Drivers of landscape evolution in northern ecosystems: Shrub expansion, vegetation-fire interaction, and permafrost degradation

Schore, Aiden I. G.

Permalink

https://hdl.handle.net/2142/132650 Copy

Description

Title

Drivers of landscape evolution in northern ecosystems: Shrub expansion, vegetation-fire interaction, and permafrost degradation

Author(s)

Schore, Aiden I. G.

Issue Date

2025-11-21

Director of Research (if dissertation) or Advisor (if thesis)

Lara, Mark J

Doctoral Committee Chair(s)

Punyasena, Surangi W

Committee Member(s)

• Diao, Chunyuan
• Fraterrigo, Jennifer M

Department of Study

Plant Biology

Discipline

Plant Biology

Degree Granting Institution

University of Illinois Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Shrub expansion
• vegetation-fire interaction
• permafrost
• ecology
• tundra
• remote sensing

Abstract

As high-latitude ecosystems warm, key biogeophysical properties controlling ecosystem carbon and energy dynamics, such as plant community composition and permafrost stability may shift in response to novel climate regimes. In the tundra of northern and northwestern Alaska, some of the most profound terrestrial responses to recent climate and environmental change have been the expansion of tall shrubs, the degradation of permafrost, and the increase in frequency and severity of wildfires. These responses, however, are not mutually exclusive. For example, changes in hydrology due to the degradation of permafrost can alter the frequency of wildfire by either inundating the soil with water and reducing fire frequency or directing water elsewhere and increasing the prevalence of wildfire. These changed wildfire frequencies along with thermokarst can free space and nutrients by removing and breaking down existing vegetation and organic matter, letting tall shrubs expand their range. Patterns tundra landscape evolution are complex, as the tundra of Alaska is far from a monolith, instead displaying extreme heterogeneity, from macro to micro-climate and topographic scales. Therefore, knowledge of fine-scale patterns and controls on land cover change dynamics across tundra ecosystems remain limited. This dissertation aims to expand our knowledge of the local-scale controls on regional-scale land cover change processes, using advances in computer vision to integrate multi-scale remote sensing observations with machine and deep learning algorithms for change detection. In the first chapter, I focus on tall-shrub expansion in the central Seward Peninsula, which is the southernmost part of the tundra region of Alaska and lies on the continuous/discontinuous permafrost boundary. By choosing a single area, I am able to isolate the environmental factors driving shrub expansion while controlling for warming. The chapter finds that there is a dynamic ceiling on potential shrub habitat controlled by climatic and edaphic factors as well as nitrogen inputs from alder shrubs. The second chapter analyzes the interaction between vegetation and fire by creating a new index to measure fire severity based on the plant functional types present on the ground. Current fire metrics are largely insensitive to ground vegetation, but different plant functional types have different properties as fuels and in recovery after a fire. Based on high-resolution satellite imagery and ground data from four recent fires, I am able to demonstrate that incorporating plant functional types into fire severity assessments improves the accuracy of the measurements. For my final chapter, I use more powerful computational tools to examine patterns in permafrost degradation across the tundra region of Alaska. I identify ice wedge degradation using object-based image analysis and a convolutional neural network and use gradient-boosted regression trees to analyze the environmental drivers of the degradation over the last 70 years. My analysis finds permafrost degradation to increase over time, but that much of this degradation is ephemeral, with older features draining as new ones develop, both processes distinctly altering landscape properties with potential impacts on hydrology, vegetation succession, and fire regimes. Understanding the causes of historical patterns of tundra land cover change will improve our ability to predict the potential consequences of future landscape reorganization as climate change and disturbance regimes intensify. In this dissertation I develop geospatial machine/deep learning models that show that broad regional Arctic land cover change patterns are controlled by predictable biophysical properties and processes at much smaller, more local scales. In doing so, I provide new tools to improve landscape change predictions going forward.

Graduation Semester

2025-12

Type of Resource

Thesis

Handle URL

https://hdl.handle.net/2142/132650

Copyright and License Information

Copyright 2025 Aiden Schore

Owning Collections