Reflectance Spectroscopy Classifies Rootstock of Peaches
Horticulturists have been grafting one species of fruit tree onto another for hundreds of years, even as far back as Roman times. The rootstock, or portion that is underground, provides an established root system onto which a closely related species (the scion) may be grafted for its desirable leaves, flowers, or fruit. Rootstock is often chosen to suit a particular soil or for its resistance to pests and disease, but its use offers many other advantages. Rootstock enables orchards with greater density to be planted and significantly reduces the time before a scion can begin to bear fruit as compared to a seedling of the same variety.
Also, the choice of rootstock can significantly affect the properties of the fruit borne by the scion, from nutritional composition to factors related to quality and maturity including sugar content and flesh firmness. Knowledge of the parent tree’s rootstock for each fruit at the warehouse could allow the optimum storage conditions to be identified, and help send that fruit to market at the peak of ripeness.
Though rootstock information is very rarely provided with fruit entering the commercial food chain, a new study from the University of Pisa, Italy, demonstrates for the first time ever that nondestructive spectral analysis of peach fruits can be used to successfully classify the rootstock from which the fruit was grown.
Knowing that many of the factors determining fruit quality and maturity manifest in the reflectance spectra of fruit skins, and that these factors are heavily dependent on the rootstock used, the team hypothesized that reflectance spectra of a fruit’s skin could be used to determine the parent tree’s rootstock.
The team looked at a single variety of peach fruit that had been grafted onto four different types of rootstock, harvested from 50 trees for a total of 630 peaches. High resolution reflectance spectra were acquired for each peach from 500-900 nm using an HR-series spectrometer configured with an HC-1 grating and a 25 μm slit (~1.1 nm resolution FWHM), a tungsten halogen light source, and a reflection probe.
The spectra were then preprocessed and analyzed using multivariate signal processing to determine whether classification of rootstock type could be automated. The team’s approach used multiple hypothesis testing and a nonparametric multivariate classifier to classify the rootstock. The nonparametric classifier allows the multivariate statistical distribution of the data spectra of each rootstock type to be estimated from the training data spectra without assuming any a priori model; such distributions are then tested against the whole set of data spectra to provide the classification outcome.
While the spectra show both considerable variation and similarity for the four rootstocks, a surface plot of the discriminant functions provided by the nonparametric classifier displayed in a 2D subspace shows that automated discrimination of the different rootstocks may be achieved. When tested, the nonparametric multivariate classification analysis achieved classification accuracy of nearly 80%, showing excellent promise as a method to automate nondestructive testing and classification of rootstocks based on skin reflectance spectra.
With an Ocean Optics reflectance test system in hand and the power of multivariate signal processing for analysis, warehouse staff may someday be able to quickly scan an incoming pallet of peaches to determine how to store or route those fruits so that they arrive at the store at the peak of ripeness, simply by understanding their roots.
Learn more about how reflectance spectroscopy and chemometrics can be used for fruit testing: Reflectance Spectroscopy Reveals the Variety and Sweetness of Apples
Matteoli, Stefania, et al. “Automated Classification of Peach Tree Rootstocks by Means of Spectroscopic Measurements and Signal Processing Techniques.” IEEE Transactions on Instrumentation and Measurement 65.3 (2016): 724-726.
Matteoli, Stefania, et al. “A Spectroscopy-Based Approach for Automated Nondestructive Maturity Grading of Peach Fruits.” Sensors Journal, IEEE15.10 (2015): 5455-5464.