Morphological, anatomical, and spatial adaptation of Calotropis procera populations across climatic and geographic gradients in Iran Authors: Mohamad Reza Kordasti, Masoud Sheidai, Fahimeh Koohdar Published: 2026 ABSTRACT Calotropis procera (Aiton) W.T.Aiton is a perennial xerophytic shrub widely distributed in arid and semi-arid regions of Iran. Understanding the spatial and phenotypic variation of morphological and anatomical traits is crucial for evaluating adaptive capacity under environmental gradients. We investigated leaf morphological and anatomical variation across 12 populations of C. procera, integrating multivariate analyses, spatial modeling, reaction norm mixed models (RRMM), and landscape adaptive indices (LAI). Morphological and anatomical traits exhibited significant population differentiation and phenotypic plasticity. Spatial principal component analysis (sPCA) revealed strong spatial structuring of traits along climatic and geographic gradients. Redundancy analysis (RDA) and canonical correspondence analysis (CCA) showed significant associations between traits and environmental variables. Kriging of LAI highlighted heterogeneous adaptive potential, with high-LAI populations serving as genetic and adaptive reservoirs, while low-LAI populations were vulnerable. Our findings provide a comprehensive framework for assessing phenotypic adaptation and guiding conservation of xerophytic species in arid landscapes.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12870-026-09048-x. Introduction Calotropis procera (Aiton) W.T.Aiton (Apocynaceae) is a xerophytic perennial shrub native to arid and semi-arid regions of Asia and Africa, including Iran, and has also colonized parts of Brazil’s Caatinga biome [1, 2]. The species typically grows to 2.5–4 m in height and features a deep taproot system, pubescent young stems, and thick, light-green leaves. Its copious latex and high concentrations of steroidal cardenolides provide strong defense against herbivory, while morphological and anatomical traits such as a thick cuticle, well-developed vascular bundles, and specialized stomatal structures enable it to tolerate high temperatures, drought, salinity, and intense solar radiation [2, 3]. Climate change is intensifying abiotic stress in dryland ecosystems worldwide, highlighting the need to understand plant adaptive mechanisms that maintain fitness under heterogeneous and changing environmental conditions. Leaf morphological and anatomical traits play a central role in regulating water-use efficiency, thermal balance, and stress resistance. However, little is known about how these traits vary and are spatially structured across climatic gradients and geographic populations of C. procera [4, 5]. These studies provide evidence for both genomic regions linked to functional traits and environmentally associated genomic offsets, underlining the species’ ecological resilience and potential for adaptive responses. Phenotypic differentiation among populations can result from genetic divergence, phenotypic plasticity, or their interaction with environmental heterogeneity. Spatially structured environmental gradients, such as those associated with latitude, longitude, and climate, can produce gradual trait clines or localized divergence, reflecting adaptive responses or plastic adjustments to contrasting conditions [5]. To disentangle these processes, integrative analytical frameworks that combine multivariate trait variation, spatial modeling, and population-level responses to environmental drivers are required, enabling a deeper understanding of both genetic and phenotypic contributions to resilience in C. procera. In this study, we assessed leaf morphological and anatomical variation and spatial structuring across twelve geographically distinct populations of C. procera in Iran. We applied univariate and multivariate statistical approaches, including ANOVA (univariate), MANOVA (multivariate), ordination methods (RDA, CCA, FAMD), spatial analyses (sPCA, variograms), and reaction-norm mixed models (RRMM), to quantify both plastic and genetically structured responses to geographic (latitude and longitude) and climatic variables. Furthermore, we evaluated population vulnerability and sensitivity using an integrative approach that combines morphological, anatomical, spatial, and genetic dimensions. Our findings provide novel insights into the adaptive potential of C. procera and inform conservation planning in arid landscapes under ongoing climate change. Materials and methods Plant m‌aterials In total, 120 samples of C. procera were studied from twelve geographical populations in Iran (Table 1). The maps related to geographical and climatic features of Iran are presented in Fig. 1A and B, and the distribution area of this study in Fig. 1C. These plants were used for morphological and anatomical measurements. Morphological traits were measured using stereo microscopes and a grid-paper method. Traits included leaf base, leaf stalk, leaf thickness, leaf length-to-width ratio, internode length, peduncle length, cyathium tail, total cyathium length, follicle tail, follicle size, seed size, and seed pappus.Table 1Collection sites, altitudes, and herbarium voucher codes of the studied populationsPopulationLocalityAlt (m)Voucher number (Herbarium: HSBU)1Sistanbalochesatan, Pashamag635HSBU-20221012Boshehr,Ab pakhsh33HSBU-20221023Kerman,kahnoje508HSBU-20221034Hormozgan,Bandar jask6HSBU-20221045Khozestan,Dezfol117HSBU-20221056Hormozgan, minab30HSBU-20221067Kerman, Bam1074HSBU-20221078KHozestan, Shoshtar65HSBU-20221089Fars, girkarzin870HSBU-202210910Boshehr,Samal72HSBU-202210111Hormozgan,goarzonak20HSBU-202211012Sistanbalochesatan,Chabahar45HSBU-2022112Fig. 1Iran maps show climatic conditions (A), land cover (B), and geographical distribution of the C. procera in Iran (C) Table 1Collection sites, altitudes, and herbarium voucher codes of the studied populationsPopulationLocalityAlt (m)Voucher number (Herbarium: HSBU)1Sistanbalochesatan, Pashamag635HSBU-20221012Boshehr,Ab pakhsh33HSBU-20221023Kerman,kahnoje508HSBU-20221034Hormozgan,Bandar jask6HSBU-20221045Khozestan,Dezfol117HSBU-20221056Hormozgan, minab30HSBU-20221067Kerman, Bam1074HSBU-20221078KHozestan, Shoshtar65HSBU-20221089Fars, girkarzin870HSBU-202210910Boshehr,Samal72HSBU-202210111Hormozgan,goarzonak20HSBU-202211012Sistanbalochesatan,Chabahar45HSBU-2022112 Collection sites, altitudes, and herbarium voucher codes of the studied populations Iran maps show climatic conditions (A), land cover (B), and geographical distribution of the C. procera in Iran (C) Anatomical traits were assessed from transverse sections prepared with double staining using Congo Red and Methylene Blue. High-resolution images of the stained sections were captured and analyzed using ImageJ software (version 1.53t). Measured anatomical traits included cuticle thickness, upper epidermis, collenchyma, cortex, and lamina. Multiple replicates were measured per individual to ensure accuracy and reliability. At the same time, the genetic variability data were obtained from our previous report (Table 2, Table S1) [6]. The plant materials were taxonomically identified by Mohamad Reza Kordasti, and voucher specimens have been deposited in the Herbarium of Shahid Beheshti University (HSBU) Voucher codes (HSBU-2022101 to HSBU-2022112) are provided in Table 1.Table 2Data set contain genetic diversity parameters, resistance (R1-R12), soil features and bioclimate data (C1-C12) across Caloteropis populations studiedPopulationNeIHe%PPop11.3680.3390.22366.67%Pop21.3420.2830.19251.28%Pop31.3100.2770.18453.85%Pop41.1500.1380.09125.64%Pop51.2480.2270.15043.59%Pop61.3720.3120.21256.41%Pop71.2570.2080.14335.90%Pop81.2790.2550.16751.28%Pop91.3160.2960.19164.10%Pop101.1550.1300.08823.08%Pop111.1850.1730.11335.90%Pop121.3010.2690.17658.97%Abbreviations: Ne = Effective number of alleles, I = Shannon Information index (I), He = Gene diversity, %P = Percentage of polymorphic loci Table 2Data set contain genetic diversity parameters, resistance (R1-R12), soil features and bioclimate data (C1-C12) across Caloteropis populations studiedPopulationNeIHe%PPop11.3680.3390.22366.67%Pop21.3420.2830.19251.28%Pop31.3100.2770.18453.85%Pop41.1500.1380.09125.64%Pop51.2480.2270.15043.59%Pop61.3720.3120.21256.41%Pop71.2570.2080.14335.90%Pop81.2790.2550.16751.28%Pop91.3160.2960.19164.10%Pop101.1550.1300.08823.08%Pop111.1850.1730.11335.90%Pop121.3010.2690.17658.97%Abbreviations: Ne = Effective number of alleles, I = Shannon Information index (I), He = Gene diversity, %P = Percentage of polymorphic loci Data set contain genetic diversity parameters, resistance (R1-R12), soil features and bioclimate data (C1-C12) across Caloteropis populations studied Abbreviations: Ne = Effective number of alleles, I = Shannon Information index (I), He = Gene diversity, %P = Percentage of polymorphic loci Data bootstrapping and collinearity test To ensure robust estimation of parameter distributions and account for potential sampling biases, bootstrap resampling was performed across a range of sample sizes (100–2700, in steps of 50) R package “boot” [7]. Resampled datasets were pooled and used for model fitting. Trait collinearity was assessed using Factor Analysis of Mixed Data (FAMD) via R package “FactoMineR” [8] and correlation coefficient estimation. Only traits with non-significant collinearity were included in subsequent analyses. Geographic distance matrix calculation Pairwise geographic distances among C. procera sampling plots were calculated from latitude and longitude coordinates using the Haversine formula, which accounts for the Earth’s curvature. Distances were computed with the distm() function from the geosphere R package [9] and converted to kilometers for standardization. The resulting distance matrix was used as input for spatial modeling of morphological and anatomical traits. All analyses were conducted in R version 4.3 using the geosphere [9] and spatstat [10] packages. Spatial principal component analysis (sPCA) Spatial PCA (sPCA) was applied to investigate spatial structuring of morphological and anatomical variation using the adegenet [11] and ade4 [12] packages [11]. sPCA identifies synthetic variables (principal components, PCs) that maximize both variance and spatial autocorrelation (Moran’s I). PCs reveal global structures (positive autocorrelation, e.g., clines or patches) and local structures (negative autocorrelation, where neighboring individuals are more distinct than random expectation). Bootstrap resampling generated robust datasets for analysis. Monte Carlo permutation tests (n = 999) assessed the significance of global and local structures using global.rtest() and local.rtest(), as described in Sheidai et al. (2016) [13] for application of sPCA in plant population studies. Positive spatial correlations (global structures) indicate similarity among neighboring individuals, while negative correlations (local structures) reflect greater-than-expected dissimilarity among neighbors. These patterns were assessed using Moran’s I on the sPCA-derived principal components. Isolation by distance (IBD) and Moran’s I Mantel tests evaluated correlations between pairwise geographic distances and trait-based distances derived from sPCA using the vegan R package [14]. Genetic distances were calculated from the sPCA trait matrix (spca-res$tab). Moran’s I quantified spatial autocorrelation along the first two sPCA axes, using k-nearest neighbors (k = 2) to define spatial relationships. The neighbor list was converted into a spatial weight matrix with nb2listw() from the spdep R package [15], and significance was tested using moran.test. In addition to Isolation by Distance (IBD), we tested for Isolation by Environment (IBE) to evaluate the influence of environmental differences on phenotypic divergence. IBE refers to increasing differentiation between populations associated with differences in environmental conditions, such as temperature, humidity, or soil properties, regardless of geographic distance. Pairwise environmental distance matrices were constructed from the measured climatic and edaphic variables and used in Mantel tests to assess the effect of environmental heterogeneity on trait variation. Spatial dependency of traits and geostatistical prediction Empirical variogram modeling quantified the spatial dependence of morphological and anatomical traits, with three theoretical models tested: Spherical, Exponential, and Gaussian. Semi-variance values were calculated from squared differences between trait values across plot pairs, binned into 11 equal-width intervals. Model parameters (nugget, sill, range) were estimated via BFGS optimization in the stats R package. Model fit was assessed with R2, and spatial structure intensity was quantified with the Spatial Dependence Index (SDI = (1 − nugget/sill) × 100). Detailed calculation procedures and model parameters are provided in the Supplementary Materials. Kriging analysis Ordinary kriging was applied for spatial interpolation of genetic diversity and Leaf Area Index (LAI) using the spmodel package [16]. Predictions were generated on a 50 × 50 grid. Kriging provides the best linear unbiased prediction (BLUP) at unsampled locations, accounting for spatial covariance [17, 18]. Modeling trait responses using random regression mixed models (RRMM) Random regression mixed models quantified population-level responses of morphological and anatomical traits to climate change, using ΔTemperature and ΔHumidity as environmental predictors (Δ = future − current values; WorldClim, [19]). Population identity was treated as a random factor with population-specific intercepts and slopes, and models were implemented using the lme4 R package [20]: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}\mathrm{Trait}_{ij} = \beta_0 +& \beta_1 \Delta \mathrm{Temp}_{ij} + \beta_2 \Delta \mathrm{Humidity}_{ij}\\& + u_{0j} + u_{1j}\Delta \mathrm{Temp}_{ij} \\& + u_{2j}\Delta \mathrm{Humidity}_{ij} + \varepsilon_{ij}\end{aligned}$$\end{document} where β are fixed effects, u are population-level random effects, and ε is residual error. Models were fit with lmer() from the lme4 R package [20] in R 4.3, and population-specific reaction norms were visualized using ggplot2. Model selection was based on AIC, and assumptions of residual normality and homoscedasticity were checked. Least-cost distance, habitat fragmentation, and genetic corridor modeling Least-cost distance analysis identified movement paths with minimum cumulative cost between populations using rasterized resistance surfaces in R (raster [21], sf [22], gdistance [23], geosphere [9], sp [24]). Habitat Fragmentation Index (HFI) quantified spatial isolation based on mean Euclidean distance to the three nearest neighbors, reflecting the degree of fragmentation and potential barriers to gene flow [25, 26]. Genetic corridors were modeled using accumulated cost surfaces (accCost()) to identify low-resistance paths facilitating gene flow. Spatially explicit resistance and population points were used to generate maps of potential connectivity. Quantifying variation in genetic connectivity: CIP network metrics Covariance-Informed Population (CIP) networks captured genetic covariance among populations. Node-based metrics included closeness centrality, betweenness centrality, mean inverse edge weight (MIW), and clustering coefficient. Two undirected, weighted CIP networks were constructed: one based on genetic distances (Nei’s) and one based on landscape resistance distances. Network congruence was assessed with Mantel tests (999 permutations), and node-level metrics were compared using Spearman correlations. Visualization of networks employed ggraph [27] and tidygraph [28] with Fruchterman-Reingold layout. Phenotypic plasticity and population differentiation ANOVA and MANOVA indicated highly significant inter-population variation (p = 0.01), highlighting pronounced phenotypic divergence. Association studies Redundancy analysis (RDA) and Canonical Correspondence Analysis (CCA), supported by 999 permutations, confirmed significant associations between traits and environmental gradients (p < 0.01) using the vegan R package [14]. FAMD explained ~ 45% of morphological variance and ~ 52% of anatomical variance, with key traits (e.g., cyathium length, stomatal pore size, epidermal thickness) correlated with latitude, longitude, temperature, and humidity gradients using FactoMineR [8]. Mantel tests indicated significant isolation by distance, confirming spatial structuring of phenotypic variability in C. procera. Results Morphological and anatomical phenotypic plasticity and population differentiation Comparisons of morphological and anatomical traits across twelve geographical populations of Calotropis procera are presented in Tables 3, 4 and 5. Representative box plots illustrate both within-population variability (phenotypic plasticity) and among-population differentiation for selected morphological and anatomical characters (Figs. 2 and 3).Table 3Morphologcall traits mean and standard deviation across C. procera populations studiedPopulationleaf baseleaf stalkleaf thicknessL/W leafinternodepedunclecyathyum tailTotal cyathyumfollicele tailfollicele sizeseed sizeseed pappus111 ± 02 ± 01.33 ± 0.582 ± 0.48 ± 1.732.07 ± 0.128.33 ± 2.5216 ± 4.362 ± 07 ± 0.50.6 ± 02 ± 0222 ± 01 ± 02 ± 01.5 ± 04 ± 02 ± 06.5 ± 014 ± 02 ± 06 ± 00.6 ± 01.5 ± 0331 ± 01.33 ± 0.581.67 ± 0.581.67 ± 0.067.33 ± 2.022.33 ± 0.159 ± 215.67 ± 1.153.17 ± 0.296 ± 00.6 ± 01.5 ± 0441 ± 01 ± 01 ± 01.83 ± 0.066.43 ± 0.123.03 ± 0.069 ± 124 ± 12.5 ± 0.16.47 ± 0.060.75 ± 0.052.47 ± 0.06551 ± 01 ± 02 ± 01.7 ± 0.13 ± 01.6 ± 0.15.83 ± 0.2913.33 ± 2.311.93 ± 0.126.17 ± 0.290.73 ± 0.122.13 ± 0.12661.33 ± 0.581.67 ± 0.581 ± 01.73 ± 0.296.5 ± 1.732.07 ± 0.126.33 ± 1.1515.33 ± 2.892.87 ± 0.126 ± 00.6 ± 02.77 ± 0.46771.67 ± 0.581 ± 02 ± 01.67 ± 0.125.33 ± 0.581.83 ± 0.297.17 ± 1.0413.67 ± 0.582.5 ± 06.67 ± 0.580.73 ± 0.122.67 ± 0.29881 ± 01 ± 01 ± 01.87 ± 0.156 ± 1.731.73 ± 0.067 ± 111.33 ± 2.312.33 ± 0.297 ± 10.73 ± 0.122.67 ± 0.58992 ± 01 ± 01.67 ± 0.581.3 ± 0.173.5 ± 0.871.77 ± 0.064.33 ± 0.2911 ± 1.732.33 ± 0.297 ± 00.8 ± 01.73 ± 0.2310101.67 ± 0.582 ± 02 ± 01.5 ± 0.12.33 ± 0.581.6 ± 0.14.67 ± 0.5810 ± 12.07 ± 0.128.33 ± 0.580.8 ± 03 ± 111111.33 ± 0.581 ± 01.33 ± 0.581.73 ± 0.255.17 ± 0.292.1 ± 0.569.33 ± 1.1515 ± 12.43 ± 0.127.33 ± 0.580.8 ± 03.07 ± 0.1212121 ± 01 ± 01 ± 02.07 ± 0.214.67 ± 0.581.7 ± 0.267.33 ± 1.5312 ± 12.43 ± 0.127.33 ± 0.580.8 ± 03.23 ± 0.25Units: leaf, stem, and internode = cm; seed and pappus = mm; ratios and counts are unitlessTable 4Preliminary statistics of morphological traits studiedPopulationleaf baseleaf stalkleaf thicknessL/W leafinternodepedunclecyathyum tailTotal cyathyumfollicele tailfollicele sizeseed sizeseed pappusMin11111.221.5491.860.61.5Max122222.49.53.111253.590.84Sum23448455461.7186.871.5254.551485.7243.925.6586.2Mean6.51.3333331.251.51.7138895.1888891.9861117.06944414.277782.3805566.7750.71252.394444Std. error0.58350340.079681910.073192510.084515430.043852710.32324690.071063660.31819980.64808770.063598370.13149780.016351180.1119059Variance12.257140.22857140.19285710.25714290.069230163.7615870.18180163.6450415.120630.14561110.62250.0096250.4508254Stand. dev3.501020.47809140.4391550.50709260.26311621.9394810.4263821.9091993.8885260.38159020.78898670.098107080.6714353Median6.5111.51.7527142.56.750.82.45Coeff. var53.8618535.8568635.132433.8061715.35237.3775821.4681827.0063527.2348116.0294611.6455613.7694228.04138Table 5Anatomical traits mean and standard deviation across C. procera populations studiedPopulationCuticule thicnessupper epidermiscollenchymacortexlamina150.8 ± 17.5117.8 ± 52.48398.92 ± 17.37963.43 ± 730.641084.55 ± 156.89249.67 ± 4.1865.74 ± 6.42382.5 ± 45.311149.17 ± 269.91306.69 ± 95.75322.94 ± 5.6931.41 ± 2.8690.72 ± 5.67280.77 ± 2.85209.54 ± 17.01452.1 ± 17.3686.28 ± 0.91374.99 ± 78.18780.79 ± 208.33829.93 ± 69.69517.01 ± 025.49 ± 2.83125.14 ± 45.36391.18 ± 85.03277.86 ± 51.02617.01 ± 026.46 ± 6.55117.16 ± 54.47266.44 ± 105.9213.53 ± 41.79751.82 ± 1.1574.99 ± 17.91301.54 ± 79.131395.51 ± 182.071236.89 ± 247.15824.57 ± 3.2739.68 ± 5.67117.16 ± 13.09334.47 ± 76.69247.54 ± 17.32919.58 ± 2.8736.75 ± 8.5141.53 ± 11.34257 ± 14.27272.34 ± 34.021028.87 ± 0.9139.9 ± 0.37157.8 ± 22.73476.68 ± 68.03304.24 ± 28.531117.69 ± 1.1828.45 ± 5.67144.63 ± 2.84276.96 ± 70.95226.61 ± 34.011220.79 ± 11.834.01 ± 19.6490.7 ± 19.64311.79 ± 66.84213.53 ± 55.93Populationupper epidermislower epidermismidribscleranchymapithlaticifer cellouter pheloem125.81 ± 13.29151.74 ± 19.03806.71 ± 100.423819.44 ± 3646.37659.72 ± 435.07688.66 ± 170.443.92 ± 5.84213.85 ± 0.27113.17 ± 8.7731.82 ± 57.98866.76 ± 251.121149.48 ± 473.582680.27 ± 1061.8851.22 ± 10.21310.36 ± 0.931.45 ± 2.87107.25 ± 14.769270.83 ± 4090.78902.78 ± 433.68896.99 ± 236.9975.85 ± 14.93415.79 ± 0.53112.56 ± 8.69487.75 ± 2.835214.12 ± 70.161053.24 ± 280.661267.36 ± 30.0759.08 ± 2.3258.38 ± 0.6336.75 ± 2.84141.72 ± 05306.71 ± 905.831186.34 ± 70.16688.66 ± 178.1847.06 ± 6.5768.53 ± 2.4243.46 ± 16.36147.39 ± 54.0810,925.93 ± 5692.86868.06 ± 125.191030.09 ± 411.3339.77 ± 8.89715.01 ± 1.21126.42 ± 13.93597.36 ± 165.071889.76 ± 346.02267.84 ± 89.28855.6 ± 71.7547.51 ± 12.27813.74 ± 2.0841.57 ± 6.55251.32 ± 27.9622,189.41 ± 19,948.572295.78 ± 1967.52778.17 ± 2842.3681.89 ± 36.84912.89 ± 2.7431.22 ± 2.84198.72 ± 17.025734.95 ± 4135.35706.02 ± 207.12896.99 ± 323.0943.35 ± 10.511014.75 ± 0.9943.08 ± 8.17238.62 ± 31.444282.41 ± 320.75422.45 ± 50.12312.5 ± 90.2138.08 ± 14.19119.59 ± 2.1534.01 ± 0124.72 ± 09751.16 ± 1726.911001.16 ± 325.88821.76 ± 595.749.98 ± 8.551211.29 ± 8.7130.23 ± 8.66160.62 ± 29.096516.2 ± 1899.94387.73 ± 184.01266.2 ± 140.3371.85 ± 25.03Fig. 2Box plots of representative morphological characters show within-population trait plasticity and among-population differences. The values below the plots show the plasticity value of the traitFig. 3Box plots of representative anatomical characters show within-population trait plasticity and among-population differences. The values below the plots show the plasticity value of the trait Table 3Morphologcall traits mean and standard deviation across C. procera populations studiedPopulationleaf baseleaf stalkleaf thicknessL/W leafinternodepedunclecyathyum tailTotal cyathyumfollicele tailfollicele sizeseed sizeseed pappus111 ± 02 ± 01.33 ± 0.582 ± 0.48 ± 1.732.07 ± 0.128.33 ± 2.5216 ± 4.362 ± 07 ± 0.50.6 ± 02 ± 0222 ± 01 ± 02 ± 01.5 ± 04 ± 02 ± 06.5 ± 014 ± 02 ± 06 ± 00.6 ± 01.5 ± 0331 ± 01.33 ± 0.581.67 ± 0.581.67 ± 0.067.33 ± 2.022.33 ± 0.159 ± 215.67 ± 1.153.17 ± 0.296 ± 00.6 ± 01.5 ± 0441 ± 01 ± 01 ± 01.83 ± 0.066.43 ± 0.123.03 ± 0.069 ± 124 ± 12.5 ± 0.16.47 ± 0.060.75 ± 0.052.47 ± 0.06551 ± 01 ± 02 ± 01.7 ± 0.13 ± 01.6 ± 0.15.83 ± 0.2913.33 ± 2.311.93 ± 0.126.17 ± 0.290.73 ± 0.122.13 ± 0.12661.33 ± 0.581.67 ± 0.581 ± 01.73 ± 0.296.5 ± 1.732.07 ± 0.126.33 ± 1.1515.33 ± 2.892.87 ± 0.126 ± 00.6 ± 02.77 ± 0.46771.67 ± 0.581 ± 02 ± 01.67 ± 0.125.33 ± 0.581.83 ± 0.297.17 ± 1.0413.67 ± 0.582.5 ± 06.67 ± 0.580.73 ± 0.122.67 ± 0.29881 ± 01 ± 01 ± 01.87 ± 0.156 ± 1.731.73 ± 0.067 ± 111.33 ± 2.312.33 ± 0.297 ± 10.73 ± 0.122.67 ± 0.58992 ± 01 ± 01.67 ± 0.581.3 ± 0.173.5 ± 0.871.77 ± 0.064.33 ± 0.2911 ± 1.732.33 ± 0.297 ± 00.8 ± 01.73 ± 0.2310101.67 ± 0.582 ± 02 ± 01.5 ± 0.12.33 ± 0.581.6 ± 0.14.67 ± 0.5810 ± 12.07 ± 0.128.33 ± 0.580.8 ± 03 ± 111111.33 ± 0.581 ± 01.33 ± 0.581.73 ± 0.255.17 ± 0.292.1 ± 0.569.33 ± 1.1515 ± 12.43 ± 0.127.33 ± 0.580.8 ± 03.07 ± 0.1212121 ± 01 ± 01 ± 02.07 ± 0.214.67 ± 0.581.7 ± 0.267.33 ± 1.5312 ± 12.43 ± 0.127.33 ± 0.580.8 ± 03.23 ± 0.25Units: leaf, stem, and internode = cm; seed and pappus = mm; ratios and counts are unitless Morphologcall traits mean and standard deviation across C. procera populations studied Units: leaf, stem, and internode = cm; seed and pappus = mm; ratios and counts are unitless Table 4Preliminary statistics of morphological traits studiedPopulationleaf baseleaf stalkleaf thicknessL/W leafinternodepedunclecyathyum tailTotal cyathyumfollicele tailfollicele sizeseed sizeseed pappusMin11111.221.5491.860.61.5Max122222.49.53.111253.590.84Sum23448455461.7186.871.5254.551485.7243.925.6586.2Mean6.51.3333331.251.51.7138895.1888891.9861117.06944414.277782.3805566.7750.71252.394444Std. error0.58350340.079681910.073192510.084515430.043852710.32324690.071063660.31819980.64808770.063598370.13149780.016351180.1119059Variance12.257140.22857140.19285710.25714290.069230163.7615870.18180163.6450415.120630.14561110.62250.0096250.4508254Stand. dev3.501020.47809140.4391550.50709260.26311621.9394810.4263821.9091993.8885260.38159020.78898670.098107080.6714353Median6.5111.51.7527142.56.750.82.45Coeff. var53.8618535.8568635.132433.8061715.35237.3775821.4681827.0063527.2348116.0294611.6455613.7694228.04138 Preliminary statistics of morphological traits studied Table 5Anatomical traits mean and standard deviation across C. procera populations studiedPopulationCuticule thicnessupper epidermiscollenchymacortexlamina150.8 ± 17.5117.8 ± 52.48398.92 ± 17.37963.43 ± 730.641084.55 ± 156.89249.67 ± 4.1865.74 ± 6.42382.5 ± 45.311149.17 ± 269.91306.69 ± 95.75322.94 ± 5.6931.41 ± 2.8690.72 ± 5.67280.77 ± 2.85209.54 ± 17.01452.1 ± 17.3686.28 ± 0.91374.99 ± 78.18780.79 ± 208.33829.93 ± 69.69517.01 ± 025.49 ± 2.83125.14 ± 45.36391.18 ± 85.03277.86 ± 51.02617.01 ± 026.46 ± 6.55117.16 ± 54.47266.44 ± 105.9213.53 ± 41.79751.82 ± 1.1574.99 ± 17.91301.54 ± 79.131395.51 ± 182.071236.89 ± 247.15824.57 ± 3.2739.68 ± 5.67117.16 ± 13.09334.47 ± 76.69247.54 ± 17.32919.58 ± 2.8736.75 ± 8.5141.53 ± 11.34257 ± 14.27272.34 ± 34.021028.87 ± 0.9139.9 ± 0.37157.8 ± 22.73476.68 ± 68.03304.24 ± 28.531117.69 ± 1.1828.45 ± 5.67144.63 ± 2.84276.96 ± 70.95226.61 ± 34.011220.79 ± 11.834.01 ± 19.6490.7 ± 19.64311.79 ± 66.84213.53 ± 55.93Populationupper epidermislower epidermismidribscleranchymapithlaticifer cellouter pheloem125.81 ± 13.29151.74 ± 19.03806.71 ± 100.423819.44 ± 3646.37659.72 ± 435.07688.66 ± 170.443.92 ± 5.84213.85 ± 0.27113.17 ± 8.7731.82 ± 57.98866.76 ± 251.121149.48 ± 473.582680.27 ± 1061.8851.22 ± 10.21310.36 ± 0.931.45 ± 2.87107.25 ± 14.769270.83 ± 4090.78902.78 ± 433.68896.99 ± 236.9975.85 ± 14.93415.79 ± 0.53112.56 ± 8.69487.75 ± 2.835214.12 ± 70.161053.24 ± 280.661267.36 ± 30.0759.08 ± 2.3258.38 ± 0.6336.75 ± 2.84141.72 ± 05306.71 ± 905.831186.34 ± 70.16688.66 ± 178.1847.06 ± 6.5768.53 ± 2.4243.46 ± 16.36147.39 ± 54.0810,925.93 ± 5692.86868.06 ± 125.191030.09 ± 411.3339.77 ± 8.89715.01 ± 1.21126.42 ± 13.93597.36 ± 165.071889.76 ± 346.02267.84 ± 89.28855.6 ± 71.7547.51 ± 12.27813.74 ± 2.0841.57 ± 6.55251.32 ± 27.9622,189.41 ± 19,948.572295.78 ± 1967.52778.17 ± 2842.3681.89 ± 36.84912.89 ± 2.7431.22 ± 2.84198.72 ± 17.025734.95 ± 4135.35706.02 ± 207.12896.99 ± 323.0943.35 ± 10.511014.75 ± 0.9943.08 ± 8.17238.62 ± 31.444282.41 ± 320.75422.45 ± 50.12312.5 ± 90.2138.08 ± 14.19119.59 ± 2.1534.01 ± 0124.72 ± 09751.16 ± 1726.911001.16 ± 325.88821.76 ± 595.749.98 ± 8.551211.29 ± 8.7130.23 ± 8.66160.62 ± 29.096516.2 ± 1899.94387.73 ± 184.01266.2 ± 140.3371.85 ± 25.03 Anatomical traits mean and standard deviation across C. procera populations studied Box plots of representative morphological characters show within-population trait plasticity and among-population differences. The values below the plots show the plasticity value of the trait Box plots of representative anatomical characters show within-population trait plasticity and among-population differences. The values below the plots show the plasticity value of the trait Both univariate (ANOVA) and multivariate analyses (MANOVA based on the Adonis test) revealed significant differences among populations for the studied morphological and anatomical traits (p = 0.01). These results demonstrate substantial phenotypic differentiation among populations as well as considerable trait variability within local populations, indicating a high degree of phenotypic plasticity across the species’ geographical range. Morphological traits showed pronounced variation among populations, particularly in reproductive and vegetative traits such as cyathium size, internode length, peduncle length, follicle size, and seed-related traits (Table 3). Preliminary descriptive statistics (Table 4) further indicated moderate to high coefficients of variation for most morphological traits, highlighting heterogeneous trait expression across populations. Anatomical traits also exhibited marked population-level differentiation (Table 5). Traits related to protective and functional tissues, including cuticle thickness, epidermal layers, cortex, lamina, vascular tissues, sclerenchyma, and laticifer cells, varied substantially among populations. Descriptive statistics (Table 5) revealed particularly high variability in traits associated with vascular and secretory tissues, suggesting differential anatomical responses across environmental gradients. Association analyses: RDA, CCA, and FAMD To assess associations between morphological and anatomical traits and geographical and climatic variables, Redundancy Analysis (RDA) and Canonical Correspondence Analysis (CCA) were performed using 999 permutations. Both analyses revealed significant associations between phenotypic traits and environmental variables (p < 0.01), indicating non-random covariation between trait variation and geographic–climatic gradients. Factor Analysis of Mixed Data (FAMD) further illustrated these relationships. For morphological traits, the first three FAMD axes explained approximately 45% of the total variance (Fig. 4A). The ordination based on the first two axes separated populations into three major groups (Fig. 4B). Morphological traits such as cyathium size, internode length, and peduncle length were closely associated with longitude and humidity, whereas leaf base and leaf thickness were more strongly associated with latitude and temperature.Fig. 4FAMD plots of morphological and anatomical traits. A and C = The Eigenvalues for the FAMD axes. B and D = The contribution of traits in differentiation of populations (1–12), and their association with geographical and climate variables FAMD plots of morphological and anatomical traits. A and C = The Eigenvalues for the FAMD axes. B and D = The contribution of traits in differentiation of populations (1–12), and their association with geographical and climate variables For anatomical traits, the first three FAMD axes accounted for approximately 52% of the total variance (Fig. 4C). The ordination revealed associations between stomatal length and upper epidermis thickness with longitude, while pith size and laticifer cell traits were associated with latitude. Additionally, stomatal pore size was associated with humidity, whereas stomatal aperture size was associated with temperature (Fig. 4D). Mantel tests based on 999 permutations showed significant correlations between morphological and anatomical trait dissimilarities and geographical distances (longitude, latitude, and altitude; p = 0.01). These results indicate that phenotypic divergence among populations increases with geographical distance. Together, these findings demonstrate that both geographical position and climatic variables are significantly associated with morphological and anatomical variation among Calotropis procera populations, a pattern also supported by spatial principal component analysis (sPCA) presented below. Spatial principal component analysis (sPCA) Spatial principal component analysis of morphological and anatomical traits revealed a clear large-scale spatial structure along the first principal axis (Fig. 5). Moran’s I values were significantly positive for both morphological (I = 0.77) and anatomical traits (I = 0.958), indicating strong spatial autocorrelation and similarity among geographically proximate populations.Fig. 5The representative plots of sPCA analysis in morphological and anatomical traits in Caloteropis procera geographical populations studied. A = The connection plot, B and D = The anatomical and morphological clines formed due to spatial variables. C = The Eige value plot of anatomical traits show the importance of longitude in spatial Shaping of the studied characters The representative plots of sPCA analysis in morphological and anatomical traits in Caloteropis procera geographical populations studied. A = The connection plot, B and D = The anatomical and morphological clines formed due to spatial variables. C = The Eige value plot of anatomical traits show the importance of longitude in spatial Shaping of the studied characters Global Monte Carlo tests on sPCA eigenvalues confirmed the presence of significant global and local spatial structure (p = 0.001), demonstrating both broad-scale and fine-scale spatial differentiation based on phenotypic traits. Moran’s I was calculated to quantify spatial autocorrelation of morphological and anatomical traits, with values ranging from − 1 (perfect dispersion) to + 1 (perfect clustering). Positive Moran’s I indicates that geographically proximate populations exhibit similar phenotypes, suggesting shared environmental influences or historical genetic connectivity. Trait loadings on the first sPCA axis showed that seed size and peduncle length (negative loadings), as well as follicle tail and follicle size (positive loadings), were the main contributors to spatial structuring. These traits, largely associated with reproductive structures, exhibited pronounced spatial organization. The second sPCA axis was primarily influenced by cyathium length, cyathium tail, and internode length, which showed weak or no spatial structure, suggesting stochastic variation or microenvironmental effects. For anatomical traits, spatial structuring was predominantly aligned along a longitudinal gradient. Random Forest analysis identified longitude as the most influential predictor for anatomical variation, suggesting a strong east–west spatial pattern. In contrast, morphological traits were influenced by both longitude and latitude, indicating a more complex spatial response. Overall, the combined evidence from Moran’s I, sPCA, and Mantel tests supports the presence of both Isolation by Distance (IBD) and Isolation by Environment (IBE) in shaping phenotypic divergence across Calotropis procera populations. Spatial dependency, variogram modeling, and climatic responses Spatial Dependency Index (SDI) analyses and variogram modeling for morphological and anatomical traits are summarized in Table 6. Trait-specific variogram analyses using spherical, exponential, and Gaussian models revealed varying degrees of spatial dependence and spatial scale.Table 6Variogram analyses results for morphological traits studiedTraitModelNuggetSillRangeR_squaredSpatial_Dependenceleaf L/WSpherical0.0580.0681696.6460.014515.086leaf L/WExponential0.0600.0711696.6460.00515.176leaf L/WGaussian0.0700.0461696.6460.034−51.848leaf thicknessSpherical0.1560.3681696.6460.36257.549leaf thicknessExponential0.1500.5261696.6460.39771.358leaf thicknessGaussian0.2110.5191696.6460.33459.358internodeSpherical2.4154.1471690.9880.15141.761internodeExponential2.6075.0051686.2810.09947.906internodeGaussian3.3183.9681695.7470.00916.379peduncleSpherical0.1990.2241696.6470.00111.045peduncleExponential0.2120.2161696.6471.7271.835peduncleGaussian0.2370.1431696.6470.011−65.372cyathyumSpherical17.50716.4311781.7380.0008−6.547cyathyumExponential18.026714.6611755.1720.002−22.951cyathyumGaussian19.4124.5122113.6550.043−330.228foliceleSpherical0.6800.4931696.6460.085−37.885foliceleExponential0.6790.3651696.6470.085−86.049foliceleGaussian0.6490.3091696.6470.124−110.008Seed.sizeSpherical0.0080.0081696.6460.0012.463Seed.sizeExponential0.0080.0091696.6460.00083.432Seed.sizeGaussian0.0080.0081696.6460.0003−2.053seed_pappusSpherical0.0080.0081696.6460.0012.463seed_pappusExponential0.0080.0091696.6460.00083.432seed_pappusGaussian0.0080.0081696.6460.0003−2.053TraitModel_TypeSpatial_Dependence(%)Range(m)Structure_ScaleInterpretation_SummarySeed SizeIsotropic66.6810Mid-scaleStrong spatial structure; anisotropy confirmed patternSeed SizeAnisotropic66.6810Mid-scaleConsistent with isotropic; directional pattern likelyLeaf L/W RatioIsotropic16.4–28.11696Broad/weakWeak spatial structure; likely overestimated rangeLeaf L/W RatioAnisotropic42.2581Local/midImproved fit; structure detected at finer scaleTotal CyathyumIsotropic−12 to −4111721–1998UnrealisticModels likely invalid; spatial structure poorly capturedTotal CyathyumAnisotropic39.76.1Micro-scaleClear localized structure; anisotropy gives valid modelpappus SeedIsotropic−80 to −11191696UnrealisticSpatial dependence misestimated; models likely invalidpappus SeedAnisotropic65.94.1Micro-scaleStrong local structure; directional dispersal plausibleLeaf ThicknessIsotropic57.3–711696BroadConsistent spatial pattern; possibly over-generalizedLeaf ThicknessAnisotropic39.78.1Local scaleSpatial structure clearer and more realistic Table 6Variogram analyses results for morphological traits studiedTraitModelNuggetSillRangeR_squaredSpatial_Dependenceleaf L/WSpherical0.0580.0681696.6460.014515.086leaf L/WExponential0.0600.0711696.6460.00515.176leaf L/WGaussian0.0700.0461696.6460.034−51.848leaf thicknessSpherical0.1560.3681696.6460.36257.549leaf thicknessExponential0.1500.5261696.6460.39771.358leaf thicknessGaussian0.2110.5191696.6460.33459.358internodeSpherical2.4154.1471690.9880.15141.761internodeExponential2.6075.0051686.2810.09947.906internodeGaussian3.3183.9681695.7470.00916.379peduncleSpherical0.1990.2241696.6470.00111.045peduncleExponential0.2120.2161696.6471.7271.835peduncleGaussian0.2370.1431696.6470.011−65.372cyathyumSpherical17.50716.4311781.7380.0008−6.547cyathyumExponential18.026714.6611755.1720.002−22.951cyathyumGaussian19.4124.5122113.6550.043−330.228foliceleSpherical0.6800.4931696.6460.085−37.885foliceleExponential0.6790.3651696.6470.085−86.049foliceleGaussian0.6490.3091696.6470.124−110.008Seed.sizeSpherical0.0080.0081696.6460.0012.463Seed.sizeExponential0.0080.0091696.6460.00083.432Seed.sizeGaussian0.0080.0081696.6460.0003−2.053seed_pappusSpherical0.0080.0081696.6460.0012.463seed_pappusExponential0.0080.0091696.6460.00083.432seed_pappusGaussian0.0080.0081696.6460.0003−2.053TraitModel_TypeSpatial_Dependence(%)Range(m)Structure_ScaleInterpretation_SummarySeed SizeIsotropic66.6810Mid-scaleStrong spatial structure; anisotropy confirmed patternSeed SizeAnisotropic66.6810Mid-scaleConsistent with isotropic; directional pattern likelyLeaf L/W RatioIsotropic16.4–28.11696Broad/weakWeak spatial structure; likely overestimated rangeLeaf L/W RatioAnisotropic42.2581Local/midImproved fit; structure detected at finer scaleTotal CyathyumIsotropic−12 to −4111721–1998UnrealisticModels likely invalid; spatial structure poorly capturedTotal CyathyumAnisotropic39.76.1Micro-scaleClear localized structure; anisotropy gives valid modelpappus SeedIsotropic−80 to −11191696UnrealisticSpatial dependence misestimated; models likely invalidpappus SeedAnisotropic65.94.1Micro-scaleStrong local structure; directional dispersal plausibleLeaf ThicknessIsotropic57.3–711696BroadConsistent spatial pattern; possibly over-generalizedLeaf ThicknessAnisotropic39.78.1Local scaleSpatial structure clearer and more realistic Variogram analyses results for morphological traits studied For morphological traits, anisotropic variogram models generally provided more realistic and ecologically interpretable estimates of spatial structure than isotropic models. Traits such as leaf thickness, leaf length/width ratio, seed size, and seed cob exhibited moderate to strong spatial dependence at local to mid spatial scales, whereas isotropic models often overestimated spatial ranges or failed to detect valid spatial structure. These results indicate the importance of directional spatial processes and localized environmental filtering. Anatomical traits showed heterogeneous spatial dependency patterns. Traits such as vascular bundle size, stomatal length, and laticifer cell area exhibited strong spatial structuring, whereas cuticle thickness, lower epidermis thickness, and outer phloem area showed weak or inconsistent spatial dependence. Several traits displayed pronounced anisotropy, suggesting directional environmental gradients or regional-scale processes influencing anatomical differentiation (Fig. 6).Fig. 6Spatial dependency plots of anatomical characters, show differences in isotropic versus anisotropic variograms Spatial dependency plots of anatomical characters, show differences in isotropic versus anisotropic variograms Kriging analysis: predicting adaptive potential in unsampled regions Kriging prediction maps based on genetic diversity and the Landscape Adaptive Index (LAI) are shown in Fig. 7. Kriging based solely on neutral genetic diversity produced relatively uniform spatial predictions, reflecting weak spatial structure and limited environmental responsiveness of neutral markers.Fig. 7Comparison of Kriging prediction maps based on (A) genetic diversity and (B) LAI. Unlike the genetic diversity map, the LAI-based prediction identifies distinct spatial variation in adaptive potential across both sampled and unsampled regions Comparison of Kriging prediction maps based on (A) genetic diversity and (B) LAI. Unlike the genetic diversity map, the LAI-based prediction identifies distinct spatial variation in adaptive potential across both sampled and unsampled regions In contrast, the LAI—integrating morphological, anatomical, genetic, and environmental components—exhibited strong spatial structure and generated heterogeneous prediction maps. The LAI-based Kriging surface closely matched observed population-level LAI values while providing continuous predictions across unsampled regions. This approach allows identification of areas with high adaptive potential, potential conservation priorities, and candidate refugia beyond sampled populations. RRMM analyses: trait responses to climate change Random Regression Mixed Models (RRMM) revealed heterogeneous population-specific responses of morphological and anatomical traits to projected changes in temperature and humidity (Tables 7 and 8; Fig. 8).Table 7Results of variogram analyses results for anatomical traits studiedTraitVariogram_ModelNuggetSillRange_mR_squaredSpatial_Dependence_%CuticleIsotropicSpherical0.0430.04514.490.00043.8IsotropicExponential0.0400.05114.490.00520.09IsotropicGaussian0.0420.05214.540.00619.36AnisotropicSpherical142.965446.7812880.49831.99921.475Vascular bundleIsotropicSpherical0.0070.04914.9110.27085.7409984065299IsotropicExponential0.0050.08114.6140.29793.2508841926537IsotropicGaussian0.0130.09114.6810.40285.3967851829292AnisotropicSpherical136,654.8636,325.34379.354Stomata lengthIsotropicSpherical0.0030.00914.4940.60464.4581646035284IsotropicExponential0.0030.01314.4870.64776.5413418959441IsotropicGaussian0.0040.01414.4870.76468.6713871696612AnisotropicSpherical0.00030.001809.87215.969CortexIsotropicSpherical0.0170.02614.5190.12731.5411099556009IsotropicExponential0.0170.03214.4870.13744.8001928040067IsotropicGaussian0.0180.03414.4870.22245.7020286680713AnisotropicSpherical44,265.88857,264.95075.0125.1636Lower epidermisIsotropicSpherical0.0690.08114.7780.00614.6278246195178IsotropicExponential0.0650.09514.4880.01531.2701623316606IsotropicGaussian0.0700.09414.6460.01225.8137133935488AnisotropicSpherical607.46219,214.538526.793.16Upper epidermisIsotropicSpherical0.0430.06014.6050.02028.2203174396111IsotropicExponential0.0390.07814.4870.03649.6563328811102IsotropicGaussian0.0430.08314.9020.04547.8720389100588AnisotropicSpherical2.02462.1531743.0473.258MataxaylemIsotropicSpherical92,772,510.092126,682,890.02143,365.1680.014926.7679241631721IsotropicExponential92,807,787.863123,867,111.48827,074.4260.01425.0747137408171IsotropicGaussian92,702,936.406146,011,688.6596780.4750.04636.509921050132AnisotropicSpherical2.02462.1531743.0473.258outer.phloem_IsotropicSpherical2611.0662553.9581504.5360.131−2.23603056222852IsotropicExponential2605.2532529.6481375.6760.096−2.98875462098771IsotropicGaussian2602.8672548.269865.8730.143−2.1425440886756AnisotropicSpherical1418.43326,805.2511,716.545.291ScleranchymaIsotropicSpherical0.1180.1081696.6460.583−8.9014776437123IsotropicExponential0.1180.1011696.6460.598−16.2482320103542IsotropicGaussian0.1160.1001696.6460.712−16.1261154164208AnisotropicSpherical606,365.471−1,712,235.155197,264.0910.492135.413679593453Laticifer cell areaIsotropicSpherical606,365.471−1,712,235.155197,264.0910.492135.413679593453IsotropicExponential606,424.721−571,997.67766,008.6450.492206.018738506126IsotropicGaussian591,836.332608,067.705294,168.9470.4722.66933646286815AnisotropicSpherical264,127.18,637,03313,970.173.058Table 8RRMM response of morphological characters in C. procera populations against climate changePopulationdelta_temp_Leaf_basedelta_temp_Leaf_LWdelta_temp_Leaf_stalkdelta_temp_Leaf_thicknessdelta_temp_Leaf_internodedelta_temp_Total_cyathiumdelta_temp_Follicle_size10.0962.0470.0260.193−0.0491.4680.7167772−0.151−1.144−0.009−0.4550.0140.515−2.024168530.097−0.2020.003−0.089−0.0381.262−2.34975544−0.0010.123−0.0010.061−0.003−0.855−0.375058750.096−0.142−0.009−0.4780.0390.047−2.20470860.0010.1730.0110.582−0.0201.008−2.07730117−0.119−0.849−0.009−0.5070.003−0.312−0.563427880.0941.245−0.0090.489−0.017−1.5281.38924249−0.159−2.599−0.009−0.1320.024−0.8240.496900510−0.062−1.2620.024−0.4670.042−1.3094.2017076110.0090.142−0.0090.256−0.00071.0041.4104501120.0972.468−0.0090.5480.007−0.4751.3793418delta_temp_Seed_sizedelta_humidity_Leaf_basedelta_humidity_Leaf_LWdelta_humidity_Leaf_stalkdelta_humidity_Leaf_thicknessdelta_humidity_Leaf_internodedelta_humidity_Total_cyathiumdelta_humidity_Follicle_sizedelta_humidity_Seed_size−2.21741168−0.51680.3452.064−0.0533.5111.0400.412−3.143−1.30356090.811−0.193−0.6930.126−1.0010.365−1.164−1.847−1.6693493−0.525−0.0340.2550.0242.7640.894−1.351−2.3660.24697170.0070.020−0.080−0.0160.246−0.606−0.2150.3500.20854967−0.517−0.024−0.7170.132−2.7700.033−1.2680.295−1.36304222−0.0060.0290.909−0.1611.4840.714−1.195−1.9320.046334220.642−0.143−0.7640.140−0.212−0.221−0.3240.0650.35905598−0.5070.210−0.732−0.1351.221−1.0830.7990.50891.384443160.855−0.438−0.7030.036−1.730−0.5840.2851.9621.403971690.334−0.2131.8780.129−3.017−0.9272.4171.9901.4497371−0.0530.0240−0.707−0.0700.0540.7110.8112.05491.45430057−0.5240.416−0.707−0.152−0.549−0.3370.7932.0614Fig. 8Representative plots of RRMM analyses for morphological traits in Caloteropis procera populations. A and C = Intercept (No effect of temperature and humidity) of the character in the populations studied. B and D = Dot plots showing the slopes (responses) of populations against climate change Table 7Results of variogram analyses results for anatomical traits studiedTraitVariogram_ModelNuggetSillRange_mR_squaredSpatial_Dependence_%CuticleIsotropicSpherical0.0430.04514.490.00043.8IsotropicExponential0.0400.05114.490.00520.09IsotropicGaussian0.0420.05214.540.00619.36AnisotropicSpherical142.965446.7812880.49831.99921.475Vascular bundleIsotropicSpherical0.0070.04914.9110.27085.7409984065299IsotropicExponential0.0050.08114.6140.29793.2508841926537IsotropicGaussian0.0130.09114.6810.40285.3967851829292AnisotropicSpherical136,654.8636,325.34379.354Stomata lengthIsotropicSpherical0.0030.00914.4940.60464.4581646035284IsotropicExponential0.0030.01314.4870.64776.5413418959441IsotropicGaussian0.0040.01414.4870.76468.6713871696612AnisotropicSpherical0.00030.001809.87215.969CortexIsotropicSpherical0.0170.02614.5190.12731.5411099556009IsotropicExponential0.0170.03214.4870.13744.8001928040067IsotropicGaussian0.0180.03414.4870.22245.7020286680713AnisotropicSpherical44,265.88857,264.95075.0125.1636Lower epidermisIsotropicSpherical0.0690.08114.7780.00614.6278246195178IsotropicExponential0.0650.09514.4880.01531.2701623316606IsotropicGaussian0.0700.09414.6460.01225.8137133935488AnisotropicSpherical607.46219,214.538526.793.16Upper epidermisIsotropicSpherical0.0430.06014.6050.02028.2203174396111IsotropicExponential0.0390.07814.4870.03649.6563328811102IsotropicGaussian0.0430.08314.9020.04547.8720389100588AnisotropicSpherical2.02462.1531743.0473.258MataxaylemIsotropicSpherical92,772,510.092126,682,890.02143,365.1680.014926.7679241631721IsotropicExponential92,807,787.863123,867,111.48827,074.4260.01425.0747137408171IsotropicGaussian92,702,936.406146,011,688.6596780.4750.04636.509921050132AnisotropicSpherical2.02462.1531743.0473.258outer.phloem_IsotropicSpherical2611.0662553.9581504.5360.131−2.23603056222852IsotropicExponential2605.2532529.6481375.6760.096−2.98875462098771IsotropicGaussian2602.8672548.269865.8730.143−2.1425440886756AnisotropicSpherical1418.43326,805.2511,716.545.291ScleranchymaIsotropicSpherical0.1180.1081696.6460.583−8.9014776437123IsotropicExponential0.1180.1011696.6460.598−16.2482320103542IsotropicGaussian0.1160.1001696.6460.712−16.1261154164208AnisotropicSpherical606,365.471−1,712,235.155197,264.0910.492135.413679593453Laticifer cell areaIsotropicSpherical606,365.471−1,712,235.155197,264.0910.492135.413679593453IsotropicExponential606,424.721−571,997.67766,008.6450.492206.018738506126IsotropicGaussian591,836.332608,067.705294,168.9470.4722.66933646286815AnisotropicSpherical264,127.18,637,03313,970.173.058 Results of variogram analyses results for anatomical traits studied Table 8RRMM response of morphological characters in C. procera populations against climate changePopulationdelta_temp_Leaf_basedelta_temp_Leaf_LWdelta_temp_Leaf_stalkdelta_temp_Leaf_thicknessdelta_temp_Leaf_internodedelta_temp_Total_cyathiumdelta_temp_Follicle_size10.0962.0470.0260.193−0.0491.4680.7167772−0.151−1.144−0.009−0.4550.0140.515−2.024168530.097−0.2020.003−0.089−0.0381.262−2.34975544−0.0010.123−0.0010.061−0.003−0.855−0.375058750.096−0.142−0.009−0.4780.0390.047−2.20470860.0010.1730.0110.582−0.0201.008−2.07730117−0.119−0.849−0.009−0.5070.003−0.312−0.563427880.0941.245−0.0090.489−0.017−1.5281.38924249−0.159−2.599−0.009−0.1320.024−0.8240.496900510−0.062−1.2620.024−0.4670.042−1.3094.2017076110.0090.142−0.0090.256−0.00071.0041.4104501120.0972.468−0.0090.5480.007−0.4751.3793418delta_temp_Seed_sizedelta_humidity_Leaf_basedelta_humidity_Leaf_LWdelta_humidity_Leaf_stalkdelta_humidity_Leaf_thicknessdelta_humidity_Leaf_internodedelta_humidity_Total_cyathiumdelta_humidity_Follicle_sizedelta_humidity_Seed_size−2.21741168−0.51680.3452.064−0.0533.5111.0400.412−3.143−1.30356090.811−0.193−0.6930.126−1.0010.365−1.164−1.847−1.6693493−0.525−0.0340.2550.0242.7640.894−1.351−2.3660.24697170.0070.020−0.080−0.0160.246−0.606−0.2150.3500.20854967−0.517−0.024−0.7170.132−2.7700.033−1.2680.295−1.36304222−0.0060.0290.909−0.1611.4840.714−1.195−1.9320.046334220.642−0.143−0.7640.140−0.212−0.221−0.3240.0650.35905598−0.5070.210−0.732−0.1351.221−1.0830.7990.50891.384443160.855−0.438−0.7030.036−1.730−0.5840.2851.9621.403971690.334−0.2131.8780.129−3.017−0.9272.4171.9901.4497371−0.0530.0240−0.707−0.0700.0540.7110.8112.05491.45430057−0.5240.416−0.707−0.152−0.549−0.3370.7932.0614 RRMM response of morphological characters in C. procera populations against climate change Representative plots of RRMM analyses for morphological traits in Caloteropis procera populations. A and C = Intercept (No effect of temperature and humidity) of the character in the populations studied. B and D = Dot plots showing the slopes (responses) of populations against climate change Morphological traits exhibited both positive and negative slopes across populations, indicating contrasting plastic responses. Traits such as leaf base, leaf thickness, internode length, reproductive structures, and seed size showed strong population-specific sensitivities to climatic variables, highlighting divergent adaptive or sensitive responses among populations (Tables 9, 10 and 11).Table 9RRMM results on the anatomical traits response to climate change (Delta Temperature, and Delta Humidity)PopulationDelta TempCuticleDelta HumidityCuticleDelta TempStomataDelta HumidityStomataDelta TempLaticifer cellDelta HumidityLaticifer cell1−0.161−0.2081.1160.9591.8812.8292−0.186−0.239−1.092−0.939−13.554−20.35630.0240.030.520.447−1.401−2.10640.00.00.00.00.00.050.1170.1511.1140.9572.5243.80160.0370.0481.3581.167−2.733−4.0997−0.025−0.0320.8050.6920.9651.49180.0420.054−1.703−1.463−13.156−19.82490.0890.115−0.935−0.803−1.193−1.79610−0.029−0.038−0.331−0.28412.2218.379110.0520.0670.5110.4391.8652.799120.0410.052−1.363−1.17212.58118.881PopulationMetaxylemDelta TempMetaxylemDelta HumidityLaminaDelta TempLaminaDelta HumidityLower epidermisDelta_TempLower epidermisDelta Humidity11.3351.178−7.809−4.1509.0617.17622.3222.047−10.003−5.3357.9676.3043−1.270−1.1203.1651.686−3.347−2.6494−4.740488E-10−4.547264E-10−6.075415E-09−7.726843E-093.29718E-092.49326E-0952.0661.8223.1311.659−3.061−2.4256−3.099−2.7330.6620.3550.9650.7627−0.447−0.394−1.619−0.8361.9391.5418−4.278−3.7724.5102.386−2.481−1.96691.5791.3923.3631.781−4.387−3.476102.3992.1151.3100.696−1.581−1.25211−2.671−2.3551.5070.805−1.97−1.562122.0631.81951.7810.952−3.099−2.451Table 10Resistance-based ICP network metricsPopulationClosenessBetweennessClusteringMIW110.4730110.228.877017.938.325016.3949.309017.97511.9240111.868.119315.9878.517017.49811.4550111.1497.160116.81108.725017.81118.121315.98128.1521.516.01Table 11Genetic distance-based ICP network metricsPopulationClosenessBetweennessClusteringMIW123.3930122.85216.9590114.11317.8720114.17425.0940124.78515.2500114.35618.6440114.65718.7270117.23816.4710113.56922.1730119.961026.9940127.081119.2060117.911220.1260117.4 Table 9RRMM results on the anatomical traits response to climate change (Delta Temperature, and Delta Humidity)PopulationDelta TempCuticleDelta HumidityCuticleDelta TempStomataDelta HumidityStomataDelta TempLaticifer cellDelta HumidityLaticifer cell1−0.161−0.2081.1160.9591.8812.8292−0.186−0.239−1.092−0.939−13.554−20.35630.0240.030.520.447−1.401−2.10640.00.00.00.00.00.050.1170.1511.1140.9572.5243.80160.0370.0481.3581.167−2.733−4.0997−0.025−0.0320.8050.6920.9651.49180.0420.054−1.703−1.463−13.156−19.82490.0890.115−0.935−0.803−1.193−1.79610−0.029−0.038−0.331−0.28412.2218.379110.0520.0670.5110.4391.8652.799120.0410.052−1.363−1.17212.58118.881PopulationMetaxylemDelta TempMetaxylemDelta HumidityLaminaDelta TempLaminaDelta HumidityLower epidermisDelta_TempLower epidermisDelta Humidity11.3351.178−7.809−4.1509.0617.17622.3222.047−10.003−5.3357.9676.3043−1.270−1.1203.1651.686−3.347−2.6494−4.740488E-10−4.547264E-10−6.075415E-09−7.726843E-093.29718E-092.49326E-0952.0661.8223.1311.659−3.061−2.4256−3.099−2.7330.6620.3550.9650.7627−0.447−0.394−1.619−0.8361.9391.5418−4.278−3.7724.5102.386−2.481−1.96691.5791.3923.3631.781−4.387−3.476102.3992.1151.3100.696−1.581−1.25211−2.671−2.3551.5070.805−1.97−1.562122.0631.81951.7810.952−3.099−2.451 RRMM results on the anatomical traits response to climate change (Delta Temperature, and Delta Humidity) Table 10Resistance-based ICP network metricsPopulationClosenessBetweennessClusteringMIW110.4730110.228.877017.938.325016.3949.309017.97511.9240111.868.119315.9878.517017.49811.4550111.1497.160116.81108.725017.81118.121315.98128.1521.516.01 Resistance-based ICP network metrics Table 11Genetic distance-based ICP network metricsPopulationClosenessBetweennessClusteringMIW123.3930122.85216.9590114.11317.8720114.17425.0940124.78515.2500114.35618.6440114.65718.7270117.23816.4710113.56922.1730119.961026.9940127.081119.2060117.911220.1260117.4 Genetic distance-based ICP network metrics Anatomical traits also showed marked variability in climate responses. Cuticle thickness, stomatal length, laticifer cell area, metaxylem area, lamina size, and epidermal thickness displayed population-specific slopes, indicating differential anatomical plasticity under projected warming and humidity changes. Genetic corridors and least-cost path analyses Genetic corridor and least-cost path analyses are presented in Fig. 9. Potential gene flow corridors were spatially heterogeneous across the study area, reflecting the fragmented geographical distribution of Calotropis procera populations. White areas on corridor maps indicate potential connectivity, whereas brown regions represent barriers to gene flow, consistent with significant genetic differentiation detected by AMOVA.Fig. 9Plots of the genetic corridors (A), and the least cost distance (B) of the studied Caloteropis procera populations show that the neighboring and closely situated populations may have less obstacle against gene flow Plots of the genetic corridors (A), and the least cost distance (B) of the studied Caloteropis procera populations show that the neighboring and closely situated populations may have less obstacle against gene flow Genetic corridors overlapped with least-cost pathways, indicating that geographically adjacent populations are more likely to exchange genes. These results suggest that maintaining or restoring connectivity among neighboring populations may enhance within-population genetic diversity and support adaptive potential across the species’ range. Discussion The integrated influence of environmental heterogeneity on vegetation structure and function represents a central theme in ecological and landscape genetic research. At regional scales, climatic gradients and topographic complexity generate substantial spatial variation in plant traits and soil properties [29]. At finer scales, interacting factors such as slope, aspect, and slope position strongly modulate microclimatic conditions—radiation, temperature, soil moisture, and nutrient availability—thereby shaping plant performance and adaptive responses [30]. Moreover, biodiversity patterns and ecosystem processes emerge from the interplay of biotic interactions, abiotic constraints, and anthropogenic pressures. Topographic attributes, including elevation, slope, and aspect, interact with dispersal capacity and landscape permeability to determine species distribution and abundance [31]. In this context, C. procera, a stress-tolerant xerophytic species, provides an excellent model to disentangle how genetic, morphological, and anatomical traits respond to spatially structured environmental gradients. Our results demonstrate that variation in C. procera across Iran is spatially structured, reflecting the combined effects of climatic filtering, edaphic heterogeneity, landscape fragmentation, and historical gene flow. This emphasizes that trait patterns are shaped by ecological and evolutionary processes rather than stochastic variation. In addition, genetic diversity analyses across the twelve studied populations revealed moderate to low variation overall, with marked inter-population heterogeneity. Populations 1, 6, and 9 exhibited the highest allelic richness, gene diversity, and polymorphism, suggesting their role as evolutionary reservoirs. In contrast, populations 4 and 10 displayed critically low diversity, indicating vulnerability to environmental change and anthropogenic disturbance. AMOVA confirmed significant genetic structuring among populations [6], reflecting restricted gene flow and historical isolation. While the molecular markers employed are selectively neutral, they capture demographic history and connectivity patterns shaping the spatial genetic template for potential selection. Multivariate genotype–environment association analyses (RDA and CCA) revealed significant correlations between neutral genetic variation and environmental predictors (temperature, humidity, latitude, and longitude), highlighting adaptive potential [32]. Landscape fragmentation varied across populations, with Pop 6, 11, and 12 in relatively intact habitats, while Pop 1 and 5 were embedded in highly fragmented landscapes. Resistance distance and least-cost path analyses identified Pop 1 and 5 as ecologically isolated, constrained by topography and human-modified land cover. Conversely, Pop 2, 3, 10, and 11 formed a connected cluster, suggesting functional dispersal corridors, in line with previous findings [6]. The concordance between resistance-based models and genetic differentiation underscores the role of landscape permeability in shaping gene flow. However, resistance models alone cannot fully capture functional connectivity, emphasizing the need for integrated approaches. Covariance-Informed Population (CIP) networks provided additional resolution, identifying populations with disproportionate influence on network cohesion. Some geographically marginal or environmentally isolated populations functioned as key genetic connectors, highlighting their importance for conservation prioritization. Integrating CIP networks with classical resistance models provides a more biologically realistic framework for connectivity-based conservation planning [33]. Soil properties varied substantially across the study area, spanning gradients in pH, texture, organic carbon, and cation exchange capacity. Populations with higher genetic diversity and adaptive indices (e.g., Pop 1 and 6) were associated with well-structured soils, enhancing water retention and nutrient availability. These edaphic conditions likely buffer climatic stress, facilitating phenotypic plasticity and long-term adaptation. Spatial dependency analyses revealed strong trait-specific and direction-dependent patterns. Morphological traits related to leaf geometry exhibited spatial structuring consistent with environmental filtering and dispersal limitation. Anatomical traits showed a dual pattern: hydraulically related traits (e.g., stomatal dimensions, vascular bundle thickness) displayed strong spatial dependency and climatic responsiveness, whereas structural traits (e.g., cuticle thickness, cortex width) were more conserved. These findings emphasize the balance between phenotypic robustness and adaptive flexibility. Although AMOVA and resistance-based analyses indicated restricted contemporary gene flow among some populations, positive Moran’s I values for both morphological (I = 0.77) and anatomical traits (I = 0.958) reveal strong spatial autocorrelation. This pattern likely arises from a combination of historical connectivity, environmental filtering, and phenotypic plasticity. Neighboring populations may exhibit similar phenotypes because they experience comparable climatic and edaphic conditions, or because ancestral gene flow established shared trait patterns, which are maintained even under limited current dispersal. Thus, spatial phenotypic similarity does not necessarily imply ongoing gene exchange but reflects both ecological constraints and evolutionary legacy. Furthermore, Random Regression Mixed Models highlighted population-level differences in plastic responses to climatic gradients. Populations 5, 9, 11, and 12 exhibited adaptive plasticity, enhancing conductive and defensive tissues under warming and altered humidity, while populations 1, 2, and especially 4 showed reduced plasticity. The convergence of RRMM, sPCA, RDA, and AMOVA results demonstrates that anatomical variation in C. procera is shaped by both plasticity and genetically structured spatial processes, underscoring the dual control of phenotypic traits and their role in resilience. Overall, the LAI integrated genetic diversity, environmental stressors, topography, and fragmentation into a spatially explicit measure of adaptive capacity. High-elevation and environmentally heterogeneous populations (Pop 1 and 9) exhibited the highest adaptive potential, while low-altitude, genetically depauperate populations (Pop 4 and 10) showed minimal adaptive capacity. These results highlight the importance of genetic connectivity and landscape structure in mitigating climatic constraints, providing a basis for conservation prioritization. Consistent with prior studies, C. procera exhibits substantial morphological and anatomical plasticity, enabling persistence under drought, salinity, and thermal stress [2, 34]. Our study extends this by explicitly linking plasticity and genetic structure to spatial processes and landscape configuration. Taken together, resilience in C. procera emerges from the interaction of spatially structured genetic diversity, trait-specific plasticity, and landscape complexity. While neutral markers do not directly indicate selection, significant associations with environmental gradients highlight their value as indicators of adaptive potential. Integrating spatial statistics, RRMM, CIP networks, and LAI offers a powerful framework for conserving xerophytic species under accelerating climate change. n addition to the spatial and genetic patterns identified in this study, incorporating an explicit perspective on plant adaptation, gene flow, and environmental variation further clarifies the ecological mechanisms underlying population differentiation in C. procera. Local populations exposed to heterogeneous climatic and edaphic regimes often exhibit pronounced divergence in functional traits, including leaf hydraulic architecture, cuticular properties, and stomatal regulation—traits known to mediate drought and heat adaptation in xerophytic taxa [35]. Variation in these traits among our studied populations is consistent with adaptive differentiation driven by selective environmental filters rather than neutral drift alone. At the same time, reduced contemporary gene flow in fragmented low‑diversity populations (e.g., Pop 4 and 10) may limit the introgression of advantageous alleles, reinforcing vulnerability under intensifying climatic stress [36]. Conversely, populations embedded within permeable landscapes (e.g., Pop 2, 3, 11) retain functional connectivity that can facilitate the spread of adaptive variants, enabling trait convergence with environmentally similar but geographically distant populations, as observed in previous C. procera studies across arid and semi-arid regions [37]. These combined lines of evidence highlight that spatial variation in morphological and anatomical traits reflects the interplay of local adaptation, environmentally mediated gene flow, and phenotypic plasticity, underscoring the need to incorporate both environmental context and demographic connectivity into conservation strategies for this widely distributed xerophytic species. Supplementary Information Supplementary Material 1. Supplementary Material 1.