Assessment of site characteristics as predictors of the vulnerability of Norway spruce (Picea abies Karst.) stands to attack by Ips typographus L-(Col., Scolytidae)

The intensity of bark beetle Ips typographus L. (Col., Scolytidae) attack o n Norway spruce (Picea abies Karst.) is known to vary greatly among stands. In a control strategy approach, previous studies investigated the relation ships between the variability in intensity off. typographus attack and site characteristics such as stand age and altitude, mean tree circumference, g rowth rate and nearest-neighbour distance, soil moisture, pH in H2O and KCl , and soil contents of C, N, K, P, Mg, Ca, Fe, Cu, Zn and Mn. The data anal ysis method used in these studies was mainly the multiple linear regression , with the mean number of attacks per spruce tree in a stand as Variable to explain. Previous results showed that the expected vulnerability of a Norw ay spruce stand to attack by I. typographus can be estimated on the basis o f simple information of easy access to the forester, when the data on the s tand in question is used with others for fitting the regression model. Pred iction of the vulnerability of a stand, without including its data in the f itting of the model, was shown to be more approximate. Therefore, the objec tives of this study were: (I) to improve the performance of models predicti ng the vulnerability of Norway spruce stands to attack by I. typographus, b ased on site characteristics; (2) to assess the stability of such predictiv e models when these are built using a moderate number of stands; and (3) to incorporate the resulting information in a global approach to control and prevention. Published data were re-analysed for these purposes. A jackknife d multiple linear regression procedure, in which each stand in turn is disc arded when fitting the model (jackknife replication), is presented. A great variability in the models fitted, depending on the stand discarded, is obs erved. For instance, the number of explanatory variables retained ranges fr om one (i.e. soil P content, for five jackknife replications) to 10 (for on e jackknife replication), for R-2-values ranging from 0.5 to 1.0 and for on e influential stand (i.e. the same stand characterized by an atypically low number of insect attacks compared to other stands with similar soil P cont ent) against many influential stands. Differences between the model finally selected here using the revisited data and the models proposed earlier are discussed. A path analysis diagram is proposed for a more comprehensive mo delling of Norway spruce stand vulnerability to I. typographus attack, base d on site characteristics.