Lack of evidence for Initial Upper Paleolithic attribution at Cueva Millán in Iberia’s hinterland

replying to: P. Sánchez-Yustos et al; Scientific Reports 10.1038/s41598-024-69913-3 (2024).

Introduction

Sánchez-Yustos et al.¹ investigated lithic assemblage from Cueva Millán and propose its classification as a southern European manifestation of the Initial Upper Paleolithic (IUP, ~ 50–40 kya cal BP). This paper critically evaluates that classification against the IUP, a technocomplex whose foundational phase is the Early Emiran in the Levant², defined by distinct blade technology with Levallois features and associated with Homo sapiens dispersal across Eurasia^3,4. The Cueva Millán assemblage lacks diagnostic IUP technological markers and exhibits predominantly Middle Paleolithic (MP) features, including Levallois-based core reduction and MP tool types. Such misclassification risks conflating MP and IUP techno-cultural units, thereby obscuring our understanding of the Neanderthal-to-H. sapiens transition.

The IUP was first defined by a bidirectional blade core technology and Levallois points² (Supplementary 1). In contrast, regional variants in the Balkans (e.g. Bacho Kiro Cave) and Central-Eastern Asia (e.g. Kara-Bom) lack true Levallois points and instead feature blades as final products^5,6. These lithic industries are sometimes associated with finely crafted bone tools and personal ornaments³ (Supplementary 2). Sánchez-Yustos et al.¹ propose that Layer L1 at Cueva Millán, containing both MP and Upper Paleolithic (UP) components represents Europe’s southernmost IUP manifestation, citing three indicators: (1) Prismatic core technology with lateral cresting, (2) Co-occurrence of Levallois-like points and blades in a single reduction sequence, and (3) Bladelet production associated with burin cores (Supplementary 3). Here, we show these features fail to meet established IUP criteria (3, 5, 7) and align more closely with MP technological traditions.

IUP core technology: blades vs. “Schéma Croisé” reduction

IUP core technology is characterized by volumetric- and bidirectional reduction (ex. Fig. 1b,c)³, producing blades and elongated Levallois points with parallel/subparallel edges (Fig. 2d,e)⁷. This is exemplified by Bohunician sites (e.g., Stránská skála, Czech Republic; Fig. 1b) and Temnata Cave, Bulgaria (Fig. 1c), where cores are systematically designed for elongated blade production. The Bohunician likely represents an early IUP variant in Central Europe, chronologically and conceptually close to the Emiran of the Levant^4,7.

Fig. 1.

Cueva Millán and Eurasian IUP core technology—(a) Cueva Millán- UP-like core, semi-rotated, (b) Bohunician reduction schema and its relationship to Levallois and blade technology ⁴, (c) Blade core from Temnata Cave, Layer 4⁹; Crested blanks: (d) Cueva Millán- crested blade with two prepared versants, (e) Cueva Millán—crested blade with one prepared versants, (f) Crested blades from Boker Tachtit-AH-B¹⁰, (g) Overshot bidirectional intercalated blade from Temnata Cave, Layer 4⁹.

Fig. 2.

Cueva Millán triangular convergent blanks and Eurasian IUP pointed blades and retouched tool: (a–c) Cueva Millán- pointed blanks (id. triangular flakes), (d,e) Retouched and unretouched Levallois points from Stránská skála, (f) Pointed bidirectional intercalated blade from Bacho Kiro Cave, (g–i) Cueva Millán- blades, (j) Bidirectional intercalated blade from Temnata Cave, Layer 4, (k) Endscraper on heavy overshot blade from Temnata Cave, Layer 4, (l) Bidirectional long and narrow blade from Bacho Kiro Cave, (m) Cueva Millán- retouched bladelets and fragments, (n) Bladelets from Kara-Bom, OH6.

Cueva Millán’s purported “blade” cores differ fundamentally. The reduction strategy employs a schéma croisé concept (Fig. 1a), producing triangular flakes (Fig. 2a–c) rather than blades, and lacks the parallel negatives indicative of systematic blade production (Fig. 1b). The cores are mostly flat (Fig. 1a) rather than prismatic, with scar patterns dominated by triangular negatives (Fig. 6: 1, 2, 4, 6 in ref.¹). Smaller-volume cores often exhibit hinged or plunging terminations (Fig. 6: 5, 8 in ref.¹). Blank production centers on short triangular flakes (Fig. 2a–c; Fig. 7 in ref.¹; Supplementary 4), aligning with MP flake-core technology⁸ not IUP blade production.

Levallois reduction and blade production in IUP vs. Cueva Millán

IUP blade production integrates Levallois technical attributes- facetted platforms and direct percussion with hard hammer- with volumetric prismatic cores (Fig. 2d–f) (Supplementary 5).

In contrast, the Cueva Millán Layer L1 assemblage demonstrates different reduction pattern: flake production through discoid method with limited presence of recurrent centripetal Levallois method. Quantitative analysis reveals 100 diagnostic Levallois-like and discoidal-like flakes alongside 49 technical flakes (Supplementary Table 11 in ref¹). The presence of genuine Levallois and discoidal cores, and MP technical flakes, indicates a clear MP technical system. Classifying these as 'Levallois-like’ or 'discoidal-like’ (Supplementary Tables 21, 22 in ref¹.) misrepresents their unequivocal MP identity.

IUP blade production standards absent at Cueva Millán

IUP blades, as seen in Bacho Kiro Cave, exhibit consistent morphological standards: straight profiles, triangular/trapezoidal/polyhedral cross-sections, with parallel edges (Fig. 2d–f,j–l), and standardized dimensions. Following established criteria, true blades must demonstrate ridges parallel or subparallel to edges and maintain a length–width ratio ≥ 2:1¹¹. These standards are evident in the volumetric core reduction at Temnata cave⁹.

Cueva Millán lacks these characteristics. The crested pieces exhibit irregular edge morphology and limited parallelism (Fig. 1d,e) (Fig. 5, Fig. 9 in Supplementary 6 in ref.¹), in contrast to typical IUP blade examples (Fig. 1d-l). These artifacts are elongated cortical flakes with divergent (Fig. 5–7 in ref.¹), asymmetric (Figs. 5–10 in ref.¹), and irregular edges (Fig. 1d,e). The majority of crested products in Cueva Millán were struck from small nodules or flake edges (Fig. 5—1–4, 8, 9,11 in ref.⁷), differing from true opening crests (Fig. 1f). As such, they fall short of the technical criteria for IUP blade production.

Blade metric analysis and blade-to-flake ratios

IUP assemblages exhibit typically high proportions of blades: 15.6% at Bacho Kiro Cave, 19.3% at Temnata Cave (Layer 4)⁹. In Bohunician assemblages, elongated Levallois points constitute 45% of formal tools, exceeding 60% at Ořechov IV⁷. These are more elongated than traditional MP points, with length-to-width ratio of 2.5:1 to as high as 3:1¹², comparable to the heavy-gauge blades from Bacho Kiro’s IUP (ratio ~ 3:1)⁵.

In contrast, Cueva Millán L1 layer yields only 2.37% laminar products and 3.26% bladelets, frequencies that fall directly within the range of MP variability¹³ (Supplementary 6). The technological features of the Arlanzian assemblage align with late MP assemblages across Southern Europe (Supplementary 7) reinforcing that these represent inherent MP technological variability rather than intrusive IUP techno-typologicalelements. Dominant tool types include triangular flakes, Levallois-like points (n = 34) representing merely 1.07% of the assemblage. Only 21 flakes meet blade elongation indices (EL: 0.25–0.49)¹¹ (Supplementary, Table 15 in Ref.¹). Blades are rare or entirely absent in the Cueva Millán L1 layer, failing essential IUP criteria³, (Supplementary 8).

Retouched tool typology and blank selection

In the IUP assemblages the retouched tools are manufactured on long blades with geometric shape (Fig. 2d,f) and comprise classic UP tool types^3,5. This is illustrated by the presence of UP tool types, such as endscrapers on Levallois points, in Bohunician assemblages^4,7.

At Cueva Millán, retouched blades (n = 12) are vastly outnumbered by retouched flakes (n = 61) and pointed flakes (n = 7) (Supplementary Table 8 in ref.¹), a pattern typical of MP assemblages (Supplementary 7). MP tool types further underscore this affinity: side-scrapers (n = 16), denticulates/notches (n = 11), and Mousterian points (n = 3) dominate (Supplementary Table 30 in ref.¹), while UP types (e.g., end-scrapers, backed tools) are scarce or atypical (Supplementary Fig. 23 in ref.¹). Notably, the few reported end-scrapers occur on cortical blanks with irregular, denticulated fronts, while other purported UP types (e.g., borers) are common in MP toolkits.

Broader implications of misclassification

While Sánchez-Yustos et al. provide valuable data, the techno-typological, morpho-metrical, and economic features (e.g., blank selection and tool production strategies) of the Cueva Millán L1 assemblage diverge significantly from IUP standards, considering both Early and Late Emiran (where the latter is the definitive IUP) assemblages. These technological discrepancies are further compounded by the lack of symbolic artifacts (despite preserved bones) and unresolved chronostratigraphic issues. Misattributing such assemblages to the IUP risks distorting our understanding of early human techno-cultural evolution by attributing local, late MP innovations to the suite of behaviors characteristic of H. sapiens dispersal. Accurate classification of these technologies is essential for reconstructing valid expansion patterns of the IUP and acknowledging the full spectrum of late Neanderthal technological adaptability. Misclassification risks attributing core elements of the UP transition to H. sapiens, thereby impoverishing models of technological convergence and population interactions in late Pleistocene Europe.

Conclusion

In light of the morphological, technological, and economic inconsistencies between Cueva Millán L1 lithic assemblage and established IUP criteria, its classification within this technocomplex is untenable. This inference is reinforced by broader comparative analysis with Late MP assemblages across the Mediterranean, where similar laminar features emerged independently within Neanderthal contexts^14,15. Rigorous adherence to technological definitions, rather than superficial parallels, is therefore paramount for reconstructing accurate patterns of human expansion and innovation during this critical period. Misattribution ultimately distorts narratives of Homo sapiens dispersal and obscurs the scope of late Neanderthal technological adaptability.

Author contributions

T.T. designed and wrote the reply, and constructed Figs. 1 and 2. P.S. edited and completed the text and data, and contributed images to Figs. 1 and 2. Y.D. completed, edited and validated the relevant data.

Declarations

Competing interests

The authors declare no competing interests.