A multi-year assessment of the Marine Areas 5 and 6 selective Chinook fishery: 2003-2007: Final Working Draft

Introduction

Five years of the Area 5 and 6 "pilot" mark-selective Chinook salmon (Oncorhynchus tshawytscha) fishery, including the monitoring/sampling programs needed for evaluation of the fishery, have been completed. This multi-year report has been produced to review achievement of the purpose for implementing pilot selective Chinook fisheries in Areas 5 and 6 during the 2003 through 2007 seasons. The pilot fishery purpose is:

"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 monitoring and sampling programs were designed to collect and provide data to estimate the following parameters:

• the mark rate in the fishery;
• the incidence of partial adipose clips;
• the number of fish retained or landed;
• the number of unmarked fish released;
• the number of unmarked fish retained;
• the number of marked fish released;
• the number of the Chinook encounters that are of sub-legal size;
• the stock composition of the mortalities;
• estimates of marked and unmarked mortalities of double-index tagged (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 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 levels of precision can be realistically obtained.

During the summers of 2003 through 2007, a selective Chinook recreational fishery was implemented in waters of the Strait of Juan de Fuca including Marine Area 5 and the western portion of Marine Area 6 (hereafter: Areas 5 and 6). Each year the fishery was scheduled to start in early July and run continuously until either the quota of harvested Chinook was attained or a set number of days was reached, whichever came first. Anglers were allowed to retain two marked (adipose fin clipped) Chinook salmon > 22" (56 cm) as part of their daily limit, and were required to immediately release, unharmed, any unmarked Chinook caught. During the Chinook Selective Fishery anglers were also allowed to retain pink (O. gorbuscha), sockeye (O. nerka), and marked hatchery coho (O. kisutch) salmon.

During the summers of 2003 through 2007, we implemented separate sampling programs in Areas 5 and 6 in order to collect the data necessary to estimate 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.

Creel Sampling Results

Over the 5 years of study, the combined Areas 5 and 6 fishery lasted from 30 to 49 days. The harvest quota was obtained each year, except for 2005. Total fishing effort averaged 22,000 34,000 angler trips per year (angler trips and anglers are used interchangeably throughout the document) and varied as a function of season length and catch fishing success. Chinook harvest ranged from 2,078 to 4,096 and was within 5% of the quota during years when it was met. On average, 81% of the Chinook harvest occurred in Area 5. Estimated total released Chinook encounters ranged from 6,408 to 14,841, the majority of which occurred in Area 5 each year. The number of Chinook released for every Chinook harvested declined by ~50% (from 4.2 to 2.1) across the five years of the fishery. Chinook harvest per angler (C/F) ranged from 0.06 to

0.19 and averaged 0.13 for all 5 years. For all legally harvestable salmon species combined (i.e., Chinook, coho, and pink), C/F ranged from 0.18 to 0.81; angling effort appeared to be correlated with total (all salmon species) C/F but not C/F for any particular salmon species.

Based on dockside sampling of landed catch and angler-reported release estimates for known mark-status Chinook, overall Chinook mark rates (legal + sublegal) were consistently higher in Area 6 than Area 5 and increased consistently from 2003 (0.24) to 2007 (0.45). The percentage of harvested Chinook that were unmarked (sublegal-size and legal-size) ranged from 0.14% to 3.03%.

During the five seasons, dockside samplers measured the lengths of 3,517 Chinook. Harvested Chinook in Area 6 were signficantly larger than those taken in Area 5. Over 92% of the Chinook harvested were legal-size and marked; 4-7% of Chinook harvest was sublegal-size and 0-3% were unmarked.

Test Fishery Results

Test boat samplers averaged 37 days on fishing annually in Area 5 and 40 days in Area 6, yielding over 1,000 Chinook encounters. Samplers fished predominantly using downriggers (>69%), as this was the predominant private-fleet fishing mode, and caught over 90% of their Chinook using this method. Season-total Chinook encounters averaged 266 for the pooled areas and total mortalities attributable to test fishing ranged from 25 to 82 annually. The majority of Area 5 test-fishery encounters were legal-size, except during 2003; Area 6 encounters were almost exclusively legal-size. Test-fishing data indicated that marked proportions were higher in Area 5 than in Area 6 in all years; Area-5 mark rates increased over the last 5 years but showed no apparent trend in Area 6. Although Chinook mark-status/size proportions differed markedly between areas, both showed progressive increases in the legal-size and marked proportion over the course of the study. Chinook encountered by test boats were significantly larger in Area 6 than in Area 5.

Voluntary Trip Report Results

The number of Chinook reported on Voluntary Trip Reports (VTRs) varied dramatically over the 5 years of the fishery, ranging from 37 (2006) to 213 (2003). VTR-based estimates of legal-size Chinook mark rates ranged from 20 to 100% and, similar to test boats, VTRs suggested that this value was higher in Area 6 than Area 5. Further, VTRs indicated that very few (< 20%) sublegal-size fish were present in Area 6. Marked, legal-size Chinook release rates estimated from VTRs ranged from 0 to 14% and averaged 5% for all years and both areas.

Encounters and Total Mortalities

Annual encounter estimates for both areas combined ranged from 8,558 to 18,662 using Method 1 and 6,362 to 13,476 using Method 2. Method-1 estimates of total encounters for the combined areas were consistently higher Method-2 values for all years. Season-total mortality estimates (harvest and release) for the pooled areas ranged from 3,465 to 6,356 using Method 1 and 3,078 to 5,449 using Method 2. Estimated total (both methods) unmarked encounters and unmarked mortalities generally declined across the 5-year study period. The ratio of unmarked mortalities (Method 1 or 2) to harvested marked legal-size Chinook dropped steadily from 2003 through 2007, e.g. from 0.73 to 0.32.

CWT analysis

Over 540 coded wire tags (CWTs) were collected during the Areas 5 and 6 selective Chinook fisheries. Puget Sound and Columbia River stocks contributed the highest proportion of CWTs. Only five of the recovered CWTs were from stocks originating from rivers on the Washington side of the Strait of Juan de Fuca. The number of Double Index CWT recoveries ranged from 33 to 41, which translates into an unmarked DIT mortality estimate that ranged from 11 to 16.

Enforcement

The number of contacts made by enforcement officers ranged from 439 to 846 annually. Of those contacts, the proportion with sublegal-size Chinook was less than 0.01 for all areas and years. The proportion of contacts with unmarked Chinook ranged from 0.00 to 0.03.

Catch and Effort

The Areas 5 and 6 selective Chinook fisheries were driven by catch rate. During years that fishing was good, angler trips were up; during years that fishing was poor, angler trips were down. Surprisingly, the Chinook catch rate does not appear to be the main factor, but rather it appears that the overall salmon catch rate is the main factor responsible for how many angler trips are expended each season.

Selective fisheries effort was higher than effort during previous non-selective periods. For 2006 and 2007, the selective fisheries effort in Area 5 was lower than the effort in 2001, but was higher than the effort in 2002. Results of this study suggest that given the low catch rate of coho in 2003 through 2007, the addition of the selective Chinook fishery increased effort (angler trips) in each year of the fishery relative to what effort would have been without the selective Chinook fishery. Effort in Area 5 increased over what was seen during the 1994-2000 period, when no Chinook retention was allowed and coho fishing was closed at times. However, effort was considerably less than that seen for the 1984-1993 period. Effort in Area 6 does not show an increase compared to the 1994 through 2000 period. Opening a selective Chinook season did not increase effort to levels that were higher than or even near historical values.

Another issue that concerned managers and anglers prior to implementation of the 5/6 fishery was whether or not Chinook salmon mark rates would be adequate for successful fishing under mark-selective regulations; legal-size Chinook mark rates for 5/6 have been good and increased in recent years. Anglers have been able to retain 50% of all legal-size Chinook encountered.

Average daily havest has ranged from approximately 50 to 115 fish per day. For future fisheries planning, assuming 100 Chinook harvested per day would be a good conservative estimate. Assuming 100 fish harvested per day for a thirty-day fishery would equate to 3,000 Chinook. For comparison, if the actual number harvested was 116 fish per day (the highest value observed), the harvest would be 3,480 or 16% over the predicted value.

CWT Analyses

Based on CWT recoveries, the Area 5 and 6 selective Chinook fishery appears to impact mainly Puget Sound and Columbia River stocks. For Strait of Juan de Fuca hatchery stocks, less than 1 percent of all CWT recoveries occur in Washington recreational fisheries; in contrast, nearly 29 percent of the recoveries occurr in Canada and Alaska. Based on our estimates of unmarked DIT Chinook mortalities, the overall bias introduced to the CWT program due to this fishery is extremely low.

Enforcement Compliance Compared to Creel Compliance

For most areas and years, creel-survey results suggest a higher proportion of either unmarked or sublegal-size Chinook retention than enforcement reports indicate. However, both creel survey and enforcement data suggest compliance was high (90% during all years and in both areas).

Sampling intensity-related questions:

In general, we successfully met sampling objectives. During the five study years, the precision of both harvest and effort estimates approached or exceeded the 0.15 precision objective. Sample-size objectives for dockside encounters (100 / month) were met in most cases, the exception being Area 6 (2005-2007 during August). Weekly sample rates (n fish examined / estimated harvest) exceeded the CWT sample-rate goals, ranging from 0.154 to 0.544 in Area 5 and 0.162 to 0.777 in Area 6. At the season-total level, sample rates ranged from 0.227 to 0.276 in Area 5 and from 0.326 to 0.558 in Area 6. Finally, test-fishery sampling objectives were generally well met in Area 5, but not Area 6.

Comparing Private Fleet, Test Fishing, and VTR data

A key assumption of our monitoring program is that the test-fishery and private-fleet encounter composition (i.e., frequency by size/mark-status class) is the same (Assumption 6). To evaluate this assumption, we compared the mark rates, length, and mark-status/size composition of fish caught by the private fleet (from creel surveys and VTRs) and and test fishers.

Lengthâ€"frequency distributions (for legal-marked Chinook) were similar when compared between test fishery and creel samples within areas and years, but remarkably different when compared between areas and within sampling methods and years. Thus, while both the test fishery and fleet "sampled" legal-marked Chinook in a manner that could discriminate gross differences between areas, they produced statistically indistinguishable length results within areas. Mark rate and mark-status/size comparison results were more variable than those from length comparisons. First, overall mark rates estimated from creel surveys occasionally differed with those estimated from both test-fishery and VTR datasets during most years in Area 5, but only one year in Area 6. Second, test-fishery and VTR mark-rate estimates differed infrequently. Third, for the majority of area-year combinations, legal-size Chinook mark rate estimates produced from test-fishery and VTR data were statistically indistinguishable. Finally, markstatus/size comparisons suggested similarlity between groups but produced inconsistent results.

Estimation of Total Encounters, Method 1 versus Method 2

To determine whether Method 1 or Method 2 provides a more accurate estimate of total Chinook encounters in selective fisheries, 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 approximately 5% (range: 0-14%) of the legal-marked Chinook that they encounter. Finally, we considered the likelihood of meeting Assumption 6 in our test fishery vs. creel/VTR comparison described above. This evaluation suggested that Assumption 6 is reasonably met in the Areas 5 and 6 sampling program.

FRAM Performance in Selective Fishery Planning

FRAM predictions were relatively accurate for the Area 5 and 6 selective Chinook fishery. Whereas estimated marked legal-and sublegal-size Chinook catch often exceeded FRAM predictions, unmarked legal-size Chinook catch never exceeded predicted values. Unmarked sublegal-size fish exceeded the FRAM predictions (zero harvest) in 3 of 5 years. Total unmarked landings were 1 to 26% of FRAM predictions.

For encounters, Method 1 estimates exceeded model predictions in 4out of 5 years for marked legal-size and 1 out of 5 years for unmarked sublegal-size fish. Despite the fact that Method 1 estimates are likely biased high (see M1 vs. M2 section), estimated total unmarked Chinook encounters never exceeded FRAM predictions. Method 2 estimates of total marked legal-size encounters exceeded FRAM predictions less frequently. Total unmarked encounters estimates (Method 2) never exceeded model predictions, though unmarked sublegal-size encountes did so in 1 year. For mortalities (harvest + release mortality), Method 1 estimates exceeded FRAM in most cases for marked legal-size and total marked fish and in one year for marked and unmarked sublegal-size fish. Despite the fact that Method 1 estimates are likely biased high, estimated total unmarked Chinook mortality never exceeded FRAM predictions. Across the four mark-status/size categories, comparisons of Method 2 mortality estimates with FRAM predictions yielded similar results as the encounters comparsions.

For selective fishery parameters used in model runs, unmarked retention error ranged <1% to 2%, well below the FRAM value (8%). Marked release error ranged 24-37% (M1 estimates), much greater than the FRAM value (6%). Whereas unmarked and marked sublegal-size retention error are modeled as zero in FRAM, empirical estimates for these respective parameters were 0-8% and 6-19%.

Sampling Intensity

Our monitoring and sampling programs were designed to collect the data needed to reliably estimate several selective fishery parameters. With few exceptions, our monitoring program was effective at sampling the 5 and 6 selective Chnook fisheries. Harvest estimates met the 15% precision objective in all years that the quota was achieved, effort estimates always met the objective; and the CWT sample rate goal was always met. Based on these results, we believe that our dockside program for summer selective Chinook fisheries with quotas should remain unchanged. Although sampling success was high for the Area 5 test fishery, Area 6 test fishers met sampling objectives less than 50% of the time. However, the objective of 100 encounters is probably unrealistic for Area 6 given that total fleet encounters ranged 683-1,614 during years when the goal was not met. An alternative test fishing objective for short duration, low catch rate fisheries should be investigated.

Comparing Private Fleet, Test Fishing, and VTR data

Based on our results, we conclude that test boat catches are representative of angler catches for the following reasons:

• The mean lengths and lengthâ€"frequency distributions of legal-size marked Chinook caught by test fishers were similar to those for Chinook caught by private fleet anglers.
• Length samples acquired via test fishing and from the private fleet (creel) both displayed clear between-area (within year) differences.
• Test fishery and VTR estimates of overall mark rates were similar.
• Legal-size Chinook mark rate estimates from VTRs and the test fishery were similar; where differences occurred, they were not in a single and consistent direction (i.e., +/-)
• Mark-status/size composition estimates from VTRs differed from test-fishery estimates in only 3/10 year/area comparisons.

Prior to implementation of the 2003 selective Chinook fishery, managers identified several questions about the magnitude and impacts of such fisheries that needed to be addressed through monitoring and evaluation.

Fishery Monitoring and Results

The use of the Murthy type estimator and test fishing worked well to describe the fishery. Opening of this selective Chinook fishery did not lead to effort levels that are substantially higher than historical levels and in fact, effort was well below historical levels. The mark rate during this fishery ranged from about 40-60%, and for legal-size fish from about 35-65%, which increased over time. The proportion of sublegal-size fish in Area 5 dropped from 0.54 in 2003 to

0.33 in 2007, and never exceeded 0.06 of the catch in Area 6. The number of fish released per landed dropped throughout the duration of fisheries from 4.2 to 2.0. Total fishing related Chinook mortalities ranged from 2,839 to 6,193. Puget Sound and Columbia River origin stocks comprise the bulk of the fishery. Very few Strait of Juan de Fuca origin stocks are caught in this fishery. Very few DIT fish are caught in this fishery and the effect on the DIT mark rates appears undetectable. Angler compliance exceeded 90% at all times

Use of FRAM to Predict Selective Fishery Impacts

FRAM encounter predictions were generally higher than analogous creel estimates. They were on average 16% higher than Method 1 and 64% higher than Method 2 estimates; however, FRAM tended to underestimate marked, and overestimate unmarked, encounters and catch. Mean unmarked retention error estimates ranged 0.7% (Method 1) to 0.9% (Method 2), with no single year/method exceeding 2%; FRAM's value (8%) significantly exceeds these estimates. FRAM uses 6% for legal-marked release error in selective Chinook fisheries. Suvey estimates for this parameter ranged from 24-37%, with a mean of 28%; VTRs yielded 5% as an estimate. FRAM models 150 encounters per test fishing boat per month. The average number of actual test fishing encounters per area and month was 157 in Area 5 and 71 in Area 6.

Method 1 versus Method 2

Though it is impossible to know with certainty the true number of Chinook salmon encountered in a particular fishery, both Method 1 and Method 2 have the potential to yield biased estimates. For this reason, it may be more productive to define the set of conditions under which one method is expected to yield better (i.e., less biased) estimates than the other and/or determine defensible means for adjusting for measurable biases when they occur.

Length and Duration of Monitoring

Very little additional knowledge was gained after the first three years of monitoring and evaluation. Since catch per effort (C/f) can be computed from baseline sampling, it could be used along with relative changes in effort to monitor gross changes in the fishery in lieu of the intensive sampling that has occurred to date.

Conservation Objectives

The estimated mortalities of unmarked Chinook were less than predicted in FRAM models used during the pre-season planning process for every year of the fishery.

• With the existing sampling program and Methods 1 and 2 as starting points, WDFW and tribal technical staff should work towards a mutually agreeable encounters and mortalities estimation framework.
• The dockside interview process should be modified to quantify the extent of intentional legal-marked Chinook release activity for the entire recreational fleet.
• In areas with sufficient test boat samples, VTRs add relatively little additional information. However, VTRs can provide useful information on mark rate and sublegalsize to legal-size ratios when test fishing is not conducted.
• We recommend utilizing the most efficient method of catching fish on test boats in order to boost sample size and increase precision.
• With the high mark rate of legal-size Chinook (40-60%), the low rate of sublegal encounters, the absence of local stock CWTs, and the low number of fish released per harvested fish, the Area 6 fishery would be a very good choice for expanded angler opportunity.
• We recommend a maximum of 3 years of monitoring for short-duration (less than 3 months) selective fisheries unless inter-year variation suggests additional years of monitoring are necessary.
• Adjust the FRAM input parameter for unmarked retention error to a value of 2%.
• Defer a decision on a new value for mark release error pending resolution on methodology.
• Continue to model 150 Chinook encounters per test fishing boat and month if necessary.
• No FRAM change to model sublegal retention is proposed.