Confronting the ghosts of the fading record: multi-proxy data and analytical advancements reveal historical fire severity dynamics

Authors: Cameron Naficy*, Oregon State University, Ben Bloem-Reddy, Department of Statistics, University of British Columbia, Lori D. Daniels, Forest & Conservation Sciences, University of British Columbia
Topics: Mountain Environments, Environment, Landscape
Keywords: fire severity, dendroecology, fire regime, historical ecology, paleoecology
Session Type: Virtual Paper
Day: 4/7/2021
Start / End Time: 1:30 PM / 2:45 PM
Room: Virtual 9
Presentation File: No File Uploaded

Paleoecology is haunted by problems that arise from missing data. Because paleoecological record quality often deteriorates with time since the present, this “fading record” problem can introduce substantial uncertainty and nonstationary constraints on meaningful ecological inference over time. However, a monolithic view of the missing data problem fails to recognize that there are different classes of missing data problems and appropriate solutions. Here, we apply a well-established statistical framework to highlight distinct classes of missing data problems and potential solutions in paleoecology, with specific emphasis on dendroecological reconstruction of historical fire severity. We review and evaluate new analytical fire severity reconstruction methods using a network of dendroecological data from western North America and a Bayesian logistic growth model that accounts for nonrandom missing data and temporal patterns of uncertainty. We use these methods to quantify fire regime space, defined by fire frequency and severity, for major forest types across western North America and to evaluate geographic patterns of fire regime diversity within these forest types. We also present initial results of a marked Poisson point process model of historical fire-fire interactions that incorporates fire severity, frequency, and biophysical co-variates. Finally, we demonstrate how combination of dendroecological fire severity reconstructions, spatial patterns of forest structure and composition derived from historical aerial imagery, and machine learning can improve spatial inferences of historical fire severity. Although reconstruction of historical fire severity remains a wicked problem in paleoecology, we highlight the potential for new research frontiers made possible by these recent quantitative, multi-proxy advancements.

Abstract Information

This abstract is already part of a session. View the session here.

To access contact information login