Although the mid-domain effect (MDE) model for species richness in bounded geographical domains has proved controversial, several studies have revealed its explanatory potential for patterns of species richness. This paper investigates unexplored aspects of one-dimensional MDE, based on a data set of latitudinal distributions of New World birds (3706 species) on a 1° scale. Two previously published data sets for other taxa are also considered. We adjusted band sums (number of species per latitudinal band) for longitudinal area by constructing species-area curves for each band. Area-corrected richness patterns differed substantially from raw band sums, although both confirmed a strong, mid-tropical peak in richness. An MDE model accounted for 47% of the adjusted pattern, whereas area alone explained 13% of variation. Area-adjusted band sum data proved preferable to coastal transect data from the same data set. MDE was relatively more important in smaller latitudinal domains and/or for taxa with relatively large ranges. On fundamental grounds, we concluded that MDE randomizations based on empirical ranges are more appropriate than those based on theoretical range size distributions. Models that, like MDE, produce quantitative richness predictions should be evaluated statistically against the null hypothesis of equality, not simply correlation, of empirical vs. predicted richness values.
Keywords: band sum area, birds, geographical domain extent, latitudinal gradient, Madagascar, MDE model, mid-domain effect, New World, randomizations, range size distribution, species–area curves, species richness