Artificial time-averaging of marsh foraminiferal assemblages: linking the temporal scales of ecology and paleoecology

Abstract

Multiple regression models were developed for seasonal test inputs to, and preservation of, marsh foraminiferal assemblages for a two-year period at Bombay Hook National Wildlife Refuge (BHNWR; Smyrna, Delaware). Seasonal assemblages were quite variable and yielded poor regression models. However, signal/noise ratios were amplified using artificially time-averaged (ATA) assemblages, in which separate dead and live abundances of the most abundant species were summed for all seasons. Regression models that used ATA species abundances to retrodict original sample depths accounted for up to ~99% (p < 0.0001) and ~91% (p < 0.023) of the variation of dead and live ATA assemblages, respectively, and usually retrodicted sample depths within 2–3 cm of actual depths.Artificially time-averaged assemblages were also used to extract multidecadal- to centennial-scale sea-level signals from near-surface assemblages at BHNWR formed during the past few centuries. The BHNWR sea-level curve closely resembles one previously published for marshes in Clinton, Connecticut (also based on foraminifera). The technique of artificial time-averaging therefore links the temporal scales of ecology and paleobiology by extracting high-resolution paleoenvironmental signals preserved in the fossil record.