Document Type

Dissertation - Open Access

Award Date

2018

Degree Name

Doctor of Philosophy (PhD)

Department / School

Geospatial Science and Engineering

First Advisor

Niall P. Hanan

Keywords

Africa, ecosystem modeling, remote sensing, savanna, woody cover, woody encroachment

Abstract

African savannas cover roughly half of the continent, are home to a great diversity of wildlife, and provide ecosystem services to large populations. Savannas showcase a great diversity in vegetation structure, resulting from variation in climatic, edaphic, topographic, and biological factors. Fires play a large role as savannas are the most frequently burned ecosystems on Earth. To study how savanna vegetation structure shifts with environmental factors, it is necessary to gather site data covering the full gradient of climatic and edaphic conditions. Several earlier studies have used coarse resolution satellite remote sensing data to study variation in woody cover. These woody cover estimates have limited accuracy in drylands where the woody component is relatively small, and the data cannot reveal more detailed information on the vegetation structure. We therefore know little about how other structural components, tree densities, crown sizes, and the spatial pattern of woody plants, vary across environmental gradients. This thesis aimed to examine how woody vegetation structure and change in woody cover vary with environmental conditions. The analyses depended on access to very high spatial resolution (m) satellite imagery from sites spread across African savannas. The high resolution data combined with a crown delineation method enabled me to estimate variation in tree densities, mean crown size and the level of aggregation among woody plants. With overlapping older and newer imagery at most of the sites, I was also able to estimate change in woody cover over a 10-year period. I found that higher woody plant aggregation is associated with drier climates, high rainfall variability, and fine-textured soils. These same factors were also indicative of the areas where highly organized periodic vegetation patterns were found. The study also found that observed increases in woody cover across the rainfall gradient is more a result of increasing crown sizes than variation in tree density. The analysis of woody cover change found a mean increase of 0.25 % per year, indicating an ongoing trend of woody encroachment. I could not attribute this trend to any of the investigated environmental factors and it may result from higher atmospheric CO₂ concentrations, which has been proposed in other studies. The most influential predictor of woody cover change in the analysis was the difference between potential woody cover and initial woody cover, which highlights the role of competition for water and density dependent regulation when studying encroachment rates. The second most important predictor was fire frequency. To better understand and explain the dominant ecosystem processes controlling savanna vegetation structure, I constructed a spatially explicit model that simulates the growth of herbaceous and woody vegetation in a landscape. The model reproduced several of the trends in woody vegetation structure earlier found in the remote sensing analysis. These include how tree densities and crowns sizes respond differently to increases in precipitation along the full rainfall range, and the factors controlling the spatial pattern of trees in a landscape.

Library of Congress Subject Headings

Savannas -- Africa.
Savanna ecology -- Africa.
Vegetation dynamics -- Africa.
Landscape changes -- Africa.
Remote sensing -- Africa.

Description

Includes bibliographical references

Format

application/pdf

Number of Pages

138

Publisher

South Dakota State University

Comments

The African Savanna Vegetation Model developed with this dissertation can be found in the GSCE Graduate Student Datasets community here.

URL: https://openprairie.sdstate.edu/gsce_grad_data/1/

Download

Share

COinS
 

Rights Statement

In Copyright