The general aim of this study is a synmorphological analysis of the loess flora and vegetation of Mezőföld region in Hungary, including the evaluation and interpretation of patterns at different spatial scales. At larger (geographical) spatial scale, mainly the autophenetic patterns (species distribution) were examined, but at smaller scales mainly the synphenetic (community level) patterns were analysed. In this case, the task was to study the pattern organization and textural coordination of the plant communities.
Three spatial scales determine the basic phenomena of patterns of loess vegetation in Mezőföld. The biogeographical scale is the scene of the regional preferences; at macrocoenological scale the stands of the communities, their spatial repetition, transition and mosaics can be observed; finally, at microcoenological scale the coexistence of populations, that is pattern organization can be explained.
The general questions are summarized as follows.
- At regional (biogeographical) scale [cf. Results 1:1-2] What are the general characteristics of the loess flora of Mezőföld? [1:3] Which asymmetries (floristic gradients) appear in the regional pattern? [1:4-5] What are the regional preferences of the plant species?
- At macrocoenological scale: [2:1-4] Which coenotaxa can be identified, what are their specifications? [2:5] How do the species differentiate between coenotaxa? [2:7-16] What are the species preferences along the grazing and the microclimatic gradients?
- At microcoenological scale: [3:1-6] Which are the characteristic values and the characteristic scale values of the spatial dependence of loess grasslands?
- Correlation between behaviour of species at different spatial scales: [4:1-4] Is there any correlation between the regional and macrocoenological preference of species groups? [4:5-6] Do the coalitions identified by different regional preferences differ from each other in degree of spatial organization?
Material and methods
Spatial scaling can be evaluated in a discontinuous way in the present study, because data only from separated spatial scale intervals were available. The spatial domains given by geographical and land-use historical factors are: (i) area of Mezőföld region (ca. 4400 km2); (ii) territory of sub-regions with constant geographical attributes (100 to 1000 km2); (iii) extension of isolated valley-systems wedged in the loess plateau, earthworks and loess cliffs (several hectares to several km2); (iv) slopes of loess valleys with homogeneous geomorphological and microclimatic characteristics (0.1-1 hectare). The sampling conditions are as follows:
- A number of valley-systems, loess cliffs and earthworks can be found all over the Mezőföld region, in these refuges many species of loess flora could survive. I recorded floristic samples in 47 localities, with 36 detailed species lists. The number of the days for floristic and coenological samples was 249.
- Many types of loess vegetation grow on the slopes of loess valleys, and the repetitions of coenotaxa are realized in many situations. For differentiating and characterizing coenotaxa, I recorded 170 coenological relevés using 2 m x 2 m quadrats, at different combinations of exposition and slope. For studying the repetition of coenotaxa in the landscape, I mapped the vegetation of five valley-systems.
- I carried out microclimate-measurements and soil sampling (with analysis of several physical and chemical variables) to determine correlations between coenotaxa and abiotic factors. The number of soil samples was 142. The depth of soil cores was between 0 and 25 cm, and in several cases 25-50 cm.
- Since the anthropogenous impact of grasslands has different intensity in the different valleys, different stages of degradation can be observed. In several cases (e.g. in the valley-system at Belsőbáránd) the degradation gradient is obvious along the series of sub-valleys of a main-valley; in this situation the biogeographical conditions are constant. I made 66 coenological relevés in NE and SW exposition, in 5 and 6 stages of degradation gradient, with 6-6 repetitions each. The coenological samples are completed by floristic samples and microcoenological transects in all sampling locations. Each transect consisted of 500 microquadrats of 10 cm x 10 cm area. The presence - absence data of species were recorded.
- A grid sample was taken in a xeromesophilous forest-steppe grassland in which grazing was cancelled many years ago, but the invasion of Brachypodium pinnatum started at the same time. The extension of the grid was 10 m x 5 m, the area of the cells was 20 cm x 20 cm, so the number of the cells was 1250. In the cells, the abundance data (relative cover) of species were recorded. A secondary sampling procedure was employed to evaluate spatial scaling.
- In sites where two sub-walleys are close to each other in a walley-system, an exposition gradient can be recognized that corresponds to a microclimatic gradient for the vegetation. If the gradient is relatively long enough (e.g. it runs from NE to W exposition) then the transition of coenotaxa, that is the coenocline can take place. I examined such a microclimatic gradient in the valley-system at Belsőbáránd. I made 22 coenological relevés situated at the same distance from each other along the ca. 200 m long line. Parallel with it, a 173.2 m long transect was also sampled, which consisted of 1732 microquadrats. In the microquadrats of 10 cm x 10 cm size the binary data of species were recorded.
- For the microcoenological study of primary succession I sampled a total of 9 transects from three vegetation stages: pioneer communities of loess cliffs; transitional, open community on steep slopes; and more or less closed, xerophilous loess grasslands in S or SW exposition. Each 50 m long transect contained 500 microquadrats of 10 cm x 10 cm area.
For analysing floristic samples, I characterized the sampling locations with distance attributes (distance from NW and SW border of Mezőföld), and the species with syntaxonomic, vertical distribution and area type attributes. The correlations between distances and biotic variables were tested by nonparametric statistics.
The coenological relevés were ordinated and grouped by multivariate-procedures (principal component and principal coordinate analysis, divisive and agglomerative clustering), and I performed a so-called coalition-ordination by using summarized cover data of species groups that have the same coenological preferences. I described the coenotaxa with several textural attributes (e.g. coordination, coenological and floristic heterogenity, Shannon-diversity, evenness), and obtained a diversity ordering for them. For analysing the relationship between coenological attributes and different abiotic factors, nonparametric statistics were used.
The basis of the microcoenological studies was formed by 21 samples which comprise a ca. 1.1 km long transect consisting of 11232 microquadrats. For information statistical analysis (including spatial series approach) I applied my INFOTHEM program. To remove the effect of rare species I reduced the species number of samples in different proportions, so I used different sub-communities for analyses. For random references, in most cases the complete randomization, the random shift, and the secondary reference were selected. I characterized the samples with the maximum values of syncretic functions and their spatial scale points. In several cases the diacretic functions of species were also calculated.
1. Characteristics of regional floristic patterns in the Mezőföld region
 A total of 331 native, non-ruderal steppe, forest-steppe and forest species have been detected. On the basis of data from 47 sampling locations I made regional distribution maps for 9 species having different specific distributional patterns.
 The flora of Mezőföld can be described and distinguished from the other loess regions by spectra of species attributes. The proportion of submediterranean and continental species is relatively large (24% and 27%). A quarter of the species is forest or forest-steppe element, there are 60 herbaceous forest/forest-steppe species. 29 species have their vertical distribution from colline or montane regions and higher.
 A floristic gradient can be demonstrated regarding the proportion of forest/forest-steppe and colline-montane species. The number of these species decreases eastwards from the NW and SW borders of Mezőföld. On the other hand, there is no significant spatial trend in the distribution of submediterranean or continental species. It is hypothesized that regions of Dunántúli-középhegység and Dunántúli-dombvidék are propagulum sources for colline-montane species which have the most characteristic asymmetrical distribution.
 The rare species with less than 20% frequency, and species with frequency of 20-60% have asymmetrical distribution along the W-E gradient in Mezőföld. The number of species distributed in western part of region is many times the number of species that are common in the eastern part, so the disappearance of western species is the most important factor to form a floristic gradient. This phenomenon indicates the montane origin of the lowland flora.
 There are many different factors causing and influencing the characteristics of loess flora of Mezőföld. Most important are the macroclimatic and geomorphological factors, but land-use history also plays similar role. Moreover, the Quaternary and Tertiary sand sediments between loess layers also have to be considered.
2. Macrocoenological characters of loess grasslands
 The compositional continuity of coenological relevés is broken in the state space of summarized cover of coalitions that have different coenological preferences. However the ordination methods using population abundances yielded indistinct segregation only. This result emphasises the importance of coalitions to form compositional nodes of vegetation.
 The multivariate analyses of the 170 coenological relevés of loess grasslands resulted two main coenotaxa, and 11 sub-types. The first main type is the xeromesophilous forest-steppe grassland containing the following sub-types: Brachypodium-type (Euphorbio pannonicae - Brachypodietum association), Festuca rupicola-type, transitional type between the previous two, and degraded, Agropyron intermedium-dominated forest-steppe. The second main type is the shortgrass, xerophilous grassland, with the sub-types dominated by Festuca pseudovina, Bothriochloa ischaemum, Salvia nemorosa, Bromus inermis, Stipa capillata or Chrysopogon gryllus, and the coenotaxa of upper margin of valleys. The types can be characterized by physiognomical attributes, dominant and fidelic species. For the last feature, the species have to be ordered by their differentiating ability.
 The two main coenotaxa well differentiate between slope expositions. While the occurence of the Brachypodium-dominated forest-steppe can be expected in N-NE exposition at 15-40° angle of slope, the shortgrass, xerophilous grassland can be found nearly exclusively in SW-W exposition.
 The Brachypodium-type gasslands have the largest number of species and strongest coordination. Both transitional coenotaxa can be characterized with the most uniform dominance distribution, so they have the largest diversity values.
 The proportion of continental species is almost constant in each coenotaxon, they represent a third part of species pools. In the forest-steppe coenotaxa, the number of colline-montane species is larger (4-6% vs. 0-3%), while the proportion of submediterranean species is smaller compared to xerophilous grasslands (9-12% vs. 13-19%). The ratio of the forest-steppe species in Brachypodium-type is twice as large as in the other coenotaxa.
 The difference of average temperature in the microclimate of soil under bare surface between slopes with NE and SW exposition is 1°C, the probable background of the marked coenological difference in vegetation. To explain this observation, positive feedback mechanisms can be supposed, in which the humus content (originated from different quantity of biomass), and the humidity of soil (reflecting the quantity of humus) can play important role.
 Along a degradation gradient, that the intensity of grazing increases so much the proportion of forest and forest-steppe species decreases. This feature confirms the existence of process by which soil humidity decreases during degradation.
 Regarding several textural parameters (e.g. evenness) there is no simple relationship (monotonity or unimodality) to degradation, in neither NE nor SW exposition. Considering fire also as degrading factor, diversity goes through a maximum in forest-steppe, but it decreases in xerophilous loess grassland. With the increase of grazing intensity, in the forest-steppe the total cover, but in the steppe the species number decreases.
 The species number of coenological relevés is usually larger in N facing slope than in southern slopes, independently of the presence of Brachypodium. If Brachypodium is present the species number is larger as long as the cover of Brachypodium pinnatum remains below 20%. When the cover of Brachypodium exceeds 20%, the number of species will decrease.
 The invasion of Brachypodium pinnatum influences the species number and diversity of local patterns depending on spatial scale. At smaller scale (0.04 m2 - 1 m2) the effect is significantly negative, but at larger scale (4 m2) the correlations are not significant.
 A topographical allocation of similarity pattern of a forest-steppe stand in the sampling grid was performed. A speciality of the allocated similarity pattern is that the cells belonging to the same cluster exhibit non-random but aggregated pattern on the grid, otherwise some of them occur separated.
 An aspect of local invasive nature of Brachypodium pinnatum that is its intense invasion limits the local spreading of forest-steppe species, however they are typical elements of Brachypodium dominated coenotaxa. This feature manifests itself not only in the number of these species, but in proportion, as well. The importance of scaling could be demonstrated again.
 The effect of the examined microclimatic gradient can be revealed at the traditional coenological spatial scale, because it forms a coenocline.
 Whereas we can expect monotonous decrease of species number along microclimatic gradient between forest-steppe and shortgrass steppe, in the transition it has its maximum value indicating coenotone. In the transition the species-cover evenness and the coordination values decrease. These characteristics indicate the organization in community at macrocoenological scale.
 Some variables, e.g. the proportion of forest and forest-steppe species exhibit monotonous decrease along a coenocline. This shows that the behaviour of the species depending strongly on the microclimatic factors is more or less independent of community organization, but it follows only the continuous change of abiotic factors.
 Several species can be found mainly or exclusively in transitional zone of coenotaxa along microclimatic gradient. These species probably utilize the possibilities in coenotone situation given by decreasing competition pressure of grass species predominating the different loess communities.
3. Pattern organization of loess grasslands at microcoenological scale
 The actual values of associatum indicates the organization of spatial pattern in all loess grassland stands. The spatial dependences are more than results of textural constrains; they originated firstly from non-random (aggregated) distribution of species, and partly from non-random co-occurences (but spatial dependences) of species patterns.
 In some cases the sub-communities (which are made by reducing the complete sample into given number of species) exhibit different spatial dependence values: the associatum-relation of compared stand can turn over. This phenomenon indicates that the dominant species have very different role to form spatial patterns.
 During primary succession in loess vegetation, the organization, complexity and heterogenity of the community increase, reaching their maximum values at smaller spatial scale. This shift of scale is valid in case of sub-communities, as well.
 The structural parameters (e.g. patchiness and spatial dependences) of the forest-steppe (growing in more humid and well balanced microclimate) reach their maximum values at smaller spatial scale than the xerophilous shortgrass steppe.
 Along the NE-W gradient the spatial pattern indicates the existence of coenotone. It can be manifested in increasing associatum values in the transitional zone of two characteristic coenotaxa, but we can expect a monotonous trend along the gradient.
 The response of forest-steppe and steppe to degradation is different. In forest-steppe a relatively moderate grazing causes an increase of associatum and florula diversity, but intensive grazing leads to decreasing of spatial dependences. In the xerophilous steppe the associatum value decreases with degradation. Otherwise, in case of sub-communities containing only a few species pattern organization is the strongest in the most degradated stages indicating that the pattern forming role of dominant species becomes stronger.
4. Relations between patterns at different spatial scales
 The species which have asymmetrical regional distribution, thus they show differences between NW-SW and eastern subregion of Mezőföld, have differenciating capability regarding their coenological preferences, as well.
 In the formation of species pool of the different loess grassland stands, their membership in a coenotaxon is less important than the floristic situation given by regional location. It means that the communities assembling from the given subregional flora become well-determined coenotaxa in any biogeographical situation.
 During degradation the communities „select" the regionally frequent species in constant ratio, but the number of regionally non-frequent and rare species becomes smaller and smaller.
 The species having regional asymmetry in Mezőföld also exhibit trend-like behaviour during degradation and along the microclimatic gradient. This phenomenon indicates that the process of the degradation at smaller (microclimatic) scale is very similar to the floristic process at larger (macroclimatic) scale across the Mezőföld.
 The behaviour of the two coalitions (which are responsible for the separation of two main coenotaxa) at microcoenological scale differs from their roles at macrocoenological scale, so the preferences of coalition are not the same regarding between- and within stand patterns.
 The preferences of species at macrocenological and regional scales are not coherent with their microcoenological behaviour. The role of the species belonging to different species group-pairs (identified at regional or macrocoenological scale: forest-steppe vs. steppe, western vs. eastern, frequent vs. non-frequent, and two characteristic coalitions) are not different in the average participation in spatial pattern organization of communities.