[Coral-List] Fractal analysis and reef complexity
Sam Purkis
purs at geo.vu.nl
Mon May 31 14:37:18 EDT 2004
Dear Coral-List,
In response to Harry Goudge's recent posting concerning the application of
fractal analysis to quantify habitat complexity, I have recently looked
into the spatial complexity of reefs on Ikonos satellite imagery. Although
at a much larger spatial scale to that relevant to fish communities, the
methodology is likely to be transferable.
Fractal patterns have certainly been observed in the small-scale structures
of coral (relevant papers would be Bradbury and Reichelt 1983; Basillais
1997) and using Ikonos satellite imagery in the Arabian Gulf I observe
behaviour in patches of different substrates types that scale with power
laws within certain thresholds. The pattern is robust over three decades of
magnitude both for the frequency-size distribution of patches and using the
box-counting metric to quantify patch boundary 'wiggliness' (complexity).
To look at habitat complexity at scales relevant to reef fish, I'd start by
recording the rugosity of the seabed through photography and then analyse
the relief using box-counting. On the positive side, the fractal dimension
is elegant in so far as it represents a single index of complexity. On the
negative side though, to really say that a system is a candidate for a
fractal, the scale-invariant behaviour must hold for numerous decades of
magnitude and I'd say that three is really the minimum. Keep this in mind
when deciding over which scales you'll collect your data.
Handy refs:
Basillais E (1997) Coral surfaces and fractal dimensions: a new method. C R
Acad Sc Paris, Life Sci 320:653-657
Bradbury RH, Reichelt RE (1983) Fractal dimension of a coral reef at
ecological scales. Mar Ecol-Prog Ser 10:169-171
All the best,
Sam
Sam Purkis Ph.D.
National Coral Reef Institute
Oceanographic Center
Nova Southeastern University
8000 N. Ocean Drive, Dania
FL 33004
Formerly: Institute of Earth Sciences, Vrije Universiteit, Amsterdam, The
Netherlands
More information about the Coral-List
mailing list