Search for: All records

Abstract Climate is a fundamental driver of macroecological patterns in functional trait variation. However, many of the traits that have outsized effects on thermal performance are complex, multi‐dimensional, and challenging to quantify at scale.To overcome this challenge, we leveraged techniques in deep learning and computer vision to quantify hair coverage and lightness of bees, using images of a diverse and widely distributed sample of museum specimens.We demonstrate that climate shapes variation in these traits at a global scale, with bee lightness increasing with maximum environmental temperatures (thermal melanism hypothesis) and decreasing with annual precipitation (Gloger's Rule).We found that deserts are hotspots for bees covered in light‐coloured hairs, adaptations that may mitigate heat stress and represent convergent evolution with other desert organisms.These results support major ecogeographical rules in functional trait variation and emphasize the role of climate in shaping bee phenotypic diversity. Read the freePlain Language Summaryfor this article on the Journal blog. 

Atmospheric gravity waves are produced when gravity attempts to restore disturbances through stable layers in the atmosphere. They have a visible effect on many atmospheric phenomena such as global circulation and air turbulence. Despite their importance, however, little research has been conducted on how to detect gravity waves using machine learning algorithms. We faced two major challenges in our research: our raw data had a lot of noise and the labeled dataset was extremely small. In this study, we explored various methods of preprocessing and transfer learning in order to address those challenges. We pre-trained an autoencoder on unlabeled data before training it to classify labeled data. We also created a custom CNN by combining certain pre-trained layers from the InceptionV3 Model trained on ImageNet with custom layers and a custom learning rate scheduler. Experiments show that our best model outperformed the best performing baseline model by 6.36% in terms of test accuracy.