This experiment utilized 56 varieties of Chinese kale from Guangdong and Fujian (which were preserved and propagated by our research team), abbreviated as GW (30 samples), GY (5 samples), FW (10 samples), and FY (11 samples) according to their flower color and origin. Each sample was as-signed a specific number, as detailed in Table 1.

Selected full-grained seeds were soaked for 2 h at room temperature, placed in petri dishes lined with moist filter paper, and cultured in the dark at 25°C for 3 days. When the root tips reached a length of 10–15 mm, they were harvested at 9 am [20].

Root tips were harvested and pretreated in 2 mmol·L⁻¹ 8-hydroxyquinoline solution for 5 h at 4°C. The pretreated root tip materials were fixed in Carnoy solution (volume ratio of absolute ethanol to acetic acid was 3:1) at 4°C for 24 h, washed with 95%, 85%, and 75% ethanol gradient, and finally kept in 75% ethanol solution. The root tip was dissociated with 1 mol·L⁻¹ HCl at 60°C for 8 min, then washed with distilled water for 3 times, stained with Carbol-fuchsin solution for 5–10 min in dark, and squashed for viewing under 100 times oil lens and 10 times eyepiece of OlympusCX21 microscope.

Thirty well-spread, morphologically clear metaphase cells were selected for chromosome counting and karyotype analysis, following the criteria of Li et al. [21]. Chromosome measurements and classifications were done based on the standards of Levan et al. [22]. The karyotype asymmetry coefficient was calculated using the method of Arano [23], and karyotype classification was performed according to the regulations of Stebbins [24].

Experimental data were processed using Excel 2019. Partial least squares discriminant analysis (PLS-DA) was conducted using SIMCA-P 14.1. Heatmaps were generated using MeV 4.9.0, and box plots were created using Origin 2021. Correlation analysis was performed with Excel 2019, and the results were visualized using Cytoscape 3.9.1 [25].

All 56 varieties of Chinese kale were diploid with 2n = 2x = 18, and no cells with abnormal chromosome numbers were detected (Fig. 2). Each variety possessed a pair of satellite chromosomes, with variations in length and location among different varieties. The satellites were predominantly located on chromosomes 7, 8, and 9, present in approximately 78.57% of all Chinese kale varieties (Fig. 3). Chromosome 8 had the highest frequency of satellites, observed in 17 varieties. Additionally, the satellite-bearing chromosomes were primarily of the sm type, representing 64.29% of all varieties, including 56.67% of GW, 100% of GY, 50% of FW, and 81.82% of FY.

As shown in Table 2, most Chinese kale varieties consisted of both median-centromere (m) and submedian-centromere (sm) chromosomes, while a few varieties contained only median-centromere (m) chromosomes. The specific karyotype formulas varied among different varieties of Chinese kale.

When analyzed together with Tables 2 and S1, the relative lengths of chromosomes ranged from 7.39% to 14.58% in GW, 8.54% to 14.33% in GY, 8.35% to 14.09 in FW, and 8.69% to 14.29% in FY. The centromere indices for GW, GY, FW, and FY ranged from 25.67–49.64, 27.61–48.7, 27.13–48.93, and 31.86–49.35, respectively. The ranges of relative chromosome lengths and centromere index of GW represented all the groups’ ranges in this experiment. However, the range of ratios of the longest and shortest chromosomes for GW was observed at 1.31–1.87, encompassing those of GY (1.31–1.62) and FW (1.42–1.59), but not fully encompassing that of FY (1.21–1.61).

Among 56 samples, the highest mean arm ratio was 1.68, and the lowest was 1.24. In GW, the arm ratio varied from 1.01 to 2.90, with 8.89% of chromosomes having an arm ratio greater than 2 and a mean arm ratio of 1.47 for the entire group. In GY, the arm ratio ranged from 1.05 to 2.62, with 17.78% of chromosomes having an arm ratio greater than 2 and a mean arm ratio of 1.55 for the group. In FW, the arm ratio ranged from 1.04 to 2.69, with 10.00% of chromosomes having an arm ratio greater than 2 and a mean arm ratio of 1.48. In FY, the arm ratio varied from 1.03 to 2.14, with 45.45% of chromosomes having an arm ratio greater than 2 and a mean arm ratio of 1.47. Interestingly, although GW had the largest range in arm ratios among the four groups, it had the smallest proportion of chromosomes with an arm ratio greater than 2. In contrast, FY, which had the smallest range in arm ratios, had the largest proportion greater than 2. The karyotype asymmetry index was similar across the four groups, ranging from 55.16% to 61.29% for GW, 59.17% to 61.39% for GY, 56.42% to 61.93% for FW, and 56.66% to 61.17% for FY. Chinese kale exhibited two karyotype types: 1A, accounting for 44.64%, and 2A, accounting for 55.36% (Table 2).

To analyze the relationships among the four groups of Chinese kale based on karyotypic parameters, partial least squares discriminant analysis (PLS-DA) was conducted. As shown in Fig. 4, PLS-DA1 explained 12.8% of all parameter information, effectively differentiating between Chinese kale from Guangdong (GW and GY) and Fujian (FW and FY). PLS-DA2 explained 12.5% of all parameter information, distinguishing GY from FW. Overall, there was overlap among the groups, making complete clustering challenging. However, some regional differences were observed, providing a basis for further differential analysis.

To comprehensively study the karyotype differences among the four groups of Chinese kale, nine differential indicators were selected from VIP values greater than 1, based on PLS-DA. Heatmaps and box plots were generated to visually observe group differences (Fig. 5). As shown in Fig. 5A, the karyotype indicators were generally clustered into two major groups. One group included the relative length of chromosome 9, the relative length of the short arm of chromosome 7, and the minimum centromere index, while the other six indicators clustered into the second group.

From the nine box-plots in Fig. 5, although the ranges of GW indicators were broad, their median values were centered among all groups. The GY group showed higher median values than the other groups for the relative length of the short arm of chromosome 3, the relative length of chromosome 3, and the maximum arm ratio (Fig. 5B,C,G), but had significantly lower values for the relative length of the short arm of chromosome 7 and the minimum centromere index (Fig. 5I,J). The FW group had the lowest median value for the relative length of chromosome 2 among all groups, while the FY group had a significantly higher median value for the relative length of chromosome 9 compared to the other groups (Fig. 5D,H).

To investigate the correlations among kale karyotype parameters, 41 relationships were identified (Fig. 6). The relative length of the long arm and the relative length of the short arm of chromosome 2 to 8 were negatively correlated. The chromosome type was positively correlated with the number of that chromosome type within the set. There was a strong negative correlation between the minimum centromere index and the maximum arm ratio, as well as between the maximum centromere index and the minimum arm ratio. Although karyotype classification is typically associated with chromosome length ratio, correlations were found between chromosome karyotype and maximum arm ratio, minimum centromere index, karyotype asymmetry, and the proportion of arm ratios greater than 2, but not with chromosome length ratio (Fig. 6).

To assess the evolutionary degree of Chinese kale, karyotypic evolutionary trends were analyzed using the average arm ratio and karyotypic asymmetry index as references. As observed in Fig. 7, the overall evolutionary level of GY was the highest, while the remaining three groups of Chinese kale varieties were evenly distributed across the plot. GW, the group with the most varieties, included the oldest Chinese kale, with GW9 notably distinct from the other varieties. Overall, GY was the most evolved among the four groups of Chinese kale.

In this study, the number of chromosomes for all Chinese kale was 2n = 2x = 18, each with two satellites, consistent with previous research [26–31]. However, the distribution of these satellites differed. Earlier studies typically localized most satellites on chromosome 7 [27,29,31]. In the study by Song et al. [26], satellites were found to be located on chromosomes 6 and 7, while Yuan et al. [30] reported their presence on chromosome 9. In contrast, our study, which examined multiple varieties, found satellites on chromosomes 4 to 9. Xia et al. [28] reported that the chromosome types of satellites were all sm, whereas this study revealed the presence of satellites on the m chromosome type in 35.71% of all varieties. Previous studies classified all karyotypes as “2A”. However, we observed that many Chinese kale varieties had a “1A” karyotype, providing a reference for future studies on the karyotype of Chinese kale.

The karyotype formulas of the four groups was primarily composed of multiple m chromosomes and a few sm chromosomes, with some groups having only m chromosomes. In GY, the predominant karyotype formula was 2n = 2x = 18 = 14m + 4sm (2SAT), with no exclusively m chromosomes. In both GY and FY, most chromosomes with satellites were of the sm type. This phenomenon was less pronounced in GW and FW, where only about half of the chromosomes with satellites were sm type. Further investigation is needed to determine if there is a relationship between satellite chromosome type and flower color in Chinese kale.

The average arm ratio of each group of Chinese kale was approximately 1.5, indicating genetic stability. The proportion of chromosomes greater than 2 differed among groups, with FY exhibiting a significantly higher proportion at 45.45% compared to the other three groups. Additionally, the relative length of chromosome 9 in FY was generally the highest among four groups, possibly contributing to its low chromosome length ratio (Fig. 5F,H). GY exhibited higher relative length of the short arm of chromosome 3 and maximum arm ratio compared to other groups (Fig. 5B,G), but significantly lower relative length of the short arm of chromosome 7 and the minimum centromere index (Fig. 5I,J). FW had the lowest relative length of chromosome 2 among the four groups (Fig. 5D). These variations constitute key karyotypic differences among the different Chinese kale groups.

According to studies by Li et al. [32] and Zhang et al. [33], yellow flower color dominance over white flower color in Chinese kale followed single-gene inheritance as a qualitative trait. Li et al. [32] identified genes related to flower color inheritance on the long arm of chromosome 3. Although differences in short arm length and total length for chromosome 3 were observed among groups (Fig. 5), the long arm length of this chromosome did not differ significantly. Therefore, it can be concluded that flower color gene in Chinese kale was genetically stable and rarely mutate.

The asymmetry index is widely recognized as a key indicator for assessing the evolutionary level of plants. In this study, the asymmetry index of the 56 varieties of Chinese kale was low, ranging from 55.16% to 61.93%, indicating that Chinese kale is an ancient plant.

In Fig. 6, there was no correlation between the karyotypes of Chinese kale and the chromosome length ratio, as the chromosome length ratios of all kale varieties were less than 2, consistent with previous studies [26–31]. It can be concluded that the karyotype of Chinese kale is determined by the proportion of arm ratios greater than 2. The stability of the chromosome length ratio in Chinese kale indicated the stability of its chromosomal inheritance.

As observed in Fig. 7, the scatter points representing the four groups of Chinese kale were evenly distributed on the karyotypic evolutionary trend map, with GY appearing to be the most evolved, likely due to subsequent hybridization. GW, however, included several primitive varieties, with GW9 identified as particularly primitive and distinct from the group. Geographically, genetic exchanges between Guangdong and Fujian have likely contributed to frequent genetic exchanges among Chinese kale varieties grown in these regions [34]. In practice, most varieties grown in Guangdong are white-flower Chinese kale, while those in Fujian are yellow-flower Chinese kale [35]. Combined with Fig. 4, GW and FY were located on the left and right sides of the distribution, respectively, while GY and FW were positioned between them. Therefore, it was hypothesized that white-flower Chinese kale originated in Guangdong, yellow-flower Chinese kale originated in Fujian, and GY and FW were hybrids of GW and FY. However, relevant molecular evidence is still lacking, and further research on the evolution of Chinese kale is needed.