Late Cenozoic sedimentary sequences in Acre state, southwestern Amazonia: Fluvial or tidal? Deductions from the IGCP 449 fieldtrip

Journal of South American Earth Sciences 21 (2006) 120–134 www.elsevier.com/locate/jsames

Late Cenozoic sedimentary sequences in Acre state, southwestern Amazonia: Fluvial or tidal? Deductions from the IGCP 449 fieldtrip Rob Westaway Faculty of Mathematics and Computing, The Open University, Eldon House, Gosforth, Newcastle-upon-Tyne NE3 3PW, UK Received 1 February 2004; accepted 1 August 2005

Abstract The IGCP 449 fieldtrip in June 2003 drew attention to the Late Cenozoic fluvial sequences of western Amazonia. In Acre state in western Brazil, underlain by relatively mobile crust, rivers have incised up to 70 minto the stacked latest Miocene (?)/Early Pliocene (?) sediments of the Solimo˜es Group, creating staircases of fluvial terraces and indicating regional uplift on this time scale. In contrast, in western Rondonia state, the Madeira River flows through the Early Proterozoic western part of the Amazon Craton, where Late Cenozoic vertical crustal motions seem minimal. The evidence in Acre suggests that the Solimo˜es Group was deposited by an ancestral river system associated with the incipient development of the modern eastward Amazon drainage. q 2005 Elsevier Ltd. All rights reserved.

1. Introduction With a length of 6516 km and a drainage area of approximately 7.05 million km2, the Amazon is the world’s largest river system (Fig. 1), discharging around 209,000 m3 sK1 (w6900 km3 yrK1) of water and transporting 1.2 billion tonnes of sediment annually (Meade et al., 1985). Its present eastward course from the Andes mountains to the Atlantic Ocean is known to be young and developed no earlier than the Late Miocene (cf. Dobson et al., 2001). Its lower reaches presumably flowed eastward ever since the South Atlantic Ocean basin began to open in the Early Cretaceous. However, its upper reaches previously flowed northward along the Andean foreland basin through Maracaibo Gulf into the Caribbean Sea (Hoorn et al., 1995) (Fig. 1) and thus crossed the modern Eastern Cordillera of the Andes in northern Colombia, where several kilometers of surface uplift have occurred since the Middle Miocene (e.g. Gregory-Wodzicki, 2000). In contrast, most other large rivers outside glaciated regions are ancient and have maintained near-constant geometries for many millions of years; once established, they tend to erode or aggrade fast enough to maintain their gradients in response to whatever crustal deformation may occur. The cause and timing of this diversion of the Amazon drainage thus represent important questions. However, due E-mail address: [email protected]. 0895-9811/$ - see front matter q 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.jsames.2005.08.004

to practical difficulties—including the vast scale and remoteness of this region, the near total absence of outcrop due to vegetation cover, and the health hazards caused by tropical diseases—few people have ever investigated the sediments of the ancestral Amazon system. To introduce this region to about 30 international specialists, Edgardo Latrubesse and Jose´ Stevaux led a 7-day field excursion to it in June 2003, as part of International Geological Correlation Programme (IGCP) 449. This trip provided a good illustration of the fluvial record in this region and enabled considerable insights into the associated literature, much of which is contentious. This excursion was based in Rio Branco in Acre state, SW Brazilian Amazonia (Figs. 1 and 2), which is a good place to study the Late Cenozoic evolution of the Amazon system because of its accessibility and the exposure of sedimentary sections. First, it forms part of the vast depocenter of sediments of the Solimo˜es Group, which record the ancestral Amazon system in the (?) Late Miocene–Early Pliocene (Fig. 3). Second, as a result of Late Cenozoic surface uplift and the associated fluvial incision, approximately 70 m of local relief has developed (from w210 to w140 m above sea level), more than in most of the Amazon Basin (Fig. 4), which facilitates the exposure of the Solimo˜es Group and the subsequent staircase of river terraces in riverbank sections and other settings. Third, this region adjoins the cratonic part of Amazonia in Rondonia state farther east (Fig. 2), where greater crustal stability is evident. Fourth, unlike in most of Amazonia, a road network facilitates access to natural

R. Westaway / Journal of South American Earth Sciences 21 (2006) 120–134

121

sections and has created new sections in the road cuts. Fifth and finally, Acre has been studied more than most other parts of Amazonia and has indeed been the origin of disputes about the ages (Fig. 3) and depositional environments of sediments, in particular, whether the exposed Solimo˜es Group sediments have a fluvial or tidal origin. 2. Fieldtrip itinerary The field excursion visited rivers of the Andean foreland basin, notably, the Acre, Purus, and Iaco (Fig. 2), which are incising the Solimo˜es Group, and also addressed Late Pleistocene–Holocene fluvial processes and environmental hazards. In addition, the excursion visited the Madeira River in Rondonia state (Fig. 2), where it flows through the more ancient landscape of the Amazonian Craton. The total distance traveled was approximately 1500 km. 2.1. Acre, Purus, and Iaco rivers Fig. 1. Summary map of the study region in northern South America, showing the Amazon and adjacent major river systems, plus localities discussed in this study. Line shading indicates the Andes, with individual ranges of the northern Andes separatelyidentified. Dashshading indicatescratonicregions.The Andean foreland basin and Amazon Rift are unshaded, except that structural highs demarcating individual subbasins are labeled. Adapted from Fig. 1(c) of Hoorn et al. (1995).

The Acre, Purus (Fig. 5), and Iaco rivers have headwaters in the Andean foreland basin (Fig. 2). The Purus is approximately 1600 km long (Fig. 1), drains an area of 0.37 million km2, discharges around 350 km3 of water, and transports 29 million tonnes of sediment annually

Fig. 2. Map of the excursion area. Dashed lines mark borders between Brazil, Bolivia, and Peru and between the Brazilian states of Acre, Amazonas, and Rondonia. Solid lines mark rivers, and dot-dashed lines mark roads. Selected vertebrate biostratigraphic sites (Latrubesse et al., 1997), Ar–Ar dating sites (Campbell et al., 2001), and the Macusani volcanic field (thought to be the source of the tuffs at these sites; cf. Noble et al., 1984; Campbell et al., 2001) are also shown. Thick solid lines mark the field excursion route: Rio Branco-Abuna-Guajara-Mirim-Pacaas Novos Mts.-Guajara-Mirim-Abuna-Rio Branco-Boca do Acre-Rio Branco-Sena Madureira-Rio Branco.

122

R. Westaway / Journal of South American Earth Sciences 21 (2006) 120–134

Fig. 3. Cross-sections illustrating the diversity of chronostratigraphic interpretations in the study region (see Fig. 2 for locations). (a) Acre River in Brazil, redrawn from Frailey et al. (1988, Fig. 1). (b) Generalized section for the Upper Purus and Las Piedras rivers in the adjacent part of Peru, redrawn from Campbell et al. (2001, Fig. 1). In (a), Tertiary redbeds are shown overlain on the Upper Acre by the ‘Acre Conglomerate’ (AC), a clay-ball conglomerate, and by three informally designated members of the Madre de Dios Formation, considered by Frailey et al. (1988) as latest Pleistocene–Holocene. Unit A consists of clay with nodular iron concretions and calcareous stringers that have yielded Late Pleistocene radiocarbon dates. Unit B is more homogenous silty clay; and Unit C is more complex crossbedded sand, clay, and silt. Paleoenvironmental interpretations (Frailey et al., 1988) are shown: C, channel; D, deltaic; F, floodplain; L, lacustrine. The same units were recognized by Campbell et al. (2001) in Peru, who considered them to be of regional extent. The vertical extent of exposure is typically about 30 m in Peru and in the Upper Acre (see (b) and left side of (a)) but about 70 m around Rio Branco (right side of (a)). In (b), the magnetostratigraphy is followed by formation names and informal subdivisions, their ages, and the observed thickness range of each subdivision. The stratigraphic positions of the Late Miocene–Early Pliocene mammal sites (in [a], Latrubesse et al., 1997) contradict the very young ages assigned by Frailey et al. (1988), as do the Ar–Ar dates and magnetostratigraphy in (b).

(cf. Latrubesse and Stevaux, 2003a). The Acre is much smaller (Fig. 2), but both it and the Purus (Fig. 5) have a similar overall appearance. During the dry season (winter in the southern hemisphere), the rivers flow within roughly 10 m deep incised channels, cut into Late Pleistocene–Early Holocene fluvial deposits or older deposits of the Solimo˜es Group. At high stages, they usually flood the surrounding land surface, which is the lowest of a staircase of river terraces that rise to the uppermost depositional surface of the Solimo˜es Group, about 70 m above the low-stage river level (Fig. 6). Many paleochannels are evident within the youngest of these terraces, some of which are much larger than modern channels and thereby indicate greater discharge in the past. However, the modern entrenched channels seem quite stable, generally

experiencing only slow migration due to bank collapse and lateral accretion. Detailed descriptions of the Late Pleistocene–Holocene sedimentary processes along such rivers have been provided by Latrubesse and Kalicki (2002) for the Purus near Boca do Acre (which was visited during this excursion, thereby enabling many field descriptions to be confirmed by direct inspection) and Latrubesse and Rancy (1998) for the Jurua´ to the west. 2.2. Madeira River The Madeira is the largest Amazon tributary (drainage area w1.36 million km2; water discharge w1010 km3 and sediment transport 450 million tonnes per annum; Latrubesse

R. Westaway / Journal of South American Earth Sciences 21 (2006) 120–134

123

Fig. 4. Montage of a relief-shaded altimetry image of northwestern South America, prepared from individual image swathes provided by the US National Imagery and Mapping Agency (NIMA). Based on Digital Terrain Elevation Data level 0 data, with each pixel representing a 30 arc second!30 arc second or w1 km! w1 km area; obtained from the NIMA Web site at http://geoengine.nima.mil. Note the badland landscape across most of western Amazonia, indicative of tens of meters of typical fluvial incision.

124

R. Westaway / Journal of South American Earth Sciences 21 (2006) 120–134

Fig. 5. View to the west up the Purus River from Boca do Acre (Fig. 2) at the confluence of the Acre and Purus rivers, illustrating the characteristic geomorphology of rivers in the Solimo˜es Group depocenter within the Andean foreland basin. The river is at its low stage, confined by w10 m high banks. During March–April, it floods the surrounding land that consists of a terrace of presumed latest Pleistocene–Early Holocene age. At higher levels, locally up to w40–50 m above river level, older lateritized fluvial sediment is present, presumed as part of the Solimo˜es Group.

and Stevaux, 2003a), with headwaters in the Andes (Fig. 2). One of its principal affluents, the Beni (Fig. 2), drains a 68,000 km2 area of NW Bolivia and transports about 100 million tonnes of sediment annually (Aalto et al., 2002, 2003), indicating a spatial average erosion rate of approximately 1 mm aK1 in the 50,000 km2 area of the Andes drained by this river (estimated as w1011 kg yrK1 of sedimentOw2000 kg mK3 sediment densityO50,000 km2 area). To investigate the Madeira and the geomorphology of the nearby Pacaas Novos mountains (Fig. 2), a trip 400 km east was required. This reach of river, downstream of Guajara-Mirim (Fig. 2), is not navigable due to the many rapids where it flows through the crystalline basement of the Amazonian Craton (Guapore Shield), in dramatic contrast with the geomorphology observed in Acre state (Figs. 5 and 6) in much younger crust. Downstream of Guajara-Mirim, there is a single low terrace about 8 m above present river level, typically consisting of an 8 m thick, 20–25 km wide expanse of gravel

(MME, 2000). The coarseness of this material contrasts strongly with the present-day sediment load of the Madeira, which is mainly silt and fine sand, the coarser components currently being deposited farther upstream near the front of the Andes (cf. Guyot et al., 1999). This terrace is thus thought to represent much higher than present discharge and sediment transport by this rivers (Fig. 2) during the latest Pleistocene, possibly related to melting of ice upstream in the Andes. The terrace appears mainly on the western, Bolivian side of the river, but around Abuna, the Madeira bends westward (Fig. 2) and crosses to the opposite side of this terrace. It was investigated in alluvial gold workings 80 km north of GuajaraMirim, at KP 4551 8806, where its uncemented fine gravels, composed mainly of clasts of Andean origin (e.g. DeCelles and Hertel, 1989), were locally observed to overlie thin, ferruginously cemented coarse quartz gravels. Similar cemented residual gravels are also observed around 1000 km farther east in Mato Grosso state, central Brazil, at the southern margin of the Amazon Craton in the headwaters of

Fig. 6. Idealized typical transverse profile across river valleys in Acre state showing the observed sand-dominated river terrace staircase inset into Solimo˜es Group deposits. The pattern shown is typical for all major rivers in SW Brazilian Amazonia (i.e. Acre, Iaco, Jurua, and Purus; Fig. 2). Based on Latrubesse et al.’s (1997) Fig. 2, with additional information from Latrubesse and Stevaux (2003c). The lowest terrace consists of four closely spaced treads no more than 1–2 m apart. The youngest of these, most widely exposed along modern river banks, is seasonally flooded and evidently Holocene.

R. Westaway / Journal of South American Earth Sciences 21 (2006) 120–134

the Parana´ River (e.g. Bibus, 1983), where they are thought to be of considerable antiquity (no younger than Pliocene or Early Pleistocene). A similar age thus can be estimated for the deposits along the Madeira, which implies that this river has not incised significantly into the cratonic crust for hundreds of thousands, or possibly millions, of years. It thus resembles other rivers in cratonic regions, such as India, southern Africa, and western Australia, which have maintained their courses at the same level for long periods of time. This effect is presumed to reflect the much greater stability of cratonic crust compared with normal continental crust (Westaway et al., 2003). Upstream of Guajara-Mirim and Riberalta (Fig. 2), outside the craton, there is instead clear evidence of young fluvial incision (e.g. Ra¨sa¨nen et al., 1987, 1992; Campbell et al., 2001; Fig. 4), including evidence of river terrace staircases (e.g. MME, 2000), which suggests that this region is also uplifting. However, no visit to any of these localities was possible. 2.3. Pacaas Novos mountains The 150 km long Pacaas Novos mountain range (Fig. 2), which runs west–east from Guajara-Mirim through the

125

southern part of the Amazon Craton and has been studied by Latrubesse et al. (2000), was visited around KP 5542 1514, 15 km NE of Guajara-Mirim. It consists of up to 400 m of gently dipping arkosic sandstones and conglomerates that seem to have been deposited in a normal fault zone within the metamorphic basement, which is Early Proterozoic (e.g. Texeira et al., 1989). However, Cenozoic chemical weathering has had much more effect on the basement than on the sediments, thus lowering the level of the basement (but leaving many residual features, such as corestones and tors) and creating a topographic inversion. The north flank of the range is now about 150 m high, rising from 200 to 350 m a.s.l. Latrubesse et al. (2000) suggested that these sediments are Late Precambrian. However, this sedimentary rock can be traced eastward more or less continuously to the Parecis Sandstone of Mato Grosso state, which overlies the earliest Cretaceous Tapirapua Basalt (e.g. Bibus, 1983), the NW part of the Parana´ flood basalt that erupted during the rifting that preceded the opening of the South Atlantic Ocean basin. Nonetheless, clear evidence of differential rates of denudation in this region makes it potentially useful for calibrating surface exposure dating techniques.

Fig. 7. Part of the SW face of a w6 m deep cutting on highway BR-364 at EK 5684 7102, w32 km SE of the Iaco River at Sena Madureira and w100 km NW of Rio Branco. Red nodular clay (left) is overlain by lithified yellow sand that aggraded both laterally and vertically. This exposure can be explained as a consequence of deposition in a river channel that flowed through a preexisting lake bed. The basal sand adjacent to the steep part of the clay contact contains clasts of nodular clay, consistent with local erosion and redeposition by bank collapse—a common present-day process along rivers in this region and widely observed in the Purus around Boca do Acre. The individual channel sand bodies revealed, which are presumed to represent successive seasonal floods, are approximately 2 m thick.

126

R. Westaway / Journal of South American Earth Sciences 21 (2006) 120–134

Fig. 8. Cross-section through the Solimo˜es Group sediments exposed in the left bank of the Acre River at Seringal Amapa (approximate coordinates FJ 235 910; w8 km SW of Rio Branco city), as logged by Ra¨sa¨nen et al. (1995; their Fig. 3(a)). Alternations of sand and clay are shown; sand units are up to 50 cm thick, and approximately 55 sand (light)/mud (dark) couplets are evident. No evidence was observed of plant root development in the silt units, suggesting they were not exposed subaerially for any significant time before the next cycle of sand deposition. Cross-bedding was not observed in the sand units (cf. Fig. 7).

2.4. Solimo˜es Group Sediment of the Solimo˜es Group underlies much of western Amazonia (Fig. 1). These sediments were found in roadside exposures east of Rio Branco and in river banks and terrace bluffs at Boca do Acre and Sena Madureira (Fig. 2). However, they were best exposed in cuttings along highway BR-364, west of Rio Branco (Fig. 7), and in the outsides of incised meanders along the Acre River south of this city (Fig. 8 and 9). The Solimo˜es Group vertebrate fossil collection at the Federal University of Acre was also visited (for a full species list, see Latrubesse et al., 1997). Although some taxa (e.g. crocodiles, turtles, fish) could have lived in a tidal estuary, the majority have been considered indicative of a terrestrial environment, probably a mixture of woodland and grassland, such as might be expected along a major river or wetland system (cf. Latrubesse et al., 1997).

3. The evolution history of the Amazon system: evidence from Acre and its regional context Tributaries of the modern Amazon originate in, or cross, the Andean foreland basin (Fig. 1). One might thus expect the sedimentary record in this basin to be a significant source of information about the evolution of the river system. However, no sufficiently detailed literature exists for most of this sequence, and the small parts of it that have been studied in detail have been the subject of major disputes about both its age (Fig. 3) and its depositional environment. This stable depocenter, sourced primarily from the Andes, was superseded in the relatively recent geological past by fluvial incision. Digital altimetry imagery (Fig. 4) indicates that the badland landscape observed in Acre state (cf. Figs. 2(b) and 3), caused by this young incision, covers most of western Amazonia. In the part of this foreland basin in NW Brazil, the thickness of Tertiary sediments reaches 1.8 km (e.g. de Matos and Brown, 1992; Latrubesse et al., 1997), with up to 1 km (Neogene [?]; e.g. Hoorn, 1994b) assigned to the Solimo˜es Group (thus deposited at a typical rate of 0.05 mm aK1, for 1 km deposition

in 20 Ma). It was thus a vast depositional system, with 1 km or more of sediment deposited over more than 2 million km2. As Latrubesse and Stevaux (2003b) have summarized, the literature on the Solimo˜es Group is full of contradictions. Frailey et al. (1988) proposed that its upper part indicates a lacustrine environment of latest Pleistocene or Early Holocene age (Fig. 3(a)). Subsequently, Kronberg et al. (1990, 1991) determined radiocarbon dates for Niteroi, one of the main Solimo˜es Group vertebrate sites in Acre state (on the Acre River, 60 km SW of Rio Branco; Fig. 2), obtaining an age in the range 40–50 ka. They suggest that aridity during the last climate cycle caused the lake level to fall, so mammals and reptiles died around isolated waterholes. The first study to contest these views appears to be that by Latrubesse (1992), on the grounds that the vertebrate species documented at Niteroi and other sites are much more ancient and pre-date the faunal influx that followed the establishment of the land bridge with North America in the Middle–Late Pliocene. Latrubesse (1992) interpreted the uppermost 70 m of the Solimo˜es Group exposed in Acre state as indicative of fluvial deposition, sourced from the Andes, during the latest Miocene and/or Early Pliocene (see also Latrubesse et al., 1997). The Solimo˜es Group in Acre state (Figs. 7–9) was reexamined by Ra¨sa¨nen et al. (1995), who accepted the Late Miocene age assignment by Latrubesse (1992) but claimed that it represented tidal, not fluvial, deposition. Ra¨sa¨nen et al. (1995) thus raised the possibility that in the Late Miocene a seaway may have extended along the entire length of the northern and central Andes, linking the western Caribbean Sea with the Atlantic Ocean off Argentina. This reinterpretation was based on the rhythmic character of the sediments (silt/ sand interbeds; Figs. 8 and 9); the identification of detailed textural characteristics that Ra¨sa¨nen et al. (1995) thought precluded the other obvious explanation for such cyclicity, namely, seasonal alternations between flood and slack-water conditions; and the tentative identification of a thickness modulation that Ra¨sa¨nen et al. (1995) thought marked the 14day cycle between spring and neap tides. According to Webb (1995), this hypothesis presented the first reliable evidence for the Middle Miocene ‘Paranese Seaway’ along the eastern flank of the Andes, which was previously considered

R. Westaway / Journal of South American Earth Sciences 21 (2006) 120–134

127

Fig. 9. Part of the exposure in the left bank of the Acre River at Seringal Amapa (Fig. 8). Rhythmic alternations of sand and clay are shown; the sand units being up to w50 cm thick. Ra¨sa¨nen et al. (1995, Figs. 2 and 3(a)) interpreted the sedimentary textures revealed as evidence of tidal sedimentation, in which each sand–mud couplet marks a single tidal cycle. An alternative explanation (Hoorn, 1996) is that each sand unit aggraded in a river floodplain during a seasonal flood. Each overlying clay bed can thus be interpreted as a mud drape deposited after the river current subsided but before the water level dropped. This process requires a mechanism to stabilize the local water level, such as the presence of a nearby seasonal lake, which is not implausible. No evidence was observed of plant root development in the silt units, suggesting they were not exposed subaerially for any significant length of time before the next cycle of sand deposition. Cross-bedding was not observed in the sand units, unlike the locality in Fig. 10.

hypothetical. However, as Hoorn (1996) has pointed out, others (notably, Hoorn, 1993, 1994b) had previously deduced intermittent marine influence in western Amazonia during the Middle Miocene. Hoorn (1993, 1994a,b) indeed presented sedimentological analyses and paleoenvironmental reconstructions using pollen, which indicated that the late Middle Miocene–early Late Miocene sediment exposed along the Amazon around Iquitos in NE Peru is predominantly fluvial/ lacustrine but has intermittent marine incursions, a view since reinforced by stable isotope studies (Vonhof et al., 1998; 2003). The key conclusion by Ra¨sa¨nen et al. (1995) was not that marine conditions existed in the Middle Miocene but that they supposedly existed in the Late Miocene, which Hoorn

(1996) considered contentious. Paxton et al. (1996) pointed out that the fish fauna in the sediments studied by Ra¨sa¨nen et al. (1995) indicates a freshwater environment, and Marshall and Lundberg (1996) argued that there is no geological evidence for a marine strait throughout the length of the Andes even in the Middle Miocene (a view that now seems less certain; see subsequent discussion), though a marine embayment open only to the north was plausible, as also suggested by Hoorn (1993, 1994b) and Hoorn et al. (1995). Ra¨sa¨nen and Linna (1996) responded to these criticisms by suggesting, for example, that if a marine embayment was plausible in the Middle Miocene but not the Late Miocene, then the mammalian biostratigraphy might be wrong, and the sediments exposed in Acre may be Middle Miocene and not Late

128

R. Westaway / Journal of South American Earth Sciences 21 (2006) 120–134

Miocene. Despite these difficulties, claims continue to be made that the sediments exposed in Acre indicate tidal marine environments in the Late Miocene (e.g. Hovikoski et al., 2003). Gingras et al. (2002a) also proposed on a similar basis that late Middle Miocene–early Late Miocene sediments exposed along the Amazon around Iquitos in NE Peru (near localities studied by Hoorn, 1993, 1994a,b) are tidal and not predominantly fluvial, thus questioning contrary stable isotope evidence (cf. Vonhof et al., 1998) due to possible calibration difficulties. Other recent work on this region (e.g. Hamilton et al., 2001) has cited Ra¨sa¨nen et al. (1995) but not their critics, suggesting that the view that the Solimo˜es Group in Acre represents tidal deposition has become quite widely accepted. However, Kronberg et al. (1998) obtained more radiocarbon dates from sediments of the Solimo˜es Group from both surface exposures in Acre and boreholes east of Iquitos (Fig. 1). They, therefore, restated the earlier argument (cf. Kronberg et al., 1990, 1991) that these sediments are Late Pleistocene, not Miocene. Kronberg et al. (1998) accepted that the Solimo˜es Group was deposited in fluvial environments similar to those of the modern Amazon (cf. Latrubesse et al., 1997) but dismissed the arguments for a Miocene age on the grounds that ‘No quantitative chronological controls for the ostensible Tertiary fossils are documented for any of the Amazonian or more southerly fossil occurrences’. Kronberg et al. (1998) used radiocarbon dates—such as 52,780G1980 and 51,950G1340 years—from depths of around 200 m in boreholes to estimate Late Pleistocene deposition rates of 4 mm aK1. However, even if the entire modern sediment load of the Amazon was trapped in the Solimo˜es Group depocenter, the spatially averaged sedimentation rate would be much less (1.2 billion tonnes per yearO2000 kg mK3 sediment densityO2 million km2 area gives a spatial average deposition rate of 0.3 mm aK1); such high deposition rates, spatially averaged across the region, would thus be implausible. The range of depths at which these radiocarbon ages have been obtained place them within sedimentary units that Hoorn (1993) had previously determined as Miocene from pollen analyses of adjacent boreholes. Like the previous set of radiocarbon dates, this new set thus appears to consist of limiting ages imposed by the technique. Most recently, Campbell et al. (2001) dated the Madre de Dios Formation, the lateral equivalent in eastern Peru, directly south of Acre (Fig. 2), of the upper part of the Solimo˜es Group exposed in Acre. They obtained Ar–Ar dates of 9.01G0.28 Ma near the base and 3.12G0.02 Ma near the top of the exposed section (Fig. 3(b)), which indicates that these sediments and the associated faunas are Late Miocene–earliest Pliocene and that the switch to incision in this region occurred around 3 Ma. Gingras et al. (2002b) also recently described an additional site, which they called Boca de Santa Pedro, on the Acre River, approximately 250 km upstream of Rio Branco, just east of Assis Brasil (Fig. 2; cf. Fig. 3(a)), where the Acre forms the Brazil-Bolivia border near the NW corner of Bolivia. Exposure at this site reveals interbedded sands and muds, similar to Seringal Amapa (Figs. 8 and 9). Gingras et al. (2002b) reported these sediments as Middle Miocene and interpreted the interbeds as reflecting seasonal variations in sediment input

into a channel; they nonetheless asserted that these deposits are marginal marine on the basis of apparent brackish water trace fossils and textural evidence attributed to tidal influence. The extent of bioturbation observed in the interbeds in this locality precludes the interpretation that they each represent semidiurnal tides. However, this reach of the Acre also was surveyed by Frailey et al. (1988), who considered the sediments exposed there to be equivalent to those in Peru that have been Ar–Ar dated (Fig. 3(a)), as well as by Latrubesse et al. (1997) who considered the sediments part of a fluvial unit emanating from the Andes and Late Miocene, according to mammal fauna, not tidal deposits that indicate a marine incursion from the north in the Middle Miocene. The Ar–Ar dating results of Campbell et al. (2001) contradict a Middle Miocene age for the sediments studied by Gingras et al. (2002b). Several lines of evidence have a bearing on whether the Solimo˜es Group sediments in Acre are fluvial or tidal and, indeed, whether they are Middle or Late Miocene. Sedimentology was emphasized in the field, but other relevant information comes from biostratigraphy and biogeography. Each of these arguments is summarized next. 3.1. Insights into Solimo˜es Group sedimentology from the June 2003 fieldtrip Several of the localities regarded as tidal by Ra¨sa¨nen et al. (1995) were reexamined in June 2003. A notable example is the riverbank section at Seringal Amapa near Rio Branco, which exhibits cyclic alternations between sand and silt deposition (Figs. 8 and 9). Ra¨sa¨nen et al. (1995) described this exposure in detail, so a description is omitted here. The principal argument made by Ra¨sa¨nen et al. (1995) in support of tidal deposition was that the transitions from deposition of sand to deposition of silt or mud are abrupt, whereas at the end of a seasonal flood, a more gradual fining of the sediment would be expected. However, the detailed sedimentary texture must depend on where one is within a channel system, as well as on the surrounding environment. For example, fine-grained sediment may be deposited not because a channel itself is in flood but because the water level is raised by a flood on a neighboring river or through fluctuations in the level of a lake farther downstream. A possibly analogous effect was observed from the air on the way to the field excursion, near Manaus (Fig. 1). The Amazon was locally in flood, due to high seasonal discharge from left bank tributaries located in the northern hemisphere, at a time of year when its right bank tributaries in the southern hemisphere were at a low stage, which in turn caused floodwater to back up along the right bank tributaries. In my opinion, no evidence was observed at Seringal Amapa of anything other than alternations of flood and slack water deposits, and other fluvial and lacustrine deltaic sediments were also observed elsewhere in the region (e.g. Fig. 7). With hindsight, several additional difficulties are apparent with the Ra¨sa¨nen et al. (1995) hypothesis. First, they estimated the tidal range at Seringal Amapa as 4 m. However, if this site was

R. Westaway / Journal of South American Earth Sciences 21 (2006) 120–134

2000 km from the mouth of a marine embayment leading from the nearly tideless Caribbean Sea, it is difficult to understand why it should have had such a strong tidal range. Second, if the sediments observed by Ra¨sa¨nen et al. (1995) are tidal, they require deposition rates of at least several tens of centimeters per day, or at least 100 m per year. The logic of assuming tidal deposition requires such rates to be maintained throughout the interior of the presumed marine embayment. If deposited at such rates, the whole 1000 m thickness of the Solimo˜es Group would have accumulated in just 10 years, which would create a major difficulty, because the required sediment influx would be vast for this brief interval of time, but nonexistent for the rest of the Late Cenozoic. Nonetheless, it could be argued that the preservation potential of the sediment deposited in any particular locality during any particular tidal cycle was low, and where this sediment is preserved, it represents brief intervals of local deposition separated by hiatuses that span most of this time scale. Third, the period of the modulation observed by Ra¨sa¨nen et al. (1995) is not well resolved due to the short rhythmic sequences they studied. It could contain 22 cycle, suggesting possible modulation of seasonal climate by the 22 year sunspot cycle, rather than 28 cycle indicative of modulation of semidiurnal tides by the fortnightly spring-neap tidal cycle. Aalto et al. (2003) indeed suggested that sedimentation in recent decades along the Beni River in NW Bolivia has experienced a seasonal modulation with a typical period of 8 years, caused by El Nin˜o climate oscillations. Finally, the upper part of the Solimo˜es Group was deposited around the time that the modern eastward drainage developed from the older northward drainage (e.g. Latrubesse et al., 1997; Westaway, 2005). However, tidal deposition would require the sedimentary surface east of the Andes to be at sea level, so there can have been no overall eastward topographic gradient for the incipient Amazon to follow. In contrast, the requirement for a topographic gradient in this sense means that the land surface east of the Andes must have been some distance above the contemporaneous sea level at the time when the modern Amazon system developed. Therefore, the apparently stable and long-lived depositional environment that existed around sea level until the Middle Miocene (or early Late Miocene) was replaced by net aggradation, raising the depositional surface above sea level in the Late Miocene. This aspect of this region’s evolution is considered further by Westaway (2005). 3.2. Biostratigraphic evidence In Europe, mammalian biostratigraphy is extremely useful in the study of Late Cenozoic sediments (e.g. Schreve, 2001) because appropriate taxa provide age control and indicate the environment. The value of this technique in application to the Solimo˜es Group is currently much more limited. The vertebrate fauna (e.g. Frailey, 1986; Rancy, 1989; Latrubesse et al., 1997) includes many species considered indicative of the Huayquerian and Montehermosan mammal biozones. In contrast to the impression created by Ra¨sa¨nen et al. (1995), who emphasized the aquatic species (notably, crocodiles, turtles, and fish), the taxa listed as represented also

129

include forest- and grassland-dwelling forms, such as might be expected in a subaerial region with forests in well-watered parts but grassland elsewhere, not in a tidal estuary (cf. Latrubesse et al., 1997). The Huayquerian is generally regarded as spanning 9–6 Ma (e.g. Madden et al., 1997) or 9–7 Ma (e.g. Cione and Tonni, 2001; Cione et al., 2000), whereas the Montehermosan spans the time from then until the Chapadmalalan stage that began at approximately 4 Ma and marks the initial influx of North American ‘waif’ species in the early Middle Pliocene. The subsequent main faunal influx followed the closure of the former Central American Seaway across the Panama Isthmus and is dated to 2.8–2.5 Ma (Marshall, 1988), with the corresponding influx of South American taxa into North America dated to 2.7–2.5 Ma (Lundelius, 1987). It is evident (e.g. Madden et al., 1997) that major problems exist with the pre-Middle Pliocene South American mammal biozones, because they lack proper stratotypes and are based on data from Argentina, where the climate would have been very different from northern South America. Nonetheless, these data suggest that by the latest Miocene or Early Pliocene, the former northward axial drainage system along the Andes had been disrupted and drainage was now generally eastward along the line of the incipient Amazon system. It can also be presumed that significant sedimentation ceased by around 4 Ma, as there is no evidence of North American taxa (e.g. sigmodontine rodents, which arrived in the Chapadmalalan; cf. Pardin˜as et al., 2002) in the uppermost part of the Solimo˜es Group in Acre state, and that the subsequent 70 m of incision revealed by river terrace staircases (Fig. 6) began shortly afterward (possibly by the Middle Pliocene, w3 Ma). Regarding age, Latrubesse et al. (1997) offered no resolution finer than Huayquerian–Montehermosan (i.e. w9–4 Ma); although Latrubesse (1992) implied that the top of the sequence in Acre is near the younger limit of the range. Many species are common to both the Huayquerian and Montehermosan mammal stages, the Solimo˜es Group indeed having species in common with other deposits considered Huayquerian elsewhere, such as the ‘Conglomerado osı´fero’ of northern Argentina (e.g. Negri and Ferigolo, 1999; Cione et al., 2000). However, Latrubesse et al. (1997) listed the toxodont Trigodon sp. as present in Acre. Trigodon gaudryi is considered (e.g. Cione and Tonni, 2001; Cione et al., 2000) as an index fossil for the Late Montehermosan (5–4 Ma), so if this attribution is confirmed at the species level, it would be extremely useful for age control, other vagaries in the taxonomy of the toxodonts (South American endemic herbivores) notwithstanding (cf. Nasif et al., 2000). Attempts to use lineages of other taxa, such as rodents, for age control, are hampered by the lack of comparative material (cf. Negri and Ferigolo, 1999). The best estimate from mammalian biostratigraphy, that these deposits in Acre span the latest Miocene–earliest Pliocene, is consistent with the dating in eastern Peru (Campbell et al., 2001). Regarding the environment, the mammalian fauna revealed in the Solimo˜es Group deposits in Acre includes species for which habitats can be inferred. Some have living relatives, such as the rodent Kiyutherium orientalis, well known from

130

R. Westaway / Journal of South American Earth Sciences 21 (2006) 120–134

Argentina and Uruguay (e.g. Francis and Mones, 1965; Cione et al., 2000), a relative of the capybara, and thus a terrestrial herbivore that lived near water (e.g. Pascual and Bondesio, 1985). The large mammal fauna consists mainly of notungulates, endemic herbivores now extinct as a group, which are believed to have adapted to ecological niches analogous to those populated in other continents by herbivores such as horses, bovids, and rhinoceroses. For example, the toxodont Trigodon (mentioned above) weighed an estimated 2 tonnes and was probably a grazer, analogous to a rhinoceros; in contrast, Litopterns were smaller, more horse-like, and thought to be adapted for browsing (e.g. Colbert et al., 1995). Some reported (cf. Frailey, 1986) taxa are of less clear environmental significance, such as Synastropotherium amazonense, another giant (w2 tonne) notungulate. Scott (1937, 1941) suggested that astrapotheres were aquatic, like the hippopotamus, and even though some of the supporting reasoning has been contested, this interpretation remains possible (cf. Wall and Heinbaugh, 1999), although Latrubesse et al. (1997) reported it as a forest or grassland dweller. Overall, this fauna is similar to others in Late Miocene fluvial environments elsewhere in South America, such as the ‘Conglomerado osı´fero’ (e.g. Cione et al., 2000), although the former is accepted as (?) Early Huayquerian, whereas the sequence in Acre may persist into the Montehermosan. The suggestion by Ra¨sa¨nen et al. (1995) that this fauna could have inhabited flats between tidal channels does not seem like a natural interpretation of the evidence. 3.3. Biogeography and molecular genetics evidence Ra¨sa¨nen et al. (1995) suggested that their reported evidence of marine conditions in the Late Miocene deposits of Acre may have implications for the evolution of aquatic taxa. Two key studies have since addressed this point: the investigation of potamotrygonid fish (i.e. freshwater stingrays) by Lovejoy et al. (1998) and that of iniid cetaceans (i.e. river dolphins) by Hamilton et al. (2001). Many studies (e.g. Cione et al., 2000) have noted fossils that resemble Amazonian river dolphins in Argentina, notably in the ‘Conglomerado ossı´fero’, suggesting a hydrological connection through the interior of South America. The closest living relative of the modern Amazon river dolphin Iniia geoffrensis is the La Plata river dolphin Pontoporia blainvillei from the Rio de la Plata estuary (Hamilton et al., 2001). By comparing evidence from molecular genetics and the fossil record, Hamilton et al. (2001) estimated that these two species began to differentiate around 13 Ma and therefore that the hydrological connection (whether marine, the so-called Paranese Sea, or via interconnected wetlands) ceased to exist by then. Hamilton et al. (2001) also established that the Bolivian river dolphin Iniia geoffrensis boliviensis (which inhabits the Madeira River system upstream of its reach through the Guapore´ Shield, Fig. 2) is a distinct subspecies from that which appears in the rest of the Amazon system, Iniia geoffrensis geoffrensis, the two having apparently been effectively isolated for several million years. This isolation

evidently resulted from the disruption of a former northwestward connection between the upper Madeira and the rest of the Amazon system along the Andean foreland basin, presumably by uplift of the land in between, providing a minimum age for the establishment of the present eastward transcontinental drainage in the Madeira. The timing of this isolation has been independently estimated as Late Pliocene (e.g. Dumont et al., 1991; Latrubesse et al., 1997), corresponding to the ‘Diaguita’ phase of uplift in the adjacent Andes. Hamilton et al. (2001) inferred from the fossil record that the Iniid dolphins entered the former South American internal seaway from the north. This river dolphin story thus supports the view of Hoorn et al. (1995) that western Amazonia was connected to the Caribbean Sea in the Middle Miocene and that the modern throughgoing eastward drainage of the Amazon is young, but it does not indicate whether the Acre region was at sea level during the Late Miocene (as suggested by Ra¨sa¨nen et al., 1995) or whether the last marine connection anywhere in western Amazonia occurred in the late Middle Miocene (as suggested by Hoorn et al., 1995). Further insights are provided by the consideration of freshwater stingrays. These distinctive fish are now found in Maracaibo Gulf and in the Magdalena, Orinoco, Amazon, and Parana/La Plata rivers (e.g. Lundberg, 1997), which are now isolated but were interconnected in the Middle Miocene if the drainage reconstructions of Hoorn et al. (1995) and the concept of the Paranese Seaway are correct. Lovejoy et al. (1998) estimated using molecular genetics that the Amazonian freshwater stingrays began to differentiate around 25 Ma from modern marine species (found in the Caribbean and the Pacific coast of central America) and then began to radiate further, around 14 Ma, into the present range of species. This pattern thus also provides strong evidence that modern western Amazonia developed from a marine embayment that was connected to the Caribbean in the Early–Middle Miocene (cf. Hoorn et al., 1995) but that this throughgoing marine connection was disrupted (presumably into wetlands) by the end of the Middle Miocene. It does not support the idea (cf. Ra¨sa¨nen et al., 1995) that such a marine connection persisted in Acre well into the Late Miocene; if it had, the species radiation would have begun much later and would have not had time to reach its present diversity. Fossil potamotrigonid tail stings have indeed been found in the Solimo˜es Group deposits in Acre (e.g. Frailey, 1986; Lundberg, 1997), consistent with a freshwater origin (cf. Paxton et al., 1996) and consistent with the view that these deposits postdate the isolation of this former marine depocenter from the Caribbean. 3.4. Deductions from this evidence The biogeographic and molecular genetic evidence strongly supports the view that marine conditions existed in western Amazonia in the Middle Miocene but ceased before the end of this era, consistent with the evidence provided by sediments of this age (cf. Hoorn, 1993, 1994a, b). Isotopic analysis (Vonhof et al., 2003) suggests a similar interpretation; in the Middle

R. Westaway / Journal of South American Earth Sciences 21 (2006) 120–134

Miocene, western Amazonia was connected to the sea in the north, but the restricted character of this marine connection, together with the magnitude of the freshwater influx from the Andes and other adjacent land areas, led to the development of a freshwater environment inhabited by endemic descendents of marine taxa. Notwithstanding the difficulties noted regarding precise age calibration, the biostratigraphy of the sediments in Acre suggests deposition in the latest Miocene and possibly also the earliest Pliocene, making it younger than the time span for which marine conditions can be expected from the biogeography. This biostratigraphy is indeed entirely consistent with a fluvial, or wetland, environment. As already noted, the alternative suggestion (cf. Ra¨sa¨nen et al., 1995)—that this fauna could have inhabited tidal flats—does not seem like a natural interpretation of this evidence. In principle, the rhythmic character of the sediments at Seringal Amapa (Figs. 8 and 9) could represent either semidiurnal tidal cycles or annual cycles of flooding. Taken in isolation, the sedimentary textures observed by Ra¨sa¨nen et al. (1995) could indicate tidal deposition, or instead seasonal flooding (cf. Hoorn, 1996), a view that is reinforced by the

Fig. 10. Cartoons indicating a possible mechanism for the development of rhythmic alternation of sand (grey) and silt (black) observed in the Solimo˜es Group sediments in Acre (e.g. Fig. 8). A channel is assumed initially to have a low gradient (w0.03 m kmK1) along reach AB (estimated length, tens of km) that flattens to an even lower gradient downstream along BC. (a) During a summer flood, channel deposition concentrates around where the channel gradient decreases, producing a wedge of sand with a maximum thickness (B– D) of w2 m (Fig. 7) at B, which tapers upstream over tens of km. (b) The channel bed gradient is reversed over reach AD, and the channel ponds, facilitating silt deposition during the subsequent winter. (c) The modification to the channel bed gradient created in (a) concentrates sand deposition in the next summer farther upstream than before, with its maximum thickness at E, upstream of D. During the following winter, silt deposition concentrates upstream of E. A cyclic alternation of sand and silt results, until the overall aggradation is so thick that the channel migrates elsewhere and a similar sequence is reestablished.

131

similarity between the facies observed in sections in the Solimo˜es Group (e.g. Fig. 7) and those created by modern sedimentary processes related to seasonal flooding (cf. Latrubesse and Kalicki, 2002; Latrubesse and Rancy, 1998). Furthermore, given that they are Late Miocene or earliest Pliocene, the view that these sediments are marine is inconsistent with, or permits no natural interpretation of, the remaining non-sedimentological evidence regarding the regional context. Latrubesse (1992; cf. Latrubesse and Stevaux, 2003b; Latrubesse et al., 1997) regarded the Solimo˜es Group deposits in Acre as transported ENE from the Peruvian Andes by rivers or low-angle alluvial fans under the influence of a strongly seasonal climate in which flooding alternated with slack water deposition. This description seems to be the most satisfactory to date of the depositional environment. Supporting evidence for a strongly seasonal paleoclimate in western Amazonia is provided by stable isotope studies of fluvial molluscs from the Solimo˜es Group deposits in NE Peru and SE Colombia (Kaandorp et al., 2005). As already noted, these deposits are Middle Miocene–early Late Miocene (cf. Hoorn, 1994b) and thus older than the deposits visited in Acre state. Nonetheless, because the climate of western Amazonia was already strongly seasonal in the Middle Miocene and remains strongly seasonal at present, it is reasonable to infer that it was strongly seasonal in the late Late Miocene–Early Pliocene, when the Solimo˜es Group sediments in Acre were deposited. However, the depositional paleoenvironment of the Solimo˜es Group differed from that of the modern Amazon, because a stacked sedimentary sequence was aggrading, whereas the modern Amazon is essentially a conduit that transports material eroded from the Andes to the Atlantic Ocean, notwithstanding temporary sediment storage within the system (cf. Meade et al., 1985). Modern floodplains in the Amazon foreland basin are sites of net aggradation as a result of deposition during seasonal flooding (cf. Latrubesse and Kalicki, 2002; Latrubesse and Rancy, 1998; Aalto et al., 2002, 2003). However, other processes, such as bank collapse and channel migration (cf. Latrubesse and Kalicki, 2002; Latrubesse and Rancy, 1998), rework material back into the active channels. As already noted, Figs. 3 and 4 indicate that, in most of western Amazonia (including the Beni River in NW Bolivia; cf. Aalto et al., 2002, 2003), rivers have incised significantly during the Quaternary into older stacked fluvial sequences. This geomorphology indicates that on long timescales, the region is experiencing net erosion, not net deposition. It can thus be presumed, by analogy with observations along many rivers in Europe (cf. Bridgland, 1994; Lewis et al., 2004), that much of the Holocene sediment accumulation in floodplains (e.g. along the Beni in NW Bolivia; cf. Aalto et al., 2002, 2003) will be eroded and reworked downstream during incision in future climate cycles and will thus not form part of the long-timescale geological record. The modern Amazon river system has a very low gradient, roughly 0.03 m kmK1 (w100 m/w3000 km). If the Solimo˜es Group river system was similar, the 0.5 m (Figs. 8 and 9) to 2 m (Fig. 7) thicknesses of sand inferred as locally deposited in successive summer floods could disrupt the gradients of both

132

R. Westaway / Journal of South American Earth Sciences 21 (2006) 120–134

the trunk and tributary channels for tens of kilometers upstream. Fig. 10 illustrates this mechanism schematically. The resulting rapid aggradation would promote instability, leading to the lateral migration of channels. The fine-scale sedimentary laminations interpreted as tidal (cf. Gingras et al., 2002b) may instead relate to ‘repiquetes’, or rapid fluctuations in the water level (and thus current) over a few days (cf. Paxton et al., 1996). Repiquetes reach approximately 2 m on the main Amazon channel (e.g. Paxton et al., 1996) but only 30 cm on the Purus (Latrubesse and Kalicki, 2002). Their magnitude in the Late Miocene is unknown, but some effect can be anticipated, superimposed onto larger magnitude seasonal variations. A possible present-day analog for this environment might be the Ganges plains in the Himalayan foreland basin, where both stacked aggradation and channel migration are observed (cf. Jain and Sinha, 2003), although the typical channel gradients (w0.3 m kmK1) and time-averaged deposition rates (w1 mm aK1) in this region are an order of magnitude greater than the values estimated for the Solimo˜es Group. Another possible analog is the large rivers of the eastern Siberian Arctic, such as the Lena and Kolyma (cf. Alekseev and Drouchits, 2004; Patyk-Kara and Postolenko, 2004). Like the Amazon, they have extremely low gradients (w0.05 m kmK1), and they likewise experience strongly seasonal flow and sediment transport, which can cause the channel systems to become choked with their own sediment. However, in these cases, the direct effects of the low river gradients on seasonal sedimentation are supplemented by other processes related to cryogenic conditions, such as the formation of ice dams, which would not have affected the Solimo˜es Group. That such aggradation persists in the Himalayan foreland basin but was superseded in the Solimo˜es Group depocenter in the Andean foreland basin in the (?) latest Miocene–Early Pliocene may relate to differences in the crustal properties. The Himalayan foreland basin is thought to be underlain by ancient (Archaean) crust that lacks a mobile lower-crustal layer, whereas the Andean foreland basin is underlain by younger crust, which permits more variable patterns of vertical crustal motions (cf. Westaway et al., 2003). The depositional environment of the Solimo˜es Group thus may have no precise modern fluvial analog, but the dynamic instabilities that can be anticipated during aggradation of a stacked fluvial sequence with the conditions of an extremely low river gradient and strongly seasonal climate seem worth investigating further to determine if they can account for the specific textural details that have been attributed to tidal deposition by Ra¨ sa¨ nen et al. (1995). However, such investigations are beyond the scope of this study, the aim of which is instead to access the use of these fluvial data to investigate large-scale topography. For this purpose, these data indicate that in the Late Miocene–earliest Pliocene, the land surface in Acre was not at sea level; instead, it was already above sea level by an unknown distance, having probably been uplifted from sea level near the end of the Middle Miocene.

4. Conclusions The IGCP 449 fieldtrip to western Amazonia in June 2003 has drawn attention to the Late Cenozoic fluvial sequences in this region, which had previously been obscure. In Acre state, underlain by relatively mobile crust, rivers have incised by up to 70 m into the stacked latest Miocene (?)–Early Pliocene (?) sediments of the Solimo˜es Group, creating staircases of fluvial terraces, as noted by Latrubesse et al. (1997), and indicating regional uplift on this time scale. In contrast, in western Rondonia state, the Madeira flows through the Early Proterozoic western part of the Amazon Craton. Late Cenozoic vertical crustal motions seem minimal, as might be anticipated from the expected much greater stability of this more ancient crust (Westaway et al., 2003). The evidence in Acre state suggests that the Solimo˜es Group was deposited by an ancestral river system associated with the incipient development of the modern eastward Amazon drainage, as also proposed by Latrubesse et al. (1997). Acknowledgements Financial support for travel to and fieldwork in Brazil was provided by the Royal Society. The author thanks David Bridgland for his encouragement to participate and Edgardo Latrubesse and Jose´ Stevaux for the effectiveness of their organization of this visit to such a daunting region. He also thanks many other participants for stimulating discussions in the field, though the views expressed in this paper are the author’s alone. Rolf Aalto, Allen Archer, and an anonymous reviewer provided thoughtful and constructive reviews. This study contributes to IGCP 449 ‘Global Correlation of Late Cenozoic Fluvial Deposits’ and to IGCP 518 ‘Fluvial sequences as evidence for landscape and climatic evolution in the Late Cenozoic’. References Aalto, R., Dunne, T., Nittrouer, C.A., Maurice-Bourgoin, L., Montgomery, D.R., 2002. Fluvial transport of sediment across a pristine tropical foreland basin; channel—flood plain interaction and episodic flood plain deposition. In: Dyer, F.J., Thoms, M.C., Olley, M. (Eds.), The Structure, Function and Management Implications of Fluvial Sedimentary Systems, vol. 276. International Association of Hydrological Sciences Publication, Wallingford, UK, pp. 339–344. Aalto, R., Maurice-Bourgoin, L., Dunne, T., Montgomery, D.R., Nittrouer, C.A., Guyot, J.-L., 2003. Episodic sediment accumulation on Amazonian flood plains influenced by El Nin˜o/Southern Oscillation. Nature 425, 493–497. Alekseev, M.N., Drouchits, V.A., 2004. Quaternary fluvial sediments in the Russian Arctic and Subarctic: Late Cenozoic development of the Lena river system, northeastern Siberia. Proceedings of the Geologists’ Association 125, 339–346. Bibus, E., 1983. Reliefgenerationen am oberen Paraguai in Mato Grosso (Brasilien). Zeitschrift fu¨r Geomorphologie N.F. Supplement 48, 261–274. Bridgland, D.R., 1994. The Quaternary of the Thames, Geological Conservation Review Series, vol. 7. Chapman & Hall, London p. 441. Campbell, K.E., Heizler, M., Frailey, C.D., Romero-Pittman, L., Prothero, D.R., 2001. Upper Cenozoic chronostratigraphy of the southwestern Amazon Basin. Geology 29, 595–598.

R. Westaway / Journal of South American Earth Sciences 21 (2006) 120–134 Cione, A.L., Tonni, E.P., 2001. Correlation of Pliocene to Holocene southern South American and European vertebrate-bearing units. Bolletino de la Societa´ Paleontologica Italiana 40, 167–173. Cione, A.L., Azpelicueta, M., de las, M., Bond, M., Carlini, A.A., Casciotta, J.R., Cozzuol, M.A., de la Fuente, M., Gasparini, Z., Goin, F.J., Noriega, J., Scillato-Yane´, G.J., Soibelzon, L., Tonni, E.P., Verzi, D., Vucetich, M.G., 2000. Miocene vertebrates from entre rı´os province, eastern argentina. El Neo´geno de Argentina, vol. 14, pp. 191. Colbert, E.H., Morales, M., Minkoff, E.C., 1995. Colbert’s evolution of the vertebrates: a history of the backboned animals through time, fifth ed. Wiley, New York, p. 576. de Matos, R.M.D., Brown, L.D., 1992. Deep seismic profile of the Amazonian Craton (northern Brazil). Tectonics 11, 621–633. DeCelles, P.G., Hertel, F., 1989. Petrology of fluvial sands from the Amazonian foreland basin, Peru and Bolivia. Geological Society of America Bulletin 101, 1552–1562. Dobson, D.M., Dickens, G.R., Rea, D.K., 2001. Terrigenous sediment on Ceara Rise: a Cenozoic record of South American orogeny and erosion. Palaeogeography, Palaeoclimatology, Palaeoecology 165, 215–229. Dumont, J.F., Deza, E., Garcia, F., 1991. Morphostructural provinces and neotectonics in the Amazonian lowlands of Peru. Journal of South American Earth Sciences 4, 373–381. Frailey C.D. 1986. Late Miocene and Holocene mammals, exclusive of the Notoungulata, of the Rio Acre region, western Amazonia. Los Angeles County Museum Contributions in Science, 374, 46 pp. Frailey, C.D., Lavina, E.L., Rancy, A., de Souza Filho, J.P., 1988. A proposed Pleistocene/Holocene lake in the Amazon basin and its significance to Amazonian geology and biogeography. Acta Amazonica 18, 119–143. Francis, J.C., Mones., A.,1965. Sobre el hallazgo de Kiyutherium orientalis n.g, n.sp. (Rodentia, Hydrochoeridae) en la Formacio´n Kiyu´, de las Barrancas de San Gregorio, Departamento de San Jose´, Repu´blica Oriental del Uruguay. Kraglieviana (Montevideo), 1, 45–54. Gingras, M.K., Rasanen, M.E., Pemberton, S.G., Romero, L.P., 2002a. Ichnology and sedimentology reveal depositional characteristics of baymargin parasequences in the Miocene Amazonian foreland basin. Journal of Sedimentary Research 72, 871–883. Gingras, M.K., Ra¨sa¨nen, M., Ranzi, A., 2002b. The significance of bioturbated inclined heterolithic stratification in the southern part of the Miocene Solimoes Formation, Rio Acre, Amazonia Brazil. Palaios 17, 591–601. Gregory-Wodzicki, K.M., 2000. Uplift history of the central and northern Andes: a review. Geological Society of American Bullatin 112, 1091–1105. Guyot, J.-L., Jouanneau, J.M., Wasson, J.G., 1999. Characterisation of river bed and suspended sediments in the Rio Madeira drainage basin (Bolivian Amazonia). Journal of South American Earth Science 12, 401–410. Hamilton, H., Caballero, S., Collins, A.G., Brownell, J.r., R, L., 2001. Evolution of river dolphins. Proceedings of the Royal Society of London Series B 268, 549–556. Hoorn, C., 1993. Miocene incursions and the influence of Andean tectonics on the Miocene depositional history of northwestern Amazonia: results of a palynostratigraphic study. Palaeogeography, Palaeoclimatology, Palaeoecology 105, 267–309. Hoorn, C., 1994a. Fluvial palaeoenvironments in the Amazonas Basin (Early Miocene–early Middle Miocene, Colombia). Palaeogeography, Palaeoclimatology, Palaeoecology 109, 1–54. Hoorn, C., 1994b. An environmental reconstruction of the palaeo-Amazon river system (Middle–Late Miocene, NW Amazonia). Palaeogeography, Palaeoclimatology, Palaeoecology 112, 187–238. Hoorn, C., 1996. Comment on ‘Late Miocene tidal deposits in the Amazonian foreland basin’ by Ra¨sa¨nen, M., Linna, A.M., Santos, J.C.R., Negri, F.R.,. Science 273, 122–123. Hoorn, C., Guerrero, J., Sarmiento, G.A., Lorente, M.A., 1995. Andean tectonics as a cause for changing drainage patterns in Miocene northern South America. Geology 23, 237–240. Hovikoski, J., Ra¨sa¨nen, M., Gingras, M., 2003. Late Miocene tidal bundle sequences in the Solimoes Formation, Acre, Brazil: preliminary notes. In: de F. Rossetti, D., Truckenbrot, W., Goes, A.M. (Eds.), Third Latin

133

American Congress of Sedimentology, Belem, Brazil, 8–11 June 2003, Abstract Volume. Federal University of Para Press, Belem, Brazil, ISBN: 85-7098-100-7, pp. 271–273. Jain, V., Sinha, R., 2003. River systems in the Gangetic Plains and their comparison with the Siwaliks: a review. Current Science 84, 1025–1033. Kaandorp, R.J.G., Vonhof., H.B., Wesselingh, F.P., Romero Pitman, L., Kroon, D., van Hinte, J.E., 2005. Seasonal Amazonian rainfall variation in the Miocene climatic optimum. Palaeogeography Palaeoclimatology, Palaecology, 221, 1–6. Kronberg, B., Benchimol, R., Bird, M., 1990. Evidence for SW Amazonia lake system drying up—50000 yr BP. Special Publication 2, PICG-218, UNESCO, Medellin, Colombia, 11 pp. Kronberg, B., Benchimol, R., Bird, M., 1991. Geochemistry of Acre subbasin sediments: window on ice-age Amazonia. Interciencias 3, 138–141. Kronberg, B.I., Fralick, P.W., Benchimol, R.E., 1998. Late Quaternary sedimentation and palaeohydrology in the Acre foreland basin, SW Amazonia. Basin Research 10, 311–323. Latrubesse, E., 1992. El Cuaternario fluvial de la cuenca del Purus en el Estado de Acre, Brasil. PhD thesis, National University of San Luis. San Luis, Argentina, 219 pp. Latrubesse, E., Kalicki, T., 2002. Late Quaternary palaeohydrological changes in the upper Purus basin, southwestern Amazonia, Brazil. Zeitschrift fu¨r Geomorphologie NF Supplement 129, 41–59. Latrubesse, E., Rancy, A., 1998. The Late Quaternary of the upper Jurua´ River, southwestern Amazonia, Brazil: geology and vertebrate palaeontology. In: Rabassa, J., Salemme, M. (Eds.), Quaternary of South America and Antarctic Peninsula. Balkema, Rotterdam, pp. 27–46. Latrubesse, E., Stevaux, J., 2003a. Southwestern Amazonia. In: Latrubesse, E., Stevaux, J. (Eds.), GLOCOPH and IGCP 449 Field Conference Guide—Amazon, 2003. Federal University of Goias Press, Goiania, Brazil, pp. 6–13. Latrubesse, E., Stevaux, J., 2003b. The Solimo˜es Formation. In: Latrubesse, E., Stevaux, J. (Eds.), GLOCOPH and IGCP 449 Field Conference Guide—Amazon, 2003. Federal University of Goias Press, Goiania, Brazil, pp. 14–18. Latrubesse, E., Stevaux, J., 2003c. The Acre River. In: Latrubesse, E., Stevaux, J. (Eds.), GLOCOPH and IGCP 449 Field Conference Guide—Amazon, 2003. Federal University of Goias Press, Goiania, Brazil, p. 46. Latrubesse, E., Bocquentin, J., Santos, J.C.R., 1997. Ramonell, Paleoenvironmental model for the Late Cenozoic of southwestern Amazonia: paleontology and geology. Acta Amazonica 27, 103–118. Latrubesse, E.M., Rossi, A., Franzinelli, E., 2000. Geomorphology of the Pacaa´s Novos Range, southwestern Amazonia, Brazil: One example on the importance of geomorphological evidences to the reconstruction of Quaternary paleoenvironmental scenarios in Amazonia. Revista Brasileira de Gioscieˆncias 30, 513–517. Lewis, S., Maddy, D., Glenday, S., 2004. The Thames Valley sediment conveyor: Fluvial system development over the last two interglacial–glacial cycles. Quaternaire 15, 17–28. Lovejoy, N.R., Bermingham, E., Martin, A.P., 1998. Marine incursion into South America. Nature 396, 421–422. Lundberg, J.G., 1997. Freshwater fishes and their paleobiotic implications. In: Kay, R.F., Madden, R.H., Cifelli, R.L., Flynn, J.J. (Eds.), Vertebrate Paleontology in the Neotropics: The Miocene Fauna of La Venta, Colombia. Smithsonian Institution Press, Washington DC, pp. 67–91. Lundelius Jr., E.L., 1987. The North American Quaternary sequence. In: Woodburne, M.O. (Ed.), Cenozoic Mammals of North America. University of California Press, Berkeley, CA, pp. 211–235. Madden, R.H., Guerrero, J., Kay, R.F., Swisher III., C.C., Walton, A.H., 1997. The Laventan stage and age. In: Kay, R.F., Madden, R.H., Cifelli, R.L., Flynn, J.J. (Eds.), Vertebrate Paleontology in the Neotropics: The Miocene Fauna of La Venta, Colombia. Smithsonian Institution Press, Washington DC, pp. 499–519. Marshall, L.G., 1988. Land mammals and the Great American Interchange. American Scientist 76, 380–388. Marshall, L.G., Lundberg, J.G., 1996. Comment on ‘Late Miocene tidal deposits in the Amazonian foreland basin’ by Ra¨sa¨nen, M., Linna, A.M., Santos, J.C.R., Negri, F.R. Science 273, 123–124.

134

R. Westaway / Journal of South American Earth Sciences 21 (2006) 120–134

Meade, R.H., Dunne, T., Richey, J.E., Santos, U., de M. Salati, E., 1985. Storage and remobilization of suspended sediment in the Lower Amazon River of Brazil. Science 228, 488–490. MME, 2000. Zoneamento ecolo´gico-econoˆmico da regia˜o fronteiric¸a BrasilBolı´via, mapa geomorfolo´gico integrado, 1:500,000 scale, NW part, sheet 3a. Ministe´rio de Minas e Energia, Servic¸o Geolo´gico do Brasil, Brasilia. Nasif, N.L., Musalem, S., Cerden˜o, E., 2000. A new Toxodont from the Late Miocene of Catamarca, Argentina, and a phylogenetic analysis of the Toxodontidae. Journal of Vertebrate Paleontology, 591–600. doi:10.1043/ 0272-4634. Negri, F.R., Ferigolo, J., 1889. Anatomia craniana de Neoepiblema ambrosettianus (Ameghino, ) (Rodentia, Caviomorpha, Neoepiblemidae) do Mioceno Superior—Plioceno, Estado do Acre, Brasil, e revisa˜o das espe´cies do geˆnero. Boletim do Museu Paraense Emı´lio Goeldi, Se´rie Cieˆncias da Terra 11, 3–80. Noble, D.C., Vogel, T.A., Peterson, P.S., Landis, G.P., Grant, N.K., Jezek, P.A., McKee, E.H., 1984. Rare-element-enriched, S-type ash-flow tuffs containing phenocrysts of muscovite, andalusite, and sillimanite, southeastern Peru. Geology 12, 35–39. Pardin˜as, U.F.J., D’Elia, G., Ortiz, P.E., 2002. Sigmodontinos fo´siles (Rodentia, Muroidea, Sigmodontinae) de Ame´rica del Sur: Estado actual de su conocimiento y prospectiva. Mastozoologia Neotropical Journal of Neotropical Mammals 9, 209–252. Pascual, R., Bondesio, P., 1985. Mamiferos terrestres del Mioceno MedioTardo de las cuencas de los Rios Colorado y Negro (Argentina): evolucion ambiental. Ameghiana (Buenos Aires) 22, 133–145. Patyk-Kara, N.G., Postolenko, G.A., 2004. Structure and Cenozoic evolution of the Kolyma river valley, eastern Siberia, from its upper reaches to the continental shelf. Proceedings of the Geologists’ Association 125, 325–338. Paxton, C.G.M., Crampton, W.G.R., Burgess, P., 1996. Comment on ‘Late Miocene tidal deposits in the Amazonian foreland basin’ by Ra¨sa¨nen, M., Linna, A.M., Santos, J.C.R., Negri, F.R.. Science 273, 123. Rancy, A., Boquentin Villanueva, J., Pereira de Souza Filho, J., Santos, J.C.R., Negri, F.R., 1989. Lista preliminar da fauna do Neo´geno da regia´o oriental do estado do Acre, Brasil (Material depositado em Rio Branco). VII Jornadas Argentinas de Paleontologı´a de Vertebrados. Ameghiniana (Buenos Aires) 26, 249. Ra¨sa¨nen, M., Linna, A.M., 1996. Reply to comments by C, Hoorn, by C.G.M. Paxton, W.G.R. Crampton, and P. Burgess, and by L.G. Marshall and J.G. Lundberg, on Late Miocene tidal deposits in the Amazonian foreland basin. Science 273, 124–125.

Ra¨sa¨nen, M., Salo, J., Kalliola, R.J., 1987. Fluvial perturbance in the western Amazon Basin: regulation by long-term sub-Andean tectonics. Science 238, 1398–1401. Ra¨sa¨nen, M., Neller, R., Salo, J., Jungner, H., 1992. Recent and ancient fluvial deposition systems in the Amazonian foreland basin. Peru, Geological Magazine 129, 293–306. Ra¨sa¨nen, M., Linna, A.M., Santos, J.C.R., Negri, F.R., 1995. Late Miocene tidal deposits in the Amazonian foreland basin. Science 269, 386–389. Schreve, D.C., 2001. Differentiation of the British late Middle Pleistocene interglacials: the evidence from mammalian biostratigraphy. Quaternary Science Reviews 20, 1693–1705. Scott, W.B., 1937. The Astrapotheria. Proceedings American Philosophical Society 77, 309–394. Scott, W.B., 1941. Part 5, Perissodactyla. In: Scott, W.B., Jepsen, G.L (Eds.), The mammalian fauna of the White River Oligocene Transactions of the American Philosophical Society, vol. 28, pp. 747–980. Texeira, W., Tassinari, C.C.G., Cordani, U.G., Kawashita, K., 1989. A review of the geochronology of the Amazonian Craton: tectonic implications. Precambrian Research 42, 213–227. Vonhof, H.B., Wesselingh, F.P., Ganssen, G.M., 1998. Reconstruction of the Miocene western Amazonian aquatic system using molluscan isotopic signatures. Palaeogeography, Palaeoclimatology, Palaeoecology 141, 85–93. Vonhof, H.B., Wesselingh, F.P., Kaandorp, R.J.G., Davies, G.R., van Hinte, J.E., Guerrero, J., Rasanen, M., Romero-Pittman, L., Ranzi, A., 2003. Paleogeography of Miocene western Amazonia; isotopic composition of molluscan shells constrains the influence of marine incursions. Geological Society of America Bulletin 115, 983–993. Wall, W.P., Heinbaugh, K.L., 1999. Locomotor adaptations in Metamynodon planifrons compared to other Amynodontids (Perissodactyla, Rhinocerotoidea). In: Santucci, V.L., McClelland, L., (Eds.) National Park Service Paleontological Research, vol. 4. Technical Report NPS/NRGRD/GRDTR99/03, U.S. National Park Service, Lakewood, Colorado, pp. 8–17. Available online from:http://www2.nature.nps.gov/geology/paleontology/ pub/grd4/nsp_paleo_vol4.pdf. Webb, S.D., 1995. Biological implications of the Middle Miocene Amazonian seaway. Science 269, 361–362. Westaway, R., in press. A coupled surface process/lower-crustal flow model for the Late Cenozoic evolution of the Andes/Amazon system. Global and Planetary Change, in press. Westaway, R., Bridgland, D., Mishra, S., 2003. Rheological differences between Archaean and younger crust can determine rates of Quaternary vertical motions revealed by fluvial geomorphology. Terra Nova 15, 287–298.