A GRID-BASED, SATELLITE-IMAGE SUPPORTED, MULTI-ATTRIBUTED VEGETATION MAPPING METHOD (MÉTA)

Transcription

1 Folia Geobotanica 42: , 2007 A GRID-BASED, SATELLITE-IMAGE SUPPORTED, MULTI-ATTRIBUTED VEGETATION MAPPING METHOD (MÉTA) Zsolt Molnár 1), Sándor Bartha 1), Tibor Seregélyes 2), Eszter Illyés 1), Zoltán Botta-Dukát 1), Gábor Tímár 3), Ferenc Horváth 1), András Révész 1), András Kun 1), János Bölöni 1), Marianna Biró 4), László Bodonczi 5), Áron Deák József 6), Péter Fogarasi 7), András Horváth 1), István Isépy 8), László Karas 9), Ferenc Kecskés 10), Csaba Molnár 11), Adrienne Ortmann-né Ajkai 12) & Szilvia Rév 13) 1) Institute of Ecology and Botany of the Hungarian Academy of Sciences, H-2163 Vácrátót, Hungary; molnar@botanika.hu (Molnár), sanyi@botanika.hu (Bartha), illyese@fre .hu (Illyés), bdz@botanika.hu (Botta-Dukát), jboloni@botanika.hu (Bölöni), kun@botanika.hu (Kun), ahorvath@botanika.hu (Horváth A.), horvfe@botanika.hu (Horváth F.), revesz@botanika.hu (Révész); 2) Tünde köz 5, H-2481 Velence, Hungary; calypso@mail.datanet.hu; 3) State Forest Service, H-2600 Vác, Hungary; timar.gabor@aesz.hu; 4) Alkotmány u. 2-4., H-2163 Vácrátót, Hungary; mariann@botanika.hu; 5) Alszer 28/a, H-9941 Õriszentpéter, Hungary; bodonczi@axelero.hu; 6) University of Szeged, Department of Climatology and Landscape Ecology, H-6722 Szeged, Hungary; aron@geo.u-szeged.hu; 7) Fazekas u. 64/a, H-2890 Tata, Hungary; kickingbird@fre .hu; 8) Eötvös Loránd University, Botanical Gardens, H-1083 Budapest, Hungary; isepy@yahoo.com; 9) Team-Work Consulting and Training, H-2000 Szentendre, Hungary; laszlo.karas@axelero.hu; 10) Budai Secondary School Táncsics Mihály, English Bilingual Grammarschool, H-1126 Budapest, Hungary; kecskes@tmgbp.hu; 11) István u. 52., H-3036 Gyöngyöstarján, Hungary; minuartia@fre .hu; 12) Báránytetõ u. 2., H-7635 Pécs, Hungary; aadrienn@dravanet.hu; 13) E-misszió Environmental Association, H-4400 Nyíregyháza, Hungary; revszilvi@e-misszio.hu Abstract: In this paper we present the main characteristics of a new, grid-based, landscape-ecology-oriented, satellite-image supported, field vegetation mapping method, called MÉTA (MÉTA stands for Magyarországi Élõhelyek Térképi Adatbázisa: GIS Database of the Hungarian Habitats). The goals of the MÉTA method based vegetation mapping program (MÉTA mapping) include the following: (1) to map the actual (semi-)natural vegetation of Hungary; (2) to evaluate Hungarian (semi-)natural vegetation heritage for conservation purposes; (3) to evaluate the present state of Hungarian landscapes from a vegetation point of view; (4) to collect vegetation and landscape ecological data for the prognosis of future changes of vegetation and the landscape. Spatial resolution, mapped attributes and mapping methods were developed to meet these goals. The MÉTA method uses a hexagon grid with cells of 35 hectares. In the hexagons, habitat types are listed, then the area, naturalness-based habitat quality, spatial pattern in the hexagon, effect of the neighbourhood, connectedness, and threats are recorded for each habitat type. Other attributes are recorded in the hexagons: potential natural vegetation, area occupied by invasive plant species, area of old fields, land use of grasslands, and landscape health status (naturalness and regeneration potential of the landscape in general). One hundred hexagons form a quadrat mainly for practical, organizational reasons, but also for collecting certain vegetation data at this spatial scale. For standardization of mapping, three different pre-printed data sheets and two different kinds of guides have been composed (Mapping Guide and Habitat Guide) and field trainings were organized. For standardization of estimation of naturalness-based habitat quality and regeneration potential field examples were prepared for each habitat type and each category of these attributes.

2 226 Zs. Molnár et al. Keywords: GIS database, Habitat quality, Landscape ecology, Large area survey, Nature conservation, Regeneration potential, Standardization Nomenclature: SIMON (2000) INTRODUCTION For conservation of our natural heritage and for developing sustainable landscape management strategies for Europe, it is essential to know what kind of interventions must be, should be or should not be done and in which particular areas (Habitats Directive, 92/43/ECC). Partly for these reasons the phytosociology-based knowledge of actual vegetation has re-gained its importance in many countries of Europe. Responses of vegetation science to the increasing landscape and vegetation knowledge are manifested in the recent phytosociological monographs and swiftly developing databases with detailed documentation of European vegetation and habitat types (MUCINA et al. 1993, RODWELL , SCHAMINÉE et al , FREMSTAD 1997, CHYTRÝ et al. 2001, RODWELL et al. 2002, STANOVÁ & VALACHOVIÈ 2002), the potential natural vegetation maps of countries (e.g. ZÓLYOMI 1989, NEUHÄUSLOVÁ & MORAVEC 1997), regions (NIKLFELD 1973) and of the whole of Europe (BOHN & NEUHÄUSL ), the ecological map of Europe (OZENDA & BOREL 2000) and vegetation mappings covering larger areas (e.g. BARR et al. 1993, RUZ IÈKOVÁ et al. 1996, LOIDI 1999, MÄGI &LUTSAR 2001, KALIGARIÆ et al. 2003, NOWOTNY 2003, GUTH &KUÈERA 2005). To satisfy at least part of the demands for actual landscape knowledge, the European Union has established the CORINE Land Cover mapping project (CLC). This is the one and only large area actual landscape mapping project for all of Europe (ANONYMOUS 1995, BÜTTNER et al. 1995, 2000, 2002). CLC developed a new, efficient and standardized methodology to accurately map the landscape of large areas (based on the interpretation of satellite images). Although land-cover maps can be considered as a certain kind of detailed actual vegetation map, they have serious deficiencies from the vegetation and landscape ecologists point of view: too many (semi-)natural habitat types are aggregated in one land-cover category, so they are not really suitable for the analysis of (semi-)natural areas (CUDLÍN et al. 2005). Land-cover maps also do not provide information for example on naturalness and diversity of vegetation patches. These features can only be observed in the field (MILLINGTON & ALEXANDER 2000) and with the proper background in vegetation science (FULLER et al. 1998, WYATT 2000). Though satellite images provide ever greater help for field botanists, and methodological developments have promising results (KERAMITSOGLOU et al. 2005, FANELLI et al. 2005), even the newest images (IKONOS, Quickbird) with the newest technologies have serious limitations in mapping habitat quality and diversity (see the very limited results of BOCK et al and KERAMITSOGLOU et al. 2005). Field mapping of actual vegetation has old traditions in Europe because a vegetation map is one of the easiest ways of communicating knowledge on vegetation (KÜCHLER 1967, FEKETE 1998). Nowadays, however, the traditional methodology has to face new challenges: larger areas, increased number of desired attributes (not only vegetation type, but also quality, representativity or fragmentation), and more frequent repetition (monitoring). For scientific and applied purposes actual vegetation data are needed for ever larger areas in ever higher

3 MÉTA vegetation mapping 227 spatial and thematic resolution, and in multi-attributed, GIS databases. Fortunately, the development of technology (remote sensing, digital data processing, advanced sampling methods) equips us to elaborate and utilize new, more efficient and more complex, more standardized vegetation mapping methods (KÜCHLER & ZONNEVELD 1988, MILLINGTON & ALEXANDER 2000). At the same time, however, very few publications deal with the problems of the methodological improvement and the necessary standardization (CHERRILL & MCCLEAN 1995, 1999, MILLINGTON & ALEXANDER 2000). However, there are some case studies on the repeatability of local field vegetation mapping (JANSSEN 2004, SANDERS et al. 2004) and on the use of satellite images for habitat mapping (BOCK et al. 2005, FANELLI et al. 2005, KERAMITSOGLOU et al. 2005). Two field-survey based fine-scale large area vegetation mappings exist in Europe, both motivated by the European Natura 2000 program: in Spain (RIVAS-MARTÍNEZ 1994, LOIDI 1999), and in the Czech Republic (GUTH & KUÈERA 2005). Neither of them discusses the methodological problems of their methods. In this paper we present (1) the development of a new vegetation mapping method; (2) the attributes to be mapped; (3) the rules of field mapping and (4) the process of standardization. Because the method was developed for large area field mappings, and was used in the so-called MÉTA mapping project (BARTHA et al. 2002), we also present those aspects of the MÉTA project that we think are relevant for future application and further development of the MÉTA method. DEVELOPMENT OF THE MÉTA METHOD Having reviewed the recent and near-future scientific and practical needs for any kind of Hungarian vegetation data, we realized that a country-level vegetation mapping project has to reach the following goals: (1) mapping in detail the actual (semi-)natural vegetation of Hungary ( km 2 ); (2) evaluating our (semi-)natural vegetation heritage for conservation purposes; (3) evaluating the present state of our landscapes from a vegetation point of view; (4) collecting vegetation and landscape ecological data for the prognosis of future vegetation and landscape changes. The MÉTA method was developed between 2001 and We had two different sources of experience: (1) the skills gained by the long tradition of phytosociological vegetation mapping (e.g. KÜCHLER 1967, KÜCHLER & ZONNEVELD 1988, DIERSCHKE 1994, FALIÑSKI 1994a,b, GRECO et al. 1994, WATERTON 1997; in Hungary: SOÓ &ZÓLYOMI 1951, ZÓLYOMI et al. 1954, JAKUCS 1965, FEKETE 1980, 1998, BAGI 1991, 1997), and (2) the experiences of non-traditional vegetation mappings and vegetation database constructions, such as: - CORINE Land Cover and Biotope mapping (BÜTTNER et al. 1995, 2000, 2002, KOVÁCS-LÁNG et al. 1997), - British Countryside Survey (ANONYMOUS 1990, BARR et al. 1993, CHERRILL & MCCLEAN 1995, 1999, SMART et al. 2003), - Hungarian Biodiversity Monitoring System (KOVÁCS-LÁNG et al. 2000, MOLNÁR et al. 1998, KUN & MOLNÁR 1999), - Preparation for Implementing the Habitats Directive (92/43/ECC) in Hungary (HORVÁTH et al. 2000, MOLNÁR et al. 2001a,b),

4 228 Zs. Molnár et al. Fig quadrats covering the whole area of Hungary. There are approximately 100 hexagons in each quadrat ( hexagons in the country). The habitat types and their areal proportion in the hexagon are shown. The radius of the circle corresponds to the total area of (semi-)natural vegetation in the hexagon, the proportions of the different habitat types are shown as well. - woodland-type mapping (DANSZKY &ROTT 1964) - conservation-oriented vegetation mappings (e.g. SEREGÉLYES & CSOMÓS 1995, BIRÓ et al. 2006a,b). The method was developed in an iterative process based on a series of workshops, field tests (in 14 landscapes) and 9 trainings in 34 different landscapes with the contribution of 160 botanists. Constraints As usual in large-area biological surveys, the most important constraint to overcome during the development of the MÉTA method were time and money (BURBIDGE 1991). The actual vegetation map of Hungary had to be prepared as soon as possible. The need for this map arose as early as in the 1950s, but the mapping could not start because of the lack of capacity (SOÓ & ZÓLYOMI 1951, FEKETE 1998), because the estimated time of work required to accomplish this map on the scale of 1: is about field days. Prior to the MÉTA mapping, a questionnaire survey was established, which revealed that the total capacity of ca. 300 Hungarian botanists is ca person-days of field work annually, from which ca person-days can be spent on this vegetation mapping (taking the load, means of subsistence, field experience and the distribution of the experts age into consideration see MORAVEC 1995). If we have 2.5 growing seasons to accomplish the

5 MÉTA vegetation mapping 229 Table 1. Applicability of different Hungarian external data sources for vegetation mapping. +++ very well applicable, ++ well applicable, + poorly applicable, - gives no suitable information. Habitat type Military Aerial Satellite Forestry Soil/geol. map photo image database maps Euhydrophyte and non-woody fen and mire vegetation Phragmites/Typha/Schoenoplectus/Carex beds in marshes Alkali vegetation + +/ Molinia meadows, mesotrophic marsh and hay meadows Annual salt pioneer swards, open sand and rock grasslands Steppe grasslands Swamp and mire woodlands Riverine willow-poplar woodlands Closed oak and beech woodlands Acidophilous woodlands Quercus pubescens scrub woodlands Forest-steppe oak woodlands Ravine, slope and rock woodlands Acid coniferous woodlands project, then each mapper should map ca ha per day (15% of that is semi-natural vegetation, 85% arable land, settlements etc.). Mapping of ca ha a day on average can be achieved only if we put ca. eight times less effort on mapping the man-made habitats than on the mapping of natural parts of the landscape. It is similar to the approach of GUTH &KUÈERA (2005), in which the so-called contextual mapping served to complete detailed habitat mapping in areas where no occurrence of natural or semi-natural habitats was assumed. Our previous experiences in the Hungarian vegetation monitoring program (KOVÁCS-LÁNG et al. 2000) and the Natura 2000 program (HORVÁTH et al. 2000, MOLNÁR et al. 2001a,b) made us aware of the considerable, yet often hidden, heterogeneity of the botanists perception. This was the second biggest contraint after time limitation. Since the MÉTA mapping was to be accomplished by the joint work of ca. 200 botanists, methodological guides for the standardization were prepared and field trainings were planned from the very beginning. External data sources Increasing the efficiency of the mapping (which mostly means the efficiency of the survey route) was a fundamental requirement. After reviewing the possible sources of background data (Table 1), field mapping was supported by the following external sources: (1) topographic maps (scale 1:25000, and partially ), (2) false-coloured, enhanced but not processed SPOT4 satellite image-maps from for orientation, planning of mapping route and interpretation (pixel size: m), with the superimposed hexagonal grid, (3) some other additional GIS layers (e.g. roads, brooks, contour lines, borders of settlements); (4) data of the National Forestry Database of Hungary (list of tree species and their abundance per forestry unit). GPS-devices and aerial photos were not used due to their high price, while soil and geological maps at a suitable scale were inaccessible.

6 230 Zs. Molnár et al. Testing the trade-off between spatial resolution and thematic richness We decided to use a grid for the MÉTA mapping to overcome the problems of patch delineation. A hexagonal grid was chosen from the different possible shapes (e.g. circle, triangle, quadrat, hexagon), because: (1) it covers the surface without overlappings and gaps; (2) is most similar to the circular panorama that is observed in the field; and (3) is more easily handled in landscape ecological analysis, because a hexagon has six neighbouring cells, all the same distance from the centre of the hexagon, while e.g. a square-shaped cell has four closer neighbours and four distant ones (WHITE et al. 1992, SCHMIDTLEIN 2000, DE CLERCQ & DE WULF 2004). One hundred hexagons form a quadrat mainly for practical, organizational reasons, but also for collecting certain vegetation data at this spatial scale. The first testing used 25 ha hexagons and the attributes of the Hungarian Natura 2000 program (the list of vegetation types, the area covered by each type in the hexagon, naturalness and the chance of local survival of vegetation patches; HORVÁTH et al. 2000, MOLNÁR et al. 2001a,b). The selection of hexagon size was based on expert judgement, but was successfully tested in Natura 2000 habitat data collection (HORVÁTH et al. 2000). The first MÉTA testing (ILLYÉS, unpubl.) provided the following basic results: (1) in one day of field work, ha could be mapped with these hexagons by combining walking together with the use of a car; (2) vegetation in 80 95% of the study area could be mapped, i.e., 5 20% of the hexagons are left out intentionally or unintentionally from the field observations, e.g. some of those lying in an isolated area, those that are hard to reach (fence, military area), or certain inside parts of vast homogeneous patches (e.g. reed beds, homogeneous pastures), or solitary hexagons missed by chance, but these missing data could be interpreted quite effectively based on the satellite image and/or the previous knowledge of the particular landscape; (3) important characteristics of the landscape are not documented by the Natura 2000 attributes at all (e.g. area of invasive species and old fields, connectedness, regeneration potential); (4) though the documentation of the vegetation attributes on the data sheets takes much time (ca. 30% of the daily work), even more time is needed for the exploration of the area; so recording more vegetation attributes would not lengthen the work excessively. Based on the results of the testing, we included new vegetation attributes (see below) and enlarged the size of the hexagons to 35 ha, thus decreasing the time required for exploring and documenting vegetation of a given area. DESCRIPTION OF THE MÉTA METHOD Data collection Mapping route To standardize mapping and to increase mapping efficiency, the mappers had to follow these rules: (1) there are compulsory and non-compulsory hexagons, the former containing more than 25% of (semi-)natural vegetation (in most cases satellite images and maps help decide whether a hexagon is compulsory or not);

7 MÉTA vegetation mapping 231 (2) during the field mapping each compulsory hexagon has to be crossed and its most common (dominant) habitat type recorded, as well as those types covering at least 25% of the hexagon; moreover, the vegetation patches found on the way should also be recorded (note that this rule led to the omission of many small fragments); (3) in fragmented landscapes (dominated by arable fields, settlements and plantations) all grassland and woodland patches of at least 12 ha that are expected to be more or less natural, should be recorded; (4) vegetation data of non-compulsory hexagons should be documented, if these hexagons are crossed by the mapping route or the data can be derived from the satellite image; (5) at least 80% of the compulsory hexagons and the same proportion of the vegetation data recorded in the compulsory hexagons has to be mapped; ca. 20% of the data can be interpreted based on satellite image and local field experience; (6) if the mappers have earlier local vegetation data (not older than 10 years), these can also be used. Documentation Three different pre-printed data sheets have to be filled in during the field mapping: the data sheet of compulsory hexagons, the data sheet of non-compulsory hexagons and the data sheet of the quadrat. In the case of non-compulsory hexagons only the area of old fields and invasive species, and the potential vegetation have to be recorded. Data are recorded as codes, thus minimizing the amount of text within the database. About 20% of our data is extrapolated and/or interpreted (based on the satellite images). Extrapolation means that a homogeneous vegetation patch mapped in a hexagon continues in the neighbouring hexagons. Interpretation means that the colour and pattern of the patch on the satellite image is similar to that of another one already mapped. In the database it is documented whether a particular data derives from recent field mapping, former vegetation data or interpretation/extrapolation. Mapped attributes Habitat types In a wider sense there are three different approaches used for mapping landscapes and documenting the actual vegetation in Hungary: (1) The phytosociological plant associations (SOÓ & ZÓLYOMI 1951, SOÓ , BORHIDI et al. 1999) proved to be suitable for local mapping of areas with (semi-)natural vegetation, on a fine (1: :10 000) scale, though degraded patches or those of a transitionary state are usually hard to classify (VIRÁGH & FEKETE 1984, BAGI 1991, 1997, SEREGÉLYES &CSOMÓS 1995). The number of categories is ca. 400, many of them weakly defined, with frequent but undocumented overlaps and gaps. (2) For a general mapping of the whole landscape (scale 1: : ) the CORINE Land Cover classification was developed (BÜTTNER et al. 1995, 2000, 2002). The new, 1: scale version of this classification uses 78 well-defined categories, but only 22 of them are related to (semi-)natural vegetation. These categories are similar to the categories of topographic maps.

8 232 Zs. Molnár et al. Table 2. Summary of data collected by the MÉTA method. Hexagon level Quadrat level For each habitat type area, naturalness-based habitat invasive species, quality, pattern, neighbourhood, connectedness, connectedness, threats regeneration potential For the landscape in the hexagon potential natural vegetation, area of - invasive species and old fields, land-use type, landscape health status (3) For an intermediate spatial and thematic resolution, a new habitat classification system was developed in Hungary in 1996, called Á-NÉR (the Hungarian abbreviation stands for General National Habitat Classification System; FEKETE et al. 1997). This system has 112 habitat types, all with detailed and standardized descriptions. For the purposes of the MÉTA the Á-NÉR system seemed to be the most adequate: (1) relatively simple; (2) with a limited number of categories; (3) well-tested in many different landscapes (ca. 200 vegetation maps); (4) vicariant geographic variants are omitted; (5) knowledge of all the plant species is not as vital as for the phytosociological system. For the MÉTA method the Á-NÉR system was partly extended and thoroughly revised (BÖLÖNI et al. 2003). We record these Á-NÉR habitat types as lists for each hexagon. Stands larger than the determined minimal size (differing between the types, but usually some m 2 ) are to be recorded. Mapping includes all habitat types that belong to our natural vegetation heritage. We document the patches of the same habitat type in the hexagon as a compiled record. Patches of transitional state should be classified as two (or more) separate habitats. According to our experiences, the mean number of habitat types in a hexagon is in the whole country, and 4 6 (not more than 8 12) in hexagons where the proportion of natural vegetation is high. Area and spatial pattern Without a thorough exploration of the hexagon we can only roughly estimate the area of the habitat types. Therefore the areal cover of each recorded habitat type has to be given as a proportion of the hexagon using the categories < 1, 1, 10, 50, 100%. Satellite images help make the estimation. Though the exact pattern of the vegetation in a hexagon is not mapped, spatial pattern of each type should be documented as follows: it forms only 1 2, 3 or several distinct patches, or it has a diffuse spatial pattern in the hexagon. Naturalness-based habitat quality To prepare the natural vegetation heritage inventory for Hungary, we had to evaluate the habitat quality of each vegetation type in the hexagon. Our evaluation was standardized, though scientifically not fully sound (see Discussion). The following system of naturalness-based habitat evaluation was used, which has proven to be efficient during the 15 years of its application: (1) totally degraded state; (2) heavily degraded state; (3) moderately degraded state; (4) semi-natural state; (5) natural state (for definitions see: NÉMETH & SEREGÉLYES 1989). This system is more-or-less in accordance with the

9 MÉTA vegetation mapping 233 conventions used in some other European countries (DIERSCHKE 1984, BASTIAN 1996, RUZ IÈKOVÁ et al. 1996, GRABHERR et al. 1998, MÄGI &LUTSAR 2001, BARTHA 2003a). Naturalness-based habitat quality was recorded separately for each habitat type. The naturalness-based habitat quality of different patches of the same habitat type in the same hexagon had to be merged into one (or two) value(s). If for up to 10% of the stands a higher category was relevant, it had to be indicated (e.g. coding 5r4 means: 4 dominates, but with less than 10% of the total area of a habitat type in category 5 is also present). That 10 25% of a particular patch belongs to a lower category of naturalness is not documented. Selection of the proper category of naturalness-based habitat quality is supported by a large set of examples in the Habitat Guide (see below). Threats In many cases, human disturbances threaten the survival of the remnant vegetation patches. From 22 threat types (listed below) the most characteristic ones had to be selected that actually threaten the survival and maintenance of the habitat type in the hexagon in the next years. The strength of the threats is not recorded. The threatening factors (listed on the data sheet) are as follows: improper water management, improper pasturing or mowing, drainage, encroachment of shrubs and trees, burning, afforestation with improper species, woodland patches managed homogeneously, improper selection of trees for timber extraction, logging trees at low age, inappropriate plantation, keeping high densities of game, colonization by invasive plant species, tillage, building and construction, gardening, mining, establishment of a pond, trampling, pollution, rubbish, commercial collection of plants. Neighbourhood Prediction of future changes of vegetation patches can be supported by the evaluation of the direct effect of neighbourhood (< 200 m) on the mapped stands. This evaluation means deciding whether the neighbouring patches will aid or hinder the survival of the particular patch in the next few (10 15) years. The categories are: (1) definitely positive (sustaining neighbourhood), (2) slightly positive, (3) indifferent, (4) slightly negative, (5) definitely negative (destructive neighbourhood). The neighbourhood is negative, e.g., if there is an intensively used arable field (chemicals, infiltration of fertilizer), expanding settlement, or spreading populations of invasive species surrounding the patch. Also an alder woodland around a patch of Molinia meadow may accelerate the colonization of trees thus causing the disappearance of the meadow. Neighbourhood is positive, if it serves as a source of species (a forest rich in species around a well-established bush vegetation), provides proper micro-climate (e.g. woodland around a bog), buffers against degrading factors (forest strip between a grassland and an arable field). Selection of the proper category is supported by a set of examples in the Mapping Guide (see below). Connectedness In the field, we document connectedness at two spatial scales: several hundred meters (hexagon), and several kilometers (quadrat). Connectedness at coarser scales can be derived from the database. By connectedness we mean the potential of dispersal of the species of one vegetation stand from the surrounding areas. We look for the presence and quantity of the

10 234 Zs. Molnár et al. species of the particular habitat type in the surroundings within the distance of several hundreds of metres. We record whether the patches are (1) isolated (typical species of the habitat are not present in the surroundings), (2) connected (species are abundant) or (3) the connectedness is intermediate. Connectedness is also documented at the quadrat level. We examine the area between the stands of the particular habitat type for two different aspects: whether the species pool of the habitat type exists there and whether there is any possibility for dispersal through this area (motorways or tree lines hindering dispersal). Categories indicate that stands are properly connected, moderately connected or isolated. For example, in marshes it is evaluated whether the marsh patches are connected by brooks or channels, or in steppes whether there are steppe species in the forests, mesic grasslands or hedges between the steppe patches. Selection of the proper category is supported by a set of examples in the Mapping Guide (see below). Regeneration potential To gather information on which landscape patches have the possibility to survive for a long time or to regenerate after degradation (BARTHA 2003b), regeneration potential should be documented. We estimate it on the level of quadrats for each habitat type, assessing three different aspects: (1) potential for regeneration of the existing stands in the case of mean degradation; (2) potential for regeneration in the place of a neighbouring habitat type (e.g. woodland encroaching on an abandoned pasture, meadow developing in the place of a drying marsh); (3) potential for regeneration on a nearby abandoned arable field. Categories of regeneration potential are the following: good, mean, poor, or it has no place to regenerate (BÖLÖNI et al. 2003). A large set of examples in the Habitat Guide (see below) also support selection of the proper category of regeneration potential. Potential natural vegetation Mapping of potential natural vegetation (TÜXEN 1956, KOWARIK 1987, HÄRDTLE 1995, CHYTRÝ 1998) does not have a long tradition in Hungary (but see PÓCS et al. 1958, SZMORAD 1997). From the many different approaches to potential vegetation, we use the following definition in the MÉTA project: potential natural vegetation is defined as the most complex (climax-like) vegetation that would develop in a site in no time, so ignoring the constraints of succession, if we exclude the present and future direct influence of humankind, but taking the history of previous changes of site conditions into consideration (KOWARIK 1987, HÄRDTLE 1995). The potential climax woodland vegetation, not the potential replacement vegetation (CHYTRÝ 1998), should be documented except in naturally non-wooded areas. If the mapper is unsure whether the grassland could be potentially afforested, then he/she has to choose some kind of forest as potential vegetation. We are aware that this is quite an arbitrary decision, thus detailed quality assurance is planned before publishing the potential vegetation map of the country. The vegetation categories of Á-NÉR are applied for the documentation. Selection of potential vegetation type must be based on the investigation of the macro-, mesoand microclimate, bedrock, soil type, hydrology and present and past land use. Detailed guidelines for each vegetation category are given in the Mapping Guide. Potential vegetation is recorded in each hexagon, giving not more than five habitat types, listed in the order of their dominance.

11 MÉTA vegetation mapping 235 Invasive alien plant species We have stressed the importance of the invasive species among the threats, because in the coming decades they might cause the most serious damage to the natural and semi-natural vegetation in Hungary (MACK et al. 2000, TÖRÖK et al. 2003). The most important invasive plant species in Hungary (TÖRÖK et al. 2003, and references therein) include Acer negundo, Ailanthus altissima, Amorpha fruticosa, Asclepias syriaca, Aster spp., Celtis occidentalis, Echinocystis lobata, Elaeagnus angustifolia, Fraxinus pennsylvanica, Phytolacca spp., Prunus serotina, Reynoutria spp., Solidago spp., Robinia pseudoacacia, Vitis vinifera and V. rupestris. We estimate the total area occupied by invasive species in each hexagon, based on the categories < 1, 1, 20, 100%. On the quadrat level we record the existence and the potential negative effect of the 15 most important invasive plant species for each habitat type separately. Old fields About 10% of the territory of Hungary is covered by old fields, but no data are available about their exact distribution, actual vegetation, regeneration potential etc. In each hexagon the total area of old fields abandoned after World War II and at least two years ago, is recorded with the categories < 1, 1, 20, 100%. Military maps drawn years ago are especially useful for recognizing old fields that can also be discerned in the field because of their typical vegetation pattern and species composition. Land use As a result of decreasing numbers of domestic animals, the use of grasslands is declining all over the country, which leads to successional changes (BÖLÖNI et al. 2003). There are no exact data on the process, because on the country scale, statistics are given only for the gross number of animals per settlement or per farmer, while there is no information on the way of keeping (e.g. stalling or grazing) or feeding (e.g. hay or other forage). Therefore we also document pasturing or mowing of the grasslands at the hexagon level: mowing, pasturing with cattle, sheep, or with other domestic animals (e.g. goats, geese, horses). More than one category can be marked if appropriate. Data on pasturing or grazing are obtained by observation: either the site has just been pastured or mowed at the time of the mapping or the mapper makes a decision based on the presence of excrements (pasturing) or the lack of dry stalks from the previous year (mowing). If evidence of threats, invasive species, old fields and land use is lacking in the field, the code not observed should be marked. Thus we can avoid the misinterpretation due to mere lack of data. Landscape health status This is a synthetic attribute that describes the (semi-)natural vegetation in its landscape context. It is based on the actual naturalness and regeneration potential of the whole landscape. The hexagons are classified into one of the following categories: (1) region with high value of naturalness and with little or no potential for regeneration ( relict landscapes); (2) region with high value of naturalness that has the potential to regenerate in case of future

12 236 Zs. Molnár et al. Table 3. Preliminary evaluation of the accuracy of identification of habitat types, based on the expert judgement after quality check of 1500 quadrats (53%). High High-medium Medium Poor Nymphaea, Nuphar, Utricularia, Aquatic communities Trapa, Lemna, Salvinia Phragmites beds of fens Stratiotes euhydrophyte habitats of fens and Ceratophyllum communities Artemisia salt steppes Eu- and mesotrophic Helophyte beds Lowland rich fens reed and Typha beds Tussock sedge communities Non-tussock beds of Salt marshes River bank vegetation large sedges Transition mires and raised bogs Alkali mud habitats Rich fens on highlands Lowland steppe thickets Salt meadows Closed rocky grasslands Molinia meadows Open sand steppes Slope steppes on stony Mesotrophic floodplain grounds meadows Open rocky grasslands Closed steppes on loess, Tall-herb communities clay and tufa Lowland alder woodlands Oak-hornbeam Arrhenatherum hay woodlands on highlands meadows Lowland hornbeam woodlands Closed thermophilous Forest-steppe meadows oak woodlands and other semi-dry grasslands Beech woodlands Rock woodland habitats Closed sand steppes White oak scrub woodlands Willow mire woodlands Sand steppe oak woodlands Riverine woodland habitats Poplar-juniper steppe woodlands Acidophilous woodland habitats Thicket vegetation of rocks Turkey oak woodlands Acidophilous coniferous woodlands degradation; (3) region with mean value of naturalness, but with the potential of regeneration; (4) region with mean value of naturalness, lacking the potential of regeneration; (5) degraded, depreciated region (see also BASTIAN 1996). Selection of the proper category of landscape health status is supported by a large set of examples in the Mapping Guide. STANDARDIZATION IN THE MÉTA METHOD Methodical Guides and field trainings Two different kinds of guides have been compiled: Mapping Guide (MOLNÁR 2003) and Habitat Guide (BÖLÖNI et al. 2003). The Mapping Guide accurately defines the terms, provides detailed descriptions of the terms often used inconsistently in Hungary, gives pieces of advice for the planning of the mapping route, determines the rules of the data sheet completion, and lists the most common mapping mistakes. Each attribute has an entry constructed as follows: (1) why do we collect this attribute?; (2) definition of the attribute; (3) detection of the attribute in the field; (4) categories of the attribute; (5) abbreviations used on the data sheet.

13 MÉTA vegetation mapping 237 The Habitat Guide includes the modified version of the Á-NÉR habitat classification system with descriptions of the habitat types with the following structure ( words for each type): definition, site conditions, characteristic species, vegetation context, subtypes (with short descriptions), types not belonging here (the correct category is given), the pattern on the satellite image characteristic for the certain type, characterization of naturalness-based habitat quality and regeneration potential with examples. The chapters of types not belonging here and the examples of naturalness-based habitat quality and regeneration potential were prepared to help standardization. To decrease differences in perception, compulsory field trainings were organized. Conceptual ground for the field trainings was the idea of experience learning: mapping in shuffled pairs, followed by discussions, in four to six different landscapes for three days. Standardization of habitat identification The basic tool for this was the preparation of the Habitat Guide written by many botanists and reviewed by all mappers. Field trainings were used to motivate mappers to use the Habitat Guide in every case when they are uncertain in the identification. Habitat types of rather extreme site conditions, very species-poor sites and habitats very rich in characteristic species proved to be the least problematic ones, as well as some zonal habitat types. Most of the habitat types comprising different transitional states have lower accuracy; these will be analyzed better after aggregating with similar/related habitat types (Table 3). Standardization of naturalness-based habitat quality and regeneration potential After the preparation of the general definitions for the categories of naturalness-based habitat quality (based on NÉMETH & SEREGÉLYES 1989), we compiled field examples for each habitat type, for each category of naturalness-based habitat quality and regeneration potential. The overall number of examples given were 522 for naturalness-based habitat quality and 678 for regeneration potential. Advantages of this approach are as follows: (1) habitat specific differences (species richness, structural and dynamic features) can be handled more easily; (2) the mapper can compare the actually mapped vegetation patch to certain habitat-specific examples instead of abstract general criteria; so (3) examples (instead of definitions) are easier to use and more reliable. Disadvantages of this approach include the following: (1) much harder work to prepare; and (2) more specialists are needed than for writing five abstract definitions. Comparing the experiences of the Natura 2000 mapping project (only categories were used and definitions for the categories were given; MOLNÁR et al. 2001b) and the MÉTA mapping, degree of standardization improved considerably, but further efforts are still needed. When preparing the descriptions for the examples of naturalness-based habitat quality, the following criteria were taken into account: species composition (e.g. number and abundance of characteristic species, dominant species, weeds and disturbance tolerant species), structural characteristics (e.g. patchiness, horizontal structure), origin and age (e.g. primary or secondary vegetation), site conditions (e.g. water-supply, erosion), landscape context (e.g. species richness) and land use (e.g. overgrazing by game, burning). When elaborating the

14 238 Zs. Molnár et al. examples of regeneration potential, the following attributes were considered: the conditions of the stand (e.g. species richness, regeneration ability of the species populations, competitive ability of disturbance tolerant ones), site conditions (e.g. water supply, soil-nutrient balance), conditions of the landscape (e.g. propagule sources, mobility of species related to their distance to the propagule source), type of land use (e.g. forest management, pasture). The standardization was successful, but not perfect (see Table 4 for examples of naturalness-based habitat quality). The proportion of the values given by the mappers and considered as unacceptable errors, decreased from 23.7% to 0% (values in bold), while the proportion of identical values increased from 21.1% to 63.4%. The standardization of naturalness-based habitat quality and the regeneration potential is more difficult than the standardization of habitat identification, because the habitat type first has to be identified, and then the habitat quality and the regeneration potential can be chosen from the examples. It is, however, easier to do, because the examples of similar habitats are also similar, thus in case of incorrect identification of the habitat type the naturalness-based habitat quality and the regeneration potential can be still correct. Selection of mappers In the announcement for the participation in the mapping it was stated that at least few years of field experience is needed. Nearly all Hungarian botanists indicated their interest, though there were many withdrawals during the standardization process (because of the lack of capacity, knowledge or true interest). The selection was not based on any exam, but on voluntary withdrawal in parallel with the standardization process. Before the field trainings, all mappers had to read the Mapping Guide (54 pages) and at least the descriptions of those habitat types from the Habitat Guide that occur in the landscape they were going to map (ca pages). Ten habitat types had to be reviewed thoroughly and the reviews had to be sent to the editors of the Habitat Guide. Later we examined the accuracy of the content of each mapper s documentation of her/his first quadrat, and corrected and sent back the mapper-specific mistakes and misunderstandings. We compiled the questions and deficiencies in the lists and asked the mapper for corrections. We estimated that by this time about 30 35% of the potential future mappers withdrew from the mapping. The selection proved to be successful; 85% of the finished quadrats are acceptable, approximately 10% needed supplementary mapping (for too few attributes or too few hexagons), and only 2% needed to be re-mapped. DISCUSSION Many practical decisions made during the development of the methodology were mostly determined by the available resources. Such decisions included the following: selection of either patch or grid map, spatial resolution, size of the hexagon, type of data sources, number of mapped attributes, and recording basic or synthetic attributes. Supported by the outcomes of similar projects, our own previous experiences and the literature, we thoroughly discussed all of these issues before the final decisions. Here we would like to present some thoughts from this discussion.

15 MÉTA vegetation mapping 239 Table 4. The values of naturalness-based habitat quality given by the pairs of mappers on the first and on the last (usually the 7th) site of the field trainings. Mapper pairs were shuffled 5 6 times during the three days. Each number indicates the values given by one pair (for coding see section: Naturalness-based habitat quality). The proportions of identical, acceptable and non-acceptable values given for naturalness-based habitat quality are significantly different on the first site and on the last site of the field trainings (chi-square = 19.44, P = ). Values given by the pairs of participants of the mapping course Standard value according to the Habitat Guide First site of the field training 4r2, 4r3, 3r2, 4, 2 3r2 5-4, 5, 5, 3, 5r4 5r4 4, 5-3, 4r3, 4, 3, 3 4 4, 3, 4, 2, 3r2, 4 3r2 4r3, 4r3, 4r3, 4r3, 3-2 4r3 5, 5r4, 5r4, 3r2, 4, 4 4r3 5, 5r3, 4r3, 5, Last site of the field training 3, 4, 3, 4, 4, 4, 4 4 4, 4-3, 4, 4r3, 3, 4r3, 4 4r3 5r4, 4, 4, 4, 4, 4, 4 4 3, 3, 4, 3, 3 3 5, 5, 5-4, 5r4 5r4 4, 4, 5r4, 5r4, 4 4 4, 4, 4, 4, 4, 3 4 Scaling and resolution Principally any scale of mapping, any kind of resolution can be good. The important thing is that the aim of the mapping, the mapped attributes and the scale/resolution must match. Generally, a map with finer resolution can be used better and for more things, and it is possible to make a coarser map from the finer, while the opposite way is limited. Though there are relevant scales in each particular type of landscape, in an area of the size of Hungary these relevant scales change between different landscapes (e.g. fine-scale alkali steppe mosaic vs. coarse-scale mosaic of zonal woodland communities). Any scale chosen for vegetation mapping of a country is therefore arbitrary. We have chosen the finest possible scale according to our capacity. We have also taken into consideration that CLC has a fine polygon-pattern and the CLC mapping will be repeated regularly, so MÉTA was designed to complement it with actual detailed vegetation data. Traditional vegetation mapping methods document the landscape in patch maps, which is a product of a thorough exploration of the area (KÜCHLER & ZONNEVELD 1988). Delineation of patches is time-consuming, and it needs much attention, hence the parallel mapping of attributes with different patch patterns is rather difficult, and therefore rarely done (KÜCHLER &ZONNEVELD 1988). To meet our four goals, we had to map many different attributes of the vegetation and landscape at the same time, therefore we were forced to find a compromise between spatial resolution and thematic richness. If there is a given amount of resource for the mapping of an area, the patch map and the grid map will mainly differ from each other in the emphasis on the spatial or thematic resolution. We chose the grid map, because the high thematic resolution was more important for our purposes.

16 240 Zs. Molnár et al. Advantages of grid mapping include the following: (1) the parallel mapping of many different attributes is easier (BAGI 1991, 1997, DIERSCHKE 1991); (2) areas with vegetation transitions and with differently scaled mosaics are easier to document (e.g. grasslands with encroachment of bushes). Disadvantages include the following: (1) the patterns finer than the grid are blurred, while (2) the patterns much coarser than the grid are documented redundantly cell by cell; (3) it is hard to visualize many different attributes at the same time. Combination of data sources Aerial photographs and satellite images can highly improve the spatial resolution and data reliability of vegetation maps (KÜCHLER & ZONNEVELD 1988, MILLINGTON & ALEXANDER 2000). If (semi-)natural regions are dominantly field surveyed, while fragmented regions partly interpreted with remotely sensed material (as is the case in the MÉTA method), it is a good example of the combined data source utilization proposed by MILLINGTON & ALEXANDER (2000). Actually, the MÉTA method uses two main data sources: (1) the actual landscape and (2) the satellite image. It is important to emphasize that in the MÉTA method we combine two approaches: (1) in-door vegetation mapping by satellite image interpretation (e.g. CORINE Land Cover mapping; BÜTTNER et al. 2002) (although we did it out-door as well!); (2) field mapping aided by remotely sensed material (KÜCHLER & ZONNEVELD 1988). We agree with KÜCHLER & ZONNEVELD (1988) and others (e.g., FALIÑSKI 1994a, MILLINGTON & ALEXANDER 2000, STEVENS et al. 2004) that the combination and local calibration of these background data sources allows for the preparation of more accurate maps containing more information. According to our experiences, the accuracy of extrapolations and out-door satellite image interpretations is satisfactory particularly at this spatial scale, if it is based on good knowledge of the landscape. Basic vs. synthetic data Both basic (inductive) and synthetic (deductive) data have advantages and disadvantages: collection of synthetic data takes less time, so the spatial resolution can be much higher (calculated for the same time spent in the field), but the possibilities of subsequent correction and re-interpretation are limited. Recording basic data demands much more effort, but they can be analyzed in a more flexible way, including various finer analyses (SANDERS et al. 2004). Basic data are scientifically more sound and easier to repeat (monitoring). Nature conservation, however, often needs a large quantity of spatially detailed data that cannot be produced reasonably based on basic data collection (e.g. site evaluation). Large differences in the possibilities of standardization in these two data types exist. These types differ so much that CHERRILL & MCCLEAN (1999) consider the collection of basic data the most efficient (or even the one and only) way of full standardization, because this way the mapper working in the field is excluded from the final decisions on the mapping process. Basic data are worthy of being collected if the mapped object can be classified objectively and its quantity or pattern can be given exactly (e.g. ten individuals of Acer tataricum). However, collecting synthetic data is worthy if many field data and/or many different attributes are needed for the categorization (e.g. naturalness-based habitat quality,

17 MÉTA vegetation mapping 241 regeneration potential). If the habitat identification is made in the traditional way, based on species composition, the data collected in the field can be considered as basic data in a well-documented, unambiguous case. However, in a hardly identifiable, transitional or unique situation the collected data can be considered as synthetic. Vegetation maps are based on synthetic data, unless there are phytosociological relevés taken in the patches with adequate regularity, and the categorization of the patches is created a posteriori, after the analysis of the relevés (CHERRILL & MCCLEAN 1999). The data are synthetic if we document historical events or predicted future changes without having documented the vegetation dynamics with basic data. In the MÉTA method we made a compromise between resolution, type (basic or synthetic) and number of attributes and scientific soundness. We decided to use synthetic data for attributes such as habitat type, naturalness-based habitat quality, landscape health status and potential vegetation, and basic data for the attributes such as area of habitat types, area of invasive plant species, area of old fields and threats (Table 5). Scientific vs. practical approach An important task of vegetation science is to make botanical evaluations and formulate suggestions responding to the challenges of the society. Quite often science is not prepared enough to fulfill the demands of the society, e.g. there are no tested adequate methods or capacity. In these cases compromises are needed. In our days the evaluation of habitats, their naturalness and prognosing their possible future is one of the most important demands of nature conservation. Because we do not have scientifically sound, objective methods for measuring habitat quality (or conservation value), the comparison of the present and an ideal or abstract state is usually used to estimate habitat quality for use in nature conservation (for the first time: BERNÁTSKY 1905, later e.g. JALAS 1955, ELLENBERG 1963, SUKOPP 1969, DIERSCHKE 1984, ANDERSON 1991, 1992, GÖTMARK 1992, BASTIAN 1996, GRABHERR et al. 1998, MÄGI &LUTSAR 2001, BARTHA et al. 2003, BARTHA 2003a, GUTH &KUÈERA 2005). In these cases we decide according to our consciously subjective and anthropocentric nature-conservation-based evaluation, what is considered to be in a better or worse natural state: richness in preferred species (e.g. habitat specialists), state of the structural elements (e.g. canopy layers) and of the natural processes (litter accumulation, small-scale successional patchiness). If more positive values can be related to a certain patch and fewer negative ones, we consider it to be of higher quality. While defining the good state, sometimes the spontaneous vegetation dynamics are emphasized more (e.g. in pioneer habitats), sometimes the high species diversity (e.g. in some meadow communities) or the structural diversity (e.g. woodlands) are. Therefore the habitat-quality categories were defined for each vegetation type separately in the MÉTA mapping. The practice-oriented attributes of MÉTA will be used for evaluations and strategic planning. In the determination of preciousness or nature conservation priority of an area beyond the habitat type, the size of the site, naturalness, surrounding landscape and zoological and aesthetic aspects should be taken into consideration. Therefore collective analysis of several different attributes will nearly always be needed. Although any kind of evaluation

18 242 Zs. Molnár et al. Table 5. Different characteristics and planned applications of the MÉTA attributes (B basic; SY synthetic; S scientific; P practical; <, <<,, >>, > comparative suitability for these purposes; +++ high; ++ medium; + low; - none). Basic/ Possibilities Scientific/ Scientific 1st goal: 2nd goal: 3rd goal: 4th goal: synthetic of practical soundness atlas, evaluation evaluation prognosis data standardization vegetation of of of geography vegetation landscape vegetation for and conservation landscape changes Habitat type B<SY +++ S>P Area B +++ S P Naturalness-based habitat quality SY ++ S<<P Pattern B + S P Neighbourhood B<SY ++ S<<P Connectedness (hexagon scale) B<SY ++ S<P Threats B<SY ++ S<<P Potential natural vegetation B<SY ++ S P Landscape health status B<<SY + S<<P Invasion (%, hexagon scale) B ++ S P Old fields B +++ S P Land use B ++ S<<P Invasion (effect, quadrat scale) B<SY ++ S P Connectedness (quadrat scale) B<<SY + S<P Regeneration potential of existing stands B<<SY ++ S<P Regeneration potential on neighbouring areas B<<SY + S<<P Regeneration potential on old fields B<<SY + S<<P

19 MÉTA vegetation mapping 243 involves weighting different attributes, the results will never be objective. Consequently, for nature conservation purposes different evaluations with different weighting systems can be or even should be made. There are some attributes (e.g. connectedness and neighbourhood), which can be used for basic studies in landscape ecology and for designing ecological networks as well. The country-scale distribution of different habitat types is suitable for basic studies in geobotany and for replenishing the Natura 2000 databases. Based on the data on landscape health status, regions can be selected where the nature protection should have priority even on the agricultural areas (zones with different levels of protection). During any analysis scientific soundness and standardization should always be taken into consideration. We indicate the possible future applications of the attributes in Table 5. FUTURE PROSPECTS At the beginning of year 2006, 91% of the area of Hungary was already surveyed. Preparation of the MÉTA maps has already started (see After the detailed quality assurance, the first version of The Atlas of Hungarian Habitat types is going to be published. It will serve as a basis for phytogeography (e.g. the relationships between flora and vegetation, and climatic, geological, hydrological and soil factors) and landscape ecology (e.g. the study of fuzzy classification of landscape types or landscape gradients). We are also going to publish the considerably extended version of the Habitat Guide. We have the 1st opportunity to prepare the detailed map of naturalness-based habitat quality of the country, to evaluate the threat of alien plant invasion, to delineate the potential ecological networks or the sites of Important Plant Areas (IPA). We can also enrich the database of Natura 2000, support the strategic regional planning procedures, and participate in the preparation of training packages for primary, secondary or higher education. Acknowledgements: Many persons have supported the elaboration of the method and the writing of the article by providing advice, opinions, questions, own data and the aid in finding local manuscripts and publications: all MÉTA participants, Gábor Fekete, Milan Chytrý, Karel Prach, Rense Haveman, Miklós Kertész, István Bagi, John Rodwell, Jiøí Guth, Helge Bruelheide, Gabriella Magyar, Dobromil Galvánek, Eli Fremstad, Andrew C. Millington, Petr Petøík, Branka Trèak, Ferenc Csillag, and three anonymous reviewers. The English translation is by Zsolt Erõs-Honti, and was improved by Mark Searle. REFERENCES ANDERSON J.E. (1991): A conceptual framework for evaluating and quantifying naturalness. Conservation Biol. 5: ANDERSON J.E. (1992): Reply to Götmark. Conservation Biol. 6: ANONYMOUS (1990): Handbook for Phase 1 habitat survey. A technique for environmental audit. Joint Nature Conservation Committee, London. ANONYMOUS (1995): CORINE land-cover: methodology and nomenclature. European Environmental Agency, Copenhagen. BAGI I. (1991): Limitations and possibilities of the methodology of the Zürich-Montpellier phytosociology school in vegetation mapping. Phytocoenosis N.S. 3, Suppl. 2: BAGI I. (1997): A vegetációtérképezés elméleti kérdései (Theoretical questions concerning vegetation mapping). PhD Thesis, MTA ÖBKI, Vácrátót.

20 244 Zs. Molnár et al. BARR C.J., BUNCE R.G.H., CLARKE R.T., FULLER R.M., FURSE M.T., GILLESPIE M.K., GROOM G.B., HALLAM C.J., HORNUNG M.& HOWARD D.C. (1993): Countryside survey 1990: Main report. Department of the Environment, London. BARTHA D. (2003a): Die Naturnähe der Wälder Bewertung auf Bestandebene. Allg. Forst-Jagd-Zeitung 175: BARTHA S. (2003b): A természetvédelmi kezeléseket alapozó vegetációkutatásról (Vegetation science for nature conservation management). Nature Conservation Council, Budapest. BARTHA D., BÖLÖNI J., ÓDOR P., STANDOVÁR T., SZMORAD F. & TÍMÁR G. (2003): Mapping on the naturalness of the Hungarian forests. Erdész. Lapok 138: BARTHA D., KIRÁLY G.& MOLNÁR Z. (2002): A botanikus szakma nagy terve: Magyarország természetes növényzeti örökségének felmérése és összehasonlító értékelése (Mapping and comparative analysis of natural botanical heritage of Hungary). In: SALAMON-ALBERT É. (ed.), Magyar botanikai kutatások az ezredfordulón. Tanulmányok Borhidi Attila 70. születésnapja tiszteletére (Recent botanical research in Hungary), PTE Növénytani Tanszék, Pécs, pp BASTIAN O. (1996): Biotope mapping and evaluation as a base of nature conservation and landscape planning. Ekológia (Bratislava) 15: BERNÁTSKY J. (1905): Anordnung der Formationen nach ihrer Beeinflussung seitens der menschlichen Kultur und der Weidetiere. Bot. Jahrb. Syst. 34: 1 8. BIRÓ M., PAPP O., HORVÁTH F., BAGI I., CZÚCZ B.& MOLNÁR ZS. (2006a): Élõhely-változások az idõ folyamán (Habitat changes through time). In: TÖRÖK K.& FODOR L. (eds.): A nemzeti biodiverzitás monitorozás eredményei I. Élõhelyek, mohák és gombák (Results of the Hungarian biodiversity monitoring system I. Habitats, mosses and fungi). KvVM TVH, Budapest, pp BIRÓ M., RÉVÉSZ A., HORVÁTH F. & MOLNÁR Z. (2006b): Point based mapping of the actual vegetation of a large area in Hungary description, usability and limitation of the method. Acta Bot. Hung. 48: BOCK M., XOFIS P., MITCHLEY J., ROSSNER G.& WISSEN M. (2005): Object-oriented methods for habitat mapping at multiple scales Case studies from Northern Germany and Wye Downs, UK. J. Nat. Conservation 13: BOHN U.& NEUHÄUSL R. (eds.) ( ): Map of the natural vegetation of Europe. Federal Agency for Nature Conservation, Bonn. BORHIDI A., KEVEY B.& VARGA Z. (1999): Checklist of the higher syntaxa of Hungary. Ann. Bot. (Rome) 57: BÖLÖNI J., KUN A.&MOLNÁR Z. (2003): Élõhely-ismereti Útmutató (Habitat guide). MTA ÖBKI, Vácrátót. BURBIDGE A.A. (1991): Cost constraints for nature conservation. In: MARGULES C.R. & AUSTIN M.P. (eds.), Nature conservation: Cost effective biological surveys and data analysis, CSIRO, Canberra, pp BÜTTNER G., BIRÓ M., MAUCHA G.&PETRIK O. (2000): Land-cover mapping at scale 1: in Hungary: Lessons learnt from the European CORINE programme. In: BUCHROITNER M.F. (ed.), A decade of trans-european remote sensing cooperation, Proceedings of the 20th EARSel Symposium Dresden, June 2000, pp BÜTTNER G., CSATÓ É.& MAUCHA G. (1995): The CORINE Land Cover Hungary project. In: HARNOS ZS. (ed.), Proceedings of International Conference on Environment and Informatics, FÖMI, Budapest, pp BÜTTNER G., FERANEC J.& JAFFRAIN G. (eds.) (2002): Corine land-cover update 2000: Technical guidelines. European Environment Agency, Copenhagen. CHERRILL A.& MCCLEAN C. (1995): An investigation of uncertainty in field habitat mapping and implication for detecting land-cover change. Landscape Ecol. 10: CHERRILL A.& MCCLEAN C. (1999): Between-observer variation in the application of a standard method of habitat mapping by environmental consultants in the UK. J. Appl. Ecol. 36: CHYTRÝ M. (1998): Potential replacement vegetation: an approach to vegetation mapping of cultural landscapes. Appl. Veg. Sci. 1: CHYTRÝ M., KUÈERA T.&KOÈÍ M. (eds.) (2001): Katalog biotopù Èeské republiky (Habitat catalogue of the Czech Republic). Agentura ochrany pøírody a krajiny ÈR, Praha. CUDLÍN P., PROKOPOVÁ M., FRANCÍRKOVÁ T., BUREŠOVÁ R., SMRZ T.&BOUCNÍKOVÁ E. (2005): Biotope evaluation in the upper Stropnice watershed (South Bohemia) using modified Hessian method. Ekológia (Bratislava) 24 Suppl.: