Page
1
Memorandum
To:
Debbi
Hart
and
Lynne
Tudor,
U.
S.
EPA
From:
Dave
Cacela
and
Liz
Strange,
Stratus
Consulting
Inc.

Date:
2/
13/
2004
Subject:
Summary
of
Peer
Reviews
of
I&
E
Methods
Chapter
A5
1.
Introduction
Chapter
A5
describes
the
methods
used
by
EPA
to
evaluate
I&
E.
Professor
Gary
Grossman,
Professor
Pete
Raimondi,
and
Dr.
Shuter
are
fisheries
scientists
who
were
contracted
by
EPA
(
through
Versar,
Inc.)
to
critique
the
chapter
with
respect
to
several
specific
topic
areas.
This
memorandum
summarizes
the
reviewers'
main
comments
and
provides
responses
to
those
comments.
Some
of
the
reviewers'
criticisms
are
no
longer
pertinent
to
the
rulemaking
because
EPA
has
made
changes
to
Chapter
A5
since
the
time
that
the
three
reviews
were
conducted.

2.
Critique
by
Professor
Gary
Grossman
A
criticism
expressed
by
Professor
Grossman
concerns
the
possibility
of
incorporating
stochasticity
in
the
survival
modeling
that
informs
the
estimation
of
foregone
fishery
yield.
Professor
Grossman
suggests
that
the
assessment
would
be
improved
if
environmental
stochasticity
was
explicitly
included
in
the
model.
He
does
not
suggest
exactly
how
it
should
be
included,
so
it
is
not
possible
to
comment
on
the
feasibility
or
utility
of
adding
a
particular
kind
of
stochasticity.
EPA
acknowledges
that
fish
populations
are
affected
by
ecological
processes
that
cannot
be
well­
described
except
in
stochastic
terms.
However,
EPA
maintains
that
an
adequate
assessment
of
the
economic
benefits
of
the
316(
b)
rulemaking
does
not
require
a
highly
detailed,
highly
realistic
ecological
model.
The
practical
requirements
of
developing
models
along
the
lines
suggested
by
Professor
Grossman
are
quite
substantial,
even
for
modeling
the
dynamics
of
a
single
fish
stock.
Inclusion
of
stochasticity
in
a
model
does
not
automatically
make
the
model
better
or
more
useful
unless
the
nature
of
the
stochasticity
is
correctly
formulated.
The
limitations
in
this
regard
are
not
simply
computing
power,
as
Professor
Grossman
suggests,
but
sufficient
understanding
of
the
systems
in
question.
The
benefits
assessment
is
addressed
at
estimating
effects
on
hundreds
of
stocks
nationwide
and
inclusion
of
the
stochastic
details
specific
to
each
of
these
stocks
is
not
a
realistic
goal.
Stratus
Consulting
2/
13/
04
Page
2
Although
EPA
has
not
included
stochasticity
within
the
I&
E
model
itself,
EPA
has
employed
stochasticity
in
a
parallel
investigation
of
uncertainty
associated
with
the
I&
E
modeling
procedure
(
see
Chapter
A6
of
the
Regional
Analysis
Document).
The
investigation
uses
Monte
Carlo
methods
to
handle
plausible
degrees
of
uncertainty
associated
with
all
of
the
numerous
parameters
in
the
model.
The
investigation
is
not
complete,
due
in
part
to
uncertainty
about
selection
of
probability
distributions.

Professor
Grossman
states
that
the
absence
of
reporting
about
stock
sizes
is
a
problem.
The
concern
is
apparently
motivated
by
a
belief
that
estimating
population
impacts
is
the
objective
of
the
assessment,
but
that
is
not
the
case.
EPA's
core
I&
E
modeling
approach
is
to
use
a
static
set
of
population
parameters
to
generate
simple
point
estimates
of
foregone
yield.
EPA
has
intentionally
generated
benefits
estimates
that
are
linked
directly
to
the
estimates
of
I&
E
losses,
not
indirectly
through
estimates
of
population
level
impacts.
This
approach
is
intended
to
provide
reasonable
estimates
of
the
general
magnitude
of
losses
associated
with
I&
E
mortality,
and
not
to
develop
realistic
population
dynamic
models.
Projections
of
future
fish
population
dynamics
is
simply
out
of
the
scope
of
the
benefits
assessment.

Another
criticism
expressed
by
Professor
Grossman
concerns
the
regional
extrapolation
of
loss
estimates
at
model
facilities
to
other
facilities.
Professor
Grossman
indicates
that
extrapolations
are
invalid
unless
conditions
at
unmodeled
facilities
are
identical
to
conditions
at
model
facilities.
EPA
acknowledges
that
the
extrapolations
introduce
additional
uncertainty
into
the
aggregate
estimates.
However,
EPA
maintains
that
this
is
unavoidable
unless
every
in­
scope
facility
were
modeled
independently.
In
order
to
avoid
the
most
significant
sources
of
bias,
EPA
applied
extrapolations
within
particular
regions
prior
to
aggregation
on
a
national
scale.

Professor
Grossman
expresses
valid
concerns
about
the
methods
used
to
generate
annualized
I&
E
loss
estimates
at
individual
model
facilities.
EPA
acknowledges
that
numerous
facility­
specific
details,
including
details
about
the
I&
E
sampling
programs
and
regional
ecological
conditions,
affect
the
validity
of
the
annualized
loss
estimates.
However,
EPA's
benefits
assessment
did
not
include
development
of
the
original
I&
E
monitoring
procedures.
EPA
relied
upon
the
loss
estimates
as
reported
by
the
facilities,
based
on
the
premise
that
the
original
investigators
had
the
necessary
understanding
of
the
facility
and
the
local
ecological
conditions
to
develop
appropriate
sampling
procedures
and
interpretations
of
the
I&
E
data.

Professor
Grossman
questions
EPA's
treatment
of
I&
E
losses
with
respect
the
presence
of
older
fish
(>
1
year
old)
among
the
I&
E
losses.
This
concern
has
been
raised
by
other
commenters,
and
EPA
has
revised
the
modeling
procedures
by
eliminating
the
assumption
that
impinged
fish
are
all
age
1.
The
revised
procedure
uses
an
assumption
that
impinged
fish
include
individuals
ranging
from
age
0
to
age
5.
Stratus
Consulting
2/
13/
04
Page
3
3.
Critique
by
Professor
Pete
Raimondi
Professor
Raimondi
has
criticized
EPA's
model
in
very
general
terms
that
relate
to
his
beliefs
that
(
a)
the
benefits
assessment
has
too
much
of
a
"
fisheries
dependent
mindset"
and
a
"
fisheries
bias,"
and
(
b)
the
correct
approach
to
the
assessment
employs
an
empirical
transport
model
(
ETM),
rather
than
the
foregone
yield
model
that
EPA
used.

EPA
acknowledges
that
the
benefits
assessment
focuses
on
foregone
use
values
of
fish
killed
by
I&
E.
Professor
Raimondi
indicates
that
EPA
has
ignored
values
associated
with
"
ecosystem
function."
EPA
interprets
this
class
of
values
to
be
among
those
commonly
known
as
non­
use
values,
which
include
a
variety
of
other
societal
values
in
addition
to
roles
in
ecosystem
functioning.
EPA
has
had
difficulty
incorporating
non­
use
values
into
the
assessment
due
to
a
lack
of
relevant
information
about
how
to
monetize
the
non­
use
values
of
I&
E
losses.

Despite
Professor
Raimondi's
claim
that
EPA's
approach
is
"
at
least
one
generation
behind"
current
I&
E
assessment
methods,
EPA
disagrees
with
his
assertion
that
ETM
is
a
preferable
method
for
conducting
the
benefits
assessment.
The
ETM
method
enjoys
certain
aspects
of
ecological
realism
and
significance
for
highly
site
specific
ecological
assessments,
but
it
is
not
well
suited
to
the
purpose
of
the
316(
b)
national
benefits
assessment.
The
advantages
of
ETM
that
Professor
Raimondi
alleges
have
no
apparent
bearing
on
the
benefits
assessment
because
the
results
of
an
ETM
assessment,
as
Professor
Raimondi
spells
out,
are
in
terms
of
foregone
equivalent
reproductive
effort
expressed
as
acreage,
mileage,
or
a
proportion
of
the
areal
extent
of
spawning
for
fishes
exposed
to
I&
E
at
a
particular
facility.
EPA
maintains
that
losses
expressed
in
this
manner
are
much
less
amenable
to
monetization
than
the
direct
approach
to
estimate
foregone
yield.
Moreover,
implementation
of
an
ETM
is
a
highly
site­
specific
exercise
that
requires
intimate
knowledge
of
both
(
a)
the
hydrology
of
a
particular
system,
and
(
b)
the
actual
spawning
habits
of
every
species
exposed
to
I&
E.
The
large
information
requirements
of
ETM
modeling
combined
with
the
off­
point
nature
of
the
results
of
the
modeling
are
clear
reasons
why
EPA
did
not
elect
to
use
ETM
modeling
in
the
assessment.

4.
Critique
by
Dr.
Shuter
Dr.
Shuter's
comments
were
generally
favorable
with
respect
to
the
appropriateness
of
EPA's
modeling.
EPA
agrees
with
Dr.
Shuter's
suggestion
that
the
chapter
would
benefit
from
a
clearer
statement
about
the
objectives
of
the
modeling
and
expectations
about
how
the
results
of
the
model
can
be
interpreted,
and
revised
the
chapter
accordingly.
A
large
part
of
his
criticisms
Stratus
Consulting
2/
13/
04
Page
4
concern
the
actual
structure
of
the
chapter
and
the
fact
that
a
number
of
relevant
facts
required
to
implement
the
model
are
physically
located
in
inconvenient
places,
such
as
in
an
unattached
appendix.
EPA
believes
that
some
of
the
practical
difficulties
noted
by
Dr.
Shuter
would
be
absent
for
readers
that
have
the
complete
assessment
in
hand,
rather
than
simply
Chapter
A5
in
isolation.

Another
general
concern
expressed
by
Dr.
Shuter
concerns
the
level
of
detail
provided
about
the
modeling
procedures,
which
he
perceived
as
being
too
general
for
rigorous
assessment.
EPA
has
revised
Chapter
A5
in
response
to
these
comments,
and
believes
that
the
current
version
of
the
chapter
is
now
more
complete.

Dr.
Shuter
expresses
concern
about
how
life
history
parameters
(
e.
g.,
survival
rates)
were
identified
and/
or
generated.
The
majority
of
the
survival
rates
that
EPA
employed
were
derived
from
literature
sources
that
are
fully
referenced
in
the
assessment
report.
EPA
acknowledges
that
there
are
numerous
difficulties
associated
with
applying
survival
rates
reported
in
published
records
to
novel
situations,
some
of
which
may
be
quite
influential
in
the
benefits
assessment.
However,
EPA
believes
that
it
has
been
completely
transparent
in
this
regard
by
publishing
all
of
the
values
that
were
used
and
their
original
sources.
Part
of
Dr.
Shuter's
concern
in
this
regard
may
relate
to
the
wording
used
in
earlier
versions
of
the
chapter
that
unfortunately
overstated
the
role
of
the
"
Fa
method"
in
identifying
survival
rates.
This
wording
made
it
appear
that
the
Fa
method
was
used
as
a
primary
tool
and
that
reference
values
for
the
majority
of
survival
rates
could
not
be
identified.
In
fact,
EPA
identified
reference
values
for
the
vast
majority
of
survival
rates
employed,
and
used
the
Fa
method
only
as
a
last
resort.
EPA
acknowledges
that
substantial
uncertainties
remain
about
proper
survival
rates
(
with
respect
to
both
M
and
F)
and
that
uncertainty
in
these
values
causes
significant
portion
of
the
uncertainty
in
the
benefits
assessment.

Dr.
Shuter
raises
some
interesting
conceptual
questions
about
the
relationships
among
I&
E
losses,
the
mortality
rates
M
and
F,
and
the
possible
decomposition
of
M
into
components
that
explicitly
describe
M
attributable
to
I&
E.
Dr.
Shuter
also
identifies
relationships
between
interpretation
of
these
concepts
and
EPA's
use
of
static
rather
than
dynamic
values
for
these
parameters.
In
particular,
Dr.
Shuter
questions
the
implications
of
a
preconceived
notion
that
the
relative
magnitude
of
I&
E
mortality
is
small
compared
to
other
forces
of
mortality.
EPA
acknowledges
that
these
are
valid
concerns.
However,
EPA's
statements
that
I&
E
mortality
is
minor
are
not
intended
to
be
a
value
judgement,
nor
a
presumption
that
they
can
be
ignored
out
of
hand.
Rather,
they
are
intended
to
indicate
that,
as
a
general
rule,
I&
E
losses
are
not
expected
to
be
dominant
driving
forces
in
the
dynamics
of
most
populations.
If
this
presumption
is
valid,
then
the
commonly
perceived
need
to
proceed
directly
to
sophisticated,
population­
specific
fisheries
models
that
include
compensation
and
other
kinds
of
dynamics
may
be
discounted.
EPA's
philosophy
in
this
regard
is
that
it
is
legitimate
to
use
the
direct
foregone
yield
approach
because
Stratus
Consulting
2/
13/
04
Page
5
these
losses
are
"
minor"
enough
to
be
considered
simply
as
incremental
differences
that
occur
concurrently
and
additively
with
other
more
dramatic
forces
that
drive
large
scale
population
changes.
Thus,
conceiving
of
these
losses
as
"
relatively
minor"
refers
only
to
importance
within
the
suite
of
other
population
drivers,
but
not
necessarily
to
importance
in
economic
terms.
The
value
of
fisheries
on
a
national
scale
is
tremendous,
so
a
(
possibly)
"
minor"
reduction
of
that
resource
could
actually
be
very
large
in
absolute
terms.

Dr.
Shuter
noted
some
inconsistencies
in
notation
with
regard
to
the
terms
"
ages"
and
stages."
EPA
regrets
the
inconsistency
and
possible
confusion
resulting
from
it.
In
fact,
EPA's
models
are
best
described
as
"
stage
specific"
throughout
the
assessment.
The
confusion
arose
because
the
actual
stages
that
EPA
employed
correspond
directly
to
"
ages"
for
all
fish
age
1
and
older,
i.
e.,
adulthood
was
not
modeled
as
a
single
stage
but
as
multiple
stages
corresponding
to
years.
