An original procedure and a new mathematical technique have been developed which allowed us to obtain more reliable climatic reconstructions than with prior methods. They have been tested for different sites in Switzerland and Morocco. First, cores that were too short and cores that were not coherent enough with others were excluded, using cross-spectral analysis. Second, detrending and master chronologies were calculated. Three methods were compared: polynomial, high-pass filter, and spline indexing. An optimal detrending was obtained through comparison with climate, but there seems to be no general rule for it. More reliable and longer climatic reconstructions are made possible. A response functions technique in three steps is presented: regression after extracting principal components on monthly climatic parameters, on seasonal parameters, and on more biological parameters such as potential evapotranspiration, multiple spectral regression introducing frequency domain. This procedure provides a more complete and more dynamic view of tree growth. The transfer function method and its verification are illustrated for different sites in Morocco: three climatic parameters in Tetouan (February, May, and June temperatures) and one in Marrakech (May temperature) have been successfully reconstructed. These reconstructions are confirmed by different verifications.

A general discussion regarding problems inherent to developing climatically sensitive tree-ring chronologies from eastern North America is presented. Tree-ring collections from eastern forests are typically not as climatically sensitive as western collections. Collections have been made from a diversity of sites, but it seems that collections from wet sites or sites of extremely shallow soils may have limited potential. The detrimental effect of crown crowding on sensitivity suggests preference be given to shade-tolerant species and to trees with less crowded crowns exposed in the canopy. Nonclimatic trends in tree-ring data are classified as growth trends and competition trends. Standardization of ring widths removes much of the growth trends, and merging individual tree chronologies into a mean collection chronology eliminates much of the competition trends of individual trees. Separation of ring width into earlywood and latewood widths, where possible, may be quite beneficial for non-pored and diffuse-porous species. However, this procedure seems to be of little value for ring-porous species.