Draft June 2005
Evaluate Factors Influencing Bias and Precision of Chinook Salmon Redd Counts
Statement of Work and Budget FY2005

BPA Project Number:  2002-049-00
BPA Project Title: Evaluate Factors Influencing Bias and Precision of Chinook Salmon Redd Counts      
Contract Number:  00010762
Contract Title: Evaluate Factors Influencing Bias and Precision of Chinook Salmon Redd Counts
Performance/Budget Period: July 2005-June 2006

Research Project Description
The ability to detect trends in fish populations depends upon obtaining reliable estimates of abundance in an efficient manner. Simply relying on a relative count that has not been adjusted for undetected individuals, may lead to misleading conclusions about population trends, spatial distribution, and habitat associations because of the unknown magnitude of the sampling bias. Redd counts in index areas are commonly used to monitor annual trends in Chinook salmon (Oncorhynchus tshawytscha) populations where total adult escapements are unknown. The assumption is that uncorrected redd counts represent a constant proportion of true numbers of redds across time, which is unlikely given the myriad of environmental and other factors affecting redd sightability or redd distribution. Further, an index count provides a single number with no measure of precision, i.e., it does not include sampling variation. Despite the widespread use of redd counts to calculate measures of population performance, little is known regarding the accuracy of Chinook salmon redd counts or the factors that decrease precision and introduce bias. Therefore, we will more rigorously evaluating factors influencing bias and precision of Chinook salmon redd counts. We will determine the true number of redds within a series of study reaches; apply the true counts to determine the accuracy of aerial and ground-based redd counts; measure environmental and habitat factors and model which variables most influence redd sightability; assess inter- and intra- year sources of variation in redd counts; quantify inter- observer variation in ground-based surveys; compare accuracy of single versus multiple pass counts; evaluate and compare the effectiveness of a modified two-sample, Lincoln-Petersen mark-resight estimator for obtaining unbiased and precise abundance estimates of redds, and evaluate the potential for crew training to reduce the bias in redd counts. Results from this research will have important implications for improving Chinook salmon redd surveys conducted across the Snake River basin.

This research addresses at least three critical needs identified in Regional Program documents. 1) the need for long-term information to assess trends in wild Chinook salmon populations; 2) the need for evaluation of broadscale population sampling and inventory methods; and 3) the need for analysis of the spatial structure of wild Chinook salmon populations.

Project History and Accomplishments
In FY 2001 we submitted and were granted a request for a within year increase in funds as part of BPA Project #1999-020-00 to conduct pilot research to evaluate the bias and precision of redd counts, which is our primary method for monitoring Chinook salmon populations. In 2002, we submitted a full proposal (#28001) to expand the redd count analysis research. In FY 2002, BPA Project # 2002-049-00 was funded and we expanded the research we had begun in 2001. From 2002 through 2004 we have annually collected field data that will enable us to rigorously evaluate the bias and precision of aerial and ground-based redd counts. In 2002, observers completed consecutive redd counts in 20 reaches of five major tributaries. Crews recorded the locations of all redds with GPS units. A total of 154 independent redd counts were completed in 17 reaches and IDFG biologists collaborated with us to complete 9 counts in 3 reaches. In 2003, observers completed consecutive redd counts in 15 reaches of six major tributaries. A total of 110 independent redd counts were completed in the 12 reaches and IDFG biologists collaborated with us to complete 9 counts in 3 additional reaches. In 2004, observers completed consecutive redd counts in 19 reaches of six major tributaries. A total of 131 independent redd counts were completed in 16 reaches and IDFG biologists collaborated with us to complete 9 counts in 3 additional reaches. From 2002-2004, variation in redd counts as a result of interobserver variability was examined by having multiple independent observers conduct single pass ground counts in several study reaches after the completion of Chinook salmon spawning. Reach- and redd-scale characteristics affecting sightability were measured by field crews at more than 800 redds from 2002-2004. These data have been assembled into a database in preparation for detailed analysis and modeling of factors affecting sightability.    

Results to date have been presented at Idaho Chapter AFS meetings (Thurow and Isaak 2003), AFS National meetings (Thurow and Isaak 2004), and Western Division AFS meetings (Thurow and Isaak 2004). Several manuscripts destined for peer-reviewed journals are in preparation and Quarterly and Annual reports have been submitted to BPA.

Location of Project:
The Middle Fork Salmon River (MFSR) drainage has been selected as the study area; see Thurow 2000 and Servheen et al. 2001 for a detailed study area description. This area was selected based on several considerations. First, opportunities exist for extensive collaboration with other agencies and tribes. Second, as part of BPA project #199902000 RMRS biologists plan to conduct annual aerial redd counts in the entire MFSR through 2004. The aerial surveys funded in the existing research can be applied at no extra cost to facilitate the mark-resight approach described in this new proposal. Third, one of the principal investigators has more than 20 years of experience working in this drainage and an intimate knowledge of the MFSR and the spawning ecology of its Chinook salmon.