Evolution of the Western Amazon Lowland Relief: Impact of Andean Foreland Dynamics

Evolution of the Western Amazon Lowland Relief: impact of Andean foreland dynamics

M.E. Rasanen*, J.S. Salo**, H. Jungnert & L. Romero Pittmantt *Department of Quaternary Geology; **Departmentof Biology, University of Turku, SF-20500 Turku, Finland; fRadiocarbon Datinx Laboratory, Unizwrsity of Helsinki, SF-001 70 Helsinki, Finland; tflnstituto Geolligica Mznero y Metalzirgico, Pablo Bernzudez 22 1, Lima 11, Peru

has also been proposed for the origin of ABSTRACT the terra firme deposits in the southwest- The congruency in the depositional origin and age of the uppermost ern Amazon (Campbell and Frailey, sedimentary strata forming non-flooded rainforest ground (terra firme) 1984; Campbell et al., 1985). Based on in the western and central Amazon lowlands is a much debated the same deposits a more recent sugges- tion is that the terra firme is derived from subject. Here we conclude from the study of remote sensing imagery sedimentation in a late Pleistocene- that active Andean foreland dynamics have played a major role in the Holocene delta-lake environment, in- evolution of the Plio-Pleistocene fluvial landscape in the western itiated after 36,000 yr RP and with the Amazon. Foreland dynamics have resulted in a terra firme composed of latest lacustrine phases occurring as re- late Tertiary alluvium and younger alluvial terraces and plains. In cently as at 2800 yr BP (Frailey et al., Peru, thermoluminescence and I4C dating show local aggradation of 1988). This ’Lago Amazonas’ would this younger alluvium between 180 and 30 ka. documented high have covered approximately the same The area in central Amazon as its predeces- age heterogeneity of the terra firme has implications for considerations sor ’Belterra lake’. Clearly, with such of the biogeography of the Amazon forest. large discrepancies in the geohistory of the Amazon, there is no sound basis on Terra Nova, 2,320-332 which to make meaningful deductions about the factors behind Amazon biogeography. INTRODUCTION All previous concepts of the origin of portant for studies on the biogeography the surficial beds follow the ‘layer-cake’ The topmost sedimentary deposits of the Amazon basin. In many cases concept, according to which homogene- forming the present non-flooded rain- (Haffer, 1985; Haffer and Fitzpatrick, ous deposits were formed in a uniform forest ground (terra firme, Sioli, 1984) in 1985) the biogeographical tradition still environment. Little attention has been the western and central Amazon basin relies on the assumption that in large paid to the possible diachronism of the have been widely reported as a rela- areas of the lower and central Amazon deposits. The Amazon basin has a long tively thin (10-40 m thick) surficial, (east of the Iquitos arch, Fig. l),the terra fluvial history (Klammer, 1984; Irion, horizontal, sandy lithostratigraphic firme deposits result from sedimenta- 1984b; Rasanen et al., 1987) and despite unit (Oppenheim, 1946; Kummel, 1948; tion in a large inland ’Belterra lake‘ their lithostratigraphic similarity the Almeida, 1Y74; Projeto RADAM- during the high Calabrian (early Pleisto- surficial sequences may differ substan- BRASIL, 1977; ONERN, 1977, 1Y84; cene) sea-level (Sombroek, 1966). Thus tially in age. This supposition is sup- Khobzi et a/., 1980; Eden et al., 1982; biogeographers have regarded the ported by studies invoking Quaternary DNPM, 1984; Campbell and Frailey, Quaternary history of the lowland fluvial origins for some of the central 1984; Campbell ct nl., 1985; Frailey rt nl., Amazon as geodynamically relatively Amazon non-flooded terra firme, 1988). The extensive documentation of stable, especially in the central Brazilian formed when the sea levels were higher the surficial lithostratigraphic unifor- Amazon, which has long acted as a than they are during the present inter- mity of the terra firiiic has created an model for the palaeogeography of the glacial (Klammer, 1984; Irion, 198413). In illusion of the presence of a wide- whole basin. The ’Belterra lake’- the western Amazon (west of the spread ‘surficial mantle‘ in the Amazon hypothesis was originally based on clay Iquitos arch, Fig. l), though, the land- (Campbell and Frailey, 1984; Frailey et sediments found up to 180 m above sea scape is mostly unaffected by Quater- a/., 1988). This view has given rise to level in the eastern Amazon. The lake nary eustatic sea-level fluctuations. In several controversial hypotheses on the was supposed to have covered the Ecuador, in the Andean forelands, the genesis of these beds, the first being major part of the lowland Amazon basin surficial beds have been attributed to Agassiz’s erroneous proposal that they below this elevation. More recently the fluvial deposition following late- are glacigenous (Agassiz, 1865). existence of ’Belterra lake’ has little Pliocene Andean tectonism (Baldock, The sedimentary history of the beds support (Irion, 1984a). 1985). In Peru, some of the beds are forming the tcwa firmr has been very im- Catastrophic Holocene flashflooding postulated to be products of more re-

320 4'

2 0'

Fig. 1. Major continental andsubmarinestructures in northwestern South America (Unesco, 1975;Jordan et al., 1983;Cunha, 1988). Thelocations of the structural highs, the Andean retroarcand thepericratonicand intracratonic basins areshown in the Amazon sedimentary basin. Dotting depicts the relative thickness oftheir Tertiary-Quaternary deposits. The flat subducting segment of the Nazca plate is thrown, and the oceanic ridges are depicted according to their 3000 m bathymetric contours.

321 cent Pleistocene fluvial sedimentation from the convergence of the Nazca sub- tions forming the 'Planalto rebaixado da under the influence of.Sub-Andean tec- oceanic plate under the continental Amazonia' in Brazil (Almekia, 1974; tonics (Rasanen et al., 1987). In south- South American plate and giving rise to Projeto RADAMBRASIL, 1977; DNPM, eastern Colombia (Hoorn, 1988), north- the Andean orogeny slowly developing 1984) the 'Monte and Sierra Realms' in western Brazil (Almeida, 1974) and eastwards. Along the Andean strike, Bolivia (Campbell rt al., 1985) and the northern Peru (Rasanen et al., in prep.), the deformed Sub-Andean foreland belt Madre de Dios (Oppenheim, 1946), In- the alluvium forming the major terra is widest in the western Amazon in apari, Pagorene, Iquitos (Onern, 1977, firme has an eastern provenance. In Peru, where the tectonic stress is re- 1984; Campbell and Frailey, 1984) or Colombia, these sediments have a prop- leased in a thin-skinned thrust and fold Ucayali (Kummel, 1948) Formations in osed Oligo-Miocene age, and only the belt with reverse and thrust faults Peru. The unconformities exhibit ab- Quaternary terra firme terraces have dipping westwards. rupt changes in degree of consolidation obtained their material from the Andes The shallow subducting segment of and weathering and, at the marginal (Hoorn, 1988). the Nazca plate (Jordan et al., 1983) has parts of the foreland basins, even in the Thus it seems clear that over an area caused the present tectonic situation in dip of the strata (Oppenheim, 1975; of continental size such as the western the Peruvian foreland (Fig. 1). The flat Frailey, 1986). Unconformities are more and central Amazon, which has subduction is considered to have been parallel in the central parts of the basins, affinities with the Andes and the initiated only 10 to 5 Ma (Isacks, 1988, where differences in facies, degree cratonic shields, different parts of the Barberi rt a!., 1988) due to subduction of of weathering and consolidation are surficial deposits forming the terra firmr the Nazca aseismic ridge (Fig. 1).The in- usually less marked (Fig. 2). have aggraded diachronously. Sedi- itiation has been dated by the cessation In places, 0.2-2 m-thick conglomer- mentation has been controlled by Mio- of volcanism of the Andes along this ates occur above the basal unconformity, Pleistocene uplift and deformation of segment and by the reconstructed sub- with large-scale trough crossbedding in the Andes and their foreland and also duction geometry of the Nazca ridge the intercalated sands. These conglom- by eustatic sea-level changes in the (Pilger, 1984). This change in subduc- erates are frequently strongly cemented central Amazon. Although foreland de- tion geometry is among the most recent by iron oxides. The conglomerates are formation covers only a restricted area along the Cordilleran trench, and the the source beds for the majority of the of the western Amazon, it affects the low dip of the Benioff zone evidently Plio-Pleistocene vertebrate fossils and headwaters of the Amazon drainage effectively directs the compressional fossilized wood found in the area system, and hence Andean dynamics forces to the foreland crust (Suarez et al. , (Simpson and Paula Couto, 1981). Fre- have strongly modified both the 1983). The present high relative rate of quently documented in the western palaeohydrological conditions and sur- subduction, up to 11 cm yr-' at the Amazon, the conglomerates seem to face erosional patterns in the Amazon trench (Minster et al., 1974), further pro- phase out eastwards towards the lowlands. For this reason, the terra firme motes deformation (Jordan et d.,1983). central Amazon (Silva, 1988). The fauna forest beds are considerably more The foreland shortening is estimated to in the conglomerates seem to be some- heterogeneous in origin and age than have been recently in the range of 1-2 what younger (Simpson and Paula biogeographers have previously mm yr-' (Suarez et al., 1983). Deforma- Couto, 1981) than the fauna found in assumed. tion is further evidenced by modern the Oligo-Pliocene Pebas Formation Our evidence supporting this view is upper plate seismicity in the foreland, (Williams 1949) that is frequently based on analyses of side-scanning with the highest activity along the reported to underly the surficial forma- radar (SLAR) images of Peru and Brazil Cordilleras (Jordan et al., 1983; Suarez et tions in the western Amazon. and Landsat Planimetric Maps and on a/., 1983; Cisternas et af., 1988). The uppermost parts of the 10-40 m- site-specific Landsat Thematic Mapper thick beds (above the basal conglomer- (TM) imageries of the Peruvian area. ates) constitute a sequence with varying THE TERRA FIRME DEPOSITS Ground-truth stratigraphic data with numbers of fluvial cycles (multistoried thermoluminescence (TL) and I4C age aggradational cycles) often composed of General stratigraphy determinations were collected to date more sandy material than below the the beds of terra firrne in the Peruvian In the Pastaza-Marafion, Ucayali, basal unconformities (Fig. 2). The indi- Amazon. Madre de Dios-Beni and Acre intrafore- vidual upward-fining fluvial cycles are land basins the surficial sandy beds of predominantly composed of channel terra firnie are reported at erosional lag (coarse sand, gravel), pointbar (fine SHALLOW SUBDUCTION banks, the majority of which are carved to medium sand) and overbank facies MAINTAINS DYNAMIC into terraces along present-day rivers. (silty-clayey) with occasional, weakly FORELAND BASINS They are frequently reported to rest un- developed palaeosols. The western Amazon lowland (< 500 m conformably on older Quaternary and a.s.1.) is structurally composed of at Tertiary deposits as various structural Deposition of the alluvium least four large retroarch intraforeland or surficial formations: the 'Terciario basins, delineated by emerging struc- Superior Amazonico' and 'superficia de The original aggradational surface of tural highs (arches, Fig. 1). The evolu- arcila' composed of 'Trinidad Beds' in the surficial beds has been preserved in tion of the Andean foreland is controlled Colombia (Khobzi et nl., 1980; Eden ct the central parts of the Madre de Dios- by W-E compressive forces originating al., 1982), the I@ or Sanozama Forma- Beni and Ucayali basins in Peru. Here

322 Fig. 2. Stratigraphy of surficial alluvial beds in a terrace deposit with a multistoried sequence of upwavdfiningfluvial cycles (a) exposed in a roughly 15 m-high erosional bank along the Madre de Dios (22"15'50 S, 70"48'25" W).The lower part of the bank 1s covered by colluvium (b). The basal unconformity (c) to older moreconsolidatedand?nt-lyingfine-graineddeposzts (d) is parallel, and the basal conglomerate is poorly developed. Thefluvial cycles of the surficial sandy beds were originally inclined (e). the terra firme relief also reveals aban- depocentre migration, and we interpret (Sombroek, 1966; Campbell and Frailey, doned channels, indicating that the the basal conglomerates as having de- 1984; Frailey ef al., 1988), these forma- material was deposited by the laterally posited during major stages of incision. tions of Western Amazonia do not have mobile channels of meandering rivers As the present-day Madre de Dios is documented lateral continuity and can- (IFG, 1984). In the Madre de Dios-Beni incised and deposit resembling con- not be correlated across the foreland basin, terraces have developed asym- glomerates, incision may have occurred either lithostratigraphically or tempor- metrically along the northeastern side during interglacials. The overlying ally. Older sediments are exposed along of the Madre de Dios (Rasanen et al., fluvial cycles in the terraces were de- the emerged areas (Fitzcarrald, Serra do 1989, whereas the southwestern side of posited when the river system was Moa and Vaup6s arches) (ONERN, the river shows younger Pleistocene more aggrading (certain phases of glaci- 1977; Hoorn, 1988) owing to gradual and Holocene aggradational surfaces ations?). From the way the basin is de- offlap (regressive overlap) of the strata. partly formed by the transversal rivers formed, the Madre de Dios river system These sediments were deposited in draining the Andes (IFG, 1984; DeCelles was more ready to relocate its course rather similar fluvial environments, and and Hertel, 1989). These geomorpho- southwestwards during maximum ag- lithostratigraphically they resemble the logical features show that the river gradation. As the river can be incised younger terra firme sediments in the system migrates laterally to the south- more than once, the cycle is repeated central parts of the foreland basins. west becoming increasingly parallel to and the river system may migrate Their deposition was also controlled by the Andean strike owing to active fault- laterally. foreland subsidence. ing along the Sub-Andean thrust zone The extensively documented surficial We thus conclude that there is no (Fig. 4) (Rasanen ef al., 1987). formations (with or without basal con- single surficial 'alluvial mantle' in the Both the lower and higher terraces glomerates) at the foreland, which Amazon as was assumed by the flash- along the river show the above strati- partly derive from similar terrace flood, 'Belterra lake' or 'Lago Amazonas' graphy (Fig. 2). We suggest that the systems should be considered as litho- depositional models for the Amazon surficial sequence (with basal conglom- stratigraphic units each with its own terra firme. We suggest that more or less erates and overlying fluvial cycles) on diachronic depositional history. In gradual, regressive, late-Cenozoic time- the Madre de Dios is due to slow lateral contrast to the 'layer-cake' concept transgressive deposition took place

323 Fig. 3. Blockdia~rainlOOkit7g tonilardsthei~~estern and central Amazon loriilandsover the Andesfrom thesouth-uiest. Thefigure shows, in a highlygeneralizedmanner, the different states of landscapeevolution immediately east oftheconsiderablyelevatedpart ofthe Sub-andean foldand thrustbelt. (A) Theareas where theoriginally planar d~osurfaceoffZat-lying terra firme alluviumis locallywell-preserved. For Peru, thecoverageis based on ourfieldand laboratoryinvestigations; for theother areasseveral sourceshave been usedBrazil (Projefo RADAMBRASIL, 1976,1977),Bolivia campb belletal., 1985)Colombia (Khobzi, etal., 1980).Themoresporadicpresenceofterra firmealluvium in theshieldareasandin fhelowerAmazon east ofManausis notdepicted.Aquestion mark(?) indicates lackofdata. (Bi Stronglydissectedarea riihere theoyiginalde~surfaces havebeen erodedandthealderalluvium is exposed.(C) Large modernto subrecent aggradationalplains in the Pastaix-Maranonand Ucayalif2oodbasinsand in thePastazafan between theMorona and Tigrerivers. Thehighermesas atthe Vaupisarcharealsoshown. 2,4,5a,b,6a-earethesitesofthefigures. TheRDand TLcodenumbers indicate thesites of the respective14CandTLsamples. Thefrontalgeneralizedstrafigraphicalsequence withvertical exagqeration roughlydepicts the deep stratigraphy ofthe Pastaza-Marnnon, Ucayali and Madre de Dios-Beni basins andis projectedto oneplane Wnesco,1975, Guizado,7 986). away from the most actively emerging both a finite I4Cage (39,300 yr BP) and an in the level of sedimentation during the arch areas during partitioning of the infinite age (> 40,000 yr BP) (RD 87- late Pleistocene. intraforeland. The effects of climatic 1A,B). At other sites along the lower The second site for TL dating was on oscillations are superimposed on the Madre de Dios (RD 87-17) and Tambo- the northern side of the River Solimoes, tectonic control of sedimentation and pata (Campbell and Romero, 1989) (Fig. 20 km downstream from Iquitos at full diachrony of the alluvial deposits 3), wood collected from basal conglom- Santa Teresa, northeastern Peru (Fig. 3 thus may not exist. Remnants of depo- erate deposits, which may belong to the and Table 1). A sample (TL-86-2) was surfaces cover the Serra do Moa uplift initial depositional phases of this ter- collected from an erosional bank carved and the Fitzcarrald and Vaupks arches race, also gave finite dates of around into a 15-m-high dissected hill of un- and are pre-uplift evidence of the latest 36,600 yr BP. However, results with a known original aggradational level. The aggradation in these areas (Fig. 3) finite (32,000 yr BP) and an infinite (> material at 5 m above the contact bet- (Koch, 1959; Instituto Geologic0 del 40,000 yr BP) age were obtained from a ween the Pebas Formation and the over- Peru, 1975; Khobzi et al., 1980; Frailey et simple trunk from another basal con- lying terra firme alluvium was brownish al., 1988). In the intensively dissected glomerate deposit (RD 87-21) nearby sand, with palaeocurrent from west, areas greater uplift has exposed strata (Table 1).Since all dated wood has been intercalated by layers of horizontally ores deeply buried. Irrespective of collected from groundwater saturated bedded clay 10-15 cm thick. The dates whether or not the original deposurface sediments, and may be contaminated for the two mineral fractions are in good is preserved, we here refer to the by younger carbon, wood of infinite age agreement, placing the age of deposi- various surficial formations as terra firme may give ages between 32,000 and tion at around 100 ka (Table I). Thus it alluvium. 39,300 yr BP (Irion, G., pers. comm.). may represent the age of the youngest These controversial results seem to part of the similarly dissected terra firme imply that even this clearly defineable on the northern side of the Solim6es TL AND 14C METHODS DATE low terrace can only be determined with channel. LATE PLEISTOCENE TERRA certainty to be > 30,000 yr BP old. Distinct fluvial terraces of terra firme FlRME DEPOSITION The models according to which terra are common throughout the lowland To achieve data on the age hetero- firme originated from late Pleistocene- area (Koch, 1959; Projeto RADAM- geneity of Western Amazon terra firme, Holocene flashfloods or delta-lake BRASIL, 1977; Khobzi et al., 1980; Rasa- we applied TL and I4C methods to date 'Lago Amazonas' deposition are based nen et al., 1987; Cunha, 1988). They may some terrace levels (sites as in Fig. 2) on I4C data, some of which were col- be climatically associated with cyclic covering extensive parts of the non- lected from the same rivers and basal glaciations in the Andes but owing to flooded relief in Peru and representing conglomerate deposits as ours. Five of the limited number of absolute dates the latest depositional phases of the the dates were infinite when only two available their exact provenance is still terra firme alluvium. TL was used for the gave dates around 36,000 BP (Campbell poorly known. The TL dates from the first time to date terra firme deposits of and Romero, 1989). The dates are as un- two sites indicate early glacial times. the Amazon basin, thus enabling the certain as ours and thus even the central However, these data are still sparse and age determinations to be extended conclusion that terra firme deposited TL dates may also be subject to uncer- further back into the Pleistocene. after 36,000 yr BP is open to question. tainties, which makes a detailed at- In northeastern Peru, infinite I4C ages To date the fluvial sediments forming tempt to interconnect the depositional (RD 86-1,86-2A, 86-6,87-10, Table 1 and terra firme levels, TL samples were col- ages and the known climatical oscilla- Fig. 3) were obtained from organic lected at two localities expected to yield tions premature. However, the 14C ages material lying in the uppermost parts of infinite I4C ages. The first site datedwas of a few low terraces along the River the Pebas Formation (inferred Mio- at Laberinto, on the Madre de Dios, Ucayali in Peru seem to cluster at Pliocene brackish-lacustrine origin, southeastern Peru. Here an erosional around 13 ka (Dumont, 1989), at times Williams, 1949; Hoorn, 1988) and at the bank is carved into deposits forming a of increased discharge during the last contact with (or under) the terra firme flat terra firme 40 m above the low water deglaciation, which started at around alluvium. In southeastern Peru, in a sec- datum of the river. Similar deposurface 13.5 and peaked at around 9.6 ka tion along the River Tambopata, a tree is well developed over large areas in the (Showers and Bevis, 1988). trunk well in the middle of a 40 m-thick Madre de Dios-Beni basin (Figs 3 and Evidence also exists for the strong in- terra firme alluvium with a well-pre- 6e). A sample (TL-87-48)was taken from fluence tectonics has had on the origin served original deposurface also gave the lower part of the section, some 8 m of many asymmetric lowland terrace an infinite age (RD 87-13). Our I4C above the river, from the lower systems (Rasanen et al., 1987; Cunha, results thus indicate that most of the de- homogeneous channel sediments of a 1988; Dumont, 1989). In the immediate posits forming the terra firme levels are fluvial cycle. Dating revealed deposi- Sub-Andean zone along the upper outside of the range of the I4C dating tion at around 176 ka. Although we con- course of the Tambopata, a well-de- method. sider these sediments to be among the veloped three-stage terrace system is A clearly definable 13-15 m-high ter- youngest terra firme deposits in the area, incised into an emerging anticlinal race level (similar and close to the site in they are still much older than the structure in the Sub-Andean thrust Fig. 2) which has asymmetrically de- flashflood or 'Lago Amazonas' delta- zone (Fig. 4). This indicates that tectonic veloped along the northeastern side of lake depositional models would imply. movements played an important role in the Madre de Dios at Titipisco, yielded This TL result indicates major changes the origin and preservation of the late-

325 ~~

Table 1. Dates for surficial formationsin Peruvian lowland Amazon. The "C samples were all of wood except sample RD 86-2Awhichconsisted of compressed remnants of peat in detritus gyttja. As the wooden samples were in a state of decay, which made the recovery of cellulose difficult, samples were subject to acid-alkaline-acid pretreatment. To diminish the influence of any contaminants, the innermost core of each stump was used for dating. Special attention was paid to the"Cmeasurementsof samplesRD87-1.87-17and87-21.To check for any contamination, parallel samples were submitted toacceleratordatingat the University of Uppsala, Sweden. Samples Ua-1213,Ua-1214 and Ua-1215 were redated by accelerator techniques. Pretreatment in 2M NaOH for6-8 h and 2M HCI for 1 h (with additional NaOH treatment when necessary) was performed. The dated sites are shown in Fig. 3. The TL samples were collected using an enclosed tube sampler that fed the samples directly into black plastic bags. The datings were performed using potassium feldspar (FK),except for sample TL 86-2, for which the plagioclase (FN) fraction was also used. For the laboratory technique used see Jungner et al., (1989).The dates given were not corrected for long-term fading. With the value of 711 kyr for the mean life of Scandinavian feldspars (Mejdahl, 1989), corrections of about 10 kyr and 40 kyr should be added to the age given in the table for samples TL 86-2 and TL 87-48, respectively. The date for TL 87-48 is based on an exponential growth curve, and the natural TL level comprisesabout 65% of the level reached when a radiation dose of 3000Gyr was given to the feldspar sample after a thorough artificial UV bleach.

Radiocarbon dates Obpdhbd L.b.no. Fieldw. sik Fviplfmxmbrid Lhkd a% AssreP (lCq.W.lltS.) musriJ

kl-2473 RD87-13A Tunbcph 69WW 12039'40' wmbrrlradimnt wood -29.0 >42ooo

kbl-2584 RD87-1A TitipLoo 70045'45'' 12017'15" cb.rmsl pdinans gad ' wmd -30.1 UI-1213 RD87-1B ' kbl-2586 RD87-17 FwbM.ldayQ 69OlU15' 12"3W(r' ~ly.mvd wind -28.2 UB-I214 ' W-wn RDm-21 bItOM.lhldayd0 69011'W 12QJ8'5YCh.mal1UFVd wmd -28.6 Ua-I215 '

-29.6 -28.8 -31.9

73WlC' MW'O5'' fhpowhtzk PK 226 11.7 1.99 7.56 337 3.55 95t15 PN 0.81 328 3.27 1WlS 69"U'lY' 12"42'W &mumIOU PK 289 14.2 202 14A 737 4.19 176f2.5

75006u" 03°49'11" Mad? PK 210 8.71 0.91 4.64 53 231 21t5

Pleistocene lowland terraces in the the Madre de Dios-Beni and Ucayali DEFORMATION REORGANIZES Madre de Dios-Beni basin, as we basins), is partly due to the offlap of the THE LOWLAND RIVER assume in our depositional model for weakly deformed terra firme alluvium SYSTEMS the asymmetric terraces along the and the sequential system of younger Madre de Dios. aggradational terra firme plains and Foreland deformation has modified the In conclusion, the diachronism of the terraces at different levels. This general drainage of the older non-aggrading western Amazon terra firme alluvium, relief diversity is accentuated by the and dissected terru finne areas. The which spans the period from the existence of other younger Holocene strong tectonic partitioning of the Ucay- suggested Oligo-Miocene (Vaupes depositional regimes in the basins ali basin (Fig. l), has had the most wide- arch) to late Pleistocene (central parts of (Fig. 3). reaching consequences for the overall

326 69057' 69O32'

:ig. 4. A sequenceof terraces (b,c,d) along the upper Tambopata in theMadrede Dios-Beni basin. The terraces have developedon theflankof a rising nticlitie in the Sub-Andean zone. Theanticline trending NW-SE is elevatedalong the thrust zone (e) in relation to the eastern lowlands. Original deposurface of the anticline (a)

relief evolution of the western Amazon, tively downwarped at the western edge recent in age and probably indicates caused by Plio-Pleistocene upward of the increasingly westward-tilting modern tectonic activity. thrusting of the reverse faults exposed monocline between the Ucayali and the Other evidence that part of the uplift on the eastern side of the Sierra do Moa Serra do Moa. Recent sedimentation is of Serra do Moa is younger than some uplift (Cunha, 1988; Dumont, 1989) transgressive, and a few blocked valley terra firme deposurfaces comes from the (Fig. 3). The partitioning was promoted ria lakes (Inuria) have formed in the sub- area between the southern Ucayali, by uplift of the igneous peralkaline merged erosional relief of the ferra firme Acre and Madre de Dios basins (Fig. 6a), massif crystallized 5-10 Ma, and now alluvium (Fig. 5a). The Ucayali and sev- where geomorphic features and general exposed close to the Serra do Moa eral other rivers (e.g. Tamaya, Fig. 5b drainage indicate reversal from NE to E (Stewart, 1971). During the general and Sheshea) have recently relocated into the Acre basin and from NW to W uplift around the other anticlinal struc- their courses towards the locus of rela- towards the Ucayali basin (Fig. 6b). tures and the Fitzcarrald and Vaupes tive subsidence, close to the reverse During the rainy season this divide is arches, the old deposuriaces of the terra fault zone along the eastern edge of the covered with extensive forested wet- firme alluvium gradually reached their Shira mountains (Fig. 3). As suggested lands with poor drainage (Oppenheim, present higher elevations above the by the poorly developed new flood- 1937). The Ucayali-Urubamba river level of modern sedimentation (Fig. 3). plain, the latest great avulsion west of system has captured the earlier north- In the Ucayali basin, the upper depo- the Ucayali channel (Rasanen et al., eastern flowing drainage by gullying its surface of the terra firme alluvium is rela- 1987, Neller et al., in press) is fairly headwaters into 100-m deep canyons

327 8'40'

Fig. 5. False dour Landsat Thematic mapper images (path 6, row 66QX July 14, 1985) of the downwarped central part of the Ucayali basin. a The dozonu~nrprd,bruwnish, dissrcted upper surface topography of the terra firme alluvium (a) is buried under the modern orlerbunk sedimentation ofthe rrccntlycut izewrourseufthe Ucayali(RdsBnen et al., 19871, forminga brown dendviticnet of burirdcrests (b)in thegreenishareaofmodern ooerbnnk sedimentation (c). A purt ofthe large blocked valley lake (Znuria)formed in the submerging erosional topography (d). b, The Tamaya (b)has abandoned itsformerusyninzefricfloodplairz(c) and relocated its new channel through thedissected terra firme alluvium (a)owirig to the incrensedwestward tilt of the landscape. Arrows show the direction of the pow.

(Oppenheim, 1975) to the south (Fig. south have been documented (Khobzi et the Acre basin they indicate a N-NW- 6b). During this process the modern al., 1980; Eden et al., 1982). In Peru, the NE tilt towards the main River Amazon Ucayali-Urubamba became one of the Nanay has faced a decrease in water (Cunha, 1988). largest longitudinal trunk rivers in the flow due to some as yet undocumented area. change in its upper drainage area, evi- The development of the headwaters denced by a Pleistocene cutoff channel DEFORM ATlON REGULATES of the Jurua into a major longitudinal belt wider than the present one pre- DENUDATION OF THE TERRA trunk river was also initiated by thrust- served only on the northeastern flood- FIRME ALLUVIUM ing along the faults of Bata-Cruzeiro plain margins (IFG, 1984). Southward (Fig. 3), which is the easternmost, and tilting is also indicated by the way the Like the late Cenozoic spatially migrat- also the youngest, clear reverse fault headwater tributaries of the Amazon ing fluvial aggradation of the terra firinr cutting through the terra firme alluvium. have captured those of the Orinoco, and alluvium and the relocations of the in- The adjustment of the Jurud floodplain by the lateral erosion of the southern cised river systems, the denudational shows the evolution of the former trans- edges of the floodplains of the Caqueta history of the lowland relief is control- versal river system of a peri-cratonic and upper Guaviare (Fig. 3), whose led by Plio-Pleistocene foreland defor- basin into the longitudinal system of an major drainage area is on the northern mation. Enlargement of the Ucayali independent foreland basin, parallel to side of the rivers (Khobzi ef al., 1980). basin southwards and uplift of the Serra the Andean strike. The southeastern drainage system do Moa and Fitzcarrald arches have Similar adjustments of incised river south of the Vaupes arch has flood- markedly constricted the Acre drainage systems to changing landscape gra- plains with higher long-term channel area. Earlier, when the terra firme allu- dients controlled by Pleistocene base- relocation rates to the south due to the vium of the SolimBes Formation in the ment tectonics have been reported predominant local tilting towards the Acre basin was deposited (Petri and elsewhere. Abandonments of the main Amazon trough. Similar drainage Fulfaro, 1988), the river systems at the southeastern courses of the Caguan and features characterize several other terra divide of the Ucayali and Acre basins several smaller rivers in Colombia to the firme areas of Amazonia (Fig. 3), and in discharged to a greater extent north-

328 basins is the latest phase in the relief Lehtovaara for constructive comments ogy of fluvial sands from the Amazo- evolution which changed the Tertiary on the manuscript, Th. van der Ham- nian foreland basin, Peru and Bolivia, Neotropical (Hoorn, 1988, Petri and men, V. Lappalainen and H. Valdivia Bull. geol. Soc. Am. 101, 1552-1562. FuIfaro, 1988) lowlands into the present for discussions on the subject, and K. DNPM (1984) Geologia do Brad, Departa- Andes. On a large scale, the Andes have Andree, J. Hietaranta, R. Kalliola, M. mento Nacional da Producao Mineral, been formed through a mechanism of Puhakka, M. Rajasilta, H. Ryhanen, M. Brasflia, pp. 501. foreland shortening, leading to crustal Valtonen, H. Venho and V. Wanderley Dumont J.F. (1989) Neotectonic of the thickening and subsequent uplift for their help in various phases of the Peruvian jungle as related to geo- (Suarez et al., 1983; Megard, 1987; work. morphology and fluvial dynamics, In: Isacks, 1988). In the central Peruvian lnternational Symposium on Global Andes, late Cenozoic shortening ac- Changes in Southern America during the REFERENCES counts for about 115 km (Megard, 1987). Quaternary: Past-Present-Future, Abstr. In the central Andes, which constitute Agassiz L. (1865) On the Drift in Brazil, pp. 140-144. ABEQUA, Sao Paulo. the southern margin of the present and on Decomposed Rocks under the Eden M.J., McGregor L.D.F.M. and Amazon rainforest, the shortening has Drift, Am. I. Sci. Ser., 2, 40, 389-390. Morelo V.J.A. (1982) Geomorphology of been about 100 km during the last Almeida L.F.G. (1974) A drenagem the middle Caqueta basin of eastern 12 Myr, and 15-20 km during the last festonada e seu significado fotogeo- Colombia, 2. Geomorph., 26,343-364. 3 Myr (Isacks, 1988). Instead of reflect- logico, In: Congresso Brasiliero de Frailey C.D. (1986) Late Miocene and ing locations (and secondary Holocene Geologia, 28, Porto Alegre, pp. 274-276. Holocene Mammals, exclusive of the border zones) of isolated Pleistocene Sociedade Erasileira dc Geologia, Porto Notongulata, of the Rio Acre Region, broad-leaved forests (climatic refuges) Alegre. Western Amazonia, Contrib. Sci. 374, (Whitmore and Prance, 1987), modern Baldock J.W. (1985) The Northern Andes: 1-46. species ditributions in the Western A Review of the Ecuadorian Pacific Mar- Frailey C.D., Lavina E.L., Rancy A. and Amazon may be a result of historical gin, In: The Ocean Basins anda Margins, Souza F.J.P. (1988) A proposed Pleisto- species dynamics controlled by land- Vol. 7A;The Pacific Ocean (ed. by A.E.M. ceneiHolocene lake in the Amazon scape evolution. Nairn, F.G. Stehli and S. Uyeda), pp. basin and its significance to Amazonian Owing to late Cenozoic palaeogeo- 181-218. Plenum Press, New York. geology and biogeography, Acta graphical changes in relief evolution, Barberi F., Coltelli M., Ferrara G., In- Amazonica, 18, 119-143. the forest biota of the western Amazon nocenti F., Navarro J.M. and Santacroce Gibbs R.J. (1967) The geochemistry of the have probably alternated between R. (1988) Plio-Quaternary volcanism in Amazon River System: Part I. The fac- allopatry and sympatry, commonness Ecuador, Geol. Mag., 125'1-101. tors that Control the Salinity and the and rarity, and continuous distribution Bartlein P.J. and Prentice I.C. (1989) Orbi- Composition and Concentration of the and fragmentation in much the same tal variations, Climate and Paleoecol- Suspended Solids, Bull. geol. Soc. Am., way as has taken place in temperate ogy, Trends Evol. Ecol., 4, 195-199. 78,1203-1232. zones as a result of glacial oscillations Campbell K.E. Jr. and Frailey D. (1984) Guizado J.J. (1986) Las Formaciones Cor- (Bartlein and Prentice, 1989). The Holocene flooding and species diversity rientes y Maranon del Cuaternario y biological distribution patterns may also in Southwestern Amazonia, Quat. Res., Plioceno del nororiente Peruano, Reun- be closely connected to present edaphic 21,369-375. id11 Anual a Nivel de Expertos, Petroperu, differences. These should be consider- Campbell K.E. Jr. and Romero P.L. (1989) Lirna, Vol. 2. able due to the age heterogenity of the La Geologia del Cuaternario del de- Haffer J. (1985) Avian Zoogeography of forest bed, showing variability in partamento de Madre de Dios, Bol. Geol. the Neotropical Lowlands, Ornithol. weathering status and in degree of pre- Soc. Peru. 79,53-62. Monogr., 36,113-146. servation of the original fluvial facies. Campbell K.E. Jr., Frailey C.D. and Arel- Haffer J. and Fitzpatrick J.W. (1985) Geo- lano L.J. (1985) The geology of the rio graphic Variation in some Amazonian Beni: further evidence for Holocene Forest Birds, Ornithol. Monogr., 36, ACKNOWLEDGEMENTS flooding in Amazonia, Contrib. Sci., 364, 147-168. The work was financed by FINNIDA, 1-18. Hall M.L. (1977) €I Volcanismo en el the Academy of Finland and the Turku Cisternas A., Dorbath L. and Dorbath C. Ecuador, Inst. Panamer. Geogr. Histo., University Foundation. The field work (1988) A study of subandean seismicity Quito, 120 pp. was carried out together with the Peru- in central Peru. In: VZZ Congreso Hoorn C. (1988) Nota preliminar sobre la vian institutions: ONERN, PETRO- Latinoamericano de Geologia. Resumenes, edad de 10s sedimentos Terciarios de la PERU, OXYPERU, IIAP. Technical pp. 322. Sociedade Brasileira de zona de Araracuara (Amazonas), Bol. cooperation was made with the Tech- Geologia, Belem. Geol., 29, 87-95. nical Research Centre of Finland (In- Cunha F.M.B. (1988) Controle tectbnico- IFG (1984) Mapa Planimitrico de Imhgenes de strument Laboratory) and Tandem estructural na hidrografia da regi5o doi Satellite 1 :250000. Institute for Applied Accelerator Laboratory, Uppsala, Alto Amazonas. In: Anais do XXXV Con- Geosciences, Neu Isenburg. Sweden. We wish to thank Heini gresso Brasileiro de Geologia, 5,2267-2277. Instituto Geologico del Peru (1975) Mapa Ervanne for help with the TL dating, S. Sociedade Brasileira de Geologia, Geoldgico del Peru, Ministerio de Energia Kroonenberg, J. Haffer, R. Neller, G. Belem. y Minas, Lima. Gluckert, V.-P. Salonen and J. DeCelles P.G. and Hertel F. (1989) Petrol- Irion G. (1984a) Sedimentation and sedi- 331 FEBBARESEARCH

ments of Amazonian Rivers and evolu- Neller R., Salo J. and Rasanen M. (1990) 238,1398-1401. tion of the Amazonian landscape since On the Formation of Blocked Valley Showers W.J. and Bevis M. (1988) Pliocene times, In: The Amazon (ed. by Lakes by Channel Avulsion In Upper Amazon Cone Isotopic Stratigraphy: H. Sioli), pp. 537-579. Dr. W. Junk, Amazon Foreland basins, 2. Geomorph., Evidence of the Source of the Tropical Dordrech t. in press. Fresh Water Spice, Palaeopgr. Palaeo- Irion G. (1984b) Clay minerals of the ONERN (1977) Innerrtario, ezraluaribn de climatol. Palaeoecol., 64, 189-199. Amazonian soils, In: The Amazon (ed. by integracidn de 10s recursos naturales de la Silva L.L. (1988) A Estratigrafia da For- €3. Sioli), pp. 201-214, Dr. W. Junk Pub- zona Iberia-lnapari, Oficina National de macao SolimBes: Uma Analise Critica, lishers, Dordrecht. Evaluacion de Recursos Naturales, In: Anais do XXXV Congreso Brasileiro de Isacks B.L. (1988) Uplift of the Central Lima, pp. 334. Geologia, Vol. 2., pp. 725-737. Andean Plateau and bending of the ONERN (1984) Inventario, eoaluacicin e Sociedade Brasileira de Geologia, Bolivian orocline, 1. geophys. Res., 93, integracidn de 10s recursos naturales de la Belem. 321 1-3231. microregion Pastaza-Tip, Oficina Simpson G.G. and Paula Couto C. (1981) Jordan T.E., Isacks B.L., Allmendinger National de Evaluacion de Recursos Fossil mammals from the Cenozoic of R.W., Brewer J.A., Ramos V.A. and Naturales, Lima, pp. 224. Acre, Brazil, 111, Iheringia, 6, 11-73. Ando C.J. (1983) Andean tectonics re- Oppenheim V. (1937) Geological explora- Sioli H. (1984) The Amazon and its main lated to geometry of subducted Nazca tion between upper Jurua river, Brazil, affluents: Hydrography, morphology of plate, Bull. grol. Soc. Am., 94,341-361. and middle Ucayali river, Peru, Bull. the river courses, and river types, In: Jungner H., LandvikJ.Y. andMangerud J. Am. Ass. Petrol. Grol., 21,97-110. The Amuzon (ed. by H. Sioli), pp. 127- (1989) Thermoluminescence dates of Oppenheim V. (1946) Geological recon- 163. Dr W. Junk. Publishers, Dordrecht. Weichselian sediments in western naissance in southeastern Peru, Bull. Sombroek W.G. (1966) Amazon soils, A re- Norway, Boreas, 18,23-30. Am. Ass. Petrol. Geol., 30, 254-264. connaissance of the soils of the Brazilian Khobzi J., Kroonenberg S., Faivre P. and Oppenheim V. (1975) The first (1944) Amazon region. Center for Agricultural Weeda A. (1980) Aspectos geomorfo- geological exploration of the Upper Publication and Documentation, logicos de la Amazonia y Orinoquia Col- Amazon Valley of Peru, Bol. Soc. Geol. Wageningen, pp. 292. ombianas, Revista CMF, 5,97-126. Peru, 45,83-94. Stewart J.W. (1971) Neogene Peralkaline Klammer G. (1984) The relief of the extra- Petri S. and Fulfaro V.J. (eds) (1988) Igneous Activity in Eastern Peru, Bull. Andean basin. In: The Amazon (ed. by H. Geologia doBrad, Editora da Univer- ged. SOC.Am., 82,2307-2312. Sioli), pp. 49-83 Dr. W. Junk Pub- sidade de Sao Paulo, Sao Paulo, 631 pp. Suarez G., Molnar P. and Burchfield B.C. lishers, Doldrecht. Pilger R.H. Jr. (1984) Cenozoic plate (1983) Seismicity, fault plane solutions, Koch E. (1959) Unos apuntes sobre la kinematics, subduction and mag- depth of faulting, and active tectonics of geomorfologia del Iiio Ucayali (Oriente matism: South American Andes, J. geol. the Andes of Peru, Ecuador, and south- l’eruano), Bol. Soc. Grol. Peru, 34,32-41. Soc. London, 141,793-802. ern Colombia. ]. geophys. Res., 88, Kummel 8. (1948) Geological reconnais- Projeto RADAMBRASIL (1976) Levarl- 10403-10428. sance of the Contamana region, Peru, tarnento de Recursos Naturais, Folha SC.19 Unesco (1975) tectonic map of South America Bull. Geol. Soc. Am., 59, 1217-1266. Rio Branco. Ministerio das Minas e Ener- I :5,000,000 Departamento National da Megard F. (1987) Structure and Evolution gia, Rio de Janeiro, 458 pp. Producao Mineral and Commission for of the Peruvian Andes. In: The Anatomy Projeto RADAMBRASIL (1977) Levan- the Geological Map of the World. of Mountain Ranges (eds by J.-P. Schaur tamento de Recursos Naturais, Falhas Unesco, Brasilia. and J. Rodgers), pp. 179-210. Princeton SBISC 19 1avarilContamana. Ministerio Whitmore T.C. and Prance G.T. (eds) University Press, Princeton. das Minas e Energia, Rio de Janeiro, (1987) Biogeography and Quaternary His- Mejdahl V. (1989) How far back: Life times 413 pp. tory in Tropical America. Clarendon estimated from studies of feldspars of Projeto RADAMBRACIL (1977) Levan- Press, Oxford, 213 pp. finite ages, In: Long bshorf range limits in tamiento de Recursos Naturuis, Folha SB19 Williams M.D. (1949) Depositos Terciarios luminescence dating (ed. by M. Aitken). Iurua‘. Ministerio das Minas e Energia, Continentales del Valle del Alto Oxford University Press, Oxford. Rio de Janeiro, 430 pp. Amazonas, SOC. Geol. Peru. Vol. lubilar, Minster J.B., Jordan T.H., Molnar P. and Rasanen M., Salo J. and Kalliola R. (1987) 5,1-13. IIaihes E. (1974) Numerical Modelling Fluvial Perturbance in the Western of Instantaneous Plate Tectonics, Amazon Basin: Regulation by Long- Geophys. I. R. astr. SOC.36, 541-576. Term Sub-Andean Tectonics, Science, Received 22 May 1990; accepted 6 ]uly 1990

332