© K. Douka,  T. Higham, E. Mastora

Reaffirming the integrity of Grotta del Cavallo and the association of modern humans with the Uluzzian: a reply to Zilhão et al. PLoS ONE 2015


We saw today the publication of a paper in PLoS ONE titled "Analysis of Site Formation and Assemblage Integrity Does Not Support Attribution of the Uluzzian to Modern Humans at Grotta del Cavallo", and authored by Zilhão and colleagues. Our reply has been already posted as a comment, and we re-blog it here. 


This is a long paper that is littered with ‘if’s ‘tentative’s, ‘possible’s and ‘maybe’s and it is a testament to the sheer doggedness of the authors that they play this out over so many pages. Facts are few on the ground. They attempt at every turn to shroud and to confuse with respect to the original excavations and reports. The aim is to create a large element of doubt in the reader’s mind regarding just about everything to do with Cavallo Cave (Figure 1). Prior to the publication of Benazzi et al. [1] these authors were virtually silent on these issues. The reason they attack the integrity of the site is because the new data does not support their own view of the Middle to Upper Palaeolithic sequence of Europe.


Zilhão’s and colleagues’ model of recent human evolution involves Neanderthals completely independently reaching "behavioural modernity" at the same time as modern humans were entering Europe. This has been coined the "impossible coincidence" [2] and has found little support amongst the archaeological community over the past 20 years. This model requires than no Upper Palaeolithic industry or modern human fossils existed in Europe prior to 41.5 ka cal BP (an arbitrary limit these authors set). The findings reported from Cavallo [1] which showed that the teeth discovered at the lowermost Uluzzian layers E (EIII and EII-I) belonged to modern humans, with shell beads pre-dating the 41.5 ka cal BP "barrier", have shaken this idea.


When the Cavallo teeth were thought to be Neanderthal, their association with implements and ornaments normally found with anatomically modern human populations was not questioned. Now that this attribution has been overturned, Zilhao et al. attack virtually all aspects of the site’s excavation and post-excavation analysis. Similar criticisms were attempted for other sites that do not fit their own model: Willendorf, Geissenklosterle, Kebara Cave, Fumane Cave, Szeleta Cave and so on.


By attempting to draw doubt on Cavallo and its stratigraphic sequence, Zilhao et al. avoid altogether having to deal with the question of the taxonomic affiliation of the teeth, now calling this merely a “moot point”.


Their approach in this paper is; 1) to question the competence of the excavators; 2) to develop the idea that there is a fully-fledged Proto-/Aurignacian assemblage (layer D) at the site; 3) to demonstrate the likelihood that post-depositional movement derived from prehistoric digging, looters pits and the like, caused significant mixing in the Uluzzian deposit (layer E), and 4) thereby question the provenance of any of the teeth excavated from the site and raise doubts over their original location.


Figure 1: Cavallo Cave, S. Italy.  Location and stratigraphic sequence. Note the difference in colour between the Uluzzian and much later Upper Palaeolithic layers. 



Limited studied material


The bulk of the material from the site is in Siena, where it has curated and under study. With the exception of a joint study on the bone implements and a much-debated typological study of the lithics by P. Gioia in the 1980s (see further comments below), the Cavallo Uluzzian material, which comprises around 2000 lithic tools, several shell ornaments, faunal remains, as well as the excavator's personal field notes, was not studied by these authors.


Instead, in their present article, these authors focus on the Cavallo collection housed in Taranto, where remnant material from the site was left by Palma di Cesnola in the late 1960s. This assemblage is limited and has had an unknown post-excavation history. Mixed bags, misplaced and lost material, non-correspondence between artefacts, labels and inventory,  multiple packaging and re-packaging, were noted by us in 2009 when we visited the Taranto depot (hence our very limited sampling from that depot), a fact also noted by Zilhão et al. in their current article. This assemblage simply cannot be taken at face value in reconstructing the site's inventory.

Even more surprising is the fact that using this selectively picked and mixed limited assemblage, Zilhão et al. go on to infer that the site's excavator indulged in "post-hoc reassignment of finds to a “layer” different from that to which they had been referred at the time of excavation". There is no such evidence, this is an unfounded and rather distasteful allegation.



Stratigraphic selectivity and the invention of an Aurignacian layer


The discussion of the site's stratigraphy is a selective translation of odd sentences picked from Palma di Cesnola's numerous publications, omissions of important -but inconvenient facts- and false interpretations that Zilhão et al. come up with to justify their idea that Cavallo must be significantly mixed. As an example, in their figure 7 they write: "This reconstruction is provided as a graphical aid to the in-text discussion and should not be taken to represent an accurate rendering of the actual layout of site and trenches". In other words their figure is not realistic nor accurate.


Zilhão et al. claim that "no geological study of the sequence exists, and illustrations of its stratigraphy have been restricted to schematic renderings". This is standard practice for most sites, even current, where a schematic rendering of the stratigraphy is published along with a detailed description of the geological layers (see Figure 1 above). This information was published by Palma di Cesnola since the first reports on the site [3-4]. In addition, the stratigraphy identified in 1963 was fully confirmed in 1966 and by Gambassini’s rescue excavations in the late 1970s and 1980s.


What is absurd in the reconstruction displayed in Zilhão et al.'s Fig. 7 is that the pit -recognized by one of us (Paolo Gambassini) as an erosive event- has been unrealistically represented by an increasing in diameter over time (from 1964 to 1966) following the widening of Palma di Cesnola’s trench. It is well-known that many Palaeolithic sites are affected by pits and biological or natural disturbances, and this is no reason to dismiss so widely the evidence from a site like Cavallo. The “so-called” pit was identified from the beginning of the excavations (1963) at Grotta del Cavallo (field season 1963). Moreover, that year Palma di Cesnola went through the Uluzzian layers only over a very limited surface (1 square meter) [4]. The hypothesis of Zilhao et al. that, in 1964, the edges of the pit had not been "accurately delimited" is a conjecture without any documentary evidence to support it.


The majority of the article is concerned with the nature of layer D (the uppermost Uluzzian layer at the site) which according to Zilhão et al. is a truly Aurignacian assemblage. There is no strong evidence for an Early Aurignacian presence in southern Italy, south of Gargano (~250km north of Cavallo), an important but rather inconvenient detail Zilhão et al. fail to mention in their article. These authors even suggest, without any actual evidence, that some of the diagnostic Uluzzian pieces found in D are intrusive from below.

However, while making every effort to "establish" that layer D yielded primarily an intact Aurignacian assemblage, they also insist that the site was so badly disturbed by the aforementioned pit so extensive that it affected the entire deposit from Epigravettian layer B to late Mousterian layer F. Somehow layer D, sandwiched between B and F, remained intact and is of Aurignacian affinities. In their own words, this is a geological impossibility.


The presence of elements attributed to “Aurignacian I” at Cavallo layer D was initially brought forward by Gioia [5,6]. Based on a selective Bordesian approach, each retouched artifact from layer D was assigned to either an Uluzzian or Aurignacian I component purely on typological grounds. Yet, specific artefacts were not included in the Aurignacian assemblage (e.g. bladelets) and no hard data (refits which may indicate mixing across layers, differential patterns of surface alteration) were presented. Gioia’s original suggestion that there is Aurignacian presence in the Bay of Uluzzo is linked to her speculation that a similar assemblage exists at the site of Fontana Nuova di Ragusa in Sicily [7] - this is now widely believed to be of Epigravettian age.


As it happens, layer D of Cavallo and its wordy description is irrelevant to the discussion of the Uluzzian human teeth, which were recovered from the lowermost Uluzzian layers E [8]. Indeed, there is very little discussion regarding these early layers.

Which leads us to another observation.

Zilhão et al. completely fail to mention the presence of a series of intact hearths discovered in layer E, which were identified and described by Palma di Cesnola, were illustrated in the published sections and more importantly, provided charcoal for the first radiocarbon date for the site in 1969 (RM-352: >31000) [9]. This sample comprised a piece of charcoal recovered from Layers E II-I excavated in 1966. The hearths were discernible in section and in his description of the teeth provenance, P. di Cesnola mentions that the oldest tooth comes from below the lowermost hearth of layer E-III, at the bottom of the first Uluzzian layer at the site. This key detail is again conveniently left out by Zilhão et al. Had the pit disturbance been so extensive, how is it possible that the teeth were retrieved from below one of the several stratified hearths that remained intact?



Uluzzian techno-typological variability


Completed and ongoing studies, which include a doctoral dissertation [10-12], focus on the lithic technology of the Uluzzian layers and their results suggest strongly that this is a coherent assemblage. Instead, Zilhão et al. favour a typological, Bordesian system applied to part of the material by P. Gioia back in 1985-86 [5-6]. In addition, they seem happy to base their own interpretations on 2-dimensional illustrations of artefacts. It is surprising how confident they are in attributing these illustrations to one or another Palaeolithic industry especially since, in our previous reply to their criticism [13], we showed that such inferences are erroneous.


Irrespective of the biological data, the notion that the Uluzzian was made by AMHs is supported by a variety of evidence involving both technological behaviour and subsistence practices (lithic production, carcass processing, bone tools, the use of pigments and ornaments). Compared with the lithic technology of the final Mousterian, the transformation (retouching) phase takes precedence whilst blank production appears to be less elaborated [10-12]. Mousterian fine (thin) elongated cutting edges (often identified as multifunctional tools) are not the primary objective. The Uluzzian lithic assemblage shares the Upper Palaeolithic “dynamic” technological character with rapid innovations and a variety of new production features when compared with the preceding period.  Bipolar technique on anvil (in the form of both cores and splintered pieces), characterized by opportunistic production schemes, is widespread. In the Cavallo series blade/bladelet débitage gradually increases from EIII to D. In terms of dimensions, bladelets are often close to small blades and were frequently used as blanks for the lunates, a completely new tool linked to the productions of composite implements. With the exception of very few cases reminiscent of “Dufour” elements, the so-called bladelets cited by Palma di Cesnola and exacerbated by Zilhao et al. are of a generic kind. Given their common use as blanks for the most typical Uluzzian tool, the lunate, there is no reason to invoke a possible Aurignacian mixing to explain their occurrence in the Uluzzian sequence.



Radiocarbon dates, shells beads, and provenance


It is important to remember that prior to the Benazzi et al. [1] paper Zilhão et al. have previously used the Cavallo shell beads as evidence for Neanderthal engagement with personal ornaments [14] and to suggest that the Neandertals of Europe were already using Dentalium shell beads (the ones found in Cavallo), thereby showing that it was incoming AMH who were copying them -rather than the reverse [15]. Clearly, Zilhão was not particularly concerned with the site's integrity at the time, after all, the co-presence of "Neanderthal" teeth and shell beads at Cavallo served to strengthen and support their own ideas. For example, Zilhao writes [15] the Uluzzian shell beads reflect to the kinds of personal ornaments in use among the Neandertal societies that immigrating modern humans encountered in Europe.


In the present article, Zilhão et al. inform their readers of "the possibility that at least some, if not all of the six “1978–84” shells [we previously dated] are in fact from 1964". We are not sure what is not clear to these authors especially after we have stated in [1] that the shells we radiocarbon dated come from the new excavations and not from the 1963-66 digs. Maybe we ought to prove that we can read labels (which accompany the ornaments of the Siena collection).


Zilhão et al. state that the shell beads from the Uluzzian layers are similar in terms of species and form to the ones used in the later Aurignacian of Mediterranean Europe, implying that they must be from this period. Similar species are also used, however, in the Aterian of northern Africa, the Initial Upper Palaeolithic of the Levant, the Neolithic and Bronze Age of Europe and the MSA of South Africa. Their argument that similar species were used in the Aurignacian that therefore that the layer must be Aurignacian is simply wishful thinking.


Zilhão et al. also claim that assigning layer D to the Uluzzian is in evident contradiction to chronostratigraphic patterns. They state that; "the radiocarbon evidence therefore corroborates the presence in the Cavallo assemblages of Protoaurignacian- or Aurignacian-aged material". It seems to us to be dangerous to link a cultural horizon to an industry simply on the basis of its age. This approach would ignore variability in the dates that are evident from varied industries across Europe during this period when we see multiple lithic industries, and also radiocarbon dates of variable quality.

In an equally mystifying paragraph Zilhão et al. go on to suggest that the collection of the dated shells from Cavallo occurred during the “Aurignacian I and/or the Aurignacian II”, as if these lithic technocomplexes (a pure archaeological construction to facilitate study and comparison) can be used as an absolute measure of time. Even more alarming is their suggestion that the radiocarbon ages simply corroborate their speculation for the presence of Protoaurignacian and Aurignacian technology and typology of the stone tools assigned to layer D, obviously a circular argument. We repeat here, that there is still no evidence to support the presence of Early Aurignacian south of Gargano (~250km from the Uluzzo Bay).

It is finally worthwhile reminding the reader that, so far, no shell ornaments have ever been found associated with Neanderthals or present in Mousterian contexts anywhere in Eurasia. Zilhão et al. talk a lot about parsimony in their paper but fail to invoke it with reference to the presence of shell beads in the lowest levels of Cavallo and the likelihood that AMHs produced these artefacts.



Human teeth


In their current paper, Zilhão et al. do not dispute the taxonomical reassessment of the two deciduous teeth from the Uluzzian levels of Grotta del Cavallo [1], suggesting instead that their status is now a “moot point”. This is despite having previously claimed that; “the taxonomic affiliation of the recovered human teeth [is] problematic” [16] and that “their morphology is insufficient to establish that they are indeed from modern humans instead of Neandertals” [17]. They do not add anything more concrete to their speculative criticisms in this or of Zilhão’s other recent work [18]. Interestingly, they also fail to mention recent work that has further confirmed the attribution of Cavallo B and C to modern humans [19-20].


Still, an example of the fog that they seek to create regarding the site is the inclusion in the discussion of a tooth from the site of Grotta del Poggio. Zilhão et al. refer to this tooth in the text and show it in their Figure 11, despite the fact that it is not from Cavallo at all. Why? They argue that, in the 1964 report of Palma di Cesnola [3: 27] a mention of human teeth finds includes a reference to a molar having a Carabelli’s cusp. They argue that the only such tooth known from the south of Italy is one from Poggio, which was analysed and published at the same time as the Cavallo teeth [8]. This is then used to raise doubt regarding the exact provenance of the teeth from Cavallo based on accusations of misidentification, mixups, uncertainties and ambiguity.


While Palma di Cesnola, who is not an anthropologist, might have easily confused the tuberculum molare with a Carabelli’s cusp, the personal excavation notes from Grotta del Poggio (Figure 2), attest to the fact that Grotta del Poggio was excavated in April/May 1966 and March 1967 [Camerota is the district in the “Cilento” region where Grotta del Poggio is located (province of Salerno, Campania)], and not in 1964, when the Uluzzian teeth from Grotta del Cavallo were found.



Figure 2: From Palma di Cesnola's personal excavation notebook on Grotta del Poggio: the cover (left) and the text described below (right).


The tooth from Grotta del Poggio was unearthed on May the 7th 1966. A translation of the discovery follows: “In the external part of the trench (sub-layer 6/1) it has been unearthed (by Miss Tavanti) a tooth that looks human (?), but with unusual size (Neandertal?). The crown is rather worn. The roots are powerful, quadruple, merged till almost the apices”.

There is no tooth from Grotta del Cavallo that fits this morphological description. All human remains from Grotta del Cavallo are deciduous teeth with resorbed roots (even the deciduous tooth from the Mousterian level). This description corresponds perfectly with the morphology of the molar from Grotta del Poggio. Zilhão et al. demonstrably fail to check the facts here, preferring instead the simple art of speculation and of raising doubt where none exists.



Final remarks


Initially, according to Zilhão and his colleagues, the Cavallo radiocarbon dates were wrong and variable [16, 18] and the attribution of the teeth to humans was erroneous and problematic [16-18]. The teeth's actual identification, modern humans or Neanderthals -previously a major aspect of these authors’ argument against the Cavallo case- has now become "a moot point". But now that the teeth are human, the site is so badly mixed that the teeth must be intrusive.  Or even worse, at least one of the Cavallo tooth may have been found at a completely different site (Grotta del Poggio). We are also told here that the radiocarbon dates may be correct, but despite the accompanying labels, the shells we dated must come for a different collection, therefore our dates do not correspond to what we say they are.

The impossible nature of all these arguments, how easy they change from time to time and how each and all of them were "right" at one point or another give us a glimpse to Zilhão et al.'s modus operandi. Their case against the integrity of Cavallo is a multi-headed beast: we prove one speculation is wrong and they come up with another, equally unfounded and even more speculative. We have attempted here to set the record straight.


Once more, we are struck by the wording that Zilhão et al. use throughout their paper and take it as evidence for the strength of the data that they muster. They "infer", "assume", "conceive", "find it more likely" or "quite possible", there are "indications", and arguments to be accepted "to further strengthen an inference". They cite "personal communication" with one of the authors over a single artefact 30 years ago and 20 years after the actual excavation took place. This paper is wafer thin on fact and data and concomitantly high on speculation and inference.


If we are to further our understanding of the Middle to Upper Palaeolithic transition in the Italian peninsula, our energy should be directed on new excavations and sound interpretations based on new data from post-excavation research, rather than this type of agenda-driven approach. Indeed, current excavations in the bay of Uluzzo are already being undertaken by teams of colleagues, and we hope that with the recovery of fresh material new light will be shed over the Uluzzian of southern Europe.



References cited


1. Benazzi S, Douka K, Fornai C, Bauer CC, Kullmer O, Svoboda J, et al. Early

dispersal of modern humans in Europe and implications for Neanderthal behaviour.

Nature. 2011;479: 525–529. doi:10.1038/nature10617.


2. Mellars P. The impossible coincidence: a single species model for the origins of modern human behavior in Europe. Evol. Anthropol. 2005;14: 12–27.


3. Palma di Cesnola A. Seconda campagna di scavi nella Grotta del Cavallo presso

Santa Caterina (Lecce). Riv Sci Preistoriche. 1964;19: 23–39.


4. Palma di Cesnola A. Prima campagna di scavi nella Grotta del Cavallo presso Santa Caterina (Lecce). Riv Sci Preistoriche. 1963;18: 41–74.


5. Gioia P. Problems related to the origins of Italian Upper Palaeolithic: Uluzzian and

Aurignacian. In: Kozlowski JK, editor. L’Homme de Néandertal 8: La Mutation.

Liege; 1988. pp. 71–101.


6. Gioia P. An aspect of the transition between Middle and Upper Paleolithic in Italy:

the Uluzzian. In: Farizy C, editor. Paléolithique moyen récent et Paléolithique

supérieur ancien en Europe Ruptures et transitions: examen critique des documents

archéologiques. Nemours: APRAIF; 1990. pp. 241–250.


7. Gioia P. L’industria litica di Fontana Nuova nel Quadro dell’Aurignaziano Italiano. Origini. 1984-1987;13: 27-58.

8. Palma di Cesnola A, Messeri P. Quatre dents humaines paléolithiques trouvées dans des cavernes de l’Italie Méridionale. L’Anthropologie. 1967;71: 249–262.


9. Palma di Cesnola A. Datazione dell'Uluzziano col Metodo del C–14. Riv Sci Preistoriche. 1969;24: 341–348.


10.  De Stefani M, Dini M, Klempererova H, Peresani M, Ranaldo F, Ronchitelli A, Ziggiotti S. Continuity and replacement in flake production across the Middle-Upper Palaeolithic transition: A view over the Italian Peninsula. In: Pastoors A & Peresani M, editors. Flakes not Blades: The role of flake production at the onset of

the Upper Palaeolithic in Europe. Mettmann: Wissenschaftliche Schriftendes Neanderthal Museums; 2012. pp. 135–151.


11. Klempererová H. Functional analysis applied to the lithic assemblage of the final Middle Palaeolithic and beginning of the Upper Palaeolithic with the aim to reconstruct behavior of ancient human groups. Unpublished PHD Thesis, Scuola di Dottorato in Scienze della terra – Preistoria, XXIV ciclo, Università degli Studi di Siena; 2012.


12. Ranaldo F, Kemplererova H, Ronchitelli A. Evoluzione dei sistemi tecnici nell’Uluzziano di Grotta del Cavallo (Nardò-LE). Atti della XLVII Riunione Scientifica IIPP “Preistoria e Protostoria della Puglia”, Ostuni 9 -13 ottobre 2012. In press.


13. Ronchitelli A, Benazzi S, Boscato P, Douka K, Moroni A. Comments on "Human climate interaction during the Early Upper Paleolithic: Testing the hypothesis of an adaptive shift between the Proto-Aurignacian and the Early Aurignacian” by William E. Banks, Francesco d’Errico, João Zilhão. J Hum Evol. 2014;73: 107–111. doi:10.1016/j.jhevol.2013.12.010.


14. Zilhão J. The Emergence of Ornaments and Art: An Archaeological Perspective on

the Origins of “Behavioral Modernity.” J Archaeol Res. 2007;15: 1–54.



15. Zilhão J. Neandertals and Moderns Mixed, and It Matters. Evol. Anthropol. 2006;15: 183-195.


16. Banks WE, d'Errico F. and Zilhão J. Human-climate interaction during the Early Upper Palaeolithic: testing the hypothesis of an adaptive shift between the Proto-Aurignacian and the Early Aurignacian. J Hum Evol. 2013;64: 39-55.


17. Trinkaus E, Zilhão J. Paleoanthropological Implications of the Peştera cu Oase and Its Contents. In: Trinkaus E, Constantin S, Zilhão J, editors. Life and Death at the Peştera cu Oase: A Setting for Modern Human Emergence in Europe. Oxford: Oxford University Press; 2013. pp. 389-400. 


18. Zilhão J. Neandertal-Modern Human Contact in Western Eurasia: Issues of Dating, Taxonomy, and Cultural Associations. In: Akazawa T, Nishiaki Y, Aoki K, editors. Dynamics of Learning in Neanderthals and Modern Humans Volume 1: Cultural Perspectives. Replacement of Neanderthals by Modern Humans Series. Tokyo: Springer; 2013. pp 21-57.


19. Bailey SE, Benazzi S, Souday C, Astorino C, Paul K, Hublin J-J. Taxonomic differences in deciduous upper second molar crown outlines of Homo sapiens, Homo neanderthalensis, and Homo erectus. J Hum Evol 2014;72: 1–9.


20. Fornai C, Benazzi S, Svoboda J, Pap I, Harvati K, Weber GW. Enamel thickness variation of deciduous first and second upper molars in modern humans and Neanderthals. J Hum Evol. 2014;76: 83–91.



Please reload

Recent Posts








Please reload