Notes and Discussion Summaries from the
October 27th CRI Workshop entitled:
Data-poor, rapid analysis assessments for other ESU's
in the Columbia River System
9:30-10:30 EDT/CRI
Peter Kareiva of NWFSC presented the CRI model. He opened with the idea
that EDT and CRI are not necessarily dueling models, rather they work
at different levels, using different approaches, and they can complement
one another. He then summarized the CRI approach. Since CRI uses quantitative
methods, it can be limited by a lack of data, whereas EDT may be able
to pursue some issues further by using expert opinion in the absence of
data.
Chip McConnaha from NWPPC summarized the Ecosystem Diagnosis
and Treatment (EDT) model. EDT is an explanatory model rather than quantitative
or statistical model.
The presentations slides are also available (in PowerPoint only):
10:20-11:10 Beginning the analyses of Puget Sound
management options and risks
Beth Sanderson (CRI of NWFSC) presentation:
Methods. The analyses presented by Beth Sanderson
used a simple extinction risk model (Dennis et al. 1991) to illustrate
the importance of correctly identifying populations. The data used in
the model were abundance time series from Puget Sound chinook stocks.
The abundance data were provided by WDFW and tribal entities, and are
maintained and updated as part of a coastwide chinook salmon database
by the NWFSC. To explore the effects of combining abundance data at different
geographic scales on estimates of extinction risk, the time series data
were analyzed at two different spatial scales: (1) as individual SASSI
stocks (WDF et al. 1993), used by comanagers to set escapement levels
and to help set hatchery broodstock policies, and (2) as harvest management
units, used by comanagers to set harvest rates and to predict exploitation
rates. The small (SASSI stock) and large (harvest management units) units
for analyses were purposefully chosen for the purposes of illustration
because they are spatial scales actually used by comanagers to make management
decisions regarding chinook salmon. The true population sizes for Puget
Sound chinook salmon are not yet known; they may in some cases be similar
to SASSI stocks, in other watersheds, populations may be closer in size
to harvest management units, or they may be somewhere in between those
scales. The point of the analyses presented by Beth was to illustrate
how defining the spatial scale chosen to represent a population has a
big effect on the estimated viability of that population. In other words,
the step of correctly identifying populations of salmonids is critical
for correctly estimating their risk of extinction.
The Puget Sound chinook salmon data were analyzed as separate
SASSI stocks and combined into harvest management units using the Dennis
et al. (1991) model. For individual and combined series, only those data
years that were represented in all of the constituent streams in a harvest
management unit were included in the analyses of single and combined groups.
The quasi-extinction threshold we used was the probability that a group
of fish would decline to one fish in a single year. The starting population
size for the model was set at N = 500 fish, and the likelihood of extinction
was based on a population projection 100 years into the future. Biologists
at the NWFSC are exploring other means of estimating extinction probabilities
by incorporating the effects of age structure into the Dennis et al. (1991)
model. Those results are not reported in Beth's talk, but the main conclusion-that
how one combines streams has a large effect on estimates of extinction
risk-does not change.
To look at the sensitivity of the effects of spatial scale
on viability predictions, the analyses Beth presented also included two
additional experiments with the data. First, the effect of having data
from fewer streams within a harvest management unit was explored by systematically
excluding data from one stream at a time from the extinction risk analyses
on the harvest management units. Second, the effect of which time interval
was used in estimating viability was estimated by performing extinction
risk analyses on complete data series (1968 - 1997) and on those same
data series shortened to include only those years from 1980 - 1997. The
more complete analysis, asking what the effect of short vs. long time
series is on viability estimates, has recently been completed, but those
results were not presented in Beth's talk.
More detailed analyses including the effects of age structure
and the length of time series on estimates of the viability of chinook
salmon populations will be published in a paper to be sent out for public
review soon. The results from the preliminary analyses described above
are contained in the Power Point presentation link below.
Discussion:
The point was re-emphasized that the absolute extinction risks are not
the focus of our results, they are simply used to illustrate how viability
estimates are affected by the way in which streams are combined for analyses.
Beth emphasized that the purpose of the analyses she presented was to
argue for the primary importance of correctly identifying population boundaries
for estimating risks for salmonids.
11:10-12:00 What will make Puget Sound recovery science
most effective from a policy perspective Walt Reid (coordinator,
Puget Sound Salmon Collaboration) is trying to achieve better coordination
on the policy side. He summarized some of the main points of discussion
from the Puget Sound Salmon Leaders Workshop.
Discussion:
There was criticism that his ideas were overly simplified. For example,
the point was brought up that productivity must also be considered rather
than the absolute numbers of returning adult spawners. Some one else argued
that scientists must also consider fish diversity (spatially and genetically)
to determine health and viability.
Walt agreed that other factors like spatial distribution
and genetic diversity must be considered to establish de-listing criteria.
As with escapements, these criteria can be set based strictly on science.
Scientists, however, should avoid establishing delisting criteria that
involve policy judgments, such as "how much harvest is enough" or how
much "habitat restoration is enough." Scientific information can inform
those choices and can be helpful in analyzing their consequences, but
the final determination of levels of harvest (and the amount of habitat
needed to support that harvest) is a policy choice, not a scientific choice
and this boundary should be respected. If citizens have access to the
scientific information, they are the ones who should hold policy makers
accountable for using that science to make wise choices.
Lunch 12:00-1:30
1:30-2:20 pm First Steps towards analyzing the Willamette
recovery issues: a data-poor environment
Michelle McClure (CRI of NWFSC)
Discussion:
Michelle emphasized that her presentation did not address the potentially
negative direct effects of hatchery fish on naturally spawning fish. Rather,
her talk pointed out that the presence of hatchery fish on natural spawning
grounds can mask the true productivity or abundance of a naturally-spawning
population. Without knowing whether the naturally spawning population
is naturally replacing itself, our ability to conduct viability estimates
can be severely compromised.
2:20-3:10 pm The perils of assessing extinction risks
or evaluating hatchery impacts in a data poor environment
Peter Kareiva (CRI of NWFSC) opened with the statement
that we must do extinction models because risk is critical to quantify
for a number of management decisions, including recovery planning, jeopardy
consultations, and setting acceptable harvest levels and habitat quality/quantity
standards. The problem is that there will never be a single "right" extinction
model. Moreover, even if someone did have the correct model, it would
yield huge confidence intervals (if using the standard 95% confidence
level). Two possibilities to shorten confidence intervals are to: 1) look
at a smaller time interval; and 2) lower the confidence level from standard,
but fairly restrictive 95%.
Discussion:
Other possibilities of decreasing the intervals were brought up. One idea
was to raise the definition of extinction (or where bad things happen)
from one fish. Peter agreed that you could raise your quasi- extinction
threshold but it is difficult to know where this is--he said CRI efforts
include multiple thresholds, especially in considering the effects of
multiple age classes on extinction risk. The results you heard today were
simplified for the purposes of illustration.
There was a clarification that these extinction risks are
the absolute minimum. They are based on the assumption that the conditions
that created the trends and absolute abundances in the data series used
do not change into the future.
One question was: How does the CRI approach include diversity,
etc.? Answer: The analyses reported to date do not--they are just a starting
point looking at abundance and variability in abundance. These results
give us a sense of urgency. The more detailed viability analyses as part
of CRI efforts will include the effects of life history diversity and
other more biologically relevant assumptions.
Someone stated that for policy considerations, it would
be ideal to have some context, some understanding of the consequences
of certain decisions. Peter agreed that analyses could be done asking
the question: if you take a certain action, what are the effects? These
sorts of analyses will be conducted under the CRI approach as part of
the Phase II, "implementation" part of recovery planning.
Compiled by:
Amy Robinson
Contact Information:
NWFSC.workshops@noaa.gov
NWFSC-NMFS-NOAA
2725 Montlake Blvd. E.
Seattle, WA 98112
Fax: 206-860-3267
last modified
02/13/2007
Web site owner: Northwest Fisheries Science Center
|
