The South American Monsoon System (SAMS) is generally considered to be highly sensitive to Northern Hemisphere (NH) temperature variations on multi-centennial timescales. The direct influence of solar forcing on moisture convergence in global monsoon systems on the other hand, while well explored in modeling studies, has hitherto not been documented in proxy data from the SAMS region. Hence little is known about the sensitivity of the SAMS to solar forcing over the past millennium and how it might compete or constructively interfere with NH temperature variations that occurred primarily in response to volcanic forcing. Here we present a new annually-resolved oxygen isotope record from a 1500-year long stalagmite recording past changes in precipitation in the hitherto unsampled core region of the SAMS. This record details how solar variability consistently modulated the strength of the SAMS on centennial time scales during the past 1500 years. Solar forcing, besides the previously recognized influence from NH temperature changes and associated Intertropical Convergence Zone (ITCZ) shifts, appears as a major driver affecting SAMS intensity at centennial time scales.

Mineral magnetism of Pau d'Alho cave sediments, soils outside the cave, and in the stalagmite #6 (ALHO6) in Midwest Brazil is presented. This high growth‐rate speleothem (~168 mm/ka) encompasses the past 1355 years. Oxygen and carbon isotope data from the same stalagmite allow for a direct comparison of the magnetic signal with changes in paleoprecipitation and soil dynamics at the surface. Magnetic experiments include isothermal remanent magnetization, anhysteretic remanent magnetization, hysteresis loops, first‐order reversal curves, and low‐temperature superconducting quantum interference device magnetometry. The main magnetic remanence carriers in ALHO6 are magnetite and goethite, with a nearly constant relative proportion. Remanent coercivities of magnetite in all our samples are within 14–17 mT for an average grain‐size of ~1–2 µm, in the range of pedogenic magnetite, thus suggesting the detrital grains deposited in the stalagmite were produced in the soil above the cave. Magnetic remanence variations follow δ13C and δ18O data, suggesting a climatic control on the input of magnetic minerals into the Pau d'Alho cave system. The concentration of magnetic minerals in the stalagmite is governed by soil erosion above the cave, which by its turn is controlled by soil erosion and vegetation cover. Dry periods are associated with less stable soils and result in higher mineral fluxes carried into karst systems. Conversely, wetter periods are associated with soils topped by denser vegetation that retains micrometer‐scale pedogenic minerals and thus reduces detrital fluxes into the cave.

The diminishing strength of the Earth’s magnetic dipole over recent millennia is accompanied by the increasing prominence of the geomagnetic South Atlantic Anomaly (SAA), which spreads over the South Atlantic Ocean and South America. The longevity of this feature at millennial timescales is elusive because of the scarcity of continuous geomagnetic data for the region. Here, we report a unique geomagnetic record for the last ∼1500 y that combines the data of two well-dated stalagmites from Pau d’Alho cave, located close to the present-day minimum of the anomaly in central South America. Magnetic directions and relative paleointensity data for both stalagmites are generally consistent and agree with historical data from the last 500 y. Before 1500 CE, the data adhere to the geomagnetic model ARCH3K.1, which is derived solely from archeomagnetic data. Our observations indicate rapid directional variations (>0.1°/y) from approximately 860 to 960 CE and approximately 1450 to 1750 CE. A similar pattern of rapid directional variation observed from South Africa precedes the South American record by 224 ± 50 y. These results confirm that fast geomagnetic field variations linked to the SAA are a recurrent feature in the region. We develop synthetic models of reversed magnetic flux patches at the core–mantle boundary and calculate their expression at the Earth’s surface. The models that qualitatively resemble the observational data involve westward (and southward) migration of midlatitude patches, combined with their expansion and intensification.

The most recent Giant Gaussian Process (GGP) model, based on the last 5 Ma, has been used as a reference for directional distribution of paleomagnetic record of older rocks as Cenozoic and Proterozoic. However, for Paleozoic times, its validity has not yet been tested. Here we evaluate the validity of this recent GGP model for the Kiaman superchron. We present new paleomagnetic results from a late Pennsylvanian section of glacial rhythmites (Mafra Formation) from southern Brazil. The 5-m section sampled spans more than 800 kyr, as evaluated by cyclostratigraphic analysis. Thermal demagnetization revealed a reversed characteristic component carried by single domain magnetite. Anisotropy of anhysteretic remanent magnetization indicated a small shallowing correction of f = 0.97. The final paleomagnetic pole position is located at 51.9°S, 344.3°E (N = 111, R = 109.0, K = 55.9, A 95 = 1.8°), with a mean direction of Dec = 144.2°, Inc = 69.5°(N = 111, R = 110.2, k = 134.4, α 95 = 1.2°, Paleolat = 53.2°S). The shape of the distribution of magnetization directions (elongation E = 2:08 3:13 1:44 Þ and the dispersion of virtual geomagnetic poles (S V ¼ 10:9 11:8 9:8 Þ are incompatible with the recent model. The reduced dispersion, also found in other studies, implies a different shape in directional distributions for any GGP model describing the Kiaman interval. This result alerts us that we should abandon the use of the recent GGP model as a reference for inclination shallowing correction of Carboniferous sedimentary data.