1. Variation of Mantel test?

Hello. I am new to this forum but it looks like it will be a great reference!

For my thesis I used Mantel tests to compare non-parametric behavioral data against genetic relatedness data. So, I had behavior rates in a square matrix and compared them against genetic relatedness values in a square matrix. The matrices had the same lay out (all individuals were in the same cells of the matrix). This told me if there was a relationship with the behavior rates and genetic relatedness between individuals.

Now, I have taken the same matrices and broken the individual animals up into 2 groups based on their location. So I have Group A and Group B. I need to compare the different behavioral matrices of Group A with A and then Group A with B to see if their location influenced their behavior rates. There are different numbers of individuals within Groups A and B. With a Mantel test the matrices have to be the same size, I also believe they are the same individuals represented in the 2 compared matrices. The test I need would be comparing matrices of 2 different groups against each other. Does anyone know of such a test? Or does anyone have other advice?

Hopefully this is clear, let me know if anything needs clarified!

2. Re: Variation of Mantel test?

hi,
assuming that the Mantel test is the right one for your research question I would try the following technique, related to the bootstrap method.

I would pick two random samples ( with replacement) from the two populations so, that the samples satisfy the requirements for the test, run the test and calculate the the test statistic. Repeat the process a large number of times (500? 1000?) and calculate the average of the the test statistic and report that, together with the method used to obtain the result.

I am not 100% sure that the bootstrap can be recommended for this test and/or whether it will be accepted in your thesis. Maybe someone else will also chime in and help. A negative argument is never a strong one, but I see no reason why this should not work .

regards
rogojel

 Tweet