Whether the climate of tropical South America during the Last Glacial Maximum (LGM) was colder and drier or colder and wetter than present day has been widely debated. It is accepted, however, that the LGM in tropical South America was 2–9 °C colder than today (e.g., Betts and Ridgway, 1992; Bush et al., 2001). Without debating the merits of the following choices, if we assume a lapse rate in the LGM similar to the modern one of ~0.6 °C·100 m−1, then an intermediate cooling of 5 °C would lower the boundary between montane cloud forest and the overlying puna grasslands by ~800 or 900 m. Palynologists on both sides of the wet/dry debate have come to similar conclusions about forest-boundary lowering due to temperature decrease (reviewed by Flenley, 1998). In the Eastern Cordillera of Bolivia the modern puna–cloud forest boundary lies ~3400 m above sea level (masl). Ignoring any other environmental changes, LGM cooling would have lowered this boundary to 2500 or 2600 masl.
Mourguiart and Ledru (2003) presented an interesting pollen diagram from a late Quaternary sedimentary sequence from a peat bog near Siberia, Bolivia. The site (2920 masl) is located within the modern cloud forest. Based on cooling alone during the LGM, it is expected that their site would have been well above cloud forest, and therefore well within the puna vegetation zone. Mourguiart and Ledru (2003) indeed observed just such an expected change at their site: the full glacial had lower representation by cloud forest taxa (e.g., Podocarpus and Myrtaceae) and higher representation by puna taxa (e.g., Poaceae). Although not referenced by Mourguiart and Ledru (2003), it should be mentioned that two previously published pollen records from sites within 40 km of Siberia at 2700 masl (Cala Conto) and 2720 masl (Wasamayu) show continuous moist-forest taxa throughout the LGM (Graf, 1989, 1992). We were thus surprised that Mourguiart and Ledru (2003, p. 195) concluded that their pollen record indicates a “drastic decrease of the Amazonian moisture source,” rather than that the upper cloud forest boundary had simply migrated to elevations well below their site due to cooling. Indeed, we see nothing in the pollen or algal record to support inferences of wide-spread aridity.
To corroborate their interpretation from Siberia, Mourguiart and Ledru (2003) present data from a second coring site at <19 m water depth in Lago Huiñaimarca, a shallow arm of Lake Titicaca (3810 masl). From changes in the abundance of Isoetes, Pediastrum, and Botryococcus in this core, they conclude that Lago Huiñaimarca was shallower during the LGM than before or afterward. Although lake-level change may be one mechanism to account for the observed patterns in Lago Huiñaimarca, there are other possible explanations for these data (e.g., changes in temperature, nutrient availability, or water clarity; Jankovska and Komarek, 2000) that are fully consistent with the hypothesis of a cold and wet LGM, as suggested by other paleoecological studies from the Altiplano. Based on data from multiple proxies (diatoms, pollen, stable isotopes, inorganic and organic carbon) in many sediment cores that we recovered from multiple locations in the main part of Lake Titicaca as well as in Lago Huiñaimarca, we have shown that the main basin of Lake Titicaca was a deep, freshwater lake during the LGM and that it overflowed via its outlet on Lago Huiñaimarca (Baker et al., 2001a; Seltzer et al., 2002; Tapia et al., 2003; Paduano et al., 2003). In fact, the greatly enhanced discharge from the lake via the Río Desaguadero (Cross et al., 2001) contributed to the flooding of the central Altiplano and the formation of a large and deep paleolake that existed throughout the LGM from ca. 25,000 to 16,000 cal. yr B.P. (Baker et al., 2001b). Thus, Lago Huiñaimarca was filled to its present-day (shallow) depth at the LGM. Indeed, as long as the outlet of Lake Titicaca was at its present-day depth, it could hardly have been otherwise.
A parsimonious explanation for all available data is that the Altiplano was cold and wet during the LGM, not a dry environment as Mourguiart and Ledru (2003) concluded. Furthermore, given the clear pacing of wet-dry cycles at precessional frequencies in regional records and the absence of evidence for LGM aridity at the Siberia site, we do not feel that it is necessary to revise our well-supported conclusions (Baker et al., 2001a, 2001b) about the contributory causes of increased precipitation on the Altiplano during the LGM.