The gambin model provides a superior fit to species abundance distributions with a single free parameter: evidence, implementation and interpretation

A species abundance distribution (herein ‘ SAD ’ ) describes the abundances of all species sampled within a given community (Ulrich et al. 2010). Because SADs characterise the structure of ecological communities, they constitute the foundation of much (macro-) ecological and biogeographical theory (Preston 1948). Th us, considerable eff ort has been made to characterise empirical SADs in a statistically tractable framework. To date around 30 diff erent SAD models have been published (McGill et al. 2007). Th e two most commonly used are the logseries (Fisher et al. 1943) and the lognormal (Preston 1948). Th e Poisson lognormal ( ‘ PLN ’ ; Bulmer 1974) is usually preferred to the continuous lognormal as the PLN incorporates a sampling theory and the continuous lognormal allows fractional abundances. Th e neutral model of Hubbell (2001) also predicts a SAD, termed the zero-sum multinomial distribution ( ‘ ZSM ’ ). Th e logseries has often been found to provide a poor fi t to empirical SADs (Ulrich et al. 2010; but see White et al. 2012), and the use of the PLN has been criticised, both in general terms (Williamson and Gaston 2005), and particularly in comparative analysis, as the parameters are not as intuitively interpretable as those of the logseries (but see Engen 2001, S æ ther et al. 2013). Th us, there is a need for a model that provides a good fi t to a variety of empirical data, is analytically tractable and possesses an easily interpreted parameter (i.e. one that simply describes the shape of the SAD), which can then be used as a system descriptor in further analyses. In a recent paper, Ugland et al. (2007) © 2014 Th e Authors. Ecography © 2014 Nordic Society Oikos Subject Editor: Joaquin Hortal. Accepted 3 April 2014 Ecography 37: 1002–1011, 2014 doi: 10.1111/ecog.00861