Conceptual and empirical advances in Neotropical biodiversity research

Taxonomic diversity

Relatively good estimates of taxonomic diversity are only available for well-studied Neotropical taxa, as in other parts of the world. These estimates have been used to identify the best predictors of diversity at large scales (Jenkins et al., 2015; Moura et al., 2016). Although sampling across taxa is comparable or even greater in the Neotropics than in other tropical regions (Fig. 2; Table 1), taxonomic diversity is generally underestimated within the Neotropics, especially for poorly sampled organisms such as fungi, invertebrates, and micro-organisms.

Several examples of species-rich, yet incompletely-documented faunas are available, including large clades of freshwater fishes, amphibians, and some groups of reptiles. Although about 5,600 species of freshwater fishes are currently known in the Amazon, the Orinoco, and adjacent river basins of tropical South and Central America, more than 100 new species are described every year (Van Der Sleen & Albert, 2017). In other words, approximately two new species are described per week, although an even higher number of new species would be expected if more trained taxonomists were available. This rapid pace of species description is not slowing down, and recent estimates for the total number of Neotropical freshwater fishes exceed 8,000 species (Reis et al., 2016). This estimate is remarkable, implying that more than 2,400 fish species might remain to be described in the Neotropics alone, a number that exceeds the combined number of rodent species currently known on Earth. This large number of expected, but still hidden, lineages represents an example of the “known unknowns” of Neotropical biodiversity (Table 5), which may have different underlying explanations (such as lack of taxonomic training, uneven distribution of resources across taxa and habitats, morphological complexity, among others).

Current knowledge of taxonomic limits of Neotropical amphibians and reptiles is gradually improving. Several molecular studies have detected high levels of cryptic diversity, that is, the existence of two or more lineages within a known species (Bickford et al., 2007), indicating that the known taxonomic diversity is still underestimated in many orders (Funk, Caminer & Ron, 2011; Fouquet et al., 2012). Even in the much more densely sampled and well-studied Atlantic Rainforest of Brazil, charismatic species of frogs are still being discovered. For instance, seven new species of Brachycephalus were recently described for this region (Ribeiro et al., 2015). Likewise, intraspecific analyses of Neotropical lizards show that the occurrence of cryptic diversity is often manifested across biomes. This subdivision of broadly distributed taxa into multiple cryptic species with more restricted geographic distributions increases the perception of biological diversity of a given region, and has numerous implications for biogeography (Werneck et al., 2012b) and conservation (Simões et al., 2014).

For plants, a quantitative assessment on the discovery of Amazonian trees during the last three centuries was compiled by Ter Steege et al. (2016), showing clear peaks in herbarium collections and new species descriptions. Although the data show a drop in the collection of unknown taxa after the 1980s (Ter Steege et al., 2016), there are still enormous discoveries to be made. For example, in a few years of increased collection efforts, the Guide of the Ducke Reserve (Brazil; Ribeiro, 1999), which covers one of the most thoroughly studied areas of Amazonian forest, increased the number of known vascular plants from 825 (Prance, 1990) to 2,079 (Hopkins, 2005).

One difficulty in assessing taxonomic diversity is that taxonomic units may vary according to the preference of the taxonomist revising a particular group (e.g., whether a “splitter” or a “lumper”), and by the data and methodologies underlying taxonomic revisions and species circumscriptions. This issue becomes obvious when taxonomic treatments of the same group are produced by different researchers independently. For example, the Neotropical palm genus Attalea included 29 species in one monograph (Henderson, Galeano & Bernal, 1995), and 65 species in another taxonomic treatment published just 4 years later (Glassman, 1999). Similarly, the Caribbean palm genus Coccothrinax included 14 species in one taxonomic treatment (Henderson, Galeano & Bernal, 1997) and 53 species in another (Dransfield et al., 2008). Personal preferences to “lumping” vs. “splitting” among taxonomists may have large consequences for biodiversity estimates and have been shown to strongly affect diversification rate estimates (Faurby, Eiserhardt & Svenning, 2016). Taxonomic practices should therefore be considered when comparing taxonomic biodiversity at any scale, and whenever adequate, researchers should take advantage of explicit and reproducible criteria for species delimitations.

Besides lumping vs. splitting, species lists may vary among authorities depending on the inclusion criteria, such as whether or not to include rare occurrences of a species common elsewhere, and how to classify the life forms of species (e.g., primarily herbaceous plants rarely recorded as trees). For large regions such as Amazonia, these are some of the reasons why the number and contents of species lists may differ substantially (Cardoso et al., 2017; H. Ter Steege et al., 2018, unpublished data).

Phylogenetic diversity

Many Neotropical clades are known from just one or a few species that may represent relictual survivors of ancient and otherwise extinct groups (Table 2). This phenomenon is known from most organism groups, from Neotropical fishes (Albert et al., 2011) to plants (Wilson et al., 2012). To study how differences in diversity arise among taxa, some researchers have turned their attention to the study of early-branching, low-diversity clades. Examples of such clades include the leafy cacti (Pereskia and Leuenbergeria spp.; Cactaceae), the South American lungfish (Lepidosiren paradoxa; Lepidosirenidae), the hoatzin (Opisthocomus hoazin; Opisthocomidae), and the coral pipe snake (Anilius scytale; Aniliidae). In contrast, other species are members of species-rich Neotropical clades still in the full bloom of their diversification, like the lianas of tribe Bignonieae with more than 400 species (Lohmann & Taylor, 2014), palms with over 730 species (Dransfield et al., 2008), armoured catfishes (Loricariidae) with 680 species (Armbruster, Van Der Sleen & Lujan, 2018), and tanagers (Thraupidae) with 371 species (Burns, Unitt & Mason, 2016).

The first attempts to map phylogenetic diversity (PD) over continental and global scales were conducted for select vertebrate groups for which phylogenies were available and for which distribution patterns are relatively well known, such as amphibians, birds, and mammals (Safi et al., 2011). Other than these, large-scale phylogenetic and FD studies with focus and dense sampling in the Neotropics are scarce. Some progress has been made in mapping PD patterns in the Neotropics for specific clades (Lehtonen et al., 2015; Lovejoy et al., 2010; Rossatto, 2014; Fenker et al., 2014) or at the intraspecific level in the search for areas of high phylogeographic diversity and endemism (Carnaval et al., 2014; Smith et al., 2017; Melo et al., 2018). Several ongoing studies by independent research groups are now working to broaden our knowledge on the spatial distribution of Neotropical PD.

Early ideas about Neotropical biogeography

The Prussian naturalist Alexander von Humboldt was among the first to realize that biotic and abiotic processes interact to constrain species distributions, and to place these influences into a geological and climatic framework. He came to this notion in the Neotropics, most famously during his study of the Chimborazo volcano in Ecuador, where he carefully documented the location of different species along elevational zones (Humboldt & Bonpland, 1805). It was in this trip that he first observed that physical parameters such as topography and climate were key for geographic distributions.

A century later, Wegener (1912) advanced the incipient field of historical biogeography with the theory of continental drift, based in part on past geographic distributions of biotas linked by previously connected continental plates. The striking fit between the coastlines of South America and Africa was one of the pieces of evidence inspiring Wegener’s theory of dynamic, non-static landmasses. In the 1960s, a geophysical mechanism for plate tectonics was proposed (Vine & Matthews, 1963), placing studies of plant and freshwater fish biogeography into a plate tectonic framework, where vicariance was assumed as a major biogeographic force (Raven & Axelrod, 1974; Rosen, 1975).

At first, the explanatory power of vicariance biogeography was the ability to predict biogeographic distributions of individual taxa and that of whole biotas from knowledge of how landscapes changed through time (Rosen, 1978). The paradigmatic example is the geological fragmentation of the Gondwana supercontinent, and the resulting fragmentation of the resident Gondwanan biotas. The vicariance biogeography approach satisfies the scientific impulse of systematists and biogeographers for general explanations of organismal distributions, rather than ascribing each distribution to the vagaries of idiosyncratic evolutionary histories (Humphries & Parenti, 1999).

Soon after, the challenge to vicariance biogeography as a general theory was the commonplace observation that vicariant cladogenesis (i.e., allopatric speciation) is only one of three general macroevolutionary processes, along with dispersal and extinction (MacArthur & Wilson, 1967; Ree & Smith, 2008). Indeed, ecologists have long understood dispersal to be a pervasive process influencing biogeographic distributions (Cowie & Holland, 2006). Long-distance dispersal has been documented in the formation of many biotas worldwide (Sanmartín & Ronquist, 2004; Bacon, Baker & Simmons, 2012) including those in the Neotropics (Smith et al., 2014; Tagliacollo et al., 2015b; Hawlitschek, Ramírez Garrido & Glaw, 2017; Antonelli et al., 2018).

Inferring landscape evolution in the Neotropics

Neotropical historical biogeography increasingly relies on geological models that specify the landscape configurations on which species originate, disperse, and go extinct. Understanding phylogeny and biogeography in the context of landscape evolution requires assessment of geological data, including sedimentary environments, sedimentation rates, palaeontological records, and geochronological ages, among others (Salo et al., 1986; Räsänen et al., 1995; Räsänen, Salo & Kalliola, 1987; Hoorn et al., 1995; Lundberg et al., 2000; Figueiredo et al., 2009; Hoorn et al., 2010b; Lagomarsino et al., 2016; Sanín et al., 2016; Jaramillo et al., 2017; Hoorn et al., 2017).

Some recent reconstructions of the Neogene landscape in Amazonia are based on dynamic topography, in which mantle movements through time are quantified (Shephard et al., 2010). The effects of these movements are estimated on surface subsidence and are then related to environmental and landscape changes, such as the model applied to explain the genesis of the Pebas wetland in western Amazonia (Hoorn et al., 2010a). Another approach is to use numerical modeling and create reconstructions from physical parameters such as rates of erosion and mountain uplift. An example is the reconstruction of the flow of the Amazon River which incorporates surface processes, flexural isostasy and crustal thickening due to orogeny into a mathematical model to explain the drainage reversal in the Miocene (Sacek, 2014). However, this study did not incorporate the synergic effects of plate movements and surface dynamics, that are known to have impacted on the formation of mega-wetlands and ecosystems (Horton, 2018).

Landscape evolution models (LEMs) can be useful in a biological context but often lack spatial and temporal precision. Biological data can help to infer past landscapes, by testing alternative geological models and increasing their precision. However, we caution that the evolutionary history of one clade might represent an idiosyncratic story, rather than inform the general evolution of an entire landscape in which the clade occurs (Cruz-Neto, Garland & Abe, 2001; Ribas et al., 2012).

In recent years, integrated approaches have built integrative LEMs based on geological, climatic and biodiversity data (Craw et al., 2016; Badgley et al., 2017; Costa et al., 2017). Some studies make use of geographic information systems and combine these with well-dated palynological databases, such as Neotoma (https://www.neotomadb.org/). These models are mainly applied to reconstruct landscapes across Quaternary time scales (i.e., the past 2.6 million years). For example, reconstruction of changes in connectivity across the northern Andes enabled the inference of cyclic phases of biotic dispersal and speciation vs. extinction (Flantua et al., 2015). Molecular phylogenetic data can be used to statistically evaluate the likelihood of competing geological models on longer time scales, such as the closure of the Central American Seaway dividing South and Central America (Bacon, 2013), and the roles of the Caribbean plate margins as dispersal corridors between South and Central America (Tagliacollo et al., 2015a). Similar approaches based on both terrestrial (Baker & Couvreur, 2012) and aquatic taxa (Hrbek, Seckinger & Meyer, 2007) may provide important insights when geological data are insufficient or ambiguous.

The impact of the Andean uplift on Neotropical diversification

Neotropical biodiversity can only be properly understood when considering the Andean uplift and the effects of this orogeny on the landscape (Fig. 3) and regional climate (Räsänen, Salo & Kalliola, 1987; Gentry, 1982; Hoorn et al., 2010b). Although the Andes are entirely confined to South America, their formation has led to far-reaching effects across the Neotropics, and there are clear links with orogenies in Mesoamerica caused by plate tectonics.

Niche-based studies

The general idea that species are adapted to their environment (i.e., have different niches) has two important consequences. First, species distributions are expected to reflect the distribution of suitable habitats. Second, species composition in local communities should reflect the environmental characteristics of the site, as unsuitable environmental characteristics or biotic interactions make it impossible for a species to establish and/or survive. Along these lines, many studies have aimed to characterize the edaphic associations of tropical plant species (Tuomisto & Poulsen, 1996; Phillips et al., 2003; Costa, Magnusson & Luizao, 2005; Roncal, 2006; Zuquim et al., 2009; Kristiansen et al., 2012; Tuomisto et al., 2016; Cámara-Leret et al., 2017; Figueiredo et al., 2017) and the elevational ranges of many taxa (Kluge, Bach & Kessler, 2008). If there are more species adapted to some environmental conditions than others and dispersal is generally not a limiting factor, a species richness gradient should result. However, it is also possible that some environmental conditions may allow more species to coexist than others. Several studies have analyzed species richness gradients along environmental gradients such as elevation (Kluge, Kessler & Dunn, 2006; Brehm, Colwell & Kluge, 2007), rainfall (Clinebell et al., 1995; Esquivel-Muelbert et al., 2017), and soil fertility (Costa, Magnusson & Luizao, 2005; Ter Steege et al., 2006; Tuomisto, Zuquim & Cárdenas, 2014). In general, these studies have shown that Neotropical plant species richness tends to be highest in warm, humid, and aseasonal environmental niches at low to middle elevations.

Neutral and non-neutral perspectives

In contrast to niche-based processes, spatial patterns in the abundance of anurans from Central Amazonia have been shown to conform to the expectations of Hubbell’s neutral theory of biodiversity and biogeography (NTBB; Hubbell, 2001; Diniz-Filho et al., 2011). More recently, a study demonstrated that the incorporation of population genetic dynamics into NTBB support the hypothesis that biodiversity dynamics are out of equilibrium (Manceau, Lambert & Morlon, 2015). Additional research is needed to assess the relative roles of niche constraints, neutral, and non-neutral processes, in explaining and predicting Neotropical biodiversity.

Ecological interactions

Early theoretical ecologists conceived the role of ecological interactions (e.g., hervivory, pollination, frugivory) in shaping natural communities, mainly through mechanisms of competition and predation (Hazen, 1964; Boucher, 1988). Later, theoretical ecology shifted to a broader perspective, when facilitation (i.e., positive interactions such as mutualism and symbiosis) was envisioned as a mechanism that affects processes in both population and community levels (Bruno, Stachowicz & Bertness, 2003). This broad spectrum linked many ecological concepts and type of data (such as geo-referenced occurrences and DNA sequences), creating a multi-layer framework for investigating macro-evolutionary processes and patterns.

The use of multi-layered data have shed light on the role of climate gradients in pollinator turnover (Correa Restrepo et al., 2016), the role of frugivory traits in palm diversification (Onstein et al., 2017) and demographic and spatio-temporal distribution of species interactions (Beck, 2006). Coupling time-calibrated phylogenetic and ecological data of ant-plant interactions in the Neotropics also allowed the reconstruction of the geographical origin of the Acacia-ant interaction (Gómez-Acevedo et al., 2010) and the identification of ecological and macro-evolutionary patterns in ant symbioses (Chomicki, Ward & Renner, 2015). In addition, phylogenetic and network analyses disclosed that specialized pollination interactions can display asymmetrically dependent diversification (Ramírez et al., 2011), revealing that specialized interactions might dilute the ecological signal in macro-evolutionary processes. Furthermore, comparative phylogenetic analyses using multi-layered data suggest that mutualistic interactions drive the relative higher diversification rates of frugivorous bats in some Neotropical regions (Rojas et al., 2012) and highlight the putative role of bat seed dispersal in shaping species-rich meta-communities.

Single clade approaches

Detailed reconstructions of the temporal and spatial evolution for individual clades are obtained through “single clade” approaches. These approaches focus on contingencies or events that are idiosyncratic to the group under study, instead of generalities across groups. Methodological advances in single lineage approaches have undergone major developments with parametric methodologies (Ree & Smith, 2008; Lemey et al., 2009; Landis et al., 2013; Matzke, 2014; Landis, 2017; Table 7).

Typical biogeographic analyses use time-trees and parametric models of biogeographic evolution to estimate ancestral ranges of lineages (branches) and speciation events (nodes), and to infer rates of biogeographic processes (e.g., dispersal, speciation, and extinction). To date, probably hundreds of studies have examined the biogeographic history of particular Neotropical clades in this way. Biogeographic hypotheses or models about the relative role of biogeographic processes in the geographic evolution of particular groups can be compared statistically using methods for model selection in phylogenetics, such as the Akaike information criterion or Bayes factors (Bozdogan, 1987). Moreover, the rates of these processes may be modified (scaled) to reflect the changing connectivity among the areas of analysis over time (Ree & Smith, 2008). These advances have contributed to the integration of landscape dynamics and geological history into taxon biogeography in the Neotropics (Givnish et al., 2014; Palazzesi et al., 2014; Tagliacollo et al., 2015b; Chazot et al., 2016).

Cross-taxonomic (multi-clade) approaches

These approaches (sometimes under the umbrella of “comparative biogeography”; Antonelli, 2017b) aim to extract generalities on the evolution of a biogeographic region or whole biota, or generalities on the relationships among biogeographic regions or biotas, by reconstructing the history of their individual components. The focus of this approach is on inferring shared biogeographic histories, such as general patterns of colonization and diversification or a common response to climatic and landscape changes. A recent cross-taxonomic analysis for six major clades of terrestrial plants and animals, across all major Neotropical biomes and regions, was presented by Antonelli et al. (2018). That study showed an unexpectedly high number of dispersal events across the entire Neotropics, which took place for tens of millions of years and often involved shifts in major environmental types (in particular from forests to savannas). The high frequency of dispersal events identified in that study reflects patterns reported for tree communities across Neotropical rain forests (Dexter et al., 2017) and between rainforests and savannas (Simon et al., 2009).

Multi-clade approaches were traditionally known as “area biogeography” and were the focus of the cladistic biogeographic school for decades (Nelson & Platnick, 1980; Humphries & Parenti, 1999). The first methods used for cross-taxonomic biogeographic approaches were based on parsimony, which does not allow the formal integration of a temporal dimension (Crisci et al., 1991; Marshall & Liebherr, 2000; Sanmartín, 2016). The incorporation of time into event-based methods then allowed the identification of dispersal corridors and barriers, such as those underlying the assembly of freshwater fish faunas in South American river basins (Dagosta & De Pinna, 2017). Recently developed parametric approaches (Ronquist & Sanmartín, 2011; Sanmartín, 2016) offer now a powerful way to obtain generalities about patterns of dispersal and diversification in biotas, allowing us to test between alternative geological or spatial scenarios (Sanmartín, Van Der Mark & Ronquist, 2008). An interesting methodology bridging community ecology and cross-taxonomic biogeographic analysis is the phylogeographic concordance factor analysis (Satler, Zellmer & Carstens, 2016), which uses Bayesian concordance analysis (Ané et al., 2007) to test for shared evolutionary history among co-distributed species and the existence of strong ecological interactions or dependence (Satler & Carstens, 2016, 2017).

Assembling biodiversity: from communities to biotas

The Theory of Island Biogeography (TIB; MacArthur & Wilson, 1967) introduced parameters such as rates of colonization (immigration) and extinction within a mathematical framework, allowing the prediction of the number of species present on an island based solely on its distance from a mainland species source and its area (Losos & Ricklefs, 2009; Warren et al., 2015). New models inspired by the TIB are now attempting to integrate additional parameters, such as speciation and island age (Whittaker, Triantis & Ladle, 2008), population abundances (Rosindell & Harmon, 2013), and trophic interactions (Gravel et al., 2011).

Community ecology and an expanded TIB are now also adopting a more evolutionary approach by integrating phylogenetic data to the study of community assembly and dynamics, including the role of in situ adaptation or speciation vs. dispersal in community assembly, the temporal sequence of species interactions, and the role of abiotic and biotic factors in diversification of specific lineages (Webb et al., 2003; Sanmartín, Van Der Mark & Ronquist, 2008; Kursar et al., 2009; Valente, Etienne & Phillimore, 2014; Valente, Phillimore & Etienne, 2015, 2018; Cabral, Valente & Hartig, 2017). Importantly, this requires denser voucher sampling of specimens, which in turn will lead to denser phylogenies, better estimates of species boundaries under multi-species coalescent approaches and tackling the common problem of cryptic species (Fig. 4; Bravo et al., 2018). By adopting a more historical focus, community ecology methods are explicitly trying to reconstruct the sequence of events leading to modern-day communities, such as the island-like patches of white-sand savannas in Amazonia (Alonso, Metz & Fine 2013). These approaches relax the assumption of ecological neutrality, and focus on the distinctive properties of individual lineages, historical contingency, and particularities of present-day outcomes (Emerson & Kolm, 2005; Sanmartín, Van Der Mark & Ronquist, 2008). In their most recent forms, these models incorporate ecological parameters such as competition and species interactions (Clarke, Thomas & Freckleton, 2016) or landscape dynamics (Aguilée, Claessen & Lambert, 2013).

Scaling up community ecology approaches

The original goals of community ecology, as established in the early 20th century, were to predict species distributions and abundances, species richness and equitability, community productivity, food web structure, predator-prey dynamics, succession, and community assembly. However, this discipline has not yet succeeded in meeting most of these goals (Ricklefs, 2008; Ritchie, 2009; Vellend, 2010; Ricklefs & Jenkins, 2011; Weber & Strauss, 2016). The reasons are many, but may be especially associated to the non-equilibrium condition of most local assemblages, in which the effects of historical contingencies of dispersal, extirpations, and other stochastic processes override the equilibrium expectations generated by local functional processes such as predation and competition (Fig. 5). In other words, the species composition and equitability of most local assemblages are more strongly governed by regional and historical factors than by local ecological interactions (Mittelbach & Schemske, 2015; Manceau, Lambert & Morlon, 2015; Fukami, 2015; Weeks, Claramunt & Cracraft, 2016).

This “crisis” in community ecology has fueled the rise of alternative functionally-neutral theories, like the NTBB (Hubbell, 2001), and the metacommunity theory (Leibold et al., 2004). However, neutral theories have also been criticized for their simplistic assumptions and lack of predictive power under the non-neutral conditions frequently observed in nature (McGill et al., 2006). In general, the field of community ecology appears to be ripe for a paradigm shift (DeAngelis & Grimm, 2017).

While many studies conducted at continental to global scales aim to test broad hypotheses about drivers of biodiversity gradients (Tuomisto, Zuquim & Cárdenas, 2014; Fine, 2015), others rely on analyses of region-wide field data collected over decades (Ter Steege et al., 2013). These surveys set the stage for analyses on the environmental and historical correlates of diversity (Benavides et al., 2005; Stropp, Ter Steege & Malhi, 2009; Ter Steege et al., 2013). Detailed explanations of the heterogeneity found at multiple scales remain a major challenge.

One recent topic of concern is whether Neotropical biodiversity patterns documented today have resulted from purely “natural” processes, or have been largely influenced by human activities (Levis et al., 2017). Evidence from archaeological, remote sensing, biodiversity data, and modeling approaches suggest that humans may have had a much deeper impact on Neotropical biodiversity, both in time and space, than traditionally conceived (Table 8).

Exploring the tripod: ecological interactions-macroevolution-biogeography

The multi-layer analytical framework developed by recent eco-genetic research granted the combination of multiple data layers from three different scale dynamics (local mechanisms, macro-evolutionary processes, geographic patterns). The synergy of those layers illustrates a tripod that gathers ecological, evolutionary and biogeographical factors of populations, communities, and meta-communities, respectively (Ricklefs & Jenkins, 2011; Hanson et al., 2012; Connolly et al., 2017). Detecting ecological signals across multi-layered data, such as the contribution of mutualisms in biogeographic processes (speciation, extinction, migration) remains a major challenge. Tackling this challenge will require linking spatio-temporal data with models that detect common signatures of ecological interactions across layers (Pilosof et al., 2017). Although recent theoretical advances have unveiled phylogenetic signals from community processes (Rezende, Jordano & Bascompte, 2007; Minoarivelo et al., 2014; Bastazini, 2017), we urge for new models that can identify ecological signal from multiple layers. The exploration of ecological factors that are associated to positive and negative interactions (i.e., network structure, taxonomic associations) might reveal important insights on the dynamics and complexity of ecological interactions for producing and maintaining Neotropical biodiversity.

Incorporating fossils into biogeography

One important shortcoming of molecular-based biogeographical analyses in general, and parametric models of range evolution in particular, is the fact that it is almost always based on extant data alone. Because of the effects of extinction, the pattern of geographic distribution we observe today may be a poor representation of the actual biogeographic history, especially if extinction rates have been unequal among areas (Sanmartín & Meseguer, 2016) and taxa (Silvestro et al., 2016). One way to solve this issue is to include extinct lineages in biogeographic analyses (Mao et al., 2012), or to use their past distribution inferred from the fossil record to constrain inferences of ancestral ranges (Meseguer et al., 2015). This approach has often revealed different biogeographic histories for the study group as compared to analyses based on extant data only (Mao et al., 2012; Meseguer et al., 2015). A recent development is the development of the dispersal-extinction-sampling model, to infer rates of dispersal and area extinction exclusively from fossil data (Silvestro et al., 2014, 2015). Under this approach, a separate sampling parameter is used to account for the unevenness of the fossil record both spatially and temporally. When the fossil record is sufficiently abundant, it provides more accurate measures of changes in rates of geographic evolution and less biased extinction rates, than when exclusively extant taxa are used (Silvestro et al., 2015).

Another challenge to understanding current patterns of evolutionary diversity is the absolute dating of phylogenies, which relies heavily on fossils. This shortcoming complicates a detailed understanding of the ages of tropical taxa, especially those from rainforests (Wing et al., 2009). New methodological developments to directly integrate fossil (extinct) lineages into phylogeny reconstruction (Ronquist et al., 2012; Heath, Huelsenbeck & Stadler, 2014; Zhang et al., 2016; Silvestro et al., 2016) offer new hope in the quest to retrieve more accurate depictions of evolutionary patterns.

Finally, estimating the tempo of diversification is difficult without fossil constraints. In simulated phylogenies, the resulting shape of lineage-through-time plots vary significantly when the fossil record is added as compared to phylogenies that incorporate extant taxa exclusively (Matos-Maraví, 2016; Sanmartín & Meseguer, 2016). The inferred macroevolutionary dynamics estimated from molecular phylogenies may thus be misleading if fossil taxa are neglected, or when macroevolutionary tools do not acknowledge the rare sampling of fossil lineages. Clearly, fossils are crucial to not only understand past dynamics, but also for an improved understanding of current patterns (Fritz et al., 2013). We therefore urge for a much tighter integration between the palaeontological and neontological research communities in the Neotropics.

Integrating landscape evolution models into biotic diversification

A potential problem with single clade and cross-taxonomic biogeographic analyses as discussed above is that areas are treated as traits of organisms evolving along phylogenetic trees. Geology is often used to inform the model but does not form its core. For instance, area connectivity is often used in parametric methods to constrain or scale migration rates but not as an actual part of the model. A new generation of methods that use the power of LEMs to study the full panoply of evolutionary processes, at both microevolutionary (population) level (Byrne & Hopper, 2008; McRae et al., 2008) and macroevolutionary (interspecific) scales (Tagliacollo et al., 2015b; Badgley et al., 2017) are now being developed. For example, uplift of a dissected landscape and river capture are two landscape evolution processes with great power to generate high species richness (Albert et al., 2018). Both of these processes simultaneously and continuously merge and separate portions of adjacent landscape areas, allowing biotic dispersal and larger geographic ranges, vicariant speciation and smaller geographic ranges, and extinction when range sizes are subdivided below a minimum persistence threshold (Albert et al., 2017).

The origins of Neotropical biodiversity

There are often mixed definitions and questions related to the timing and mode of biotic evolution in the Neotropics. The “origin of Neotropical biodiversity” encapsulates at least two contrasting subjects: the timing of origin of hyperdiversity (i.e., “when did the Neotropics reach globally outstanding levels of species richness?”), and the actual age of extant species (i.e., “when did the species that we see today split from their most recent common ancestors, as defined by their stem age?”; Hoorn et al., 2011). It is clear that there have been unusual periods of time throughout geological history, both in terms of biotic and abiotic events (Jaramillo, 2006; Hoorn et al., 2010b; Jaramillo et al., 2010). However, all periods of time have contributed to the current biodiversity, if seen in the perspective of the “evolutionary continuum” that bridges the core biodiversity disciplines (Fig. 4).

Examples of studies that have sought for “special” periods of time often come from time-calibrated molecular phylogenies. For instance, butterfly species-pairs seem to be relatively young in origin (i.e., <2 Ma), suggesting that the Pleistocene and Holocene may have represented “extraordinary times” for Neotropical butterfly speciation (Garzón-Orduña, Benetti Longhini & Brower, 2014) and following the refugium theory (Table 4). However, time-calibrated phylogenies may not fully address the potential impact of extinction and species duration (Hoorn et al., 2011). In other words, if we were able to travel back in time to any period and sequence species around us, the odds are that most species alive might also be around two Ma old or less. In addition, the definition of “species” may vary considerably depending on data source (e.g., based on the fossil record and extant populations) and across taxa. Highly structured populations with considerable genetic divergences may be seen as “incipient species” that have not yet completed the speciation process (Craig et al., 2018). Variable species concepts and adequate sampling of extant and extinct taxa represent a serious barrier for our understanding of Neotropical diversification (Tables 1–3 and 5).

Advancing Neotropical research

Comparative biology has experienced major advancements in the theory and practice of biogeography and molecular phylogenetics during the past decades. However, we still need to increase sampling of organisms drastically in order to advance our knowledge on the patterns and processes underlying Neotropical biodiversity (Feeley, 2015). However, fieldwork in the Neotropics, especially in pristine areas, is time consuming and logistically demanding. Research funding for exploratory inventory projects is also becoming increasingly harder to obtain, despite the fact that highly successful projects (i.e., sequencing the first human genome and creating the Amazon Tree Diversity Network) were initially discovery-driven, rather than focused on testing specific hypotheses. Furthermore, obtaining permits to collect and export biological samples is also challenging, involving many differences across national legislations. Finally, some authors have seen the need of fieldwork as less relevant in this era of museomics (Buerki & Baker, 2016; Zedane et al., 2016).

Despite all these obstacles, fieldwork remains absolutely essential for biodiversity data generation and monitoring (Albert, 2002). Fieldwork also provides students and researchers with a deeper understanding of their study systems, often providing new ideas and questions, while facilitating the establishment of new collaborations, enabling the exchange of knowledge, fueling the development of new methods, and increasing the possibilities of major discoveries (Fleischner et al., 2017). We should seriously consider new strategies for the generation of new biodiversity data, as well as for the syntheses of the already available data.

Multi-taxon field campaigns could provide unique opportunities for intensive sampling, while optimizing resources, bureaucratic, and logistic efforts. However, this vision requires a radical re-thinking and re-organization. We need to provide young generations with the training, tools and resources needed to carry out research on all aspects of biodiversity. We also need to support taxonomic specialists and institutions in order to adequately study, archive and facilitate free access to biological collections and associated data. In addition, we also need to join forces across nations and disciplines for mutual benefit and joint scientific growth. Clearly, these investments would be worthwhile from a global perspective. The future of Neotropical biodiversity research depends on extensive collaborations and coordinated efforts.