What the deer comparison establishes
Deer made Malta a usable food destination, but the available comparison does not establish better hunting than Sicily. Latnija preserves processing of small deer; Uzzo shows that deer remained important prey in Sicily too. The useful next hypothesis is that Malta offered predictable access to several resources within a compact area, rather than unusually large animals or a demonstrated surplus.
Comparison LAT-DEER-001, October 28, 2026. This first comparison of accessible evidence is complete. A matched quantitative comparison of kill ages, processing intensity and returns remains unavailable. Latnija’s detailed fauna study is a January 2026 preprint, including its supplementary tables and code. [S323 · Mata-González et al. (2026); S330 · Mata-González et al. (2026)]
Choose comparable occupations first
| Assemblage | Published time and context | Role in this comparison |
|---|---|---|
| Latnija Phase V | Approximately 8,500–8,000 years ago in the fauna preprint; earliest occupation phase, accumulating over centuries | Main Maltese assemblage. Its published phase age is not a direct date on every deer or a single visit. [S323 · Mata-González et al. (2026)] |
| Uzzo transition deposits | Trench F, spits 14–11; published overview approximately 8,770–7,850 cal BP | Best overlapping Sicilian phase in this comparison. Cultural classification is debated; the name does not make every occupant a farmer. [S325 · Mannino et al. (2015)] |
| Uzzo earlier Mesolithic | Trench F, spits 22–15; approximately 11,110–8,500 cal BP | Earlier baseline, not a fully contemporary hunting sample. [S325 · Mannino et al. (2015)] |
| Uzzo Neolithic I | Trench F, spits 10–6; approximately 8,050–7,400 cal BP | Later comparison, overlapping only the late edge of the Latnija Phase V overview. Domestic animals change the denominator of faunal percentages. [S325 · Mannino et al. (2015); S332 · Tagliacozzo (2005)] |
These rounded envelopes combine uncertainty in phase boundaries; they are not single specimen confidence intervals or proof of continuous residence. Uzzo’s transition start is modeled at 8,770–8,500 and its end at 8,050–7,850 cal BP. No new calibration was performed. Latnija’s unresolved chronology inputs remain footnoted in the occupation audit.1
Two tempting alternatives are less useful for this particular match. Fontana Nuova is geographically closer to a possible southern departure coast, but its collection is unstratified and the youngest overlapping dated specimen is wild boar, not a matched deer assemblage. Grotta d’Oriente’s main archaeological snapshots are around 14.2, 9.7–9.6 and 7.8 cal ka BP; its marine-resource study therefore does not supply a securely matched Phase V deer mortality sample. Neither is evidence for a specific Malta-bound community. [S326; S333, archaeological setting]
What the deer actually tell us
| Question | Latnija Phase V | Sicilian evidence and comparison limit |
|---|---|---|
| Were deer used for food? | Deer/ungulate remains occur with burning and processing evidence. This supports food use, although many surfaces are damaged. | Deer and wild boar were important at Uzzo. The transition-phase summary reports a deer minimum number of individuals of 38; this is an assemblage estimate, not annual kills or herd size. [S323; S331, Table 5] |
| Were hunters choosing particular ages? | The restricted Dp4–M3 dental profile has 2 juvenile, 5 prime and 1 old observation. The broader pool has 9, 41 and 9, but combines tooth types and can count an animal repeatedly. The small restricted sample leaves considerable uncertainty. | A comparable primary, phase-specific dental count table was not recovered for Uzzo. Claims about selective hunting cannot yet be tested between the sites using equivalent age definitions and independent individuals.2 |
| How much meat did each animal offer? | Published allometric body-mass estimate: about 43 kg, from 30 measurements, not necessarily 30 separate animals. Body mass is not edible meat yield. | Latnija’s statistical size comparison uses Late Pleistocene Sicily, not contemporary Uzzo. Uzzo’s separate 185-kg illustrative weight comes from a general reference and is explicitly considered an overestimate for its small deer. Dividing these values would create a misleading yield ratio. [S323, Table 2 and Fig. 6; S331, Table 5] |
| Were whole carcasses brought back? | Head, trunk and limb elements are represented. The authors favor whole-carcass deposition, but an input inconsistency makes the numerical model provisional.3 | Comparable Uzzo element counts, density controls and transport analysis remain missing from this review. Whole-carcass transport would be compatible with small prey or nearby kills; it does not measure abundance. |
| Was marrow extraction unusually intensive? | Percussion evidence supports some processing. Yet most assessed long-bone fractures are dry fractures, so fragmentation alone cannot be counted as marrow extraction. | No matched Uzzo rate with the same inspected-surface denominator was recovered. At Latnija, the supplementary table separates two percussion impacts, two potential impacts and eight potential cut marks; its peeling observations concern birds, not deer. [S323; S330, Tables 9–10] |
| Did deer decline under hunting pressure? | A centuries-long aggregate cannot supply a trajectory of hunting returns. Small body size and persistence are not measurements of hunting pressure. | The Uzzo analyst describes deer hunting continuing through the Early Neolithic and gradually giving way to sheep/goat herding. A falling deer percentage can reflect more domestic animals or other foods without demonstrating declining deer stocks. [S332, pp. 435–436] |
The Latnija age profile is compatible with human hunting, but cannot establish a dense, easy-to-catch population. Likewise, the foot-bone fusion counts primarily help distinguish younger animals from subadults and older animals; they should not become a precise count of prime adults. The published 59-tooth pool is not a census of 59 deer. [S323, Fig. 5 and methods; S330, Table 21]
Preservation changes the comparison
Latnija’s recovery methods included plotting remains over 20 mm and wet-sieving sediment fractions down to 0.5 mm. ZooMS sampling prioritized dense fragments, mostly from the 2-mm fraction. Morphological identifications, selected ZooMS identifications and spatial observations consequently have different sampling frames. [S323, methods]
In the supplementary table, 832 of 1,914 specimens have less than half their surface readable. Modification assessment excludes these and teeth/tooth fragments. The reported corrosion and cut/percussion percentages therefore do not use the entire collection as their denominator, and are not deer-only rates. Published skeletal survival estimates also depend on density and utility proxies from other deer taxa. A low observed mark rate cannot establish little butchery, and an absent fragile bone cannot establish selective transport. [S323; S330, Tables 9, 17–20]
The public spatial CSV provides specimen labels, contexts, coordinates, elements and overlapping taxonomic/taphonomic categories. It offers a route back to context, but its rows are not independent animals or even a deduplicated specimen inventory. It does not provide the tooth-age and wear-stage fields needed to place the mortality observations into separate episodes. [S330, code archive]
New hypotheses worth testing
A predictable stop within a wider territory. Smaller animals could be easier to carry whole, while fresh water and coastal foods could make the surrounding area useful even when a hunt failed. This combines a transport hypothesis with the discovery paper’s local environmental setting; it is not a demonstrated reason for the first crossing. Test whether distinct secure contexts repeatedly combine deer processing with independently identified harvest seasons. [S80; S106; S323]
A reserve when another coast was less rewarding. This requires evidence that resources in Malta and a candidate Sicilian region failed at different times. More Maltese bones, or climate change somewhere in Sicily, would not show that complementarity. The strongest result would be recurring Maltese harvest opportunities during a demonstrated seasonal gap at the comparison locality.
Different ways of using the same prey. Similar dependence on deer could coexist with different mobility: small carcasses processed near Latnija versus selected parts brought to another site. Test equivalent skeletal elements and tooth categories in overlapping contexts, then control for bone density, surface visibility, sediment volume and recovery methods. Such a difference would identify behavior before it identifies motive or origin.
The next practical step is a context-linked deer and seasonality ledger: specimen ID, element/side, tooth and wear stage, age-class definition, marks and surface visibility, recovery fraction, context and date basis. Build Latnija entries only where actual specimen joins are possible, and obtain the original Uzzo tables before filling its missing age and transport cells. Compare rates only after matching denominators. If those data remain inaccessible, retain the empty cells and advance the independently available shell-seasonality comparison; do not replace them with values quoted in later summaries.
Sources and qualifications
- S323: Mata-González et al. 2026, preprint v1, especially methods, Table 2 and Figures 4–6. No journal successor verified.
- S330: The same preprint’s supplementary tables and code/data archive. These are dependent supporting material, not an independent replication.
- S325 / S331: Mannino et al. 2015 and supplement, including Table 5 and chronology.
- S332: Tagliacozzo 2005, Animal exploitation in the Early Neolithic in Central-Southern Italy, pp. 435–436 and Tables 5–6. Used for the later comparison, not transferred to earlier foragers.
- S333: Colonese et al. 2018, Grotta d’Oriente, author manuscript, archaeological setting. Used to assess chronological fit.
- Earlier records: discovery S80, Fontana Nuova S326, and the occupation audit.
Footnotes
-
LAT-DATE-001 and LAT-DATE-002 remain unresolved table/model disagreements. The published phase overviews are retained without new calibration. Any revised first-arrival chronology would require revisiting the overlap assessment. ↩
-
Tagliacozzo’s 1993 Uzzo monograph and the original Accordia article’s mortality/body-part tables were not accessible in full during this comparison. The publisher lists the article in volume 5 (1994), pp. 7–38; later sources vary in year and pagination. The often-repeated claim of hunting young adults aged 3–5 is retained only as a lead in the structured record, not a verified matched dataset. This gap does not negate the better-supported evidence of deer exploitation at Uzzo. ↩
-
LAT-DEER-INPUT-001: Supplementary Table 16 gives tibia MAU 4.00 and a maximum MAU of 5.50, but prints 36.37% rather than the approximately 72.73% implied by those entries. The distributed
Bayesian_SP.Ruses 36.37 as its fifteenth input, so the inconsistency reaches the supplied model code. Source values are preserved; no replacement or model rerun was performed. The whole-carcass interpretation remains plausible from broader skeletal representation, but the published model outputs, including estimated survival, are provisional. Separate count inconsistencies in Table 1 and Supplementary Table 21 are preserved in the structured record and excluded from quantitative comparisons. ↩