Evaluating the post-fire natural regeneration of Mediterranean-type scrublands in Central Spain

. We performed a five-year assessment of the natural vegetation restoration capacity following the 2012 fires in Valdemaqueda (Madrid, Spain) via the characterization of the post-fire and residual vegetation and the analysis of soil physico-chemical characteristics. Six pilot-plots were established in the affected site. Forest species, representative of the potential natural vegetation of the area ( Juniperus oxycedrus subsp. lagunae and Quercus rotundifolia woodlands) and broom shrubs ( Cytisus scoparius, Retama sphaerocarpa ) were planted to assess the relationship among the stages of ecological succession, competition, and soil restoration processes following devastating fire events. The fire-driven alteration of the soil’s physico-chemical properties was evident, given the increased pH and reduced C/N ratio in the first years of the study. However, we observed an increased soil enrichment in the last years of study, accompanied by the propagation of herbaceous species, supporting our seed bank findings, showing a clear difference in the sprouting rate between burnt and control plots (80% vs. 20%, respectively). The establishment of robust, pyrophyte shrub species ( Cistus ladanifer, C. laurifolius, Rosmarinus officinalis ) rather than natural succession evidenced the clear conversion of the vegetation in burnt areas. These findings in the pilot-plots allowed evidencing the high vulnerability of the natural vegetation to the settling of pyrophytes, given their low survival rate under the strong competitive pressure of these pyrophytic species. The proliferation of these pyrophytes could translate into changes in soil macro- and microbiota, nutrient dynamics, species diversity, and interaction, added to the alteration of fire regimes in the area. Overall, these results highlight the risk for soil impoverishment and possible erosion of the fire-affected sites. Moreover, they underline the importance of the establishment and regeneration of Genisteae species to outcompete pioneer pyrophytic species, favoring the restoration of the area’s potential natural vegetation.


Introduction
In arid and semi-arid areas worldwide, regarded as hotspots of biodiversity, severe vegetation and soil losses are estimated given the significant vulnerability of these habitats to the climate crisis and desertification (Diffenbaugh et al., 2007;Salvati et al., 2013). In particular, these effects of climate aggravation are exacerbated in the Mediterranean basin, given the predicted increase in temperatures and rainfall scarcity by the end of this century, deriving in prolonged drought periods during summertime (Giorgi & Lionello, 2008). The combination of higher temperatures, elevated solar radiation, and evaporation, together with the low relative humidity and rainfall in Mediterranean-climate zones, could promote an increased susceptibility of desertification in these areas. Furthermore, these adverse climatic conditions could foster fire regime changes, resulting in an increased frequency of more severe fires added to the lengthening of the fire season (Amatulli et al., 2013;Pérez-Fernández et al., 2016).
Nowadays, fire-prone regions (e.g., Australia, California (USA), Portugal, Spain) face a substantially increased incidence of destructive and wide-ranging wildfires, threatening both biodiversity and human lives. The impact, the level of damage, and the magnitude of these fires have been critically influenced by anthropogenic effects on climate (i.e., climate crisis) and land use (Moreira et al., 2020). The latter being of particular importance since it directly translates into the alteration of fire regimes as a consequence of an enhanced buildup of fuel loads, caused by the conversion of natural areas for logging purposes or the expansion of the urban-wildland interface (Kraaij et al., 2018;Moreira et al., 2020;San Miguel-Ayanz et al., 2013). Furthermore, this deliberate conversion of natural vegetation tracts for logging purposes can increase post-fire physical damage and soil erosion, further restricting the restoring capacity of natural environments (Kraaij et al., 2018).
Fires of low to moderate severity, as most of those prescribed in forest management are, stimulate the renovation of the dominant vegetation via the removal of undesired species, transient increase of soil pH, and available nutrients (Certini, 2005;Hinojosa et al., 2016). However, severe fires, such as wildfires, generally have numerous adverse effects on the soil properties such as the significant removal of organic matter, the deterioration of both structure and porosity, a considerable loss of nutrients through volatilization, ash entrapment in smoke columns, leaching and erosion, and a marked alteration of both quantity and composition of microbial and soildwelling invertebrate communities (Certini, 2005). Furthermore, post-fire soil physical damage correlates with the degree of invasion of alien species, thus having a higher disturbance rate in uninvaded or lightly invaded areas than on heavily invaded sites (Kraaij et al., 2018;van Wilgen & Scott, 2001). Additionally, the environmental conditions in the affected areas have a significant role in vegetation restoration capacity. The latter is especially crucial in Mediterranean-type climate areas where the disadvantageous conditions (i.e., low rainfall, high temperatures) can significantly challenge the vegetation recovery processes (Pérez-Fernández et al., 2016;Vallejo et al., 2012).
Short-term responses of soil characteristics and vegetation to fire incidence are well-studied in most biomes across the globe. However, little information is available on the post-fire effects in Mediterranean scrublands from central Spain, despite the higher incidence of wildfires in the latest years and the expected increase in fire severity and frequency outcome of the estimated reduction in rainfall and prolonged annual drought periods, as a consequence of the climate crisis. In turn, this research aims to provide a preliminary overview of early soil and vegetation responses of a fire-affected scrubland charac-teristic of Central Spain, including post-fire ecological successions and restoration processes of these habitats.

Study site
Our study was set in the territory affected by the forest fires of 2004, 2012, and 2013 in the municipality of Valdemaqueda, Madrid. The affected region is delimited by cultivated areas of Pinus pinaster s.l., for resin extraction. The area's potential vegetation comprises holm oaks (Quercus rotundifolia) with cade junipers (Juniperus oxycedrus subsp. lagunae), still present in intact natural areas neighboring to those destined for pine-resin extraction. It is worth mentioning the immediate post-fire measurements taken (outsourced, independent from the study) consisted of the extraction of burnt and deadwood present in the affected area, using heavy equipment on repeated occasions, imperiling the soil to erosion processes. The affected area is located along the sloping terrain, with different exposures and aspects (Table 1; Table  S2). In turn, six permanent pilot-plots were established through the affected site based on fire incidence, altitude, and vegetation type. We set four permanent plots (25 m 2 ) on burnt sites at different elevations, encompassing the entire macroclimatic spectrum and the main types of natural vegetation found in the affected areas. Two control plots were also established in unburnt areas adjacent to the burn sectors, following the protocol mentioned above (Table 1).

Soil sampling and analyses of physicochemical characteristics
We took five soil cores from the upper 10 cm of the soil profile, taken at the four corners of each pilot-plot (20 cores per plot), and merged them into one sampling bag per plot. At the laboratory, samples were air-dried at room temperature for two weeks and subsequently sieved. Once all soil samples were dried out, we proceeded to perform all the analyses of physicochemical characteristics (Anon., 2006;Soil Survey staff, 1999).

Particle size and classification of soils
The previously dried soil samples were used to assess the granulometric parameters. The following soil fractions were estimated and expressed in percentage: fine soil fraction for total natural soil (<2 mm), fine gravel (2-20 mm), and gravels (>20 mm). We determined the proportion of the various size fractions for the mineral part of the soil and expressed them in percentages. The sand (50-2000 µm), silt (20-50 µm), and clay (<2 µm) fractions of each soil sample were separated and determined following the protocol by van Reeuwijk (2002). The textures in the sand fraction (only coarse sand and fine sand) were further identified following the limits established in the Soil Survey Staff (1999). The proportions of sand, clay, and slit of the soil samples were plotted on the International Soil Society texture triangle to obtain the soil sample texture (Table S1). The pH was also measured in H2O, with a suspension of soil-DIO water (1:2.5 by volume), using a glass electrode.

Micro-element analysis of soil samples
Total soil organic carbon (C Soil , %), total soil nitrogen (N Soil , %), soil organic matter content (SOM, %) of the fine earth fraction, and sulfur (S, %) content were determined using the "Flash Combustion Method" using a FLASH 2000 analyzer (Thermo Scientific).

Determination of the Cation-exchange capacity of soil samples
The cation-exchange capacity (CEC) of soil samples was used as an indicator of soil fertility. The measurements of the potassium (K) and sodium (Na) adsorbed to colloid (cmol/kg soil) were measured following the protocol by Burt (2004). While calcium (Ca) and magnesium (Mg) were measured using a gas chromatographer with a FID detector (Agilent Technologies, Santa Clara, CA, USA).

Evaluation of the aboveground vegetation
The characterization of the vegetation types and the estimation of fire's influence on potential changes in the taxonomic diversity of the study site were assessed in the surrounding perimeter and inside each pilotplot. Visual estimation of cover and identification of the species encompassed in these plant communities was performed following the established by Mucina and van der Maarel (1989) in 20 subplots (100x100 cm) inside and in the surroundings of the pilotplots during the spring-summer periods 2014-2018.
Throughout the study, we assessed the survival and growth rates of woody individuals planted inside the pilot-plots in spring 2014 (mainly shrub species representative of the potential vegetation). The growth of these planted individuals was assessed from autumn 2014 and on. We also registered the establishment of new individuals growing in these plots (recruitment) and the new species colonizing within the pilot-plots, the outcome of the natural regeneration of the area. The growth and survival rate of these recruited individuals were registered as they established in the pilot-plots (i.e., after 2014).
The taxonomy usually follows the proposals of Euro+Med PlantBase (2006), except for the cade juniper on the territory, Juniperus oxycedrus subsp. lagunae (Pau) Rivas Mart. Higher syntaxonomical units are accepted from Mucina et al. (2016), while species pools are structured following the The study area is circled in red.

Soil seed bank
Soil samples were taken to assess the fires' effects on the temporal seed bank and, consequently, obtain a good indication of the regeneration of vegetation of the fire-affected area. Thirty soil samples were taken from the topsoil (5 cm), since this is the area where most seeds accumulate in the soil (Vilà & Gimeno, 2007), in the perimeter outside each pilot-plots (control and treated) in 2015 and 2017 using a metallic core (diam: 5 cm, height: 5 cm). After collection, all samples were kept at 4 ℃ for two months. Afterward, all individual soil samples were sieved through a 0.5 cm mesh to retain the coarse fraction, with one additional sieving through a 0.2 cm mesh to reduce the volume of fine material and favor the scarification of seeds to promote germination (Thompson et al., 1997).
Next, each soil sample was placed in a small tray (14x11x5 cm, one core sample per tray) and thoroughly mixed with a sterile substrate of vermiculite and peat (1:1:0.5) to improve the water-retention capacity of the soil. Afterward, all trays were transferred to the greenhouse to monitor seedling emergence. Greenhouse conditions were set to maintain a mean 25 ℃ day/night temperature, with individual tray irrigation every three days. Seedlings were identified as soon as they emerged in the trays and subsequently removed. When the identification was not feasible at the species level at this early stage, the seedlings were kept in their respective trays until they attained a sufficient development that allowed their proper identification.
When growing under favorable conditions, seed germination should occur in its majority within two months (Graber & Thompson, 1978;Thompson et al., 1997). In our case, we maintained and followed-up the seed bank for three months in both years to allow the germination and identification of all potentially viable seeds present in the soil samples. After the third month, we added a solution of gibberellic acid (1000 ppm) to stimulate seed emergence in those with endogenous dormancy (Hartman & Kester, 1991). Emergence was monitored for another month in the gibberellic acid-treated trays.

Effects of fire on soil physico-chemical characteristics
Minor differences in the soil pH were observed between 2015 and 2017 (Table 2), maintaining a moderately acidic soil in the measurements closer to the fire events. Nevertheless, in the latest samples (2017), a slight increase in the pH was observed in most of the burned sites, and the increase in the lime fraction and reduction of the clay fraction in all sites (Table S1).
The CEC analyses from burned plots (i.e., Sta. Cat B , Q B , CC B ) show a high content of alkaline cations Ca 2+ and Mg 2+ in years closer to the fire events, contrasting their lower values in the 2018 samples (Table 3). In each plot, minor variations were observed in the CEC throughout the study period while showing considerable differences among sites, independent of fire incidence. Furthermore, the C/N ratio decreased between 2015 and 2017 in all plots ( Table 2), regardless of fire incidence. However, we observed a broader range on the C/N decrease (58.07-72.75%) in burned plots than in the unburnt sites (64.94-63.39%). The slope could have influenced the latter since the burned plots are located on steeper slopes than those unaffected by the fire.

Soil seed bank
Results from the 2015 soil seed bank reported the germination of 48 species from 20 different families (Table 4). In this round, the highest-incidence families were the Asteraceae (10 species), Fabaceae (6 spe- Table 2. Comparison of the general characteristics of topsoil (0-10 cm) in pilot-plots between the year after the last fire event (2015) and the three following years (2017). Values of the total soil carbon content (C), nitrogen (N), sulfur (S), carbon/nitrogen ratio (C/N), and soil organic matter content (SOM) are given in percentage. EC, Electric Conductivity.
Year sample  pH, (H 2 O) 5.7 5.7 5.8 5.9 6.3 6.0 6.8 6.9 5.9 6.1 6.0 6.7 cies), and Caryophyllaceae (6 species). In contrast, we observed a substantial increase in species richness in the 2017 soil seed bank, with a highly similar number of families (22). Nonetheless, we included two more families among those with the highest incidence of germination in our seed bank, with the most abundant being the following: Caryophyllaceae (6 species), Scrophulariaceae (7 species), Fabaceae (8 species), Poaceae (7 species) y Asteraceae (6 species). Compared to prior seed bank results (i.e., 2015), we observed substantial differences in the germination rate between treatments in the 2017 samples. Our results indicated that 80% of the total sprouted seedlings corresponded to samples from burned plots, while only 20% of the seedlings emerged in control plots (not burned).

Post-fire stages of ecological succession, competition, and soil restoration processes
During four years (2014-2018), we assessed the rates of persistence, recruitment, and growth rates of the four main shrub species present within the pilot-plots to determine the fire effects on post-fire ecological successions and restoration processes. Persistence and growth rates were recorded for the individuals of each species planted since the establishment of the pilotplots (i.e., 2014), while persistence, recruitment, and growth were recorded for species and new individuals establishing inside the pilot-plots after 2014. Our results revealed a substantial and steady increase in plant density and height for Cistus ladanifer at all elevations (Figure 2, Figure S2a), with the highest individuals found at the lowest elevationss (pilot-plots CC B , A B ). Similarly, Cytisus scoparius and Rosmarinus officinalis showed a steady increase in plant height (Figures S1a, S1b, respectively) at mid-and low-altitude plots.
Compared to unburnt plots, those affected by wildfire registered a higher density of individuals was registered in plots affected by wildfire (Sta. Cat. B , Q B , CC B , A B ), particularly in those located at mid-and low-elevations ( Figure 2). Densities of pioneer pyrophyte species, mainly Cistus ladanifer (Figure 2d), were considerably higher than the density of the potential natural vegetation and Genisteae (Cytisus scoparius, Retama sphaerocarpa; Figure 2a, b) in burned sites. Conversely, unburnt plots (Sta. Cat. NB , Er NB ) showed minor variations in the density of individuals. Furthermore, we observed a greater density of individuals and species diversity in the mid-and low-elevation plots burned in recent years (i.e., Q B , CC B ), compared to those that suffered the incidence of wildfires in earlier years (Sta. Cat. B ).
A common occurrence of plants from Mediterranean  Table  9), together with several perennial species distinctive of dry grasslands (Stipo giganteae-Agrostietea castellanae) and, similar to the observed at mid-and low-altitudes, a vast set of annuals from Helianthemetea guttati.

Discussion
In fire-affected areas, recovery of SOM initiates with the natural or artificial re-establishment of vegetation, a commonly fast process owing to the high net primary productivity of secondary ecological successions (Certini, 2005). Nevertheless, despite the positive, shortterm effects of leaf matter burning (e.g., higher soil N available for plants, nitrate availability), its long-term effects have a more pondering effect. The latter fosters an enhanced direct exposure of the topsoil to environmental factors, making it more prone to erosion. Moreover, the burning of the vegetal cover and roots prompts the loss of mechanical support in the soil, leading to the decapitation of the soil profile and, ultimately, resulting in the loss of the soil's ecological utility, hindering the natural recovery of the burned area. Rapid plant recolonization is crucial for safeguarding soil N in fire-affected areas (Weston & Atwill, 1996). Expressly, the recovery of the organic N pool can occur at a relatively fast rate if N-fixer species are within the recolonizing flora, a regular instance in burnt areas (Johnson & Curtis, 2001;Certini, 2005). The increased total soil N found in all burned sites allows inferring a positive effect of establishing N-fixing species, Cytisus scoparius and Retama sphaerocarpa (Adams & Atwill, 1984;Spehn et al., 2002;Temperton et al., 2007;Fornara & Tilman, 2009). The lower N TOTAL measured in years closer to the burning (i.e., 2015) compared to the higher values in later years (i.e., 2017) implies the increase in N TOTAL is not related to the enhanced release of N from burning residues (Alcañiz et al., 2018;Pellegrini et al., 2018). Thus, the observed increase in N TOTAL could suggest the N-fixers' potential contribution in our pilot-plots to the soil N pools (Johnson & Curtis, 2001;Fornara & Tilman, 2009). Moreover, the minor changes in the soil C of burned pilot-plots (Table 2) could be suggested as an additional prospective benefit of establishing N-fixing species since their enhanced N input has been linked to facilitating C sequestration (Johnson & Curtis, 2001). Nevertheless, further studies assessing the organic N fraction are still required. Fire intensity is strongly associated with the alterations in the amount and availability of soil nutrients due to the severe influence of fire on the biological, chemical, and physical properties of soils (James et al., 2018). Resembling the findings by Tomkins et al. (1991) and Santín et al. (2018), our results show a high availability of nutrients in years closer to the fire events (i.e., 2015; Table 3) than in later years, as an outcome of the combustion of residues and organic matter (Certini, 2005). The similar Ca 2+ , Mg 2+ , K + , and Na + levels in control and more-recently burned plots (i.e., Sta. Cat B , Q B , CC B ) allows suggesting a persistent fire-induced availability of these nutrients up to 3 years after fire (Simard et al., 2001;Certini 2005). However, this persistence effect shows to decline >3 years after the fire, as seen in plots burned more than ten years ago (i.e., A B ), possibly due to processes such as leaching or soil erosion. In line with the latter, the increase in the soil pH could be related to the increase in SOM observed in 2017 samples, enhancing the buffering capacity of the soils in the affected area, which is linked to the CEC.
The recovery of the natural balance is associated with species diversity. The latter relating to the observed increase in species diversity in the latest seed bank (i.e., 2017; Table 4, Table S1) compared to the lower species richness found in samples from the year following the last fire event (i.e., 2015). The high species diversity observed in plots burned in recent years (Table 4) could be related to the reduced soil acidity, higher SOM, and lower C/N (Table 3). In recent years, the increase in N TOTAL could have facilitated the abundant sprouting of herbaceous species, with inclinations for nitrophylous taxa. Additionally, vegetation plots recorded in 2017 indicate a higher incidence of plants from Mediterranean ephemeral grasslands than the rate of purely ruderal, indicating the preference of the former type of plants for such nutrients. The observed increase in species richness might be associated with the observed soil enrichment, and the decrease in soil acidity, the outcome of the spread of N-fixing species in the study plots (Fornara & Tilman, 2009). In Mediterranean-type bioclimates, the vegetation typeconversion is mostly shaped by combining two plant stress factors: fire regimes and the summer drought period characteristic of these habitats (Jacobsen & Pratt, 2018). Resembling the findings by Dickens and Allen (2014), our results evidence the harmful effects of forest fires on maintaining the natural vegetation-types. The latter is supported by the enhanced establishment and rapid growth of pyrophytic pioneer species favored by the fire-removal of the vegetation cover and not due to the natural succession stages of vegetation (Supplementary Figure 2). Additionally, the enhanced species richness found in the seed bank samples from burned plots, compared to those untouched by fire, suggests fire and summer drought as stress factors that act as strong drivers of species richness in this area (Jacobsen & Pratt, 2018). Furthermore, studies in chaparral habitats have demonstrated that the vegetation-types' conversion always leads to functional-type conversions (Jacobsen & Pratt, 2018). The observed vegetation-type conversion could lead to changes in ecosystem services, such as alterations in nutrient-cycles, species interactions, and even influence the fire regimes (Schoennagel et al., 2004;Dickens & Allen, 2014). For instance, our results from the soil seed bank confirmed the observations from the vegetation plots, indicating that, albeit being adjacent zones, the plots located in the "Camino de los Corrales" area (CC B ) denoted a slightly higher species richness than that of the "Atalaya" area (AB). The latter could be attributed to the "Atalaya" ranch establishment, which altered the zone for intensive hunting purposes, thus, fostering a vegetation-poor area in the forest layer before it burned. Consequently, the "Atalaya" site required a longer time to be colonized by the diaspora of surrounding areas, e.g., CC B , to regain species diversity. These alterations in species richness and vegetation-type conversions can be further associated with the enhanced vulnerability of Mediterranean-type ecosystems to the effects of climate change.
The forecasted lengthening of the drought period and increase in temperature regimes could escalate the risk of fires, particularly in areas with higher fuel loads (San . Additional fuel resulting from the invasion of pyrophyte species can increase the intensity and severity of fires in the area, intensifying the extent of fire destruction and the difficulty of controlling them (Keeley, 2009;San Miguel-Ayanz et al., 2013). The latter is particularly threatening for areas where non-endemic species have been introduced for primary activity purposes. Similar to the occurred in the study area, the substitution of natural Fynbos shrublands in South Africa by plantations of Pinus pinaster s.l. and Eucalyptus globulus enhanced the fuel loads and, consequently, the fire risk and intensity of wildfires (Kraaij et al., 2018).

Final remarks and future perspectives
• The introduction of species with forestry exploitation purposes produces changes in the composition of the natural species. This leads to a higher fire-prone biomass content, which heightens these areas' vulnerability to forest fires.
• Identification and establishment of fire-risk zones are indispensable, combined with creating efficient fire-prevention policies that incorporate the paradigm shift proposed by Moreira et al. (2020): rating policy effectiveness as a function of prevented socio-ecological damage and loss rather than as a function of area burned.
• Use the present study results regarding the alteration of edaphic properties to create a suitable strategy and protocols to restore the fire-affected scrublands in Central Spain.
• Highlight the importance of studying the post-fire evolution of the soil microbial communities, given their fundamental role in biological processes, which directly influences soil quality. The latter being particularly vital in areas significantly affected by the fire since improving soil quality will allow the rooting and, subsequently, the restoration of the aboveground vegetal cover. soil analyses and the Alfonso XIII Botanical Garden (UCM) for providing the greenhouse infrastructure and facilities. Additional thanks to GEDESMA for their aid in the field sampling campaigns. Special thanks to the forest rangers from the Government of Madrid Autonomous Community, specially Mónica Gómez, José Araujo, Miguel Higueras, Abel López, Conrado de Mirande, and Xurxo Xoan, for their vital assistance with the field work. Thanks are also due to María Nuñez Munáiz for her enthusiasm, inestimable help, and company during the sampling campaigns. Additional thanks to Bernardo de Haro and Rodrigo Alonso for their contributions in the early stages of the project.    Rumex sp.