A Multi-year Assessment of the Marine Areas 8-1 and 8-2 Selective Chinook Fishery: 2005-2007: Final Working Draft

Two complete years of the Areas 8-1 and 8-2 "pilot" mark-selective fishery, including the monitoring/sampling programs needed for evaluation of the fishery, have been completed and a third year of the fishery is currently in progress. This multi-year report has been produced to review achievement of the purpose for implementing pilot selective Chinook fisheries in Areas 8-1 and 8-2 during the October-April 2005-06 and October-April 2006­07 seasons. The pilot fishery purpose is stated in the State-Tribal agreement documents (Northwest Treaty Tribes and the Washington Department of Fish and Wildlife, 2007):

"The purpose of the ‘pilot’ fishery is to collect information necessary to enable evaluation and planning of potential future mark-selective fisheries. The ‘pilot’ fishery provides a basis for determining if the data needed to estimate critical parameters can be collected and if the sample sizes needed to produce these estimates with agreed levels of precision can be realistically obtained."

These mark-selective fisheries were planned making assumptions about the performance of the fishery and how the fishery will affect wild (unmarked) and hatchery (marked) Chinook salmon. For example, the total number of marked and unmarked Chinook salmon encountered in these fisheries was estimated during the pre-season planning process using the Chinook FRAM and assumptions about fish abundance and angler effort levels. The sampling and monitoring programs in place for the "pilot" fisheries will aid verification of these assumptions. More fundamentally, results of the programs will be used to determine if the data needed to provide usable estimates of critical parameters can be collected.

These monitoring and sampling programs were designed to collect and provide data to estimate the following parameters, as listed in the State-Tribal agreement documents (Northwest Treaty Tribes and the Washington Department of Fish and Wildlife, 2007):

• the mark rate in the fishery: marked and unmarked encounters estimated by both on-the-water and shore-based programs;
• the incidence of partial adipose clips: estimated by both shore-based and on-water programs;
• the number of fish retained or landed: marked and unmarked fish estimated using a shore-based program, including CWT and scale-age sampling;
• the number of unmarked fish released: estimated by shore-based and on-water programs;
• the number of unmarked fish retained: estimated by a shore-based program and compared to enforcement program estimates;
• the number of marked fish released: estimated by a shore-based program in conjunction with on-water mark rate encounter estimates;
• the number of the chinook encounters that are of sub-legal size: estimated by shore-based and on-water programs;
• the stock composition of the mortalities: estimated by CWT recoveries via dockside sampling and DNA samples in the test fishery;
• estimates of marked and unmarked mortalities of double-index tag (DIT) and other CWT stocks.

With the exception of partial adipose-clip incidence (bullet 2) and DNA-based stock composition (bullet 8), we evaluate each of the above parameters in this multi-year review document. Additionally, we present analyses of several other parameters of significance to the evaluation and future management of selective Chinook fisheries.

This report was completed by WDFW, while incorporating extensive review and input from the Tribes. We review and analyze results of the monitoring/sampling program to evaluate if the intended objectives of the first two years of pilot fisheries in Areas 8-1 and 8-2 have been achieved. These objectives include: 1) collect information necessary to enable evaluation and planning of future potential Chinook mark-selective fisheries; and 2) determine if the data needed to estimate critical parameters can be collected and if the sample sizes needed to produce these estimates with agreed-to levels of precision can be realistically obtained. We initiated our review efforts with the intent of completing a thorough and timely evaluation that could help inform managers as they plan the 2008 season.

Our multi-year report contains two sections, each of which addresses separate aspects of the Areas 8-1 and 8-2 selective fisheries. In Section I, we present the modeling, sampling, and estimation methods that were employed in our evaluation of these two fisheries; provide resulting estimates of key fishery parameters; and discuss their patterns and significance on both a within-and between-area and season basis. In Section II, we address four topical questions relating to how the sampling, estimation, and modeling of the Areas 8-1 and 8-2 fisheries has been conducted over the past two seasons. These questions and their associated analyses are presented and discussed in a manner that aims to facilitate discussions for improved selective fisheries monitoring in the future.

Section I: Within-and Between-Year Patterns in Fishery Parameters

From October 1 to April 30 of 2005-6 (the "05-06 Season" hereafter) and October 1 to April 30 2006-7 (the "06-07 Season" hereafter), we implemented separate sampling programs in Areas 8-1 and 8-2 in order to collect the data necessary to estimate critical fishery parameters. For each area, the general study design was built around Murthy’s population-total estimator (Murthy 1957, Cochran 1977) and was focused specifically on obtaining daily estimates of total catch (landed and released) and total effort which could be expanded to weekly, monthly, and ultimately season-total values. Our sampling program incorporated comprehensive and complementary data collection strategies, including: 1) dockside-based angler interviews and catch sampling ("creel sampling"); 2) on-the-water total (instantaneous) effort surveys; 3) test fishing; and 4) voluntary reports of completed trips provided by charter boats and private anglers. We combined datasets collected through each of these sampling efforts within a rigorous estimation framework to characterize the behavior of the private recreational fleet (catch, effort, etc.) and characterize the overall impacts of the Areas 8-1 and 8-2 pilot selective fisheries.

Additionally, we quantified and analyzed the biological attributes (size and age) of landed catch sampled in the creel as during catch test-fishery sampling.

Creel Sampling Results

Estimates of total fishing effort, total landed catch, and average catch per unit of effort (CPUE) were remarkably consistent for the first two seasons of the pilot Areas 8-1 and 8­2 selective blackmouth fishery. Approximately 12,000 anglers participated in the combined fishery during both the 05-06 (12,495) and 06-07 (11,302) seasons; the majority of effort (two thirds) of effort occurred in Area 8-2. Within-season (i.e., month­to-month) effort patterns were also consistent between the two pilot seasons. On average, peak effort occurred in October in both areas, followed by a late-winter/early-spring effort peak (February-April). Only a limited amount of fishing effort occurred effort between November and the end of January.

Though nearly twice as many Chinook were harvested in Area 8-2 compared to Area 8-1, monthly average and season-total landed catch differed little between the 05-06 and 06­07 selective winter blackmouth seasons (pooled areas: 1,152 in 05-06, 1,210 in 06-07). Within seasons, there was limited Chinook harvest during October and November, followed by increased catches from December through to the end of the season; in both years, there was a February-March catch peak. Catch per unit effort (CPUE; estimated total landed catch / estimated total angler trips) averaged 0.10 Chinook retained per angler trip in both areas and years; however, there was evidence of considerable within-season variation in CPUE. Though total catch and effort were lowest at this time, the highest values of monthly CPUE were observed during mid-winter (Dec/Jan on average).

During creel interviews, dockside samplers measured the lengths of 1,215 marked, 15 unmarked, and 4 unknown mark-status Chinook that were harvested. From this, 99% (05-06: 596/601) and 98% (06-07: 619/629) of Chinook harvested from 8-1 and 8-2 combined were adipose clipped and 93% and 90% (05-06 and 06-07, respectively) retained marked fish were legal in size. With the exception of fish sampled in Area 8-2 during 06-07, there was little difference in Chinook total length between areas and seasons. However, we documented clear and systematic within-season size patterns whereby the monthly mean total length of landed-marked Chinook increased by 4 to 8 cm between October and April. The majority of marked salmon harvest consisted of individuals that were either 2 or 3 years in age, with little between-area and -year variation (80.1% in 05-06, 86.3% in 06-07; age-4 individuals accounted for the remainder of catch in both years (19.9% and 13.7% in 05-06 and 06-07, respectively).

The 05-06 and 06-07 pilot blackmouth seasons differed markedly in terms of estimated total Chinook releases. This result was consistent for both of the estimation approaches that we employed (i.e., "Method 1", relies solely on interview-based estimates; and "Method 2", relies on creel survey estimates of legal-marked retained Chinook expanded by test fishery proportions). Combining both areas and all release categories, between 4 (Method 2) and 7 (Method 1) times as many Chinook were hooked and released in the 06-07 season than during the 05-06 season. When apportioned to mark-status groups using test-fishery data, 5-9 (Method-1 to Method-2 range) times as many marked and 3-5 (Method1-Method 2) times as many unmarked Chinook were encountered during 06-07 compared to 05-06. Approximately 253-281 (Method 1-Method 2) unmarked and 267­301 (Method 1-Method 2) marked Chinook were encountered and released during each month of the 05-06 selective season, with little month-to-month variability. During the 06-07 season, 831-1,279 (Method 2-Method 1) unmarked and 1,515-2,438 (Method 2­Method 1) marked Chinook were encountered and released during each month on average, with October constituting the greatest number of releases for the season. Given the consistency of landed catch between areas and years, total 06-07 Chinook encounters (retained + released) were 4 (Method 2) to 7 (Method 1) times greater than for the 05-06 season.

Based on dockside sampling of landed catch and angler-reported release estimates for known mark-status groups, mark rates varied little between months and areas within years but considerably so between years. 2005-06 mark rates were 0.61 in Area 8-1 and

0.60 in Area 8-2; averaging an absolute 10% higher in 06-07, mark rates for the two respective areas were 0.71 and 0.73. Thus, between two thirds and three quarters of all Chinook encountered were visibly of hatchery origin.

Test Fishery Results

Over the two areas and two seasons, test fishers spent 2,476 hours and 496 days pursuing Chinook salmon for WDFW monitoring purposes. These efforts yielded a total of 3,727 Chinook encounters, the majority of which occurred during the 06-07 season. Monthly test-boat encounters averaged 133 across the two areas and seasons and ranged from 24 to 615. Using assumed mortality rates, we estimated total test-fishing impacts at 715 Chinook mortalities (253 unmarked, 462 marked) for the two areas and seasons, the majority of which were for the 06-07 season.

The size/mark-status composition of test-fishery encounters was similar between the two areas, but differed markedly between seasons. For 05-06, the overall mark rate (i.e., marked encounters / all encounters) was 0.58 in 8-1 and 0.62 in 8-2. In 06-07, values were higher in both areas, at 0.65 and 0.67, respectively. Legal mark-rates (i.e., legal-marked encounters / all legal encounters) were even more disparate between years: 8-1 test-fishery estimates were 0.62 in 05-06 and 0.72 in 06-07; 8-2 legal-mark rates were

0.56 in 05-06 and 0.79 in 06-07. Although the size/mark-status composition of test-fishery encounters was varied from month to month, there was a tendency towards an increased legal-sized proportion towards the close of the fishery.

We analyzed length data for Chinook encountered in the Areas 8-1 and 8-2 test fisheries and found that a significant proportion of total-length variation was due to area, season, mark-status effects. In particular, we documented a trend towards smaller Chinook sizes during 06-07 relative to 05-06 â€" especially for Area 8-2. We also found that between 6 and 10% of all encountered marked Chinook were within 2 inches of the legal length limit (i.e., 20 < x < 22 in). Finally, we the average size of test-boat encountered Chinook increased as the season progressed during both years, with mean total length of marked fish increasing from 35-40 cm to approximately 50 cm over the seven month test fishery.

Similar to mark-rates and other fishery attributes, we found little difference in the age composition of test fishery encounters (marked and unmarked) within seasons and between areas but considerable differences between seasons. In particular, there was a clear shift towards increased age-1 and age-2 relative abundance in 06-07 compared to 05-06. In 05-06, 55% of marked and 63% of unmarked encounters were age 2 or less; in 06-07, these same two age (1 and 2) classes comprised 72 and 81% of all marked and unmarked Chinook encountered in the test fishery.

Total Fishery Impacts

We estimated total mortality due to the combined 8-1/8-2 selective fishery by combining creel-based estimates of Chinook encounters, test-fishery data on the size/mark-status composition of the pool of fishable Chinook, and agreed-to selective fishing mortality rates (sfm). For the 05-06 season, total Chinook mortality for the combined fishery was estimated at 1,840 (based on Method 1 encounters) to 1,941 (based on Method 2). 06-07 mortality was estimated to be 2-3 times greater than the 05-06 season, with estimates ranging from 4,481 (Method 2) to 6,311 (Method 1) for this latter season. During both seasons, the majority of mortality was comprised of marked (relative to unmarked), sublegal (relative to legal), and Area 8-2 Chinook (relative to 8-1). In an attempt to characterize selective fishery impacts in a manner independent of assumed sfm values, we also evaluated released-to-retained ratios for the Areas 8-1 and 8-2 fisheries for both seasons. Released-to-retained corroborate that the 8-1/8-2 fishery had substantially greater impacts during 06-07 compared to 05-06. During the first pilot season, an average of 2-3 unmarked and 1-3 total (marked and unmarked) releases occurred for each Chinook retained. In 06-07, estimates averaged 21-24 total and 7-9 unmarked releases per kept fish, respectively.

Based on coded-wire tag (CWT) recoveries (unexpanded), Puget Sound hatchery stocks comprised the majority of marked, tagged Chinook harvested during the 05-06 and 06-07 selective seasons. Out of the 209 CWTs recovered during the first two pilot seasons, only three came from hatcheries from outside of Puget Sound (two from Canadian facilities and one from the Columbia River). During the 05-06 season, 29 of 101 CWT recoveries were double index tags (DITs); 20 of 108 CWTs recovered in 06-07 were DITs. Unmarked-DIT mortality estimates (using λ at release) due to selective fishing were low for both seasons. We estimated that 9 and 5 unmarked-DIT Chinook perished as a result of the 05-06 and 06-07 selective seasons, respectively.

Angler Compliance and Enforcement Summary

For the two pilot seasons that Areas 8-1 and 8-2 were under mark-selective rules for Chinook retention, available information suggests that angler compliance with regulations was quite high. For anglers sampled at dockside, we estimated an unmarked retention error (no. unmarked [legal and sublegal] Chinook landed / no. unmarked [legal and sublegal] Chinook encountered) of 0.0% and 0.9% for 05-06 and 06-07 in Area 8-1 and 5.2% and 1.0% in Area 8-2 during the same respective seasons. Yearly enforcement reports compiled for the North of Falcon season-setting process corroborate these sample-based estimates of compliance. Overall compliance with salmon rules for Area 8-1 was 95.7% for 2005 and 97% for 2006 and there were no citations issued for possession of wild Chinook. In Area 8-2, compliance with salmon rules was 86.6% during 2005 and 90% for 2006, and three fishery-related arrests were made during the latter season (two for wild Chinook and one for over-limit [salmon] possession).

SECTION I: SUMMARY AND DISCUSSION

Based on two years of experience with implementing and intensively monitoring the pilot Areas 8-1 and 8-2 mark-selective blackmouth fisheries, we note and conclude the following:

• Monthly and season-total patterns fishing effort, CPUE, and total Chinook landings were relatively stable for the two areas and years.
• The first two pilot seasons differed considerably in total estimated impacts, due primarily to increased sublegal-sized Chinook (marked and unmarked) abundance.
• The combined Areas 8-1 and 8-2 selective fishery generally operated at or below expected (i.e., FRAM-modeled) level of impact.
• The impacts of the Areas 8-1 and 8-2 selective fisheries on the coast-wide CWT programâ€"assessed in terms of estimated capture-and-release mortalities inflicted upon unmarked-DIT Chinook encounteredâ€"were minor for both seasons.
• Estimated mark rates were high relative to what is deemed acceptable for implementing successful mark-selective fisheries.
• Dockside data and WDFW-Enforcement summary reports indicate that anglers closely followed mark-selective Chinook harvest regulations during both seasons of the pilot fisheries.

Section II: An Assessment of Selective Fishery Sampling and Analysis Methods

To better understand the quality of existing monitoring data and to guide future work, we addressed four topical questions relating to how the planning (i.e., Fishery Regulation Assessment Model application), sampling, and evaluation (i.e., data analysis) of the Areas 8-1 and 8-2 fisheries has ensued over the past two seasons:

1. Have the sampling programs performed at a level sufficient to characterize fishery impacts within acceptable bounds of precision?
2. Have the test-boat anglers succeeded at emulating the private recreational fleet, in terms of fishing methods and Chinook encounters (i.e., size/mark-status composition)?
3. Which method [i.e., "Method 1" (creel-only based) or "Method 2" (creel-based landed catch expanded by test fishery proportions)] is most likely to yield unbiased estimates of total Chinook encounters?
4. How well has the Fishery Regulation Assessment Model (FRAM) performed in planning the combined Areas 8-1 and 8-2 selective Chinook fisheries?

Question 1: Adequacy of the Areas 8-1 and 8-2 Selective Fishery Sampling Program

To answer Question 1, we: 1) characterized the intensity of sampling efforts in both Areas 8-1 and 8-2, 2) evaluated the adequacy of dockside and test-fishery sampling programs relative to pre-determined and agreed-upon sample-size objectives, 3) described the relative precision of key quantities estimated from sample-program data, and 4) evaluated the effects of reduced sampling on the precision of season-wide estimates of test-fishery parameters.

During the first two seasons of the 8-1 and 8-2 selective fisheries, we directly sampled 4,950 angling parties, yielding data on a total of 9,580 angler-trips and 11,223 Chinook encounters. We sampled Chinook encounters (retained and released) and fishing effort at a level commensurate with the stated goal (100 encounters per month), with few exceptions. Relative to sample-rate objectives defined for CWT sampling in selective Chinook fisheries, we met our target (20% of all harvested Chinook) for 25 of 28 Area-month combinations. We were also successful at sampling completed fishing trips at a high rate (20-50%). Finally, coefficients of variation (CVs) for season-total and monthly estimates of fishing effort, Chinook landings, and released Chinook encounters averaged 10-20%. Overall, these findings illustrate that the dockside component of our monitoring program is successful at achieving sampling objectives and delivering precise estimates of catch and effort.

Relative to Question 1, we also assessed the ability of our test-fishing program to meet specified objectives. As test-fishery encounters consistently exceeded the stated objective of 100 Chinook encounters per management regime, we evaluated whether or not opportunities exist for scaling back efforts without significantly compromising the precision of parameter estimates. This re-sampling exercise demonstrated that the variance around test fishery-based estimates of mark rates and legal-marked proportions decreases with increasing sampling intensity, but not at a constant rate. The sharpest variance reductions were observed for sample rates that were 10-40% of the present level; variance decreased little at sample rates that were 50% or greater. Thus, clear opportunities exist for scaling back test fishing efforts without significantly compromising the precision of estimates.

Question 2: How well does the test fishery emulate the private recreational fleet?

The test-fishing component of the Areas 8-1 and 8-2 selective fisheries monitoring program supplies critical information used for fishery characterization and total encounters and mortalities estimation. In using an experimental fishery to fulfill these data needs, we have by default assumed that the size/mark-status composition of test-fishery Chinook encounters approximates that experienced by the private recreational fleet.

While emulating the fleet is generally achieved in practice, we formally addressed Question 2 by comparing parameters describing the composition of Chinook encounters between test-fishery and private-fleet datasets. For all Chinook encounters, we compared overall mark rates between test-fishing and dockside datasets; for known mark-status fish, test-fishery and dockside-based estimates of overall mark rates were virtually identical for both areas during 05-06 but not 06-07. We separately compared mean sizes and length-frequency distributions between test-fishery legal-marked Chinook encounters and those retained by anglers that were inspected during creel surveys for each Area-season combination. While lengthâ€"frequency distributions were similar in shape, lengths differed for 3 of 4 test-fishery vs. fleet comparisons; test-fishery lengths were 1-2 cm smaller than those estimated for the fleet. Finally, we compared the age composition of legal-marked Chinook observed at dockside and sampled in the test-fishery. From this, the age composition of legal-marked Chinook encountered in the test fishery appears similar to that experienced by the private fleet. With some comparisons illustrating similarities and other suggesting differences in measured attributes of Chinook encounters, it remains equivocal as to whether or not the 8-1 and 8-2 test fisheries perfectly mimic the private fleet in its angling behavior. For this reason, future evaluation may be necessary to completely answer Question 2.

Question 3: Does Method 1 or 2 provide a better estimate of total encounters?

To answer Question 3, we evaluated: i) Method-1 and -2 total-encounters estimators and their associated assumptions, ii) the sensitivity of estimators to assumption violations, and iii) the validity of assumptions based on indirect evaluations using empirical data. Method 1 (M1, sum of creel-based estimates for all Chinook encounters categories) and Method 2 (M2, creel-based estimate of legal-marked Chinook landed catch expanded by test-fishery legal-marked proportion) differ computationally and in terms of the assumptions they require for accurate encounters estimation. M1 accuracy relies on the ability and/or willingness of anglers to accurately recall and/or report caught-and-released Chinook encounters (Assumption 3). The accuracy of M2 estimates depends on whether or not anglers report all legal-marked Chinook encountered (Assumption 5) and the extent to which the size/mark-status composition of test-fishery encounters mirrors that seen by private anglers (Assumption 6).

Our M1 vs. M2 sensitivity analysis revealed that: i) when Assumptions 3 and 5 are not met, M1 and M2 estimates are affected similarly, ii) estimates are most sensitive to Assumption 6 departures, and iii) due to compensating effects, M2 has the potential to yield accurate encounters estimates when both Assumption 5 and 6 are imperfectly met. Next, we considered available empirical evidence to gauge the plausibility of Assumptions 3, 5, and 6. For Assumption 3 ("Anglers accurately report released Chinook encounters"), we reviewed pertinent literature, considered patterns in M1 relative to M2 estimates, and inspected raw interview data (i.e., releaseâ€"frequency distributions). Based on this, we concluded that Assumption 3 is unlikely to be perfectly metâ€"particularly during high-encounters periodsâ€"and that in general anglers probably over-report released Chinook encounters. Though few data exist for evaluating Assumption 5, available information suggests that it is violated to a minor degree. Based on voluntary trip reports, we estimate that anglers may release as many as 10% of the legal-marked Chinook that they encounter. Finally, we considered the likelihood of meeting Assumption 6 under Question 2 above. Though this evaluation did not provide uniform support indicating that Assumption 6 is perfectly met, initial findings suggest that it is reasonably approximated but should be assessed further in the future.

Question 4: FRAM vs. Observed Estimates of Selective Fishery Parameters

In this section we evaluated how well the Fishery Regulation Assessment Model (FRAM) predicted fishery outcomes (landings, encounters, mortalities) and we evaluated modeled selective fishery parameters relative to empirical estimates from creel surveys (hereafter referred to as "observed" values). Evaluated parameters include: i) encounters by size (legal-size and sublegal-size) and mark status (marked and unmarked) and associated mortalities; ii) landed catch (i.e., Chinook that are kept); iii) unmarked retention error (legal unmarked kept/legal-unmarked encounters); iv) mark release error (legal-marked released/legal-marked encounters); v) unmarked sublegal retention error (sublegal unmarked kept/sublegal-unmarked encounters); and vi) marked sublegal retention error (sublegal marked kept/sublegal-marked encounters).

FRAM’s prediction of total Chinook encounters during the 2005-06 season was more than three-fold higher than Method 1 and 2 creel survey estimates. For the 2006-07 season, the FRAM estimate of 19,062 total Chinook encounters fell within the range of total Chinook encounters estimated via Methods 1 and 2. For both seasons, FRAM overestimated unmarked Chinook encounters. FRAM overestimated marked Chinook encounters in 05-06 for all categories; 06-07 modeled encounters for marked fish were an underestimate relative to observed values, with the exception of Chinook landings (which were over-predicted by FRAM). For both seasons, predicted (FRAM) vs. observed (creel) mortality comparisons yielded results that were comparable to those observed for Chinook encounters.

In addition, we considered FRAM’s ability to predict total Chinook encounters and landed catch by comparing predictions to historical (1994-2005 for encounters, 1989­2005 for landed catch) estimates derived from a combination of CRC harvest estimates and Baseline creel sampling information about released salmon. FRAM encounters predictions were lower than the CRC-based 11-year average but well within the 95% confidence interval for this parameter. 05-06 creel estimates were approximately five­fold lower than the average estimate of Chinook encounters, whereas the 06-07 estimates (Method 1 and 2) straddled the historical average. Observed total Chinook landings, when adjusted to make them comparable to historical non-selective values, were consistently less than historical levels and FRAM predictions. Despite this variability, overall FRAM performed relatively well in predicting total Chinook encounters for average years.

In addition to comparing predictions to observations, we also compared parameter values used in modeling to empirical (creel) estimates. First, FRAM uses an unmarked retention error (legal unmarked retained / total legal unmarked encountered) rate of 8% to calculate the number of unmarked legal-size fish that are retained in a selective fishery. Creel estimates of unmarked retention error for 05-06 were 5.3-5.4%, whereas 06-07 season estimates were 3.4-9.2%. Second, mark release errorâ€"defined as the number of legal-marked Chinook released divided by legal-marked Chinook encountersâ€"is modeled at 6% in FRAM. Creel-based estimates of legal-marked release error (Method 1 only) were estimated at 8.5% during the 05-06 season and 55.6% during the 06-07 season of the Areas 8-1 and 8-2 selective Chinook fishery. While the 8.5% creel-based value for the 2005-06 season is comparable to the 10% value obtained from the voluntary trip reports, we believe the 06-07 estimate is unrealistically high and probably an artifact of the creel interview process (See Question 3 above). Finally, while neither unmarked nor marked sublegal retention error (sublegal Chinook retained for a given mark-status category / sublegal Chinook encountered for a given mark-status category) is modeled in FRAM (i.e., algorithms assume no sublegal fish are retained), creel survey estimates of unmarked sublegal retention error were 0.0% and 0.1% for 05-06 and 06-07, respectively; marked sublegal retention errors were 0.7% and 4% for 05-06 and 06-07 seasons, respectively.

SECTION II: CONCLUSIONS and RECOMMENDATIONS

Question 1: Sampling Adequacy

• Dockside sampling and test-fishery efforts were successful at achieving agreed-to sampling objectives.

• Dockside sampling and test-fishing efforts yielded precise estimates of key fishery parameters.

• Sampling efficiencies should be pursued where possible, assuming such efficiencies do not affect the integrity/reliability of estimates. As a start, we recommend that a single test fishing vessel be shared between Areas 8-1 and 8-2 to achieve cost savings and sampling efficiencies.

Question 2: Test Boats Emulating the Fleet?

• Whether or not the Areas 8-1 and 8-2 test fisheries perfectly mimic the private fleet remains equivocal. We characterized the ability of test-boat anglers to fish like the fleet and demonstrated similarity in some fishery parameters but we also found evidence of small but statistically significant differences in other parameters.
• Opportunities for improved and more efficient collection of test fishing data should be considered in the future. For example, as instituted in November 2007, spatial evaluations of test-fishery and private-fleet effort patterns should be pursued for both in-season guidance and post-season evaluation.
• Given that it is the most reliable (i.e., in terms of control over how data are collected, logged, etc.) dataset on Chinook encounters available and the lack of strong evidence suggesting otherwise, we recommend that the analytical assumptions associated with test fishery data be accepted at the present time. If discrepancies are detected in future analyses, appropriate measures should be taken to modify sampling and/or correct for biases.

Question 3: Evaluating Method 1 versus Method 2

• With the existing sampling program and Methods 1 and 2 as starting points, WDFW and tribal co-managers should work towards a mutually agreeable encounters and mortalities estimation framework.
• The actual percent of released marked legal-size fish remains an unknown parameter. We recommend modifying the dockside creel surveys to query anglers specifically about how many marked legal-size fish they released.

Question 4: Evaluating FRAM vs. Observed Estimates of Selective Fishery Parameters

• FRAM predicted total Chinook encounter estimates that were within the range of historical encounters but sometime over-and under-predicted encounters in particular years. Given this variability, we believe adjustments to the inputs and methods by which FRAM predicts encounters are unwarranted at his time.
• FRAM overestimated unmarked Chinook encounters during both seasons of the selective Chinook fishery in Areas 8-1 and 8-2, when compared with both Method 1 and Method 2-based creel estimates.
• FRAM overestimated landed catch of unmarked and marked Chinook for both seasons, when compared with both Method 1 and Method 2-based creel estimates.
• FRAM is not designed to estimate sublegal retained catch. However, creel survey estimates produced from the 2005-06 and 2006-07 seasons in Areas 8-1 and 8-2 provided low estimates of unmarked sublegal retention error, which are considered to have a minor impact on exploitation rates, especially after being converted to adult-equivalency. To account for sublegal retention error in FRAM would require a major restructure to program catch algorithms, which we do not recommend at this time.
• Currently the exploitation rate scalars in FRAM characterize fishing power during 1989-1993 as estimated in FRAM post-season runs relative to FRAM base period "catch" and stock abundances used in the 2002 and 2005 model calibrations. We recommend continuing the current method of developing fishery input scalars for at least one more year until a pattern is apparent.
• Based on two seasons of observed results, we recommend reducing the FRAM input parameter for unmarked retention error to a value of 6%, to calculate the predicted number of unmarked legal-size Chinook that are retained in a selective fishery.
• We recommend increasing the FRAM input parameter for mark release error to a value of 10%, based on the two seasons of observed results in Areas 8-1 and 8-2.
• FRAM currently models 150 encounters per test fishing boat and month. The average number of actual test fishing encounters per area and month was very close to the modeled number of encounters. We recommend continuing to model 150 Chinook encounters per test fishing boat and month.