Grassland with tradition: sampling across several scientific disciplines

The traditional, low-input use of grassland in Central and Eastern Europe has provided high-quality food, clothing and manure for millennia. As an outcome of sustainable low-intensity agriculture, some rural areas have globally significant species richness. Traditional farming is still well preserved in several regions of the Carpathian Mountains. This is a unique opportunity to use the wisdom of our ancestors to keep grassland biodiversity for our descendants. We present a sampling methodology to survey traditionally managed grassland ecosystems holistically, including abiotic, biological and cultural phenomena, and reflect thus the multidimensionality of traditional farming. Our main objective was to reveal the connection between particular management practices and precisely measured plot plant diversity. Our motivation was to identify traditional farming approaches that result in both high biodiversity and sustainable grassland utilization in particular region, and confirm their impact also using statistical tests. The multitaxon vegetation sampling at seven spatial scales combined with soil analyses, detailed land-use information derived from interviews with the land parcel owners, satellite pictures and historical materials provide potentially valuable data for several scientific disciplines including syntaxonomy, plant ecology, environmental anthropology and ethnology. Examples of grassland management practices based on traditional ecological knowledge can serve as an inspiration for developing modern biodiversity conservation strategies applicable for rural regions. The database Grassland with Tradition is registered in Global Index of Vegetation-Plot Databases (GIVD) with the identifier ID EU-00-032. To date it contains data from 31 study sites in 7 countries (Austria, Czech Republic, Slovakia, Hungary, Poland, Romania, Ukraine). Syntaxonomic reference: Mucina et al. (2016).


Introduction
A standard approach to vegetation survey is a single visit of the selected study site, particularly when the aim is classification of a certain vegetation type. Approaches based on re-sampling, permanent study plots or manipulative experiments allow for the study of ecological processes and temporal changes to vegetation. However, some goals of vegetation survey cannot be reached without cooperation with other scientific disciplines. This is particularly the case for studies of habitats that depend on human influence, such as traditionally managed grassland.

International Association for Vegetation Science (IAVS)
All semi-natural grassland habitats are partly or fully dependent on agricultural management (Halada et al. 2011) and their land-use history is usually crucial for conservation of their biodiversity (Diacon-Bolli et al. 2012;Bürgi et al. 2013). Traditional farming conditioned the emergence of all semi-natural grasslands in Europe, as it has contributed to forming their species composition and diversity over several centuries or even millennia (Bonn and Poschlod 1998;Poschlod and WallisDeVries 2002;Poschlod et al. 2009;Hejcman et al. 2013;Poschlod 2015). Grassland history is reflected in present grassland, but it is difficult to detect its traces clearly. Moreover, in traditional landscapes, each piece of land has different ownership structure, different management schemes and different histories Kun et al. 2019), and it is not easy to make generalized management recommendations. The simple typology distinguishing between meadows (mown grasslands) and pastures (grazed grasslands) used nowadays in Western Europe does not cover the whole scale of applied management techniques and does not reflect the multidimensionality of traditional farming. It is often the case that each traditional land parcel represents a category of its own (cf. Janišová et al. 2020).

Traditional farming and local ecological knowledge
Traditional farming includes various types of low-input land utilization on private farms. This type of farming typically occurs within small parcels belonging usually to a single family, or, in some cases, to a commune or farmer associations. Different owners usually apply different farming practices according to the family customs and personal traditional knowledge, so that the landscape itself is very diverse (see also Johansen et al. 2019;Wehn et al. 2019). However, neither the size of land parcels nor the level of labour mechanization is decisive for our delimitation of traditional farming. The most important criterion is rural culture and traditions passed down from generation to generation, which underpin local farming approaches. In this aspect, traditional grassland management differs from modern high-input grassland management, as well as from low-intensity conservation grassland management. The main difference between traditional grassland management and a conservation approach is that local farmers prioritise temporal stability in fodder quantity and quality over biodiversity, which is the main assumption of a long-term ecological sustainability (Janišová et al. 2020). In most European countries, traditional farming is no longer practiced; it has been either substantially modified or replaced by modern farming approaches. However, in some remote mountain areas and in several regions of Eastern Europe, historic land use patterns and farming approaches have survived to the present day ( Figure 1). Similarly, the local-ecological knowledge of the rural inhabitants has been preserved in these areas, which may become a tool for effective conservation of grassland biodiversity Ivașcu et al. 2016;Kun et al. 2019;Janišová et al. 2020).

Anthropologic research status in the Carpathian countries
Studies of traditional farming systems are well represented in the fields of cultural and environmental anthropology and its branches of ethnography (the in-depth study of a particular cultural group) and ethnology (the comparative study of ethnographic data, society and culture). For example, Romania has a long history of ethnographic research of peasant households and beliefs concerning the natural environment. Since the mid-19 th century, folk knowledge regarding plants was systematically studied, firstly by folklorists such as Simeon Mangiuca (1831-1890) and Simion Florea Marian (1847Marian ( -1907. The latter compiled an impressive Encyclopaedia of traditional culture with information concerning plants used by the local Romanian population, in a massive manuscript (12.000 pages and a herbarium, see Marian 2008Marian -2010. Later, with the development of ethnobotany, well-known Romanian botanists conducted studies (Borza 1935(Borza , 1968 and some ethnologists trained in botany (Butură 1979). Despite communist censorship, a considerable number of studies were published starting from the mid-19 th century up until 1990, after which new studies were scarce and were carried out by only few researchers (e.g., Drăgulescu 2013, for Southern Transylvania). However, Romanian ethnobotanical research focuses mostly on the linguistic topics (vernacular names and its origin), empirical use, and spiritual importance of plants. Data on the traditional management of hay meadows and other grasslands, or the local economic importance of certain meadow-associated wild plant species, are absent. While the number of ethnographic and ethnological studies on the material culture of the peasant population in Romania increased in the 20 th century, these studies provide limited information regarding traditional grassland management (hay meadows and pastures). Detailed information about tools and constructions can be found in Butură (1978), whilst in Vuia (1964) we find the first and the most important classification of pastoral practices for Romania.
In the aforementioned studies, the information on grassland use is scarce and can be partially conceived from the other agricultural practices described. For example, piecemeal information on traditional agricultural practices can be found in studies concerning beliefs and/or feasts of the Romanian villages that involve work regulations, such as interdictions and calendar of the labours (see Pamfile 1997Pamfile , with the first edition in 1914Marian 1994, with the first edition in 1898-1899). Important contributions can also be found from the interwar Rural Sociology School studies, that occurred between 1925-1948, under the guidance of Dimitrie Gusti (1880Gusti ( -1955. The main method used by sociologists was the monographic research (for example Conea 1940;Ionica 1944;Bernea 1985). The topic of hay meadow management was first addressed by the Rural Sociology School of Bucharest in their exhaustive monograph from Clopotiva (Conea 1944, which described the local mowing time and some other practices related to hay meadow management) and incidentally in the complex studies from Drăguș (Bărbat 1944).
The situation is fairly similar in the Ukrainian Carpathians. Despite active research by the ethnographers and geographers, studies on traditional farming practices are scarce. An exception is the widely studied Carpathian pastoralism (e.g. Kubijovyč 1936;Podolák 1966;Tyvodar 1994). Recently, some topics, such as various aspects of haymaking traditions and constructions, or glade farming, were studied by Igor Boiko and Janusz Łach (Boiko 2008(Boiko , 2009aBoiko and Łach 2020).
Within the Western Carpathians, most ethnographical studies were published during the 1960s and 1970s by Polish and Slovak authors. Bronisława Kopczyńska-Jaworska focussed mainly on mountain shepherding traditions and undertook studies in Poland, Slovakia and Romania (e.g., Kopczyńska-Jaworska 1959. In the 1960s, Ján Podolák developed a methodological manual for ethnographic research of livestock farming in Slovakia (Podolák 1962), which, besides the detailed description of grassland management practices, included a basic typology of traditional meadows and pastures. His detailed regional studies (e.g. Podolák 1961) as well as his national overviews (Podolák 1965(Podolák , 2008 contributed significantly to recent knowledge on traditional grassland management practices. Meanwhile, traditional rural cultures and local traditions became extinct in substantial areas of the Carpathian Mountains, limiting further ethnological research. Recently, traditional farming systems and local-ecological knowledge concerning meadows and pastures have been of interest for ethnologists (Boiko 2008(Boiko , 2009aFruntelată et al. 2016;Iuga 2016 ;Boiko and Łach 2020), but also for botanical, agricultural, and landscape-focussed researchers. Numerous studies have taken place in the Carpathian regions with well-maintained traditional agriculture Page 2006, 2011;Dahlström et al. 2013;Molnár 2014, 2016;Biró et al. 2014;Molnár et al. 2015;Plieninger et al. 2015;Hartel et al. 2016;Ivașcu et al. 2016Ivașcu et al. , 2018Sõukand and Pieroni 2016;Špulerová et al. 2019;Kun et al. 2019). Some of these studies are the result of joint research: either a collaboration by biologists and ethnologists, or biologists who recognize the importance of integrating social aspects into landscape studies. This type of scientific approach was encouraged by some of the founding fathers of Romanian biology even at the beginning of the 20 th century (like Antipa 1916), or by other biologists even later (Filipașcu 1981) but has been limited in its uptake. The findings of the aforementioned studies show how the collaboration between biology and ethnology is desirable and beneficial for both disciplines. We therefore see a need to develop an interdisciplinary approach to survey the still maintained traditionally managed grassland ecosystems holistically, including abiotic, biological and cultural phenomena, and reflect thus the multidimensionality of traditional farming.

Basic aims of our study approach
1. Survey of vegetation composition and diversity in traditionally managed grasslands. Recording species composition of vascular plants, bryophytes and lichens at seven spatial scales together with environmental factors. Estimation of species richness, beta diversity and species pool in grasslands of each study site. 2. Record the local bio-cultural heritage and traditional ecological knowledge of the particular region. 3. Explanation of the impact of each single management practice used in traditional agriculture on plant species composition and diversity parameters at the plot level. Understanding the interaction of different management measures and their impacts on plant biodiversity and land productivity. 4. Clarification of the impacts of historical land use and landscape parameters on the structure, composition and diversity of current grassland vegetation. Examination of the relationship between phytosociological grassland classification and grassland typology based on factors other than floristic criteria, such as grassland location, its original purpose, historical land use and applied management practices.

Outline of our study approach
We developed a specific methodological approach to study traditionally managed grassland in the Carpathian Mountains which encompasses most of its context-specific characteristics. It includes the study of biological objects (plants, vegetation and their formative environment) as well as sociological objects (people, human communities and rural cultures). Therefore, it combines several scientific disciplines: botany and ecology with anthropology and history.
The described methodology has been developed and first tested in 2017. Since then, it was customized and applied to study bio-cultural heritage in 31 sites during two projects (NGS-288R-18, VEGA 02/0095/19). While the inspiration for botanical sampling of plant diversity using the nested-plot-series comes from the standard EDGG methodology , our approach includes novel aspects such as broadening the scale to a site level, including the study of current grassland management practices by interviewing plot owners, gathering local ecological knowledge in particular region, and including historical information. In the following sections we describe the different elements of our methodology and provide recommendations for their successful implementation.

Site and plot selection
The study sites are to be selected in regions with long-term extensively managed grasslands ( Figure 1). For the purpose of our study, the site is defined as a circle with an area of 25 km 2 with the proportion of managed rural landscape being at least 10% ( Figure 2). In accordance with the study focus, the sites may represent different situations along the elevation gradient, different bedrock types or different political and agricultural histories within the investigated area ( Table 1).
Instead of preferential plot selection within a site, a stratified random plot selection is applied in our approach. As land-use type is frequently determined by the terrain configuration and topography, the stratification is based exactly on these criteria. To maximise the variability in Table 1. Country overview of site and plot numbers, elevation range, geology, typology and phytosociological affiliation of grasslands sampled as of 31 October 2020. vegetation composition on plots, each site is stratified by slope inclination (flat, moderate, steep) and slope exposition (W-N-E vs. E-S-W). In each combination of inclination and exposition, a series of 7 nested plots of increasing size ( Figure 3) is randomly placed in a homogeneous veg-etation patch. Six nested-plot series are thus placed within each site ( Figure 2): two in a flat area, one on a moderate W-N-E slope, one on a steep W-N-E slope, one on a moderate E-S-W slope and one on a steep E-S-W slope. For our specific purpose, flat areas have inclination up to 5°, moderate slopes between 5 and 25°, and steep slopes above 25° (for flatter sites a threshold of 15° is used). As the plot selection is made a priori, two sets of coordinates are prepared for each site: 1) six sampling coordinates and 2) six reserve coordinates. If during the sampling the pre-selected location is found unsuitable for sampling due to various reasons (lack of access, land destruction, freshly ploughed, etc.) the reserve plot coordinates are used within the given stratification category.

Plot-scale plant diversity sampling
During the fieldwork, the location of the predetermined coordinates is identified and the actual vegetation is checked for homogeneity before a nested-plot-series is established. The nested plots cover seven spatial scales (0.0001 m 2 , 0.001 m 2 , 0.01 m 2 , 0.1 m 2 , 1 m 2 , 10 m 2 , 100 m 2 ) and their arrangement is shown in Figure 3. In each of the nested plots, the shoot presence of all species of vascular plants, bryophytes and lichens is recorded. Species relative abundance is estimated visually as a percentage cover in 10-m 2 plots. For this plot size, the detailed environmental data are obtained. Height of the herb layer is measured at five points -in the middle of the edges and in the plot centre. Percentage cover of herb layer, cryptogam layer and bare soil (plot surface not covered Figure 2. Sampling design within a site. Design consists of a circle of 25 km 2 with a minimum of 10% managed grassland area. Topographic map is used for stratified random selection of sampling plots (red pins with numbers): two in a flat area, one on a moderate W-N-E slope, one on a steep W-N-E slope, one on a moderate E-S-W slope and one on a steep E-S-W slope. Sampling plots are selected exclusively in the CORINE grassland habitats (blue-coloured patches) and a satellite map is used to check the active land-use and avoid abandoned land parcels. Figure 3. Arrangement of nested plots covering seven spatial scales. The sampling plots are established from the diagonal; after setting the location of the outer NE and SW corners, the inner NE and SW corners are set at 483 and 930 cm. Then the measuring tape is used to fix the 100-m 2 plot and a special rope with nodes is used to fix the 10-m 2 and the two 1-m 2 plots. The 1-m 2 sampling plots are placed in the NW and SE corners of the 10-m 2 plot and their outer corners are used for permanent marking by metal nuts of 20 mm (NW corner) and 24 mm (SE corner) buried to a depth of 5 cm.
by vegetation) are estimated and other details on the particular vegetation and surrounding location are recorded in a purpose-designed form ( Figure 4). Sampling equipment also includes a camera for documentation of the actual vegetation before the sampling start, GPS for precise measurement of SE and NW coordinates, and some other tools shown in Figure 5.

Data on topography, climate and soil
During the fieldwork, topography of each 10-m 2 plot is characterised by elevation (m), inclination (°), and microrelief (small-scale variability of microtopography expressed as a deviation from a smooth plane in cm). Further topographic and climatic data can be calculated a posteriori (e.g. calculation of solar radiation from the latitude, slope and aspect data is possible using the approach of McCune (2007)) or obtained from specialized databases (e.g. mean annual temperature, annual precipitation sum, precipitation of the driest quarter, and other climatic variables can be obtained from the WorldClim (Hijmans et al. 2005; http://www.worldclim.org/) or Chelsa (https:// chelsa-climate.org) databases. Soil depth is measured on each plot at five random points with a steel rod of 1 cm diameter. Cover of stones and rocks (particle size diameter >63 mm) and cover of gravel (particle size diameter 2-63 mm) is estimated on the plot surface (%). A mixed soil sample of the uppermost 10 cm of the mineral soil is taken from five random locations. Air-dried soil samples are analysed in the lab for the following soil parameters: pH (measured in KCl), phosphorus (P), potassium (K) and magnesium (Mg) content in mg/kg, and calcium (Ca), organic carbon (C), and total nitrogen (N) content in g/kg. Humus soil content is calculated from the organic carbon content. The C/N ratio is calculated as a surrogate of soil accessible nutrients.

Data on management practices
Basic information on the management practices is obtained during the fieldwork using semi-structured outdoor or indoor interviews with the local farmers and landowners (Molnár et al. 2008). Our questions (see the section on question guidelines for the details) focus on agricultural practices used recently (since 2010) and during the two historical periods, 1950-1990 (the period of centrally planned economy in all investigated countries except Austria) and 1990-2010 (the period of market economy in all investigated countries). Based on the information from landowners and/or their neighbours we were able to derive several management variables with potential short-or long-term effects on grassland ecosystems, including the long-term effect of mowing, grazing, burning or ploughing, management stability, or the cumulative impact of recent traditional management practices which reflects both their heterogeneity and frequency. The cover of litter (cover of dead biomass on the plot surface) is estimated during the field sampling and can be used as a surrogate of grassland abandonment. Further important management variables include type of grazing animal(s) (e.g. horse, cow, sheep, goat, pig, goose), type of grazing system (e.g. spring grazing, autumn grazing, combination of spring and autumn grazing, whole-season grazing in enclosures, common pasture with whole-season free grazing, whole-season herding with a shepherd, folding/corralling), year of the last ploughing and the type of crop(s) grown at that time.

Data on landscape structure and heterogeneity
In addition to the management variables, we derived several variables that characterise the surrounding landscape and reflect the local historical grassland development.
Along with the size of the actual land parcel around the sampling plot, we calculated the mean size of grassland parcels (in km 2 ) within 1-km 2 plot surroundings (a circle with a radius of 0.564 km centred at the plot) and the distance (in m) of the sampling plot to the nearest forest or hedge, using the satellite images (Google Earth). Proportion of non-forest habitats and habitat diversity in 4-km 2 plot surroundings (a circle with a radius of 1.128 km centred at the plot) were derived from the CORINE land cover maps with spatial resolution of 10-50 m (Bossard

Historical land-use information
Historical data on particular grassland parcels can be obtained from historical maps and, more recently, orthophotos ( Figure 6). In our study, we used the georeferenced historical maps from the three military surveys of the Habsburg empire: Josephinische, Franziszeische and Franzisco-Josephinische Landesaufnahmen (https://mapire.eu/en/; Timár et al. 2010). These maps are available for the entire territory of the Carpathians, although the particular region might have changed its state affiliation several times during the last 250 years. For most of the map sheets, cropland and grassland area can be distinguished, the latter usually differentiated between meadows and pastures. Based on these data, we estimated the approximate grassland age of the studied grassland parcels. We considered grassland age to be the minimum age of the grassland patch at the plot location in years. Values usually ranged between 100 and 250 years. The aerial and satellite photographs provide valuable plot-based management data, as they are usually available for several years or seasons. Radical interventions in the parcel vegetation by plowing and mowing can be usually dated and these data supplement information from landowners. Vegetation succession following abandonment or cessation of management activities can also be detected from aerial and satellite images.
Regional historical information is available from chronicle and archive materials. Chronical data mainly provides the demographic data and livestock numbers, while cadastral, konkretual (drawn up for the purpose of collecting taxes), or consolidation maps can be found in archives. The explanatory notes to these documents often contain valuable information on the historical management of individual land parcels. These data help to illustrate the context of the study region.
Along with the abovementioned sources, recent and historical literature is available on traditional agricultur-al practices, the latter mainly available in local languages only. Many pre-industrial agricultural textbooks are increasingly accessible online, as they fall out of copyright and become digitized by libraries (Burton and Riley 2018). Information on the inclusion of particular land parcels in subsidy schemes is usually available on the agricultural web portals of the respective country.

Ethnological field survey on local management practices and ecological knowledge
An anthropologist or ethnologist investigates contemporary social and cultural characteristics of a community, but also often looks into the community's history in order to understand the dynamics of the local culture. Qualitative methods are commonly used to carry out a social survey, because they are comprehensive (see Rotariu and Iluț 1997). These methods are i) observation; ii) interactive methods (such as interview, oral history, and narratives of life); and iii) ethnographic description (see Laplantine 2010, Geertz 1973. All these methods provide scientific knowledge created by means of the social interactions that happen between the researcher and the natives of a community. The first strategic method to gather information during a field research is the observation. The goal is to understand the cultural phenomena by observing and recording as much information as possible about the social context in study locations. Direct observation implies recording (using a notebook, dictaphone, camera or video recorder) what the researcher is seeing, focusing on each detail. This means paying attention to everything that the persons who are being observed are doing. An easy example of direct observation we used during our research is the filming of the creation of a haystack, which will allow the researcher to access the information at any time afterwards. The second type of observation is participant observation (see also Spradley 1980 andBernard 2006), a contextual observation where the researcher becomes immersed in the culture (s)he is studying. Participant observation allows researchers to better understand the techniques and the actions of the community.
In order to gather local traditional ecological knowledge, interviews are frequently used, which is an interactive method that takes the form of a dialogue between the researcher and the members of the communities (see Spradley 1979, Bernard 2006. When conducting an interview, the researcher should be aware of several points. First of all, it is important for the researcher to focus on the personal experience of the interlocutor, taking into account that traditional values are usually abstract or self-implied for a member of a community, but are present in the actions that the interlocutor is describing. His/ her biography becomes the "embodiment" of tradition (Bot 2015). It is therefore recommended to avoid "general" questions, such as "What actions are done in spring to clean the meadows?" but ask, instead a more personal question, such as: "What actions do you and your family take for cleaning the meadows in spring?" In this manner, the respondent is giving an answer that (s)he knows is correct, otherwise (s)he may search in his memory for an answer that (s)he believes would please the researcher. By doing this, the interlocutor is personally involved in the interview and may find it easier to express their thoughts and experiences. Similarly, when a temporal dimension is in question, it is advisable to appeal to the respondent's own experience and memories. For example, to ascertain continuity of practices through a community, the following questions could be posed: "From whom did you learn …?" or "Have you seen somebody else in your family that has done this/that…? Whom?" Responses to these questions could reveal the way practices are handed down or introduced from one generation to the next.
There are several types of interviews that a researcher can choose from (see Bernard 2006: 210-250). First, there is the "informal interview", which does not have a clear structure and is mainly used to establish a first contact and a preliminary discussion with the interlocutors. Usually it is not recorded, but the researcher in his field notebook writes down the interesting ideas. It can lead to an appointment for a second interview. The second type of interview is the "unstructured interview", which is done according to a broad and very flexible plan. This type of interview consists of giving a topic to a person and letting her/him talk. This type of interview is very well adapted to the narratives of life. A third type of interview, the "semi-structured interview", is the most used in ethnological surveys. In this type, the structure and the guidelines are more understandable, there are clearer topics approached, sometimes in a special order, the questions are conceived in such a manner to cover all possible aspects of the theme discussed. Some of the questions are even repeated moments later, if the topic allows it, using different words, in order to verify the coherence of the dialogue, or to stir up the memory of the interlocutor. The fourth type of interview is the "structured interview", which is frequently used in sociological surveys, applied under the form of questionnaires with questions formulated a priori, where the subject responds to questions with formulated answers.
After choosing the right type of interview that would fit the research, the next challenge is to find interviewees. It is recommended that the researcher is transparent about the reasons for the research and why (s)he would like to talk with people. In order to find out the best interlocutors when the community is unknown, it is advisable to make preliminary interviews with the local key stakeholders, such as the mayor, teachers, priests, or even veterinarians as we found out in the community of Șișești (Maramureș, Romania). These people could give valuable information about their own experiences concerning the topics of the interview, but they also can point to the people whom they think are suitable to participate in the research. This is similar to the snowball sampling technique, when research participants are asked to assist in identifying other potential local experts. This method is often applied in studies of traditional ecological knowledge, since it is known that local experts within a community are acknowledged by the community (Berkes 2018). Where the researcher does not speak the local language, it is recommended to have a translator familiar with local dialect. This step is particularly important given that traditional ecological knowledge is embedded in the local speech, certain local words have different meanings in different areas and can sometimes describe different practices or ecological concepts (e.g. vegetation succession, habitats, etc.).
When conducting an interview, the researcher should ideally talk in a simple language, without using obscure or complex scientific terms that the interviewee may not understand. After explaining the reason for the interview, the researcher should first gain verbal permission for the interview to be recorded. Written permission can often be granted following the interview, once trust has been built. In some cases it is necessary to encourage the interlocutor in speaking, e.g by demonstration of importance of the information obtained, or using the probing techniques (see Bernard 2006: 217-223) that stimulate the interview, such as: the silent probe (which implies we should not rush for the interlocutor to answer); the echo probe (repeating the last thing said by the interviewee and then asking her/ him to continue; the "uh-huh" probe; the "tell-me-more" probe; and so on. In parallel to recording the dialogue, ethnologists commonly use a notebook during interviews to write down fieldnotes (see Sanjek 1990), which may include interviewees' ideas, behaviour, emotions or other details. These notes can be a backup in case of failure to record the interview, but are also helpful as memory aids for recalling the interview content when analysing the data.
The ethnological field survey is followed by transcription and translation of the recorded interviews, and interpretation of the obtained information. It should be noted that this phase of ethnological research usually require more time than the field survey.

Question guidelines: thematic areas for investigation of traditional farming
Traditional farming can be studied at several hierarchical levels, such as the land parcel, the farm, the commune or the regional/national levels. Detailed knowledge about particular land parcels is often insufficient to understand the complexity of local traditions and requires additional historical information to set it in context. In order to guide researchers in this interdisciplinary approach, we have prepared a set of questions for interviewees with different foci.
1. Topics to be addressed in the questions asked to the parcel owner related to the sampled nested plot series. ; has the parcel been cleaned from litter, tree leaves, branches, stones, and rocks; has the parcel been burnt for cleaning; j) Abandonment: is the parcel managed continuously; are there periods with management interruption; when and for how many years; how abandonment affects grassland quality and productivity; what are the advantages and disadvantages of grassland abandonment; k) Restoration: what activities improve grassland quality and productivity, what kind of restoration has taken place in the parcel; how can an abandoned grassland/ overgrazed pasture/ degraded meadow be restored; how to restore a grassland digged on the surface by wild boars; l) Other grassland management techniques: irrigation, amelioration, anthill or molehill removal, others -details on their application. ; what is the level of self-sufficiency in the farm; details on marketing; rentability of small farming; ideas for improvement; d) Haymaking and grazing system: details on spatio-temporal organization of mowing, grazing and manuring during the year; how is it related to the farm area, number of animals and distances between the parcels.
2. Questions to the parcel owners, local inhabitants and town hall officers related to farming and animal husbandry in the village/commune/region: a) Animal husbandry in the village: how many animals do people usually have in the household; what animals; which breeds do they prefer and why; what changes happened during the last decades (increasing/decreasing number of animals); b) Job opportunities: how many people work abroad, in the cities, how many are making a living from agriculture, how many receive subsidies for working the land or keeping animals; c)  parison to 1900, 1950, 2000; is the proportion of inhabitants living from agriculture the same/lower/higher in comparison to 1900, 1950, 2000; e) Animal numbers: is the current number of livestock in the village the same/lower/higher in comparison to 1900, 1950, 1980, 1990, 2000, 2010, 2020 (

Pros and Cons of our study approach
Pro 1: Teamwork enriches each of the participants Participation of researchers from several disciplines allows for new insights and addresses the multifunctionality of traditional farming landscapes as a study object.

Pro 2: Clear, detailed and unified methodology used for a huge area
The value of the data increases with the size of the area on which the method is applied. Methods with a similar degree of detail have so far been applied only locally, often without the possibility of comparison with other areas.

Pro 3: Simple but robust sampling design
Thanks to the stratified random plot selection, the data are suitable for estimation of multiple diversity parameters. Species richness can be estimated at seven increasing spatial scales, while beta diversity and species pool can be estimated at the site level.

Pro 4: Question guidelines are provided for ethnobotanical studies focussing on traditional farming
Our question guide was prepared by botanists and anthropologists in collaboration and both natural and social sciences qualifications were valued equally in its production. It is widely applicable in both disciplines.

Pro 5: Historical and landscape information opens new horizons
Including historical and landscape information allows evaluation of vegetation change from new perspectives. History, landscape context, and management are increasingly seen by environmental scientists as key for understanding grassland biodiversity. Gathering both vegetation and management data for all the plots allows for proofing the importance of particular management methods for biodiversity values.

Pro 8: Capturing the grassland dynamics
In contrast to phytosociological sampling based on a single visit of the sampled area, our approach tries to capture the vegetation development by asking about the past. This methodology proves the dynamic management of grasslands, since it records precise agricultural practices that were carried out by individual farmers on the grasslands they own. Moreover, it shows how the continuity or discontinuity in traditional grassland management may correlate with current biodiversity patterns.

Con 1: It is labourious and time consuming
The interview protocol was developed during the three years of our project. It was clear from the beginning that it is not possible for a single team to perform both the vegetation sampling and the interviews in numerous study sites, as the period between flowering of the grasses and the first cut is too short and also because the interviews need special skills to obtain high quality information. We started with a collection of questions to determine the workflow associated with the management of local grasslands by consulting agricultural workbooks (Klapp 1965;Opitz von Boberfeld 1994). We began with more than 100 questions (including over 70 questions to the landowners) and reduced them during the following years to those questions that connected with the plot biodiversity. Experience has shown that with good logistics, it is possible to carry out the vegetation survey part of our methodology on one study site in three days, concentrating on two land parcels (and nested plot series) per day. However, more than one visit to communities was necessary in order to carry out the anthropological part of our methodology (e.g. in certain situations where the local people do not have time to respond due to their daily activities, the researcher has to come back another time and try again). We found that the research team should ideally consist of two anthropologists to support each other in data collection particularly when the study site is new to the researchers. The best strategy is to make two teams, the anthropologists and the biologists, logistically independent, so that they can work together, but also in different places during a single day: either to go to the sampling parcel and to ask about the specific management practices for that very place, or to return to the village if there are no people nearby the parcel, or to stay in the settlement, visit the town hall and to take interviews that would provide more context concerning the local practices in general. Spending more time with the local communities would help the researchers to understand better the local practices and find suitably knowledgable interviewees.

Con 2: It is concise and thus not always covering all important aspects
By focusing only on specific parcels in the landscape where the vegetation surveys are carried out, researchers may miss the wider picture of the agricultural practices and landscape management of the community. It is therefore recommended to integrate individual site-specific management with the wider agricultural and economic activities at the landscape level. Some of the activities at this level are organized by the community according to their own local rules. Consequently, it is usually necessary to obtain additional oral or historical information on the common use of resources in the recent past.

"Grassland with Tradition" database
The database "Grassland with Tradition" is registered in Global Index of Vegetation-Plot Databases (GIVD, https://www.givd.info/faces/database_details.xhtml) with the identifier ID EU-00-032. As to 31 October 2020 it contains 186 nested plot series recorded in the Carpathian Mountains within seven different countries. The sampling was conducted during 2017-2020. The sampled plots are distributed at elevation between 147 and 1422 m (Figure 7) and over various geological bedrock types including plutonic and volcanic rocks, limestone, flysch, loess, and sediments (Table 1). The main land-use categories (meadows, pastures and abandoned land) are further divided according to the combination of applied management practices (Figure 8). The meadows include five basic categories regarding the additional application of ploughing and grazing, and its timing. Permanent meadows were never ploughed and are used exclusively by regular mowing. Meadows on former cropland include grassland in active crop-grassland rotation systems or older fallows which are now used exclusively as meadows. Three types of grazed meadows can be distinguished, either with exclusive spring or autumn grazing and those grazed during both these periods. In the regions with best-preserved farming traditions (Figure 9) grazed medows represent the absolute majority of local meadow types. Most of the sampled pastures are used for grazing cows or sheep as single grazing animal (Figure 8). Regarding the phytosociological affiliation (Table 1, Figure 8), mesic grasslands of the Arrhenatherion elatioris alliance (Molinio-Arrhenatheretea class) and Violion caninae (Nardetea strictae class) prevail, followed by semidry grasslands of the Bromion erecti and Cirsio-Brachypodion alliances (Festuco-Brometea class). According to the actual land-use consulted with the landowners we distinguished three categories of grassland management: i) management according to continuous living tradition, ii) management according to modified tradition, and iii) management not based on tradition. Although the classification criteria of traditional land-use are sometimes fuzzy and subjective, there is a clear decreasing trend of traditional farming along the Carpathian arch from the southeast to the northwest (Figure 9).   . "Grassland with Tradition" database: Distribution of living, modified, and abandoned tradition in grassland management of the sites studied so far (as to 31 October 2020). In each study site, six parcels were selected by a stratified random selection considering the topographic situation (aspect and slope) and habitat type (managed grassland). According to the actual land-use type on those parcels (consulted with the landowners) we distinguished three categories of grassland management: i) management according to continuous living tradition, ii) management according to modified tradition, and iii) management not based on tradition. Although the classification criteria of traditional land-use are sometimes fuzzy and subjective, clear differences in farming approaches are obvious between the Western and Eastern Carpathians.