John G Lundberg

Direct fossil evidence shows that by the Middle Miocene the freshwater fish fauna of South America, including Amazonia, was essentially modern across a wide taxonomic and ecological range. Much diversification of modern Neotropical fishes occurred during at least the roughly 60 million year period from the Late Cretaceous to the Miocene. Miocene to Holocene Earth history events played little or no role in creating the great diversity of Neotropical fishes at or above the genus level. The record of Neotropical fishes contains almost no documented extinctions of distinct lines of fishes that are phylogenetically close to living groups. There are, however, many cases of Late and post-Miocene local extirpations of modern groups from areas now peripheral to the large, lowland cis-Andean (Eastern-slope) rivers. The fossil fish assemblages of the Pebas and Acre stages in the Miocene mega-wetland are markedly similar. Further, freshwater fishes of the Middle Miocene La Venta fauna and Late Miocene Urumaco fauna are similar to those of the Miocene mega-wetland assemblages and thus support a high degree of biotic and hydrological relatedness. The majority of Amazonian Neogene fishes were strictly freshwater inhabitants with little or no tolerance for brackish-water conditions. Virtually no Neogene fish fossils from the deep interior of Amazonia, including those of carcharinid sharks, sawfishes, stingrays, ariid catfishes and drums, are certain indicators of marine or even brackish waters.

In northern South America the Cenozoic was a period of intense tectonic and climatic interaction that resulted in a dynamic Amazonian landscape dominated by lowlands with local and shield-derived rivers. These drainage systems constantly changed shape and size. During the entire Cenozoic, the Brazilian and Guiana Shields were stable mountainous areas. Andean-derived river systems increased in importance especially in the Neogene. A remarkable feature in western Amazonian history is the waxing and waning of large lake systems and embayments. By the Late Miocene (about 11 Ma), the Andes were connected with the Atlantic through an incipient Amazon River, and from c. 7 Ma Andean-derived river systems became fully established in central and eastern Amazonia and the modern landscape configuration had developed. Rainforests already existed in northern South America during the Paleogene, but the modern rainforests - with resemblance to the Present forest only developed during the Miocene. The western Amazonian Miocene record contains very diverse aquatic faunas (molluscs, ostracods, turtles, crocodiles, fishes) as well as terrestrial mammals. Remarkable gigantic forms thrived in Amazonian ecosystems at the time. Since the Late Miocene, edaphically heterogeneous lands emerged in western Amazonia in areas previously occupied by lake systems. At the same time nutrient-rich deposits spread over central and eastern Amazonia, an event that, based on molecular phylogenetic studies on extant taxa, coincided with diversification of terrestrial taxa. Molecular-based time estimates confirm the steady diversification and mostly pre-Quaternary origin of extant Amazonian taxa. A significant portion of the current species richness is attributed to a combination of relatively constant wet and warm climates and a heterogeneous edaphic substrate. The Quaternary was a time of distribution shifts, but can no longer be considered a time of diversification in Amazonia.

A megafan is a low-angle, partial cone of fluvial sediments that can reach hundreds of kilometres in length. Its size and depositional processes differentiate the megafan from alluvial fans and deltas. Megafans are a mesoscale, entirely continental depositional feature. This landform has received little attention in either modern or palaeogeographic reconstructions of northern South America. We present a model for landscape development in Neogene Amazonia examining the role of the 'inland delta' or megafan. This environmental setting should be taken into account in mesoscale palaeogeographic reconstructions because it is so widespread in modern foreland landscapes in South America and elsewhere. We describe both megafan and multi-megafan patterns, the latter introduced by means of a roughness map of northern South America, which also assists in identifying landscapes of different types. We distinguish seven landscape phases; in each we evaluate a specific set of physical environmental processes in megafan settings that have biological significance. The processes are based primarily on stream avulsion, but also on stream incision and drainage-net breakdown. We conclude that fluvial sedimentation of the Andean foreland depression probably was accomplished primarily by megafan sedimentation, with declining estuarine and lacustrine sedimentation as the Neogene progressed. Because megafans are often juxtaposed, hydrological connections between neighbouring megafan rivers were probably repeatedly established - with all the associated biological ramifications of range expansion, population mixing, introgression, reinforced genetic isolation and competition. Regional connections between suites of nested megafans, and between major drainage basins, probably existed at times via critically situated megafans. Some modern fish distributions coincide with the mountain-front zone of megafan development, suggesting that megafan-mediated biological processes could have contributed to speciation. Accordingly, our attempt to reconstruct Miocene and younger landscapes in greater detail, with the benefit of the modern megafan analogue, seems justified.

Not long ago, discussions of biotic relationships between Africa and South America were dominated by the issue of biological evidence for or against continental drift (e.g., Mayr et al. 1952; Darlington 1965). Modem freshwater fishes were leading subjects in those discussions. Based on modem fish distributions, Regan (1922) had insisted on a union of southern continents years before plate tectonics revolutionized earth science. Indeed, the apparent close phylogenetic relations within certain Recent groups of African–South American freshwater fishes has been noted in the scientific literature for over a century: Günther (1880), Eigenmann (1909, 1910), Myers (1938, 1967), Gosline (1944,