Susan J. Sprague

Accurate assessment of the presence and severity of plant diseases is essential for effective crop monitoring and management. This study presents a deep learning-based pipeline for quantifying blackleg crown canker disease severity in canola stems by combining image segmentation and severity prediction tasks. Three architectures (ResUNet, UNet and SegFormer) were compared for the first step of stem segmentation to isolate relevant regions. The disease severity scores of four experts, and their aggregated median, were used to train models which were evaluated for label consistency, ambiguity, and model robustness. Among the three segmentation architectures, SegFormer achieved the best performance (mean IoU = 0.939, F1 score = 0.962), outperforming ResUNet and UNet. There was a high correlation in disease severity scores across expert labels, with the median-trained model achieving correlation coefficients of 0.924–0.963 against individual expert assessors on the evaluation dataset. Confusion matrix analysis further demonstrated reliable classification across severity levels. This work highlights the influence of segmentation quality, label aggregation strategies and data imbalances on downstream prediction tasks. This study uses controlled imaging conditions, but the proposed framework provides a strong foundation for future application in field environments. The framework enhances model interpretability by generating severity scores that align closely with expert assessments to support users such as agronomists and plant breeders for better decision-making and help track disease resistance by providing consistent, objective disease measurements over time. Future work will focus on exploring multi-task learning for greater efficiency, alongside validating the approach in field conditions to enable broader adoption.

Plant diseases impair the yield and quality of crops and threaten the health of natural plant communities. Epidemiological models can predict disease and inform management. However, data are scarce, because traditional methods to measure plant diseases are resource intensive, which often limits model performance. Optical sensing offers a methodology to acquire detailed data on plant diseases across various spatial and temporal scales. Key technologies include multispectral, hyperspectral, and thermal imaging, as well as light detection and ranging; the associated sensors can be installed on ground-based platforms, uncrewed aerial vehicles, airplanes, and satellites. However, despite enormous potential for synergy, optical sensing and epidemiological modeling have rarely been integrated. To address this gap, we first review the state of the art to develop a common language accessible to both research communities. We then explore the opportunities and challenges in combining optical sensing with epidemiological modeling. We discuss how optical sensing can inform epidemiological modeling by improving model selection and parameterization and providing accurate maps of host plants. Epidemiological modeling can inform optical sensing by boosting measurement accuracy, improving data interpretation, and optimizing sensor deployment. We consider outstanding challenges in (A) identifying particular diseases; (B) data availability, quality, and resolution; (C) linking optical sensing and epidemiological modeling; and (D) emerging diseases. We conclude with recommendations to motivate and shape research and practice in both fields. Among other suggestions, we propose standardizing methods and protocols for optical sensing of plant health and developing open access databases including both optical sensing data and epidemiological models to foster cross-disciplinary work. [ ] Copyright © 2025 The Author(s). This is an open access article distributed under the CC BY-NC-ND 4.0 International license .

Context: Australian canola (Brassica napus) growers have shifted towards earlier sowing and flowering times in addition to the widespread adoption of stubble conservation practices. Aims: This study determined the consequent impact of these crop and stubble management practices on blackleg crown canker and upper canopy infection caused by the stubble-borne pathogen Leptosphaeria maculans. Methods: Ascospore maturation and release were measured across seasons in commercial canola crops in south-eastern Australia on standing/erect and lying/knocked down stubble. We determined the severity of blackleg crown canker and upper canopy infection in crops with early or late sowing and flowering windows exposed to different stubble treatments. Key results: Ascospore maturation was delayed and ascospore release was 10-fold lower in standing stubble in the first season post-harvest. In the second season when standing stubble was knocked down, total ascospore production was up to 4-fold higher than in the first season. Stubble orientation had no effect on blackleg disease severity. The main driver of infection at vulnerable crop stages was determined by sowing and flowering times such that disease severity was reduced by 35–76% and 98% in early sown and later flowering crops, respectively. Conclusions: Canola grown in short rotations or intensively within a landscape will be exposed to high blackleg inoculum loads from retained stubble. Revision of current blackleg recommendations are required to account for modern crop management practices. Implications: Innovative blackleg control strategies are urgently required to safeguard host genetic resistance and fungicide efficacy in modern farming systems to ensure sustainable intensification of canola production in Australia and globally.

Management of plant disease in agro-ecosystems ideally relies on a combination of host genetic resistance, chemical control, and cultural practices. Growers increasingly rely on chemical and genetic options, but their relative benefits in disease control, yield, and economic outcomes are rarely quantified. We explore this relationship for blackleg crown canker disease (caused by Leptosphaeria maculans), a major biotic constraint limiting canola production globally. Data from 20 field trials conducted from 2013 to 2015 in canola-growing regions of Australia were used to assess the effects of host resistance and fungicide treatment on blackleg severity, grain yield, and gross margin. In the absence of fungicide, blackleg disease was 88% lower in the most resistant compared with the most susceptible blackleg resistance category. In the most susceptible resistance category, the most effective fungicide treatment significantly reduced blackleg severity (from 50 to 6%) and increased grain yield (478 kg/ha, 41%) and gross margin (AU$120/ha, 17%). However, the mean benefits of fungicide tended to decrease with increasing levels of genetic resistance, to the point that yield, disease, and gross margin benefits were close to zero in the most resistant cultivars. Overall, these findings suggest that fungicides can reduce blackleg severity, but the benefits of application strongly depend on associated levels of genetic resistance. Canola cultivars with higher genetic resistance reliably reduced blackleg disease and maintained grain yield without the associated cost of fungicide application. The intensification of canola production to meet increasing global demand will require strategies to sustainably manage and protect finite genetic resistance resources to control blackleg disease.

Crop diseases are a significant constraint to agricultural production globally. Plant disease phenotyping is crucial for the identification, development, and deployment of effective breeding strategies, but phenotyping methodologies have not kept pace with the rapid progress in the genetic and genomic characterization of hosts and pathogens, still largely relying on visual assessment by trained experts. Remote sensing technologies were used to develop an automatic framework for extracting the stems of individual plants from RGB images for use in a pipeline for the automated quantification of blackleg crown canker (Leptopshaeria maculans) in mature Brassica napus plants. RGB images of the internal surfaces of stems cut transversely (cross-section) and vertically (longitudinal) were extracted from 722 and 313 images, respectively. We developed an image processing algorithm for extracting and spatially labeling up to eight individual plants within images. The method combined essential image processing techniques to achieve precise plant extraction. The approach was validated by performance metrics such as true and false positive rates and receiver operating curves. The framework was 98% and 86% accurate for cross-section and longitudinal sections, respectively. This algorithm is fundamental for the development of an accurate and precise quantification of disease in individual plants, with wide applications to plant research, including disease resistance and physiological traits for crop improvement.

Genetic resistance forms the foundation of infectious disease management in crops. However, rapid pathogen evolution is causing the breakdown of resistance and threatening disease control. Recent research efforts have identified strategies for resistance gene deployment that aim to disrupt pathogen adaptation and prevent breakdown. To date, there has been limited practical uptake of such strategies. In this paper, we focus on the socio-economic challenges associated with translating applied evolutionary research into scientifically informed management strategies to control pathogen adaptation. We develop a conceptual framework for the economic valuation of resistance and demonstrate that in addition to various direct benefits, resistance delivers considerable indirect and non-market value to farmers and society. Incentives for stakeholders to engage in stewardship strategies are complicated by the uncertain timeframes associated with evolutionary processes, difficulties in assigning ownership rights to genetic resources and lack of governance. These interacting biological, socio-economic and institutional complexities suggest that resistance breakdown should be viewed as a wicked problem, with often conflicting imperatives among stakeholders and no simple cause or solution. Promoting the uptake of scientific research outcomes that address complex issues in sustainable crop disease management will require a mix of education, incentives, legislation and social change. This article is part of the theme issue ‘Infectious disease ecology and evolution in a changing world’.

Background: Blackleg disease, caused by the fungal pathogen Leptosphaeria maculans, is a serious threat to canola (Brassica napus) production worldwide. Quantitative resistance to this disease is a highly desirable trait but is difficult to precisely phenotype. Visual scores can be subjective and are prone to assessor bias. Methods to assess variation in quantitative resistance more accurately were developed based on quantifying in planta fungal biomass, including the Wheat Germ Agglutinin Chitin Assay (WAC), qPCR and ddPCR assays. Results: Disease assays were conducted by inoculating a range of canola cultivars with L. maculans isolates in glasshouse experiments and assessing fungal biomass in cotyledons, petioles and stem tissue harvested at different timepoints post-inoculation. PCR and WAC assay results were well correlated, repeatable across experiments and host tissues, and able to differentiate fungal biomass in different host-isolate treatments. In addition, the ddPCR assay was shown to differentiate between L. maculans isolates. Conclusions: The ddPCR assay is more sensitive in detecting pathogens and more adaptable to high-throughput methods by using robotic systems than the WAC assay. Overall, these methods proved accurate and non-subjective, providing alternatives to visual assessments to quantify the L. maculans-B. napus interaction in all plant tissues throughout the progression of the disease in seedlings and mature plants and have potential for fine-scale blackleg resistance phenotyping in canola.

Blackleg, caused by Leptosphaeria maculans, is the main disease constraint for canola production in Australia. The fungus infects all aboveground and belowground parts of the canola plant. Yield loss in Australia and worldwide is generally associated with cankers at the crown, which arise from leaf infections during the early stages of seedling growth. Infection of flowers, peduncles, siliques, main stems and branches, with resultant lesions and canker formation, are typically uncommon symptoms. We propose the term ‘upper canopy infection’ to encompass symptoms on all of these plant parts because they generally occur together on the same plant and appear after the plant has undergone elongation. Branch and stem lesions observed in a commercial crop in 2010 were confirmed as L. maculans. Since then, assessment of upper canopy symptoms at 25 sites across the canola-producing regions of Australia between 2011 and 2016 show that symptoms are more prevalent, although they differed between sites and seasons. In 2011, symptoms were present at a single site, and this increased to seven sites in 2013 and 23 sites in 2016. Preliminary data indicate that infection arises from both ascospore and pycnidiospore inoculum, and that earlier onset of flowering is a key risk factor for more severe upper canopy infection. Evidence suggests that host genetic resistance may be an effective control for upper canopy infection.

Cropping has recently expanded into arable areas of the high rainfall zone (HRZ) of Australia. We assessed the suitability of canola varieties of winter, winter × spring and spring-maturity at six sites across the south-eastern, northern and western HRZ of Australia for their suitability for dual-purpose production. Experiments measured potential forage production and the effect of defoliation or grazing on grain yield of crops sown from mid-March to mid-May. Overall, these experiments demonstrated the potential for dual-purpose canola across a wide area of the HRZ. In the south-eastern HRZ where winter conditions were sufficient for vernalisation and spring conditions were mild, winter and winter × spring types outperformed spring types as they provided an extended vegetative period for ‘safe’ grazing (prior to stem elongation), producing 3.0–6.8 t dry matter (DM) ha–1 of forage and recovered to produce 2.5–4.9 t ha–1 of grain yield. In the south-eastern region, early-sown winter types produced more forage than other canola types for grazing in late autumn and winter. In one experiment with four sowing times, consecutive delays in sowing of 2 weeks reduced forage available for grazing by 58%, 72% and 95% compared with the earliest sowing time of 10 March (6.1 t DM ha–1). Although spring types in this region provided some potential for grazing, the phenology was unsuitable for early sowing as the rapid onset of flowering reduced the period of safe grazing. Winter types were not suited to the western region, but the winter × spring and spring types produced >1.0 t DM ha–1 of forage and grain yield of 2.3 t ha–1. In the northern region, spring types produced the highest grain yield (>3.0 t ha–1) but suffered significant yield penalties associated with grazing. In other regions there was generally little or no effect of grazing on grain yield when crops were grazed or defoliated before stem elongation. These experimental studies confirm the potential for dual-purpose canola across all regions of the HRZ when suitable maturity types are sown, managed and grazed appropriately.