Data banking and elaboration of large data sets.

Data represent the basis of any scientific research; large data sets are particularly necessary when addressing large-scale research questions in biogeography and ecology of plant species and communities. It is well known that phytosociological syntaxa are vegetation models that can constantly be refined thanks to improved knowledge and improved analytical and statistical tools. In this sense, relevé databanks play a special role in monitoring and analyzing vegetation (Knollová et al., 2005). With the purpose to create an Italian database for archiving and sharing, the project AnArchive (http://www.anarchive.it) has been developed in the last decade. AnArchive is a collaborative project that involves several Italian universities coordinated by the University of Perugia (Venanzoni et al., 2011). The natural evolution of this project is VegItaly (http://vegitaly.anarchive.it), a web database designed to store, retrieve and analyze vegetation data of Italy. Designed for both floristic and phytosociological data, it can handle any field plot according to the definition of “vegetation database” suggested by the Global Index of Vegetation-Plot Databases (GIVD) and the IAVS Eco-informatics working group. VegItaly is supported by Società Italiana di Scienza della Vegetazione (former Società Italiana di Fitosociologia) (Venanzoni et al., 2011). Besides a number of useful tools, VegItaly offers the possibility to sort and extract large thematic data sets sharing similar floristic and/or ecological attributes, starting off new chances for phytosociological reviews at a wide geographical scale. As a direct consequence, the need to improve data management techniques, with special reference to vegetation analysis, became prominent and led to screen different methods for vegetation classification as alternatives to those traditionally used in Italy. In fact, many well known numerical classification algorithms do not always produce satisfactory results and often require further subjective interpretations and manipulations, such as manual re-allocation of relevés among clusters (Bergmeier, 2002; Podani, 2007). One alternative to the numerical classification algorithms is the Cocktail method (Bruelheide, 2000) modified by Kocí et al. (2003), successfully used in central Europe for over a decade. This method is designed to be applied to large data sets simulating the Braun-Blanquet approach and to create formal definitions of vegetation units. On this ground, a large amount of data including the vegetation of central Italy was stored into AnArchive (about 4,500 relevés) in order to test the Cocktail method. To perform the method, the database system was interfaced with the program Juice 7.0 (Tichý, 2002) creating a compatible exporting data system. A case study restricted to aquatic plant communities and wetland vegetation of central Italy is here presented. Formal definitions were created for plant communities included in the classes Lemnetea, Potametea, Charetea and Phragmito-Magnocaricetea. Formal definitions will be integrated in the project VegItaly in order to crate a simple classification method available to users.