Fig 1.
LIG (A) and Early Holocene (B) study area.
Legend: 1–Elevations (in meters above sea level, m a.s.l.); 2–No data; 3–Case studies indicating possible vegetation burning by LIG and Early–Middle Holocene hunter-gatherers [4,9–12,16,17]. List of case studies: a–Neumark-Nord; b–Bonfield Gill Head; c–Campo Lameiro; d–Dudka Island; e–Dumpokjauratj; f–Ipmatisjauratj; g–Kunda-Arusoo; h–Lahn valley complex; i–Lake Miłkowskie; j–Meerstad; k–Mesolithic site at Soest; l–North Gill; m–Pulli; n–Rottenburg-Siebenlinden sites; o–Star Carr; p–Vingen sites; q–Wolin II.
Fig 2.
Vegetation openness: CARbon Assimilation In the Biosphere (CARAIB) LIG (A), CARAIB 8700–8200 BP (B); Regional Estimates of VEgetation Abundance from Large Sites (REVEALS) mesocratic I (C), REVEALS 8700–8200 BP.
Vegetation openness for other time windows available in Supporting Information (S1 and S2 Figs S1 File.). Legend: 1–No data; 2–Vegetation openness (%).
Fig 3.
Distribution of dominant plant functional types (PFTs): CARAIB LIG (A), CARAIB 8700–8200 BP (B); REVEALS mesocratic I (C), REVEALS 8700–8200 BP.
PFT distribution for other time windows available in Supporting Information (S3 and S4 Figs S1 File.). Legend: 1–No data; 2–Herbs; 3–Shrubs; 4–Broadleaf trees; 5–Needleleaf trees.
Fig 4.
Overview of research steps including the comparison of CARAIB (climate-driven potential natural vegetation) and REVEALS (pollen-based vegetation reconstruction) data, the development and upgrade of the HUMLAND ABM, its integration with a genetic algorithm, and the generation of scenarios to quantify the impacts of Neanderthals, Mesolithic population, megafauna, natural fires, and climate on vegetation.
Table 1.
HUMLAND 2.0 parameter overview.
Table 2.
Genetic Algorithm setup details. A black dot indicates that a variable was optimized within its specified minimum and maximum values (as outlined in Table 1), whereas a white dot signifies that the variable remained constant. The experiment subsets are categorized as follows: 1) megafauna impact; 2) megafauna impact combined with natural fires; and 3) megafauna impact, natural fires, and human-induced fires.
Fig 5.
CARAIB–REVEALS comparison of mean vegetation openness (black dots) and the mean percentage of grid cells dominated by herbs (yellow) and trees (green) for the LIG and the Early Holocene.
Table 3.
Percentage of possible scenarios with output similar to REVEALS without anthropogenic fires. In these scenarios humans do not engage in vegetation burning, but they exert hunting pressure on herbivores.
Fig 6.
Summary statistics and values’ distribution of the “Hunting_pressure” parameter values required to generate HUMLAND scenarios with output similar to REVEALS without anthropogenic fires.
Humans do not engage in vegetation burning, but they exert hunting pressure on herbivores. The dot indicates the mean value for each dataset. For the LIG, most simulations matching REVEALS outputs have “Hunting_pressure” values around 20–25%, whereas for the Early Holocene, they typically cluster around 80–90%.
Table 4.
Percentage of possible scenarios with output similar to REVEALS with anthropogenic fires. These scenarios include the combined direct impact of all agents on vegetation: human induced and natural fires, and megafauna plant consumption.
Fig 7.
Summary statistics and distribution of the parameter values required to generate scenarios with output similar to REVEALS for PFT distribution (A, C, E, G) and vegetation openness (B, D, F, H) with hunting and anthropogenic fires.
The dot indicates the mean value for each dataset.
Fig 8.
Mean percentages of grid cells modified by different agents during the HUMLAND equilibrium state: A–LIG most frequent scenarios; B–Early Holocene most frequent scenarios.