The increasing interest in the use of palacoenvironmental indicators and in particular diatoms to reconstruct past changes in sea level has highlighted the need for a more precise methodology that (a) provides quantitative reconstructions, and (b) is applicable to a wide range of sedimentary environments. Despite the widespread and increasing recent interest in the use of diatoms as indicators of estuarinep alaeoenvironmentsa nd sea-levelc hange,e xisting interpretative models, basedo n simple classification of taxa into freshwater, brackish or marine forms, provide only _qualitative estimates of past conditions. Resulting palaeoenvironmental reconstructions are at best crude, offering some indication of past mean sea level height, and at worst erroneous, as they fail to consider the effect of post-mortem transport and other taphonomic processes. This study aims to address these problems for the coast of Britain by developing a more robust quantitative method for using diatoms as indicators of estuarine palaeoenvironments and sea-level change. More specifically, it aims to develop a quantitative predictive model (transfer function) that relatesd iatom assemblagec omposition to salinity, habitat, depositional environment and tidal level around the coast of Britain. This is done through the collection and analysis of a training data set of diatom assemblagesa nd environmental variables (salinity class, elevation, grain size, habitat type and sediment organic content) from 25 sites around the coast of Britain. Qualitative and quantitative relationships within the diatom assemblagesa nd between the diatom assemblagesa nd coastale nvironmentalv ariables are explored using TWINSPAN and canonical correspondencea nalysisr espectively. The key environmental variables driving diatom assemblagesin the intertidal environment are shown to be elevation, salinity and sediment particle size. Habitat type and site location also explain a significant amount of variation in the diatom data, suggestingr egional differences in diatom assemblagesn ot accounted for by geornorphological and sedimentological differences between sites. The final transfer function for inferring normalised tidal height has a root mean squared error (RMSE) of 0.26, and a squared correlation (1-2) between observed and diatom-inferred normalised tidal height of 0.61. Corresponding error estimates under cross-validation by leave-one-out are 0.34 and 0.35 for RMSEjack and rjack 2 respectively. The poor performance of the model in comparison to published regional transfer functions is concluded to be due to the mergt:i In g of data from a large number of sites over a large geographical area. Such merging - has apparently introduced a large amount of noise into the diatom / elevation relationship, and is probably related to the increased heterogeneity and interaction of sediment typeý and elevation, and to the observed regional overprint in the diatom assemblages. Analogue matching to infer Sample habitats from the diatom data performs with a success Z): rate of 59%. Further merging of the habitat types based on ecologically similarities increases the success rate to 82%. The correct scale of trade off between coverage of palaeoenvironments, fossil diatom species 4): and reduction of regionality in the modem training dataset is an issue that needs further research before this model may be applied to core material to assist in palaeoenvironmental reconstructions.
Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:391251 |
Date | January 2001 |
Creators | Lewis, Mary Gwendolyn |
Publisher | University of Newcastle Upon Tyne |
Source Sets | Ethos UK |
Detected Language | English |
Type | Electronic Thesis or Dissertation |
Source | http://hdl.handle.net/10443/565 |
Page generated in 0.0019 seconds