I've seen some authors using "R*" (e.g.: Tilman, 1982) as a way to rank different species in competition studies, but I want to know if there are other ways to do it.
Whit distance measures?. Jaccard? Riemannian distances are multivariate....
Hi, it is very good question. How height ist forest stand, how much ist structured, how much tree species is in forest stand and whot is your goal?. :-)
I am carrying an indoor study to evaluate the interspecific competition among 3 egg parasitoid species of a moth. I found covariance between environmental conditions (temperature) and the competitive ranking. I am using mortality rate, development speed, emergence rate and sex ratio to compare the different species, constructing a ranking for each one of these parameters. However, I want to know if there is a way to match these four parameters and construct a single ranking.
I would be very careful with this ranking concept. Very very careful. I'll lay my main cards out front: intraspecific diversity will be a huge issue and frequency dependent selection can lead to seemingly nonsense results. A competitive ranking might not even exist if the experiment is performed perfectly. For some added bonuses, maternal effect/non-genetic inheritance/environmental acclimation may be a challenge.
First, if you have a stable, homogeneous environment in which you desire to get the measure (and this isn't easy) I would encourage you to acquire mutants with a range of each continuous variable in the different genetic backgrounds. 'Strain level' variation may easily swamp your species signals. Second, you should be watching the dynamics with shifts in density and pay careful attention to paths and outcomes. You may be able to make certain assumptions about path independence in the dynamics, you might not. In any case, frequency dependence is the rule, not the exception. In such cases, there may not exist a ranking; or your ranking may be an illusion, part of a larger system you did not characterize. Non-transitivity in competition is more common than you might think.
Of course, if you have good data, and you want to make a model to combine four real valued variables with network node location/structure in a digraph (directed graph) that represents competition outcomes, it can certainly be done. I would encourage you to start with this more general model in mind, with the null hypothesis that the edges in the graph exist with non-zero weight in every possible pair, before you conclude that there exists a single ranking (ie. three nodes, two non-zero edges; or so on).
I would guess this answer raises questions. Feel free to discuss.
Dear Cherre,
If you have sufficient data on the entire network of interactions surrounding your suite of competitors, there are effective ways of ranking limiting transactions.
The best, Bob Ulanowicz
Dear Bob, thanks for your help, but the link is not working.
Dear Cherre, Please try copying and pasting in your browser. (http://people.biology.ufl.edu/ulan/pubs/Cnp.pdf). If that doesn't work, please email me at .
- Badges
- Science method
- Similar topics
- Biological Science
- Ecology