The database was assembled by the BIOME Lab of the University of Bologna and is described in a data paper in the journal Vegetation Classification and Survey. Data collection began in early 2005, with material stored on a local server, and the various collections were merged into a single database only in 2018. The first nucleus came from a survey of vegetation on small circum-Italian islands, which was then expanded with data from projects in the Alps, the Apennines and the Po valley. Plots were drawn from formal publications such as articles, books and monographs, but also from a large body of grey literature, including degree theses, doctorate theses and technical reports. The team deliberately avoided sources already held in other Italian databases in order to limit duplication. At the time of the report, 245 sources had been used to extract data.

Almost all of the plots were collected according to Braun-Blanquet’s phytosociological method, the classical European approach in which a botanist records every plant species within a defined area and estimates the cover of each, although other approaches, such as estimating percentage cover for all occurring species, were also considered. The data are stored in Turboveg, software that is recognised as the official tool of the European Vegetation Archive. For each plot, the site information reported in the original source, such as locality, elevation, slope and aspect, is preserved exactly as published, even where those values may be wrong. The original taxon names recorded in each plot are also maintained, generating a list of 20,682 taxa, some identified at the genus level but most at the species or subspecies level, including vascular plants, bryophytes and lichens. The researchers kept original names to ensure traceability and to avoid accumulating taxonomic interpretations, leaving standardisation to the researchers who reuse the data.

Georeferencing required substantial detective work. Not all sources provided coordinates, particularly those published before the 2000s, so the team inferred coordinates where possible from locality names, converting authors’ textual indications into the best approximated geographic coordinates with varying degrees of accuracy estimated in metres. Accuracy is estimated through geographic and topographic checks using elevation, slope and exposition supplied by the source, together with remote sensing information in a GIS environment, and plot locations on maps are always cross-checked. The team is also dedicating significant effort to contacting original authors to locate coordinates and improve the quality of plot geolocalisation. The precision of georeferencing spans from very accurate coordinates obtained by high-precision GPS to a maximum of 30,000 metres for some old plots lacking detailed sampling information, although precision values above 10,000 metres are exceptional, at roughly 1.4 percent of plots. In most cases, precision falls between 201 and 1,000 metres, which accounts for 32.4 percent of plots, while 22.8 percent fall between 0 and 50 metres. A total of 502 plots, 2.9 percent of the database, could not be georeferenced at all, and 2,124 plots, 12.1 percent, lack an accuracy estimation because the original sources did not indicate one or the value was not credibly estimable.

The content of the database reflects decades of shifting sampling effort. The distribution of plots per decade shows a progressive increase over time, with the lowest values in the first half of the twentieth century and the highest in the 1990s, when 5,010 plots were recorded in the decade 1991 to 2000. Of the total, 12,490 plots have been formally published and 5,015, or 28.6 percent, remain unpublished. Published sources account for 71.4 percent of the data, theses for 11.9 percent, and unpublished data from specific vegetation-sampling projects for 14.4 percent, with the remainder coming from ongoing fieldwork by members of the BIOME Lab. Geographically, the plots are overwhelmingly Italian, with 16,274 plots from Italy, followed by 916 from Malta, 134 from Austria, 97 from the Republic of San Marino and 84 from Slovenia. Within Italy, the plots principally come from Veneto, Tuscany, Emilia-Romagna and Sicily, which together account for 68.0 percent of the Italian total, and the highest density of plots occurs at low altitudes between 0 and 400 metres above sea level.

Species richness per plot ranges from 1 to 103 species, with an average of 19.3 plus or minus 12.2 standard deviations, and the most frequently observed value is 16 species per plot. Plot areas range from 0.01 square metres to 1,000 square metres, with 100 square metres the most frequent value, one of the standards used by the team during field sampling; for 2,111 plots, or 12.1 percent of the total, no plot surface was indicated in the original sources. The database is organised into thematic collections. The largest, Small Mediterranean Islands, holds 5,723 plots, of which 1,048 are still unpublished, sampled between 1950 and 2021 on 54 islands and islets of the central Mediterranean, with the Tuscan Archipelago, the Maltese Islands, the Aeolian Islands, the Pelagian Islands and the Tremiti Islands the most heavily sampled archipelagos. The Tuscan Archipelago alone contributes 1,616 plots, and the corpus has been used to test fundamental biogeographical theories such as island species-area relationships and the role of habitats as predictors of species richness and composition within islands.

The Dolomites collection contains 2,387 plots gathered between 1955 and 2019, of which 221 are unpublished. The majority, 1,962 plots or 82.2 percent, were recorded by Sandro Pignatti and Erika Wikus Pignatti over several decades and largely used to produce a monographic book, with the original data files kindly provided by the Pignatti family and part of two bachelor’s theses devoted to georeferencing each plot. These plots span altitudes from 350 to 2,950 metres, with the highest concentration between 1,500 and 2,500 metres, and describe montane and alpine vegetation ranging from broadleaved forests and steppe grasslands to alpine grasslands, conifer formations and pioneer vegetation of the highest altitudes. The Italian National Parks collection includes 998 plots recorded between 1930 and 2021, 415 of them unpublished, many from the Foreste Casentinesi National Park with its old-growth mixed forests of Fagus sylvatica and Abies alba. The 12 plots performed before 1950, the oldest dating to 5 August 1930, are likely the most ancient data of such type at the Italian level and among the most ancient at a European level, and the collection includes data from the Sasso Fratino Forest, the first integral nature reserve of Italy.

Further collections broaden the thematic reach. The Protected Areas of Emilia-Romagna dataset holds 1,681 plots recorded between 1940 and 2007, of which 1,505 are available at a dedicated data centre of the regional administration, covering nearly 2,000 square kilometres with vegetation from halophytic coastal communities to Apennine grasslands. A serpentine outcrops dataset counts 231 plots from 1994 to 1998 on mafic and ultramafic substrates in Tuscany and eastern Liguria, already published and hosting specialised endemic species such as Alyssum bertolonii. A segetal vegetation dataset counts 779 plots from 1952 to 2019 in cereal and other cultivated areas of northern and central Italy, all published, and part of this group was already used in a comprehensive study of arable-habitat vegetation of central Europe, although the researchers note there is no overlap between the resulting European Weed Database and their own data. The Maremma Regional Park dataset holds 90 unpublished plots from 2007 to 2009, recorded according to a probabilistic sampling design and therefore spatially representative of park conditions, all with spatial accuracy of less than one metre. A miscellaneous collection of 5,616 plots from 1951 to 2021 covers quarries, springs, river beds, rocky hills, forests and marine coasts.

The team stresses that data entry is ongoing and that all figures were updated to 24 June 2022. Active projects include the digitization of a 2015 survey of gypsum outcrops in Emilia-Romagna of roughly 160 plots, a project on Apennine chestnut groves of about 500 plots, a project on drainage canals of the Po valley of about 120 plots, a project on Natura 2000 protected areas in the province of Siena of about 600 plots, and resurveys of historical plots, alongside efforts to avoid duplication and taxonomic inconsistencies with other major Italian databases. Within the European Vegetation Archive, data accessibility is semi-restricted, so a formal request with a project proposal must be submitted. With 13 Italian vegetation databases registered in the Global Index of Vegetation-Plot Databases at the time of writing, the authors position AMS-VegBank as an important reference for making the vegetation-plot data of many Italian and other Mediterranean areas available to the global scientific community, and as a tool for studying plant biodiversity within continental and global vegetation-plot databases.

Subject of Research: A new vegetation-plot database compiling Italian and Mediterranean plant co-occurrence observations from 1930 to 2021

Article Title: AMS-VegBank: a new database of vegetation plots for the Italian territory

Article References: Alessi, N., Bruzzaniti, V., Buldrini, F., Centomo, E., Cervellini, M., Enea, M., Landi, S., Lelli, C., Montanari, I., Nascimbene, J., Pezzi, G., Virzí, G., Zannini, P., & Chiarucci, A. (2022). AMS-VegBank: a new database of vegetation plots for the Italian territory. Vegetation Classification and Survey, 3, 177-185. https://doi.org/10.3897/vcs.85083

Image Credits: AI Generated

DOI: 10.3897/VCS.85083

Keywords: vegetation plots, plant biodiversity, Italy, phytosociology, georeferencing, Mediterranean islands, Dolomites, national parks, database, Braun-Blanquet, EVA, species records