Developing A Geospatial Protocol For Coral Epizootiology

This dissertation explores how geographic information systems (GIS) and spatial statistics, specifically the techniques used to map, detect, and spatially analyze disease epidemics, could be used to advance our understanding of coral reef health. Given that different types of spatial analysis, as well as different parameter settings within each analysis, can produce noticeably different results, poor selection or improper use of a given technique would likely lead to inaccurate representations of the spatial distribution and false interpretations of the disease. For this reason, I performed a comprehensive review of the following types of exploratory spatial data analysis (ESDA): mapping and visualization methods; centrographic and distance-based point pattern analyses; spatial kernel density estimates (KDE) using single and dual versions of adaptive and fixed-distance KDEs in which the fixed-distance KDEs were performed using bandwidths calculated using 12 different estimation methods; SaTScan’s spatial scan statistic using both the Bernoulli and Poisson probability models; and last, local and global versions of the Moran’s I and Getis-ord G spatial autocorrelation statistics. Each technique was applied to an artificial dataset with known cluster locations in order to determine which methods provided the most accurate results. These results were then used to develop different geospatial analytical protocols based on the types of coral data available, noting that the most meaningful results would be produced using local spatial statistics to analyze data of diseased colonies and colonies from the underlying coral population at risk. Last, I applied the techniques from one of the protocols to data from a 2004 White-Band Disease (WBD) outbreak on a population of <i>Acropora palmata</i> corals in the US Virgin Islands. The results of this work represent the first application of geospatial analytical techniques in visualizing the spatial nature of a coral disease and provides important information about the epizootiology of this particular outbreak. Specifically, the results indicated that WBD prevalence was low with numerous significant disease clusters occurring throughout the study area, suggesting WBD may be caused by a ubiquitous stressor. The material presented in this dissertation will provide researchers with the necessary tools and information needed to perform the most accurate geospatial analysis possible based on the coral data available.

Name thesis:degree_name

Doctor of Philosophy (PhD)

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Oceanography and Atmospheric Sciences and Meteorology

Grantor

Oceanography and Coastal Sciences

Year dc:date.available

2012

Author dc:creator

• Lentz, Jennifer Anne

• local spatial statistics
• prevalence
• USVI
• colony
• GPS
• scale
• global spatial statistics
• medical geography
• ArcMap
• CrimeStat
• GeoDa
• sampling scale
• Monte Carlo simulation
• Ripleys K
• Disease Mapping and Analysis Program (DMAP)
• spatial analysis
• disease
• White-band Disease (WBD)
• transect
• spatial epidemiology
• epizootiology
• Buck Island
• Acropora palmata
• Caribbean
• coral reefs
• GIS

Statement dc:rights

• unrestricted
• Release the entire work immediately for access worldwide.

Identifier

etd-04182012-151443
https://repository.lsu.edu/gradschool_dissertations/551

OAI identifier oai:identifier

oai:repository.lsu.edu:gradschool_dissertations-1550