- Spatial patterns of plant association in grazed and ungrazed shrublands in the semi-arid Karoo - 253 Journal of Vegetation Science 11: 253-258, 2000 © IAVS; Opulus Press Uppsala. Printed in Sweden Abstract. The investigation of vegetation pattern and plant association by spatial statistics has become increasingly popu- lar among plant ecologists. Recently, Individual-centered analy- sis (ICA) has been introduced as a new tool for analysis of multi-species co-occurrence patterns. We tested this new tech- nique by applying it to spatial data from grazed and ungrazed shrub communities in the semi-arid Great Karoo, South Af- rica. There were substantial but complex and scale-dependent differences in pattern between grazed and ungrazed vegeta- tion. Unpalatable species that increase in abundance in grazed vegetation possibly play a key role in the change of vegetation pattern. At small scales we found indications of aggregation (< 30 cm) at the ungrazed, but of repulsion (30 - 40 cm) at the grazed site. An additional non-random pattern at 60 - 170 cm at the grazed site was probably due to the clumped distribu- tions of some species on broader scales. We show that the interpretability of ICA results is improved when the actual observed and expected frequencies of species combinations are added to the program output. The main strength of ICA is that it has the potential to detect association patterns that involve more than two species. Keywords: Aggregation; Community structure; Disturbance; Grazing; Heterogeneity; Individual-centered analysis; Point pat- tern analysis; Scale; Spatial statistics; Vegetation composition. Abbreviations: CSR= complete spatial randomness; ICA = Individual-centered analysis; MCI = Mean compositional in- formation. Nomenclature: Chesselet & de Wet (1997) for succulents; Gibbs Russell et al. (1987) for non-succulents. Introduction Ecologists that have dedicated themselves to the study of biological communities must in some way address the problem of the tremendous complexity of their objects of study. A rather extreme way is to lay down one’s arms by declaring the community as inappropriate level of study (Lawton 1999), another is to go one step back and to first demonstrate that the community is not merely a product of chance (Harvey et al. 1983; Drake 1990; Weiher & Keddy 1995; Wilson 1995a). Detecting non-random patterns in communities (e.g. Silvertown & Wilson 1994; Wilson & Roxburgh 1994; Weiher et al. 1998) justifies the scale of study and at the same time it is likely to generate hypotheses that can be experimentally tested. Tools for the investigation of spatial patterns in vegetation have in the last decade become more power- ful through development of spatial statistical techniques where point maps of plant individuals are analysed for deviation of the spatial distribution of plants from complete spatial randomness (CSR). A popular tech- nique of point pattern analysis has been Ripley’s K- function (Ripley 1981, 1987) by which pattern is ana- lysed using circles of increasing radius so that the characteristic scales of pattern can be detected (see e.g. Prentice & Werger 1985; Skarpe 1991; Haase 1995; Goreaud & Pélissier 1999). Analyses are restricted, though, to single- or two-species patterns. Recently, Podani & Czárán (1997) introduced a new technique, the so-called Individual-centered analysis (ICA), that also uses variable radii to analyse pattern but, unlike the K-function, focuses on the diversity of species combinations rather than on one or two-species pat- terns. Individual-centered analysis may therefore be expected to detect non-random multi-species patterns in point maps of plant communities. To investigate the utility of this new technique, we applied it to data from semi-arid shrublands in the Great Karoo, South Africa. Karoo vegetation is very suitable for point pattern analyses since vegetation is sparse and largely dominated by a few shrub species. Thus, sampling and computational effort can be re- duced without losing much information by restricting analysis to the most abundant species. An important feature of Karoo vegetation is the ubiquity of grazing by domestic livestock, which, like in many other semi-arid ecosystems (Schlesinger et al. 1990; Mil- chunas & Lauenroth 1993), is a major cause of veg- etation change and degradation (Milton et al. 1994). To investigate the effect of grazing on spatial pattern, our study included data both from grazed and ungrazed shrublands. Spatial patterns of plant association in grazed and ungrazed shrublands in the semi-arid Karoo, South Africa Bossdorf, Oliver, Schurr, Frank & Schumacher, Jens * Institute of Ecology, University of Jena, Dornburger Str. 159, D-07743 Jena, Germany; * Corresponding author; Fax +493641949402; E-mail jens.schumacher@uni-jena.de