Ecological niche divergence and specialization in Dianthus pseudocrinitus, a neo-endemic species: ecological evidence challenges sister-taxon delimitation

Abstract

Clarifying species boundaries within taxonomically complex groups like Dianthus is paramount for understanding evolutionary processes and biodiversity. This study employed a spatially explicit, integrative ecological approach to assess the niche dynamics of Dianthus pseudocrinitus Behrooz. & Joharchi, a neo-endemic species with debated taxonomic status, alongside Dianthus crinitus subsp. turkomanicus (Schischk.) Rech. f. and Dianthus orientalis subsp. stenocalyx (Boiss.) Rech. f. in northeastern Iran. Utilizing high-resolution environmental variables (bioclimatic, soil, NDVI, topographic) within ecological niche modeling, multivariate analyses (MANOVA, LDA), hypervolume-based quantifications, and density overlap metrics, I demonstrated notable and statistically significant ecological niche differentiation among the three taxa. These findings suggested soil characteristics as the likely primary drivers of niche segregation, with NDVI also playing a potentially important role. Crucially, D. pseudocrinitus exhibited a substantially narrower ecological niche (478.33 units, 18.23–22.66%) and minimal overlap with its congeners (Jaccard indices ≤ 0.162), providing strong ecological indications of its distinct species status despite morphological similarities and geographic proximity, and supporting its previously described functional divergence. The strong ecological segregation and narrow niche of D. pseudocrinitus relative to its geographically adjacent congeners provide compelling evidence for parapatric speciation. This pattern of strong fine-scale ecological divergence, alongside some broader bioclimatic niche conservatism, highlights the dynamic interplay of factors shaping species boundaries. The results underscore the potential importance of microhabitat specialization and local adaptation in driving diversification in complex montane environments and advocate for integrative taxonomic approaches to resolve intricate delimitation challenges.

Supplementary information

The online version contains supplementary material available at 10.1186/s12862-026-02491-2.

Introduction

Understanding how ecological niches diverge and drive species emergence and distribution is a core pursuit in evolutionary ecology, particularly within landscapes shaped by complex environmental gradients and intricate historical processes [1, 2]. Mountains, with their pronounced topographic relief, environmental heterogeneity, and microhabitat variability, serve as exceptional natural laboratories for studying processes of speciation and ecological differentiation [3, 4]. Within such dynamic systems, closely related taxa may evolve distinct environmental preferences that facilitate coexistence or drive speciation [5], while others may retain overlapping niches due to niche conservatism [6, 7]. However, the relative roles of niche differentiation and conservatism in shaping biodiversity patterns remain context-dependent and frequently unresolved [7–9].

The genus Dianthus L. (Caryophyllaceae), comprising over 300 species across Eurasia and Africa with a major center of diversity in the Mediterranean region [10, 11], exemplifies this complexity. Taxonomic delimitation within Dianthus is notoriously challenging due to high morphological diversity, rapid species radiation, and frequent cases of subtle morphological and low molecular divergence [10, 12–15]. These difficulties are compounded by evolutionary processes such as polyploidy, hybridization, and genome duplication, which obscure phylogenetic relationships and confound traditional classification schemes. Given these challenges, resolving species boundaries within Dianthus necessitates integrative taxonomic frameworks that incorporate detailed morphological data alongside cytological, anatomical, molecular, and increasingly, ecological information [10, 15–21].

Ecological niche theory offers a powerful conceptual lens for addressing these taxonomic ambiguities. In groups where genetic or morphological divergence may be recent or subtle, patterns of ecological niche conservatism or differentiation can provide critical insights into lineage independence, adaptive strategies, and evolutionary trajectories [22–24]. Multidimensional niche analyses, especially when applied in tandem with spatially explicit models, are particularly valuable for clarifying species limits and identifying environmental factors driving diversification [6, 25–27].

The Khorassan–Kopet Dagh floristic region represents one of the major centers of plant diversity within the Irano–Turanian domain, hosting numerous endemic and subendemic taxa across its mountainous landscapes. Its semi-arid to temperate montane climate, pronounced elevational gradients, and heterogeneous rocky–calcareous substrates have produced a fine-scale environmental mosaic that underlies this exceptional level of endemism [28, 29]. Within this region, five closely related Dianthus taxa occur across different mountain systems: D. polylepis subsp. polylepis, D. polylepis subsp. binaludensis, D. crinitus subsp. turkomanicus, D. orientalis subsp. stenocalyx, and D. pseudocrinitus. According to the phylogenetic and morphological assessment of [30], D. pseudocrinitus was described as a newly recognized neo-endemic lineage most closely allied to D. crinitus subsp. turkomanicus and D. orientalis subsp. stenocalyx, consistent with the absence of introgression among these species [31]. Nevertheless, its taxonomic status remains uncertain due to considerable morphological and cytogenetic (2n = 2x = 60) overlap with the D. crinitus subsp. turkomanicus (2n = 2x = 60), despite its ecological similarity and geographic proximity to D. orientalis subsp. stenocalyx (2n = 2x = 30) [16, 30, 32]. To date, the extent of ecological differentiation among these taxa has not been assessed comprehensively using multidimensional environmental datasets or modern ecological niche modeling approaches.

Accordingly, this study focuses on the Khorassan–Kopet Dagh floristic region in northeastern Iran, encompassing the known distribution ranges of the focal taxa (D. pseudocrinitus, D. crinitus subsp. turkomanicus, and D. orientalis subsp. stenocalyx). While some congeners occur in adjacent regions, the populations of these three taxa included in this study represent the full documented range within the province. Although geographically proximal, the taxa exhibit clear ecological differentiation shaped by strong environmental gradients. D. orientalis subsp. stenocalyx is confined to cooler, high-elevation, and stress-prone habitats, whereas D. crinitus subsp. turkomanicus shows broader ecological tolerance, occurring from lower montane slopes to mid-elevation steppes [33, 34] (Fig. 1b). In contrast, D. pseudocrinitus occupies highly disturbed and fragmented anthropogenic habitats, a niche shift potentially facilitated by disturbance regimes. This complex, characterized by overlapping yet environmentally structured habitats and the narrow geographic range of D. pseudocrinitus, provides an ideal model for examining ecological niche divergence and species delimitation within the mountainous Irano–Turanian region [33, 35].

Fig. 1.

Study area and geographical distribution of Dianthus taxa. (a) Geographic location of the study area in northeastern Iran and Turkmenistan. (b) Distribution of three Dianthus taxa in the main mountain systems of the Khorassan–Kopet Dagh (KK) region, including (1) Kopet Dagh and Hezar-Masjed, (2) Binalood, (3) Aladagh and Eastern Alborz. (c) Records and calibration area (M area) for D. crinitus subsp. turkomanicus. (d) Records and calibration area (M area) for D. orientalis subsp. stenocalyx. (e) Records and calibration area (M area) for D. pseudocrinitus

This study is situated within a broader context of recent ecological and functional investigations in the region. Notably, [33, 35] demonstrated that climate, soil conditions, topographic heterogeneity, and NDVI significantly influence the ecological strategies of D. pseudocrinitus and its close relative taxa. Despite their geographic proximity and similarly narrow endemic ranges within the montane steppes of northeastern Iran, these species exhibit contrasting ecological strategies—D. pseudocrinitus tending toward ruderal behavior in disturbed environments, and D. orientalis subsp. stenocalyx displaying traits consistent with stress tolerance in rocky, nutrient-poor habitats [33]. These contrasting strategies likely reflect divergent ecological filtering and recent speciation events, including those driven by polyploidy, common in montane floras experiencing complex environmental gradients [36]. Building upon these insights, this study aims to assess the ecological niche dynamics of D. pseudocrinitus within its phylogenetic context by comparing its environmental preferences and niche breadth to those of its closest relatives. Variables such as elevation, temperature seasonality, precipitation variability, soil nutrient content, and NDVI are known to structure plant distributions in these systems and are expected to underlie observed patterns of niche differentiation [6, 37]. Additional studies of D. polylepis in the same region have confirmed the role of environmental variables, such as temperature seasonality, precipitation regimes, soil fertility, and elevation, in structuring ecological niches of montane Dianthus species (e.g., CCA analysis). These findings support the hypothesis that microhabitat specialization and environmental adaptation, rather than mere geographic isolation, drive niche divergence [31, 33, 38].

I propose that microhabitat specialization and local environmental adaptation, rather than geographical isolation alone, are key drivers of the ecological and taxonomic divergence observed among these taxa. Through this comprehensive, multidimensional analysis, I aim to provide robust ecological evidence to clarify taxonomic boundaries within this morphologically and genetically complex Dianthus group and, more broadly, to contribute to a deeper understanding of the eco-evolutionary processes that promote diversification and endemism in the montane ecosystems of the Irano–Turanian region [28, 29]. Specifically, this study aims to: (1) assess the ecological niche similarity of D. pseudocrinitus with D. crinitus subsp. turkomanicus and D. orientalis subsp. stenocalyx to clarify its taxonomic affinity and species boundaries; (2) evaluate the extent of ecological niche divergence among these taxa to test species-level distinctiveness; (3) identify the key environmental drivers (soil, NDVI, topography, and macroclimate) shaping their niche differentiation using spatial modeling and multivariate statistical approaches; and (4) examine signals of niche conservatism versus ecological shifts to elucidate the ecological mechanisms underlying recent divergence and speciation in montane habitats.

Materials and methods

Plant material and data collection

The occurrence records of D. crinitus subsp. turkomanicus, D. orientalis subsp. stenocalyx, and D. pseudocrinitus were compiled from their natural ranges within the Khorassan–Kopet Dagh (KK) mountains in northeastern Iran and a small area of southern Turkmenistan (Fig. 1a, b). This region corresponds to the Khorassan–Kopet Dagh floristic province, a transitional phytogeographical unit at the intersection of the Irano-Turanian and Central Asian regions [39, 40]. The KK province is characterized by complex topography, climatic heterogeneity, and diverse geological formations that contribute to its outstanding floristic richness [28]. It hosts numerous endemic and subendemic taxa adapted to semi-arid to montane environments and functions both as a center of floristic differentiation and a refugium for Irano-Turanian elements with narrow ecological niches [41]. Owing to its unique geographical position and environmental mosaic, the KK province serves as an ideal natural laboratory for studies of plant diversification, ecological adaptation, and the evolutionary history of southwestern Asian flora [28].

Occurrence data for the three species were obtained from field observations, the literature [17, 42], herbarium collections (FUMH, TARI), GBIF, and SpeciesLink. Initial datasets included 104, 108, and 40 records for D. crinitus subsp. turkomanicus, D. orientalis subsp. stenocalyx, and D. pseudocrinitus, respectively. After removing duplicates, records outside known ranges, and entries lacking coordinates or showing high spatial uncertainty, the datasets were reduced to 90, 68, and 27 points. To reduce spatial clustering and avoid model overfitting, spatial thinning was applied using the spThin package in R (version 3.5.1 [43]), retaining one point per ~1-km grid cell. Final datasets contained 66, 61, and 20 unique occurrences for the three taxa, respectively. These cleaned records were then split into training and evaluation subsets in a 50:50 ratio (see Fig. 1c, d, e).

Plant material was collected from public, non-protected areas in Iran, where non-commercial scientific sampling of non-protected species requires no formal permits. No specimens were taken from protected areas or taxa under conservation regulation.

Environmental predictors and data processing

To develop ecologically robust species distribution models, I assembled a comprehensive set of environmental predictors spanning four categories: climate, vegetation productivity, topography, and soil properties (Table S1). Variable selection was guided by previous studies on Dianthus communities in northeastern Iran, which highlight the strong influence of these environmental domains on species distributions and functional divergence [33, 38, 44]. To standardize spatial resolution and reduce georeferencing uncertainty, all predictors were resampled to 30 arc-seconds (~1 km), following recommended practices in ecological niche modeling [45].

Climate variables were obtained from CHELSA v2.1 [46, 47]. Vegetation productivity was quantified using MODIS V6 NDVI 16-day composites (250 m), with a four-year time series (2018–2021; 78 scenes) capturing seasonal and interannual variation [48, 49]. Soil variables (pH, CEC, total nitrogen, organic carbon) were extracted from SoilGrids at 250 m for the 5–15 cm depth layer [50]. Topographic layers (elevation, slope, aspect) were derived from the SRTM 1 Arc-Second DEM and generated using the Spatial Analyst toolbox in ArcGIS.

To address multicollinearity and reduce dimensionality, I performed separate PCAs for each environmental group, synthesizing correlated variables into independent axes. This approach reduced the original, highly correlated environmental variables into a set of orthogonal Principal Components that captured most of the underlying variance. Using these independent composite predictors in niche modeling improved analytical robustness by minimizing spurious correlations and avoiding truncated or biased niche estimates caused by interdependent variables. The first three components from each PCA, explaining 98.68% (climate), 96.99% (NDVI), 99.69% (topography), and 94.93% (soil) of the variance, were retained, resulting in 12 orthogonal predictors for subsequent niche modeling and ordination analyses (Table S1).

Calibration areas

The selection of an appropriate calibration area is critical in ecological niche modeling, as overly broad regions can artificially inflate model performance metrics [51]. To mitigate this, I defined specific calibration areas, referred to as the accessible area (M), for each three species based on their natural histories. I delineated these M regions (Fig. 1c, d, e) through dispersal simulations using the grinnell package in R [51–53]. These simulations incorporated occurrence data and all 12 environmental predictors at a 30 arc-second spatial resolution. While most grinnell settings remained at their defaults, the kernel spread parameter and simulation period were adjusted to reflect the distinct dispersal capabilities of each subspecies. For D. crinitus subsp. turkomanicus, kernel spread values ranged from 1 to 7, with simulations run over 60-time steps and replicated 10 times. For D. orientalis subsp. stenocalyx, the kernel spread ranged from 1 to 5, with identical temporal and replication settings. In contrast, D. pseudocrinitus was modeled with kernel spread values from 1 to 6, but over a shorter duration of 25-time steps, also replicated 10 times. Final M areas were derived by averaging and binarizing replicate outputs to capture consistent patterns of potential dispersal across simulations [54, 55].

Ecological niche models

The potential distribution of D. crinitus subsp. turkomanicus, D. orientalis subsp. stenocalyx, and D. pseudocrinitus was modeled using Maxent implemented in the kuenm R package [56]. Maxent was selected because it is the most reliable and widely validated method for presence-only data, especially when sample sizes are small, and occurrences show spatial clustering [57, 58]. The modeling workflow consisted of two phases to optimize performance and address limitations associated with small sample sizes, particularly for D. pseudocrinitus.

In the first phase, I generated candidate models using 29 combinations of feature types (l, q, p, t, h) and 13 regularization multipliers (0.25, 0.5, 0.75, 1, 1.25, 1.5, 1.75, 2, 3, 4, 6, 8, 10), exploring a broad spectrum of model complexities. For D. crinitus subsp. turkomanicus and D. orientalis subsp. stenocalyx, all initial models achieved high AUC values ( > 0.9). For D. pseudocrinitus, however, the best model produced a lower AUC of 0.71, indicating weaker discriminatory power and potential overfitting or underfitting due to the limited number of records.

To improve performance for D. pseudocrinitus (20 occurrences), a second modeling phase was conducted with parameter settings tailored for small datasets. I tested nine RM values (0.25–5) and five feature class combinations (l, lq, lqp, lqpt, lqpth), prioritizing simpler features (e.g., l, lq) to minimize overfitting [59] and applying higher regularization values ( > 1) to enhance generalization [60]. Maxent’s ability to assign zero lambda coefficients allowed inclusion of all 12 PCA axes without increasing model complexity [57].

Given the small sample sizes ( < 60 points for all species), I employed jackknife cross-validation [61], excluding one locality at each iteration and testing model accuracy with that point omitted. This procedure generated 377 models for D. crinitus subsp. turkomanicus and D. orientalis subsp. stenocalyx, and 55 models for D. pseudocrinitus. Continuous predictions were converted to binary suitability using the minimum training presence threshold [61]. Model evaluation followed a two-step protocol: (1) statistical significance based on partial ROC [62] at p ≤ 0.05, using the Pearson small-sample test [61] R.G. Pearson, pers. comm.); and (2) among significant models, selection of those with the lowest AICc values (ΔAICc < 2) and omission rates < 0.05 [56, 60].

Final models were constructed using the selected parameter settings and 10 bootstrap replicates, with the “logistic” output format and 10,000 background points. Median values across replicates were retained as the final suitability estimate, and model uncertainty was calculated as the per-pixel range (max–min) of median values [62]. Final replicates were thresholded using an E = 5% omission error [63]. All analyses were performed in kuenm [56, 64] and ArcGIS 10.3.1.

Binary suitability maps derived from the best Maxent models and thresholded using the minimum training presence (MTP) were used to classify suitable areas for each species. Niche overlap among all pairwise species combinations (D. crinitus subsp. turkomanicus vs. D. orientalis subsp. stenocalyx; D. crinitus subsp. turkomanicus vs. D. pseudocrinitus; D. orientalis subsp. stenocalyx vs. D. pseudocrinitus) was then quantified and visualized using Raster Calculator in ArcGIS.

Environmental niche differentiation and ecological niche metrics

A comprehensive ecological niche assessment was conducted to examine niche characteristics and environmental relationships among D. crinitus subsp. turkomanicus, D. orientalis subsp. stenocalyx, and D. pseudocrinitus, using four complementary analytical approaches.

First, all 12 PCA-derived variables (Bioclimate, NDVI, Soil, Topography) were standardized using z-score normalization (scale() function in R [64]). This step ensured an equal contribution from each variable by centering them on a mean of zero and scaling them to unit variance, following best practices in multivariate analyses such as MANOVA and PCA [65, 66]. The normalized dataset was used for MANOVA, univariate ANOVAs, and Tukey’s HSD post hoc tests.

I then performed a MANOVA on the 12 standardized variables to test for overall environmental niche differences among the three species [67]. Before running MANOVA, assumptions of multivariate normality, homogeneity of variance–covariance matrices, and absence of multicollinearity were evaluated. Wilks’ Lambda was used to assess the significance of species effects. Univariate ANOVAs for each variable followed significant MANOVA results and for LD1 and LD2, with Tukey’s HSD applied for pairwise species comparisons when p < 0.05. All MANOVA, ANOVA, and post hoc tests were performed in R (version 4.2.2).

To identify linear combinations of variables that best discriminate among species, I conducted a Linear Discriminant Analysis (LDA) using only variables significantly contributing to species differentiation (pc1_bio, pc3_bio, pc2_NDVI, pc1_soil, pc2_soil, pc3_soil). Variables with zero within-group variance were removed prior to analysis. LDA was implemented using the MASS package in R and produced discriminant functions (LDs), whose significance was tested with Wilks’ Lambda. Variance explained by each LD was calculated, and predictive performance was evaluated via cross-validation. Scatterplots of the LDs were used to visualize species separation in environmental space.

To complement the multivariate analyses, I performed univariate density overlap analysis for each environmental variable. Kernel density estimation was used to model the probability distribution of each variable within each species. Overlaps (0–1) were calculated as the area under the minimum of two density curves, and permutation tests (n = 1000 [9, 25]); assessed significance. This analysis highlighted the contribution of individual variables to niche differentiation. All computations were performed in R using appropriate packages (e.g., overlap).

Finally, ecological niche hypervolumes were constructed for each species using significant environmental variables and Gaussian kernel density estimation in the hypervolume package [68]. Hypervolumes were estimated with 10,000 samples and data-driven bandwidth selection. Pairwise comparisons quantified niche volume, intersection, union, and species-specific unique fractions. Jaccard and Sørensen similarity indices were calculated to measure niche similarity and divergence. Hypervolume boundaries and overlaps were visualized using pairwise scatterplots with kernel contours.

To investigate whether D. pseudocrinitus occupies a unique ecological niche or shares significant environmental space with its congeners, I employed a spatially explicit, integrative approach. I used high-resolution environmental variables, encompassing bioclimatic parameters, vegetation indices (NDVI), soil characteristics, and topographic factors, within an ecological niche modeling (ENM) framework using Maxent [57]. Complementary multivariate statistical analyses, including Multivariate Analysis of Variance (MANOVA) and Linear Discriminant Analysis (LDA), were applied to test for environmental differentiation among taxa. Furthermore, I used hypervolume-based niche modeling and kernel density overlap metrics to quantify niche overlap and identify environmental axes driving niche segregation [27, 68].

Results

Occurrence data and environmental niche modeling for three Dianthus taxa

Accessible areas simulated under both changing frameworks of climatic conditions, including for most of KK for D. crinitus subsp. turkonamicus and the northern parts of KK for D. orientalis subsp. stenocalyx, and D. pseudocrinitus. Fig. 1b shows the georeferenced occurrence locations used in this study and the areas identified by the simulations to be accessible to three Dianthus over time (M).

The contributions of environmental variables to the ecological niche models differed substantially among the three Dianthus taxa (Table S1), and the PCA-derived predictors represent distinct ecological gradients. In D. crinitus subsp. turkomanicus, topography was the dominant predictor: pc2_topography (elevation) contributed 40.2% (PI = 43.9%), and pc1_topography (aspect) contributed 18.8% (PI = 18.9%). Bioclimatic and soil effects were weaker, with pc1_bio (bio1) at 9.8% (PI = 3.8%) and pc3_soil (n) at 6.4% (PI = 8.9%) (Table S1; S2). According to response curves, Suitability increased sharply from ~0.10 to > 0.50 along both pc1_topography (aspect) and pc2_topography (elevation), indicating preference for mid-to-high topographic values (Fig. S1). In D. orientalis subsp. stenocalyx, soil variables were the strongest predictors: pc2_soil (OC) and pc1_soil (CEC) contributed 39.4% (PI = 11%), 12.6% (PI = 0.6), respectively. Bioclimatic factors also influenced its niche, especially pc2_bio (bio12) with 10.5% (PI = 24.1%) and pc1_bio (bio1) with 9.4% (PI = 11.6%). Topography played a smaller role, with pc2_topography (elevation) contributing 11.6% (PI = 14.7%) (Table S1, S2). Suitability increased monotonically along soil gradients (pc1_soil, pc2_soil), rising from < 0.10 at low values to ~0.75–0.80 at high values (Fig. S1). D. pseudocrinitus showed a balanced influence of soil and topography. pc1_soil (CEC) contributed 23.6% (PI = 8.6%), while pc3_topography (slope) contributed 12.1% (PI = 18.5%). Additional effects were observed for pc3_bio (bio2) (9.9% PC, 9.3% PI) and pc2_NDVI (NDVI2-2018) (6.2% PC, 7.6% PI) (Table S1; S2). Suitability declined from ~0.70 to < 0.20 along pc3_topography (slope), indicating sensitivity to fine-scale topography, but increased from < 0.20 to > 0.75 along pc1_soil (CEC), reflecting strong dependence on soil fertility (Fig. S1).

377 candidate models were evaluated for D. crinitus subsp. turkomanicus and D. orientalis subsp. stenocalyx, using combinations of 29 feature classes, 13 regularization multipliers, and one environmental dataset. Of these, 346 models for D. crinitus subsp. turkomanicus and 333 for D. orientalis subsp. stenocalyx performed significantly better than random expectations based on jackknife and partial ROC tests (p < 0.001; Table S3). Final models were selected using AICc: two models met the criteria for each subspecies. The optimal model for D. crinitus subsp. turkomanicus used two feature types (product; product + threshold) with RM = 4, whereas D. orientalis subsp. stenocalyx performed best with three feature types (linear, product, threshold) and RM = 3. For D. pseudocrinitus, 55 models were generated using five feature classes and 11 regularization multipliers; 34 were statistically significant (p < 0.001). Only one met AICc requirements, incorporating linear, quadratic, and product features with RM = 1.25 (Table S3).

High habitat suitability for all three taxa was concentrated in montane regions (Fig. 2a). D. crinitus subsp. turkomanicus showed the highest suitability in Kopet Dagh, Hezar-Masjed, Eastern Alborz, and Binalood, while D. orientalis subsp. stenocalyx and D. pseudocrinitus were most suitable in Kopet Dagh, Aladagh, and Eastern Alborz (Fig. 1b; Fig. 2a, left panels). Model uncertainty was generally low in core high-suitability zones and highest along peripheral and ecotonal boundaries, reflecting environmental heterogeneity or extrapolation (Fig. 2a, middle panels). Binary maps confirmed stable high-suitability areas with low uncertainty across KK mountains (Fig. 2a, right panels).

Fig. 2.

Predicted suitable areas for the distribution of three Dianthus taxa based on Maxent model outputs. (a) Left panels: median predictions of habitat suitability for each taxon. Middle panels: associated prediction uncertainty. Right panels: binary presence–absence maps based on the minimum training presence (MTP) threshold. (b) Overlapping suitable areas for the three Dianthus taxa based on MTP thresholds from the best-performing models

Overlap analysis based on binary maps revealed that D. crinitus subsp. turkomanicus and D. orientalis subsp. stenocalyx shared the greatest extent of suitable habitat, especially in the central and western parts of their predicted ranges. Overlaps involving D. pseudocrinitus were more fragmented and spatially restricted, indicating more specialized habitat preferences. Overall, suitability overlap patterns showed varying degrees of niche sharing among the three Dianthus taxa (Fig. 2b).

Environmental niche differentiation and ecological niche metrics

The MANOVA revealed a highly significant overall species effect on the combined environmental variables (Pillai’s trace = 0.89, F = 8.90, p < 0.001; Table S4), confirming distinct environmental niches among the three Dianthus taxa. Principal component–based MANOVA (Table 1) showed significant interspecific differentiation in bioclimatic (pc1_bio: F = 6.86, p = 0.001; pc3_bio: F = 3.58, p = 0.030), NDVI (pc2_NDVI: F = 6.78, p = 0.001), and especially soil variables, where all components were highly significant (pc1_soil: F = 26.65, p = 1.43e − 10; pc2_soil: F = 9.31, p = 0.0001; pc3_soil: F = 10.77, p = 4.36e − 05). Boxplots (Fig. S2) visually reflected these differences, highlighting species-specific distributions—particularly for pc1_bio and pc1_soil in D. crinitus subsp. turkomanicus and for pc2_soil and pc3_soil in D. orientalis subsp. stenocalyx and D. pseudocrinitus. Tukey’s HSD tests (Table 2) clarified pairwise contrasts: D. crinitus subsp. turkomanicus differed from D. orientalis subsp. stenocalyx in pc1_bio (−0.67, p = 0.001) and pc3_bio (−0.14, p = 0.04), while D. pseudocrinitus did not differ significantly from either taxon in these variables. NDVI (pc2_NDVI) significantly separated D. crinitus subsp. turkomanicus from both D. orientalis subsp. stenocalyx (−1.36, p = 0.003) and D. pseudocrinitus (−1.61, p = 0.022). Soil variables showed the strongest discrimination: pc1_soil differed significantly between D. crinitus subsp. turkomanicus and both D. orientalis subsp. stenocalyx (1.18, p < 0.001) and D. pseudocrinitus (1.33, p < 0.001), with similar patterns for pc2_soil (0.74 and 0.89) and pc3_soil (−0.47 and −0.48; all p < 0.01). No significant soil differences were detected between D. orientalis subsp. stenocalyx and D. pseudocrinitus, indicating overlapping edaphic preferences.

Table 1.

Multivariate analysis of variance (MANOVA) results for environmental predictors (bioclimatic, NDVI, topographic, and soil variables). Asterisks indicate significance levels (*p < 0.05, **p < 0.01, ***p < 0.001)

Variable	Df	Sum.Sq	Mean.Sq	F.value	p-value
pc1_bio	2	12.71	6.35	6.86	0.001***
pc2_bio	2	0.48	0.24	0.24	0.787
pc3_bio	2	6.91	3.46	3.58	0.030*
pc1_NDVI	2	3.03	1.52	1.53	0.221
pc2_NDVI	2	12.57	6.28	6.78	0.001***
pc3_NDVI	2	2.67	1.34	1.34	0.264
pc1_topography	2	0.98	0.49	0.49	0.614
pc2_topography	2	0.45	0.22	0.22	0.802
pc3_topography	2	0.95	0.48	0.47	0.624
pc1_soil	2	39.44	19.7	26.65	1.43e-10***
pc2_soil	2	16.71	8.36	9.31	0.00016***
pc3_soil	2	19.00	9.50	10.77	4.36e-05***

Table 2.

Pairwise comparisons of environmental variables and linear discriminants among Dianthus taxa using Tukey’s HSD post hoc test following ANOVA. The table includes estimated differences (diff), confidence intervals (lwr and upr), adjusted p-values, and corresponding ANOVA p-values. Asterisks indicate significance levels (*p < 0.05, **p < 0.01, ***p < 0.001)

Variable	Comparison	diff	lwr	upr	p adj	ANOVA_p
pc1_bio	D. crinitus subsp. turkomanicus & D. orientalis subsp. stenocalyx	−0.68	−1.12	−0.23	0.001**	0.001***
pc1_bio	D. crinitus subsp. turkomanicus & D. pseudocrinitus	−0.54	−1.18	0.09	0.111	0.001***
pc1_bio	D. orientalis subsp. stenocalyx & D. pseudocrinitus	0.13	−0.51	0.78	0.881	0.001***
pc3_bio	D. crinitus subsp. turkomanicus & D. orientalis subsp. stenocalyx	−0.14	−0.29	−0.00	0.045*	0.030*
pc3_bio	D. crinitus subsp. turkomanicus & D. pseudocrinitus	0.02	−0.18	0.23	0.954	0.030*
pc3_bio	D. orientalis subsp. stenocalyx & D. pseudocrinitus	0.17	−0.04	0.37	0.129	0.030*
pc2_NDVI	D. crinitus subsp. turkomanicus & D. orientalis subsp. stenocalyx	−1.36	−2.35	−0.38	0.003**	0.001***
pc2_ NDVI	D. crinitus subsp. turkomanicus & D. pseudocrinitus	−1.61	−3.03	−0.19	0.021*	0.001***
pc2_ NDVI	D. orientalis subsp. stenocalyx & D. pseudocrinitus	−0.25	−1.68	1.186	0.913	0.001***
pc1_soil	D. crinitus subsp. turkomanicus & D. orientalis subsp. stenocalyx	1.18	0.76	1.61	2.27e-09***	1.43e-10***
pc1_soil	D. crinitus subsp. turkomanicus & D. pseudocrinitus	1.33	0.72	1.94	2.43e-06***	1.43e-10***
pc1_soil	D. orientalis subsp. stenocalyx & D. pseudocrinitus	0.15	−0.47	0.76	0.839	1.43e-10***
pc2_soil	D. crinitus subsp. turkomanicus & D. orientalis subsp. stenocalyx	0.73	0.28	1.20	0.0006***	0.0001***
pc2_soil	D. crinitus subsp. turkomanicus & D. pseudocrinitus	0.89	0.23	1.56	0.005**	0.0001***
pc2_soil	D. orientalis subsp. stenocalyx & D. pseudocrinitus	0.16	−0.51	0.82	0.847	0.0001***
pc3_soil	D. crinitus subsp. turkomanicus & D. orientalis subsp. stenocalyx	−0.47	−0.72	−0.21	9.33e-05***	4.36e-05***
pc3_soil	D. crinitus subsp. turkomanicus & D. pseudocrinitus	−0.48	−0.85	−0.11	0.006**	4.36e-05***
pc3_soil	D. orientalis subsp. stenocalyx & D. pseudocrinitus	−0.02	−0.39	0.36	0.995	4.36e-05***
LD1	D. crinitus subsp. turkomanicus & D. orientalis subsp. stenocalyx	4.09	3.67	4.51	0.00000***	6.10e-53***
LD1	D. crinitus subsp. turkomanicus & D. pseudocrinitus	4.29	3.63	4.84	0.00000***	6.10e-53***
LD1	D. orientalis subsp. stenocalyx & D. pseudocrinitus	0.15	−0.47	0.76	0.839	6.10e-53***
LD2	D. crinitus subsp. turkomanicus & D. orientalis subsp. stenocalyx	−0.23023	−0.65084	0.190385	0.399	0.003**
LD2	D. crinitus subsp. turkomanicus & D. pseudocrinitus	0.650294	0.045816	1.254771	0.031*	0.003**
LD2	D. orientalis subsp. stenocalyx & D. pseudocrinitus	0.880522	0.27031	1.490734	0.002**	0.003**

Linear discriminant analysis (LDA) further supported these findings. Two significant discriminant functions explained 100% of variance (LDA1 = 99.13%, LDA2 = 0.87%), showing that separation is driven by one dominant axis (Fig. 3). LDA1 clearly separated D. crinitus subsp. turkomanicus from the other taxa (vs. D. orientalis subsp. stenocalyx: 4.09, p < 0.001; vs. D. pseudocrinitus: 4.24, p < 0.001), while LDA2 distinguished D. pseudocrinitus from D. crinitus subsp. turkomanicus (0.65, p = 0.031) and D. orientalis subsp. stenocalyx (0.88, p = 0.002). Soil variables dominated the loadings on LDA1, pc1_soil (8.60) and pc2_soil (−8.32), whereas bioclimatic and NDVI variables contributed mainly to LDA2 (e.g., pc3_bio = 0.89), reinforcing the primary role of soil gradients in niche separation (Table S5).

Fig. 3.

Linear discriminant analysis of environmental niches among three Dianthus taxa

Density overlap analysis (Table S6; Fig. 4) revealed high but non-significant overlap in bioclimatic variables (pc1_bio = 0.82, p = 0.125; pc3_bio = 0.83, p = 0.067), suggesting broadly shared climatic conditions. NDVI (pc2_NDVI) exhibited moderately high overlap (0.78) but significant differentiation (p = 0.022). Soil components showed the clearest divergences: pc1_soil had the lowest overlap (0.52, p < 0.001), while pc2_soil (0.73, p = 0.001) and pc3_soil (0.68, p = 0.001) also differed significantly, reaffirming soil as a major axis of niche partitioning.

Fig. 4.

Density overlap of environmental variables among Dianthus taxa based on bioclimatic, NDVI, and soil gradients

Hypervolume analyses (Table 3; Tables S7; Fig. 5; Figs. S3–S4) quantified these patterns. Niche volume varied strongly: D. orientalis subsp. stenocalyx had the largest hypervolume (2570.79 units, 78.71–97.99%), followed by D. crinitus subsp. turkomanicus (1933.12 units, 59.19–91.57%) and D. pseudocrinitus (478.33 units, 18.23–22.65%), indicating much narrower ecological breadth for D. pseudocrinitus (Table 3). The greatest pairwise overlap occurred between D. crinitus subsp. turkomanicus and D. orientalis subsp. stenocalyx (intersection: 1237.90 units; Jaccard = 0.38; Sørensen = 0.55), though both maintained substantial unique fractions (0.21 and 0.41). D. pseudocrinitus showed minimal overlap with D. crinitus subsp. turkomanicus (300.41 units; Jaccard = 0.14; Sørensen = 0.25) and D. orientalis subsp. stenocalyx (425.36 units; Jaccard = 0.16; Sørensen = 0.28). Unique niche fractions were strongly asymmetric: D. crinitus subsp. turkomanicus retained 1632.71 units (0.77) versus D. pseudocrinitus (177.91 units; 0.08), and D. orientalis subsp. stenocalyx retained 2145.43 units (0.82) versus D. pseudocrinitus (52.97 units; 0.02) (Tables S7). Volume ratios highlighted restricted niche breadth for D. pseudocrinitus (4.04 vs. D. crinitus subsp. turkomanicus; 5.37 vs. D. orientalis subsp. stenocalyx) (Table 3).

Table 3.

Hypervolume overlap analysis of ecological niche differentiation among Dianthus taxa. A quantitative comparison of ecological niche volumes and overlaps among three closely related Dianthus taxa

Comparison	D. crinitus subsp. turkomanicus (sp1) & D. orientalis subsp. stenocalyx (sp2)	D. crinitus subsp. turkomanicus (sp1) & D. pseudocrinitus (sp2)	D. orientalis subsp. stenocalyx (sp1) & D. pseudocrinitus (sp2)
Niche Volume_sp1	1933.12	1933.12	2570.79
Niche Volume_sp2	2570.79	478.33	478.33
sp1% in Union	59.19%	91.57%	97.99%
sp2% in Union	78.71%	22.65%	18.23%
Ratio of sp1 and sp2	0.75%	4.04%	5.37%
Intersection	1237.9	300.41	425.36
Union of niche	3266.01	2111.03	2623.76
Unique_sp1	695.22	1632.71	2145.43
Unique_sp2	1332.89	177.91	52.97

Fig. 5.

Pairwise comparisons of ecological niche hypervolumes for D. crinitus subsp. turkomanicus, D. orientalis subsp. stenocalyx, and D. pseudocrinitus

Overall, the results consistently indicate that soil variables represent the primary axis of environmental niche differentiation, NDVI contributes secondarily, and climatic conditions are broadly shared. D. pseudocrinitus occupies a narrow, distinctive niche, whereas the other two taxa exhibit broader, partially overlapping ecological spaces.

Discussion

Spatial patterns and habitat suitability

This study employed an integrative approach to elucidate the ecological niche dynamics of three closely related Dianthus taxa in northeastern Iran: D. pseudocrinitus, D. crinitus subsp. turkomanicus, and D. orientalis subsp. stenocalyx. The findings revealed clear patterns of environmental niche differentiation, particularly influenced by soil characteristics, which provide important ecological context for clarifying taxonomic boundaries within this morphologically and genetically complex group.

The ecological interpretation of PCA-derived predictors clarified how different environmental gradients structure niche differentiation among the three Dianthus taxa [25]. In this framework, topography-related axes primarily summarize elevational and exposure-driven microclimatic gradients, while soil-related axes capture variation in nutrient availability and soil fertility. NDVI and bioclimatic axes, on the other hand, reflect broad patterns of vegetation productivity and broad-scale climatic regimes [4, 69]. In D. crinitus subsp. turkomanicus, the dominance of elevational and aspect-related axes indicated specialization along broad-scale topographic gradients, likely reflecting constraints imposed by temperature regimes, radiation exposure, and associated moisture availability in mountainous environments. In contrast, the niche of D. orientalis subsp. stenocalyx was primarily structured by soil fertility gradients, suggesting stronger edaphic specialization and reduced dependence on topographic heterogeneity. The intermediate niche structure of D. pseudocrinitus, shaped by both soil properties and fine-scale topography, supported its interpretation as an ecologically flexible taxon occupying transitional habitats. Together, these patterns demonstrate that PCA-derived axes capture ecologically meaningful gradients rather than purely statistical constructs, and they highlight contrasting strategies of niche specialization and habitat filtering among closely related taxa [70, 71].

The spatial analyses precisely delineated the potential distributions of three closely related Dianthus taxa in northeastern Iran, revealing significant variation in habitat suitability across environmental gradients within the region. The models consistently predicted high habitat suitability with relatively low model uncertainty within the currently known distribution ranges of all three taxa. These optimal areas, predominantly situated across montane landscapes, represent localized zones with ideal environmental conditions (Fig. 2a). However, certain scattered regions exhibited elevated uncertainty, which may stem from limitations in spatial resolution, inherent environmental heterogeneity, or the scarcity of georeferenced and verified occurrence records. These factors are well-documented to impact the predictive performance of species distribution models [2, 26, 58].

The projection of habitat suitability, particularly when calibrated using the environmental niche of D. pseudocrinitus, provided crucial insights into potential ecological overlap and divergence among the taxa (Fig. 2b). Notably, areas of spatial overlap (depicted in purple) among the predicted niches of species pairs suggest the presence of niche conservatism. This ecological affinity reflects shared environmental preferences and has been extensively documented in other recently diverged or closely related plant taxa [8, 72, 73]. Such patterns may reflect phylogenetic constraints on niche evolution, where closely related species retain similar environmental requirements inherited from a common ancestor [71].

Specifically, the detection of overlapping high-suitability areas within the accessible region (M) of D. orientalis subsp. stenocalyx and D. pseudocrinitus suggests that both taxa may respond similarly to certain environmental drivers, despite their morphological divergence. These overlapping zones could function as potential ecological contact areas where interspecific or inter-subspecific interactions, such as competition, hybridization, introgression, or limited gene flow, could occur, particularly in regions lacking significant physical dispersal barriers [9, 74, 75]. These findings highlight the relevance of integrating spatial ecology with evolutionary frameworks to gain a more comprehensive understanding of species boundaries and the complex processes shaping ecological diversification in montane environments. Furthermore, investigating these contact zones with genetic analyses could provide empirical evidence for the predicted interactions and shed light on the mechanisms driving speciation or maintaining species integrity [76].

Environmental drivers of niche differentiation

By integrating ecological niche modeling, multivariate analyses including MANOVA and LDA, and hypervolume-based niche quantifications, I demonstrated consistent and significant niche differentiation of D. pseudocrinitus from its morphologically and genetically similar congeners, D. crinitus subsp. turkomanicus and D. orientalis subsp. stenocalyx. The Multivariate Analysis of Variance (MANOVA) resulted in unequivocally demonstrated distinct environmental niches among the three Dianthus taxa (p < 0.001). While bioclimatic variables and NDVI showed some significant differentiation, soil variables emerged as the primary drivers of niche separation, exhibiting highly significant differences across all three principal components of soil (pc1_soil, pc2_soil, pc3_soil). This is illustrated by the distinct species-specific distributions in boxplots for soil variables (Fig. S2). The strong influence of soil properties aligns with other studies highlighting the critical role of edaphic factors in shaping plant distributions and facilitating diversification, especially in montane and semi-arid environments where soil heterogeneity can create distinct microhabitats [33, 35, 77]. For instance, research on serpentine endemics often points to the unique soil chemistry as a primary driver of niche differentiation and speciation [77].

Linear Discriminant Analysis (LDA) further reinforced the pronounced environmental distinctiveness of these Dianthus taxa. LDA1, which accounted for 99.13% of the variance, strongly separated D. crinitus subsp. turkomanicus from the other two taxa, while LDA2 primarily distinguished D. pseudocrinitus. The LDA loadings explicitly identified soil variables (pc1_soil, pc2_soil) as the dominant contributors to species separation along LDA1. These statistical results support the interpretation that soil characteristics are the most influential environmental factors driving niche divergence, with bioclimatic and NDVI variables playing a secondary, albeit still significant, role in shaping finer-scale distinctions. These findings indicate that microhabitat specialization and local environmental adaptation, rather than geographic isolation alone, are the main drivers of ecological divergence and species distinctiveness among these taxa.

The density overlap analysis further elucidated patterns of niche partitioning. While some bioclimatic variables showed relatively high overlap, indicating shared climatic conditions, statistically significant differentiation emerged in NDVI and, most strikingly, across all soil principal components. The lowest overlap was observed for pc1_soil (p < 0.001), suggesting clear separation in soil characteristics. This empirical evidence of pronounced environmental preferences directly addresses my aim to evaluate the extent of ecological niche divergence and rigorously test species-level distinctiveness (Purpose 2). It implies that even if these taxa share broad climatic envelopes, their specific ecological boundaries are predominantly defined by soil properties and vegetation characteristics, thereby confirming soil and NDVI as key environmental drivers of segregation (Purpose 3). These findings provide strong support for previous studies indicating that vegetation characteristics and edaphic properties are key drivers of local adaptation and functional divergence (from stress tolerance to ruderal strategies) in D. pseudocrinitus. This is particularly evident in degraded habitats, where anthropogenic disturbances, such as agricultural activities, are likely to exert disproportionately negative effects on the species’ populations [33, 35].

Furthermore, the observed dichotomy between high bioclimatic overlap (suggesting a degree of niche conservatism) and strong differentiation in edaphic and vegetation-related factors (indicating ecological shifts) provides critical insights into the ecological mechanisms behind recent divergence and speciation in these montane habitats (Purpose 4). This pattern is important in understanding how species may coexist in sympatry; subtle differences in resource use or abiotic tolerances, such as soil pH or nutrient availability, can reduce interspecific competition and allow for the stable co-occurrence of closely related species [7, 22, 26, 78].

Ecological niche overlap, specialization, and taxonomic implications

The hypervolume-based niche quantification provided strong quantitative support for the ecological niche differentiation among the three Dianthus taxa, directly addressing the objectives of evaluating species distinctiveness and understanding underlying ecological mechanisms. D. orientalis subsp. stenocalyx exhibited the largest ecological niche volume (2570.79 units; 78.71–97.99% of the union), reflecting a broad ecological amplitude, followed by D. crinitus subsp. turkomanicus (1933.12 units; 59.19–91.57%). In sharp contrast, D. pseudocrinitus displayed a substantially narrower niche (478.33 units; 18.23–22.66%), indicating its occupation of a much more restricted portion of the available environmental space. This pattern was further reinforced by hypervolume ratios: the niche of D. crinitus subsp. turkomanicus was 4.04 times larger than that of D. pseudocrinitus, while D. orientalis subsp. stenocalyx occupied a niche 5.37 times larger, representing the greatest asymmetry observed among the taxa. The strong ecological segregation observed between D. pseudocrinitus and the geographically adjacent D. orientalis subsp. stenocalyx provides compelling evidence for a parapatric mode of speciation. Although the two taxa occur in immediate spatial proximity, the extremely narrow hypervolume niche of D. pseudocrinitus (478.33 units) and the large hypervolume differences (4.04× and 5.37×) indicate that divergent ecological selection overrides potential gene flow. Such asymmetric niche differentiation along fine-scale environmental gradients is a classic signature of parapatric ecological speciation, where adaptation to distinct microhabitats leads to the formation of a narrowly specialized and evolutionarily independent lineage. Together, these contrasts provide compelling evidence that D. pseudocrinitus possesses a markedly restricted ecological niche, likely reflecting a high degree of specialization to specific microclimatic or edaphic conditions. This restriction not only reinforces its unique ecological identity and supports its debated neo-endemic status within the D. crinitus complex (Purposes 1 and 2) but also suggests greater susceptibility to habitat alteration and climatic variability.

This narrowness implies either a more recent divergence into a specialized niche or adaptation to a highly specific and restricted range [23, 24]. The broader niches of the other two taxa indicate greater ecological amplitude and occupancy across more heterogeneous environments, suggesting a more generalist strategy or an older lineage with wider adaptation.

Further strengthening the evidence for species-level distinctiveness and clarifying taxonomic affinities, pairwise comparisons of niche overlap quantified the degree of shared environmental space among the taxa. The highest overlap occurred between D. crinitus subsp. turkomanicus and D. orientalis subsp. stenocalyx, with moderate to high environmental similarity (Jaccard = 0.38, Sørensen = 0.55). Despite this overlap, both taxa retained substantial unique niche components, suggesting shared but partially differentiated ecological spaces. This pattern of partial overlap and unique niche components is common in closely related species that have undergone recent divergence but still share ancestral environmental preferences, a phenomenon often referred to as niche conservatism with some degree of niche evolution [7, 22].

Importantly, D. pseudocrinitus exhibited minimal niche overlap with both congeners (Jaccard = 0.143 with D. crinitus subsp. turkomanicus; Jaccard = 0.162 with D. orientalis subsp. stenocalyx). This ecological separation provides quantitative support for its distinct ecological niche (Purpose 2) despite morphological similarities, thereby robustly resolving ambiguities in its taxonomic placement and clarifying species boundaries.

Hypervolume and LDA analyses indicate that D. pseudocrinitus is environmentally closer to D. crinitus subsp. turkomanicus, showing lower overlap with D. orientalis subsp. stenocalyx. These findings highlight the important role of ecological boundaries in clarifying the taxonomic position of D. pseudocrinitus, and the identification of its narrow, unique niche provides strong support for its species-level distinctiveness.

Moreover, the patterns of limited and unique niche overlap among taxa, particularly for D. pseudocrinitus, indicate a combination of retained ancestral niche components and ecological shifts. Together with the pronounced differences in soil and NDVI, these patterns illuminate the ecological mechanisms underlying recent divergence and species-level differentiation in montane habitats, directly addressing the study objectives of evaluating taxonomic affinities, ecological distinctiveness, and key environmental drivers of niche differentiation.

Limitations of the study

While this study provides strong evidence for ecological niche differentiation among the three Dianthus taxa, two primary limitations should be acknowledged. As with most ecological niche modeling approaches, the associations identified here are interpreted within a correlation-based framework, and their ecological significance is discussed in light of this methodological context.

First, the environmental predictors used, including climate, NDVI, topography, and soil, are constrained by their relatively coarse spatial resolution, particularly for soil variables derived from SoilGrids (250 m). Such broad-scale datasets cannot fully represent fine-scale microhabitat heterogeneity, including edaphic microsite variation, substrate rockiness, localized hydrological patterns, or micro-topographic complexity. These factors may be ecologically significant for narrowly distributed taxa such as D. pseudocrinitus, whose habitat preferences may depend on subtle edaphic or geomorphological features not captured by the available environmental layers. Consequently, while the analyses reliably detect broad-scale patterns of niche differentiation, they may underestimate microhabitat specialization. Future work incorporating high-resolution environmental layers or field-based microhabitat measurements would refine these estimates.

Second, the study relies on static contemporary environmental data, which limits the ability to evaluate temporal aspects of niche evolution. Without integrating paleoclimatic reconstructions or future climate projections, it is not possible to assess historical niche stability, past habitat connectivity, or potential responses of these taxa to ongoing climate change. Incorporating dynamic, time-explicit environmental datasets in future analyses would provide a more comprehensive understanding of long-term niche dynamics and species persistence.

These two limitations represent the most critical considerations for interpreting the results; nonetheless, the current study provides a robust foundation for understanding ecological divergence in this Dianthus complex.

Conclusion

This study employed a robust, integrative ecological approach to rigorously assess the niche dynamics of D. pseudocrinitus, D. crinitus subsp. turkomanicus, and D. orientalis subsp. stenocalyx in northeastern Iran, directly addressing the long-standing taxonomic ambiguities of D. pseudocrinitus. The comprehensive analyses, encompassing ENMs, multivariate statistics (MANOVA, LDA), and quantitative niche metrics (hypervolume and density overlap), indicate consistent and statistically significant environmental niche differentiation among these closely related taxa. Soil characteristics were identified as the main drivers of this ecological segregation, with NDVI also contributing, highlighting the role of fine-scale environmental heterogeneity in shaping species distributions in montane ecosystems.

The results provide ecological support for the distinctiveness of D. pseudocrinitus. Its narrower ecological niche and limited overlap with its congeners support its recognition as an ecologically distinct taxon. This pattern may reflect ecological shifts and microhabitat specialization in D. pseudocrinitus, suggesting that ecological divergence can contribute to species differentiation, even when morphological or genetic differences are subtle, as often seen in rapidly radiating groups like Dianthus. This study emphasizes the value of integrating spatial ecology with evolutionary approaches to better understand species boundaries and diversification processes. Future research should integrate these ecological insights with extensive genetic and genomic analyses to fully disentangle the evolutionary processes, including gene flow, hybridization, and selection, that have driven the observed patterns of niche differentiation and the emergence of species integrity in this intriguing plant group.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

M.B. conceptualized the study, designed the methodology, and collected the data. She performed all analyses and led the manuscript writing process.

Funding

Not applicable.

Data availability

The datasets generated and analyzed during this study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

No specific ethics or collection permits were required by Iranian law for non-commercial sampling of non-protected plant species from public or privately accessible lands, as conducted in this study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.