Search for: All records

Abstract Drought events may increase the likelihood that the plant water transport system becomes interrupted by embolism. Yet our knowledge about the temporal frequency of xylem embolism in the field is frequently lacking, as it requires detailed, long‐term measurements.We measured xylem embolism resistance and midday xylem water potentials during the consecutive summers of 2019 and 2020 to estimate maximum levels of embolism in leaf and stem xylem of ten temperate angiosperm tree species. We also studied vessel and pit membrane characteristics based on light and electron microscopy to corroborate potential differences in embolism resistance between leaves and stems.Apart fromA.pseudoplatanusandQ.petraea, eight species experienced minimum xylem water potentials that were close to or below those required to initiate embolism. Water potentials corresponding to ca. 12% loss of hydraulic conductivity (PLC) could occur in six species, while considerable levels of embolism around 50% PLC were limited toB.pendulaandC.avellana. There was a general agreement in embolism resistance between stems and leaves, with leaves being equally or more resistant than stems. Also, xylem embolism resistance was significantly correlated to intervessel pit membrane thickness (T_PM) for stems, but not to vessel diameter and total intervessel pit membrane surface area of a vessel.Our data indicate that low amounts of embolism occur in most species during moderate summer drought, and that considerable levels of embolism are uncommon. Moreover, our experimental andT_PMdata show that leaf xylem is generally no more vulnerable than stem xylem. 

A key assumption in multi-task learning is that at the inference time the multi-task model only has access to a given data point but not to the data point’s labels from other tasks. This presents an opportunity to extend multi-task learning to utilize data point’s labels from other auxiliary tasks, and this way improves performance on the new task. Here we introduce a novel relational multi-task learning setting where we leverage data point labels from auxiliary tasks to make more accurate predictions on the new task. We develop MetaLink, where our key innovation is to build a knowledge graph that connects data points and tasks and thus allows us to leverage labels from auxiliary tasks. The knowledge graph consists of two types of nodes: (1) data nodes, where node features are data embeddings computed by the neural network, and (2) task nodes, with the last layer’s weights for each task as node features. The edges in this knowledge graph capture data-task relationships, and the edge label captures the label of a data point on a particular task. Under MetaLink, we reformulate the new task as a link label prediction problem between a data node and a task node. The MetaLink framework provides flexibility to model knowledge transfer from auxiliary task labels to the task of interest. We evaluate MetaLink on 6 benchmark datasets in both biochemical and vision domains. Experiments demonstrate that MetaLink can successfully utilize the relations among different tasks, outperforming the state-of-the-art methods under the proposed relational multi-task learning setting, with up to 27% improvement in ROC AUC. 

Langmuir monolayers at gas/liquid interfaces provide a rich framework to investigate the interplay between multiscale geometry and mechanics. Monolayer collapse is investigated at a topological and geometric level by building a scale spaceM from experimental imaging data. We present a general lipid monolayer collapse phase diagram, which shows that wrinkling, folding, crumpling, shear banding, and vesiculation are a continuous set of mechanical states that can be approached by either tuning monolayer composition or temperature. The origin of the different mechanical states can be understood by investigating the monolayer geometry at two scales: fluorescent vs atomic force microscopy imaging. We show that an interesting switch in continuity occurs in passing between the two scales, CAFM MAFM 6¼ CFM M. Studying the difference between monolayers that fold vs shear band, we show that shear banding is correlated to the persistence of a multi-length scale microstructure within the monolayer at all surface pressures. A detailed analytical geometric formalism to describe this microstructure is developed using the theory of structured deformations. Lastly, we provide the first ever finite element simulation of lipid monolayer collapse utilizing a direct mapping from the experimental image spaceM into a simulation domain P. We show that elastic dissipation in the form of bielasticity is a necessary and sufficient condition to capture loss of in-plane stability and shear banding.