Introduction
A central attribute of ecological communities is that only few species and interactions may be essential to maintain community structure and ecosystem processes, even though many species might coexist and interact (Ellison et al. 2005; Valiente-Banuet et al. 2015). The loss of essential species and the disruption of essential interactions may have cascading effects in the community and increasing species extinction rates (Gilbert 1980; Koh et al. 2004; Dunn et al. 2009; Rodriguez-Cabal et al. 2013). Thus, studying the loss of native species and gains of non-native invasive species is crucial to understanding the structure and functionality of communities in a changing world.
In particular, mutualistic interactions such as pollination and seed dispersal are essential in supporting terrestrial ecosystems because most of the plant reproduction success relies on these interactions (Bond 1994; Ashman et al. 2004). Ecological networks provide a systematic way of representing, characterizing and comparing the complexity of ecological communities, in which species are represented by nodes and interactions by links between nodes (Pimm 1982; van Veenet al. 2008). In the last two decades, the study of mutualistic networks has been critical to understand the structure and functionality of communities (Bascompte et al. 2003; Bascompte & Jordano 2007; Rohr et al. 2014). However, although several studies have focused on the impacts of human-caused environmental change on mutualistic interactions (Traveset & Richardson 2006; Rodriguez-Cabal et al.2007; Tylianakis et al. 2008; Traveset & Richardson 2014), few studies have focused on how such change triggers the disassembly of mutualistic webs (Fortuna et al. 2013; Rader et al. 2014). Moreover, the indirect effects on interaction webs are not well understood because indirect effects are complex to study and because perturbations can be attenuated by functional redundancy (Lawton & Brown 1993; Terborgh & Feeley 2010).
The temperate forest of Patagonia is known for its highly endemic flora and high proportion of plant species requiring animal mutualists for pollination and seed dispersal (Aizen & Ezcurra 1998). Almost 60% of plant species require at least one animal species for pollination (Riveros 1991) and nearly 50% of plants produce fleshy fruits suggesting animal seed dispersal (Armesto et al. 1987). In the northern portion of this temperate forest occurs a unique mutualistic interaction involving a hummingbird (Sephanoides sephaniodes ), a marsupial (Dromiciops gliroides ), a mistletoe (Tristerix corymbosus ), and its main host (Aristotelia chilensis ), a common understory shrub. The nectar produced by the mistletoe is the primary food resource for the hummingbird during winter (Smith-Ramírez 1993; Aizen et al. 2002). In spring and summer the hummingbird pollinates almost 20% of the endemic woody genera in this region (Aizen & Ezcurra 1998; Aizen 2003). The marsupial provides an efficient seed dispersal service by dispersing at least 16 fleshy-fruited species, including large fruits (>11 mm) not dispersed by the native birds (Amico et al. 2009), and is the only seed disperser of the mistletoe (Amico et al. 2011), which thus depends fully on the marsupial for its persistence. Previous studies found that herbivory onA. chilensis by non-native ungulates leads to the disassembly of the hummingbird-mistletoe-marsupial interaction (Rodriguez-Cabalet al. 2013). Owing to the high degree of asymmetry of plant-animal mutualisms in the temperate forest of Patagonia, where most of the plants depend only on a few mutualistic partners (Armestoet al. 1987; Riveros 1991; Aizen & Ezcurra 1998), the disruption of this interaction could trigger a cascade of linked extinction events throughout the community.
In this study, we evaluate whether the disruption of a hummingbird-mistletoe-marsupial interaction by non-native ungulates can have cascading effects on pollination and seed dispersal networks in the temperate forest of Patagonia, Argentina. Specifically, we address the following questions: (1) Is the hummingbird-mistletoe-marsupial mutualism a keystone interaction? We expect that the complexity and functional redundancy among generalist species of pollination and seed dispersal networks will be greater in sites with this interaction than in sites without the interaction. We also expect that the ecological importance of the species involved in this proposed keystone interaction will be greater in sites with this interaction, occupying a high number of positions in the networks. (2) Does the disruption of this proposed keystone interaction produce cascading effects on the community? We expect that the complexity and functional redundancy among generalist species of pollination and seed dispersal networks will be lower in sites invaded by non-native ungulates than in intact forest sites.