Cheakamus River Project Water Use Plan

Transcription

1 Cheakamus River Project Water Use Plan 2007 Cheakamus River Chum Salmon Escapement Monitoring and Mainstem Spawning Groundwater Survey Reference: CMSMON#1b Cheakamus River Water Use Plan Monitoring Program: 2007 Cheakamus River Chum Salmon Escapement Monitoring and Mainstem Spawning Groundwater Survey Study Period: October to December 2007 Peter M. Troffe, Don McCubbing, Caroline Melville and Jason Ladell InStream Fisheries Research Inc.

2 2007 Cheakamus River Chum Salmon Escapement Monitoring and Mainstem Spawning Groundwater Survey Cheakamus River Monitoring Program #1b Submitted by: Fisheries Research Inc. Peter M. Troffe, Don McCubbing, Caroline Melville and Jason Ladell InStream Fisheries Research Inc

3 EXECUTVE SUMMARY In this inaugural year of monitoring Cheakamus River chum salmon the aggregate recovery rate of spawners PIT tagged at the lower river Stables location (River KM 1.5) was low at 3.8%. As a result, estimates of the total number of spawners upstream of river KM 1.5 was calculated by simple pooled models using the number of known tags applied and ratio of PIT tag detections to the number of net upstream spawners counted at each resistivity counter site. An estimate of the number of spawners above the tag application site (River KM 1.5) was based on a total of 28 PIT unique detections and 6551 spawners counted in the Moody s, Upper Paradise, and BC Rail side channels, for a total detection (recapture) rate of 0.43% tags per spawner. The observed variance in the detection ratio of tagged to untagged fish at different side channels in 2007 ranged from a high in BC Rail channel of 0.84 % (5/595), to a low at Moody s side channel with 0.28 % (10/3629). Using a simple pooled recapture efficiency technique an escapement population of 186,029 (or 179,840 ± 64,212 if using pooled Petersen technique) chum salmon was estimated to be present upstream of River KM 1.5 during the 2007 monitoring period. In addition, we derived an estimate of 16,929 (16,365 if using pooled Petersen) adult chum spawners above the RST juvenile monitoring location based on distribution data from the radio telemetry survey which identified that 9.1% (5/69) radio tagged chum moved to spawning habitats upstream of the juvenile monitoring site. In this first survey year we make the assumption based on previous visual survey observations that all net upstream post-spawned migrants die in the side channels above the counter sites. In future years the PIT antennae arrays will be upgraded to provide detection directionality allowing inference about this assumption by counting the number of tagged out-migrant kelts moving downstream over the resistivity counters. This will allow for differentiation between spawners who remained resident in the side channels and those that out-migrated as kelts. Many technical challenges were evaluated and overcome during the primary year of this study, resulting in proposed revisions to fish tagging locations, capture methodology and enumeration locations. These recommendations should allow for much improved precision on subsequent annual chum salmon escapement estimates

4 ACKNOWLEDGEMENTS We would like thank the following people for their cooperation and help on this project: Ian Dodd, BC Hydro Coastal Generation Brent Mossop, BC Hydro Coastal Generation Carl Halverson, North Vancouver Outdoor School Peter Campbell, DFO Tenderfoot Hatchery Brian Klassen, DFO Tenderfoot Hatchery L.J. Wilson, InStream Fisheries Research Lloyd Buroughs, Riverbank Resources Randal Lewis, Squamish First Nation Josh Korman, Ecometric Research CITATION Troffe, P.M., D. McCubbing, C. Melville and J. Ladell Cheakamus River Chum Salmon Escapement Monitoring and Mainstem Spawning Groundwater Survey; Cheakamus River Monitoring Program #1b. Technical report for BC Hydro Coastal Generation. 42 p. + appendix

5 TABLE OF CONTENTS 1.0 Introduction Background Experimental design Methods Capture upper and lower sites Tagging and release Spawner enumeration, tag recapture, tag recovery Spawning channel and fish enumeration Fish counters and video validation PIT tag detection Radio telemetry, mobile and fixed station Channel walks and tag recovery Escapement analysis Assumptions Results Capture and tagging Sex ratio, spawner condition and length Radio telemetry and spawner distribution Channel walks and tag recovery PIT tag detection Total counts, run time and video validation Escapement estimates Upstream of Stables pool Upstream of Juvenile Monitoring site Discussion Recommendations Literature cited Appendix

6 LIST OF TABLES Table 1 Numerical example to illustrate the monitoring approach and calculations used in the whole river population estimate. Table 2 Number of tags applied, sex ratio and average standard length of male and female chum salmon tagged per week at the Stables location during October 15 November 26, Table 3 Number and fate of chum salmon tagged with radio telemetry tags at the lower river Stables tag application site. Table 4 Proportional distribution of chum salmon tagged with radio telemetry tags among main stem detection locations. Spawner distribution was assessed through a combination of fixed station and mobile tracking data. Table 5 Number and proportion of tags recovered during channel walk surveys for live and dead chum spawners in all side channels during weeks starting October 29, through December 12, 2007 Table 6 Example of PIT detection log data from two unique spawners detected on the Upper Paradise array illustrating multiple detections for each spawner over a 2-3 hour period on two separate dates. Table 7 Number and total proportion of unique PIT tag detections in side channels with the average number of days from spawner tag application to first detection for fish tagged at the lower river Stables and upper river tagging sites. Table 8 Total number of net cumulative upstream chum spawners enumerated at the side channel resistivity counters with correction for counter efficiency. Table 9 Total number and proportion of unique PIT tag detections per corrected net cumulative upstream chum spawner enumerated at the side channel resistivity counters and Tenderfoot Creek trap. Table 10 The monitoring approach and calculations used in the whole river and above RST chum spawner escapement estimate - Cheakamus River

7 LIST OF FIGURES Figure 1 Study area for Cheakamus River chum salmon escapement monitoring with tagging sites, side channel resistivity counter / PIT detection sites, and fixed radio telemetry. Figure 2 The Cheakamus River at river KM 1.5 at the lower river Stables pool tagging locations. Figure 3 Application of a brightly coloured 30 cm spaghetti and 22 mm PIT tag. A radio telemetry tag was gastically implanted into approximately 10% of tagged fish. Figure 4 Upper Paradise side channel enumeration site. Spawners pass through multiple counter chutes with fixed resistivity counter pads and full-duplex PIT detection antennas. Figure 5 Moody s side channel enumeration site. Spawners pass through a single slottype resistivity counter chute with a full-duplex PIT detection antennas. Figure 6 BC Rail side channel enumeration site. Spawners pass through a counter chutes with a fixed resistivity counter pad and full-duplex PIT detection antenna. Figure 7 Cheakamus River relative discharge as estimated at the Brackendale WSC gauge from October 15- December 4, Figure 8 Number of male and female chum salmon tagged weekly at the Stables tagging location during October 15 November 26, Figure 9 Proportional distribution of the condition of chum salmon spawners tagged weekly at the Stables tagging location during October 15 November 26, Figure 10 Weekly averaged counts of live and dead chum spawners counted during channel walks during weeks starting October 29, through December 12, Figure 11 The proportional distribution of peak signal size for up and down counts recorded at BC Rail, Moody s and Upper Paradise spawning channels. Figure 12 Total cumulative daily up and down counts at the Upper Paradise, Moody s and BC Rail fish counters. Tenderfoot Creek counts are total cumulative counts of spawners captured in a trap just downstream of Tenderfoot Lake

8 1.0 INTRODUCTION 1.1 BACKGROUND The Water Use Plan (WUP) for the Cheakamus River (BC Hydro 2005) includes a flow regime for the Cheakamus River designed to balance environmental, social and economic values. One of the fundamental objectives of the Cheakamus River WUP was to maximize wild fish populations, and the WUP recommended an operating alternative and associated river flow regime based in part on expected benefits to wild fish populations. However, the benefits to fish populations from the new river flows were uncertain because they were modeled based on uncertain relationships between fish habitat and flow, and assumed relationships between fish habitat and fish production (Parnell et al. 2003). To reduce this uncertainty, the Cheakamus WUP Consultative Committee recommended a number of environmental monitoring programs. The Cheakamus River chum salmon population was identified during the consultative process as a key-stone indicator species, and the effect of flow on chum salmon spawning and incubation was of particular concern. To reduce this uncertainty, a recommendation was to link adult chum salmon spawner escapement to juvenile out migration data and use the resultant spawner-fry index (H ) as an indicator of flow effects. The potential value of this index was highlighted during an exercise that modeled alternative monitoring designs (Parnell et al. 2003). BC Hydro has monitored Cheakamus River juvenile chum fry out-migration for the last eight years (see Melville and McCubbing ) and monitoring of out-migrant fry is continuing at several locations. However, no accurate adult chum salmon spawner escapement data exists for the Cheakamus watershed and the linkages between adult escapement and juvenile outmigration are currently poorly understood. Another important uncertainty during the consultative process was the relation between river discharge and groundwater upwelling in mainstem spawning areas. The effective spawning area Performance Measure for chum salmon and other salmon species was influential in the selection of flow alternatives during the consultative process. The performance measure was calculated using a model based on River 2-D simulations, depth, velocity and substrate preference curves, and redd stranding calculations. This model identifies those areas where spawning is likely or unlikely to occur based on depth, velocity and substrate criteria, and thus the approach will tend to overestimate the area of spawning habitat relative to empirical measures (Parnell et al. 2003). The model does not predict the precise location of spawning. Thus, the model is useful for comparing alternative flows, but does not provide precise measures of spawning habitat. Modeling suggested that lower and more stable flows during the fall (relative to the existing Interim Flow Order) would provide a larger area suitable for spawning that would also remain wetted during incubation, resulting in relatively greater effective spawning area. These findings and the modeling approach in general, was uncertain because chum spawning habitat selection can also be driven primarily by groundwater upwelling, and not the surface flow characteristics of water depth/velocity and spawning gravel suitability. It was suggested by some committee members that lower flows during the fall spawning period would result in reduced surface water-to-groundwater exchange, reduced - 8 -

9 upwelling, poorer spawning site selection and thus lower chum egg to fry survival, and that the River 2-D modeling had greatly overestimated suitable spawning area under low flows. This monitor was developed to examine the effects of the WUP flow regime on chum salmon spawning in the mainstem of the Cheakamus River and major side channels and includes two components: i) Estimating annual escapement of adult chum salmon in the Cheakamus River, and distribution within the mainstem and off channel habitats. ii) Examining the relation between discharge, groundwater upwelling, and the selection of spawning habitat by chum salmon in the mainstem. Data from this study will also be used in conjunction with data from other monitoring programs to develop stock-recruitment relationships that are critical for separating effects of spawning escapement from flow-related changes in survival during incubation. The key water use decision that would potentially be affected by the results of the monitoring is the seasonal flow release from the Daisy Dam, in particular, releases during the chum spawning and incubation period. Such changes would affect power generation and other social and environmental values in the Cheakamus River. 1.2 EXPERIMENTAL DESIGN There are many challenges to estimating chum escapement and spawning distribution in the Cheakamus watershed due to its size and environmental conditions. Observations of considerable downstream movement of spawned-out moribund fish among mainstem spawners combined with restricted water visibility and poor access to some river/channel reaches when river discharges are high (see: Melville and McCubbing 2000; Korman et al. 2002) create challenges for traditional visual tag mark recapture approaches that are commonly employed in smaller coastal systems. Traditional visual mark recapture escapement surveys involve tagging salmon with external tags followed by detailed foot carcass surveys of all possible spawning grounds. To be robust, this type of survey generally would require tagging of upwards of 10 per cent of the total estimated population (historically estimated at > 80,000 chum, DFO data on file; DFO 1957). Instead, this monitor uses a passive mark recapture technique in place of a traditional mark recapture carcass recovery or visual estimation study methods. This passive tag recovery approach involves the use on side channels of fixed location resistivity fish counters to enumerate all fish entering the side channel, coupled with PIT (Passively Integrated Transponder) scanning tag readers to scan for tags on all fish entering the side channel. PIT tags are small sealed electronic modules with unique identification codes that can be implanted in, or externally attached to juvenile and adult fish. Fixed station river pass-through antennas passively monitor movements of fish with tags and record data with logging equipment. PIT technology has many advantages over externally mounted visual tag techniques and has been extensively used as an accurate adult and juvenile salmonid monitoring tool - 9 -

10 since the mid 1980s in the Columbia River basin (e.g. Zydlewski et al. 2006; Prentice et al. 1986; Prentice et al. 1990; McCutcheon et al. 1994; Downing et al. 2001; Matter and Stanford 2003) and is currently used in a wide variety of aquatic and terrestrial monitoring programs worldwide (see: biomark.com for a bibliography and Thorsteinsson (2002) for additional references). This study employs a single mark, multiple recapture location mark-recapture design (e.g. Schwarz and Taylor 1998). PIT tags were applied to adult chum salmon through out the spawning season downstream of the Cheekeye River confluence with subsequent detections of tagged and untagged fish at all side channel complexes with sizable chum spawning habitat. In addition radio telemetry was used to determine spawner distribution upstream and downstream of the current juvenile out-migration monitoring site as well as assisting in evaluating spawner residence time during the initial four years of the monitor. This approach simultaneously parses mainstem and individual side channel spawning populations based on the ratio of tagged fish detections to untagged fish detections and allowed for detailed collection of run time data in each side channel complex. 2.0 METHODS 2.1 CAPTURE - Upper and lower sites We selected a tagging site based on a the following factors: i) the requirement to apply tags in the lower reach of the river downstream of major spawning areas, in order to estimate escapement for the majority of spawners in the Cheakamus River, and ii) site specific characteristics for access, fish capture and processing. One suitable site was selected, located upstream of the Cheakamus/Squamish confluence at River KM 1.0, commonly known as the Stables pool (Figure 1). Data from this site were used to provide an estimate of the majority of chum spawners in the Cheakamus River as suitable spawning/incubation habitat downstream of this area is typically limited and of poor quality, suffering from high bed-load movement and siltation from the Cheekeye River, though spawning was observed downstream of the Stables pool during river drift surveys. Fishing effort directed at the capture of chum salmon for tag application was conducted during daylight hours from mid October through late November, 2007 at the primary Stables pool location in the lower river. The majority of the tagging occurred at this lower river capture and tagging location (>90% of effort and capture). However, due to in-season observations of limited chum spawner distribution upstream of Moody s channel via visual and radio telemetry observations and low PIT tag detections in the spawning channels during the study, a second experimental upper river capture site was evaluated in the mainstem Cheakamus upstream of the Moody s channel confluence near River KM 4.0 (Figure 1). Fish were only captured and tagged at this location on one day, November 15, Chum salmon were captured using 18 x 4.5 m or 13.5 x 3.6 m tangle type floating gill net hung with 15 cm stretched length Alaska twist tangle mesh. As often as river conditions

11 were appropriate (discharges < 45 CMS at Brackendale WSC gauge), a two person crew deployed and drifted a tethered net from a small pontoon raft at the upstream section of the fishing run, while another shore based two person crew walked the tethered line through the 120 m run to the bank side landing location whereupon any captured fish were quickly placed into floating fish tubes for holding before processing (Figure 2). While we were targeting chum spawners, all species captured were recorded. Fishing effort was recorded as the number of fish captured during each standardised net set

12 Figure 1 Study area for Cheakamus River chum salmon escapement monitoring (River KM ) with tagging sites, side channel resistivity counter / PIT detection sites, and fixed radio telemetry receiver locations

13 Figure 2 The Cheakamus River at river KM 1.0 at the lower river Stables pool. In the foreground are fish holding tubes, the tagging cradle table, one of two holding pens. In the back ground the crew of four deploys the tangle net from a small raft. Prior to tag application, fish were removed from holding tubes, sexed, and measured for standard fork length. To increase likelihood of tagging spawners destined for upstream migration body condition was assessed according to a five point scale and tags were only applied to the high condition spawners categorized as Condition 0, 1 or 2 using the following criteria: Condition 0 and 1 - fish were silver uncoloured spawners, which appeared to have entered the river recently and Condition 0 fish displayed sea lice on their opercular flap. Condition 2 fish exhibited some spawning colouration, but were in fresh condition and free body decay. Condition 3 fish clearly display spawning colouration and are showing early signs of body decay. Condition 4 fish are heavily coloured, have some body deterioration, and may show signs of previous spawning activity

14 2.2 TAGGING and RELEASE All Condition 0, 1 and 2 chum salmon were placed in a portable tagging cradle with the dorsal surface exposed and tagged through the leading edge of the dorsal fin with a uniquely numbered 30 cm external nylon spaghetti tag (Floy Ltd.). Tag colours were changed weekly to assist with run time estimation during visual surveys. In addition to the visual spaghetti tag, each fish was implanted with a 22mm 1420 SX Destron-Fearing khz full-duplex glass encapsulated PIT tag into muscle tissue on the lateral surface just below the dorsal fin (Figure 3). Approximately every tenth tagged fish was also gastrically implanted with a 90 day life span Lotek radio tag (model MCFT-3A in a methodology similar to Brown and Eiler (2005). Tagging time, from holding tube removal through tag application to placement in the recovery pen was usually less than one minute. Tagged fish were held in two 2.5 x 2.5 m recovery pens and released once the day s fishing and tagging sessions were complete. Tagged spawners remained vigorous after tagging and no recovery problems were observed during the tagging potion of this survey. Figure 3 Biometric data is recorded once the chum salmon is placed in the table cradle followed by application of a brightly coloured 30 cm spaghetti and 22 mm PIT tag. A radio telemetry tag was gastrically implanted into approximately 10% of tagged fish. 2.3 SPAWNER ENUMERATION, TAG RECAPTURE, and TAG RECOVERY Briefly, the enumeration technique involved the use of three full spanning fish fences at the lower reaches of different side channel sites fitted with one or more fish counters and PIT antenna arrays at openings in the fences (see Figures 4, 5, 6). The fish counters and PIT receivers continuously monitor and log the number of tagged and untagged spawners entering each side channel

15 2.31 SPAWNING CHANNEL - FISH ENUMERATION Tag recapture through PIT tag logging and spawner detection through resistivity counter monitoring and trap operations were conducted from October 15 through to December 17, 2007 in the lower reaches of side channels and included: Upper Paradise spawning channels, BC Rail channel, Tenderfoot Creek (DFO trap), and Moody s channel (Figure 1). Downloads of electronic equipment were conducted up to twice a week and each site was visited daily by maintenance crews to monitor any debris build-up along fish fencing FISH COUNTERS, TRAPS and VIDEO VALIDATION The primary method for evaluating the numbers of spawning salmon entering side channels was a resistivity fish counter. A resistivity fish counter operates by detecting the change in resistance caused by a fish as it passes a fixed point and close to a set of electrode sensors submerged in water. The change in resistance observed occurs because the fish is more conductive than the water it is displacing and therefore allows a slight increase in conductance while present between a pair of electrodes. The electrode sensors in any resistivity counter are designed to encourage migrating fish to pass close enough to the sensors to be detected and in a uniform manner, such that each fish passage can be recorded consistently. The Logie 2100C fish counter uses these changes of electrical resistance between electrodes pairs caused by fish passage to provide counts. The date, time, conductivity, channel, direction of movement (up or down) and peak signal size (PSS) are recorded by the counter when a change in electrical resistance above threshold setting is encountered. If a change of resistance occurs which is not interpreted as a fish count by the counter s fish algorithm, the direction of count is substituted with the character E which denotes an unclassified event. Such events may be fish which have been miss-classified, or failed to pass completely over the counter as well as debris flow, and air entrainment noise (Aprahamian et al. 1995). To each change of resistance the counter assigns a peak signal size which relates to the maximum deviation from baseline resistance observed during the event. PSS is a function of the fish size, counter gain setting (electrode sensitivity), river conductivity conditions and of the channel bulk resistance (a measure of the instantaneous background resistance created by water flowing over the electrodes). To avoid collecting a multitude of events with low PSS due to background noise a threshold PSS is selected for each channel and each type of counter record. The counter is then able to evaluate records which are at least 0.5 seconds apart and can enumerate fish passing over all enabled channels simultaneously, up to a maximum of four per unit. The Logie counter is designed to re-calibrate every 30 minutes for changes in bulk resistance and conductivity. These calibrations alter the gain (sensitivity) setting so that a fish of a standard size will be attributed a similar PSS, under a wide range of environmental conditions. Data are stored on the fish counter memory and downloaded periodically by laptop computer

16 SITE SPECIFIC DESIGN AND SETTINGS Resistivity fish counters were used for monitoring chum spawners in the Upper Paradise (100 m upstream of mainstem confluence), Moody s (1200 m upstream of mainstem confluence) and BC Rail (200 m upstream of mainstem confluence) channels and a trap was used to enumerate spawners at Tenderfoot Creek near the inlet to Tenderfoot Lake (700 m upstream of mainstem confluence) (Figure 1). Each side channel had a different design tailored to the channels environment and discharge regime during the spawning season. Briefly, the Upper Paradise spawning channel counter consisted of four pseudo Crump weirs of 12 cm height and 60 cm width, affixed to a sill constructed across the channel bed. Into these and in a high density polyethylene (HDPE) sheet were set three stainless steel electrodes (12 by 4mm) at 30 cm spacing. These electrodes were connected to the Logie 2100C counter by copper wire. The weirs were placed at the leading edge rather than the rear of the sill as originally planned. This was undertaken to avoid flow induced air entrained noise, which on testing resulted from previous sill slumping (Figure 4)

17 Figure 4 Upper Paradise side channel enumeration site. Spawners pass through multiple counter chutes each with a fixed resistivity counter pad (white, upstream end) and full-duplex PIT detection antennas (black, downstream end)

18 The Moody s side channel sensor electrodes (of the same type as in Upper Paradise) were contained within a single 1m long and 1.5 m high U shaped slot type channel of 30 cm width. The electrodes were placed at 30 cm centres on the base and up the sides of the channel. Fish passage at the counter location was restricted to the sensor area by an Alaskan style picket fence (Figure 5). Figure 5 Moody s side channel enumeration site. Spawners pass through a single slottype resistivity counter chute with a full-duplex PIT detection antennas BC Rail channel sensor electrodes were placed on the base of a 60 cm wide, 2.0 m long flume. The sensor unit was placed flat on the base of the flume and consisted of electrodes set in HDPE as in Upper Paradise. Fish were directed through the flume by means of an Alaskan picket fence (Figure 6). Spawners were visually enumerated at Tenderfoot Creek through capture with an aluminum vee-type slot trap near Tenderfoot Hatchery (Figure 1). An experimental floating PIT tag antenna was placed immediately downstream of the trap to detect tagged fish. Each day the number, sex and presence of any Floy spaghetti tagged chum spawners were visually assessed by hatchery staff and recorded before the fish were released through an upstream trap gate to spawning habitat located in the groundwater fed Tenderfoot Lake

19 Figure 6 BC Rail side channel enumeration site. Spawners pass through a counter chutes with a fixed resistivity counter pad and full-duplex PIT detection antenna. Conductivity calibration was not utilized at any site in 2007 as conductivity was expected to vary little and was low (circa 50 μs), resulting in the counter generating large peak signal sizes for fish passage while utilizing a predetermined fixed gain setting of 100. In this study we were not attempting to evaluate the ability of the counter to determine fish size, rather just fish passage. As such, fine hourly adjustment of gain settings to calibrate the relationship between fish size and PSS size were not required. In this study and although each sites electrode arrays were of different designs a minimum threshold PSS of 30 (on a scale of 1-127) was selected in 2007 for both upstream and downstream counts and events. This threshold was visually observed to minimize background noise triggers while evaluating all fish passage. Lower threshold levels while allowing for the potential enumeration of smaller fish, tend to pick up resistance noise created by water turbulence and entrained air bubbles so are best avoided. As our target species were adult chum salmon with weights in excess of 3.5 kg all fish created PSS well in excess of this threshold as observed visually and by remote video

20 VIDEO VALIDATION Counter data obtained at Upper Paradise were analysed in relation to video footage recorded using a digital video recorder (Capture DVMS 400) linked with infra red microcameras, using the methods of set up as described in Aprahamian et al. (1995). Similar studies in the United Kingdom and British Columbia (Fewings 1987; Welton et al. 1987; Dunkley 1991) have demonstrated the utility of this video validation methodology. Counter efficiencies were based on the number of fish viewed passing completely over the counter in relationship to the number correctly assigned as upstream or downstream counts by the electronic counter. This Logie counter was programmed to trigger recording when an event, upstream or downstream record was observed. Video records were then compared with counter records to establish a counter efficiency PIT TAG DETECTION To detect PIT tags applied to upstream migrant spawners, full-duplex PIT tag detection and logging equipment comprised of Destron-Fearing FS khz readers/loggers and 0.5 X 0.5 m Biomark Inc. pass-through river antennas. These were deployed concurrently with each fish counter channel such that each upstream migrant spawner would be monitored by both the PIT antenna and the fish counter (Figures 4, 5, 6). As for the Logie fish counters the PIT array and loggers were operated continuously through the monitoring period. Each PIT antenna and receiver was individually tuned to reduce any background signals and periodically tested by floating a drone tag taped to a 3m piece of twine up and down through the detection field to confirm a cm tag detection range RADIO TELEMETRY MOBILE and FIXED SITES Three directional fixed station Lotek W31 radio receivers and one mobile Lotek radio tracking unit were used to survey the side channels and mainstem habitats to determine spawner distribution and to assist in evaluating spawner residence time data. Fixed station logging receivers were located at the juvenile monitoring (RST) site, 50 m downstream of the Bailey bridge, and downstream of the Cheekye River confluence near the Sunwolf Recreation Centre (Figure 1). Mobile tracking was performed by foot and raft once per week from 200 m upstream of the Bailey bridge (River KM 7.0) downstream to the Cheakamus River confluence (River KM 0.0) to assess spawner movements between fixed telemetry stations and to monitor possible downstream movements of fish out of the Cheakamus River CHANNEL WALK ENUMERATION and TAG RECOVERY Channel walks were conducted by a three to four person field crew twice a week during the October 15 through December 23, 2007 survey period. The intent of the channel walks was to visually estimate and tally the total number of live, dead, and tagged chum spawners in all assessable portions of spawning habitat upstream of the fish counter and PIT tag detection sites. The areas surveyed include:

21 Upper Paradise channel upstream from counter to Sue s Channel, Kisutch Channel, and the Gorbuscha Channels. BC Rail channel upstream from Tenderfoot Creek outlet culvert through to Dave s Pond. Moody s channel upstream of counter site through to Big House and beyond Canoe Pond to channel end. 2.4 ESCAPEMENT ANALYSIS Escapement estimates of chum salmon spawners into the Cheakamus River are required for hypothesis testing at a variety of levels. Our study aims to provide three key estimates of spawner abundance: 1) A whole river chum salmon spawner estimate this estimate accounts for all spawners upstream of river KM 1.5 the Stables tag application location, including all side channel complexes (e.g. Upper Paradise, BC Rail, Tenderfoot, Moody s, see: Table 1 for example). This estimate provides a density dependent context for year-to-year comparisons of chum spawner escapement to assess if spawning distribution varies with fish density. 2) An estimate of chum salmon spawning upstream of RST juvenile monitoring site using spawner distribution information from telemetry survey assign a proportion of the whole river estimate to the sections of river upstream of the RST juvenile monitoring site. 3) Estimates of individual spawning channel chum escapement on BC Rail, upper Paradise and Moody s channel to complement trap counts from Tenderfoot Creek. To determine the actual number of chum salmon arriving back to the watershed to spawn, in a given sample time period, a known number of marked fish are released into the population downstream of the side channel enumerating locations with the assumption that these fish will move upstream past the enumeration station (resistivity counter with PIT tag receiver or manual trap) and that a portion of these fish will be recaptured (i.e. reobserved). Assuming that fish do not lose their marks before recapture, that no marks are missed during sampling, and that the chance of capturing any marked fish is equal to unmarked fish, the efficiency of a capture trap on sampling marked fish can be calculated for a given time period (Seber 1982; AFS 2007). Combined with these data, when the total number of unmarked fish are also evaluated at the same locations, it is then possible to statistically model the numbers of total fish in the study population from which the sub-sample was derived (see: equations below for Pooled Petersen estimator or Table 1 for an example from the monitors terms of reference for the simple mark recapture models considered herein)

22 Pooled Peterson population estimates can be calculated from the basic mark recapture equation provided by Ricker (1975): Where N = population estimate C = total catch N = (M+1)*(C+1) + (mortalities) (R+1) R = number of marked fish recaptured M = number of marks released If random mixing of marked and unmarked individuals is assumed, then the variance of recovered marks has a binomial distribution. In these cases it is best to obtain approximate confidence intervals from a table or equations that approximate the binomial distribution using recovered marks as the key parameter. Ricker (1975) derives the confidence intervals for N in large sampling regimes (>25) as in equation 1 as approximately equal to: R(V) = R ± 1.96 (R+1) Where V = the variance of R R = number of recaptures By substituting the upper and lower calculated values of R (equation 2) the confidence limits for Peterson population estimates can be derived. Factors which may confound such estimates include, the death of marked fish prior to spawning, loss of marked fish to predators prior to an opportunity for recapture, recycling of marked or unmarked fish through open enumeration sites (excludes fixed traps) or marked fish bypassing the enumeration site without the potential for capture (AFS, 2007; Frith et al. 1995). Our study design utilizes best practice methods to minimize the risk of challenging these assumptions, which include but are not limited to: marking fish while minimizing handling stress, remote sensing of tags while fish are completing their spawning migration and prior to possible predator removal, providing gated validation of recycling rates and providing 100% stream coverage for marked and unmarked fish enumeration and tag scanning. Variations in observed recapture rates amongst different recovery locations indicate differential susceptibility of tagged fish to being recaptured at the counter/trap sites. These variances are expected to be high where sample size is low and may be the result of equipment efficiency or incomplete mixing of tagged and untagged cohorts of fish

23 Table 1 - Numerical example to illustrate the originally proposed monitoring approach and calculations used in the whole river population estimate (1200 PIT tags applied). Moody s spawning channel 45 PIT detections per 1,500 net upstream counts BC Rail Creek 50 PIT detections per net upstream 2,600 counts Tenderfoot Creek 178 PIT detections per 12,000 net upstream counts Upper Paradise spawning channel 276 PIT detections per net upstream 22,000 counts 1. Total of 549 PIT detections per 38,100 spawners counted= 69 spawners per detection 2. Assume equal mix of tagged and untagged fish in side channel and mainstem tags applied 549 detected = 651 PIT tagged individuals not detected in side channel complexes x 69 spawners per detection = 44,919 mainstem fish 5. Total population estimate: 38,100 side channel + 44,919 mainstem = 83,019 spawners In this spawner survey, PIT tag detectors, resistivity counters and a spawner trap monitors different side channels for each fish entering the spawning habitat. By design, the resistivity counter allows fish to move freely upstream and downstream over the directional counter electrodes. Based on the literature for chum salmon spawning behaviour, we expected a directional upstream migration in to the channels, where chum would spawn and die, and carcasses would remain in the channel. In this simple case for fish moving in a single direction upriver to spawn the total number of up counts is the total spawning escapement to that channel. However, we also suspected that spawners may also move downstream past the counters, and some carcasses may move downstream, particularly in the variable flow in Moody s channel. In this case, the total up counts can be subtracted from the total down counts resulting in the number of net upstream spawners. Before the net number of upstream spawners is calculated, the total numbers of up and down resistivity counts are corrected for counter efficiency through video validation. The counters cannot identify specific individuals to determine when an individual fish moved upstream then downstream past the counter, or if individuals were alive or dead. The PIT readers detect individual fish by their unique tag code when they pass by (or are within the detection range of) the readers. However, the single antenna PIT readers do not give the direction of movement. A large number of downstream counts were enumerated in the counters (see Results). For these downstream counts, we could not distinguish the downstream movement of a fish after spawning (kelt), vs. pre-spawning movements and other migratory behaviours. Therefore, in this first survey year we make the conservative assumption that spawning escapement is equal to the net upstream count. In future years the PIT antennae arrays will be upgraded to provide directionality allowing calculation of

24 residence time of individual fish upstream of the counter, and inference about this assumption by evaluating if down counts are likely to be kelts and in what quantity. This will allow differentiation between kelted spawners and fish which may recycle across the counter prior to spawning. Radio telemetry provides inference on spawner distribution and, relative to the PIT detection data, an independent verification of spawner distribution between mainstem and side channel spawning areas, and distribution, upstream of the RST juvenile monitoring site ASSUMPTIONS Mark-recapture assumptions During our analysis we assume that: 1. The population is closed during the period of the study. For adult spawners death and emigration affect tagged and untagged fish equally, and that all components of the population are vulnerable to either capture or recapture. For this assumption to be valid, it is critical that marks be applied to chum salmon throughout the entire period of adult migration, and that tagged individuals are well mixed within the population at time of recapture. Spawners were tagged in the lower reach of the river to promote equal mixing and tag application was conducted throughout migratory period except when river discharges were too high to allow for fish capture. 2. Tagged and untagged fish are correctly identified. If tagged or untagged fish are not detected, the proportion of tagged fish is underestimated in recapture samples, and population abundance is overestimated. The detection efficiency of resistivity counters has been demonstrated to be >90 per cent and of low variance in several other river systems in British Columbia (McCubbing et al. 1999; McCubbing and Ignace 1999). Remote scanning PIT read detection efficiencies ranging from 88 to 100 per cent, have been reported in the literature with efficiency largely dependant on antenna design and migration aperture. Most studies observed detection efficiencies of >95 per cent (Prentice et al. 1990; McCutcheon et al. 1994; Castro-Santos et al. 1996). 3. No tags are lost. If tags are lost (due to poor application technique or aggressive behaviour during spawning), the proportion of tagged fish will be also underestimated in the recapture samples, and population abundance will be overestimated. For visual tags, Schubert et al. (1996) found loss rates from 0 to 2.7 per cent from adult pink salmon, but tag loss of up to 30 per cent has been documented in adult chum salmon (Lister and Harvey 1969). Retention of PIT tags is usually extremely high (96.6 per cent in juvenile brown trout, Salmo trutta, Ombredanne et al. 1998; 99 to 100 per cent in chinook salmon, Oncorhynchus tshawytscha, Prentice et al. 1990). Most salmonid studies indicate PIT tag loss is lowest (< 2 per cent) when tags are

25 properly positioned in the peritoneal cavity. (Prentice et al. 1990; McCutcheon et al. 1994; Buzby and Deegan 1999; Dare 2003) and recent advances in radio frequency identification (RFID) technology have produced larger glass encapsulated tags that might be capable of oralgastric placement similar to radio tags used in telemetry surveys. 4. Tagging does not change the availability of fish for detection. The stress of capturing, holding and marking fish could lead to behavioural changes which affect a fish s ability to swim upstream, change its availability for detection, or in some cases even cause mortality. Such effects would again cause an underestimate of the percent of fish tagged, and an overestimate of population abundance. Visual surveys provide some inference on behaviour of both untagged and tagged spawners. 5. Tagged and untagged fish have an equal probability of initial capture and detection. This assumption is generally violated to some extent in all mark-recapture studies (Otis et al. 1978), but can be minimized by making tag application and recovery as representative as possible, through standardized effort and the use of gear with minimal selectivity. 3.0 RESULTS PIT detection arrays and resistivity counters are passive type technologies which are, in this study, located in very close proximity to each other by design to maximize the likelihood that tagged and untagged fish are detected equally. 3.1 CAPTURE AND TAGGING Tagging effort at the lower river Stables location was directed over a total of 31 fishing days during October 15 through November 30, The best opportunities for fishing were related to river discharge and more fish and higher catch per unit effort (CPUE) were encountered when river discharges were falling after a period of recent increased discharge conditions, especially during the two weeks starting October 22, 2007 and November 12, 2007 (Figures 7, 8). A total of 795 (446 Male, 349 Female) chum salmon were tagged with PIT and visual coloured Floy tags, and of these fish 69 (37 Male, 32 Female) were gastrically implanted with radio telemetry tags (Table 2). Fishing effort was directed for one day, November 15, 2007, at the upper river tagging locations just downstream of Moody s channel and a total of 75 (45 Male, 30 Female) fish were PIT and visually tagged, and of those tags 7 fish were implanted with radio tags

26 Figure 7 Cheakamus River relative discharge as estimated at the Brackendale WSC gauge from October 15- December 4, (N.B. periods of gauge outages during October 27, and November 15, 2007). 100 # MALES TAGGED # FEMALES TAGGED # of spawners tagged per week Oct. 15 Oct. 22 Oct. 29 Nov. 5 Nov. 12 Nov. 19 Nov. 26 Figure 8 Weekly number of male and female chum salmon tagged at the Stables tagging location during October 15 November 26,

27 Table 2 Number of PIT, radio tags (RT) applied, catch per unit effort (CPUE) expressed as the number of fish tagged per number of weekly net sets, sex ratio, and average standard length of male and female chum salmon tagged per week at the Stables location during October 15 November 26, Tagging date FEMALE MALE SEX RATIO CPUE FEMALE MALE M:F # TAGGED/ LENGTH ± LENGTH ± S.D. (weeks start) # TAGGED / RT # TAGGED / RT # WEEKLY SETS S.D. (mm) (mm) 15 Oct / 3 37 / ± ± Oct / 7 92 / ± ± Oct / 6 85 / ± ± 6 5 Nov / 2 39 / ± ± Nov / 4 54 / ± ± Nov / 8 93 / ± ± Nov / 2 46 / ± ± 5 TOTALS 349 / / SEX RATIO, LENGTH AND CONDITION Sex ratio of chum salmon captured for tag application was generally skewed towards male, but the difference in sex ratio moved towards being neutral toward the end of the tagging period as more females moved into the fishing site (Table 2, Figure 8). The average standard length of males was approximately 50 mm larger (801 ± 20 mm S.D.) than females (749 ± 34 mm S.D.) and the average sizes of male and female fish were consistent from week to week through the tag application period (Table 2). High condition fish were targeted for tag applications and the majority (>70 %) of chum salmon captured were in Condition 1 and 2, although during the last week of tagging sessions approximately 29 % of fish encountered were in Condition 3 (Figure 9)

28 100% Condition 0 Condition 1 Condtion 2 Condition 3 Condition 4 75% Proportion of catch 50% 25% 0% Oct. 15 Oct. 22 Oct. 29 Nov. 5 Nov. 12 Nov. 19 Nov. 26 Figure 9 Proportional distribution of the condition of chum salmon spawners tagged weekly at the Stables tagging location during October 15 November 26, RADIO TELEMETRY and SPAWNER DISTRIBUTION All of the 69 radio tagged fish moved upriver and were detected at least one time at the Cheekeye confluence receiver and of those fish enough distributional data was collected from 54 spawners to estimate the spawner distribution upriver of the tagging site (Tables 3 and 4). A Squamish Nation chum salmon food fishery occurs annually near the Moody s side channel confluence (River KM 3.5) and we received anecdotal information from fishers that a small number of radio and PIT tagged spawners had been encountered in harvesting nets and it is unclear how many of these fish were harvested and the condition of any released fish is unknown. The vast majority (90.9%) of radio tagged fish were identified as spawning in a section of river stretching from the Cheekeye River confluence upstream past the Moody s side channel confluence (River KM ) but below the RST juvenile monitoring site. The remaining 9.1% of radio tagged spawners spawned between the RST juvenile monitoring site and a section of river 500 m upstream of the Bailey bridge (River KM ) (Table 4). Two spawners with radio tags were detected by PIT receivers in the Upper Paradise side channel

29 Table 3 Number and fate of chum salmon tagged with radio telemetry tags at the Stables tag application site. Sex # Telemetry tags Detected at Killed and applied at 'Stables' Cheekeye at least once reported by fishers Females Males Total Table 4 Proportional spawning distribution of chum salmon tagged with radio telemetry tags among main stem Cheakamus River detection locations. Spawner distribution was assessed through a combination of fixed station and mobile radio tracking.. Sex Between Cheekeye confluence and Moody s confluence (RKm 3 4) Between Moody's Confluence and RST site (RKm 4 5) Between RST monitoring site RST site (RKm 5 6) Upstream of Bailey bridge (RKm 7+) Females 16 (29.1%) 6 (10.9%) 2 (3.6%) 0 (0%) Males 23 (41.8%) 5 (9.1%) 2 (3.6%) 1 (1.8%) Total 39 (70.9%) 11 (20.0%) 4 (7.3%) 1 (1.8%) 3.3 CHANNEL WALKS The portions of BC Rail, Moody s and the Upper Paradise channels which were assessable by foot were surveyed by three to four person crews approximately twice a week through the October 15 December 12, 2007 survey period. The total number of live and dead chum salmon spawners in each channel was recorded during each foot survey and any tags recovered or observed were noted. Counts of live chum spawners increased through early November with numbers peaking for the BC Rail and Upper Paradise channels during the week starting November 12, and a week later for spawners in Moody s channel complex (Figure 10). Carcass counts slowly increased through early November, and numbers for all channels peaked in the weeks through November 20-30, approximately one to two weeks after peak live spawner counts were observed, except in Moody s channel where they were coincident (Figure 10)