• biodiversity patterns;
  • elevational gradients;
  • hypotheses;
  • mammals;
  • mountains;
  • Philippines;
  • species richness;
  • tropical forest


1 It is widely accepted that tropical lowland rain forest holds the greatest diversity of organisms, and it is often implied that this general pattern is also true for virtually all individual higher-level taxa. Standardized elevational transect surveys of non-flying small mammals (Insectivora and Rodentia) on geologically old, species-rich islands in the Philippines consistently show maximum diversity and relative abundance in upper montane/lower mossy forest at 1500–2200 m, often exceeding lowland species richness and relative abundance by a factor of three or more.

2 On mountains where maximum elevation exceeds 2000 m, there is a decline in species richness above about 1500–2000 m, yielding a curvilinear pattern of species richness along the elevational gradient. The peak in species richness occurs at the area of transition from montane to mossy forest, which is also the point at which rainfall probably peaks. In parallel with species richness, relative abundance of small mammals in the Philippines also increases from the lowlands to 1500–2200 m, increasing by a factor from two to 10.

3 Twelve hypotheses concerning patterns of diversity along elevational gradients, plus the null hypothesis, are evaluated. The null hypothesis of no variation and the hypotheses that maximum diversity is in the lowlands, that diversity is highest in areas of least perturbation, and that diversity increases with increasing area, are rejected. There is weak or ambiguous support for the hypotheses that diversity is greatest in areas of community interdigitation, that diversity is highest in the area of highest productivity, that diversity is correlated with habitat complexity, and that diversity is correlated with habitat diversity. There is strongest support for the hypotheses that diversity is correlated with annual rainfall, with total abundance of individuals in the community, with food resource diversity, with areas of reduced competition from other organisms, and with areas characterized by high rates of speciation.

4 Causality is difficult to evaluate because many hypotheses make non-exclusive predictions, they probably represent non-independent aspects of causal factors (in other words, there is much interaction among the processes highlighted by the various hypotheses), and they represent the range from proximate to ultimate and from descriptive to causal. All the hypotheses probably represent phenomena that exist in nature, but few (or none) represent phenomena found in all taxa. The primary challenge in the future will not be simply to accept or reject individual hypotheses, but rather to determine the circumstances under which the various causal factors are most important, how they interact, and how they can be combined into a more comprehensive and general multi-factorial model.