摘要
Plants do not confront pathogens as isolated organisms. Their roots are embedded in microbial communities whose composition and activity can alter the probability, severity and persistence of disease. The 'cry for help' hypothesis proposes that pathogen attack can reprogramme plant root exudation and other host traits in ways that favour protective microorganisms, creating an indirect defence layer that may operate through antagonism, resource competition, immune priming or disease-suppressive soil legacies. This critical narrative review evaluates how strongly the available evidence supports that causal sequence. Literature was selected through transparent searches of multidisciplinary, biomedical and agricultural scholarly sources, with emphasis on experiments that linked pathogen perception, changes in root chemistry, microbiome assembly and disease outcomes. The evidence is strongest where chemical perturbation, host genetics, microbial isolation or synthetic-community experiments, transplantation and disease phenotyping converge. Studies in Arabidopsis thaliana, tomato and wheat show that pathogen challenge or defence activation can alter root exudates and enrich microorganisms that subsequently reduce disease. Yet the generality of a host-adaptive recruitment programme remains uncertain. Many influential experiments use simplified substrates, hydroponic exudate collection, relative-abundance sequencing or controlled-environment systems that do not reproduce the sorption, diffusion, microbial transformation and ecological competition operating in field soils. Pathogens can also manipulate host metabolism, and disease-associated microbial shifts may be by-products of tissue damage, nutrient leakage or altered plant physiology rather than adaptive recruitment. Root exudates therefore function less as unambiguous 'signals' than as context-dependent chemical filters whose ecological effects depend on microbial traits, soil chemistry, host genotype and prior community assembly. The most defensible interpretation is that plants possess mechanisms capable of biasing microbiome assembly during disease, but only a subset of documented cases currently demonstrates the full causal chain required by a strong cry-for-help model. Progress will depend on in situ metabolite flux measurements, absolute and activity-based microbiome quantification, causal genetic tests, multikingdom community reconstruction and replicated field validation across soils and crop genotypes.