TY  - UNPB
N1  - Thesis digitised by ProQuest.
A1  - Goldsmith, Benjamin John
PB  - UCL (University College London)
KW  - Biological sciences; Diatoms; Lowland rivers
Y1  - 2002///
M1  - Doctoral
TI  - A diatom-based model to monitor trophic status in lowland rivers using artificial substrata
AV  - public
N2  - The aim of this thesis was to develop the methodologies required to sample diatoms from lowland rivers in southern England, and to use these techniques to establish a diatom-based model to assess trophic status. A number of different artificial and natural substrata were compared from four lowland river sites of contrasting nutrient chemistry to assess the most suitable means of obtaining representative, repeatable and diverse diatom assemblages. Natural substrata supported more variable diatom assemblages than the artificial substrata and could not be kept constant between all lowland river sites. Thus two artificial substrata were chosen for diatom sampling: rough ceramic tiles and frayed polyethylene rope. From a chemical survey of 115 lowland river sites, 61 were chosen which covered a wide phosphorus gradient (3-7530 ?gL-1 FRP). From these sites the two artificial substrata were used to collect diatom samples, in order to develop two independent diatom training sets. Multivariate statistical techniques (PCA, DCA and CCA) were used to investigate the chemical relationships, species relationships and environmental-species interactions. From these analyses a strong, independent, relationship was identified between the diatom assemblages and the phosphorus concentration of the river sites. These data were demonstrated to be suitable for weighted average (WA) modelling. Weighted average regression and calibration techniques were used to generate diatom-based transfer functions, from which river phosphorus levels could be estimated with an acceptable level of error (r2 > 0.80, RMSEP < 0.30). When tested on independently derived diatom samples, the models performed well if the diatoms had been collected at the same time of year as the training set samples (autumn). Considerable seasonal variation was demonstrated within the diatom assemblages however, and the predictive power of the autumn-based models was greatly reduced when applied to samples collected at other times of the year. This thesis demonstrates that with carefully chosen sampling techniques, diatoms can be used as a reliable bio-monitoring tool for the assessment of phosphorus levels in lowland rivers. It is suggested that with the temporal and spatial extension of the training sets developed in this thesis, diatoms could provide a fast and effective means of assessing and monitoring the trophic status of lowland rivers.
UR  - https://discovery.ucl.ac.uk/id/eprint/10099996/
EP  - 356
ID  - discovery10099996
ER  -