RW Moriarty Science Seminars

R.W. Moriarty Science Seminar Presents: The Datafied Animal: Big Data, Machine Learning, and Wildlife Conservation

Speaker: Emily Wanderer, University of Pittsburgh

Abstract: Over the past twenty years, scientists have developed an ever expanding “internet of animals.” This internet of animals is composed of GPS-tracking and biologging technologies and machine learning and AI tools; they are being deployed globally to produce what has been called a regime of benign surveillance of non-human life. As miniaturized GPS tags, camera traps, and bioacoustics monitoring now capture exponentially more data about wildlife, scientists and engineers are developing AI and machine learning tools to process that data. These include things like facial recognition systems for bears, an analogue to FaceNet, Google’s human facial recognition system, and machine learning translation of sperm whale sounds, part of a project to develop models for the study of non-human communications. There have been many studies of the social construction of machine learning and the effects of datafication on human lives. However, these studies leave unexamined what happens when these technologies are developed for or transferred to the study of animals. In this talk, Wanderer brings an anthropological perspective to tech for conservation, analyzing the ideas, cultural categories, and histories that shape machine learning and AI about non-human animals.
.

Register Now

R.W. Moriarty Science Seminar Presents: Working the Face: Integrative Evolutionary Morphology & Physiology in the Mammalian Face

Speaker: Anne Burrows, Duquesne University

Abstract:Faces are ubiquitous in the mammalian and general vertebrate clades. However, our understanding of the evolutionary morphology and physiology of the face remains unclear. We have many paleontological pieces of evidence for how the bony and dental aspects of mammalian faces changed over time, especially in regard to dietary functions, brain development, and even the hearing and vestibular mechanisms. Similar evidence for the muscles of facial expression, the “mimetic muscles”, and associated connective tissue are lacking. This talk integrates evidence from the fossil record, comparative anatomy, physiology, biomechanics, plastic surgery, genetics, social networks, and artificial intelligence to understand the evolution of the mammalian face.
.

Register Now

R.W. Moriarty Science Seminar Presents: Modern mineralogy at the Smithsonian NMNH

Speaker: Gabriela Farfan, Smithsonian National Museum of Natural History

Abstract:The Smithsonian National Museum of Natural History is renowned for its National Gem and Mineral Collection that encompasses the American “crown jewels,” such as the Hope Diamond, world class mineral specimens, and hundreds of thousands of reference specimens for use in research. This presentation will be a tour of some of the most recent highlights that have been added to the National collection, as well as a behind-the-scenes tour of modern mineralogical research being done by Dr. Gabriela Farfan and her colleagues in the Department of Mineral Sciences. The first research example is a gem science project on the origin of trapiche-like inclusion patterns in quartz from Inner Mongolia, China. The second is an example of environmental biomineralogy and how we can use crystallography and mineralogy to learn more about how coral skeletons and mollusk shells grow under changing aquatic environments.
.

Register Now

R.W. Moriarty Science Seminar Presents: Dead shells speak: detecting unsuspected ecosystem change without being there

Speaker: Susan Kidwell, University of Chicago

Abstract:One of the major challenges for environmental management and conservation biology is simply discovering ‘what was natural’ before human impacts. This problem is especially pressing in marine systems, where biological monitoring and other records are brief or lacking. This question can be tackled locally in a geologically common-sensical way. First, we compare local dead-shell assemblages — sieved from the surface mixed layer — to a census of the local living community, identifying which species occur only as dead individuals (suggesting their populations have waned) and those that occur only alive, despite having mineralized tissues that should leave a trace, suggesting they are new arrivals to the habitat or region. We can then “age un-mix” the dead-shell assemblage using radiocarbon or other methods, which permits us to establish when taxa declined (or increased) relative to local timelines of natural and cultural stressors. This simple approach, strongly supported by meta-analysis and mechanistic modeling, provides a powerful tool for recognizing ecological change retrospectively. Importantly, it has the power to exonerate as well as to incriminate human activities. The method will be illustrated with some case studies from around the world.
.

Register Now

R.W. Moriarty Science Seminar Presents: Eavesdropping on wildlife: Realizing the potential of autonomous acoustic surveys for ecology and conservation biology

Speaker: Tessa Rhinehart, University of Pittsburgh

Abstract:: In an era of rapid global change, it is critically important to understand how anthropogenic actions like habitat disturbance and restoration affect wildlife populations. Ecologists and conservation biologists are increasingly applying autonomous survey methods to study these questions at large scales. Autonomous surveys involve using passive sensors such as camera traps and autonomous acoustic recorders to gather data on the species present in a location. To interpret the large data sets that such sensors produce, many practitioners are turning to machine learning algorithms that can identify the species captured in these data.
However, these machine learning-driven approaches come with several drawbacks. For example, algorithms regularly misidentify species and their accuracy can differ between datasets. Interpreting this error-prone output poses challenges that this work aims to address. This talk will describe three case studies applying autonomous acoustic recorders to study bird populations. It will identify common obstacles in using such approaches and discuss how to overcome these obstacles to test ecological theories and conserve biodiversity in a changing world.

.

Register Now

R.W. Moriarty Science Seminar Presents: Integrating the living and the dead: how functional traits can improve our understanding of early Cenozoic fern evolution

Speaker: Molly Ng, Carnegie Museum of Natural History

Abstract:Plants have different economic strategies to grow, survive, and reproduce. In leaves, these strategies are important in producing energy for the entire plant. Leaf mass per area (LMA), the ratio of leaf area to dry mass, is one way we can compare different strategies and has been shown to provide valuable insights about relative growth rates, productivity of plants within their environment, and even evolution. Ferns are important environmental and ecological indicators yet are not as well studied as angiosperms. The majority of modern ferns are considered ecological opportunists, hypothesized to have evolved as a response to angiosperm dominance ~55 million years ago, which also coincides with early Cenozoic climatic events. It is unclear whether ecological opportunity or climate is responsible for modern fern radiation, but LMA may provide insights into ecological opportunity. Using collections, both living (Powdermill Nature Reserve) and dead (herbarium and fossil specimens), Molly Ng estimates LMA across ferns to test whether “new” ferns strategize differently from “old” ferns. She determines plant response to atmospheric carbon dioxide and temperature, and tests phylogenetic conservation of traits across lineages. Finally, she introduces the work she is doing with fossils, which will ultimately improve our understanding of trait evolution and function within ferns.
.

Register Now