INTRODUCTION

A coordinated approach to the monitoring and evaluation of status and trends in anadromous and resident salmonid populations and their habitats is needed to support restoration efforts in the Columbia Plateau.  This project will focus on the Steelhead component of this need.  Currently, independent research projects and some monitoring activities are conducted by various state and federal agencies, tribes, and to some extent by watershed councils or landowners, but there is no overall framework for coordination of efforts or for interpretation and synthesis of results.  This SOW extends the structure and methods employed by the Oregon Plan for Salmon and Watersheds Monitoring Program (Nicholas, 1997a; 1997b; 1999) to the John Day subbasin of the Columbia Plateau. This approach, successfully implemented in Oregon's coastal watersheds, applies a rigorous, Tier-2 sampling design to answer key monitoring questions, provides integration of sampling efforts, and has greatly improved coordination among state, federal, and tribal governments, along with local watershed groups. Because the John Day subbasin populations of steelhead are part of the federally listed Mid-Columbia steelhead ESU status and trend monitoring needs to be integrated with ongoing projects in this subbasin. This project is high priority based on the high level of emphasis the NWPPC Fish and Wildlife Program, Subbasin Summaries, NMFS, and the Oregon Plan for Salmon and Watersheds have placed on monitoring and evaluation to provide the real-time data to guide restoration and adaptive management in the region.

The sampling program described in this document fulfills critical monitoring needs for Council's Fish and Wildlife Program and help fulfill population and environmental monitoring requirements under the NMFS 2000 FCRPS Biological Opinion (Action 180).  The work under this statement of work will meet most of the BiOp requirements for Tier 1 and 2 monitoring throughout the John Day River subbasin.  Current population and environmental monitoring in the Province is based on a highly variable application of a combination of index surveys and periodic monitoring of some status and trend indicators.  For example, most adult steelhead monitoring is based on a small number of index sites relative to the number of streams steelhead use for spawning.  The index approach only allows us to draw inference about trends in adult abundance for the surveyed streams and provides little information on abundance (status) or distribution at the subbasin or plateau spatial scales.  This is due to the fact that index reaches are not randomly selected and represent an unknown proportion of the total population.  In addition, there are no systematic programs in place to collect information on the status, trends, and distribution of fish habitat/riparian conditions or juvenile salmonids.  

The ISRP, in their guidance on monitoring, strongly recommended that the region move away from index surveys and embrace probabilistic sampling for most population and habitat monitoring.  The ISRP stated "the Council's Fish and Wildlife Program calls for monitoring and evaluation of biological and environmental conditions at the scale of provinces and subbasins.  Tier 2-level monitoring is required to provide inductive inferences to entire provinces, subbasins, and many watersheds, because it is impossible to survey every square meter of every stream bottom, riparian zone, and uplands area in these large regions every month of every year for decades.  Many of the Columbia Basins' projects for "monitoring" fish and wildlife species (redds, spawners, juveniles, etc.) currently limit surveys to "index sites" selected by professional judgment in past years.  The objectives of these projects can only be met with Tier 2-level monitoring using probabilistic selection of survey sites with limited replication".  The sampling approach outlined in this study fulfills these requirements.

By implementing the program we will address many of the goals for Tier 1 monitoring, such as defining areas currently used by adult steelhead and spring chinook for holding and spawning habitats and summer rearing habitats for juvenile O. mykiss and spring chinook (adult/juvenile salmonid monitoring), determining range expansion or contraction of summer rearing and spawning populations of O. mykiss and spring chinook (adult/juvenile salmonid monitoring), and identifying associations between salmonid presence (adult/juvenile monitoring) and habitat attributes (habitat monitoring).  The BiOp describes Tier 2 goals as defining population growth rates (adult monitoring), detecting changes in those growth rates or relative abundance in a reasonable time (adult/juvenile monitoring), estimating juvenile abundance and survival rates (juvenile/smolt monitoring), and identifying stage-specific survival (adult-to-smolt, smolt-to-adult) and environmental attributes (habitat monitoring). This project provides much needed Tier-2 monitoring for the two anadromous focal species in the John Day basin.

Integration with on-going monitoring is accomplished in the following manner.  In annual reporting, we use data from on-going projects to develop a more complete picture of status and trends in resources and life stage-specific survival.  This project will be the vehicle to pull all related fish population and habitat monitoring data together into a synthetic analysis of resources at the provincial and subbasin scales.  For example, we use data from ongoing smolt and adult monitoring to calibrate surveys and to track trends in survival and productivity at life-stages not targeted under the EMAP program.  More detailed studies at finer scales will inform the EMAP program and vice-versa.  To accomplish these tasks, we work with co-managers and other interested publics to establish a monitoring oversight committee for the region that is tasked with coordinating and integrating on-going efforts into a comprehensive reporting system of regional resources.

This project provides information as directed under two measures of the Columbia Basin Fish and Wildlife Program.  Measure 4.3C specifies that key indicator naturally spawning populations should be monitored to provide detailed stock status information.  In addition, measure 7.1C identifies the need for collection of population status, life history, and other data on wild and naturally spawning populations.  This project was developed in direct response to the recommendations and needs of regional modeling efforts, the Independent Scientific Review Panel (ISRP), the Fish and Wildlife Program, and the Columbia Basin Fish and Wildlife Authority Multi-Year Implementation Plan.


APPROACH 1:  Implement the EMAP sampling framework, a statistically based and spatially explicit sampling design, to quantify the status and trends in the abundance of steelhead redds.  Based on the strong relationship between cumulative redd counts and adult steelhead abundance, cumulative redd counts is used to index the abundance and distribution of adult steelhead at the provincial and subbasin scales (Susac and Jacobs, 1999; Jacobs et al., 2000; Jacobs et al., 2001).  Fifty spatially balanced, randomly selected reaches are sampled and steelhead redds are quantified in the John Day subbasin from about March 1 through June 15 annually. Dam counts and index surveys tell only part of the story and include unknown biases.  Adding a statistically-based sample program gives unbiased estimates of abundance in addition to data on distribution and habitat use and life history patterns (timing of spawning, spatial distributions).  This information cannot be derived from dam counts or index surveys.

Sampling domains and site selection:  ODFW in cooperation with co-managers and other interested parties refine the sampling universe for steelhead redd surveys based on current ODFW distribution maps. The sampling domain is defined for the upper and lower ends of distributions based on available data and best professional judgment on the potential distribution of spawners.  The delineation of the sampling domain is currently liberal in its' extent at the outset to encompass all potential habitat.  To balance the needs of status (more random sites) and trend (more repeat sites), we implement a rotating panel design based on recommendations from the EPA EMAP Design Group.  The 50 sites drawn on an annual basis are assigned to the rotating panel design as follows:

- 3 panels with different repeat intervals
- 17 of the sites are sampled every year
- 16 sites are allocated to a 4 year rotating panel (sites visited once every 4 years on a staggered basis)
- 17 sites are new sites each year

With this sampling strategy, 50 sites were drawn the first year and 33 new sites are drawn in subsequent years because 17 of the originally drawn sites are repeated each year.  Once annual sample sites are drawn, the site is assigned to the river reach file based on site coordinates.  From these point coverage's, ODFW develops landowner contacts based on county plat maps.  Based on ownership maps, project personnel work with ODFW District Biologists and Co-Managers to obtain permission from landowners and set up sites.  A Geographic Information System (GIS) incorporating a 1:100,000 digital stream network is used to insure an unbiased and spatially balanced selection of sample sites across each subbasin.  The GIS site selection process provides the geographic coordinates (i.e. latitude and longitude) of each of the candidate sites.  We then produce topographic maps showing the location of each sample point.  Field crews use a handheld Geographic Positioning System to find the approximate location of the EMAP selected sample point, and then establish 2 km long survey reaches that encompass the sample point.  Site reconnaissance is conducted in the fall in preparation for spawning surveys the following spring.  Site reconnaissance involves obtaining landowner permission, verifying the presence of suitable habitat (e.g., presence of spawning gravel, barriers to upstream migration, gradient, etc.), marking the upper and lower boundaries of the survey with spawner survey signs, take Universal Transverse Mercator (UTM) coordinates of the upper and lower boundaries, and attempting to define upper and lower boundaries by distinctive landmarks.

Adult Steelhead Redd Surveys:  Adult steelhead redd surveys are conducted from March 1 - June 15 annually based on standard ODFW methods for conducting steelhead redd surveys (Susac and Jacobs, 1999; Jacobs et al., 2000; Jacobs et al., 2001).  Fifty sites arebe selected and visited every 2-3 weeks throughout the season to quantify the cumulative redd count at each site. Surveyors sample upstream from the downstream end of each survey reach.  Each surveyor counts live fish and determines the fin-mark status of all live fish through observations.  All redds are counted and flagged.  Data are recorded on the spawning survey form, redd longevity form, and spawning location description form.  Survey crews review survey forms daily and deliver hard copies bi-weekly to the crew chief.  Crew chiefs conduct weekly site visits with each crew.  Data entry is conducted as time allows throughout the survey season and is completed within one month of the end of fieldwork.  The population status is indexed through cumulative redd counts.  Expected precision is ±40% at the subbasin scale.  Hatchery: wild ratios are estimated by observing the occurrence of adipose fin-clipped and unmarked live fish on spawning grounds.

To quantify observer error we implement the following procedures.  Each site is visited every two weeks with the surveyors swapping sample reaches every survey.  The surveyor records the number of flagged redds, new redds, and redds missed during the previous survey.  Missed redds are distinguished from new redds by the amount of periphytic growth in the redd pocket.  New redds are devoid of periphyton whereas older redds become obscured by periphytic growth.  The independent estimate of marked versus unmarked redds from survey to survey provides an estimate of the error associated with identifying steelhead redds.  To validate whether cumulative redd counts are a reliable indicator of populations status, we will begin exploring where we can develop the data to allow the conversion of redd counts to population estimates.  The necessary data would include the sex ratio of returning adults and redd:female ratios. This validation effort will need to be conducted outside of the John Day subbasin so it can be calibrated with permanent weir counts.

Where we have on-going index surveys, these surveys will continue through a transition period from index surveys to probabilistic sampling.  We are developing a dataset that covers the range of abundance seen under the historic index surveys to examine the relationship between the two.  From this analysis we should be able to develop a strong relationship that will allow us to index the historic surveys to the probabilistic surveys.  This will take an unknown length of time but will probably be on the order of 5-10 years.


APPROACH 2:   Implement the EMAP sampling framework, a statistically based and spatially explicit sampling design, to quantify trends in juvenile trout and spring chinook and status and trends in stream and riparian habitats.  Juvenile and habitat data collected in coastal watersheds and the Great Basin were critical to NOAA Fisheries and US Fish and Wildlife Services decisions to not list Klamath Mountain Province steelhead and Great Basin Redband Trout.  In addition, NOAA Fisheries Technical Review Teams and Oregon's Salmon Recovery Task Force have used these data sets extensively in their status reviews and recovery planning efforts. Fifty spatially balanced, randomly selected reaches will be sampled for juvenile salmonids and stream and riparian condition in the John Day subbasin from late June through September annually.  The sample selection process is described above. There will be near complete site overlap with steelhead spawner surveys except that juvenile/habitat survey sections will be 0.5-1 km in length instead of 2 km.

Juvenile Salmonid Survey Methodology:  Snorkel surveys involve a single upstream pass through each pool during daylight along a 1-km survey reach.  The number of snorkelers employed will be based on what is needed to effectively cover the pool being snorkeled on a single upstream pass.  To reduce problems associated with snorkeling in shallow or fast water habitat, only pools > 6 m2 in surface area and > 40 cm deep are snorkeled.  Counts of the number of juvenile and adult trout (O. mykiss and O. clarki) and salmon (O. tshawytscha) are recorded for each pool.  Trout and salmon will be categorized as juvenile (1+ years or greater), or adult based on size classes developed from local data and/or standards used by ODFW and co-managers.  Other species will be noted as present and recorded.  Crewmembers either alternate the pools that they snorkel or one crewmember snorkels the entire reach.  After snorkeling, the underwater visibility of each pool during the snorkel count is ranked on a scale of 0 to 3 where: 0 = not snorkelable due to an extreme amount of hiding cover or zero water visibility; 1 = high amount of hiding cover or poor water clarity; 2 = moderate amount of hiding cover or moderate water clarity neither of which were thought to impede accurate fish counts; and 3 = little hiding cover and good water clarity.  Only pools with a visibility rank of two or three are used in data analysis.  If all pools in a reach have visibilities < 2, then as many pools in the reach as possible will be electrofished using Smith-Root model 12-B backpack electrofishers following NMFS electrofishing guidelines for juvenile salmonid presence/absence.  Electrofishing will be conducted by making a single pass upstream in each pool that meets the size and depth criteria for conducting snorkel surveys.  No block nets will be used for this sampling.  Electrofishing data will be combined with snorkeling data to determine the presence/absence of juvenile O. mykiss and spring chinook.  The presence/absence data will be analyzed to quantify the percent of sites where juvenile O. mykiss and spring chinook are present as an estimate of juvenile distribution in the sample frame annually (e.g., 40% site occupancy).

To quantify the measurement error in the snorkel data, and to provide information on temporal changes in abundance during the course of the sampling season, supervisory staff will resurvey a random sample of 10 to 20 percent of the sites surveyed in each subbasin.  Our goal is to limit between diver error to ± 20% or less with intensive presurvey training of field crews and regular random resurveys.  Our approach in coastal watersheds has been to check crews early and often to ensure that the surveys are meeting the target precision levels.  Once this is done, we have found no need to adjust the data.  Since the crews know that any site may be re-surveyed at any time the focus on quality data has remained high.  Five years of data and over 1000 sites surveyed have required no post-survey adjustment of the data.  Re-surveyed sites that do not meet our precision goals are evaluated with the crew and re-done to meet the QC criteria.

Data analysis involves calculating the percentage of survey sites that contain at least one juvenile fish for O. mykiss and spring Chinook and the percentage of pools per site that contain juvenile O. mykiss and spring Chinook to quantify changes in the relative distribution interannually.  Analysis from coastal watersheds indicate that snorkeling data from pools has the strongest explanatory power regarding the overall trend is juvenile steelhead and coho populations (Pers. Comm, Jeff Rodgers, ODFW Research Lab, Corvallis).  We will quantify the number of juvenile O. mykiss and spring chinook observed per square meter for use in population trend analysis within and among individual subbasins.  Confidence limits for summary estimates will be developed based on quantifying the measurement error in the snorkel data (see paragraph above) and site-to-site variability based on a variance estimator developed by the EPA EMAP Program for this application.  Because juvenile salmonids have more diverse habitat requirements (rearing habitats are often different and dispersed relative to spawning habitat), evaluating their trends through time are necessary as an independent indicator of salmonids status.  ODFW will use the data developed in this project to evaluate the types of questions put forth in BPA's 9/27/2003 memo.  These are important questions to answer in the evaluation but should not be conditions for implementation.  Plausible outcomes and interpretations should not be required a priori.

Habitat and Riparian Survey Methodology:  Channel habitat and riparian surveys are conducted as described by Moore et al. (1997) with some modifications.  Modifications include: survey lengths of 500-1000 m and measurement of all habitat unit lengths and widths (as opposed to estimation).  Survey teams collect field data based on stream, reach, and channel unit characteristics.  Each field crew is comprised of two people with each member responsible for specific tasks.  The "Estimator" focuses on the identification of channel unit characteristics.  The "Numerator" focuses on the counts and relative distribution of several unit attributes and will verify the length and width estimates for a subset of units.  The "Estimator" and "Numerator" share the responsibility for describing reach characteristics, riparian conditions, identifying habitat unit types, and for quantifying the amount of large woody debris.  Crewmembers may switch responsibility for estimator or numerator when they start a new stream.  They do not, however, switch estimator and numerator jobs on the same stream.  The methods and indicator variables collected with this protocol are far too detailed to include in a work statement, but can be viewed at: http://osu.orst.edu/Dept/ODFW/freshwater/inventory/pdffiles/habmethod.pdf.  
These variables are consistent with the core indicators for US Forest Service and BLM surveys in the region.  In fact, the BLM contracts with ODFW to conduct BLM habitat surveys using the ODFW protocol in Oregon.  The core variables are consistently used and accepted throughout the Pacific Northwest, not just in coastal watersheds.  ODFW's program has been implemented and refined for the past 12 years and has formed the basis for several recent EDT analyses in Oregon.  The most recent example being the work of Chip McConnaha in Johnson Creek where he noted that our habitat data was particularly useful for EDT modeling.  ODFW will work with BPA and other partners to refine the list of habitat indicators as needed.

To quantify within-season habitat variation and differences in estimates between survey crews, ten percent of the sites will be resampled with a separate two-person crew.  Repeat surveys will be a randomly selected sub-sample from each subbasin and each survey crew.  Variation in survey location was assumed minimal because survey starting and ending points were marked in the field.  The precision of individual metrics will be calculated using the mean variance of the resurveyed streams "Noise" and the overall variance encountered in the habitat surveys "Signal". Three measures of precision are calculated, the standard deviation of the repeat surveys SDrep, the coefficient of variation of the repeat surveys (CVrep), and the signal to noise ratio (S:N).  S:N ratios of < 2 can lead to distorted estimates of distributions and limit regression and correlation analysis. S:N ratios > 10 have insignificant error caused by field measurements and short term habitat fluctuations (Kauffman et al. 1999).

Habitat conditions in each subbasin will be described using a series of cumulative distributions of frequency (CDF). The variables described are indicators of habitat structure, sediment supply and quality, riparian forest connectivity and health, and in-stream habitat complexity.  The specific attributes include but are not constrained to:

Density of woody debris pieces (> 3 m length, >0.15 m diameter)
Density of woody debris volume (> 3 m length, >0.15 m diameter)
Density of key woody debris pieces (>10 m length, >0.6 m diameter)
Density of wood jams (groupings of more than 4 wood pieces)
Density of deep pools (pools >1 m in depth)
Percent pool area
Density of riparian conifers (>0.5 m DBH) within 30 m of the stream channel
Percent of channel shading (percent of 180 degrees)
Percent of substrate area with fine sediments (<2 mm) in riffle units
Percent of substrate area with gravel (2-64 mm) in riffle units

While these attributes do not describe all of the conditions necessary for high quality
salmonid habitat, they do describe important attributes of habitat structure within and adjacent to the stream channel. The attributes are also indicative of streamside and upland processes.  Water quality and quantity, as well as food production, are not addressed in the discussion of physical habitat, but are critical elements for the Oregon Department of Environmental Qualities EMAP program.  The median and first and third quartiles will be used to describe the range and central tendencies of the frequency distributions of the key habitat attributes used in the analysis of current habitat conditions (Zar 1984).  Frequency distributions will be tested to determine if significant differences (p<0.05) exist between subbasins for each habitat attribute (Thom et al. 2000).


LITERATURE CITED
Bisson. P.A., J.A. Nielsen, R.A. Palmason, and E.L. Grove.  1982.  A system of naming
habitat types in small streams, with examples of habitat utilization by salmonids during low stream flow.  Pages 62-73 in: N.B. Armantrout, ed. Acquisition and utilization of Aquatic Habitat Inventory Information.  Western Division, American Fisheries Society, Portland OR.

Chilcote, M.W. 2001. Conservation assessment of steelhead populations in Oregon. Oregon Department of Fish and Wildlife, Portland, OR.

Cupp, C.E.  1989.  Stream corridor classification for forested lands of Washington.  Hosey and Assoc.  Bellevue, WA  46 p.

Dambacher, J.M. and K.K. Jones.  1997.  Stream habitat of juvenile bull trout populations in Oregon and benchmarks for habitat quality.  Pages 353-360 in Mackay, W. C., Brewin, M. K., and M. Monita, editors.  Friends of the bull trout conference proceedings.  Bull Trout Task Force (Alberta), c/o Trout Unlimited Canada, Calgary.

Dempson, J.B., and D.E. Stansbury. 1991. Using partial counting fences and a two-sample stratified design for mark-recapture estimation of an Atlantic salmon population. North American Journal of Fisheries Management. 11:27-37.

Everest, F.H., R.L. Beschta, J.C. Scrivener, K.V. Koski, J.R. Sedell, and C.J. Cederholm. 1987. Fine sediment and salmonid production: A paradox. Pages 98-142 In: E.O. Salo and T.E. Cundy eds., Streamside Management: Forestry and Fishery Interactions. Contribution No. 57. Institute of Forest Resources, University of Washington, Seattle, Washington.

Firman, J.C., and S.E. Jacobs.  2001.  A survey design for integrated monitoring of salmonids.  First Int. Symp. On GIS in Fishery Science.

Frissell, C.A., W.J. Liss, C.E. Warren, and M.D. Hurley.  1986.  A hierarchical framework for stream habitat classification: viewing streams in a watershed context.  Environ. Manage. 10: 199-214.

Grant, G.E. 1988. Morphology of high gradient streams at different spatial scales, Western Cascades, Oregon. Pages 1-12 in: Shizouka Symposium on Geomorphic Change and the Control of Sedimentary Load in Devastated Streams, Oct. 13-14, 1988. Shizouka University, Shizouka, Japan.