Premium
Retrospective Sampling
Author(s) -
Irvine James R.,
Chen DinDing Geng,
Schnute Jon T.
Publication year - 2003
Publication title -
fisheries
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.725
H-Index - 79
eISSN - 1548-8446
pISSN - 0363-2415
DOI - 10.1577/1548-8446(2003)28[25:rs]2.0.co;2
Subject(s) - sampling (signal processing) , streams , statistics , abundance (ecology) , sample (material) , systematic sampling , sampling design , computer science , fishery , mathematics , biology , demography , population , computer network , chemistry , filter (signal processing) , chromatography , sociology , computer vision
Retrospective sampling can help fishery scientists improve the cost‐effectiveness of field sampling programs. This technique uses historical data gathered intensively from a few sites to set realistic future sampling goals. We illustrate the method with 10 years of salmon smolt emigration data from three streams in British Columbia, Canada. When we used data from one‐third to one‐half the days actually sampled during each run, the error associated with individual annual estimates rarely exceeded 30%. Our mean 10‐year abundance estimates were never significantly different from the true numbers, regardless of sampling effort, indicating that estimates were unbiased. Although errors were highest with infrequent sampling, the incremental gains achieved by sampling more than half of each run were small. In retrospect, a reduced sampling effort in each stream would have created an opportunity to examine more than three streams with the same sampling effort, and may have achieved a better balance between sample size and habitat coverage.