An official website of the United States government
The appropriate selection of rootstock-scion combinations to improve yield and fully realize grafting benefits requires an in-depth understanding of rootstock-scion synergy. Toward this end, we grafted two determinate-type scions [grape tomato (‘BHN 1022') and beefsteak tomato (‘Skyway')] onto four rootstocks with different characteristics to examine plant growth, yield performance, biomass production, and fruit mineral nutrient composition. The study was conducted during two growing seasons (spring and fall plantings in Florida) under organic production in high tunnels with the non-grafted scions as controls. Rootstocks had previously been designated as either “generative” (‘Estamino') or “vegetative” (‘DR0141TX') by some commercial suppliers or had not been characterized [‘RST-04-106-T' and ‘SHIELD RZ F1 (61-802)']. Also, ‘Estamino', ‘DR0141TX', and ‘RST-04-106-T' had been described as more vigorous than ‘SHIELD RZ F1 (61-802)'. In both planting seasons (with low levels of soilborne disease pressure), the “vegetative” and “generative” rootstocks increased marketable and total fruit yields for both scions except for the beefsteak tomato grafted with the “vegetative” rootstock in fall planting. Positive effects of ‘RST-04-106-T' on fruit yield varied with scions and planting seasons, and were most manifested when grafted with the beefsteak tomato scion in fall planting. ‘SHIELD RZ F1 (61-802)' led to similar yields as the non-grafted controls except for grafting with the grape tomato scion in fall planting. For vegetative and fruit biomass, both the “vegetative” and “generative” rootstocks had positive impacts except for the beefsteak tomato in fall planting. For fruit mineral composition, the “vegetative” and “generative” rootstocks, both highly vigorous, consistently elevated fruit P, K, Ca, Zn, and Fe contents on a dry weight basis, whereas the other rootstocks did not. Overall, although the more vigorous rootstocks enhanced tomato plant productivity and fruit minerals, the evidence presented here does not support the suggestion that the so-called “vegetative” and “generative” rootstocks have different impacts on tomato scion yield, biomass production, or fruit mineral contents. More studies with different production systems and environmental conditions as well as contrasting scion genotypes are needed to further categorize the impacts of rootstocks with different vigor and other characteristics on plant biomass production and their implications on fruit yield development.
more »
« less
Predicting Grapevine Physiological Parameters Using Hyperspectral Remote Sensing Integrated with Hybrid Convolutional Neural Network and Ensemble Stacked Regression
Grapevine rootstocks are gaining importance in viticulture as a strategy to combat abiotic challenges, as well as enhance scion physiology. Direct leaf-level physiological parameters like net assimilation rate, stomatal conductance to water vapor, quantum yield of PSII, and transpiration can illuminate the rootstock effect on scion physiology. However, these measures are time-consuming and limited to leaf-level analysis. This study used different rootstocks to investigate the potential application of aerial hyperspectral imagery in the estimation of canopy level measurements. A statistical framework was developed as an ensemble stacked regression (REGST) that aggregated five different individual machine learning algorithms: Least absolute shrinkage and selection operator (Lasso), Partial least squares regression (PLSR), Ridge regression (RR), Elastic net (ENET), and Principal component regression (PCR) to optimize high-throughput assessment of vine physiology. In addition, a Convolutional Neural Network (CNN) algorithm was integrated into an existing REGST, forming a hybrid CNN-REGST model with the aim of capturing patterns from the hyperspectral signal. Based on the findings, the performance of individual base models exhibited variable prediction accuracies. In most cases, Ridge Regression (RR) demonstrated the lowest test Root Mean Squared Error (RMSE). The ensemble stacked regression model (REGST) outperformed the individual machine learning algorithms with an increase in R2 by (0.03 to 0.1). The performances of CNN-REGST and REGST were similar in estimating the four different traits. Overall, these models were able to explain approximately 55–67% of the variation in the actual ground-truth data. This study suggests that hyperspectral features integrated with powerful AI approaches show great potential in tracing functional traits in grapevines.
more »
« less
- PAR ID:
- 10542268
- Publisher / Repository:
- MDPI
- Date Published:
- Journal Name:
- Remote Sensing
- Volume:
- 16
- Issue:
- 14
- ISSN:
- 2072-4292
- Page Range / eLocation ID:
- 2626
- Subject(s) / Keyword(s):
- aerial hyperspectral remote sensing deep learning ensemble learning stacking regressor UAV high throughput phenotyping grapevine rootstock carbon assimilation stomatal conductance transpiration
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Understanding the genetic basis of leaf size and shape is essential for evaluating and selecting for plant adaptability and performance in variable and shifting climatic conditions. This study maps the leaf size and shape phenotypic variation as influenced by the genetic architecture of a rootstock population and its conferred influence on these traits in a common scion. The influence of the root system genotype was studied using two different presentations of an F1 rootstock population (F1_Vruprip;V. rupestrisScheele ‘B38’ (USDA PI#588160) XV. ripariaMichx. ‘HP1’ (USDA PI#588271)); 1) the F1_Vruprip grapevine progeny on their own roots and 2) a F1_Vruprip cohort that was grafted with the common scion scion 'Marquette'. Three leaf positions (apical, middle, and basal) were sampled in both presentations at two timepoints in two consecutive growing seasons. A twenty-one-point leaf morphological landmark coordinate analysis was conducted, and ten leaf size and six derived shape phenotypes were used for QTL mapping. Genetic analysis identified five distinct hotspots associated with size-related leaf area attributes in own-rooted and grafted vines. The identification of multiple leaf-growth-associated pathways in these hotspot regions strengthened the correlation between genetics and phenotypic traits. Shape related QTL accounted for 12-48% of the shape phenotypic variation but did not cluster as QTL hotspots. Three QTL hotspots captured the genetic influence of the rootstock conferred onto the scion leaf area traits. The results showed that the leaf position and the rootstock population’s genetic composition significantly impacted leaf morphological attributes and that there was a measurable rootstock genotype influence conferred on the grafted scion leaves. This reveals the genetic loci and gene pathways underlying leaf morphological phenotypes in own-rooted progeny and also verifies the potential of rootstock genetics to confer modulation of scion canopy features, providing greater potential to select for climate-resilient grapevines.more » « less
-
Proximal remote sensing offers a powerful tool for high-throughput phenotyping of plants for assessing stress response. Bean plants, an important legume for human consumption, are often grown in regions with limited rainfall and irrigation and are therefore bred to further enhance drought tolerance. We assessed physiological (stomatal conductance and predawn and midday leaf water potential) and ground- and tower-based hyperspectral remote sensing (400 to 2,400 nm and 400 to 900 nm, respectively) measurements to evaluate drought response in 12 common bean and 4 tepary bean genotypes across 3 field campaigns (1 predrought and 2 post-drought). Hyperspectral data in partial least squares regression models predicted these physiological traits ( R 2 = 0.20 to 0.55; root mean square percent error 16% to 31%). Furthermore, ground-based partial least squares regression models successfully ranked genotypic drought responses similar to the physiologically based ranks. This study demonstrates applications of high-resolution hyperspectral remote sensing for predicting plant traits and phenotyping drought response across genotypes for vegetation monitoring and breeding population screening.more » « less
-
Abstract Estimates of plant traits derived from hyperspectral reflectance data have the potential to efficiently substitute for traits, which are time or labor intensive to manually score. Typical workflows for estimating plant traits from hyperspectral reflectance data employ supervised classification models that can require substantial ground truth datasets for training. We explore the potential of an unsupervised approach, autoencoders, to extract meaningful traits from plant hyperspectral reflectance data using measurements of the reflectance of 2151 individual wavelengths of light from the leaves of maize (Zea mays) plants harvested from 1658 field plots in a replicated field trial. A subset of autoencoder‐derived variables exhibited significant repeatability, indicating that a substantial proportion of the total variance in these variables was explained by difference between maize genotypes, while other autoencoder variables appear to capture variation resulting from changes in leaf reflectance between different batches of data collection. Several of the repeatable latent variables were significantly correlated with other traits scored from the same maize field experiment, including one autoencoder‐derived latent variable (LV8) that predicted plant chlorophyll content modestly better than a supervised model trained on the same data. In at least one case, genome‐wide association study hits for variation in autoencoder‐derived variables were proximal to genes with known or plausible links to leaf phenotypes expected to alter hyperspectral reflectance. In aggregate, these results suggest that an unsupervised, autoencoder‐based approach can identify meaningful and genetically controlled variation in high‐dimensional, high‐throughput phenotyping data and link identified variables back to known plant traits of interest.more » « less
-
Ensemble methods that average over a collection of independent predictors that are each limited to a subsampling of both the examples and features of the training data command a significant presence in machine learning, such as the ever-popular random forest, yet the nature of the subsampling effect, particularly of the features, is not well understood. We study the case of an ensemble of linear predictors, where each individual predictor is fi t using ordinary least squares on a random submatrix of the data matrix. We show that, under standard Gaussianity assumptions, when the number of features selected for each predictor is optimally tuned, the asymptotic risk of a large ensemble is equal to the asymptotic ridge regression risk, which is known to be optimal among linear predictors in this setting. In addition to eliciting this implicit regularization that results from subsampling, we also connect this ensemble to the dropout technique used in training deep (neural) networks, another strategy that has been shown to have a ridge-like regularizing effect.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.3390/rs16142626
