X-ray computed tomography to quantify tree rooting spatial distributions

Abstract

Poor root development due to constraining soil conditions could be an important factor influencing health of urban trees. Therefore, there is a need for efficient techniques to analyze the spatial distribution of tree roots. An analytical procedure for describing tree rooting patterns from X-ray computed tomography (CT) data is described and illustrated. Large irregularly shaped specimens of undisturbed sandy soil were sampled from various positions around the base of trees using field impregnation with epoxy resin, to stabilize the cohesionless soil. Cores approximately 200 mm in diameter by 500 mm in height were extracted from these specimens. These large core samples were scanned with a medical X-ray CT device, and contiguous images of soil slices (2 mm thick) were thus produced. X-ray CT images are regarded as regularly-spaced sections through the soil although they are not actual 2D sections but matrices of voxels ∼0.5 mm×0.5 mm×2 mm. The images were used to generate the equivalent of horizontal root contact maps from which three-dimensional objects, assumed to be roots, were reconstructed. The resulting connected objects were used to derive indices of the spatial organization of roots, namely: root length distribution, root length density, root growth angle distribution, root spatial distribution, and branching intensity. The successive steps of the method, from sampling to generation of indices of tree root organization, are illustrated through a case study examining rooting patterns of valuable urban trees.

Introduction

In the urban environment, the health of large trees is important for a variety of reasons. In many cases, the cost of replacing ailing or dead trees is significant. Observations along trenches dug for maintenance of subterranean infrastructure consistently showed that the root systems of urban trees were not fully developed compared to what would be expected under natural conditions. The spatial extent of roots of these urban trees is often restricted to the volume of the pit in which they were initially planted. This poor root development is a threat to the growth and health of the above-ground part of the tree. This aberrant root development is believed to decrease the life expectancy of urban trees by 10 to 20 years because it affects water and nutrient uptake and potentially leads to extra-sensitivity to diseases. Another problem is associated with trees whose roots spread excessively towards leaking parts of the water supply network and cause considerable damage, which is expensive to remedy. Therefore, urban management authorities are keen to develop techniques to minimize ill-health of trees and to provide a favorable growth environment. The spatial distribution of tree roots and their relationship with the surrounding environment is important information for understanding how to manage trees to meet these objectives. An additional restriction is that the value of the trees constrains the extent to which the root system of any one tree can be sampled. If the soil lacks cohesion it is not possible to take specimens to estimate root length density with any confidence. To obtain the required information, undisturbed specimens must be extracted and analyzed non-destructively.

Spatial organization of plant or tree roots is most commonly assessed by mapping the contact of roots with horizontal or vertical planes (Tardieu, 1988a, Tardieu, 1988b; Commins et al., 1991; Bruckler et al., 1991; Logsdon and Allmaras, 1991; McBratney et al., 1992; Van Rees et al., 1994; Stewart et al., 1994; Pellerin and Pagès, 1996, Stewart, 1997). This approach appears well suited to the analysis of the spatial patterns of roots, and several of these studies provided conclusive evidence of interaction between root distribution and soil structure. As an alternative to these manual, tedious and destructive investigations, some studies have been focused on automatic and non-destructive assessment of soil structural features. X-ray computed tomography (CT) with low energy medical scanners has been demonstrated as a valuable method for documenting soil macro-porosity (e.g., Warner et al., 1989; Peyton et al., 1992), root growth dynamics (Tollner et al., 1994), and earthworm burrow system development (Joschko et al., 1993, Capowiez et al., 1998). Difficulty in accessing facilities and imaging limitations (e.g., limited spatial resolution of medical X-ray CT devices, low penetration capability leading to low signal/noise ratios, or problems related to soil Fe content when using nuclear magnetic resonance (NMR)) has meant that efforts have seldom been put into reconstructing plant root 3D distributions using non-destructive methods (MacFall and Johnson, 1994; Liu et al., 1994). Heeraman et al. (1997)recently demonstrated that a quantitative description of roots of experimentally grown plants can be achieved with high-energy industrial tomography equipment. However, X-ray CT has not been used before for looking at the in situ morphology of root-networks.

We present here a technique for mapping roots in 3D using low-energy X-ray CT images of undisturbed soil cores. Even though X-ray CT resolution is usually only around 0.5 to 1 mm with medical equipment, this is a sufficient resolution to examine a large proportion of the population of tree roots. The drawback of using a low-energy source of X-rays (leading to low penetration capability and associated artifacts) was counterbalanced by using long integration times. X-ray CT images are used as the basic information to generate an equivalent of horizontal root contact maps, at regularly spaced depths, but representing the projection of a volume rather than an actual section. These maps are used to reconstruct the skeletons of three-dimensional objects assumed to be roots. The resulting connected objects are then used to derive indices of the spatial organization of roots.