group of major actors. Selection must therefore occur on a stage with continuously changing combinations of environmental variables and competitors, and the local setting (specific soils, relief, etc.), local players, and recent arrivals generally have only a short-term influence on the evolutionary play.
Within this chapter, I begin by describing pollen data and their ability to record patterns in present and past vegetation. I then summarize what recent mapping studies of late Quaternary pollen data have illustrated about past changes in vegetation and their link to climate change. Simulations of past climates by climate models have aided these studies (Barnosky et al., 1987; Webb et al., 1987). These models are being used as tools for generating hypotheses that the data can test (COHMAP, 1988; Wright et al., 1993). In the final sections, I discuss (1) how an understanding of the space and time scales for taxonomic ecological units leads to a temporal separation of ecological units from evolutionary units for many plant taxa (McDowell et al., 1990) and (2) the advantages of a 4-dimensional space-time perspective of the data.
Pollen data are sensitive to a wide spectrum of spatial and temporal variations in the vegetation (Figure 13.1). These can vary from local succession (Janssen, 1967; Bradshaw, 1988; Edwards, 1986) up to long-term continental-scale changes in plant formations (Jacobson et al., 1987; Huntley, 1990). The data can also record how the vegetation was influenced by humans (Behre, 1988; Birks et al., 1988; McAndrews, 1988), disease (Davis, 1981; Webb, 1982; Allison et al., 1986), fire (Patterson and Backman, 1988; Clark, 1990), soils (Webb, 1974; Brubaker, 1975; Jacobson, 1979; Bernabo, 1981; Tzedakis, 1992), topography (Janssen, 1981; Gaudreau et al., 1989; Lutgerink et al., 1989; Jackson and Whitehead, 1991), and climate (Bartlein et al., 1984; Webb et al., 1987; Huntley and Prentice, 1988; Prentice et al., 1991). With such a variety of possible processes and influences, palynologists must choose appropriate methods of data collection, analysis, and display to obtain results indicative of the vegetational variations of particular interest (Webb et al., 1978, 1993; Jacobson and Bradshaw, 1981; Grimm, 1988).
A key metaphor for understanding the interpretation of pollen data is to think of them as remotely sensed vegetation data (Webb, 1981; Webb et al., 1993). Just as the current vegetation emits or reflects radiation that remote sensors on satellites intercept, so too does (and has) the current (and past) vegetation shed pollen that accumulates "remotely" (i.e., well away from the source) in lakes and bogs. Both types of "remote" sensors record data with certain sampling characteristics (e.g., spatial and temporal resolution), and their data need calibration in terms of climate or vegetation variables. One major thrust in palynology, therefore, has been the analysis of modern pollen data to see what features of the modern vegetation are recorded (Figures 13.1 to 13.3). Palynologists have attempted to learn what the modern vegetation looks like in pollen terms (Webb, 1974), so that they can better visualize the past vegetation, which is only represented in pollen (or other fossil) terms. Palynologists have also studied modern data in order to learn how varying the sampling characteristics of the data can alter what vegetational features or processes are represented (Janssen, 1966; Andersen, 1970; Webb et al., 1978; Heide and Bradshaw, 1982; Bradshaw and Webb, 1985; Prentice et al., 1987; Prentice, 1988; Jackson, 1990, 1991).
Recent mapping studies at the subcontinental scale (Peterson, 1984; Webb, 1988; Huntley, 1990; Anderson et al., 1991) have shown how well the modern pollen data match contemporary vegetation patterns. In eastern North