Wheat stripe rust (WSR) is a foliar disease that causes destructive damage in the wheat production context. Accurately estimating the severity of WSR in the autumn growing stage can help to objectively monitor the disease incidence level of WSR and predict the nationwide disease incidence in the following year, which have great significance for controlling its nationwide spread and ensuring the safety of grain production. In this study, to address the low accuracy and the efficiency of disease index estimation by traditional methods, WSR-diseased areas are segmented based on Segformer, and the macro disease index (MDI) is automatically calculated for the measurement of canopy-scale disease incidence. The results obtained with different semantic segmentation algorithms, loss functions, and data sets are compared for the segmentation effect, in order to address the severe class imbalance in disease region segmentation. We find that: (1) The results of the various models differed significantly, with Segformer being the best algorithm for WSR segmentation (rust class F1 score = 72.60%), based on the original data set; (2) the imbalanced nature of the data has a significant impact on the identification of the minority class (i.e., the rust class), for which solutions based on loss functions and re-weighting of the minority class are ineffective; (3) data augmentation of the minority class or under-sampling of the original data set to increase the proportion of the rust class greatly improved the F1-score of the model (rust class F1 score = 86.6%), revealing that re-sampling is a simple and effective approach to alleviating the class imbalance problem. Finally, the MDI was used to evaluate the models based on the different data sets, where the model based on the augmented data set presented the best performance (R² = 0.992, RMSE = 0.008). In conclusion, the deep-learning-based semantic segmentation method, and the corresponding optimization measures, applied in this study allow us to achieve pixel-level accurate segmentation of WSR regions on wheat leaves, thus enabling accurate assessment of the degree of WSR disease under complex backgrounds in the field, consequently providing technical support for field surveys and calculation of the disease level.

Traditional assessments of grapevine susceptibility to grapevine downy mildew (GDM) caused by Plasmopara viticola rely on the visual evaluation of leaf symptoms. In this study, we used a well-established quantitative real-time PCR TaqMan assay (real-time PCR) to quantify the number of P. viticola infecting 12 grapevine cultivars under controlled conditions. The molecular disease index (MDI), derived from molecular detection methods, reflects the relative abundance of pathogens in plant tissues during the latent infection phase. Our application of the MDI revealed a progressive increase in latent P. viticola levels over time, indicating dynamic levels of latent P. viticola infection across the inoculation processes. We calculated the area under the disease progress curve in terms of the MDI (AUDPCMDI) to evaluate the susceptibility of each cultivar to GDM. Cultivars with lower AUDPCMDI values consistently exhibited reduced pathogen establishment, suggesting higher levels of innate resistance. Correlation analysis revealed a significant correlation between the visual disease index and the AUDPCMDI values ( r = 0.790, P = 0.002), indicating that higher levels of latent P. viticola infections were associated with higher disease severity. Grapevine cultivars were clustered into distinct groupings, indicating variability in their susceptibility to the pathogen. Cultivars with similar levels of susceptibility were grouped, highlighting that the real-time PCR assay used in this study represents a robust, rapid, and standardized method for quantifying pathogens, which significantly improves the efficiency of evaluating the susceptibility of grapevine cultivars to GDM. This quantitative protocol provides practical guidelines for selecting resistant cultivars and implementing effective disease management strategies.

The obligate biotrophic fungus Puccinia striiformis f. sp. tritici (Pst) frequently overcomes host immunity through rapid evolution. The wheat landrace Paozimai exhibits high-temperature all-stage resistance (HTAS) to Pst, a type of resistance expressed at relatively high temperatures across all growth stages. In this study, we investigated host–pathogen interactions using integrated transcriptomic and data-independent acquisition proteomic analyses. Dual RNA-seq analysis at 192 and 216 h post-inoculation (hpi) revealed a shift in enriched Pst pathways during the late latent period, from endoplasmic reticulum protein processing and fatty acid metabolism at 192 hpi to carbohydrate and glutathione metabolism at 216 hpi. In the host, differentially expressed genes (DEGs) were enriched in defense-related signaling pathways at 192 hpi and in metabolic pathways at 216 hpi. Co-expression network analysis identified ribosomal and protein biosynthesis processes that may contribute to defense. Proteomic analysis showed greater overlap across time points, with oxidative phosphorylation enriched at 192 hpi and glycan biosynthesis at 216 hpi. Furthermore, integrated multi-omics and co-expression network analysis identified 16 candidate wheat genes, including genes encoding ribosomal proteins and stress-response proteins. These findings describe temperature-associated host and pathogen responses during HTAS and nominate candidates for functional validation.

Early warning of grapevine downy mildew (GDM) remains a critical challenge for precision crop protection. Here, we develop a stage-aware fusion framework that aligns qPCR-derived pathogen pressure, qPCR-based mechanistic infection risk modeling, and hyperspectral canopy sensing across phenological stages. The cumulative risk index (CRI) provided a measurable temporal prior: its association with leaf qPCR increased across lags and peaked at a 5-week lead (r = 0.665, P < 0.001), with regional optima spanning 2–5 weeks (r = 0.53–0.98, P < 0.001). The early-stage spectral disease index (SDI) reflects the observability of molecular disease variation. The SDI consistently outperformed single source predictors across phenological stages. At the early stage (n = 795), the SDI achieved a predictive accuracy of 71% (R² = 0.71, ρ = 0.77, P < 0.001), exceeding the performance of the mechanistic CRI (R² = 0.69, ρ = 0.80, P < 0.001) and the hyperspectral physiological stress indicator (PSI; R² = 0.09, ρ = 0.17, P < 0.001) when applied in isolation. At the late stage (n = 375), the SDI performance decreased to 46% (R² = 0.46, ρ = 0.50, P < 0.001), closely tracking the CRI (R² = 0.42, ρ = 0.48, P < 0.001), whereas the PSI remained weakly informative (R² = 0.04, ρ = 0.20, P < 0.001). Bayesian SDI stratification further revealed that incidence acceleration was greatest at intermediate SDI levels (mean delta incidence = 36.8, 95% HDI 36.4–37.2). These results support a stage-aware early warning framework in which the integration of epidemiological risk accumulation with canopy-level spectral responses can enhance early disease warning and disease severity estimation.