1371 Direct and indirect interactions co-determine species composition in nurse plant systems Christian Schöb, Cristina Armas and Francisco I. Pugnaire C. Schöb (christian.schoeb@hutton.ac.uk), C. Armas and F. I. Pugnaire, Estación Experimental de Zonas Áridas, Consejo Superior de Investigaciones Científcas, EEZA-CSIC, Carretera de Sacramento s/n, ES-04120 La Cañada de San Urbano, Almería, Spain. Present address for CS: Te James Hutton Inst., Craigiebuckler, Aberdeen, AB15 8QH, UK. Facilitation by nurse plants plays an important role in determining community composition in severe environments. Although the unidirectional efect of nurses on benefciary species has received considerable research interest, nurse- mediated interactions among benefciary species (so-called indirect interactions) are less known. Consequently, com- munity composition in nurse plant systems is generally considered as a simple consequence of the facilitative efect of the nurse even though benefciary species may signifcantly contribute to community assembly and modulate the direct nurse efects on the community. In an observational study we assessed nurse efects and nurse-mediated benefciary inter- actions in two contrasting nurse plant systems in dry environments using a newly developed framework. We quantifed plant–plant interaction intensity using the relative interaction index (RII) at the community and species level for three Retama sphaerocarpa shrub size-classes in a semiarid shrubland and four Arenaria tetraquetra cushion plant communities difering in aspect and elevation in dry alpine gravel habitats. Te observed RII was split into nurse and benefciary efects, and related to individual mass, species frequency and abundance using generalized linear mixed models. Results showed predominantly positive nurse efects and negative benefciary interactions. Te efect size of nurse plants, however, was signifcantly higher than the efect size of benefciary species in both systems. Individual plant mass and abundance of species was dependent on the combined efects of nurse and benefciary species whereas species occurrence was related to nurse efects only. Despite evident diferences, the semiarid and alpine nurse plant systems showed strong functional par- allelisms. We found interdependence between the efects of nurse and benefciary species on benefciary plant assemblages emphasizing their combined role on community assembly in both systems. Our results highlight the need to consider indirect interactions to understand fully plant community dynamics. Since all plants use the same resources, competition for light, nutrients, and water as well as for physical space is common in plant communities (Whittaker 1965). However, an increasing number of studies have shown that at least some species do not only consume resources but can also increase their availability or at least dampen their loss to other species (Callaway 2007) thereby benefting them. For example water in shallow soil layers can be increased by plants due to redistribution of soil water (Prieto et al. 2012) or reduced evaporation caused by shading (Moro et al. 1997), while nutrient availability can be increased by nitrogen-fxing species (Temperton et al. 2007), the efect of mycorrhiza (Nara and Hogetsu 2004, Casanova-Katny et al. 2011) or accretion of soil organic matter (Pugnaire et al. 2004). Tese and other positive efects of plants on resource availability may generally be to their own beneft (Kylafs and Loreau 2011) but, as a side efect, they can also beneft other species. A notable example of this phenomenon is facilitation in nurse plant systems where a dominant species creates environmental conditions that often beneft a large number of subordinate species (Pugnaire et al. 2011). Nurse plants are so termed because of their positive ‘nursing’ efects on their own seedlings (Niering et al. 1963), although current defnitions of nurse plants generally include both intra- and inter-specifc efects. Te positive efect of nurses on subordinate, benefciary species has been widely addressed (reviewed by Callaway 2007). However, community assembly in nurse plant systems and its consequences for species composition and diversity is not only determined by the unidirectional, positive efect of nurses. All individuals sharing resources potentially interact with each other implying also efects of benefciaries on the nurse as well as interactions among benefciaries. Te few studies available on the efect of benefciaries on the nurse suggest a full range of inter- actions: positive (Pugnaire et al. 1996b), neutral (Lortie and Turkington 2008), or negative (Holzapfel and Mahall 1999, Michalet et al. 2011). Furthermore, these inter- actions may also change in sign and intensity with onto- geny and do likely become more competitive with increasing age of the benefciary (Barnes and Archer 1999). Most empirical evidence, however, shows predominantly Oikos 122: 1371–1379, 2013 doi: 10.1111/j.1600-0706.2013.00390.x © 2013 Te Authors. Oikos © 2013 Nordic Society Oikos Subject Editor: James M. Bullock. Accepted 25 January 2013