Elevational variation in the spatial configuration (α, β, and γ components) of multiple aspects of taxonomic biodiversity: The role of variation in productivity and abundance

Background/Question/Methods

Although taxonomic biodiversity is a multidimensional concept, most research at broad spatial scales has solely considered gradients of species richness.  In contrast, we quantified spatial variation in multiple metrics of taxonomic biodiversity along an extensive tropical elevational gradient.  We explored the extent to which biotic and abiotic characteristics mold patterns of taxonomic biodiversity (species richness, evenness, diversity, dominance, and rarity, transformed into Hill Numbers) for gastropods by quantifying hierarchical compartmentalization (α, β, and γ,) along two parallel transects, one passing through tabonuco, palo colorado, and elfin forest, and one passing through only palm forest.  If climatic factors primarily mold elevational variation in animal biodiversity, then patterns should be indistinguishable between transects.  If variation in plant community composition or physiognomy primarily mold elevational variation, patterns along the two transects should be different.  γ for each transect was partitioned using the multiplicative model and two spatial scales (γ_transect = α_plots × β_plots × β_elevations).  Significance of partitions was determined via simulations that randomized individuals separately within each elevation or that randomized individuals from an entire transect.  Orthogonal polynomial regression was used to capture elevational responses (linear and non-linear) of α, β, and γ components for each metric of biodiversity.

Results/Conclusions

Despite considerable differences in vegetative composition and structure along these two transects, contributions of scale – within plots (α_plots); among plots within elevations (β_plots); among elevations (β_elevations) – to total biodiversity (γ_transect) were similar for all metrics.  Nonetheless, the form and parameterization of elevational gradients were quite variable, including five patterns: (1) flat or random; (2) monotonic and linear; (3) monotonic and non-linear; (4) quadratic with a hump; and (5) quadratic with a trough.  Generally, the fit of data to the best model was greater along the mixed forest transect than along the palm forest transect.  Moreover, differences between transects in the form of gradients were most apparent for metrics that were insensitive to abundance (species richness and rarity) compared to those that weighted estimates of richness by abundances (species diversity, evenness, and dominance).  Unweighted measures at the α- and γ-scale generally had linear or random elevational relationships, whereas measures weighted by abundances consistently evinced modal patterns with mid-elevational peaks.  After controlling for variation in abundance among plots, transect-related differences in elevational relationships of hierarchical components of each metric were non-significant, suggesting a dominant role for environmental factors that affect spatial variation in productivity rather than those causing forest zonation.