Relationship between observations in mini-rhizotrons and true root length density

This project has resulted in the development of two models which can be applied to the mini-rhizotron technique for root observations. The models are based on probabilistic assumptions concerning root growth directions in the soil. The models apply to the average number of roots which intersect the wall of several tubes buried in the soil, and not to individual observation tubes. The usefulness of the mini-rhizotron technique has been expanded to include indications of the orientation of the root system through the ratio hypothesis. The conversion of root counts to root length density (RLD) has been given a mathematical basis, with few assumptions. The primary disadvantage of the system is the number of samples required.

By determining the ratio of the number of roots which intersect the top to the number which intersect the bottom, the model predicts the direction of deviation from a random orientation. Observations made on the wall of a trench confirmed the horizontal orientation of a cotton root system and the existence of upward growth which had been predicted by the model from observations in mini-rhizotrons.

A model, relating the number of root intersections on a mini-rhizotron to the bulk soil RLD, predicted a linear relationship. The average length to associate with each intersection was determined assuming that root growth was affected only after an intersection occurred, that roots can be represented by straight line segments, and that root growth direction followed some probability density function. The assumption that roots can be represented by straight lines can be removed if the tortuousity of the path of root growth is known. The length can be weighted by the probability of root growth in that direction, if it is known.

The correlation coefficient between RLD determined by applying the model to mini-rhizotron observations and that from soil cores was low but significant at the 99% level, when all treatments and angles were considered. The correlation was largest, 0.70, in the dryland treatment at the 30 installation angle. The higher correlation in the dryland treatment may have resulted from of the reestablishment of the natural soil structure at the interface as the soil dried.

The statistical properties of the mini-rhizotron system emphasize the need for a large sample size. The number of samples required to detect specific differences in RLD is large for both soil cores and mini-rhizotrons when the difference is small. The decision about the number of tubes to install will be affected by the reported variance, the magnitude of the expected RLD, the difference in RLD which is important to the project, and the resources of the project. The variance of the mini-rhizotron was larger than soil cores in this project, but in another project reported in the literature this result was reversed.