Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Comparison of some physical characteristics of salmonids under culture conditions using underwater video… Jones, Rachael Elizabeth 1997

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

Item Metadata

Download

Media
831-ubc_1997-0180.pdf [ 4.2MB ]
Metadata
JSON: 831-1.0058938.json
JSON-LD: 831-1.0058938-ld.json
RDF/XML (Pretty): 831-1.0058938-rdf.xml
RDF/JSON: 831-1.0058938-rdf.json
Turtle: 831-1.0058938-turtle.txt
N-Triples: 831-1.0058938-rdf-ntriples.txt
Original Record: 831-1.0058938-source.json
Full Text
831-1.0058938-fulltext.txt
Citation
831-1.0058938.ris

Full Text

C O M P A R I S O N OF S O M E PHYSICAL CHARACTERIST ICS OF SALMONIDS UNDER C U L T U R E CONDITIONS USING U N D E R W A T E R VIDEO IMAGING TECHNIQUES by R A C H A E L ELIZABETH J O N E S B.Sc. (Biology), McGill University, 1990 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE D E G R E E OF M A S T E R OF S C I E N C E in THE FACULTY OF G R A D U A T E STUDIES D E P A R T M E N T OF CHEMICAL AND B I O - R E S O U R C E ENGINEERING We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA February, 1997 © Rachael Elizabeth Jones, 1997 In presenting this thesis in partial fulfilment of the requirements for 1 an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of C H E M \ C A L 4 g \ Q - R b S O Q £ C b fcl\&l^£rRlNG The University of British Columbia Vancouver, Canada Date DE-6 (2/88) ABSTRACT Body dimensions and swimming speeds of Atlantic salmon (Salmo salar) and Chinook salmon (Oncorhynchus tshawytscha) were measured in order to try to explain some of the observed growth differences between the two species under similar husbandry conditions, and to find an accurate estimator of individual fish mass from body dimensions. Size distribution of individual Chinook salmon in sea cages was also examined. Data was collected using a pre-existing non-invasive underwater video imaging system (VICASS) based on the principles of stereoimagery. An existing database of physical measurements was also analyzed (consisting of 1539 Atlantic salmon ranging in size from 0.42 to 8.50 kg and 840 chinook salmon ranging in size from 0.009 to 4.91 kg). Average fish size varied significantly with position in six out of fourteen cages of chinook salmon. This result suggested the presence of a dominance hierarchy, in which the largest fish in the cage are found in the best apparent location in the cage. Swimming speeds for chinook salmon ranged from 0.37 to 1.06 m s"1 or 0.72 to 2.04 body lengths s"1 (bl s"1). Atlantic salmon swam between 0.25 to 0.98 m s"1 or 0.40 to 1.82 bl s' 1 . The swimming speeds (in bl s 1 ) were compared for fish between 1.0 - 3.5 kg in size, and chinook salmon were found to be swimming approximately 20% faster than Atlantic salmon on average. After testing several different models, M=BL 2H was found to be the best estimator of mass. B varied from 39.58 to 73.86 in Atlantic salmon and 45.93 to 76.52 in chinook salmon. This model decreased the variability in calculating mass by between 38 and 44% as compared to the conventional fisheries model M=KL 3. Chinook salmon were significantly taller and thicker than Atlantic salmon of similar mass, while Atlantic salmon were significantly longer than Chinook salmon of similar mass. The ratios of fork length to height and fork length to girth were also found to be significantly different for the two species. The drag forces acting on Chinook salmon were found to be approximately 40% higher than those for Atlantic salmon as a result of the above morphological and swimming speed differences, and the power needed to overcome these drag forces was found to be 53% higher. Due to this increased demand on the metabolic component of the energy budget of chinook salmon, their growth was decreased by 5.2 to 10.3% (depending on fish size) as compared to Atlantic salmon. This difference in growth explains over 50% of the observed difference in the F C R s of the two species. Results suggest that farmers should: a) choose chinook salmon stock that swim slower and/or have a body form that is more similar to that of Atlantic salmon, thereby decreasing the energetic demands of swimming in chinook salmon, or b) develop a higher energy feed for chinook salmon. iii TABLE OF CONTENTS ABSTRACT ii TABLE OF CONTENTS iv LIST OF TABLES vi LIST OF FIGURES vii LIST OF APPENDICES ix ACKNOWLEDGEMENT x INTRODUCTION 1 OBJECTIVES 6 LITERATURE REVIEW 7 Swimming Speed 7 Condition and Shape Descriptors 9 Fish Distribution and Behaviour 12 MATERIALS AND METHODS 15 Equipment 15 Data Acquisition 15 Sampling Procedures 18 Video Sampling 18 Physical Sampling 19 Sites 20 Site A 20 Site B 22 S i t e C 22 Data Manipulation 23 Units of Measurement 24 iv Numerical and Statistical Analysis 24 RESULTS 27 Terminology 27 Density 27 Population Distribution Patterns 27 Position Within the School 31 Size Distribution 32 Site A 32 Site B 34 Site C 37 Swimming Speed 38 Chinook salmon 38 Atlantic and chinook salmon 50 Body Dimensions 54 Length-Weight Models 57 Body Dimensions and Shape 65 Drag 71 DISCUSSION 75 Size Distribution 75 Site A 75 Site B 76 Site C 77 Swimming Speed 79 Body Dimensions 81 CONCLUSIONS 89 AREAS FOR FURTHER STUDY 91 REFERENCES 93 PERSONAL COMMUNICATIONS 99 APPENDICES 100 v LIST OF TABLES T A B L E 1: CHARACTERIST ICS OF C A G E S S A M P L E D 21 TABLE 2: FILMING CONDITIONS OF C A G E S S A M P L E D AT SITE A 28 TABLE 3: FILMING CONDITIONS OF C A G E S S A M P L E D AT SITE B 29 TABLE 4: FILMING CONDITIONS OF C A G E S S A M P L E D AT SITE C 30 TABLE 5: C O M P A R I S O N OF A V E R A G E WEIGHTS F R O M V ICASS, DIP NET AND H A R V E S T DATA. 33 T A B L E 6.1: ATLANTIC S A L M O N SWIMMING S P E E D DATA (ADAPTED F R O M SHIEH, 1996) 40 T A B L E 6.2: ATLANTIC S A L M O N SWIMMING S P E E D DATA (ADAPTED F R O M SHIEH, 1996) CONT. 41 T A B L E 7: SWIMMING S P E E D STATISTICAL R E S U L T S FOR SITE A 46 TABLE 8: SWIMMING S P E E D STATISTICAL R E S U L T S FOR SITE B 48 TABLE 9: SWIMMING S P E E D STATISTICAL R E S U L T S FOR SITE C 49 TABLE 10: R E G R E S S I O N COEFFICIENTS FOR S H A P E CHARACTERIST ICS OF CHINOOK S A L M O N 63 T A B L E 11: R E G R E S S I O N COEFFICIENTS FOR S H A P E CHARACTERIST ICS OF ATLANTIC S A L M O N 64 vi LIST OF FIGURES FIGURE 1 VIDEO IMAGE CAPTURING AND SIZING S Y S T E M (VICASS) 16 F IGURE 2 F R E Q U E N C Y P O L Y G O N OF FISH SIZE DISTRIBUTION FOR C A G E 3, SITE B, ILLUSTRATING A C A S E W H E R E T H E R E IS NO STATISTICAL D I F F E R E N C E B E T W E E N A N Y OF THE FIVE S A M P L E S . 36 F IGURE 3 F R E Q U E N C Y P O L Y G O N OF FISH SIZE DISTRIBUTION FOR C A G E 4, SITE C, ILLUSTRATING A C A S E W H E R E T H E R E A R E STATISTICAL D I F F E R E N C E S B E T W E E N THE FIVE S A M P L E S . 39 F IGURE 4 THE E F F E C T OF C A M E R A INTRODUCTION INTO A S E A C A G E ON SWIMMING S P E E D . S T R O N G EVIDENCE OF INCREASED SWIMMING S P E E D DUE TO C A M E R A INTRODUCTION IN C A G E 5-P1, SITE A. THE A R R O W INDICATES THE POINT AT WHICH THE SWIMMING S P E E D S B E C O M E UNIFORM. 43 F IGURE 5 A F R E Q U E N C Y P O L Y G O N OF SWIMMING S P E E D FOR C A G E 5-P1, SITE A, ILLUSTRATING THE L E F T - S T E E P E D , FLAT ON THE RIGHT A P P E A R A N C E OF THE DISTRIBUTION. 44 FIGURE 6 VARIATION IN SWIMMING S P E E D WITH RELATION TO TIME OF DAY FOR (A) CHINOOK S A L M O N AND (B) ATLANTIC S A L M O N . E A C H POINT R E P R E S E N T S THE A V E R A G E SWIMMING S P E E D FOR A POSITION. 51 F IGURE 7 VARIATION IN SWIMMING S P E E D (MEASURED IN M S"1) IN RELATION TO A V E R A G E FISH SIZE FOR (A) CHINOOK S A L M O N AND (B) ATLANTIC S A L M O N . E A C H POINT R E P R E S E N T S THE A V E R A G E SWIMMING S P E E D FOR A POSITION. NOTE THAT T H E R E W A S NO DATA AVAILABLE FOR ATLANTIC S A L M O N B E T W E E N 2 AND 3 K G IN SIZE. 52 F IGURE 8 THE RELATIONSHIP B E T W E E N SWIMMING S P E E D (MEASURED IN BL S"1) AND A V E R A G E FISH SIZE FOR CHINOOK (C) AND ATLANTIC (A) S A L M O N . 53 F IGURE 9 THE RELATIONSHIP B E T W E E N SWIMMING S P E E D AND A V E R A G E W A T E R T E M P E R A T U R E FOR (A) CHINOOK AND (B) ATLANTIC S A L M O N . E A C H POINT R E P R E S E N T S THE A V E R A G E SWIMMING S P E E D FOR A POSITION. THE LINE OF BEST FIT W A S G E N E R A T E D BY SIMPLE LINEAR R E G R E S S I O N OF THE DATA. 55 F IGURE 10 THE RELATIONSHIP B E T W E E N SWIMMING S P E E D AND A V E R A G E DEPTH FOR (A) CHINOOK AND (B) ATLANTIC S A L M O N . E A C H POINT R E P R E S E N T S THE A V E R A G E SWIMMING S P E E D FOR A POSITION. 56 vii FIGURE 11 S C A T T E R DIAGRAMS SHOWING THE VARIABILITY OF C A L C U L A T E D M A S S A R O U N D T R U E M A S S FOR CHINOOK S A L M O N . (A) INCREASED VARIABILITY OF M A S S C A L C U L A T E D USING K A S C O M P A R E D TO (B) M A S S C A L C U L A T E D USING B. 58 FIGURE 12 S C A T T E R DIAGRAMS SHOWING THE VARIABILITY OF C A L C U L A T E D M A S S A R O U N D T R U E M A S S FOR ATLANTIC S A L M O N . (A) INCREASED VARIABILITY OF M A S S C A L C U L A T E D USING K A S C O M P A R E D TO (B) M A S S C A L C U L A T E D USING B. 59 F IGURE 13 S C A T T E R PLOT OF (A) STANDARDIZED B V A L U E S AND (B) STANDARDIZED K V A L U E S AGAINST M A S S FOR CHINOOK S A L M O N . 60 F IGURE 14 S C A T T E R PLOT OF (A) STANDARDIZED B V A L U E S AND (B) STANDARDIZED K V A L U E S AGAINST M A S S FOR ATLANTIC S A L M O N . 61 F IGURE 15 S C A T T E R PLOT OF (A) LENGTH C U B E D AND (B) LENGTH S Q U A R E D X HEIGHT AGAINST M A S S FOR CHINOOK S A L M O N . 62 F IGURE 16 THE VARIABILITY OF HEIGHT WITH LENGTH FOR (A) CHINOOK AND (B) ATLANTIC S A L M O N . 66 FIGURE 17 THE RELATIONSHIP B E T W E E N FORK LENGTH (L) AND M A S S (M) FOR BOTH S P E C I E S , W H E R E C DENOTES CHINOOK AND A D E N O T E S ATLANTIC S A L M O N . 68 FIGURE 18 THE RELATIONSHIP B E T W E E N HEIGHT (H) AND M A S S (M) FOR BOTH S P E C I E S , W H E R E C D E N O T E S CHINOOK AND A D E N O T E S ATLANTIC S A L M O N . 69 F IGURE 19 THE RELATIONSHIP B E T W E E N GIRTH (G) AND M A S S (M) FOR BOTH S P E C I E S , W H E R E C D E N O T E S CHINOOK AND A D E N O T E S ATLANTIC S A L M O N . 70 F IGURE 20 THE RELATIONSHIP B E T W E E N R O U N D N E S S AND M A S S FOR BOTH S P E C I E S , W H E R E C D E N O T E S CHINOOK AND A D E N O T E S ATLANTIC S A L M O N . 72 F IGURE 21 THE RELATIONSHIP B E T W E E N DRAG AND M A S S FOR BOTH S P E C I E S , W H E R E C D E N O T E S CHINOOK AND A D E N O T E S ATLANTIC S A L M O N . 74 F IGURE 22 THE RELATIONSHIP B E T W E E N M A S S AND F O R K LENGTH FOR BOTH S P E C I E S , W H E R E C D E N O T E S CHINOOK AND A D E N O T E S ATLANTIC S A L M O N . 84 F IGURE 23 THE E N E R G Y BUDGET OF ATLANTIC AND CHINOOK S A L M O N 87 viii LIST OF APPENDICES A-1: STATISTICAL ANALYSIS OF A V E R A G E M A S S E S F R O M VIDEO S A M P L E S , DIP NET S A M P L E S , AND H A R V E S T DATA - A N O V A AND BARTLETT 'S TEST100 A-2: STATISTICAL ANALYSIS OF A V E R A G E M A S S E S F R O M VIDEO S A M P L E S AND DIP NET S A M P L E S , SITE A - T U K E Y T E S T 101 A-3: STATISTICAL ANALYSIS OF A V E R A G E M A S S E S F R O M VIDEO S A M P L E S , DIP NET S A M P L E S AND H A R V E S T DATA, SITE B - T U K E Y T E S T 102 A-4: STATISTICAL ANALYSIS OF A V E R A G E M A S S E S F R O M VIDEO S A M P L E S , DIP NET S A M P L E S AND H A R V E S T DATA, SITE C - T U K E Y T E S T 103 A-5: STATISTICAL ANALYSIS OF A V E R A G E S P E E D S F R O M VIDEO S A M P L E S , M E A S U R E D IN M/S - A N O V A AND HARTLEY T E S T 104 A-6: STATISTICAL ANALYSIS OF A V E R A G E S P E E D S F R O M VIDEO S A M P L E S , M E A S U R E D IN BL/S - A N O V A AND HARTLEY TEST 105 A-7: STATISTICAL ANALYSIS OF A V E R A G E S P E E D S F R O M VIDEO S A M P L E S , M E A S U R E D IN M/S - T U K E Y T E S T 106 A-8: STATISTICAL ANALYSIS OF A V E R A G E S P E E D S F R O M VIDEO S A M P L E S , M E A S U R E D IN BL/S - T U K E Y T E S T 107 ix ACKNOWLEDGEMENT In appreciation of the support without which this research project would not have been possible, I would like to extend my thanks to the following people and organizations. Dr. Royann Petrell provided guidance, support and patience throughout this project. I am grateful to my committee members, Dr. Daniel Pauly and Dr. Sie-Tan Chieng, for their suggestions and comments. Finally, I would like to thank Pacific National Group Ltd. and Paradise Bay Seafarms for allowing me to conduct my research on their sites and for their willing help during those visits. x INTRODUCTION The history of farmed Atlantic salmon dates back to the early 1960's, when it was introduced in both Norway and Scotland. Salmon farming was first attempted in B.C. in 1972, primarily with chinook and coho salmon (Oncorhynchus kisutch) (Bjorndal, 1990). The introduction of domesticated Atlantic salmon into the B.C. salmon farming industry began in 1988, and by 1993 production of Atlantic salmon exceeded that of chinook salmon, the only Pacific species that has been farmed here with any degree of success, by approximately 5000 metric tonnes (BCMAFF, 1993). In 1994 the annual production (in dressed pounds) of Atlantic salmon in B.C. was more than double that of chinook salmon, at 11,000 metric tonnes (BCSFA, 1994). Some of the reasons for this trend away from farming chinook salmon are that they have higher feed conversion ratios (FCRs), higher mortality rates, and lower growth rates than Atlantic salmon. The F C R is the ratio of feed fed (kg) over mass gain (kg). According to the Cooperative Assessment of Salmonid Health program which is being conducted through the B.C. Salmon Farmers' Association, typical FCR values in British Columbia are approximately 1.5 for Atlantic salmon and 2.0 for chinook salmon over the entire production period. Atlantic salmon also have a much higher average harvest size, at 4.3 kg as compared to 2.5 kg for chinook salmon (BCMAFF, 1993). Finally, Atlantic salmon command a higher market price than chinook salmon, at an average price of $3.55 per pound as compared to $3.31 per pound for chinook salmon (BCSFA, 1994). The severe decline in chinook salmon prices over the past ten years places a major constraint upon the economic viability of the species, compounded by the fact that chinook salmon prices drop every summer to compete with those of the 1 wild fishery, which are generally below the cost of production (BCSFA et al., 1992). Atlantic salmon prices are much better cushioned against the annual summer drop in prices. All of these factors add up to the higher production costs associated with chinook salmon, leading to decreased profits. Although it has been considered that in general the biophysical criteria on which site selection is based are the same for all salmonids, there are in fact some small but significant differences. The most environmentally sensitive species of farmed salmonid is the chinook salmon (Ricker & Truscott, 1989). Chinook salmon have higher oxygen requirements than Atlantic salmon and show symptoms of oxygen stress when dissolved oxygen is below saturation, or less than 9 mg I"1. Atlantic salmon can tolerate oxygen levels as low as 44% saturation or less than 4.5 mg I"1 (Caine et al., 1987). Chinook salmon are much more sensitive to temperature fluctuations than Atlantic salmon, but Atlantic salmon become increasingly susceptible to Vibrio infections as salinity levels drop below 15 parts per thousand (Caine et al., 1987). Finally, Atlantic salmon show somewhat faster growth than chinook salmon at temperatures lower than 14°C, which is the optimum temperature for growth for both species (Pennell, 1992). It is, therefore, possible to grow Atlantic salmon at sites where chinook salmon would not be a viable option, for example at a site with lower dissolved oxygen levels or greater temperature fluctuations. The overall purpose of this research was to gain insight into the behaviour, condition and morphology of chinook salmon in sea cages in an attempt to explain the observed differences in growth between chinook and Atlantic salmon. The underwater cameras and image analysis technology (Video Image Capturing and Sizing System, or VICASS) utilized in this study had previously been developed in the Fish Image 2 Laboratory, Chemical and Bio-Resource Engineering, U.B.C. (Petrell et al., 1996). Earlier research carried out by Shieh (1996) characterized the size distribution of individual Atlantic salmon in sea cages over a wide range of sizes and under varying husbandry conditions. This research utilized the knowledge gained from these earlier studies, in addition to a large database of physical measurements (length, height, girth and mass) taken from both Atlantic and chinook salmon, and took the study one step further in an attempt to explain as well as document the growth patterns exhibited by farmed salmon. The size distribution of fish in sea cages may have important implications regarding growth as size segregation could be evidence of competition and/or territoriality. There have been many studies carried out on the importance of size in determining the status of an individual (Fenderson et al., 1968; Jenkins, 1969; Li & Brocksen, 1977; Ejike & Schreck, 1980; and Abbott & Dill, 1989). Magurran (1986) states that "the competitive ability of an individual determines its rank in a dominance hierarchy, with high-ranking individuals having preferential access to food, mates or whatever commodity is in demand". Therefore, if a size-based dominance hierarchy is in existence in chinook salmon sea cages, it could have implications for the scope for growth of both the dominant and subordinate individuals. Another possible explanation for the lower F C R s and lower growth potential of chinook salmon as compared to Atlantic salmon is that they use up more energy in other areas, such as swimming. This could be due to several different factors. The most obvious is that chinook salmon swim at a higher absolute speed than Atlantic salmon. Weihs and Webb (1983) defined the optimal constant speed as that at which the rate of energy expended per unit distance is minimal, and Weihs (1973) predicted 3 on theoretical grounds that this speed is approximately 1.0 bl s"1 (see literature review for further explanation). Therefore, if chinook salmon are found to swim substantially faster than this theoretical optimum speed, they are clearly not swimming in an energy-efficient manner, and will have less resources available for growth. Less energy-efficient locomotion could also be related to morphology. A poorly streamlined fish will have larger drag forces associated with its movement through water, again resulting in higher energy requirements for propulsion. By investigating the body dimensions of the two species in terms of fork length, girth, and height and, through study of the interrelationships between these parameters, comparing the relative 'shapes' of these fish, estimates of the drag forces acting on the fish can be calculated and compared. Length-weight models can be used by aquaculturists in two ways. First, to predict growth (in terms of mass as some function of length) and second, to assess nutritional status as defined by condition. The importance of an accurate model for determining both growth and condition is immense. Feed rations are based largely on the estimated size of fish in a cage. Overestimation of fish size leads to overfeeding, which in turn translates into feed wastage, a huge concern for the fish farmer both economically and environmentally, as uneaten food accumulates underneath the farm where it affects the benthic flora and fauna and attracts wild fish. Conversely, underestimation of fish size leads to underfeeding, resulting in slower growth, smaller fish at harvest and decreased profits. Condition is also crucial for determining the health of a population. A model that is sensitive to and reflects small changes in condition can alert a farmer to the onset of disease, stress due to overcrowding, or other physiological effects before high mortality rates are suffered. 4 Mass-balance equations provide a means of modelling the partitioning of food energy between the various metabolic processes within a fish, such as tissue function and repair, synthesis of new tissue and swimming. An energy budget for an individual fish takes the basic form: I=M+G+E where / is the energy content of the food consumed over the time period, M is the energy lost in the form of heat produced during metabolism, G is the energy in both somatic and reproductive growth and E is the energy lost in faecal and excretory products (Wootton, 1990). It has been shown that roach reduce locomotor activity to compensate for gonadal production (Koch & Wieser, 1983). Perhaps a similar tradeoff exists for chinook salmon between swimming activity and somatic growth, whereby growth is reduced to meet the energetic requirements of swimming. In summary, it was hoped that a combination of the various aspects of behaviour, condition and morphology would facilitate a better understanding of some of the factors affecting the growth of chinook and Atlantic salmon, and possibly allow some recommendations to be made regarding the future of chinook salmon farming in British Columbia. 5 OBJECTIVES The objectives of this thesis were as follows: 1. To determine whether chinook salmon segregate according to size within a cage, using underwater cameras and image analysis techniques, and if so, to determine where the subpopulation swims which is most similar to the average weight as determined by physical data from either harvesting or dipnetting. 2. To compare swimming speeds of Atlantic and chinook salmon. 3. To compare the body dimensions and shape of chinook and Atlantic salmon. 4. To relate body dimensions to mass to determine the relationship between them, thereby finding an appropriate estimator of mass from body dimensions. 5. To apply the above information to facilitate an understanding of the differences in growth and F C R between chinook and Atlantic salmon. 6 LITERATURE REVIEW Swimming Speed Boisclair and Tang (1993) performed a study on the energetic costs associated with different types of swimming and found that routine swimming was 4.9 to 7.5 times more expensive than directed swimming for similar weight and speed combinations. Routine swimming was defined as swimming behaviour characterized by marked changes in swimming speed and direction, while directed swimming occurred when the fish were trained to remain under the shaded area of a semi-circular plate rotating above their aquarium where swimming speed equals the speed of the plate's rotation. Another study looked at the metabolic costs of spontaneous swimming by measuring the oxygen consumption of free-swimming fish (Krohn & Boisclair, 1994). They investigated the relationship between spontaneous swimming costs (measured directly by respirometry) and forced swimming (swimming with constant speed and direction) costs. Spontaneous swimming costs were six times higher than costs predicted by the forced swim relationship for the same speeds, suggesting that the metabolic costs of turning and accelerating can be substantial. Many researchers have carried out experiments in an attempt to discover the optimal swimming speed for maximum growth, which would logically be the natural swimming speed in the normal environment. The optimum cruising speed was mathematically derived by Weihs (1973) to be 1.0 bl s"1, based on experimental data from rainbow trout (Salmo gairdneri). Weihs states that the derivation should hold for many other species using carangiform and subcarangiform locomotion. However, this does not take temperature and size effects into consideration. Within a species, 7 individual fish have a higher metabolism and therefore swim faster with increasing water temperature, at least within their normal temperature regime. Larger individuals will swim slower (in terms of bl s'1) than smaller fish of the same species. In other words, there is a negative correlation between the optimum speed in bl s 1 and body mass: uopt = U0AT°M (Videler, 1993). Rainbow trout were trained to swim in a flume at 1, 2, and 3 bl s 1 and maximum growth was attained at 1 bl s"1 (Greer Walker & Emerson, 1978). East and Magnan (1987) trained brook trout (Salvelinus fontinalis) at water velocities of 0, 0.85, 1.72, and 2.5 bl s"1 and found that growth (wet weight), food conversion efficiency, and deposition of lipids on the digestive tract was maximized at 0.85 bl s"1. A similar study was concerned with the growth rates of brown trout (Salmo trutta) when exercised continuously at swimming speeds of 1.5, 3, and 4.5 bl s"1 (Davison & Goldspink, 1977). Again, the best growth was found at the swimming speed closest to 1.0 bl s 1 . Jobling (1994) states that there is an increasing body of evidence that indicates that salmonids show improved growth rates when they are forced to swim at moderate speeds for prolonged periods of time. He suggests that this may be due to behavioural changes that occur when groups of salmonid fish are exposed to water currents. These adaptations include orientation against the current, formation of schools, and highly reduced levels of aggression as compared with groups of fish held in still water. Improved weight gain in the exercising fish is the result of these fish displaying better food conversion. There have also been many field studies done on migrating fish. Malinin (1975) tracked acoustically tagged Atlantic salmon migrating in a reservoir at sustained speeds of 1.2 bl s 1 . Migrating adult sockeye salmon (Oncorhynchus nerka) were tracked 8 ultrasonically as they returned to the Fraser River at speeds of 66.75 cm s 1 (1.0 bl s"1 ) (Quinn, 1988). Smith et al. (1981) tracked six Atlantic salmon with an average length of 69 cm off the coast of Scotland and found that they swam at speeds of 0.58 bl s"1 in fixed directions regardless of the current. Condition and Shape Descriptors In general, the change in weight of fish can be described by the relationship W = aLh, where W is observed fish weight, L is observed fish length, and a and b are estimated by logW = loga + M o g Z (i.e. a is the regression intercept and b is the regression slope). Therefore, the case where b < 3 represents fish that become less rotund as length increases, whereas when b > 3 fish become more rotund as length increases. These are both examples of allometric growth. When b = 3, growth may be isometric (growth with unchanged body proportions and specific gravity), although it is possible for shape to change when b = 3, due to changes in a (Anderson & Gutreuter, 1983; Cone, 1989). The study of condition assumes that heavier fish of a given length are in better condition. Condition indices have been used by fisheries biologists as indicators of the general 'well-being or fitness' of the population under consideration. In a review paper, Bolger and Connolly (1989) identified eight forms of index that have been used to analyse and measure the condition of fish. These indices fall into two categories: those which measure the condition of individual fish, i.e. 'condition factors', and those which measure the condition of subpopulations as a whole. However, several publications have pointed out the problems inherent with the condition factors currently 9 in use (Bolger & Connolly, 1989; Cone, 1989; Hayes et al., 1995). In particular, Cone (1989) raises the question that the "conversion of the two-dimensional weight-length relationship into a single statistic results in a loss of information and, in many cases, an inaccurate representation of that relationship." Hayes et al. (1995) performed simulations of length-weight regressions and found that for sample sizes commonly used in fisheries research, estimates of the mean-weight-at-length were biased low, whereas estimates of the intercept were biased high. The earliest condition factor developed was Fulton's condition factor (K), which assumes isometric growth and is of the form K = xWL^ where x is an arbitrary scaling constant that varies with units of measure (Fulton, 1911). It is still one of the most commonly used indices today (Kristinsson et al., 1985; Kjartansson et al., 1988; Fries, 1994). However, there are many instances when the assumption of isometric growth will not be met, i.e. the slope of the weight-length relationship will not be 3.0. If, for example, b > 3, there is a significant positive relationship between K and fish length, indicating that K will increase with increasing length (Anderson & Gutreuter, 1983; Cone, 1989). Conversely, there is a negative relationship between K and fish length when b < 3, which leads to a decrease in K with increasing length (Cone, 1989). Ricker (1975) developed an extension of Fulton's condition factor that assumes allometric growth, known as the relative condition factor: K = WL~b. The relative condition factor must be confined in its uses to comparisons between fish which are homogeneous for b however, as it is based on the assumption that the slopes of all samples to be compared are equal to some value specific for that sample or set of samples (Bolger & Connolly, 1989; Cone, 1989). 10 The third commonly used condition index is relative weight, which is described by the following equation: Wr = \00WW~1 where I/Vis the weight of an individual and Ws is a length-specific standard weight. Ws is derived from Ws=a'Lh' where a ' a n d b' ideally account for genetically determined shape characteristics of a species and yield Wr values of 100 for fish that have been well fed. Therefore, this statistic assumes the same slope and intercept as the ideal population, rendering determination of the ideal relationship extremely critical. In summary, all of these indices relate the actual weight of an individual fish to some 'expected weight' which is calculated as a function of its length. Fulton's condition factor relates the actual weight to a hypothetical expected weight, the relative condition factor relates actual weight to a calculated average weight for the population, and relative weight relates it to a weight which is related to the genetically determined shape characteristics of the species. All three of the indices increase as condition increases. There is also some debate as to the method of estimating a and b. Due to the fact that length cannot be considered to be truly independent of weight, Ricker (1975) proposed using the geometric mean of the regression of weight on length and the inverse of the regression of length on weight as the appropriate linear regression for weight-length comparisons. Anderson and Gutreuter (1983) also recommend geometric mean regression techniques, however Cone (1989) favors the use of ordinary least-squares regression due to the difficulty of interpretation of geometric mean regression. 11 Fish Distribution and Behaviour There has been little research directed at the study of the distribution of salmon in sea cages, and what research has been conducted has been done almost exclusively on Atlantic salmon. However, the few behavioural studies that have been done on Atlantic salmon in sea cages "stress that the fish in a pen should not be regarded as one unit" (Ferno et al., 1988). Juell (1989) concurs, stating that "variations in behaviour between and within individuals are found to reflect e.g. phenotypic or environmental differences, rather than being accidental. If there exist e.g. size related differences in behaviour within a population in intensive aquaculture this might call for practices considering this". Stocking density also seems to play a role in determining fish behaviour. Although it has usually been assumed that high fish density prevents territorial behaviour, Atlantic salmon parr stocked at densities of 10 kg m 3 were studied and a few dominant individuals were found to defend specific areas of the tank, thereby gaining preferential access to food arriving in those areas (Ferno & Holm, 1986). Kjartansson et al. (1988) studied Atlantic salmon stocked at three different densities (35-45, 65-85, and 100-125 kg m 3 ) and found that "within fish communities, increased crowding often enforces social hierarchies with the appearance of dominant and subordinate individuals." Both the swimming speed and vertical distribution of Atlantic salmon in sea cages exhibit daily rhythms. Swimming speed shows a marked diurnal rhythm, being highest in the early morning and evening, and lowest in the early afternoon (Kadri et al., 1991). Vertical distribution is influenced by a diel rhythm, characterized by a downwards migration at dawn and a upwards migration at dusk (Ferno, 1989; Juell, 12 1989). However, while Jueil (1989) found that the acoustically tagged fish he was studying showed a definite preference for being in the deepest part of the population at all hours, Ferno (1989) detected the highest concentrations of fish somewhere in the middle of the cage, with the lowest concentrations observed at 5-7 m depth at the bottom of the cage. Bjordal et al. (1993) also found that before feeding the highest densities of Atlantic salmon were at medium depths. Juell (1989) found that the behaviour of adult Atlantic salmon changed drastically when the number of individuals in a cage was increased from 30 to 530. With only 30 fish in the cage, the swimming pattern was unstructured, characterized by long periods of drifting along the side of the cage. When the number of fish increased, structured behaviour initiated, which included swimming as part of the school with higher and less variable swimming speed. Weihs (1973a) developed 'wake' theory, which predicted a rigid diamond lattice structure based on five assumptions about spacing and behaviour in fish schools. He theorized that there was a hydrodynamic advantage to this type of structured schooling that would enable fish to make energy savings of up to 65%. However, extensive research on the three-dimensional structure of fish schools has failed to support this theory (Partridge & Pitcher, 1979; Pitcher & Parrish, 1993). A lateral 'push-off effect could be beneficial for lateral neighbours in a fish school, and would be maximized if the fish chose to swim next to neighbours of similar size (Pitcher & Parrish, 1993). Evidence of this tendency towards a choice of neighbour size has been found in mackerel and herring schools (Pitcher et al., 1985) and bluefin tuna schools (Partridge etal., 1983). 13 Utilizing echo sounders placed below sea cages, Ferno (1989) also observed two distinct types of fish behaviour. Fish either formed a polarized school, swimming relatively rapidly in a ring-formed horizontal distribution, or were classified as unstructured with no common orientation, low swimming speed and even dispersal throughout the cage. However the author offers a different explanation for the two behavioural modes than Juell (1989), who believes it to be density related. Ferno (1989) suggests that fresh water rearing conditions might affect the later performance of salmon in sea cages. The group of fish that exhibited unstructured behaviour was raised in small tanks with consistent current directions, and then parr from tanks with different current directions were mixed in the sea cages. The schooling fish were raised in large tanks with variable current directions, so they may not have formed such a strong imprint and were more amenable to forming a structured school with a common swimming direction. Functional density in a cage is influenced by both the behaviour and group structure of the fish and by environmental factors. Seghers (1974) found that guppies from streams with higher numbers of predators lived in tighter schools than guppies from streams with smaller predator populations. It is possible to have a local density of up to ten times higher than the mean fish density in the cage for restricted periods of time (Ferno, 1989). Unstructured fish could quite possibly experience a higher functional density than schooling fish, which swim regularly with a kind of free movable space around them. Dunn and Dalland (1993) found that fish will avoid water layers containing algae, unstable temperature or unstable salinity, which can reduce the effective volume of the cage or create barriers that prevent normal feeding, leading to more variation in fish size. 14 MATERIALS AND METHODS Equipment The filming apparatus consisted of two Panasonic WV-BD400 surveillance cameras with Cosmicar 4.8 mm 1:1.8 TV lenses in watertight metal housings. The cameras were mounted on a metal plate spaced 23 cm apart. Umbilical cords approximately 23 m long which enclose the power cable and control cables connected the cameras to a Panasonic TR-930 C B video monitor on the surface. The remainder of the equipment required for videotaping in the field consisted of a customized camera control box, a Panasonic WJ-FS 10 digital frame switcher (DFS), and a Panasonic A G -1960 s-VHS recorder (Figure 1). The videotapes were analysed in the imaging laboratory once the footage had been obtained. The computer system in the lab consisted of a 486-33 MHz DX with 16 Meg RAM, a 600 M.B. HD and an overlay frame grabber (OFG) card. This was connected to a standard A S T S V G A video monitor and a specially adapted MG-11 monochrome video monitor that has been resynchronized to the O F G card. The tapes were played back on a J V C BR-S822U S-VHS recorder with a built in time-based corrector, which also allowed for frame-by-frame playback. Data Acquisition The two underwater cameras were manually controlled from the surface by means of the customized camera control unit, which manipulated shutter speed, lens aperture, etc. The video signal travelled from the cameras via the umbilicals to the digital frame switcher, which enabled the operator to control the sequencing and speed 15 S-VHS Video Recorder Monitor 13 | 12:00 » | —I • • • • • o o o o o o o o o Digital Frame Switcher E 3 Personal Computer • Video Cameras In Waterproof Housings Weighted Base Figure 1 Video Image Capturing and Sizing System (VICASS) 16 of the two signals as they were recorded in stereo with the s-VHS recorder. The images were then grabbed from the videotape by the 486DX-33 computer and O F G video card. The next step was to correct the images that were stored as image files on the computer's hard drive for any distortions caused by the camera lens, dome port of the camera housing, and water. Finally, a program based on the theory of stereoimaging was used to process the pairs of fish images and produce estimates of fish length, height, and weight. For a complete discussion of the physics behind this theory, please refer to Petrell et al. (1997). One additional parameter was measured using the video footage, that of coasting swimming speed. A fish is said to be coasting when it stops its swimming movements, straightens its body, and coasts along in a straight line, decelerating slowly (Videler, 1993). To calculate swimming speed, the shift in position between two consecutive fish images captured with the same camera was measured over the time interval between the two frames. Measurement of current speed was beyond the scope of this research, and as such was not taken into consideration when swimming speed was calculated. The effect of current on swimming speed was taken to be negligible, as in most cases the video sampling was done under minimal current conditions (as illustrated by the nets, which were hanging straight in the water). In the few cases where the nets were billowed due to currents, the fish were always seen to be swimming at right angles to the current, which would therefore not affect the swimming speed significantly. Finally, the average current speed measured on farm sites is 0.08 m s 1 (Ok-Hyun Ahn, research data for M.Sc thesis in progress) which is not high enough to drastically alter the measured swimming speeds. 17 Sampling Procedures Video Sampling The first step in the video sampling technique was to lower the cameras into the sea cage in the centre of the cage to determine the apparent horizontal and vertical distribution of the fish throughout the cage. From this information, the bottom, middle and top of the swimming aggregation in each cage was identified and then three filming depths were chosen which corresponded to the above subdivisions of the school of fish. Next the cameras were turned around to face the net and positioned to ensure that the majority of the fish were swimming perpendicular to the cameras and were of a good density, i.e. dense enough to provide an adequate sample size, but not so dense as to layer on top of each other and obscure their silhouettes. The cameras were always positioned at the side of the cage facing the net at a distance of 2-3 m from the net, which allowed the net to be used as a backdrop, thereby sharpening the outlines of the fish by providing contrast. The second advantage of this filming position was that it encouraged the fish schools to thin out, facilitating higher quality footage. In two out of a total of fourteen cages filmed, where there were poor natural lighting conditions, a 12.2 m x 3 m white tarpaulin was suspended in the water in front of the net to act as a backdrop. Each depth was filmed for an average of 30 minutes to secure enough usable footage. Measurements of several physical parameters and a description of farming conditions, such as feeding schedules and stocking densities, were also noted during the videotaping. The environmental parameters included weather, position of the sun, visibility in water (using a Secchi disk), and water temperature, salinity (YSI Model 33) 18 and dissolved oxygen content (YSI Model 57). The last three measurements were taken at two metre intervals to a depth of ten metres. Physical Sampling A physical sample of the fish in a pen was taken using the most commonly used method employed for size estimation on commercial fish farms, known as "dip netting" or "seining". A seine net is used to capture a sample of the population in the sea cage, and then the fish are captured with a dip net, anaesthetized, weighed and measured. The target is to capture approximately one quarter of the cage population in order to get a true representative sample. Of the dip net samples, 50 of the fish had length (measured as fork length, the length from the nose to the fork of the tail), height (measured just before the dorsal fin at the tallest part of the fish), and girth measurements taken. This dip net sample served a dual purpose: first, as a check of the accuracy of the V ICASS system, which was still being tested under field conditions. Second, the physical measurements were added to a growing database of chinook and Atlantic salmon of varying sizes, that was used in the second half of this research to compare the body conformation of these two species. Sites B and C had the added advantage of being harvested soon after the video sampling. This harvest data acted as an even better accuracy check of the V ICASS system, as every fish in the cage was sized, resulting in a very accurate measure of the average fish size per cage. In addition to the physical samples obtained in the course of this study, physical measurements of both chinook and Atlantic salmon were obtained from dip net samples 19 over a wide range of sizes and under varying husbandry conditions. A total of 1539 Atlantic salmon were measured from 18 sea cages, ranging in size from 0.42 - 8.50 kg. There were 840 chinook salmon measured from 17 sea cages, ranging in size from 0.009 to 4.91 kg (data not included for either species). Sites Video footage of chinook salmon was filmed at three different locations near Vancouver Island. Site A The first site was located northwest of Tofino, off Vancouver Island. The video sampling took place over a period of three days in September, 1994 (Table 1). A total of six cages were sampled, and all six also had a physical sample of 100 fish seined and dip netted for size measurements within a week of the video sampling. All of the cages had the same dimensions, 1 5 m x 1 5 m x 1 6 m deep. The stocking densities in the cages ranged from 4.4 to 6.8 kg m"3, and the fish were all approximately the same size, between 1.5-2.5 kg. The fish had not been graded (i.e. size sorted) within the six months previous to the sampling. The six cages were all fed twice a day, once in the early morning and again in the late afternoon, and their feeding schedule did not change during the video sampling. The weather was sunny and clear for all three days of filming. Underwater visibility (measured with a Secchi disk) was fair, ranging between 4.5-5.5 m. The water temperature was fairly constant at 14°C, and salinity levels averaged 22.5 ppt (Table 1). 20 CD g 0) C < ° c 15 CO CD n E 3 (D T 3 Is 3 to o O - r-5 E .a > O ^ E <u .2 O J CD T J CD C _ ^ > CD O d d ) CO ft* E C 3 x3 T J CD CO O J C CD . £ O 3 E o E o) So - g o j ? > , c o O C -E O CD C D to -° ^ 03 c CD .2 D ) «5 ro £ ••=, - J CD -4—« CO Q T J CD CD O ) E 2 CO CD CO C L MIS' CO CO CO CO CO CO <D CD CD CD CD CD o o o o o o o8 08 o<5 o<5 08 o& >» >» >, >, >> c c c c c c c c c c c c 3 3 3 3 3 CO CO CO CO CO CO i n CO CM m i n i n c\i CM CM CM CM CM i r i CO CO j D _CD O O •i i 0 8 0 8 c c c c 3 3 CO CO CO CO 10 m i n i n i n i n i r i i r i T r i r i i r i m i n i r i i n t Tj" Tf TT CO CO CO CO CO CO c c c: c "c c .><>>.>>.><>< CO CO CO CO CO CO T ) T 3 T 3 T 3 T J T 3 CD CD CD CD CD <D O O O O O O 'S !S 'S '5 'S '2 O ) O ) O ) >«.>».>> >. CO CO CO CO T I T I T I T J CD CD CD CD O O O O C C C C O O O O > T J T J .2 S ^ O ^ O C 1 co 5? — co S co CD S- E CD S- CD J : £ b r £ E r o £ t- o -o ~ E o _ c o _ r- O E o •= - - -0 d T 0 c 0 - - r» o O O O G CI O O O E 0 o aj CD E CO _ CD 8 E s 0 .S> 0 CD © >» co > o. CO c <" 5 E 8 o — Xi O sz CO T J T J 0 ) 3 3 O 0 0 03 C O O >. 2 <* 08 C 3 3 g CO CO ^ i n . i n 00 • h~ • T - c o 00 i n i n „ m • CO CM CM T— CM c o 00 00 00 h~ r - r - r--Tf t >^ O) <J> G) O) <J> G) O O CM CM T— T-a a a a a a CD CD CD CD CD CD V) CO CO CO CO CO C O C O C M CM C O C O T— CM c o m c o I I I I 1 I <<<<<< CM CO CM Y CNI c o m m m oa -3- CO CO CO O) C7> C7> C7> CD CD CO CO T 3 T J TO TO CD CD to tn o o c o o o •o 2 >. o > ro ^ ^ co C L E E CD E L- i - I— 0 0 O £ £ C " c c g > 3 = C Z O O J - Q Ti- CM CO c o i n CM 0 CO i n CO i r i ed CD CD CO i r i i r i i r i i r i CO CO CO CO CD CO CO CD CO CO CD CD CD CD 0 0 0 0 ^ — ^ — ^ — ^ — ^ — ^ — ^ — ^ — — ^ — CM CM CM CM X X X X X X X X X X X X X X m i n m m m i n i n m i n i n CO CO CO CO ^ — x— X X X X X X X X X X X X X X m m i n i n LO m i n m 10 m CO CO CO CO m m m i n LO m m m CT> c» a> OJ g> o> cn CO 25 LO CM CM CO CM CM CM CM CM CM CM CM CM Feb Feb Feb Feb Mar Mar Mar Mar CO CO CM CO (M CO i f 1 1 1 1 O O O O 21 Site B The second site was also located northwest of Tofino, off Vancouver Island. The video sampling took place over a period of three days in February, 1995. A total of four cages were sampled, and all four also had a physical sample of 100 fish seined and dip netted for size measurements in the same time period. The four cages were also completely harvested within a month of the video sampling. All of the cages had the same dimensions, 15 m x 15 m x 16 m deep. .The stocking densities in the cages ranged from 5.0 to 5.5 kg m 3 , and the fish were all approximately the same size, between 3.0-3.5 kg. The fish had not been graded within the six months previous to the video sampling. The cages were all on a once daily feeding schedule, however they were not fed during the sampling visit due to the concurrent physical sampling. The weather was overcast and rainy for the first two days of filming, and clear for the last day. Underwater visibility was fair for the first two days (5.5 m), due to heavy rainfall during the previous week which caused high silt levels in the water around the site. Visibility improved for the last day of filming, however, reaching 8 m. There was a creek mouth adjacent to the site, which explains the low salinity levels of 13-15.5 ppt. The water temperature was fairly constant at 8°C, and dissolved oxygen (DO) levels were approximately 9 mg I"1 (Table 1). Site C The third site was located adjacent to Quadra Island, which is situated in the Strait of Georgia east of Campbell River. The video sampling took place over a period of two days in March, 1995. A total of four cages were sampled, and all four also had a 22 physical sample of 100 fish seined and dip netted for size measurements in the same time period, except for Cage 3, which only had 82 fish physically sampled. The four cages were also completely harvested within two weeks of the video sampling. All of the cages had the same dimensions, 1 3 m x 1 3 m x 2 0 m deep. The stocking densities in the cages ranged from 3.1 to 3.5 kg m 3 , and the fish were all approximately the same size, between 2.0-2.5 kg. The fish had not been graded within the six months previous to the video sampling. Two of the cages (Cages 1 & 2) were fed once daily, in the morning, while the other two cages (Cages 3 & 4) were on "starve" and were not fed at all. This feeding schedule did not change during the video sampling. The weather was a mix of sun, rain and cloud for both days of filming. Underwater visibility was excellent, with readings of 12.5-13 m. The water temperature was 7°C, salinity levels ranged between 17-18.5 ppt, and dissolved oxygen (DO) levels were approximately 9 mg I"1 (Table 1). Data Manipulation The pressure drag acting on coasting fish was calculated using the formula Df = y2CdApV2 where D f is the drag force on the fish, C d is the drag coefficient, A is the cross-sectional area of the fish, p is the water density, and V is the velocity of the fish. The power needed to overcome the drag forces was found using the standard power formula P' = y1pV'iACd where P is the mean power output and the other variables are as defined earlier. The balanced energy budget of carnivorous fish (based on published data for fifteen species of fish) can be written as 1007= ( 4 4 ± 7 ) M + ( 2 9 ± 6 ) G + ( 2 7 ± 3 ) £ (see page 5 for definitions of the variables) 23 (Brett & Groves, 1979). The total metabolism can be further broken down into standard metabolism (Ms), feeding metabolism (MF), active metabolism (MA), and the heat increment, or energy of specific dynamic action (SDA). Units of Measurement Standard SI units (mass in kilograms (kg); force in Newtons (N); power in Watts (W), length in meters (m); and time in seconds (s)) were used wherever possible. However, in many cases mass was expressed in grams (g) and length was expressed in centimeters (cm), due to the conventional usage in the fisheries and aquaculture fields. The units used in the energy budget model were kcal kg fish"1 day"1. Note also that the terms fish weight (denoted as W and used in the literature review) and fish mass (M) are interchangeable. Swimming speed (in units of m s"1) was converted to units of bl s 1 by dividing the individual fish's swimming speed by fish length (as each individual fish's length is equal to one body length). The units for K are g c m 3 , and then the value is multiplied by 100 to get values near unity. The units for B are the same, but the value is multiplied by 1000, as this gives a value corresponding to that achieved with SI units (i.e. kg m"3). Numerical and Statistical Analysis The minimum number, n, of fish images required per video sampling position <r21.962 was calculated using a standard sample size determination equation: n = ——— 24 where cr is the population standard deviation, kg, D equals the maximum acceptable difference between the sample mean mass and the true population mass (5% of the mean, kg), and 1.96 is the critical value of the cumulative normal variable at the 95% confidence interval (Zar, 1984). The sample size varied between 56 and 119 fish. Homogeneity of variance was tested using Bartlett's test for the size data and Hartley's test for the speed data. The Hartley test was used in the second instance as it is a much simpler way to test for differences between sample variances when the sample sizes for all treatments are equal, as is the case for the speed data (Pfaffenberger & Patterson, 1977). The Kolmogorov-Smirnov goodness of fit procedure was used to determine whether the swimming speed distributions (measured in both m s~1 and bl s"1) were normal. One way analysis of variance (ANOVA) was used to test for differences among sample means for both the speed and size data. If the hypothesis of equal means was rejected by the ANOVA, Tukey's w-procedure was performed to determine which mean or means were different from the others. Although the data sets can be seen to violate two of the underlying assumptions of the A N O V A procedure in several instances (i.e. normality and homogeneity of variances), A N O V A was still used to compare the means, instead of using a non-parametric test. The validity of the analysis is known to be affected only slightly by even considerable departures from normality, especially as n increases. A N O V A is also robust to substantial heterogeneity of variances, as long as all n are equal or nearly equal (as is the case with the speed data) (Zar, 1984). 25 The coefficient of variation (Vs = SDImean) and relative coefficient of variation (F r[w%] = SD/™ean ioo) (where n is the sample number) are useful for comparison of samples of those population types when mean and variance vary together (Sachs, 1984). The aquaculture industry exhibits this type of situation, where typically larger variances are associated with larger fish sizes (Sylvain Alie, personal communication). The t-test (as given in the Quattro Pro v 6.0 software package) was used to compare the average swimming speeds (in bl s"1) of the two species over a comparable size range. Regression techniques were used to find the parameters of the three length-weight models that were compared. For the model M = KL\ Fulton's condition factor (K) was estimated using a linear regression of M vs L3. In the next model, M = BL2H, B was estimated with a linear regression of M vs L2H. For M = aLh, a and b were estimated using a linear regression of log (M) vs log (L). A modified version of the t-test developed specifically to compare the slopes and elevations of regression equations was used to compare the regression equations generated for the natural log of mass versus the natural logs of girth, fork length and height for both species (Zar, 1984). The z-test (as given in the Quattro Pro v 6.0 software package) was used to compare the ratios of fork length to height and fork length to girth for the two species, and to compare the roundness of both species. The z-test was chosen instead of the t-test because n > 30 and the standard deviation was known (Bluman, 1995). All of the above statistical tests were performed at the 95% confidence level unless otherwise noted. 26 RESULTS Terminology Density The density of the swimming aggregation, or school, in the sea cage was classified as either sparse, medium, or dense according to two qualitative indices of density. The first index was related to the number of individual fish present in the cameras' viewing area at any one time, while the second was based on the percentage of the screen that was taken up by fish images (Tables 2, 3 & 4). Sparse refers to the condition where a single individual or only a few fish were present in a video frame at any instance. In other words, a situation where 10 percent or less of the screen was occupied by fish images. Medium density describes a situation in which there were always between 10 and 20 fish present in the cameras' viewing area at any one time, which is equivalent to approximately 50 percent of the screen being occupied by fish images. If the conditions were described as dense, then more than 20 fish could be detected in the cameras' viewing area at any instant, and over 90 percent of the screen was occupied by fish images. Population Distribution Patterns Several different structures of swimming aggregations within sea cages were observed during the course of this research. They were classified in this study as: bottom heavy, bottom compacted, highly compacted, bell-shaped, and uniform 27 co .9 .g 15 S £ O ) o f ^ CO ^ o ^_ o c •2 Q - . ra 2 cn —-u- C <J cn 0 <*- Q) c o ^ 0 O ) < CD £ 1 2 E ra O a> i a> C L p 0 3 ra 0 T3 0 £ E 0 £ j TO — ra 5 O c o o a . ® E C L - O O fc - O ro Q) 0 Q) £Z C C in in ( D O N <u aj it: ft: ro 0 3 ro E E E C L co ro o m o CO ^ CO csi o T — CM co a> E Q . T 3 O O T J e — F O fc X I 0) 0 0 CD to co CO CO c c. c c CD cu CD CD TD " D " O CD CD CD c c c 0 0 T _ ^ CD CD CO CO t I 03 05 ds >> C L CO >> o CD o CD o CD o (0 ion ecu ion ecu co .—' CO (0 CO CD o CD o Q . N C L N C L CO C L 'co O -*-» o O -*—• O «J E T J o T J e F O fc n E E . T J T J CD CD E E in OT fore fore fore fore Fore bel bel bel bel bel pm pm pm pm pm i n o o CM o •"3-o i n CN CO CM CM CO CM CO CM CO CM CM CM CO CO a? E - a o - a & " F O fc X I CD CO CD C CO CD C TJ T J cT CD > in ^ CD CD 4=: * ; ro co E E ro ro m o in o CM CO i i u E C L T J O o -a H ~ F O fc X I c c c C ~5 "5 "5 "5 -t—• co co ro ro CD C L C L CL C L c i— I L I— ro ra ra ro CD CD 0 CO CO CO c c c CD CD CD TJ t ) T J ^ in » ^ OT 2 0 CD CD i~ i— i— o o o t f- "4-0 0 0 X I X I X I E E E C L C L C L lO O O T - Tl; O CM CO CO T — CM CO i i i i n m i n E C L T J o O X I s — F O fc x i CD E E CD .2 . 2 <o T J T J C 0 0 0 E E " ° 0 0 0 C C C i n O CM T — \ i— CD CD CU it: it: ro CO ro E E E ro CO CO o o i n CO T — m o T — T~ x— CM CO CD CD CD 28 to 0 o 2 c <= = .Q .£ co H SZ CD (o .-i=: cu ro 2 ^ co • <= o D>0) t> cp o < CD -o o 8>c2 cn -a cn < CL a) 2 ^ ro E CD E ro O Q- c CD S-TD O CD CD CQ D) g TD CD CD •a CD £ E CD E CD CO ro 5 O c o *•«-» CO o CL CD TO CO o TD ro -«—< CD a> TD E E CD E ro ition o CD sz sh fil >. 0 CO -*—' <= ro o CL CD with ugh ccur c: CO o ro o E c CD t= .>< CD o cz XI o o o i c SO TD CL TJ CL 0 ro CD -t—< Q> O o O TD CD CD T J q= tde tecf tde E CD CD CD N c w 4—* c CO TJ TJ CD (O 1= CD T J E v E Q. o CL TD O o is o n s ~ O ~ F O XI t XI CD CD CO (O 2-1 CO CD t CD co 2 co TO "° £ CL CD CD co c: TJ -4—• CD 0 c -«—' 0 C •4-» 0 c CD m CN o CO TJ 2 TJ TJ 2 TJ o cr o c o c -4—< O c pm pm pm pm o o o CN o o o CO CN CO T — CN CO T— CN CM CO E E o o in in CN T-* CN CO TJ 0 E o cz o '-*—' CO o CL E ' o _ £ >, o ro XI TJ • - M— TJ O 2 * o o 0 ro cn 0 TJ 0 ZJ TJ 0 CL TJ O TJ 0 0 w io a s CO 0 0 in in co r-^  TJ TJ 0 m O c E E CL ea-rn o o co t - CM • i 29 CO CD C ° .2 .<= "co co .-t; £ 5 OJ CD i_ O) O) CO o *-*—» CO . cn -CD o ^ o X _ E - c o O) CD C cn -e o c o CD "O < Q. CD 2 CL *- TJ ro ro CD E ro O CL C TJ O CD CD CQ cn CD u C CD CD £ O cn — ro S co O o CL "O CD 5 o 2 CD > -4—' CD c .a E CL T3 O O TJ Hi — F O fc _Q CD E 0 0 2 co co "D <= C 0 CD CD C "CJ "O CD 0 CD c c c CO LO OJ .2 E CL TJ O o TJ ±2 -~ F O C .Q E E E .2 .2 .2 TJ TJ TJ CD CD CD E E E CD CD CD c c c CD s CD 1 CD 1— 0 i 0 s CD * i it; CO ro ro ro ro ro pm pm am pm pm pm o o LO o o o x — LO LO CN T— LO CO CN CO CN CO CN CN CN ™ 2 ro ^ k 'w £ ci " 8" CD -2 CO . -ro g "D -*-< 33 ro CD £ •«= TJ -C CD CO CO ro -P. 0 o M— 0 ro g> x: 5 CO E 0 2 CL ro o 'c x: o 0 E o o X3 0 0 CO CO c cz 0 0 TJ TJ 0 0 C C CN TJ TJ 0 .0 o c ro LO E ro LO CO CN CO I I CO CO TJ 0 o 0 .2 E CL TJ O C O S — F O t XI E 3 0 _ _ W CO TJ S Q. 0 0 co cr TJ 0 0 0 c c c LO TJ TJ TJ .0 .0 .0 O E E E CL CL CL LO O LO T O CO CN CN t- CN CO i i i •"3- "tf 30 (Table 1). Bottom heavy describes the situation where most of the fish are to be found in the bottom half of the cage, with only a few individuals near the surface. Bottom compacted means that all of the fish are swimming together in the bottom third of the cage, with no obvious density change within the school. Highly compacted describes essentially the same phenomenon as bottom compacted, but the school does not occur in the bottom of the cage. Bell-shaped describes a swimming aggregation where the fish swim around the perimeter of the cage creating a hole in the cage center. The bottom heavy aspect refers to the fact that concentrated numbers of fish are found in the bottom of the cage. A uniform distribution is characterized by an even dispersal of fish throughout the cage, in both the horizontal and vertical planes, but is not indicative of a random distribution offish. Position Within the School The three positions in each cage are referred to as P1 , P2, and P3. P1 was filmed at the top of the school, the shallowest depth at which fish were found in a cage. P2 is the middle filming position, or mid school. Finally, P3 was filmed at the bottom of the sea cage, or bottom of the school. As these three filming positions were dictated by the vertical position and extent of the swimming aggregation within a cage, the depths of the filming positions vary between cages. For example, at Site A, Cage 1 had a moderately compacted and bottom heavy swimming aggregation, so that the top of the school (P1) was found at 6.5 m (Tables 1 & 2). However, at Site C, Cage 2 had a swimming aggregation with a vertical extension of 11 m and a uniform distribution. This resulted in the top of the school occurring at 4 m depth (Tables 1 & 3). All of the 31 positions except two were filmed with the camera lenses pointing at the side of the cage facing the net approximately 2-3 m from the net, therefore utilizing the net as a backdrop for filming (Tables 2, 3 & 4). The two exceptions (B1-2 and C4-1) were filmed with the cameras in the centre of the cage, facing the side, due to a lack of fish swimming around the perimeter of the cage. Size Distribution Site A In summary, four out of the six cages sampled showed significant differences in average fish size between positions. The differences ranged from 13-28%. Seven out of a total of 16 positions were significantly different from the dip net. The value of the video sample ranged from 24% less than the dip net to 17% greater than the dip net (Table 5 & Appendix A-1). One out of the six cages (Cage 3) showed a significant difference between the variances of the video positions and the variances of the video positions and the dip net (Appendix A-1). The coefficients of variation ranged from 0.19-0.27, while the relative coefficients of variation were between 1.92 and 2.59%. For Cage 1, all four sample means (P1=6.5m, avg =1.83±0.43 kg; P2=10.5m, avg.=2.00±0.41 kg; P3=12.5m, avg.=1.91±0.51 kg; and the dip net, avg.=1.99±0.41 kg) were not significantly different from each other. For Cage 2, the sample mean of P3 (14m; avg.=1.58±0.35 kg) was significantly different from the means of the other two positions and the dip net. The dip net (avg.=2.02±0.39 kg) and P2 (10m; avg =2.02±0.41 kg) did not have significantly 32 i n 00 * — CO co o 00 Tt cn o •* Tt Tt CM CO CO T— o 00 m i n O co Tf Tt m CM csi CO C\i CM csi CM csi csi csi CM CM CM csi CM Tt Tt Tt co CM cn i n CD CN CM CO CM CM CM CM CM CM CM T — CM .— CM CN CM CM CM CN o O d c i d c i c i c i c i c i d d d d d XI ro CO ra ra XI CO XI CO ra ra CO XI o ra ^ — CM CD CO ro XI cn CO i n i n N- r- i n m i n i n i n m co i n CO Tt Tt m m i n i n o d c i c i c i i n d i n c i d d d d d d CJ +i •H +l +1 CO +i c> +i c i XI +1 +i +1 CO •H -H -H +i XI CN Tt CD Tl- CM 00 -H CO M .— CO o> rv. co f - 00 co 00 CO CO CM CM v— CO o CM CM o T . — co t T - Tt CM c\i C\i csi csi CM CM csi CM CM CM CM CM csi csi CM csi CM csi csi (O o ) 2 o m cn i f CM ^ i n co o cu to CM co c 3? - - ° - £ oo cn o o ^ N CM CM a 13 j oo •<-O T -m CM Tt oo o m cn Tt CM co c § CO CO C L J= ro SZ •st Tl- Tl- O . "O o CO CM m OO Tj- Tt T— 00 o CO CN T T Tf 00 1--; 00 T T i n CO Csi csi csi csi csi csi csi csi csi csi csi csi CD CM co Tf CD m Tf 00 co CM CM CM CM CM CM CN CN CN CN CN CM d d d d d d d d d d d d x> XI XI ro ro ra co XI ro ra ro ro ro ro ro CO CO o CO ra o T— oo T T o o CD 00 00 00 00 00 00 a> cn 00 d d d d d d d d d d d d -H -H CO +t •H -H XI •H •H -H M CO -H -H -H XI Tt T - CO CD cn t-~ Tf T~ Tf cn cn CM a> i— T— CO CN CN cn CN CO T T CM CM Tt T T CO T— CO i n co CO CO CO csi CO CO CO CO CO CO CO CO CO CO CO <-> co cn 9 Ti-oo ° •«-C D in CM c « " # « x : r- CN CO C I I I T ~ T ~ 1 - Q . TJ 00 S 9 ™ o o .— r— m Tt co cn 9 i n oo oo oo — i n ^ C D co m o o o o •>- CO * cu ™ tM .9- % "O x : — CO T — CM CO c CO CO CO Q . TO T — CM C CO Tt CO co i n T T i n co o i n m CO CN o Tt T T CN cn CN oo m CN m o t— CM CO cn i n CM CO i n i n CO i n i n i n i n i n co csi csi csi Cvi csi csi csi T~ csi csi csi csi csi csi CN csi csi CN Cvi csi CO o o CM cn CM cn Tt CN CM CO CO CO CN cn CO CN CN CM CM CN CM CN CM T — CN CM CM CM CM CN CM CN CM CM d d d d d d d d d o d d d d d d d d d d d CO ra ra ro ca ra o XI CO CO T — T x— r>- i n cn cn Tt Tl- uo Tf co Tl- CO co Tf d d d d d d d d d •H -H -H -H +l -H •H -H -H co O * — cn 1"- CN 00 CM cn CO o cn o> 00 O m O T — T- csi , " _ T- csi csi csi CD Tf ,— o i n CN cn o o oo CO t— o 1— 00 00 00 x— co XI XI CD Tl-CO Tf d d -H -H CO t~-cn oo o o x i o co i n co 00 TT Tf " d d O -H -H -H CD N O CO T ™ CS| T - " oo i n 9 r~- ° i - CM co CN CN CM cu CM CO c CO CO Q. T3 TI- a. T3 co co ro ra co co oo i n Tl" Tf co co d d d d •H -H -H -H oo oo m CM C O 00 N CO CO T - CO 9 CO CO h-i n i n m o . T3 o ro x i o co f— CM Tl" CO CO CO CO Tt d d d d -H -H -H -H CO CM T - oo I— Tt co oo oo o cn 9 to CO s ^ T - CM CO to co co 33 different means. The mean of P1 (8.5m; avg.=1.87±0.37 kg) was not significantly different from P2, but it was significantly different from the dip net (Appendix A-2). For Cage 3 only two positions were used. P2 (9m; avg.=2.19±0.49 kg) had a significantly different mean from P3 (14.5m; avg.=1.93±0.36 kg) and the dip net (avg.=1.87+0.44 kg), but P3 and the dip net did not have significantly different means. For Cage 4 only two positions were used. All three sample means (P2=9.5m, avg.=1.70±0.38 kg; P3=14m, avg.=2.06± 0.45 kg; and the dip net, avg.=1.87±0.43 kg) were significantly different from each other. For Cage 5, all four sample means (P1=7m, avg.=1.88±0.43 kg; P2=9.5m, avg =1.88±0.43'kg; P3=12.5m, avg =1.75±0.38 kg; and the dip net, avg.=1.82±0.35 kg) were not significantly different from each other. For Cage 6, P2 (10m; avg.=1.42±0.32 kg) had a significantly different mean from the other two positions and the dip net. The dip net (avg.=1.88+0.43 kg) and P1 (7.5m; avg.=1.73±0.37 kg) did not have significantly different means. P3 (12m; avg.=1.61+0.34 kg) did not have a significantly different mean from P1, but it was significantly different from the dip net. Site B In summary, none of the four cages sampled showed a significant difference between positions. None of the total of eight positions were significantly different from the dip net. Two out of the eight positions were significantly different from the harvest data. In both cases, the value of the video sample was 10% less than the harvest data (Table 5 & Appendix A-1). 34 None of the cages showed a significant difference between the variances of the video positions, dip nets and harvest data (Appendix A-1). The coefficients of variation ranged from 0.21-0.28, while the relative coefficients of variation were between 2.21 and 2.88%. For Cage 1, only one position was used. None of the means (P2=9m, avg.=3.14±0.83 kg; dip net, avg.=3.31±0.73 kg; and the harvest data, avg.=3.23 kg) were significantly different from each other. For Cage 2, only two positions were used. P1 (2-3 m; avg.=3.26+0.70 kg), P3 (10m; avg =2.99±0.83 kg) and the dip net (avg =3.27±0.80 kg) were not significantly different from each other. The harvest data ( avg.=3.34 kg) was not significantly different from P1 and the dip net, but it was significantly different from P3 (Appendix A-3). For Cage 3, all five sample means (P1=3m, avg.=3.41±0.90 kg; P2=7m, avg =3.24±0.81 kg; P3=12m, avg.=3.29±0.88 kg; the dip net, avg =3.49±0.84 kg; and the harvest data, avg.=3.42 kg) were not significantly different from each other (Figure 2). For Cage 4, only two positions were used. The mean of P2 (7.5m; avg.=3.17±0.90 kg) was not significantly different from the mean of P1 (3.5m; avg.=3.39+0.90 kg) and the dip net (avg =3.31 ±0.86 kg), but it was significantly different from the mean of the harvest data (avg.=3.51 kg). The means of P1 , the dip net and the harvest data were not significantly different from each other. 35 4 Individual Fish Mass (kg) dip net - ® - harvest - • - P1 — i — P2 -^w- P3 Figure 2 Frequency polygon offish size distribution for Cage 3, Site B, illustrating a case where there is no statistical difference between any of the five samples. 36 Site C In summary, two out of the four cages sampled showed significant differences between positions. The differences ranged from 11-22%. One out of a total of eleven positions was significantly different from the dip net. The value of the video sample was 11% less than the dip net. Two out of a total of eleven positions were significantly different from the harvest data. The value of the video sample ranged from 8% less than the harvest data to 17% greater than the harvest data (Table 5 & Appendix A-1). Two out of the four cages showed significant differences between the variances of the video positions, dip nets and harvest data (Appendix A-1). The coefficients of variation ranged from 0.16-0.25, while the relative coefficients of variation were between 1.58 and 3.25%. For Cage 1, none of the five sample means (P1=3m, avg.=2.32±0.51 kg; P2=5m, avg.=2.14±0.52 kg; P3=9m, avg =2.36±0.56 kg; the dip net, avg.=2.24±0.53 kg; and the harvest data, avg.=2.22 kg) were significantly different from each other. For Cage 2, P1 (4m; avg.=2.18±0.50 kg), P2 (7m; avg.=2.30±0.51 kg), the dip net (avg.=2.20+0.55 kg) and the'harvest data (avg.=2.21) did not have significantly different means. P3 (11m; avg =2.03±0.39 kg) did not have a significantly different mean from P1 and the dip net, but it was significantly different from the harvest data and P2 (Appendix A-4). For Cage 3, only two positions were used. None of the four sample means (P2=9m, avg =2.13±0.35 kg; P3=12m, avg.=2.09±0.45 kg; the dip net, avg.=2.17±0.47 kg; and the harvest data, avg.=2.16 kg) were significantly different from each other. 37 For Cage 4, P3 (11m; avg =2.18±0.51 kg) and the harvest data (avg =2.28 kg) did not have significantly different means. The mean of P1 (7m; avg.=2.67±0.57 kg) was not significantly different from the mean of the dip net (avg.=2.46+0.55 kg). P2 (9.5m; 2.41+0.57 kg) did not have a significantly different mean from the harvest data and the dip net, but it was significantly different from P1 and P3 (Figure 3). When the data from all three sites was combined, five out of the 14 cages showed significant differences between the average masses of the dip net and at least one of the video samples. Of those five cages, in three cases the dip net was not significantly different from the largest subpopulation in the cage. In one cage the dip net sample was not significantly different from the smallest subpopulation in the cage, and in the last cage the dip net sample was significantly different from both of the video samples, but fell in between them. Swimming Speed The average swimming speeds calculated for chinook salmon (within the size range of 1.5-3.5 kg) were between 0.37-1.06 m s"1 (0.72-2.04 bl s"1). Average speeds for Atlantic salmon (between 1.0 and 5.5 kg) were in the range of 0.25-0.98 m s 1 (0.40-1.82 bl s 1 ) . The average swimming speeds of the Atlantic salmon were calculated from data presented in Shieh, 1996 (Tables 6.1 & 6.2). Chinook salmon Average swimming speeds were calculated for ten pens of chinook salmon, or a total of 25 different positions (i.e. different depths within the pens). Speeds were 38 2 3 4 Individual Fish Mass (kg) dip net -m- harvest - P1 — i — P2 P3 Figure 3 Frequency polygon offish size distribution for Cage 4, Site C, illustrating a case where there are statistical differences between the five samples. 39 TABLE 6.1 Atlantic Salmon Swimming Speed Data (adapted from Shieh, 1996) Site with Date Time Camera Water Position Average Average Average cage and filmed depth temperature within fish speed speed position # (degrees C) aggregation size (kg) (m/s) (bl/s) H1-1 Nov 24/93 11:00 am 10 7 middle 4.68 0.28 0.40 H1-2 9:45 am 5 8 . fop 5.36 0.37 0.50 H1-3 12:07 pm 8 7 middle 4.64 0.28 0.40 H1-4 3:00 pm 8 7 middle 4.60 0.60 0.80 H2-1 Nov 24/93 3:05 pm 3 8 top 5.25 0.55 0.75 H2-2 3:20 pm 5 7 top 5.26 0.42 0.56 H3-1 Nov 24/93 11:20 am 6-7 7 middle 5.22 0.25 n/a H4-1 July 9/94 5:50 pm 7.5 8 mid/bot 4.65 0.61 n/a H4-2 6:40 pm 4 11 top/mid 4.63 0.59 n/a H5-1 July 9/94 2:30 pm 8 8 mid/bot 4.46 0.61 n/a H5-2 3:25 pm 3.5 11 top 4.51 0.63 n/a 1-1 Mar 24/94 3:15 pm 15 7 mid/bot 0.91 0.79 1.82 1-2 3:42 pm 15 7 mid/bot 1.15 0.71 1.48 2-1 Mar 26/94 2:50 pm .- 15 7 mid/bot 1.6 0.29 n/a 2-2 3:20 pm 15 7 mid/bot 1.54 0.32 n/a 3-1 Mar 28/94 2:35 pm 8 7 top/mid 1.33 0.31 0.66 3-2 3:15 pm 18 7 bottom 1.43 0.28 0.55 3-3 4:00 pm 15 7 mid/bot 1.39 0.27 0.53 4-1 May 10/94 12:26 pm 15 8 bottom 1.35 0.67 1.32 5-1 May 11/94 12:26 pm 7 8 middle 1.47 0.39 0.74 6-1 May 10/94 4:23 pm 7 8 middle 1.64 0.76 1.41 7-1 May 10/94 11:45 am 7 8 middle 1.46 0.95 1.81 8-1 May 10/94 2:06 pm 7 8 middle 1.33 0.90 1.74 9-1 May 11/94 4:08 pm 6-7 7 middle 1.57 0.75 1.42 n/a: Not available. 40 TABLE 6.2: Atlantic Salmon Swimming Speed Data (adapted from Shieh, 1996) Site with Date Time Camera Water Position Average Average Average cage and filmed depth temperature within fish speed speed position # (degrees C) aggregation size (kg) (m/s) (bl/s) 10-1 July 6/94 4:45 pm 7.5 8 mid/bot 3.77 0.86 1.29 10-2 5:30 pm 6 10 middle 4.02 0.98 1.46 11-1 July 7/94 12:00 pm 2.5 12 top 3.92 0.42 0.60 11-2 1:05 pm 5.5 10 middle 3.84 0.83 1.21 11-3 2:00 pm 3 12 top 3.92 0.61 0.88 12-1 July 6/94 2:00 pm 3 12 top 3.76 0.74 1.09 12-2 2:40 pm 8 8 bottom 3.82 0.55 0.80 12-3 3:20 pm 5 10 middle 3.61 0.41 0.61 13-1 July 5/94 4:00 pm 3,5 11 middle 3.73 0.79 1.17 13-2 4:40 pm 8 8 bottom 3.59 0.78 1.19 13-3 5:30 pm 1.5 14 top 3.77 0.76 1.14 14-1 July 9/94 3:20 pm 5 10 top 3.36 0.70 1.08 14-2 4:00 pm 10 8 bottom 3.40 0.63 0.98 14-3 4:40 pm 7 9 middle 3.49 0.71 1.08 15-1 July 9/94 11:46 am 2 13 top 3.52 0.7:0 1.07 15-2 12:30 pm 5 10 middle 3.40 . 0.60 0.93 15-3 1:35 pm 9 8 bottom 3.15 0.70 1.10 16-1 Aug 26/94 12:50 pm 3 12 top 4.64 0.61 0.85 16-2 1:35 pm 11 8 bottom 4.58 0.73 1.03 17-1 Aug 26/94 2:30 pm 5.5 10 top 4.61 0.54 0.76 17-2 3:45 pm 10 8 mid/bot 4.81 0.61 0.85 18-1 Aug 26/94 10:35 am 11 8 bottom 4.22 0.71 1.04 18-2 11:35 am 1.5 14 top 4.32 0.85 1.26 19-1 Aug 25/94 4:20 pm 9 8 bottom 5.28 0.68 0.91 19-2 5:50 pm 2.5 12 top 4.62 0.61 0.85 20-1 Aug 25/94 12:40 pm 3.5 11 top 3.96 0.61 0.91 20-2 1:50 pm 8 8 middle 4.11 0.56 0.81 20-3 2:50 pm 10 8 bottom 4.01 0.67 0.98 41 calculated only on fish that were swimming at more or less the same speed as the rest of the school: in other words, fish that were darting or swimming particularly slowly were not used, so as to get an average swimming speed for each position that was as accurate as possible . In each case, average speeds were calculated in both m s~1 and bl s \ The first important question regarding these data was whether or not the fish were affected by the introduction of the cameras into the sea cage, reflected by an increase in swimming speed (Figure 4). When the data for each position were plotted on a linegraph as a function of time, it became clear that in some cases there was an obvious effect. In an attempt to mitigate this, only the last 50 swimming speeds for each position were used, as in all cases the swimming speeds were apparently randomized by that point (Figure 4). At first glance, many of the swimming speed frequency polygons appeared to be left-steep (flat on the right) distributions (Figure 5). One out of the 25 positions was found to have a swimming speed distribution that was non-normal when calculated in m s 1 , but normal when the swimming speed was expressed in bl s"1. None of the other positions were found to significantly differ from a normal distribution. Five out of nine cages showed significant differences between the variances of swimming speeds at different positions within the sea cage for both sets of data (i.e. m s 1 and bl s"1). In one cage, there was a significant difference in variances for the data measured in m s 1 but it was not significant for the data in bl s~1. The remaining three cages did not have significantly different variances for either set of data. ANOVA 's were performed on the swimming speed data from each position to 42 0 -I 1 1 1 1 1 1 1 _+-0 20 40 60 80 Temporal sequence of fish sampled Figure 4 The effect of camera introduction into a sea cage on swimming speed. Strong evidence of increased swimming speed due to camera introduction in Cage 5-P1, Site A. The arrow indicates the point at which the swimming speeds become uniform. 43 40 30 c 20 CD cr cu 10 0.37 0.64 0.91 1.18 1.45 1.72 Swimming speed (m/s) 1.99 2.26 2.53 Figure 5 A frequency polygon of swimming speed for Cage 5-P1, Site A, illustrating the left-steeped, flat on the right appearance of the distribution. 44 determine if the fish swam at different speeds at different depths in the sea cage. Eight out of a total of nine cages showed significant differences in swimming speeds at different depths (Appendices A-5 & A-7). Cage 5 and 6 are the only cages from Site A where swimming speed was measured, as the other four cages were analysed earlier using prototype software that was incapable of calculating swimming speeds for individual fish. In Cage 5 it was found that the fish were swimming fastest at the bottom of the aggregation (0.94±0.33 m s'1). Swimming speeds at the top and in the middle of the aggregation did not differ significantly from each other (0.79±0.17 m s"1 and 0.82±0.21 m s"1 respectively). In Cage 6, the fish were swimming fastest at the top of the aggregation, and slowest at the bottom. The average speed at the top was 0.90±0.22 m s 1 , which was not significantly different from the swimming speed in the middle of the aggregation (0.82±0.19 m s 1 ) , although it was significantly different from the average speed at the bottom (0.74±0.21 m s 1 ) . The middle and bottom speeds were not significantly different from each other (Table 7). For Site B, Cage 1, there was only one position sampled. The average swimming speed of fish located in the middle of the aggregation was 0.59±0.19 m s"1. In Cage 2, fish were swimming fastest at the bottom of the aggregation, at 1.06±0.31 m s \ which differed significantly from their speed of 0.69±0.18 m s 1 at the top of the aggregation. Cage 3 also exhibited the fastest average speed at the bottom of the aggregation, 0.85±0.17 m s"1, which was significantly different from the swimming speeds at the other two positions. The slowest fish were at the top of the aggregation (0.57±0.21 m s"1), and their speed did not differ significantly from the speed at the middle of the aggregation (0.66±0.27 m s"1). Cage 4 was the only cage studied that did 45 * O § 1 %.g -2 8 ^ CD o 1 o i 8 5 o CD L _ CD fl) Q. > CO co jo CD o o CD -iZ 8 5 C= c CD g 1 o l S 5 o co 0 u. CD 0 Q. > CO co o «= 0 £ O c n ~ ro 5 "w O o in CO CM CD CM •sl- CM CO 00 CO ed iri CO CO CO co CM CM co CN CM CM d d d d d d ro ro .a r-- o> T -n co s o d d +1 -H -H CM CM CM CD CO CD O +1 ro .a i - CO CM CO d d +1 +1 cr> CM h- oo cn o d d o o o LO LO lO CM CO CM oo S co CO ^ LO T— T— CO CJ) CM o CD LO CD Ti- CO o CN CO ed CO •<* m in CO CM CM CO CM CN CM d d d d d d CO 19ab XI CM 19ab T— CN 19ab CN d -H 82+0. d -H o cn 82+0. •SJ-d d d o o o LO LO LO T - CN CO I I I CD CD CO 4 6 not have significantly different average speeds. Fish at the top and middle of the aggregation swam at 0.74±0.29 m s"1 and 0.71 ±0.20 m s"1 respectively (Table 8). For Site C, Cage 1, the fastest speed occurred in the middle of the aggregation (0.57±0.17 m s"1), and was significantly different from the other two average speeds. The bottom of the aggregation exhibited the slowest average speed (0.45±0.17 m s~1), which did not differ significantly from the speed at the top of the aggregation (0.49±0.13 m s"1). The bottom of the aggregation was also swimming slowest in Cage 2, at 0.50±0.14 m s 1 , which did not differ significantly from the speed at the top of the aggregation (0.52±0.15 m s 1 ) . The fastest position was the middle of the aggregation, 0.80±0.18 m s 1 , and it differed significantly from the other two positions. Cage 3 also displayed the slowest swimming speed at the bottom of the aggregation, 0.37±0.12 m s"1. This was significantly different from the swimming speed in the middle of the aggregation, 0.71 ±0.22 m s"1. In Cage 4 all three speeds were significantly different from each other. The fastest speed occurred at the top of the aggregation (1.05±0.23 m s"1), while the slowest speed was in the middle of the aggregation (0.60±0.15 m s"1). The speed at the bottom of the aggregation was 0.92±0.28 m s"1 (Table 9). Although positional differences in swimming speed are only discussed here as measured in m s'\ statistical analysis of swimming speeds in bl s 1 was also performed (Appendix A-6 & A-8). This analysis was not presented here as the results were identical for A N O V A and very similar for both the Hartley test and the Tukey test. 47 CD o J9 0 CD O <- o c o c: CD g o ^ -r-"1 X >- CD CD C L > CO CD CD O CD O >- O c o CD c o it o £ S 5 o ^ -r-TO CD CD 0 C L S w ro O x: • CO o «= CO E T f o L O C D C N C O I S - C M co T f C N co T f C M T f C O X T L O C O iri T f C D C N C N O C O T t C O o T t C O C M o T f o C O O d O ci d d d d O C O ro C N C O X I h-L O ro L O C O ro C D T J X I C O C O CO C M m ro o T f ,04±0. o +i o C M o +1 C D 0 0 C D +1 C O o d +i m d +i C D T f d +i o C O d +i C O C O co T f T - C D r-0 0 C O 0 0 0 0 in o o •st C O L O L O C M iri T f C N C O C D C M C D C M h-C O C M T f o C N C D C O 0 0 C M o O O d d d d d C D ro C O x— X I C O ro C M ro h-C N X I i - - ro C D C M ro o C M 0+69 O +1 O ) C D d +i C D o d +1 m d +1 C D co d +1 in 0 0 d +i T f d +i o O T — d d d d d o C O o C O o L O o L O o in o L O o in o in £ .2 ro $ <o O o C L C M T - C O I I C M C M T - C N 0 0 C O C O C O T - CM I I T f T f 48 CO o CD O •- o crCD CD O O CD T3 1- CD CD Q_ > CO CO CD O CD O •- o 0) o it CD O O 0 3 - r -w- CD CD CL co o x: • CO o «= CO O o C L 0 CO O CN CO •sr OJ 00 0 CO 00 LO •sj-co OJ CN LO CD 1--T f van --- . co •sr LO CO CO T * T f CO CO T f van cr 0 *•*-« CO CD CN OJ CM co CO CO CM CM CM OJ CN CM CO CO CO CO CN CO CM CN CO van d d d d d d d d d d d van 25 co LO CM XJ CO CO CO •sj-CO co 00 CM XJ T T CO CO OJ CN CO LO T t XI T f CN co 00 XJ CO CO OJ LO d +1 LO CD d +1 CO T -d •H OJ 00 d +1 CM O d +1 00 LCJ d +1 0 0 d +1 00 CO d +1 CM d +1 T J -0 d +1 OJ d +1 CD 00 d T — d x — T - d CN X— tr 0 + J CO OJ CO r- CM CM OJ 00 CO OJ 00 CO CO T t 0 CO CD CN CO van CO •4 LO T J CO CO T f CO CO T f van cr O *.*-» co CO CM OJ CM CO 0 CO CO CM 00 CN CO CO CM CM CD CM CO van d d d d d d d d d d d van CO CO X> co co LO X— XJ CO CO co CN CM XI CN x — co CO CN XJ LO 0 00 CN d +1 OJ T f d +1 LO d -H LO ••a-O +1 CN LO d +1 0 00 d +1 0 LO d +1 r-O +1 h-CO d +1 LO 0 d +1 0 CD d +1 CN OJ d d d d d d d d x — d d 0 LO 0 LO 0 LO 0 LO 0 LO 0 LO 0 LO 0 LO 0 LO 0 LO 0 LO * t ion X— CM CO , - CM CO CN CO x - CM CO -4—* CO T — CM CM CN CO CO T T 4 4 49 Atlantic and chinook salmon The effect of time of day on swimming speed (m s"1) was analysed for both Atlantic and chinook salmon. Regression analysis produced an r2 of 0.02 for chinook salmon and 0.05 for Atlantic salmon, so it was concluded that time of day had no bearing on the average swimming speed, at least over the time period that we looked at (0900-1900 hrs) (Figure 6). The relationship between average fish size and average swimming speed (in m s 1 ) was also investigated for both species (Figure 7). The correlation coefficient for chinook salmon was -0.07 and -0.04 for Atlantic salmon, which indicated no correlation between size and speed measured in m s~1. However, there was an obvious relationship between average fish size and average swimming speed measured in bl s"1 for both salmonid species. When the data from the two species was combined, the correlation coefficient was -0.72 (Figure 8). The trend was for the fish to swim more slowly as they became larger. If we look at the swimming speed data for both species over the same size range (i.e. 1.0 - 3.5 kg) in terms of bl s"1 (Tables 6.1, 6.2, 7, 8, & 9) an interesting trend becomes apparent. Of a total of 43 average swimming speeds measured for Atlantic salmon, 17 were measured on fish within this size range. All of the 25 average speeds measured for chinook salmon were measured on fish within this size range. When all of the average values were combined, the overall result was that Atlantic salmon swam 1.16 bl s"1, while chinook salmon swam 1.38 bl s"1. When a t-test was performed on these average swimming speeds, a significant difference was not found between the two species at the 95% confidence level, but the speeds were found to be significantly 50 900 1100 1300 1500 Time of day (hrs) 1700 1900 900 1100 1300 1500 Time of day (hrs) 1700 1900 Figure 6 Variation in swimming speed with relation to time of day for (a) chinook salmon and (b) Atlantic salmon. Each point represents the average swimming speed for a position. 51 2 3 4 Average fish size (kg) 1.1 1 I (b) wTJ.9 | 0.8 + g.0.7 CO 0)0.6 | 0.5 I 0.4 0.3 0.2 _ l 1 1 , h 2 3 4 Average fish size (kg) Figure 7 Variation in swimming speed (measured in m s 1 ) in relation to average fish size for (a) chinook salmon and (b) Atlantic salmon. Each point represents the average swimming speed for a position. Note that there was no data available for Atlantic salmon between 2 and 3 kg in size. 52 2.2 2 1.8 4 A A AC |T1.6 1.4 cu CD &1 .2 E E 0% 0 .8 0.6 0.4 0.2 1 4-C A A * \ A C C C C A ^ a A A A A ^ A C A A A /A _| , j 1 (_ 2 3 4 Average fish size (kg) A A Figure 8 The relationship between swimming speed (measured in bl s 1 ) and average fish size for chinook (C) and Atlantic (A) salmon. 53 different at the 90% confidence level. Although there was a great deal of variability among swimming speeds at the same temperature, the general trend was for swimming speed to increase with increasing water temperature for both Atlantic and chinook salmon (Figure 9). Neither species showed any relationship between swimming speed and average depth (Figure 10). There was also no evidence of any relationships of a positional nature in individual sea cages. In other words, there were no trends between swimming speed and the top, mid, and bottom positions or between swimming speed and the largest, medium, and smallest positions. For chinook salmon, the type of swimming aggregation seemed to have an effect on the distribution of swimming speed within a cage. In general, the most compacted part of the swimming aggregation was seen to be swimming the fastest (Tables 1, 2, 3 & 4). In the cases of uniform distribution, the fish were observed to swim fastest in the middle of the aggregation. There was no evidence of a relationship between type of swimming aggregation and swimming speed for Atlantic salmon, however. Body Dimensions The shape data included in the database was used for two purposes: the first was to investigate the relationship between the different physical dimensions and mass, and the second was to compare the body shapes of Atlantic and chinook salmon. 54 Figure 9 The relationship between swimming speed and average water temperature for (a) chinook and (b) Atlantic salmon. Each point represents the average swimming speed for a position. The line of best fit was generated by simple linear regression of the data. 55 6 8 10 12 14 Average depth (m) 20 8 10 12 Average depth (m) 14 16 18 Figure 10 The relationship between swimming speed and average depth for (a) chinook and (b) Atlantic salmon. Each point represents the average swimming speed for a position. 56 Length-Weight Models Three different models for calculating mass were compared to find the most accurate model to compare a state of 'condition' over a wide range of fish sizes. The three models were: M = KL3, M = aLh and a newly proposed model, M = BL2H. In every case, there was a tighter relationship (expressed as the coefficient of determination, r2, from the regression equation) using M = BL2H than either of the other two equations (Tables 10 & 11). This can also be seen when the scatter diagrams for mass generated with B and K are compared to the true measured mass. For both species, there is a tighter relationship with B than with K (Figures 11 & 12). The average values of B for each of the 18 sea cages ranged from 51.95 - 61.81 for Atlantic salmon and from 56.55 - 69.09 for the 17 sea cages of chinook salmon. The overall average B value for all of the fish was 62.4 for chinook salmon and 56.8 for Atlantic salmon. The range of average K values was 0.97 - 1.36 for Atlantic salmon and 1.22 -1.76 for chinook salmon. When standardized K and B values are plotted against mass for both species (Figures 13 & 14), two observations can be made. Firstly, as already mentioned, there is considerably less scatter for the B values than for the K values. Secondly, the B values do not show any dependency on or relationship with size, whereas it is obvious that the K values decrease with increasing size in chinook salmon and increase with increasing size in Atlantic salmon. One possible reason for this is illustrated in Figure 15. The relationship between mass and length cubed is non-linear, which means that the K values that describe the relationship between these two variables will change with fish size (Figure 15a). When the relationship between mass 57 True mass (g) Ideal case line + Cal. mass, K True mass (g) Ideal case line + Cal. mass, B Figure 11 Scatter diagrams showing the variability of calculated mass around true mass for chinook salmon, (a) Increased variability of mass calculated using K as compared to (b) mass calculated using B. 58 Ideal case line + Cal. mass, K Ideal case line + Cal. mass, B Figure 12 Scatter diagrams showing the variability of calculated mass around true mass for Atlantic salmon, (a) Increased variability of mass calculated using K as compared to (b) mass calculated using B. 59 1.6 (a) 1.4 + 1.6 (b) 1.4 * 1.2 -E . > ^ 1 0.8 .-jf . •- -0.6 1000 2000 3000 Mass (g) 4000 5000 Figure 13 Scatter plot of (a) standardized B values and (b) standardized K values against mass for chinook salmon. 6 0 2000 4000 6000 Mass (g) 8000 10000 Figure 14 Scatter plot of (a) standardized B values and (b) standardized K values against mass for Atlantic salmon. 6 1 CO CO CD 5000 4000 3000 2000 1000 0 (a) • • • • • • I Ki-JmSP-i--4^ H— 1 1 1 1 1 1 r— 50 100 150 200 Length cubed (cm3) (Thousands) -t- h 250 300 5000 4000 4-S 3 0 0 0 4 to to CO 2000 1000 20 40 60 Length squared x height (cm3) (Thousands) Figure 15 Scatter plot of (a) length cubed and (b) length squared x height against mass for chinook salmon. 62 TABLE 10: Coefficients of Determination for Length-Weight Models and Shape Characteristics of Chinook Salmon Coefficients of Determination (r2) Sample # M=BL2H M=KL3 M=aLb L=mH+b 1 0.96 0.90 0.91 0.79 2 0.93 0.88 0.88 0.82 3 0.96 0.93 0.93 0.81 4 0.90 0.70 0.71 0.51 5 0.93 0.83 0.84 0.70 6 0.96 0.87 0.87 0.68 7 0.96 0.89 0.92 0.77 8 0.94 0.84 0.85 0.70 9 0.90 0.83 0.86 0.62 10 0.92 0.85 0.86 0.65 11 0.92 0.80 0.88 0.65 12 0.91 0.78 0.80 0.63 13 0.96 0.91 0.94 0.77 14 0.94 0.78 0.77 0.52 15 0.99 0.96 0.96 0.86 16 0.96 0.94 0.94 0.70 17 0.99 0.95 0.95 0.82 63 TABLE 11; Coefficients of Determination for Length-Weight Models and Shape Characteristics of Atlantic Salmon Coefficients of Determination (r2) Sample* M=BL2H M=KL3 M=aLb L=mH+b 1 0.91 0.78 0.81 0.65 2 0.93 0.80 0.82 0.68 3 0.88 0.71 0.75 0.54 4 0.89 0.84 0.86 0.61 5 0.94 0.88 0.87 0.66 6 0.94 0.84 0.85 0.66 7 0.92 0.83 0.81 0.57 8 0.95 0.79 0.81 0.57 9 0.94 0.83 0.83 0.63 10 0.94 0.86 0.86 0.68 11 0.94 0.87 0.86 0.72 12 0.85 0.61 0.65 0.54 13 0.91 0.80 0.81 0.71 14 0.91 0.80 0.80 0.65 15 0.96 0.89 0.89 0.75 16 0.95 0.86 0.85 0.63 17 0.90 0.77 0.77 0.55 18 0.91 0.72 0.71 0.59 64 and length squared x height is considered, however, it appears to be linear (Figure 15b). Although it might be argued that B and K are not truly independent of each other, as they both describe to some degree the relationship between length and mass, it can be seen that they are independent measurements by inspection of Figure 16. In both species, it is obvious that the relationship between length and height is not particularly tight, nor does it appear to be linear, especially in the case of Atlantic salmon. This could be another reason that the M = KI} model works well for fish of smaller sizes, but becomes less accurate at larger sizes. A second proof of this can be seen in Tables 10 and 11 , by the low r2 values for the regression of length/height on mass. The coefficients of variation for the B and K values of Atlantic salmon were 8.08 and 14.35 respectively, and were found to be significantly different from each other. For chinook salmon, the coefficients of variation for the B and K values were 7.65 and 12.34 respectively, and were also found to be significantly different from each other. The percentage decrease in variability by using B instead of K was 43.7% for Atlantic salmon and 38.0% for chinook salmon. Body Dimensions and Shape To look at the morphological differences between the two species, 592 fish (300 Atlantic salmon and 292 chinook salmon) from the same approximate size range (0.1-6 kg) were chosen from the overall data set. The relationship between mass and the three variables length, height and girth was examined by performing a least squares regression on the natural logs of the variables in question, generating equations 65 10 20 30 40 50 60 70 80 90 Length (cm) Figure 16 The variability of height with length for (a) chinook and (b) Atlantic salmon. 66 describing these relationships (Figures 17, 18 & 19). For chinook salmon, the natural log of length explained 97.6% of the variation in the natural log of mass, In height explained 97.3% of the variation in In mass, and In girth explained 94.4% of the variation in In mass. In the case of the Atlantic salmon data, In length explained 93.8% of the variation in In mass, In height explained 96.6% of the variation in In mass, and In girth explained 97.5% of the variation in In mass. When a t-test was used to test the hypothesis that the slopes of the regression equations were the same for the two species, it was found that the first two variables had significantly different slopes, but the slopes of the regression lines of In girth on In mass were not significantly different. A second t-test was then used to compare the elevations of the two regressions, to see whether the two lines coincided. In this case, the null hypothesis that the two elevations were not significantly different was rejected, indicating that the two regression lines were in fact different. However, a common population regression coefficient (slope) for the two lines was computed, and found to be 2.72. When both species were examined together, some interesting trends became evident. For fish of the same mass, Atlantic salmon were seen to be longer than chinook salmon, but chinook salmon were both taller and had a larger girth than Atlantic salmon of the same mass (Figures 17, 18 & 19). The ratio of fork length to height was 4.10 for chinook salmon and 4.86 for Atlantic salmon, while the ratio of fork length to girth was 1.56 for chinook salmon and 1.87 for Atlantic salmon. In both cases, the null hypothesis that the two values were statistically equal was rejected by the z-test. One other parameter was looked at, that of roundness. Roundness is defined as "a measure of the sharpness of the corners of a solid" (Mohsenin, 1980) and is a typical criterion used to describe shape. For the purposes of this study, roundness was 67 80 20 -I 1 1 1 1 1—• 1 1 1 1 1 1 1 0 1000 2000 3000 4000 5000 6000 Mass (g) C: M = 0.0053L 3 2 7,r 2 =0.98 A : M = 0.0154L 2 9 4,r 2 =0.94 Figure 17 The relationship between fork length (L) and mass (M) for both species, where C denotes chinook and A denotes Atlantic salmon. 6 8 A A A CC CA AC A A g, /C A /C CC AAA A A 1 3000 Mass (g) 0 1000 2000 4000 C : M = 1.0681H 2 9 9 ,r 2 =0.97 A : M = 4 .0433H 2 5 5 , r 2 =0.97 5000 6000 Figure 18 The relationship between height (H) and mass (M) for both species where C denotes chinook and A denotes Atlantic salmon. 69 50 45 4-40 E" 35 o § 3 0 25 20 15 c c A c ccc C r<S£ A * ^ A A OCT A A A AA A AA A ^ A A ^ A . A"A M A A A A A «.» fl 1000 2000 3000 Mass (g) 4000 5000 C : M = 0 .1310G Z / B , r ' =0.94 A : M = 0 .2012G 2 7 2 , r 2 =0.98 6000 Figure 19 The relationship between girth (G) and mass (M) for both species, where C denotes chinook and A denotes Atlantic salmon. 70 calculated as girth (or circumference) / n / height (or diameter). A roundness of one would indicate that the fish was perfectly round. The average roundness of chinook salmon was 0.837, while that of Atlantic salmon was 0.836 (Figure 20). These two values were not found to be significantly different from each other. Drag One component of the drag force (that of pressure drag) acting on coasting chinook and Atlantic salmon was calculated. As the two species were not significantly different with respect to shape (i.e. roundness), it was assumed that they would have the same drag coefficients, because "bodies with the 'same characteristics and alignment in flow have identical drag coefficients" at similar swimming velocities, water density and viscosity (Petrell & Bagnall, 1991). Drag was calculated for fish between 1.0 - 3.5 kg in size, as this was the size range over which the average swimming speed (in bl s 1 ) was known for both species. The actual swimming speed (in m s 1 ) could then be calculated for each individual fish by multiplying the fork length of the fish by the average swimming speed (in bl s"1). The range of fork lengths for fish between 1.0 - 3.5 kg is approximately 40 - 60 cm (Figure 17). Reynolds numbers for 50 cm long fish in 10°C water with velocities ranging from 0.1 -1 m s"1 were found to be between 10 4 - 105. Based on this, the drag coefficient was estimated to be 0.09 (Blake, 1983). The cross-sectional (or frontal) area was calculated using formulae for the circumference and area of an ellipse, as this was considered to be the closest approximation to the shape of the frontal area. When the drag force was calculated for both species, it was found to be 41 .44% greater for chinook salmon than for Atlantic salmon of comparable mass. The 71 1 A A A A A A A „ A A m A A .~Aa»rA/r A A yfcCCC^C c * > c < C A ^ C A 0 0 CP R ^ -AAAA C CA A A c " A < A C A A A A A C A . ^ A AA A A < A ^ A A A A A A 1000 2000 3000 Mass (g) 4000 5000 6000 Figure 20 The relationship between roundness and mass for both species, where C denotes chinook and A denotes Atlantic salmon. 72 magnitude of drag ranged between 0.05 - 0.35 N (Figure 21), which is within the range of experimental values found by other researchers (Blake, 1983). It was determined that approximately half of the difference in drag forces was due to the increased frontal area of chinook salmon as compared to Atlantic salmon, with the balance attributable to the difference in swimming speed between the two species. The other component of drag on a coasting fish, skin friction drag, was not calculated as the wetted surface area of the fish was required and had not been measured. Measurement of the drag forces acting on an actively swimming fish was beyond the scope of this thesis. The power needed to overcome the drag force ranged between 0.03 - 0.19 W for Atlantic salmon and 0.06 - 0.27 W for chinook salmon, which are similar to values found for Atlantic salmon (Tang & Wardle, 1992). The power requirements of chinook salmon were found to be 53% higher than those for Atlantic salmon of the same mass. 73 c?0.15 4-0.1 0.05 4-1000 cc ; cc cc C C C A /C r t e c CAA £ M A A % * A A A A ^ C A A A ^ A A A C c*c C C c c cc A A A * « A A CC c c c c C A A i A 1500 2000 2500 Mass (g) 3000 3500 Figure 21 The relationship between drag and mass for both species, where C denotes chinook and A denotes Atlantic salmon. 74 DISCUSSION Size Distribution Size distribution of chinook salmon within sea cages estimated from video sampling seemed to be controlled by two factors, the first being time of video sampling in relation to time of feeding, and the second factor being the degree of net deflection present at the time of filming. Site A As the fish on this site were on a twice daily feeding schedule, video sampling was classified as having occurred after feeding if it took place in the morning and before feeding if it took place in the afternoon. In two of the six cages (Cages 2 and 3), the biggest fish were found in P1 and P2, or in other words the top half of the swimming aggregation (Tables 2 & 5). Both of these cages were filmed in the afternoon, before feeding, which suggests that the larger, more dominant and aggressive fish were higher up in the school in anticipation of a feeding event. Cage 4 was filmed after feeding and here the largest fish were in P3, at the bottom of the sea cage. This trend implies that there may be some degree of territorial and dominant/submissive behaviour occurring among chinook salmon in sea cages, as in these cases the largest fish were found in the best locations in the cage. Another interesting point about these three cages was that they were the ones with the least net deflection at the bottom of the cage (in each case, filming of P3 took place at at least 14 m depth). This lack of deflection increased the usable net volume available 75 to the fish and may also have encouraged the separation of the fish into sub-populations of significantly different size classes offish. Both Cages 1 and 5 showed no significant difference between positions 1, 2, and 3. Cage 1 was sampled after feeding, while Cage 5 was filmed before feeding. In both cases, the cameras were lowered so that they were directly above the net bottom, which should have been at 16 m. Due to net deflection, however, the cage depth was only 12.5 m, which caused a substantial decrease in net volume. This finding suggests that in situations where space limitations are in effect, the dominant hierarchy breaks down and the separate sub-populations offish become mixed. Site B The fish at Site B were not fed during the video sampling, as they were undergoing a physical sampling at the same time. In all three cages that had more than one position filmed, the largest fish were found at the top of the swimming aggregation (Tables 3 & 5). This is consistent with the findings for Site A, in that the largest and most dominant fish were again in the premium location in the sea cage, at the top of the school and first in line for feeding. Although there was a fair degree of net deflection observed during filming, mixing of sub-populations did not occur. This might be due to the fact that the fish at Site B were already distributed higher up the water column than the fish at Site A, and therefore had more space to utilize. The shallowest depth that Site A fish were found at was 6.5 m, while Site B fish were filmed at 2-3 m on three separate occasions. Stocking densities at Site B were, in general, lower than those at 76 Site A, which could also explain the ability of the Site B fish to stay within their sub-populations (Table 1). Site C At this site, two of the cages (Cage 3 and 4) were on starve, and in both the largest fish were found at the top of the aggregation (Tables 4 & 5). Cage 1 had an extremely billowed net (one of the corner anchors had come loose) and the fish were compacted and well mixed, resulting in no significant differences between positions. Cage 2 did not follow the general trend. It was filmed in the late afternoon (which was still after feeding, as the Site C fish were only being fed once a day) when the largest fish were expected to be at the bottom of the aggregation. The large fish were found in the middle of the school, however. Perhaps there was a different advantage to being in the center of the school in effect, such as predator avoidance (there were several seal hits on the sea cages during the sampling visit). These findings contradict those of Kadri et al. (1996) who found that one-sea-winter Atlantic salmon exhibited marked differences in food acquisition among individual fish that were unrelated to size or gender. Size was also found to be a poor predictor of status (assessed by monopolisation of food and intimidation of siblings) for Atlantic salmon parr (Huntingford et al., 1990). However, Shieh (1996) suggests the possibility that size-based dominance hierarchies exist in Atlantic salmon sea cages, based on the occurrence of larger fish at the surface prior to feeding. Studies done on other salmonids, namely rainbow trout, steelhead (Oncorhynchus mykiss) and brown trout also found status to be largely affected by size (Jenkins, 1969; Abbott & Dill, 1989). 77 The presence of size-based dominance hierarchies within chinook salmon sea cages can have several different results (Magurran, 1986; Ruzzante, 1994). The first is that sub-dominant individuals may grow at less that maximal rates, due to limited access to food. This, of course, results in dominant fish receiving a disproportionate amount of food, so that they become larger even in the presence of excess food. Secondly, the presence of dominant fish may lead to physiological stress in the subordinate fish, resulting in loss of appetite and low efficiency of food conversion. Thirdly, subdominants may exhibit higher activity levels and therefore higher metabolic expense than dominants, again resulting in slower growth. All of these possible mechanisms for behavioural adaptation in the face of a dominance hierarchy result in a single outcome: the growth of a significant proportion of the cage population will be retarded, resulting in poorer overall growth and higher F C R s . It is interesting to note that out of the five cages (based on all three sites) where there were statistical differences between the average size of the fish in the dip net sample and at least one of the positional samples in a sea cage, in three cases the dip net was not found to be different from the largest subpopulation in the cage. This was probably due to the fact that the fish were always placed on starve before a dip net sample was taken, to reduce the stress on the fish. The large dominant fish would therefore be schooling at the top of the swimming aggregation in the optimum feeding location, and consequently would be most likely to be caught by the seine net, which often does not reach all the way to the cage bottom. This could obviously lead to large biases in the dip net sample, and in fact there are reports of over estimation of the true mean fish size by as much as 15-25% (Klontz, 1993). 78 Swimming Speed The range of average swimming speeds measured for chinook and Atlantic salmon in sea cages was very similar. The speeds measured using the V ICASS system were in line with values quoted for research done on salmonid species both in the wild and under farming conditions (e.g. Malinin, 1975; Davison & Goldspink, 1977; Greer Walker & Emerson, 1978; and Quinn, 1988). Sub-populations of fish in a sea cage did not swim at the same speed. In eight out of nine cages of chinook salmon sampled, there were significant differences in swimming speed at different depths. Although this was not statistically tested for the cages of Atlantic salmon, it was evident that there were also large differences between speeds at different positions. Several factors were examined in an attempt to explain these differences in swimming speed, such as water temperature, depth, position within the school, and shape of swimming aggregation. Both species exhibited a tendency to swim at higher speeds as the water temperature increased, although with a high degree of variability with respect to temperature. The other clear trend was that of type of swimming aggregation with swimming speed for chinook salmon. In most cases, the location of highest fish density within the cage coincided with the fastest average swimming speed. This agrees with the findings of Pitcher and Partridge (1979), who conducted research on the density and volume of schools of saithe. They discovered that for swimming speeds of 1.6 bl s 1 and up, there was a strongly significant curvilinear trend towards smaller volumes per fish. As none of the above parameters (with the exception of temperature) were clearly associated with swimming speed differences in Atlantic salmon, it is possible that another factor was of more importance than any of the parameters that were investigated in this research. 79 These differences in swimming speed within a cage are important with regard to the growth and F C R of the population of fish within that cage. The optimum speed (defined in Videler, 1993, as the swimming velocity "where the amount of work per metre reaches a minimum") has been mathematically derived to be approximately 1.0 bl s"1, so presumably fish that are swimming above the optimum are using a greater proportion of their energy budget for swimming than is absolutely necessary. The best growth rates in fish have been found to occur at swimming speeds approximating 1.0 bl s 1 (Davison & Goldspink, 1977; Greer Walker & Emerson, 1978; and East & Magnan, 1987). A 19% difference was found between the average swimming speeds of the two species. On average, chinook salmon were found to be swimming at 1.38 bl s 1 , while Atlantic salmon were only swimming at speeds of 1.16 bl s 1 . Obviously, this will place a greater demand on an individual chinook salmon's resources, leading to higher FCR values. There was no evidence that time of day had any effect on swimming speed. Previous research done on Atlantic salmon indicated that the fish swam fastest in the early morning (0400-0600 hrs) and evening (2000-2200 hrs), and slowest in the middle of the afternoon (Kadri et al., 1991). However, as filming was restricted to daylight hours (0900-1900 hrs), it was not possible for us to measure swimming speed over a complete 24 hour period. A strong relationship was found to exist between average fish size and swimming speed (measured in bl s 1 ) for both species. The salmonids swam slower as they increased in size. There were two major drawbacks to measuring the swimming speed offish using V ICASS. The first was that by attempting to gauge the average speed of a sub-80 population of fish one did not take into account fish that were swimming either much faster or much slower than their neighbors. The second problem was that there was no way to quantify turning or acceleration, which are known to have substantial metabolic costs (Krohn & Boisclair, 1994). It was observed from the footage that chinook salmon exhibited far more darting, acceleration, and directional changes than Atlantic salmon; in other words, they seemed 'wilder' and responded much more readily to outside stimuli, while the Atlantic salmon behaved in a more domesticated fashion. By observation, it could be said that chinook salmon exhibited a swimming behaviour that could be characterized as routine swimming, while Atlantic salmon tended to demonstrate a directed swimming pattern (as defined by Boisclair & Tang, 1993). The aforementioned researchers found that routine swimming was 4.9 to 7.5 times more expensive than directed swimming for fish of similar weight. Body Dimensions When two length-weight relationships commonly used in fisheries research ( M = A2 3 and M = aE) were compared with a new model (M = BL2H) developed during the course of this study, it was found that the new model consistently gave more accurate results over a much wider size range than either of the other models. This is of course largely due to the fact that instead of attempting to describe the growth and condition offish using only one body parameter, that of length, the model incorporates a second body parameter, fish height. There are two different ways in which the higher degree of accuracy of this model can be illustrated. For example, Atlantic salmon have been observed to continue 81 to grow in length (but not in height or girth) when fed a maintenance diet, leading to a population of 'eely' fish, i.e. fish that are very long and thin (Ang & Petrell, 1996). In this situation, using a model based only on length would lead to an overestimation of the mass of the population. The model M = BL2H is also more sensitive to changes in the condition offish because it is invariant with size (Figures 13, 14 & 15), and is therefore a better indicator of nutritional status. This model accounts for two parameters of the fish body, leaving only the thickness to be captured by B (the condition factor). Therefore, even slight changes in thickness will be reflected in the B factor. The one possible drawback to the model is that changes in condition must be accompanied by comparative changes in thickness. Chinook salmon have been shown to be taller, thicker and shorter than Atlantic salmon of the same weight, and a deteriorating chinook salmon must therefore exhibit a much greater proportional change in thickness than an Atlantic salmon before a change in the condition factor will become evident. If this fact presented a problem in the model's estimation of the condition of chinook salmon, we would expect to see little change in the B factor measured at the different sites, even though the fish were visually observed to be in varying states of condition. However, the wide range of average B values (56.55 - 69.09) for the 17 sea cages studied demonstrates that, in fact, chinook salmon do express a significant proportion of their nutritional status in terms of thickness. Atlantic salmon were found to be significantly longer than chinook salmon of the same mass, while chinook salmon were found to be significantly taller and of larger girth than Atlantic salmon of the same mass. The ratios of fork length to girth and fork length to height were both found to be significantly different for the two species. However, the shape of both species, as described by the criterion of roundness, was 82 not found to be significantly different. These facts have important implications on the drag forces that act on the fish as they move through the water. It has already been shown that chinook salmon must overcome significantly larger drag forces (approximately 40% higher) than Atlantic salmon of the same mass, due to their larger frontal area and higher swimming speeds (over the size range considered, 1.0 - 3.5 kg). The drag of a steadily swimming fish is approximately proportional to the square of the speed (Videler, 1993). When we use this relationship to calculate the drag on chinook and Atlantic salmon using 1.38 bl s"1 and 1.16 bl s~1 respectively, the drag on chinook salmon is 41.5% higher than that on Atlantic salmon, which is almost identical to the difference found using the data. Videler (1993) also states that the energy needed to overcome the drag will be nearly proportional to the cube of the speed. This translates into a 68.4% higher energy requirement for chinook salmon. When the power requirements were calculated the actual difference was found to be slightly less than this theoretical value, at 53%. However, there is a second element to this equation. For a given length, chinook salmon are heavier than Atlantic salmon (Figure 22). Assuming that gill area and the efficiency of the pumping mechanism and gaseous exchange are similar for salmonids of a given length, the metabolic scope for activity will decrease with an increase in weight, placing another demand on the chinook salmon's energy budget (Beamish, 1978). Although in absolute terms, chinook and Atlantic salmon may swim at comparable speeds, chinook salmon must expend a great deal more energy than Atlantic salmon to maintain these speeds, leaving a smaller proportion of the energy budget for growth. Commercial salmon feeds are high energy diets comprised of 83 6000 5000 4000 3 ^3000 CD 2000 1000 l 20 cc C cc A £ A C ^ A ^ A A A I * 2 ? 8 t A A A A A A A j A A A A A 50 Fork length (cm) 70 80 Figure 22 The relationship between mass and fork length for both species, where C denotes chinook and A denotes Atlantic salmon. 84 approximately 44% protein and 23% oil (Jackson, 1988). The theory behind this is that by providing fat in abundance and protein at a level just sufficient to meet growth requirements, the fish will be forced to use fat for energy, thereby sparing the protein for growth. This is desirable as protein is the most expensive source of energy, and also the first to be utilized in fish, followed by fat and then carbohydrate. The average caloric value of these two major energy sources, corrected for digestibility and the portion present as nitrogen (and therefore unavailable as energy) is 3.9 kcal g"1 of protein and 8.0 kcal g"1 of fat (Phillips, 1972). Based on these values and the percent breakdown of the diet, one kg of food has a total of 3556 kcal of energy from protein and oil. This translates into a total of 16.0 kcal kg fish"1 day"1 (calculated with a ration level of 4.5 x 10"3 kg of feed kg fish"1 day"1, which is the recommended level for salmonids weighing over two kg in 8°C water (Moore Clark feed tables)). I conservatively assumed that caged salmonids expend somewhere between 50 - 1 0 0 % of their budget for activity on swimming, with the remaining percentage spent on feeding behaviours. However, the percentage of energy spent feeding under culture conditions is probably much less than 50%, as "space limitations and the usual feeding protocols clearly obviate any great expenditure of energy upon search and capture of rations..." (Kerr, 1971). The active metabolism is going to be proportionately larger than that explained by the drag force calculated here, as the pressure drag makes up only a portion of that acting on the fish. Therefore, the energy needed to overcome this drag is also only a portion of the total energy required for swimming. When the energy budgets of both species were modelled for three different sizes offish (1, 2, and 3 kg), it became clear that regardless of these two variable factors (i.e. 85 the proportion of energy spent on swimming (as opposed to foraging); and the proportion of the total energy required for swimming that was explained by the power needed to overcome the pressure drag on a coasting fish), the overall difference in scope for growth remained constant at any one fish size. At one kg, chinook salmon had 5.2% day"1 less energy for growth (kcal kg fish"1 day"1) than Atlantic salmon, whereas at three kg, this difference had increased to 10.3% (Figure 23). Therefore, over the production cycle, chinook salmon have on average 7.8% less energy for growth on a daily basis than Atlantic salmon do. Ultimately, we are interested in what this daily difference in growth between the two species adds up to over the entire production period. The average growth rate of Atlantic salmon at 8°C ranging in size from 150 g to 3000 g is 0.67% day 1 (Wootton, 1990). The average growth rate of chinook salmon was then calculated as 0.62% day 1 (i.e. 7.8% less than that of Atlantic salmon). The average growth rates were multiplied by the average time the fish spent at sea (24 months for Atlantic salmon; 22 months for chinook salmon (Cooperative Assessment of Salmonid Health data)) and an overall difference of 18% was found. This difference in growth explains over 50% of the 33% difference between typical F C R values of 1.5 for Atlantic salmon and 2.0 for chinook salmon. Obviously, this is one of the main causes of the consistently poorer performance of chinook salmon as compared to Atlantic salmon. The fact that the difference in scope for growth increases as fish size increases may also offer some explanation as to why chinook salmon, which are the largest of the Pacific salmon and are known to exhibit high growth rates in the wild, cannot attain the same harvest weight as Atlantic salmon at approximately the same age. The swimming behaviour of wild chinook salmon in the ocean has not been well documented. Perhaps they tend to 86 1 CM CO CO CD O CD CM co" CM LO^  enl oo E CD co Incr Los , - t • co co CD CD I X in CD CO - >>•-O P CD ^ 3 O 00 CD .9? 2 0> C O T J - — C fl) CO CO c o ^ g CD ° E X _co o CD c _ CO T j OT O .9 S-c CD J ? i § < ° c5 CD .E co co CD 3 0) CD - 5 C co > CD T J SZ CD I- o 1 C O o E CO CO CD O c TO 2 I T J CO CD _ i _ CZ CO — Q . CD c co £ ro 8 2 i ? £ T J CO C cn o Is 1 8 2 co >r co CO > CO CD o o c J C o T J c CO CO CD CD CO CO CD ro g £ .S o c S o CD ^ X I >Ss CD b o <u c E co c CD CD C L CD CO -= CO 2 £ CD .!= £ CD — CO CO CD c ra w < i T l <= CD ro N CD co x : j? o) T J CO „ CD O x , CD CO CD 9> T J 21 U CO x» > cov L— >» CD CO C T J CD CD x : sz w h- «-CO CN 81 3 cn LE 2 o z: c o o -4—' CD CD is . ro x " CD O x : £ ** x : £ ° T J c5 CD CO cn CO — " CD .?2 CO T J CD *- x : <D • £ 3 CO Xi CL o c c ^ ro ai c • - CJ) CO r _ 3 co" CO CD CD > o CD o3 o "S CO CD p E c o O Jz fc- T J «= £ C T J J= O CO ^ CO S> 87 swim with the current rather than against it, or shelter behind objects such as rock outcroppings to escape the strongest currents. When forced to swim continuously in a cage, it is possible that the costs of swimming become so great for chinook salmon once they reach a certain size that they simply have no energy left for growth. 88 CONCLUSIONS In six out of fourteen cages of chinook salmon, fish size varied significantly with position. This result suggested a size-based dominance hierarchy. In general, larger fish were found at the top of the swimming aggregation prior to commencement of feeding, and were otherwise located in the bottom portions of the cage at the bottom of the aggregation. Under conditions of severe net deflection a well-mixed aggregation devoid of discrete subpopulations was observed. This was probably due to increased fish densities, and actually offers the best situation under which to obtain an accurate size sample of the population. The presence of a dominance hierarchy within sea cages will affect the feeding opportunity of fish within the cage, thereby affecting the growth of the fish. It will also adversely affect the accuracy of a physical sample of the cage. The range of relative swimming speeds of chinook and Atlantic salmon was not found to differ under varying environmental conditions and husbandry techniques. Neither species exhibited any relationship between swimming speeds and several environmental factors (time of day, water temperature, and depth) although both showed a strong negative correlation between swimming speed measured in bl s"1 and fish size. When average swimming speeds of both species (bl s 1 ) were compared, chinook salmon were seen to swim proportionately faster than Atlantic salmon. This will place larger energetic demands on chinook salmon than Atlantic salmon, decreasing the amount of energy available for growth. There were large differences between swimming speeds at different places in the swimming aggregations of chinook salmon 89 in sea cages, but in general the fish in the densest part of the aggregation swam the fastest. A model for assessing the condition and growth of salmonids was tested (M = BL2H) and found to be more accurate than two commonly used fisheries models (M = KL3 and M=aLh). One feature of this model that makes it useful for aquaculturists is its increased sensitivity to changes in the nutritional status of a fish, reflected in changes in the condition factor, B. Another advantage is its accuracy over a wide range of sizes and ages, from smolts up to harvest-sized fish. Several shape parameters of Atlantic and chinook salmon were compared, and the two species were found to be morphologically distinct. Chinook salmon were shown to be taller, shorter, and thicker than Atlantic salmon of the same mass. Finally, the drag forces acting on the two species were calculated, and chinook salmon were found to have a substantially higher drag than Atlantic salmon due to their larger cross-sectional area. This translates into greater energy requirements for propulsion, and consequently less energy available for growth. A simple energy budget was used to model the partitioning of energy for these species, and the decreased growth of chinook salmon as compared to Atlantic salmon was found to explain over half of the observed differences in F C R s for the two species. There are several possible solutions to this problem. The first two involve selective breeding of chinook salmon: a) that swim slower and with less acceleration and directional changes, and b) that have a higher fork length to height ratio, i.e. with a body conformation that is closer to that of Atlantic salmon. Another viable solution would be to formulate a higher energy feed for chinook salmon. 9 0 AREAS FOR FURTHER STUDY Results from this study suggested the possibility of size-based dominance hierarchies in chinook salmon sea cages. This hypothesis should be tested to determine whether, in fact, such hierarchies do exist and under what conditions they might break down. Quantifying the degree of acceleration, directional changes and darting was beyond the scope of this thesis. However, as was stated in the Discussion, through observation of many hours of video footage chinook salmon were seen to exhibit these behaviours a significantly higher proportion of the time than Atlantic salmon did. Acceleration and deceleration could be physically measured for individual fish using an accelerometer, or calculated from the existing video footage by double differentiation of displacements as a function of time. Directional changes would be harder to quantify, but acoustic tags placed on an individual fish would allow for continuous tracking of the fish over time, which would create a clearer picture of how a fish moves within the swimming aggregation. There are various interpretations of the meaning of condition in fish. Most researchers agree that it indicates to some degree the health of an individual or population, but do not define it further. Is condition a measure of the amount of flesh, flesh quality, proportion of flesh mass to skeletal mass, or some other parameter? The determination of exactly what attribute is being quantified when we assess condition would be valuable to aquaculturists. 91 Finally, there is enormous scope for investigation into the nutrition of chinook salmon, as feeds formulated for Atlantic salmon do not seem to provide the energy requirements for optimal growth in chinook salmon. Hopefully research into one or more of the above areas will resolve some of the problems facing the production of chinook salmon and allow for continued profitable farming of a native species on B.C.'s west coast. 92 REFERENCES Abbott, J .C . & Dill, M. (1989). The relative growth of dominant and subordinate juvenile steelhead trout (Salmo gairdneri) fed equal rations. Behaviour, 108, 104-113. Anderson, R.O. & Gutreuter, S .J . (1983). Length, weight, and associated structural indices. In Fisheries Techniques (eds L.A. Nielsen & D.L. Johnson), American Fisheries Society, Bethesda, Maryland, pp. 283-300. Ang, K.P. & Petrell, R.J. (1996). Control of feed dispensation in seacages using underwater video monitoring: effects on growth and food conversion. (Aquacultural Engineering, in press, 28 p.). B.C. Ministry of Agriculture, Fisheries and Food. (1993). Aquaculture: Its role in British Columbia's Seafood Industry. B.C.M.A.F.F., Victoria, B.C., 7 p. B.C. Salmon Farmers Association et al. (1992). Aquaculture: British Columbia's Future. B.C.S.F.A. , Vancouver, B.C., pp. 29-53. B.C. Salmon Farmers Association. (1994). Weekly Volume Report of B.C. Farmed Salmon Processing. B.C.S.F.A. , Vancouver, B.C. Beamish, F.W.H. (1978). Swimming capacity. In Fish Physiology, vol. VII, Locomotion (eds W.S. Hoar, D.J. Randall & J.R. Brett), Academic Press, London, pp. 101-187. Bjordal, A., Juell, J .E . , Lindem, T. & Ferno, A. (1993). Hydroacoustic monitoring and feeding control in cage rearing of Atlantic salmon (Salmo salar L.). In Fish Farming Technology (eds H. Reinertsen, A.D. Dahle, L. Jorgensen, K. Tvinnereim), Balkema, Rotterdam, pp. 203-208. Bjorndal, T. (1990). The Economics of Salmon Aquaculture. (1990). Blackwell Scientific Publications, Oxford, England, pp. 9-20. Blake, R.W. (1983). Fish Locomotion. Cambridge University Press, New York, 208 p. Bluman, A . G . (1995). Elementary Statistics, 2nd ed. Wm. C. Brown Communications, Inc., Dubuque, Iowa, pp 311-319. Boisclair, D. & Tang, M. (1993). Empirical analysis of the influence of swimming pattern on the net energetic cost of swimming in fishes. J. Fish Biol., 42(2), 169-183. Bolger, T. & Connolly, P.L. (1989). The selection of suitable indices for the measurement and analysis offish condition. J. Fish Biol., 34, 171-182. 93 Brett, J.R. & Groves, T.D.D. (1979). Physiological energetics. In Fish Physiology, vol. VIII, Bioenergetics and Growth (eds W.S. Hoar, D.J. Randall & J.R. Brett), Academic Press, London, pp. 279-352. Caine, G., Truscott, J . , Reid, S. & Ricker, K. (1987). Biophysical critieria for siting salmon farms in British Columbia. B.C. Ministry of Agriculture and Fisheries, Aquaculture and Commercial Fisheries Branch, Victoria, B.C. and Aquaculture Association of British Columbia, 50 p. Cone, R.S. (1989). The need to reconsider the use of condition indices in fishery science. Trans. Am. Fish. Soc, 118(5), 510-514. Davison, W. & Goldspink, G. (1977). The effect of prolonged exercise on the lateral musculature of the brown trout (Salmo trutta). J. exp. Biol., 70, 1-12. Dunn, M. & Dalland, K. (1993). Observing behaviour and growth using the Simrad F C M 160 fish cage monitoring system. In Fish Farming Technology (eds H. Reinertsen, A.D. Dahle, L. Jorgensen, K. Tvinnereim), Balkema, Rotterdam, pp. 269-274. East, P. & Magnan, P. (1987). The effect of locomotor activity on the growth of brook charr, Salvelinus fontinalis Mitchill. Can. J. Zool., 65, 843-846. Ejike, C. & Schreck, C.B. (1980). Stress and social hierarchy rank in coho salmon. Trans. Am. Fish. Soc, 109, 423-426. Fenderson, O.C., Everhart, W.H. & Muth, K.M. (1968). Comparative agonistic and feeding behaviour of hatchery-reared and wild salmon in aquaria. J. Fish. Res. Board Can., 25, 1-14. Ferno, A. & Holm, M. (1986). Aggression and growth of Atlantic salmon parr. I. Different stocking densities and size groups. FiskDir. Ser. Havilnders., 18, 113-122. Ferno, A., Furevik, D., Huse, L. & Bjordal, A. (1988). A multiple approach to behavior studies of salmon reared in marine net pens. International Council for the Exploration of the Sea, C M . 1988/F15, 15 p. Ferno, A. (1989). Behavioural studies of salmon reared in marine net pens. Ethology in Aquaculture: Application of Behavioral Studies in Aquaculture. Proceedings from minisymposium, Trondheim, Norway, October 1989, Norwegian Society for Aquaculture Research, 6 p. Fries, J .N. (1994). S A M C A L C : A computer program for fish culturists. Prog. Fish-Cult, 56, 62-64. Fulton, T.W. (1911). The Sovereignty of the Sea. London: Edinburgh. 94 Greer Walker, M. & Emerson, L. (1978). Sustained swimming speeds and myotomal muscle function in the trout, Salmo gairdneri. J. Fish Biol., 13, 475-481. Hayes, D.B., Brodziak, J.K.T. & O'Gorman, J.B. (1995). Efficiency and bias of estimators and sampling designs for determining length-weight relationships of fish. Can. J. Fish. Aquat. Sci., 52, 84-92. Huntingford, F.A., Metcalfe, N.B., Thorpe, J .E. , Graham, W.E. & Adams, C .E . (1990). Social dominance and body size in Atlantic salmon parr, Salmo salar L. J. Fish Biol., 36, 877-881. Jackson, A. (1988). Growth, nutrition and feeding. In Salmon and Trout Farming (eds L.M. Laird &T . Needham), Ellis Horwood Ltd., Chichester, England, pp. 202-216. Jenkins, T.M. (1969). Social structure, position choice and microdistribution of two trout species (Salmo trutta and Salmo gairdneri) resident in mountain streams. Anim. Behav. Monogr., 2, 57-123. Jobling, M. (1994). Biotic factors and growth performance. In Fish Bioenergetics, Fish and Fisheries Series 13, Chapman and Hall, London, pp. 169-201. Juell, J .E . (1989). Swimming patterns of salmon in sea cages. Ethology in Aquaculture: Application of Behavioral Studies in Aquaculture. Proceedings from minisymposium, Trondheim, Norway, October 1989, Norwegian Society for Aquaculture Research, 9 p. Kadri, S., Metcalfe, N.B., Huntingford, F.A. & Thorpe, J .E . (1991). Daily feeding rhythms in Atlantic salmon in sea cages. Aquaculture 92, 219-224. Kadri, S., Huntingford, F.A., Metcalfe, N.B. & Thorpe, J .E . (1996). Social interactions and the distribution of food among one-sea-winter Atlantic salmon (Salmo salar) in a sea-cage. Aquaculture, 139, 1-10. Kerr, S.R. (1971). Analysis of laboratory experiments on growth efficiency of fishes. J. Fish. Res. Bd. Canada, 28, 801-808. Kjartansson, H., Fivelstad, S., Thomassen, J .M. & Smith, M.J. (1988). Effects of different stocking densities on physiological parameters and growth of adult Atlantic salmon (Salmo salar L.). Aquaculture, 73, 261-274. Klontz, G.W. (1993). Producing a Marketable Fish. Part V: Inventory Techniques. Northern Aquaculture 10, 21-25. Koch, F. & Wieser, W. (1983). Partitioning of energy in fish: can reduction of swimming activity compensate for the cost of production? J. Exp. Biol., 107, 141-146. 95 Kristinsson, J.B. , Saunders, R.L. & Wiggs, A . J . (1985). Growth dynamics during the development of bimodal length-frequency distribution in juvenile Atlantic salmon (Salmo salar L ) . Aquaculture, 45(1), 1-20. Krohn, M.M. & Boisclair, D. (1994). Use of a stereo-video system to estimate the energy expenditure of free-swimming fish. Can. J. Fish. Aquat. Sci., 51, 1119-1127. Li, H.W. & Brocksen, R.W. (1977). Approaches to the analysis of energetic costs of intraspecific competition for space by rainbow trout (Salmo gairdneri). J. Fish Biol., 11, 329-341. Magurran, A . E . (1986). Individual differences in fish behaviour. In Behaviour of Teleost Fishes (ed T.J. Pitcher), John Hopkins Univ. Press, Baltimore, pp. 338-365. Malinin, L.K. (1975). Rates of movement of salmon (Salmo salar) during anadromous migration in fresh water. Zoologicheskii Shurnal, LIV, 1729-1731. Mohsenin, N.N. (1980). Physical Properties of Plant and Animal Materials. Gordon and Breech, New York, p 56. Partridge, B.L. & Pitcher, T.J . (1979). Evidence against a hydrodynamic function of fish schools. Nature, 279, 418-419. Partridge, B.L., Johansson, J . & Kalish, J . (1983). The structure of schools of giant bluefin tuna in Cape Cod Bay. Env. Biol. Fishes, 9, 253-262. Pennell, W. (1992). Site Selection Handbook. Ministry of Agriculture, Fisheries and Food, Victoria, B.C. Petrell, R.J & Bagnall, L.O. (1991). Hydromechanical properties of water hyacinth mats. Aquacultural Engineering, 10, 133-147. Petrell, R.J. , Shi, X., Ward, R.K., Naiburg, A. & Savage, C R . (1997). Measuring fish size in cages and tanks using simple video techniques and fish movement. (Aquacultural Engineering, in press, 34 p.). Pfaffenberger, R.C. & Patterson, J . H. (1977). Statistical methods for business and economics. Richard D. Irwin, Inc., Homewood, Illinois, 750 p. Phillips, A.M.Jr. (1972). Calorie and energy requirement. In Fish Nutrition (ed J .E. Halver), Academic Press, New York, pp. 2-27. Pitcher, T.J . , Magurran, A . E . & Edwards, J.I. (1985). Schooling mackerel and herring choose neighbours of similar size. Marine Biology 86, 319-322. 96 Pitcher, T.J . & Parrish, J.K. (1993). Functions of shoaling behaviour in teleosts. In Behaviour of Teleost Fishes, 2nd edn. (ed T.J. Pitcher), Chapman & Hall, New York, pp. 363-439. Pitcher, T.J . & Partridge, B.L. (1979). Fish school density and volume. Marine Biology 54, 383-394. Quinn, T.P. (1988). Estimated swimming speeds of migrating adult sockeye salmon. Can. J. Zool., 66, 2160-2163. Ricker, K.E. & Truscott, S .J . (1989). Biophysical suitability of the Sunshine Coast and Johnstone Strait/Desolation Sound areas for salmonid farming in net cages. B.C. Ministry of Agriculture and Fisheries, Victoria, B.C., 98 p. Ricker, W.E. (1975). Computation and interpretation of the biological statistics off ish populations. Bull. Fish. Res. Board Can., 191, 1-382. Ruzzante, D.E. (1994). Domestication effects on aggressive and schooling behavior in fish. Aquaculture, 120, 1-24. Sachs, L. (1984). Applied Statistics: A Handbook of Techniques. Springer Verlag, New York, 707 p. Seghers, B.H. (1974). Schooling behaviour in the guppy Poecilia reticulata: an evolutionary response to predation. Evolution, 28, 486-489. Shieh, A. (1996). Fish size estimation in sea cages using a fish image capturing and sizing system, unpublished M.Sc. Thesis. University of British Columbia, 120 p. Smith, G.W., Hawkins, A.D., Urquhart, G .G. & Shearer, W.M. (1981). Orientation and energetic efficiency in the offshore movements of returning Atlantic salmon, Salmo salar L. Scott. Fish. Res. Rep. No. 21. Tang, J & Wardle, C S . (1992). Power output of two sizes of Atlantic salmon (Salmo salar) at their maximum sustained swimming speeds. J. exp. Biol., 166, 33-46. Videler, J . J . (1993). Fish Swimming. Chapman and Hall, New York, 260 p. Weihs, D. (1973a). Hydromechanics and fish schooling. A/afare, 241, 290-291. Weihs, D. (1973). Optimal fish cruising speed. Nature, 245, 48-50. Weihs, D. & Webb, P.W. (1983). Optimization of Locomotion. In Fish Biomechanics (eds D. Weihs & P.W. Webb), Praeger, New York, pp.339-371. Wootton, R.J. (1990). Ecology of Teleost Fishes. Chapman and Hall, New York, 404 P-97 Zar, J.H. (1984). Biostatistical analysis. Prentice-Hall, Inc., Englewood Cliffs, New Jersey, 718 p. 98 PERSONAL COMMUNICATIONS Sylvain Alie, British Columbia Packers Ltd., Campbell River, B.C. 99 APPENDICES to O 3 to CD CD I | x 0} o CO CD CD E o £ CD D) CD CD .{= T3 ro -H "5 > o ^ ro m o 3 CO CD S - i " LL ro 'E o c CD •o CD TJj P CD CD f0 <D >J= T3 O O p s S I CD CD [p. E i _ CD CD CD =5 CD J= C T3 XJ CD CD ro - i ° , ro l J -o CD 2i D J ro o o O o O o o o o o o X X X X X X X X X X X ad a: O ad of cd cd ad ad ad cd z z Reje z z z z z z z z d Q Reje a Q ci ci ci ci Q Q CO CO CM CM CO CO CM CO •sT CO O X ad O O O O X X X X ooood CD CD CD aT aT aT CtC CtC Di Q o X •*-* o CD aT o X ad o o o X X X o t t o CD CD '67 • 'aT ££ Q Cd CO CO CM CN CO CO L O L O CM CO -sT CO CN CM CO CO co CO CO •sl- CO CN ed CO CM T— CN CN T - CM CO -sj- LO CO i i i i i i <<<<<< O O X X o ad CD CD • ad cd a o CD LO LO LO LO T LO CO LO 00 00 LO 00 x— x— CD CD x— X— CD X— 00 X— 00 00 x— 00 00 CO CD CD 00 00 OJ CO •sr 00 •sr •sr CO •sT LO LO 1^ LO a> 1^ CD CD ai o CO CO CO O 00 o LO •sr LO r -CN o CD CO CM LO CM CO CD CN CO CM LO T— CO CM C\i •sr -st- co -sr i 0 0 o o o o X X X X ad o cd o z eje z eje ci ad d Cd C M C M C O C O C N C N C O C N r~- C N O J C O C O C O O o co C D O C D C O co co C O C D co C N C N C O C O C M C N C O C N C M C N C N C N C N C N r-- CN co O CO CO CD 00 x— r-- CD CM CO LO •ST LO CO CN oo in O LO •sr Oi LO co CO CM CN CO CO CN CO •sf •sr •st- LO CO •sr •sj- -sf CO -si-•sr 00 O) O •sr CN CD OJ CN CD •sr CO T CN CN Tr o r— CN CO •st- T— CN CO CO CD CO CCI 6 6 6 o CD '57 i— o c o II ad z ci hi o z CD TD -t—' a CD O OJ 'to o c CD \ CO CO CD O c ro 1 ro > CO c: ro CD E II o x 100 A-2: Statistical Analysis of Average Masses from video samples and dip net samples, Site A - Tukey Test site & positions calculated critical result cage # tested q value q value P3 vs dip net 11.31 ' 3.633 Reject Ho P1 vs dip net 3.75 3.633 Reject Ho A-2 P2 vs dip net 0.02 3.633 D.N.R.Ho P3 vs P2 1.0.79 3.633 Reject Ho P1 vs P2 3.60 3.633 D.N.R.Ho P3 vs P1 7.22 3.633 Reject Ho dip net vs P2 6.92 3.314 Reject Ho A-3 P3 vs P2 5.11 3.314 Reject Ho dip net vs P3 1.27 3.314 D.N.R.Ho P2 vs P3 7.33 3.314 Reject Ho A-4 dip net vs P3 4.12 3.314 Reject Ho P2 vs dip net 3.68 3.314 Reject Ho P2 vs dip net 11.52 3.633 Reject Ho P3 vs dip net 6.61 3.633 Reject Ho A-6 P1 vs dip net 3.51 3.633 D.N.R.Ho P2 vs P1 7.13 3.633 Reject Ho P3 vs P1 2.68 3.633 D.N.R.Ho P2 vs P3 4.63 3.633 Reject Ho Note: D.N.R. = do not reject Ho = meanl is not significantly different from mean2 101 A - 3 : Stat is t ical A n a l y s i s of A v e r a g e M a s s e s f rom v ideo s a m p l e s , dip net s a m p l e s and harves t data , S i te B - Tukey Tes t site & posi t ions ca lcu la ted cri t ical result c a g e # tested q va lue q va lue P 3 v s harvest 4 .53 3 .633 Re jec t H o P1 v s harvest 0.96 3 .633 D . N . R . H o B-2 dip net v s harves t 0.91 3 .633 D . N . R . H o P 3 v s dip net 3.61 3 .633 D . N . R . H o P1 v s dip net 0.11 3 .633 D . N . R . H o P 3 v s P1 3.27 3.633 D . N . R . H o P 2 v s harvest 4 .06 3 .633 Re jec t H o dip net v s harvest 2 .32 3.633 D . N . R . H o B-4 P1 v s harvest 1.37 3 .633 D . N . R . H o P 2 v s P1 2 .73 3 .633 D . N . R . H o dip net v s P1 0.99 3.633 D . N . R . H o P 2 v s dip net 1.71 3 .633 D . N . R . H o Note: D . N . R . = do not reject H o = m e a n l is not signif icantly different f rom m e a n 2 1 0 2 A - 4 : Stat ist ical A n a l y s i s of A v e r a g e M a s s e s f rom v ideo s a m p l e s , d ip net s a m p l e s and harves t da ta , S i te C - T u k e y Tes t site & pos i t ions ca lcu la ted cri t ical result c a g e * tes ted q va lue q va lue P 3 v s P 2 6.01 3.858 Re jec t H o P1 v s P 2 2 .52 3.858 D . N . R . H o dip net v s P 2 2 .18 3.858 D . N . R . H o harves t v s P 2 1.88 3 .858 D . N . R . H o C - 2 P 3 v s harvest 3.91 3.858 Re jec t H o P1 v s harvest 0.54 3.858 D . N . R . H o dip net v s harves t 0.30 3.858 D . N . R . H O P 3 v s d ip net 3 .60 3.858 D . N . R . H o P1 v s dip net 0 .23 3.858 D . N . R . H o P 3 v s P1 3.52 3.858 D . N . R . H o P 3 v s P1 8.29 3.858 Re jec t H o harvest v s P1 6 .85 3.858 Re jec t H o P 2 v s P1 4.41 3.858 Re jec t H o dip net v s P1 3.64 3.858 D . N . R . H o C - 4 P 3 v s d ip net 5.07 3.858 Re jec t H o harves t v s d ip net 3.42 3.858 D . N . R . H o P 2 v s d ip net 0.87 3.858 D . N . R . H o P 3 v s P 2 4 .15 3.858 Re jec t H o harvest v s P 2 2 .49 3.858 D . N . R . H o P 3 v s harvest 1.78 3.858 D . N . R . H o Note : D .N .R . = do not reject H o = m e a n l is not signif icantly different f rom m e a n 2 103 CO CD H >» 0) - £ CO X O CO CD — OJ co 3 •c > "8 <D I ! o 00 CD CO 3 O •j= co n > O LL ro ° £ c £ ° E £ T> |&2i 1^  O O p s 8 I (D <D S E L- ( U CO CD 3 Q) i t X J CD CD « - i 3 > O , ro u -o aj g, o o X X "5 or CD or Q o o X X o o CD CD 'cF "oT rr or N- r--T f T f CN CO CN TJ-iri m CD i • < < o o o X X X o o o CD CD CD <j7 aT 'cuT or or or O O O X X X tj o n: CD CD a: ct: Q oo CD O Tf ° ° . T? Tt O Q i n CN C M C O T t • i • co m m o o X X or or z z Q ci o o r x *-* •*-* o o CD CD 'cF CD or or m m Is- i n Is- i n m Is- i n 00 00 co 00 CD 00 00 CD 00 t Is- CD 00 Is- CD o oo CN oo i n T— co i n CO CO csi cvi CN T ~ CO CO o o o o X X X X -*—' -1—' -4—' -*—' o o o o CD CD CD CD 'cF cF cF CD or or or or C D O C D o T t cn C D o T t C D C D O C D o T t C D C D o C O C O C O C O C O C O C O C O C O Is- Is-T t t CM CN T - CN 00 T t T - C M C O T f I I I I o o o o Is- o T t c T t i n C O o C O C O ,-: TD m a> m II c CD 1 T J CO o to o c CD I ro CO CD o c: ro CD ro CO cr CO or o x CD O 104 00 cu r -•t: ro x < > O z < oo 0) — CU CO ZJ •£ ro c > " x o 00 cu — CD s - i •j= ro c > O LL ^•5 ro c E o tz 0 00 o CD -D fU CD c5> 2> CD H= X ) O -4—» ro - CD ° ID o) S E l— CD cn CD 3 CO t c -a "a CD CD o ro u-o CD CD & CO o o X X o DC • ^ Z CD o o x x -«—> -*—> o o CD CD cu cF cc or CD CD O O CO CO Is- Is-Tf Tf CM CN T - N -0 0 CD i r i od LO CD • i < < o o o X X X t5 o or CD CD a? CD _ cc cc a o o o x x x t3 cc CD CD ^ CD CD • CC DC Q ^ CD i T 0 ) 0 0) CO CO CO 0 0 CD CM CO CD CD CD i r i co lO T-CN CO Tf i i i CD CQ CD O O X X DC DC z z ci ci o o CD CD a? CD DC DC LO LO LO Is- LO LO Is- lO 0 0 0 0 CD 0 0 co 0 0 oo CD 0 0 o N-Is- Is- CD 0 0 CD CN CM 0 0 CD oq Tf lO CD 0 0 T— CO T— T ~ CO CO O O o o X X X X •*-» 1 - -*—> o o CJ o CD CD CD CD CD CD CD CD DC CC CC DC CD CD Tf CD O O CD O CO CO CO CO Tf Tf 0 0 CN CM T — CM , n Tf Is- CO Q CN O LO n I S (D U) i n oo Tf oo T — CM CO Tf I I I I O O O O 105 A - 7 : Stat is t ical A n a l y s i s of A v e r a g e S p e e d s f rom v ideo s a m p l e s , m e a s u r e d in m/s - T u k e y T e s t si te & posi t ions ca lcu la ted crit ical result c a g e # tested q va lue q va lue P1 v s P 3 4 .26 3.314 Re jec t H o A - 5 P 2 v s P 3 3.57 3.314 Re jec t H o P1 v s P 2 0.69 3.314 D . N . R . H o P 3 v s P1 5.47 3.314 Re jec t H o A - 6 P 2 v s P1 2.79 3.314 D . N . R . H o P 3 v s P 2 2.68 3.314 D . N . R . H o P1 v s P 3 8.91 3.314 Re jec t H o B-3 P 2 v s P 3 6.19 3.314 Re jec t H o P1 v s P 2 2.72 3.314 D . N . R . H o P 3 v s P 2 5.41 3.314 Re jec t H o C-1 P1 v s P 2 3.43 3.314 Re jec t H o P 3 v s P1 1.98 3.314 D . N . R . H o P 3 v s P 2 13.15 3.314 Re jec t H o C - 2 P1 v s P 2 12.12 3.314 Re jec t H o P 3 v s P1 1.03 3.314 D . N . R . H o P 2 v s P1 13.97 3.314 Re jec t H o C - 4 P 3 v s P1 4 .20 3.314 Re jec t H o P 2 v s P 3 9.76 3.314 Re jec t H o Note : D .N .R . = do not reject H o = m e a n l is not signif icantly different f rom m e a n 2 106 A-8: Statistical Analysis of Average Speeds from video samples, measured in bl/s - Tukey Test site& cage# positions calculated tested q value critical q value result P2 vs P3 4.20 3.314 Reject Ho A-5 P1 vs P3 4.16 3.314 Reject Ho P2 vs P1 0.04 3.314 D.N.R.Ho P3 vs P1 5.84 3.314 Reject Ho A-6 P2 vs P1 2.23 3.314 D.N.R.Ho P3 vs P2 3.61 3.314 Reject Ho P1 vs P3 7.96 3.314 Reject Ho B-3 P2 vs P3 5.82 3.314 Reject Ho P1 vs P2 2.14 3.314 D.N.R.Ho P3 vs P2 5.46 3.314 Reject Ho C-1 P1 vs P2 4.07 3.314 Reject Ho P3 vs P1 1.39 3.314 D.N.R.Ho P3 vs P2 13.28 3.314 Reject Ho C-2 P1 vs P2 12.93 3.314 Reject Ho P3 vs P1 0.35 3.314 D.N.R.Ho P2 vs P1 12.80 3.314 Reject Ho C-4 P3 vs P1 2.69 3.314 D.N.R.Ho P2 vs P3 10.11 3.314 Reject Ho Note: D.N.R. = do not reject Ho = meanl is not significantly different from mean2 107 

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:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0058938/manifest

Comment

Related Items