Use this URL to cite or link to this record in EThOS:
Title: Using statistical modelling to link disparate sources of available information to study factors that influence bird distributions
Author: Mammides, Christos
ISNI:       0000 0004 2729 1311
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2012
Availability of Full Text:
Access from EThOS:
Access from Institution:
The aim of this project is to develop a method to successfully link together various sources of disparate information and data in order to study and understand which factors influence bird communities in different areas. For many areas, especially those of high conservation priority, an enormous amount of information and data may already be available. By using the right tools they could potentially be linked to provide improved understanding of the mechanisms that influence the biodiversity in a region of interest. To test this hypothesis I used structural equation modelling (SEM) to link data from Kakamega Forest in order to study the effects of various socio-ecological factors on the bird species richness in twenty-two different parts of the forest. Kakamega Forest is Kenya’s only rainforest and despite its rich biodiversity, and especially avian diversity, the forest is highly threatened with less than half of its area containing indigenous vegetation. A similar modelling approach is used to study the factors that influence migratory birds found in thirty-eight Natura 2000 sites in Cyprus, designated under the EU’s Habitats Directive (officially known as the Council Directive 92/43/EEC of 21 May 1992 on the conservation of natural habitats and of wild fauna and flora). Using GIS analysis and available data on land use, vegetation cover, habitat diversities, human population densities, and road densities, I developed a path model explaining the observed bird species richness in those areas. Through this type of analysis I identified and quantified the impact of various habitat variables on bird species richness, which is one of the many measures of species diversity. Compared to other diversity indices, species richness is not influenced by species abundance and therefore it is an appropriate measure for studying the distribution of the species. This type of analysis however, does not allow us to indentify which species in particular are impacted by the variables identified so I used generalized linear mixed modelling to study interactions between habitat variables and certain species specific behavioural, morphological, and life history characteristics (including food choice, body length and clutch size), to examine how abundances and presence/absence are influenced by those variables in each site. The results are of significant conservation importance as they give us valuable insight on: a) which factors are most important in determining species richness and b) what species characteristics make birds more vulnerable to change in these factors. Moreover, the results demonstrate that by using an appropriate statistical method there is potential to successfully utilize the enormous amount of available information to derive important conservation conclusions.
Supervisor: Coulson, Tim Sponsor: A.G. Leventis Foundation ; Research Promotion Foundation (Cyprus)
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral