Fungal symbionts are increasingly targeted as tools for crop management, but their use in the field requires an understanding of how fungi interact in a community context. Fungal interactions may result in additive effects on the host plant, which could be predicted simply based on individual fungal behavior. Alternatively, interactions among fungi may result in non-additive synergistic or antagonistic effects on plant performance that are more challenging to predict. Here, we hypothesized that the effects of fungal interactions on the plant host could be predicted from their niche overlap. To test this idea, we used foliar fungal endophytes with a range of niche overlap based on trait dissimilarities to examine the effects of six fungal species pairs compared to the corresponding individual fungal species on switchgrass (Panicum virgatum L.) in water-stressed and well-watered conditions. Mixtures of endophytes had either no effect or predictable, additive effects on plant tiller number, but effects on plant growth rate and wilting were largely non-additive. Moisture level, fungal stress, and metabolic trait dissimilarity predicted 51 to 92% of the deviation of fungal effects from additive, with less similar fungi likely to have more synergistic effects on the plant host. Furthermore, we identified indicator metabolites for fungal interaction outcomes. However, the effects of endophyte interactions on the plant host were environment dependent making single community applications more challenging. Overall, future development of microbial tools for use in agriculture must consider their interactions to optimize application.