DIGITALNA ARHIVA ŠUMARSKOG LISTA
prilagođeno pretraživanje po punom tekstu
ŠUMARSKI LIST 3-4/2011 str. 52 <-- 52 --> PDF |
Z. Sedlar, V. Hršak, R. Šoštarić: NUMERICALAND PHYTOSOCIOLOGICALANALYSIS OF ... Šumarski list br. 3–4, CXXXV (2011), 1-3 Figure 1.Geographical position of the study area. Islands of Brač (1), Hvar (2), Korčula (3), the Pelješac Peninsula (4) and Biokovo Mountain (5). Slika 1. Zemljopisni položaj proučavanih područja. Otoci Brač (1), Hvar (2), Korčula (3), poluotok Pelješac (4) Biokovo (5). ranean we used data from Dalmatia, Eastern, Central and West Mediterranean taken from Bergmeier 1990 and 2002,Brullo &al. 2001Sánchez-Gómez & Data processing – The relevé table based on the Braun-Blanquet scale was transformed into a data matrix using theVan der Maarel 1979 ordinal transformation for further statistical analysis.The table was purified before transformation by removing species with only one and two appearances, thus reducing the total number of species from 239 to 193 species.A cluster analysis was performed on this data matrix using the Bray-Curtis distance (Bray & Curtis 1957) as a resemblance measure, and the UPGMA(group average) linkage as clustering method. Indicator species analysis (Dufrene &Legendre 1997) was then performed on the groups defined by the clustering results to obtain the optimal number of clusters. Relevés were clustered into a different number of clusters (3, 5, 6, 7, 9, 12 and 15) using a cluster dendrogram.The optimal number of Results – According to the ISA analysis and its method of dendrogram pruning, the obtained optimal number of clusters was 7. Figure 2 represents the group average clustering dendrogram divided according to this result. In our research, on these newly formed clusters we can distinguish different syntaxa, determined by the relevé authors and grouped in clusters based on the cluster analysis and ISA. Inspection of this dendrogram shows a clear separation of the cluster marked as number 7, which includes relevés of the higher (1150–1400 m) parts of Biokovo, Alcaraz 1992.As in all mentioned areas, from which data was taken, there is a different taxon of Pinus nigra complex.All were treated as one taxonPinus nigras.l. Obrada podataka clusters in the cluster analysis was obtained using the lowest average probability value, computed in separate indicator species analyses for each number of clusters (McCune & Grace 2002). The probability value was computed using the Monte Carlo permutation test with 4999 random permutations. Species with indicator values over 50.0 (IV> 50) and a probability level under 0.05 (p<0.05) were then marked as indicator species (McCune & Grace 2002). Synoptic table with combined data from Dalmatia and other parts of Mediterranean was used to make data matrix on which Nonmetric multidimensional scaling (NMS) using the Bray-Curtis distance (Bray &Curtis 1957) as a resemblance measure was performed. Numerical analyses were done using PCOrd 5.0 software (McCune & Mefford1999). Rezultati taken by Domac 1965 and phytosociologically defined as theJunipero sibiricae -Pinetum dalmaticae association. This cluster is separated at the beginning of clustering and is completely distinguished from all other relevé groups. The syntaxonomical classification of relevés from cluster 7 is the object of this research. Clusters 1-6 are separated on the other side and are composed of relevés from the coastal part of the Dalmatian pine distribution area and of three relevés from the lower part of Biokovo Mountain.Cluster 1 comprises relevés taken by Domac 1965 on the islands of |