Frequent disturbances enhanced the resilience of past human populations

Understanding human societal responses to past disturbances is crucial for addressing future crises. While numerous case studies exist on cultural collapse, transformation, and persistence, a lack of global comparability hinders a comprehensive understanding of population resilience. Previous research often focuses narrowly on extreme events, potentially overlooking long-term vulnerabilities. This study addresses this gap by establishing a global comparative approach to long-term resilience, examining factors that structure the response of prehistoric populations to disturbances over time. The study utilizes radiocarbon time-frequency data as a proxy for population change, defining downturns as periods when population levels are significantly below expected growth trajectories. The core questions investigated are: (1) how quickly do past populations recover after downturns, and (2) what factors mediate resistance and resilience to downturns?

The authors review existing literature highlighting the lack of systematic global comparisons of human resilience to disturbances. They criticize the tendency to focus on extreme events and the limited comparability between case studies of population resilience. The limitations of existing approaches in modeling impacts, recovery, and adaptation, and failing to account for long-term historical variations, are also discussed. The Greenland Norse example of a short-term dietary shift that ultimately increased long-term vulnerability is presented as a key illustration of the need for long-term perspectives. The authors introduce their proposed global comparative approach as a solution to the limitations of previous research.

This meta-analysis involved a synthesis of data from 16 published study regions using archaeological radiocarbon data to reconstruct regional palaeodemographic trends. Studies were selected based on three criteria: evidence of significant downturns, their scope, and the inclusion of radiocarbon datasets. Where necessary, data was supplemented from the People3k database. The global sample included regions spanning the Arctic to the tropics, covering 30,000 years of history. Population downturns were identified by comparing SPDs (summed probability distributions) of calibrated radiocarbon dates to expected growth trajectories. Resistance and resilience metrics were calculated for 154 periods of population downturn. Disturbances were classified into general categories (environmental, cultural, mixed) and specific drivers were identified. Hierarchical linear mixed models were used to test for significant associations between parameters while controlling for regional variability. Metrics such as overall downturn duration, time to minimum SPD value, resistance (depth of downturn), and resilience (rate of recovery) were calculated and analyzed. The frequency of downturns was also calculated and analyzed as a potential mediating factor.

The meta-analysis revealed that the frequency of past downturns is a significant predictor of both resistance (ability to withstand disturbances) and resilience (rate of recovery). Agricultural and agropastoral populations experienced significantly more frequent downturns than other land-use types. While most downturns lasted for multi-decadal to centennial periods, with median duration of 98 years, the time to reach minimum population levels skewed towards shorter timescales. Although around a third of downturns showed resistance values dropping more than 50% from pre-downturn levels, resilience was relatively high across all cases (median = 0.64). Regions showing notably lower resilience included the Central China Plains, the Caribbean archipelago, and the Korean Peninsula. Further analysis indicated that agricultural and agropastoral land-use patterns are strongly associated with higher frequencies of downturns, suggesting a trade-off between increased vulnerability and enhanced adaptive capacity through repeated exposure. The effect size of downturn frequency on resistance was larger than its effect on resilience, suggesting that the ability to withstand downturns is distinct from the ability to recover from them.

The findings suggest a common mechanism across human populations linking frequent disturbances to increased resilience. The strong relationship between the frequency of downturns and both resistance and resilience is highlighted, although this does not imply a deterministic relationship. The study demonstrates that while food production increased population vulnerability, it also enhanced adaptive capacity through repeated exposure to downturns. The authors draw parallels to macroecology, where similar resistance-resilience patterns are observed at long timescales. Historical examples of long-term depopulation in the Americas and the Roman Empire support the study's findings on the typical duration of downturns. The authors note that the observed relationship might be explained by mechanisms like biased cultural transmission, enhanced social cohesion, technological advancement, and landscape learning.

This study demonstrates a strong correlation between the frequency of population downturns and the resilience of past human populations. Agricultural and agropastoral societies, while experiencing more frequent disturbances, exhibited higher resilience. Future research should focus on disentangling the complex interplay of factors influencing this relationship, including the role of cultural transmission, technological innovation, and socio-political structures. The findings underscore the importance of considering long-term historical contexts when evaluating societal resilience in the face of environmental change.

The study acknowledges limitations inherent in using radiocarbon data as a proxy for population size, as the data may not always directly reflect population numbers. The classification of disturbances into categories and specific drivers relies on interpretations from original studies and may be subject to some uncertainty. The sample size, while large compared to previous studies, could be further enhanced to increase statistical power and refine understanding of the identified relationships. Additionally, the model does not fully account for the complex interactions between various factors influencing resilience.