Presymptomatic detection with multispectral imaging to quantify and control the transmission of cassava brown streak disease
Lead Research Organisation:
University of Manchester
Department Name: Electrical and Electronic Engineering
Abstract
Assuring food security for 8 billion people is one of the most pressing challenges of the 21st century. Insurance crops like cassava, which can withstand droughts and grow in nutrient-poor soil, are projected to play a key role in these efforts. However, cassava production in East Africa it is limited by RNA viruses that cause cassava brown streak disease (CBSD). CBSD causes subtle to no symptoms on stems and leaves while destroying the root tissue, which means farmers may be unaware their field is infected until they have a failed harvest. Distressingly, the visible symptoms are so slight that cuttings provided by 'clean seed' programs may be infected. While molecular diagnostics like PCR can determine if plants are infected, they are economically inviable for testing many plants in a field. Since there are no cures for CBSD, farmers can only rogue infected plants (remove them from the field before they can spread infection), but models of rouging based on visible symptoms show the method to be ineffective. Because it is difficult to observe infection prior to harvest, we do not know how quickly CBSD spreads in fields due to transient association with whitefly vectors, which has harmed modeling efforts. CBSD has been spreading in East Africa since it became epidemic in the 1990s, the development of CBSD-resistant cassava clones has been slow, and the disease is at high risk for spread to West Africa.
We propose to use an engineering advancement, our multispectral imager (MSI), to rapidly determine the infection status of plants in the field in Tanzania. It observes the leaves of plants with many different light spectra, which are then interpreted by machine learning models trained on cassava leaf scans. Under laboratory conditions, the MSI detects CBSD infection with 95% accuracy at 28 days post infection, when plants have no visible symptoms. We will experimentally study the spread of CBSD and parameterize transmission models to assess the efficacy of rouging with different detection methods in the most critical fields with large downstream effects within the clean seed propagation system. Finally, using these parameterized models we will look at how host susceptibility and environmental factors driving vector abundance affect disease pressure, and model interventions by farmers and agricultural agents to minimize regional spread and standing disease pressure from CBSD.
Intellectual Merit
Our study will be the first to track CBSD spread in the field, the first to accurately model the plant-to-plant spread of this emergent and damaging cassava pathogen, and the first to integrate methods with different limits of detection into a plant pathosystem with significant vegetative propagation. We will create both accurate tools for use in cassava and frameworks for employing our technology and models for other economically significant pathosystems in vegetatively propagated crops, such as potato, sweet potato, taro and yam.
Broader Impacts
This research will significantly impact the food security of people living in areas affected by CBSD and reduce the likelihood of further spread into other regions of Sub-Saharan Africa, where cassava is a staple crop for 800 million people. We will partner with Tanzanian cassava researchers with connections to the Tanzanian clean seed program to assure a high likelihood of translation of the conclusions of our work to real applications. We will share our results through accessible recorded talks, at regional meetings in Africa, presentations at scientific conferences. We will actively recruit researchers from traditionally excluded groups and will train at least four postdocs, a PhD student, a research associate, and at least eight undergraduate students. All trainees will be exposed to a multinational research team working on a critical problem of African agriculture.
We propose to use an engineering advancement, our multispectral imager (MSI), to rapidly determine the infection status of plants in the field in Tanzania. It observes the leaves of plants with many different light spectra, which are then interpreted by machine learning models trained on cassava leaf scans. Under laboratory conditions, the MSI detects CBSD infection with 95% accuracy at 28 days post infection, when plants have no visible symptoms. We will experimentally study the spread of CBSD and parameterize transmission models to assess the efficacy of rouging with different detection methods in the most critical fields with large downstream effects within the clean seed propagation system. Finally, using these parameterized models we will look at how host susceptibility and environmental factors driving vector abundance affect disease pressure, and model interventions by farmers and agricultural agents to minimize regional spread and standing disease pressure from CBSD.
Intellectual Merit
Our study will be the first to track CBSD spread in the field, the first to accurately model the plant-to-plant spread of this emergent and damaging cassava pathogen, and the first to integrate methods with different limits of detection into a plant pathosystem with significant vegetative propagation. We will create both accurate tools for use in cassava and frameworks for employing our technology and models for other economically significant pathosystems in vegetatively propagated crops, such as potato, sweet potato, taro and yam.
Broader Impacts
This research will significantly impact the food security of people living in areas affected by CBSD and reduce the likelihood of further spread into other regions of Sub-Saharan Africa, where cassava is a staple crop for 800 million people. We will partner with Tanzanian cassava researchers with connections to the Tanzanian clean seed program to assure a high likelihood of translation of the conclusions of our work to real applications. We will share our results through accessible recorded talks, at regional meetings in Africa, presentations at scientific conferences. We will actively recruit researchers from traditionally excluded groups and will train at least four postdocs, a PhD student, a research associate, and at least eight undergraduate students. All trainees will be exposed to a multinational research team working on a critical problem of African agriculture.
Technical Summary
Cassava brown streak disease causes $1 billion annual losses in cassava production and severely impacts affected small stakeholder farmers. The symptoms on growing cassava plants are often very subtle, and infected plants are frequently considered healthy by experienced cassava farmers and even by cassava breeders cultivating 'clean seed' for sale to small stakeholder farmers. This prevents timely roguing, and modelling predicts roguing based on visual symptoms is ineffective for the control of CBSD.
Increased understanding of the epidemiology of CBSD would benefit cassava seed systems, and the application of field-level technology that increases the accuracy of CBSD detection would have a tremendous economic impact and improve food security in sub-Saharan Africa. This proposal aims to deliver on both targets and will model the efficacy of different detection strategies in the Tanzanian clean seed system. Because there are similar impacts in all other countries affected by CBSD, improvements to CBSD diagnosis and control achieved in Tanzania will have direct scaling potential across the wider geographical region of East, Central and Southern Africa.
Our team has developed a hand-held active multispectral imaging (MSI) device that takes images with a wide range of wavelengths of light. Our laboratory studies show that the model predictions were 95% accurate in a susceptible cultivar at 28 days post infection (dpi), a time when most plants exhibited no symptoms and PCR failed to detect any of the infected plants.
We will use our MSI device for CBSD early detection, develop field experiments to accurately parametrise CBSD disease spread through the cassava seed system, and develop models of CBSD spread under different detection and surveillance scenarios. In this way we will uncover key data about the biology and evolvability of CBSIs.
Increased understanding of the epidemiology of CBSD would benefit cassava seed systems, and the application of field-level technology that increases the accuracy of CBSD detection would have a tremendous economic impact and improve food security in sub-Saharan Africa. This proposal aims to deliver on both targets and will model the efficacy of different detection strategies in the Tanzanian clean seed system. Because there are similar impacts in all other countries affected by CBSD, improvements to CBSD diagnosis and control achieved in Tanzania will have direct scaling potential across the wider geographical region of East, Central and Southern Africa.
Our team has developed a hand-held active multispectral imaging (MSI) device that takes images with a wide range of wavelengths of light. Our laboratory studies show that the model predictions were 95% accurate in a susceptible cultivar at 28 days post infection (dpi), a time when most plants exhibited no symptoms and PCR failed to detect any of the infected plants.
We will use our MSI device for CBSD early detection, develop field experiments to accurately parametrise CBSD disease spread through the cassava seed system, and develop models of CBSD spread under different detection and surveillance scenarios. In this way we will uncover key data about the biology and evolvability of CBSIs.