New Insights into Mousterian Faunal Assemblages from Uluzzo C (Apulia, Southern Italy)

Abstract

Grotta-Riparo di Uluzzo C (Uluzzo Bay, Apulia, southern Italy) preserves a long and complex stratigraphic sequence spanning from the Middle Palaeolithic to the Bronze Age, offering a valuable context for investigating depositional dynamics and human–environment interactions during the Late Pleistocene. Although recent multidisciplinary research has substantially advanced knowledge of the Uluzzian occupations, the Mousterian faunal record of the site has remained largely unexplored from zooarchaeological and taphonomic perspectives. This study examines the faunal assemblages from the Mousterian layers (E, F, and G), integrating material from historical excavations with those recovered during recent fieldwork. Zooarchaeological, taphonomic, and Zooarchaeology by Mass Spectrometry (ZooMS) analyses are combined to reconstruct local environmental conditions, evaluate the relative contribution of human and non-human agents to bone accumulation, and assess patterns of site use and deposit formation. The faunal spectrum indicates an ecologically heterogeneous landscape, consistent with a Mediterranean refugial setting during the Late Pleistocene. Taphonomic evidence points to complex and cumulative formation processes resulting from repeated, short-term human occupations interspersed with carnivore activity and natural depositional processes. The Mousterian deposits are therefore best interpreted as brief palimpsests rather than the result of continuous or intensive occupation. Placed within a regional framework, the Uluzzo C assemblages contribute to broader discussions on site formation processes and environmental variability in southern Italy and provide an important comparative baseline for the Middle to Upper Palaeolithic period.

1. Introduction

Faunal assemblages recovered from caves and rock shelters represent one of the primary datasets for investigating human–environment interactions and site formation processes during the Late Pleistocene. However, their interpretative potential critically depends on the recognition and disentanglement of multiple accumulating agents and post-depositional processes that often operate within the same stratigraphic contexts. In caves and rock shelters, Mousterian occupations commonly form stratigraphic successions resulting from recurrent and often alternating use by human groups and carnivores over more or less extended periods of time, frequently characterised by short-term and discontinuous occupations ([1,2,3,4,5,6], among others). In these contexts, the cumulative nature of the archaeological record intersects with issues of temporal resolution and cultural attribution, as well as with questions related to the Neanderthal–Sapiens relationships and variability in occupation intensity [7,8].

Within this interpretative framework, reconstructing human subsistence strategies is essential. Archaeological consensus has long held that Neanderthals relied predominantly on ungulates and other large game across much of Eurasia ([9,10,11,12,13,14,15,16,17,18,19,20,21,22,23], among others). However, an expanding body of evidence indicates that Late Pleistocene human groups also exploited plant resources and a wide range of small and medium-sized prey, including tortoises, birds, small game and marine resources, including molluscs [1,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40], challenging long-standing assumptions about regional dietary variability and suggesting that Neanderthal diets were broader than previously recognised. At the scale of the Mediterranean Basin, recent case studies further clarify this variability, documenting small-game use and freshwater/coastal fishing at several Late Pleistocene sites [39,41,42,43,44].

Southern Italy, and Apulia in particular, offers an exceptional setting for investigating Neanderthal lifestyles and adaptations to climate and environment. The abundance of archaeological sites with stratified Mousterian levels indicates a continuous Neanderthal presence in the region during the Late Pleistocene, spanning from MIS 5 to mid MIS 3 [45,46,47], until their abandonment of those territories just before or during the arrival of modern humans [48,49,50]. Apulia may have served as a refugium when thermophilous and temperate species migrated southwards [46,48], thereby ensuring the persistence, conservation and survival of biodiversity, including plant species, animal species and human populations [49,50]. Numerous Mousterian sites in Apulia document repeated human occupation within ecologically heterogeneous landscapes [51,52], yet zooarchaeological studies highlight substantial variability in the relative contribution of anthropogenic and non-anthropogenic agents to faunal assemblages across sites and stratigraphic units [46,53,54,55].

Within this regional framework, Grotta-Riparo di Uluzzo C (hereinafter Uluzzo C) represents a key site for investigating Middle Palaeolithic dynamics within a long and well-stratified sequence spanning from the Mousterian to the Bronze Age. Excavations carried out in the 1960s by Borzatti [56,57,58] established the stratigraphic framework and documented the importance of the Mousterian deposits (layers E, F, and G), although the associated faunal assemblages were never subjected to systematic zooarchaeological and taphonomic analysis. For decades, the presumed loss of the assemblage prevented any reassessment using modern analytical approaches. The recent relocation of the faunal collection from the 1960s excavations now makes it possible to address this gap and to apply integrated zooarchaeological, taphonomic, and biomolecular analyses, including Zooarchaeology by Mass Spectrometry (ZooMS), allowing for the taxonomic identifications where morphological criteria alone are insufficient. Since 2016, renewed multidisciplinary research at Uluzzo C has significantly refined the stratigraphic, chronological, and geoarchaeological framework of the site through new excavations, absolute dating, sedimentological and micromorphological analyses, as well as detailed studies of the lithic industries, faunal remains and pollen [59,60,61,62,63,64,65].

This paper presents the first comprehensive zooarchaeological, taphonomic, and ZooMS-based study of the faunal assemblages from the Mousterian layers (E, F, and G) at Uluzzo C, bridging materials from both historical and recent excavations. Our study aims to reconstruct local palaeoenvironmental conditions, identify the agents responsible for bone accumulation and modification, and assess patterns of hominin site occupation. The results are discussed within a regional and diachronic framework, enabling direct comparison with the overlying Uluzzian phases at Uluzzo C and with other Mousterian assemblages across southern Italy, and contributing to a broader understanding of Neanderthal behavioural and ecological variability in the region.

2. The Archaeological Setting

Uluzzo C is located in Nardò, province of Lecce in southern Italy, along the western Apulian coastline overlooking the Ionian Sea (40°9′27.84″ N, 17°57′35.34″ E). It is situated within the Parco Naturale di Porto Selvaggio, in the centre of Uluzzo Bay, with the nearby Grotta del Cavallo (Uluzzo A) and Grotta di Uluzzo B located less than 300 m away (Figure 1A). The current configuration of Uluzzo C comprises a narrow central hall alongside an external sedimentary deposit that rests against the rock face and slopes downward toward the sea. These elements constitute the preserved portions of what was originally a more spacious cave and shelter system that experienced vault failure [56,57,58].

The site was first identified in 1960 when Carlo Cosma and the Gruppo Speleologico Salentino conducted a survey of caves in Uluzzo Bay. During these explorations, lithic artefacts, abundant faunal remains, and human teeth, recently morphometrically attributed to Neanderthals [61], were recovered [66]. Systematic stratigraphic excavations were conducted at Uluzzo C between 1963 and 1968 under the direction of Edoardo Borzatti von Löwerstern, carried out by the Italian Institute of Prehistory and Protohistory (IIPP) of Florence in collaboration with the Provincial Administration of Lecce and the “S. Castromediano” Archaeological Museum [56,57,58]. During the initial phase of fieldwork (1963–1966), a 2.6 × 1.5 m trench was opened in the central portion of the deposit, extending to a depth of 2.5 m (Sector B, Figure 1B) and revealing a continental sequence spanning from the Mousterian to the Bronze Age. This trench was later enlarged to a depth of about 6 m, until a marine deposit was identified by Borzatti von Löwerstern as a Tyrrhenian beach. The uppermost units, associated with Bronze Age occupation, had been heavily disturbed by looting, resulting in the loss of most deposits [65]. Two further campaigns were undertaken in 1967 and 1968, extending the trench one meter seaward. This area yielded markedly fewer lithic and faunal materials, despite sedimentological continuity with the interior deposits [56,57,58]. The stratified deposit excavated in the 1960s comprised a sequence approximately 6.40 m thick, divided into ten layers (Figure 2, Table 1). Several units were culturally attributed to Romanellian (layer A) and Uluzzian occupations (layers C and D) as well as an extensive Mousterian sequence (layers E, F, G and H). Intervening sterile sediments (layers B, I, and L) and thin volcanic ash lenses designated by Greek letters (β and γ) completed the sequence [56,57,58].

The new excavations, started in 2016, defined three sectors: A, B, and C (Figure 1B). Sectors A and B lie within the rock shelter, corresponding, respectively, to the present-day surface of the deposit and to the base of Borzatti’s original trench. Sector C, a test trench on the terrace outside the shelter, yielded only reworked sediments, capped by cemented breccia [59]. The first phase of renewed fieldwork (2016–2019) focused on the Uluzzian deposits in Sector A, corresponding to SUs 3, 15 and 17 (Borzatti’s layer C) and SUs 21, 22, 23 and 25 (Borzatti’s layer D). These layers were divided into three distinct phases [62,63,64,65]. The second phase (2022–ongoing) targeted the Middle Palaeolithic sequence, including SUs 23 and 24 (Borzatti’s layer E) and SU 30 (Borzatti’s layer F). The Middle Palaeolithic sequence examined in this study exhibits considerable variation in sedimentary properties and archaeological content, while the analysis of the macrofaunal assemblages focuses on layers E, F, and G from Borzatti’s excavation, together with SUs 23, 24, and 30 (Table 1).

Samples for sedimentological analysis and OSL dating (Table 1) were collected from the entire sequence accessible in sector A (Borzatti’s layers B, C–L) [65]. OSL ages were obtained for the Uluzzian units C (42.7 ± 2.6 ka) and D (38.1 ± 2.2 ka) and for the Mousterian units F (41.6 ± 2.7 ka) [65]. The date of 46 ± 4 ka in the uppermost part of G is, for now, considered too recent if compared with the characteristics of the lithic industry and requires further validation [60]. The lower Mousterian levels (spits 21–6 of unit G) are not supported by direct chronologic data and their age is inferred through correlation with Grotta del Cavallo, where U/Th dating of the base of layer M (overlying deposits comparable to layers I and L at Uluzzo C) yielded ages of c. 116 ka [67], indicating a post-Tyrrhenian chronology for the entire Mousterian sequence at both sites ([61], but for a different view see [68]). A key chronostratigraphic marker is the recent identification at Uluzzo C of the PGT/Y-6 tephra layer (SU 27), situated between units D and E. This tephra derives from the Pantelleria eruption, dated to ~45.0–44.9 ka [60]. The same tephra at Grotta del Cavallo marks the end of the Mousterian [69], providing a terminus ante quem for both the sequences. Collectively, these data constrain the Mousterian occupation at Uluzzo C between the post-Tyrrhenian phase (MIS 5) and c. 45 ka (mid MIS 3).

Figure 1. (A) Location of Uluzzo C in the Bay of Uluzzo (Apulia, southern Italy) from Google Earth, with the indication of the nearby Grotta del Cavallo and Grotta di Uluzzo B; (B) Map of the excavation areas ([68], modified).

Figure 1. (A) Location of Uluzzo C in the Bay of Uluzzo (Apulia, southern Italy) from Google Earth, with the indication of the nearby Grotta del Cavallo and Grotta di Uluzzo B; (B) Map of the excavation areas ([68], modified).

Figure 2. Stratigraphic sequence of Uluzzo C: Sector B (Borzatti’s trench) ([59,60], modified) and Sector A ([63], modified).

Figure 2. Stratigraphic sequence of Uluzzo C: Sector B (Borzatti’s trench) ([59,60], modified) and Sector A ([63], modified).

Table 1. Schematic archaeological sequence ([59,61,63,68,70,71,72], modified).

Borzatti’s Layers	New Excavation (SUs)	Depth (cm)	Volcanic Layer	Sediment	OSL Dates (From Sector A)	Cultural Attribution
A		0–17	α	Reworked topsoil with blue-coloured volcanic pumiceous sand lenticels (α).		Romanellian
B		17–62	β	Volcanic pumiceous sand mixed with red detrital quartz sand. At the base a thin layer of silvery-grey pumiceous sand (β).	ULOC 2= 22± 2 ka	Sterile
C	3, 15, 17	62–80		Strongly cemented red sediment with debris.	ULOC 3= 42.7 ± 2.6 ka	Uluzzian
D	21, 22, 26 cs (combustion structure)	80–90	γ (25, 26, 27)	Red sandy sediment with fine debris. At the base lenses of greenish pumice sand (γ = SUs 25 and 27). At the top and within the body of the structure (26cs), a lens of pumice and sand mixed with stones fine sediment, yellowish to olive-green in colour and bright in the light, was identified and recorded as SU26 in recent excavations, 2015–2022. SU 27 at the bottom of the deposit is a Tephra layer (PGT/Y-6) originated during the Pantelleria eruption, dated to ca. 45.0-44.9 ka.	ULOC 4= 38.1 ± 2.2 ka	Uluzzian
E	23, 24	90–100		Cemented red sandy-silty sediment with few debris. Above a thin violaceous clay layer (δ).		Mousterian
F	30	100–150		Strongly cemented red sediment with debris capped by a thin sandy-clayey yellowish layer (ε). Between G and F there is a 15-20 cm gap (ζ) formed by a hollow zone.	ULOC 5= 41.6±2.7 ka	Mousterian
G		165–250		Dark brown loose sandy-clayey sediment with few debris sealed by a horizon rich in large and small stones, slightly cemented. Unit G was divided in 22 spits (each spit is 10-15 cm thick) by Borzatti. Between spits 4 and 5 there is a layer (η) made of often broken and corroded stones and pebbles generally covered with thick patinas of manganese and iron oxides.	ULOC 1= 46 ± 4.0 ka	Mousterian
H				Strongly cemented red ochraceous clayey sandy sediment, which lies on a dark, loose, fine sand layer. In lateral contact with spit 21 of layer G, was interpreted as a possible remnant of an older, partially eroded deposit.		Mousterian
I				A series of large calcareous rock blocks up to 1 m long exhibiting signs of bioerosion (lithodomes).
L				Conglomerate of much smaller, more or less rounded blocks, contained in a coarse white predominantly calcareous sand tenaciously cemented and travertinised, with limestone fragments and fossils of marine shells. The marine event represented by layers I and L was attributed by Borzatti to MIS5e, providing a terminus post quem for human occupation, as observed at other Mousterian sites in the region.

Table 1. Schematic archaeological sequence ([59,61,63,68,70,71,72], modified).

Borzatti’s Layers	New Excavation (SUs)	Depth (cm)	Volcanic Layer	Sediment	OSL Dates (From Sector A)	Cultural Attribution
A		0–17	α	Reworked topsoil with blue-coloured volcanic pumiceous sand lenticels (α).		Romanellian
B		17–62	β	Volcanic pumiceous sand mixed with red detrital quartz sand. At the base a thin layer of silvery-grey pumiceous sand (β).	ULOC 2= 22± 2 ka	Sterile
C	3, 15, 17	62–80		Strongly cemented red sediment with debris.	ULOC 3= 42.7 ± 2.6 ka	Uluzzian
D	21, 22, 26 cs (combustion structure)	80–90	γ (25, 26, 27)	Red sandy sediment with fine debris. At the base lenses of greenish pumice sand (γ = SUs 25 and 27). At the top and within the body of the structure (26cs), a lens of pumice and sand mixed with stones fine sediment, yellowish to olive-green in colour and bright in the light, was identified and recorded as SU26 in recent excavations, 2015–2022. SU 27 at the bottom of the deposit is a Tephra layer (PGT/Y-6) originated during the Pantelleria eruption, dated to ca. 45.0-44.9 ka.	ULOC 4= 38.1 ± 2.2 ka	Uluzzian
E	23, 24	90–100		Cemented red sandy-silty sediment with few debris. Above a thin violaceous clay layer (δ).		Mousterian
F	30	100–150		Strongly cemented red sediment with debris capped by a thin sandy-clayey yellowish layer (ε). Between G and F there is a 15-20 cm gap (ζ) formed by a hollow zone.	ULOC 5= 41.6±2.7 ka	Mousterian
G		165–250		Dark brown loose sandy-clayey sediment with few debris sealed by a horizon rich in large and small stones, slightly cemented. Unit G was divided in 22 spits (each spit is 10-15 cm thick) by Borzatti. Between spits 4 and 5 there is a layer (η) made of often broken and corroded stones and pebbles generally covered with thick patinas of manganese and iron oxides.	ULOC 1= 46 ± 4.0 ka	Mousterian
H				Strongly cemented red ochraceous clayey sandy sediment, which lies on a dark, loose, fine sand layer. In lateral contact with spit 21 of layer G, was interpreted as a possible remnant of an older, partially eroded deposit.		Mousterian
I				A series of large calcareous rock blocks up to 1 m long exhibiting signs of bioerosion (lithodomes).
L				Conglomerate of much smaller, more or less rounded blocks, contained in a coarse white predominantly calcareous sand tenaciously cemented and travertinised, with limestone fragments and fossils of marine shells. The marine event represented by layers I and L was attributed by Borzatti to MIS5e, providing a terminus post quem for human occupation, as observed at other Mousterian sites in the region.

3. Materials and Methods

This study examines macrofaunal remains of the Mousterian sequence of Uluzzo C, combining material from two excavation campaigns: 1577 remains from the historical excavations by Borzatti in the 1960s (layers E, F, and G), stored at Museo di Storia Naturale di Firenze, and 1233 remains from the recent investigations initiated in 2016 (stratigraphic units 23, 24, and 30). A total of 2810 faunal remains from the Mousterian layers were analysed using an integrated zooarchaeological and taphonomic approach, complemented by targeted ZooMS (Zooarchaeology by Mass Spectrometry) analysis. The study included all bone fragments, from larger pieces (>20 mm) to identifiable and indeterminate smaller fragments (<20 mm).

3.1. Taxonomic and Anatomical Analysis

Taxonomic and skeletal identifications were carried out in the Bones Lab (Laboratory of Osteoarchaeology and Palaeoanthropology) at the Department of Cultural Heritage of the University of Bologna (Ravenna, Italy), using both the reference collection, osteological atlases [72,73,74,75] and online resources, including 3D model repositories and photographic databases (e.g., Sketchfab.com; boneid.net; vertebres3d.fr; boneidentification.com). Faunal remains were categorised into determinate specimens, identified both taxonomically and anatomically, and indeterminate specimens, for which only anatomical identification, or no identification, was possible. Taxonomically unidentifiable specimens were grouped according to their body size into five mammals categories: size 1: smaller than a rabbit; size 2, small mammals weighing < 20 kg (Lagomorphs, Rodents and small carnivores); size 3, medium mammals between 20 and 100 kg (e.g., Capreolus capreolus, Canis lupus); size 4, medium/large mammals between 100 and 300 kg (e.g., Cervus elaphus, Dama dama, Crocuta spelaea); and size 5, large mammals of >300 kg (e.g., Bos/Bison). When species determination was indeterminate, elements were reported to their family (e.g., Cervidae, Caprinae), genus (e.g., Bos/Bison) or class (e.g., Aves) level.

To evaluate species abundance, we considered the number of identified specimens (NISP), the minimum number of individuals (MNI), the minimum number of elements (MNE), and the standardised minimal animal units (MAU) [76,77,78]. For the quantification of the long bones, the coding method established by Romandini [79] after Marean et al. [80] was followed. Animal age at death was determined by examining dental eruption and wear and assessing bone fusion, following Silver’s established criteria [81]. For cervids, we applied species-specific ageing frameworks [82,83,84,85]. Caprids and bovids were analysed using known criteria from Couturier [86], Habermehl [87,88] and Pflieger [89] and equids have been classified following the Fernandez and Legendre [90] protocol. Leporids were classified using Broekhuizen and Maaskamp [91], Gardeisen and Valenzuela [92] and Jones [93]. After establishing the age at death, each individual was assigned to one of five age categories: F/NB (foetal/new born); J (juvenile—characterized by light wear on deciduous teeth and eruption of the first molar, M1); SAd (subadult—showing moderate wear on deciduous teeth and eruption of the second molar, M2); Ad (adult—with all permanent teeth erupted and exhibiting light to moderate wear); and S (senile/old adult—marked by advanced dental wear). Age-at-death analyses were investigated through ternary diagrams showing the proportions of senile, adult, and young individuals for the most abundant ungulate taxa in each stratigraphic layer [94,95]; only taxa with a minimum number of individuals (MNI) greater than 4 per layer were considered.

To evaluate transport strategies and attrition processes, the correlation between %MAU and bone density was calculated only in layer G for large bovids (Bos/Bison) and cervids (Cervus elaphus and Dama dama), the most abundant species in the assemblage. In addition, to assess the exploitation strategies, %MAU of ungulates was also correlated with other utility indices, such as the Modified General Utility Index (MGUI; [96]), MAX density, the Food Utility Index (FUI; [97]), the Meat Grease and Marrow Indexes, and the Unsaturated Marrow Index (UMI; [98]). For all of them, Spearman’s and Pearson’s correlation coefficients were calculated using PAST [99]. Assemblage diversity was examined using the Inverse of Simpson’s Index (1/D) with an MNI [100]. The higher the value (starting from 1), the wider the diet breadth [101].

3.2. Taphonomic Analysis

The bone surface of every specimen over 3 cm long was examined under a Leica S9i stereomicroscope (0.75–70× magnification range, Leica Microsystems, Wetzlar, Germany) and a Hirox HR-2016 3d digital microscope (Hirox Co., Ltd., Tokyo, Japan) to identify biostratinomic and diagenetic alterations. Evidence of anthropogenic modification observed on the faunal remains includes cutmarks, intentional bone breakage and burning [102,103,104,105,106]. Cutmarks (incisions and scraping marks) have been identified and recorded considering the following characteristics for each bone specimen: quantity, location, distribution (isolated, clustered or crossed), and orientation (longitudinal, transverse and oblique) on the bone surface. Moreover, to determine specific butchering activities (skinning, evisceration, dismemberment, defleshing, periosteum removal, and tendon cutting), experimental and ethnological studies on medium/large prey were used [27,102,107,108,109,110,111,112,113]. Intentional bone breakage was inferred from type and angle of fracturing (fresh-green versus old-dry) following Villa and Mahieu [114], and percussion marks, such as cortical percussion notches (negative flake scars), impact flakes (positive flake scars), adhering flakes, percussion pits and crushing marks [109,115,116,117,118,119,120,121,122]. As for impact flakes, the distinction from those made by carnivores was made following Coil et al. [123]. Burning damage was recorded based on presence/absence (macroscopically) and assessed using a modified version of the criteria described by Stiner et al. [124].

Carnivore-induced modifications were identified by diagnostic features, such as pits, scores, punctures, furrowing, gnawing, crenulated edges, and evidence of gastric acid etching. These were classified following the criteria established by Blumenschine [125], Domínguez-Rodrigo and Piqueras [126], and Domínguez-Rodrigo and Barba [127]. Additionally, a range of other biotic and abiotic surface alterations was recorded [128], including weathering [129], root etching, insect or fungal damage, carbonate encrustations, polishing [78,105,108], and the development of mineral coatings, particularly manganese oxides [128]. To distinguish trampling-induced modifications from anthropogenic cut marks, diagnostic criteria proposed by Blasco et al. [130] and Domínguez-Rodrigo et al. [131] were applied.

3.3. ZooMS Analysis and Glutamine Deamidation

A total of 24 bone specimens were sampled and analysed at the Palaeoproteomics Laboratory of the Bones Lab (Department of Cultural Heritage, University of Bologna, Ravenna) for ZooMS identification. Approximately 30–50 mg of bone was sampled. Collagen extraction followed the protocol established by Buckley et al. [132]. Samples were demineralised in 0.6 M HCl (hydrochloric acid) and subsequently incubated in ammonium bicarbonate (AmBic). The resulting supernatant was collected for trypsin digestion, after which the tryptic digests were purified and concentrated using in-house STAGE tips (Empore™). Following peptide elution, samples were dried overnight under a hood, then resuspended and sonicated. A matrix of α-cyano-4-hydroxycinnamic acid (CHCA, 10 mg/mL Sigma- Aldrich, St. Louis, MO, USA) was prepared, and for each sample, 1 µL of matrix was mixed with 1 µL of sample directly on a ground-steel target plate and allowed to air dry. Each specimen was spotted in duplicate. Analyses were performed using an Autoflex maX MALDI-TOF mass spectrometer (Bruker, Billerica, MA, USA), calibrated at regular intervals throughout the analytical session.

Spectra, each acquired from 3000 to 5000 laser shots, were manually inspected and averaged using mMass [133], after setting a signal-to-noise ratio equal to 3.5, a baseline correction and smoothing with the default parameters (baseline correction precision 100 and relative offset 0; smoothing method Savitzky–Golay algorithm, window size 0.2 m/z and 1.5 cycles; peak picking with deisotoping tool, relative intensity threshold 0.5% and picking height 80%) [134]. Taxonomic identification was carried out manually by comparing collagen peptide markers with reference spectra from Pleistocene and Holocene medium- to large-sized mammals [135,136,137,138].

Deamidation ratios were calculated for all the analysed samples. It is represented by a number ranging from 0 to 1, with 0 corresponding to complete deamidation and 1 representing a non-deamidated peptide. Deamidation of glutamine and asparagine residues is a common, non-enzymatic modification resulting in a +0.98402 mass shift caused by the conversion of a side-chain amide group to a carboxylate via a cyclic intermediate (glutarimide/succinimide) under alkaline and neutral conditions or direct hydrolysis under acidic conditions (or in the presence of metal cationic species) [139]. Peptides with m/z values of 1105.6 and 1706.7 were selected to determine the percentage of deamidation [139] using the R 4.0.1 q2e package [140].

4. Results

4.1. Taxonomic Composition

In total, the analysed assemblage comprises 2810 remains, distributed unevenly across the three layers under study: 968 from layer E, 380 from layer F and 1462 from layer G (

Table 2. Number of identified specimens (NISP), minimum number of elements (MNE) and minimum number of individuals (MNI) from layers E, F and G of Uluzzo C.

	Layer E + SUs 23, 24							Layer F + SU 30						Layer G								Total Mousterian Layers
	NISP	%NISP	MNE	MNI				NISP	%NISP	MNE	MNI			NISP	%NISP	MNE	MNI					NISP	%NISP
				J	SAd	Ad	S				J	SAd	Ad				F/NB	J	SAd	Ad	S
Bos primigenius														4	0.3	4				1		4	0.2
Bos/Bison	4	2.8	2			1	1	2	1.7	2			1	178	12.4	43		2	1	4		184	10.8
Equus sp.	3	2.1	2			1		23	19.8	12	1		2	22	1.5	10			1	2	1	48	2.8
Cervus elaphus	18	12.7	5	1	1	2	1	33	28.4	14	1	1	3	35	2.4	21	1	1		2		86	5.1
Dama dama	7	4.9	5	1		2		15	12.9	9	1	1	1	35	2.4	20	1	1		1		57	3.4
Sus scrofa	4	2.8	3	1			1							5	0.3	5				1		9	0.5
Capreolus capreolus	1	0.7	1			1		6	5.2	3			1	3	0.2	2				1		10	0.6
Cervidae (size 3-4)	23	16.2						10	8.6					6	0.4							39	2.3
Caprinae														1	0.1							1	0.06
Ungulata	44	31						13	11.2					15	1							72	4.2
Total Ungulata	104	73.2	18	3	1	7	3	102	87.9	40	3	2	8	304	21.1	105	2	4	2	12	1	510	30
Crocuta spelaea	1	0.7	1			1								2	0.1	2				1		3	0.2
Panthera pardus														2	0.1	2				1		2	0.1
Canis lupus	3	2.1	1			1								1	0.1	1				1		4	0.2
Vulpes vulpes	21	14.8	11		1	2		7	6	6			1	2	0.1	2				1		30	1.8
Lynx lynx														1	0.1	1				1		1	0.06
Meles meles														1	0.1	1				1		1	0.06
Martes sp.	1	0.7	1			1																1	0.06
Carnivora	5	3.5						1	0.9					3	0.2					1		9	0.5
Total Carnivora	31	21.8	14		1	5		8	6.9	6			1	12	0.8	9				6		51	3
Lepus sp.	1	0.7	1			1								200	13.9	151	1	3		16		201	11.8
Oryctolagus cuniculus														51	3.5	48		1		8		51	3.0
Lagomorpha														164	11.4			3		10		164	9.7
Total Lagomorpha	1	0.7	1			1								415	28.8	199	1	7		34		416	24.5
Testudo hermanni														114	7.9	20		1		6		114	6.7
Aves	6	4.2						6	5.2					596	41.4							608	35.8
Total identified	142	100	33	3	2	13	3	116	100	46	3	2	9	1441	100	333	3	12	2	58	1	1699	100
Mammal size 2-3	13	1.3						13	3.4					2	0.1							28	1
Mammal size 4-5	8	0.8						5	1.3					10	0.7							23	0.8
Indeterminate	805	83.2						246	64.7					9	0.6							1060	37.7
Gran Total	968	100						380	100					1462	100							2810	100

; Figure 3; Figure S1). Sixty percent were identifiable to taxa, 2% to mammal body sizes and the remaining 38% were indeterminate fragments. The faunal assemblage comprises 3 vertebrate classes (Mammalia, Reptilia and Aves), with 16 species distributed across 17 genera. Mammals are represented by four orders (Lagomorpha, Carnivora, Perissodactyla, and Artiodactyla) and reptiles by chelonians; for birds, taxonomic identification did not proceed below the class level. Due to the fragmented state of the assemblages, a minimum number of elements (MNE) of 412 was quantified and a total MNI of 111, with adults predominant (Table 2).

Among the ungulates, Cervus elaphus represents the most abundant taxon across the assemblage and dominates layers E and F, while Dama dama is consistently represented in both levels. Equids are particularly abundant in layer F, whereas other ungulates, including Bos/Bison, Sus scrofa and Capreolus capreolus, occur in low proportions. Layer G shows a markedly different taxonomic composition, being dominated by birds and lagomorphs, while ungulates represent a minor component and are restricted to Bos/Bison and cervids (C. elaphus and D. dama). Other ungulate taxa have a scarce impact on the assemblage.

Among carnivores, remains are scattered. Vulpes vulpes is the most common taxon and is present in all three layers, with its highest abundance in layer E. The same layer also includes isolated remains of Canis lupus, Crocuta spelaea and Martes sp. In contrast, layer G shows a slightly higher taxonomic diversity among carnivores, with the occurrence of Panthera pardus, Lynx lynx and Meles meles, although each taxon is represented by one or two specimens only and does not affect the overall faunal composition.

4.2. Age-at-Death, Seasonality and Sex Determinations

As shown in Table 2, the three layers are dominated by adult individuals. Ungulates in layers E and F show comparable age structures dominated by adults, with a minor contribution of juvenile and senile individuals, whereas layer G displays a more diversified profile including foetal/newborn, juvenile, and adult individuals, with rare senile specimens. Carnivores are represented by very low MNI values in all layers and are almost exclusively composed of adult individuals, with no evident age structuring. Lagomorphs are negligible in layers E and F but show a marked increase in layer G, where the assemblage is dominated by adult individuals with a limited juvenile component. Testudo hermanni is recorded only in layer G and is represented by one juvenile specimen among all adults.

To further explore the age structure of the most abundant ungulate taxa, a ternary plot was constructed based on the proportions of senile, adult, and young individuals in each stratigraphic layer (Figure 4). Only taxa with an MNI > 4 per layer were included. Cervus elaphus, represented above this threshold in all layers, shows variable age structures, ranging from a relatively balanced profile in layer E to assemblages dominated by adults and young individuals in layers F and G, with senile individuals absent in the latter. Bos/Bison and Equus sp. exceed the inclusion threshold only in layer G. Bos/Bison is characterised by a dominance of adults and young individuals, whereas Equus sp. displays a more heterogeneous profile, including senile individuals.

Seasonality can be tentatively inferred from the presence of very young individuals. Two foetal/newborn cervids (one Cervus elaphus and one Dama dama) suggest mortality between summer and late autumn, while a juvenile Bos primigenius mandible with dp2–dp4 present and the first molar erupting, estimated at approximately six months of age, indicates death in early winter. Although based on a limited sample, these observations point to repeated or prolonged site use from early summer through early winter.

Regarding sex determination, it was only possible to infer the presence of a Cervus elaphus male specimen in layer G, where two antler fragments (MNE = 1) were recovered.

4.3. Taphonomic Observations

The bone assemblage shows very high fragmentation in layers E and F, where complete skeletal elements represent only 1% (N = 7 and 5, respectively), consisting primarily of teeth and compact bones, such as phalanges, carpals, tarsals, and sesamoids. In layer G, the proportion of complete elements rises to 15% (N = 218), reflecting the dominance of small game (lagomorphs, chelonians, and birds) whose small, compact bones are less prone to fragmentation than those of larger animals (Table S1). The high fragmentation observed in layers E and F may be related to the collapse of the cave roof, which would have left the remains exposed to weathering. However, more intensive exploitation of prey cannot be ruled out.

Low-density axial elements (ribs, sternum, and vertebrae) and girdle bones (shoulder and pelvic) are virtually absent in layer E and scarce in layer F (2%), likely due to their poor preservation potential. In layer G, however, fragmented remains of these elements reach 11%, a proportion largely driven by chelonian shell elements, which, despite being classified as axial elements, are considerably more resistant to fragmentation and destruction than other low-density bones (Table S1). In contrast, long bone shaft fragments are abundant (E = 5%; F = 14%; G = 10%; Table S1). Epiphyses, spongy, and flat bones are consistently underrepresented across all layers, with the lowest values recorded in layers E and F and virtually absent in layer G (Table S1).

Overall, the bone assemblage shows generally good preservation, with post-depositional fractures representing less than 10% of the total across all layers, though more frequent in layer G (Table S1). Fresh fractures are very limited in layers E and F (<1% and 2%, respectively), but more common in layer G (12%). Green breakage was generally identified on long bones and metapodials (Table S2).

Anthropogenic modifications were identified exclusively on faunal remains from layers E and G (

Table 3. Number of remains showing anthropogenic modifications: cut marks (CM) and percussion marks (PM) with their percentage as butchery marks (%BM) relative to NISP, and thermal alteration (TA).

Anthropogenic Modifications
	Layer E+ SUs 23, 24					Layer G
	NR	CM	PM	%BM	TA	NR	CM	PM	%BM	TA
Bos primigenius						4				1
Bos/Bison	4					178	2		1.1	49
Equus sp.	3					22	1		4.5	3
Cervus elaphus	18	1		5.6		35	2		5.7	5
Dama dama	7					35		1	2.9	5.0
Sus scrofa	4					5				1
Capreolus capreolus	1					3				1
Cervidae (size 3-4)	23					6				1
Caprinae						1
Ungulata	44					15				2
Crocuta spelaea	1					2
Panthera pardus						2
Canis lupus	3					1
Vulpes vulpes	21					2
Lynx lynx						1				1
Meles meles						1
Martes sp.	1
Carnivora	5					3
Lepus sp.	1					200	7		3.5	36
Oryctolagus cuniculus						51				5
Lagomorpha						164	2		1.2	11
Testudo hermanni						114	8		7	29
Aves	6					596
Mammal size 2-3	13					2
Mammal size 4-5	8					10				3
Indeterminate	805	1	3	0.5	45	9				1
Total	968	2	3	0.5	45	1462	22	1	1.6	154

), including butchery marks (cutmarks and intentional bone breakage) and evidence of thermal alteration (TA). Cutmarks (CM) and percussion marks (PM) were analysed according to their distribution, orientation, and morphology to identify butchering activities. The recorded marks primarily consist of short incisions associated with skinning, defleshing, and periosteum removal, along with scraping marks related to periosteum stripping and deep striae mainly associated with tendon extraction. Anthropogenic bone breakage was inferred from two types of percussion-related traces: impact notches and flakes.

In layer E, cutmarks were recorded on one red deer tibial shaft (%BM = 5.6) and on one flat bone of an indeterminate mammal. In layer G, cutmarks were observed on ungulates, leporids, and chelonids. Among ungulate remains, two bones of large bovines (%BM = 1.1), one equid bone (%BM = 4.5), and two red deer bones (%BM = 5.7) show evidence of butchery. All three taxa exhibit multiple obliquely oriented incisions on the hemimandible. In B. primigenius and Equus sp., cutmarks are located on the medial surface just below the last premolar (dp3 in the case of Equus) and first molar, indicating mandible disarticulation and/or tongue removal (Figure 5D,E). The C. elaphus hemimandible shows cutmarks on the lateral surface beneath the condylar process, consistent with joint disarticulation or flesh removal (Figure 5A). The diaphysis of one Bos/Bison metatarsal bears multiple longitudinal incisions and oblique scraping marks associated with skinning and periosteum removal. Similarly, transverse incisions and deep striae were recorded on one red deer metatarsal near the proximal epiphysis (Figure 5B), suggesting hide removal and tendon extraction. Percussion marks include two flakes and one impact point on indeterminate remains from layer E, and one D. dama metatarsal with consecutive and opposite notches from layer G, all related to marrow extraction (Figure 5C). Among lagomorph remains, cutmarks were identified on nine bones: seven on Lepus sp. (%BM = 3) and two on indeterminate Lagomorpha (%BM = 1). These traces were distributed across nearly all anatomical districts, except for cranial and axial regions (Figure 6A). Regarding chelonians, cutmarks were recorded on eight bones (%BM = 7), predominantly on shell elements and humeri (Figure 6B).

Thermal alterations (Table 3) indicate a higher degree of combustion in layer G (10% of finds) than in layer E (5% of finds). In the uppermost layer, fire damage is limited to indeterminate remains and a single bone of an indeterminate ungulate (Figure S2A). In contrast, in layer G all ungulates, leporids and chelonians are affected by thermal alteration, together with a single specimen of L. lynx (Figure S2B). Fire damage is documented across all anatomical regions, including teeth, cranial fragments, small skeletal elements (such as phalanges and sesamoids) and chelonian shell elements. Overall, the degree of combustion is moderate; most of the affected specimens exhibit brown and/or black colouration, whereas only five calcined specimens (grey to white) were recorded, all originating from layer G.

Biostratinomic modifications are rare across all three stratigraphic units, affecting less than 5% of bone remains in each layer (

Table 4. Number of remains and percentage for the biostratinomic and diagenetic alterations recorded in the Mousterian layers.

		Layer E + SUs 23, 24		Layer F + SU 30		Layer G
		NISP	%NISP	NISP	%NISP	NISP	%NISP
Biostratinomic alterations	Rodent marks	1	0.1			21	1.4
	Carnivore marks	1	0.1			17	1.2
	Digestion marks	2	0.2	5	1.3	3	0.2
	Weathering	3	0.3			1	0.1
	Dermestid pupal chambers	3	0.3			7	0.5
	Trampling marks	5	0.5			11	0.8
Diagenetic alterations	Concretions	295	30.5	325	85.5	489	33.4
	Manganese staining	277	28.6	145	38.2	624	42.7
	Root marks	27	2.8	12	3.2	181	12.4
	Corrosion	1	0.1	3	0.8	8	0.5
	Exfoliation			1	0.3	6	0.4
	Water dissolution	3	0.3
Total NISP		968	100	380	100	1462	100

). This pattern is consistent with relatively rapid burial and stable depositional environments. Evidence of carnivore activity is limited to isolated tooth marks (one in layer E, 17 in layer G) and digestion marks (two in layer E, five in layer F, and three in layer G). Rodent gnawing is similarly scarce (one in layer E, 21 in layer G). Weathering and trampling marks are infrequent (<1%) and occur only in layers E and G, while dermestid’s pupal chambers were recorded on a few remains from layer E (N= 3) and layer G (N= 7).

In contrast, diagenetic alterations are abundant throughout the bone assemblage and reflect notable variability in post-depositional processes across layers E, F, and G (Table 4). Concretions and manganese staining are the most frequent modifications, with concretions peaking in layer F (85%) and manganese oxide deposits most prominent in layer G (43%). These patterns likely reflect the specific microenvironmental conditions of Uluzzo C, interpreted as a karstic habitat where the sheltered setting promoted localised water percolation and mineral precipitation. Root etching is moderately represented, particularly in layer G (12%), suggesting periodic penetration of plant roots into the sediment, possibly facilitated by minor fissures in the overlying rock.

Despite the prevalence of mineral-related alterations, chemical weathering features, including corrosion, exfoliation, and water dissolution, occur at consistently low frequencies throughout the sequence (<1%). This pattern suggests that while the karst environment provided conditions conducive to mineral deposition through percolating groundwater, bones were sufficiently protected from prolonged subaerial exposure and aggressive chemical weathering, likely due to relatively rapid burial within the cave sediments.

4.4. Skeletal Profiles

Ungulates’ skeletal composition across the three layers is characterised by a significant presence of cranial elements, particularly isolated teeth, along with some fragmented parts of mandible, cranium and horn/antler (Table S3). In contrast, the axial skeleton is nearly absent, with few fragments of ribs and vertebrae from the three layers and two fragments of red deer sternum from layer G. The scarcity of post-cranial bones is more evident in layers E and F, whereas in layer G, skeletal profiles of both ungulates and carnivores are more complete.

As concerns main medium-large mammal taxa, the three layers show some differences in the distribution of forelimb, hindlimb, and extremities. Layer E is particularly poor, as red deer is only represented by one fragment of tibia, while the red fox skeletal profile only includes a few remains of fibula, metatarsal and phalanx II. In layer F, red deer are completely missing postcranial elements, whereas equids and fallow deer skeletons are better represented. Equids show some remains of limb bones (radius, carpals and metapodials) as well as phalanges (I and II), while fallow deer show more remains of the anterior extremity (humerus, radius and metacarpal) than the posterior (tibia and metatarsal). In layer G, post-cranial elements are more abundant and allow for more refined estimates of attrition processes and prey transport strategies.

The large bovines (Bos/Bison) skeletal profile is mostly complete, as only a few elements are missing (i.e., radius, ulna and phalanx II). Based on NISP, the mandible (NISP = 12) stands out among cranial elements (excluding teeth) and horns (NISP = 6) are well represented. Likewise, in the appendicular skeleton, the most abundant elements are the tibia (NISP = 6), phalanx I (NISP = 5) and metatarsal (NISP = 4) (Table S3). The skeletal profile based on %MAU corroborates this pattern, with mandible and horn reaching the highest values (100% and 86%, respectively). Among post-cranial elements, humerus, tibia and astragalus display the highest %MAU values (43%) (Figure 7; Table S4). Correlation analysis between %MAU and the bone density for Bos/Bison shows no significant correlation (p > 0.05), nor are significant correlations found with utility indices (Table S5).

Regarding the remains of red deer and fallow deer from layer G, to increase statistical consistency, they were aggregated into a single category of medium-large cervids (NISP = 70) (Table S3). Although these are different genera, their similar size and bone morphology make density and utility indices comparable. The results should therefore be interpreted as indicative of exploitation strategies applied generically to medium-large cervids, without necessarily distinguishing between species. The cervid’s skeletal profile is generally complete, except for the absence of scapula, ulna, carpals, femur and phalanx III. According to NISP, the best represented post-cranial element is the metatarsal (NISP = 23), followed by the radius (NISP = 7) and phalanx I (NISP = 5), while in the cranial region, the mandible (NISP = 7) is the most abundant (not considering teeth). The %MAU profile shows a marked overrepresentation of metatarsal (100%) and mandible (87%), followed by radius (50%) and astragalus (25%), whereas phalanx I displays low values (<10%) (Figure 7, Table S4). Correlation analyses indicate statistically significant positive relationships between %MAU and several parameters (Table S5). A significant correlation is observed with bone density (ρ = 0.58; p = 0.003) (Figure 8). Significant positive correlations are also recorded with the Marrow Index (ρ = 0.56; p = 0.003) and UMI (ρ = 0.64; p = 0.017), while the correlation with the Grease Index approaches statistical significance (ρ = 0.40; p = 0.051) (Figure 8).

4.5. ZooMS Analysis

Out of the 24 samples analysed, only four obtained a taxonomic attribution, conferring a success rate of 16% (

Table 5. Taxonomic results obtained from ZooMS analysis for Uluzzo C’s samples by layers.

Species	SUs			Grand
	E	F + 30	G	Total
Bos/Bison		1	1	2
Ungulate		1		1
Total Ungulate		2	1	3
Human?	1			1
Total determined		2	1	3
Indeterminate	1	7	12	21
Grand Total	2	9	13	24

). One has been categorised as a possible human (Human? in Table 5) due to the presence of a single peptide associated with this species. However, given the very low spectral quality and the detection of only one human-specific peptide, this evidence is insufficient to support a reliable taxonomic determination. Further analyses, such as genetic investigations, are therefore required to achieve a secure identification of the fragment; for this reason, the sample has already been selected. The deamidation rates calculated for these samples are comprised between 0.26 and 1 for the peptide m/z 1105.57 and 0.07 and 0.42 for the peptide m/z 1706.77 (Figure S3). This wide range of values is contradictory to the low success rate obtained; further analysis would be required to identify the reason for such a gap.

5. Discussion

The palaeoenvironmental and zooarchaeological patterns documented at Uluzzo C must be interpreted in light of several factors that unevenly affect the number, preservation, and representativeness of the faunal remains across layers. Differences in specimen counts partly reflect post-depositional processes, including episodes of the vault collapse [58,65] that influenced both bone preservation and the visibility of surface modifications relevant for taphonomic analysis. In addition, the preliminary results of ZooMS analyses further support the interpretation that the local depositional context is unfavourable to organic preservation. The overall poor preservation of collagen in the samples is consistent with past analysis, which demonstrated that the pH of the sediment, the proximity of the sea (marine spray) and the average annual temperature of 17.6 °C explain the scarce preservation of the organic component at Uluzzo C [62,65]. Further uncertainty derives from the excavation history: because Borzatti did not provide a total count of the recovered faunal remains, it cannot be determined whether the material currently available corresponds to the entirety of what was originally excavated. Furthermore, the restricted horizontal exposure of the deposit, excavated primarily through narrow trenches, limits spatial comparability and may accentuate sampling biases. These constraints caution against overly fine-grained layer-by-layer comparisons and favour an interpretative framework centred on broader trends within the stratigraphic sequence. Nevertheless, despite these limitations, the assemblages preserve sufficient taxonomic and taphonomic information to support meaningful, although necessarily partial, palaeoecological and behavioural inferences. Ongoing excavations at the site will provide a more complete and spatially coherent record, allowing future research to confirm or revise these provisional interpretations.

5.1. Paleoenvironmental and Chronological Reconstruction of the Mousterian Layers

The taxonomic composition of the Mousterian sequence at Uluzzo C reflects dynamic environmental conditions and provides insight into the ecological and chronological framework of the site. Large bovids indicate the presence of open environments as well as wooded areas with clearings, typical of temperate conditions [141,142], while equids point to more extensive steppe and grassland environments under continental climatic regimes [141,142]. Cervids document a variety of wooded habitats, from dense forests for roe deer to Mediterranean scrub for fallow deer, with red deer, generally dominant in cold temperate phases, indicating the presence of mature forests [143]. Other taxa, such as wild boar, hare, rabbit, wolf, lynx and fox, are broadly adaptable and provide more limited palaeoecological resolution. Overall, the assemblage points to a heterogeneous mosaic landscape, combining open grasslands, woodland patches, Mediterranean vegetation and coastal wetlands.

Comparisons of the layers reveal subtle but significant ecological shifts (Figure S1). Layer E is dominated by C. elaphus, together with V. vulpes and D. dama, pointing to open woodland and ecotonal settings. Layer F, while still dominated by red deer, shows an increase in equids, indicating a more open grassland-forest mosaic. Layer G differs markedly, with abundant leporids and a significant representation of large bovids, alongside T. hermanni, pointing to increased openness, while maintaining temperate to warm-temperate niches suitable for tortoises.

Taxonomic diversity values further refine these interpretations. Layer E shows the highest Inverse Simpson’s Index (1/D = 6.67), reflecting a relatively heterogeneous faunal community typical of transitional ecotonal settings. Layer G exhibits similarly high diversity (1/D = 6.52), consistent with an open environment that nevertheless accommodated multiple ungulate taxa in comparable proportions. Layer F displays the lowest diversity (1/D = 4.26), which might suggest more ecologically constrained conditions dominated by a reduced suite of species.

The faunal evidence is supported by pollen data from layers E and F, which document a highly diverse and stable vegetation mosaic, including savannas with pine, oak, and juniper, steppe-like saltmarshes, riverine forest patches, open woodlands, shrubby grasslands with conifers and mesophytes, rocky scrub, grasslands with heaths and litoral vegetation [60]. These shifts outline a gradual transition from comparatively more closed and humid environments in layer E to increasingly open conditions in layers F and G.

The combination of taxonomic composition, species-specific climatic affinities, diversity patterns, and pollen data reinforces the view of Uluzzo C as a dynamic landscape, with each layer capturing a distinct yet complementary snapshot of paleoenvironmental variability. Compared with higher-latitude European contexts, the Italian peninsula, especially the south, maintained relatively mild conditions and high ecological diversity even during the cold stages of the Quaternary [144,145]. The proximity of the Ionian Sea at Uluzzo C likely helped moderate local climates, adding complexity to faunal patterns.

Most of the identified species have been present in Italy since the Middle Pleistocene, but the record also captures evolutionary and chronological dynamics. Modern forms of red deer and fallow deer are known only from the Late Pleistocene [146,147], while P. pardus, C. spelaea, and D. dama disappeared shortly before MIS 3 or during the Last Glacial Maximum. The latest occurrences of E. ferus date to the early Holocene, B. primigenius survived into historic times, and other taxa are still extant [143,148]. In Borzatti’s work [57], the presence of Rhinoceros mercki, currently referred to as Stephanorhinus kirchbergensis Jäger, 1839, was reported within layer G (spit 17), whereas no evidence of this taxon was identified in the present re-examination of the faunal record. Nevertheless, the possible occurrence of S. kirchbergensis would represent a highly significant palaeoecological and chronological indicator. This rhino is considered a grazer or mixed feeder, adapted to open environments and temperate to cold climatic conditions [149,150]. In the Italian and European record, the species is documented mainly during interpleniglacial phases of the late Middle Pleistocene, spanning from MIS 15 to MIS 5 [151,152]. Given that the base of layer G overlies a Tyrrhenian beach deposit attributed to MIS 5, its identification within layer G, and more specifically in spit 17, would therefore allow a considerable refinement of the chronological framework, support regional-scale stratigraphic correlations and contribute to a more precise reconstruction of the local paleoenvironment. Of chronological significance is the association of cave hyena in layers E and G with fallow deer, found across all layers. This combination firmly situates the sequence within a Late Pleistocene interval, between MIS 5 and MIS 3 [146,148]. Taken together, the palaeoecological and faunal patterns observed at Uluzzo C are fully compatible with the chronometric framework established with the OSL ages for the upper Mousterian units and the constraint provided by the PGT/Y-6 tephra [60,65], supporting a Mousterian occupation spanning from MIS 5 to late MIS 3.

A comparison between the Mousterian and Uluzzian layers at Uluzzo C highlights both continuity and significant ecological transformations, consistently reflected in the faunal and pollen records. The Uluzzian sequence (layers C and D; [63,64]) records a progressive environmental transition from relatively wooded and humid conditions to increasingly open and fragmented habitats. From a quantitative perspective, diversity metrics further underscore these shifts. Despite this, diversity values for the Uluzzian assemblage (1/D = 6.25) remain comparable to the most heterogeneous Mousterian layers, indicating sustained ecological complexity. This suggests that, despite the increasing openness of the landscape, the Uluzzian phases maintained a high degree of ecological heterogeneity, comparable to the more diversified Mousterian horizons and distinct from the more constrained conditions reflected in layer F.

In this context, Uluzzo C offers both a local picture of habitat changes and a representative example of broader regional palaeoecological dynamics during the late Middle Palaeolithic.

5.2. Uluzzo C Within the Apulian Mousterian Framework

Considering the Mousterian layers of Uluzzo C (E, F, and G) collectively, the faunal assemblage exhibits a consistent pattern of dominance by cervids, equids, and large bovids, reflecting the exploitation of heterogeneous landscapes that combine open areas with patches of woodland. When compared with coeval Mousterian sites across southern Italy, particularly in Apulia (Table S6), the Uluzzo C Mousterian sequence aligns closely with the ecological patterns documented at Grotta del Cavallo [153,154,155], Riparo l’Oscurusciuto [53,156,157], and Grotta Santa Croce [46,55,158], where cervids, equids, and aurochs dominate, indicating semi-open to open woodland environments. Broader consideration of other sites, including Grotta dei Giganti [159], Tana delle Iene [141], Grotta Romanelli [160,161,162,163], and Avetrana [142,164,165], reveals a consistent regional pattern: faunal assemblages frequently combine large bovids, cervids, and equids, reflecting heterogeneous landscapes that range from ecotonal woodlands and semi-open forests to more open, steppe-like habitats.

Zooarchaeological and taphonomic analyses demonstrate that bone accumulation at Uluzzo C was predominantly anthropogenic throughout the Mousterian sequence, as highlighted by diagnostic fracture patterns, butchery marks, and thermal modifications. These findings corroborate the evidence for human occupation already attested by the considerable lithic assemblage recovered throughout all levels, a production fully consistent with the Mousterian technocomplex documented across the Apulian region [56,166], as well as by the hearths identified in level G [58].

Nevertheless, the possible involvement of carnivores, particularly hyenas, warrants examination to assess their contribution to the assemblage. Borzatti already reported the presence of hyenas in layer G based on the recovery of coprolites [57]. In line with this, the faunal assemblage includes coprolites preliminarily attributed to cave hyena, currently under study, from layers E and G. In addition, fossil remains of cave hyena, including a mandible of a juvenile individual, have been recovered from the same layers (E and G). Despite the low frequency of carnivore tooth marks and other diagnostic modification traces (7 in the whole assemblage), the presence of coprolites and cave hyena remains suggests that the cave was used as a den, albeit sporadically, most likely when Neanderthal groups were elsewhere. Such alternating use of caves by humans and large carnivores is a well-documented pattern in Late Pleistocene contexts in Italy, including Grotta Paglicci [167], Grotta Guattari [168], Grotta del Fossellone and Grotta Breuil [169], Buca della Iena [170] and Grotta dei Santi [171] among others.

Within this framework, the limited impact of carnivore activity on bone surfaces at Uluzzo C may reflect short-term or episodic carnivore occupations rather than intensive carcass accumulation and processing. Overall, this integrated evidence provides a coherent framework for interpreting the formation processes and post-depositional history of the site assemblage, emphasising a predominantly anthropogenic origin of the deposit, with a secondary contribution from carnivore use of the site. Having established the predominantly human origin of the faunal assemblage, it becomes possible to reconstruct the subsistence strategies that drove its formation.

The results of this study may indicate that, in layers E and F, Neanderthal hunting strategies were primarily focused on medium- to large-sized ungulates, with a clear emphasis on large bovines, cervids, and equids. During the earlier occupational phase (layer G), human predation appears to have been more variable, spreading from large bovines and cervids to leporids and chelonians. A substantial presence of butchery marks and thermal alteration on small game remains show clear evidence of human consumption, demonstrating that small prey exploitation was part of Neanderthal behaviour at the site.

The Mousterian assemblage from Uluzzo C can be further contextualised within the broader framework of Middle Palaeolithic sites in Apulia, which show consistent patterns of resource exploitation and site use. Comparable strategies are documented at Riparo l’Oscurusciuto (Ginosa, Taranto), where high-utility portions of large ungulates were preferentially transported and intensively processed, and where lithic and faunal evidence suggest structured activity areas and repeated use of space for butchery and other tasks [53,156,157]. Similarly, Grotta del Cavallo (Nardò, Lecce) exhibits systematic butchery of cervids and other large ungulates, along with a persistent presence of anthropogenic modifications across multiple layers, reflecting repeated and structured Neanderthal occupations [54,154,155,172]. Grotta di Santa Croce (Bisceglie, Bari) further confirms these regional patterns, showing dominance of adult large ungulates, intensive bone breakage to access marrow and grease, and selective transport of high-utility skeletal elements [55,173,174,175].

When considered within a diachronic framework, the age-at-death patterns of the main prey taxa at Uluzzo C (Figure 4) can be discussed in relation to broader changes in Neanderthal hunting behaviour, while remaining mindful of the interpretative constraints imposed by the limited dataset. In the earliest occupation (layer G), the predominance of prime-age adults among large bovines and equids, together with the scarcity of old-adult individuals among red deer, may suggest some degree of selective exploitation of high-return prey. A similar pattern could be tentatively identified in layer F, where the apparent focus on prime adult red deer might reflect a continuation, or even a mild intensification, of such strategies. In the most recent occupation (layer E), however, the U-shaped mortality profile observed for red deer, characterised by an increased representation of juveniles and older individuals, may indicate a shift towards more vulnerable animals, possibly linked to changes in prey availability, hunting opportunities, or efficiency. Taken together, these patterns tentatively suggest a diachronic trend in prey selection strategies across the sequence, though the available evidence remains limited and alternative explanations cannot be ruled out.

When tentatively considered alongside preliminary indications of summer hunting activity, these demographic patterns could suggest a degree of forward planning that exploited seasonally predictable aspects of prey behaviour, such as migration routes or aggregation sites. However, it is important to emphasise that the seasonal signal within the assemblage remains ambiguous. The current evidence does not permit other periods of site occupation or hunting activity to be confidently excluded, nor does it establish a definitive seasonal focus. Both the mortality data and the seasonal indicators should therefore be regarded as suggestive of possible behavioural patterns rather than as conclusive demonstrations of planned seasonal hunting strategies.

Regarding carcass processing and transport strategies, several hypotheses may be advanced based on the available evidence. The skeletal representation of large ungulates across the Mousterian sequence is dominated by cranial remains, primarily teeth, with limb bones occurring less frequently and axial elements being extremely scarce. While dental survivorship partly reflects the superior durability of tooth enamel, the broader pattern suggests deliberate choices in which body parts were introduced to the site. Layer G, which yielded the most anatomically diverse assemblage, offers the clearest window into these transport behaviours, particularly for large bovids and cervids where sample sizes exceeded the threshold for meaningful statistical evaluation (MNE > 35).

For large bovids, the skeletal profile reveals a relatively complete representation of appendicular elements alongside a near-complete absence of axial bones and a marked underrepresentation of extremities. Although statistical tests between %MAU values and both bone mineral density and food utility indices failed to produce significant correlations, this may reflect the complex interplay of multiple formation processes rather than the absence of patterned behaviour. The profile itself, emphasising limbs and crania while excluding lower-value body portions, could tentatively suggest field butchery practices in which hunters may have prioritised the transport of meat-bearing and marrow-rich elements over bulkier, lower-yield axial portions.

Unlike large bovids, the cervids assemblage from layer G yields stronger statistical support for nutritionally driven transport strategies. While bone mineral density shows a significant positive correlation with skeletal part representation, indicating that attritional loss preferentially affected less robust elements, the assemblage also exhibits significant correlations with multiple food utility indices. The relationship is particularly pronounced for marrow values, suggesting that elements rich in this high-calorie resource may have been preferentially transported or retained. This is further supported by the presence of percussion marks indicative of marrow extraction, including anvil-assisted fracturing (Figure 5C). Weaker but still significant positive correlations with grease and general meat utility (UMI) indices also suggest that soft tissue exploitation may have played a role, a pattern tentatively corroborated by the presence of cutmarks. Such preferences would have been influenced by factors including prey body size, group composition, and the distance between kill and consumption locations. In this regard, the abundance of cranial remains at Uluzzo C warrants consideration in relation to ethnoarchaeological models linking head transport to group size and mobility [176,177]. The pattern at Uluzzo C, particularly in layer G, with a pronounced representation of cranial elements, implies that entire heads may have been transported rather than discarded during field processing. Therefore, if layer G represents occupations by groups large enough to manage the transport of heavy, lower-utility body parts, cranial abundance would align with reduced selectivity characteristic of larger aggregations [178]. Alternatively, heads may have been specifically targeted for secondary exploitation requiring controlled heat and time investment, such as brain extraction or rendering of cranial soft tissues [84,179]. Both scenarios imply a structured approach to carcass management and thermal processing strategies. The nature and intensity of such thermal processing can be further explored through the distribution and characteristics of fire-related bone modifications across the sequence.

Although not abundant, the proportion of thermally altered remains in layer G (10%) is nonetheless noteworthy, especially when compared to the uppermost Mousterian layers, among which only layer E yielded more limited evidence (5%) (Figure S2). Fire-related modifications are almost entirely of moderate intensity: charred bones dominate the thermally altered assemblage, while calcined specimens are virtually absent. This pattern closely resembles that documented for the Uluzzian layers at Uluzzo C, where 13% of faunal remains exhibit burning [62,63], but differs from other regional Mousterian sites, such as Grotta del Cavallo, Grotta Santa Croce, and Riparo l’Oscurusciuto, where thermally altered specimens account for less than 5% of their respective assemblages [55].

These differences reflect contrasting approaches to fire management and carcass processing across Apulian Neanderthal sites. At Oscurusciuto, despite the presence of multiple hearth features, bone remains show no evidence of having been used as fuel, leading researchers to attribute the scarcity of vertebrae and ribs to selective transport strategies rather than combustion-related destruction [157]. Conversely, at Grotta del Cavallo and Grotta Santa Croce, both interpreted as residential camps where intensive butchery activities took place, the low frequency of thermal alterations has been linked to systematic bone grease rendering. This inference rests on the combination of skeletal profiles dominated by high-utility elements, mortality patterns focused on prime adults, absence of carnivore modifications, and the minimal thermal alteration consistent with controlled, low-temperature heating required for grease extraction and exploitation, which seemed to have been particularly significant for Apulian Neanderthals [158].

Within this framework, Uluzzo C presents a more ambiguous picture. Like Oscurusciuto, the assemblage exhibits selective transport favouring crania and limbs, with no clear evidence that bone was systematically used as fuel. However, unlike the bone grease rendering scenario documented at Cavallo and Santa Croce, characterised by minimal thermal modification, layer G displays higher frequencies of moderate-intensity burning. This pattern does not align neatly with either model: the thermal alterations are too abundant to reflect dedicated low-temperature grease rendering, yet too moderate in intensity to suggest use of bone as primary fuel. The distribution and character of burning may instead point to incidental exposure during other cooking or heating activities, or to more varied fire-related processing practices than those documented elsewhere in the region.

It is worth noting that the combination of frequent, albeit moderate-intensity, thermal alterations, structured hearth features, and evidence for on-site marrow extraction points toward a residential occupation pattern. This interpretation is consistent with the hypothesis of Uluzzo C as a residential camp based on the lithic evidence, which records a complete chaîne opératoire [180]. Uluzzo Bay offered key advantages for repeated occupation, combining access to high-quality flint, with a plateau overlooking the bay suitable for hunting and foraging, where northern Ionian coastal sites, such as Uluzzo C and Grotta del Cavallo, were used as prolonged residential bases, while southern locations fulfilled more specialised logistical roles [180]. In addition, skeletal part representation and butchery trace frequencies may indicate selective transport of high-utility elements and limited on-site primary carcass processing. The low frequency of axial elements and extremities implies that the early stages of carcass dismemberment may have taken place elsewhere in the landscape, with only selected portions brought into the cave for final consumption and marrow extraction. This pattern likely reflects logistical decisions related to prey body size, kill-site distance, and mobility constraints, suggesting that even residential bases functioned within a broader spatial framework in which different stages of carcass processing were distributed across multiple locations. When considered together, the lithic and faunal evidence identify Uluzzo C as a central node of intensive and diverse technological production, while subsistence activities may have been primarily focused on late-stage carcass processing from selected skeletal elements.

6. Conclusions

The Mousterian faunal record from Uluzzo C highlights the complexity of depositional processes operating in cave and rock shelter contexts in southern Italy during the Late Pleistocene. The assemblages reflect a dynamic interplay between environmental conditions, human presence, and carnivore activity, resulting in stratified deposits characterised by palimpsest conditions rather than by single, discrete occupational events.

The taxonomic and ecological composition of the faunal remains points to a heterogeneous landscape combining open and wooded environments, consistent with a Mediterranean refugial setting. Such environmental diversity likely contributed to the recurrent use of the area over time, while also favouring the presence of multiple carnivore taxa, whose activities played a significant role in shaping the faunal assemblages. The alternating signatures of human and carnivore modification underscore the importance of considering non-anthropogenic contributions when interpreting Mousterian faunal records in similar settings. From a behavioural and spatial perspective, the evidence supports an interpretation of Uluzzo C as a location subject to intermittent and short-term human visits rather than prolonged or continuous occupation. This pattern is consistent with broader models of Middle Palaeolithic site use in ecologically rich but highly competitive landscapes, where caves and rock shelters functioned as repeatedly accessed places within wider mobility systems. Placed within a diachronic framework, the Mousterian deposits of Uluzzo C provide a critical point of reference for evaluating changes and continuities in site formation processes across the Middle to Upper Palaeolithic transition. Differences observed between the Mousterian and the overlying Uluzzian levels suggest that shifts in depositional dynamics and site use cannot be attributed solely to cultural change but must also be considered in relation to environmental conditions and the role of non-human agents.

More broadly, these results emphasise the interpretative value of zooarchaeological and taphonomic analyses for understanding the formation of archaeological deposits in Mediterranean contexts. Careful consideration of accumulation agents, palimpsest formation, and environmental proxies remains essential for constructing robust behavioural and palaeoenvironmental reconstructions, particularly in long and complex stratigraphic sequences, such as that of Uluzzo C. Ongoing excavations of Mousterian levels of the site will further contribute to refining these interpretations and to expanding the available dataset, allowing a more detailed assessment of depositional dynamics and environmental variability during the Middle Palaeolithic occupation of the site.