(good points from Diana Carolina)

"I was talking to Charles, because I am a little concerned about the use
of windows (for the landscape analysis), because which is that ideal
distance, or ideal area that take the landscape process that affect the
movement of the organisms? Also, when we are talking of animal
disperssion, it is very complex talk about an ideal area of study
because it depends in the taxa...



What I´m trying to say is that the dispersal process in tropical rural
landscapes are complex process that are characterized by the landscape
characteristics which they are in. The spatial patterns of the
landscape influence this process, is for this, that you have to take
into account the fragmentation degree, understanding of this: the
isolated degree of the forests fragment, the functional distance
between them, the cover types that conform the matrix, the friction of
the matrix in the movement and disperssion of the organisms. Also, have
to take into account the human intervention in the forests, because
forest in rural landscapes are used by the farmers (in summer, they get
the cattle in the forest for shadow, they get wood from it, and more
things...), and the most important thing is to take notice of the
pasture management (This is another characteristic, because when we are
working with active pastures, the natural regeneration is modified or
altered by the cattle, but when the pastures are abandoned, the process
is different. This kind of antropic activities must be controled or be
identified for not to get in a subestimation of the dispersal process
in open or arboreal pastures)."

Diana Carolina -- you're absolutely right on a number of points:

• the window size can be critical -- scale dependence! The spreadsheet that charles and i put together was motivated by the question: "how many non overlapping analysis windows will we have if each window has a radius of X?" We were wondering how many actual plots we might have based on their location.
  
• If dispersal mechanisms are the processes of interest that generate the patterns we see, these mechanisms likely operate on different scales -- so a one sized fits all analysis window probably won't cut it.
• Finally, the human component is VERY important. None of our landscape analysis may matter if there is  substantial variability in the anthropogenic use where plots were sampled. If cattle ate all the plants except those unpalatable we'll run into some problems for sure!