Fig 1.
The black dots show the spatial distribution of the archaeological sites included in the EUBAR database and used to derive population data (Software: ArcGIS10.3).
Fig 2.
Regions investigated in the present study: (a) Swiss Plateau; (b) Po Plain; (c) Massif Central; (d) Southern French coast and Northeastern Iberian Peninsula. The latter two regions are also jointly referred to as ‘Northwestern Mediterranean’ in the present paper. White dots: sites included in the regional SCPDs, selected from the EUBAR dataset. Black dots: sites used for climatic reconstructions (coordinates reported in Table 1). (1) Bibersee; (2) Durchenbergried; (3) Feuenried; (4) Hornstaad/Bodensee; (5) Lac de Clairvaux; (6) Lac du Mont d'Orge; (7) Lac d’Annecy; (8) Lobsigensee; (9) Montilier; (10) Rotsee; (11) Castellaro Lagusello; (12) Forcello; (13) Lago Piccolo di Avigliana; (14) La Taphanel; (15) Lac du Mont de Belier; (16) Lastioulles; (17) Peyrelevade; (18) Tourbière de Chabannes; (19) Banyoles; (20) Embouchac; (21) Etang d'Ouveillan; (22) Pla de l`Estany; (23) Salada Pequeña.
Fig 3.
Testing the effect of taphonomic bias on the original SCPD from settlements (Software: PAST 3.18).
Black line: Original data; red line: Null-Hypothesis.
Fig 4.
Simulated SCPD (dark grey, same number of dates at each temporal bin) compared with the IntCal13 calibration curve.
Table 1.
Name and location of the pollen archives used for climatic reconstructions.
Coordinates are expressed in decimal degrees (WGS84 reference system).
Table 2.
Performance summary for the h-block cross validation exercise.
For each variable, we report the coefficient of determination (r2), the root of the mean squared error of the prediction (RMSEP) and the standard deviation of the observed climate variables.
Fig 5.
SCPDs of pre-screened radiocarbon dates from the EUBAR database, juxtaposed curves: (a) 1233 dates from the filtered dataset; (b) 852 14C dates from settlements; (c) 283 14C dates from funerary contexts (IntCal13 calibration curve).
Fig 6.
Probability density distributions of the SCPD with dates from settlements (black) and the simulated SCPD (red) after having been standardized (Software: PAST 3.18).
Fig 7.
LOWESS function applied to SCPD data from settlements, the 95% confidence interval is marked by the blue lines (Software: PAST 3.18).
Fig 8.
Standard logistic model fitted to the interpolated LOWESS function obtained from the SCPD data from settlements (Software: PAST 3.18).
Fig 9.
Linear trend of the interpolated LOWESS function obtained from the SCPD data from settlements (Software: PAST 3.18).
Fig 10.
(a) SCPD of 208 14C dates originating from sites located on the Swiss Plateau; (b) LOESS model of reconstructed summer precipitations; (c) LOESS model of reconstructed summer temperatures; (d) High lake level z-scores digitized from Magny (2013). The shaded areas in (b) and (c) outline the 95% confidence interval of each LOESS model.
Fig 11.
(a) SCPD of 134 14C dates originating from sites located in the Po Plain; (b) LOESS model of reconstructed summer precipitations; (c) LOESS model of reconstructed summer temperatures; (d) Water level fluctuation for Lake Ledro and Lake Accesa, digitized from Magny et al. (2013). The shaded areas in (b) and (c) outline the 95% confidence interval of each LOESS model.
Fig 12.
(a) SCPD of 57 14C dates originating from sites located in the Massif Central; (b) LOESS model of reconstructed summer precipitations; (c) LOESS model of reconstructed summer temperatures; (d) Duration of high detrital phase in Lake Aydat (Lavrieux et al., 2013); (e) SCDP-based peat initiation events digitized from Cubizolle et al. (2012). The shaded areas in (b) and (c) outline the 95% confidence interval of each LOESS model.
Fig 13.
(a) SCPD of 158 14C dates originating from sites located on the northeast Iberian Peninsula; (b) SCPD of 72 14C dates originating from sites located on the Southern French coast; (c) LOESS model of reconstructed summer precipitations; (d) LOESS model of reconstructed summer temperatures; (e) Annual temperature (ΔTann) and precipitation (ΔPann) reconstructions from the site of Montou (SW-France), digitized from Terral and Mengual (1999). The shaded areas in (b) and (c) outline the 95% confidence interval of each LOESS model.