Author's personál copy _ General Ecotogy 1 Phyto socio fogy 2767 Author's personal copy 2768 General Ecology I Phytosociology_ Phytosociology J Dangler. University of Hamburg, Hamburg. Germany M Chytrý, Masaryk University, Brno. C^ech Republic J Ewald. University of Applied Sciences Wethen step han. Freising. Germany © 2008 Elsevier B.V. All rights reserved. Introduction Phytosocioiogical Data Classification of Vegetation Applied Phytosociology Further Reading Introduction Phytosociology is a subset of vegetation science, in. which it stands out by focusing on extant (vs. fossil), taxonomic (vs. physiognomic or functional) plant assemblages at the scale of vegetation stands (vs. landscapes or biom.es). Its principal gnal is the definition and functional characterization of vegetation types based on the total florist ic composition of stands. Phytosociology distinguishes between concrete vegetation stand (phyiocoenosis), which can be represented by a plot record (relevc), and abstract vegetation type (syntaxnn), representing a group of all stands sharing certain attributes. The classification framework (syn taxonomy) is designed in close analog}' to plant taxonomy, with association as the basic unit. The fundamental concepts of phytosociology were developed by Josias Braun-Blanquer in the 1920s. He combined a standardized protocol for plot sampling, sorting of species-by-plot matrices, demarcation of community types, and their hierarchical ordering into a practical and efficient framework for the study of vegetation. In this article, wc use the term phytosociology for the Braun-Blanquct approach and its modem extensions. Phytosociology is the mainstream vegetation classification scheme in Europe, as well as in several countries outside Europe, and has become increasingly popular worldwide from the 1900s onward. Within modern ecology, phytosociology represents the most comprehensive and consistent methodology for vegetation classification. Releves are the most widely used standardized protocol lor sampling plant species co-occurrences at the stand scale, Being derived from the vast body of relcve data, syntaxonomy provides a comprehensive yet open system nf vegetation types, which arc indispensable in land-use management and natuTc conservation. Consisting of abundance data on individual plant species, rcleves and vegetation types organi7x'd in large phytosocioiogical databases are an enormous source of fine-scale biodiversity informarion. If linked to the growing body of plant trait or indicator value data or environmental information in geographical information systems (GTSs), phytosocioiogical data open new avenues for exploring large-scale ecological patterns and processes, and provide spatially explicit information necessary for environmental management. Phytosocioiogical Data Data Records In phytosociology, the data of a single plot are called & relcve (French for record, sec Tabic 1), which consists of 'header' and species data. The 'header' comprises plot identification, methodological information, and metric, ordinal, or categorical data on geographic position, environmental conditions, and overall vegetation structure. Some of these data are essential, others optional, depending on the purpose and resources of a proiect (Table 2). The species data are composed of a list of plant taxa (species and infraspccific taxa; further referred to as 'species') and their attributes, A full releve lists all plant species occurring in the plot and growing on soil, including bryophvtes, lichens, and macroalgae, Additional recording of species growing on substrata other than soil, such as on living plants (epiphytes), rocks (saxicolous plants), or dead wood (lignicolous plants), is desirable, but not standard in phvtosociology. Ever)' species observation is assigned to a vertical stratum (e.g., tree layer, shrub layer, herb layer, and cryptogam layer). Woody species occurring in different layers are recorded separately for each layer. For each species observation in a layer, an importance value is estimated and usually expressed on a simplified scale of abundance (number of individuals/ ramets) and/or cover (area of the vertical projection of all aerial parts of a species relative to the total plot area) (Tahle 3). As mixed cover-abundance scales pose problems in data analysis, pure cover scales arc preferred when precise quantitative estimates arc required, for Tahle 1 Example of a forest relevewilh five vegetation layers distinguished: upper tree layer (T1), lower treelayer(12). shrub layer (S). herb layer (H), and cryptogam layer (C) Plot ID/methodology Field number Author Plot size (m2) Plot shape Sampling date Preliminary syntax on Geographic data UTM coordinates Lccalrty Ertv/ronmenra/ date Elevation (m a.s.l.) Slope aspect (') Slope inclination {') Soil type Parent material Management Stand age {year) Structural data Height upper tree layer (m) Height lower IreB layer (m) Height shrub layer (m) Cover upper tree layer (%) Cover lowBr tree layer (%) Cover shrub layer {%) Coverherb layer {%) Cover cryptogam layer {%) Layer T1 Spec/es Fagus syivatica Picea abies T2 S 11 Picea abies Acerpseudopfa tanus Aconitum vulpana Adenostyles allianae Adoxa moschatellina Athyrium filix-femina Cardamine flexuosa Cnaemphyltum hirsutism Chrysasplenium aftemifolium Gcarbita alpina Deschampsia cespřfosa Dryopierís dilaiaia Dryopteris fílix-mas Epifobium montanum Caleopsts tetrzihit Galium odoratum Geranium robertianum Gymnocarpium dryopteris Impatiens noli-tangere Lamiastrum montanum Luzuia syivatica sebsp. sieben Lysimachia nemorum MercuriaOs perermis Mycetts muraiis Myosotis syivatica Importance 3 3 1 1 291 J Ewafd 144 square 3 June 1997 Galio-Fagetum adenostyletosum 32 U 4434393 E - 5272800 N Ettaäer Manndl, Höllenstein, 3km W from Eschanlohe, Garmisch-Partenkirchen, Bavaria. Germany 1300 35 32 Cambisol Cretaceous sandstone Protective Forest 140 30 6 3 75 3 1 20 3 Layer H Species Oxalis acetosella Pans quadrifolia Potypodium vulgare Prenanthes purpurea Primula eiatior Ranunculus lanuginosa Rum ex alpestris Salvia gluttnosa Senicula europaea Saxifraga rotundifolia Senecio fuchssi SteUaria nemorum Theryptens limbosperma Veronica urticifolia Viola biflora Atrichum unduiatum Brachythecium rutabulum Conocepbalum conicum Ctenidium moduscum Dicranella neteromalla Dicranum scoparium Hssidens taxifolius Mmum spinosum Plagiochila porelloides Pfagiomnlum unduiatum PiagioihecSum curvifolium Pofyinchum formosum Fht'7omnium punctalum Tnuidium tamariscinum Importance 2 Author's personal copy _general Ecology I Phytosoclology 2769 Author's personal copy 2770 Gqnoral Ecology I Phytosociology_ Table 2 Essential 0 and selected optional data to ba included in the "header' of a phylosociological relevfi Group Data Comment ID/methodology Field number" Authors)' Plot size* Plot shape Sampling date* Preliminary assignment to a syntaxon Geographic data Geographic coordinates" For Bxampie, Greenwich coordinates, UTM Locality in textual farm' including political and/or natural geographic- units Environmental data Elevation (m a.s.lj* Slope aspect' Inclination' Soil For example, type, texture, depth, pH, humus form, humus content, C/N ratio Geology (parent matenal) Management Structural data Height of vegetation layers (m) For example, tree layer, shrub layor, herb layer, cryptogam layer Cover of vegetation layers (%)* Cover of each layer and total cover Cover of other surfaces (%) For example, bare soil, litter, woody debris, rocks, open water Table 3 Customary version ol an extended Braun-Blanquet cover-abundance scalewilh ordinal values, which are often used for numerical interpretation. In the original Sfaun-Blanquet scale, 2m, 2a, and 2b ware joined underlhe symbof '2' Cover Abundance {number of interval Ordinal Symbol individuals/ramets) t*l value r 1 0-5 1 + 2-5 0-5 2 1 6-50 0-6 3 2nl Mora than 50 0-5 A 2a Any 5-12,5 5 2b Any 12.5-25 6 3 Any 25-50 7 4 Any 50-75 a 5 Any 75-100 9 example, in studies of vegetation change in permanent plots. Sometimes, additional character-sties of the species - such as sociability (degree of clustering of the individuals), vitality, fertility, age class (e.g., seedling or juvonilcj, and phonological status - are recorded, but these arc of little or no importance for standard analyses. Selection and Size of Plots Plot sites in the field art,- positioned in vegetation stands that are relatively homogeneous in tonus of structure, species composition, and environment, so that vaiiation is minimized within and maximized between plots. The traditional sampling strategy m phytosociology, preferential sampling, in which the researcher selects stands that are considered as representative of some vegetation units, has several disadvantages: it is not repea-table by other researchers, tends to neglect some vegetation types and oversample others, and produces a n on representative sample of vegetation diversity in the study area. In spite of these disadvantages, probabilistic sampling strategies, such as random or systematic sampling, have never received wider acceptance in phytosociology. While providing reliable estimates of vegetation attributes, probabilistic sampling is less suited to phytosociology's goal of representing maximum variation in vegetation diversity across a study area, as it tends to uudcrsample or even miss rare types. GIi> and global positioning system (GPS) technology have made strati-fied-random sampling schemes increasingly popular in phytosociology; Based on the overlay of digital maps in a GlS, the study area can be stratified into patches with certain combinations of land-cover types and environmental variables that are supposed to correlate with plant distribution. Within each of these strata, plot positions are randomly placed and subsequently found in the field with a GPS receiver. A related sampling strategy is a gradient-oriented transect or gradscct, which establishes plot sites along a landscape transect that runs parallel to an important environmental gradient. Phytosociological plots are usually squares or rectangles, which, as a rule of thumb, are roughly as large in square meters as the vegetation i.s high in decimeters (e.g., 200 m" for a forest of 20 m height). Despite this rule and other suggestions in textbooks, a cm a I plot sizes used may span more than one order of magnitude within the same vegetation type. Standardization of plot sizes is hindered by the vague and misleading concept of'minimal area', which is thought to be a certain plot size specific for each vegetation type, beyond which any further enlargement has negligible effects on species richness and composition. However, plot size strongly influences estimates of species richness and other vegetation parameters. Joint use of differently sized releves in a single analysis may thus produce artifacts in classification, ordination, and calculation of fidelity of species to vegetation units. Tu safeguard data compatibility, standard plot sizes have been proposed for use within certain srrucrural formations, for example, 200 m* in forest vegetation; 50 m' in scrub vegetation; 16 m' in grassland, heaihland, and other herbaceous vegetation; and 4 m" in aquatic and low-growing herbaceous vegetation. Vegetation Databanks Phytosociology has a long tradition of publishing, archiving, and re-analyzing releves as its basic primary data. Many phytosociological journals print full tables including all relevant releves, thus making data accessible for future compilation and analysis, which was traditionally performed as synoptic tables on paper. The limitations of manual data management were overcome by using rable editing and databank software, which allows seizing, storing, managing, filtering, and analyzing releve data in niultiple ways. Compilation in a databank requires that all information obeys stringent formal and technical rules laid down in reference lists, meta-data and data models, Databanks of different formats and complexity were established, ranging from simple spreadsheets to relational and object-based data models that allow flexible definitions and comprehensive documentation of meta-data. Simple databanks are able to exchange data freely if the same standards, database formats, definitions, and reference lists are used. The success of phytosociological databanks is so far due to rather simple management software packages such as TUTCBOVEGt which is currently the most widespread program in Europe and beyond, distributed free of charge or at small cost along with taxonomic reference lists and tools to create, edit, and analyze phytosociological tables. While early databank development revolved around fixing standards for data types and references for plant taxon concepts and names> modern ccoinformatics provides tools to exchange data of different formats and taxouomic reference and, ultimately, link up databanks of any format in networks. Rather than enforcing standard formats, these systems require that data are recovered and stored with as much original information as possible. including meta-data on sampling design and methods, cover-abundance scales, definition of layers, laxonomic references, and original data sources. Classification of Vegetation Aims and Criteria Vegetation classifications arc performed with three fundamental goals: (1) delimiting and naming parts of the vegetation continuum to enable communication about them; (2) predicting a multitude uf ecosystem attributes (e.g., species composition, site conditions, and ecological processes) Irom the assignment of a particular stand to a vegetation unit; and [?) making multi-species co-occurrence patterns representable by verbal descriptions, tables, diagrams, and maps, l-'loristically defined vegetation types are thus suitable reference entities for ecological research, bioindication, and nature conservation. Reaching these aims requires of the classification approach: L coherence of units with respect to major ecosystem properties; 2. simple and clear discernabiliry of units; 3. completeness of the system (ie., coverage of all vegetation types of the given area); 4. robustness (i.e., minor changes of the data should not considerably change the classification); 5. tolerance against varying data quality; 6. supra-regional applicability; 7. applicability for a range of dlfferenr purposes; 8. hierarchical structure, allowing for different degrees of generalization; 9. equivalence of units of the same hierarchical level; and 10. adequate number of units with respect to practical use. As no single classification can ideally meet all of these criteria at the same time, and their relative importance depends on the purposes* competing classifications of the same objects and data ate a reality. Thus, the imerpreta-tion of local data will change with scaling up from local to regional and supra-regional context. However, there is also a practical requirement to have a unified supra-regional classification to enable communication among scientists, managers, and authorities between regions. Braun-Blanquet Approach The 'Braun-Blanquet approach' provides a methodological framework for vegetation classification that seeks an optimal combination of the above criteria and that reconciles conflicting requirements of different scales and