Absolute dating of the European Neolithic using the 5259 BC rapid $^{14}$C excursion

The study addresses the need for precise, absolute dating of Neolithic settlements in the Balkans and wider Eastern Mediterranean to clarify neolithisation processes, settlement durations, and environmental impacts. Conventional dendrochronology in the region lacks continuous master chronologies extending back to the 6th millennium BC, and radiocarbon dating often relies on charcoal with potential inbuilt age, yielding broad calibrated ranges and chronological inconsistencies. Recent discovery of rapid, short-term atmospheric 14C spikes (Miyake events) recorded in annually dated tree rings provides absolute year tie-points. The authors aim to use the 5259 BC Miyake event to anchor a floating tree-ring chronology from the waterlogged Neolithic lakeside site of Dispilio (Lake Kastoria, Northern Greece) to exact calendar years, thereby establishing the first absolute, calendar-year dates for a European Neolithic site and providing a regional chronological anchor.

• Dendrochronology allows annual-resolution dating but typically depends on continuous regional master chronologies, which are rare and geographically limited. Several long tree-ring records exist in the Eastern Mediterranean, but they do not extend to 7500 years BP, leaving many prehistoric chronologies only approximately placed by conventional 14C wiggle-matching.
• Annual 14C datasets have identified abrupt Δ14C spikes (Miyake/SEP events) at AD 774/775, AD 993/994, 660 BC, and in the early Holocene at 7176 BC and 5259 BC. These events enable absolute annual dating of wooden materials when detected in single-year 14C series.
• Prior absolute Miyake-event dating has been demonstrated mainly for sites in the Common Era; prehistoric European applications have been lacking. Previous work at wetland sites in the southwestern Balkans produced floating chronologies dated approximately by 14C.
• The Dispilio site is archaeologically significant with extensive waterlogged wooden remains and rich material culture; earlier radiocarbon (mostly charcoal) suggested occupation from later Middle Neolithic (~5600 cal BC) through Bronze Age (~2100 cal BC), but lacked precise anchoring due to inbuilt age and calibration uncertainties.
• IntCal20 shows sparse and block-averaged data in the 53rd–52nd centuries BC, which can yield misleading wiggle-matching outcomes for annual datasets. Incorporating annually resolved reference data (e.g., BR22) can improve calibration precision around the 5259 BC event.

• Fieldwork and sampling: In 2019, 787 whole cross-section discs were collected from wooden piles at Dispilio’s East Sector. Wood taxa were identified anatomically. Tree-ring widths (TRW) were measured (0.01 mm precision), two to four radii averaged per sample, and chronologies constructed and cross-dated.
• Dendrochronology: 23% of samples cross-dated into two master TRW chronologies: juniper (Juniperus spp.) 303 years (118 series; mean segment 86 years; mean inter-series r=0.62), and oak (Quercus subgenus Quercus spp.) 120 years (58 series; mean segment 66 years; mean inter-series r=0.51). Waney edges preserved mostly in juniper; felling season inferred as late summer/autumn to early spring based on latewood terminations. Juniper and oak chronologies were cross-dated over a 108-year overlap (t-values 4.95–5.1; GLK 63%). Conventional 14C wiggle-matching supported the cross-date for oak.
• Targeting the Miyake event: A preliminary wiggle-match of 11 14C dates from the juniper chronology against IntCal20 provided a coarse end-date range (5233–5137 cal BC, 95%), guiding selection of rings likely spanning 5259 BC. Four juniper samples (DISP-10206, -10611, -10070, -10063) were dissected for annual 14C around the predicted event.
• Single-year 14C measurements: 115 individual ring 14C measurements were made. At LARA (University of Bern), cellulose extraction used BABAB with modifications; measurements on MICADAS with standards and blanks; an identified oxalic acid II standard batch offset (+30.9 ± 3.2 14C years) was corrected (−4.2‰ shift) via intercomparison; repeated subset ensured consistency. At ETH Zürich, modified cellulose preparation, holo-cellulose freeze-dried, graphitised with AGE-3, measured on MICADAS with OX1/OX2 standards and blanks.
• Event identification and modelling: Δ14C year-to-year increase was assessed sensu Miyake (difference between 5260 and 5259 BC). A χ2 goodness-of-fit wiggle-matching against a mean annual reference curve derived from Brehm et al. 2022 (BR22) was used to determine optimal waney-edge placement. Annual datasets were also wiggle-matched in OxCal 4.4 (D_Sequence) against IntCal20 and a modified IntCal20plus (replacing ca. 82 years around 5259 BC with the annual BR22 average).
• Site plan analysis: Dendrodated piles with waney edges within 1–2 years were mapped in GIS to infer felling phases and possible structure outlines, considering spatial clustering and orientation relative to the present shoreline.

• Detection of the 5259 BC Miyake event: All four juniper samples recorded a sharp Δ14C increase in the same cross-dated ring (relative year 184 of the juniper chronology). The mean year-to-year increase was −15.8‰ Δ14C, with sample-level increases of −11.1‰ (DISP-10070), −13.1‰ (DISP-10206), −14.8‰ (DISP-10063), and −18.6‰ (DISP-10611).
• Absolute dating: χ2 fits to the BR22 annual reference placed end-dates at 5240 BC (DISP-10070, DISP-10063), 5153 BC (DISP-10206), and 5155 BC (DISP-10611), consistent with cross-dated positions. The juniper master chronology’s last ring is 5140 BC; the cross-dated oak chronology’s last ring is 5311 BC.
• Occupation/construction span: Felling dates range from 5328 to 5140 BC, indicating at least 188 years of construction activity. Oak sapwood reconstructions may extend the start by ~30 years. Distinct felling phases suggest rebuilding cycles of ~29 years (west) and 35–37 years (east), aligning with independent estimates of Neolithic house lifespans.
• Calibration performance: OxCal wiggle-matching against IntCal20 failed to include true felling years within 95% ranges for the annual datasets (short series biased younger by >40 cal years; long series biased older by ~15–20 cal years), reflecting poor IntCal20 replication in the 53rd–52nd centuries BC. Using IntCal20plus (with BR22 annual data) yielded accurate and more precise end-date ranges.
• Growth and environment: Both juniper and oak TRW show a ~20-year growth suppression ca. 5360–5340 BC, potentially reflecting decreased precipitation; may correlate with short-term eutrophication signals in Lake Kastoria’s paleoecology.
• Cross-dating potential: The absolute, calendar-anchored juniper and oak chronologies from Dispilio can serve as reference datasets for dendro-dating contemporaneous wetland sites within ~100 km, facilitating regional chronological harmonization.

Anchoring the Dispilio tree-ring chronology to the 5259 BC Miyake event resolves longstanding uncertainties in the chronology of a key Neolithic lakeside settlement. The absolute placement of building phases between 5328 and 5140 BC refines regional periodisation of the later Middle to Late Neolithic and provides an independent check on settlement durations typically inferred from modelled radiocarbon sequences. The results demonstrate that Miyake events are powerful tie-points for prehistoric European contexts, overcoming the absence of long regional master chronologies and calibration curve limitations around the 53rd–52nd centuries BC. The precisely dated felling phases and intervals between construction events are consistent with expected Neolithic house lifespans and suggest patterns of episodic construction, possible timber storage, and potentially nonperennial occupation. Environmentally, preliminary TRW signals indicate multi-year moisture stress episodes, showing how absolutely dated tree rings from archaeological wood can inform high-resolution climate–society interactions. The study also highlights the need to integrate annual 14C datasets into calibration curves to improve wiggle-matching accuracy in sparsely constrained intervals.

This study provides the first absolute, calendar-year dating of a European Neolithic site using a Holocene Miyake event. By detecting the 5259 BC Δ14C spike in juniper tree rings from Dispilio and cross-dating associated oak, the team anchored a 303-year juniper chronology (ending 5140 BC) and a 120-year oak chronology (ending 5311 BC). The precisely dated felling phases document at least 188 years of construction activity and offer a robust chronological framework to refine Neolithic periodisation and settlement histories in the southern Balkans. The Dispilio chronologies now serve as regional references for dendrochronological cross-dating at nearby wetland sites, potentially extending absolute dendrochronologies deeper into prehistory. Future work should: (1) cross-date additional regional sites to expand the absolute dendro network; (2) produce more annual 14C series to enhance calibration curve fidelity around 5259 BC and other SEP events; (3) integrate dendroclimatic indicators from archaeological wood with paleoenvironmental records to reconstruct local hydroclimate and assess links to societal change.

• Calibration curve sparsity: IntCal20 is poorly replicated in the 53rd–52nd centuries BC (dominated by multi-year blocks), causing inaccurate wiggle-matching of annual data without supplementary annual reference datasets.
• Wood anatomy/species resolution: Species-level identification was not possible for junipers and deciduous oaks; functional grouping may mask ecological nuances.
• Tree-ring anomalies: Juniper can have false and missing rings due to lobate growth; mitigated by abundant full cross-sections and cross-dating with oak, but residual uncertainty is possible.
• Oak sapwood preservation: Limited waney edges and sapwood loss in oak hinder exact felling dates for some elements; reconstructions add uncertainty.
• Site-scale inference: The mapped outlines of felling-phase clusters suggest structures, but interpretations are tentative and constrained by excavation extent and uncertain Neolithic shoreline position.
• Occupation continuity: The 188-year span of felling dates indicates activity but not necessarily continuous occupation; gaps could reflect preservation biases or true hiatuses.