Biodiversity will probably be threatened by climate change effects and the Mediterranean area is a well know hotspot of genetic diversity. Climatic data are a very important source of information for those studies and the aim of this work was to study and compare eight methods for spatial interpolation of climatic data and indices including parametric and non-parametric methods, deterministic, regressive and geostatistical. The Root Mean Square Error (RMSE), relative RMSE (rRMSE) and relative BIAS (rBIAS) were calculated to assess algorithm’s performances in a Mediterranean region. None of the eight methods performed much better than others with a very complex physiographic environment. The range of errors was very high and rRMSE varied from 3.8% to 295%. Anyway, even in case of low differences among methods and despite the necessity of the assumption of normality of data, the interpolation at local scale with parametric and geostatistical methods (e.g. kriging or cokriging) should be preferred to globally-interpolated climatic data due to the possibility to obtain the distribution of prediction’s error. © 2017, Patron Editore S.r.l. All rights reserved.

In young black pine plantations, the most valuable and interesting thinning scheme is mainly based on the positive selection of dominant and well-shaped trees to be candidates for carbon sequestration, timber production and natural regeneration. The mechanical stability of candidate trees is here a fundamental skill that must be taken into account and the slenderness ratio (HD) is one of the main indicators. HD has been recently proved to be correlated to the living whorl number (LWN) by Cantiani & Chiavetta (2015). In this study, the statistical model was re-calibrated in order to study the influence of soil fertility on the HD - Living whorls number (LWN) relationship. The fertility-balanced models estimated a different LWN threshold. The model for the highest fertility class (Site index 24) estimated 12 LWN (RMSE of 20%). Similarly, a lower value were detected for the other two fertility classes, SI20 and SI16, where 10 LWN were considered enough with an associated RMSE of 16% and 17% respectively. Compared to the general model provided by Cantiani & Chiavetta (11 LWN with 18% of RMSE) the site index approach improved the accuracy and reliability.