Amazonia is the primary source of Neotropical biodiversity

Abstract

The American tropics (the Neotropics) are the most species-rich realm on Earth, and for centuries, scientists have attempted to understand the origins and evolution of their biodiversity. It is now clear that different regions and taxonomic groups have responded differently to geological and climatic changes. However, we still lack a basic understanding of how Neotropical biodiversity was assembled over evolutionary timescales. Here we infer the timing and origin of the living biota in all major Neotropical regions by performing a cross-taxonomic biogeographic analysis based on 4,450 species from six major clades across the tree of life (angiosperms, birds, ferns, frogs, mammals, and squamates), and integrate >1.3 million species occurrences with large-scale phylogenies. We report an unprecedented level of biotic interchange among all Neotropical regions, totaling 4,525 dispersal events. About half of these events involved transitions between major environmental types, with a predominant directionality from forested to open biomes. For all taxonomic groups surveyed here, Amazonia is the primary source of Neotropical diversity, providing >2,800 lineages to other regions. Most of these dispersal events were to Mesoamerica (∼1,500 lineages), followed by dispersals into open regions of northern South America and the Cerrado and Chaco biomes. Biotic interchange has taken place for >60 million years and generally increased toward the present. The total amount of time lineages spend in a region appears to be the strongest predictor of migration events. These results demonstrate the complex origin of tropical ecosystems and the key role of biotic interchange for the assembly of regional biotas.

Keywords: Neotropics; biogeography; biome shift; evolution; phylogenetics.

Fig. 1.

Neotropical regions used in the biogeographic analyses, grouped by major biome type and plotted together with the abbreviations used for subsequent analyses. These regions are derived from the classification of terrestrial biomes and ecoregions by Olson et al. (48), but adapted to differentiate disjunct biomes and to simplify the complexity in Mesoamerica and the West Indies, which comprise both forested and open habitats.

Fig. 2.

(A) Biotic interchange among Neotropical regions estimated from dated molecular phylogenies, comprising 4,450 species in six taxonomic groups (angiosperms, birds, ferns, frogs, mammals, squamates). Arrows indicate the direction and number of dispersal events, with line thickness proportional to the number of events. Only connections with more than 10 events are shown (see SI Appendix, Tables S5 and S6 for a full list). The position of the circles in the layout reflects the biotic connection among regions, as calculated by a force-directed placement algorithm implemented in the R package qgraph (56). Dispersal events out of Amazonia are highlighted in red. The color of the circles corresponds to those in Fig. 1. AGL, Andean Grasslands; AMA, Amazonia; ATF, Atlantic Forests; CAA, Caatinga; CEC, Cerrado and Chaco; DNO, Dry Northern South America; DWE, Dry Western South America; Mes, Mesoamerica; PAS, Patagonian Steppe; WIN, West Indies. (B) Number of nonambiguous dispersal events associated with shifts in major biome types compared with shifts to other regions within the same biome type. Shifts involving Mesoamerica and the West Indies are considered ambiguous (n = 2,068; SI Appendix, Table S6).

Fig. 3.

Clade-specific lineage interchange among Neotropical regions, using the same color scheme and methodology as in Fig. 2. Fern image courtesy of Pixabay/Geralt. Flower image courtesy of Pixabay/821292. Animal images courtesy of Malin Laggren.

Fig. 4.

Taxon-specific interchange between Amazonia and (A) Mesoamerica, (B) Cerrado and Chado, (C) Andean Grasslands, and (D) Atlantic Forests throughout the Cenozoic (older events are pruned from the figure but included in the total calculation of events). Dispersal rates are normalized in relation to the number of available lineages in each phylogeny. Age in millions of years ago (Ma).

Fig. 5.

Relationship between the number of dispersal events out of a region (emigration) and into a region (immigration), and predictor variables. (A and B) Regression coefficients of a Bayesian multiple linear mixed effects model calculated based on all regions and clades. (C–E) Partial fit of the number of dispersal events for area and cumulative branch length, calculated for all regions and clades except ferns and frogs. The fitted lines show the posterior mean predicted values from the same model, assuming average values for all other predictors, whereas the points show the raw data. Shaded lines show 95% credible intervals. Dispersal events into or out of Amazonia are shown in red.