web comes from Wootton’s (1994) work on inter-
actions among some of the species living in the rocky
intertidal zone of Washington State in the USA
(see Fig. 1.1c). The search for patterns in interaction
networks has only begun rather recently. For a
recent review, see Ohgushi (2005), who focuses on
indirect interactions in plant-based terrestrial food
webs.
1.1.3 Three general questions
For each type of ecological network, we suggest that
there are three general types of questions: those
about whether and what structural regularities
exist within and among observed networks; those
about which mechanisms are responsible for the
structure of the networks and of any structural reg-
ularities that exist; and lastly, those about the gaps in
our knowledge. The aim here is to define a set of
general contemporary questions that we will use to
organize each of the following sections.
1.2 Competitive networks
1.2.1 Structural regularities
Real-world competitive networks remain poorly
studied and documented. A fundamental question
about their structure remains quite unresolved:
given a set of co-occurring species that appear
to consume a similar class of resources (such as
different-sized seeds, forming a potential guild,
sensuRoot 1967), how many pairs of these species
really compete? That is, how many of the plausible
links are actually realized in competitive networks?
Groups of taxonomically similar species are often
assumed to be potential competitors, but experi-
ments show that only a small fraction of the poten-
tial competitors actually compete. Sometimes this
can be generalized to simple integrative traits,
building on the early work on limiting similarity
(for a review, see Brown 1981). For savanna large
herbivores, Prins and Olff (1997), for example,
found that large herbivores overlap more in
resource use when they are more similar in body
size. Thus, more similar-sized species are more
likely to interact.
Hairston (1981) studied a group of six species of
plethodontid salamanders found in moist forests of
the mountains of North Carolina, USA. Based on
their morphological similarity and similar life-
styles, all being carnivores feeding on small inver-
tebrates, it seemed plausible that all six species
might compete for food (Fig. 1.2). Over a 5 year
period Hairston removed each of the two most
common species at this site. Only one of the remain-
ing five species responded positively to each re-
moval, and this was the most common species at
the site. These results imply that only the two most
common species compete. Hairston suggested that
competition might be for nest sites, rather than for
shared prey, but subsequent research showed that
direct territorial (interference) interactions between
salamanders constituted the primary mechanism of
competition (Nishikawa 1985). In this case, the net-
work of competitors is considerably simpler than
the completely connected system that would result
if all species shared and competed for common
resources (Fig. 1.2).
Another general and important question about
the structure of competitive networks is whether
the interactions form a hierarchy, or an intransitive
loop. In a hierarchy, for example, species A out-
competes species B and C, B outcompetes C, and
C cannot outcompete either A or B. This will mostly
lead to predominance of a single competitively su-
perior species (A) and therefore will foster relative-
ly low biodiversity. In an intransitive loop, in which
A outcompetes B, B outcompetes C, and C outcom-
petes A, all three species can potentially coexist.
Consequently, mechanisms of interaction that
allow intransitive competitive networks might ex-
plain high biodiversity in some ecosystems.
At first hand, intransitive competitive networks,
especially within one trophic level, would be un-
likely to emerge, based on the theory of life history
and competitive trade-offs. An organism’s life his-
tory generally reflects a compromise in the use of
available resources. Species generally specialize
along environmental axes of resources axes and
stress factors in such a way that investing in some
life history/functional trait (e.g. the ability to crack
big seeds in birds) occurs at the expense of some
other function (e.g. the efficiency with which small
seeds can be collected). In plants, for example, theTHE TOPOLOGY OF ECOLOGICAL INTERACTION NETWORKS 11