Variability in population and community biomass in a grassland community affected by environmental productivity and diversity Jan Leps ˇ Leps ˇ, J. 2004. Variability in population and community biomass in a grassland community affected by environmental productivity and diversity.  / Oikos 107: 64  /71. Temporal variability of the biomass of individual populations and total community biomass were studied in a species rich meadow (35  /40 species 0.25 m 2 plot). Communities were experimentally subjected to fertilization and removal of the dominant Molinia caerulea in a factorial design. The typical dominance structure, with few dominants and many sub-ordinate species, developed in all the treatments. Variability was measured by the coefficient of variation betweenyears, from the fifth to the eighth year of the experiment (to avoid the initial response to the manipulations). Dominant removal increased the diversity measured by the reciprocal of the Simpson index, while fertilization decreased both species number and diversity and also caused a shift in community composition. The variability of both the total community and of individual species was higher in the fertilized plots. The coefficient of variation decreased with species mean biomass in all the plots. In non-fertilized plots, the dominant species had lower variability than total biomass. The biomass values of individual species fluctuated in a concordant mannerover the years (i.e. were positively correlated). All of these factors will decrease the strength of the expected portfolio effect. It is argued that the effect of environmental productivity on variability is more pronounced than the effect of diversity. Jan Leps ˇ, Department of Botany, Faculty of Biological Sciences, Univ. of South Bohemia, and Inst. of Entomology, Czech Academy of Sciences, Na Zlate ´ stoce 1, CZ-370 05 C ˇ eske ´ Bude ˇjovice, Czech Republic (suspa@bf.jcu.cz). The idea that diversity begets stability has been prevalent in ecological literature for a long time (Odum 1953, Elton 1958). For example, the well-known Shannon Wiener index of diversity H? was introduced to ecology by MacArthur (1955) as an index of stability. The idea behind it was that, if there were more redundant species in a community, the failure of one of them could be compensated for by the other species present. In fact, this is a mechanism, which roughly corresponds to what is today known as the portfolio effect (Doak et al. 1998, Tilman et al. 1998). Various mechanisms leading to increasing stability with increasing diversity were sug- gested (Yachi and Loreau 1999, McCann 2000, Loreau et al. 2002), however, all are (explicitly or implicitly) based on the differential response of species to environ- mental fluctuation. A competing view, however, suggests that the life-history characteristics of the constituent species, and particularly of dominants, are the most important determinants of community functioning, in- cluding its stability (Leps ˇ et al. 1982, MacGillivray et al. 1995, Wardle et al. 2000, Grime 2001). Completely different view was presented in May’s (1973) classic book. May used mathematical models to show that stability should decrease with system complex- ity. However, a closer look shows that, in fact, there is a decrease in the probability that the equilibrium of a system with randomly generated interaction coefficients will be stable a in Liapunov sense. How simple models with randomly generated coefficients correspond to ecological reality and what is the relationship between Accepted 19 February 2004 Copyright # OIKOS 2004 ISSN 0030-1299 OIKOS 107: 64  /71, 2004 64 OIKOS 107:1 (2004)