Data S1. Material and Methods: Non-metric multidimensional scaling (nMDS) is an ordination technique which is particularly useful for the exploration of patterns of variation in multivariate species data sets. The main advantage over other ordination techniques [e.g. principal components analysis (PCA) or correspondence analysis (CA)] is that it does not impose an underlying response model for the species as nMDS can use any dissimilarity matrix as input. nMDS aims to preserve the rank ordering of the original pair-wise distances rather than the distances themselves. It was preferred over other ordination techniques as it has been shown to be a more robust technique for indirect gradient analysis and is generally a very effective method for ordinating ecological community data (Minchin, 1987; McCune & Grace, 2002). The Bray Curtis distance coefficient was used to calculate the dissimilarity matrix, as it has been demonstrated to provide a robust monotonic relationship with ecological distances as well as a robust, linear relationship over a wide range of distances (Faith et al., 1987). The stress value produced by nMDS is an evaluation on how well the computed data matrix is representing the original data set in n-dimensional space and can therefore be understood as a measure of the variation explained. According to Clarke (1993) and McCune & Grace (2002), stress values in the range of 10–20% are common in most ecological community data and indicate a decent fit between the original distance of objects and the fitted values. Procrustes rotation analysis and the associated PROTEST permutation test were used to explore the similarity/dissimilarity between different data sets and to test the significance of any relationship found. Procrustes rotation assesses the overall degree of correlation between two data sets on the basis of their ordination (nMDS) scores through a Procrustean superimposition approach (Gower, 1971; Peres-Neto & Jackson, 2001). Geometrically, Procrustes rotation can be viewed as if one ordination configuration remains fixed in space, whilst the other configuration is rotated, flipped and scaled in a series of translations relative to the fixed configuration. The degree of similarity between the configurations is then computed between the fixed and translated configuration. The justification for Procrustes rotation is that an nMDS ordination may be rotated arbitrarily without loss of information and that the absolute spacing of the samples in nMDS space is a function of the scale of the data and the dissimilarity used. As such the ordination configuration provided by nMDS is data set specific and a means is needed by which to such-derived configurations may be compared. Diagnostics such as Procrustes rotation sum of squares and the root mean square error (RMSE) indicate the performance of the model; the correlation-like statistic PROTEST (r) and the associated P value indicate the quality and likelihood of the fit.

gcb2474-sup-0001-f1-AA.tifimage/tif19018KFigure S1. Radiocarbon age (reservoir effect corrected, in bp) against calibrated ages (ad). Illustrated is the sigma 2 range.
gcb2474-sup-0002-f2-AA.tifimage/tif4170KFigure S2. nMDS ordination biplots (2D) on interpolated and resampled data sets. Dashed line indicates division between samples older 1880 ad and sample younger 1880 ad.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.