To our minds, the last decade has seen at least two highly significant broad theoretical developments that address the core principles of ecology. The first of these has been the theory of metabolic scaling developed by G. B. West, J. H. Brown, B. J. Enquist and their colleagues. Based on the premise that metabolic rate is the most fundamental biological rate process, a large and growing body of work has sought to explore how, through geometrical constraints on exchange surfaces and distribution networks, relationships arise between body size and metabolic rate (West, Brown & Enquist 1997, 1999a,b; Gillooly et al. 2001; West, Woodruff & Brown 2002), developmental time (Gillooly et al. 2002), population growth rate (Savage et al. 2004), abundance and biomass (Enquist & Niklas 2001), production and population energy use (Ernest et al. 2003), and species diversity (Allen, Brown & Gillooly 2002). Whilst some of this work remains contentious (e.g. Dodds, Rothman & Weitz 2001; Kozłowski, Konarzewski & Gawelczyk 2003; Ricklefs 2003a; Storch 2003), the links to many of the topics routinely addressed by research papers published in Functional Ecology are obvious, and led to a recent special forum in the journal (Volume 18, Number 2) being dedicated to the topic. Indeed, much of the metabolic scaling research and numerous papers in the journal fall comfortably within the field of macrophysiology, which has been defined as ‘the investigation of variation in physiological traits over large geographic and temporal scales and the ecological implications of this variation’ (Chown, Gaston & Robinson 2004). The second significant broad theoretical development in ecology has been the neutral theory of biodiversity and biogeography proposed and championed by S.P. Hubbell (1979, 1997, 2001). This builds on the foundations of the classical theory of island biogeography, based on an assumption of demographic neutrality of individuals, to explain patterns of species richness, abundance and distribution. The connections between this work and the research typically published in Functional Ecology may perhaps be less apparent to many of the journal's readers. However, we suggest that these connections may, in practice, be equally strong as those to the theory of metabolic scaling. Indeed, there may also be much that can be gained from a joint consideration of the two theoretical developments, especially the interaction between those processes responsible for large-scale spatial variation in the numbers of individuals in a community or metacommunity (Allen et al. 2002), and those that produce the characteristics of many local communities that are a function of dispersal, speciation rate and numbers of individuals (the neutral theory) (see also Hubbell & Lake 2003). For these reasons, the present issue of the journal contains a second special forum, this time devoted to the neutral theory of biodiversity and biogeography. In this editorial, we briefly explore some of the reasons that the neutral theory is likely to be particularly relevant to functional ecology.
The neutral theory of biodiversity and biogeography provides an elegant unified theory of ecology and evolution. Although the mathematical developments are increasingly sophisticated (e.g. Vallade & Houchmandzadeh 2003; Volkov et al. 2003; Etienne & Olff 2004a; McKane, Alonso & Solé 2004), at its heart the theory is rather simple. It is the ecological analogue of genetic drift (Kimura 1983), has roots in a venerable history of the application of neutral models in ecology, and is one of a family of such models, albeit perhaps the most popular to date (Chave 2004).
The neutral theory concerns groups of trophically similar species typically occurring in sympatry and potentially competing for similar resources. The diversity of the assemblage results from stochastic ecological and evolutionary processes acting on both local and regional scales. The local communities within a region form a metacommunity, the evolutionary biogeographical unit within which member species originate, live and become extinct (Hubbell & Lake 2003), whose diversity is governed by a neutral speciation–extinction process. The metacommunity comprises a pool of individuals, from which local communities are assembled at random. Ecological processes operate at the scale of a local community, which experiences immigration, births and deaths. Species differentiate in relative abundance, and are lost to extinction, through stochastic processes or ecological drift.
There are two fundamental assumptions shaping the way in which the theory works. First, all individuals belonging to all species are equivalent with regards to the probabilities of birth, death, dispersal and speciation. That is, individuals exhibit no traits associated with their species identity that have any influence over their reproductive success, longevity, movements or likelihood of speciation. The absolute numbers of births, deaths and dispersal events will, however, vary between species, which will vary in abundance and distribution. Second, the community is saturated with individuals, such that if one dies its place is taken by another, and individuals are thus engaged in a zero-sum game. As a consequence, competition in the model is intense. Again, individuals exhibit no traits associated with their species identity that have any influence over the contribution they make to community saturation.
From the neutral theory, predictions can be derived about many basic ecological patterns (Hubbell 2001; Bell 2003). These include, for example, the form of species–abundance distributions, patterns of species richness, species–area relationships, and turnover in species composition with distance. It also makes evolutionary predictions, including ones related to speciation rates, times to extinction, age distributions of species, and relationships between abundance and the likelihood of speciation.
As with features of the theory of metabolic scaling, the neutral theory of biodiversity and biogeography has proven controversial (for commentaries see Abrams 2001; Brown 2001; Whitfield 2002; Harte 2003; Chave 2004). The principal criticisms concern, on the one hand, how realistic are the assumptions of the theory (for discussion see, e.g. Zhang & Lin 1997; Yu, Terborgh & Potts 1998; Hubbell 2001; Enquist, Sanderson, & Weiser 2002; Chase & Leibold 2003; Hubbell & Lake 2003; Ricklefs 2003b; Chave 2004; Poulin 2004) and, on the other hand, how well the theory describes observed properties and patterns in assemblage and community structure (for discussion see, e.g. Bell 2000; Hubbell 2001, 2003; Condit et al. 2002; Clark & McLachlan 2003, 2004; Fargione, Brown & Tilman 2003; Hubbell & Lake 2003; Lande, Engen & Sæther 2003; McGill 2003; Ricklefs 2003b; Tuomisto, Ruokolainen & Yli-Haila 2003; Volkov et al. 2003, 2004; Chave 2004; Gilbert & Lechowicz 2004; Olszewski & Erwin 2004). Plainly, taken literally, some of the assumptions are strictly invalid, such as the assumption of equivalence of all individuals (Hubbell 2001; Chave 2004). Plainly also, the fit of theory and observation is often debateable. The fundamental questions remain, however, as to the extent to which the neutral theory captures the essence of the structuring of assemblages and communities, and the extent to which it captures the mechanism behind this structuring.
Bell (2001) has distinguished two interpretations of the neutral theory: ‘weak’ and ‘strong’. The weak version of the neutral theory can generate patterns that resemble those observed in natural systems, but there is no recognition that it correctly identifies the mechanisms responsible for those patterns. The theory essentially serves as a null hypothesis, which could itself shape the way in which analyses of assemblage and community structure are conducted, although the weak interpretation might equally be regarded as a gentler way of saying that the theory is irrelevant. The strong version of the neutral theory regards any match between prediction and observation as evidence that the theory captures the mechanism driving assemblage or community structure.
From the perspective of functional ecology, the most important thing about the neutral theory is what it says about the niches of species. Species-level traits such as physiological tolerances, habitat preferences, energy usages, growth patterns, reproductive strategies, dispersal abilities and body sizes are all ignored. If the structures of assemblages and communities were well predicted by the neutral theory, then the strong interpretation would suggest that variations in the niches of species are relatively unimportant in these regards (some would say that it means that variations in niches are irrelevant, but this presumes that the predictions of neutral theory are perfect). Indeed, trophically similar species will be rather unspecialised, in as much as they can potentially exploit the resources freed by the death of an individual of any other species in the community.
Such a position is, of course, entirely contrary to that inherent in the vast majority of papers in the field of functional ecology. These are typically intimately concerned with variation between species in their physiological and life-history attributes (reviews in Bartholomew 1987; Spicer & Gaston 1999; McNab 2002; Chown & Nicolson 2004), what shapes the niches of individual species, and the consequences of the form that they take for where and when those species occur, how they interact with other species (as competitors, resource-providers, predators, etc.; Tokeshi 1999), and whether or not species can be considered functionally equivalent (reviews in Loreau, Naeem & Inchausti 2002; Naeem & Wright 2003).
There are three principal ways in which these two positions might be reconciled.
(i) Neutral theory is wrong. First, the neutral theory of biodiversity and biogeography might simply be wrong, in the sense that it fails to account for the patterns observed in nature (rather than that the mathematics is flawed). Whilst apparently close matches between the assemblage and community patterns observed and those predicted by neutral theory are suggestive that the processes embodied in the theory are shaping the observed patterns, this need not be so. The failure to demonstrate such matches implies that a theory is flawed, but the converse is not necessarily true. Certainly a number of other models predict patterns of assemblage and community structure very similar to those predicted by the original neutral theory (e.g. Chave, Muller-Landau & Levin 2002; Mouquet & Loreau 2003), although the interpretation of some of these alternative models has been a topic of discussion (e.g. Hubbell & Lake 2003).
Niche theory can explain many of the patterns of assemblage and community structure that are observed (e.g. Tokeshi & Schmid 2002; Brose et al. 2003; Cattin et al. 2003). However, there is as yet no unified theory of biodiversity that is broadly equivalent to the neutral theory but rooted in niche theory. Although some degree of unification has been achieved (Chase & Leibold 2003), it is difficult at present to see this being much more substantially realised in a highly generalisable form.
Whilst the neutral theory of biodiversity and biogeography has been the topic of much discussion and debate, the evidence in support of the theory remains rather sparse. Some of the predictions have been shown to be consistent with observed patterns of assemblage structure, but even here matters have not always proven clear cut. The debate surrounding the form taken by the species–abundance distribution for an assemblage of trees on a plot on Barro Colorado Island is salutary. McGill (2003) claimed that the lognormal distribution was a better descriptor than the zero-sum multinomial (ZSM) predicted by the neutral theory. Volkov et al. (2003) produced a simpler algorithm for the ZSM, and concluded that this gave a better fit than the lognormal. Etienne & Olff (2004a), using a new method for the ZSM, found once again that the lognormal did better, albeit only weakly so. The only safe conclusion would seem to be that the lognormal and the ZSM fit tropical forest tree (and possibly other) data about equally well (Williamson & Gaston, in press). Given the numbers of parameters involved, a more robust test of the neutral theory would seem to be, having fitted the model to abundance or other such data, to verify that the parameters estimated are realistic.
(ii) Niche theory is wrong. Equally, it could be that niche theory is wrong, again in the sense that the variation in niches really is not that important in structuring assemblages and communities. Most ecologists have been educated to regard niches as central to such structuring, and it is difficult to comprehend that this may not be so; ‘How can we have a neutral theory of biodiversity that not only ignores, but seems to contradict, the single most pervasive feature of life: the incredible variety of size, form, and function?’ (Brown 2001). However, the development of the field of macroecology, in particular, has revealed the great importance of regional scale processes (which may involve niche relationships to varying degrees) in shaping local assemblages and communities, and the difficulty of explaining the structure of local assemblages and communities based solely on processes that operate at the local scale (Ricklefs 1987, 2004; Brown 1995; Lawton 1999; Gaston & Blackburn 2000). The influence of niche variation in structuring assemblages and communities may thus, in principle, be much weaker than is widely supposed, with much of the diversity of size, form and function itself in some sense representing a kind of neutral variation.
(iii) Neutral theory only applies under some circumstances. The roots of the neutral theory lie in attempts to explain the diversity of tree species in tropical forests, where large numbers of species occur in close sympatry (Hubbell 1979; Chave 2004). A number of constraints of the theory which may make it appropriate in such circumstances arguably limit its wider applicability. This raises the possibility, scarcely unusual in ecology, that there is some truth both in neutral and in niche theory. These constraints on the neutral theory, as presently formulated (it may prove possible to generalise the model further in the future) include, first, that the neutral theory is foremost concerned with largely sessile (or space limited) organisms such as trees and brachiopods (e.g. Hubbell 2001; Condit et al. 2002; Olszewski & Erwin 2004). Whilst the outcomes of the tests of neutral theory using these organisms are contentious (see above), those investigating more mobile organisms tend not to support it (e.g. Engen et al. 2002; Poulin 2004). In a related vein, the neutral theory is concerned with resident organisms (Chave 2004). Most obviously this excludes the complications provided by migratory species, but the principle can be extended further to organisms that apportion time as a niche axis (Kronfeld-Schor & Dayan 2003).
Second, the theory concerns trophically similar groups of species. Attempts have been made, with some success, to develop models in which the dynamics of functionally more similar groups of species are neutral, but between groups the vital rates (birth, death, dispersal, speciation) differ (e.g. Etienne & Olff 2004b). However, how well this approach generalises, and the extent to which trophic and other interactions between these groups can be ignored, remains unclear.
A parallel can perhaps be drawn between ecological and population genetic theory. Neutral theory applies rather well to a wide range of apparently synonymous genetic variation, which has enabled population geneticists to use such markers to explore evolutionary history. However, stochastic variation (mutation and drift) becomes less important depending on the strength of selection. Ecological systems might operate on the same continuum of neutral to non-neutral forces, with neutral (i.e. stochastic processes) being more important in some assemblages, communities, or parts thereof.
Shaping the future
Which ever of these ways of reconciling neutral and niche theory ultimately proves to be correct, consideration of the neutral theory of biodiversity and biogeography should serve to focus attention on a number of questions in functional ecology. We would identify the following.
(i) Why has a unified theory of biodiversity and biogeography founded on niches not arisen, and is the explanation adequate that ‘the world is too complex a place’?
(ii) Is energy variation across space (Hawkins et al. 2003) and time (Crame 2001) responsible for variation in the number of individuals of all species in a local assemblage and the numbers in the metacommunity, and can this constitute a link between the scaling models of West et al. (1997) and neutral community models?
(iii) Is there spatial variation in speciation rate per birth, which is a consequence of spatial variation in energy (Rohde 1992)? The few tests of this idea based on sister-group comparisons (e.g. Bromham & Cardillo 2003) may well have confounded variation in the number of individuals of all species in the metacommunity and the speciation rate per birth.
(iv) Apparent fits are often observed between the physiological tolerances and capacities of species and the environmental conditions that they experience. What is the relative contribution to these fits that arises from selection on individuals of the species to adapt to prevailing conditions and that which arises from the costs to individuals of maintaining broader tolerances and capacities? It is clear that a balance between selection and the mean dispersal rate of individuals over the metacommunity landscape, the latter a critical parameter of the neutral theory, is of considerable importance in determining this contribution, as well as in determining both local abundance and range size (for review see Lenormand 2002; Butlin, Bridle & Kawata 2003).
(v) Given the huge range of ways in which individuals of different species vary from one another, which are the characteristics that are most important in influencing assemblage and community structure, and how does this vary with spatial scale? Much of the work on the relationship between biodiversity and ecosystem functioning is occupied with this question (Loreau et al. 2002; Rosenfield 2002; Naeem & Wright 2003), but rarely in the context of neutral theory.
Two final questions that are perhaps of less direct interest to functional ecologists, but are nonetheless essential for understanding spatial and temporal variation in the distribution, abundance and richness of species are as follows.
(vi) At the local scale, what proportion of species diversity is maintained by local processes and what by regional ones, and is it essential to include a third, global scale? Davis (2003) suggested that the addition of a global community level to the metacommunity and local community levels, so creating a hierarchical system, would allow the phenomenon of global biological invasion to be more adequately addressed than is the case with the present two-tiered model. The model developed by He (2005) goes a considerable way towards addressing this question by showing that a continuum exists between large-scale and small-scale (or regional and local) processes, eliminating the artificial distinction between the two.
(vii) What is the relationship between abundance and speciation probability, and what form does speciation take? Although there have been long-standing arguments that species with high abundances and large ranges should have greater rates of speciation, the converse has also long been advanced as the correct relationship (reviews in Chown 1997; Gaston & Chown 1999). Recent work suggests that the latter is the case (Jablonski & Roy 2003). Moreover, even taking into consideration the longer lifetimes bestowed on species by large ranges (the relationship between range size and abundance is strongly positive, though with variation; Gaston et al. 2000), they do not appear to have the highest speciation rates. Predictions from the neutral theory may provide further insight into this relationship as well as to the predominant mode of speciation (see Gaston & Chown 1999; Hubbell & Lake 2003; Ricklefs 2003b).
Addressing these and the many related issues should keep the pages of Functional Ecology filled for a long time to come.
We thank K.L. Evans, M.A. McGeoch and R.E. Ricklefs for comments on the manuscript. K.J.G. is grateful for an Ellerman Foundation award from the University of Stellenbosch. S.L.C. is supported by the DST Centre of Excellence for Invasion Biology.