ecological modelling 205 ( 2 0 0 7 ) 270–275 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/ecolmodel Short communication Quantifying positional importance in food webs: A comparison of centrality indices Ferenc Jord ´ an a,b,* , Zs ´ ofia Benedek c ,J´ anos Podani c a Collegium Budapest, Institute for Advanced Study, Szenth´ aroms ´ ag u. 2, Budapest, Hungary b Animal Ecology Research Group, HAS, Hungarian Natural History Museum, Budapest, Hungary c Department of Plant Taxonomy and Ecology, E¨ otv ¨ os University, Budapest, Hungary article info Article history: Received 10 November 2006 Received in revised form 12 February 2007 Accepted 23 February 2007 Published on line 9 April 2007 Keywords: Food web Network analysis Centrality Keystone species Indirect effect Clustering Ordination Ranking abstract Conservation biology should focus more on the importance rather than the rarity of species, although the definition and quantification of importance are not easy. One approach involves measuring the positional importance (e.g. centrality) of species in ecological inter- action networks to provide a basis for species ranking. However, there are many centrality indices, each reflecting a particular aspect of positional importance and therefore giving a rank order of species different from those provided by alternative formulations. Thus, there is a strong need for comparing the available indices and for examining their relative merits in network analysis. In this paper, we apply 13 centrality indices to the “species” (trophic components) of methodologically comparable trophic flow networks, in order to answer the following questions: (1) What is the disagreement between different indices regarding the rank of a given species in a given network? (2) How is this disagreement in performance influenced by the choice of the network? (3) What is the overall relationship among these indices and, in particular, which are the most similar to degree (the simplest index of all, being equal to the number of links pertaining to a given node)? We compare the 13 indices based on the data of nine networks using metric and rank statistics and multivariate anal- ysis procedures. We conclude that (1) different centrality ranks differ in each network; (2) different webs can be characterized by different relationships between ranks but there is a robust pattern of relationships among the indices, some index pairs behaving very similarly in all networks; and (3) it is the index of closeness centrality which provides a rank most similar to that based on degree. © 2007 Elsevier B.V. All rights reserved. 1. Introduction Future conservation biology needs to be more functional and should be outlined within a multispecies context. For exam- ple, the conservation of rare species should be gradually replaced by conserving the most important species that play key role in maintaining ecosystem functions (Wilson, 1987). ∗ Corresponding author. Tel.: +36 1 22 48 300; fax: +36 1 22 48 310. E-mail address: jordan.ferenc@gmail.com (F. Jord´ an). A fundamental requirement in such an approach is the more quantitative, less subjective and, hopefully, more predictive study on keystone species (Paine, 1969; Mills et al., 1993; Power et al., 1996). The importance of species in a community is not easy to define and is even more complicated to quantify. The problem is discussed within a community ecology framework: we 0304-3800/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.ecolmodel.2007.02.032