INTRODUCTION
The Neotropics are home to some of the world’s most diverse mammal faunas. A recent estimate is that the region houses 1617 recognized species (Burgin et al. 2018) or roughly one-quarter of the world’s total (mammaldiversity.org). Some represent truly ancient clades, including the earliest-diverging extant lineages of both marsupials (Didelphimorphia and Paucituber culata) and placentals (Cingulata and Pilosa) (see Fig. 1; Meredith et al. 2011; Esselstyn et al. 2017; Upham et al. 2019). Others represent comparatively recent but explosive radiations, such as the Neotropical shrews (He et al. 2015) and the sigmodontine rodents (Steppan et al. 2004; Parada et al. 2013; Maestri et al. 2019). Fortunately for our science, many Neotropical groups have truly eloquent fossil records (Lemon & Churcher 1961; Kay et al. 1997; Madden et al. 2010; Carrillo et al. 2015). Fos sils allow the disentanglement of diversity (as a state) and diversification (as a process or rate), because diversity in a region increases both with rising diversification rates (via spe ciation) as well as by lowered extinction rates (Rabosky 2009). These rate shifts have different causal agents, although they may have equiva lent effects on diversification rate (Rabosky & Lovette 2008). For example, originations in ferns appear to depend more on within-group diversity rather than on environmental changes, whereas extinctions are strongly af fected by external factors such as climate and geology. Thus, environmentally driven extinction is a prime driver of fern diversity dynamics, with origination rate representing a dynamic response to variable ecospace occupancy (Lehtonen et al. 2017).
Mammals resemble other vertebrates in having spatial patterns of species richness that are strongly affected by the interaction of climate and vegetation. At larger spatial and temporal scales, climate effects are more evident in the species richness patterns of good dispersers, such as bats (Willig & Bloch 2006; Stevens 2011), or older radiations, such as the cavio morph rodents (Maestri & Patterson 2016). In contrast, less vagile groups have ranges that are limited by the continuity of their preferred habitat (Kisel & Barraclough 2010). Indeed, much of the variation in vegetation that ultimately determines vertebrate richness is climatically structured, supporting the productivity hy pothesis (Dunn et al. 2015; Moura et al. 2016). Diversification needs qualification, as many biologists reject the notion that lineage splitting to form daughter species is the only noteworthy evolutionary event. To taxonomic diversity (among vertebrates generally estimated via species richness), ecologists now routinely add phylogenetic diversity (assessed by phylogenetic branch lengths) and functional diversity, which gauges differences in ecological roles (Cisneros et al. 2014; Stevens & Gavilanez 2015; Cernansky 2017; González-Maya et al. 2017). These scale-dependent facets of diversity vary independently—functional diversity typically reaches an asymptote as species diversity con tinues to rise, so that species-rich areas often support many species that are functionally redundant (Oliveira et al. 2016; Jarzyna & Jetz 2018). Disparity, a measure of morphospace occupation, is a related concept, and allometry (Álvarez et al. 2015), heterochrony (Wilson & Sanchez-Villagra 2009), and modularity (Marroig et al. 2009; Porto et al. 2009; Marroig et al. 2012) all constitute developmental drivers of disparity. In this perspective, I will focus chiefly on extrinsic drivers of diversification.
DIVERGENCE
At its simplest, divergence begins whenever genetic continuity is interrupted, and divergence is an essential feature of diversification. The role of geographic isolation in divergence and speciation is now well understood (Mayr 1963), but the roles of other factors, including environmental gradients (Patton et al. 1990; Smith et al. 2001), sexual selection (Brindle & Opie 2016; Orr & Brennan 2016), and saltational chromosomal rearrangements (Ortells 1995; Patton 2004; Borodin et al. 2006; Ojeda et al. 2015) are more variable and dimly per ceived. Phylogenetic niche conservatism, or the tendency of species to retain ancestral niche traits, may be a general principle (Wiens 2004). However, a recent analysis of Neotropical canids demonstrated that when ecological processes drive species divergence, even closely related species may have distinctive climatic tolerances (Zurano et al. 2017).
Early studies of allopatric speciation in Amazonia, dating from Alfred Russel Wallace’s 1848 trip there, emphasized the roles of South America’s mighty rivers in isolating populations on opposing banks (Hershkovitz 1987; Bonvicino & Weksler 2012). The “riverine barrier” hypothesis, postulating greater divergence downstream from the headwaters, has been elegantly addressed for various mammals along Brazil’s Rio Juruá (Gascon et al. 2000; Patton et al. 2000), evincing stronger evidence for geomorphic barriers to gene flow than for riverine effects. However, the role of rivers in limiting terrestrial mammal distributions, effectively confining them to interfluves, has been convincingly demonstrated for primates (Hershkovitz 1977; Ayres & Clutton-Brock 1992; van Roosmalen et al. 2000) and undoubt- edly applies to other less-studied groups. Some newly described pygmy anteater species appar ently arose via fluvatile vicariance (Miranda et al. 2017), which has also played important roles in the diversification of dryland mammals, such as Thylamys (Giarla & Jansa 2014), Dasypus (Feijó & Cordeiro-Estrela 2016), and Thrichomys (Nascimento et al. 2013).
Rivers as principal drivers of Amazonian and other Neotropical diversity were challenged by Pleistocene vicariance or “refuge theory,” which hypothesized a diversity pump resulting from range contraction and genetic divergence in forest refuges during cool, dry episodes of the Pleistocene and subsequent reexpansions dur ing warm, wet interglacial times (Haffer 1969; Vanzolini & Williams 1970). Climatic oscillations of the Pleistocene have been overwhelmingly important to mammalian diversification, as shown by molecular phylogenies for many groups: e.g., Carollia perspicillata (Pavan et al. 2011); Brazilian Alouatta (Martins et al. 2011); Pteronotus parnellii (Clare et al. 2013); and various Lycalopex species (Tchaicka et al. 2016). Nevertheless, time-trees show that many speciation events predated the Pleistocene (e.g., Upham 2014), and at least some paleoclimatic reconstructions suggest that forests weren’t hugely reshuffled during glacial cycles (Colinvaux 2007). Pleistocene glacial episodes also emptied oceans, exposing continental shelves and facilitating island colonization. Noctilio leporinus was apparently recently derived from N. albiventris, and its colonization of Caribbeanislands depended on glacial-aged exposures (Pavan et al. 2013), as did that of Molossus molossus (Loureiro 2019). Ancestors of the Sturniraspecies S. angeli and S. paulsoni reached remote Antillean islands during the Pleistocene where they diverged by the same mechanism (Velazco & Patterson 2013). Glaciation fostered divergence by permitting the colonization of remote islands, but when glaciers melted and sea-levels rose, reductions in the area of islandsand effective population sizes could lead to divergence or local area-dependent extinctions. Curiously, this globally recognized mechanism (Case et al. 2002) appears not to explain the case of Holocene Antillean bat extinctions (Soto-Centeno & Steadman 2015; see Dávalos and Russell 2012).
The Andes constitute the world’s longest terrestrial mountain chain and present extended elevational gradients (Fig. 2). Its sections have arisen at different times, sundering lowland populations on either side and creating dispersal corridors along its slopes. Patterson et al. (2012) reviewed the roles of Andean orogeny on mammalian diversification, finding that speciation has been recent and rapid on the Eastern Versant and apparently slower and more relictual on the Western Slope. The fun damental distinction of cis-and trans-Andean lineages (i.e., lowland lineages to the east and west, respectively, of the Andes’ north-south axis) can be attributed to orogenic vicariance, and the Andes played a critical role in the di versification of Vespertilionidae (López-Aguirre et al. 2018). Yet middle-elevation faunas remain poorly known, highly diverse, and substantially endemic (Voss 2003), and species found there may be critical to understanding the diversification patterns of widespread Neotropical groups (Patterson & Velazco 2008). Parapatric speciation also needs further investigation, despite its rejection for Peruvian Akodon (Patton et al. 1990). Poison-dart frog (Oophaga) sister taxa are known to neatly replace one another along elevational gradients (Posso-Terranova & Andres 2016), but few mammals have been assessed in this light. Finally, ecotones or transi tion zones, located at the boundaries between biogeographic regions, deserve special attention because they represent areas of intense biotic interactions (Morrone 2010); in such circumstances, lineages may undergo introgression or character displacement and/or reinforcement. Complex topography and high habitat turnover can act jointly to limit gene flow.
DISPERSAL VERSUS VICARIANCE
Since the ascension of phylogenetic systematics in the 1980s, vicariance has been a preferred explanation for divergence. When multiple lineages show spatially and temporally congru ent patterns of differentiation, vicariance offers a more parsimonious explanation of diversity than does dispersal. Yet few studies employ general-area cladograms, and dispersal-driven divergences are equally parsimonious for individual clades. Smith et al. (2014) argued that because most species-level diversity in birds postdated episodes of major Andean uplift, dispersal and differentiation on a pre-existing environmental matrix was a major driver of avian speciation in lowland rainforests. More recently, Crouch et al. (2018) attributed South American passerine radiations to dispersal limits shaped by geophysical and biotic features of the landscape. Highly vagile mammal groups also evince the effects of dispersal: within-area events, not vicariance of pre-existing distributions, provided the principal mode of speciation for New World emballonurid bats (Lim 2008), and noctilionoid bats appear to show the same pattern (Rojas et al. 2016).
Antonelli et al. (2018) compiled the timing and origin of species from six major clades of plants and vertebrates. They found that about half of all events involved transitions between major environmental types, predominantly from forested to open biomes. For all taxonomic groups, Amazonia was the primary source of Neotropical diversity (see Upham et al. 2013 for evidence that the Andes have also played a ma jor role). Dispersal is integral to the exploration of ecological opportunity and is a dominant feature of community assembly. Maestri and colleagues (2019) investigated spatial patterns of evolutionary relatedness and diversification rates to address the historical biogeography of Sigmodontinae. They found a negative correlation between mean phylogenetic distance and diversification rates, meaning assemblages of closely related species also contain the fastest diversifying ones. Subregions of the Neotropics where mice are on average more slowly diversi fying include Central America, northern South America, and the Atlantic forest, whereas recent species turnover appears to have been higher in temperate South America. Each dispersal event both transformed the lineage and presented the group with new ecological opportunities (see also Dias & Perini 2018).
ECOLOGICAL OPPORTUNITY
All adaptively driven diversification must hinge on ecological opportunity (Schluter 2000)—that is, vacant niches or unexploited resources that are accessible to the focal lineage and in turn trigger positive diversification rates. Proximately, those opportunities may arise from colonization of a new area, the acquisition of a novel trait, the appearance of a new resource, or even the extinction of an unrelated but codistributed lineage (Simpson 1953; Stroud & Losos 2016). These are considered here seriatim. Certainly the colonization of South America, which was practically a huge island for much of the Cenozoic, represented a continent-sized opportunity for those groups that managed to colonize it. Marsupials arrived from North America across an ephemeral island arc in the Late Cretaceous, and caviomorph rodents and platyrrhine monkeys arrived from Africa during the Paleogene (Bond et al. 2015; Upham & Patterson 2015; Goin et al. 2016). Noctilionoid bats probably colonized from North America in the Oligocene (Rojas et al. 2016), and sigmodontine rodents surely did so at the dawning of the Great American Biotic Interchange (Pardiñas 1999). Each subsequently underwent exten sive radiations in South America. In keeping with Simpson’s notions of adaptive radiations (Simpson 1944), analyses of platyrrhines and sigmodontines have suggested an “early burst” of variation followed by a decelerating rate of diversification (Schenk et al. 2013; Parada et al. 2015; Rocatti et al. 2017), although evidence for adaptation in the case of sigmodontines is equivocal (Maestri et al. 2017). Among noctilionoids, the highest rates of speciation coincided with the appearance of the stenodermatine bats ~18 Ma (Fig. 3; Rojas et al. 2016). On the other hand, the didelphid marsupial radiation was marked by a long mid-Miocene interval of zero net lineage accumulation, implicating a mass extinction event (Jansa et al. 2014).
Novel traits can also admit organisms to eco logical opportunity. Rodents employ a unique mastication system in which the cheekteeth do not occlude when the incisors are engaged in gnawing, and viceversa in cases of chewing (Cox & Hautier 2015). Krentzel & Angielczyk (2018) demonstrated that incisor size and shape are unrelated to cheekteeth size and shape across rodent phylogeny, corroborating the functional independence of these feeding modules. Myomorphy is a versatile combination of the functional elements in both gnawing- adapted sciuromorphy and chewing-adapted hystricomorphy. Acquisition of myomorphy is associated with explosive and contemporaneous rodent radiations worldwide (e.g., Steppan & Schenk 2017). Certainly, 20 million years of prior incumbency by the diverse hystricognathous caviomorphs (Fig. 4) did not eclipse opportunities for the >85 genera and >400 species of myomorphous sigmodontines that now share their Neotropical landscapes. Ecological opportunities also triggered the adaptive diversification of phyllostomid bats via dietary specialization, spurring the origination of novel feeding habits (Rossoni et al. 2017).
The appearance of a new food resource may also constitute a trigger for diversification. Classic examples include the concerted rise of angiosperms and herbivorous beetles and of grasses and mammalian hypsodonty (but see Madden et al. 2010 on the role of volcanic grit in South America). Any trophic specialist offers the potential for evolutionary radiations in synchrony with its host plants. Sturnira fruit bats with their Solanum fruits and Bradypus sloths with their Cecropia trees offer plausible models for resource-driven divergences (Moraes Barros et al. 2011; Velazco & Patterson 2013; Schetino et al. 2017). Lobato et al. (2014) traced the evolution of species-rich clades of reef fishes to their switch to low-quality foods, which is consistent with an opportunity-limited (density-dependent) model of diversification.
Finally, the extinction of one group is thought to open the gates for another’s diversification, as implied by the Cenozoic radiation of mammals in the wake of the K/Pg archosaur extinction (O’Leary et al. 2013; Lyson et al. 2019). Halliday et al. (2019) documented the unequal rates of molecular and morphological diversification of crown eutherian lineages on either side of that geological boundary. Neo tropical radiations spurred by extinction were thought to include the endemic metatherian carnivore group Sparassodonta (Goin et al. 2016), but that group disappeared long before the Pliocene arrival of putative North American replacements in the form of placental carnivores (Eizirik 2012). The cataclysmic extinctions accompanying the arrival of humans around the Pleistocene-Holocene boundary led to numerous extinctions of large-bodied mammals, including glyptodonts, ground sloths, notoungulates, and gomphotheres, whose ecological roles remain at least partly unfilled (Janzen & Martin 1982; Howe 1985).
LOOKING FORWARD
There is so much more to learn. Ever since graduate school, I have been fascinated by the processes driving diversification in montane systems (e.g., Patterson 1981), but many ele ments that are operational in those contexts are more general. Genomic and transcriptomic data will surely permit new insights and greater resolution (Lessa et al. 2014), especially regard ing the various avenues taken by different clades (e.g., Sadier et al. 2018). Phylogenomic and demographic analyses now offer tests on whether Andean orogeny and geographic isolation were the principal drivers of diversification and the extent to which gene flow was involved (e.g., Nevado et al. 2018). How much gene flow links the lower and upper reaches of species distrib uted along Andean slopes? Xing & Ree (2017) found evidence in the Hengduan Mountains of Asia that tectonic uplift creates environmental conditions (new habitats, dispersal barriers, etc.) that increase the rate at which residents speciate. Is the episodic orogenic history of the Northern, Central, and Southern Andes reflected in the radiations of thomasomyines, akodontines, and abrotrichines in those sectors? These are the sorts of questions that I continue to ponder, but some others might be listed here for more ambitious readers:
• Are there geographic patterns in the relative contributions of in situ diversification vs. immigration from neighboring regions to the composition of Neotropical faunas (Liu et al. 2016)?
• What regions serve as “cradles of diversity” and where are the “museums” located? Do high elevations and high latitudes serve as centers of lineage diversification, and are the lowlands really museums (cf. Vijaya-Kumar et al. 2016)?
• Are there meaningful differences between open and closed habitats in speciation, extinction, and dispersal rates (Pinto- Ledezma et al. 2017)? Is there an interaction of habitat with latitude (Lessa et al. 2010)?
• Within diversified clades, is there a positive correlation between phylogenetic distance and spatial co-occurrence (cf. Chaves et al. 2013)?
• Do non-physical barriers to gene flow play a role in driving the divergence of mam mals (e.g., Taylor et al. 2018)?
• Does climatic niche divergence serve as a driver of diversification in homeotherms (see Matuszak et al. 2016)?
• Do lineage-specific life-history traits, such as rapid evolutionary shifts in dispersal ability (Moyle et al. 2009), play a role in mammalian diversification?
• What spatial, temporal, and biological features trigger rapid lineage turnover?
Certainly, new insights can be expected from the continuing discovery of Neotropical mammals and documentation of their genetic, morphological, ecological, behavioral, and physiological diversity.