Embedded Files
Western Flower Thrips (WFTs) are one of the major threats to the Australian Strawberry farms, mainly in the southern growing districts of Victoria, Western Australia and South Australia. It is challenging to monitor and control WFTs due to their small size and their strong resistance to most chemicals. Consequently, strawberry growers typically adopt a sophisticated and mostly manual Integrated Pest Management (IPM) process to control WFTs. Intending to address some labour-intensive steps in the current IPM practice, we are researching on automated monitoring of WFTs and modelling their population dynamics under various biotic and abiotic factors.
Why is it important to manage WFTs in crop field?
Western Flower Thrips ( WFTs) as a pest. Western Flower Thrip (Frankliniella Occidentalis) are one of the major insect pests that can be found on a wide range of crops around the world. They can be found in all states and territories in Australia except the Northern Territory. Typically, WFT negatively affects for almost all crops including vegetables (e.g., tomato, capsicum, cucumber), fruits (e.g., strawberry, plum) and ornamental plants (e.g., chrysanthemum) particularly green-house crops. They induce damages to plants i) directly by feeding, and ovipositing on plant parts, ii) indirectly by transmitting viruses. The aforementioned damages by WFT reduce yield, aesthetic value, and storage quality of crops result in million of annual losses in Australia and overseas. Hence, it is important to monitor and control WFT in the crop field thoroughly. In this research, we mainly aim at managing WFTs in Australian strawberry farms.
Why it is challenging to monitor and control WFTs. Nearly 6,500 thrip species has been reported in globally, and around 900 thrips species have been found in Australia. Even though there are a large number of thrip species, only around 10% of them are identified as crop pests. And, some thrip species are beneficial as they act as pollinators or biological control agents(i.e., prey on other insect pests). However, mix-population of thrip species can be found on plants. For example, WFTs, plague thrips and haplorethrips can be found in strawberry farms in Victoria. Moreover, treatments to the different thrip species are varied as some thrip species are resistant to the insecticides. As a result, biological control agents are introduced to the crop field to control the population of some thrip species. Hence, species-level identification of thrips is required, especially where integrated pest management (IPM) strategies are exercised. However, thrips are small in size, and thrip species have minute morphological differences, which are invisible to the naked eye. This makes it is hard to distinguish WFTs from other thrip species.
WFTs undergo different life stages: (1) eggs; (2) two wingless larval stages; (3) two pupal stages; and (4) adult stage. Only the larval and adult stages of WFTs damages plants via feeding. Hence, regular monitoring of both larva and adult stages of WFTs are required. Typically, WFTs are controlled using biological control agents (i.e., predatory bugs) such as Neoseiulus Cucumeris and Orius Insidiosus as WFTs resist to most of the chemicals. However, biological control agents effective only on particular life-stages of WFTs. For example, Cucmeris feed only on the first instar larval of WFTs. Hence, it is challenging to decide the correct controlling strategy (i.e., when, where and what amount of biological control agents/chemicals are needed to be applied).
The conventional WFTs monitoring process. WFTs are currently monitored by collecting thrip samples from the crop field and observing them in a lab environment. Generally, thrips samples are collected weekly or fortnightly, covering the entire crop field. These samples are examined under a microscope with at least 40X magnification for species-level discrimination. If the number of thrips is exceeded a pre-defined threshold, then relevant actions are taken place to control the thrip population. This process is time-consuming, labour-intensive, and requires the domain-experts knowledge for species-level identification. However, there is a potential to automate some parts of this monitoring process to mitigate these drawbacks.
Our approach. In responding to aforementioned these challenges, and to improve the decision-making process, this research attempts to:
- Automate the species-level classification of thrips using their images;
- Simulate population dynamics of WFTs with the changes of biotic and abiotic factors like climate condition (e.g., temperature, humidity, precipitation), the population density of predators (e.g.,Neoseiulus cucumeris and Orius insidiosus), the host pant stage, and chemicals.
The main focuse of the project
Classification of WFTs from other thrip species using images:
It is challenging to distinguish different thrip species due to their small size (~1mm) and similar morphological characteristics between thrip species. Conventional WFTs and other thrips identification are based on visual inspection of external morphology. Since different thrip species are generally similar in appearance, the diagnostic aids called ’keys’ are used to identify the thrip species. Those keys represent different morphological characters, which can be used to discriminate two or more thrip species from each other. For example, WFTs have eight antenna segments, while plague thrips have seven (Figure 1.3). This process is time-consuming and labour-intensive. This process can be conducted only by entomologists or trained technicians. These drawbacks in conventional methods could be mitigated through the development of an automated system to detect those keys and to classify thrip species based on the identified keys, which has not been addressed by previous studies.
In this research, we aim to propose a system to automate the species-level classification of thrips mainly WFTs from other thrip species using their images. Moreover, by working with thrip, we expand the potential of these tools to perform fine-grained classification between morphologically close species, and also microscopic insect species. Also, it is challenging to capture high details images of microscopic insects. Hence, we thoroughly analyse the different image capturing devices and propose a cost-effective device that can be used to capture detailed images of microscopic insects with less human intervention. This also will be benefited to a similar kind of research that involve tiny insects.
Image of a Western Flower Thrip, captured using a mobile zoom lens (24X magnification)
Some differentiating characteristics (keys) of Western Flower Thrips from other thripspecies. Image was captured by the author using a Dino-Lite USB microscope camera after mounting thrip on a microscope slide
Simulation of WFTs population dynamics in Strawberry Farms:
Insect population dynamics models can be used to forecast insect population with the changes of different biotic (e.g., natural enemies, plant growth stage) and abiotic factors (e.g., temperature, chemicals) over time. These models help the decision-making process of IPM in many ways such as: (1) to give early warning to framers based on the forecasted population; (2) to decide the timing and measurement of pesticide/biological control agents applications; and (3) to conduct the sensitivity analysis of the different factors.
WFTs population dynamics with the presence of multiple predators have not been well studied in the literature, and it is challenging due to factors such as: (1) interactions between multiple life stages of predators and prey; (2) immigration and emigration of predators and prey; (3) influence of temperature and plant growth stage; and (4) availability of alternative food sources to predators. Hence, in this research, we aim to simulate the WFT population dynamics in strawberry farms with the changes of (1) temperature and humidity; (2) population density of two predatory bugs (Cumeris and Orius) and (3) strawberry plant growth.
Simulation of WFT population variation with the temperature
About the author
Don Chathurika Amarathunga is a PhD student at Monash University, Australia. She belongs to the Computational & Collective Intelligence group and studying under the supervision of Prof. Alan Dorin, Prof. John Grundy from Monash University, and Dr Hazel Parry from CSIRO. Her current research focuses on automatic monitoring and population dynamics modelling of insect pests.
Email: don.amarathunga@monash.edu
Email: don.amarathunga@monash.edu
Page updated
Google Sites
Report abuse