Human response to the Younger Dryas along the southern North Sea basin, Northwest Europe

Abstract

Currently in NW Europe little is known about the human response to the extensive cold reversal at the end of the Pleistocene, the Younger Dryas (ca. 12,850 till ca. 11,650 cal BP), mainly due to the poor chronological resolution of the archaeological sites belonging to the Ahrensburgian Culture. Here we present a series of 33 radiocarbon dates performed on the seminal cave site of Remouchamps, situated in the Belgian Meuse basin. Combined with a revision of the available radiocarbon evidence along the southern North Sea basin (Belgium, southern Netherlands, western Germany), it is suggested that the first half of the Younger Dryas, characterized as extremely cold and wet, faced a significant population reduction. Repopulation started around the middle of the Younger Dryas, from ca. 12,200 cal BP onward, probably in response to a slight climatic improvement leading to somewhat warmer summers. This might be considered a prelude to the subsequent population boost of the Early Holocene (Mesolithic).

Introduction

The Younger Dryas or GS-1 represents a cooling event at the end of the Pleistocene, which lasted over 1200 years, from ca. 12,850 till ca. 11,650 cal BP^1,2. It is generally assumed that this major climatic event, most likely caused by meltwater outbursts into the Atlantic Ocean and a resulting decrease in the North Atlantic thermohaline circulation^3,4, had a considerable but variable impact on contemporaneous hunter-gatherer societies globally^5,6. This particularly holds for Western and Northern Europe, occupied during the Younger Dryas by the Ahrensburgian Culture, the most western tradition of the Tanged Point Groups or Complex (TPC) (Stielspitzen-Gruppen)⁷. Sites of the Ahrensburgian Culture are found from the Belgian Meuse basin in the (south)west to the lowlands of NE Germany and Poland^7–13 (Fig. 1).

Figure 1.

(A) Distribution map of the Ahrensburgian Culture; (B) The Belgian-Dutch Meuse basin with indication of Remouchamps (star) and Ahrensburgian Culture sites mentioned in the paper. 1. Coléoptère; 2. Zonhoven; 3. Eersel; 4. Geldrop; 5. Budel; 6. Kartstein.

Except for NE Germany and Poland, in most regions of Western and Northern Europe a sharp decrease of archaeological sites is observed compared to the Allerød, the warm phase preceding the Younger Dryas¹⁴. In most regions, the number of sites reduces by roughly half. This is generally interpreted as an indication of a drastic population decline and/or a (north)eastward population movement in response to the long-term severe cooling of the Younger Dryas. However, for the moment, our understanding of the human responses to this long-term climate deterioration and the climatic variability that occurred within the Younger Dryas is severely hampered by the fact that we lack a fine-grained chronology for the Ahrensburgian culture^9,10,13. Except for the renowned kettle hole site of Stellmoor in N Germany with up to 26 radiocarbon dates¹³, very few sites have yielded reliable dates, mostly limited to one or two dates per site. The vast majority of Ahrensburgian sites generally do not contain organic dating materials, such as animal and human bone or charcoal intimately associated with occupation remains and anthropogenic features (e.g. structured hearths). Hence they can only be assigned to the Younger Dryas (and eventually the start of the Preboreal) based on the presence of specific armature types (i.e. tanged points and microliths, mainly consisting of obliquely truncated points) and/or standardized blade and bladelet productions, typical of the Ahrensburgian culture^15–18.

In the framework of a large interdisciplinary research project, called ROAM (A Regional Outlook on Ancient Migration), the Ahrensburgian cave site of Remouchamps in the Belgian Meuse basin¹⁹ is currently being re-evaluated. Besides ongoing reanalyses of the faunal and human remains and the lithic industry, an extensive dating program has been conducted aimed at refining the chronology of its occupation(s). The site of Remouchamps can be considered a key-site for research into the Ahrensburgian Culture in NW Europa^19,20, thanks to its rich faunal assemblage dominated by reindeer²¹, the presence of perforated tertiary shells, ochre-covered lithic artefacts and engraved bone objects^19,22. Above all the site is famous because of a radiocarbon date situating the earliest occupation at the Allerød/Younger Dryas transition²³ (Table 1), making it one of the oldest Ahrensburgian contexts in NW Europe. However, the reliability and meaning of this early date have been questioned by some scholars¹⁰, since three other radiocarbon dates from Remouchamps yielded much younger results, placing the main occupation in the later phase of the Younger Dryas. Lastly, Remouchamps is also known as one of the few Ahrensburgian sites in NW Europe which yielded human remains and the only site on which the latter were found in context¹⁰. However, their cultural attribution so far was solely based on their stratigraphical position, hence the need for an assessment by means of a radiocarbon date.

Table 1.

List of radiocarbon dates performed in the late twentieth century^23,29.

Lab code	BP date	St. dev	95.4% Probability (cal BP) (IntCal20)	C:N	Dated species	Dated fragment	Human traces	Excavation	References
OxA-4191	10800	110	13056–12505	3.4	R. tarandus	Metacarpal (R)	Cut marks	Rahir 1902	23
Lv-535	10380	170	12724–11622		R. tarandus	Bone fragments		Dewez 1969–1970	29
OxA-4190	10330	110	12613–11751		Tetrao urogallus	Proximal humerus	Cut marks	Rahir 1902	23
OxA-3634	10320	80	12480–11825		R. tarandus	Maxilla	Cut marks	Dewez 1969–1970	29

In order to deal with these dating issues, it was decided to set up an extensive dating program, the results of which will be presented in this paper. Subsequently, the current dating evidence available for the Ahrensburgian culture along the southern North Sea basin (Belgium, the southern Netherlands and western Germany) will be discussed, as well as the new insights this provides concerning the occupation history of the region during the Younger Dryas.

The cave of Remouchamps is located in Southern Belgium (province of Liège) on the right bank of the Amblève river, ca. 9 km upstream of its confluence with the Ourthe, one of the major tributaries of the Meuse (Fig. 1). The extensive cave system was formed by the passage of the Vallon des Chantoirs through the Givetian and Frasnian limestone bedrock and consists of a dry upper gallery and a lower gallery through which an underground river flows¹⁹ (Fig. 2).The first systematic excavations at the site were held by Rahir and Van den Broeck in 1902^24,25. In the entrance hall of the upper gallery at a depth of ca. 50 cm underneath a loam layer of variable thickness, they found the remains of a “dark and charcoal-rich layer” containing besides numerous lithic artefacts (ca. 5000 pieces) and faunal remains among others, two charcoal patches identified as hearth-remnants.

Figure 2.

Excavated sectors and stratigraphy within the cave of Remouchamps (based on^19,24,25). 1. Plan of the first part of the cave with upper and lower gallery; 2. Entrance hall with position of the Dewez’s excavation trenches (green) and profiles; area excavated by Rahir and Van den Broeck (purple) and disturbed areas by nineteenth and twentieth century activity in the cave (rastered); 3 & 4: stratigraphy consisting of (0) Disturbed top layer; (1) Homogenous clayish loam lens; (1') Clayish loam with angular limestone debris; (2 and 2') Discontinuous brown/black sandy loam layer or lens (= artefact-bearing horizon), charcoal rich and associated with a stalagmite crust; (3) "Rose" loam layer with small limestone blocks; (4) Large limestone block layer with limited interstitial sediment composed of sandy/loam sediments; (5) Similar to 4, but the limestone blocks are smaller and more numerous.

In 1969 and 1970, new excavations were undertaken by Dewez and colleagues who dug five trenches in the entrance hall (trench A, B, C, D and DD) and four test-pits in the rest of the cave (E–H) (Fig. 2). Other than precisely locating the Rahir/Van den Broeck excavations and confirming the fact that the archaeological finds were limited to a single, homogenous artefact-bearing layer, they also provided much needed additional stratigraphic information¹⁹. Based on their report, the Ahrensburgian occupations seem to be essentially associated with a discontinuous charcoal-rich brown sand layer with at places thick charcoal lenses and calcitic concretions. In the eastern part of the site (trench D and DD), this layer is found at a depth of 20 cm more or less directly beneath a layer of recent (nineteenth and twentieth century) backfill and on top of a layer of massive limestones with limited interstitial sediment. Towards the western part of the site (Trench A), however, the Ahrensburgian level seemingly slopes and is located at a depth of ca. 50 cm, at the base of a clayey loam layer with small limestone blocks. A second loam layer separates the prehistoric finds found there, from the underlying layer of massive blocks. The discrepancies between the profiles in the western and eastern part of the site seem to suggest that the Ahrensburgian levels were partially truncated or affected by slope erosion in the eastern part of the site. Finally, from the account of Rahir²⁵, it is clear that the human remains, shells and lithics found in a fissure in the eastern wall were collected from the same stratigraphic position as the Ahrensburgian remains.

Results

Altogether 33 bone samples have been radiocarbon dated (Table 2), among which 31 newly selected samples and two older samples that have been redated. Combined with the four dates obtained in the twentieth century (Table 1) this makes 37 radiocarbon dates for the site of Remouchamps. All collagen samples meet the quality criteria for radiocarbon dating (Table 2). Collagen yields are above 4% and the atomic C:N values range between 2.9–3.6. After screening (cf. Methods), 30 radiocarbon dates were retained as belonging to the Ahrensburgian occupation(s) of Remouchamps.

Table 2.

List of radiocarbon dates performed on animal and human bones from the Remouchamps cave.

Excavation sector	REM ID	Taxon	Fragment	Element	Taxon	Human modification	C:N	14C lab ID	BP date	St. dev	95.4% Probability (cal BP) (IntCal20)
D	REM 69.1	V. vulpes	Mandible/Tooth	MAN/TTH (L)	Fox	–	3.3	RICH-34321	12958	36	15646–15320
DD	REM 238	R. rupicapra	Tibia	TIB (L)	Chamois	Cut marks, percussion and green fracturation	3.3	RICH-34314	10364	31	12470–12001
D	REM 28	E. caballus	Tooth	UP3/4 (R)	Horse	–	3.2	RICH-34327	10330	31	12458–11945
D	REM 10050	R. rupicapra	Humerus	HUM (L)	Chamois	Cut marks and green fracturation	3.2	RICH-34332	10308	31	12454–11936
A	REM 10006	R. tarandus	Tooth	LM3	Reindeer	–	3.1	RICH-34317	10293	32	12447–11835
D	REM 10052	R. tarandus	Metapodial	METAP (L)	Reindeer	Cut marks and green fracturation	3.2	RICH-34324	10290	31	12443–11832
A	REM 10,043	R. rupicapra	Radius	RAD (L)	Chamois	Possible cut marks	3.0	RICH-34334	10275	31	12431–11826
D	REM 10	E. caballus	Mandible	MAN _ LP3/4 (L)	Horse	Cut marks and percussion	3.2	RICH-34330	10258	31	12421–11822
D	REM 312	R. tarandus	Metatarsal	MT (R)	Reindeer	Green fracturation	3.2	RICH-34323	10255	31	12420–11821
DD	REM 231	R. tarandus	Metacarpal	MC	Reindeer	Percussion and green fracturation	3.2	RICH-34311	10234	31	12040–11764
A	REM 243–1	Cervidae	Tooth	UM (R)	Cervid	–	3.2	RICH-34325	10216	31	11993–11758
DD	REM 225	R. tarandus	Metacarpal	MC (L)	Reindeer	Cut marks and green fracturation	3.3	RICH-34309	10212	35	11996–11752
D	REM 301	R. tarandus	Metapodial	METAP	Reindeer	NA	3.1	RICH-34319	10,209	31	11969–11755
A	REM 204	R. tarandus	Tooth	LM3 (R)	Reindeer	–	3.4	RICH-34307	10200	31	11949–11750
DD	REM 235	R. tarandus	Metacarpal	MC	Reindeer	Green fracturation	3.2	RICH-34306	10197	34	11967–11746
DD	REM 223	R. tarandus	Metacarpal	MC	Reindeer	Percussion and green fracturation	3.4	RICH-34322	10191	40	11972–11653
A	REM 388	Lagopus sp.	Humerus	HUM (R)	Snow grouse	Cut marks	3.3	RICH-34329	10170	31	11939–11653
DD	REM 230–1	R. rupicapra	Tibia	TIB (L)	Chamois	Cut marks and green fracturation	3.1	RICH-34308	10139	31	11932–11510
A	REM 10007	R. tarandus	Tooth	LM3 (L)	Reindeer	–	3.0	RICH-34312	10118	31	11876–11406
D	REM 10068	R. tarandus	Tooth	LD4 (L)	Reindeer			RICH-34333	10116	31	11874–11406
Redate OxA-4191	New sample	R. tarandus	NA		Reindeer	Cut marks	3.3	RICH-35064	10113	37	11878–11403
A	REM 10012	R. tarandus	Tooth	LM3 (L)	Reindeer	–	3.1	RICH-34326	10109	31	11835–11404
A	REM 10111	R. tarandus	Humerus	HUM (L)	Reindeer	Cut marks, percussion and green fracturation	3.2	RICH-34310	10103	31	11828–11403
Redate OxA-4190	New sample	Tetrao urogallus	Humerus		Wood grouse	Cut marks	3.4	RICH-35063	10091	39	11822–11402
D	REM 214	R. tarandus	Tooth	LD4 (L)	Reindeer	–	3.2	RICH-34328	10058	30	11804–11398
A	REM 10040	B. primigenius	Metacarpal	MC	Auroch	Cut marks and green fracturation	3.2	RICH-34318	10032	31	11738–11344
A	REM 10039	R. tarandus	Humerus	HUM (L)	Reindeer	Cut marks and green fracturation	3.2	RICH-34313	10018	30	11709–11321
D	REM 82	C. lupus	Mandible	MAN (R)	Wolf	–	3.1	RICH-34320	10018	30	11709–11321
DD	REM 256	Lepus sp.	Femur	FEM (R)	Hare	Cut marks and green fracturation	3.4	RICH-34315	9917	30	11401–11239
D	REM 10072	F. silvestris	Mandible/Tooth	MAN/TTH	Cat	–	3.2	RICH-34316	9182	30	10486–10245
		Human	Tooth			–	3.3	RICH-35065	8980	36	10235–9920
A	REM 10026	C. fiber	Tooth	UPM (L)	Beaver	–	3.1	RICH-34331	8264	28	9409–9127
D	REM 199	S. scrofa	Tooth	LI2 (R)	Wild boar	–	3.2	RICH-34335	369	22	495–319

Redating of the reindeer metacarpus with cut marks, previously dated (OxA-4191) to the Allerød-YD transition, yielded a significantly younger result (RICH-35064). Using the C:N ratio as an indicator of the collagen quality, it is clear that the latter date is more reliable. Well-preserved, uncontaminated fresh bone collagen has a C:N ratio of 3.1–3.2²⁶. The more a C:N ratio of archaeological bone collagen approaches these values, the less it is contaminated, since an increase of the C:N ratio with 0.1 means the presence of 2% exogenous carbon. The C:N ratio of RICH-35064 is 3.3 and 3.4 for OxA-4191, implying that the collagen of OxA-4191 is more contaminated.

This may be due to the pre-treatment methods used in the ’90s, which consisted in washing with 0.1 M NaOH. Nowadays, at least in the RICH-laboratory²⁷, 0.25 M NaOH is standardly used, which is a stronger alkali that should remove all exogenous humic contaminants. At first sight it may seem strange that the better cleaned sample (RICH-35064) yielded a younger age compared to the Oxford sample, as it may be expected that the former contains lesser modern carbon contaminants. However, in literature there are rare examples of bone samples which yielded younger ages after improving the sample quality by using nanofiltration^26,28. One of the possible explanations is contamination with old carbon from the environment or another so far unknown source, which might not have been fully removed with the 0.1 M NaOH in the old Oxford sample.

In sum, as the newly obtained date (RICH-35064) is perfectly in line with the other dates from the site, was pre-treated according to the current high(er) standards and has a reliable C:N ratio (3.3), it seems very likely that this date is the most acceptable one. This means that the existence of an earlier Ahrensburgian occupation phase at the Allerød-Younger Dryas transition could not be confirmed, and Remouchamps should no longer be considered as one of the oldest Ahrensburgian sites of NW Europe.

The remaining radiocarbon dates all fall within the limits of the Younger Dryas and the transition to the Early Holocene. More specifically the calibrated dates (Tables 1 and 2) refer to a start of the occupation(s) at ca. 12,700/12,500 and an end at ca. 11,400/11,200 cal BP, corresponding to most of the YD and the early Preboreal. A further refinement of this chronology is seriously hampered by the large standard deviation of the two previously dated samples which have not been redated (OxA-3634; Lv-535; Table 1), ranging between 80 and 170 BP-years, but above all by the presence of two important plateaus in the calibration curve. The first is situated between ca. 10,400 and 10,300 uncal BP while the second one runs from ca. 10,100 to 10,000 uncal BP. As a result, after calibration numerous dates span a time period of up to 400–600 calendar years, hindering the development of a fine chronology for the Ahrensburgian occupation at Remouchamps in particular and the Ahrensburgian culture in general.

To address this problem, a Bayesian phase-model (Fig. 3; Table SI1) was constructed, yielding an overall agreement index of 69.1%. This model allows to narrow down the chronology of the occupation substantially. Based on it, the Ahrensburgian occupation(s) are situated between ca. 12,180/11,990 cal BP (mean 12,080 cal BP) and 11,645/11,360 cal BP (mean 11,540 cal BP). This covers the second half of the YD and the very start of the Holocene. A further refinement can be obtained if date RICH-34315, which yielded a very poor agreement (8.3%), is omitted from the model. In this new model (Fig. and Table SI2), yielding a higher overall agreement (94.6%), the end date of the occupation is moved to ca. 11,715/11,505 cal BP (mean 11,620 cal BP) which coincides with the very end of the YD.

Figure 3.

Bayesian Phase model of the Remouchamps dates compared to the NGRIP climate curve².

The radiocarbon date obtained on a human bone (metacarpus), which most likely belongs to an intentionally interred individual (burial), proves that it cannot be attributed to the Ahrensburgian Culture but rather belongs to the Early Mesolithic (Table 2). The calibrated date, ranging between 10,235 and 9920 cal BP, fits perfectly with other dates from cave burials in the Meuse basin, the vast majority of which are attributed to the early (Preboreal-Boreal) Mesolithic³⁰. That being said, the burial from Remouchamps is the first to have been discovered in the Ourthe/Amblève region.

Discussion

Human occupation during the late Allerød/early YD?

From the above, it is clear that the Ahrensburgian presence at Remouchamps is temporally more homogenous than initially thought. In addition, as the oldest radiocarbon date definitely needs to be rejected, there are currently only very few Ahrensburgian Culture sites within NW Europe which can be reliably dated to the Allerød-YD transition or early YD^10,13. While that is true, along the southern North Sea basin, several open-air sites situated in the coversand area of the Meuse basin, including Budel IV, Zonhoven-Molenheide, Geldrop 1, Geldrop A2 and Geldrop-Mie Peels (Fig. 1), yielded very early radiocarbon dates (Table 3; Fig. 4). Although some scholars accept these dates claiming that these sites are amongst the oldest Ahrensburgian sites within NW Europe^31,32, other researchers have questioned (most or some of) these early dates based on the fact that they have been performed on charcoal from open-air contexts^9,10,33. These samples are not always entirely reliable as some may result from natural wildfires and eroded contexts. This certainly holds for the very old dates of Budel IV (GrN-1687) and Geldrop 1 (GrN-1059), which were most likely performed on charcoal eroded from the underlying Usselo soil³³. This soil has been found at many spots throughout NW Europe and northern central Europe (there called Finow soils) (for an overview see³⁴), and is characterized by the presence of numerous charcoal fragments mainly from Scotch pine (Pinus sylvestris). This charcoal has been produced by repeated and large-scale forest fires which occurred over a long time-period spanning the late Allerød and early YD^34,35. Locally this soil has been eroded in the subsequent YD, leading to the deposition of Usselo charcoal in younger aeolian sediments. This most likely was the case at Geldrop 1 since the dated charcoal was retrieved in a secondary position, i.e. from the top of a thin layer of drift sand³³. In addition the charcoal date is not at all compatible with a date obtained on cremated bones from the same site, which is markedly younger (cf below).

Table 3.

List of radiocarbon dates from different Ahrensburgian sites along the southern North Sea basin. Italic: Holocene dates that are considered as outliers.

SITE	Context	Feature	Dating material	Lab code	14C BP date	St. dev	95.4% Probability (cal BP) (IntCal20)	Tanged points	References
Bone dates
Geldrop 1	Open-air	Hearth-pit	Calcined bone fragments	GrA-15177	10500	70	12694–12101	23%	33
Kartstein	Cave		R. tarandus, femur	OxA-9031	10220	75	12459–11507	30%	10
Coléoptère	Cave	Layer 6b	Lepus sp.	RICH-34338	10196	32	11946–11749	?	This paper
Geldrop 3–1	Open-air		Calcined bone fragments	GrA-15181	10190	60	12423–11407	20%	33
Remouchamps	Cave			R_Combine	10175	6	11933–11752	22%	This paper
Geldrop 3–2 Oost	Open-air		Calcined bone fragments	GrA-15182	9970	60	11701–11245	0.7%	33
Eersel-Panberg 436–207	Open-air		Calcined bone fragments	GrA-15175	9810	70	11596–10881	0	33
Geldrop 3–2 Oost	Open-air		Calcined bone fragments	GrA-15183	8800	60	10150–9559	0,7%	33
Charcoal dates
Budel IV	Open-air		Charcoal fragments	GrN-1687	11070	90	13156–12769	?	68
Geldrop 1	Open-air		Charcoal patch	GrN-1059	10960	85	13071–12755	23%	33
Geldrop A2	Open-air	Feature s49	Pinus charcoal cm-size	SUERC-37211a	10955	35	12970–12757	0	32
Geldrop A2	Open-air	Feature s2	Pinus charcoal cm-size	SUERC-37209	10905	35	12891–12751	0	32
Geldrop A2	Open-air		Pinus charcoal cm-size	SUERC-37211b	10885	35	12881–12745	0	32
Zonhoven-Molenheide 2	Open-air		Scattered charcoal fragments	UtC-3720	10760	70	12834–12621	5%	31
Geldrop/Mie Peels/85–2	Open-air	Feature 7	Pinus charcoal patch	OxA-2563	10610	100	12758–12107	0	33
Geldrop/Mie Peels/85–1	Open-air	Feature 3	Pinus charcoal patch	GrN-16507	10090	110	12042–11248	0	33
Geldrop 3–3	Open-air		Charcoal patch	GrN-6841	8055	75	9254–8642		33
Zonhoven-Molenheide 2	Open-air		Scattered charcoal fragments	UtC-3195	7060	70	8015–7732	5%	31

For Kartstein the radiocarbon dates performed on bulk bone samples have not been included as there is doubt about their reliability ¹⁰.

Figure 4.

Summed probabilities distribution (SPD) of the charcoal dates and bone dates from different Ahrensburgian Culture sites along the southern North Sea basin, compared with the SPD of charcoal dates from the Usselo soil (list of dates, cf. SI3).

At the nearby Geldrop A2-site a possible similar situation was observed³². Here an Ahrensburgian Culture level was excavated in the top of an initial soil embedded in Younger Coversand II sediments situated ca. 5–30 cm above the Usselo soil. Three dates performed on charcoal collected from the edges of red sand features and lithic artefact clusters place the occupation at the very beginning of the YD (Table 3). The excavators argue there is little doubt that these dates are closely related to the occupation given their spatial distribution and the absence of bioturbation by dung beetle which could have moved the charcoal upwards. Notwithstanding, this does not seem to guarantee the anthropogenic nature of the charcoal features and the dispersed charcoal fragments given the fact that other, similar features at the site exist that appear less closely linked to the artefact clusters³⁶. Furthermore, the three charcoal dates are perfectly in line with the 12 radiocarbon dates performed on natural charcoal fragments from the underlying Usselo soil and lowest Younger Coversand II sediments^32,37 (Fig. 5). In fact, they are even slightly older than the highest positioned samples of natural charcoal, retrieved from the latter sediments (GrA-49570 and GrA-49521)³⁷. According to the excavators this is due to an old-wood effect, as cm-sized pine charcoal was used to date the Ahrensburgian occupation, while mm-sized samples were selected for dating the underlying sediments. Yet, Van Hoesel et al.³⁷ mention that an old wood-effect likely also affected the mm-sized charcoal samples from the Usselo-layer. So, despite the clear stratigraphical position of the Ahrensburgian level at Geldrop A2 above the Usselo soil one cannot completely exclude the possibility that the three dated samples were performed on charcoal from the underlying sediments, which were eroded in the vicinity of the site. At the nearby site of Geldrop 1, situated hardly 100m north of Geldrop A2, the Usselo soil in the highest part of the site was indeed eroded³³. Kasse et al.’s argument that cm-sized charcoal cannot have been displaced by wind has been contradicted by empirical and experimental research^38,39. These demonstrated that depending on the wind intensity and velocity, but also landscape topography, large charcoal fragments can be transported over distances of several 100m and even in certain conditions over several kilometres.

Figure 5.

Calibrated dates from the Ahrensburgian Culture site of Geldrop A2, ordered stratigraphically (list of dates, cf. SI4).

Besides movement by wind erosion, charcoal at some Ahrensburgian sites in the Meuse basin might have been displaced by bioturbation. This probably applies to the sites of Geldrop-Mie Peels and Zonhoven-Molenheide, as both are confronted with an intra-site incompatibility between the dates (Table 3). Although both dated samples at the former site came from the same stratigraphical layer, sealed by ca. 50 cm of Younger Coversand II sediments, according to the excavators the age difference can only be explained by intrusion of younger charcoal as a result of bioturbation⁴⁰. They conclude that the oldest date (OxA-2563) is the only correct one belonging to the excavated Ahrensburgian occupation. At Zonhoven-Molenheide the age difference between both dates performed on scattered charcoal fragments is even bigger. The youngest date (UtC-3195) goes back to the Late Mesolithic, pointing to a re-occupation of the site during this period, as also suggested by the occurrence of Mesolithic artefacts intermixed with the Ahrensburgian lithic industry¹⁸.

So, in conclusion, based on the above arguments, the dates performed on charcoal samples from open-air sites in the sandy lowland of the Meuse basin need to be addressed with utmost caution as they might not be performed on humanly produced charcoal (e.g. from fire places) but rather on natural charcoal from wildfires. This is reinforced by the fact that the chronological range of these dates is in perfect agreement with the chronology of charcoal found in Usselo soils within the coversand lowland of Belgium and the Netherlands^34,35, covering the late Allerød and early YD (Fig. 4). This can be considered an extra argument against an anthropogenic connection of most, if not all, charcoal dates from the Belgian-Dutch Ahrensburgian sites.

Increased population from the mid-YD onwards?

Even if one would hypothetically accept that (some of) the late Allerød/early YD charcoal from the Belgian-Dutch Meuse basin, as discussed above, were produced anthropogenically, it would imply that Ahrensburgian hunter-gatherers were almost completely absent from the area during the later phases of the YD. This is however in sharp contrast with the chronology obtained on samples of burnt (calcined) and unburnt animal bones from open-air and cave sites. These dates clearly cluster in the later phase of the YD, with five of the six dated Ahrensburgian sites, Remouchamps included, younger than ca. 12,500 cal BP. Most of them are even younger than 12,200/12,000 BP showing no overlap with the majority of charcoal dates (Table 3; Fig. 4). This is also the case for the W-German cave site of Kartstein¹⁰, situated hardly ca. 70km to the east of the Meuse (Table 3). It might be argued that samples of calcined bones are not the most reliable for radiocarbon dating Late and Final Palaeolithic sites, as they often yield dates that are obviously too young^41–44. However, a recent extensive inter-comparative study on Final Palaeolithic and Mesolithic sites in the Belgian coversand area⁴⁵ has demonstrated that reliable dates can be obtained if the calcined bones have been quickly covered, protecting them from post-depositional processes and contamination. This certainly applies to the Geldrop-sites where most of the bones were collected from 20 to 50 cm depth in Younger Coversand II sediments, suggesting that they have been covered by aeolian sands during the later stages of the YD. This is also confirmed by extensive OSL-dating on the soil sequence at Geldrop A2, which yielded a mean date of 11,9 ± 0.9 ka for the coversands immediately above the Ahrensburgian level³². Furthermore, the chronology from the bone samples seems to fit better with the common assumption of a gradual decrease in the number of tanged points from the early stages of the YD (Remouchamps, Geldrop 1, Geldrop 3–1, Kartstein) towards the early Holocene (Geldrop 3–2 Oost, Eersel-Panberg) and ultimately the start of the Mesolithic^7,9,20,46 (Table 3).

Based on the bone chronology, it seems more likely that the majority of Ahrensburgian sites along the southern North Sea basin date to the later stages of the YD (from ca. 12,220/12,000 cal BP) and start of the early Holocene. The temporal difference in occupation may be related to changes in climatic conditions during the YD. Several studies^4,47–51 have identified a bi-partition of the YD in continental Europe, consisting of an initial cold and wet phase followed by a second phase of climatic amelioration and drier conditions. This climatic shift is generally dated to the mid-YD and seems to have occurred rather rapidly within a few decades or even shorter. At the lake Meerfelder Maar, situated less than 100km southeast of the Meuse valley, high-resolution varve analyses have demonstrated that the transition occurred within a year, at 12,240 cal BP^49,52,53. Preliminary analyses of a speleothem record from the Père Noël cave in the Meuse basin also suggest a transition from wet to drier conditions in the late YD⁵⁴.

This climatic improvement in the later YD was probably caused by a resumption of the North Atlantic Meridional Overturning Circulation (AMOC), inducing a northward shift of the oceanic polar front and westerly wind systems⁴⁹. Based on the appearance of pollen and macroremains of specific thermophilous plants, such as yellow water-lily (Nuphar), white waterlily (Nymphaea alba) and cattail (Typha latifolia), an increase of summer temperatures by 1°–2°C is suggested for NW Europe⁴⁸. This implies a mean July temperature between 13° and 15°C. However, summers were short as they remained restricted to July and August; significant cooling started from September onwards leading to severe cold during the winter, which lasted until late May or early June⁵⁵. These strong and long winters probably were the main reason why reindeer remained the dominant species even during the late YD.

Conclusions

An extensive re-dating of animal bones from the seminal cave of Remouchamps, in combination with a critical assessment of available radiocarbon dates for the Ahrensburgian Culture, demonstrates that, considering only reliable dates, sites from the southern North Sea basin all range within the second half of the YD and the very beginning of the Holocene. The reliability of earlier (charcoal) dates can be questioned either on behalf of potentially remobilised charcoal from the Usselo soil present at these sites, or due to the fact that the charcoal samples were not incontestably associated with anthropogenic structures. Therefore, we propose that the observed patterning reflects a decrease in population densities in the first half of the YD in response to severe coldness and large-scale wildfires. Population probably increased again, albeit still slowly, in the second half of the YD, when summers became somewhat warmer, announcing the start of early Holocene warming. This probably allowed hunter-gatherers from the Ahrensburgian Culture to settle again in open-air, at least seasonally. The above conclusions will however need to be further corroborated by the discovery of new Ahrensburgian sites that can be securely dated, as well as by increased research on palaeo-environmental proxies (e.g. speleothem records of the Meuse basin) that could help to improve our understanding of the internal climatic and environmental variability within the YD and how this might or might not have affected human occupations.

Methods

Sample selection

Altogether 32 radiocarbon dates were performed on animal remains coming from the excavations of M. Dewez, except for two older samples from the Rahir excavations (OxA-4191 and OxA-4190) which have been redated (Table 2). In addition a date was performed on a human bone (a metacarpus) collected by Rahir. Animal bones were selected from sectors A, D and DD, excavated by M. Dewez (Fig. 2), approximately the same amount each. Sectors B and C were not sampled since they were entirely disturbed by the backfill of the previous excavations¹⁹. The sample selection was performed with specific research questions in mind (species representation, traces of exploitation, baseline data for stable isotope studies etc.), and it also had to ensure that the remains were suitable for optimal collagen extraction. Thus, given the prevalence of reindeer (R. tarandus) remains in the collection^21,56 the sampling strategy primarily targeted this species to ascertain the most accurate dates for the occupation(s). The majority of the selected bones (N = 18) show evidence of human exploitation, such as cut marks, percussion traces, or green-bone fractures. None of them exhibit traces of carnivore activity. Additionally, other species were included to confirm whether they belong to the Ahrensburgian occupation(s). This is particularly the case for potential temperate/woodland taxa that could date to the Allerød/YD transition or the subsequent YD/Preboreal transition, such as cervids (Cervidae), wild boar (Sus scrofa), beaver (C. fiber), wild cat (F. silvestris), among others.

Sample preparation

The faunal and human bones were sampled using a diamond saw mounted on a Dremel drill tool. On average, 500 mg of bone material was sampled, by scraping approximately 2 µm of the outer surface of each of the bones.

Collagen extraction was implemented at the ArcheOs research laboratory at Ghent University (Belgium) following a modified version of Longin’s method⁵⁷. The demineralized samples were washed with 0.25 M NaOH before the hydrolyzation step and then, filtered with Ezee filter separators (pore size: 60–90 µm) to eventually be freeze-dried.

At the Isotope Bioscience Laboratory (ISOFYS) of Ghent University, total organic carbon and nitrogen content of the samples were measured using an Elemental Analyser (EA ISOLINK, Thermo Finnigan, Bremen, Germany). To assess the reliability of the ¹⁴C dates, C:N ratios of all samples, except one (RICH-34333) yielding too little collagen, were calculated using Ambrose’s⁵⁸ and DeNiro’s⁵⁹ indicators of collagen quality but also following the indications from Wojcieszak et al.²⁷, as applied at the radiocarbon dating laboratory at the Royal Institute for Cultural Heritage (Brussels, Belgium).

Radiocarbon dating and modelling

Samples were transferred into quartz tubes with CuO and Ag and combusted to CO₂. Graphitization of CO₂ was carried out using H₂ over a Fe catalyst on the Automated Graphitisation Equipment (AGE)^60–62 which is linked to the Elementar Vario Isotope Select. Targets were prepared and ¹⁴C concentrations were measured with accelerator mass spectrometry (AMS) at the Royal Institute for Cultural Heritage in Brussels (Belgium)^63,64. ¹⁴C results are expressed in pMC (percentage modern carbon) and indicate the percent of modern (1950) carbon normalized to δ ¹³C = –25‰ VPDB using the δ ¹³C measurements⁶⁵.

All radiocarbon dates were calibrated using IntCal20⁶⁶ and OxCal version v4.4.4. Before modelling first the most obvious outliers were eliminated, including samples RICH-34321 (12,958 ± 36 BP) and RICH-34335 (369 ± 22 BP), resp. much too old and too young with respect to the other dates. As the former predates the Late Glacial it most likely does not refer to a human activity in the cave, while the latter date clearly indicates a modern intrusion. In addition, both samples do not exhibit traces of human modification. Secondly, two dates that were obtained in the twentieth century (Table 1) were omitted as the new dates obtained on the same samples fit better with the remaining dates (OxA-4191 versus RICH-35064) and have much smaller standard deviations. Third, it was decided to not include the three youngest dates (RICH-34316; RICH-35065; RICH-34331), among which a date from a human metacarpus, as they clearly belong to the Mesolithic and cannot be attributed to the Ahrensburgian occupation(s) of the cave.

Bayesian modelling was performed on this filtered database consisting of 30 dates (two formerly and 28 newly performed dates) using the Phase function in Oxcal⁶⁷. An Agreement Index (AI) of > 60% was used as criterium for the model validity.

Supplementary Tables

Supplementary Table 1. ^{(293.1KB, pdf)}

Supplementary Figure and Table 2. ^{(305.8KB, pdf)}

Supplementary Table 3. ^{(19.5KB, docx)}

Supplementary Table 4. ^{(16.6KB, docx)}

Author contributions

P.C.: Writing—review & editing, Writing—original draft, Conceptualization, Formal analysis, Investigation, Methodology, Project administration, Supervision; C.P.: Writing—review & editing, Writing—original draft, Data curation, Formal analysis, Investigation, Methodology; P.R.: Writing—review & editing, Writing—original draft, Data curation, Formal analysis, Investigation, Methodology; P.V.D.S.: Writing—review & editing; M.B.: Writing—review & editing, Formal analysis, Investigation, Methodology; I.D.G.: Writing—review & editing, Supervision; S.V.: Writing—review & editing; H.V.D.D.: Writing—review & editing, Writing—original draft.

Data availability

Data will be made available on request after sending an e-mail to Philippe.crombe@ugent.be.

Competing interests

The authors declare no competing interests.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-68686-z.