Toward a theory for diversity gradients: the abundance–adaptation hypothesis