henhaobio
Sign in

PLoS ONEEcologyPeer-reviewed journal article

Demographic responses to climatic changes during the Final Palaeolithic in Europe

Isabell SchmidtiD, Birgit GehleniD, Katja WinkleriD, Álvaro Arrizabalaga ValbuenaiD, Nico ArtsiD, Nuno BichoiD, Philippe CrombéiD, Berit Valentin ErikseniD, Sonja B. GrimmiD, Katarína KapustkaiD, Mathieu LanglaisiD, Ludovic MeveliD, Nicolas NaudinotiD, Zdeňka NerudováiD, Marcel J. L. Th. NiekusiD, Marco PeresaniiD, Felix RiedeiD, Florian SaueriD, Werner SchöniD, Iwona Sobkowiak‐TabakaiD, Hans VandendriesscheiD, Mara-Julia WeberiD, Annabell ZanderiD, Andreas ZimmermanniD, Andreas MaieriD

PDFDOIPreprint

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 2 (17 Sept 2024) · CC BY 4.0

Cologne Protocol estimates from 1788 Final Palaeolithic sites show Europe's hunter-gatherer population roughly halving from GI-1d-a to the Younger Dryas, with occupation shifting eastward.

Regional population dynamics of Final Palaeolithic Europe remain poorly resolved, and an earlier study could only treat the period as one block because the data were coarse. An expert-compiled site database now allows separate regional estimates for each phase, testing how hunter-gatherers responded to Younger Dryas cooling and comparing this with the Gravettian decline.

Key results

  1. ShownThe total Final Palaeolithic meta-population in the study area fell by roughly half from GI-1d-a to GS-1. The decline appears in both the all-sites dataset and the chronologically resolved dataset, and radiocarbon-dated site counts show the same trend.

    Dataset A: mean 8100 people (4800 to 14,300) in GI-1d-a vs 4250 (2200 to 8800) in GS-1; Core Area density 1.3 vs 0.9 p/100 km². Dataset B: 5700 (3100 to 9900) vs 3500 (1900 to 6200). Radiocarbon-dated sites (Radiocarbon Palaeolithic Europe Database v.27) fell by more than 50 %. · Results; Discussion; Demographic dynamics, minimal viable population…

  2. SuggestedIn GI-1d-a, occupation centred north of the Alps rather than in the Franco-Cantabrian region, and Central Europe became the main growth area. The lead of the Czech Republic and SE Germany depends on including generically dated sites.

    Dataset A, GI-1d-a: Czech Republic and SE Germany around 2500 people (2.2 p/100 km²), northern Spain and French Pyrenees 1200 (1.7), Poland and NE Germany 1100 (0.8). Dataset B ranks northern Spain and French Pyrenees first (1100), then Benelux and NW Germany (820), with Poland and NE Germany at 770. · Abstract; Results; Discussion

  3. ShownRegions followed different paths under GS-1 cooling. Most declined, and some may have fallen below a hundred people. Poland with NE Germany and NE Italy grew instead, and Poland with NE Germany became the largest regional population.

    GS-1: Czech Republic and SE Germany fell to around 460 people (dataset A) or 80 (dataset B), at 0.9 p/100 km²; Poland and NE Germany held around 1200 in all estimates, almost twice northern Spain and French Pyrenees (740 dataset A; 590 dataset B); Switzerland and SW Germany 260 (A) and 230 (B). · Results

  4. SuggestedCore Areas shrank in GS-1 and, outside Iberia and southwestern France where they stayed in place, shifted east. The authors read the eastern increases as people moving from west to east, not as local growth, because Younger Dryas conditions were harsher toward the west.

    Core Area extent fell by about 25% in datasets A and B (dataset A: about 635,000 to 481,000 km²); Core Areas vanished in western Britain, Brittany and Denmark and appeared in eastern Poland (Masuria); the growing season shortened by more than a month on the Atlantic coast vs 2 to 10 days in northern Central Europe. · Results; A large-scale shift to the east as a potential…; Conclusion

  5. SuggestedUnlike the late Gravettian collapse, the Final Palaeolithic decline differed between regions and left population totals within the range proposed for a viable human population.

    GS-1 mean estimates of 4250 (dataset A) and 3500 (dataset B) people vs a suggested minimum viable threshold of about 1500 individuals; Gravettian means dropped from 2800 (early phase) to 1000 (late phase), with reductions in all regions and collapse north of 50° latitude. · Results; A large-scale shift to the east as a potential…; Demographic dynamics, minimal viable population…

Methods, data and reproducibility

Cologne Protocol geostatistics: site-density isolines mark Core Areas within a 2.6 million km² study region. Each Core Area's size is divided by raw material catchment area (RMCA) quartiles to count groups, then multiplied by an ethnographic mean group size of 42.5. The model is run on all sites (dataset A) and on chronologically assigned assemblages only (dataset B).

  • Expert-compiled open-access database of Final Palaeolithic sites (excavations and surface collections) across Southern, Western, Northern and Central Europe · human (archaeological record) · n = 1788 sites; 1548 phase-assigned assemblages
  • Assemblages assigned to GI-1d-a (14-12.7 ka BP) · human (archaeological record) · n = 988 assemblages; 1356 sites in dataset A
  • Assemblages assigned to GS-1 (12.7-11.6 ka BP) · human (archaeological record) · n = 560 assemblages; 1003 sites in dataset A
  • Generically dated assemblages (Final Palaeolithic; Final Palaeolithic plus Preboreal; late GS-1 to Preboreal) kept only in dataset A · human (archaeological record) · n = 237, 131 and 75 assemblages
  • Lithic raw material catchment areas (RMCA) compiled from the literature, per phase · human (archaeological record) · n = 108 (53 used) for GI-1d-a; 97 (39 used) for GS-1
  • Radiocarbon-dated sites per phase, used as an independent check on visibility bias · human (archaeological record)
Methods & tools
Cologne Protocol, Optimally describing isolines (equal site-density isolines), Raw Material Catchment Area convex hulls with 5 km site buffer, Radiocarbon Palaeolithic Europe Database v.27, Ethno-historic mean group size (42.5 people)
Reproducibility gaps
The text mentions an open-access online site database but gives no URL or identifier for it, and does not say whether the analysis code or GIS workflow is available.
Limitations (6)
  • Discovery biases (geomorphology, modern land use, taphonomy, research history, how recognisable artefacts are) shape site distributions. The central void from SW France to the Upper Rhine may partly reflect such biases. · stated by the authors
  • Chronological resolution varies by region (calibration plateau; undated surface collections in SE Germany and the Czech Republic). Dataset A may overestimate and dataset B likely underestimates the population of each phase. · stated by the authors
  • Submerged coastlines and Doggerland are almost absent from the dataset, so the estimates rest only on evidence from today's landmass. · stated by the authors
  • The GS-1 increases in NE Italy, Poland and NE Germany could be inflated because GS-1 assemblages are easier to recognise typologically. · stated by the authors
  • Raw material data were scarce in some Core Areas (e.g., Great Britain, Denmark, northern Spain), so RMCA sizes were transferred from adjacent regions; after exclusions, estimates rest on 53 of 108 (GI-1d-a) and 39 of 97 (GS-1) RMCAs. · stated by the authors
  • The min-max ranges reflect only RMCA quartiles; the fixed group size (42.5) and the isoline modelling choices are never varied. The inferred west-to-east migration is not tested against any direct mobility evidence. · noted in this summary

Read this if you study Late Glacial hunter-gatherer demography or climate-driven human mobility during the Younger Dryas, or apply the Cologne Protocol to archaeological site data.

Citations
11
FWCI
24.61
Field percentile
top 1%

Abstract

The European Final Palaeolithic witnessed marked changes in almost all societal domains. Despite a rich body of evidence, our knowledge of palaeodemographic processes and regional population dynamics still needs to be improved. In this study, we present regionally differentiated estimates of absolute numbers and population densities for the Greenland Interstadial 1d-a (GI-1d-a; 14-12.7 ka BP) and the Greenland Stadial 1 (GS-1; 12.7-11.6 ka BP) for Southern, Western, Northern and Central Europe. The data were obtained by applying the Cologne Protocol, a geostatistical approach for estimating prehistoric population size and density, to a newly compiled dataset of Final Palaeolithic sites. On a large spatio-temporal scale, we observe a shift of the main areas of human occupation from the Franco-Cantabrian region, which was intensely occupied during most phases of the preceding Upper Palaeolithic, to regions north of the Alps. At smaller scales, we observe divergent regional trends in the Final Palaeolithic meta-population: during GI 1d-a, a decreasing population in southwestern Europe and an increasing population in north-eastern Central Europe. For the first time since the dispersal of anatomically modern humans into Europe, we see that Central Europe becomes the dominant demographic growth area. Subsequently, the climatic cooling of GS-1 coincides with a pronounced population decline in most parts of the study area. An apparent increase in population density occurs only in north-eastern Central Europe and north-eastern Italy. Our estimates suggest that the total population was reduced by half. Similar results, with a relationship between decreasing temperatures and decreasing populations, have already been observed for the late phase of the Gravettian, when populations were reduced to only one third of those estimated for the early phase. Yet, in contrast to the collapse of local populations during the late Gravettian, the increase in population densities in Central Europe during GS-1 indicates population movements eastwards, possibly in response to deteriorating climatic conditions, particularly in western regions during the Younger Dryas.

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISGelabert '25Roebroeks '17Baca '22Xue '25Racimo '19Bortolini '21
68% semantic matchGenomics
Pere Gelabert, Susanna Sawyer, Olivia Cheronet+33 more·bioRxiv·Sep 26, 2025· 1 citations