Third, they provide new perspectives on food
web and community properties, and potentially
a more complete understanding of the mechan-
isms that generate them. We provided several
examples of such applications in sections 12.2.2
and 12.2.3. Community evolution models are
capable of giving as good a match to binary
data sets as classical food web models such as
the Cascade and Niche models. But, additionally,
they provide the dynamics of food web structur-
ing whereas other models are only able to repro-
duce empirical data at a given time. Finally,
community evolution models describe species
interactions based on individual-level traits, so
that community properties are emergent proper-
ties of processes that take place at a smaller
scale. As a consequence, the mechanisms under-
lying emerging structures are much clearer than
inthecaseoftheNicheorCascademodels,
which use large-scale patterns, such as species
diversity and connectance, to predict other large-
scale patterns, but cannot account for species
diversity and connectance in the first place.
12.4.1 Possible extensions of community evolution models
As discussed in section 12.3.3 community evolution
models can include other traits than body size. What-
ever other traits are chosen, however, body size seems
a natural candidate for a primary trait. Body size has
well-documented effects on many life-history traits
and trophic interactions in all taxonomic groups, on
both plants and animals. It has been suggested as a
good proxy for a species’ trophic level, and has been
used abundantly in both static food web models and
the new community evolution models.
Although the importance of body size is undis-
puted, species interactions are the product of several
traits. Therefore, a straightforward extension of
these models would be to include one or several
other traits to better account for species interactions.
Some of the good candidates, such as elemental
ratios, habitat choice and dispersal rates, are dis-
cussed in section 12.3.3. In addition to these, another
important trait is niche width, which encapsulates a
species’ ability to consume a more or less large array
of prey species. In the model presented in section
12.2.1.2, we made the simplifying assumption that
niche width is constant among species and does not
evolve. We are currently working to add evolution
of niche width in this model.
Another possible extension of the model is the
incorporationofothertypesofinteractions.Cur-
rent community evolution models account for tro-
phic interactions, and sometimes interference
competition. There is increasing evidence that
other types of interactions, such as mutualism
and parasitism, play an important role in the struc-
ture and dynamics of natural communities (e.g.
Callawayet al. 2002; Laffertyet al.2006;Michalet
et al.2006).Networksofmutualisticinteractions
are now documented, and some recent studies
suggest a possible role of evolution in constraining
their structure (Jordanoet al. 2003; Va ́zquez and
Aizen 2004; Bascompteet al. 2006). The biomass of
parasites is sometimes comparable to the biomass
of predators, so that nutrient flows involved in
parasitic interactions may no longer be neglected
(Laffertyet al. 2006). The main problem with the
inclusion of such interactions in community evo-
lution models is to find traits that can be linked to
them unambiguously in the same way as body size
is for trophic interactions. Goudard and Loreau
(2007) recently proposed a first community assem-
bly model that includes all types of species inter-
actions. Their model could be extended to include
evolutionary dynamics.12.4.2 Empirical and experimental implications of community evolution models
When community evolution models are based on
well-defined traits, it is possible to include physio-
logical or genetic information on these traits. The
benefits and costs of these traits are then assessed
from empirical or experimental knowledge, and the
evolutionary trade-offs that constrain them are
built as assumptions into the models. This is both
a blessing and a curse. The advantage is the possi-
bility to play with the various fitness components to
determine how each trait influences emerging pat-
terns. On the other hand, evolutionary trade-offs
are notoriously difficult to obtain, and their shape
strongly influences the results of the evolutionaryEMERGENCE OF COMPLEX FOOD WEB STRUCTURE 177