Which food web model in agriculture ? Bridging the gap between agronomists and food web ecologists Philippe TIXIER CIRAD, UPR26, Martinique, France Networks and interactions in agriculture Dijon, France, September 28th, 2012
[email protected]
Outline • Which models to link food web mechanisms and agricultural performances? • Integrate ecosystem functioning knowledge in predicting models to explore management options • How to deal with complex webs and data issued from new empirical methods e.g. molecular tools?
Expected feedbacks of food webs on agroecosystem functioning
• Pest regulation – top-down – Bottom-up
• Biogeochemical cycles • Indirect/non trophic effects – Soil structure – Pollination
Changes in agroecosystem view Intra-Guild predator Intra guild predation
Predator
Pests
Cultivated crop
Predator 3 Predation / parasitism: • Biological control Prédator 2 • Natural ennemies Top-down effect Apparent competition
Alternative prey
Prey 2
Associated plants Bottom-up effect
What should we expect by taking into account food webs in agroecosystems? • Clarify and hierarchize the role of species or trophic groups in biological regulations • Optimize beneficial feedbacks on ecosystem functioning • Propose innovative management options • Maximize ecosystem services
What type of food web model? • Dynamic models – Population models: Lotka-Volterra based … huge theoretical analysis of simple models – Generalized population models – Mass balance models (e.g. Ecopath-Ecosim)
• Descriptive and structural models – Cascade, niche models (Williams and Martinez, Nature 2000; Allesina et al Science 2008…)
– Motif detection (Milo et al Science 2002…)
Application to a concrete management question • How the addition of a cover crop (a new primary resource) alters – food web structure – functions, e.g. pest regulation – ecosystem services ?
Case of cover crops in banana agroecosystem
How the cover crop alter the food web? D
?
Régulation ?
C
C
?
? Cosmopolites sordidus
A
B
B
A
? Banana
Banana
Cover crop
Quantification of processes before designing a predictive model
Does the new primary resource is consumed ? araignées araignées carabes carabes scolopendres scolopendres forficules forficules fourmis fourmis litière banane litière banane litière Bracharia C. sordidus C. sordidus
10.00
8.00
δ15N
15 δδ15 N N (‰) (‰)
6.00
4.00
Banana
Cover crop
2.00
0.00 -28.00 -26.00 -24.00 -22.00 -20.00 -18.00 -16.00 -14.00 -12.00 -10.00
δ13C (‰)
δ13C Banana with bare soil
à Banana with cover crop
à Shift of the trophic niche of generalist predators: consumption of the cover crop primary resource Duyck et al. 2011 Basic & Applied Ecology
Does the addition of a new primary modify predators abundance? Solenopsis geminata Bare soil
Cover crop
5 3mes more ants in cover crops Site 1 Site 2 Mollot et al. 2012. Agricultural & Forest Entomology
Does the addition of a new primary modify the predation of the pest? Predation rate sordidus eggs Preda3on rate of by C. ants
Bare soil
Cover crop
2 to 7 3mes more preda3on with cover crop
Site 1 Site 2
Mollot et al. 2012. Agricultural & Forest Entomology
P1 Ants
H1 Weevils
H2 Alternative prey
Multi-cycle banana growth model
Cover crop growth model
Nitrogen balance model
CROP MODEL FOOD WEB MODEL
P2 Spiders
Linking a food web model (LotkaVolterra based)
with a cropping system model (Tixier et al 2008 Agr Syst)
Tixier et al Submitted
Effect of cultural practices on ecosystem services 25
Agricultural performances
Yield
§ Fertilization level + +/- § Cover crop management (mowing)
15 120
Mean weevil populations
Pest populations
100
5
20 0 0
5
-1
60 1400 10 15 20 25 1200 Mowing40interval (weeks)
5
1000
-1
0
800
tons ha y
0
80
600 10
Primary productivity Mowing interval vs cumul H1 1.0
Productivity
Mowing interval vs cumul H1 Mowing interval vs cumul H1 1.4
15
20
25
1,2
Mowing400 interval (weeks)
Arthropod diversity
1,0
200 0 0
5
10
Shanon index
10 Abundance
tons ha
-1
20
Biodiversity
0,8
15 20 0,6 Mowing interval (weeks) 0,4
25
0,2 0,0 0
5
10
15
20
Mowing interval (weeks)
25
Tixier et al Submitted
Generalize this framework to link agronomists and community ecologists models • Use the best of agronomic models: crop models: – which accurately describe the soil-plant relations regarding water and nutrients
• Linked with food web models: – which adequately describe interactions between communities associated with the crop
Top predator
Agricultural management
• Biological control
Parasitoids
Pests
• Pesticide application
• Biomass exportation • Pruning
Generalist predator
Herbivores
Cultivated plants
Associated plants
Photosynthesis
Photosynthesis
Detritivores
Bacteria & fungi
Food web model Crop model
• Planting
Soil • Fertilization
Mineral nutrients
Organic matter • Tillage
Physical properties
Tixier et al Submitted
How trophic groups may alter soil-plant processes?
Tixier et al Submitted
But in reality… webs are much more complex
Unraveling the quantitative banana food web by DNA barcoding • Barcoding of plant DNA (trnL chloroplastic gene) and animal DNA (mini-CO1 mitochondrial gene) • 928 individuals sampled • 753 consumption links at the individual scale, corresponding to 33 taxa (mostly identified at species or genus level) • 98 trophic links identified
The complex quantitative banana food web unraveled by DNA barcoding
Mollot et al submitted
A generlized Lotka-Volterra food web model? s s dX i = ∑ q Xi a XiXi−1 X i X i −1 − ∑ a Xi+1 Xi X i +1 X i − u Xi X i dt j =1 j =1
S number of species per trophic level aXY per capita consumption rate qX conversion efficiency uX death rate KHiRi carrying capacity of Hi supported by resource Ri rHi growth rate of Hi
It becomes possible to establish real adjacency matrix
Mollot et al submitted
But the whole community modelling has limits… 800
weevil cover crop beatle myriapod earwig ant spider
700 600 500
Kg ha-1
• Sensitivity of errors in adjacency matrix on final simulations • Biological parameters not available (conversion efficiency…) • Data demanding for validation
400 300 200 100 0 1
201
401
601
801
1001
Days after planting
1201
1401
1601
Simplify the web using functional trait/
trophic group approach Example of the arable ecosystem model (Caron-Lormier et al Ecol. Model. 2009)
Link structure to functions using motifs detection
From: Stouffer & Bascompte 2010. Eco. Lett.
Occurrence of motifs to assess whole web functions Analyze the position and the frequency of species of interest (pests) in motifs
Example of the position of the banana weevils in size 4 motifs 1
2
A) Bi-fan
B) Generalism
97
11
0
0
56 16
0
C) Vulnerability
76 36
0
2
16
à Higher occurrence of weevils inside the quantitative motifs was always associated with the higher interaction strength Mollot et al To be submitted
Key messages • The perfect food web model for agricultural issues does not exist! • A combination of different approaches is needed to address the range of agricultural issues – Theoretical à general rules – Simple dynamic models in link with soil-plant functioning à explore management options – Structural models and motifs approaches à better description of complex webs
• New methods to measure trophic links (molecular tools) offer great perspectives but taking account all these information in predictive models remain an issue
Thank you for your attention … and many thanks to my colleagues: Pierre-François Duyck, Grégory Mollot, Elsa Canard, Raphaël Achard…