Kharchia does show a significantly reduced level of Na+ uptake in the 5th leaf when compared to salt-sensitive cultivars, which likely explains the stable harvest index of this line . Based on these contradictory findings, it appears that the most informative way to conventionally assess salt tolerance is to measure multiple growth and physiological parameters simultaneously. However, this is laborious, expensive, and not feasible for many lines or populations simultaneously as is required for breeding programs. In contrast, our method could be used to quantitatively and objectively rank salt tolerance of individual wheat lines in a non-destructive and cost-effective manner. Moreover, the proposed method could be successfully used to detect subtle differences between lines, such as between alloplasmics and euplasmics. A mixed pixel contains spectral information derived from a combination of objects that inherently have varying spectral characteristics. Mixed pixels are often located along the edges of objects in images, especially when the spatial resolution of an image is finer than the individual objects contained in the image . In this study, mixed pixels along the edges of leaves were eliminated from the binary vegetation mask using morphological image processing. Without masking, these mixed pixels would have led to reduced accuracy through the end member selection and subsequently in computing the similarity of pixels to the salt end member on which further analyses were based. Prior to using morphological operations,plant pot with drainage filtering based on vegetation indices allowed us to appropriately mask pixels that were covered by shadows from other leaves or objects.
The interaction between incident light and the leaf surface is complex due to leaf angle and curvature. The intensity of the reflected light from leaves is both a function of physiology and leaf geometry, and tends to cause a significant change in the spectral response of vegetation. Behmann et al. reported 60% of the spectral response of sugar beets could be influenced by the geometry of plant leaves. This provides explanation for observing several control and salt pixels having similar spectral properties . Although we did not explicitly address the issue of leaf geometry, we could mitigate its negative effects on the results by considering the similarity distribution of pixels to the salt end member. By deploying a fuzzy concept rather than a hard classification , downstream processing techniques were possible, such as posterior probability and histogram distance. In this research, the curse of dimensionality in analysis of hyperspectral images was addressed by projecting from a high dimension feature space onto one dimension. The one-dimensional feature space, which can be represented as a gray-scale image, denoted the similarity of each pixel to the salt end member while maintaining the required information for making further inferences. As a result, interpretation of hyperspectral images could be performed in a much more efficient manner with more meaningful information. By reducing dimensionality, results are simpler and are easier to comprehend. Because agricultural research groups are oftentimes comprised of many scientific disciplines, this directness is imperative. In general, the proposed strategy helps to interpret complex hyperspectral images by discovering the underlying hidden features caused by salt stress.
This strategy will also provide more flexibility in selecting ML methods to analyze images. For instance, MDPA was utilized to measure the distance between the similarity distributions of control and salt pixels to the salt end member. Another benefit of conversion to a onedimensional feature space is that visualization of relevant spectral information is possible in a more meaningful manner while involving all bands in similarity measurement and maintaining spatial integrity. The other important advantage is that the required computational time and data storage space for analyzing and storing hyperspectral data was significantly reduced due to converting from a high-dimensional to one-dimensional space. For instance, the average size of a hyperspectral image was ∼400 MB while the size of obtained gray scale image was less than 1 MB. It is also noteworthy that the VSM method used for computing the coefficient matrix proposed herein significantly reduced the processing time compared to the quadratic optimization problem that is commonly used. Further increases in wheat grain yield are required to feed a growing population, but losses generated by water stress are increasing with global warming, and are eroding progress in other areas of wheat improvement. Root depth and biomass distribution in the soil profile are critical traits for adaptation to water stress and have been prioritized for improving drought resilience in wheat. However, the gene networks that regulate these traits in wheat remain largely unknown. This has prompted new efforts to understand and modify wheat root architecture to optimize water acquisition in both common and durum wheat .A tested source for improving these traits in wheat is the introgression of the short arm of rye chromosome one into common wheat , which induces higher root biomass and confers a yield advantage under drought stress. Unfortunately, this translocation reduces bread baking quality.
To eliminate the negative effect on quality, a recombinant 1RS chromosome with two wheat 1BS interstitial introgressions was developed twenty years ago in the wheat cultivar Pavon. We introgressed the engineered 1WW chromosome into the 1RS.1BL cv. Hahn by six backcrosses, and then generated three lines with either the complete 1RS , the proximal 1BS introgression removing the rye secalins , or the distal 1BS introgression restoring the Gli-B1 / GluB3 locus . Using the wheat Illumina 90 K SNP iSelect array, we demonstrated that the 1RS and 1RW lines were 99.3% identical. In large physiological and agronomic field experiments replicated over four years , in two locations, and under both full irrigation and terminal drought, lines carrying the complete 1RS arm showed significantly higher grain yield relative to 1RW. The 1RS lines showed higher carbon isotope discrimination and increased stomatal conductance, indicating improved access to residual water in the soil. Increased water content was also evident in replicated split-plot field experiments where 1RS plants showed improved water indexes and absence of rolled leaves and dry leaf tips that were evident some years in the adjacent 1RW plots. In a subsequent field study, we excavated three ~2 m deep trenches cutting perpendicular across the middle of the plots in three blocks including the different genotypes and took horizontal soil core samples from the center of each plot at 20 cm intervals. These data confirmed that the 1RS lines have a higher root density, with roots that reach deeper in the soil and can access more water than the 1RW lines. The differences in wheat architecture were also evident under hydroponic conditions,pot with drainage holes where the elongation of the 1RW seminal roots decreased from nine days after germination and almost stopped by ~16 DAG while the 1RS roots continue to grow. The 1RW roots also showed lateral roots forming close to the seminal root apical meristem , a phenotype that was not observed in the near isogenic sister lines with the complete 1RS arm. In summary, these results indicated that the small wheat introgression in 1RW was associated with drastic changes in root architecture both in short term hydroponic assays and in field experiments across the growing season. More recent exome sequencing of the different lines showed that the 1RW chromosome has a complex structure generated by structural differences between the 1RS and 1BS distal chromosome regions. The 1RW chromosome has a duplicated 1RS region flanking an insertion of a colinear region from wheat chromosome arm 1BS , which resulted in increased dosage of the genes included in the triplicated region. Figure 1A shows the effect of these structural changes on a gene family located within the duplicated region. The previous study also showed that different 1RS-1BS recombinant lines with longer distal regions from 1BS or 1RS have long seminal roots, suggesting that changes in gene dosage are responsible for the shorter seminal roots in 1RW rather than specific 1BS or 1RS genes.
This hypothesis was also supported by the intermediate seminal-root length of heterozygous 1RW plants, and by the restoration of normal root growth in a 1WW line carrying a small deletion that eliminated part of the distal 1BS introgression and the proximal duplicated 1RS segment. This radiation mutant further delimited a candidate gene region including 38 wheat and rye duplicated orthologs, but the causal gene was not identified. In this study, we present the functional validation of the genes responsible for the differences in root development using CRISPRCas9 induced mutants and transgenic lines. We also demonstrate that these genes encode cytoplasmic functional enzymes involved in the biosynthesis of JA-Ile, and that a JA bio-synthetic inhibitor can abolish the differences between 1RS and 1RW in root architecture. Finally, we present high-throughput transcriptomic data for the isogenic and transgenic lines that reveal the pathways affected by the differences in dosage or expression levels of these genes.To prioritize the candidate genes for functional validation, we performed additional characterizations of the effects of the 1BS introgression and 1RS duplication on nodal roots growth and aerial tissues dry weight. Nodal roots started to appear at 18 DAG and initially grew similarly in 1RS and 1RW. However, by 27 DAG the 1RS longest nodal roots were 2.7 cm longer than the equivalent root in 1RW, a difference that increased to 10.8 cm at 31 DAG and to 18.1 cm at 34 DAG . These results indicate that the 1RW nodal roots experience a similar developmentally-regulated growth arrest as the seminal roots. Taken together, these results are consistent with the higher root density and increased water access of the 1RS plants relative to 1RW in the field experiments. In replicated hydroponic experiments showing highly significant differences in seminal root length between 1RS and 1RW, we analyzed the aerial part of the plants. A combined ANOVA for two experiments showed no significant differences in leaf number, length of the youngest fully expanded leaf or aerial dry weight at 26 DAG . We observed a slight decrease in both aerial and root dry weight in 1RW relative to 1RS , which was marginally significant in one of the experiments but not in the combined analyses . Based on these results, we concluded that the differences between the 1RS and 1RW genotypes mainly involve changes in root elongation and therefore, we prioritized candidate genes expressed predominantly in roots for functional characterization. To test if 1RS and 1RW exhibit differential physiological responses under osmotic stress, we compared them in hydroponic experiments with and without mannitol . We found no significant differences between 1RS and 1RW for leaf length and relative water content , but detected significant reductions in root length and root and shoot dry weight in 1RW relative to 1RS. All traits showed highly significant differences for the mannitol treatment, but no significant genotype x mannitol interactions , indicating that both genotypes respond similarly to the osmotic stress induced by mannitol . Finally, we tested the effect of the 1RS duplication on seminal root architecture in the absence of the 1BS segment introgression. To do this, we developed a diisosomic 1RS addition line with two fused 1RS arms into Hahn-1RS . The seminal roots from plants carrying extra 1RS arms were significantly shorter than those from the wildtype , confirming that duplicated 1RS gene dosage is sufficient to induce shorter seminal roots.We measured levels of the bioactive JA-Ile and its precursors JA and OPDA in the terminal 1 cm of the seminal roots at 6 DAG. We observed a 2.5 to 3.3-fold increase in JA and JA-Ile concentrations in 1RW relative to 1RS and a 4.3 to 4.8-fold increase in UBI::OPRIII-R5 relative to 1RS . These results demonstrate that increases in OPRIII gene dosage or expression are associated with increases in JA and JA-Ile in early seminal root development.To explore the effect of the increases in JA-Ile on wheat genes known to be involved in the JA-signaling pathway, we compared the root transcriptome data from 1RW and UBI::OPRIII-R5 with 1RS . For each gene, we analyzed simultaneously the three homoeologs in a factorial ANOVA including genotypes and homoeologs as factors and RNA-seq samples as replications. No significant interactions between homeologs and genotypes were detected, indicating similar responses among homoeologs. All four genes showed significant differences in expression between 1RS and at least one of the two genotypes with increased JA-Ile .