Abstract

The timing and relative contributions of precipitation versus temperature changes is crucial to understanding the tropical hydrological cycle over the central Andes since the Last Glacial Maximum. Despite its location on the lee side of the Andes on the western margin of the tropical rainfall belt, the Rio Salado basin receives almost 90% of its scant precipitation during the summer months, depending on upper-air conditions that either favor or suppress transport of moist air masses from the Amazon and Gran Chaco Basins across the Altiplano onto the Pacific slope. Previous results from the central Atacama Desert indicate that century to millennial-scale variability of summer rainfall have been shown to produce dramatic elevational displacements in plant distributions. Plant macrofossils were analyzed in 35 rodent middens from the lower Rio Salado basin that span the last 22,000 calendar years (22 ka). Middens were collected at three different localities, two from between 2950-3050 m with almost exclusively riparian vegetation with sparse low shrubs, and one between 3100-3150 m with diverse dry slope plant communities. Based on the presence of steppe grasses (some of which are found several degrees further north today) and high species richness, we infer wet phases between 17.4-16.4, 11.2-9.6, 7.4-6.7, and 0.7-0.5 ka. Dry phases, characterized by low diversity and/or dominance of riparian species, occurred at 21.9, 14.2 and 4.8-2.3 ka. Summer precipitation variability inferred from Rio Salado middens generally agrees with other midden and lake records throughout the central Andes between 22-10 ka. Increased rainfall inferred from middens dated between 7.4-6.7 ka is in conflict with high Andean lake records which show evidence of severe drought beginning at 8 ka. Increased temperature at the end of the Pleistocene, however, was followed by colonization of the high Andes by considerably more productive plant communities. This may have had a major impact on lake hydrology budgets through increased evaporation, evapotranspiration and decreased surface runoff. Thus, part of the discrepancy between paleoclimate records in the central Andes during the early to middle Holocene could stem from confounding climate factors.