Quantifying the causal effects of climate change on land use and bird biodiversity

Chen, Luoye

Permalink

https://hdl.handle.net/2142/115905 Copy

Description

Title

Quantifying the causal effects of climate change on land use and bird biodiversity

Author(s)

Chen, Luoye

Issue Date

2022-07-13

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

Khanna, Madhu

Doctoral Committee Chair(s)

Khanna, Madhu

Committee Member(s)

• Christensen, Peter
• Deryugina, Tatyana
• Hutchins, Jared

Department of Study

Agr & Consumer Economics

Discipline

Agricultural & Applied Econ

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Land Use
• Climate Change
• Biodiversity

Abstract

Climate change is emerging as a pervasive threat to natural resources such as land and biodiversity that provide the principal basis for human livelihoods and well-being, including the supply of food, energy, and ecosystem services. Accurately quantifying the impacts of climate change on these natural resources is critical to informing the design of agricultural and environmental policies to mitigate these impacts. However, estimating the causal effects of climate change on land use and biodiversity is complicated due to well-documented challenges, including the lack of comprehensive data and the difficulty of constructing reliable measurements for land use change and biodiversity. In addition, the relative importance of different climate variables may vary across species and land types as well as across space and time, which hampers our ability to accurately assess the impacts of climate change on these natural resources. In this dissertation, I provide the first large-scale causal evidence on the heterogeneous effects of climate change on agricultural land use, farmers’ production decisions, and the biodiversity of bird species that provide intangible value to humanity. I overcome the challenges documented above by utilizing novel fine-scale datasets and both econometric and machine learning techniques. By providing insights on these topics, I contribute to a broad literature on climate change and natural resource economics. This dissertation is composed of three chapters as follows. Bird population in the United States is declining at alarming rates. While a few studies have investigated the association between climate change and bird population trends, these studies are either investigating correlational relationships and hence, subject to confounding factors, or focus only on individual species or specific regions. The first chapter provides the first continental-scale evidence of the effect of a changing climate on bird diversity and the extent to which birds can adapt to these changes across species and regions. Using a long-term dataset of the North American bird population from 1980 to 2015, I find robust, statistically significant negative effects of additional high-temperature days on bird biodiversity. The effects are mainly confined to the species richness and abundance rather than Shannon-weaver index, which implies the potential heterogeneous effect of climate change on different components and attributes of bird biodiversity. Furthermore, I do not find statistically significant evidence of the ability for US bird population to adapt and offset the negative impact of climate change in the long run. This work contributes to the understanding of climate change impacts on an important but often ignored topic – species loss. These findings explain the role of climate change on the loss of bird population, especially on species and biodiversity metrics, which could help to support the design and implementation of related conservation policies in the US. Land use change (LUC) caused by the conversion of grasslands to cropland is a pressing concern due to its implications for ecosystem services such as carbon emissions, biodiversity, and water quality. However, there is limited understanding of where, at what rates, and why this land use change occurs. In the second chapter, I examine how climate change and crop prices affect the land conversions between cultivated cropland and grassland. I compile a novel dataset that includes crop-grass conversions and rich biophysical characteristics for each 500-meter land pixel covering the eastern US from 2000-2019. Utilizing this high-resolution data, I provide the first large-scale investigation of how US farmers and landowners adjust crop-grass conversions in response to weather shocks and crop price fluctuations, and how these adjustments vary across land pixels with different biophysical attributes. This chapter contributes to the literature by showing that crop prices, short-term rainfall shocks, and long-term changes in local precipitation conditions jointly affect the likelihood of crop-grass conversion. Using machine learning inference, I also characterize heterogeneity in these adjustments across land parcels with different biophysical attributes. The effects of rainfall shocks and price fluctuations on crop-grass conversions are mainly confined to land parcels at the margin of agricultural production, with relatively lower water storage capacity, less productivity, and steeper slopes. which implies the importance of incorporating these indicators into agricultural policies such as crop insurance to encourage sustainable farming methods to build soil health and resilience against extreme weather events. These findings suggest that crop-grass conversion is not a static phenomenon and that it is important to examine how it evolves in response to various factors that may change over time. Increasing crop prices may lead to an expansion of cropland, raising significant concerns about the effects of policy interventions such as biofuel mandates on land use change (LUC). The recent explosion in the availability of satellite imagery provides an opportunity to track the land use change at the field level. However, assessments of LUC from satellite products are only as good as the data and assumptions upon which they are based. Classification of landscapes into various categories of land use from satellite data is often subject to significant variation in definitions of land categories and substantial measurement errors, which could lead to uncertainty in LUC estimates. The previous LUC studies have relied exclusively on the CDL, which provides detailed land use classification for crops, but that may be at the expense of accuracy of land categories, especially for noncropland. In the third chapter, I provide the first systematic comparison of the accuracies of three dominant satellite land cover data---CDL, MODIS, and LCMAP---and examine the extent to which measurement errors can lead to uncertainty in estimating price-induced LUC in empirical analysis. I find that crop prices have a positive and statistically significant effect on total cropland acreage, regardless of the data source. However, the magnitudes of estimates are substantially different across data sources. Specifically, the price-acreage elasticity estimated by CDL is at least two times larger than those from MODIS and LCMAP. This difference is statistically significant and persists even after controlling for the variations in the definition of land categories. In addition, I also find that LUC estimates on noncropland are highly sensitive to data sources and the methods used for land category aggregations. This chapter contributes to the literature by underlining the importance of constructing and comparing inferences from different satellite land cover datasets. It also provides a practical approach to understanding and quantifying the uncertainty in LUC estimates and useful policy insights.

Graduation Semester

2022-08

Type of Resource

Thesis

Copyright and License Information

Copyright 2022 Luoye Chen

Owning Collections