Flessa, K.W., 2003.  Beyond baselines in historical ecology: estimating the natural range of environmental variation.  Geological Society of America Abstracts with Programs 35: 84.

Few, if any, extant communities are completely free of the direct or indirect effects of human activity. As a consequence, fossil (pre-human influence) communities are often the best source of information on pristine, baseline conditions and natural environmental and biotic variability. Reconstructing environmental baselines using paleontological approaches is only the first step in assessing the impact of environmental change and establishing targets for ecological restoration. Because environments and their biotas vary under natural conditions baseline reconstructions must be accompanied by estimates of the natural range of environmental variability so that human-induced change can be distinguished from natural variation. Characterizing variation at the scale of seasons to centuries is probably most important: shorter-term daily variation may be too noisy while longer-term variation may encompass directional environmental and biotic changes. Because most direct-dating techniques, even within in the Holocene, cannot resolve seasonal to centennial differences in age, dendrochronological, sclerochronological, and high-resolution stratigraphic approaches must be used to establish the natural range of environmental variation in baseline communities. Stable isotopic and other geochemical environmental indicators preserved within accretionary hard parts and laminated sediments are natural archives of environmental variation; reading and interpreting these archives requires both an understanding of the controls of biotic or sediment accretion, and high-resolution sampling and analysis. Characterizing natural variation in species composition and abundance is also important. Fortunately, as Kidwell has shown, sample-to-sample variation in rank abundance largely reflects natural variation in composition and rank abundance. In addition, variation in the frequency distribution of time-since-death may reflect natural temporal variation in original abundance.