I, f' af

33 1

'0- '0- IS,

o1 PALAEOMAGNETIC STRATIGRAPHY OF PLIOCENE k- i CONTINENTAL DEPOSITS OF THE BOLIVIAN ALTIPLANO m

:y al N.THOUVENY' and hf. SERVANT'

J. LGQ/CIL'RS, Luminy, Case 907, 13288 Marseille Cedex O9 (France) 11- 'UR 103 ORSTON, 70-74 Route d'Aulnay, Bondy 9.3 (France)

n- (Received April 28, 19%; revised and accepted September '26, 19SS) e,

.5. 1 Abstract. 31 2: Thouveny, N.and Servant, LI., 19S9. Palaeomagnetic stratigraphy of Pliocene continental deposits of the Bolivian .Altiplano. Palaeogeogr., Palaeoclimatol., Palaeoecol., 'io: 331-344.

Fluvio-lacustrine post-Miocene deposits of the Bolivian Altiplano have been sampled in 400 m of section in the and basi'ns in order to provide a magnetostratigraphic scheme for chronostratigraphic interpretations. Stable characteristic components were isolated after stepwise .AF or thermal demagnetizations. The directions of cleaned magnetization were compared to the normal and reverse dipole field directions at the site. The plot OE the mean directions calculated for each straiigraphic level define successive magnetozones which can be identified on the standard geomagnetic polarity scale with the help of the K/Ar ages obtained on tuff layerã. The deposition of n the middle and upper part of the La Paz Formation covers the Gauss epoch. The Gauss/bIatuyama (2.4s Ma) and the GilbertiGauss (3.4 Ma) limits are located in the top and the middle OF the formation. The occurrence of the first g!acial/interglacial oscillation can be placed in the early Matuyama (Ca 2.2 Ma) in agreement with the oxygen iãotopic record from the equatorial ocean.

Introduction Ayo basins, 100 km south from La Paz (Evern- den et al., 1977). This can be considered as a The Altiplano plateau at 4000 m spreads maximum age since remains of Pliocene mam- between the Western and the Eastern cordil- mals were discovered at Ayo Ayo, in the lower lera of the Andes. The La Paz basin () part of the Formation, correlatable to situated at 16'30'S, 6S'W (Fig.1) was cut in the the La Paz Formation (Hoffstetter, 1971), and Altiplano by a regressive erosional phase dur- at La Paz in the lower part of the La Paz I l ing the Late Pleistocene related to the setting formation (Villarroel, 1978). The K/Ar age was I of the upper hydrologic basin of the Amazon. recently confirbed on a correlatable tuff layer I First reported by Troll and Finterwalter (1935), near the base of the La Paz Formation called the lacustrine, fluvial and glacial deposits were Toba Cota Cota: the age given by Servant et al. described in detail by Dobrovolny (1962). The (1988) is 5.5 I: 0.02 Ma. stratigraphy was recently reinterpreted by Servant (1977) and Balliviàn et al. (1978). Stratigraphy and recent chronological I The first chronological data obtained on the data I post-hfiocene deposics (5.5 Ma) results from i E(/& analyses on a tuff layer (Toba Umala) The lower sequence (La Paz Formation sensu 1 underlying the deposits in the Umala and Ayo stricto) is more than 500m thick, and can be I i' oo:~l-ols~js9/so:3.so c, 1989 Elsevier Science Publishers B.V. AC : PU: Purapura

O 3 km u 'i .2900 Fig.1. Geographical situation of the La Paz Basin and sampling localities. described at successive outcrops (Figs2 and 6). Ayo and 2.SF0.1 at La Paz (Lavenu et al., It consists of lacustrine silty and clayey beds 19SS). interlayered with fluvio-lacustrine sandy and Above the Chijini tuff, the La Paz Formation gravelly layers. The sediments are grey, often is enriched with conglomeratic units. The lightly coloured in yellow (silts) or green upper part of the formation is affected by a (clays). Some clayey levels are brightly col- major discontinuity (erosional phase) overlain oured (red, white or black). by fluvio-glacial or glacial deposits: till In the upper part of this sequence, a white (50-100 m thick) of the Calvario glaciation on bed (2-12m thick) was recognized as a tuff which lies a thick (100-200 m) interglacial (Toba Chijini, sometimes presenting an ignim- conglomeratic bed (Purapuråni Formation). X britic facies), which was formerly reported as second tuff layer (Toba Sopari) can be observed 1 overlying the La Paz Formation (Dobrovolny, in the upper valley of the Rio Kaluyo (North 1962) but can clearly be observed as inter- from La Paz) interlayered between the Cal- I bedded in the upper part of it. After unsuccess- vario till and the Purapurani conglomerates. i full K/Ar dating attempts (ages ranging Dobtovolny (1962) did not differentiate this i from 0.5 to 11 Ma), two coherent ages were tuff from the Chijini tuff and considered the finally obtained by K/Ar analyses on bio- lowest till older than the Chijini tuff but he tite by MacIntyre (quoted in Clapperton, noted that in other sections of the basin, the I 1979): 3.2'7k0.14 and 3.2840.13 Ma. Other Chijini tuff directly overlies the La Paz Forma- ! i ages now available are 2.840.1 Ma at Ayo tion. Clapperton (1979) also claimed to have i 1 ! 333

sw NE /

IR : lrpavi section VS : Viscachani section AC: Achocalla section PU: Purapura section

Fig.?. Summary stratigraphy of the post->riocene deposits of the La Paz basin. I =Paleozoic or Cretaceous basement. 2=La Paz formation s.s. 3=Tills. .i=interglacial conglomerates. 5= tuff layers. LP= La Paz. C= Calvario. FP= Purapurani formation. K= Kaluyo. S=. Ch= Choqueyapu. I-IV design the erosional phases. observed glacial deposits below the Chijini integrated in the dated samples and so falsify tuff. Following stratigraphical arguments and the results. As such inclusions were not using the ages provided by McIntyre in Clapper- observed in the Chuquiaguillo valley, the age ton (19í9) Clapperton dated this glacial phase obtained from this site might be considered as prior to 3.2 Ma. However, Clapperton did not more reliable than the other two. present any precise description or stratigraphic Other successive tills can be observed in log to support his arguments. All the clear the upper part of the Kaluyo valley attributed and complete sections of the Basin clearly to the Kaluyo, Sorata and Choqeyapu gla- show that the Chijini Tuff not only overlies the ciations. These tills are separated by inter- La Paz formation, but is interlayered in its glacial deposits, erosional phases and palaeo- upper part (note that Clapperton’s new section soils. is situated between the Capellani section and the Viscachani section of Ballivian et al., 1978) Magnetostratigraphic study This problem is further complicated by con- tradictory KIAr ages obtained on the Sopari Magnetostratigraphic investigations of post- tuff 2.7 f0.1 and 2.8 f0.1 Ma (remarkably con- Miocene deposits from Bolivia were success- sistent with the ages of the Chijini tuff) on fully carried.out by MacFadden et al. (1983) on two samples from the Kaluyo valley and the Quaternary Tarija Formation (south 1.650.1Ma on one sample from the Chuquia- Bolivia). The palaeomagnetic investigations of guillo valley. However, the observation of the La Paz Formation were initiated by Ser- multiple inclusions of granitic materials vant in 1981: specimens were tested for the (pebbles) in the outcrops of the Kaluyo valley intensity and sta-lsility of the remanence at the suggests that old biotites might have been LGQ. A complete sampling program was car- ,

334 ried out in May 19S4 on five different sections laminated clayey levels only provided one or of the La Paz basin (Figs2 and 6). two specimens each. (1) The IRPAVI section (Fig.6a). 130 m thick. comprises the bottoni part of the La Paz Palaeomagiiet ic ai1 a ljses Formation: 31 levels were sampled. Unfortu- nately. the link with the middle and upper part Measurements were made with a Balanced of the formation could not be determined until Flusgate Spinner Magnetometer (Molyneux. now. 19ïl) connected to a microcomputer (Tucholka (2) The VISCACHANI section (Fig.Gb), about and Tessier, 1957). Natural Remanent Magneti- 210 ni high. comprises the middle aiid upper zation (NRM) intensities range from 0.5 to part of the La Paz Forgation (Chijini tuff 20 mA ni- (1 millihinpere/nieter = 1 niicro- included) and the Calvario till. However, for Gauss). The highest values (0.35.4 ni-’) are reasons of accessibility. the levels above the recorded in the tuff layers. The NRlI values Chijini tuff could not be sampled. Twenty-nine are not plotted on the figures because un- levels were sampled in the l‘ioni below the cleaned remanent magnetizations are not tuff. representative: furthermore, characteristic (3) The levels above the Chijini tuff were remanent magnetizations were isolated at dif- sampled in the basin of the Rio Achocalla ferent thermal steps providing a large and (tributary to the La Paz basin) (Fig.6~).Twelve meaningless spectrum of intensity values. levels were sampled in a Som section called Susceptibility nieasurements were carried Achocalla. out with a Bartington Als2 meter: the values (4) Rare and thin clayey layers could be range from 5 to 200 x SI units (Fig.6a-c): sampled in the Calvario till and in the Purapu- the highest values (>SO x IO-’ SI) have not rani interglacial complex (Pura Pura section. been recorded in rhe tuff layers but in the silty Fig 6d). The scarcity of the subsampled levels and clayey layers of the Irpavi section. is due to the difficulty of finding fine-grained The uncleaned NRM directions seem to be layers and safe sampling sites. strongly influenced bh7 the present-day field. (5) Finally. the upper tuff (Toba Sopari) were Viscosity tests (Thellier and Thellier. 1959) sampled in the high valley of the Rio Kaluyo. were carried out on the levels of the Irpavi As close correlations were sho\?rn between section. The specimens (1 per level) were set in the Laz Paz formation at Laz Paz and the a normal field (X north: 2 dou7n) for 6 days; Umala formation at Ayo Ayo, a sampling was the magnetic moment A12 was measured; the also carried out on rare suitable layers in the specimens were then reset in a reverse field latter: 6 levels were sampled in an 18 ni section (X south; 2 up) for 6 days; the moment A12 from clayey and silty layers and one level from was measured. The rate of viscosity: the overlying tuff (unnamed): 14 in of unsani- (All - M2/M1 f 1\42) x 100. varies from 2 to 45°/o pled section separate the sampled sedimentary (Fig.6a). As the viscous remanent inagnetiza- layers from the tuff. tion acquired on a few million years could The stratigraphy presented here results from be strong enough to hide the characteristic the observations made during the sampling. remanent magnetization. stepwise demagnetiza- the description of the sampled levels and the tions were applied. Twin pilot specimenstaken data presented by Balliviàn et al. (19ïS). from each level were submitted to alternating Sampl in g tech n i qu e: t h e ni a gn et i c n ort li field (AF) and/or thermal demagnetization at direction was marked on the horizontal surface respectively 5-1OOniT or 50-650 C. The AF of each block. Two or three blocks per level demagnetization was performed \vith a Schoii- were collected on a total of about 100 levels. In stedt GSD - 1 demagnetizer. High temperature general d-6 cubes (8 cm3) \vere cut for each treatments were made 111 a furnace protected level ivith a rotary saw. but a few of the by a triple magnetic shield: proper cooling \vas 33.5

insured by pulsed air circulating in the plots revealed that thermal demagnekization shielded cooling unit. The median destructive (Fig.4) gives more accurate results than AF. field generally lies between 10 and 50 mT and The 300'C step was retained for the routine half of the initial intensity is removed between demagnetization as it insures a sufficient 150' and 300". In some levels, less than 10% of removal of the secondary magnetization and the initial intensity remained after heating at leaves a residual magnetization strong enough 600'C suggesting a small contribution of high to be accurately measured. blocking temperature minerals. Progressive The individual directions, retained after inductions of isothermal remanent magnetiza- treatment, were each compared to directions tions (IRM) show that the saturation is nearly corresponding to normal and reverse dipole acquired after application of a 0.2T field fields at the site (resp. Dec=360', Inc= -30'; . (Fig.3); the coercivity of remanence ranges Dec= MO", Inc = 30'). As the sedimentation from 20 to 35mT. This suggests the major rates could be high enough to allow a record of contribution from pseudosingledomain (1-13 medium to long period palaeosecular varia- pm) grains of magnetite in the NRM. tion, the vectors lying inside a 50' arc around Comparative analysis of the orthogonal these directions are considered as normal or

I .. I kr -2 -1 #! 1 2 H (1O-I T) 1 2 i '-

IRl2A IRZOA

i

VS16A VSZOA Fig.3. Acquisition of Isothermal Remanent Magnetization and back field IRM (.- 0.2 to 0.2 T). The saturation magnetization SIRM is not reached for specimens IRIZA and IR20A but is completely acquired before 0.2 T for specimens VS16.4 and VS2Oil. The coercivity of remanence-(c-r) ranges around 25, 30 mT. 336

IR15A

Y,Z 1 grad. - 0.4 mA.m-I X

AClOA 500

Y, z 1 graduation- 1 PA,~-IPA. y,z :i/

VS26A T li VS24A

1 grsd.. 1 r,A.r.-I xKxY,Z 1 grad.= 0.6 rrY4.m-I O io Y,Z 1 grad.= 10 rrA.n-l

Fig.4.Orthogonal projection of the magnetic vectors during thermal demagnetization. The thermal steps are indicated in 'C. For IRl5.4. IRI.%A7I:S2'4.4 and l'S26A the normal component is stable. For IR36A and AClOA: the reverse magnetizztion appears after eliminarion of spurious normal components (200 to 300:~). reverse; the directions lying outside this 50' istic magnetizations. This is explained by the arc are considered as intermediate. When addition of vectors of opposite direction in the intermediate directions occur from the same first case and of the same direction in the level as reverse directions the former are second case. interpreted as due to incomplete removal of the VRAI: they are rejected for polarity interpreta- tion. TABLE I The cleaned directions are plotted on a Statistical parameters for each section stereographic projection (Fig.5) and the statis- tical parameters are given in Table I for each Dec.(') Inc.(') AT Y 95 K section. The mean inclinations of the ra'erse groups do not reach the dipole field inclination Irpa ci Normal O -26 S7 4.1 13.5 (30') ivhich confirms that, in some cases, small ReFXerse lÏ9 15 40 5.2 9.2 secondary components still persist. It was I'iscochani N 3 -25 29 4.8 15.8 noted that weak SRM intensity and strong R 168 13 16 12.9 6.9 viscosity rates were generally associated with Achocalla S S -34 12 12.9 9.0 reversed. characteristic magnetizations while R 1Ï8 27 20 10.9 ï.9 -4s 2 -._ stronger XRl1 intensity and weaker viscosity 4x0 A?.O s 1.5 rates nere associated nith normal character- R lÏ5 35 16 11.5 10 .- - .I. __

..a--

33;

O O

X

-1 90 270 90 90 60 30

180 180

a) IRPAVI section b) VISCACHANI section

O

-1 ,. n e e e 180

c) ACHOCALLA section Fig.% Stereographic projection of the individual vectors after 300'C treatment (the characteristic magnetizations). For each section normal and reverse groups clearly appear. Black dots are vectors pointing downward (positive inclination) and white dots are pointing upward (negative inclination). Stars represent the mean direction of each group, the ellipse drawn around the stars are the projection of the cone of confidence at 95%.

Magnetostra t igra p hic interpretat i on period palaeosecular variation. The succession ) ! of polarity zones is given for each separate The mean directions, calculated from all the stratigraphic section. The limits of the magne- ! individual cleaned directions retained at each I tozones are arbitrarily set midway between level, are plotted along the stratigraphic logs two sites of opposite polarity. The normal and l in Fig.Ga-d. The changes in polarity are reverse magnetozones are respectively labelled clearly deduced from the declination record N and R,followed by an integer. As the Irpavi while the sign of the inclination at such a low section cannot be directly connected with the latitude is affected by the medium to long others, the magnetozones were labelled I(1rpa- 338

\'!SC. RATE JCEPT!ß!;IT! POLARITY 4Q)5 2 110 I 11 - --- * 40 -

39 - ?a - +. 1R2 31 - 26 - .-L . 35 - . .-. + .-._. i: - --. -. ...-...... __.- 33 - ..,. o _II ;.,.... -::: I' .. - . - ... t. 3i - f. +

30 - i o

28 - c. 21 - .i 26 - t 25 - + 24 - 23 - 12 - 21 - t $3 1 t ia - 17 - t 16 - + 1'. - .+ 14 - + __ia - + 12 - li - ìo -

9- A

a- .+ + 7- + +-

6- + --a-- 5- + - IR1 a- + ï-

CLAYS a CONELOYERATES

SILTS @TILL TLIFF

Fig.6a. For explanation see p. 310. vi)R and IhT.The zone IR1 is defined in the first specimen and 8 m remain uncovered. Zone IR2 18m (5 sites). the IN1 zones extends to ï5m comprises the upper 3Om of the section (12 (24 levels): for reason of cohesion of the -sites). sediment. site 39 did not provide any reliable The Viscachani section starts rvith a rc-\.erse 3

I.;, &'

a-- ...

LITHOLOGY SUSCEPTIBILITY DECLINATION I NCL I NAT i O11 AAA 90 AA A h -80 -O 180 - 29 - 2a - 27 - 26 - 25 - 24 -

23 - 22 '- 21 -

20 - 19 - 18 - 33 17 - o

N2

16 - o 15 -

+ R2 14 -

+ 13 - 12 -

11 -

10 -

9- E;?

8-

7-

6-

3- O

R?

2- -0-

1-

Fig.6b. For explanation see p. 310. 310

LITHOLOGY SUSCEPTIBILITY DECLIiiATION IKCLIKATION POikfilTY O EO O leo -00 O0 - o

r- +

9

+ +

o o

+ -0-

o +

o o

o +

b

DECLI !

b

o +

Eig.6. Summary lithology, viscosity rates (if available), susceptibility (SI units) average declination and inclination (degree) at each level; z95 is given by the horizontal bars or integrated.in the size of the point. The magnetostratigraphic interpretation is given as a succeãsion of normal (black) and reverse (white) magnetozones. Vertical intervals are 10 m. a. Irpavi section (see p. 33s). b. Viscachani section (see p. 339). c. Achocalla section (p. 340, upper). d. Purapura section (p. 340: Io\ver).

zone RI (30m,3 sites) followed by zone ìV1 the 3 specimens from two different blocks. (50 m, 6 sites). Zone A2 is defined on 22 m at 4 Zone R3 is also defined at 1 site by 3 speci- sites, a thin normal zone .Y2 is defined at one mens. The 43 uppermost meters of the sec- site only but the polarity is firmly established tion represent a normal polarity zone (Arg, ---by a highly stable normal magnetization on 11 sites) if \ve except site 23 which shows a .. _-

*-.-

341

large standard deviation; the weak residual Correlation to the standard geomagnetic remanence ( < 0.5 m-\ m- ’1 may have been polarity time scale and chronological covered by the background noise of the magne- implications tometer. The sequence sampled on the Viscachani Palaeontological arguments and radiometric section ends just below the tuff: it is completed datings already cited suggest a Pliocene age on the Achocalla section (Fig.6~):normal for the bottom of the lacustrine formation. and polarity zone N3 covers the Chijini tuff and a middle Pliocene age 3.7-3.2 Ma for the 20 m above it (3 sites). The polarity scale of the Chijini tuff, e.g. the upper part of the forma- La Paz formation ends with zone R4, recorded tion. On this basis, the magnetostratigraphic by 60 m of lacustrine sediments. The fluviatile results of the La Paz formation can be com- conglomerates at the top of the formation pared to the standard geomagnetic polarity could not be sampled. In the Calvario till and time scale (GPTS) of the last 5 Ma (Mankinen the Purapurani conglomerates, the sampling and Dalrymple, 1979). The polarity succession was restricted because of the lack of suitable N1 to IV3 is remarkably identical to that lithologies. One site in the middle of the till recorded during the Gauss epoch. The RIIN1 carries reversed directions. The 30 m sampled boundary can be interpreted as the Gilbert/ in the interglacial conglomerates (Fig.6d) pre- Gauss boundary (3.4 Ma), magnetozone R2 sent a reverse polarity interrupted by a normal and R3 as the Mammouth (3.05-3.15 Ma) and polarity recorded on two sites. Kaena (3.01-2.92 Ma) events and the lV3,’R4 Finally, the four specimens of the Sopari tuff boundary as the Gauss/Matuyama boundary (Kaluyo valley) present a very high intensity (2.48 Ma). The absence of stratigraphic con- (2 A m- ’), significantly different from the in- nection between the Irpavi section and the tensity of the Chijini tuff (0.2 X m-’). in spite Viscachani section raises the problem of the of identical susceptibility values (Ca 100 x identification of the IRl, IN1 and IR2 magneto- SI units). The declinations before cleaning are zones. The altitudinal situation (165 ni below significantly deviated from the normal dipole the bottoni of the Viscachani section). the declination (D= 300’). -4cleaning at 30 mT was palaeontological arguments and the age 5.5 Ma necessary to change the declination to 330’. At of the tuff underlying the Pliocene fossiliferous this stage. the intensity equals 3% of the initial deposits of the Uniala Formation at Ayo Ayo intensity. It is thus reasonable to conclude that and La Paz Formation at Cota Cota allo\y to a high secondary magnetization was super- place this section between 5.2 and 3.1SIa imposed on the original one. The position during the Gilbert epoch. of the sampled tuff section in the Kaluyo In Fig.7, the magnetozones are plotted ver- valley suggests the possibility of a lightning sus their age on the GPTS. The five chronologi- effect. The paleomagnetic data do not provide cal data respectively assigned to the polarity an argument to differentiate the two tuff reversals (Rl/Nl, Nl/R,3, N2lR2, R3jN3 and layers. N3/R4)define five points A-E. Stable deposition The sis levels sampled from the Umala rates throughout the whole Gauss epoch are Formation at Ayo Ayo present a well-defined suggested by the alignment of the points, in reversed polarity while the overlying tuff is of spite of slightly different slopes between AC normal polarity (directions and statistical and CE. The average value equals I7 cm per parameters are listed in Tabler). The faunal thousand years. If extrapolated, the slope of AC arguments tend to associate this reverse mag- locate the bottoni of the Viscachani section netization to zone IR1 of the Irpavi section around 3.6 Na. Such an extrapolation is invalid while the normal polarity of the tuff is in for the upper part of the formation because of agreement with the normal polarity of the the higher frequency of conglomeratic beds. Toba Chijini andior the Toba-Sopari. By interpolation, the Chijini tuff, located in 342

MATUYAMA GAUSS GILBERT 5.0 Age (Ma) ._-- ca'2.7 I I 25( I l R4 Il I Il I

( 12( IR2

IN 1

IR 1 c ? RELATIVE HEIGTH (meters) CT= Chijini Tuff

Fig.?. Magnetostratigraphic results (versus height) against the standard Geomagnetic Polarity time scale (Mankinen and Dalrymple, 19i9). This representation show the remarkable constance of the deposition rate across the middle and upper part of the La Paz formation and the high sedimentation rates implied by identification of zone Ih'l to any of the normal events in the Gilbert epoch the middle part of zone N3, can be dated defined during the Gauss epoch. An alternative around 2.7 Ma e.g. not significantly younger hypothesis would consist of an interruption of than the ages given by Lavenu et al. (1988; ca sedimentation during the reversed period sep- 2.8Ma); on the other hand, the ages given by arating two successive normal events. Mac Intyre in Clapperton (1979) (Ca 3.27) are The reverse polarity of the oldest till (Cal- O.ï Ma too old. vario till) and the Purapurani conglomerates The interpretation of the succession of and their position above the La Paz Formation magnetozone IRl: INl, IR2 raises problems place them in the Matuyama Epoch. The because of the lack of connection between the normal event, recorded in a few meters (8-15 mj, Irpavi and the Viscachani sections. As it is must be related to one of the normal events in clearly older than the succession RI and R4 the early Matuyama (Reunion 1, 2 or Olduvai). and younger than 5 Ma (palaeontological and We must emphasize that, in our strati- radiometric arguments), it must be related to graphic interpretation, the Sopari tuff is located part of the Gilbert epoch. Magnetozone IN1 at about the same position as the Purapurani recorded in 75, of lacustrine to fluviolacus- conglomerates (e.g. lying on the Calvario till) trine sediments, must be related to one of the and also shows a normal polarity; these two normal events in the Gilbert epoch. Howe\~er. normal polarity levels may be considered as the duration of such events does not exceed nearly contemporaneous. The possibility that about 0.15 Ma; this implies minimum values of these represent the Olduvai event would be in deposition rates of ca 50 cm per thousand years agreement with the age of ca 1.6 Ma obtained e.g. three times higher than the average on the tuff. - 343

The major erosional phase which affected Lacustrine and fluviolacustrine conditions the La Paz Formation may have also removed prevailed during the interval 5.5-2.48 Ma in the tracks of eventual high amplitude climatic the Altiplano basin. oscillations preceding the Calvario glacial. Consequently, this till might be related to any Acknowledgements cold phase occurring after 2.48 hIa (Gauss/ Matuyama limit) and before 1.8 Ma (Olduvai event). Contacts with B. MacFadden (Florida Uni- versity and Florida state Museum) before the Discussion field work are acknowledged. This work was financed by ORSTOM (Institut Français de Late Neogene climatic variations recorded Recherche Scientifique pour le dévellopement in the equatorial Pacific Ocean have been en coopération). The help of J. Argollo, and D. reconstructed by oxygen isotope stratigraphy and J. Wirrmann as well as the technical with a time calibration provided by the magneto- support of the University of La Paz in the field stratigraphy (Shackleton and Opdyke, 1976, were greatly appreciated. Susceptibility 197'7) in two sites (cores V28-239, 3'N-l60'E measurements were carried out by Ulrike and V28179, 4"--1IO'W). These studies pre- Thouveny. sent the evidence of high amplitude oscilla- tions of 6"O during the late Pliocene. The References authors interpret these variations of 0.76 and l%o as reflecting glacial phases with magni- Ballivihn, O., Bles, J. L. and Servant, M., 19'78. El Plio- Cuaternario de la region de la Paz (Andes Orientales, I and 1 tudes at least 2/3 of the last glaciation of the Bolivia). Cah. ORSTOX SQr.Géol., lO(1): 101-113. wer i upper Pleistocene. The d''O progressively Clapperton, C. &I., 1979. Glaciation in Bólivia before 3.2'7 Jrmal I i increases during the whole Gauss epoch and Myr. Nature, 27'7: 375-3'7'7. i reaches its maximum values after the Gauss/ Dobrovolny, E., 1962. Geologia del valle de la Paz. Publ. i Matuvama boundary. Several maxima are then Min. hlinaspetrol. Bol., 3, 153 pp. 1 Evernden, J., Kreg, S. and Cherroni. C., 19'77. Potassium- tive I recorded between the GausslMatuyama bound- i argon ages of some Bolivian rocks. Econ. Geol., 72: n of i ary and the end of the Olduvai event. In an 1042-1061. equatorial Atlantic site (DSDP site 397, 26'N- Hoffstetter, R., Martinez, C., Nunoz-Reyes. J. and Tomasi, l.5'W) Shackleton and Cita (1959) showed that P., 19'71. Le gisement d'.Avo Ayo (Bolivie), une succes- the first significant increase in d1'0 of Plio- sion stratigraphique Pliocène-Pléistocène datée par des mammifères. C.R. Acad. Sci., 273: 2172-2475. cene age occurs between 2 and 2.4Ma; they Hooghiemstra, H., 1951. Vegetational and climatic history 1 ;ion interpreted it as evidence of the onset of of the high plain of Bogota, Colombia: a continuous The glaciations in the Northern Hemisphere. record of the last 3.5 million years. In: Diss. Bot., 79, m), Hooghiemstra et al. (1954) have drawn a 368 pp. I Lavenu, A., Bonhomme, hI. G., Vatin-Perignon, N. and De 3 in curve of fluctuations of forest pollen through Pachtère, P., 1955. Neogene magmatism in the Bolivian ai). the last 3.5 Ma, from the study of the Funza Andes (between 16' and 183). Part I: numerical strati- sti- core (300m) drilled in a sequence of tuffs graphy (K/Ar) and tectonics. J. South Am. Earth Sci., ted and lacustrine deposits located at 2600m l(2): in press. ani near Bogota in Colombia (5'N, 75'W). In MacFadden, B. J., Siles, O., Zeitler, P., Johnson, N. M. and Campbell, K. E., Jr., 1953. Magnetic polarity stratigra- 'ill) agreement with the oceanographic data, large phy of the middle Pleistocene (Ensenadan) Tarija wo amplitude oscillations appear around 2.4 Ma formation of southern Bolivia. Quat. Res., 19: 1'72-187. as reaching their maximum between 1.9 and Mankinen, E. A. and Dalrymple, G. B., 19'79. Revised iat 1.5 Ma. Finally, the present study attests geomagnetic polarity time scale for the interval 0-5 my. B.P. J. Geophys. Res., 81: 615-625. in that the glacial maximum recorded in the Molyneus, L., 1971. A complete results magnetometer for ed Bolivian Andes occurred between 2.48 and measuring Remanent Magnetization of rocks. Geophys. 1.S m.y. B.P. J.R. Astron. Soc., 21: 429-433. i i 344

Servant: M., 1977. Le cadre stratigraphique du Plio-Quater- Shackleton, N. J. and Opdyke, N. D.,1977. Oxygen isotope naire de I'AltipIano des Andes tropicales en Bolivie. and palaeomagnetic evidence for early Northern hemi- Rech. Fr. Quat. Bull. Assoc. Fr. Etud. Quat., 50: 324-327. sphere glaciation. Nature, 270: 216-219. Servant, h4.. Sempéré, T., Argollo. J.: Bernat, Al., Ferraud, Thellier, E. and Thellier, O., 1959. Sur l'intensité du champ G. and Lo Bello, P., 1989. hlorphogenèse et soulèvement magnétique terrestre dans le passé historique et géologi. de la cordillère orientale des Andes centrales de Bolivie que. Ann. Geophys., 15: 265-376. au Néogène. C.R. .4cad. Sci. Paris, in press. Troll, C. and Finterwalter, R., 1935. Die Karten der Shackleton. N. J. and Cita, hl. B., 19ï9. Oxygen and carbon . Cordillera Real und des Talkessel von La Paz (Bolivien) isotope stratigraphy of benthic foraminifera at Site 39ï: und die Diluvialgeschichte der zentralen .Anden. detailed history of climate changes during the late Petermanns Mitt., 81: 393-399. Neogene. In: Initial Reports of the Deep Sea Drilling Tucholka, P. and Tessier, J., 1987. Interfacing the DIGICO Project. 47. U.S. Goyernment Printing Office, Washing Fluxgate Spinner magnetometer to an Apple IIE micro- ton, D.C., 1, pp. 433-445. computer. Geophys. J.R. Astron. Soc., 69: 705-712. Shackleton, N. J. and Opdyke, N. D., 1976. Oxygen isotope Villarroel, C., 1978. Revalidacion y redescription de and palaeomagnetic stratigraphy of Pacific core V28-239, Posnanslqtherium desaguaderoì Liendo. 1943; Tosodon- late Pliocene to latest Pleistocene. Geol. Soc. Am.Mem., tidae (Notoungulata) del Plioceno superior Boliviano. 145: 449-464. Rev. Serv. Geol. Bol. 16.

I- C