HAL Id: hal-01214153
https://hal.archives-ouvertes.fr/hal-01214153v3
Submitted on 10 Nov 2015
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of
sci-entific research documents, whether they are
pub-lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Diversity in plant breeding: a new conceptual framework
Isabelle Litrico, Cyrille Violle
To cite this version:
Isabelle Litrico, Cyrille Violle. Diversity in plant breeding: a new conceptual framework. Trends in
Plant Science, Elsevier, 2015, 20 (10), pp.604-612. �10.1016/j.tplants.2015.07.007�. �hal-01214153v3�
Opinion
Diversity
in
Plant
Breeding:
A
New
Conceptual
Framework
Isabelle
Litrico
1,*
and
Cyrille
Violle
2Faced with anacceleratingrate of environmentalchange andtheassociated needforamoresustainable,low-inputagriculture,theurgentnewchallengefor cropscienceistofindwaystointroducegreaterdiversitytocroppingsystems. However,thereisadearthofgenericformalisminprogramsseekingtodiversify crops.Inthisopinion,weproposea newframework,derived fromecological theory,thatshouldenablediversitytargetstobeincorporatedinto plant-breed-ingprograms.Whileecologicaltheoryprovidescriteriaformaintainingdiversity andoptimizingtheproductionofmixtures,suchcriteriaarerarelyfullyrealized innaturalecosystems.Conversely, cropbreedingshouldoptimize both agro-nomicvalueandtheabilityofplantstoperformandlivealongsideoneanother. Thisframeworkrepresentsanopportunitytodevelopmoresustainable crops andalsoaradicalnew waytoapplyecologicaltheorytocroppingsystems.
ANewFrameworkforBreedingDesign
Onequestionhasbeenmuchdebatedinagronomyandecology[1–3]:isitbettertoliveamong strangersorrelatives;inotherwords,areplantmixtures(seeGlossary)moreproductive,more resistanttopestsandpathogens,andmoresustainablecomparedwithplantmonocultures? The answer will have crucial implications for plant-breeding programs. The introduction of geneticand/orspeciesdiversity(geneticheterogeneitysensu lato)tocroppingsystems isa pressingissueinagriculture[4].Therearetwomainchallenges:(i)ageneralneedtooptimizethe multifunctionality of crop systems [5,6]; and (ii) a need to adapt existing crop systems to acceleratingratesofenvironmentalchanges.
Tofacethesechallengeswitharesponseofchoice,ecologicalstudiessuggestthatagricultural systemscontaininghighgeneticheterogeneityshouldbefavored[1,3].Thispathshouldalso favorlow-inputagriculture, which isanotherofthemajorchallenges facedbycrop science today. Compared with a monoculture, the combination of legumes and grasses in sown pasturesisagoodexample ofhowsignificantproductivityimprovements canbeobtained, withouttheinputofexternalnitrogen[7].Nevertheless,despiteimportantongoingeffortsto mergeecologyandagriculturalsciencewithinacommondiscipline(agroecology),the introduc-tionofgeneticheterogeneitytocropsstillpresentsasignificantchallengetobreeders. Intheoreticalecology,diversity(especially,speciesdiversity)hasanimportantroleinthefunctioning andresilienceofecosystems[8–10].Overthepasttwodecades,numerousecologicalexperiments havebeeninitiatedthatexaminethecontrolofecosystemprocessesbybiodiversity.Theseinclude controlofplantproductivity[2,11,12].Despitesometimes-heateddebate,theevidence neverthe-less suggests a positive effect of biodiversity on primary productivity. Strikingly,most crop-breedingprogramsarepersistentlymonospeciesand/or-genotypeculturalintheiroutlook. Themainstreamapproachtocropbreedingisstilltoimprovegenotypesbycultivatingindividuals inpureculture,eitherinmono-genotypicormulti-genotypicpopulationsofasinglespecies.This
Trends
To face societal and environmental challenges, agriculture is changing. Promotingspeciesand/orgenetic het-erogeneityinagronomiccoversisone ofthekeydevelopments.
Mostcropcoverscomprise mono-gen-otype varieties, except synthetic or populationvarietiesthat,through selec-tionstrategy,containsomeresidualand noncontrolledgeneticdiversity. Thereare currentlyfeworno plant-breedingapproachesappliedto agro-nomiccoversthatincorporatemultiple genotypesand/orspecies.
Ecologytheoryprovidesthekeyforthe maintenance and optimized applica-tionofheterogeneouscovers. Wecallforecologicalassemblyrulesto proposea novelparadigmfor plant-breeding programs when selecting foragronomicmixtures.
1
P3FUR004–INRA–LeChêneRD 150,F-86600Lusignan,BP86006, France
2
CEFEUMR5175,CNRS–Université deMontpellier–UniversitéPaul-Valéry Montpellier–EPHE-1919routede Mende,F-34293Montpellier,CEDEX 5,France
*Correspondence:
isabelle.litrico@lusignan.inra.fr
(I.Litrico).
604 TrendsinPlantScience,October2015,Vol.20,No.10 http://dx.doi.org/10.1016/j.tplants.2015.07.007 ©2015ElsevierLtd.Allrightsreserved.
mirrorsthemostcommonstructureofcurrent varieties(Figure1),whosecovercomprises a singlegenotype(purelinesorhybrids)ormultiplegenotypes,inthecaseofsyntheticvarieties. However,evenforsyntheticvarieties,geneticdiversityislowanddirectionalselectionisusedto reducecover-widephenotypicvariance.Competitionexperimentsdescribedintheecological andevolutionaryliteraturehavelongstressedthattheperformanceofagenotypeinpureculture candifferradicallyfromitsperformanceinmixedculture[13].Similarfindingshavebeenreported intheagronomyliterature[14–16].The differenceinperformancesbetween pureandmixed culturesarisesfromlocalselectionpressuresgeneratedbyintra-and/orinterspecificinteraction withneighborsinthemixture[17–20].Therefore,there isanurgentneedforplant-breeding schemestobegintoformalizetheintegrationoftheseinteractions,ratherthanignorethem(with theexceptionofresearchdoneonbi-speciesmixtures;e.g.,[21,22]).
Breedingprogramsusuallyseektooptimizekeyagronomictraits,suchasseedqualityand quantity,biomassproduction,andpestanddiseaseresistance.Clearly,thisobjectiveshould persistwhenacropmixtureisthebreedingtarget,butothertraitsmustalsobeincorporatedand optimized,including,critically,anabilityto‘liveandperformwithothers’.Interestingly, commu-nityecologyhasmadeimportantrecentprogressinthis,helpedbytheoreticaladvancesin trait-basedecology[23–25].Here,webrieflyreviewearlierattemptsinbreedingdesigntoimprove cropmixtures.Wealsoproposeanewframeworkforbreedingdesignthatexplicitlyincorporates geneticheterogeneitythroughtheinclusionofmechanismsderivedfrombiodiversitytheory.
Glossary
Agronomictrait:phenotypicfeature ofaplantthatatleastpartially determinesitsagronomicvalue (biomassproductionand/orseed production).
Artificialselection(orselective breeding):breedingofplantswith thepurposeofimprovingtraitsof particularimportancetohuman beings.
Assemblyrulesinecology:these rulesrefertothesetofprocesses thatexplainsthelocalcoexistenceof species.Thisincludestheroleof bioticinteractionsandabiotic constraintsonspeciesperformance. Thisideaisintroducedhereinthe contextofplantbreeding
programmes,tohelpdevelopasetof rulestoassistbreederstocreate productiveandsustainable multi-species(ormulti-genotype)mixtures (‘Ideomixes’).
Compositemulti-linepopulation: apopulationofhybridsarisingfrom crossesbetweenmultipleinbred lines.
Functionaltrait:anymorphological, physiologicalorphenologicalfeature, measurableatsomelevel(fromcell towhole-organism)withoutreference totheenvironmentoranyotherlevel oforganisation[50].
Inbredline:apoolofindividuals derivingfromasinglehomozygote genotype.
Interactiontrait:phenotypicfeature ofaplantinvolvedinitsinteractions withneighboringconspecificor heterospecificplants. Mixture:amulti-speciesand/or multi-genotypeculture.
Monoculture:single-speciesculture. Herethedefinitionisrestrictedtoa single-genotypeculture.
Multi-lines:amixtureofinbredlines. Resource-usecomplementarity: thetendencyforcompetingspecies toexploredifferentportionsofthe resourcesavailablelocally. Tradional selecon
schemes with genealogic and recurrent selecon Test of the ability of a
variety to perform in ‘associaon’ with
other variees Evoluonary plant
breeding
Increasing genec and species heterogeneity Approaches to plant breeding
One genotype
• Hybrid Fvariety (maize, sunflower…)
• Pure line variety (wheat, soya...) • Clonal culvar (olive, potato…)
=> most common situaon in CA
Several genotypes
• Mix of hybrid F variees (rare situaon in CA)
• Mix of pure line variees (barley, wheat, rare situaon in CA) • Mix of synthec variees (rare situaon in CA, but occurs in
forestry and grasslands)
• Synthec variees (alfalfa, leek, rare situaon in CA, but occurs in
forestry and grasslands)
Single-species cover Bi-species cover
One genotype per species
• Mix of pure lines and/or hybrids from different species (rare situaon in CA but occurs in agroforestry )
Several genotypes per species
• Mix of synthec variees (rare situaon in CA, but occurs in agroforestry or temporary grasslands; e.g., ryegrass and white clover)
Mul-species cover (more than two species)
Several genotypes per species
• Mix of synthec variees (Cocksfoot, Ryegrass, Alfalfa, Clover, …)
(extremely rare in CA but occurs in temporary grasslands)
(C) (D) (E) (F) (G) (J) (I) No development yet (B)
One genotype per species
• Extremely rare situaon in CA
(A)
(H)
Figure1.DifferentLevelsofSpeciesandGeneticDiversitywithinAgronomicCoversUsedinConventional Agriculture(CA)andtheirRelatedPlant-BreedingApproach.Differentlevelsofdiversitywithinagronomiccoversdo existinCA,fromsingle-genotypeandsingle-speciessituationstomulti-speciesandmulti-genotypessituations.The advancement(green,advanced;orange,weak;red,nodevelopment)inplant-breedingapproachesdependsonthelevelof diversity.Seemaintextformoredetailoneachtypeofplant-breedingapproach.Photos:(A)Multi-speciessowngrassland (LotuscorniculatusL.,Medicagosativa,TrifoliumpratenseL.,TrifoliumrepensL.,Dactylisglomerata,Loliumperenne, Festucaarundinacea);(B)Mixofcereals(TriticosecaleWittm.exA.Camus,AvenasativaL.,Hordeumvulgare)andlegume (PisumsativumL.);(C)Mixofgrass(L.perenne)andlegume(T.repensL.);(D)Mixofcereal(Triticumaestivum)andlegume (PisumsativumL.)(E)Mixinagroforestry(TriticumdurumandhybridJuglans);(F)MixofT.durumcultivars;(G)M.sativaL.; (H)Zeamays;(I)T.durum;(J)OleaeuropaeaL.Reproduced,withpermission,fromD.Denoue(A),C.Maitre(B),B.Cauvin (C),M.C.Lhopital(D),C.Duprax(E),P.Saulas(F),M.Preudhomme(G),J.Niore(H),H.Cochard(I),andJ.Weber(J). CopyrightINRAinallcases.
LivingandPerformingwith Others:TheChallengesofPlant-Breeding Programs
Despitealong-terminterestintheimprovementofmixturesthroughplantbreeding[26],little formalismhasbeendevelopedtodealwiththebreedingofmultiplecroppingandmulti-species mixtures [27]. Untilnow, breeding programs have attemptedto select varieties and create ideotypes basedonan ‘average’ability to perform whengrown in ‘association’ with other varieties[28].Toestimatethisaverageabilitytoperforminassociation,theapproachusedhas beenderivedfromdialleltestswithmaize(Zeamays)forthedeterminationoftheabilityoflinesto hybridize[29,30].Notably, this has beenused to improvemulti-genotype blends, including multi-line[31–33]andmulti-species[29,34,35]mixtures.However,thisapproachhaslimited portabilitybeyondthespecifictestmaterialandinthecontextofaparticularstudy.Tosome extent, these limitations can be offset by using large sample sizes and a broad range of genotypes[28].Nevertheless,evenif‘biodiversity’isrestrictedtotheconsiderationofbinary associations,thelargenumberofdifferentbinarycombinationsrapidlymakestestingthemall difficult[23].Furthermore,usingpairwisecombinationshasoftenbeendescribedinthe eco-logicalliteratureasposinganimpasse,becauseagenotypebehavesdifferentlyinthepresence ofoneormorecompetinggenotypes[23].Thus,McGilletal.[23]callfortheanalysisofthe ‘interactionmilieu’asawhole,toaccountforallkindsofdiffusebioticinteractionoccurringinan assemblage.Italsoavoidsthelogisticproblemofanalyzingapotentiallyvastnumberofpairwise combinations.
Insteadofoptimizingapoolofgenotypesbyestimatingtheiraverageabilitiestoliveandperform with others,a proposed alternative has beento directly exposecrop mixtures to complex selectionpressures[17,36,37].Thisaccountsforthelocalselectionpressuresduetogenotype genotype interactionswithin themixture.This isthemain objectiveofevolutionary plant-breeding methods[28,38,39].Evolutionary plant breeding has beenappliedto composite multi-line populations but never to multi-species mixtures [14,21] The efficiency of this selectionapproachinmulti-linemixturesdependsonpotentialnegativecorrelationsbetween fitnessandagronomicvalueinselectedcoexistingindividuals[40].Tomitigatethe consequen-cesofanegativecorrelation(oralackofone)betweenfitnessandagronomicvalue,artificial selectionmanipulationscanbeusedwithinthesemixturesinparallelwithnaturalselection[41]. Twomajorlimitationsareassociatedwiththisapproach:(i)drifteffects,whichcanleadtotheloss ofgeneticvariability[39];and(ii)thelongperiodrequiredtoallownaturalselectionprocessesto act;thisisespeciallyaproblemiftheselectiontargetsareperennialplants.
Altogether,attemptstoselectplantsforanabilitytoliveandperformwellwithotherplantshave rarelybeenexploredordevelopedatthisstage.Attheveryleast,theyseemratherinefficient waystorespondtoactualagroecologicalneedswhenappliedtomixtureswithmorethantwo components.Here,wearguethattheuseofknowledgebasedonecologicaltheoryshouldbeof assistanceinpromotingwithin-cropcoverdiversity.Thisshouldimprovetheperformanceof mixturesthroughredesignedbreedingprogramsand,atthesametime,maycircumventsome ofthelimitationsofearlierapproaches.
ImprovingthePerformanceofCropMixtures:Insightsfrom Ecological
Theory
Foralongtime,ecologistsandagronomistshavestudiedtheyieldandstabilityofmixturesof genotypesorofspecies[42–46].Overthepast20years,experimentalandtheoretical under-standingshaveconvergedtowardsacceptanceofthegeneralprinciplethatamixtureofspecies or genotypes is inaverage more productive, and its production is more stable with time, comparedwitheachspeciesgrowninisolation[2,47,48].Thisprinciplehascrucial consequen-cesforartificial(crop)mixtures,grownundercontrolledconditions.Akeyunderlyingconceptis thatanecosystemcomprisingseveralspecieseachexploringaportionoftheavailablelocal
resources(resource-usecomplementarity),isexpectedtobemoreefficientintheutilization ofthetotalpoolofavailableresourcesand,thus,shouldbeabletotransformthemintoagreater amountofbiomass[49].Thishypothesishasalsobeenappliedtothetemporalresponseof ecosystems ina fluctuatingenvironment. Interms of composition andyield, aspecies-rich ecosystemisexpectedtobethemoststableovertime[43]because,ateachtimestep,there should beat least one persistent species thatis well adaptedto any given environmental condition[2,44,48].Althoughtheunderlyingmechanismsofthestabilizingroleofbiodiversityare notcompletelyelucidated[45],thediversity–stabilityhypothesisispivotalforthedevelopmentof sustainable crops. Recently, the plus-value of species and genetic diversity for biomass productionanditstemporalstabilityinsowngrasslandshasbeendemonstrated[46].
Inrecentdecades,theriseoftrait-basedecologyhasallowedtestingandquantificationofthe mechanismsofresource-usecomplementarybetweenspeciesand,thus,anunderstandingof thelinks between plant biodiversityand primaryproductivity.A keypostulate isthat some functionaltraits(defined as any morphological, anatomical, physiological,orphenological featuremeasurableattheindividuallevel[50])arerelatedtotheresource-useaxesofthespecies
[51].Inotherwords,measuringthefunctionaltraitsofaplantisexpectedtoreflectthetypeand quantityoftheresourcesitconsumes.Asaconsequence,beyondthesignificanteffectofthe numberofspeciesorgenotypes[2,52,53]ontheperformanceofanecosystem,thereisgrowing consensusthattheroleoffunctionaltraitdiversity(functionaldiversity)iskeytothemaintenance ofaproductiveandstableecosystem[54].Maximizingbetween-speciesfunctionaltrait differ-ences in a particular ecosystem is a way of maximizing ecosystem-level resource-use complementarity.
However, this resource-economics perspective to ecosystems is conditioned by complex between-traitcorrelations,as emphasizedinthenext section.Furthermore,thisperspective ofecosystemfunctioningnecessarilyimpliesafocusonthe‘speciesunit’andadescriptionof speciesby trait averages(the mean-field approach [24]). This approachhas recently been challengedintheecologicalliterature[12,24,55,56],whereitisarguedthatindividualdifferences maybeasimportantasbetween-speciesdifferencesinexplainingthemaintenanceofspecies coexistenceand,subsequently,thecontrolofecosystemfunctioningbybiodiversity.Toquantify resource-usecomplementaritywithinecologicalcommunitiesbytheintraspecific:interspecific phenotypicvarianceratio, Violleetal.[24]usefullyproposed avariance decomposition.This echoestheuseofWright'sstatisticsinpopulationgenetics.Thisapproachisofinterestforplant breeding,whichalreadyquantifiestwodistinctkindsofvariance:geneticandphenotypic.
FromNatural PlantCommunitiesto‘Optimized’SpeciesAssemblagesin
CroppingSystems
Itisstrikingthatthefunctionalrelationswithinnaturalandexperimentalecosystemsareoften verydifferent.Inexperimentalecosystems,therelationsaremostoftenpositiveandsaturating, whereas in natural ecosystems, a range of relation forms are found (positive, neutral, and negative)[57–59]. Onereason for this wouldseem to bea lack ofa true ‘optimization’ of resource-usecomplementaritywithinmanynaturalecosystems.This,inturn,islikelybecause thereare severalkinds oftrade-offwithinorganisms andpopulations [60–62].Asa conse-quence,thediversityeffectonecosystemprocesses,suchasproductivity,isoftenweakunder natural conditions. Conversely,one may expect greater diversity effects inan agronomical context,inparticularunderlow-inputconditions[46,63].Here,weproposethatplant-breeding programscanoptimizecomplementaritybetweengenotypesinamixture,to‘force’thepositive relationbetweendiversityandyield.Thisisthecornerstoneofthenovelframeworkwenow proposeforplantbreeding,wherebybreedersshouldfocusonthetraitsidentifiedasthemajor contributorstoresource-usecomplementarityinecologicalstudies.Asastartingpoint,rooting depth, vegetative architecture, and phenology, including growth rhythms, are particularly
important[64–67].Thesetraitsinvolvedininteractionsbetweengenotypesandspecies,weterm ‘interactiontraits’.Fromniche-basedecologicaltheory,wewouldexpectgreater resource-usecomplementaritybetweenspecies(orgenotypes)whenthebetween-speciesvariance(or between-genotypesvariance)ofinteractiontraitsishighandthewithin-species variance (or within-genotypesvariance) of interactionstraits is low[24,55,68] (Figure 2). A mixture that maximizesdifferencesininteractiontraitvaluesbetweenindividualswillbemoreproductiveand morestable with time [69], especially under suboptimal environmental conditions[46].We suggestthatthisshouldbeakeyobjectiveofthenewgenerationofdiversity-orientedbreeding programs.
Itisimportanttonotethatdifferenttypesofcompetitiveinteractiondoexist[70–72]. Resource-usecomplementaritycanbepromotedwhencompetingspeciesdisplaycontrastingtraitvalues. Conversely, these species can suffer from competitive exclusion in cases of competition hierarchy(if one specieshas acompetitive advantage overanother). This is detrimental in thecaseofcropmixtures,inwhicheachcomponentneedstobemaintainedtoperformwell.As a consequence, those competitive exclusion processes will have to be removed by plant breeding(seebelow).
Beyondtheroleofresource-usecomplementarity(and,thus,between-speciestraitdifferences) inthe regulationofecosystem processes,trait-basedecologyhasalso clearlyidentifiedthe importantroleofdominantspeciesinthecontroloftheoverallyieldofamixture[73].Indeed,the short-termyieldofgrasslandcommunitiesisprimarilyexplainedbythetraitvaluesofthemost abundantspecieswithinthemixtures [74].Inotherwords,ecosystemsthatcontainspecies havingtraitsthatconvergetowardsthetraitsofthemostproductivespeciesofthemixture[75], expressthebestyields[76,77].Hereafter,thetraitsinvolvedinplantyieldaretermed‘agronomic
Trait axis Trait axis Agronomic traits Interacon traits
σ
IC 2σ
IC 2σ
IG 2σ
IG 2 (A) (B)Figure 2. Variance Partitioning AppliedtoPhenotypicDiversityina MixtureofCropGenotypesor Spe-cies.Consideramixturecontaining multi-plecomponents(genotypesorspecies). Foranyphenotypictrait,thereisa within-componentvariance(sG2
)andatotal var-iance(sM2
)(thevariancebetweenall indi-vidualsinthemixtureirrespectiveoftheir taxonomicorgenotypicidentity).(A)The optimumagronomicvalueofthemixture shouldfirstbemaximized.Todothis,both thewithin-componentvariance(sG2
)and thetotalvariance(sM2
)shouldbe mini-mizedaroundasingleoptimalvaluefor eachofthekeyagronomictraits.(B)Next, resource-use complementarity between thecomponents shouldbe maximized. Todothis,thevarianceofeachinteraction traitshouldbemaximizedbetween com-ponents (divergent selection) and mini-mizedwithin eachcomponent (without overlapbetweenthecurvesforeach com-ponent).Inthisway,thesG2:sM2
ratiowill be minimized, following the framework proposedbyViolleetal.[24].
traits’.Agronomictraitscanbeeitherdirectmeasuresofyield(e.g.,vegetativebiomass,grain production,orseedquality)orproxiesofyield,asiscommoninfunctionalecology(e.g.,specific leafareaisakeydeterminantofthegrowthrateofaspeciesandecosystemproductivity[74,78]). Plant-breedingprogramsshouldworktooptimizethemeanvalueoftheagronomictraitsofeach ofthecomponentsofamixture,becausethisshouldleadtoagreateryieldforthewholemixture.
FromIdeotypestoIdeomixes
Theincorporationofplantbreedingintoagroecologicalprogramsinsuchawayastobestexploit theadvantagesofmixinggenotypesandspeciesrepresentsthenextmajorstepinagroecology. Indeed,breedingcanbreaktheconstraintsinherenttowildspeciesandnaturalecosystems[79], withtheresultthatseveralfunctionscanbeoptimized.However,thisalsoimpliesaparadigm shift for thinking inplant breeding.Since the green revolution started,plant breedershave focusedonideotyping; thatis, the creationof elite genotypesselected for their agronomic performanceinaparticularartificialenvironment.Giventhegrowingimportance(asdiscussed above)ofusingmixturestomeetthechallengesposedbytheneedforsustainabilityina fast-changingworld,weproposeherethecreationofelitemixtures,whichweterm‘ideomixes’.A criticaladvantageofbreedingisthatitcanovercomethelimitationsandcomplexityofnatural ecosystemsby decouplingvarioustraits(hereinteraction andagronomictraits)bybreaking geneticconstraintsandoptimizingvariousmechanismsviathetraitsthathavebeenfoundto displayantagonistvaluesinthefield(e.g., resource-usecomplementarityversuscompetition hierarchy).Inecology,ithasbeenarguedthatthecurverepresentingtherelationbetweenyield andthenumberofspecies,issaturating[11,80,81](Figure3).Thisisthoughttobebecauseofan increaseinfunctionalredundancyamongspeciesasthenumberofspeciesincreases;thatis,as speciesnumbersincrease,thespeciesaremorelikelytobeecologicallysimilartoeachother (see[82]foranexample).However,theoreticallyatleast,breedingshouldbeabletocreatean artificialmixturehavingnoredundancyandallcomponentsofthemixtureswouldcompletelyfill theecologicalspace.Thesaturationpointcanbeonlyreachedathigherlevelsofdiversitythanks tothe selectionof contrastingandmorespecializedgenotypesand/orspecies, as demon-stratedinmicrobialexperiments[83].Weseeplantbreedingashavingthepotentialtoallow escapefromtheenvelopecurveidentifiedasaconstraintinecology(Figure3).Next,weshow howbreedingcanproduceideomixes.
Genec heterogeneity
Natural ecosystems
Crop mixtures
Funconing
Figure3.HypotheticalGenetic Het-erogeneity Functional Relations in NaturalEcosystems(Gray)andCrop Mixtures(Red).Thegenetic heteroge-neity of an assemblage can increase throughanincreaseineitherthenumber ofspecies(orgenotypes)orinthegenetic distancesbetweenthespecies(or geno-types).Asaturatingcurveisexpectedfor natural ecosystems due to increasing functional redundancy (see main text). Saturationisnotexpectedforcrop mix-tureswherebreedingisabletomaximize the functioning of all components and theirabilitytocohabitandperformwith each other (notably if ‘specialists’ are selected:seemaintextformoredetail). The creation of crop mixtures through breedingrepresentsanewandpromising avenueforagroecologicalresearch.Inthis schema,productivitylevelsforboth nat-uralandartificialmixturesandthe hierar-chy between them are represented diagrammatically.
Optimizationof CropMixtures: TowardsRedesignedBreedingPrograms
Weproposeanovelframeworkforimprovingcropmixturesbybreeding.Basedonideasfrom theoreticalecology,aswellasrecentadvancesintrait-basedecology,weproposethatbreeding programmes be refocused by basing them on the decoupling and optimization of both interactiontraitsandagronomictraits.Thisimpliesthatselectionisdonenotonlyforthemean valuesofagronomictraits(andoccasionallyalso forthevariance,inthecase ofagronomic qualityofmixtures:seebelow),butalsoforthemeansandvariancesofinteractiontraits.The choiceoftargettraitsisacentralelementofthisapproach. Itissuggestedthatthefocusis placedon:(i)asmallsetofmajortraits,suchasvegetativebiomassorgrainproduction(theseare centralagronomictraitsandtheirchoicewilldependontheselectionobjectives);and(ii)three typesofinteractiontrait.Asapioneeringchoiceofinteractiontraitstofocuson,weproposetraits relatedto resourceforaging (e.g., rooting depth), phenology(e.g., the period of vegetative production),andabovegroundarchitecture(e.g.,stembranching).Thesetraitsrelatetospatial resource-usecomplementarity,temporalresource-usecomplementarity,andlightpartitioning withinthecover,respectively.Preliminaryworkiscriticaltochoosethebestcandidateinteraction traitsandthisisbestguidedbyalternatingbetweenthetraditionalapproachesofecologyand cropscience.Indeed,theidentificationofinteractiontraitsisalong-standingquestioninecology
[71,84,85]butremains puzzling[86].Specificecophysiologicalstudiesandrecurrent use of physiologymodelinginagronomycanprovide insightsintothissearch forthemostrelevant interactiontraits.
Thegeneralobjectiveofabreedingprogramrefocusedonamixtureisto:(i)optimizethemean valuesoftheagronomictraitsofthemixturebyforcingallgenotypestoconvergetowardsan optimalvalue(Figure2);and(ii)maximizethevariancebetweenthecomponentsofthemixturefor theinteractiontraitswhilesimultaneouslyminimizingthewithin-componentvariance(Figure2).In thecase where agronomic quality is the targetof selection, the objective can beto select agronomictraits(e.g.,foragechemistry)thatmayormaynotbethesameforeachcomponent ofthemixture.Thus,maximizationofthevariancecanalsobetargetedforagronomictraits.To reachthesegoals,weproposeatwo-stepframework,laidoutindetailbelow.
StepOne(SelectionStep)
Thefirststepreliesonimprovingtheagronomicandinteractiontraitsforeachcomponentofthe mixture.Inthecaseofspeciesmixtures,theideaistoapplyclassicalbreedingschemestoselect specificvaluesofagronomicandinteractiontraitsforeachspeciesseparately.Inmostcases,for agronomictraits,selectiontowardsthesametraitmeansforallspeciesisneeded.Importantly, asdescribedinsteptwo(assemblystep),thevaluesofinteractiontraitshavetodifferamong speciessoastomaintainthediversityofthemixtureandtheabilityofeachspeciestoliveand performwithothers.Thesamelogiccanbeappliedforamixtureofgenotypesofagivenspecies. Overall, thisstepisbasedona multi-traitselection approachand, thus,needs toconsider correlations amongthe traits ofinterest, notably between agronomic andinteraction traits. Severaltypesofcorrelationcanbeencountered,including:(i)Ifagronomicandinteractiontraits aregeneticallyindependent,thesetwodifferenttraitscanbeindependentlyselected(selection ononetraitwithoutcorrelativeresponseoftheother,andpossiblesimultaneousimprovementof both);and(ii)ifagronomicandinteractiontraitsaregeneticallycorrelated,selectionononetrait resultsinacorrelativeresponseoftheother.Ifthecorrelationcanbebrokenusingexisting breeding tools, particularly genetic andchromosome shuffling, selection of agronomic and interactiontraitscouldbedoneseparately.Bycontrast,ifthecorrelationisstrong,especiallyif pleiotropyisinvolved,simultaneousimprovementappearsdifficult.Inthatcase,genetic recom-binationdrivenbybreedingconstraints,suchasindexselection[87,88],canbeused.Thishas beenachievedinmanysituationsbyplantbreeding,suchasthesimultaneousimprovementof yieldandearlinessinmaizeandofforageyieldandseedyieldinforagegrasses.Theprincipleof
indexselectionisamulti-traitselectionaccordingtothevalueofanintegratedindex;thatis,a linearcombination oftheestimated breeding valuesofcandidate traitsbeing selected. The coefficientsoftheindexaredeterminedusingaconstraintsystembasedonexpectationsof geneticgains.
StepTwo(AssemblyStep)
Thisstepreliesonapplyingassemblyrulesfromecologytobuildinselectedgeneticpoolson thebasisoftheconvergencevaluesofagronomictraitsandthedivergencevaluesofthemajor interactiontraits. To evaluate thedegree ofdiversity required for optimalfunctioning ofthe mixture,process-basedmodelscanbeusedtotestandvalidatetherangeandvarianceofthe interactiontraits.Anexampleofthismightbeindividual-basedmodels[89]thataccountfor relationsbetweena genotype,itsenvironment, anditsbioticneighborsto developavirtual platformfortheoptimizationofassemblagesofgenotypesandspecies.Syntheticvarietiescan becreatedwhendealingwithauto-incompatiblespecies.Inthatcase,thissecondstepofour novelschemacanbeavoidedandreplacedbyaselectionontraitvarianceinstepone.Toreach thatgoal,wecanapplyadditionalconstraintstomaintainenoughgeneticvarianceofinteraction traitsintheconstraint systembasedonexpectationsofgenetic gains.Thisimproved index selectionmethod willenablebreeders todeal withthe trade-offbetweenselection towards certaintraitmeansandmaintenanceofdiversityforothertraitswithinagivenspecies.
ConcludingRemarks
Theproposedframeworkforplantbreedersisbasedoninsightsgainedfromecologicaltheory. Bothbreedingtoolsandmodelingapproachescanbeusedtohelpavoidtheinherenttrade-offs betweenagronomicandinteractiontraitsandtodeterminethedegreeoftraitdiversityrequired tomaintainsustainableproduction(seeOutstandingQuestions).Thisplant-breedingframework enables ideomixesto be created that may beused as the reference mixtures required by agroecology.
Acknowledgments
WethankJean-PaulSampouxandPhilippeBarreforhelpfuldiscussions.Wearegratefultoreviewersforconstructive commentsonearlierversionsofthismanuscript.ThisworkwasfundedbytheAgenceNationaldelaRecherche(ANR), France(PRAISE,ANR-13-BIOADAP-0015).C.V.wassupportedbytheEuropeanResearchCouncil(ERC)StartingGrant Project‘Ecophysiologicalandbiophysicalconstraintsondomesticationincropplants’(Grant ERC-StG-2014-639706-CONSTRAINTS).
References
1. Hajjar,R.etal.(2008)Theutilityofcropgeneticdiversityin main-tainingecosystemservices.Agr.Ecosys.Environ.123,261–270 2. Hooper,D.U.etal.(2005)Effectsofbiodiversityonecosystem
functioning:aconsensusofcurrentknowledge.Ecol.Monogr.75, 3–35
3. Malézieux,E.etal.(2009)Mixingplantspeciesincropping sys-tems:concepts,toolsandmodels.Areview.Agron.Sustain.Dev. 29,43–62
4. Isbell,F. (2015)Agroecosystemdiversification.Nat.Plants1, 15041
5. Tilman,D.etal.(2001)Forecastingagriculturallydrivenglobal environmentalchange.Science292,281–284
6. Rosenzweig,C.etal.(2013)Assessingagriculturalrisksofclimate changeinthe21stcenturyinaglobalgriddedcropmodel inter-comparison.Proc.Natl.Acad.Sci.U.S.A.111,3268–3273 7. Turkington,R.andJolliffe,P.(1996)Usingresponsefunction
modelstostudythecompetitiveinteractionsofTrifoliumrepens andLoliumperenneinmixtures:short-termandlong-term analy-sis.J.Ecol.84,563–571
8. Loreau,M.(1998)Biodiversity andecosystem functioning:a mechanisticmodel.Proc.Natl.Acad.Sci.U.S.A.95,5632–5636
9. Loreau,M.etal.(2001)Biodiversityandecosystemfunctioning: currentknowledgeandfuturechallenges.Science294,804–808 10.Tilman,D.etal.(1997)Plantdiversityandecosystem produc-tivity:theoreticalconsiderations.Proc.Natl.Acad.Sci.U.S.A. 94,1857–1861
11.Tilman,D.etal.(2001)Diversityandproductivityinalong-term grasslandexperiment.Science294,843–845
12.Cardinale,B.etal.(2012)Biodiversitylossanditsimpacton humanity.Nature486,59–67
13.Goldberg,D.E.andBarton,A.M.(1992)Patternsand consequen-cesofinterspecificcompetitioninnaturalcommunities-areviewof fieldexperimentswithplants.Am.Nat.139,771–801 14.Eagles,C.F.(1983)Relationshipbetweencompetitiveabilityand
yieldingabilityinmixturesandmonoculturesofpopulationsof DactylisglomerataL.GrassForageSci.38,21–24
15.Francis,C.A.etal.(1978)Genotypebyenvironmentinteractionsin bushbeancultivarsinmonocultureandassociatedwithmaize. CropSci.18,237–242
16.Zimmerman,M.J.O.etal.(1984)Aheritabilityandcorrelationstudy ofyield,yieldcomponentsandharvestindexofcommonbeanin solecropandintercrop.FieldCropsRes.9,109–118
OutstandingQuestions
Canresource-usecomplementaritybe optimizedinagronomiccovers,based on novel plant-breeding approaches andthe choiceof particular species andgenotypes?
Whataretheagronomicand interac-tiontraitspertainingtoagronomic cov-ers? Here, weprovide a preliminary coresetoftraits,butmore (experimen-talandtheoretical)researchisneeded toprovideamoredetailedlist.Crop physiologymodelscanbeconsidered virtualexercises to simulatedifferent situationsandtraitresponses. Can thelinkage betweenagronomic and interaction traits be broken? Studyingthetypeoflinkagesinvolved isapriorityresearcharea.
How can we address pragmatically geneenvironmentinteractionsina multi-species and multi-genotype selectionframework?
Methods based on selection index havebeenusedtooptimizeexpected genetic gains, and these methods impose constraints on trait means. Howcanweoptimizethesemethods toalsoselectfortraitvariance?
17.Allard,R.W.andAdams,J.(1969)Populationstudiesin predomi-nantlyself-pollinatingspeciesXIII.Intergenotypiccompetition& populationstructureinbarley&wheat.Am.Nat.103,621–645 18.Clay,R.E.andAllard,R.W.(1969)Acomparisonofthe
perfor-manceofhomogenousandheterogenousbarleypopulations. CropSci.9,407–412
19.Hill,J.(1996)Breedingcomponentsformixtureperformance. Euphytica92,135–138
20.Phillips,S.L.etal.(2005)Theeffectofpotatovarietymixtureson epidemicsoflateblightinrelationtoplotsizeandlevelof resis-tance.Ann.Appl.Biol.147,247–252
21.Wright,A.J.(1985)Selectionforimprovedyieldininter-specific mixturesorintercrops.Theor.Appl.Genet.69,399–407 22.Francis,C.A.(1981)Developmentofplantgenotypesformultiple
croppingsystems.InPlantBreedingII(Frey,K.J.,ed.),pp.179– 232,IowaStateUniversityPress
23.McGill,B.J.etal.(2006)Rebuildingcommunityecologyfrom functionaltraits.TrendsEcol.Evol.21,178–185
24.Violle,C.etal.(2012)Thereturnofthevariance:intraspecific variabilityincommunityecology.TrendsEcol.Evol.27,244–252 25.Garnier,E.andNavas,M-L.(2012)Atrait-basedapproachto comparativefunctionalplantecology:concepts,methodsand applicationsforagroecology.Areview.Agron.Sustain.Dev.32, 365–399
26.Mayo,O.(1980)TheTheoryofPlantBreeding,OxfordUniversity Press
27.Francis,C.A.(1990)Potentialofmultiplecroppingsystems.In Agroecologyand SmallFarmDevelopment (Altieri,M.A.and Hecht,S.B.,eds),pp.137–150,CRCPress
28.Dawson,J.C.andGoldringer,I.(2012)Breedingforgenetically diversepopulations:varietymixturesandevolutionary popula-tions.InOrganicCropBreeding(LammertsvanBueren,E.T. andMyers,J.R.,eds),pp.77–94,Hoboken,NY,USA, Wiley-Blackwell
29.Gallais,A.(1970)Modèlepourl’analysedesrelationsbinaires. Biométrie11,51–80
30.Griffing,B.(1956)Conceptofgeneralandspecificcombining abilityinrelationtodiallelcrossingsystems.Aust.J.Biol.Sci. 9,463–493
31.Gallandt,E.R.etal.(2001)Diallelanalysisofcultivarmixturesin winterWheat.CropSci.41,792–796
32.Lopez,C.G.andMundt,C.C.(2000)Usingmixingabilityanalysis fromtwo-waycultivarmixturestopredicttheperformanceof cultivarsincomplexmixtures.FieldCropRes.68,121–132 33.Mille,B.etal.(2006)Assessingfour-waymixturesofwinterwheat
cultivarsfromtheperformancesoftheirtwo-wayandindividual components.Eur.J.PlantPathol.114,163–173
34.Finckh,M.R.etal.(2000)Cerealvarietyandspeciesmixturesin practice,withemphasisondiseaseresistance.Agronomie20, 813–837
35.Finckh,M.R.andWolfe,M.S.(2006)Diversificationstrategies.In TheEpidemiologyofPlantDisease(Cooke,B.M.etal.,eds),pp. 269–308,Springer
36.Evans,D.R.etal.(1989)Coexistenceandtheproductivityof whiteclover-perennialryegrassmixtures.Theor.Appl.Genet. 77,65–67
37.Hill,J.andMichaelson-Yeates,T.P.T.(1987)Effectsof competi-tionupontheproductivityofwhiteclover-perennialryegrass mix-tures.Seasonaltrends.PlantBreed.99,251–262
38.Suneson,C.A.(1956)Anevolutionaryplantbreedingmethod. Agron.J.48,188–191
39.Döring,T.F.etal.(2011)Evolutionaryplantbreedingincerealsinto anewera.Sustain3,1944–1971
40.Allard,R.W.andHansche,P.E.(1964)Someparametersof pop-ulationvariability&theirimplicationsinplantbreeding.Adv.Agron. 16,281–325
41.Patel,J.D.etal.(1987)Naturalselectioninadoublehaploid mixture&acompositecrossofbarley.CropSci.27,474–479 42.Sanderson,M.A.etal.(2004)Plantspeciesdiversityand
man-agementoftemperateforageandgrazinglandecosystems.Crop Sci.44,1132–1144
43.deMazancourt,C.etal.(2013)Predictingecosystemstabilityfrom communitycompositionandbiodiversity.Ecol.Lett.16,617–625 44.Jiang,L.andPu,Z.(2009)Differenteffectsofspeciesdiversityon temporalstabilityinsingle-trophicandmultitrophiccommunities. Am.Nat.174,651–659
45.Loreau,M.anddeMazancourt,C.(2013)Biodiversityand eco-systemstability:asynthesisofunderlyingmechanisms.Ecol.Lett. 16,106–115
46.Prieto,I.etal.(2015)Complementaryeffectsofspeciesand geneticdiversityonproductivityandstabilityofsowngrasslands. Nat.Plants1,15033
47.Cardinale,B.J.etal.(2012)Biodiversitylossanditsimpacton humanity.Nature486,59–67
48.Gross,K.etal.(2014)Speciesrichnessandthetemporalstability ofbiomassproduction:ananalysisofrecentbiodiversity experi-ments.Am.Nat.183,1–12
49.Cardinale,B.J.(2013)Towardsageneraltheoryofbiodiversityfor theAnthropocene.ElementaPublishedonlineDecember4,2013. http://dx.doi.org/10.12952/journal.elementa.000014 10.3410/ f.718201253.793488195
50.Violle,C.etal.(2007)Lettheconceptoftraitbefunctional!Oikos 116,882–892
51.Violle,C.andJiang,L.(2009)Towardsatrait-basedquantification ofspeciesniche.J.PlantEcol.U.K.2,87–93
52.Hughes,A.etal.(2008)Ecologicalconsequencesofgenetic diversity.Ecol.Lett.11,609–623
53.Vellend,M.(2006)Theconsequencesofgeneticdiversityin com-petitivecommunities.Ecology87,304–311
54.Cadotte,M.W.etal.(2011)Beyondspecies:functionaldiversity andthemaintenanceofecologicalprocessesandservices.J. Appl.Ecol.48,1079–1087
55.Bolnick,D.I.etal.(2011)Whyintraspecifictraitvariationmattersin communityecology.TrendsEcol.Evol.26,183–192 56.Jung,V.etal.(2010)Intraspecificvariabilityandtrait-based
com-munityassembly.J.Ecol.98,1134–1140
57.Adler,P.etal.(2011)Productivityisapoorpredictorofplant speciesrichness.Science233,1750–1753
58.Jiang,L.etal.(2009)Speciesdiversityandproductivity:whydo resultsofdiversity-manipulationexperimentsdifferfromnatural patterns?J.Ecol.97,603–608
59.Mittelbach,G.G.etal.(2001)Whatistheobservedrelationship between species richness and productivity? Ecology 82, 2381–2396
60.Chapin,F.S.etal.(1993)Evolutionofsuitesoftraitsinresponseto environmentalstress.Am.Nat.142,S78–S92
61.Reich,P.B.(2014)Theworld-wide‘fast-slow’planteconomics spectrum:atraitsmanifesto.J.Ecol.102,275–301 62.Vasseur,F.etal.(2012)Acommongeneticbasistotheoriginof
theleafeconomicspectrumandmetabolicscalingallometry.Ecol. Lett.15,1149–1157
63.Finn,J.A.etal.(2013)Ecosystemfunctionenhancedbycombining fourfunctional typesofplantspeciesinintensivelymanaged grasslandmixtures:a3-yearcontinental-scalefieldexperiment. J.Appl.Ecol.50,365–375
64.Eviner,V.T.andChapin,F.S.(2003)Functionalmatrix:a concep-tualframeworkforpredictingmultipleplanteffectsonecosystem processes.Annu.Rev.Ecol.Evol.Syst.34,455–485 65.Schenk,H.J.(2006)Rootcompetition:beyondresource
deple-tion.J.Ecol.94,725–739
66.Wolkovich,E.M.andCleland,E.E.(2011)Thephenologyofplant invasions:Acommunityecologyperspective.Front.Ecol.Environ. 9,287–294
67.Wolkovich,E.M.etal.(2014)Progresstowardsaninterdisciplinary scienceofplantphenology:buildingpredictionsacrossspace, timeandspeciesdiversity.NewPhytol.201,1156–1162 68.MacArthur,R.H.andLevins,R.(1967)Thelimitingsimilarity,
convergenceanddivergenceofcoexistingspecies.Am. Nat. 101,377–385
69.Zuppinger-Dingley,D.etal.(2014)Selectionforniche differentia-tioninplantcommunitiesincreasesbiodiversityeffects.Nature 515,108–111
70.Grime,J.P.(2006)Traitconvergenceandtraitdivergencein her-baceousplantcommunities:mechanismsandconsequences.J. Veg.Sci.17,255–260
71.Navas,M-L.andViolle,C.(2009)Planttraitsrelatedto competi-tion:howdotheyshapethefunctionaldiversityofcommunities? Com.Ecol.10,131–137
72.Mayfield,M.M.andLevine,J.M.(2010)Opposingeffectsof com-petitiveexclusiononthephylogeneticstructureofcommunities. Ecol.Lett.13,1085–1093
73.Grime,J.P.(1998)Benefitsofplantdiversitytoecosystems: immediate,filterandfoundereffects.J.Ecol.86,902–910 74.Garnier,E.etal.(2004)Plantfunctionalmarkerscaptureecosystem
propertiesduringsecondarysuccession.Ecology85,2630–2637 75.Enquist,B.J.etal.(2015)Scalingfromtraitstoecosystems: developingageneralTraitDriverTheoryviaintegratingtrait-based andmetabolicscalingtheories.Adv.Ecol.Res.52,249–318 76.Norberg,J.etal.(2001)Phenotypicdiversityandecosystem
functioninginchangingenvironments:atheoreticalframework. Proc.Natl.Acad.Sci.U.S.A.98,11376–11381
77.Savage,V.etal.(2007)Ageneralmulti-trait-basedframeworkfor studyingtheeffectsofbiodiversityonecosystemfunctioning.J. Theor.Biol.247,213–229
78.Vile,D.etal.(2006)Ecosystemproductivitycanbepredictedfrom potentialrelativegrowthrateandspeciesabundance.Ecol.Lett.9, 1061–1067
79.Milla,R.etal.(2015)Plantdomesticationthroughanecological lens.TrendsEcol.Evol.30,463–469
80.Balvanera,P.etal.(2006)Quantifyingtheevidencefor biodi-versityeffectonecosystemfunctioningandservices.Ecol.Lett. 9,1146–1156
81.Hector,A.etal.(1999)Plantdiversityandproductivityexperiments inEuropeangrasslands.Science286,1123–1127
82.Lamanna,C.A.etal.(2014)Functionalspaceandthelatitudinal speciesrichnessgradient.Proc.Natl.Acad.Sci.U.S.A.111, 13745–13750
83.Gravel,D. et al. (2011) Experimental niche evolution alters the strength of the diversity-productivityrelationship. Nature 469,89–92
84.Violle,C.etal.(2009)Competition,traitsandresourcedepletionin plantcommunities.Oecologia160,747–755
85.Grace,J.(1990)Ontherelationshipbetweenplanttraitsand competitiveability.InPerspectivesonPlantCompetition(Grace, J.andTilman,D.,eds),pp.51–65,AcademicPress 86.Violle,C.etal.(2014)Theemergenceandpromiseoffunctional
biogeography.Proc. Natl. Acad.Sci. U.S.A. 111,13690– 13696
87.Henderson,C.R.(1975)Bestlinearunbiasedestimationand pre-dictionunderaselectionmodel.Biometrics21,423–447 88.Hazel,L.N.(1943)Thegeneticsbasisforconstructingselection
indices.Genetics28,476–490
89.Soussana,J-F.etal.(2012)GEMINI:Agrasslandmodelsimulating theroleofplanttraitsforcommunitydynamicsandecosystem functioning.Parameterizationandevaluation.Ecol.Model.231, 134–145