Research@StAndrews
 
The University of St Andrews

Research@StAndrews:FullText >
Research Centres and Institutes >
Centre for Research into Ecological & Environmental Modelling (CREEM) >
Centre for Research into Ecological & Environmental Modelling (CREEM) Masters Theses >

Please use this identifier to cite or link to this item: http://hdl.handle.net/10023/1636
This item has been viewed 31 times in the last year. View Statistics

Files in This Item:

File Description SizeFormat
Payal_Bal_MRes_thesis.pdf1.1 MBAdobe PDFView/Open
Title: Estimating gopher tortoise abundance through design-based and model-based methods
Authors: Bal, Payal
Supervisors: Thomas, Len
Issue Date: 2010
Abstract: Gopher tortoises are a small land tortoise that inhabit south-eastern United States and are listed as threatened due to habitat loss. They are rather hard to survey, since they occur at low density. This project was based on extensive data from systematic line transect surveys for gopher tortoises at Fort Gordon and Ichauway in Georgia, USA, collected by the Joseph W. Jones Ecological Research Centre. Despite excellent survey designs and field methods, the resulting estimates of abundance have shown high variance. Alternate methods for abundance and variance estimation that could lower the variance were explored. Both, design-based and model-based approaches to abundance estimation were attempted. For the design-based approach, abundance and associated variance estimates were obtained using the CDS/MCDS analysis engines in DISTANCE (Version 6.0). The cluster size estimation technique to scale burrow to tortoise abundance was used. Variance of the encounter rate component, that usually dominates the overall variance estimate for line transect data was originally calculated using the R2 estimator (used in DISTANCE 6.0 as the default) that is suited for random line placement. This was compared against encounter rate variance estimators developed for systematic designs (Fewster et al. 2009). As expected, the latter produced substantially lower variance estimates. For the model-based approach, abundance as well as occupancy was modelled by specifying GAMs using environmental covariates (where available) for the study sites. Resulting predictions were subjected to non-parametric and parametric bootstrapping for variance estimation. Parametric bootstrap suited to the model-based approach did not perform well because the underlying GAMs specified for burrow occupancy were found to be inaccurate. Due to the excellent design of the survey and the lack of sufficient information to model burrow occupancy accurately, design-based methods appeared to do better than the model-based methods for the data. The final estimates for both the study sites and the surface maps (only for Ichauway) produced need to be reviewed and must be considered in conjunction with the accompanying uncertainty.
URI: http://hdl.handle.net/10023/1636
Type: Thesis
Appears in Collections:Centre for Research into Ecological & Environmental Modelling (CREEM) Masters Theses



This item is protected by original copyright

This item is licensed under a Creative Commons License
Creative Commons

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

 

DSpace Software Copyright © 2002-2012  Duraspace - Feedback
For help contact: Digital-Repository@st-andrews.ac.uk | Copyright for this page belongs to St Andrews University Library | Terms and Conditions (Cookies)