Abstract Questions Positive and negative associations among species influence the structure of plant communities. Yet how these plant associations are assembled at the community level is poorly understood. We propose a new approach that combines spatial ecology, network theory and trait‐based ecology to examine the assembly of plant–plant associations at the community level. Location Gemmipass, Swiss Alps. Methods We fully mapped an alpine plant community at the individual plant level, recording both plant coordinates and functional traits for each individual. We identified non‐random species associations using spatial point‐pattern analysis and partialled out the effect of abiotic heterogeneity. We then analysed the plant network structure and used plant traits to predict species associations. Results We identified 36 significant spatial associations between plant species, 34 positive and two negatives. Dominant, stress‐tolerant species such as Dryas octopetala , Linaria alpina and Leontodon montanus were highly connected in the network, whereas rare, water‐ and nutrient‐demanding species such as Saxifraga aizoides , Galium anisophyllon and Thymus praecox were less connected compared to random expectation. The plant network was clustered, meaning that species were overall more connected among each other than expected by chance. Conclusions Positive associations among species characterized the studied plant community. Besides the primary effect of associations of the “foundation” species D. octopetala with other species, these “subordinate” plants were also associated with each other. Our study reveals the assembly of plant communities as driven by positive associations among stress‐tolerant pioneer species, highlighting their role in supporting the cohesiveness of alpine plant communities.