Within biological communities, species interact in a wide variety of ways. Species interactions have always been noted and classified by naturalists in describing living organisms and their ways. Moreover, they are essential to characterize ecological communities as functioning entities.<br> Biodiversity databases, as a rule, are comprised of species records in certain localities and times. Many, if not most, originated as databases of museum specimens and/or published records. As such, they provide data on species occurrences and distribution, with little functional information. Currently, online databases for species interaction data are being formed or proposed. Usually, these databases set out to compile data from actual field studies, and their design reflects the singularities of particular studies that seed their development. In two online databases: the Web of Life (2021) and the Interaction Web DataBase (2020) (IWDB), the categories of interactions are quite heterogeneous (Table 1). For instance, they may refer explicitly to certain taxonomic groups (e.g., anemone-fish), or do so implicitly (host-parasitoid; parasitoids are all holometabolous insects with arthropod hosts); conversely, they may encompass almost any taxon (food webs). In another example, the Global Biotic Interactions database (Poelen et al. 2014) (GloBI) offers a choice of relational attributes when entering data, ranging from undefined to quite restricted (Table 2).<br> Here we intend to contribute to the development of interaction databases, from two different points of view. First, what categories can be effectively applied to field observations of biotic interactions? Second, what theoretical and applied questions do we expect to address with interaction databases? These should be equally applicable to comparisons of studies of the same kind or mode of interaction, and to contrasts between interactions in multimodal studies.
