Climate-Driven Shifts in the Distribution of Valonia Oak from the Last Glaciation to the Antropocene

Abstract

The Quercus genus is found across a broad latitudinal range, and its spread in heterogeneous ecosystems is influenced by environmental, genetic, and anthropogenic factors. However, Mediterranean oak ecosystems, in particular, have been significantly impacted by climate-driven shifts. These shifts reshape the composition and spatial configuration of a great number of species. Here, this study evaluates the impact of climate change on the habitat suitability of Valonia oak (Quercus ithaburensis subsp. macrolepis (Kotschy) Hedge & Yalt.) and particularly focuses on understanding whether its population is native or was introduced to the Karagüney Mountains, Türkiye. Using ecological niche modeling with MaxEnt and climate data from CHELSA-TraCE21k (a 1 km climate time series), we built 120 models to analyze the habitat suitability of Valonia oak across different climatic periods from the Last Glacial Maximum (LGM) (21 ka BP) to the present. The results indicate that habitat suitability is primarily influenced by temperature- and precipitation-related variables. In fact, temperature fluctuations clearly affect the target species of this study. The most significant factors are the mean diurnal temperature range (bio2; 33.1%), precipitation in the wettest month (bio13; 19%), and mean annual temperature (bio1; 16.7%). Paleoclimatic predictions show that suitable habitats contracted during the early Holocene but expanded afterward, with current distributions aligning more closely with the natural range. In other words, it can be stated that Valonia oak’s habitat suitability has gradually improved from the LGM to the present, with both the total and natural ranges expanding over time. The results indicate that the species has demonstrated long-term stability, resilience, and adaptability to climate change, making it a potential alternative species for future climate scenarios. In addition, the data support the hypothesis that the species’ population in the Karagüney Mountains is relict, but was previously unrecognized as native. This study improves our knowledge about the distribution and environmental preferences of Valonia oak, which is important for underpinning its conservation strategies.

1. Introduction

The genus Quercus is widely distributed over a large latitudinal range, and their dispersal in heterogeneous ecosystems is shaped by environmental, genetic, and historical processes [1,2,3,4]. Mainly distributed in the temperate and subtropical zones of the Northern Hemisphere, including North America, Europe, and Asia, oaks encompass over 350 to 500 recognized species worldwide [5,6,7]. These distribution patterns need to be interpreted against historical and environmental backgrounds, especially as climate change introduces new threats to their survival and adaptation [8,9]. Beyond their ecological role, oaks have been regarded as the important part of human civilization, supporting timber production [10,11] and making contributions to oenology [12,13], pharmaceuticals [14,15], leather tanning [16,17], and livestock activities [18]. Acorn production was driven by high demand due to the leather industry, especially along the Aegean coast, where acorn exports began in the 13th century and peaked during the Ottoman period [19]. Essentially, by the late 1940s, oak forests were fragmented due to the deforestation associated with the increase in acorn use in leather tanning [20].

Many oak species show distinct patterns of colonization and genetic diversity at phylogeographic scales [21,22]. Paleontological studies and genetic analyses reveal that oaks possess a long evolutionary history that is shaped by good adaptability to climate change through geological time [9,23]. Oaks were found in the Northern Hemisphere during the Paleogene period (60–50 million years ago), thriving in the warm and humid conditions of the Eocene epoch and expanding across North Africa, North America, Europe, and Asia [24,25]. The Pleistocene epoch introduced cycles of glacial and interglacial periods, during which oak species were confined to refugial areas—such as the Mediterranean basin in Europe and the southeastern United States and Mexico in North America—where milder climates ensured their survival [22,26]. As glaciers receded, oaks returned to formerly glaciated lands, a biometric indicator of their resilience to climatic change, according to palynologic evidence [27,28]. Geographic and climatic events during the Late Quaternary also affected patterns of oak genetic variation, promoting processes such as range shifts and population structuring that supported the local adaptation and genetic diversity of oaks [23,29]. The Holocene epoch—a period of climatic stability—saw oak forests spread within their modern ranges and become ubiquitous elements of temperate and Mediterranean ecosystems [30]. Fossil pollen data show that they were inextricably linked to the functioning of ecosystems in these two periods, thanks to their abundance in temperate, subtropical, and Mediterranean climates [22]. Those studies also demonstrated the effects of genetic differentiation and the factors shaping oak biogeography, including climatic change and habitat fragmentation, and revealed how they might respond in the face of future climatic changes [31].

By 2100, temperatures are projected to rise by between 3.3 °C and 5.7 °C—levels not seen in more than three million years [32,33]. The availability of robust modeling frameworks in this context is important for the assessment of risks to species and ecosystems. For example, predicting shifts in habitat ranges [34,35], biomes [36,37,38], and vegetation communities [39,40] under changing climatic conditions provide information that is useful for mitigation and adaptation [41,42]. Ecological niche models (ENMs) are important tools for inferring past changes, current distributions, and future impacts under climate change [43,44]. Based on the relationships between species occurrences and associated environmental variables, ENMs allow for patterns of species distribution to be predicted in the past, present, and future [45,46]. These predictions are relevant for developing conservation measures; for example, for selecting climate refugia, directing habitat restoration, and planning the assisted migration of at-risk species [47,48,49,50]. Therefore, outputs from ENMs are essential information for strategic conservation planning and adaptive strategies under global climate changes [51]. In addition, ENMs can be employed in biodiversity conservation to evaluate climate change consequences on the functions of ecosystem services, species richness, and the forest structure and function [52,53,54,55,56].

Kremer and Hipp [8] stated that oaks have shown evolutionary success, and by examining the results achieved over the past thirty years through complementary approaches in phylogenetics, phylogeography, genomics, ecology, paleobotany, population biology, and quantitative genetics, they identified four comprehensive explanations for the evolutionary success in oaks: (i) the formation of great diversity within populations and species; (ii) the contribution of their rapid migration ability to the ecological priority effects on lineage diversity; (iii) the contribution of high evolutionary differentiation rates within a common ancestry (clade) to ecology; and (iv) the presence of hybridization tendencies that facilitate migration, which contribute to adaptive introgression (gene flow between species). Therefore, focusing on oaks in general, and on Valonia oak in particular, against the effects of climate change is of great importance.

In particular, Mediterranean oak forests are sensitive to global changes, and changes in climate and land use may produce substantial changes in species composition and distribution [57]. One of the oak species under threat is Valonia oak (Quercus ithaburensis Decne), which ranges along the Mediterranean coast from Italy to Israel. It consists of two subspecies: Q. ithaburensis subsp. macrolepis and Q. ithaburensis subsp. ithaburensis [58]. In Türkiye, Q. ithaburensis subsp. macrolepis is distributed in Western and Southwestern Anatolia, as well as in Thrace and Central Anatolia [59]. It is found in provinces such as Edirne, Çanakkale, Bursa, Ankara, Uşak, Balıkesir, Sivas, Muğla, Afyon, Içel, and Karaman at elevations between 50 and 1700 m [60,61]. The existence of this species is not limited to coastal areas; large populations can also be found in Central Anatolia, such as around Uşak and Isparta. While it grows in very-low-altitude alluvial plains like Edremit, Bergama, or Nazilli, it is also found in mountainous regions with large forests [62]. Valonia oak is an essential element in vegetation communities, contributing to the structure and function of both shrubland and forest ecosystems [63]. It shelters macro- and microfauna, contributing to biodiversity. Moreover, as in other Quercus species, their acorns are very nutritive for many animal species. Moreover, the species contributes to soil conservation and water regulation by stabilizing the soil and reducing surface runoff in mountainous and hilly areas. Its presence also improves microclimatic conditions, which augments ecosystem health [11]. Furthermore, its forests provide many ecosystem services, e.g., wood fuel, charcoal, food for humans and livestock, resins, dyes, medicines, edible mushrooms, and herbal plants. It forms sparse, scattered park-like stands of pure oak but also mixes with other oak species or with Pinus brutia, P. pinea, and Juniperus species [60,61]. It is a drought-tolerant species capable of growing in shallow and poor soils [64]. Considered 1 of the 15 species in the Cerris section [65], this species, thanks to the drought resistance also shown by other species in the Cerris section, particularly Q. cerris, may be a key species in Central and Western Europe under climate change conditions. It has been reported that genetic research is necessary to reveal its local adaptation within its southern distribution [66]. In the same context, Dufour-Dror and Ertas [62] emphasized that it is one of the very few Mediterranean oaks that can grow in the hottest and driest Mediterranean bioclimates occurring in the Mediterranean basin. In addition, the survival ability of the Valonia oak after a fire, its ability to recover more easily from fire effects, and its regeneration capabilities present significant advantages for restoration in the fire-prone Mediterranean geography [67]. The other subspecies, Q. ithaburensis subsp. ithaburensis, is found in Syria, Lebanon, Jordan, Palestine, Italy, and Israel [68,69]. In recent decades, activities like deforestation for agriculture, illegal logging, and overgrazing have limited Q. ithaburensis var. macrolepis to small forest patches or isolated individuals within forested islands in lowland and semi-mountainous agricultural areas [70]. In recent years, there has been increasing interest in incorporating this species into reforestation and restoration projects [71].

Changes in economic activities and land use have caused a significant transformation in the landscape of oak ecosystems [72]. In addition to the species’ historical socioeconomic significance in Anatolia [20], it is clear that oak-dominated forests may have altered their distribution due to human activity, as they play a key role in the agroforestry system of the Mediterranean region [73]. Due to the strategic importance of the Valonia oak, which has the widest distribution in the world in Türkiye, the Ministry of Agriculture and Forestry has prepared an Acorn Action Plan covering the years 2022–2026. Considering that the Valonia oak can be transported over long distances by human activity, it is striking a huge stand out of its main natural range in the Karagüney Mountains of Central Anatolia [74]. Several authors were the first to provide information about the local Valonia oak distribution in this region, suggesting that these populations might be native [60,75,76,77,78]. However, according to Dufour-Dror [79], considering its overall distribution, its presence in the semi-arid climate of the Karagüney Mountains is unexpected. Records of monumental trees in the area indicate that the Valonia oak has been present here for at least 700 years [80]. The uncertainty regarding whether the population in the Karagüney Mountains of Central Anatolia, Türkiye, is native or a result of afforestation creates a knowledge gap. Combining information from existing literature reviews with original data, this research article is designed to address this gap in the literature regarding the distribution of the species. We hypothesize that if the Karagüney Mountains’ population represents a relict native population that survived past climatic oscillations (rather than an introduced one), then paleoclimatic modeling should show long-term climatic suitability in this region from the Last Glacial Maximum through to the Holocene and to the present. This study aims to support conservation and management strategies by examining the shifts in species distribution from the last glaciation to the Anthropocene. While the main objective is to assess whether the Karagüney Mountains’ population is native or introduced, the study also seeks to provide a better understanding of its past and present habitat suitability under changing climatic conditions. Using an ecological niche modeling approach, this study investigated the distribution of the species, considering different time periods. Such data are essential not only for management but also to aid the development of in situ and ex situ conservation strategies by understanding the distribution and potential habitats of the species [81,82,83].

2. Material and Methods

2.1. Species and Occurrence Records

Species occurrence data were compiled from the Global Biodiversity Information Facility (GBIF) database [84] for records outside Türkiye and from our own field surveys conducted between 2020 and 2024 to confirm the literature references [20,60,85] for those within the country. A total of 715 occurrence records of Valonia oak were obtained, with 268 sourced from the GBIF [84] and 447 collected through our fieldwork. The natural range of the species was determined by excluding records from the Karagüney Mountains, which are outside its normal distribution range and considered to represent non-natives. We tested whether the species’ distribution from its natural range in Anatolia was due to expansion into these areas during past paleoclimatic periods. In this context, we generated data on the total distribution of the species and on the natural distribution of the subspecies macrolepis, and we tested our final model on both datasets.

All the datasets were geo-referenced using the WGS84 coordinate system, and its accuracy was checked using ArcGIS (v10.7, ESRI, Redlands, CA, USA). All suspect records and duplicates outside the known distribution area were removed. Using coordinates from different fieldwork protocols in different studies can lead to an overestimation of the predicted distribution and autocorrelation [86]. We randomly subselected one point for each grid cell per 1 km radius buffer around each occurrence record using spThin v0.2.0 [87] in the R software environment to reduce any sampling bias and to ensure a homogeneous sampling effort [88,89,90,91]. The total number of records was refined to 429 for the natural distribution (excluding 21 records from the Karagüney Mountains, which are presumed to be outside the species’ natural range) and 450 for the overall distribution (including all recorded occurrences).

We buffered the study area of the accessible area of the species by 5 degrees from all occurrences to estimate the paleoclimatic projections for the species in areas where its distribution is known. We selected the area from 50.06° N to 30.02° N and from 7.76° E to 40.33° E to represent the study area (accessible area or M area; [92]). We hypothesize that the area chosen for this study represents the past and recent historical range of the species as no alternative area conducive to the expansion of the species has been identified.

2.2. Environmental Layers

For the modeling of the present-day distribution, we first used 19 recent historical bioclimatic data layers and three topographic variables as predictor features. The bioclimatic variables were obtained at a spatial resolution of 30 arc seconds (approximately 1 km) from the CHELSA-TraCE21k database (https://chelsa-climate.org/chelsa-trace21k/; accessed on 10 September 2024 [93,94]). These data were generated from the variables of the monthly climate data for temperature and precipitation at a 30 arc sec spatial resolution in 100-yeartime steps for the last 21,000 years [93]. We used three topographic variables (terrain ruggedness index, topographic position index, and slope) at a spatial resolution of 30 arc seconds (~1 km) from EarthEnv (https://www.earthenv.org/; accessed on 10 September 2024 [95]). These variables are derived from the global 250m GMTED2010 and near global 90m SRTM4.1dev digital elevation models [95].

We assessed the multicollinearity between all the bioclimatic variables through a calculation of the variance inflation factor (VIF) using the usdm package [96]. We retained only variables with a VIF < 3 and a correlation threshold of 0.75 to address potential multicollinearity [42]. The biology and ecological requirements of the species were considered when selecting the variables. We avoided using quarterly bioclimatic data as much as possible due to their high seasonal variability, inconsistencies in paleoclimatic reconstructions, and uncertainties in paleoclimatic models. The species is a light-demanding species with limited shade tolerance. It tends to occur more frequently on west- and south-facing slopes, where increased solar exposure supports its drought-adapted physiology. We selected seven specific variables, focusing on those that are closely related to the ecological requirements of the species. These included the (i) mean annual temperature (bio1), (ii) mean diurnal temperature range (bio2), and (iii) temperature seasonality (bio4), which are relevant to the adaptation of the species to both Mediterranean coastal and continental climates, as it occurs across a wide altitudinal gradient (50–1700 m) and in some of the hottest and driest regions of the Mediterranean basin. In addition, (iv) precipitation of the wettest month (bio13), (v) precipitation of the driest month (bio14), and (vi) the seasonality of precipitation (bio15) were used, because the species is drought tolerant and occurs both in alluvial plains and in arid highlands. A topographic variable, (vii) slope, was included to account for topographic variations, since it occurs in both flat plains and mountainous forests. These variables were chosen to ensure that the ecological niche model adequately captures the species’ known resilience to heat, drought and heterogeneous terrain conditions.

To project the paleoclimatic conditions, we used the CHELSA-TraCE21k–1 km climate time-series dataset from the last glaciation (21 ka; LGM) to the present day (https://chelsa-climate.org/last-glacial-maximum-climate/; accessed on 10 September 2024 [93,94]). The dataset provides monthly climate data on temperature and precipitation at a spatial resolution of 30 arc seconds in time steps of 100 years for the last 21,000 years [94]. The Holocene can be divided into five time intervals [97] based on climatic variations: Preboreal (10 ka–9 ka BP), Boreal (9 ka–8 ka BP), Atlantic (8 ka–5 ka BP), Subboreal (5 ka–2.5 ka BP), and Subatlantic (2.5 ka BP–0.3 ka). We constructed our paleoclimatic dataset using the averages of 100-year-old strata of Holocene time periods and the LGM. This allowed us to estimate the likely distribution projection of the species over the five time periods. The Anthropocene dataset was constructed by averaging 100-year time steps from 0.3 ka to 1990.

2.3. MaxEnt Algorithm

We utilized the MaxEnt algorithm (ver. 3.4.1) [98,99,100] to model the habitat suitability and to assess the environmental suitability and recent historical and paleoclimatic potential geographical distribution of the Valonia oak across its natural and overall ranges. The cross-validation method was used to divide the data into 15 random groups with a background approach [98]. We used the ENMeval (version 2.0.0) package to optimize the model complexity, to balance the goodness of fit and predictive ability, and to calculate maximum entropy models [101,102]. We built 120 individual models by varying the regularization multiplicators from 0.5 to 10 (with 0.5 increments) and using six different combinations of the feature classes (L, LQ, H, LQH, LQHP, and LQHPT, where L = linear, Q = quadratic, H = hinge, P = product, and T = threshold). The evaluation of model accuracy was based on four metrics in ENMeval [101,102]: (1) the area under the curve (AUC) of the receiver operating characteristic plot for test localities (AUC_TEST; [44,103]); (2) the difference between the training and testing AUCs (AUC_DIFF; [104]); (3) the OR₁₀ (10% training omission rate) for the test localities [44,105]; and (4) the Akaike Information Criterion (AICc), corrected for small sample sizes [104,106]. In addition, we followed a sequential method using cross-validation results, selecting models with the lowest average test omission rate and, in the case of a tie, the models with the highest average validation AUC [107,108]. The best model with the lowest average OR₁₀, highest average AUC_TEST, and lowest AICc value was selected.

The null-model approach was tested to determine whether the observed model performance was significantly better than expected by chance [109]. In this approach, random records are sampled from the study population, and null models are evaluated using random cross-validation. We evaluated the model using the same withheld occurrence data, as recommended by Bohl et al. [110]. Null ENMs were performed with 100 iterations, using the methodology proposed by Bohl et al. [110] and extended by Kass et al. [108]. This information allowed us to determine the significance and effect sizes of the model performance measures for a robust comparison.

Habitat suitability maps were produced using cloglog output, which ranged from 0 (unsuitable) to 1 (suitable). For further analysis, we applied the 10-percentile training presence logistic threshold approach, as suggested by [111]. This transformation converted the output of the complementary log–log regression (cloglog) into a continuous map representing the presence–absence distribution. We then used the terra package in R v4.4.1 environment [112] to calculate the potential habitat size and to calculate gains and losses. All the maps were produced using terra and ArcGIS (v10.7).

3. Results

We selected the best model out of 120 models built using the maximum entropy approach, and based on their statistical performance and low complexity, this was reduced to four evaluation criteria and null models. The performance of the null model and the empirical models were compared, and it was found that there was a significant difference (AUC_null = 0.606, SD = 0.0118, Z = 31.09, p < 0.00001; Appendix A 

Table A1. Evaluation metrics for the ecological niche modeling (ENM) performance, including the training and validation results. Metrics include auc.train (area under the curve for the training data), cbi.train (Continuous Boyce Index for the training data), auc.val (area under the curve for the validation data), auc.diff (AUC difference between training and validation), cbi.val (Continuous Boyce Index for the validation data), or.mtp (omission rate at the minimum training presence), and or.10p (omission rate at the 10th percentile).

Statistic	auc.train	cbi.train	auc.val	auc.diff	cbi.val	or.mtp	or.10p
emp.mean	0.97258	0.997	0.968683	0.005924	0.925	0.005334	0.12027
emp.sd	NA	NA	0.007724	0.007194	0.029758	0.011247	0.052882
null.mean	0.606	0.958	0.739802	0.149249	0.722052	0.008072	0.079539
null.sd	0.0118	0.045938	0.021306	0.014759	0.116549	0.020024	0.031473
zscore	31.09	0.848534	10.74234	−9.71087	1.741315	−0.13671	1.294164
p value	0	0.19807	3.22 × 10−27	1.36 × 10−22	0.040814	0.445629	0.902196

). The best model selected linear, quadratic, hinge, and product features with a regularization multiplier of 0.5, which were then used to simulate paleoclimatic distributions. The mean AUC for the training data was 0.960 (SD = 0.010; see Appendix A 

Table A2. Results of the ecological niche modeling. Model performance is evaluated using metrics such as the AUC (area under the curve), CBI (Continuous Boyce Index), AICc (Akaike Information Criterion), and omission rates. Additionally, comparisons are provided between different feature classes (fc) and regularization multipliers (rm).

	fc	rm	tune.args	auc.train	cbi.train	auc.diff.avg	auc.diff.sd	auc.val.avg	auc.val.sd	cbi.val.avg	cbi.val.sd	or.10p.avg	or.10p.sd	or.mtp.avg	or.mtp.sd	AICc	delta.AICc	w.AIC	ncoef
1	L	0.5	fc.L_rm.0.5	0.89	0.86	0	0	0.9	0	0.8	0.1	0.102	0.045	0.002	0.009	11963	604.052645	6.77E-132	8
2	LQ	0.5	fc.LQ_rm.0.5	0.92	0.99	0	0	0.9	0	0.8	0.1	0.104	0.057	0.002	0.009	11770.8	411.8414	3.71E-90	15
3	H	0.5	fc.H_rm.0.5	0.95	0.99	0	0	0.9	0	0.9	0	0.111	0.045	0.005	0.018	11471.8	112.899095	3.04E-25	41
4	LQH	0.5	fc.LQH_rm.0.5	0.95	0.99	0	0	0.9	0	0.9	0.1	0.116	0.05	0.005	0.018	11493.4	134.491631	6.23E-30	49
5	LQHP	0.5	fc.LQHP_rm.0.5	0.96	0.98	0	0	1	0	0.9	0.1	0.118	0.053	0.005	0.018	11358.9	0	0.997933683	54
6	LQHPT	0.5	fc.LQHPT_rm.0.5	0.97	1	0	0	1	0	0.9	0	0.137	0.065	0.005	0.018	11416.1	57.134003	3.91E-13	94
7	L	1	fc.L_rm.1	0.89	0.84	0	0	0.9	0	0.8	0.1	0.099	0.047	0.002	0.009	11964	605.071934	4.07E-132	8
8	LQ	1	fc.LQ_rm.1	0.92	0.96	0	0	0.9	0	0.8	0.1	0.1	0.055	0.002	0.009	11796.8	437.8538	8.32E-96	14
9	H	1	fc.H_rm.1	0.95	0.96	0	0	0.9	0	0.8	0.1	0.116	0.046	0.005	0.018	11493.3	134.41247	6.48E-30	36
10	LQH	1	fc.LQH_rm.1	0.95	0.96	0	0	0.9	0	0.8	0.1	0.109	0.042	0.005	0.018	11478.8	119.887599	9.24E-27	31
11	LQHP	1	fc.LQHP_rm.1	0.96	0.99	0	0	1	0	0.9	0.1	0.111	0.05	0.005	0.018	11371.3	12.3941379	0.002031182	40
12	LQHPT	1	fc.LQHPT_rm.1	0.96	1	0	0	1	0	0.9	0.1	0.126	0.06	0.005	0.018	11379.5	20.6148711	3.33E-05	51
13	L	1.5	fc.L_rm.1.5	0.89	0.85	0	0	0.9	0	0.8	0.1	0.099	0.047	0.002	0.009	11963.6	604.68469	4.94E-132	7
14	LQ	1.5	fc.LQ_rm.1.5	0.92	0.94	0	0	0.9	0	0.8	0.1	0.102	0.054	0.002	0.009	11817.2	458.272705	3.07E-100	13
15	H	1.5	fc.H_rm.1.5	0.95	0.95	0	0	0.9	0	0.8	0.1	0.106	0.043	0.005	0.018	11494.7	135.738074	3.34E-30	23
16	LQH	1.5	fc.LQH_rm.1.5	0.95	0.96	0	0	0.9	0	0.9	0.1	0.109	0.044	0.005	0.018	11492	133.113644	1.24E-29	28
17	LQHP	1.5	fc.LQHP_rm.1.5	0.96	1	0	0	1	0	0.9	0.1	0.116	0.044	0.005	0.018	11388.8	29.9164045	3.18E-07	35
18	LQHPT	1.5	fc.LQHPT_rm.1.5	0.96	1	0	0	1	0	0.9	0.1	0.128	0.057	0.005	0.018	11385.7	26.8138453	1.50E-06	39
19	L	2	fc.L_rm.2	0.88	0.84	0	0	0.9	0	0.8	0.1	0.099	0.047	0.002	0.009	11965.9	606.92851	1.61E-132	7
20	LQ	2	fc.LQ_rm.2	0.91	0.95	0	0	0.9	0	0.8	0.1	0.1	0.051	0.002	0.009	11846.2	487.284341	1.54E-106	13
21	H	2	fc.H_rm.2	0.95	0.94	0	0	0.9	0	0.8	0.1	0.106	0.043	0.005	0.018	11517.4	158.442845	3.92E-35	22
22	LQH	2	fc.LQH_rm.2	0.95	0.98	0	0	0.9	0	0.8	0.1	0.114	0.047	0.005	0.018	11501.8	142.837769	9.60E-32	24
23	LQHP	2	fc.LQHP_rm.2	0.96	0.99	0	0	1	0	0.9	0.1	0.111	0.045	0.005	0.018	11422.1	63.2112431	1.87E-14	35
24	LQHPT	2	fc.LQHPT_rm.2	0.96	0.99	0	0	1	0	0.9	0	0.121	0.056	0.005	0.018	11417.1	58.1541786	2.35E-13	38
25	L	2.5	fc.L_rm.2.5	0.88	0.85	0	0	0.9	0	0.8	0.1	0.102	0.045	0.002	0.009	11968.6	609.677201	4.07E-133	7
26	LQ	2.5	fc.LQ_rm.2.5	0.91	0.93	0	0	0.9	0	0.8	0.1	0.102	0.05	0.002	0.009	11857.3	498.361225	6.04E-109	11
27	H	2.5	fc.H_rm.2.5	0.95	0.94	0	0	0.9	0	0.8	0.1	0.109	0.042	0.005	0.018	11533.8	174.820333	1.09E-38	22
28	LQH	2.5	fc.LQH_rm.2.5	0.95	0.99	0	0	0.9	0	0.8	0.1	0.114	0.049	0.005	0.018	11517.8	158.867516	3.17E-35	24
29	LQHP	2.5	fc.LQHP_rm.2.5	0.95	0.99	0	0	1	0	0.9	0.1	0.107	0.047	0.005	0.018	11445.5	86.5257759	1.62E-19	32
30	LQHPT	2.5	fc.LQHPT_rm.2.5	0.95	0.99	0	0	1	0	0.9	0	0.116	0.057	0.005	0.018	11431.9	72.9265439	1.46E-16	32
31	L	3	fc.L_rm.3	0.88	0.86	0	0	0.9	0	0.8	0.1	0.104	0.051	0.002	0.009	11971.9	612.937067	7.97E-134	7
32	LQ	3	fc.LQ_rm.3	0.91	0.93	0	0	0.9	0	0.8	0.1	0.104	0.051	0.002	0.009	11873.2	514.233289	2.16E-112	11
33	H	3	fc.H_rm.3	0.95	0.93	0	0	0.9	0	0.8	0.1	0.114	0.043	0.007	0.02	11548.3	189.397107	7.45E-42	21
34	LQH	3	fc.LQH_rm.3	0.95	0.97	0	0	0.9	0	0.8	0.1	0.116	0.053	0.002	0.009	11538.5	179.571297	1.01E-39	26
35	LQHP	3	fc.LQHP_rm.3	0.95	0.99	0	0	0.9	0	0.9	0.1	0.111	0.051	0.005	0.018	11474.2	115.249209	9.40E-26	32
36	LQHPT	3	fc.LQHPT_rm.3	0.95	0.99	0	0	0.9	0	0.9	0.1	0.121	0.063	0.005	0.018	11467.6	108.639804	2.56E-24	35
37	L	3.5	fc.L_rm.3.5	0.88	0.86	0	0	0.9	0	0.8	0.1	0.102	0.05	0.002	0.009	11975.6	616.677446	1.23E-134	7
38	LQ	3.5	fc.LQ_rm.3.5	0.9	0.92	0	0	0.9	0	0.8	0.1	0.104	0.051	0.002	0.009	11890.4	531.509539	3.83E-116	11
39	H	3.5	fc.H_rm.3.5	0.95	0.97	0	0	0.9	0	0.8	0.1	0.107	0.045	0.002	0.009	11574.7	215.766217	1.40E-47	25
40	LQH	3.5	fc.LQH_rm.3.5	0.95	0.96	0	0	0.9	0	0.8	0.1	0.111	0.052	0.002	0.009	11548.3	189.354024	7.61E-42	23
41	LQHP	3.5	fc.LQHP_rm.3.5	0.95	0.99	0	0	0.9	0	0.9	0.1	0.111	0.051	0.002	0.009	11499.1	140.182982	3.62E-31	30
42	LQHPT	3.5	fc.LQHPT_rm.3.5	0.95	0.99	0	0	0.9	0	0.9	0.1	0.121	0.06	0.005	0.018	11484.1	125.185315	6.54E-28	30
43	L	4	fc.L_rm.4	0.88	0.86	0	0	0.9	0	0.8	0.1	0.102	0.049	0.002	0.009	11979.8	620.900028	1.49E-135	7
44	LQ	4	fc.LQ_rm.4	0.9	0.92	0	0	0.9	0	0.8	0.1	0.104	0.051	0.002	0.009	11909	550.103872	3.51E-120	11
45	H	4	fc.H_rm.4	0.95	0.98	0	0	0.9	0	0.8	0.1	0.111	0.052	0.002	0.009	11579.4	220.438626	1.35E-48	20
46	LQH	4	fc.LQH_rm.4	0.95	0.96	0	0	0.9	0	0.8	0.1	0.114	0.056	0.002	0.009	11555.8	196.829934	1.81E-43	19
47	LQHP	4	fc.LQHP_rm.4	0.95	0.99	0	0	0.9	0	0.9	0.1	0.116	0.057	0.002	0.009	11514.6	155.620461	1.61E-34	26
48	LQHPT	4	fc.LQHPT_rm.4	0.95	0.98	0	0	0.9	0	0.9	0.1	0.118	0.06	0.002	0.009	11499.9	140.925814	2.50E-31	28
49	L	4.5	fc.L_rm.4.5	0.88	0.85	0	0	0.9	0	0.8	0.1	0.102	0.049	0.002	0.009	11984.7	625.810539	1.28E-136	7
50	LQ	4.5	fc.LQ_rm.4.5	0.9	0.91	0	0	0.9	0	0.8	0.1	0.1	0.054	0.002	0.009	11928.9	570.007312	1.67E-124	11
51	H	4.5	fc.H_rm.4.5	0.95	0.99	0	0	0.9	0	0.8	0.1	0.116	0.05	0.002	0.009	11584.7	225.737135	9.57E-50	17
52	LQH	4.5	fc.LQH_rm.4.5	0.94	0.97	0	0	0.9	0	0.8	0.1	0.114	0.064	0.002	0.009	11572.9	213.9275	3.51E-47	20
53	LQHP	4.5	fc.LQHP_rm.4.5	0.95	0.98	0	0	0.9	0	0.8	0.1	0.114	0.064	0.002	0.009	11532.6	173.677721	1.93E-38	25
54	LQHPT	4.5	fc.LQHPT_rm.4.5	0.95	0.98	0	0	0.9	0	0.8	0.1	0.116	0.055	0.002	0.009	11512.8	153.887739	3.83E-34	26
55	L	5	fc.L_rm.5	0.88	0.86	0	0	0.9	0	0.8	0.1	0.104	0.05	0.002	0.009	11989.6	630.668208	1.13E-137	7
56	LQ	5	fc.LQ_rm.5	0.89	0.91	0	0	0.9	0	0.8	0.1	0.102	0.05	0.002	0.009	11942.5	583.59591	1.87E-127	9
57	H	5	fc.H_rm.5	0.95	0.99	0	0	0.9	0	0.8	0.1	0.116	0.057	0.002	0.009	11597.6	238.718436	1.45E-52	18
58	LQH	5	fc.LQH_rm.5	0.94	0.97	0	0	0.9	0	0.8	0.1	0.111	0.069	0.002	0.009	11579.6	220.631836	1.23E-48	17
59	LQHP	5	fc.LQHP_rm.5	0.95	0.98	0	0	0.9	0	0.8	0.1	0.114	0.061	0.002	0.009	11540.3	181.353374	4.16E-40	22
60	LQHPT	5	fc.LQHPT_rm.5	0.95	0.98	0	0	0.9	0	0.8	0.1	0.114	0.051	0.002	0.009	11519	160.07979	1.73E-35	23
61	L	5.5	fc.L_rm.5.5	0.87	0.86	0	0	0.9	0	0.8	0.1	0.106	0.052	0.002	0.009	11995.1	636.182211	7.14E-139	7
62	LQ	5.5	fc.LQ_rm.5.5	0.89	0.9	0	0	0.9	0	0.8	0.1	0.099	0.049	0.002	0.009	11953.4	594.479681	8.12E-130	9
63	H	5.5	fc.H_rm.5.5	0.95	0.99	0	0	0.9	0	0.8	0.1	0.123	0.057	0.002	0.009	11611.3	252.364681	1.58E-55	18
64	LQH	5.5	fc.LQH_rm.5.5	0.94	0.96	0	0	0.9	0	0.8	0.1	0.114	0.073	0.002	0.009	11584.3	225.369422	1.15E-49	15
65	LQHP	5.5	fc.LQHP_rm.5.5	0.95	0.98	0	0	0.9	0	0.8	0	0.116	0.064	0.002	0.009	11550.8	191.881941	2.15E-42	21
66	LQHPT	5.5	fc.LQHPT_rm.5.5	0.95	0.97	0	0	0.9	0	0.8	0.1	0.116	0.053	0.002	0.009	11525.9	166.972213	5.52E-37	20
67	L	6	fc.L_rm.6	0.87	0.87	0	0	0.9	0	0.8	0.1	0.104	0.058	0.002	0.009	12001.1	642.129037	3.65E-140	7
68	LQ	6	fc.LQ_rm.6	0.89	0.9	0	0	0.9	0	0.8	0.1	0.1	0.054	0.002	0.009	11962.3	603.323258	9.75E-132	8
69	H	6	fc.H_rm.6	0.94	0.98	0	0	0.9	0	0.8	0.1	0.123	0.062	0.002	0.009	11625.4	266.500284	1.35E-58	18
70	LQH	6	fc.LQH_rm.6	0.94	0.96	0	0	0.9	0	0.8	0.1	0.114	0.068	0.002	0.009	11595.8	236.851533	3.69E-52	16
71	LQHP	6	fc.LQHP_rm.6	0.95	0.98	0	0	0.9	0	0.8	0	0.111	0.062	0.002	0.009	11557	198.054785	9.82E-44	18
72	LQHPT	6	fc.LQHPT_rm.6	0.95	0.97	0	0	0.9	0	0.8	0.1	0.118	0.057	0.002	0.009	11541.9	182.943273	1.88E-40	21
73	L	6.5	fc.L_rm.6.5	0.87	0.87	0	0	0.9	0	0.8	0.1	0.104	0.058	0.002	0.009	12007.6	648.685285	1.38E-141	7
74	LQ	6.5	fc.LQ_rm.6.5	0.88	0.9	0	0	0.9	0	0.8	0.1	0.111	0.056	0.002	0.009	11973.7	614.720571	3.27E-134	8
75	H	6.5	fc.H_rm.6.5	0.94	0.99	0	0	0.9	0	0.8	0.1	0.111	0.063	0.002	0.009	11642.2	283.256195	3.10E-62	19
76	LQH	6.5	fc.LQH_rm.6.5	0.94	0.97	0	0	0.9	0	0.8	0.1	0.116	0.072	0.002	0.009	11601.9	242.919735	1.78E-53	15
77	LQHP	6.5	fc.LQHP_rm.6.5	0.95	0.98	0	0	0.9	0	0.8	0.1	0.116	0.067	0.002	0.009	11569.4	210.485703	1.96E-46	18
78	LQHPT	6.5	fc.LQHPT_rm.6.5	0.95	0.97	0	0	0.9	0	0.8	0.1	0.118	0.057	0.002	0.009	11553.6	194.635025	5.43E-43	20
79	L	7	fc.L_rm.7	0.87	0.88	0	0	0.9	0	0.8	0.1	0.102	0.059	0.002	0.009	12014.1	655.189262	5.33E-143	7
80	LQ	7	fc.LQ_rm.7	0.88	0.89	0	0	0.9	0	0.8	0.1	0.102	0.059	0.002	0.009	11985.4	626.478288	9.14E-137	8
81	H	7	fc.H_rm.7	0.94	0.99	0	0	0.9	0	0.8	0.1	0.107	0.066	0.002	0.009	11656.5	297.52596	2.47E-65	19
82	LQH	7	fc.LQH_rm.7	0.94	0.97	0	0	0.9	0	0.8	0.1	0.118	0.074	0.002	0.009	11610.6	251.632562	2.28E-55	15
83	LQHP	7	fc.LQHP_rm.7	0.94	0.98	0	0	0.9	0	0.8	0.1	0.118	0.069	0.002	0.009	11584.4	225.485799	1.09E-49	19
84	LQHPT	7	fc.LQHPT_rm.7	0.95	0.97	0	0	0.9	0	0.8	0.1	0.121	0.055	0.002	0.009	11566.6	207.630951	8.18E-46	20
85	L	7.5	fc.L_rm.7.5	0.86	0.86	0	0	0.9	0	0.8	0.1	0.1	0.055	0.002	0.009	12019	660.057611	4.67E-144	6
86	LQ	7.5	fc.LQ_rm.7.5	0.88	0.89	0	0	0.9	0	0.8	0.1	0.102	0.059	0.002	0.009	11993.4	634.440378	1.71E-138	7
87	H	7.5	fc.H_rm.7.5	0.94	0.99	0	0	0.9	0	0.8	0.1	0.104	0.064	0.002	0.009	11666.8	307.900975	1.38E-67	18
88	LQH	7.5	fc.LQH_rm.7.5	0.94	0.97	0	0	0.9	0	0.8	0	0.116	0.076	0.002	0.009	11620.7	261.803234	1.41E-57	16
89	LQHP	7.5	fc.LQHP_rm.7.5	0.94	0.98	0	0	0.9	0	0.8	0.1	0.116	0.076	0.002	0.009	11588.7	229.739381	1.29E-50	16
90	LQHPT	7.5	fc.LQHPT_rm.7.5	0.95	0.97	0	0	0.9	0	0.8	0.1	0.116	0.061	0.002	0.009	11573.2	214.254895	2.98E-47	18
91	L	8	fc.L_rm.8	0.86	0.86	0	0	0.9	0	0.8	0.1	0.1	0.055	0.002	0.009	12022.2	663.2353	9.54E-145	5
92	LQ	8	fc.LQ_rm.8	0.88	0.89	0	0	0.9	0	0.8	0.1	0.102	0.059	0.002	0.009	11996.3	637.382393	3.92E-139	6
93	H	8	fc.H_rm.8	0.94	0.99	0	0	0.9	0	0.8	0.1	0.097	0.071	0.002	0.009	11674.7	315.745167	2.73E-69	16
94	LQH	8	fc.LQH_rm.8	0.94	0.97	0	0	0.9	0	0.8	0	0.116	0.076	0.002	0.009	11629.4	270.433028	1.89E-59	16
95	LQHP	8	fc.LQHP_rm.8	0.94	0.98	0	0	0.9	0	0.8	0.1	0.111	0.071	0.002	0.009	11596.2	237.308908	2.94E-52	15
96	LQHPT	8	fc.LQHPT_rm.8	0.95	0.97	0	0	0.9	0	0.8	0.1	0.116	0.061	0.002	0.009	11584.9	226.003463	8.38E-50	19
97	L	8.5	fc.L_rm.8.5	0.86	0.87	0	0	0.9	0	0.8	0.1	0.1	0.055	0.002	0.009	12025	666.096195	2.28E-145	5
98	LQ	8.5	fc.LQ_rm.8.5	0.88	0.89	0	0	0.9	0	0.8	0.1	0.102	0.059	0.002	0.009	12001.3	642.414022	3.17E-140	6
99	H	8.5	fc.H_rm.8.5	0.94	0.99	0	0	0.9	0	0.8	0.1	0.1	0.069	0.002	0.009	11684.4	325.466205	2.11E-71	15
100	LQH	8.5	fc.LQH_rm.8.5	0.94	0.97	0	0	0.9	0	0.8	0	0.116	0.076	0.002	0.009	11636.9	277.92398	4.45E-61	15
101	LQHP	8.5	fc.LQHP_rm.8.5	0.94	0.98	0	0	0.9	0	0.8	0.1	0.107	0.072	0.002	0.009	11605.9	246.947532	2.37E-54	15
102	LQHPT	8.5	fc.LQHPT_rm.8.5	0.94	0.97	0	0	0.9	0	0.8	0.1	0.116	0.061	0.002	0.009	11595.4	236.450022	4.51E-52	19
103	L	9	fc.L_rm.9	0.86	0.87	0	0	0.9	0	0.8	0.1	0.102	0.059	0.002	0.009	12027.9	669.008047	5.32E-146	5
104	LQ	9	fc.LQ_rm.9	0.87	0.88	0	0	0.9	0	0.8	0.1	0.1	0.055	0.002	0.009	12006.6	647.635068	2.33E-141	6
105	H	9	fc.H_rm.9	0.94	0.99	0	0	0.9	0	0.8	0.1	0.104	0.071	0.002	0.009	11696.6	337.620855	4.85E-74	15
106	LQH	9	fc.LQH_rm.9	0.94	0.97	0	0	0.9	0	0.8	0	0.114	0.077	0.002	0.009	11646.5	287.577673	3.57E-63	15
107	LQHP	9	fc.LQHP_rm.9	0.94	0.98	0	0	0.9	0	0.8	0.1	0.107	0.072	0.002	0.009	11617.7	258.750803	6.49E-57	16
108	LQHPT	9	fc.LQHPT_rm.9	0.94	0.97	0	0	0.9	0	0.8	0.1	0.116	0.063	0.002	0.009	11603.9	244.921973	6.53E-54	18
109	L	9.5	fc.L_rm.9.5	0.86	0.86	0	0	0.9	0	0.8	0.1	0.102	0.059	0.002	0.009	12030.9	671.970204	1.21E-146	5
110	LQ	9.5	fc.LQ_rm.9.5	0.87	0.88	0	0	0.9	0	0.8	0.1	0.102	0.059	0.002	0.009	12012	653.039672	1.56E-142	6
111	H	9.5	fc.H_rm.9.5	0.94	0.99	0	0	0.9	0	0.8	0.1	0.104	0.071	0.002	0.009	11713.1	354.135594	1.26E-77	17
112	LQH	9.5	fc.LQH_rm.9.5	0.94	0.97	0	0	0.9	0	0.8	0	0.109	0.072	0.002	0.009	11654.1	295.133581	8.16E-65	14
113	LQHP	9.5	fc.LQHP_rm.9.5	0.94	0.98	0	0	0.9	0	0.8	0.1	0.104	0.072	0.002	0.009	11630.7	271.802462	9.50E-60	17
114	LQHPT	9.5	fc.LQHPT_rm.9.5	0.94	0.97	0	0	0.9	0	0.8	0.1	0.109	0.063	0.002	0.009	11614.7	255.758502	2.90E-56	18
115	L	10	fc.L_rm.10	0.86	0.87	0	0	0.9	0	0.8	0.1	0.102	0.059	0.002	0.009	12033.9	674.982053	2.68E-147	5
116	LQ	10	fc.LQ_rm.10	0.87	0.88	0	0	0.9	0	0.8	0.1	0.102	0.059	0.002	0.009	12017.6	658.622323	9.57E-144	6
117	H	10	fc.H_rm.10	0.94	0.99	0	0	0.9	0	0.8	0.1	0.104	0.071	0.002	0.009	11722.4	363.42287	1.21E-79	16
118	LQH	10	fc.LQH_rm.10	0.94	0.98	0	0	0.9	0	0.8	0	0.111	0.08	0.002	0.009	11666	307.047532	2.11E-67	15
119	LQHP	10	fc.LQHP_rm.10	0.94	0.98	0	0	0.9	0	0.8	0	0.104	0.072	0.002	0.009	11639.2	280.226565	1.41E-61	16
120	LQHPT	10	fc.LQHPT_rm.10	0.94	0.98	0	0	0.9	0	0.8	0.1	0.107	0.065	0.002	0.009	11625.7	266.723292	1.20E-58	18

). The maps presented in this study correspond to the projections derived from this selected model. Figure 1 illustrates the recent historical habitat suitability of Valonia oak (Quercus ithaburensis subsp. macrolepis), showing its overall and natural ranges.

The seven environmental factors shaping the habitat suitability of the Valonia oak were the mean diurnal temperature range (bio2, 33.1%), precipitation in the wettest month (bio13, 19%), the mean annual temperature (bio1, 16.7%), the seasonality of temperature (bio4, 13.8%), precipitation in the driest month (bio14, 7.9%), the seasonality of precipitation (bio15, 6.1%), and the slope (slope, 3.2%). Its distribution is primarily influenced by the mean diurnal range (bio2), precipitation in the wettest month (bio13), mean annual temperature (bio1), and the seasonality of temperature (bio4).

When the distribution of the species in the paleoclimatic projections was analyzed (Figure 2 and Figure 3), the LGM (21 ka BP) period estimates were quite similar, and the overall data set is very close to the present-day estimate and higher than the natural distribution estimate (6.32%).

In the Pre-Boreal (10 ka–9 ka BP) period at the beginning of the Holocene, the habitat suitability of the species had narrowed in comparison with the LGM, and it was also narrower (7.86%) than the present-day distribution. In the Boreal (9 ka–8 ka BP) and Atlantic (8 ka–5 ka BP) periods, the habitat suitability of the species was estimated to be narrower (8.26% and 8.68%, respectively) than in the present-day distribution, and broader in the natural distribution dataset (Appendix A Figure A1). Across both datasets, the Sub-Boreal (5 ka–2.5 ka BP) period exhibited narrower habitat suitability. The natural distribution dataset showed that 9.29% were suitable areas, while the overall dataset showed 10.96%. During the Subatlantic period, the estimates displayed a trend similar to the present-day habitat suitability. Paleoclimatic projections initially indicated restricted habitat suitability for the species, with contraction at the start of the Holocene followed by expansion. The natural distribution data were projected to align more closely with the present-day distribution than the overall dataset.

The comparison of the overall range (a) and natural range (b) across binary maps of the paleoclimatic projections revealed key differences in the habitat suitability of Valonia oak throughout the Mediterranean. The general range consistently indicated larger areas with suitable habitat for the species, particularly in Corsica (France), Tunisia, and parts of southern Italy and Greece, where the natural range appeared narrower and more fragmented. In the central Mediterranean (Italian Peninsula, Sicily, and Sardinia), the overall range highlighted significant potential for habitat suitability while the natural range was more localized with concentrated distributions. In the eastern Mediterranean, the overall distribution suggested high habitat suitability; however, the natural range was more focused on key areas such as coastal Türkiye, Greece, the Aegean, and Cyprus compared with the broader extent of the overall range. Similarly, in northern Africa, the overall range revealed substantial potential in Tunisia and northern Libya, yet the natural range reflected more restricted and fragmented habitat suitability. The binary maps presented in Figure 3 also illustrate temporal differences in habitat suitability across the various periods. From the LGM to the present, the habitat suitability steadily increased, with both the total and natural ranges expanding progressively over time.

4. Discussion

The habitat suitability of the species is particularly related to temperature-related variables (bio1, bio2, and bio4). The mean annual temperature determines the growth and distribution limits of the species. In addition, the seasonality of temperature may affect the seasonal cycles and seed production of the species. The relationship between the habitat suitability of the species and rainfall (bio13, bio14, and bio15) determines the growth period of the species and its need for water. Fluctuations in rainfall regimes can affect seed germination rates and seedling survival. Among the topographic factors, slope affects the water-holding capacity and soil depth. This converges with findings on Quercus ithaburensis susbp. Ithaburensis, whose species distribution is also associated with soil depth [69]. In particular, the mean diurnal range (bio2), precipitation in the wettest month (bio13), mean annual temperature (bio 1), and the seasonality of temperature (bio4) are most important factors affecting the distribution of the species. The model results indicate that bio2 is the most significant factor in habitat suitability, explaining 33.1% of the variation. The species thrives best in coastal zones and low elevation valleys close to the Mediterranean Sea, which is reflected in its short diurnal range [4]. These findings show that temperature fluctuations play a major role in the ecological requirements of the species. Furthermore, bio13 contributes 19% to the habitat suitability. This stresses the need to maintain adequate water availability during critical periods of the year. The bio1 variable accounts for 16.7%, followed by bio4 at 13.8%, which implies that the species has been affected by climate change over a long period of time. This remains consistent with previous studies focused on Mediterranean oak species that identified temperature and precipitation as significant factors in determining suitable habitats [54,62]. In addition, Kenar [73] stated that the species is expected to experience slight declines in suitable habitats and to be more affected by climate change than by the combined effects of land-use and climate change, particularly with minor temperature increases. Although the results of our ecological niche modeling can offer us some valuable information on the habitat suitability and potential historical range shifts of Valonia oak, it is important to acknowledge the limitations inherent to such models: (1) They do not consider biotic interactions, dispersal constraints, or evolutionary limitations. Therefore, while our results imply a native or relict lineage for the Karagüney Range population, this conclusion should be treated with caution. (2) To confirm the autochthonous status of this population, robust evidence such as fossil pollen data, dendrochronological records, or genomic analyses are required. These additional datasets might directly indicate historical persistence, local adaptation, or long-term isolation. In addition, while historical and biogeographic factors have been discussed regarding habitat contraction and expansion, the mechanisms that underpin these dynamics, such as the role of human land use, competition with co-occurring species, or fire regimes, should be further considered. Future study integrating historical land-use data, vegetation dynamics, and long-term ecological monitoring will be essential to better understand these patterns.

Regarding conservation strategies, although the suggestion to improve habitat connectivity is still valuable, it is certainly vague. These findings strongly indicate that in the future, conservation efforts should focus on climatically stable refugia, particularly regions in northeastern Central Anatolia that are less likely to disappear, and even fragmented habitats remain potentially suitable, according to our projections. The identification of such microrefugia could lead to identification and protection efforts that support genetic resilience and population viability in the face of ongoing climate change.

For ENMs, general occurrence data along with the observed distribution at the Karagüney Mountains and the natural distribution were used. The comparison of overall (Figure 1a) and natural (Figure 1b) species distributions showed that the natural dataset fits the known pattern. In other words, this study demonstrated substantial congruence between observed occurrences and current climatic suitability projections for the target species, which supports the effectiveness of the ENM. In addition, according to the overall dataset, suitable habitats were predicted to occur in Marmara, Central Anatolia, northward along the Anatolian diagonal, and in the central Black Sea. The suitability maps depicted that the species’ natural habitat was particularly restricted to certain marginal and relict locations, such as the Karagüney Mountains. Evidence from the dataset suggested that current climatic conditions in these areas might be inadequate to fully support the species’ ecological needs. The overall dataset emphasized these potential discrepancies. Hence, this is an important consideration for future conservation efforts, as these regions may experience climatic challenges that could limit the establishment or survival of Valonia oak populations. In this context, identifying and prioritizing marginal habitats are necessary for robust conservation strategies, as these areas could become increasingly vulnerable under future climate scenarios.

As mentioned in the Introduction, the primary aim of this study was to assess whether the species population in the Karagüney Mountains was native or introduced. The diameters and morphological characteristics of the monumental trees observed during field studies in this region have reflected their exceptional lifespan. Their presence in the area may date back centuries. Understanding whether the species occurs naturally in this area is complex. In both cases, the Subatlantic period was favorable for the species in northeastern and central Anatolia, which might indicate that the species was able to find a marginal azonal distribution area where multiple climatic and biogeographic factors were in favor of its possible presence. The conditions during the Subatlantic period could have been similar in this part of Anatolia, with a suitable diurnal range or suitable quantities of precipitation during the wettest month. Another marginal aspect might be the altitudinal distribution of the species. While this species occurs at elevations of 500–600 m along the Uşak-Kütahya line, where it has its broadest distribution in Türkiye’s Mediterranean region, it reaches up to an elevation of 1000 m in Emirdağ in the east. In this region, it can grow at elevations of up to 1300 m, thriving particularly between elevations of 1000 and 1100 m. Besides climatic and biogeographic factors, another factor is the economic value of the species, which has led to its protection by local villagers for the past forty years, casting doubt on the concept of competition. Although evaluating or demonstrating competition is beyond the scope of this study, changing competitive conditions could still have influenced the species’ current distribution. Nevertheless, evidence exists to suggest the natural presence of this species in the region. With ongoing deforestation in Central Anatolia, only limited extensive forest remnants have survived. These forests are predominantly located in rugged, agriculturally unsuitable areas in the eastern and northeastern parts of Central Anatolia. Almost all the intermediate areas, particularly alluvial and valley floors where Valonia oak thrives, have been entirely converted into agricultural lands [74].

Valonia oak, an element of the Mediterranean region, typically co-occurs with other species of Mediterranean oaks—Turkey oak (Q. cerris) and Downy oak (Q. pubescens). The target species appears to have ranged more widely in distribution historically, although currently, Turkey oak has a limited range in a few remnant forests in Central Anatolia. This distribution pattern suggests that Valonia oak might have historically occupied a more extensive range in the region and subsequently contracted due to habitat loss, climate changes, or human disturbances. One of the most important abiotic stressors in the Mediterranean region is the increase in drought frequency. Drought-tolerant species are favored over less-drought-tolerant oaks, which would have affected the ability of the target species to regenerate and to create associations within the ecosystem. Other similar Mediterranean oak communities encounter the same issues, whereby environmental stressors reshape community structures, resulting in less biodiversity and forest resilience. This observation strengthens the idea of historical range reductions for Valonia oak in Central Anatolia [74].

The historical context provided by Birks [113] discusses how oak species have demonstrated a high degree of adaptability and resilience by modifying their distribution ranges in response to Quaternary climatic oscillations (e.g., glacial/interglacial patterns). The survival of oak populations during unfavorable periods of climatic conditions has been assumed to depend on their retention in macro- and microrefugia [113]. During the LGM, the extent of their occurrence is similar to that of the current distribution, suggesting that the species has adapted relatively well to significantly lower temperatures and drier conditions, namely for populations located at higher elevations and in eastern and central Anatolia. During the Atlantic period, higher temperatures and increased precipitation occurred; however, the species exhibited a narrower distribution, suggesting that the species shifted its range to relatively drier and cooler areas. Hence the Valonia oak has tolerance to drought and cold, explaining its possible expansion (and further retraction) toward the northeast during the Atlantic and Subatlantic period. The current distribution is mainly shaped by temperature seasonality, temperature diurnal variation, and the amount of precipitation in the wettest month. Despite divergent results in distribution with a mesic species like Juniperus drupacea, the factors shaping species distribution are the same [114,115,116,117].

Disentangling the relationships between species and their environments is complex, as these involve biological traits and environmental influences. Species’ ability to potentially follow suitable climate conditions under climate change is largely controlled by their dispersal abilities, landscape connectivity, and the velocity of climatic shifts. The degree of change in temperature gradients, particularly in mountainous areas, plays an important role in the survival and movement of species. At higher elevations, the slower movement of temperature gradients due to lapse rates and topography suggests that mountains may provide climatic refugia for Valonia oak. However, although these mountainous refugia facilitate short-term survival, they may also contribute to thermal isolation. As a result, sections within the valleys and at lower altitudes of these massifs become unsuitable. This isolation could alter genetic exchange and increase the risk of local extinction cases. The level of topographic diversity in the study region implies that the observed Valonia oak distribution patterns are likely influenced by local microclimates [74]. Taking this factor into account, conservation efforts should focus on habitat connectivity, particularly in mountainous refugia that would allow for species dispersal, genetic diversity, and greater resiliency toward climatic changes in the future.

5. Conclusions

This study quantified changes in the habitat suitability of Valonia oak (Quercus ithaburensis subsp. macrolepis), considering different time periods from the LGM to the Anthropocene. Using high-resolution historical datasets and applying the MaxEnt algorithm, we modeled present and paleoclimatic distributions of the species. Spatial heterogeneity in habitat suitability was observed between the overall and natural distribution datasets, particularly in the central and eastern Mediterranean regions. Paleoclimatic estimations indicated substantial shifts in suitable habitats throughout different historical climatic periods. During the LGM, suitable habitats showed a similar pattern with the present distribution but contracted considerably during the early Holocene. Subsequently, a gradual expansion of suitable areas occurred, aligning more closely with the current natural range. According to this result, this confirms that the species has shown a stable resistant and adaptive character to the effects of climate change in the long term, so it can be considered a potential alternative species under climate change. This resilience suggests that the species may possess key ecological traits, such as drought tolerance and niche flexibility, which can be strategically utilized in forest restoration and climate adaptation programs. The classification of the species as either relict or introduced remains complex due to its significant geographical isolation from known natural ranges in the Karagüney Mountains. However, the population is considered marginal or peripheral to its current species range, and for some characteristics, it can be, in some cases, considered endangered because of its isolation from important sources of genetic adaptation. We recommend incorporating dendrochronological analyses, pollen records, historical documents, genetic analyses, and long-term monitoring in further studies to make broader interpretations and to support conservation strategies of the species in response to climate change. Moreover, integrating these approaches with predictive habitat modeling could enhance our ability to forecast range dynamics and to identify climate refugia for effective in situ and ex situ conservation planning.