Search for: All records

Hagfishes defend themselves from gill-breathing predators by producing large volumes of fibrous slime when attacked. The slime's effectiveness comes from its ability to clog predators' gills, but the mechanisms by which hagfish slime clogs are uncertain, especially given its remarkably dilute concentration of solids. We quantified the clogging performance of hagfish slime over a range of concentrations, measured the contributions of its mucous and thread components, and measured the effect of turbulent mixing on clogging. To assess the porous structure of hagfish slime, we used a custom device to measure its Darcy permeability. We show that hagfish slime clogs at extremely dilute concentrations like those found in native hagfish slime and displays clogging performance that is superior to three thickening agents. We report an extremely low Darcy permeability for hagfish slime, and an effective pore size of 10–300 nm. We also show that the mucous and thread components play distinct yet crucial roles, with mucus being responsible for effective clogging and low permeability and the threads imparting mechanical strength and retaining clogging function over time. Our results provide new insights into the mechanisms by which hagfish slime clogs gills and may inspire the development of ultra-soft materials with novel properties. 

Abstract Classifying images using supervised machine learning (ML) relies on labeled training data—classes or text descriptions, for example, associated with each image. Data‐driven models are only as good as the data used for training, and this points to the importance of high‐quality labeled data for developing a ML model that has predictive skill. Labeling data is typically a time‐consuming, manual process. Here, we investigate the process of labeling data, with a specific focus on coastal aerial imagery captured in the wake of hurricanes that affected the Atlantic and Gulf Coasts of the United States. The imagery data set is a rich observational record of storm impacts and coastal change, but the imagery requires labeling to render that information accessible. We created an online interface that served labelers a stream of images and a fixed set of questions. A total of 1,600 images were labeled by at least two or as many as seven coastal scientists. We used the resulting data set to investigate interrater agreement: the extent to which labelers labeled each image similarly. Interrater agreement scores, assessed with percent agreement and Krippendorff's alpha, are higher when the questions posed to labelers are relatively simple, when the labelers are provided with a user manual, and when images are smaller. Experiments in interrater agreement point toward the benefit of multiple labelers for understanding the uncertainty in labeling data for machine learning research.