New Ag International JUN/JUL 2020
The BioSuccess app is the combined efforts of CAB International (CABI), an organization founded in 1910 in the UK, and Assimila an Earth observation company that creates bespoke software for modelling. The partnership came about through a brainstorming event and, after a short preliminary project, has now resulted in a working app that could help fight against locust outbreaks in China using a mycoinsecticide.
In China, the government takes responsibility for locust monitoring and management through the National Agro-tech Extension and Service Center (NATESC). In 2004, China suffered from a plague of locust that destroyed nearly one million hectares of farmland across 10 provinces and municipalities, and reportedly affected millions more hectares of farmland.
Locust lay their eggs in undisturbed areas of grassland, such as reed beds, pasture and desert, and emerge as flightless hoppers, also called nymphs (see box).
“We wanted to help NATESC control locusts with mycoinsecticides before they move to farmland,” noted Luke. “When locust are juveniles, they can’t fly. So, we need to control them before they become adults.”
One of the drivers for the project is that China is looking to prevent the increase of the use of chemical pesticides beyond 2020. This is an official objective that also applies to its consumption of fertilizers.
This has led to research involving biocontrol applications, and specifically a mycoinsecticide – a Metarhizium fungus that kills an insect by growing inside its body. Whereas a chemical control will kill in a matter of hours, the biocontrol will take seven to 14 days to kill a hopper on average, according to Luke, and this will largely be dependent on temperature.
The association with the ground temperature is really where Assimila – a specialist in Earth observation and remote sensing founded in 2006 – comes into the story.
Into the sandpit Assimila’s modelling experience and CABI’s biocontrol knowledge and expertise came together as part of a three-year project with £1.6 million of funding from the UK’s Science and Technology Facilities Council (STFC) and Newton Agri-Tech Fund. The project ended in June 2019 and involved seven partners in total, three from China and four from the UK.
Before the three-year project, there was a one-year project involving a number of institutions from the UK and China. During this project, there was a “sandpit” exercise, which was the brainstorming event that brought Assimila and CABI together. The one-year project, which focused on the control of wheat rust and locust, was essentially the proof-of-concept that led to the three-year project, which ended with two models that predicted how quickly a locust reaches adulthood and after application how quickly a mycoinsecticide will kill locusts. There was interest from NATESC to turn this information into an in-field useable app. This led to further funding from the EC’s Copernicus Climate Change programme (C3S) to develop a C3S use case for “BioSuccess” and to allow the team to focus on the market suitability of developing the app into a commercial product.
Model behaviour When talking about how Assimila handles the data, Cornelius describes it in terms of making the data operational, which means it can be interrogated.
“The data that drives our models is the ‘surface temperature’ parameter provided as part of the European Centre for Medium-range Weather Forecasts’ (ECMWF) numerical weather prediction system,” said Cornelius, who studied environmental science and remote sensing. “This surface temperature is not a direct observation from satellites or by people in the field but is actually an output of their modelling system. This system is a hugely complicated but accurate mathematical model of the surface, ocean and atmosphere, which is constantly being calibrated and validated by observation from weather stations or satellites. It calculates over 300 different meteorological parameters in ~28km regular grid cells globally. This is the huge advantage of using climate reanalysis data as it has consistent distribution of data across the world, compared to the inconsistent coverage of traditional Earth observation data.
“Reanalysis data is also provided every hour and not interrupted with cloud cover like a lot of satellite observations. This makes it perfect for accurate biopesticide calculations,” Cornelius added.
Showing the province of Shangdong, the colours move from red through green and into blue as you move inland, indicating regions that would be more suitable for the biopesticide, as it would kill the locust population more effectively.
Inside the app The first thing to notice about the BioSuccess app is the user-friendly interface, designed to be used by non-scientists. The map of China is covered in a grid, with each square representing 27.75 x 27.75 km. Cornelius indicated that soon a 9 x 9 km grid would be possible with the utilization of the newly released enhanced ERA5-Land product. Under the section for historical analysis in the app, there is data stretching back to 1997. This is where Assimila’s work has created a continuous series for the ground temperature using the ECMWF data, which in simple terms takes discrete pieces of data and merges them into a continuous time series.
Cornelius explains the map on the previous page. “The coloured pixels on the map show how effective the biopesticide would be at controlling a locust population in those areas, given the estimate of the ground temperature after the spray date. In this image, the province of Shangdong is partially covered in red squares, showing areas with environmental conditions not favourable for the biopesticide. These regions could be susceptible to an infestation if the biological agent was solely relied on, as it wouldn’t work well. In this time period, the colours move from red, through green and into blue as you move inland, indicating regions which would be more suitable for the biopesticide, as it would kill the locust population a lot more effectively.”
The prescriptive part of the app, in the “current analysis” section, is its ability to predict the kill rate of the biocontrol at a given ground temperature at a specific location, and then return the number of days the biocontrol will take to kill 90 percent of the locust population.
It is important to be clear about what the app is “predicting”’. As Luke explains, it is not an actual measurement showing that a given percentage of locust have been killed. Instead it is giving guidance as to how long the biopesticide will take to kill 90 percent of the locusts.
The interface of the BioSuccess app.
The surface temperature will indicate the speed of operation of the biocontrol, and CABI’s experiments tell the effectiveness – in other words the “time to kill” – of the biocontrol at those temperatures.
The historical data is more for NATESC, for the organization to decide on how best to advise local pest control officers on whether to use a biocontrol or chemical pesticide, noted Luke. Their resources are limited, so the advice for blue squares would likely be the use of biological products.
The current analysis will be of more interest to users on the ground in a specific location. This will plug into the forecast, said Cornelius, and will effectively answer the question – “ If I spray tomorrow, how long will it take the locusts to die?”
This is where another piece of the jigsaw comes in – the life development model (LDM). The algorithm employed by Assimila will combine the estimation of ground and soil temperature with the LDM to produce an estimation of the lifecycle stage. This is then meshed together with the model for biocontrol efficacy, which gives the life stage of the locust by the time of the biopesticide’s completion. This helps pest control officers use the biocontrol to prevent pests reaching the more damaging adult stage. A potentially very useful add-on for pest control officers that has been developed within the BioSuccess tool is the “last chance” tool, which returns the last date that biopesticide can be sprayed to prevent the locusts reaching this adult stage.
Ground control The three inputs – ground temperature, LDM and biocontrol efficacy – are combined in a “data cube” to help find the optimum time to apply the biocontrol given an indicated ground temperature in a locust breeding ground.
The aim is to prevent the locust getting to the 6th stage in their growing cycle, which is when they get wings. So ideally you want to control the locust while they are still hoppers on the ground. But there are a couple of complications. The first is if a warmer period is coming, the hoppers will develop faster. That’s why it is desirable to have data linked to a weather forecasting model. Added to this, there are early and late hatchers.
“It’s a bit like a race – the biocontrol product likes warmer temperatures, but that will mean the locust develop faster,” said Luke.
Ground truthing CABI has conducted experiments to measure the speed and effectiveness of the biocontrol product in situ at various ground temperatures. Luke refers to this as “ground truthing”.
While the ECMWF reanalysis will give an estimate of the ground temperature and the LDM an estimate for the speed of development of the locust given the ground temperature, it is not possible to extrapolate the total numbers of locust involved. The CABI contribution is essential in estimating the kill rate of the biocontrol at a given temperature, and therefore placing the final piece of the jigsaw in estimating the optimum time to apply and maximize the kill rate percentage.
Building on the previous work CABI and Assimila have done in this field, getting to a working BioSuccess app was the objective of Phase 1 of the current C3S BioSuccess use case project. Its use will be by NATESC advisers in the first instance, although there is a much broader market potential, according to Luke, such as distributors in China offering the app as a service. The app also has great potential in assisting biopesticide use for other pests, not just locust, and in other regions, not just China. Spreading the word of the efficacy and practical benefit of the app is the current challenge for the team, concluded Luke.
Locust are herbivorous insects that belong to the same order of insects as grasshoppers, katydids and crickets – Orthoptera. Depending on location, locust can produce two generations per year.
The desert locust (Schistocerca gregaria) typically leads a solitary lifestyle in west Africa and India, and will breed after periods of rain.
The female needs moist soil to lay her eggs in the ground, burrowing to 50-60 mm. In China, reed beds make common breeding sites, or undisturbed pasture or grassland. In general, the eggs hatch about two weeks after they were laid, depending on temperature and moisture.
The flightless juveniles, called nymphs or hoppers, go through five moulting stages, known as an instar, to become an adult. The instar stages can take between one to two months and will depend on environmental factors such as ground temperature. After the fifth instar the wings are developed. The locusts are now known as “fledglings” but they cannot fly just yet – it takes another seven days for their bodies to harden to make them capable of flight.
During development, locusts will continually feed to store energy for flight and reproduction. The swarming effect does not always occur. If food supply is sufficient for an individual, the locusts will live their lives separately, as do grasshoppers, and remain a green colour. Crowding triggers a change in behaviour, known as gregarisation, and this results in a change of colour – becoming the familiar black and yellow.
When swarms migrate, females will lay eggs when they land.
When it comes to the control of locust, the early instar phase is when the locusts are most vulnerable. But they are only in this phase for a few weeks and hoppers will be emerging in waves, so any control would ideally work for both early and late stage hoppers.
Insecticides must either be ingested by the locusts or have contact with the outside of the body. Putting insecticide on a food, as prepared bait, is one method, however this requires good knowledge of where the locust have laid eggs, and monitoring on the emergence of hoppers.
When spraying an insecticide directly on locusts, an oil is used to help penetrate the cuticle. Spraying can be done from the ground and also from the air. Products sprayed from an aircraft are at ultra-low volume (ULV). Various countries issue advice on locust control – in New South Wales, Australia, for example, fenitrothion is registered for use against locusts on pasture and a wide range of cereal and other crops. It is available as an emulsifiable concentrate EC (for ground control) and as a ULV formulation for aerial control. Fipronil, also a ULV formulation for aerial control, is advised for pasture and sorghum situations by the Department of Primary Industries, NSW government. Chlorpyrifos is suggested as suitable for EC ground control only in crop or pasture situations. Metarhizium, sold as Green Guard, is a biological control agent that can be used in environmentally sensitive areas and in areas of organic farming or chemical sensitivity, according to the department’s website.
In some regions of the world there can be a delay in transporting or procuring the necessary control products. This is time lost in combatting locust.
References
https://www.dpi.nsw.gov.au/climate-and-emergencies/locusts/chemicals/faq-insecticides https://www.nature.com/articles/d41586-020-00725-x https://www.agriculture.gov.au/ http://www.biology-resources.com/locust-01.html https://phys.org/news/2019-01-locusts-colours.html
At ABIM 2019, you presented information on a project at ZHAW on automated airborne pest monitor (AAPM). Could you give a brief overview of the main objectives and the duration of the project? The project Automated Airborne Pest Monitoring (AAPM) of Drosophila suzukii is a collaborative project between University of Aberdeen (UoA), Prof. David R. Green, Wageningen University and Research (WUR), Prof. Lammert Kooistra, and Zurich University of Applied Sciences (ZHAW). The project was aimed at developing an automatable monitoring system for pests. This should be exemplified in one of the major pests of soft berries and cherries named spotted wing drosophila (SWD), scientifically named Drosophila suzukii. The insect should be trapped by photographable traps. The imagery should be taken by
means of a UAV, and insects counted on the imagery using deep learning methodology.
What is the problem with current monitoring systems for Drosophila suzukii? Current monitoring of SWD is conducted with cup traps that are filled with a liquid lure consisting of a mixture of wine and vinegar. Traps have to be filled and hang in the field; the lure containing the caught flies is then emptied and analyzed in the lab once per week over the whole harvest period. Since the insect is polyphagous, it kind of hops from one ripe crop to the next. Therefore, monitoring season is relatively long starting in May with strawberries until ending in October with grape harvest. Therefore, the current method is very labour intensive. Producers seldom monitor their measures fighting the pest and, hence,
automatization would be highly appreciated.
In the AAPM project, a drone flies from base to the traps and takes images from close distance, then back to base.
How does your system work in stages? We bring special traps to the field. A drone flies from base to the traps and takes images from a close distance, then back to base. Currently, data is transferred manually from drone to computer and then analyzed automatically by the algorithm that was developed by my colleagues at WUR. However, several issues have to be solved in coming projects to make the system work automatically. To name the most important: (1) trap
efficiency needs to be increased; (2) autonomous drone flight needs to be more precise; (3) automated data transfer; and (4) integration into decision support systems (DSS) to deliver advice directly to farmers.
The software used to analyze images can “learn” – can you describe this process and why it is important? We used a deep learning approach called ResNet-18 to identify the small flies to the trap imagery. These methods need a high number of training imagery. In our case, we prepared about 250 images of traps and labelled all SWD on the imagery; in total 4,750 labels of male and female flies, and an additional 16,400 bycatch labels. Seventy percent of these labelled object were used to train the algorithm. That means the algorithm “studies” these objects and “learns” how to distinguish the
target objects. Another 20 percent were used for validation and, finally, 10 percent that have never been “seen” by the software were used to test the algorithm. The results showed a strong dependency on high quality imagery to be offered to the algorithm to receive a high umber of true positives.
One of the major pests of soft berries and cherries is SWD.
Does the software have the ability to extrapolate from the images and give a ranking for the level of pest infestation, e.g., low, medium, high? Up till now, we haven’t programmed anything like this. But at a later stage we will – for sure. Depending on trap efficiency, we may have only two classes (e.g. present/not present) or more classes like you suggested (e.g. no fly, some, medium, many flies). However, the relation of catch numbers to population density and/or infestation (egg deposition
in fruits) is almost impossible because the trap location has a very strong influence on catch rate.
Presumably, the number of trapped flies is the main parameter, and so I’m thinking it would be relatively simple to program the software, but are there other considerations based on the distribution of the numbers? That relates to the previous answer. More research is needed to make the relation between number of caught flies with the important information for farmers: infestation level and what is the proper measure. The latter could be a suggestion, taking into account the number of flies, weather forecast, crop status (crop maturity, days to harvest), available measure (e.g. insecticides, their retention times to harvest, biologicals), etc. This could be solved if the output from our
AAPM is fed into existing DSS. However, DSS are often not yet suitable because of lack of data (e.g. crop species specific phenology) and therefore there is still a lot of work to do.
In the AAPM project, a photograph of the trap was taken by means of a UAV, and insects identified and counted on the imagery using deep learning software.
What did you learn from the project? I recall “red” traps were optimum colour, plus you were trying to use off-the-shelf drones, but you thought the optics weren’t good enough. We could confirm with our data that red is the most attractive colour to attract SWD, but odour outweighs the visual stimulus. That was the reason why we developed a new photographable trap that contained an attractive lure in addition to the colour red. From the technical side, we’ve seen some difficulties while developing the new traps that worked well in the lab but were less attractive in the field, resulting in a poor catch rate.
Additionally, since the insect is rather small (2-4mm) high resolution imagery is needed to allow automatic identification. We aimed at using easy to handle off-the-shelf UAVs. However, the currently available camera systems do not deliver high quality imagery taken in short distances (50-80cm in our case needed). As mentioned above, there are still some issues to be solved.
Where next for the AAPM application? Do you have a project proposal in the pipeline? We always have plenty of ideas! However, funding is not yet secured. We would like to work further with the original idea of an automated pest monitoring system. Since time goes by, farmers adopted to SWD and monitoring is not as highly important as other pests popping up in the fields such as the brown marmorated stink bug (Halyomorpha halys), Japanese beetle (Popillia japonica), or vectors of diseases such as American grapevine leafhopper (Scaphoideus titanus) or the various leafhoppers and sharpshooters.
The colour red was discovered as being the best colour to lure SWD to the sticky trap. An odour was also used.
Dr. Gadi V.P. Reddy
Why do wheat stem sawfly and Hessian fly pose a particularly challenging threat to wheat crops? Wheat stem sawfly has been one of the topmost pests of wheat around the world. This pest is an internal borer so difficult to monitor and control. One of the main approaches of the management of wheat stem sawfly is planting of resistant solid-stemmed wheat varieties and some cultural practices. However, resistant wheat varieties have been not widely adopted by wheat growers and
reported to cause lower yield and protein levels. Cultural practices such as tillage partially reduces populations. Synthetic chemical insecticide-Thimet 20-G (an organophosphate insecticide) has recently been registered in Montana against wheat stem sawfly but this chemical poses many health and environmental risks. Being a stem borer, other insecticides cannot kill the larvae of this pest. There are two parasitoids, Bracon cephi and Bracon lissogaster available for the wheat stem sawfly but they are effective only when the sawfly population is low. So, to date there is no effective control method for wheat stem sawfly. On the other hand, Hessian fly is not a serious pest at this moment, but this situation may change at any time in the future.
When did your research begin and why? My research began in 2012 when Montana State University started an entomology and ecology unit at the research center in Conrad. There has been severe insect pest incidence in the Golden Triangle area of Montana. Therefore, growers and stakeholders wanted the research on these insect pests and their management in that region. Based on the request from the community, Montana State University started the entomology/ecology unit at Western Triangle Agricultural Research Center in Conrad in June 2012. I was appointed as associate professor of entomology/ecology and superintendent of that center. In 2017, I became full professor and head of the entomology/ecology unit at the research center.
Prior to this new, patented fungi, what methods were used to battle wheat stem sawfly and Hessian fly? Were these methods effective? Why or why not? None of the available control methods are reliable and effective in managing the wheat stem sawfly. This is because all stages of the insect occur inside the plant stem and emerged adult are short-lived, but emergence occurs for a long period of time. The most damaging stage of this insect is larvae and contact insecticides are not effective because the larvae and pupae are protected inside the plant stem.
What makes these patented fungi so successful in controlling wheat stem sawfly and Hessian fly? The patented fungus is specific to wheat stem sawfly (for that matter to Hessian fly as well). Added to this, the fungus can remain in the soil and recycle so growers may not find it necessary to repeat the application of fungus. Moreover, this fungus has no effect on non-target organisms such as pollinators, insect natural enemies and the environment.
Where does the patented fungi go from here? i.e. licensing through ag companies? Currently, Montana BioAgriculture Inc., from Missoula, Montana, is doing some initial work before it can be licensed and commercially available to growers. Once this fungus is available to farmers, they can apply this to manage the wheat stem sawfly not only in Montana but also at other places. This way farmers can save spending millions of dollars on insecticides and also save the environment.
Biotalys NV announced the results from more than 100 field trials with its first biofungicide, BioFun-1, which is on track to launch in the United States in 2022, followed by global market introductions.
In 2018, Biotalys demonstrated that BioFun-1 provided competitive and consistent protection against Botrytis cinerea when compared with commercial chemical fungicides and outperforming biologicals, in multiple crops and regions. The 2019 field trial program took place across the United States and key European countries and included more than 50 efficacy trials against major pests such as Botrytis cinerea and powdery mildew.
In solo applications, BioFun-1, provided protection against multiple pathogens in the majority (over 85 percent) of the trials compared to the untreated control. BioFun-1 showed a clean dose response curve, allowing dose rate modulation to adapt to the disease pressure conditions. Under severe disease pressure a higher dose rate provides comparable protection to the chemical reference without the challenge of residues for the growers.
In 89 percent of the trials, the IPM program with BioFun-1 in rotation with commercial fungicides performed on par with the standard chemical IPM program, resulting in comparable yield, fruit quality and post-harvest shelf-life while chemical residues were reduced by up to 68 percent.
The 2020 field trial program, already underway, spans more than 150 field trials in various crops and different environmental conditions in Europe, South Africa and the United States. This field trial program complements ongoing product safety studies that will support the registration dossiers on track to be submitted later this year in the United States and Europe, as well as the development and implementation of the product supply chain.
A sugarcane variety from the Saccharum complex, widely cultivated in Brazilian farms, when exposed to the attack of the sugarcane borer (Diatraea saccharalis), produces chlorogenic acid, a substance that works against the aggressive insect. That was the finding from a study by Embrapa in partnership with Brazilian and international research institutions published in the journal Industrial Crops & Products on ScienceDirect. The sugarcane borer is deemed the main pest to the crop in Brazil.
Metabolic analyses showed that the sugarcane variety SP791011, which was developed by the Sugarcane Technology Center (CTC) and is currently in the public domain, produced chlorogenic acid on its own when it was attacked by the sugarcane borer. In the same environment and in a control group, plants of the same variety that were not submitted to the herbivorous pest attack did not present high acid expression.
The substance is harmful to pests to several crops, such as maize, coffee and tomato, as it affects their development and neutralizes their economic impacts on farms. In the study conducted with sugarcane, which also involved experiments adding chlorogenic acid to the borer's diet at their caterpillar stage, Diatraea saccharalis showed quicker development at the pupal stage; however, it was associated with wing deformity at the moth stage when exposed to all acid concentrations.
The researchers say chlorogenic acid can be considered a natural biopesticide, and its production can be induced to develop more sugarcane-borer resistant sugarcane varieties.
a
STK Bio-Ag Technologies’ hybrid fungicide REGEV is now registered in Chile for the control of powdery mildew in tomatoes and apple scab in pome fruits. Label extension has also been submitted for gray mould, powdery mildew in grapes, brown rot and sour rot in stone fruits and Alternaria in cherries.
STK REGEV is the world’s first hybrid fungicide, and is currently used successfully in many countries and several regions of the world, with plans for a U.S. rollout this year and global expansion in the future.
STK REGEV is a ready to use pre-mix fungicide but with the added benefits of reduced chemical residues and much better resistance management due to its highly complex formulation of tea tree oil and Difenoconazole. This fungicide serves as a bridge, providing farmers an easy to use crop protection solution, thereby expanding the use of biologicals products for sustainable agriculture.
According to STK Country Manager for Chile, Alvaro Arroyo, REGEV will be a game-changer for Chilean growers, making their crops more environmentally-friendly and more exportable. “Many have experienced great success with STK’s TIMOREX GOLD, the botanical-based biofungicide. Now hundreds of tomato, apple, and soon grape and stone fruits growers, who are moving to healthier production, will be able to use REGEV and enjoy the benefits of biologics in the same easy way they use their current chemical pesticides,” said Arroyo.
