Near-infrared reflectance spectroscopy for nursery eucalyptus seedlings and cuttings production.

Abstract

A critical function and purpose of a nursery is for seedlings and cuttings production for farming systems. Nurseries provide farms with known pure varieties that are healthy and pathogen-free. Therefore, seed and subsequent seedling quality assessments are of great importance. Traditional methods for assessing seed (i.e. germination, and vigour) and hedge (i.e. clonal purity and nutrient content) either require long and laborious procedures or expensive chemical/molecular analysis, making them less accessible to nursery managers. Near-infrared reflectance spectroscopy (NIRS) is a rapid low-cost analysis tool easily integrated into various processes and industries, making it a viable evaluation technique. Hence, this study developed prediction models for seed germination, foliar nutrient content and hedge/clonal purity and tested whether NIRS can improve specific nursery quality control tasks, namely, seed germination and vigour testing, foliar nutrient content analysis and hedge clonal purity assessments via clonal identification. Seed germination rate and vigour are important characteristics that impact the success rate of plantation establishment. Hence, a study aimed to evaluate the feasibility of NIRS in predicting Eucalyptus seed germination and seed vigour was initiated. Species included in the study were E. nitens, E. dunnii, E. benthamii and E. macarthurii, which were accessed from 191 seed lots. Seed germination and vigour were determined through seed germination assays in 96-well culture plates for 10 days and used as the reference data for model development and validation. The results indicated that NIRS can predict seed germination with high accuracy in most of the species tested (RMSEP: 0.357 – 73.5%, R2: 58.46 – 95.42%) and collectively as a global species model (RMSEP: 11.4 – 18.8%, R2: 73.68 – 84.99%). The results for seed vigour also indicated that NIRS can predict with accuracy for the species tested (RMSEP: 5.68 – 126%, R2: 78.91 - 83.57%) and collectively as a global species model (RMSEP: 18.8%, R2: 71.37%). These NIR models showed promise for enhancing Eucalyptus seed germination and vigour testing. This study was the first to report the use of NIR as a viable tool for seed germination and vigour in nursery operations prioritising Eucalyptus plantations. Eucalyptus clones are intensively propagated in nurseries and planted to establish commercial plantations. The nutritional status of Eucalyptus cuttings impacts their ability to establish viable hedges and subsequently impacts the success of plantation establishment. Therefore, a study was conducted to determine whether NIRS can be used to measure and assess Eucalyptus foliar nutrient levels. Foliar material was sampled from the field and pot trials that were part of long-term monitoring nutrient depletion trials, sampled from 1 year-old coppice trees and pot trial samples, with a total of 382 samples with a total of 1146 spectra excluding independent test data set. The spectra were captured using a Bruker MPA NIRS. Foliar nutrient quantification was done on an Agilent MP-AES 4200 and Leco CNS and was used as the reference data to create NIR nutrient calibration models. Global models were developed for key nutrients (N, P and K) and correlated well with the reference nutrient data, having R2 > 89% for E. dunnii, G×N and G×U. This indicate that NIRS was able to model certain elements well, whilst others to a lesser extent. The significance of this study was that spectroscopic techniques, like NIRS, have promise to measure plant nutrient levels. However, this requires more investigation and development. The accurate identification of clonal purity is a critical step in forestry industries to ensure uniformity, optimize productivity, and maintain genetic integrity. Therefore, the ability of NIRS to determine hedge clonal purity was evaluated. The foliar material of 4 Eucalyptus varieties, G×U400, G×NPP, GL222 and G×U488, were sampled from commercial hedges at Top Crop Nursery (Cramond, Pietermaritzburg). The samples were dried, milled and presented to the NIRS for spectra acquisition, and the reference data used for model calibration relied on the nursery label of the hedges used. Principal component analysis (PCA) of the spectral signatures was able to cluster the samples to their respective groups using 2 and 3-dimensional PCA plots. Categorical models were created using the Ident module of the Bruker OPUS software. Upon validation of the clonal purity model, a confusion matrix indicated that all varieties except G×U400 could be correctly identified, possibly due to contamination with GL222 at this nursery, which is the problem this approach aims to resolve. The model accuracy for G×NPP, G×U400, G×U488 and GL222 were 100.00, 60.00, 92.00 and 68.00%, respectively. Considering the potential contamination of the reference material, the models performed reasonably well and warrant further development. Further development would include the use of DNA validation of calibration and validation samples. This study highlights the potential of NIRS as a transformative tool for clonal verification in large-scale forestry operations, with implications for improving efficiency and reducing operational costs. Finaly, the study demonstrated the capability of NIRS to expedite the analysis of various nursery processes. Once this research is refined and optimized, it can save on nursery analysis costs and improve the monitoring of seed lots and hedges. Therefore, the current study has contributed to advancing forestry nursery operations.