The Mahajanga Basin Project, initiated in 1993 and centered in Upper Cretaceous strata of northwestern Madagascar, has resulted in the discovery of some of the most complete, well-preserved, and significant specimens of Late Cretaceous vertebrate animals from the Southern Hemisphere and indeed the world. Among the most important finds are various specimens of crocodyliforms, non-avian dinosaurs, and mammals; these finds have the potential to provide key insights into the biogeographic and paleogeographic history of Gondwana. Madagascar has been physically isolated from Africa for over 160 million years and from all other major landmasses for more than 85 million years. The closest known relatives of many of the Late Cretaceous Malagasy taxa are penecontemporaneous forms from South America (primarily Argentina) and India, thus documenting a previously unrecognized high level of cosmopolitanism among Gondwanan vertebrates near the end of the Cretaceous. The family-level taxa that are shared among Madagascar, South America, and the Indian subcontinent are not known from penecontemporaneous horizons in mainland Africa, but it cannot yet be confidently determined if this is due to differential extinction, poor sampling, true absence (i.e., the taxa were never present on Africa), or some combination thereof. Nonetheless, currently available geologic and paleontologic data are most consistent with the Africa-first model, suggesting that Africa was the first of the major Gondwanan landmasses to be fully isolated prior to the Albian/Cenomanian boundary, and that its terrestrial vertebrate faunas became progressively more provincial during the Cretaceous, while those on other Gondwanan landmasses remained relatively cosmopolitan until the later stages of the Late Cretaceous.

Based on the unequaled southern Late Cretaceous–Cenozoic record, the history of South American land mammals is divisible into two major episodes we term the Gondwanan Episode and the South American Episode. The former episode is distinguished by the northern and central Patagonian Argentinian record, while the latter is mostly based on the extra-Patagonian Argentinian record. The Gondwanan Episode is so termed because it is exclusively represented by endemic mammals of Gondwanan origin, i.e., Mesozoic lineages. In contrast, the South American Episode is almost exclusively distinguished by endemic Therian mammals whose ancestors emigrated from the Laurasian North American Continent.

These two major, successive, and for some time superimposed episodes are the result of the geologic history of the South American Plate. (1) This plate was part of the Gondwana Supercontinent until about 120 millions of years before Present (mybP), when it began to separate and drifted westward (but always south and relatively near the North American Plate). Until about 30 mybP it was connected to Western Antarctica, and through it to most of the Eastern Gondwanan continents. (2) By approximately 125 mybP, Donnelly's “Flood Basalt” was initiated. After its cessation by 85 mybP a variety of compressive tectonic features around the Caribbean began, i.e., subduction and island arcs or continental margin magmatism. These features probably permitted the first known inter-American exchange of tetrapods, dinosaurs such as Hadrosauridae from the rising North American continent and Titanosauridae Saltasaurinae from the South American continent. Probably by the latest Cretaceous these geologic features also permitted the immigration of the first Therian, which gave rise to the native South American land mammals. (3) The geologic definition of southern Central America is the last and most important phenomenon related to the final connection of both Americas. By 12 mybP the submarine connection of Central America with South America began, and subsequent volcanic island arcs permitted the beginning of The Great American Biotic Interchange.

According to the Patagonian record, the absence of tribosphenic mammals and the total extinction of the endemic non- and pre-tribosphenic mammals are the most outstanding events characterizing the Gondwanan Episode. Up to the beginning of the South American Episode (early Paleocene), few Gondwanan lineages of mammals (a native Gondwanatheria and an endemic Dryolestoid) survived in the South American continent; another related Gondwanatheria taxon survived up to the Late Eocene in Western Antarctica. These Gondwanan survivors lived together with the first, but advanced, immigrant Therian that initiated the South American Episode. Although not conclusively demonstrated, there are some suggestions that this superposition began earlier, probably by the pre-Campanian Cretaceous. The history of endemic Therian mammals characterizing the South American Episode began to be known in 1948 thanks to the efforts of the Ameghinos and G. G. Simpson. These represented the only known higher taxa until 1985 when Bonaparte recorded the first non-therian Gondwanan lineages in Late Cretaceous (Campanian) Patagonian beds. Successive authors contributed to this history, ratifying Simpson's statement that the South American Cenozoic mammal evolution was episodic. Thus, the most characteristic geobiotic features of the two major episodes are based almost exclusively on the most eloquent Campanian-Quaternary Argentinian record.

The fauna of the West Indies includes more than 1300 native terrestrial vertebrate species and is characterized by high levels of endemism. Several theories have been proposed to explain how these animals arrived to the islands, including dispersal, vicariance, and land bridges. The dispersal theory proposes that most of the West Indian terrestrial biota arrived by flying or by flotsam. The vicariance theory suggests that there was a proto-Antillean land mass, or masses, connecting North and South America in the late Cretaceous that traveled eastward as the Caribbean geologic plate developed and carried an ancient biota with it. One widely discussed land bridge theory proposes that much of the Antillean biota originated by dispersal over an unbroken dry land connection, the Aves Ridge, between South America and the Greater Antilles 35–33 Ma. Geologic evidence cannot unambiguously support or refute any of these models, despite claims to the contrary. Other evidence bearing on these three theories, or mechanisms, comes from the taxonomic composition of the fauna, their phylogenetic relationships, fossil record, paleogeography, ecology, climate, water currents, and divergence time estimates from molecular clocks. This evidence supports an origin by overwater dispersal for most or all of the West Indian terrestrial vertebrate fauna. The strongest support comes from a reduced higher-level taxonomic composition of the fauna (now and in the past), the presence of unusually large adaptive radiations, and the finding of divergence time estimates between island and mainland groups that are not clustered around a particular time. In addition, the majority of terrestrial (non-flying) groups have closest relatives in South America, which is consistent with the direction of water currents and most hurricane tracks. Some of the same evidence in support of dispersal argues against a mid-Cenozoic land bridge from South America. Several ancient, relictual groups (e.g., xantusiid lizards and solenodontoid shrews) may have arisen through proto-Antillean vicariance, and molecular clock analyses—here revisited—provide support for this, but an origin by dispersal can also be argued. No model can be completely discounted. Although the general pattern (dispersal) has emerged, many details remain to be determined concerning the origin of the fauna.

When the Panamanian land bridge was emplaced about 2.7 Ma, it triggered the Great American Biotic Interchange (GABI), a major mingling of land mammal faunas between North and South America. Four families of northern immigrants (Procyonidae, Felidae, Tayassuidae, and Camelidae) diversified at moderate rates, while four others, Canidae, Mustelidae, Cervidae, and especially Muridae, evolved explosively. As a consequence, half of living South American genera are descendants of northern immigrants. The other major consequence of the interchange was the conquest of tropical North America by immigrants from Amazonia, an episode that justifies the term Neotropical Realm.

The detailed nature of climatic change over the late Quaternary remains poorly understood for northern and central Mexico. A scarcity of records in the former and great complexity in the latter have hindered a thorough reconstruction of changing environments. Previously published research by Metcalfe et al. highlighted questions relating to conditions at the last glacial maximum (LGM), the nature of the transition from glacial to interglacial conditions, and change over the Holocene, including the role of phenomena such as El Niño–Southern Oscillation (ENSO). Here, data from the Sonoran Desert, the Chihuahuan Desert, and the Trans-Mexican Volcanic Belt (TMVB) (and adjacent oceans) are reviewed. In the desert regions, the mid-Pleistocene may have been drier than the late Pleistocene, which was significantly cooler than present and saw more winter precipitation derived from midlatitude frontal systems. There was a significant expansion southward of woodland taxa, although many fossil vegetation assemblages apparently have no modern analogues. Extensive paleolakes existed in the modern desert. Conditions wetter than present persisted into the Holocene, but the modern summer rainfall regime may not have become established until after 9000 uncalibrated radiocarbon-dated years before present (¹⁴C yr. BP). Fully modern conditions started about 4000 ¹⁴C yr. BP. In the TMVB, sparse lake sediment records indicate that the mid-Pleistocene may have been wetter than the late Pleistocene. Further data are still required to confirm whether the proposed pattern of a wet west and a dry east around the LGM holds true. Most lake sediment records show major anthropogenic influence from the mid-Holocene on, although there is evidence for increasing climatic variability in the late Holocene. New deep sea core records indicate the glacial meltwater was re-routed into the Gulf of Mexico after the Younger Dryas cool event, helping to explain the delayed onset of the modern summer rainfall pattern in relation to general warming. High-resolution records are still confined to deep sea cores and tree rings, but highlight the region's vulnerability to climatic change.

The corpus of historical data from lake sediments relating to the climate, vegetation, and human land use of the lowland Central American tropical forest between ca. 20,000 BP and the time of European contact is reviewed. Pollen, phytolith, and charcoal records identify the distribution and composition of tropical vegetation and fire patterns during the late Pleistocene, when they were significantly altered from today's, and earliest Holocene, when plant communities reassembled and interglacial representatives began to coalesce on the landscape. The significance of the environmental perturbations that occurred during the transition from the Pleistocene to the Holocene for human occupation of the lowland tropical forest and the geography and chronology of agricultural origins is discussed. Fire was employed by hunters and gatherers and farmers alike during the past 11,000 years as a primary tool of forest modification. The profound effects of an ancient pre-Columbian development of plant food production and, subsequently, slash and burn agriculture between ca. 10,000 BP and 4000 BP can be seen on lowland forests from Mexico to the Amazon Basin.

Plant microfossil and macrofossil associations obtained from six dated sections from the area of the basin of Bogotá (2550 m, Eastern Cordillera, Colombia) show the evolution of the late Neogene Andean montane forest, triggered by the Andean orogeny. Progressive adaptation of warm tropical taxa to cool montane conditions, evolution of new neotropical montane taxa, and immigration of temperate Laurasian, Holarctic, and Austral-Antarctic elements gave shape to the present-day montane forest. Vegetational characteristics inferred from fossil plant associations reveal the altitude at the time of deposition. Neogene forests are floristically compared with contemporary forests at comparable altitudinal intervals in the surroundings of the Bogotá basin; however, the absence of taxa that had not yet arrived, or evolved, is most salient and shows that non-analogue plant communities are common.

The main phases of montane forest development are: (1) pre-uplift phase of the late Miocene with abundant lowland taxa with tropical or neotropical affinities (Mauritia Kunth, Amanoa Aubl., Ceiba Miller, and representatives of Humiriaceae); montane forest rich in Podocarpaceae (potentially including Nageia Gaertn., Podocarpus L'Hér., Prumnopitys Phil., and Retrophyllum C. N. Page) covered other previously uplifted areas in the region; (2) toward the early Pliocene the area was uplifted to ca. 1000 m; the relative proportion of temperate taxa of North American and southern South American stock increased and occupied the slopes of the low mountains; and (3) by the middle Pliocene uplift had proceeded to ca. 2000–2200 m and tropical lowland taxa, which are now restricted to altitudes below 1000 m, are no longer recorded in the fossil plant associations; the increase in the number of newly appearing montane taxa (Myrica L., Turpinia Vent., Gunnera L., Bocconia L., Gaiadendron G. Don f., Daphnopsis C. Martius, and Monnina Ruiz & Pav.) suggests a significant increase of diversity.

Until the late Pliocene there is little to no evidence for the existence of the páramo; taxa such as Xyris L., Hypericum L., Carex L., Montia L., and Ranunculus L. might have formed swamp or bog vegetation only. It is believed that these taxa colonized mountaintops with half-open vegetation; these areas extended when the mountains reached above upper forest line elevations.

The distribution areas of the endemic genera of the Espeletiinae largely coincide with the youngest parts (< 5 Ma) of the northern Andes. Changing climatic conditions forced individual plant species to migrate vertically. Composition of plant communities changed continuously and vegetation belts frequently were altitudinally squeezed or offered possibilities for expansion. This long process stimulated speciation and provoked sequential non-analogue vegetation types. Thus, the biodiversity hotspot of the northern Andes has a dynamic history.

The fossil record of the Rhizophoraceae includes numerous questionable identifications from strata of unconfirmed age, together with others representing clear evidence of the family in deposits whose age is well constrained by independent or multiple lines of evidence. Bruguiera or the Bruguiera lineage, and Ceriops are known from the early Eocene, Kandelia from the middle Eocene, and Rhizophora from the late Eocene. Combretocarpus is likely present by the middle Miocene, but it belongs to a family (Anisophylleaceae) earlier placed in the Rhizophoraceae that is now considered unrelated to that family. Rhizophora, as a genus, has existed essentially uninterruptedly in the Caribbean region since the late Eocene, but cooling events at ca. 15–14 Ma, ca. 3.4 Ma, and ca. 1.6 Ma, and the subsequent 18 to 20 glacial intervals provide a mechanism for development of the cold(er) tolerant populations toward the northern part of its range in the New World. In reconstructing the biogeographic history of Rhizophora in the Caribbean region, it is worthwhile noting that molecular-defined lineages (with little morphological or palynological variation) may have been introduced and gone extinct on multiple occasions. This allows for variation in molecular patterns and apparent stasis in the fossil record—a possibility not widely considered in the literature. The most recent molecular-defined common ancestor to existing populations is estimated to have arrived in the New World ca. 11 Ma, using an origin in the Paleocene (ca. 60 Ma) as the basis for calculation.

The combination of factors that account for present-day distributions of organisms is unique to individual lineages and varies over time. An observation relevant to some lineages at some point in their history is that hurricane frequency and intensity appears to be increasing with global warming. If so, then directional winds from Africa to the Caribbean region (the trade winds), from South America to Africa (the westerlies), the ocean currents they induce, and the transport of the floating islands they carry, were likely more intense during the generally warmer-than-present Tertiary Period and especially at peaks of exceptional warmth. These peaks occurred in the Paleocene/Eocene (65–45 Ma), in the early to middle Miocene (broadly between ∼23–12 Ma), in the middle Pliocene (3–4 Ma), and probably extended slightly later in the lower latitudes. The Paleocene/Eocene interval includes the time when the distance between Africa and South America was one-half to two-thirds that of the present, and when the Greater Antilles island arc was first becoming emergent as increasing target areas for propagules. The second and third intervals of warming include the times when molecular evidence suggests divergence between several African and New World lineages. Thus, wind and ocean transport of organisms and propagules throughout the Tertiary, and especially at peaks of warmth, was likely a more important means of dispersal than would seem plausible under present conditions.

The historical biogeography of the tribe Sinningieae (Gesneriaceae) was analyzed based on distribution data and a molecular species-level phylogeny of 76 species. This plant group is distributed from Mexico to northern Argentina, but by far the highest diversity occurs within the Brazilian Atlantic forest. The dispersal-vicariance analysis method and a cladistic approach were used to infer ancestral areas as well as patterns of dispersal and vicariance. Results indicate that the Sinningieae probably arose in the coastal rain forest or in the neighboring tropical area delimited by the São Francisco river in Brazil. The majority of the dispersal-vicariance episodes were reconstructed between the Brazilian Atlantic rain forests and their neighboring inland areas (i.e., Paraná and São Francisco regions). In contrast, few dispersal-vicariance events were reconstructed between the tropical and subtropical areas of the Atlantic rain forest and between the Paraná and São Francisco regions. These results, together with ancestral areas inferred at the root of the main lineages, indicate an early north-south disjunction in Sinningieae. Occurrence of Sinningieae species in other areas, such as the Andes, the cerrado, Amazonia, northern South America, and Central America, are mainly explained by independent range expansions of single species from either the Paraná or the São Francisco regions. According to the dispersal-vicariance optimization, 20% of the speciation events are subsequent to inter-area dispersal and range expansion, 23% are associated with inter-area vicariance, and 57% occurred at a lower geographical scale within a single biogeographic area. These results pinpoint the need for a phylogenetic framework to correctly understand area relationships and the relative contribution of dispersal and vicariance events in present-day distribution patterns.

We studied the Araceae of the Cabo Corrientes region on the Pacific Coast of Colombia and compared its aroid flora with those of La Selva (Costa Rica), Barro Colorado Island (Panama), Bajo Calima (Colombia), Río Palenque (Ecuador), and Iquitos (Peru). We found 114 native species in 14 genera at Cabo Corrientes; the largest genera were Anthurium (38 species), and Philodendron (36), which together accounted for 65% of the species. Seventy-four percent of the species were exclusively epiphytic or hemiepiphytic. Most of the epiphytic species belonged to the genera Anthurium, Philodendron, Rhodospatha, Syngonium, Monstera, and Stenospermation, the latter three containing exclusively epiphytic or hemiepiphytic species. The flora of Araceae of Cabo Corrientes was most similar to that of La Selva, Barro Colorado Island, and Bajo Calima; similarity with Río Palenque was low. Our findings support Lellinger's view that the Chocó biogeographic region extends to the Nicaragua-Costa Rica border and is divided into a northern and a southern flora.