CADIC   02618
CENTRO AUSTRAL DE INVESTIGACIONES CIENTIFICAS
Unidad Ejecutora - UE
congresos y reuniones científicas
Título:
A new peat bog testate amoeba transfer function and quantitative palaeohydrological reconstructions from southern Patagonia.
Autor/es:
VAN BELLEN S, ; MAUQUOY D.,; PAYNE R., ; ROLAND T., ; PANCOTTO V., ; DALEY, T., ; HUGHES P., .; LOADER N., ; STREET-PERROT A
Reunión:
Congreso; American Geophysical Union (AGU) Fall Meeting.; 2013
Resumen:
Testate amoebae have been used extensively as proxies for environmental change and palaeoclimate reconstructions in European and North American peatlands. The presence of these micro-organisms in surface samples is generally significantly linked to the local water table depth (WTD) and preservation of the amoeba shells downcore allows for millennial length water table reconstructions. Peat bog archive records in southern Patagonia are increasingly the focus of palaeoecological research due to the possibility of detecting changes in the Southern Westerlies. These Sphagnum magellanicum-dominated peat bogs are characterised by a wide range of water table depths, from wet hollows to high hummocks (>100 cm above the water table). Here we present the first transfer function for this region along with ~2k-year palaeorecords from local peat bogs. A modern dataset (155 samples) was sampled along transects from five bogs in 2012 and 2013. Measurements of WTD, pH and conductivity were taken for all samples. The transfer function model was based on the 2012 dataset, while the 2013 samples served as an independent test set to validate the model. Besides the standard leave-one-out cross-validation, we applied leave-one-site-out and leave-one transect-out cross-validation, which are effective means of verifying the degree of clustering in the dataset. To ensure that the environmental gradient had been evenly sampled we quantified the root-mean-squared error of prediction (RMSEP) individually for segments of this gradient. Ordinations showed a clear hydrological gradient in amoeba assemblages, with the dominant Assulina muscorum at the dry end and Amphitrema wrightianum and Difflugia globulosa at the wet end. Canonical correspondence analysis showed that WTD was the most important environmental variable, accounting for 18% of the variance in amoeba assemblages. A weighted averaging-partial least squares model showed best performance in cross-validation, using the 2013 data as an independent test set. Any spatial autocorrelation was minimal although the model still appeared less effective in predicting WTD for sites not included in the training set. The segment-wise RMSEP showed that the WTD gradient was generally evenly sampled with RMSEP below 15 cm for most of the gradient, much lower than the standard deviation of the mean of all WTDs. Preliminary results from peat cores sampled from the same peat bogs show surprisingly stable water tables over the last ~2k years in Andorra bog but more variation in nearby Tierra Australis bog. Peat accumulation rates in Andorra bog are among the highest recorded in temperate bogs with around 4 m of peat accumulated during the last 2k years.