Unifying community detection across scales from genomes to landscapes

Authors: HUDON, STEPHANIE - Boise State University; ZAIATS, ANDRII - Boise State University; ROSER, ANNA - Boise State University; ROOPSIND, ANAND - Boise State University; BARBER, CRISTINA - Boise State University; ROBB, BRECKEN - Boise State University; PENDLETON, BRITT - Boise State University; CAMP, MEGHAN - Washington State University; Clark, Pat; DAVIDSON, MERRY - Boise State University; FRANKEL-BRICKER, JONAS - Boise State University; FORBEY, JENNIFER - Boise State University; HAYDEN, ERIC - Boise State University; RICHARDS, LORA - University Of Nevada; RODRIGUEX, OLIVIA - Boise State University; CAUGHLIN, TREVOR - Boise State University

Submitted to: Oikos
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/5/2021
Publication Date: 4/18/2021
Citation: Hudon, S., Zaiats, A., Roser, A., Roopsind, A., Barber, C., Robb, B., Pendleton, B., Camp, M., Clark, P., Davidson, M., Frankel-Bricker, J., Forbey, J., Hayden, E., Richards, L., Rodriguex, O., Caughlin, T. 2021. Unifying community detection across scales from genomes to landscapes. Oikos. 130(6):831-843. https://doi.org/10.1111/oik.08393.
DOI: https://doi.org/10.1111/oik.08393

Interpretive Summary: While biodiversity science increasingly encompasses multiple disciplines and biological scales, biodiversity data are often analyzed separately with discipline-specific methodologies and resulting inferences may be constrained across scales. Using topic models, we evaluate how biodiversity inference from disparate datasets can inform the conservation of interacting plants and vertebrate herbivores. We show how topic models can identify members of molecular, organismal, and landscape-level communities that explain the health and population dynamics of threatened herbivores. We present a future vision for how topic modeling could be used to design cross-scale studies that promote a holistic approach to detect, monitor, and manage both threatened species and biodiversity.

Technical Abstract: Biodiversity science increasingly encompasses multiple disciplines and biological scales from molecules and landscapes. Each scale has potential to inform conservation strategies and nested interactions between scales are common. Nevertheless, biodiversity data are often analyzed separately with discipline-specific methodologies and resulting inferences may be constrained across scales. To overcome this, we present a topic modeling framework to analyze community composition in crossdisciplinary datasets, including those generated from metagenomics, metabolomics, field ecology and remote sensing. Using topic models, we demonstrate how biodiversity inference from disparate datasets can inform the conservation of interacting plants and vertebrate herbivores. We show how topic models can identify members of molecular, organismal, and landscape-level communities that explain the health and population dynamics of threatened herbivores. We conclude with a future vision for how topic modeling could be used to design cross-scale studies that promote a holistic approach to detect, monitor, and manage both threatened species and biodiversity.