Introduction
To disentangle the mechanisms producing the biological diversity seen in nature, ecologists increasingly seek to integrate ecology and evolution (Wiens and Donoghue 2004; Jetz et al. 2012; McGill et al. 2019). Mapping traits onto phylogenies is essential for such integration, as mapping traits reveals the rates and tempo of evolution of behavioral, morphological and ecological characteristics (Bollback 2006). Knowledge of trait evolution has often been applied to evaluate the evolutionary mechanisms producing, for example, the appearance of ecological innovations and the bursts behind evolutionary radiations (Cantalapiedra et al. 2014; Joy et al. 2016; Maestri et al. 2017). Nevertheless, despite extensive study on rates of trait evolution over time and across clades (Gingerich 2009; Joy et al. 2016 and references therein), understanding how these rates vary over space is equally challenging, and still little understood.
In terms of species diversification, rates are heterogeneous over space. Between-biome comparisons suggest that some biomes are more speciation-prone than others (Goldberg et al. 2005; Davies and Buckley 2011; Antonelli et al. 2018). For example, Amazonian tropical forests were inferred to be the main source of Neotropical biodiversity due to high speciation and low extinction rates, yielding species accumulation within tropical forests (Davies and Buckley 2011; Antonelli et al. 2018). Marine tropical biomes appear to be sources of temperate-region bivalves owing to the dispersal of taxa that evolved in tropical regions (Goldberg et al. 2005). Montane portions of the Andes and also of the Atlantic Rainforest were shown to be centers of early rodent diversification and diversity accumulation into the Neotropics (Leite et al. 2014; Maestri et al. 2019). Although these findings implicate cradles and museums of biodiversity, we still need to know the situations where diversification results in trait evolution (e.g., Oliveira et al. 2016), as well the roles of historical and ecological factors in producing spatial heterogeneity in trait evolution.
Here, we tested whether within-biome heterogeneity in species trait evolution would be related to the distance from spatially and temporally unstable ecotones. If this relationship holds, then assemblage position relative to ecoregion boundary, or its interaction with habitat type, should be the main predictor of evolutionary speed relative to other ecological and historical variables like habitat type, neighborhood characteristics, and location (either in the Andes or Atlantic Rainforest). Ecotone is a concept used at several spatial scales to characterize the boundary between habitat patches; the environmental contrast between adjacent patches can produce boundaries that organisms may perceive (Cadenasso et al. 2003). We hypothesize that assemblages located at ecoregion ecotones have species with more changes from the ancestral character state (higher transition rates), have maintained the ancestral character state for shorter time (lower stasis time) and have more recent transitions to the current character state (lower last transition times) than assemblages from ecoregion cores.
Region cores are more homogeneous environmentally and in terms of selection pressures exerted on organisms, since environmental changes are buffered before they reach cores (Mayle et al. 2004; Mayle and Power 2008; Donoghue and Edwards 2014). Populations inhabiting region cores should be large and stable in size over time, as well as occur under environmental conditions similar to the conditions found in the ancestral range (Wiens and Graham 2005; Pearman et al. 2008; Davies and Buckley 2011). In contrast, region ecotones are more heterogeneous, both environmentally and in relation to selection pressures exerted on organisms, because environmental challenges are first noticed in ecotones, which leads to changes in vegetation development and in the location, quality and type of habitats and limiting resources on which individuals depend (De Vivo and Carmignotto 2004; Eckert et al. 2008; Sexton et al. 2009; Donoghue and Edwards 2014). Populations inhabiting ecotones should be smaller, be under stronger extinction pressure, and have less stable population size than core populations (Karanth et al. 2006). They also should show shifts from ancestral characters as these shifts may be needed to persist in ecotones (Pearman et al. 2008; Sexton et al. 2009; Benton 2010; Donoghue and Edwards 2014).
Spatial heterogeneity in the rates of species diversification and trait evolution is well known (Benton 2010; Jetz et al. 2012; Oliveira et al. 2016). But while there are a few metrics to evaluate spatial heterogeneity in rates of diversification (including the tip-based metrics reviewed in Title and Rabosky 2019), there is no consensus or metric on how to evaluate spatial heterogeneity in rates of trait evolution. In order to evaluate our hypothesis, we propose three tip-based metrics for calculating trait transition rates, stasis time and last transition time (TR, ST and LT, respectively). All three metrics aim to calculate species-specific direction and time of character-state transitions from the phylogeny tips to the root. Transition rates indicate how many times the ancestral character has changed over time. Stasis time indicates the maximum branch length (time interval) over which the current tip-character was maintained across the whole phylogeny. Finally, last transition time is the sum of branch lengths from the tip to the prior/previous node with a reconstructed character equal to current tip-character (Fig. 1). To calculate the three tip-based metrics, we mapped and estimated ancestral states using stochastic mapping of discrete traits via Bayesian inference (Bollback 2006), which allows calculating the time at which a trait changed along phylogeny branches and not just at the nodes. Tip-based metrics such as TR, ST and LT can be later summarized across assemblages, allowing assessments of the effect of spatial, environmental, and historical factors on trait evolution rates. Here, we averaged tip-based metrics across all species occurring in a given assemblage to obtain assemblage-level TR, ST, and LT –hereafter aTR, aST, and aLT– to then test whether evolution has been faster at ecotones. The test involved a thorough consideration of phylogenetic uncertainty from character reconstruction to hypothesis testing (Fig. 1).
We tested our hypothesis of faster trait evolution at ecotones by integrating data on distribution, diet, and phylogeny of sigmodontine rodents. Sigmodontinae is a subfamily within the family Cricetidae (Musser and Carleton 2005) that arrived in South America before the final closure of Panama Isthmus (~10 Ma; Leite et al. 2014; Steppan and Schenk 2017). They are a useful group for testing our hypothesis because the species are sensitive to habitat stability at fine scales due to their small body-size, short generation time, small geographic range, and narrow microhabitat requirements for feeding, reproducing, and avoiding predation (Patton et al. 2015). One notable aspect about sigmondontine rodents is the uncertain phylogenetic relationships among species, genera and tribes. Different phylogenies show equally plausible but differing topologies (e.g., Weksler 2006; Machado et al. 2013; Leite et al. 2014; Steppan and Schenk 2017), suggesting that reliance on only one phylogeny may be insufficient to understand the evolution of the group (Rangel et al. 2015; Upham et al. 2019). Another remarkable aspect about them is that, since their colonization of South America, they experienced a rapid evolutionary radiation that has allowed sigmodontine species to spread into many types of habitats (Patton et al. 2015), without radical changes from their ancestral morphology (Maestri et al. 2017). However, sigmodontine rodents do show an impressive variation in diet (Paglia et al. 2012; Missagia et al. 2019). Many species are highly specialized to consume a few items from specific habitats, such as herbs and seeds from open habitats or leaves and fruits from forested habitats (Paglia et al. 2012; Pardiñas et al. 2020). Thus, diet should evolve as a response to the spatial heterogeneity and temporal instability of ecotones.