Premium
Testate amoebae as proxies for mean annual water‐table depth in Sphagnum ‐dominated peatlands of North America
Author(s) -
Booth Robert K.
Publication year - 2008
Publication title -
journal of quaternary science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.142
H-Index - 94
eISSN - 1099-1417
pISSN - 0267-8179
DOI - 10.1002/jqs.1114
Subject(s) - testate amoebae , peat , sphagnum , water table , physical geography , hydrology (agriculture) , environmental science , table (database) , substrate (aquarium) , calibration , ecology , sediment , snow , taxon , water column , water level , geology , geography , statistics , biology , paleontology , mathematics , geomorphology , cartography , geotechnical engineering , computer science , data mining , groundwater
Peatland‐inhabiting testate amoebae are sensitive indicators of substrate‐moisture conditions and have increasingly been used in palaeohydrological studies. However, to improve accuracy of testate‐amoeba‐based hydrological inferences, baseline ecological data on rare taxa, a larger geographic network of calibration sites, and incorporation of long‐term estimates of water‐table depth are needed. Species–environment relationships at 369 sites from 31 peatlands in eastern North America were investigated. Long‐term estimates of water‐table depth were obtained using the method of polyvinyl (PVC) tape‐discolouration. Transfer functions were developed using a variety of models, and validated through jackknifing techniques and with an independent dataset where water‐table depths were directly measured throughout the growing season. Results indicate that mean annual water‐table depth can be inferred from testate amoeba assemblages with a mean error of 6 to 8 cm, although there is a slight systematic bias. All transfer function models performed similarly and produced similar reconstructions on a fossil sequence. In a preliminary effort towards development of a comprehensive North American calibration dataset, data from this study were combined with previous studies in Michigan and the Rocky Mountains ( n = 650). This combined dataset had slightly larger mean errors of prediction (8–9 cm) but includes data for several rare taxa. Copyright © 2007 John Wiley & Sons, Ltd.