Machine Learning Algorithms to Predict Forage Nutritive Value of In Situ Perennial Ryegrass Plants Using Hyperspectral Canopy Reflectance Data

Abstract

Nutritive value (NV) of forage is too time consuming and expensive to measure routinely in targeted breeding programs. Non-destructive spectroscopy has the potential to quickly and cheaply measure NV but requires an intermediate modelling step to interpret the spectral data. A novel machine learning technique for forage analysis, Cubist, was used to analyse canopy spectra to predict seven NV parameters, including dry matter (DM), acid detergent fibre (ADF), ash, neutral detergent fibre (NDF), in vivo dry matter digestibility (IVDMD), water soluble carbohydrates (WSC), and crude protein (CP). Perennial ryegrass (Lolium perenne) was used as the test crop. Independent validation of the developed models revealed prediction capabilities with R2 values and Lin’s concordance values reported between 0.49 and 0.82, and 0.68 and 0.89, respectively. Informative wavelengths for the creation of predictive models were identified for the seven NV parameters. These wavelengths included regions of the electromagnetic spectrum that are usually excluded due to high background variation, however, they contain important information and utilising them to obtain meaningful signals within the background variation is an advantage for accurate models. Non-destructive field spectroscopy along with the predictive models was deployed infield to measure NV of individual ryegrass plants. A significant reduction in labour was observed. The associated increase in speed and reduction of cost makes targeting NV in commercial breeding programs now feasible.

Applying RGB- and Thermal-Based Vegetation Indices from UAVs for High-Throughput Field Phenotyping of Drought Tolerance in Forage Grasses

“…Digestibility can be measured routinely with NIRS (near-infrared spectroscopy) analyses, from which parameters as the D-value (digestibility of organic matter in dry matter), WSC (water soluble carbohydrates) or similar feed quality features can be predicted [96]. However, NIRS uses destructive sampling and is too costly and time-intensive for large scale application in breeding programs [97]. Several studies indicate that hand-held hyperspectral sensors can provide a non-destructive and fast alternative, possibly suited for breeding purposes [97][98][99].…”

“…However, NIRS uses destructive sampling and is too costly and time-intensive for large scale application in breeding programs [97]. Several studies indicate that hand-held hyperspectral sensors can provide a non-destructive and fast alternative, possibly suited for breeding purposes [97][98][99]. Hyperspectral sensing from UAVs has been demonstrated to be suitable for estimating forage grass digestibility [96].…”

Bottlenecks and opportunities in field-based high-throughput phenotyping for heat and drought stress

“…This methodology has not been implemented with a row crop; however, it was recently tested with perennial ryegrass. The authors used a hyperspectral radiometer as well as a light shield in order to capture the spectra under stable light conditions from 960 different plants, comprised of 50 experimental perennial ryegrass varieties ( Smith et al , 2020 ). The light shield was manually placed on each plant, and artificial light within the shield was used as the light source.…”

Reference Measurements in Developing UAV Systems for Detecting Pests, Weeds, and Diseases

“…After weighing, the sample is dried in an oven to determine the amount of dry biomass [67]. Cut and dry is the most common reference data collection method in biomass estimation studies, and it has been used in various recent studies [69][70][71][72][73]. The method provides objective reference data, but the cutting phase can include lot of variation.…”

“…The method provides objective reference data, but the cutting phase can include lot of variation. In studies where biochemical parameters such as the nitrogen or digestibility of grass were estimated, the reference samples were analysed in a laboratory, most commonly using the NIRS (near-infrared spectroscopy) technique [69,72,73], providing numerical comparable values.…”