Hill coefficients from different sources of sediment core from Lake Bayan Nuur

Estimating aquatic and pollen-assemblages, and community diversity. Comparison of multivariate datasets of bioproxies. To evaluate the similarity or dissimilarity of aquatic (lacustrine) and pollen assemblages, and their response to environmental changes, different datasets were compared using the statistical methods applied in modern species ecology. The richness and diversity of all bioproxies (pollen, diatoms, cladoceran, and chironomids) were calculated as the effective taxon numbers N0, N1, and N2 proposed by Hill (1973, doi:10.2307/1934352). Investigators using Hill numbers should report, at least, the diversity of all species (q0), of ''typical'' species (q1), and dominant species (q2) (Chao et al., 2014, doi:10.1890/13-0133.1). where S is the number of taxa in the sample count and the ith taxon has a relative abundance pi. The parameter q determines how sensitive the estimate is to taxon frequencies, where q = 0 is simply the total number of taxa in a community (Hill N0, 0D), q = 1 is the number of common taxa represented by the exponential of the Shannon-Wiener diversity index (Hill N1, 1D), and q = 2 is the number of dominant taxa represented by the inverse of Simpson's diversity index (Hill N2, 2D) (Chao et al., 2014, doi:10.1890/13-0133.1). These measures give easily interpretable numbers and provide information at three different levels based on how rare and abundant taxa are weighted. Diversity analyses for pollen, diatom, chironomid, and cladoceran (original counts) were conducted using the iNEXT package version 2.0.12 for R. Integrated curves that allow rarefaction and extrapolation were used to standardize samples based on sample size or sample completeness, and facilitate the comparison of biodiversity data; and the Hill numbers (N0, N1, and N2) were calculated using a non-asymptotic approach. A comparison of multivariate datasets can be performed using Procrustes rotation, which assesses the overall degree of correlation between two or more ordination results, and finds an optimal superimposition that maximizes their fit. PROTEST performs a random permutation test and assesses the degree of concordance between two matrices, producing the significance of the Procrustes fit as an r value with an associated p-value to indicate the likelihood of the relationship occurring by chance (Jackson, 1995, doi:10.1080/11956860.1995.11682297); and this approach has been employed in evaluating the similarities among different proxies or between environmental signals and their driving factors. We evaluated the similarity in temporal evolutions among pollen, diatom, chironomid, and cladoceran using Procrustes rotation and tested the significance of any relationship found with the associated PROTEST permutation test for the non-metric multidimensional scaling (NMDS) results of these datasets. Preliminarily to NMDS, Procrustes analyses, and PROTEST, all datasets were temporally standardized for each 100-year interval between 2850 and 50 cal yr BP. The NMDS, Procrustes analyses and PROTEST were carried out using the vegan package version 2.5-2, and linear interpolation was achieved by interp.dataset function in the rioja package version 0.9-15.1. Evenness (E) was calculated as the N2/N0 ratio. Evenness is a measure of the relative frequency of species in the community (sample). If evenness is at maximum, then all species are likely to be represented in a sample comprising only a few per cent of the individuals of the community. In contrast, if evenness is very low only a fraction of the species present in the entire population is likely to be found in even a large sample comprising almost half the individuals of the community. Interpretations of fossil assemblages in terms of evenness dynamics prove more rewarding than studies of past species richness (N0). Detrended canonical correspondence analysis (DCCA), the direct form of DCA, with species assemblage changes constrained to sediment age as the sole environmental variable, was used to develop quantitative estimates of compositional turnover, scaled in standard deviation (SD) units for each taxonomic group. The analysis is performed in CANOCO 5. Pollen percentages were square-root transformed. No down weighting of rare species was selected in this study because of the many rare pollen taxa present in our data. The change in weighted average (WA) sample scores reflects compositional change or turnover in standard deviation (SD) units along the temporal gradient