Determining the environmental limits that govern how engineered metabolic traits control whole-plant carbon (C) allocation is essential for developing climate-resilient bioenergy systems. Oil-enhanced sorghum offers a promising strategy to boost aboveground energy density, yet its system-wide effects on root C investment and soil C delivery, particularly under drought, are largely uncharacterized. Here, we provide the first comprehensive assessment of these interactions using whole-plant 13CO2 continuous labeling, depth-specific sampling of roots and rhizosphere soil, and coordinated measurements of photosynthesis, stomatal anatomy, tissue nitrogen, and root non-structural carbohydrates (NSCs). Under well-watered conditions, oil enhancement produced distinct aboveground physiological changes, including slightly longer stomata, lower stomatal density, significantly higher leaf nitrogen concentration and biomass, and greater leaf 13C enrichment. Importantly, these aboveground changes did not translate into detectable shifts in the allocation of recent photosynthate to belowground pools. In contrast, drought acted as a dominant regulatory factor, reorganizing C flow across both genotypes by suppressing photosynthesis and leaf water status, increasing root nitrogen and NSC reserves, and significantly promoting the retention of recent assimilates in shallow root systems. Depth-specific 13C patterns showed that drought reduced new C incorporation into deep roots while increasing 13C enrichment in deep rhizosphere soil, suggesting that drought altered the balance between C investment in root growth and rhizodeposit C inputs at depth. Ultimately, soil moisture availability was an overriding determinant of belowground C partitioning and vertical C delivery, superseding the influence of the engineered lipid sink. These findings provide new mechanistic insight into the environmental constraints on C flow in an engineered bioenergy crop and identify moisture-driven limitations as the primary bottleneck for translating synthetic metabolic innovations into robust, ecosystem-level C outcomes.
Target-Side Paraphrase Augmentation for Sign Language Translation with Large Language Models
arXiv:2605.31393v1 Announce Type: cross Abstract: Sign language translation (SLT) remains constrained by limited paired sign-video/text corpora and heavy-tailed target vocabularies. We study target-side augmentation in



