ability to compete well for nutrients has been sug-
gested to happen at the expense of the ability to
compete for light (Tilman 1990). Also, competing
needs for growth and reproduction result in differ-
ent optimal combinations in different species, de-
pending on the conditions to which they are
adapted. Such evolution-driven trade-offs make
competitive hierarchies much more likely than
transitive networks in most cases. Said simply, it
is unlikely that the species adapted to the ‘left-tail’
of some underlying niche axis will outcompete a
species from the ‘right-tail’ or vice versa.
An important class of exceptions exists in organ-
isms in which the outcome of interactions is not
driven by resource competition, but by chemical
warfare, such as allelopathy. In this case, few
trade-offs are expected, as the defence and compe-
tition is much more strongly information-based
(what is the opponent sensitive to) than energy-
or nutrient-based (which requires morphological
adaptations, such as allocation shifts).
One example of such an intransitive competitive
network involves different strains of bacteria
competing in spatially structured environments.
Kerret al.(2002) explored how a mixture of me-
chanisms can allow three strains ofEscherichia coli
to coexist via a set of intransitive interactions that
are akin to a game of rock–paper–scissors. There
are three key features of this system – one strain
produces a toxin, called colicin, whereas the two
other strains are either sensitive or resistant to the
toxin (Kirkup and Riley 2004). The three strains
interact as an intransitive network in the following
way. All of the strains compete consumptively for
a carbon source, such as glucose, but they differ in
competitive ability such that colicin-sensitive
strains are the best consumptive competitors,
followed by colicin-resistant strains followed
by colicin-producing strains. However colicin-
producing strains can displace colicin-sensitive
strains by poisoning them, but colicin-producing
strains can in turn be displaced by competitively
superior colicin-resistant strains. These patterns of
sequential displacements occur only in an un-
mixed spatially complex environment, such as
the surface of a culture plate. In a well-mixed
spatially homogeneous environment, such as a
well-stirred liquid culture, only one strain per-
sists. It appears that limited dispersal can promote
diversity by allowing an intransitive competitive
network to exist (Fig. 1.3). A similar role is played
by dispersal limitation in maintaining diversity in
neutral communities (i.e. where all species are as-
sumed to be functionally equivalent; e.g. Hubbell
2001).
Qualitatively similar but much more complex
networks may explain the high diversity of bacteria
that manage to coexist in soils, where dispersal is
limited. For example, Torsviket al.(1990) made
initial estimates using DNA hybridization techni-
ques that suggested that thousands of bacterial taxa
could occur per 30 g of soil. These conservative
estimates were re-evaluated by Dykhuizen (1998),
who estimated that 30 g of forest soil could contain
over half a million species, depending on the as-
sumptions made in the analysis.Plethodon
glutinosusPlethodon
glutinosusDesmognathus
ochrophaeusDesmognathus
ochrophaeusEurycea
cirrigeaEurycea
cirrigeaPlethodon
serratusDesmognathus
wrightiDesmognathus
wrightiPlethodon
jordaniPlethodon
jordaniPlethodon
serratusFigure 1.2Potential (top) and realized (bottom) networks
of competing salamanders in the study of Hairston
(1981). Dashed lines in the lower graph indicate
unrealized competitive interactions; the heavy solid line
indicates that only two species,Plethodon glutinosus and
Plethodon jordani, actually compete.
12 SHAPE AND STRUCTURE