Contrasting short‐ and long‐term direct and indirect interactions in Mediterranean oak savannah communities

Co‐occurrence data are often used to infer interactions between species in plant communities. However, the robustness of these inferences is debated and benefits from a joint assessment of both co‐occurrence data and empirical measurements of biotic interactions. We used the Mediterranean oak‐savannah communities from Spain to address whether co‐occurrence patterns coincide with short‐term assessments of biotic interactions quantified through experimental removal of dominant neighbours. We quantified, with an observational procedure, species co‐occurrences between herbaceous species and an evergreen Quercus , a broom‐like leguminous shrub and an evergreen Cistaceae shrub. Direct and indirect (in presence of Quercus ) short‐ and long‐term effects of the two shrubs were quantified using the removal method. Species were grouped in co‐occurrence groups with a cluster analysis based on co‐occurrences data. Species responses to neighbours were calculated at species‐group level and differences between groups were analyzed with linear mixed effect models. Seven morphological traits were quantified on herbaceous species to assess differences in size and growth‐rate traits between groups. The five delimited co‐occurrence groups had specific responses to the effects of the shrubs, with two groups showing dominant negative indirect responses and three other groups dominant positive indirect ones. The direct and indirect responses could easily be related to co‐occurrence patterns. Short‐ and long‐term effects of the two shrubs were also highly species‐group specific but need further ecological measurements to be fully understood. Co‐occurrence groups had significantly different size traits with higher values for the species group the most positively associated to neighbours and showing an indirect competition outcome, but there were no significant differences in growth‐rate traits between groups. The results of our study show a high consistency between short‐term experimental assessments and long‐term observational assessments of plant interactions and brings additional evidence that size traits are more explanative than growth‐rate traits in plant–plant interactions.