
elimination
data
as
measures
of
effect
when
neither
of
these
are
either
target
or
non­
target
species
being
considered
in
the
assessment.

EFED
response:
Those
data
are
discussed
in
the
risk
assessment.
The
data
tabulated
are
half­
lives
and
retention
times
(
days).
Those
values
are
not
obtainable
from
Diaz
and
Whitacre
(
1976);
as
stated
in
the
risk
assessment,
nearly
a
third
of
the
dose
administered
was
not
recovered
in
that
study.

Tables
40
&
41
Source
of
data
not
shown.
EPA
does
not
cite
the
source
of
the
retention
time
values
listed
in
this
table,
which
are
subsequently
used
to
derive
measure
of
effect
values.
The
values
cannot
be
verified
without
this
information.

EFED
response:
Attachment
C
states
the
source
of
data
used
in
the
decision
analysis.

41
GIT.
See
GIT
comment
above
for
page
ii.

EFED
response:
Previously
addressed.

45,
46
2
gram
pellets,
erroneous.
Describing
the
"
average
pellet"
weight
as
2
grams
is
in
error
by
an
order
of
magnitude
or
more
and
the
large
number
of
LD50s
per
pellet
is
incorrect
and
misleading.
For
example,
Talon
3/
16
inch
pellets
weigh,
on
average,
0.2
g,
and
a
smaller
3/
32
inch
pellet
that
weighs
less
is
also
available.
With
the
exception
of
mouse­
sized
or
smaller
animals,
it
is
not
true
that
one
or
two
pellets
of
brodifacoum
bait
will
kill
a
single
animal.
Ingestion
of
7
to
14
of
the
larger
pellets
is
required
to
kill
a
rat
and
considerably
more
for
the
larger
non­
targets.
This
error
is
repeated
throughout
the
document
and
leaves
a
false
impression
that
one
granule
will
kill
an
organism.
All
assumptions,
calculations,
and
conclusions
based
on
this
statement
must
be
corrected.

EFED
response:
This
correction
has
been
made
in
the
revised
risk
assessment.

46,
Table
26
Source
of
data
not
shown.
EPA
does
not
cite
the
source
of
the
LD50
values
listed
in
this
table,
which
are
subsequently
used
to
derive
measure
of
effect
values.
The
values
cannot
be
verified
without
this
information.

EFED
response:
Attachment
C
states
the
source
of
data
used
in
the
decision
analysis.

47
b/
c
The
dietary
RQs
should
be
defined
as
HQs.
The
footnotes
do
not
provide
sufficient
background
and
justification
for
the
rate
(
100%
or
20%
intake
of
daily
intake)
or
timeframe
of
exposure
(
i.
e.,
"
several
days").
The
bottom
line
is
that
the
concentration
in
the
bait
does
not
provide
an
estimate
of
exposure
and
the
PRA
does
not
provide
a
case
Page
RRTF
comment
17
for
the
probability
of
actual
exposure.
The
latter
is
dependent
on
use
pattern
that
is
ignored
in
these
calculations.

EFED
response:
See
previous
EFED
responses
on
page
2,
for
Hazard
not
risk,
and
page
5
for
Presumption
of
equal
exposure.

47,
Table
27
Source
of
data
not
shown.
EPA
does
not
cite
the
source
of
the
LC50
values
listed
in
this
table,
which
are
subsequently
used
to
derive
measure
of
effect
values.
The
values
cannot
be
verified
without
this
information.

EFED
response:
Attachment
C
states
the
source
of
data
used
in
the
decision
analysis.

48,
Table
28,
Figure
1,
Attachment
C
Error
in
calculation
of
summary
values.
EPA
has
made
a
significant
calculation
error
when
calculating
the
summary
values
for
primary
risks
to
birds
due
to
a
mistake
in
weighted
average
values
for
the
second
measure
of
effect
(
grams
of
bait
needed
for
a
50
g
bird
LD50
dose).
EPA
calculated
the
weighted
average
values
for
this
measure
of
effect
by
indexing
to
the
least
toxic
rodenticide
(
diphacinone
50
ppm)
rather
than
the
most
toxic
one
as
was
done
for
the
other
measures
of
effect
[
Note:
similar
incorrect
calculations
were
also
done
for
primary
risk
to
mammals,
this
will
be
discussed
below].
EPA's
method
of
calculation
is
presented
on
page
133
(
Step
3,
substep
B)
in
Attachment
C.
This
method
skewed
results
so
that
both
brodifacoum
and
zinc
phosphide
were
given
the
same
weighted
average
of
5.0,
even
though
it
is
clear
from
the
data
for
this
measure
of
effect
that
zinc
phosphide
(
LD50
dose
=
0.03
g)
is
almost
10
times
more
hazardous
than
brodifacoum
(
LD50
dose
=
0.26
g)
and
should
be
weighted
accordingly.
Results
for
the
other
rodenticides
were
also
skewed
in
a
similar
manner.
The
table
below
presents
EPA's
values
and
corrected
results
based
on
indexing
to
the
most
toxic
rodenticide
by
using
the
inverse
of
the
LD50
dose
(
i.
e.,
weighted
averages
are
calculated
by
indexing
to
the
value
of
33.33
g
for
zinc
phosphide,
rather
than
400
g
for
diphacinone
50
ppm).
After
the
values
have
been
corrected,
zinc
phosphide
has
the
highest
ranking
based
on
summary
values.

EFED
response:
The
Agency
agrees
with
using
the
inverse
of
the
number
of
bait
pellets
equal
to
an
LD50
dose
and
indexing
based
on
the
highest
number.
The
necessary
corrections
have
been
made
in
the
revised
risk
assessment.

49
Incorrect
presentation
of
bait
concentrations.
In
all
graphs
and
figures
in
the
document,
the
assay
of
active
ingredient
is
listed
incorrectly
for
all
products
as
x
mg.
This
should
be
correctly
listed
as
x
mg/
kg
bait
or
ppm.
This
must
be
corrected
as
it
gives
the
reader
a
false
sense
of
what
is
being
stated.
These
are
concentrations
(
i.
e.,
rates
of
exposure),
not
fixed
amounts.

EFED
response:
A
change
to
ppm
has
been
made.
Page
RRTF
comment
18
51
Table
29.
The
footnote
for
Coumatetralyl
and
Difenacoum
should
be
(
b)
not
registered
in
the
United
States
EFED
response:
Correction
has
been
made.

54
Nomenclature.
The
taxonomy
entry
for
Chaffinch
should
have
the
scientific
name
Fringilla
coelebs
following
the
entry
as
this
is
the
first
reference
to
the
species.

EFED
response:
Correction
has
been
made.

55
Speculation
of
sub­
lethal
effects.
The
authors
of
the
PRA
refer
in
several
places
to
speculations
by
authors
that
there
might
be
long­
term
physiological
or
behavioral
effects.
There
are
no
substantive
data
that
support
these
sub­
lethal
effects.
All
of
the
references
cited
by
EPA
regarding
this
issue
are
speculative
and
no
data
are
available.
The
only
behavioral
effects
are
associated
with
lethal
levels
of
anticoagulants.
Discussion
of
sublethal
effects
must
be
highly
qualified
as
speculative
in
this
document.

EFED
response:
We
disagree
that
all
the
references
cited
indicating
the
potential
for
sublethal
effects
are
speculative.
EFED
acknowledges
that
additional
data
are
needed
to
confirm
the
reasoned
arguments
that
sublethal
effects
adversely
impact
nontarget
organisms
exposed
to
rodenticides.
Toward
this
end,
and
as
previously
stated,
the
potential
for
adverse
sublethal
effects
will
be
addressed
through
a
data
call­
in.

56
Products
not
comparable.
Paragraph
2.
"
calciferol"
(
vitamin
D2)
in
the
UK,
is
an
entirely
different
product
and
is
unrelated
to
the
cholecalciferol
(
vitamin
D3)
in
the
U.
S.
This
comparison
should
be
removed
as
irrelevant.

EFED
response:
The
products
may
be
different,
but
the
toxicologically
active
metabolites
may
have
similar
effects,
including
hypercalcemia
and
degeneration
of
bone
matrix.
Eason
et
al.
(
2000)
state
that
cholecalciferol
(
Vitamin
D3)
must
undergo
metabolic
conversion
to
25­
hydroxycholecalciferol
(
25OHD)
to
gain
biological
and
toxicological
activity.
If
the
RRTF
has
information
that
calciferol
metabolizes
in
a
different
manner
or
does
not
have
comparable
toxicological
effects
in
animals,
documentation
should
be
provided.

56
Using
correct
toxicity
data.
Second
paragraph
from
the
bottom,
reference
to
the
decision
table
analysis
should
be
based
on
the
above­
mentioned
7.0
mg
a.
i./
kg
acute
oral
dose.

EFED
response:
As
previously
discussed,
EFED
has
checked
the
values,
and
the
values
used
in
the
risk
assessment
are
correct.
Page
RRTF
comment
19
57
Using
correct
toxicity
data.
Table
31.
Diphacinone
should
be
listed
as
7.0
mg
a.
i./
kg.
(
acute
oral
rat),
instead
of
the
2.3
found
in
the
EPA
unacceptable
study.
The
entries
in
the
table
for
diphacinone
should
be:

This,
of
course,
changes
the
summary
values
for
diphacinone.
Diphacinone
100
ppm
moves
to
under
Chlorophacinone
100
ppm,
and
Diphacinone
50
ppm
moves
under
Chlorophacinone
50
ppm.

EFED
response:
See
previous
comment.

57,
Table
31
Source
of
data
not
shown.
EPA
does
not
cite
the
source
of
the
LD50
values
listed
in
this
table,
which
are
subsequently
used
to
derive
measure
of
effect
values.
The
values
cannot
be
verified
without
this
information.

EFED
response:
LD50
values
used
in
the
comparative
risk
assessment
are
listed
in
the
revised
Table
31
in
the
revised
risk
assessment.

58­
59,
Table
32,
Figure
2,
Attachment
C
Error
in
calculation
of
summary
values.
EPA
has
made
a
significant
calculation
error
when
calculating
the
summary
values
for
primary
risks
to
mammals
due
to
a
mistake
in
weighted
average
values
for
the
measure
of
effect
(
grams
of
bait
needed
for
a
25
g
mammal
LD50
dose).
EPA
calculated
the
weighted
average
values
for
this
measure
of
effect
by
indexing
to
the
least
toxic
rodenticide
(
chlorophacinone
50
ppm)
rather
than
the
most
toxic
one
as
was
done
for
the
other
measures
of
effect.
This
method
skewed
results
so
that
both
brodifacoum
and
zinc
phosphide
were
given
almost
the
same
weighted
average,
even
though
it
is
clear
from
the
data
for
this
measure
of
effect
that
zinc
phosphide
(
LD50
dose
=
0.03
g)
is
almost
7
times
more
hazardous
than
brodifacoum
(
LD50
dose
=
0.20
g)
and
should
be
weighted
accordingly.
Results
for
the
other
rodenticides
were
also
skewed
in
a
similar
manner.
The
table
below
presents
EPA's
values
and
corrected
results
based
on
indexing
to
the
most
toxic
rodenticide
by
using
the
inverse
of
the
LD50
dose
(
i.
e.,
weighted
averages
are
calculated
by
indexing
to
the
value
of
33.33
g
for
zinc
phosphide,
rather
than
3.10
g
for
chlorophacinone
50
ppm).
Note
that
because
there
is
only
one
measure
of
effect
for
evaluating
risk
to
non­
target
mammals,
the
summary
values
are
identical
to
the
average
weighted
values
for
this
measure
of
effect.

EFED
response:
The
Agency
agrees
with
using
the
inverse
of
the
number
of
bait
pellets
equal
to
an
LD50
dose
and
indexing
based
on
the
highest
number.
The
necessary
corrections
have
been
made.

59
Figure
2.
Same
changes
as
Figure
1.

EFED
response:
The
necessary
changes
have
been
made.
Page
RRTF
comment
20
59
Spelling.
Fifth
line
from
the
bottom,
correct
spelling
is
cholecalciferol.

EFED
response:
Correction
has
been
made.

60
Correction
of
footnotes.
Table
33
­­
the
footnotes
for
(
c)
and
(
d)
are
missing
from
the
table.

EFED
response:
The
footnotes
have
been
added
within
the
table.

61
Selective
and
misleading
presentation
of
data.
The
entire
presentation
of
the
toxicokinetics
(
absorption,
metabolism,
and
excretion)
is
based
on
a
selective
and
misleading
interpretation
of
the
data.
The
half­
life
of
residues
of
second­
generation
anticoagulants
cannot
be
characterized
by
a
single
number.
The
elimination
from
the
body
is
biphasic.
The
rapid
initial
(
a­)
phase
(
a
few
days)
is
related
to
toxicity
and
the
extent
of
exposure
in
this
phase
is
the
determinant
factor
in
toxicity.
The
PRA
discusses
the
residues
in
various
tissues
and
the
longer
b­
phase
of
elimination
(
hundreds
of
days)
leading
the
reader
to
conclude
that
toxic
residues
are
present
for
hundreds
of
days.
There
are
two
distinct
half­
lives
and
it
is
incorrect
and
misleading
to
discuss
toxic
residues
as
having
the
longer
half­
lives
(
hundreds
of
days).
The
authors
of
the
PRA,
however,
ignore
discussion
from
the
same
articles
referenced
for
residue
data
(
e.
g.,
Batten
and
Bratt,
1987)
that
present
observations
that
the
b­
phase
is
not
dose­
related
and
not
related
to
toxicity
(
i.
e.,
coagulopathy).
When
exposure
occurs
at
non­
toxic
levels,
only
the
b­
phase
of
elimination
is
evident,
indicating
that
low­
level
exposure
may
occur
without
being
toxicologically
significant.
This
is
an
important
point
in
a
balanced
and
complete
discussion
of
the
toxicokinetics
data.

EFED
response:
The
existence
of
biphasic
kinetics
in
the
liver
is
now
discussed
in
a
comprehensive
and
balanced
way
in
the
document.
However,
the
RRTF
should
be
aware
that
not
all
studies
have
demonstrated
biphasic
elimination.

64
Correction.
The
first
sentence
on
this
page
is
incorrect.
Two
animals
did
not
die
in
the
top
dose
level.
All
animals
that
exhibited
marked
toxicity
were
euthanized
according
to
the
protocol.
The
authors
of
the
PRA
ignore
one
of
the
major
points
of
this
paper,
that
toxicity
is
associated
with
the
rapid
a­
phase
of
clearance
and
not
the
b­
phase
of
clearance.
The
b­
phase
residues
are
associated
with
long­
term
liver
residues
and
are
independent
of
dose.
This
makes
liver
residues,
especially
low­
level
residues,
a
good
marker
of
exposure,
but
a
poor
indicator
of
causative
agent.

EFED
response:
See
previous
comment.
As
previously
noted,
the
issue
of
low­
level
exposure
will
be
addressed
through
a
data
call­
in.
Page
RRTF
comment
21
70
GIT.
See
GIT
comment
above
for
page
ii.

EFED
response:
Previously
addressed.

70
Secondary
hazard
vs.
risk.
The
authors
discuss
laboratory
data
as
a
basis
for
determining
secondary
risk.
This
is
not
possible.
Risk
cannot
be
determined
without
an
estimate
or
probability
of
exposure.
As
discussed
by
the
SAP
(
SAP
Report
No.
99­
01A,
Jan.
22,
1999)
in
reviewing
the
Decision
Analysis,
this
is
a
hazard
assessment,
not
a
risk
assessment.
The
SAP
Report
states:
"
The
Panel
encourages
the
Agency
to
change
the
term
risk
to
"
hazard."
The
calculation
of
the
RQ
does
not
include
elements
of
risk.
.
.
."

EFED
response:
See
previous
EFED
responses
above.

71,
Table
40
Correction,
Table
40.
For
diphacinone,
secondary
mortality,
EPA
has
used
an
active
ingredient
blended
rate
of
mortality
of
9.0%.
This
should
be
a
product­
specific
value,
however.
As
in
the
previous
Table
13,
"
Secondary
Toxicity
of
Diphacinone
to
Birds
.
.
.,"
it
is
clearly
shown
that
there
is
a
difference
in
secondary
toxicity
to
the
predator
if
the
prey
receives
bait
containing
50
ppm
versus
products
or
prepared
diets
with
higher
concentrations.
When
predators
consume
prey
that
fed
upon
bait
containing
50
ppm
or
less,
there
is
"
0%"
mean
mortality
to
the
secondary
species.
This
demonstrates
the
problem
of
confusing
active
ingredient
vs.
formulated
product
and
should
be
corrected
before
the
PRA
is
released
for
public
comment.

EFED
response:
Nine
percent
mean
mortality
was
assigned
to
both
formulations
of
diphacinone
as
a
measure
of
secondary
risk
based
on
the
assumption
that
the
target
would
contain
approximately
the
same
residue
regardless
of
which
formulation
it
was
exposed
to.
Data
providing
more
detailed
information
by
formulation
is
needed
to
refine
this
assessment.

72
Field
data
taken
out
of
context.
In
this
document,
numerous
types
of
field
studies
are
referenced,
but
none
of
these
studies
are
directly
applicable
to
assessing
the
risks
of
products
used
to
control
commensal
rodents
(
i.
e.,
"
in
and
around
buildings").
It
is
unclear
to
the
reader
that
these
field
data
are
from
research
and
development
studies
for
products
never
registered
or
for
localized
island
restoration
projects.
This
false
impression
must
be
corrected
prior
to
public
release
of
this
document.

EFED
response:
The
Agency
believes
that
the
description
of
the
field
studies
and
their
results
provide
accurate
information
on
the
effects
of
rodenticides
outdoors
and
and
does
not
leave
the
reader
with
a
false
impression.
There
is
no
clear
distinction
between
commensal
uses
and
field
or
other
outdoor
uses.
Labels
for
commensal­
use
products
do
not
limit
bait
placements
to
any
specified
distance
from
buildings,
and
"
in
and
around
buildings"
may
be
interpreted
differently
among
rodenticide
users.
Thus,
some
Page
RRTF
comment
22
commensal
uses,
especially
in
rural
areas,
might
have
comparable
exposure
scenarios
to
some
field
uses.

73
Relevance
of
field
studies
conducted
outside
the
United
States.
Studies
such
as
Duckett,
1984,
involving
Asian
owls
in
Malaysian
oil­
palm
plantations,
are
not
relevant
to
the
labeled
use
in
the
United
States.
Asian
owls
are
larger
and
take
larger
prey
(
e.
g.,
rats)
than
do
North
American
owls.
Rats
were
present
due
to
the
monoculture
of
this
crop
and
owls
were
encouraged
to
inhabit
the
plantations
using
nest
boxes.
At
the
same
time,
anticoagulants
are
used
to
control
rats.
This
is
completely
opposite
of
the
use
pattern
labeled
in
the
United
States.
Although
some
hazard
information
can
be
gleaned
from
such
studies,
there
is
no
relevance
to
the
exposure
to
wildlife
from
current
use
patterns
(
i.
e.,
commensal
uses).
This
must
be
clearly
stated.

EFED
response:
These
studies
are
presented
in
a
hazard
context
and
confirm
the
potential
for
adverse
effects
of
exposure.
See
also
previous
comment.

75
The
author's
use
of
unrelated
data
to
justify
a
position
that
is
incorrect
and
misleading.
The
"
Incident
Data
Birds
and
Non­
target
Mammals,"
page
77
through
86
of
the
PRA
is
used
as
a
reference.
The
author
references
the
RRTF's
proposal
for
a
0.7
ppm
threshold
of
toxicity
for
brodifacoum
in
liver
tissue
­­
a
concept
clearly
based
on
brodifacoum
data.
The
author
cites
a
study,
Savarie
et
al.,
1979,
in
which
liver
tissue
from
coyotes
was
examined
and
found
to
contain
residues
of
<
0.7
ppm
diphacinone.
The
use
of
unrelated
data
(
brodifacoum
vs.
diphacinone)
to
justify
a
position
(
rejection
of
the
toxicity
threshold
of
0.7
ppm
based
on
brodifacoum
data,
the
largest
body
of
data
for
any
of
the
rodenticides)
is
not
scientifically
justified.

EFED
response:
The
reference
to
the
diphacinone
study
was
deleted
from
this
discussion
in
the
revised
risk
assessment.
The
RRTF
is
correct
in
stating
that
such
a
"
threshold
of
toxicity"
would
need
to
be
established
for
each
of
the
nine
rodenticides
if
the
concept
were
to
be
of
any
value.
Addressing
the
issue
of
adverse
effects
from
sublethal
toxicity
also
needs
to
be
considered
in
such
a
threshold.

75
Inappropriate
comparisons.
Table
41.
Decision
Table
Analysis
for
Secondary
Risk
to
Bird.
It
is
not
scientifically
justified
to
compare
rodenticides,
when
values
for
Blood
Retention
and
Liver
Retention
are
taken
from
different
species,
e.
g.,
cattle
and
humans
vs.
rats.

EFED
response:
EFED
acknowledges
the
variable
nature
of
the
retention
data;
however,
these
are
the
only
data
available.
Additional
data
on
retention
in
tissues
of
target
organisms
would
greatly
facilitate
a
refinement
of
the
risk
assessment.

76
Figure
4.
The
same
correction
as
in
previous
figures.
Page
RRTF
comment
23
EFED
response:
The
necessary
changes
have
been
made.

77
Errors
in
the
EIIS
database
carry
over
to
the
PRA.
The
EPA,
in
conducting
the
PRA
of
anticoagulant
rodenticides,
emphasizes
the
number
of
wildlife
mortality
incidents
reported
to
EPA,
particularly
by
California
and
New
York.
EPA
data
on
wildlife
mortality
incidents
were
obtained
through
a
request
for
information
by
the
RRTF
under
the
Freedom
of
Information
Act
(
FOIA).
These
data
have
been
reviewed
and
analyzed
by
cross­
referencing
to
EPA
and
state
(
California,
New
York,
and
other
states)
incident
numbers,
the
report
date,
the
species
reportedly
involved,
the
compound(
s)
reported,
the
number
of
individual
mortalities
reported
per
incident,
tissue
residue
levels,
the
presence
of
raw
data,
the
presence
of
necropsy
information,
the
relative
condition
of
carcasses,
and
any
indication
of
intentional
or
unintentional
misuse
(
off­
label
use)
of
the
rodenticide
products.
The
underlying
data
was
also
analyzed
using
a
threshold
of
toxicity
based
on
liver
residues
(
Kaukeinen,
Spragins,
and
Hobson,
2000)
that
differentiates
residues
that
are
clearly
acutely
toxic
and
very
low
residues
that
are
simply
a
marker
of
exposure.

This
analysis
demonstrates
that
the
toxicological
and
ecological
significance
of
the
wildlife
mortality
incident
data
for
anticoagulant
rodenticides
is
greatly
overstated.
There
are
numerous
factors
that
restrict
the
number
of
wildlife
mortality
incidents
that
can
be
accurately
attributed
to
anticoagulant
rodenticides.
There
are
as
many
as
30%
redundant
reports
(
i.
e.,
multiple
reports
of
the
same
incident)
in
the
EIIS
database
relative
to
anticoagulants.
EPA
conclusions
relative
to
many
incident
reports
are
not
supported
by
the
underlying
data.
For
many
of
the
incident
reports
the
residue
levels
of
anticoagulants
are
very
low
and
are
not
indicative
of
anticoagulant
toxicity.
Reported
pathological
observations
are
often
not
diagnostic
of
anticoagulant
toxicity
and
often
do
not
provide
a
basis
for
attributing
mortality
to
anticoagulant
rodenticides.
The
role
of
misuse
(
intentional
or
unintentional)
is
not
consistently
documented
in
incident
reports,
but
may
play
a
role
in
the
many
incidents
attributable
to
anticoagulant
rodenticides.
The
primary
conclusion
of
this
analysis
is
that
the
magnitude
of
reported
incidents
alleged
to
be
caused
by
anticoagulant
rodenticides
is
significantly
over­
estimated.

When
incident
numbers
for
the
226
incidents
referred
to
in
the
PRA
were
requested
by
the
RRTF,
the
authors
did
not
have
the
incident
numbers,
suggesting
that
the
analysis
by
the
RRTF
was
not
reviewed
or
considered
by
EPA
prior
to
finalizing
the
current
draft
of
the
PRA.
If
the
RRTF
analysis
of
the
EIIS
database
had
been
thoroughly
reviewed,
the
incident
numbers
would
have
been
obtained
for
comparison.
The
RRTF
believes
that
the
errors
pointed
out
in
the
EIIS
database
constitute
a
serious
error
in
the
PRA
and
must
be
corrected
before
this
document
is
released
for
public
review.

EFED
response:
The
RRTF's
assertion
that
the
authors
did
not
provide
incident
tracking
numbers
is
incorrect.
Incident
tracking
numbers
for
all
incidents
cited
in
the
assessment
Page
RRTF
comment
24
were
provided
(
via
the
Special
Review
and
Reregistration
Division)
to
the
RRTF
when
requested
in
November
of
2001.
The
Agency
is
now
aware
of
258
rodenticide
incidents.

81
Correct
spelling
is
Contra
Costa
County.

EFED
response:
The
change
from
"
Contra
costa
County"
to
"
Contra
Costa
County"
has
been
made.

81
Unbalanced
review
of
data.
The
authors
of
the
PRA
spend
more
than
half
of
a
page
justifying
the
reference
to
one
Golden
eagle
mortality
arguing
that
it
is
a
brodifacoum
mortality
with
0.04
ppm
in
the
liver.
Brodifacoum
was
only
"
implicated,"
however.
This
raises
three
important
issues:
1)
the
majority
of
residues
reported
in
wildlife
are
below
0.7
ppm
in
the
liver
and
one­
third
are
below
0.1
ppm,
making
interpretation
of
low­
level
residues
a
very
important
issue
requiring
a
comprehensive,
scientifically
defensible
discussion;
2)
residues
below
0.7
ppm
are
frequently
reported
in
healthy
feral
animals;
and
3)
pathology
is
not
diagnostic
of
anticoagulant
toxicosis
and
cannot
be
used
in
combination
with
low­
level
liver
residues
as
the
determinative
criteria
of
a
causative
agent
in
a
wildlife
mortality
incident.

The
majority
of
reported
liver
residues
of
anticoagulant
rodenticides
in
the
EIIS
database
are
well
below
0.7
ppm;
therefore,
it
is
important
to
understand
the
significance
of
such
residues.
If
there
is
no
consistency
in
association,
causality
cannot
be
confirmed
and
must
be
rejected.
California
and
New
York
incident
data
were
analyzed
by
the
RRTF
utilizing
a
threshold
of
0.7
ppm
brodifacoum
(
and
possibly
other
anticoagulants)
in
liver.
Applying
this
threshold
to
the
data
from
both
states
(
which
is
primarily
for
brodifacoum)
indicates
that
approximately
two­
thirds
of
all
incidents
with
residue
data
are
below
0.7
ppm
in
the
liver.
One­
third
of
incidents
had
reported
liver
residues
below
0.1
ppm.
The
predominance
of
low­
level
residues
in
mortality
reports
emphasizes
the
importance
of
accurate
interpretation
of
their
significance.
This
merits
a
balanced
discussion.

There
are
numerous
reports
in
the
literature
and
by
state
agencies
that
document
measurable
liver
residues
of
brodifacoum
and
other
anticoagulants
in
perfectly
healthy
feral
animals.
In
the
analysis
of
ten
coyotes,
the
conclusion
of
an
unpublished
California
Department
of
Fish
and
Game
(
CDFG)
report
was:
"
the
residue
concentrations
in
these
otherwise
healthy
animals
may
suggest
that
background
levels
of
anticoagulant
rodenticides
are
found
in
urban
carnivores "
(
Table
81a)
(
p­
2051,
Hosea,
1999).
In
other
incident
reports
by
CDFG,
however,
lower
level
residues
of
second­
generation
anticoagulants
are
cited
as
diagnostic
of
the
anticoagulant
as
the
causative
agent
of
observed
mortality
(
Hosea,
1999).
These
inconsistencies
demonstrate
the
difficulty
of
ascribing
causality
in
these
cases,
and
the
value
of
agreed
protocols
for
pathology
and
chemical
analysis
(
Brown
et
al.,
1996).
Detection
of
low­
level
residues
may
represent
Page
RRTF
comment
25
the
slow
terminal
phase
of
clearance
with
residues
sequestered
in
the
liver,
and
must
be
carefully
interpreted
with
respect
to
any
forensic,
diagnostic,
or
toxicological
significance.
Long­
term
anticoagulant
feeding
studies
in
rats,
such
as
with
diphacinone,
for
example,
failed
to
find
consistent
effects
on
clotting
times
or
general
health
and
feeding
behavior
at
levels
of
0.03
to
0.5
ppm
over
90
days
of
continuous
feeding
(
Elias
and
Johns,
1981).

All
of
the
animals
were
free
of
any
apparent
trauma
or
disease,
and
necropsy
revealed
no
evidence
of
hemorrhage
(
other
than
one
hematoma
caused
by
the
administration
of
the
lethal
injection).
All
5
of
these
animals
carried
residues
of
brodifacoum
in
the
liver
and
4
of
the
5
carried
multiple
anticoagulant
residues
(
Table
81b).
It
is
apparent
that
liver
levels
of
brodifacoum
characterized
in
many
wildlife
reports
as
diagnostic
of
toxicity
and
fatality
are
also
found
as
background
levels
in
the
livers
of
healthy
wildlife.

Finally,
low­
level
residues
of
anticoagulants
are
often
used,
regardless
of
the
magnitude
of
the
residue,
to
confirm
pathological
observations.
In
combination,
these
low­
level
residues
and
pathology
cannot
be
used
to
determine
that
an
anticoagulant
rodenticide
is
the
causative
agent.
Pathology
is
often
the
primary
criteria
in
wildlife
incident
reports
used
to
conclude
anticoagulant
toxicity.
Although
the
lesions
observed
in
incident
reports
may
be
indicative
with
anticoagulant
toxicity,
they
are
general
and
not
diagnostic.
Pathology,
necropsy,
and
clinical
signs
of
toxicity
following
anticoagulant
exposure
reported
in
published
literature
were
compared
by
the
RRTF
to
the
information
in
the
EIIS
wildlife
incident
reports
(
Berny
et
al.,
1997;
DuVall
et
al.,
1989;
Elias
and
Johns
et
al.,
1981;
Gray
et
al.,
1994;
Hegdal
and
Colvin,
1988;
Huckel
et
al.,
1988;
Meehan,
1984;
Newton
et
al.,
1990;
Rammell
et
al.,
1984).
The
descriptions
of
anticoagulant
toxicity
in
controlled
studies
were
for
the
most
part
general.
These
descriptions
include
external
hemorrhage
and
internal
hemorrhage
in
a
number
of
organs,
including
brain,
kidney,
lungs,
heart,
and
gut.
Major
organs,
including
the
liver,
may
exhibit
diffuse
pallor.
First
signs
often
include
bloody
diarrhea
or
urine.
A
number
of
articles
cautioned
that
care
must
be
taken
in
diagnosing
anticoagulant
poisoning
both
because
obvious
symptoms
may
be
lacking
and
not
every
hemorrhagic
lesion
denotes
anticoagulant
poisoning.
Other
causes
of
coagulopathy
noted
in
these
articles
include:
infectious
canine
hepatitis,
hemorrhagic
disease
of
pigs,
cows,
and
chicks,
heat
stroke,
aflatoxicosis,
vitamin
K
deficient
diet,
trauma,
inherited
clotting
factor
deficiencies,
and
consumption
of
naturally
occurring
anticoagulants
(
e.
g.,
dicumarol
in
sweet
vernal
hay).

In
summary,
pathologic
observations
should
be
used
as
secondary
indicators
of
anticoagulant
toxicity
and
not
in
combination
with
low­
level
anticoagulant
residues.
Although
they
may
be
indicative,
they
are
not
diagnostic.
There
are
other
causes
of
these
generic
types
of
lesions.
Page
RRTF
comment
26
EFED
response:
As
previously
stated,
the
issue
of
low­
levels
of
residue
will
be
addressed
through
a
data
call­
in.

82
Correction.
The
reference
Savarie
et
al.,
1979
included
oral
doses
of
0.63,
1.25,
2.5,
5,
and
19
mg
a.
i./
kg.
Table
44
lists
the
doses
as
0.63,
1.25,
2.5,
5,
and
10
mg
a.
i./
kg
EFED
response:
The
doses
listed
in
the
table
are
correct.
The
doses
are
not
cited
in
the
text
in
the
revised
risk
assessment.

83
Footnote
for
Dicoumarol
is
missing.
This
product
is
not
registered
in
the
United
States
as
a
rodenticide
EFED
response:
Footnotes
are
used
only
in
the
tables,
not
in
the
text.
Dicoumarol
is
an
anticoagulant
compound.

87,
Table
46
Errors
in
EPA's
calculations
(
Table
46).
EPA
has
made
a
significant
calculation
error
when
deriving
the
summary
values
for
primary
risks
to
both
birds
and
non­
target
mammals.
EPA
incorrectly
calculated
the
weighted
average
values
for
the
following
measures
of
effects:
1)
grams
of
bait
needed
for
a
bird
LD50;
and
2)
grams
of
bait
needed
for
a
non­
target
mammal
LD50.
Values
were
indexed
to
the
least
toxic
rodenticides
rather
than
the
most
toxic
ones.
This
error
changes
the
overall
rankings
of
the
nine
rodenticides,
as
well
as
the
magnitude
and
spread
of
the
summary
values
among
the
nine
products.
The
correct
summary
values
are
presented
in
Table
5
(
Note:
this
is
a
revised
Table
46
from
the
PRA).
Based
on
the
new
overall
summary
values,
brodifacoum
is
still
ranked
first
(
i.
e.,
"
most
hazardous").
Zinc
phosphide,
formerly
fifth,
is
now
ranked
second,
although
its
summary
value
is
almost
the
same
as
brodifacoum.
The
two
diphacinone
baits
are
now
ranked
fourth
and
fifth
overall
as
opposed
to
formerly
being
ranked
as
third
and
sixth,
and
their
overall
summary
values
are
lower.
The
two
chlorophacinone
baits
are
now
ranked
eighth
and
ninth
overall
versus
previous
rankings
of
ninth
and
eleventh,
but
again
the
relative
numbers
and
differences
are
lower
than
before,
indicating
less
overall
hazard
than
previously
expressed.

EFED
response:
The
necessary
changes
have
been
made.

88
Figure
5.
The
same
correction
as
previous
figures.

EFED
response:
The
necessary
changes
have
been
made.

89
Figure
6.
The
same
correction
as
previous
figures.

EFED
response:
The
necessary
changes
have
been
made.
Page
RRTF
comment
7
Ibid.

27
90,
Figure
7
Flaw
in
sensitivity
analysis.
Because
several
of
the
measures
of
effect
were
significantly
correlated,
the
sensitivity
analysis
would
not
be
expected
to
show
differences
in
rankings
when
values
for
the
measures
of
effect
were
varied.

EFED
response:
EFED
disagrees
that
several
the
measures
of
effect
are
in
fact
correlated.
Specifically,
the
two
measures
of
effect
for
primary
risk
to
birds
and
the
retention
times
for
blood
and
liver
were
tested
for
correlation
using
the
`
Correlation
and
Regression
Calculator'
at
http://
www.
ebook.
stat.
ucla.
edu/
cgibin
php.
cgi/
calculators/
correlation.
phtml,
and
the
correlation
coefficients
were
0.105801
and
0.272307,
respectively,
indicating
little
linear
correlation.
7
91
Distinctions
between
50
and
100
ppm
baits.
EPA
states
that
distinctions
cannot
be
made
between
50
ppm
and
100
ppm
chlorophacinone
and
diphacinone
baits
using
the
incident
data,
"
but
the
100
ppm
baits
are
likely
to
present
greater
risk
than
50
ppm
baits."
This
may
seem
like
an
obvious
statement,
but
in
fact
it
may
not
be
true
because
of
differences
in
the
formulations
and
use
patterns
between
the
baits
containing
100
ppm
and
50
ppm
of
these
active
ingredients.
It
is
not
correct
to
assume
that
they
are
used
interchangeably.
Some
pelleted
baits
containing
50
ppm
active
ingredient
are
used
"
in
and
around"
homes
for
commensal
control
while
other
50
ppm
grain­
based
baits
are
used
in
bait
stations
for
control
of
ground
squirrels
and
other
field
rodents.
The
100
ppm
baits
are
grain­
based
and
only
used
in
agricultural
settings
for
control
of
ground
squirrels
and
field
rodents.
The
100
ppm
baits
are
applied
by
broadcast
methods
(
mechanical
or
hand)
and
are
not
used
in
bait
stations.
Secondary
risks
to
birds
and
non­
target
mammals
are
dependent,
in
part,
on
residues
in
the
target
species
and
could
potentially
be
higher
for
50
ppm
baits
because
of
greater
bait
availability
in
bait
stations
and
many
other
factors.

EFED
response:
EFED
believes
that
this
discussion
actually
argues
that
the
100
ppm
baits
are
likely
to
present
greater
risk
to
non­
target
organisms
than
the
50
ppm
baits.
In
addition,
it
seems
counter­
intuitive
to
argue
that
bait
stations
would
present
greater
availability
to
non­
target
organisms
than
broadcast
applications
without
some
supporting
data.

91
Use
of
6­
g
pellets
for
ground
squirrel
control.
Ground
squirrels
are
selective
feeders
much
of
the
year
and
bait
acceptance
can
greatly
limit
the
time
available
for
applications.
Regardless
of
size,
use
of
pellets
instead
of
grains
for
spot
and
broadcast
baiting
will
likely
reduce
bait
acceptability
and
efficacy
from
the
current
formulations.
More
importantly,
concentrating
the
amounts
of
active
ingredients
through
use
of
large
pellets
may
increase
the
potential
primary
and
secondary
risks
to
non­
target
mammals,
thus
offsetting
the
potential
benefit
of
reduced
risks
to
birds.
Risk
to
non­
target
mammals
Page
RRTF
comment
28
could
be
increased
for
several
reasons.
Use
of
larger
baits
will
make
it
easier
for
non­
target
mammals,
such
as
coyotes
and
Kit
foxes,
to
obtain
a
lethal
dose
through
direct
ingestion
of
pellets.

EFED
response:
This
discussion
has
been
deleted
from
the
document.

93
Use
pattern
and
market
share.
The
Decision
Table
Analysis
ranks
the
relative
hazard
of
the
different
active
ingredients,
but
does
not
estimate
exposure,
without
which
risk
cannot
be
estimated.
The
assumption
of
equal
exposure
is
totally
inappropriate
considering
the
divergent
use
patterns
of
products
included
in
this
analysis
and
the
high
market
share
held
by
certain
active
ingredients.
Use
pattern
is
a
key
factor
in
any
pesticide
risk
assessment
and
the
exposure
is
use
pattern
and
often
chemical­
specific.
Brodifacoum
(
50
ppm),
bromadiolone
(
50
ppm),
difethialone
(
25
ppm),
diphacinone
(
50
ppm),
chlorophacinone
(
50
ppm),
bromethalin
(
100
ppm),
cholecalciferol
(
750
ppm),
warfarin
(
250
ppm),
and
zinc
phosphide
(
20,000
ppm)
baits[
1]
are
all
registered
for
"
commensal
uses"
in
the
U.
S.
Commensal
use
is
defined
as
"
in
and
around
buildings,
transport
vehicles
and
other
manmade
structures."
Commensal
rodents
exist
because
man
has
provided
highly
desirable
conditions
for
them
to
do
so
(
i.
e.,
structures
which
provide
food,
water,
and/
or
harborage).
In
the
absence
of
control
measures,
commensal
rodent
populations
will
escalate
because
the
highly
favorable
environment
provided
by
man
is
not
balanced
by
the
rodents'
natural
predators.
Farmers,
consumers,
and
professional
exterminators
for
the
protection
of
health
and
property
from
commensal
rodents
use
commensal
use
rodenticides.
Commensal
rodents
typically
include
the
house
mouse,
the
Norway
rat,
and
the
roof
rat.
In
some
instances,
other
rodent
species,
e.
g.,
the
deer
mouse,
can
become
commensal
(
i.
e.,
invade
structures).
Certain
rodenticides
are
also
approved
for
field
uses.
Field
use
constitutes
use
against
rodents
living
"
in
the
field,"
i.
e.,
not
associated
with
man­
made
structures.
Most
typically
these
rodenticides
are
used
for
crop
protection,
but
can
also
be
used
against
public
health
pests
(
e.
g.,
California
ground
squirrel
control
for
plague
prevention).
Zinc
phosphide
bait
(
typically
20,000
ppm)
is
the
most
widely
used,
being
federally
registered
for
a
range
of
uses
against
a
comparatively
broad
range
of
rodent
and
related
pests.
Some
Section
24(
c)
registrations
exist
for
diphacinone
and
chlorophacinone
and
also
for
some
non­
federally
registered
uses
of
zinc
phosphide.
Warfarin
was
recently
approved
for
use
against
moles
below
ground.
A
below
ground
use
diphacinone
pocket
gopher
bait
was
federally
registered,
but
it
is
unclear
whether
this
registration
is
still
active.
There
was
also
a
24(
c)
for
cholecalciferol
that
is
inactive.
A
few
highly
specialized
uses
also
exist
for
certain
products
for
the
purpose
of
natural
ecosystem
restoration.
Brodifacoum
has
been
used
on
uninhabited
islands
in
the
U.
S.
and
elsewhere
to
remove
non­
native
rats
(
arriving
originally
by
ship)
that
predate
and
significantly
endanger
local
fauna,
typically
birds.
Diphacinone
is
used
in
Hawaii
for
controlling
mongoose
and
rats
that
predate
native
birds.
These
uses
are
highly
regulated,
being
carried
out
by
government
personnel
only,
and
constitute
an
extremely
small
proportion
of
overall
rodenticide
use.
There
are
Page
RRTF
comment
29
no
other
field
uses
approved
in
the
U.
S.
for
brodifacoum,
and
no
field
uses
at
all
for
bromadiolone,
difethialone,
or
bromethalin.

EFED
response:
As
previously
noted,
the
risk
assessment
is
based
on
the
available
data.
Registrants
have
not
submitted
the
data
that
would
be
needed
to
assess
the
probability
of
exposure.
These
data
have
been
outlined
in
a
section
on
Uncertainty
and
Data
Needs.
in
the
revised
assessment.

96
Incorrect
term.
The
term
PCO's
is
used
and
this
term
is
not
defined
and
is
incorrect.
See
discussion
above.

EFED
response:
Previously
addressed.

127
&
128
Missing
data.
Where
no
data
were
available,
the
specific
measure
of
effect
was
not
included
in
the
analysis
for
that
particular
active
ingredient.
This
causes
an
over
weighting
of
data
for
those
measures
of
effect
where
data
were
available.

EFED
response:
Missing
data
does
add
uncertainty
to
the
results
of
the
assessment.
This
is
acknowledged
and
the
data
needed
to
refine
this
assessment
are
presented
in
a
section
on
Uncertainty
and
Data
Needs.
In
addition,
many
of
the
studies
required
in
the
Rodenticide
Cluster
Reregistration
Eligibility
Decision
(
RED)
have
not
yet
been
submitted,
even
though
the
RED
was
issued
in
July,
1998.

134
Correct
calculation
error.
Attachment
1,
Results
of
the
Comparative
Analysis,
Step
3,
Substep
B,
the
entry
for
Bromethalin
100
mg
should
read
((
400.0­
2.30)/
400)*
5=
4.97,
the
LD50
for
Bromethalin
from
Table
1
is
2.30
EFED
response:
The
necessary
changes
have
been
made.

135
Correct
calculation
error.
Attachment
1,
Results
of
the
Comparative
Analysis,
Step
4,
the
entry
for
Bromethalin
100
mg
should
read
0.04+
4.97=
5.01,
the
sum
from
Step
3,
Substep
A,
and
Substep
B.

EFED
response:
The
necessary
correction
has
been
made.

135
Correct
calculation
error.
Attachment
1,
Results
of
the
Comparative
Analysis,
Step
4,
the
entry
for
Diphacinone
100
mg
should
read
0.10+
2.50=
2.60,
a
simple
math
computation
error,
this
may
or
may
not
have
effects
on
the
overall
hazard
assessment.

EFED
response:
The
sum
of
the
weighted
average
values
for
Diphacinone
100
mg
should
be
0.01+
2.50=
2.51.
No
change
was
made.
Page
RRTF
comment
30
147
HD5
data.
The
method
used
to
extrapolate
the
HD5
(
50%)
from
one
bait
concentration
to
another
is
not
appropriate
as
it
does
not
take
into
consideration
the
slope
of
the
dose­
response
relationship
for
the
active
ingredient.
For
example,
reducing
the
concentration
of
active
ingredient
by
50%
will
not
necessarily
reduce
toxicity
by
50%
depending
on
the
slope
of
the
dose­
response
relationship.

EFED
response:
The
extrapolation
from
one
bait
concentration
to
another
does
depend
upon
the
slope.
However,
since
slope
information
was
not
available,
the
assumption
is
that
the
slope
is
consistent
with
a
50%
reduction
in
toxicity
when
the
concentration
of
the
active
ingredient
is
reduced
50%.

151
Correct
reference
to
Table
42.
Total
incidents
of
271
does
not
match
Table
42
summary.

EFED
response:
The
number
of
incidents
has
been
updated
in
the
revised
risk
assessment.

153­
155
Graphs
9,
10,
11
Lack
of
correlation.
These
plots
do
not
show
a
strong
relationship
between
summary
risk
values
and
the
number
of
incidents,
suggesting
that
the
two
are
not
highly
correlated
and
that
EPA's
measures
of
effect
may
not
be
good
predictors
of
incidents.
This
should
not
be
surprising
since
EPA's
analysis
did
not
account
for
exposure
and
product­
specific
use
pattern
differences,
whereas
the
incident
data
better
reflect
actual
exposure,
including
factors
such
as
market
share.
Note
that
the
relationships
in
the
graphs
will
become
even
weaker
once
the
"
corrected"
summary
values
in
Table
46
are
plotted
against
the
number
of
incidents.
Note
also
that
if
data
for
brodifacoum
are
removed
from
the
graphs,
the
data
become
an
almost
random
scatter
gram
with
no
predictive
power.

EFED
response:
This
graph
was
not
meant
to
show
an
overall
correlation
between
summary
risk
values
and
the
number
of
incidents.
Rather,
the
graphs
show
that
the
rodenticide
baits
with
the
greatest
number
of
reported
incidents
and
the
largest
summary
risk
values
should
appear
in
the
upper
left
of
the
graph.
In
all
three
graphs
brodifacoum
is
the
only
bait
to
appear
in
this
position.
Thus,
the
graph
confirms
that
brodifacoum
poses
the
greatest
overall
potential
risk
to
birds.
The
corrected
summary
values
do
not
significantly
weaken
this
confirmation.
