Vulnerability of Mediterranean forests to climate change, innovative methodologies for remote and in field monitoring of their health status and resilience to extreme events

Studies conducted over the past decades have shown that climate alterations, such as droughts and heat waves, are increasingly causing forest stress and dieback, affecting changes in the composition, structure and biogeographic distribution of forests. Dieback phenomena have been observed worldwide and are particularly affecting the Mediterranean basin, which is considered a climate change hotspot. Therefore, these phenomena can make our forests vulnerable, i.e. unable to tolerate stressors in time and space. This manifests itself in the reduction/absence of forest response to such climatic events, such as loss of resilience or the ability to recover, even to the dieback and death of entire forests. Thus, it is necessary to understand how forests respond to extreme climatic events and how such phenomena can be analysed. This report intends to present the research work carried out to investigate the vulnerability of forests to climate change, with the aim of improving current knowledge on the subject and understanding forest dynamics that could lead to future scenarios. Careful “review” work was carried out, which not only made it possible to analyse the current state of the art on forest vulnerability, but also to identify any uncertainties and critical issues. The assessment of forest vulnerability is very complex due to the many factors influencing the response of forests to climate change and the different methods used to analyse forest health. Through the review, we have highlighted the problems, needs and possible solutions that could be adopted to overcome some of the critical issues that research is facing. Specifically, the research activities involved the study of six forest sites located in Basilicata region (Southern Italy). The stands analysed were affected by the severe drought of 2017, showing symptoms of dieback (leaf yellowing, crown thinning and death), and thus appear to be suitable for investigating and assessing the response of forests to extreme climatic events, such as droughts and heat waves. In these areas, we attempted to understand whether previous favourable climatic conditions might have stimulated growth by making forests more vulnerable to subsequent drought stress. That is, we investigated possible structural overshoot phenomena that could contribute to forest dieback. This phenomenon attracted our attention because the studies and dynamics driving structural overshoot phenomena are still poorly understood. The six forest sites analysed fall within the municipalities of: Savoia di Lucania 'Orto Siderio site' (hereafter OS) and 'Grotta dell'Angelo site' (hereafter GA), Accettura 'Accettura Palazzo site' (hereafter AP), Vietri di Potenza (hereafter VP), Castelmezzano (hereafter CM) and Pietrapertosa (hereafter PI). These are mixed woods in which various Mediterranean species vegetate, such as Quercus pubescens Willd., Fraxinus ornus L., Acer monspessulanum L. and Pinus pinaster Aiton. For each site, an area representative of the entire stand (5000 m2) was identified, in which structural (diameter at breast height 1.30 m, plant height) and qualitative (to assess the state of defoliation of the crowns) surveys were carried out. At each site, 15 mature individuals of each of the two most abundant tree species were randomly selected and wood samples useful for the dendrochronological study were extracted from each tree. A total of 180 trees were sampled and 8531 annual growth rings were measured. In addition to the description of each site (exposure, soil, elevation, etc.) and the qualitative assessment of canopy transparency (percentage of defoliation), we also used remotely sensed data in particular the NDVI index (Sentinel-2 collections, spatial resolution 10 mX10 m) to further characterise the status of the canopies following the 2017 impact. Furthermore, by considering the Basal Area Increment (BAI) of trees before, during and after droughts, it was possible to understand whether and how stands respond to such events, by calculating the resilience indices (resistance, recovery, resilience, relative resilience and impact) for each species at each site. In order to have a broader overview of the drought events that occurred in the study area, resilience indices were also calculated considering previous drought years 2003 and 2012. In addition, the legacy effects of drought were highlighted by subtracting the observed increase in basal area (BAI) from that expected. The calculation of resilience indices showed that resilience and relative resilience after 2017 tended to be lower than during the droughts of 2003 and 2012, while impact tended to increase. F. ornus showed a high recovery after the 2017 drought at most sites, while oak shows greater resistance to the drought event. A. monspessulanum responded to drought similarly to Q. pubescens. Overall, all broadleaves studied showed a better response to the disturbance than P. pinaster (originated by artificial planting). For most sites and species, the reduction in growth due to drought was followed by a rapid recovery and positive legacy effects, particularly in the case of F. ornus. We found negative two-year drought legacies mainly in VP site (Q. pubescens and A. monspessulanum) and GA site (only in P. pinaster). A slight negative legacy effect is also observed in OS site (Q. pubescens only) probably caused by structural damage and widespread mortality affecting oak recovery in this area. Conversely, a slight positive drought legacy is observed on F. ornus at the OS site probably due to post-drought release in the competition (widespread oak mortality). At GA and VP sites, where a negative legacy effect is most observed, the high growth rates of forest stands prior to drought (i.e. structural overshoot) in response to previous wet winter-spring conditions may have predisposed trees to drought damage and the resulting negative legacy effect. Overall, the response of forest stands to the 2017 drought certainly depended on site conditions and the characteristics of the species studied. Indeed, not only the resilience indices, but also the NDVI values showed site-specific responses. Furthermore, the observed legacy effects allowed us to find evidence for our structural overshoot hypothesis at some experimental sites, highlighting that predrought conditions can influence and exacerbate drought damage in overbuilt stands (i.e. characterised by significant pre-drought vegetation growth probably beyond the normal carrying capacity of the site-specific ecosystem). In order to obtain more information on forest dynamics in response to climate change, we considered it appropriate to deepen the study, using wood anatomy and remote sensing, but focusing the analysis on four sites (AP, OS, CM, PI) where the same species, oak and ash, coexist, i.e. the most widespread species in Mediterranean thermophilic mixed deciduous forests. In this way, we were able to obtain quantitative anatomy elements, useful for analysing xylem traits in detail, and remote sensing information through the use of vegetation indices (NDVI, EVI, NDWI) with moderate resolution (Landsat collections, spatial resolution 30m X 30m). The aim was to observe how and whether radial increments and wood anatomy correlate with the remote sensing information. To perform the earlywood anatomical analyses (hereafter EW) we selected 6 wood samples, with the highest intercorrelation, for each of the two species analysed for each site (oak and ash). An analysis of the anatomical traits of the wood from 2001 to 2021 was performed for each of these. Dendro-anatomical studies are very laborious and timeconsuming to develop, but they are able to return detailed information, in fact anatomical variables (lumen size, number of xylem vessels, etc.) are considered important proxies for studying the response of forests to environmental changes. The anatomical analysis involved 48 woody samples, 1008 growth rings and a total of 17028 xylem vessels measured. We quantified anatomical traits such as: ring area, areas of the EW and late wood (LW), area (%) occupied by vessels in the EW, vessel lumen area and vessel density. These measurements allowed the calculation of hydraulic diameter (Dh) and potential hydraulic conductivity (Kh). The remotely sensed data were obtained for the period 2001-2021, using the Landsat collections (L5, L7, L8 and L9). The vegetation indices estimated are NDVI and EVI used to characterise canopy cover and photosynthetic activity and NDWI used to examine canopy water content. The result was the calculation of 960 (monthly) values for each index with a total of 2880 values analysed. In addition, using the thermal infrared bands (TIRS 1 and TIRS 2) from the Landsat collections, we assessed the thermal emissivity of the soil to further validate the impact of the 2017 drought. In general, growth decreased during the drought year (2017), especially for Q. pubescens, which showed considerable defoliation. Both species (oak and ash) showed a decrease in Dh in 2018 after the drought. Positive relationships were observed between the area of growth rings, Dh and remote sensing data (NDVI, EVI, NDWI) for Q. pubescens, confirming it as a very reactive species. At highly affected sites, such as the OS site, a covariation between Dh and growth (i.e. significant positive trends) was observed, such relations may be further confirmation of highly stressed conditions. Overall, even in this case, the results showed a high variability in drought response between the species and sites analysed, but ash seems to show less growth decline than oak. Therefore, analysing the anatomical variables, it would appear that F. ornus shows a better response to water stress phenomena and lower mortality than Q. pubescens. This also further confirms the results returned by the resilience indices, i.e. in general a better post-drought recovery of ash than oak. Thus, the use of dendrochronology and the study of resilience indices seem to indicate anomalies that have been further confirmed through the use of wood anatomy and remote sensing. This type of approach is relevant, as there are very few studies that employ a multiscale analysis of this type. Furthermore, this study contributes to documenting the ecological consequences of drought, i.e. probable changes in the composition of Mediterranean forests in response to climate change with the formation of communities dominated by species better adapted to drought events. The information obtained could, in addition to implementing the current state of the art on Mediterranean forest vulnerability and structural overshoot phenomena, help clarify the relationships between dendrochronology, wood anatomy and remote sensing, highlighting the opportunity to work at different multi-temporal scales using different techniques. Furthermore, our work provides information on widespread but little-studied Mediterranean species such as A. monspessulanum and F. ornus, and in particular the latter would seem to have a good chance of occupying forest sites significantly impacted by drought. In order to have more detailed information and to understand what the evolution of Mediterranean forests might be and thus what the future scenarios might be, multi-proxy analyses and long-term studies will be needed to monitor the state of forests and the dynamics of competition between different Mediterranean species.