In lieu of an abstract, here is a brief excerpt of the content:

CHAPTER EIGHT From Niches to Distributions Species’ potential geographic distributional areas GP often differ from their occupied distributional areas GO. In this chapter, we discuss the conceptual bases for this discrepancy, and summarize methodological approaches to addressing the consequent problems. First, we discuss the meaning of the potential distribution GP, and describe reasons why a niche model may not estimate it correctly. Next, we explore reasons why a species may not be at equilibrium with its potential distribution GP (or GA, see the following), but rather inhabits only some subset of areas suitable for it (see chapter 3; Araújo and Pearson 2005). In terms of the BAM diagram (see figure 3.1), nonequilibrium situations may arise for three reasons. The first is that the set M is frequently less than universal, meaning that the species will not be present in all suitable areas of the landscape. Second, M may change through time; here, dispersal limitation reduces GP to GO. Third, B may often differ from M; that is, in some situations, the Eltonian Noise Hypothesis does not hold, and negative interactions reduce GA to GP, or to GO if M is not limiting. (Bear in mind that this reasoning is limited to cases of negative effects of biotic interactions, and that the situation will differ if interactions are positive.) In some cases, both M and B can contribute to nonequilibrium distributions. Finally, we outline procedures for further processing of a niche model, which expresses GP or GA, to yield an estimate of GO. In this chapter, as in most of this book, we focus on the case of conditions when the Eltonian Noise Hypothesis is true, but attempt to note necessary modifications when it is not. POTENTIAL DISTRIBUTIONAL AREAS Estimates of species’ potential geographic distributions GP are critical to myriad niche modeling applications (chapters 10 to 15), and can also be used to estimate the species’ occupied distribution, although under certain assumptions (see the following; Anderson and Martínez-Meyer 2004, Peterson 2006c). GO is observable in nature, and is frequently of interest in theoretical and empirical FROM NICHES TO DISTRIBUTIONS 139 studies, particularly applications in conservation biology where it is crucial to know the actual distribution of the species. In contrast, GP is a theoretical construct (see chapter 3), and as such its full extent will probably remain unknowable in practice, although in theory it would be possible to perform broad suites of experiments to outline it. Now, remembering that that the environments of η(GO)  EO constitute the occupied niche space, it is clear that those environments may occur outside of GO. As we saw in chapter 3, a map of η–1(EO) may well be larger than GO, extending into the potential distributional area. Niche modeling exercises often strive to estimate GP (see chapter 3), but numerous factors may cause the predicted (modeled) potential distribution to be different from, and typically smaller than, the species’ true GP. In strategizing for model development, investigators should consider each of these factors and aim to eliminate, or at least minimize, as many of them as possible. First, the calibration data Gdata on which the model is based are drawn from GO, which may not encompass the full breadth of environmental conditions that the species can inhabit (and may encompass some conditions where it cannot persist, of course, if sink populations are included). Unfortunately, the species’ full biotically reduced niche EP  η(GO ∪ GI), the projection of which onto geography would constitute GP, is unknowable. GO (or a representative sample from it) will underrepresent EP and GP if: 1. The data Gdata used to estimate GO fail to include a representative sampling of the environments that the species uses. 2. The algorithm μ(Gdata, E) used to calculate GO overfits to Gdata. 3. The universe G is limited, such that M is limited with respect to η(GP). 4. The Eltonian Noise Hypothesis is not true, and some environmental conditions suitable for the species exist only in regions where biotic interactions do not permit the species to inhabit its full abiotically suitable distributional area GA. With correlative models (the main subject of this book), researchers must generally accept this reality (i.e., that niche models may not estimate GP completely), and deal with the fact that the resulting models may be conservative—in other words, that ÊP will generally be a subset of the true EP (contra Soberón and Peterson 2005), and as such will identify as suitable an area smaller than...


Additional Information

Related ISBN
MARC Record
Launched on MUSE
Open Access
Back To Top

This website uses cookies to ensure you get the best experience on our website. Without cookies your experience may not be seamless.