A vanishing imprint? Modelling the present and future distribution of the enigmatic Quercus crenata Lam., a Mediterranean sporadic tree species

A vanishing imprint? Modelling the present and future distribution of the enigmatic Quercus crenata Lam., a Mediterranean sporadic tree species | Authorea try { document.documentElement.classList.add('js'); } catch (e) { } var _gaq = _gaq || []; _gaq.push(['_setAccount', 'G-8VDV14Y67G']); _gaq.push(['_trackPageview']); (function() { var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true; ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js'; var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s); })(); Skip to main content Preprints Collections Wiley Open Research IET Open Research Ecological Society of Japan All Collections About About Authorea FAQs Contact Us Quick Search anywhere Search for preprint articles, keywords, etc. Search Search ADVANCED SEARCH SCROLL Ecology and Evolution This is a preprint and has not been peer reviewed. Data may be preliminary. 1 March 2025 V1 Latest version Share on A vanishing imprint? Modelling the present and future distribution of the enigmatic Quercus crenata Lam., a Mediterranean sporadic tree species Authors : Giuseppe Antonelli , Giuseppe Puddu [email protected] , Chiara Cipollini , Gianluca Sabatini , Maurizio Conticelli , and Marco Cosimo Simeone 0000-0003-3798-9580 Authors Info & Affiliations https://doi.org/10.22541/au.174084852.27969033/v1 Published Ecology and Evolution Version of record Peer review timeline 602 views 250 downloads Contents Abstract Supplementary Material Information & Authors Metrics & Citations View Options References Figures Tables Media Share Abstract Poorly known, rare species are important biodiversity elements; understanding the relationships between their effective numbers, geographical distribution, ecology, and adaptive potential is an unquestionable, critical aspect to reverse biodiversity decline under climate change. Quercus crenata Lam. is a sporadic Mediterranean tree species with debated taxonomy and evolutionary history. Confusing identifications and a scattered distribution combine to an incautious lack of comprehensive and reliable information on its spatial distribution, ecology, and genetic resources, thereby hindering correct management and conservation efforts. This work undertook a first decisive step to address these knowledge gaps, integrating all previous dispersive information and presenting a comprehensive map of Q. crenata occurrences, with 923 established records: 495 new field observations and 428 verified from all available literature and online databases. The taxon occurs with extremely low individual numbers across central and northern Italy, southern France, western Slovenia and Croatia, mainly at altitudes between sea level and 1100 m a.s.l. The largest part of the species records are outside current networks of protected areas. The MaxEnt-based distribution model highlights Q. crenata adaptation to mild Mediterranean climates with moderate temperature fluctuations, moderate-to-high water requirements, and diverse soil types. A broader present potential distribution than currently assessed is suggested, underlining the possibility to identify new occurrences with accurate searches in targeted sites. Future range projections under three carbon emission scenarios with increasing severity (SSP1, SSP3, and SSP5) predict substantial range losses by 2100, ranging from 32% to a drastic 99% reduction under the most severe scenario (SSP5). On these bases, our findings underline the urgent need to improve current conservation practices, which should be conveniently implemented by exhaustive genetic investigations. TITLE: A vanishing imprint? Modelling the present and future distribution of the enigmatic Quercus crenata Lam., a Mediterranean sporadic tree species Authors: Giuseppe Antonelli 1 , Giuseppe Puddu 1,2* , Chiara Cipollini 3 , Gianluca Sabatini 4 , Maurizio Conticelli 5 , Marco Cosimo Simeone 1 Affiliations: 1: Department of Agricultural and Forest Sciences (DAFNE), University of Tuscia, 01100 Viterbo (VT - Italy) 2: Regione Lazio – Lago di Vico Natural Reserve – Regione Lazio, Caprarola (VT - Italy) 3: Unaffiliated, Via Pierpaolo Pasolini 7, Fiano Romano (RM - Italy); [email protected] 4: Trifolium Coop, via del Paradiso 3 Viterbo (VT - Italy); [email protected] 5: Unaffiliated, Vicolo della pace 8, Orvieto (TR - Italy) [email protected] *: corresponding Author: [email protected] Giuseppe Antonelli 0009-0007-7059-6742 Giuseppe Puddu 0000-0001-9705-8257 Gianluca Sabatini 0009-0006-7498-8789 Marco Cosimo Simeone 0000-0003-3798-9580 Abstract Poorly known, rare species are important biodiversity elements; understanding the relationships between their effective numbers, geographical distribution, ecology, and adaptive potential is an unquestionable, critical aspect to reverse biodiversity decline under climate change. Quercus crenata Lam. is a sporadic Mediterranean tree species with debated taxonomy and evolutionary history. Confusing identifications and a scattered distribution combine to an incautious lack of comprehensive and reliable information on its spatial distribution, ecology, and genetic resources, thereby hindering correct management and conservation efforts. This work undertook a first decisive step to address these knowledge gaps, integrating all previous dispersive information and presenting a comprehensive map of Q. crenata occurrences, with 923 established records: 495 new field observations and 428 verified from all available literature and online databases. The taxon occurs with extremely low individual numbers across central and northern Italy, southern France, western Slovenia and Croatia, mainly at altitudes between sea level and 1100 m a.s.l. The largest part of the species records are outside current networks of protected areas. The MaxEnt-based distribution model highlights Q. crenata adaptation to mild Mediterranean climates with moderate temperature fluctuations, moderate-to-high water requirements, and diverse soil types. A broader present potential distribution than currently assessed is suggested, underlining the possibility to identify new occurrences with accurate searches in targeted sites. Future range projections under three carbon emission scenarios with increasing severity (SSP1, SSP3, and SSP5) predict substantial range losses by 2100, ranging from 32% to a drastic 99% reduction under the most severe scenario (SSP5). On these bases, our findings underline the urgent need to improve current conservation practices, which should be conveniently implemented by exhaustive genetic investigations. Keywords: Quercus , Mediterranean Flora, Forest Biodiversity, Global Change, Species Distribution Model, MaxEnt Introduction Quercus crenata Lam. [= Q. x hispanica Lam. Quercus (Fagaceae), subgen. Cerris , Sect. Cerris (Denk et al. 2017)], is a sporadic tree species of the South-central Mediterranean region that has long captivated the attention of researchers due to its iconic appearance, controversial taxonomy, enigmatic distribution and recently questioned evolutionary history. It is a large (up to 2 m of diameter and over 20 meters height), spreading tree, with crenate leaves that are shiny green above and whitish pubescent below, and large acorns (maturing in two-years) with slender, non-appressed cupscales (Pignatti et al. 2017). The silvery corky bark (1-2 cm thick) and the semi-evergreen habit, together with its architecture and longevity, are probably the main features making this tree particularly attractive to the point that many plants are presumed to have been intentionally dispersed (or preserved) by man as symbols of the local natural and cultural heritage (Armiraglio et al. 2003). Many majestic plants today stand in numerous arboreta, parks and botanical gardens in Central Europe (https://www.internationaloaksociety.org/). Although widely adopted by the nursery trade in the form of numerous vegetatively propagated cultivars (https://www.treesandshrubsonline.org/), only a few hundred plants are known to thrive in the wild as sporadic members of mixed deciduous oak forests and, quite frequently, isolated trees in forest margins along roadways, private gardens, edges and open spaces in country contexts (Grossoni et al. 2020). Known in historic floras since antiquity (Cristofolini et al. 2017), the taxonomy of Q. crenata , first formally described by Lamarck (1785), has been the subject of considerable debate, adding several alternative names to a rather complex nomenclature (reviewed in Cristofolini et al. 2017). Although generally considered a natural hybrid between Q. cerris and Q. suber (Schwarz 1936-39; Pignatti et al. 2017), two oaks with partially overlapping distributions, its occurrence in regions lacking Q. suber and recent phylogenomic data (Hipp et al. 2020; Denk et al. 2023) challenge the simple hybridization hypothesis, revealing a more intricate evolutionary history. In particular, Denk et al. (2023) inferred a dated phylogeny of Quercus Sect. Cerris using Restriction site-Associated DNA sequencing (RAD-seq) and fossilized birth–death model calibrated with 47 fossils. Unexpectedly, the obtained phylogenomic tree resolved Q. crenata as a less-derived species, likely differentiated ca. 10-5 Ma (Late Miocene to Early Pliocene), within an autonomous evolutionary lineage (subsection Suber ) that includes Q. suber and diverged from the remaining Western Eurasian members of the same section (including Q. cerris ) ca. 24 Ma (Late Oligocene to Early Miocene). Denk et al. (2023) also tested the introgressive hybridisation hypothesis with the Patterson’s D-statistic test (Durand et al. 2011) and recorded consistent support for admixture with Q. cerris as well as with other oak species within the same section, suggesting ancient introgression among precursors of the modern species and retention of ancient polymorphisms as likely explanations. In agreement with previous nuclear ribosomal DNA sequence data (Simeone et al. 2018; see also Bellarosa et al. 2005) and further supported by several symplesiomorphic traits of the leaf morphotype (e.g., the strongly developed teeth with mucronate to cuspidate tips), and other putatively ancestral traits within Sect. Cerris (including the corky bark and semi-evergreen habit), new hypotheses on the origins of Q. crenata were thus advanced: the species may represent the remainder of an ancestral form rather than the product of secondary contacts between Q. suber and Q. cerris , as traditionally assumed based on a mere phenotypic intermediacy. Although based on low numbers (< 10 individuals analysed), these findings contribute to solidify “ Q. crenata ” as the correct and valid name, stabilize its classification and point to a remarkable importance as a relict species among West Eurasian oaks. Nevertheless, coexistence of Q. crenata with hybrid/introgressed Q. cerris x Q. suber individuals is feasible (Cristofolini & Crema 2005; Conte et al. 2007; Simeone et al. 2018). The two forms are morphologically, genetically and ecologically hardly traceable, further complicating a reliable comprehension of the true nature of this taxon . An adequate assessment of the species’ occurrence would certainly assist larger samplings and deeper studies to definitely assess whether we should consider Q. crenata as an irreplaceable evolutionary leftover within West Eurasian oaks or as an exemplary model of hybridization/introgression, possibly facilitating adaptation, migration, and demographic resilience (Petit et al. 2004; Dodd & Afzal-Rafii 2004). Nonetheless, the sporadic nature and complex taxonomy of Q. crenata have contributed to a significant uncertainty regarding its distribution. Current knowledge generally reports sparse occurrences extending from central Italy to southern France, Slovenia, and Croatia (Jalas & Suominen, 1976). Further (historical) records are disputed, particularly those from the Iberian Peninsula, Albania, Montenegro and Sicily where the taxon identity and presence are considered either uncertain or non-native (Amaral Franco 1990; Cristofolini et al. 2017; Barina et al. 2018). In Italy, preliminary assessments report Q. crenata at approximately 200 sites, concentrated along the central and northern Tyrrhenian coast of Italy, with scattered occurrence in central Apennines and Prealps, and very limited information for the Adriatic coast and southern Italy (Mercurio 1985; Cresta & Salvidio 1991; Armiraglio et al. 2003; Cristofolini et al. 2007). Today, Q. crenata is considered “not evaluated” in the Red List of Oaks (Eastwood & Oldfield 2007), and “Data Deficient” in the IUCN Red List of the European Trees (Rivers et al. 2019). Today, Q. crenata is considered “not evaluated” in the Red List of Oaks (Eastwood & Oldfield 2007), and “Data Deficient” in the IUCN Red List of the European Trees (Rivers et al. 2019). It is however listed as “Vulnerable” in the National Red Lists of vascular flora of France (UICN France 2018), cited as “Endangered” in the Slovenian report of the Global Forest Resources Assessment 2005 (FAO 2005) and is under protection, by law, in seven Italian administrative regions. Understanding the relationships between species’ geographical distribution and climatic factors is a critical ecological aspect for biodiversity conservation, especially under the current climate change (Pecl et al. 2017; Adam et al. 2022). In this view, poorly known species with small geographical ranges and low numbers of individuals should be given high conservation priority because they might face a high extinction risk due to demographic, genetic and ecological stochasticity (Vincent et al. 2020). Ideally, sound assessment of a species’ population counts, distribution and ecology would allow evaluating the impact of climate change and the consequent application of efficient conservation strategies. Recent research with Species Distribution Models (SDMs) has demonstrated that climate warming will significantly reduce and reshape the geographical distribution of many European tree species (Dyderski et al. 2025), including oaks (Vessella et al. 2017; Suicmez & Avci 2023; Özcan et al. 2024; Sadeghi et al. 2024). Because of the absence of a thorough study, Q. crenata is lagging behind in knowledge and detailed analyses, with only a handful of surveys focusing on local and regional territories within its range. SDMs predict the geographic distribution of species by analysing the relationship between known occurrences and environmental variables (Franklin 2010). These models generate probabilities that reflect the likelihood of a species’ presence in a given area, identifying areas with similar environmental conditions. These probabilities can be interpreted as relative habitat suitability (Guisan et al. 2017), providing insights into current distributions, potential range shifts under climate change, and informing conservation planning. MaxEnt (Maximum Entropy; Phillips et al. 2006) is a powerful machine learning tool well-suited for species with sparse and uneven occurrence data, especially those with limited presence-only records (e.g., Gao et al. 2022; Zhu et al. 2024; Dong et al. 2025). In this study, we developed a statistically robust MaxEnt-based SDM to model the current and future distribution of Q. crenata using verified and operable occurrence data. Our objectives were to: (a) provide an overdue exhaustive assessment of Q. crenata actual distribution, evaluating whether if its sporadic nature might entail rarity; (b) determine the primary environmental factors influencing its potential distribution, highlighting regions where the species may have been overlooked; and (c) predict its vulnerability under future climatic conditions. The findings will contribute to a better understanding of the ecology of Q. crenata , recommend priority areas for its effective conservation and serve as a baseline for future field searches and further investigations on its genetic diversity. Materials and methods Presence data and study area All Q. crenata presence data were compiled from georeferenced records. The primary data source were field surveys conducted by the authors during the winter seasons from 2008 to 2024 in central Italy. Scientific publications and grey literature (including technical reports, policy and working documents, newsletters) documenting Q. crenata punctual occurrence and unequivocal identity all over its range were critically reviewed and supplemented with free contributions from botanists and forestry professionals. The dataset was further enriched with verified data from publicly accessible online databases (Alfeus et al. 2025), including iNaturalist (https://www.inaturalist.org/), GBIF (https://www.gbif.org/), WikiPlantBase #Italia (https://bot.biologia.unipi.it/wpb/italia), and the French National Inventory of Natural Heritage (https://inpn.mnhn.fr/). Due to the heterogeneity of the collected presence data, each report was treated as a unique occurrence. INSERT HERE FIGURE 1 The study area for the distribution model includes all regions with verified occurrences of Q. crenata , namely peninsular Italy, southern France, Slovenia, and coastal Croatia. To account for historical range dynamics and potential future shifts (Barve et al. 2011), we expanded the study area to include the rest of the French Mediterranean coast, parts of the Catalonian coast, sections of inland France and Switzerland, the Balkan coasts extending to Albania, and parts of the inland Balkans, as well as Sicily, Sardinia, Corsica, and smaller islands. This expanded area spans a total of 811,900 km². To mitigate potential biases in spatial distribution of Q. crenata , which could have led to an over-representation of specific environmental parameters in the model, we implemented a standardized sampling approach: the study area was divided into 1 km² grid cells using the WGS 84 / UTM 32N (EPSG:32632) coordinate system, and the centroid of each grid cell containing at least one occurrence was retained, thus ensuring only a single representative location per cell. Environmental data A combination of bioclimatic, soil, and topographic data was employed to describe the environmental niche and the current potential distribution of Q. crenata . Bioclimatic variables representing present conditions were sourced from the CHELSA repository (Karger et al. 2017), version 2.1 (Karger et al. 2018). They consist of nineteen bioclimatic variables, recognized as key drivers of species distributions computed over the period 1981-2010, with a native resolution of 30 arc-seconds. Soil data, extracted from the Harmonized World Soil Database v2.01 (FAO & IIASA 2023), also had a native resolution of 30 arc-seconds. This dataset classifies soils according to the World Reference Base for Soil Resources (IUSS Working Group WRB 2022). Topographic data, including elevation, slope, and aspect, were obtained from ESA Copernicus GLO-90 v2023.1 (European Space Agency, 2024), with a finer native resolution of 90 meters. All these raster layers were resampled to a common resolution of 1 km using the study area grid as a reference. Their descriptions are reported in Appendix A, Table A1. To address potential multicollinearity, a statistical issue that arises when independent variables in a regression model are highly correlated, we conducted an iterative variable selection process using Pearson’s correlation coefficient. A correlation matrix was calculated for all 23 environmental variables (3 topographic, 1 soil, and 19 bioclimatic) and pairs of variables with a correlation coefficient (|r|) greater than 0.7 (Dormann et al. 2013) were identified as highly correlated. To remove redundancy, we iteratively removed the variable with the highest average correlation to the others, recalculated the correlation matrix, and repeated the process until all remaining variables had a correlation coefficient below the threshold. During this process, BIO12 (annual precipitation), BIO4 (temperature seasonality), and BIO15 (precipitation seasonality) were retained due to their ecological meaning: BIO12 was included as it provides a fundamental measure of precipitation, capturing the overall influence of precipitation without bias toward specific ecological hypotheses, while BIO4 and BIO15 were retained for their value in understanding climatic patterns, as they help differentiate between Mediterranean and temperate climates. Additionally, to further assess potential multicollinearity among the selected variables, a Variance Inflation Factor (VIF) analysis was performed as a secondary check, as Pearson’s correlation analysis may not always identify complex forms of multicollinearity; a threshold of 5 was used, where values below 5 indicate low correlation among predictors (James et al. 2013). MaxEnt modelling The distribution of Q. crenata was modelled using MaxEnt version 3.4.4 (Phillips et al. 2006) within an R environment. Given the unplanned sampling and heterogeneous distribution of presence data, MaxEnt was a suitable choice for its robustness in handling such data (Phillips & Elith 2013). To optimize MaxEnt model parameters and reduce overfitting, 50 unique combinations of feature classes and regularization multipliers were tested. The feature classes tested included linear (L), quadratic (Q), product (P), hinge (H), and threshold (T), in the following configurations: L, LQ, LQH, LQHP, LQHPT; regularization multipliers ranged from 0.5 to 5.0 in 0.5 increments. All models were run using 20,000 background points. To account for spatial autocorrelation, we employed a block cross-validation approach, dividing the study area into four spatially distinct blocks; this ensured independence between training and testing data by preventing the random assignment of occurrences and background points. By leveraging data heterogeneity, this approach minimizes spatial autocorrelation, making it particularly suitable for model transfer across space or time (Wenger & Olden, 2012; Radosavljevic & Anderson 2014). For each iteration, the model was trained on three blocks and tested on the remaining one. Different metrics were assessed to choose the best-performing model, specifically: Akaike Information Criterion correction (AICc; Burnham & Anderson 1998) was used for model selection, balancing model fit with complexity; the absolute difference between training and validation AUC (AUC diff ; Warren & Seifert 2011) was used to detect overfitting, with smaller differences indicating better generalization; and the Continuous Boyce Index (CBI; Hirzel et al. 2006), particularly useful for presence-only data, was used to assess the agreement between predicted and observed distributions, providing insight into the model’s ability to capture ecological patterns. Finally, the response curves of the best-performing combinations were examined to select the most ecologically interpretable model. These curves visualize how the predicted probability of occurrence changes with environmental conditions, providing insights into the species’ niche (Elith et al. 2006). This visualization allowed us to assess the plausibility of the model’s predictions and directly inform the selection of the model that best balanced high predictive performance, low overfitting, and ecological relevance. A detailed description of the overall modelling workflow including Overview, Data, Model, Assessment and Prediction (ODMAP protocol; Zurell et al. 2020) can be found in Appendix A, Table A2. Potential distribution analysis To better understand the spatial variation in the probability of Q. crenata occurrence, we categorized the predicted probabilities into four classes. These classes were defined using the following thresholds: Minimum Training Presence (MTP), the lowest probability at which the model correctly predicted the species’ presence during training; 10th Percentile Training Presence (P10), representing the probability below which 10% of the training presence points fell; and a fixed threshold of 1–1/e ≈ 0.632, corresponding to a predicted abundance of one individual per cell in a cloglog output (Phillips et al. 2017). Future projections To project future distributions of Q. crenata , we considered three time periods: 2011-2040, 2041-2070, and 2071-2100. Future bioclimatic data for these periods were also obtained from the CHELSA repository at a native resolution of 30 arc-seconds and then resampled to 1 km resolution. Specifically, these data were derived from the MPI-ESM 1.2 climate model at high resolution (Gutjahr et al. 2019) and describe global climate changes under three Shared Socioeconomic Pathways (SSP; IPCC, 2021) scenarios: SSP1-2.6 (optimistic), SSP3-7.0 (moderate), and SSP5-8.5 (pessimistic). Given the relatively slow rates of change in topographic and pedological features, these variables were assumed to remain constant in future projections. To better visualize the areas in which Q. crenata could be present in the future, and to delineate possible conservation areas, in the future projection maps the predicted probability was categorized into only two classes: above and below the fixed threshold of 0.632, representing high and low suitability, respectively. We quantified the number of raster cells with high suitability and computed the percentage difference between current and future cells across all occurrence probability classes. The reliability of both present and future projections was assessed using a Multivariate Environmental Similarity Surface (MESS) analysis (Elith et al. 2010), a method that compares the environmental conditions of the projected areas to those used in model training, thereby highlighting areas of dissimilarity. Updated map of Q. crenata distribution The data collection process resulted in a total of 923 verified and georeferenced occurrences of Q. crenata (dataset description available at: https://zenodo.org/records/14945097). These included 480 records obtained through the authors’ field surveys, 213 extracted directly from scientific and grey literature, 215 sourced from online databases, and 15 contributed by forestry professionals and botanists. These data form the basis for the first comprehensive map of Q. crenata actual distribution (Figure 1). Combined, the 495 newly field-observed and contributed records (53.6% of total) represent fresh data that integrate previously dispersive information and improve the knowledge on the present species’ range. This total number of the occurrences is the result of a selection process aimed at removing duplicates whenever the available information was sufficient, such as when the location was clearly repeated. Overall, the regions along the Tyrrhenian and Ligurian-Provençal Sea coast collectively exhibited the highest number of recorded occurrences, totalling 647 observations: 430 in Latium (West-central Italy), 218 in Tuscany (West-central Italy), 58 in Liguria (West-northern Italy), and 48 in Provence-Alpes-Côte d’Azur (South France). Significant occurrences were also observed across the entire arc of Italian Prealps, whereas only scattered specimens were found on the eastern Apennines slopes, in Southern Italy, and along the Italian, Slovenian, and Croatian Adriatic coasts. Occurrences were recorded across a broad elevational range, from 1 to 1086 m a.s.l. with a mean of 473 m a.s.l., while isolated occurrences were observed up to 1623 m a.s.l. Many individuals appeared to be in a state of vegetative health, and seed production was reported in many cases, although its extent varied. However, limited natural regeneration was evidenced in stands mostly located in the Central (Lazio, Tuscany) and Northern (Liguria, Piedmont) Italian regions. Additional 136 occurrences were not included in the analyses due to a lack of precise location data and/or insufficient reliability. These were mostly collected in historic herbarium specimens and online databases. Overall, the excluded records did not provide additional geographic information to the described distribution pattern, since the largest part of these records were from areas largely well-covered by georeferenced data. They however also included a small number of interesting records (nine in total) from Southern Italy (Basilicata and Calabria), the Italian Adriatic coast (Abruzzo), and Southern French coast (Occitanie), which are regions where records of the species are comparatively scarce. INSERT HERE TABLE 1 Environmental variable selection and model performance After filtering the data to ensure unique occurrences within 1 km² grid cells, 705 records were used to model the taxon’s potential distribution. The iterative selection process used to address potential multicollinearity identified a final set of 9 variables: slope, aspect, soil type, and six bioclimatic variables (BIO3, BIO4, BIO8, BIO9, BIO12, BIO15; see Appendix A, Figures A1 and A2). This reduced set effectively represents the environmental conditions of the study area while minimizing redundancy. Moreover, selected variables had a maximum VIF value of 2.2, indicating that multicollinearity was adequately mitigated (see Appendix A, Figure A3). The results from evaluation of the 50 candidate models showed that the LQHPT models outperformed others in terms of AICc (see Appendix A, Table A3). However, after analysing the response curves, the LQHPT models displayed irregular, step-like patterns that lacked ecological interpretability. LQHP models, on the other hand, while performing well in terms of AICc, exhibited smoother, more biologically plausible response curves, which facilitated clearer ecological interpretation and a better understanding of the species’ environmental preferences. The best-performing LQHP model in terms of AUC diff and CBI, selected for subsequent SDM development, was the one with a regularization multiplier of 2, showing an AUC diff of 0.047 ± 0.032 (the lower the better) and an average CBI on the validation data of 0.876 ± 0.140 (maximum = 1). INSERT HERE FIGURE 2 Modelled environmental niche of Q. crenata The most influential variables contributing the MaxEnt model (Table 1) were precipitation seasonality (BIO15), which had the highest contribution at 32.3%, followed by temperature seasonality (BIO4) at 22.9%, and annual precipitation (BIO12) contributing 14.9%. Pedological factors (Soil) and mean daily temperature of the driest quarter (BIO9) contributed similarly, with 10.2% and 10.1%, respectively; however, BIO9 exhibited a higher permutation importance (10.4%) compared to Soil (3.6%), indicating that BIO9 had a stronger influence on model performance. The response curves of the three most contributing variables (Figure 2) illustrate how the species’ probability of presence changes with each environmental variable, while the variables are kept constant at their mean values. For precipitation seasonality (BIO15), a unimodal relationship is observed, with peak suitability around 40 and 50 mm/year, indicating a preference for Mediterranean climatic conditions. Temperature seasonality (BIO4) shows a similar trend, with an optimum between 610 and 660 (6.1 and 6.6°C), suggesting a preference for a mild Mediterranean climate with moderate temperature fluctuations. BIO12 (annual precipitation) exhibits a maximum suitability for values between 650 and 1100 mm while showing slightly asymmetrical shape, suggesting that the species has moderate-to-high water requirements and is not well adapted to drought. Regarding the edaphic conditions, Q. crenata looks adaptable to different soil types. All obtained response curves are reported in the Appendix A (Figure A4). INSERT HERE FIGURE 3 Predicted current potential distribution Based on the model, the thresholds chosen to better visualize the spatial variation in the probability of Q. crenata occurrence were as follows: 0.0078 for Minimum Training Presence and 0.1746 for the 10th Percentile Training Presence, in addition to 0.632 as a fixed threshold. Probability classes were then defined as follows: High (values exceeding 0.632), Moderate (0.632 to 10P), Low (10P to MTP), and Negligible (values below the MTP). INSERT HERE TABLE 2 The areas resulting with high probability of presence (or high suitability, Figure 3) extend across central and northern Italy, particularly in well-documented areas such as northern Latium, west Umbria, Tuscany, and Liguria, highlighting a continuity of suitable habitats stretching up to the French Riviera in southern France. In Italy, more restricted and fragmented areas with high suitability were also detected in regions with fewer documented occurrences, such as coastal areas in central Abruzzo, Samnite and Campanian Apennines (Molise and Campania) and Pollino and Sila massifs in Calabria. In Croatia, areas of high suitability are scattered along the coast, with limited overlap with the locations of the recorded occurrences. In France, besides the well-documented Provence-Alpes-Côte d’Azur, high suitability was also predicted in eastern Occitanie and southern Auvergne Rhône-Alpes regions, particularly in the Rhône River basin, where no occurrences have been recorded. Other areas, including Sicily, Sardinia, Corsica, and parts of Albania, also exhibited scattered high-suitability areas despite the lack of any historical or current documented occurrence. The total extent covered by high-suitability areas resulted 35,273 km 2 , corresponding to the 4.4% of the study area extent (Table 2). Areas predicted to have moderate probability of presence, or moderate suitability, often border regions with high suitability. However, some areas of moderate suitability were also identified in disjunct locations, such as across the Italian Prealps up to Slovenia, where numerous occurrences were documented, and along the Adriatic coast in northern Marche, and in Gargano Promontory and Murge Plateau in Apulia, where only a few sparse occurrences were found. INSERT HERE FIGURE 4 INSERT HERE FIGURE 5 The impact of climate change on the distribution of Q. crenata Projections for Quercus crenata under three SSP scenarios (SSP1-2.6, SSP3-7.0, and SSP5-8.5) reveal different trajectories in habitat suitability over the 21st century (Figures 4 and 5). Under the optimistic SSP1 scenario, high-suitability areas initially shrink by 2070 but recover slightly by 2100, reaching 24100 km², with a loss of approximately 32% of the current total (Table 3). This fluctuation reflects the scenario’s assumption of stringent climate mitigation measures, resulting in a long-term stabilization of temperatures and precipitation patterns. However, the areas with recovered suitability are not always spatially aligned with current habitats, shifting slightly to the north and to higher altitudes (Figure 4c and 6). In contrast, the intermediate SSP3 scenario predicts a progressive decline in high-suitability areas, culminating in a reduction from 35273 km² to 15121 km² by 2100, a 55% loss compared to the present (Table 3). Central Italian regions like Latium and Tuscany, which currently host the densest occurrences, lose much of their suitability, while areas with high suitability, such as those present in the Ligurian Apennines and Calabrian Pollino and Sila massifs, shift to higher altitudes (Figures 4f and 6). Despite the absence of present occurrences, the Rhône River basin in France is suggestive of a great potential for high ecological suitability at northern latitudes. The SSP5 scenario, characterized by unmitigated emissions and rapid warming, presents the most dramatic habitat loss, with high-suitability areas dwindling to just 251 km² by 2100, a 99% reduction compared to the present (Table 3). These remaining suitable habitats are confined almost exclusively to the coastal area between northern Tuscany and the France-Italy border, highlighting the catastrophic impact of this high-emission scenario. Notably, the largest contiguous high-suitability area (25 km²), along with additional sparse areas (totaling 33 km², 13% of the total high-suitability areas extent), would fall within a protected area (the Alpi Apuane Regional Natural Park, in Tuscany). In this scenario, the Rhône River basin no longer offers suitable habitat (Figure 4i and 5). INSERT HERE FIGURE 6 INSERT HERE TABLE 3 In Appendix A, Figure A5, are presented the results of the MESS analysis, which was used to assess the reliability of the present and 2100 future projections. Discussion The geographical distribution and evolutionary history of Q. crenata present several unanswered questions, making it a particularly interesting case study of forest biodiversity in the Euro-Mediterranean region. This work fills the existing knowledge gap on the species’ distribution, integrating previous cluttered data with new information gathered through direct field observations and in-depth literature analysis. Hence, punctual distribution and modelled ecological niche of Q. crenata are put forward as crucial steps to foster future conservation programs and drive novel studies aimed at clarifying the species’ identity, origin and standing genetic diversity. However, MaxEnt-based models are obviously not devoid of some limitations. The bioclimatic, edaphic, and topographic variables generally used may not encompass all factors influencing a species’ distribution. For instance, species interactions, pathogens, and human activities (such as indiscriminate logging opposed to historical preservation of impressive individuals) could be significant as well, but the complexity of these factors, combined with a lack of spatially explicit data, prevents their incorporation into current modelling approaches. Despite these limitations, our study represents the first application of a distribution model to estimate the potential range of a European sporadically distributed tree species. The relevant number of the collected records was essential in ensuring a well-supported model and make the obtained results extremely valuable to suggest adequate conservation strategies. To these aims, the crucial role of field surveys, combined with comprehensive data mining and verification in herbaria and publicly accessible online data is remarked, especially for yet poorly known and sparsely distributed species. We are confident that our approach can serve as a guideline for similar studies on other sporadic or scattered species deserving protection, at both continental and national scales. Current distribution and conservation status The resulting 923 occurrences significantly increase the available knowledge on the actual distribution of Q. crenata and accurately delineate its current range. However, total documentation accounts for an alarming count of <2000 individuals (Frankham 2022; Hoban et al. 2023). Sparse groups and isolated plants of Q. crenata currently extend from the central and northern Apennines to the coasts of the Tyrrhenian Sea, passing through the Ligurian Apennines to the French Riviera. Even fewer, scattered individuals occur in the Italian Prealps, along the Adriatic coasts of Italy, Slovenia, and Croatia, and in southern Italy. While confirming previous generic maps (Jalas & Suonimen 1976; Sicily deleted by Brullo et al. 1999), our results significantly highlight Latium and Tuscany as the core regions of Q. crenata distribution and further provide a fundamental baseline for assessing its current conservation status and identifying key areas for protection. In facts, the updated distribution map, overlaid with data from Natura 2000 network of protected areas shows that a significant portion (approximately 70%) of Q. crenata occurrences is located outside existing protected sites, emphasizing the need for increased conservation efforts. This urgency is further stressed by the varying levels of protection conferred to the species across its range. Although Q. crenata is officially recognized and protected in seven Italian regions (Piedmont, Lombardy, Emilia-Romagna, Tuscany, Umbria, Marche, Molise), with a total of 354 occurrences, and further 21 individuals are under strict protection regime, listed as “Monumental trees” by Italian Minister of Agriculture and Forests (https://www.politicheagricole.it; updated to Nov. 2024), still a significant proportion of the Italian records (including those from Latium region, Liguria and the rare north-eastern and southern individuals) remains without protection. On the other hand, all French and the few Slovenian occurrences are nationally classified as Vulnerable or Threatened and consequently protected. However, the species is not currently included in any conservation list for Croatia, despite the documented rare occurrences. The low numbers of all extant Q. crenata individuals and the disparity in conservation status across countries and regions (Table 4) clearly urge a coordinated and comprehensive conservation strategy at the international level. INSERT HERE TABLE 4 Ecology of Q. crenata and potential species range The key environmental factors influencing Q. crenata distribution revealed by the model point towards a strong adaptation to Mediterranean climatic conditions. The importance of precipitation seasonality and the preference for moderate temperature fluctuations suggest a species thriving in a mild Mediterranean climate, where seasonal variability is present, but droughts are not severe. The apparent need for sufficient annual precipitation indicates moderate to high water requirements and a potential vulnerability to water scarcity, although it is adaptable to a variety of soil types. These results align well with previous ecological descriptions, which provided a partial understanding of Q. crenata ecology at a local scale (Mercurio 1985; Cresta & Salvidio 1991; Armiraglio et al. 2003), particularly regarding its water requirements and soil tolerance. Furthermore, our model is in agreement with the general pattern produced by Denk et al. (2023) who, even if based on a limited number of occurrences (84) retrieved from GBIF database (https://www.gbif.org/), characterized Q. crenata as a sub-Mediterranean member of the Temperate Broadleaf and Mixed Forest biome with extension into the Mediterranean Forests, Woodlands and Scrublands biome (Olson et al. 2001), with a Köppen–Geiger climate profile (Kottek et al. 2006; Peel et al. 2007) comprising Warm temperate summer-dry to fully humid climates, with warm or hot summers (Csa, with extensions into Csb, Cfb). Based on our model, the potential range of Q. crenata appears to extend beyond its currently known distribution, including vast portions of South Italy (northern Sicily, much of the territories of Calabria, Campania, and Molise), and the Rhone basin in the French regions of Rhône-Alpes and Languedoc-Roussillon. Although the poorly recorded species’ presence in these high-suitability areas matches the scarce historical and present evidence, and provided the species has not been progressively decimated by men, future targeted searches may reveal if the species is actually thriving in inaccessible areas or if it is being overlooked by foresters and botanists, thereby leading to an improved assessment of the species’ distribution and ecology; however, although desirable, a partial increase in the number of records would hardly be decisive to reverse the critical low census size detected in this study. At this regard, it is relevant to note that Brullo et al. (1999) excluded Q. crenata from the Sicilian flora and reported instead the presence of a highly similar taxon , Q. fontantesii Guss., occurring as small groups or isolated plants in the submontane belt of the north Sicilian Mountain systems (between 200 and 800 m a.s.l.), and considered the stabilized hybrid between Q. suber and Q. gussonei (Borzì) Brullo. Both Q. fontanesii and Q. gussonei are Sicilian endemic oaks currently accepted in main Italian and online World floras (Pignatti et al. 2017; https://www.ipni.org/; https://powo.science.kew.org/). However, Q. gussonei appears as a mere thermophilus form of Q. cerris and Q. fontanesii is morphologically undistinguishable from Q. crenata (Schicchi et al. 2000; Cristofolini et al. 2017). The potentially suitable areas of Q. crenata we identified in Sicily perfectly overlap those of Q. fontanesii , thereby pointing to a close correspondence between the two taxa . A thorough geno-taxonomic investigation would therefore be needed to disentangle such a questionable nomenclature and eventually expand the current distribution of Q. crenata to Sicily. ” Tyrrhenian” affinity and evolutionary questions Our data suggests a possible ”Tyrrhenian” affinity of the Q. crenata distribution, given that most of the high-suitability areas are located along the Tyrrhenian coast or in close-by regions. This could reflect a historical pattern of the species establishment in suitable (mesic) habitats and survival through times, likely determined by the peculiar geo-morphological and climate changes of the Euro-Mediterranean area since Miocene and in Pliocene, when a mosaic of different routes and local/regional ecological conditions occurred (Suc 1984; Krijgsmann 2002; Magri 2010), followed by persistence in local Pleistocene refugia (Médail & Diadema 2009; for further evidence of Italian Glacial refugia of oaks along the Tyrrhenian coast see also Di Pietro et al. 2024). At the same time, the concentration of Q. crenata occurrences within the Tyrrhenian regions, also characterized by overlapping ranges of its presumed parental species ( Q. cerris and Q. suber ), lends support to the hypothesis of a recent hybrid origin. However, recent phylogenomic studies and the presence of Q. crenata individuals in areas where Q. suber is absent reinforce questions about the complexity of the evolutionary processes that led to the formation of this taxon and support the hypothesis of an autonomous, today relict species among West Eurasian oaks, possibly coexisting with few, morphologically indistinguishable, hybrid individuals (Cristofolini & Crema 2005; Denk et al. 2023). Resolving the complex evolutionary puzzle of Q. crenata is not easy, given the current lack of diagnostic standard molecular markers (Conte et al. 2007; Bagnoli et al. 2016), but it would be critical to understand the species’ adaptability to future environmental changes. As an ancient evolutionary leftover, it may possess deep-time genetic legacies (“trait conservatism”; Cavender-Bares et al. 2016), possibly making it pre-adapted to face multifaceted ecological changes. However, the possibility of reduced genetic diversity and drifting, due to its extremely low population numbers, should be carefully evaluated. Conversely, as a (either recent or stabilised) hybrid taxon , it may possess an increased genetic diversity and be favoured by the transfer of adaptive alleles from the genetic backgrounds of both parents, enhancing local adaption and allowing to respond quickly to selective pressures (Rieseberg & Willis 2007; Taylor et al. 2015). In this case, the roles of climate, landscape, demography, and genetic diversity of the parental species should be adequately considered (Bolte et al. 2024). Our results show that Q. crenata has some phenotypic plasticity, despite the extremely low population numbers and the current lack of population genetic studies. Combining the here assessed species ecology and demographic data with novel studies on the species evolutionary history and the geographic structure of its genetic diversity would be essential to comprehend how Q. crenata will tackle future environmental changes. As an intriguing add-on, hindcasting Q. crenata distribution since the last Interglacial could reveal whether it experienced range stability or significant shifts and eventually indicate larger range overlaps with the presumed parental species (Vessella et al. 2015; Bagnoli et al. 2016). Climate Change impacts and conservation efforts Future projections regarding changes in high-suitability areas suggest that rising temperatures and changes in rainfall patterns could significantly impact Q. crenata habitats, with varying degrees of severity depending on future greenhouse gas emission scenarios. Under the more optimistic SSP1 scenario, while initial habitat shrinkage is expected, some recovery is possible by the end of the century, though the location of suitable areas may shift, potentially complicating conservation. The moderate SSP3 scenario projects a substantial decline in suitable habitat, particularly areas with currently numerous occurrences, with remaining suitable areas shifting to higher altitudes and more northerly latitudes, including regions where the species is not currently found. The most severe SSP5 scenario paints a dire picture, with a near-total loss of suitable habitat by the end of the century, leaving only small, fragmented refugia. These projections underscore the critical need for effective climate change mitigation strategies to preserve suitable habitat for this species. The coastal area between northern Tuscany and the France-Italy border, especially the Alpi Apuane Regional Park in Tuscany (despite currently few occurrences), could emerge as a priority area for the conservation of Q. crenata . Therefore, for the future, an in-depth assessment of the possible migration routes and strategies necessary to facilitate the movement of the species is crucial, along with a thorough implementation of current conservation measures. In addition, if the hybrid origin hypothesis for Q. crenata will be confirmed, comparing its future distribution to those of the putative parent species ( Q. cerris and Q. suber ) will be critical as well. Recent studies showed that under intermediate emission scenarios Q. cerris is projected to expand northward and to higher altitudes, potentially increasing its presence in Central European regions, while its distribution in Italy may contract (Illés and Móricz, 2022). Quercus suber , conversely, is expected to experience moderate range contraction without significant shifts (Vessella et al. 2017). Despite some contraction in areas suitable for Q. crenata , the continued presence of both parent species in close-by areas may thus help maintain some hybridization, possibly allowing demographic renewal. However, under extreme emission scenarios, Q. cerris is likely to experience greater area losses in Italy, particularly at lower altitudes (Illés & Móricz, 2022), while Q. suber is expected to face severe range reduction, with high-suitability areas shrinking to few isolated refugia (Vessella et al. 2017). This near-total loss of Q. suber suitable habitats across much of its range and the consequent limitation of genetic exchange with Q. cerris therefore raises important concerns about Q. crenata future presence. The extinction risk could be further exacerbated by the restricted ecologically suitable areas for Q. crenata , which are projected to become extremely fragmented. In this scenario, recommendations for the sustainable conservation and management of Q. crenata should certainly consider a prompt assessment of the species’ status at the international level and involve steering policy and actions to search and catalogue, monitor, and protect all Q. crenata individuals across its range. Embracing the recent European Nature Restoration Law (http://data.europa.eu/eli/reg/2024/1991/oj), a key element of the EU Biodiversity Strategy for 2030 (European Commission, 2021), coordinated international plans should pursue maintenance, connection, restoration, and possibly an increase, of the forest habitats where Q. crenata (and the putative parental species, Q. cerris and Q. suber ) grows. Germplasm banking and plantations addressing afforestation/reforestation initiatives and tree planting in urban and peri-urban contexts could play an important role as well, to enlarge Q. crenata population quantity. Finally, studies on allelic diversity and genetic structure of Q. crenata would be of utmost importance. Besides revealing the efficacy of gene flow, which is especially critical for long-term survival of sporadic tree species, these studies could allow the identification of particularly vulnerable individual/populations or those with higher adaptive potential. Conclusion In this work, we provide an exhaustive assessment of the actual geographical distribution of Q. crenata , describe its ecology, and explore the consequences of climate change on its potential habitats to recommend conservation actions. Quercus crenata mostly thrives along west central and northern regions of Italy, with scattered occurrences in confining regions, generally outside protected areas. As a member of moderately mesic forests and open habitats, with strong adaptation to Mediterranean climatic conditions, the species ecology is primarily determined by sufficient precipitation and moderate temperature fluctuations; both features are projected to be highly impacted by global change in the next decades. Furthermore, the overall count of the recorded individual suggests an alarming census size of no more than 2000 adult trees, highlighting serious risks of species extinction and/or inbreeding depression with loss of genetic diversity and adaptive potential. Promoting individual and stand protection, monitoring health and reproduction, appear thus mandatory; at the same time, artificial plantations could improve genetic connectivity, a fundamental issue especially for species having small, scattered populations and a fragmented distribution. In Italy, all gathered Q. crenata records become gradually rarer moving southwards of the core distribution range, despite the potential ecological suitability we estimated in some regions. One additional, major outcome of our results is therefore to stress future searches in all those suitable areas, to definitely assess the species range and individual numbers, with a crucial impact on the species’ conservation issues. In this view, we further call for deeper analyses to understand the true nature and origin of this oak and possibly resolve all detrimental cases with a controversial nomenclature. Acknowledgements This study was done within the frame of the National Recovery and Resilience Plan (NRRP), Project ECS 0000024 Rome Technopole CUP B83C22002820006, Investment 1.5 Author contributions Giuseppe Antonelli : conceptualization (equal); data curation (equal); formal analysis (lead); methodology (lead); investigation (supporting); writing – original draft (equal); writing – review and editing (equal). Giuseppe Puddu : conceptualization (equal); data curation (equal); formal analysis (supporting); methodology (supporting); investigation (lead); supervision (supporting); writing – original draft (equal); writing – review and editing (equal). Chiara Cipollini ; data curation (equal); investigation (supporting); writing – review and editing (equal). Gianluca Sabatini : investigation (supporting); writing – review and editing (equal). Maurizio Conticelli : investigation (supporting); writing – review and editing (equal). Marco Cosimo Simeone : conceptualization (equal); data curation (equal); formal analysis (supporting); methodology (supporting); investigation (supporting); supervision (lead); writing – original draft (equal); writing – review and editing (equal). Data availability statement The detailed list of all the Q. crenata occurrences, relevant R scripts, and the environmental variable layes used for the model, are available at https://zenodo.org/records/14945097 All other relevant information is included within the manuscript. Conflicts of Interest The authors declare no potential sources of conflicts of interest. Bibliography Adam, M.M., B. Lenzner, M. van Kleunen, & F. Essl. 2022. Call for integrating future patterns of biodiversity into European conservation policy. Conservation Letters 15: e12911. https://doi.org/10.1111/conl.12911 Alfeus, M., J. Irish, & K. Birkhofer. 2025. Recognition and completeness metrics from iNaturalist and GBIF can inform future citizen science and research projects: a case study on arthropods in Namibia. Biodiversity and Conservation 34: 467–480. https://doi.org/10.1007/s10531-024-02981-z Amaral Franco, J. 1990. Quercus. In Flora iberica, S. Castroviejo (ed.). Madrid: Real Jardín Botánico de Madrid, CSIC. Armiraglio, S., C. De Carli, C. Ravazzi, F. Di Carlo, G. Lazzarin, S. Scortegagna, & F. Tagliaferri. 2003. Nuove stazioni di Quercus crenata Lam. nelle Prealpi Lombarde. Note ecologiche e distribuzione in Italia Nordorientale. Informatore Botanico Italiano 35(2): 289-300. Bagnoli, F., Y. Tsuda, S. Fineschi, P. Bruschi, D. Magri, P. Zhelev, L. Paule, M.C. Simeone, S.C. González-Martínez, & G.G. Vendramin. 2016. Combining molecular and fossil data to infer demographic history of Quercus cerris: insights on European eastern glacial refugia. Journal of Biogeography 43: 679-690. https://doi.org/10.1111/jbi.12673 Barina, Z., G. Somogyi, D. Pifkó, & M. Rakaj. 2018. Checklist of vascular plants of Albania. Phytotaxa 378(1): 1-339. https://doi.org/10.11646/phytotaxa.378.1.1 Barve, N., V. Barve, A. Jiménez-Valverde, A. Lira-Noriega, S.P. Maher, A.T. Peterson, J. Soberón, & F. Villalobos. 2011. The crucial role of the accessible area in ecological niche modeling and species distribution modeling. Ecological Modelling 222(11): 1810-1819. https://doi.org/10.1016/j.ecolmodel.2011.02.011 Bellarosa, R., M.C. Simeone, A. Papini, & B. Schirone. 2005. Utility of ITS sequence data for phylogenetic reconstruction of Italian Quercus spp. Molecular Phylogenetics and Evolution 34: 355-370. https://doi.org/10.1016/j.ympev.2004.10.014 Bolte, C.E., T. Phannareth, M. Zavala-Paez, B.N. Sutara, M.F. Can, M.C. Fitzpatrick, J.A. Holliday, S.R. Keller, & J.A. Hamilton. 2024. Genomic insights into hybrid zone formation: The role of climate, landscape, and demography in the emergence of a novel hybrid lineage. Molecular Ecology 33(14): e17430. https://doi.org/10.1111/mec.17430 Brullo, S., R. Guarino, & G. Siracusa. 1999. Revisione tassonomica delle querce caducifoglie della Sicilia. Webbia 54: 1–72. https://doi.org/10.1080/00837792.1999.10670670 Burnham, K.P., & D.R. Anderson. 1998. Model Selection and Inference: A Practical Information-Theoretic Approach. New York: Springer-Verlag. Cavender-Bares, J., D.D. Ackerly, S.E. Hobbie, & P.A. Townsend. 2016. Evolutionary legacy effects on ecosystems: biogeographic origins, plant traits, and implications for management in the era of global change. Annual Review of Ecology, Evolution, and Systematics 47: 433–462. https://doi.org/10.1146/annurev-ecolsys-121415-032229 Conte, L., C. Cotti, & G. Cristofolini. 2007. Molecular evidence for hybrid origin of Quercus crenata Lam. (Fagaceae) from Q. cerris L. and Q. suber L. Plant Biosyst 141: 181–193. https://doi.org/10.1080/11263500701401463 Cresta, P., & S. Salvidio. 1991. Stazioni di Quercus crenata Lam. in Liguria: l’alta Val Bormida. Italia Forestale e Montana 46: 133-148. Cristofolini, G., & S. Crema. 2005. A morphometric study of the Quercus crenata species complex (Fagaceae). Bot. Helv. 115: 155-167. https://doi.org/10.1007/s00035-005-0728-5 Cristofolini, G., A. Managlia, & S. Crema. 2017. Quercus crenata – the correct if long-forgotten name of the hybrid Quercus cerris × Quercus suber. Israel Journal of Plant Sciences 64(1-2): 57-63. https://doi.org/10.1080/07929978.2016.1253981 Denk, T., G.W. Grimm, A.L. Hipp, J.M. Bouchal, E.-D. Schulze, & M.C. Simeone. 2023. Niche evolution in a northern temperate tree lineage: biogeographic legacies in cork oaks (Quercus sect. Cerris). Annals of Botany 131: 769-787. https://doi.org/10.1093/aob/mcad032 Denk, T., G.W. Grimm, P.S. Manos, M. Deng, & A.L. Hipp. 2017. An updated infrageneric classification of the oaks: review of previous taxonomic schemes and synthesis of evolutionary patterns. In Tree physiology. Oaks physiological ecology. Exploring the functional diversity of genus Quercus L., 13–38, E. Gil-Pelegrín, J.J. Peguero-Pina, & D. Sancho-Knapik (eds.). Cham (Switzerland): Springer. https://doi.org/10.1007/978-3-319-69099-5_2 Di Pietro, R., L. Quaranta, C. Mattioni, M.C. Simeone, P. Di Marzio, E. Proietti, P. Fortini. 2024. Chloroplast Haplotype Diversity in the White Oak Populations of the Italian Peninsula, Sicily, & Sardinia. Forests 15: 864. https://doi.org/10.3390/f15050864 Dodd, R.S., & Z. Afzal-Rafii. 2004. Selection and dispersal in a multispecies oak hybrid zone. Evolution 58: 261–269. https://doi.org/10.1111/j.0014-3820.2004.tb01643.x Dong, Z.-H., H. Jiang, W. Zhang, J. Wu, Y. Yang, T. Yang, J. Zhao, C. Luo, X. Yang, & G. Li. 2025. Potential distribution prediction of Terminalia chebula Retz. in China under current and future climate Scenarios. Ecol Evol 15: e70908. https://doi.org/10.1002/ece3.70908 Dormann, C.F., J. Elith, S. Bacher, C. Buchmann, G. Carl, G. Carré, J.R. García Marquéz, B. Gruber, B. Lafourcade, P.J. Leitão, T. Münkemüller, C. McClean, P.E. Osborne, B. Reineking, B. Schröder, A.K. Skidmore, D. Zurell, & S. Lautenbach. 2013. Collinearity: a review of methods to deal with it and a simulation study evaluating their performance. Ecography 36: 27-46. https://doi.org/10.1111/j.1600-0587.2012.07348.x Durand, E.Y., N. Patterson, D. Reich, & M. Slatkin. 2011. Testing for ancient admixture between closely related populations. Molecular Biology and Evolution 28: 2239–2252. https://doi.org/10.1093/molbev/msr048 Dyderski, M.K., S. Paź-Dyderska, A.M. Jagodziński, & R. Puchałka. 2025. Shifts in native tree species distributions in Europe under climate change. Journal of Environmental Management 373: 123504. https://doi.org/10.1016/j.jenvman.2024.123504 Eastwood, A., & S. Oldfield. 2007. The Red List of Oaks. Cambridge: Flora and Fauna International. Elith, J., C.H. Graham, R.P. Anderson, M. Dudík, S. Ferrier, A. Guisan, R.J. Hijmans, F. Huettmann, J.R. Leathwick, A. Lehmann, J. Li, L.G. Lohmann, B.A. Loiselle, G. Manion, C. Moritz, M. Nakamura, Y. Nakazawa, J.Mc.M. Overton, A.T. Peterson, S.J. Phillips, K. Richardson, R. Scachetti-Pereira, R.E. Schapire, J. Soberón, S. Williams, M.S. Wisz, & N.E. Zimmermann. 2006. Novel methods improve prediction of species distributions from occurrence data. Ecography 29: 129–151. https://doi.org/10.1111/j.2006.0906-7590.04596.x Elith, J., M. Kearney, & S. Phillips. 2010. The art of modelling range-shifting species. Methods in Ecology and Evolution 1: 330-342. https://doi.org/10.1111/j.2041-210X.2010.00036.x European Commission: Directorate-General for Environment. 2021. EU biodiversity strategy for 2030 – Bringing nature back into our lives. Luxembourg: Publications Office of the European Union. https://data.europa.eu/doi/10.2779/677548 European Space Agency. 2024. Copernicus Global Digital Elevation Model. Distributed by OpenTopography. https://doi.org/10.5069/G9028PQB. Accessed 2024-08-24. FAO, & IIASA. 2023. Harmonized World Soil Database Version 2.0. Rome and Laxenburg. https://doi.org/10.4060/cc3823en FAO. 2005. Global forest resources assessment 2005: Country report Slovenia (FRA2005/094). Rome: FAO. Frankham, R. 2022. Evaluation of proposed genetic goals and targets for the Convention on Biological Diversity. Conservation Genetics 23: 865–870. https://doi.org/10.1007/s10592-022-01459-1 Franklin, J. 2010. Mapping Species Distributions: Spatial Inference and Prediction. Cambridge: Cambridge University Press. Gao, X., J. Liu, & Z. Huang. 2022. The impact of climate change on the distribution of rare and endangered tree Firmiana kwangsiensis using the Maxent modeling. Ecology and Evolution 12: e9165. https://doi.org/10.1002/ece3.9165 Grossoni, P., P. Bruschi, F. Bussotti, M. Pollastrini, & F. Selvi. 2020. Trattato di botanica forestale. 2. Angiosperme. Milano: Wolters Kluwer Italia. Guisan, A., W. Thuiller, & N.E. Zimmermann. 2017. Habitat Suitability and Distribution Models with applications in R. Cambridge: Cambridge University Press. Gutjahr, O., D. Putrasahan, K. Lohmann, J.H. Jungclaus, J.-S. von Storch, N. Brüggemann, H. Haak, & A. Stössel. 2019. Max Planck Institute Earth System Model (MPI-ESM1.2) for the High-Resolution Model Intercomparison Project (HighResMIP). Geosci. Model Dev. 12: 3241–3281. https://doi.org/10.5194/gmd-12-3241-2019 Hirzel, A.H., G. Le Lay, V. Helfer, C. Randin, & A. Guisan. 2006. Evaluating the ability of habitat suitability models to predict species presences. Ecological Modelling 199(2): 142-152. https://doi.org/10.1016/j.ecolmodel.2006.05.017 Hipp, A.L., P.S. Manos, M. Hahn, M. Avishai, C. Bodénès, J. Cavender-Bares, A. Crowl, M. Deng, T. Denk, S. Fitz-Gibbon, O. Gailing, M. Socorro González-Elizondo, A. González-Rodríguez, G.W. Grimm, X-L. Jiang, A. Kremer, I. Lesur, J.D. McVay, C. Plomion, H. Rodríguez-Correa, E.-D. Schulze, M.C. Simeone, V.L. Sork, & S. Valencia-Avalos. 2020. Genomic landscape of the global oak phylogeny. New Phytologist 226: 1198-1212. https://doi.org/10.1111/nph.16162 Hoban, S., J.M. da Silva, A. Mastretta-Yanes, C.E. Grueber, M. Heuertz, M.E. Hunter, J. Mergeay, I. Paz-Vinas, K. Fukaya, F. Ishihama, R. Jordan, V. Köppä, M.C. Latorre-Cárdenas, A.J. MacDonald, V. Rincon-Parra, P. Sjögren-Gulve, N. Tani, H. Thurfjell, & L. Laikre. 2023. Monitoring status and trends in genetic diversity for the Convention on Biological Diversity: an ongoing assessment of genetic indicators in nine countries”. Conservation Letters 16: e12953. https://doi.org/10.1111/conl.12953 Illés, G., & N. Móricz. 2022. Climate envelope analyses suggests significant rearrangements in the distribution ranges of Central European tree species. Annals of Forest Science 79: 35. https://doi.org/10.1186/s13595-022-01154-8 IPCC. 2021. Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge: Cambridge University Press. IUSS Working Group WRB. 2022. World Reference Base for Soil Resources. International soil classification system for naming soils and creating legends for soil maps. 4th edition. Vienna: International Union of Soil Sciences (IUSS). Jalas, J., & J. Suominen (eds.). 1976. Atlas Florae Europaeae. Distribution of Vascular Plants in Europe. Vol. 3. Salicaceae to Balanophoraceae. Helsinki: Committee for Mapping the Flora of Europe, Societas Biologica Fennica Vanamo. James, G., D. Witten, T. Hastie, & R. Tibshirani. 2013. Linear Regression. In “An Introduction to Statistical Learning: with Applications in R”, 59–126. New York: Springer. Krijgsman, W. 2002. The mediterranean: mare nostrum of earth sciences. Earth and Planetary Science Letters 205: 1–12. https://doi.org/10.1016/S0012-821X(02)01008-7 Karger, D.N., O. Conrad, J. Böhner, T. Kawohl, H. Kreft, R.W. Soria-Auza, N.E. Zimmermann, H.P. Linder, & M. Kessler. 2017. Climatologies at high resolution for the earth’s land surface areas. Scientific Data 4: 170122. https://doi.org/10.1038/sdata.2017.122 Karger, D.N., O. Conrad, J. Böhner, T. Kawohl, H. Kreft, R.W. Soria-Auza, N.E. Zimmermann, H.P. Linder, & M. Kessler. 2018. Data from: Climatologies at high resolution for the earth’s land surface areas. EnviDat. https://doi.org/10.16904/envidat.228.v2.1 Kottek, M., J. Grieser, C. Beck, B. Rudolf, & F. Rubel. 2006. World map of the Köppen-Geiger climate classification updated. Meteorologische Zeitschrift 15(3): 259–263. https://doi.org/10.1127/0941-2948/2006/0130 Lamarck, J.-B.P.A. de M. de. 1785. Encyclopédie Méthodique Botanique. Vol. 1 Part 2. Paris adn Liege: Panchoucke and Plomteux. Magri, D. 2010. Persistence of tree taxa in Europe and Quaternary climate changes. Quaternary International 219: 145–151. https://doi.org/10.1016/j.quaint.2009.10.032 Médail, F., & K. Diadema. 2009. Glacial refugia influence plant diversity patterns in the Mediterranean Basin. Journal of Biogeography 36(7): 1333–1345. https://doi.org/10.1111/j.1365-2699.2008.02051.x Mercurio, R. 1985. Note sulla distribuzione e sulla ecologia della Quercus crenata Lam. nella Toscana meridionale. Italia Forest. Mont. 40(4): 213-112. Olson, D.M., E. Dinerstein, E.D. Wikramanayake, N.D. Burgess, G.V.N. Powell, E.C. Underwood, J.A. D’Amico, I. Itoua, H.E. Strand, J.C. Morrison, C.J. Loucks, T.F. Allnutt, T.H. Ricketts, Y. Kura, J.F. Lamoreux, W.W. Wettengel, P. Hedao, & K.R. Kassem. 2001. Terrestrial ecoregions of the world: A new map of life on Earth. BioScience 51(11): 933–938. https://doi.org/10.1641/0006-3568(2001)051\%5b0933:TEOTWA\%5d2.0.CO;2 Özcan, A.U., D. Gülçin, G. Tuttu, J. Velázquez, S. Ayan, J. Stephan, U. Tuttu, A. Varlı, & K. Çiçek. 2024. The Future Possible Distribution of Kasnak Oak (Quercus vulcanica Boiss. and Heldr. ex Kotschy) in Anatolia under Climate Change Scenarios. Forests 15: 1551. https://doi.org/10.3390/f15091551 Pecl, G.T., M.B. Araujo, J.D. Bell, J. Blanchard, T.C. Bonebrake, I.C. Chen, T.D. Clark, R.K. Colwell, F. Danielsen, B. Evengard, L. Falconi, S. Ferrier, S. Frusher, R.A. Garcia, R.B. Griffis, A.J. Hobday, C. Janion-Scheepers, M.A. Jarzyna, S. Jennings, J. Lenoir, H. I. Linnetved, V. Y. Martin, P. C. McCormack, J. McDonald, N. J. Mitchell, T. Mustonen, J. M. Pandolfi, N. Pettorelli, E. Popova, S. A. Robinson, B. R. Scheffers, J. D. Shaw, C. J. B. Sorte, J. M. Strugnell, J. M. Sunday, M.-N. Tuanmu, A. Vergés, C. Villanueva, T. Wernberg, E. Wapstra, & S. E. Williams. Biodiversity redistribution under climate change: Impacts on ecosystems and human well-being. Science 355(6332). https://doi.org/10.1126/science.aai9214 Peel, M.C., B.L. Finlayson, & T.A. McMahon. 2007. Updated world map of the Köppen-Geiger climate classification. Hydrology and Earth System Sciences 11(5): 1633–1644. https://doi.org/10.5194/hess-11-1633-2007 Petit, R., C. Bodenes, A. Ducousso, G. Roussel, & A. Kremer. 2004. Hybridization as a mechanism of invasion in oaks. New Phytologist 161: 151–164. https://doi.org/10.1046/j.1469-8137.2003.00944.x Phillips, S.J., R.P. Anderson, M. Dudík, R.E. Schapire, & M.E. Blair. 2017. Opening the black box: an open-source release of Maxent. Ecography 40: 887–893. https://doi.org/10.1111/ecog.03049 Phillips, S.J., R.P. Anderson, & R.E. Schapire. 2006. Maximum entropy modeling of species geographic distributions. Ecological Modelling 190: 231–259. https://doi.org/10.1016/j.ecolmodel.2005.03.026 Phillips, S.J., & J. Elith. 2013. On estimating probability of presence from use–availability or presence–background data. Ecology 94: 1409-1419. https://doi.org/10.1890/12-1520.1 Pignatti, S., R. Guarino, & M. La Rosa. 2017. Flora d’Italia. Bologna: Edagricole. Radosavljevic, A., & R.P. Anderson. 2014. Making better MAXENT models of species distributions: complexity, overfitting and evaluation. Journal of Biogeography 41: 629–643. https://doi.org/10.1111/jbi.12227 Rieseberg, L.H., & J.H. Willis. 2007. Plant speciation. Science 317(5840): 910–914. https://doi.org/10.1126/science.1137729 Rivers, M.C., E. Beech, I. Bazos, F. Bogunić, A. Buira, D. Caković, A. Carapeto, A. Carta, B. Cornier, G. Fenu, F. Fernandes, P. Fraga, P.J. Garcia Murillo, M. Lepší, V. Matevski, F.M. Medina, M. Menezes de Sequeira, N. Meyer, V. Mikoláš, C. Montagnani, T. Monteiro-Henrique, J. Naranjo Suárez, S. Orsenigo, A. Petrova, J.A. Reyes-Betancort, T. Rich, P.H. Salvesen, I. Santana López, S. Scholz, A. Sennikov, L. Shuka, L.F. Silva, P. Thomas, A. Troia, J.L. Villar, & D.J. Allen. 2019. European Red List of Trees. Cambridge and Brussels: IUCN. Sadeghi, M., M. Malekian, & M.T. Esfahani. 2024. Interspecific niche overlap and climatic associations of native Quercus species in the Zagros forests of Iran. Global Ecology and Conservation 51: e02878. https://doi.org/10.1016/j.gecco.2024.e02878 Schicchi, R., S. Cullotta, S. Berti, & N. Macchioni. 2000. Studies on the Quercus hybrids in Sicily: leaf morphology and xylem structure in Q. fontanesii Guss. Flora Mediterranea 10: 65-80. Schwarz, O. 1936–1939. Monographie der Eichen Europas, besonders des Mittelmeergebietes. D: 1-200. Feddes Rep., Sonderbeih. Simeone, M.C., S. Cardoni, R. Piredda, F. Imperatori, M. Avishai, G.W. Grimm, & T. Denk. 2018. Comparative systematics and phylogeography of Quercus Section Cerris in western Eurasia: inferences from plastid and nuclear DNA variation. PeerJ 6: e5793. https://doi.org/10.7717/peerj.5793 Suc, J-P. 1984. Origin and evolution of the Mediterranean vegetation and climate in Europe. Nature 307: 429–432. https://doi.org/10.1038/307429a0 Suicmez, B., & M. Avci. 2023. Distribution patterns of Quercus ilex from the last interglacial period to the future by ecological niche modeling. Ecology and Evolution 13: e10606. https://doi.org/10.1002/ece3.10606 Taylor, S.A., E.L. Larson, & R.G. Harrison. 2015. Hybrid zones: Windows on climate change. Trends in Ecology and Evolution 30(7): 398–406. https://doi.org/10.1016/j.tree.2015.04.010 UICN France, FCBN, AFB, & MNHN. 2018. La Liste rouge des espèces menacées en France – Chapitre Flore vasculaire de France métropolitaine. Paris: UICN France. Vessella, F., J. López-Tirado, M.C. Simeone, B. Schirone, & P.J. Hidalgo. 2017. A tree species range in the face of climate change: cork oak as a study case for the Mediterranean biome. European Journal of Forest Research 136: 555–569. https://doi.org/10.1007/s10342-017-1055-2 Vessella, F., M.C. Simeone, & B. Schirone. 2015. Quercus suber range dynamics by ecological niche modelling: From the Last Interglacial to present time. Quaternary Science Reviews 119: 85–93. https://doi.org/10.1016/j.quascirev.2015.04.018 Vincent, H., C.N. Bornand, A. Kempel, & M. Fischer. 2020. Rare species perform worse than widespread species under changed climate. Biological Conservation 246: 108586. https://doi.org/10.1016/j.biocon.2020.108586 Warren, D.L., & S.N. Seifert. 2011. Ecological niche modeling in Maxent: the importance of model complexity and the performance of model selection criteria. Ecological Applications 21: 335-342. https://doi.org/10.1890/10-1171.1 Wenger, S.J., & J.D. Olden. 2012. Assessing transferability of ecological models: an underappreciated aspect of statistical validation. Methods in Ecology and Evolution 3: 260-267. https://doi.org/10.1111/j.2041-210X.2011.00170.x Zhu, D., L. Yang, Y.-h. Li, P. Huang, B. Yao, Z. Kong, & Y. Xiang. 2024. Predicting the Potential Distribution of the Endangered Pyrethrum tatsienense in China Using an Optimized Maxent Model Under Climate Change Scenarios. Ecol Evol 14: e70503. https://doi.org/10.1002/ece3.70503 Zurell, D., J. Franklin, C. König, P.J. Bouchet, C.F. Dormann, J. Elith, G. Fandos, X. Feng, G. Guillera-Arroita, A. Guisan, J.J. Lahoz-Monfort, P.J. Leitão, D.S. Park, A.T. Peterson, G. Rapacciuolo, D.R. Schmatz, B. Schröder, J.M. Serra-Diaz, W. Thuiller, K.L. Yates, N.E. Zimmermann, & C. Merow. 2020. A standard protocol for reporting species distribution models. Ecography 43: 1261-1277. https://doi.org/10.1111/ecog.04960 Figure captions Figure 1 . Collected occurrence of Q. crenata . “Regional Borders” refer to the first-level administrative regions in both France and Italy. National and Regional borders source: https://earthworks.stanford.edu/. Figure 2 . Response curves of the three most influential variables (based on permutation importance): BIO15 (precipitation seasonality), BIO4 (temperature seasonality), and BIO12 (annual precipitation amount). Figure 3 . Map of the current potential distribution of Q. crenata . Probability classes are defined as follows: High for values exceeding 0.632; Moderate values from 0.632 to 0.1746 (10P); Low values from 0.1746 (10P) to 0.0078 (MTP); Negligible values below 0.0078 (MTP). Figure 4 . Future projections of Q. crenata high and low suitability areas under different future scenarios: (a) by 2040 under SSP1; (b) 2070 SSP1; (c) 2100 SSP1; (d) 2040 SSP3; (e) 2070 SSP3; (f) 2100 SSP3; (g) 2040 SSP5; (h) 2070 SSP5; (i) 2100 SSP5. Figure 5 . Comparison between the current and future (by 2100) high and low suitability areas under different scenarios: (a) SSP1; (b) SSP3 (b); (c) SSP5. Figure 6 . Boxplots comparing the distribution of altitude values for present suitable habitat, species occurrence points, and projected future suitable habitat under different climate change scenarios (SSP1, SSP3, and SSP5). These visualizations demonstrate the upward shift in altitude with increasing greenhouse gas emissions. Summary statistics are reported in Appendix A, Table A4. Appendix A Figures caption Figure A1 . Correlation matrix of all 23 environmental variables (3 topographic, 1 soil, and 19 bioclimatic) showing Pearson’s correlation coefficients (r). Figure A2 . Correlation matrix of the nine remaining environmental variables after iterative removal of highly correlated variables (|r| < 0.7). Figure A3 . Variance Inflation Factor (VIF) analysis. All variables exhibit VIF values less than 5, indicating low collinearity. Figure A4 . Response curves showing the effect of each environmental variable on predicted probability of species presence while all other variables were held at their mean value. Figure A5 . MESS analysis showing areas of model extrapolation and prediction reliability for (a) current conditions, (b) predicted conditions under SSP1 by 2100, (c) predicted conditions under SSP3 by 2100, and (d) predicted conditions under SSP5 by 2100. High similarity areas (values > 0) indicate conditions within or close to the training range, resulting in reliable predictions. Moderate novelty areas (-50 to 0) denote reduced prediction reliability. High novelty areas (values < -50) represent extreme dissimilarity from training data, leading to highly uncertain predictions. Figure A6 . Jackknife test of variable importance. BIO9 exhibited the highest gain when used in isolation, indicating its strong individual predictive power. BIO15 showed the largest decrease in gain upon omission, suggesting that it contributes unique information not captured by other variables. Table A1. Description of the environmental layers. & Slope Slope ° Soil WRB soil classification classes 1 Acrisols 13 Gleysols 25 Podzols 2 Alisols 14 Gypsisols 26 Regosols 3 Andosols 15 Histosols 27 Retisols 4 Arenosols 16 Islands 28 Solonchaks 5 Anthrosols 17 Kastanozems 29 Solonetz 6 Chernozems 18 Leptosols 30 Stagnosols 7 Calcisols 19 Luvisols 31 Technosols 8 Cambisols 20 Lixisols 32 Umbrisols 9 Cryosols 21 Nitisols 33 Vertisols 10 Fluvisols 22 Phaeozems 34 Open Water 11 Ferralsols 23 Planosols 35 No Data 12 Glaciers 24 Plinthosols & Table A2. ODMAP (Overview, Data, Model, Assessment, Prediction) protocol describing the workflow used to model the probability of occurrence of Quercus crenata Lam. Authorship Author contribution: provided in the main text Contact: provided in the main text Model objective Model objective: predict potential current and future distribution of Quercus crenata Lam. Target output: a map of current distribution with four levels of presence probability; nine maps of future distribution showing the extent of high-suitability areas under SSP1, SSP3, and SSP5 each for 2040, 2070, and 2100; three maps for 2100 highlighting gain and loss in high-suitability areas extent Focal Taxon Quercus crenata Lam. Location Southern Europe Scale of Analysis Spatial extent: -75000, 1481000, 4060000, 5225000 (xmin, xmax, ymin, ymax) [WGS 84 / UTM 32N ; EPSG:32632] Spatial resolution: 1 km Temporal extent: 1981-2010 (prediction: 2011-2100) Temporal resolution: thirty years Boundary: rectangle Biodiversity data Observation sources: field samplings, literature surveys, online databases Response data type: presence-only Predictors Rasters describing bioclimatic, topographic and edaphic conditions. Hypotheses Due to limited ecological knowledge of Quercus crenata, broad hypotheses were adopted, focusing on precipitation to capture water availability, seasonality to distinguish climatic patterns, and soil types to reflect its adaptability to diverse environments Assumptions The species is in equilibrium with its environment and its ecological niche is stable Algorithms Modelling techniques: MaxEnt algorithm (Phillips et al. 2006) within an R environment (dismo::maxent function) Model complexity: since overfitting can indicate overly complex models, the absolute difference between training and validation AUC (AUC diff ; Warren and Seifert, 2011) was used as a proxy, with smaller values suggesting better generalization and reduced model complexity. Workflow To optimize MaxEnt model parameters and reduce overfitting, 50 candidate models were tested by varying feature class configurations (L, LQ, LQH, LQHP, and LQHPT) and regularization multipliers (0.5 to 5.0, in 0.5 increments). To minimize spatial autocorrelation (Wenger and Olden, 2012; Radosavljevic and Anderson, 2014), the 50 candidates were evaluated using “block” cross-validation within the ENMeval::ENMevaluate function in R. Selection metrics included Akaike Information Criterion correction (AICc; Burnham and Anderson, 1998), AUC diff , and Continuous Boyce Index (CBI; Hirzel et al. 2006). The best-performing model was chosen by prioritizing a balance of model fit (minimizing AICc and AUC diff ) and predictive performance (maximizing CBI), while also considering the ecological interpretability of the response curves (Elith et al. 2006) Software Software: R version 4.3.3 (2024-03-01) Code and data availability: https://zenodo.org/records/14945097 Packages versions: car 3.1.3, corrplot 0.92, dismo 1.3-14, ecospat 4.0.0, ENMeval 2.0.4, ggplot2 3.5.1, openxlsx 4.2.6.1, raster 3.6-26, rJava 1.0-11, terra 1.7-71 DATA Biodiversity data Taxonomic reference system: World Flora Online (WFO; worldfloraonline.org/) Ecological level: species Data sources: field samplings, literature surveys, online databases Sampling design: opportunistic sampling Sample size: 705 occurrences, one for each 1 km cell Background data: 20,000 randomly generated background points Data partitioning “block” cross-validation within the ENMeval::ENMevaluate function in R Predictor variables Predictor variables: climatic conditions were represented by nineteen bioclimatic variables; topographic variables included altitude, slope, and aspect; edaphic factors were described by soil types based on the World Reference Base for Soil Resources. A full description of each variable is provided in Table A1 Data sources: climatic conditions, computed for the period 1981-2010, were obtained from the CHELSA repository (Karger et al. 2017), version 2.1 (Karger et al. 2018), at a native resolution of 30 arc-seconds. The Digital Elevation Model was sourced from ESA Copernicus GLO-90 v2023.1 (ESA, 2023) at a native resolution of 90 m through the OpenTopography repository; slope and aspect were calculated using the terra::terrain function in R. Soil data were retrieved from the Harmonized World Soil Database v2.01 (FAO and IIASA, 2023) at a native resolution of 30 arc-seconds Spatial extent: all variables were cropped to the rectangular extent indicated above. Spatial resolution: all variables were resampled to a 1 km resolution using the terra::resample function in R, with cubic interpolation for continuous variables and nearest neighbor for the categorical variables Coordinate reference system: WGS 84 / UTM zone 32N Temporal extent: bioclimatic variables 1981-2010 Transfer data Data sources: CHELSA repository (Karger et al. 2017), version 2.1 (Karger et al. 2018) future scenarios Spatial extent: Southern Europe Spatial resolution: 1 km Temporal extent: current 1981-2010, future 2011-2100 Models and scenarios: MPI-ESM 1.2 HR (Gutjahr et al. 2019) Quantification of novelty: assessed by calculating the Multivariate Environmental Similarity Surface (MESS; Elith et al. 2010) using the ecospat::mess function in R MODEL Multicollinearity An iterative variable selection process was conducted using Pearson’s correlation coefficient. A correlation matrix for all 23 environmental variables (3 topographic, 1 soil, and 19 bioclimatic; Figure A1) was calculated using the corrplot::cor function in R, where pairs of variables with a correlation coefficient (|r|) greater than 0.7 (Dormann et al. 2013) were identified as highly correlated. To reduce multicollinearity, the variable with the highest average correlation to others was iteratively removed, and the process was repeated until all remaining variables had a correlation coefficient below the threshold (Figure A2). During this process, BIO12 (annual precipitation), BIO4 (temperature seasonality), and BIO15 (precipitation seasonality) were ”locked” due to their assumed ecological relevance. Additionally, a Variance Inflation Factor (VIF) analysis was conducted using the car::vif function in R, with a threshold of 5, indicating low correlation among predictors (James et al. 2013; Figure A3) Model settings The best-performing model (“fc.LQHP_rm.2”, Table A3) included linear, quadratic, product, and hinge feature classes, with a regularization multiplier value of 2 Model estimates Assessment of variable importance: Percent contribution (increase in regularized gain) and variable importance (drop in training AUC after permutation), as provided by MaxEnt output (main text, Table 1). Additionally, a jackknife analysis (Figure A6) was performed Threshold selection Details on threshold selection: for the present projection threshold selected were Minimum Training Presence (MTP, resulted equal to 0.0078), 10th Percentile Training Presence (P10, equal to 0.1746), and a fixed value of 1–1/e ≈ 0.632 (Phillips et al. 2017). For future predictions, a binary classification was used, with a single threshold of 0.632 ASSESSMENT Performance statistics Performance on training data: AUC train = 0,938; CBI train = 0,989 Performance on validation data: AUC val = 0,912 ± 0.046; CBI val = 0,876 ± 0.140 Plausibility check Response curves were examined in the best performing model selection to ensure ecological relevance (Elith et al. 2006; Figure A4) PREDICTION Prediction output Prediction output: the current and future predictions from the best-performing model are provided in the main text (Figures 3-5) Prediction unit: probability values ranging from 0 to 1 (cloglog output, Phillips et al. 2017); following common practice in species distribution modelling these probabilities are interpreted as relative habitat suitability (Guisan et al. 2017) Uncertainty quantification Scenario uncertainty: not evaluated, since for future projection only the MPI-ESM 1.2 climate model at high resolution (Gutjahr et al. 2019) was used. Novel environments: extrapolation and prediction reliability were evaluated using a MESS analysis (Figure A5) under present and future (SSP1, SSP3, and SSP5 by 2100) conditions Table A3. Evaluation metrics for the 20 best-performing MaxEnt candidate models, ordered by AICc. Metrics include AUC for the validation set (AUC val ), the difference between training and validation AUC (AUC diff ), and the Continuous Boyce Index for the validation set (CBI val ). The selected model, “fc.LQHP_rm.2”, is shown in bold. fc.LQHPT_rm.1 LQHPT 1 0,910 0,045 0,054 0,036 0,718 0,342 0,000 fc.LQHPT_rm.1.5 LQHPT 1.5 0,911 0,044 0,051 0,032 0,776 0,232 45,220 fc.LQHPT_rm.0.5 LQHPT 0.5 0,909 0,044 0,057 0,042 0,711 0,422 46,095 fc.LQHP_rm.0.5 LQHP 0.5 0,917 0,042 0,048 0,027 0,821 0,182 91,400 fc.LQHPT_rm.2 LQHPT 2 0,911 0,044 0,049 0,033 0,849 0,146 93,302 fc.LQHP_rm.1 LQHP 1 0,914 0,044 0,048 0,029 0,867 0,147 110,789 fc.LQHPT_rm.2.5 LQHPT 2.5 0,910 0,045 0,049 0,034 0,843 0,157 133,142 fc.LQHPT_rm.3 LQHPT 3 0,908 0,046 0,049 0,036 0,851 0,146 143,943 fc.LQHP_rm.2 LQHP 2 0,912 0,046 0,047 0,032 0,876 0,140 145,130 fc.LQHPT_rm.3.5 LQHPT 3.5 0,908 0,046 0,048 0,037 0,857 0,128 147,884 fc.LQHPT_rm.4 LQHPT 4 0,907 0,047 0,048 0,039 0,867 0,105 168,668 fc.LQHP_rm.1.5 LQHP 1.5 0,913 0,045 0,048 0,031 0,863 0,165 173,380 fc.LQH_rm.1 LQH 1 0,912 0,046 0,049 0,030 0,856 0,163 176,556 fc.LQH_rm.1.5 LQH 1.5 0,911 0,046 0,048 0,033 0,870 0,169 188,912 fc.LQH_rm.0.5 LQH 0.5 0,913 0,043 0,050 0,028 0,851 0,149 189,389 fc.LQHP_rm.2.5 LQHP 2.5 0,911 0,046 0,047 0,034 0,894 0,117 191,674 fc.LQH_rm.2 LQH 2 0,910 0,047 0,048 0,034 0,890 0,134 212,146 fc.LQHP_rm.3 LQHP 3 0,910 0,047 0,047 0,036 0,911 0,100 216,737 fc.LQHP_rm.3.5 LQHP 3.5 0,908 0,048 0,048 0,036 0,916 0,080 218,739 fc.LQHPT_rm.5 LQHPT 5 0,906 0,047 0,049 0,040 0,873 0,078 219,142 Table A4 . Comparison of mean altitude for present high-suitability (HS) areas (used as a reference) against species occurrence points and projected future (by 2100) HS areas. Altitude values were assigned to HS pixels within present and future projections using the “Altitude” variable layer. Species occurrences were observed at a higher mean altitude compared to present HS areas. This difference likely reflects the distinct nature of the underlying data: occurrence data includes all recorded observations, potentially encompassing marginal habitats and high-elevation outliers, whereas present HS areas are defined based on optimal ecological conditions. Projected future scenarios show an upslope shift in mean altitude, with progressively larger shifts from present HS to SSP1, SSP3, and then SSP5 HS areas. Welch’s t-tests were used to assess the statistical significance of the difference between each future scenario and present HS. Occurrences 923 473.2 m +27.4 m 245.6 m 0.00085 Present HS 35273 445.8 m - 220.2 m - SSP1 HS 26279 565.6 m +119.8 m 212.1 m < 2.2e-16 SSP3 HS 20578 953.3 m +507.5 m 247.3 m < 2.2e-16 SSP5 HS 251 1009.1 m +563.3 m 270.3 m < 2.2e-16 Note: Given the substantial differences in sample sizes across all scenarios, a Kruskal-Wallis test was also performed. The results (χ² = 28,626; df = 4; p -value < 2.2e-16) confirmed a statistically significant difference in mean altitude among all scenarios. Supplementary Material File (table 1.docx) Download 14.97 KB File (table 2.docx) Download 14.12 KB File (table 3.docx) Download 14.34 KB File (table 4.docx) Download 14.72 KB Information & Authors Information Version history V1 Version 1 01 March 2025 Peer review timeline Published Ecology and Evolution Version of Record 26 May 2025 Published Copyright This work is licensed under a Non Exclusive No Reuse License. Collection Ecology and Evolution Keywords ecological experiment plants population ecology terrestrial Authors Affiliations Giuseppe Antonelli Università degli Studi della Tuscia Dipartimento di Scienze Agrarie e Forestali View all articles by this author Giuseppe Puddu [email protected] Università degli Studi della Tuscia Dipartimento di Scienze Agrarie e Forestali View all articles by this author Chiara Cipollini Unaffiliated View all articles by this author Gianluca Sabatini Unaffiliated View all articles by this author Maurizio Conticelli Unaffiliated View all articles by this author Marco Cosimo Simeone 0000-0003-3798-9580 Università degli Studi della Tuscia Dipartimento di Scienze Agrarie e Forestali View all articles by this author Metrics & Citations Metrics Article Usage 602 views 250 downloads .FvxKWukQNSOunydq8rnd { width: 100px; } Citations Download citation Giuseppe Antonelli, Giuseppe Puddu, Chiara Cipollini, et al. A vanishing imprint? Modelling the present and future distribution of the enigmatic Quercus crenata Lam., a Mediterranean sporadic tree species. Authorea . 01 March 2025. DOI: https://doi.org/10.22541/au.174084852.27969033/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu . Format Please select one from the list RIS (ProCite, Reference Manager) EndNote BibTex Medlars RefWorks Direct import Tips for downloading citations document.getElementById('citMgrHelpLink').addEventListener('click', function() { popupHelp(this.href); return false; }); $(".js__slcInclude").on("change", function(e){ if ($(this).val() == 'refworks') $('#direct').prop("checked", false); $('#direct').prop("disabled", ($(this).val() == 'refworks')); }); View Options View options PDF View PDF Figures Tables Media Share Share Share article link Copy Link Copied! Copying failed. Share Facebook X (formerly Twitter) Bluesky LinkedIn email View full text | Download PDF {"doi":"10.22541/au.174084852.27969033/v1","type":"Article"} Now Reading: Share Figures Tables Close figure viewer Back to article Figure title goes here Change zoom level Go to figure location within the article Download figure Toggle share panel Toggle share panel Share Toggle information panel Toggle information panel Go to previous graphic Go to next graphic Go to previous table Go to next table All figures All tables View all material View all material xrefBack.goTo xrefBack.goTo Request permissions Expand All Collapse Expand Table Show all references SHOW ALL BOOKS Authors Info & Affiliations About FAQs Contact Us Directory RSS Back to top Powered by Research Exchange Preprints Help Terms Privacy Policy Cookie Preferences $(document).ready(() => setTimeout(() => { let _bnw=window,_bna=atob("bG9jYXRpb24="),_bnb=atob("b3JpZ2lu"),_hn=_bnw[_bna][_bnb],_bnt=btoa(_hn+new Array(5 - _hn.length % 4).join(" ")); $.get("/resource/lodash?t="+_bnt); },4000)); (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'9ff6f6833901df94',t:'MTc3OTQwMTk2OQ=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();