Community Server

hello, I am biologist and trying to establish the level of significance at which I can identify trophic guilds in a fish community. I have a matrix of resource use with 12 predators and 17 prey items. I have read that a bootrapping technique can give me the solution. To do so I am suppose to create a programm that:

1. randomly re-assign (with replacement) values of the percentage of prey that a species is eating. The conditions to this re-arregaments is that values of zero should not be changed and that al re-arregements should yield to a fixed value of diet breadth (already calculated) for each species. I have to do this boottraping prodedure 1000 times and with this new data matrices calculate again dietary similarities between species.

Finally i will get a frequency distribution of pseudovalues of diet similarity that will help  me to calculate at which level of similarity I can consider guilds. I am not a programmer and would like to know if somebody can tell me how can I create a programm like this in statistics101 (if it is possible). can try to explain better if someone think that can help me.

thanks,

Latimeria,

I'll need more information to help work this out.

From what you wrote, you have a matrix of 12 X 17 = 204 entries representing consumption percentages, some of which are zero. And I assume that by a "fixed value of diet breadth" you mean that for each predator, the sum of the amounts of all its prey stays constant. But I would need details on how values are to be re-arranged. I'm guessing that you mean for example, if predator A has five prey items, totaling 100%, then you want to randomly select new percentages for those same prey items, again totaling 100%. Also, I'm not clear on what is meant by "dietary similarities" and how those are measured.

I'd suggest that you first write down a list of steps, in words, in great detail as to how to perform the computation you want to do, including the math calculations needed for any step that needs them. This would constitute a description of a procedure as to how you would do the entire bootstrap process by hand. You can post that description here, or if you prefer, email it to me. From there, it could be rewritten as a program in Statistics101.

John