Fewer species because of uncommon habitat? Testing the species pool hypothesis for low plant species richness in highly productive habitats

Abstract

Numerous studies have reported a significant pattern of decreasing plant species richness across regional habitat gradients of increasing productivity or community biomass (Grace 1999, Waide et al. 1999). The interpretation of this pattern has been the focus of several hypotheses (see reviews by Rosenzweig and Abramsky 1993, Abrams 1995, Grace 1999, Aarssen 2001) but none that have emerged with strong supporting evidence. In this article we propose a simple method for evaluating one of these hypotheses and illustrate its application using published data. The 'species pool' hypothesis was proposed by Taylor et al. (1990) in response to the 'competitive exclusion' hypothesis of Grime's (1973, 1979) 'hump-backed' model. Grime's model postulates that the commonly reported decline in plant species richness with increasing habitat productivity is explained by an increasing intensity of competition for resources, causing increasing competitive exclusion. The species pool hypothesis, however, explains this variation in plant species richness in terms of variation in the historical opportunity for origination of adapted species, i.e. species that can complete their life cycle and recruit offspring under the prevailing selection pressures, including competition, associated with a given habitat type. Because speciation requires space and time, we may expect that relatively few species have evolved with adaptation to the selection pressures of habitat types that have been uncommon over evolutionary time, i.e. habitat types that are relatively rare (in terms of spatial extent) or relatively young (with a short geological history). Such habitat types may be expected, therefore, to have relatively few contemporary resident species (see also Ricklefs and Schluter 1993). Accordingly, the species pool hypothesis predicts that resident plant species richness in highly productive habitats is relatively low because these habitat types are relatively young in geological age and/or small in terms of historical land area on a global scale (Taylor et al. 1990, Aarssen 2001). The central question from this hypothesis remains unanswered: Are habitat types with relatively high productivity not only relatively low in species richness but also relatively uncommon? The principal determinant of variation in potential vegetation productivity between habitats is the resource-supplying power (e.g. substrate fertility) of the habitat (Taylor et al. 1990). The species pool hypothesis predicts that the frequency distribution of historical land area of different habitat fertility or productivity types should approximate some kind of unimodal central tendency; i.e., with the most common habitat productivity type over evolutionary time being of some intermediate level and with the most extreme (both low and high) productivity types being the least common. Given that this distribution has no definable upper limit and is strictly bounded only at the left end (i.e. habitats cannot have less than zero productivity), the predicted frequency distribution of habitat productivity types should be right-skewed unimodal, analogous to the 'left wall' effect proposed for the evolution of organismal complexity (Gould 1988). On this premise, the species pool hypothesis would predict that past opportunity for origination of adapted species and hence, contemporary resident species richness should also be generally highest in habitats with somewhat less than intermediate productivity (compared with extremely low or high productivity) (Taylor et al. 1990; see also Rosenzweig and Abramsky 1993, Abrams 1995, VanderMeulen et al. 2001). The latter is indeed the case for contemporary plant species richness patterns recorded in numerous studies reporting a unimodal or right-skewed unimodal 'hump-shaped' relationship (Grace 1999).