Modelling the bacterial colonization of leaves

Photo courtesy of Jan Tech

Our world is a quite green world: a sea of trees, bushes, grasses, or, if you happen to live in the Midwest, corn fields… What is less obvious, though, is the fact that this green vastness harbors a huge community of microbes. Yeasts and filamentous fungi are often found on plant surfaces, but the most numerous inhabitants are first and foremost bacteria. Indeed, a centimeter square of leaf can contain as many as 10 millions of them! No worries, most of them are harmless to us or their plant hosts. On the contrary, many are required to maintain a healthy plant environment, by stimulating plant growth or by preventing the plant colonization by pathogens (they compete for the same space and the same resources). 

Because plants are so vital to us (think food, raw materials, landscapes, etc.), there is a real interest in understanding what the microbial contribution to the plant ecosystem is. One lingering question, for instance, is how bacteria colonize the surface of leaves (what we call the phyllosphere). What we do know is that bacteria on leaf surfaces appear as clusters of cells, rather than an even layer of bacteria covering the surface; the mechanisms that lead to this colonization pattern, however, is not well understood. I have already written about this question in a previous post that dealt with the use of bacterial bioreporters. Another way to explore these mechanisms of cluster formation is computer-based modelling, which enables us to test different scenarios and compare it with what has been observed on real plants. 

Bacteria (in green) colonize the surface of a snap bean leaf and form clusters.

In Explaining bacterial dispersion on leaf surfaces with an individual-based model, a new study that I co-authored with Annemieke van der Wal (lead author), Jan-Ulrich Kreft, Wolf Mooij and Johan Leveau, we present a computer simulation that helps explain how bacteria can colonize the surfaces of plant leaves. It’s just fresh out of the (virtual) press here at PLoS ONE! 

To predict the clustering of bacteria growing on a surface, Annemieke has created a so-called individual-based (or agent-based) model¹. It differs from a ‘classical’ model in that it doesn’t try to express the behavior of a whole population using appropriate mathematical expressions. Instead, a population pattern emerges from the behavior of individual agents (for instance bacteria) that obey a few simple rules. In our case, we give the bacteria rules as to how they can grow and form microcolonies. The advantage of such models is that they do not impose any predetermined behavior on the virtual population, but rather allow complexity to be derived from simple, individualistic actions.

Bacteria (in green) clustering next to a leaf stomate.

The main goal of our model was to explain the clustering patterns of bacteria that were observed in a classic paper by Jean-Michel Monier and Steve Lindow (2004), from UC Berkeley. Bacteria can grow on pant leaf surfaces thanks to the availability of plant sugars such as glucose and fructose. These sugars are located in the sap, but some of it diffuses from the interior of the plant to the exterior, at least if water is available on the surface to collect these molecules! For this reason, the presence of water is a key parameter in our model: if the surface is dry, bacteria cannot grow! When water is present, sugar is available to the bacteria; they use it, grow and produce offspring that colonize the surface. Annemieke thus tested the importance of different ‘waterscapes’ on bacterial growth, for instance comparing a continuous water film with individual droplets spread across the leaf surface.

The video below shows a simulation of colonization run with our model. The surface is represented in 2D (right side in the video), and the round elements represent droplets of water. When the droplets turn dark, it indicates that the concentration of sugar has increased. Individual bacteria are seen in green. They will consume the sugars (the color in the droplet becomes white), divide and form microcolonies containing many individuals.

This mechanism (droplet distribution, sugar diffusion, bacterial growth), however, was not sufficient to explain the experimental patterns seen by Monier and Lindow. What was missing was bacterial detachment: when bacteria can detach and disperse from the microcolony in which they originated (as shown in the video), the model predictions match the experimental results very accurately! (See figures below.) To say it differently, if there is no bacterial movement (either active or passive), we cannot explain the experimental observations, namely that we see some large clusters of thousands of cells and many smaller clusters of only a couple of cells. 

 

Experimental results by Monier and Lindow.

Predictions of the model without detachment.

Predictions of the model with bacterial detachment.

The main finding of our study was thus that bacterial detachment from clusters (and re-attachment) is an important mechanism in the colonization of plant leaves, and potentially in the colonization of other natural and artificial surfaces. This gives us a better view of what it means to live a microbe's life!

Notes:

1. The model was built in NetLogo, a platform developed at the Northwestern University (Illinois) that is freely available! 

References:

• van der Wal A., Tecon R., Kreft J.-U., Mooij W. M. and J. H. J. Leveau (2013). Explaining bacterial dispersion on leafsurfaces with an individual-based model (PHYLLOSIM). PLoS ONE 8(10): e75633. doi:10.1371/journal.pone.0075633. 
• van der Wal A., and J. H. J. Leveau (2011). Modelling sugar diffusion acrosse plant leaf cuticles: the effect of free water on substrate availability to phyllosphere bacteria. Environmental Microbiology 13(3): 792-797. DOI: 10.1111/j.1462-2920.2010.02382.x
• Monier J.-M., and S. E. Lindow (2004). Frequency, size, and localization of bacterial aggregates on bean leaf surfaces. Applied and Environmental Microbiology 70(1): 346-355. doi: 10.1128/AEM.70.1.346-355.2004