Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Compass orientation in migrating Fraser River sockeye salmon Dat, Claire Germaine 1994

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1994-0338.pdf [ 1.98MB ]
Metadata
JSON: 831-1.0053154.json
JSON-LD: 831-1.0053154-ld.json
RDF/XML (Pretty): 831-1.0053154-rdf.xml
RDF/JSON: 831-1.0053154-rdf.json
Turtle: 831-1.0053154-turtle.txt
N-Triples: 831-1.0053154-rdf-ntriples.txt
Original Record: 831-1.0053154-source.json
Full Text
831-1.0053154-fulltext.txt
Citation
831-1.0053154.ris

Full Text

COMPASS ORIENTATION IN MIGRATING FRASER RIVER SOCKEYE SALMONByClaire Germaine DatB.H.A., York University, 1990A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF SCIENCEinTHE FACULTY OF GRADUATE STUDIESDEPARTMENT OF OCEANOGRAPHYWe accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAMay 1994© Claire Germaine Dat, 1994In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission._______________________________Department of_______________The University of British ColumbiaVancouver, CanadaDate IDE-6 (2/88)AbstractThree numerical models were used to investigate whether compass orientation is aneffective guidance mechanism for sockeye salmon migrating to the Fraser River from the openocean. Daily surface ocean currents, simulated by the Ocean Surface Currents Simulations(OSCURS) model, were used in each model to test the influence of currents on the return oceanicmigration of the Fraser River sockeye salmon. The high seas tagging and coastal recovery data ofthe Fraser River sockeye salmon were used for the migration simulations.The Fraser River sockeye salmon were shown to move in a northeastward directionduring the first phase of their oceanic return migration and in a southeastward direction during thesecond phase of their migration. The surface daily currents were shown to increase the speed ofthe homeward migrating sockeye salmon. Most importantly, compass orientation alone wasshown to be a possible orientation mechanism for the Fraser River sockeye salmon.11Table of contentsAbstract.Table of contents .List of Tables vList of Figures viiAcknowledgements xChapter 1 - Introduction 11.1 Objectives 11.2 Northeast Pacific ocean circulation 21.3 Overview of salmon migration patterns 41.3.1 Juveniles: seaward migration 41.3.2 At sea 41.3.3 Homeward to spawn 71.3.4 Northern Diversion Rate 81.3.5 Upstream Migration 91.4 Orientation mechanisms 101.5 Environmental influences 121.6 Migration models 16Chapter 2- Data and Models 202.1 Fish migration data 202.2 Ocean Surface Current Simulations (OSCURS) model 232.3 Fish migration models 25Chapter 3 - Results and Analysis 363.1 The oceanic distribution of the Fraser River sockeye salmon at tagging 363.2 Model 1 433.2.1 Spatial accuracy 433.2.2 Migration speed 46ifi3.2.3 Migration direction .513.2.4 Influence of the currents 553.3 Model 2 583.3.1 Migration speed 613.3.2 Migration direction 633.4 Model 3 653.4.1 Direction = 900 653.4.2 Direction = 450 653.4.3 Direction = 22.5° 69Chapter 4 - Discussion 734.1 Direction 734.2 Speed 744.3 Currents 754.4 Model 3 75Chapter 5 - Conclusion 775.1 Summary 775.2 Possibilities for further research 78Bibliography 79Appendix 1 83ivList of TablesTable 3.1Mean and standard deviation of the latitude and longitude of the salmon attagging 37Table 3.2Mean and standard deviation of the latitude and longitude of the salmon attagging by year 39Table 3.3Mean and standard deviation of the aiming error and of the latitude andlongitude of recovery of the sockeye in model 1 45Table 3.4Mean and standard deviation of the optimal speed of migration of the FraserRiver sockeye in model 1 47Table 3.5Mean and standard deviation of the optimal direction of migration of the FraserRiver sockeye in model 1 53Table 3.6Mean and standard deviation of the latitude and longitude of recovery of thesockeye in the absence of currents in model 1 56Table 3.7Mean and standard deviation of the speed of the sockeye migrating fromtagging to recovery location with and without daily currents in model 1 58Table 3.8Mean and standard deviation of the recovery latitude and longitude of thesalmon at the end of phase 1 and phase 2 in model 2 60Table 3.9Mean and standard deviation of the speed, direction and aiming error of theFraser River sockeye salmon in model 2 61Table 3.10Mean and standard deviation of the latitude and longitude of recovery at the endof phase 1 and phase 2 of the sockeye migrating at 20 krn/d in a 450 direction inphase 1 in model 3 66Table 3.11Mean and standard deviation of the speed, direction and aiming error of theFraser River sockeye salmon migrating in a 450 direction at 20 km/d inmodel 3 68vTable 3.12Mean and standard deviation of the latitude and longitude of recovery of thesockeye migrating at 20 knu/d in a 22.5° direction in phase 1 in model 3 69Table 3.13Mean and standard deviation of the speed, direction and aiming eor of theFraser River sockeye salmon migrating in a 22.5° direction at 20 km/din phase 1 in model 3 71viList of FiguresFigure 1.1Schematic diagram of surface circulation 3Figure 1.2Ocean migration of northeastern pacific sockeye salmon 6Figure 1.3Presumed distribution of mature Fraser River sockeye, obtained from oceantagging data between 1958 and 1983 7Figure 1.4Migratory routes of adult sockeye salmon returning to the Fraser River 9Figure 1.5Simulated sockeye migration paths, for 1982 and 1983 17Figure 2.1Tagging location of the Fraser River sockeye salmon 21Figure 2.2Recovery location of the tagged sockeye salmon 22Figure 2.3Grid of the OSCURS numerical model 23Figure 2.4Schematic diagram of the migration of the Fraser River sockeye in model 1 29Figure 2.5Contour plot of the aiming error for one salmon in model 1 30Figure 2.6Schematic diagram of the migration of the Fraser River sockeye in model 2 31Figure 2.7Contour plot of the aiming error for one salmon in model 2 32Figure 2.8Schematic diagram of the migration of the Fraser River sockeye in model 3 33Figure 2.9Contour plot of the aiming error for one sockeye migrating at 20 krn/d in a 45°direction in phase 1 in model 3 34Figure 2.10Contour plot of the aiming error for one sockeye migrating at 20 km/d in a22.5° direction in phase 1 in model 3 35vuFigure 3.1Distribution of sockeye salmon at tagging (depending on catch area) 37Figure 3.2Distribution of the Fraser River sockeye at tagging by year 38Figure 3.3Distribution of the Fraser River sockeye at tagging (models 2 and 3) 40Figure 3.4Distribution of the sockeye salmon at tagging depending on the number ofmigration days 41Figure 3.5Tagging location of the Fraser River sockeye salmon depending on their taggingdate 42Figure 3.6Start date versus migration days 43Figure 3.7Recovery of the Fraser River sockeye salmon in model 1 44Figure 3.8Frequency histogram of the aiming error of the sockeye salmon in model 1 46Figure 3.9Frequency histogram of the migration speed of the sockeye salmon in model 1.. 46Figure 3.10Migration days versus speed in model 1 48Figure 3.11Start date versus speed in model 1 48Figure 3.12Histograms of the speed of the sockeye depending on the number of daysspent migrating in model 1 49Figure 3.13Histograms of the speed of the sockeye depending on the start date of theirhomeward migration in model 1 50Figure 3.14Sockeye speed versus distance travelled in model 1 51Figure 3.15Frequency histogram of the migration direction of the of the Fraser Riversockeye in model 1 52Figure 3.16Direction angle versus number of migration days in model 1 53vmFigure 3.17Direction angle versus start date in model 1 54Figure 3.18Direction angle versus distance travelled in model 1 54Figure 3.19Direction angle versus speed in model 1 55Figure 3.20Distribution of the Fraser River sockeye salmon at recovery in the absence ofcurrents in model 1 57Figure 3.21Distribution of the Fraser River sockeye salmon at the end of phase 1 andphase2inmodel2 59Figure 3.22Speed versus number of migration days in model 2 62Figure 3.23Speed versus distance travelled in phase 1 in model 2 62Figure 3.24Speed versus distance travelled in phase 2 in model 2 63Figure 3.25Direction versus distance travelled in phase 1 in model 2 64Figure 3.26Direction versus distance travelled in phase 2 in model 2 64Figure 3.27Distribution of the Fraser River sockeye swimming at 20 km/d in a 45°direction in phase 1 in model 3 67Figure 3.28Distribution of the Fraser River sockeye swimming at 20 km/d in a 22.5°direction in phase 1 in model 3 70ixAcknowledgementsI would like to thank my supervisor Dr. Paul H. Leblond for his great advice and constantencouragement throughout this project. I also wish to thank the other members of my researchcommittee, Dr. Doug W. Oldenburg and Dr. Carl Walters.I acknowledge James Ingraham for allowing me to use the OSCURS model and theInternational North Pacific Fisheries Commission for granting me permission to use the taggingand recovery data of the Fraser River sockeye salmon.My biggest thanks go to Ian Jardine for his very valuable help, advice and patience duringthe preparation of this thesis.Thank you to Keith Thomson for all the discussions on sockeye migration.I wish to thank Deirclre and all those in my office : Ma, Debby, Renee, Ian and Tim forthe excellent company and support during the past two years.I am very grateful to two very special friends : Anou and Bill for their faithfullencouragement.I wish to thank my mom and dad very profoundly for all the support and confidence theygave me.Lastly, I wish to dedicate this thesis to my French grandmother, whom I will alwaysdearly miss.xChapter 1Introduction1.1 ObjectivesSalmon migration is a primary concern to British Columbia’s fisheries management. Thisstudy will concentrate on the species Oncorhynchus nerka, commonly called sockeye salmon.This species is the second most abundant in British Columbia waters as well as the most preferredcommercially due to the sockeye’s excellent flavoured, firm and attractive orange-red flesh (Grootand Margolis, 1991).Sockeye salmon migration has been extensively investigated because of the ability of thefish to home rapidly to their river of origin with very precise timing after spending two to fouryears in the open ocean. Sockeye salmon must thus possess some precise orientation abilities.However, these orientation mechanisms used by salmon during their oceanic migration are stilllargely controversial and unknown (Pearcy, 1992). Therefore in this study, three numericalmodels are used to investigate whether compass orientation (moving in a specific direction even inunfamiliar territory) is an effective guidance mechanism for sockeye salmon migrating to theFraser River from the open ocean, Daily surface currents are accounted for in each model to testthe influence of currents on the return oceanic migration of the Fraser River sockeye salmon.The following introduction will review the biology of sockeye salmon, their migrationpatterns, orientation mechanisms, the importance of environmental influences and models whichare pertinent to salmon migration.11.2 Northeast Pacific ocean circulationThe Northeast Pacific ocean circulation is dominated by the Alaska Gyre : a long-termmean cyclonic (counterclockwise) flow driven by cyclonic surface winds of the Aleutian Lowaround the Gulf of Alaska (Thomson et al., 1992). Both the Alaska Gyre and the Aleutian Lowdisplay interannual variability (Emery and Hamilton, 1985 ; Hsieh et al., 1991).The Subarctic Current (a broad, slow eastward drift with speeds up to 5 to 10 cm/s), theAlaska Current (a broad, slow northward drift with speeds of approximately 30 cm/s) and theAlaskan Stream (a narrow, southeastward boundary current with speeds greater than 100 cm/s)make up the Alaska Gyre (Thomson et al., 1992) (figure 1.1).The North Pacific Current (with speeds of 5 to 10 cmls) flows south of the SubarcticCurrent in an eastward direction. The Subarctic Boundary, just south of the Subarctic Current andnorth of the North Pacific Current at about 40°N, delineates the southern limit of salmondistribution in the North East Pacific (Favorite et al., 1976 ; Blackbourn, 1987 ; Quinn, 1990;Thomson et al., 1992). Recent studies by Welch et al. (1993) have also shown that each salmonspecies has a distinct southern boundary limited by temperature.2Figure: 1.1 Schematic diagram of surface circulation (From Dodimead et al., 1963).31.3 Overview of salmon migration patterns1.3.1 Juveniles : seaward migrationSockeye salmon spawn mostly from September to November. The females select thespawning site at the bottom of their home river, lay a batch of eggs subsequently fertilized by oneor more males. This process is repeated one to several times. After emergence, the fry migrate totheir feeding grounds. The length of their stay in the rearing lake depends on their growth rate andvaries between different stocks and years. The temperature of the water, the distribution of foodin the water column, the light and diseases are all factors that determine the growth of the fish(Groot and Margolis, 1991).In April or May, after having spent one year in freshwater and leaving the Fraser River,the juveniles migrate northward along the coast through the Strait of Georgia. Some fish aredirected westward across the Strait of Georgia by the Fraser River plume and tidal currents. Whenthey hit the Gulf Islands, these smolts migrate northwestward to join the others. It takesapproximately thirty days for the smolts to swim through the Strait of Georgia at an average speedof 6 to 7 km/day swimming 6 to 8 hours/day on a direct course (Healey and Groot, 1987). Byfall, all migrating fish have reached offshore waters.1.3.2 At seaFrench et al. (1976) used data collected by seine sampling in the Northeast Pacific tosummarize the oceanic migration of sockeye salmon.After entering the ocean, juveniles are found near the west coast of Vancouver Island andoff southeastern Alaska in late June. In July, they migrate northwestward in the Gulf of Alaskaremaining within 40 km off the coast. By August, by following the Alaska current, they haveprogressed off the central and southwestern coast of Alaska. The average rate of travel for Fraser4River sockeye during this coastal movement is approximately 18.5 km/day (Hartt and Dell,1986).The distribution of these fish in the fall is not well documented but according to theirlocation during the following winter, it is assumed that they must travel mostly southwestward(French et al., 1976). Age .1* fish are distributed across the North Pacific ocean (between 500and 45°N) in the winter. According to gilinet catches, in the spring, the sockeye have migratedthrough the Subarctic current system to their southernmost location (just south of 44°N). FromJune to August, these fish migrate northward. In the summer, they are found across the NorthPacific between 50° and 53°N. Their migration is in a westward direction as they approach theAleutian Islands. In the fall, they migrate southward to attain their winter distribution (figure 1.2).In the winter, maturing (i.e. starting their homing migration in the spring) age .2 fish arefound north of 49°N and east of l65°W whereas immature age .2 sockeye are displaced slightlysouth. In the spring the maturing age .2 fish commence their inshore migration and have left thehigh seas by the end of July. By winter their distribution overlaps that of the immature .1 and .2age fish but slightly north of the latter. By the end of the spring, most mature fish have left theoffshore waters for their spawning grounds. The immature age .2 sockeye show a broaddistribution throughout the Northeastern Pacific and have started a northward movement; they arefound from 150°W to 160°E and between 50°N to 53°N.Immature age.3 fish are distributed slightly northward of the previous year. By the end ofspring, they have started their homeward migration. A very small number of maturing age .4sockeye can be found the following year distributed as age .3 immatures and they migratehomeward in spring (figure 1.2).* The fish’s age is denoted by a point followed by a digit indicating the number of winters spent in saltwater.5Figure 1.2: Ocean migration of northeastern pacific sockeye salmon (From French et aL, 1976).61.3.3 Homeward to spawnMaturing salmon leave their feeding areas at population-specific times in spring andsummer and migrate rapidly to the coastal area near the mouth of their home river (sockeyeaverage 46-56 kmlday) (Quinn, 1989). Their homing migration can be divided into two phases:an ocean phase that carries the fish from offshore feeding grounds to the coast and a nearshorephase that carries the fish along the shore to their river (Hasler, 1966 ; Healey and Groot, 1987).During the fmal stretch of their ocean migration in May and June, maturing Fraser River sockeyeswim northeastwards towards the coast and then southeastwards along southeastern Alaska andthe Queen Charlotte Islands toward Vancouver Island (fig.1.3). According to Neave (1964), thispattern is meant to avoid an ocean area of high temperature (surface isotherm of 140 or 15°C)from which salmon of all species are almost absent after June.•4’k170• 1. 140._1I0•PmDonit abiSoaiiO°W•05..S2•4..44.40°1700 110° 1500 140° 130°WFigure 1.3 Presumed distribution of mature Fraser River sockeye, obtained from oceantagging data between 1958 and 1983 (From French et al., 1976, adaptedfrom Groot and Quinn, 1987).7Fraser River sockeye make landfall along the west coast of Vancouver Island or in QueenCharlotte Sound. According to Groot and Quinn (1987), the fish that make landfall on the upperpart of Vancouver Island and in Queen Charlotte Sound migrate to the Fraser River via JohnstoneStrait whereas the sockeye found along the southern part of the west coast of Vancouver Islandmigrate homeward through the Juan de Fuca Strait.1.3.4 Northern Diversion RateThe percentage of fish that migrate through the Johnstone Strait as opposed to the Juan deFuca Strait is called the Johnstone Strait Diversion or Northern Diversion Rate (hereafter NDR).According to the International Pacific Salmon Fisheries Commission (IPSFC), between 1953 and1977, the majority of salmon homed via the Juan de Fuca Strait (average of 84%) but from 1978to 1987, more Fraser River sockeye migrated via the Johnstone Strait (figure 1.4). Thefluctuations in the percentage of fish choosing the northern route over the southern route seems tobe due to oceanographic and climatological conditions. Indeed, after anomalously warm years,when the water temperature is higher off the British Columbia coast, a high percentage of fishmigrate through the Johnstone Strait (Blackbourn, 1987 ; Groot and Quinn, 1987 ; Healey andGroot, 1987 ; Wickett, 1977 ; Xie and Hsieh, 1989).810090 9080 eo70 70i-60 60k-z zw w50 5040 4O30 3020 20I0Figure 1.4: Migratory routes of adult sockeye salmon returning to the Fraser Riveraround Vancouver Island via the northern and southern routes. The bar graphindicates the proportion of the total run that used the northern route (adaptedfrom Groot et a!., 1984 and Healey and Groot, 1987).1.3.5 Upstream MigrationFraser River sockeye salmon migrate upstream to their home river or lake by travelling inschools along the sides and near the bottom of the Fraser River where the currents are slower.Run timing appears to be synchronized with the specific temperature regime of the home streamso that spawning will occur at an appropriate time for development and emergence of the alevin inspring (Miller and Brannon, 1982). Sockeye salmon rely on their energy reserves for theupstream migration, the maturation of their gonads, spawning and nest defence.Sockeye salmon undergo physiological changes when they approach their home river. Atthe beginning of their home journey, females and males appear identical : their body is silvery andthe top of their head blue-black. They have silvery white jaws and their silvery sides extend abovethe lateral line (across their body) and change to white on their belly (Groot and Margolis, 1991).PERCENT FRASER SOCKEYE USNG NORTHERN PASSAGE,eanSoAN,d56 58 60 62 64 66 68 70 72 74 76 78 80 82 84YEARS9the top of their head blue-black. They have silvery white jaws and their silvery sides extend abovethe lateral line (across their body) and change to white on their belly (Groot and Margolis, 1991),Near the home river, the female’s abdomen enlarges (because of the growth of the eggs),its gums recede around the teeth and its snout elongates. Its head becomes olive green, its backbright red and sides greyish black (Groot and Margolis, 1991).The male’s head becomes greenish, its back bright red and sides blackish-red with whiteventrally. The male’s body compresses and a hump of flesh forms anterior to the dorsal fin. Itssnout and upper jaw elongate (Groot and Margolis, 1991).1.4 Orientation mechanismsHealey and Groot (1987) found that juvenile sockeye are able to utilize compassorientation (move in a specific direction even in unfamiliar territory by means of celestial or otherreference cues) as fry during dispersal into lakes and as smolts during migration out of lakes. Itappears that these directional preferences are innate and involve the use of both the sun and theearth’s magnetic field. However Royce et al. (1968) pointed out that salmon migrate both duringthe day and night and often under cloudy skies ; thus it is highly doubtfull that sockeye smoltsrely only on the sun for orientation. Groot and Cooke (1987) found that Fraser River sockeyesmolts have a general directional tendency to move north after reaching salt water.No orientation mechanism is known for the one to four year oceanic motions of sockeyesalmon within the Gulf of Alaska. More work has been done on the return migration of sockeyesalmon, which is generally divided into three phases: an oceanic phase, a coastal phase and anupriver phase (Healey and Groot, 1987).During their homeward oceanic phase, Fraser River sockeye are thought to swimnortheastward and then southeastward before making landfall. The first part of this returnmigration was described by Thomson et al. (1992) as a non-directed phase during which thesockeye may swim randomly or with very weak orientation. However, the highly precise timing10of landfall at the end of the second phase implies salmon must have the ability to direct themselvesat a later stage. Moreover, their travel times are too rapid to be consistent with poor orientation,given their swim speeds and the ocean currents (Quinn and Groot, 1984).I shall now briefly review the results found and mechanisms deduced from variousexperiments conducted to elucidate the “mysteries” surrounding the orientation abilities of salmon.Groot and Cooke (1987) tested the hypothesis that the route chosen by adult sockeye returning tothe Fraser River might be determined by the pathway followed as seaward migrating juveniles.However, sockeye migrate seaward via the Strait of Georgia but homeward via either the northernor southern route around Vancouver Island. Thus adult sockeye do not need to rely on previousexperience to find their way back through coastal waters.By observing the horizontal and vertical movements of ultrasonically telemetered sockeyesalmon, Quinn, Terhart and Groot (1989) and Pascual and Quinn (1991) observed that when fishentered coastal waters and hit land, they did not follow the shoreline toward their home river butswam in the opposite direction (northwest), then turned southeast and finally turned round againto continue their homeward migration. Sockeye salmon also seem incapable of detecting watermass movements (i.e. they swam into the currents and not deeper to avoid them). Indeed, in theabsence of reference points, currents are not directly detectable by salmon.Royce et al. (1968) investigated the idea of salmon being guided by the earth’s magneticfield. Sea water is an electrical conductor moving through the earth’s magnetic field, thus theproduction of an electrical voltage can be expected. Because these voltages are directly related tothe current and are polarized with respect to its direction, electrical cues seem to be a possiblenavigational device for salmon on the high seas (Royce et al., 1968).Hasler and Wisby (1951) proposed that juvenile salmon learn odors of their freshwaterhabitat, are attracted to these odors as juveniles during their outward migration and retrace them asadults. Harden Jones (1978) proposed that salmon learn a sequence of odors as juveniles duringtheir outward migration and retrace them as adults. Juveniles imprint odors emanating from therocks, plants and soil in their home stream and when they return as adults, the exposure to these11imprinted odors stimulates their upstream migration. Healey and Groot (1987) also suggested thathomeward migrating adults may be directed upstream by population specific pheromone trails laiddown by juveniles on their way to sea.Thus outward migrating juveniles perform compass orientation by moving in anorthwestern direction (up the Strait of Georgia and through Johnstone and Queen CharlotteStraits) upon entering saltwater. Adults, on the other hand, have some knowledge of their goal(the breeding grounds they came from) and are able to locate it from either a northern or southernposition in response to oceanographic conditions (Groot and Cooke, 1987).1.5 Environmental influencesThree main oceanographic properties have been the topic of extensive research aspotentially important on salmon migration : the currents, the temperature of the water and itssalinity.Thomson et al. (1992) examined the influence of the current fields in the North Pacific onthe latitude of landfall and migration speed of sockeye salmon returning to the Fraser River. Theyhypothesized that the interannual variations in the Northeast Pacific circulation would account forthe variations in the NDR. Their study concentrated on two years : 1982 with a low NDR and1983 with a high NDR. The strong 1983 circulation deflected salmon to the north ; there was adifference of 400km between the latitude of landfall between the two years due to the currentfields. The latitude of landfall also depended on the start location the further south the fish startedtheir migration the more they were deflected by currents. Also the slower the swim speed and theearlier the start date of their migration, the greater the difference in latitudes between 1983 and1982. The results supported their hypothesis and Thomson et a!. (1992) found that the ttmeandifference in landfall between the two years was of sufficient magnitude to suggest that theinfluence of ocean currents accounts for at least a portion of the NDR”.12In a second paper, Thomson et al. (1993) tested the influence of ocean currents on thereturn timing of Fraser River sockeye salmon. Again they focused on 1982 and 1983 with a weakand strong (respectively) circulation pattern. Thomson et al. (1993) divided the return oceanicmigration of the sockeye into three phases: a non-directed and a directed oceanic phase and adirected coastal phase. In the first two phases, the ocean currents were found to affect the fishdirectly by advecting it ; in the last phase, the ocean currents indirectly affected the fish byinfluencing the latitude of landfall along the outer coast of Vancouver Island and in QueenCharlotte Sound. Thomson et a!. concluded that the interannual variability of the Northeast Pacificsurface ocean currents is sufficient to affect the return timing of Fraser River sockeye. Howeverthe magnitude of these effects depend upon pre-migration position, swim speed, compassorientation and migration start date.Hamilton and Mysak (1986) also studied the effects of the interannual variability in oceancurrents on the homeward migration of sockeye salmon. They concentrated on the Sitka eddywhich is a large (—300 km in diameter) clockwise rotating vortex. It is only observed in someyears and is centered near 57°N, 1380W. The currents in the eddy can exceed 0.5 rn/s which isthe typical swim speed of an adult sockeye. Hamilton and Mysak (1986) hypothesized that thepresence of the eddy could significantly affect the migration route of homeward bound sockeyesalmon. By comparing catch data from 1957 when the eddy was absent and 1958 when it waspresent , Hamilton and Mysak concluded that the Sitka eddy significantly deflected the sockeyesalmon swimming through it to the south or southwest. However their analysis remained entirelyqualitative and their conclusions were not based on trajectory calculations.Thomson et al. (in preparation for submission) simulated the migration of Fraser Riversockeye migration swimming through the Sitka eddy. They concluded that the eddy did notdeflect the sockeye and that any deflection that may occur would have to be accounted for by theinterannual variability in the Alaska Stream.Blackbourn (1987) presented a temperature displacement model, discussed in detail in thenext section. He divided the oceanic homeward migration of the Fraser River sockeye into two13parts and found that the position of the salmon prior to stage 2 is determined by temperature. Thesecond phase of salmon migration usually begins in the spring when the temperature of the waterdepends on the previous winter: after a warm winter, the optimum range of temperature for thesalmon will be further north than after a cold one and the fish will be found further north, thusfurther from the Fraser River, delaying their return timing. After cold winters, the salmon will bedisplaced south and thus return early to the coast. For salmon returning northward to theirhomestream, the opposite relationship held: after a cold winter, salmon arrived late and early aftera warm winter. Blackbourn tested this model by comparing sea surface temperature in the Gulf ofAlaska with the return timing of various stocks of Fraser River sockeye from 8 to 31 years ofdata. His results were consistent with the predictions and he concluded that the sea surfacetemperature in the Gulf of Alaska is significantly statistically related to adult run timing.Changes in temperature also affect the NOR. Hamilton (1985) used commercial catch datato determine the NDR from 1906 to 1952. He found a strong correlation between the NDRpercentages and the temperature changes in the coastal waters off the British Columbia coast.Furthermore, according to him, the NDR rates are better correlated with long period temperaturetrends than with the temperature at the time of the return migration. Therefore as discussedpreviously with Blackboum’s work, the water temperature that influence the return migration ofthe sockeye salmon may be that which occurs while the salmon is still in his feeding stage in theGulf of Alaska and not so much the temperature the fish encounters during its return migration.Groot and Quinn (1987) found that during warm years when the salmon were displaced tothe north in the Gulf of Alaska, they tended to make landfall in Queen Charlotte Sound and alongthe north side of Vancouver Island, thus migrating to the Fraser River through the JohnstoneStrait (a high NDR). During cold years, sockeye made landfall along the west side of VancouverIsland, migrating through Juan de Fuca Strait (a low NOR).Temperature is thus a determining oceanographic feature in the landfall timing and choiceof route of the Fraser River sockeye. Salinity is another feature which seems to have some14influence on the sockeye salmon’s homeward migration. However its impact is not as clear andleads to more debate among researchers.Favorite (1961) observed the surface conditions off the Washington and British Columbiacoasts during 1958 and 1959 to determine “if any significant differences occurred during thesetwo years that could have affected the 1958 migration of the salmon (through Johnstone Straitrather than Juan de Fuca Strait)”. Temperature could have been a significant factor but Favorite(1961) also hypothesized that the change might have been due to the unusual extent of the FraserRiver runoff flowing out Queen Charlotte Sound and Juan de Fuca Strait. Any definite conclusionseemed impossible due to the lack of knowledge on threshold recognition of salinity differencesby sockeye salmon (Groot and Quinn, 1987). Nevertheless, the NDR from 1953 to 1977 washighly correlated with the springtime Fraser River discharge, more so than temperature, whichmay not be so surprising considering temperatures during these years were cooler and thus maynot have influenced the migration route of the sockeye.Xie and Hsieh (1989) developed a numerical model to predict the migration routes of theFraser River sockeye salmon. Xie and Hsieh used a nonlinear regression model for the NDR. Themonthly average sea surface temperature at Kains Island (50026’ N, 128°2W) and the monthlyaverage Fraser River runoff (measured at Hope, B.C., 49°23’ N, 121027’ W) were the physicalvariables used to correlate with the diversion. As the temperature and runoff have very differentmagnitudes, the authors divided the physical variables by their respective means beforeperforming a regression analysis. When they then drew a scatter plot of the diversion and theMarch temperature data from 1953 to 1987, they found that the scatter was independent oftemperature for temperatures less than 8.5°C whereas for temperatures greater than 8.5°C, thediversion steeply increased with temperature. Thus to model this nonlinear behaviour, theyintroduced the river runoff in the regression with the diversion by a stepwise multiple regressionmethod. A systematic procedure was then used to determine which terms were statisticallysignificant enough to be retained in the regression formula. This procedure eliminated the termsthat failed to meet the significance criterion. This elimination procedure was iterated until15convergence to a final regression formula (Xie and Hsieh, 1989). The authors then verified thevalidity of the model by using data from 1955-78 to make predictions for 1979-88. The latterwere slightly below observed values, but the high NDR rate (80%) of 1983 was predictedaccurately.The currents, the temperature and the salinity of the Northeast Pacific waters have all beencorrelated to some degree with the sockeye’s homeward migration behaviour. However, theextent to which salmon rely on these three variables has not yet been determined and it is highlydoubtfull that such changes could account wholly for the homeward route chosen by adultsockeye.1.6 Migration modelsThomson et al. (1992) used the Ocean Surface CURrents Simulations (OSCURS) model,which is an empirical model developped to examine the variability of surface drift in the NorthPacific Ocean and Bering Sea (Ingraham and Miyahara, 1988). It computes surface currents as thevector sum of climatological geostrophic currents and daily surface wind drift with a spatialresolution of about 85 km and a time step of one day (Ingraham and Miyahara, 1988). Thomsonet al. (1992) simulated salmon migration paths for two years: 1982, with a weak Alaska Gyrecirculation and low NDR and 1983, with a strong circulation and high NDR. Twenty sevendifferent behavioural scenarios of sockeye migration were chosen. The authors chose reasonablelow, medium and high swim speeds, three start dates : May 1, June 1 and July 1 and compassorientations of 900T (east), 1 12.5°T (east, southeast) and 135°T (southeast). To proceed withthese simulations, they seeded the OSCURS model with an array of active drifters to simulateswimming compass-oriented sockeye (Figure 1.5). This model enabled them to deduce that “theinterannual variability of the Northeast Pacific Ocean circulation affects the latitude of landfall andmigration speed of adult sockeye salmon returning to the Fraser River” (Thomson et al., 1992).16NI.N15011 14011Figure 1.5 : Simulated sockeye migration paths, for 1982 and 1983, with a swim speed of20.8 cm/s. a compass orientation of 900T, and a migration start of May 1.The map projection is polar stereographic. The large dots indicate the startlocations. The small dots along the paths are the daily positions. The mid-sized dots along the paths are the positions at the beginning of indicatedmonths.Blackbourn (1987) set up a temperature displacement model solely based on oceantemperature variations in the Gulf of Alaska. He assumed that the Fraser River sockeye salmonstay within an optimum range of temperature which changes with the seasons: northward (ornortheastward) in spring and summer and in the reverse direction in fall and winter. This optimumtemperature range also shows interannual variability, thus the position of the salmon will changeNN15011 1401117from year to year. The author assumed two migration phases: “1) a relatively slow non-directedmigration occupying the first several months in response to seasonal changes in surfacetemperature or in the boundaries of ocean domains, 2) later, a relatively rapid, directed migrationtoward the home stream, largely independent of the boundaries of ocean parameters, and lastingno more than two months”. The position of the salmon prior to phase 2 is determined by theposition of the optimum temperature at that time which is determined by conditions in the priorwinter. This simple and straight forward model enabled Blackbourn to make importantconclusions (discussed in the previous section) on the return timing of the Fraser River sockeye.The last models I will describe are the ones developped by Pascual and Quinn (1991) to“investigate whether the preference of a compass direction is an effective guidance mechanism forsalmon in geographically complex coastal areas”.These models are based on a grid representing area covering the mouth of the FraserRiver, the Strait of Georgia and Johnstone and Queen Charlotte Straits (123 to l29°W and 49 to51°20’N) and a set of rules for the sockeye movements. The authors used ultrasonic tracking datafor their values of direction and speed in the model (Pascual and Quinn, 1991). Four differentstrategies of fish migration were modelled. In the first strategy, the fish had no direction ; this wasto determine whether they could migrate homeward randomly. In the second strategy, thedirection was added to the fish trajectory by modelling its heading at each time interval as arandom deviate from a normal distribution. For the third strategy, the fish could change directionat each time interval and finally in the last case, the fish randomly swam as in the first model butwhen encountering land, it milled around for a while before migrating again. All the fish startedalong the same line at the mouth of Queen Charlotte Strait, but their start position along that linewas randomly determined. Their goal was a vertical boundary 45 km west of the Fraser Rivermouth and they were allowed to migrate during four weeks to make it to their goal before beingconsidered lost. To render these models feasible,the authors also chose to make the followingassumptions : the fish movements were only two dimensional, the direction and speed of the fishwere considered independent parameters and the movements of each individual fish were18considered to be unaffected by the movements of other fish in the proximity. The authorsconcluded that compass orientation was not a sufficient orientation mechanism for the sockeye inthe last phase of their homeward migration.The work accomplished to date emphasizes the importance of oceanographic conditions inthe oceanic homeward migration route chosen by Fraser River sockeye salmon. However, theoceanic orientation mechanisms utilized by homeward migrating salmon have yet to bedetermined. Thus the objective of this thesis is to investigate the oceanic orientation skills ofhomeward migrating Fraser River sockeye salmon.After a presentation of the data and numerical models in chapter 2, the work is divided intothree sections. Chapter 3 describes and analyzes the results given by the models, chapter 4encompasses a discussion of these findings and chapter 5 states the conclusions and possibilitiesfor further research.19Chapter 2Data and Models2.1 Fish migration dataThe data used in this project consists of the open ocean tagging and coastal recovery dates,latitude and longitude of homeward migrating adult Fraser River sockeye. The tagging wasconducted from May 1st through to August 31st by Canada, the United States and Japan between1961 and 1968 (figure 2.1). The sockeye were all recovered by August 31st of the same year theywere tagged (figure 2.2). These data were made available by the International North PacificFisheries Commission (Appendix 1).20+ Fraser River. Juan de Fuca StraitFigure 2.1: Tagging location of the Fraser River sockeye salmon.Symbols refer to areas where the salmon were caught, as defined infigure 2.2.x Queen Charlotte Islands Strait of Georgia21Figure 2.2 Recovery location of the tagged sockeye salmon.Queen Charlotte Islands222.2 Ocean Surface Current Simulations (OSCURS) modelThe Ocean Surface CURrent simulations (OSCURS) model simulates the surface currentsin the North East Pacific from 1946 to the present. OSCURS computes the current vector field asthe sum of the geostrophic and wind drift components using an orthogonal grid (40x 104) with aspatial resolution of about 85 km and a daily time step (Ingraham and Miyahara, 1988). However,in this study, only the eastern half of the OSCURS model was necessary thus reducing the grid to40x50 (figure 2.3).150E 165E 180W 165W 150W 135W 120W 105W 90W/K4..,--Ic45 1’ j J\/// r30/ ___________—_._ _.— —I___7_/I. . IIFigure 2.3 : Grid (83-95 1cm) of the OSCURS numericalmodel (40*50).23The surface geostrophic current vector components were obtained by Ingraham bycalculating the geopotential anomalies (from 3000m to the surface) from the long-term meantemperatures and salinities of Bauer and Robinson (1985). The resulting circulation pattern of thislong-term mean flow is consistent with the descriptions of Favorite et al. (1976) and Reed andSchumacher (1987), with the exception of the Alaskan Stream. Low flow speeds were calculatedfor the Alaskan Stream due to its narrow width compared with the relatively large grid spacing.The grid size limitation of the model was compensated for by standardizing the geostrophicvelocity at specified grid points (Ingraham and Miyahara, 1989).The wind-induced surface drift, generally the largest component of the surface velocity,was computed by Ingraham as follows : the daily sea level pressure data from the U.S. NavyFleet Numerical Oceanography Center on their standard 380km Northern Hemisphere grid wasinterpolated to fit the OSCURS model grid points. These data were used to calculate the surfacewind vector field and wind-induced surface current vectors from an empirical function of the windspeed and direction described in Ingraham and Miyahara (1989).Thus a relatively simple summation of the geostrophic and wind drift componentsproduces a daily representative total current vector field at any date (from 1946 on) and point onthe model grid. These data were made available by James Ingraham.The Fraser River sockeye salmon swim in the upper lOm of the water column, thus it isreasonable to assume that they will only be affected by the surface circulation. However, theOSCURS model only simulates the geostrophic currents and winds and not anomalous oceansurface currents such as mesoscale eddies. Hamilton and Mysak (1986) studied the effect of theSitka eddy on homeward migrating sockeye salmon and found that the eddy significantlydeflected the sockeye swimming through it to the south and southwest. On the other hand,Thomson et al. (in preparation for submission), simulated the migration of Fraser River sockeyeswimming through the same eddy and found that the eddy did not deflect the fish, but rather thatmigrating salmon would “benefit from the eddy currents, with reduced metabolic costs fromadvection in the direction of migration”.24Thus it is reasonable to assume that surface geostrophic currents and winds of theNortheast Pacific are sufficient to account for the effects of the ocean circulation on the FraserRiver sockeye salmon migrations.2.3 Fish migration modelsThe three models I developped for fish migration simulate the oceanic return migration ofFraser River sockeye salmon tagged in the North East Pacific (figure 2.1) and recovered in theQueen Charlotte Islands, in the Strait of Georgia, in the Juan de Fuca Strait and in the FraserRiver (figure 2.2).The models compute the sockeye’s displacement as the vector sum of fish movements andcurrents determined by OSCURS with a time step of one day. Given the longitude and latitude ofthe fish, the four closest current vectors are interpolated to calculate the current vector componentsat the fish’s position. The models calculate the fish’s daily displacement and repeat this processfor a given number of travelling days starting at each new daily position. The high seas taggingdata used in these models gives us the longitude, latitude and date of tagging and recovery of thefish, thus the number of travelling days. The goal of these models is to find the optimal speed anddirection of the sockeye in order for them to home to a given position in a fixed number ofmigration days.Quinn, Terhart and Groot (1989) and Pascual and Quinn (1991) observed that whensalmon entered coastal waters and hit land, they did not follow the shoreline toward their homeriver but swam in the opposite direction (northwest), then turned southeast and finally turnedround again to continue their homeward migration. Thus, when the sockeye encountered land(i.e., when the four closest current vectors were equal to 0), they were made to turn around : wereredirected south (-90° direction, due east), migrated in this direction for one day, andsubsequently started migrating in the direction they were originally swimming in. However, aswill be seen in model 3 in chapter 3, some salmon found themselves caught between the Queen25Charlotte Islands and the British Columbian coast and were unable to swim out. Since the numberof sockeye to which this happened was very small, their migration simulations were consideredunsuccessfull and these fish were discarded in the analysis of the results.Coastal currents are poorly defined and highly variable (Thomson et al., 1992) ; theOSCURS model does not include such variability, except as directly caused by the winds.Therefore the sockeye recovered in the Queen Charlotte Islands were given an effective landfallposition just off the Queen Charlotte Islands (53°OO’N, 133000’W) ; similarly the sockeyerecovered in the Straits of Georgia and Juan de Fuca were assumed to have aimed at the mouth ofthese straits (51°OO’N, 129°OO’W and 48°30’N,125°30’W respectively).The majority of the sockeye recovered in the Straits of Georgia and Juan de Fuca werecaught close enough to the mouth of the straits (figure 2.3) that the distance between their aimed-at latitude and longitude and their recovery latitude and longitude was minimal, thus it wasassumed that the number of days spent travelling would not be affected significantly by thechange in the distance covered by the salmon. This relocation of the aimed-at latitude andlongitude was also assumed to have no significant effect on the choice of oceanic return migrationdirection and speed. Indeed the Fraser River sockeye must have aimed for the same latitude andlongitude during their homeward migration as during their simulated migration, since they had toswim through the mouths of the Straits of Georgia and Juan de Fuca to get to their recoverylocation and since both straits have very narrow mouths (27 km for the Strait of Georgia and24km for the Juan de Fuca Strait). The Queen Charlotte sockeye’s aimed-at latitude and longitudewere located on the west side of the Queen Charlotte Islands half-way up the coast, thus similarlyfor these fish the slight change in aim was assumed to have no effect on their homeward migrationpath.As for the sockeye recovered in the Fraser River, the model simulated their migration tothe mouth of both the Strait of Georgia, north end Queen Charlotte Strait and the Juan de FucaStrait, since their choice of route around Vancouver Island was unknown. Quinn (1988) estimatedthe average swim speed through coastal waters of a returning adult Fraser River sockeye to be 5826kmld. At this speed, the fish would spend 9 days migrating through the Strait of Georgia and 5through the Strait of Juan de Fuca. Thus, to account for the time spent by the sockeye migratingthrough the Straits of Georgia and Juan de Fuca, I respectively substracted 9 and 5 days from thetotal number of migration days.The three models, which are about to be described solve an inverse problem by forwardmodeffing.Model 1 simulates the Fraser River sockeye salmon’s migration assuming they migratehomeward with only one speed and one direction. The model simulates the daily displacement ofthe salmon as the vector sum of the fish movements and currents (figure 2.4). The magnitude ofthe fish movement vector is given by the fish’s speed and the direction of this vector is equal tothe direction given to the fish. The model simulates the migration of each fish for a time equal tomigration duration for a wide range of directions and speeds 55 speeds ranging from 0 to 88km/d in steps of 2km)d and 119 directions ranging from 45° to -86° due east in steps of 1.1°. Themodel therefore simulates 5355 migration scenarios for each fish. It then calculates the distancebetween the landfall position at the end of each migration and the point of landfall the fish wereaiming for. Model 1 results consists of the optimal speed and angle of migration (i.e. thecombination of both that enables the sockeye to make landfall with the greatest spatial accuracy),the latitude and longitude of the fish’s landfall position as well as the distance between the aimed-at position and the model landfall location for that optimal swim vector, hereafter referred to as theaiming error. Figure 2.5 shows a contour plot of the aiming error for all the angles and speeds forone particular sockeye recovered in the Strait of Georgia.The oceanic return migration of Fraser River sockeye salmon is sometimes believed to bedivided into two phases : one in which the salmon migrate randomly or with very weakorientation in a northeastward direction and a second one that lasts for two months in which thefish swim in a southeastward direction (French et al.,1976 ; Blackbourn, 1987 ; Thomson et al.,1992 ; Pascual and Quinn, 1991). Thus models 2 and 3 will simulate a two phase migration of theFraser River sockeye. The second phase of the oceanic return migration will last 60 days for the27salmon recovered in the Straits of Juan de Fuca and Georgia, 65 for the Fraser River sockeyemigrating through the Juan de Fuca Strait and 69 for the Fraser River salmon swimming throughthe Strait of Georgia. The first phase of the salmon’s migration lasts for the total number ofmigration days of each salmon minus 60 for the sockeye recovered in the Strait of Georgia andJuan de Fuca Strait, minus 65 for the Fraser River salmon migrating through the Juan de FucaStrait and minus 69 for the Fraser River fish swimming through the Strait of Georgia.In model 2, the sockeye are given no direction and no speed during phase 1, thus they areleft to drift with the currents (figure 2.6). For the second phase of the salmon’s return migration,model 2 runs the same number of migration simulations as model 1 with the same range of speedsand directions and similarly selects the optimal combination of angle and speed for the sockeye tohome with accurate timing as well as calculates the aiming error. Figure 2.7 shows a contour plotof the aiming error for one sockeye (same sockeye as in model 1) at all combinations of speedsand angles.Hartt (1966), French et al. (1976) and Groot and Quinn (1987) estimated the meanmigration speed of adult sockeye migrating from the open ocean to coastal waters to be 23 km/d.Also, according to laboratory studies of swimming energetics led by Brett (1983), swim speedsranging from 24 to 62 kmld were found to be relatively efficient. Thus I chose three migrationspeeds to represent reasonable low, medium and high speeds: 20 krn/d, 35 km/d and 55 kmld.The swim speed of the sockeye refers to the fish’s movement, regardless of the water’s motion(Thomson et al., 1992).In model 3, during the first phase of their homeward migration, the sockeye are giveneach one of the three swim speeds and three migration directions : 22.5° (northeast-east), 45°(northeast) and 90° (north) (figure 2.8). For the second phase of the salmon’s return migration,model 3 runs the same number of migration simulations as models 1 and 2 with the same range ofspeeds and directions and similarly selects the optimal combination of angle and speed for thesockeye to home with accurate timing as well as calculates the aiming error. Figure 2.9 shows acontour plot of the aiming error at all combinations of speeds and angles for one sockeye (the28same one as in models 2 and 3) migrating at 20 km/d in a 45° in phase 1 and figure 2.10 shows acontour plot of the aiming error for the same salmon migrating at 20 km/d in a 22.5° direction inphase 1.Models 2 and 3 results consists of the optimal speed and direction, the distance betweenthe aiming landfall position and the model landfall location, the latitude and longitude after the firstand second phase of the return migration.F = daily displacement of sockeye = current vector + sockeye vectorS = sockeye vector (optimal speed and direction of the sockeye)C = current vectorFigure 2.4: Schematic diagram of the migration of the Fraser Riversockeye in model 1,Sdayl ‘ day2model recoverypositionaiming errorlast day of migratidnaimed-atposition29I-85 —I/7IIIIII,III 111111o io 20 30 40 50 60 70 80speed (kmld)Figure 2.5 : Contour plot of the aiming error (km) for onesalmon in model 1.30Sç’4s model recoverypositionaiming errorday 1 day 2 day 1 day 2 last day POSitiOflof migrationphase 1 phase 2C = current vector = sockeye vector in phase 1= daily displacement of sockeye in phase 1S = sockeye vector in phase 2 (optimal direction and speed)F = daily displacement of sockeye in phase 2Figure 2.6: Schematic diagram of the migration of the Fraser River sockeye in model 2.31-85 —-68 —-51—a-—-17-• I/0-17—I I34—0 10 20 30 40 50 60 70 803peed mld)Figure 2.7 : Contour plot of the aiming error (1cm) for one salmon in model 2.32dayl day2’I IC = current vectorFl = sockeye vector in phase 1 (given fixed direction (22.5, 45 and 900) andspeed (20, 35 and 55 krnld))Si = daily displacement of sockeye in phase 1F2 = sockeye vector in phase 2 (optimal direction and speed)S2 = daily displacement of sockeye in phase 2Figure 2.8: Schematic diagram of the migration of the Fraser River sockeye in model 3.F--1(model recoveryNi \ positionaiming error\aimed-atdayl day!2positionphase 11last ay ofmigrationphase 233;o io 20 30 40 50 60 70 803peed (kmld)Figure 2.9 : Contour plot of the aiming error (kin) for a sockeyemigrating at 20 km/d in a 450 direction in phase 1in model 3.34E‘6’ -—-17-0 10 20 30 40 50 60 70 80speed (kmid)Figure 2.10 : Contour plot of the aiming error (1cm) for onesockeye migrating at 2Okm/d in a 22.5odirection in phase 1 in model 3.35Chapter 3Results and AnalysisThe object of this thesis is to examine the ability of the Fraser River sockeye salmon tohome by compass-orientation (i.e., to move in a particular compass direction even in unfamiliarterritory (Healey and Groot, 1987)). In this chapter, I use three models to test the hypothesis thatcompass orientation by itself is sufficient to ensure accurate homing. The results of the threemodels will be statistically analysed in the hope of elucidating the oceanic homeward migration ofFraser River sockeye salmon.3.1 The oceanic distribution of the Fraser River sockeye salmon at taggingDifferent stocks of Fraser River sockeye salmon occupy different feeding areas of theGulf of Alaska prior to their homeward migration (French et al., 1976). More specifically,Blackboum (1987) characterized the feeding areas of seven stocks of Fraser River sockeye. Thusdocumenting the distribution of the sockeye in the Northeast Pacific Ocean prior to theirhomeward migration could indicate what stocks the salmon belong to and therefore what river thesalmon originated from. The longitude and latitude of tagging of each salmon are given inAppendix 1. The mean longitude of the Fraser River sockeye at tagging is 145°37’W with astandard deviation equal to 6040? and the mean latitude is 51°54’N with a standard deviation of2°OT. The standard deviations are large showing a disparate distribution of the sockeye at tagging(table 3.1, figure 3.1). The mean and standard deviation of the latitude and longitude of tagging ofthe sockeye by year (table 3.2) show no systematic differences in the distribution of the fish(figure 3.2). Furthermore, the number of tagged Fraser River sockeye is limited (305) but also the36number of migration days varies indicating that the salmon were tagged at different stages of theirreturn migration preventing any hope of stock recognition.60 1’45 1165W 150W 135W 120WFigure 3.1: Distribution of sockeye salmon at tagging (depending on catch area).Area of catch Latitude LongitudeStrait of Georgia Mean 52°28’ N 145°47W108 2°05’ 6°04’Standard deviation (2321cm) (411 1cm)JuandeFucaStrait Mean 52°03’N 145°02’W53 1°50’ 5°36’Standard deviation (203 1cm) (383 km)Queen Charlotte Islands Mean 5307?N 137°1 1’W8 2026? 2°43’Standard deviation (270 1cm) (180 1cm)1Fraser River Mean 51021 N 46013?w68 2°06’ 7021?Standard deviation (2341cm) (510 km)Table 3.1 : Mean and standard deviation of the latitude and longitude ofthe salmon at tagging.Fras Riverj \376045Juan de Fuca Queen CharlotteC Georgia -I-. FraserFigure 3.2 : Distribution of the Fraser River sockeye at tagging by year.38Year Latitude Longitude1961 Mean 53°14’N 132°l6tW6 1°42’ 1°30’Standard deviation (189 km) (100 km)1962 Mean 52°25’N 146°l9tW48 1°41’ 6°33’Standard deviation (186 1cm) (444 1cm)1963 Mean 49°59’N 150°40W34 1°48’ 4°51’Standard deviation (201 km) (346 1cm)1964 Mean 50°59’N 147°4ltW23 2°25’ 4°04’Standard deviation (268 1cm) (285 km)1965 Mean 51°23’N 142°27’W10 2°50’ 3°06’Standard deviation (316 1cm) (216 1cm)1966 Mean 52°09’N 147°12’W129 2°08’ 5032’Standard deviation (238 km) (3771cm)1967 Mean 52°43’N 139°56’W54 1°29’ 4030?Standard deviation (166 km) (303 km)Table 3.2 : Mean and standard deviation of the latitude and longitude of theFraser River sockeye at tagging by year.Model 1 simulates the homeward migration of all the Fraser River sockeye, however inmodels 2 and 3, only the migration of the salmon travelling over 60, 65 and 69 days (seedescription of models 2 and 3 in chapter 2 for details) are simulated. Figure 3.3 shows thedistribution at tagging of the Fraser River sockeye used in models 2 and 3. As in model 1, nosignificant differences characterize the distribution of these fish in the North East Pacific prior totheir homeward migration.396045The distribution of the fish depending on the number of migration days as well as on theirtagging date was next investigated to see if the oceanic distribution of the Fraser River sockeyechanged through the season. Indeed, at the time of tagging, the mean and standard deviations ofthe latitude and longitude show that the distribution of the sockeye shifts to the northeast as thenumber of migration days decreases (figure 3.4) as well as when their start date advances towardsthe fall (figure 3.5). The shift in the mean latitude and longitude of the distribution of the FraserRiver sockeye as the number of migration days decreases and as the migration start date advancesis ifiustrated in figures 3.4 and 3.5. Furthermore the date of tagging and the number of migrationdays are negatively correlated by a coefficient equal to 0.8 (figure 3.6) the later the salmon wereFigure 3.3 : Distribution of the Fraser River sockeye at tagging (model 2 and 3).40tagged, the least time they spent migrating. The value of the correlation coefficient above whichthe null hypothesis that the correlation between the variables is zero is rejected at the 0.1% level is0.188, indicating a very strong correlation between these two variables. Thus the Fraser Riversockeye salmon seem to migrate from the open ocean to the West coast of North America in ageneral northeastward direction.Figure 3.4: Distribution of the sockeye salmon at tagging, depending onthe number of migration days and showing the mean and standarddeviation of the latitude and longitude at tagging.416045Figure 3.5 : Tagging location of the Fraser River sockeye salmon depending on theirtagging date and showing the mean and standard deviation of the latitudeand longitude at tagging.42120 1203.2 Model 1100806040200Figure 3.6: Start date versus migration days.100806040200Model 1 selects the optimal combination of speed and angle for each sockeye to home in agiven number of days with the most accuracy and calculates the distance between the longitudeand latitude the fish were aiming for and the longitude and latitude they arrived at (hereafter namedthe aiming error). The statistical analysis of the results will indicate the plausibility of thehypothesis.3.2.1 Spatial accuracyAt the end of their simulated homeward migration, the Fraser River sockeye are closelygathered at the mouth of the Straits of Georgia (north end: Queen Charlotte Strait) and Juan deFuca and in the Queen Charlotte Islands (figure 3.7). The mean latitude and longitude of recoveryof the salmon (table3.3) indicate how close the sockeye migrated to their goal : 5 1°N, 129°W forthe mouth of the Strait of Georgia, 48°30’N, 125°30’W for the mouth of the Juan de Fuca Strait0 20 40 60 80 100 120number of migration days43and 53°N, 133°W for the Queen Charlotte Islands. The standard deviations of these latitudes andlongitudes of recovery (table3.3) are small showing the density of the distribution of the salmonupon arrival.The mean aiming error is given in table 3.3. The mean ratio of this distance over the totaldistance travelled by the migrating sockeye is 0.0165, thus the aiming error of these fish can beconsidered minimal when compared with the length of their homeward migration, indicating theaccuracy of the aim. Furthermore, the width of the mouths of the Strait of Georgia (north end:Queen Charlotte Strait) and the Juan de Fuca Strait are 27km and 24km respectively. The meanaiming error for sockeye migrating through the Strait of Georgia and the Juan de Fuca Strait are19 and 18 km respectively (figure3.8). It therefore seems reasonable to assume that all theFigure 3.7 : Recovery of the Fraser River sockeye salmon in model 1.44homeward migrations simulated by model 1 would allow the sockeye to migrate to the FraserRiver through the appropriate strait around Vancouver Island, even in the presence of daily oceancurrents.Area of catch Latitude Longitude Aiming error (1cm)Straitof Georgia Mean 51°00’N 128°56’W 18108 0006’ 0016’Standard deviation (12 1cm) (18 km) 12Juan de Fuca Strait Mean 48°33’N 125°33’W 2153 0010’ 0015’Standard deviation (18 km) (18 Jçm) 17Fraser River to Juande Fuca Strait Mean 48°29’N 125°31W 1668 0°06’ 0011’Standard deviation (12 km) (14 Jcin) 12Fraser River to Straitof Georgia Mean 50°59’N 128°59’W 1968 0°06’ 0°17’Standard deviation (10 1cm) (20 1cm) 13Queen CharlotteIslands Mean 52°59’N 132°57’W 78 0°03’ 0°04’Standard deviation (51cm) (51cm) 4Table 3.3 : Mean and standard deviation of the aiming error and of the latitude andlongitude of recovery of the sockeye.45100fI3.2.2 Migration speed160-140-8) 120-. 100-0Cl)C60-40-20-0-I8060402005 15 25 35 45 55 65 75 85distance (km)Figure 3.8 : Frequency histogram of the aiming errorof the sockeye salmon in model 1.0The mean calculated optimal swim speed of the sockeye salmon is 21 km/d with manyswim speeds between 9 km/d and 32km/d (standard deviation equal to 11 kmld) (figure 3.9).III—160140-120Hi 00-806040-20IIII.I.I,II.5 15 25 35 45 55 65 75 85speed (kmld)Figure 3,9 : Frequency histogram of the migration speedof the sockeye salmon in model 1.46The mean migration speed of the sockeye varies from 14 km/d for the Queen Charlottefish to 23 km/d for the Juan de Fuca Strait salmon with standard deviations ranging from 9 to 15km/d indicating variability in the speed of the homeward migrating sockeye (table 3.4).The swim speed of the sockeye is related to the number of migration days by a correlationcoefficient of 0.6 (figure 3.10). The value of the correlation coefficient above which the nullhypothesis that the correlation between the variables is zero is rejected at the 0.1% level is 0.188,indicating a high correlation between the number of migration days and the speed. Figure 3.12shows that salmon migrating 0 to 30 days have swim speeds ranging from 0 to 70 km/d withmost salmon swimming between 20 and 50 kmld. The speed of the salmon slowly decreases asthe number of migration days increases and all salmon migrating for more then 90 days have aswim speed ranging from 0 to 20 kmld. Similarly, the sockeye’s swim speed is correlated withtheir tagging date (the correlation coefficient is equal to 0.5 and as previously this value shows astrong correlation between the start date and the speed) : the later the sockeye were tagged, thefaster their migration speeds (figure 3.11). Figure 3.13 shows that the majority of sockeye taggedin May had speeds ranging from 0 to 15 km/d, however in July the salmon’s speeds rangedmostly between 20 and 50 kmld.Area of catch Speed (km/d)Strait of Georgia Mean 18108 Standard deviation 12Juan de Fuca Strait Mean 2353 Standard deviation 15Fraser River to Juan de FucaStrait Mean 2168 Standard deviation 9Fraser River to Strait ofGeorgia Mean 2068 Standard deviation 9Queen Charlotte Islands Mean 148 Standard deviation 13Table 3.4: Mean and standard deviation of the optimal speed ofmigration of the Fraser River sockeye in model 1.47ioo -100y = 37.994 + -0.26667x R= 0.59434+80 80+60 + 604. +++40 40+ + +* ++*w, * 2020÷ 4 1- ÷ *+4. *0 00 20 40 60 80 100 120number of migration daysFigure 3.10 : Migration days versus speed in model 1.100 100y = 11.497 + 0.27691x R= 0.54821+80 80++60 60t++ + +++ 44.4020 204* 1-0o 20 40 60 80 100 120start date (1=may1.123ugust31)Figure 3.11: Start date versus speed in model 1.4861-74 migration days5 15 25 35 45 55 65speed (km/d)F84035302520151050302520I*151005 15 25 35 45 55 65 75 85speed (km/ti)75-90 migration days25 35 45 55 65 75 85speed (km/ti)Figure 3.12 : Histograms of the speed of the sockeye depending on the numberof days spent migrating in model 1.49Sspeed (1cm/ti)15-30 migration daysF8•1IIIIII-50-40-30E1L0U: -20•101:-05 15 25 35 45 55 65 75 85speed (kin/ti)2531-44 migration days25 2520 2091-160 migration days20151075 8545-60 migration daysa,I105-120 migration days10 1015 155 5(1 (15 155 15 25 35 45 55 65 75 85speed (km/d)Figure 3.13 : Histograms of the speed of the sockeye depending on the start date oftheir homeward migration in model 1.50May 1-15 July 1-1560 6050504030::.[ 402010i:0I L11105 15 25 35 45 55 65 75 85speed (km/d)May 16-31IISIiJuly 16-31-707060605040so soE 3020 2010 10n 0:i 41 1ij 12-1145 15 25 35 45 55 65 75 85speed (kn/76543201. 105 15 25 35 45 55 65 75 85speed (km/d)August 1-15. 2.52.521.5o5speed (km/d)21.50.5June 16-305 15 25 35 45 55 65 75 85speed (km/d)5 15 25 35 45 55 65 75 85speed (km/d)The value of the correlation coefficient above which the null hypothesis that the correlationbetween the speed of the Fraser River sockeye and the distance travelled by these fish during theiroceanic return migration is zero is rejected at the 10% level is 0.095 thus this correlation isextremely questionable and these two variables should not be considered correlated (figure 3.14).Therefore it is not how far the fish are from their goal that influences their choice of speedbut rather how much time they have to return and what the date at the time of tagging is.100 100y = 17.862 + 0.0023044x R= 0.09898880 80+60 60+ +40 +++*+++++++++ +++++++40+ +++ +++4++ 4++ ++ + *.+ +20__.__—4 + + 20++ + :ê4 4$t .%+ +4.++* + 4: + +* * +t +0 00 500 1000 1500 2000 2500distance (1cm)Figure 3.14: Sockeye speed versus distance travelled in model 1.3.2.3 Migration directionThe mean migration direction of the salmon is 240 with a standard deviation equal to 18°84% of the Fraser River sockeye have a migration direction ranging from 60 to 420 (figure3.15), where angles are defined as positive counterclockwise from due east.51140- -140120- -120. 100- -10080- -8060- -6040- -4020- -200-56 34 11 -11 -34 -56 -79angle (o)Figure 3.15 : Frequency histogram of the migration direction of theFraser River sockeye in model 1.The mean migration direction of the sockeye varies from -15° for the sockeye recovered inthe Fraser River and aiming for the mouth of the Strait of Georgia, to -32° for the Fraser Riversalmon aiming for the mouth of the Juan de Fuca Strait with standard deviations ranging from 12°(Juan de Fuca Strait) to 59° (Queen Charlotte Islands) (table 3.5). The standard deviation for thefish aiming for the Queen Charlotte Islands is very large in comparison with the other standarddeviations (table 3.5) which may be due to the very different aim these salmon have.The correlation coefficient between the number of migration days and the direction of thesockeye is 0.1 (figure 3.16). The value of the correlation coefficient above which the nullhypothesis that the correlation between the variables is zero is rejected at the 5% level is 0.113indicating a reasonable probability of the null hypothesis not being rejected. Thus no relationshipexists between these two variables. Similarly figure 3.17 shows that the tagging date of thesockeye and their direction are very poorly correlated, the correlation coefficient is equal to 0.2.However, the value of the correlation coefficient above which the null hypothesis that thecorrelation is zero is rejected at the 0.1% level is 0.188, showing these two variables to be slightlycorrelated.52Area of catch Angle (0)Strait of Georgia Mean -21108Standard deviation 15Juan de Fuca Strait Mean -2853Standard deviation 12Fraser River to Juan deFuca Strait Mean -3268Standard deviation 17Fraser River to Strait ofGeorgia Mean -1568Standard deviation 16Queen Charlotte Islands Mean -218Standard deviation 59Table 3.5 : Mean and standard deviation of the optimal direction ofmigration of the Fraser River sockeye in model 1.57 - 57—_---y=-498.73+ 1.4032x R=O.1148429 ÷ 29** 4:*++4• ++ +0- * + ++4 *‘-+*** 4, * * + -++ I- ++ +-29+ ÷*—*•::*-.‘ 14 :-::-29+÷ ÷: : ÷+: + +***++++++ ++ ++ 57+ ++ +—86- :::I:..I.,.I...I:..I..:-860 20 40 60 80 100 120number of migration daFigure 3.16 : Direction angle versus number of migration days in model 1.53a0 20 40 60 80 100 120start date (1..mayl.123=august3l)Figure 3.17 : Direction angle versus start date in model 1.However, the distance travelled by the sockeye and their angle of direction are correlatedwith a small coefficient equal to 0.3 (figure 3.18): as the distance travelled by the fish increases,the angle of direction also increases. The value of the correlation coefficient above which the nullhypothesis that the correlation between the variables is zero is rejected at the 0.1% level is 0.188implying a very high probability that the null hypothesis will be rejected and thus a smallcorrelation between these variables.a5729-29-57-860 500 1000 1500 2000 2500d,stance (km)Figure 3.18 : Direction angle versus distance travelled in model 1.54On the other hand, figure 3.19 shows that the speed and the direction of the salmon aretwo absolutely non correlated variables (the correlation coefficient is equal to 0 and the value ofthe correlation coefficient above which the null hypothesis that the correlation is zero is rejected atthe 50% level is 0.039).3.2.4 Influence of the currentsThe latitude and longitude of landfall of the sockeye were determined in the absence ofcurrents using the optimal migration angle and speed calculated in model 1 to estimate theinfluence of the currents on the Fraser River sockeye’s homeward migration. The mean latitudeand longitude of recovery of the fish in the absence of currents are generally south of thosecalculated in the presence of currents (table 3.6). The standard deviations vary from 115km to57290-29-57-8657290-29-57-860 20 40 60 80speed (km/d)Figure 3.19 Direction angle versus speed in model 1.10055245km for the latitude and from 111km to 638km for the longitude indicating a very disparatedistribution of the fish upon arrival (table 3.6).The recovery latitude and longitude calculated in this case are far from the aimed-at latitudeand longitude (figure 3.20). The distance travelled by the salmon is shorter and the latter aredeflected to a more southerly latitude._______________Latitude LongitudeStrait of Georgia Mean 52°26’N 132°13’W108 2005’ 3004’Standard deviation (2321cm) (208 km)JuandeFucaStrait Mean 46°50’N 127°53’W53 2°11’ 8°23’Standard deviation (243 km) (638 km)Fraser River toJuandeFucaStrait Mean 46°06’N 132°21’W68 1007’ 3049’Standard deviation (124 km) (279 1cm)Fraser River toStrait of Georgia Mean 49°00’ 130°14’W68 1°02’ 3035’Standard deviation (115 km) (261 1cm)Queen CharlotteIslands Mean 52°07’ 134°27’W8 2°12’ 1°38’Standard deviation (245 1cm) (111 1cm)Table 3.6: Mean and standard deviation of the latitude and longitude of recoveryof the sockeye in the absence of currents in model 1.The speed of the Fraser River sockeye if they migrated from their tagging location to therecovery latitude and longitude determined by model 1 in the absence of currents is greater than inthe presence of currents. Indeed, in the presence of daily currents, the mean speed of the sockeyeranged from 14 km/d for the Queen Charlotte sockeye to 23 km/d for the Juan de Fuca Strait fish,whereas in the absence of currents the mean speed varies from 19 km/d for the Queen Charlottesalmon to 28 km/d for the Strait of Georgia and Fraser River to the Juan de Fuca Strait salmon56(table 3.7). The currents of the Gulf of Alaska deflect the sockeye towards the northeast and thesalmon must, to some extent, swim with the currents. These results clearly show the importanceof currents for a compass-oriented navigator.45Figure 3.20 : Distribution of the Fraser River sockeye salmon at recoveryin the absence of currents in model 1.6057Area of catch Speed without currents Speed with currents(kmld) (kmld)Strait of Georgia Mean 28 21108Standard deviation 14 12Juan de Fuca Strait Mean 27 2353Standard deviation 11 15Fraser River toJuan de Fuca Strait Mean 28 2168Standard deviation 10 9Fraser River toGeorgia Strait Mean 27 2068Standard deviation 10 9Queen CharlotteIslands Mean 19 148Standard deviation 12 13Table 3.7 : Mean and standard deviation of the speed of the sockeye migrating from taggingto recovery location with and without daily currents in model 1.3.3 Model 2Model 2 divides the return oceanic migration of the Fraser River sockeye salmon into 2phases the first one, which lasts the total number of migration days minus 60 days and in whichthe sockeye are not given a speed nor a direction (i.e., drift with the currents), the second phase inwhich model 3 selects the optimal combination of speed and angle for each sockeye to homethrough the appropriate strait around Vancouver Island and calculates the aiming error. All the fishexcept for 2 Strait of Georgia sockeye completed their oceanic migration under these conditions.Figure 3.21 illustrates the distribution of the Fraser River sockeye salmon at the end ofphase 1 and phase 2. The mean latitude and longitude range from 51°48’N and 147°1’TW for theFraser River sockeye swimming to the mouth of the Strait of Georgia to 5301 1’N and 145°14Wfor the Juan de Fuca Strait salmon.586045Distribution of sockeye after Distribution of sockeye afterphase 1 phase 2Figure 3.21 : Distribution of the Fraser River sockeye salmonat the end of phase 1 and phase 2 in model 2.59At the end of their simulated return migration, the sockeye salmon are closely gathered atboth ends of Vancouver Island (figure 3.21). The mean latitude and longitude of recovery of thesalmon and the standard deviations (table 3.8) show how close the fish migrated to their goal.Catch area Latitude 1 Longitude 1 Latitude 2 Longitude 2Strait of Georgia Mean 52°39’N 147°14’W 51°00’N 129°07’W53/55 1°58’ 4044’ 0012’ 0°18’Standard deviation (218 km) (319 km) (22 1cm) (221cm)JuandeFucaStrait Mean 53°11’N 145°14’W 48°33’N 125°41’W38/38 1°53’ 3057’ 0011’ 0°20’Standard deviation (209 kin) (263 1cm) (21 kin) (25 1cm)Fraser River toJuan de Fuca Strait Mean 51°57’N 146°58’W 48°31’N 125°34W35/35 2025’ 4019’ 0°10’ 0023’Standard deviation (269 1cm) (296 1cm) (18 1cm) (29 1cm)Fraser River toStraitof Georgia Mean 51°48’N 147°17’W 50°52’N 129°13’W32/32 2°25’ 4°16’ 0°20’ 0°29’Standard deviation (269 kin) (2941cm) (371cm) (34km)Table 3.8 : Mean and standard deviation of the recovery latitude and longitude of the salmonat the end of phase 1 and phase 2 in model 2.The mean and standard deviations of the recovery latitude and longitude and the meanaiming errors (table 3.9) are larger in this model than in model 1, however it still seemsreasonable to assume that the salmon are recovered close enough to the mouth of the Straits ofGeorgia, north end: Queen Charlotte Strait and Juan de Fuca to be able to reach their spawninggrounds with accurate timing and in the presence of daily ocean currents.60Catch area Angle (0) Speed (kmld) Aiming error(km)Strait of Georgia Mean -20 15 1553/53Standard deviation 18 4 19Juan de Fuca Strait Mean -32 19 3238/38Standard deviation 12 4 15Fraser River toJuandeFucaStrait Mean -24 18 3035/35Standard deviation 14 4 17Fraser River toStrait of Georgia Mean -13 12 4632/32Standard deviation 20 4 28Table 3.9 : Mean and standard deviation of the speed, direction and aiming error ofthe Fraser River sockeye salmon in model 2.3.3.1 Migration speedThe mean calculated optimal migration speed of the sockeye salmon varies from 12 km/dfor the Fraser River sockeye aiming towards the mouth of the Strait of Georgia to 19 kmld for theJuan de Fuca Strait salmon (table 3.9). The standard deviations are 4 km for all the fish thusindicating a small variability in the migration speeds (table 3.15).Unlike in model 1, the swim speed of the sockeye is only very slightly correlated with thenumber of migration days by a coefficient equal to 0.2 (figure 3.22). The number of migrationdays in phase 2 is 60, 65 or 69 depending on the recovery location of the fish (as described inchapter 2), however it varies in phase 1, thus this slight correlation tells us that as the number ofmigration days in phase 1 increases, the swim speed of the salmon also increases. The value ofthe correlation coefficient above which the null hypothesis that the correlation between the numberof migration days and the swim speed is zero is rejected at the 1% level is 0.208 indicating a veryprobable but weak correlation between these two variables.6130 30Speed versus number of migrationdays in model 2.I0 100 200 300 41))distance travelled in phase 1 (inn)Figure 3.23 : Speed versus distance travelledin phase 1 in model 2.252015I105.0-—y = 8.9034 + 0.086325x R= 0.22133 ++ a +1- + ++ ++++ +4-ta+ + + + + ++ 4-4—4 +4-4252015105050 60 70 80 90number of migeation daysFigure 3.22:100 110 120The distance travelled in phase 1 is also only slightly correlated with the optimal calculatedspeed of the salmon in phase 2. The correlation coefficient is equal to 0.2 and the value of thecorrelation coefficient above which the null hypothesis that the correlation between the variables isrejected at the 2% level is 0.189 indicating a probable correlation. As the distance increases, theswim speed of the salmon also increases (figure 3.23).I302520151050— y = 14371 + 0.00g7853x R= 0.20602+ + +++ +4 + + + + + ++4- + + +4 ++ + 0 + ++ 4 + + +++ + +++++ ++ +25201510550062However, the migration speed and the distance travelled in phase 2 are highly correlated(coefficient equal to 0.9) and the value of the correlation coefficient above which the nullhypothesis that the correlation between the variables is rejected at the 0.1% level is 0.264. Thus asthe distance travelled increases, the speed of the salmon increases as well (figure 3.24).0 500 1000 1500 2000 2500dialance Iravelled in phase 2 (km)Figure 3.24: Speed versus distance travelledin phase 2 in model 2.3.3.2 Migration directionThe mean calculated optimal direction of the Fraser River sockeye varies between 130 and-32° (table 3.9). Absolutely no correlation was found between the number of migration days, thestart date, the speed of the sockeye in phase 1 and the optimal direction of the fish in phase 2.However, the direction of the Fraser River sockeye in phase 2 is correlated with thedistance travelled by the fish in phase 1 by a correlation coefficient equal to 0.2. The value of thecorrelation coefficient above which the null hypothesis that the correlation between the distancetravelled in phase 1 and the migration direction in phase 2 is zero is rejected at the 0.5% level is0.227. Thus the correlation is definite even though it is weak (figure 3.25).6323 * . 23—y .508.73 + 0.65849x R 0.23094÷++**+** + * 0-23÷,-23: +÷i *-46 *-46÷.69 690 100 200 300 400 500distance travelled in phase 1(kin)Figure 3.25 : Direction versus distance travelledin phase 1 in model 2.The optimal migration direction is also correlated with the distance travelled in phase 2:the greater the distance travelled, the greater the angle of migration of the salmon (figure 3.26).The correlation coefficient is equal to 0.3 and the value of the correlation coefficient above whichthe null hypothesis that the correlation between these two variables is zero is rejected at the 0.1%is 0.264 indicating a strong correlation.23 • 23—y = -842.14 +4).3236x R 0.34868+ * ++0 + 5 + S •00 500 1000 1500 2000 2500distance travelled in phase 2 (kin)Figure 3.26 : Direction versus distance travelledin phase 2 in model 2.643.4 Model 3Model 3 divides the homeward migration of the sockeye into two phases the first one inwhich the sockeye are given compass orientations of 900 (north), 450 (northeast) and 22.5°(north-northeast) and swim speeds of 20, 35 and 55 km/d and a second phase in which model 2selects the optimal combination of speed and angle for each sockeye to home through theappropriate strait around Vancouver Island and calculates the distance between the longitude andlatitude the fish were aiming for and the longitude and latitude they arrived at (i.e., the aimingerror).Model 3 simulates nine homeward oceanic migration scenarios for the Fraser Riversockeye. The following three sections will describe the homeward oceanic migration of the FraserRiver sockeye when given each direction in phase 1.3.4.1 Direction = 900The majority of the Fraser River sockeye make landfall on the Alaskan coast at the end ofphase 1 when given a northern direction (90°) and the greater the swim speed of the fish, thegreater the number of them found on the Alaskan coast prior to phase 2 (see chapter 2 for detailedexplanation). The reality of such a migration behaviour seems highly doubtfull since Fraser Riversockeye salmon are believed to make landfall around Vancouver Island and in the Queen CharlotteIslands prior to their coastal migration and not as far north as the Alaskan coast. Thus a 90°northern direction in phase 1 for the Fraser River sockeye migrating homeward was ruled out.3.4.2 Direction = 450146 of the 160 sockeye, whose homeward oceanic migration were simulated in model 3,completed their migration when given a direction of 45° (northeast) and a speed of 20 kmld: 14of the fish were “lost’t, i.e. they got stranded along the coast in phase 2 (most likely in the QueenCharlotte Islands) and were not recovered at the end of their second phase of migration (seechapter 2 for detailed explanation). Similarly, 130 were recovered at the end of phase 2 whenmigrating at 35 krn/d in phase 1 and 91 were recovered at the end of phase 2 when migrating at 55km/d in phase 1. The number of sockeye to complete their homeward migration with a speedequal to 20 km/d in phase 1 is greater than in the other two cases and these fish were also the onesto accomplish their homeward migration with the most accuracy, i.e., to be recovered the closestto their goal, so I chose to describe their homeward migration in more detail.65Figure 3.27 shows the distribution of the Fraser River sockeye at the end of phase 1 and 2when swimming at 20 km/d in phase 1. The mean latitude of distribution of the sockeye prior tophase 2 ranges from 56°10’N for the Juan de Fuca Strait sockeye to 54°41’N for Fraser Riversalmon aiming to the mouth of the Strait of Georgia (table 3.10). The mean longitude varies from140°26W for the Juan de Fuca fish to 143°30’W for the Strait of Georgia salmon (table 3.10).The standard deviations (table 3.10) of the latitude and longitude of distribution of the fish at theend of phase 1 are large indicating a disparate disthbution of the Fraser River salmon at this stageof their homeward migration (figure 3.27). However, at the end of their oceanic migration, all thesockeye are recovered very close to their goal (latitude 2 and longitude 2 in table 3.10) and thestandard deviations of these latitudes and longitudes are small (table 3.10) showing the density ofthe fish at the mouth of the Straits of Georgia and Juan de Fuca (figure 3.27).Catch area Latitude 1 Longitude 1 Latitude 2 Longitude 2Strait of Georgia Mean 55°18’N 143°30’W 50°56’N 129°08’W49/55 2°40’ 4°51’ 0°36’ 0°21’Standard deviation (297 km) (307 km) (66 km) (25 km)Juan de Fuca Strait Mean 56°10’N 140°26’W 48°29’N 125°26’W36/38 1°56’ 4°10’ 0024’ 0017’Standard deviation (215 km) (258 km) (44 km) (21 km)Fraser River toJuandeFucaStrait Mean 55°05’N 142°31’W 48°32’N 125°32’W34/35 2°26’ 3047’ 0°14’ 0°17’Standard deviation (270 1cm) (240 km) (27 1cm) (21 1cm)Fraser River toStrait of Georgia Mean 54°41’N 142°13’W 51°00’N 129°04W27/32 2°19’ 3°48’ 0°20’ 0°14’Standard deviation (258 kin) (244 km) (38 km) (16 kin)Table 3.10: Mean and standard deviation of the latitude and longitude of recovery at the endof phase 1 and phase 2 of the sockeye migrating at 20 km/d in a 450 direction inphase 1 in model 3.666045Distribution of sockeye afterphase 1Distribution of sockeye afterphase2Figure 3.27 : Distribution of the Fraser River sockeye swimming at 20 km/dwith a northeastern direction (450) in phase 1 in model 3.67The mean aiming errors for the fish migrating at 20 km/d range from 31 to 38 km/d (table3.11). The standard deviations seem reasonably small (table 3.11) which allows us to assume thatthe salmon could migrate through the appropriate strait around Vancouver Island. These aimingerrors are the smallest for the sockeye swimming at any of the three speeds in a northeasterndirection (45°) in phase 1, thus not all the salmon migrating at 35 and 55 km/d in phase 1 wouldbe able to migrate to their spawning grounds with accurate timing.The mean direction of migration of the salmon in phase 2 ranges from 430 for the FraserRiver sockeye aiming to the mouth of the Strait of Georgia to -57° for the Juan de Fuca Straitsalmon (table 3.11). The mean speed of these fish varies from 10 to 19 km/d with small standarddeviations, indicating a very narrow range of speeds (table 3.11). These mean speeds are,however, all smaller than 20 km/d implying that the Fraser River sockeye migrate faster in phase1 than in phase 2. The differences between the speeds in phase 1 and 2 for the fish migrating at 35and 55 km/d in phase 1 are greater.Catch area Angle () Speed (kmld) Aiming error(1cm)Strait of Georgia Mean -46 14 3749/55Standard deviation 25 3 59Juan de Fuca Strait Mean -57 19 3836/38Standard deviation 15 3 30Fraser River toJuan de Fuca Strait Mean -48 17 3134/35Standard deviation 18 3 15Fraser River toStrait of Georgia Mean -43 10 3327/32Standard deviation 24 3 23Table 3.11: Mean and standard deviation of the speed, direction and aiming errorof the Fraser River sockeye salmon migrating in a northeastern directionat 20 km/d in a 45° direction in phase 1 in model 3.As will be discussed in chapter 4, the occurrence of the salmon migrating at greater speedsin phase 1 than in phase 2 is very unlikely. Therefore, Fraser River sockeye most likely do notmigrate in a northeastern direction (45°) at 20, 35 or 55 km/d in phase 1 when migratinghomeward from their ocean feeding grounds.683.4.3 Direction = 22.5°143 of the 160 Fraser River sockeye that migrated at 20 km/d and in a northeast-easterndirection (22.50) in phase 1 were recovered at the end of phase 2. However only 97 of the salmonswimming at 35 km/d and 51 of those migrating at 55 km/d in phase 1 were recovered at the endof phase 2. Not only is the number of sockeye who completed their simulated oceanic migrationgreater when the fish migrated at 20 km/d in phase 1 but also these sockeye were recovered theclosest to their goal, therefore I will concentrate my analysis on these migration simulations.Figure 3.28 shows the distribution of the Fraser River sockeye at the end of phase 1 and 2when swimming at 20 km/d in phase 1. The Juan de Fuca Strait salmon have the most northerlyand easterly centre of distribution prior to phase 2 (table 3.12) and the Fraser River sockeyeaiming to the north end of Vancouver Island have the most southerly mean latitude and the Straitof Georgia salmon the most westerly mean longitude (table 3.12). All the salmon are recoveredvery close to their respective goals at the end of their oceanic migration (figure 3.28, table 3.12).The standard deviations are small (table 3.12) indicating the proximity of all the salmon to theirgoal.Catch area Latitude 1 Longitude 1 Latitude 2 Longitude 2Strait of Georgia Mean 53°57’N 142021’W 50°56’N 129°07’W46/55 2°19’ 4°44’ 0015’ 0016’Standard deviation (257 km) (310 km) (27 km) (19 km)JuandeFucaStrait Mean 54°30’N 138°09’W 48°35’N 125°31W36/38 1°48’ 5015’ 0°26’ 0°12’Standard deviation (200 km) (339 km) (42 km) (15 km)Fraser River toJuandeFucaStrait Mean 53°47’N 140°11W 48°33’N 125°32W35/35 2°15’ 3054’ 0°12’ 0°16’Standard deviation (250 km) (256 km) (22 km) (20 km)Fraser River toStrait of Georgia Mean 53°29’N 141°22’W 50°57’N 129°10’W26/32 2°23’ 3°06’ 0°22’ 0°26’Standard deviation (264 km) (205 1cm) (40 km) (31 1cm)Table 3.12 : Mean and standard deviation of the latitude and longitude of recovery of thesockeye migrating at 20 km/d in a 22.5° direction in phase 1 in model 3.696045Distribution of sockeye afterphase 1Distribution of sockeye afterphase 2Figure 3.28 : Distribution of the Fraser River sockeye swimming at 20 kmldin a northeast-eastern direction (22.5°) in phase 1 in model 3.70The mean aiming errors for these Fraser River sockeye range from 31 to 16 km (table3.13) and the standard deviations of the latitude and longitude of recovery of the fish (table 3.13)are small allowing us to conclude that these fish would have been able to migrate to theirspawning grounds with accurate timing. These aiming errors are the smallest for the sockeyeswimming at any of the three speeds in a northeast-eastern direction (22.5°) in phase 1: not all thesalmon swimming at 35 and 55 km/d in phase 1 would have been able to be at their spawninggrounds in time.The mean direction of migration of the salmon in phase 2 varies from -56° for the Juan deFuca Strait salmon to -42° for the Strait of Georgia fish (table 3.13). The mean calculated optimalspeed ranges from 8 km/d for the Fraser River salmon aiming towards the mouth of the Strait ofGeorgia to 16 krnid for the Juan de Fuca Strait fish (table 3.13). The standard deviations are equalto 3km/d for all groups of sockeye (table 3.13). As in the previous section, the mean migrationspeeds of the Fraser River sockeye are smaller in phase 2 than in phase 1, which seems unrealisticas will be discussed in the next chapter. The difference between the migration speeds of phase 1and 2 for salmon migrating at 35 and 55 kni/d in phase 1 are greater than for the sockeyeswimming at 20 kmld in phase 1.Catch area Angle (°) Speed (kmld) Aiming error(1(m)Strait of Georgia Mean -42 12 2846/55Standard deviation 29 3 18Juan de Fuca Strait Mean -56 16 3136/3 8Standard deviation 16 3 33Fraser River toJuan de Fuca Strait Mean -50 14 2535/35Standard deviation 22 3 16Fraser River toStrait of Georgia Mean -43 8 1626/32Standard deviation 31 3 32Table 3.13 : Mean and standard deviation of the speed, direction and aiming errorof the Fraser River sockeye salmon migrating in a 22.5° direction at20 km/d in phase 1 in model 3.Even though these simulations would enable the Fraser River sockeye to migrate to theirspawning grounds, this pattern of migration seems highly unlikely, since too many of the salmon71made landfall along the Alaskan coast at the end of phase 1 and their swim speeds were greater inphase 1 than in phase 2. Indeed, we found that the salmon’s swim speed decreased as theirnumber of migration days increased, and increased the later their migration start date, indicatingthat the salmon’s speed increased as they got closer to their goal and not decreased.Chapter 4 will gather all the information these results have given us and discuss whatthese models might tell us on the ability of the Fraser River sockeye to migrate by compassorientation.72Chapter 4DiscussionThe results obtained from the three numerical models will be discussed in this chapter toinvestigate whether compass orientation by itself is a sufficient return oceanic migrationmechanism for the Fraser River sockeye salmon.The first two sections will concentrate on the information we can deduce from the speedand direction of the salmon during their modelled homeward migration. The third section willdiscuss the effect of the currents on the sockeye salmon. Lastly, the migration simulations ofmodel 3 will be examined.4.1 DirectionOnly one combination of speed and angle gives the optimal recovery latitude and longitudeof each Fraser River sockeye in models 1 and 2. Thus, even though in models 1 and 2 all theFraser River sockeye salmon were able to home with one direction and one speed (i.e., compassorient), the choice of direction varied between fish (i.e., not one single specific directionrepresented the choice of all salmon). This indicates that compass orientation by itself is probablynot a sufficient migration mechanism for the Fraser River sockeye salmon ; nevertheless,combined with a direction selection mechanism, which would somehow dictate to the fish whatdirection to migrate in, it could be a satisfactory explanation. Pearcy (1992) proposed that a senseof location based on oceanographic features such as temperature and salinity combined withcompass orientation could be a sufficient explanation for the ability of sockeye salmon to homewith such geographic and timing accuracy. Blackboum (1987) found that different stocks ofFraser River sockeye were found in different areas of the North East Pacific, which indicates thatstock-specific compass orientation could also be a sufficient explanation for the migration abilitiesof Fraser River sockeye.However, even though only one combination of speed and angle gives the optimalrecovery location, a range of different combinations of angles and speeds give close enoughrecovery points (within the 30 km contour of figures 2.5 and 2.7, as examples of one sockeye inmodels 1 and 2), which would enable the fish to home with accurate timing. The mean calculatedoptimal direction of the Fraser River sockeye in model 1 is -24° with a standard deviation equal to7318° and in model 2, it is -22° with a standard deviation equal to 17°. Tn both models, all theoptimal directions are within a narrow range (± 18° in model 1 and ± 17° in model 2). Thisallows us to conclude that there is a considerable overlap in the ranges of angles that enable eachfish to reach their goal. Therefore, it seems reasonable to assume that all Fraser River sockeyewould be able to home with a similar migration direction, indicating that compass orientation mayafter all be a sufficient migration mechanism for the Fraser River sockeye salmon.4.2 SpeedThe mean calculated optimal swim speed of the Fraser River sockeye salmon in model 1 is21 km/d with a standard deviation equal to 11 km/d and in model 2, it is equal to 16 km/d with astandard deviation of 4 km/d.Figures 3.10 and 3.11 allowed us to conclude that the optimal calculated migration speedof the Fraser River sockeye salmon was strongly correlated with their start date and the number ofmigration days. Indeed as the salmon started their oceanic migration later in the season, theirspeed increased as well as when the number of days spent migrating decreased. Thus in model 2,the range of optimal speeds is extremely narrow (±4 km/d) due to the fact that all the fish migratefor 60, 65 and 69 days during the second phase of their migration.Therefore, we just found that the Fraser River sockeye salmon could migrate home withapproximately the same direction, but that their speed depended on the time they had left tomigrate. Thus, if the Fraser River sockeye have knowledge of the time they have left untilspawning (which could be indicated to them by morphological changes) then we could concludethat compass orientation is a sufficient oceanic return orientation mechanism.Quinn and Groot (1984) estimated that maturing sockeye tagged in the open ocean oftentravel 40 to 60 km/d. In both these models, the sockeye swim at speeds slower than expected, butare, however, recovered close enough to their goal to be able to migrate to their spawninggrounds with accurate timing. In model 2, in which the salmon only actively started migrating(i.e., started compass-orientating) 60, 65 and 69 days (for the Strait of Georgia and Juan de FucaStrait salmon, for the Fraser River sockeye swimming through the Juan de Fuca Strait and for theFraser River salmon aiming towards the mouth of the Strait of Georgia , respectively) prior totheir recovery date, their migration speed is slower than in model 1 in which the fish orientedduring their entire oceanic return migration. This allows us to conclude that the Fraser Riversockeye do not need to swim as fast as thought of by Quinn and Groot (1984) and also thatpassive drifting (i.e., being “carried” by the currents : the salmon have neither speed nor directionof their own) is less energy demanding during the first part of the salmon’s oceanic returnmigration than compass orientation started at the beginning of the salmon’s homeward migration.74Homeward migrating sockeye drifting with the currents are displaced in a northeastwarddirection, which is the direction sockeye salmon are believed to follow during the first part of theirhomeward oceanic migration (French et al., 1976 ; Blackbourn, 1987 ; Pascual and Quinn, 1991;Thomson et al., 1992). The distribution of the Fraser River sockeye at tagging depending on theirnumber of migration days (figure 3.4) and on their start date (figure 3.5) confirmed this belief.4.3 CurrentsThomson et al. (1992) found that the currents assisted the shoreward migration of sockeyesalmon starting south of 550N. The majority of the tagged Fraser River sockeye salmon used forthe migration simulations of the three models were distributed south of this latitude (appendix 1)at the start of their homeward migration. In model 1, the sockeye started compass-orienting attheir tagging latitude and longitude and, indeed, their speed was greater in the absence of dailycurrents than in their presence, indicating that the currents did assist the fish in their homewardmigration and thus their importance in the modelling of the salmon’s homeward oceanicmigration.4.4 Model 3In model 3, the Fraser River sockeye salmon migrated in one of three fixed directions (90,45 and 22.5° due east) and three speeds (20, 35 and 55 km/d) in phase 1 and compass orientedwith the best combination of speed and direction during phase 2. The speeds were chosen torepresent reasonable slow, medium and high swim speeds for sockeye salmon.Fraser River sockeye salmon are believed to move northeastward in May and June andsoutheastward in July and August towards Vancouver Island during their homeward oceanicmigration (French et al., 1976). The salmon then make landfall along the west coast of VancouverIsland and north along the British Columbian coast prior to their coastal migration. In themigration simulations of model 3, the sockeye do migrate northeastward in the first phase of theirhomeward oceanic migration and southeastward in the second phase. However, an importantnumber of the fish make landfall along the Alaskan coast prior to phase 2 (the more northern theirdirection in phase 1, the greater the number of sockeye found on the Alaskan coast) and themajority of them also have slower swim speeds in phase 2 than in phase 1 (the greater their swimspeed during phase 1, the greater the difference between swim speeds in phase 1 and 2). Little isknown about the ocean phase of sockeye salmon migration except that it is rapid, well directedand well-timed (Royce et a!., 1968 ; French et al., 1976 ; Groot and Quinn, 1987). Thus the75likelyhood of the Fraser River sockeye salmon making landfall on the Alaskan coast, millingaround while waiting for the right time to start the last phase of their return migration andswimming slower as they approach the coast seems contradictory with the general beliefs onsockeye migration. These simulations also assume the salmon to be able to navigate since theyhave to change their compass-orientation between phase 1 and 2.Since models 1 and 2 have shown to simulate reasonably realistic migration paths for theFraser River sockeye salmon assuming less orientation abilities on the part of the fish than model3, we conclude that the simulations of model 3 are highly questionable and demand more of thesalmon in their ability to orient than is necessary.76Chapter 5Conclusion5.1 SummaryCompass orientation was tested as a mechanism for Fraser River sockeye salmon tomigrate home from the open ocean to the mouths of the Strait of Georgia and of the Juan de FucaStrait. The oceanic migration of the salmon was simulated for sockeye tagged in the open oceanand recovered around Vancouver Island, accounting for daily surface currents, modelled by theOSCURS model. Three migration scenarios were simulated : a first one in which the salmonmigrated with only one direction and one speed, a second one in which the fish were given nodirection and no speed (i.e., drifted with the currents) during the first phase of their oceanicmigration and migrated with one direction and one speed during the second phase and finally athird one in which the salmon migrated in one of three fixed directions and one of three fixedspeeds during the first phase of their oceanic migration and migrated with just one direction andone speed in the second phase.The results are summarized as follows:- The belief that the Fraser River sockeye salmon migrate in a northeastward directionduring the first phase of their homeward oceanic migration and then in a southeastward directionwas confirmed by the shift to the northeast of the center of distribution of the sockeye at taggingas the fish were tagged later in the season and as the number of migration days decreased.- The influence of the currents was shown to be important and thus should be accountedfor. Indeed the daily surface currents increased the migration speed of the Fraser River sockeyeand deflected the latter to the southeast.- Most importantly, compass orientation was shown to enable Fraser River sockeyesalmon to home with accurate timing if the Fraser River sockeye have knowledge of the time leftuntil spawning to determine their migration speed.775.2 Possibilities for further researchThe ability of Fraser River sockeye salmon to recognize the time they have left untilspawning should be further investigated. This would allow us to conclude whether compassorientation alone is a sufficient oceanic orientation mechanism for the Fraser River sockeye.To further enhance our knowledge of orientation mechanisms in sockeye salmon, theinfluence of other environmental parameters such as temperature, salinity and food availabilityshould be included in the modelling of ocean migration simulations, since these parameters arebelieved to affect the migration of sockeye salmon.New numerical models including other environmental parameters to determine what theFraser River sockeye salmon and other salmonids do in the open ocean and where they are wouldbe of a great benefit to fisheries research and fisheries management.78BibliographyBauer, R., and Robinson, M, (1985) Description of the Bauer-Robinson numerical atlas,version VIII, February 1985. Compass Systems, Inc., p. 13.Blackbourn, D. J. (1987) Sea surface temperature and the pre-season prediction of return timingin Fraser River sockeye salmon (Oncorhynchus nerka). Can. Sp. Pub!. Fish. Aquat. Sci,96: 296-306.Brett, J. R. (1983) Life energetics of sockeye salmon, Oncorhynchus nerka. In: BehaviouralEnergetics: The Cost ofSurvival in Vertebrates. W. P. Aspey and S. I. Lustick (eds.)Columbus: Ohio State University Press, pp. 29-33.Dodimead, A. J., Favorite, F., and Hirano, T. (1963) Salmon of the North Pacfic Ocean-PartII. Review of Oceanography of the Subarctic Pacific Region. mt. North Pac. Fish.Comm. Bull. 13:195.Emery, W. J., and Hamilton, K. (1985) Atmospheric forcing of interannual variability in theNortheast Pacific Ocean: connections with El Nino. J. Geop. Res. 90(C1):857-868.Favorite, F. (1961) Surface temperatures and salinity off the Washington and British Columbiacoasts, August 1958 and 1959. J. Fish. Res. Board Can. 18:311-3 19.Favorite, F., Dodimead, A. J., and Nasu, K. (1976) Oceanography of the Subarctic PacificRegion, 1962-72. mt. N. Pac. Fish. Comm. Bull. 33:187.French, R., Bilton, H., Osako, M., and Hartt, A. (1976) Distribution and origin of sockeyesalmon (Oncorhynchus nerka) in offshore waters of the North Pacific Ocean. mt. N. Pac.Fish. Com. Bull. 34:113.Groot, C., and Cooke, K. (1987) Are the migrations of juvenile and adult Fraser River sockeyesalmon (Oncorhynchus nerka in near-shore waters related? Can. Sp. Publ. Fish.Aquat. Sci. 96:53-60.Groot, C., and Quinn, T. P. (1987) The homing migration of sockeye salmon to the FraserRiver. Fish. Bull. 85(3) :455-469.Groot, C., and Margolis, L. (1991) Pacific Salmon Life Histories. University of BritishColumbia Press, Vancouver. p.564.Hamilton, K. (1985) A study of the variability of the return migration route of Fraser Riversockeye salmon (Oncorhynchus nerka). Can. J. Zoo!. 63:1930-1943.79Hamilton, K., and Mysak, L. A. (1986) Possible effects of the Sitka eddy on sockeye(Oncorhynchus nerka) and pink salmon (Oncorhynchus gorbuscha) migration offsoutheast Alaska. Can. J. Fish. Aquat. Sci. 43:498-504.Harden Jones, F. R. (1968) Fish migration. London: Arnold, p. 325.Hartt, A. C., (1966) Migrations of salmon in the North Pacific Ocean and Bering Sea asdetermined by seining and tagging, 1959-1960. mt. North Pac. Fish. Comm. Bull.19: 14 1.Hartt, A. C., and Dell, M. B. (1986) Early oceanic migrations and growth of juvenile Pacificsalmon and steelhead trout. mt. North Pac. Fish. Comm. Bull. 46:105.Hasler, A. D., and Wisby, W. J. (1951) Discrimination of stream odors by fishes and relation toparent stream behavior. American Naturalist 85:223-238.Hasler, A. D. (1966) Underwater guideposts. University of Wisconsin Press, Madison,Wisconsin, U.S.A..Healey, M. C., and Groot, C. (1987) Marine migration and orientation of ocean-type chinook andsockeye salmon. Am. Fish. Soc. Sym. 1:298-3 12.Hsieh, W.W., Lee, W.G., and Mysak, L.A. (1991) Using a numerical model of the NortheastPacific Ocean to study the interannual variability of the Fraser River sockeye salmon(Oncorhynchus nerka). Can. J. Fish. Aquat. Sci. 48:623-630.Ingraham, W. J., and Miyahara, R. K. (1988) Ocean surface current simulations in the NorthPacific Ocean and Bering Sea (OSCURSQ Numerical Model). NOAA TechnicalMemorandum NMFS S/NWC-130, p. 155.Ingraham, W. J., and Miyahara, R. K. (1989) Tuning the OSCURS numerical model to oceansurface current measurements in the Gulf of Alaska. NOAA Technical MemorandumNMFS F/NWC-168, p. 67.International Pacific Salmon Fisheries Commission (IPSFC) 1954-1986. Annual Reports for1953- 1985.Miller, R.J., and Brannon, E.L. (1982) The origin and development of life history patterns inPacific salmonids. In Salmon and Trout Migratory Behavior Symposium, ed. E.L.Brannon and E.O. Salo (School of Fisheries, University of Washington, Seattle), p.296-309.Neave, F. (1964) Ocean migrations of Pacific salmon. J. Fish. Res. Board Can. 21:1227-1244.Pascual, M.A., and Quinn, T.P. (1991) Evaluation of alternative models of coastal migration ofadult Fraser River sockeye salmon (Oncorhynchus nerka). Can. J. Fish. Aqat. Sci.48:799-810.Pearcy, W.G. (1992) Ocean ecology of North Pacific Salmonids. Washington Sea Grant.University of Washington Press, Seattle. p. 179.80Quinn, T, P. (1988) Estimated swimming speeds of migrating adult sockeye salmon. Can. J.Zool. 66(10):2160-2163.Quinn, T. P. (1990) Current controversies in the study of salmon homing. Ethology, Ecologyand Evolution 2:49-63.Quinn, T. P., and Groot, C. (1984) Pacific salmon (Oncorhynchus) migrations : orientationversus random movement. Can. J. Fish. Aquat. Sci. 41:1319-1324.Quinn, T. P., Terhart, B. A., and Groot, C. (1989) Migratory orientation and vertical movementsof homing adult sockeye salmon, Oncorhynchus nerka, in coastal waters. Anim. Behav.37:587-599.Reed, R. K., and Schumacher, 3. D. (1987) Physical Oceanography. In: The GulfofAlaska:Physical Environment and Biological Resources. W. D. Hood and S. T. Zimmerman(eds.) Washington: U.S. Gov. Print. Off., pp.57-76.Royce, W. F., Smith, L. S., and Hartt, A. C. (1968) Models of oceanic migrations of Pacificsalmon and comments on guidance mechanisms. U.S. Fish and Wildlife Service Fish.Bull. 66:441-462.Saila, S. B., and Shappy, R. A. (1963) Random movement and orientation in salmon migration.J. du Conseil International pour l’Exploration de laMer 28:153-166.Thomson, K.A., Ingraham, W.J., Healey, M.C., LeBlond, P.H., Groot, C., and Healey C.G.(1992) The influence of ocean currents on latitude of landfall and migration speed ofsockeye salmon returning to the Fraser River. Fish. Oceanogr. 1(2): 163-179.Thomson, K.A., Ingraham, W.J., Healey, M.C., LeBlond, P.H., Groot, C., and Healey C.G.(1993) Computer simulations of the influence of ocean currents on Fraser River sockeyesalmon return times. Can. J. Fish. Aquat. Sci.Thomson, K.A., LeBlond, P.H., Huato, L., Walters, C.J., Healey, C.M., and Hinch, S.G.(1994) Computer simulations of the effects of the Sitka eddy on sockeye salmon returningto British Columbia. In preparation for submission to : Can. J. Fish. Aquat. Sci.Ware, D. M., and McFarlane, G. A. (1989) Fisheries production domains in the NortheastPacific Ocean. In: Effects of Ocean Variability on Recruitment and an Evaluation ofParameters Used in Stock Assessment Models. R. 3. Beamish and G. A. McFarlane(eds.) Can. Spec. Publ. Fish. Aquat. Sci. 108:359-379.Welch, D.W., and Parsons, T.R. (1993) 13C- ‘5N values as indicators of trophic position andcompetitive overlap for Pacific salmon (Oncorhynchus spp.). Fish. Oceanogr. 2:11-23.Wickett, W. P. (1977) Relationship of coastal oceanographic factors to the migration of FraserRiver sockeye salmon (Oncorhynchus nerka ). International Council for the Exploration ofthe Sea. CM 1977/M 26: p. 18.81Xie, L., and Hsieh, W. W. (1989) Predicting the return migration routes of the Fraser Riversockeye salmon (Oncorhynchus nerka). Can. J. Fish. Aquat. Sci. 46:1287-1292.82Appendix 1The latitudes and longitudes are given in °N and OW, respectively. The tagging dates startas 1 for May 1st to 123 for August 3 1st.Sockeye salmon recovered in the Fraser RiverLatitude Longitude Start # of Year Latitude Longitudeof tagging of tagging date migration of recovery of recoverydays50.667 152.37 50 50 63 49.080 123.0053.067 150.92 28 60 64 49.200 123.0050.917 151.20 27 55 64 49.060 123.2249.167 147.00 26 62 64 49.200 123.0047.083 145.75 25 80 64 49.200 123.0052.167 142.67 10 85 64 49.200 123.0049.100 143.55 23 87 64 49.200 123.0048.383 147.07 9 69 65 49.050 123.0553.067 150.92 28 60 64 49.200 123.0050.917 151.20 27 55 64 49.060 123.2251.000 160.00 1 105 66 49.150 123.0050.000 157.50 30 63 66 49.100 123.3351.050 147.33 17 69 66 49.200 122.9250.000 147.50 32 62 66 49.100 123.3251.050 147.33 17 69 66 49.200 122.9254.000 142.50 34 38 66 49.170 122.9754.000 142.50 34 59 66 49,150 123.0054.050 142.50 51 70 66 50.670 121.8353.000 145.00 29 86 66 49.100 123.3351.000 140.00 36 36 66 49.170 122.4850.000 137.58 37 58 66 49.920 121.5049.000 137.50 28 38 66 49.100 122.7055.050 145.58 26 88 67 49.100 123.0853.000 133.00 52 46 67 49.100 123.3053.000 133.50 86 20 67 49.150 122.3849.000 137.50 28 38 66 49.100 122.7052.000 131.17 57 14 61 49.100 123.0849.000 157.00 34 87 62 49.080 123.2249.700 156.83 52 55 62 49.100 122.5852.100 153.00 24 61 62 49.220 123.7852.833 153.67 2 49 62 49.010 123.1852.917 149.00 4 66 62 49.170 122.5751.750 149.08 12 102 62 50.230 121.5851.750 149.08 12 95 62 49.080 123.2251.750 149.08 12 107 62 49.080 123.2255.000 143.00 32 45 62 49.080 123.228354.967 142.50 68 39 62 49.080 123.2252.083 135.33 54 34 62 49.570 121.4252.083 138.67 55 29 62 49.100 123.3252.000 137.77 70 43 62 49.180 123.1852.067 155.85 4 109 63 49.080 123.2247.950 156.00 2 110 63 49.170 122.5047.950 156.00 2 97 63 49.080 123.2247.950 156.00 2 88 63 49.150 122.4047.950 156.00 2 94 63 49.100 122.6547.950 156.00 2 110 63 49.080 123.2247.000 154.92 21 85 63 49.100 123.3351.250 150.00 15 81 63 49.130 123.1851.017 148.33 28 70 63 49.280 123.1751.017 148.33 28 72 63 49.170 123.2549.000 145.50 25 86 63 49.010 123.0151,117 144.78 30 78 63 49.370 121.4554.867 142.00 24 68 65 49.010 123.0854.867 142.00 24 78 65 49.080 122.0850.050 137.42 34 32 65 49.010 122.4250.000 157.50 30 63 66 49.010 123.3351.000 160.00 1 105 66 49.010 123.0051.050 147.33 17 69 66 49.200 122.9350.000 147.50 32 62 66 49.010 123.3254.000 142.50 34 38 66 49. 170 122.9754.000 142.50 34 59 66 49.150 123.0054.050 142.50 51 70 66 50.670 121.8353.000 145.00 29 86 66 49.010 123.3351.000 140.00 36 67 66 49.170 122.4850.000 137.58 37 58 66 49.920 121.5053.567 140.88 30 89 67 49.100 123.3053.000 133.00 52 48 67 49.100 123.3053.000 133.00 86 20 67 49.150 122.40Sockeye salmon recovered in the Strait of Georgia53.200 154.13 20 33 64 50.800 126.0248.383 147.07 9 67 64 50.600 127.0855.067 144.50 37 63 64 50.830 127.6752.167 142.67 10 66 64 50.600 127.0853.200 154.13 20 82 64 50.800 126.0250.000 160.00 11 99 66 50.730 127.4853.417 153.42 8 93 66 50.670 127.0052.817 152.35 60 42 66 50.500 126.3352.817 152.35 60 55 66 50.500 126.3352.000 147.50 30 72 66 50.670 127.0052.983 147.50 59 52 66 50.550 126.7852.000 147.50 76 32 66 50.580 126.7751.083 147.60 58 51 66 50.730 126.7548.117 150.00 12 88 66 50.330 125.4753.417 153.42 8 93 66 50.670 127.0052.817 152.35 60 42 66 50.500 125.3352.000 152.50 28 79 66 50.480 126.338451.000 152,50 10 100 66 50.380 125.8352.000 147.50 30 72 66 50.670 127.0052.983 147.50 59 44 66 50.550 126.7852.000 147.50 76 39 66 50.380 125.8352.000 147.50 76 32 66 50.580 126.7751.000 150.00 68 26 66 50.670 127.0051.000 150.00 68 41 66 50.330 125.4247.000 150.00 34 60 66 50.670 127.0054.950 142.58 95 25 66 50.500 126.2051.533 136.33 49 28 66 50.630 126.9851.000 137.50 66 18 66 50.630 127.1755.050 145.58 26 80 67 50.500 126.3352.000 143.17 12 87 67 50.500 126.0852.000 143.17 14 93 67 50.480 126.5852.000 143.17 14 59 67 50.580 127.1752.533 143.88 16 89 67 50.580 127.1751.800 143.72 9 89 67 50.330 125.4353.000 140.00 28 31 67 50.670 127.2853.500 139.17 31 88 67 50.500 126.3352.500 139.43 31 35 67 50.830 127.6050.683 136.33 7 59 67 49.630 124.0050.683 136.33 7 110 67 50.730 127.4849.883 132.58 6 74 67 50.630 127.1757.000 142.52 31 35 67 50.860 127.9057.000 142.50 50 59 66 50.500 126.0856.500 142.50 70 31 66 50.550 126.5856.500 142.50 70 32 66 50.550 126.7556.500 142.50 70 31 66 50.670 127.0055.000 142.50 35 58 66 50.480 126.5855.000 142.50 55 72 66 50.600 126.9257.000 142.52 31 35 66 50.860 127.9057.000 142.50 50 59 66 50.500 126.0856.500 142.50 70 31 66 50.550 126.7556.500 142.50 70 31 66 50.670 127.0052.083 144.78 48 55 66 50.670 127.0055.067 140.03 55 43 66 49.570 124.4354.967 135.97 58 29 68 50.600 126.9352.000 131.17 57 62 61 50.500 126.3352.000 131.17 57 10 61 50.500 126.3352.100 153.00 24 83 62 49.550 124.3350.250 153.08 23 81 62 49.800 124.6350.783 153.42 38 25 62 50.520 126.5554.417 149.08 31 75 62 49.550 124.3354.417 149.08 31 68 62 50.670 127.0553.000 147.00 33 65 62 50.670 127.2851.000 149.50 3 91 62 50.500 126.3254.967 142.50 68 52 62 49.550 124.3354.033 136.08 81 13 62 50.750 126.1754.033 136.08 81 18 62 50.550 126.5852.050 139.07 80 19 62 50.570 126.6852.000 136.00 81 16 62 50.920 127.8351.250 150.00 15 84 63 50.500 126.5851.250 150.00 15 97 63 50.330 125.4351.017 148.33 28 69 63 50.370 125.728551.367 145.83 29 69 63 50.370 125.6851.367 145.83 29 69 63 50.500 126.6349.000 150.00 24 79 62 50.480 126.5852.500 139.00 14 86 63 50.370 125.6750.000 160.00 11 99 66 50.730 127.4853,417 153.42 8 93 66 50.670 127.0052.817 152.35 60 42 66 50.500 126.3352.817 152.35 60 55 66 50.500 126.3352.000 152.50 28 79 66 50.480 126.5851.000 155.00 29 81 66 50.670 127.0051.000 152.50 10 100 66 50.380 125.8350.000 152.50 30 73 66 50.550 126.7852.000 147.50 30 72 66 50.670 127.0052.983 147.50 59 44 66 50.550 126.7852.000 147.50 76 39 66 50.380 125.8352.000 147.50 76 32 66 50.580 126.7751.083 147.60 58 20 66 50.730 126.7551.000 150.00 68 26 66 50.670 127.0051.000 150.00 68 41 66 50.330 125.4248.117 150.00 12 88 66 50.330 125.4747.000 150.00 34 60 66 50.670 127.0054.950 142.58 95 25 66 50.500 126.2051.000 137.50 66 18 66 50.630 127.1755.050 145.58 26 80 67 50.500 126.3352.000 143.17 14 59 67 50.630 127.1752.000 143.17 14 93 67 50.480 126.5852.000 143.17 12 87 67 50.500 126.0852.533 143.88 16 89 67 50.630 127.1752.967 140.28 28 86 67 50.550 126.6851.800 143.72 9 89 67 50.330 125.4353.000 140.00 28 31 67 50.670 127.2853.500 139.17 31 88 67 50.500 126.3352.500 139.43 31 35 67 50.830 127.6050.683 136.33 7 59 67 49.630 124.0052.000 152.50 28 79 66 50.480 126.5850.000 152.50 30 73 66 50.550 126.7851.000 155.00 29 81 66 50.670 127.00Sockeye salmon recovered in the Juan de Fuca Strait48.250 144.53 8 89 64 48.300 124.0048.250 144.37 8 89 64 48.300 124.0052.133 144.83 15 92 67 48.230 124.6250.467 128.27 100 12 67 48.500 124.5055.000 142.50 49 65 66 48.500 124.5052.083 144.78 48 71 66 48.530 124.4250.750 154.10 57 42 62 48.600 124.7550.783 153.42 38 52 62 48.470 124.3052.917 149.00 4 120 62 48.580 124.7253.000 147.00 33 89 62 48.500 124.5052.950 147.92 58 64 62 48.370 123.9254.967 142.50 68 45 62 48.330 123.758652.050 139.07 80 41 62 48.500 124.5052.050 139.07 80 37 62 48.330 123.7547.000 154.92 21 89 63 48.500 124.5051.250 150.00 15 83 63 48.580 124.7451.050 142.67 22 80 65 48.420 124.6652.000 152.50 28 67 66 48.500 124.5051.000 152.50 10 98 66 48.500 124.5050.000 152.50 30 89 66 48.500 124.5053.033 147.58 19 84 66 48.420 124.7051.000 150.00 68 46 66 48.300 124.7250.883 145.42 69 40 66 48.500 124.6649.950 149.95 16 110 66 48.500 124.7454.000 142.50 34 59 66 49,150 123.0053.000 145.00 29 73 66 48.500 124.6651.067 140.13 72 24 66 48.420 124.0552.500 144.00 11 96 67 48.420 124.0553.050 144.75 12 94 67 48.500 124.6652.817 144.68 14 87 67 48.500 124.5053.500 139.17 31 76 67 48.500 124.6652.350 135.18 33 79 67 48.500 124.5054.000 133.50 85 23 67 48.500 124.5049.100 143.55 23 78 65 48.370 123.9052.000 152.50 28 67 66 48.500 124.5050.000 152.00 30 89 66 48.500 124.5053.033 147.58 19 84 66 48.420 124.7050.883 145.42 69 40 66 48.420 124.6749.950 149.95 16 110 66 48.500 124.7353.033 147.58 19 84 66 48.420 124.7051.000 150.00 68 46 66 48.600 124.7249.950 149.95 16 110 66 48.500 124.7354.000 142.50 34 80 66 48.520 124.6253.000 145.00 29 73 66 48.420 124.6751.067 140.13 72 24 66 48.420 124.5055.000 140.00 35 73 66 48.420 124.6753.050 144.75 12 94 67 48.420 124.6752.983 144.68 14 87 67 48.500 124.5053.500 139.17 31 76 67 48.420 124.6754.000 133.50 85 23 67 48.500 124.5052.083 144.78 48 70 66 48.530 124.4255.000 140.00 35 74 66 48.500 124.3355.050 145.58 26 88 67 48.500 124.67Sockeye salmon recovered in the Queen Charlotte Islands56.250 136.50 58 32 67 53.015 132.9650.683 136.33 7 29 67 52.953 132.9051.100 139.67 8 95 67 52.919 132.8755.100 133.92 44 8 61 53.023 133.0155.100 133.92 44 8 61 53.023 133.0155.000 141.17 18 24 62 53.026 133.0350.683 136.33 7 29 67 52.953 132.9051.100 139.67 8 95 67 52.919 132.8787

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

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.831.1-0053154/manifest

Comment

Related Items