Figure 1. Study site and design: A) habitat photo of the sampling plot; B) Map of Sabah with elevation gradient presented; C) The top and side view of the transect established within the forest plot (modified from Dial et al. 2004a and Dial et al. 2006).

Figure 2. Heatmaps along with scatter plot showing the horizontal and vertical distribution at 5 m vertical interval of (A) air temperature, (B) relative humidity, (C) photosynthetic photon flux density (PPFD), and (D) total leaf area. Color with higher saturation indicate higher value of the environmental variables.

Figure 3. Heatmaps along with scatter plot showing vertical and horizontal distribution of (A) species abundance (ln( x+1) transformed)) and (B) species richness. Color with higher saturation indicate higher value.

Figure 4. Linear relationships between pairwise distance and abundance based pairwise dissimilarity of ant assemblage when using A) maixmum surface travel distance and B) straight line distance between sampling points. The pairwise dissimilarity (green points) is partitioned by nestedness (yellow points) and turnover (blue points). Grey bands show 95% confidence interval. Solid line indicates significant relationship. Dissimilarity close to one indicates that two assemblages are dissimilar, while dissimilarity close to zero indicates that they are similar.

Figure 5. Effects of horizontal (circle points) and vertical distance (triangle points) on abundance based pairwise dissimilarity of ant assemblage between transects and vertical strata. Pairwise dissimilarity (green points) is partitioned by nestedness (yellow points) and turnover (blue points). Grey bands show 95% confidence interval. Solid line indicates significant relationship. Dissimilarity close to one indicates that two assemblages are dissimilar, while dissimilarity close to zero indicates that they are similar.

Figure 6. A) Relationship between vertical distance and pairwise dissimilarity of ant assemblages within each transect. Different colors of dot point and trend line indicate different transect. B) Relationship between horizontal distance and pairwise dissimilarity of ant assemblages within each vertical strata. Darker points and trend line indicate higher vertical stratum. C) Boxplot showing the median and interquartile ranges (IQR) of the estimated very small scale beta diversity (model intercept) and distance effects (model coefficient) on pairwise dissimilarity between sampling points within transect/stratum. Model intercept and coefficients are from the linear regressions between distance and assemblage dissimilarity within each horizontal transect and vertical stratum (see Fig. S3 and Fig. S4). Whiskers incorporate data that are 1.5 * IQR. Data beyond this are plotted individually.