Discussion
The Origin of Neotropical Biodiversity. All evidence generated in our analyses suggests that Neotropical biodiversity attained outstandingly high levels prior to the Quaternary. The age of the MRCA of most clades analyzed here (146 clades out of 150, 97%) largely predates the Quaternary (Fig. 1 ; mean age of 30.4 Myrs), supporting the results of (Hoorn et al. 2010). The MRCA of clades and genera, however, do not inform on the diversification of their descendant species (Rull 2011b). This is instead informed by the diversification rates calculated here, which for most Neotropical lineages have remained constant through time (75 clades, 50%; 2903 species, 23%) or were higher in the past and diversification decreased toward the present (42 clades, 28%; 4486 species, 36%) (Fig. 3 ). This result suggests that the total number of species in the Neotropics was probably as high (or higher) before the Quaternary than today, a result previously only reported from the fossil record (Jaramillo et al. 2006; Salas-Gismondi et al. 2015).
Diversification Drivers for Tetrapods. A substantial number of Neotropical tetrapod clades studied here (44%) evolved under a constant rate during the Cenozoic (Fig. 2 ), although this proportion is smaller when the number of species is accounted for (1,204 species, 19%). Phylogenies best fitting constant models in our study tend to be smaller and younger than phylogenies fitting other models, and as so, we find a higher proportion of bird and mammal clades’ diversification dynamics (>50%) explained by constant-rate models than amphibian and squamates clades (<30%,Fig. 1 and SI Appendix , Table S7 and Fig. S2). This could be partly explained by the observed differences in clade ages between these groups, with younger clades having, arguably, less time to being exposed to geological and environmental changes during their history (Smith et al. 2014), but also by our reduced power to detect rate variation as the number of species decreases (Davis et al. 2013). The remaining tetrapod diversification is generally best explained by speciation rates decreasing through time and extinction remaining constant (Fig. 3 and SI Appendix , Table S9), a pattern that is especially manifest in endotherms, and among them in birds. Regardless of their ecology or distribution, all bird clades exhibit diversification slowdowns (Fig. 3 and SI Appendix , Tables S3 and S9).
The pattern of decreasing diversification in Neotropical endotherms is consistent with previous studies suggesting a general trend for slowdowns in speciation at continental scales and across taxonomic groups (McPeek 2008; Phillimore & Price 2008; Morlon et al.2010; Luzuriaga-Aveiga & Weir 2019), which has been often interpreted as the signal of ecological limits on the number of species within a clade (Rabosky 2009). Among the phylogenies supporting diversification slowdowns here, time-dependent models explain 8% of all tetrapod phylogenies, but the proportion is highest for mammals (25%,Fig. 2 and SI Appendix , Table S10 and Figs S4-5). Time-dependent models with decreasing speciation have been suggested to be a good approximation of diversity-dependent diversification (Raboskyet al. 2014). Our results lend support to an alternative explanation for diversification slowdowns: the idea that clades fail to keep pace with a changing environment (Moen & Morlon 2014; Condamineet al. 2019). Our temperature-dependent models explain a substantial proportion of phylogenies supporting diversification slowdowns in our study (17% of mammal and 38% of birds; Fig. 4 and SI Appendix , Figs S6-7). The positive correlation between diversification and past temperature indicates that these groups diversified more during periods of global warming such as the greenhouse Eocene or the MMCO, and speciation decreased during colder periods (such as the Eocene-Oligocene transition and the late Miocene onwards).
Temperatures can influence diversification in different ways. According to the Metabolic Theory of Biodiversity, high temperatures can increase enzymatic activity, generation times and mutation rates (Gilloolyet al. 2001), which may in turn affect diversification positively, and conversely (Allen et al. 2006; Condamine et al. 2019). Climate cooling could also decrease global productivity, resource availability, and population sizes (Mayhew et al. 2012) or even species interactions (Jaramillo & Cárdenas 2013; Chomickiet al. 2019). Reduction of the tropical forest biome (Jaramillo 2019) could have also contributed to this pattern of decreasing diversification in association with climate changes, caused by decreased precipitation in the Neotropics during Cenozoic cooling (Silva et al. 2019). Only one group of endotherms, the New World monkeys (Platyrrhini), had increased diversification as temperature dropped. This result could reflect the role of Quaternary events on primate speciation (Rull 2011a), or could, however, be an artifact of taxonomy over splitting species in this clade (Springer et al. 2012).
Compared with endotherms, ectotherm tetrapods show a mixed diversification trend (Fig. 3 and SI Appendix , Table S8). We find a similar fraction of amphibians decreasing and increasing diversification rates through time (668 species and 5 phylogeniesvs. 823 species and 6 phylogenies, respectively). In squamates, the diversification of most clades has decreased, but for a non-negligible proportion it increased (679 species and 13 phylogeniesvs. 353 species and 4 phylogenies). Contrasting with endotherms and plants (see below), the triggers of rate variations in ectotherm diversification are mostly associated with the Andean orogeny (50% of amphibian clades and 1,166 species; 33% of squamate clades and 663 species) rather than with global temperature changes or time (Figs. 2, 5 and SI Appendix , Figs. S6 and S7 and Tables S6 and S9). This is in agreement with the dominant view on Neotropical ectotherm diversification (Santos et al. 2009; Hutter et al. 2013, 2017; Esquerré et al. 2019).
The Andes host twice the amphibian richness as the entire Amazonian lowland rain forest, with more endemic frog species than any other region in the world (Hutter et al. 2017). They also comprise the third most endemic squamate fauna, after the Caribbean Islands and Madagascar (Mittermeier et al. 2011). Interestingly, we find that diversification in many lineages (i.e. , 649 amphibian and 353 squamate species) correlated positively with the Andean uplift, such that cladogenesis progressively increased with the orogeny (Figs. 2, 5 ). Some of these clades show an Andean-centered distribution, such as Liolaemidae and Tropiduridae lizards or Centrolenidae frogs, but others are predominately distributed outside the Andes such as Leptodactylidae frogs or Hoplocercidae lizards (SI Appendix , Tables S4-5). Sustained diversification in the context of Andean orogeny into and out of the Andean region could be explained by increasing thermal and environmental gradients from equatorial areas to Patagonia or in a west-east gradient, as it has been suggested to affect Leptodactylidae and other frogs (Fouquet et al. 2014; Moen & Wiens 2017). Other possible correlates include changes in elevational distributions of lineages and concomitant shifts in climatic regimes (Kozak & Wiens 2010; Hutter et al. 2017), recurrent migrations from (and within) the Andes into other regions, and allopatric fragmentation (Santos et al. 2009; Esquerré et al. 2019). The explosive generation of new land area that occurs with mountain building could also explain this pattern (Elsen & Tingley 2015; Antonelli et al. 2018b), e.g. , each small ravine down the mountain has a microtopography with a variety of slopes in multiple directions.
In a significant proportion of ectotherm clades, however, we also detected that diversification was elevated only during the early stages of the orogeny and then decreased with progressive uplift (i.e.,negative correlation between diversification and Andean orogeny for 472 amphibian and 310 squamate species; Fig. 5 and SI Appendix , Figs. S6 and S7), in agreement with diversification slowdowns detected in recent studies (Santos et al. 2009). These include lineages that are diverse in the Andes, such as dendrobatid frogs, but also non-Andean lineages, such as Odontophrynidae frogs or Leiosauridae and Xantusiidae lizards, which abound in the Cerrado, Chaco, temperate South America, and Mesoamerica.
Andean uplift started in the late Eocene with the formation of moderate elevation uplands (1,000–1,500 m) in a non-continuous belt, and then accelerated in the late Miocene, with the majority of the Tropical Andes reaching its modern elevation ~4.5 Mya (Garzioneet al. 2008; Hoorn et al. 2010; Chen et al. 2019). Initial Andean uplift might have stimulated diversification in the lowland transition zone, with new ecological opportunities in tropical-like habitats formed at moderate elevation and increased rates of geographical isolation for species with cross-Andean distributions (Santos et al. 2009). Post-Miocene uplifts, however, built a major ecological and geological barrier for biotic dispersals of many groups, with strong physiological constraints limiting adaptations to new upland environments, or dispersal across unsuitable habitats (Santoset al. 2009; Olalla‐Tárraga et al. 2011; Hutter et al. 2013; Pie et al. 2017). Taken together, these results suggest that elevation of the Andes impacted ectotherm diversification at the continental scale.
Diversification Drivers for Plants. In contrast to tetrapods, the diversification of a significant fraction of Neotropical plants shows an expanding trend toward the present (4,019 species, 64%; 22 clades, 33%), either due to increasing speciation, decreasing extinction, or both (Fig. 3 and SI Appendix , Table S9). Another substantial proportion of plants evolved under a constant rate model (1,699 species, 27%; 38 phylogenies, 58%). Only a small proportion of Neotropical plant diversity experienced slowdowns of diversification (537 species, 8.6%; 6 phylogenies, 9%). Changes in plant diversification are also mostly associated with paleotemperatures (16 clades; 24%) and with time alone (11 clades; 16%), while the effect of the Andes is negligible (1 clade, Gesnerioideae) (Figs. 2, 4 and SI Appendix , Figs S6 and S7). This result is surprising given that Andean uplift has often been considered the main driver behind the radiation of Neotropical plants, especially in the Páramo, but also in other regions (Hughes & Eastwood 2006; Antonelli et al. 2009; Drummond et al. 2012; Luebert & Weigend 2014; Lagomarsino et al. 2016; Pérez-Escobar et al. 2017; Bacon et al. 2018; Pouchon et al. 2018). Lineages including species distributed in the Páramo are not well represented in this study, but the few included here – Lupinus(Fabaceae), centropogonids (Campanulaceae), Pleurothalis andStelis (Orchidaceae), Solanum, Cestrum , and Sessea(Solanaceae) – do not follow an uplift model. This result contrasts with a previous study for Cymbidieae orchids supporting uplift-dependent diversification (Pérez-Escobar et al. 2017), while diversification in this clade is best explained by temperature-dependent models in our study (SI Appendix , Table S9). Results are however not directly comparable as ref. (Pérez-Escobar et al. 2017) only evaluated uplift models. Similarly, centropogonids diversification is best explained by temperature-dependent models in ref. (Lagomarsinoet al. 2016), but by time-dependent models in our study (SI Appendix , Table S9). Time-dependent models are not evaluated in previous studies (Lagomarsino et al. 2016; Pérez-Escobar et al. 2017), though these models probably represent more realistic null hypotheses than constant-rate scenarios, especially when half of the diversity in our study is found to have changed diversification rates through time (Fig. 2 ).
Among the plant clades increasing diversification rates through time, in 10 phylogenies (15% of the clades; 2,180 species, 35%), we found a negative correlation between diversification and temperature changes, indicating that these groups diversified more during cold periods and due to increasing speciation or decreasing extinction (Figs. 3, 4 and SI Appendix , Figs. S4-6). In the other 11 clades (1251 species, 20%), diversification increased as a function of time alone (SI Appendix , Table S10), and generally due to decreases of extinction while speciation remained constant (SIAppendix , Table S9).
Plant phylogenies increasing diversification rates through time mostly correspond to clades distributed in the Andes, Chocó and Central America – also termed “Andean- centered” groups (Gentry 1982) – from lowlands to highlands, such as Bactris palms (Arecaceae),Fuchsia (Onagraceae), or Cymbidieae orchids, among others (SI Appendix , Table S9). Many species within Andean-centered groups were likely able to adapt to the new conditions that increasingly appeared in the mountain foothills as the Andes uplifted and global temperatures dropped (Luebert & Weigend 2014; Antonelli et al.2018b). It has been proposed that the first Quaternary glaciation could have acted as a major evolutionary bottleneck, whereby many warm‐adapted lineages succumbed, while those that survived could have diversified and better cope with subsequent climatic oscillations (Silva et al.2018). Cenozoic climate cooling could also have created a “biotic corridor” for pre-adapted lineages to montane conditions to increase their range and colonize new montane environments (Antonelli et al. 2009; Pérez-Escobar et al. 2017; Meseguer et al.2018). Many iconic high-elevation Andean clades (e.g. Espeletia ) had major radiations within the past 2.5 Myrs (Madriñán et al.2013; Pouchon et al. 2018), about 2 Myrs after the Andes had reached their modern elevation, suggesting that the onset of the Quaternary climate could have played a much stronger role in increasing speciation rates than the generation of high topography in itself. In addition, some of these clades represent textbook examples of ongoing explosive radiations; e.g. centropogonids (Lagomarsino et al. 2016), Lupinus (Hughes & Eastwood 2006; Drummond et al. 2012), and Inga (Richardson et al. 2001). Their diversification has been previously associated with biotic drivers, such as species interactions (Kursar et al. 2009), the evolution of key adaptations (Drummond et al. 2012), or pollination syndromes (Lagomarsino et al. 2016). Although we have not explicitly tested the impact of intrinsic biological traits in the generation of this diversity, these results add support to the role of climate and biotic factors as not mutually exclusive drivers of macroevolutionary changes in the Neotropics, with the rise of the Andes acting mostly indirectly by providing the necessary conditions for species to expand and diversify in new climatic regimes.
In contrast, we detected six clades (537 species) that have decreased diversification, all associated with global cooling (includingSideroxylon [Sapotaceae], Guatteria[Annonaceae], Myrcia [Myrtaceae], among others). These clades are mostly distributed in lowland regions of Amazonia, the Brazilian and Guiana Shields – the Amazonian-centered elements in Gentry’s sense (Gentry 1982) (SI Appendix , Table S9). These groups diversified more during warm periods and climate cooling negatively impacted their diversification, a result consistent with paleobotanical studies showing a positive correlation between Neotropical plant paleodiversity and past temperatures (Jaramilloet al. 2006). Slowdowns of diversification were mainly due to decreases in speciation (Fig. 4 and SI Appendix , Figs. S5 and S6). Such pattern could be explained by decreases in primary productivity, but also by the expansion of several biomes, including Páramo, cloud forests, savannas, and dry/xerophytic forest at the expense of the reduction of the rainforest during the late Neogene (Jaramillo 2019).
Conclusion. Environmental perturbations have long been recognized as fundamental for regulating diversity, although progress toward understanding how has been slow. Here, we demonstrate that diversification for a significant fraction of Neotropical clades correlates with deep-time environmental trends, especially with temperature changes and to a lesser extent with Andean uplift. The effect of these environmental perturbations extends to the continental scale, modulating the pace of Neotropical diversification across organisms and biomes. Yet, the specific mechanisms by which they impact diversification are clade-dependent and remain to be understood. The other fraction of Neotropical diversity evolved at constant rates or is associated with different extrinsic/intrinsic factors.
Our results have implications for discussing the future of biodiversity in the context of current environmental changes, and on how it might recover from human-induced extinction. As global change accelerates, ecosystems will face an increasing rate of perturbations, e.g.temperatures increase, drought or habitat loss, with current deforestation rates staggering across all Neotropical biomes. If one quarter of the Neotropical diversity in our study (~3000 species) follows a constant diversification mode as in the past, it may take tens of millions of years for biodiversity to reach its pre-extinction level. For most of the remaining Neotropical diversity, but especially tetrapods and Amazonian plants, past climate cooling triggered coordinated slowdowns of their speciation rates, suggesting that the pace of diversification in the world insignia of biodiversity, the Neotropics, has been in deceleration. Whilst this study found that ancient climate warming triggered increases in diversification on these lineages, this relationship must not be extended to the present, as the pace of current environmental changes is the fastest in geological history and acting in synergy with multiple biotic stressors lacking past equivalents (Condamine et al. 2013).