The virtual fruit: towards a tool to design ideotypes for sustainable production systems

Transcription

1 The virtual fruit: towards a tool to design ideotypes for sustainable production systems Bénédicte Quilot Mohamed Ould Sidi Abdeslam Kadrani Michel Génard Françoise Lescourret Nadine Hilgert MISTEA

2 Context Controlling storage disease is a priority to reduce fruit residues and increase safety Critical question : finding the best combinations of genetic resources and cultural practices adapted to, and respectful of specific environments

3 Monilinia brown rot caused by Monilinia species, can provoke as much as 30 to 40% of crop losses no other alternative than chemical treatment is available : fungicide applications are generalized and occur till pre-harvest all cultivated peaches are more or less sensitive to brown rot mechanisms of resistance may be complex and largely linked to fruit characteristics including fruit craking Monilinia laxa

4 Cracking and practices cuticular cracks Monilinia laxa Fruit mass (g) 250 Low load S cracking (cm 2 ) Trees in container High load Days after bloom

5 Cracking and practices cultivars crack occurrence sensitivity growth and quality characteristics thinning irrigation Practices Genotype humidity temperature radiation Climate fruit growth (carbon and water) Models sugar partitioning infection transpiration cuticular cracking

6 Virtual fruit Leaf water potential Temperature Stem water potential State of the system Radiation fruit fresh mass proportion of flesh flesh dry matter content Relative humidity Carbon photosynthesis stem carbon balance stone and flesh dry growth C flow Sugars sucrose, sorbitol, glucose and fructose amounts of sugars Water osmotic pressure turgor pressure tissue plasticity flesh dry mass stone dry mass C reserves Submodel variables Respiration sustain and growth C0 2 emission Temperature 4 sugars content Transpiration skin conductance Skin microcracking cuticule extension rates Matter or information fluxes Main outputs Environmental inputs H 2 0 loss microcracking area Adapted from Bertin et al, 2010

7 Virtual fruit Fruit growth variations between tree load 6 fruits 1 fruit Fruit mass (g) Cuticular cracking for different scenarii Days after bloom Sucrose accumulation for different tree loads 25 Low load S cracking (cm 2 ) 30 leaves/fruit 18 leaves/fruit 6 leaves/fruit Irrigated 15 Water stressed 5 High load 0 Days after bloom Days after bloom

8 Sensibility analysis 55 parameters 6 genotypic parameters involved in different processes fruit mass sweetness crack density Cuticular conductance Photosynthesis and vegetative growth Fruit dry matter growth and fruit photosynthesis Sugar accumulation Stone-pulp partition Water fluxes parameters

9 Genetic parameters PARAMETERS DEFINITION PROCESSES REFERENCE MINI MAXI coeff.transfo1 Part of the leaves in the structural part of the leafy shoots Vegetative structure * MuF Initial relative fruit growth rate Fruit demand ** ca1 FSsat cut.be Lx Proportion of carbon as sucrose in phloemic sap Parameter of the relationship between stone dry mass and total fruit dry mass Parameters of the equation of the relative elongation rate of the cuticle Conductivity of the composite membrane for water transport Sugar metabolism Stone-pulp partition *** ** Transpiration * Water fluxes ** * ± 10 % variations ** Observed variability within BC2 population (Quilot et al 2005) *** Extreme values from litterature

10 Criteria to optimize CRITERIA OPTIMISATION MINI MAXI Fruit mass (g) maximiser Sweetness (%) maximiser 4 20 Crack density (%) minimiser 0 20

11 8 scenarii CLIMATE IRRIGATION FRUIT LOAD 4 fruits/stem LC Well irrigated (WI) 20 fruits/stem HC Avignon 4 fruits/stem LC Water deficit (WD) 20 fruits/stem HC same for Bordeaux leaf water potential (MPa) days of simulation

12 Optimization NSGA-II : a multiobjective evolutionary algorithm, a Pareto approach Non-dominated sorting 100 loops P t + 1 P t 40 ind F_1 F_2 F_3 Crowding distance sorting Selection Mutation recombination Q t criteria Rejected R t Evaluation using the Virtual fruit model Dominance relation between solutions A large choice of solutions: 40 optimum genotypes over 4000 tested per scenario

13 Results AV_WI_LC AV_WI_HC AV_WD_LC AV_WD_HC Criterion are contradictory CrackDensity alternative actual Sweetness FruitMass

14 Parameter correlations Optimum are concentrated in the space No correlations between parameters AV_WI_LC AV_WI_HC AV_WD_LC AV_WD_HC coeff.transfo1 FSsat cut.be ca1 Lx FSsat cut.be ca1 Lx mu.fruit

15 AV_WI_LC AV_WI_HC AV_WD_LC AV_WD_HC Complex interactions Direct influence of Lx Criteria and parameters coeff.transfo1 mu.fruit ca1 FSsat cut.be Lx Crack density Sweetness Fruit mass (g)

16 Conclusions Lx FSsat Specific genotypes for high crop load High variability of ideotypes for WD_LC mu.fruit ca1 oeff.transfo1 cut.be AV_WI_LC AV_WD_LC AV_WI_HC AV_WD_HC

17 Perspectives explore genotypes selected in Bordeaux climate test the best genotypes of one scenario in the other scenarii: - are their genotypes good in different scenarii? try to develop methods from Operations Research and compare results with those from Artificial Intelligence methods (NSGA)

18 Thank you for your attention