L. Hably szerk.: Studia Botanica Hungarica 19. 1986 (Budapest, 1986)

Németh, Ferenc; Iványi, E.: Morphometrical studies on the Hungarian representatives of Ophrys scolopax Cav. agg. (Orchidaceae)

dicates the relative phenophase of the plant at the time of observation. This can be used subsequently to correct the other (especially quantitative! data. The present analysis, however, does not incorporate this correction feature. The randomized sampling has been a general problem of this subject. In the small population at Pécs we have observed all the measurable specimens but these samplings are certainly disturbed by genetic drift. At Kunpeszér, on the contrary, the rather large population and area caused some problems, as we did not want randomize in the whole area in order to find the markings more easily. Starting from a randomly chosen plant, we observed the 30-50 nearest neighbour around it. Unfortunately the spatial distribution of the different morphological types seems to be contagious, i.e. it is uncertain whether the sampling method is correct. An examination of the spatial distribution of the Ophrys-types is in the process (LOCSMANDI & BORI pers. comm.). Transitional Fig. 3. Shape variants of the appendix ANALYSIS The observed plant individuals belong to two subsamples on the basis of their locality (Pécs and Kunpeszér) and to three ones based on their time of observation (1983, 1984, 1985). There is no overlap between the first two groups. We did not consider the occurrence of the same plants in the consequent years to be a disturbant factor, because they contribute to the reproduction of the population to two three times greater extent than the other specimens, flowering only one time in the three years; it was considered to be a kind of self-weighting. Furthermore, the number of the consequently flowering plants was relatively small (Table 1). The univariate analyses were carried out on a COMMODORE 64 personal computer by a commercial statistical programme packaged SES 1, 2, 3). The multivariate analyses ran on a Hungarian TP A computer and the programme for it was written by KOVÁCS L. and KOVÁCS G. Both of the programmes required a certain transformation of the qualitative data, i.e. they have been transformed onto a binary nominal scale, as follows: - colour of the sepals and petals was considered as a combination of presence and absence of two colour factors, that of the green and pink ones. The possible code numbers of the colours are 1,0 (green), 0,1 (pink), 1,1 (greenish pink) and 0,0 (white). - colour of the appendix was definied similarly: 1,0 (yellow), 0,1 (brown), 1,1 (brownish-yellow). The code 0,0 was not interpreted. - shape of the appendix was definied as the combinations of two shape-factors, i.e. tridentate and cut. The codes were 1,0 (tridentate), 0,1 (cut) and 1,1 (transitional), see Fig. 3. The code 0,0 was not interpreted. Such coding may not be fully correct for the two last characters, but we could not find any other method to transform them from an ordinal to a nominal scale. First we determined the distribution of the variables in the subsamples and then in the total sample. An example of the distribution of the length of the horn is shown in histograms (Fig. 4). The frequency data of the qualitative characters are given in Table 2. Normality of the distribution of the quantitative characters was tested by chi^ test and most of the tests were negative. We calculated the means and standard deviations of the characters (Table 3). Paired significance tests were made for each character between the following subsamples: Pécs-Kunpeszér (locality), 1983-1984, 1983-1985, 1984-1985 (generation), Pécs 1983-1984, Pécs 1983-1985, Pécs 1984-1985, Kunpeszér 1983-1984, Kunpeszér 1983-1985, Kunpeszér 1984-1985 (generations at the same place), i.e. F and T tests for the qualitative characters and chi^ test for the qualitative ones (Tables 4. and 5). The multivariate analysis was made in two steps. First a similarity matrix was composed including both the qualitative and quantitative characters (GOWER 1971). The calculation is as follows: V V S..= w s / IT w. íj H ilk ljk / l-l ljk ,