Matskási István (szerk.): A Magyar Természettudományi Múzeum évkönyve 83. (Budapest 1991)

Dobolyi, K., Kováts, D. , Szerdahelyi, T. ; Szollát, Gy.: Vegetation studies on the rocky grasslands of Odvas Hill (Budaörs, Hungary)

facing slopes too (here the composition of species partly changes together with the ratio of domination, while the grassland becomes more open). Disregarding the natural erosive phenomena of the steep slopes, notable signs of degradation have been found along the path which is climbing up to the ridge of the hill from the E side, but first of all from the W side and it can be found on the top of the hill too (weeds, natural pioneer species, native species tolerated disturbance). This degradation is naturally caused by heavy tourism. This kind of disturbance has been found on the S, SW side, respectively on SE side of the hill in a lower degree. RESULTS OF COMPUTER ANALYSES We gained 4 larger sample groups as well as some smaller ones from the results of classification. The dendrogram has been cut at 0.18 similarity level. The 4 larger sample groups consist of: I = 21, II = 19, HI = 16, IV = 18 samples (Fig. 2). PCA diagram shows also the considerably similar grouping of samples (Fig. 3) where the samples have 4 separate point groups. Fig. 2. Dendrogram of a matrix of Jaccard index of similarity The principal component analysis - according to its nature - usually gives a more realistic feature about the position of the sample plots. This analysis is especially important in solving problematic arrangements. It is interesting that serial number of samples from 22 to 31 (W exposure, 5-10 angle of slope, 95 % grass covering) is separated by the classification method from sample III groups (NW exposure, 35 angle of slope, 45 % grass covering), while they are not separated so sharply by ordination. It is remarkable that the samples marked on the southfacing slopes as groups I and IV being far from each other by every numerical methods were contradictory to expection. Their similarity refers to the sample position only on the horizontal axis in the diagram of ordination. At the same time sample group IV shows the greatest similarity with sample group III, according to both analysis. The reason of difference in numerous further characters between sample groups I and IV (over the subjective factors,