Objectives This study investigated the effects and underlying mechanisms of different biocontrol agents in alleviating continuous cropping obstacles in greenhouse-grown watermelon. By analyzing and comparing rhizosphere microbiome changes driven by bacterial and fungal biocontrol agents, as well as their relationships with watermelon growth promotion and Fusarium wilt suppression, this research provides technical support and theoretical foundations for the cultivation of self-rooted watermelon seedlings.

Methods Self-rooted watermelon cultivar 'Su Meng No. 6' was used as the experimental material. Two biocontrol agents, Bacillus amyloliquefaciens and Trichoderma harzianum, are applied in a field experiment conducted in continuously cropped greenhouse soil. Three treatments were established: control without microbial agents (CK), B. amyloliquefaciens agent (BA), and T. harzianum agent (FA). Plant growth indices, disease index, and the structure and diversity of rhizosphere soil microbial communities were measured.

Results Compared with CK, both BA and FA treatments significantly alleviated growth inhibition caused by continuous cropping specifically, stem diameter increased by 18.7% and 27.1%, plant biomass by 97.0% and 129.8%, and single fruit weight by 60.8% and 77.1% under BA and FA treatments, respectively. For disease control, both BA and FA treatments significantly reduced the disease index of watermelon Fusarium wilt, with relative control efficacies of 50.1% and 45.6%, respectively. The application of biocontrol agents significantly altered the bacterial and fungal community structures in the watermelon rhizosphere soil. In addition, FA treatment significantly decreased and increased the alpha diversity of bacteria and fungi, respectively. Correlation analysis revealed a significant positive correlation between the fungal Shannon index and watermelon plant biomass. Fusarium oxysporum was the second most abundant fungal taxon in the f rhizosphere, with an average relative abundance of 13.8%. However, no significant differences in its abundance were observed among treatments. The relative abundances of beneficial genera such as Neobacillus and Linnemannia, which were significantly negatively correlated with the disease index, were significantly increased in both BA and FA treatments. Random forest analysis identified these two genera as key predictors of disease incidence. The disease suppression effect of BA treatment on continuously cropped watermelon was also likely attributed to a significant enhancement of key ecological modules within bacterial co-occurrence networks, which included beneficial taxa such as Sphingomonadales, Gemmatimonadales, and Lysobacterales.

Conclusions Both biocontrol agents significantly alleviated continuous cropping obstacles in greenhouse watermelon, primarily by reshaping rhizosphere microbial community structure and regulating key functional microbial taxa in the rhizosphere, rather than through direct suppression of F. oxysporum population. The differential mechanisms of the two biocontrol agents are as follows: the B. amyloliquefaciens agent significantly enriched key bacterial ecological modules, whereas the T. harzianum agent significantly altered the alpha diversity of microbial community.