The El Cañizar de Villarquemado pollen record covers the last part of MIS 6 to the Late Holocene. We use Tolerance-Weighted Averaging Partial Least Squares (TWA-PLS) to reconstruct mean temperature of the coldest month (MTCO) and growing degree days above 0°C (GDD0) and the ratio of annual precipitation to annual potential evapotranspiration (MI), accounting for the ecophysiological effect of changing CO2 on water-use efficiency. Rapid summer warming occurred during the Zeifen-Kattegat Oscillation at the transition to MIS 5. Summers were cold during MIS 4 and MIS 2, but some intervals of MIS 3 had summers as warm as the warmest phases of MIS 5 or the Holocene. Winter temperatures declined from MIS 4 to MIS 2. Changes in temperature seasonality within MIS 5 and MIS 1 are consistent with insolation seasonality changes. Conditions became progressively more humid during MIS 5, and MIS 4 was also humid, although MIS 3 was more arid. Changes in MI and GDD0 are anti-correlated, with increased MI during summer warming intervals. Comparison with other records shows glacial-interglacial changes were not unform across the circum-Mediterranean region, but available quantitative reconstructions are insufficient to determine if east-west differences reflect the circulation-driven precipitation dipole seen in recent decades.

Oxygen isotope stage 3 (OIS 3), encompassing the long middle section of the last glacial interval, has been the focus of an intensive high-resolution climate modeling effort for Europe. These model simulations produce substantially colder climates than modern simulations; however, the temperatures appear warmer than many proxy indicators suggest. In order to evaluate the importance of the model boundary conditions, comparable simulations are completed for the last glacial maximum (LGM). The LGM simulation produces a much colder European continent than OIS 3, despite similarities in the specification of sea-surface temperatures (SSTs). Ice-sheet dimension is evidently a key factor in explaining the difference in European climates over the past 40,000 yr. However, underestimates in specified OIS 3 ice sheets cannot be invoked to explain the discrepancies, since data strongly indicate small ice-sheet extents at that time; this leaves errors in specified OIS 3 SSTs as the most likely cause.