Late Quaternary climate change in the Awatere Valley, South Island, New Zealand using a sine model with a maximum likelihood envelope on fossil beetle data

摘要

We present a new climatic reconstruction method appropriate for biological proxies where modern distributions are poorly defined and data sets are small. The technique uses a sine function in conjunction with maximum likelihood estimates of best high and best low values for the distribution of each species. To demonstrate the model we present temperature reconstructions for the Last Glacial Maximum (LGM) and Holocene from beetle fossil assemblages from the Awatere Valley, New Zealand. The temperature estimates are determined by the mutual overlap of the climate range for all the species in the assemblage. The overlap is then compared with modern physio-chemical conditions.For our example, we estimate the LGM summer (February) mean temperature was about 3.5-4degreesC cooler, and July (winter) mean daily minimum temperature was about 4-5degreesC cooler than present day temperatures. The maximum likelihood estimates broaden the reconstructed temperature ranges to 2.5-5degreesC cooler for February temperatures and 3.5-6.0degreesC cooler for mean minimum daily temperature of the coldest month (July). These estimates are consistent with LGM temperature estimates of 4-7degreesC from other climate proxy indicators.Estimates of Holocene temperatures are very similar to modern. Estimates are compared with results from the established mutual climatic range (MCR) technique and the results are compatible. MCR is less robust than the sine model approach for these data because it requires the pre-determination of the critical physiochemical controls and assumes Gaussian distributions in climate space. The sine model is conceptually superior to traditional BIOCLIM modelling, with which it shares many features, because BIOCLIM also assumes Gaussian distributions and the sine model allows attribute testing of the data sets which are not possible with BIOCLIM. (C) 2004 Elsevier Ltd. All rights reserved.