Lithostratigraphic and magnetostratigraphic data from late Cenozoic glacial and proglacial sequences underlying the Altiplano at La Paz, Bolivia

We provide lithostratigraphic and magnetostratigraphic data derived from a Plio-Pleistocene continental sediment sequence underlying the Altiplano plateau at La Paz, Bolivia. The record comprises six sections along the upper Río La Paz valley, totaling over one kilometre of exposure and forming a ~20-km transect oblique to the adjacent Cordillera Real. Lithostratigraphic characterization includes lithologic and stratigraphic descriptions of units and their contacts. We targeted gravel and diamicton units for paleomagnetic sampling to address gaps in the only previous magnetostratigraphic study from this area. Paleomagnetic data – magnetic susceptibility and primary remanent magnetization revealed by progressive alternating field demagnetization – are derived from 808 individually oriented samples of flat-lying, fine-grained sediments. The datasets enable characterization of paleo-surfaces within the sequence, correlation between stratigraphic sections, and differentiation of asynchronous, but lithologically similar units. Correlation of the composite polarity sequence to the geomagnetic polarity time scale supports a range of late Cenozoic paleoenvironmental topics of regional to global importance: the number and ages of early glaciations in the tropical Andes; interhemispheric comparison of paleoclimate during the Plio-Pleistocene climatic transition; timing of and controls on inter-American faunal exchange; and the variability of Earth's paleomagnetic field.


Data
Geologic sequences underlying the Altiplano plateau in the South American Andes provide extensive, but underexplored records of late Cenozoic continental paleoenvironments. Due to the Altiplano's long history as an internally drained basin [1,2], its sequence of up to 12 km of Tertiary sediments is relatively complete [3]. The low-energy depositional environments represented by many units [3][4][5] makes these sediments suitable recorders of variations of the ancient geomagnetic field on a wide range of time scales [6], as demonstrated by the results of the small number of paleomagnetic investigations in the region [7][8][9][10].
Despite the importance of records from the sub-Altiplano fill sequence, their ages are generally poorly constrained. Current chronologic control is based largely on radiometric dating of volcanic beds within the sequence [11][12][13], but many of the ages are unreliable [14]. Magnetostratigraphy at a few localities across the Altiplano constrains ages of non-volcanic units as well as their accumulation rates [7][8][9][10].
Here we present chronostratigraphic data from the upper part of the fill sequence at La Paz, Bolivia, where it is extensively exposed. The data come from six sections within and adjacent to the city of La Paz ( Fig. 1 and Table 1). Our lithostratigraphic descriptions and magnetostratigraphic data, respectively, build on the Plio-Pleistocene stratigraphic framework developed by previous workers [5,[15][16][17][18][19][20][21] and greatly expand upon the only previous paleomagnetic study in the area [8]. The data provide new insights into several aspects of the Altiplano and adjacent Cordillera Real area during the late Pliocene and Early Pleistocene [24]. Specifically, these data can be used to: demonstrate facies relationships of coeval units fining away from the high cordillera; constrain the number and ages of recurrent Pliocene and Early Pleistocene glaciations of the tropical Andes; revise the ages of several gravel sequences; quantity rates of sediment accumulation, the spatial variability of which help to characterize syndepositional tectonism; constrain the onset of incision of the local Altiplano surface and thus the approximate time of drainage capture of the eastern Altiplano by headwaters of the Amazon River system; and compare the timing of environmental change related to early glaciation of the Central Andes with patterns of Plio-Pleistocene faunal evolution and dispersal in the Americas, including occasional migrations leading up to the Great American Biotic Interchange.  Table 2; Figs. 5 and 6A of Ref. [14]) is located on the west bank of Río Kaluyo, where it curves east above the Limanpata landslide. It is the farthest upstream exposure in the Río Kaluyo/Choqueyapu valley (Fig. 1C) and consists of two exposures, each containing a 10-m-thick tuff at 4260-4270 m a.s.l. The upper exposure (137 m) extends from the base of the Chijini Tuff to the Altiplano surface and is exposed in a gully entering the west side of Quebrada Aquatiña. The lower part of the exposure [22] was uncovered during recent (ca. 2007) roadwork just downvalley of Quebrada Aquatiña. The top of this lower section aligns with Dobrovolny's [5,18] 7-m type section of the Patapatani Drift on the opposite side of the valley and thus greatly extends exposure of the Patapatani Drift. Nineteen metres of the lower exposure are covered by spoil dumped downslope during road construction. A 3-m-high exposure on the west side of Quebrada Aquatiña, which does not appear to have slumped [22], provides the only details on stratigraphy and paleomagnetism in this largely covered zone (Fig. 6A of Ref. [14]). The sequence below the base of the exposed section (4165 m a.s.l.) is buried beneath Late Pleistocene and Holocene glacial and colluvial deposits down to Río Kaluyo (4125 m a.s.l.). Most clasts in gravel units are rounded to subrounded. b The paleosols are classified as either strong or weak, depending on the degree of development of soil horizons and pedogenic structure. Strong paleosols are those with B horizons thicker than 50 cm (e.g. Fig. 5C of Ref. [14]) and pedons coated with clay (e.g. Fig. 5I of Ref. [14]). In contrast, weak paleosols are those with B horizons thinner than 30 cm, and little or no clay translocation and pedon formation.    [14]) and for mean directional data by stratigraphic section (Fig. 3D and Table 1 of Ref. [14]). e Primary remanence directions determined by intersection of great circles (GC) used for group mean directional data ( Fig. 3D and Table 1 of Ref. [14]), but not in overall statistics of entire sample collection ( Fig. 3A-C and Table 1 of Ref. [14]). Most clasts in gravel units are rounded to subrounded. b The paleosols are classified as either strong or weak, depending on the degree of development of soil horizons and pedogenic structure. Strong paleosols are those with B horizons thicker than 50 cm (e.g. Fig. 5C of Ref. [14]) and pedons coated with clay (e.g. Fig. 5I of Ref. [14]). In contrast, weak paleosols are those with B horizons thinner than 30 cm, and little or no clay translocation and pedon formation.  Table 4; Fig. 6C of Ref. [14]) is located in Quebrada Capellani, a deeply incised network of gullies on the east bank of Río Choqueyapu, 0.5 km upstream from the hairpin curve on the autopista (Fig. 1C). The uppermost 80 m of the main section are exposed only in vertical inaccessible cliffs; exposures in branches of Quebrada Tangani, 300 m to the southeast, provide access to units 10-12, which we correlate to the upper part of the Tangani section by elevation and by a thick, laterally persistent silt bed 185 m above the base of the main section. Most clasts in gravel units are rounded to subrounded. b The paleosols are classified as either strong or weak, depending on the degree of development of soil horizons and pedogenic structure. Strong paleosols are those with B horizons thicker than 50 cm (e.g. Fig. 5C of Ref. [14]) and pedons coated with clay (e.g. Fig. 5I of Ref. [14]). In contrast, weak paleosols are those with B horizons thinner than 30 cm, and little or no clay translocation and pedon formation. Weakly stratified, matrix-supported diamicton with thin basal zone of poorly sorted pebble-cobble gravel (10% granitic clasts)

173
Strong paleosol b 2 Weakly stratified, poorly sorted, clast-supported, multi-lithic gravel, coarsening upward from pebble-cobble to cobble-boulder (granite content increases in the upper part of the unit from o20% to 450%); 10-mthick zone near the base of the unit is covered 134 1 Rhyolitic tuff 10 a All diamicton units contain striated and faceted clasts; most clasts are subrounded to subangular.
Most clasts in gravel units are rounded to subrounded. b The paleosols are classified as either strong or weak, depending on the degree of development of soil horizons and pedogenic structure. Strong paleosols are those with B horizons thicker than 50 cm (e.g. Fig. 5C of Ref. [14]) and pedons coated with clay (e.g. Fig. 5I of Ref. [14]). In contrast, weak paleosols are those with B horizons thinner than 30 cm, and little or no clay translocation and pedon formation.  Table 6; Fig. 6E of Ref. [14]) mostly follows the old railway ascending to the Altiplano along the west slope of the Río Choqueyapu valley (Fig. 1C). The lower part of the section crosses this slope obliquely between quebradas Jacha and Pantisirca along the old railway route where it passes under the aqueduct. The upper part of the section follows the railway route on its final (60-m elevation gain) approach to the plateau. This section roughly coincides with the 'Pura Pura' (Aqueducto) section of Bles et al. [19] and Ballivián et al. [20], particularly the lowest 80 m. The Purapura section is~2 km up-valley of the approximate location Thouveny and Servant [8] give for their Purapura magnetostratigraphic section. Most clasts in gravel units are rounded to subrounded. b The paleosols are classified as either strong or weak, depending on the degree of development of soil horizons and pedogenic structure. Strong paleosols are those with B horizons thicker than 50 cm (e.g. Fig. 5C of Ref. [14]) and pedons coated with clay (e.g. Fig. 5I of Ref. [14]). In contrast, weak paleosols are those with B horizons thinner than 30 cm, and little or no clay translocation and pedon formation.  Table 7; Fig. 6F of Ref. [14]) is exposed in the gullies of a ridge west of Laguna Jacha Kkota, rising from the floor of the Achocalla basin to the Altiplano surface (Fig. 1C). The ridge is an intact remnant of the fill sequence below the Altiplano, which was left behind when a gigantic, early Holocene earthflow created the Achocalla basin [23] shortly before 11,485-10,965 cal yr BP [24]. The section includes the Chijini Tuff and 42 m of fine-grained sediments directly below it. The Achocalla magnetostratigraphic section of Thouveny and Servant [8] extends from the base of the tuff 80 m upslope along the same ridge, but does not reach the Altiplano surface. We included the magnetostratigraphy of the uppermost 10 m of the sedimentary sequence at a nearby section (~3.2 km to the southeast); Fig. 1 to extend the Jacha Kkota section to the Altiplano surface. The exact stratigraphic alignment of the two sections is uncertain, but in view of the similar elevations of the Altiplano surface and the Chijini Tuff at both sites, the upper part of the sequence probably starts 40-45 m above the top of Thouveny and Servant's [8] Achocalla section, in agreement with the stratigraphy of the Achocalla basin margins reported by Bles et al. [19] and Ballivián et al. [20]. Weakly laminated fine sandy silt 10 a All diamicton units contain striated and faceted clasts; most clasts are subrounded to subangular.
Most clasts in gravel units are rounded to subrounded. b The paleosols are classified as either strong or weak, depending on the degree of development of soil horizons and pedogenic structure. Strong paleosols are those with B horizons thicker than 50 cm (e.g. Fig. 5C of Ref. [14]) and pedons coated with clay (e.g. Fig. 5I of Ref. [14]). In contrast, weak paleosols are those with B horizons thinner than 30 cm, and little or no clay translocation and pedon formation.

Paleomagnetic data
See Tables 2-7.     Fig. 6B of Ref. [14] for stratigraphy and stratigraphic positions of sample groups. Position, sampling height in metres above base of section; χ, mean magnetic susceptibility of collected samples (×10 −6 SI units); n, number of samples; D and I, mean declination and inclination, respectively; k, precision parameter; α 95 , circle of confidence (P¼ 0.05); p, polarity. a Error between 10°and 20°underlined; error greater than 20°double underlined.  Fig. 6C of Ref. [14] for stratigraphy and stratigraphic positions of sample groups. Position, sampling height in metres above base of section; χ, mean magnetic susceptibility of collected samples (×10 −6 SI units); n, number of samples; D and I, mean declination and inclination, respectively; k, precision parameter; α 95 , circle of confidence (P¼ 0.05); p, polarity. * Magnetic enhancement of paleosol compared to the parent material in which it formed. a Error between 10°and 20°underlined; error greater than 20°double underlined. b Colluvium draping incised valley slope (possible mass flow deposit).

Lithostratigraphic characterization
We measured and described six sections along the western margin of the La Paz and Achocalla basins, totaling 1100 vertical metres of exposure of the sediment sequence underlying the Altiplano plateau ( Table 1). The sections are exposed in steep valley slopes, gullies, and road cuts and form ã 20-km-long transect through the eastern Altiplano margin, oblique to the trend of the Central Andes (Fig. 1). Sedimentologic and stratigraphic characterization include texture, structure, lithology, colour, clast size and shape, sorting, weathering features, and the nature of contacts. We divided units on the basis of major changes in material properties and on the occurrence of major hiatuses indicated by paleosols or erosional contacts. We measured unit thicknesses using a TruPulse 200 Laser Range Finder, and stratal thicknesses and sizes of clasts using a graduated metric scale. We measured the long-axes orientations (trend and plunge) of 50 elongate clasts from each of eight units at the two sections closest to the Cordillera Real (seven units at the Patapatani West section and one unit at the Patapatani East section: Fig. 2).  Fig. 6F of Ref. [14] for stratigraphy and stratigraphic positions of sample groups. Position, sampling height in metres above base of section; χ, mean magnetic susceptibility of collected samples (×10 −6 SI units); n, number of samples; D and I, mean declination and inclination, respectively; k, precision parameter; α 95 , circle of confidence (P¼ 0.05); p, polarity. * Magnetic enhancement of paleosol compared to the parent material in which it formed. a Error between 10°and 20°underlined; error greater than 20°double underlined. b Remanence directions obtained by the intersection of great circles.

Materials
The sample collection comprises 808 oriented cylindrical samples (2.1 cm diameter, 1.8 cm length) collected typically in groups of six (ranging from three to 16) from 124 stratigraphic levels at the sections (Table 1). Sampling gaps due to limited exposure, inaccessibility, or unsuitably coarse sediments were filled where possible by sampling closely aligned units at nearby exposures. We collected larger numbers of samples in gravel and diamicton units to provide a more complete magnetostratigraphic record; coarse units are more likely to yield problematic paleomagnetic results [25] and are thus less commonly sampled in magnetostratigraphic studies, including the only previous paleomagnetic study in the La Paz area [8]. During subsequent field visits, we re-sampled sites that produced indeterminate polarity or incoherent magnetization characteristics.
Samples were typically taken in horizontally bedded zones of predominantly silt and fine to medium sand. Where these were not available, we collected samples from the matrices of gravel and diamicton units, avoiding granules and pebbles. Where possible, sampling included material both above and below unit boundaries. Samples were stored in magnetic shields at the University of Lethbridge following transport from the field and between measurements.

Magnetic susceptibility
Prior to demagnetization, we measured bulk magnetic susceptibility of each sample with a Sapphire Instruments SI-2B magnetic susceptibility meter.

Magnetic remanence
We measured natural remanent magnetization of each sample with an AGICO JR-6A spinner magnetometer. We re-measured remanence after stepwise alternating field (AF) demagnetization with an ASC Scientific D-2000 alternating-field demagnetizer in fields up to 200 mT. One or two pilot samples, having either representative or relatively high magnetic susceptibility, were selected from each group. These pilots were demagnetized at 10 to 16 closely spaced steps (intervals of 2.5-10 mT up to 80 mT, and 10-30 mT above 80 mT). The remaining samples from each group were then demagnetized at 4 to 10 steps (5-30 mT spacing) guided by characteristic magnetizations of pilot samples. Each sample was demagnetized to 20% or less of the natural remanent magnetization. Median destructive fields for most samples range from 10 to 80 mT, although a small number of samples included hard components of magnetization that remained following demagnetization at 200 mT AF (the limit of the equipment used).
We determined remanence directions for most samples by principal component analysis [26] and for a small number of samples ( o2%) by the intersection of great circles [27] (Table 1). We calculated mean remanence directions by group (Tables 2-7), stratigraphic unit (Tables 2-7), and polarity (Table 1 and Fig. 3 of Ref. [14]). Sample-specific and mean remanence directions were calculated using AGICO's Remasoft v. 3.0.

Transparency document. Supporting information
Transparency data associated with this article can be found in the online version at https://doi.org/ 10.1016/j.dib.2018.05.038.