Stream Habitat and Fish Distribution in the Upper Old Franks Watershed, Prince of Wales Island
John Hannon, Craig Ranger District

Craig Ranger District personnel completed a basin wide survey of anadromous fish habitat in the Upper Old Franks Creek watershed during June and July of 1994. The Ketchikan Area and Regional Watershed Assessment Team chose Upper Old Franks as a site for watershed analysis. The basin wide survey was conducted to identify how Old Franks habitat compares to other southeast Alaska streams, ie. Regional habitat management objectives. Previous habitat surveys were conducted in 1990, 1991, and 1992. Snorkel fish surveys have been conducted yearly since 1990. These surveys were conducted primarily for monitoring changes in salmonid abundance and distribution following installation of fishpasses. These fishpasses give coho and sockeye salmon access to the Old Franks stream and lake system. This report documents the results of the 1994 fish and habitat survey.
 

The Old Franks watershed lies 32 kilometers east of Craig. It flows into Polk Inlet on the east side of Prince of Wales Island. The watershed covers about 65 square kilometers. The upper watershed, above Old Franks Lakes, covers 41 square kilometers and is the location of the survey. The upper watershed was chosen because National Forest timber harvest has occurred above the lakes and the lower watershed contains private land. The Old Franks Creek Watershed Analysis document (Ketchikan Area 1994) contains detailed maps of the watershed.
 

Channel Type Distribution
 

The surveyed reach is 7.4 kilometers long and extends from the upper lake to the top of anadromous habitat, which stops at a 10 meter waterfall. The reach consists of four channel types: low gradient floodplain channel (FP4), wide low gradient floodplain channel (FP5), low gradient contained channel (LC1), and narrow shallow contained channel (MC2) (Paustian 1992) . Four distinct reaches were identified along the surveyed channel based on process group (Figure 1). The first reach is a floodplain channel (FP4 and FP5 channel types) and extends from the upper lake for 4.0 kilometers. (Note: FP4 and FP5 channel types were distinguished during the survey but the two channel types are not considered as separate reaches because the surveyed section is near the 20 meter width dividing point throughout much of its length. Distinctions between the two channel types would probably not be useful. Channel type verification was not conducted during the survey so distinctions between FP4 and FP5 channel types may not be accurate throughout the survey.) At that point the stream becomes contained by bedrock in a large contained channel (LC1) for 0.7 kilometer. The next 2.2 kilometers is uncontained in a floodplain channel (FP4 and FP5 channel types). The final 0.5 kilometer to the top of anadromous habitat is a moderate gradient contained channel (MC2 channel type). Survey methods were similar to those described by Wright (1994).
 
 

Figure 1. Channel type reach distances in Upper Old Franks Creek, 1994.
 

Habitat Type Distribution

Side channel habitat is an important fish habitat component so habitat units that are side channels of the main surveyed channel were identified. The floodplain process group contained most of the side channel habitat area (17,041 square meters) while the large contained process group contained 1,155 square meters. No side channel habitat was identified in the moderate gradient contained stream reach (Figure 2). Old Franks also contains much off-channel habitat in beaver ponds, sloughs, and small tributary streams. This habitat has been unquantified but a preliminary estimate from aerial photos is greater than 50,000 square meters.

Figure 2. Side channel and main channel habitat in Upper Old Franks Creek, 1994.
 

All habitat units were classified as either fast water or slow water and then further broke them down into micro habitat types according to Hawkins et al (1993). A glide classification was added to both the slow and fast water habitats. The slow water glides would be classified as mid-channel scour pools and the fast water glides would be runs according to Hawkins et al (1993). The wetted surface area in the surveyed section of Old Franks Creek consisted of 65% slow water habitat. The percent slow water habitat by channel type was as follows: FP4 = 70%, FP5 = 59%, LC2 = 57%, and MC2 = 23%. These values are above the 50th percentile (Alaska Region Riparian Habitat Management Objectives) for percent of stream in pool habitat in each of these channel types.
 

Lateral scour and mid-channel scour pools dominated the FP4 channel type. They made up 28 and 26 percent of the habitat area respectively (Figure 3). The FP5 channel type was dominated by the riffle habitat type (21%) which was followed by lateral scour (19%) and mid-channel scour (17%) habitat types. The LC2 channel type was dominated by mid-channel scour pools, which made up 45% of the area. The MC2 channel type reach was dominated by fast water habitat, which made up 77% of the area. No scour or dammed pools were identified in the MC2 reach. The total wetted surface area by habitat type and channel type is presented in Figure 4.
 

Average habitat unit sizes are presented in Figure 5 and Table 1. In the FP4 and FP5 channel types riffle habitat units were the most numerous with 45 and 41 units respectively. Lateral scour pools were the next most numerous in both channel types. Glides had the largest average habitat unit size in both floodplain channel types at 590 square meters. The faster moving glides (runs) were the largest units in FP5 reaches at 1,389 square meters. In the LC2 reach mid-channel scour pools were the most numerous (six units) and the largest units (564 square meters). In the MC2 channel type glides were most numerous with five units and bedrock riffles were the largest with an average size of 639 square meters. Off-channel and side channel units made up 42%, 33%, and 32% of habitat units in FP5, FP4, and LC2 channels respectively. The high percentage of off channel and side channel habitat represents the complexity of much of the fish habitat along Upper Old Franks Creek.
 

Scour pools, predominantly mid-channel and lateral scour with some additional eddy and plunge, were the most numerous pool types throughout the floodplain and large contained reaches.
 

Note: The following habitat type abbreviations are used in tables and figures throughout this report. PL=pool, RF=riffle, DRY=dry channel, BW=backwater, ED=eddy, GL=glide, LSC=lateral scour, MCS=mid-channel scour, PP=plunge pool, TR=trench, B=boulder, BR=bedrock, CB=cobble, CS=cascade, RP=rapid.
 
 

Figure 3. Wetted surface area (percent of area covered) by habitat type and channel type in Upper Old Franks Creek, 1994.
 
 

Figure 4. Wetted surface area by channel type and habitat type in Upper Old Franks Creek, 1994 .
 
 

Figure 5. Average habitat unit size in Upper Old Franks Creek, 1994.
 

Table 1. Number and average area (square meters) of habitat types by channel type in Upper Old Franks Creek,1994.

Habitat Type	FP4  Count	Area	FP5 Count	Area	LC2 Count	Area	MC2 Count	Area	Total Count	Mean  Area
DRY	0		1	370	0		0		1	370
PL-BW	7	300	10	170	2	94	0		19	210
PL-ED	9	63	4	86	0		0		13	71
PL-GL	9	590	11	581	1	111	5	147	26	483
PL-LSC	39	421	39	215	3	130	0		81	311
PL-MCS	28	533	16	465	6	564	0		50	515
PL-PP	9	129	6	145	0		2	173	17	140
PL-TR	0		0		1	205	0		1	205
RF	45	211	41	223	1	511	3	532	90	231
RF-B	1	132	0		3	147	3	280	7	202
RF-BR	0		0		0		1	639	1	639
RF-CB	9	361	11	289	5	270	0		25	311
RF-CS	2	98	0		4	202	0		6	167
RF-GL	11	351	4	1389	1	58	1	566	17	590
RF-RP	0		0		1	21	0		1	21
Total	169	340	143	303	28	267	15	315	355	318

Pool Forming Agents
 

The stream feature that resulted in pool formation was identified when possible in each pool unit. Originally this survey component was intended to identify those pools formed by large wood, but it was expanded to include any identifiable pool forming feature. The feature responsible for pool formation was identified in 100 of the 207 pools in the survey. When the specific pool forming agent could not be identified, the item was left blank. Large wood (not including rootwads) was the forming agent in most floodplain pools with rootwads and flow the next most dominant forming agents (Figure 6). In the FP4 and FP5 channel types large wood and rootwads combined were the identified forming agents in 31% and 52% of the pools respectively. Bedrock, flow, and boulders were the only identified forming agents in the LC2 and MC2 reaches.

Figure 6. Pool forming agents by channel type in Upper Old Franks Creek, 1994 (B=boulders, B.Dam=beaver dam, BR=bedrock, Flow=water flow, LW=large wood, RW=rootwad).
 
 
 

Water Depths
 

Mean and maximum depths were measured in all habitat units and riffle crest depth for pools was measured to calculate residual pool depth. The upper habitat unit in the MC2 reach was left out of the analysis. It was a four meter deep plunge pool at the barrier falls and greatly skewed the summary for the channel type. Mean, maximum, and residual pool depths were averaged for all habitat units in each channel type. Mean depths were similar for all channel types (Figure 7 ). Maximum depth and residual pool depth were highest in the FP4 reaches and lowest for the MC2 reach. Surprisingly maximum and residual pool depths were higher in the bedrock controlled LC2 channel type then in the FP5 channel type.

Figure 7. Water depth by channel type at low flow in Upper Old Franks Creek, 1994.
 

Woody Debris
 

Woody debris comparisons were made based on the total number of pieces in all size classes per 1,000 square meters wetted surface area of channel type (Table 2). The FP5 channel type had the highest density of wood with all pieces combined and in every category except the large category where the FP4 channel had the highest density. The FP5 channel type, with 17.7 pieces per 1000 square meters had well above the regional 75th percentile (six pieces). The FP4 channel was at the low end of the regional range while the LC2 and MC2 channel types were near the 50th percentile in the regional range.
 

Table 2.. Woody debris counts by size class and channel type for Upper Old Franks Creek, 1994, SM=1-7.6m long by .1-.5m dia., MED=1-7.6m long by .5-.9m dia., LRG=7.6-15m long by .1-.9m dia., GNT>15m long, SRW=small rootwad<1m dia, LRW=large rootwad>1m dia.
 

Size	FP4	FP5	LC2	MC2	Grand Total
SM	314	365	22	33	734
MED	127	176	4	14	321
LRG	112	81	7	0	200
GNT	129	144	7	12	292
SRW	54	72	2	1	129
LRW	104	145	1	2	252
Total Wood	840	983	43	62	1928
Kilometers	4	3	1	1	7
Pieces/km	188	294	61	114	209
Miles	2	2	0	0	5
Pieces/mile	303	473	98	184	336
Area	57427	43398	7472	4724	113021
Pcs/1000 sq m	11.9	17.7	5.4	12.5	13.7

Cover
 

Fish cover was recorded by percent of the habitat unit comprised of the dominant and subdominant cover types. The LC2 channel had the largest portion of the channel type with cover at 48%, with 17% cover in FP4, 15% in FP5, and 5% in MC2. The LC2 cover consisted of mainly substrate components, boulders, and bedrock (Figure 8). FP4 and FP5 cover percentages both consisted primarily of depth and large wood with smaller percentages of overhanging vegetation, rootwads, undercut banks, and slash. The only cover components present in the MC2 channel type were depth and boulders.
 
 

Figure 8. Percent of total area of each channel type covered by each cover type in Upper Old Franks Creek, 1994 (AV=aquatic vegetation, B=boulder, BR=bedrock, C=cobble, D=depth, DJ=debris jam, LW=large wood, OV=overhanging vegetation, RW=rootwad, SL=slash, SU=substrate, UC=undercut bank).

Most of the cover was contained in the FP4 and FP5 channel types and consisted predominantly of depth and large wood (Figure 9). Substrate cover components were found in roughly equal amounts between the LC2 channel and floodplain channel types. Cover in the MC2 channel made up little of the total cover present in the stream.

Figure 9. Fish cover in Upper Old Franks Creek by wetted area covered in each channel type, 1994 (AV=aquatic vegetation, B=boulder, BR=bedrock, C=cobble, D=depth, DJ=debris jam, LW=large wood, OV=overhanging vegetation, RW=rootwad, SL=slash, SU=substrate, UC=undercut bank).

Substrate
 

The FP4 and FP5 channel types had very similar dominant and subdominant substrate sizes by percentage of channel type area (Figure 10). Substrate types consisted predominantly of gravel, fine gravel, and cobble in these floodplain channels. The FP4 channel had more of the larger substrates (bedrock, boulder, and rubble), while the FP5 channel had more of the smaller depositional substrates (fines and organic). The larger substrates dominated the LC2 and MC2 channels and consisted predominantly of bedrock and cobble.
 

Figure 10. Substrates by percent of channel type area covered in Upper Old Franks Creek, 1994 (BR=bedrock, B=boulder, R=rubble, C=cobble, GR=gravel, FG=fine gravel, S=sand, F=fines, O=organic).
 

Riparian Vegetation
 

Riparian vegetation was classified by plant association (Demeo et al 1992). Timber harvest was noted only when it occurred to the stream banks. Harvest units were buffered at least 200 feet so riparian vegetation along the stream was all classified as Old Growth and plant associations were identified. The mixed conifer/blueberry plant association dominated the riparian vegetation in all channel types except the LC2 (Figure 11). The LC2 channel type was dominated by the Sitka spruce/blueberry/devil's club association and this association was the second most prevalent in both floodplain channel types. Both floodplain channel types also contained reaches of non-forested habitat, primarily grass/sedge meadows at the confluence with the lake and around beaver ponds. In addition, the FP4 channel type had a short reach of the shore pine/spruce sedge association.

Figure 11. Riparian plant association along the surveyed section of Upper Old Franks Creek, 1994 (bb=blueberry, dc=devil's club, Sh=shore, S=spruce).
 
 
 

Fish Abundance and Distribution
 

Snorkel fish counts were conducted in 60 habitat units spaced at regular intervals along Upper Old Franks Creek. Every fifth pool and every eighth riffle was snorkeled. One or two divers were used, depending on habitat unit size. Fish density estimates were produced by summing numbers of fish counted in units of each habitat type and dividing this total by the total area of stream snorkeled in the habitat type. No correction factors were applied to snorkel counts to account for missed fish. Density estimates should be considered minimums. When extrapolated into population estimates, the estimates will be low.
 

Cutthroat Trout
 

Cutthroat and rainbow trout are both present in the Old Franks system. These species are difficult to differentiate with snorkel surveys so cutthroat density estimates also include the rainbows. Cutthroat density was three times higher in mid-channel scour pools of the LC2 channel type (based on only one habitat unit snorkeled in this habitat type) than in any other habitat type (Figure 12). When entire channel types were pooled, cutthroat densities were highest in the MC2 channel followed by the LC2 channel type, FP5 channel type, and FP4 channel type. The cutthroat seemed to prefer, or were more visible in, the higher energy channel types with larger substrates.

Dolly Varden
 

Dolly varden is the most difficult species in the surveyed area to enumerate by snorkeling because they tend to sit motionlessly on the stream bottom or under cobbles. Dolly varden density was highest in the LC2 channel type in rapids (Figure 13). Dollys were also observed in the floodplain channel types in lateral scour, mid-channel scour, and backwater pools. The density of dolly varden is very low throughout the surveyed area. They were observed in only 20% (12) of the habitat units snorkeled.

Figure 13. Dolly varden density by habitat type in Upper Old Franks Creek, 1994.
 

Coho Salmon
 

Coho salmon were counted during the fish survey for the second consecutive year. The presence of juvenile coho shows that some indigenous adults are moving through the entire lake system and spawning in the stream above. All coho observed were age 1 fish (>60 mm). No progeny of the 1993 coho run were observed. Coho were seen only in the floodplain channel types in glides, lateral scour pools, and mid-channel scour pools (Figure 14). Density was about one fish per 500 square meters in the floodplain channels. Bioenhancement of the coho run began in August with the stocking of coho fingerlings into the top lake, at the downstream end of this survey. These fish came from Karta stock reared at the Klawock River Hatchery. This survey was conducted prior to the coho stocking.
 
 

Figure 14. Coho density by habitat type in Upper Old Franks Creek, July 1994.
 

Species Comparisons
 

Cutthroat densities exceeded densities of other species in all habitat types throughout the stream (all channel types combined) (Figure 15). Snorkel surveys have been conducted each year from 1990 to 1994 and timing has varied from early June to mid-October. The 1994 survey showed the highest densities of cutthroat of all surveys conducted (Figure 16). This is partially due to the 1994 survey occurring immediately after emergence of cutthroat fry, but age 1+ densities were also much higher than in any other year. Consistently low densities of cutthroat indicate that cutthroat are not heavily using the mainstem of Old Franks Creek as a rearing area. Dolly varden density remained fairly consistent from 1991 to 1994. No dolly varden were noted in the 1990 survey.
 

Lake
 

The lake cutthroat population is high (as indicated by angling success) in comparison with the stream and with similar lakes. Cutthroat, rainbow, dolly varden, and kokanee are all present in Old Franks Lakes. A 1992 rod and reel survey of fish in the upper lake revealed an average cutthroat size of 261mm (n=63 fish), average rainbow size of 231mm (n=30 fish), average dolly varden size of 201mm (n=13) and kokanee size of 191mm (one fish) (Figure 17).
 
 

Figure 15. Upper Old Franks Creek salmonid densities from snorkel surveys, 1994.
 
 

Figure 16. Salmonid densities from snorkel surveys in Upper Old Franks Creek, 1990-1994, (CT = cutthroat/rainbow, DV = dolly varden, CO = coho).
 
 

Figure 17. Representative angling sample of fish in Upper Old Franks Lake, 1992 (CTH=cutthroat, RNB=rainbow, DV=dolly varden, KOK=kokanee).
 
 
 

Comparison of Old Franks Salmonid Densities to Other Streams
 

The 1994 overall Old Franks cutthroat density with all habitat types combined (.05 fish per square meter) was below the average density (.063 fish per square meter) from similar surveys conducted on southern Prince of Wales Island between 1992 and 1994 (Figure 18). The 1994 Old Franks density was more than double that from any previous year. The overall Old Franks dolly varden density (.001 fish per square meter) was also below the average density (.02 fish per square meter) from similar surveys on southern Prince of Wales Island. The average coho density from surveys on southern Prince of Wales (not including Old Franks) was 0.45 coho per square meter while the Old Franks value was .002 fish per square meter (Figure 19). It is anticipated that when the system is fully colonized by coho the density will approach this figure.
 

Old Franks is the only non-anadromous (previously) lake system with a stream survey conducted above the lake. Upper Cable Creek, Kassa Creek, and Dog Salmon Creek all have some similarities with Old Franks. Upper Cable Creek and Kassa Creek are both non-anadromous systems without lakes. The upper Cable Creek survey showed similar densities of cutthroat and dolly varden compared to Old Franks. Cutthroat were not identified in Kassa Creek and dolly varden densities were slightly higer than Old Franks. Dog Salmon Creek is the only other surveyed stream associated with a lake, although the system is anadromous. Densities of cutthroat/steelhead and dolly varden in Dog Salmon Creek were more than double the Old Franks values.
 
 

Figure 18. Average overall stream densities of cutthroat (combined with rainbow and steelhead) and dolly varden in streams on southern Prince of Wales Island.
 

Figure 19. Juvenile salmonid densities from snorkel surveys in streams on southern Prince of Wales Island 1992-1994.
 
 
 

References
 

Demeo, T. J. Martin, and R. West. 1992. Forest plant association management guide. Ketchikan Area, Tongass National Forest. 405p.
 

Hawkins, C.P. et al. 1993. A hierarchical approach to classifying stream habitat features. Fisheries. vol.18 no.6. 3-12.
 

Paustian, S.J., editor. 1992. A channel type users guide for the Tongass National Forest, southeast Alaska. Alaska Region Technical paper, R10-TP-26, USDA, Forest Service. 179 p.
 

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

Wright, B.E. 1994. Habitat survey and salmonid distribution, Kadake Creek watershed, Kuiu Island. USDA Forest Service, PNW Research Station. unpublished manuscript. 25p.
 

Thanks to Herb Roerick and                                       for ensuring accurate and timely completion of the stream surveys.
 
 
 
 

Table 1. Percent habitat area by channel type in Upper Old Franks Creek, 1994.
 

	FP4	FP5	LC2	MC1	Grand Total
DRY	0.0	0.9	0.0	0.0	0.3
PL-BW	3.7	3.9	2.5	0.0	3.5
PL-ED	1.0	0.8	0.0	0.0	0.8
PL-GL	9.3	14.7	1.5	15.6	11.1
PL-LSC	28.6	19.3	5.2	0.0	22.3
PL-MCS	26.0	17.2	45.3	0.0	22.8
PL-PP	2.0	2.0	0.0	7.3	2.1
PL-TR	0.0	0.0	2.7	0.0	0.2
RF	16.6	21.1	6.8	33.8	18.4
RF-B	0.2	0.0	5.9	17.8	1.2
RF-BR	0.0	0.0	0.0	13.5	0.6
RF-CB	5.7	7.3	18.1	0.0	6.9
RF-CS	0.3	0.0	10.8	0.0	0.9
RF-GL	6.7	12.8	0.8	12.0	8.9
RF-RP	0.0	0.0	0.3	0.0	0.0
Grand Total	100.0	100.0	100.0	100.0	100.0
Percent Pools	70.5	58.8	57.3	22.9	63.

 
 
 

Table 2. Wetted surface area (square meters) by habitat type and channel type in Upper Old Franks Creek, 1994.
 

MICRO	FP4	FP5	LC2	MC2	Grand Total
DRY	0	370	0	0	370
PL-BW	2098	1699	189	0	3986
PL-ED	571	346	0	0	917
PL-GL	5313	6395	111	737	12557
PL-LSC	16419	8394	391	0	25205
PL-MCS	14911	7448	3386	0	25744
PL-PP	1165	868	0	346	2379
PL-TR	0	0	205	0	205
RF	9514	9141	511	1595	20761
RF-B	132	0	440	840	1411
RF-BR	0	0	0	639	639
RF-CB	3252	3184	1352	0	7787
RF-CS	196	0	809	0	1005
RF-GL	3856	5554	58	566	10035
RF-RP	0	0	21	0	21
Grand Total	57427	43398	7472	4724	113021
Percent Pools	70.5	58.8	57.3	22.9	63.1

 
 
 

Table 3. Number and average area (square meters) of habitat types by channel type in Upper Old Franks Creek, 1994.

Habitat Type	FP4  Count	Area	FP5 Count	Area	LC2  Count	Area	MC2  Count	Area	Total Count	Mean  Area
DRY	0		1	370	0		0		1	370
PL-BW	7	300	10	170	2	94	0		19	210
PL-ED	9	63	4	86	0		0		13	71
PL-GL	9	590	11	581	1	111	5	147	26	483
PL-LSC	39	421	39	215	3	130	0		81	311
PL-MCS	28	533	16	465	6	564	0		50	515
PL-PP	9	129	6	145	0		2	173	17	140
PL-TR	0		0		1	205	0		1	205
RF	45	211	41	223	1	511	3	532	90	231
RF-B	1	132	0		3	147	3	280	7	202
RF-BR	0		0		0		1	639	1	639
RF-CB	9	361	11	289	5	270	0		25	311
RF-CS	2	98	0		4	202	0		6	167
RF-GL	11	351	4	1389	1	58	1	566	17	590
RF-RP	0		0		1	21	0		1	21
Total	169	340	143	303	28	267	15	315	355	318