Geospatial Analysis of Florida Attacks – summary information

Identifying factors that attract sharks or keep them away from a person or an area is

paramount to the understanding of shark attacks. Although the research literature

is far away from comprehending what attracts sharks in the first place, the opposite

is even less understood, as the history of finding a functioning shark repellent reveals

(reviewed in Sisneros and Nelson 2001). This project did not aim to reveal causal

links for shark attack but to identify potential high- and low-risk clusters along the

Florida east coast. The ability to identify high- and low-risk areas for shark attack

along a very large and heavily populated coastline opens the possibility of searching

for causal links, such as environmental pollution and other anthropogenic triggers

(Adams and McMichael 1999). Large-scale changes in physical parameters, such as

water temperature, could also be examined for their potential to facilitate shark

Marine and Freshwater Behaviour and Physiology 11

Downloaded by [University of West Florida], [Raid Amin] at 07:13 23 August 2012

migration and aggregation (Heupel and Simpfendorfer 2005) or meteorological

phenomena like short-term air pressure changes, which have been known to affect

the swim patterns of sharks (Heupel et al. 2003). Being able to identify low-risk areas

may also provide an invaluable baseline from which causal factors could be


Value of cluster analysis for analyzing shark attacks

Spatial statistics methodology has been well established in recent years (e.g., Roddick

and Spiliopoulou 1999; Kalnis et al. 2005; Chi et al. 2007; Rosswog and Ghose 2008).

The methodology has been applied in many fields (Sudakin et al. 2002; Coulston and

Riitters 2003; Amin et al. 2010). This project is the first to use it to analyze shark

attack data and to compare surfing and non-surfing attack rate data on a heavily

populated coast. The methodology does not identify specific links between

immediate risk and encounter outcomes but does permit comparisons across large

territories to reveal previously hidden correlations. The outcomes provide baseline

data and methodology to generate and test new hypotheses.