Any way the wind blows - frequent wind dispersal drives species sorting in ephemeral aquatic communities Bram Vanschoenwinkel, Saı ¨dja Gielen, Maitland Seaman and Luc Brendonck B. Vanschoenwinkel (bram.vanschoenwinkel@bio.kuleuven.be), S. Gielen and L. Brendonck, Laboratory of Aquatic Ecology, Katholieke Univ. Leuven, Ch. Deberiotstraat 32, BE-3000 Leuven, Belgium.  M. Seaman, Centre for Environmental Management, Univ. of the Free State, Mandela Drive, 9300 Bloemfontein, South Africa. Despite an upsurge of interest in spatial interactions between communities and in the impact of dispersal on ecological and evolutionary processes, dispersal patterns and dynamics in natural metacommunities remain poorly understood. Although passive aerial dispersal of freshwater invertebrates is generally accepted, the frequency and relative importance of wind as a vector is still subject of considerable debate. We assessed the importance of wind dispersal in an invertebrate metacommunity in a cluster of 36 temporary rock pools on an isolated mountaintop in South Africa. Wind dispersal was quantified every four days using nine windsocks (about 1.5 m above rock base), placed in the field during one month. Distance to the nearest pool varied from 2 up to 16 m. Wind direction and speed were monitored for the entire period. About 850 propagules (mostly resting eggs) of 17 taxa were captured. The presence of water in the pools (level of exposure of the dormant propagule bank) and the dominant wind direction were the key factors affecting the yield. Wind speed was much less important. Our results suggest that wind dispersal of propagules from temporary aquatic systems is more frequent than previously thought. This may stabilise the metacommunity by mediating gene flow among populations and facilitating rapid (re)colonisation of patches. On the other hand, wind erosion of the dormant propagule bank may lead to egg bank depletion and local extinction. The measured frequent wind dispersal most likely fuels strong species sorting processes ultimately shaping the structure of the local communities as observed in an earlier study. To elucidate the link between local dispersal rates and their contribution to long range dispersal is a major challenge for future research on aerial dispersal of aquatic invertebrates. Different researchers have raised interest in the passive overland dispersal of small aquatic organisms ever since Darwin (1859) pointed at the quasi cosmopolitan distribu- tions of many protozoan and zooplankton species. Tradi- tionally, animals, water connections and wind are recognised as dispersal vectors (Bilton et al. 2001). Maguire (1963) suggested that the link between distribution and dispersal ability can be studied either by observing the colonization of isolated water bodies or by capturing organisms during transport. A third way is to infer dispersal rates indirectly from patterns of genetic differentiation (Neigel 1997, Bossart and Prowell 1998). However as dispersal and gene flow are not synonymous due to local interactions determining establishment success (De Meester et al. 2002), such indirect estimates are often not sufficient to describe and understand dispersal patterns (Bossart and Prowell 1998). Louette and De Meester (2005) have shown that some zooplankton species are able to colonise new water bodies very quickly, relying solely on passive dispersal. The general importance of passive dispersal in zooplankton, the relative importance of different dispersal vectors and the efficiency of subsequent colonization, however, is still subject of considerable debate. Zooplankton are known to disperse readily between aquatic systems that have either permanent (Cladocera; Michels et al. 2001) or temporary connections (large branchiopods; Hulsmans et al. 2006). Additionally a large amount of evidence has accumulated confirming the importance of mostly waterbirds (Figuerola and Green 2002, Green and Figuerola 2005) but also amphibians (Bohonak and Whiteman 1999), reptiles (Lopez et al. 1999) and crayfish (Moore and Faust 1972) as animal vectors. Wind is probably the least studied dispersal vector. Experimental studies which excluded animal vectors sub- scribe the importance of wind as a main vector mediating stochastic dispersal events (Ca ´ceres and Soluk 2002, Cohen and Shurin 2003). Relatively few researchers have tried to measure wind dispersal in a straightforward way by intercepting dispersing propagules. Brendonck and Riddoch (1999) measured only limited short distance dispersal ( B1 m) in Anostraca of temporary rock pools in Botswana during a three day period, using sticky surfaces mounted Oikos 117: 125134, 2008 doi: 10.1111/j.2007.0030-1299.16349.x, # The authors. Journal compilation # Oikos 2007 Subject Editor: Jeremy Fox, Accepted 4 September 2007 125