CLIMATIC SIGNALS IN TREE-RINGS OF FAGUS-SYLVATICA L FROM THE CENTRAL APENNINES, ITALY

Tree-ring analysis provides empirical information on how trees respond to climate at scales, both spatial and temporal, that are not easily accessible to ecophysiological models. Dendroecological research in Italy has focused on conifers, but a prominent species in the Apennines is European beech (Fagus sylvatica L.), which forms pure and mixed forests over a broad elevation range. This research was aimed at assessing climatic influences on stem growth of old dominants in pure beech stands, after removing tree-ring variation related to stand development and anthropic manipulation. Stem samples collected at Abruzzo National Park, in the central Apennines, Italy, were analyzed using dendroecological methods. Ring width was modeled as an exponential function of both a deterministic component, the intrinsic growth trend, and a stochastic component, the autoregressive process. A tree-ring chronology was developed from the original ring-width measurements. Climatic data were provided by die nearby station of Pescasseroli, located within the beech altitudinal zone, at 1 150 m elevation. Surrounding weather stations were used to test reliability of the Pescasseroli station. Response functions were computed between the tree-ring chronology and monthly climatic variables for the 1950-1985 period, using a 12-month dendroclimatic window, from the current August to the previous September. The bootstrap method was used to assess significance of relations between tree growth and climate. The final tree-ring chronology was derived from a total of 8,554 ring widths and it spans more than three centuries, from 1672 to 1987. Ring-index series were best modeled as an AR(2) process with a positive first-order parameter and a negative second-order parameter. Precipitation and temperature regimes at the study area could be described as Mediterranean mountain climate. Beech growth showed a higher correlation with climatic data from Pescasseroli than with averages of all available stations. Response functions explained about half of the total variance in tree-ring data. December and July precipitations were positively related to beech growth, but April temperature was negatively related to it. December precipitation acts as an indicator of snowpack accumulation, which melts at lower rates when April temperature is lower, allowing a greater supply of water during the dry summer season. Radial growth of mature Fagus sylvatica L. trees at the study sites is limited by summer drought, and relies heavily on accumulation and melting of winter snowpack to meet its moisture demand in the growing period. Response functions computed using the bootstrap procedure were identical in shape to those computed using normality assumptions. However, bootstrapped confidence intervals were broader and allowed a clearer interpretation of climatic signals embedded in tree rings. The interpretation of response functions was tested by an ordinary least squares (OLS) multiple correlation model between beech radial growth and two climatic predictors, i.e. total precipitation from December until May and mean temperature during March and April. The model was highly significant, well behaved, and it explained about one third of the total variance. A three-dimensional plot of model residuals suggested the existence of nonlinear responses and trade-offs between variables.