How Rapidly Does Vegetation Respond To Changes In Climate

An important question that arises in studies of high resolution (e.g., varved) sediments is: What is the lag response of vegetation to climate change? Can the pollen record provide information on short period, large amplitude changes in climate, or to put the question more generally: What are the frequency response characteristics

Paleoclimatology MapsPaleoclimatology Maps

FIGURE 9.3 Generalized vegetation map of eastern North America and isopoll maps of selected taxa and groups of taxa at 500 yr B.P. Contours shown are 5 and 10% for forbs, 1,5, and 10% for Cyperaceae, Fagus, and Tsuga, 1,5, and 20% for Picea and Quercus, 1,10, and 20% for Betula, 20 and 40% for Pinus, and 1,3, and 6% for Carya. Forb pollen is the sum of Ambrosia, Artemisia and other Compositae,Chenopodiaceae,and Amaranthaceae pollen (Webb, 1988).

FIGURE 9.3 Generalized vegetation map of eastern North America and isopoll maps of selected taxa and groups of taxa at 500 yr B.P. Contours shown are 5 and 10% for forbs, 1,5, and 10% for Cyperaceae, Fagus, and Tsuga, 1,5, and 20% for Picea and Quercus, 1,10, and 20% for Betula, 20 and 40% for Pinus, and 1,3, and 6% for Carya. Forb pollen is the sum of Ambrosia, Artemisia and other Compositae,Chenopodiaceae,and Amaranthaceae pollen (Webb, 1988).

of the pollen record? Davis and Botkin (1985) attempted to answer this question by the use of a forest growth model to simulate changes in forest composition (basal areas of particular tree species) after steplike changes in temperature. They compared large amplitude, short duration events with smaller amplitude, longer-term changes and concluded that the forest response to climatic cooling lagged 100-150 yr behind, due to "community inertia." There is a natural delay in colonization by new species, due mainly to the shading effects of mature canopy trees. Consequently, they anticipate that changes in forest composition in response to the warming from -1850-1990 will continue for at least another century (Overpeck et al., 1990). They also found that there were similar vegetation responses to large, short events as there were to longer, smaller amplitude changes (Fig. 9.6). The smallest change necessary for a detectable forest response was a ~50-yr change in mean annual temperature of 2 °C, or a 1 °C change sustained over -200 yr. In short, these model experiments suggest that one should not expect the pollen record of (mid-latitude forest) vegetation to resolve the influence of climatic changes "more closely than within a century or two"; rather it provides a "running mean of climatic variation" (Davis and Botkin, 1985).

\ 6CPEAL FOnrST ' UIXC0 FWEST

E3 RSIST1

\ 6CPEAL FOnrST ' UIXC0 FWEST

E3 RSIST1

Paleoclimatology Maps

FIGURE 9.4 Maps of the percentages of (a) hickory, (b) oak, (c) elm, and (d) ash in the vegetation of Michigan (V) compared to the percentages of pollen (P) from the same trees in the modern pollen rain (based on arboreal pollen sum). Modern pollen data based on analysis of uppermost lake sediments (Webb, 1974).

The question of vegetation response to climatic change also involves the notion of ecosystems: Have they always been the same as we see them in the landscape today? Webb (1988), Huntley and Webb (1989), and Huntley (1990a) provide persuasive arguments that modern vegetation should not be viewed as fixed units with a constant composition, which moved regimentally across a region in response to changes in climate. Rather, individual taxa respond differently, leading to a constant change in vegetation composition, at times producing vegetation formations with no modern analogs (e.g., in Late Glacial time in much of western Europe; Huntley, 1990b). This is due partly to the individualistic responses of taxa to climatic changes and their different abilities to migrate, but also due to the fact that in many areas late Pleistocene and early Holocene climates were quite unlike anything seen today (COHMAP members, 1988). According to Webb (1988) "ecosystems and plant assemblages are to the biosphere what clouds, fronts and storms are to the atmosphere . . . features that come and go . . . [with] . . . internal dynamics . .. [but] . . . not of sufficient strength to overcome major changes from the outside." This has important implications for anticipating future changes in ecosystems that may accompany global warming. New communities may well be unlike those of today, or those in the past (Overpeck et al., 1990; Davis, 1991).

With such a view of vegetation change, we inevitably turn to the controversial issue of whether vegetation can ever be considered "in equilibrium" with climate.

Firs! Principal Component

Firs! Principal Component

type percentages (P) in Michigan.The principal components of vegetation reflect major vegetation formations in the state. PC I has a distribution reflecting the change from deciduous forest in the south to mixed coniferous-hardwood forest in the north. It accounts for 25% of variance in the original data set. PC2 depicts primary divisions within the two major vegetation formations, differentiating the northern hardwoods from the pine-birch-aspen forests in the north and the beech-maple and elm-ash-cottonwood forests from the oak-hickory forests in the south. It accounts for a further 35% of variance in the original data set.The first two principal components of pollen mirror the principal components of vegetation, indicating that the spatial pollen data may be used as a reliable indicator of vegetation distribution. Lakes shown by dots (Webb, 1974).

type percentages (P) in Michigan.The principal components of vegetation reflect major vegetation formations in the state. PC I has a distribution reflecting the change from deciduous forest in the south to mixed coniferous-hardwood forest in the north. It accounts for 25% of variance in the original data set. PC2 depicts primary divisions within the two major vegetation formations, differentiating the northern hardwoods from the pine-birch-aspen forests in the north and the beech-maple and elm-ash-cottonwood forests from the oak-hickory forests in the south. It accounts for a further 35% of variance in the original data set.The first two principal components of pollen mirror the principal components of vegetation, indicating that the spatial pollen data may be used as a reliable indicator of vegetation distribution. Lakes shown by dots (Webb, 1974).

This seems to be very much a question of one's definition of equilibrium, and the scale at which one examines the question. Webb (1986, 1987, 1988) argues that vegetation is in dynamic equilibrium with climate; on the timescale of 103-105 yr, climate has changed continuously, and vegetation (viewed on subcontinental scales, at ~103 yr intervals) has kept pace with these changes. Changing the temporal and spatial focus to shorter intervals and smaller areas would no doubt reveal "disequilib-ria" related to migrational, successional, or edaphic influences (Prentice, 1986; Davis et al., 1986). However, this issue is further complicated by differences related to particular types of vegetation or environment (e.g., in a forest environment, if a species can tolerate shade by canopy trees, or if the landscape being invaded is open). In particular, the extent of disturbance in an ecosystem can have an important effect on the ability of a species to occupy new environments even under the pressure of a change in climate (Davis, 1991). Furthermore, in some environments such as desert and semiarid regions, where vegetation has adapted to a high variability of rainfall, a sustained change in precipitation may generate an almost immediate response in vegetation cover30 (Ritchie, 1986). Notwithstanding these specific problems, the evidence is that in midlatitudes, at least, a synoptic view of vegetation at ~2000-3000 yr intervals reveals changes, which are consistent with the changes that occurred in climate, as we currently understand them (Webb et al., 1987).

30 This has implications for vegetation-induced changes in trace gas concentrations (CH4, C02) in relation to climatic forcings, as a rapid response in vegetation over large areas (such as the tropical desert margins) could have more or less immediate feedback effects on the climate system (Petit-Maire et al., 1991).

Second Principal Component

Second Principal Component

Brian Warming

FIGURE 9.6 Upper diagram: Basal area for dominant species of tree plotted against time in a 1200-yr simulation of forest growth on good soils with a reduction of 600 growing degree days at year 400 for a 100-yr interval, then a return to previous conditions, as shown at top of graph. Lower diagram: Basal area for dominant species of tree with a reduction of ~300 growing degree days at year 400 for 200 yr, then a return to previous conditions.The (model) vegetation response to a large, short event is similar to a longer, smaller amplitude change (Davis and Botkin, 1985).

Years

FIGURE 9.6 Upper diagram: Basal area for dominant species of tree plotted against time in a 1200-yr simulation of forest growth on good soils with a reduction of 600 growing degree days at year 400 for a 100-yr interval, then a return to previous conditions, as shown at top of graph. Lower diagram: Basal area for dominant species of tree with a reduction of ~300 growing degree days at year 400 for 200 yr, then a return to previous conditions.The (model) vegetation response to a large, short event is similar to a longer, smaller amplitude change (Davis and Botkin, 1985).

As a test of the hypothesis that vegetation is in dynamic equilibrium with climate, Bartlein et al. (1986) and Webb et al. (1987) used equations that relate modern pollen rain to contemporary climate (see Section 9.6) and applied them to climatic conditions in the past (derived from a general circulation model) to predict what the pollen rain should have been if vegetation was in equilibrium with climate. This approach assumes (a) the equations adequately characterize the pollen-climate relationship; and (b) the models accurately reconstruct past climate. The results show fairly good correspondence between the simulated and observed pollen rain, demonstrating that vegetational lags are not significant on the space and timescales being considered. In another approach, Prentice et al. (1991) used "response surfaces" (see Section 9.6) to predict climate (from 18 ka to 3 ka B.P.) directly from the fossil pollen data. The results were compared with (independently derived)

model-simulated climate, and good agreement was found. As there were no major anomalies between the pollen-derived climate and that derived from the model, they concluded that "continental scale vegetation patterns have responded to continuous climatic changes during the past 18 ka, with lags no greater than -1500 yr" (Prentice et al., 1991). In fact, pollen can clearly register changes over much shorter intervals. For example, many sites show a pronounced change in the pollen spectra during the Younger Dryas interval, which seems to have lasted <1000 years. It seems safe to conclude that in some locations pollen may be a sensitive indicator of abrupt, short-lived climatic changes (lasting perhaps a few hundred years). Elsewhere, the site may be poorly located or the sedimentation rate may be too low to reveal such transient changes in climate.

9.4 POLLEN ANALYSIS OF A SITE: THE POLLEN DIAGRAM

Pollen data from a stratigraphic sequence are generally presented in the form of a pollen diagram composed of "pollen spectra" from each level sampled (Fig. 9.7). A pollen spectrum consists of the number of different pollen grains at a particular level expressed as a percentage of the total pollen count (the pollen sum). Actually, the pollen sum is not always made up of all pollen types counted. For paleoclimatic purposes, the objective of the analysis is to depict climatically significant regional vegetation change, so both arboreal and nonarboreal (shrub and herb) species are included in the pollen sum. Species that commonly grow in wet (lowland) environments around the sample site are usually excluded, though difficulties arise when a particular genus has different species (not easily distinguished by their pollen) that grow in both wet lowland and drier upland environments (e.g., Picea mariana, black spruce, and Picea glauca, white spruce, respectively; Wright and Patten, 1963).

In the pollen diagram, changes in the percentage of one species are assumed to reflect similar changes in the vegetation composition (due consideration being given to the factors of over- and under-representation already discussed here). The problem with this is that apparent changes in one species may occur in the percentage data as a result of changing receipts of pollen from other species because the total must always equal 100% (what Prentice and Webb [1986], call the "Fagerlind effect"). The following example from Faegri and Iversen (1975, p. 160) succinctly summarizes the difficulty:

If we visualize a forest consisting of equal parts of oak and pine, and we use the pollen production figures quoted, we find that the corresponding spectrum will contain 15 percent oak, 85 percent pine. If beech is substituted for pine (apart from the botanical improbability of that succession) the same quantity of oak will give 60 percent of the pollen as against 40 percent beech. If the beech is then replaced by a tree, e.g. Acer spp. or Populus balsamifera, which is scarcely, or not at all, registered in the spectra, we shall find almost 100 percent oak pollen, although the quantity of oak has not changed at all. It is necessary to take into account not only the curve under discussion, but the others as well.

To circumvent such problems, palynologists may calculate pollen flux density (sometimes, incorrectly, called absolute pollen influx values), which is the number

Arboreal taxa sc

100 110

120130-

100 110

120130-

Pollen Diagram

20 40 60 20 20 Percent

FIGURE 9.7 Pollen diagram from Carp Lake, Oregon, spanning the last 125,000 yr, showing I I distinct zones, which were derived objectively from the spectrum of pollen represented. Three types of pine are recorded in the first column: P. contorta or P. ponderosa (white); P. monticola or P. albicautis (black) and indeterminate (shaded) (Whitlock and Bartlein, 1997).

40 60 SD

20 40 60 20 20 Percent

20 20 40 20

FIGURE 9.7 Pollen diagram from Carp Lake, Oregon, spanning the last 125,000 yr, showing I I distinct zones, which were derived objectively from the spectrum of pollen represented. Three types of pine are recorded in the first column: P. contorta or P. ponderosa (white); P. monticola or P. albicautis (black) and indeterminate (shaded) (Whitlock and Bartlein, 1997).

of grains accumulating on a unit of the sediment surface per unit time (Fig. 9.8). However, to do this the sediment accumulation rates must be known. Generally samples are 14C dated at close intervals to establish a mean sedimentation rate. In North America, the rise of Ambrosia (ragweed) pollen at the time of colonial settlement (when forests were being cleared and herbaceous plants were increasing rapidly in numbers) is clearly seen in lake sediments; because the dates of settlement are known, sedimentation rates since then are readily calculated (Bassett and Teras-mae, 1962; McAndrews, 1966; Davis et al., 1973). Clearly the latter method is useful only for obtaining modern pollen flux statistics whereas 14C dating enables influx to be calculated over earlier periods.

Pollen flux values can often clarify a stratigraphic record but they also have some important disadvantages. Sediment focusing in lakes may result in unrealisti-cally high values of total pollen flux, leading to erroneous conclusions. Most importantly, flux calculations require that sediments be closely and accurately dated so that reliable sedimentation rates can be obtained. Relatively few records are sufficiently well-dated, and for large regional climate reconstructions, in which dozens if not hundreds of sites are used, percentage pollen values are always employed. There is strong evidence that in spite of their inherent limitations, such data provide reliable, reproducible, and verifiable paleoclimate reconstructions (T. Webb et al.,

SEDIMENT POLLEN

0 0

Responses

  • kacper
    How does vegetation reflect climate?
    7 years ago
  • Enrico
    How does climate causes the variation in vegetation?
    6 years ago

Post a comment