Premium
Spawning migrations in landlocked Atlantic salmon: time series modelling of river discharge and water temperature effects
Author(s) -
Trépanier S.,
Rodríguez M. A.,
Magnan P.
Publication year - 1996
Publication title -
journal of fish biology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.672
H-Index - 115
eISSN - 1095-8649
pISSN - 0022-1112
DOI - 10.1111/j.1095-8649.1996.tb01487.x
Subject(s) - salmo , ordinary least squares , landlocked country , fishery , fish migration , fish measurement , regression , discharge , series (stratigraphy) , regression analysis , autocorrelation , biology , time series , fish <actinopterygii> , environmental science , oceanography , geography , statistics , geology , drainage basin , mathematics , paleontology , cartography , political science , law
River discharge and water temperature are frequently cited as controlling the upstream migration of adult salmonids to their spawning areas. The results of earlier studies on the effect of these environmental factors were examined. The statistical methods employed in some of these studies failed to consider the serial correlation often found in migration time series. To assess the effects of discharge and temperature on the migratory activity of the landlocked Atlantic salmon (ouananiche, Salmo salar ), 12 years of data on spawning migrations in the Mistassini R., Quebec, Canada, were analysed and the results obtained by ordinary least squares regression and time series regression were compared. In six of the 12 years, upstream migratory movement was related negatively to changes in river flow, suggesting that fish favour falling water phases for ascent. Water temperature appeared to have little effect on migratory movement. The mean body size of migrating fish decreased significantly throughout the summer; early migrants were on average 11.4% larger (mean fork length 522mm) than late migrants (469 mm). Larger, 3‐lake‐year salmon migrated 7.2 days earlier than 2‐lake‐year salmon. Because the residuals from ordinary regression exhibited strong autocorrelation, time series regression was more appropriate than ordinary regression for the analysis of migration time Series.