Fig. 8: Summary of results (last 300 ka).

OSL and radiocarbon ages from floodplains/islands and fluvial terraces (a) are compared to plots showing population sizes through time estimated in the EBSP analyses for bird species specialized in islands or riverbanks (b), and for bird species specialized in floodplain forests (c), sea level variation derived from benthic foraminifera δ¹⁸O stack from globally distributed marine sites (d), precipitation variation in western Amazonia (e), atmosphere methane (CH₄) concentration (f) and precession-free variation of titanium-to-calcium ratio (ln(Ti/Ca)) in marine sediments (core M125-55-7) offshore eastern Brazil (g). Data generated in this study (a, b, c) are presented in Supplementary Information. In panel a, dots represent sediment sample ages and horizontal bars represent age errors calculated according to the Gaussian law of error propagation. The unit named as “Mid-Lower terraces” in panel a can include higher elevation still susceptible to flooding. Confidence intervals (shaded areas) in b, c correspond to 95% High Posterior Density. Higher amplitude variations (d) in stable oxygen isotopes (δ¹⁸O) stack from benthic foraminifera⁴² indicate sea level changes representative of the equatorial Atlantic Ocean. Precipitation variation in western Amazonia (e) corresponds to stable oxygen isotopes (δ¹⁸O) measured in speleothems from the Diamante cave⁴⁴. Atmospheric concentrations of CH₄ (f) correspond to data from Antarctic ice cores^64,65. Precession-free variation of ln(Ti/Ca) in marine sediments offshore eastern Brazil (g) represents the intensity of the SAMS⁵⁸.