Buffer zone 'shape' & state-space matrix rationale

[Paolo Strampelli]

Hi all,

I am currently working on utilising SPACECAP to obtain a density estimate for a leopard population in Mozambique. I had two quick questions I was hoping someone could help me with, both regarding the rationale for certain choices.

1. It seems that similar studies are fairly evenly between having either a "minimum area rectangle" encompassing the traps, upon which to then build a (rectangular) buffer, or building the buffer upon a polygon which is not rectangular in shape, but rather follows more closely the outline of the camera traps. Given that my array is not exactly rectangular in shape, the resulting buffers would be fairly different, and although I'm aware in the end the difference during analyses shouldn't be massive, it would be great to understand which of the two methods should be employed. 

2. The second question regards the state-space matrix - I understand it must be smaller than the radius of the species' home range, but was hoping to understand better the rationale between choosing, say, having the points 500 m apart versus 2 km apart (both of which I have encountered for large felids). Does varying this have an effect solely on computational time, or is there a risk of choosing the 'wrong' pixel size? 

Thank you so much in advance to anyone who can help. I'm aware these are not critical issues, but it would be great to understand more about the reasoning behind these choices.

Best,

Paolo Strampelli

[Arjun Gopalaswamy]

Hi Paolo:

My answers are in CAPS below. For further questions, feel free to write to me off the forum.

Arjun

On 19/01/2016 02:13, "Paolo Strampelli" <paolo.st...@gmail.com> wrote:

Hi all,

I am currently working on utilising SPACECAP to obtain a density estimate for a leopard population in Mozambique. I had two quick questions I was hoping someone could help me with, both regarding the rationale for certain choices.

1. It seems that similar studies are fairly evenly between having either a "minimum area rectangle" encompassing the traps, upon which to then build a (rectangular) buffer, or building the buffer upon a polygon which is not rectangular in shape, but rather follows more closely the outline of the camera traps. Given that my array is not exactly rectangular in shape, the resulting buffers would be fairly different, and although I'm aware in the end the difference during analyses shouldn't be massive, it would be great to understand which of the two methods should be employed. 

BOTH THESE APPROACHES SHOULD YIELD SIMILAR ESTIMATES. HOWEVER, IN YOUR SECOND APPROACH (WHERE BUFFER CONTOURS FOLLOW ARRAY) WILL BE MORE EFFICIENT IN THE SENSE THAT DUE TO LESSER NUMBER OF PIXELS,  SUBSEQUENTLY A SMALLER AREA, THE VALUE YOU WILL SET FOR “M” WILL BE REDUCED. THIS WILL LEAD TO A FASTER ANALYSIS. YOU SHOULD NOTE THOUGH THAT USING THIS SECOND APPROACH SHOULD STILL ENSURE THAT THE BUFFER IS LARGE ENOUGH AND THE SETTING OF M SHOULD NOT ALLOW FOR ANY PROBABILITY THAT AN INDIVIDUAL OUTSIDE THIS BUFFER GETS CAUGHT BY YOUR CAMERA TRAPS.  

2. The second question regards the state-space matrix - I understand it must be smaller than the radius of the species' home range, but was hoping to understand better the rationale between choosing, say, having the points 500 m apart versus 2 km apart (both of which I have encountered for large felids). Does varying this have an effect solely on computational time, or is there a risk of choosing the 'wrong' pixel size? 

THE CHOICE REPRESENTS A TRADE-OFF BETWEEN HOW ACCURATELY WOULD YOU LIKE TO DESCRIBE THE LOCATION OF ACTIVITY CENTRES IN YOUR STUDY AREA AND THE RESULTING COMPUTATION TIME.  SETTING A VERY FINE RESOLUTION ON YOUR STATE-SPACE WILL RESULT IN GOOD ESTIMATES OF ACTIVITY CENTRE LOCATIONS AND OTHER PARAMETER ESTIMATES SUCH AS SIGMA, BUT WILL RESULT IN MASSIVE COMPUTATION TIME.

[Paolo Strampelli]

Hi Arjun,

Thank you very much for your response and availability, appreciate it.

Very informative answers - great to have a better understanding of it all!

Thanks again,

Paolo