The probability of cannibalism


I am interested in calculating the probability of cannibalism occurring in groups of fish (25). The species I am looking at are known to cannibalize each other via consumption of their prey head first and whole. This type of cannibalism is limited by the mouth gape of the predator compared to the maximum body depth of the prey. Mouth gape and total length of the fish are related and thus knowledge of the potential for cannibalism can be gained from the variation in lengths within the group.

I have taken weekly measurements of total length of each of the fish in each group, and would like to use these data, and my knowledge of the relationship between length and gape, to show the probability of cannibalistic mortality at each of the weekly timepoints so I can compare it to observed mortality.

Logistic regression has been used to determine cannibalism when looking at multiple pairs of fish; my question is, can I use it for groups, and if so, how?

Your input will be gratefully received!



TS Contributor
maybe others wil help a bit more, to me it feels like a poisson regression - you could model the number of cannibalistic death for each group. I guess, if the groups are karge and cannibalism relatively rare such a model could work well.