Journal of Ecology 2009, 97, 603–608 doi: 10.1111/j.1365-2745.2009.01503.x © 2009 The Authors. Journal compilation © 2009 British Ecological Society Blackwell Publishing Ltd FORUM Species diversity and productivity: why do results of diversity-manipulation experiments differ from natural patterns? Lin Jiang 1 *, Shiqiang Wan 2 and Linghao Li 2 1 School of Biology, Georgia Institute of Technology, 310 Ferst Drive, Atlanta, GA 30332, USA; and 2 Key Laboratory of Vegetation and Environmental Change, Institute of Botany, Chinese Academy of Sciences, Beijing 100093, China Summary 1. Experiments that directly manipulate species diversity often report a positive diversity effect on productivity, whereas observations of natural communities reveal various productivity–diversity relationships and nutrient addition to natural plant communities generally results in negative productivity–diversity relationships. 2. We hypothesize that this apparent paradox may be potentially explained by the reduced roles of complementarity and positive selection effects, and the increased importance of competitive exclusion in natural communities compared to diversity-manipulation experiments. This hypothesis arises from the difference in species distribution and abundance patterns between immature synthetically assembled communities in diversity-manipulation experiments and more mature natural communities. 3. Our hypothesis applies best to small-scale studies within homogenous habitats and comple- ments the environmental heterogeneity hypothesis that explains diversity–productivity patterns across heterogeneous habitats. 4. Synthesis. Our analysis highlights important differences between synthetic communities in diversity-manipulation experiments and natural communities that may translate into different diversity–productivity patterns, and cautions against indiscriminate extrapolations of results of diversity-manipulation experiments to natural communities. Key-words: biodiversity and ecosystem functioning, competitive exclusion, complementarity, environmental heterogeneity, productivity, selection effects, species diversity Introduction During the past two decades ecologists have devoted con- siderable effort to understanding functional significances of biodiversity (Kinzig et al. 2002; Loreau et al. 2002; Hooper et al. 2005; Srivastava & Vellend 2005). Much of the experimental work in this area has been conducted in synthetically assembled plant communities involving direct manipulations of plant diversity. This research has thus far revealed generally positive effects of biodiversity on ecosystem functioning and in particular positive effects of plant species diversity on primary productivity (Balvanera et al. 2006; Cardinale et al. 2006, 2007). This positive relationship between plant species diversity and productivity, however, is in striking contrast with emerging patterns in natural communities that are revealed by studies exploring productivity as a driver of species diversity – a perspective that has historically received more attention (e.g. Hutchinson 1959; Connell & Orias 1964). Such studies include observations of diverse forms of productivity–diversity relationships (Waide et al. 1999; Mittelbach et al. 2001; Gillman & Wright 2006), and nitrogen fertilization experiments that reported nearly ubiquitous negative productivity–diversity relationships (Gough et al. 2000; Crawley et al. 2005; Suding et al. 2005). Understandably, these patterns of different diversity–productivity relationships have led to debates on whether results of diversity-manipulation experiments are relevant for natural communities (Thompson et al. 2005; Hector et al. 2007) and some conceptual models have been proposed to reconcile these different patterns (Loreau et al. 2001; Schmid 2002; Hector et al. 2007). The central message *Correspondence author. E-mail: lin.jiang@biology.gatech.edu