Plant–soil feedbacks have been observed in many forest communities, but the role of the mycorrhizal community in perpetuating feedback loops is still poorly understood. Mycorrhizal community composition is closely linked to soil properties and host plant composition, which highlights their potential importance in plant–soil–fungus loops. Eastern hemlock (hemlock; Tsuga canadensis) seedlings were grown in soil bioassays in growth chambers and transplanted under closed forest canopy to examine the effect of hardwood and hemlock forest soil on seedling growth, survival, and ectomycorrhizal fungi (EMF) colonization. Seedlings propagated in hemlock forest soil had greater height growth compared with sterile control soil and achieved greater mycorrhizal colonization than seedlings grown in hardwood forest soils after 9 months in a growth chamber. Outplanted seedlings grown in hemlock communities achieved significantly greater increment growth than those seedlings grown in hardwood communities (mean height difference (95% CI) = 0.39cm (0.14–0.63 cm)), although final survival and EMF colonization was similar between forest types. EMF diversity (Shannon-Wiener index (SE) = 1.88 (0.28) and 1.23 (0.44) for hardwood and hemlock, respectively) and community assemblage (Jaccard index (SE) = 19.0%(4%)) differed between the two forest communities. EMF community assemblage was associated with both the forest type (i.e. plant community/ microsite effects) and initial soil type (i.e. soil characteristics/resistant inoculum). The results support previously observed positive feedbacks between conspecifics under hemlock forest communities and provides evidence for the role of the EMF community within this feedback loop. Alternatively, the reduced growth of hemlocks under hardwoods may be attributed to the different EMF community associated with that forest.