The origin of oriented lakes: Evidence from the Bolivian Amazon

    The origin of oriented lakes: Evidence from the Bolivian Amazon Umberto Lombardo, Heinz Veit PII: DOI: Reference:

S0169-555X(13)00434-0 doi: 10.1016/j.geomorph.2013.08.029 GEOMOR 4473

To appear in:

Geomorphology

Received date: Revised date: Accepted date:

7 March 2013 22 August 2013 24 August 2013

Please cite this article as: Lombardo, Umberto, Veit, Heinz, The origin of oriented lakes: Evidence from the Bolivian Amazon, Geomorphology (2013), doi: 10.1016/j.geomorph.2013.08.029

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT The origin of oriented lakes: Evidence from the Bolivian Amazon Umberto Lombardo1 and Heinz Veit1 Institute of Geography, University of Bern Hallerstrasse 12, CH-3012 Bern – Switzerland

IP

T

1

Corresponding author: Umberto Lombardo, Institute of Geography, University of Bern

SC R

Hallerstrasse 12, CH-3012 Bern – Switzerland Tel.: +41 (0)31 631 85 78; e-mail: [email protected]

NU

ABSTRACT

The presence of hundreds of rectangular and oriented lakes is one of the most striking

MA

characteristics of the Llanos de Moxos (LM) landscape in the Bolivian Amazon. Oriented lakes also occur in the Arctic coastal plains of Russia, Alaska and Canada and along the Atlantic Coastal Plain from northeast Florida to southeast New Jersey and along the coast of

D

northeast Brazil. Many different mechanisms have been proposed for their formation. In the

TE

LM, Plafker’s (1964) tectonic model, in which subsidence results from the propagation of bedrock faults through the foreland sediments, is the most accepted. However, this model has

CE P

not been verified. Here, we present new results from stratigraphic transects across the borders of three rectangular and oriented lakes in the LM. A paleosol buried under mid-Holocene sediments is used as a stratigraphic marker to assess the vertical displacement of sediments on

AC

both sides of the alleged faults. Our results show that there is no vertical displacement and, therefore, that Plafker’s model can be ruled out. We suggest that, among all the proposed mechanisms behind lake formation, the combined action of wind and waves is the most likely. The evidence from the LM provides new hints for the formation of oriented lakes worldwide.

Keywords: Oriented lakes; Rectangular lakes; Llanos de Moxos; Bolivian Amazon

1 Introduction Geometric and oriented lakes are intriguing geomorphological features that are still poorly understood. They occur along the coast of northeast Brazil (Bezerra et al., 2001) in the Arctic coastal plains of Russia, Alaska and Canada; in the Atlantic coastlands of Maryland, Georgia and the Carolinas, where the well-known Carolina Bays are found; and in the Llanos de Moxos (LM), in north-eastern Bolivia (Burn, 2004). The LM (Fig. 1) is a seasonally inundated savannah landscape (Hamilton et al., 2004; Mayle et al., 2007) interspersed with

ACCEPTED MANUSCRIPT several hundred perennial lakes (Plafker, 1964; Hanagarth, 1993; Dumont and Fournier, 1994). Many of these lakes have been noted for their rectangular shape and markedly uniform SW-NE orientation. Although they vary considerably in size, they are characterized by being

T

very shallow, usually less than two meters deep, and having a flat bottom. Many different

IP

mechanisms have been proposed for their formation. These potential mechanisms fall into two categories: i) those that involve one mechanism explaining both the formation of the lakes and

SC R

their morphological characteristics, and ii) those that invoke a two-step process, with the first step forming the lake basins and the second accounting for their shape and orientation. The first group of hypotheses include the following: 1) tectonic movements resulting from the

NU

propagation of bedrock faults through the foreland sediments (Plafker, 1964, 1974; Allenby, 1988); 2) water erosion caused by large-scale flooding in the Llanos de Moxos after Andean

MA

deglaciation (Campbell et al., 1985); 3) wind deflation during the dry Last Glacial Maximum (LGM) (Clapperton, 1993); and 4) anthropogenic origin (Barba, 2003; Belmonte and Barba, 2011) in line with other well-documented pre-Columbian earthworks (see Erickson, 2006;

D

Mann, 2008; Lombardo et al., 2011). A two-step model was proposed by Dumont and

TE

Fournier (1994) suggesting that the lakes developed in mid- to late Holocene deflation areas and were later shaped by waves. Similarly, Langstroth (1996) suggested that deflation during

CE P

dry periods may have initiated the formation of the lakes and that the shorelines and lake beds were later modified by wave action during wet phases. The wide range of hypotheses about the processes behind lake formation is a function of the lack of field data and limited studies, which, so far, have been largely based on the

AC

interpretation of aerial photography or first-generation Landsat imagery. However, the most accepted hypothesis to date is that of tectonic control, as proposed by Plafker almost 50 years ago (Plafker, 1964; Price, 1968; Plafker, 1974; Allenby, 1988; Gonzales and Aydin, 2008). A small cluster of oriented lakes in the northeast of Brazil which resemble the LM lakes has also been interpreted as resulting from tectonic control (Bezerra et al., 2001). According to Plafker, the lakes’ rectangular shape results from the propagation of bedrock fractures through unconsolidated sediments (Fig 2). Plafker suggested that the lakes are the projection on the surface of subsiding (Fig. 2A) or vibrating (Fig. 2B) basement blocks. The basement blocks would be the result of a system of orthogonal faults with a northeast – southwest and a northwest-southeast orientation. This system of faults would be the result of the re-activation of old fractures linked to those outcropping in the Brazilian Shield. The major northwest-southeast and northeast-southwest trending lineament sets are interpreted to be the expression of "tension" fractures.

ACCEPTED MANUSCRIPT The debate about the origins of rectangular and oriented lakes is not confined to the LM. While there is a general consensus that oriented lakes in the Arctic coastal plains initially form as thermokarst lakes (French, 2007), why they share the same orientation is a highly

T

debated question (Hinkel, 2006; Pelletier, 2006). The most accepted explanation for their

IP

similar orientation is that wind induces a differential erosion that eventually elongates the lakes (Livingstone, 1954; Rex, 1961; Carson and Hussey, 1962; Côté and Burn, 2002; Hinkel

SC R

et al., 2005). However, several other hypotheses have been put forward, including marine ice sheets transgression (Grosswald et al., 1999), thaw slumping (Pelletier, 2005), differences in solar insolation (Ulrich et al., 2010) and a combination of hillslope processes with heat

NU

conduction and thaw-driven subsidence (Jorgenson and Shur, 2007; Plug and West, 2009). Among the lakes in the Arctic region, the oriented and rectangular lakes found in the Old

MA

Crow Plain in the Yukon Territory in Canada are the ones that most resemble the Bolivian lakes (Plafker, 1964; Allenby, 1988, 1989). The lakes in these two regions share unique characteristics. Most of them have square or rectangular shapes and form groups of adjacent

D

rectangles. When the lakes are elliptical, they often have one side more strongly curved than

TE

the other. In both regions lakes are very shallow and are associated with stream lineaments which are parallel to lake shores (Plafker, 1964).

CE P

The shape and orientation of the lakes in the Old Crow Plain are generally believed to be the result of tectonic control (Price, 1968; Allenby, 1989; Morrell and Dietrich, 1993); however, Roy-Léveillée and Burn (2010) have recently proposed that their shape and orientation could depend on the combination of wind and patterns of ice-wedge development. According to

AC

some authors, the resemblance between the Old Crow and the Bolivian lakes suggests that a similar process is likely to have generated both (Plafker, 1964; Allenby, 1988, 1989). Assuming this to be the case, this mechanism must be independent from local factors such as periglacial processes, differential sun exposure or human agency. Tectonics and wind/wave action are the only two geomorphological agents that can be acting at both arctic and tropical latitudes. Excluding one of these agents would provide a strong case in favour of the other being responsible for the orientation of the lakes. Understanding the formation and evolution of oriented lakes can allow us to constrain and accurately interpret the environmental conditions during their formation and, in turn, improve our ability to use sediments from these lakes for paleo-environmental reconstructions.

2 Regional setting

ACCEPTED MANUSCRIPT The Llanos de Moxos (LM) is a seasonally flooded savannah interwoven with gallery forests and forest island archipelagos; it covers most of the Bolivian Amazon (Hanagarth, 1993) (Fig. 1). The forested areas often coincide with fluvial deposits such as modern and relict splays

T

and levees (Mayle et al., 2007). Geologically, the LM is the foredeep of the Bolivian foreland

IP

basin system, between ~11° and 15° S and 67° and 63°W (Espurt et al., 2007). The plains are limited by the Andean piedmont to the south-west, and by the Brazilian Shield to the east and

SC R

northeast. The Brazilian Shield dips gently towards the Andes, underlying unconsolidated foreland deposits of mostly Quaternary age. Very little is known about the stratigraphy and thickness of these foreland deposits. Geophysical prospections point to flat-lying foreland

NU

deposits of mainly fluvial origin (Plafker, 1964). The basement reaches a depth of more than 5500 m along the Andean foothills and a depth of ~800 m at 150 km from them (Plafker,

MA

1964). Annual rainfall ranges from 1600 mm to 3500 mm and is concentrated in the rainy season, which lasts from November to April. The average annual temperature is 25 °C. Mild northern winds are the norm during the rainy season, while strong winds (locally called

D

surazos) blow from the south during the dry season (Hanagarth, 1993).

TE

The lake basins studied are located in the south-eastern part of the LM (Fig. 1). This area was covered by sediments deposited by a distal fluvial distributary system of the paleo-Río Grande

CE P

during the mid- to late Holocene (Lombardo et al., 2012). These sediments buried the original soil. Calibrated radiocarbon ages indicate that the soil formed between 7166 and 4628 cal. yrs. BP.

AC

3 Material and methods

All the geometric and oriented lakes have been digitalized manually on the basis of eleven Landsat 7 ETM+ and Google Earth images, covering the whole study area, between 11°S and 17°S and between 68°W and 62°W. All the Landsat images used were taken during the dry season, between the 22nd of June and the 19th of September. Lakes that were totally or partially filled with water have been classified as “full” lakes. Lakes have been classified as “dry” when there was no water or it was not clearly defined because the lakes were in an advanced state of infilling. As suggested by Plafker, drilling across lake margins could provide the data required to determine conclusively whether faulting or differential compaction predominates in the LM (Plafker, 1964). In fact, if faulting is involved, the displacement should be visible and measurable through sediment profiling. The only element needed is a stratigraphic marker that allows the measurement of the vertical displacement (Fig 2). In the south-eastern LM, a

ACCEPTED MANUSCRIPT paleosol below the lakes provides such a stratigraphic marker (Lombardo et al., 2012). This paleosol is easily detectable in stratigraphic profiles. The formation of the lakes followed the deposition of the Río Grande sediments, which, in turn, followed the formation of the buried

T

soil. Therefore, if the tectonic hypothesis is correct (Plafker, 1964), the presence of the

IP

supposed faults that generated the lakes should be readily visible as a difference in the depth of the paleosol outside and below the lakes. In order to test whether or not this difference

SC R

exists, three oriented lakes located in the south-eastern LM have been cored: Lake Perotó, Lake San José and Lake Villa Banzer (Fig. 1). Due to the extremely stiff clay sediments that constitute the bottom of the lakes, attempts to core them beyond a depth of 20 or 30 cm had

NU

been unsuccessful (Mayle et al., 2007; Whitney et al., 2012). However, using a Waker motor vibracorer operated from a fixed wooden platform, we have been able to penetrate the hard

MA

bottom of the lakes. The main stratigraphic units of the profiles were defined in the field based on the visual analysis of colour, texture and pedogenic features. Radiocarbon ages were measured at Poznan Radiocarbon Laboratory (www.radiocarbon.pl). They were calibrated

D

using OxCal 4.1 (Bronk Ramsey, 2009) (https://c14.arch.ox.ac.uk/oxcal/OxCal.html) and the

CE P

4 Results and discussion

TE

ShCal04 (McCormac et al., 2004) calibration curve.

In this study, approximately 700 geometric and oriented lake basins have been digitized in the LM, 40% of them are totally dry or in an advanced state of infilling. They cover a total area of ~1900 km2 (2600 km2 if we include the dry lakes) (Fig. 3). Almost all of them are oriented

AC

NE-SW (Fig. 4). They vary considerably in size and shape. The average size of full lakes is 4.7 km2, with 30% of them smaller than 1 km2. The average size of dry lakes is 2.25 km2, with 66% of the lakes smaller than 1 km2 (Fig. 3). Besides the rectangular lakes, oval and triangular lakes are also common (Fig. 4). The range of shapes shown by the LM lakes matches well the different forms of oriented lakes described by French (2007) in the Arctic region. The analysis of the retrieved cores shows that in all three cases the paleosol below the bottom of the lakes is at the same depth as the paleosol in the surrounding area (Fig. 5). This challenges the tectonic model. According to this model, the paleosol at the centre of Lake San José should have been found at a depth of about 4.4 m; as the thickness of the overlying sediments is 2.3 m, the water column 1.5 m and the thickness of the lacustrine sediments is 60 cm (Fig. 5). The supposed vertical displacement caused by the subsidence of a basement block would be equal to the water column plus the thickness of the lacustrine sediments. However, this is not the case. The paleosol below the lake is found at a depth of 2.3 meters, at

ACCEPTED MANUSCRIPT the same depth at which it is found in the surrounding area. Radiocarbon ages confirm that the paleosol below the lake is the same paleosol that is found below the savannah (Fig. 5). These data constitute strong evidence that no subsidence or faulting was involved in the formation of

T

Lake San José. The stratigraphic evidence from lakes Perotó and Villa Banzer is not as strong

IP

as for Lake San José because we could not core the very centre of the lakes. Nevertheless, cores 78 and 209_b (Fig. 5), that were taken inside the lakes at 200 m and 250 m from the

SC R

shores of Lake Perotó and Lake Villa Banzer respectively, do not show any vertical displacement either. Even if we cannot affirm that the paleosol at the centre of these two lakes is at the same depth as outside the lakes, we can say that these two lakes have been

NU

dramatically re-shaped by erosive processes. All the profiles inside the lakes show that the lacustrine deposits lay on layers made of clay / fine silts that belong to the paleo-Río Grande

MA

deposits. During the mid to late Holocene, Río Grande covered the southeast of the LM with a sedimentary lobe (Lombardo et al., 2012). Layers belonging to this sedimentary lobe separate the lacustrine sediments (above the continuous line in fig. 5) from the paleosol (below the

D

dotted line in fig. 5) and are laterally continuous, extending from below the lake basins to

TE

below the savannah surrounding the lakes. In the case of Lake San José, the depth at which these layers occur goes from about 1.4 m close to the lake margin to about 2 m at its centre

CE P

(Fig. 5). Consequently, lacustrine sediments thicken towards the centre of the lake. In the case of Lake San Josè our stratigraphy is consistent with the data presented by Whitney et al. (2012) who recently cored Lake San Josè. From a core located at about the same distance from the shore as our core 169, they recovered only 31 cm of sediments because of the

AC

stiffness of the clay. No pollen was found below the first 25 cm (Whitney et al. in press). Data from lakes Perotó and Villa Banzer seem to suggest that in these cases the depth of the layers belonging to the paleo-Río Grande also increases towards the centre of the lake. However, cores from the centre of these lakes would be needed in order to confirm this. The flatness of the paleosol, combined with the concave profiles of the overlying Río Grande sediments, suggests that the basins of the oriented lakes result from erosional processes of mostly Holocene alluvium. The cap of “indurated clay” underlying the softer lacustrine sediments has been interpreted as indicating a drier climate (Dumont and Fournier, 1994). Instead, our data show that the “indurated clay” is actually part of the Río Grande alluvium and is not a proxy for “drier climate”. The avulsive phase of the Río Grande, which covered the south-eastern LM with an up to 4 m thick layer of sediments, took place around 4700 cal. yrs. BP (Lombardo et al., 2012). Therefore, the rectangular lakes that are located on these alluvial sediments of the Río Grande

ACCEPTED MANUSCRIPT must have formed after 4700 cal. yrs. BP, otherwise they would have been submerged or eroded by the Río Grande. If tectonics did not originate the lakes studied, other forces must have been at play.

T

Alternative hypotheses involving a single mechanism for the formation of the lakes include

IP

LGM deflation (Clapperton, 1993), catastrophic flooding (Campbell et al., 1985) and the hypothesis that the lakes are human-made (Barba, 2003; Belmonte and Barba, 2011).

SC R

The LGM deflation hypothesis is not compatible with the relative chronology of the lakes we studied as inferred by their stratigraphic relation to the underlying paleosol, which formed during the mid- to late Holocene (Lombardo et al., 2012). The catastrophic flood hypothesis

NU

that Campbell (1985) proposed for the formation of the aligned lakes in the north-western LM is hardly applicable to the rest of the LM in light of the spatial distribution of the lakes (Fig. 6), which should have followed the flood path instead of being scattered all over the LM

MA

(Langstroth 1996). Moreover, Campbell’s hypothesis assumes that the catastrophic flood was triggered by climate change during the Pleistocene-Holocene transition, which is incompatible

D

with the radiocarbon ages from the underlying paleosol. The anthropogenic lake formation

TE

hypothesis is also at odds with the spatial distribution of the lakes, as there is no clear overlap between areas where archaeological earthworks have been identified and areas where the

CE P

lakes are located (Fig. 6). For instance, in the north of Santa Ana, the area of the LM where the largest amount of earth was moved by pre-Columbians to build thousands of hectares of raised fields (Lombardo, 2010), no geometric and/or oriented lakes are found at all. In turn, near the central part of the LM, an area with one of the highest densities of geometric oriented

AC

lakes, no evidence of pre-Columbian human occupation has yet been discovered. Our data suggest that none of the above proposed mechanisms for the formation of geometric lakes in the LM can, in themselves, explain all the characteristics of these lakes. Whereas our data challenge the tectonic hypothesis and are at odds with other “one-step” models, they are compatible with the hypothesis that the lakes were formed in a two-step process, as proposed by Dumont and Fournier (1994): a first stage of lake basin formation and a second stage responsible for their reshaping and orientation. Currents induced by strong winds can relocate sediments within lakes, inducing morphological changes of the basins (Ashton et al., 2009). Preferential erosion of the ends of the lakes and redistribution of sediments along the long-axis shorelines has been observed in Arctic Alaska (Carson and Hussey, 1962). In the LM, topographic depressions could form by a variety of processes such as uneven compaction of the underlying unconsolidated sediments, basement subsidence and fluvial related features such as water ponding in and between paleo-channels, oxbows,

ACCEPTED MANUSCRIPT backswamps and bar-top hollows (Dumont, 1993; Mertes et al., 1996; Best et al., 2006). These processes could initially form lake basins in the LM as the thermokarst processes in the Arctic region form the thaw lakes. The differences in the processes involved in the initial

T

formation of the basins could be responsible for the spatial patterns, clustering and size that

IP

oriented lakes have in different parts of the world. In a second phase, wind/wave action could elongate and orient the lakes.

SC R

Dumont and Fournier (1994) observed that, in the LM, water entered the lakes free of sediments and left laden with them. They concluded that wave action mobilizes particles and excavates the lakes up to a depth in which the waves lose their energy; the fine sediments are

NU

carried away by the lake outflow. According to them, wave action would therefore maintain the depth of the lakes constant and the bottom flat. In fact, wave energy, and therefore its

MA

capacity to erode and rework sediments, decreases with depth (Bye, 1965); this would explain why all the LM lakes have a similar depth and are very shallow. The reworking of the sediments along the lake margins would entrain clays into suspension and would mobilize

D

silts and sands. Over time, suspended clays might leave lakes through outlets (Dumont and

TE

Fournier, 1994) or by overflow during the wet season. Alternatively, sediments eroded from the lake banks would remain in the lake, eventually turning lakes into swamps. The almost

CE P

300 lakes that we have classified as dry or in advanced state of infilling could have undergone this process. In smaller lakes wave action would have been less intense and its capacity to erode the bottom of the lakes reduced. This could explain why small basins are more likely to be dry or in advanced state of infilling. It has been noted that many lakes are surrounded by

AC

elevated berms, on top of which forest grows (Plafker, 1964). These berms could be beach ridges formed by strong waves. In the LM, the angle of about 45° that exists between the direction of the major winds and the orientation of the lakes is not consistent with what has been observed elsewhere (Carson and Hussey, 1962; Côté and Burn, 2002). This could be due to the complexity of the relation between wind patterns and erosion (Roy-Léveillée and Burn, 2010) or to changes in the wind patterns that could have occurred during the last 3000-4000 years. The assumption that the shape and orientation of the lakes in the LM are primarily due to the wind/wave action does not mean that local factors are not important in order to understand the specific characteristics of the lakes in different regions. For example, faults like those envisaged by Plafker could be responsible for the markedly straight shorelines of some of the oriented lakes, in the same way that faults are probably responsible for the orientation of other lineaments, such as the reaches of small rivers (Plafker, 1964; Allenby, 1988). Likewise, in

ACCEPTED MANUSCRIPT some areas of the LM, pre-Columbians may have contributed to maintain the depth of the lakes by building drainage canals (Denevan, 1966; Erickson, 2006; Lombardo et al., 2012). Langstroth (1996) reports that, when villagers cleared an old channel linking a lake to a river,

T

the increased flow into the river lowered the level of the lake and further incised the channel.

IP

As river water levels are several meters below lake water levels for most of the year, preColumbians could have removed large amounts of sediments from the lake simply by opening

SC R

the canals and letting the water (and fine sediments) flow out of the lake. In fact, numerous pre-Columbian canals that could have worked in this way have recently been mapped in the area of the three lakes studied (Lombardo and Prümers, 2010; Lombardo et al., 2012).

NU

Satellite images and the stratigraphy shown in Fig 5 indicate that in all the three lakes deposition is currently greater than erosion and lakes are infilling. This is the case for all the

MA

oriented lakes we have surveyed in the LM. The change from erosion (lake formation) to deposition (lake infilling) could be related to some sort of environmental change that must have taken place after the mid-Holocene. More detailed studies of the lacustrine sediments are

TE

D

needed in order to constrain the age of this change and unravel what might have caused it.

5 Conclusions

CE P

For the first time stratigraphic cores have been taken across the margins of oriented lakes in the LM, allowing us to test existing hypotheses about their formation. Our data show the need for a reassessment of these hypotheses. The most accepted theory, which explains the formation of the lakes as resulting from the propagation of faults through the overlying

AC

sediments, is not compatible with the field evidence. Our findings suggest that, among all the proposed mechanisms for the formation of the oriented lakes in the LM, a two stage process is the most likely. During stage one, a variety of processes (e.g. neotectonics, sediment compaction or water ponding) could have created topographic depressions which subsequently filled with water; during stage two, wind/wave action could have reshaped and “oriented” these lakes. Further research is needed in order to confirm this hypothesis. Evidence from the LM supports the idea that the Old Crow oriented lakes in Yukon Territory, Canada, could have also been reshaped and oriented by wind/wave action (Roy-Léveillée and Burn, 2010). However, when extrapolating to rectangular and oriented lakes elsewhere, it must be acknowledged that different processes could have originated lakes with similar shapes, as it is suggested by the case of the rectangular and oriented lakes in north-eastern Brazil that are of tectonic origin (Bezerra et al., 2001). These new data provide i) a starting point for future research on the reconstruction of regional geomorphological histories in the

ACCEPTED MANUSCRIPT LM, ii) a context for the interpretation of lake-related paleo-environmental archives, and iii) important elements to estimate the future evolution of the oriented lake basins.

T

6 Acknowledgments

IP

The present study has been funded by the Swiss National Science Foundation (SNSF), grants N° 200021-122289 and 200020-141277/1. We thank the communities of Villa Banzer,

SC R

Miraflores, Somopae and Perotó, Dr. O. Sikuajara, Mr. M. Mostajo and Mr. J. Soruco, for granting us free access to their land. Many thanks to J. Barba and J-H May for several years of stimulating discussions on the topic. Field work assistance by B. Vogt is gratefully

NU

acknowledged. E. Canal-Beeby and two anonymous reviewers helped improve earlier

MA

versions of the manuscript.

7 References

Allenby, R.J., 1988. Origin of rectangular and aligned lakes in the Beni Basin of Bolivia.

D

tectonophysics, 145, 1-20.

TE

Allenby, R.J., 1989. Clustered, rectangular lakes of the Canadian Old Crow Basin. Tectonophysics, 170(1–2), 43-56.

CE P

Ashton, A.D., Murray, A.B., Littlewood, R., Lewis, D.A., Hong, P., 2009. Fetch-limited selforganization of elongate water bodies. Geology, 37(2), 187-190. Barba, J., 2003. Las lagunas de Moxos. In: CEAM (Ed.), Moxos: una limnocultura. Gràfiques 92, Barcelona, pp. 37-58.

AC

Belmonte, J.A., Barba, J., 2011. Can nature align? The enigma of Moxos' Lagoons— astronomy and landscape in south-western Amazonia, International Astronomical Union. International Astronomical Union, Oxford, pp. 135-143. Best, J., Woodward, J., Ashworth, P., Sambrook Smith, G., Simpson, C., 2006. Bar-top hollows: A new element in the architecture of sandy braided rivers. Sedimentary Geology, 190(1–4), 241-255. Bezerra, F.H.R., E. Amaro, V., Vita-Finzi, C., Saadi, A., 2001. Pliocene-Quaternary fault control of sedimentation and coastal plain morphology in NE Brazil. Journal of South American Earth Sciences, 14(1), 61–75. Bronk Ramsey, C., 2009. Bayesian analysis of radiocarbon dates. Radiocarbon, 51(1), 337360. Burn, C.R., 2004. Oriented Lakes. In: A. Goudie (Ed.), Encyclopedia of Geomorphology. Routledge Ltd, London, pp. 730.

ACCEPTED MANUSCRIPT Bye, J.A.T., 1965. Wind-driven circulation in unstratified lakes Limnology and Oceanography, 10(3), 451-458. Campbell, K.E., Frailey, C.D., Arellano, J.L., 1985. The geology of the Rio Beni: Further

T

evidence for Holocene flooding in Amazonia. Contributions in Science, Natural

IP

History Museum of Los Angeles County, 364, 1-18.

Carson, C.E., Hussey, K.M., 1962. The oriented lakes of Arctic Alaska. The Journal of

SC R

Geology, 70(4), 417-439.

Clapperton, C., 1993. Quaternary geology and geomorphology of South America. Elsevier, Amsterdam.

NU

Côté, M.M., Burn, C.R., 2002. The oriented lakes of Tuktoyaktuk Peninsula, Western Arctic Coast, Canada: a GIS-based analysis. Permafrost and Periglacial Processes, 13(1), 61-

MA

70.

Denevan, W.M., 1966. The Aboriginal Cultural Geography of the Llanos de Mojos of Bolivia. University of California Press, Berkeley.

D

Dumont, J.F., 1993. Lake patterns as related to neotectonics in subsiding basins: The example

TE

of the Ucamara Depression, Peru. Tectonophysics, 222(1), 69-78. Dumont, J.F., Fournier, M., 1994. Geodynamic environment of Quaternary morphostructures

CE P

of the subandean foreland basins of Peru and Bolivia: Characteristics and study methods. Quaternary International, 21(0), 129-142. Erickson, C.L., 2006. The domesticated landscape of the Bolivian Amazon. In: W. Balée, C.L. Erickson (Eds.), Time and complexity in historical ecology: studies in the

AC

neotropical lowlands. Columbia University Press, New York, pp. 236-278. Espurt, N., Baby, P., Brusset, S., Roddaz, M., Hermoza, W., Regard, V., Antoine, P.O., SalasGismondi, R., Bolaños, R., 2007. How does the Nazca Ridge subduction influence the modern Amazonian foreland basin? Geology, 35(6), 515. French, H.M., 2007. The periglacial evironment. John Wiley and Sons, Ltd., Chichester, UK. Gonzales, J., Aydin, A., 2008. The origin of oriented lakes in the Andean foreland, Parque Nacional Torres del Paine (Chilean Patagonia). Geomorphology, 97(3–4), 502-515. Grosswald, M.G., Hughes, T.J., Lasca, N.P., 1999. Oriented lake-and-ridge assemblages of the Arctic coastal plains: glacial landforms modified by thermokarst and solifluction. Polar Record, 35(194), 215-230. Hamilton, S.K., Sippel, S.J., Melack, J.M., 2004. Seasonal inundation patterns in two large savanna floodplains of South America: the Llanos de Moxos(Bolivia) and the Llanos del Orinoco(Venezuela and Colombia). Hydrological Processes, 18(11), 2103-2116.

ACCEPTED MANUSCRIPT Hanagarth, W., 1993. Acerca de la geoecología de las sabanas del Beni en el noreste de Bolivia. Instituto de ecología, La Paz. Hinkel, K., 2006. Comment on “Formation of oriented thaw lakes by thaw slumping” by Jon

T

D. Pelletier. Journal of Geophysical Research: Earth Surface, 111(F1).

IP

Hinkel, K.M., Frohn, R.C., Nelson, F.E., Eisner, W.R., Beck, R.A., 2005. Morphometric and spatial analysis of thaw lakes and drained thaw lake basins in the western Arctic

SC R

Coastal Plain, Alaska. Permafrost and Periglacial Processes, 16(4), 327-341. Jorgenson, M.T., Shur, Y., 2007. Evolution of lakes and basins in northern Alaska and discussion of the thaw lake cycle. Journal of Geophysical Research: Earth Surface,

NU

112(F2).

Langstroth, R.P., 1996. Forest islands in an Amazonian savanna of northeastern Bolivia,

MA

University of Wisconsin-Madison.

Livingstone, D.A., 1954. On the orientation of lake basins [Alaska]. American Journal of Science, 252(9), 547-554.

D

Lombardo, U., 2010. Raised Fields of Northwestern Bolivia: a GIS based analysis. Zeitschrift

TE

für Archäologie Außereuropäischer Kulturen, 3, 127-149. Lombardo, U., Canal-Beeby, E., Veit, H., 2011. Eco-archaeological regions in the Bolivian

CE P

Amazon: Linking pre-Columbian earthworks and environmental diversity Geographica Helvetica, 66(3), 173-182. Lombardo, U., May, J.-H., Veit, H., 2012. Mid- to late-Holocene fluvial activity behind preColumbian social complexity in the southwestern Amazon basin. The Holocene,

AC

22(9), 1035-1045.

Lombardo, U., Prümers, H., 2010. Pre-Columbian human occupation patterns in the eastern plains of the Llanos de Moxos, Bolivian Amazonia. Journal of Archaeological Science, 37(8), 1875-1885. Mann, C.C., 2008. Ancient earthmovers of the Amazon Science, 321, 1148-1152. Mayle, F.E., Langstroth, R.P., Fisher, R.A., Meir, P., 2007. Long-term forest-savannah dynamics in the Bolivian Amazon: implications for conservation. Philosophical Transactions of the Royal Society B: Biological Sciences, 362(1478), 291-307. McCormac, F.G., Hogg, A.G., Blackwell, P.G., Buck, C.E., Higham, T.F.G., Reimer, P.J., 2004. SHCal04 Southern Hemisphere calibration, 0–11.0 cal kyr BP. Radiocarbon, 46(3), 1087-1092.

ACCEPTED MANUSCRIPT Mertes, L.A.K., Dunne, T., Martinelli, L.A., 1996. Channel-floodplain geomorphology along the Solimões-Amazon River, Brazil. Geological Society of America Bulletin, 108(9), 1089-1107.

T

Morrell, G., Dietrich, J.R., 1993. Evaluation of the Hydrocarbon Prospectivity of the Old

IP

Crow Flats Area of the Northern Yukon. Bulletin of Canadian Petroleum Geology, 41(1), 32-45.

SC R

Pelletier, J.D., 2005. Formation of oriented thaw lakes by thaw slumping. Journal of Geophysical Research: Earth Surface, 110(F2).

Pelletier, J.D., 2006. Reply to comment by Kenneth Hinkel on “Formation of oriented thaw

NU

lakes by thaw slumping”. Journal of Geophysical Research: Earth Surface, 111(F1). Plafker, G., 1964. Oriented Lakes and Lineaments of Northeastern Bolivia. Geological

MA

Society of America Bulletin, 75(6), 503.

Plafker, G., 1974. Tectonic implications of the oriented lakes and lineaments in notheastern Bolivia. In: R.A. Hodgson, P.S.J. Gay, J.Y. Benjamins (Eds.), First international

TE

City, Utah, pp. 519-527.

D

conference on the new basement tectonics. Utah Geological Association, Salt Lake

Plug, L.J., West, J.J., 2009. Thaw lake expansion in a two-dimensional coupled model of heat

CE P

transfer, thaw subsidence, and mass movement. Journal of Geophysical Research: Earth Surface, 114(F1).

Price, W.A., 1968. Oriented lakes. In: R.W. Fairbridge (Ed.), Encyclopedia of Earth Sciences. Hutchinson and Ross, Pennsylvania, USA, pp. 784-796.

AC

Rex, R.W., 1961. Hydrodynamic analysis of circulation and orientation of lakes in northern Alaska. In: G.O. Raasch (Ed.), Geology of the Arctic. University of Toronto Press, Toronto, pp. 1021-1043. Roy-Léveillée, P., Burn, C., 2010. Permafrost conditions near shorelines of oriented lakes in Old Crow Flats, Yukon Territory, Canadian Permafrost Conference, 12 – 16 September 2010. Canadian Geotechnical Society, Calgary, AB, pp. 1509-1516. Ulrich, M., Morgenstern, A., Günther, F., Reiss, D., Bauch, K.E., Hauber, E., Rössler, S., Schirrmeister, L., 2010. Thermokarst in Siberian ice-rich permafrost: Comparison to asymmetric scalloped depressions on Mars. Journal of Geophysical Research: Planets, 115(E10). Whitney, B.S., Rushton, E.A., Carson, J.F., Iriarte, J., Mayle, F.E., 2012. An improved methodology for the recovery of Zea mays and other large crop pollen, with

ACCEPTED MANUSCRIPT implications for environmental archaeology in the Neotropics. The Holocene, 22(10), 1087-1096. Whitney, B.S., Dickau, R., Mayle, F.E., Soto, J.D., Iriarte, J., (in press). Pre-Columbian

T

landscape impact and agriculture in the Monumental Mound Region of the Llanos de

IP

Moxos, lowland Bolivia. Quaternary Research.

SC R

Figure Captions

Figure 1 Digital Elevation Model of the southeastern part of the LM, including the mid-

NU

Holocene Río Grande paleo-rivers and location of the three lakes studied (detailed in Fig. 5.).

Figure 2. The sinking/vibrating basement blocks model. Bedrock faults propagate through the

MA

overlying Quaternary sediments and create oriented and rectangular lake basins in the surface. In A the subsidence is due to a sinking basement block; in B the subsidence is due to a vibrating basement block. Regardless of the underlying cause, in both cases the effect on the

TE

D

surface is the same. Adapted from Plafker (1964).

CE P

Figure 3 Frequency distributions of permanent and dry lakes by size

Figure 4 Examples of the diversity of lake shape and size in the Llanos de Moxos. A and B, rectangular; C, oval; D and E, elliptical; F, coalesced; G-I, triangular; K and J trapezium; L,

AC

square. Based on Landsat imagery.

Figure 5 Stratigraphic transects from the outside to the inside of the lakes. Dotted white lines define the lakes’ basins. The early to mid-Holocene paleosol acts as a stratigraphic marker (see Fig. 1). Cores 52, 63, 81, 170, 205 and 210 provide the reference depth of the paleosol outside the lakes; cores 77 and 204 have been performed in areas of the original lakes’ basins that have been infilled; cores 78, 169, 171 and 209_b come from inside the lakes. Continuous black lines reconstruct the original lake bottom (previous to lacustrine infilling); dashed black lines connect the paleosol. Source of digital images: Google earth.

Figure 6 Spatial distribution of geometric lakes in the LM and overlap between lakes and major archaeological features (based on Lombardo et al., 2011).

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

Figure 1

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

Figure 2

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Figure 3

AC

CE P

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

Figure 4

Figure 5

AC

CE P

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

Figure 6

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

SC R

IP

T

Highlights  Cores were taken across the margins of 3 oriented and rectangular lakes in Bolivia  Stratigraphic markers indicate that these lakes are not tectonic  Radiocarbon ages show that lakes formed in the mid to late Holocene  Wind action is the most likely explanation for their shape and orientation