2.1 Plant materials, soil and chemicals

Seven rice (Oryza sativa) cultivars, Huagan-3, Huagan-8, Huafeng, Lingyou-6173, Lingyou-6365, Lingyou-3826 and Liaojing, were used for this study. The pairs Huagan-3/Huagan-8 and Lingyou-6173/Lingyou-6365 were bred from reciprocal parents or identical male sterile, represent the closest relatedness and have the ability for kin recognition at the cultivar level (Xu et al., 2021; Yang et al., 2018). Huafeng and Lingyou-3826 were obtained from other parents or male sterile lines that do not have consanguinity ties with the two closely related pairs. These cultivars were divided into two sets of genotypes, indica-inbred and indica-hybrid. In the indica-inbred rice set, Huagan-3 was the focal cultivar. Huagan-8 and Huafeng were the closely related and distantly related cultivars, respectively. In the indica-hybrid rice set, Lingyou-6173 was the focal cultivar with Lingyou-6365 and Lingyou-3826 as the closely related and distantly related cultivars, respectively. In addition, the japonica-inbred cultivar Liaojing was used as a more distantly related cultivar for both sets. The relatedness of cultivars in both sets has been confirmed by genetic distances to the focal cultivar with simple sequence repeat markers (Yang et al., 2018). The seeds of two common paddy weeds Echinochloa crus-galli and Leptochloa chinensis were collected from local rice fields for use in weed competition experiments.

Soil was collected at Nanling Experimental Station of Rice Research, Anhui Province, China (30°56′52″ N, 118°23′15″ E), which located in the low mountains on the southern side of the Yangtze River. Soil was collected randomly from the surface (0–10 cm) of a paddy field. The soil is a typical fluvaquent, Etisol (US taxonomy) with pH 5.6, organic matter of 24.5 g·kg^-1, total nitrogen of 1.4 g·kg^-1, available phosphourus of 29.3 mg·kg^-1 and available potassium of 56.8 mg·kg^-1. Soil samples were air-dried, sieved (2 mm mesh) to remove plant tissues, and used in the series of greenhouse experiments described below.

(–)-Loliolide was isolated from root exudates using previously developed methods (Kong et al., 2018) and verified with its authentic standard obtained from Yuanye Biology Corporation (Shanghai, China). Organic solvents and other chemicals were purchased from China National Chemical Corporation (Beijing, China) and were of the highest purity available.

2.2 Field trials

Two paddy fields were selected at the experimental station described above during the 2021 growing season. The paddies had previously been planted with rice for several seasons. Both fields were divided into a series of 3 × 5 m plots arranged in a completely randomised design of two rice sets and four combinations with four replicates accounting for 32 plots. Plots were separated by trenches with at least 30 cm discard strips on each side. Each focal cultivar was paired with itself (the same cultivar) or one of the closely or distantly related cultivars. The pairs were planted in randomly selected plots at a density of 50 plants/m² by direct seeding in a field. Cultivar mixtures were conducted in a fully crossed design at a 1:1 ratio. All plots received fertiliser (N, P₂O₅ and K₂O at the rates of 7.5, 9.0 and 8.5 g/m², respectively) 1 day before the fields were saturated with water. All other field management followed the rules of the rural administration for the local rice industry. Rice growth was checked weekly before flowering. The number of days from sowing when the first flower appeared in the main panicle was recorded daily in 256 haphazardly selected plants for each focal cultivar and 128 plants for each neighbour cultivar during the flowering stage. Finally, grain yields were recorded at mature stage.

2.3 Pot-culture experiments

Four experiments were conducted in a greenhouse with 20–30°C night and day temperatures and 65%–90% relative humidity. Each experiment was conducted in a completely randomised design with five replicates for all treatments and control for a total of 140 pots. Seeds were surface-sterilised with 5% H₂O₂, placed in Petri dishes with moistened filter paper, and germinated in an environmental chamber at 28°C in the dark.

A cultivar mixture experiment was conducted to evaluate the performance of focal rice cultivars to neighbour relatedness in a series of 15 (diameter) × 20 cm (height) plastic pots containing 4 Kg of soil with base fertiliser (60 mg N·kg^-1, 30 mg P·kg^-1, 90 mg K·kg^-1) in each pot. Three germinated seeds of the focal cultivar were spaced uniformly in the centre of each pot while three seeds of the neighbour cultivar were sown in the surrounding area. Pots with monocultures in the same planting pattern served as controls. All pots were placed in the greenhouse, watered daily and their positions randomised weekly. Focal cultivars in 1/3 of pots were sampled at both the seedling and flowering stages, and their shoots and roots used for the quantification of (–)-loliolide, or for quantitative real-time PCR analysis of flowering-related genes as described below. In the remaining 1/3 of pots, flowering time and grain biomass were recorded.

An exudate incubation experiment investigated the impact of root exudates from the cultivars of varying relatedness on focal cultivar flowering time and grain yield. Root exudates of rice cultivars were collected hydroponically. Sterilised and germinated rice seeds were sown in nursery seedling plates and placed in a growth chamber at 22°C with a 14 h light and 10 h dark photoperiod. Seedlings of each cultivar at the 3-leaf stage were inserted into a styrofoam float, transplanted to a hydroponic container (12 cm diameter × 5 cm height) containing 500 ml distilled water and placed in a growth chamber. After 7 days, the solution was filtered with sterile filter paper (GE Healthcare Whatman) and yielded the root exudates for use in subsequent experiments: Six pre-germinated seeds of each focal cultivar were sown in a series of pots (15 cm diameter × 20 cm height) containing 4 kg soil as described above and placed in the greenhouse. The pots were divided into four relatedness groups. Each group was watered with 300 ml of the root exudates from the same cultivar, closely or distantly related cultivars once every 2 days. Pots treated with sterile distilled water served as controls. Finally, flowering time and grain yield were determined.

A weed competition experiment addressed endogenous (–)-loliolide induction of rice in response to common paddy weeds. Germinated seeds of each focal rice cultivar and two weeds (Echinochloa crus-galli and Leptochloa chinensis) were sown in a plastic pot (12 cm × 10 cm × 10 cm) containing 800 g of soil. Three germinated seeds of rice were placed uniformly in the centre of each pot, and three seeds of one weed were sown surrounding them. Pots with rice monocultures in the same planting pattern served as controls. The pots were placed in the greenhouse and water daily. Rice seedlings were sampled at the 3-leaf stage for the quantification of (–)-loliolide as described below.

A whole-growth-period-sampling experiment was to detect temporal changes in the level of (–)-loliolide in roots, leaves, and rhizosphere soils with varying growth stages. One germinated rice seed was sown into a plastic pot (15 cm diameter × 15 cm height) containing 2 kg of the soil described above. Pots were placed in the greenhouse, watered daily and their positions randomised weekly. Roots, shoots, and rhizosphere soil were sampled in three pots each at the 1-leaf, 3-leaf, 5-leaf, tillering, elongation, booting, heading, flowering, and mature stages. Rhizosphere soil was collected from the soil portion tightly adhering to roots by taking the plants from the soil and shaking off the loosely adhering soil (Guo et al., 2011). Plant and soil samples were taken for the quantification of (–)-loliolide as described below.

2.4 Rhizobox experiments

To examine root growth and placement patterns of focal cultivars in response to different neighbour cultivars, a series of window rhizoboxes were made of a 10 (width) × 200 (length) × 300 mm (height) polyvinyl chloride box containing 800 g of soil. Each focal cultivar was paired with the same cultivar, or closely and distantly related cultivars with five replicates. The rhizoboxes were divided into two groups. The first group was in the presence of neighbour cultivars. To remove edge effect on root growth, each rhizobox was vertically divided into four equal parts, and two germinated seeds of the focal cultivar were sown in the left while two neighbour cultivar seeds were sown in the right (Figure 2a). The second group were sown only with a focal cultivar in the centre of each rhizobox (Figure S2a) and were incubated with the root exudates of one of cultivars with varying genetic relatedness every 3 days. Incubation was carried out with two syringes 5 cm away on each side of the seedling at a depth of 2 cm. The left syringe injected root exudates while the right injected 1/10 Hoagland solution to control for the effect of nutrition solution in the root exudates.

All rhizoboxes were placed in the greenhouse. When the rice reached the 3-leaf stage, rhizoboxes were opened and the central root systems of focal cultivar were processed with WINRHIZO (Regent Instruments, Quebec, Canada) to determine horizontal asymmetry in root length. The roots were subsequently freeze-dried for biomass determination.

2.5 (–)-Loliolide incubation

Surface-sterilised rice seeds were germinated in a growth chamber at 28°C until their roots reached 1 cm in length. Three seedlings of focal cultivars were transferred into a 50 ml bottle with 1/2 Hoagland solution, each with three replicates. The bottles were placed in a growth chamber at 28°C with a 14 h-light/10 h-dark photoperiod. After 7 days, rice roots were incubated with (–)-loliolide at a concentration of 50 μM (Li et al., 2023b). Roots incubated without (–)-loliolide served as controls. Roots and shoots were sampled at different times after the start of incubation, and samples were used for the RNA-seq and quantitative real-time PCR analysis described below.

2.6 Quantitative analysis of (–)-loliolide

The quantification of (–)-loliolide was performed by liquid extraction/solid-phase extraction, followed by high-performance liquid chromatography (HPLC). Plant tissues and rhizosphere soils were each freeze-dried and ground with liquid nitrogen. 250 mg of the resulting powder was extracted with 10 ml of a MeCN (acetonitrile)-H₂O-HOAC mixture (90:9:1, v/v/v), and vortexed for 5 min at 25°C. After the solution was centrifuged at 2800 × g for 10 min, the supernatant was filtered with a 0.22 μm nylon syringe filter (Sterlitech, Kent, WA, USA). The filtrates of tissue and extracts of soil were evaporated to dryness individually under vacuum. Dry residues were dissolved in 50% aqueous methanol and loaded onto reversed phase C₁₈ Sep-Pak cartridges (Waters, Co., Milford, MA, USA), equilibrated with water, and eluted with MeOH. The MeOH fraction was concentrated with nitrogen gas to a final volume of 100 μl. The concentrated samples were subsequently subjected to HPLC analysis (Waters 1525) equipped with a C₁₈ reverse-phase column (Hypersil 100 mm×4.0 mm, 5 μm) and a diode array UV detector at 220 nm. Elution was performed with a mixture of 1% acetic acid and MeOH (70:30, v/v) at a constant flow rate of 1.0 ml min^-1 at 35°C. The peak of (–)-loliolide was identified by its retention time (9.8 min) and coelution with an authentic standard (Yuanye Biology Co., Shanghai, China). Working standard solutions ranging from 0.1 mg/L to 50 mg/L were prepared to establish a calibration curve. (–)-Loliolide was quantified by regression analysis of the peak areas against standard concentrations.

2.7 RNA-seq and transcriptome

Rice samples at 1 h from the (–)-loliolide incubation experiment described above were used for RNA-seq with three biological replicates. The extraction of total RNA used Trizol reagent (Invitrogen, CA, USA) following the manufacture's protocol. Library construction and sequencing were carried out by Lianchuan Biotechnology Corporation (Hangzhou, China) on an Illumina Hiseq. 4000 platform. The clean reads were mapped to the reference genome (Du et al., 2017).

2.8 Quantitative real-time PCR

The expression levels of genes involved in flowering [Heading date 3a (Hd3a), Rice flowering locus T 1 (RFT1), Early heading date 1 (Ehd1), Days To Heading on chromosome 8 (DTH8), GIGANTEA (GI), Grain number, plant height and heading date 7 (Ghd7)] in rice were analysed with quantitative real-time PCR. Rice shoots were sampled at the 3-leaf stage in the mixed-cultivar experiment, and at 0, 3, 6, 12 and 24 h in the (–)-loliolide incubation experiment described above. Total RNA was extracted from samples using a RNA extraction kit (Tiangen Biotech Co., Beijing, China), following the manufacturer's protocol. Agarose gel electrophoresis and spectrophotometry were used to determine the quality and concentration of RNA. First-strand cDNA synthesis was made with a cDNA synthesis kit (Tiangen Biotech Co., Beijing, China). qPCR was conducted with a Talent qPCR premix kit (Tiangen Biotech Co., Beijing, China) and StepOnePlus Real-Time PCR System (Thermo Fisher Scientific, Waltham, MA, USA). The total reaction mixture consisted of 10 μl 2×Talent qPCR PreMix, 0.6 μl forward and reverse primers, 2 μl 50×ROX Reference Dye, 2 μl diluted cDNA, and 4.8 μl RNase-free ddH₂O. Reaction conditions consisted of 95°C for 3 min followed by 40 cycles of 30 s at 95°C, 30 s at 52°C, and 30 s at 72°C. The relative level of expression was calculated using the comparative C_T method (Schmittgen & Livak, 2008). The UBQ was used as an internal control. Each sample was represented by three biological replicates, and each biological replicate by three technical replicates. Primer sequences are provided in Table S1.

2.9 Data analyses

Data were presented as means ± SE from at all replicates for each experiment and determination. The normality and homogeneity of variances were verified before statistical analysis. The effects of weed neighbours and developmental stages on the (–)-loliolide concentration in rice were analysed using one-way ANOVA. The effects of relatedness level and rice set on the flowering time, grain yield, (–)-loliolide concentration, root biomass and root length of focal cultivars were examined using two-way ANOVA. Tukey's honestly significant difference (HSD) test was conducted for multiple comparisons if ANOVA terms were significant. In addition, a simple Pearson's correlation was used to test whether the (–)-loliolide concentration across focal rice cultivars correlated with root biomass and flowering time. All data analyses were performed with SPSS v.25.0 programme (SPSS, Chicago, IL).

For gene expression analysis, the transcript abundance of each gene was normalised by the fragments per kilobase per million fragments mapped reads (FPKM) method. Differentially expressed genes (DEGs) were selected that had log₂ (fold change) >1 or <-1 with statistical significance (p < 0.05) using the edge package in R (R Foundation for Statistical Computing, Vienna, Austria). The identified DEGs were subjected to enrichment analysis of the Kyoto Encyclopaedia of Genes and Genomes (KEGG) pathway.

3.1 Genetic relatedness determines rice flowering and grain yield in cultivar mixtures

When focal cultivars (Huagan-3 and Lingyou-6173) were each paired with themselves (the same cultivar) or their closely and distantly related cultivars in paddy fields, flowering time and grain yield significantly varied with relatedness of paired cultivars. The presence of the same and closely related cultivars advanced flowering time 1–3 days (Figure 1a), and increased grain yields by at least 5% (Figure 1b) over target plants with more distantly related neighbours. The two focal cultivars responded similarly to relatedness in flowering time (interaction was nonsignificant) but the response was greater in Huagan-3 for yield (a significant cultivar × relatedness interaction) in a field setting (Figure 1, Table S2).

The greenhouse trial yielded consistent changes in flowering time and grain yield directly associated with relatedness as the field trial (Figure S1). Focal cultivars growing with closely related cultivars significantly earlier flowering and increased grain yield compared to those growing with distantly related cultivars (Table S2), indicating overall identical responses in the field and lab with the exception that yield responded similarly to relatedness across focal cultivars (a nonsignificant cultivar × relatedness interaction).

3.2 Root growth and distribution in response to genetic relatedness

When rice cultivars interacted, root growth and distribution varied significantly with relatedness (Figure 2a, Table S3). Focal cultivars growing with closely related cultivars had reduced root biomass (Figure 2b) and root length (Figure 2c), increasing both when grown with distantly related cultivars. In particular, focal cultivars’ root placement changed with neighbour relatedness, growing towards the roots of distantly related neighbours. However, focal cultivars tended to avoid the roots of themselves or closely related cultivars (Figure 2b, c). These root placement patterns were consistent across focal cultivars and should minimise root competition in closely related cultivar mixtures.

3.3 Root exudates mediate neighbour relatedness discrimination

The observed yield, root behaviour, and flowering time responses appear to be mediated by neighbour's root exudates and dependent on their relatedness. Similar to the physical presence of neighbours, flowering time and grain yield significantly varied with application of root exudates of different relatedness cultivars (Table S4). Application of root exudates from the same cultivars and closely related cultivars resulted in faster flowering (Figure 3a, b), greater grain yield (Figure 3c), and optimised root growth and placement patterns in focal cultivars (Figure 3d & Figure S2) than when they were exposed to root exudates of distantly related cultivars.

3.4 The role of (–)-loliolide in relatedness-mediated neighbour discrimination

Rice plants consistently produced more (–)-loliolide in response to the presence of unrelated neighbours regardless of whether they were intraspecific or interspecific competitors. As expected, the concentration of (–)-loliolide in focal cultivars significantly increased with genetic distance of neighbours (Figure 4a & Table S5). The most distant japonica cultivar led to the highest (–)-loliolide concentration, followed by the indica-distantly related cultivars in both rice sets. Focal cultivars grown by themselves or with closely related cultivars contained the lowest (–)-loliolide concentrations (Figure 4a). Interestingly, the signal transduction of relatedness by (–)-loliolide also extended to interspecific competition. The presence of the competing paddy weeds Echinochloa crus-galli and Leptochloa chinensis also induced (–)-loliolide production in rice (Figure 4b).

The activity of (–)-loliolide as a relatedness signal was greatest at early developmental stages when competition may be most important. The production and release of (–)-loliolide in rice significantly varied with developmental stage ((Figure 4c & Table S6). (–)-Loliolide was produced and released in early growth stages, dramatically declining as plants matured (Figure 4c). There were significantly positive relationships between (–)-loliolide concentrations and both root biomass and flowering time in rice cultivar mixtures (Figure S3), implying that (–)-loliolide stimulated root growth and delayed flowering.

3.5 Transcriptomic profiles of rice flowering and roots in response to (–)-loliolide

To address the regulatory networks and molecular mechanisms associated with flowering time and root behaviour, we performed a transcriptome analysis in rice exposed to (–)-loliolide. RNA-seq identified a total of 1780 (598 upregulated and 1182 downregulated) and 2323 (1649 upregulated and 674 downregulated) differential expressed genes (DEGs) in rice roots and shoots after incubation with (–)-loliolide (Figure S4a). Among these DEGs, 46 DEGs genes were related to root growth, of which 20 genes were upregulated and 26 genes were downregulated. KEGG analysis showed that most of the enriched genes were in the zeatin biosynthesis, RNA transport, and brassinosteroid biosynthesis pathways (Figure S4b). The enrichment of DEGs related to the zeatin biosynthesis pathway suggest that (–)-loliolide could directly promote root growth by upregulating zeatin biosynthesis.

Similarly, 33 genes were related to flowering (24 upregulated and nine downregulated). These were mainly enriched in circadian rhythm, MAPK signalling and proteasome pathways (Figure S4c). The 21 genes (17 upregulated and four downregulated) with significant changes in the circadian rhythm pathway included: CONSTANS-LIKE (OsR498G0306913700; OsR498G0815539300), GHD7 (OsR498G0713935400; OsR498G1019192800), PRR73 (OsR498G0305710000), PRR95 (OsR498G0917850800), TCP7 (OsR498G0204385100), RUP2 (OsR498G0202844500; OsR498G0261422600), and FT (OsR498G0917792400) (Figure S4c).

To clarify the effect of (-)-loliolide on rice flowering, key genes in the circadian rhythm pathway were screened by transcriptome analysis and mapped onto a signalling pathway. Under long-day conditions, the negative flowering regulators OsCOL16, Ghd7, TCP7, OsPPR73 and OsPPR95 were significantly upregulated in leaf, inhibiting the expression of RFT1/Hd3a which codes florigen. Moreover, the upregulation of RUP2 also repressed photomorphogenesis (Figure 5). This analysis indicates that (–)-loliolide exposure may directly inhibit rice flowering by regulating the expression of flowering-related genes.

3.6 (–)-Loliolide regulates rice flowering gene expression

Similar to the results of the cultivar mixture and transcriptome analysis, the presence of distantly related rice cultivars delayed flowering. Compared to plants grown with distantly related cultivars, the expression levels of positive flowering regulators Hd3a, RFT1 and Ehd1 were increased in focal cultivars grown with closely related cultivars. However, the expression of the negative flowering regulators DTH8, GI and Ghd7 were reduced in focal cultivars grown with closely related cultivars (Figure 6a).

In incubation experiments, (–)-loliolide upregulated or downregulated the expression of key flowering genes based on their functional roles and focal cultivar identity over a time-course (Figure 6b & Table S7). In general, (–)-loliolide inhibited the expression of positive regulators and promoted the expression of negative regulators related to flowering in Huagan-3 after 6 h incubation (Figure 6b). (–)-Loliolide also inhibited the expression of positive regulators and promoted the expression of negative regulators related to flowering in Lingyou-6173 at 12 h (Figure 6b). Therefore, (–)-loliolide may delay the flowering time of rice by modulating the expression of flowering-related genes in response to neighbour relatedness.