Climates of the Mediterranean During the Neolithic Transition

A Quantitative Approach Using the Köppen Classification and Radiocarbon Dating

Thomas Huet

University of Oxford

Niccolò Mazzucco

Università di Pisa

with the collaboration of Andrea Manica

University of Cambridge

Introduction

An ecological and economic revolution

Holocene

Late Foragers and Early Farmers

Late Mesolithic (LM), Late Foragers.

@ChatGPT

• Residenciality increase

• Circular mobility to access discontinuous resources in time and space (seasonality)

• Low demography

Early Neolithic (EN), Early farmers.

@ChatGPT

• Permanent settlements

• Logistic mobility to access localized resources for farming economy

• High demography

Cold weather favors a greater mobility. A la fin du Dryas III, vers 9,600 BC: amélioration du climat, et diminution des grands herbivores clairement documentée au Moyen-Orient et dans le Nouveau Monde, force les HG à intensifier leur foraging Climates: Near East is temperate and wet, North Africa transitioned to arid conditions, Southern Europe experienced is warmer and wetter Fertile Crescent | cultigens. , soil fertility for agriculture (cereals) and climates for livestock. Climates: broadly warm and stable The different periods defines characteristic anthropological moments. Conventional division in 3 phases, both for the Mesolithic and Neolithic. Standardized and heuristic way to represent the evolution of a culture. We represent here, using large brushes

Framework

Data: radiocarbon dates and palaeoclimatic data.

Neonet project

• Meso/Neo
• Dataset
• Online app
• Functions
• Openness

Early Mesolithic (EM)
Middle Mesolithic (MM)
Late Mesolithic (LM)
Early Neolithic (EN)
Middle Neolithic (MN)
Late Neolithic (LN)

Baume de Montclus, stacked SPD

Franchthi cave, stacked SPD

http://shinyserver.cfs.unipi.it:3838/C14dev/

source("R/neo_spd.R")
source("R/neo_spdplot.R")

neo_spd(df.c14 = df.c14)

• Open Data, Open Source, Open Access
• Scalable, Incrementable, Reusable, Analyzable

Materials and Methods

Data: radiocarbon dates and palaeoclimatic data.

Radiocarbon dataset

• Sources
• Alignment
• Outliers
• ROI
• Worflow

https://github.com/ropensci/c14bazAAR

https://github.com/zoometh/neonet

ref_table_per.xlsx

EM - Early Mesolithic

MM - Middle Mesolithic

LM - Late Mesolithic

EN - Early Neolithic

MN - Middle Neolithic

LN - Late Neolithic

• 8 different databases and datasets. Among which 2 are the Neonet datasets. We collect using the c14bazAAR and curated with NeoNet functions.
• mapping table with period and culture coming from different datasest to be mapped to the Neonet main classes (10 classes, among which 4 are undefined and period transition).
• Using

c14_aberrant_dates.tsv

• With R functions coming from the C14bazaar R package and curated with the NeoNet R series of functions, we are gathering some 10,000 radiocarbon dates coming from 8 different datasets and databases.
• We map these dates to the Neonet periods (here on the left class).
• Then we remove duplicated dates and discard outliers (currently in process).
• Some 4,000 dates are now available for the study.

neonet-data-2023-10-22-select-aera.geojson

neonet-data-2023-10-22.geojson

Here it’s a leaflet view of these exported files. On the left hand side the selection rectangle, on the right hand side the dates. Dates can undergo a secondary curation and subsetting process using a GIS or any other appropriate application.

Radiocarbon modelling

• Date selection
• Dates medians
• Dates interpolation

The most recent LM date median

The most ancient EN date median

weighted.median <- matrixStats::weightedMedian(x = ages1$Date1$ageGrid,
                                               w = ages1$Date1$densities)
df.c14[i, "median"] <- -(weighted.median - present)
if(stat.mean){
  weighted.mean <- matrixStats::weightedMean(x = ages1$Date1$ageGrid,
                                             w = ages1$Date1$densities)
  df.c14[i, "mean"] <- -(weighted.mean - present)
}
df.c14[i, "tpq"] <- -(min(ages1$Date1$ageGrid) - present)
df.c14[i, "taq"] <- -(max(ages1$Date1$ageGrid) - present)

• For a given site, we select either the most recent LM date (ex: Monte Frignone) or simetrically the most ancient EN date (ex: Le Baratin)
• Extract on the radiocarbon date weighted median from the Summed probability density (SPD)

• With R functions coming from the C14bazaar R package and curated with the NeoNet R series of functions, we are gathering some 10,000 radiocarbon dates coming from 8 different datasets and databases.
• We map these dates to the Neonet periods (here on the left class).
• Then we remove duplicated dates and discard outliers (currently in process).
• Some 4,000 dates are now available for the study.

interpolated <- akima::interp(x = df$longitude, 
                                y = df$latitude, 
                                z = df$median, 
                                duplicate = "mean")
  
  interp_df <- tidyr::expand_grid(i = seq_along(interpolated$x), 
                                  j = seq_along(interpolated$y)) %>% 
    dplyr::mutate(lon = interpolated$x[i],
                  lat = interpolated$y[j],
                  date.med = purrr::map2_dbl(i, j, ~interpolated$z[.x, .y])) %>% 
    dplyr::select(-i, -j)

Climates data

• Sources
• Methods

Mean annual temperature (ºC)

Annual precipitation (mm year ^-1)

Biome (pollen-based)

Beyer et al. 2020¹

https://github.com/EvolEcolGroup/pastclim

B <- MAP < 10 * Pthresh
BW <- B & MAP < 5 * Pthresh
BWh <- BW & MAT >= 18
BWk <- BW & MAT < 18
BS <- B & MAP >= 5 * Pthresh
BSh <- BS & MAT >= 18
BSk <- BS & MAT < 18

A <- Thot >= 18 & !B
Af <- A & Pdry >= 60
Am <- A & !Af & Pdry >= 100 - MAP / 25
Aw <- A & !Af & Pdry < 100 - MAP / 25

• pastclim R package to recreate the Koppen Climate classification (KCC) for past periods, currently from 8,000 BC to 5,000 cal BC (expressed here in cal BP)
• KCC based on average temperature and precipitation patterns, dividing them into five main types: tropical, dry, temperate, continental, and polar. Each type is further subdivided based on seasonal precipitation and temperature patterns (30 classes).
• The pastclim KCC current resolution is 30” = 0.5 degrees = c. 50 km. It uses the Beyer et al. 2020 dataset.

Climate dating by site

• Extract KCC
• EF Pioneer front KCC

52: Le Baratin, Ly-4725

• For a given site, we select either the most recent LM date (ex: Monte Frignone) or simetrically the most ancient EN date (ex: Le Baratin)
• Extract on the radiocarbon date weighted median from the Summed probability density (SPD)

• For a given site, we select either the most recent LM date (ex: Monte Frignone) or simetrically the most ancient EN date (ex: Le Baratin)
• Extract on the radiocarbon date weighted median from the Summed probability density (SPD)

Spread of Neolithic the Mediterranean

Radiocarbon dates are still being revised, and many of them need to be verified. First modeling

Eastern and Central Mediterranean (in BC)

• 9000
• 8400
• 7600
• 6800
• 6200
• 6000
• 5800
• 5600

Western Mediterranean (in BC)

• 5600
• 5300

Climates evolution

• Late Foragers
• Early Farmers
• Comparisons

• Late foragers: Drop of the inhabited cold climates from 11k BP to 7k BP. In 9k BP newly occupied occupied Cold, dry summer, cold summer, possibly corresponding to the higher mountain range (Atlas, Alps, Pyrenees, etc.)
• Early Farmers: 11k and 10k 2 main climates, continuous area: Fertile Crescent. Arid climates increases in 9k: arid margins (oasis). 8k and 7k: neolithisation of cold and temperated areas (Bulgaria, Southern Germany, etc.)
• Comparison: LM occupied a larger diversity of ecozones. EF replace HG in cold climate over 4,000 years
  

Discussion

Rproductible methods. Enhancement of cultural and climatic data

productible methods

• Modeling
• Cultural and Climate data

Ammerman, A. J., & Cavalli-Sforza, L. L. (1971)²

Fort, J. (2022)³

Betti, L., Beyer, R. M., Jones, E. R., Eriksson, A., Tassi, F., Siska, V., … & Manica, A. (2020)⁴

Binder, D., Angeli, L., Gomart, L., Huet, T., Maggi, R., Manen, C., … & Tagliacozzo, A. (2022)⁵

• All models are wrong but some are usefull
• Dates curation
• Possibility of having 1 km of resolution and aggregating in 5” (~10 km) with TraCE21k

Thank you

https://github.com/zoometh/neonet

• Contribution: mapping table (archaeologists), interpolation methods (mathematician), aggregating data from other sources (aDNA, etc.), etc.

Footnotes

1. Beyer, R. M., Krapp, M., & Manica, A. (2020). High-resolution terrestrial climate, bioclimate and vegetation for the last 120,000 years. Scientific data, 7(1), 236.

2. Ammerman, A. J., & Cavalli-Sforza, L. L. (1971)[^2]. Measuring the rate of spread of early farming in Europe. Man, 674-688.

3. Fort, J. (2022). The spread of agriculture: quantitative laws in prehistory?. In Simulating Transitions to Agriculture in Prehistory (pp. 17-28). Cham: Springer International Publishing.

4. Betti, L., Beyer, R. M., Jones, E. R., Eriksson, A., Tassi, F., Siska, V., … & Manica, A. (2020). Climate shaped how Neolithic farmers and European hunter-gatherers interacted after a major slowdown from 6,100 BCE to 4,500 BCE. Nature Human Behaviour, 4(10), 1004-1010.

5. Binder, D., Angeli, L., Gomart, L., Huet, T., Maggi, R., Manen, C., … & Tagliacozzo, A. (2019, March). L’Impresso-cardial du nord-ouest et ses rapports avec la «zone-source»: une synthèse chrono-culturelle. In Céramiques imprimées de Méditerranée occidentale. Matières premières, productions, usages.