A statistical method for the robust detection of interannual changes in plankton abundance: analysis of monitoring data from the Bay of Fundy, Canada

Plankton abundance time series were analysed with an emphasis on the detection of interannual variability. These data covered the period 1988–1999 and were collected in the western Bay of Fundy, Canada as part of an ongoing regional monitoring program. The abundance observations considered here were obtained from water samples at a single station (44°59.57′N, 66°44.36′W) and at four depths (0, 10, 25 and 50 m). The analysis was based on log-transformed abundance of the taxonomic groups diatoms, dinoflagellates and zooplankton. The time-varying abundance level and its error variance were estimated using a statistical time series method designed for the identification of interannual variability using irregularly sampled and noisy data. To this end, plankton abundance measurements were fitted to a cyclic model using a Kalman filter and associated fixed interval smoother. Maximum likelihood procedures were used to determine all unknown system parameters. The model provided a good fit to these data and allowed identification of year to year changes in the timing and magnitude of the annual abundance cycle. The greatest interannual variability was found in the dinoflagellates and diatoms, with only slight changes in the zooplankton. It was shown that while the annual cycle is readily determined from the observations, interannual variations were near the detection limit. For much of the study period, plankton abundance was not statistically distinguishable from a fixed and repeating mean annual cycle at the 90% confidence level. However, during many time periods dinoflagellates and diatoms, and to a lesser extent zooplankton, exhibited significant interannual variations in abundance. A retrospective analysis of the monitoring program suggested that to achieve robust detection of interannual abundance changes, the variability associated with the water sample based point measurements must be decreased, either through replication, more frequent sampling, or alternative sampling methodologies.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.