The “veiled gradients” problem in ecology

Abstract

Apparent regular variation in the distributions of organisms along geographical gradients has always fascinated ecologists. Variation in species richness with latitude has received the most attention, of course, but variation in species richness with elevation also has been of considerable interest and study. Precisely how and why species richness varies with elevation remains controversial (Wolda 1987, McCoy 1990, Colwell and Hurtt 1994, Rahbek 1995, 1997, Fleishman et al. 1998). Part of the reason for the continuing controversy is the fact that species richness is only an index, a manageable substitute for the set of distributions of individual species along an elevational gradient. Any explanation proposed for apparent regular variation in species richness with elevation can be, at best, only a loose reflection of the complicated autecological relationships between the individual species and elevation. The more species the proposed explanation is able to accommodate, the closer it will come to explaining variation in species richness, but any single explanation may not be suitable for a wide range of species (Lawton 1996). Another part of the reason for the continuing controversy about how and why species richness varies with elevation involves sampling problems (McCoy 1990, Rahbek 1995). Simply comparing species richness from one location to another can involve many sampling problems (Gotelli and Colwell 2001), but when such comparisons are made along gradients, the problems multiply. Understanding the causes of variation in species richness with elevation may have been hampered, for example, because different studies typically do not encompass the same elevational range and because the elevation of true maximum species richness sometimes occurs outside the sampled range (McCoy 1990). Ultimately, sampling problems even could lead to erroneous views of how species richness varies along elevational gradients (Rahbek 1995). Sampling problems such as those just described for studies of variation in species richness along elevational gradients are instances of a general problem in ecology that I term gradients. Veiled gradients arise from the fact that the outcome of any examination of a local gradient elevation, in this case is contingent upon the temporal and spatial frameworks employed. Over short time periods or across small geographical scales, only a portion of the gradient is likely to be revealed, the remainder of the gradient being veiled in time or space. Failure to realize that only a portion of the gradient is likely to be revealed in studies that are temporally or spatially restricted could obscure any true relationship that might exist (Allen and Starr 1982, McCoy 1990, McCoy and Bell 1991, Korner 2000). A ready example of how veiled gradients can influence interpretation of the variation in species richness with elevation can be found in the recent literature. Fleishman et al. (2000) determined that butterfly species richness increased with elevation in one of the canyons they studied but decreased in the other (Fig. 1). They attributed the apparent difference in the distribution of species richness with elevation to differences in climatic severity between the two canyons. The elevational ranges that were sampled were not the same in the two canyons, however, raising the possibility of a sampling problem. Samples (n = 102) from the first canyon (Toiyabe Range) spanned an elevational range of 1917-3272 m, whereas samples (n = 49) from the second canyon (Toquima Range) spanned an elevational range of 1872-2750 m. Ideally, we would like to unveil the distribution of species richness with elevation by extending the gradient in the Toquima Range to 3272 m, but we cannot do so with the available data. We can find out what the distribution of species richness with elevation would look like if the gradient in the Toiyabe Range were veiled to 2750 m, however. Examining changes in richness with elevation when the gradients are of the same length dramatically alters the interpretation of the relationship (Fig. 1). The apparent decline in species richness with elevation in the Toiyabe Range essentially disappears. Furthermore, an alternative interpretation of the data now suggests itself: that