Quaternary Science Reviews 20 (2001) 1289}1308 Mid- to Late-Holocene pollen-based biome reconstructions for Colombia Robert Marchant*, Hermann Behling, Juan Carlos Berrio, Antoine Cleef, Joost Duivenvoorden, Henry Hooghiemstra, Peter Kuhry, Bert Melief, Bas Van Geel, Thomas Van der Hammen, Guido Van Reenen, Michael Wille Hugo de Vries-Laboratory, Faculty of Biology, Institute of Biodiversity and Ecosystem Dynamics, University of Amsterdam, Postbus 94062, 1090 GB Amsterdam, Netherlands Centre for Tropical Marine Ecology, Fahrenheitstrasse. 1, D-28359 Bremen, Germany Arctic Centre, University of Lapland, P.O. Box 122, FIN-96101 Rovaniemi, Finland Fundacion Tropenbos Colombia, Carrera 21 39-35, Santafe de Bogota & , Colombia Abstract The assignment of Colombian pollen data to biomes allows the data to be synthesised at 10 &time windows' from the present-day to 6000radiocarbonyearsbeforepresent(BP).Themodernreconstructedbiomesarecomparedtoamapofmodernpotentialvegetation to check the applicability of the method and the a priori assignment of pollen taxa to plant functional types and ultimately biomes. The reconstructed modern biomes are successful in describing the composition and distribution of modern vegetation. In particular, altitudinal variations in vegetation within the northern Andean Cordilleras are well described. At 6000 BP the biomes are mainly characteristic of warmer environmental conditions relative to those of the present-day. This trend continues until between 4000 and 3000 BP when there is a shift to more mesic vegetation that is thought to equate to an increase in precipitation levels. The period between2500and1000BPrepresentslittleornochangeinbiomeassignmentandisinterpretedasaperiodofenvironmentalstability. The in#uence attributed to human-induced impact on the vegetation is recorded from 5000 BP, but is particularly important from 2000 BP. The extent of this impact increases over the Late-Holocene period, and is recorded at increasingly high altitudes. Despite these changes, a number of sites do not change their biome assignment throughout the analysis. This asynchronous vegetation response is discussed within the context of site location, non-linear response of vegetation to Late-Holocene environmental change, regionally di!erential signals, localised human impact and methodological artefacts. 2001 Elsevier Science Ltd. All rights reserved. 1. Introduction Biome reconstructions from pollen data have been produced for most regions of the world at a continental (Prentice et al., 1996; Jolly et al., 1998b) or sub-continen- tal scale (Williams et al., 1998). These spatially extensive reconstructions have demonstrated the ability of the biomisation technique to provide an objective recon- struction of vegetation using pollen data derived from multiple sites. In addition to these large-scale reconstruc- tions, a focus on the Arctic region (Edwards et al., 2000), where the pollen data are of su$cient quality and quantity, has enabled a regional reconstruction that of- * Corresponding author. E-mail address: marchant@bio.uva.nl (R. Marchant). fers greater ecological de"nition to the reconstructed vegetation than available at the scale of eastern North America (Williams et al., 1998). The biome reconstruc- tion for Latin America (Marchant et al., in review) demonstrated the variable spatial distribution of pollen- based records in this region, with sites particularly being concentrated along the Andean spine. One country with- in Latin America that has an extensive coverage of fossil pollendataisColombia(Fig.1),andmuchofthesepollen data are of a high quality with a number of localities having multiple radiocarbon-dated cores. However, in keeping with much of tropical palaeoecology, the initial research was focused on the abundant sedimentary basins located at relatively high altitudes (Table 1) where the sites are relatively accessible and the climate temper- ate. Furthermore, the sedimentary basins within lowland situations are mostly associated with riverine migration. 0277-3791/01/$-see front matter 2001 Elsevier Science Ltd. All rights reserved. PII:S0277-3791(00)00182-7