Toward multiplex ecological networks: Accounting for multiple interaction types to understand community structure and dynamics

Introduction In drylands, there is often not enough water for the whole land to be covered by vegetation. Instead, vegetation occurs in patches (Figure 6.1a), where established plants provide a favorable environment for the recruitment of new individuals, for example, by creating shading and locally favoring soil and resource retention (Aguiar and Sala, 1999). This facilitation mechanism is well documented in drylands and known to be of great importance for dryland plant communities (Soliveres et al., 2015). At the same time, plant species compete with each other for water, the main limiting resource in those ecosystems, and are consumed by herbivores. This is only one example of the variety of ecological interactions that co-occur in ecological communities (Figure 6.1b, c). As early as 1859, Charles Darwin highlighted the diverse interaction types that link species in nature. One of his famous examples on how cats influence seed set in red clover (Darwin, 1859) illustrates how indirect effects can percolate through a variety of ecological interactions that co-occur in ecological communities. Cats eating mice is a trophic interaction, mice building nests that are later used by bumble bees is an ecological engineering interaction, and these bees in turn increasing seed set in clover is a mutualistic pollination interaction. Despite the recognized importance of this variety of interaction types in nature, ecological research has largely focused on analyzing one single type of interaction at a time, for example trophic networks or food webs (Pimm, 1982; Cohen et al., 1993; de Ruiter et al., 1995; Brose et al., 2005, 2006; Neutel et al., 2007), mutualistic communities (Jordano et al., 2003; Blüthgen et al., 2007), host–parasite and host–parasitoids webs (Vázquez et al., 2005; Krasnov et al., 2012), and facilitation networks (Verdú and Valiente-Banuet, 2008). Studies on networks of single interaction types – greatly dominated by food-web studies (e.g., Ings et al., 2009) – have suggested that single-interaction ecological networks exhibit predictable structural regularities with important consequences for their dynamics (Williams and Martinez, 2000; Bascompte et al., 2003; Montoya et al., 2006; Verdú and Valiente-Banuet, 2008; Thébault and Fontaine, 2010). For example, mutualistic (Bascompte et al., 2003) and facilitation networks (Verdú and Valiente-Banuet, 2008) tend to be more nested than expected by chance, a feature that has been shown to give rise to positive complexity–stability relationships in dynamic models (Okuyama and Holland, 2008; Thébault and Fontaine, 2010).