Forest Vegetation Sampling
We sampled the vegetation in the park between 1993 and 2000 with one-hectare permanent plots; study sites were established in a range of forest formations identified by field expeditions, over-flights, and LandSat images (Killeen et al 1998b). We sampled evergreen, semi-deciduous, and deciduous forest types, in upland and seasonally inundated landscapes, and in continuous forests and forest islands. Plots are grouped within the tables of this paper according to an a priori classification based on physiognomy and edaphic criteria (Table 1).
In order to minimize within plot heterogeneity, we installed the plots in visually homogeneous forest. The form of plots at most sites was a rectangle measuring 500 m x 20m, though at five sites, (CP-1, CP-2, CR-1, IH2-3, and IH2-4) the plot shape was altered to fit the 1-ha study area into forest patches that were not long enough for the 500 m design. We installed 1-ha plots at 19 different sites; to document within forest-type variability, we installed an additional plot at 10 of these sites, yielding a total of 29 one ha plots. At the locations where we installed two plots, these were placed parallel to each other 200 m apart within a 500 x 500 m study area, with the exception of CH-1 and CH-2 and CR-1 and CR-2 where plots are approximately 8 km apart. All freestanding plants greater than 10 cm dbh (1.4 m above ground level, or above buttresses) were registered with identification numbers using aluminum tags; diameter, height (triangulated using a clinometer), position within the plot, and species name were recorded. We excluded lianas but included palms in this portion of the study.
Voucher specimens are on deposit at the herbarium of the Museo de Historia Natural Noel Kempff Mercado in Santa Cruz, Bolivia (USZ). We made specimen determinations by comparison with specimens in the collection at USZ, at the Missouri Botanical Garden (MO), and with the assistance of many family specialists (see Killeen & Schulenberg (1998) for a list of specialists consulted). Primary data from the voucher collections, taxonomic authors, and synonyms are available at
Savanna Vegetation Sampling
We selected 9 savanna formations based on LandSat imagery and aerial recognizance (Table 1). The sites represent both seasonally inundated and upland formations and within each of these groups, we selected sites with differing physiognomic characteristics, from rocky grassland (campo rupestre) to tall woodland (cerradão) (Table 1). These plots are widely distributed around the park, with a maximum distance of 192 km between plots (Figure 1).
We used line intercept and point intercept techniques to sample the composition and structure of the vegetation. We marked a 500 m baseline at each site and randomly located between nine and fifteen 20 m lines perpendicular to this baseline. For the line intercept sampling, we recorded the horizontal coverage of each plant along these 20 m lines, yielding a total linear coverage of 180 to 300 m for each site. A botanical voucher collection was made the first time that a species was encountered in each plot. Voucher specimens are on deposit at the herbarium of the Museo de Historia Natural Noel Kempff Mercado in Santa Cruz, Bolivia (USZ). Specimen determinations were made by comparison with specimens in the collection at USZ, at the Missouri Botanical Garden (MO), and with the assistance of many family specialists (see Killeen et al (1998) for a list of specialists consulted.) Specimen label information is available at