Search

Accurate forecasting of crop yields holds paramount importance in guiding decision-making processes related to breeding efforts. Despite significant advancements in crop yield forecasting, existing methods often struggle with integrating diverse sensor data and achieving high prediction accuracy under varying environmental conditions. This study focused on the application of multi-sensor data fusion and machine learning algorithms based on unmanned aerial vehicles (UAVs) in wheat yield prediction. Five machine learning (ML) algorithms, namely random forest (RF), partial least squares (PLS), ridge regression (RR), k-nearest neighbor (KNN) and extreme gradient boosting decision tree (XGboost), were utilized for multi-sensor data fusion, together with three ensemble methods including the second-level ensemble methods (stacking and feature-weighted) and the third-level ensemble method (simple average), for wheat yield prediction. The 270 wheat hybrids were used as planting materials under full and limited irrigation treatments. A cost-effective multi-sensor UAV platform, equipped with red–green–blue (RGB), multispectral (MS), and thermal infrared (TIR) sensors, was utilized to gather remote sensing data. The results revealed that the XGboost algorithm exhibited outstanding performance in multi-sensor data fusion, with the RGB + MS + Texture + TIR combination demonstrating the highest fusion performance (R2 = 0.660, RMSE = 0.754). Compared with the single ML model, the employment of three ensemble methods significantly enhanced the accuracy of wheat yield prediction. Notably, the third-layer simple average ensemble method demonstrated superior performance (R2 = 0.733, RMSE = 0.668 t ha−1). It significantly outperformed both the second-layer ensemble methods of stacking (R2 = 0.668, RMSE = 0.673 t ha−1) and feature-weighted (R2 = 0.667, RMSE = 0.674 t ha−1), thereby exhibiting superior predictive capabilities. This finding highlighted the third-layer ensemble method’s ability to enhance predictive capabilities and refined the accuracy of wheat yield prediction through simple average ensemble learning, offering a novel perspective for crop yield prediction and breeding selection.

Grain yield (GY) prediction for wheat based on canopy spectral reflectance can improve selection efficiency in breeding programs. Time-series spectral information from different growth stages such as flowering to maturity is considered to have high accuracy in predicting GY and combining this information from multiple growth stages could effectively improve prediction accuracy. For this, 207 wheat cultivars and breeding lines were grown in full and limited irrigation treatments, and their canopy spectral reflectance was measured at the flowering, early, middle, and late grain fill stages. The potential of temporal spectral information at multiple growth stages for GY prediction was evaluated by a new method based on stacking the multiple growth stages data. Twenty VIs derived from spectral reflectance were used as the input feature of a support vector regression (SVR) to predict GY at each growth stage. The predicted GY values at multiple growth stages were trained by multiple linear regression (MLR) to establish a second-level prediction model. Results suggested that the prediction accuracy (R2) of VIs data from single growth stages ranged from 0.60 to 0.66 and 0.35 to 0.42 in the full and limited irrigation treatments, respectively. The prediction accuracy was increased by an average of 0.06, 0.07, and 0.07 after stacking the VIs of two, three, and four growth stages, respectively, under full irrigation. Similarly, under limited irrigation, the prediction accuracy was increased by 0.03, 0.04, and 0.04 by stacking the VIs of two, three, and four growth stages, respectively. Stacking of VIs of multiple important growth stages can increase the accuracy of GY prediction and application of a stable stacking model could increase the usefulness of data obtained from different phenotyping platforms.

Grain yield is increasingly affected by climate factors such as drought and heat. To develop resilient and high-yielding cultivars, high-throughput phenotyping (HTP) techniques are essential for precise decisions in wheat breeding. The ability of unmanned aerial vehicle (UAV)-based multispectral imaging and ensemble learning methods to increase the accuracy of grain yield prediction in practical breeding work is evaluated in this study. For this, 211 winter wheat genotypes were planted under full and limited irrigation treatments, and multispectral data were collected at heading, flowering, early grain filling (EGF), and mid-grain filling (MGF) stages. Twenty multispectral vegetation indices (VIs) were estimated, and VIs with heritability greater than 0.5 were selected to evaluate the models across the growth stages under both irrigation treatments. A framework for ensemble learning was developed by combining multiple base models such as random forest (RF), support vector machine (SVM), Gaussian process (GP), and ridge regression (RR). The R2 values between VIs and grain yield for individual base models were ranged from 0.468 to 0.580 and 0.537 to 0.598 for grain yield prediction in full and limited irrigation treatments across growth stages, respectively. The prediction results of ensemble models were ranged from 0.491 to 0.616 and 0.560 to 0.616 under full and limited irrigation treatments respectively, and were higher than that of the corresponding base learners. Moreover, the grain yield prediction results were observed high at mid grain filling stage under both full (R2 = 0.625) and limited (R2 = 0.628) irrigation treatments through ensemble learning based stacking of four base learners. Further improvements in ensemble learning models can accelerate the use of UAV-based multispectral data for accurate predictions of complex traits like grain yield in wheat.

The simultaneous improvement of protein content (PC) and grain yield (GY) in bread wheat (Triticum aestivum L.) under low-input management enables the development of resource-use efficient varieties that combine high grain yield potential with desirable end-use quality. However, the complex mechanisms of genotype, management, and growing season, and the negative correlation between PC and GY complicate the simultaneous improvement of PC and GY under low-input management. To identify favorable genotypes for PC and GY under low-input management, this study used 209 wheat varieties, including strong gluten, medium-strong gluten, medium gluten, weak gluten, winter, semi-winter, weak-spring, and spring types, which has been promoted from the 1980s to the 2010s. Allelic genotyping, performed using kompetitive allele-specific polymerase chain reaction (KASP) technology, found 69 types of GY-PC allelic combinations in the tested materials. Field trials were conducted with two growing season treatments (2018–2019 and 2019–2020) and two management treatments (conventional management and low-input management). Multi-environment analysis of variance showed that genotype, management, and growing season had extremely substantial effects on wheat GY and PC, respectively, and the interaction of management × growing season also had extremely significant effects on wheat GY. According to the three-sigma rule of the normal distribution, the GY of wheat varieties Liangxing 66 and Xinmai 18 were stable among the top 15.87% of all tested materials with high GY, and their PC reached mean levels under low-input management, but also stably expressed high GY and high PC under conventional management, which represents a great development potential. These varieties can be used as cultivars of interest for breeding because TaSus1-7A, TaSus1-7B, TaGW2-6A, and TaGW2-6B, which are related to GY, and Glu-B3, which is related to PC, carry favorable alleles, among which Hap-1/2, the allele of TaSus1-7A, and Glu-B3b/d/g/i, the allele of Glu-B3, can be stably expressed. Our results may be used to facilitate the development of high-yielding and high-quality wheat varieties under low-input management, which is critical for sustainable food and nutrition security.