Applied Vegetation Science 19 (2016) 291–303 Are trees of intermediate density more facilitative? Canopy effects of four East African legume trees Anja Linstadter, Zinabu Bora, Adugna Tolera & Ayana Angassa Keywords Bush encroachment; Canopy density; Facilitation; Shading; Sub-canopy; Understorey vegetation Nomenclature The Plant List (http://www.theplantlist.org/; accessed on 2 Mar 2015), except for the polyphyletic, recently revised genus Acacia Miller s.l. (Fabaceae: Mimosoideae) where it follows Kyalangalilwa et al. (2013). Received 5 May 2014 Accepted 6 November 2015 Co-ordinating Editor: David Ward Linstadter, A. (corresponding author, anja. linstaedter@uni-koeln.de) 1 , Bora, Z. (zinabu_bora@yahoo.com) 2 , Tolera, A. (adugnatolera2@gmail.com) 3 , Angassa, A. (ayana.angassa@gmail.com) 1,3 1 Botanical Institute, Range Ecology and Management Group, University of Cologne, Zulpicher Str. 47b, D - 50674 Cologne, Germany; 2 Oromia Pastoral Area Development Commission, P.O. Box 20120, AA Finfinne, Addis Ababa, Ethiopia; 3 Department of Animal and Range Sciences, Hawassa University, P.O. Box 1591, Awassa, Ethiopia Abstract Questions: Do East African legume trees differ in their canopy effects? Have species with intermediate canopy density more pronounced facilitative effects on understorey primary production, community composition and diversity? How are canopy effects related to species’ encroachment status, and modulated by a local aridity gradient? Location: Upland and lowland environments of the Borana rangelands, an arid thornbush savanna in southern Ethiopia. Methods: We harnessed pastoralists’ local ecological knowledge to rank the encroachment status of six legume tree species previously known as Acacia, and correlated this rank to a dendrometric proxy of canopy density. Vegetation releves (1 m²) were placed in sub-canopy and adjacent inter-canopy habitats of four legume tree species that differed in canopy density and encroachment sta- tus (Vachellia bussei, V. drepanolobium, V. seyal, V. tortilis). Using mixed-effects ANOVA, we evaluated effects of tree species and habitat (sub-canopy or inter- canopy) on total, forb and grass biomass, and on species diversity, comparing results for lowland and upland sites. Effects on floristic composition were assessed via PERMANOVA and NMDS. Results: Species’ encroachment status and canopy density were negatively cor- related. Most pronounced facilitative effects (more frequent and larger differ- ences between sub-canopy and inter-canopy habitats) were found for a species with intermediate canopy density (V. bussei). Diversity in its understorey vegeta- tion increased up to 32% (total biomass: 29%, forb biomass: 40%). In contrast, the species with the most open canopy (V. drepanolobium; also the most encroaching species) never exerted significant facilitative effects. In contrast to expectations, canopy effects were more frequent in (climatically less arid) upland environments, probably due to a redistribution of water within the land- scape. Large differences in facilitative effects across tree individuals indicate that a high intraspecific variability of canopy traits and/or small-scale differences in abiotic site conditions have partly overridden species-specific differences in canopy architecture. Conclusion: Our study provides new insights on why encroaching legume tree species may decrease herbaceous production: they tend to have more open canopies, with small facilitative effects on sub-canopy herbaceous production. We thus recommend promoting an open savanna with the tree layer dominated by species with a high facilitative potential to maintain forage provision and spe- cies diversity. 291 Applied Vegetation Science Doi: 10.1111/avsc.12218 © 2015 International Association for Vegetation Science