LETTER Stochastic trajectories of succession initiated by extreme climatic events J. Kreyling, 1 * A. Jentsch 2 and C. Beierkuhnlein 1 1 Biogeography, University of Bayreuth, Universita ¨ tsstraße 30, D-95440 Bayreuth, Germany 2 Disturbance Ecology, University of Bayreuth, Universita ¨ tsstraße 30, D-95440 Bayreuth, Germany *Correspondence: E-mail: juergen.kreyling@uni-bayreuth.de Abstract Deterministic or rule-based succession is expected under homogeneous biotic and abiotic starting conditions. Effects of extreme climatic events such as drought, however, may alter these assembly rules by adding stochastic elements. We monitored the succession of species composition of 30 twin grassland communities with identical biotic and abiotic starting conditions in an initially sown diversity gradient between 1 and 16 species over 13 years. The stochasticity of succession, measured as the synchrony in the development of the species compositions of the twin plots, was strongly altered by the extreme warm and dry summer of 2003. Moreover, it was independent from past and present plant diversity and neighbourhood species compositions. Extreme climatic events can induce stochastic effects in community development and therefore impair predictability even under homogeneous abiotic conditions. Stochastic events may result in lasting shifts of community composition, as well as adverse and unforeseeable effects on the stability of ecological services. Keywords BIODEPTH, climate change, compositional legacy, drought, ecological memory, temperate grassland. Ecology Letters (2011) 14: 758–764 INTRODUCTION The investigation of temporal dynamics is one of the fundamental aims of ecology, especially in times of global change (Coreau et al. 2009). Deterministic concepts for target communities are increasingly under debate (Chiarucci et al. 2010), although most arguments are hypothetical as long-term time series on vegetation trajectories, especially after extreme climatic events, are rare. The concept of ecological succession refers to more or less deterministic (rule-based) developments in the composition or structure of an ecological community after disturbance (Connell & Slatyer 1977). Ecologists generally used to assume that deterministic succession occurs under identical abiotic starting conditions, especially when residual species and individuals from the former community are present (Turner et al. 1998). Multiple trajectories are presumed to be the result of spatial or temporal heterogeneity, with chance being an important factor mainly in initial phases in the absence of residual species and individuals (Turner et al. 1998; Baasch et al. 2009). Chance, or the concept of stochasticity, is a basic feature of ecological theories (Hubbell 2001; McGill 2010). However, it should not be confounded with lack of knowledge on underlying processes (Clark 2009). Here, we argue that any probabilistic process such as biotic interactions, colonisation or in particular speciation inherits a certain degree of stochasticity (Hubbell 2001; Chave 2004). Assessing the degree of stochasticity in field studies is challenging (Ellwood et al. 2009) and we suggest here that the monitoring of initially identical twin communities represents a valuable novel tool for research on stochasticity in succession. Short-term events may determine the long-term legacy of an ecosystem by directing it towards alternate stable states (Scheffer & Carpenter 2003). Extreme climatic events are such drivers that can alter successional trajectories (Jentsch et al. 2007). In contrast to spatially restricted extreme weather conditions, extreme climatic events may impact large regions (Ciais et al. 2005), and even though residual species and individuals from former communities are generally abundant, they can lead to sustained shifts in successional pathways (Allen & Breshears 1998; Stampfli & Zeiter 2004). It remains unknown if events, such as extreme drought or periods of heavy rain, are resulting in deterministic trajectories at the ecosystem scale or if they rather promote stochastic developments, which may possibly initiate multiple successional pathways. The extreme warm and dry summer of 2003 in Central Europe displays one example of such extreme climatic events. With a statistical recurrence time under current climate conditions of about 10 000 years (Scha ¨r et al. 2004), the summer of 2003 resulted in strong reductions in primary productivity (Ciais et al. 2005) and even complete dieback of drought sensitive species (Bragazza 2008). More diverse communities are theoretically expected to be more resilient against external perturbations (Yachi & Loreau 1999), and regime shifts of ecosystems due to external stressors are less likely to occur in diverse communities (Folke et al. 2004). For temperate grasslands, a linear relationship between species diversity and the recovery after severe drought has been shown (Tilman et al. 2006; van Ruijven & Berendse 2010). Invasion ecology adds evidence that the diversity of a resident community plays a determining role for shifts in species composition, i.e. succession (Sax et al. 2007; Kreyling et al. 2008). More diverse communities generally decrease invasibility at fine spatial scales where all occurring plant individuals interact (reviewed by Fridley et al. 2007). In stochastic niche models, low change in species composition is predicted to result from uniformly low levels of resources or open niches (Tilman 2004). More diverse systems generally fulfil this condition better than less diverse systems (see reviews and meta-analyses by Balvanera et al. 2006; Cardinale et al. 2006). It is, however, highly controversial which facet of diversity (i.e. species richness, functional group richness, presence of key species, presence of specific functional groups such as legumes) is mainly attributed to the observed diversity effects (Hooper et al. 2005). Ecology Letters, (2011) 14: 758–764 doi: 10.1111/j.1461-0248.2011.01637.x Ó 2011 Blackwell Publishing Ltd/CNRS