Evaluating the Relationships of Phenological and Inter-Annual Landscape Dynamics with Farmland Biodiversity using Multi-Spatial and Multi-Temporal Remote Sensing Data

It appears your Web browser is not configured to display PDF files. Download adobe Acrobat or click here to download the PDF file.

Click here to download the PDF file.

Creator: 

Alavi-Shoushtari, Niloofar

Date: 

2019

Abstract: 

Agricultural landscapes are highly variable ecosystems and are home to many species. Farmland spatial heterogeneity has been shown to negatively impact biodiversity. However, phenological and inter-annual agricultural landscape dynamics also have potential to affect species diversity. Remote sensing provides data that enable monitoring landscape changes at multiple temporal and spatial scales. The goal of this research was to determine the response of biodiversity to phenological and inter-annual landscape dynamics.

The study area is the predominantly agricultural region of eastern Ontario. Ninety-three sample landscapes were selected prior to this research. Biodiversity data were collected during the summers of 2011 and 2012 within a 1 × 1 km area at each landscape. This extent and 3 × 3 km were selected for this research to analyze the impacts of spatial scale on biodiversity response.

Relationships between biodiversity and vegetation phenology were modelled using MODIS NDVI, while relationships between biodiversity and long term inter-annual vegetation changes were modelled using Landsat NDVI and Tasseled Cap components. Random Forest Regression was used to determine variable relative importance over all biodiversity models. The most important variables were identified and subsequently used in step-wise regression to determine model significance, the variables entered, and the direction of their relationship with biodiversity.

Results demonstrated that phenological and inter-annual changes in vegetation dynamics were related to biodiversity. For MODIS, most 3 × 3 km models were significant, whereas most 1 × 1 km models were not. For Landsat, model performance was not consistently different for the two extents, indicating that model performance can depend on landscape extent when coarse spatial resolution data are used. Plants and beetles showed the strongest relationships with MODIS phenological variables, particularly greenness onset date (negative) and NDVI (positive), as well as Landsat mean NDVI (positive) and mean Tasseled Cap Brightness (negative) and Wetness (negative). The inter-annual variability of Landsat NDVI and the percentage of pixels in a landscape with significantly decreasing trends in NDVI demonstrated a negative relationship with plant diversity. This thesis emphasized the importance of spatial and temporal variations in the landscape on biodiversity, especially in highly variable landscapes such as agricultural lands

Subject: 

Agronomy
Remote Sensing
Biostatistics

Language: 

English

Publisher: 

Carleton University

Thesis Degree Name: 

Doctor of Philosophy: 
Ph.D.

Thesis Degree Level: 

Doctoral

Thesis Degree Discipline: 

Geography

Parent Collection: 

Theses and Dissertations

Items in CURVE are protected by copyright, with all rights reserved, unless otherwise indicated. They are made available with permission from the author(s).