Maize root complexity analysis using a Support Vector Machine method
Introduction
The ability of plants to grow and produce seeds is directly related to a healthy, functional and efficient root system. Generally, root complexity and root development depend on genetic and environmental factors and their interactions (O’Toole and Bland, 1987). To assess the genetic basis of root complexity, earlier research determined the Fractal Dimension (FD) of thousands of maize roots recovered from specifically designed field trials using images of the roots (Bohn et al., 2006). A combined analysis of molecular linkage information and FD results led to the identification of Quantitative Trait Loci (QTL) for FD on most of the ten maize chromosomes. QTL are regions in the genome that carry genes involved in the inheritance of a quantitative trait, in this case root complexity. The FD has been shown suitable to describe the complexity of natural objects (Mandelbrot, 1983). A considerable amount of work has been done to capture biological complexity using FD, including studies on root systems (Tatsumi et al., 1989, Lynch et al., 1993, Shibusawa, 1994, Nielsen et al., 1997, Masi and Maranville, 1998, Oppelt et al., 2000, Eghball et al., 2003, Walk et al., 2004, Lontoc-Roy et al., 2006, Soethe et al., 2007), soil clod formation (Shibusawa, 1992), shoot systems and canopies of young trees (Morse et al., 1985, Foroutan-pour et al., 1999), seaweeds (Kubler and Dugeon, 1996), plant foliage (Da Silva et al., 2006), sponges (Abraham, 2001), neurons (Fernandez et al., 1994), and fungal mycelia (Mihail et al., 1995).
A disadvantage of the use of FD is that the complexity of the whole root as contained in gray scale images is captured in a single indicator. Therefore as an alternative, a method was devised which transforms the two-dimensional gray scale image into a set of parameters. This was accomplished by drawing circles around the known centre location of a root image, and to accumulate the intercepting pixels of these circles with the root branches. This method yielded a characteristic function where the accumulated number of intercepting pixels was plotted against the radius of the circles. This characteristic function was approximated by fit curves and the parameters of these curves were used to classify the roots among their original genotypes using the Support Vector Machine (SVM) algorithm (Vapnik, 1995). The SVM method is essentially a binary classifier based on finding the maximal margin hyper-plane between two or more classes (Burges, 1998, Suykens and Vandewalle, 1999). The SVM method has been applied in a variety of applications such as in weed and nitrogen stress detection (Karimi et al., 2005), tissue classification (Furey et al., 2000, Pavlidis et al., 2004), face detection (Osuna et al., 1997), gene selection for cancer (Guyon et al., 2002), as well as shape extraction and classification (Cai et al., 2001).
The objective of this study was to develop an alternative method of root image analysis, based on the Support Vector Machine method, enabling classification of maize roots among their original genotypes.
Section snippets
Materials and methods
Maize plants were grown in Urbana, IL, USA, using an incomplete block design with 235 entries (genotypes), 2 replications, and 47 incomplete blocks at 5 entries per block. Each plot was a single row measuring 4.6 m in length at a distance of 0.76 m separating the rows. Plots were composed of 25 plants/row or 71,525 plants/ha. The roots were harvested at the R1 (silking) stage. The first plant per row was discarded and the next five consecutive plants were trimmed at the third node and uprooted
Results and discussion
Although the order of the chosen polynomial shown in Eq. (1) was three, the curve fitting process showed that the value of parameter ‘a’ was consistently close to zero, and therefore the value was ignored. This left nine potentially useful features from the curves, being “b, c, d”, “A, B, C, D” and MaxPoint as well as Density.
To determine the most influential parameters, among the major features from parameter group 1 (b, c, d), group 2 (A, B, C, D) and MaxPoint, experiments were carried out
Conclusions
The complexity of maize roots was captured in images from which root characteristic functions were derived by drawing circles around the centre of the roots and accumulating the number of circle intercepting pixels as a function of the radius of the circle. The characteristic functions were approximated by fitting two curves and a characteristic MaxPoint where these two curves intersected. In addition to the fit curve related parameters, a Density parameter was evaluated. The curve related
Further research
The SVM algorithm was trained using 235 classes (genotypes). When a new unknown root is evaluated the network will classify this root in a class that closely resembles the phenotype of the root, represented in its architecture and complexity. An interesting extension of the method would be to evaluate crosses among the genotypes and to evaluate if the offspring classifies closely (and equally) to its parents.
The fact that the three main parameter were d, A and MidPoint indicates that the
References (32)
- et al.
Temperature dependent changes in the complexity of form of Chondrus crispus fronds
Journal of Experimental Marine Biological Ecology
(1996) - et al.
Advances in the acquisition and analysis of CT scan data to isolate a crop root system from the soil medium and quantify root system complexity in 3-D space
Geoderma
(2006) - et al.
Fractal geometry of diffuse mycelia and rhizomorphs of Armillaria species
Mycological Research
(1995) - et al.
Genotypic variation in crop plant root systems
Advances in Agronomy
(1987) Fractals in clods formed with rotary tiller
Journal of Terramechanics
(1992)- et al.
Root tapering between branching points should be included in fractal root system analysis
Ecological Modelling
(2007) The fractal branching of an arborescent sponge
Marine Biology
(2001)- et al.
Genetic evaluation of root complexity in maize
Acta Agronomica Hungarica
(2006) - et al.
Multicategory classification by support vector machines
Computational Optimization and Applications
(1999) A tutorial on support vector machines for pattern recognition
Data Mining Knowledge Discovery
(1998)
Support vector machines for predicting protein structural class
Bioinformatics
A critical appraisal of the box counting method to assess the fractal dimension of tree crowns
Lecture Notes in Computer Science, vol. 4291
Spatial and temporal variability of soil nitrate and corn yield: multifractal analysis
Agronomy Journal
Complexity and scaling properties of amacrine, ganglion, horizontal, and bipolar cells in the turtle retina
Journal of Computational Neurology
Soybean canopy development as affected by population density and intercropping with corn: fractal analysis in comparison with other quantitative approaches
Crop Science
Cited by (22)
Feedbacks between soil penetration resistance, root architecture and water uptake limit water accessibility and crop growth – A vicious circle
2018, Science of the Total EnvironmentCitation Excerpt :Such approaches enabled to show how different root phenotypic traits are related to plant productivity in response to low nutrient and water availability (Miguel et al., 2013; Saengwilai et al., 2014; Gao and Lynch, 2016). In addition to manual quantifications, automated image processing may be used to extract more complex information about the structure and spatial configuration of entire root systems or single roots at high throughputs (Zhong et al., 2009; Grift et al., 2011; Bucksch et al., 2014; Colombi et al., 2015). These studies clearly demonstrated that combining plant phenotyping with quantifications of soil properties is a suitable approach to improve our understanding on how belowground processes affect plant growth and crop productivity.
Automatic individual identification of Holstein dairy cows using tailhead images
2017, Computers and Electronics in AgricultureCitation Excerpt :Each one of these classifiers was summarized in the following sections. Support vector machine (SVM) (Zhong et al., 2009; Rumpf et al., 2010) is a state-of-the-art classifier that applies appropriate nonlinear mapping of the data into a sufficiently high dimensional space. Two classes can be separated by an optimum hyperplane.
Plant root spatial distribution measurements based on the hough transformation
2014, NeurocomputingCitation Excerpt :It is important to study the spatial distribution of roots for several reasons: it can reflect root growth conditions, different distributions can provide a classification for distinguishing among different plant species, nutrient absorption can be promoted by changing the root architecture in the soil, and it has important theoretical value for protecting slopes, intercropping plants, and increasing production. Complexity and overlapping in root architecture restrict the accuracy of root measurements [3–5]. Both two-dimensional (2D) [6–8] and three-dimensional (3D) [9–14] quantitative studies of root architecture have been conducted, which include manual methods, Minirhizotron technology, CT technology, root analysis software and root reconstruct methods.
From lab to field, new approaches to phenotyping root system architecture
2011, Current Opinion in Plant BiologyCitation Excerpt :To improve throughput in analyzing excavated roots, image analysis methods coupled to data mining approaches can be applied to characterizing RSA. For example, a support vector machine method using single top view digital images of excavated maize roots was reported to correctly classify more than 99% of roots among 235 genotypes [69]. Although of proven utility, excavation approaches are labor-intensive, they destroy much of the RSA information including that of fine roots, and they do not allow repeated observations of the same plant.
Digitalizing breeding in plants: A new trend of next-generation breeding based on genomic prediction
2023, Frontiers in Plant Science