Limitations
Environmental variation unrelated to the distance from ecotones could alternatively explain the lower transition rates and higher stasis time and last transition times we find for ecotone species. For instance, ecotones of many ecoregions may be formed by other environmental factors (e.g., soil moisture and type) rather than climate, which might make them temporally more stable than the ecotones formed by climate (Cantidio and Souza 2019). In addition, patches of stable habitat currently found at ecotones could rather be at the core of ecoregions in the past. Thus, shifts in ancestral area could alternatively influence trait evolution. Combining the tip-based metrics we developed here with approaches that incorporate shifts in ancestral area (e.g., Maestri and Duarte, 2020) can help to show whether trait evolution is produced by shifts in distribution.
We used diet because it varies across sigmodontine species (Paglia et al. 2012) and habitat types (de Vivo and Carmignotto 2004), and is available for virtually all species (e.g., Wilman et al. 2014). In addition, each diet type is subjected to a particular set of selection pressures and presents different probabilities of transition and speeds of evolution (Price et al. 2012; Maestri et al. 2017). Although other important traits like life-mode could produce different results, we believe our results are robust to trait choice as diet and life-mode were shown to produce similar macroevolutionary patterns of morphological disparity in sigmodontine rodents (Maestri et al. 2017).
Results were mostly robust when considering small-ranged species, whose rates and time of diet transition respond to the position x habitat type interaction —similarly to overall results. The location of species assemblages in Andean and Atlantic Rainforest ecoregions has a large influence on its tip-based metrics. This result is largely expected as most of their small-ranged species both speciated and subsequently evolved within these regions (D’Elia and Pardiñas 2015).