In recent times, network theory has become a useful tool to study the structure of the interactions in ecological communities. However, typically, these approaches focus on a particular kind of interaction while neglecting other possible interactions present in the ecosystem. Here, we present an ecological network for plant communities that consider simultaneously positive and negative interactions, which were derived from the spatial association and segregation between plant species. We employed this network to study the structure and the association strategies in a semiarid plant community of Cabo de Gata-Níjar Natural Park, SE Spain, and how they changed in 4 sites that differed in stocking rate. Association strategies were obtained from the partitions of the network, built based on a relaxed structural balance criterion. We found that grazing simplified the structure of the plant community. With increasing stocking rate species with no significant associations became dominant and the number of partitions decreased in the plant community. Independently of stocking rate, many species presented an associative strategy in the plant community because they benefit from the association to certain 'nurse' plants. These 'nurses' together with species that developed a segregating strategy, intervened in most of the interactions in the community. Ecological networks that combine links with different signs provide a new insight to analyze the structure of natural communities and identify the species which play a central role in them. © 2014 Elsevier Masson SAS.

We thank the authorities of Cabo de Gata-Níjar Natural Park for their cooperation in the study. The study was made possible by the financial support of ECO-CC project Ref. CGL2011-27259 of the Spanish Ministry of Science and Innovation, and grant JAEPre from the National Research Council of Spain. We thank Bruce McWhirter and Yolanda Pueyo for critically reading on the manuscript and the editor and three anonymous referees for its helpful suggestions.

Peer Reviewed