Drought as a possible contributor to the Visigothic Kingdom crisis and Islamic expansion in the Iberian Peninsula

The study investigates whether climate variability—particularly drought—contributed to the decline of the Visigothic Kingdom and facilitated the Islamic expansion in the Iberian Peninsula during the 5th–10th centuries CE. Motivated by the scarcity of high-resolution paleoclimate records for Early Medieval Iberia and the ongoing debate about climate’s role in historical crises, the authors compile and analyze a large set of paleoenvironmental proxies. Prior research indicates precipitation is the main driver of Mediterranean environmental variability, strongly influencing human populations. This work uses a high-resolution pollen stack from Iberia and Morocco, alongside paleohydrological, archaeological, and historical data, to test the hypothesis that persistent droughts coincided with and possibly contributed to economic stress, famine, disease, and socio-political instability during the Visigothic and early Islamic periods.

Previous Holocene studies at decadal–centennial scales identify precipitation as the key climatic factor in Mediterranean environmental change, with notable societal impacts. The Late Antique Little Ice Age (LALIA, starting mid-6th century) has been linked to volcanic eruptions and the Justinianic plague, with broader implications for European settlement crises. Regional reconstructions (NAO, TSI) suggest complex, debated relationships between solar forcing, atmospheric circulation, and Mediterranean hydroclimate. Prior archaeological and historical scholarship on the Visigothic period documents rural settlement dynamics, agricultural strategies, pastoral expansion, and governance changes, but the explicit role of climate in the Visigothic crisis and Islamic expansion in Iberia has been underexplored. Recent work relating drought to societal change in Arabia during the 6th–7th centuries provides comparative context for climate’s role in political transformations.

- Study area and datasets: Compiled 107 continuous lacustrine fossil pollen records from Spain, Portugal, Andorra, and Morocco (Neotoma and European Pollen Database; additional records via author contact). Focused on aridity indicators: Artemisia and other xerophyte taxa. The Artemisia stack includes 3977 samples; the xerophyte stack includes 4115 samples (Artemisia absent in seven records). Detailed site metadata provided in Supplementary Data 1; stacked data in Supplementary Data 2. - Age chronologies: Used original chronologies from databases and EPD MADCAP outputs. Calibration curves (IntCal98/04/09/13/20) differences are negligible (<20 calibrated years) for 450–950 CE. Composite stack-specific age uncertainties cannot be assigned; large sample size (107 records) mitigates age error. - Pollen-precipitation calibration: Employed 8217 modern pollen sites (EMPDv2) with WorldClim v2.1 precipitation to validate Artemisia as an aridity proxy. Sites with MAP < ~400 mm/yr show mean Artemisia >7%; with MAP > ~400 mm/yr, mean Artemisia <3%, supporting Artemisia as an indicator of dry conditions. - Data processing and normalization: Per-record z-score normalization of Artemisia and xerophyte percentages prior to stacking. Applied 45-point and 15-point moving averages for comparison with other records. A 3-point moving average was used to reduce noise before SnSiZer analysis. - Correlation analyses: Compared stacked Artemisia with high-resolution speleothem δ18O from Chaara cave (N Morocco) and stacked relative δ13C from Cobre, Kaite, and Mayor caves (N Spain). Time series were linearly interpolated to common resolutions (4 or 5 years). Pearson correlations: r=0.802 (p=0.0071) for Artemisia vs Chaara δ18O; r=0.768 (p=0.0106) for Artemisia vs Spanish δ13C. Analyses performed in PAST 4.12b. - SnSiZer analysis: Implemented Scale-normalized Significant Zero crossing (SnSiZer) to detect statistically significant increases/decreases in Artemisia (interpreted as aridity trends; inverted axis). Used 95% confidence to mitigate autocorrelation risks from pre-smoothing. Drought periods defined from the onset of significant increases (red) to onset of decreases (blue) in Artemisia; maximum onset/termination resolution ~5 years. Identified four strong droughts within 450–950 CE. - Additional paleohydrological and archaeological context: Compared pollen results with Iberian/Moroccan speleothems (e.g., Asiul, Buraca Gloriosa, GP5/Grotte de Piste), quantitative precipitation reconstructions, and lake records (Medina non-pollen palynomorphs; Zoñar lake levels). Compiled historical sources (6th–11th century chronicles, Council records) under annual/biannual resolution, and reviewed key archaeological evidence (rural settlements, irrigation/marriage of terrain, crop and pastoral strategies).

- Long-term aridification: Artemisia and xerophyte stacks show an overall increasing trend over the last 5000 years, indicating progressive aridification in the western Mediterranean. - Four strongest droughts (450–950 CE) identified via SnSiZer on Artemisia stack: I) 545–570 CE; II) 695–725 CE; III) 755–770 CE; IV) 900–935 CE. The 695–725 CE interval exhibits the maximum aridity peak of the last 5000 years. - Proxy coherence: Strong positive correlations between Artemisia and speleothem isotopes support hydroclimate interpretation: r=0.80 with Chaara cave δ18O (N Morocco, p=0.007) and r=0.76 with stacked δ13C from N Spain (p=0.010), with higher isotopic values consistent with reduced precipitation and lower vegetation density. - Independent evidence of drought: Moroccan GP5 speleothem shows hiatuses at ~620–750 and 900–940 CE; Iberian precipitation reconstructions indicate declines of 0–100 mm/yr (700–800 CE) in northern Iberia and ~200 mm/yr around 680 CE in southern Iberia. Lake Medina records peak Gelasinospora abundance ~710 CE (dry lake indicator), and Zoñar lake shows its lowest level of the last 2500 years at ~650–750 CE. - Historical linkage: Drought I coincides with volcanic-driven cooling (536, 540, 547 CE), onset of LALIA, and reports of famine and plague (e.g., Justinianic plague). Drought II aligns with reported harsh winters, famine, and plague in Hispania (late 7th–early 8th c.), the Visigothic civil strife, and the 711 CE Muslim invasion. Drought III coincides with mid-8th-century famines and political turmoil, including the Berber uprising and establishment of the Emirate of Córdoba (756 CE). Drought IV overlaps with the first fitna in al-Andalus and prolonged internal rebellions (e.g., Omar Ibn Hafsun, 880–928 CE). - Climate mechanisms: The most arid late 7th–early 8th century conditions occurred during a high positive NAO reconstruction and the lowest TSI of the last 5000 years, consistent with a mechanism linking low solar irradiance to altered NAO states and reduced storm activity over the Mediterranean. Mid-6th century cold/drought likely volcanically forced. - Socioeconomic implications: Given that ~80% of the population worked in agriculture and at least ~60% of wealth derived from agricultural products, multi-decadal droughts plausibly reduced food production, intensified famines and disease, and exacerbated socio-political instability in both the Visigothic state and early al-Andalus.

The integrated pollen-speleothem-historical approach indicates that repeated, severe droughts coincided with major episodes of social stress and political transformation in Early Medieval Iberia. The strongest drought (695–725 CE) aligns with extensive historical testimony of famine and plague, the Visigothic political crisis, and the Islamic conquest, suggesting climate acted as a compounding stressor on an already fragile, agriculture-based economy. Earlier (545–570 CE) volcanically forced cooling and aridity plausibly contributed to food shortages and disease (Justinianic plague), while subsequent mid-8th and early-10th century droughts overlapped with internal crises in al-Andalus, including famines and civil wars. Mechanistically, the study situates these droughts within a broader framework involving NAO variability and low solar irradiance during the late 7th–early 8th century, which may have suppressed storm tracks and reduced precipitation over the western Mediterranean. While human land-use and political dynamics clearly modulated outcomes, the multi-proxy congruence strengthens the inference that hydroclimate variability was an important contributor to instability. The findings address the core question by demonstrating statistically significant, regionally coherent droughts temporally aligned with documented crises, thus supporting a climate-society linkage without asserting monocausality.

By stacking 107 Iberian-Moroccan pollen records and combining them with speleothem isotopes, lake-level and non-pollen palynomorph indicators, quantitative precipitation reconstructions, and contemporaneous historical sources, the study identifies four statistically significant droughts (545–570, 695–725, 755–770, and 900–935 CE) between 450 and 950 CE. The late 7th–early 8th century interval represents the most severe aridity of the last 5000 years and coincides with the Visigothic Kingdom’s crisis and the Islamic expansion into Iberia. The results support the view that repeated droughts likely stressed agricultural production, intensified famines and disease, and contributed to socio-political instability in both the Visigothic and early Islamic polities. Future research should refine age models to reduce chronological uncertainty, expand high-resolution multi-proxy coverage (especially in North Africa and understudied Iberian regions), quantify hydroclimate impacts on specific economic sectors, and further test climate mechanisms (NAO–TSI–volcanic forcings) with climate model–proxy data integration.

- Human impact on vegetation (deforestation, land-use change, fire) can influence pollen abundances, potentially confounding purely climatic interpretations, though regional stacking reduces localized anthropogenic noise. - Composite age uncertainties cannot be precisely quantified due to differing site-specific chronologies; while large sample size mitigates, residual offsets may affect event alignment and lag interpretation. - SnSiZer analysis involved pre-smoothing that can induce autocorrelation and risk false positives; a 95% confidence level was used, but statistical significance cannot be guaranteed. - NAO reconstructions show imperfect agreement with pollen-inferred aridity for 450–950 CE, and the solar–NAO–hydroclimate linkage remains debated. - Some key records (seven pollen sites) are not openly accessible; spatial heterogeneity means localized deviations from regional trends may exist. - The study infers contribution rather than causation; political, military, and economic factors also played major roles in historical outcomes.