Introduction
Hainan Shanlan upland rice is a unique rice germplasm domesticated and cultivated over generations by the ethnic minorities of Hainan, such as the Li and Miao peoples, through long-term traditional agricultural practices. It is widely distributed across the central and western mountainous regions of Hainan Province. As a representative upland rice from Hainan, Shanlan upland rice has become the subject of extensive scientific research. Previous studies have screened Shanlan upland rice for drought-tolerant germplasm (Xu et al. Reference Xu, Yang and Xu2018; L-Q et al. Reference L-Q, Q-J, Han, Xu, Zhu, Yan, Wang and Wang2019), investigated its drought-resistant physiological characteristics (Zheng et al. Reference Zheng, Chen and Huang1997; Liu et al. Reference Liu, Xu, He, He, Wen, Zhang and Yuan2014) and elucidated the underlying molecular mechanisms of its drought resistance (Yang et al. Reference Yang, Niu, Li, Wang, C-Y, J-N, Yuan and Pei2023b). In addition to drought-related studies, researchers have also explored Shanlan upland rice’s resistance to pests and diseases, along with its adaptability to various environmental stresses. The resistance of Shanlan upland rice to brown planthopper is largely due to strengthened sclerenchyma tissue and hormonal regulation (Liu et al. Reference Liu, Lou and Lv2024). Through the evaluation of salt tolerance in 46 rice accessions, including 40 Shanlan upland rice varieties, nine highly salt-tolerant varieties were identified and key salt tolerance evaluation indices at the tillering stage were established (Qu and Yuan Reference Qu and Yuan2024). Through the evaluation of resistance to bacterial blight in 17 Shanlan upland rice varieties under both irrigated and drought conditions, it was found that drought conditions conferred stronger resistance. This resistance was also associated with the presence of specific resistance genes in the varieties (Liu et al. Reference Liu, Huang, Zhai, Li, Chi, Chen and Yuan2021). The origin and evolution of Shanlan upland rice are also key research topics. The sequence diversity and haplotypes of three nuclear genes, one mitochondrial gene and one chloroplast gene were analysed. Their findings revealed that Shanlan upland rice has a closer genetic relationship with common wild rice (Oryza rufipogon) from Guangdong and Hunan provinces, while showing a more distant relationship with common wild rice from Hainan (N-n et al. Reference N-n, Wei, D-Y and Q-w2013).
In recent years, there have been numerous reports on the genetic dissection of rice agronomic traits. Through grey relational analysis, the relationships between agronomic traits, quality and yield of rice in Xinjiang were evaluated. The results revealed that yield was primarily influenced by 1,000-grain weight, plant height (PH) and seed setting rate (SSR). These findings provide a basis for multi-trait coordinated breeding of rice in arid regions (Zhao et al. Reference Zhao, Zhu, Tang, Kang, j, Zhang, Wang, Du and Hou2025). A genome-wide association study (GWAS) was conducted on eight agronomic traits across 124 newly developed rice lines from Yunnan, resulting in the identification of 39 significant quantitative trait locus’ (QTLs) (P ≤ 1.0 × 10−⁵). These findings provide new targets for marker-assisted selection in rice breeding (Yang et al. Reference Yang, Niu, Li, Wang, C-Y, J-N, Yuan and Pei2023). However, research specifically targeting the agronomic traits of Shanlan upland rice remains relatively limited. By investigating the effects of different planting densities and cultivation methods on the agronomic traits and weed control efficacy of Hainan upland rice, this study aims to provide technical support for organic upland rice cultivation (He et al. Reference He, Zhao, He, Ke and Yuan2018). Our team has long been dedicated to the collection, conservation and innovative utilization of Shanlan upland rice germplasm resources (Tang et al. Reference Tang, Yan, Yang, Zhong and Tang2018). We have conducted surveys and preliminary investigations on its agronomic traits. Studies revealed that the rice performed exceptionally well in key traits such as panicle length (PL), SSR and 1000-grain weight, contributing valuable insights for breeding superior Shanlan rice lines (Yang et al. Reference Yang, Yang, Zhong, Huang, Wang and Tang2018). Additional analysis of 65 Shanlan upland rice resources highlighted significant genetic diversity, with agronomic traits such as PH and PL showing distinct variability. This was quantified by coefficients of variation that ranged from 10.6% to 30.5%, reflecting the substantial trait differences within the studied germplasm (Yang et al. Reference Yang, Yan, Zhong, Huang, Wang and Tang2019). A separate study on 56 red rice resources from the Shanlan variety provided crucial information for the improvement and breeding of Shanlan red rice varieties (Tang et al. Reference Tang, Yang, Yan, Chen and Tang2020). Additionally, through quality analysis of Shanlan upland rice, we have screened and identified two germplasms meeting the high-quality glutinous rice standards (Yang et al. Reference Yang, Huang, Wang, Zhong, Yan and Tang2017).
In this study, 162 Shanlan upland rice accessions were used as research materials. Through field experiments combined with statistical analysis, the agronomic traits of Shanlan upland rice were systematically identified, and phenotypic variation and correlations between traits were revealed. This research is expected to deliver valuable breeding materials for future studies and establish a theoretical foundation for conserving and utilizing Shanlan upland rice resources.
Materials and methods
Experimental materials
The materials used in this study consisted of 162 accessions of Shanlan upland rice germplasm resources systematically collected by our team from multiple geographical regions in Hainan. The detailed geographical distribution of rice germplasm is presented in Figure S1.
Investigation of agronomic traits
All rice accessions were planted continuously for 2 years (2023–2024) in Sanya. Each rice material was planted in 2 rows with 10 plants per row, and the plant spacing was 25 × 15 cm. Five representative plants per accession were measured, with the averaged value used for analysis. Agronomic traits evaluation was performed in accordance with the Descriptors and Data Standards for Rice (Oryza sativa L.) Germplasm Resources (Han and Wei Reference Han and Wei2006). Phenotypic measurements covered 11 traits: PH, PL, flag leaf length (FLL), flag leaf width (FLW), number of primary branches (NPB), grain number of primary branches (GPB), number of secondary branches (NSB), grain number of secondary branches (GSB), grain number per main panicle (GNP), SSR and effective tiller number (ETN).
Statistical analysis of trait data
Population distribution frequency, the assessment of normality for population phenotypic distribution and correlation analyses were conducted using GraphPad Prism 9.0.0, and cluster analysis was performed with OriginPro 2025b.
Results
Characterization of Shanlan upland rice
Statistical analysis of 11 agronomic traits in 162 Shanlan upland rice germplasm resources (Tables 1 and 2) revealed the following variation characteristics: In 2023, the coefficient of variation (CV) for the traits ranged from 14.11% to 46.04%, while in 2024 it ranged from 11.45% to 44.82%, indicating the rich genetic diversity of Shanlan upland rice. Among these, the PL showed relative stability over the two years, with a range of 15.68 cm (17.12–32.98 cm) and 17.27 cm (15.75–33.02 cm), respectively. The CV for this trait was the lowest among all traits (CV = 14.11%/11.45%), suggesting a high level of genetic conservatism in this trait within the Shanlan upland rice population. Notably, the NSB exhibited significant phenotypic differences, with ranges of 9.8–73.2 in 2023 and 11.4–76.8 in 2024, with range values of 63.4 and 65.4, respectively, and its CV was the highest among all traits (CV = 46.04%/44.82%). Further analysis showed significant differences in the mean values for PH, FLL, NSB, GSB and GNP between the 2 years, suggesting that these traits are more affected by cultivation environment and water-fertilizer management practices. In contrast, traits such as PL, FLW, NPB, GPB, SSR and ETN remained relatively stable across the two years, indicating strong genetic stability in these traits.
Table 1. Phenotypic characterization summary of Shanlan upland rice in 2023
Table 2. Phenotypic characterization summary of Shanlan upland rice in 2024
Correlation analysis of agronomic traits
This study employed phenotypic association analysis to elucidate the covariation network characteristics among agronomic traits (Fig. 1). The data revealed that ETN exhibited marked phenotypic independence, as no significant correlations with other traits were detected across years (|r| < 0.20, P > 0.05), and most of the associations were negative. SSR showed year-specific correlation pattern: in 2024, it showed no significant correlation with any other traits (|r| < 0.20, P > 0.05), whereas in 2023, it was significant and negatively correlated with the GSB (r = −0.21, P < 0.01). Importantly, yield-related components, including NBP, GBP, NSB, GSB and GNP, exhibited a highly stable co-expression pattern across years (r ≥ 0.42, P < 0.01), suggesting these traits are tightly linked within the genetic regulatory network.
Figure 1. Correlation analysis of Shanlan upland rice. PH, plant height; PL, panicle length; FLL, flag leaf length; FLW, flag leaf width; NPB, number of primary branches; GPB, grain number on primary branches; NSB, number of secondary branches; GSB, grain number on secondary branches; GNP, grain number per main panicle; SSR, seed setting rate; ETN, effective tiller number.
We further analysed the correlations among traits other than yield-related traits mentioned above. The results indicated that most of the traits showed positive correlations, though the significance levels varied between growing seasons. In 2024, most of the agronomic traits were significantly positively correlated (r > 0.2), except for the correlations between FLL and FLW (r = 0.16), and FLL and GBP (r = 0.15), which were not statistically significant. In 2023, a similar trend of positive correlations was observed, but the correlations of PL with NBP (r = 0.08), GBP (r = 0.02), NSB (r = 0.10), GSB (r = 0.15) and GNP (r = 0.12), as well as between GPB and PH (r = 0.14) or FLL (r = 0.13), were not significant.
Cluster analysis
To systematically elucidate the genetic characteristics and variation patterns of yield-related traits in Shanlan upland rice, GNP and SSR were taken as target traits, Euclidean distance was employed as the metric for measuring similarity among samples, and hierarchical cluster analysis was conducted (Figs. 2 and 3). Cluster analysis revealed that when the Euclidean distance threshold was set at 30, followed by sorting in ascending order of GNP, the Shanlan upland rice population in both years could be clearly divided into five major groups (Table 3). In the two growing seasons, Group I included 41 and 43 accessions, respectively; Group II included 43 and 39 accessions; Group III included 39 and 25 accessions; Group IV included 26 and 39 accessions; and Group V included 13 and 16 accessions. Although the overall classification results were consistent between the two years, there were differences in the composition of Group III and Group IV. This observation suggests that yield-related traits in Shanlan upland rice are significantly influenced by interannual variations in climatic factors such as light duration and rainfall distribution, as well as by precise management practices related to water and nutrient input.
Figure 2. Cluster analysis of yield-related traits (grain number per main panicle and seed setting rate) in Shanlan upland rice (2023).
Figure 3. Cluster analysis of yield-related traits (grain number per main panicle and seed setting rate) in Shanlan upland rice (2024).
Table 3. Comparison among clusters
Screening for superior germplasm
The study revealed that Group V comprised 13 and 16 accessions in the two years, respectively, all of which had more than 240 grains per panicle and a SSR above 50%. The average GNP in these groups was 264.2 and 277.9, respectively, representing the highest values among all clusters for each year (Table 3). Among these, three accessions consistently exhibited superior yield traits across both growing seasons (Table 4). These accessions originated from Yinggen Town (2 accessions) and Wanling Town (1 accession) in Qiongzhong Li and Miao Autonomous County. Compared with the population averages, these accessions had significantly higher GNP, showing a strong potential for substantial sink capacity and providing a solid phenotypic foundation for high-yield breeding. Among the identified superior accessions, Line 69 stood out in particular (Figure S2). It achieved 261.4 and 305 grains per panicle over the two years, with consistently high SSRs of 93.79% and 90.07%, respectively.
Table 4. Accessions with high-yield potential
Discussion
Through a systematic analysis of 11 core agronomic traits across 162 Shanlan upland rice germplasm accessions, this study revealed the rich phenotypic diversity present in this population. The results showed high coefficients of variation for agronomic traits, with yield-related traits exhibiting especially wide variation – likely attributable to genetic differentiation driven by geographic isolation across diverse ecological zones. The abundant genetic diversity within the Shanlan upland rice germplasm offers a valuable gene pool for breeding programs. Additionally, differences were observed in the interannual correlation analysis of agronomic traits, and such variations in trait correlations likely reflect the influence of environmental factors on the trait interaction network. Moreover, variations in light intensity, temperature, precipitation and other environmental conditions across years may alter the regulatory mechanisms underlying trait expression, thereby affecting trait correlations. In particular, the accessions from Group V in two years, with more than 240 grains per panicle, suggest a strong yield potential and highlight the unique advantages in terms of productivity. The identification of three elite germplasms not only enriches the genetic pool for rice breeding but also provides crucial material support for the development of broadly adaptable rice varieties suited to diverse environments.
Landraces are characterized by rich genetic heterogeneity, strong environmental adaptability, and unique agronomic trait combinations. These germplasms serve as a “natural variation reservoir” for plant genomes and possess irreplaceable strategic value in modern crop genetic improvement. For example, researchers have cloned the major gene Wx(lv), which controls high amylose and low viscosity in rice, from a local indica variety using map-based cloning (Zhang et al. Reference Zhang, Zhu, Chen, Fan, Li, Lu, Wang, Yu, Yi, Tang, Gu and Liu2019). Another study conducted GWASs on 117 landraces to identify the nitrogen-use-related gene OsTPP6 (Ren et al. Reference Ren, Tian, Liu, Yu, Tang, Chen, Lin, Li, Wang and Wang2022). Moreover, blast-resistance genes Bsr-d1 and pid3 from highly resistant local germplasms were identified (Shang et al. Reference Shang, Tao, Chen, Zou, Lei, Wang, Li, Zhao, Zhang, Lu, Xu, Cheng, Wan and Zhu2009; Li et al. Reference Li, Zhu, Chern, Yin, Yang, Ran, Cheng, He, Wang, Wang, Zhou, Zhu, Chen, Wang, Zhao, Ma, Qin, Chen, Wang, Liu, Wang, Wu, Li, Wang, Zhu, Li and Chen2017). As a characteristic landrace resource of Hainan, Shanlan upland rice exhibits rich phenotypic and genetic diversity. A systematic investigation of 11 agronomic traits across the Shanlan upland rice population over two consecutive growing seasons, showing that all traits except SSR exhibited either strict or approximate normal distribution (Figure S3). This widespread normality strongly supports that these traits are typical quantitative traits, whose expression is jointly regulated by polygenic inheritance and random environmental fluctuations. This not only validates the rationality of selecting this population as a research object for agronomic trait genetic analysis but also provides a statistically reliable phenotypic foundation for subsequent QTL mapping of important agronomic traits in Shanlan upland rice.
It is important to acknowledge the limitations of the present study: our analysis relied exclusively on phenotypic data, without integrating transcriptomic or genomic information. Nevertheless, the phenotypic date of Shanlan upland rice investigated in this study lay a critical groundwork for future integrative research that combines phenomic and genomic data. Specifically, GWAS can be employed to identify key QTLs for yield-related traits. Future work will include genome analysis of elite accessions such as Line 69 to identify structural variations or novel genes associated with large panicle size and high SSRs.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1479262125100397.
Author contributions
Q.T. and F.X. conceived and designed this research. Y.L performed the experiments. Y.L. and L.Z. wrote the original draft. X.Y., Q.T. and and F.X., contributed to materials collection, preservation, classification. Y.L., L.L., X.X., M.Z., Y.Y. and Y.H. engaged in field sowing and other tasks. All authors have read and agreed to the published version of the manuscript.
Funding statement
This work was supported by Hainan Provincial Natural Science Foundation of China (324QN345); Hainan Province Science and Technology Special Fund (ZDYF2024KJTPY027, ZDYF2024XDNY165); Species and Variety Resources Protection Fund Project of Department of Agriculture (111821301354052033); Basic Scientific Research Project of HAAS (HAAS2023RCQD19, HAAS2025KJCX01, HAAS2025CYFH02, 2024-LZSQN005).
Competing interests
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as potential conflicts of interest.
Disclosure statements of AI tools used
During the writing of the paper, I utilized AI tools solely for non-substantive polishing purposes. The specific AI tool employed is Deepseek. No other AI tools were used in the research design, data collection, analysis, interpretation of results, or drafting of core content.
