[R-sig-eco] point process analysis

[Dylan Irion]

Hi list,

This is more of a general spatial statistics question, but I think 
someone here might be able to point me in the right direction. I have a 
dataset collected from active tracks of a marine species and am 
interested in looking at the the swimming depth of the animals across 
various sizes and crepuscular periods. Initially I wanted to just test a 
simple hypothesis with a two-way ANOVA, but decided to rather model the 
relationship, and also to include location as an explanatory variable. 
I've been skimming Bivand's Applied Spatial Data Analysis book for some 
clues and thought maybe section 7.5.3 Binary Regression Using 
Generalised Additive Models, was what I should be looking at. My case 
would certainly be a linear regression rather than binary, but is this 
the right approach? a GAM with a smoothed spatial function on the 
coordinates from the mgcv package? Essentially I would have swimming 
depth (absolute or percentage of bottom depth) as the  dependent 
variable, and shark name (indicator of size class), crepuscular cycle 
(day, night, dusk, dawn), and the smoothed coordinate function as 
independent variables. Am I headed in the right direction? Spatial Point 
Processes?

Kind Regards,

Dylan Irion

-- 
University of Cape Town (MSc. Candidate)
Oceans Research
PO 1767
Mossel Bay, Western Cape
South Africa 6500
www.oceans-research.com
+27(0)723531503