Mixed messages: Unmixing sedimentary molecular distributions reveals source contributions and isotopic values

Abstract

Plant leaf waxes and their isotopic composition are important tracers of ecological, environmental, and climate variability, with strong preservation potential in sedimentary archives. However, they represent an integrated, and often complicated, signal of vegetation and hydrology within a watershed. Here, we report a new approach for examining complex mixtures of n-alkanes in sediments and their isotope values: non-negative matrix factorization (NMF). NMF identifies the endmembers in a mixture from the integrated n-alkane data and provides quantitative information on the relative importance of those endmembers across samples. We apply this approach to a synthetic dataset and two previously published datasets to illustrate its uses. Our application of NMF to re-analyse previously published data reveals new insights into past climate and ecological change. We demonstrate that NMF allows a user to 1) identify potential mixing problems, 2) evaluate which specific compounds in a mixture carry the isotope signal that can best address a given scientific objective, 3) determine compound concentrations after excluding contributions from particular endmember sources, and 4) calculate isotope values of different sources. NMF provides a quantitative approach for evaluating the influence of endmember mixing on molecular concentrations and isotope values within a dataset. The re-analysis of two published datasets reveals new quantitative insight into Holocene Arctic climate and Neogene vegetation change.