Ecology, 94(1), 2013, pp. 169–179 Ó 2013 by the Ecological Society of America Identity effects dominate the impacts of multiple species extinctions on the functioning of complex food webs ERIC HARVEY, 1,4 ANNIE SE ´ GUIN, 1,2 CHRISTIAN NOZAIS, 3 PHILIPPE ARCHAMBAULT, 2 AND DOMINIQUE GRAVEL 1 1 Canada Research Chair in Continental Ecosystem Ecology, Universite´ du Que ´bec a ` Rimouski, Rimouski, Quebec G5L 3A1 Canada 2 Institut des Sciences de la Mer de Rimouski (ISMER), Universite ´ du Que ´bec a ` Rimouski, Rimouski, Quebec G5L 3A1 Canada 3 De ´partement de Biologie, Chimie et Ge´ographie, Universite´ du Que ´bec a ` Rimouski, Rimouski, Quebec G5L 3A1 Canada Abstract. Understanding the impacts of species extinctions on the functioning of food webs is a challenging task because of the complexity of ecological interactions. We report the impacts of experimental species extinctions on the functioning of two food webs of freshwater and marine systems. We used a linear model to partition the variance among the multiple components of the diversity effect (linear group richness, nonlinear group richness, and identity). The identity of each functional group was the best explaining variable of ecosystem functioning for both systems. We assessed the contribution of each functional group in multifunctional space and found that, although the effect of functional group varied across ecosystem functions, some functional groups shared common effects on functions. This study is the first experimental demonstration that functional identity dominates the effects of extinctions on ecosystem functioning, suggesting that generalizations are possible despite the inherent complexity of interactions. Key words: biodiversity; ecosystem functioning; food web; functional groups; marine and freshwater food webs; species extinctions; species identity; species interactions. INTRODUCTION There has been a great deal of effort over the past two decades to understand the impact of community structure on ecosystem functions (see reviews in Loreau et al. 2001, Cardinale et al. 2011). In spite of some controversies about interpretations of the data (e.g., Huston 1997), the vast number of experiments per- formed allowed quantitative analyses of the general trend (Balvanera et al. 2006, Cardinale et al. 2012, Naeem et al. 2012) and confirmed that a positive relationship between biodiversity and ecosystem func- tioning (BEF) emerges from simple and conceptually tractable mechanisms (i.e., selection effect and comple- mentarity among species; Loreau and Hector 2001, Loreau 2010). Despite the impressive progress in that research area, there is still a wide array of crucial issues to resolve before we can provide satisfying recommen- dations for ecosystem management (Srivastava and Vellend 2005). Among them, the discipline must recognize the important complexity of natural ecosys- tems (Polis and Strong 1996, Duffy et al. 2007, Schmitz 2010) and, consequently, develop predictive tools to assess the impacts of species loss on ecosystem functioning. The incorporation of trophic interactions to the BEF theory clarified the importance of top-down and bottom-up constraints upon the diversity effect (see Duffy et al. 2007). For instance, in simple tri-trophic food webs, it is predicted that increasing predator diversity will increase primary producer biomass (Byrnes et al. 2006). Theoretical studies in relatively more complex systems have shown that many relations between ecosystem function and diversity could emerge from simple mechanisms (Duffy et al. 2007). The effect of adding plant species, for example, could either enhance (stoichiometric hypothesis; sensu DeMott 1998) or inhibit (resource dilution hypothesis; sensu Root 1973) consumption by herbivores, depending on the mechanism involved. Moreover, an ecosystem perspective explicitly integrating nutrient dynamics can strongly moderate simple predictions such as the effect of removing a predator on resource control by an herbivore (Schmitz 2008). Moving forward in complexity from a food chain to a food web perspective remains a challenging task (Bascompte 2009) for both theoreticians and empiricists. Some theoretical studies of complex food webs suggest- ed that the increasing number of ecological interactions (intraguild predation, omnivory, and indirect effects) with species richness promotes the emergence of diffuse and indeterminate reactions to disturbances (Polis and Strong 1996, Yodzis 2000, Berlow et al. 2009). Accord- ingly, results from experimental BEF studies in multi- trophic communities revealed much higher variability between ecosystems and processes than in simpler Manuscript received 12 March 2012; revised 3 July 2012; accepted 16 August 2012. Corresponding Editor: E. Van Donk. 4 Present address: Department of Integrative Biology, University of Guelph, Guelph, Ontario N1G 2W1 Canada. E-mail: eharve01@uoguelph.ca 169