A useful model of auxin transport in the root apex.

A popular quotation among computational modelers is that “Essentially, all models are wrong, but some are useful” (Box and Draper, 1987). Like themodels referred to, thestatement itself is an oversimplification of the process of computational modeling, but it is arguably useful, for example, as an introduction to—and an easyway of remembering—amore complex discourse on some of the fundamental principles of modeling. Saying that all models are “wrong” is a clever abbreviation for noting that all models are based on a host of assumptions, many of which are undoubtedly inaccurate to varying degrees. Embedded in any model is a list of parameters or input variables that are estimated, which often translates to “wholly unknownandpurely guessedat.” Evenso, such models may be extremely useful, if when compared with real-world situations they accurately predict the existence of previously unknowncomponents or unrecognized features of the process or system under investigation and help toguideproductive avenuesof subsequent research. Some would go so far as to turn this quotation on its head and say that models are useful only when they are wrong, implying that it is only when a model disagrees with empirical data that we learn something new and gain insight into the real world. A case in point is presented by Band et al. (862–875) in a new model of auxin transport in the Arabidopsis thaliana root apex. The model employs a system of ordinary differential equations to describe auxin dynamics based on actual cell geometries, obtained through a newly developed method of extracting information from confocal microscopy images, and the use of the DII-VENUS auxin reporter (see figure). The authors incorporated localization patterns of PIN auxin efflux transporters (based on previously published reports and their own immunolocalization studies) and tested various model predictions against empirical data obtained through the use of DII-VENUS. The results showed that patterns of auxin accumulation based on DII-VENUS, which is closely related to auxin concentrations, are more complex than those obtained through the use of the more indirect auxin response reporter DR5:GFP. Importantly, the observed pattern of auxin accumulation could not be predicted based solely on PIN transporters, as previousmodels have suggested. The authors found that the model more accurately described auxin distribution when expression patterns of the auxin influx carriers (AUX1, LAX2, and LAX3) were included, and model predictions were supported by experiments with an aux1 mutant line. The results suggest that the polar PIN transporters control the direction of auxin movement, while the nonpolar AUX/LAX transporters are crucial for determiningwhich cells have high auxin levels. An online graphical user interface is presented called SimuPlant (www.simuplant.org), where readers can further explore the workings of the model and how the dynamic auxin and DII-VENUS distributions are affected by the presence and distributions of the auxin carriers and by other model parameters. This provides greater opportunity for understanding the model and assessing its usefulness. Like all models, this one is still inaccurate in some respects; for example it considers auxin movement in a static domain of root tip geometry, which clearly is not the case in a growing root tip. Although auxin dynamics may be affected by cell proliferation and growth, the authors argue that the assumption is reasonable because the time scale of auxin transport is much shorter than that of growth and cell movement. Future studies may consider the dynamic interplay between auxin distributionpatternsandcell growth inachanging environment, in response to a variety of environmental signals. Meanwhile, the model as presented is highly useful, as it provides insight into the important role of the AUX/LAX transporters in auxin distribution and suggests a number of avenues for further investigation.