PowerPoint Presentation

Estimating Steelhead Redd Abundance and Variance in the Wenatchee River BasinAndrew Murdoch WDFWChad Herring WDFWKevin See QCIChris Jordan NOAAAgendaBackgroundObjectivesMethodsPreliminary resultsPlans for 2012

BackgroundUCR Steelhead listed in 1999Need estimate of spawner abundance to estimate productivityUCR steelhead, like most others, have a huge variation in migration patternsOvershooting natal streamsHatchery straysDifferential overwintering mortality Tributary versus Columbia

Goals and ObjectivesGoalEstimate the total number of redds with an estimate of variance Objective Develop models to estimate redd observer efficiency based on environmental, biological, or habitat variablesEstimate number of redds using AUC and redd lifeEstimate variance based on total uncertainty in the models

Survey MethodologyConduct weekly surveys of index areasGeo-reference all reddsCollect environmental data each surveyConduct habitat survey of each index area (one time)Conduct single survey of non-index areas at or near peak spawning of associated index areaCollect environmental and habitat data

Index and Non-index reachesIndex reaches are not randomIncludes all major spawning areasNumber of index reaches depends on spawn timing and redd lifeNon-index reachesEverywhere elseSame spawn timing and redd life as index reachRedd Life SummarySteelhead spawning timing progresses from low to upper elevations, redd life is positively correlated with elevation.Redd life is only as good as survey frequencyMay require some intensive surveys in the first few years Some surveys may need to be performed after spawning is complete

2010 Steelhead Redd Life ReachMeanNSDCVRange% of all reddsMinMaxP116.2448.753.534170%I119.5476.734.483242%W918.238949.534536%W1018.0509.150.634536%N115.1108.153.652836%N320.413524.5143315%Observer Efficiency Results by ReachCensus reachNRedd statisticsReddsOmissionFalse IDMeanCVRange%SD%SD%SD P1 916383 - 2448.818.351.218.317.217.6 N310344517 - 5461.216.238.816.214.011.3 I110242912 - 3649.814.350.214.322.5 9.0 W9 821779 - 5230.922.369.122.340.611.7W1011344018 - 6141.213.158.813.120.713.2

This shows the relative influence of each possible covariate in the suite of possible models.Top 7 are clearly important: reach experience, Mean Max Distance, density, channel complexity index, mean discharge, depth, widthChannel Complex Index is the most informative measure of channel complexity10Overall Best ModelReach ExperienceVisibility Redd densityChannel Complexity Index

11Statistical Analysis Redd Abundance Estimate of redds by reach for both index and non-index areasAUCUse normal approximation of redd counts through timeApply some of the same parameters to non-index reachesGet estimate of total number of redd-daysDivide by redd lifeDivide by observer efficiencyIndex Reach (Icicle River)

Raw dataEventually these will be adjusted for observer efficiency using model predicted estimates of observer efficiency13Index Reach (Icicle River)

The area under the curve is the total number of redd-days.For index reaches, where counts are only of new redds, we divide redd-days by 7 to get the total number of redds14Index Reach (Icicle River)

The shaded area is the same as the area under the normal curve.Curve gives estimate of peak spawning, spread around that timing, and a measure of the magnitude of the run for that reach.15Non-Index Reach (W3)

Here are the raw counts and the fitted model to the index reach, W2.16Non-Index Reach (W3)

Here are the raw counts and the fitted model to the index reach, W2.And here is the one count from the non-index reach, W3Assume the same peak spawning time, and the same spread of that timing; the only difference is the magnitude.Fit a whole new curve for the non-index reach, and then divide the area under the red curve by observer efficiency and operational redd-life to get an estimate of the total number of redds in the non-index reach.17Statistical Analysis Variance Sources of varianceRedd lifeObserver efficiency Parameter uncertainty in estimating the normal curveAll sources can be included using the delta methodRedd life is most important in the non-index reachesObserver efficiency may be the biggest source of uncertainty, but were working to make it more precise18ConclusionsObserver efficiency can be estimated reasonably wellMethodology can be used with a reasonable amount of effortRedd life is important, but somewhat subjectiveTraining, training, trainingNext StepsCollect more observer efficiency dataMaximize contrast to the greatest extent possibleEvaluate importance of survey effort (hours) into modelEvaluate alternate metrics for channel complexity (thalweg variation)Model validationConduct weekly surveys on all census reachesEstimate redd abundance and variance for the Wenatchee Basin in 2012 Conduct same study in the Methow using two surveyors instead of one (2012 -2014)

AcknowledgementsFunding provided by BPA and Chelan County PUDAll the WDFW spawning ground surveyorsChris Jordan conceived the projectJody White and Chris Beasley at QCI for valuable insight into the projectKevin See for graciously accepting the project with no warning