Jump to ContentJump to Main Navigation
Ecological Niches and Geographic Distributions (MPB-49)$

A. Townsend Peterson, Jorge Soberón, Richard G. Pearson, Robert P. Anderson, Enrique Martínez-Meyer, Miguel Nakamura, and Miguel B. Araújo

Print publication date: 2011

Print ISBN-13: 9780691136868

Published to Princeton Scholarship Online: October 2017

DOI: 10.23943/princeton/9780691136868.001.0001

Show Summary Details
Page of

PRINTED FROM PRINCETON SCHOLARSHIP ONLINE (www.princeton.universitypressscholarship.com). (c) Copyright Princeton University Press, 2022. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in PRSO for personal use. Subscriber: null; date: 04 July 2022

Linking Niches with Evolutionary Processes

Linking Niches with Evolutionary Processes

(p.238) Chapter Fifteen Linking Niches with Evolutionary Processes
Ecological Niches and Geographic Distributions (MPB-49)

A. Townsend Peterson

Jorge Soberón

Richard G. Pearson

Robert P. Anderson

Enrique Martínez-Meyer

Miguel Nakamura

Miguel Bastos Araújo

Princeton University Press

Abstract and Keywords

This chapter examines how the process of ecological niche evolution and diversification helps us better understand ecology, biogeography, and biodiversity. It first considers how species respond to changes in the environmental substrate on which the niches are manifested before discussing the concept of niche conservatism as well as tests of conservatism in areas such as species invasions and comparison of the ecological niche requirements of sister–species pairs. It then explores how temporal change in niche dimensions occurs, how it can be studied, and what can be learned. It also describes some of the challenges associated with applications of ecological niche modeling in the realm of evolution and concludes by outlining future directions for research.

Keywords:   ecological niche evolution, diversification, ecology, biogeography, biodiversity, species invasions, niche conservatism, ecological niche modeling, sister–species pairs

As methods for modeling and understanding ecological niches and geographic distributions of species have become increasingly robust and well-understood, evolutionary biologists have begun to pay attention. That is because a critical dimension of the evolutionary biology of species is precisely their ecological requirements, as biogeography, distribution, and genetic variation all hinge rather critically on the ecological niche. Evolutionary studies of ecological niches have thus begun to appear in numbers, amplifying the diversity of challenges to which these techniques have been applied.

Changes in the Available Environment

Since the envelope of environmental space available to a species [i.e., environments represented within M or η‎(M)] changes through time, to avoid extinction a species must either track the geographic extent of its scenopoetic existing fundamental niche, or be able to change it via evolutionary responses in physiological or behavioral traits (Holt 1990). One of the important advantages of expressing Grinnellian niches as subsets of an E-space is that the issue of constancy or change of the environmental substrate on which the niches are manifested becomes apparent (Jackson and Overpeck 2000, Ackerly 2003). For example, Jackson and Overpeck (2000) showed changes in what they termed the “Realized Environmental Space,” our η‎(G) or E available in the study region, measured using two extreme temperatures in a G covering all of North America, from modern times back to 21,000 years before the present (figure 15.1). The actual, existing environmental combinations of a species’ niche will shift in spatial location and extent, and a species must track suitable conditions, adapt to suboptimal ones, or go extinct.

The way in which species respond to these challenges is varied. Ackerly (2003) pointed out that the “leading” and “trailing” range edges may pose contrasting (p.239)

Linking Niches with Evolutionary Processes

Figure 15.1. Illustration of the changing climate conditions in North America at four points in time over the past 21,000 years. The black points in all panels show current-day conditions, while the gray points indicate the conditions at the given earlier period.

Redrawn from Jackson and Overpeck (2000).

selective pressures during episodes of change. Imagine the retreat of glaciers in the Northern Hemisphere and the associated northward advance of vegetation. Along the northern edge of the range, populations of these plant species would encounter habitats with few competitors but novel environmental conditions; along the trailing edge, however, populations experience already-known environmental conditions and combinations of species already coadapted but with environmental conditions becoming unsuitable (Ackerly 2003, Brown et al. 2003). The selective pressures are bound to be different in these scenarios. An extremely important question is whether populations can adapt quickly to changes in E-space or whether they must geographically “track” sites with the right conditions. When several correlated environmental variables that are (p.240) indeed important for the species are considered, tracking them simultaneously may become impossible. Even without extinctions, when the scenopoetic fundamental niches of members of a community of species correlate in different ways with a suite of environmental variables, a change in climate may lead to wholesale rearrangements of species assemblies along gradients (Graham et al. 1996), as illustrated in figure 15.2.

The preceding ideas illustrate the importance of understanding how fast species can adapt to environmental changes in G, which can take place over time spans of a few thousand years or even much shorter, over centuries or even decades (Balanyá et al. 2006). This process is that of niche evolution, although niches can evolve for other reasons not related to adaptation (e.g., owing to genetic drift or linkage with other traits under selection). This task would appear to be relatively innocuous: physiological tolerances and habitat associations are clearly features of the evolved phenotype of organisms (Angilletta et al. 2002). However, the concept of niche evolution as a consequence of the evolution of the broader phenotype forms the basis for many key insights from ecological niche modeling, and indeed, since some sort of conservatism in niche features would be required to make possible most of the predictions treated in the last several chapters of this book (Peterson 2006c), the success of those applications appears to provide evidence for conservatism (Peterson et al. 1999). Niche conservatism has many implications (Wiens and Graham 2005), such as the feasibility of forecasting the geography of species’ invasions (Peterson 2003a) and effects of climate change on species’ occupied and potential distributional areas (Peterson et al. 2005b, Araújo and Rahbek 2006), and for understanding speciation processes (Wiens 2004).

Clearly, though, once we conceive of niches as evolving as part of the overall phenotype of the organism, considerable interest will focus on the circumstances under which niches have evolved. That is, if all niches were conserved strictly, then all of life would have the same ecological niche, which is clearly far from the case. Rather, ecological niches of species do evolve, and this diversification has been key in structuring life on Earth. Understanding the process of ecological niche evolution and diversification would thus offer key insights into ecology, biogeography, and biodiversity. In this chapter, we offer comments and ideas regarding how the notions of this book can be oriented toward addressing this challenge.

Niche Conservatism

A body of theoretical ecological work offers a framework in which to consider ecological niche evolution (Brown and Pavlovic 1992, Holt and Gaines 1992, (p.241)

Linking Niches with Evolutionary Processes

Figure 15.2. Illustration of how environmental change can affect species associations. At time 1, existing fundamental niches of species 1 and 2 overlap, so the two can potentially coexist within that intersection. However, at time 2, the existing fundamental niches of the two species do not overlap, so they would not be able to coexist.

Redrawn from Jackson and Overpeck (2000).

Kawecki and Stearns 1993, Kawecki 1995, Holt 1996a and 1996b, Holt and Gomulkiewicz 1996, Holt 2003). The essence of these arguments is captured by the following idea: populations outside of the biotically reduced niche (here, EP) of a species are “sink populations” (Pulliam 1988) that will eventually go extinct without immigration or adaptation (by definition, fitness w is (p.242) less than unity). Other areas outside of M will be uninhabited, but could be colonized by invasive populations (i.e., in GI).

However, adaptation may take place as well. Using a series of models of rather different scenarios, Holt and Gomulkiewicz (1996) have shown that the adaptive process is, generally speaking, slower than extinction dynamics. Therefore, adaptation seldom rescues sink populations from extinction—in which case, evolution mostly takes place within NF. These models predict that a combination of large initial populations, small degrees of maladaptation and limited immigration rates from source populations are required for populations to adapt to new niche conditions (Holt 2009, Sexton et al. 2009). The spatial structure of the selective pressures is also predicted as an important factor driving niche evolution (Bell and Gonzalez 2009).

Empirical testing of niche conservatism and broader contemplation of the concept has now begun to fill out the picture considerably: niches act as long-term stable, evolved constraints on species’ physiological tolerances and needs, as well as on their geographic distributions (Peterson et al. 1999, Martínez-Meyer et al. 2004a, Nogués-Bravo et al. 2008b). The evidence for conservatism in ecological niche characteristics comes from diverse studies: geographic variation in niche characteristics across species’ ranges (Peterson and Holt 2003); comparisons of native and invaded ranges of invasive species (Peterson 2003a); longitudinal (i.e., over time) comparisons of niche characteristics within species (Martínez-Meyer et al. 2004a, Martínez-Meyer and Peterson 2006, Waltari et al. 2007, Nogués-Bravo et al. 2008b); cross-phylogeny comparisons (Ackerly 2003, Martínez-Meyer et al. 2004b, Eaton et al. 2008); and transplant experiments showing that fitness is lower at the margins of distributions (Crozier 2004, Angert and Schemske 2005). These diverse perspectives on niche conservatism are reviewed later.

Obviously, despite their evolution, ecological niches are not wildly variable over evolutionary time periods. That is, physiologically challenging environmental realms such as the air and land have been invaded only a relatively few times (Gordon and Olson 1994, Padian and Chiappe 1998, Larson 1982). On the flip side of the coin, however, Eltonian and Grinnellian niches, and major morphological and physiological traits that determine, for example, trophic position or other fundamental ecological adaptations, have obviously changed on some scale: marine organisms invaded land at several points in evolutionary history, and terrestrial clades have even invaded back into marine environments (e.g., sea snakes and whales). In this sense, ecological niches are not especially static, and do evolve, just not frequently or wildly (Holt 2009).

Of course, in most ecological niche modeling studies, more subtle ecological changes are where the interest lies—that is, as with the broader panorama (p.243) of macroevolution versus microevolution (Wright 1982), what we now perceive as macroevolutionary changes occurred far in the past, deep in evolutionary history, yet through microevolutionary processes (Lande 1986). As such, microevolutionary changes occurring in the recent past are much more tangible, and certainly far more accessible to study and analysis, especially given the poor and uneven fossil record available for most groups. Microevolutionary ecological niche change, which we can define for the purposes of this discussion as relatively minor changes in ecological niche parameters that permit the occupation of new environmental situations and new distributional areas, are the focus of the remainder of this chapter.

Tests of Conservatism

Within Distributional Areas

The simplest and most frequently applied test of niche conservatism is that of testing whether niche characteristics are constant across species’ geographic distributions. This idea quite simply builds niche models based on one sector of the occupied distributional area of a species and tests whether those niche models are predictive with respect to distributions in other sectors (i.e., equivalent to spatially stratified validation; Peterson and Holt 2003). Although the first formal presentation of the approach as a test of conservatism (Peterson and Holt 2003) showed examples of nonconservatism, the vast bulk of the examples examined to date have shown conservatism—that is, different sectors of species’ distributions can reliably be predicted based on the remainder of the species’ distributions (Peterson 2001, Peterson 2005a), although counterexamples exist (Raxworthy et al. 2008). A note of caution is that these tests must consider carefully the range of environments over which the model is calibrated, to assure that it is representative of the areas onto which the model is projected (see the discussion of transferability and extrapolation in chapter 7), because if not, niche comparisons may not be valid (see chapters 7 and 9; Kambhampati and Peterson 2007, Peterson and Nakazawa 2008).

Species’ Invasions

Invasive species offer an additional level of complexity to testing ecological niche conservatism. Here, in addition to the element of spatial diversity provided by testing across different sectors of species’ geographic distributions, introduced ranges generally also present distinct biotic environments. In the context of the BAM diagram, if native- and introduced-range occupied niches EO are highly similar, then either AB (the Eltonian Noise Hypothesis) or at (p.244) least features of B correlate closely with the variables used to define A. In such situations, a model based only on scenopoetic variables can recover a meaningful and consistent biological signal about AB across very diverse regions (Peterson 2003a). Many such tests have been developed, and most with positive results—that is, native-range ecological characteristics generally have excellent predictive power regarding invaded-range distributions (Richardson and McMahon 1992, Martin 1996, Peterson and Vieglais 2001, Peterson et al. 2003a, Peterson and Robins 2003, Iguchi et al. 2004, Hinojosa-Díaz et al. 2005, Roura-Pascual et al. 2005, Thuiller et al. 2005b, Nyári et al. 2006, Zambrano et al. 2006, Benedict et al. 2007, Peterson et al. 2007d). Apparent exceptions to this predictive nature of species’ invasions are treated in chapter 13.

Chapter 13 presents invasive species applications of ecological niche modeling in greater detail. The coincidence of the spatial extent of predictions of GI with the areas actually invaded by the species is impressive. Results of these studies suggest that (1) niches are frequently conserved across species’ invasions in ecological time, and (2) biotic interactions do not shift dramatically among distributional areas to the extent that predictivity is negated. (We note, however, that the species that do invade may be a nonrandom selection from the overall pool of possible invaders, though this possibility will require creative exploration to resolve.) Some recent studies ostensibly documenting negative results in this realm are discussed in chapter 13 and in the following.

Single Lineages through Time

Perhaps the most direct tests of niche conservatism available are situations in which models can be developed and tested “before and after” some period of time in which change could be manifested. Here, the same dimensions of the ecological niche are estimated and compared in the same region, so many of the caveats of other tests are avoided. The drawback, however, is that opportunities for such tests are relatively rare, so only a few such studies have been developed to date.

Martínez-Meyer and Peterson (2006) developed so-called longitudinal tests for eight plant taxa discernable at least to genus from pollen samples across North America. Ecological niche models were developed for the present-day distributions of each taxon and tested using occurrence data from within 3000 years of the Last Glacial Maximum (21,000 years ago), and vice versa. Of the total of 16 reciprocal tests conducted, all models predicted the independent evaluation data from the other time period better than would be expected at random. In general (see figure 15.3), the cross-time predictions matched distributional expectations quite closely, suggesting that these plant species were tracking a highly conserved ecological niche closely over the past 21,000 (p.245) years. Other such “before and after” studies include an analysis of mammal distributions in the present and in the Pleistocene (Martínez-Meyer et al. 2004a), and detailed analysis of the extinction of Woolly Mammoths (Mammuthus primigenius) from Europe at the end of the Pleistocene (Nogués-Bravo et al. 2008b).

A particularly intriguing study was that which failed to predict the historical distribution of the Spotted Hyena (Crocuta crocuta; Varela et al. 2009). In that analysis, the current distribution of the species was accurately predicted, implying that the species is in equilibrium with its environment, and also that climatic variables used for modeling were adequate; however, projecting the niche model back to the Last Interglacial period (126,000 years ago) failed to predict the spatial distribution of records from across western Eurasia, where extensive fossil records are known for that period, suggesting that its current occupied distributional area represents a subset of its scenopoetic fundamental niche.

Nogués-Bravo (2009) reviewed published studies in which niche models were projected to past scenarios. An important point to make is that such tests of conservatism are unidirectional. In other words, one can find support for the hypothesis of conservatism by finding that present and past distributions are environmentally coincident, but failure to demonstrate such overlap does not provide support for the alternative hypothesis of no conservatism. This asymmetric nature of the test results because several ecological and methodological reasons can be invoked to explain nonoverlap between occupied niches in time (Peterson 2011).

Sister-Species Comparisons

The limitation of the longitudinal approach, of course, is that few situations lend themselves readily to such testing, particularly over longer periods of time, primarily due to the paucity of past occurrence records. As a consequence, Peterson et al. (1999) explored the possibility of building such tests over evolutionary time, comparing the ecological niche requirements of sister-species pairs, in essence asking the question of whether ecological niche characteristics had been conserved over a time period of twice the time since the advent of allopatric conditions for sets of previously contiguous conspecific populations. They assessed 37 sister-species pairs of birds, mammals, and butterflies that are distributed on either side of the Isthmus of Tehuantepec in southern Mexico—in all, 74 reciprocal comparisons, testing whether the occupied niche characteristics of one of the species (EO) were able to predict the geographic distribution (GO) of its sister-species better than random expectations. Parallel analyses not of sister-species, but rather of confamilial species for each of (p.246)

Linking Niches with Evolutionary Processes

Figure 15.3. Ecological niche models derived from relating Last Glacial Maximum (LGM) detections of pollen of eight tree species to general circulation model reconstructions of LGM climatic parameters. Shown are the LGM reconstructed distribution (potential distributional area GP; left-hand column) and the projection to present-day climate conditions (GP; right-hand column). Darker shading indicates greater estimated suitability for the species; independent occurrence data for model evaluation are shown for the present-day projections.

Adapted from Martínez-Meyer and Peterson (2006).

(p.247) the focal species, showed very low levels of predictive ability, suggesting that ecological niche conservatism is strong between sister-species, but breaks down over longer timescales. Reanalyses of these results (Warren et al. 2008) confirmed that the niches modeled for these species pairs indeed are generally more similar than would be expected at random, although they are basically never identical.

Tracing Niche Characteristics over Phylogeny

A still more general approach to the question of ecological niche conservatism is to extend analysis back over phylogenies more complex than simple 2-taxon statements (i.e., sister-species pairs). Later, we will treat some more complex questions that can be addressed within this framework. However, for the moment, we focus on the question of niche conservatism, and how it can be studied using phylogenies.

A study that illustrates such approaches is an analysis of ecological niche diversity across the Icteridae (figure 15.4; Eaton et al. 2008)—the American blackbirds—a diverse clade for which detailed phylogenies are increasingly available (Lanyon 1994, Johnson and Lanyon 1999, Lanyon and Omland 1999, Omland and Lanyon 2000, Price and Lanyon 2002 and 2004). In this analysis, ecological niches and potential distributions (GP) were estimated for each of the >100 species of blackbirds, centroids of species’ niches were calculated and compared in environmental space, and species’ occupied and potential distributional areas were characterized in geographic space. Results showed that ecological niches were dramatically differentiated only between relatively distantly related species (see figure 15.4B), but that convergent evolution can make distant relatives appear rather similar in environmental space (see figure 15.4D). Most interesting, perhaps, is the point that very close relatives are invariably only slightly differentiated in ecological niche characteristics. Other such studies, in which niche characteristics (expressed in binary format) are mapped onto phylogenetic trees, are beginning to appear in greater numbers (Prinzing et al. 2001, Graham et al. 2004b, Hoffmann 2005, Knouft et al. 2006), although these studies are divided in their conclusions regarding the (p.248) (p.249)

Linking Niches with Evolutionary ProcessesLinking Niches with Evolutionary Processes

Figure 15.4. Illustration of phylogenetic approaches to studying ecological niche divergence and nondivergence in New World blackbird lineages. This figure shows plots of all pairwise species comparisons within each of three blackbird lineages (vertical columns), comparing (A) genetic distance versus geographic distance, (B) genetic distance versus ecological niche distance, (C) geographic distance versus ecological niche distance, and (D) phylogenetic (patristic; Fitch branch lengths) distance versus ecological niche distance.

Redrawn from Eaton et al. (2008).

(p.250) generality of niche conservatism and must confront serious methodological challenges regarding how best to reconstruct ancestral character states of non-binary ecological characteristics. Warren et al. (2008) presented novel randomization tools that will prove useful in such studies.


The question of evolutionary conservatism of ecological niches of species is in many senses independent of the conceptual framework laid out in the introductory chapters of this book. That is, our conceptual framework is described as two associated spaces—geographic and environmental—at a single point in time. A common assumption that will certainly bear closer examination is that among-population variation in inherited niche characteristics is negligible. This chapter attempts to lay out the panorama of ecological niche change through time, which represents yet a third dimension to the question. We now address how temporal change in niche dimensions occurs, how it can be studied, and what can be learned.

The ecological niches of greatest interest initially in this chapter are clearly Grinnellian in nature (see chapter 2)—that is, at a first level, we are most likely to be intrigued by how species’ requirements in scenopoetic niche dimensions either change or remain static. As such, we have focused on coarse-resolution, largely climatic scenopoetic environmental dimensions, and have generally neglected the Eltonian, interactive dimensions that may depend critically on other species. Nonetheless, clearly, the evolution of interactions between species is also of potentially great interest—how have the diverse interactions among elements of biodiversity come to be? For example, how did Mallophaga (feather lice) colonize bird feathers, and how did that association become so obligatory over time? Such questions can—in time and with thought—be addressed within these frameworks as well.

Learning More about Ecological Niche Evolution

A recent review argued that the question of conservatism of ecological niche characteristics is not particularly interesting (since we know that niches evolve), but rather that the important and interesting questions regard how often, how much, and under what circumstances they evolve (Wiens and Graham 2005). We agree. That is not to say that niche conservatism is not important: if and only if ecological niches are relatively conserved in a particular lineage can (p.251) many of the predictive approaches outlined in this book be informative. If, on the other hand, the ecological niche of a species were to vary wildly through time and across space, then even the simple idea of predicting an occupied distributional area would not be likely to succeed, so the conservatism question is extremely relevant. As mentioned earlier, a contribution by Warren et al. (2008) presents a useful clarification of the null hypotheses being tested in a diversity of studies of ecological niche conservatism versus evolution, and may be able to reconcile the different points of view that have been presented in the literature. Nonetheless, any evolved character will tend to be more similar among close relatives than among distant relatives. Wiens and colleagues (Wiens 2004, Wiens and Graham 2005, Kozak and Wiens 2006) have correctly pointed out that while “conservatism” sounds like no change, ecological niche conservatism can be the agent of distributional constraint and therefore isolation, divergence, and even speciation.

More interesting, however, is the diversity of questions regarding evolutionary biology of ecological niches that can be addressed using phylogenetic frameworks. Given information regarding the present-day diversity of niche characteristics among species in a clade, understanding the ecology of ancestral forms becomes feasible by means of phylogenetic methods that allow reconstruction of ancestral character states (Cunningham et al. 1998, Martins 2000, Pagel et al. 2004), although these methods are not without uncertainties. In particular, reconstructions of continuous character states (e.g., preferences with respect to temperature and rainfall) have been complicated by assumptions of averaging of ancestral character states, and generally are quite imprecise (Garland et al. 1999).

Graham et al. (2004b) applied these approaches to understanding speciation in dendrobatid frogs in Ecuador. They used both maximum likelihood and least-squares approaches to estimate ancestral niche dimensions as maximum and minimum values for each climatic variable. The result was a detailed view of ecological niche (putatively EA, but probably between EA and EO) shifts between ancestors and present-day forms (figure 15.5), which permitted reconstruction of ancestral distributional areas (albeit based on present-day climate conditions). No other studies have taken this general approach to understanding ecological niche evolution, to our knowledge.

Another interesting line of inquiry is that of viewing ecological niche characteristics on a phylogenetic framework to detect lineages along which ecological niches have changed dramatically or have remained constant. An example is an analysis of niches across the Neotropical manakins (Pipridae) undertaken by estimating amounts of ecological niche change in comparison to branch lengths on an independent phylogenetic framework (Anciães and Peterson 2009). (p.252)

Linking Niches with Evolutionary Processes

Figure 15.5. Example analysis of geographic and environmental distributions of species at present and in the past. From Graham et al. (2004b), this figure shows analyses of the dendrobatid frog lineage that includes Epipedobates boulengeri, E. sp., E. tricolor, Colostethus machalilla, and E. anthonyi. The left-hand panel shows the modeled potential distributional area GP for E. boulengeri and for the ancestor of E. anthonyi, C. machalilla, E. tricolor, and E. sp., where gray shows the distribution of E. boulengeri and black shows the ancestor. (Areas with diagonal shading are areas of overlap between past and present potential distributional areas.) The right-hand panel shows the distribution of E. boulengeri and the ancestral taxon in a principal components manipulated E-space.

This study found that niches had been generally conservative, but identified a few lineages (see, e.g., Chiroxiphia boliviana in figure 15.6) in which niches have changed dramatically. Such studies have the potential to reconstruct patterns of ecological innovation and to permit insight into when and under what circumstances ecological characteristics do change, and many more are now being developed and published (Martínez-Meyer et al. 2004b, Eaton et al. 2008).

The recent development of paleoclimatic reconstructions and readily available digital data layers describing environmental variables in the past opens doors to new understandings of the geography of speciation and the paleogeography of species. In particular, once ecological niche conservatism has been tested and established in a particular lineage over a particular span of time, it becomes possible to reconstruct past potential distributional patterns, such as (p.253)

Linking Niches with Evolutionary Processes

Figure 15.6. Analysis of niche characteristics across the Neotropical manakins (Pipridae), depicted as branch lengths summarizing amounts of ecological niche change on an independent phylogenetic framework.

Adapted from Anciães and Peterson (2009).

(p.254) Pleistocene refugia or dispersal corridors (Waltari et al. 2007). For example, Peterson and Nyári (2007) used ecological niche models projected onto climatic information from the Last Glacial Maximum to reconstruct putative Pleistocene refugia for the Thrush-like Mourner (Schiffornis turdina) across the Neotropics. They tested the degree to which these patterns of connectivity and isolation correspond to present-day genetic differentiation and found a close correspondence—in other words, Pleistocene connectivity estimated using ecological niche modeling techniques proved very informative regarding present-day genetic breaks in this highly structured species. These hypotheses, which are independent of the usual suites of molecular data that have traditionally been applied to such questions (i.e., the field of phylogeography), can in turn be used as hypotheses to be tested with those other datasets (Carstens and Richards 2007, Knowles et al. 2007, Strasburg et al. 2007, Jakob et al. 2009).

In the last few years, paleodistributions generated with niche modeling techniques have been combined with phylogeographic methods to test hypotheses regarding patterns of population-genetic structure of different taxa (Carstens and Richards 2007, Knowles et al. 2007). In this new approach, niche models based on current distribution data are “hindcasted” to identify potential refugia and likely routes of dispersal, which are then contrasted with population-genetic signals of demographic processes (e.g., bottleneck effects; Jakob et al. 2007, Buckley et al. 2010). Combination of these two independent lines of evidence has proven useful to understanding the roles of geography and ecological constraints on species in responses to climate change (Cordellier and Pfenninger 2009). Although only in initial stages, this fusion of fields promises fruitful future insights (Hickerson et al. 2010).

Future Directions and Challenges

This realm of applications of ecological niche modeling is only beginning to open and be explored rigorously. Several important questions remain for analysis. A potentially fruitful realm concerns the question of how ecological niches interact with processes of evolution, geography, and environmental history to produce biological diversification, including how the available environmental space itself may change through time. An important area of inquiry not yet explored in detail is how dynamic ecological landscapes (i.e., changing available environmental spaces) may constrain the geographic and evolutionary possibilities of species (Jackson and Overpeck 2000). These questions would certainly rank among some of the most fundamental and fascinating in systematics and evolutionary biology, so exciting insights lie ahead.

(p.255) A second suite of “next steps” looks back to the question of ecological niche conservatism. Here, as we have reviewed earlier, a growing mass of evidence points to the generality of conservatism of ecological traits over short-to-moderate periods of evolutionary time (e.g., Peterson et al. 1999), particularly in the context of species’ invasions (Peterson 2003a). We point to exceptions to the conservatism “norm” as being some of the most intriguing and exciting situations encountered, and consider that more research is required to assert the generality of these results, and most interestingly, to explore under what circumstances ecological niche change does occur.

Finally, we offer the contrast between Grinnellian and Eltonian niches as an additional fertile field of exploration. Do scenopoetic dimensions of species’ ecological niches show greater evolutionary conservatism than linked, interactive dimensions? The existence of many long-term ecological associations that show impressive constancy over evolutionary time periods (DiMichele et al. 2004, McGill et al. 2005), coupled with growing evidence about the fast pace of local adaptations in Eltonian variables (Thompson 2005), makes these questions relevant and interesting.