Long-term plant community dynamics has been investigated by analyses of macro- and microfossil remains found in unconsolidated limnic and terrestrial sediments, chiefly of Late-Quaternary and Recent age. Plant macrofossils preserved in packrat (Neotoma) midden sites in semi-arid regions have yielded the most informative data for the period between the lastest glacial to the present day. The remains can, with few exceptions, be identified to species level, they represent a large (> 75%) proportion of the species composition of vegetation within a radius of 30-50 m of a site, and they can be dated by the accelerator mass-spectrometric radiocarbon method. Most sites have been found in the arid southwest of the United States and significant contributions have been made recently to understanding of long-term changes in species composition and shifts in range and elevational limits of species. Similarly, analysis of macrofossils in lake, mire and river sites can draw on the taxonomic precision, small source area and potential for precise radiocarbon dating of samples, to provide detailed reconstructions of the local vegetation of lacustrine, mire and fluvial habitats. The vast literature reporting these investigations deals with sites chiefly in northern temperate regions, but recent studies from tropical and warm-temperate regions show the promise of the approach. The use of analyses of microscopic charcoal preserved in sediments has increased significantly in the past few decades as part of attempts to reconstruct forest fire history and, when combined with other analyses (pollen, physical chemistry), to investigate past community dynamics. The major limitations of the method are poor chronological control in many sediments, uncertainties about source area and limited taxonomic precision. On the other hand, macroscopic remains of charcoal can usually be identified to species level, can be dated accurately and have readily defined source areas. However, sites with high frequencies of sample occurrence spanning long, continuous time intervals are very rare, as a remarkable example from the humid tropics of Africa illustrates. Pollen analysis of lake and mire sediments has a long history and has yielded a vast literature. Its weaknesses as a tool in the three main fields of interest where it is used - palaeoclimatology, cultural history and plant community dynamics - are well known and increase in regions farther north and south of the boreal and temperate regions, as the frequency of anemophily declines among the major elements of the terrestrial plant communities. The most promising developments in pollen analysis as applied to community dynamics are in small, within-canopy sites in forested regions, and small (< 5 ha) lakes with fine-scale temporal resolution.
