Botanical concepts have traditionally viewed the environment as a static box containing plants. In this box, plants compete with one another and act as passive resource consumers subjected to the environment in a top-down manner. This entails that plants have only negative effects on other plants and have no influence on the environment. By contrast, there is increasing evidence that plants have positive, bottom-up engineering effects and diversity effects on other plants and on the environment. Here, to overcome the limitations of top-down environmental control, antagonistic-only and pairwise interactions, I propose the concept of constructive networks. Constructive networks unify niche construction and network theory recognizing that (i) plants have manifold ecological functions and impacts on their neighbours, and (ii) the environment shapes and is shaped by diverse organisms, primarily plants. Constructive networks integrate both plant–environment and plant–plant interactions in a relational context. They address how plants influence the environment and support or inhibit other plant species by physically, biochemically and ecologically shaping environmental conditions. Constructive networks acknowledge the fact that diverse plants change and create novel environmental conditions and co-produce, share and transform resources, thereby influencing biological communities and the environment in constructive ways. Different interaction types are considered simultaneously in constructive networks. Yet, the main limitation to understanding constructive networks is the identification of plant links. This barrier may be overcome by applying complexity theory and statistical mechanics to comparative data and experimental field botany. Considering multiple interaction types and feedback between plants and the environment may improve our understanding of mechanisms responsible for biodiversity maintenance and help us to better anticipate the response of plant systems to global change.