Quantitative analysis of the impact of droughts and floods on internal wars in China over the last 500 years

Abstract

Although many large-N quantitative studies have evidenced the adverse effects of climatic extremes on social stability in China during the historical period, most of them rely on temperature and precipitation as major explanatory variables, while the influence of floods and droughts on social crises is rarely measured. Furthermore, a comparison of the climate-society nexus among different geographic regions and at different temporal scales is missing in those studies. To address this knowledge gap, this study examines quantitatively the influence of floods and droughts on internal wars in three agro-ecological (rice, wheat, and pastoral) regions in China in AD1470–1911. Poisson regression and wavelet transform coherence analyses are applied to allow for the non-linear and non-stationary nature of the climate-war nexus. Results show that floods and droughts are significant in driving internal wars in historical China, but are characterized by strong regional variation. In the rice region, floods trigger internal wars at the inter-annual and multi-decadal time scales. In the wheat region, both floods and droughts cause internal wars at the inter-annual and multi-decadal time scales. In the pastoral region, internal wars are associated with floods only at the multi-decadal time scale. In addition, the multi-decadal coherence between hydro-climatic extremes and internal wars in all three of the agro-ecological regions is only significant in periods in which population density is increasing or the upper limit of regional carrying capacity is being reached. The above results imply that the climate-war nexus is mediated by regional geographic factors such as physical environmental setting and population pressure. Hence, we encourage researchers who study the historical human-climate relationship to boil down data according to geographic regions in the course of statistical analysis and to examine each region individually in follow-up studies.