INVESTIGATIONS ON THE CORRELATION PATTERN IN EVEN-AGED STANDS OF LARCH .3. STATIC DESCRIPTION OF PHENOTYPIC CORRELATIONS BETWEEN NEIGHBORING OBSERVATIONS

The main topic of this study is a static description of spatial neighb ourhood correlation patterns of stands of larch, i.e. an analysis of t he correlations between neighbours at only one point in time. For thes e investigations extensive data sets from 9 field trials with a varyin g number (8 to 23) of ''entries'' (Larix europaea, Larix leptolepis, h ybrids) have been used. The trials are slightly different in their age s (7 to 11 years). Single tree measurements of these trials were avail able for the traits height and diameter at breast height. Additionally , the diameter values have been transformed and analysed as individual basal areas. The correlative structure of these trials for measuremen ts of neighbouring individuals (regular square spacing with spacings 1 .5 m x 1.5 m, artificial thinning procedures for only 2 of the trials) has been described by 12 different correlation coefficients. These co efficients are defined by considering quite different spatial configur ations of competitive neighbourhoods. In these procedures and analyses the diagonally located neighbours of a subject tree and its missing n eighbours too are explicitly considered. For the trait height one obta ins for all trials a quite uniform correlation pattern: positive corre lations with intermediate or small numerical amount between the trait measurement of a subject tree and the corresponding Value of its neigh bourhood (for different spatial definitions of neighbourhood) and nega tive correlations of similar numerical amount between the trait measur ement of a subject tree and the number of missing neighbours in its ne ighbourhood. The correlations for the trait diameter are throughput ve ry small. In most of the field trials one obtains stronger spatial nei ghbourhood relationships for height than for diameter. The correlation between height and diameter varies between 0.63 and 0.94. This correl ation, therefore, exhibits a considerable variability. The estimates o f the correlation coefficients as well as the detected relationships a re of high reliability since the correlation calculations of this stud y are based on large numbers of observations (smallest sample size=199 , largest sample size=6649).