COMMENTS ON
NATIONAL MARINE FISHERIES SERVICE
Resource Management Plan 4(d) Rule Evaluation and Recommended Determination:
for the Puget Sound Comprehensive Chinook Management Plan:
Harvest Management Component,
provided by Washington Department of Fish and Wildlife and The Puget Sound Treaty Tribes
February 27, 2001

Submitted March 23, 2001 by
Washington Trout
PO Box 402
Duvall, WA 98019
425/788-1167; fax 425/788-9634
wildfish@washingtontrout.org
Prepared by
Nick Gayeski, WT Resource Analyst

1.0 INTRODUCTION
The Puget Sound Comprehensive Chinook Management Plan: Harvest Management Component (the RMP) would, if approved, constitute the Resource Management Plan sufficient to secure protection from ESA Section 9 take prohibitions for fisheries management activities under Take Limit 6 of NMFS’ Final 4(d) Rule for the Puget Sound chinook salmon ESU. NMFS’ Resource Management Plan 4(d) Rule Evaluation and Recommended Determination (RMP Evaluation, or Evaluation) is required to assess whether the RMP adequately provides for fishery management activities that will not impede recovery of the listed ESU.

At a minimum, it would appear to be incumbent upon the RMP Evaluation to assess to the maximum extent possible the likely impacts of activities pursuant to the RMP on individual and aggregate ("management units") wild populations of chinook salmon within the ESU, and to satisfy the public that fishery activities conducted under the RMP are appropriately risk-averse and will be constructively integrated into recovery planning efforts. It would also be expected that the Evaluation list or outline key requirements for risk-averse chinook population management in general and harvest management in particular and then discuss the extent to which the RMP addresses such requirements.

Appropriate guidelines and national standards from the Magnuson-Stevens Act as amended by the Sustainable Fisheries Act (Public Law 104-297; October 11, 1996) and the precautionary approach to fisheries management as outlined by the United Nations Food and Agricultural Organization (FAO) in several recent documents would seem relevant, particularly standards
pertaining to the definition of "overfishing" and "overfished" stocks, and the setting of target and limit biological reference points for harvest management, and their proper interpretation and employment.

The Evaluation fails to provide a significant discussion of these matters in regard to the RMP. Rather, it reads largely like a grammar school first-effort book report that resorts to summarizing and listing, page by page, the assertions made in the book in lieu of ever giving the reader a critical synopsis of the plot or any summation and assessment of the book’s key claims and supporting arguments. Simply put, the Evaluation consists principally of a recitation of the claims and assertions made in the RMP, accompanied by little critical or substantive editorial commentary.

Not surprisingly, therefore, the Evaluation concludes that the RMP "adequately addresses all of the criteria established for an RMP under Limit 6 of the 4(d) Rule" (p. 31), and recommends granting a take limitation for harvest activities pursuant to the RMP for a period of two years (through April 30, 2003).

Based upon a review of the RMP itself and of the details of the RMP parroted in the Evaluation, we believe that the RMP not only fails minimal criteria for doing no harm, but would positively impede recovery of most populations and management units in the Puget Sound ESU. We, therefore, strenuously object to NMFS evaluation of the RMP and its recommended determination to grant a take exemption to the co-managers under limit 6.

2.0 TERMINOLOGICAL ISSUES
2.1 Escapement Level Terminology
Before discussing substantive issues, we note some issues regarding terminology in both the Evaluation and the RMP. The RMP is not at all careful or consistent in its choice of terminology regarding several escapement levels. This makes it difficult for the reader/reviewer to always be sure exactly what level is being referred to in critical contexts. Whether this is due to any inherent difficulty in the application of the several concepts regarding escapement to the conditions of specific populations or management units, merely to poor editing, or a deliberate effort to be vague is irrelevant. The result is that the document is vague, making it difficult to pin co-managers down as to precisely what management targets are being set and under what circumstances they are to apply. This will make it difficult at best to evaluate whether harvest is proceeding in accordance with the plan or not.

At least four terms are employed to describe differing levels of escapement: critical thresholds, low abundance thresholds, viable thresholds, and long-term abundance thresholds. Nowhere within the main body of either the RMP or the Evaluation are these terms defined and clarified in relation to each other. Only low abundance threshold is listed in the Glossary of the RMP.

The terms viable and critical are to be interpreted in terms of the Viable Salmonid Populations document (NMFS-NWFSC-42, June 2000). "Low abundance thresholds" are alleged to be levels of escapement which "trigger additional management actions, and meet or significantly exceed the general guidance in VSP" (Evaluation p.12). Table 4. (Evaluation p. 15) lists Escapement Thresholds under the categories of Critical and Viable. Table 5 (Evaluation p. 16) lists "Spawning Escapement Threshold" and "Low Escapement Threshold."

A Low Abundance Thresholds is defined as "A spawning escapement level below which the co-managers will exercise maximum regulatory effect to minimize fishery related impacts and maximize spawning escapement" (RMP Glossary, p. 33). The relationship between this threshold and critical and viable ones is rarely made clear and is at best inconsistent.

For example, Table 5 lists the low escapement threshold for the Stillaguamish summer/fall chinook at 500, which appears to correspond to the critical thresholds listed for the North and South Fork stocks in Table 4 (300 and 200, respectively). The current (= recent) escapement threshold listed in Table 5 for this stock is 2000. The viable threshold from Table 4 is 852 (552 North Fork, 300 South Fork).

For the Snohomish summer/fall chinook Table 5 lists a low escapement threshold for three component stocks of 1530 (582, 621, and 327 each), where Table 4 lists a critical threshold of 1100 (400, 300, and 200, respectively). Table 5 also lists a low escapement threshold for the entire Snohomish summer/fall management unit of 2000 and a current (= recent) spawning escapement threshold of 5250. Table 4 lists viable thresholds only for the first two of the three stocks that are provided with low thresholds in Table 5. These viable thresholds are 2325 and 1600 (=3925).

Further confusion arises when the levels for the Skagit summer/fall chinook given in Table 5 are considered. The spawning escapement threshold is 4700 (in contrast to the recent escapement goal of 14,900) and the low abundance threshold is 4800. This inconsistency is acknowledged but inadequately addressed in a footnote on page 14.

These four terms pertaining to escapement levels should be employed consistently and with clear meaning throughout the RMP. At the very least such terminological sloppiness contributes to making the intentions and goals of the intended plan less than perfectly transparent. For example, what exactly is the "maximum regulatory effect" managers will exercise at the Low Abundance Threshold? The Evaluation indicates that the thresholds "trigger additional management action," but again, without identifying or defining the action. If "maximum regulatory effect" or "additional management action" meant closing or severely curtailing a fishery (a reasonable response, and certainly implied in the phrase "minimize fishery related impacts and maximize spawning escapement") why not simply say so? It appears that neither the RMP applicants nor NMFS want to be locked into such an unsavory choice, or provide the public an objective standard for judging whether or how conditions of the RMP are being met. An "additional management action" could mean almost anything, including aggregating a population into a still larger management unit, or simply revising the threshold. Transparency, if not ease of understanding, should be a minimum requirement of any acceptable harvest-management plan, even were it not targeting or encountering ESA-listed populations.

2.2 Harvest-Rate Terminology
Terms pertaining to harvest rates are also employed with less than reasonable clarity. The RMP and the Evaluation both refer to Exploitation Rates (ERs) and to Recovery Exploitation Rates (RERs). Both rates are generally understood to be measured in terms of "adult equivalents" (AEQs). However, neither document takes the trouble to make it clear whether or not the specific rates discussed and listed in tables are annual rates -- in which the denominator quantity is the total stock size susceptible to fishery mortality in a given year -- or whether it is the size of the cohort at the time (year) at which members of a brood year are first recruited to the fishery. This is critical to know.

The RMP's Glossary definitions of harvest rate and exploitation rate are of little help in this regard. Harvest rate is defined as "total fishing mortality in a fishery expressed as a proportion of the total fish abundance available (standing stock) in a given fishing area at the start of a time period." In other words, it is the proportion of total fish caught and/or killed in/by the fishery in a particular year (fishing season) divided by the total number of fish susceptible to being caught and/or killed by the fishery at the start of the season. It is clearly an annual rate.

Exploitation rate is defined as "total mortality in a fishery expressed as the fraction of the potential escapement removed due to the fishery". It is unclear what the numerator and denominator of this rate are according to this definition. The numerator is an adult equivalent of fish killed by the fishery, but it is unclear over what time period this adult equivalent is to be calculated. If the adult equivalent is the adult equivalent of the numerator of the harvest rate defined above, then the ER is merely the adult equivalent version of the HR and it is likewise an annual rate calculated on the total number of fish in a given season that are susceptible to fishery-induced mortality. In other words, it is the HR with the numerator and denominator both discounted by the age-specific AEQs.

What is most relevant to chinook management and recovery, however, is the Cohort (or Brood Year) rate. This is the total number of fish (either raw or as AEQs) from a cohort (brood year) suffering harvest-induced mortality summed over all seasons during which the cohort is susceptible to fishery-induced mortality as a proportion of the number of fish in the cohort (or their adult equivalents) when it first becomes recruited to the fishery. It is approximately the rate that would be calculated in terms of the stock-recruit relationship.

This latter rate is significantly higher than the corresponding annual rates. As a rough indication of the difference, an MSY harvest rate expressed in terms of the stock-recruit relationship of 65 to 70% on a chinook stock with the approximate adult age composition of current Puget Sound populations is realized by an annual AEQ harvest rate in the neighborhood of 40%.

It is critical to know whether the ERs discussed and listed in the RMP are Cohort rates or simple annual rates. The reviewer should not be left to assume or guess this, especially in a document for public review.

3.0 SUBSTANTIVE ELEMENTS AND ISSUES OF THE RMP
The standard that the RMP must meet is rather minimal, though nonetheless substantive. The RMP must reasonably demonstrate that implementing and enforcing the plan "will not appreciably reduce the likelihood of survival and recovery of [the] affected threatened ESU."

It would seem reasonable that any action judged to "not appreciably reduce the likelihood of survival and recovery" of the ESU would be consistent with the listing decision and the reasons given therein for listing the stock in the first place. We believe the RMP fails this simple consistency test.

The RMP sets "recovery exploitation rates " that are purported to secure a high probability of not reducing population or management unit abundances below newly chosen "Low Abundance Thresholds." These thresholds, however, are set at levels that are generally less than one-half the previously established, generally MSY-based, escapement goals.

These latter escapement goals for most Puget Sound wild chinook stocks (previously established and agreed upon in the Puget Sound Salmon Management Plan) have consistently not been met in 95% of the past 20 or more years. The RMP and the Evaluation attempt to imply that the RERs will allow for current or better escapement, but both documents steadfastly refuse to commit to attaining the long-standing escapement targets for the management units in the ESU.

It is these escapement failures in addition to declining trends in both estimated run size and escapements during most of the recent past that motivated and justified the listing decision in the first place. NMFS clearly listed harvest impacts among the causes of declines in abundance in its decision to list the Puget Sound ESU (FR 64 14316 and 14319). (and in its Status Review, pp. 222 and 251, Response to Comments to the Proposed 4(d) Rule, and in the Final 4(d) Rule.)

The RMP abandons these previous escapement goals as management targets altogether, and would replace them with exploitation rates that are believed on the basis of FRAM model-based simulations to have a "low" probability of producing escapements lower than less than half the previous escapement goals.

Such an approach would appear to be clearly inimical to contributing to the rebuilding of at-risk, listed populations within the ESU that are critical to the recovery of the ESU. Worse, such an approach has a high probability of driving marginal small wild populations within the ESU to extinction.

Moreover, the entire issue of hatchery-origin fish in the escapement and on the spawning grounds is inadequately addressed in the RMP and essentially not even raised in the Evaluation. It is unclear whether only NORs ("natural origin recruits") are the object of each of the several "escapement levels", or whether hatchery strays will be included in determining whether or not some of the escapement levels are achieved.

This raises a fundamental inadequacy of the RMP and the Evaluation which we note here, but do not elaborate upon due to the severe time-constraint of the 14 day comment period, as opposed to the 30 day period explicitly declared for public comment on FMEPs in the 4(d) Rule. It is fundamentally unclear how hatchery management plans are to be integrated into the RMP! The two cannot be considered in isolation from one another, yet the public is being forced to consider the RMP in isolation from hatchery plans and policies that will likely drive many aspects of harvest management. NMFS should require the co-managers to provide an integrated hatchery/harvest management policy for public review as a condition for issuing a limit to take under Limit 6.

3.1 Critical Assumption
A critical and unjustified assumption of the RMP that underlies the general approach of the RMP is suggested by several remarks in the Evaluation regarding population productivity:

The Puget Sound Chinook RMP states that:
"It is the goal of the Parties to protect, restore, and enhance the productivity, abundance, and diversity of Puget Sound chinook salmon and their ecosystems to sustain ceremonial, subsistence, commercial, and recreational fisheries, non-consumptive fish benefits and other cultural and ecological values. Achievement of this goal requires that harvest be constrained within limits appropriate to the productivity of each stock." (p. 4) (emphasis added.)

This appears to assert the reasonable and conservative view that harvest rates should not exceed "productivity" of the stock. But this is not stated. The choice of words is intentionally open-ended and telling: harvest is to be constrained within limits "appropriate to" stock productivity. This is a hedge against having to "under-harvest" an "unproductive" stock as well as against overharvesting one. This is evident in the discussion of Population Size on page 13 and the discussion of Population Growth Rate on page 17 of the Evaluation:

"The RMP reflects a variety of methods used to derive long-term spawning and low abundance escapement thresholds for Puget Sound chinook populations. Long-term thresholds in the RMP represent (1) established MSY escapement goals agreed to under the PSSMP; (2) MSY escapement consistent with current conditions; or, the escapement for which there is a 1% probability of extinction at the end of 100 years…" (p.13) (emphasis added.)

"Productivity is driven by habitat quality and reproductive fitness, not by fishery actions. However, harvest management objectives must be appropriate to the habitat capacity and productivity experienced by the individual populations. Although the Puget Sound chinook RMP includes no explicit management objective for productivity, the interim viable thresholds and exploitation rates are based on current survival and productivity rates with adjustments to account for data uncertainty and management imprecision. By ensuring that harvest management objectives are consistent with current environmental conditions, and accounting for known sources of error, fisheries are not expected to impede recovery.
"As part of the scheduled RMP evaluation, management objectives will be revised as necessary to reflect changes in environmental conditions. The intent is to increase spawners in concert with the recovery of the system's productivity and capacity resulting from habitat restoration efforts, thereby annually providing sufficient escapement to enable the management unit to generate maximum surplus under progressively improving habitat conditions. In this way, harvest is linked to recovery efforts in the other H's…" (p. 17)(Emphases added)

This leading idea is clear (and decidedly at odds with current ecological thinking and salmon life history): only so many fish can use the existing amount of habitat, so there is no point allowing any "extra" fish to escape to try to spawn. This reflects a very crude deterministic view of salmon productivity and of the relationship between salmon and freshwater habitats.

Under such a view it is positively pointless and wasteful to fail to harvest all of the "excess" individuals from a run size greater than the habitat’s capacity. Harvest management can only be an impediment to recovery if it oversteps this line and prevents escapement for which there is usable habitat up to the MSY recruitment point.

This point of view implicitly assumes that current spawning and rearing habitats within the ESU have by and large been completely occupied. This is decidedly not the case for many key populations within the ESU. More importantly, neither the RMP nor the Evaluation provide any credible data or methodology appropriate to an attempt to establish the claim that spawning and rearing habitats have been at or near capacities, and that consequently escapements at or above current MSY-based goals would not result in additional recruitment.

The fundamental problem with such reasoning in the case of most populations within the ESU is that escapements have generally been significantly less than extant escapement goals for over 5 generations of chinook. Harvest rates have simply failed to decline enough to secure a consistent attainment of apparently minimal and by and large MSY-based levels of escapement.

Quite simply there is no evidence based upon recruitment from escapements at or above these existing escapement goals for as long even as one generation of chinook for claims that habitat capacity and productivity are insufficient to support such levels of escapement. Harvest has systematically prevented attainment of such escapement levels.

Neither does the RMP acknowledge, much less consider and incorporate, recent data concerning the role of ocean-derived nutrients supplied by salmon carcasses in increasing the productivity of freshwater rearing habitats, and the relevance of such inputs to the setting of escapement goals adequate to recovery. (cf., e.g., Bilby et al. Fisheries vol. 26, # 1, 2001 and Michael, Northwest Science vol. 72: 239-248)

Rather than commit to attaining such minimal escapement levels through a fundamental revision in chinook harvest management as a component of a suite of comprehensive recovery initiatives, the co-managers are attempting to use the loss and impairment of freshwater habitat as an excuse to set lower escapement targets! Neither the RMP nor the Evaluation evidences any extant data or peer-reviewed literature to support the implication that attaining hitherto-existing escapement goals (goals stated, by the way, in the co-managers’ Puget Sound Salmon Management Plan) would exceed existing habitat capacity or productivity!

The Evaluation should require the RMP to explicitly address this issue and to thoroughly document evidence that 1) extant escapement goals are incapable of yielding MSY escapements; and 2) current MSY escapements lie on the lower escapement side of the existing escapement goals.

In point of fact, the only real proof of such claims and hypotheses is to provide escapements at or above the existing target escapement levels for several generations. Even one generation of attainment of these escapements combined with rigorous monitoring of fingerling/smolt production would provide meaningful data for assessing whether freshwater habitats are capable of producing increasing recruitment from escapements relatively higher than those in the recent past.

In light of such considerations, we strenuously disagree with the assertion on page 22 of the Evaluation that:
"[t]he use of exploitation rates rather than fixed escapement goals for most Puget Sound chinook populations will allow the possibility of larger run sizes, which are needed to explore population productivity and habitat capacity. Rather than always harvesting down to escapement when abundance is high, a proportion of the run size is always allocated to escapement regardless of run size … an exploitation rate approach is more resilient to data uncertainty and environmental variability than a fixed goal approach."

In a fishery with a management history of consistently attaining minimum escapement goals through exploitation rate management, this argument may be a reasonable one. When management history is otherwise as is the case for most Washington Coast and Puget Sound fisheries over the past 25 years, there is little reason to believe such a claim. The Evaluation disingenuously leaves out the flip side of its argument, that even in periods of low abundance a fixed proportion of the run size is always allocated to harvest, regardless of actual escapement numbers. Exploitation rate management on ESA-listed populations is essentially an approach that assures fishing opportunity regardless of stock size.

(Even when populations fall below the putative "Low Abundance Thresholds," neither the RMP nor the Evaluation will commit to actually ending harvest on that population. It is hard to escape the conclusion that given their management history, WDFW and the Treaty Tribes have little confidence in their ability to meet established escapement objectives, but they are comfortable with new "low abundance" targets at less than half those objectives.)

Not only is the underlying premise of tying the ERs to unsubstantiated estimates of habitat-driven productivity unsupported and misleading, it is expressed in a way that will make objective evaluation of RMP performance impossible. "Management objectives" can be "revised" in any number of ways. Identifying the "intent… to increase spawners" and provide "sufficient escapement" may imply lowering exploitation rates to increase escapements, but the RMP and the Evaluation decline to lock themselves into that minimally responsible course. Under the RMP’s and Evaluation’s theory of exploitation management, fixed ER’s can automatically take advantage of "changes in environmental conditions." Under this theory (and the vague language employed in the Evaluation), revisions to management objectives in response to "improving habitat conditions" could conceivably include raising ER ceilings to take advantage of the newly generated "maximum surplus."

Moreover, there is nothing in the principle of escapement-goal (a spawning biomass Target Reference Point) management that necessitates restricting escapement to the goal when run sizes are higher than expectations. Except in the case of exclusively terminal fisheries, escapement-goal management is largely realized by employing gear, time and area restrictions i.e., exploitation rates. The difference from exploitation-rate management is simply that under escapement-goal management exploitation rates are adjusted on a regular basis (in-season, annually and/or over a period of several years) to assure that the escapement-goal minimum (the floor target) is attained. The option is always open to set the ER at zero when returns are predicted to be low and to keep them low when returns are predicted to be high in order to realize escapements in excess of the escapement floor. The claims of the Evaluation in the above passage are inaccurate and misleading, if not actually disingenuous in regards to escapement-goal management.

It should also be pointed out that exploitation-rate management does not inherently result in above-floor (goal) escapements when run sizes are higher than expected. Recent history provides a clear example.

In 1998 the Snohomish summer/fall chinook stock met its escapement goal of 5250 for the first time since 1980. Escapement was 6304 from a post-season estimated terminal run size of 6400 natural and 1100 hatchery chinook. The pre-season estimates were 5600 and 6500, respectively (Table II-9. PFMC Salmon Technical Team Preseason Report I February 2001, page II-20.)

In 1996 exploitation rates on Snohomish Basin chinook were alleged to be at or below the RMP target of 0.32. (Table 2 of Appendix A of the RMP (p. 59) lists AEQ ERs for the Snohomish Stock of .27 in 1993 and .21 in 1994, the most recent brood year data available from the Chinook Technical Committee of the Pacific Salmon Commission). The post-season estimated terminal run size was 8000 natural and 9200 hatchery. Pre-season estimates were 4200 and 6700, respectively. Despite a natural run size 1600 larger than in 1998 and a total run size 9700 larger than in 1998, escapement in 1996 was only 4851, 400 below the escapement goal. Exploitation rate management not only failed to secure that a run size greater than expectation resulted in additional escapement in excess of the escapement goal, it failed to attain the goal.

3.2 Insufficiently Risk-Averse
The RMP and the Evaluation propose to proceed in the direction of subjecting populations of the listed ESU to continuing risk of excess harvest, under the guise of uncertainty regarding population productivity and the thesis that by staying near or above estimated "low abundance thresholds" the maximal amount of harvest can be attained while not "appreciably reducing the likelihood of survival and recovery" of the ESUs. This is to be done by abusing the concepts of "viable" and "critical" thresholds employed by the VSP and in general by proceeding contrary to the precautionary approach embodied in the Sustainable Fisheries Act and numerous recent United Nations FAO documents.

The RMP fails to couch the plan in terms of Biological Reference Points, which have become a near-universal frame of reference for discussions of risk-averse fisheries management throughout the world. It fails to make clear the purpose of distinguishing between a Limit Reference Point and a Target Reference Point for fisheries management, and it fails to adopt an approach relating harvest management to recovery that is consistent with the Sustainable Fisheries Act policies regarding overfishing as embodied, for example, in National Standard One (which apply to WA coast fisheries under PFMC). The Evaluation makes no case, let alone a compelling one, for allowing these omissions in the RMP.

In general, it appears that the RMP’s "low abundance thresholds" are established in order to "safeguard populations from declining to the point of instability…," and are to "trigger additional management actions" when such thresholds are crossed (Evaluation p.12). Such points of instability appear to correspond to critical thresholds as characterized in the VSP. It might therefore be inferred that the management actions that are to be triggered will be intended to prevent the population or management unit from descending to critical level escapements, although this is not made clear. In any case, it does not appear at all likely that many of these thresholds will achieve a high probability of avoiding the kinds of declines they are alleged to be designed to prevent.

A glance at Tables 4 and 5 (Evaluation pp. 15 and 16), reveals that low abundance thresholds are generally within 100-200 spawners of critical thresholds. Even if one grants for the sake of argument that the critical abundance levels are appropriate, the low abundance levels that are to trigger corrective management action are dangerously close to the level that the action is to avoid. Given the likely inability of any corrective action to compensate for this knife-edge approach, and the RMP’s and Evaluation’s disinclination to explicitly define what the corrective action would be in any case, it will again be difficult if not impossible for the public to judge whether and how "additional management actions" will be in compliance with the RMP or "safeguard populations from declining to the point of instability…."

Moreover, harvest which reduces a population to such low levels will impact several year classes from several cohorts. By the time this overharvest manifests itself in a return below the low-abundance threshold, returns for several future years will have already been impacted. Such cumulative impacts must be factored into the setting both of limit thresholds (such as the so-called "low abundance thresholds") and target thresholds set well above the limits that are the object of active management and which are intended to reduce the risk of approaching the limit boundary.

Contrast this with the Magnuson-Stevens Act definition of "overfishing" and the corrective actions to be taken when a healthy stock becomes "overfished." (See FR 62 , no. 149; 41907-41910 and FR 63, no. 84; 24212-24213 and 24229-24233.) National Standard 1 (Optimum Yield, OY) requires that "optimum yield" from a fishery be defined with respect to MSY (maximum sustained yield) "as reduced by any relevant economic, social, or ecological factor; and in the case of an overfished fishery, that provides for rebuilding to a level consistent with producing the MSY in such fishery" (FR63, no. 84; 24232) (emphasis added). Optimum yield therefore can be no greater than MSY.

An "overfished" stock is a key concept in the Act and is defined with respect to the definition of MSY used to specify Optimum Yield and the "control rule" chosen to achieve OY. A stock is "overfished" if it is reduced to a size that is "sufficiently small that a change in management practices is required in order to achieve an appropriate level and rate of rebuilding" (FR 63; no. 84, 24230). That size is "[o]ne-half the MSY stock size, or the minimum stock size at which rebuilding to the MSY level would be expected to occur within 10 years if the stock or stock complex were exploited at the maximum fishing mortality threshold…" (emphasis added).

This criteria is to be used to determine when a healthy stock becomes overfished and triggers a statutory requirement for initiating a rebuilding action. The one-half MSY level is a lower limit to the specification of this lower threshold. The real limit is greater than or equal to one-half MSY stock size (escapement) and is the minimal size that would permit the stock to rebuild to MSY escapement within 10 years if fished at the MSY exploitation rate.

This leads to the specification of the minimum rebuilding time for an overfished stock. This is "the length of time in which a stock could be rebuilt in the absence of fishing mortality on that stock" (FR 63; no. 84, 24212) for which National Standard 1 provides an upper boundary: "the outside limit of the rebuilding period is the no-mortality period plus one mean generation time…" (ibid., 24213) (emphasis added). In the case of chinook this upper bound would be 14 or 15 years!

The stock size below which a stock is said to become overfished, according to National Standard 1 is a precautionary Limit Reference point. The purpose of fishery management plans (FMPs) under the Sustainable Fisheries Act is to insure that such Limit points are never approached. If despite extant FMPs such a Limit is crossed, prompt action on a short timeline is required to reverse the overfished condition and rebuild the stock to the OY level.

In order to insure that the FMPs not result in the Limit being approached, Target Reference Points based upon the OY control rule are required. These are the points that must be actively managed for by the FMP!

The employment by the RMP and the Evaluation of low abundance thresholds and exploitation rates intended to have "low probability" of "falling below the critical abundance threshold…" Evaluation p.19) is at the least inconsistent with the approach embodied in the Sustainable Fisheries Act in that it treats these thresholds essentially as Targets, not as Limits.

Language throughout the Evaluation reinforces the implication that the "Low Abundance Thresholds are the new management objective. It is the Low Abundance Thresholds that "trigger additional management actions" (p.12). Exploitation rates are identified as the backbone of the RMP, governed not by moderate to high escapement levels they are purported to produce, but only by the Low Abundance Thresholds: "The basic harvest management strategy is to keep exploitation rates at or below a management unit-specific ceiling rate, as long as the unit’s spawning escapement is expected to be above the low abundance threshold." (p.12.) (Emphasis added.) The Thresholds are again identified as targets in discussion re Criteria (C): "… should management units or associated populations fall below or be projected to fall below their low abundance thresholds, Southern U.S. exploitation rates will be reduced…" (p.19) (Emphasis added). While the ERs are represented as designed to provide current or higher escapements, the only clear escapement targets identified in the Evaluation are the Low Abundance Thresholds.

Furthermore, the time frame over which impacts of the target ERs has apparently been modeled is longer than that required of fisheries subject to the Sustainable Fisheries Act. The Evaluation states that harvest at the target exploitation rate "assures an 80% probability of the management unit exceeding the recovery [viable] escapement level in 25 years." (p. 19) Here the viable escapement levels, which per Tables 4 and 5 are shown to be perilously close to the low abundance thresholds and less than half of the existing MSY-based escapement goals are treated as Targets. Simulation modeling by the co-managers, further, indicates that there is a 20% probability that these levels will not be attained within 25 years!

We can therefore hardly concur with the Evaluation that the "general approach to setting exploitation rate ceilings described in the RMP… is risk averse in that it is designed to provide high probabilities of survival and recovery." (p. 19). To the contrary, by steadfastly refusing to specify Target escapement levels significantly above low abundance thresholds and to plan to actively manage to achieve or exceed such target escapements, the RMP would impose continuing harvest risks on populations and management units within the ESU that will impede recovery.

We recommend that the RMP be redrafted using a template that at least follows the precautionary fishery-management principles embodied in the Magnuson-Stevens Act as revised by the Sustainable Fisheries Act in the reliance upon biological reference points, stated in terms of Limit and Target reference points (cf. e.g., Caddy 1998. "A short review of precautionary reference points and some proposals for their use in data-poor situations"; FAO Fisheries Technical Paper 379).

3.3 Required Revision to Chinook Harvest Management
The RMP and the Evaluation fail to adequately address uncertainty in modeling populations and management units within the ESU, and fail to consider aspects of life history that are unique to chinook and are critical to incorporate into any harvest management regime capable of contributing to stock rebuilding and the eventual de-listing of the ESU. We note the principal issues here and recommend that NMFS require their incorporation into a revised RMP as a pre-requisite to its approval.

Chinook have a considerably more complex life history than other Pacific salmon. Most populations mature at multiple ages. In addition chinook are subjected to harvest mortality at sub-adult (generally age-2) and young adult (ages 3 and 4) life stages, in addition to fully mature stages (ages 3 - 6). This has been the case for nearly a century for most stocks, including those within Puget Sound. In addition, for most of the historic record, chinook have been subject to size-selective fisheries that disproportionately targeted older, larger fish.

The net result of these two factors - fishing on immature ages and differential selection of older, larger individuals -- has been a decline in mean body size, age-at-maturity and mean population age, and a shift in the sex ratio towards males. It is now widely acknowledged that chinook are smaller and younger than they were even 20 years ago, and females therefore are depositing fewer eggs.

These conditions have a two-fold effect on the life history of chinook populations: a more simplified age-structure renders each population less resilient to environmental variability; and given levels of escapement result in fewer eggs deposited and therefore fewer adults recruited per spawner even when assuming no changes in post-egg deposition survival rates. The reduction in age and size of female spawners however are likely to result in reductions in post-deposition survival rates due to changes in egg size and quality, and to the inability of smaller females to bury eggs as deeply in the substrate.

Smaller female body size also results in available potential spawning habitat being unused due to the inability of smaller fish to spawn at depths and velocities or utilize substrate size accessible only by larger individuals. Despite the claims of the Evaluation, these direct impairments of population "productivity" are the result of harvest impacts alone.

It is also important to recognize that harvest mortality applied to sub-adult and immature 3,4, and 5 year old chinook will significantly skew the age structure in the direction of a younger mean generation time even in the absence of direct size-selectivity in the fishing regime. Once altered in the direction of fewer adult age classes, earlier age-at-maturity, smaller body size, and increased proportion of males, moderate harvest rates (ERs) are capable of suppressing directional change toward a more historic, complex age structure. Harvest rates have to be significantly reduced below rates that may be required merely to achieve minimal escapement goals. Exploitation rates that result in escapements well above current goals may be necessary to significantly reduce the directional force that sustains current simplified age structures, which likely include altered maturation schedules that are to some extent under genetic control.

This belies the Evaluation’s contention that the RMP "preserves the existing diversity and spatial structure of natural populations within Puget Sound," or that it appropriately manages for the ESU’s "full range of genetic diversity and life history traits."

Such issues clearly indicate the need to integrate harvest management (and hatchery policy) as an active component of a truly "comprehensive" approach to chinook recovery. However, they receive no mention in the RMP or the Evaluation. We recommend that a revised RMP include the following:

&Mac183; Escapement goals derived on the basis of potential egg deposition and stated in terms of numbers of female spawners.
&Mac183; Escapement goals stated in terms of the age composition of females need to be developed.
&Mac183; Exploitation rates adequate to recovering more complex, historic population age-structure need to be devised and evaluated. This requires that ERs be clearly determined as cohort rates expressed as the fraction of the total number of individuals in the cohort at the age at which they are first recruited to the fishery.

Models used to set harvest regimes and to evaluate their performance need to be more fine-grained than current models appear to be (including FRAM). Each fishery (by area and gear type) inflicts mortality on several age classes, each of which is at a different stage of maturity. Over the duration of the season for each gear type and area, the size and maturity of each age class of each stock susceptible to the fishery changes.

Calculation of harvest impacts on age classes must account adequately for the impacts of harvest mortality within each area on current and future year spawning. A simple assumption of AEQs for each immature age is inadequate to this kind of calculation of impacts. AEQs change during the season within each area depending upon which stocks are being encountered by the fishery and their state of maturity at the time they encounter the fishery and they are different for each sex.

At a minimum each area should be subdivided into area-time segments and age-size distributions estimated on the basis of stock-specific maturity schedules applied within each area-time segment to estimate AEQ impacts. These impacts then need to be integrated over the life of each cohort of each population.

This requires a level of modeling and data collection beyond those currently in use in chinook harvest management. But these requirements are not beyond the limits of current technology. In fact, such a model was developed in the early 1970s by the Washington Department of Fisheries but abandoned as bureaucratically inconvenient. Regardless, as long as chinook are fished as immatures and away from terminal areas, such detailed management is necessary in order to secure a responsible fishery regime that is biologically appropriate to the complex life history of chinook. Evaluations of submitted RMPs must require either this level of detail in modeling and management, or alternative fisheries regimes.

The alternatives to this kind of complex management are few, but include:
&Mac183; Fish only on hatchery chinook using selective technologies capable of avoiding wild stocks and/or of releasing them unharmed.
&Mac183; Fish only in terminal areas where encounters will only be with mature members of known populations or small aggregates of populations, and where harvest impacts by stock, size, age, and sex can be closely monitored in-season.
&Mac183; Employ a maximum size limit to eliminate harvest mortality on older aged wild chinook in all fisheries.

3.4 Uncertainty in Stock Assessments
The Evaluation asserts that the RMP acknowledges and addresses uncertainty and imprecision in stock productivity and in harvest management. As an illustration, we quote more fully from page 19 of the Evaluation cited previously:

"The general approach to setting exploitation rate ceilings described in the RMP is risk-averse in that it is designed to provide high probabilities of survival and recovery. Harvest at the target exploitation rate:
&Mac183; "will not increase the probability of the management unit falling below the critical abundance threshold, in 25 years, by more than 5 percentage points than if the exploitation rate were modeled as zero (original emphasis);
&Mac183; "assures an 80% probability of the management unit exceeding the recovery [viable] escapement level in 25 years.
"For each management unit, the exploitation rate objectives are developed that reflect the current productivity of its associated population. A simulation model was used to project escapements of the management unit or population over a 25-year period under a range of exploitation rates. The simulations included variability in data estimates, management error and survival conditions."

And on page 13 (previously quoted) "long-term thresholds" include escapements "for which there is a 1% probability of extinction at the end of 100 years."

No documentation, however, is provided either in the Evaluation or the RMP regarding critical details of the simulation modeling enterprise itself or of the assessment methodology used to determine "current productivity of [management unit] associated populations." Key parameters, such as life-stage specific survival rates, are either not given or not made fully clear; nor are ranges and distribution of such parameters discussed. Neither are error terms and their distributions provided or discussed.

Uncertainty in parameters such as life history stage-to-stage survival and mortality rates, and stock-recruitment function parameters are generally best characterized as parametric or non-parametric distributions. Both probability mass functions and cumulative distributions are particularly valuable in characterizing uncertainty. (cf. e.g., W.J. Overholz "Precision and Uses of Biological Reference Points Calculated from Stock Recruitment Data" North American Journal of Fisheries Management vol. 19, no. 3, August 1999, pp.643-658).

No such features of common discussions and analyses of uncertainty in population modeling occur in the RMP or the Evaluation. Without knowing the relevant details and assumptions made in the simulation modeling there is simply no way to evaluate the accuracy or reasonableness of claims such as those made in the above-quoted passage and elsewhere in the RMP and the Evaluation. (Our previous criticism of this passage is based on taking the claims at face value).

We also note that recruitment estimates based upon stock-recruit functions, which largely underlie the estimation of MSY harvest rates and escapement levels for salmon, are notoriously subject to the impacts of environmental variability. A dominant impact of environmentally-driven variability in life stage survival rates is a right-skewed or log-normal distribution in recruitment at any given level of spawner abundance (See, e.g., Cramer 2000 "The effects of environmentally driven recruitment variability on sustainable yield from salmon populations"; chapter 31, pp. 485-505, in Sustainable Fisheries Management , CRC press). This results in the majority of recruitments being lower than the mean recruitment value estimated by the stock-recruit function. It is not at all apparent that the various escapement thresholds and ERs mentioned in the RMP and the Evaluation have taken this into account. It can only be assumed that all values referenced are mean values (point estimates) and therefore seriously under-estimate the likelihood of escapement levels being attained or, in the case of critical and low-abundance thresholds, avoided!

Claims regarding the probabilities of reaching extinction thresholds are, in addition, extremely sensitive to the definition of 'extinction'. Common definitions of extinction range from zero spawners for one mean generation time to the population size at which the decision would be made to take all spawners into a captive broodstock program, which may be as high as 50 spawners/year. Without knowing how 'extinction' is defined it is not possible to judge exactly how risk-averse a 1% probability of extinction within 100 years is.

In addition, when several management units and populations within the ESU are considered, claims concerning extinction probabilities such as the 1% probability of extinction in 100 years sound much more conservative than they really are. If we accept at face value the claim that escapement goals for each of 13 Puget Sound management units are established so that each management unit has an independent probability of surviving for 100 years of 99%, then the probability of all 13 management units surviving for 100 years is 0.99^13 or about 88%. This means that there is a 12% probability that at least one of the MUs will be extinct in 100 years! If what we really desire is that all MUs together have no less than a 99% probability of surviving for 100 years, each MU must have a probability of surviving for 100 years of 99.923%; in other words, each population must have a probability of becoming extinct within 100 years of .077%, not of 1%. This is a considerably higher standard and represents a considerably more risk-averse condition!

The claim of the Evaluation that "exploitation rate objectives are developed that reflect the current productivity of [management unit] populations" is simply bereft of any detailed documentation, either data-based or model-based. But even if documentation were provided, a range and distribution of parameter values relevant to the alleged determination of exploitation rates "reflective of" population productivity would be required in order to appropriately assess the uncertainty in such parameter values and in the extrapolation from those values to the allegedly risk-averse exploitation rates themselves.

Only after distributions are provided with which to characterize the uncertainty of assessment data, models, and model parameters does the issue of risk-aversion arise. Risk-aversion is exhibited by where along the cumulative distribution curve management points are chosen. Without a clear idea of the shape of the distribution and specific numerical values at specific points along the cumulative curve (such as mean, mode, median, quartiles), it is nearly impossible to judge how risk-averse a target value such as an ER or a "low-abundance escapement level" is.

The failure of the RMP to provide such detail itself renders the RMP a risky plan upon which to base harvest management of populations within a listed ESU. It is difficult, therefore, to avoid the suspicion that such a discussion of uncertainty is largely missing from the Evaluation and the RMP because a more discrete discussion of the relevant uncertainties connected with harvest of Puget Sound chinook populations would reveal that the approach the co-managers wish to take entails considerably more risk to chinook stocks than they wish to make known.

A self-evident principle of risk-averse resource management would be to move forward on exploitation actions very circumspectly, if at all, in the face of uncertain or insufficient data. As in the 4(d) Rule itself, the Evaluation not only approves exploitation actions in the face of uncertain and insufficient data, it appears to allow the applicants to use uncertain and insufficient data to justify action. Not only does this create a disincentive to collect potentially useful data, it again creates an environment that makes public evaluation of compliance with an approved RMP difficult or impossible.


4.0 SUMMARY
As with the 4d Rule itself and the criteria for FMEPs (applicable to Joint RMPs submitted under take limit 6), Washington Trout objects to the overall tone of the RMP Evaluation. In the Evaluation, NMFS appears to uncritically concede the implicit premises underlying the Joint RMP: that salmon harvest and its management have not been a factor in population declines within the ESU; that insofar as harvest should be revised, it is only in reaction to other factors limiting productivity, spatial structure, and diversity; therefore, status quo exploitation and escapement levels (or even reduced escapement represented by new Low Abundance Thresholds effectively replacing minimum escapement objectives as a new management "target") are not necessarily inconsistent with recovery. Under these premises, increased biological risks to populations and management units within the ESU can be dismissed against increased economic and bureaucratic risks to fisheries and managers. In other words, harvest proposals can be designed, at least in part, to assure that "treaty and non-treaty fishermen do not bear the burden of conservation beyond the effects of their actions." (Evaluation p.24.) By conceding these premises, the Evaluation allows the RMP applicants to decrease their conservation "burden" by diminishing or dismissing the "effects of their actions."

But in its Status Review, Listing Decision, Response to Comments to the Proposed 4(d) Rule, and Final 4(d) Rule, NMFS implicitly and explicitly acknowledges the role of harvest and harvest management in the decline of the ESU and in its risk of extinction. That acknowledgement does necessarily require significant revisions in harvest-management policies and their underlying assumptions. It does necessarily require that social, economic, and bureaucratic risk be weighed appropriately against the biological risk of any proposed harvest or management action, and that fisheries and managers assume an appropriate "burden of conservation."

Under the 4d Rule NMFS is required to critically evaluate proposed harvest regimes and the full "effects of their actions" before it approves applied-for take limitations. The Evaluation appears to have merely reviewed the submitted RMP to determine whether it has filled in the proper blanks on an application form. It offers little in the way of actual evaluation, and almost no objective analyses within the document itself to justify the Recommended Determination. The Evaluation fails to acknowledge all the scientific evidence regarding the effects of harvest on the abundance, productivity, spatial structure, and genetic and life-history diversity of populations within the ESU, does not attempt to analyze or weigh the impacts of those effects, and does not adequately justify its determination that the proposed RMP will address and mitigate these impacts.

In conclusion, we reiterate that the Evaluation fails to adopt a reasonably critical attitude toward the RMP. The RMP fails to reflect a sound comprehensive and ecologically-based framework within which harvest management of Puget Sound chinook can be pursued in a manner consistent with preservation and recovery of the ESU.

The RMP provides no valid basis upon which NMFS should grant a take limitation under Limit 6. We recommend that NMFS revise its Evaluation and Recommended Determination to indicate that the submitted RMP does not adequately address all of the 4(d) Rule criteria for Take Limit 6. We further recommend that NMFS request a revised Joint RMP from WDFW and the Puget Sound Treaty Tribes.

5.0 ATTACHMENTS
1. SUPPLEMENTAL COMMENTS:
Comments on a Proposed National Marine Fisheries Service Evaluation and Recommended Determination Under Limit 6 of the Endangered Species Act 4(d) Rule for the Listed Puget Sound Chinook Salmon ESU Re: Puget Sound Comprehensive Chinook Management Plan: Harvest Management Component, Sam Wright, Washington Trout, March 23, 2000
1.(A):
A PERSPECTIVE: on Fish Population Management Elements that Must be Addressed in Order to Achieve the Successful Recovery of Naturally Spawning Puget Sound Chinook Salmon, Sam Wright. Washington Trout, 2000