No abstract

. Patterns of β‐diversity in a highly diverse tropical dry forest tree community are described; the contribution of environmental heterogeneity and distance to β‐diversity was assessed. Significant differences in elevation, insolation, slope and soil water holding capacity (p < 0.01), variables related to water availability, were found among 830 m × 100 m transects laid along contrasting slopes of a system of three parallel microbasins. A gradient in elevation and insolation was found within north‐facing transects, among 10 m × 10 m sites; south‐facing transects showed an elevation gradient while crest transects showed a gradient in water holding capacity. In total 119 species were registered, with 27 to 64 species per transect, and 4 to 16 species per site. A large β‐diversity was found among and within transects; two indices of β‐diversity consistently showed a higher β‐diversity within transects than among them. Among transects, 64% of the variance in species composition could be attributed to the environmental variables; an additional 22% to the spatial distribution of sites. Within transects, 42% of the deviance in β‐diversity values was explained by insolation, and 19% by distance. β‐diversity increased with distance and with difference in insolation among sites; north‐facing transects, those with most contrasting insolation conditions, had the steepest increase in β‐diversity with distance. Such increase was clearly associated with changes in species composition, not with changes in species richness.

Landscape level variability of structure and tree species diversity was analyzed in a tropical deciduous forest at Chamela, Mexico. Trees with DBH ≥5 cm were sampled in 21 0.24 ha plots (5.04 ha in total) distributed among six different morpho-pedological land units. Average density was 1,385 individuals ha-1, basal area 15.9 m2 ha-1, and canopy height 6.8 m. Trunks with DBH ≤14 cm accounted for 90% of the entire set. A total of 148 species, 102 genera, and 43 families were recorded. Seventy percent of all species were poorly represented (&lt; 10 individuals ha-1). A Principal Component Analysis (PCA) based on structure and diversity variables showed that plots from the same morpho-pedological land unit were not always located close to each other along the two first axes, but a further PCA based on dominant species clearly divided two groups of plots. Although canopy structure and tree species diversity varied continuously across the landscape, -diversity (evaluated through species similarity between plot pairs) and the identities of dominant species exhibited the clearest distinction. The dichotomy between granitic vs. non-granitic lithology was the condition most clearly related with a lower similarity in species composition and the strongest contrast in the dominant species group.