The southern Atacama Desert (27–30°S) is a climatic transitional region between the tropics and the extratropic source of moisture. As such, it is not certain whether the influence of one, or the other, or both, controls glacier and climate dynamics in the region (e.g., Zech et al., 2006). There is still debate about which source of moisture was responsible for driving glacier dynamics and the associated hydrographic evolution during the Local Last Glacial Maximum (LLGM) and termination (e.g., Ammann et al., 2001). Therefore, determining the former geographic influence of extratropical winter westerly and/or tropical easterly wind belts in this region remains a major open question. The limited dated records in the southern Atacama region still hinder reaching firm conclusions. Veit (1993), Veit (1996), and Ammann et al. (2001) pointed to extra-tropical westerlies as responsible for driving former ice/climate fluctuations in this part of the Andes. Jenny and Kammer (1996) remark on the tropical influence through the easterlies origination in the Amazonian and the Chaco Region (e.g., D’Arcy et al., 2019), whereas Zech et al. (2017) suggest that both seasonal circulation systems were operating together. These interpretations are based on geomorphological and geochronological studies of landforms sculpted by ancient glaciers, but often lack a holistic view of the evolution of arid landscapes.

One of the valleys in the Atacama Desert where glacial landforms which originated from former glacial advances are preserved is El Encierro valley. Pioneer studies carried out by Veit (1993), Veit (1996), Jenny and Kammer (1996), and Grosjean et al. (1998) suggest that the El Encierro River Valley (ERV, 29.1°S–69.9°W, 3,750–4,150 m asl, Figure 1) was occupied by mountain glaciers that extended several kilometers during the LLGM. More recently, investigations by Zech et al. (2006) and Zech et al. (2017) have dated ¹⁰Be the outer moraine (MII) to 20–22 ka during the LLGM inboard moraines (MIII–MV) been interpreted as recessional glacial re-advances dated between 16 and 18 ka. Zech et al. (2017) suggest this glacial re-advance was related to the humid phase recorded in the tropical Andes known as the Central Andean Pluvial Event (CAPE; Latorre et al., 2006; Quade et al., 2008). With this evidence, these authors suggest that this glacial re-advance is linked to a tropical supply of moisture reaching the southern Atacama Desert (Zech et al., 2006; Zech et al., 2017).

This research report presents a geomorphological map and reports seven new cosmogenic ¹⁰Be exposure ages obtained from boulders and an ice-molded bedrock surface in the paleo-drift of the ERV. The ERV is a tributary of the El Huasco River (Figure 1), one of the southern Atacama Desert's main Andean catchments where the geomorphological evolution associated with Late Pleistocene and Early Holocene hydro-climatic changes is remarkably well preserved. Based on our geomorphological mapping in the field, different glacial landforms were characterized and dated with ¹⁰Be to reconstruct the former glacial advances and deglaciation. We include geomorphic interpretations and recalibrated ¹⁰Be data from previous works in this valley (Jenny and Kammer, 1996; Grosjean et al., 1998; Zech et al., 2006). Finally, the results are discussed to provide an updated perspective of understanding the spatio-temporal variability of the former El Encierro Glacier during the LLGM and the Pleistocene–Holocene transition.

The geomorphological map allows the recognition of two moraine systems (ENCs) with a telescopic plan-view of arc morphologies. The two-terminal moraines of these systems are situated at 16 (ENC 1a) and 9 km (ENC 2a) downstream of the catchment head.

In total, three ages were obtained in moraine boulders in the outermost system (ENC 1) and two in the internal system of moraines (ENC 2). An additional ¹⁰Be sample was obtained from an ice-molded bedrock surface on the valley bottom in ENC 2 and also in a lateral moraine (ENC 1L) situated between ENC 1 and ENC 2. Table 1 summarizes the characteristics of the sample, the measured ¹⁰Be concentrations, the exposure ages, and the correlation with moraine stages defined by Zech et al. (2006) in view of their recalibrated and incorporated ages in this article. ¹⁰Be exposure ages range between 39.1 ± 1.7 ka and 17.2 ± 0.8 ka, without considering one outlier of 193.7 ± 7.3 ka which is not considered in the following sections of this report.

Some boulders sitting on top of moraines have a length of up to 3 m along their main axis. Some of these boulders are partially buried and emerge a few tens of centimeters above the surface, so the possibility of a prior complete burial condition of them cannot be excluded. The same post-glacial conditions of burying could be assigned to the ¹⁰Be-dated ice-molded bedrock surface. So, boulders and bedrock surface ages should be considered always as minimum ones.

Rock glaciers (RGs) cover large areas of the headwaters in the tributary valleys of ERV (Figure 2). Some of these RGs are still actively indicated with accumulations of boulders at their front. Although they are not the focus of this brief research report, RGs are mapped to highlight that the potential preservation of moraines of tributary glaciers has been compromised. The non-consolidated Neogene Gravels Units on the hillslopes of the ERV (12 Ma, Salazar and Coloma, 2016) are included (Figure 2) because its distribution was not available at the time of Zech et al. (2006) and are remains of the infilling history of the valley with fluvial processes.

Tills fill the valley along 16 km as low as 3,650 m a.s.l., while a preserved Neogene Gravel Units formed by non-consolidated polymictic alluvial deposits are identified downstream on the hillslopes of the ERV (Figure 2; Salazar and Coloma, 2016; Rossel et al., 2018). In the ERV, these Neogene Gravel Units hanging on valley slopes mapped in works from the Chilean Geological Survey (Salazar and Coloma, 2016) and in Miocene tectonostratigraphic Andean interpretations (Rossel et al., 2018) have been dated with interlocked tuff levels indicating a minimum age of 12 Ma. These recently published observations deny the interpretations that considered these deposits as till-forming moraines (MI) associated with a pre-LLGM glacial advance that reached 3,450 m a.s.l. and distant almost 20 km from the glacial cirque (Jenny and Kammer, 1996; Grosjean et al., 1998).

The presence of these Neogene gravels in the ERV identified by Salazar and Coloma (2016) suggests that the tills are in part composed of reworked Neogene gravels and so, the glacial forms could be in part the result of the redistribution of former non-consolidated alluvial deposits during glacial advances. The lateral moraines preserved on the hillslopes of ERV occur at a similar altitude as the non-consolidated Neogene gravels in relation to the valley floor and thus might have led to misinterpretations in the past. Thus, moraines distributions are indicators of the lateral limits of the glaciers that occupied the valley, but the supply of sediments was controlled, at least in part, by the Neogene gravels. This implies that particular attention should be paid when interpreting the spatial extent of glacial advances in arid valleys with well-preserved and thick ancient alluvial gravel successions.

In total, nine out of eleven ages of ENC 1 give a range between 40–25 ka (4 samples) and 18–22 (5 samples) (Figure 2, Figure 3), whereas one sample is younger (15 ka) and another is an outlier (193.7 ± 7.3). Thus, the oldest geomorphological evidence of glaciations in the ERV is the arc related to ENC 1, which represents the LLGM. The LLGM in the ERV was punctuated by several glacial advances (ENC 1a–d), which extended 16–14 km from the headwaters and down to 3,650–3,750 m a.s.l. In that sense, and based on the available data and supported by the preference of depositional ages assignment to the oldest-one-model (e.g., D'Arcy et al., 2019), we define that the external terminal moraine arc ENC 1a and the inner terminal moraine arc ENC 1d were deposited by full glacial conditions during ice expansions at ∼40 ka and 33–25 ka respectively. Glacial advances recorded in this valley have been reported in other nearby valleys of the Atacama Desert (e.g., Zech et al., 2011; Zech et al., 2017) and on the eastern slope of the Andes at similar latitudes (e.g., D'Arcy et al., 2019, 27°S).

Overall, five ages from moraine boulders on ENC 1 system are between 17–24: Four ages on a lateral moraine ENC 1L (EE33, EE34, EE42, and EE51) in the western slope of the ERV, and one on the inner terminal moraine arc ENC 1d (EE62). We interpret that these ages constrain the glacial recession or the paraglacial readjustment of moraines after the deglaciation (Figure 3). In fact, the tributaries situated in the west developed several rock glaciers that filled tributary valleys and reached the trunk valley after the deglaciation. The ridges on the eastern slope that avoid the influence of rock glaciers are dated at 33.4 ± 1.5 (HPL-26) and are much better preserved than the ones on the western side of the valley. The age of the lateral moraine agrees with the ages between 33–25 ka of inner-terminal moraine arc ENC 1d (Figure 3).

To recognize the age of the onset of deglaciation of glacial drift ENC 1 in the ERV, Zech et al. (2006) relied on exposure ages of lateral moraines that were developed as recessional positions of the glacier, recalibrated in this work to around 17–18 ka (EE22, EE24; Figure 2). This moraine ridge has been reinterpreted in this work as a lateral moraine related to the glacial advance of the tributary valley. The preservation of this ridge was only possible if the trunk valley was already deglaciated by that time. So, deglaciations of ENC 1 started at ∼24 ka and finished before 18–17 ka. Indeed, this time span of deglaciation is consistent with our ice-molded bedrock age of 18.8 ± 0.7 ka (sample HPL-31) and suggests that the valley was almost completely deglaciated at that time.

Nonetheless, the time span for the deglaciation of ENC 1 coincides with the ages of boulders on moraines ENC 2a (EE11, EE12) obtained by Zech et al. (2006) a few kilometers down-valley of the ice-molded bedrock samples (HPL-31; Figure 2). The ages of these two samples of moraine boulder is within 1 sigma and thus statistically the same (20–18 ka; Figure 3B). We interpret that the glacial advances of ENC 2a reached 9 km from the head of the valley, but the glacial erosion was not efficient to reset the nuclide concentrations of the ice-molded bedrock surface (sample HPL-31). In fact, ages of recessional moraine ridges ENC 2d (HPL07 and HPL33) suggest this glacial drift could be much thinner and retreat at 18–17 ka up the valley of the sampling site of HPL-31.

We suggest that valley deglaciation was completed by 18–17 ka. Thus, no major advances occurred in the ERV after 17 ka as proposed previously by Zech et al. (2017). Andean glaciations in the northern segment of the Atacama Desert were linked to CAPE humid events that occurred 16–10 ka (Quade et al., 2008). Any effect of this pluvial event on local glaciation was negligible in the valleys of the southern Atacama Desert, at least in the ERV (29°S), as already observed in the Cordon Doña Rosa (30°S; Zech et al., 2017). Deglaciation by 18 ka is a global warming signal at the start of the Heinrich Stadial Event 1 (HS1; 18–14.5 ka; Rasmussen et al., 2006), which is thought to play a direct role in the last glacial termination in the Andes through the high-latitude migration of the southern westerly winds and the CO₂ ventilation from the Southern Ocean (e.g., Denton et al., 2010).