Past legacies and future trajectories: climatic refugia, range shifts, and conservation gaps for the endangered Oglethorpe oak (Quercus oglethorpensis)

Abstract

Introduction:

The endangered Quercus oglethorpensis is a narrowly distributed tree species endemic to the southeastern United States, persisting in fragmented, poorly drained hardwood forests. Reconstructing climatic suitability across past, present, and future periods provides insight into how long-term climatic legacies and ongoing warming constrain its range dynamics.

Methods:

We modeled climatic suitability during the Last Glacial Maximum (LGM), Mid-Holocene (MH), present, and a projected future period (2071–2100, SSP3-7.0) using species distribution models parameterized with bioclimatic variables.

Results:

Results indicate a restricted Gulf Coast refugium during the LGM, with a secondary extension into the southern Appalachian foothills, followed by northward expansion into the interior Southeast during the MH. Present-day suitability is highly fragmented, and future projections suggest a pronounced northeastward shift toward the Piedmont and Appalachian regions, with an estimated centroid displacement of ∼672 km from the LGM to the projected future. Although total climatically suitable area is projected to decline modestly (∼5%), the representation of suitable habitat within protected areas declines disproportionately (∼64%), from 0.9% of the current suitable range to 0.3% under future conditions. Multi-period overlap analysis identifies only a single persistent refugium (2.83 km²), underscoring the rarity of long-term climatic stability.

Discussion:

These findings demonstrate how Quaternary climate dynamics and ongoing warming interact to drive range shifts and fragmentation, exposing substantial conservation gaps. Effective conservation of Q. oglethorpensis will require integrating in situ protection of remnant populations with ex situ conservation, connectivity planning, and adaptive forest management under future climate change.

Introduction

Oaks (Quercus spp.) are foundational species in North American forests. Their ecological roles are multifunctional, supporting diverse species, contributing to nutrient cycling, providing habitat for wildlife, and supplying food, wood, and timber for humans. However, many oak species are now endangered or threatened due to increasing anthropogenic and natural disturbances. One species of conservation concern is the Oglethorpe oak (Quercus oglethorpensis), which persists in a highly fragmented disjunct patches of poorly drained forest across five southeastern states. Fewer than 1,000 mature individuals remain in the wild, scattered across small and isolated populations, highlighting the species’ rarity and fragmented distribution, which are associated with reduced genetic diversity and increased conservation concern (Lobdell and Thompson, 2017; Beckman et al., 2019; Spence et al., 2021). Oaks (Quercus spp.) have a long evolutionary history spanning tens of millions of years and have persisted through repeated climatic oscillations that shaped their geographic distributions (Backs and Ashley, 2021; Barrón et al., 2017). Reconstructing climatic suitability for species across the Last Glacial Maximum (LGM), Mid-Holocene (MH), present, and projected future (2071–2100) may provide valuable insight into how climatic factors have shaped its range dynamics (Adhikari et al., 2026). In this study, we reconstructed climatic suitability for Q. oglethorpensis for those time periods using species distribution models. This information may be critical for conservation planning and management, including identifying historical climatic refugia and potential genetic diversity hotspots, understanding range shifts and persistence, and informing climate adaptation strategies.

During previous glacial cycles, many species persisted in localized refugia that acted as reservoirs of biodiversity under unfavorable climatic conditions (Médail and Diadema, 2009; Brighenti et al., 2021). Following glacial retreats, terrestrial plant populations expanded outward from these refugia, reshaping their distribution under post-glacial warming. These historical dynamics underscore strong sensitivity of plant distribution to climatic variation. Today, anthropogenic climate change is driving similarly structured but far more rapid range shifts, with many species projected to track suitable climates toward higher latitudes and elevations (Médail and Diadema, 2009; Subedi et al., 2015; Brighenti et al., 2021; Adhikari et al., 2023; Ross et al., 2024). A growing body of evidence indicates that contemporary climate change is already producing comparable, yet accelerated, effects on biodiversity worldwide (Bellard et al., 2012; Pecl et al., 2017; Keppel et al., 2012; Das et al., 2019). Species may respond through range shifts, modifications in life-history strategies, or genetic adaptation to novel environments (Bellard et al., 2012). However, when environmental changes exceed physiological tolerance thresholds and dispersal or adaptation is not possible, local populations face high risk of extinction (Thomas et al., 2004; Berg et al., 2010; Urban, 2015; Das et al., 2019; Subedi et al., 2019). The unprecedented rate of ongoing climate change increases this risk, raising concern that Q. oglethorpensis may be unable to adapt to avoid significant population declines or increased extinction risk.

Q. oglethorpensis is one of the endangered oak species endemic to the southeastern U.S., with a small highly fragmented population. Identifying geographic areas that can serve as climate refugia is a vital conservation strategy for future (Reside et al., 2014), particularly where such areas can complement existing protected-area networks (Keppel et al., 2012; Das et al., 2019). In situ refugia-locations that are currently climatically suitable and projected to remain so, are especially important for species with limited dispersal ability, fragmented distributions, or few alternative habitats (Ashcroft, 2010). Species distribution models (SDMs) have emerged as effective tools for analyzing species-environment relationships, assessing habitat quality, and informing conservation strategies (Guisan et al., 2013; Subedi et al., 2023, 2024, 2025). SDMs define ecological requirements to explain and predict the potential extent of a species’ distribution across space and time (Guisan et al., 2013). In this study, we aim to address three key objectives: (1) to evaluate past and present climatic suitability for Q. oglethorpensis and identify the bioclimatic variables influencing its spatial patterns, (2) to project shifts in climatic suitability under future climate change scenarios, and (3) to assess how climatically suitable habitat is represented within the current protected area network under present and future conditions. We hypothesize that the climatic suitability of Q. oglethorpensis is highly concentrated within its current southeastern range and will decline substantially under future climate change, exacerbating risks associated with increasing anthropogenic pressures.

Materials and methods

Species occurrence data

We compiled occurrence records for Q. oglethorpensis from multiple herbaria through the Southeast Regional Network of Collections¹, encompassing 130 herbarium records from 22 herbaria, and from the Global Biodiversity Information Facility (GBIF). After removing duplicate entries and points with obvious geolocation errors, 129 records remained. To reduce spatial autocorrelation and sampling bias, we applied a 1-km spatial thinning filter in R (R Core Team, 2023) using the spThin package (Aiello-Lammens et al., 2015). Localities falling within the same 1-km grid cell or outside the known species range were removed. As a result, 64 Q. oglethorpensis present locations were retained in subsequent modeling, with only one data point at each grid cell (1 km²) (Figure 1).

FIGURE 1

Environmental predictors

We used 19 bioclimatic variables (bio01–bio19) from the Climatologies at High Resolution for the Earth’s Land Surface Areas (CHELSA) dataset at ∼1 km resolution as environmental predictors for species occurrence, characterizing climatic conditions across four time periods relevant to the distribution of Q. oglethorpensis: the LGM, MH, Present, and Future (late-21st-century). Paleoclimate layers (LGM and MH) were obtained from CHELSA-TraCE21k centennial reconstructions. Future predictors were derived from CHELSA v2.1 downscaled CMIP6 projections for 2071–2100 under the SSP3-7.0 medium-high emissions pathway. We selected the MPI-ESM1-2-HR global climate model because it is one of the high-resolution CMIP6 models available in CHELSA v2.1 and is supported by published model documentation demonstrating improvements in physical parameterizations and large-scale climate representation (Gutjahr et al., 2019). CMIP6 evaluations also show that the suite of CMIP6 models, including MPI-ESM1-2-HR, reasonably reproduces U.S. temperature patterns and precipitation extremes at broad spatial scales (Akinsanola et al., 2020; Srivastava et al., 2020), making it suitable for regional ecological applications. We used SSP3-7.0 because it represents a plausible “regional rivalry” trajectory characterized by limited climate mitigation, expanding land-use pressures, and substantial warming—conditions consistent with ongoing socioeconomic and environmental trends in the southeastern United States (O’Neill et al., 2017; Riahi et al., 2017). SSP3-7.0 is also widely applied in biodiversity and conservation assessments because it avoids the extreme socioeconomic assumptions of high-end forcing scenarios while still capturing meaningful climatic change relevant for ecological forecasting (IPBES, 2019). The CHELSA bioclimatic and climatology layers are openly accessible for download through the CHELSA data portal (version 2.1) (Karger et al., 2021).

All 19 CHELSA bioclimatic variables (bio01–bio19) were initially considered as candidate predictors. Environmental layers for each time slice were resampled to a common projection and resolution and masked to a buffered convex hull around the spatially thinned occurrence records to ensure consistent spatial coverage. To reduce multicollinearity, we calculated variance inflation factors (VIF) and retained only predictors with VIF < 5, following established best practices in ecological niche modeling (Phillips and Dudík, 2008; Kramer-Schadt et al., 2013). The final predictor set comprised six ecologically interpretable and statistically independent variables representing key seasonal temperature and precipitation gradients: bio05 (maximum temperature of the warmest month), bio06 (minimum temperature of the coldest month), bio08 (mean temperature of the wettest quarter), bio09 (mean temperature of the driest quarter), bio18 (precipitation of the warmest quarter), and bio19 (precipitation of the coldest quarter).

Model calibration

Species distribution models (SDMs) were developed in Wallace v2.0 (Kass et al., 2023) using the MaxEnt algorithm implemented through the maxnet package (Phillips and Dudík, 2008; Kramer-Schadt et al., 2013). MaxEnt allows different feature classes that define the functional form of species–environment relationships, including linear (L), quadratic (Q), hinge (H), and product (P) features. To reduce overfitting and improve model generality, we evaluated multiple combinations of feature classes (L, LQ, H, LQH, LQHP) and regularization multipliers (0.5–3.5). This approach is commonly used to balance model complexity and predictive performance (Phillips and Dudík, 2008; Muscarella et al., 2014). Although some feature classes (e.g., linear and hinge) may partially overlap in functional form, evaluating multiple combinations allows identification of the best-performing model based on empirical fit and information-theoretic criteria. The regularization multiplier controls model smoothness by penalizing complexity, with higher values producing simpler models. Models were trained using 2,000–4,000 randomly sampled background points drawn from within the study extent, defined as a 1° buffer surrounding all spatially thinned occurrence records to approximate the species’ accessible area. Occurrence records were partitioned using the block method to produce spatially independent training and testing datasets. Continuous suitability outputs were generated on the cloglog scale, which provides an interpretable estimate of occurrence probability. All transfers were run with clamping enabled to prevent extrapolation beyond the calibration range.

Model evaluation and selection

Model performance was evaluated using 4-fold block cross-validation (Muscarella et al., 2014; Roberts et al., 2017). We calculated the area under the receiver operating characteristic curve (AUC), the continuous Boyce index (CBI), and omission rates at the minimum training presence (mtp) and 10th percentile (10p) thresholds. The p10 threshold balances sensitivity and specificity, minimizes marginal overprediction, and ensures consistent comparability across time periods (Liu et al., 2005; Jiménez-Valverde and Lobo, 2007; Pearson et al., 2007). Across all feature class and regularization multiplier combinations, mean validation AUC values ranged from 0.89 to 0.94, indicating strong discriminatory power. Mean CBI values ranged from 0.51 to 0.69, suggesting consistent reliability across spatial partitions. Omission rates at the 10p threshold varied between 0.17 and 0.39, with most models falling below the commonly accepted cutoff of 0.5.

Candidate models were further ranked using Akaike Information Criterion corrected for small sample size (AICc) (Akaike, 1974; Warren and Seifert, 2011). Among all tested models, hte LQH feature class with RM = 1.5 achieved the lowest AICc (1489.73), ΔAICc = 0, and the highest model weight (AICc weight = 0.92). This model also showed strong performance across evaluation metrics (mean AUC = 0.93, CBI = 0.51, 10p omission = 0.36). Competing models (e.g., H with RM = 2.5) had ΔAICc > 10 and negligible weights, confirming LQH_rm1.5 as the best-supported candidate. The final cloglog output from this model was retained for subsequent analyses. Presence-absence maps were derived using the 10th percentile training presence threshold (p10 = 0.299), which balances sensitivity and specificity and facilitates comparability across time periods.

Post-processing and interpretation

Predicted climatic suitability was generated for four climatic periods (LGM, MH, Present, and Future) and post-processed in R using terra (Hijmans et al., 2026). For each time slice, we calculated the geographic centroid of suitable habitat, providing a spatially explicit measure of the horizontal displacement of climatically favorable conditions through time. Centroid positions were compared across periods to quantify directional shifts in potential habitat, helping identify areas that may have functioned as historical refugia or future climate pathways for Q. oglethorpensis. Binary maps and continuous suitability surfaces were visualized to qualitatively evaluate patterns of contraction, expansion, or fragmentation. To evaluate how climatically suitable habitat is represented within the protected area network, we overlaid binary suitability maps (based on the 10th percentile training presence threshold) with protected area polygons from the World Database on Protected Areas (WDPA; UNEP-WCMC and IUCN, 2023). For each time period (present and future), we calculated the total area of climatically suitable habitat and quantified the extent of suitable habitat within each protected area. We then computed the proportion of each protected area that was climatically suitable, as well as the proportion of total suitable habitat contained within the protected area network.

Results

Our study shows the shifting climatically suitable habitat of Q. oglethorpensis across time (Figure 2). During the Last Glacial Maximum (21,000 ybp), climatically suitable habitat for Q. oglethorpensis was primarily concentrated along the Gulf Coast, with a secondary area of suitability extending into the southern Appalachian foothills. By the Mid-Holocene (6,000 ybp), it expanded northward into the interior Southeast. At present, climatically suitable locations of this species occur in fragmented patches across the Southeast. The projections for the year 2100 suggest a further northward shift, concentrating mainly in the Appalachian region.

FIGURE 2

Study-area habitat dynamics

Across the Southeastern U.S. region of interest, climatically suitable habitat for Q. oglethorpensis under the p10 threshold was estimated at 63,259.8 km² in the present and 60,070.4 km² in the future, with a modest overall decline of −3,189.4 km² (−5.0%).

Habitat representation within protected areas

A total of 19 protected areas (PAs) currently contain climatically suitable habitat for Q. oglethorpensis, encompassing approximately 560 km², which represents <1% of all climatically suitable area across the region (Table 1). By 2100, the extent of climatically suitable habitat within these PAs is projected to decline to ∼200 km² (0.33% of the total), a 64.2% reduction. Although the regional extent of suitable climate decreases by only ∼5%, representation within the PA network declines disproportionately, with the protected share dropping from 0.89% to 0.33%—a 63% loss relative to present.

Based on our current distribution data, Q. oglethorpensis is reported from four protected areas - Bienville National Forest (MS), Oconee National Forest (GA), George L. Smith State Park (GA), and Sumter National Forest (SC). In addition, our model projection showed that present-day suitability is concentrated in a few large PAs. Cedar Creek (157 km², 98% of its area climatically suitable), Piedmont (139 km², 98%), and B. F. Grant (59 km², 101%) together hold more than half of all climatically suitable habitat currently protected. Under future projections, suitability becomes more restricted and shifts eastward, with Grant (57 km², 98%), Crockford-Pigeon Mountain (32 km², 41%), and Croft (25 km², 92%) emerging as the main refugial sites.

Only four protected areas currently contain both current and future climatically suitable habitat: Grant, Ogeechee, Elbert, Wilkes (Table 1). Although the average proportion of each PA that remains climatically suitable changes only slightly (from 0.58 to 0.53), the absolute extent of protected climatically suitable habitat declines sharply – from ∼560 km² at present to ∼200 km² by 2100. This represents a loss of ∼360 km² (≈64%), indicating a marked contraction of protected climatic space into a small number of isolated refugial areas (Table 1).

Climatic centroids and directional shifts

Area-weighted centroid analysis indicated that the geographic core of climatically suitable habitat for Q. oglethorpensis remained relatively stable through the late Quaternary but is projected to undergo a pronounced east-northeastward shift in the future (Figures 2, 3). During the Last Glacial Maximum (LGM), the centroid was located near −86.68°, 30.20°, moving north-northeast by approximately 236 km to −86.12°, 32.27° in the Mid-Holocene. From the Mid-Holocene to the present, the centroid shifted only 57 km northward (359.5°), remaining within the Alabama-Georgia Piedmont region. In contrast, future projections indicate a much larger displacement of about 438 km northeast (57.6°), reaching −82.09°, 34.83° in the western Carolinas and northern Georgia corridor. The net movement from the LGM to 2100 represents a 672 km north-northeast shift (38.8°), highlighting a major redistribution of suitable conditions toward the Appalachian foothills under future climate scenarios (Figure 3).

FIGURE 3

Climate refugia

Analysis of multi-period overlaps (LGM ∩ MH ∩ Present) identified a single climatic refugial patch totaling 2.83 km², with a centroid near 34.42°, −80.86° in the upper Piedmont/foothills region (Figure 3), which has remained suitable continuously across late-Quaternary to modern conditions.

Discussion

Our species distribution modeling reveals how past climate oscillations and future warming are reshaping climatic suitability for an endangered oak species. Here, we interpret the temporal distribution shifts from the LGM through the MH to the present, and into projected future climates. Our results indicate a pronounced loss of climatically suitable areas within protected lands despite only a modest decline in total suitable area. In addition, we detected a substantial displacement of the centroid of climatic suitability through time and identified only a single, extremely small persistent climatic refugium. Together, these patterns highlight a growing mismatch between areas of long-term climatic stability and existing conservation networks. Our findings underscore the vulnerability of Q. oglethorpensis to ongoing climate change and point to an urgent need for conservation strategies that account for both historical climatic legacies and future shifts in climatic suitability.

Our results partially support our initial hypothesis. While climatic suitability remains largely concentrated within the southeastern United States, we did not observe a substantial decline in the total area of suitable habitat under future climate scenarios. Instead, the primary changes involve a pronounced spatial displacement of suitable conditions and a marked reduction in overlap with existing protected areas. This shift results in climatically suitable habitats occurring far from the species’ current distribution, suggesting that natural migration is unlikely given its limited dispersal capacity. Consequently, future persistence of Q. oglethorpensis will likely depend on human-assisted movement and proactive conservation planning.

Historical contraction and post-glacial expansion

Our reconstructions indicate that Q. oglethorpensis was primarily confined to a Gulf Coast climatic refugium during the LGM (∼20,000 ybp), with a secondary area of climatically suitable habitat extending into the southern Appalachian foothills, followed by a northward expansion into the interior Southeast by the MH (∼6,000 ybp). This pattern is consistent with the classic refugial dynamics documented for temperate tree species during glacial-interglacial cycles (Bennett and Provan, 2008). For Q. oglethorpensis, the Gulf Coast likely represented an area of climatic suitability in a colder glacial condition. As post-glacial warming progressed, climatically suitable conditions expanded northward into newly suitable habitats of the Southeastern U.S. This emergence of climatic suitability in the Piedmont and Coastal Plain by the MH suggests that dispersal and establishment were possible once climatic constraints relaxed. However, climatically suitable areas remained geographically restricted to the Southeast, reflecting the species’ narrow ecological niche and potential dispersal limitations. The relatively limited extent of climatic suitability during the MH period implies that Q. oglethorpensis never achieved widespread distribution, but instead tracked a shifting envelope of suitable moist hardwood forest conditions. These historical patterns indicate that Quaternary climate oscillations drove repeated range contractions and expansions, setting the stage for the species’ current distribution. The persistence of Q. oglethorpensis through past climatic fluctuations, despite a consistently restricted climatic niche, suggests an inherent vulnerability, whereby long-term survival was closely associated with the availability of climatically suitable conditions. While climatically suitable areas identified by the models may have supported persistence during past climatic fluctuations, alternative mechanisms may also have contributed to the present-day distribution of Q. oglethorpensis. In particular, human-mediated dispersal by Indigenous peoples represents a plausible but underexplored mechanism, as acorns of many oak species are edible and have historically been transported and utilized by humans. Although there is no direct evidence for such dispersal in this species, this possibility highlights the uncertainty surrounding its historical distribution and suggests that both ecological and anthropogenic processes may have influenced its current range. Loss or degradation of climatically suitable conditions would therefore be expected to substantially increase extinction risk.

Current distribution and future range shifts

Our results indicate that climatically suitable habitat for Q. oglethorpensis at present is confined to highly fragmented patches across the southeastern U.S. Modeled suitable range is concentrated within disjunct, poorly drained forested areas distributed across Georgia, South Carolina, Alabama, Mississippi, and Louisiana, consistent with the species’ known geographic limitations. This fragmentation is further reinforced by the species’ narrow ecological requirements, typically occupying subcanopy positions in seasonally wet Piedmont oak-hickory forests, which restricts its ability to naturally recolonize newly suitable areas. Biologically, these small and isolated populations may experience Allee effects, including limited pollen transfer among distant populations, potentially limiting reproductive success. These populations also remain vulnerable to stochastic disturbances, as evidenced by recent surveys documenting local extirpations at historically occupied sites (Beckman et al., 2019). It is important to emphasize that SDMs estimate the climatic niche (i.e., climatically suitable habitats) rather than the realized niche. Thus, actual occupancy of climatically suitable areas may be constrained by additional factors, including biotic competition, limited dispersal, and natural or anthropogenic disturbances, all of which may influence the observed distribution relative to the model-predicted suitable habitat.

Our models forecast a pronounced northeastward shift of Q. oglethorpensis’ suitable range by the year 2100 (SSP3-7.0 scenario). Specifically, the climatically suitable habitat is projected to contract from the current Coastal Plain and lower Piedmont and re-center farther north and east, toward the uplands of the Piedmont and the southern Appalachian foothills. By 2100, much of the species’ current range in the deep south may become too warm or otherwise climatically unsuitable, while new areas in northern Georgia, South Carolina, and North Carolina emerge as potential refuges. This projected range shift is substantial in geographic extent as our centroid analysis indicates a ∼438 km displacement to the northeast from the present day to 2100, part of an overall ∼672 km north-northeast shift since the LGM. Such a dramatic shift reflects the rapid pace of 21st-century climate change. The future climate niche of Q. oglethorpensis lies outside its historical (Holocene) range, implying that without migration the species will face a severe climate mismatch in its current distribution.

These modeled shifts align directionally with the well-documented global trend of species moving poleward and to higher elevations under warming climates (Chen et al., 2011). Many taxa are already shifting their distributions at rates averaging ∼17 km per decade northward (for latitude) and ∼11 m per decade upward in elevation (Chen et al., 2011). However, the magnitude and speed of Q. oglethorpensis’ projected shift are particularly striking. A 438 km shift over the coming 80 years corresponds to roughly 55 km per decade, far exceeding typical observed rates. This would require dispersal and establishment far beyond the species’ known dispersal capacity. Q. oglethorpensis acorns are large, not wind-dispersed, and the species has limited regeneration even under current conditions. While our models assumed unlimited dispersal to identify suitable future areas, Q. oglethorpensis is unlikely to naturally migrate fast enough to track its shifting climate suitability. Tree regeneration from acorns tends to lag significantly behind climate change, especially in fragmented landscapes. Moreover, successful regeneration in Q. oglethorpensis depends on both seed stratification and animal-mediated dispersal. Without human intervention, Q. oglethorpensis may thus experience a reduction in distribution range, with dieback in the south and failure to colonize newly suitable areas in the north. This scenario is a major concern for immobile or dispersal-limited species under rapid climate. For instance, Thomas et al. (2004) projected that 15%–37% of species could be committed to extinction under mid-range warming by 2050, especially under limited dispersal scenarios. Q. oglethorpensis is a narrowly distributed, poorly dispersing specialist in a rapidly changing climate. Therefore, while our models highlight the Piedmont and southern Appalachians as future climatically suitable regions (which may function as new refugia or escape habitats), proactive steps will be needed to facilitate the species’ range shift into these areas.

Climate refugia, protected-area gaps, and management implications

Fragmentation and small population size can greatly reduce a species’ capacity to adapt or migrate in response to changing climate. When a species is already stressed and confined to isolated habitat patches, its resilience to additional climatic extremes (e.g., prolonged droughts or heat waves) is low. For Q. oglethorpensis, this means that without intervention, climate-driven habitat shifts could lead to local extinctions in parts of its current range long before the species manages to establish new populations elsewhere. Q. oglethorpensis faces other threats as well, such as habitat loss, competition from invasive plants, and pests/pathogens including susceptibility to diseases like chestnut blight (Beckman et al., 2019). These existing stresses compound the risks posed by climate change. The present-day fragmentation, therefore, not only has inherent conservation implications (e.g., needing genetic rescue or increased in small populations) but also indicates a sign of serious climate vulnerability. This underscores the urgency of enhancing habitat connectivity and implementing assisted gene flow or translocation strategies to prevent remnant populations from becoming climatically stranded as conditions shift (Subedi et al., 2020, 2022). Therefore, the fragmented status of Q. oglethorpensis reflects the combined legacy of past climate change and ongoing anthropogenic change, making it a prime example of how habitat fragmentation and climate change can threaten species survival.

For Q. oglethorpensis, the mismatch between its range and the protected areas is pronounced. Only four protected areas are projected to retain both current and future climatically suitable habitat, highlighting their importance as potential climate refugia. Many of the future suitable locations in the Piedmont/Appalachian region are not currently designated for conservation, or they fall in small, isolated pockets of public land. Meanwhile, some reserves that currently support the species may become climatically suboptimal, leading to local population declines even within protected sites. This calls for climate-aware conservation planning. Land managers should consider expanding or shifting protected area coverage to encompass predicted future habitats (for example, prioritizing land acquisition or protection in the Carolina foothills where our model indicates suitable climate will emerge). At minimum, connectivity between current protected sites and future habitats should be enhanced, so that natural or assisted dispersal can occur. The 64% collapse in protected habitat also highlights the importance of integrating climate projections into reserve design: conservation areas need to be evaluated and possibly realigned under scenarios of climate change to avoid severe gaps. Without such proactive measures, Q. oglethorpensis could effectively lose the thin safety net that protected areas currently provide, leaving the species even more vulnerable to land-use change and other threats in unprotected zones. In summary, our results illustrate a critical conservation gap - the areas that will remain suitable for Q. oglethorpensis are largely outside the current protected network, demanding strategic planning to safeguard those climate refugia and migration corridors before they are lost or degraded.

In addition to formally designated protected areas included in our analysis (WDPA), previous studies have identified sites where Q. oglethorpensis is actively managed, including Bienville National Forest, Oconee National Forest, Sumter National Forest, and George L. Smith State Park (Beckman et al., 2019). Some of these sites were not identified in our analysis, likely due to differences in protection classification and management objectives, as certain areas may not meet IUCN criteria for inclusion in the WDPA database or may be designated as multi-use landscapes rather than strictly protected areas. In addition, some sites may not overlap with modeled climatically suitable habitat under the applied threshold. Nonetheless, these examples highlight that targeted management actions, such as prescribed burning and selective clearing, can support population persistence outside formally recognized protected areas.

Overall, these findings show Q. oglethorpensis as a species highly vulnerable to ongoing climate change and illustrate how past and future range dynamics create new conservation challenges. Several key implications emerge for conservation planning can be implemented such as (i) establishing ecological corridors and managing landscapes to enable natural or assisted dispersal of Q. oglethorpensis populations toward climatically suitable areas, (ii) protecting current and future refugia to safeguard both existing and projected suitable habitats that may serve as long-term refuges under climate change, (iii) integration of in situ and ex situ conservation by combining habitat-based conservation with cultivation in arboreta to preserve genetic diversity, and (iv) incorporating climate adaptation, land-use management, and species recovery strategies into regional and national conservation policies.

Conclusion

The projections for Q. oglethorpensis highlight the complex challenges of conserving an endangered species in an era of rapid climate change. Although the species persisted through past climatic fluctuations, this persistence was likely associated with the availability of climatically suitable conditions. In contrast, the pace and magnitude of projected future change are unprecedented. Our modeling results reveal a scenario of significant range displacement, increasing fragmentation, and a severe mismatch with the current protected-area network, placing the species at elevated risk. At the same time, our results identify priority regions for future conservation, including potential refugia and areas where connectivity will be critical. Proactive, climate-informed management—including enhancing habitat connectivity, facilitating dispersal toward future climatically suitable areas, and integrating in situ and ex situ conservation strategies—will be essential to promote the long-term persistence of Q. oglethorpensis. Timely action to protect existing populations, secure climatically suitable areas, and support assisted dispersal can substantially improve the species’ ability to persist under ongoing climate change.

References

• 1

  AdhikariB.AlstonJ. M.BurgerJ. R. (2026). Patchy distribution of a Madrean Sky Island squirrel shaped by historical habitat configuration.bioRxiv [Preprint] 10.64898/2026.02.23.707456

  • CrossRef
  • Google Scholar
  • View reference in article

• 2

  AdhikariB.SubediS. C.BhandariS.BaralK.LamichhaneS.MaraseniT. (2023). Climate-driven decline in the habitat of the endemic spiny babbler (Turdoides nipalensis).Ecosphere14:e4584. 10.1002/ecs2.4584

  • CrossRef
  • Google Scholar
  • View reference in article

• 3

  Aiello-LammensM. E.BoriaR. A.RadosavljevicA.VilelaB.AndersonR. P. (2015). spThin: An R package for spatial thinning of species occurrence records for use in ecological niche models.Ecography38541–545. 10.1111/ecog.01132

  • CrossRef
  • Google Scholar
  • View reference in article

• 4

  AkaikeH. (1974). A new look at the statistical model identification.IEEE Trans. Automatic Control19716–723. 10.1109/TAC.1974.1100705

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 5

  AkinsanolaA. A.KoopermanG. J.PendergrassA. G.HannahW. M.ReedK. A. (2020). Seasonal representation of extreme precipitation indices over the United States in CMIP6 present-day simulations.Environ. Res. Lett.15:094003. 10.1088/1748-9326/ab92c1

  • CrossRef
  • Google Scholar
  • View reference in article

• 6

  AshcroftM. B. (2010). Identifying refugia from climate change.J. Biogeogr.371407–1413. 10.1111/j.1365-2699.2010.02300.x

  • CrossRef
  • Google Scholar
  • View reference in article

• 7

  BacksJ. R.AshleyM. V. (2021). Quercus conservation genetics and genomics: Past, present, and future.Forests12:882. 10.3390/f12070882

  • CrossRef
  • Google Scholar
  • View reference in article

• 8

  BarrónE.AveryanovaA.KvačekZ.MomoharaA.PiggK. B.PopovaS.et al. (2017). “The fossil history of Quercus,” in Oaks Physiological Ecology. Exploring the Functional Diversity of Genus Quercus L, edsGil-PelegrínE.Sancho-KnapikD.Peguero-PinaJ. J. (Cham: Springer International Publishing), 39–105.

  • Google Scholar

• 9

  BeckmanE.LobdellM.MeyerA.WestwoodM.BeckmanE.MeyerA.et al. (2019). Conservation Gap Analysis of Native U.S. Oaks. Lisle, IL: The Morton Arboretum.

  • Google Scholar

• 10

  BellardC.BertelsmeierC.LeadleyP.ThuillerW.CourchampF. (2012). Impacts of climate change on the future of biodiversity.Ecol. Lett.15365–377. 10.1111/j.1461-0248.2011.01736.x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 11

  BennettK. D.ProvanJ. (2008). What do we mean by ‘refugia’?Quaternary Sci. Rev.272449–2455. 10.1016/j.quascirev.2008.08.019

  • CrossRef
  • Google Scholar

• 12

  BergM. P.KiersE.DriessenG.Van Der HeijdenM.KooiB. W.KuenenF.et al. (2010). Adapt or disperse: Understanding species persistence in a changing world.Glob. Change Biol.16587–598. 10.1111/j.1365-2486.2009.02014.x

  • CrossRef
  • Google Scholar

• 13

  BrighentiS.HotalingS.FinnD. S.FountainA. G.HayashiM.HerbstD.et al. (2021). Rock glaciers and related cold rocky landforms: Overlooked climate refugia for mountain biodiversity.Glob. Change Biol.271504–1517. 10.1111/gcb.15510

  • CrossRef
  • Google Scholar

• 14

  ChenI. C.HillJ. K.OhlemüllerR.RoyD. B.ThomasC. D. (2011). Rapid range shifts of species associated with high levels of climate warming.Science3331024–1026. 10.1126/science.1206432

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 15

  DasS.BaumgartnerJ. B.Esperon-RodriguezM.WilsonP. D.YapJ. Y. S.RossettoM.et al. (2019). Identifying climate refugia for 30 Australian rainforest plant species, from the last glacial maximum to 2070.Landsc. Ecol.342883–2896. 10.1007/s10980-019-00924-6

  • CrossRef
  • Google Scholar

• 16

  GuisanA.PetitpierreB.BroennimannO.DaehlerC.KuefferC. (2013). Unifying niche shift studies: Insights from biological invasions.Trends Ecol. Evol.29260–269. 10.1016/j.tree.2014.02.009

  • Pubmed Abstract
  • CrossRef
  • Google Scholar
  • View reference in article

• 17

  GutjahrO.PutrasahanD.LohmannK.JungclausJ. H.von StorchJ. S.BrüggemannN.et al. (2019). Max planck institute earth system model (MPI-ESM1.2) for the high-resolution model intercomparison project (HighResMIP).Geoscientific Model Dev.123241–3281. 10.5194/gmd-2018-286

  • CrossRef
  • Google Scholar
  • View reference in article

• 18

  HijmansR. J.BrownA.BarbosaM. (2026). terra: Spatial Data Analysis (R package version 1.9-17).

  • Google Scholar
  • View reference in article

• 19

  IPBES (2019). Summary for Policymakers of the Global Assessment Report on Biodiversity and Ecosystem Services of the IntergovernmentalScience-Policy Platform on Biodiversity and Ecosystem Services, edsDíazS.SetteleJ.BrondízioE. S.NgoH. T.GuèzeM. (Bonn: IPBES secretariat), 56.

  • Google Scholar
  • View reference in article

• 20

  Jiménez-ValverdeA.LoboJ. M. (2007). Threshold criteria for conversion of probability of species presence to either–or presence–absence.Acta Oecol.31361–369. 10.1016/j.actao.2007.02.001

  • CrossRef
  • Google Scholar
  • View reference in article

• 21

  KargerD. N.ConradO.BöhnerJ.KawohlT.KreftH.Soria-AuzaR. W.et al. (2021). Climatologies at High Resolution for the Earth’s Land Surface Areas.Sereetz: EnviDat. 10.16904/envidat.228

  • CrossRef
  • Google Scholar
  • View reference in article

• 22

  KassJ. M.Pinilla-BuitragoG. E.PazA.JohnsonB. A.Grisales-BetancurV.MeenanS. I.et al. (2023). wallace 2: A shiny app for modeling species niches and distributions redesigned to facilitate expansion via module contributions.Ecography2023:e06547. 10.1111/ecog.06547

  • CrossRef
  • Google Scholar
  • View reference in article

• 23

  KeppelG.Van NielK. P.Wardell-JohnsonG. W.YatesC. J.ByrneM.MucinaL.et al. (2012). Refugia: identifying and understanding safe havens for biodiversity under climate change.Glob. Ecol. Biogeogr.21393–404. 10.1111/j.1466-8238.2011.00686.x

  • CrossRef
  • Google Scholar
  • View reference in article

• 24

  Kramer-SchadtS.NiedballaJ.PilgrimJ. D.SchröderB.LindenbornJ.ReinfelderV.et al. (2013). The importance of correcting for sampling bias in MaxEnt species distribution models.Diver. Distributions191366–1379. 10.1111/ddi.12096

  • CrossRef
  • Google Scholar
  • View reference in article

• 25

  LiuC.BerryP. M.DawsonT. P.PearsonR. G. (2005). Selecting thresholds of occurrence in the prediction of species distributions.Ecography28385–393. 10.1111/j.0906-7590.2005.03957.x

  • CrossRef
  • Google Scholar
  • View reference in article

• 26

  LobdellM. S.ThompsonP. G. (2017). “Ex-situ conservation of Quercus oglethorpensis in living collections of arboreta and botanical gardens,” in Proceedings of a Gene Conservation of Tree Species—Banking on the Future. Workshop on Gene Tech. Rep. PNW-GTR-963, Vol. 963edsSniezkoR. A.ManG.HipkinsV.WoesteK.GwazeD.KliejunasJ. T. (Portland, OR: US Department of Agriculture, Forest Service, Pacific Northwest Research Station: 144-153), 144–153.

  • Google Scholar

• 27

  MédailF.DiademaK. (2009). Glacial refugia influence plant diversity patterns in the Mediterranean Basin.J. Biogeogr.361333–1345. 10.1111/j.1365-2699.2008.02051.x

  • CrossRef
  • Google Scholar

• 28

  MuscarellaR.GalanteP. J.Soley-GuardiaM.BoriaR. A.KassJ. M.UriarteM.et al. (2014). ENM eval: An R package for conducting spatially independent evaluations and estimating optimal model complexity for Maxent ecological niche models.Methods Ecol. Evol.51198–1205. 10.1111/2041-210X.12261

  • CrossRef
  • Google Scholar

• 29

  O’NeillB. C.KrieglerE.EbiK. L.Kemp-BenedictE.RiahiK.RothmanD. S.et al. (2017). The roads ahead: Narratives for shared socioeconomic pathways describing world futures in the 21st century.Glob. Environ. Change42169–180. 10.1016/j.gloenvcha.2015.01.004

  • CrossRef
  • Google Scholar

• 30

  PearsonR. G.RaxworthyC. J.NakamuraM.Townsend PetersonA. (2007). Predicting species distributions from small numbers of occurrence records: A test case using cryptic geckos in Madagascar.J. Biogeogr.34102–117. 10.1111/j.1365-2699.2006.01594.x

  • CrossRef
  • Google Scholar

• 31

  PeclG. T.AraújoM. B.BellJ. D.BlanchardJ.BonebrakeT. C.ChenI. C.et al. (2017). Biodiversity redistribution under climate change: Impacts on ecosystems and human well-being.Science355:eaai9214. 10.1126/science.aai9214

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 32

  PhillipsS. J.DudíkM. (2008). Modeling of species distributions with Maxent: New extensions and a comprehensive evaluation.Ecography31161–175. 10.1111/j.0906-7590.2008.5203.x

  • CrossRef
  • Google Scholar

• 33

  R Core Team (2023). R: A Language and Environment for Statistical Computing.Vienna: R Foundation for Statistical Computing.

  • Google Scholar

• 34

  ResideA. E.WelbergenJ. A.PhillipsB. L.Wardell-JohnsonG. W.KeppelG.FerrierS.et al. (2014). Characteristics of climate change refugia for Australian biodiversity.Austral Ecol.39887–897. 10.1111/aec.12146

  • CrossRef
  • Google Scholar

• 35

  RiahiK.Van VuurenD. P.KrieglerE.EdmondsJ.O’neillB. C.FujimoriS.et al. (2017). The shared socioeconomic pathways and their energy, land use, and greenhouse gas emissions implications: An overview.Glob. Environ. Change42153–168. 10.1016/j.gloenvcha.2016.05.009

  • CrossRef
  • Google Scholar

• 36

  RobertsD. R.BahnV.CiutiS.BoyceM. S.ElithJ.Guillera-ArroitaG.et al. (2017). Cross-validation strategies for data with temporal, spatial, hierarchical, or phylogenetic structure.Ecography40913–929. 10.1111/ecog.02881

  • CrossRef
  • Google Scholar

• 37

  RossM. S.StoffellaS. L.RuizP. L.SubediS. C.MeederJ. F.SahJ. P.et al. (2024). Transient vegetation dynamics in a tropical coastal wetland: Sea-level rise, glycophyte retreat, and incipient loss in plant diversity.J. Vegetation Sci.35:e13267. 10.1111/jvs.13267

  • CrossRef
  • Google Scholar

• 38

  SpenceE. S.FantJ. B.GailingO.GriffithM. P.HavensK.HippA. L.et al. (2021). Comparing genetic diversity in three threatened oaks.Forests12:561. 10.3390/f12050561

  • CrossRef
  • Google Scholar

• 39

  SrivastavaA.GrotjahnR.UllrichP. A. (2020). Evaluation of historical CMIP6 model simulations of extreme precipitation over contiguous US regions.Weather Climate Extremes29:100268. 10.1016/j.wace.2020.100268

  • CrossRef
  • Google Scholar

• 40

  SubediS. C.BhandariS.AdhikariB. (2025). Mapping oak diversity and habitat suitability using multi-species niche modeling.Front. Forests Glob. Change8:1624716. 10.3389/ffgc.2025.1624716

  • CrossRef
  • Google Scholar

• 41

  SubediS. C.BhattaraiK. R.ChaudharyR. P. (2015). Distribution pattern of vascular plant species of mountains in Nepal and their fate against global warming.J. Mountain Sci.121345–1354. 10.1007/s11629-015-3495-9

  • CrossRef
  • Google Scholar

• 42

  SubediS. C.DrakeS.AdhikariB.CoggeshallM. V. (2024). Climate-change habitat shifts for the vulnerable endemic oak species (Quercus arkansana Sarg.).J. For. Res.35:23. 10.1007/s11676-023-01673-8

  • CrossRef
  • Google Scholar

• 43

  SubediS. C.HoganJ. A.RossM. S.SahJ. P.BaralotoC. (2019). Evidence for trait-based community assembly patterns in hardwood hammock forests.Ecosphere10:e02956. 10.1002/ecs2.2956

  • CrossRef
  • Google Scholar

• 44

  SubediS. C.RustonB.HoganJ. A.CoggeshallM. V. (2023). Defining the extent of suitable habitat for the endangered Maple Leafoak (Quercus acerifolia).Front. Biogeogr.15:e58763. 10.21425/F5FBG58763

  • CrossRef
  • Google Scholar

• 45

  SubediS. C.SternbergL.DeAngelisD. L.RossM. S.OgurcakD. E. (2020). Using carbon isotope ratios to verify predictions of a model simulating the interaction between coastal plant communities and their effect on ground water salinity.Ecosystems23570–585. 10.1007/s10021-019-00423-4

  • CrossRef
  • Google Scholar

• 46

  SubediS. C.WallsS. C.BarichivichW. J.BoylesR.RossM. S.HoganJ. A.et al. (2022). Future changes in habitat availability for two specialist snake species in the imperiled rocklands of South Florida, USA.Conserv. Sci. Pract.4:e12802. 10.1111/csp2.12802

  • CrossRef
  • Google Scholar

• 47

  ThomasC. D.CameronA.GreenR. E.BakkenesM.BeaumontL. J.CollinghamY. C.et al. (2004). Extinction risk from climate change.Nature427145–148. 10.1038/nature02121

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 48

  UNEP-WCMC, and IUCN. (2023). Protected Planet: The World Database on Protected Areas (WDPA).Cambridge: UNEP-WCMC.

  • Google Scholar

• 49

  UrbanM. C. (2015). Accelerating extinction risk from climate change.Science348571–573. 10.1126/science.aaa4984

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 50

  WarrenD. L.SeifertS. N. (2011). Ecological niche modeling in Maxent: The importance of model complexity and the performance of model selection criteria.Ecol. Appl.21335–342. 10.1890/10-1171.1

  • CrossRef
  • Google Scholar