UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The early life history and ecology of Columbia Lake burbot Taylor, Joshua L. 2001

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

Item Metadata


831-ubc_2001-0121.pdf [ 6.43MB ]
JSON: 831-1.0099526.json
JSON-LD: 831-1.0099526-ld.json
RDF/XML (Pretty): 831-1.0099526-rdf.xml
RDF/JSON: 831-1.0099526-rdf.json
Turtle: 831-1.0099526-turtle.txt
N-Triples: 831-1.0099526-rdf-ntriples.txt
Original Record: 831-1.0099526-source.json
Full Text

Full Text

T H E E A R L Y LIFE HISTORY A N D E C O L O G Y OF C O L U M B I A L A K E BURBOT b y JOSHUA L. T A Y L O R B. Sc. The University of British Columbia, 1997 A THESIS SUBMITTED IN PARTIAL F U L F I L M E N T O F T H E REQUIREMENTS FOR T H E D E G R E E OF M A S T E R OF SCIENCE In T H E F A C U L T Y OF G R A D U A T E STUDIES (Department of Zoology) We accept this thesis as conforming to the Required standard T H E UNIVERSITY OF BRITISH C O L U M B I A January 2001 © Joshua L. Taylor, 2001 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. The University of British Columbia Vancouver, Canada Department DE-6 (2/88) Abstract - This thesis examines aspects of the early life history and ecology of the Columbia Lake burbot population in southeast British Columbia. Growth was examined using otoliths from burbot sampled in shoreline habitats, a recreational ice fishery, and the spawning population. Columbia Lake burbot begin to recruit to the fishery, and no longer inhabit shoreline habitats, by age 2 or 3 (39 cm in length). The mean density of age 0 burbot per km of shoreline on 12 sites sampled from 1997 to 1999 ranged from 400 in 1997 to 62 in 1999. This suggests that recruitment is not seriously depressed and, thus, the decline in the burbot population may be best explained by changes in environment that affect post settlement life stages. Recruitment, as estimated by shoreline surveys of juvenile abundance and the age distribution of adults in the fishery and spawning population, varied greatly among cohorts. This variation is not well explained by spawner numbers and, presumably, is partly driven by environmental fluctuations. Reduced lifespan, as occurs with decreasing latitude and increasing fishing pressure, should reduce the survival of adults over periods when conditions are unfavorable for recruitment and, thus, reduce population stability. Egg development and early larval life were investigated under laboratory conditions. Egg survival peaked about 3°C and all embryos died at temperatures above 6°C. This narrow temperature tolerance during incubation may cause density independent mortality, especially when egg incubation temperatures are variable and borderline for survival. The minimum time to hatch at 5, 4, and 3°C was 28, 32, and 38 days, respectively. Newly hatched larvae averaged 3.47 mm in length and were positively phototactic. The mouth and swimbladder form between 5 and 10 days after hatching. After 27 days (mean length 4.25 mm) all larvae were neutrally buoyant and feeding exogenously. Habitat use by juvenile burbot was modeled statistically. Juveniles use crevices, especially the interstitial spaces between substrate particles, as cover. Because much of the shoreline habitat suitable for juvenile burbot is above water from late fall to early spring, competition for cover may regulate recruitment in Columbia Lake. ii Table of Contents ABSTRACT II LIST OF TABLES V TABLE OF FIGURES V ACKNOWLEDGEMENTS VIII GENERAL INTRODUCTION 1 STUDY SITE 4 CHAPTER 1: GROWTH AND RECRUITMENT 8 INTRODUCTION 8 METHODS 9 Shoreline sampling 9 Creel and spawner sampling 9 A g i n g 1 0 Fitting a growth function 1 3 Comparison of cohort size 1 5 RESULTS 1 5 Size and age of recruitment 1 5 Growth pattern 1 7 Variat ion i n cohort size 1 7 DISCUSSION ._. 1 9 CHAPTER 2: EGGS AND LARVAE 24 INTRODUCTION 2 4 METHODS 2 4 Egg incubation 2 4 Rearing larvae 2 6 Phototaxis experiments 2 7 iii R E S U L T S 2 8 Egg size 2 8 Survival of eggs to hatching 2 8 Time to hatch 2 9 Early larval development : 3 0 Phototaxis 3 2 D I S C U S S I O N 3 2 CHAPTER 3: JUVENILE HABITAT USE 36 I N T R O D U C T I O N 3 6 M E T H O D S 3 7 Habitat use survey 3 7 Model ing habitat use 3 9 R E S U L T S 41 Burbot, torrent sculpins, and sediment size 4 3 Detailed models of burbot habitat use 4 7 D I S C U S S I O N 5 1 CONCLUSIONS 59 LITERATURE CITED 62 iv List of Tables T A B L E I. PERCENT E G G SURVIVAL FROM FERTILIZATION TO H A T C H I N G FOR BURBOT E G G BATCHES INCUBATED A T 3, 4, A N D 5 ° C IN 1999 28 T A B L E II. T H E SIZE R A N G E A N D R A N K OF T H E SUBSTRATE CATEGORIES MEASURED IN T H E HABITAT USE SURVEY 38 T A B L E III. SUMMARY OF T H E SIZE A N D A B U N D A N C E OF A G E 0 BURBOT, BURBOT OLDER T H A N A G E 0, A N D TORRENT SCULPINS SAMPLED IN T H E HABITAT USE TRANSECT SURVEY 41 T A B L E IV. DEFINITIONS A N D SUMMARY OF T H E OBSERVED VALUES FOR T H E FIRST DEGREE TERMS CONSIDERED AS FACTORS POTENTIALLY AFFECTING BURBOT A B U N D A N C E (N = 153) 42 T A B L E V . SUMMARY OF T H E VARIABLES INCLUDED IN T H E DETAILED M O D E L OF HABITAT USE FOR A G E 0 BURBOT (N = 129, R 2 = 0.45) 47 T A B L E VI . SEMIPARTTAL R 2 VALUES FOR VARIABLES INCLUDED IN T H E DETAILED M O D E L OF A G E 0 BURBOT HABITAT USE 50 T A B L E VII. SUMMARY OF T H E VARIABLES INCLUDED IN T H E DETAILED M O D E L OF HABITAT USE FOR BURBOT OLDER T H A N A G E 0 (N = 153, R 2 = 0.71) 50 Table of Figures FIGURE l . M A P O F COLUMBIA L A K E 5 FIGURE 2. L E N G T H FREQUENCY HISTOGRAM FOR T H E 230 JUVENILE BURBOT C A U G H T O N T H E TWELVE TRANSECTS USED TO SAMPLE BURBOT DENSITY FROM 1997 TO 1999. N O T E T H E NON-OVERLAPPPNG SUMMER L E N G T H DISTRIBUTIONS OF A G E 0 A N D OLDER BURBOT... 10 FIGURE 3. A N OTOLITH REMOVED FROM A N A G E 2 BURBOT SAMPLED FORM COLUMBIA L A K E O N N o v . 20,1999. N O T E T H E NARROW Z O N E OF OPAQUE GROWTH O N ITS EDGE (SHOWN BY A N ARROW) 12 FIGURE 4. T H E RELATIONSHIP BETWEEN T H E L O G OF ROUTINE O X Y G E N CONSUMPTION A N D T H E L O G OF BODY MASS FOR BURBOT ESTIMATED USING D A T A FROM SHODJAI (1980)... 14 V FIGURE 5. OBSERVED A N D PREDICTED SIZE AT A G E FOR POST-LARVAL BURBOT SAMPLED O N -A N D OFFSHORE IN COLUMBIA L A K E . N O T E T H E THRESHOLD OF ABOUT 39 C M (DASHED LINE), ABOVE WHICH THEY BEGIN T O RECRUIT TO T H E ICE FISHERY A N D N O LONGER INHABIT SHORELINE HABITATS 16 FIGURE 6. M E A N (± SE) L O G A B U N D A N C E OF FOUR COHORTS OF BURBOT FROM COLUMBIA L A K E AS SAMPLED O N TWELVE SHORELINE TRANSECTS IN 1997,1998, A N D 1999 18 FIGURE 7. L E N G T H A N D YEAR CLASS DISTRIBUTION OF 83 C O L U M B I A L A K E BURBOT SAMPLED FROM T H E FISHERY A N D SPAWNING POPULATION IN T H E WINTER OF 1995/1996 19 FIGURE 8. T H E NUMBER OF BURBOT CAPTURED IN A WEIR ACROSS T H E U N N A M E D SPAWNING TRIBUTARY TO T H E SOUTHWEST CORNER OF COLUMBIA L A K E DURING T H E SPAWNING SEASON FROM 1996 TO 2000 (ARNDT 2000, A R N D T A N D H U T C H I N S O N 2000) 20 FIGURE 9. COMPARISON OF LENGTH-FREQUENCY IN 1999 A N D 2000 FOR BURBOT CAPTURED A T T H E TRIBUTARY WEIR O N COLUMBIA L A K E 21 FIGURE 10. T H E INFLUENCE OF INCUBATION TEMPERATURE O N MINIMUM TTME FOR BURBOT EGGS TO H A T C H 29 FIGURE l l . M E A N TIME TO H A T C H , FIT TO A REGRESSION LINE WITH 95 PERCENT CONFIDENCE INTERVALS, FOR COLUMBIA L A K E BURBOT EGGS INCUBATED A T 3, 4, A N D 5 ° C 30 FIGURE 12. A PHOTOGRAPH OF TWO COLUMBIA L A K E BURBOT H A T C H E D A N D REARED IN T H E LAB: O N E FIVE DAYS O L D (TOP), A N D T H E OTHER 27 DAYS O L D (BOTTOM). A S A CONSEQUENCE OF PRESERVATION IN 95 % E T H A N O L FOR 18 M O N T H S PRIOR TO TAKING THIS PHOTOGRAPH, SOME SHRINKAGE OF T H E OIL GLOBULE A N D YOLK OCCURRED 31 FIGURE 13. FREQUENCY DISTRIBUTION OF SEDIMENT SIZE INDEX (SED) VALUES FOR 153 TRANSECTS SAMPLED IN SHORELINE HABITATS OF COLUMBIA L A K E DURING T H E SUMMER OF 1998. N O T E T H A T S E D C A N R A N G E F R O M ZERO W H E N A T R A N S E C T S SUBSTRATE IS ENTIRELY FINES TO 100 W H E N A TRANSECT'S SUBSTRATE IS ENTIRELY BOULDERS (TABLE II) 43 vi FIGURE 14. OBSERVED A N D PREDICTED SHORELINE A B U N D A N C E OF A G E 0 BURBOT VERSUS A SEDIMENT SIZE INDEX (SED) FOR COLUMBIA L A K E IN 1998. N O T E T H A T S E D C A N R A N G E FROM ZERO W H E N A TRANSECT'S SUBSTRATE IS ENTIRELY FINES TO 100 W H E N A TRANSECTS SUBSTRATE IS ENTIRELY BOULDERS (TABLE II). PREDICTIONS WERE M A D E USING BOTH A CUBIC FUNCTION OF S E D [Y = EXP ( - 1.83 -0.39 X S E D + 0.0223 X SED2 -0.00028 X SED3), P < 0.0001, R2 = 0.24], A N D A MORE DETAILED M O D E L OF HABITAT USE (TABLE V ) . PREDICTIONS FROM T H E MORE DETAILED M O D E L ASSUME T H A T D E P T H = 50 C M , TIME = DAY 46 (JULY 13), A N D BOTTOM COVER (BC) = 0 % 44 FIGURE 15. OBSERVED A N D PREDICTED SHORELINE A B U N D A N C E OF BURBOT OLDER T H A N A G E 0 VERSUS A SEDIMENT SIZE INDEX (SED) FOR COLUMBIA L A K E IN 1998. N O T E THAT S E D C A N R A N G E FROM ZERO W H E N A TRANSECT'S SUBSTRATE IS ENTIRELY FINES TO 100 W H E N A TRANSECT'S SUBSTRATE IS ENTIRELY BOULDERS (TABLE II). PREDICTIONS WERE M A D E USING BOTH A QUADRATIC FUNCTION OF S E D [Y = EXP ( -2.12 - 0.015 X S E D + 0.00092 X SED 2 ) , P < 0.0001, R 2 = 0.66], A N D A MORE DETAILED M O D E L OF HABITAT USE (TABLE VII). PREDICTIONS FROM T H E MORE DETAILED M O D E L ASSUME T H A T THERE ARE N O UNDERCUT BANKS (UC) A N D T H A T SURFACE COVER (SC) = 0% 45 FIGURE 16. OBSERVED A N D PREDICTED SHORELINE A B U N D A N C E OF TORRENT SCULPINS VERSUS A SEDIMENT SIZE INDEX (SED) FOR COLUMBIA L A K E IN 1998. N O T E T H A T S E D C A N R A N G E FROM ZERO W H E N A TRANSECT'S SUBSTRATE IS ENTIRELY FINES TO 100 W H E N A TRANSECT'S SUBSTRATE IS ENTIRELY BOULDERS (TABLE II). PREDICTIONS WERE M A D E USING A POLYNOMIAL FUNCTION OF S E D [Y = EXP ( - 2.12 - 0.015 X S E D + 0.00092 X SED2), P < 0.0001, R2 = 0.48] 46 FIGURE 17. SHORELINE A B U N D A N C E OF A G E 0 BURBOT VERSUS BOTTOM COVER (BC) A N D TIME W H E N T H E SEDIMENT SIZE INDEX (SED) = 0 FOR COLUMBIA L A K E IN 1998, AS OBSERVED, A N D AS PREDICTED BY A DETAILED M O D E L OF HABITAT USE (TABLE V) . N O T E T H A T T H E SUBSTRATE O N A SITE WITH S E D = 0 i s ENTIRELY HNES (TABLE II) 48 FIGURE 18. SHORELINE A B U N D A N C E OF A G E 0 BURBOT VERSUS D E P T H A N D TIME FOR COLUMBIA L A K E IN 1998, AS OBSERVED, A N D AS PREDICTED BY A DETAILED M O D E L OF HABITAT USE (TABLE V ) . T H E PREDICTIONS ASSUME T H A T T H E SEDIMENT SIZE INDEX (SED) A N D BOTTOM COVER (BC) EQUAL THEIR A V E R A G E VALUES FOR T H E 129 TRANSECTS SAMPLED, 18.4 A N D 26.1 %, RESPECTIVELY 49 vii Acknowledgements This study was only part of a larger project investigating the biology of the Columbia Lake burbot population. The entire project was funded by the Columbia Basin Fish and Wildlife Compensation Program (CBFWCP) and organized by CBFWCP employees Harald Manson and Steve Arndt. Employees of the British Columbia j Ministry of Environment, Land and Parks, especially Bill Westover and Jay Hammond, helped initiate this study and continued to provide support through later stages. The thoughtful guidance of my thesis supervising committee members — Dr. J. Donald McPhail, Dr. Eric Taylor, Dr. Scott Hinch, and Dr. John Dower — is greatly appreciated. The people who helped me in the field include Peter Mylecreest, Tara Fleming, Mark Phillpotts, Larry Taylor, Derek Whyte, Mark Stevenson, Joyce Hutchinson and Colin Spence. viii General Introduction Fish conservation depends largely on our ability to understand and predict the responses of fish to environmental perturbations. These responses result from the biology of individual species and the interactions between species. Both factors, however, change drastically through a fish's life. This is a consequence of the range in body size spanned over the life of most fish and the scaling of morphological, physiological, and behavioral constraints with body size (Evens et al. 1987, Ebenman and Persson 1988). Thus, without a thorough understanding of their life history, it can be very difficult to predict the impacts of environmental perturbations on fish. Burbot, Lota lota, are the only truly freshwater member of the cod family, Gadidae (Nelson, 1994). They occupy lakes and rivers throughout the Holarctic region. Adult burbot are demersal, nocturnal piscivores that can grow to over one meter in length (McPhail 1997). Recreational fisheries for burbot occur during winter, often through ice. Typical fishing methods are jigging, set lines, or spears. The quality of burbot flesh is high (comparable to that of the commercially exploited marine Gadids) and, since burbot reach a large size, they can give anglers a good fight. Despite these merits, burbot are not widely esteemed by North America's recreational fishers. This may be due to the burbot's arguably unattractive appearance, and the recreational fishing culture's general obsession with salmonids. Accordingly, burbot fisheries have traditionally had few, if any, regulations. This scarcity of regulations for burbot fisheries sets up the potential for overexploitation. While marked declines in burbot fisheries appear common, the evidence is usually anecdotal, and fishing can rarely be isolated as the only contributing factor (McPhail 1997). An exception is the well documented collapse of burbot fisheries in the interior of Alaska (Lafferty and Vincent-Lang 1991, Taube and Bernard 1995). Several interacting factors are suspected to have contributed to declines of burbot populations in the Columbia River system. Besides recreational fishing pressure, these include impoundments and associated changes in flow and nutrient regimes (Hildebrand 1991, McPhail 1997, Paragamian et al. 2000), dyking in the lower reaches of the Kootenay River (S. Arndt, Columbia Basin Fish and Wildlife Compensation Program, personal communication), and climate change (McPhail 1997, Cohen et al. 1 2000). Damaging factors suspected elsewhere include channelization and impoundment of the Missouri River (Fisher 2000), acid rain in Ontario and Scandinavia (Gunn 1982, Berqquist 1991), eutrophication of Finnish Lakes (Kjellman et al. 2000) and sea lampreys (Petromyzon marinus) in the Great Lakes (Smith and Tibbies 1980). In some cases, when corrective measures have been undertaken, such as lime treatment of acidified lakes in Scandinavia (Berqquist 1991), lamprey control in the Great Lakes (Bruesewitz et al. 1989), and strict fishing regulations in the interior of Alaska (Taube and Bernard 1995), burbot populations have rebounded. Declines in burbot populations and increased interest in the conservation of biodiversity have stimulated a new approach to the management of burbot. For instance, the Kootenay region of southeastern British Columbia first imposed region wide catch limits for burbot in 1968 (Bill Westover, British Columbia Ministry of the Environment, personnel communication). The daily limit was 12 and possession limit was 24. In 1981, British Columbia officially recognized burbot as a sports fish (Bill Westover, British Columbia Ministry of the Environment, personnel communication), and today, the daily catch limit for burbot is five in all regions of the province. In addition, burbot have been Red Listed1 in areas of the Kootenay region, and are under consideration for protection as an Endangered Species in Idaho (V. Paragamian, Idaho Department of Fish and Game, personal communication). Although spear fishing and set lines were banned in the Kootenay Region of southeast British Columbia in 1974 and 1987, respectively, these highly effective fishing methods are still allowed in some other regions of the province. A possible impediment to the further development of rational strategies for the management of burbot is the meager data on their life history and ecology. Since local adaptations and species interactions will presumably vary in different environments, this dearth of information is compounded by the diversity of environments inhabitated by burbot populations. Furthermore, much of what is known in North America is based on studies conducted to the east of the Continental Divide. Typically, wide ranging northern species like burbot survived glaciation in several refugia (McPhail and Lindsey 1970, Bernatchez and Wilson 1998). Consequently, burbot 1 Red Listed is the name of the highest provincial conservation status given to indigenous species or subspecies in British Columbia 2 from west of the Continental Divide may differ slightly from those to the east, and such differences may be reflected in their life history characteristics. Burbot eggs average about one millimeter in diameter and batch fecundity can be as high as 3.5 million (McPhail 1997). Burbot hatch at a length of 3-4 mm and pass from a pelagic larval stage to a demersal juvenile stage during their first year of life. The juvenile lifestyle is assumed by a length of about 4 cm (Meshkov 1967, Fischer 1999). The larvae are positively phototactic and often school (McPhail 1997). Juveniles are nocturnal and solitary (McPhail 1997) and their morphology is essentially the same as their parents (Meshkov 1967). The burbot's late winter spawning time, enormous fecundity, and extended larval period make it unique amongst the freshwater fishes of North America. This rest of this thesis is organized into three chapters proceeded by a detailed description of the study site, Columbia Lake in the southeast corner of British Columbia. The first chapter will examine growth and variation in recruitment. The second chapter will investigate egg development and early larval life under laboratory conditions. The final chapter will investigate the shoreline habitat use of juveniles. This thesis will not . only increase our knowledge of burbot early life history and ecology in general, but also address the specific concerns of inadequate knowledge in the west and population declines in the Columbia River system. 3 Study Site Columbia Lake, the source of the Columbia River, is located in the Kootenay region of southeastern British Columbia at 50° 15' N, 115° 50' W and an elevation of about 809 m. Columbia Lake has a surface area of 25.7 km 2, a length from north to south of about 13 km, a mean depth of 2.9 m, and a maximum depth of 6 m (Figure 1). Columbia Lake is mesotrophic and does not form a summer thermocline. During the summers from 1997 to 1999, the peak offshore surface temperatures in Columbia Lake ranged from 19.5 to 24.0°C. There are about 15 inlets draining into Columbia Lake but most of these are ephemeral. The two major inlets to Columbia Lake are Dutch Creek in the northwest corner and a short unnamed spring fed stream in the southwest corner (Figure 1). The only outlet stream is the Columbia River at the northeast corner of the lake (Figure 1). The Columbia River enters Windermere Lake about 15 km downstream from Columbia Lake. Maps show a man made canal connecting the southwest corner of Columbia Lake to the Kootenay River (Figure 1) but I have been told by locals that this connection is incomplete. During deglaciation, a large lake formed in the area now occupied by Columbia and Windermere lakes (Ryder 1981). This glacial lake was 40 to 50 m higher than the modern lakes (Ryder 1981). Thus, it is not surprising that the dominant substrate of Columbia Lake, including its shoreline, is fine particles of silt. Gravel and rocky substrates are rare. Human activities, the building of a railway (Figure 1) and jetties, have provided most of the gravel and rocky substrate in Columbia Lake. This artificial substrate rarely extends to much over a meter below the high water mark and, due to an average water level fluctuation of about one meter, is mostly above water for part of the year. The largest natural location of rocky substrate is the southeast corner of the lake where it is bordered by rock bluffs for approximately one km. Because of its fine substrate, moderate nutrient levels, and shallow depth, Columbia Lake has an abundance of aquatic macrophytes. Submergent macrophytes-mainly pondweed (Potamogeton ssp.), stonewort (Chara ssp.) and northern millfoil (Myriophyllum exalbescens) - cover approximately 80% of the main body of the lake (R.L. & L. Environmental Services 1993). The nearshore area is dominated by bulrush (Scirpus spp). 4 Fourteen fish species occur in Columbia Lake. The twelve native species are rainbow trout (Oncorhynchus mykiss), bull trout (Salvelinus confluentus), kokanee (Oncorhynchus nerka), mountain whitefish (Prosopium williamsoni), burbot (Lota lota), large scale sucker (Catostomous macrocheilus), longnose sucker (Catostomous catostomous), northern pike minnow (Ptychocheilus oregonensis), peamouth chub (Mylocheilus caurinus), redside shiner (Richardsonius balteatus), longnose dace (Rhinichthyes cataractae), and torrent sculpin (Cottus rhotheus). Sunfish (Lepomis gibbossus), and small mouth bass (Micropterus dolomieu) were introduced to the upper Columbia but have not yet become well established in Columbia Lake. A 1992 fish inventory study of Columbia Lake (R.L. & L. Environmental Services 1993) caught 11 of the 14 species present. Sampling methods used were gill netting, beach seining, minnow traps, and set lines. Burbot were the fifth most abundant fish sampled (2.7% of catch) after peamouth chub (35.9%), northern pike minnow (30.4%), mountain whitefish (18.3%), and largescale sucker (6.5%). Although Columbia Lake currently supports a popular winter ice fishery for burbot, local anglers report that both the size and abundance of burbot in Columbia Lake have declined from historic levels. Liberal fishing regulations and the burbot's local popularity have probably contributed to heavy exploitation in the past. For instance, before it became illegal, spears were used to harvest burbot congregating in Dutch Creek and shallow water areas of Columbia Lake during the spawning season. Even during the 1980's, however, poaching of burbot spawning in Dutch Creek was still a concern of conservation officers (G. Oliver, G. G. Oliver and Associates, personal communication). Therefore, over-exploitation by recreational fishers probably contributed to the apparent decline. Additionally, although substantial numbers of burbot used to spawn in a side channel of Dutch Creek's alluvial fan (B. Harmsworth, former British Columbia Conservation Enforcement Officer, personnel communication), few burbot are though to spawn in the creek anymore (S. Arndt, Columbia Basin Fish and Wildlife Compensation Program, personal communication). Air photo's from 1976, 1988 and 1995 support the claim by locals that the course of the Dutch Creek channel has shifted from a southerly 6 coarse, towards Columbia Lake, to an easterly course, towards the Columbia River (Urban Systems 1997). Local land owners and the railway have attempted to channelize the upper portion of the alluvial fan in order to resist this natural movement (R. Dubielewicz, Canadian Pacific Railway, personal communication). Currently, the spring freshet deposits gravel in front of the side channel flowing into Columbia Lake where burbot spawning used to take place, almost completely blocking it off under low flow conditions. During the 1980's, however, locals kept it open to ensure surface flow for their personal use (G. Oliver, G. G. Oliver and Associates, personal communication). Presently, Columbia Lake burbot are known to spawn in both the lake and the spring-fed tributary that enters the southwest corner of the lake (Arndt and Hutchinson 2000). Although no burbot appear to be spawning in the historic site in the alluvial fan of Dutch Creek, a few age 0 burbot were sampled in the lower reaches of Dutch Creek in 1997 (Baxter 1998). This suggests either that there is still some spawning in Dutch Creek or that some age 0 burbot move into the stream from the lake. 7 Chapter 1: Growth and recruitment Introduction Two examples of size related events in the life history of many fish are the movement from a juvenile rearing habitat to an adult habitat, and recruitment to a fishery. Presumably, the movement between habitats is meditated by some body size related tradeoff between growth and predation risk (Werner et al. 1983). Recruitment to a fishery occurs when fish become large enough to be susceptible to fishing gear and deemed worthy of keeping. Body size at recruitment to a fishery is important because it affects the potential for production and recruitment overfishing. Fisheries with within-age size selectivity can also affect population growth patterns on both ecological and evolutionary time scales. Because of environmental fluctuations, the rates of mortality due to starvation and predation during the early life stages of fishes are expected to vary from year to year (Sissenwine, 1984). Given their tiny eggs and larvae, this variation could be especially drastic in burbot. An understanding of variation in recruitment success affects how a fisheries manager interprets trends in population abundance. For instance, if a population is known to have large variation in recruitment success and the individuals in the population have long lifespans, then one may still hold out hope for a turn around after a short term trend of poor recruitment (Warner and Chesson 1985). If relative cohort strength is largely determined by a certain life history stage, then that life history stage might be surveyed by fisheries managers as an index of future abundance (Sunby et al. 1989, de Lafontaine et al. 1992). For small populations, recruitment variation is of special concern because it can contribute to extinction (Primack 1993). In this chapter the growth pattern of Columbia Lake burbot is described through the juvenile and adult life stages. The size and age distribution of Columbia Lake burbot is compared among shoreline habitats, the spawning population and the fishery. Variation in recruitment success between cohorts is also investigated. 8 Methods Shoreline sampling The abundance of juvenile burbot was measured on transects during the summers from 1997 to 1999. Each transect was a 27.5 m stretch of shoreline sampled out to either 1 m depth or 27.5 m from shore, whichever came first. The sampling gear was a DC backpack electrofisher with a 20 cm diameter anode, a pulse frequency of 100 Hz, and a current of 0.5 A. The abundance of burbot on a transect was measured as the combined catch from two passes separated by 15 min. Electrofishing was always performed by the same person, while wearing polarized sunglasses, and only when light and wind conditions allowed adequate visibility. In 1997, forty transects were sampled from June 28 to August 1. These forty sites were evenly distributed around the lake and well represented the variety of habitats present. A random sample of 12 transects was chosen from the original 40 transects and re-sampled in 1998 and 1999. One of the twelve transects was 55 mm long rather than 27.5 m long. The abundances measured at this transect have been divided in half. In 1998, the twelve transects were sampled between June 29 and August 1, and no more than two days from when they had been sampled the previous year. In 1999, all twelve transects were sampled over five days starting on July 11. From 1997 to 2000, opportunistic backpack electrofishing of shoreline habitats was conducted to obtain additional size at age data. This opportunistic sampling was targeted at larger and generally older juveniles than those sampled in the transect survey. Size at age data from 128 of these opportunistically sampled burbot were used in this study. Creel and spawner sampling Burbot from Columbia Lake's ice fishery and spawning population were also sampled for size at age data. A creel survey of the burbot ice fishery on Columbia Lake in the winter of 1995/1996 provided 81 burbot. In addition, twelve spawners were collected at a trap and weir set across the unnamed tributary at the southwest corner of the lake (Arndt & Hutchinson 2000); 2 in 1996 and 10 in 1997. 9 Aging The hatching date, when a burbot is 0 days old, was assumed to be March 15. This date was based on the assumptions that fertilization would occur in early February (Arndt and Hutchinson 2000) and that incubation would take almost 1.5 months (Chapter 2). Since the length of age 0 and age 1 burbot did not begin to overlap until the end of August, burbot caught before the end of August could be assigned an age of either 0 or older based on length alone (Figure 2). The remaining burbot, those which were not assigned an age of 0 years based on length alone, were assigned ages based on the number of annuli on their saggital otoliths. 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 Total length (mm) Figure 2. Length frequency histogram for the 230 juvenile burbot caught on the twelve transects used to sample burbot density from 1997 to 1999. Note the non-overlapping summer length distributions of age 0 and older burbot. 10 Otoliths were removed with forceps, wiped clean, and stored either in 15 percent ethanol or dry in scale envelopes. All otoliths were first examined with a dissecting microscope under reflected light against a black background. Otoliths of burbot older than age 3 were also split and examined with oblique light. This was because annuli that were ambiguous on whole otoliths were sometimes more obvious on split otoliths. The first annulus was verified using length frequency data, as described above. Hyaline (clear) growth was seen at the edge of the otolith during the summer (Figure 3). Opaque edge growth occurred during the winter (Figure 3). Since otoliths were not collected from burbot sampled during the shoreline transect survey, age could only be estimated from length-frequency data (i.e. as age 0 or older than age 0). Of the 230 burbot sampled on the twelve transects from 1997 to 1999, 30 were over age 0. These 30 fish ranged in length from 11.7 to 22.7 cm (Figure 2). Of the burbot opportunistically sampled from shoreline habitats and aged using otoliths many age 1 individuals were over 20 cm in length, a few age 1 individuals were over 23 cm in length and the smallest age 2 individual was 20 cm long. Therefore, most of the 30 fish sampled on transects that were older than age 0 were probably age 1. To make inferences about relative cohort abundance, I assumed 29 of these fish were age 1. The 22.7 on individual sampled in 1997 probably was older. Because enough burbot older than age 0 were obtained during opportunistically shoreline sampling, I decided not to include these 30 fish when estimating a growth curve. 11 Figure 3. An otolith removed from an age 2 burbot sampled form Columbia Lake on Nov. 20,1999. Note the narrow zone of opaque growth on its edge (shown by an arrow). 12 Fitting a growth function The shoreline, creel and spawner samples were combined to estimate the growth pattern through the juvenile and adult life stages. The shoreline samples consisted of burbot from more recent cohorts (1995 to 1999) than the creel or spawner samples (1987 to 1993). Therefore, the validity of fitting a growth curve to the combined sample rests on the assumption that growth pattern did not differ much between older and more recent cohorts. The model of growth used was a generalization of the Von Bertalanffy growth function (VBGF) (Moreau 1987). The VBGF assumes that oxygen consumption of fish is proportional to a 2/3 power of body mass. The generalized Von Bertalanffy growth function (GVBGF) (Pauly 1981, Moreau 1987) allows flexibility in assigning the power to which oxygen consumption is proportional to body mass and, thus, can provide a better fit than the VBGF. The GVBGF in terms of length growth is expressed as L t = L - {1 - exp [ - K D (t - to)]}1/0 where L t is length at time t, L„ is the mean asymptotic length, K is a stress factor, D is a surface factor, and to is the hypothetical age the fish would have had at zero length. Time was measured in months (30.4 days). The surface factor D is calculated as D = 3 (1 - d) where d is the power to which oxygen consumption is proportional to body mass. Shodjai (1980) estimated routine oxygen consumption for wild reared burbot of different sizes. Although his total sample size was twelve burbot, he only presents four data points. Each of these data points, however, represented the mean routine oxygen consumption and body mass for two to four burbot of similar size. The mean body size for the four data points ranged from 79.4 to 305.4 g. The slope of the weighted G M regression line (Ricker 1973) of log oxygen consumption on log body mass for these four data points was 0.84 (Figure 4). This slope, 0.84, is an estimate of d (the power to which oxygen consumption is proportional to body mass) and, consequently, can be used to estimate that D (the surface factor in the GVBGF) equals 0.48. Pauly (1981) empirically derived a formula that allows the estimation of d from a fish species' maximum body size. McPhail and Paragamian (2000) presented a photo from the 1930's of a 15.44 kg 13 burbot caught in Windermere Lake, only about 15 km downstream from Columbia Lake. This burbot is one of the largest known caught from British Columbia. Since, the value of d estimated by Pauly's formula for a fish stock with a maximum size of 15 kg is 0.82, the d value of 0.84 estimated from Shodjai's data seems realistic. 1.55 "i 1 0.95 H 0.85 -I 1 1 1 1 1 1 1 1 1.75 1.85 1.95 2.05 2.15 2.25 2:35 2.45 2.55 Log 1 0 body mass (g) Figure 4. The relationship between the log of routine oxygen consumption and the log of body mass for burbot estimated using data from Shodjai (1980). All other parameters in the GVBGF, besides D, were estimated using the nonlinear regression platform of JMP™ statistical software. Heterodasticity was corrected for using iteratively reweighted least squares regression (Neter et al. 1996). A variance function was used to obtain the weights for the iteratively reweighted least squares process. The variance function was estimated by squaring the predicted values 14 from a regression of absolute residuals against the reciprocal of age. The estimated regression coefficients stabilized after three iterations. Comparison of cohort size The abundances of age 0 and age 1 burbot in the 1997,1998, and 1999 cohorts was compared using the abundances measured on the 12 transects sampled over all three years. A repeated measures A N O V A was used to statistically compare the age 0 abundances between years. A repeated measures A N O V A was necessary because the same 12 sample sites were sampled each year. The logio (Y + 1) transformation was used to reduce the skewness and heteroscedasticity of the observed abundances. Multiple comparisons were done using the Tukey test. The critical level was set at a = 0.05. The potential effect of differences in sampling dates between years was examined by plotting the residuals from the A N O V A model against sampling date. The age composition of the 93 burbot in the creel and spawner samples, described above, was also examined for evidence of year-class dominance. Results Size and age of recruitment For Columbia Lake's burbot, 39 cm appears to be a threshold above which they begin to recruit to the ice fishery and no longer inhabit shoreline habitats (Figure 5). This threshold is crossed sometime between ages 2 and 3 (Figure 5). Consequently, between ages 2 and 3, burbot from the creel and spawner samples were larger than burbot from the shoreline samples. In other words, the fishery appears to select for the largest, and presumably the fastest growing, individuals. 15 70 0 - I 1 1 1 1 1 1 1 1 1 1 0 12 24 36 48 60 72 84 96 108 120 A g e (months) Figure 5. Observed and predicted size at age for post-larval burbot sampled on- and offshore in Columbia Lake. Note the threshold of about 39 cm (dashed line), above which they begin to recruit to the ice fishery and no longer inhabit shoreline habitats. 16 Growth pattern As described above, burbot size varied between the shoreline, creel, and spawner samples during ages 2 and 3. As a consequence, the average size of age 2 and 3 burbot could not be determined accurately, and the 67 burbot of ages 2 and 3 were not included when fitting the GVBGF presented below. Incidentally, however, the estimated growth function turned out to be almost identical with or without the inclusion of the size at age data for age 2 and 3 burbot. The estimated GVBGF was: L t = 56.36 {1 - exp [ - 0.0907 * 0.48 (t + 5.76)]}V048 The predicted mean length of Columbia Lake burbot when 3.5 months old (i.e., at the end of June) was 5.7 cm (Figure 5). The growth function had an inflection point at about 11 months (mid-February) and 14.3 cm (Figure 5). The estimated mean length when 48 months old was 45.6 cm (Figure 5). Mean asymptotic length was estimated as 56.36 cm with a 95 % likelihood confidence interval of 54.68 to 58.20 cm. Variation in cohort size Large variations in year class strength were apparent from the transect survey (Figure 6). The difference in abundance of age 0 burbot among years was statistically significant (repeated measures ANOVA, P = 0.0005, R 2 = 0.63). No trend in the residuals was apparent when plotted against sampling date. A Tukey test suggested that age 0 abundance was larger in 1997 than in either 1998 or 1999, but that there was no difference between 1998 and 1999. Visual comparison of age 1 burbot abundances (Figure 6) strongly suggests that the 1996 cohort was also much smaller than the 1997 cohort. 17 1.2 Z 0.8 A 0.6 0.4 0.2 a g e c l a s s (years) Figure 6. Mean (± SE) log abundance of four cohorts of burbot from Columbia Lake as sampled on twelve shoreline transects in 1997,1998, and 1999. The mean, untransformed abundance per transect for age 0 burbot from the 1997, 1998 and 1999 cohorts were 11.0, 2.7 and 1.7, respectively. This is a four times difference between 1997, the year of the highest estimate, and 1998, the year of the next highest estimate. The mean untransformed abundance per transect for age 1 burbot from the 1996,1997 and 1998 cohorts were 0,1.5 and 0.5, respectively. The age distribution of the 83 Columbia Lake burbot sampled from the ice fishery and spawning population during the winter of 1995/1996 suggests that the 1991 and 1993 cohorts were considerably larger in abundance than the 1992 cohort (Figure 7). 18 Earlier cohorts, 1987 to 1990, are poorly represented in the 1995/1996 sample (Figure 7). Of the 10 fish sampled from the spawning tributary in 1997, one came from each of the 1989,1990 and 1993 cohorts, and the remaining 7 came from the 1991 cohort. 415 445 475 505 Total length (mm) 535 565 595 Figure 7. Length and year class distribution of 83 Columbia Lake burbot sampled from the fishery and spawning population in the winter of 1995/1996. Discussion A few of the spawners sampled from the trap and weir across the unnamed tributary to Columbia Lake from 1996 to 2000 were in the 60 to 70 cm length range (Arndt 2000, Arndt and Hutchinson 2000). Only one spawner over 70 cm in length was sampled, but it was 84 cm (Hutchinson 1996). Therefore, the estimated mean asymptotic size of Columbia Lake burbot estimated in this study, 56.36 cm with a 95 % likelihood confidence interval of 54.68 to 58.20 cm, may be slightly low. One potential reason for underestimating L„ is the apparent selectivity of Columbia Lake's ice fishery for faster growing individuals (Figure 5). When a fishery is selective for the faster growing individuals within an age group, and a growth curve is estimated from the average size 19 at capture of successive age groups, then the rate of decline in growth rate with age will tend to be overestimated (Ricker 1979). The resulting growth curve represents the population growth pattern rather than the growth pattern of individual burbot. One way to reduce the effect of within-age class selective mortality on the estimation of the rate of decline in growth rate with age is the use of lengths back-calculated from otoliths rather than lengths measured at capture (Ricker 1979). There is evidence of large variation in recruitment success between cohorts in Columbia Lake. The 1992 cohort appears to have been much smaller than the adjacent 1991 or 1993 cohorts. Also, the 1996 cohort appears to have been much less abundant than the 1997 cohort (Figure 6), despite the presence of almost as many spawners in 1996 as in 1997 (Figure 8). In fact, since a few burbot are thought to have burrowed under the weir in 1996 and, so, avoided capture (Arndt and Hutchinson 2000), the number of spawners in 1996 may have been as high or higher than in 1997. Presumably, variation in recruitment success is driven largely by the density independent affect of environmental fluctuation on mortality. 200 A 1996 1997 1998 Year 1999 2000 Figure 8. The number of burbot captured in a weir across the unnamed spawning tributary to the southwest corner of Columbia Lake during the spawning season from 1996 to 2000 (Arndt 2000, Arndt and Hutchinson 2000). 20 The abundance and length frequency distribution of burbot spawning in the spring fed tributary entering the southwest corner of Columbia Lake was measured from 1996 to 2000 (Arndt 2000, Arndt and Hutchinson 2000). From 1996 to 1998, the length frequency distribution was unimodal and the average length increased each year (Arndt and Hutchinson 2000). This suggests that the bulk of the spawners captured from 1996 to 1998 came from one, or perhaps two, strong closely spaced cohorts. The age distributions of burbot sampled from the fishery and the spawning population in the winters of 1995/1996 (Figure 7) and 1996/1997 suggest that the 1991 and 1993 cohorts were particularly abundant during this period. Additionally, the length distributions of these two cohorts appears to strongly overlap (Figure 7). In 1999 (Arndt and Hutchinson 2000) and 2000 (Arndt 2000), the length frequency distribution of the burbot spawners became bimodal (Figure 9), and the average length decreased. This appears to reflect the entrance of a strong new cohort into the spawning population. Since early maturing burbot in Columbia Lake spawn after only two growth seasons (Arndt and Hutchinson 2000), this new cohort at the fence is likely the 1997 cohort. To confirm the suspected age of these new spawners, a few fish should be sacrificed at the weir for otoliths in 2001. 1999 2000 LnJ 250 300 350 400 450 500 550 600 650 700 750 800 850 Length (mm) Figure 9. Comparison of length-frequency in 1999 and 2000 for burbot captured at the tributary weir on Columbia Lake. 21 The apparent maintenance of the 1997 cohort's especially high abundance right up to the time of spawning suggests that sampling juvenile abundance might be a useful method for managers to forecast burbot recruitment. In cod, Gadus morhua, also, the idea that year-class strength is largely determined prior to the juvenile stage has been used as an argument for the use of early juvenile abundance estimates as a means to predict future stock size (Sunby et al. 1989). It is curious, however, that the spawner abundance of burbot in Columbia Lake at the fence in 1999 and 2000 did not rebounded to 1996 and 1997 levels despite the addition of the 1997 cohort. The temporal proximity of two strong year classes, 1991 and 1993, could partially explain the relatively high spawner abundance of 1996 and 1997. Alternatively, relative to the 1991 and 1993 cohorts, the 1997 cohort may have been less numerous at settlement, suffered greater juvenile mortality, or matured more slowly. Additionally, yearly differences in the distribution of spawners amongst alternative spawning sites might lessen the accuracy of the spawner counts at the fence as a measure of relative spawner abundance in the lake as a whole. For, instance there is evidence of individual fish spawning in the unnamed tributary at the south end of the lake one year and at the north of the lake in the following year (Arndt and Hutchinson 2000). Consequently, part of the decline in spawner abundance at the fence (Figure 8) may reflect a shift in spawner distribution. The majority of the individuals in the strong 1991 and 1993 cohorts should have become sexually mature after four growth seasons (Arndt and Hutchinson 2000), that is in 1995 and 1997, respectively. Correspondingly, the strongest year class measured as juveniles (Figure 6), and the highest spawner abundance measured at the weir (Figure 8) both occurred in 1997. Also, the 1997 cohort appears to have maintained its high abundance up to the time of spawning (Figure 9). Consideration of this short time series hints at cyclical population dynamics; one strong cohort propagating the next in succession. This, however, does not discount the role of environmental fluctuation in generating and modifying patterns of variation in population abundance. Carl (2000a) conducted an electrofishing survey of age 0 burbot density along randomly chosen stretches of shoreline in ten southern Ontario Lakes. Shoreline abundance average about 100 per km in five lakes without lake herring, Coregonus artedi, 22 and less than one per km in 5 lakes with lake herring. This pattern was taken as evidence of a demographic bottleneck during the larval life stage of burbot due to predation by lake herring. The mean density of age 0 burbot per km of shoreline sampled in Columbia Lake during this study ranged from 400 in 1997 to 62 in 1999. None of the lakes sampled in Carl's study had densities as high as 400 per km. Since choice of sampling sites in Columbia Lake was not randomly chosen, however, mean densities estimated in this study are probably somewhat biased. Even allowing for some degree of error, however, the above comparison suggests that from 1997 to 1999 in Columbia Lake the recruitment of age 0 juvenile burbot to shoreline habitats was not alarmingly low. The reduction in the impact of recruitment failures on population size due to the ability of adults to survive over periods of poor recruitment has been called the "storage effect" (Warner and Chesson 1985). Since, the fecundity of a cohort will peak before a cohort reaches its maximum age (Evans et al. 1987), the benefit from the storage effect depends on what specific age a strong cohort experiences conditions favorable to recruitment. In this study, the oldest burbot sampled were eight years old (Figure 5) and the time span between two years of good recruitment, 1993 and 1997, was 4 years (Figures 6 and 7). Thus, the storage effect has some ability to reduce population size variability in Columbia Lake. 23 Chapter 2: Eggs and Larvae Introduction Despite decades of intensive research and theoretical development there is still much debate, and little consensus, about the causes of mortality during the egg and larval stages of fish. Furthermore, empirical attempts to correlate recruitment variation with environmental fluctuations usually fail to predict post-publication events (Sissenwine 1984). Even for the most intensively studied fish populations, such as those of commercially exploited marine species, the study of eggs and larvae as a means of predicting recruitment variation may be an intractable problem (Pauly 1994). Since recreational freshwater fisheries generally are not studied or managed as intensively as commercial marine fisheries, attempting to predict variation in the recruitment by egg and larval stages for most freshwater fisheries probably is even more hopeless. To be useful, the study of the egg and larval ecology of fishes does not, however, have to allow the prediction of recruitment variation. A more attainable goal is the identification of factors that potentially limit a fish's distribution and abundance. These factors could be physical, such as temperature and turbidity, or biological, such as predators and competitors. Once defined, environments favorable for eggs and larvae of fish can be protected, restored or even created. This study investigates the egg development and early larval life of Columbia Lake burbot under laboratory conditions. The objectives were twofold: 1) to measure the influence of temperature on egg incubation time and survival, and 2) to document the development and behavior of newly hatched larvae. Methods Egg incubation In February, 1998 and 1999, eggs were stripped from burbot captured at the fish weir and trap on the unnamed tributary that emerges from the present spawning springs. Eggs were fertilized using the dry fertilization method (Bagenal and Braum 24 1978). Each female was crossed to a single male. Fertilized eggs were transported to Vancouver, B.C. in plastic thermoses filled with stream water. The elapsed time from fertilization to arrival in Vancouver was 1 d. In Vancouver, the eggs were transferred to 2-L glass jars and incubated in dechlorinated city tap water. The jars were held in water-filled aquaria. The aquaria were cooled with ice in 1998, but a 2°C water bath was installed in 1999. In both years the temperature inside each aquarium was adjusted and maintained with thermostats and heaters. Each incubation jar was aerated with a gentle stream of bubbles. In 1998, eggs were collected from three females (each fertilized with a different male). Subsamples of these eggs were incubated at 2.0, 4.0, 6.0, and 8.0°C. No water changes were made during the incubation period. Dead eggs were picked from the jars on a daily basis. The 1998 experiment was abandoned shortly after the eggs began to hatch because only a few eggs survived to hatching. Thus, for the 1998 eggs, only the relative survivals at different temperatures are presented. In 1999, the eggs were treated differently during incubation. Twice a week, from 30 to 50% of the water in each jar was changed. In addition, the jars were treated with 2 mg/L of malachite green for 1 h approximately twice a week. About 90% of the water was changed after each malachite green treatment. The eggs were mixed once each day by gently stirring the water. After stirring, dead eggs remained suspended longer than live eggs, and were selectively removed from the jars by changing water immediately after stirring. Once hatching started, the malachite green treatments were ended. Eggs from only two females were fertilized in 1999; however, the eggs of one of the females were split into two lots and fertilized separately, but with milt from the same male. The first and second egg batches from this female, and the egg batch from the second female are referred to as batches la, lb, and 2, respectively. These eggs were incubated at 3.0, 4.0, 5.0 and 7.0°C. The volume of eggs per jar was 8.0 mL for egg batches la and lb. Egg batch 2 was smaller and the volume of eggs per jar was reduced to 3.0 or 4.8 mL. Single subsamples from batches la and lb were incubated at each temperature. For batch 2 a subsample was incubated at 3.0 and 7.0°C. In addition, two subsamples of batch 2 were incubated at 5.0°C. No batch 2 eggs were incubated at 4.0°C. Egg volume was measured using wet eggs in a graduated cylinder. 25 Two 1.0 ml samples were taken from each batch and preserved in 2% formaldehyde. These samples were taken at the beginning and end of the jar filling process. Estimates of the mean number of eggs per mL and mean egg diameter were made from these 1.0-mL samples. Once eggs began hatching in 1999, all jars were checked for larvae every 2 days. Larvae were removed from the jars using a glass tube and suction bulb. Egg survival was estimated for each jar as the total number of larvae hatched divided by the estimated starting number of eggs in the jar. Between February 25 and March 1,1999, egg samples were removed from three jars and transferred to a flowing water incubation system. Thus, for these jars, the number of larvae hatching in the jar was an underestimate of survival. Also, survival of eggs was higher in the flowing water system. Hence, including larvae hatched from the flowing water system probably over estimates survival in these three jars. Consequently, for these jars, the two estimates of survival are presented as upper and lower limits. Rearing larvae In 1999, three separate batches of larvae were reared. The first batch contained 910 larvae that hatched over the first 10 d, March 8-17. The second batch contained 4,051 larvae that hatched over the next 10 d, March 18-27. Observations on development after hatching for these two batches use their mean hatching dates as day zero. The mean hatching date was March 13 for the first batch and March 23 for the second batch. A third batch contained 505 larvae all hatched on March 21. This third batch was used to examine developmental changes over the first 10 d after hatching. Length measurements were made on larvae preserved in 2% formaldehyde. All three batches contained larvae from both the static and flowing water incubators. Each batch of larvae was reared in a 10-L aquarium. The tank was lit for 15 h per day. Water circulation was promoted by a gentle stream of bubbles. The original rearing temperature was 4.0°C; however, when a batch reached 7 d the rearing temperature was raised over 4 days to 9.5° C. Larvae were fed Cryptomonas and rotifers for the first 10 d after hatching. After 10 d they were also fed filtered pond water and Paramecium. The filtered pond water contained plankton in the 164 to 475 um size range. The most abundant plankton in the filtered pond water included rotifers, Volvox, 26 copepods, and copepod nauplii . Some copepods that were initially too large to be eaten by the larvae produced nauplii that were eaten by the larvae. Phototaxis experiments A light gradient was set up in a 20 L glass aquarium. Flat black Plexiglas was placed against the inside of the glass on the back and sides of the tank. The bottom of the tank was painted flat black on the outside. A clear sheet of scintered plastic with one half painted black was placed over the top of the tank. The light source was a 25 W full spectrum fluorescent lamp fastened to the inside of a white enamel tray. The tray was placed on the top of the tank over the unpainted half of the plastic sheet. Only a negligible amount of light escaped from around its edges. All other sources of light were eliminated. The light intensity at the surface of the water was 0.80 x 1015 quanta sec1 nv2 in the center of one half of the tank and 24.0 x 1015 quanta sec1 nv2 in the center of the other half of the tank. Light intensity was measured with a full scale irradiance meter (Biospherical Instruments, Inc.). First batch larvae were tested at 13 and 27 d (six trials on both days). Trials were done in the afternoon. Each trial began by removing larvae from the rearing tank and depositing them in the center of the test tank. The test tank was kept at the same temperature as the rearing tank. After 30 min, the number of larvae in each half of the tank was recorded. The light gradient then was reversed and after another 20 min, the number of larvae in each half was again recorded. If the larvae are photo-tactic, they should move from one end of the tank to the other when the light gradient is reversed. The paired t-test was used to test for phototaxis. The critical level used was a = 0.05. The paired observations consisted of the number of larvae in the half of the test tank originally on the positive end of the light gradient compared with the number of larvae at the same end of the tank after the light gradient was reversed. 27 Results Egg size The number of eggs in the six 1.0-ml samples ranged from 541 to 734. The mean (±SE) across the six samples was 603 ± 32 eggs/mL. The mean (± SE) egg diameter was 1.16 ± 0.02 mm (N = 7) for batch la and 1.19 ± 0.02 mm (N = 7) for batch lb. Batches la and lb came from a female with a total length of 573 mm and a mass of 1.2 kg. Survival of eggs to hatching Batch la had less than 1% survival at all temperatures (Table 1). Survival to hatching was higher for eggs from the other two 1999 batches (Table 1). This difference in survival occurred despite the majority of eggs in all three batches being alive when first added to the incubation jars. Thus, some undetermined factor aside from temperature had a large negative effect on the survival of batch la. Therefore, this batch was excluded when considering the effect of temperature on survival. Table I. Percent egg survival from fertilization to hatching for burbot egg batches incubated at 3, 4, and 5°C in 1999 Egg batch 3°C 4°C 5°C la < 1 < 1 <1 lb 48*(71) 28* (35) 7 2 44 Not available 5 and 8 a (14) aThis survival estimate is only the lower limit for a single jar. The upper limit is given in brackets. Over the temperature range of 2-6°C, larvae hatched from all jars. At temperatures of 7°C and 8°C, all eggs died within the first 2 weeks of incubation. The majority of the eggs at 7°C and 8°C died with little outward signs of disease or fungus. In 1999, survival increased with decreasing temperature over the range of 5 to 3°C (Table I). Survival in 1998 was lower than in 1999. Survival in 1998 was greatest at 4°C, intermediate at 2°C, and lowest at 6°C. Survival appears to peak at about 3°C. 28 Time to hatch Time to first hatch was 28 d at 5°C, 32 d at 4°C, and 38 d at 3°C (Figure 10). Mean time to hatch ranged from 39 d in one jar at 5°C to 51 d in one jar at 3°C. The fitted regression function (Figure 11) predicts that the mean time to hatch will increase from 41 to 46 d if the incubation temperature is reduced from 5°C to 3°C (a =0.013, r 2 = 0.91). In this regression analysis, the hatching times of larvae from egg batches la and lb were pooled (same mother and father) and treated as a single observation. X o This study + Anderson (1942) • Bjorn (1940) o Ehrenbaum (1909) A Jager et al . (1981) X McCr immon (1959) X Meshkov (1967) - Muth & Smith (1974) Regress ion line 1/Y = 0.0134 + 0.00474 * X , r2 = 0.80 2 4 6 Incuba t i on t e m p e r a t u r e ( ° C ) Figure 10. The influence of incubation temperature on minimum time for burbot eggs to hatch. 29 50 -. 48 H 38 -36 H 1 1 1 1 1 1 2.5 3 3.5 4 4.5 5 5.5 Incubation temperature (°C) Figure 11. Mean time to hatch, fit to a regression line with 95 percent confidence intervals, for Columbia Lake burbot eggs incubated at 3, 4, and 5°C. Early larval development Newly hatched larvae lack a mouth and a swimbladder. Larvae that hatched on March 21,1999, had a total length of 3.47 ± 0.09 mm (mean ± SE; N = 12). Five days after hatching, larvae still lacked a swimbladder or an open mouth, but internal jaw development was visible (Figure 12). After 10 d, the majority of the larvae had an open mouth and a swimbladder. The mean (± SE) length of 13-d old larvae was 4.15 ± 0.04 mm (N = 12). The yolk began to be absorbed before the oil globule. Among 13-d old larvae, the yolk was slightly smaller than the oil globule, and swimbladder size varied greatly. Less than half of the 13-d old larvae were neutrally buoyant. First feeding occurred between the third and fourth week after hatching. By that time, the diameter of the yolk sac was less than half the diameter of the oil globule. After 27 d all the larvae were feeding and neutrally buoyant; however, a bit of yolk and most of the oil globule were still present (Figure 12). In some cases the oil globule was breaking up. The mean (± SE) length of larvae after 27 d was 4.25 ± 0.04 mm (N = 10). Approximately 75% of the larvae in the first batch died within the first four weeks. 30 Figure 12. A photograph of two Columbia Lake burbot hatched and reared in the lab: one five days old (top), and the other 27 days old (bottom). As a consequence of preservation in 95 % ethanol for 18 months prior to taking this photograph, some shrinkage of the oil globule and yolk occurred. The swimming behavior of the larvae gradually changed over the first four weeks. Larvae without swimbladders were oriented head-up (vertically) in the water. They swam in bursts, the body moving much like mosquito "wrigglers," and when they stopped wriggling the larvae sank head first. At any time, about a quarter of the larvae were resting motionless on the bottom but would move if disturbed. As larvae filled their swimbladders and approached neutral buoyancy they sank less rapidly between bursts of swimming and their orientation in the water shifted toward horizontal. Larvae 31 that were neutrally buoyant after 13 d held position at roughly a 45 0 angle to the surface, and after 27 d all larvae were horizontal. Phototaxis The mean (± SE) difference between the number of larvae on the original illuminated half of the tank before and after the light gradient was switched was 11.67 ± 0.33 after 13 d and 13.00 + 0.37 after 27 d. Both measures are significantly different from zero (a < 5 x 10"7 for both) and strongly imply positive phototaxis. Even before 13 d, when a flashlight was shone against the tank, the larvae would move toward the light. Discussion Jaeger et al. (1981) estimated that survival to hatching of European burbot eggs peaked at about 4°C and that embryonic development did not occur outside the range of 1 to 7°C. My data closely agree with this estimate: peak survival was at about 3°C with no survival above 6°C. Kainz and Gollman (1996), however, found that although in the first week of incubation burbot eggs die at temperatures over 6°C, temperatures as high as 9°C can be tolerated after the third week of development. Thus, although all our eggs died within a week or two at temperatures above 6°C, higher temperatures may be tolerated later in incubation. Curiously, burbot are often reported to spawn at temperatures outside the apparent range of peak egg survival (3-4°C). Typically, water temperatures measured at the bottom of lakes, when burbot are spawning, range from about 1 to 4°C (Fabricius 1954; Hewson 1955; Lawler 1963). Burbot in the Tanana River, Alaska, are thought to spawn at near 0°C (Chen 1969). The average daily water temperature at the fish weir where eggs were collected for this study ranged from 4°C to 6°C at the time of spawning from 1996 through 1998 (Arndt and Hutchinson 2000). Temperatures measured in the springs where the burbot congregate were about 5°C at the time of spawning in 1998 (S. Arndt, Columbia Basin Fish and Wildlife Compensation Program, personal communication). Water temperature in the alluvial fan of Dutch Creek was about 1°C during the spawning season of 2000. The water temperature at the bottom of Columbia 32 Lake at the time of spawning is not known but is probably closer to 3 to 4 °C, the optimal temperature for burbot egg survival, than in these two tributaries to Columbia Lake. Thus, the observation that some Columbia Lake burbot spawn in tributaries, despite presumably better temperature for egg survival in the lake, suggests that flowing water may have a positive effect on egg survival. Burbot spawning in the unnamed tributary to the southwest corner of Columbia Lake are spawning at temperatures near the maximum for egg survival. Consequently, for this population, the water temperature during the first two weeks of egg incubation may be critical. If the temperature in the spawning springs is 3 or 4°C zygote survival may be high. In contrast, if the temperature is 5 or 6°C zygote survival may be poor. The average daily water temperature at the fish weir where eggs and spawners are sampled was no higher in 1996 than in 1997 (Arndt and Hutchinson 2000). Therefore, the affect of temperature on egg survival does not appear to explain the apparent low survival to the juvenile stage for the 1996 cohort relative to that of the 1997 cohort (Chapter 1). The egg survival measured in this study does not take predation into account. Predation by planarians, Turbellaria, for instance, appeared to be a large source of egg mortality during a failed attempt to incubated burbot eggs in containers immersed in the unnamed spawning tributary following the 1998 spawning season (P. Mylecreest, Columbia Basin Fish and Wildlife Compensation Program, personal communication). It could be argued that under completely natural conditions, increased incubation temperature might lower egg predation by causing eggs to hatch earlier. This reduced predation might even compensate for the increased mortality observed at high temperatures in this study. On the other hand, increased temperature would also increase the metabolism and, thereby, the food demand of the predators. The literature on hatching in burbot suggests that there is a slightly curvilinear relationship between minimum time to hatch and temperature over the range of 0.5 to 7°C (Figure 10). Variation about the trend line could be due to random sampling error, variation in phenotype among populations, and differences in methods among studies. Only studies that incubated eggs at a constant temperature throughout incubation were included in Figure 10. A partial exception is the study by Muth and Smith (1974). They induced hatching after 46 d by warming the water "somewhat" above their constant 33 incubation temperature of 1.0 ± 0.5 °C (Figure 10). Thus, their estimate of time to hatch is slightly low. However, their results are included because the 43 d incubation time at 1°C reported by Meshkov (1967) was even lower (Figure 10). Drift net sampling at the spawning site stream from February 27 to April 5,1998, caught seven larvae between March 31 and April 5 (P. Mylechreest, Columbia Basin Fish and Wildlife Compensation Program, unpublished data). They ranged in length from 4.0 to 4.2 mm, were just beginning to form an open mouth and had a small swimbladder. Their yolk was as large as their oil globule and they had not begun to feed. Two of the larvae were kept in a bowl and observed for a few days. They had not achieved neutral buoyancy and still swam with the characteristic wriggling movement observed in the laboratory. They also spent time resting motionless on the bottom of the bowl. Based on their morphology, size, and behavior, these larvae probably hatched about 1-2 weeks before capture. That is about 43-50 d after the peak of the spawning run on February 2, 1998 (Arndt and Hutchinson 2000). This is slightly longer than the 41 d predicted as the mean hatching date for larvae incubated at 5°C in this study (Figure 11). Larval fish are often born without mouths (Blaxter 1969). Burbot in our study formed a mouth between 5 and 10 d after hatching. This is similar to the data of Muth and Smith (1974) who reported that larvae form a mouth between 5 and 12 d after hatching. In contrast, Meshkov (1967) observed that burbot larvae had a mouth when they hatched. Size at first feeding appears to vary in burbot. In a field study, Ghan and Sprules (1993) found that burbot started feeding at a length of 3.2 mm. Burbot in this study hatched at a length of about 3.5 mm and did not start feeding until 3-4 weeks after hatching. By this time they were almost 4.3 mm in length. Also, the seven larvae (4.0-4.2 mm in length) caught in the Columbia Lake spawning stream had not begun feeding. This suggests that even under natural conditions, Columbia Lake burbot do not feed at sizes of less than 4.3 mm. In another laboratory study, burbot larvae did not start feeding until they reached a length of 4.4 mm, despite hatching at a length of about 3.8 mm (Kainz and Gollman 1996; Steiner et al. 1996). Comparison of these three studies suggests that larger size at first feeding is associated with larger size at hatching. Depending on species, larvae that use up their yolk reserves can still retain the 34 potential to feed for some days (Blaxter 1969). Indeed, yolk may be present even after the yolk sac is gone because many fish, including cod, transport and store yolk in subdermal spaces and mesentaries (Blaxter 1969). If larvae do not begin feeding by a certain period they become too weak to feed and are said to have reached the "point of no return" (Blaxter and Hempel 1963). Based on the timing of yolk sac absorption, burbot in this study probably would not have reached a point of no return until more than 4 weeks after hatching, and a length of over 4.3 mm. Muth and Smith (1974) found that burbot larvae that hatched at an average length of 3.9 mm had absorbed their yolk sac when 4.6 mm in length. Since first feeding in this study occurred between the third and fourth week after hatching, the critical period over which Columbia Lake larval burbot must find food or starve may last a week or more. Positive phototaxis has been observed before in larval burbot from eastern North America (P.J. Colby, as cited by Ryder and Pesendorfer 1992) and Europe (Girsa 1972) but never directly in newly hatched burbot. The positive phototaxis and associated wriggling behavior of newly hatched larvae in this study may cause them to swim up from the substrate. Swimming up from the substrate may reduce the potential for suffocation within fine substrate, and aid in dispersal downstream to Columbia Lake. Because the larvae hatched with a large yolk sac and without a mouth, immediate exogenous feeding was not possible. No data are available on the time it takes the larvae to reach the lake, but they probably do not begin feeding before they reach the relatively plankton rich lake. The inferences made in this study are based on observations made from the offspring of only a few mated pairs: three in 1998 and two in 1999. Unfortunately, the number of ripe spawners that passed through the weir in a single night limited this sample size. Obviously, a larger sample size would have given a better estimate of the variability between individuals and improved accuracy. Close agreement between the findings of previous studies on relative survival of eggs at different temperature conducted in Europe (Jaeger et al. 1981, Kainz and Gollman 1996) and the findings of this study (Table I), however, suggests that this studies' conclusions on relative survival of eggs at different temperatures are broadly applicable. 35 Chapter 3: Juvenile Habitat Use Introduction While most large scale-variations in fish recruitment success probably result from environmental fluctuations acting on the egg and larval stages, there is also the potential for further small-scale control of recruitment during the juvenile life stage (Sissenwine 1984, Houde 1987, de Lafontaine et al. 1992, Walters and Juane 1993). One group of fish in which there is support for the importance of juvenile stages in determining recruitment success is the Gadids (Houde 1987, Mehl 1989, Sunby et al. 1989, Bogstad et al. 1994, Tupper and Boutilier 1995a, Tupper and Boutilier 1995b, Hussy et al. 1997). Juvenile mortality is probably due mainly to predation (Sissenwine 1984, Houde 1987, Tupper and Boutilier 1995a) and, thus, should be affected largely by two factors: predator abundance, and availability of cover from predators. Considerable predation on juvenile burbot in Columbia Lake is likely. The decline in shoreline abundance of juvenile burbot in Columbia Lake between age 0 and age 1 (Figure 6) probably resulted from both mortality and movement to offshore habitats. Cannibalism has been observed in both burbot (Chen 1969), and cod (Mehl 1989, Bogstad et al. 1994, Scott and Brown 1998). Other potential fish predators of juvenile burbot in Columbia Lake include torrent sculpins, pike minnows, and trout. Scratches, puncture wounds, and scars were common on the juvenile burbot sampled in this study. Columbia Lake's diverse and abundant bird community probably inflicted most of these injuries. Ospreys occasionally take even adult Columbia Lake burbot that have ventured too close to the surface. In lakes where juvenile burbot habitat is inadequate, the addition of cover may reduce juvenile mortality, and thus, increase recruitment. Before management agencies start building cover for juvenile burbot, even on an experimental basis, habitat use by juvenile burbot needs to be understood. This chapter tests the usefulness of several cover-related habitat characteristics as predictors of shoreline juvenile burbot abundance. The sediment size in shoreline habitats used by torrent sculpins and juvenile burbot is also compared. 36 M e t h o d s Habitat use survey A total of 153 transects were sampled during the day from May 29 to August 13 of 1998. Transects sampling dates were chosen opportunistically but locations were chosen using a stratified random sampling procedure. The lake was divided perpendicular to its north-south axis into four equal sized sections. These sections were sampled in a cycle, starting at the north end, progressing south through each section, and then starting back at the north end again. Three transects were sampled each time a section was visited. One of the three transects occurred at each of three lake depths: 0.17, 0.5 and 0.83 m. The order and location of the three transects within a section were chosen randomly. Transects were sampled with a DC backpack electrofisher. A 20 m string with a weight at each end was used to measure a 5 m approach distance followed by 15 m of sampled transect. The string was laid out behind the people sampling during the first pass on each transect. The string was left in the water after the first pass, allowing the same area to be sampled during a second pass 10 min later. The abundance of burbot on a transect was measured as the combined catch from the two passes. Fish electrofished more than 0.3 m from the longitudinal axis of the transect were not counted. The size frequency distribution of age 0 burbot did not begin to overlap with that of older burbot until the end of August (Figure 5). Therefore, length was used to categorize the burbot sampled as either age 0 or older. Initially, electiofishing was done with a 10 cm diameter anode, a pulse frequency of 120 Hz, and a current of about 0.6 A. The concentrated, rapidly pulsing, electric field produced by this set up was effective for sampling even tiny larval fish; some as small as 1 cm long. As the age 0 burbot grew, however, current and pulse frequency were decreased, and a 20 cm anode replaced the 10 cm anode. By the end of the survey we were using a pulse frequency of 75 Hz, and a current of about 0.4 A. The electrofishing was always performed by the same person, wearing polarized sunglasses, and only when light and wind conditions allowed adequate visibility. 37 Four cover related habitat characteristics were estimated visually at each site: bottom cover (BC), water surface cover (SC), the presence of undercut banks (UC), and sediment size distribution. Bottom and surface cover were both measured as the percent area covered by vegetation and debris. Four substrate size categories were used: fines, gravel, cobble, and boulders (Table II). Sediment size was compared across sites using a sediment size index (SED) calculated as the sum of the products of rank (Table II) and proportion of transect surface area covered for each substrate size category. SED = [(33 x proportion of transect surface area covered by gravel) -(- (67 x proportion of transect surface area covered by cobbles) + (100 x proportion of transect surface area covered by boulders)] SED can range in value from zero when the substrate is entirely fines, to 100 when the sediment is entirely boulders. Table II. The size range and rank of the substrate categories measured in the habitat use survey Category Size Range Rank Fines < = 2 mm 0 Gravel > 2 mm to 64 mm 33 Cobble > 64 mm to 256 mm 67 Boulder > 256 mm 100 The utility of SED as a measure of habitat quality is demonstrated by the comparison of two hypothetical sample sites: one where the sediment is entirely gravel and another where the sediment is 60 percent gravel, and 20 percent each of fines and cobbles. These two sites would have a similar average substrate particle size and, appropriately, share a SED value of 33. Furthermore, on the second site the cobble substrate would probably be somewhat compacted due to the higher incidence of fine substrate. Consequently, the size distribution of the interstitial spaces in the substrate should also be relatively similar on these two sites. 38 Modeling habitat use The fish abundances measured on transects in this study are count data. Count data generally have several characteristics — non-negativity, non-normality, and heterodasticity — that make them unsuitable for analysis by ordinary least squares methods. A number of other statistical methods have been identified as being generally more appropriate for modeling count data (McCullagh and Nelder 1989, Cameron and Trivedi 1998). Of these, the simplest and most commonly used method is Poisson regression. Consequently, the method chosen to model habitat use from the transect data collected this study was Poisson regression. Poisson regression parameters were estimated by the iterative process of maximum likelihood estimation. The maximum likelihood calculations were performed using the nonlinear fitting platform of IMP™ statistical software. The Poisson model assumes that the variance equals the mean; a condition known as equidispersion. In reality, however, data rarely show equidispersion. Fortunately, equidispersion is not required for the Poisson maximum likelihood estimator to obtain accurate estimates of the Poisson regression coefficients (Cameron and Trivedi 1998). The accuracy of standard errors and likelihood ratio statistics obtained from the Poisson maximum likelihood estimator, however, is affected by violation of the equidispersion assumption (Cameron and Trivedi 1998). The simplest method of accommodating the relaxation of the Poisson assumption of equidispersion is called Poisson pseudo-maximum likelihood estimation and involves using the standard output from the Poisson maximum likelihood estimator to obtain regression coefficients but re-scaling the output to obtain standard error and log-likelihood statistics (Cameron and Trivedi 1998). How the re-scaling is done depends on what alternative assumptions are made about the relationship between the variance and the mean. In this study it was assumed that the variance simply equals a multiple of the mean; this is also the generalized linear model approach to Poisson regression (McCullagh and Nelder 1989). This multiple, called the dispersion parameter (())), was estimated using the standard formula (McCullagh and Nelder 1989, Cameron and Trivedi 1998). Correction of the standard error estimates and likelihood ratio statistics involved multiplication by the square root of <[> and the division by <\>, respectively. The 39 significance of the likelihood ratio statistics were judged at a critical level of a - 0.05 after this re-scaling occurred Substantive significance, or goodness of fit, was estimated using a pseudo R 2 measure that was calculated as the fraction of the potential log-likelihood gain that was achieved with inclusion of regressors (Cameron and Trivedi 1998). In ordinary least squares multiple regression the multiple semipartial r 2 (r2sp) measures the proportion of the variation in the dependent variable that is explained by a subset of predictor variables while partialing out the effect all other predictor variables. The r 2 s p for a subset of variables is equal to the reduction in R 2 when that subset is removed from a model. In this study a pseudo r 2 s p measure was analogously calculated as the reduction in the pseudo R 2 measure with the removal of a subset of variables. Violation of the equidispersion assumption does not affect the calculation of these measures of substantive significance (Cameron and Trivedi 1998). Poisson regression was first used to predict abundance versus sediment size for three types of fish: age 0 burbot, burbot older than age zero, and torrent sculpins. For each type of fish a cubic polynomial function of SED was initially assumed. Then, backward stepwise elimination was used to select a reduced model. Lower degree terms were not removed if they were part of statistically significant higher degree terms. Additionally, two more detailed models of burbot habitat use were built: one for each age category of burbot. Initially the same full model was used for both age categories. It included 11 variables: an intercept, the four cover related variables, sampling day (Time), water depth (Depth), two second degree polynomial terms (SED2, Time2), and two interaction terms (Time x BC, Time x Depth). Again, backward step-wise elimination was used to select two separate reduced models, one for each age category. Degree of collinearity amongst the predictor variables was assessed for both full models using the eigenvalues of the predictor variable correlation matrix as described in Kleinbaum et al. (1988). This examination of eigenvalues suggested that the levels of collinearity were low and should not be problematic. Data points sharing the same x and y co-ordinates were very common in the juvenile habitat use data. Standard two dimensional graphical representation of this data, therefore, would have hidden much of the information. For this reason, data points in 40 figures describing juvenile habitat use data have been "jittered". Jittering is the addition of a small random amount of horizontal and vertical variation to each data point. The random amount added to each data point ranged from positive one to negative one percent of the length of the figure axis. Results The first age 0 burbot was caught on June 16 and, thus, the age 0 habitat use model was built using only the 129 transects sampled after June 15. Of the three types of fish surveyed, torrent sculpins were most abundant, followed by burbot older than age 0, and then age 0 burbot (Table III). Torrent sculpins and age 0 burbot are similar in length but burbot older than age 0 were much longer (Table III). Table III. Summary of the size and abundance of age 0 burbot, burbot older than age 0, and torrent sculpins sampled in the habitat use transect survey. Fish type Number Total length Total Per transect Range Mean Age 0 burbot 43 0.33* 39 to 96 mm 66.2 mm Age 1 burbot 68 0.44 116 to 327 mm 173.0 mm Torrent sculpins 180 1.18 5.5 to 9.5 cm Unknown * For age 0 burbot only, the number of transects sampled was 129 rather than 153 41 Data collected on the independent variables for all 153 transects are summarized in Table IV. The sediment size distribution was trimodal (Figure 13) with a mean SED value of 18.7 (Table IV). One half (50%) of the transects had SED values of 10 or less (Figure 13), indicating that their dominant substrate was fines (Table II). The next largest peak in transect frequency (33% of the transects) occurred largely over the SED range of 26 to 45 (Figure 13). These transects with intermediate values of SED occurred mostly along the western edge of the lake in areas where rocks were added during railway construction (Figure 1). The third mode, a clump of three outliers, occurred within the SED range of 75.1 to 85 (Figure 13). These three sites with the largest values of SED occurred in the southeast corner where the lake is bordered by rock bluffs. Table IV. Definitions and summary of the observed values for the first degree terms considered as factors potentially affecting burbot abundance (n = 153). Variable Definition Range Mean SD SED An index of sediment size (see page 38) 0 to 83.3 18.7 20.1 BC Percent of bottom (substrate surface) area covered by vegetation or sunken debris 0 to 100 % 27.2 % 38.1 % SC Percent of water surface area covered by vegetation or floating debris 0 to 100 % 5.3 % 16.9 % UC Presence of undercut banks 0 or 100 3.3 17.8 Time Day of the transect survey 1 to 77 d 40.4 d 23.3 d Depth Water depth at the transect 0.17 to 0.83 cm 50.0 cm 27.0 cm 42 2.5 7.5 12.5 17.5 22.5 27.5 32.5 37.5 42.5 47.5 52.5 57.5 62.5 67.5 72.5 77.5 82.5 Sediment size index (SED) Figure 13. Frequency distribution of sediment size index (SED) values for 153 transects sampled in shoreline habitats of Columbia Lake during the summer of 1998. Note that SED can range from zero when a transect's substrate is entirely fines to 100 when a transect's substrate is entirely boulders (Table II). Burbot, torrent sculpins, and sediment size The relationship between age 0 burbot abundance and SED was best fit by a cubic function with highest abundance on transects wi th intermediate values of SED and a second smaller peak on transects with the smallest values of SED (Figure 14). The relationship between the abundance of burbot older than age 0 and SED was best fit by a quadratic function wi th peak abundance occurring on transects wi th the largest values of SED (Figure 15). Torrent sculpin abundance was also best predicted by a quadratic function of SED and appears to peak on transects with intermediate values of SED (Figure 16). 43 CD Q . O XI — JO 5.5 5 4.5 1 3.5 CO CD .O E 2.5 1.5 0.5 0 -0.5 O c u b i c func t ion of S E D deta i l m o d e l of habi ta t u s e o o b s e r v e d o ° Q o S e d i m e n t s i z e i ndex ( S E D ) 70 80 9P Figure 14. Observed and predicted shoreline abundance of age 0 burbot versus a sediment size index (SED) for Columbia Lake in 1998. Note that SED can range from zero when a transect's substrate is entirely fines to 100 when a transect's substrate is entirely boulders (Table II). Predictions were made using both a cubic function of SED [Y = exp (-1.83 -0.39 x SED + 0.0223 x SED2 - 0.00028 x SED3), P < 0.0001, R2 = 0.24], and a more detailed model of habitat use (Table V). Predictions from the more detailed model assume that Depth = 50 cm, Time = day 46 (July 13), and bottom cover (BC) = 0 %. 44 22 20 18 16 A o 14 c CO r 12 CD Q . •1 10 2 8 CD § 6 4 2 0 -10 quadratic function of S E D detailed model of habitat use o observed 60 70 80 9b Sediment size index (SED) Figure 15. Observed and predicted shoreline abundance of burbot older than age 0 versus a sediment size index (SED) for Columbia Lake in 1998. Note that SED can range from zero when a transect's substrate is entirely fines to 100 when a transect's substrate is entirely boulders (Table II). Predictions were made using both a quadratic function of SED [Y = exp (-2.12 - 0.015 x SED + 0.00092 x SED2), P < 0.0001, R 2 = 0.66], and a more detailed model of habitat use (Table VII). Predictions from the more detailed model assume that there are no undercut banks (UC) and that surface cover (SC) = 0%. 45 22 20 A 18 16 A o 14 CO 12 CO §. 10 8 ^ — quad ra t i c func t ion of S E D o o b s e r v e d o o o o ' O r— 50 60 Sed imen t s i ze index ( S E D ) Figure 16. Observed and predicted shoreline abundance of torrent sculpins versus a sediment size index (SED) for Columbia Lake in 1998. Note that SED can range from zero when a transect's substrate is entirely fines to 100 when a transect's substrate is entirely boulders (Table II). Predictions were made using a polynomial function of SED [Y = exp (- 2.12 - 0.015 x SED + 0.00092 x SED2), P < 0.0001, R2 = 0.48]. 46 Detailed models of burbot habitat use For the detailed model of age 0 burbot habitat use, all variables from the full model except two, SC and UC, were found to have a significant relationship with the abundance of age 0 burbot (Table V). In the absence of BC, the detailed model of habitat use for age 0 burbot predicts that abundance peaks at intermediate values of SED (Figure 14). In the detailed model, the five sites with above zero abundance of burbot at SED = 0 (Figure 14) are explained by the positive effect of BC (Figure 17). Initially, age zero burbot abundance increased with both BC and Depth (Figures 17 and 18). However, this positive association of abundance with BC and Depth did not persist over time (Figures 17 and 18). Age 0 burbot abundance peaked at about the end of the first week in July (Figure 18). Table V. Summary of the variables included in the detailed model of habitat use for age 0 burbot (n = 129, R 2 = 0.45) Variable3 Coefficient Approx. SE Significance Intercept -14.77129 3.14118 SED 0.24979 0.07545 <0.0001 SED 2 -0.00299 0.00106 <0.0001 BC 0.08681 0.02390 <0.0001 Time 0.26985 0.08792 0.0005 Time2 -0.00185 0.00074 0.0075 Depth 0.08244 0.02340 <0.0001 Time x BC -0.00114 0.00041 0.0010 Time x Depth -0.00131 0.00048 0.0015 a Described in Table IV. 47 4 H 0.5 H predicted when BC = 100 % and Depth = 50 cm predicted when BC = 75 % and Depth = 50 cm o observed when BC >= 75 % ( N = 26) A observed when BC < 75 % (N = 27) " " T ^ A ^ °*> °AA ^P, ftfr> i A f f , rs-din-98 28-Jun-98 12-Jul-98 26-Jul-98 9-Aug-98 Date Figure 17. Shoreline abundance of age 0 burbot versus bottom cover (BC) and Time when the sediment size index (SED) = 0 for Columbia Lake in 1998, as observed, and as predicted by a detailed model of habitat use (Table V). Note that the substrate on a site with SED = 0 is entirely fines (Table II). 48 5.5 4 .5 3.5 o _ 2 .5 1-5 4 0.5 A -0 .5 2 o 83 cm predicted — 17 cm predicted o 83 cm observed x 50 cm observed A 17 cm observed -*—*>•» 14-Uun 2 8 - J u n 12-Ju l 26 -Ju l 9 -Aug Date Figure 18. Shoreline abundance of age 0 burbot versus Depth and Time for Columbia Lake in 1998, as observed, and as predicted by a detailed model of habitat use (Table V). The predictions assume that the sediment size index (SED) and bottom cover (BC) equal their average values for the 129 transects sampled, 18.4 and 26.1 %, respectively. 49 Table VI. Semipartial r 2 values for variables included in the detailed model of age 0 burbot habitat use. Group r 2 s p Subgroup r 2 s p Variables3 included Cover 0.306 SED 0.248 SED, SED 2 BC 0.157 BC, Time x BC Other 0.259 Time 0.202 Time, Time2, Time x BC, Time x Depth Depth 0.156 Depth, Time x Depth a Described in Table IV. The r 2 s p value for the group of cover related variables in the age 0 burbot habitat use model was 0.306, while that of the non-cover related variables was 0.259 (Table VI). Of the four subgroups of variables in the age 0 burbot habitat use model, SED had the highest r 2 s p value (Table VI). In fact, due to the sharing of interaction terms between the other three subgroups of variables, their r 2 s p values are inflated relative to that of the SED subgroup (Table VI). Table VII. Summary of the variables included in the detailed model of habitat use for burbot older than age 0 (n = 153, R 2 = 0.71) Variable3 Coefficient Approx. SE Significance Semipartial r 2 Intercept -4.6094 0.7840 SED 0.0888 0.0114 <0.0001 0.703 SC 0.0475 0.0133 0.0096 0.066 UC 0.0347 0.0112 0.0218 0.051 a Described in Table IV. For the detailed model of habitat use for burbot older than age 0, only three variables from the full model were kept: SED, SC, and UC (Table VII). The near equality of the semipartial r 2 value of SED (0.703) and the overall R 2 value (0.71) indicates that most of the explained variation in the abundance of burbot older than age 0 can be attributed to SED. Mean abundance increases with SED (Figure 15). However, there are 50 six non zero observations at about SED = 0 that are not well predicted by sediment size alone (Figure 15). Two of these six observations, one count of one and one count of three, came from sites where burbot were hiding beneath undercut banks. Three others, two counts of one and one count of two, are associated with SC values of 50 percent or greater. The SC used was floating mats of the dead stems of Scirpus spp (bulrush). The remaining count of two came from a transect where two burbot older than age 0 were beneath a flat piece of log that covered only 5 % of the transect's bottom area. The highest observed value of SED was 83.3 (Table IV). For the detailed model of habitat use, the predicted mean abundance of burbot older than age 0 when SED = 83.3, SC - 0%, and UC = 0 is 15 per transect (Figure 15). The highest observed abundance was 20 on a transect where SED = 82.6, SC = 0 %, and UC = 0 (Figure 15). Discussion Meshkov (1967) observed that major changes delineating the transformation from larva to juvenile burbot, in feeding, in habitat location, and in external and internal structures, occur over the 23 to 39 mm length range. Girsa (1972) reported that the shift from positive to negative phototaxis occurs over the same size range. A gradual shift in diet from predominantly pelagic prey to predominantly benthic prey was observed in burbot in the length range of 21 to 30 mm (Ryder and Pesendorfer 1992), and 25 to 50 mm (Hartman 1983). The shift in depth distribution during this transition has not been described adequately but may also be gradual. The smallest burbot caught from shoreline habitats in this study were juveniles about 4 cm in length. They were already strongly associated with the bottom and microhabitats providing cover. This suggests that selection of settlement habitats with cover occurs during the transition period. In addition to the electrofishing described in this study, visual observation, pole seining, and dip netting were tried but proved unsuccessful as methods for sampling burbot smaller than 4 cm in Columbia Lake's shoreline habitats. Thus, pelagic larval burbot do not seem to congregate in shoreline habitats before settlement in Columbia Lake. Furthermore, the highest initial abundance of age 0 burbot was on the deepest transects (Figure 18). After examining growth sectors in otoliths of larval and juvenile burbot from Europe Fischer (1999) speculated that just 51 before the juvenile stage, larval burbot move to the profundal zone and migrate to shore. Similarly, in Columbia Lake, it seems probable that late in the transition from larva to juvenile burbot assume an epi-benthic lifestyle while searching out cover providing settlement habitat. It is unclear, however, if the increase in abundance of age 0 burbot on shallow transects over the course of the survey was due entirely to the movement of fish. The water level of Columbia Lake during 1998 peaked in late May. From June 16 to Aug 13, the period when sampling for the age 0 burbot habitat use model was performed, the water level dropped by about 40 cm. This amounts to an average rate of about 0.68 cm per day. Therefore, a receding shoreline may also have contributed to the eventual concentration of burbot in the shallowest transects (Figure 18). Rocky substrate has been suggested as an important component in the habitat of juvenile burbot (Robins and Deubler 1955, Lawler 1963, Boag 1989, Ryder and Pesendorfer 1992, Fischer and Eckman 1997). Since there was considerable variation in sediment size between transects sampled in Columbia Lake (Figure 13), the regression models with only polynomial terms of sediment size as predictor variables were satisfactory for the simple prediction of juvenile burbot abundance (Figures 14 and 15). Floating algal mats, sunken ligneous debris, and vegetation, however, are also thought to be components in the habitat of juvenile burbot (Robins and Deubler 1955, Meshkov 1967, Hanson and Quadri 1980, Boag 1989, Ryder and Pesendorfer 1992). The detailed models of juvenile burbot habitat use (Tables V and VII) support this notion that other habitat characteristics besides sediment size — variables such as vegetation, floating and sunken debris, and undercut banks — are important. Thus, compared to the models with sediment size as the only predictor variable, the detailed models provides not only a better overall fit to the data (higher R2) but also a better functional understanding of the factors affecting juvenile burbot abundance. Vegetation, floating and sunken debris, and undercut banks are abundant in Columbia Lake on shores with fine substrate but largely absent on shores with larger substrate. Despite the abundance of these potential sources of cover on shores with fine substrate, sites with intermediate sized substrate (predominantly gravel and cobble) had the most age 0 burbot (Figure 14) and sites with the largest sized substrate 52 (predominantly cobble and boulders) had the most burbot older than age 0 (Figure 15). This implies that the interstitial spaces in the substrate are the preferred habitat of juvenile burbot. Measuring cover in units of percent area covered fails to recognize important qualitative habitat characteristics. For instance, the vegetation occasionally used as cover by age 0 burbot in Columbia Lake is not the bulrush (Scirpus spp) which dorninates the shoreline; but rather, more extensively branching vegetation such as bushy pondweed, Najasflexis. The degree of aggregation of cover on a site also seems important. For instance, a fixed volume of sunken debris provides better cover, especially for larger juveniles, if it is concentrated in one spot rather than scattered. Perhaps the most important qualitative consideration for juvenile burbot in Columbia Lake is the size of the crevices or physical 'niches' provided by habitat features. This is demonstrated by older (larger) juveniles occupying larger substrate (Figures 14 and 15). Vegetation provides mostly small crevices. Therefore, it is not surprising that only age 0 burbot, and especially the small newly settled ones, were found hiding in vegetation (Figure 17). The spaces beneath undercut banks and floating mats of dead bulrush stems roughly matched the body depth of the older juveniles. Accordingly, age 0 burbot are less likely than older juveniles to be found living beneath an undercut bank, or a floating mat of dead bulrush stems (Tables V and VII). When juvenile cod were confronted with predators in an experimental setting they selected coarse substrate and kelp in which they could successfully hide from predators (Gotceitas and Brown 1993, Gotceitas et al. 1995). Furthermore, Tupper and Boutilier (1995a) demonstrated that post settlement survival of cod was highest in structurally complex habitats due to the increased shelter availability and decreased predator efficiency in these structurally complex habitats. Presumably, the smaller the crevice chosen as cover the fewer potential predators that can invade the crevice. Therefore, the scaling of shelter size with juvenile burbot size is likely an adaptation for predator avoidance. Since age 0 burbot hide in small crevices and small crevices tend to be much more abundant than large ones (Caddy and Stamatopoulos 1990), it is not surprising that the age 0 burbot sampled in this study appeared far from samrating the available 53 cover. Furthermore, due to the large size of the transects used (9 m2) there was some form of seemingly adequate cover for age 0 burbot on just about every transect. This may partly explain why the detailed model of age 0 habitat use had a somewhat low R 2 value of 0.45. (Table V). If smaller transects had been used then the proportion of variation in age 0 burbot abundance that was explained by habitat variables probably would have increased. Unfortunately, if smaller transects had been used, a much larger and, perhaps, impractical number of transects would have been needed to sample the same number of fish. Torrent sculpins and burbot are the dominant benthic fish species along the shoreline of Columbia Lake. They also both share a preference for habitats with rocky substrate over habitats with smaller particle sizes. The sediment particles in the habitats preferred by sculpins (Figure 16), however, appear to be intermediate in size relative to those prefered by the two age groups of juvenile burbot (Figures 14 and 15). Since age 0 burbot are about 4 cm in length at settlement while some sculpins approach 10 cm in length, predation on newly settled juvenile burbot by sculpins could be substantial. Shortly after settlement age 0 burbot grow to the same size as the average sculpin and, although inter-specific lifestyle differences undoubtedly exist at this time (Fischer 2000), some level of inter-specific competition for space and food might still occur. Competition for space would be especially likely if juvenile burbot show the same site attachment and territorial tendencies as those shown by juvenile cod (Tupper and Boutilier 1995b). Intraspecific competition appears able to affect growth in juvenile cod (Tupper and Boutilier 1995b). Decreased growth could increase the time that juveniles remain small and vulnerable to predators. This may lead to density dependent mortality (Houde 1987, Sissenwine 1984). The number of appropriately small crevices for age 0 burbot to hide in probably does not decrease as mean substrate size increases (Caddy and Stamatopoulos 1990). Therefore, the lower density of age 0 burbot and sculpins in sample sites with the largest substrate relative to sites with intermediate sized substrate (Figure 14 and 16) may be partly the consequence of past or present predation from larger juvenile burbot on sites with the largest substrate (Figure 15). Indeed, prey items resembling sculpins can sometimes be felt through the distended abdomens of the larger juvenile burbot 54 sampled from Columbia Lake. Scott and Brown (1998) observed that age 1 cod began feeding on age 0 cod when age 1 cod where approximately three times larger than their prey. This is about the ratio of age 1 burbot length to age 0 burbot length in Columbia Lake (Figure 5). The idea that juvenile Gadid behavior may serve to avoid cannibalism has already been suggested by researchers studying the foraging cycle of juvenile cod in eelgrass habitats (Scott and Brown 1998). These cod shift from feeding predominantly on zooplankton by day at age 0 to feeding predominantly on benthos at night at age 1. For age 0 burbot, a preference for a temperature of 21.2°C has been observed in laboratory experiments (Ferguson 1958). Field observations suggest that older juveniles and adults prefer much cooler water, in the 10-12°C range, and may avoid temperatures above 13°C (Ferguson 1958, Hackney 1973). Thus, during the summer in thermally stratified lakes adult burbot are usually found below the thermocline (Lawler 1963, Sandlund et al. 1985, Carl 1992, McPhail 1997, Carl 2000b). This movement offshore may start as early as the first year of life (Meshkov 1967, Fischer and Eckman 1997). In many lakes, however, habitat structural complexity decreases with depth. Since this complexity provides shelter, offshore movement to cooler water by juvenile burbot may increase their predation risk. Besides increasing mortality, an increased predation risk might also negatively affecting growth rate by increasing physiological stress levels (Fischer 2000) or reducing time spent foraging. During the summer in Columbia Lake, the largest burbot living in shoreline habitats are about 39 cm in length and age 3 (Figure 5). Since Columbia Lake does not develop a thermocline, the summer depth distribution of juvenile burbot in Columbia Lakes should not be affected by a substantial vertical temperature gradient. Accordingly, no time or depth related variables were retained in the reduced model of shoreline abundance for burbot older than age 0 (Table VII). However, the association between the highest densities of juveniles older than age 0 and the transects with the largest substrate (Figures 15), suggests that a shortage of large crevices (Figure 13) might be mediating the offshore movement of larger juvenile burbot in Columbia Lake. Robins and Deubler (1955), suggested that the important feature of the habitat used by an adult burbot is still "the presence of shelter of sufficient size to hide itself completely. " Robins and Deubler were studying adult burbot in lotic populations. 55 These burbot were living in areas of rock slides, large sunken woody debris, and where erosion had honeycombed the bottom with tunnels. This strong association of adult burbot with cover might not be expected to occur when burbot are living deep in a lake away from the natural structural complexity of shoreline habitats. However, the use by adult burbot of tunnels or burrows in fine substrates has been noted in Lake Superior (Boyer et al 1989), Okanagan Lake, British Columbia (B. Shepard, personal communication, as cited by McPhail 1997), and Lake Laberge, Yukon Territory (D. Davidge, Environment Canada, personal communication). Boyer et al. (1989) provide evidence that burbot may excavate these structures. Juvenile burbot also sometimes occupy excavations (Ryder and Pesendorfer 1992). These excavations, however, are usually built beneath more substantial shelter such as a tree or boulder (Ryder and Pesendorfer 1992). It seems that burbot retain a strong tendency to occupy crevices throughout their life. It is difficult to imagine a great risk of predation for adult burbot living deep in a lake. Therefore, the shelters may have other purposes for adults. Perhaps they act to attract prey and make prey easier to capture (Boyer et al. 1989), or to provide shelter from light during the day. The potential for cover availability to limit juvenile burbot survival in Columbia Lake probably depends on how strong competition for cover is between those juveniles that are still quite vulnerable to predators. Although, the maximum length of burbot in shoreline habitats of Columbia Lake during the summer is about 39 cm (Figure 5), cover availability would presumably begin to limit the abundance of juveniles when they were less than this length. Burbot are very poor swimmers (Jones et al. 1974) and, thus, even 39 cm burbot that cannot find adequate cover may be easy prey for piscivorous birds, or even especially large trout, pike minnows, and burbot. From late fall to early spring, when water levels are at their lowest, and much of the complex shoreline habitat of Columbia Lake is above water, the potential for a demographic bottleneck is great. A next step in studying this possible bottleneck would be to measure over-winter survival of juveniles. An impediment to accurately measuring over-winter survival, however, would be the difficulty of sampling juveniles as they move into offshore habitats. Backpack electrofishing is impractical in depths much over a meter. During this 56 study, boat shocking, trapping and diving with dip nets were attempted as methods for sampling juvenile burbot in Columbia Lake. They proved unsuccessful. Rather than directly estimating juvenile over-winter survival, a relative index of juvenile survival might be compared for several years before and after the addition of rocky substrate below the winter low water level. If shallow water cover is limiting, juvenile survival should increase after such a manipulation. Perhaps the decline in shoreline juvenile abundance from age 0 to age 1 could be estimated by backpack electrofishing and compared among cohorts. Adults are easier to sample than offshore juveniles. Adults could be sampled from a weir across the spawning stream, from the ice fishery, or by using set lines or trap nets. Therefore, survival from settlement to adulthood would be another more attainable, if less direct, measure of the proposed bottleneck. An important concern about generalizing the results of this study to other burbot populations is that the relative importance of different habitat variables in predicting juvenile burbot abundance is influenced by the sample variances of these variables. For instance, in this study the r 2 s p value of sediment size was higher than that of any other variable in the detailed models of habitat use for both age categories of burbot (Tables VI and VII). This was due not only to the strong substrate size preferences of juvenile burbot but also the considerable variation in sediment size between sample sites in Columbia Lake (Figure 13). The importance of sediment size in predicting variation in juvenile burbot abundance would be much less within a lake that had only fine substrate. Other forms of cover, such as vegetation or undercut banks, however, might become more important predictors. Obviously, the specific abundances predicted by the habitat use models developed in this study are only applicable in Columbia Lake during the summer of 1998. To make predictions over a broader spatial or temporal scale additional environmental variables probably need to be included in the model. Some additional variables useful in a more widely applicable model might include predator community, climate, or productivity. The general patterns identified in the habitat use models of this study, however, should be broadly applicable. Moreover, the comparison of such general patterns between population could itself yield insights into burbot ecology. Of 57 the general patterns identified in this chapter the most fundamental and broadly applicable are the importance of crevices as cover for juvenile burbot and the scaling shelter size with burbot body size. 58 Conclusions Burbot recruitment, as estimated by shoreline surveys of juvenile abundance and the age distribution of adults in the fishery and spawning population, varies greatly among cohorts in Columbia Lake. This variation is not well explained by variation in spawner numbers. Presumably, some of this variation is driven by environmental fluctuations. The narrow tolerance of burbot eggs during incubation is one possible cause of such density independent mortality. Competition among juvenile burbot for cover, in contrast, is likely to have a density dependent or regulating affect on recruitment. Both temperature related egg mortality and juvenile competition for cover, however, have the ability to limit mean population size. At the beginning of this study two factors were suggested as potentially causing the decline in Columbia Lake burbot: the degradation of the historic Dutch Creek spawning site, and overfishing. The relatively high abundance of age 0 juveniles sampled in this study relative to densities found in other populations (Carl 2000a) suggests that spawning habitat for burbot is adequate and that recruitment overfishing is not occurring. It is possible, however, that conditions were particularly favorable for the recruitment to the settlement life history stage during my study and, therefore, the age 0 abundances measured in this study may be uncharacteristically high. Another possibility is that historically the shoreline abundance of age 0 juveniles in Columbia Lake was much more than was observed in this study. This might explain why, what I assumed to be a strong yearclass in 1997, did not return spawner numbers at the fence in 1999 and 2000 back to 1996 and 1997 levels. Consequently, it is possible that recruitment to settlement in Columbia Lake may be limited by spawning habitat and fishing pressure. Hence, the potential benefits of both the rehabilitation of the historic Dutch Creek spawning site and more strict catch restrictions for Columbia Lake's burbot anglers can not be ruled out. The decline of the burbot population in Columbia Lake might also be explained by reduced survival of burbot after the early settlement life-history stage. There is no reason to suspect that habitat for older juvenile and adult burbot in Columbia Lake has declined from historic levels. In fact, the rocky substrate added to the lake's western shoreline when the railway was built probably increased the survival of juveniles. It is 59 possible, however, that the composition of Columbia Lake's fish community has changed resulting in increases in the abundances of burbot competitors or predators, or reductions in the abundance of burbot prey. Studies of the food web in Columbia Lake might help determine potential competitors, predators and prey of burbot. Evidence of fish community changes might be obtained from old fish inventory reports. The potential for competion between adult burbot and pike minnows is created by three shared characteristics: large size, piscivory, and nocturnal feeding (Scott and Crossman 1973, Chisholm 1975). I suspect, therefore, that a reduction in the burbot population size due to fishing mortality could lead to an increased pike minnow population size. Indeed, pike minnows were the second most abundant fish species captured in a 1992 fish inventory of Columbia Lake (R.L. & L Environmental Services 1993). What is more, pike minnows may have an advantage over burbot in Columbia Lake because of the tolerance of minnows to high temperatures (Nikcevic 2000). An increase in pike minnows could increase predation on juvenile burbot and, because of reduced cover availability during the low water period (late fall to early spring), juvenile burbot in Columbia Lake may be especially vulnerable to pike minnow predation. Thus, competition and predation due to pike minnows might inhibit the rebuilding of the burbot population even if burbot fishing pressure is reduced. I hope, though, that this conjecture does not lead anyone to attempt to control (slaughter ) pike minnows. As mentioned above, further research is needed. Age and size at easily recognizes stages in early development such as hatching, mouth formation, first feeding, and yolk sac absorption tend to vary among burbot populations. In this study, I suggested a positive relationship between size at hatching and size at first feeding for burbot. Ultimately, this relationship may be determined by egg size because size at hatching and time from hatching to yolk sac absorption are both positively related to egg size in fish (Miller et al. 1988, Duarte and Alcaraz, 1989). Pepin et al. (1997) demonstrated that for cod time to death by starvation, as well as larval size at hatching and time to yolk sac absorption, are positively related to egg size. There is, however, also probably an environmental influence on these developmental characteristics. For instance, there is a positive correlation between egg size at hatching and incubation temperature in cod (Pepin et al. 1997) 60 Larger eggs tend to hatch larger larvae which suffer lower mortality (Duarte and Alcaraz 1989). This is, perhaps, partly because larger eggs have more yolk, which translates into a longer time to yolk sac absorption and, consequently, a longer time to make the transition to exogenous feeding. Survival potential of individual offspring, however, is not the only factor affecting the fitness of parents. For instance, if survival after hatching varies greatly between individuals due to a patchy environment then increased egg number may increase parental fitness (Duarte and Alcaraz 1989). Given that the energy available for reproduction remains constant, any substantial increase in egg number must result in a decrease in egg size and this would decrease the survival potential of individual offspring. Thus, much of the variation in early larval development among burbot populations might be explainable in the context of the adaptive significance of egg size and number. The "storage effect" (Warner and Chesson 1977) could reduce the negative impact of periodically poor recruitment success on burbot population size. Furthermore, since the maximum age of burbot has been observed to decrease with decreasing latitude (McPhail 1997) and the "storage effect" is dependent on adults surviving over periods of poor recruitment, the ability of the "storage effect" to stabilize population size may decrease toward the southern distribution limit of burbot. As fishing pressure could further reduce burbot lifespan, southern burbot populations may be especially susceptible to overfishing. 61 Literature Cited Anderson, K. A. 1942. Fiskar och fiske i Norden. Natur och kultur, Stockholm. Arndt, S. 2000. Data summary for the 2000 Upper Columbia burbot study. Report to the Columbia Basin Fish and Wildlife Compensation Program, Nelson, B.C. Arndt, S. K. A., and J. Hutchinson. 2000. Characteristics of burbot spawning in a tributary of Columbia Lake, British Columbia, over a four year period. Pages 48-60 in V. L. Paragamian and D. W. Willis, editors. Burbot biology, management, and ecology. Fisheries Management Section, Special Publication 1, American Fisheries Society, Bethesda. Bagenal, T. B., and E. Braum. 1978. Eggs and early life history. Pages 165-201 in T. Bagenal, editor. Methods for assessment of fish production in fresh waters. Blackwell Scientific Publications, Oxford. Baxter, J. S. 1998. Summary of juvenile burbot sampling in the upper Columbia River Catchment, 1997. Report to the Columbia Basin Fish and Wildlife Compensation Program, Nelson, B.C. Bernatchez, L., and C. C. Wilson. 1998. Comparative phylogeography of Neartic and Paleartic fishes. Molecular Ecology 7: 431-452. Berqquist, B. C. 1991. The extinction and natural recolonization of fish in acidified and limed lakes. Nordic Journal of Freshwater Research 66: 50-62. Bjorn, E. E. 1940. Preliminary observations and experimental study of the ling, Lota maculosa (LeSuer), in Wyoming. Transactions of the American Fisheries Society 69: 192-196. Blaxter, J. H. S. 1969. Development: eggs and larvae. Pages 178-252 in W. S. Hoar and D. J. Randall, editors. Fish physiology, volume 3. Academic Press, New York. Blaxter, J. H. S., and G. Hempel. 1963. The influence of egg size on herring larvae (Clupea harengus L). Journal du Conseil, Council International pour l'Exploration de la Mer 28: 211-240. Boag, T. D. 1989. Growth and fecundity of burbot, Lots lota L. in two Alberta Lakes. Masters Thesis, University of Alberta, 100 p. Bogstad, B., G. R. Lilly, S. Mehl, O. K. Palsson, and G. Stefansson. 1994. Cannibalism and year-class strength in (Gadus morphua) in Arcto-boreal ecosystems (Barents Sea, 62 Iceland, and eastern Newfoundland). ICES Marine Science Symposium 198: 576-599. Boyer, L. F., R. A. Cooper, D. T. Long, and T. M . Askew. 1989. Burbot (Lota lota) biogenic sedimentary structures in Lake Superior. Journal of Great Lakes Research 15(1): 174-185. Bruesewitz, R. E., T. Fratt, F. Copes, and D. W. Cobble. 1989. Age, growth, and population dynamics of burbot in Green Bay and Lake Michigan. 32nd Conference on Great Lakes Research (summary only), p. 34. Caddy, J. F. and C. Stamatopoulos. 1990. Mapping growth and mortality rates of crevice dwelling organisms onto a perforated surface: the relevance of cover to the carrying capacity of natural and artificial habitats. Estuarine, Coastal and Shelf Science 31: 87-106. Cameron, A. C , and P. K. Trivedi. 1998. Regression analysis of count data. Economic Society Monograph No. 30, Cambridge University Press. Carl, L. M. 1992. The response of burbot (Lota lota) to changes in lake trout (Salvelinns namaycush) abundance in Lake Opeongo, Ontario. Hydrobiologia 243/244: 229-235. Carl, L. M. 2000a. Effects of two planktivores on the fish communities in Precambrian Shield Lakes in southern Ontario. Pages 81-89 in V. L. Paragamian and D. W. Willis, editors. Burbot biology, management, and ecology. Fisheries Management Section, Special Publication 1, American Fisheries Society, Bethesda. Carl, L. M. 2000b. Sonic tracking of burbot in Lake Opeongo, Ontario. Pages 102-104 in V. L. Paragamian and D. W. Willis, editors. Burbot biology, management, and ecology. Fisheries Management Section, Special Publication 1, American Fisheries Society, Bethesda. Chen, L. C. 1969. The biology and taxonomy of the burbot, Lota lota leptura, in interior Alaska. University of Alaska, Biological Papers No. 11. Fairbanks. Chisholm, B. A. 1975. An experimental study of nocturnal feeding i n the northern squawfish (Ptychocheilus oregonensis). Honours Thesis, University of British Columbia, 46 p. 63 Cohen, S. J., K. A. Miller, A. F. Hamlet, and W. Avis. 2000. Climate change and resource management in the Columbia River Basin. Water International 25(2): 253-272. de Lafontaine, Y., T. Lambert, G. R. Lilly, W. D. McKone, and R. J. Miller (editors). 1992. Juvenile life stages: the missing link in fisheries research. Canadian Technical Report of Fisheries and Aquatic Sciences 1890,137 p. Duarte, C. M . , and M . Alcaraz. 1989. To produce many small or few large eggs: a size independent reproductive tactic of fish. Oecologia 80: 401-404. Ebenman, B., and L. Persson (editors). 1988. Size structured populations: ecology and evolution. Springer-Verlag, Berlin. Ehrenbaum, E. 1909. Eier und larven von fischen, zweite halite (Eggs and larvae of fishes, second half). Pages 217-413 in Nordisches Plankton, Zoologischer Teil, Erster Band. Verlag von Lipsius und Tischer, Kiel und Leipzig. (Nordisches Plankton reprinted in 1964 by Neudruck A. Asher and Company, Amsterdam). Evens, D. O., B. A. Henderson, N. J. Bax, T. R. Marshall, R. T. Oglesby, and W. J. Christie. 1987. Concepts and methods of community ecology applied to freshwater fisheries management. Canadian Journal of Fisheries and Aquatic Sciences 44: 448-470. Fabricius, E. 1954. Aquarium observations on the spawning behavior of the burbot, Lota vulgaris L. Report of the Institute of Freshwater Research, Drottingholm 35: 51-57. Ferguson, R. G. 1958. The preferred temperature of fish and their midsummer distribution in temperate lakes and streams. Journal of the Fisheries Research Board of Canada 15(4): 607-624. Fischer, P. 1999. Otolith microstructure during the pelagic, settlement, and benthic phases in burbot. Journal of Fish Biology 54:1231-1243. Fischer, P. 2000. Test of competitive interactions for space between two benthic fish species, burbot Lota lota, and stone loach Barbatula barbatula. Environmental biology of fishes 58(4): 439. Fischer, P, and R. Eckmann. 1997. Seasonal changes in fish abundance, biomass and species richness are characteristic for the littoral zone of a large European lake, Lake Constance, Germany. Archives of Hydrobiology. 139(4): 433-448. 64 Fisher, S. J. 2000. Early life history observations of burbot utilizing two Missouri River backwaters. Pages 96 -101 in V. L. Paragamian and D. W. Willis, editors. Burbot biology, management, and ecology. Fisheries Management Section, Special Publication 1, American Fisheries Society, Bethesda. Ghan, D., and W. G. Sprules. 1993. Diet, prey selection, and growth of larval and juvenile burbot Lota lota (L.). Journal of Fish Biology 43: 47-64. Girsa, 1.1.1972. The effect of photoperiod and water temperature on photoreaction in some fish. Journal of Ichthyology 12: 505-510. Gotceitas, V., and J. A. Brown. 1993. Substrate selection by juvenile Atlantic cod (Gadus morh.ua): effects of predation risk. Oecologia 83: 346-351. Gotceitus, V., S. Fraser, and J. A. Brown. 1995. Habitat use by juvenile Atlantic cod (Gadus morh.ua) in the presence of an actively foraging and non-foraging predator. Marine Biology 123: 421-430. Gunn, J. M . 1982. Acidification of lake trout (Salvelinus namaycush) lakes near Sudbury, Ontario. Page 351 in, R, E Johnson, editor. Acid rain/fisheries. Proceedings of the International Symposium on Acid Rain and Fishery Impacts in Northeastern North America. (Summary only) Hackney, P. A. 1973. Ecology of burbot (Lota lota) with special reference to its role in the Lake Opeongo fish community. Doctor's Thesis, University of Toronto. Hanson, J. M. , and S. Quadri. 1980. Morphology and diet of young-of-the-year burbot, Lota lota, in the Ottawa River, Canadian Field-Naturalist 94: 311-314. Hartmann, J. 1983. Two feeding strategies of young fishes. Archives of Hydrobiology 96: 496(4): 496-509. Hewson, L. C. 1955. Age, maturity, spawning, and food of burbot, Lota lota, in Lake Winnipeg. Journal of the Fisheries Research Board of Canada 12: 930-941. Hildebrand, L. 1991. Lower Columbia River fisheries inventory. Report of R. L. & L. Environmental Services to the British Columbia Ministry of the Environment, Lands and Parks, Nelson. Houde, E. D. 1987. Fish early life dynamics and recruitment variability. American Fisheries Society Symposium 2:17-29. 65 Hussy, K, M. A. St. John, and U. Bottcher. 1997. Food resource utilization by juvenile Baltic cod Gadus morhua:a mechanism potentially influencing recruitment success at the juvenile stage? Marine Ecology Progress Series 155:199-208. Hutchinson, J. 1996. Upper Columbia Lake burbot study. Report to the Columbia Basin Fish and Wildlife Compensation Program, Nelson. Jaeger T., W. Nelson, W. Schofer, and F. Shodjai. 1981. Influence of salinity and temperature on early life stages of Coregonns albula, C. lavaretus, R. rutilius, and L. lota. Pages 345- 348 in R. Lasker and K. Sherman, editors. The early life history of fish: recent studies. Rapport Proces-vebaux des Reunions, International Council for the Exploration of the Sea, Copenhagen. Jones, D. R., J. W. Kiceniuk, and O. S. Bamford. 1974. Evaluation of the swimming performance of several species of fish from the Mackenzie River. Fisheries Research Board of Canada 31:1641-1647. Kainz, E., and H. P. Gollman. 1996. Artificial hatching and first rearing experiments with burbot (Lota lota). Osterreichs Fisherei 49:154-160. Kjellman, J., J. Lappalainen, M. Vinni, L. Uusitalo, and J. Saren. 2000. Occurrence of burbot in a eutrophic lake. Pages 105-110 in V. L. Paragamian and D. W. Willis, editors. Burbot biology, management, and ecology. Fisheries Management Section, Special Publication 1, American Fisheries Society, Bethesda. Kleinbaum, D. G., L. L. Kupper, and K. E. Muller. 1988. Applied regression analysis and other multivariable methods. 2 n d ed. Duxbury Press, Belmont, California. Lafferty, B., and D. Vincent-Lang. 1991. Status of lake burbot stocks in the upper Copper/upper Susitna Management Area. Report to the Alaska Department of Fish and Game, Sport Fish Division, Anchorage. Lawler, G. H. 1963. The biology and taxonomy of the burbot, Lota lota, in Heming Lake, Manitoba. Journal of the Fisheries Research Board of Canada 29: 417-433. McCrimmon, H. R. 1959. Observations on spawning burbot in Lake Simcoe, Ontario. Journal of Wildlife Management 23: 447-449. McCullagh, P., and J. A. Nelder. 1989. Generalized linear models. 2 n d ed. Chapman and Hall, London. 66 McPhaiL J. D. 1997. A review of burbot (Lota lota) life history and habitat use in relation to compensation and improvement opportunities. Canadian Manuscript Report of Fisheries and Aquatic Sciences 2397. McPhaiL J. D., and C. C. Lindsey. 1970. Freshwater fishes of northwestern Canada and Alaska. Bulletin of the Fisheries Research Board of Canada 173, 381 p. McPhail, J. D., and V. L. Paragamian. 2000. Burbot biology and life history. Pages 11-24 in V. L. Paragamian and D. W. Willis, editors. Burbot biology, management, and ecology. Fisheries Management Section, Special Publication 1, American Fisheries Society, Bethesda. Mehl, S. 1989. The northeast arctic cod stock's consumption of commercially exploited prey species in 1984-1986. Rapport Proces-vebaux des Reunions, International Council for the Exploration of the Sea 188: 185-205. Meshkov, M. M . 1967. Developmental stages of burbot (Lota lota L.). Voprosy Ikhtiologii I Gidrobiologii Vodnemov 62:181-194 (in Russian, English translation in Zoology Reprint Library, Univ. of Toronto). Miller, T. J., L. B. Crowder, J. A. Rice, and A. A. Marschall. 1988. Larval size and recruitment mechanisms in fishes: toward a conceptual framework. Canadian Journal of Fisheries and Aquatic Sciences 45: 1657-1670. Moreau, J. 1987. Mathematical and biological expression offish growth in fishes: recent trends and further developments. Pages 81-113 in C. Summerfelt and G. E. Hall, editors. The age and growth of fish. The Iowa State University Press, Ames. Muth, K. M. , and L. L. Smith. 1974. The burbot fishery in Lake of the Woods. University of Minnesota, Agricultural Experimental Station, Technical Bulletin 296, St. Paul. Nelson, J. S. 1994. Fishes of the world. 3 r d ed. Wiley, New York. Neter, J., M. H. Kutner, C. J. Nachsheim, and W. Wasserman. 1996. Applied Linear statistical models. 4 th ed. Irwin, Chicago. Nikcevic, M. 2000. Thermal acclimation capacity of the burbot Lota lota L. Pages 71-77 in V. L. Paragamian and D. W. Willis, editors. Burbot biology, management, and ecology. Fisheries Management Section, Special Publication 1, American Fisheries Society, Bethesda. 67 Paragamian, V. L., V. Whitman, J. Hammond, and H. Andrusack. 2000. Collapse of burbot fisheries in Kootenai River, Idaho, USA, and Kootenay Lake, British Columbia, Canada. Pages 155-164 in V. L. Paragamian and D. W. Willis, editors. Burbot biology, management, and ecology. Fisheries Management Section, Special Publication 1, American Fisheries Society, Bethesda. Pauly, D. 1981. The relationship between gill surface area and growth proportions in fish: a generalization of Von Bertalanffy's theory of growth. Meeresforschung 28: 251-282. Pauly, D. 1994. On the sex of fish and the gender of scientists. Fish and Fisheries Series 14, Chapman and Hall, London. 250 p. Pepin, P., D. C. Orr, and J. T. Anderson. 1997. Time to hatch and larval size in relation to temperature and egg size in Atlantic cod (Gadus morhua). Canadian Journal of Fisheries and Aquatic Sciences 54(Suppl. 1): 2-10. Primack, R. B. 1993. Essentials of conservation biology. Sinauer Associates, Sunderland, Massachusetts. R. L. and L. Environmental Services. 1993. A fisheries investigation of Columbia Lake. Report to the Mica Fisheries Compensation Program, Nelson, B.C. Ricker, W. E. 1973. Linear regression in fisheries research. Journal of the Fisheries Research Board of Canada 30: 409-434. Ricker, W. E. 1979. Growth rates and models. Pages 677-743 in W. S. Hoar and D. J. Randall, editors. Fish physiology, volume 8. Academic Press, New York. Robins, C. R., and E. E. Deubler. 1955. The life history and systematic status of the burbot, Lota lota lacustris (Walbaum), in the Susquehanna River System. New York State Museum and Science Service Circular 39,44 p. Ryder, J. M. 1981. Biophysical resources of the east Kootenay area: terrain. Assessment and Planning Division Bulletin 7, British Columbia Ministry of the Environment, Victoria. Ryder, R. A., and J. Pesendorfer. 1992. Food, growth, habitat, and community interactions of young-of-the-year burbot, Lota lota (L), in a Precambrian Shield lake. Hydrobiologia 243/244: 211-227. 68 Sandlund, O. T., L. Klyve, and T. F. Naesje. 1985. Growth, habitat and food of burbot, Lota lota, in Lake Mjoesa. Fauna Blindern 38: 37-43. (in Norwegian with English summary) Scott, M., and J. A. Brown. 1998. Diel foraging cycles and interactions among juvenile Atlantic cod (Gadus morhua) at a nearshore site in Newfoundland. Canadian Journal of Fisheries and Aquatic Science 55:1307-1316. Scott, B., and E. J. Crossman. 1973. Freshwater fishes of Canada. Bulletin of the Fisheries Research Board of Canada 184, 966 p. Shodjai, F. 1980. Entwicklungs-, stoffwechsels- und ernahrungs-physiologische untersuchungen an der aalquappe (Lota lota L.) unter berucksichtigung ihrer eignung als kulturfisch. Doctoral Thesis, University of Kiel, 201 pp. Sissenwine, M . P. 1984. Why do fish populations vary? Pages 59-94 in May, R.M., editor. Exploitation of marine communities. Springer-Verlag, Berlin. Smith, B. R. and J. J. Tibbies 1980. Sea lamprey Petromyzon marinus in Lakes Huron, Michigan, and Superior: History of Invasion and control, 1936-78. Canadian Journal of Fisheries and Aquatic Sciences 37:1780-1801. Steiner, V., N. Schotzko, M . Kletzl, and E. Kainz. 1996. A contribution to the economical raising of small, sensitive fish larvae by the example of burbot (Lota lota L.). Osterreichs Fischerei 49:160-172. Sunby S., H. Bjorke, A. V. Soldal, and S. Olsen. 1989. Mortality rates during the early life stages and year-class strength of northeast Arctic cod. Rapport Proces-vebaux des Reunions, International Council for the Exploration of the Sea 191: 351-358. Taube, T., and D. R. Bernard. 1995. Stock assessment and biological characteristics of burbot in Lake Louise and Tolsona Lake. Alaska Department of Fish and Game, Fisheries Data Series no. 9, Anchorage. Taylor, C.C. 1958. Cod growth and temperature. Journal du Conseil, Council International pour l'Exploration de la Mer 23: 366-370. Tupper, M . , and R. G. Boutilier. 1995a. Effects of habitat on settlement, growth, and post settlement survival of Atlantic cod (Gadus morhua). Canadian Journal of Fisheries and Aquatic Science 52:1834-1841. 69 Tupper, M. , and R. G. Boutilier. 1995b. Size and priority at settlement determine growth and competitative success of newly settled Atlantic cod. Marine Ecology Progress Series 118: 295-300. Urban Systems. 1997. Columbia Lake management strategy. Report to Regional District of East Kootenay, Cranbrook, British Columbia. Walters W. J. and F. Juanes. 1993. Recruitment limitation as a consequence of natural selection for use of restricted feeding habitats and predation risk taking by juvenile fishes. Canadian Journal of Fisheries and Aquatic Sciences 50: 2058-2070. Warner, R. R., and P. L. Chesson. 1985. Coexistence mediated by recruitment fluctuations: a field guide to the storage effect. The American Naturalist 125: 769-787. Werner, E. E., J. F. Gilliam, D. J. Hall, and G. G. Mittelbach. 1983. An experimental test of the effects of predation risk on habitat use in fish. Ecology 64(6): 1540-1548. 70 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items