A quantitative approach for classifying fish otolith strontium:calcium sequences into environmental histories

Hedger, Richard D. and Atkinson, Peter M. and Thibault, Isabel and Dodson, Julian J. (2008) A quantitative approach for classifying fish otolith strontium:calcium sequences into environmental histories. Ecological Informatics, 3 (3). pp. 207-217. ISSN 1574-9541

Full text not available from this repository.


Zoning algorithms were used to quantitatively classify strontium:calcium otolith sequences into fish environmental histories. Otoliths were acquired from 162 American eels (Anguilla rostrata) caught in the Gaspé region of Québec, Canada, and Sr:Ca ratios were determined at an interval of 10 µm along a transect from the core to the edge of each otolith (the otolith sequence) using an electron probe microanalyzer. Changes between freshwater and brackish water occupancy were determined with reference to a sample of non-anadromous species including brook char (freshwater) and Fundulus sp. (brackish water). Three algorithms were then applied separately to zone the sequences into environmental histories: (i) a local zoning algorithm, which used a split-moving window; (ii) a global zoning algorithm, which used a recursive method; and (iii) an optimization zoning algorithm, which maximized the combined value of selected statistics of the fitted model within a decision-rule framework. Zones were further classified into being of either freshwater or brackish water. All algorithms produced classifications that were not significantly different to those determined using the standard approach of qualitative interpretation, demonstrating the applicability of a quantitative approach. The advantages of the quantitative approach are that (i) the statistics of the model fit provide information on environmental history patterns that is generally not available from qualitative interpretation, and (ii) the parameters of the algorithm can be reported, allowing methodological consistency between different researchers, enabling the potential for more robust meta-analyses.

Item Type:
Journal Article
Journal or Publication Title:
Ecological Informatics
Additional Information:
M1 - 3
Uncontrolled Keywords:
ID Code:
Deposited By:
Deposited On:
21 Dec 2015 15:22
Last Modified:
22 Nov 2022 02:41