Analysis of Floral Trait Diversity and Its Correlation with Fertility in Hibiscus Syriacus Cultivars

1. Introduction

Hibiscus syriacus L., commonly known as the rose of Sharon or hedge flower, is a deciduous shrub in the Malvaceae family [1]. This species is highly valued in landscaping for its diverse floral colours, extended flowering period, and strong stress resistance [2]. In addition to its ornamental appeal, it possesses significant edible and medicinal value [3], establishing it as an excellent greening plant for a variety of ecological applications [4]. Numerous cultivars of H. syriacus exist, displaying extensive diversity in floral phenotypes and palynological characteristics. However, most cultivars exhibit an extremely low natural fruit-setting rate [5], with obvious differentiation in fertility observed among cultivars, representing a critical technical barrier that hinders germplasm innovation, limits cross-breeding efficiency, and restricts the large-scale industrial upgrading of the species. Recent research on H. syriacus has mostly focused on germplasm resource investigation, phenotypic diversity evaluation, and cultivation application. Although preliminary studies on fertility have been undertaken [6], some research has examined pollen morphology and fruiting separately. However, no study has systematically combined floral phenotype and pollen morphology in H. syriacus to identify which traits contribute to fertility. Studies on plant fruit-setting rate and fertility have shown that floral phenotypes directly influence pollination efficiency and fertility [7]. Furthermore, pollen morphology is genetically stable, serving as an important indicator for taxonomic classification and the study of genetic relationships [8]. It is also closely linked to pollination and breeding outcomes [9,10,11], pollen morphology is correlated with pollen viability, stigma fertility, and gametophyte development [12,13,14]. These are the key factors related to fruit set rate. Our team previously reported that the fertility of H. syriacus is closely associated with specific phenotypic traits and palynological characteristics [6], thereby providing an important basis for subsequent studies. Building upon previous research, this study has introduced new metrics, including field fruit-set rates and pollen viability statistics, conducted field surveys on pollinating insects, expanded the observation indicators for floral characteristics, and further refined the comprehensive evaluation system for hibiscus fertility.

Despite the challenges of low fruit-setting rates and poor fertility in H. syriacus, a systematic correlation analysis among floral phenotype, pollen morphology, and fertility has yet to be conducted. To address this, we selected 20 cultivars with significant differences in fruit-setting rates as experimental materials. We evaluated fertility based on fruit-setting data collected over three consecutive years and measured 50 trait indicators, including floral phenotype and pollen morphology. Through this approach, we identified the correlations between these traits and fertility, revealing the key factors that influence fruit-setting rates. These results provide technical support for the breeding of high-fertility H. syriacus cultivars and the sustainable development of the industry.

2. Methods

2.1 Experimental Materials

Twenty cultivars of Hibiscus syriacus L., covering diverse phenotypic traits, were collected from Hunan Forest Botanical Garden (Table 1).

2.2. Trait Evaluation and Measurement

2.2.1. Measurement of Floral Phenotypic Traits

The investigation was performed following the Guidelines for the Conduct of Tests for Distinctness, Uniformity and Stability—Hibiscus syriacus and Hibiscus rosa-sinensis, and the methodology described by Pan et al. [15,16,17]. Healthy H. syriacus specimens, free from pests and diseases, were selected for the evaluation of 38 floral traits. These comprised 17 quantitative traits (Figure 1, Table 2), 18 qualitative traits (Figure 2, Table 2), and 3 colorimetric parameters. For the quantitative traits, measurements were taken from 10 flowers at the full-bloom stage, and mean values were calculated. Qualitative traits were assessed through visual assignment (Table S1). Petal colour groups were determined using the Royal Horticultural Society Colour Chart, and colour coordinates were subsequently recorded using a KONICA MINOLTA colorimeter (Model: CM-5) [18].

2.2.2. Pollen Morphological Traits

Five randomly selected full-flowering hibiscus flowers from each variety were collected, and their pollen was mixed and uniformly mounted on conductive adhesive to prepare samples for electron microscopy, ensuring species representativeness. The samples were examined and photographed using a JSM 6360-LV electron microscope, with 12 pollen morphological parameters measured (Table 3). For each parameter, five randomly chosen pollen grains were analysed, with three measurements taken per parameter and the average value calculated. Twelve pollen morphological indicators were measured (Table 3), with mean values derived from five randomly selected pollen grains per indicator. These indicators comprised both qualitative traits—including pollen grain shape (PGS), wrinkle degree (WD), spine curvature (TC), and spine surface microstructure (TS)—and quantitative traits.

2.3. Assessment of Fertility Grades and Observation of Insect Flower Visit Characteristics

A 0.2% benzidine solution, 0.2% α-naphthol solution and 0.25% sodium carbonate solution were prepared and mixed in equal volumes to obtain the staining solution for later use; a 0.3% hydrogen peroxide solution was prepared separately.

Pollen samples were collected on sunny days between 8:00AM and 9:00AM during the full-bloom stage and spread evenly onto glass slides. The staining solution and hydrogen peroxide solution were then added dropwise to the pollen samples, followed by coverslip placement and gentle pressing. After standing for 10 min, microscopic examination was performed. Three independent visual fields were observed for each sample. Pollen viability was calculated as follows: Pollen viability = (number of viable pollen grains / total number of pollen grains) × 100%. Three biological replicates were analysed for each cultivar, and the average value was adopted for the experimental results.

The fruit-setting rate serves as the primary indicator of fertility classification [19]. In this study, fruit set was considered successful if the ovary had expanded and remained attached two weeks after pollination.

Field trials were conducted in three consecutive summers from 2023 to 2025. Twenty Hibiscus syriacus cultivars were divided into two treatment groups: artificial pollination and open natural pollination. For each cultivar, 100–200 flower buds were tagged and labelled.

Artificial pollination group: The petals and stamens were removed from the flower buds. Pollen collected from different plants of the same cultivar was applied to the stigmas between 8:00 and 12:00 on sunny mornings. Immediately after hand pollination, the flowers were enclosed in nylon mesh bags; the bags were removed 2 weeks later.

Open natural pollination group: This group was only tagged, without receiving additional treatments.

Fruit-set rate (%) = (total number of set fruits / total number of tagged flowers) × 100.

Based on three consecutive years of field observations, the fruit-set rate classification method proposed by Yu Xianghe et al. was adopted in this study [20]. Fertility was classified based on the fruit-setting rate, with a lower limit of >0. The criteria were defined as follows: Grade 1 (0 < fruit-setting rate ≤ 10%), Grade 2 (10% < fruit-setting rate ≤ 30%) and Grade 3 (fruit-setting rate > 30%). This classification method, grounded in long-term field data, objectively and accurately reflected the fertility characteristics of H. syriacus cultivars.

Observations were conducted continuously for 10 days under sunny and breezy conditions across 20 Hibiscus syriacus cultivars, representing different fertility grades. Daily observations were conducted from 8:00 to 18:00. Observers minimised human interference throughout the whole observation period. The single-plant flower visitation counting method was adopted. A valid flower visit was recorded when an insect landed on a flower and then departed. The species and hourly visitation frequency of flower-visiting insects for each cultivar were statistically quantified.

2.4. Data Analysis

Statistical analyses were performed using Microsoft Excel 2021. The coefficient of variation (CV) represented the degree of trait dispersion, in which a lower CV indicated greater trait stability. This was calculated using the formula CV (%) = σ / μ × 100%.

The Shannon-Wiener diversity index (H′) was employed to evaluate phenotypic diversity and was calculated as H′ = －∑Pi lnPi. In this equation, Pi represents the proportion of cultivars in the i-th category of a trait relative to the total number of cultivars, and ln denotes the natural logarithm.

Spearman correlation analysis was performed to assess the relationship among floral traits of Hibiscus syriacus using SPSS version 25.0. The resulting P values were subjected to FDR correction in Excel, with a corrected q ≤ 0.1 considered as statistically significant [21], principal component analysis, and hierarchical cluster analysis were conducted on the floral traits of H. syriacus using SPSS 25.0 software. The cluster analysis included R-type clustering for traits and Q-type clustering for cultivars. Visualisation charts were generated using Origin 2021 software to support the screening of fertility-related traits and the germplasm classification of H. syriacus.

3. Results

3.1. Floral Phenotypic Traits of Hibiscus syriacus Cultivars

3.1.1 Diversity of Floral Phenotypic Traits in Hibiscus syriacus cultivars

The 20 tested H. syriacus cultivars exhibited rich diversity in quantitative and qualitative floral phenotypic traits.

Seventeen quantitative traits of the 20 cultivars showed considerable variation (Table S2). The CV ranged from 13.11% to 52.62%, with an average of 31.03%, in the following descending order: CW > HLL > PL1 > SN > SL > HL2 > HL > HL1 > BL > PW > SL1 > FD > CL > PL > PI > PL2 > BN. These results indicated that calyx width (CW, 52.62%) and heart line length (HLL, 51.85%) were prone to variation, whereas bract number (BN, 13.11%) was the most stable trait. The Shannon-Wiener diversity index (H′) of quantitative traits ranged from 1.23 to 1.98, with an average of 1.67, in the following descending order: SL1 > BL > PL > PW > PI > FD > PL2 > SL > PL1 > HLL > CL > HL1 > CW > SN > HL > HL2 > BN. Among these, stigma length (SL1, 1.98), bract length (BL, 1.96), and petal length (PL, 1.94) had the highest diversity indices, whereas the ratio of heart line length to heart length (HL2, 1.36) and bract number (BN, 1.23) had the lowest. Their quantitative trait diversity may facilitate adaptation to pollination requirements across heterogeneous environments, thereby sustaining population reproduction and environmental adaptability.

Qualitative floral traits exhibited significant variation among cultivars (Table S3), with flower colour displaying up to six distinct morphotypes. The CV ranged from 22.80% to 75.286%, with a mean of 43.633%, in the following descending order: SC > STS > OP > SPD > HBS > FT > FC > RSB > OPP > PN > SS > FCN > RBH > BS > PWD > OS > HP > PF. Furthermore, the diversity index (H′) for these qualitative traits ranged from 0.19 to 1.67, with an average of 0.729, according to the descending order: FC > OP > SPD > FT > RBD > RSB > SS > BS > HBS > OPP > SC > STS > PN > PF > PWD > FCN > OS > HP. These results indicated that stigma colour (SC, 75.286%), sepal tube shape (STS, 75.285%), and petal overlap (OP, 63.664%) possessed the highest CV values. Additionally, the highest diversity indices were observed for flower colour (FC, 1.67) and petal overlap (OP, 1.238), and the lowest for ovary shape (OS, 0.325) and heart presence (HP, 0.199). Traits uninvolved in pollination, such as bract number (BN) and petal wrinkling degree (PWD), possess low diversity indices, which may help in preventing nutrient and energy consumption by non-functional morphological variations.

Based on visual observation and colour chart comparisons, the flower colours of the 20 cultivars were categorised into six distinct groups: white, pinkish-white, blue-purple, purple, pinkish-purple, and purplish-red. Based on colorimeter parameters, two-dimensional distribution maps of a* and b* values (Figure 3a) and three-dimensional distribution maps of L*, a*, and b* values (Figure 3b) were generated for each group. Within the two-dimensional a*–b* coordinate system, the colour groups occupied four quadrants: pinkish-white was located primarily in the first and second quadrants, white was concentrated in the second and third quadrants, and the remaining colour groups were situated mainly in the third and fourth quadrants, with white positioned near the origin. In the three-dimensional L*–a*–b* map, each category was clearly separated into concentrated clusters, suggesting that the six colour groups defined by CIE Lab were suitable for the classification of H. syriacus flower colours.

3.1.2. Correlation and Cluster Analysis of Floral Phenotypes in Hibiscus syriacus

Heatmap analysis of the correlations among 38 traits in the 20 H. syriacus accessions (Figure 4) demonstrated that quantitative traits were predominantly positively correlated, whereas qualitative traits were mainly negatively correlated.

Among the traits related to the heart line, heart line length (HLL) was significantly positively correlated with heart length (HL,0.743**), heart length/petal length (HL1,0.589**), heart line length/heart length (HL2,0.448*),together constituting a coordinately varied heart characteristic system. Regarding pistil length-related traits, pistil length (PL2) was significantly positively correlated with flower diameter (FD,0.582**), pedicel length (PL1,0.577**), and heart line length (HLL,0.558*), indicating that pistil development was closely associated with overall flower size and heart traits. For flower type-associated traits, flower type (FT) and stamen petalisation degree (SPD,0.948**) were significantly correlated with petal length (PL,-0.452*). These results suggested that a higher degree of doubling was associated with reduced petal length and stamen development. Meanwhile, FC, a qualitative trait, was significantly correlated with colour parameters such as yellow-blue value (b*,-0.592**).

Based on the cluster analysis of floral phenotypic traits, the 20 H. syriacus germplasm accessions were categorised into five distinct groups (Figure 5). Group I consisted of one accession (5%; H06-PurP), characterised by a heart length (HL) and heart line length (HLL) significantly greater than those of other cultivars. Group II comprised four accessions (20%; H10-LavC, H16-MarI, H15-BaiF, H01-RedH), which exhibited a relatively large flower diameter (FD) and petal index (PI). Group III included one accession (5%; H09-FreC) with a high degree of doubling. Group IV contained three accessions (15%; H20-DiaN, H19-AphR, H13-RedW), which were distinguished by a short pedicel length (PL1). The remaining 11 accessions were classified into Group V (55%), characterised by relatively short pistil length (PL2).

3.2. Hibiscus Syriacus Pollen Morphology

3.2.1. Analysis of Pollen Morphological Diversity in Hibiscus syriacus

H. syriacus pollen morphology was examined using scanning electron microscopy. Pollen grains across all cultivars were monad, subspherical, and characterised by spinose surface ornamentation. The spines were conical with obtuse apices, and scattered pores were evenly distributed around the microspines. Granular protuberances were also observed on the pollen surface (Figure S1, Figure 6a–c).

Detailed data regarding qualitative pollen traits are presented in Table S4. Pollen shape was categorised as subspherical (80%) or spherical (20%). Surface wrinkling varied significantly among cultivars, with most appearing wrinkled and only a few identified as smooth. Furthermore, pollen spines were classified into straight or curved types, with some accessions possessing microspines (Figure S2, Figure 6).

Quantitative pollen traits of H. syriacus are detailed in Table S4 and Figure 7. The equatorial axis length (EAL) averaged 129.40 μm (range: 115.52–154.75 μm), whereas the polar axis length (PAL) averaged 128.01 μm (range: 115.52–149.74 μm). The PAL/EAL (P/E) ratio averaged 0.99 (range: 0.96–1.04), displaying a low CV; the maximum P/E ratio was observed in “Bai Fengche” and the minimum in “Blue Bird.”

The mean spine number (TN) for H. syriacus pollen was 36.2 (range: 28–58), with significant differences noted among cultivars (Figure 6). Although most cultivars exhibited a relatively low TN, “Red Woodbridge,” “Raspberry Smoothie,” and “Pink Giant” displayed elevated counts. The mean spine distance (TD) was 23.12 μm (range: 17.5–31.04 μm), with “China Chiffon” and “Raspberry Smoothie” exhibiting the maximum and minimum values, respectively.

The mean spine length (TL) of the 20 cultivars was 19.38 μm (range: 15.39–24.75 μm). Variation was observed among varieties, with “Diana” exhibiting the maximum TL and “Marina” the minimum. The mean spine basal width (TBW) was 9.34 μm (range: 7.81–10.92 μm); “French Cabaret Purple” had the largest TBW, whereas “Pink Giant” had the smallest. Finally, the mean TL/TBW ratio was 2.08 (range: 1.68–2.46), with the highest ratio found in “Diana” and the lowest in “Blue Bird.”

3.2.2. Correlation and Cluster Analysis of Pollen Morphology in Hibiscus syriacus

Heatmap analysis of significant correlations among 12 pollen morphological traits across 20 H. syriacus accessions (Figure 8) revealed four groups of positively correlated variables. Positive correlations were observed among EAL, polar axis length (PAL), spine number, and wrinkle degree (WD); among spine length (TL) and the TL/TBW ratio (0.802**); among spine curvature (TC) and TN (0.605**); and finally between TC and TS (0.464*). Three groups of negatively correlated variables were also identified. The first included pollen grain shape (PGS) and EAL (−0.520*);in the second group, TN was correlated with TD (−0.488*) and TBW (−0.45*); and the third involved TBW and TC (−0.444*).

These results indicated that the pollen traits of H. syriacus do not vary independently, and that significant synergistic or antagonistic correlations exist among pollen size, exine spinose ornamentation, surface microstructure, and other morphological characteristics.

Based on cluster analysis of 12 palynological traits, the 20 H. syriacus germplasm accessions were categorised into four distinct groups (Figure 9). Group I comprised seven accessions (35%; H19-AphR, H12-SpeC, H09-FreC, H07-FenY, H11-StaC, H08-XiaN, and H01-RedH), characterised by an EAL and polar axis length (PAL) below 120 μm. Group II consisted of a single accession (5%; H17-RasS) distinguished by a narrow spine distance (TD). Group III included three accessions (15%; H10-LavC, H13-RedW, and H05-PinG), defined by EAL and PAL measurements exceeding 140 μm and a high spine number (TN). Finally, Group IV contained nine accessions (45%; H02-BluB, H18-HamB, H03-PaeF, H06-PurP, H04-ChiC, H14-FenC, H16-MarI, H20-DiaN, and H15-BaiF) with EAL and PAL values ranging from 120 μm to 140 μm.

3.3 Fertility Evaluation and Observation of Insect Flower-Visiting Behaviour for Hibiscus syriacus Cultivars

3.3.1. Evaluation of Hibiscus Syriacus Cultivar Fertility

Pollen viability differed among the 20 Hibiscus syriacus cultivars. Overall, single-petaled cultivars exhibited notable higher pollen viability than that of semi-double-petaled cultivars, whereas semi-double-petaled cultivars exhibited higher pollen viability than that of double-petaled cultivars. Regarding performance, the cultivars ranked in the following order: ‘Starlight Chiffon’ > ‘Species’ > ‘Red Heart’ > ‘Diana’ > ‘Blue Bird’ > ‘Aphrodite’ > ‘French Cabaret Purple’ > ‘Lavender Chiffon’ > ‘Marina’ > ‘Fen Chiffon’ > ‘Bai Fengche’ > ‘Xiao Nvhai’ > ‘Pink Giant’ > ‘Red Woodbridge’ > ‘Purple Pillar’ > ‘Hamabo’ > ‘Paeonyflorus’ > ‘Raspberry Smoothie’ > ‘Fen Yun’ > ‘China Chiffon’ (Figure 10).

Fertility grades were classified based on fruit-set rates under open and artificial pollination conditions, as determined from their three consecutive years of field observations. The fruit-set rate under artificial pollination served as the primary evaluation criterion. The results showed that there was significant fertility differentiation among the 20 H. syriacus cultivars, with 13 low-fertility cultivars, 3 medium-fertility cultivars, and 4 high-fertility cultivars identified (Table 4).

3.3.2. Observation of Insect Flower Visitation

Field observations showed that the main flower-visiting insects of H. syriacus were flies, bees, and butterflies, with the overall visitation frequency in the order of bees > flies > butterflies. There were significant differences in insect visitation times among different H. syriacus cultivars, and cultivars with higher fruit-set rates corresponded to higher flower-visiting frequencies (Figure 11).

3.4. Correlations Between Fertility and Floral Phenotype and Pollen Morphology in Hibiscus syriacus

In this study, we employed correlation coefficient analysis via SPSS 25.0 to identify 9 indicators significantly associated with fertility in H. syriacus, encompassing floral phenotype and pollen morphology (Table S5).

3.4.1. Correlation Analysis Between Floral Phenotype and Fertility

Floral phenotypic traits were divided into quantitative and qualitative traits, and Spearman correlation analyses were conducted separately to assess the association between each trait category and fertility. After FDR correction, five floral quantitative traits were found to be significantly correlated with fertility, while no significant correlations were observed for floral qualitative traits. Among them, heart length (HL, r = 0.63**), the ratio of heart length to petal length (HL1, r = 0.579**), style length (PL2, r = 0.509*) and stamen number (SN, r = 0.475*) showed significant positive correlations with fertility. The ratio of heart line length to heart length （HL2，r = −0.549*） was significantly negatively correlated with fertility.

3.4.2. Correlation Analysis Between Pollen Morphology and Fertility

Spearman correlation analysis was conducted between 12 pollen morphological indicators and fertility. A total of 4 pollen morphological traits showed significant or extremely significant correlations with H. syriacus fertility (Figure 13). Among them, spine curvature (TC，r = 0.628**), spine surface (TS，r = 0.588**) and spine number (TN, r = 0.503*) were significantly positively correlated with fertility, whereas spine length (TL，r = −0.494*) was negatively correlated with fertility.

3.4.3. Principal Component Analysis of Pollen Morphological and Floral Phenotypic Traits

Correlation analysis between pollen morphological and floral phenotypic traits (Table S6) identified 19 fertility-related indicators for subsequent principal component (PCA) and comprehensive clustering analyses of the 20 hibiscus cultivars. PCA was used to reduce the data dimensions to identify key fertility-related factors, whereas cluster analysis grouped the cultivars to reveal differences in fertility-related phenotypic characteristics.

KMO and Bartlett’s test of sphericity were assessed before conducting the PCA. The indices met the required criteria (KMO ≥ 0.6, p < 0.01) [22], indicating that the dataset was suitable for PCA.

The floral phenotypic indicators consisted of 13 parameters: pedicel length (PL1), pistil length (PL2), outer petal posture (OPP), colorimetric value L*, heart presence (HP), sepal tube shape (STS), petal form (PF), stamen number (SN), stigma length (SL1), flower type (FT), stamen petaloidy degree (SPD), heart length (HL), and the ratio of heart length to petal length (HL1). For pollen morphology, the indicators included six parameters: the ratio of polar axis length to equatorial axis length (P/E), spine length (TL), spine number (TN), spine surface (TS), spine curvature (TC), and wrinkle degree (WD).

Principal component analysis performed on 20 H. syriacus varieties using these 19 indicators yielded five principal components (Table 5, Figure 14) with a cumulative contribution rate of 77.514%. This indicates that the selected components effectively represent the core information regarding fertility-related traits in this species.

Principal Component 1 (PC1:flower size + pollen spine structure factor) accounted for 26.651% of the total variance, representing the most explanatory component. Traits significantly influenced by its eigenvalues included pistil length (PL2), heart length (HL), and stamen number (SN), followed by pollen spine surface microstructure (TS), flower type (FT), and spine curvature (TC). PC2 (pedicel length + floral and pollen proportional structure factor) contributed 19.013%, with primary factors consisting of pedicel length (PL1), the ratio of polar axis length to equatorial axis length (P/E), and the heart length/petal length ratio (HL1). PC3 (flower lightness + calyx shape + petal posture + heart presence factor) explained 14.217% of the variance, driven by brightness value (L*), sepal tube shape (STS), outer petal posture (OPP), and heart presence (HP). PC4 (petal shape + pollen spine length factor) contributed 10.066%, with significant traits including petal form (PF) and pollen spine length (TL). Finally, PC5 (stigma length + heart presence factor) contributed 7.566%, primarily characterised by stigma length (SL1) and heart presence (HP).

3.4.4. Cluster Analysis

Based on R-type cluster analysis (Figure 15), the 19 traits were categorised into three distinct clusters at an Euclidean distance of 3. Within these clusters, pistil length (PL2) was strongly correlated with pollen spine number (TN), whereas pedicel length (PL1) showed a strong association with heart length (HL). Furthermore, the heart length/petal length ratio (HL1) demonstrated a strong correlation with pollen spine surface (TS) and pollen spine curvature (TC). These seven traits were inferred to be important indicators for distinguishing fertility levels in H. syriacus. The scattered distribution of traits shown in Figure 15 indicated that the selection was appropriate for use in subsequent Q-type cluster analysis.

Based on the 19 floral traits, Q-type cluster analysis classified the 20 H. syriacus accessions into five distinct groups (Figure 16).

Group I, comprising seven cultivars (35%), exhibited high fruit-set rates. These cultivars were characterised by high values for pedicel length (PL1), pistil length (PL2), stamen number (SN), heart length (HL), and heart length/petal length ratio (HL1), alongside the presence of microspines on the pollen spine surface. Group II (two cultivars, 10%) demonstrated low fruit-set rates, double FT, and low pollen spine numbers (TN). Group III (one cultivar, 5%) showed a high fruit-set rate and was associated with significant pistil length (PL2), high stamen number (SN), and high spine number (TN). Group IV (two cultivars, 10%) presented low fruit-set rates, spine numbers (TN) exceeding 50, and high rates of pollen abnormality. Finally, Group V (eight cultivars, 40%) displayed low fruit-set rates, accompanied by low values for pistil length (PL2), stamen number (SN), and heart length (HL).

4. Discussion

4.1. Diversity of Floral Traits in Hibiscus syriacus

Phenotypic diversity serves as the foundation for biodiversity research, and relevant data are widely utilised in the identification, classification, conservation, and genetic utilisation of germplasm resources. This approach has been effectively applied to species including Hibiscus mutabilis, Paeonia suffruticosa, Paeonia lactiflora, Hibiscus rosa-sinensis, and Gardenia jasminoides [23,24,25,26,27,28].

Previous investigations by our team confirmed that the floral traits of H. syriacus exhibit substantial diversity. Building upon those findings, the current study incorporated floral traits highly relevant to fertility, including pistil length (PL2), stigma colour (SC), and ovary shape (OS). At the palynological level, fertility-related indicators such as spine curvature (TC) and pollen spine surface (TS) were also included. When compared with the Guidelines for the Conduct of Tests for Distinctness, Uniformity, and Stability of New Varieties of Hibiscus syriacus and existing phenotypic indicators for Malvaceae and Hibiscus, the parameters selected in this study are more scientifically rigorous. We proposed that highly variable floral traits, including style length (PL2) and stamen length (SL), alongside pollen traits, can serve as auxiliary markers for early fertility screening in breeding programmes, thereby improving germplasm evaluation in H. syriacus. In subsequent phenotypic studies of H. syriacus, high-variability traits should be prioritised to facilitate the efficient differentiation of cultivars, while traits with low coefficients of variation, such as calyx length (CL), bract shape (BS), and petal wrinkling degree (PWD), should be avoided.

Among the 20 tested H. syriacus accessions, all quantitative trait diversity indices exceeded 1, reflecting rich phenotypic diversity [29]. Traits with high diversity indices included stigma length (SL1, 1.98), bract length (BL, 1.96), petal index (PI, 1.90), flower colour (FC, 1.67), and the degree of overlap between petals (OP, 1.23). The high diversity indices for stigma length (SL1) and BL suggest that these parameters represent important quantitative floral traits, providing essential morphological evidence for cultivar evolution, pollination adaptation, and germplasm classification.

Traits exhibiting large CV values included calyx width (CW, 52.52%), heart line length (HLL, 51.85%), pedicel length (PL1, 43.88%), sepal tube shape (STS, 75.286%), and stigma colour (SC, 75.28%). Specifically, the high CVs for CW and HLL indicate that these traits may be susceptible to genetic background or environmental conditions, reflecting strong phenotypic plasticity. We found that, in H. syriacus, the average CV for qualitative traits was higher than that for quantitative traits, whereas the diversity index showed the opposite trend. The negative relationship between the diversity index and the CV for the same trait is consistent with findings in Nymphaea tetragona, Elaeagnus angustifolia, and other species [30,31].

Coded quantification of H. syriacus quantitative traits can improve cultivar identification [32,33], and a comprehensive set of morphological indicators can establish a more complete classification basis. In this study, principal component and cluster analyses were conducted on a dataset that combined floral phenotypic and pollen morphological traits. The results showed that in addition to key traits identified in previous work—including spine number (TN), pedicel length (PL1), heart length (HL), and the heart length to petal length ratio (HL1)—newly added indicators such as pistil length (PL2), pollen spine curvature (TC), and pollen spine surface (TS) also serve as important references for these analyses.

Pistil length (PL2) has been widely used as an indicator for principal component and cluster analyses in various plant species, whereas pollen spine curvature (TC) and spine surface (TS) are rarely employed as core analytical indicators. The inclusion of these two traits enriches the quantitative dimensions of pollen morphology. Twenty H. syriacus cultivars were investigated, with five pollen grains measured per cultivar. This relatively small sample size may not adequately capture intraspecific variation in pollen morphology, representing a key sampling limitation. Therefore, the available data can only demonstrate correlative relationships. Future research should increase the sample size and incorporate design approaches to further functionally verify the hypothesis that pollen spine curvature (TC) and the presence or absence of small spines on pollen spine surfaces (TS) can improve fertility in H. syriacus.

4.2. The Significance of Key Traits for Fertility in Hibiscus syriacus

Research has shown that fertility in H. syriacus is governed by a multifaceted interplay among insect pollination patterns, nutrient composition, hormone levels, evolutionary adaptation, and floral phenotypic characteristics. Although in our previous work we identified a potential relationship between pollen characteristics and fertility, we did not undertake a systematic correlation analysis between floral traits and reproductive success. To address this limitation, multiple floral phenotypic and palynological indicators were integrated in this investigation to comprehensively examine their relationships with fertility. Pollen within the Malvaceae family is typically spinose, characterised by either straight or curved spines, with some featuring secondary microspines. Building upon these familial traits, pollen spine curvature (TC) and spine surface microspines (TS) were introduced as two targeted and innovative palynological indicators.

External plant morphology exerts a primary regulatory influence on fertility. This regulation is derived from evolutionary adaptations that determine the fundamental pollination mode. Moreover, quantifiable variations in traits among conspecific individuals directly influence fertility levels. In this study, H. syriacus flowers exhibited nectaries, large stigmas, and approach herkogamy, with stigmas protruding beyond the anthers. Because the stigma lobes cannot curl and the anthers are positioned lower on the monadelphous staminal column at varying heights, this species demonstrates a low rate of self-pollination [34]. However, these diverse floral traits effectively attract insect visitors, which promotes frequent interspecific gene flow and supports the diversity of H. syriacus genetic resources.

In this work, we identified five external floral morphological traits significantly correlated with fertility in H. syriacus: Stamens Number (SN), pistil length (PL2), heart length (HL), Heart line length over heart length(HL2), and the heart length to petal length ratio (HL1). A higher number of stamens enhances pollen supply and improves fertility. A longer pistil facilitates pollen adhesion and germination, and supports the selection of high-quality male gametes [35]. The association between pistil length and fertility has been widely confirmed in other ornamental species [36,37,38] and crops [39,40]. Regarding heart-related traits, research has shown that Hibiscus rosa-sinensis individuals with a longer heart length and a higher heart length to petal length ratio attract pollinators, such as honeybees, more effectively [41]. The floral eye pattern serves as a visual signal that increases pollination efficiency and fertility. Other plant species similarly attract pollinators and enhance reproductive success through variations in flower colour [42,43], indicating that visual floral traits, such as the H. syriacus heart, act as pollination signals and represent a significant pollination regulatory strategy.

Pollen morphological characteristics, such as size, shape, and surface structure, directly influence the efficiency of insect-mediated pollen transport and dispersal [44,45]. When insect foraging behaviour matches pollen morphology, transfer efficiency is greatly improved [46]. In this investigation, we found that four pollen morphological indicators were significantly correlated with fertility: spine length (TL), spine number (TN), spine curvature (TC), and spine surface microspines (TS). Research has suggested that spines on pollen enhance adhesion and reduce the risk of pollen being groomed off by insects, thereby benefiting pollination [47]. Here, we found that H. syriacus individuals with shorter pollen spines had higher fruit-set rates, consistent with the preference of honeybees for collecting short-spined Malvaceae pollen [48,49]. Furthermore, research by Vaissière showed that only 16% of honeybees selected long-spined upland cotton pollen, as long spines hinder dense packing, require additional binding fluid, and increase behavioural costs, thus lowering honeybee preference [50]. Conversely, a higher number of pollen spines increases the probability of deposition and adhesion on the stigma, thereby improving the fruit-set rate [51]. Flies, butterflies and other insects are common flower visitors, but do not represent the dominant pollinator groups. Therefore, analyses on floral-visitation preferences of these insects were not performed in the current study, constituting a limitation of this research. Follow-up studies should investigate the foraging preferences of different flower-visiting insect taxa in greater depth.

Pollen spine curvature significantly influences pollen adhesion. Mildly curved spines demonstrate a higher adhesion capacity, a finding that supports previous hypotheses and aligns with existing literature [6]. Lunau et al. observed that after the long spines of Alcea rosea pollen were bent, bumblebees began collecting pollen they had previously rejected. Research has demonstrated that the straight, rigid spines found on Malvaceae pollen hinder collection by honeybees, serving as a specific mechanical defence mechanism [52]. However, some H. syriacus cultivars (e.g., Pink Giant, Raspberry Smoothie) display excessively curved, malformed and shrunken pollen spines, together with poor pollen viability. We hypothesise that pollen malformation is correlated with reduced pollen viability in H. syriacus and that these two factors jointly contribute to the low fruit-set rate [53].

We propose that microspines located on pollen spines enhance the probability of pollen adhesion. Comparable secondary microspines and attached particles have been observed on the pollen spines of Abutilon and Gossypium (Malvaceae), as well as in several plant families. In this study, H. syriacus cultivars possessing these microspines—specifically Red Heart, Red Woodbridge, and Hamabo—demonstrated high fruit-set rates. We suggest that these microspines represent a mechanical adaptation through which plants improve fertility and fruit-set rates. In addition, we found that the degree of pollen wall wrinkling is also associated with fertility, and that this physical trait correlates with specific colorimetric parameters [54,55]. A significant positive correlation was observed between the colorimetric a* value and the pollen wrinkling degree. We infer that a higher a* value reflects an elevated pigment content, reduced pollen wall stability, and greater wrinkling degree, and vice versa.

5. Conclusions

The floral phenotype and pollen morphology of H. syriacus exhibited significant diversity. Diversity indices for quantitative traits ranged from 1.23 to 1.98, whereas those for qualitative traits ranged from 0.19 to 1.67. Flower colour was categorised into six distinct colour groups. Furthermore, the cultivars showed significant differences in pollen grain size, spine morphology and number, surface wrinkle degree (WD), and other morphological indicators.

A total of 9 fertility-related indicators were identified, including four pollen morphological indicators ( spine length, spine number, spine curvature, and the presence of microspines on the pollen spine surface) and five floral phenotypic indicators (pistil length, heart length, heart length to petal length ratio, heart line length over heart length, and stamen number).

These findings clarify the primary floral phenotypic basis for fruit setting in H. syriacus and deepen the current understanding of reproductive regulatory mechanisms. The results also highlight the important contributions of pistil length, pollen spine curvature, and the presence of surface microspines on the pollen spines to H. syriacus fertility. The 9 screened indicators serve as reliable morphological markers for the rapid screening of hybrid parents. They provide a scientific foundation and key technical support for H. syriacus genetic improvement and breeding programmes, markedly contributing to germplasm innovation and the development of high-quality industrial cultivars.