Sustained productivity and agronomic potential of perennial rice

The study addresses the need to develop sustainable, high-yielding grain crops that reduce inputs and environmental degradation relative to annual systems. Annual grains dominate global croplands but require intensive tillage and inputs, leading to soil erosion, nutrient loss, and ecosystem impacts. Perennial crops can provide continuous soil cover, deeper root systems, improved nutrient retention, and reduced input requirements. The research question is whether perennial rice, bred from annual Oryza sativa and perennial Oryza longistaminata, can deliver sustained grain yields comparable to annual rice while improving livelihoods and soil health. The paper reports breeding progress, multi-year performance across locations and regrowth cycles, socioeconomic outcomes, soil effects, suitable ecological zones, and prospects and limitations for adoption and further development.

Background literature highlights the environmental costs of annual grain systems and the potential of perennial grains to enhance ecosystem services through longer photosynthetic seasons, deep roots, and mixed grain-forage systems. Two main breeding pathways for perennial grains are de novo domestication (e.g., Kernza from Thinopyrum intermedium) and interspecific hybridization. Previous work demonstrated progress in perennial grain breeding and the adaptive diversity of rice ecosystems (irrigated, rainfed lowland, flood-prone, upland). The authors position perennial rice within this broader effort, citing studies on perennial grain benefits (reduced nitrate leaching, increased soil carbon), rice ratooning limits in annual cultivars, and climate constraints on rice fertility at high temperatures. This literature frames the rationale for interspecific hybridization to create perennial rice suitable for irrigated systems, with potential broader adaptation.

Breeding: Perennial rice was developed via interspecific hybridization between Oryza sativa ssp. indica RD23 (female) and O. longistaminata (male), using embryo rescue (1996). The F1 displayed strong rhizomes and partial fertility. Extensive pedigree selection targeted short rhizomes and high pollen fertility while retaining domestication traits. From 7,200 F2 individuals, an exceptional F2 (36-1) was selected (pollen fertility >85%, seed set >60%). Successive selfing from F1 to F7 reduced rhizome number/length while improving agronomic traits toward cultivated rice. Next-generation sequencing showed 16.16% of the PR23 genome derived from O. longistaminata. Multi-year, multi-location evaluations (2012–2017) identified PR23 as a stable, high-yield line (broad-sense heritability for yield 0.87–0.94; regrowth 0.88–0.96). Additional cultivars PR24, PR25, PR101 were bred; a second breeding round used backcrossing of perennial donor PRB3 (MP3-235) to elite indica DR449 to produce PR107 (released as Yunda107/NARORICE-1). Selection emphasized perenniality, short rhizomes, high pollen fertility, and agronomic performance. Field experiments: Five experiments in China and Laos assessed performance, adaptation, socioeconomics, soil effects, and temperature constraints.

• Experiment 1 (2012–2013): 22 genotypes across 12 environment combinations (Yunnan, China; Vientiane, Laos); randomized complete block design; traits included regrowth, yield, components.
• Experiment 2 (2014–2016): 5 PR genotypes + 3 annual controls across 19 environments in Yunnan; RCBD; measured yield and components; N split dressing at boot stage.
• Experiment 3 (2016–2020): Large farm-scale plots (1–13 ha) at Mengzhe, Menglian, Xinping comparing PR23 to local elite annual rice (AR). PR: initial transplanting; after harvest, stubble cut to 10 cm; no tillage between regrowth cycles. AR: replanted each season with ploughing and transplanting. Recorded yields, regrowth rates, agronomic traits; detailed labour and non-labour costs collected for socioeconomic analysis.
• Experiment 4 (2018–2020): PR25 and PR107 performance across multiple locations with PR23 as control; measured yields and agronomic traits.
• Experiment 5: Low-temperature tolerance for regrowth. Sub-experiment A: field exposure at 16 locations; regrowth vs January mean temperature fitted by quadratic model. Sub-experiment B: growth chamber exposures to 0 °C and 4 °C for 3–7 days; recorded regrowth. Socioeconomic measurements: Tracked labour categories (seedling nurseries, ploughing, transplanting, crop management, harvesting) and non-labour inputs (seed, pesticides, herbicides, fertilizers). Calculated outputs using market prices and net economic gain (output minus input). Sensitivity analysis assessed effects of yield and input changes on net gain. Soil sampling and analysis: Annual post-harvest sampling (0–40 cm, four 10-cm increments) under PR and AR. Measured pH, bulk density, porosity (capillary/non-capillary), water retention (saturation upper limit, wilting lower limit, plant-available water capacity), soil organic carbon (SOC via dichromate oxidation), total nitrogen (TN via Kjeldahl). Calculated SOC and TN stocks using bulk density across depths. Statistical analysis: Genotype-by-environment interactions via cluster and ordination (CropStat 7.2); variance components and heritability estimates. For yields and soil parameters, ANOVA with LSD (P ≤ 0.05). Figures generated with Origin; data analyzed in IBM SPSS v20. Risk assessment and management: Evaluated threats from pests, diseases, weeds, and climatic extremes; discussed mitigation (integrated pest management, water/fertilizer management, occasional tillage and herbicide to reset stands, rotations).

• Sustained yields from single planting: PR23 produced grain for eight consecutive seasons (four years) with average 6.8 Mg ha−1 per harvest, comparable to replanted annual rice (AR) at 6.7 Mg ha−1. In Mengzhe across five years, PR23 first-season yields were 10.9, 8.7, 8.2, 8.1, 5.3 Mg ha−1 and second-season yields 6.6, 6.4, 6.2, 6.0, 3.1 Mg ha−1; yields declined in the ninth season, suggesting resowing after ~4 years.
• Performance across sites: PR23 matched or exceeded local elite AR in Mengzhe and Menglian over eight seasons; in Xinping, PR23 recovered after pest damage to yield averages comparable to AR for several seasons, then stabilized at slightly lower but stable yields. PR25 and PR107 achieved high, stable yields for at least two years (four seasons) from a single planting, comparable to PR23.
• Regrowth and agronomy: High regrowth rates (>75%) maintained over eight seasons with stable agronomic traits; decline in regrowth and yield occurred in year 5 (seasons 9–10). Broad-sense heritability for yield 0.87–0.94 and regrowth 0.88–0.96 indicates stability of key traits.
• Genomic composition: PR23 retained 16.16% of its genome from O. longistaminata, conferring perenniality while agronomic traits converged toward domesticated O. sativa.
• Socioeconomic benefits: In regrowth seasons, PR saved 58.1% of labour and 49.2% of input costs compared with AR, eliminating need for nurseries, ploughing, and transplanting. Monetary savings were US$1,177–1,401 ha−1 per regrowth season (46.8–51% of AR costs). Labour days saved were ~68–77 d ha−1 per regrowth. Net economic gain was consistently higher for PR, even where yields were lower than AR (e.g., Xinping). Sensitivity analysis showed net gain was ~5× more responsive to yield increases in PR regrowth cycles than in re-transplanted AR (US$2,308 vs US$436 per additional 1 t ha−1).
• Soil improvements: Over four years (0–40 cm), SOC and TN accumulated at 0.95 and 0.11 Mg ha−1 yr−1, respectively. Soil pH increased by 0.3–0.4 units into optimal ranges for nutrient uptake and microbial activity. Plant-available water capacity increased (reported 7.2 mm), with associated improvements in porosity and water retention; C/N ratio declined in topsoil, implying enhanced N cycling with concurrent N immobilization due to SOC accrual.
• Environmental implications: Reduced tillage likely lowers methane emissions relative to annually puddled systems; intermittent flooding could further reduce methane but may increase nitrous oxide pulses, affecting net global warming potential.
• Adoption and scale: By 2020 PR23 covered >8,400 ha; in 2020, 3,818 ha across >11,000 farms; by 2021, 15,533 ha with 44,752 smallholder farmers in southern China. Farmers strongly preferred PR cultivars.
• Ecological zoning: Suitable zones for regrowth identified largely within 40° N to 40° S. Regrowth probability >75% associated with mean coldest-month temperature ≥13.5 °C; exposure to 0 °C for ≥7 days reduced regrowth to ~22%; 4 °C for 7 days reduced to ~55%. High temperatures (~31 °C) around flowering pose sterility risks similar to AR. Short rhizomes were selected to limit weediness while aiding survival.

The results demonstrate that perennial rice, developed via interspecific hybridization, can achieve yields comparable to elite annual cultivars over multiple regrowth cycles from a single planting, while markedly reducing labour and input costs. These sustained yields, coupled with improved soil properties (SOC, TN, pH, water-holding capacity) and potential reductions in methane emissions from reduced tillage, directly address the research aim of enhancing productivity and livelihoods with fewer resources and less environmental impact. High trait heritability and stable regrowth indicate genetic and agronomic robustness, though resowing after approximately four years is advisable. The substantial adoption in southern China underscores practicality and farmer preference. Ecological zoning suggests broad potential in frost-free regions within 40° N–S, with low-temperature duration thresholds and high-temperature fertility constraints guiding management. Perennial rice thus presents a viable step-change for irrigated and favourable lowland systems and a platform for expansion into rainfed and upland environments with attention to pest, disease, and weed management and greenhouse gas trade-offs.

Perennial rice cultivars PR23, PR25, and PR107 deliver sustained grain production over multiple cycles from a single planting, maintain yields comparable to annual rice, reduce labour and input costs, and improve soil quality. The work establishes a commercially viable perennial grain through interspecific hybridization, with significant farmer adoption and a broad potential ecological range. Future research should refine weed-management agronomy in regrowth cycles, monitor and manage pest and disease dynamics (including virus risks), optimize water and nitrogen management to balance methane and nitrous oxide emissions, and continue breeding for broader adaptation, especially rainfed lowland and upland systems. Genomic tools (marker-assisted selection, genomic selection, gene editing) can accelerate perennial trait introgression and quality improvements, extending perennialization to other major crops.

• Regrowth longevity: Yields declined notably by the ninth–tenth seasons; current cultivars typically require resowing after ~4 years.
• Weed control: Regrowth cycles required 1–2 additional herbicide applications compared with transplanted annual rice.
• Pests and diseases: Risks include pests favored by high nitrogen (e.g., hoppers), diseases with high pustule loads, and insect-transmitted viruses (e.g., tungro). While not observed as major issues in these trials, they remain potential threats; PR107 carries resistance to rice yellow mottle virus via O. longistaminata.
• Climatic constraints: Low temperatures limit regrowth (sustained temperatures ≤4 °C for >5 days markedly reduce regrowth), and high temperatures (~31 °C) around flowering can cause spikelet sterility; thus, suitable zones are largely frost-free regions within 40° N–S.
• Greenhouse gas trade-offs: While reduced tillage may lower methane emissions, intermittent flooding could increase nitrous oxide pulses, affecting net global warming potential.
• Generalizability: Most performance data are from irrigated/favourable lowland environments; broader testing and breeding for rainfed lowland and upland conditions are ongoing.
• Stand termination: Under some conditions, intentional termination (tillage and/or herbicide) may be needed to reset stands before re-sowing or rotation.