Understanding and Managing Frogeye Leaf Spot through Network-Based Modeling in Soybean

Frogeye Leaf Spot (FLS), caused by Cercospora sojina, poses a significant threat to soybean production, with yield losses of 30-60%. Traditional mass-action models assume homogeneous mixing, which rarely holds in real fields and limits their ability to inform FLS management. To address this, we developed a network-based model that incorporates real-field structure to improve FLS management in soybeans. Using approximate Bayesian computation, we estimated key epidemiological parameters and found that infection origin can shift the balance between transmission routes. Data analyses indicated that tillage and non-tillage plots did not differ significantly in fungal spread, decay, or disease severity. Finally, we show that early, targeted roguing is more effective than delayed or random removal. Together, these findings offer science-based guidance for FLS management and highlight the value of network-based models to inform agricultural disease control.

💡 Research Summary

Frogeye leaf spot (FLS), caused by the fungus Cercospora sojina, remains a major threat to soybean production, with reported yield losses ranging from 30 % to 60 %. Traditional epidemiological models for this disease have relied on mass‑action (homogeneous mixing) assumptions, which ignore the spatial heterogeneity inherent in real fields and therefore limit the models’ utility for guiding management decisions. In this study, the authors develop a spatially explicit, network‑based SEIRB (Susceptible–Exposed–Infectious–Removed–Soil‑borne inoculum) model that captures both plant‑to‑plant (direct) transmission and soil‑mediated (primary) infection pathways.

The model construction proceeds in two stages. First, the classic SEIRB framework of Yang and Wang (2022) is described, where the environmental compartment B(t) represents the density of viable spores in soil and crop residue. The differential equations include a direct transmission rate θ, a soil‑mediated transmission rate β, a latent period σ⁻¹, an infectious period γ⁻¹, and a logistic growth term for the inoculum reservoir (intrinsic growth r₀, carrying capacity k₀, decay τ). Second, the authors embed this compartmental system in a contact network derived from the actual geometry of a field experiment. The field consists of six sub‑plots (three tillage, three no‑tillage), each containing 1 728 soybean plants spaced at 12 seeds m⁻¹. Plant coordinates are used to compute Euclidean distances; an edge is placed between any two plants whose distance is ≤ d, where d is a tunable threshold. This yields an adjacency matrix A that defines which plants can directly infect each other, preserving realistic row and alley spacing. By varying d, the authors can explore regimes from very local (nearest‑neighbor) to more densely connected networks.

Parameter inference is performed using Approximate Bayesian Computation (ABC), a simulation‑based approach suitable for complex stochastic network models where likelihoods are intractable. Summary statistics derived from field observations (disease severity, soil spore counts) are compared to simulated outputs, and the posterior distributions of the key transmission parameters (θ, β, ξ – the contribution of infectious plants to the soil reservoir) are obtained. The ABC analysis also quantifies uncertainty, allowing the authors to assess the robustness of model predictions.

The results address three central questions. (1) Effect of tillage vs. no‑tillage – The estimated soil inoculum dynamics (initial B, decay τ, and growth r) do not differ significantly between tillage and no‑tillage plots. Consequently, the overall disease trajectories (cumulative infections, peak prevalence) are statistically indistinguishable, suggesting that current tillage practices alone are insufficient to curb FLS spread under the conditions studied. (2) Influence of infection origin – Simulations initialized with an infected plant at the network’s core show a dominance of direct transmission (θ) because many neighbors are immediately exposed. When the seed infection is placed at a peripheral location, the contribution of soil‑mediated infection (β) rises, reflecting the longer time required for the pathogen to percolate through the sparse edge connections. This origin‑dependence highlights the importance of early scouting to locate initial foci. (3) Management via roguing – The authors compare three roguing strategies: (a) random removal of a fixed proportion of plants, (b) delayed removal after disease has progressed, and (c) early, targeted removal of plants with the highest infection status and network centrality (degree). The targeted early roguing consistently outperforms the other two, reducing final disease prevalence by 40–60 % and lowering the soil inoculum reservoir. The benefit stems from breaking key transmission pathways before secondary cycles amplify.

Beyond these specific findings, the study demonstrates the broader value of network epidemiology in plant pathology. By explicitly modeling spatial contact structure, the approach captures clustering effects, edge‑effects across subplot boundaries, and the interplay between local and reservoir‑driven transmission. The integration of ABC provides a principled way to calibrate such models to field data, yielding credible intervals for management‑relevant parameters.

The authors acknowledge several limitations. The distance threshold d is a simplification of real dispersal kernels; wind‑driven spore movement and vector behavior are not modeled. The soil inoculum compartment follows a simple logistic growth and decay, ignoring potential seasonal or moisture‑dependent variability. Climate variables (temperature, humidity) that modulate infection rates are held constant. Future work could incorporate stochastic weather drivers, multi‑pathogen interactions, and economic cost‑benefit analyses of roguing versus fungicide applications.

In conclusion, this paper introduces a rigorously calibrated, spatially explicit SEIRB network model for Frogeye Leaf Spot in soybean. It shows that (i) tillage alone does not markedly affect disease dynamics, (ii) the spatial origin of infection shifts the balance between direct and soil‑mediated transmission, and (iii) early, targeted roguing is a highly efficient, low‑cost management tactic. The methodology sets a template for applying network‑based epidemiology to other crop diseases, bridging the gap between theoretical disease dynamics and actionable field‑level recommendations.