One of the distinctive features of the LIFE AForClimate project is its provision of a decision-support tool for forest managers to assist in planning silvicultural interventions.

This tool, an online platform, offers two main functionalities:

• It identifies areas with lower-than-average potential forest growth rates;
• It estimates the probability of increased seed production (mast seed).

With this information, forest managers can implement silvicultural actions aimed at enhancing stand resilience.

The platform is published on the LIFE AForClimate website, and access is granted upon user registration.

There are three access levels within the platform:

• Administrator: has full access to the platform, including the ability to configure tool parameters;
• Moderator: can manage user requests and approve or deny registration requests;
• User: can register on the platform, submit requests, and monitor them through a personal dashboard.

Since the functionalities of this tool require input from the project’s scientific reference group, a two-step usage procedure has been established:

• Platform Registration: Users complete a specific form, after which their access is enabled by the service manager through a control panel;
• Request Creation: Once logged into the platform, users can access their personal dashboard to create and manage requests.

In the dashboard, users can track the status of submitted requests, create new ones, and view all requests either in a tabular format or graphically via the webGIS tool provided on the platform. Additionally, users can export a list of parcels, including the data from their requests, in Excel format.

The project focuses on increasing the average radial wood growth by defining one or more algorithms that correlate potential forest growth with recent trends in specific climate variables.

The platform has three data entry sections:

• Climate Database Section: This section can be populated with the manager's own data or directly with online data from the Copernicus Climate Database (https://cds.climate.copernicus.eu/cdsapp#!/home).
• Algorithm Input Section: Currently complete for the project's three application sites (Tuscany, Molise, and Sicily).
• Forest Management Plan Mapping Section: Allows for the entry of forest management plan maps.

The platform’s processing workflow involves four phases:

1. Importing the Forest Management Plan Intervention Map: The map, acquired in SHP format, must follow a specific structure. Key informational fields include an identification code for each plot, the planned silvicultural intervention, the intervention year, and the forest unit’s position relative to altitude and aspect (e.g., high-altitude, north-facing; low-altitude, north-facing, etc.). The platform includes a validation process to correct any topological issues with the input map.
2. Selecting the Algorithm: The algorithm used is specific to the geographic area of the Forest Management Plan. If a site-specific algorithm is unavailable, the project's scientific staff can assist in defining an appropriate algorithm for the application area. Currently, the platform contains algorithms for the three pilot forests where the project was implemented.
3. Selecting the Climate Data Set: The platform includes multiple data sets. It is designed to receive climate series data directly collected in situ from meteorological stations, as well as data from major online climate databases that provide both time series and spatial analyses of climate parameters.
4. Processing: Once the baseline conditions are defined, the processing phase is initiated.

The processing results in a table showing, for each plot or sub-plot subject to intervention, the anticipated potential growth level for the next growing season. In addition to the table, a GIS interface within the platform color-codes the plots: red for areas where a slowdown in potential growth is expected, and green for areas where growth potential is likely to be more favorable.

The platform includes a section dedicated to applying a mast prediction algorithm (ForeMast package) through the ShinyForeMast interface, developed during the AForClimate project. This set of functions calculates the probability of seed production for the current year, using only meteorological data. The data encompass average temperature and precipitation from the two previous years, calculated for the summer months (June, July, and August), as these factors have proven crucial in triggering years with abundant seed production.

The ShinyForeMast pages provide a guide for applying the algorithm to the specific geographic area where the user needs an assessment.

The algorithm generates a graph that represents the trend in mast probability over the climate period considered. The last reference year is the one relevant for prediction.

For management purposes, it is recommended to consider probabilities above 75% as favorable for implementing regeneration-promoting interventions.

The tools available on the platform have been designed and developed specifically for beech forests managed as high forests with uniform shelterwood cuts, where one goal is timber extraction. The dendrometric growth models may also be useful in initiating the conversion to high forest through a gradual reduction of coppice shoots.

As previously described, the platform provides two main outputs:

• An indication of whether the stand is likely to experience a period of reduced dendrometric growth responsiveness;
• The probability of abundant fruiting in the following growing season.

For growth responsiveness, the platform’s output provides two possible values: reduced growth or stable growth. A result indicating expected reduced growth should be interpreted as a sign that the stand has recently experienced suboptimal climatic conditions for growth. This alert signal, by itself, should prompt the manager to consider—within the broader management plan-whether an immediate mitigation action is warranted to prevent other stress factors. For example, if an intermediate cut is planned in the near future for this plot (e.g., within five years, meaning a currently dense stand), this intervention (which would reduce competition for water in a drought scenario) could be advanced.

Alternatively, a plot showing stable growth that is scheduled for thinning within the same year may not require immediate intervention. However, if another plot shows an alert signal, and the manager cannot (or does not wish to) intervene in both plots, they could choose to postpone the scheduled intervention on the stable plot and advance the one on the alerted plot.

Finally, this platform component is useful not only during the implementation of a management plan but also in developing future plans. Before defining management objectives and methods, the algorithms can be applied retrospectively with historical meteorological data, such as that from Copernicus, which the platform utilizes. In this case, plots repeatedly showing phases of reduced growth over recent years could be identified. For these plots, it would be advisable to adopt a different management strategy than that used for a typical beech forest with successive cuts, considering alternative interventions that introduce one or more adaptive strategies, as outlined in Chapters 3 and 4 of the AForClimate Guidelines. As a last resort, a gradual species change might also be considered.

Regarding growth modeling, it should be noted that the algorithms have local validity and should be developed during management plan drafting and applied within its scope. However, for practical purposes, it may be feasible to extend their application geographically and temporally. In any case, it is recommended that usage be limited to a geographically confined area and that model revision be considered after 20 years, equivalent to approximately two management plan cycles.

For mast prediction, the algorithm should be applied when carrying out a seed cut at the beginning of the regeneration phase in a beech high forest. In this case, the algorithm’s output is a probability value between 0% and 100%. A high mast probability (above 75%) indicates a low likelihood that insufficient seed will be produced for future regeneration after the seed cut. Alternatively, even probabilities above 50% may be a calculated risk in cases where (for example, due to the stand’s homogeneity and maturity) regeneration is still desirable in part of the forest.

It is important to note that the algorithm has been tested and validated based on percentile seed production across various field observations in Europe. Therefore, the algorithm has pan-European territorial validity and can be used across the range of beech distribution. However, since the algorithm provides percentile seed production relative to the average production at each site, managers should assess whether a 50% probability (sufficient for sites with generally abundant production) is acceptable, or whether a 75% threshold (recommended for sites with lower absolute seed production) should be awaited.

The above examples are general considerations that must be adapted to the specific context of the approach recommended by AForClimate. Consulting with the CREA Forests and Wood research team is recommended during both the drafting of the plan and the platform’s application to an existing plan.