Figure 1: A. Location map, showing 12 vegetation sample sites. Shaded areas = water. B, The westernmost sample sites superimposed on USGS 30 arc-second DEM data for the Central Andes, as processed by the Cornell Andes Project.

Figure 2. A. Mean annual temperature vs. percentage entire-margined species for different databases: Bolivia (this study) (MAT = -3.83 + 0.379 *NoT, r2 = 94 %, s = 1.1 °C, F = 174). Bolivia + Peru: this is the database composed of the Bolivian sites from this study and 7 sites from Bolivian and Peru from Wilf (1997) (MAT = 1.73 +0.283 * NoT, r2 = 84 %, s = 1.8 °C, F = 94), CLAMP 3B, CLAMP 3A, New Zealand, and east Asia. B. Mean annual temperature vs. percentage entire-margined species for the Bolivian sites, the CLAMP 3B database, and a subset of the CLAMP 3B database consisting of sites with the same range of MAT and mean growing season precipitation as the Bolivian sites.

Figure 3. A. Mean annual temperature vs. mean annual temperature vector score (= 0.906 * axis 1 score + 0.372 * axis 2 score) for canonical correspondence analysis of the CLAMP 3B dataset. The subalpine and Bolivian sites run as passive samples, meaning that their leaf morphology scores were not used to constrain the ordination axes. B. Mean growing season precipitation vs. mean growing season precipitation vector score (= -0.066 * axis 1 score + 0.911 * axis 2 score) for canonical correspondence analysis of the CLAMP 3B dataset, with Bolivian sites run as passive samples. Recall that for the Bolivian sites, mean growing season precipitation = mean annual precipitation.

Figure 4. Natural log of mean annual precipitation vs. the mean natural log leaf area (MlnA) for different databases: Bolivia (lnMAP = 2.99 + 0.255 * MlnA, r2 = 48 %, s = 0.23, F = 11); CLAMP 3B (lnMAP = 1.78 + 0.484 *MlnA, r2 = 61 %, s = 0.5, F = 226); and Western Hemisphere + Africa (lnMAP = 0.76 + 0.549 *MlnA, r2 = 76 %, s = 0.36, F = 152).