Premium
Robust Harvest Estimates from On‐Site Roving–Access Creel Surveys
Author(s) -
Bernard David R.,
Bingham Allen E.,
Alexandersdottir Marianna
Publication year - 1998
Publication title -
transactions of the american fisheries society
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.696
H-Index - 86
eISSN - 1548-8659
pISSN - 0002-8487
DOI - 10.1577/1548-8659(1998)127<0481:rhefos>2.0.co;2
Subject(s) - statistics , sampling (signal processing) , variance (accounting) , sample (material) , environmental science , fishing , recreation , econometrics , mathematics , fishery , computer science , economics , ecology , biology , telecommunications , chemistry , accounting , chromatography , detector
Roving–access creel surveys to estimate harvest of migratory or nonmigratory fish in recreational fisheries are described as two‐stage statistical designs based on sampling with equal probability without replacement. Harvest is estimated for each first‐stage unit (a sampling period) with information from roving counts of anglers and completed‐trip interviews at access locations. Approximations are provided for sample variances when sampling periods or counts are systematically scheduled. Stratification and poststratification are discussed, and covariance equations are developed for summing poststratified estimates to avoid bias. We develop methods to correct failures to sample scheduled periods or failures to interview anglers during a period. Methods are demonstrated with examples drawn from past on‐site creel surveys in Alaska and with simulations. Covariances from summing poststratified estimates proved negligible relative to overall variance, whereas corrections for missing data represented a potentially significant portion of variance. Estimated harvests are still subject to significant length‐of‐stay bias when the fishing day is subsampled, particularly when harvest is restricted by a daily bag limit. Length‐of‐stay bias can be avoided by setting the sampling period equal to the fishing day.