INTRODUCTION
Biodiversity is a crucial yet complex concept in ecological research. Beta diversity, representing species turnover across spatial gradients, plays a key role in understanding ecosystem functioning and conservation planning. Studies suggest that environmental factors such as altitude, latitude, and geographical distance drive beta diversity patterns. However, large-scale analyses may not directly inform local conservation efforts. Therefore, fine-scale assessments within specific administrative regions are essential for effective biodiversity management and protected area planning. In FORCING project (Geri et al.2016), the Edmund Mach Foundation and the University of Trento recovered a huge database of vegetation surveys, build in the 1970s to represent the "Schmid's vegetation belts" in the forests of Province of Trento a region of about 6.212 km² in the northeastern Italian Alps with a huge flora and fauna biodiversity (Tattoni et al. 2021). The surveys and the cartographic materials were digitized and organized in a geographic geodatabase with QGIS and postGIS. The sampling design of this archive lends itself perfectly to being analyzed from a beta diversity perspective, permitting to compare several environmental gradients in terms of species turnover and species richness. The archive was created using FOSS4G and is permanently available and stored in a web-GIS hosted on servers maintained by Fondazione Edmund Mach and is accessible at http://meteogis.fmach.it/forcing/ (unfortunately due to a technical problem related to an ongoing general software update the access maybe unavailable until the end of 2025).
The aim of this paper is to test the use of FOSS4G software (Ciolli et al. 2017) and the FORCING geodatabase (Geri et al.2016) to perform an exploratory analysis of the floristic species turnover in a relative small area, and to try to underline some patterns and driving forces.

METHODS
The data coming from the original sampling project were managed and prepared using Qgis software, using the Spatialite format geographic database and georeferenced in the WGS84 UTM 32N coordinate reference system (srid: 32632). 517 linear transects and a total of 190761 species records were analysed (Geri et al. 2016). The statistical analysis were performed with R software. In each linear transect, considered a single ecological community, the beta diversity using site as simple point were calculated evaluating in this way the degree of species turnover across the environmental gradient that is created along the transect (Tuomisto, 2010). The basic statistical properties extracted for each transect were put in relation with the beta diversity index and with the corresponding values of species richness, producing graphs that shows the various relations trend. The significance of the linear relations were tested using the Pearson correlation coefficient. Each belt was compared in terms of species composition using the Sørensen’s coefficient of similarity. The behavior of the beta diversity and species richness were deepened in terms of variance partitioning. It was tested the variance explained by the four variables: mean altitude, mean slope, range of altitude and range of slope against beta diversity and species richness. The analysis should stress the role of the variable in single or in multiple way to drive the species turnover. The variance partitioning analysis were processed using the Vegan library of the R statistical software (R Core Team, 2024), and in particular the module “varpart”. This function partitions the variation of response data table with respect to two, three, or four explanatory tables, using redundancy analysis ordination (RDA). To simplify the results interpretation the variance partitioning in combinations of group of three variables was applied. Both terrain altitude and slope data and both vegetation beta diversity and species richness data were transformed with a log transformation in order to obtain a normal distribution of data. QGIS was used also for data exploration and representation.

RESULTS
Pearson indices show that all the variables are significative except the slope variance for both beta diversity and species richness and the mean altitude only for species richness. Generally both species richness and beta diversity grow increasing altitudinal range, slope range and slope mean while variance doesn’t show a definite trend considering in particular way the species richness. Sorensen statistic shows how the similarity decreases with increasing the altitude separation from the lower level, and highlights the pairwise comparison between altitude adjacent belts. The latter statistic shows how the similarity presents a different behavior with the increase of altitude, rising very fast in the first step (between 0 and 600 meters) then leveled off and finally decrease in correspondence to the last two steps, between 1500 meters to 2100 meters. The variable that explain much more variance is the altitude variance for both beta diversity and species richness. Regarding beta diversity the greater joined effect is due to the combination of altitude range and slope mean while altitude range, slope mean and slope range presents an higher joined effect for three variables.

DISCUSSIONS AND CONCLUSIONS
The results confirm that the transects characterized by a wider range of slope and elevations show a higher rate of beta diversity. This is reasonable, since the more are the different environments the transects cross, the more pronounced should be beta diversity. This is also confirmed by the linear relation of the single variables highlighted both as graphical trend and by the pearson tests and moreover, by the fact that the variance is explained as a joint action of variables.
Finally this work confirmed that FOSS4G software is perfectly suitable to be used to perform spatial statistical analysis to study beta diversity both from the point of view of numerical statistic and from the point of view of geostatistics (Ciolli et al. 2017) showcasing the power and versatility of these tools.
Further future developments and analysis will include the comparison of beta diversity of the present vegetation with other historical floristic archives sampling (Lelli et al 2023), statistical analysis of the data using different set of statistical and geostatistical techniques and finally to include remote sensed data.

The transition towards more sustainable transport together with a worldwide push for decarbonization promotes the adoption of light-duty electric vehicles (EVs). Nevertheless, for EVs to run on par with or better than internal combustion engine vehicles, they require convenient enough charging infrastructure (Knez et al., 2019). EV charging infrastructure must accommodate shifting demands in terms of density (queuing), frequency (coverage gaps), and dependability (outage) (Hanig et al., 2025). Even if only a small fraction of all car trips are longer than 50 miles (well within the range of today's EVs), long-distance drivers' concerns about charging tend to have a disproportionate effect on their decision to buy a car (Haidar et al., 2022). Moreover, changing stations availability can be critical when choosing a turistic destination.
This research project analyzes the availability of EV charging stations in the Provincia Autonoma di Trento (PAT), a region in the Italian eastern Alps, a popular touristic destination for Italians and northern Europeans.
While an Italian national repository, PUN, "Piattaforma Unica Nazionale dei punti di ricarica per i veicoli elettrici" of the Ministero dell'Ambiente e della Sicurezza Energetica (Single National Platform for Charging Points for Electric Vehicles of the Italian Ministry of Environment and Energy Security) is available for consultation, its dataset cannot be downloaded as a map or a table for processing. Therefore the Open Charge Map dataset, available under the Creative Commons Attribution 4.0 International license (CC-BY 4.0) license, has been used. While this charging points database is far from complete, it is fairly representative of the distribution and density of the charging stations. The JSON dataset for Italy has been converted to CSV and the points within the Provincia Autonoma di Trento have been extracted.
The road network has been provided by the local government, Provincia Autonoma di Trento, with a 1:10000 scale, again under the CC-BY 4.0 license. Only the paved roads have been used.
The road network and the charging stations have been combined, placing a node in each station, at the each road intersection and on each road extremity.
With this configuration, the distance of each road to the closest charging station, defined as the minimum distance of the starting or ending node of the arc representing the road, has been evaluated: the minimum distance is below 1 km for most of the roads, with only a few roads above 7 km.
To provide a better representation of the distance between charging stations and potential users a set of points has been created along the roads with a distance of 500m. The distance to charging points has been evaluated for these 8975 points. Nodes belonging to roads shorter than 100m have been removed because they would have too mach influence on the distance distribution.
The mean distance from the charging points is 4749.4 m, with a standard deviation of 4592.6 m. The maximum distance of 36766.6 m, and, as expected the minimum is 100 m. Only 3161 (35.22%) points have a distance above 5 km and 1104 (12.30 %) above 10 km.
To analyze the distribution of the charging stations their density has been evaluated by extracting the charging points for each municipality. The province has 166 municipalities, ranging from relatively large cities in the main valleys to very small municipalities in secondary valleys.
The number of charging stations per municipality is quite low, 1.9 on average, but 72 (43.4%) municipalities have no charging points at all. For the other 94 (56.6%) municipalities which do have at least one charging station, the average number is of 3.32 charging point per municipality, with a standard deviation of 3.79.
Results are compatible with a recent Italian national report (MOTUS-E, 2025) indicating that more than 40% of the municipalities have no charging stations. Moreover, around 30% of Italy has a distance to the nearest charging station above 5 km, 6% above 10 km. However, results are not really comparable because the national report does not employ network analysis but a coarse raster analysis with 1 km resolution and, more importantly, it takes advantage of the access to a more complete charging stations dataset.
Future developments include the repetition of the analysis for other Italian regions, the differentiation of the analysis per types of EV chargers and the use of a more comprehensive charging stations dataset. The availability of traffic data is being investigated since it would make it possible to verify whether the charging stations distribution match the traffic distribution or it is possible to optimize its configuration to serve the largest number of vehicles.
The main limitations of the analysis come not from the processing tools but from the insufficient availability of data, which are often in fragmented, proprietary and inaccessible datasets.
All analyses and statistical and spatial processing were carried out using only FOSS, demonstrating the power and versatility of these software tools. In particular, topological analysis has been implemented using python with numpy and geopandas for data processing and igraph for network analysis. The Matplotlib library has been used for data visualization. QGIS has been used for coordinate conversion, map representation, table processing and geoprocessing.