In most mycorrhizal symbioses, including most adult green orchids, fungi provide soil minerals to the plant, in exchange for photosynthetic sugar. Yet, during orchid germination, the fungi, which belong to the polyphyletic rhizoctonias, provide carbon to germinating seedlings that have no reserve. Moreover, in some orchid species, adult plants also recover carbon from their fungi: some are achlorophyllous and non-photosynthetic (mycoheterotrophic species) while others are green but mix photosynthesis and exploitation of fungal carbon (mixotrophic species). Mixotrophic and mycoheterotrophic orchids have lost association to rhizoctonias, but connect (often with high specificity) to ectomycorrhizal fungi, so that their carbon issues from surrounding trees. But this scenario is mostly based on temperate studies – how is it challenged in tropical regions, where additionally most orchids are epiphytic? Our recent researches provide clues on this question. First, and although this was sometimes questioned, epiphytic orchids are always mycorrhizal. We characterized the orchid fungi across the natural habitats of Reunion Island (Pacific) and investigated the architecture of bipartite plant-fungal networks for 73 orchid species and 95 taxonomic units of mycorrhizal fungi. Unlike some recent evidence for nestedness in mycorrhizal symbioses, we found a highly modular architecture that largely reflected an ecological barrier between epiphytic and terrestrial sub-networks. By testing for phylogenetic signal in this network, it was stronger for both partners in the epiphytic subnetwork. Moreover, in the sub-network of epiphytic angraecoid orchids, the signal in orchid phylogeny was stronger than the signal in fungal phylogeny. Epiphytic associations are therefore more conservative and may coevolve more than terrestrial ones. Second, our study of tropical mycoheterotrophic and mixotrophic orchids in the paleo- and neotropics revealed that specificity is not the rule and that, if rh