Ecoregion-scale variables
We tested whether tip-based metrics varied relative to ecoregion ecotone and cores, as well as to other variables characterizing the ecological and biogeographic context of ecoregions (Table 1). First, we superimposed the points on the ecoregion shapefile (Olson et al. 2001) to determine whether points were in ecoregion cores or ecotones, and whether their habitat was either forested or open. The distinction between forested and open habitats reflects broad differences in vegetation structure and in the type of available niches and resources (Vivo and Carmignotto 2004). Ecoregions belonging to forest, woodland, and mangrove biomes were considered forested habitats, while ecoregions belonging to grassland, shrubland, desert, savanna, inland-water, and the rock and ice biomes were considered as open habitats.
We considered the predominant habitat type and number of neighboring ecoregions in our analyses. Neighborhood can be important because an ecotone assemblage can have values of tip-based metrics that resemble a core assemblage when the ecotone lies between two ecoregions with similar habitat. That ecotone assemblage is then expected to have lower transition rates, higher stasis time, and longer last transition times than an assemblage located in the ecotone between ecoregions with contrasting habitats.
We considered the importance of the Central Andes and Atlantic Forest in predicting aTR, aST, and aLT because sigmodontine diversification and richness have a close relationship with these regions (Patton et al. 2015; Maestri and Patterson 2016; Maestri et al. 2019). We superimposed ecoregions on the shapefiles of Central Andes (Löwemberg-Neto 2015) and the Atlantic Rainforest (Muylaert et al. 2018, available at https://github.com/LEEClab/ATLANTIC-limits) to distinguish their ecoregions from others (Fig. S1, see Supplementary Results in Supporting Information). We treated the southernmost portions of the Andean region (mainly southern Argentina and Andean piedmont), as well as extreme northern and southern portions of Atlantic Rainforest as belonging to other regions (e.g., Uruguayan Savanna, Cerrado, Caatinga), not as primary loci of sigmodontine diversification (Maestri et al. 2019).