range_partition() splits the variation in observed richness along a
gradient into the part explained by the mid-domain null model
(geometry), the part explained by environmental variables, the part
they share, and the unexplained rest. It is variation partitioning
(Borcard et al. 1992; Peres-Neto et al. 2006) with adjusted R², applied
to the debate on whether geometric constraints or the environment
drive richness gradients (Colwell et al. 2004; Dunn et al. 2007).
Unique fractions are tested with an F test of the nested linear models. With few sites (as on most single-mountain transects) tests have little power and adjusted fractions can be negative (read negative values as zero); report them as descriptive. Sites along a gradient are also spatially autocorrelated, which makes the p-values too small.
Arguments
- mde
A
range_mde()result.- env
A formula of environmental predictors, e.g.
~ temperature, evaluated inmdejoined withdata.- data
Optional site-level data frame with the environmental variables and the site (and grouping) columns of
mde.
Value
A tibble with the grouping columns and, for each fraction
(geometry only, environment only, shared, residual), adj_r2
and, for the unique fractions, p_value.
References
Borcard, D., Legendre, P. and Drapeau, P. (1992). Partialling out the spatial component of ecological variation. Ecology, 73, 1045-1055.
Colwell, R. K., Rahbek, C. and Gotelli, N. J. (2004). The mid-domain effect and species richness patterns: what have we learned so far? American Naturalist, 163, E1-E23. doi:10.1086/382056
Dunn, R. R., McCain, C. M. and Sanders, N. J. (2007). When does diversity fit null model predictions? Scale and range size mediate the mid-domain effect. Global Ecology and Biogeography, 16, 305-312.