De novo design of future rapeseed crops:Challenges and opportunities

Shengyi Liu *,Hrsh Rmn ,Yng Xing ,Chunji Zho ,Junyn Hung ,Yunyun Zhng

a Key Laboratory of Biology and Genetic Improvement of Oil Crops,the Ministry of Agriculture and Rural Affairs,Oil Crops Research Institute,Chinese Academy of Agricultural Sciences,Wuhan 430062,Hubei,China

b NSW Department of Primary Industries,Wagga Wagga Agricultural Institute,Wagga Wagga,NSW 2650,Australia

c Guizhou Rapeseed Institute,Guizhou Academy of Agricultural Sciences,Guiyang 550008,Guizhou,China

Keywords:Oilseed rape Seed yield Stress tolerance Physiological basis Ideotype Gene editing Breeding by genome design

ABSTRACT To address the global demand for rapeseed while considering farmers’ profit,we face the challenges of making a quantum leap in seed yield and,at the same time,reducing yield loss due to biotic and abiotic stresses.We also face the challenge of efficiently applying new transformative biotechnology tools such as gene editing and breeding by genome design to increase rapeseed productivity and profitability.In this Perspective,we review advances in research on the physiological and genetic bases of both stress factorsaffected yield stability and seed yield potential,focusing on source–sink relationships and allocation of photosynthetic assimilates to vegetative growth and seed development.We propose research directions and highlight the role of plant architecture in the relative contributions of the root system,leaves,and pods to seed yield.We call for de novo design of new rapeseed crops.We review trait variation in existing germplasm and biotechnologies available for crop design.Finally,we discuss opportunities to apply fundamental knowledge and key germplasm to rapeseed production and propose an ideotype for de novo design of future rapeseed cultivars.

1.Introduction

Brassica oilseed crops,comprising oilseed rape(also called rapeseed or canola,Brassica napus L.),turnip rape (B.rapa L.),Indian mustard (B.juncea L.) and Ethiopian mustard (B.carinata L.),have been cultivated for thousands of years.Rapeseed is grown on more than 36 Mha with an annual seed yield of 72 million tons and accounts for 13% of global oil production (FAOSTAT,https://www.fao.org/faostat/).It is now the third-largest oilseed crop after oil palm and soybean.Rapeseed is grown mainly in temperate regions of Australia,Canada,China,and Europe,whereas turnip rape and Indian mustard are cultivated in warmer climates,especially in India and China.Brassica crops play versatile roles in the circular economy and provide sources of healthy edible oil for human consumption,biodiesel,and industrial oil.After oil extraction from seeds,the byproduct(meal)is used as a nutritionally rich source of protein,mainly for stock feed.

Global oilseed production and acreage trends show strong growth in the rapeseed industry since 1961.However,almost no increase in seed yield has occurred,at least in the past five years(Fig.1).The daily energy demand of the growing human population,expected to reach more than 9 billion by 2050,requires large areas for rapeseed production and increased crop productivity.Accelerating rates of urbanization and shrinkage of resources(land,water,and human resources available for agriculture)pose further challenges.Rapeseed cultivation also relies on global prices,as farmers make decisions based on profit margin.Although the recent increase in the global price of rapeseed (＞US $800 per ton in 2021)has motivated some farmers to continue rapeseed cultivation;it must compete with other staple crops such as wheat,rice,and maize.Therefore,there is an urgent need to boost rapeseed productivity and profitability by innovative genomic,biotechnological and agronomical interventions.

Although progress has been made in conventional genetic improvement and technologies of rapeseed cultivation,the challenges of alleviating yield stability and increasing productivity remain.In alignment with and support from the nine research articles published in this issue with a special focus on ‘‘Genomic and Gene-editing Technologies to Boost Brassica Oilseed Productivity”,in this Perspective we discuss broader issues beyond current rapeseed plant architecture with a focus on seed yield-related physiology and genetics,and offer forward-looking prospects for enhancing crop performance including yield stability and yield potential.Leveraging available information on germplasm,trait variation,QTL,causal genes and biotechnologies,we propose a de novo design to allow future rapeseed crops to meet global demands.

Fig.1.Average seed yield of the major rapeseed producing countries in comparison to global production during the past 60 years.Data Source:https://www.fao.org/faostat/.

2.Enhancing resistance to stresses for sustainable rapeseed production

A sustainable and secure rapeseed industry requires a reduction in yield losses due to environmental/abiotic(salinity,acidity,alkalinity,drought,heat and water-logging)and biotic(insect pest,disease and weeds) stress factors while protecting the environment.Fluctuations in global rapeseed production across years (Fig.1)are likely driven by various stress factors,particularly drought and heat episodes in Australia,North America,Europe,and China.These stress factors account for large losses;for example,drought and heat stress account for the US $200 billion per annum[1].We must tackle these stress factors to ensure yield stability in rapeseed production.

Conventional and biotechnology-assisted breeding methods have led to increases in rapeseed yield by producing cultivars for increased resistance to diseases,in particular blackleg (casual organism:Leptosphaeria maculans and L.biglobosa),clubroot (casual organism:Plasmodiophora brassicae),and Sclerotinia stem rot(SSR;casual organism:Sclerotinia sclerotiorum).Several sources of qualitative and quantitative resistance to diseases and matching genomic regions have been identified in rapeseed populations [2–15] (Table S1).Five race-specific resistance (R) genes for blackleg resistance have been cloned [6–9].Both qualitative and quantitative trait loci (QTL) for clubroot resistance have been reported in Brassica germplasm [10–12].Major QTL were identified for SSR resistance [13–15] and an RNAi strategy was tested for control of SSR [16].Several R and quantitative resistance loci have been deployed in commercial cultivars to reduce yield losses.However,some R genes deployed in commercial cultivars have already become ineffective for conferring resistance,owing to evolution and change in the pathogen virulence patterns [17,18].Climatic stress factors such as high temperatures and weather patterns threaten sustainable rapeseed production worldwide.Resistance to blackleg caused by L.maculans is compromised with increased temperature [19,20].There is a need to identify novel loci for broad-spectrum and durable disease resistance that could be deployed in new cultivars.

Soil constraints such as acidity,salinity,compaction and sodicity impair rapeseed production.These problem soils are distributed worldwide and are rapidly increasing with mechanization and intensive agriculture.Soils are being increasingly acidified due to nitrogenous fertilizers,nitrate leaching,and rising CO2levels in the atmosphere.QTL and genome-wide association studies(GWAS) have identified genomic regions for tolerance to manganese (Mn2+) and aluminum (Al3+) toxicities,which occur in acidic soil [21],and salinity [22] in rapeseed.The studies showed the possibility of increasing tolerance by genetic approaches,and more investigations are needed to reduce soil constraints limiting genetic yield potential.

Environmental constraints such as drought,heat,frost,and waterlogging also restrict rapeseed yield potential.The extent of losses varies with the time of sowing and the severity and timing of stress.There is difficulty in elucidating the genetic basis of tolerance to environmental constraints in rapeseed,owing to a lack of robust phenotyping methods,significant G × E interactions,and the quantitative nature of trait variation.However,research is in progress to elucidate the genetic,physiological and molecular bases of stress tolerance in rapeseed.Genetic loci associated with drought avoidance,such as carbon isotopic ratio(δ13C)/carbon isotope discrimination (Δ13C) and root architectural traits,were recently identified in rapeseed populations [23–26].Some of the Δ13C genomic loci were also mapped for plant fitness traits such as flowering time and plant height.But these fitness traits often show trade-offs with plant biomass and yield,particularly in water-limited environments.Breeding programs need to select suitable combinations of alleles for increased yield in target environments.

In summary,stress factors severely affect yield stability and restrict yield potential.Some genetic research progress has been made toward tackling the problems,but more is needed.For breeding tolerance or resistance to multiple stress factors,it is challenging to combine the identified loci because they are distributed in different germplasm accessions.Genomic selection or more comprehensively breeding by genome design (see more description below) is a promising tool for combining loci or genomic regions that contain target genes and can be integrated with gene editing if the function of genes is known.

3.Prioritizing traits for de novo design of new rapeseed plant architecture

We can see from Fig.1 not only that seed yield fluctuates among years,probably owing to the stress factors discussed above,but also that yield improvement seems to have reached a plateau in recent years.In high-yielding environments,for example,in Germany and France,seed yield appears to have remained at the same level since 2004,nor has there been a pronounced increase in the past five years in other major rapeseed producing countries such as Australia,China and Canada.As a result,the world average rapeseed production has remained stagnant.The seed yield trend in the developed countries suggests that the potential for further increase in seed yield will be limited in developing countries (Table S2).

If high seed-yield records in a large farm field (the achievable highest yield from a high-yielding cultivar under near-ideal growing conditions such as optimum irrigation,nutrition,and protection from all stress factors) represent theoretical yield potential(YP) of the current plant type or architecture,the ratio of average farm yield (FY) in a large region to YP reflects the extent of YP achieved in farm production.The highest rapeseed yield record of 6.1 t ha1was achieved in a 1.27-ha field in China,whereas the highest yield of 6.7 t ha1in an 8.2-ha area was recorded in the UK (Table S2).The FY/YP ratio for rapeseed is 50.7% in the UK,higher than those of all other staple crops reviewed;in China,this ratio is 32.8%,ranked in the middle among crops (Table S2).These statistics suggest that average farm production in rapeseed has reached a high level relative to that of other crops.The high FY/YP ratio may be determined by the low harvest index (HI) of rapeseed [27,28],suggesting that current plant architecture and population structure in rapeseed are at a low yield potential level and have been a factor limiting a further increase in seed yield.The route to lifting crop productivity and profitability must rely upon the de novo design of new plant architecture optimized for achieving higher yield and making rapeseed cultivation more convenient and lucrative than other main winter season crops.

To identify a target ideotype of plant architecture for the de novo design to achieve ideal population structure and thereby elevate seed yield level,we must examine the relationships among plant architecture,HI,physiology,vegetative growth,reproductive development,and seed yield of oilseed rape.

3.1.Relationship between vegetative growth and seed development

The studies[29–33]have shown that plant biomass at or before the flowering stage is significantly correlated with seed yield;the number of leaves is highly correlated with the number of branches.The plant-vigor measures,such as normalized difference vegetation index(NDVI)and leaf area index(LAI),also showed a positive correlation with seed yield[23,31,32].Current rapeseed cultivation practice in the Yangtze River Valley indicates [29,30] that plants with approximately 12 leaves in conventionally cultivated soil or around nine leaves in direct-drill cultivation at a high plant density(about 4.5 × 105plants ha1) before the winter solstice (2–3 days before Christmas Eve)are crucial to achieving high seed yield.Bud differentiation is generally initiated 20 days before the winter solstice and bolting occurs in spring.The increase in leaf number due to genetic and agronomic interventions dramatically elevated seed yield in China,with a 2–5-fold increase from 1977 to 1987 and an average seed yield of up to 3.7 t ha1in some counties [30,33].However,there were no substantial changes in plant architecture.It is reasonable to infer that for a given genotype (the current cultivars),there were few changes in HI among plant densities or farm fields,and accordingly,higher seed yield requires larger plants with more leaves to support the development of more branches that ensure a thick canopy of pods.

A series of detailed physiological experiments[34–41]indicated that photosynthates from pods (mainly pod walls) contribute approximately 70% of seed yield.The other green parts account for 30%of seed yield,comprising 10%from leaves and 20%from the main stem and branches (of which about 5% is from the main inflorescence and about 9% is already stored in the main stem and branches and formed before seed development).Mainly leaf photosynthates contribute to the early development of young pods with embryos [34].Inanaga et al.[38,39] compared LAI with pod area index (PAI) and found that PAI at the middle stage of pod development was much higher than the highest LAI formed at the beginning of flowering;net production(gross production–respiration,g CO2m2d1) increased with PAI and solar radiation.These findings indicated that the photosynthesis of pods,rather than leaves,is the determinant of seed dry matter.Physiological and molecular experiments[42,43]further supported the idea that pod wall photosynthesis is a key determinant,contributing to the majority (＞80%) of all lipid accumulation in the seeds.

The leaf number per plant decreased at a high plant density,but population LAI and PAI increased [44].At the pod development stage,there was a strong correlation between radiation use efficiency (RUE) and seed yield [45].In a comparison of high plant density (e.g.,4.5 × 105plants ha1) with low density (e.g.,1.5 × 105plants ha1,as standard practice in China),population biomass and seed yield showed a more significant increase in a compact plant type (genotype 1301 with smaller leaf and branch angle)than in a normal genotype,HZ62[46].There was an increase in Rubisco and sucrose phosphate synthase activities and the carbohydrate contents of leaves of the compact genotype 1301 compared with HZ62.The plants of the compact type were shorter(164 vs.185 cm) and had fewer branches per plant (7.8 vs.8.7),but produced more photosynthates and biomass.Rapeseed is a magical crop (Fig.S1);plants withered due to freezing injury (depending on the degree of cold temperature in a year in some regions of Shaanxi province,China)or those lost leaves due to grazing recover well and produce expected seed yield.These findings again supported the importance of PAI to seed yield,but whether the number of leaves or LAI is as important as suggested above to seed yield awaits more comprehensive investigations,particularly in rapeseed cultivars with different plant architectures.

3.2.Relationship between vegetative growth,seed yield traits,and HI

HI,which refers to resource allocation of reproductive organs,measured by the ratio of seed yield to total above-ground plant biomass,is a critical trait for plant fitness in the agricultural context and determines the economic return on farmer investment.HI differs widely among crops:tuber crops have higher HI (more than 0.6 for cassava,sugarbeet and sweet potato)than cereal crops(0.4–0.5 for rice,wheat,maize)and soybean (0.4–0.5).Still,the HI of rapeseed cultivars is around 0.24[27,28]although recent studies have shown that HI could be increased up to 0.44 in some genotypes [47].Since the first Green Revolution,grain yield increase in rice and wheat came mainly from increased HI [48].High broad-sense heritability of up to 75% are reported for HI in rapeseed [49].QTL and GWAS analyses identified several genomic regions controlling HI,each accounting for up to 11% of the total phenotypic variance [47,49].

However,the relationships between HI and traits related to vegetative growth and seed yield are complex [47,49–52]:The relationship between HI and seed yield,and seed yield-related traits may vary depending upon environment or plant architecture;Generally,HI is higher in short-statured genotypes than in tall ones;High HI does not consistently correspond to high seed yield;a breakthrough in achieving high seed yield requires high HI.The results suggest that only under the condition of relatively high varietal biomass,the comparison of HI between varieties will make sense and improvement of HI is of significance.As pods (mainly pod walls)are the primary source of photosynthates for seed yield and oil accumulation [34–40,42,43],allocation of leaf assimilates into seeds via flux is not as essential as expected,as long as there is enough carbon source and nutrients for the initial development of embryos and young pods.In this context,de novo design for enhancing HI may aim to create new plant architectures to optimize vegetative growth toward higher seed yield.However,what is the rational route remains to be established.

3.3.Relationship between the root system and seed yield

The plant root system functions in absorbing nutrients from soil and synthesizing some hormones for plant growth and development,supporting plants for resistance to lodging and environmental stresses,and improving soil’s ecological system.Although phenotypic characterization for root architectural traits is complex and labor-intensive [31,53,54],some research progress has been made.In old and recent winter and semi-winter rapeseed varieties,root biomass generally increases rapidly until the bolting stage,and subsequently,leaf biomass peaks at or after flowering [39].Root volume and shoot biomass are positively correlated with seed yield [53].There is wide variation in root architectural traits and biomass with a heritability range of 0.4–0.8,and several genes with minor allelic effects have been identified [25,55,56].GWAS analyses have detected many QTL in different rapeseed germplasm panels.For example,16 QTL regions for root development,32 QTL for stage-specific root growth measures,and 40 QTL for mature root traits(primary root length,diameter of root,fresh root weight,root dry weight,total root length,total root surface area,total root volume) were identified in rapeseed populations [25,55–57].However,previous studies and cultivation practice did not reveal the role of root architecture in fast vegetative growth and flower development in spring.The relationships between above-ground plant architecture,root architecture,and the timing of the establishment of maximal root volume should be investigated to uncover the role of these root traits in the vegetative and reproductive growth of rapeseed.

Given that rapeseed leaves die in harsh winters and still produce expected seed yield,it may be reasonable to hypothesize that some of leaf assimilates are transported into roots for storage before leaves die.However,detailed information on the proportion of the leaf assimilates transported to and stored in roots for nourishing plants in favorable springtime needs to be determined.In considering the limited role of leaves in seed yield,root system growth may be a primary consumer (sink) of leaf photosynthetic assimilates.In taking account of the time course of all sources and sinks (root-leaf-stem-pod),we drew attention to the spatial–temporal relationship of the source to sink.Therefore,for the de novo design of new plant types,identifying or creating a large and early established root system is essential,especially for a variety with a shortened growth period.

3.4.Matching plant architecture with a trade-off between earliness and seed yield

Under the given ecological condition of a region,rapeseed with a shorter growth period and early maturity produces lower seed yield.But earliness or shorter growth period are essential for crop rotation in many rapeseed regions,particularly in China’s winter rapeseed regions.In some other areas,early flowering and early maturity are preferred traits to avoid damage from environmental vagaries such as frost,high temperature,and drought events,especially during flowering and seed development.Based on the relationships between vegetative growth and seed yield or HI,excessive leaf number and vegetative growth could be reduced without compromising seed yield,and thereby the growth period could be shortened by rationalizing plant architectures.For a same variety,high-density planting allows later sowing than the usual,also resulting in a shorter growth period.However,other problems arise.Early flowering and resistance to SSR,a devastating disease leading to yield loss in oilseed rape,are negatively correlated[15,58–60].Studies have shown that early flowering was genetically associated or tightly linked with disease susceptibility,possibly owing to genetic pleiotropy [15,58,61].High-density planting required for high seed yield in current varieties also causes shade syndrome and makes plants hungry for sunlight,which is needed for carbon assimilation.This practice also makes rapeseed more prone to SSR infection and lodging,mediated through Della-JAZphyB protein interactions.These issues may be resolved by breeding compact plant types with upward-oriented (narrow angle)leaves,branches and pods.

3.5.Genetic advance for seed yield-related traits

The above sections considered some of the genetic aspects of physiology-related seed yield traits.This section highlights the advances not addressed in seed yield traits.Many traits involved in adaptation,plant growth and development,and agronomic performance[62,63]influence seed yield in rapeseed(Table S1).Traits include number of pods per plant(per unit area),number of seeds per pod,thousand-seed weight,plant height,first branch height,inflorescence length,pod length,shoot and root biomass,leaf and branch angle,and seedling vigor.

QTL or association mapping has been conducted to understand the genetic basis of the trait variation (Table S1).Several genes such as BnARF18 and BnGRF2that affect seed weight,pod length,and oil production in rapeseed have been cloned [64,65].In this issue,Hussain et al.[66]identify additional key genes and mechanisms underlying the natural variation of pod length in rapeseed germplasm.Generally,larger pods have more and bolder seeds.However,seeds may be more spaced(distanced)in long pods than shorter pods.Larger pods are likely to be more efficient at sunlight interception due to more chloroplasts located in the outer pod wall layers and may contribute to increased yield.Oil content is directly related to seed yield,and high oil content has long been a major breeding objective.Several genes involved in oil content have been identified in rapeseed and related species [67–70].In this issue,Yan et al.[70] identify four major QTL for oil content in rapeseed by integrating genome resequencing and transcriptomics approaches.These studies suggest that several pod traits contribute to seed oil content.In addition to its own regulatory system,seed oil content may be determined mainly by the abundance of photosynthates transported to seeds,and therefore pod characteristics are essential.

Pod strength,a measure of shatter resistance,is another priority trait to reduce preharvest yield losses that have been targeted in rapeseed breeding programs for several decades.However,due to this trait’s multigenic nature,compounded with multiple copies of pod dehiscence genes [71,72],genetic improvement for pod shatter resistance has been slow.Several QTL and candidate genes associated with pod shatter resistance have been identified using genetic mapping populations and GWAS panels of B.napus,B.rapa,and B.carinata [73–75].Mutagenesis and gene-editing have also been applied to developing pod-shatter resistance germplasm in rapeseed [76,77].Bayer Crop Science has developed pod shatterresistant varieties(such as Pod Guard)and made them available to farmers for commercial cultivation in several countries.In this issue,Chu et al.[78] report that a single recessive gene controls a lignified-layer bridge and is associated with high pod-shatter resistance in rapeseed.

To develop de novo design for future rapeseed crops,further research is required on the following priority traits to determine their contribution to the final seed yield:pod angle,pod size,pod thickness,pod density on the primary inflorescence,pod photosynthetic efficiency,PAI,node number,internode length,early root system establishment,and root architecture traits (root length,root mass,and root volume).

4.Opportunity and challenge for de novo design of ideal architecture

4.1.Proposed high yield plant architecture for de novo design

Based on the present knowledge (some reviewed above),three key conclusions can be drawn:(1) current farm rapeseed yield is high and has reached a plateau,and theoretical yield or yield potential has to have a breakthrough to create space for farm yield increase;(2) in terms of the spatial–temporal relationship of the source to sink,analyses of the interrelationships of the root system,vegetative growth,photosynthates and allocation of green tissues,reproductive development,HI,and seed yield have revealed that the plant architectures of current varieties are a factor limiting both farm and theoretical yield breakthroughs;and(3)considering the spatial–temporal relationship of the source to sink,the principle for an ideotype design might have to consider the relative roles of the root system,LAI at various growth stages,and PAI in vegetative and reproductive growth in field populations.Some ideotype characters have been described in previous studies [79–82].

Taking the above knowledge together with existing phenotypic variation in morphological and seed yield-related traits (some described in Fig,3),we propose a de novo design of future rapeseed presented in Fig.2 for making a breakthrough in genetic improvement for seed yield.This ideal-plant architecture option may be particularly relevant to Chinese semi-winter rapeseed breeding programs.The key characteristics of plants and their field population at the pod development stage are summarized in the left column of Fig.2(three alternative plant types shown differ only in the presence or absence of branches).The brief description of the design is as follows:(1)short plants(about 1.5 m)in a field population with very compact morphology,seedlings with high vigor(fast growth rate),the optimum number of rosette leaves before bolting and the optimum number of stem leaves for targeting short plants after bolting (the total leaf number may not be so many as those of current varieties),and early established stout and extensive root system.Upward-oriented and/or incised leaves are better,although the influence on seed yield remains to be investigated under the optimized/reduced number of leaves.(2) The novel canopy of pods:long primary inflorescence with dense,long (e.g.,10 cm including beak) and upward-oriented pods,no or 1 to 2 upward-oriented branches,and thick and firm stem to support heavy pods.It must be pointed out that this plant design is entirely conceptual;not all the characteristics(e.g.,combining the attribute of dense and upward-oriented pods with an extensive root system)have been brought together into one design.

There are a few advantages over the current plant architecture for this conceptual rapeseed crop.The most important advantage is that PAI and the rapeseed plant population’s photosynthetic efficiency are maximized by the much smaller number of overlapping pods in the vertical canopy.Dense,long,and upward-oriented pods together with no or one to two branches with narrow angles allow higher plant density and dramatically increase PAI and sunlight interception,thereby avoiding shade-avoidance syndrome [83]due to severe branch overlap in the canopies of current cultivars.Such plants also avoid wastage of nutrients,and photosynthetic assimilates by branches that produce no or just a few pods.Shading or indirect sunlight (reflected light) reduces the ratio of red to far light,resulting in the decrease of both photosynthetic efficiency and disease resistance [83] through a dynamic balance of Della-JAZ-phyB proteins and may impair the development of pods and seeds.Plants with reduced height (about 1.5 m) and wide stem diameter and strength can support heavy pods and thus have higher resistance to lodging.A long pod,including its beak can produce more photosynthates to increase seed weight per pod,either through the increase in seed number or seed size or a balance between the two.

Fig.2.The proposed ideal architecture of rapeseed plants for de novo design.Plants at the pod development stage are shown.The description of key traits is shown in the left column.Each variable is based on existing genotypes for these morphological or agronomic traits.(A–C)The three alternative plant types.(D)A representative of the current cultivars.

The second most important advantage of the de novo design is to target genotypes whose plants with fewer leaves and especially with early establishment of a stout and extensive root system required to support fast vegetative growth in spring,and thereby the whole growth period can be shortened (e.g.,to 180 days) to fit into flexible crop rotations.Shortened growth period does not inevitably compromise seed yield,owing to the re-constructed relationship between leaf number/vegetative growth/HI and seed yield described above.

The above design does not exclude any design for taller plants(e.g.,2.8 m) with several upward-oriented branches that are expected to spatially better utilize sunlight to produce more photosynthates.However,no germplasm shows that such tall plants can hold heavy seeds without lodging.There is no information available to breed for lodging resistance by gene editing or other biotechnologies.Further,such tall plants generally have long growth periods and are unsuitable for cultivation under current crop rotations or weather conditions in many rapeseed growing areas.

In a field test,a genotype (Fig.3A) at the density of 4.5 × 105plants ha1produced ＞1.2 × 108pods ha1,and corresponded to the theoretical seed yield of the new plant type (Fig.2A–C) of＞12 t ha1(calculated by multiplying pod number per hectare,seed number per pod,and grams per thousand seeds).If the YP/FP ratio can be as high as 42% (the average YP/FP of the UK and China.Table S2),the farm yield will be 5.0 t ha1.The previous studies [84] showed that at densities from 1.5 × 105to 9.0 × 105plants ha1,both total biomass and seed yield of cultivars with compact plant type increased with the plant density of up to 6.0×105or 7.5×105plants ha1,but total biomass and seed yield decreased when density was increased in normal commercial cultivars.Considering that the new plant architecture is especially suitable for high density,planting at a density of 6.0 × 105plants ha1will produce 1.4 × 108pods ha1;this would translate to 15 t ha1seed yield and double the rapeseed yield records(the current yield potential).With the new plant architecture yield potential and additional advantages such as more accessibility to mechanized harvest and shorter growth period,growing rapeseed crops will have an advantage over other staple crops in many regions of China as the de novo designed rapeseed cultivars will fit in rapeseed–mid-maturity rice or rapeseed–rice–rice rotation systems.

Although the de novo design of such crop architecture is challenging,particularly for established breeders who will have to shift their priorities from conventional plants to the new type,we should aim for innovation,as partly accomplished in rice [85].The de novo design will rely largely on the integrated utilization of existing germplasm resources and new biotechnologies on the horizon,such as breeding by genome design and genome editing.

4.2.Broadening genetic diversity in rapeseed germplasm

There is wide variation in traits contributing to seed yields,such as seedling vigor,flowering time,root and shoot biomass,resistance to biotic and abiotic stress,and resistance to pod shattering in the existing rapeseed germplasm.There is wide variation among natural and mutant accessions for plant architecture traits.Several mutants show compact inflorescence with dense and upwardoriented pods [79–81] and multiple inflorescences [82],and some are shown in Fig.3.A comprehensive investigation of 33 traits of 96 genotypes indicated wide variation in seed yield-related pod attributes [79].It revealed that the main inflorescence is the principal source of seed yield,producing a good number of ovules and a large number of long pods with a concomitantly high number of seeds per pod.However,it is widely accepted that modern rapeseed breeding pools have extremely narrow genetic diversity due to genetic bottlenecks posed by the small number of founders involved in allopolyploidization events during the origin of B.napus,strong adaptive selection in strict eco-geographic gene pools,and intensive agronomic selection made in recent breeding programs for essential seed quality traits [86,87].To broaden the genetic diversity,the diploid species,which have more genomic divergence and harbor novel variation [11,88–90] have been exploited for heterosis and other traits in rapeseed breeding programs.As a result,many resynthesized allotetraploid accessions have been produced from interspecific/intergeneric crosses [91].Although very few of these resynthesized lines have been bred into elite commercial varieties to date,owing to their generally poor performance,they have the potential to be used in the de novo design of new rapeseed.Cytogenetic and molecular analysis revealed that resynthesized rapeseed lines are subject to extreme structural variation [92],particularly homeoelogous exchange,allowing selection for target traits.Current findings [87] suggest that the natural and resynthesized germplasm of oilseed rape presents rich DNA sequence variations such as SNPs,small InDels,large segmental deletions,and duplication,which showed linkage to traits among varieties.The genomic variation termed polyploid genome plasticity allows high-pressure selection to create new phenotypes.We anticipate that rapeseed has a novel source of genome plasticity waiting for the intensive selection and development of the ideal plant types.

Fig.3.Representative genetic variation in rapeseed pod attributes.(A)Genotype corresponding to the main inflorescence of the type II design.(B)genotype corresponding to the main inflorescence of the type I and type III designs,but the type III design has two to three primary stems that look like the genotype displayed in G.(C,D)Two genotypes with long inflorescence,small angle and dense pods.(E) Genotype with two parallel and closely connected inflorescences and thus producing more pods.(F,G) Genotypes with dense and upward pods.(G)Genotype with two or more primary stems.(H)Genotype representing pod length of current commercial cultivars.(I)Genotype with long pods.Note:the sizes of plants (A–G) are not to scale.

4.3.The power of genome/gene-editing technology

We have witnessed the potential of conventional transgenic technology by introducing exogenous or endogenous genes into plants in achieving vast economic benefits in the past twenty years,including in rapeseed [91].Examples include but are not limited to resistance to the herbicides glufosinate and glyphosate and barnase-mediated sterile hybrid of rapeseed,producing massive profit in North America.Gene editing and genome engineering technologies provide an unprecedented opportunity to manipulate crops more precisely and efficiently.Gene editing enables us to knock in,knock out,and delete a gene or multiple genes for traits of interest via the processes of non-homologous end-joining and homologous recombination of chromosomes [93,94].Single-base editing and gene expression level manipulation can also be performed [95–97].It is exciting that this technology becomes applicable just after most crop reference genomes have been de novo assembled,enabling DNA sequences of target genes to be ready for primer design for gene editing.Fortunately,community efforts have developed rapeseed genetic transformation and gene editing methods,although not all rapeseed cultivars are amenable to genetic transformation.Gene editing has enabled the alteration of many traits in rapeseed.Examples include resistance to pod shattering [98],glucosinolate transport (low seed glucosinolates)[99],yellow seed [100],high oleic acid content [101],herbicide resistance [102],branch number and angle for compact plant architecture [103],multilocular pod [104],and plant height[105].For the de novo design of new rapeseed plant architecture,it is promising that editing of certain target genes to create new variations/traits is combined with the use of other already existing target traits and variations.

4.4.Breeding by genome design

The de novo design of rapeseed cultivars with stress resistance and ideotype architecture requires dissection of their genetic loci or cloning of genes.To date,at least 1500 genetic loci of traits have been mapped (Table S1).Besides those reviewed above,many more causal genes influencing canopy architectural traits and seed yield have been cloned in rapeseed.Examples include TFL1,D14,RGA and lAA7 for plant architecture,CYP78A9,EOD3,UPL3.C03 for seed weight and pod length,SFAR4,ORF188,and TPK for oil content(Table S1) [62,91].Further efforts are required to uncover other ground-breaking plant architectural traits contributing to high yield and oil content.For traits that are beyond visual phenotyping(e.g.,physiological characteristics)and time-consuming and laborintensive,applying new high-throughput phenotyping technologies such as LiDAR,hyperspectral imaging,NDVI,and automated imaging could accelerate trait dissection.

Conventional and molecular marker-aided selection may be hard to perform when many loci or genes are involved in selection or combination from different germplasm accessions.In this case,high throughput technologies have to be developed or adopted.Several genomic technologies are promising.Comprehensive knowledge of the genome,phenome and gene network involved in trait expression would enable the design of crops in silico[106]and then validation across growing environments.A practical technology is a genomic selection in animal and crop breeding[107],which uses genome-wide DNA markers and target trait phenotypic values from a training population to predict genomic estimated breeding values of candidates in a test population with genotypic data.A comprehensive technology is breeding by genome design that integrates molecular markers linked to or genes controlling target traits with ‘‘elite” genomic regions.Breeding by genome design is a kind of synthetic biology of crops when it combines with gene-editing technology.Both technologies require genomic information.A large number of rapeseed accession genomes have been sequenced [108–110].These technologies and genomic resources provide bases for the de novo design of new rapeseed cultivars.

In summary,the prediction shows the possibility of increasing rapeseed yield by creating ideal plant architecture.The rich genomic and phenotypic diversity in natural and mutant germplasm provides trait variation required for rapeseed architecture de novo design.The challenge now is efficiently and rapidly combining different traits to breed rapeseed varieties with increased productivity and profitability.We anticipate that gene editing and breeding by genome design methodologies will facilitate the creation of high-yielding rapeseed ideotypes for future productivity and sustainable development.

CRediT authorship contribution statement

Shengyi Liu:Conceptualization,Funding acquisition,Investigation,Methodology,Resources,Writing– original draft,Writing–review &editing.Harsh Raman:Conceptualization,Investigation,Writing– original draft,Writing– review &editing.Yang Xiang:Data curation,Funding acquisition,Investigation,Methodology,Resources.Chuanji Zhao:Data curation,Investigation.Junyan Huang:Data curation,Investigation.Yuanyuan Zhang:Data curation,Investigation,Writing– review &editing,Project administration.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

We thank Professor Yonghong Li at the Hybrid Rapeseed Research Center of Shaanxi Province for providing the photo presented in Fig.S1,and Mr.Shuntai Zhang at the College of Humanities and Law of Huazhong Agricultural University for drawing the model of rapeseed architecture in Fig.2A–C.We highly appreciate the critical discussion and valuable comments from Professors Qiong Hu and Ni Ma at the Oil Crops Research Institute of CAAS,Professor Yonghong Li,and two anonymous reviewers.This research was supported by the National Natural Science Foundation of China (U20A2034 and 32070217) and the Agricultural Science and Technology Innovation Program of the Chinese Academy of Agricultural Sciences (CAAS-ZDRW202105 and CAASASTIP-2013-OCRI).

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2022.05.003.