Influences of Experts' Personal Experiences in Fuzzy Logic Modeling of Atlantic Salmon Habitat

• Published in: North American Journal of Fisheries Management
• Main Authors: Mocq, J., St‐Hilaire, A., Cunjak, R. A.
• Other Authors: Natural Sciences and Engineering Research Council of Canada
• Format: Article in Journal/Newspaper
• Language: English
• Published: Wiley 2015
• Subjects: Canada; Atlantic salmon; Salmo salar
• Online Access: http://dx.doi.org/10.1080/02755947.2014.996684; https://afspubs.onlinelibrary.wiley.com/doi/pdf/10.1080/02755947.2014.996684

Abstract The knowledge of scientific experts, which is regularly used in modeling, is acquired by training, education, and practical experiences that modify the experts' perceptions. Using a case study dealing with fish habitat modeling, we investigated the possible influences and potential biases imparted by some of these personal experiences. Thirty salmon experts with different backgrounds and nationalities defined fuzzy sets and fuzzy rules in a fuzzy habitat model of three Atlantic Salmon Salmo salar life stages. Weighted usable area (WUA) curves were calculated for each expert by coupling the fuzzy model with a hydraulic model applied to the Romaine River (Quebec, Canada). Experts were then split into subgroups, and three possible experiential biases were tested: the experts' main geographic region of expertise (Europe versus North America), their primary source of knowledge (fieldwork, scientific literature, or both), and their employment sector (public or private). A confidence interval was calculated around the median WUA curve for each subgroup by bootstrap resampling. A divergence in the confidence intervals (i.e., no overlap) indicated a significant influence of the tested experience. For all three considered life stages, we observed no significant impact of employment sector or knowledge source on modeled WUA. However, the experts' geographic region of expertise had a significant influence on the output of the spawning adult habitat model. Consequently, the use of local expert knowledge in modeling is recommended. Received June 16, 2014; accepted December 1, 2014