Evidence Of Temperature Change

The first evidence that Amazonia experienced substantially cooler-than-modern conditions during the last ice age came from the discovery of Podocarpus timbers in an exposure of silty peat near the town of Mera, Ecuador, at 1,100-m elevation (Liu and Colinvaux, 1985). Podocarpus spp. are generally found in cloud forests above 1,800 m, and this observation was used to draw the inference of a c. 800-m descent of Podocarpus populations at c. 30-36 kcal. yr bp. Indeed, because Podocarpus is not restricted to modern montane forests, Gentry (1993) lists four species that occur above 1,800 m and one that is found in the lowlands—the use of this genus as a paleoecological indicator of cooling has been criticized (van der Hammen and Hooghiemstra, 2000). However, further work on Mera and the site of San Juan Bosco that lies about 160 km to the south of Mera (Bush et al., 1990) revealed a suite of macro- and microfossils of additional taxa that are similarly most abundant in modern montane forests. Drimys, Alnus, Weinmannia, and Hedyosmum, were found to be abundantly represented in glacial age samples that also were depauperate in the taxa currently associated with this elevation—for example, Cecropia, Urticaceae/Moraceae, Iriartea.

This finding reinforced the probability that these sites supported cold-adapted elements at the peak of the last ice age. Every Pleistocene-aged pollen record recovered from Amazonia provides similar evidence of cool-tolerant populations moving into the lowland forests. In each case, floral elements became abundant 800 m to 1,500 m below the modern centers of their population. Similar patterns have been found in regions adjacent to the Amazon Basin—for example, southern and central Brazil (de Oliveira, 1992; Ledru, 1993; Salgado-Labouriau, 1997), the Andes (reviewed in Chapter 2), and Central America (Bartlett and Barghoorn, 1973; Bush and Colinvaux, 1990).

Quantifying this cooling has relied on translating the altitudinal descent of species into temperature change via the moist air adiabatic lapse rate, generally taken to be c. 5.5-6° C. Hence, the observed 800-m to 1,500-m descent of thermally sensitive populations translates into a 4°C to 7°C cooling. A similar estimate of cooling was developed from the isotopic analysis of groundwater in eastern Brazil. Stute et al. (1995) found the temperature of "fossil" groundwater to be c. 5°C cooler than that of groundwater formed under modern conditions.

Van der Hammen and Hooghiemstra (2000) and Wille et al. (2000) have argued for a flexible lapse rate during the ice ages that would significantly steepen the temperature gradient from the lowlands to the highlands. Their contention is that the Colombian Andes at 2,580 m cooled by c. 8°C while the lowlands cooled only between 2.5 and 6°C (according to the data source). If these values are taken at face value, one interpretation is that the moist air adiabatic lapse rate must have steepened. The implied lapse rate to accommodate the difference in montane versus lowland temperature increases for a modern rate of c. 6°C in Colombia (Wille et al., 2000) to unrealistically high values of between 6.7 and 8.1 °C. Such high lapse rates are unlikely as they imply very dry air, and—given that no mid-elevation Andean setting with that kind of aridity has been documented so far—a flexible lapse rate is not the solution to the observed variability in data.

The moist air adiabatic lapse rate is controlled by atmospheric humidity and is not seen to vary greatly from one tropical setting to another despite differences in precipitation and seasonality—that is, it is almost always 5.8±0.5°C per 1,000m of elevation. While narrow fluctuations can be expected through time, lapse rates are unlikely to vary beyond a constrained range (Rind and Peteet 1985).

On first principles it is difficult to envisage a very strong change in lapse rate in humid sections of the Andes, and yet the foothill regions consistently provide a slightly lower (typically 5°C change) temperature reconstruction than the highlands (typically 8°C change). Given that the LGM in Colombia was dry (Hooghiemstra and van der Hammen, 2004), while it was wet in Peru and Bolivia (Baker et al., 2001), we can assume that lapse rates may have risen close to c. 6.3°C in Colombia and been near modern (i.e., 5.5°C) in the central Andes. However, these changes are inadequate to describe the c. 3-6°C discrepancy between the lowlands and the uplands. We advocate taking a step back and considering other mechanisms than lapse rate change to account for the observed data.

We observe that paleovegetation response in mountains does not provide a pure temperature signal and is likely to be exacerbated by factors correlated with elevation.

If not parsed out, these factors can lead to an exaggerated paleotemperature change estimate. For example, changes in black body radiation due to lowered atmospheric C02 content (Bush and Silman, 2004 and Chapter 10) and feedback mechanisms involving ultraviolet radiation (Chapter 8) are more extreme with increasing elevation. With a thinner atmosphere—that is, LGM conditions of 170 ppm C02 and 350 ppb CH4 compared with pre-industrial Holocene concentrations of 280 ppm C02 and 650 ppb CH4—more heat is lost during nighttime re-radiation of stored heat than under modern conditions. This black body radiation effect would increase with elevation, be strongest under cloudless skies, and might add an apparent c. 2.3°C of cooling to the true change in temperature (Bush and Silman, 2004). As it is the coldest nighttime temperatures that a plant must survive, the black body radiation imposes an additive thermal stress that can cause mortality as physiological thresholds are exceeded. This mechanism provides one example; more probably exist, contributing to the differential migration distances in montane and lowland settings.

Although the seminal work on the High Plains of Bogota set a new path for Neotropical paleoecology, many other records now exist that suggest somewhat more modest temperature departures both in the lowlands and in the mountains than the reported 7-9°C. Thus, we suggest that the migration of tree line as inferred from the pollen records is accurate, but that there may be more mechanisms at work than simply temperature change contributing to those shifts. If estimates of glacial descent and ELA (Seltzer, 1990; Rodbell, 1992; Seltzer et al., 2003; Smith et al., 2005) are also considered, the actual LGM cooling at all elevations may have been c. 4-6° C, making it more consistent with data obtained from marine paleotemperature reconstructions (Ballantyne et al., 2005).

It is important to note that this degree of cooling was not temporally uniform throughout the last ice age and that there were other high-magnitude changes (comparable with the Pleistocene/Holocene transition) during this period. That all the lowland records contain gaps in sedimentation makes it difficult to put a firm timeline on when Amazonia was coldest. At the Hill of Six Lakes, montane taxa are clearly abundant in samples that are radiocarbon-infinite in age, and they have their peak occurrence between 21 kcal. yr bp and 18 kcal. yr bp (Bush et al., 2004a). Looking farther afield, Lake Titicaca and Lago Junin in the Peruvian Andes were probably coldest between c. 35 kcal. yr bp and 21 kcal. yr bp (Hansen et al. 1984; Seltzer et al., 2000; Smith et al., 2005). But, at the lowest elevation of modern cloud formation, the record from Lake Consuelo (1,360-m elevation) indicates a protracted, steady cooling of about 6°C between 40,000 and 22,000 calyr bp (Urrego et al., 2005).

In some records, particularly those in eastern Amazonia and coastal Brazil (de Oliveira, 1992; Ledru, 1993; Behling, 1996; Behling and Lichte, 1997; Haberle and Maslin, 1999; Ledru et al., 2001; Sifeddine et al., 2003), cold-tolerant taxa persist and in some cases reach their peak abundance as late as 14 kcal. yr bp, when western Amazonian and selected Andean records are showing considerable, steady, warming (Paduano et al., 2003; Bush et al., 2004b). Such short-term variability has yet to be explained, but it is a marked characteristic of these records that climatic events are neither synchronous nor basin-wide.

0 0

Post a comment