Brushing aside doubts: an evaluation of the beat-sheet brushing technique for detecting the Nearctic tree trunk sheetweaver (Araneae: Linyphiidae).

It is important to have reliable information on the presence/absence, population structure, and density of animals across their natural range. Detecting small organisms, however, such as the Nearctic tree trunk sheetweaver spider Drapetisca alteranda Chamberlin 1909 (Araneae: Linyphiidae), presents challenges due to its diminutive size and cryptic nature. We used a capture/recapture study to determine the detection and recapture probabilities of this spider using a standard beat sheet technique adopted for surveying tree trunks. Spiders were released on 3 different tree species that provided a range of microhabitats, including variable bark surface area and furrow depth/width. Microhabitat features played a small role in the timing of spider recapture (i.e., slower rate of recapture as furrowing increased). However, our results demonstrated 100% detection across replicate experiments and individual recapture probabilities exceeding 90% in most situations, with no significant differences in recapture observed among tree species and with respect to tree circumference. Furthermore, we show that most spiders could be recaptured within 2 sampling revolutions around the tree trunk, and there was no difference in the probability of collecting male and female spiders (although they differ markedly in size). Finally, we found no difference among brushers, supporting the idea that this method is replicable across collectors and studies. Collectively, we establish confidence in the ecological knowledge obtained with this technique and encourage its application with similar species and systems.

Machine-Learning-Assisted Determination of the Global Zero-Temperature Phase Diagram of Materials.

Crystal-graph attention neural networks have emerged recently as remarkable tools for the prediction of thermodynamic stability. The efficacy of their learning capabilities and their reliability is however subject to the quantity and quality of the data they are fed. Previous networks exhibit strong biases due to the inhomogeneity of the training data. Here a high-quality dataset is engineered to provide a better balance across chemical and crystal-symmetry space. Crystal-graph neural networks trained with this dataset show unprecedented generalization accuracy. Such networks are applied to perform machine-learning-assisted high-throughput searches of stable materials, spanning 1 billion candidates. In this way, the number of vertices of the global T = 0 K phase diagram is increased by 30% and find more than ≈150 000 compounds with a distance to the convex hull of stability of less than 50 meV atom-1 . The discovered materials are then accessed for applications, identifying compounds with extreme values of a few properties, such as superconductivity, superhardness, and giant gap-deformation potentials.