Information Technology Reference
In-Depth Information
alteration on the network of interactions. In the meantime, we again must rely on
numerical simulations. In the most common methodology a progressive number of
species goes extinct and the inuence of network structure on the size of the co-
extinction cascade, or the subsequent loss of evolutionary history, is investigated.
The consensus result, as shown by Memmott et al. (2004) building on the papers
by Albert et al. (2000) for the Internet and Sole and Montoya (2001) and Dunne
et al. (2002a) for food webs, is that the heterogeneous, nested, structure of mutu-
alistic networks makes them robust to the loss of specialists or to the random loss
of species [Memmott et al. (2004)].
When one examines, however, the identities of the coextinct species and their
phylogenetic positions, there is a signicant phylogenetic signal on both the number
of interactions per species and on whom they interact with [Rezende et al. (2007b)];
this is true for as many as half of the communities examined. Because of this, coex-
tinction cascades do not involve randomly picked species but phylogenetically close
species. The loss of evolutionary history then proceeds faster than expected in the
absence of such phylogenetic signal, leading to a biased pruning of the evolutionary
tree [Rezende et al. (2007b)].
Another driver of global change is habitat loss. There is only one study we
are aware of that empirically studies the consequences of habitat transformation
on network structure. Tylianakis et al. (2007) explored how the structure of host-
parasitoid networks changes across an environmental gradient. They demonstrated
that even without a reduction in the number of species, habitat loss changes the
structure of interaction networks with important implications for their collapse. To
further assess the inuence of habitat transformation, requires the use of models.
An important area of research is the study of metacommunities, species in-
teractions across discrete habitat patches maintained by a balance between local
processes and regional dispersal across patches [Leibold et al. (2004); Holyoak et al.
(2005)]. This important area has mainly focused on theoretical work consisting
of a small number of interacting species. Melian et al. (2005) and Fortuna and
Bascompte (2006), on the other hand, have studied real ecological networks across
space from a theoretical perspective but with added realism by using the structure
of the real ecological networks.
A rst step towards understanding the consequences of habitat loss on mutu-
alistic networks was the study of a spatially implicit model of metacommunities
in which two real mutualistic networks were used as a skeleton for the theoretical
model [Fortuna and Bascompte (2006)]. The heterogeneous, nested, structure of
mutualistic networks confers a higher level of robustness to habitat loss. While
species start to go extinct sooner than for expected at random, the network as a
whole persists for greater values of habitat loss [Fortuna and Bascompte (2006)].
This is only a rst step, however, because the metacommunity is assumed to live
in an idealized spatially implicit model composed by an innite number of patches
with similar dispersal to any other patch. It maintains the structure of the mutu-
