The East-West Mediterranean Moisture Gradient: Multi-Decadal Phasing Changes from Mediterranean Tree Rings

Diagnosing the causes of the Mediterranean drying trend now being observed requires a temporal perspective that is not possible through relatively short instrumental climate records. In the simplest case, it is possible to take long tree-ring chronologies from the east-west poles of Mediterranean Basin and evaluate these records for changing patterns of variability since medieval times. This has been done for four chronologies, representing three tree species, from Morocco and Algeria in the West (CDAT) and Turkey, Greece, and Cyprus in the East (JUEX and PINI).  The four chronologies, each standardized using signal free methods, are all drought sensitive (PINI is the weakest) and have a useful common period (EPS>0.85) from 1246 to 2010. Unrotated and varimax rotated EOF analyses reveal clear evidence for an East-West gradient in the tree-ring chronology estimates of the Mediterranean moisture gradient, with distinct periods of multi-decadal phase-coherent variability based on cross-wavelet coherency analysis.  Orthogonality constraints imposed on the data by both unrotated and rotated EOF analysis were evaluated by conducting a parallel evaluation of the gradient on just the Morocco CDAT and Turkey JUEX chronologies since 1100 (EPS>0.85). Since only two chronologies were used, no EOF analysis was conducted and therefore no orthogonality constraints applied to the data. Cross-wavelet analysis of the Morocco and Turkey chronologies reveals the even stronger presence of multi-decadal phase-coherent East-West variability centered on 50-60 years.  From roughly 1500 to 1800, this time scale of variability is directly out-of-phase (strongly East-West). In contrast, the 1100-1350 and post-1900 periods have nearly in-phase behavior (a pan-Mediterranean signal) at periods of 50-60 years. A major perturbation in the Morocco chronology centered on 1450 serves as a marker for the transition from in-phase to out-of-phase behavior, but it is unclear if this “event” has contributed to the change in phasing. Because these in-phase periods are at least partly in the “cone of influence” of the cross-wavelet analysis, they are necessarily less certain.  However, if coincidental timing of anti-phased behavior from 1500 to 1800 (roughly during the Little Ice Age) and in-phase behavior from 1100 to 1350 and after 1900 (roughly during the Medieval Warm Period and 20-21st century warming) suggest a relationship to thermal conditions affecting the Mediterranean region and the changes in the identified East-West moisture gradient. If so, the currently observed Mediterranean drying trend needs to be explained within the context of this identified long-term multi-decadal variability.