Reconstruction of Holocene sea-levels using diatom- and pollen-based microfossil transfer functions, west coast of Scotland, UK

Abstract

Due to its sensitivity to isostatic and eustatic sea-level change the west coast of Scotland has been an important area for studying sea-level change. Recently, data from north west Scotland have been used to refine GIA models because of the presence of the ice sheets of the last glacial maximum (LGM) and its long record of sea-level change exceeding 10 000 yr. The Holocene highstand record of the area studied in Scotland is preserved within the many isolation basins, tidal marsh sediments and raised beaches which collectively preserve a diachronous record of marginal marine sedimentation over the course of the Holocene. The Arisaig area preserves one of the longest records of relative sea-level change in Great Britain, in excess of 16 ka (e.g. Shennan eta/., 1993, 1994, 1995a and 1995b). The microfossil sediment sequence from Mointeach Mhor North, is an important bio-stratigraphical sequence from the north west coast because it records both the rise and fall of sea level throughout the mid Holocene more precisely than many other sites. Where until recently the data used to reconstruct RSL were sea level index points taken from regressive and/or transgressive contacts, verified by the microfossil data, this investigation aims to for the first time reconstruct sealevel change through the sediment sequence using diatom, pollen and multiproxy microfossil-based transfer functions.This thesis also highlights some of the problems encountered associated with the transfer function technique and microfossil choice that may limit or constrain the applicability of transfer functions in sea-level studies, including spatial autocorrelation, over- and under-estimation of optima, multiple analogues, no analogues and the problem of allochthonous and autochthonous species.This research project has determined that regional and local multi-proxy transfer functions have the potential to reconstruct reference water levels (RWLs) more precisely than diatom and pollen training sets (Table 4.16). When the regional transfer functions were ran again using only the samples within the multi-proxy training set to allow direct comparison, the multi-proxy training set still produced the best performance statistics. When the training sets were used to calibrate the fossil cores/monoliths to produce estimates of RWL (Chapter 6) the original samples containing all available samples gave more sensible and reliable estimates of RWL than the training sets that only contained the 75 assemblages in the multi-proxy training set. In Chapter 6 the original regional diatom training set produced the most reliable and sensible estimates of RWL. The excellent statistical parameters produced during regression in Chapter 4 that indicated multi-proxy training sets performed better than diatom and pollen training sets are not reflected in the RWL reconstruction when the multi-proxy training sets are used to calibrate the fossil assemblages. This indicates that diatom-inferred RWL reconstructions are the most reliable but when diatoms are not present there is the potential to use a multi-proxy approach. When considering whether local or regional training sets are the most precise, the local diatom, pollen and multi-proxy training sets from Saideal nan Ceapaich in Morar continually provided the most precise estimates of RWL. However, regional training sets provided the most reliable estimates of RWL.