Optically Stimulated Luminescence (OSL) Dating of Alluvial Deposits from the Cahuachi Archaeological Site (South Peru)

Abstract

Cahuachi (Nazca River Valley, South Peru) was the major ceremonial center of the Nasca civilization. According to previous studies, it was struck and destroyed by three El Niño-Southern Oscillation events, which would have occurred around 100 BCE, 600 CE and 1000 CE, respectively. At the end of the series of events, the ceremonial center would have been buried by a cap of conglomerates. Despite this hypothesis raised well-founded doubts regarding its geochronological and paleoenvironmental implications, it is uncritically used as a reference in geoarchaeological research. In the present study, optically stimulated luminescence (OSL) dating results of some samples taken from alluvial deposits at Cahuachi are reported, with the aim to evaluate the literature’s hypothesis. Since the obtained ages are older than the Holocene epoch, such a hypothesis must be rejected. A number of field evidences corroborate this result. Finally, the advancement in fluvial geomorphology knowledge of the Nazca River Valley is briefly discussed.

1. Introduction

Cahuachi (Nazca River Valley, Pampa of Nazca, South Peru) was the major ceremonial center of the Nasca Culture. Such a civilization, which arose around 200 Before Common Era (BCE), spread its dominance over southern Peru under the influence of a theocratic elite. Cahuachi became a place of pilgrimage, preserving this function at least until 450 Common Era (CE) [1,2]. This monumental site is spread over 150 ha and consists of a number of smoothed mounds forming the core of pyramids and temples made of adobe [3].

As a result of a geoarchaeological investigation, Grodzicki (1990, 1992, 1994) [4,5,6] claimed that the Pampa of Nazca, including Cahuachi area, was struck by three catastrophic El Niño-Southern Oscillation (ENSO) events, which occurred around 100 BCE, 600 CE and 1000 CE, respectively. According to the author, the second event caused severe damage to buildings and the destruction of the Nasca Culture; furthermore, the third event would have caused the burial of the site under a cap of conglomerates [6]. Grodzicki’s reconstruction raised doubts regarding its geochronological and paleoenvironmental implications (see, e.g., refs. [7,8]). On the other hand, possible evidence of hydrogeological disasters ascribable to ENSO events are reported, with different confidence levels, for many Pre-Hispanic Peruvian sites [9]. Some of these evidence are quite controversial and caused contrasting opinions. Thus, exploring the reliability of the literature hypotheses is a decisive issue in both the geoarchaeological and paleoclimatic perspectives. As highlighted by previous authors “more geological work is necessary in Nazca and full publication data is crucial” [7].

Some years ago, an archeological excavation inside Cahuachi allowed us to study the lithostratigraphy of the bedrock underlying pyramids and temples (Delle Rose et al., 2019 [10]). The main finding was to refer the examined layers to the top of the Tertiary regional succession rather than Holocene ENSO-related deposits. A research target planned by the authors of ref. [10] was the dating of alluvial sediments investigated by Grodzicki by means of Optically Stimulated Luminescence (OSL) methods. In the present study, OSL dating results of some samples taken at Cahuachi in April 2019 are reported (Section 3) with the aim to evaluate the catastrophe hypothesis of Grodzicki (Section 4.1). As this dating method is generally addressed to quartz and feldspars of homogeneous grainsize (Section 2.4.1), a thin section mineralogical study and a grainsize analysis (Section 2.3) have been previously carried out for testing the suitability of the sediment samples to be dated through OSL. Finally, the advancement in fluvial geomorphology knowledge of the study area is briefly discussed (Section 4.2).

2. Materials, Methods and Sample Descriptions

2.1. Study Area

The study area is placed at the northern margin of the Atacama desert [3]. From a morpho-structural point of view, it belongs to the Cuenca Pisco Unit (a sub-domain of the Andean forearc) and shows the typical flat landscape of the pampas [11]. Early lithostratigraphic studies on the Tertiary regional succession were published during the 80s [12,13]. They defined two major lithostratigraphic units at the top of the succession, named Changuillo Fm (upper Pliocene) and Canete Fm (upper Pliocene-lower Pleistocene), respectively, without providing suitable information regarding criteria to define their lithostratigraphic boundary.

Since the end of 90s, a stratigraphic revision of the Tertiary regional succession has been underway by the Ministerio de Energía y Minas [11]. Thus, it has been established that the top of the succession is made up by interstratified sandstones, siltstones, mudstones, volcaniclastic layers, conglomerates and breccias, and shows a marine to continental transition from the bottom to the top (the depositional environment changes from shallow marine water to shoreface to alluvial fan) [10]. The stratigraphic revision also provided first information on the boundary between Changuillo and Canete Fms [11,14,15], which was almost completely unknown before Grodzicki’s studies (see Appendix A).

Throughout the Early–Middle Pleistocene, south–west Peru emerged from the sea to be shaped by geomorphological processes under different climate conditions. The uplift of the Andes led to abundant supplies of clasts that formed large alluvial fans, while the river basins evolved through cycles of erosion and sedimentation. During the late Middle Pleistocene, a series of planation surfaces and fluvial terraces developed [16]. For the Andean forearc of the southern Peru, a regional-scale fluvial response to climate and tectonic forcings has been recently highlighted [17]. Since the beginning of the Upper Pleistocene, moisture transport changes led to repeated transition from grassland to desert, and vice versa [18]. The last glacial cycle was characterized by phases with enhanced precipitation [19,20] which influenced, in turn, the fluvial dynamics before the deglaciation [17].

The modern ENSO atmospheric conditions were established at the beginning of the late Holocene. They determined precipitation decreasing and desert expansion [21,22]. Over the last few millennia, the final shaping of the smoothed mounds of the Cahuachi area was characterized by periods of drought interspersed with rainfall events. Such conditions allowed the last phase of formation of a widespread desert pavement, lag deposits, eolian and colluvium layers along the sides of the mounds, and debris flows inside small ephemeral watercourses [3]. These deposits completely cover the bedrock and have thicknesses ranging from a few decimeters to a few meters.

Figure 1 displays bedrock formations and major fluvial bodies, including active alluvial-slope systems, of Cahuachi and surrounding areas. The valley floor deposits of the Nazca River are distinguished in floodplain deposits and sediments of the present riverbed.

2.2. Sediment Sampling

During April 2019, a geological survey was carried out at Cahuachi and the surrounding areas with the main goal to collect samples of sandy sediments suitable for OSL analyses. In the meantime, field observations on Tertiary regional succession, Quaternary alluvial deposits, and bedrock covering were performed (see Appendix A).

To summarily characterize the main sedimentological features of the sampled alluvial deposits, the facies classification codes of Miall (1978, 2006) have been used [24,25]. These codes are composed by two or three letters. A first capital letter indicates dominant grain size (G = gravel; S = sand), while subsequent lowercase letters indicate texture and sedimentary structures (c = clast-supported; h = horizontal bedding, imbrication; i = inverse grading; l = low-angle cross-bedding; m = massive).

For OSL dating, samples were collected by hammering, horizontally, 22.5 cm-long and 4.5 cm-diameter aluminum-based alloy cylindrical tubes into the sand levels to prevent exposure of the sediments to the sunlight. Only the central parts of the sandy samples contained in the tubes were then used for OSL dating (Section 2.4.1), thus minimizing the possibility to analyse the last centimetres of sands of both the ends of the tubes, which could have had the chance to be exposed to sunlight during the sampling before closing them with thick caps not filtering the sunlights. The same sandy levels for OSL analyses were also sampled in separate plastic bags to perform grain-size and minero-petrographic analyses by thin sections (Section 2.3).

Regarding the selected sample for radiocarbon dating (Section 2.4.2) by means of an accelerator mass spectrometer (AMS) charcoals were wrapped in aluminum foil (to avoid contact with organic material during transport).

Five sampling points were selected for the aim of this study (Figure 1). They are described below from the lowest to the highest in elevation to form a composite profile (

Table 1. Sampling profile and information on the sampling points.

Sampling Point	Samples	Latitude *	Longitude *	Altitude **
1	NZ16	14°48′41.58″ S	75°08′43.66″ W	335
2	NZ7-8	14°48′38.45″ S	75°07′49.86″ W	345
3	NZ11	14°48′40.26″ S	75°07′46.96″ W	355
4	NZ1a	14°49′07.94″ S	75°07′00.38″ W	390
5	NZ3	14°49′10.53″ S	75°06′57.05″ W	390

* values extracted from Google Earth Pro; ** approximated values extracted from Google Earth Pro [m a.s.l.].

and Figure A3 in Appendix A). These samplings were carried out without affecting or involving structures and other archaeological remains.

2.2.1. Sampling Point 1

The sampling point 1 is placed inside Estaquería (Figure 1), the archaeological site that was the only one to remain active after the decline of Cahuachi [26]. On the wall of an illegal excavation (Figure 2; for the looting problem see Lasaponara and Masini, 2016 [27]), the top of the floodplain deposits is exposed (estimated elevation above the valley floor about 15 m). From the bottom to the top of this stratigraphic exposure, massive sand (Sm facies) and clast-supported bedded gravel (Gh facies) occur in succession. The NZ16 sample (Table 1) was extracted from the Sm facies.

2.2.2. Sampling Point 2

The sampling point 2 coincides with a terraced deposit remnant (Figure 3). Its estimated elevation above the valley floor is about 7.5 m. This deposit is made up by laminated sand (Sh facies) containing scattered pebbles. Two sandy samples (NZ7 and NZ8, see Table 1) were gathered in vertical profile from this sedimentary body.

2.2.3. Sampling Point 3

The section of the sampling point 3 coincides with the upper part of the composite section reconstructed in Figure 8 of Grodzicki (1990) [4]. According to this author, the top of this exposure is made up by the sediments produced during the 600 CE ENSO event. It is of particular importance for this study. The estimated elevation above the valley floor is about 17.5 m. From the bottom to the top, clast-supported gravel with inverse grading (Gci facies), cross-bedded sand (Sl facies), and clast-supported bedded imbricated gravel (Gh facies) occur in succession. The upper conglomerate layer determines the flat shape of the top of the mound. Next to the slope edge, above the alluvial deposit, the foundation of a weathered adobe wall is placed (Figure 4). The NZ11 sample (Table 1) was extracted from the Sl facies.

2.2.4. Sampling Point 4

The exposure of the sampling point 4 coincides with the one described in Figure 5 of Grodzicki (1992) [5]. Its importance is crucial for this study because the author states that the deposition of the uppermost layer was caused by the 1000 CE ENSO event. The estimated elevation above the valley floor is about 40 m. From the bottom to the top of this section, faint laminated sand (Sm facies) and a tabular bed (about 1.2 m thick) of structureless poorly sorted clast-supported gravel (Gcm facies) occur in succession. The conglomerate apparently shaped the flat top surface of Gran Piramide temple. Its depositional characteristics are similar to the ones of analogous bodies observed inside the floodplain deposits (see Appendix A, Figure A4a). Above the tabular conglomerate, the foundation of a weathered adobe wall is placed (Figure 5); a small part of this wall (at the left in the figure) has slipped on the ground surface. The NZ1a sample (Table 1) was extracted from the Sm facies.

It must be noted that the alluvial bodies of the points 2, 3, and 4 are too small to be drawn in Figure 1 and, however, their mapping is beyond the scope of this work (see Section 1 and Section 4.2).

2.2.5. Sampling Point 5

At the sampling point 5 (Figure 6), a finer-grained colluvium level containing charcoal fragments has been sampled (NZ3 sample, Table 1). The level is part of a layered colluvium and is covered by the youngest, coarse-grained colluvium at the top. In addition to the NZ3 sample for OSL dating, several charcoal fragments were gathered for radiocarbon dating. The sampled level has no visible relation to the adobe walls of Figure 6 (that belong to the “Piramide Sur” temple [28]), and was taken for comparison with the alluvial samples.

Summarizing the sampling collection for OSL dating, the samples NZ1a and NZ11 come from alluvial deposits ascribed to ENSO events by Grodzicki, while the samples NZ3, NZ7-8, and NZ16 were taken for comparison. For each sampling points, latitude, longitude, and an approximated value of altitude extracted from Google Earth are reported in Table 1.

2.3. Grain-Size and Mineralogical Features

The grain-size of the sediment samples for OSL dating was determined through dry sieving at 2–1–0.5–0.250–0.125 and 0.625 mm. Organic fractions were eliminated as much as possible, the remaining amount of the organic components were however considered as a sum with pelite-loss. Thin section mineralogy and texture were determined though a polarized light optical microscope (one polarizer only and crossed nicols). They are medium- to fine-grained sands whose grainsize variations are shown below.

NZ1a, NZ11 and NZ16 samples are mostly unimodal medium-grained sands with the grains interval component <2 mm–125 μm comprised between 84.8 and 95.3 wt.%. The modes of these samples are represented by <500–250 μm (NZ1a: 35.7 wt.%; NZ11: 47.9 wt.%) or <1 mm–500 μm (NZ16: 44.9 wt.%). Thin section observation of these three samples highlighted quartz, plagioclase and k-feldspar crystals in decreasing order of abundance. Mafic minerals (pyroxene, amphibole and micas) are rare. Among the lithic component both intrusive (mostly granitoid rocks often with subsolidus exsolution textures such as perthite/antiperthite/mirmechyte) and extrusives (mostly rhyolites/dacites lavas and pyroclastics and subordinate andesites) are well represented (Figure 7) with both micro-cryptocrystalline to glassy groundmass. Due to devitrification processes, spherulitic texture are common among the volcanic lithotype grains. Metamorphic lithologies mainly consist of grains of quartzites and feldspathic quartzites (often with triple junction equilibrium texture among the quartz grains) and subordinate schists. Arenites and siltites represent the main sedimentary lithotypes. Most of the crystals and lithics of these three samples are sub-angular to slightly rounded thus emphasizing a low to medium degree of textural maturity.

Mineralogy of the sample NZ3 is very similar to that of the three samples described above, both concerning crystals and lithics (Figure 8) as well as their roundness. However, this sample is characterized by a finer sandy grainsize. The wt.% fractions are uniformly distributed among the sandy components (comprised between 5.5 wt.% for >1–2 mm and 14.1 wt.% for >62.5–125 μm), with no distinct modes. The most abundant fraction is represented by the organic and pelite-loss (27.9 wt.%) and the component > 2 mm is 11.2 wt.%.

NZ7 and NZ8 sediments are unimodal fine-grained sands, with the grains interval component <500–62.5 μm comprised between 76.9 and 82.2 wt.% (Figure 9). The fractions <62.5 microns are comprised between 15.9 wt.% (NZ7) and 21.2 (NZ8). Also these two finer grained sediments show the same crystal phases and lithic lithologies distribution already described for NZ1a-NZ11-NZ16 and the same degree of textural maturity, however with a clear enrichment in the mafic components, mainly pyroxenes and amphiboles (cf. Figure 7).

2.4. Analysis Methods Description

2.4.1. OSL Dating

OSL can estimate the time elapsed since buried sediment was last exposed to daylight [29,30]. It is based on the measurement of the charges trapped in mineral grains like quartz and feldspars since the time of the sediment deposition, as a consequence of the irradiation due to the natural radioactivity field. The burial age is estimated by dividing the equivalent dose (${\mathrm{D}}_e$, a measure of the radiation energy absorbed by grains during their period of burial) by the environmental dose rate (the rate of supply of ionizing radiation to the grains over the same period). One of the most widely adopted procedures for measuring ${\mathrm{D}}_e$ values from quartz OSL signals is the single-aliquot regenerative dose (SAR) procedure [31,32,33].

The alluvial sediment samples were dated with OSL at the Lambda laboratory (Laboratory of Milano Bicocca University for Dating and Archaeology, Department of Materials Science) [34]. In laboratory, sediments from the middle of the sampling tubes were used for OSL measurements to avoid the light-exposed ends that instead were used for the determination of dose rates. Sample preparation and OSL measurements were carried out under subdued red-light conditions. Quartz grains of 180–250 μm were obtained from each sediment sample using routine preparation procedures, including treatment with 37% HCl and ${\mathrm{H}}_{\mathrm{2}}$${\mathrm{O}}_{\mathrm{2}}$ solutions overnight, followed by dry sieving and etching in 40% HF to remove any feldspar contamination and the alpha-irradiated outer layer of each grain [30]. The etched grains were rinsed in HCl acid to remove precipitated fluorides, sieved again, and then mounted on stainless-steel discs using silicon oil. The purity of the quartz samples was checked by stimulating the aliquots with IR light after laboratory dosing (OSL IR depletion ratio test; see ref. [35]). The ${\mathrm{D}}_e$ were calculated using the arithmetic mean for the samples showing a Gaussian distribution of the ${\mathrm{D}}_e$ estimated on the individual aliquots analyzed, indicating complete bleaching of the sample before burial.

If the bleaching is incomplete, the age will be overestimated, and the dose distribution will be different from a normal distribution. Small aliquots were prepared because it has been demonstrated that in poorly bleached samples, a small aliquot performs as well as single grains for obtaining a model age [36,37], allowing the use of Minimum Age Model (MAM) or Finite Mixture Model (FMM) on the data [31].

The minimum age limit for luminescence dating is influenced by the OSL sensitivity and the detection limit of the OSL signal in relation to the background. The upper age limit is controlled by saturation of the luminescence signal, with the saturation level differing between samples.

OSL measurements and ${\mathrm{D}}_e$ determinations were made using the SAR procedure and were performed with a Risø TL/OSL DA20 reader (Risø DTU, Roskilde, Denmark). Quartz luminescence were stimulated by blue LEDS (470 ± 30 nm, 54 mW/cm²) and the emitted photons were detected with an EMI 9235QB photomultiplier (ET Enterprises, Ltd, Uxbridge, UK) coupled with a UV filter (7.5 mm Hoya U-340; UQG Optics Ltd, Cambridge, UK). Samples were stimulated for 40 s at 125 °C and the pre-heat value was experimentally derived on the basis of the results of a dose recovery pre-heat plateau test (Figure A6). The first 0.6 s of OSL decay was used for ${\mathrm{D}}_e$ determination and the mean count recorded over the last 10 s was subtracted as background. Irradiations were made using ⁹⁰Sr/⁹⁰Y beta sources (5.50 GBq, dose rate: 0.11 Gy/s for 180–250 μm grains).

In order to calculate the annual dose, the Th and U concentrations of the sediments were measured with total alpha counting using Zn scintillator discs [38], assuming a concentration ratio Th/U equal to 3 ⁴⁰K content was deduced from the total concentration of K measured with flame photometry.

Attenuation of the beta dose was taken into account [39]. The investigated samples have a relatively low (≤11%) gravel (>2 mm) content and medium sand-silty component (<1 mm) always ≥80%, with extremely abundant (between 36 and 98%) grains ≤ 500 µm. In this grain size range, the variation in the beta dose does not change significantly [40]. The beta and gamma dose rates were calculated using the conversion factors from Guerin et al. [41].The contribution of the cosmic rays to the dose rates were taking into account according to Prescott and Hutton [42]. In calculating the dose-rates, it was assumed that the average saturation water content over the entire burial period was about 30% due to the relatively dry condition of the site, similarly to the H₂O % content established for similar sandy sediments of droughty areas (see, e.g., ref. [43]).

2.4.2. Radiocarbon Dating

A charcoal sample (Laboratory Code LTL21547) taken at the sampling point 5 (Section 2.2), was submitted to radiocarbon dating by AMS (Accelerator Mass Spectrometry) at the Centre of Applied Physics, Dating and Diagnostics (CEDAD)-Department of Mathematics and Physics, University of Salento, Lecce, Italy [44]. The sample underwent a chemical processing consisting of alternate Acid-Alkali-Acid (AAA) attacks in order to remove any source of contamination. Purified sample material was then sealed in evacuated quartz tube together with copper oxide and silver wool and combusted to carbon dioxide at 900 °C for 8 h.

Extracted carbon dioxide was then cryogenically purified and catalytically reduced to graphite by using H₂ as reducing agent and iron powder as catalyst [45]. The obtained graphite (∼1 mg) was then pressed into an aluminum sample holder of the AMS system where the ¹⁴C/¹²C isotopic ratio was measured. The conventional radiocarbon age was then calculated from the ¹⁴C decay law after correction for isotopic fractionation and machine and processing blank [46]. The measured radiocarbon age was finally calibrated to calendar one by using the last internationally accepted calibration curve (INTCAL20) and the OxCAl ver. 4.4 Calibration software.

3. Results

The OSL dating results are reported in

Table 2. Saturation water content (W), U, Th and K₂O contents, dose rates, Overdispersion (OD), ${\mathrm{D}}_e$ and ages obtained on Cahuachi samples.

Sample	W	ppm U (±5%)	ppm Th (±5%)	K2O (±3%)	Dose Rate (mGy/a)	OD (%)	${\mathbf{D}}_{\mathit{e}}$ (Gy)	Age
NZ1a	0.22	2.32	7.33	1.73	2.65 ± 0.13	16 ± 5	67 ± 4	>25,280 ± 1970
NZ3	0.21	2.96	9.36	1.89	3.08 ± 0.15	33 ± 5	7.0 ± 0.7	2270 ± 250
NZ7	0.20	2.31	7.31	1.7	2.64 ± 0.13		nd
NZ8	0.20	2.99	9.45	1.75	2.98 ± 0.15		nd
NZ11 1	0.30	3.05	9.63	1.51	2.70 ± 0.13	35 ± 4	66 ± 3	24,440 ± 1650
							143 ± 15	52,960 ± 6140
NZ16	0.23	3.21	10.13	2.04	3.29 ± 0.16	37 ± 7	63 ± 9	19,150 ± 3100

¹ ${\mathrm{D}}_e$ calculated with FMM with two components. The lower component as a relative proportion of 81% and the upper 19%.

. In Figure 10 some examples of OSL decay and dose response curved were reported. In Figure 11 the ${\mathrm{D}}_e$ distributions obtained on the sediments are shown.

The ${\mathrm{D}}_e$ of NZ1a sample was near to saturation as shown in Figure 10b (70% of the aliquots were at saturation); therefore the age obtained on this sample can only be considered a minimum age. The natural OSL signals of the NZ7 and NZ8 samples, however, are very low, at the limit of the minimum detection limit of the instrument (see Figure 10e). For this reason it was not possible to determine their ${\mathrm{D}}_e$ and therefore their ages. ${\mathrm{D}}_e$ distribution of sample NZ11 instead shows a double distribution (Figure 11c). Most likely its luminescence was not completely erased before deposition or older sediments were incorporated by post-depositional mixing. In this case the ${\mathrm{D}}_e$ was calculated using the FMM statistical models using a two components fitting. Two possible dates were calculated.

Despite the above problems, the estimated dates for the sampled alluvial deposits (column Age in Table 2) appear to be substantially consistent with the sampling profile (Table 1). The age of NZ16 sample (taken at the top of the floodplain deposits; see sampling point 1) is around 19 ky and predates those estimated for both the NZ11 and NZ1a samples. NZ11 sample (taken 17.5 m above the valley floor; see sampling point 3) gives two possible age of 23–26 and 47–59 ky, respectively. The dating of NZ1a sample (taken about 40 m above the valley floor; see sampling point 4) results in a minimum age of 23–27 ky.

The OSL age obtained on NZ3 sample (natural colluvium; sampling point 5) is around 2.3 ky (Table 2). This dating matches the calibrated radiocarbon age of 1960 ± 93 BP (before the present: 1950 CE; conventional radiocarbon age: 2018 ± 35 years BP; see Figure 12) of the examined charcoal sample (Section 2.4.2), which was determined with a confidence level of two standard deviation.

4. Discussion

Luminescence dating in Peru faces methodological challenges, particularly with the OSL signal from quartz. Studies show that this signal can be unreliable for coastal fluvial sediments due to a thermally unstable medium component of the OSL signal [47] that caused the underestimation of the OSL ages. This fact leads the researchers to prefer post Infrared-Infrared Stimulated Luminescence (pIR-IRSL) dating of K-feldspar [48,49,50]. While some studies report reliable quartz OSL ages in the Andes and Amazon [51,52], inconsistencies exist, prompting the use of K-feldspar in problematic cases. Furthermore, challenges in effectively bleaching sediments complicate K-feldspar dating due to daylight exposure issues and an unbleachable component.

In our case study, to check the stability of the medium OSL quartz component, the ${\mathrm{D}}_e$(t) plots of the samples [53,54] were obtained. While for ideal sample ${\mathrm{D}}_e$ should not depend on signal integration interval, a falling ${\mathrm{D}}_e$ value towards the later intervals may indicate an unstable medium component [55]. Samples NZ1a, NZ11 and NZ16 show a trend of the ${\mathrm{D}}_e$ that does not vary with the increase of the integration interval used for the ${\mathrm{D}}_e$ calculation (see Figure A7a in Appendix B), indicating that the medium OSL component is thermally stable. Their OSL ages are indeed reliable. Only sample NZ3 shows a decreasing ${\mathrm{D}}_e$(t) (Figure A7b) but the date obtained with OSL is in agreement with that obtained with radiocarbon (Section 3). This excludes that the presence of an unstable medium OSL component could lead to an underestimation of the age for this sample, as reported in some cases in literature [47,50].

4.1. Evaluation of the Catastrophe Hypothesis of Grodzicki

Grodzicki [4,5,6] reconstructs three catastrophic river floods and heavy rains around 100 BCE, 600 CE, and 1000 CE, respectively, which the author claims to be the result of as many ENSO events. These events, would cause the deposition of widespread thick conglomerates made up by debris flows and alluvial deposits. The first conglomerate would be the base of the Holocene succession (cf. Figure A1d and Figure A4d) and would not have affected archaeological structures ([6], pp. 95–96). Differently from the former, Grodzicki states the second event severely damaged the adobe constructions of Cahuachi, abruptly changed the surrounding landscape features, and finally caused the abandonment of the ceremonial center and the end of the Nasca theocracy ([6], pp. 96–97). But that time Cahuachi, as noted by Silverman and Proulx [7] “had ceased to function as the great early Nasca ceremonial center”. Again, considering “the good state of preservation of geoglyphs, even where they cross valleys or erosion rills”, also Eitel et al., 2005 [8] “disagree with ideas that catastrophic El Niño events destroyed the Nasca Civilization”.

The disagreement of the above authors appears even more justified considering the landscape impact which would have caused the third hypothetical catastrophe. According to Grodzicki, around 1000 CE “the largest cataclysm that occurred on the Nazca plateau at the end of the Holocene” ([6], p. 98) would have produced such a quantity of sediments as to “cover the entire area of the ceremonial center [...] burying the monuments of the same, of which some ruins have remained. This effect is seen, in particular, in the Gran Piramide, which dominates the river valley” [5], p. 122. However, in coastal Peru adobe buildings deteriorate within a few decades because of the scarce rain and moisture, and their conservation is a crucial issue in pre-Columbian sites for cultural tourism [56]. If the adobe buildings of Cahuachi had been hit by such a catastrophe, they would have been lost. Differently, they are usually preserved under thin layers of colluvium [3].

A crucial point of the catastrophe hypothesis of Grodzicki, is a supposed younger age of the geoglyphs in comparison with the one of the Cahuachi temples [5]. However, several studies have shown that the Nazca Lines were constructed up to 650 CE, that was near the end of the Nasca Civilization ([57,58,59,60]). Although Grodzicki’s reconstruction is based on this biased chronology and considered inconsistent by the above-mentioned authors [7,8], it is uncritically used as a reference in geoarchaeological research (cf. e.g. [61,62,63,64,65]).

A further issue regarding the reliability of Grodzicki’s hypothesis arose during the geological survey for sediment sampling (Section 2.2). At both the sampling points 3 and 4, it was evident that the adobe walls were built on the investigated alluvial deposits (Figure 4 and Figure 5). To confirm this gathered relationship between deposits and walls, targeted geoarchaeological excavations are suggested. However, since the age obtained by OSL on the samples NZ11 (sampling point 3) and NZ1a (sampling point 4) are older than the Holocene epoch (Table 2), the present study adds new evidence against Grodzicki’s reconstruction. In fact, these samples were taken from the two alluvial deposits ascribed by this author to ENSO events occurred around 600 and 1000 CE, respectively [4,5,6]. Even considering possible inaccuracies in dating, the obtained result implies that the discussed hypothesis must be rejected.

4.2. Advancement in Fluvial Geomorphology Knowledge

The failure of the Grodzicki’s reconstruction has been mainly caused by the lack in knowledge about the geomorphology and geochronology of the alluvial deposits present in the study area. To the best of our knowledge, this is the first study that reports chronological and morpho-sedimentological data on the late Quaternary alluvial deposits of Nazca River Basin. In the study area, we identified three alluvial deposits arranged at different heights above the valley floor (Section 2.2, Table 1). Each of these could belong to a different fluvial terrace, the oldest at the top and the youngest at the bottom. However, since this study has another purpose (see Section 1 and Section 4.1), these findings are incomplete and must be carefully verified.

The literature on fluvial geomorphology of the south Peru has expanded considerably in recent years (see e.g. [16,17,47,66,67]). Considering these works, the herein studied alluvial deposits can be interpreted as fluvial terraces resulting from the interplay between tectonic uplift and climate changes. This working hypothesis needs to be tested by specific research to achieve an advancement in fluvial geomorphology knowledge. First, detailed mapping is mandatory to define the complete fluvial terrace sequence of the Nazca River Basin. As an example, to establish the average height of the fluvial terraces above the current river course, detailed field topographic survey and high resolution DEM processing have to be performed (cf., e.g., refs. [16,17]).

Establishing the formation age of fluvial terraces is an intriguing challenge in geomorphology. For reliable dating of terrace formation, several terrace profiles must be methodically sampled and analyzed [68]. Frequently, a single age is assigned to a terrace (e.g. the one of the sediments atop the terrace strath), but this could be not reflect the actual geomorphological process [69,70]. To obtain more accurate results, series of samples should be collected from different fluvial facies belonging to a single terrace [68]. Finally, to establish confident timing of the late Quaternary fluvial evolution in Peru, the pIR-IRSL method is recommended [17,48].

5. Conclusions

Possible evidence of meteo-hydrogeological disasters ascribed to ENSO events are reported for many Pre-Hispanic Peruvian sites. A catastrophe hypothesis regarding the ceremonial center of Cahuachi had already raised criticism from some researchers and made further geological work in Nazca River Basin and the publication of geochronological data necessary. With this study we have addressed such a requirement (Section 1 and Section 2).

OSL dating does not reconcile with the ENSO-related catastrophic events emphasized by previous studies. The conglomeratic bodies in Cahuachi ascribed by Grodzicki to the first millennium CE have a late Pleistocene age (Section 3 and Section 4.1). They belong to a fluvial terrace system identified and briefly described in this work (Section 2.2 and Section 4.2). Other conglomeratic bodies found in the surrounding of Cahuachi belong to the upper part of the Tertiary regional succession (Appendix A). Nevertheless, specific studies are needed to deepen our knowledge of the fluvial geomorphology of the Nazca River Basin.

To date, the main evidence of damage due to meteo-hydrogeological phenomena at Cahuachi is that caused by heavy rains towards the end of the 4th century CE. They partially destroyed the roof and walls of the “Piramide Sur” [28].