Stillaguamish Summer Chinook:

Productivity Estimates from Coded-Wire Tag Recoveries and

A Simple Model for Setting Interim Exploitation Rate Objectives

Kit Rawson

Tulalip Fisheries

7615 Totem Beach Rd.

Marysville, WA 98271

Krawson@tulalip.nsn.us

May 17, 2000

Introduction

The joint state tribal Comprehensive Chinook working group (COOKS[1] draft plan, November 28, 1999) has established the following objective for management of the harvest of Puget Sound chinook salmon stocks:

Manage fisheries to ensure that harvest related mortality will not impede recovery of the productivity, abundance, and diversity of Puget Sound chinook salmon stocks to levels consistent with treaty-reserved fishing rights and salmon –related cultural and ecological values.

The Stillaguamish chinook management unit is comprised of two stocks (WDF et al. 1993): Stillaguamish summer chinook and Stillaguamish fall chinook. Consistent with the COOKS draft plan, the management guidelines include an interim exploitation rate objective and a low abundance threshold. The interim exploitation rate objective is the maximum fraction of the production from any brood year that is allowed to be removed by all sources of fishery-related mortality, including direct take, incidental take, and non-landed mortality. The exploitation rate is expressed as an adult equivalent rate (see below for definition). The low abundance threshold is the minimum natural spawning escapement allowable in any return year. Whenever spawning escapement is projected to be below this level, fisheries will be managed to a very low exploitation rate (to be described in the COOKS plan in terms of specific fishery restrictions), lower than the interim exploitation rate objective.

In this report I will document analysis of available data pertaining to the Stillaguamish summer chinook stock and a simple model that can be used to project the consequences of employing different maximum exploitation rates. Using this model, managers can select an interim rate to use in management. The exploitation rate objective is termed “interim” because the comanagers expect and intend to collect new information pertaining to the current productivity, capacity, and diversity of the chinook salmon stocks. The ultimate management plan will be based on analysis of this new data.

The interim exploitation rate objective is developed from the most recently available information concerning productivity and capacity. For Stillaguamish chinook, there are 8 complete brood years of coded-wire tag (CWT) recoveries from the Stillaguamish Tribe’s indicator stock project. From recoveries of these tags, it is possible to estimate the total numbers of recruits produced from each brood year 1986-1993. Using spawner escapement estimates for those years, it is possible to estimate recruit per spawner values. I used the 8 estimates of recruits per spawner to simulate future patterns of production and harvest under adult equivalent exploitation rates, varying from zero to .80 in intervals of .05.

Methods

Coded-wire tag codes used. I used summarized and expanded recoveries of coded-wire tags from the Stillaguamish indicator stock project for brood years 1986-1993 to estimate actual exploitation rates. Because this project involved taking broodstock from the summer run fish in the North Fork Stillaguamish, this analysis is specific to the summer chinook stock. All escapement numbers, reconstructed run sizes, and estimates of recruits per spawner, are for the Stillaguamish Summer chinook stock only. However, because of a current lack of data for the Stillaguamish fall chinook stock, the maximum exploitation rate objective recommended for the summer chinook stock may be applied to the fall chinook stock as well. Tag codes and release numbers are in Table 1.

Table 1. Tag codes used in reconstruction of Stillaguamish chinook exploitation rates.

Brood Yr.	Tag Code	Tags released	AD-Only	Unmarked
1986	21-22-21	23,904	996	0
1987	21-25-55	127,910	9,333	7,923
1988	21-31-47	36,599	4,524	0
1989	21-18-26	44,964	1,873	0
1990	21-20-26	63,019	5,852	219
1991	21-22-05	165,620	5,303	5,833
1992	21-22-40	24,091	771	848
1992	21-22-51	89,207	5,708	5,206
1993	21-23-30	200,664	6,912	8,424
1993	21-20-45	3,855	133	162
1993	21-18-56	5,048	174	212

I obtained estimated CWTs by age for each tag code from the CRAS database management system[2]. I used the estimated contribution of CWTs to the fisheries directly from the CRAS reports and estimated the escapement of tagged fish to spawning escapement based on information on spawning escapement estimates and sampling rates assembled by Will Beattie (personal communication, 1999). Estimated CWTs for each brood year are shown in Table 2. Estimated CWTs were summed over all codes for the brood year for brood years with more than one tag code.

Table 2. Estimated contribution of Stillaguamish CWTs to fisheries and escapement for each brood year. The Nominal exploitation rate (ER) in the righthand column is simply the sum of the estimated fishery CWTs divided by the sum of fishery and escapement CWTs.

Brood	Est. Fishery CWTs^a				Est. Escapement CWTs^b				Nominal
Year	Age 2	Age 3	Age 4	Age 5	Age 2	Age 3	Age 4	Age 5	Total	ER
1986	79.5	157.7	32.1				82.9	22.2	374.5	0.719
1987	131.7	256.9	77.8	1.0		40.2	395.8		903.4	0.517
1988	71.7	145.7	83.8			104.0	37.4	2.3	444.9	0.677
1989	75.7	259.9	97.2			28.9	50.0		511.7	0.846
1990	286.2	481.8	116.0	4.0		160.0	196.4	3.2	1247.7	0.712
1991	3.1	42.3	44.3			23.9	43.7	3.6	160.9	0.558
1992	11.2	72.1	35.1		7.8	67.9	87.3	10.7	292.2	0.405
1993	140.8	284.8	94.1		62.9	172.0	141.5	43.6	939.7	0.553

^aFrom CRAS summary output.

^bFrom information supplied by W. Beattie, see Appendix A.

Exploitation Rates. The total of the estimated fishery CWTs divided by the total fishery CWTs plus the escapement CWTs is the nominal exploitation rate for the brood year. To make exploitation rates comparable among brood years when the distribution of fishing mortality among ages may have been diffrent, it is necessary to adjust fishery recoveries by adult equivalency (AEQ) factors[3]. I obtained AEQ factors for these tag codes form the Pacific Salmon Commission’s Chinook Technical Committee (M. Alexandersdottir, Northwest Indian Fisheries Commission, personal communication). Applying these to the fishery recoveries, I computed AEQ exploitation rates as the sum of the AEQ-adjusted fishery CWT contribution, divided by this sum plus the sum of the estimated escapement CWT contribution. These values are given in Table 3.

Table 3. Computation of AEQ exploitation rates from recoveries of Stillaguamish coded-wire tags.

Brood	AEQ Fishery CWTs^a				Fshry	Escp.	AEQ
Year	Age 2	Age 3	Age 4	Age 5	Total	Total	ER
1986	48.1	129.1	31.3		208.5	105.2	0.665
1987	77.7	209.6	77.6	1.0	365.8	436	0.456
1988	45.9	129.4	83.3		258.7	143.7	0.643
1989	44.9	216.9	97.0		358.9	78.9	0.820
1990	179.5	415.6	115.2	4.0	714.4	359.7	0.665
1991	1.8	35.4	44.0		81.2	71.2	0.533
1992	7.0	64.3	34.7		106.0	173.8	0.379
1993	84.2	238.8	93.4		416.3	420.0	0.498

^aComputed by applying AEQ factors from the PSC chinook technical committee to the estimated tag recoveries by age (Table 2).

Reconstruction of Total Recruits and Recruits per Spawner. Based on the above, I reconstructed the total brood year run size by expanding the spawning escapement by the estimated AEQ exploitation rate. Because the Stillaguamish Tribe’s indicator stock project includes acquisition of broodstock from the escapement, there are four categories of spawning escapement that must be considered: 1) natural origin fish spawning in natural areas, 2) natural origin fish taken for broodstock, 3) fish originating from the indicator stock program spawning in natural areas, and 4) fish from the indicator stock program taken as broodstock. I separated the escapement into these categories based on age and mark data from the Stillaguamish Tribe (J. Drotts, personal communication, 1999). The spawning escapement that was expanded to reconstruct total recruitment included all natural origin fish (categories 1) and 2) ). In order to estimate recruits per spawner I divided this total recruitment by the naturally spawning fish (categories 1) and 3) ) in the brood year. These computations are summarized in Table 4.

Table 4. Reconstruction of total recruitment and recruits per spawner.

Brood Year	Original Escapement^a		Subsequent Escapement^b		Brood Yr Exp. Rate	Recruitment		Recruits Per Spawner
1986	980	505		0.66		1,505	1.54
1987	1,065	695		0.46		1,278	1.20
1988	516	654		0.64		1,832	3.55
1989	510	458		0.82		2,544	4.99
1990	575	488		0.67		1,457	2.53
1991	1,331	486		0.53		1,040	0.78
1992	466	596		0.38		959	2.06
1993	563	585		0.50		1,165	2.07

^aIncludes all fish spawning naturally in the brood year.

^bIncludes all fish produced by the naturally spawners from the brood year which returned as escapement to the river (natural origin recruits).

Marine and Freshwater survival indices. The recruits per spawner values in Table 4 are the product of a component representing survival in freshwater and marine survival. Jim Scott (Washington Department of Fish and Wildlife, personal communication, November 1999) has developed a set of marine survival indices for Puget Sound chinook salmon from commonalties in recovery patterns of coded-wire tagged chinook from hatcheries throughout Puget Sound. These are shown in Table 5, along with a Stillaguamish freshwater index, which is simply the recruit per spawner value from Table 4 divided by the marine survival index value for the particular brood year.

Table 5. Puget Sound marine survival indices and Stillaguamish freshwater indices derived from recruits per spawner and marine survival indices. For each brood year, the recruits per spawner is the product of these indices.

Brood Year	Mar. Surv. Index^a	Freshwater Index^b
1986	2.35	.655
1987	.74	1.622
1988	.78	4.551
1989	.80	6.238
1990	1.66	1.524
1991	.29	2.690
1992	.37	5.568
1993	.68	3.044

^aFrom Jim Scott, Wash. Dept. Fish and Wildlife.

^bRecruits per spawner (Table 4) divided by marine survival index.

The Simulation Model and Results

In order to compare the effects of different exploitation rate strategies, I developed a simple simulation model using the above information. For each year simulated, the model selects a freshwater index and marine survival index at random from the values in Table 5. The first stage in computing total recruitment for the brood year is to multiply brood year escapement times the chosen freshwater index. If the resulting product exceeds a maximum value (5,000 in these simulations[4]) then the value of recruitment at this stage is reduced to that maximum . This is a simple spawner-recruit relationship that does not include any density dependence, other than the maximum value of total recruitment[5]. The first stage recruits are then multiplied by a marine survival index to obtain the total brood year recruitment. Each year’s marine survival index is the average of the randomly selected index and the index used for the previous year. This averaging mimics some degree of autocorrelation in marine survivals.

The brood year recruitment is then subjected to fishing mortality, based on the target exploitation rate chosen for the run. For each year of each run, the exploitation rate was adjusted by a management error rate randomly chosen from 36 values computed for the 1988 through 1993 fishing years by comparing expected with actual exploitation rates for six Puget Sound chinook management units[6]. Surviving fish return to spawning escapement according to the average escapement age distribution observed for the 1986-1993 brood years[7].

Each model run projects the return and the escapement for 25 years. I replicated 1000 model runs at each exploitation rate ranging from 0 to .80 in increments of .05 (increments of 0.1 above 0.50). All model runs were seeded with the observed North Fork Stillaguamish natural chinook escapement values for the first four years of the run (years 0 through 3).

Examples of individual runs of this model at four different exploitation rates are shown in Appendix C.

For each target exploitation rate I computed the average percentage of years that the escapement was below a critical threshold (500 spawners in these runs)[8]. I also averaged the spawning escapements in years 0-3 and 22-25. To compare rebuilding rates, I computed the ratio of years 22-25 escapement to years 0-3 escapement. These quantities are tabulated (Table 6) below.

Table 6. Summary of results of 1000 runs at each exploitation rate.
Exploitation	Percent	Percent	Median	Median
Rate	< crit.	Escap. ratio>1	Escap. ratio	Escapement
0.00	1%	96%	2.75	3,597
0.05	1%	96%	2.81	3,377
0.10	1%	96%	2.76	3,165
0.15	2%	95%	2.66	2,964
0.20	2%	95%	2.56	2,758
0.25	3%	93%	2.57	2,418
0.30	4%	92%	2.48	2,210
0.35	6%	92%	2.46	1,920
0.40	7%	91%	2.29	1,686
0.45	11%	87%	2.14	1,444
0.50	17%	80%	1.92	1,180
0.60	41%	52%	1.04	648
0.70	73%	12%	0.27	259
0.80	94%	0%	0.02	55

Conclusion

The results from these simulations can be compared with criteria for evaluating fishery management plans under consideration by NMFS. The first test is whether the proposed action increases the probability of the spawning escapement dropping below a threshold level by more than 5 percentage points, when compared with no action. The comanagers have chosen a threshold of 500 spawners for the Stillaguamish summer chinook stock based on historical spawning escapement data showing that in 6 of 30 years (1969-1998) the estimated escapement for this stock was below this value. Because of reduced fishing mortality, escapements below 500 have only occurred twice in the last 15 years (1984 and 1992). Until detailed information on capacity and productivity is assembled, the comanagers have chosen to use 500 as the critical threshold in the interim management plan and I used the same value as the critical escapement level to test in these simulations. Using the NMFS criterion and the comanagers’ critical escapement value of 500, exploitation rates up to 0.35 would be acceptable based on the results of these simulations (Table 6, comparing the percentages of runs with escapements < critical for exploitation rates of 0 and 0.35).

A second criterion proposed by NMFS is to test whether the probability of rebuilding under a proposed action is 80% or greater. For these simulations, I defined rebuilding as occurring when the escapement ratio (years 22-25 escapement divided by years 0-3 escapement) is greater than 1. Using this definition and the NMFS criterion, exploitation rates of up to 0.50 would be acceptable (Table 6).

Using the most restrictive of the two criteria, target exploitation rates of up to 0.35 would be acceptable for the Stillaguamish summer chinook stock. Other factors, not included in the simulation analysis, might dictate a maximum rate lower than 0.35. The most important of these is that the exploitation rate chosen will be applied to the entire Stillaguamish management unit, which includes both the Stillaguamish summer and Stillaguamish fall stocks. Therefore, the chosen rate must be appropriate for both stocks.

At the present time, there is very little information concerning the productivity of the Stillaguamish fall stock other than the fact that the average abundance of this stock is approximately 50% of the Stillaguamish summer stock (Appendix A, Table A.1, comparing percent North Fork escapement with estimated total escapement). I ran the model described above, with initial escapements and maximum recruitment reduced to 50% of the values used for the Stillaguamish summer stock and the critical escapement threshold set at 250 spawners. These resulted in an exploitation rate of 0.35 showing a probability of 4% of escapements below critical, as compared with a 1% probability for an exploitation rate of 0. The probability of rebuilding at this exploitation rate was 96%. This rough analysis indicates that a target exploitation rate of 0.35 would also be appropriate for the Stillaguamish fall stock.

The comanagers have selected an exploitation rate guideline of 0.30 for the Stillaguamish chinook management unit and a critical threshold of 500 natural origin spawners for the Stillaguamish summer chinook stock. According to the above analysis, these guidelines should meet the NMFS criteria for an acceptable harvest management strategy for bot the Stillaguamish summer and Stillaguamish fall stocks.

References Cited

Morishima, Gary. 1999. When is a fish not a fish? Metrics of exploitation for chinook salmon. Briefing paper prepared for WDFW/NWIFC modeling workshop. Quinault Management Center, Mercer Island, WA, October 1999.

Washington Department of Fisheries (WDF), Washington Department of Wildlife (WDW), and Western Washington Treaty Indian Tribes. 1993. 1992 Washington State salmon and steelhead stock inventory. 212 p. (Available from the Washington Department of Fish and Wildlife, Olympia, WA.)

Appendix A. In-river tag recovery information (assembled by W. Beattie, NWIFC) A1. Escapement estimates and spawning ground CWT sampling summary
				total	Est.	CWT		Est. CWT
	Natural esc.		# Carcass	# CWT's	Pct.	exp.	NF % of	in
Year	N. Fork	Total	sampled	Recov'd	Natrl.	factor	total esc	Escp.
1988	516	717	0				72.0%
1989	510	811	72			7.08	62.9%
1990	575	842	65			8.85	68.3%
1991	1331	1632	187			7.12	81.6%
1992	466	780	84	11	86.9%	5.55	59.7%	61.0
1993	563	928	270	94	65.2%	2.09	60.7%	196.0
1994	659	954	206	65	68.4%	3.20	69.1%	207.9
1995	561	822	192	54	71.9%	2.92	68.2%	157.8
1996	918	1384	270	84	68.9%	3.40	66.3%	285.6
1997	771	1156	88	30	65.9%	8.76	66.7%	262.8
1998	955	1540	212	111	47.6%	4.50	62.0%	500.0
Average							67.0%
A.2 Detail of observed and estimated Stillaguamish indicator stock CWT recoveries from stream surveys
	Recovery Year
Brood Yr	1989	1990	1991	1992	1993	1994	1995	1996	1997	1998
1985
1986	0	9	3
1987	0	4	49
1988			13	6	1
1989				5	23
1990					70	56	1
1991						7	14	1
1992						2	19	21	1
1993							20	47	15	9
1994								15	11	83
1995										8
1996										11
total	0	13	65	11	94	65	54	84	27	111

A.3 Estimated recoveries expanded for sampling rate (total carcasses sampled / estimate of escapement)
	Recovery Year
Brood Yr	1990	1991	1992	1993	1994	1995	1996	1997	1998
1986	79.6	21.4
1987	35.4	348.8
1988		92.5	33.3	2.1
1989			27.7	48.0
1990				146.0	179.1	2.9
1991					22.4	40.9	3.4
1992					6.4	55.5	71.4	8.8
1993						58.4	159.8	131.4	40.5
1994							51.0	96.4	373.9
1995									36.0
1996									49.6
A.4. Escapement recoveries expanded for tagging rate at release
	Recovery Year
Brood Yr	1990	1991	1992	1993	1994	1995	1996	1997	1998
1986	82.9	22.2
1987	40.2	395.8
1988		104.0	37.4	2.3
1989			28.9	50.0
1990				160.0	196.4	3.2
1991					23.9	43.7	3.6
1992					7.8	67.9	87.3	10.7
1993						62.9	172.0	141.5	43.6
1994							53.0	100.2	388.7
1995									43.6
1996									106.9
Total	123.1	522.0	66.3	212.3	228.2	177.7	316.0	252.4	582.9
% of NF	0.214	0.392	0.142	0.377	0.346	0.317	0.344	0.327	0.610

Appendix B. Management errors based on comparison of post-season exploitation rates with preseason predictions. All values, except those for White River spring chinook from J. Gutmann’s memo (24 Feb 1998) were used in the simulations. White River values were rejected because they were significantly different from the remaining values in the array (1-tailed t-test, see p-values on right).

		Management	Percent
Management Unit	Type	Year	Error
LAKE WASHINGTON	S/F	1988	-0.116	t-test p-values
LAKE WASHINGTON	S/F	1989	-0.033	0.000426	WR vs S/F
LAKE WASHINGTON	S/F	1990	0.012	5.41E-05	WR vs rest
LAKE WASHINGTON	S/F	1991	-0.144	0.078335	SP vs S/F
LAKE WASHINGTON	S/F	1992	-0.085
LAKE WASHINGTON	S/F	1993	0.213	Averages
SKAGIT	S/F	1988	0.226	Lk WA	-0.02550
SKAGIT	S/F	1989	-0.101	Skagit S/F	-0.02217
SKAGIT	S/F	1990	0.126	Sno S/F	-0.00750
SKAGIT	S/F	1991	-0.071	Stil S/F	0.03900
SKAGIT	S/F	1992	-0.127	Nook SP	-0.02467
SKAGIT	S/F	1993	-0.186	Skagit SP	-0.02083
SNOHOMISH	S/F	1988	0.052	WR SP	0.26283
SNOHOMISH	S/F	1989	0.130
SNOHOMISH	S/F	1990	0.162	S/F	-0.004042
SNOHOMISH	S/F	1991	-0.134	non-WR	-0.010278
SNOHOMISH	S/F	1992	-0.182	SP	0.0724444
SNOHOMISH	S/F	1993	-0.073
STILLAGUAMISH	S/F	1988	-0.009	1988	0.06286
STILLAGUAMISH	S/F	1989	0.117	1989	0.04714
STILLAGUAMISH	S/F	1990	-0.252	1990	0.07257
STILLAGUAMISH	S/F	1991	0.067	1991	-0.04557
STILLAGUAMISH	S/F	1992	0.008	1992	-0.07986
STILLAGUAMISH	S/F	1993	0.303	1993	0.11529
NOOKSACK	SP	1988	0.009	1994	0.09267
NOOKSACK	SP	1989	0.049
NOOKSACK	SP	1990	-0.044	All	0.02874
NOOKSACK	SP	1991	-0.173
NOOKSACK	SP	1992	-0.028
NOOKSACK	SP	1993	0.039
SKAGIT	SP	1988	0.081
SKAGIT	SP	1989	-0.177
SKAGIT	SP	1990	-0.006
SKAGIT	SP	1991	-0.162
SKAGIT	SP	1992	-0.069
SKAGIT	SP	1993	0.208
WHITE RIVER	SP	1988	0.197
WHITE RIVER	SP	1989	0.345
WHITE RIVER	SP	1990	0.51
WHITE RIVER	SP	1991	0.298
WHITE RIVER	SP	1992	-0.076
WHITE RIVER	SP	1993	0.303

Appendix C. Examples of individual simulations at exploitation rates of 0.0, 0.05, 0.30, and 0.50.

Exploitation Rate = 0

---

[1] The Committee to Organize Options for King Salmon (COOKS) is developing a harvest management plan for Puget Sound chinook salmon to be completed in early 2000.

[2] The Coded-wire tag Retrieval and Analysis System is maintained by the Northwest Indian Fisheries Commission in Olympia, WA (www.nwifc.wa.gov). The tag recovery data are compiled and stored by the Pacific States Marine Fisheries Commission in Portland, OR.

[3] “The adult equivalence factor is an age- and stock-specific scalar that converts nominal fish into potential adults.” (Morishima 1999).

[4] Based on the upper quartile of total recruitment values inferred from observed escapements 1965-1993 and adjusted by the geometric mean of the marine survival indices so that the maximum total recruitment will average to approximately 4000.

[5] Because of the shape of the graph of recruits on spawners this has been called the “hockey stick” model.

[6] J. Gutmann, memorandum to Puget Sound management biologists, 24 February 1998. See Appendix B for further detail.

[7] 30% age 3, 60% age 4, 10% age 5

[8] Only years 4-25 were counted, since years 0-3 were always seeded with the same values.