[Babase] Questions re. maturity in ranker

Susan Alberts alberts at duke.edu
Fri Sep 29 17:52:37 EDT 2006


Jun, your proposal is very clear. See comments below.

On Sep 29, 2006, at 4:11 PM, Jun Yang wrote:

> Thanks Susan.  That helps a lot.
>
> Based on this thread of discussion, I am proposing
> a possible ranker design:
>
> 1. User can choose any one of the six rank types.
>
> 2. The ranking will be automatically checked against
> other rank types for which ranking info already exists
> in Babase. Such checks prevent user from producing
> a ranking that is not consistent (in terms of relative
> order of individuals) with existing rank types.
>
> For example, suppose we already have ALF and ADM
> and are working on ALL. The ranker will automatically
> warn about inconsistency with ALF and ADM rankings,
> and disallow an inconsistent ALL ranking to be saved
> in Babase.

Yes this captures the logic of the situation. However, I think we do  
not want to disallow an inconsistent ALL ranking to be saved because  
I think that we will confront circularities the arise in ALL that  
don't occur in ALM or ALF.

For instance, let's say that juvenile male number 1 (called him JAM1)  
consistently ranks over JAM2. In ALM, they are and always have been  
(since their respective births) in that order, JAM1>JAM2. However,  
let's say that JAM2 happens to begin challenging adult females at an  
earlier age than JAM1. This will create a circularity in ALL that  
does not occur in ALM, in that JAM2 will simultaneously rank above  
some adult females but below JAM1, who himself is still below adult  
females. We might want to deal with this by allowing a inconsistency  
between ALL and ALM. In general these circularities have not impinged  
upon babase because we have not, for the most part, entered ALL ranks  
into babase. But we need a structure that deals with this. So, my  
thought is that we do not need to enforce consistency between ALL and  
the other rnktypes, but we do need to enforce consistency within each  
sex. I would like Jeanne to comment on this.

Related to this is the fact the the relationships between the  
different rnktypes is not strictly hierarchical, in that ALL is not a  
"master" ranking. ALM and ALF are the "master" rankings from which we  
want to derive other rankings. ALL is in a somewhat different category.

Also related to this is the fact that I think that the rnktypes  
should be:

ADF  Adult females
ADM Adult males
ALF All females
ALM All males
ALL all group members
Non-adult females (NAF?)
Non-adult males (NAM?)

I am not sure why we put all females and males < 7 in a rnktype but  
it is not making obvious sense to me at the moment, so I would also  
like Jeanne's input on this.

> This design gives some flexibility to the users
> in choosing which type of rank to work on first.
> For example, one can still work on ADF first, and then
> work on ALF, which will be constrained by ADF.

Yes the logic of this is right.
>
> Optionally, we can also implement an automatic
> procedure for producing rankings that are
> subsumed by bigger ones. For example, when
> an ALL ranking is entered, then all other types
> of rankings will be automatically populated.

See comments above about the fact that ALL is not at the top of the  
hierarchy, but in a different category I believe.
>
> The consistency checking makes implementation
> a bit more messy, but it looks like the right thing
> to do. How does this sound?


>
> On 9/29/06, Susan Alberts <alberts at duke.edu> wrote:
>> >A more general question along the same line:
>> >There are currently 6 types of rankings:
>> >
>> >ADF    Adult Females
>> >ALL    All group members
>> >ADM    Adult Males
>> >ALF    All Females
>> >ALM    All Males
>> >FYM    All females and males < 7
>> >
>> >Besides enforcing the consistency between ADF and ALF
>> >as you described above, should we also enforce the consistency,
>> >say, between ADM and ALM, and, in general,
>> >ALL with everything else?
>>
>> Yes; males and females represent a similar situation in that ADM
>> should be extractable from ALM in the same way that ADF should be
>> extractable from ALF
> ...
>> in practice we would not need to disentangle the ranks
>> of young males from those of adults.
> ...
>> why don't we just create one ranking -- the ALL ranking -- that
>> includes all the members in the group, and then just have a rule that
>> allows us to extract each of the other rnktypes from this ALL. We
>> don't do this because it is actually quite complicated to rank all
>> group members together,
> ...
>> In addition, the biological evidence suggests that within-sex
>> rankings are likely to be the relevant rankings from the animal's
>> perspective, rather than whole-group rankings.  So, although we have
>> all group rankings for some periods of time,, we don't create these
>> routinely.
>>
>> In general, age-sex classes rank as follows from highest to lowest:
>> Adult males, subadult males, adult females, juvenile males, juvenile
>> females. However, as I mentioned above, animals mature in a
>> non-systematic fashion and as they grow and climb ranks, brief
>> circularities result.
> _______________________________________________
> Babase mailing list
> Babase at www.eco.princeton.edu
> http://www.eco.princeton.edu/mailman/listinfo/babase

-----------------------------------------------
Susan Alberts, Dept. Biology, Duke University, Durham NC 27708. Phone  
919-660-7272, Fax 919-660-7293. alberts at duke.edu





More information about the Babase mailing list