Feature

Programming Plants

Computer simulation of plant root structures can provide new insights into plant function and health.

By Terry N. Brown

Plant root systems are easily overlooked, but they are just as important to a plant's development as the leaves, flowers and seeds above ground. As well as supplying the plant with water and nutrients from the soil, roots anchor and support the plant, increase soil resistance to erosion, influence a plant's competition with its neighbours, and often host fungi and bacteria that perform specialist tasks of benefit to the whole plant.

Plants use a considerable amount of the energy they acquire from photosynthesis in maintaining their root systems, often feeding root-consuming insects and rots with raw materials that could otherwise contribute to above ground production. The positioning and contact between roots of adjacent plants may influence the spread of root diseases that can seriously reduce plant productivity. Clearly a detailed understanding of plant root behaviour is required for effective and efficient plant growth management.

Even for people with extensive knowledge of overall plant behaviour, understanding of the root system is often based on assumption and deduction rather than experience and observation. Plant roots are notoriously difficult to examine in situ. The patience and precision of an archaeological excavation is required to uncover a root system without damaging or losing the finer roots. Indirect techniques such as core sampling and trenching are useful, but inevitably conceal some details. Buried sections of transparent pipe allow growing roots to be observed directly, but these often distort the system they're intended to reveal, and important fungi and bacteria cannot be identified by simple visual inspection.

Overall, plant roots are complex systems that perform a variety of functions affecting plant growth, and yet our knowledge of the basic mechanisms is limited by the nature of the system itself.

Quick Simulations

As is often the case when expense and practicality restrict the viability of field studies, computer simulation offers a rapid, cost-effective alternative for root system studies. In my work on the spread of disease between the roots of adjacent trees, the long time-scale involved also favours computer simulation -- a trial that might take years in the field can be simulated in hours.

Of course, simulation has its own drawbacks. The simulation must be based on existing knowledge of the system under study, and where this knowledge is sketchy, assumptions and educated guesses may be required. Confidence in the results of a computer simulation can only be justified if it can be shown to correctly predict the outcome of trials that have been carried out in the past, for which the true response is known. Given this ability, a simulation can provide a valuable indication of the behaviour of a system under new, untested conditions without the expense of creating those conditions in the field.

While many studies of roots look only at the total mass of the system, or the amount of water and various nutrients consumed by it, studying the spread of disease by direct inter-root contact between trees requires information on the position of the individual roots.

The overall shape of the root system is controlled by the length of the roots and the number of times they bend and branch. These traits can be characterised by statistical methods, with various random distributions reflecting the root system's structure. The program being developed for this study reads a list of statistical instructions for a specific type of tree, and builds a network of interconnecting points which define the shape of the root system. The list includes formulae specifying the number of forks each root is likely to have, how long each root will be, and how the direction a root is growing in changes along its length.

By tracking the expansion of the network as the tree grows and checking for intersection between networks representing adjacent trees, the potential progress of a root disease can be estimated.

This information is presented in images of the trees and their neighbours. The images are produced by ray-tracing, a technique used to produce many of the computer-generated effects seen in the media. Being able to visualise the output of the simulation in this way is essential in a situation where the complexity of the results makes a concise numerical summary impractical.

Simulation of branch and root architecture is occurring more frequently as suitable computer equipment becomes cheaper. It is applied to other disease systems, fruit load distribution studies, shading studies, and even landscape architecture.

With a trend towards generalised descriptions of trees (making the simulation of new types of tree almost "plug in and go"), and increasingly fast computers (particularly parallel processing systems) it may not be long before the physical structure of complete plant ecosystems can be simulated satisfactorily. Simulation will never replace field observations and experimentation, but as a tool for manipulating ideas and testing hypothesis quickly and cheaply, it is becoming an increasingly flexible and powerful tool.

Terry Brown is a graduate student at Lincoln University.