Methods
Survey data. We used information on well-surveyed avian communities from published studies and our own field work (Sol et al. 2020b, c). The term community is used to describe assemblages within a region that could be unambiguously assigned to one of five land-use classes: native vegetation, rural, little urbanized (e.g. urban parks), moderately urbanized (suburbs) and highly urbanized (city centers). Our criteria to accept data were that, at each study region, bird species were exhaustively surveyed and that at least one of the communities was invaded by one or more non-indigenous bird species. The resulting database comprised information on 1367 species —63 non-indigenous in one or more communities— in 277 communities from 50 study regions. In 47 of these study regions, surveys were conducted both in urbanized areas and in the surrounding non-urbanized areas, allowing us to assess the effect of human-related disturbances on invasiveness. The entire dataset is available in Sol et al. (2020a).
Introduced species information. We extracted information on bird species introduced and established in regions outside their native range from the Global Avian Invasion Atlas GAVIA (Dyer et al. 2017), the most complete dataset of historical introductions currently available. The dataset covers > 27,000 introductions of 971 bird species and also provides information of the year of introduction for established exotics.
Invasion success. We used two classic measures of invasion success: 1) the presence/absence of the invader in communities from a same region, 2) the abundance of exotic species relative to native species within the invaded communities. The Gavia dataset (see above) confirmed that the exotic species analyzed were introduced sufficiently time ago (range from 1853 to 1975) to make their abundances little influenced by time lags or introduction effort. The density of occurrences of an animal in a given place may however be influenced by their size, either directly (e.g. by increasing the demands of food) or indirectly (by correlating with the fast-slow continuum of life history variation). We evaluated the importance of body size by including it as a co-variate in the models (see below).
Phylogenetic information. We extracted from the BirdTree database (http://www.birdtree.org) (Jetz et al. 2012), two samples of 5,000 phylogenies that included all native and exotics species included in this study. Each sample was based two alternative backbone topologies: Hackett et al. (2008) and Ericson et al. (2006). Phylogenetic distances were estimated based on two summary trees calculated from both samples. These were computed as the maximum clade credibility tree using the program TreeAnnotator (included in the package BEAST v1.8.0) (Drummond et al. 2012). Because results from both phylogenies were highly coincident, we only present the results of analyses based on Ericson et al. (2006).
Functional traits information. We described the foraging niche of species based on three types of functional traits extracted from Pigot et al. (Pigot et al. 2020): 1) eight morphological traits (all log-transformed), 2) seven diet categories, and 3) 30 categories of foraging behavior collected from the literature, as described in Sol et al. (Sol et al. 2020b). Diet and foraging behavior were described as fuzzy variables, ranking each category from 0 to 10 as a function of the degree of use (Pigot et al. 2020). In addition, we extracted information on relative brain size, resource niche breadth and brood value from Sol et al. (Sol et al. 2014a, 2020b) and Sayol et al. (Sayol et al. 2016). To estimate niche breadth, we used Rao’s quadratic entropy (De Cáceres et al. 2011), based on the functional distances matrix (see below) derived from niche information.
Phylogenetic and functional distance metrics. We used two complementary metrics of phylogenetic distance, the average and nearest phylogenetic distance between each exotic species and all the native species of the community (Tucker et al. 2017). For each invader from a given site, we calculated the mean (MPD) and nearest phylogenetic distance (NPD) between invaders and all native species of each community regardless of whether the exotic species was present or not (see justification below). Summary trees were pruned down to the species present in the communities and phylogenetic distances among species were calculated by means of the function “cophenetic” in the R package “Picante” (Kembel et al. 2010). To consider the fact that most native species in the community were rare, we used weighted versions of MPD and NPD in which the distance of the invader to each native species was multiplied by 1 – Abr , whereAbr is the relative abundance of the invader in the community. This increases the distance of rarer species relative to more common species. Functional distances were estimated the same way, but applying Gower’s distance (Gower 1971) to morphological, diet and foraging behavior traits (the last two coded as fuzzy variables).
Spatial scale and biases. We used our community dataset to assess how the nearest phylogenetic distance between exotic and native species varied with spatial scale (i.e. community, region, country and continent). We used simulations to explore whether spatial patterns were affected by the increase in species with scale or by the fact that exotic and native species came from different regions. The former was investigated by randomly sampling without replacement pairs of species in different numbers (from 1 to 125) from the entire avian phylogeny (n=9993). We compared all species as well as the subset that were introduced in regions outside their native range and the subset of them that have become successfully established in the new region. The analysis of geographic effects was conducted comparing species within and across biogeographic realms that have been either donors (Palearctic) or receptors of invaders (Nearctic and Australian).