0,33 ha

~ 460 trees/ha6 m

4 m

(A)

Bare ground

(B)

Tarpaulin mulching

(C)

High weed cover

© A. Ratnadass - Cirad

Assessment of mango blossom gall midge management solutions

from in-silico experiments : overview of an on-going modeling approach

Isabelle Grechi1,3,*, Laurie Saint-Criq1,2, Christian Soria1,3, Alain Ratnadass1,3, Frédéric Normand1,3,

Paul Amouroux5, Frédéric Boudon2,3

1 CIRAD, UPR HortSys, F-97455 Saint-Pierre, La Réunion, France2 CIRAD, UMR AGAP, F-34398 Montpellier, France 3 University of Montpellier, F-34090 Montpellier, France5 Departamento de Fruticultura y Enología, Pontificia Universidad Católica de Chile, Santiago, Chile

* Corresponding author: isabelle.grechi@cirad.fr

2. The modelling framework

Mango (Mangifera indica L.), a major fruit production in tropical and subtropical regions, is facing many production constraints: mango yield is irregular across

years, fruit quality is heterogeneous at harvest, and mango tree exhibits phenological asynchronisms within and between trees that result in long periods with

phenological stages susceptible to pests and diseases. Among them, the mango Blossom Gall Midge (BGM, Procontarinia mangiferae (Felt)) is a major pest

of mango. It can cause significant yield losses by damaging mango inflorescences. Management solutions to improve fruit yield and quality while reducing the

use of pesticides are considered. A crop-pest model applied to the mango-BGM system is currently developed from experimental data and will be used for in

silico assessment of BGM management levers (e.g. soil mulching and manipulation of mango phenology). The on-going modeling approach is presented.

1. Introduction

► Model description

A mango-BGM model simulates the dynamics of inflorescence and BGM populations of an

orchard at a daily time-step during the period of mango flowering (Fig.1). The orchard is

structured into three patches (A, B and C) according to soil mulching treatments (Fig.2).

In a first approach, the model is defined at the patch scale and considers:

Inflorescence age-structured population dynamics within each patch, accounting for

natural development and BGM-induced mortality of inflorescences: steps to

BGM stage-structured population dynamics within each patch, differentiating the effect of

soil mulching treatments on the BGM life cycle for each patch: steps to

Orchard colonization and movements of BGM adult individuals between patches, at

constant rates or driven by inflorescence abundance in each patch: steps to

© A. Franck - Cirad

Fig.2. The experimental orchard

Movements of

endogenous females

between patches

soil

3rd instar larvae

falling to the soil

𝐿𝑡+𝑑𝑙

Adult females

𝑁𝑡

New adult

females emerging

from the soil

𝑁𝑡+𝑑𝑙+𝑑𝑝

Fig.3. 3-D representation of simulated vegetative and reproductive

architectural development and phenology of a mango tree over a

growing cycle using a functional-structural plant model (FSPM) [1]

Female

egg-laying

Egg and larval

survival and

development

Resource abundance

within a patch

Inflorescence

mortality due

to larval

damages

1

2

1

2

Fig.1. Schematic representation of the mango-BGM model (bold text: estimated

parameters)

30th International Horticultural congress, 12-18 August 2018, Istanbul, Turkey

soil mulching

1

3

2

Orchard colonization by exogenous females (𝝀𝒕)

Local BGM dynamic

within a patch

Pupation rate and pupal survival and development

3 Larval & Adult

survival to soil mulching treatments

Pupae

females

leaving the

patch

4

5

Fig.4. Simulated (blue lines) vs observed (black points) BGM dynamics,

assessed by trapped larvae, in each of the three patches A, B and C

► Model parameterization and simulations

• Model parameterization from existing knowledge [2] and experimental data [3] collected in

Reunion Island on Cogshall cultivar (see experimental design; Fig.2)

Survival to soil mulching treatment, 𝜇𝑆𝑀 = 𝜇𝑆𝑀1 𝜇𝑆𝑀2 : 1 for bare ground (patch A), 0 for

tarpaulin mulching (patch B), and 0.11 for high weed cover (patch C).

• Qualitative validation based on experimental data and model simulations (Fig.4).

© A. Franck - Cirad

4. Conclusion

• This first modeling approach at the population scale gave promising results. However, further investigations are required to

assess the benefits of i) considering inflorescence phenological stages, and ii) changing from a population to an individual-

based and spatially explicit modelling approach, using a mango FSPM [1] for instance.

• Furthermore, relying on the mango FSPM can be useful to assess the effects of cultural practices on mango tree flowering

and their indirect effects on BGM dynamics. Eventually, the mango-BGM model should be used for the design of

management solutions for a sustainable mango production.

References: [1] Boudon et al. (2017). Acta Horticulturae, 1160: 83-

90; [2] Amouroux (2013). PhD Thesis dissertation. University of La

Réunion; [3] Brustel (2018). Master Thesis dissertation. Purpan.

Virtual experiments will be performed with the mango-BGM model to assess the effects of BGM

management levers on flowering level and fruit yield, according to exogenous pest pressure.

Tested environmental and management levers and their corresponding variables will be :

- Exogenous pest pressure: 𝜆𝑡- Soil mulching treatment: 𝜇𝑆𝑀- Manipulation of mango phenology: 𝐼𝑏𝑡- Pesticide applications: 𝜇𝑎 (adult survival rate to pesticides)

Acknowledgements. This work has been carried out as

part of the Cirad DPP COSAQ and ECOVERGER project

30th International Horticultural congress, 12-16 August 2018, Istanbul, Turkey

3. Future prospects: virtual experiments

Dynamic of inflorescence bud burst

(𝐼𝑏𝑡, input from experimental data)

Inflorescence development & aging

(constant lifespan T)

Inflorescence dynamic can also be simulated from a FSPM

model (Fig.3)

Inflorescences

born on day d,

𝐼𝑡𝑑 :

• 𝐼𝑏𝑑 if d≤t <d +T• 0 otherwise

𝐸0 𝑓 𝒌,𝑁𝑡 , 𝐼𝑡

0 ≤ 𝑓 ≤ 1

𝜇𝑙

0

500

1000

1500

2000

2500

3000

3500without BGM

with BGM

01-July 01-August 01-Sept. 01-Oct.

BGM-induced

mortality

Number of inflorescences, 𝐼𝑡 = 𝑑 𝐼𝑡𝑑

© L. Brustel - Cirad

𝝁𝑺𝑴𝟏 𝝁𝑺𝑴𝟐

𝑷𝒎𝒐𝒗

1 − 𝑃𝑚𝑜𝑣

𝜇𝑝𝑃𝑝

𝐼𝑡𝑑𝑔 𝜸,𝑁𝑡 , 𝐼𝑡 ,

0 ≤ 𝑔 ≤ 1

Nu

mb

er

of tr

ap

pe

dla

rva

e

(A) Bare ground (B) Tarpaulin mulching (C) High weed cover