PALEOGEOGRAPHY OF SOUTH AMERICA AND
ITS EFFECTS ON THE DISTRIBUTION & PHYLOGENY
By Mike Wise
In general, the geographic distribution, or biogeography, of fishes in tropical South America is caused by the many hydrologic pathways and barriers. This is particularly true for small fishes. Favorable hydrologic pathways allow easy distribution of fauna. On the other hand, hydrologic barriers produce endemism, or a restriction of a fauna from becoming widely dispersed. For smaller fishes, like Apistogramma species, such barriers can include wide, deep river channels, waterfalls and rapids, and even variations in water chemistry and temperature.
Schmettkamp (1982) was the first author to report that most Apistogramma-species-groups have a regional pattern of distribution. Of course, this shouldn’t be a surprise. By definition, a monophyletic species-group arises from a single species and then migrates outward from its original location. Where a distributional range is disjunct, it is an indication that the species-group might be either polyphyletic (having several different ancestors) or that there has been some kind of change in the hydrologic history of the region.
Apistogramma species typically inhabit the shorelines of small streams, lakes, pools, in flooded forests, and floating meadows. In these locations they are typically found in shallow water, living among snags, leaf litter, and aquatic or bog plants. They are almost never found in the open main channel of major rivers. Their small size prevents them from readily crossing major channels without some kind of assistance.
Apistogramma species with extensive ranges of over 500 km (300 mi.) are anomalies. The distribution patterns of such species also tend to be quite linear. Typically such species are highly adaptable in their habitat requirements and water chemistry. They typically inhabit all types of white-, clear-, or mixed clear/blackwater environments
Many highly endemic Apistogramma, on the other hand, tend to be blackwater species. For blackwater species, endemism is most likely due to their water chemistry requirements. Blackwater habitats, streams in particular, are usually not very extensive.
We need to realize that distribution patterns are not based only on modern drainage patterns, but also on those that occurred in the past. Major alterations in drainage patterns have occurred in South America over the past several million years. These changes affected the distribution patterns of the fishes in them. Kullander (1986), for example, noted the presence of the same species - or very closely related species - in the Amazon drainage of northern Peru and in the headwaters of the Río Orinoco along the eastern slopes of the Andes. If one looks at present day hydrogeography, there is no connection between these two areas. Geological studies, however, indicate that several million years ago the rivers of the western Amazon Basin actually flowed northward along the foothills of the Andes and emptied into the Caribbean Sea through Lake Maracaibo. Similar alterations in the past have affected the distribution patterns of many Apistogramma species-groups
The following is a proposed distribution and phylogeny for the Apistogramma species-groups. Papers on the paleogeography of South American rivers by Lundberg et al. (1998) and in Hoorn & Wesselingh (2010), along with genetic studies on Apistogramma species by Miller and Schleiwen (2005) and Ready et al. (2006) have modified many of my ideas about the ages and relationships of the various species-groups.
As I have written previously, since this isn't a juried scientific paper, I will allow myself much more license in suggesting relationships than would a true taxonomist/systematist! I urge those with questions or criticisms to please write me. Only exchanges of views can improve our knowledge of the Apistogramma-species-groups, their distribution, and phylogenic relationships.
Paleogeography of South America
It should be obvious that evolution of the different Apistogramma species-groups is strongly linked to the evolution of the drainage patterns that occurred in tropical South America over at least the past 25 million years, during which time the genus arose and spread throughout much of tropical South America. The phylogeny of Apistogramma species-groups is easier to understand when set on the backdrop of the paleogeography of South America.
If we could see the South American continent of 25 million years ago, it would look nothing like the continent of today. The direction in the flow of interior rivers was influenced not only by the rising Andes Mountains and the Guiana and Brazilian Highlands, but also by structural arches and sedimentary deposition on the ancient interior craton.
Oligocene Paleogeography (25 Million Years BP) (Fig. 1)
Prior to 25 million years ago, the Andes Mountains, which started as a volcanic island arc (similar to present day Japan), finally welded to the South American continent. To the east, the Guiana and Brazilian Highlands, together with the adjoining Purus Arch formed highlands on the South American craton. These upland features produced a hydrological continental divide through the northern half of South America. Rivers to the west of the divide flowed westward in a sub-Andean River system that followed the eastern margins of the rising Andes. The system flowed northward into the Carbonera-Roblecito Embayment in northern Colombia. This embayment was a freshwater delta much of the time, but occasionally was inundated by marine waters. Rivers draining the western Guiana Highland flowed westward into the same river system. East of the Purus Arch, rivers originating in the southern Guiana Highlands and northern Brazilian Highlands flowed into the proto-Amazon and proto-Tocantins river systems. The Michicola Arch in central Bolivia appears to have been a highland area that formed a hydrologic divide between the sub-Andean River system of western Amazonia and the proto-Paraná system to the south. During much of the period, the proto-Paraná was a freshwater system that emptied into a shallow marine embayment at the mouth of the present day Río de la Plata. Several times during the Late Oligocene to Middle Miocene, major inundations by shallow epicontinental seas (Paranan Sea) filled large parts of the proto-Paraná system.
Fig. 1. Late Oligocene Paleogeography
Early Miocene Paleogeography (24 -16 Million Years BP) (Fig. 2)
Major mountain building in the central and (locally) northern parts of the Andes Range occurred at the beginning of the Miocene Epoch. Uplift of the range caused down-warping of the eastern foreland basins, creating a roughly north-south trough in which the sub-Andean River flowed. Mountain building in the northern Andes cut off the sub-Andean River system from its delta near present day Lake Maricaibo. The waters of the river were forced to flow eastward where the El Baul Arch blocked its flow to the Atlantic. This caused much of the old sub-Andean River basin to be flooded with freshwater. This proto-Pebas Wetland is believed to have been more like an Everglades-like biotope than a large lake. Eventually the El Baul Arch was breached and the waters of western Amazonia formed an outlet to the Atlantic Ocean near the present day Orinoco Delta. Periodically, northern parts of the system were inundated by marine waters from the Atlantic Ocean. Waters of the lower Amazon basin, east of the Purus Arch, flowed eastward at least as far as the present day Rio Xingu where it emptied into a marine embayment.
Fig. 2. Early Miocene Paleogeography
Middle Miocene Paleogeography (16 – 11 Million Years BP) (Fig. 3)
The Middle Miocene Epoch started with continued uplift of the central and northern Andes. With increased height, the mountain trapped more moisture from the Atlantic, forming a wetter climate than is seen in the Amazon Basin today. Rivers flowing eastward out of the Andes meandered back and forth over wide areas, depositing thick layers of sediment at the base of the mountain range. These alluvial fans expanded to form megafans at the base of the range, particularly in area of the present day central and northern Andes foothills. Megafans can be thought of as being similar to river deltas except that they are deposited above water. The megafans buried the western parts of the sub-Andean River valley, forcing the wetlands to the east. With increased rainfall, more of western Amazonia, which sat close to sea level, was flooded in freshwater swamps called the Pebas Megawetlands. Periodically, marine waters invaded the wetlands from northern inlets. Little change occurred in the proto-Amazon system, east of the Purus Arch.
Fig. 3. Middle Miocene Paleogeography
Late Miocene Paleogeography (11 – 5 Million Years BP) (Fig. 4)
The late Miocene Epoch was characterized by increasingly rapid uplift of the central and northern Andes, including the uplift of the Vaupés Arch. The Vaupés Arch blocked northward flow of the southern part of the sub-Andean River system and initiated the onset of separate Orinoco and Amazon systems. Megafans expanded, becoming regional in size. The rivers on them meandered over wide swaths of land. Two of these megafans appear to have had major effects on the distribution of Apistogramma species-groups. These are the Pastaza Megafan, which was deposited between the present day Río Putumayo and the Río Tigre drainages, and the Guayabero-Caquetá Megafan, which formed between the Iquitos Arch and the Vaupés Arch. The Guayabero-Caquetá Megafan includes the upper parts of the present day Río Caquetá and Río Guaviaré systems. The southern part of the sub-Andean River system was blocked to the north by the Vaupés Arch, to the east by the Purus Arch, and to the south by the Michicola Arch. This blockage inundated parts of the central Andean foreland basins, forming the Acre Megawetlands. Eventually, the Purus Arch was breached. Waters from the southern part of the former sub-Andean River system flowed eastward into the proto-Amazon, forming the present transcontinental Amazon River. In the Rio Negro region, first the Rio Branco and later the Rio Uaupés systems were pirated from the sub-Andean River system and incorporated into the present Rio Negro system.
Fig. 4. Late Miocene Paleogeography
PM = Pastaza Megafan; G-C = Guayabero-Caquetá Megafan
Pliocene Paleogeography (5-2 Million Years BP) (Fig. 5)
By the start of the Pliocene Epoch, the river drainages were similar to that of present day South America. The only exception was in the upper drainages of the present day Rio Acre and Rio Madre de Dios, which flowed westward into the Peruvian Amazon. About 4 million years ago, a series of east-west trending submarine volcanoes, resting on the Nazca oceanic tectonic plate west of South America, were subducted below the South American continental plate. This caused a bulge in the continent called the Fitzcarrald Arch. The Fitzcarrald Arch blocked northward flow of the Acre and Madre de Dios systems into the Peruvian Amazon. These two drainages were forced to change their flow to the southeast, into the Rio Purus and Rio Madeira systems. In the Orinoco drainage, continued uplift of the Andes repeatedly filled in the channel of the Orinoco, forcing it eastward as far as the western edge of the Guiana Highland.
Fig. 5. Pliocene Paleogeography
Pleistocene Epoch (2 million to 10,000 Years BP) (Fig. 6)
The Pleistocene Epoch was influenced more by radical climatic fluctuations than by tectonic processes. The Pleistocene Epoch is known for a series of ice ages in the Northern Hemisphere, yet these had a worldwide effect - even in the tropics. At the beginning of the Pleistocene Epoch, 2 million years ago, most of the rivers in South America were flowing in the same general channels in which they flow today. Sediments eroded from the Andes by the headwaters of the Amazon, Orinoco, and Paraguay had filled in marine embayments at their mouths.
At least four major and an equal number of minor ice ages, with equally spaced warmer interglacial periods, occurred during the Pleistocene Epoch in the Northern Hemisphere. So much water was trapped in continental and mountain glaciers that during glacial periods sea level dropped approximately 100 m (300 ft.) worldwide. Not only did it expose wide areas of previously submerged coastal areas, but it increased the gradients of the rivers. The Amazon cut its channel downward by up to 100 m (300 ft.) as far west as present day Peru. By the end of the Pleistocene, the main channel of the Amazon in many locations flowed 100 m (300 ft.) below the surrounding land. Tributaries cut deeper channels, too, but not as deep. High waterfalls and rapids separated the lower parts of these rivers from their upper parts and isolated the fauna of the northern banks of the Amazon from those of the southern banks.
Although the most severe climatic changes occurred in the Northern Hemisphere, the Southern Hemisphere was affected, too. In South America, during periods of maximum glaciation, many tropical areas cooled. Andean mountain glaciers expanded and trapped precipitation. This led to a drier climate. In many places, the tropical rainforests of the Amazon basin was replaced by savanna and open montaine forests that were more tolerant of the cooler, drier environment. Tropical rainforests retreated to areas where the orographic effects of mountains and highlands still provided enough rainfall to supply their needs.
Fig. 6. Pleistocene Paleogeography
Holocene Epoch (10,000 Years BP-Present) (Fig. 7)
Warmer temperatures, increased rainfall, and a rise in sea level came with the end of the Pleistocene ice ages. The climate became wetter and warmer. Increased rainfall helped fill the deep channel of the Amazon with freshwater. Rainforests again invaded the central Amazon Basin.
Fig. 7. Holocene Geography
A Proposed Distribution of the Apistogramma-Species-Groups
The Apistogramma species-groups listed in this paper are my attempt to break the genus in to groupings of species that share physical and behavioral characteristics. For the most part, they also show a common pattern of distribution. Those interested in the descriptions of my species-groups can find them in the online article, “A description of Apistogramma species-groups” (Wise, 2011a). A list of species can be found in the online article, “Apistogramma Species List By Species-Groups/Complexes August, 2011” (Wise, 2011b). Both of these works are based primarily on the works of Koslowski (2002) and Stawikowski (2005), which includes the preliminary genetic findings of Miller & Schleiwen (2005). New species and distribution data have been used to modify their works. The Apistogramma species-groups used in these articles and the proposed distribution of these species-groups tend to conform closely to the paleohydrology of South America over the past 25 million years. Altogether, they indicate that the species-groups are valid – and can be used to predict which groups/complexes can be expected in unexplored regions of South America.
Although we have no direct fossil evidence, the genus Apistogramma appears to have a rather long history, probably arising from its geophagine ancestors sometime during the late Oligocene or early Miocene Epoch, possibly over 25 million years ago. The ancestral forms probably originated somewhere along the northern periphery of the Brazilian Highlands. They probably were casually monogamous, with little sexual dimorphism - similar to species in the regani-lineage. This area seems to be the source of ancestors for the pertensis- and trifasciata-lineages, too, since primitive examples of these lineages also occur in the region today. The ancestors of these lineages eventually radiated outward in all directions.
Ancestral species of the regani-lineage appear to have established themselves along the northern margins of the Brazilian Highlands very early. From there, species of the regani-group migrated first north, then west and south to inhabit waters of the Amazon and Paraguay systems. The regani-group became the source for species of the alacrina- and macmasteri-groups in the Orinoco system.
Early migration was northeastward into the eastern part of the proto-Amazon system, where the caetei- and Xingu-complexes arose. Other species-complexes moved south and west.
Species of the caetei-complex probably originated from an ancestral species of the regani-group somewhere the northern Brazilian Highlands. From there it migrated down the Rio Tocantins system into rivers of the Atlantic coastal plain near the mouth of the Tocantins. This migration probably occurred very early in the history of the genus, sometime during the early Miocene. First a marine embayment at the mouth of the proto-Amazon and later the broad main channel of the Amazon prevented the caetei-complex from migrating to the north side of the Amazon. Lowering of sea level during glacial periods of the Pleistocene opened new areas of the Atlantic coast, allowing the caetei-complex to invade the Ilha de Marajó and areas north of the Serra de Tiracambu uplands (Fig. 8).
Fig. 8. Distribution of the caetei-complex
The species of the Xingu-complex show many features in common with those of the caetei-complex. It is likely that they arose from an ancestral regani-group species in the northern Brazilian Highlands at about the same time as did the caetei-complex, sometime during the early Miocene. This locus appears to be in the upper reaches of the Xingu and Tapajós systems, where species of the Xingu-complex still occur (Fig. 9.). From there, species of the Xingu-complex migrated down the Rio Xingu. It is possible that other species of the complex migrated down the Rio Tapajós, too. There is a species (A. sp. Jabuti) found in the lower Tapajós. None are known to occur in the middle reaches of the river, however. This could be due to insufficient collection. On the other hand, it is also possible the Xingu-complex migrated into the lower Tapajós from the lower Xingu.
Fig. 9. Distribution of the Xingu-complex
The ancestor of the resticulosa-complex probably originated in the western Brazilian Highlands (Fig. 10). From there, species of the complex migrated to both the southwest and southeast along the southern part of the sub-Andean River system. This probably occurred early in the evolution of the genus, at least by the middle Miocene. Species of the complex reached southeastern Peru by the start of the Pliocene, but further migration along the sub-Andean river system was blocked by the rise of the Fitzcarrald Arch. Other species of the complex started migrating southeastward into the upper reaches of the present day Rio Mamoré and lower Rio Guaporé, then a part of the sub-Andean river system. During the late Miocene or Pliocene, these eastern rivers of the sub-Andean river system were pirated by the lower present day Rio Madeira. Species of the complex were now able to migrate down the Madeira and along the southern banks of the Amazon where they inhabit the lower reaches of its southern tributaries. It was only able to migrate into a limited number of northern tributaries of the Amazon located downstream of the Rio Negro. One questionable species of the complex (A. pleurotaenia) occurs in the Río Paraguay. It probably arrived during the Pleistocene, together with species of the commbrae-complex and trifasciata-group, when the uppermost part of the Río Guaporé was pirated by the Río Paraguay.
Fig. 10. Distribution of resticulosa-complex
The ancestor of the commbrae-complex appears to have been a species of the resticulosa-complex that lived in southern part of the sub-Andean River system. From there, species of the commbrae-complex migrated east and west (Fig. 11). The rise of the Fitzcarrald Arch during the Pliocene prevented further westward migration. At this time the commbrae-complex was restricted to the upper part of the present day Madeira (Beni, Mamoré, and Guaporé). Eventually, part of the uppermost Madeira (Guaporé) was pirated by the Río Paraguay. This probably occurred during the Pleistocene, when lower sea levels increased stream gradients and increased downward cutting of stream channels. Once the uppermost Guaporé became part of the Paraguay, species of the commbrae-complex eventually migrated into the lower Río Paraguay and Uruguay systems.
Fig. 11. Distribution of commbrae-complex
Apistogramma borellii (sensu lato) is widely distributed down the main channel of the Río Paraguay system of Brazil, Bolivia, Paraguay, and northernmost Argentina (Fig. 12). Koslowski (2002) wrote that A. borellii had more features in common with species of the macmasteri-group than the regani-group. Miller and Schleiwen (2005), however, could not find any close relationship between the two groups. I believe that A. borellii arose from a resticulosa-complex ancestor that had found its way into the Paraguay. Similarities with species of the macmasteri-group might be explained by species convergence. Both groups live in more open savanna/plains environments. Each group would select similar features to best adapt to similar environments. It is very recent in origin, probably first appearing during the Pleistocene.
Fig. 12. Distribution of borellii-group
The Genus Apistogrammoides
The genus Apistogrammoides consists of only one species that is distributed from the upper Ucayali downstream as far as the Colombian Amazon (Fig. 13). This species probably arose from a resticulosa-complex species that became separated from other resticulosa-complex species by the uplift of the Fitzcarrald Arch during the Pliocene. In isolation, it developed into a separate genus and has migrated downstream as far as the Colombian Amazon.
Fig. 13. Distribution of Apistogrammoides
The regani-complex species seems to have originated somewhere along the northwestern flanks of the Brazilian Highlands (Fig. 14). There are still several unusual species with a mixture of features of both regani- and eunotus-complexes that still live in this region. Its ancestors appear first to have moved westward together with other species-groups/complexes along the southern section of the sub-Andean River system. This certainly occurred before the rise of the Fitzcarrald Arch during the Pliocene. Species of the regani-complex occur on both sides of this divide. At first, it was unable to expand past the Pastaza Megafan that started to form sometime during the middle Miocene. When the Purus Arch was breeched during the late Miocene, and the present Amazon drainage formed, the regani-complex followed the main channel downstream along the northern shoreline to its mouth. Populations entered the Rio Negro and reached the present day upper Negro drainage. The regani-complex migrated up the Rio Branco, crossed over the Branco-Rupununi divide, and entered Atlantic coastal rivers in Guyana, Surinam, and eastern Venezuela during the late Miocene or Pliocene.
Fig. 14. Distribution of the regani-complex
The Winkelfleck-complex appears to be closely related to the regani-complex. It probably arose from a regani-complex species somewhere along the eastern edge of the Pastaza Megafan, south of the main channel of the Peruvian Amazon. This probably occurred sometime after the middle Miocene. (Fig. 15).
Fig. 15. Distribution of the Winkelfleck/Angle-patch-complex
The eunotus-complex, like the regani-complex, appears to have originated from species living along the northwestern flanks of the Brazilian Highlands, where eunotus-like species still exist. It followed the same arcuate course of the sub-Andean River system westward and then north into western Amazonia, at least as far as the Pastaza Megafan, as did the regani-complex (Fig. 16). This probably occurred before the middle Miocene, when the Pastaza Megafan started to form. Species of the eunotus-complex (cruzi-subcomplex) occur in areas along the northern edge of the Pastaza Megafan in Ecuador at elevations above 500m (1600 ft.). This would indicate that these typically lowland species inhabited the area prior to the continued uplift of the Andes and development of the megafans. It also appears that the complex existed in the region prior to full development of the Pastaza Megafan. One or two species of the eunotus-complex appear to be the ancestors of the alacrina- and macmasteri-groups. Unlike the regani-complex, when the modern Amazon drainage formed in the late Miocene, the eunotus-complex does not appear to have migrated farther down the Amazon than the western border of Brazil.
Although species of the eunotus-complex cannot be subdivided genetically, it can be subdivided by physical features and distribution patterns. Species of eunotus-subcomplex occur mostly south of the Pastaza Megafan and the main channel of the Amazon. Species of the cruzi-subcomplex occur mostly in river systems that drain the Pastaza Megafan (Río Nanay, Napo and Putumayo). Species of the very closely related Pebas-subcomplex occur mostly on the northeastern periphery of the Pastaza Megafan, north of the main channel of the Amazon, between the Río Napo and Río Ampiyacu.
Fig. 16. Distribution of the eunotus-complex
Based on physical characteristics, species of the alacrina-group appear closely related to cruzi-subcomplex species of the eunotus-complex. Species of the alacrina-group and cruzi-subcomplex are both found along northern margin of the Pastaza Megafan. The alacrina-group probably arose during the late Miocene after the Purus Arch was breeched and the modern Amazon formed. The area north of Pastaza Megafan and Iquitos Arch - which now includes the Caquetá/Japura system, a northwestern outlier of the Amazon basin - became an isolated region of the Peruvian Amazon (Fig. 17). In isolation, the alacrina-group probably originated from an A. cruzi-like species. More recently species of the group migrated into nearby headwaters of the Río Guaviare, part of the Orinoco, and the Río Vaupés, part of the Rio Negro.
Fig. 17. Distribution of alacrina-group
Physically, species of the macmasteri-group appear to be quite similar to species of the eunotus-complex. It probably arose during the middle Miocene, before the formation of the modern Amazon. Once the Purus Arch was breeched and the modern Amazon formed, the northern part of the sub-Andean River system - along with its fish fauna - became isolated from the southern part. North of the Caquetá/Japura system, a eunotus-complex species was probably cut off from species in the Amazon. In isolation, it gave rise to the species of the macmasteri-complex.The species of the macmasteri-group can be split into two species-complexes, the macmaster- and hongsloi-complexes. For the most part, species of the macmasteri-complex are distributed in the upper reaches of the Orinoco system, while those of the hongsloi-complex occur farther downstream.
The macmasteri-complex species probably originated during the middle Miocene, probably between the Guayabero-Caquetá Megafan and the Vaupés Arch (Fig. 18). Many species still occur in this area. From this locus, the complex followed the western shoreline of the northern sub-Andean River system, entering rivers north of the Vaupés Arch. From there, the species-complex migrated around the northern edge of the Guiana Highland and into the newly formed Orinoco delta.
The hongsloi-complex probably arose from a macmasteri-complex species after the main channel of the early Orinoco was forced eastward by sediments from the Andes filling in its original western channel. The hongsloi-complex forms tend to occur along the main channel of the modern Orinoco, along the western flanks of the Guiana Highlands (Fig. 18). It is possible that the hongsloi-complex originated along the southern shoreline of a large llanos lake that existed in the middle Miocene, prior to the second breaching of the rejuvenated El Baul Arch during the late Miocene. It would explain its distribution in mostly llanos regions.
Fig. 18. Distribution of macmasteri-group
The ancestors of the pertensis-lineage appear to have originated in the northern part of the Brazilian Highlands. Species of the lineage then migrated down the proto-Madeira and into the headwaters of the proto-Amazon, east of the Purus Arch. There, it split into several different species-groups. This probably occurred during the middle Miocene, prior to the development of the present transcontinental Amazon system that formed in the late Miocene. Over time it has migrated as far as the middle Rio Orinoco.
The ancestor of the steindachneri-group originated along the northern edge of the Brazilian Highlands (Fig. 19). A primitive representative (A. sp. Rio Preto) still exists in the area. Members of the group then migrated down the proto-Madeira, into the upper reaches of the proto-Amazon east of the Purus Arch. Its distribution included the proto-Negro, which at the time had its headwater below the Rio Branco. This probably occurred sometime during the early or middle Miocene. The present upper Negro, the Rio Branco and farther upstream, called here the proto-Uaupés-Branco, was then an eastern tributary of the sub-Andean River system and flowed to the west. Sometime during the late Miocene the Rio Branco system was pirated from the sub-Andean River system by the Rio Negro. Species of the steindachneri-group migrated up the Rio Branco and crossed over the Branco-Rupununi divide into Atlantic coastal rivers between eastern Venezuela and Surinam. This probably occurred together with species of the regani-complex during the late Miocene or Pliocene.
Fig. 19. Distribution of steindachneri-group
According to Miller and Schleiwen (2005), A. wapisana appears to represent a pertensis-lineage species more closely related to the steindachneri-group than the pertensis-group. It certainly exhibits attributes of both of theses species-groups. It is also found in the middle of the present ranges of these two species-groups. It might be an ancestral remnant of an original pertensis-lineage form, but more likely is an offshoot of the steindachneri-group that originated in the upper proto-Negro. Once the proto-Negro pirated the Rio Branco from the sub-Andean River system during the late Miocene, A. wapisana was able to migrate into the upper Rio Branco. It must have arrived in the upper Branco after the closure of the Branco-Rupununi divide because it was not able to enter the Guianas. It now inhabits the present day middle Rio Negro and Rio Branco (Fig. 20).
Fig. 20. Distribution of A. wapisana
It is possible that the ancestor of the pertensis-group arose along the western edge of the Brazilian Highlands at about the same time as the steindachneri-group, but it is more likely to have occurred elsewhere and at a later time. Although there are no known species that occur in the upper Madeira, an undescribed species (A. sp. Erdfresser/Earth-eater) is found in the Rio Purus. The pertensis-group produced two different species-complexes – the older and more widespread pertensis-complex and the more recent and endemic velifera-complex.
It is uncertain where the ancestor of the pertensis-complex originated, but the group spread widely throughout the proto-Amazon. This probably occurred during the early or middle Miocene, prior to the breaching of the Purus Arch and development of the present transcontinental Amazon during the late Miocene. Once the present upper Rio Negro was pirated from the sub-Andean River system, sometime in the late Miocene, the pertensis-group migrated into the upper Rio Negro and eventually into the Orinoco (Fig. 21). For some reason, the pertensis-group never spread up the Rio Branco.
The two species of the velifera-complex appear to be endemic species, specialized to extreme blackwater biotopes. They are found only in the headwaters of the present day upper Rio Negro/Orinoco where they probably originated from a pertensis-complex species (Fig. 21). The velifera-complex possibly is the ancestor of the iniridae-group.
Fig. 21. Distribution of pertensis-group
The species of the iniridae-group probably arose from a species of the velifera-complex. Like the velifera-complex species, members of the iniridae-group are restricted to blackwater biotopes, but occur only in the upper Río Orinoco drainage (Fig. 22).
Fig. 22. Distribution of iniridae-groups
The species-groups in the trifasciata-lineage represent a monophyletic group that exhibit a reduced number of cephalic pores when compared to species of the other lineages. It can be split into two sublineages based mainly on the number of dental pores on the head. The species-groups of the trifasciata-sublineage show more plesiomorphic (ancestral) features than do the species-groups in the agassizii-sublineage. There is little doubt that the trifasciata-sublineage is older and that the agassizii-sublineage arose from it at a later time.
It appears that the ancestors of many of the species-groups in the trifasciata-sublineage originated in the northwestern Brazilian Highlands, since most species-groups are found along the sub-Andean River system of the western Amazon Basin. All species-groups except one occur in western Amazonia. The brevis-group appears to be an older group that was able to cross into the paleo-Madeira and enter the upper reaches of the proto-Amazon. The species A. arua, A. salpinction, and A. sp. Doppelfleck/Double-spot may represent relicts of this migration into the proto-Amazon.
Based on both physical features and distribution, the brevis-group is the most difficult to define of all of the species-groups in the trifasciata-sublineage. Its range is far from that of any other species-group in the sublineage. Presently, the species of the brevis-group are endemic to blackwater rivers of the upper Rio Negro/Orinoco region (Fig. 23). They appear most closely related to the cacatuoides-group, which they resemble closely. The brevis-group probably originated from a cacatuoides-group species that found its way into the proto-Madeira during the early or middle Miocene. From there it migrated downstream and into areas of the proto-Amazon, including the proto-Negro. During the early and middle Miocene, the proto-Negro ended below the present day Rio Branco. It is possible that A. arua, A. salpinction and A. sp. Doppelfleck/Double-spot represent relicts of this migration. Migration into the upper Negro/Orinoco probably occurred together with that of the pertensis-group, no earlier than the late Miocene or Pliocene, after the proto-Uaupés-Branco was pirated by the Negro from the sub Andean River system.
Fig. 23. Distribution of A. brevis-group
The ancestors of the nijsseni-group appear to have migrated into western Amazonia quite early in the history of the genus, at about the same time as ancestors of the eunotus- and regani-complexes, by following the sub-Andean River system. This probably occurred during the early to middle Miocene, prior to the development megafans along the rising Andean foreland. Species of this group are presently known to occur in Ecuador, together with cruzi-subcomplex species, at elevations over 500m (1600 ft.). The only method in which this could occur is by the fish existing in the area before its uplift. It appears that the nijsseni-group was blocked from further migration by the growth Pastaza Megafan and Acre Megawetlands in the region of the present day Río Marañon and the lower Río Ucayali during the late Miocene (Fig. 24). Members of the nijsseni-group tend to be species that are endemic to blackwater and mixed clear-/blackwater biotopes with rather restricted ranges. It is possible that the species-group originally had a wider range in the region, but at a later time other species (possibly of the cacatuoides-group) entered into the region and successfully evicted nijsseni-group species from more favorable environments, forcing the species of the nijsseni-group into scattered and more extreme biotopes.
Fig. 24. Distribution of nijsseni-group
The ancestor of the cacatuoides-groups appears to have originated in the western Brazilian Highlands. From there, it first migrated into streams in the upper reaches of the sub-Andean River system (the present day Río Mamoré/Guaporé system) (Fig. 25). It then followed this river system into western Amazonia as far as the Pastaza Megafan. This occurred after ancestors of the regani- and nijsseni-groups had arrived, probably early in the late Miocene. Further migration was blocked by the Iquitos Arch and/or the Guayabero-Caquetá Megafan to the north and the Pebas Megawetlands to the east. With the breaching of the Purus Arch during the late Miocene, forming the transcontinental Amazon River, the cacatuoides-group was able to expand eastward into the Rio Solimões.
Fig. 25. Distribution of the cacatuoides-group
According to Miller and Schliewen (2005) the species of the atahualpa-group are an early offshoot of the cacatuoides-group. They probably originated from a cacatuoides-group species somewhere along the outer edge of the Pastaza Megafan. They are primarily blackwater species and are restricted to such biotopes. The atahualpa-group can be split into two different complexes by their breeding behavior and some minor physical differences, but not genetically. The older atahualpa-complex species, found south of the main channel of the Amazon (Fig. 26), breed in the manner typical for Apistogramma. The barlowi-complex species of the atahualpa-group appear to have undergone fairly recent modification to aid in a mouthbrooding style of reproduction. Like the atahualpa-complex, the species of the barlowi-complex are adapted for and endemic to blackwater tributaries. They, however, are presently known only north of the main channel of the Peruvian Amazon (Fig. 26).
Fig. 26. Distribution of atahualpa-group
The trifasciata-group is another species-group that appears to have originated from a species of the cacatuoides-group. It probably arose in the western Brazilian Highlands, in what is now the lower Río Mamoré/Guaporé (Fig. 27). Some species migrated into the upper Guaporé while it was still the eastern headwaters of the sub-Andean River system, while other populations found their way west, as far as the Río Beni system of Bolivia. Westward migration was blocked by the rise of the Fitzcarrald Arch during the Pliocene. When the Río Paraguay pirated streams in the headwaters of the Río Guaporé (probably during the Pleistocene), A. trifasciata crossed into streams of the Paraguay system, migrating downstream as far as northern Argentina.
Fig. 27. Distribution of trifasciata-group
All species of the agassizii-sublineage exhibit two commonly held diagnostically unique features: a reduction of cephalic pores from the plesiomorphic 4 infraorbital pores to 3 and 5 dental pores to 4, and a metallic, usually blue, patch at the corner of the mouth. The only species-group outside of the trifasciata-lineage that consistently exhibits a reduced number of infraorbital pores is the iniridae-group of the steindachneri-lineage. The iniridae-group species, however, exhibit a complete set of 5 dental pores. Behaviorally, all species in the agassizii-sublineage display aggression with the mouth open and gill covers closed. The only other species-groups that show this behavior are the brevis- and trifasciata-groups of the trifasciata-sublineage.
The species-groups that comprise the agassizii-sublineage appear to be monophyletic, that is, to have arisen from a single ancestor. It is not known for certain from which species-group the sublineage arose. It probably was from either an iniridae-group species or, more likely, a trifasciata-sublineage species. It is also uncertain where the agassizii-sublineage originated. Based on present distribution, it appears to have arisen somewhere along the southwestern edge of the Guiana Highlands in the proto-Uaupés-Branco system, which then was part of the sub-Andean River system. All of the species-groups within the agassizii-sublineage still have representatives in the present day upper and middle Rio Negro system. If it did originate in the proto-Uaupés-Branco, then it was there before it was pirated by the Rio Negro during the late Miocene. I believe that an ancestral member of the trifasciata-sublineage was able to cross from the proto-Negro into the proto-Uaupés-Branco or that the ancestor entered the northern sub-Andean River system by finding a way around the Purus Arch. Either way, it occurred during the early or middle Miocene and followed the west-flowing proto-Uaupés-Branco downstream into the Pebas Megawetlands This would explain how the agassizii- and bitaeniata-groups entered western Amazonia.
Wherever it originated, species-groups of the agassizii-sublineage appear to have radiated outward throughout much of the present day Amazon Basin. The only regions of the present Amazon Basin in which agassizii-sublineage species-groups do not occur are in the upper reaches of the former sub-Andean River system south and east of the Pastaza Megafan and north of the Guayabero-Caquetá Megafan. This indicates that the sublineage did not migrate into western Amazonia until these megafans had formed hydrologic barriers, probably early in the late Miocene.
Morphologically, A. elizabethae has many characteristics in common with species of the iniridae-group. Several species of the iniridae-group exhibit a similar body shape, broad lateral band, and caudal fin development with A. elizabethae. It might indicate a relationship, but it could also be explained by evolutionary convergence. The ancestor of the A. elizabethae appears to have originated in the upper reaches of the proto-Uaupés-Branco system, then part of the northern sub-Andean River system. This probably occurred during the middle Miocene, prior to it being pirated by the proto-Negro in the late Miocene. Presently A. elizabethae is known only from the Rio Uaupés and Rio Içana drainages, both rightbank blackwater tributaries of the Rio Negro (Fig. 28).
I am uncertain if the elizabethae-group is monotypic. Although A. elizabethae is well known, there is a photo of another A. elizabethae-like fish that shows a fish with a low, even dorsal fin. No one knows if this is a separate form or just an A. elizabethae with a damaged dorsal fin.
Fig. 28. Distribution of elizabethae-group
The gibbiceps-group appears to be another early offshoot of the ancestor of the agassizii-sublineage. It probably existed in the proto-Uaupés-Branco, probably in the proto-Branco, prior to the Branco being pirated by the proto-Negro during the late Miocene. When the modern Negro formed, it expanded into biotopes in the middle and lower Negro (Fig. 29). The gibbiceps-group exists in the uppermost Branco but for some reason was unable to cross the Branco-Rupununi divide and enter into Guyana, as species of the steindachneri-group and regani-complex had done during the late Miocene/Pliocene.
Fig. 29. Distribution of gibbiceps-group
The ancestor of the bitaeniata-group probably originated in the upper reaches of the proto-Uaupés-Branco at least as early as those of the elizabethae- and gibbiceps-groups. The bitaeniata species-group can be split into two complexes based on the shape of the dorsal and caudal fins. Species of the older bitaeniata-complex have either round or double-tipped caudal fins. The dorsal fin is moderately high to high, with serrated or highly extended anterior dorsal lappets. Species of the younger paucisquamis-complex all exhibit lyrate caudal fins and low, even to only slightly serrated dorsal fins. In other species-groups, modification of the caudal fin appears to progress from round (oldest) to truncate to double-tipped to lyrate/lance/spade shape (youngest). If one considers evolution of the caudal fin in the bitaeniata-group to have followed the same progression, then the oldest forms are now found in the upper Rio Negro and Río Apaporis (Caquetá system). The next oldest, double-tipped forms are found in the Peruvian Amazon, the Solimões, and even in southern tributaries of the Brazilian Amazon as far as the city of Santarém. Species with lyrate caudal fin, the youngest, only occur in the middle and lower Rio Negro.
The distribution of the bitaeniata-complex – migrating first westward and then east and south – seems counter to the present stream flow of the modern Amazon. If, however, the bitaeniata-complex arose in the upper reaches of the proto-Uaupés-Branco sometime during the middle Miocene or even earlier, then it is possible that it followed westward flowing rivers of the sub-Andean River system out of the Guiana Highlands. It found its way somewhere south of the Guayabero-Caquetá Megafan, but was unable to move farther south than the southern boundary of the Pastaza Megafan. When the Purus Arch was breached, forming the modern Amazon River, the bitaeniata-complex followed the river into the lower Amazon. Ancestors of the paucisquamis-complex entered the proto-Negro when the river pirated the paleo-Uaupés-Branco from the sub-Andean River system during the late Miocene (Fig. 30).
The paucisquamis species-complex remained in the Rio Negro, possibly due to being adapted to more blackwater biotopes and/or out competed by more adaptable species (Fig. 30).
Fig. 30. Distribution of bitaeniata-group
The original ancestor of the agassizii-group probably inhabited the proto-Branco in the upper proto-Uaupés-Branco system along the southwestern edge of the Guiana Highlands. A relict species (A. sp. Branco) still occurs in the upper Rio Branco. Species of the agassizii-group exhibit many characteristics found in species of the bitaeniata-group. It is even possible that the two species-groups arose from the same ancestral line. Based on the shape of the caudal fin, the agassizii-group also can be split into two species-complexes. Species of the presumed older pulchra-complex all show round caudal fins, usually with a pattern of interior spots. Species of the younger agassizii-complex have a lance/spade shaped caudal fin, usually with one or more submarginal bands around the outer margin of the fin.
Today, the agassizii-group shows an unusual cross-shaped distribution pattern (Fig. 31). The Rio Negro-Madeira (north-south) is dominated by the pulchra-complex. The main channel of the Amazon (east-west) is dominated by the agassizii-complex. This seems to indicate a series of successive migrations over time.
The agassizii-group probably arose in the proto-Uaupés-Branco system. When the proto-Branco was pirated by the proto-Negro, probably sometime early in the late Miocene, species of the pulchra-complex were able to migrate into the proto-Amazon prior to the breaching of the Purus Arch. The species-complex was then able to migrate up the proto-Madeira. Like the gibbiceps-group, the pulchra-complex did not follow the regani-complex and steindachneri-group over the Branco-Rupununi divide into Guyana. This might be due to the species being more adapted to blackwater biotopes.
Although there presently is no proof for the agassizii-complex, it probably migrated westward, together with the bitaeniata-complex, down the proto-Uaupés-Branco into western Amazonia as far as the southern edge of the Pastaza Megafan. This probably occurred during the early or middle Miocene. Once the Purus Arch was breached in the late Miocene, forming the present Amazon River, the agassizii-complex followed the main channel of the Amazon all of the way to its mouth. The species-complex proved highly adaptable and during the Pliocene invaded many different biotopes on both sides of the Amazon River. During the Pleistocene, the downward erosion of the main channel of the Amazon isolated different populations of agassizii-complex. This isolation produced three major faunal realms: the Peruvian/Solimões realm upstream of the Purus Arch, the left bank realm north of the main channel and east of the Purus Arch, and the right bank realm south of the main channel and east of the Purus Arch. In isolation, forms in each realm developed slightly different sets of characteristics. Presently, the forms which are commonly recognized as A. agassizii can be considered a superspecies that is splitting into several different species.
Fig. 31. Distribution of agassizii-groups.
The diplotaenia-lineage is an odd lineage in the genus Apistogramma. Species show many morphological similarities with the agassizii-sub-lineage, but genetically are separate from all other species-lineages.
Miller & Schleiwen (2005), note that the two known species in the diplotaenia-group are genetically in a separate lineage from all other species-groups. It is unclear where it originated, but it is now endemic to drainages of the Rio Negro and Río Orinoco (Fig. 32). The diplotaenia-group appears to be a very old species-group. Based on its pattern of distribution, its ancestor, like that of the agassizii-lineage, probably inhabited the proto-Uaupés-Branco system. Once the proto-Uaupés-Branco system was captured by the proto-Negro to form the present Rio Negro system, the diplotaenia-group was able to migrate into the middle and lower Rio Negro. With the opening of the Casiquiare Canal between the Rio Negro and Río Orinoco, the species-group was able to migrate northward into the upper Orinoco. In this respect, it may have followed a path similar to that of the brevis-group and at about the same time. The species of the diplotaenia-group are specialized to biotopes in the main channels of larger rivers with sandy bottoms.
Fig. 32. Distribution of diplotaenia-group
Although not part of the genus Apistogramma, it appears that the genus Taeniacara has a distant ancestor in common with it. Taeniacara is distributed in the Solimões and Amazon drainage between the Rio Tefé and the Rio Tapajós, and in the lower and middle Rio Negro. It probably originated in an upper tributary of the proto-Amazon, east of the Purus Arch prior to development of the present Amazon River during the late Miocene. From there it migrated into the lower Amazon as far as the Rio Tapajós (Fig. 33).
Fig. 33. Distribution of Taeniacara
The species-groups used in this article are for the most part based on those proposed by Koslowski (2002). Genetic studies by Miller and Schliewen (2005) have corroborated most of his groupings. Recent discoveries of new species and their distribution have only reinforced the validity of these species-groups. It is, in my opinion, strong evidence that the species-groups in this article are the most valid at present. In addition, the distribution of the species-groups, as proposed in this article, also appear to agree much more closely with the paleogeographic history of South America than that of any other study published on the genus. With it, it should be possible to predict which species-groups occur in areas as yet not examined thoroughly.
This article would be impossible without the resources of so many apistophiles – too many to list, or for my old brain to remember them all. Rest assured that all of you have had a part in this project, the results of over four decades of research, correspondence, and, friendship. Still, I want to personally thank two special people. First, is my mentor Ingo Koslowski. Without Ingo’s books and personal correspondence, the idea of a project like this would be virtually impossible. Second, is my good friend Tom Christoffersen. I doubt that this project would have ever been more than a daydream without his many questions that made me rethink parts of this article – and the years of prodding me to write it.
Hoorn, Carina & F. P. Wesselingh, (eds.). 2010. Amazonia: Landscape and species evolution – A look into the past. 447 pp. Wiley-Blackwell, Chichester.Koslowski, Ingo. 1985. Die Buntbarsche der Neuen Welt — Zwergcichliden. 192 pp. Stuttgart. Verlag Eugen Ulmer GmbH & Co.Lundberg, J. G., L. G. Marshall, J. Guerrero, B. Horton, M. C. S. L Malabarba, & F. Wesselingh. 1998. The stage for Neotropical fish diversification: A. history of tropical South American rivers. p. 13-48. in Reis, R. E., R. P. Vari, Z. M. Lucena, & C. A. S. Lucena (eds.). Phylogeny and classification of
Neotropical fishes. 603 pp. EDIPURUS, Porto Alegre, Brazil.Miller, Michael & Ulrich Schliewen. 2005. The molecular phylogeny of the genus Apistogramma – a working hypothesis. p. 22-25. in Stawikowski, Rainer (ed.). Südamerikanische Zwergcichliden / South American Dwarf Cichlids
(DATZ Sonderheft). 129 pp. Stuttgart. Eugen Ulmer KG. Ready, J. S., I. Sampaio, H. Schneider, C. Vinson, T. Dos Santos, & G. F. Turner. 2006. Colour forms of Amazonian cichlid fish represent reproductively isolated species. J. Evol. Biol. 19(4): 1139-1148.Schmettkamp, Werner. 1982. Die Zwergcichliden Südamerikas. 176 pp. Landbuch-Verlag GmbH. Hannover.Stawikowski, Rainer (ed.). 2005. Südamerikanische Zwergcichliden / South American Dwarf Cichlids (DATZ Sonderheft). 129 pp. Stuttgart. Eugen Ulmer KG.Wise, Mike. 2011a. A DESCRIPTION OF APISTOGRAMMA SPECIES-GROUPS. Published electronically at: http://apisto.sites.no/page.aspx?PageId=116.Wise, Mike. 2011b. Apistogramma Species List by Species-Groups/Complexes, as of August 2011. Published electronically at: