Accurate catch estimates are integral to the assessment of fish stocks and the development and application of harvest rules for sustainable fisheries management. As with spawning escapement estimates, accurate and timely catch information is the currency upon which all higher level management functions are based.  The need to routinely produce precision estimates (confidence limits and standard errors) in addition to point estimates of catch is an expected standard practice. In recent times, salmon managers often find themselves sequestered in post-season review exercises to account for “fish have gone missing”. These assessments are often based on comparisons of point estimates of (a) fish estimated to have escaped past Mission; (b) catch estimates in the gauntlet of fisheries (ocean commercial, commercial in-river, First Nation in-river, ocean and in-river sport); (c) estimates of pre-spawn / en-route migration mortality; and (d) estimates of final spawning escapement. Error estimates in all these variables must be understood to set a proper context for what proportion of any “missing fish” can simply be explained by sample variability and measurement error (Figure 1).

Figure 1: Significance of precision estimates in post-season accounting

Generating standard errors and confidence limits for multi-stage ratio estimators is a relatively advanced statistical problem. MERCI employs a non-parametric, bias-corrected bootstrap approach (see further details and examples in Alexander 2002a/b). The bootstrap is a simulation‑based method that can answer many statistical questions without the need to derive complex formulas. It avoids the restrictive and often-dangerous parametric assumptions about the distribution of the underlying population that are made when using classical (usually normal‑theory based) statistical estimators (Efron and Tibshirani 1993). 

MERCI Precision Estimation Tool (PET) is a stand-alone software package designed to estimate precision of bootstrap simulation data generated by the MERCI software.  The outputs generated by MERCI PET include estimates for catch, CPUE, and effort with the corresponding estimates of precision: standard error and (bias-corrected) confidence intervals.  The program uses a combination of a Visual Basic .NET User Interface coupled with the statistical computing power of R: A Language and Environment for Statistical Computing.  

This online help site provides a walkthrough of 
  • where to get MERCI PET, 
  • How to install MERCI PET, 
  • How to use MERCI PET, and 
  • The output generated by MERCI PET.  
Subpages (2): Download Install