Old Franks Fishpasses--1995 Monitoring Update
Prince of Wales Island, Southeast Alaska
John Hannon, Craig Ranger District

The Craig Ranger District fisheries department continued to monitor fish populations associated with the Old Franks fishpass project during 1995. The monitoring program's purpose is to assess the effectiveness of the Old Franks fishpass project and to quantify large scale effects on resident fish populations. The objective of installing fishpasses in the Old Franks system was to increase the number of coho and sockeye salmon available to the common property fishery. The 1995 monitoring activities included the following:
 

o Operation of a weir at the upper fishpass site to validate the accuracy of fish counts made by the electrical fish counting tunnel

o Fish surveys in Lake Mary, small lake south of Lake Mary, and Old Franks Lake

o Snorkel fish surveys in upper Old Franks Creek

o Stream survey of lower Old Franks Creek and tributary stream to the north

o Steelhead escapement counts in Old Franks Creek

o Water temperature monitoring above and below the lake system

Background Information

Since at least the 1960's, the Old Franks system had been identified for a fishpass project to provide access to spawning and rearing habitat for anadromous salmonids. A nearly total barrier waterfall blocked anadromous access at one km above saltwater and another large waterfall is at three km above saltwater. Production estimates for the system were derived for coho salmon (Oncorhynchus kisutch), sockeye salmon (O. Nerka), and a combination of pink salmon (O. gorbuscha) and chum salmon (O. keta). The Forest Service in cooperation with the private landowners and contributors constructed two fishpasses over the two large waterfalls on the system in June and July of 1992. Coho and pink salmon both began using the fishpasses in late July and August of 1992.
 

Stocking of the system dates back to 1952 when sockeye salmon were planted (Table 1). A run was not established due to the barrier falls. This stocking may have established the present kokanee (O. nerka) population in the lakes. King salmon (O. tshawytscha) enhancement was pursued in 1962-63 but was unsuccessful. Sockeye fingerlings were stocked in 1992 in conjunction with the fishpass project but since then sockeye have been unavailable. The Alaska Department of Fish and Game began a coho bioenhancement program in 1993 with egg takes from the Karta system, above Salmon Lake. Coho fingerlings were stocked in both 1994 and 1995 and will be stocked again in 1996. Resident salmonids present above the barrier waterfalls in the system prior to fishpass construction and stocking were: cutthroat trout (O. clarki) , rainbow trout (O. mykiss), and dolly varden char (Salvelinus malma).

Table 1. Stocking History of the Old Franks system

Date	Species	Age	Number	Location Planted	Source
1952	Sockeye	Green Eggs	60,000	Unknown	Buschmann Creek
1952	Sockeye	Eyed Eggs	35,000	Unknown	Buschmann Creek
1953	Sockeye	Green Eggs	165,550	Unknown	Buschmann Creek
1962	King Salmon	Fry	33,750	Lakes	Soos Cr., WA
1963	King Salmon	Fry	46,223	Lakes	Karta (Beaver Falls reared)
5-14-92	Sockeye	Fry	227,200	Lakes	Karta (Klawock H reared)
8-16-94	Coho	Fry	96,632	Upper Lake	Karta (Klawock H reared)
7-21-95	Coho	Fry	78,719	Bridge	Karta (Klawock H reared)
7-26-95	Coho	Fry	139,405*	O.F. Lake	Karta (Klawock H reared)

34,735 cohos were coded wire tagged in 1994 and 32,728 were tagged in 1995.

*9,223 cohos with the same tag code were accidentally released into the Klawock River in 1995.

162,000 coho eggs obtained for 1996 coho release
 

Methods

Site Description

The Old Franks watershed is 32 km east of Craig and covers 65 km². It flows into Polk Inlet on the east side of Prince of Wales Island Alaska. The lakes in the system cover over 300 ha and there are over 19 km of anadromous streams made accessible by the fishpasses (Figure 1). Monitoring work has been conducted primarily on the mainstem of Old Franks Creek both below the lakes (escapement) and above the lakes (juvenile fish) and within the lakes (resident fish).
 

Escapement Counts

A Smith Root Model 1100 fish counting tunnel was installed in the upper fishpass and has been used since 1993 to count returning adult salmon. Periodic snorkel counts from Lake Mary down to saltwater were used to help validate results on the fish counter. We used a fish trap in the exit pool above the fish counter to further validate results in 1995. The trap allowed us to refine the setup and calibration of the fish counter (Appendix contains information on fish counter accuracy trials). Adult salmon captured in the trap were tagged behind the dorsal fin with 30 cm yellow spaghetti tags to enable easy identification of tagged fish coming through the fishpass and entering the upper reaches of the system.
 

Lake Fish Surveys

Lake fish populations were estimated with Schnabel and Peterson mark recapture estimates (Ricker 1975) in 1995. Fyke nets and hook and line were used to capture fish. All fish captured were marked with caudal fin clips and measured to the nearest five millimeters. Fish sampling occurred between June and August in Lake Mary and Old Franks Lakes for cutthroat, rainbow, dolly varden, and coho parr.
 

Stream Fish and Habitat Surveys

Habitat surveys were conducted in 1990, 1991, 1992, and 1994 on upper Old Franks Creek from the upper lake to the top of anadromous habitat. Fish densities (all ages and species) were determined annually since 1990 by snorkeling established habitat units. New snorkel units were marked during each survey to adjust for natural changes in habitat units. The draft Region 10 stream survey protocol (Coghill 1995) was used in 1995 to survey lower Old Franks Creek and the tributary from the north.
 

Results

Coho and Sockeye Escapement into Old Franks Creek

We expected the first increase in coho salmon escapement attributable to the fishpasses to occur in 1995. These returns would be from coho that spawned in 1992 with the offspring smolting in 1994. Returns in 1993 and 1994 averaged around 300 coho. Sockeye from a release in 1992 were also expected to return in 1995. We counted and tagged 343 coho, 31 sockeye, and one ardent pink salmon coming through the fishpass (Figure 2). An additional 29 salmon were incidentally observed jumping over the falls during higher flows. High flows on September 11-12 required removal of the weir pickets. The pickets were replaced September 12, but one loose picket resulted in some fish escaping. The count of 75 cohos for September 12 to October 26 is based only on the fish counter.

During juvenile fish counts on September 22 in upper Old Franks Creek we observed 36 adult coho salmon (four tagged). If this tagged fraction is representative of the entire coho spawning population with a 10% tag loss (no marks from lost tags were noted on any of these fish) then there were over 2,400 coho adults in the system at that time. We assume this is an overestimate of the population. We counted 380 adult coho below Lake Mary on September 8 during a snorkel survey. Less than half the total coho count through the fishpass had passed through the fishpass by September 8. An escapement of around 700 coho may be more plausible.
 

The sockeye return from the 227,200 fingerlings released in 1992 was very low. It is possible that the majority of sockeye will return in 1996 but based on results from Margaret Lake (Bryant etal 1995) it may be that the cutthroat and rainbow in the lakes took advantage of the easy prey or zooplankton populations were not high enough to support the sockeye.

We snorkeled lower Old Franks Creek on April 26 to check steelhead escapement levels and verify whether steelhead are actually present in the system. We conducted the count during a moderate flow caused by snowmelt. The water level prevented viewing of some habitat areas with high water velocity. Two steelhead were observed between saltwater and Lake Mary. This is the first time the presence of steelhead has been verified in the system. Old Franks is not catalogued for steelhead so an update to the anadromous stream catalog will be submitted to the Alaska Department of Fish and Game.
 

Lake Fish Populations
 

We surveyed resident fish (cutthroat, rainbow, dolly varden, and kokanee) in the lower lakes in the system to assess changes in fish populations and size structure following introduction of anadromous salmonids. A continuous mark-recapture procedure was used to estimate population in each lake. Fish were marked with a fin clip and measured to the nearest five millimeters. We captured approximately 1,500 resident fish in the lower lakes using fyke nets and rod and reel surveys. Past fish surveys were conducted in 1978 (Zadina and Haddix 1993), 1990 (Bryant 1991), and 1992. Each survey has occurred between mid-June and August when surface water temperatures are highest. The primary sampling method, fyke nets, sampled only the littoral zone of the lakes so population estimates are considered minimums. Old Franks Lake is the largest lake in the system and is the lake with the most prior resident fish information. Population estimates are available for Old Franks Lake for 1990 and 1995.

Cutthroat

The Old Franks Lake cutthroat length/frequency distribution in 1995 (Figure 3) included larger fish than in past surveys (1978 and 1990) and showed a normal distribution with a peak in numbers at the 290 mm size class. The peak in 1990 was at 170 mm. The average length was largest in 1978 and smallest in 1990. Differences in populations in Old Franks Lake were not significant between any species during 1995 (Figure 4). During the 1978 survey cutthroat and rainbow were not differentiated between but there are two peaks in the 1978 data that may be roughly the peak rainbow and cutthroat size classes. Lake Mary had a very similar size distribution to Old Franks except there was an additional peak at 190 mm (Figure 5). The cutthroat population estimate was higher than that in 1990 on Old Franks Lake but the confidence intervals overlap so differences are not significant.

Coho

This was the first year that coho have been sampled in the lake system. The coho averaged 81 millimeters in length. Using a 70 millimeter cutoff between age 0 and age 1 fish, we estimate that 17% are age 0 fish and 83% are age 1 fish. The age 0 fish are fish spawned in Old Franks and the age 1 fish are a combination of 1994 stocked fish and Old Franks produced fish. The age 1 coho are staying in the lakes but not staying in upper Old Franks Creek in high numbers as evidenced by the pre-stocking snorkel survey. Part of this may be because the coho stocked in 1994 were stocked only into the lake system, at Upper Old Franks Lake. The Old Franks Lake coho length-frequency produced a normal distribution with a peak at 80 mm (Figure 6). The average size of coho is slightly larger in Old Franks Lake than in Lake Mary but the difference is not significant. The population estimate of coho in Lake Mary was nearly the same as rainbow (Figure 7). The Old Franks population estimate showed slightly more rainbow than coho, but the difference is not significant.

Rainbow

Rainbow trout length-frequency distributions in 1990 and 1995 were both bi-modal. The 1995 sampling captured fish that were slightly larger than in 1990 (Figure 8). The 1978 data with rainbow and cutthroat combined showed a peak at 170 mm that appears to match the peak in 1990 and 1995. Rainbows in Lake Mary showed a nearly identical distribution to the Old Franks distribution with the same size ranges. The Old Franks Lake population estimate was higher than in 1990 but the confidence intervals overlap slightly so differences cannot be considered significant. Population estimates were higher for rainbow than for cutthroat in both Old Franks Lake and Lake Mary, but the difference is significant only in Lake Mary. The rainbow population was significantly higher than the dolly varden population in both Old Franks Lake and Lake Mary.

Dolly Varden

During 1995 very few dolly varden were captured relative to sampling effort in comparison with other years. They were fairly evenly distributed throughout the size ranges with no real peaks. The 1990 survey produced many more dolly varden and the distribution was bi-modal with peaks at 130 and 190 mm (Figure 9). The 1978 survey produced larger dolly varden than in other years and again more dollys than in 1995 relative to sampling effort. Dolly varden in Lake Mary were captured in low numbers as well and showed a peak at 90 and at 170-190 mm. Dolly varden in the 230 mm size class have been captured during each survey but the only ones larger than that were two dollys captured in 1978. The low numbers of dolly varden captured are probably due to differences in sampling methodology as indicated by the low recapture rate in 1995. Sampling in 1992 produced higher catch per unit effort than in 1995. Large minnow traps were used in 1992 to sample the lake bottom (Upper Old Franks Lake) and they were successful in capturing only dolly varden. The minnow traps were not used in 1995 because they were ineffective for capturing cutthroat and rainbow.
 

Kokanee

Kokanee have been captured each year that sampling has taken place, but they have never been captured in sufficient numbers to conduct population estimates. No recaptures have ever occurred. The low numbers of captures may be due to the sampling methods used; they do not target kokanee because kokanee primarily inhabit the pelagic zone which is not effectively sampled. Kokanee have never been observed during surveys in upper Old Franks Creek. The mean sizes of kokanee captured each year in the lakes are as follows (all lakes combined): 1978 = 154 mm, 1990 = 152 mm, 1992 = 191 mm (only one kokanee captured), 1995 = 142 mm.
 

Salmonid Abundance and Distribution in Upper Old Franks Creek
 

We snorkeled permanent stations in upper Old Franks Creek before and after stocking of coho juveniles from the Klawock River Hatchery (Karta River Stock). Cutthroat and rainbow cannot be reliably differentiated using snorkel survey techniques. Therefore, whenever cutthroat are mentioned it is also inclusive of rainbow. Prior to the stocking in mid-August, densities of cutthroat, both age 0 and age 1+ were lower than after stocking (Figure 10). Densities by habitat type prior to stocking were not calculated because only 24 of the 60 marked habitat units were surveyed before the stocking. One load of 78,719 coho, averaging 3.25 grams, was stocked into the stream at the bridge crossing on July 21. These fish dispersed downstream to approximately 0.5 km above the upper lake by August 17. They were observed in both the mainstem and off-channel (beaver ponds and tributaries) habitat.
 

Cutthroat and Rainbow

The resident cutthroat/rainbow populations in the stream are much lower than that observed during 1994, but at a similar level to 1991-1993 densities (Figure 11). Mid-channel scour pools, cobble riffles, and plunge pools held the highest densities of cutthroat in the floodplain channel types. Floodplain channel types make up the majority of the mainstem habitat (CRD 1994). The highest densities of cutthroat in the stream were found in the LC2 channel type reach in mid-channel scour and backwater pools (Figure 12). The contained process group (LC2 and MC2 channel types) is bedrock controlled with generally higher water velocities than in the floodplain process group. Cutthroat in the contained channels were generally close to fast water but out of the thalweg where water velocity is highest. Cutthroat showed a preference for the FP5 channel type and the LC2 channel type reaches and a selection against the FP4 and MC2 reaches (Table 2).
 
 
 

Coho

Coho were found almost exclusively in the floodplain process group. The highest densities of coho were in the backwater pool, lateral scour pool, and mid-channel scour pool habitat types (Figure 13). The coho on the mainstem are staying in very tight groups in dense cover. We counted 1,190 coho on the mainstem below the bridge in the sample of habitat units snorkeled (every 5th pool and every 8th riffle). Dispersal upstream was limited with 23 juvenile coho counted above the bridge. Coho showed an obvious preference for the FP5 channel type over all others (Table 2). Some of this preference is because the stream is primarily FP5 immediately downstream from the stocking location at the bridge and they had not yet fully dispersed. Coho were observed in off-channel habitats (beaver ponds and tributary streams) downstream of the bridge but no density or population estimates were conducted.

Populations

Population estimates for cutthroat and coho in the stream were extrapolated from the density and habitat area measurements (Figures 14 and 15). The population estimates are considered minimums. Cutthroat are often missed in snorkel counts because they tend to be more associated with the substrate than coho. The coho were stocked within one month of the snorkel surveys so they were not fully dispersed at the time of the surveys. This caused an underestimate of the population because although some habitat units had very high coho densities most units had no or few coho. The cutthroat were more evenly distributed throughout the habitat types than the coho. Mid-channel and lateral scour pools both had high numbers of coho and cutthroat and appear to be habitats where the two species habitat use currently overlaps the most.
 

Table 2. Upper Old Franks Creek coho and cutthroat estimated population expressed as a percent of the total population by channel type and habitat area by channel type expressed as a percent of the total surveyed habitat area.
 

	FP4	FP5	LC2	MC2
Coho	15%	85%	0.1%	0%
Cutthroat	28%	51%	21%	1%
Habitat Area	51%	38%	7%	4%

Lower Old Franks Creek Salmonid Abundance
 

We snorkeled every fifth macro pool in lower Old Franks Creek to obtain a fish density estimate (Figure 16). We found cutthroat/rainbow/steelhead densities generally higher in lower Old Franks Creek than in upper Old Franks Creek. This could be due to a number of factors: 1) the steelhead run contributes more juvenile fish to the lower stream. We do not know if steelhead have passed through all the lakes and spawned in the upper stream following installation of the fishpasses. Fish densities in 1994 indicated that they may have but densities were lower again in 1995. 2) resident fish in lower Old Franks are more of a stream rearing population than lake rearing because the upper falls has blocked access to the lake system. 3) some cutthroat in lower Old Franks may be anadromous. Some smaller cutthroat may be offspring from these fish.
 

Coho in lower Old Franks Creek were much more evenly distributed than in upper Old Franks Creek. They were present in every unit sampled. This was to be expected because the run is established in the lower stream. Based on past observations, coho densities in lower Old Franks Creek appear similar to other streams in these channel types (LC2 and MM2).
 
 
 

Stream Habitat Surveys
 

We conducted level I habitat surveys (Coghill 1995) in lower Old Franks Creek (saltwater to Lake Mary) and in the main tributary from the north (Old Franks Creek up to Kidney Lake). These stream surveys provide some quantification of habitat conditions in these streams prior to timber harvest in the vicinity. The survey of the stream to the north was conducted from the confluence with Old Franks Creek up to Kidney Lake. This reach is 3.54 km long and includes good overwinter beaver pond habitat for coho as does the 12 hectare Kidney Lake. The surveyed reach contains MM1, PA1, and MC1 channel types. The reach from saltwater to Lake Mary is 3.63 km long and is made up of LC2 and MM2 channel types. This reach is a higher energy channel because of the high degree of channel containment. Habitat parameters are presented in table 3. No ranges of variability have been established for the parameters collected under this survey, but this information provides a baseline of the current habitat conditions in these streams. When ranges of variability are established we will be able to better evaluate habitat quality. Presently the habitat appears to be in good condition for these channel types.
 

Table 3. Level I survey statistics for the Old Franks tributary stream from Kidney Lake and for Old Franks Creek from saltwater up to Lake Mary.
 

C-Type	Pools Per Km	Pool len/ stream length	BFW/ Resid Pool Dep	Average resid pool depth	LWD Cluster #1/km	LWD Cluster #2/km	LWD Cluster #3/km	LWD Cluster #4/km	Side Channel Length	Length of Channel Type, m
		Tributary	From	Kidney	Lake
MC1	41.3	0.33	19.3	23.3	11.8	20.6	5.7	0		339
MM1	58.3	0.51	14.4	31.9	27.6	16.9	3.1	0.8	267	1,304
PA1	22.2	0.50	6.4	49.8	0	0	0	0	16	1,844
		Lower	Old	Franks	Main	Stream
LC2	22.6	0.54	18.9	127.0	37.9	0	0	0.4		2,742
MM2	24.8	0.63	30.8	68.2	22.6	0	0	1.1		886

Water Temperature Monitoring
 

Since 1991, water temperature has been recorded during the summer and fall. The extensive lake system warms the water to sometimes critical temperatures during periods of low rainfall and sunny days. Critical water temperatures have not been documented in the streams above the lakes. The area where high temperatures can affect fish, primarily returning adult salmon, is between Lake Mary and saltwater. During 1995, the water temperature was above 20 C between July 19 and July 22 with a high temperature of 23.5 C (Figure 17). Fish kills were not recorded during this period. Fish kills have occurred in the past when high temperatures coincide with returning salmon. We counted 37 dead coho on 15 August 1994 during a period of high water temperature.
 

Discussion
 

Three years after construction of the fishpasses it appears that coho are readily colonizing the available habitat. The coho run in Old Franks prior to fishpass construction was larger than we originally thought. We estimate that the natural run in Old Franks averaged around 300 coho. The first increase in the coho run due to the fishpasses occurred in 1995. The 1995 escapement estimate of 700 coho is a doubling of escapement compared to 1992 - 1994. This increase is due to natural reproduction from fish that passed through the fishpass in 1992. Coho adults have been found throughout the system; in Kidney Lake, above the small lake south of Lake Mary, and in upper Old Franks Creek. It is expected that coho juveniles will quickly exploit all available rearing habitat. They are now present throughout the lake system and in many off-channel habitats. Stocked coho are expected to start returning in 1996 so a larger run is anticipated. Returns from coded wire tags in 1996 will give a picture of survival of stocked coho and show which fisheries the sytstem is contributing to. Sockeye may take longer to establish than originally predicted as returns from the 1992 stocking were lower than needed to fully seed the system. The fate of stocked sockeye in Old Franks is probably similar to what occurs in Margaret Lake (Bryant et al 1995) where resident cutthroat find the sockeye to be easy prey. In addition, Old Franks has rainbow trout, not present in Margaret Lake, which may add to the predation on sockeye. Old Franks also has extensive littoral habitat where many sockeye-trout interactions probably occur.
 

Passage for pink salmon through the cascade falls between the two fishpasses appears to be flow dependent. Only one pink salmon has been documented through the upper fishpass so full habitat utilization by pink and chum salmon will probably not occur. The majority of pink salmon spawning is currently in the first one km above the lower fishpass. Steelhead have now been positively documented in the lower stream. Population increases may occur, but based on other escapement counts on Craig Ranger District (CRD 1994 and 1995) steelhead populations do not appear to be ultimately limited by habitat availability so predictions are difficult to make.
 

Rainbow populations in the lakes are consistently higher than cutthroat as indicated by catch per unit effort and population estimates. Reasons for the difference in population sizes are unknown. The size distribution of cutthroat has been consistently larger than rainbow throughout the years surveyed.

The entire lake system offers a high quality fishing experience for those willing to make the effort to access the lakes with a canoe. The lakes receive very little utilization by anglers although use will probably increase due to the introduction of coho and the recently completed road which comes to about one km from Upper Upper Old Franks Lake.
 

Resident fish densities in upper Old Franks Creek have been consistently lower than other Prince of Wales streams, although Old Franks is the largest of the streams we have surveyed. Observations in tributaries and palustrine habitats appear to show higher densities of resident fish, especially dolly varden, than the mainstem. However, no quantitative surveys of these "off the main channel" habitats have been conducted. The primary area of overlap of habitat utilization on the mainstem currently appears to be in the floodplain channel types in mid-channel scour pools, lateral scour pools, and some backwater pools. In these areas, competition between juvenile cutthroat/rainbow and the larger and bolder juvenile coho could result in smaller trout. It appears that nearly all rearing after the first winter occurs in either the tributaries or lakes, not in the mainstem. Resident fish less than 130 mm were captured in low numbers in the lakes as well, but this may be due to sampling methods rather than actual lower numbers of fish. In future lake surveys both small and large minnow traps and possibly beach seines could be used to determine numbers of smaller trout in the lakes.
 

Timber harvest in the upper watershed, above the lakes, will occur again in 1997. Currently, there are 14 harvest units of between 9 and 132 acres planned. These units were planned prior to the completion of the Old Franks Watershed Analysis (KTN Area 1994) and many units lie within the riparian habitat conservation areas. The watershed analysis will be taken into account during unit layout, but the recommendations may not be fully implemented in some high sediment transfer hazard areas. This could have some effect on the upper Old Franks stream system.
 
 

Thanks to Herb Roerick, Martin Becker, and John Frank for completing many days and nights in the field for this project.
 
 

Bryant, M.D., S.J. McCurdy, and B.J. Frenette. 1995. The Margaret Lake monitoring program: An assessment of the resident cutthroat trout, dolly varden char, and introduced anadromous salmonids, progress report for 1994. FSL Juneau. 41p.
 

Bryant, M.D. 1991. Evaluation of the zooplankton and resident salmonid populations of Old Franks Lake before the introduction of an anadromous salmonid population. Final Report for 1990. FSL Juneau. 21p.
 

Coghill, K. 1995. Stream survey protocol for the Tongass National Forest. Draft 3-21-95. FSL Juneau. 38p.
 

Hannon, J. 1995. Old Franks watershed: 1994 happenings. U.S. Forest Service. Craig Ranger District.  Craig, AK. 23p.
 

Hannon, J. 1994 and 1995. Steelhead escapement counts in seven streams on southern Prince of Wales Island. unpublished document. USFS Craig Ranger District.
 

Ketchikan Area. 1994. Watershed analysis for Old Franks Creek watershed. USDA Forest Service, Ketchikan Area. 70p.
 

Ricker, W.E. 1975. Computation and interpretation of biological statistics of fish populations. Bulletin of the Fisheries Research Board of Canada. Bulletin 191., Ottawa, Canada. 382p.
 

Zadina and Haddix. 1993. Summary of limnological and fisheries investigations of the Old Franks Lake system 1978-1989. Alaska Department of Fish and Game FRED Division. 37p.
 
 

Fish Counter Accuracy test

Escapement estimates for 1993 and 1994 were made using a Smith-Root Model 1100 Fish Counter in conjunction with snorkel fish counts. The fish counting tunnel is designed to be placed in an area of relatively steady water flow with few bubbles. We installed the tunnel in a horizontal fishpass flume at the top of the upper fishpass.
 

Numbers on the counter sometimes appeared high relative to numbers obtained with snorkel counts. These instances occurred primarily during periods of relatively low flow, when fish would not be expected to be moving significantly. Staffing the camp at the fishpass during the 1995 coho and sockeye run enabled us to assess the accuracy of the fish counter.
 

Fish counts were made using both the weir and the fish counter between June 21 and September 11. The fish counter indicates number of fish passing upstream and number passing downstream. The final counts were made by subtracting downstream counts from upstream counts. During this entire period we caught 367 salmon in the trap in the exit pool above the fish counter while the fish counter indicated that 608 fish passed through. The actual number of fish passing through was 60% of the number on the counter.
 

Counting errors are due to a combination of three factors: 1) Fish are able to hold out inside the tunnel when water velocities are sufficiently low. Over extended periods (hours) this seems to result in additional upstream or downstream counts. 2) Counter sensitivity settings too high can result in counts of smaller fish, debris, or bubbles. Counter sensitivity settings too low can result in missed fish. 3) Flow conditions in the fishpass flume can produce bubbles which result in false counts on the counter.
 

We experimented with a variety of tunnel configurations to improve counter performance. Initially, we tried higher and lower sensitivity settings which did not result in more accurate counts. When fish started coming through in greater numbers we began to observe fish holding out in the tunnel and above or below the tunnel. We then moved the tunnel as far upstream as possible so that after fish passed through the tunnel they were only 0.5 m downstream from the fish trap funnel. This prevented fish from passing upstream through the tunnel and then turning around and going back downstream through the tunnel. It also produced a higher water velocity in the tunnel because most of the flow was funnelled through the tunnel. The problem was that the weir pickets and other roughness elements at the top of the fishpass flume produced bubbles that we were unable to prevent from entering the counting tunnel. The bubbles produced faulty counts.
 

The final tunnel configuration was with the counting tunnel as far downstream as possible so that when fish emerge from the steeppass section they immediately go into the tunnel. We attempted to increase water velocity in the tunnel by putting a 2"x6"x20" board above the upstream end of the tunnel, perpendicular to the water flow. This location eliminated any problem with air bubbles as nearly all were dissipated by the time water got to the tunnel. It did not eliminate the problem of fish holding out inside the tunnel as they were observed holding inside the tunnel. A high flow began two days after the tunnel was moved to this location and this triggered movement of the fish that had been holding out in the stream. During the four day period after the tunnel was moved, September 8-11, we captured 151 salmon in the trap and the counter counted 151 salmon for 100% accuracy. The fishpass was then unstaffed for a four day period and the pickets were removed. When the pickets were replaced one of them was not properly inserted so fish were escaping from the trap for the next four days. Following this, the fishpass was unstaffed for the remainder of the year so we do not have an actual assessment of counter accuracy during a lower flow when fish are not moving so quickly.
 

During 1996 we should be able to further assess fish counter configuration. The current configuration appears to produce acceptable results when the water is high and many fish are moving upstream but is untested during low flows. If flows, staffing and fish timing allow we should be able to determine if an electrical fish counter will be a reliable alternative to a fully staffed weir. An option to be considered is a datalogger attached to the fish counter. The datalogger would show when counts are produced by the counter. This would provide a potential means to determine when faulty counts are produced.