Status of Burbot (Lota lota) in Arrow Lakes Reservoir Victoria St., Nelson, B.C.

Transcription

1 Status of Burbot (Lota lota) in Arrow Lakes Reservoir Steve Arndt 1, M.Sc. and James Baxter 2, M.Sc. 1 Columbia Basin Fish & Wildlife Compensation Program Victoria St., Nelson, B.C. 2 BC Hydro, th St., Castlegar, B.C. March 2006

2 Executive Summary The status of burbot populations in reservoirs is of interest with respect to evaluating both footprint and operational impacts of BC Hydro dams. Arrow Lakes Reservoir (ALR) is located in the West Kootenay and is located downstream of two other large impoundments. Prior to this study, no directed assessment of burbot had been conducted on the reservoir. The primary objective of this study was to provide an initial assessment of the status of burbot in ALR including distribution, relative abundance, and size and age structure. Secondary objectives were to identify spawning locations and evaluate the use of an underwater video camera for line transect or mark-recapture population estimates. Distribution of burbot was determined using data from incidental catches during sturgeon set line sampling in Relative abundance was determined using catch-per-unit-effort (CPUE) in cod traps baited with dead kokanee. Traps were set in two widely separated locations (the narrows between the two main basins, and Galena Bay) from late October to early November of 2003, and one location (narrows) in late October Biological data were obtained from fish caught in the cod traps. All sampled fish were marked with Floy tags applied posterior to the dorsal fin prior to release. Ten fish were surgically implanted with radio tags during the trapping in 2003 and flights were conducted around the presumed time of spawning in an attempt to locate spawning areas. Underwater video camera surveys were conducted in the narrows area on February Age was assessed using otoliths collected during a creel survey. Set line surveys indicated that burbot were widely distributed throughout the reservoir from Revelstoke Reach to Hugh Keenleyside Dam. Average number of burbot per overnight trap set was 4.5 and 4.7 for the narrows and Galena Bay respectively in 2003, and 8.5 for the narrows in Relative abundance, as inferred from trap CPUE, was the highest reported in the province. Median CPUE for baited traps in the ALR was more than 10 times higher than other B.C. lakes sampled to date with similar gear. Burbot in the ALR were relatively large in comparison with other B.C. lakes, with more than 90% of the trap catch exceeding 530 mm. Age structure showed some evidence of dominant and weak cohorts, but suggested relatively stable recruitment compared to Columbia Lake in the East Kootenay. There was large variation in length-at-age among individual fish; however, average growth rates in ALR were higher than most other B.C. populations for which data are available. Mature males ranged from 598 to 875 mm with two immature males at 510 mm. Mature females ranged from 630 to 860 mm and immature females from 510 to 640 mm. These mature fish would be age-5 or older. Flights located three of the radio-tagged fish after release. One was found near the mouth of Mosquito Creek and this is a likely spawning area for burbot. Another was located near the release site, and the third about 9 km north of the release site. Fish that were not detected may have spawned in deeper locations, or may not have spawned during the surveyed period. No aggregations of spawning burbot were observed in the narrows with the underwater video camera in February, however, the camera showed the types of habitat occupied and the daytime behaviour of burbot. Most burbot were resting inactive on the bottom and could be examined closely for the presence of a Floy tag. Therefore the camera might be useful for mark-recapture population estimates, if it was used immediately after a period of intense trapping and Floy tagging. Key Words: burbot, Lota lota, fish, age, growth, maturity, cod, abundance, Floy tag, radiotelemetry, spawning, stock density, habitat, depth, substrate Columbia Basin Fish and Wildlife Compensation Program ii

3 Acknowledgements Funding and in kind support for this project was provided by the Columbia Basin Fish & Wildlife Compensation Program (CBFWCP), BC Hydro (Castlegar), and the BC Ministry of Environment (Nelson). Thanks to Colin Spence, Glen Olson, Matt Neufeld, Mike Hildebrand, Demitria Burgoon, Bob Chapman, Dean den Biesen, Jeremy Baxter, Duncan Wassick, Dave Mair, Josh Taylor, and Janice Arndt. CBFWCP is a joint initiative between BC Hydro, the BC Government (Ministry of Environment) and Fisheries and Oceans Canada to conserve and enhance fish and wildlife populations affected by the construction of BC Hydro dams in Canada s portion of the Columbia Basin. Cover photo credit: Ernest Keeley Columbia Basin Fish and Wildlife Compensation Program iii

4 Table of Contents Executive Summary... ii Acknowledgements... iii Table of Contents... iv List of Tables...v List of Figures...v 1.0 INTRODUCTION METHODS Field Sampling Radio Telemetry Underwater Video Lab and Data Analysis RESULTS AND DISCUSSION Distribution and Depth of Capture Relative Abundance Growth, Condition, and Maturity Size and Age Structure Movements and Spawning Locations...17 Radio Telemetry...17 Tag Returns Underwater Camera Observations CONCLUSIONS AND RECOMMENDATIONS REFERENCES...22 Appendix 1. Summary of burbot trapping data for the narrows south of Nakusp in Appendix 2. Summary of burbot trapping data for Galena Bay in Appendix 3. Summary of burbot trapping data for Nakusp area in Appendix 4. Length-frequency of burbot used for ageing (A) and from 2003 (B) and 2004 (C) trapping in the Nakusp area of Arrow Lakes Reservoir. Appendix 5. Depth and habitat description for burbot observed by underwater camera on February Columbia Basin Fish and Wildlife Compensation Program iv

5 List of Tables 1. Summary of burbot trapping effort and depths sampled on Arrow Lakes Reservoir in 2003 and Further details are provided in Appendices 1 to Summary of burbot implanted with radio tags in the Arrow Lakes Reservoir in Oct, Summary of tracking locations of radio tagged burbot in Arrow Lakes Reservoir during the spring of Summary of recaptures for burbot tagged in Arrow Lakes Reservoir in 2003 and Summary of sampling data for LBV underwater camera in the narrows area of Arrow Lakes Reservoir in List of Figures 1a. Locations of burbot traps set in the narrows between the upper and lower basins of Arrow Lakes Reservoir in Set numbers correspond with Appendix b. Locations of burbot traps set in Galena Bay of Arrow Lakes Reservoir in Set numbers correspond with Appendix Locations of burbot traps set in Arrow Lakes Reservoir in Trap numbers correspond with Appendix Locations (squares) where burbot were caught on set lines in Arrow Lakes Reservoir in 1995 (data from Ministry of Water, Land, and Air Protection, and Kootenay Wildlife Services Ltd Comparison of CPUE (number of burbot/trap hour) for cod traps in Arrow Lakes Reservoir in 2003 and Comparison of Arrow Lakes Reservoir CPUE to other British Columbia Lakes sampled with baited cod traps Length-at-age of burbot from Arrow Lakes Reservoir Comparison of mean length-at-age of burbot in Arrow Lakes Reservoir to five other British Columbia Lakes Length-weight relationships for Arrow Lakes Reservoir burbot captured by cod trap in 2003 and Comparison of mean Wr by length category for Arrow Lakes Reservoir and four other British Columbia Lakes Length frequency distributions for burbot taken in cod traps from Arrow Lakes Reservoir Mean proportional and relative stock density of burbot from Arrow Lakes Reservoir compared to four other British Columbia Lakes sampled with cod traps. Length frequency distributions for burbot taken in cod traps from Arrow Lakes Reservoir Year class distribution of burbot sampled by cod traps and creel survey in Arrow Lakes Reservoir...16 Columbia Basin Fish and Wildlife Compensation Program v

6 1.0 INTRODUCTION Burbot have become a species of concern in Kootenay Region due to major declines of certain populations. In Kootenay Lake and the lower Kootenay River, for example, burbot were very abundant during the late 1960s and early 1970s, but currently these populations are red-listed (British Columbia Conservation Data Centre). Fish habitats in the Canadian portion of the Columbia River basin have been impacted substantially by the construction of several large reservoirs from the 1960s through to the 1980s for flood control and hydroelectric power production. In other parts of the world, creation of reservoirs has sometimes been accompanied by increased burbot density (McPhail 1995). However, changes in flow regimes, temperature, or nutrient loading may be detrimental to burbot downstream of impoundments (McPhail 1995; Paragamian 2000; Paragamian et al. 2000; Ahrens and Korman 2002). Consequently, the status of burbot populations is of concern from both footprint and operational hydroelectric perspectives. Arrow Lakes Reservoir (ALR) is located in the West Kootenay between the Selkirk and Monashee mountain ranges. It is 230 km long with a surface area of 49,500 ha at full pool (Sebastian et al. 2000). Two main basins at the north and south are separated by a shallower narrows section approximately 30 km long near the midpoint of the reservoir. The reservoir is located downstream of two other large impoundments (Kinbasket and Revelstoke Reservoirs). Burbot assessments have occurred in a number of other lakes in the Kootenay Region (e.g., Arndt and Hutchinson 2000; Baxter et al. 2002a, 2002b; Bisset et al. 2002; Spence and Neufeld 2002; Neufeld and Spence 2004b), but prior to this study there was no directed stock assessment of burbot in ALR. Some limited information was available from a creel survey and incidental catches during sturgeon assessment. The creel survey indicated a small but apparently growing fishery for burbot, most of which occurs in the narrows south of Nakusp (Arndt 2004). Catch rates are high, and there was concern that the fishery might be based on an unusual concentration of fish similar to that which is believed to have occurred in the West Arm of Kootenay Lake. The primary goal of this study was to provide an initial assessment of burbot status in the ALR. Key population parameters including distribution, relative abundance, and size and age structure are described, with comparisons to other British Columbia lakes where data are available. A second aim was to identify spawning locations, and the third was to evaluate the use of underwater video camera in mark-recapture or line-transect population estimates (Bergstedt and Anderson 1990; Edsall et al. 1993). 2.0 METHODS 2.1 Field Sampling Distribution of burbot in the reservoir was inferred from incidental catches during sturgeon sampling in 1995 (BC Ministry of Environment; data on file). In that study, the whole length of the reservoir was sampled at roughly equal intervals using overnight set lines baited with dead kokanee. Cod traps were used in late October/early November of 2003 and late October 2004 to determine relative abundance and size distribution. We used traps similar to those used in coastal Columbia Basin Fish and Wildlife Compensation Program 1

7 commercial cod fisheries, baited with dead kokanee collected the previous spawning season. These traps are easy to deploy and effective for capturing burbot (Spence 2000). Resources were not sufficient to allow a randomized survey covering the whole reservoir. Therefore we focused on two areas in The first was in the narrows south of Nakusp between the two main basins where the majority of known angling occurs; the second was in Galena Bay about 60 km to the north. In 2004, trapping was done again in the area south of Nakusp as part of another program to obtain fish for fish culture experiments. The majority of sets were in the narrows area but three traps were also set close to the Town of Nakusp. The two years of trapping near Nakusp allow a comparison of relative abundance over two years. All traps were left overnight before lifting, with two exceptions (Appendices 1 and 3). A total of 43 sets were made in 2003 and 15 in 2004 (Table 1; Figures 1a, 1b, 2). In order to ensure an adequate sample of fish for size distribution data in 2003 and fish culture in 2004, a number of traps were clustered around one of the more productive angling areas at Nakusp. Possible influences of this on catch rates are considered in the Results and Discussion section. Table 1. Summary of burbot trapping effort and depths sampled on Arrow Lakes Reservoir in 2003 and Further details are provided in Appendices 1 to 3. Year Location No. Sets Depth (m) Time Fished (h) Average Range Average Range 2003 Nakusp Galena Bay Nakusp An overnight decompression procedure (Neufeld and Spence 2004a) was used to reduce decompression trauma to burbot. This involved retrieving each trap and moving it quickly to a depth of approximately half of the original pressure. The number of fish in the initial pull was recorded as well as the number when the trap was processed on the following day. Most burbot had gas accumulation in the abdominal cavity when retrieved, even after the decompression period. This was relieved in the majority of fish by puncturing the body cavity with a hollow needle so that fish were able to submerge to deeper water. Bruesewitz et al. (1993) found that this did not affect survival of burbot. Total length and weight were recorded and all fish were tagged with numbered Floy tags before release. Ten fish were selected for radio tagging. These fish were assessed for sexual maturity during the surgery, as well as a few fish that died during operations. Twelve otoliths were obtained from burbot at Nakusp, six more were obtained from set line captures in a sturgeon study, and 38 from the ALR creel survey at the Nakusp access point. All otoliths were stored dry in envelopes. Columbia Basin Fish and Wildlife Compensation Program 2

8 Upper Arrow Dog Creek Baerg Creek , , , 26 1, McDonald Creek 14 Figure 1a. Locations of burbot traps set in the narrows between the upper and lower basins of Arrow Lakes Reservoir in October Set numbers correspond with Appendix 1. Burbot Trap Locations - set October 2003 Columbia Basin Fish & Wildlife Compensation Program Kilometers March 29, 2006 w:\\srm\nel\workarea\cbfwcp\fish\burbot\burbot_arrowa2003.mxd $MAPS/fish/maps2005/burbot_arrow2003a.pdf 3

9 Hi ll Cree k Galena Bay Mackenzi e Cre ek 32 Figure 1b. Locations of burbot traps set in Galena Bay of Arrow Lakes Reservoir in November Set numbers correspond with Appendix 2. Burbot Trap Locations - set November 2003 Columbia Basin Fish & Wildlife Compensation Program March 29, Kilometers w:\\srm\nel\workarea\cbfwcp\fish\burbot\burbot_arrowb2003.mxd $MAPS/fish/maps2005/burbot_arrow2003b.pdf 4

10 15 Nakusp Upper Arrow Lake : : : : : ::: 8 : :: 7 Burbot Trap locations - set October 2004 : Columbia Basin Fish & Wildlife Compensation Program :: : : : :: :: MacDonald Creek Dog Creek Figure 2. Locations of burbot traps set in Arrow Lakes Reservoir in October Set numbers correspond with Appendix Kilometers March 30, 2006 w:\\srm\nel\workarea\cbfwcp\fish\burbot\burbot_arrow.mxd $MAPS/fish/maps2005/burbot_arrow.pdf 5

11 2.2 Radio Telemetry Of the burbot that were captured in the area south of Nakusp, we selected ten individuals for surgical implantation of radio tags. Radio tags used for this study were manufactured by Lotek Engineering Inc of Newmarket, Ontario. The tags used were model MCFT-3A (16 mm in diameter, 50 mm in length, 6.2 g in weight, operation life > 680 days). They were digitally coded and transmitted on either frequency or We selected fish that were a minimum of 1 kg in weight so that the weight of the tag did not exceed 2% of the weight of the fish. Sterile conditions were maintained at the site with the person performing the surgery scrubbing with Betadine liquid soap and donning sterile gloves. All operating instruments were sterilized in a container of germiphene. Once fish were ready for processing, they were immersed in an anaesthetic bath containing water and clove oil at a concentration of 100 ppm. The burbot were taken to stage IV of anaesthesia, which generally took four to five minutes. Each fish was then removed from the bath, laid on its back in a v-shaped operating trough lined with foam, and the gills were irrigated with water using spray bottles. The incision into the abdominal cavity (left hand side of the body wall) was then made with a scalpel fitted with a curved (No. 12) blade. After the incision was complete, a 16 gauge stainless steel needle was inserted through the abdominal wall posterior to the incision, and back out the incision. The antenna of the radio tag was then threaded out through the needle, and the needle was pulled out leaving the antenna coming out the side wall of the fish. The radio tag was then inserted into the abdominal cavity and the incision was closed with three interrupted sutures of 40% monofilament on a cutting needle. We then applied Betadine to the closed incision and the antenna exit point. Biological data was then collected, the fish was Floy tagged, and the fish was allowed to recover in a flow through fish tube (generally 30 minutes) prior to release. 2.3 Underwater Video We used a remotely operated vehicle (Little Benthic Vehicle LBV300, Seabotix Inc.) that was equipped with a video camera and halogen light. The unit was attached to the surface by a cable that transferred data to and from the unit. The camera image was projected in colour on a TV monitor on the boat where an operator controlled the depth and direction of the LBV. Results were then recorded on 8 mm video tape. Fifteen sites were surveyed by LBV in the narrows area. At each site, the unit was lowered to the bottom at the boat and moved away from the boat in a relatively straight transect until reaching a distance of up to 250 m or less (the limit of the cable). In areas where there was a noticeable current, it was necessary to travel either directly against, or in the same direction as the current, to maintain a fairly consistent direction of travel. Underwater video camera work included four objectives. The first was to determine whether there were spawning aggregations of burbot in the narrows area where angling effort is concentrated. The second was to determine the feasibility of using the unit for line transect sampling such as that described by Bergstedt and Anderson (1990). The third was to test whether the unit could be used for the recapture phase of a mark-recapture population estimate. For this objective it is necessary to clearly observe whether or not a burbot carried a tag from the earlier Columbia Basin Fish and Wildlife Compensation Program 6

12 trapping session. The last objective was to describe the daytime behaviour and habitats (substrate types) occupied by burbot. Our original intention was to conduct the LBV sampling soon after the trapping phase of the project was completed in early November. In this way, the likelihood of immigration and emigration of fish from the narrows would be minimized and assumptions for the mark-recapture estimate would be more nearly met. However, the equipment was not available until late January, and then malfunctioned on the first attempt, so that the surveys did not occur until February, three months after the trapping was completed in the narrows. 2.4 Lab and Data Analysis Otoliths were aged by an independent consultant with previous burbot experience (Josh Taylor, Vancouver, B.C.). Age was assessed first by looking at the otoliths whole and then splitting them if necessary (about 2/3 were split). Each split otolith was burned and brushed with mineral oil to highlight the rings. Examination was done under a microscope with illumination from a fibre optic light. Trapping results and fish size and age data were analyzed using Systat software (Version 10 for Windows). Relative abundance was measured by catch per unit effort (CPUE) in cod traps and expressed as number of burbot/trap hour. Since baited cod traps have been used on several other B.C. lakes, CPUE should provide a reasonable index of differences in relative abundance between locations or over time. The 95% confidence limits around the median CPUE were compared among locations and years using the notched box plots procedure in Systat (McGill et al. 1978, cited by Systat 1999). Size structure of the population was compared within the ALR using length frequency plots, and to other lakes using relative stock density plots (Fisher et al. 1996). The maximum length of ALR burbot was approximated using the mean length of the ten largest fish in the sample. Ultimate length (L ) could not be estimated using Walford plots (Everhart and Youngs 1981) because of large variation in length-at-age and a lack of younger fish in the aged sample. The length-weight condition of ALR burbot was compared to other populations using relative weight as recommended by Fisher et al. (1996). 3.0 RESULTS AND DISCUSSION 3.1 Distribution and Depth of Capture Incidental catches of burbot during baited set line sampling for sturgeon in 1995 show that the species is widely distributed throughout the reservoir, from Revelstoke Reach in the north to Keenleyside Dam in the south (Figure 3). Burbot were also a common incidental catch on set lines for sturgeon in other years, although sampling was conducted in more restricted areas [White Sturgeon (WST) database from ; Golder Associates, Castlegar, B.C.]. Depth of capture for burbot in the cod traps in 2003 ranged from 11.3 to 30 m in the narrows area (the only trap with a catch of zero was at 10 m) and 11.5 to 31 m in Galena Bay (again the only trap with a catch of zero was at 10 m). In 2004, cod trap depth of capture ranged from 23 m Columbia Basin Fish and Wildlife Compensation Program 7

13 to 30 m (one trap at 18.5 m had no fish). For overnight set lines in Galena Bay and Beaton Arm, capture depths ranged from 5 m to 62 m with an average of ~ 23 m (n=102; WST database ; Golder Associates, Castlegar, B.C.). In summary, burbot in the ALR show a preference for water over 10 m deep, but this may vary at different times of year and different reservoir levels. Depths occupied may relate to suitable temperatures, however, we did not measure temperature during our sampling. It is also worth noting that our study areas did not include the deeper parts of the reservoir. Fall cod traps in Duncan Reservoir captured no burbot in less than 16 m depth; highest capture rates were from 26 to 30 m, although again the deepest portions of the reservoir were not sampled (Spence and Neufeld 2002; Neufeld, pers. comm.). Columbia Lake, in the East Kootenay, contrasts with the above results. It has a maximum depth of 6.0 m, and trapping on this lake has captured adult burbot as shallow as 1 m (Bisset et al. 2002). Columbia Basin Fish and Wildlife Compensation Program 8

14 Status of Burbot in Arrow Lakes Reservoir Figure 3. Locations (squares) where burbot were caught on set lines in Arrow Lakes Reservoir in 1995 (data from Ministry of Water, Land, and Air Protection, and Kootenay Wildlife Services Ltd.). Map adapted from Pieters et al. (2000). Cod traps were deployed in the Narrows and Galena Bay in 2003 and Columbia Basin Fish and Wildlife Compensation Program 9

15 3.2 Relative Abundance Cod trap CPUE was similar between Nakusp and Galena Bay in 2003, with medians of about 0.2 burbot/trap hour (Figure 4). The highest CPUE value (1.03, Nakusp 2003; Figure 4) was from a trap that was lifted after less than 5 hours. This indicates that many burbot were able to locate the bait and enter the traps in a short time, before nightfall. The CPUE for the Nakusp sampling in 2004 was significantly higher than both 2003 sites (note non-overlapping confidence limits, Figure 4). Average number of burbot after an overnight trap set was 4.5 and 4.7 burbot per trap for Nakusp and Galena Bay in 2003, with a maximum of 9 and 12 burbot, respectively. For Nakusp in 2004, average and maximum number of burbot per overnight set was 8.5 and 18, respectively. These catch rates are the highest observed to date in Kootenay Region CPUE (burbot/hour) Galena 2003 Nakusp 2003 Nakusp 2004 Location and Year Figure 4. Comparison of CPUE (number of burbot/trap hour) for cod traps in Arrow Lakes Reservoir in autumn of 2003 and Box includes the central 50% of values, notch indicates the median with 95% confidence limits, and whiskers extend to 1.5 times the interquartile range. Actual values for individual sets are indicated by dots. The outlier for Nakusp 2003 was retrieved after less than 5 hours, whereas other sets were approximately 20 hours. The significant increase in CPUE in 2004 could be a reflection of different sampling conditions (e.g., temperature, current, trap spacing), burbot movement into the narrows, or possibly a population increase. However, the most important observations are: (1) CPUE was similar at two Columbia Basin Fish and Wildlife Compensation Program 10

16 widely separated locations in 2003, and (2) the unexpectedly high catch rates in 2003 were not an anomaly. The high catch rates in Galena Bay, and also in traps set well north of the narrows near Nakusp in 2004, indicate that the narrows area is not the only concentration of burbot in the reservoir. Differences between 2003 and 2004 in the ALR, although statistically significant, were minor compared to the difference between ALR and other British Columbia lakes sampled with baited traps (Figure 5). Median CPUE in the ALR was more than ten times higher than other lakes, and in fact median values for the ALR were higher than maximum values seen in most other lakes. This is a remarkable finding, and if cod trap CPUE is a reliable indicator of relative abundance, the sampled areas in the ALR have the highest burbot abundance measured in the province thus far CPUE (burbot/hour) Arrow -0.2 Nak03 Nak04 Dunc01 Dunc03 Koot Col01f Roche Gal03 Dunc00 Dunc02 Trout Col01s Location and Time Birch Figure 5. Comparison of Arrow Lakes Reservoir CPUE to other British Columbia Lakes sampled with baited cod traps. Box includes the central 50% of values, notch indicates the median with 95% confidence limits, whiskers extend to 1.5 times the interquartile range, and dots indicate outliers. Lakes and sample times are as follows: Duncan (Dunc00, Dunc01, Dunc02, Dunc03), Trout 2001 (Trout), Kootenay 2000/01 (Koot), Columbia spring and fall 2001 (Col01s, Col01f), Birch 2002 (Birch) and Lac Des Roche 2002 (Roche). Data for other lakes are from Bisset et al. (2002), Reddekopp et al. (2003), Baxter et al. (2002a, 2002b), and Neufeld and Spence (2004b). Columbia Basin Fish and Wildlife Compensation Program 11

17 Bernard et al. (1993) compared CPUE of baited hoop traps to mark-recapture density estimates for 15 burbot populations in small to moderate-sized lakes (up to 6500 ha) in Alaska. They found that changes in abundance could be detected with mean CPUE in all lakes. On a finer scale, they found a mild asymptototic relationship between mean CPUE and burbot density, which they believed was most likely from partial saturation of their sampling gear. Mean number of fish per trap was 7 in their higher density lakes, with up to 30 in a single trap. Since our traps rarely exceeded 10 burbot, it seems unlikely that gear saturation would have had a significant influence on CPUE. However, the hoop traps used in their study have a different configuration than the cod traps in this study and the relative capacity of the two trap types is not known 1. Two other factors that may have influenced CPUE at Nakusp are clustering of traps leading to gear competition, and avoidance of shallow water. Clustering in areas of suspected high density was related to the need to collect a sufficient sample of fish with limited time and resources, but this may have slightly decreased CPUE, especially in Bernard et al. (1993) estimated the effective fishing areas for their baited traps to be a little less than one hectare. This would correspond to a spacing of about 55 m between traps in locations where there is no direct current. Most of our traps in the narrows were greater than 55 m apart, and there is a downstream current that might carry scent for some distance. Most trap sites in the narrows were also selected to be within the deeper channel formed by the historical streambed. Outside of this channel a substantial portion of the area is less than 10 m. The avoidance of shallow water probably increased CPUE in relation to what it would have been if the sites had been selected randomly in the narrows. However, the CPUE should still be fairly representative of areas greater than 10 m depth. At Galena Bay, neither clustering nor avoidance of shallow water is likely to have substantially influenced CPUE. As noted earlier, this study did not attempt to sample the whole reservoir due to cost constraints. Instead, we targeted two areas. The narrows area was suspected to have higher densities due to the concentration of angling there; Galena Bay is 60 km to the north and is not known as a popular burbot angling area. Relative abundance was high in both sampled locations. To obtain a crude idea of actual abundance, the average catch in ALR cod traps of burbot/24 hr set is in the order of hoop trap catches of 5 to 7 burbot/48 hr set, which corresponded to densities of 10 to 15 burbot/ha in the Bernard et al. (1993) study. 2 This density is high in relation to studied Alaska lakes (Bernard et al. 1993), although burbot densities as high as /ha have been recorded in Lake Michigan and other waters (Edsall et al. 1993). 3.3 Growth, Condition, and Maturity There was large variation in length-at-age of ALR burbot (Figure 6), but mean length-at-age was higher than most other B.C lakes (Figure 7). The exception is Kootenay Lake during a period when the lake was enriched by nutrients from an upstream fertilizer plant. Due to the large variation in length-at-age and the relatively small sample of fish, it was not possible to obtain a 1 See Spence (2000) for diagrams and a comparison of hoop and cod traps. 2 The translation from CPUE to density for this study should be viewed with caution given the differences in trap type, geographical location, and study methods. Columbia Basin Fish and Wildlife Compensation Program 12

18 reasonable estimate of L for the population. Back-calculation of lengths at younger ages using otolith annuli might be a feasible method of obtaining the required information for younger ages, but was beyond the scope of this study. The mean length of the 10 largest fish in the sample was 886 mm, and the largest sampled was 935 mm. This suggests that the ultimate length of ALR fish is greater than Columbia and Kootenay Lakes (630 mm and 779 mm respectively), but less than Birch Lake (1073 mm) and Lac Des Roche (1235 mm) in the Thompson- Nicola Region (Arndt and Hutchinson 2000; Ahrens and Korman 2002; Reddekopp et al. 2003). The heaviest burbot weighed in this study was 5.1 kg. A photograph of an unusual 19 pound (8.6 kg) specimen from the ALR was featured in a local newspaper in The oldest age assessed in our sample was 13. Otoliths from the ALR were difficult to age. Part of the difficulty was due to variation in banding patterns among and within sampling locations. For instance, two burbot sampled at Nakusp that were of a similar size could have otoliths with very different banding patterns that suggested much different ages (J. Taylor; pers. comm.). In addition, the difference between clear and opaque rings was not as distinct as has been observed for other burbot otoliths (e.g., Columbia Lake). This could be due to natural differences among populations, or differences in storage methods (dry vs. dilute ethanol). Other factors that might influence banding patterns are altered temperature regimes due to changes in flow for power production and flood control, and the availability of opossum shrimp, Mysis relicta, as a winter food source. There may also be a large degree of burbot movement (Bernard et al. 1993) that could expose fish to varying levels of food availability. Another possible contributing factor to growth and otolith variation is the change in productivity of the reservoir since fertilization began in Length (cm) Mean Length (mm) Arrow Columbia Moyie Birch Roche Kootenay Age Figure 6. Length-at-age of burbot from Arrow Lakes Reservoir Figure 7. Comparison of mean length-at-age of burbot in Arrow Lakes Reservoir to five other British Columbia Lakes. Data are from Martin (1976), Arndt and Hutchinson 2000, Reddekopp et al. (2003), and K. Andreashuk (pers. comm., Moyie Lake). Age Columbia Basin Fish and Wildlife Compensation Program 13

19 The length-weight slope for ALR burbot was < 3 (Figure 8), indicating that burbot become less rotund as they increase in length (Anderson and Neumann 1996). Relative weight (Wr) is a method suggested by Fisher et al. (1996) for comparing condition across populations. When averaged by length category, Wr shows a lower value for ALR fish > 530 mm (Figure 9) which is somewhat similar to data for Trout and Duncan lakes in the West Kootenay, but different from the pattern for Birch Lake and Lac Des Roche. The slight trend towards lower condition in older fish is supported by both the slope of the length-weight relationship and the Wr averages, however, there was a substantial degree of variability among individual fish within the size categories (data not shown) that somewhat reduces confidence that there are strong population trends. Higher condition of smaller burbot in the ALR could be a result of feeding on Mysis relicta shrimp. Scott and Crossman (1979) note that deepwater invertebrates, especially Mysis, make up the majority of the diet for burbot up to 500 mm, and Mysis availability is likely high in the ALR, especially in the narrows areas. The situation may be somewhat similar to that at the upstream end of the West Arm of Kootenay Lake, which held high densities of burbot in the 1960s (Martin 1976). Martin (1976) attributed the fast growth of burbot in the West Arm of Kootenay Lake to large abundances of Mysis and kokanee. The diet of burbot in the ALR has not been investigated, however, we commonly observed large numbers of Mysis being expelled from burbot as they were raised from depth during the field sampling, and in one fish of 770 mm length the stomach was full of Mysis. Larger burbot are thought to be more piscivorous (McPhail 1995) and it may be that prey availability decreases at larger sizes y = 2E-08x R 2 = Arrow LR Birch Lake Lac Des Roche Trout Lake Duncan LR Weight (kg) Relative Weight (Wr) Length (mm) Figure 8. Length-weight relationship for Arrow Lakes Reservoir burbot captured by cod trap in autumn 2003 and Length Category (mm) Figure 9. Comparison of mean Wr by length category for Arrow Lakes Reservoir and four other British Columbia Lakes. Seventeen burbot were examined for sexual maturity in Mature males ranged from 598 to 875 mm, and two immature males were 510 mm. Mature females ranged from 630 to 860 mm, Columbia Basin Fish and Wildlife Compensation Program 14

20 and immature females from 510 to 640 mm. The mature fish would be age-5 or older (Figure 6). This size and age range for maturity is similar to that reported by Reddekopp et al. (2003) for Birch Lake and Lac Des Roche, although they did not find immature fish over 614 mm. Burbot from Palmer Lake, Washington, also mature at a similar size to ALR fish (Bonar et al. 2000). Lakes with earlier maturing burbot include Columbia Lake, where most mature at a length of 400 to 530 mm (age 3 or 4), and as small as 250 mm (age-2), and Lake Cle Elum, Washington (Arndt and Hutchinson 2000; Bonar et al. 2000). 3.4 Size and Age Structure Length distributions from Nakusp and Galena Bay in 2003 were similar in range although the Galena sample had more fish under 600 mm (Figure 10). Distinct peaks occurred in both locations in 2003 that may correspond to strong cohorts. The 2004 sample from Nakusp was similar to 2003 considering that fewer fish were measured (Figure 10). Stock density comparisons show that the ALR has a higher proportion of larger fish than four other British Columbia lakes (Figure 11). Count Nakusp Galena Nakusp Length (mm) Length (mm) Length (mm) Proportion per Bar Count Proportion per Bar Count Proportion per Bar Figure 10. Length frequency distributions for burbot taken in cod traps from Arrow Lakes Reservoir. Year class data (Figure 12) show that the 1993 cohort was the youngest that was fully recruited to the gear. Additional peaks at 1995 and 1997 suggest that these cohorts may have been stronger than the 1994 and 1996 cohorts, for example. Taylor and Arndt (in preparation) found large differences in burbot cohort abundance in Columbia Lake that were evident at the juvenile and adult stages. Differences between cohorts in the ALR are not as marked as those in Columbia Lake indicating that recruitment is probably more consistent in the ALR. Columbia Basin Fish and Wildlife Compensation Program 15

21 120 PSD RSD-P RSD-M Stock Density (%) Arrow LR Trout Duncan Birch Lake Lac Des Roche Figure 11. Mean proportional and relative stock density of burbot from Arrow Lakes Reservoir compared to four other British Columbia Lakes sampled with cod traps (PSD= percentage of sample >38 cm; RSD-P= percentage >53 cm; RSD-M= percentage >67 cm). Categories are based on recommendations of Fisher et al. (1996) Number of Burbot Year Class Figure 12. Year class distribution of burbot sampled by cod traps (n=16) and a creel survey (n=40) in Arrow Lakes Reservoir in 2003 and Columbia Basin Fish and Wildlife Compensation Program 16

22 3.5 Movements and Spawning Locations Radio Telemetry A total of ten burbot were implanted with radio tags and one fish died during surgery (Table 2). Tagged fish ranged in size from 600 to 875 mm, and all burbot were tagged in the narrows area near Nakusp. Several fish appeared to be maturing and were expected to spawn in late winter or spring of One point of note is the long period of time that it took for burbot to be anaesthetized and recover in the clove oil bath; this was also observed during a similar study at Columbia Lake (Arndt 2001). Table 2. Summary of burbot implanted with radio tags in the Arrow Lakes Reservoir in Oct, Length (mm) Weight (kg) Sex Floy Tag Radio Tag Comment 875 M (mat.) W Code M (mat.) W Code ? (imm.) W Code M (mat.) W Code F (mat.) mortality M (mat.) W Code M (mat.) W Code ? (imm.) W Code ? (imm.) W Code ? (imm.) W Code ? (imm.) W Code 8 We conducted two tracking surveys in the spring of 2004, which were carried out from an aircraft while conducting other surveys in the area. These aerial flights occurred on February 16 and March 19, On the first flight two burbot were located, and on the second flight three burbot were located (Table 3). The burbot located near the mouth of Mosquito Creek ( Code 10) was about 200 m away from the main channel of the narrows. Movement into this location is unlikely to be related to feeding, and therefore the creek inlet should be considered a likely spawning site. Burbot typically spawn in shallow bays or gravel shoals in lakes, but will also move into rivers or creeks to spawn (Scott and Crossman 1979; Arndt and Hutchinson 2000). The signal for the burbot located on the west side of ALR across from Nakusp ( Code 9) was strong and it is possible that this fish had died in very shallow water or perhaps been eaten by a terrestrial predator with the remains left on the shore. The failure to locate other radiotagged fish is likely because they were in deeper water that prevented signal detection. This may be an indication that they did not spawn during the surveyed period, or that they spawned in deeper locations. Columbia Basin Fish and Wildlife Compensation Program 17

23 Table 3. Summary of tracking locations of radio tagged burbot in Arrow Lakes Reservoir during the spring of Date Radio Tag Location Comments February 16, Code 9 West Side of ALR In shallow water across from Nakusp Code 10 West Side of ALR at Mosquito Creek In shallow water under ice March 19, Code 10 West Side of ALR at Mosquito Creek In shallow water under ice Code 8 Narrows Release site Code 9 West Side of ALR across from Nakusp In shallow water Tag Returns In the 2003 sampling, 223 burbot were tagged with spaghetti-type Floy tags and released in the vicinity of their capture location (153 in the narrows, and 70 in Galena Bay). From the time of their release to the end of 2004, a total of 88 burbot caught in the Nakusp area were examined. Of these, 5 were tagged (6%; Table 4). One additional tag was returned by mail. Burbot were not recorded at the two other creel sites during the above period. During the 2004 trapping, one of 166 sampled burbot retained a tag from Two fish from the 2004 sampling were detected in the creel in the following 8 months (Table 4). All recaptured fish were taken in the same area (narrows) as they had been tagged. Table 4. Summary of recaptures for burbot tagged in Arrow Lakes Reservoir in 2003 and Unless otherwise noted all tags were obtained from the Nakusp creel survey. Tagging Recapture Tag No. Date Location Length (cm) Weight (kg) Date Locality Length (cm) Weight (kg) W Oct-03 Narrows Dec-04 Nakusp W00491* 27-Oct-03 Narrows Feb-04 Nakusp na 3.6* W Oct-03 Narrows Jun-04 Nakusp W00171** 28-Oct-03 Narrows na na 27-Oct-04 Nakusp na na W Oct-03 Narrows Feb-04 Nakusp 67 na W00373*** 29-Oct-03 Narrows Dec-03 MacDonald Creek Park na Oct-04 Narrows Dec-04 Nakusp W Oct-04 Narrows May-05 Nakusp * Caught and released; weight estimated at ~8 lb ** Caught by angling during 2004 cod sampling *** returned by mail Columbia Basin Fish and Wildlife Compensation Program 18

24 3.6 Underwater Camera Observations The LBV was employed at 15 locations in the narrows area south of Nakusp in February, 2004 (Table 5). A total of 72 burbot was observed of which only one was tagged. In most cases the burbot were resting motionless on the substrate near pieces of sunken wood, although a few were seen swimming actively a meter or more above the substrate (Appendix 5). It was possible to approach resting burbot closely with the camera before they would swim away, allowing a clear determination of the absence or presence of a tag in almost all cases. In a few cases, individual burbot approached the LBV, apparently curious or interested in the possibility of food in the sediments that were stirred up by the propulsion of the machine. Large numbers of suckers (Catostomus spp.) were also observed, sometimes in large schools and sometimes in small groups. Usually suckers were observed where the current was stronger than was typically the case for burbot. The suckers were very wary of the LBV and could not be approached closely. Two bull trout were also observed. The LBV was useful for observing the behaviour and types of habitat used by burbot during the day, and the manoeuvrability of the unit, combined with the inactive behaviour of most burbot would allow the unit to be used as the recapture phase of a mark-recapture estimate once an initial batch of fish had been marked with external tags. A potential advantage of this method is that it could eliminate, or test for, violation of statistical assumptions such as trap avoidance or trap-addiction by previously captured fish. A disadvantage would be that the individual tag numbers could not be read. Sampling would be limited to locations and times of day when natural illumination was sufficient to allow the camera to operate without the halogen light. We sometimes switched on the light in deeper water (i.e., 30 m) or under low light conditions (late in day) but visibility was poor because of the reflection from many small particles in the water column. We did not attempt to make a mark-recapture estimate for this study because of the long time between tagging and LBV sampling. The LBV would not be suitable for attempting to estimate population size using a line transect estimation method because of difficulty in quantifying the width of observed transect. The angle of the LBV and its distance from the substrate cannot be held constant as it can be when a camera unit is mounted on a sled and towed on the bottom (e.g., Edsall et al. 1993). The relatively low number of burbot counted in February was surprising, given the number of fish caught in cod traps in the same area three months previously. This could be an indication of high efficiency of the baited traps, or a change in the density of burbot between November and February. Burbot are winter spawners and there may have been a movement to spawning sites. We did not observe any aggregations of spawning burbot in the narrows. A few burbot appeared to have recently completed spawning based on their thin appearance, but others were quite rotund and may have been gravid. Columbia Basin Fish and Wildlife Compensation Program 19

25 Status of Burbot in Arrow Lakes Reservoir Table 5. Summary of sampling data for LBV underwater camera in the narrows area of Arrow Lakes Reservoir in See Appendix 5 for depth and habitat descriptions of individual fish. Stations designated with A or B had transects in more than one heading direction. Station Location Date Depth Easting Northing Start Total time Heading Estimated Number Other species Bottom Type range (m) time (minutes) Distance (m) burbot 1 Narrows deep channel 12 Feb :33 20 South Few suckers Mostly fine sand or silt, small gravel, and some small sunken wood 2 Narrows deep 12 Feb :15 32 South & Many suckers Gravel and sand channel return (~1,000) 3 Narrows on 12 Feb :58 60 South Sucker Sand, gravel, cobble flat 4 Narrows near 12 Feb :37 23 South Suckers Sand, cobble, gravel white house 5 Narrows 12 Feb :22 12 South None Mostly sand with some gravel shallow site 6 Narrows near 12 Feb :50 22 South None Mostly sand Macdonald Ck 7 Narrows N of Macdonald Ck 12 Feb :22 66 South suckers Mostly silt with some sunken wood; one tagged burbot 8 N of Narrows 13 Feb :09 22 North None Sand and silt on east side 9A N of Narrows 13 Feb :44 31 North None Silt with small wood on west side 9AB N of Narrows 13 Feb :00 15 West sucker Silt with small wood on west side 10 N of Narrows 13 Feb :29 27 North Bull trout Fine silt with bits of wood west side 11 N of Narrows 13 Feb :16 19 North suckers Fine silt with small amount of wood near middle 12A Upstream end 13 Feb :07 20 South suckers, Fine silt with some wood of Narrows 1 bull trout 12B Upstream end 13 Feb :39 17 Northwest None Fine silt with some wood of Narrows 13A South of white 13 Feb :22 14 South Sucker schools Fine silt and wood house (hundreds) 13B South of white 13 Feb :48 11 Southeast Lots of suckers Fine silt, cobble house 14 Near Mosquito 13 Feb :29 10 South None Fine silt, cobble Creek 15A Off 13 Feb :11 14 South None Fine silt, few pieces of wood Macdonald Ck 15B Off Macdonald Ck 13 Feb :32 16 Northwest None Fine silt, few logs Columbia Basin Fish and Wildlife Compensation Program 20

26 4.0 CONCLUSIONS AND RECOMMENDATIONS 1. Set line captures indicated that burbot are widely distributed in the ALR, and cod trapping showed that burbot were abundant in two widely separated sampling locations over two years of sampling. Relative abundance, as indicated by trap CPUE, was substantially higher than other British Columbia lakes sampled to date. 2. The area south of Nakusp, in which the majority of burbot angling occurs, does not appear to be a single area of highly concentrated fish. Trap CPUE was similarly high in Galena Bay. 3. Burbot in the ALR were relatively large in size compared to other British Columbia lakes with more than 90% of the catch exceeding 530 mm (~ 1 kg). Sexually mature fish were typically 600 mm or greater in length, which corresponds to about age Age structure of the population showed some evidence of dominant and weak cohorts but the differences in cohort abundance were not as large as has been observed in Columbia Lake. 5. Radiotelemetry studies were not as successful as hoped in locating spawning areas likely due to burbot spawning in deep habitat where picking up a radio signal was difficult, or due to fish not spawning. The tracking did show that the mouth of Mosquito Creek is a likely spawning area for burbot. This area should be protected from development that would alter the existing habitat. No spawning aggregations were observed in the narrows during LBV video camera surveys in February. 6. The LBV was not useful for line transect population estimates because of difficulties in determining the exact width and length of transects. However, in most cases, burbot allowed the LBV to approach closely and determine with certainty whether a dorsal tag was present. Thus the LBV might be useful for mark-recapture estimates if it was used immediately after an intense period of trapping and tagging. 7. Trends in burbot abundance could be monitored by conducting cod trapping sessions every 3 to 5 years in the two locations for this study. Expansion to additional locations would improve the description of relative abundance. 8. Current methods have primarily sampled depths less than 50 m. Future assessment should consider incorporating sampling in deeper areas of the reservoir. 9. The habitats and growth of younger burbot in the ALR should be investigated. Columbia Basin Fish and Wildlife Compensation Program 21