Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Racial analysis of Skeena River steelhead trout (Salmo gairdneri) by scale pattern features Cox-Rogers, Steven Frank 1985

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
UBC_1985_A6_7 C69.pdf [ 8.41MB ]
[if-you-see-this-DO-NOT-CLICK]
Metadata
JSON: 1.0096059.json
JSON-LD: 1.0096059+ld.json
RDF/XML (Pretty): 1.0096059.xml
RDF/JSON: 1.0096059+rdf.json
Turtle: 1.0096059+rdf-turtle.txt
N-Triples: 1.0096059+rdf-ntriples.txt
Original Record: 1.0096059 +original-record.json
Full Text
1.0096059.txt
Citation
1.0096059.ris

Full Text

RACIAL ANALYSIS OF SKEENA RIVER STEELHEAD TROUT (SALMO GAIRDNERI) BY SCALE PATTERN FEATURES by STEVEN FRANK COX-ROGERS B.Sc.,University Of British Columbia,1981 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Department of Zoology) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA September 1985 © Steven Frank Cox-Rogers, 1985 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of The University of British Columbia 1956 Main Mall Vancouver, Canada V6T 1Y3 Date OCT" / o/fitf DE-6(3/81) i i ABSTRACT The feasibility of using freshwater and first marine year scale patterns to identify component stocks of steelhead trout (Salmo gairdneri) in the Skeena River was investigated. Scale samples and sex and size data were attained from adult steelhead originating from five Skeena River tributaries (Zymoetz, Kispiox, Morice-Bulkley, Babine, Sustut) over a series of different years. Adult scale samples were also collected from the 1984 incidental steelhead catch in the Area 4 commercial salmon fishery for potential stock classification purposes. Significant differences in scale pattern growth, age composition, and sizes at age were found between the five Skeena River steelhead stocks. Linear discriminant function analysis indicated that the five stocks could be classified to correct river of origin with between 45% and 62% average classification accuracy (range Zymoetz 29%-60%, Kispiox 35%-60%, Morice-Bulkley 44%-76%, Babine 54%-64%, Sustut 56%-72%) depending upon the classification model used. Juvenile morphometric analysis for three of the stocks (Kispiox, Morice-Bulkley, Zymoetz) indicated the presence of significant between stock differences in standardized body form. These results support the notion that Skeena River steelhead exist as quantifiably discrete stocks. Classifying the 1984 mixed stock commercial fishery catches to probable stock of origin indicated that distinct peaks of stock abundance and run-timing occur through the fishery. In general, Morice-Bulkley and Sustut River steelhead were predicted to be most abundant with run-timings during the earlier portions of the fishery. Kispiox, Babine, and Zymoetz River steelhead were predicted to be less abundant with later run-timings through the fishery. Potential commercial fishery impacts to steelhead are briefly discussed. These observations suggest that the technique of scale patterns is a feasible method for stock separation in Skeena River steelhead. Further study is required to clarify yearly variance in the technique and to better establish stock specific differences. i v TABLE OF CONTENTS ABSTRACT ii LIST OF TABLES vLIST OF FIGURES . . viii ACKNOWLEDGEMENTS xIntroduction 1 Description of the Skeena River Drainage 5 Life History of Skeena River Steelhead 8 The Skeena River Commercial Salmon Fishery 10 Materials and Methods 17 Scale Data Collection and Preparation 17 Determination of Sample Sizes 23 Juvenile Analysis 24 Analytical Techniques" for Scale Pattern Analysis .... 25 Results 30 Discrimination of Skeena River Steelhead 30 Verification of scale aging 3Descriptive statistics 0 Age composition 3Sizes at age 5 Scale Pattern Features of Skeena River steelhead .... 40 Scale features of smolt age 3 adult steelhead .... 41 Scale features of smolt age 4 adult steelhead .... 48 Scale features of age 3.2+ and 4.2+ steelhead .... 52 Scale pattern variation between years 55 Plus growth 57 Stock DiscriminationSeparate freshwater age discriminant models 58 Pooled freshwater age discriminant models 72 Discrimination by sex 75 Commercial Fishery Stock Composition 76 Juvenile Analysis 84 Discussion 97 Biological Considerations 9Theoretical Considerations 105 Commercial Fishery Considerations 108 Applications to Steelhead Management 114 LITERATURE CITED 1 1 8 LIST OF TABLES Table 1.Riverine features of the five Skeena River tributaries used in the study 9 Table 2. Variables measured from the adult steelhead scales for each stock . 22 Table 3. Age Composition features of Skeena River steelhead by sex, smolt age, and ocean age 34 Table 4. Mean lengths and weights of steelhead at various ocean age for the five stocks used in the study 40 Table 5. One way ANOVA results for comparison of differences in mean scale zone widths at age for steelhead from the Kispiox, Zymoetz, and Morice-Bulkley Rivers 44 Table 6. One way ANOVA results for comparison of differences in mean scale zone features for smolt age 3 steelhead by pooled ocean age 47 Table 7. One way ANOVA results for comparison of differences in mean scale zone features for smolt age 4 steelhead by pooled ocean age 51 Table 8. One way ANOVA results for comparison of differences in mean scale zone features for age 3.2+ and 4.2+ steelhead 55 Table 9. One way ANOVA results for comparison between years in mean scale zone features for Sustut and Zymoetz River steelhead 56 Tables 10-16. Classification matrices for the linear discriminant models used to classify Skeena River steelhead to stock of origin 62 Table 17. 1984 commercial fishery steelhead catches in Area 4 79 Table 18. Ocean age composition by sex for steelhead sampled from Area 4 in 1984 79 Table 19. Mean lengths and weights of steelhead sampled from Area 4 in 1984 80 Table 20. Classification results to stock of origin by week for steelhead sampled from Area 4 in 1984 81 Table 21. Means and the results of one way ANOVAS for comparison of differences in morphological features in juvenile steelhead 95 vi i i LIST OF FIGURES Figure 1. The Skeena River Drainage 6 Figure 2. 1963 to 1984 mean annual steelhead harvest by month in Area 4 13 Figure 3. 1963 to 1984 mean steelhead escapement by month through Area 4Figure 4. 1963 to 1984 mean annual steelhead harvest+escapement in Area 4 15 Figure 5. Adult steelhead scale from the Sustut River ..... 19 Figure 6. Discriminant space 28 Figure 7. Age composition structure for the five steelhead stocks used in the study 31 Figure 8. Mean lengths of age 3.2+ and 4.2+ steelhead from the five stocks used in the study 36 Figure 9. Mean weights of age 3.2+ and 4.2+ steelhead for the five stocks used in the study 36 Figure 10. Mean lengths by sex for age 3.2+ steelhead from the five stocks used in the study 38 Figure 11. Mean scale zone widths for steelhead of smolt ages 3 and 4 2 Figure 12. Yearly freshwater scale zone widths in adult steelhead of smolt age 3 by pooled ocean age 45 Figure 13. Yearly freshwater scale zone circuli counts in steelead of smolt age 3 by pooled ocean age 45 Figure 14. Yearly freshwater scale zone widths in adult steelhead of smolt age 4 9 Figure 15. Yearly freshwater scale zone circuli counts in adult steelhead of smolt age 4 9 Figure 16. Yearly freshwater scale zone widths in steelhead of age 3.2 + 53 Figure 17. Yearly freshwater scale zone widths in steelhead of age 4.2+Figure 18. Discriminant function analysis describing scale pattern variation in adult steelhead of smolt age 3 .... 59 Figure 19. Discriminant function analysis describing scale pattern variation in adult steelhead of smolt age 4 .... 64 Figure 20. Discriminant function analysis describing scale pattern variation in steelhead of age 3.2+ 67 Figure 21. Discriminant function analysis describing scale pattern variation in steelhead of age 4.2+ 70 Figure 22. Discriminant function analysis describing scale pattern variation in adult steelhead of pooled smolt age 73 Figure 23. Age composition structure by week for steelhead sampled in the 1984 commercial fishery 77 Figure 24. Predicted run-timing 85 Figure 25. Predicted run-timingFigure 26. Predicted run-timing 87 Figure 27. Predicted run-timing . .. 87 Figure 28. Predicted run-timing 89 Figure 29. Predicted run-timingFigure 30. Predicted run-timing 91 X Figure 31. Predicted run-timing • 91 Figure 32. Discriminant function analysis describing morphological variation among juvenile steelhead from the Kispiox,Zymoetz,and Morice-Bulkley Rivers 94 xi ACKNOWLEDGEMENTS I would like to thank my supervisor Dr N.J. Wilimovsky and committee members Drs. Art Tautz, J.D. McPhail, and Carl Walters for their advice and assistance in the development of this thesis. I would also like to thank Mike Lough, Eric Parkinson, Bruce Ward, and Angelo Fachin of the Provincial Fisheries Branch and Les Janz of the Federal Department of Fisheries and Oceans for their helpful comments through numerous discussions. The B.C Fisheries Branch and the Department of Fisheries and Oceans provided financial support for this study. A note of thanks is extended to the staff of B.C. Packers Prince Rupert Division for the use of their facility during this project. Finally, I would like to express sincere gratitude to my wife Marie who was always willing to provide field assistance and support when needed most. 1 INTRODUCTION Fisheries biologists have long been interested in determining subpopulation structure in mixed population fisheries (Clutter and Whitesel 1956, Worlund and Fredin 1962, Anas and Murai 1969, Bilton 1971, Cook and Lord 1978, Pella and Robertson 1979, Lear and Sandeman 1980, Maclean and Evans 1981, McDonald 1981, Conrad 1984). Whenever different spawning populations of single or several species intermix, the harvesting of one may differentially affect the other (McDonald, 1981). Not surprisingly, this has led to extensive applications of the "stock concept" (Simon and Larkin 1972, Ricker 1972, Utter 1981) in mixed fishery management. The various spawning populations of a given species are taken to represent local stocks possessing genetic differences that are adaptive (Maclean and Evans, 1981) and which should be maintained (Larkin, 1972). Because less productive stocks are particularly susceptible to overfishing in a mixed fishery (McDonald, 1981), effective management requires knowledge of which stocks are contributing and how their distributions change over time. Mixed stock fisheries analyses in North America have primarily concentrated on the salmonids, although not exclusively (Hill 1959, Parsons 1971, Misra and Ni 1983). Because of their commercial importance, all of the Pacific salmon (genus Oncorhynchus) as well as the Atlantic salmon (Salmo salar) have received considerable attention. Less studied have been non-target species harvested incidentally in mixed stock fisheries. The incidental interception of steelhead 2 trout (Sa-lmo gairdner i) in net fisheries for salmon along the Pacific coast of North America is one such example. Steelhead occur along the Pacific coast from northern California into Alaska (Withler, 1966). In British Columbia, steelhead are harvested throughout their range (Oguss and Andrews 1977, Oguss and Evans 1978, Parkinson 1984a) with major incidental fisheries occurring in areas adjacent to the Fraser River and Skeena River estuaries. With regard to the latter, the Skeena River hosts various "stocks" of summer run steelhead which are incidentally harvested during the commercial sockeye (Oncorhynchus nerka) and pink salmon (Oncorhynchus qorbuscha) each year (June-September). An estimated 30%-60% of the total Skeena River steelhead return in any given year is harvested as incidental catch (unpublished data, BCFB 1984). Little is known of steelhead stock dynamics through the fishery nor of how commercial fishing may be affecting the biological integrity of each stock. Preliminary investigations by the B.C Fisheries Branch (unpublished data, 1982, 1984) suggest that the major Skeena River steelhead stocks show distinct "peaks of temporal abundance through the ' commercial fishery. The identification of each stock has been, however, quite difficult. Of concern is how each stock contributes proportionally to the weekly incidental catch. This, in turn, determines the overall pattern of stock specific run-timing. Without such knowledge the management of Skeena River steelhead has been limited, especially for those stocks believed to coincide with peak sockeye and pink salmon 3 run-timing. This suggests the need for ways of identifying the stock origins of Skeena River steelhead in the commercial salmon fishery. Several techniques are available for identifying the racial origins of salmonids in natal environments and in mixed stock commercial fisheries. Mark and recapture methods have been widely applied in various studies (Hartt 1962); however, in the case of wild steelhead, they present substantial logistic problems for both juvenile tagging and later adult recapture. An alternative technique is to use naturally occurring variation in one or more biological systems that are hypothesized or known to differ between populations (Worlund and Fredin, 1962). Electrophoretic variation (steelhead: Utter and Allendorf 1977, Chilcote et al. 1980, Parkinson 1984a, 1984b; Atlantic salmon: Nyman and Pippy 1972, Thorpe and Mitchell 1981; sockeye salmon: Grant et al. 1980; chum salmon: Fournier et al. 1984,), body morphology and meristics ( steelhead: Smith 1969, Winter et al. 1980; sockeye salmon: Fukuhara 1962, Dark and Landrum 1964 chinook salmon: McGregor 1924; pink salmon: Amos et al. 1963; chum salmon: Fournier et al. 1984; Atlantic salmon: Riddell and Leggett 1981; coho salmon: Taylor 1984), elemental composition (sockeye salmon: Caliprice 1971, Mulligan et al. 1983), age structure (Ricker 1972) and parasitic infestations (sockeye salmon: Margolis 1958) have all been used with varying degrees of success to characterize different spawning populations. Perhaps the most widely applied technique has been the use of calcareous structures such as otoliths (steelhead: 4 Mckern et al. 1974), fin rays (see Ihssen et al. 1981) and especially scales (Atlantic salmon: Lear and Misra 1978, Lear and Sandeman 1980, Reddin and Misra 1985; Pacific salmon: Clutter and Whitesel 1956, Henry 1961 , Rowland 1969, Mosher 1963, Anas and Murai 1969, Tanaka et al. 1969, Bilton 1971, Bilton and Messinger 1975, Cook and Lord 1978, Krasnowski et al. 1978, McBride and Marshall 1983, McGregor et al. 1983, Conrad 1984). Scale analysis has certain advantages over other stock identification techniques. Scales are generally easier to collect and prepare, do not require killing of the specimen, and are applicable to large scale stock identification studies (Ihssen et al. 1981). Steelhead scales have been read by many authors and have proven reliable in those populations studied (Neave 1944, Shapovalov and Taft 1954, Maher 1954, Maher and Larkin 1955, Chapman 1958, Bali 1958, Withler 1966, Narver 1969, Narver and Withler 1974, Whately 1977, Whately et al. 1978, Horncastle 1981, among others). Few however, (Bali 1958, Keating 1959) have used scale patterns to characterize particular steelhead stocks. Scale pattern analyses rely on stock specific variations in the widths and patterns of scale circuli and yearly scale growth zones. Environmental differences between freshwater rearing environments are hypothesized to result in differential scale growth during the freshwater period. The degree of scale pattern difference between stocks determines the accuracy of statistical models used to separate them, often by discriminant analysis. Both 5 parametric (Anas and Murai 1969, Major et al. 1975, Bilton and Messinger 1975, Conrad 1984) and nonparametric (Cook and Lord 1978, Cook 1982) discriminant analyses have been applied to a wide range of mixed salmonid fishery problems. The potential of discriminant analysis by scale patterns is particularly suited to Skeena River steelhead as they rear in natal environments for long periods of time and are subject to longterm watershed specific growth regimes. This thesis examines the use of scale pattern analysis as a practical method for differentiating between steelhead trout stocks from the Skeena River. The goals of the study were two fold. Firstly, scale pattern analysis was used to test the hypothesis that steelhead from the Skeena River exist as racially separable stocks. Secondly, scale pattern analysis was used to assess the potential for identifying the weekly steelhead contributions by stock to the commercial salmon f ishery. Description of the Skeena River Drainage The Skeena River drains an area of approximately 30,500 square kilometers lying in the central western portion of British Columbia (figure 1). Climatic patterns vary in an east-west direction with light precipitation and extremes of temperature near the interior plateau and heavy precipitation and moderate temperatures nearer the coast (Larkin and McDonald, 1968). Seven Skeena River tributaries, as well as their sub-tributaries, can be considered as hosting the major steelhead 6 Figure 1. The Skeena River Drainage. Shown are the major steelhead tributaries: the Lakelse, Kitsumkalum, Zymoetz, Morice-Bulkley, Kispiox, Babine, and Sustut Rivers (after Whately, 1977). 7 8 stocks; in ascending order upstream from the mouth these are the Lakelse, Kitsumkalum, Zymoetz (Copper), Morice-Bulkley-Suskwa, Kispiox, Babine, and Sustut rivers respectively. In addition, various other tributaries (Ecstall, Khyex, Eschamsiks, Gitnadoix, Khtada, Exstew, Kitwanga, Kitseguecla, Sicintine, Squingula etc), smaller creeks, and the mainstem Skeena itself are known to support steelhead production. Two of the larger tributaries, the Babine and Morice-Bulkley Rivers, headwater in large lake systems. Table 1 summarizes the major riverine features for the five Skeena tributaries considered in this study (Zymoetz, Morice-Bulkley, Kispiox, Babine, and Sustut Rivers). Life Hi story of Skeena River Steelhead Skeena River steelhead taken incidentally in the commercial fishery are primarily of summer and fall run origin which return to the Skeena River as adults from June through September in their fourth, fifth, sixth, seventh, or eighth plus years of life. After overwintering in natal streams the adults generally spawn from mid April through June. Fry emergence occurs from mid to late summer with the parr remaining in freshwater for one to five years (winters) before smolting and migrating to the ocean. Not all adults die following spawning and many are taken as kelts in the commercial fishery during their seaward migration. Winter and spring run steelhead (November-April) are found in the lower Skeena River tributaries below Hazelton and Table 1.Riverine features of the five Skeena River tributaries used in the study. Feature Zymoetz Long. 128 27 W Lat. 54 32 N Drainage (sq km) 3080 Upstream (km) 115 Distance app. Mean Flow (m3/s) Peak Flow Minimum Flow Length (km) 1 38 June Jan. 80 Summer Turbidity (JTU) Summer Water (C) Temperature Water Hardness (mg/1 CaCo3) Mean Annual (cm) Prec ipi tat ion Mean January (C) air temp. Mean July (c) air temp. Frost free (days) period 6-15 11-16 22-64 Morice/ Bulkley 126 43 W 1 54 24 N 12300(M+B) 1911(M) 200(B) 315(M) 164(M+B) 76(M) June Jan. 120(B) 75(M) 1 5-24 6-1 1 Ki spiox 27 40 W 1 55 1 5 N 2086 Babine 26 42 W 55 25 N 6790 1 5-25 10-16 24-33 9- 1 4 Sustut 127 20 W 56 00 N .3000 220 270 350 46 51 -June June June Feb. Mar . Mar. 1 37 85 65 24-52 6-10 1 00 to 40 to 40 to 40 to 50 to 350 250 1 00 75 75 _ 1 1 -15 -16 -18 -22 1 5 1 6 1 6 <1 4 < 1 4 60 to <60 to 60 to <70 <50 1 40 1 00 1 00 10 are not subject to any appreciable incidental (commercial) fishery. The Morice-Bulkley river system and its tributaries is believed to support the majority of Skeena river steelhead production followed by the Babine, Zymoetz, Sustut, and Kispiox river systems respectively (BCFW Branch, unpublished data, 1984) The Lakelse and Kitsumkalum rivers are primarily winter run streams although their contribution to summer run production is recognized. Mainstem Skeena River steelhead production is not known; however, it may have an important role in rearing the larger parr originating from several of the less productive tributaries (Tredger, 1984). Skeena River steelhead have been previously examined for life history features in the Morice-Bulkley River (Whately et al. 1978), the Kispiox River (Whately, 1977), and the Babine River (Narver, 1969). Both Taylor (1968) and Pinsent and Chudyk (1973) described the Skeena River system with regards to steelhead. The Skeena River Commercial Salmon Fishery The commercial salmon fishery on the Skeena River has had a diverse history (see Milne, 1955) characterized by fluctuating catches of the two principle target species, sockeye and pink salmon (Larkin and McDonald 1968, Todd and Larkin 1971, McDonald 1981). The majority of fishing effort occurs by gillnet in Fisheries statistical Area 4 adjacent (within 25-30 km) to the Skeena River estuary. An increasing proportion of seiners participate in the fishery although they are primarily 11 restricted to the outer regions of Area 4. Other salmonid species taken in the fishery include chinook (0 tshawytscha), chum (0 keta), and coho (0 kisutch) salmon as well as small numbers of searun Dolly Varden char (Salvelinus malma) and cutthroat trout (Salmo clarki). Oguss and Andrews (1977) and Oguss and Evans (1978) reviewed the incidental catches of steelhead in the Skeena River commercial fishery. Both sockeye and pink salmon are believed to pool in area 4 for considerable lengths of time before migrating upstream into the Skeena River (5 and 3 days respectively, Aro and McDonald, 1968) although variations can occur depending upon tidal action and river flows. Based on limited information, steelhead pass through Area 4 on a daily basis and may take three to four days to do so. The effects of fluctuating fishing effort in Area 4 (harvest rates, geartypes, fishing locations, duration etc) on steelhead escapement is not well understood. Seiners are typically abundant only at the height of the sockeye fishery (late July). Normal fishery openings for all gears generally occur on Sunday evenings and can last from one to four or more days (24 hours/day). Department of Fisheries and Oceans catch and test fishery records for the years 1963 to 1984 show average catches of steelhead peaking from early to mid August just after peak sockeye and just prior to peak pink salmon harvests. The annual average steelhead catch for all gear types in area 4 has been just over 13,000 pieces with extremes in catch occurring in 1966 (20,000) and again in 1984 (31,000). The average annual harvest+escapement for the same time period has been estimated 1 2 at 37,000 pieces with extremes again occurring in 1966 (55,000) and 1984 (85,000). Figures 2 and 3 outline the general temporal distribution of the commercial and test fishery steelhead catches by month. Figure 4 outlines the fluctuating nature of the total Skeena River steelhead harvest+escapement for the years 1963 to 1984. Upstream escapement calculations for steelhead are based on Department of Fisheries and Oceans test fishery indices and multiplication factors generated on best estimated escapement figures for a ten or more year period (BCFW Branch, unpublished data, 1984). Skeena River steelhead are also harvested by native net fisheries in much of the Skeena itself and by major sport fisheries in all of the mainstem tributaries. 13 Figure 2. 1963 to 1984 mean annual steelhead harvest by month in Area 4. The week beginning codes are Week 7=July 1, Week 8=July 8, Week 9=Julyl5, Week lO=July 21 Week 11=July 29, Week 12=Aug 5, Week 13=Aug12 Week 14=Augl9 Week l5=Aug26 (Source, unpublished data, BCF Branch, 1984). Figure 3. 1963 to 1984 mean steelhead escapement by month through Area 4. The week beginning codes are Week 7=July 1, Week 8=July 8, Week 9=July15, Week 1U=July 22 Week 11=July 29, Week 12=Aug 5, Week 13=Aug12 Week 14=Augl9 Week l5=Aug26 (Source, unpublished data, BCF Branch, 1984). • i..-oc; i i snnnu i \mc B I / I <VT ST fcT £T ?;T TI OT 6 8 /. 15 Figure 4. 1963 to 1984 mean annual steelhead harvest+escapement in Area 4. ( Source, unpublished data, BCF Branch, 1984). 16 * = harvest + - escapement 17 MATERIALS AND METHODS Scale Data Collection and Preparation Ninety to one hundred adult steelhead scale samples taken in the late fall (1975-1983) from each of the five major Skeena River stocks (Kispiox, Zymoetz, Babine, Sustut, Morice-Bulkley) were selected from existing B.C. Fisheries Branch data bases for scale pattern analysis. Stock definition was limited to the major Skeena River tributaries. Most scales had been previously mounted in acetate and represented angler caught steelhead taken during various Fisheries Branch projects. The majority of scales had been once read for age and included length, weight, and sex data. These scales represented the learning samples for subsequent discriminant analysis. Scale samples were selected from years having adequate (n>100) sample sizes for each stock; these were: Kispiox River 1975 n=l03, Zymoetz River 1975 n=30 1978 n=62, Morice River 1976 n=30 1977 n=60, Babine River 1978 n=9l, Sustut River 1977 n=30 1983 n=60. The availability of yearly time series scale data for between years comparison was limited. A major a priori assumption for this study was that the existing data base adequately represented the true population structure of each stock. Sixty scale samples were attained and analysed for each of the Lakelse and Kitsumkalum Rivers but were not used in later discriminant analyses because of their likely winter-run origins. Previously prepared scales and those prepared by the author 18 were sampled from the preferred area (Clutter and Whitesel, 1956) on the left side of each steelhead two to four scale rows above the lateral line just posterior to the dorsal fin. Two nonregenerate scales were mounted in acetate following the methods of Chuganova (1963) and the two selected scales were then projected at 34X magnification under a 3M microfiche reader-printer. Initial ages were assigned following the criteria of previous workers (Maher 1954, Clutter and Whitesel 1956, Maher and Larkin 1955, Henry 1961, Chuganova 1963, Narver 1969, Major et al. 1972). Freshwater and first marine year scale growth zones and annuli were distinguished along the posterior-anterior scale axis through the scale focus (Figure 5). Prints were made of one scale from each steelhead and used for subsequent analysis. The criteria for establishing annuli, false checks, freshwater plus growth, circuli counts, and spawning checks on each scale followed the methodology of Chuganova (1963), Narver (1969), and Tanaka et al (1969). Freshwater annuli were identified by any narrowing of circuli and/or the space between circuli including cutting over of the first circulus of new year's growth. Saltwater annuli were identified as the last circulus in a region of narrowing which preceded marked increases in circulus spacing. Given the subjective nature of scale reading (Conrad, 1984), the author's aging technique was verified by an independent source for a random sample of fifty scales. In addition, a subsample of one hundred scales was reread by the author six months after the initial reading. Validation of 19 Figure 5. Adult steelhead scale from the Sustut River. Total age is 3.2+. Shown is the measurement axis used for aging and measurement of scales in this study. Each annulus is marked by the horizontal lines; a region of spring plus growth precedes ocean entry (34X magnification). 20 21 scale growth at age for this study was not attempted. Age designations followed the methodology of Narver (1969). The time of annulus deposition was taken to be March 31, after Maher (1954). As an example of age designation, a steelhead of age 4.1S1+ is in its seventh plus full year of life. It spent four complete winters in freshwater (4) before smolting to sea (.) where it spent the next winter (1) and part of the next summer in saltwater before returning in the fall and spawning (S) the next spring. It then survived, migrated back to sea and spent the next winter (1) and part of the next summer again in saltwater before returning in the fall (+) to potentially spawn again the next spring. All scales were analysed for four measurements and two circuli counts in each yearly freshwater scale zone and in the first marine scale zone (table 2). Measurements were made to the nearest 0.01mm using Helios calipers on each scale print print held to low power under a Wild M5 stereo microscope. The scale variables used in this study were selected for analysis because of their successful use in other scale pattern studies (Anas and Murai 1969, Bilton 1971, Lear and Sandeman 1980, Conrad 1984). As Skeena River steelhead spend from one to five years in freshwater and from one to five years in saltwater, the number of scale variables recorded for each steelhead was dependent upon freshwater age. Only steelhead of the dominant freshwater Skeena River age groups (3 and 4) were used in this study. One hundred seventy five scale samples per week were 22 Table 2. Variables measured from the adult steelhead scales for each stock. Variable Definition PG Presence (1) absence (2) of plus growth A1 Distance to second circulus in year 1 A2 Distance to fourth circulus in year 1 A3 Distance to sixth circulus in year 1 A4 Total width of scale zone in year 1 A5 Number of circuli half across year 1 A6 Number of circuli full across year 1 B1 Distance to second circulus in year 2 B2 Distance to fourth circulus in year 2 B3 Distance to sixth circulus in year 2 B4 Total width of scale zone in year 2 B5 Number of circuli half across year 2 B6 Number of circuli full across year 2 CI Distance to second circulus in year 3 C2 Distance to fourth circulus in year 3 C3 Distance to sixth circulus in year 3 C4 Total width of scale zone in year 3 C5 Number of circuli half across year 3 C6 Number of circuli full across year 3 D1 Distance to second circulus in first ocean year D2 Distance to fourth circulus in first ocean year D3 Distance to sixth circulus in first ocean year D4 Total width of scale zone in first ocean year D5 Number of circuli half across first ocean year D6 Number of circuli full across first ocean year E1 Distance to second circulus in year 4 E2 Distance to fourth circulus in year 4 E3 Distance to sixth circulus in year 4 E4 Total width of scale zone in year 4 E5 Number of circuli half across year 4 E6 Number of circuli full across year 4 Additional variables = L Length WT Weight Sex Sex FWA freshwater age SWA saltwater age 23 attained from incidentally caught steelhead in the six week area 4 commercial salmon fishery during mid-July through August of 1984. Seiner and packer offloads from gillnetters were randomly sampled at the end of each two to four day weekly fishery opening. Fork length (to the nearest 0.5cm), weight (to the nearest 0.5kg) and sex were recorded for each scale sample. All sampling was conducted at the Prince Rupert plant of B.C. Packers Limited. An examination of sales slips indicted that 20 to 60% of the total area 4 incidental catch passes through B.C Packers facility. Attempts to use Department of Fisheries and Oceans test fishery steelhead scale data for 1984 and past years were limited by small sample sizes and a high incidence of regenerate scales present in the data base. Determination of Sample Sizes Required sample sizes for this study followed the methodology of Clutter and Whitesel (1956). Using sockeye salmon as an example they showed that scale sampling variation could be kept to within plus or minus one half a circulus of a true population mean (95% confidence level) with a sample of sixty scales. Previous estimates of scale pattern variance in Skeena River steelhead were not available. The author used a maximum expected standard deviation (in circuli count) from the true mean in any scale zone of one. From the modified formula of Clutter and Whitesel (1956, pages 75-82) and assuming that the sample means in this study were normally distributed, 95% of the sample means of size n from a given stock should lie within 24 two standard errors of a given sample mean. For 95% of the means to lie within plus or minus one circulus of a given sample mean, 136 scales from each stock were required for stock separation purposes. For 90% of the means to lie within the same confidence interval, a sample of 97 was required. Sample sizes from the commercial fishery were difficult to determine because of the number of stocks involved, the diverse age structure of steelhead in the catch, and the highly variable nature of the fishery. Anas and Murai (1969) utilized Worlund's (1960) precision curves (page 172) for maximum expected error of classification in deducing favorable sample sizes for classifying sockeye salmon on the high seas. Following their methodology I chose an estimated error rate in correct classification for Skeena River steelhead of between 15 and 30 percent. The weekly mixed fishery samples required for this study were then calculated at between 150 and 200 (90% confidence level). Juvenile Analysis In August of 1983, thirty steelhead parr were collected by electroshocker and seine from the lower reaches of the Morice-Bulkley, Kispiox, and Zymoetz rivers respectively. Morphological comparisons were conducted between the juveniles in order to assess morphological features and to compare overall body form in the different rivers. Ten body measurements, following Hubbs and Lagler (1967) were attained from each specimen. These were head length (HL), head depth (HD), head 25 width (HW), caudal peduncle depth (CD), caudal peduncle width (CW), body depth (BD), body width (BW), predorsal length (PrDL), and post dorsal length (PoDL). As the parr were of various ages and size, the effects of allometry were removed by standardizing the data to pooled grand mean standard length (Thorpe, 1976). Log-log (base 10) regressions for each variable on standard length were adjusted according to the correction procedure: log(y)=logy-b*(logX-logX') (1) where log(y) was the adjusted variable value, log Y was the initial variable value, b was the regression coefficient for the regression of each variable against standard length, and log X' was the grand mean standard length. Antilogs (log(y)) were used in a discriminant analysis of morphological features to assess the separability of juveniles from the three systems. Analytical Techniques for Scale Pattern Analysis Linear discriminant function analysis (Fisher 1936, Dixon 1981) was applied to the adult scale data for calculating the decision rules for stock separation and classification. Linear versus quadratic discriminant analysis was chosen because a) other studies had used linear models successfully b) the underlying distributions of scale pattern features seemed to be normal and c) linear analysis was readily implementable. Models utilizing steelhead of the two dominant freshwater age classes (3 and 4) were constructed using those scale variables which 26 were both normally distributed in univariate comparisons and which had high F scores in one way analyses of variance. Univariate ANOVAS, multivariate ANOVAS, and the discriminant analyses performed in this study were generated using BMDP (Dixon, 1981) software. Discriminant analysis is a multivariate technique for separating and analyzing differences present in previously established groups of objects (Pimental, 1979). A discriminant function is the linear combination of p observed variables which maximizes between group variance relative to within group variance (Fisher, 1936). The rationale for using discriminant analysis stems from the usual inability to statistically distinguish between known groups using univariate methodology (Jolicouer, 1959). For Skeena River steelhead, each stock represents an established group of known origin (a learning sample) in multivariate space which can be represented by a multivariate normal probability density function. The linear array of scale measurements (vector) from each steelhead describes the location of that individual in multivariate space. Individual steelhead from the same stock should occupy a common region in multivariate space defined by the dispersion (variance-covariance) of individuals about the common stock average for all variables (the stock centroid). Multivariate analysis of variance was used to test the significance of differences between stock centroids in this study. The rejection of equality between centroids is a prerequisite for discriminant analysis. Appendix A outlines the 27 methodology of discriminant analysis as it applied to this study. Figure 6 shows the basic relationship between euclidean and discriminant space for a hypothetical three variable, three stock analysis. 28 Figure 6. Discriminant space. Euclidean three variable space for three hypothetical steelhead stocks. The multivariate swarms of data points (individuals), considered one variable at a time, fail to to separate in euclidean space along any single variable plane: wx, xy, or yz. Linear combinations of the original variables and projection of the resulting canonical variables to two axis discriminant space best separates the groups. The + denotes centroids for each group, the ( + ) denotes the grand mean centroid with a mean of 0 and a standard deviation of one in discriminant space. 3 variab.le euclidean space w a a + a a b b b + b b b < + ) c c c + c c c function 2 functi on Two axis discriminant space 30 RESULTS Discrimination of Skeena River Steelhead Verification of scale aging The steelhead scales used in this study exhibted variable readability. Some scales had to be reread two and three times because of poor annular definition and scale clarity. In general, all scales exhibited narrow freshwater growth zones which often made annular placement difficult. Still, of the fifty randomly selected scales read for age by an outside source, 93% were in agreement with the authors' designation of age. A sample of one hundred scales reread by the author approximately six months after the initial reading resulted in nine scales being changed for designation of age. Descriptive statistics Age composition Age composition structure within and between each of the five major Skeena River steelhead stocks was found to be diverse (Appendix table 1 and figure 7). Of the original 475 scales collected for analysis, 466 had readable fresh and saltwater scale growth zones. For all stocks six dominant age classes (3.1+, 3.2+, 3.3+, 4.1+, 4.2+, 4.3+) were evident from the data as well as six minor ones (2.1+, 2.2+, 3.4+, 4.4+, 5.1+, 5.2+) 31 Figure 7. Age composition structure for the five steelhead stocks used in the study. RS denotes repeat spawners (compiled from appendix table 1). AGE COMPOSITION KISPIOX RIVER : n=103 I U I III aV x%" yVy kS* t.y »>* & AGE CLASS SUSTUT RIVER : n=90 JUuLu 1>* %V V y1-* ^" l>* Ll* U>* & ACE CLASS BABINE RIVER : n=91 2 O 0.4 , P._^_P-_.»-.. a>* i>* v^* y** y^* »>* kV k>* & AGE CLASS I yi> ki* k>* & AGE CLASS MORICE/BULKLEY RIVER : n=89 O £ 0.2 I 1>" i%* V>* yl* y** O* (..V »>* ^ AGE CLASS ZYMOETZ RIVER : n=92 I UJUL-JK aV V yi* yV *>* AGE CLASS 33 and six or seven repeat spawner age classes (3.1S1+, 3.251+, 4.1S1+, 4.2S1+, 4.1S1S1+, 3.S1+ etc). From Appendix table 1, the most common age classes over all stocks were 4.2+ (31%) and 3.2+ (27%). Repeat spawners were apparent in 12% of the total data base. Maiden spawners to the five Skeena stocks had spent, on average, two (1%),three (45%) and four (54%) years in freshwater prior to smolting and one (18%), two (66%) and three (15%) years in saltwater prior to spawning. Reduced maturation and growth rates in harsher northern environments (Ricker, 1972) would explain the older freshwater ages of Skeena River steelhead compared to southern steelhead stocks (see Shapovalov and Taft 1954, Withler 1966, Horncastle 1981). By stock, steelhead from the Zymoetz River were predominantly of ages 4.2+ (35%) and 3.2+ (25%); those from the Kispiox River were predominantly of ages 3.2+ (29%), 3.3+ (26%) and 4.2+ (25%); those from the Morice-Bulkley River were predominantly of ages 4.1+ (43%), 4.2+ (22%) and 3.1+ (15%); those from the Babine River were predominantly of ages 3.2+ (62%) and 4.2+ (26%); and those from the Sustut River were predominantly of ages 4.2+ (50%), 4.3+ (16%), 3.2+ (13%), and 3.3+ (11%). Table 3 summarizes the age composition features of each stock as read from their scales according to smolt age, ocean age, and contributions by sex. Kispiox River steelhead had long ocean residencies (32% 3+) while those from the Morice-Bulkley River had relatively short ocean residencies (64% 1+). 90% of the Babine River steelhead had spent 2+ years in the ocean while only 2% had spent three or more. . The incidence of 34 Table 3. Age Composition features of Skeena River steelhead by sex, smolt age, and ocean age (compiled from appendix table 1, ** denotes maiden spawners only). PROPORTION OF STOCK FEATURE: Zymoetz Ki spiox Mor ice Babine Sustut n = 92 n=1 03 n=90 n=9l n=90 1)adults of:** -smolt age 3 46% 55% 27% ' 67% 27% -smolt age 4 54% 45% 73% 33% 73% -ocean age 1+ 1 4% 7% 64% 8% 1% -ocean age 2 + 76% 61% 34% 90% 70% -ocean age 3+ 1 0% 32% 3% 2% 29% 2)repeat spawners 22% 1 4% 1 0% 2% 9% 3)sex ratio (f/m) 1 .2/1 1.1/1 1 .5/1 1 .8/1 1 .5/1 4)females of:** -ocean age 1+ 8% 1 0% 70% 7% 0% -ocean age 2+ 87% 69% 30% 91% 85% -ocean age 3+ 5% 3% 0% 2% 1 5% 5)ma1es of -ocean age 1+ 20% 4% 53% 9% 2% -ocean age 2+ 65% 55% 38% 88% 49% -ocean age 3+ 1 5% 41% 6% 3% 49% repeat spawning was highest in those stocks closest to the ocean (eg Zymoetz 22%) and least in those stocks farthest away (eg Babine 2%). This suggests a higher incidence of kelt survival in downstream Skeena River stocks. Limited sample sizes made testing the hypothesis of within stock age class homogeneity between years difficult. Several studies; however, support such a trend in steelhead (Maher 1954, Maher and Larkin 1955). Narver (1959) found slight differences in the proportions of age 3.2+ steelhead (73% in 1967, 60% in 1968), 3.3+ steelhead (10% in 1967, 23% in 1968), and 4.2+ steelhead (8% in 1967, 11% in 1968) in the Babine River between 35 years. 62% of the Babine River steelhead used in this study (1977) were of age 3.2+ (1% were of age 3.3+ and 26% were age 4.2+). For both the Morice and Sustut River steelhead data used in this study the proportional dominance of the major age classes changed little between years (Morice-Bulkley 1976 4.1+=37% , 4.2+=24% 1977 4.1+=43% 4.2+=21%: Sustut 1977 4.2+=46% 1983 52%). Given that age at maturity has a heritable basis in salmonids (Ricker, 1972) the age class structure of Skeena River steelhead may reflect selection for successful reproduction in river specific environments. Sizes at age Appendix table 2 summarizes the mean sizes at age for the five Skeena River steelhead stocks. Size at age was found to be a function of saltwater and not freshwater residence time. Both the mean lengths and weights of 3.2+ and 4.2+ steelhead were similar within stocks but significantly different between stocks (ANOVA length P<0.001, weight P<0.001) for the sexes combined (figures 8 and 9) and by sex alone (age 3.2+, figure 10). For a given ocean age stock differences by sex were quite pronounced; Kispiox River ocean age 2+ males were 11% longer (mean length= 88.3cm) and 40% heavier (mean weight =7.9kg) than Morice River ocean age 2+ males (mean length=79.4cm, mean weight=4.7kg). Over all stocks and ages, Kispiox and Sustut River steelhead were predominantly the largest, Morice-Bulkley River steelhead were predominantly the smallest. Variations in size between Skeena River steelhead stocks 36 Figure 8. Mean lengths of age 3.2+ and 4.2+ steelhead from the five stocks used in the study. Shown are the means +/- one standard error about the mean, the 95% confidence interval about the mean, and the sample size for each age class. Figure 9. Mean weights of age 3.2+ and 4.2+ steelhead for the five stocks used in the study. Shown are the means +/- one standard error about the mean, the 95% confidence interval about the mean, and the sample size for each age class. 37 -i-33 121 12 12 44 I -I-I •I-I 4.2+ 3.2+ 4.2+ 3.2+ 4.2< I I I I 3.2+ 4.2+ 2+ 3.2+ 4. IKispioxI IZymoetzI IMorIce I I Bablnel ISustut I Rivar and Ag« 27 22 33 42 12 I -I -+ -I-II II II till 3.2• 4.2+ 3.2+ 4.2+ 3.2+ 4.2+ 3.2» 4.2+ 3.2+ 4.2+ IKispioKl IZymootzi ;Moricp I I Babine! River and rtge I bustict I 38 Figure 10. Mean lengths by sex for age 3.2+ steelhead from the five stocks used in the study. Shown are the means +/- one standard error about the mean, the 95% confidence interval about the mean, and the sample size .for each sex. 90-lO L E 85-! N G T H 80-! (cm) • 17 I 20 75-I 70- ! 65-1 I M M F M F M F 11 F iKispioxl iZymoetzl IMorice i t Babine! River and Sex ! i M F ISustut I 40 have been noted (Whately et al. 1978) and probably reflect genetic differences in ocean growth rates, variable ocean feeding behaviors, or both. Table 4 reports the stock specific mean lengths and weights of males and females by ocean age for each stock use in the study. Table 4. Mean Lengths and Weights for Skeena River steelhead of various ocean age (standard errors about the mean available from appendix table 2). RIVER FEATURE Zymoetz Kispiox Mor ice Babine S u s t LENGTH cm) 1)males: ocean age 1+ 57.3 63.5 59.0 58.8 55.9 ocean age 2+ 81.0 88.3 79.4 77.3 84.4 ocean age 3+ 94.0 97. 1 91 .5 91 .4 94.7 2)females: ocean age 1+ 64.9 57.8 56.3 60.3 63.5 ocean age 2+ 75.2 80.4 72.5 76.4 77.2 ocean age 3+ 84.2 87.3 — — 87.0 WEIGHT (kg) 1)males ocean age 1+ 2.2 2.2 1.8 2.0 1.8 ocean age 2+ 7.9 7.9 4.7 4.4 6.4 ocean age 3+ 8.1 9.6 7.4 7.4 8.9 2)females ocean age 1+ 2.6 2.4 1 .7 2.0 2.7 ocean age 2+ 4.4 5.6 3.3 4.5 4.3 ocean age 3+ 5.7 7.2 — — 6.0 Scale Pattern Features of Skeena River steelhead Analysis of the scale variables used in this study revealed them to generally be normally distributed (using BMDP7D, Dixon, 1981). The following sections summarize only the width and 41 circuli count scale features found within each scale zone for the five Skeena River steelhead stocks. Intracircular distance differences, which reflect both zone widths and circuli counts, are presented separately in the appendix summary tables. The adult steelhead of younger smolt age used in this study had both wider freshwater scale zones and more circuli in each scale zone than did the adults of older smolt age (figure 11), which supports the notion of slower growth rates in older smolts (Ricker, 1972). It was also found that adult steelhead of the same smolt age but of different ocean age had similar within stock freshwater scale pattern features. For example, Kispiox and Zymoetz River age 3.1+, 3.2+, and 3.3+ steelhead exhibited nonsignificant differences (ANOVA P>0.10) in yearly freshwater scale zone widths and circuli counts for each of the three age classes within each stock respectively. The same was was found for Morice-Bulkley River age 4.1+ and 4.2+ steelhead (Table 5). This suggested that scale pattern comparisons could be made using adult steelhead of similar smolt age but of pooled ocean age from each of the five Skeena River stocks. Scale features of smolt age 3 adult steelhead Appendix table 3 summarizes the descriptive statistics for scale growth in adult steelhead of smolt age 3 from the five Skeena River stocks by pooled ocean age (3.1+, 3.2+, 3.3+ etc). Significant differences for the majority of measured scale features were found. Table 6 summarizes the scale zone width and circuli count differences between the five stocks. Both 42 Figure 11. Mean scale zone widths for steelhead of smolt ages 3 and 4. Shown are the yearly scale zone means (mm) for all stocks combined +/- one standard error about the mean, the 95% confidence interval, and the sample sizes for each smolt age. FWA3=smolt age 3, FWA4= smolt age 4. The width of the first ocean year scale zone is given with the standard error about the mean in brackets. 43 FRESHWATER I OCEAN I 44 Table 5. F statistics and associated probabilities for one way ANOVAs within stocks for differences in mean yearly freshwater scale zone widths for various age classes in the Kispiox, Zymoetz, and Morice rivers. (* denotes no significant difference at the 5% level of significance). River Age classes Variable F D.F P compared Kispiox 3.1+,3.2+,3.3+ A4 0 .34 2,45 * 0.79 B4 2 .06 2,45 * 0.12 C4 1 .46 2.45 * 0.27 D4 1 .34 2,45 * 0.27 Zymoetz 3. 1 + ,3.2+,3.3 + A4 0 .54 2,33 * 0.66 B4 1 .51 2,33 * 0.23 C4 2 .20 2,33 * 0.11 D4 2 .10 2,33 * 0.12 Morice 4.1+, 4.2+ A4 0 .25 1 ,62 * 0.78 B4 1 .19 1 ,62 * 0.31 C4 0 .89 1 ,62 * 0.78 D4 1 .34 1 ,62 * 0.27 E4 0 .21 1 ,62 * 0.80 scale zone widths and circuli counts in the second year differed the most between the five stocks. Figures 12 and 13 summarize these differences graphically. Adults of smolt age 3 from the Morice-Bulkley River had the widest first year scale zones while adults from the Zymoetz River had the smallest first year scale zones. This suggests either earlier emergence times and/or better first year growth in productive rearing environments for the former and vice versa for the latter. Interestingly, both scale zone widths (figure 12) and scale zone circuli counts (figure 13) decreased markedly in Morice-Bulkley River smolt age 3 adults after the first year of growth. This was the only stock to show such a trend and suggests either high competition for food or displacement of parr into areas of less 45 Figure 12. Yearly freshwater scale zone widths in adult steelhead of smolt age 3 by pooled ocean age. Shown are the means and the 95% confidence interval about the means. Figure 13. Yearly freshwater scale zone circuli counts in steelead of smolt age 3 by pooled ocean age. Shown are the means and the 95% confidence interval about the means. River and Yearly Scale Zone 47 Table 6. F statistics and associated probabilities for one way ANOVAS between stocks for smolt age 3 steelhead by pooled ocean age (* no significant difference at the 5% level of significance)) Variable DF=4,186 F P A4: year 1 width. 3.80 P<0. 02 B4: year 2 width. 8.18 P<0. 001 C4: year 3 width. 4.50 P<0. 005 D4: 1st ocean " . 1 .67 *P>0. 20 A5: year 1 circ. 6.26 P<0. 005 B5: year 2 circ. 6.48 P<0. 001 C5: year 3 circ. 5.59 P<0. 001 D5: 1st ocean " . 1 .67 *P>0. 20 productivity. Conversely, Babine River adult steelhead of smolt age 3 showed large incremental scale growth between years (figures 12 and 13) which suggests parr growth in highly productive environments. Scale pattern features (zone widths and circuli counts) were most similar in adult smolt age 3 steelhead from the Kispiox and Zymoetz rivers (figures 12 and 13) which suggests growth in comparable environments. Scale growth in the first marine year was not significantly different between the five Skeena River stocks for adults of smolt age 3 (ANOVA: widths 0.10 <P< 0.20, circuli counts 0.20 <P< 0.50). This may be attributable to a large level of within stock variance for marine growth in steelhead of different ocean ages. Invariably, the author found the widths of the first marine zone in 3.1+ steelhead to be notably narrower than those of age 3.3+ steelhead from the same stock. This suggests faster maturation rates in the younger adults and would result in nonsignificant differences between the stocks when combining the 48 ages in a pooled ocean age analysis. Between stock comparisons of first marine year scale growth may be valid only when individuals of the same ocean age are used. Interestingly, steelhead of smolt age 4 showed the opposite trend. Scale features of smolt age 4 adult steelhead Appendix table 4 summarizes the descriptive statistics for scale growth in adult steelhead of smolt age 4 from the five Skeena River stocks by pooled ocean age (4.1+, 4.2+, 4.3+ etc). Significant differences for the majority of measured scale features were found. Table 7 summarizes the F statistics generated for the scale zone width and circuli count differences between the five stocks in a one way analysis of variance for adults of smolt age 4. First year circuli counts and the widths of the fourth year differed the most between the five stocks. From the F scores, adult steelhead of smolt age 4 were more different between the five stocks than were adults of smolt age 3. The slower growth rates and longer residence times of 4 year olds may enhance stock differentiation by scale pattern analysis. Figures 14 and 15 summarize the between stock dfferences for scale widths and circuli counts in adults of smolt age 4 and pooled ocean age graphically. Incremental scale zone growth (widths and circuli counts) was again large in the first year and small in subsequent years for Morice River steelhead. Sustut River adults of smolt age 4 showed wide incremental scale growth zones during all freshwater years. Wider but fewer circuli were 49 Figure 14. Yearly freshwater scale zone widths in adult steelhead of smolt age 4. Shown are the means and the 95% confidence interval about the means. Figure 15. Yearly freshwater scale zone circuli counts in adult steelhead of smolt age 4. Shown are the means and the 95% confidence interval about the means. 50 . 40-n=37 n=4B n=65 n=30 n=5<? W I D T H (mm) . 30-. 20- 1/ !./! V 15-I 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 Kispiox Zymoetz Morice Babine Sustut River and Yearly Scale Zone I n=37 n=4B n=65 n=30 n=59 ! 5-1 i i 0 1234 1234 1234 1234 1234 Kispiox Zymoetz Morice Babine Sustut River and Yearly Scale Zone 51 Table 7. F statistics and associated probabilities for one way ANOVAS between stocks for smolt age 4 steelhead by pooled ocean age (* no significant difference at the 5% significance level) Variable DF=4,234 F P A4: B4: C4: D4: E4: A5: B5: C5: D5: E5: year 1 width year 2 width year 3 width 1st ocean " year 4 width year 1 #circ year 2 #circ year 3 #circ 1st ocean " year 4 #circ 8.61 7.40 7.64 4.32 1 2.82 18.45 3.40 0.93 5.79 4.99 P<0.001 P<0.001 P<0.001 P<0.005 P<0.001 P<0.001 P<0.05 *P>0.50 P<0.001 P<0.002 apparent in Kispiox River smolt age 4 steelhead after the first year (figure 15) even though the scale zones were increasing in width (figure 14). Kispiox and Zymoetz River adults of smolt age 4 again had similar patterns of scale zone growth. In contrast to the results of the previous section, first marine year scale growth (width) was significantly different between the five ' Skeena River stocks for adult steelhead of smolt age 4 (ANOVA width P<0.05, circuli count P<0.05). This suggests less within stock variance of first year marine growth between smolt age 4 adults of different ocean age than for smolt age 3 adults. Variable feeding and/or migrational patterns for 4 vs 3 year old smolts from each stock may explain the differences. Healey (1983) notes that different "types" of salmonid smolts (by stock, size, age, etc) may respond characteristically to marine environments by growing differently or similarily. 52 Scale features of age 3.2+ and 4.2+ steelhead Skeena River steelhead of specific age were also analyzed to assess stock differences in scale pattern features. Age specific stock identification is usually desirable so as to remove any possible variations in scale growth attributable to variations in age. However, as noted previously, freshwater scale growth was found to vary nonsignificantly in steelhead of different ocean age. Still, the two dominant steelhead age classes (3.2+, 4.2+) were analyzed to satisfy general methodology and to better compare scale growth in the first marine year. Appendix tables 5 and 6 summarize the univariate statistics for scale features in age 3.2+ and 4.2+ steelhead from each of the five Skeena River stocks. The results of one way analyses of variance for differences in certain scale features are summarized in table 8 for the two age classes respectively. Figures 16 and 17 summarize the between stock differences for scale zone widths alone. Significant between stock differences were found in all zones except for circuli counts in years two and three for age 4.2+ steelhead and, interestingly, first marine year widths in both age 3.2+ and 4.2+ steelhead. Figures 16 and 17 show that the zone differences were similar to those of the pooled ocean age analyses (Figures 12 and 14). Concerning the nonsignificant differences in first marine year widths for 3.2+ and 4.2+ steelhead, this result suggests comparable between stock scale growth in the first ocean year. However, the numbers of circuli (table 8) and the distances to 53 Figure 16. Yearly freshwater scale zone widths in steelhead of age 3.2+. Shown are the means and the 95% confidence interval about the means. Figure 17. Yearly freshwater scale zone widths in steelhead of age 4.2+. Shown are the means and the 95% confidence interval about the means. 54 River and Yearly Scale Zone 55 Table 8. F statistics and associated probabilities for one way ANOVAS between stocks for age 3.2 + and 4.2+ freshwater and first marine year scale zone widths and circuli counts (* indicates no significant difference at the 5% significance level) . Variable Age 3. F 2+ DF= P 4, 1 36 Age 4. F 2+ DF= P 4, 185 A4: year 1 width 3. 97 P<0. 01 7. 41 P<0. 001 B4: year 2 width 6. 64 P<0. 001 4. 75 P<0. 005 C4: year 3 width 5.' 68 P<0. 001 2. 93 P<0. 05 D4: 1st ocean " 2. 68 *P>0. 05 0. 1 6 *P>0. 50 E4: year 5 width -- -- 1 1 . 25 P<0. 001 A5: year 1 circ. 5. 22 P<0. 002 18. 93 P<0. 001 B5: year 2 circ. 5. 15 P<0. 002 2. 75 *P,0. 10 C5: year 3 circ. 8. 22 P<0. 001 0. 79 *P<0. 50 D5: 1st ocean " 3. 06 P<0. 05 4. 46 P<0. 005 E5: year 5 width -- — 5. 26 P=0. 001 circuli in the first marine zone (variables D1, D2, and D3) were all significantly different between the stocks. These results are difficult to explain. Either yearly variations in ocean growth are being reflected in the data base or, alternatively, the differences are real and reflect stock specific genetic and/or feeding differences. Scale pattern variation between years Comparisons were made to assess the degree of scale pattern variation within stocks between years even though steelhead of different ages (eg 3.1+, 3.2+,3.3+ etc), and thus brood years, sampled in the same year exhibited non-significant differences. Sustut River steelhead of smolt age 4 (1977 n=24, 1983 n=38) and Zymoetz River adult steelhead of smolt age 3 (1975 n=l9 1978 n=20) revealed significant between years differences (table 9) 56 only for second year and first marine year scale growth in the Sustut River stock. While it is . difficult to draw strong Table 9. Results of one way ANOVA'S for between years differences in scale growth for Sustut River smolt age 4 and Zymoetz River smolt age 3 adult steelhead A=Sustut 1977 n=24, 1978 n-38 B=Zymoetz 1975 n=l9 1978 n=20 (* no significant difference at the 5% level) Scale Variable df 1,60 (A) df 1,39 (B) A4 A5 B4 B5 C4 C5 D4 D5 E4 E5 F .03 .16 4.25 5.92 .04 .04 17.68 15.22 2.23 5.71 P .85 * .69 * .04 .02 .84 * .85 * <. 00 1 <. 00 1 . 14 * .02 F .95 1 .59 2.85 .01 .00 .17 .04 1 .62 - -P .33 * .23 * .10 * .92 * .97 * .69 * .84 * .21 * - -conclusions from these results, small yearly variations in scale features may be expected in all Skeena River steelhead stocks if rearing conditions remain fairly stable between years. As an index of environmental stability, an examination of flow rate data revealed considerable variation between years for each of the five major Skeena River tributaries. The influence of such fluctuations on instream productivity and scale growth is not known. For this study the primary concern was that any changes in scale pattern growth between years within stocks be smaller than the changes in scale pattern growth between years between stocks. 57 Plus growth The incidence of freshwater plus growth prior to onset of the first ocean year was highest in steelhead from the Kispiox (35.6%) and Morice-Bulkley (33.3%) rivers followed by the Babine (21.7%), Sustut (20.7%), and Zymoetz (20.6%) rivers respectively. Plus growth reflects rapid growth to smolt size and may differ between stocks (and between years) according to the level of maturity reached in the last freshwater year. While significant trends were not readily apparent, Skeena River adult steelhead of smolt age 3 tended to show a higher incidence of plus growth than did adults of smolt age 4. Stock Discrimination Twelve discriminant analysis models were constructed in this study for separating the five Skeena River steelhead stocks. As the majority of scale variables used were normally distributed, the assumption of multivariate normality was accepted. The null hypothesis of equal dispersion matrix equality was rejected for several of the models, a finding often observed by other workers (Conrad, 1984). However, discriminant analysis is still justified in most cases because of the power of MANOVA in detecting significant ' and nonsignificant differences between groups (Pimental, 1979). 58 Separate freshwater age discriminant models Figure 18 summarizes the results of a five stock discriminant analysis using Skeena River adult steelhead of smolt age 3 and pooled ocean age. Only scale variables are included. Multivariate analysis of variance indicated highly significant differences among centroids for the the five stocks (approximate F= 5.24 DF= 40 676 P<0.001). Pairwise comparison of centroids showed that all ten comparisons were significant (range of F= 3.76-7.98 DF= 10 178). The four canonical functions accounted for 38.5%, 36.3%, 17.8%, and 7.4% of the explained between stock variance respectively. Ten of the original twenty four scale variables were selected for function construction, the four best discriminating variables being B1, C1, C2, and A2. From figure 18, the first discriminant function primarily separated Kispiox and Sustut River steelhead on the basis of distances to the second circulus in years two and three. As previously noted, Sustut River steelhead of smolt age 3 had much wider scale zones than did Kispiox River steelhead of smolt age 3, this best being reflected by the interzone circuli distances.The second discriminant function primarily separated Babine from Morice-Bulkley River steelhead on the basis of scale variables C1 and A2. The proximity of centroids in figure 18 (especially of the Kispiox to the Zymoetz) portrays the relatively high degree of scale variable overlap between the five stocks for steelhead of smolt age 3. Paitwise comparisons revealed that the patterns of freshwater scale growth in smolt age 3 adults were most similar for the Sustut to Zymoetz, 59 Figure 18. Discriminant function analysis describing scale pattern variation in adult steelhead of smolt age 3. The letters indicate the stock centroids S=Sustut Z=Zymoetz K=Kispiox M=Morice B=Babine *=grand centroid, the open circles indicate the 90% confidence interval about each centroid (from Pimental, 1979), and the lines point to the next most similar stock in discriminant space. The first two standardized discriminant functions are given below. D1 = -37.23C2 +61.33C1 +0.19A5 +0.50B4 -27.19A2 +69.28B1 +0.27PG -24.17B3 +5.80D3 +25.50A1 -0.12 D2 = 25.01C2 -58.24C1 -0.09A5 -.83B4 +29.27A2 -26.15B1 +0.28PG +0.11B3 -7.04D3 -24.49A1 -4.12 60 -2.0 -2.0 -1.5 -1.0 -0.5 0 +0.5 +1.0 +1.5 FUNCTION 1 61 Zymoetz to Kispiox, Kispiox to Zymoetz, Babine to Kispiox, and Morice-Bulkley to Kispiox stocks. Using Lachenbruch's (1975) holdout procedure, 45.3% (range Zymoetz 29%-Sustut 69%) of the smolt age 3 adults were correctly classified to stock of origin (Table 10) by the classification technique. Figure 19 summarizes the results of a five stock discriminant analysis using Skeena River adult steelhead of smolt age 4 and pooled ocean age. Only scale variables are included. Multivariate analysis of variance again indicated highly significant differences among centroids for the five stocks (approximate F= 7.48 DF= 52 861 P<0.001). All pairwise comparisons between the five stocks were significant (range of F= 2.04-14.87 DF= 13 222). The first two canonical functions accounted for 50.5% and 26.1% of the explained between stock variability (17.6% and 5.7% for the third and fourth functions respectively). Thirteen of the original thirty scale variables were selected for function construction, the four best discriminating variables being C1, B1, C3, and B3. From figure 19, the first discriminant function primarily separated adults of smolt age 4 from the Sustut and Morice-Bulkley Rivers as being the most distinctly different, again primarily on the basis of distances to the second circulus in years two and three (large in the Sustut, small in the Morice-Bulkley). Kispiox and Zymoetz River steelhead were again found to be the most similar. The second discriminant function primarily separated Babine River steelhead from the other four stocks on the basis of variables C3 and B3. Pairwise comparisons revealed that the 62 •s A I? UiSi 1 ! !l« V I s:l-55S5s il..... o b I J 3 i-2SS32 I553-S5S55 u« §1 2=SS2SS 33 It Si 52521 3 S3 S-5SSS5 11 i-s==is 1! "5 SeESSSS fa -SISS? Id I-=sS2s Hi ?§Ss = s'is=i I; 1 C •A SSsrs |-i!i£S i« 3 ::::: 5 E-2 I a; si I. -I! ic I l-ssssi aq 5-55=55 1?5 -«Sss I 1 i.:; 2^ -*a5Si |- -ss?ss !J! 1 I ri *-S2s5a "r^§=? J I-5S5S5 "1 ^ « r- - r- * J: 5-'. -=531= n i i! u U 4) m > C K c 4-1 -V to O >i cn cn cu in o rtj rH u u • 4J C <0 O -ri E 4J cn CD IJ o m o •rH cn 4J 3 »M ro o u in •r-t rH Jtf •<H T3 in in (0 rH U U3 I O O 4J 6 in 4-> O C 4-> C t> •<H (0 E CD •H .C V4 rH U CU cn 0) • n 4J TJ in 1 in o> rH Numbering error. Text for leaf 63 not available. 64 Figure 19. Discriminant function analysis describing scale pattern variation in adult steelhead of smolt age 4. The letters indicate the stock centroids S=Sustut Z=Zymoetz K=Kispiox M=Morice B=Babine *=grand centroid, the open circles indicate the 90% confidence interval about each centroid (from Pimental, 1979), and the lines point to the next most similar stock in discriminant space. The first two standardized dicriminant functions are given below. D1 = -9.61E3 +0.15A5 +0.09D5 +8.50D3 -15.24B3 +24.92B1 -6.28E4 -22.50C3 +33.68C1 +0.13B6 +0.13E5 +1.52A4 -0.19D6 +0.89 D2 = -1.30E3 -0.05A5 +0.01D5 +0.42D3 -1.61B3 -1.00B1 -0.41E4 -2.22C3 +0.87C1 +0.01B6 -1.02E5 +0.07A4 -0.01D6 +5.64 + 1.5--2.0 • i i i i i i i i •  II i-2.0 -1.5 -1.0 -0.5 0 +0.5 +1.0 +1.5 FUNCTION 1 66 patterns of freshwater scale growth in smolt age 4 adults were most similar for the Sustut to Zymoetz, Zymoetz to Kispiox, Kispiox to Zymoetz, Babine to Morice-Bulkley, and Morice-Bulkley to Babine stocks. Using Lachenbruch's (1975) holdout procedure, 58.2% (range Zymoetz 40%-Sustut 71%) of the smolt age 4 Skeena adults were correctly classified to stock of origin (Table 11). The inclusion of size related data (length and weight) in the discriminant analyses for smolt age 3 and 4 Skeena increased stock discriminance. This was not too suprising as the five stocks were shown to differ greatly with respect to sizes at age. 57.1% (range Zymoetz 38%-Sustut 65%) of the smolt age 3 Skeena adults and 58.6% (range Zymoetz 50%-Sustut 66%) of the smolt age 4 Skeena adults were correctly classified (tables 12 and 13) to stock of origin with the inclusion of length and weight data. Figure 20 summarizes the results of a five stock discriminant analysis using Skeena River adults of age 3.2+ based on scale variables alone. The results were similar to the smolt age 3/pooled age analysis. Again, significant differences were found among the centroids for the five stocks (approximate F= 7.02 DF= 40 483 P<0.001) by multivariate analysis of variance and in all pairwise comparisons between stocks. Classification accuracy for the age 3.2+ discriminant model was 51.8% (range Zymoetz 26%-Babine 71%) using scale variables alone and 61% using scale variables in conjunction with size related data (range Zymoetz 38%-Babine 72%) (Table 13A). Using scale variables alone Sustut, Babine and Morice-Bulkley River age 3.2+ 67 Figure 20. Discriminant function analysis describing scale pattern variation in steelhead of age 3.2+. The letters indicate the stock centroids S=Sustut Z=Zymoetz K=Kispiox M=Morice B=Babine *=grand centroid, the open circles indicate the 90% confidence intervals about each centroid (from Pimental, 1979), and the lines point to the next most similar stock in discriminant space. The first two standardized discriminant functions are given below. D1 = +0.07C5 +0.01A5 +26.06C2 +13.90B1 -16.62A2 -0.03D5 +1.08D4 -0.08B5 +0.42B6 +5.26A1 -8.2 D2 = +0.02C5 -0.01A5 -4.28C2 +17.26B1 -10.02A2 +0.05D5 -0.35D4 -0.07B5 +0.11B6 +9.55A1 -0.4 -1.5-i -2.0 i i i i i i i > i  i i i II i -2.0 -1.5 -1.0 -0.5 0 +0.5 +1.0 +1.5 FUNCTION 1 69 steelhead were separated along the first canonical function (figure 20) primarily by scale variables C2 and A2 while the second canonical function primarily separated the stocks on the basis of variable B1. Pairwise comparisons revealed that the patterns of freshwater scale growth in age 3.2+ steelhead were most similar for the Sustut to Zymoetz, Zymoetz to Kispiox, Kispiox to Zymoetz, Babine to Kispiox, and Morice-Bulkley to Kispiox stocks. Figure 21 summarizes the results of a five stock discriminant analysis using Skeena River steelhead of age 4.2+. Significant differences between stock centroids (approximate F= 6.91 DF= 44 671 P<0.001) and in all pairwise comparisons were again evident from multivariate analysis of variance. Classification accuracy for the age 4.2+ model was 55.3% (range Zymoetz -40%-Sustut 70%) using scale variables alone and 59.5% (range Zymoetz 44%-Sustut 71%) (Table 13B) using scale variables in conjunction with size related data. Using scale variables alone, Sustut, Morice and Babine River age 4.2+ steelhead were separated along the first canonical function (figure 21) primarily ~ by scale variables C1 and B1 while the second canonical function primarily separated the stocks on the basis of variable C3 (figure 21). Pairwise comparisons revealed that the patterns of freshwater scale growth in age 4.2+ steelhead were most similar for the Sustut to Zymoetz, Zymoetz to Kispiox, Kispiox to Zymoetz, Babine to Morice-Bulkley, and Morice-Bulkley to Babine stocks. 70 Figure 21. Discriminant function analysis describing scale pattern variation in steelhead of age 4.2+. The letters indicate the stock centroids S=Sustut Z=Zymoetz K=Kispiox M=Morice B=Babine *=grand centroid, the open circles indicate the 90% confidence interval about each centroid (from Pimental, 1979), and the lines point to the next most similar stock in discriminant space. The first two standardized discriminant functions are given below. D1 = +0.21A5 -9.31E3 +9.73D3 +0.05D5 -24.11B3 +39.69B1 -3.32E4 -24.14C3 +46.52C1 +0.15A6 + 0.73 D2 = -0.09A5 -0.48E3 -0.35D3 +0.01D5 -0.46B3 -0.02B1 -0.30E4 +1.26C3 -0.76C1 +0.58A6 + 3.41 71 FUNCTION 1 72 Pooled freshwater age discriminant models Several pooled smolt age/pooled ocean age discriminant analyses were constructed under the assumption that large stock differences in age composition, sizes at age, and scale features (relative to within stock differences) would distinguish the five stocks without resorting to smolt age specific models. Because both smolt age 3 and 4 steelhead from certain stocks tended to grow similarily (eg small scale zones in the Morice River) this suggested that the various freshwater ages (smolt) be pooled to best describe overall growth in each system. Table 14 summarizes the basic discriminant analysis results for a pooled freshwater age/ocean age model. Using Lachenbruch's (1975) holdout procedure the mean classification accuracy for the pooled smolt age/pooled ocean age model was 52.5% (range Zymoetz 35%-Sustut 67%) using scale variables alone and 61.8% (range Zymoetz 50%-Sustut 72%) using scale variables in conjunction with length and weight. Figure 22 summarizes the placement of stock centroids in discriminant space for the all variable pooled age model. Pairwise comparisons revealed that the patterns of freshwater scale growth by pooled age were most similar for the Sustut to Zymoetz, Zymoetz to Babine, Babine to Zymoetz, Kispiox to Zymoetz, and Morice-Bulkley to Zymoetz stocks. 73 Figure 22. Discriminant function analysis describing scale pattern variation in adult steelhead of pooled smolt age. The circles indicate the stock centroids S=Sustut Z=Zymoetz K=Kispiox M=Morice B=Babine *=grand centroid, the open circles indicate the 90% confidence interval about each centroid (from Pimental, 1979), and the lines point to the next most similar stock in discriminant space. The first two standardized discriminant functions are given below. D1 = 0.45 WT -0.13 A5 -3.97 D1 +14.76 C3 -18.14 C -0.06 D5 -0.06 C5 +0.13 L -2.31 D3 +0.94 A2 +0.13 D6 +0.02 B6 -15.11 B1 +9.44 B3 -7.28 A +0.06 PG -1.81 D2 = 0.30 WT +0.15 A5 +7.85 D1 -17.25 C3 +44.83 C +0.12 D5 +0.11 C5 -0.02 L +5.66 D3 -24.54 A2 -0.13 D6 +0.26 B6 +35.63 B1 -11.13 B3 +18.53 +0.06 PG -4.47 74 + 1. 5-FUNCTION 1 75 Discrimination by sex As both male and female steelhead have been shown to grow at similar rates in freshwater (Parker and Larkin, 1959) little effort was made to distinguish between the five Skeena River steelhead stocks on the basis of sex. Any success in differentiating the five stocks on the basis of sex must rely on differences in size at age; for this reason, the only discriminant models constructed by sex were for a pooled smolt age/pooled ocean age analysis. For such a model, female Skeena River steelhead were correctly classified to stock of origin using Lachenbruch's (1975) holdout technique with 65.% (range Kispiox 54%-Morice-Bulkley 76%) accuracy (table 15) while male Skeena River steelhead were correctly classified to stock of origin with 52.6% accuracy (table 16) using the same model. Several points concerning all of the above models are in order. Firstly, Kispiox and Zymoetz River steelhead had consistently lower classification success in all models than did any of the other three stocks. Misclassifications of each to the other were generally responsible for lowering the mean classification results of all models. Secondly, the variables chosen for stock discrimination were consistently from the second and third years of freshwater growth, which suggests that stock differentiation is most prominent well into parr stage. Thirdly, the range of variable overlap was high between the stocks, as indicated by the fairly low rates of classification (45%-61%) even though the differences between stock centroids were highly significant in each model. Finally, stock 76 discriminance was greatest in those stocks farthest from the ocean (Sustut, Babine, Morice-Bulkley) which suggests the presence of specific growth regimes/and or selection factors for growth towards the upper regions of the Skeena River drainage. Not surprisingly, a reduction in the number of stocks used in the analyses resulted in greater classification success for all discriminant models. In general, the results of discriminant analysis support the hypothesis of stock discreteness in Skeena River steelhead. Commercial Fishery Stock Composition The results of scale analysis indicated that classification of incidentally caught steelhead in the commercial salmon fishery to stock of origin was feasible. Only one year of commercial data (1984) were available for classification. Table 17 summarizes the Fisheries and Oceans statistical area 4 steelhead catch by week for the 1984 commercial salmon fishery. Incidental catches of steelhead in 1984 were the highest on record. As shown, peak catches occurred with peak effort (Weeks ending July 21 and 28) during the peak of sockeye salmon fishing. Figure 23 and appendix table 7 summarize the sample age composition of steelhead collected over the six week period 9-14 in 1984. Steelhead of age 3.2+ and 4.2+ were predominant although shifts in the other age classes were found. A proportional abundance of ocean age .1+ males was found in 1984, compared to females which were predominantly of ocean age .2+ (Table 18) The mean lengths and weights of steelhead in the 77 Figure 23. Age composition structure by week for steelhead sampled in the 1984 commercial fishery. (compiled from appendix table 7). WEEK 9 1984 :n=123 WEEK 12 1984 :N-127 -.1 %y -iX* ^k vi*' ^* »>* ACE CLASS WEEK 10 1984 :N=130 I • •111 a.^'aV *v va* y-i* ^ »i* ACE CLASS WEEK 11 1984 :N=134 a>* a** i>* yi* *>* ^ ACE CLASS 3 0.2 O a-%* a^* y%* yl* y**' k.l* kV ^ AGE CLASS WEEK 13 1984 :N=130 .III 11 i T,.\* %%* y\* va* y-s* t.a* k>* tf> AGC CLASS WEEK 14 1984 :N=108 I 1i1 I 1.1 *' ^n* .a" vb* k%" u>" & aN*V* bV ^" b>* <-a* AGE CLASS 79 TABLE 17. Steelhead catch and escapement statistics through Area 4 for the 1984 commercial salmon fishery. S.W indicates the statistical week. (GN=gillnet, SN=seine).Source D.F.O, 1985. Week Ending S.W Gear Catch Days f ished CPUE Escape. H.R 3191 Jul 15 8 290 GN 687 1 2.4 5484 '. 1 25 Jul 21 9 649 GN 7021 4.3 10.8 5750 .619 180 SN 2340 (1.3) 13.0 Jul 28 1 0 591 GN 4362 3 7.4 3324 .735 204 SN 4881 (3) 23.9 Aug 4 1 1 504 GN 4269 3.3) 8.5 4838 .468 Aug 1 1 1 2 297 GN 2854 3 9.6 1 0090 .221 Aug 18 1 3 252 GN 431 9 5 17.0 5433 .443 Aug 25 1 4 80 GN 367 2 4.6 3880 .109 35 SN 105 (1 ) 3.0 Sep 1 1 5 55 GN 167 1 3.0 — 0 TOTAL 31 372 22.9 42989 x=.422 Table 18. Ocean age composition by sex for steelhead taken in the 1984 Skeena River commercial salmon f i shery (assembled from appendix table 7). Week Ocean age 9 1 0 1 1 1 2 1 3 14 . 1 + %M .278 .221 .333 .464 .285 .222 %F . 1 25 . 1 40 .282 . 1 58 .171 . 106 .2+ %M .544 .573 .560 .375 .457 .587 %F .708 .640 .696 .631 .800 .723 .3 + %M . 1 52 .208 . 1 06 .161 .257 .190 %F . 1 67 .220 .021 .210 .028 . 170 weekly samples generally increased through the fishery (Table 19), which suggests a general shift in size brought about by stock specific run-timing differences. Classification of the 1984 commercial fishery steelhead 80 TABLE 19. Mean lengths and weights of steelhead sampled in the 1984 commercial salmon fishery. SW indicates statistical week Week SW n Length S Weight S Ending (cm) (kg) Jul 21 9 1 32 69.4 8.2 3.7 1 .3 Jul 28 10 131 72.2 10.5 4.4 1 .9 Aug 4 1 1 127 72.3 9.2 4.4 1 .9 Aug 1 1 1 2 1 22 72.3 8.3 4.4 1 .9 Aug 18 1 3 1 33 74.3 9.5 4.8 1 .9 Aug 25 1 4 108 73.4 8.2 4.7 1 .7 samples to stock of origin utilized four of the twelve discriminant analysis models previously oultlined. In order, these were classification of adults by smolt age 3/pooled ocean age/scale variables alone (Model A), smolt age 4/pooled ocean age/scale variables alone (Model B), pooled smolt age/pooled ocean age/scale variables alone (Model C), and model C using size related data in addition to the scale variables (Model D). Classification of the commercial fishery samples by specific age class (eg 3.2+, 4.2+ etc) was not considered feasible because of the low age class specific sample sizes present in the data base. For all analyses, the five stocks under study were assumed to occur in the weekly samples in proportion to their relative overall abundance in the fishery. Table 20 summarizes the results of classifying the 1984 commercial fishery steelhead samples to stock of origin by the above four models. Temporal differences in the point estimates for all four models were found with some stocks estimated to be present in large proportions throughout the sample period. Morice River steelhead were estimated to be present in large Table 20. Classification results to stock of origin by week for steelhead sampled from Area 4 in 1984. Model A: smolt age 3/pooled ocean age/scale variables alone. Model B: smolt age 4/pooled ocean age/scale variables alone. Model C: pooled smolt age/pooled ocean age/scale variables alone. Model D: pooled smolt age/pooled ocean age/scale variables +length and weight (+/- 95% confidence limits about the estimated variances in brackets). Week Proportional Estimated Stock Composition Kispiox Zymoetz Sustut Babine Morice 9 1 0 1 1 12 13 1 4 A 065( . 181 ) 0 ( .253) .454( .091 ) .0261 .092) .454( .200 B 1 49( . 430) . 1 82 ( .440) .298( .097) . 1 17( .086) .254( .232 C 1 06( . 1 50) .038( .251 ) .439( .087) .061 ( .042) . 356( .116 D 0 ( .027) 0 ( .065) .303( .065) .099( .032) .599( .072 A 0 ( .091 ) .057( .251 ) .701 ( .091 ) .081 ( .094) . 1 61 ( . 1 73 B 255( . 424) 0 ( .368) • 395( .121) . 1 1 6( .086) .232( .112 C 091 ( .161) .068( . 1 73) • 496( .094) .0991 -.07 5) . 244 ( . 1 04 D 0 ( .056) 0 ( .072) . 4 1 2 ( .065) . 168( .056) .4 1 9( .072 A 1 05( . 181 ) 0 ( .262) • 376( .092) .223( .108) .294( . 1 66 B 0 ( .775) . 525( .798) .325( .112) 0 < .030) . 1 50( . 1 34 C 1 49( . 200) .087( .141) .425( .091 ) . 1 42 ( .075) . 1 96( . 1 00 D 052( .181) .032( .086) .378( .065) .093( .056) .443( .077 A 0 ( . 1 92) 0 ( .462) .428( .073) .307( .149) .264( .515 B 1 93< .665) .3221 .711) .290( . 1 26) . 1 93( . 1 34) 0 ( .112 C 1 38 ( .092) • 089( .313) • 422( .093) . 1 87 ( . 100) . 1 63 < .094 D 0 ( .072) .262( .086) .336( .072) .287( .072) . 1 1 5( .072 A 1 44 ( . 181 ) 0 ( .274) .490( .092) . 1 63( .079) .202( . 175 B 1 72 ( .660) .241 ( .634) .448( .120) 0 ( . 1 03) . 1 37 ( . 1 42 C 1 72! . 101 ) • 045( .113) . 443 ( .075) .2031 .095) . 1 35( .072 D • 144 1 .065) .071 ( .086) .366( .065) .21 4( .065) .204( .056 A 0 ( . 1 66) . 1 0 1 < .214) .681 ( .092) 0 ( .079) . 2 1 8 ( . 1 75 B 0 < .590) . 2 1 6 ( .634) .486( .117) .054( .103) .243( .141 C 0 ( .101) .083( .094) .639( .081 ) . 1 35( .079) . 1 67( .101 D 0 ( .046) . 1 20 ( .086) .629( .072) .083< .046) . 1 29( .065 82 proportions during the early weeks 9 through 10. Babine River steelhead were estimated to be present in large proportions during the later weeks 12 through 14. Both Kispiox River and Zymoetz River steelhead were estimated to be present in variable proportions during each week 11 through 13 depending upon the model. In several instances the point estimates for these stocks were negative. However, the confidence intervals about the estimates indicated that both the Kispiox and Zymotez stocks may have been present in small numbers. For all four models, the confidence limits were wide and varied about each point estimate in proportion to the classification success of the original discriminant analyses. Confidence in the point estimates for Babine, Sustut, and Morice-Bulkley River steelhead was notably greater than for Zymoetz and Kispiox River steelhead. The point estimates in table 20 were used to calculate the probable run-timing curves of each steelhead stock through the fishery in 1984. The diverse age class structure of the 1984 steelhead catch suggested that all four classification models be used to generate specific run-timing curves. This allowed for between model comparison and a more detailed run-timing analysis. Calculating the 1984 run-timing curves first required data from external sources. The calculation of total steelhead population size during each week of the fishery was calculated by adding Department of Fisheries and Oceans weekly steelhead catch estimates to the weekly estimated steelhead escapement 83 past the test fishery (summary, next table). The general run-Statistical Week 8 9 1011 12 13 14 15 Total Harvest 687 9361 9243 4269 2854 4319 472 167 31372 Index 44.7 26.6 15.4 22.4 46.7 25.1 18.0 - 198.9 Escapement 5484 5750 3324 4838 10090 5433 3880 - 42989 Run Size 10662 15111 12567 9107 12944 9752 4352 167 74361 timing curves were calculated by multiplying the estimated weekly point estimates for a given classification model (table 20) by the estimated total steelhead population size for a given week. Adjustments were made for the two smolt age specific classification models (A and B); here, the estimates of total weekly run size were set to reflect the age class compositions of the weekly samples. In week 9, for example, 53.5% of the steelhead sample was comprised of adults of smolt age 3. Thus 53.5% of the total steelhead harvest + escapement in week 9 was assumed to be comprised of smolt age 3 adults. Similar adjustments were made by week for each model. Appendix table 8 summarizes the results of applying the predicted weekly stock composition estimates from the four classification models (table 20) to the estimated weekly harvest + escapement run sizes in the 1984 commercial fishery. Figures 24 to 31 summarize the estimated run-timings from appendix table 8. In general, Morice River and Sustut River were found to 84 predominate in the early weeks of sampling while the other three stocks tended to predominate during the later weeks. The "best" run-timing model is difficult to identify, especially considering . the limitations imposed by only one year of commercial data and the acknowledgement that 1984 was a unique year for steelhead returns. In addition, the effect of "other" steelhead stocks in the weekly samples not considered in this study remain unknown. The specific smolt age models ( A and B) reduce the within stock scale pattern variances but may suffer from reduced sample sizes. The pooled smolt age analysis (C) likely increase the within stock scale pattern variances but has the advantage of utilizing much of the data base. The same model with the inclusion of size data (D) has the same advantages of (C) plus the added benefit of utilizing stock specific sizes at age. Model D may suffer, however, if the fishery selects for size or if sizes at age should change appreciably between years for a given stock. Juvenile Analysis Riddell and Leggett (1981) provided evidence that morphological variations between juvenile atlantic salmon from various stocks have an adaptive basis and is highly stock specific. The results of comparing Zymoetz River, Kispiox River, and Morice-Bulkley River juvenile steelhead parr suggests a similar phenomenon in Skeena River steelhead. In a univariate analysis of variance significant differences between means for seven of the ten juvenile morphological body measurements used 85 Figure 24. Predicted run-timing. Estimated run-timing composition through the 1984 Skeena River commercial salmon fishery for adult steelhead of smolt age 3/pooled ocean age using scale features alone. Figure 25. Predicted run-timing. Normalized estimated run-timing composition through the 1984 Skeena River commercial salmon fishery for adult steelhead of smolt age 3/pooled ocean age using scale features alone. Week Ending UV84) 87 Figure 26. Predicted run-timing. Estimated run-timing composition through the 1984 Skeena River commercial salmon fishery for adult steelhead of smolt age 4/pooled ocean age using scale features alone. Figure 27. Predicted run-timing. Normalized estimated run-timing composition through the 1984 Skeena River commercial salmon fishery for adult steelhead of smolt age 4/pooled ocean age using scale features alone. 88 XI00 30-N 23-u m b e r 20-s 10-Key l=Sustut 2KMori ce 3=L<abine 4«Ki spi ox 5=Zymoetz 123456!123436!123436112345611234361123456!1234361 I I I I B 7 10 11 1 July 21 August ,4 I I 13 14 August 18 I IS Week Ending (1VB4) X 100 !1234561123436112343611234361123436112343611234361 ! I I I I I ! I B V 10 11 12 13 14 13 July 21 August 4 August 18 Week Ending <1VB4> 89 Figure 28. Predicted run-timing. Estimated run-timing composition through the 1984 Skeena River commercial salmon fishery for adult steelhead of pooled smolt age/pooled ocean age using scale features alone. Figure 29. Predicted run-timing. Normalized estimated run-timing composition through the 1984 Skeena River commercial salmon fishery for adult steelhead of pooled smolt age/ pooled ocean age using scale features alone. Weols Ending (19131) 91 Figure 30. Predicted run-timing. Estimated run-timing composition through the 1984 Skeena River commercial salmon fishery for adult steelhead of pooled smolt age/pooled ocean age using scale features in conjunction with length and weight. Figure 31. Predicted run-timing. Normalized estimated run-timing composition through the 1984 Skeena River commercial salmon fishery for adult steelhead of pooled smolt age/pooled ocean age using scale features in conjunction with length and weight. Weel; Ending <19B4) Week Ending U9B4) 93 in this study were found, especially for caudal peduncle width and caudal peduncle depth (Table 21). Table 21. Adjusted geometric means (+/- one S.D) and the results of one way analyses of variance for differences in body morphology between Kispiox, Zymoetz, and Morice-Bulkley River steelhead parr. All measurements are in cm. (* indicates no significant difference at the 5% level of signi f icance) grand mean standard length= 10.29cm (DF=2,88) Variable Zymoetz Morice Kispiox F P n = 29 n= 29 n = : 30 HL 3. 00 (0. 21 ) 3.07 (0. 13) 3.06 (0. 13) 0. 08 *>0. 05 HD 2. 01 (0. 24) 2.01 (0. 88) 2.11 (0. 12) 3. 1 7 <0. 05 HW 1 . 49 (0. 29) 1.41 (0. 06) 1 .59 (0. 12) 6. 1 2 <0. 05 CD 1 . 1 1 (0. 07) 1 .04 (0. 04) 1.17 (0. 06) 12. 98 <0. 05 CW 0. 48 (0. 07) 0.43 (0. 04) 0.56 (0. 07) 14. 98 <0. 05 BD 2. 84 (0. 18) 2.70 (0. 13) 2.91 (0. 22) 7. 41 <0. 05 BW 1 . 55 (0. 17) 1 .40 (0. 07) 1 .55 (0. 14) 9. 29 <0. 05 PrDL 5. 91 (0. 20) 5.95 (0. 18) 5.92 (0. 14) 0. 42 *>0. 05 PoDL 6. 1 4 (0. 13) 6.10 (0. 19) 6.14 (0. 16) 0. 40 *>0. 05 The results of three stock morphological discriminant analysis are shown in figure 32. Multivariate analysis revealed significant differences between the stock centroids (approximate F=8.13 DF= 10 162 P<0.001) and in all pairwise comparisons between stocks. Four of the original ten variables were selected as best describing the between stock juvenile differences; these were, in order of entry, head width, caudal peduncle depth, caudal peduncle width, and body width. The first discriminant function primarily separated Kispiox River juveniles from the other stocks on the basis of caudal peduncle width and caudal peduncle depth. Kispiox juveniles had large mean values for these features and were generally more "robust" 94 Figure 32. Discriminant function analysis describing morphological variation among juvenile steelhead from the Kispiox,Zymoetz,and Morice-Bulkley Rivers. Each of the letters indicates the stock centroids Z=Zymoetz K=Kispiox M=Morice *=grand centroid, the open circles indicate the 90% confidence interval about each centroid (from Pimental, 1979), and the lines point to the next most similar stock in discriminant space. The two standardized discriminant functions are given below. D1 = 1.60HW +8.28CD +8.80CW -1.23BW -8.28 D2 = 0.03HW +0.84CD +1.47CW -1.68BW -14.25 + 1.5-+1. 0--1.5-! -2.0 -2.0 -1.5 -1.0 -0.5 0 +0.5 +1.0 +1.5 FUNCTION 1 96 in body shape at a given length. The second discriminant function separated the Morice River juveniles from the other two stocks primarily on the basis of body width. Morice juveniles had low mean body widths, and were generally less "robust" in overall body shape. 61.4% of the juveniles from the three stocks were correctly classified to stock of origin using Lachenbruch's (1975) holdout classification procedure (range Zymoetz 48.3%- Kispiox 70.0%). Misclassifications for Zymoetz River juveniles were evenly divided between the other two stocks, a result similar to the findings of adult classification by scale pattern features: These results suggest that the % Predicted Stock Actual Stock correct K M Z K n = 30 70.0 21 3 6 M n = 29 65.5 4 19 6 Z n = 29 48.3 7 8 1 4 x= 61.4 observable differences in juvenile body form for Skeena River steelhead are quite different. While extensions of such body form analysis to the adults from each stock were not made, it is likely that similar shape differences could be found. Observed adult body proportions have been noted to vary widely between the adults from several Skeena River steelhead stocks (M. Lough, pers. Comm. 1985) and may provide additional information for stock separation purposes. 97 DISCUSSION Biological Considerations The primary objective of this study was to test the racial separability of Skeena River steelhead by scale pattern analysis. Significant differences in scale growth, age composition, sizes at age, and juvenile body morphology exist between steelhead from five of the major Skeena River tributaries (Morice-Bulkley, Kispiox, Zymoetz, Babine, Sustut). Run-timing differences for each stock are also evident in incidental catches from the commercial salmon fishery. This variability confirms the subdivision of Skeena River steelhead into discrete stocks and suggests that stock discreteness is an adaptive property of the species that has arisen through natural selection. The scale pattern technique for differentiating Skeena River steelhead works better for some stocks (Sustut, Babine, Morice-Bulkley) than others (Kispiox, Zymotez). The success of the technique depends upon the observed levels of within stock compared to between stock variance. This, in turn, depends upon stock definition and the variables chosen for analysis. The diverse age class structure of Skeena River steelhead makes the construction of discrimination models difficult. The use of age specific models, which are most commonly used in stock discrimination studies, is quite restricted for this species. The mean classification success for the classification models used in this study was not high (50% to 65%: range Zymoetz 29%-98 50%- Sustut 55%-75%) but substantially better than random allocation (20%) and acceptable considering the large number of stocks (5) involved. A certain level of freshwater scale pattern "similarity" exists between all Skeena River steelhead. This may reflect a common response by all stocks to several dominant abiotic features of the Skeena River drainage (yearly freeze up, peak flows, low temperatures etc). Environmental variation contributes to within stock scale pattern variability in Skeena River steelhead and determines the success of stock separation. Sustut River steelhead, which occupy the upper regions of the Skeena River drainage, are typified by older ages at smolting, wide freshwater scale zones (= large smolt sizes at age), large adult sizes at age, and older ocean ages at maturity. Babine River steelhead, which also occupy the upper Skeena River region, are typified by intermediate ages at smolting, large freshwater scale zones (= large smolt sizes at age), intermediate to large adult sizes at age, and intermediate ocean ages at maturity. Morice-Bulkley River steelhead, which occupy the "inland" regions of the Skeena River drainage, are typified by older ages at smolting, small freshwater scale zones (= small smolt sizes at age), small adult sizes at age, and younger ocean ages at maturity. Kispiox River steelhead occupy the mid-river areas of the Skeena River drainage and are typified by older ages at smolting, intermediate freshwater scale zones (= intermediate smolt sizes at age), notably larger sizes at age, and older ocean ages at maturity. Zymotez River steelhead occupy the lower regions of 99 the Skeena River drainage and are typified by intermediate to older ages at smolting, intermediate freshwater scale zones (= intermediate smolt sizes at age), intermediate to large sizes at age, and intermediate ocean ages at maturity. Juveniles from three of the stocks (Kispiox, Zymoetz, Morice-Bulkley) display significant between stock morphological variability. Kispiox River juveniles are notably more "robust" than the more fusiform juveniles of the Morice-Bulkley River. Stock discreteness within a species depends upon the level of interaction between ecological and genetic processes in "stochastic" environments (Maclean and Evans, 1981). Various authors have suggested that site specific homing in fishes provides the potential for genotypic and phenotypic adaptation to such environments (Larkin 1972, Ricker 1972,). Parkinson (1984b) showed that genetic variation exists between steelhead populations in geographically adjacent streams in British Columbia. He concluded that "this species is subdivided into a large number of semi-isolated populations each having the genetic potential to evolve adaptations to local environments". While not all observable differences between stocks are necessarily adaptive, many may have a strong selective basis. My results suggest that this is the case for the observed patterns of variation in Skeena River steelhead. Stock discreteness by discriminant analysis depends not only upon significant differences between stock centroids but also upon the level of individual variance about each stock centroid (centroid dispersion). Sustut, Babine, and Morice-Bulkley River 100 exhibit greater separability in discrimination models beacuse they exhibit lowered levels of centroid dispersion. Conversely, Kispiox and Zymoetz River steelhead exhibit lower separability in discrimination models because they exhibit increased levels of centroid dispersion. Assuming that the learning samples used in this study are representative of each stock, then the dispersive homogeneity of some stocks could represent the presence of dominant selective forces. Steelhead from the Sustut and Babine Rivers, for example, could exhibit large freshwater scale zones (= large smolt sizes at age) and larger adult sizes at age because of hydrodynamic selection for larger size. The upper Skeena River region is turbulent and larger body size would enhance both adult and juvenile upstream/downstream migration. Hydrodynamic selection has been suggested by several authors as a potentially strong selective force in salmonids (Schaffer and Elson 1975, Thorpe and Mitchell 1981). Schaffer and Elson (1975) concluded that the larger sizes and older ages of upriver Atlantic salmon from the Miramichi River in New Brunswick are adaptations to meet the energetic costs of sustained swimming in greater flows during long and difficult upriver migration. Sustained swimming seems to have a strong genetic component. Tsuyuki and Williscroft (1977) found the swimming endurance of "upstream" Fraser River steelhead juveniles (Thompson River) to be significantly greater than the swimming endurance of "downstream" Fraser River juveniles (Chilliwack River) in treadmill type tests. They attributed the differences to greater levels of the LDH-A allele 101 in the Thompson River stock which increases the threshold of muscular fatigue and thus extends sustained swimming ability. Steelhead smolt sizes increase with ascending distance upstream in to the Skeena River drainage, which confirms my findings by scale pattern analysis. Narver (1969), Whatley (1977), and Whately et al (1978), reported, that the mean back-calculated lengths for age 3 smolts from the Morice-Bulkley, Kispiox, and Babine Rivers were 145mm, 163mm, and 187 respectively. The mean back-calculated lengths for age 4 smolts from the same three rivers were 178mm, 195mm, and 203mm respectively. Both genetic and environmental factors control smolt sizes at age in salmonids (Ricker, 1972). Although larger parents generally produce larger eggs and thus larger fry, the eventual sizes at smolting depend upon yearly growth rate and therefore food availability. McBride and Marshall (1983), in a study of Yukon River chinook salmon stocks by scale patterns, found that upriver stocks had larger adult sizes at age yet exhibited smaller freshwater scale zones (= small smolt sizes at age) than the lower Yukon river stocks. They attributed the smaller upriver scale zones to lower productivity in the upper Yukon River area. This contrasts my findings and suggests that food productivity in the upper Skeena River is sufficiently high enough to produce large smolts at age. Babine River steelhead may additionally benefit from sockeye salmon enhancement, although little information is available. Different ' selective forces may explain the observed features of scale pattern and life history variation in Morice-1 02 Bulkley River steelhead. Although the Morice-Bulkley River is the largest of the five Skeena River tributaries its flows are rather uniform over long, low gradient distances. Whately et al. (1978) attributed the small sizes and older ages of Morice-Bulkley River steelhead smolts to low instream productivity. The smaller adult sizes at age and younger ages at maturity of Morice-Bulkley River steelhead suggests that strong ecological selection for rapid adult maturation may exist. Rapid adult maturation would ensure maximal fry seeding (and parr to smolt production) in less productive environments on a yearly basis by minimizing the time between year class spawnings. Older ages at maturity would extend the time between year class spawnings and thus increase the biological risk of poor parr production in less favorable years. The early predicted run-timing of Morice-Bulkley River steelhead through the 1984 commercial fishery supports the notion of "rapid maturation" in this stock; however, early run-timing is probably better related to the long distances inland Morice-Bulkley River steelhead must travel. Sustut River steelhead were also predicted to pass through the 1984 commercial fishery quite early, which makes such a hypothesis tenable. It is possible that small sizes at age and young ages at first spawning in Morice-Bulkley River steelhead represents a cumulative genetic effect from commercial fishing. Ricker (1981) documents the decreasing sizes at age and ages at maturity for many Pacific salmon stocks and attributes the trend to size selection for older and thus more mature individuals in 103 commercial fisheries. However, the size composition of Morice-Bulkley River steelhead has remained rather constant over time, as shown by the homogenous length frequencies of steelhead passing Moricetown rapids from 1961-1967 (Harding and Buxton, 1971) and from the 1976-1977 data used in this study. While not conclusive, this evidence suggests that commercial fishery effects may be less important than ecological forces in determining the sizes at age and ages at first spawning of Morice-Bulkley River steelhead. Steelhead from the Kispiox and Zymoetz Rivers show a high degree of freshwater scale pattern overlap. This suggests that environmental growth regimes in the two sytems are somewhat similar. Both stocks inhabit "coastal" type rivers although the Zymoetz River is considerably larger and may exhibit a wider range of environments. Stock separation by discriminant analysis increases between the two only when adult sizes at age (length and weight) are introduced, which, being substantially greater in the Kispiox stock, implies either genetic differences in ocean growth rates and/or differences in ocean migration and feeding patterns. This naturally leads to the potential for discriminating the stocks on the basis of scale pattern ocean growth. However, first year ocean growth differences between the two were not that pronounced even for the age specific models developed in this study (3.2+, 4.2+). Scale growth after the first ocean year was not examined and could lead to differences for separating the two stocks. No definitive reasons for the similarity of freshwater scale patterns in 104 Kispiox River and Zymoetz River steelhead seem obvious. The Kispiox River, being glacial in its headwaters, is fed by many lakes, bogs, and creeks situated in a series of low hills and benches which provide moderate flows and high water quality (Whately, 1977). The Zymoetz River has a somewhat similar morphology except on a larger scale. It should be noted that steelhead from the Zymoetz River are proximally close to the multivariate grand centroid for all stocks which thus supports the notion of environmental heterogeneity for this system. The results of juvenile analysis bear further comment. Kispiox River juveniles are quite "robust", exhibiting deep heads, deep bodies, and "thick" caudal peduncles. In contrast, Morice-Bulkley River juveniles are quite "fusiform", exhibiting smaller heads, slender bodies, and "thinner" caudal peduncles. Zymoetz river juveniles demonstrate a broad cross-section of both body types. Body shape in salmonids, especially juveniles, has been shown to have a genetic basis and may be highly adaptive (Riddell et al., 1981). Stream habitat (substrate, flows, space, poolrriffle ratios, cover, etc) is extremely important for juvenile salmonid biology (Northcote, 1969). In general, those streams with higher flow velocity and longer migration routes may select for a more fusiform body shape in the juveniles to reduce drag and maximize sustained swimming ability (Taylor, 1984). Relating to this study, the concentration of older steelhead parr in the Morice-Bulkley River is heaviest in the lower reaches (Tredger, 1984), apparently because of limited upstream productive capacity. 105 Here, the parr are subject to higher flow velocities and less microhabitat "refuges" compared to Kispiox River parr which rear throughout the drainage. Kispiox River juveniles exhibit the typical "coastal" (Taylor, 1984) body type where hydrodynamic selection for sustained swimming abilty may be less important. Zymoetz River juveniles exhibit both body types which supports the notion of growth in a wide range of habitats. Body form differences may extend to the adults from each Skeena River stock and could provide additional information for stock separation purposes. Theoretical Considerations Errors in data interpretation, assumed representativeness of the data, and the assumptions of discriminant analysis are all of concern for the present study. Firstly, data interpretation was based on established methods. Any misinterpretation by the author is homogenous across all samples used for discrimination and classification in this study. Secondly, representativeness of the data was limited by the availability of learning scale samples. Ideally, discrimination should be achieved using fish from the same brood year and of the same age from each stock. This would limit any variability attributable to differences in age and yearly differences in growth. However, the diverse age class structure of Skeena River steelhead precludes any simple age specific discrimination approach except for the dominant age classes (3.2+, 4.2+). Even then, I would question the utility of age specific analyses for 1 06 Skeena River steelhead. The patterns of freshwater scale growth found in this study appear to indicate that steelhead of different total ages but of the same smolt age from each stock have similar patterns of freshwater scale growth. This argues against the necessity of age specific models. However, the potential effects of differential freshwater scale growth by brood year on the results of this study are harder to quantify. Based on limited evidence, it appears that freshwater scale growth is relatively stable between brood years for a given stock. Significant differences in scale growth between yearly samples (and thus brood years) for several of the stocks used in this study were not evident. This supports the use of different brood years for constructing stock specific learning samples. Further clarification of this point is needed, especially with regard to differential density effects on scale growth. Thirdly, it is possible that violation of the assumptions necessary for linear discriminant function analysis could affect the discrimination and classification models developed in the study. Each "stock" should be discrete and definable. This requirement appears to have been met, although substock structure and its effect on discrimination success was not investigated. Straying between stocks is assumed to be minimal, which should maintain group identity for discrimination purposes. The assumption of multivariate normality for the discriminating variables used in the study could have been violated because tests for multivariate normality were 1 07 unavailable. Multivariate normality is especially important for linear discriminant analysis because of the nature of the decision surfaces used to separate groups. In linear analysis, these surfaces are actually linear classification boundaries that best separate ellipsoidal (multivariate normal) hyperspheres. If the multivariate density distributions are not normal, then the distribution contours of each group can randomly "overlap" the decision surface and result in reduced classification success. I relied on univariate frequency comparisons for each stock to estimate multivariate data normality. This does not guarantee that the distributions are multivariate normal (Pimental, 1979). The assumption of homogenous variance-covariance structure between stocks was not rigorously tested in this study. Stock specific variance-covariance matrices describe the patterns of spread and linear variable association within groups on a multivariate basis (ie. Variables should show the same patterns of association for each stock). The effects of dispersive inequality on canonical axes and discrimination functions is not well known (Pimental, 1979). Gilbert (1969) notes that linear discriminant analysis is still valid for classification purposes even when the hypothesis of dispersive equality is rejected. Apparently, inequality of dispersions has no real effect on multivariate analysis of variance type I or type II errors if the sample sizes are large and of equal size (Pimental, 1979). In other words, the test of centroid equality by MANOVA is powerful enough to result in rejection even when slight 108 departures from dispersive equality are apparent. It is possible that the use of non-parametric discriminant analyses (eg quadratic analysis, Cook and Lord, 1978), which make no assumptions regarding underlying density distributions or dispersive relationships within and between groups, could have provided better results. However, quadratic analysis is primarily useful when there are significant differences between the variances of the variables used in the analysis. This did not seem to be the case for this study. The choice of which variables best separate the stocks in this study could also be subject to error. In common with the majority of discriminant analysis studies using large variable systems, I chose to use stepwise variable selection procedures. Johnson and Wichern (1982) note the problems of using stepwise variable selection techniques for constructing discriminant functions. There is no guarantee that the subset selected is "best". In fact, although discriminant analysis relies on variables that show some degree of intercorrelation (Pimental, 1979), large intercorrelations between linear combinations of variables will magnify the "the problems associated with variable selection procedures" (Johnson and Wichern, 1982). This aspect was not fully investigated. Commercial Fishery Considerations The second objective of this study was to assess the potential of scale pattern analysis for identifying Skeena River steelhead stocks caught in the commercial salmon fishery. As 109 previously noted, all five major stocks were separable in the 1984 fishery within varying bounds of confidence. Although age composition differences between the stocks are pronounced in the learning samples, no distinct stock specific patterns of age composition through the commercial fishery was evident in 1984. This stems from the composite run-timing nature of Skeena River steelhead. Although based on limited evidence, it may not be possible to use age composition data for catch allocation. In general, the four-model five stock classification analyses for 1984 predicted the early run-timing and numerical dominance of Sustut River and Morice-Bulkley River steelhead through the fishery. The same models predicted the later run-timings and less abundant dominance of Babine, Zymoetz, and Kispiox River steelhead through the fishery. The exception was for the smolt age 4/pooled ocean age/scale variable only analysis. Here, both Babine and Kispiox stocks were predicted to be prominent during the early parts of the fishery. This may reflect differential time at return for steelhead of different smolt ages or error in the analysis because of reduced sample sizes. The same trend was not seen in the pooled smolt age classification analysis. Further study is required to clarify this point. The weekly point estimates of stock abundance in 1984 are sufficiently variable enough to result in considerable temporal fluctuation for the run-timing estimates. The assumption of normalized run-timing may or may not be practical because of this. However, based on the long term patterns (normal) of 1 10 steelhead return and escapement to the Skeena River, I believe that normalized run-timing is a valid assumption for this study. All four classification analyses for 1984 resulted in several negative point estimate values for some of the stocks (eg Zymoetz, Kispiox, table 20). However, the 90% confidence intervals associated with these estimates usually included an upper positive limit. It seems unlikely that those stocks with negative point estimates were not actually present in the fishery during the sample period. Rather, the negative estimates reflect the difficulty in estimating contribution rates for stocks in low abundance by scale pattern analysis when learning sample classification success is low. Scale pattern analysis predicted the largest component of the 1984 fishery to be the Sustut River stock. This is somewhat surprising as population levels in this sytem are not believed to be high. This either suggests that previous population estimates are in error or that other stocks with scale patterns similar to the Sustut River stock but not considered for analysis were present in the fishery samples. Both possibilities need investigation. Steelhead production in the upper Skeena River region is not well defined. In addition, several downstream "stocks" (Lakelse, Kitsumkalum, Suskwa, Kitwanga) could also have scale patterns similar to the Sustut system. Modification of the method may be necessary as further information becomes available. Although size (length and weight) is a good stock discriminator for Skeena River steelhead, its use for commercial 111 fishery classification must be done with caution. Any size selectivity by the commercial fishery will bias the estimates of stock abundance in the fishery samples used for classification purposes. Scale pattern analysis itself is not affected by potential size selectivity as scale features (freshwater) in Skeena River steelhead appear to be independent of eventual adult age (and thus size). All four classification models developed in this study should be used to classify commercial fishery steelhead interceptions until variability in the technique is clearly established. Stock specific run-timing has been previously noted for both Skeena River sockeye and pink salmon (Aro and McDonald 1968, Larkin and McDonald 1968, McDonald 1981). Temporal shifts in stock specific run-timing for these species appears to be slight between years (Larkin and McDonald, 1968) although some variability is present. For Skeena River steelhead the effects Of differential brood year success and stock abundance on the applicability of the scale pattern technique is of concern. Stock abundance will fluctuate between years according to the numerical returns by brood year to each stock for each contributing age class; if the returns to a given stock happen to be low (high) in a given year because of a series of poor (good) brood year successes, then fewer (more) fish from that stock will be present in the fishery and available for classification. Assuming that each stock is sampled according to its proportional abundance and that the sampling design is adequate, then the technique of scale patterns should respond to 1 1 2 such fluctuations. However, at the present stage of development, the technique cannot distinguish between actual shifts in the predicted run-timing curve and/or simply changing abundance. For example, stock A which comprises 50% of the catch in week 1 in year 1 may have a predicted abundance in week 1 of year 2 of 20%. Either less fish from that stock are available for capture in year 2 ( different abundance, same run-timing) or the run-timing curve has shifted earlier or later (same abundance, different run-timing), or both. For the most part, I have assumed the former although further investigation is clearly required. Another aspect affecting the utility of the scale pattern technique is its overall accuracy. Discrimination success is variable enough to result in wide confidence limits ,for some of the point estimates of commercial fishery stock contribution (eg Zymotez). This reflects the level of scale pattern overlap between the stocks and cannot be modified. To increase stock discriminance and classification success, the possibility of utilizing other multivariate features in conjunction with scale patterns should be pursued." These include body morphology, meristics, parasites, gene frequencies etc. The inclusion of such character systems must be weighed against their increased difficulty of collection; however, once established, they could provide valuable information for stock separation purposes. Fournier et al. (1983) have used such an approach for distinguishing chum salmon stocks with favorable results. 1 13 Gear Selectivity Ricker (1981) notes that the mode of selection on incidental salmonid species caught in net fisheries for sockeye salmon depends upon their size. For example, chinook salmon taken incidentally are often smaller than their average size in the run at that time while pink salmon taken incidentally tend to be larger than their average size in the run at that time. This results in considerable size differences between those fish, caught and those fish which escape the commercial fishery to spawn. Over time, strong genetic selection by size is possible. The degree of similar response for Skeena River steelhead is difficult to establish although some selection for smaller sizes and younger ages at maturity no doubt exists. Generally, the gillnet fishery selects for larger four year old male sockeye salmon (2-3 kg) and larger female five year old sockeye salmon (3 kg) (L. Janz, pers. Comm., 1985). Any selective effects on steelhead by size may be somewhat reduced by the extreme levels of fishing effort in the Skeena River estuary. Oguss and Andrews (1977) found that mesh sizes have no significant effect on the numerical size of the incidental Skeena River steelhead catch. Although behavioral differences between the stocks may change their susceptability to an unknown extent (depth of swimmimng, proximity to shore etc.) steelhead caught in the 1984 fishery were more often "tangled" than gilled, regardless of size (interview data). In addition, the dense nature of gillnetting may reduce the chances of any given steelhead successfully migrating past the fishery. 1 1 4 This argues against any specific size selective effects of commercial fishing. Reductions in overall stock specific esapement may be more important. Applications to Steelhead Management A major management objective for Skeena River steelhead is to minimize the potential impacts of stock specific incidental harvesting during the commercial salmon fishery. My results provide a method for identifying which stocks are present in the fishery and thus provide the potential for structuring stock specific management objectives. However, the fishery is extremely dynamic and is regulated by complex socio-economic factors. Short of resorting to a wier system or drastically reducing the size of the commercial fleet, the problem of incidental steelhead catches in the fishery is not easily solved. The principle concern for steelhead is adverse harvest rate pressured Mean weekly percent harvest rates on steelhead appear to increase dramatically if continuous fishing is extended beyond three days per week (BCF Branch, unpublished data, 1983). In addition, the mean percent weekly harvest rate for sockeye is higher than that for steelhead in a three day per week or less fishery while it becomes lower in a four day per week or more fishery. Presumably, this relates to the fact that steelhead move into the fishery area daily whereas sockeye salmon tend to pool and can be harvested quite quickly. The problem of increasing weekly harvest rate pressure on steelhead is most prominent during peak sockeye salmon run-timing where 1 1 5 fishing may actually continue for five days per week (eg Monday, Tuesday, Wednesday, Saturday, Sunday). What is the best commercial fishing strategy that would reduce commercial harvests on steelhead? Firstly, my results suggest that peak stock specific run-timing, while composite, is somewhat compressed within a short period of time. How "short" will depend upon the estimates of variability obtained for future analyses. Any management alternatives for reducing incidental catches should focus on maximizing escapement during run-timing peaks. Three techniques are apparent. The first is to stop all fishing during the estimated peak run-timings for each stock. Logistically, such an approach is not feasible. The second is to make use of more fishery closures or "windows" on a weekly basis. This would ensure that portions of run-timing peaks escape the fishery rather than risk entire cohort removal during long fishery openings. Presently, fishing occurs 24 hours a day during any given continuous opening (two, three, four days etc). As an example of window use, three or four days of fishing interspersed by two days of closure (windows) may be more beneficial to steelhead than three or four days of continuous fishing followed by two days of closure (the present practice). This assumes, of course, that steelhead do in fact migrate through the fishery area quite quickly. An intensive tagging study of steelhead through the commercial fishery area would help to clarify the latter point. One potential problem of interspersed windows is 1 1 6 potentially apparent during the presence of the seine fleet, which is restricted to the outer regions of the fishery area. During periods of intense seiner activity (peak sockeye salmon abundance) seiners remove steelhead that normally would be caught a few days later at the river mouth had the seiners not been present. Window closures during such periods may do more harm than good by allowing fishing pressure time to build; those steelhead managing to pass the outer seine fleet negotiate the fishing area during the closure and are taken anyway by gillnetters when fishing reopens a few days later. Under such circumstances, overall steelhead escapement may be greater using a normal pattern of longer fishery openings. The third technique is to simply reduce fishing effort from 24 to 12 hours per day. This would create "nightly" windows and would not restrict the movements of the commercial salmon fleet to the same extent as full daily closures. Thus, sockeye fishing could occur for four or five days continuously while peaks of steelhead stock abundance would still be able to escape through the fishery (assuming nightly movements do, in fact, occur). In summary, I believe' that the technique of scale patterns is feasible for the identification of Skeena River steelhead in the commercial salmon fishery. The technique provides a means for statistically separating each stock and for classifying mixed stocks with measurable bounds of confidence at any point in time. Secondly, the technique can be used to construct stock specific run-timing curves through the fishery; with further 1 1 7 investigation to quantify yearly variability in run-timing, the technique can be used to predict the future impacts to any stock from various patterns of commercial fishing. Thirdly, the technique is flexible and can therefore be easily modified as new information becomes available. Fourthly, the technique is easily implementable and does not require large capital expenditure or effort. Only further extension of the results of this thesis will establish the long term usefulness of scale pattern analysis as a practical management tool. 1 18 LITERATURE CITED Amos, M.H., R. Murai and R. Pearson. 1963. The use of discriminant functions in the morphological separation of pink salmon. Int. N. Pac. Fish. Comm., Bull. 11:73-100. Anas, R.E., and S. Murai. 1969. Use of scales and a discriminant function for classifying sockeye salmon (Oncorhynchus  nerka) by continent of origin. Int. N. Pac"! Fish. Comm. Bull., 26:157-192. Aro, K.V. and J. McDonald. 1968. Times of passage of Skeena River sockeye and pink salmon through the commercial fishing area. FRB MS Rept. Series 984. Pacific Biological Station, Nanaimo, B.C. Bali, J.M. 1958. Scale analyses of steelhead trout, Salmo  gairdneri qairdneri Richardson, from various coastal watersheds in Oregon. M.Sc. Thesis. Oregon State College. Bilton, H.T. 1971. Identification of major British Columbia and Alaska runs of even and odd year pink salmon from scale characteristics. J. Fish. Res. Bd. Can. 29:295-301. Bilton, H.T., and H.B. Messinger. 1975. Identification of major British Columbia and Alaska runs of age 1.2 and 1.3 sockeye from their scale characters. Int. N. Pac."Fish. Comm., Bull. 32:109-129. Caliprice, J.R. 1971. X-ray spectrometric and multivariate analsysis of sockeye salmon (Oncorhynchus nerka) from different geographic regions. J. Fish. Res. Bd. Can. 28:369-377. Chapman, D.W. 1958. Studies on the life history of Alsea River steelhead. J. Wildlife Mangmt. 22:123-134. Chilcote, M.W., B.A. Crawford, and S. A. Leider. 1980. A genetic comparison of sympatric poulations of summer and winter steelheads. Trans. Amer. Fish. Soc. 109:203-206. 119 Chuganova, N.I. 1963. Age and growth studies in fish. Translated from Russian. National Science Foundation. Washington, D.C. 132pp. Clutter, R.I. and L.E. Whitesel. 1956. Collection and interpretation of sockeye salmon scales. Int. Pac. Sal. Fish. Comm. Bull., 9 159pp. Conrad, R.H. 1984. Management applications of scale pattern analysis methods for the sockeye salmon runs to Chignik, Alaska. Alaska Dept. Fish and Game, Informational Leaf. #233. 185pp. Cook, R.C. 1982. Stock identification of Sockeye salmon (Oncorhynchus nerka) with scale pattern recognition. Can J~. Fish. aquat~ Sci . 39:611-617. Cook, R.C, and G.E. Lord. 1978. Identification of stocks of Bristol Bay sockeye salmon (Oncorhynchus nerka), by evaluating scale patterns with a polynomial discriminant method. U.S. Fish and Wildlife Service. Bull. 76:415-423. Dark, T.A., and B.J. Landrum. 1964. Analysis of 1961 red salmon morhological data. Int. N. Pac. Fish. Comm. Annu. Rep. 1962:110-115. Dixon, W.J. (ed). 1981. BMDP Biostatistical Software (1981). Univ. of Calif. Press. Los Angeles. 726pp. Fisher, R.A. 1936. The use of multiple measurements in taxonomic problems. Ann. Eugen. 7:179-188. Fournier, D.A., T.D. Beacham, B.E. Riddell and C.A. Busack. 1984. Estimating stock composition in mixed stock fisheries using morphometric, meristic, and electrophoretic characteristics, Can J. Fish. Aquat. Sci. 41:400-408. Fukuhara, F.M., S. Murai, J.J. LaLanne, and D. Sribhibhadh. 1962. Continental origin of red salmon as determined from morpholbical characters. Int. N. Pac. Fish. Comm., Bull. 8:15-109. 1 20 Gilbert, E.S. 1969. The effect of unequal variance-covariance matrices on Fisher's linear discriminant function. Biometrics 25:505-515. Grant, W.S., G.B. Milner, P. Krasnowski and F.M. Utter. 1980. Use of biochemical genetic variants for identification of sockeye salmon (Oncorhychus nerka) stocks in Cook Inlet, Alaska. Can. J. Fish. Aquat. Sci. 37:1236-1247. Harding, D.R., and J.D. Buxton. 1971. Meristic characteristics of Pacific salmon and steelhead trout captured at Moricetown Falls 1961-1966. Manuscript report. Fisheries Service. Pacific region. Dept. of Fish, and Forestry.. Canada. 32pp. Hartt, A.C. 1962. Movement of salmon in the North Pacific Ocean and Berring Sea as determined by tagging, 1956-1958. Int. N. Pac. Fish. Comm., Bull. 6 157pp. Healy, M.C. 1983. Coastwide distribution and ocean migration patterns of stream and ocean type chinook salmon Oncorhynchus tshawytscha. Can. Field Nat. 97:427-433. Henry, K.A. 1961. Racial identification of Fraser River Sockeye salmon by means of scales and its application to salmon management. Int. Pac. Sal. Fish. Comm., Bull. 12. 97pp. Hill, D.R. 1959. Some uses of statistical analysis in classifying races of american shad (Alosa sapidissima). USFWS Fishery Bulletin 147:269-284. Horncastle, G.S. 1981. Life history of steelhead trout from the Somass River on Vancouver Island. B.C. Fish and Wildlife Branch. Fish. Tech. Circ. No. 51. 19pp. Hubbs, C.L., and K.F. Lagler. 1967. Fishes of the Great Lakes region. Univ. Michigan Press. Ann Arbor Michigan. 213pp. Ihssen, P.E., H.E. Booke, J.M. Casselman, J.M. McGlade, N.R. Payne, and F.M. Utter. 1981. Stock identification: materials and methods. Can. J. Fish. Aquat. Sci.. 38:1838-1855. 121 Johnson, R.A., and D.W. Wichern. 1982. Applied Multivariate Statistical Analysis. Prentice-Hall. Inc. Englewood Cliffs, New Jersey. 594pp. Jolicouer, P. 1959. Multivariate geographic variation in the wolf Canis lupis L. Evolution 13:283-299. Keating, J.F. 1959. Clearwater fishery investigations, exploration of methods for distinguishing steelhead trout progeny from resident races of rainbow trout. University of Idaho, Wildlife Cooperative Research Unit. 50pp. Krasnowski, P.V., and M.L. Bethe. 1978. Stock separation studies of Alaskan salmon based on scale pattern analysis. Alaska Dept. Fish and game., Inf. Leafl. No. 175. 37pp. Lachenbruch, P.A. 1975. Discriminant Analysis. Hafner Press, New York. 128pp. Larkin, P.A. 1972. The stock concept and management of Pacific salmon, pages 11-15 in: R.C. Simon and P.A. Larkin (ed). The stock concept on Pacific salmon. H.R. Macmillan Lectures in Fisheries. University of British Columbia, Vancouver, Canada, 231pp. Larkin, P.A., and J.G. McDonald. 1968. Factors in the population biology of the sockeye salmon of the Skeena River. J. Anim. Ecol. 37:229-258. Lear, W.H., and E.J. Sandeman. 1980. Use of scale characters and discriminant functions for identifying continental origin of Atlantic salmon. Rapp. Pi-v. Reun. Cons.int. Explor. Mer. 176:68-75. Lear, W.H., and R.K. Misra. 1978. Clinal variation in scale characters of Atlantic salmon (Salmo salar) based on discriminant function analysis. J. Fish. Res. Bd. Can. 33:43-47. Maclean, J.A., and D.O. Evans. 1981. The stock concept, discreteness of fish stocks and fisheries management. Can. J. Fish. Aquat. Sci. 38:1889-1898. 122 Maher, F.P. 1954. Life history of the steelhead as interpreted from the scales. M.A. Thesis. University of British Columbia. 78pp. Maher, F.P., and P.A. Larkin. 1955. Life History of the steelhead trout of the Chilliwack River, British Columbia. Trans. Amer. Fish. Soc. 84:27-38. Major, R.L., K.H. Mosher, and J.E. Mason. 1972. Identification of stocks of Pacific salmon by means of scale features, pages 209-231. In R.C. Simon and P.A. Larkin (ed). The Stock Concept in Pacific Salmon. H.R. MacMillan Lectures in Fisheries, University of British Colunbia, Vancouver, Canada. B.C. 231pp. Margolis, L. 1958. The application of parasitological data to the problem of recognizing the continent of origin of stocks of sockeye salmon in the North Pacific Ocean and adjacent seas- a summary of three years research. Fish, res. Bd. Can., Biol, Stat. Nanaimo, B.C. MS Rept. Biol. 658. 20pp. McBride, D.N., and S.L. Marshall.' 1983. Feasibility of scale pattern analysis to identify the origins of chinook salmon (Oncorhynchus tshawytscha) in the lower Yukon River commercial gillnety fishery, 1980-1981. Alaska. Dept. Fish and Game., Inf. Leafl. No. 208. 64pp. McDonald, J. 1981. The stock concept and its application to British Columbia salmon fisheries. Can. J. Fish. Aquat. Sci. 38:1657-1664. McGregor, A.J., S.A. McPherson, and S.L. Marshall. 1983. Feasibility of determining the origin of sockeye salmon in the Taku-Snettisham gillnet fishery using scale pattern analysis, 1981-1982. Alaska Dept. Fish and Game. Inf. Leafl. No. 219. 23pp. McGregor, E.A. 1924. A possible separation of the river races of king salmon in ocean-caught fish by means of anatomical characters. Calif. Fish, and Game, 9:138-150. 1 23 McKern, J.L., H.F. Horton, and K.V. Koski. 1974. Development of steelhead trout (Salmo qairdneri) otoliths and their use for age analysis and for separating summer from winter races and wild from hatchery stocks. J. Fish. Res. Bd. Can. 31:1420-1426. Milne, D.J. 1955. The Skeena River salmon fishery with special reference to sockeye salmon. J. Fish. Res. Bd. Can. 12:451-485. Misra, R.K., and I-H. Ni. 1983. Distinguishing beaked redfishes (deepwater redfish Sebastes mentella and Labrador redfish S.fasciatus) by discriminant analysis (with covariance) and multivariate analysis of covariance. Can. J. Fish. Aquat. Sci. 40:1507-1511. Mosher, K. 1963. Racial analysis of red salmon by means of scales. Int. N. Pac. Fish. Comm., Bull 11:31-56. Mulligan, T.J., L. Lapi, R. Keiser. S.B. Yamada, and D.L. Buewer. 1983. Salmon stock identification based on elemental composition of vertebrae. Can. J. Fish. Aquat. Sci. 40:215-229. Narver, D.W. 1969. Age and Size of steelhead in the Babine River, British Columbia. J. Fish. Res. Bd. Can. 26:2754-2760 Narver, D.W., and F.C. Withler. 1974. Steelhead of the Nanaimo River. Aspects of their biology and the fishery from three years of anglers catches. Fish, and Mar. Ser., Nanaimo, B.C., Circ No. 99. 25pp. Neave, F. 1944. Racial characteristics and migratory habits in Salmo gairdneri. J. Fish. res. Bd. Can. 6(3):245-251. Nie, N.H. (ed). 1975. SPSS:Statistical Package for the Social Sciences.Second edition. McGraw-Hill Book Company,(1975). Northcote, T.G. (ed). 1969. Symposium on salmon and trout in streams. H.R. MacMillan Lectures in Fisheries. University of British Columbia, Vancouver, Canada. 388pp. 1 24 Nyman, O.L. and J.H.C. Pippy. 1972. Differences in Atlantic salmon, Salmo salar, from North America and Europe. J. Fish. Res. Bd. Canada. 29:179-185. Oguss, E. and L.K. Evans. 1978. Incidental catches of steelhead trout in the commercial salmon fisheries of Barkley Sound, Johnstone Strait, and the Skeena and Fraser Rivers. B.C. Mar. Res. Branch Fish. Mangmt. Rep. No. 14.84pp. Oguss, E., and T.R. Andrews. 1977. Incidental catches of steelhead trout in the commercial salmon fisheries of the Fraser River, Skeena River, and Juan de Fuca strait. B.C. Mar. Res. Branch Fish. Mangmt. Rep. No. 7. 50pp. Parker, R.R., and P.A. Larkin. 1959. A concept of growth in fishes. J. Fish. Res. Bd. Can. 16(5). Parkinson, E.A. 1984a. Identification of steelhead stocks in the commercial net fishery of the southern British Columbia coast. B.C. Fisheries Branch Fish. Mangmt. Rep. No. 81. 15pp. Parkinson, E.A. 1984b. Genetic variation in populations of steelhead trout (Salmo qairdneri) in British Columbia. Can J. Fish. Aquat. Sci. 41:1412-1420. Parsons, L.S. 1971. Use of meristic characters and a discriminant function for classifying spring and autumn spawning Atlantic herring. Int. Comm. NW Atlant. Fish. Res. Bull. 9:5-9. Pella, J.J., and T.L. Robertson. 1979. Assessment of composition of stock mixtures. U.S. Fish and Wildlife Service, Fish. Bull. 77(2):387-398. Pimental, R.A. 1979. Morphometries, the multivariate analysis of biological data. Kendall Hunt Publishers. 250pp. Pinsent, M.E., and W.E. Chudyk. 1973. An outline of the steelhead of the Skeena River system. Unpubl. M.S. B.C. Fish and Wildlife Branch, Prince George, B.C. 26pp. 1 25 Reddin, D.G., and R.K. Misra. 1985. Hotelling's T2 to identify the origin of Atlantic salmon (Salmo salar) in a mixed stock fishery. Can. J. Fish. Aquat. ScT. 42:250-255. Ricker, W.E. 1972. Hereditary and Environmental factors affecting certain salmonid populations, pages 19-160 in: R.C. Simon and P.A. Larkin (ed). The stock concept in Pacific salmon. H.R. Macmillan Lectures in Fisheries. University of British Columbia, Vancouver, Canada, 231pp. Ricker, W.E. 1981. Changes in the average size and average age of Pacific salmon. Can. J. Fish. Aquat. Sci. 38:1636-1656. Riddell, B.E. and W.C. Leggett. 1981. Evidence for an adaptive basis for geographic variation in body morphology and time of downstream migration of juvenile Atlantic salmon (Salmo  salar). Can. J. Fish. Aqaut. Sci. 38:308-320. J Rowland, R.G. 1969. Relation of scale characteristics to river of origin in four stocks of chinook salmon (Oncorhynchus  tshawytscha) in Alaska. USFWS. Special scientific report No. 577. 6pp. Schaffer, W.M., and P.F. Elson. 1975. The adaptive significance of variations in life history among local populations of Atlantic salmon in North America. Ecology 56:577-590. Shapovalov, L., and A.C. Taft. 1954. The life histories of rainbow trout (Salmo gairdneri qairdneri) and silver salmon (Oncorhynchus kisutchT with special reference to Waddell Creek,California, and recommendations regarding their management. Calif. Fish and Game. Bull. 98. 375pp. Simon, R.C., and P.A. Larkin (editors) . 1972. The stock concept in Pacific salmon. H.R. Macmillan Lectures in Fisheries. University of British Columbia, Vancouver, Canada, 231pp. Smith, S.B. 1969. Reproductive Isolation in summer and winter races of steelhead trout, pages 21-38 in: T. G. Northcote (editor): Symposium on salmon and trout in streams. H.R. MacMillan Lectures in Fisheries. University of British Columbia, Vancouver, Canada. 388pp. 126 Tanaka, S., M.P. Shepard, and H.T. Bilton. 1969. Origin of chun salmon (Oncorhynchus keta) in offshore waters of the North Pacific in 1956-1958 as determined from scale studies. Int. N. Pac. Fish. Comm., Bull. 26:57-155. Taylor, E.B. 1984. Evidence for adaptive relationships between body morphology, body size, and swimming performance in British Columbia populations of coho salmon (Oncorhynchus  kisutch). Masters Thesis. University of British Columbia, Vancouver, Canada. 115pp. Taylor, G.D. 1968. Report on the preliminary survey of the Skeena River drainage streams. Unpubl. M.S. B.C. Fish and Wildlife Branch, 42pp. Thorpe, J.E., and K.A. Mitchell. 1981. Stocks of Atlantic salmon (Salmo salar) in Britain and Ireland: discreteness and current management. Can. J. Fish. Aquat. Sci. 38:1576-1590. Thorpe, R.S. 1976. Biometric analysis of geographic variation and racial affinities. Biol. Rev. 51:407-452. Todd, I. St. P., and P.A. Larkin. 1971. Gillnet selectivity on Sockeye (Oncorhynchus nerka) and Pink salmon (0. qorbuscha) of the Skeena River System, British"*Columbia. J. Fish. Res. Bd. Can. 28:821-842. Tredger, D. 1984. Skeena River boat shocking program:summary report. Fisheries Habitat and Improvement section. B.C. Fisheries Branch. Victoria. Tsuyuki, H., and S. N. Williscroft. 1977. Swimming stamina differences between genotypically distinct forms of rainbow trout (Salmo qairdneri) and steelhead trout. J. Fish. Res. Bd. Can. 34:996-1003. Utter, F.M. 1981. Biological criteria for definition of species and distinct intraspecific populations of anadromous salmonids under the U.S. Endangered species Act of 1973. Can. J. Fish. Aquat. Sci. 38:1626-1635. 1 27 Utter, F.M., and F.W. Allendorf. 1977. Determination of the breeding structure of steelhead populations through gene frequency analysis, pages 44-54: in T.J. Hassler and RR. VanKirk (editors) Genetic implications of steelhead management. Calif. Coop. Fish. res. Unit, Areata. Whately, M.R. 1977. Kispiox River steelhead trout: the 1975 sport fishery and life history characteristics from anglers' catches. B.C. Fish and Wildlife Branch, Fish. Tech. Circ. No. 30, 22pp. Whately, M.R., W.E. Chudyk, and M.C. Morris. 1978. Morice River steelhead trout: the 1976 and 1977 sport fishery and life history characteristics from anglers' catches. B.C. Fish and Wildlife Branch, Fish. Tech. Circ. No. 36, 25pp. Winter, G.W., C.B. Schreck and J.D. Mclntyre. 1980. Meristic variation in four stocks of steelhead trout (Salmo  qairdneri). Copeia. 1980:160-162. Withler, L. 1966. Variability in life history characteristics of steelhead trout (Salmo qairdneri) along the Pacific coast of North America. J. Fish. Res. Bd. Canada.23:365-393. Worlund, D.D. and R.A. Fredin. 1962. Differentiation of stocks, pages 143-153 in Symposium on Pink salmon, H.R. Macmillan Lectures in Fisheries, Univ. British Columbia, Vancouver, Canada. 128 APPENDICES 129 Appendix A: Discriminant Analysis and Classification Discriminant analysis reduces the variable vectors for individuals and centroids to single values (canonical variables, Di) by forming linear combinations of the original variables weighted according to their contribution to between groups discriminance (using partial one way ANOVA variable F scores as entry criteria). The discriminant functions are of the form: Di=dZ +dZ + ....+d z i 1 i1 2 i2 pip where Di is the discriminant score, for the ith individual, d1, d2, ....dp are the weighting coefficients and zi1, zi2...zip are standardized values of the measurements from the ith individual. The weighting coefficients are calculated so that the Di are standard normal variables and the grand mean discriminant score is zero with a standard deviation of one. Discriminant functions, their number being one less than the number of groups, are orthogonal to each other and describe group variation along different directional axes (figure 6). The major assumptions of discriminant analysis are a) that the groups being distinguished are identifiable (b) that the variable system being used is multivariate normal and (c) that the groups all share a common variance-covariance structure. Assumption b was tested as best possible by the examination of univariate frequency distributions. Assumption c was tested by the application of Box's multivariate M test (Nie, 1975). Classification matrices (confusion matrices, Johnson and Wichern, 1982) are derived in discriminant analysis through the use of classification functions; one for each stock. An empirical measure of group (stock) separability is obtained by classifying the individuals used to construct the discriminant functions into their most probable groups of origin using the classification functions. Lachenbruch's (1975) holdout classification procedure (jacknifing) was used in this study to reduce the bias in predicting classification error rates when using the same individuals for both discrimination and classification. Incidentally caught steelhead from the 1984 commercial fishery provided the samples of unknown stock composition to be classified to stock of origin. Of primary concern were the relative proportions of each stock predicted to be present during each week of the salmon fishery. Worlund and Fredin (1962) first described linear equations which adjust the predicted proportional estimates from the mixed sample to account for the errors in assigning individuals of known origin (the learning samples). Cook and Lord (1978) extended the procedure to more than two stocks using matrix algebra. Using their methodology, the classification accuracy estimated by the holdout procedure for a given learning sample is represented by the square matrix C, where the element Cij is the proportion of the sample from stock j that is classified as stock i. Letting r be a column vector r1,r2,r3,....ri, where ri is the proportion 130 of the mixed sample classified as stock I then: -1 U = C r where each element of the column vector U (U1, U2,...Ui) is the estimate of the proportion of stock I in the commercial sample after correcting for the errors in classifying individuals of known origin. Variances about these point estimates (Ui) were estimated using the formulae of Pella and Robertson (1979) and a 90% confidence interval was calculated for each estimate. The correction procedure of Cook and Lord (1978) is basically a modification of the two stock learning sample scenario: Classified Stock Actual Stock A B A Aa • Ab =C B Ba Bb where Aa, Ab, Ba, and Bb are the proportions of fish from their respective stocks correctly (Aa, Bb) and incorrectly (Ab, Ba) classified. Aa and Bb are the estimated probabilities of correctly classifying an unknown individual which actually belongs to one of those stocks whereas Ab and Ba are the estimated probabilities of misclassifying an individual actually belonging to one of the stocks as being from the other. In a mixed fishery sample, the proportions of fish assigned by discriminant analysis to each stock (Pa, Pb) represent both the correctly assigned individuals plus the uncorrectly assigned individuals. Solving for Na and Nb, the actual proportions of each stock present in a mixture,- is by solution of two simultaneous equations: Pa = Aa Na + Ab Nb Pb = Ba Na + Bb Nb or -1 r = C*U ..U = C r which reduces to the matrix adjustment procedure of Cook and Lord (1978). The elements of U can be greater than zero, less than zero, or equal to zero depending upon the proportion of a given stock actually in the commercial sample. Proportional estimates less than zero indicated the absence of a particular stock in the sample. Any samples resulting in proportional estimates less than zero in this study were reanalyzed using discriminant models which did not include those stocks. Appendix T.1. Age composition structure for the five stocks used in the study. AGE CLASS + + + + + + + + RI SEX 2. 1 2.2 3.1 3.2 3.3 4.1 4.2 4.3 RS TOTA1 1 M 0 0 4 9 4 3 13 1 8 42 F 0 0 1 14 1 2 19 1 12 50 % - - .05 .25 .05 .05 .35 .02 .22 1 2 M 0 0 8 4 1 10 9 1 2 35 F 0 0 5 4 0 28 10 0 7 54 % - - . 1 5 .09 .01 .43 .22 .01 .10 1 3 M 0 0 1 12 12 1 15 8 5 54 F 1 0 2 16 05 1 1 1 3 10 49 % .01 - .03 .29 .16 .02 .25 . 1 1 .14 1 4 M 0 0 2 21 1 1 7 0 2 34 F 0 0 1 35 0 3 17 1 0 57 O. -b - - .03 .62 .01 .04 .26 .01 .02 1 5 M 0 0 0 3 6 1 1 4 1 1 1 36 F 0 0 0 9 4 0 31 3 7 54 % - - - .13 . 1 1 .01 .50 . 16 .09 1 TOTALS % .002 - .05 .27 .09 . 1 1 .31 .06 . 1 2 1 RI KEY : 1 = ZYMOETZ, 1974,1978 n = 92 2 = MORICE , 1977 n = 90 3 = KISPIOX, 1975 n=l03 4 = BABINE , 1978 n = 9l 5 = SUSTUT , 1977,1983 n = 90 Appendix T.2. Sizes at age for the five stocks used in the study. Reported are the means, standard deviations and sample sizes for the major age classes. Age Kispiox Zymoetz Morice Babine Sustut L WT L WT L WT L WT L WT 3.1 + M X 61.1 2.3 56.7 1 .8 58.3 1 .8 57.3 2.0 - -s 0 0 2.63 0.5 4.14 0.5 1 .27 0 - -n 1 1 4 4 7 7 2 1 - -F X 59.7 2.9 67.0 2.7 55.8 1 .6 60.0 2.0 - -s 2.63 0.9 0 0 5.51 0.5 0 0 - -n 2 2 1 1 6 6 1 1 -3.2+ M X 86.5 7.7 82.0 5.6 75.3 4.0 81.5 5.3 84. 1 6.0 s 5.51 1 .3 8.46 2.6 3.33 0.9 6.39 1 .0 4.47 0.9 n 10 10 7 7 3 3 22 22 4 2 F X 82.8 5.9 75.5 4.1 73.2 3.3 78.5 4.6 77.0 4. 1 s 6.53 1 .8 3.49 0.6 5.83 0.4 4.72 0.9 4.23 0.9 n 17 17 15 15 5 5 20 20 8 5 3.3 + M X 99.9 10.2 91 .4 7.5 91 .5 7.5 91.4 7.4 93.3 8.9 s 8.25 2.4 6.59 1.8 0 0 0 0 3.37 0.9 n 12 12 5 5 1 1 1 1 6 4 F X 87.3 7.5 88.9 6.9 - - - - 86.4 6.1 s 8.88 2. 1 0 0 - - - - 6.58 1.2 n 5 5 1 1 - - — - 4 3 4.1 + M X 55.9 2.0 57.8 1.9 59.6 1 .8 60.3 2.0 55.9 1 .8 s 0 0 1 .36 0.4 4.68 0.4 0 0 0 0 n 1 1 3 3 10 10 1 1 1 1 X 55.9 1 .8 62.8 2.4 56.9 1 .5 60.5 2.0 63.5 2.7 s 0 0 2.47 0.7 3.43 0.4 0 0 0 0 n 1 1 2 2 30 30 1 1 1 1 4.2+ M X 90.0 8.2 80. 1 5.0 83.5 5.3 73. 1 3.4 84.6 6.8 5 9.45 2.3 3.91 0.5 7.77 1 .2 8.52 0.9 7.89 2.4 n 13 13 15 15 7 7 5 5 13 10 F X 77.9 5.3 74.9 4.6 71.9 3.4 74.3 4.4 77.2 4.6 s 8.35 1.0 4.72 1.8 2.58 0.6 1 .03 0.4 3.45 0.9 n 12 12 18 18 14 14 7 7 31 22 4.3+ M X 94.5 8.9 96.5 8.6 - - - - 96. 1 8.9 s 7.84 1.7 0 0 - - - - 4.02 1 .2 n 1 1 11 1 1 - - - — 10 10 X 87.2 6.7 79.5 4.5 - - - - 87.6 6.0 s 5.26 1.3 0 0 - - - - 1 .25 0 n 3 3 1 1 - - - — 3 1 RS M X 79.1 5.4 80.1 5.4 66.5 3.0 75.2 4.1 78.7 5.4 s 5.13 0.9 6.10 1 .7 9.19 1 .4 11.1 2.7 0 0 n 3 3 9 9 2 2 2 2 1 1 F X 89.4 7.6 84.2 5.8 82.6 5.1 - - 84.4 5.9 s 5.30 1.6 5.40 1 .2 5.37 1 .4 - - 2.91 1 .1 n 10 10 1 1 1 1 7 7 - - 7 4 Key M=males F=females X=mean s=S.D n=sample size 133 ix T.3. Variable means, standard deviations, and one way ANOVA F statistics for the five stocks used in the study by smolt age 3 (learning samples). GROUP » KISPIOX COPPER SUSTUT VARIABLE 1 PG 0. 89362 0. 4054 1 0. 26923 3 FWA 3. 00000 3 OOOOO 3. OOOOO 4 SWA 2 . 48936 2 . 27027 2 . 73077 5 L 87 . 23402 77 93782 84 . 31 152 6 WT 7 . 27234 4 . .88378 6. OOOOO 7 SEX 1 . 42553 1 . .43243 1 . 38461 8 A 1 0. 09234 0 .09351 0. 09000 9 A2 0. 14362 0 .15568 0. 15462 10 A4 0. 2 1043 0 .20189 0. 23 192 1 1 A5 6 . 72340 5. .67568 6. 846 15 12 A6 2 . 76596 2. .35135 2. 92308 13 B1 0. 04702 0. .04865 0. 03654 14 B2 0 .09745 0 .09838 0. 09500 IS B3 0. 14SS3 0. .15081 0. 15038 16 B4 0. 29043 0. .29459 0. 30269 17 65 1 1 . 59574 1 1 . .59459 1 1 . 53846 18 B6 5 . 702 13 s. .56757 5. 42 308 19 CI 0. 05128 0. .05189 0. 04385 20 C2 0. 10787 0 .10622 0. 1 1385 21 C3 0. 16362 0. .16432 0. 17692 22 C4 0. 34234 0 35514 0. 37654 23 C5 12. 51064 12. .67568 12. 57692 24 C6 5 .85106 5 .97297 5. 8846 1 25 01 0 .08489 0 .08811 0. 06731 26 D2 0 .18128 0 .18054 0. 16500 27 03 0 .28723 0 .27784 0. 26885 28 04 1 .65319 1 .69459 1 . 53615 29 D5 35 .72340 35 . 10809 32 . 15384 30 D6 16 .00000 15 .45946 15. 65385 COUNTS 47. 37. 26. F TO BABINE MORICE ALL GPS. ENTER DF = 4 186 0. 29310 0. 69565 0. 50785 3 350 3 OOOOO 3. OOOOO 3 OOOOO 0 .0 2 01724 1 . 652 17 2 . 23560 9 . 876 77 . 33275 66 21738 79 49789 22 .597 4 59403 2 . 80435 5 28534 31 . 745 1 39655 1 . 43478 1 .41361 0 .069 0 .09345 0 10087 0 .09361 0 .829 0 14983 0 16522 0 . 15194 2 .970 0 23017 0. 24783 0 22220 3 .803 7 32759 7 . 82609 6 85340 6 . 266 3 .06896 3 . 26087 2 .85864 3 . 935 0 04931 0. .04261 0 .04607 6 .677 0 .10569 0 09 130 0 .09906 3 315 0 16086 0 14 174 0 .1514 1 4 . 154 0. 3462 1 0 .24783 0 .30471 8 . 178 13. .22414 10. .47826 11 ,94764 6 . 48 1 6 .25862 4 . .65217 5 .68063 7 .080 0 .05345 0 .04087 0 .04979 7 . 169 0 .11586 0 .09174 0 .10885 9 . 277 0 17776 0 14652 0 16780 7 933 0 38672 0 .28957 0 .35660 4 . SOO 14 .05172 10. 95652 12 83246 5. 585 6 .55172 5. .08696 6. .OOOOO 5. 234 0 .08552 0 .08000 0 08272 5. 281 0 .19086 0. 17478 0. ,18105 3 332 0 .30655 0 28609 0 28864 4 . 601 1 .78207 1 .70435 1. 69057 2 . 689 34 .63792 34 .43477 34 . .63350 1 . 678 15 .91379 15 .56522 15 .76963 0. 389 58 . 23. 191 . STANDARD DEVIATIONS GROUP VARIABLE KISPIOX COPPER SUSTUT BABINE 1 PG 1 . 32261 0. 92674 0. 72430 0. 67560 3 FWA 0. 0 0. 0 0. 0 0. 0 4 SWA 0. 7481 1 0. 87078 0. 82741 0. 39698 5 L 9 . 74736 10. 74704 7 . 26857 7 . 6 1405 6 WT 2. 09032 2. 03790 1 . 71277 1 12475 7 SEX 0. 49977 0. 50225 0. 496 14 0. 49345 8 A1 0. 02098 0. 02530 0. 02 135 0. 0209 1 9 A2 0. 02462 0. 02387 0. 02970 0. 025 17 10 A4 0. 05373 0. 03865 0. 05485 0. 049 19 1 1 A5 2. 05047 1 . 27048 1 . 86959 1 . 64783 12 A6 1 . 12699 0. 58766 0. 97665 0. 93400 13 B1 0. 00998 0. 01 159 0. 01263 0. 01 197 14 B2 0. 01687 0. 01756 0. 01903 0. 01836 15 83 0. 02385 0. 02 1 1 3 0. 02457 0. 02364 16 B4 0. 06659 0. 06517 0. .06372 0 097 13 17 B5 2. 14334 2. 25412 2. 68672 2. 84 106 18 B6 1 . 12123 1 ,30257 1 . .36156 1 39624 19 C1 0. 01296 0. ,01221 0. 01061 0. 01 132 20 C2 0 .01488 0 .02086 0 .01577 0 01697 21 C3 0. .02523 0 .02882 0 ,02223 0 .02527 22 C4 0 .08352 0 .08909 0 .13323 0 .11063 23 C5 2 .91078 2 .92550 3 .59080 2 .56441 24 C6 1 .36698 1 .32259 1 .88312 1 .20193 25 01 0 .01679 0 .02459 0 .02475 0 01613 26 D2 0 .02651 0 .04007 0 .02470 0 .03074 27 03 0 .03820 0 .04995 0 .03241 0 .04024 2B D4 0 .38385 0 .30138 0 .32883 0 .26839 29 D5 6 .33039 s .58660 5 .75980 5 .15614 30 06 2 .57917 2 .28028 2 .29682 . 2 .30396 MORICE 0.97397 0.0 0.83168 1 1 .61193 1 .53637 0.5068 7 0.02314 0.03232 0.07580 2.62249 .48377 .01176 .02201 .02289 .05705 19233 1.07063 0.00793 0.01403 0.01774 .05653 .63702 94931 .01706 .03369 .05289 .39773 .47250 .08514 ALL GPS. 0.95992 0.0 0.71224 9.30948 .71625 49868 .02216 .02639 .05318 .86214 1.01828 .01150 .01842 .02327 .07577 .47624 .27336 .01150 .01685 0.02485 0.09887 2.79591 1.35152 0.01958 0.03147. .04257 .33066 .78897 .34552 1 . O O. O. O. 1 . O. 0. O. o. 2. 1 . O. 0. o. 0. 5. 2 . 134 Appendix T.4. Variable means, standard deviations, and one way ANOVA F statistics for the five stocks used in the study by smolt age 4 (learning samples). GROUP KISPIOX COPPER SUSTUT BABINE MORICE ALL GPS. ENTER VARIABLE DF= 4 23' 1 PG 0. 43243 0. 29167 0. 33898 0. 20000 0. 43077 0. 35146 0 . 539 3 FWA 4 . 00000 4. 00000 4 . 01695 4 . 00000 4 . 01538 4 . 00837 0 .464 4 SWA 2. 64865 2. 56250 2. 38983 1 . 90000 1 . 5846 1 2. 184 10 14 . 778 5 L 84 . 69188 77 . 00624 83. 38982 72 . 98666 65. 04 308 76. 01379 4 1 . 365 6 WT 6. 73513 4. 56458 5. 94915 .1 . 10333 2 . 62923 4 . 65816 50 . 272 7 SEX 1 . 51351 1 . 45833 1 . 4067B 1 . 30000 1 . 30769 1 . 39330 1 . 565 8 A1 0. 09378 0. 09687 0. 09220 0. 10000 0. 09815 0. 09598 1 .007 9 A2 0. 14568 0. 15167 0. 15559 0. 15167 0. 15800 0. 15343 1 8 .460 10 A4 0. 18216 0. 18708 0. 2 1 136 0. 22900 0. 22369 0. 20753 .607 1 1 A5 5. 48649 5. 354 17 6. 15254 7 . 40000 6. 73846 6 . 20502 18 . 457 12 A6 2 . 37838 2 . 12500 2. 52542 2. 93333 2 646 15 2 . 50628 7 . 150 13 B1 0. 04 351 0. 04750 0. 04305 0 04867 0. 04092 0. 044 14 3 . 533 14 B2 0. 09108 0. 09917 0. 10220 0. 09933 0. 09 108 0. 09649 4 . 522 15 B3 0. 13784 0. 15021 0. 16068 0. 15167 0. 13723 0. 14753 10. 842 16 B4 0. 24135 0. 24 146 0. 27339 0 28167 0. 22985 0. 25121 7 . 404 17 B5 10. 43243 9. 50000 9. 89830 1 1 . 00000 9. 78461 10. 00837 3. 401 18 B6 4 . 97297 4. 37500 4. 72881 5. 20000 4 . 67692 4 . 74059 4 . 126 19 C1 0. 04622 0. 04833 0. 04373 0. 05067 0. 04 200 0. 04544 3. 835 20 C2 0. 10432 0. 10250 0. 10949 0. 10700 0. 09462 0. 10293 5 . 570 21 C3 0. 15865 0. 15833 0. 17339 0. 16 100 0. 14338 0. 15837 1 1 . 348 22 C4 0. 27297 0. 27646 0. 29797 0. 27633 0. 23938 0. 27 113 7 . 648 23 C5 10. 16216 10. 43750 10. 25424 10. 66667 9. 96923 10. 25105 0 934 24 ce 5. 08108 4 . 87 500 4 . 94915 5. 03333 4 . 6923 1 4 . 89540 0. 959 25 01 0. 07838 0. 08167 0. 07237 0. 08900 0. 07354 0. 07757 4 . 399 26 02 0. 17568 0. 1777 1 0. 17339 0. 19767 0. 1743 1 0. 1779 1 3 . 535 27 03 0. 29108 0. 28667 0. 27661 0. 3 1533 0. 28262 0. 28736 4 . 588 28 •4 1 . 85540 1. 76479 1. 63034 1 . 79100 1 . 6663 1 1 . 72213 4 . 325 29 D5 38. 05405 36. 16666 32. 77965 35 . 29999 33 . 8 1538 34 . 87447 5 . 798 30 06 16 . 59459 15 . 89583 15. 627 12 15 . 70000 15. 33846 15 . 76 15 1 1 . 766 31 E 1 0. 05216 O. 05125 0. 04983 0. 05267 0. 04277 0. 0489 1 5. 637 32 E2 0. 1 1 162 0. 10667 0. 12220 0. 1 1500 0. 09569 0. 10933 15 . 167 33 E3 0. 16784 0. 16396 0. 19068 0. 17633 0. 14 800 0. 16837 2 1 . 306 34 E4 0. 27622 0. 31292 0. 37220 0. 34733 0. 27646 0. 31628 12 . 819 35 E5 9. 70270 10. 85417 12 . 1 1864 1 1 . 70000 1 1 . 13846 1 1 . 17 155 4 . 990 36 E6 4 . 62162 4 . 89S83 5. 81356 5 . 56667 5 . 20000 5. 24686 4 . 938 COUNTS 37 . 48 . 59 . 30. 65 . 239 . STANDARD DEVIATIONS GROUP KISPIOX COPPER SUSTUT BABINE MORICE ALL GPS. VARIABLE 83960 1 PG 0 .95860 0 .7 7069 0 .88298 0 .550B6 0 .88334 0 3 FWA 0. ,0 0 .0 0 .13019 0 .0 0 .12403 0 09 170 4 SWA 0 .85687 1 18333 0 .61635 0 .40258 0 .89952 0. 85425 5 L 10. .90086 8 .33509 8 .40368 5 .51172 10 .81063 9. 24489 6 WT 2. .14556 1 .50325 1 . ,80453 0 .76089 1 .50610 1 . 63135 7 SEX 0. .50671 0. ,50353 0 49545 0 .46609 0 .46513 0. 48720 8 A 1 0. .02487 0 .02085 0. .02009 0 .02133 0 .02098 0. 02143 9 A2 0. .02588 0 .02579 0. ,02430 0. ,02793 0 .02852 0. 02650 10 A4 0. 04905 0 ,03690 0. .04066 0 .05215 0 .05421 0. 04684 11 A5 1 21613 1 .02084 0 .99678 1 8 1 184 1 .31431 1. 24995 12 A6 0. .63907 0 .53096 0. .67864 0 . 94443 0 .75892 0. 70829 13 B1 0 01060 0 .01139 0. 01235 0 01224 0 .01142 0. 01 164 14 B2 0 .01370 0. .01699 0. .02026 0. 01660 0 01724 0. 01744 15 S3 0 .02175 0 .02274 0. .02399 0 .02183 0 .02058 0. 02223 16 B4 0. .05271 0 .04807 0. .05827 0. .08914 0 .04185 0. 05654 17 B5 1 . .95136 1 .65027 2 .09016 2 .75431 1 .60558 1. 96490 18 B6 0 .95703 0 .81541 0 .94377 1 .37465 0 .81216 0. .95384 19 C1 0 .01361 0 .01243 0 .01299 0 .01285 0 .00870 0. 01 194 20 C2 0 .01980 0 .01781 0 .02021 0 .02292 0 .01370 0 .01848 2 1 C3 0 .02594 0 .02435 0 .02577 0 .02940 0 .02 138 0 .02489 22 C4 0 .06806 0 .06406 0 .06501 0 .05980 0 .04596 0 .06005 23 C5 2 .06173 1 .79723 1 .88087 1 74856 I .63906 1 8 1472 24 C6 1 .78541 0 .93683 1 .02425 0 .85029 0 .88252 1 .10875 25 01 0 .02328 0 .02014 0 .02029 0 .01689 0 .02080 0 ,02051 26 D2 0 .01994 0 .02860 0 .03693 0 .02788 0 .03455 0 .03141 27 03 0 .03116 0 .04 138 0 .04334 0 03329 0 .04925 0 .04195 28 D4 0 .38639 0 .29673 0 .28399 0 .34343 0 .24515 0 .3028 1 29 05 7 .26080 5 .92266 5 .68745 5 .79624 4 .98087 5 .84140 30 D6 2 .66 103 2 .51158 2 .37005 2 .32156 2 .05606 2 .36067 3 1 E 1 0 .01250 0 .01315 0 .01491 0 .01230 0 .00976 0 .01260 32 E2 0 .02089 0 .01849 0 .02335 0 .01737 0 .01677 0 .01963 33 E3 0 .02678 0 . 03009 0 .02888 0 .02141 0 .02251 0 .02635 34 E4 0 .05574 0 .07252 0 .09828 0 .11020 0 .08352 0 .08568 35 ES 1 .85390 2 .19273 2 .76732 3 .97534 2 .68022 2 .70478 36 E6 0 .95310 1 .01561 1 .80492 2 .04574 1 .32523 1 .46743 ' 135 ix T.5. Variable means, standard deviations, and one way ANOVA F statistics for the five stocks used in the study by age 3.2 + (learning samples). GROUP - KISPIOX COPPER SUSTUT VARIABLE 1 PG 1 , OOOOO 0. .55556 0. 25000 3 FWA 3. OOOOO 3 OOOOO 3. OOOOO 4 SWA 1 . 92593 1 . 81481 2 . OOOOO 5 L 83. 05554 74 . 48517 79. 38332 6 WT 6. 47037 4 . 08889 4 . 90833 7 SEX 1 . .37037 1 . .44444 1 . 33333 8 Al 0. .09074 0. .08926 0. 08667 9 A2 0 .14407 0. .15185 0. 15583 10 A4 0. .21370 0 20000 0. 23167 1 1 A5 6. 74074 5, 85185 6. 91667 12 A6 2. .77778 2 . .44444 2 . 83333 13 B1 0 .04667 0. 04815 0. 034 17 14 B2 0 .09704 0. 09852 0. 09167 15 B3 0 . 14519 0 .15111 0. 14667 16 B4 0 28444 0. .29000 0. 30167 17 B5 11 .29630 11 66667 12. OOOOO 18 B6 5. .70370 5. 66667 5. 75000 19 C1 O ,05148 0. 05074 0. 044 17 20 C2 0 .10889 0. .10407 0. 1 1000 21 C3 0 .16630 0 .16037 0. 16750 22 C4 0 .32556 0 .33333 0. 30750 23 C5 11 .85185 12 .25926 10. 91667 24 C6 5 .62963 5. .85185 5 . 16667 25 01 0 .08296 0 .08444 0. 06583 26 02 0 .17852 0 .17519 0. 16500 27 03 0 .28630 0 .27148 0. 26667 28 04 1 .69963 1 .69555 1 . 47000 29 D5 37 .22221 35 .14815 30. 75000 30 D6 16 .44444 15 .59259 14 . 91667 COUNTS 27. 27. 12. F TO BABINE MORICE ALL GPS. ENTER DF = 4 136 0. 29091 0 60000 0. 51773 2 758 3 OOOOO 3. OOOOO 3. OOOOO 0 .0 1 . 94545 1 . 40000 1 . 84397 12 .547 77 . 28 18 1 63 . 72499 76 . 10779 16 .543 •4 . .58727 2 . 45000 4 . 57660 24 . 750 1 . 38182 1 . 45000 1 . 397 16 0 .200 0 .09400 0. 10150 0. 09291 1 . 259 0 .15018 0 16700 0. 15220 2 .318 0 23073 0. 25900 0. 22567 3 . 978 7 . 34545 8. 20000 7 . 02837 5 . 227 3 . .09091 3. 45O00 2 . 936 17 3 .420 0. 04927 0. .04 ISO 0. 046 17 4 .605 0 .10545 0. .08950 0. 09908 3 .277 0 16018 0. .14000 0. 15156 3 .460 0 .34509 0 .24850 0. 30553 6 .64 1 13 .23636 10 55000 12 . 07801 5 153 6. 25455 4 . 70000 5 . 77305 4 . 956 0. .05327 0 .04000 0. 04979 5 912 0 .11636 0 .09200 0. 10858 a .758 0 ,17836 0 .14700 0. 16723 6 780 0 .38909 0 .28750 0. 34489 5 .685 14 .12727 10 .90000 12 . 60284 8 . 223 6 .60000 5 .OOOOO 5 . 92 199 7 878 0 .08509 0 .08100 0. 08234 2 8 18 0 . 19145 0 .17700 0. 18 156 2 602 0 .306 18 0 .28550 0. 28943 4 . 075 1 .79054 1 .67350 1. 71106= 2 . 678 34 .87273 34 .20000 34 . 92908 3. 06 1 16 .03636 15 .65000 15. 87943 1 . 217 55. 20. 14 1. STANDARD DEVIATIONS GROUP VARIABLE PG FWA SWA L WT SEX 8 A1 9 A2 10 A4 1 1 A5 12 A6 13 B1 14 B2 15 B3 16 B4 17 B5 18 B6 19 CI 20 C2 21 C3 22 C4 23 C5 24 C6 25 DI 26 D2 27 D3 28 04 29 D5 30 06 KISPIOX 1.38675 0.0 0.26688 9.45085 2.03334 0.49210 0.02093 0.02358 0.05197 2.04925 1.01274 0.01144 01540 02486 06606 07206 13730 01292 01340 02589 08568 .93131 .44510 .01772 .02670 0.03904 0.36378 5.04847 2.25888 COPPER SUSTUT 1. 05003 0. 62158 0. 68510 0. 94032 0. 95894 0. 0 0. 0 0. O 0. 0 0. 0 0. 39585 0. 0 0. 22918 0. 50262 0. 31579 9. 55879 5. 39491 7 . 05696 9 . 84280 8 . 37883 1 . 61896 t . 01216 1 . 04084 1 . 08845 1 . 40357 0. 50637 0. 49237 0. 4903 1 0. 51042 0. 49676 0. 02464 0. 02309 0. 02 122 0. 02300 0. 02226 0. 02434 0. 03175 0. 02535 0. 03278 0. 02658 0. 03772 0. 06043 0. 05036 0. 07483 0. OS360 1 . 29210 1. 92865 1. 68015 2 . 58742 1 . 86520 0. 64051 0. 93744 0. 948 15 1 . 50350 1 . 009 13 0, .01210 0. 01505 0 01230 0. 01226 0. 01234 0. 01895 0. 02368 0. 01854 0. 02305 0. 01924 0. 02190 0. 03085 0. 0236 1 0. 02406 0. 02427 0. 06139 0. 07371 0. 09937 0. 0585 1 0. 07996 2 . 41788 3. 49024 2 . 87365 2 . 18788 2 . 62132 1 38675 1 . 71225 1 ,43007 1 . 08093 1 . 35178 0. .01207 0. .01 164 0 .01123 0 .00795 0. 01 139 0 .02080 0 .01128 0 .01671 0 .01473 0 01638 0 .02848 0 .01485 0 .02507 0 .01895 0 02451 0 .09004 0 ,07569 0 . 1 1263 0 .06034 0 .09467 3 .10821 2 .90637 2 .59667 1 .68273 2 .69177 1 .40613 1 .40346 1 .21106 0 .91766 1 .27927 0 .02375 0 .01975 0 .01620 0 .01651 0 .01849 0 .03817 0 .02505 0 .03123 0 .03278 0 .03167 0 .05013 0 .03143 0 .04039 0 .05336 0 .04356 0 .27301 0 .36449 0 .27288 0 .39001 0 .31798 5 .55187 5 .97152 5 .18563 6 .20356 5 .45005 2 .25762 1 .97522 2 .30107 2 .13431 2 .23700 136 ix T.6. Variable means, standard deviations, and one way ANOVA F statistics for the five stocks used in the study by age 4.2 + (learning samples). MEANS F TO GROUP KISPIOX COPPER SUSTUT BABINE MORICE ALL GPS. ENTER VARIABLE DF = 4 18! 1 PG 0 .44444 0 .32558 0 .33898 0 .23077 0 .26923 0 .33158 0 . 296 3 FWA 4 OOOOO 4 .OOOOO 4 .01695 4 OOOOO 4 .03846 4 .01053 0 .817 4 SWA 2 .69444 2 .74419 2 .38983 2 .03846 2 .46154 2 .48947 4 .021 S L 85 .49165 79 .01161 83 .38982 74 .39615 76 .21922 80 .58525 12 . 366 6 WT 6 .87222 4 .851 16 5 .949 15 4 .13077 4 .12692 5 .37737 19 .622 B A1 0 .09417 0 .09651 0 .09220 0 .10038 0 .09654 0 .09526 0 .723 9 A2 0 .14583 0 .15163 0 .15559 0 .15269 0 .15692 0 .15263 1 .004 10 A4 0 .18222 0 .18651 0 .21136 0 .22885 0 .22731 0 .20479 7 .408 1 1 A5 5 .47222. 5 .32558 6 .15254 7 .53846 7 .07692 6 .15263 18 . 963 12 A6 2 .36 1 11 2 .09302 2 .52542 3 .OOOOO 2 .80769 2 .50000 8 .044 13 B1 0 .04361 0 .04837 0 .04 305 0 .04692 0 .04077 0 .04458 2 . 567 14 B2 0 .09083 0 .10023 0 .10220 0 09769 0 .08923 0 .09721 3 .990 15 B3 0 .13750 0. . 15209 0 .16068 0 14885 0 .13577 0 14932 8 529 16 B4 0 .24222 0 .24442 0 .27339 0. 27346 0 22269 0 .25400 4 , 753 17 B5 10. .47222 9. .44186 9 .89830 10. 96154 9 .80769 10. .03684 2 . 759 18 B6 5 OOOOO 4 . 39535 4 72881 5. 1 1538 4 . 57692 4 . 73684 2. 945 19 C1 0. 04639 0. 04953 0 04373 0. 04846 0. 04385 0. 04621 1 , 835 20 C2 0. 10417 0. 10512 0. .10949 0. 10500 0. 09538 0. 10495 2 . 47 1 21 C3 0. 15917 0. 16256 0. 17339 0. 15846 0. 14577 0. 16242 6 19 1 22 C4 0. 27528 0. 28116 0. 29797 0. 27731 0. 24885 0. 28032 2 934 23 CS 10. 22222 10. 39535 10. 25424 10. 76923 9 . 88461 10. 3OOO0 0. 790 24 C6 5. 11111 4. 88372 4 . 94915 5. 07692 4 . 65385 4 . 94210 0. 703 25 01 0. 07917 0. 08395 0. 07237 0. 09000 0. 07000 0. 07837 5. 288 26 02 0. 17556 0. 181 16 0. 17339 0. 19962 0. 16654 0. 17821 4 . 728 27 03 0. 29083 0. 29070 0. 27661 0. 3 1846 0. 27 192 0. 28758 6 . 031 28 04 1 . 84778 1 . 77581 1 . 63034 1 . 796 15 1. 70346 1. 737 16 3. 182 29 05 37. 91666 36. 06976 32. 77965 35 65384 35. 61537 35. 27895 4 . 463 30 D6 16. 52777 15. 90698 15. 62712 IS. 80769 15. 76923 15. 90526 0. 807 31 E1 0. 05250 0. 05256 0. 04983 0. 05154 0. 04423 0. 0504 2 2 . 066 32 E2 0. 1 1 194 0. 10721 0. 12220 0. 1 1385 0. 09654 0. 1 122 1 8. 059 33 E3 0. 16778 0. 16442 0. 19068 0. 17500 0. 14923 0. 17258 12 . 139 34 E4 0. 27556 0. 30907 0. 37220 0. 34 192 0. 26846 0. 32126 1 1 . 249 35 E5 9. 66667 10. 72093 12 . 1 1864 1 1 . 7 3077 1 1 . OOOOO 1 1 . 13158 5. 262 36 E6 4 . 61111 4 . 81395 5. 81356 5 . 53846 5 . 15385 5. 23158 4 . 983 COUNTS 36. 43. 59. 26 . 26. 190. STANDARD DEVIATIONS GROUP KISPIOX COPPER SUSTUT BABINE MORICE ALL GPS. VARIABLE 83595 1 PG 0. 96937 0. 80832 0. 88298 0. 58704 0. 77757 0. 3 FWA 0. 0 0. 0 0. 13019 0. 0 0. 19612 0. 10252 4 SWA 0. 821B2 1 . 1 1468 0. 61635 0. 19612 0. 85934 0. 7962 1 5 L 9. 89349 6. 10534 8. 40368 3. 76843 7. 77529 7 . 69473 6 WT 2 . 00494 1 . 30190 1 . 80453 0. 68397 1 . 29230 1 . 56685 8 A1 0. 0251 1 0. 02080 0. 02009 0. 02254 0 02226 0. 02 190 9 A2 .0. 02623 0. 02600 0. 02430 0. 02878 0 026 19 0 02595 10 1 1 A4 0. 04975 0. 03810 0. 04066 0. 05450 0. 05604 0. 04628 A5 1 . 23024 1 . 01702 0. 99678 1 . 85969 1 38342 1 . 24845 12 A6 0. 63932 0. 52617 0. 67864 0. 97979 0 89529 0. 72292 13 B 1 0. 01073 0. 01111 0 01235 0. 01 192 0. 0089 1 0 .01130 14 B2 0. 01381 0. 01739 0 02026 0. 01704 0. .017 19 0 ,01768 15 83 0. 02 196 0. 023 15 0. 02399 0. 02142 0 02 194 0 0228 1 16 B4 0. 05319 0. 04896 0 05827 0 0897 1 0 04423 0. ,05912 17 B5 1 . 96376 1 . 70855 2. 09016 2 . 86329 1, 6497 1 2. ,05854 18 B6 0. .95618 0. 84907 0 .94377 1 . 42343 0. 94543 1 ,00523 19 C1 0. 01376 0. 01234 0. ,01299 0. 01 156 0. .00898 0 .01234 20 C2 0. .02005 0. 01653 0 .02021 0 .02319 0 .01421 0 .01913 21 C3 0. ,02612 0. .02128 0 .02577 0 .02880 0 .02062 0 .02469 22 C4 0 .06755 0. .06272 0 .06501 0 .06213 0 .04366 0 .06214 23 C5 2 .05789 1 . .80131 1 .88087 1 .79572 1 .68 1 12 1 .86133 24 C6 1 .80123 0 .93119 1 .02425 0 .89098 0 .89184 1 .16376 25 DI 0 .02310 0 .01966 0 .02029 0 ,01789 0 .02040 0 .02042 26 02 0 .02021 0 .02822 0 .03693 0 .02932 0 .02993 0 .03038 27 03 0 .03157 0 .04 17 1 0 .04334 0 .03379 0 .04656 0 .04024 28 D4 0 .38904 0 .29626 0 .28399 0 .35787 0 .28305 0 .31934 29 05 7 .31485 5 .96575 5 .68745 5 .95947 5 .02056 6 .04557 30 D6 2 .66711 2 .57102 2 .37005 2 .41693 2 .04563 2 .44179 31 E 1 0 .01251 0 .01311 0 .01491 0 .01223 0 .00902 0 .01302 32 E2 0 .02109 0 .01919 0 .02335 0 .01813 0 .01573 0 .02041 33 E3 0 .02716 0 .03165 0 .02888 0 .02232 0 .02331 0 .02773 34 E4 0 .05639 0 .07243 0 .09828 0 .11275 0 .0629 1 0 .084 10 35 E5 1 .86701 2 .11936 2 .76732 4 1042 1 2 .51396 2 .68522 36 E6 0 .96445 0 .98212 1 .80492 2 .08290 1 .22286 1 .48480 4 13 7 Appendix T.7. Age composition structure for the 1984 commercial fishery steelhead samples. AGE CLASS + + + + + + + + WK SEX 2.1 2.2 3.1 3.2 3.3 4.1 4.2 4.3 RS TOTAL 9 M 2 2 F . 0 0 % .02 .02 10 M 0 1 F 0 0 % - .01 1 1 M 2 2 F 0 0 % .01 .01 12 M 0 0 F 0 2 % - .02 13 M 1 1 F 0 2 % .01 .02 1 4 M 0 5 F 0 2 % • - .06 10 27 6 12 1 20 4 5 .08 .35 .08 .13 7 28 7 8 4 25 4 3 .08 .41 .08 .08 12 25 6 1 1 5 25 1 8 . 13 .37 .05 .14 17 16 6 9 6 25 12 3 . 18 .32 . 14 .09 17 24 1 1 2 2 20 1 4 . 1 5 .34 .09 .05 7 22 9 7 3 1 7 7 2 .08 .33 . 1 4 .08 1 4 6 2 81 1 4 4 4 52 .21 .08 .05 1 10 7 9 77 7 7 3 53 .13 . 1 1 .09 1 15 2 7 82 7 0 6 52 .16 .01 . 10 1 5 3 9 65 9 0 5 62 . 1 1 .02 . 1 1 1 7 7 14 84 6 0 1 1 46 .10 .05 .19 1 10 3 2 65 15 1 6 53 .21 .03 .07 1 TOTALS % .01 .02 .11 .36 .10 .10 .16 .05 .10 1 WEEK KEY: 9 = ending July 21 n= 133 10 = ending July 31 n = 1 30 1 1 = ending Aug. 7 n= 134 1 2 = ending Aug. 1 4 n= 127 13 = ending Aug. 21 n= 130 1 4 = ending Aug. 31 n = 1 18 138 Appendix T.8. Numerical runtiming estimates for the five stocks used in this study through the 1984 commercial fishery. Model A: smolt age 3/scale variables alone. Model B smolt age 4/scale variables alone. Model C pooled smolt age/scale variables alone. Model D pooled smolt age/all variables. 1984 estimated run timing by week model a 9 10 1 1 12 13 1 4 total stock mor ice 3670 1 244 1606 2374 1 330 555 10779 babine 210 626 1219 2762 1073 - 5890 sustut 3670 5418 2055 3850 3225 1734 19952 zymoetz - 44 1 - - - 257 698 kispiox 525 - 573 - 948 - 2046 by proportion mor ice .340 .115 . 149 .220 . 123 .052 1 .00 babi ne .036 . 106 .207 .469 .182 - 1 .00 sustut . 184 .272 . 103 .193 .162 .087 1 .00 zymoetz - .632 - - - .368 1 .00 kispiox .257 - .280 - .463 - 1 .00 model D 9 10 1 1 12 13 14 total stock mor ice 1707 1064 492 - 394 375 4032 babine 787 532 - 687 - 83 2089 sustut 2003 1812 1066 1 032 1289 751 7953 zymoetz 1223 - 1721 1 1 46 693 334 5117 kispiox 1001 1 169 - 687 495 3352 by proportion morice .423 .263 . 122 - - - 1 .00 babine .376 .254 - .329 - .040 1 .00 sustut .252 .228 . 134 . 130 .162 .094 1 .00 zymoetz .239 - .336 .224 .135 .065 1 .00 kispiox .299 .349 - .205 . 1 48 - 1 .00 model c 9 10 1 1 1 2 13 1 4 total stock morice 5164 3035 1749 2067 1 277 683 1 3975 babine 885 1232 1267 2372 1920 552 8228 sustut 6368 6171 3793 5353 4 1 90 261 4 28489 zymoetz 551 846 776 1 229 426 340 4168 k i spiox 1537 1 1 32 1329 1751 1627 303 7679 by proportion morice .369 .217 . 125 . 147 .091 .049 1 .00 babine .107 .150 . 1 54 .288 .233 .067 1 .00 sustut .224 .217 . 133 .188 .147 .092 1 .00 zymoetz . 1 32 .203 .186 .294 .102 .082 1 .00 kispiox .200 .147 .173 .228 .212 .039 1 .00 model d 9 10 1 1 12 13 14 total stock morice 8689 5213 5401 1459 1850 528 23140 babine 1436 2090 1 134 3641 1941 340 10582 sustut 4395 5126 4609 4262 3319 2573 24284 zymoetz . - - 390 3324 644 491 4849 kispiox - - 634 - 1306 - • 1940 by proportion 1 .00 morice .375 .225 .233 .063 .080 .023 babine .136 .198 .107 .344 .183 .032 1.00 sustut .181 .211 .189 .176 .137 .106 1 .00 zymoetz - .080 .685 .133 .101 1 .00 kispiox - - .327 - .673 — 1 .00 

Cite

Citation Scheme:

    

Usage Statistics

Country Views Downloads
China 10 62
United States 8 0
Russia 3 0
United Kingdom 2 0
Germany 1 2
Japan 1 0
City Views Downloads
Unknown 7 2
Shenzhen 5 62
Beijing 5 0
Ashburn 4 0
University Park 2 0
Redmond 1 0
Tokyo 1 0

{[{ mDataHeader[type] }]} {[{ month[type] }]} {[{ tData[type] }]}
Download Stats

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0096059/manifest

Comment

Related Items