Palaeogeographic Reconstruction of Minchin Palaeolake System, South America: the Inﬂuence of Astronomical Forcing

Home , Cabana (ancient lake), Harper Lake, Mataro (ancient lake), Ouki, Sajsi, Salinas (ancient lake)

Geoscience Frontiers 5 (2014) 249e259

Contents lists available at SciVerse ScienceDirect

China University of Geosciences (Beijing) Geoscience Frontiers

journal homepage: www.elsevier.com/locate/gsf

Research paper Palaeogeographic reconstruction of Minchin palaeolake system, South America: The inﬂuence of astronomical forcing

Andrea Sánchez-Saldías a,*, Richard A. Fariña b a Universidad de la República, Facultad de Ciencias, Departamento de Astronomía, Iguá 4225, 11400 Montevideo, Uruguay b Universidad de la República, Facultad de Ciencias, Laboratorio de Paleobiología, Iguá 4225, 11400 Montevideo, Uruguay article info abstract

Article history: Current palaeoclimatic reconstructions for the Río de la Plata region during the latest Pleistocene (30,000 Received 3 December 2012 e10,000 yr BP) propose dry conditions, with rainfall at the Last Glacial Maximum amounting to one-third Received in revised form of today’s precipitation. Despite the consequential low primary productivity inferred, an impressive 17 May 2013 megafauna existed in the area at that time. Here we explore the influence of the flooding from a huge Accepted 2 June 2013 extinct system of water bodies in the Andean Altiplano as a likely source for wet regimes that might have Available online 1 July 2013 increased the primary productivity and, hence, the vast number of megaherbivores. The system was reconstructed using specifically combined software resources, including Insola, Global Mapper v13, Keywords: Planetary science Surfer and Matlab. Changes in water volume and area covered were related to climatic change, assessed Climate science through a model of astronomical forcing that describes the changes in insolation at the top of the at- Quaternary mosphere in the last 50,000 yr BP. The model was validated by comparing its results with several proxies 18 Palaeoecology (CH4,CO2,D, O) from dated cores taken from the ice covering Antarctic lakes Vostok and EPICA Dome C. Megafauna It is concluded that the Altiplano Lake system drained towards the southeast in the rainy seasons and Paraná Basin that it must have been a major source of water for the Paraná-Plata Basin, consequently enhancing primary productivity within it. Ó 2013, China University of Geosciences (Beijing) and Peking University. Production and hosting by Elsevier B.V. All rights reserved.

1. Introduction Serbian astronomer Milutin Milankovitch (1941) and is considered a major factor in environmental reconstructions (Berger, 1977, 1.1. Astronomical forcing 1980; Huybers, 2006). Apart from some caution posed by the conventional climato- During Quaternary periodic cycles in glaciation-deglaciation, logical approach, which claims the Earth atmosphere has its own temperature fluctuations determined high-latitude ice accumula- physical dynamics and is only crudely influenced by these cycles, tion and major variations in sea-level, as well as climatic, envi- astronomical forcing is considered the main cause of Quaternary ronmental and even geographical changes. These cycles are glacial-interglacial episodes. Since the original proposal by explained by periodic variations in insolation patterns due to Milankovitch (1941),refinements have arisen from improvements changes in the Earth’s orbital and rotational elements. This so- in numerical, computer-based algorithms for calculating the inso- called astronomical forcing theory was first proposed by the lation and also from the gathering of proxy data e a measured variable used to infer the value of another variable of interest in climate research. In Section 3.1 we show evidence of a strong fit between insolation and proxies using a simple model without any * Corresponding author. Tel.: þ598 25258624; fax: þ598 25258630. additional assumptions about atmospheric behaviour. E-mail addresses: andrea@fisica.edu.uy, [email protected] (A. Sánchez-Saldías). 1.2. Late Pleistocene palaeoclimate in mid-latitude South America Peer-review under responsibility of China University of Geosciences (Beijing) Among other climatic changes, and as a consequence of the lower availability of water circulating in the atmosphere during the colder parts of these Milankovitch cycles, many regions of the Earth Production and hosting by Elsevier had dryer climates than today. For example, the areas neighbouring

1674-9871/$ e see front matter Ó 2013, China University of Geosciences (Beijing) and Peking University. Production and hosting by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.gsf.2013.06.004 250 A. Sánchez-Saldías, R.A. Fariña / Geoscience Frontiers 5 (2014) 249e259 the present day Río de la Plata have been reconstructed in a clas- Here we review the geological evidence and reconstruct the sical work by Iriondo and García (1993) as much dryer and slightly palaeolake using digital procedures. We also test the possibility colder during the Last Glacial Maximum, with precipitation at that it had a substantial drainage to the Paraná Basin and, hence, about 350 mm and temperatures lower by 2.5e3 C, with more could have been a major source of water through such tributaries as marked seasonality. Even though more recent studies propose a Palaeopilcomayo and Palaeoparaná, which helped increase the more complex scenario (e.g., Garreaud et al., 2008; Zech et al., primary productivity to the levels required to support the 2009; Heil et al., 2010), the general approximation remains valid megafauna. 18 for our purposes. Analysis of oxygen isotopic ratios (d Oice) and insoluble dust On the other hand, the assemblage of megamammals that were from ice cores extracted from the Sajama Desert (Thompson et al., widespread in the region at that time includes up to 12 species with 1998) provide a time constrain of about 30,000 yr before present adult body masses over one tonne found at the same site (Fariña for the existence of the palaeolake, named Minchin in honour of et al., 2013). Darwin (1839) himself was astonished by such di- the British engineer and geographer John B. Minchin (Minchin, versity of megamammals and subtitled Chapter V of his journal of 1882; Steinmann et al., 1904), whereas carbonate-encrusted the Beagle’s Voyage “Large animals do not require luxuriant vege- shoreline deposits dated by 14C and U/Th techniques yielded tation”, in which commented that the association between giant even older ages between 44e34 and 68e72 kyBP (Rondeau, 1990; animals and forests was not always necessary and that “yet in Fornari et al., 2001; Placzek et al., 2006). Previous palaeolakes former days, numerous large animals were supported on the plains have also been described as forming a complex series called now covered by a thin and scanty vegetation”. Mataro, Cabana and Ballivián (Servant and Fontes, 1978; Lavenu Darwin’s perception was later assessed with modern tools of et al., 1984). Recently, several cycles with different lake configu- energetics and palaeoecology (Fariña, 1996), from which it was rations have been recognised for the last glacial period (Lavenu concluded that the rainfall inferred would have been insufficient to et al., 1984; Mourguiart et al., 1997; Sylvestre et al., 1999; yield a primary productivity high enough to support so many large Placzek et al., 2006, 2013): Ouki/Salinas Lake cycle between mammals. 125 kyBP and 80 kyBP, Inca Huasi Lake cycle about 46 kyBP, Sajsi Here we propose a possible source of water e that coming from Lake cycle between 24 kyBP and 20 kyBP, Tauca Lake cycle be- the drainage of a palaeolake system inferred to have existed in the tween 18 kyBP and 14 kyBP and Coipasa Lake cycle between 11 Altiplano region (Thompson et al., 1998). kyBP and 12.8 kyBP. Placzek et al. (2006) also proposed dropping the original name Minchin on the grounds that it has been used 1.3. Palaeolakes with different and even contradictory meanings. However, we intend here to continue its usage and, hence, maintain the hom- The Andean Altiplano palaeolake system is reconstructed as a age to the original discoverer of the whole system, which includes drainage basin of nearly 200,000 km2 lying between the western the northern Titicaca Basin, the southern Poopó-Coipasa-Uyuni (Occidental) Cordillera and the eastern (Oriental) Cordillera at an Basin and the connecting Desaguadero River Basin (Fig. 1), based average height of 3800 m above sea-level (masl). At present, Lake on the notion that these water bodies worked and still work as a Titicaca is found in the northern Altiplano at 3812 masl. Its area is system. 8372 km2, its depth is 281 m and the whole basin covers about Indeed, due to the associated effects of increase in pre- 56,270 km2. In the southern Altiplano, connected with Titicaca by cipitations, decrease in temperature and deglaciation processes the Desaguadero River, is Poopó Lake (3686 masl, 3191 km2 in area, (Blodgett et al., 1997), Lake Titicaca overflowed and discharged into 3 m deep and 27,700 km2 in basin area). Currently, the southern Poopó during the late Pleistocene, flooding both Coipasa and Uyuni Altiplano is a dry area that contains very scarce water bodies like salares (Baker et al., 2001), a phenomenon that also takes place Poopó Lake, but instead is characterised by two huge salares (salt today (Roche et al., 1991), albeit at a smaller scale. flats): Coipasa (3657 masl, 2218 km2 in area) and Uyuni (3663 masl, These changes in water volume and area were related to climatic 10582 km2 in area). change, that is here assessed through an astronomical forcing In the past, however, several large lakes have been proposed to model that describes the changes in insolation at the top of the have existed (Minchin, 1882; Steinmann et al., 1904; Servant and atmosphere over the last 50,000 yr. The model is validated by Fontes, 1978; Blodgett et al., 1997; Argollo and Mourguiart, 2000; comparing the computed insolation with several proxies (CH4,CO2, Placzek et al., 2006, 2011, 2013). The first attempts to date these D, 18O) from dated cores taken from the ice covering Antarctic lakes palaeolakes are those of Servant and Fontes (1978). Vostok and EPICA Dome C.

Figure 1. (a) Minchin Basin, as shown in Thompson et al. (1998), adapted from Fariña et al. (2013); (b) idem with georeferences. A. Sánchez-Saldías, R.A. Fariña / Geoscience Frontiers 5 (2014) 249e259 251

2. Material and methods same time, the Earth’s albedo decreases, so that a positive feedback mechanism is established. 2.1. Insolation

In this section we detail the method used for calculating insola- 2.2.2. Insolation model tion using the software Insola (Laskar et al., 2004). The sampling step The Earth’s insolation at the top of the atmosphere during the is constant throughout the integration and allows us to obtain equi- last 450 kyr, the period for which both proxy data are available and spaced series, providing more accurate results. It is noteworthy that stratigraphy has reached a wide consensus (Jouzel et al., 2007), the software Insola allows the integration step to be included, hence was calculated using the specifically developed software Insola permitting high-resolution numerical data to be obtained in the (Laskar et al., 2004). This software analyses the influence of the appropriate time interval (50 MyBP to þ20 MyBP) for the selected three Earth parameters, orbital (eccentricity) and rotational latitude. Therefore, the results of this latitude-dependent model are (obliquity and precession), that are relevant to palaeoclimatology. more accurate when the insolation calculated and the proxies used It then calculates numerical solutions for insolation quantities by a are at the same latitude, as shown in Section 3.1. direct integration of the gravitational equation of the orbital mo- Ò The results obtained were processed with Matlab (2013) for tion. The three leading terms (Laskar et al., 2004; Crucifix et al., generating plots and grids, and to set the periods we performed a 2006) in an analytical approximation for expansion of the eccen- frequency analysis with the algorithm Spectra (Gallardo and Ferraz- tricity from e50 Myr to þ 20 Myr, are probably influenced by Mello, 1997)andPAST (PAleontological STatistics, Hammer et al., resonances: 405 kyr (VenuseJupiter resonance), 95 kyr 2001). (MarseJupiter resonance) and 124 kyr (VenuseMars resonance). The importance of the proxies taken from the Vostok core lies Little difference is obtained by including the dynamic influences of not only in the strong correlation obtained, but also on the obser- other planets (Berger et al., 2004, 2005). In regard to precession vations by Placzek et al. (2013) that some results from the Altiplano and inclination, modelling is not as simple as it is for eccentricity, (changes in temperature of 5.7 1.1 C) are incongruent with because of the tidal dissipation in the Earth-Moon system. Since models applied to tropical regions and consistent instead with the angular velocity of the Earth is slowing down and the Earth- those in the Vostok record. In fact, Placzek et al. (2013) state that Moon distance is increasing, it is necessary to make an analytical ‘we recognize the uncertainties that results from this choice to use approximation for the obliquity and precession (Berger and Loutre, such a distant temperature record; unfortunately the Vostok record 1992; Berger et al., 1992; Laskar et al., 2004) and introduce the is the best continuous approximation of temperature available at expression of climate precession esinu, where e is the orbital ec- present’. On the other hand, as will be seen in Section 3.1,itis centricity and u is the longitude of the perihelion from moving relevant to look for the best correlation between the numerical equinox. results and the proxies, which depends both on the latitude and Insolation and proxies were plotted against the age and season considered. calculated or obtained, respectively. The chosen latitude was that of Lake Vostok, 78S, where the proxies were obtained, and the chosen time of the year was the beginning of southern summer 2.2. Validation of the model with data from ice cores from Antarctic and the equinox day of the austral autumn. The latter is the ‘critical lakes season’ (Berger, 1980), since it gave the best fit between Insola output and proxies (Fig. 2). Moreover, we examined the periods To calculate insolation, the software Insola (Laskar et al., 2004) obtained by way of a frequential analysis based on the algorithm was used. This approach was validated by observing how our re- Spectra (Gallardo and Ferraz-Mello, 1997), which takes into ac- sults fit the curves obtained by integrating the values of the proxies count non equi-spaced data, as is the case with the proxies we available at the same latitude as that for which the insolation was analysed. We also performed frequential analysis using PAST calculated. The chosen places were Lake Vostok (78S) (Petit et al., (Hammer et al., 2001), with an algorithm based on the Lomb 1999) and EPICA Dome C (75.1S) in Antarctica (Monnin et al., 2001; periodogramme, which is comparable with the classical technique Siegenthaler et al., 2005), from which stratigraphically controlled of FFT (Fast Fourier Transform), but without the risk of yielding the samples of gas trapped in ice cores were obtained for a time appearance of spurious frequencies as might occur with proxy spanning from 800 kyBP to present. data.

2.2.1. Sample selection 2.2.3. Frequential analysis To check this astronomical tuning two kinds of proxies were Many methods of performing frequential analysis exist for chosen: equi-spaced series; the most common being the Fast Fourier Transform (FFT). However, this procedure has the limitation of (1) Isotopic tracers related to temperature (dD and 18O; Lorius including only the assessment of the spectrum in specific fre- et al., 1979) allow inference on the evolution of warm/cool quencies. It splits the time series in an orthonormal base defined climate events when compared to the standard (SMOW, Stan- by N/2 frequencies, where N is the number of the data. It is not a dard Mean Ocean Water). As glaciers grow worldwide, less 16O continuum transformation, so it should not be regarded as pro- is retained in liquid water due to its enhanced probability of ducing real spectral lines. For that reason we chose to work with vaporising. Hence, the 18O/16O ratio in the ocean is increased, the algorithm Spectra. Data were included in the format (t, x), and so is the d18O ratio. As the world’s glaciers retreat and where t is the estimated age and x the concentration of the proxy temperature rises, values for this ratio become larger. There- for that age, and fitted for tentative values of function f,witha fore, the oxygen isotope ratio records the size of the ice sheets. function Y: (2) Greenhouse gases (CO2 and CH4; Loulergue et al., 2008), apart Y ¼ C1 þ C2*cos ð2 pf tÞ þ C3*sin ð2 pf tÞ from being proxies, also contribute to the variation of temperature. As the volume of these gases grows in the atmo- where constants Ci define amplitude, phase and zero level, and t is sphere, a stronger greenhouse effect takes place, the in years. The Spectra Correlation Coefficient (hereafter SCC) has temperature rises, and more water turns into vapour. At the values between 0 and 1, and is associated with the probability that 252 A. Sánchez-Saldías, R.A. Fariña / Geoscience Frontiers 5 (2014) 249e259

Figure 2. Insolation (in green) vs. (a) CO2 (in blue) and (b) CH4 (in black). the frequency f is an actual value of the system. The maxima in the 2.3. Topography plot SCC (f)inFig. 3 show the spectral lines and the periods, 1 18 calculated as 1/frequency (kyr ), for the proxies O and CO2 The Shuttle Radar Topography Mission (SRTM) Data Base (NOAA database). used dual radar antennas to acquire interferometric radar data, processed to digital topographic data at 1 arc sec resolution. We used elevation data to generate high-resolution digital topographic maps adapted to the region of interest (SRTM3-South America) (http://dds.cr.usgs.gov/srtm/version2_1/SRTM3/South_America/) and produced gradient overlay maps and proﬁles to link these with the variation of insolation and drainage of lakes and rivers obtained with software Global Mapper v13 and Surfer (Fig. 1). We also used USGS/Earth Resources Observation and Science (EROS) GTOPO30 data (http://eros.usgs.gov/products/elevation/ gtopo30/gtopo30.html), a global digital elevation model (DEM) with a horizontal grid spacing of 30 arc seconds (approximately 1 km).

3. Results and discussion

3.1. Lake Vostok

Fig. 4 shows more variation in the eccentricity than that suggested by the simple sinusoid pattern of the original graph by Milankovitch (1941). As expected, the sharp lines related to eccentricity give a period of about 100 kyr. Another period found (400 kyr, as in Berger, 1977, but with lower SCC) remains an open question (Muller and McDonald, 1997), although proxies with that period have been identified in lake bed depths (e.g., Newark Basin Coring Project; Olsen and Kent, 1996). In the graphical analysis it can be demonstrated that, if just the less spaced peak intervals are taken, the 100 kyr period is obtained, but the 400 kyr period is not so clearly visible, unless the envelope curve is considered. Previous numerical models showed correlations between astronomical variables, usually measured at 65N, and proxies taken at different latitudes, which resulted in inaccuracies and ad hoc hypotheses to explain discrepancies (Petit et al., 1999). Our model shows an appropriate fit and produced strong correlations and 18 Figure 3. Periodogrammes for O and CO2. Results were not smoothed or filtered. astronomical tuning of the proxies, taking into account the latitude A. Sánchez-Saldías, R.A. Fariña / Geoscience Frontiers 5 (2014) 249e259 253

Figure 4. Eccentricity vs. time for the Quaternary (last 2 Myr). Periods for this orbital parameter are shown as P1 ¼ 97.873 kyr, P2 ¼ 127.620 kyr, P3 ¼ 401.430 kyr, which are in agreement with other findings. Figure 6. Proxies and insolation vs. time (in kyr). The first three rows correspond to 18 the behaviour of the proxies CO2,CH4 and O in the time interval studied, along with of the data and the appropriate season for insolation numerical the best-fit insolation output (78S, fourth row). In the fifth row, results for 65Nare calculation at the same latitude. It is noteworthy that the insolation included for comparison purposes. curve also depends, both in amplitude and in phase, on the latitude of the time of the year chosen for running the model. In our case, it (3SD, dotted line) were considered. For 18O there is one frequency is possible to choose among different positions of the Earth along with its corresponding period near 100 kyr (eccentricity) and, to a fl its orbit, re ected in the parameter called mean longitude, an lesser extent, three values around 20e25 kyr (climatic precession). imaginary angle that is described by the Earth at constant velocity, For CO2, Fig. 3 shows a more complex pattern. There is a very sharp proportional to the time. frequency with high SCC near 100 kyr, and many values with a 18 In Fig. 5, four plots of proxies of O and the integration of much lower SCC above, but very close to, the line defined as a limit insolation at 78 S for three starts of seasons (the beginning of of 3SD. southern summer, spring and autumn, respectively) are shown. In the frequential analyses, without further ad hoc hypothesis, we Note that the third row yielded the best correlation between found (1) a better fit between data and proxies than previously ob- proxies and insolation, whose cross-correlation algorithm gives a tained, (2) that the variation of some isotopic tracers (like 18O) was fi coef cient of 0.57. In Fig. 6, the top three rows correspond to the correlated with eccentricity but without likely effects on climate at 18 behaviour of the proxies CO2,CH4 and O in the time interval the timescales considered, (3) that the Earth’s eccentricity was a fi studied, along with the best- t insolation output (78 S, fourth row) dominant parameter, and (4) a complex pattern and better defined for them. In the bottom row, results for 65N(Petit et al., 1999), frequency lines of greenhouse gases (e.g., CO2)thatmusthaveplayed which yielded a less accurate agreement, are included for com- a major role in the glaciation-deglaciation process. Fig. 7 corroborates parison purposes. through a wavelet analysis that, from 400 to 100 kyBP, the most Potential phase-lag between insolation forcing and climate relevant frequency is 97 kyr. The differences in the periodogrammes response is not considered here. But as can be clearly seen in the 18 for OandthegreenhousegasCO2 (Fig. 3) are related to the diverse fi plots, the climate response obtained from the proxies ts the behaviour of these gases in the atmosphere. Indeed, 18Oisanisotopic insolation curve without any phase-lags. tracer that reflects the climatic conditions during the interval of time 18 In the periodogrammes for O and the greenhouse gas, CO2 under study. In other words, it is a proper proxy, a variable that is (Fig. 3), only the points above average plus 3 standard deviations important for its relationship with another that is more difficult to observe. Therefore, it is not expected to show any direct effect in the climatic conditions at the time scales studied here. The 97 kyr frequency peak observed for this isotope is close to the period of the eccentricity. The other peaks, clustered about some 20 kyr, can be associated with climate precession. On the other hand, CO2 in our results has, as expected, a more complex pattern. There is a very sharp frequency with high SCC in association with eccentricity, and many values with a much lower SCC below, but very close to, the line defined as a limit. We suggest this is a consequence of the feedback system, with astronomical forcing reaching its threshold as the associated temperature increases, and triggering, once the maximum is high enough, a break in the bal- ance of low temperatures and ice formation. Moreover, our approach predicts that glaciations end suddenly when ice starts to melt, since the Earth’s albedo decreases and less solar radiation is reflected by the Earth’s surface. The same is true for CH4, which, while of lesser emissions, has an influence on the greenhouse effect, and hence on global warming, that is 23 times greater. This is Figure 5. Insolation for Vostok latitude (78S, first three rows) at the beginning of consistent with the importance of the Southern Hemisphere evi- different seasons, proxies of 18O (in ppm, dotted line), and insolation at 65 N at the beginning of northern summer (last row) vs. time (in kyr). The time-series of proxies dence per se, and not just as a record of the global climatic variation are not smoothed or filtered. (Jouzel et al., 2007). 254 A. Sánchez-Saldías, R.A. Fariña / Geoscience Frontiers 5 (2014) 249e259

Figure 7. Wavelet analysis for CO2.

3.2. Central Andes insolation

The austral summer is the season of maximum rainfall in the Andean highlands, 75% of the annual rainfall is concentrated in the period from December to March. Wet and dry phases in this region over the past 50,000 yr are in phase with the maxima and minima of insolation respectively, particularly when the dominant parameter is the astronomical forcing precession (23 kyr cycle) (Baker et al., 2001). The variation of insolation controls the climate system in the area, known as the South American Summer Monsoon, a seasonal wind that is generated in the inter-tropical regions because the continents are heated (and cooled) faster than ocean water. During the summer, in conditions of maximum insolation in the Altiplano, the air over the land starts to rise and creates a low pressure area. Since, to equalise pressure, wind direction is from high pressure Figure 8. Insolation for 50 kyBP at 16S (Titicaca), from November to March. areas to those of low pressure, a strong wind blows in from the ocean during the Southern Hemisphere summer (December to March). Rain is produced by moist air rising and cooling as it

Figure 9. Periodogrammes for the results of insolation for 50 kyBP at 16S (Titicaca), November, December and January. A. Sánchez-Saldías, R.A. Fariña / Geoscience Frontiers 5 (2014) 249e259 255 encounters the mountains, reinforced by the source of moisture, interpretations tend to attribute the lake-level highstands to which comes from the Atlantic Ocean over the entire Amazon re- meltwater input during glacier recession (Servant and Fontes, 1978) gion, and is transferred to the Altiplano. or to precipitation higher than today (Hastenrath and Kutzbach, Some authors suggest that the North Atlantic sea-surface tem- 1985). The first hypothesis is now considered unlikely because perature (SST) can influence the tropical region of the southern Clayton and Clapperton (1997) showed that a glacial advance hemisphere and that this could be the dominant mechanism that coincided with a highstand of paleolake Tauca at 13,300 14CyrBP. determined the late Pleistocene palaeoclimate in this region. For We, too, do not propose that the lakes overspilled due only to example, Blard et al. (2011) suggested that, during the Tauca glacier melting. episode, the temperature dropped by 6 or 7 C relative to the Mourguiart and Ledru (2003) claimed that, during the wet sea- present, while humidity was between 1.5 and 3 times higher than son, rainfall is related to the ITCZ at its southernmost position, today. Those values of temperature and precipitation were approximately 17S. During the last ice age, cooling by 5 C is evident concordant with that of the Tropical Atlantic SST and with the but it is still unclear whether tropical South America experienced position of the Intertropical Convergent Zone (ITCZ) during the moist or dry conditions during this period (Bush et al., 2001), and if interval between 17 and 12 kyBP. However, the late local Last Milankovitch’s(1941)approach should be corrected, especially in Glacial Maximumm (LGM) (about 15 kyBP) did not coincide with regard to the 23 kyr precession cycle and its influence on climate. glacial retreat, which took place 25 kyBP. Based on 14C and Clearly, the monsoon system for the LGM should be reviewed and 230Th/234U dating, Sylvestre et al. (1999) identified the lacustrine atmospheric circulation needs to be better understood. phases Tauca (13e12 kyr) and Coipasa (9500e8500 kyr) as signif- Fig. 8 shows the variation in insolation at 16S (average latitude icant hydrological events in the southern tropical Andes. Former of Lake Titicaca) for the months from November to March. The

Figure 10. (a) Normalised insolation for 50 kyBP a 16S (Titicaca) and 34.5S (Luján) on the 21st December. (b) Above: Normalised insolation for Titicaca Lake. Middle: Seasonal contrast for Luján (34.5S); Equinoxes MeS (MarcheSeptember) and SeM (SeptembereMarch). Below: Seasonal contrast for Luján (34.5S); Solstices (DecembereJune). 256 A. Sánchez-Saldías, R.A. Fariña / Geoscience Frontiers 5 (2014) 249e259

Figure 11. Topographic maps ordered in the sequence of water contribution. In all cases NWeSE slopes can be observed.

Figure 12. Topographic proﬁles showing the viability of the drainage from NW to SE. (a) Titicaca, (b) Poopó, (c) Coipasa, and (d) Uyuni. A. Sánchez-Saldías, R.A. Fariña / Geoscience Frontiers 5 (2014) 249e259 257

amplitude in the appropriate period of interest (Fig. 10a). This implies that the seasonal contrast curve for the solstices is essentially in phase with that of insolation with an insolation percentage difference for Luján (IL) between December and June of 65%, 63% and 64% for the maxima of IT at 47.6, 22.7 and 1.3 kyr, respectively (Fig. 10b). Thus, the optimal periods are those of maximal insolation in the Altiplano and minimal seasonal contrast in Luján during the equinoxes (CESL). For the two oldest maxima of insolation in Titicaca (47.6 and 22.7 kyBP), the CESL corresponds to periods in which the insolation in March is higher than that in September, which is maintained in time intervals that are relevant for this paper (49e35 kyBP and 24e12 kyBP). This is consistent with what is expected, since it follows the rain and spill period in Titicaca (November to March), whose climate shows low equinoxal contrast, especially in autumn (Fig. 10b). In Placzek et al. (2013; Fig. 8C), the temperature variation in Antarctica with the insolation at 15S and several proxies in the Figure 13. Region studied with current topography and Pilcomayo headwaters. Altiplano are compared. The 5 kyr delay in the Tauca Lake cycle after the maximum elevation in the insolation curves poses a problem that we do not find in our model. maxima of insolation occur in December, January and November, in decreasing order. The periodogrammes for insolation in those months shows that November, December and January coincide 3.3. System drainage with a period of 22.2 kyr, associated with climatic precession. To determine the variation of insolation according to latitude, we In optimal conditions of insolation (and hence, as explained choose the month of December as the most representative, allow- above, of precipitation), the Minchin system was transformed from ing a two-month lag for the system’s response to variation of a closed (endorheic) basin into an open system, in which the Titi- insolation (Fig. 9). caca Lake overflowed and discharged into Poopó, flooding both The best scenario for improving primary productivity in the Río salares, Coipasa and Uyuni (Baker et al., 2001), a phenomenon that de la Plata area was judged to be that combining high insolation in also takes place today (Roche et al., 1991)(Figs. 11 and 12, all maps the Minchin system, which implies water contribution through the in sequence Titicaca, Poopó, Coipasa and Uyuni). Other authors annual monsoon, with low seasonal contrast at 34.5S, the latitude have suggested the connection Titicaca-Desaguadero-Poopó-Coi- of the Luján Formation, near Luján, Argentina, whose richly pasa/Uyuni as the natural course of the water overspilling from fossiliferous Guerrero Member was deposited between 21 and Titicaca (Argollo and Mourguiart, 2000; Placzek et al., 2006, 2013). 10 kyBP (Tonni et al., 2003). Thus, the seasonal contrast for solstices This waterway is shown here through topographic profiles and (21st December minus 21st June) and equinoxes (21st September slopes. Following this, the water accumulated must have reached minus 21st March) was calculated for that latitude over the last the headwaters of the palaeoriver Pilcomayo (Fig. 13), and hence 50 kyr (Fig. 10). The insolation calculated for the last 50 kyr for the the hydrological Paraná-Plata Basin. 21st December at 16S and 34.5S show similar behaviour, both in In Fig. 14, the region of interest is shown according to the co- the timing and position of maxima and minima, and in the ordinates in Placzek et al. (2013; Figs. 1 and 2). The appropriate

Figure 14. Minchin Basin, corrected according to the coordinates in Placzek et al. (2013; Fig. 2). (a) Topographical map with main features, particularly Laka sill and Pilcomayo headwaters; (b) contour plot. 258 A. Sánchez-Saldías, R.A. Fariña / Geoscience Frontiers 5 (2014) 249e259

4. Conclusions

Our latitude-dependent insolation model, validated against proxies taken from ice cores in Antarctic lakes, accounts for the variations in the extent of the Minchin system during the late Pleistocene, which must have flooded periodically across the basin. Astronomically controlled insolation must have played a key role in establishing the conditions in which the Minchin system developed. 75% of the annual rainfall is concentrated during the summer. Orbital precession led the astronomical forcing that determined wet and dry phases in the Andean Altiplano over the past 50,000 yr. Fig. 11 shows the four main components of Minchin system: Titicaca Lake (3812 masl), connecting with Lake Poopó (3686 masl) by way of the Desaguadero River, Coipasa Salar (3657 masl) and Uyuni Salar (3650 masl). These features are arranged in an NWeSE direction. According to these heights, the excess water in the case of Titicaca flooding, must have discharged towards the other three Figure 15. Contour plot for the Laka sill region showing that the difference in height between Poopó and Uyuni/Coipasa is at least 15 m, hence the discontinuities in the components and, eventually, into the Palaeoparaná tributaries such curve for 3685e3700 masl (in blue). as the Pilcomayo River. 3-D reconstructions in Fig. 11a, b, c and d show that there were no obstacles for this NWeSE flow direction. Moreover, the overlapping slope maps show that the inward rela- contour plot is shown. These authors state that there is a 15-m high tive slopes are small and, hence easily reversed. hydrological barrier (Laka sill) between Poopó Lake and the salares, In Fig. 12, the same zones are analysed in terms of topographic although these levels can equalise depending on the model chosen. profiles to observe whether there are geographic features that In Fig. 15, the contour plot for that region is shown. For the barrier might have prevented the flow from Titicaca to Uyuni. The highest e to be overpassed, the curve for 3685 3700 masl (in blue) should be altitudes are located in the NW of the area and that there is no continuous. However, there is a topographic discontinuity, which higher particular relief (see above). Within the Minchin area the indicates that the estimated height of the lakes is not accurate at topographic gradient goes from 3812 masl to 3657 masl, which the scale of tens of metres. On the other hand, the results in Placzek would allow flooding in the above mentioned direction, both for et al. (2013; Fig. 2A and Table 2) show that the greatest extent of the system as a whole and to each of its components. some cycles (Ouki Lake cycle in the Coipasa/Uyuni basins and Sajsi As reconstructed, the Minchin system must have drained to- Lake cycle in the Poopoó Basin) are not known. Placzek et al. (2003; wards the southeast, finding the slope that gives birth to the Pil- 2 Fig. 2B) relate surface area (km ) with height (masl). Hence, since comayo River and so contributing moisture to the otherwise dry Río the greatest extent of the lake is not known, the maximum height de la Plata region. This must have enhanced primary productivity in cannot be determined. For instance, in Placzek et al. (2003; Table 2) an area that may have received as little as 300e400 mm annual it is stated that the highstand elevation for Ouiki and Salinas is precipitation. As a consequence of this compensation of the infer- ‘ ’ variable . red low rainfall, an impressive megafauna thrived in the plains of These results indicate that the highstand elevation for the whole mid-latitudes South America during the latest Pleistocene. system, considered for time periods of the order of 100 kyr, is not necessarily 3790 masl, but higher levels could have been achieved. Acknowledgements In our case, the hydrological barrier between the Minchin system and the headwaters of the Pilcomayo River is about 35 m (Fig. 15). Sebastián Tambusso drew Figs. 4 and 5. Edwin da Costa helped The level of 3830 masl is continuous, which must have allowed with the treatment of the drainage maps. spilling towards the Pilcomayo Basin (Fig. 16). Such flooding must have been of paramount importance in enhancing primary productivity in an otherwise arid area (Iriondo References and García, 1993), whose low precipitation is inconsistent with the richness of the megamammals found there in late Pleistocene Argollo, J., Mourguiart, P., 2000. Late Quaternary climate history of the Bolivian Altiplano. Quaternary International 72, 37e51. times. Baker, P.A., Seltzer, G.O., Fritz, S.C., Dunbar, R.B., Grove, M.J., Tapia, P.M., Cross, S.L., Rowe, H.D., Broda, J.P., 2001. The history of South American tropical precipitation for the past 25000 years. Science 291, 640e643. Berger, A., 1977. Support for the astronomical theory of climatic change. Nature 269, 44e45. Berger, A., 1980. The Milankovitch theory of paleoclimates. Vistas in Astronomy 24, 103e122. Berger, A., Loutre, M.F., 1992. Astronomical solutions for paleoclimate studies over the last 3 million years. Earth and Planetary Science Letters 111, 369e382. Berger, A., Loutre, M.F., Laskar, J., 1992. Stability of the astronomical frequencies over the Earth’s history for paleoclimate studies. Science 255, 560e566. Berger, A., Loutre, M.F., Mélice, J., 2004. Milankovitch and beyond. In: Proceedings of the Milutin Milankovitch Anniversary Simposium, pp. 31e50. Berger, A., Mélice, J.L., Loutre, M.F., 2005. On the origin of the 100-ky cycles in the astronomical forcing. Paleoceanography 20, PA4019. Blard, P., Sylvestre, F., Tripati, A.K., Claude, C., Causse, C., Coudrain, A., Condom, T., Seidel, J., Vimeux, F., Moreau, C., Dumoulin, J., Lavé, J., 2011. Lake highstands on the Altiplano (Tropical Andes) contemporaneous with Heinrich 1 and the Younger Dryas: new insights from 14C, UeTh dating and d18O of carbonates. Quaternary Science Reviews 30, 3973e3989. Figure 16. Contour plot for the Pilcomayo headwaters region. No discontinuities are Blodgett, T., Lenters, J., Isacks, B., 1997. Constraints on the origin of paleolake ex- apparent in the interval 3700e3830 masl (in yellow). pansions in the Central Andes. Earth Interactions 1, 1e28. A. Sánchez-Saldías, R.A. Fariña / Geoscience Frontiers 5 (2014) 249e259 259

Bush, M.B., Miller, P.E., DeOliveira, P.E., Colinvaux, P.A., 2001. Orbital forcing signal in Mourguiart, P., Ledru, M.P., 2003. Last glacial maximum in an Andean cloud forest sediments of two Amazonian Lakes. Journal of Paleolimnology 27, 341e352. environment (Eastern Cordillera, Bolivia). Geology 31, 195e198. Clayton, J.D., Clapperton, C.M., 1997. Broad synchrony of a Late-glacial glacier Mourguiart, P., Argollo, J., Corrège, T., Martin, L., Montenegro, M.E., Sifeddine, A., advance and the highstand of palaeolake Tauca in the Bolivian Altiplano. Wirrmann, D., 1997. Limnological and climatological changes in the Lake Titi- Journal of Quaternary Science 12, 169e182. caca basin (Bolivia) during the last 30 millennia. Comptes Rendus de l’Académie Cruciﬁx, M., Loutre, M.F., Berger, A., 2006. The climate response to the astronomical des Sciences Paris 325, 139e146. forcing. Space Science Reviews 125, 213e226. Muller, R.A., MacDonald, G.J., 1997. Glacial cycles and astronomical forcing. Science Darwin, C., 1839. Narrative of the surveying voyages of His Majesty’s Ships 277, 215e218. Adventure and Beagle between the years 1826 and 1836, describing their ex- NOAA database, http://www.ncdc.noaa.gov/paleo/, (last access 12.11.12.). amination of the southern shores of South America, and the Beagle’s circum- Olsen, P.E., Kent, D.V., 1996. Milankovitch climate forcing in the tropics of Pangaea navigation of the globe. Volume III journal and remarks. 1832e1836. Henry during the Late Triassic. Palaeogeography, Palaeoclimatology, Palaeoecology Colburn, London, 615 pp. 122, 1e26. Fariña, R.A., 1996. Trophic relationships among Lujanian mammals. Evolutionary PAST (PAleontological STatistics Software Package for Education and Data Analysis), Theory 11, 125e134. http://folk.uio.no/ohammer/past/, (last access 17.05.13.). Fariña, R.A., Vizcaíno, S.F., de Iuliis, G., 2013. Megafauna. Giant Beasts of Pleistocene Petit, J.R., Jouzel, J., Raynaud, D., Barkov, N.I., Barnola, J.-M., Basile, I., Benders, M., South America. Indiana University Press, Bloomington, 464 pp. Chappellaz, J., Davis, M., Delayque, G., Delmotte, M., Kotlyakov, V.M., Fornari, M., Risacher, F., Feraud, G., 2001. Dating of palaeolakes in the central Alti- Legrand, M., Lipenkov, V.Y., Lorius, C., Pépin, L., Ritz, C., Saltzman, E., plano of Bolivia. Palaeogeoraphy, Palaeoclimatology, Palaeloecology 172, Stievenard, M., 1999. Climate and atmospheric history of the past 420,000 years 269e282. from the Vostok ice core, Antarctica. Nature 399, 429e436. Gallardo, T., Ferraz-Mello, S., 1997. Understanding libration via time-frequency Placzek, C.J., Quade, J., Patchett, P.J., 2006. Geochronology and stratigraphy of late analysis. Astronomical Journal 113, 863e870. Pleistocene lake cycles on the southern Bolivian Altiplano: implications for Garreaud, R., Vuille, M., Compagnucci, R., Marengo, J., 2008. Present-day South causes of tropical climate change. Geological Society of America Bulletin 118 American climate. Palaeogeography, Palaeoclimatology, Palaeoecology 281, (5e6), 515e532. 180e195. Placzek, C.J., Quade, J., Patchett, P.J., 2011. Isotopic tracers of paleohydrologic change Global Mapper. Blue Marbel Geographic. http://www.bluemarblegeo.com/products/ in large lakes of the Bolivian Altiplano. Quaternary Research 75, 231e244. global-mapper.php, (last access 17/5/2013). Placzek, C.J., Quade, J., Patchett, P.J., 2013. A 130 ka reconstruction of rainfall on the Hammer, Ø., Harper, D.A.T., Ryan, P.D., 2001. PAST: paleontological statistics software Bolivian Altiplano. Earth and Planetary Science Letters 363, 97e108. package for education and data analysis. Palaeontologia Electronica 4 (1), 9 pp. Roche, M.A., Bourges, J., Cortés, J., Mattos, R., 1991. Climatologia e hidrologia de la Hastenrath, S., Kutzbach, J., 1985. Late Pleistocene climate and water budget of the cuenca del lago Titicaca. In: Dejoux, C., Iltis, A. (Eds.), El Lago Titicaca: Síntesis south American Altiplano. Quaternary Research 24, 249e256. del Conocimiento Limnológico Actual. ORSTOM-HISBOL, La Paz, Bolivia, Heil Jr., C.W., King, J.W., Zárate, M.A., Schultz, P.H., 2010. Climatic interpretation of a pp. 83e104. 1.9 Ma environmental magnetic record of loess deposition and soil formation in Rondeau, B., 1990. Géochimie isotopique et géochronologie des stromatolites the central eastern Pampas of Buenos Aires, Argentina. Quaternary Science lacustres quaternaires de l’Altiplano bolivien. Unpublished M.Sc. thesis. Uni- Reviews 29, 2705e2718. versité de Québec, Montréal. Huybers, P., 2006. Early pleistocene glacial cycles and the integrated summer Servant, M., Fontes, J.C., 1978. Les lacs quaternaires des hauts plateaux des Andes insolation forcing. Science 313, 508e511. boliviennes: premières interprétations paléoclimatiques. In: Evolution récente Iriondo, M.H., García, N.O., 1993. Climatic variations in the Argentine plains during des hauts plateaux andins en Bolivie. Cahiers ORSTOM.Série Géologie, 10(1), the last 18,000 years. Palaeogeography, Palaeoclimatology, Palaeoecology 101, pp. 9e23. 209e220. Shuttle Radar Topography Mission (SRTM) Data Base, http://www2.jpl.nasa.gov/ Jouzel, J., Masson-Delmotte, V., Cattani, O., Dreyfus, G., Falourd, S., Hoffmann, G., srtm/, (last access 17.05.13.). Minster, B., Nouet, J., Barnola, J.-M., Chappellaz, J., Fischer, H., Gallet, J.C., Siegenthaler, U., Stocker, T.F., Monnin, E., Lüthi, D., Schwander, J., Stauffer, B., Johnsen, S., Leuenberger, M., Loulergue, L., Lüthi, D., Oerter, H., Parrenin, F., Raynaud, D., Barnola, J.-M., Fischer, H., Masson-Delmotte, V., Jouzel, J., 2005. Raisbeck, G., Raynaud, D., Schilt, A., Schwander, J., Selmo, E., Souchez, R., Stable carbon cycle-climate relationship during the late Pleistocene. Science Spahni, R., Stauffer, B., Steffensen, J.P., Stenni, B., Stocker, T.F., Tison, J.L., 310, 1313e1317. Werner, M., Wolff, E.W., 2007. Orbital and millennial antarctic climate vari- SRTM3-South America, http://dds.cr.usgs.gov/srtm/version2_1/SRTM3/South_ ability over the past 800,000 years. Science 317, 793e797. America/, (last access 17.05.13.). Laskar, J., Robutel, P., Joutel, F., Gastineau, M., Correia, A.C.M., Levrard, B., 2004. Steinmann, G., Holk, H., Bistram, A., 1904. Zur Geologie des suedestland Bolivien. A long-term numerical solution for the insolation quantities of the Earth. As- Zentralbl. Mineral 5, 1e4. tronomy and Astrophysics 428, 261e285. Surfer Golden Software, http://www.goldensoftware.com/products/surfer/surfer. Lavenu, A., Fornari, M., Sebrier, R.I., 1984. Existence de deux nouveaux izpisodes shtml, (last access 17.05.13.). lacustres quaternaires dans l’altiplano Peruvo-Bolivien. Cahiers ORSTOM. Série Sylvestre, F., Servant, M., Servant-Vildary, S., Causse, C., Fournier, M., Ybert, J.P., Géologie 14, 103e114. 1999. Lake-level chronology on the southern Bolivian Altiplano (18e23S) Lorius, C., Merlivat, L., Jouzel, J., Pourchet, M.A., 1979. 30,000-yr isotope climatic during Late-Glacial Time and the Early Holocene. Quaternary Research 51, record from Antarctic ice. Nature 280, 644e648. 54e66. Loulergue, L., Schilt, A., Spahni, R., Masson-Delmotte, V., Blunier, T., Lemieux, B., Thompson, L.G., Davis, M.E., Mosley-Thompson, E., Sowers, T.A., Henderson, K.A., Barnola, J.-M., Raynaud, D., Stocker, T.F., Chappellaz, J., 2008. Orbital and Zagorodnov, V.S., Lin, P.-N., Mikhalenko, V.N., Campen, R.K., Bolzan, J.F., Cole- millennial-scale features of atmospheric CH4 over the past 800,000 years. Na- Dai, J., Francou, B., 1998. A 25,000-year tropical climate history from Bolivian ice ture 453, 383e386. cores. Science 282, 1858e1864. MatlabÒ, 2013. MATrix LABoratory, http://www.mathworks.com/products/matlab/, Tonni, E.P., Huarte, R.A., Carbonari, J.E., Figini, A.J., 2003. New radiocarbon chro- (last access 17.5.13.). nology for the Guerrero Member of the Luján Formation (Buenos Aires, Milankovitch, M., 1941. Cannon of insolation and the ice age problem (in Serbian). Argentina): palaeoclimatic signiﬁcance. Quaternary International 109-110, Royal Serbian Sciences 33, 1e633. 45e48. Minchin, J.B., 1882. Notes of a journey through part of the Andean table-land of USGS/Earth Resources Observation and Science (EROS) GTOPO30, http://eros.usgs. Bolivia in 1882. Proceedings of the Royal Geographical Society and Monthly gov/products/elevation/gtopo30/gtopo30.html, (last access 16.04.13.). Record of Geography New Monthly Series 4, 671e676. Zech, W., Zech, M., Zech, R., Peinemann, N., Morrás, H.J.M., Moretti, L., Ogle, N., Monnin, E., Indermühle, A., Dällenbach, A., Flückiger, J., Stauffer, B., Stocker, T.F., Kalim, R.M., Fuchs, M., Schad, P., Glaser, B., 2009. Late Quaternary palaeosol Raynaud, D., Barnola, J.M., 2001. Atmospheric CO2 concentrations over the last records from subtropical (38 S) to tropical(16 S) South America and palae- glacial termination. Science 291, 112e114. oclimatic implications. Quaternary International 196, 107e120.