ANADROMOUS SALMONID RESOURCES
OF MENDOCINO COASTAL AND INLAND RIVERS
1989-1990
An Evaluation of Rehabilitation Efforts Based on Carcass Recovery and Spawning Activity
by Jennifer L. Nielsen
. -- Mike Maahs George Balding
This work was fundecLby the California Department of Fish and Game, Fisheries Division, Fisheries Restoration Program, Work Progress Report 1990, Contract No. FG9364. ABSTRACT
Eighty-two streams and tributaries (355 stream miles) in Mendocino County were surveyed by members of The Salmon Trollers Marketing Association for anadromous salmon on a weekly basis during the 1989-1990 spawning season (November to February). Data on carcass capture-recapture, live fish, redds, carcass species, sex, retention by habitat, carcass predation and geomorphic and environmental parameters co~tributingto salmonid spawning were measured. Three models were used to estimate the total salmon runs remaining on streams in Mendocino. When compared to a controlled release of spawners on the Noyo River, the area- under-the-curve models consistently underestimated the population. Jolly-Seber estimates analyzed by a nonparametric smoothing model gave the best results when compared to the control. Low numbers of carcass capture- recapture histories limited application of this model to three populations. No single model applied to all the streams surveyed. The standard area-under-the-curve model (AUC) estimated the total populations on 15 streams to equal 703 coho and 520 chinook. The nonparametrlc model (NPACC) using carcass data estimated total populations on 6 streams to be 2033 coho and
484 chinook,. Using a computer program (JOLLY), Jolly-Seber estimates based on recapture histories for individual fish estimated the total population on 3 streams as 870 coho and 615 chinook (AUC) or 3511 coho and 615 chinook (NPAUC) . Live counts were used to estimate a total population of 1071
coho, 714 chinook and 38 steelhead spawners on 19 streams. Run timing and spawning distribution for different anadromous salmonids were discussed for the survey streams. The relative contribution of different age-classes and sexes to the returning spawning runs was considered on the Noyo River coho and the S.F. Eel River chinook where scales were read. Carcass retention by deep pool habitat was more significant to recovery on our streams than woody debris, regardless of stream size and spawning species. However, no analysis of total available habitat was made. Carcass predation was discussed as a factor in carcass recovery and spawning success. Evaluations of enhancement efforts were made on streams
where records were available. In 1989-1990, the numbers of returning spawners were low throughout Mendocino. The Noyo River was the only stream with coho populations exceeding
500 spawners. This stream receives fry and smolts hatched from eggs taken at the Noyo station and reared off site by
the CDF&G. Results suggest an evaluation of the contribution of these fish to natural spawning should be made.
The low numbers of spawning salmon found in 1989-1990 were thought to be more a factor of declining natural populations due to continuous drought conditions in California than available spawning habitat or enhancement efforts. Acknowledgements
We wish to thank the following survey team members without whose dedication on the job and keen observation skills this data would never have evolved: G. Abbott, L.
Amos, L. Barkeley, R. Baxter, R. Behm, A. Bolin, V. Bolin,
M. Buckley, R. Bush, G. Chereau, B. Clark, J. Cresci, J. Frey, J. Gilleard, P. Gilly, A. Giusti, J. Gunter, W. Gunter, D. Hall, G. Hall, W. Hansen, J. Hasha, W. Heidinger, K. Heintz, R. Jensen, L. Kemp, M. Lippincott, J. Martin, R.
Martin, S. McDermid, J. Mendosa, T. Merriman, L. Miller, C. Platt, T. Quinn, E. Quintana, S. Ragsdale, J. Reiff, K. Sample, R. Scott, W.-Scott, D. Stamback, M. Stanback, D.
Stanley, D. Swenson, L. Vargas, J. wall, R. Wheeler, J.
Williams, J. Yearwood, G. Zackary. Weldon Jones of the California Department of Fish and Game was instrumental in getting this program going and ran the initial training surveys on Casper Creek. Brad Clark and Jim Reiff deserve special recognition for their efforts on behalf of the project. A special thanks is due Wayne Scott of the Salmon Trollers Marketing Association and Craig Tuss of the U.S. Fish & Wildlife Service for scale reading. We would finally like to thank all of the land owners with property adjacent to the survey streams for their kind assistance Grid cooperation. TABLE OF CONTENTS
Introduction ...... 1 Methods ...... 4 Results ...... 15 General Survey Data ...... 15
Results by Stream ...... 30 South Fork Noyo River ...... 30
South Fork Eel River ...... 42 Big Dan Creek ...... 52 South Fork Big River ...... 52 Bridges Creek ...... 54 .. Caspar Creek ...... 55 Cedar Creek ...... 58 Deer Creek ...... 58 Dehaven Creek ...... 59 Dutch Charlie Creek ...... 59 South Fork Garcia River ...... 60 Little North Fork Gualala River ...... 60 Hollow Tree Creek ...... 61 Howard Creek ...... 67
Indian Creek ...... 67
Jack of Hearts Creek ...... 71
Kenny Creek ...... 71
~ittle.CharlieCreek ...... 73 Little River ...... 73 Low Gap Creek ...... 73 McCoy Creek ...... 74
iii Mill Creek ...... 74 Mud Creek ...... 74 Outlet Creek ...... 75 Piercy Creek ...... 77 Pudding Creek ...... 78 Rattlesnake Creek ...... 80 Red Mountain Creek ...... 80 ... - . . Redwood Creek ...... 82 Rock Creek ...... 84 Section-4 Creek ...... 84 Ten Mile Creek ...... 85 Ten Mile River ...... 85 \ Wages Creek ...... 89 Discussion ...... 90 Literature Cited ...... 99 Appendix I . 1989-1990 Budget ...... 102 Appendix I1 . Location maps ...... 103 INTRODUCTION
The salmon runs of Northern California are thought to be on the decline. In 1989, specific runs of coho and chinook salmon were listed by the California Department of Fish and Game as species of special concern (Moyle, et al., 1990). Surveys of the coho and chinook populations in Mendocino during the 1989-1990 spwa&qy&uns.&aroduced- _- . low counts of anadromous spawners throughout the county. Despite frequent observation cycles and extensive stream walking our estimates of abundance exceeded 500 spawners on only two streams, coho in the Noyo River and chinook on the South Fork Eel River. . It is important to consider these results in light of the ongoing drought conditions in California. Low spawner abundance is probably the result of both population declines and redistribution of spawners due to low flows. The brief rains of January, 1990, were followed by chinook spawning in the upper Eel River tributaries suggesting that movement within the basin may have been limited prior to these flows. The primary objective of this study was to use carcass or live fish counts to estimate total escapement populations. Such estimates are limited by the assumptions inherent in the models used to calculate them. In this paper three different approaches were tested on the carcass and spawner data collected in 1989-1990. No one model proved to be a panacea for estimating abundance in small populations No single model applied to the data available on all the streams where anadromous spawning was recorded. This suggests the need for flexibility in estimating the salmon runs of Mendocino. Understanding changes in populations over time may depend on our ability to estimate different parameters in each basin. Monitoring several parameters which apply to different models will give a clearer understanding of long-term trends in salmonid abundance.
A second objective of this study was to identify "index" streams that could be used for future monitoring programs in Mendocino County. We feel such streams were not readily selected from the results of the carcass surveys. Our data demonstrate the need to monitor streams using several environmental indicators: estimates of the spawning populations, estimates of habitat abundance, juvenile survival and growth rates, timing of smoltification and genetic heterogeneity. None of these parameters alone can be used to predict the success or failure of salmonid populations over the next few years nor determine the reason why we continue to have low numbers of anadromous salmonids in California waters. These data provide a hard earned insight into the
salmon populations of Mendocino. Over 3,500 survey miles were walked in all types of winter weather. The extent of the surveys.and the quality of the data testify to the commitment the surveyors had to the research objective to monitor the anadromous spawning populations remaining on streams in Mendocino county. However, their results were not encouraging. Surveyors were surprised by the extent and quality of habitat available on Mendocino streams. They expected to find streams in much poorer condition. They were struck by the obvious lack of fish using the existing habitat. Such extensive information is seldom available on the distribution and timing of salmon runs in one year on so many streams. But our data present no real answer to the lack of returning salmon on Mendocino streams due to the unusual climatic condition (drought) during this winter and the lack of available comparable data from previous years. It is our hope that the streams in this study will
continue to be monitored for salmonid use in the future. A similar study conducted in a "normal" hydrologic year would give better baseline information on these populations. Since natural population numbers can change over time, it is . - important to monitor natural llcycles"in density which reflect diverse climatic and ecological conditions. We believe that as the human population increases in Mendocino County their impact on the environment will also increase. Every available factor should be considered in the pursuit of a remedy for California's declining salmon populations. Throughout history they have represented the health of our rivers. Their disappearance would represent a great ecological, economic and cultural loss for this state. METHODS
Streams in Mendocino County were surveyed for live
fish, redds, carcasses (dead fish measured for fork-length from the tip of the nose to the fork in their tail) and skeletons (dead fish which could not be measured for fork-
length). Streams were surveyed in selected reaches where historic data, local residents and professional opinion demonstrated a high likelihood of salmon spawning. Stream reaches were divided into distinct sections which could be surveyed in one day. Teams of two observers conducted the surveys on a weekly basis depending on weather conditions and access. Team partners were rotated throughout the labor pool such that no team member repeatedly surveyed the same stream section.
Surveys were conducted by consistently walking the stream bed up or down stream from a standard access point to a prescribed exit point. Observers poked into holes, under debris and checked the brush along the stream bank for carcasses. Team members carried hooked gaffs to facilitate retrieval of carcasses from deep pools or under log jams.
Chest-waders and wading boot with felts or corkers were worn to allow access into areas where stream beds were slick, woody debris abundant and stream banks steep.
Daily,records were kept for each survey. Carcasses were examined for species, sex and measured for length (cm). They were then tagged with a dolored hog-ring in the lower jaw.
Colors were alternated weekly on each creek to establish mark-recapture data. Fish remaining until the next survey were tagged again with a new color. A complete record was kept of old tags by color and number on carcasses during each visit. Whenever possible, carcasses and skeletons were also tagged on the tail using a paper punch to create a hole in the tissue between the rays of the caudal fin. Additional holes were punched on fish when they were reexamined on subsequent visits and used to supplement the mark-recapture data from jaw-tags. Carcasses were examined for adipose clips. If the adipose fin was missing, fish snouts were removed and sent to the California Department of Fish and Game for coded-wire tag examination. Scales were taken from the upper posterior body of the fish, about half-way above the lateral line between the dorsal and adipose fin. After manipulation all carcasses were replaced in the exact location where they were found. Scales were retained in individual envelopes labeled as to fish species, length, sex and stream section. All scales read for fish age were evaluated by three independent researchers: the author, Wayne Scott of the Salmon Trollers Marketing Assoc., and several government biologists under
the direction of Craig A. Tuss, U.S. Fish & Wildlife Service, Arcata. Fish age was assumed to be the consensus of evaluations determined from these three observations. All scales taken from chinook in the South Fork Eel basin were read for age. A random set of scales (25%) were read from coho recovered in the Noyo River to establish the proportional distribution of age classes in the spawning population. The location of carcasses and skeletons was recorded by habitat. Habitat types included pools, riffles, eddy, stream bank, shrubs or vegetation, large organic debris (LOD), rootwads, and gravel bars. Predation on carcasses was noted based on direct observation and signs left by predators such as scat, foot prints and residual sign of carnivore scavenging (Cederholm, et al., 1989). Counts of redds-and live fish were estimated by each observer. Team estimates were based on a consensus between the two members. Most surveys were conducted on private property. At the request of several large private land owners no effort was made to flag redds in 1989-1990. Flagging of redds was considered a possible threat to stream aesthetics and might have conflicted with ongoing commercial timber harvest flagging practices. This year's redd counts may, therefore, represent repeat enumerations, especially with significant dry periods occurring between our surveys. Live fish were estimated when the observers first encountered holding pools and were based on estimates made from bank observations. No record was kept on upstream or downstream movements of live fish during the counts. Physical parameters measured during the surveys included stream flow, water and air temperatures, the observable depth as a proportion of the water column, and time of access and exit from the stream section. Flow was measured by estimating velocity and volume at specific sites located at the lower end of each section. These values were used to calculate discharge in cubic meters per second (cms) passing through the reach. Velocity of water in the reach was estimated by measuring the time required for a standard float to travel through relatively laminar flow in a reach of approximately 10 meters (Schlosser, 1982). Laminar flow was defined as flow passing through a path of relatively uniform depth with little turbulence and no obstructions. This measure was repeated three times and the mean value was used as the estimate of meters per second of discharge. Total water volume was estimated by measuring total water column length at the site and stream widths at three equidistant channel cross-sections. Three water column depths were measured at equal distances across each width cross-section. Mean depth at each cross-section was calculated as the sum of the three depths divided by four to allow for the decrease in water depths at the stream margins (Platts, et al., 1987). Discharge was calculated as total length (meters) multiplied by the mean width (meters), multiplied by the average of the three mean depths (meters), divided by the mean time duration estimated for water velocity in seconds (cubic meters per second). These procedures tended to oveGestimate the volume of water passing through each site because surface velocity is usually greater than the average velocity of the water column (Leopold, et al., 1964). Air and water temperatures were taken at the flow site with a hand held thermometer in degrees Celsius. Observable stream depth was standardized in a deep pool within the stream section. Total depth was measured using a calibrated staff (cm). A second measurement was taken at the point. -.-. - - .- - where the tip of the staff disappears to the observer to determine the clear volume of the water column. The ratio of observable to total depth was considered the observable proportion of the stream. This measure was used to monitor relative visibility of carcasses at depth during surveys.
AS...... Three general models were used to estimate the population of spawners on each stream. Standard area-under- the-curve estimates were calculated for carcass and live fish counts based on the methods given in Beidler & Nickelson (1980). This model estimates the population as:
where, N = total estimated number of spawning fish in a given stream section, C;= count of fish in the ith period, TIG, number of day between /'and (i-I), - X Sp = the average carcass duration (days) or spawning life for fish in the given stream section. This model is extremely sensitive to the estimate of carcass duration when calculating populations from carcass data and is equally sensitive to the estimated length of an average spawner's life when working with live fish counts. Beidler & Nickelson (1980) gave an estimated spawning life for coho of
11.3 days in Oregon streams. Due to the lack of tag data on live spawner in Mendocino streams it was assumed that this number applied to fish in this study. The average time of carcass duration on the survey streams was calculated from recapture histories on individual fish where records were available. If no carcass history was available the average duration calculated for the nearest stream with a history = was used. If the first or last count in the section was greater than zero, a zero count was assumed to occur seven days before or after the given count. Confidence intervals for the population estimate were calculated using the standard error of the sample.
A second model applied to the data from these surveys was a Jolly-Seber model. The Jolly-Seber model is an open population model appropriate for carcass survey data because it allows for both immigration into the population (new deaths) and time specific capture probabilities (due to variability of run size over time). A new program, JOLLY, was used to.estimate survival rate, capture probability, population size at each time interval, new recruits into the population at each time interval, standard error and 95% confidence intervals for the population estimate (Pollock, et al., 1990). JOLLY based its estimates on the mathematical assumption that the proportion of marked animals (carcasses) in a sample should equal that in the population:
where mi and n,, are the marked and total numbers of animals captured in the ith sample. This assumption is critical to all mark-recapture models and JOLLY carried with it several of the assumptions implicit in the more traditional Lincoln- Peterson model: 1. Every subject has an equal probability of capture in each sample period. 2. Marks are not lost or overlooked.
3. All samples are instantaneous and release is made back into the same environment as capture.
The open population model used a survival rate estimator developed from the number of marked carcasses in the population immediately after sample i and the proportion of marked carcasses remaining in the population. Program JOLLY added one assumption to meet this criteria:
4. Every marked animal present in the population after each sample period has the same probability of suryival until the next sample period.
JOLLY calculated the difference between the population size in a given sample and the expected number of survivors to estimate new recruitment into the population. Capture probability was estimated as the proportion of marked or total (marked + unmarked) carcasses surviving into the next sample (i.e. retags). A thorough and in-depth explanation of the parameter estimations, calculations of variance and covariance built into this model was given in Pollock, et al. (1990) and will not be repeated .here. Seber (1982) recommended that the number of marked animals recaptured in each interval and the number of animals released within each sample period be greater than ten for satisfactory performance of the estimators in the Jolly-Seber model. Small sample sizes limited the application of this model to two streams, the South Fork Eel River chinook population and the Noyo River coho population. Recapture data was traced through the observation cycles plotting recaptures as individual case histories. A "B-Table" of capture histories was constructed for each stream and run on JOLLY. Four versions of the standard open model were offered in JOLLY which set controls for the estimated parameters. Depending on the data, parameter controls can increase the robustness of the model (Cormack, 1972). Because the time interval between surveys varied from creek to creek and within samples on any one creek, Model B, which assumed a constant survival rate per unit of time and a time specific capture probability was run on the 1990 carcass data. A comparison was made between the results from the general model and Model B using chi-square and goodness- of-fit tests. The Jolly-Seber model estimated the number of individuals within the population at given time intervals. It was still necessary to manipulate this data to get total population estimates. The Jolly-Seber estimates generated by JOLLY were run through the standard area-under-the-curve estimator and the third estimator used in this study, a nonparametric-under-the-curve estimator described in Noakes
(1989). The nonparametric approach to generating total population estimates-from interval data used a probabilistic -. model as opposed to the parametric relationships im&-icit in the standard area-under-the-curve model. Because of their robust nature, nonparametric models seemed appropriate for . -. situations where the system changes over tine and results often show significant deviations from an assumed normal probability distribution. This is particularly true of carcass data where run size often pulse over time due to flow dependent passage and bimodal spawning migration patterns. Nonparametric techniques of regression and Bayesian analysis of time series models for population estimation have been proposed throughout the recent literature (Gazey &
Stanley, 1986; Cleveland et al., 1988; Spall, 1988). These < techniques try to capture local characteristics of the population in discrete time intervals while maintaining some degree of smoothness over the total population. The degree of smoothing applied to the 1990 carcass data was determined by a kernel estimator (a simple nonparametric technique sometimes referred to as Occam's razor) which considered both the central tendencies and the spread of the data over time. The kernel estimator, k(.), was simply a function used to weigh the contribution of surrounding data to the estimated density at each observation interval. These data were smoothed using a smoothing parameter and the area under the resulting curve calculated to give an estimate of total population. Kernel nonparametric estimators of density are based on
+-.,. -:A. .. independent observatidns+f-the population size (xi , i = 1,2, ...,n), in our case carcass counts. These observations have a common but unknown density f(x). The estimate of -. ':.... , : . . . ;:. -..-;Li" .-..-2. ;.:: .. . - --?a f(x), based on the kernel and a smoothing parameter (h), was calculated as:
where k(-)was a standard normal density function. Assuming a Gaussian kernel, the estimated population at any interval was calculated as: i
1 n 1 f(x;h) = - 1 * exp [ -(x - x;)~ / 2h2] nh i=l fl The estimated density at any point (x) was a weighted sum of the contributions from all the observations with the size of the contribution due to point xi being a function of the distance between x and xi.
This model was not only sensitive to the selection of a kernel estimator but more so to the choice of (h) the smoothing parameter (Noakes, 1989):As suggested by Noakes, the smoothing parameter was determined with a modified jack- knife likelihood estimator (Duin, 1976), which maximizes fi(xi;h) as h approaches zero. After smoothing the interval estimates, the probable spawning population was estimated by maximizing the density function. Nonparametric estimates of population size were calculated for Jolly-Seber data on the Noyo River coho and South Fork Eel River chinook. It was also possible to
- . .. - . - ..< ..- ~ -,. . j..__. . +--- - __ --_ . ~.- , ... ~ . .. ~ - ...... - - - ..-A- estimate run size nonparametrically szkcarcass counts on Indian Creek and Ten Mile River coho runs due to sufficient capture counts. These streams were not included in the Jolly-Seber analysis due to poor recapture histories. Practical application of these population estimates was determined by comparing results of the various models to a known population count made at the South Fork Noyo egg- taking station. The numbers of fish released above the weir at this stapion were recorded by the California Department of Fish & Game. The estimated population based on carcass surveys above the station were compared to this number as a total count and as a fishjmile estimate. The known distance (the sum of miles walked in all surveys) was used to indicate streams where maximum benefit was derived from carcass surveys (Table 1). In many streams where no carcasses were found, redds and live fish were counted. The lack of carcass data on 43 streams (15% of the total survey miles, or 500 miles) indicated a need to establish better survey methods which would eliminate recounting of redds (by flagging redd counts between surveys) and establishing uniform procedures for live counts to improve estimates using these two variables. The calculation of tags per mile (Table 1) included both hog-rings and tail punches. These data indicated the value of a dual tagging methodology. During carcass counts the fish head is often taken or removed from the carcass by predators before the tail. Tail tags helped increase our estimate of tag recoveries on 82% __ .i_. ,--_ of the streams where carcasses were recovered. The calculation of reddslspawner using total tag data was confounded by the lack of control in redd counts, but can be used to indicate relative spawning abundance where other data are lacking. On the Garcia River this ratio takes on the unusual proportion of 123 redds for each spawner captured as a carcass. This can be interpreted to indicate poor estimates of redd abundance or that surveys were conducted in good spawning areas but carcass recovery was poor due to unidentified factors. One significant contributing factor in estimating the number of redds per spawner is the fact that steelhead, unlike coho and chinook, do not necessarily die after spawning. In systems where steelhead spawn the number of redds can easily outnumber the number of carcasses found on the stream. Factors such as -. predation, timing of the surveys or post spawning distribution of carcasses to other parts of the stream by swift flows may have also influenced this data. In any case, -. in areas where this ratio is high efforts should be made to improve the redd counts and increase carcass capture The estimated populations calculated by the various models differed greatly (Table 2). In 73% of the streams where carcass data were obtained the area-under-the-curve model estimated fewer than 50 returning fish. Using live fish counts this same criteria applies to 68% of the streams. It is important to realize that these models are expected to underestimate the total population based on simulations (Shardlow et al., 1987). The data were collected in the fourth year of drought conditions in California and do not necessarily reflect the actual run size within each basin. Many fish may have spawned at sites lower in the basin than the surveys monitored. The numbers, however, seem low even if they are off by several orders of magnitude (Figure 1) Area-under-the-curve data calculated from carcass and live fish counts (Figures 2 & 3) showed variability in the effective estimates obtained on different creeks by these . two methods. Beidler & Nickelson (1980) indicated that live counts were frequently used on Oregon streams to estimate Table 2. Population Predictions 1989 -1990. (Estimates are based on data uhich fit deL assurptions. Live counts were calculated by species uhen identification uas clear, otherwise they were calculated as a total ~uczarea-under-the-curve; NPAUC=nonoarametric area-der-the-curve.) ktream Carcass Model Jolly-Seber Model Live Fish ~ountsl I Species AUC 95% CI NPAUC 95% CI AUC 95x1 UPAUC 95% CI AUC 95% Cl I 5. F.Big River unk. Caspar Creek coho Dehaven Creek sthd. Gualala River sthd. Garcia River sthd. Howard Creek sthd. Little River coho Noyo River coho Abave Egg St. coho Outlet Creek chinook Pudding Creek coho S.F. Eel River chinook Cedar Creek coho Dutch Charlie chinook Hollou Tree chinook coho total Irdian Creek chinook Redmountain chinook Redwood Cr. coho Ten Mile River chinook Coho total
TOTALS [AUC esr. 95% CI I chinook 520 474-565 coho 709 662-759
[NPAUC est. 95~~11 chinook 484 651-517 coho 2034 1979-2087
JoLly-Seber AUC est. 95%CI chinook 394 322-466 coho 871 732-1021
HPAUC est. 95XCI chinook 615 562-669 coho 3513 3440-3581
I~iveCounrs I AUC est. 957.CIJ chinook 714 648-779 coho 1076 1035-1123 sthd. 38 29-49 unkn. . 20 17-23 able 1. Fish Data 1989 -1990 (All per mile calculation based on total survey miles; carcass counts based on measurable fish; reddslspauner based on counts of carcasses and skeletons; ~st.=meinstem: Redu.=Reduocd Cr.; TerU.=Ten Mile Cr.; !Jirdn=~indemCr.) Stream total surrey live/ reddsl tags/ carcass cotmfs reddsl survey(mi) dates mile \ mile mile chinook coho rteeihead spawner Big River - South Fork 134.70 11 0.10 (0.78 0.00 0 0 0 0.00 Anderson Gulch Daugherry Creek Gates Creek Jeep Trail Creek Hettick Creek Ramon Creek Soda Creek Casper Creek mainstem North Fork South Fork Oehaven Creek Gualala River Garcia River Howard Creek Little River Hoyo River South Fork (above) S.F. a nst. bel lo^) Kass Creek N. Fork S.F. Parlin Creek Outlet Creek Wst. Eel (at mouth) Mouth.Longvale Longvale-101 br. Baechtel Creek Bloody Run Cr. Broaddus Creek Cherry Creek Davis Creek Dutch Henry Cr. Haehl Creek Long Valley Cr. Mill Creek Reeves Creek Ryan Creek Vi i l its Creek Pudding Creek Sourh Fork Eel R. lower (~edw.-tlcCoy) mi&ile(McCoy-Tern.) upper (ier+i.-Yindm.) Big Dan Creek Bridges Creek Cedar Creek Oeer Creek Dutch Charlie Cr. Hol lou Tree Cr. trap down trap up Bear Ual low Cr Bond Cr. Butler Creek Huckleberry Cr Michaels Cr. Reduocd Cr. Indi an Creek 30.30 Table 1. Fish Data 1989 - 1990 (cont.)
Stream total survey live/ reddsl tags/ carcass counts reddsl
survey(mi) dates mile , mile mile chimok coho steelhead spawner S.F. Eel (cont.) Jack of Hearts Cr. Kenny Creek Little Charlie Cr. Lou Gap Creek McCoy Creek Mill Creek Mud Creek Piercy Creek Rattlesnqke Cr. Curmings Cr. Elk Creek Foster Creek Twin Rocks tr. Rhtn. Creek Redwood Creek Rock Creek Section-4 Creek Ten Mile Creek Ten Mile River Mainstem Mill Creek Middle Fork Bear Haven L. Bear Haven Worth Fork Bald Hills Cr. Buck Creek Cavanaugh Cr. Little North F Patsy Creek Stanley Creek Unnamed Trib. South Fork Canpbell Cr. Churchman Cr. Gulch 11 Redwood Cr. North Fork Smith Creek Wages Creek abundance. Regression analysis using a linear model (Y=
a+bX) of live and carcass counts on the two streams where we
encountered the most carcasses (S.F. Eel R. & Noyo R.)
indicated poor relationships between these two variables. : Even when the regression was calculated with a time lag
of 1 to 3 weeks for carcass counts poor R-sqqared values e. .---7- were obtained. On the South Fork Eel River the best
relationship was found without a time lag (r2 = 0.62; F-
Ratio=13.3; ~~0.05).This suggested that on large rivers
live counts may be an appropriate relative measure of
abundance or at least as valuable a relative count as those done on carcasses.
>*:-- On the Noyo River the relationship between live counts
and carcass counts did not fare as well. The best fit was
found in the relationship calculated with a two-week time .- lag, comparing live counts to carcass counts two weeks later
(r2=0.21; F-Ratio=4.5). This relationship was not
significant (p>0.05) and suggests that live counts in
smaller coastal streams in Mendocino should not be used to
replace carcass counts. This is not to say that live counts
do not represent valid data, but that both methods would
give totally different estimates of population abundance.
Results of the model estimates calculated as fishlmile values (Table 3) was perhaps the best way to compare relative abundance between streams. Counts were normalized to the baseline stream miles walked in each survey. These Carcass AUC
chinook
Carcrss NPAUC
Jolly-Srbmr NPAUC
Lius counts QUC
(X 1800) estimrted population
Figure 1. Population estimates on surveyed Mendocino County streams for coho and chinook (AUC=area- under-the-curve; NPAUC=nonparametric area-under- the-curve; bars=95% CI). Chinook Hollow Tr-9 Cr. 1 carcme-
1 iu-
Dutch Charlie Cr.
SF Em1 Riu-r
Indian Cr.
Rmdmountain Cr.
Outlet Cr.
Tmn Milm River
Figure 2. Area-under-the-curve estimates (fishfrnile) for carcass arrd live counts of chinook (bars=95% CI). Cmdmr Cr.
Coho
Hollow Trmm Cr. Carcasm
Lium Rmdwood Cr.
Cwmpar Cr.
Little Riuer
Noyo Riumr total
Noyo Riumr aboue
Pudding Cr.
Figure 3. Area-under-the-curve estimates (fishlmile) for carcass and live counts of coho (bars=95% CI). Table 3. ~ish/survey-mileestimates (95% CI). Estimates base on baseline survey length (miles).
Carcass Carcass Jolly Jolly Live Stream Spcies AUC NPAUC AUC NPAUC AUC
Caspar Cr. coho L.4 (0.4) 6.1 (0.7) Cedar Creek coho 5.0 (1.8) 7.6 (1.7) Dutch Charlie chinook 2.7 (0.9) 9.1 (1.8) Hollow Tree coho 0.L (0.1) 3.5 (0.0 chinook 0.2(0.03) 1.7 (0.1) Indian Creek chinook 24.7 (1.7) 13.7 (0.7) 13.7 (1.6) Little River coho -22.- 0.5 C0.L) Noyo R. coho 29.7 (1.5) 98.6 (2.6) 43.3 (6.5) 174.8'(3.6) 46.3 (0.6) (total) Noyo R. coho B.O(O.4) 23.8 (0.5) 13.3 (0.7) 31.6 (1.7) 8.4 (0.4) (above) Outlet Creek chinook 1.2 (0.1) Pudding Cr. coho 0.6 (0.1) 4.7 (0.6) South Fork Eel chinook 21.2 (1.9) 21 (1.5) 21.3 (3.9) 33.2 (2.9) 30.4 (2.7) Ten Mile River coho 0.5 (0.1) 0.7(0.04) 0.6(0.04) chinook 0.8(0.04) 0.6 (0.1) Redmountain chinook 2.7 (0.3) 9.7 (1.0) Redwood Cr. coho 8.5 (0.8) 12.3 (1.5)
numbers gave a good indication of where the most spawning occurred by species in the Mendocino streams surveyed. The physical parameters measured on the streams estimated mean winter flows during the surveys to range
between 0.2 and 4.6 cubic meters per second (Table 4). The low flow of the winter of 1989-1990, due to drought conditions, probably influenced the spawning distributions in the basins by blocking or deterring upstream migration of spawning salmon except at times of brief freshets. This could be seen in the data for the South Fork Eel River where most of the spawning occurred in the lower reaches of the basin until a late January storm triggered movement of fish into the smaller upstream tributaries. A second factor to consider with low winter flow and carcass counts is increased vulnerability of spawners and
carcasses to carnivore predation (Table 5). In several cases the survey teams had to walk up and down stream during their survey, there being no direct route out at the headwaters of the stream. In this situation, three carcasses counted on the way up had completely disappeared from the streambed one or two hours later on the return walk. These observations imply a bias in the estimates of mean carcass duration on any creek using only carcasses which escape predation. An improvement in the carcass duration estimate for each creek would make area-under-the-curve calculations more effective. Mean water temperatures taken during the 1989-90
surveys were relatively consistent, ranging from 5.3 to
8 OC, with the exception of Bridges Creek on the S. F. Eel ..- where stream temperatures averaged 12.3 OC. This may simply be due to the time of day water temperatures were taken. Water and air temperatures need to be standardized to time of day in future surveys. Winter water temperature was not thought to be a factor influencing spawning migration during the winter of 1989-1990. Snowfall on several surveys may have hampered carcass observation and recovery in the Eel River basin. Because of drought conditions, other climatic factors incluencing survey accuracy were minimal this year. Table 4. Physical Parameters 1989.1990 (cont.) stream Obs. Survey Mean Mean HZ0 Mean Alr Mcan Pool Observable Cycles Miles FLou (cm) Temp. ( C) Tenp. ( C) Oepth tcm) Depth (%) . Eel (cont.) Kenny Creek 8 2.80 1.1 (0.7) 5.8 1. 7.4 (2.3) 83.40 76% Little Charlie Cr. 1 1.10 0.30 5.60 11.00 45.00 100% Lou Gap Creek 5 1 .OO 0.4 (0.4) 8.3 (1.2) 10.6 (2.0) 90.00 93% nccoy Creek 5 1.50 0.8 (0.7) 8.3 (1.6) 10.6 (2.5) 62.50 73% MilL Creek 1 0.70 na 8 10 55.20 100% Mud Creek 6 1 .LO 1. 0.7 5.8 (0.8) 7.4 (4.3) 77.90 54% Piercy Creek 9 1.50 0.7 (0.4) 7.9 (1.7) 10.L (3.3) 81.50 88% Rattlesnake Cr. 7 7.80 0.8 (0.7) 6.5 (1.9) 11.4 (3.1) 76.80 82% Cmings Cr. 2 0.30 0.39 6.50 5.00 64.00 54% Elk Creek 4 1.60 0.6 (0.3) 7.5 (2.1) 11.3 (3.3) 103.34 86% Foster Creek 1 . 1.00 na ,,a . na 45.00 100% , 2- Twin Rocks Cr. 4 1.80 1.0 (0.8) 7.4 (2.3) 9.6 (2.9) 69.20 63% Redmtn. Creek 8 3.00 2.0 (1.0) 7.3 (1.5) 9.0 (2.2) 88.30 58% Redwood Creek 1 I 1.30 0.5 (0.4) 6.4 1.7 7.7 (3.1) 90.70 83% Rock Creek 1 0.60 0.04 na na 25.00 100% Section-4 Creek 1 0.50 na na na na na Ten Mile Creek 6 3.00 1.7 10 4.8 1.11.7 (4.4) 132.40 86% Ten Mile River 12 69.40 0.8 (0.6) 7.5 1.3 9.9 (2.7) 109.30 88% nainstem 3 3.40 17(1.5 6.9 (1.2) 8.3 (4.0) 100.00 89% Mill Creek 5 2.00 0.4 (0.3) 8.1 (1.5) 8.4 (4.9) 35.00 64 % Middle Fork 9 9.00 2.1 (1.2) 7.9 (1.7) 11.0 (2.3) 133.50 76% Bear Haven 4 2.50 0.5 (0.4) 7.2 (2.4) 8.4 (5.1) 82.90 100% L. Bear Haven 2 0.30 0.4 (0.3) 9.00 10.00 60.80 93% North Fork 8 15.50 1.5 1.2 7.1 (1.1) 9.6 (2.2) 141.70 81% Bald Hills Cr. 4 2.30 0.7 1. 8.3 (2.0) 10.7 (2.5) 34.00 88% Buck Creek 4 1 .OO 0.4 (0.2) 9.3 (1.6) 13.3 (5.1) 80.00 70% Cavanaugh Cr. 2 0.50 0.20 10.00 12.80 50.00 80% Little North F 9 3.50 0.8 (0.5) 7.4 (1.2) 10.6 (4.8) 76.20 8 n. Patsy Creek 4 2.50 0.5 (0.4) 6.00 7.80 51.34 81% Stanley Creek 4 1.00 na na na na na UnnamedTrib. 1 0.30 0.10 10.00 6.70 35.00 100% South Fork 10 15.20 1.4 1.9 7.1 (1.8) 9.6 (4.0) 112.30 85% Campbell Cr. 8 2.00 0.4 (0.3) 7.7 (1.7) 9.5 (4.1) 71.30 81% Churchman Cr. 7 1 .70 0.6 (0.2) 7.8 (1.8) 10.2 (4.0) 70.60 91% Gulch 11 5 0.80 0.2 (0.1) 7.8 (1.7) 9.8 (4.0) 90.00 81% Redwood Cr. 5 3.00 0.6 (0.5) 6.0 (1.6) 7.7 (5.5) 6A.80 96% North Fork 3 1.00 0.20 8.00 6.00 55.00 88% Smith Creek 9 4.50 0.4 (0.31 7.3 (1.9) 10.2 (2.7) 66.90 75% Wages Creek 8 4.80 1. 07 7.1 (2.7) 7.5 (3.1) 108.00 75% 7Table 4. Physical Parameters 1989.1990 (Msr.=Mainsrem: Redw.=Redwood Cr.: Term.=Termiie Cr.: 1st ream Obs. Survey Mean Mean HZ0 Mean Air Mean Pool observable1 1 Cycles Miles Flow (cm) Tenp. (OC) Tm. PC) De~th(cm) Depth (%)I gig River - South Fork 11 18.30 0.61 (0.6) 6.65 (1.6) 9.75 (2.6) 8L.60 71% Anderson Gulch Daugherty Creek Gates Creek U~amdTributary Mettick Creek Ramon Creek Soda Creek Caspar Creek Mainstem North Fork South Fork Oehaven Creek Guaiala River Garcia River Howard Creek LittLe River Noyo R lver S.F. & Hst. belou South Fork above Kass Creek N. Fork S.F. Parlin Creek Ourlet Creek Msr. Eel (at mouth Mouth-Longvale Longvale-101 br. Baechtel Creek Bloody Run Cr. Broaddus Creek Cherry Creek Oavis Creek Dutch Henry Cr. Haehi Creek Long Valley Cr. Mill Creek Reeves Creek Ryan Creek Uillirs Creek Pudding Creek South Fork Eel R. lower (Red".-McCoy) middle(McCoy-Term.) upper (TeMi-Uim) Big Dan Creek Bridges Creek Cedar Creek Deer Creek Dutch Charlie Cr. Hollow Tree Cr. trap down trap up Bear Uallov Cr Bond Cr. Butler Creek Huckleberry Cr Michaels Cr. Redwood Cr. Indian Creek Jack of Hearrs Cr. 11 Table 5. Observed salmon carcass predation.
Stream n Bird Man Raccoon Wild Bear Otter Unknown Car Caspar Creek 6 1 0 1 0 0 0 4 Gualala River 3 0 0 0 0 0 3 0 Noyo River 115 36 0 3 0 2 1 T3 Outlet Creek 3 0 1 0 0 0 2 0 South Fork Eel 82 L9 2 4 1 9 5 12 Cedar Creek 1 1 0 0 0 0 0 0 Hollow Tree 8 0 0 0 0 4 2 2 Indian Creek 4 0 2 0 0 0 0 2
Piercy Creek 3 0 0 0 0 -- .. P2 . . 1- 1. '1 .O Ten Mile River i3 3 0 0 0 '~-: 2 8 0
TOTAL 238 90 5 8 1 19 22 93 (%) 3m 2% 3% .1X 8% 9% 39%
On ten streams the mean portion of the water column visible to survey crews was less than 60%. This may have influenced the effectiveness of survey counts on these streams. Pools contributed 36% of the carcass retention
(Table 6) in our surveys and visibility at depth may affect carcass recovery in pools. The mean value given for observation at depth may not be as effective an indicator of relative turbidity found on streams as the range of values. In many cases survey reaches were turbid on one survey while the remainder were absolutely clear. If this turbidity occurred during periods of peak spawner mortality this may have influenced the estimate for that stream. Observation at depth should be considered a relative value related to the effectiveness of carcass counts and not a true measure of turbidity in any particular stream. Table 6. Carcass retention by habitat.
-Habitat
Stream Pools Riffles Eddy Bank Shrubs Debris Roots Bar Caspar Creek 5 0 0 1 1 0 1 3 Dehaven Cr. Gualala River Garcia River Houard Creek Nayo River Outlet Creek Pudding Creek South Fork Eel Cedar Creek Dutch Charlie Cr. Hollow Tree Indian Creek Lou Gap Cr. Piercy Creek Rehuntain Redwood Cr. Ten Mile River
Results bv Stream South Fork Noyo River (514 T18N R17W) Two species of anadromous fish were observed on the
South Fork Noyo River, coho salmon (Onco~hyncbusk/su~ch)and steelhead trout (Amyvkj~). Coho were found in the river from 11/30/89 to 2/28/90 (Figure 4), the entire survey schedule for this stream. Steelhead were found in the S.F. Noyo from I FEMALE
MALE
YEFIR i
YEAR C
8 20 4 0 6 8 8 8 180
% SEX WITHIN EACH AGE CLASS
Figure 6. Percent age-class distribution of males and females on the-Noyo River 1989-1990. number or new tags
Figure 5. New carcass tag counts by date for the Noyo River, 1989-1990. Figure 4. Observation dates for anadromous fish on the South Fork.Noyo River.
Surveys on the Noyo River were divided into two major sections by a weir located at the egg-taking station operated by the California Department of Fish and Game. Results were analyzed for stream miles surveyed below the egg-station weir, including Kass Creek, as data collected "belowM the weir. The South Fork Noyo above the weir to the spillway at McGuirels pond, and tributaries above the weir were analyzed jointly as the site "above" the weir. New carcass tags in the section below the weir showed a bimodal distribution on the Noyo River (Figure 5). Peaks were seen on 12/19/89 and on 1/22/90. New carcass tags above the egg station peaked between 1/22/90 and 1/26/90. A randon set of scales taken from coho above and below the weir were read for age at spawning. Scales from 26 female coho were examined. Twenty-one female coho (81%) were estimated to be three years of age (Figure 6). ...- Of+.the,J+~~*-.: . .- - ,. -- -- .~, -- - ., .- ___-- males whose scales were read, 23 (72%) were estimated to bee- 3-years old. Mean spawner length by sex and age showed no significant difference between males and females 3 years of age (Mann-Whitney U test; p>0.05). Two year old females were, however, significantly larger (n=5; mean=45.6 cm; SD=4.8) than their male counterparts (n=9; mean=39.7 cm; SD=3.2; Mann-Whitney; p=0.04). Length-frequency distributions plotted by sex show more males at both extremes of the distribution function (Figure 7). Fish and Game classified coho with fork lengths less than 56 cm as grisle (W. Jones, pers. comm.). Most estimates of fish length were made visually at the station and not verified by actual measurements. Fish smaller than 56mm in length were thought to be precocious males returning after only one summer at sea. Fish meeting this size criteria were found above and below the weir on the S.F. Noyo. While males predominated, females less than 56cm were also found. Below the weir 25. male (9.8% of the population tagged below) and 12 (4.7%) female grisle were tagged (Figure 8). Above the weir 10 males (17.2%) anC6 females (10.3%) were tagged. The count of grisle taken at the egg station equaled 294 or 29% FEMALE MRLE
Figure 7. Length-frequency distribution of coho carcasses found-in 1989-1990 on the Noyo River. of the population found in the trap. A higher proportion of grisle were taken at the station than were found in the population at large. The proportion of female grisle in the carcass data may reflect subjective criteria on the part of observers. Survey crews may have called any fish without a developed kype
(hooked jaws bsually indicating sexual maturity in males) a I .-..: - female. But review of the data, based on scale-age and sex,
showed that in most cases (except 2) when a grisle was identified as a female another was identified at the same site as a male, indicating other criteria were probably used to sex these smaller'fish. In the future, residual eggs or milt should be set as the standard for sexing carcasses. Smaller coho in the S.F. Noyo deserves further investigation to determine true age and sex of this proportion of the population. Mean carcass duration on all streams was calculated from tagging histories compiled from consecutive surveys. This estimate is critical to the area-under-the-curve estimates. Mean carcass duration on the S.F. Noyo varied
throughout the surveys (Figure 9). Despite a similar bimodal trend in retention data, there was no correlation between the mean time a carcass remained in the sample population and carcasspabundance (Kendall rank correlation coefficient
=o. 04 ; p>o. 05) .
/ The South Fork Noyo River was the only stream where a control count of the population allowed a true test of the BELOW GBOUE
Figure 8. Grisle (<56 cm) above and below the egg- taking station as a percent of respective population by sex. Nourrnbmr
Drcembcr
Januar~
Fmbrurry
Figure 9. Mean coho carcass retention time (days) on the Noyo River-by survey period, 1989-1990 (bars represent 1 standard deviation of the mean). population models. When the known numbers of fish placed above the weir into the upper S.F. Noyo were compared to model results the Jolly-Seber interval counts with a nonparametric under-the-curve estimate of the total run was the only model to come close to the actual count (Figure 10). Even with 95% confidence intervals the other models tended to underestimate the population. Adults and grisIe (319 and 91 respectively) were released above the weir at the Noyo egg station. The 58 tags placed on carcasses above
the weir represented only 14% of the released population. An initial release from the egg station on 11/27/89, was not picked up as carcass data in the upper drainages. The fate of these fish remained unknown. Some fish may have moved downstream over the weir after their release. However this possibility can not be confirmed. Fish were not tagged at the station before their release upstream. Downstream movement over the weir may have contributed to the underestimated population count in area-under-the-curve model on the upper S.F. Noyo River. Despite pulsed releases from the egg station (Figure ll), the recapture data exhibited a Gaussian frequency distribution. Records from the Department of Fish and Game indicated that 214,230 coho fry were released into the South Fork Noyo in 1987. These fry would return during our surveys as age- three spawners. Assuming the scales taken at random from the population represent the true age class distribution, 73% of Ye 3511 spawners (Jolly-Seber nonparametric estimate, see ==timat.
control
i Figure 10. Comparison of model estimates and control count of the coho population above the egg-taking station weir on the Noyo river, 1989-1990 (bars represent 95% CI for model estimates). 'page 20) returning to the S. F. Noyo were age-three. By
'adding the spawners sacrificed at the egg station (lib males and 246 females), we calculated the ratio of return to release as 1.33% for age-3 coho. This does not factor in any natural production. Kass Creek (19%) and the lower South Fork Noyo River (61%) contributed significantly to the Noyo carcass count, suggesting substantial natural production in this part of the basin. A tagging program would help to quantify the actual coho return due to CA Fish & Game's enhancement policy.
A comparison of'the results of two versions of the
Jolly-Seber nodel (models A & B) showed an improvement in the estimated interval values calculated by Model-B (Model-B recruitment sum=79 fish; Model-A recruitment sum=22), with time-specific capture probability (Goodness-of-fit tests: Model B vs A; total chi-square=100.8; df=8; p<0.001).
South Fork Eel River (S10 T4S R3E)
Three anadromous species of salmon were found on the South Fork Eel, including coho, steelhead and chinook
( 0ncoi'hpnct.m bi~wyixch) . Coho were first observed on the S . F. Eel on 12/19/89 and last observed on 1/25/90 (Figure 12). Chinook first and last observations were 12/4/89 and 2/27/90, respectively. Steelhead were first seen on 2/28/9, 3t the end of the surveyZcycle for the S.F. Eel River.
R3E) to McCoy Creek, a middle section from McCoy to one mile above Rattlesnale Creek and an upper section from Ten Mile
Creek to Windem creek. Most major tributaries within these
reaches were surveyed but will be discussed individually in
later sections. The 217 chinook carcass counted on the S.F.
Eel mainstem were tagged in the lower (64%) and middle
reaches (36%). Distribution according to section walked in surveys (Table 7) shows a concentration of carcasses from below Richardson Grove State Park to Mccoy Creek.
Figure 12. observation dates for anadromous fish on the South Fork EeL River. R3E) to McCoy Creek, a middle section from McCoy to one mile above Rattlesnale Creek and an upper section from Ten Mile Creek to Windem Creek. Most major tributaries within these reaches were surveyed but will be discussed individually in
later sections. The 217 chinook carcass counted on the S.F.
Eel mainstem were tagged in the lower (64%) and middle
reaches (36%). Distribution according to section walked in
surveys (Table 7) shows a concentration of carcasses from
below Richardson Grove State Park to McCoy Creek.
coho
Figure 12. observation dates for anadromous fish on the South Fork EeJ River. Table 7. Carcass recovery on the mainstem South Fork Eel by reach (TOT=total; SCL=number of scales taken; M2 to M5=males age 2 to 5; F2 to F5=females age 2 to 5).
REACH STR TOT. SCL. H2 H3 H4 H5 F2 F3 F4 F5 UNK &way 10 45 3E 6 1 0 3 0 0 1 0 1 0
Garberville 24 55 3E 1
Senbob! 34 4S.3~ 7
Wagner 1 5s 3E 7
Richardson 7 5s 4E 2
101 Bridge 13 55 3E 45
Indian Cr. 24 5s 3E 33
McCoy Cr. 35 5s 3E 38
Redmtn. 6 24W IN .19
Bridges Cr. 17 24N IN 24
Mile 64 21 24N 1N 5
Stardish H. 33 23H 1N 18
Leggert 10 23H 1N 9
Hermat ige 19 23H 16V 2
Termile Cr. 20 23N 16V 1
TOTAL 217
All scales collected from the chinook carcasses were examined for age. The distribution of age classes ranged from two to five year (one scale read by USFWS was considered a year-6 fish, but this reading was not confirmed by independent readings of other scales from the same fish). Fish fork length was positively correlated to fish age (Kendall's tau=0.63; p<0.05). The average fork length for
I YEAR 2 YEFIR 3 YEFIR 4 YEAR 6
Figure 13. Mean chinook spawner length given by age- class for the South Fork Eel River, 1989-1990 (bars represent one standard deviation of the mean). age 3, 4 and 5 chinook showed significant overlap (Figure
13). With the exception of age-2 fish, carcass data suggested that a higher proportion of females returned to
the S.F. Eel survey sites (Figure 14). Length-frequency distributions for chinook and coho carcasses found on the S.F. Eel River show a high proportion of males (coho=56%; chinook=14%) returning as grisle with fork lengths <56 cm (Figure 15). Based on recapture histories the Jolly-Seber model estimated between 562 and 669 chinook spawned in the two
lower sections of the mainstem S.F. Eel (see Table 2). As mentioned earlier, this number probably reflected the drought conditions experienced during the winter of 1939- 1990. The average retention time documented for chinook carcasses on the S.F. Eel River was eight days. Eight chinook carcasses found on the mainstem S.F. Eel were judged to have missing adipose fins. Examination of snouts taken from these fish revealed only two coded-wire tags (Bob
Reavis, Ca. Fish & Game, pers. comm., 1990). It was interesting to note the habitat factors contributing to carcass retention on the S.F. Eel and Mendocino streams in general. Unlike reports from the literature where large organic debris caught and held 56-67%
of surveyed.coho carcasses (Cederholm & Peterson, 1985;
Cederholm, et al. 1989), on the S. F. Eel, woody debris retained only 4% of the chinook carcasses (Figure 16). Deep pools contributed substantially to chinook carcass recovery 1 FEMFILE
MALE YEAR 2
YEAR 3
YEAR 4
YEAR 5
8 18 20 3 8 4 0 5 0
% OF AGE-CLASS BY SEX
Figure 14. Males and females as a percent of the population given by age-class for chinook on the S.F. Eel River- (1989-1990) . COHO
36 38 40 42 44 46 48 50 52 54 56 55 68 62 64 66 68 70 72 74 70 78 FISH FORK LENGTH (cm>
CHINOOK 0FEEHHLE t4RLE EUNIDENTIFIED
FISH FORK LENGHT (cm)
Figure 15. Length-frequency distributions for coho and chinook on the S.F. Eel River, 1989-1990. on the S.F. Eel. This trend was true in all surveys regardless of species or stream size (see table 6). It is also interesting to note the carnivores feeding on chinook carcasses in the S.F. Eel (Figure 17). Bear, 2*::;-*5 5 q+?+$: , .--. ,-- * -"-* erz.7 raccoon and otter were the primary mammalian consumers of chinook carcasses observed or documented by our study. Bear typically left the lower head or jaw (presumable after consuming the brain), gills, guts and caudal fin. Bears also frequently left the fish's air bladder. Raccoons and otters tended to peel the carcass skin back to consume the muscle leaving skin and skeletal remains. Otters were the only predator documented to kill spawning salmon before they had spawned. Despite the lack of documentation from this study, nocturnal predation by mammals must have played a major role in the dynamics of the food chain supported by salmon carcasses (Cederholm, et al., 1989). Birds were the most common predator observed feeding on carcasses in the S.F. Eel. Bald eagles, red-tailed hawks, great blue herons and egrets were seen eating salmon carcasses during our surveys. Chinook carcasses found in the vicinity of Wilderness Lodge have been documented as extraordinarily large fish (Bill Trush, pers. corn., U. C. Berkeley, 1988). The one chinook skeleton found in the Branscomb area (1/25/90) during our ,1989-90 surveys could not be measured. However, three carcasses over 130-cm were found in the reach directly below the old Nature Conservancy property after the Thanksgiving Day storm, 1988 (J.L. Nielsen unpublished POOL
RIFFLE
CODY
BANK
SHRUBS
DEBRIS
ROOTS
BAR
Figure 16. Carcass petention by habitat as a percent Of the total count, S.F. Eel River, 1989-1990. BIRD
MAN
RFICCOON
BEAR
OTTER
UNKN
% obsarvad carcass predation
~igure.17.Carcass predation given as a percent of the total observations of predation on the S.F. Eel River, 1989-1990. data). These fish may represent a unique run of large chinook still exists in the upper reaches of the S.F. Eel.
Fish of this size ma~d~~&-$g~a$jtpt~~Aa_tE?~c.F.-,- ~.. , . .-. . spawning, longer ocean residence, and extensive ocean feeding cycles. Runs of large chinook were called "Tyee" by the Native Americans of the Pacific Northwest. Tyee runs in Washington and Oregon have not beendo?&%@--=i-nce-.the 1950's and are thought to have disappeared (Dr. E. Salo, pers. comm., U. Washington, 1985). Every effort should be made to document this run and secure its survival in the S.F. Eel.
Big Dan Creek (S23 T23N R17W)
Big Dan Creek is a tributary of the S. F. Eel River (S23 T23N R17W). This stream was surveyed 5 times from 11/29/89 to 2/22/90. The terrain was steep and water levels very low. There was no sign of fish in this tributary during any survey. Two redds found on the final survey were assumed to be steelhead redds.
South Fork Big River (S33 T17N R15W) Big River is a coastal stream with its mouth at the town of Mendocino. The South Fork and seven associated tributaries.were surveyed in 1989-1990. Tributaries surveyed included Ramon Cr., Mettick Cr., Anderson Cr., Daugherty Cr., Soda Cr., Gates Cr. and one unnamed tributary. Carcasses and skeletons were not found during our surveys. Live fish counts and redds were the only indicators of salmonid use of the..-S,F. Big River, Four live fish were observed in Ramon Cr. in 1990. These fish were thought to be two female coho, one grisle and'ori5'coho of'unknown sex, but no confirming carcasses were found on subsequent visits.
Based on these cou~ntsthe -area-under-the-curve model' estimates 17 to 23 fish spawning in Ramon Cr. Four live fish were observed (1/20/90) in the river above Hansen's School. One lamprey was found in the upper South Fork around Hell's Gate.
Thirteen redds were noted on Ramon Cr. between 1/25/90 and 2/22/90. Six redds were observed on Daugherty Cr. on 2/12/90. Redd counts on the mainstem S.F. Big river equalled 58. These redds were observed from 12/15/89 to 2/22/90, with the first peak count found in December on the lower South
Fork mainstem from the mouth of Ramon Crek upstream about 1 mile. The second peak redd count was found in January on the stretch up and down stream from Hansen School. Comments made on the mainstem South Fork included excellent spawning gravels and good holding pools in the vacinity of Ramon Creek. Johnson Creek, a tributary of the S.F. Big River had a coho enhancement project in place from 1981-1987. .About 2500 coho fry were reared and released from this site in 1987. The mouth of the culvert on Johnson Creek was estimated to be 5 ft. above the water surface of the South Fork effectively establishing a blockage for returning coho. Spawning activity, based on observations of redds in the upper South Fork, was heaviest around Hansen School.
These spawners may have resulted from the small release in.. *~-, . - -?r.?- Johnson Cr. in 1987, or natural propagation due to earlier. plants from the same facility. Hillside erosion, turbidity and log jams (on Gates and Soda Creeks) were commented on by surveyors of Daugherty Cr. and its tributaries. Observation at depth was poor (mean= 59% of the water column). This basin has been the site of
recent active commercial logging. Beyond the 6 redds counted on Daugherty Creek below the mouth of Gates Creek, no other sign of active spawning was observed.
Bridges Creek (S21 T24N R17W) Bridges Cr. is a tributary of the S. F. Eel River. This creek was surveyed three times from 1/10/90 to 2/26/90. The stream was considered too steep with few deep holes to be provide quality spawning areas. Several slides were recorded from the surrounding hillsides. Sediment and logs at one slide created a 7 ft. drop in the creek flow and trapped water underground for some distance upstream effectively blocking any fish passage. No fish or redds were observed on this creek. Caspar Creek (S1 T17N R17W) Caspar Creek flows through Jackson State Forest to the coast in Xendocino. Caspar was surveyed 11time during the winter of 1989-1990. Total survey miles equaled 26.5 miles on the mainstem below the forks, 13.3 miles on the North Fork and 9.4-miles on the South Fork. Only one carcass was tagged on Caspar Creek, a coho on the mainiYemT' ,yo.. : .-- - rr>w-.i steelhead skeletons were also tagged. Skeletons counts provided sufficient data to estimate 30-35 spawning coho
(area-under-the-curve) and live counts estimated 38-43 live spawners (species unknown) in Caspar Creek. There was no obvious trend in the timing of live counts and carcasses on Casper (Figure 18). Redds, however were observed in the mainstem on the first survey (11/28/89). Redd activity peaked on the mainstem from 1/19/90 to 1/31/90. Redds were first observed on the South Fork on 1/19/90 and the North Fork on 1/26/90. The number of redds observed on the mainstem was several times greater than those found on either fork (Figure 19). The weirs operated by United States Forest Service on the North and South forks of Caspar Creek have been thought to be a blockage to salmon spawning nigrations (A. F. Grass,
CA Dept. Fish & Gane, pers. comm, Ft. Bragg, 1990). Improvements were made to the fish passage at the weirs in 1989. The spawning data Collected in 1989-1990 found two live fish and three redds above the North Fork weir. All live fish and redds observed on the South Fork were found Figwe 18. Live fish, redd and carcass counts on Caspar Creek, 1989-1990. REDO5
CARCASS ,
MAINSTEM N. FORK 5. FORK
Figure 18. Live fisif, redds and carcass counts by site, Caspar Creek, 1989-1990. below the weir. This reach of stream was probable not surveyed sufficiently. Subsequent surveys for juvenile coho during the summer of 1990 on the South Fork indicated a substantial population above the weir. Estimates of juvenile coho abundance above the weir were 0.25 fish/m2 and 0.04 fish/m2 for the South Fork and the North Fork respectively
(Rod Nakamoto, U.S. Forest Service;Redwood Sciences Lab., Arcata, CA, unpublished data).
Cedar Creek (514 T23N R17W) Cedar Cr. is a tributary of the S.F. Eel River. Six surveys were conducted on this creek from 11/29/89 to 2/22/90). One carcass was tagged, a coho (1/19/90). Four additional skeletons were counted between 11/29/89 and 1/29/90. Area-under-the-curve models using these data estimated 11-23 coho spawning on Cedar Cr. Live counts estimated 20-33 fish. However, two live steelhead were seen on 2/12/90 and redds were not seen on Cedar until the last observation on 2/22/90.
Deer Creek (S4 T21N R16W) Deer Creek, a tributary to the S.F. Eel River, was surveyed only once during 1990. A local resident said despite historic salmonid use of Deer Cr., four families now pump drinking and washing water from the creek and it dries up in late summer. The resident said "no fish have been seen in this creek for the last two years." No sign of spawning use was seen during our survey.
Dehaven Creek (S19 T21N R17W) Seven surveys were made on Dehaven Creek from 11/28/89 to 2/13/90. No carcasses were tagged, only one skeleton . f -.--.-- (unknown species) was found and 8 redds were counted. Live fish counts analyzed by the area-under-the-curve model estimated only 4 spawning adults in Dehaven Creek. Two log jams were found, one each in Section 22 and 23. The reach walked from the North Fork downstream to the ocean was considered to have excellent spawning gravel and good riparian cover, but no fish.
Dutch Charlie creek (S9 T21N R16W)
Eight surveys covering i7.9 miles were done on Dutch Charlie Creek, a tributary of the S.P. Eel River. These surveys ran from 12/4/89 to 2/20/90. Two chinook carcasses were tagged and one skeleton found on Dutch Charlie. Both carcasses were female. Area-under-the-curve estimates were 5-8 chinook and 16-24 live fish (species unknown) spawning on Dutch Charlie. The gravel in Dutch Charlie as thought to be excellent for spawning. Redd counts started on 1/17/90 and lasted to
2/7/90. A count of 0.7 redds per mile was made on Dutch Charlie Creek. South Fork Garcia River (S31 T12N R15W) The S.F. Garcia River was surveyed from the mouth
upstream 2 miles on 6 occasions from 11/30/89 to 2/22/90. No carcasses were tagged and only one skeleton (unknown species) was found (1/31/90). This fish was identified as a
female and scales aged her as a 4-year- old. Live fish were 1 - ;"*V ..+s..?'ZF - ,2 - =.k observed on the S.F. Gatcia om 1/17/90 to 2/22/90. These fish were identified as steelhead by the survey crews. Redds were observed in this reach from 1/17/90 to 2/22/90, peaking on the last survey indicating that these were probably steelhead redds. Numerous juvenile salmonids were observed during these surveys and assumed to be last year's steelhead. Flemming Creek was considered to have excellent habitat supporting many juvenile steelhead.
Little North Fork Gualala River (S23 TllN R15W) The Little North Fork of the Gualala River was surveyed from the mouth upstream 2.9 miles. Two live steelhead were observed on 1/23/90. Two steelhead carcasses were tagged on 1/31/90, one male and one female. Redds were observed in this reach of river from 1/11/90 to 2/22/90. Numerous small salmonids were observed in the pools in this reach, probably steelhead yearlings. A local sports group has been operating a small steelhead hatchery on Doty Cr. for five years. Adult steelhead are removed from the river by hook and line and placed in flow-through tubes until ripe. After milking the live adults are returned to the river. Eggs are reared on site and released the following sprlng into the stream. Three log jams, not considered to be total blockages at this time were found approximately 0.5 miles above Doty Creek. Otter predation was documented on the steelhead carcasses which ended up as skins on the stream bank. -- -
Hollow Tree Creek (S10 T23N R17W) Hollow Tree is a tributary of the S.F. Eel River. In addition to 20 miles of mainstem surveyed on Hollow Tree, six tributaries were surveyed. These tributaries included Redwood Cr. (S9 T22N R17W), Bond Creek (S15 T22N R17W), Michaels Creek (515 T22n R17W), Huckleberry Creek (523 T22N R17W), Bear Wallow Creek (S23 T22N R17W) and Butler Creek (525 T22n Rl7W). Coho salmon and chinook salmon were observed in the Hollow Tree basin. Chinook were first observed on 12/5/89. Coho were first observed on 1/24/90. Both species were observed in Hollow Tree until 2/13/90 (Figure 20). Six chinook and 14 coho were tagged on Hollow Tree Cr. An additional 71 skeletons were found. Area-under-the-curve calculations did not increase the carcass counts, estimating 6-7 chinook and 14-17 coho on Hollow Tree Cr. Area-under- the-curve estimates based on live counts (undifferentiated species)'estimated 146-158 spawning fish on this stream. coho
chin
Figure 20. Observation dates for anadromous fish on Hollow Tree Creek, 1989-1990.
Six chinook scales were read from Hollow Tree. One fish was considered a 5-year old, two were 4-years old and one each were considered 3-year and 2-year old spawners. Four of these chinook were located on the mainstem of Hollow Tree Creek between the location of an old egg taking weir and the new trap. The 2-year old, located in this area, was a male only 47 cm long (a grisle). The last chinook carcass was found on the mainstem between the mouth of Huckleberry Cr.
< and the mouth of Butler Cr. The Salmon Restoration Association of California (SRAC) runs the egg taking station on Hollow Tree Creek. During the 1989-1990 season 20 chinook adults and 22 chinook grisle
(<56 cm) were released above the weir. This represented the total population of chinook caught at the station. For the
first time in the fristcry of-the Hollow-Zree~statiom~:Lr. z. . . I _~_<_ r .I_.- ~ .. - chinook were spawned at the stationdue to the low numbers
of returning adults (Larry Barcley, president SRAC, pers. comm., 1990). The 20 chinook captured returned to the station in sexually exclusive groups. Three adult males and 12 grisle arrived first and were released from the egg station. During this period (11/28/89 to 1/12/90) only one chinook female reached the weir. She was released on 1/10/90
due to her ripe condition. Between 1/13/90 and 1/14/90, 11
males, 10 grisle and 2 females arrived. All fish, except 4 males, were released upstream. No silvers were caught until 2/2/90 - 2/4/90, when 1 female and 2 males were captured. At this time, these fish
and the 4 males being held at the station were released back into the stream above the weir. Differential migratory behavior in males and females could be a critical factor affecting natural production and enhancement facilities in populations with low numbers of returning adults. The policy of releasing small nuxbers of returninb females at the Hollow Tree station was appropriate, allowing natural production in the stream to take precedence over artificial production when numbers are low and the proportion of returning females remains uncertain. Efforts should be made to examine the factors contributing to the coordination of timing in male and female fish as they reach natural spawning sites. Aggregation of fish in holding.pools in mainstem habitats ~ -.. , ?...&. . . . -.---.*-.-- - . may be a critical factor in the natural mixing of the sexes. Loss of these pools due to aggregation of the stream channel and low flows during drought cycles may contribute to sexually different migration patterns. In May, 1987, chinook fingerlings (99,322) were released from the Hollow Tree station. Of this release 22,035 were placed in Mule Cr., a tributary of Hollow Tree and fed until fall. Coded-wire tags were placed in 8,185 of these fish. The one chinook snout taken on mainstem Hollow Tree below the weir did not contain a coded-wire tag. The
1987 brood year ( 1988 releases) included 189,600 fry. These
fish could returnI as grisle in the 1989-90 spawning run. Of these fish 43,862 were tagged as Hollow Tree releases. One tagged grisle was the only documented return of tagged fish found at the station. The 22 chinook grisle trapped and released above the weir were not recovered as carcasses. The adipose of the one grisle carcass found below the weir was intact, indicating an untagged fish. Despite excellent production levels of juvenile chinook, return data on Hollow Tree releases were pobr for the 1989-1990 spawning season Coho carcasses (no.=2) were found on the mainstem below the old welr and from the mouth of Redwood Cr. to the mouth of Bond Creek (no.=ll). Live fish (n=4), redds (n=ll), skeletons (n=3) and one coho carcass were found in Redwood Creek above the junction with Hollow Tree Creek. One coho carcass was found on Huckleberry Creek. Two of the mainstem carcasses were grisle (<56 cm in length). Based on capture- recapture data, the average carcass duration for coho on Hollow Tree was 9.5 days. Coho released above the weir in 1989-90 numbered 162,
53 males, 87 females and 22 grisle. Of the 11 coho tagged as carcasses above the weir, 2 were tagged on 2/13/90, after the weir was removed at the egg station (2/9/90). The remaining 9 coho indicated a 5.6% carcass recovery (Figure 21). Coho were spawned on two occasions at the Hollow Tree station. Eggs from these spawns were sent to Shelter Cove and Warm Springs for off-site rearing to juveniles and eventual release in the S.F. Eel River. Some progeny from the Hollow Tree brood stock reared at Warm Springs are released into the Russian River (W. Jones, pers. corn.). Historical release numbers from these sources were not available for this report. Considerable enhancement in the form of barrier removal has been done on Bond Creek, a tributary of Hollow Tree Creek. Six surveys were conducted on Bond Creek between 1/12/90 and 2/27/90'. One steelhead skeleton found on this tributary (2/12/90) was the only indication of anadromous 0 3 0 6 0 90 120 160 100
rimh counts
Figure 21. Chinook and coho released and recaptured , above the Hollow Tree Creek Station (12/5/89- 2/9/90). spawning in 1990. Two live chinook, 4 redds and 2 chinook skeletons were found on Butler Creek on 1/22/90. Other tributaries (Michaels, Huckleberry and Redwood) indicate coho spawning based on skeleton, live and redd counts. One skeleton and five new redds were found on Butler on 2/20/90, suggesting steelhead use. Steelheadspawning-was:afso observed on Bear Wallow Creek.
Howard Creek (S18 T21N R17W)
Howard Creek, a small coastal stream, was surveyed 8 tines between 11/28/89 and 2/20/90. A live steelhead was first observed on 1/29/90. Only two steelhead carcasses were tagged (2/13/90). Redds were observed on Howard from 1/23/90 to 2/20/90. No other activity by salmonids was observed on Howard Creek in 1990.
Indian Creek (S35 T5S R3E) Indian Creek is a tributary to the South Fork Eel River. This stream was surveyed 11 times from 11/29/89 to 2/26/90, covering a total of 30.3 stream miles. Baseline stream miles (3.0 mi) represents approximately 113 of the drainage potentially utilized by spawning salmonids. Five chinook'carcasses and 32 skeletons were counted. Based on recapture data the average carcass duration was 4.5 days on
Indian Creek. Area-under-the-curve estimates for the total chinook run were 69-78 chinook based on carcass and skeleton counts and 36-47 chinook based on live counts. A nonparametric estimate was obtained (39-43) by combining carcass and skeleton data. The estimate appears low compared to the area-under-the-curve estimate. This is probably due to an extreme outlier point in the data (Figure 22). This point represents one survey date (2/8/90) where 38% of the population was tagged. This indicates one problem with the nonparametric model. Outliers inflate the estimate of the smoothing factor
(h) resulting in over smoothing and inadequate estimates of the total populations (Noakes, 1989). Noakes suggests modifying the contribution of extreme points to the likelihood function by weighting the sample based on the standard deviation (SD) of the mean. Data points falling within one standard deviation are weighted by a factor of 1, points outside one SD are given a weight inversely proportional to the number of standard deviations between their count and the mean. Using this method the estimate becomes 53-57 chinook. Live counts on Indian Creek peaked on 1/23/90 (Figure
23). These observations may have included steelhead. A second cycle of redds were observed around 2/20/90, probably from steelhead spawning. The live counts estimated 41 fish of undifferentiated species spawning in Indian Creek. count.
+ smaothrd
0 2 4 6 8 10 12
time indsx
Figure 22. Indian Creek carcass and skeleton data for chinook showing the smoothed line used to calculate the total population by the nonparametric area-under-the-curve model. 8 5 10 15 20 25
counts
Figure 24. Live fish, redds and carcass counts for Jack ; of Hearts Creek, 1989-1990. Little Charlie Creek (S4 T2lN R16W) Little Charlie Creek was walked only once at the start of the observation cycle on the S.F. Eel River. No sign of anadromous fish use was seen on this visit.
Little ~iver(57 T16N R17W) Little River is a coastal stream in Mendocino County that flows through Van Damme State Park. Only two live fish were observed on Little River. These fish were seen in the lower basin and were identified as coho. Redds were observed throughout the basin from 1/17/90 to 2/9/90, peaking between 2/5/90 and 2/9/90. Subsequent surveys for juvenile coho estimated approximately 0.17 coho per m2 of surface area (W. Jones, CA DF&G, unpublished data). This suggests more frequent surveys should be made on Little River to better estimate the population using carcass data.
LOW Gap Creek (S24 T5S R3E) Low Gap Creek is a tributary of the S.F. Eel River.
This creek was walked 5 times between 12/5/89 and 1/30/90. No live observations, skeletons, carcasses or redds were counted on this creek. Observation of the first 0.5 miles of creek from the mouth on 1/30/90 recorded high turbidity. Observation at depth was calculated as only 9%. McCoy Creek (56 T24N R17W) McCoy Creek flows into the S.F. Eel River. Despite comments by local landowners that many fish used to spawn in this creek, one jaw bone found on the last survey was the only sign of anadromous use during our 5 surveys (12/4/89-
2/8/90). Water clarity was poor on the 1/30/90 survey, . :..L +Gp7+y -..-~. observation at depth was calculated as 13%. Flows were thought to be very low but good spawning gravels were found throughout the creek.
Mill Creek (S31 TZlN R14W) Mill Creek is a'tributary of the S.F. Eel River. This stream was visited once (1/25/90) as a extension of the survey of the surrounding mainstem S.F. Eel. One skeleton of unknown species was found on the 0.75 stream miles surveyed. No other sign of anadromous use was documented on this stream.
Mud Creek (S26 T21N R16W)
This tributary of the S.F. Eel River was surveyed 6 times between 11/30/89 and 2/20/90. Two live fish were observed on the stream on 1/17/90, their species was unknown. Mud Creek was unexpectedly clear when it rained and murky on the surveys made on clear days. Rainwater tends to dilute the mud springs located on this stream. On average, visibility at depth was 54% of the water column (range on clear days 13%-67%). One local lifetime resident of this creek (Mrs. Jean Tosten, Branscomb post mistress) recalled chinook, coho and steelhead spawning on her property in this creek in years past, but none in recent history (4-5 years).
Outlet Creek (531 T21N R13W)
Surveys on Outlet Creek included 8.1 miles of mainstem and 34.8 miles of tributaries. The tributaries surveyed included: Baechtel Cr. (5.1 mi); Bloody Run (1.0 mi);
Broaddus Cr. (5.4 mi); Cherry Cr. (1.0 mi); Davis Cr. (3.0 mi); Haehl Cr. (2.5 mi); Longvalley Cr. (5.9 mi); Dutch
Henry Cr. (1.5 mi) ; Ryan Cr. (0.4 mi) ; Reeves Cr:-tl; @ mi) ;
Upper Little Lake (lower Baechtel Cr. below the sewage treatment plant, the Baechtel overflow channel to Mill Cr. and lower Mill Cr., 4.0 mi) ; Willits Cr. (4.0 mi) .
Two chinook carcasses were tagged in the Outlet Basin.
.--,-;- - One on the mainstem below Cherry Cr. and one at the mouth where Outlet Cr. joins the mainstem Eel River. In addition to these carcasses, 16 skeleton were counted in the Outlet basin. Otter predation was often noted in these surveys, especially on Outlet Cr. There were numerous salmon skins found without heads or tails.
Live counts estimated from 58-66 chinook spawning in the Outlet Basin (area-under-the-curve model). Redd counts were made in the upper reaches of Outlet Cr. above the gravel pit (no.=2), the lower mainstem around Cherry Cr.
(no.=23), the mainstem around the mouth of Longvalley Cr
(no.=25), on Bloody Run Cr. (no.=l), on Longvalley Cr. i Figure 25. Red counts on Outlet Creek given by survey date, 1989-1990. (no.=5), on Baechtel Cr. (no.=15), on Broaddus Cr. (no.=6) and on Mill Cr. near the sewage plant (no.=l). Redd counts peaked around 1/5/90 (Figure 25). Comments made during the stream surveys in the Outlet basin included many concerns about the habitat and water quality in the tributaries around the town of Willits. Old car batteries were found piled in Baechtel Cr., chlorine smelling water was common downstream from the treatment plant on Broaddus Cr., numerous livestock carcasses were found piled in Davis Cr., tires, old cars and other trash were found in Haehl Cr. and behind the pistol range on Mill Cr. another pile of mammal-carcasses was found dumped in the creek. Water quality was judged poor due to oily mud and silt found in many reaches. The California Department of
Fish & Game was notified of these findings. It is important to realize that based on the comments of local residents all of these creeks supported salmon and trout until recently. The spawning populations appear to have declined rapidly in the last two years due in part to drought conditions.
Piercy Creek (S35 T5S R3E) The 9 visits made to Piercy Creek found one coho carcass and 2 dead steelhead. The coho and one steelhead were found on 2/8/90. The second steelhead was found 0: 2/26/90. Two live fish of unknown species were seen - on 1/16/90. An additional 4 live steelhead were observed between 2/12/90 and 2/20/50. Live counts and carcass data were insufficient to estimate total populations beyond the actual counts. Redds were seen on this creek from 1/16/90 to 2/20/90, with a peak count between 2/8/90 and 2/20/90. Redds were observed above several large log jams noted on this creek which were originally though to be potential blockages. Bear predation was documented on the coho carcass found on this creek.
Pudding Creek (54 T18N R17W) Pudding Creek was surveyed 8 time between 11/28/89 and 2/8/90. The stream was walked from the railroad tunnel to the headwaters near Ramsey Ridge. During these surveys only one coho grisle (42 cm) and 4 coho skeletons were found. The carcass was brought back from the field due to controversy over its species. It had several characteristics usually associated with chinook (gray gums and spots on the lower dorsal). The consensus, after some discussion, as to the species was a young mature coho male (grisle). Coho skeletons were tagged from 1/26/90 to 2/8/90. Redds were found throughout Pudding Creek with the highest counts occurring in the lower basin. Based on total survey miles it was estimated that 1.57 redds per mile were dug by coho in Pudding Creek. Peak live observations were recorded at the same time as peak redd counts (Figure 26). Live counts estimated 38-50 coho spawning in Pudding Creek in 1990. Subsequent surveys a 18 28 38 4 8
counts
Figure 26. Live fish, redds and carcass counts for Pudding Creek, 1989-1990. on Pudding Creek for juvenile coho found coho utilizing the total survey area during the summer of 1990. Estimates of abundance for juvenile coho on Pudding Creek ranged from 0.12 coho/m2 in August (J.L. Nielsen, unpublished data) to approximately 0.03 coho/m2 in October (W. Jones, unpublished data). ~hesedata indicate more spawning activity than the carcass and skeleton counts would confirm.
Rattlesnake Creek (520 T23N R16W)
Seven surveys were conducted on Rattlesnake Creek and 3 tributaries between 11/29/89 and 2/22/90. The tributaries walked were Elk Cr., Cummings Cr. and Twin Rock Cr. The total stream miles walked during these surveys equaled 47.6 miles. Only one live fish of unknown species was found on Rattlesnake Creek (1/19/90). One anadromous lamprey was found on Twin Rock Creek. Pools in this basin were not deep, often filled with silt and fine gravels. Spawning areas were heavily armored and considered poor. Local residents reported steelhead spawning in Cummings Cr. in February and March. Fish (steelhead) were reported to have spawned in Elk Cr. up to 2 years ago but not last year. No redds were observed during our surveys.
Red Mountain Creek (S17 T24N R17W)
* This tributary to the S.F. Eel River was surveyed 8 times between 1/10/90 and 2/26/90. Five chinook carcasses REDDS
CARCASS
8 2 4 6 8 18 12
counts
Figure 27. Live fish, redds and carcass counts for Red i Mountain Creek, 1989-1990. were tagged in the lower basin between 1/23/90 and 1/30/90. Live counts were high between 1/16/90 and 1/23/90, just prior to the peak carcass counts and the first redd counts (Figure 27). Chinook did not hold long in Red Mountain Cr. before spawning, suggesting the January rains may have been a factor in their movement from the mainstem into the tributary to spawn. The 2/20/90 count of live fish were judged to be chinook. One skeleton found on 2/26/90 confirmed a late chinook spawning run in Red Mountain Creek. Area-under-the-
curve estimates using live counts equaled 26-32 spawning . chinook in Red ~ountainCreek in 1990.
Redwood Creek (516 T21N R16W) Surveys done on Redwood Creek near Branscomb began on 11/30/89 and ended 2/20/90. Eleven surveys were conducted from the mouth upstream approximately 1.3 miles. Six coho carcasses were tagged on Redwood Creek. Only two of these were recaptured on subsequent surveys. One dead steelhead carcass was found during the same survey that four of the coho were tagged (1/25/90), suggesting an overlap in the . spawning of these two species on this creek. Area-under-the-curve estimates for total spawners (unknown species) using live counts equaled 34-38 fish. Live counts, 'redds and carcasses were all found for the first time on the 1/10/90 surbey (Figure 28), despite the fact 8 4 8 12 16 28
counts
Figure 28. Live fish, redds and carcass counts for , Redwood Creek, 1989-1990. that the stream had been surveyed 7 days prior to this visit. Fish moved quickly into Redwood Creek. This stream was thought to have good spawning gravels, sufficient flow and plenty of cover to support a large spawning population of coho. Comments on undercut banks and woody debris suggest good coho juvenile habitat as yell. One survey which extended up into the North Fork of Redwood Cr. found many log jams and swampy areas. The extent to which juvenile coho use this upper area should be determined before any debris removal directed at improved access for spawning adults is done. Areas such as this are often used by coho as overwinter habitat when mainstem flows are elevated.
Rock Creek (S9 T21N R16W) This tributary to the S.F. Eel River was surveyed only twice (11/30/89 and 12/4/89). No sign of spawning was observed during these surveys. Numerous small salmonids (48-
72 mm) were seen in the pools. section-4 Creek (533 T12N R14W) This tributary to the S.F. Eel River above Branscomb was walked only once on 1/25/90. There was no sign of anadromous fish use recorded during this survey. Angular gravels observed on this creek were thought to be poor spawning quality. Ten Mile Creek (516 T22N R16W) Ten Mile Creek was surveyed from the mouth upstream
13.9 miles on 6 occasions (11/30/89-2/22/90). TWO tributaries of Ten Mile Creek were surveyed, Streeter Cr. (1.5 mi) and Big Rock Cr. (1.5 mi). No carcasses were found during these surveys. Redds were not observ&;eib;.thesa. . ..z -.* --=. - . . ~ .- 2/22/90 survey on the mainstem of Ten Mile Cr. above Laytonville.
Ten Mile River This coastal stream was surveyed for spawning salmon on 12 occasions between 11/28/89 and 2/28/90. Three anadromous species were found spawning in Ten Mile River, coho, chinook and steelhead (Figure 28). Coho were first observed on 11/30/89 and last seen in the river on 2/13/90. Chinook were first found as cascasses on 1/18/90 and last seen on 2/13/90. Mill Creek, a tributary of Ten Mile River was surveyed for two miles on five occasions. Fishermen reportedly caught three steelhead on Mill Creek on 1/19/90. Spawning steelhead were not seen on Ten Mile River until 1/30/90 and were observed through 2/21/90. Based on carcass and skeleton data identified to species the area-under-the-curve model predicted 29-33 chinook and 31-36 coho spawners on Ten Mile River. he nonparametric model estimated 51-57 chinook and 49-55 coho spawners in %en Mile River. Live counts predicted between 80-92 spawners of unidentified species. Figure 29. Observation dates for anadromous fish on Ten Mile River, 1989-1990.
There is a long history of chinook enhancement on Ten Mile River. Chinook were introduced into the basin for the first time in 1979. In that year approximately 350,000 chinook yearlings from the Trinity River spring chinook run were released into the mainstem below the confluence of the Middle Fork. In the next year, approximately 199,000 fingerling (200/lb) destined for Hollow Tree Cr., escaped from holding ponds on Ten Mile River and entered the mainstem. The next introduction of chinook was made in 1981 when 20,000 "Wisconsin" chinook originating from the Sacramento River stock were released into the river. In 1982 an additional 95,000 "Wisconsin" chinook were released. Since that year, trapping of returning spawners and pond rearing operations on Ten HU-River- have supplemented the chinook run with fry relea&-& i-11-1985 (2,000 fry), 1986 (5,000 fry) and 1987 (9,000 fry) . The results of this enhancement activity established a natural spawning run which continues today. Scales taken from chinook carcasses on Ten Mile River indicated most spawners were 4-years old (no.=3). One 3-year old and one
fish thought to be 5 years old were found as carcasses on Ten Mile River. Three chinook carcasses were found on the South Fork, one on the North Fork and one on the Little North Fork, indicating a wide distribution of chinook spawning in the basin. Recent coho enhancement has included the planting of 6,000 coho juveniles in June, 1987. Scales from only one coho were read. This female coho was )-years old. Data on the recovery of coho carcasses, skeletons and redds came primarily from the lower Middle Fork and the lower South Fork of Ten Mile River. Redds found on Bear Haven in late January suggested some of the coho skeletons found on the Middle gork may have come from this tributary. Survey notes stated that the gravels on Bear Haven were superior to those in the Middle Fork. Slxteen skeletons and 36 redds were found on Campbell Creek, a tributary of the South Fork Ten Mile river. On Smith Creek we found two live fish of unknown species, 18 redds and one skeleton. Considerable habitat enhancement has been done to improve spawning passage in the tributaries of Ten Mile River. In the 1970's and 1980's stream rehabilitation in the form of barrier removal was done on 7 tributaries. Redwood Cr. and the upper South Fork have seen extensive restoration work removing total stream blockages. Although no carcasses were tagged in this area, live coho were seen, 18 redds counted and 2 skeletons found. Churchman Cr. had blockages removed in 1982 and 1983.
Three live coho spawners were seen on Churchman, 2 redds counted and 1 skeleton found. Buckhorn Creek was cleared of a blockage but no sign of anadromous use was found in the 1990 surveys. Little Bear Haven was also opened to spawning migration by extensive barrier removal. Redds were found on this stream in late February, indicating probable steelhead use. Twenty log jams were removed from Bald Hills Creek but no sign of anadromous use was indicated by our surveys. Cavanaugh Cr. was also cleared of blockages but no sign of anadromous use was found. The scale taken from a large steelhead carcass was read for life history patterns by the Craig Tuss of the U.S.
Fish & Wildlife Service and Dr. Rodger Barnhart and Dr. Thomas Hassler, two scientists from the California Cooperative Fisheries Research Unit, Humboldt State
University. This steelhead female (104 cm) had five .a+@&\-i.;;,r.... and had completed six sumers of growth. The first two summers were spent in freshwater, the second two in the ocean. At the end of the fourth summer she spawned for the
first time returning to the ocean f3r another summer. A . . -~~ -. . second spawning check was found followed by another summer in the ocean (1989), her sixth season of growth. Her return to freshwater in 1990 indicated a third spawning migration after which she was found as a carcass on the South Fork of Ten Mile River. The scale was a valued addition to the collection at the Research Unit.
Wages Creek (S29 T21N R17W) This coastal stream was surveyed 8 times between
11/28/89 and 2/20/90. A total of 25.3 miles were walked on Wages Creek. No carcasses were found. Live spawners were first observed on 1/18/90 and identified as steelhead. During this survey fresh steelhead eggs were found on the bank. The area-under-the-curve estimate for live counts on
Wages Cr. was 22 spawners (95% CI=20-25). DISCUSSION The primary motivation behind the inception of this
- 3 ,7 =.. .o*-C' '-1 ; y<"-' ' : ,,; ...... ~ ... ~ . - . study for all parties involved (the authors, the funding agency and the participants) was to better understand the results of historic enhancement and rehabilitation efforts on the streams and rivers of Mendocino County by estimating . -, ,+ -r-.~:7:i.:'~:$ye-- --.z.2==. ~- ---i. .. the current status of anadromous salmonid runs in selected streams. There was great discussion and controversy as to which streams would be surveyed during the 1989-1990 spawning season. Despite generous funding, surveys had to be limited to the 355 miles of anadromous stream included in this report due to limited resources. Our study area represented only 18% of the potential anadromous stream miles in Mendocino County. Total anadromous miles in Mendocino County were estimated in 1979 (Sherr & Griffin, CDFG) as 1980 steelhead miles, 1092 coho miles and 430 chinook miles (W. Jones, personal corrmunications). Many streams with vital salmonid runs were left out because other agencies or organizations were currently surveying them. Others were not included because historical records indicated a lack of anadromous fish and/or restoration efforts in the basin. We apologize in advance for any omissions which may seem critical to the reader. This data was designed to be the seeg of a more comprehensive database on the fishery resources of Mendocino*County to be compiled by the California Department of Fish and Game. We solicit input on streams which were not covered in this research which have received rehabilitatiomefforts in the past and still support anadromous salmon. We will make every effort to document additional streams in the future and add them to the database. Information sharing is critical to the success of such a project-2%Xher-agencies, individuals and companies are encouraged to contact the authors with information pertinent to the project. Meetings will be arranged where the corrpatibility of data and study design can be discussed.
The analysis of carcass counts and spawning
. observations to measure trends in escapement has been controversial since its inception (Brett, 1952; Ricker,
1958). Most methods developed in the late 1950's (AUC, area- under-the-curve models) have changed little despite the fact it was well reported that they tend to underestinate actual escapement (Sheridan, 1952). As a consequence the reliability of escapement measurements based on carcass data has been questionable and the resulting estircates considered relative. Where populations were primarily the result of hatchery releases, extensive tagging programs have been initiated to gain more precision in escapement estimates. However, estimates of escapenent have been especially difficult for streams with small natural spawning populations (Beidler &
Nickelson, 1980; Cederholm & Peterson, 1985). In Mendocino where the numbers of anadromous salmon have been declining steadily in recent history, all that is left are streams with small spawning populations. It was critical to our objectives to compare the available methods for quantifying these populations. In small rivers migrating adult salmon often show a bimodal distribution with an early spawning group and a late spawning group (Banks, 1969), which accelerates the error in estimating the total population. With small counts a slight error can have enormous computational implications in the under-the-curve model. The time between surveys can impact population estimates using AUC models, since this factor affects the probability of carcass recovery. Cederholm et al. (1989), suggested that the average time a coho carcass remains- in the stream was 4 days. The current study indicated that coho carcasses remained on average for 6 days and chinook carcasses 8 days on Mendocino streams. Our observations, however, were probably an overestimate since they were based on tag recoveries and did not take into account the portion of the population which was lost to predators before initial tagging. Differences in the estimate of carcass duration were critical to the AUC model and may be the major contributing factor to their underestimation of small populations. The above argument is also true for estimating populations from live counts using AUC. Estimates of average spawner life in the river are scarce and need more research. This is especially true in light of the staggered migratory patterns which develop during the low flows of winters
droughts in California. <..~-..-. Live spawners are also difficult to see and even more difficult to count with any reasonable accuracy. Differentiation of species is questionable with live count
data made from bank observations. Directsnorkel~abservatton*.~:s-z would enhance live counts (Shardlow et al., 1987). In this
study, the probability of counting a live chinook from the
bank was 18%. Snorkel observation probability for chinock
was 92%. During the winter, this nethod is dangerous and dependent on trained technical observers. Live counts made from the bank were,'however, the only data available on the spawning runs in several of our survey streams. These data were used to demonstrate occurrence, timing of the run and relative abundance, but not as actual estimates of the spawning population size. Live counts have been used exclusively in some carcass
surveys (Beidler & Nickelson). The relationship between live counts and carcass counts was not significant for the Noyo River. Only the chinook counts on the S.F. Eel were significantly related to carcass counts, suggesting that in large river systems live counts may be an appropriate measure of population. Estimates of population size using the Jolly-Seber model (Jolly, 1965; Seber, 1965) allow for increases in population due to new recruits (spawner deaths in our case) and changes in the probability of recapture over time. These factors theoretically provide a more efficient means of estimating the population parameters contributing to total estimating populations. This model, however, has been criticized because its flexibility is provided by estimated parameters (Cormack) and useful precise estimates are hard to obtain (Arnason & Baniuk, 1978). Recent-developments in open population computer models offer several adaptations of the Jolly-Seber model (JOLLY; Pollock, et al, 1990) which allow parameter estimates using specific controls which tend to improve precision (Jolly, 1982). Model-B from JOLLY holds survival rate constant per unit of time, but recapture probability was time-specific. This parameter control seemed appropriate to our carcass data with the following assumptions: 1) the probability of initial carcass discovery and tagging, i. e. carcass survival (in this case a true oxymoron), was equal for all members of the population; 2) the probability of tag recovery was specifically related to the interval of time since last tagging. We used JOLLY to predict the number of individuals in the population on any given survey. These numbers were then used in the standard AUC model and a nonparametric smoothing model (Noakes, 1989) to estimate total population size. The results of the test of these models on the control population above the egg-taking station on the Noyo River confirm the underestimation expected from AVC models. The success, however, of the nonparametric model using the JOLLY estimates does not suggest it as a panacea for all population estimates on small streams.
Recapture data on the control population lacks records for a large release made on 11/11/89 (see Figure 11). These fish were released above the weir before our surveys began.
If these fish were fresh when released it is possible that they spawned above the weir and should have contributed to our carcass counts. An estimated 11.3 days as live spawners and 6 days as carcasses puts them within reach cf our first survey above the weir (11/27/89). However, no recoveries were recorded.
Although application of estimates that allow for nonlinearity between survey counts (i.e. nonparametric estimates) seem intuitively correct, carcass counts made during this period would inflate the JOLLY estimates of the population above the weir. This in turn would probably lead to an overestimate of the total population with the nonparametric approach. Tests of this model with better controls (i.e. no releases prior to surveys) would help confirm the effectiveness of this model.
A second factor not built into the model estimates but critical to evaluation of the numbers of returning spawners is the effect of habitat on spawning and carcass recovery.
The fact that 80% of the spawning population was found below the egg-taking station may not be the result of natural e factors. Intense natural spawning by coho below a weir is common in most hatchery operations (J.L. Nielsen, unpublished data). Many fish will resist the constrictive passage leading to the holding ponds at-hatcheries.-This is ;itr .- .t- true for natural obstruction in streams as well, where coho will spawn in the first available habitat when their migration becomes blocked. On the Noyo River, fish arriving after the first wave of spawners may have found the best spawning sites occupied and selected to move downstream to spawn rather than enter the egg-taking station. The value of Kass Creek, a tributary below the weir, as a coho spawning site predates the Noyo weir (A. Grass,
CDF&G, pers. comm., 1990). However, the proportion of the total population utilizing this spawning habitat was not evaluated prior to the construction of the weir. The number of fish coming from the enhancenent program on the Noyo which spawn in Kass Creek, potentially competing with wild spawners, also remains unevaluated. A study of these fish using coded-wire tags would help us understand the interactions between the wild and enhanced populations. It is also important to consider habitat contribution to carcass counts. The variation in habitat from stream to stream will naturally affect the available spawners which contribute to the carcass population, i.e. carcasses will tend to concentrate around sites with quality spawning gravels. In our surveys we were careful to visit sites where critical concentrations were expected. But this assumes a fully seeded spawning population and normal flow conditions. Our surveys identified many areas with what was judged excellent spawning habitat and no fish. This lack of fish at critical spawning sites may be due low population levels or migration blockage by low flows. The evaluation of spawners per mile would be more critical if it vas compared to an estimate of available spawning habitat per mile and utilization throughout the basin. Carcass retention is also dependent on habitat. In our study deep pools contributed significantly to carcass recovery. Woody debris was not as critical to carcass retention in Mendocino County as reported in studies from
Washington (Cederholm, et al. 1989). Once again these data should be compared to the availability of deep pools and woody debris on Mendocino County streams before general assumptions are made from the results. The results of this survey should be considered in light of the ongoing drought conditions in California. These data represents an evaluation of returning anadromous salmon to streams impacted by low flows for four consecutive years. This climatic recession of the aquatic habitat comes at a time when fish populations were still suffering from the effects of anthropomorphic impacts on the environment. The low population levels predicted by even the best model cannot compare to historical referendes of the salmon runs in Mendocino County. Evaluation of enhancement activities undertaken in the < last few years is complicated by these effects. On one hand, we cannot report any true success stories. But on the other hand, if barrier removal or outplanted fry on a Mendocino County stream have contributed to even one pair of spawning coho or chinook, it may have increased the probability of salmon survival in the local streams. It is difficult to analyze the cost-effective value of enhancement operations when existing salmon runs are low and naturally spawning salmon may be priceless. However, without returning spawners all enhancement efforts will be money wasted. Carcass surveys were not sufficient to estimate the populations remaining on all of the streams looked at in Mendocino County. Juvenile salmon abundance on Caspar creek and Pudding Creek suggest larger populations than the carcass data estimated. It will be critical to add juvenile abundance and habitat data to the spawning results to evaluate enhancement effectiveness and secure the survival of natural runs. By recording the various habitat and developmental needs during different life history stages, we may begin to understand the factors contributing to salmon survival in Mendocino. Instead of analyzing "limiting factors" it is time to develop "contributing factors" which can lead to enhancement approaches which fit the existing opportunities. These factors may be totally different from those historically used to sustain large numbers of anadromous fish LITERATURE CITED
Arnason, A.N. & L. Baniuk (1978). POPAN-2, a data maintenance and analysis system for mark-recapture data. Charles Babbage Research Cntr., St. Pierre, Canada. Banks, J.W. (1969). A review of the literature on the upstream migration of adult salmonids. J. Fish Biol. 1:85-136.
Beidler, W.M. & T.E. Nickelson (1980). An evaluation of the Oregon Department of Fish and Wildlife standard spawning fish survey system for coho salmon. ODFW report Dec. 1980 (Xerox) . - . Brett, J.R. (1952). Skeena River sockeye escapement and distribution. J. Fish. res. Bd. Can. 8:453-468.
Cederholm, C.J. & N.P. Peterson (1985). The retention of coho salmon (Oncoi~~nch~'sk~sti~ch)carcasses by organic debris in small streams. Can. J. Fish. Aquat. Sci. 42: 1222-1225.
Cederholm, C.J., D.B. Houston, D.L. Cole & W.J. Scarlett (1989). Fate of coho salmon ( L%~o~bflcb~~i/~~'kh) carcasses in spawning streams. Can. J. Fish. Aquat. Sci. 46:1347-1355.
Cleveland, W.S., S.J. Devlin & E. Grosse (1988). Regression by local fitting: methods, properties and computational algrithms. J. Econometrics 37:87- 114. Cormack, R.E. (1972). The logic of capture-recapture estimates. Biometrics 28:337-343. Duin, R.P.M. (1976). The choice of smoothing parameters for Parzen estimators of probability density functions. IEEE Trans. Comput. C-25:1175-1179.
Gazey, W.J. & M.J. Staley (1986). Population estimation from mark-recapture experiments using a sequential Bayes algorithm. Ecology 6?(4):941-951.
Hightower, J. E. & R.J. Gilbert (1984). Using the Jolly- Seber model to estimate population size, mortality, and recruitment for a reservoir fish . population. Trans. Am. Fish. Soc. 113:633-641. Jolly, G.M. (1965). Ekplicit estimates from capture- recapture data with both death and immigration - a stochastic model. Biometrika 52:225-247. Jolly, G.M. (1982). Mark-recapture models with parameters constant in time. Biometries 38:301-321. Kope, R.G. (1987). Separable virtual population analysis of Pacific salmon with application to marked chinook salmon, OncofhyncbusLrhawy/~c,i~,from California's Central Valley. Can. J. Fish. Aquat. sci. 44:1213- 1220.
Leopold, L.B., M.G. Wolman & J.P. Miller (1964). Fluvial processes in geomorphology. W.H. Freeman, San Francisco CA.
Moyle, P.B, J.E. Williams & E.D. Wikramanayake (1989). Fish species of special concern of California. Special Report, CA DF&G, Rancho Cordova Contract No. 7337. Noakes, D.J. (1989). A nonparametric approach to generating inseason forcasts of salmon returns. Can. J. Fish. Aquat. Sci. 46:2046-2055. Platts, W.S. et al. (1987). Methods for evaluating riparian habitats with applications to management. U.S.D.A. Forest Service Gen. Tech. Rpt. INT-221. Intermountain Research Station, Ogden, Utah. 177pp.
Pollock, K.H., J.D. Nichols, C. Brownie & J. Hines (1990) Statistical inference for capture-recapture experiments. Wildl. Monogr. No. 107:l-97. Ricker, W.E. (1958). Maximum sustained yields from fluctuating environments and mixed stocks. J. Fish. Res. Bd. Can. 15:991-1006. Schlosser, I.J. (1982). Fish community structure and function along two habitat gradients in a headwater stream. Ecolo. Monogr. 52(4):395-414.
Seber, G.A.F. (1965). A note on the multiple-recapture census. Biometrika 52:249-259. Seber, G.A.F. (1982). The estimate of animal abundance and related pakameters. 2nd Ed. MacMillan Press, N.Y. 654pp.
Shardlow, T., R. Hilborn & D. Lightly (1987). Components analysis of instream escapement methods for Pacific salmon ( Uneorbynchus spp. ) . Can. J. Fish. Aquat. Sci. 44:1031-1037. Sheridan, W.L. (1962). Variability in pink salmon escapements estimated from surveys on foot. U.S. Fish & Wildl. Serv. Spec. Rpt. Fish. 408:l-7. Smith, A.K. (1973). Development and application of spawning velocity and depth criteria for Oregon salmonids. Trans. Am. Fish. Soc. 2:312-316. Spall, J.C. (1988). Baysian analysis of time series and dynamic models. Marcel Dekker, Inc. N.Y. 536pp.
Sykes, S.D. & L.W. Botsford (1986). Chinook salmon, Onco~dynchusLcb~r$scha, spawning escapement based on multiple mark-recapture of carcasses. Fish. Bull. 84:261-270. REACH LOCATION Ten Mile Cr.-Hwy 101 to Streeter Cr. (1.7 mi) Ten Mile Cr.-Streeter Cr. to Big Rock Cr. (1.5) Ten Mile Cr.-Big Rock Cr. to Laytonville (4.7 mi) Ten Mile Cr.-Laytonville to Hwy 101 (3.2 mi) Streeter Creek mouth up (1.5 mi) Big Rock Creek mouth up (1.5 mi) SF Eel-Ten Mile Cr. to Wilderness Lodge (2.3 mi) SF Eel-Wilderness Lodge to Jack of Hearts (1.7) Jack of Hearts Creek mouth up (2.8 mi) SF Eel-Jack of Hearts to Dutch Charlie (2.9 mi) SF Eel-Dutch Charlie to Redwood Cr. (0.9 mi) Dutch Charlie Creek mouth up (2.2 mi) Redwood Creek (Branscomb) mouth up (1.3 mi) Xenny Creek mouth up (2.8 mi) SF Eel-Redwood Creek to Kenny Creek (2.5 mi) SF Eel-Xenny Creek to Branscomb (1.0 mi) SF Eel-Branscomb to Windem Creek (7.2 mi) Deer Creek mouth up (0.5 mi) Section-4 Creek mouth up (0.5 mi) Outlet Creek mouth to Cherry Creek (2.8 mi) Outlet Creek-Cherry Cr. to tunnel (2.5 mi) Outlet Creek-tunnel to Longvale (2.9 mi) Outlet Creek-Longvale to Reeves Canyon (4.2 mi) Outlet Creek-Reeves Canyon to 101 bridge (3.9 mi) Cherry Creek mouth up (1.0 mi) Bloody Run Creek mouth up (1.0 mi) Longvalley Creek mouth to Dutch Henry Cr. (0.8 mi) Longvalley Cr.-Dutch Henry to Sherwood Rd. (2.5) Longvalley Cr.-Sherwood Rd. up (2.6 mi) Dutch Henry Creek mouth up (1.5 mi) Ryan Creek mouth up (0.4 mi) Reeves Canyon mouth up (1.0 mi) Willits Creek sewage plant to dam (4.0 mi) Broaddus Cr. mouth to weigh station (5.4 mi) Baechtel Cr. sewage plant up (5.1 mi) Haehl Creek sewage plant to 101 bridge (2.5 mi) Davis Creek-Berry Cr. to East Hill Rd. (3.0 mi) Baechtel overflow channel complex (4.0 mi) Howard Creek mouth up (2.0 mi) Dehaven Creek mouth up (3.0 mi) Wages Creek mouth to fork (1.8 mi) Ten Mile River-Mill Cr. to Middle Fork (3.4 mi) Mill Creek (Ten Mile R. ) mouth up both forks (2.0) South Fork Ten Mile River mouth to bridge (1.5 mi) SF Ten Mile R.-bridge to gravel pit (2.2 mi) SF Ten Mile R.-gravel pit to Redwood Cr. (9.0 mi) SF Ten Mile R.-Redwood Cr. to headwaters (2.5 mi) Smith Creek mouth to headwaters (4.5 mi) Campbell Crgek mouth to headwaters (2.0 mi) Churchman Creek mouth to headwaters (1.7 mi) Redwood Creek mouth up (3.0 mi) North Fork Redwood Cr. mouth up (1.0 mi) LOCATION
Gulch 11 mouth up (0.8 mi) Middle Fork Ten Mile R. mouth to Bear Haven (3.1) MF Ten Mile R. Bear Haven to bridge (1.5 mi) Bear Haven mouth up (2.5 mi) Little Bear Haven mouth up (0.3 mi) North Fork Ten Mile R. mouth to L.N. Fork (1.9 mi) NF Ten Mile R.-LN Fork to Bald Hills Cr. (5.2 mi) NF Ten Mile R.-Bald Hills to Patsy Cr. (5.1 mi) NF Ten Xile R.-Patsy Cr. to headwaters (3 .3 mi) Little North Fork Ten Mile R. mouth to dam (3.5) Buckhorn Creek mouth up (1.0 mi) Cavanaugh Creek mouth up (0.5 mi) Bald Hills Creek mouth up both forks (2.3 mi) Unnamed NF Ten Mile tributary mouth up (1.0 mi) Patsy Creek mouth up (2.5 mi) Stanley Creek mcuth up (1.0 mi) Pudding Creek tunnel to Little Valley Cr. (4.2 mi) Pudding Creek L. Valley to Ramsey Ridge Rd. (4.1) Little Valley Creek mouth up (0.8 mi) Unnamed Pudding Cr. tributary mouth up (0.3 mi) South Fork Noyo River mouth to Kass Cr. (0.6 mi) SF Noyo R.-Kass Cr. to egg-taking station (3.3 mi) SF Noyo R.-Egg station to Parlin Cr. (2.4 mi) SF Noyo R.-Parlin Cr. to pond spillway (2.4 mi) Kass Creek mouth to fork (2.7 mi) North Fork SF Noyo R. mouth up (4.9 mi) Parlin Creek mouth up (3.5 mi) South Fork Big R. mouth to Hell's Gate (10.1 mi) SF Big River-Hell's Gate to Orr springs (8.2 mi) Ramon Creek mouth up (2.5 mi) Mettick Creek mouth up (1.0 mi) Anderson Gulch mouth up (0.5 mi) Daugherty Creek mouth up (4.7 mi) Soda Creek mouth up (0.3 mi) Gates Creek mouth up (2.5 mi) Unnamed Daugherty Cr. tributary mouth up (1.0 mi) Little River mouth up (4.0 mi) Caspar Creek mouth to forks (3.0 mi) South Fork Caspar Creek mouth up (2.0 mi) North Fork Caspar Creek mouth up (2.5 mi) 1 LOWER N SOUTH FORK EEL RIVER 1 BASIN
Richardson Grove
/ Andersonia / I