The complex relationship between disease resistance and yield in crops

Abstract

In plants, growth and defence are controlled by many molecular pathways that are antagonistic to one another. This results in a 'growth-defence trade-off', where plants temporarily reduce growth in response to pests or diseases. Due to this antagonism, genetic variants that improve resistance often reduce growth and vice versa. Therefore, in natural populations, the most disease resistant individuals are often the slowest growing. In crops, slow growth may translate into a yield penalty, but resistance is essential for protecting yield in the presence of disease. Therefore, plant breeders must balance these traits to ensure optimal yield potential and yield stability. In crops, both qualitative and quantitative disease resistance are often linked with genetic variants that cause yield penalties, but this is not always the case. Furthermore, both crop yield and disease resistance are complex traits influenced by many aspects of the plant's physiology, morphology and environment, and the relationship between the molecular growth-defence trade-off and disease resistance-yield antagonism is not well-understood. In this article, we highlight research from the last 2 years on the molecular mechanistic basis of the antagonism between defence and growth. We then discuss the interaction between disease resistance and crop yield from a breeding perspective, outlining the complexity and nuances of this relationship and where research can aid practical methods for simultaneous improvement of yield potential and disease resistance.

Keywords: disease; growth; yield.

Figure 1

The impacts of nucleotide‐binding site leucine‐rich repeat (NBS‐LRR) gene introgression on yield stability and yield potential are possibly modulated by endogenous and exogenous interactions. This figure summarizes a preliminary, simplified overview of two recently discovered factors, and one known factor, that may modulate the impacts of NBS‐LRR gene introgression on yield potential and stability respectively. Following introgression (green arrow, top), the gene may have interactions with exogenous agents such as necrotrophic pathogens (purple arrow, upper‐left), which could enhance susceptibility and lower (‘modulation’ grey arrow and bottom‐left) overall yield stability. Inside the plant, the new NBS‐LRR gene could have endogenous interactions (blue arrows, right) with the existing NBS‐LRR complement (middle) or NBS‐LRR‐targeting microRNA content (right, purple). Existing NBS‐LRRs could form complexes that inhibit or enhance activity of the introgressed NBS‐LRR, or they could not interact with it at all. The precise nature of these interactions and the activity of the introgressed gene may impact yield potential. An extreme example of incompatibility in NBS‐LRR complements is seen in the phenomenon of hybrid necrosis. MicroRNAs targeting NBS‐LRRs can evolve from transcription of hairpin RNAs from existing tandemly duplicated NBS‐LRRs. Whether or not these microRNAs target the new NBS‐LRR and the timing and strength of that interaction, will likely impact the overall effect of it on yield potential.

Figure 2

The complex relationship between disease resistance and yield potential in crops, and a basic outline of the genomic prediction technique. (a) Crop yield is a complex trait, which is affected by several component traits related to growth, morphology and developmental timing. These traits may have complex genetic interactions with one another and the levels of both disease susceptibility and yield potential. This, coupled with strong environmental interactions, complicates the link between the molecular growth‐defence trade‐off and the genetic antagonism between yield potential and disease resistance and crops. (b) A training population is a set of varieties of the crop of interest used for creating a genomic prediction model. Both phenotype and genotype data are collected for this population and used to generate a statistical model. A testing population is a set of varieties from which selections are to be made. Genotypes from the testing population can be input into the statistical model to predict phenotypes based on the relationship between genotype and phenotype described in the model. Predicted phenotypes allow selections at various points in the breeding programme.