# Coral Puzzles and Spatial Statistics

So far, I’ve only really written about the exciting things that I get to do because of the laid-back nature of my research. Let me explain more about what I actually do…

I’m looking at the spatial patterns of coral and coral disease at different coral reef sites around ST. Croix USVI. All of the data was collected by Dr. Erinn Muller back in 2009 when I was just graduating High School, so I’ve never actually been to the study area.

It’s easiest if I just explain most things in pictures:

St. Croix with some of the sites I am using marked with pins.

Dr. Erinn Muller taking a photo-quadrat of a 10mX10m coral reef site

I identify the coral species and disease in each row, and then I merge everything together before giving colonies coordinates.

Spatial statistics used to understand coral and disease clustering

The statistics part is where people usually get a bit confused. Lets back up to why I’m using spatial statistics.  Spatial epidemiology maps the location of diseased corals and uses spatial statistics to describe and quantify coral-disease transmission. Spatial statistical methods identify spatial patterns as random, aggregated, or regular at different spatial scales (Jolles et al, 2002). Different spatial patterns are observed for different types of disease transmission. A pathogen that disperses long distances may produce random patterns in disease activity over small spatial scales (Jolles et al, 2002). Additionally, random patterns of diseased individuals may indicate a non-communicable disease. However, at small scales close to the source of the pathogen, local hydrodynamics could cause nonrandom patterns in dispersion (Jolles et al, 2002). When a disease is transmitted by direct contact, the underlying random pattern will be broken up by clusters in close proximity to each other (Jolles et al, 2002). Water borne pathogens would show up as more extensive clusters with larger gaps (Jolles et al, 2002). A contagious mode of transmission is likely if diseased individuals are consistently more clustered than the natural distribution of all individuals (Muller, 2011).

I am trying to determine if Caribbean yellow band disease is a contagious disease by using spatial epidemiology. I am comparing clustering levels of two species of coral in all of the sites and the clustering levels between healthy and diseased sites. In each site, I calculate the euclidean distance (shortest straight line distance between coral colonies) between all coral colonies of each species that I am interested in. I then run a kolmogorov-smirnoff test to compare the spatial distribution of clustering of the two species and of the healthy and diseased sites.

Caribbean Yellow Band Disease on Montastraea annularisCaribbean Yellow Band Disease on Montastraea faveolata

I do all of my statistics in R. Another part of my project entails creating a disease transmission model, but that’s a whole other blog post worth of stuff. All of my work is done on a computer, and I multi-tasked a lot during the previous weeks (as shown below). I would be identifying coral on one computer, merging sites on another, and keeping track of data on the third. Now I have all of my mapping done and I am just playing around with statistics and graphs and trying to get my model to work.

My desk on a typical day at work!

Jolles, Anna E., Patrick Sullivan, Alisa P. Alker, and Drew Harvell. “Disease Transmission of Aspergillosis in Sea Fans: Inferring Process from Spatial Pattern.” Ecology 83 (2002): 2373-378.

Muller, Erinn M. Spatial and Temporal Dynamics of Coral Disease in the US Virgin Islands and Puerto Rico. Diss. Florida Institute of Technology, 2011.