ecological modelling 212 ( 2 0 0 8 ) 522–527 available at www.sciencedirect.com journal homepage: www.elsevier.com/locate/ecolmodel Asynchronous and synchronous updating in individual-based models Geoffrey Caron-Lormier a,* , Roger W. Humphry b , David A. Bohan a , Cathy Hawes b , Pernille Thorbek c a Rothamsted Research, West Common, Harpenden, AL5 2JQ, UK b Scottish Crop Research Institute, Invergowrie, Dundee, DD2 5DA, UK c Syngenta Research Centre, Jealott’s Hill, RG42 6EY, UK article info Article history: Received 8 July 2007 Received in revised form 25 October 2007 Accepted 26 October 2007 Keywords: IBM Scheduling Simulation model Ecological modelling Trophic interaction Cellular automata abstract The use of mathematical and simulation models is widespread in ecology, and individual- based models (IBMs) have proved valuable for exploring individually-explicit interactions and behaviour. The success of a model will depend upon its design and the different assumptions made during construction. In particular, methods implemented in the model to deal with interactions between objects are of fundamental importance for producing appropriate results. Asynchronous and synchronous scheduling are two methods for updating object characteristics during interaction. The consequence of these updating methods has been investigated for cellular automata, but not for IBMs. Here, we assess the two methods for their potential to give different results in a deliberately simple IBM. We show that the two methods produce different results, particularly at high population densities and for increasing interaction complexity (e.g. increasing numbers of trophic levels). This work appears to be the first evidence of the importance of scheduling methods on emergent properties for individual-based models and consequently individually-explicit interactions and behaviour in ecology. © 2007 Elsevier B.V. All rights reserved. 1. Introduction Simulation models are widespread in population ecology (Grimm and Railsback, 2005). They are useful for predict- ing mean and variation in population abundance that result from competitive, trophic or other behavioural interactions. For instance, they can be used to investigate prey–predator dynamics (Pineda-Krch et al., 2007), pest–crop interaction (Brown et al., 2007; Nibouche et al., 2007), weed competition (Holst et al., 2007) and animal spatial behaviour (Mirabet et al., 2007). ∗ Corresponding author. Tel.: +441582 763133; fax: +441582 760981. E-mail addresses: carong@bbsrc.ac.uk (G. Caron-Lormier), Roger.Humphry@scri.ac.uk (R.W. Humphry), bohand@bbsrc.ac.uk (D.A. Bohan), cathy.hawes@scri.ac.uk (C. Hawes), pernille.thorbek@syngenta.com (P. Thorbek). The underlying structure and coding of these models is important, and difference in structure can lead to distinct sets of results even for the same biological interactions (Grimm and Railsback, 2005). Depending on the method, authors might not infer the same conclusions from their model. Furthermore, the effects of adopting particular methods on simulation model output are not widely appreciated, and modellers typically choose methods based on personal opinion and coding expe- rience rather than knowledge either of the effect on the mean and variation of the result or the duration of simulations. Recent work in cellular automata simulation models shows that the method of updating the model, within a time-step, 0304-3800/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.ecolmodel.2007.10.049