order to have a deep understanding of the contem-
porary views of metacommunity ecology, it is nec-
essary to know how the ideas have evolved. In
addition, when possible, we discuss theory and
empirical work hand-in-hand, which we feel is a
more appropriate way to discuss this information,
rather than keeping them separate. We argue that
metacommunity ecology will most rapidly advance
with an intimate, rather than superficial, connection
between theory and data.
As a first order of business, it is necessary to
recognize that, although metacommunity ecology
as a field has exploded in the past decade or so
(Tilman 1997; Leiboldet al. 2004; Holyoaket al.
2005), its historical roots are much older. For exam-
ple, early experimentalists testing Lotka–Volterra
competition and predator–prey models quickly re-
cognized that it was exceedingly difficult for com-
petitive or predator–prey species pairs to coexist
without some sort of spatio-temporal variation
that was missing from the simple models (Gause
1934; Park 1948, 1954; Huffaker 1958). Insect
ecologists have long recognized the importance of
large-scale processes and dispersal in population
dynamics, and aspects of metapopulation ecology
were inherent in the writings of Andrewartha and
Birch (1954). Similarly, during the ‘renaissance’ of
community ecology – the 1960s and 1970s – spatio-
temporal perspectives were quite common (e.g.
MacArthur and Levins 1964; MacArthur and Wilson
1967; Levins and Culver 1971; MacArthur 1972;
Horn and MacArthur 1972; Levin 1974; Slatkin 1974).
Probably as a consequence of the increase in sta-
tistical and experimental rigor that arose from a
series of critiques (e.g. Stronget al. 1984), communi-
ty ecology studies in the 1980s and early 1990s were
primarily aimed at local-scale processes, such as the
mechanisms that influence species coexistence and
relative abundances. The recent recognition of the
importance of spatial processes probably has sever-
al simultaneous origins. First, while the focus on
experiments and local-scale mechanisms offers
much to community ecology, it is not able to ac-
count for many of the patterns that are observed in
nature, at both local and larger scales (e.g. Ricklefs
2004). Second, Hubbell’s (2001) ‘neutral’ theory
served to catalyse the field by espousing the contro-
versial perspective that local interactions are irrele-
vant at larger scales, and that a perspective based
solely on stochastic processes of colonization and
extinction (and speciation) could approximate nat-
ural patterns (see Chave 2004; Alonsoet al. 2006;
Holyoak and Loreau 2006). Third, in the context of
applied community ecology, spatio-temporal per-
spectives are often necessary to understand how
communities are degraded, and how they can be
restored (see Chapter 9). Fourth, statistical analyses,
including spatial, multivariate and computational
analyses, have become more sophisticated and
powerful (e.g. Clarket al. 2007), and, when applied
in combination with theoretical and experimental
information, can provide a deeper understanding
of both the patterns and underlying processes of
community structure than could have been gained
previously.5.2 The varied theoretical perspectives on metacommunities
Recent syntheses have suggested that metacommu-
nity theory can be roughly categorized into four
general conceptual frameworks: Neutral, patch
dynamics, species sorting and mass effects (Leibold
et al. 2004; Holyoaket al. 2005). The four perspec-
tives are not exclusive. For example, mass effects
and patch dynamics can be viewed as occurring
along a gradient of organism movement intensity
and habitat heterogeneity. The relative importance
of species sorting, patch dynamics and mass effects
in metacommunities is a primary issue when exam-
ining the influence of local and regional processes
on patterns of community composition (see
Chapter 9). Furthermore, the stochastic processes
inherent in the neutral theory are one component
of more complex spatio-temporal niche models
(Chesson 2000; Adleret al. 2007).
There has been considerable attention directed
towards testing empirical data from a variety of
systems against the predictions of metacommunity
models based on neutral versus niche differences
(e.g. Chave 2004; McGillet al. 2006). Unfortunately,
patterns from one type of data are generally not
able to unambiguously separate the different
model predictions (Chase 2005; Chaseet al. 2005)
(Table 5.1). Although data on the relative abun-
dances of species in a community are often used58 SPACE AND TIME