Amino acid contents and their derived volatile compositions in Cabernet Sauvignon grapes and wines after regulated deficit irrigation (RDI) were investigated during the 2015 and 2016 growing seasons in Yinchuan (NingXia, China). High-performance liquid chromatography (HPLC) and gas chromatography-mass spectrometry (GC-MS) were used for amino acid and volatile compound analyses. Three RDI strategies were tested: 60% (RDI-1), 70% (RDI-2), and 80% (RDI-3) of grapevine estimated evapotranspiration (ETc), and 100% ETc was used as the control group (CK). RDI-treated vines had lower yields and berry weights with higher total soluble solids than the control treatment. RDI-1 increased proline levels in berries and wines. RDI-2 enhanced tyrosine and asparagine levels in wines. RDI-3 enhanced arginine, alanine, valine, leucine, and isoleucine levels in berries and wines. RDI-2 and RDI-3 increased the concentrations of 2-methyl-1-butyl acetate, benzaldehyde, 3-methyl-1-pentanol, and 3-methyl-1-butanol in wines. The accumulation of volatile compounds was closely related to the amino acid concentrations—especially isoleucine, valine, and leucine—in grapes. Our results showed that RDI treatments altered amino acid concentrations and their derived volatile compositions in wines.

Winegrape cultivation has become increasingly common throughout northwest China over the past 20 years, and such viticulture can profoundly impact the properties of the utilized soil and the associated soil microbial communities. To explore these effects in the present study, samples of soil were collected from control desert soil and from vineyards planted in different years: Cabernet Sauvignon (CS, 2014), Merlot (M, 2012), and Italian Riesling (IR, 2008). The properties of this soil and the microbial communities therein were assessed through a series of physicochemical, enzymatic, and high-throughput sequencing analyses. Compared to the control soil, respective 1033.06 U/g, 398.28 U/g, and 240.56 U/g increases in alkaline phosphatase activity levels were observed in the CS, M, and IR soil samples. Stable soil bacterial richness was observed in the control, CS, and M samples, but decreased in the 11-years old IR soil, whereas no noticeable changes in soil fungi were observed across these samples. The network analyses highlighted correlations among soil microbes, and soil organic carbon and microbial biomass carbon were found to strongly influence variability in these soil microbial communities. Overall, these findings thus provide strong evidence that the prolonged monoculture of winegrapes can contribute to the deterioration of soil quality.

Background Traditional vine variety identification methods usually rely on destructive sampling of vine leaves followed by physical, physiological, biochemical and molecular measurement, which is destructive, time-consuming, labor-intensive and requires experienced grape phenotype analysts. To mitigate these problems, this study aims to develop an application (App) running on Android client to identify wine grape automatically and in real time, which can help the growers to quickly obtain the variety information. Results Experimental results show that all Convolutional Neural Network (CNN) classification algorithms can achieve an accuracy of over 94% for 21 categories on validation data, which proves the feasibility of using transfer learning to identify grape species in field environments. In particular, the classification model with the highest average accuracy is Googlenet (97.4%) with learning rate (0.001), mini-batch size (32) and maximum number of epochs (80). Testing results of the App on Android device also confirms these results. Conclusions An App running on Android client is developed to identify the wine grape in real time and field condition, which can help the growers to quickly obtain the variety information of wine grape. A total of 4200 leaf images were first collected in the field environment, which contain 21 types of typical grapes. Then both image complement preprocessing and data augmentation are adopted to enhance image features and augment the training dataset. On this basis, a number of typical CNN models (including VGG-16, DenseNet, ResNet101, ResNet18, and GoogLeNet) are compared in transfer learning training to identify the suitable one while with model parameter tuning. It is shown that GoogLeNet model outperformed other models in terms of accuracy, model complexity, and robustness with a fine-tuned accuracy of 99.91%. The effect of image complement preprocessing is also assessed by using Grad-CAM algorithm. The developed App is also shown to be feasible for real-life applications with a processing time of less than 1 second.