Precision agriculture for the management of agripests

By Hazel Parry, Senior Research Scientist, CSIROMay 21st, 2021

Digital decision-support tools are now deployed in many aspects of agronomy, but how about for the management of weeds, pests and diseases?

A simulated pest population distribution in crop and non-crop habitat, NSW

 

In Australia, “digital agriculture” applied to weed, pest and disease (‘agripest’) control across multiple sectors has been estimated to have the potential to unlock $1 billion of economic benefits. Improved access to computers, powerful handheld devices, and the internet’s increasing reach into remote areas makes digital tools more accessible today than they have ever been.

However, underneath this headline dollar figure is a plethora of benefits that not only improve farm profitability, but also improve its underlying sustainability. Uptake of digital agriculture and the precision agriculture it enables is an opportunity to reduce the dependence on pesticides and their negative effects on ecosystems. In Holland, for example, precision or smart farming (along with innovative plant breeding) now forms the core of the Dutch Government’s vision for 2030 to protect biodiversity in agriculture.

So how exactly might this benefit be realised? The answer could be in integrated digital solutions that growers can rely on to protect their crops.

The transition to integrated pest management approaches

Integrated pest management (IPM) approaches only use chemicals as a back-up. Problems are avoided and management practices optimised through improved ecological understanding made available via decision-support tools. Such approaches are needed to break our reliance on broad spectrum chemical pesticides such as pyrethroids or glyphosate.

However, important barriers remain to this transition in terms of the requirement to ‘think beyond the farm’. Success requires multiple tactics to be employed that may interact in complex ways, and coordination of management actions both on-farm and in neighbouring habitats. Agripests rarely occur in isolation and the management strategies employed to control them are not always compatible.

Furthermore, agripests may interact synergistically, both with one another and with other inputs such as irrigation (crops) or nutrition (animals). Their combination can cause more damage than the sum of their individual impacts. However, integrating data on multiple organisms and environmental conditions into real-world management decisions is very challenging.

Digital decision support tools may offer novel means to optimise management in complex real-world situations.

The transition process to sustainable agricultural practice. After Parry et al., 2012 and Naranjo, 2011.

The power of computer models for decision-making

The application of modelling and simulation to support decision-making in agriculture, and how best to deploy such tools for decision support, is increasingly being discussed.

For example, CSIRO developed a computer simulation model of Queensland fruit fly (Bactrocera tryoni) that enabled a better understanding of the role of resource landscapes and movement strategy in shaping their population dynamics. This same model was applied to shape area-wide management control strategies, indicating that these approaches may be more effective in urban landscapes than existing ad-hoc approaches, and the best time of year to apply them.

CSIRO also recently developed a model of green peach aphid Myzus persicae and the parasitic wasp Diaeretiella rapae to explore and understand how climate drives their populations. The model allowed us to consider how the time of sowing and type of control (biological, aphid-specific or broad-spectrum chemical) interact, both within and between seasons, to indicate under what conditions IPM approaches may be advantageous.

There are moves now underway to extend such desktop simulations designed for research into applications available in the hands of farmers with data from their own fields.

RapidAIM traps detect fruit fly behaviour and provide an automatic digital alerts to users. Image credit: RapidAIM Pty Ltd.

Automated monitoring

The ‘evidence’ to support models for decision-making can often be a challenge to obtain. However, with the current revolution in “Big Data”, this is about to change. Automated monitoring has boomed, with start-ups like RapidAIM commercialising CSIRO technology with the promise to deliver data on the presence of insect pests. They can provide reports in real-time, at much larger scales possible than by manually collecting field data, and at a fraction of the cost.

Drones are also being used by CSIRO to collect data to monitor plant health and early weed detection. In addition, our scientists are using genetic techniques to develop rapid, real-time, on-farm tests for identifying pests and diseases.

Such ‘big data’ is often defined in terms of the ‘four Vs’ that it provides: a large volume of data, at a rapid velocity (often real-time), from a variety of sources (e.g. multiple monitoring devices) which give substantial value in terms of intelligence. These combine to improve forecasting and prediction for decision-support.

A schematic of ‘Digital IPM’: integration of monitoring, forecasting and decision-making across multiple devices and platforms to provide real-time pest management solutions.

A vision of the future

So, how might computer simulation models evolve into digital decision-support tools for the management of agripests in the near future?

Picture this: a farmer stands in the middle of their crop. Their mobile phone trills and they see a notification. They have been sent a report showing them the latest insect monitoring data from the multiple sensors installed around their farm. Bad news: it looks like there are early signs of green peach aphid activity in the crop.

The farmer then opens an insect pest management tab. It pulls in live data, collected from the Bureau of Meteorology, local weather stations and current crop information. All this information is fed into a computer model that provides them with a detailed, short-term forecast of what is likely to happen next for any established aphid colonies.

At the same time, the app indicates that there is a high level of beneficial insect activity in and around the field, based on the automated image recognition of parasitic wasps in sticky traps. Luckily, these wasps enjoy dining on aphids. This too is fed into the model. The farmer looks at the app’s final recommendation: it reveals there is a low risk to relying on beneficial parasitoids to deal with the potential aphid outbreak, and thus no real need to spray pesticide.

In this scenario the farmer has saved time and money by avoiding a pesticide treatment. Similarly, they’ve avoided off-target risk that could harm the farm’s beneficial insects, providing ongoing protection against pests. But they are now also armed with information and can be prepared to take further action if required.

A farm of tomorrow? It’s not so far-fetched.

Making the vision a reality

Does the above sound fanciful? In fact, the technology almost entirely exists already. The problem is you currently need to open and crosscheck data from a dozen different apps and widgets, and insect models are currently not easily accessible as decision-support tools.

Scientists at the CSIRO, in collaboration with colleagues from a range of organisations have developed a flexible digital framework, known as ‘Digiscape’, which does exactly that. It can integrate multiple types of data (from automated insect traps to weather stations: the ‘Internet of Things’ (IoT)) with forecasting models, making them available as user-friendly, decision-support .

Essentially, the final piece of the puzzle is making everything talk with one another on the one IoT platform.

There is increasing pressure to produce more with less, and we are faced with ongoing challenges of a changing climate leading to increasing and novel insect pest management problems. It is exciting to think that we are beginning to unlock the full potential promised by digital technology for the management of agripests to help us to meet these challenges.

0 comments

Leave a Reply