Temporal trends in a large marine ecosystem

Devine, Jennifer A. (2006) Temporal trends in a large marine ecosystem. Doctoral (PhD) thesis, Memorial University of Newfoundland.

[English] PDF - Accepted Version Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission. Download (7MB)

Abstract

Different scales of change imply the need for different approaches to study change in ecosystems. I determined if proximate causes for ecosystem change are apparent at different biological and spatial scales by analyzing time series at fairly short temporal scales (< 30 years). I determined whether complex dynamics involving abundance could be described for a single species whose population units are distinctly defined, for two deep-sea species, and for the demersal fish community of the northeast Newfoundland-Labrador Shelf using two newer approaches of multivariate time series analysis: minimax autocorrelation factor analysis (MAFA) and dynamic factor analysis (DFA). I analyzed trends in abundance and, for the demersal community, mean size and investigated the relationships between observed trends and external factors (environmental, exploitation and natural (non-human) predation) operating on time lags. No one factor fully explained fish population and community dynamics in the Northwest Atlantic, but rather a combination of factors operating over several temporal scales were largely responsible for the dynamics seen today. No one scale captures all the dynamics in abundance and mean size for populations, deep-sea species, or the fish community. -- Before using the techniques on deep-sea species, I wanted to determine if the data were adequate to describe trends in "non-traditional" fishery species; for this, I used generalized linear models. I discovered several deep-sea species qualified as endangered. From that analysis arose questions regarding statistical rigor: does the loss of comprehensiveness bias results? I assessed several approaches to analyzing population change that explicitly dealt with differing degrees of data filtering and the comprehensiveness of associated metrics with a detailed examination of two North Atlantic endemic deep-sea species, Macrourus berglax and Coryphaenoides rupestris. Statistical rigor is necessary, not to show that a trend exists in available data for deep-sea species, but to ensure that the trend is real. When assessing temporal trends under any conditions, ensuring the comparability of the index over time is extremely important.

Actions (login required)

View Item