Fire Affects Tree Growth, Water Use Efﬁciency and Carbon Sequestration Ecosystem Service of Pinus nigra Arnold: A Combined Satellite and Ground-Based Study in Central Italy

: The Mediterranean basin is an area particularly exposed to ﬁre risk due to its climate and ﬁre-prone vegetation. In recent decades, the frequency and intensity of wildﬁres increased, leading to negative effects on forests, such as a decrease in tree growth or an increase in tree mortality, producing a relevant loss of carbon sequestration ecosystem service. This study of the impacts of ﬁres on forests is fundamental for planning adequate forest management strategies aimed at recovering and restoring the affected areas. In this framework, our research delves into the effects of a forest ﬁre that, in 2017, affected a forest of black pine ( Pinus nigra Arnold) in Central Italy. Combining satellite and terrestrial analyses, this study evaluated the impact of the ﬁre on tree growth, water use efﬁciency and carbon sequestration capacity. Our ﬁndings highlight the importance of using remote sensing for the accurate identiﬁcation of ﬁre-affected areas and precise planning of ground-based activities. However, the integration of satellite data with forest surveys and sampling has proven crucial for a detailed understanding of ﬁre’s effects on trees. Dendrochronology and stable isotopes have revealed the post-ﬁre growth decline and altered water usage of defoliated trees. Furthermore, the quantiﬁcation of CO 2 sequestration highlighted a signiﬁcant reduction in carbon uptake by damaged trees, with severe implications for this ecosystem service.


Introduction
The Mediterranean basin is considered to be a wildfire hotspot due to its characteristic climatic conditions and fire-prone forest vegetation [1].In recent years, due to climate change, the frequency, intensity, and spatial extent of fires in this area have increased, with serious ecological, economic, and social consequences [2,3].Although fires play a crucial ecological role for Mediterranean vegetation [4], severe damage to plant tissues can have negative effects on the eco-physiological functioning of trees, resulting in decreased growth, impaired photosynthesis rates, and tree mortality [5,6].The degree of damage to trees is directly correlated with the severity of the forest fire [7].High-severity fires could lead to significant trunk damage and complete destruction of a tree's canopy, resulting in immediate tree death [8].Conversely, fires of low or moderate severity may not kill the trees in the short term, but they can affect tree's physiology and exacerbate existing climate stresses, increasing plants' vulnerability in the middle and long term [9][10][11].
The study of forest fires is of fundamental importance for the preservation of forest ecosystem services.In fact, when considered at a population level, the impacts of forest fires significantly alter the ecological functions provided by forests, such as regulation of the hydrological cycle, provision of food and timber products, soil stability, and biodiversity conservation [12].Moreover, the physiological dysfunctions or even deaths of trees caused by fires greatly reduce forests' capability to sequester CO 2 from the atmosphere and, Forests 2023, 14, 2033 2 of 17 consequently, to counteract climate change [13].Therefore, it is essential to accurately identify and promptly map areas affected by fire, as well as to have a comprehensive understanding of the eco-physiological impacts of fires on key trees species.This knowledge is crucial for implementing effective post-fire management practices in fire-affected areas and developing suitable landscape management strategies for future fire regimes [14].
To evaluate the impact of a fire on forests, surveys and land-based measurements are essential.In this context, dendrochronological analyses are widely used to study how tree species respond to various fire regimes [7,[15][16][17][18][19]. Tree rings provide valuable growth information, enabling the reconstruction of carbon allocation capacity over time and determining if it was affected by stress events such as fire [20,21].Dendrochronology studies are often combined with stable carbon isotope analyses in tree rings to assess the intrinsic water use efficiency (WUEi) of forests [22,23].The examination of tree WUEi, which is the ratio between net photosynthetic velocity and stomatal conductance, enhances our understanding of the eco-physiological effects of fires [24].
In recent years, the advancement of new technologies has facilitated the integration of land-based studies with remote sensing data.The use of satellite data in the ecological field has greatly contributed to the development of studies of forest fires, allowing a precise assessment of their extent, intensity, and severity regarding forest resources [25].From the comparison of the pre-and post-fire satellite images, it is possible not only to quantify the immediate impacts of fire on forests but also to evaluate the vegetation recovery in the following years [26].One of the most used satellite indices is the Normalized Burn Ratio (NBR), which allows us to identify burned area and measure burn severity [27,28].Similarly, the Normalized Difference Vegetation Index (NDVI), derived from optical remote sensing imagery, is often used to quantify vegetation greenness and assess variations in plant density and health [29].Therefore, this index can be highly useful for monitoring post-fire vegetation dynamics: the usage of data obtained from satellite image allows us to organize precise land-based forest survey campaigns to study fires [30] and plan immediate forest management interventions [31].
Our research aimed to understand the impacts of a forest fire that affected a stand of Pinus nigra Arnold (black pine) on Monte Cairo in Central Italy in July 2017, combining satellite information and land-based forest surveys.Our goal was to use the satellite images to discriminate, within a large forest area affected by the fire, the stands in which trees had experienced significant defoliation due to fire compared to stands where trees were not severely affected.Based on this mapping, we conducted a structural survey in the field and dendrochronological sampling to evaluate the impacts of the fire on growth and water use efficiency of the trees, as well as on the ability of the stands to sequester CO 2 from the atmosphere in the post-fire years.We expect the identification of the more damaged areas via the NBR to be confirmed via ground surveys, with a higher degree of defoliation and tree mortality.Furthermore, we hypothesize that the fire has negatively affected the efficiency of water use and the carbon allocation for the growth of defoliated trees, thus resulting in a serious negative impact on forest ecosystem services.

Study Area
The study area is located on the Monte Cairo (1669 m asl) massif of the Lazio region in Central Italy (Figure 1A).This area is characterized by a typical Mediterranean climate, with mild and wet winters and hot and dry summers: the average temperature and precipitation in the last 50 years were 13.1 • C and 1267 mm/year, respectively (the thermo-pluviometric diagram of the study area is reported in Supplementary Material Figure S1).The forest vegetation of Monte Cairo is mainly composed, at the lowest elevation, of sporadic stands of Castanea sativa and deciduous oak forests of Quercus ilex, Quercus pubescens, and Quercus cerris.At higher elevations, on the southwestern slopes, stands of Pinus nigra occur, while in the highest and less accessible areas, Fagus sylvatica forests are present (Figure 1B).
forest is located at altitudes ranging from 700 to 1550 m above sea level in the municipality of Terelle (Frosinone), characterized by even-aged trees (with an average age of 50 years) that are uniformly distributed (the pre-fire tree density was around 400 trees per hectare), as well as by an undergrowth with sparse herbaceous and shrubby plants.At the end of July 2017, this population has been hit by a severe wildfire, classifiable as a crown fire, which affected several hectares of the Pinus nigra forest, causing serious damage to the tree canopy, scorching the trunk and showing blackening on the bark.

Mapping of the Fire Damage from Satellite Imagery
In order to identify the areas burned in 2017 and provide a measure of burn severity, the Normalized Burn Ratio (NBR) was used.Two images from the Landsat 8 database were chosen for this purpose, with one representing the pre-fire condition and the other one representing the post-fire condition.These images were downloaded from the Earth Explorer portal (https://earthexplorer.usgs.gov/,accessed on 1 June 2023) as Level 2 surface reflectance data [32].For post-fire, the image recorded on 19 October 2017 was chosen.For pre-fire, the image recorded on 1 November 2016 was used to minimize the effect of vegetation phenology on the assessment procedure.
For each selected satellite image, the NBR was computed following Equation (1), and the difference between the pre-fire and post-fire (hereafter dNBR) was calculated, following Equation ( 2 where NIR is the near infrared, which corresponds to the Band 5 of the Landsat 8 OLI sensor, recorded in the 851-879 nm wavelength; while SWIR is the short-wave infrared, which corresponds to the Band 7 of the Landsat 8 OLI sensor, recorded in the 2107-2294 nm wavelength. The dNBR was used to discriminate between defoliated stands (DS) and non-defoliated stands (NS) within the Pinus nigra forest after the 2017 fire according to the table proposed by USGS [34].The procedure was performed using QGIS software, version 3.22.Our research focused on the Pinus nigra forest planted around 1970 and covering an approximate area of 223 ha along the slope of Monte Cairo (41.573 • N, 13.773 • E).This forest is located at altitudes ranging from 700 to 1550 m above sea level in the municipality of Terelle (Frosinone), characterized by even-aged trees (with an average age of 50 years) that are uniformly distributed (the pre-fire tree density was around 400 trees per hectare), as well as by an undergrowth with sparse herbaceous and shrubby plants.At the end of July 2017, this population has been hit by a severe wildfire, classifiable as a crown fire, which affected several hectares of the Pinus nigra forest, causing serious damage to the tree canopy, scorching the trunk and showing blackening on the bark.

Mapping of the Fire Damage from Satellite Imagery
In order to identify the areas burned in 2017 and provide a measure of burn severity, the Normalized Burn Ratio (NBR) was used.Two images from the Landsat 8 database were chosen for this purpose, with one representing the pre-fire condition and the other one representing the post-fire condition.These images were downloaded from the Earth Explorer portal (https://earthexplorer.usgs.gov/,accessed on 1 June 2023) as Level 2 surface reflectance data [32].For post-fire, the image recorded on 19 October 2017 was chosen.For pre-fire, the image recorded on 1 November 2016 was used to minimize the effect of vegetation phenology on the assessment procedure.
For each selected satellite image, the NBR was computed following Equation (1), and the difference between the pre-fire and post-fire (hereafter dNBR) was calculated, following Equation ( 2 where NIR is the near infrared, which corresponds to the Band 5 of the Landsat 8 OLI sensor, recorded in the 851-879 nm wavelength; while SWIR is the short-wave infrared, which corresponds to the Band 7 of the Landsat 8 OLI sensor, recorded in the 2107-2294 nm wavelength. The dNBR was used to discriminate between defoliated stands (DS) and non-defoliated stands (NS) within the Pinus nigra forest after the 2017 fire according to the table proposed by USGS [34].The procedure was performed using QGIS software, version 3.22.
To assess the dynamics of post-fire forest vegetation over the years, we computed the Normalized Difference Vegetation Index (NDVI) values for the DS and NS areas.The satellite imagery used were obtained from LANDSAT 8 via Google Earth Engine (accessed on 1 June 2023).Data from 2013 to 2022 were utilized, selecting only images not affected by cloudiness over the study area.The computation of NDVI, as a measure of canopy greenness, was performed for the DS and NS areas.Subsequently, the mean annual NDVI values, excluding outliers, along with their corresponding relative standard deviation, were graphically represented for each area.The equation used for NDVI calculation is as follows: where NIR represents the near-infrared reflectance (LANDSAT 8 band 5), and R corresponds to the visible red reflectance (LANDSAT 8 band 4).

Field Surveys and Sampling Campaign
After the elaboration of the fire damage map, land-based surveys and sampling operations were carried out.For the two identified areas (DS and NS), three random experimental plots, each with a diameter of 20 m (surface of 314 m 2 ), were selected.In order to ensure consistent comparison, the experimental plots were positioned at the same altitude of 1200 m above sea level.In each plot, the total number of trees was counted, and for each tree, the diameter at breast height (DBH) and the height were measured.Dead trees, when present, were counted and measured.Furthermore, the heights of trunk scorching were measured, as well as the potential tree crown defoliation, through visual estimation [7].Finally, dendrochronological samples were collected in each experimental plot using an incremental borer (Haglöfs, Långsele, Sweden).For both DS and NS, a total of 10 dominant trees were chosen, from which two wood-cores were taken in the southern and western directions, following standard procedures [35].

Dendrochronological Measurement and Processing Data
The collected wood cores were mounted on dedicated supports and underwent a sanding process to facilitate the identification of tree rings.The measurement of ring widths was carried out using the LINTAB system, a stereomicroscope connected to a computer equipped with TSAP-Win software, enabling the analysis and processing of dendrochronological curves that represent the growth of each individual tree.After measuring each sample, a statistical comparison of the curves was conducted, utilizing cross-dating through the Gleichaeufigkeit index (GLK), which assesses the correlation between different series [36].Afterwards, the Basal Area Increment (BAI) was calculated from the raw chronologies.BAI is a sensitive and biologically meaningful indicator of tree growth, widely utilized for evaluating the impacts of stressful events on forests and trees [23,37].Indeed, BAI series helped to minimize the influence of tree size and age on annual growth patterns while preserving the high-and low-frequency signals embedded within the tree-ring series [38].BAI was computed using the following equation: where BAI at year t is equal to the annual ring area (expressed as cm 2 ), r represents the radius of the stem at the end of the annual increment, and r t−1 represents the stem radius at the beginning of the growing season.
The BAI of the trees in DS was compared to that in NS in the five years before the fire event to test whether both groups previously had the same growth rate.It was then compared to the five years after the fire to check if any differences appeared after the fire event.The t-test was used to compare the two groups.The dendrochronological data analyses were performed in R using the dplR library [39,40].
Finally, to evaluate the influence of the climatic conditions of Monte Cairo on the Pinus nigra stands, the correlation was tested between the BAI data of DS and NS in the pre-fire period  and meteorological data (maximum, minimum, and mean temperature, and precipitation) obtained from the ERA5-Land database with a spatial resolution of 11 km [41]; the ERA5 data for the study area were extracted via Google Earth Engine (accessed 1 June 2023) [42].Data of post-fire years (2018-2022) were excluded to prevent the fire event from affecting the correlation results.The correlation test was performed using the dcc and seascorr functions of the treeclim package in RStudio environment [43] to test the correlations both monthly and seasonally [44].

Intrinsic Water Use Efficiency Analyses
For the two selected areas, DS and NS, six wood cores were selected based on their strong cross-dating (GLK > 0.70) with the corresponding reference chronology in order to perform the WUEi analysis.The annual rings of the last 10 years (2013-2022) were carefully separated using a blade cutter and grounded using a pulverizing mill (ZM 1000, Retsch, Haan, Germany), with a mesh size of 0.5 mm used to ensure homogeneity.Then, samples were precisely weighed in tin capsules for the δ 13 C measurements.The carbon isotope composition was determined at the Icona Laboratory (University of Campania, Caserta, Italy) by means of an isotope ratio mass spectrometer (Delta V Advantage, Thermo Scientific, Waltham, MA, USA) connected in continuous flow to an elemental analyzer (EA 1112 series, Thermo Scientific).The δ 13 C measurements were calibrated using certified international standards, and the standard deviation of the repeated analysis of an internal standard was less than 0.1‰ for δ 13 C. Carbon isotopic ratios are expressed using the delta notation: where a represents the fractionation factor due to CO 2 diffusion through stomata (4.4‰), while b represents the fractionation factor due to the Rubisco enzyme during photosynthesis (27‰).ca and ci are atmospheric and leaf intracellular space CO 2 concentrations, and 1.6 is the ratio of the diffusivities of water and CO 2 in the atmosphere.
After measuring the δ 13 C values, they were utilized to calculate the WUEi, which represents the ratio between net photosynthesis (A) and stomatal conductance (g s ) [45]: The calculation of carbon isotope discrimination (∆) was performed based on the difference between the isotopic carbon values of atmospheric CO 2 (δ 13 C atm ) and plant organic matter (δ 13 C plant ): The δ 13 C atm data could be obtained from online archives (http://www.esrl.noaa.gov/,accessed on 1 June 2023), while δ 13 C plant data referred to those measured in tree rings.
By combining Equations ( 5) and ( 7), we can obtain the value of ci as follows: Finally, replacing the Equation ( 8) in ( 6), we obtained the following equation: Once WUEi data were obtained, the relationships between the growth and water use of trees in DS and NS were evaluated by performing correlation tests with BAI.Furthermore, to assess the post-fire intrinsic water use efficiency responses of trees in DS and NS, we used WUEi data to calculate the resistance, resilience, and recovery indices [46].
The resistance index was calculated as the ratio between the WUEi in the year of the fire (WUEI disturbance) and the average of WUEi in the pre-fire years (WUEI pre-disturbance).
WUEi Resistance index = WUEi disturbance /WUEI pre-disturbance (10) Further, the resilience index was calculated by dividing the average WUEI value in the post-fire years (WUEi post-disturbance) by the WUEI pre-disturbance.
WUEi Resilience index = WUEi post-disturbance /WUEI pre-disturbance (11) Forests 2023, 14, 2033 6 of 17 Conversely, the recovery index was computed by dividing the values of each post-fire year by the WUEI disturbance.
WUEi Recovery index = WUEi post-disturbance /WUEI disturbance (12) Finally, Student's t-test was used to compare these indices between the two areas.All the statistical analyses were performed using R-studio [47].

Backdating of Structural Stand Data and Atmospheric CO 2 Sequestration
The tree ring width data were used to infer the past DBH of the examined trees.The DBH for each year was estimated by progressively subtracting the sum of the ring widths from the DBH measured at the end of 2022.The tree height chronology was estimated from the DBH using the Chapman Richards allometric equation [48].The total above ground biomass (AGB) of trees was then computed using the equations from the Italian Forest inventory, which are implemented in the ForIT R package [49]: Afterwards, to take into account the belowground biomass, its contribution was calculated considering a belowground-to-aboveground biomass ratio of 0.2 [50].Furthermore, the biomass was converted to its CO 2 equivalent stored in the trees by multiplying biomass by the Carbon Fraction Value (0.51) and the Carbon-CO 2 conversion factor (3.67), following the guidelines of the IPCC for assessing carbon stored in forest biomass [51]: CO 2 stored in the tree = AGB × (1 + 0.2) × 0.51 × 3.67 (14) This process allowed the computation of the yearly values of CO 2 fixed as tree biomass as the difference between the biomass at the end of the growing season and the biomass at the end of the previous growing season.CO 2 fixed t = CO 2 stored t − CO 2 stored t−1 (15) Finally, to assess the economic impacts of the fire on the carbon fixation capacity of Monte Cairo Forest, we calculated the economic value of the CO 2 fixed by the trees in the two areas for the pre-fire and post-fire years.The calculation was performed by multiplying the amount of stored CO 2 by the average spot price of CO 2 in the third quarter of 2022 (80 EUR/tCO 2 ), as reported in the Quarterly Report on European Electricity Markets released by the European Community [52].

Satellite-Based Computation Analysis
The calculation of the dNBR (difference Normalized Burn Ratio) from the satellite images allowed us to discriminate between the defoliated stands (DS) and the non-defoliated stands (NS) within the extensive Pinus nigra forest.The pixels of the map reported in Figure 2A were classified according to their dNBR values: all areas with dNBR values lower than 0.1 were identified as non-defoliated stands (NS, green pixels; total area, 69 ha).On the contrary, all areas with dNBR above this threshold were classified as defoliated stands (DS, orange pixels, total area 154 ha).While the dNBR allowed the assessing of the immediate impact of the fire on the forest, the processing of the Normalized Difference Vegetation Index (NDVI) enabled the analysis of vegetation dynamics over the past 10 years, allowing monitoring of the post-fire trends in the two areas.The time series of NDVI between the two areas (Figure 2B) had a similar trend over the years.In the years prior to the fire, while the NS curve was slightly higher than the DS one, they were not statistically different (p > 0.05).However, following the fire event in 2017, a statistically significant divergence in values between NS and DS was found.In fact, the post-fire values of the non-defoliated stands were significantly higher (p < 0.05) compared to the defoliated stands.
NDVI between the two areas (Figure 2B) had a similar trend over the years.In the years prior to the fire, while the NS curve was slightly higher than the DS one, they were not statistically different (p > 0.05).However, following the fire event in 2017, a statistically significant divergence in values between NS and DS was found.In fact, the post-fire values of the non-defoliated stands were significantly higher (p < 0.05) compared to the defoliated stands.

Forest Survey of the Experimental Plots
Identifying NS and DS enabled us to establish three experimental plots in each category for land-based analysis, all situated at the same altitude and distance from the forest perimeters.Dendrometric measurements conducted in the experimental plots allowed us to collect the structural information of the trees for both areas (Table 1).The two areas showed similar tree heights (in average 22.98 ± 2.48 m for NS and 18.30 ± 4.65 m for DS) and DBH (33.81 ± 5.79 cm for NS and 36.20 ± 8.62 cm for DS).The defoliation observed in the field confirmed the satellite classification: the trees of the NS showed no crown damage, while the trees of the DS had an average crown reduction of around 50%.Trees in both areas showed scorch along the trunk, indicating that the fire also affected the nondefoliated areas, albeit without reaching the tree canopy.Indeed, the mean scorch height

Forest Survey of the Experimental Plots
Identifying NS and DS enabled us to establish three experimental plots in each category for land-based analysis, all situated at the same altitude and distance from the forest perimeters.Dendrometric measurements conducted in the experimental plots allowed us to collect the structural information of the trees for both areas (Table 1).The two areas showed similar tree heights (in average 22.98 ± 2.48 m for NS and 18.30 ± 4.65 m for DS) and DBH (33.81 ± 5.79 cm for NS and 36.20 ± 8.62 cm for DS).The defoliation observed in the field confirmed the satellite classification: the trees of the NS showed no crown damage, while the trees of the DS had an average crown reduction of around 50%.Trees in both areas showed scorch along the trunk, indicating that the fire also affected the nondefoliated areas, albeit without reaching the tree canopy.Indeed, the mean scorch height was 2.73 + 1.94 m for NS and 4.65 + 2.59 m for DS.The post-fire tree density was higher in the NS experimental plots (392 ± 36) compared to the DS experimental plots (276 ± 120) due to the observed tree mortality rate in DS (31%), which caused its reduction.On the contrary, the pre-fire tree density between the two examined areas was more balanced (392 ± 36 for DS, 402 ± 113 for NS).The pre-fire density in DS was calculated by including standing dead trees, which were excluded from the post-fire calculation.

Dendrochronology Analysis
Dendrochronological analyses have highlighted how the Basal Area Increment (BAI) of the two examined areas was generally highly synchronous over the years (Figure 3A).However, while during the pre-fire period, the growth rates were perfectly overlapping, starting from 2018, a clear divergence in productivity is observed.While statistical tests have revealed that in the five pre-fire years (2013-2017), tree growth between the two areas was comparable (p > 0.05), in the five post-fire years (2018-2022), the growth rate was significantly higher (p < 0.05) in NS trees (Figure 3B).was 2.73 + 1.94 m for NS and 4.65 + 2.59 m for DS.The post-fire tree density was hig the NS experimental plots (392 ± 36) compared to the DS experimental plots (276 due to the observed tree mortality rate in DS (31%), which caused its reduction.O contrary, the pre-fire tree density between the two examined areas was more bal (392 ± 36 for DS, 402 ± 113 for NS).The pre-fire density in DS was calculated by incl standing dead trees, which were excluded from the post-fire calculation.

Dendrochronology Analysis
Dendrochronological analyses have highlighted how the Basal Area Increment of the two examined areas was generally highly synchronous over the years (Figur However, while during the pre-fire period, the growth rates were perfectly overlap starting from 2018, a clear divergence in productivity is observed.While statistica have revealed that in the five pre-fire years (2013-2017), tree growth between the two was comparable (p > 0.05), in the five post-fire years (2018-2022), the growth rate w nificantly higher (p < 0.05) in NS trees (Figure 3B).The correlation analyses between the pre-fire BAI (1970-2017) of DS and NS their respective meteorological data revealed the positive influence of spring temper on growth (Figure 4A-C).The temperatures of March (maximum, minimum, and for DS; maximum and mean for NS) of the current year showed positive correlation tree growth in both stands.
Furthermore, we identified negative correlations with the June temperatures f current year (maximum and mean for DS; maximum and minimum for NS).
Regarding precipitation, no correlations were found for DS, while a negative lation between growth and precipitation in December of the previous year was evid NS (Figure 4D).Finally, the seasonal correlation analysis yielded only a negative s cant correlation with precipitation of the coldest months (complete results are repor Supplementary Materials Figure S2).The correlation analyses between the pre-fire BAI (1970-2017) of DS and NS with their respective meteorological data revealed the positive influence of spring temperatures on growth (Figure 4A-C).The temperatures of March (maximum, minimum, and mean for DS; maximum and mean for NS) of the current year showed positive correlations with tree growth in both stands.
Furthermore, we identified negative correlations with the June temperatures for the current year (maximum and mean for DS; maximum and minimum for NS).
Regarding precipitation, no correlations were found for DS, while a negative correlation between growth and precipitation in December of the previous year was evident in NS (Figure 4D).Finally, the seasonal correlation analysis yielded only a negative significant correlation with precipitation of the coldest months (complete results are reported in Supplementary Materials Figure S2).

Growth-WUEi Relationship and Responses in Short-to-Medium Term
The relationship between WUEi and BAI (Figure 5A), tested over the last 10 years (2013-2022), highlighted a significant positive correlation (p < 0.05) for the trees in NS: an increase in growth was coupled with an enhancement in water use efficiency.Conversely, in trees of DS, no correlation was observed (p > 0.05).The indices of resistance, resilience, and recovery (Figure 5B) calculated using the WUEi data showed that while the resistance index revealed a similar response to the fire event in both NS and DS (p > 0.05), trees in NS showed a better response in terms of resilience and recovery in the years following the fire (p < 0.05).

Growth-WUEi Relationship and Responses in Short-to-Medium Term
The relationship between WUEi and BAI (Figure 5A), tested over the last 10 years (2013-2022), highlighted a significant positive correlation (p < 0.05) for the trees in NS: an increase in growth was coupled with an enhancement in water use efficiency.Conversely, in trees of DS, no correlation was observed (p > 0.05).The indices of resistance, resilience, and recovery (Figure 5B) calculated using the WUEi data showed that while the resistance index revealed a similar response to the fire event in both NS and DS (p > 0.05), trees in NS showed a better response in terms of resilience and recovery in the years following the fire (p < 0.05).

CO2 Sequestration and Ecosystem Service Costs
The estimation of CO2 sequestered by the trees in the NS and DS (Figure 6A) is in line with tree growth data.Although, over the years, the amount of CO2 absorbed by the two stands was not always comparable (as evident from 1980 to 1992), during the pre-fire years (2013-2017), the carbon fixation capacity of the trees was statistically the same (p > 0.05).Conversely, in the post-fire years (from 2018 to 2022), a clear divergence was evident (p < 0.05).Therefore, the economic value of the CO2 fixed in the five years before the fire was comparable between the two examined areas (ranging from 650 to 1150 EUR/ha).However, starting in 2018 and for all post-fire years, the lower carbon fixation capacity of trees in the DS resulted in a significant decrease in the CO2 fixed economic value (the values were always around/below 500 EUR/ha, Figure 6B).On the contrary, the trees in the NS had a stable carbon assimilation capacity in these years, retaining a higher service value (ranging from 500 to 1100 EUR/ha).

CO 2 Sequestration and Ecosystem Service Costs
The estimation of CO 2 sequestered by the trees in the NS and DS (Figure 6A) is in line with tree growth data.Although, over the years, the amount of CO 2 absorbed by the two stands was not always comparable (as evident from 1980 to 1992), during the pre-fire years (2013-2017), the carbon fixation capacity of the trees was statistically the same (p > 0.05).Conversely, in the post-fire years (from 2018 to 2022), a clear divergence was evident (p < 0.05).Therefore, the economic value of the CO 2 fixed in the five years before the fire was comparable between the two examined areas (ranging from 650 to 1150 EUR/ha).However, starting in 2018 and for all post-fire years, the lower carbon fixation capacity of trees in the DS resulted in a significant decrease in the CO 2 fixed economic value (the values were always around/below 500 EUR/ha, Figure 6B).On the contrary, the trees in the NS had a stable carbon assimilation capacity in these years, retaining a higher service value (ranging from 500 to 1100 EUR/ha).

Discussion
The combination of satellite-and ground-based analyses presented in this stud provided valuable information regarding the understanding of the eco-physiologica logical, and economic consequences of wildfires on forest ecosystems.Firstly, the dNBR (difference Normalized Burn Ratio) based on satellite imagery has been confi to be an effective tool for identifying and distinguishing forest areas affected by fire those remaining relatively intact.The analysis of pre-and post-fire satellite imag lowed precise discrimination between defoliated stands (DS) and non-defoliated s (NS) within a large forest of Pinus nigra.This provided a spatially detailed view extent of areas affected by the 2017 fire on Monte Cairo and facilitated the strategic ning of a targeted forest survey and ground-based sampling campaign.
The forest surveys on the ground confirmed the correct identification of the tw eas: the field measurements highlighted how the discrimination between the two ty areas reflected the crown damage, as well as the reduction in tree density determ based on post-fire mortality rates.While trees in NS did not show defoliation an mortality, in DS, we observed strong canopy damage (in average 50%), as well as a high tree mortality rate (in average 31%), which significantly reduced the tree dens the years following the fire.These findings are in agreement with previous studie assessed the effectiveness of the dNBR parameter through field surveys, demonst robust correlations with variations in crown structure and forest density [53][54][55].
Moreover, the results reported in our study highlighted the crucial importan combining remote sensing analyses with land-based forest survey campaigns.In

Discussion
The combination of satellite-and ground-based analyses presented in this study has provided valuable information regarding the understanding of the eco-physiological, ecological, and economic consequences of wildfires on forest ecosystems.Firstly, the use of dNBR (difference Normalized Burn Ratio) based on satellite imagery has been confirmed to be an effective tool for identifying and distinguishing forest areas affected by fire from those remaining relatively intact.The analysis of pre-and post-fire satellite images allowed precise discrimination between defoliated stands (DS) and non-defoliated stands (NS) within a large forest of Pinus nigra.This provided a spatially detailed view of the extent of areas affected by the 2017 fire on Monte Cairo and facilitated the strategic planning of a targeted forest survey and ground-based sampling campaign.
The forest surveys on the ground confirmed the correct identification of the two areas: the field measurements highlighted how the discrimination between the two types of areas reflected the crown damage, as well as the reduction in tree density determined based on post-fire mortality rates.While trees in NS did not show defoliation and tree mortality, in DS, we observed strong canopy damage (in average 50%), as well as a rather high tree mortality rate (in average 31%), which significantly reduced the tree density in the years following the fire.These findings are in agreement with previous studies that assessed the effectiveness of the dNBR parameter through field surveys, demonstrating robust correlations with variations in crown structure and forest density [53][54][55].
Moreover, the results reported in our study highlighted the crucial importance of combining remote sensing analyses with land-based forest survey campaigns.In fact, although the areas have been correctly identified, the field surveys showed that fire also affected the trees in the NS.The scorch found along the trunk of trees confirmed that NS has been affected by a ground fire, which, however, was not able to damage the crown.As confirmed using the NDVI data, this information was not captured via satellite image analyses in the months and years following the fire, as the low fire severity in the NS did not affect tree density or canopy structure.Indeed, while we observed a significant decline in NDVI, starting in 2017 for the DS, NDVI values were constant in the NS between the years pre-and post-fire.
Further insights have been provided via dendrochronological investigations: while in the years preceding the fire, tree productivity was comparable between the two areas, after the fire event, we observed a clear divergence in growth rates.The strong reduction in growth observed in defoliated trees represents a phenomenon widely reported and investigated in the scientific literature [5,7,11,19,[56][57][58].This decline in productivity has mainly been attributed to a marked reduction in trees' photosynthetic capacity, which triggers a mechanism known as carbon starvation [15,16].Severe defoliation results in the lower capacity of trees to uptake atmospheric carbon and produce the necessary carbohydrates to meet the physiological and metabolic demands for their growth and maintenance [59][60][61].These mechanisms make trees particularly sensitive to climatic stresses, amplifying their susceptibility to events such as heatwaves and prolonged drought periods, affecting their health and growth [62][63][64].
The correlations between basal area increment and meteorological data highlighted that although the growth of Pinus nigra on Mount Cairo was positively influenced by spring temperatures over the years, the trees were often subjected to intense heatwaves during the summer, as indicated by the negative correlations between growth and temperature observed for the month of June in both DS and NS.
The lack of positive correlation between precipitation and growth, on a monthly and seasonal scale, could be due to the low water retention capacity of the soil, as the forest is located on a steep slope [65].Furthermore, most of the precipitation occurs during the coldest months, and, thus, its effect on growth is dampened [66].
This supports the hypothesis that the combination of reduced photosynthetic capacity and increased sensitivity to climatic stress has led to a reduction in the growth of trees in DS.In contrast, non-defoliated trees that suffered no photosynthetic damage were able to withstand summer heatwaves.Furthermore, the ground fire might have benefited trees in DS by suppressing undergrowth species, thereby reducing competition for resources [7,16,58].
The WUEi data confirm the previously discussed points.Given that WUEi is an index of the amount of carbon fixed by the plant per unit of water lost through transpiration [67], the strong positive correlation observed between the growth and WUEi of trees in the NS suggests the better capacity of these trees to maintain unchanged photosynthetic rates over the years using a limited amount of water [68].This reflects a typical strategy of healthy isohydric species growing in drought prone areas such as Monte Cairo, which allows them to grow resiliently, despite climatic stresses, by overcoming periods of limited water availability thanks to their dynamic stomatal control [61,69].The absence of a correlation between growth and WUEi in defoliated trees confirmed their inability to maintain predisturbance photosynthetic rates due to severe canopy injuries [64].The calculation of tree resistance, resilience, and recovery indices, based on WUEi data, has allowed the establishment of a direct link between tree water use efficiency and their capacity to respond to wildfires in the short and medium term.The resistance index, which assesses the immediate impact of the fire, did not reveal significant differences between trees in NS and DS.Being the same species, fire resistance mechanisms are analogous [70], and this could have triggered the same effects of fire on WUEi in the short term.However, the resilience and recovery indices, which indicate the ability of trees to reach their predisturbance conditions, are significantly higher in NS compared to DS.This result suggests a negative effect on the trees' capacity to optimize water use in defoliated trees in the years after the fire [71].This phenomenon could be attributed to a strategy of greater stomatal opening to counteract the carbon deficit due to strong defoliation suffered by burned trees [15].Under normal conditions, stomata play a crucial role in regulating the exchange in CO 2 and water between the plant and its surroundings [72,73].However, the considerable defoliation experienced by these trees can disrupt this usual pattern: the necessity to assimilate carbon via the reduced canopy may have pushed the trees to increase stomata opening in order to prioritize CO 2 uptake, even at the expense of increased water loss through transpiration [74].Nevertheless, defoliated trees allocated a limited amount of carbon to growth, suggesting that the resources were prioritized for maintaining metabolic functions [75,76].Although this strategy ensured their survival, it led to a reduction in post-fire growth and water use efficiency [77,78].
The quantification of the CO 2 sequestered by the trees in the DS and NS, in addition to corroborating our previous interpretations, highlights the ecological and economic impacts of the fire.Although, over the years, the amount of carbon sequestered was not always comparable (such as from 1980 to 1992), probably due events beyond the scope of this study, in the five years preceding the fire event, the CO 2 sequestered between the two sites was statistically similar.However, after the fire, there was a marked decline in defoliated trees absorbing atmospheric carbon.
Forests, under normal conditions, act as important carbon sinks through the process of photosynthesis [79,80].However, due to fire damage, their ability to perform this crucial role has been significantly impaired in DS.In fact, it was observed that their CO 2 uptake decreased dramatically to approximately half of the levels observed in undamaged trees.A reduced capacity for photosynthesis and CO 2 sequestration implies that a greater amount of carbon dioxide will remain in the atmosphere, contributing to the accumulation of greenhouse gases, which are known to contribute to global warming [81].The forest carbon sequestration capacity can be quantified through an economic value [82,83].In the specific case, the ecosystem service of carbon sink of DS-which represent a substantial portion of the total extension of Pinus nigra forest on Monte Cairo-has been halved.Therefore, the inability of damaged trees to fully recover their physiological and photosynthetic functions over time, as highlighted in this study, could lead to a gradual reduction in or complete loss of the economic value of the carbon sequestration service.This scenario underscores the importance of adopting responsible and targeted forest management practices to promote the recovery of ecosystems impacted by fires and preserve the long-term economic value of carbon sequestration services.

Conclusions
Our insights showed how the dNBR parameter for the initial discrimination of areas affected by fire from those remaining relatively intact is a fundamental tool for assessing fire impacts and planning forest sampling and survey campaigns.Similarly, the NDVI index proved highly reliable in monitoring the evolution of a burned forest.However, our study highlights the crucial role of integrating remote sensing with land surveys for a more precise understanding of tree fire impacts over time.
Dendrochronological and stable isotope analyses shed light on the eco-physiological dynamics of trees in the years following the fire, revealing that defoliated trees reduced their growth and modified their water use due to severe canopy injury.
Finally, the quantification of CO 2 sequestration highlighted the fire's ecological impact.Fire-induced damage has substantially reduced the capacity of defoliated trees to sequester carbon from the atmosphere, reducing their contribution to climate change mitigation.The economic implications of declining carbon stocks have underscored the urgency of employing responsible forest management to mitigate the long-term losses of the carbon sequestration service.
In conclusion, the presented results, albeit based on a single case study, represent a clear example of how the integration of satellite-and ground-based analyses can provide a complete view of the consequences of fires on forest ecosystems, allowing us to achieve several goals aimed at the recovery and restoration of fire-affected areas.

Figure 1 .
Figure 1.(A) Location of Monte Cairo (red dot) within the Lazio region (in blue), in the center of the Mediterranean basin.(B) Forest vegetation map of Monte Cairo: the Pinus nigra pine forest under study is delimited in red.

Figure 1 .
Figure 1.(A) Location of Monte Cairo (red dot) within the Lazio region (in blue), in the center of the Mediterranean basin.(B) Forest vegetation map of Monte Cairo: the Pinus nigra pine forest under study is delimited in red.

Figure 2 .
Figure 2. (A) Map of the Pinus Nigra Forest of Monte Cairo showing the identification of defoliated stands (DS, orange pixels) and non-defoliated stands (NS, green pixels).The black and white dots indicate the position of the experimental plots in DS and NS, respectively.(B) Mean NDVI time series for DS (in orange) and NS (in green) from 2013 to 2022.The lowercase letters indicate statistical differences between the two areas in the pre-and post-fire periods.The bands indicate the relative standard errors.

Figure 2 .
Figure 2. (A) Map of the Pinus Nigra Forest of Monte Cairo showing the identification of defoliated stands (DS, orange pixels) and non-defoliated stands (NS, green pixels).The black and white dots indicate the position of the experimental plots in DS and NS, respectively.(B) Mean NDVI time series for DS (in orange) and NS (in green) from 2013 to 2022.The lowercase letters indicate statistical differences between the two areas in the pre-and post-fire periods.The bands indicate the relative standard errors.

Figure 3 .
Figure 3. (A) Mean BAI trend recorded in the trees of NS (in green) and DS (in orange).The indicate the relative standard error.(B) Comparison between the BAI recorded in the five yea fire (2013-2017) and the five years post-fire (2018-2022) of NS (in green) and DS (in orang lowercase letters indicate statistical differences between the two areas in the pre-and post-fir ods.

Figure 3 .
Figure 3. (A) Mean BAI trend recorded in the trees of NS (in green) and DS (in orange).The bands indicate the relative standard error.(B) Comparison between the BAI recorded in the five years pre-fire (2013-2017) and the five years post-fire (2018-2022) of NS (in green) and DS (in orange).The lowercase letters indicate statistical differences between the two areas in the pre-and post-fire periods.

Figure 4 .
Figure 4. Correlation tests between the pre-fire BAI data (1970-2017) of DS (in orange) and NS (in green) with the relative (A) maximum, (B) minimum, (C) mean temperatures, and (D) precipitation obtained from ERA5 database.Red bars indicate significant correlations (p < 0.05), while gray bars indicate no correlations (p > 0.05).

Figure 4 .
Figure 4. Correlation tests between the pre-fire BAI data (1970-2017) of DS (in orange) and NS (in green) with the relative (A) maximum, (B) minimum, (C) mean temperatures, and (D) precipitation obtained from ERA5 database.Red bars indicate significant correlations (p < 0.05), while gray bars indicate no correlations (p > 0.05).

Forests 2023 , 17 Figure 5 .
Figure 5. (A) Correlation tests between WUEi and BAI values; (B) resistance, resilience, and recovery indices calculated for trees in DS (in orange) and NS (in green).Reference period: 2013-2022.The lowercase letters indicate statistical differences between the two areas.

Figure 5 .
Figure 5. (A) Correlation tests between WUEi and BAI values; (B) resistance, resilience, and recovery indices calculated for trees in DS (in orange) and NS (in green).Reference period: 2013-2022.The lowercase letters indicate statistical differences between the two areas.

Figure 6 .
Figure 6.(A) Mean trend of the tons of CO2 per hectare fixed by the trees in NS (in green) a (in orange).The bands indicate the standard error.(B) Economic value, expressed in Euros, CO2 per hectare fixed in the five years pre-and post-fire by the trees in NS (green bars) a (orange bars).

Figure 6 .
Figure 6.(A) Mean trend of the tons of CO 2 per hectare fixed by the trees in NS (in green) and DS (in orange).The bands indicate the standard error.(B) Economic value, expressed in Euros, of the CO 2 per hectare fixed in the five years pre-and post-fire by the trees in NS (green bars) and DS (orange bars).

Table 1 .
Summary of the results of the field survey.

Table 1 .
Summary of the results of the field survey.