Monitoring, warning, and decision support in winter wheat

Publikation: Bog/antologi/afhandling/rapportRapportForskningpeer review

Background
Winter wheat occupies a large part of the conventionally cultivated area in Denmark, and farmers commonly treat against fungal diseases 2 - 3 times per season. At least 1-2 of these treatments are specifically aimed at controlling Septoria. When treating with fungicides in June, farmers often add an insecticide to the application in order to be sure to avoid problems with aphids, even though the number of aphids has not exceeded the given threshold. Therefore, treatments with fungicides in winter wheat may be the reason for unnecessary treatments with insecticides, and it makes sense to develop a decision support system targeting Septoria and aphids at the same time.

In an earlier project, a detailed, mechanistic ecosystem model (SeptoriaSim) was developed to simulate the growth of both winter wheat and Septoria and the damage caused by Septoria to winter wheat, depending on temperature, solar radiation, humidity, and rainfall. The output from this model was both the yield of winter wheat with and without fungicide treatment and the reve-nue of the treatments, taking sales price of wheat and treatment costs into account. The revenue of a treatment is an essential parameter when deciding on pesticide treatments, and the model was found to perform well in trials in 2015.

Objective
The overall objective of this project was to develop a combined decision support system for Sep-toria and aphids in winter wheat by improving SeptoriaSim and extending it with an aphid module. The work to fulfill this objective was split into four immediate objectives:
1. Develop new monitoring methods for the two pests
2. Provide a stronger scientific foundation for warning models against Septoria and aphids
3. Describe the spatial variation of aphids
4. Evaluate the reliability of the warning method/decision support tools in field trials.

Investigations
Four types of investigations were carried out with the aim of both providing data for calibration of the model and enforcing the scientific background:
1. Investigations of the background level of Septoria spores causing the initial infestation of winter wheat
2. Investigations of the growth of Septoria in the winter wheat leaves
3. Investigations of the population development of all three cereal aphids in winter wheat from late May to mid-July
4. Detailed growth analysis of winter wheat under field conditions.

In 2016 and 2017, controlled trials were carried out, where treatments against Septoria were timed by: 1 ) SeptoriaSim using the original calibration, 2) the well-known Danish decision support system Crop Protection Online (CPO), 3) a humidity model, and 4) a number of fixed predefined treatments. The controlled trials were supplemented with a few trials in farmers’ fields, where the farmers sprayed a small part of their fields according to SeptoriaSim.

Results
All controlled experiments showed positive gross effects of fungicide treatments. Concerning the net yield, which is most relevant for the farmers, all treatments showed very limited responses in 2016, ranging from averages of 3.97 hkg/ha in the triple standard scheme to 1.67 hkg/ha follow-ing the recommendations of CPO. Of the three models, CPO performed best in 2016 with the humidity model being second, but generally treating against Septoria in 2016 appeared not to increase net yields much.

In 2017, the average net yields of the Septoria treatments ranged between averages of 1.25 hkg/ha using the humidity model to 4.25 hkg/ha following SeptoriaSim. Following CPO gave the second best net yield, 3.2 hkg/ha. Following both the standard treatments and the three models gave higher average net yields in 2017 than the best performing one in 2016.

Four out of the seven experiments showed generally negative or neutral net yield responses to fungicide treatments, while the only ones showing clear positive treatments were the experiments with susceptible varieties from Flakkebjerg (Nakskov in 2016, and Hereford in 2017) and the medium susceptible variety Torp in Holeby in 2017. Both experiments with Sheriff in 2017 showed generally negative net yields. However, using the models produced slightly increased net yields in Sheriff in all cases, except in Flakkebjerg in 2017.

Looking at the averages of the 2016 treatments (across varieties), CPO performed best, the humidity model was number two and SeptoriaSim number three. In 2017, SeptoriaSim per-formed best, with CPO being second and the humidity model third. Generally, the net yields produced using the decision support systems were even to (susceptible varieties) or slightly high-er (resistant variety) than the ones produced when treating according the standard schemes with three treatments. This especially applies to SeptoriaSim and CPO.

Concerning the number of treatments, SeptoriaSim produced fewest treatments against Septoria in the trials in 2017 (average 1.80) and most treatments (average 1.33) in 2016. CPO behaved oppositely by triggering most fungicide applications in 2017 (average 2.25) and fewest in 2016 (0.67). Over the two years, SeptoriaSim and CPO triggered a very similar number of treatments (average 1.46 and 1.55, respectively), while the humidity model triggered 1.6.
The results from the trials in farmers’ fields show low and non-significant increases in yield and 1000-grain weight in the trial plots, and the Septoria treatment frequency was slightly lower in the trial plots than outside the trial plots. The aphid treatment frequency was the same in the trial plots as outside the trial plots.

The field data from the project was used to re-calibrate SeptoriaSim, and the recalibrated version was tentatively tested by repeating the projections carried out in May - June 2016 and 2017 to decide on timing of treatments. The number of treatments using weather files from 2016 reduced the average treatment frequency by 0.33 compared to the original version, while the new version reduced the average treatment frequency by 0.5 treatments in 2017. This means that the recali-brated version of SeptoriaSim would have released clearly fewer treatments than the originally used version, making it environmentally speaking clearly the most attractive model.

The aphid module of SeptoriaSim was only tentatively tested in field trials at five farmers’ field tests, where the average yield in the trial plots was even to the yield outside the trial plots. The average number of treatments against aphids was 0.2, which was reduced slightly by using the recalibrated version of SeptoriaSim. SeptoriaSim did not justify tank mixes at any time.
The results on the spatial distribution of aphids suggest that the initial density does not vary strongly between fields within a distance of 10 km.

Discussion and conclusion
The number of treatments is discussed in relation to national averages in number of treatments and/or Treatment Frequency Indices, and it is concluded that using all three decision support systems can reduce the number of Septoria treatments and that SeptoriaSim also can reduce the number of aphid treatments.

The results from the investigation of the spatial distribution of aphids suggest that it is possible to use regional estimates of initial aphid densities as input for the aphid module of SeptoriaSim.

Perspectives
This project has shown that a decision support system based on biological knowledge and projec-tions of economic revenue of treatments may produce net yields equivalent to or better than the empirically based CPO. One important difference between the two decision support systems is that SeptoriaSim does not tell the farmer to spray. Instead, it gives predictions of the economic revenue of treating, and then it is up to the farmer to decide whether to treat or not. This decision may depend on the farmer’s economic situation and his environmental attitude. This type of deci-sion support system is an innovation that has the capacity to avoid treatments that may be eco-nomically beneficial, but may only produce very low revenues that farmers do not regard worth the effort.

The model system of SeptoriaSim is a general model type that can be used for other crops and other pests. Thus, it will be relatively simple to add other winter wheat pests to the system, and a similar system can be established for oilseed rape and its insect pests without much effort.

Administrative perspectives
One of the immediate objectives of the project was to develop new monitoring systems for the two pests, Septoria and aphids, and the results might be transformed into actions, making it at-tractive for farmers to use a decision support system on these two pests.
With one or more well-functioning decision support systems and the result that necessary input parameters can be applied to larger geographical areas, it is possible to qualify the decisions concerning pesticide treatments in winter wheat. The idea is to establish a system consisting of:
1. regional reporters who monitor the basic input of initial aphid densities and Septoria spore influx for SeptoriaSim
2. the reporters enter their results to a central database
3. a central service runs SeptoriaSim weekly, based on the input from the database, and identi-fies regional demands for control operations
4. the central service emits regional alerts to the farmers.

Such a system might help farmers make qualified decisions concerning Septoria and aphid con-trol to the benefit of both farmers’ economy and the environment
OriginalsprogEngelsk
UdgivelsesstedS.l.
ForlagMiljøstyrelsen
Antal sider90
ISBN (Trykt) 978-87-7038-184-0
StatusUdgivet - 2020
SerietitelPesticide Research
Nummer186

    Forskningsområder

  • Beslutningsstøttesystem, Septoria, Bladlus

Se relationer på Aarhus Universitet Citationsformater

ID: 140698035