1000 years of population, warfare, and climate change in pre-Columbian societies of the Central Andes

Abstract

Different Andean societies underwent processes of expansion and collapse during propitious or adverse climate conditions, resource boost or depletion along with population variations. Previous studies have emphasized that demographic collapses of polities in the Central Andes Area were triggered by warfare and the negative impacts of fluctuating climate (droughts) on crop productivity. Nevertheless, the interactions between climatic variability, demography and warfare have been less thoroughly evaluated. We develop population dynamic models to test feedback relationships between population growth, climate change and warfare in the Central Andes, where considerable regional hydroclimate variations have occurred over a millennium. Through population models, we found out that the rise and demise of social polities in the northern coast of the Central Andes appear to be a consequence of climate change. In contrast, for the highlands of Peru and the Titicaca basin, population models suggest that warfare intensity has a negative effect on population growth rates.

Fig 1. Central Andes archeological sites and the analyzed time series.

(A)—Map showing the locations of archaeological sites providing radiocarbon data for each cultural tradition. (B) and (C)—Normalized (solid curves) and unnormalized (dashed curves) SPDs for northern (blue curve) and southern (orange curve) sociocultural areas. Vertical dashed bars indicate declining population growth rates. Note that colors of curves that describe SPDs for each tradition are maintained in panels below. (D) and (E)—Calculated growth rates based on normalized (solid curve) and unnormalized (dashed curve) SPDs. (F) and (G)—Raw (dashed curves) and smoothed (solid curves) time series for warfare intensity in each sociocultural area. (H)—Proxy time series for regional hydroclimate conditions. The green curve shows annually resolved lithic concentrations (see [21]), and the solid dark curve denotes the corresponding smoothed time series.

Fig 2. Predicted and observed population dynamic at northern sociocultural area.

Comparisons between population dynamic model-predicted trajectories (red solid curve) and observed SPD data (normalized SPD: solid line and blue dots; nonnormalized SPD: dotted lines and blue dots) for the northern sociocultural area. Panels (A) and (B) show model predictions that include hydroclimatic variation for normalized and unnormalized data, respectively (Eq (1) = model Climate; S2 Table). (C) and (D) show predictions for models including warfare effects for normalized and unnormalized data (Eq (2) = model Warfare; S2 Table). Further details on the model predictions are provided in S2 Table.

Fig 3. Predicted and observed population dynamic at southern sociocultural area.

Comparisons between population dynamic model-predicted trajectories (dark dashed curve) and observed SPD data (normalized SPD: solid curves; nonnormalized SPD: dotted orange curves) for the southern sociocultural area. Model predictions for the effect of changes in hydroclimate are shown for normalized and unnormalized data in Panels (A) and (B), respectively (Eq (1), model Climate). Predictions of models that include warfare effects for normalized and unnormalized data are shown in Panels (C) and (D) (Eq (2), model Warfare). Gray shaded areas show 95% confidence intervals for model predictions. Further details on the model predictions are provided in S2 Table.