Use of dynamic models for disease management
in viticulture
Tito Caffi
Vittorio Rossi
Piacenza, 14 June 2013
The ques(on should not be if we have to use PPP or not but
how we can use them reasonably
33% with insec(cides 99% with fungicides
3.5 mln ha of vineyards
~ 70000 tons of fungicides (ac(ve substances!) are sprayed each year
97% of the cul(vated areas is treated at least once/season
Integrated Pest Management (IPM) Integrated produc(on / farming Organic farming
-‐ Dynamic concepts -‐ High level of organiza(on -‐ Decisional skills (both strategic and tac(c) -‐ Constant upgrade (research > prac(ce) -‐ Specific advices
A model represents a simplified simula(on of reality
Modelling must be based on a deep knowledge
of reality
weather host
pathogen
Models can: -‐ increase efficacy and speed of the decision-‐making process; -‐ help in understanding epidemic processes and elabora(ng protec(on strategies
Ø Fundamental: explain the behavior of a system on the basis of the cause-‐effect rela@onship with the influencing variables; Ø Dynamic: the state of the system in every moment can be quan@ta@vely defined and the changes described mathema@cally; Ø Weather-‐driven: the weather variables are the main inputs of the model; Ø Simula@ons and predic@ons tools: used for extrapola(on beyond measured (mes and spaces
Many modeling approaches…
our approach is:
Grape Downy mildew is a disease of major importance worldwide and fungicide applica(ons
are necessary to avoid yield losses
With calendar-‐based schedule there is a huge amount of input (some(me useless) A key role in the epidemic is played by the primary infec(ons
Fungicide applica(ons begin in late April 14-‐18 applica(ons/year
penetra(on
invasion
sporula(on
inocula(on
dispersal (splash)
Winter
Sprin
g – early
summer
Spring-‐summer
oosphere
oogonium
anteridium
oospores
dispersal (air)
zoosporangia
macrosporangia
Late summer -‐ autumn
zoospores
- pathosystem analysis - drawing of the relational diagram - collection of necessary information/data - mathematical relationships/algorithms - dynamic simulation
A new modelling approach
Time t Ungerminated spores
Time t +1 Spores germinated
Germina(on
T
LW
LLM
MMO
PMO
GEO
OLL
MMR
DOR
GER
t
T
R
T
VPD
RH
ZLL
R
ZGL
ZCI
LW
T
ISS
RHINC
T
RHSUR
T
T
LW
INF
LW
LW
LLM
MMO
PMO
GEO
OLLOLL
MMR
DOR
GER
t
T
RR
T
VPDVPD
RH
ZLL
R
ZGL
ZCI
LW
T
ISS
RHINC
T
RHSUR
T
RHSUR
T
T
LW
T
LW
INF
LW
LW
rainfall
January February March April May June July
Primary inoculum season
rainfall
January February March April May June July
Primary inoculum season
1/4
8/4
15/4
22
/4
29/4
6/
5 13
/5
20/5
27
/5
3/6
10/6
17
/6
24/6
1/
7 8/
7 15
/7
22/7
29
/7 0
10
20
30
40
50
60
70
R (m
m)
Oospore germination
Zoospore release
Zoospore dispersal
Infection
End of incubation
Dynamic simulation
2004 – 2007 15 vineyards 1999 – 2004
19 vineyards
1998 – 2002 5 vineyards
1996 – 2004 7 vineyards
1995 – 2007 45 vineyards
2004 – 2007 9 vineyards
2004 – 2007 5 vineyards
The model valida(on was performed in different epidemiological condi(ons (locali(es x years):
105 vineyards in Italy + 43 in Quebec (Canada) + 6 in Georgia (US)
2008 – 2009 43 vineyards
2010 – 2012 6 vineyards
Yes
No
No
74%
0%
Yes
10%
16%
Obs
erve
d
Expected
1 8 15 22 29 5 12 19 26 3 10 17 24 30 Aprile Maggio Giugno
Modena
0.0% 0.1% 0.9%
0.0% 0.0% 0.0%
2006
2007
3
3 6
6
2008
4 8 0.0%
0.2% 72.2%
Farm practice Model-based strategy
DM intensity on bunches at fruit set
Phytoiatric validation
-50%
-50%
-50%
The model was: detailed it follows, step by step, the infection process accurate it estimated correctly 96% of the total cases robust it estimated early and late infections and it operated well under different epidemiological situations
Model’s application within the real time warning system produced very promising result
disease control as good as farm practice
50% reduction of fungicide applications
Conclusions I
MoDeM A web-based system for real-time Monitoring and Decision Making for Integrated Vineyard Management
VineMan.org Integration of plant resistance, cropping practices, and biocontrol agents for enhancing disease management, yield, and biodiversity in organic European vineyards
PURE Pesticide Use-and-Risk reduction in European Farming systems with Integrated Pest Management
Innovine Combining innovation in vineyard management and genetic diversity for a sustainable European viticulture
International research
A Decision Support System: vite.net®
Manager
informa(on
upda(ng
advice
decision supports
vite.net®
Mediterranean vine mealybug
American grapevine leaTopper
Grape berry moth
Pests (synte@c risk indicators)
Summary
vite.net® - main dashboard
Detail Informa(on about how many oospores overcome dormancy
Informa(on about how (already mature) oospores
are germina(ng
Detail of each key event of the infec(on cycle
Informa(on about the efficiency of each infec(on and their incuba(on period
vite.net® - downy mildew primary infections
Mediterranean vine mealybug
American grapevine leaTopper
Grape berry moth
Pests (synte@c risk indicators)
vite.net® - fungicides protection
Support Before 1st leaf
unfolded
Yes No
After 1st leaf unfolded
Preventive treatment
Does the model simulates an infection?
Yes
Did you apply any fungicide?
Treatment is not justified
high low medium
Residual protection level
high low high
Infection risk
Treatment justified
No Yes
Are grapevines between flowering
and fruit set?
No
Infection risk
low Treatment suggested
vite.net® - decision tree
24
DSS
NT
1 48
Farm
Row
Reduction in the number of treatments
5 ha plot of cv. Barbera in north-western Italy DSS
9
5
Farm
No. treat. No. Xn
IPM 9,4 1,9
vite.net® 7 1,5
8 farms - 2012
No. treat. No. Xn
IPM 10,2 1,8
vite.net® 7,9 1,5
Downy mildew scheduling
Powdery mildew scheduling
Pilot application in organic: vitebio.net®
21 organic farms 2011-2012
Same protection against the disease
No. of treatment / season : 9 vs 12 - 24%
Kg of Cu++ / application: 0.56 vs 0.68 - 18%
Kg of Cu++ / season: 5.1 vs 8.1 - 37%
Cost of the disease management strategy: -195 €/ha
Mechanis@c (explanatory) models explain the rela(ons within a pathosystem and describe how the system changes over @me and
space, depending from external (weather variables)
These models produce simula@ons of plant disease epidemics and can be successfully used for decision making in plant disease management at different scales of complexity, from territorial
warning systems to precision farming
A Decision Support System based on such models was used in a wide range of environmental condi(ons and reduced the number of
fungicides applica@on and increased the ra@onale and the sustainability of the fungicide scheduling
Final conclusions
Thank you for your attention