Journal
SCIENCE OF THE TOTAL ENVIRONMENT
Volume 760, Issue -, Pages -Publisher
ELSEVIER
DOI: 10.1016/j.scitotenv.2020.143320
Keywords
Otolith chemistry; Concentration-discharge; Water chemistry; Random forests
Categories
Ask authors/readers for more resources
This study evaluated the spatiotemporal variability of trace metal ratios in the St. Lawrence River and its major tributaries, confirming their stability over time. Results demonstrate the influence of element selection and geochemical similarity of tributaries on the accuracy of river classification based on element ratios.
Tracemetals found in the calcified structures of fish (i.e. otolith, scales and vertebrae) serve as proxies for the ambient water composition at the time of mineralization, and these trace metals are increasingly used as a tool for assessing population structure and the migratory patterns of fish. However, the appropriate scale (e.g. resolution) for such applications can be uncertain because of a poor understanding of the spatiotemporal variations of metal-to-calcium ratios (Me:Ca) in the studied watersheds. This study aims to assess Me:Ca spatiotemporal variability within the St. Lawrence River and nine major tributaries and evaluate the ability of random-forest models to correctly identify rivers on the basis of their elemental composition. We tested the influence of daily discharge on four measured ratios (Sr:Ca, Ba:Ca, Mg:Ca and Mn:Ca) to document local and regional trace element sources and dynamics. The four element ratios displayed a low spatiotemporal variation, reflecting a marked stability over time. We observed that most element- and tributary-specific concentration-discharge relationships were either not significant or showed a weak influence, thereby confirming a stable point source dynamic. The classification performance based on a four-element model (Sr:Ca, Ba:Ca, Mg:Ca and Mn:Ca) produced a classification accuracy of 92.5%, which correspond to a small decrease of accuracy compared to the full model (25 elements, 96.6% of correct classification). A classification based on two elements (Sr:Ca and Ba:Ca) produced a lower classification accuracy (72.6%). Classification errors related mainly to tributaries in close proximity, a problem tempered by grouping these geochemically similar watersheds. Our results show that surveys of the elemental fingerprint of regional tributaries within a given region can provide critical information to determine the appropriate scale (tributary or watershed) for trace metal analysis of the hard-calcified parts of fish. (C) 2020 Elsevier B.V. All rights reserved.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available