Use Convert software freely downloaded. it exports data into different formats.

I used RDA (dbRDA and RDA on various dependent variables) for landscape genetics analysis and a methods paper last year (individual-based). It depends what your dependent variable is going to be. From what I saw in the literature, measures of genetic diversity, sPCA spatially lagged scores from adegenet, and connectivity indices between populations were the most common. 

To use pair-wise measures of population differentiation between populations, you would have to utilize a dbRDA (distance-based RDA), and can use measures such as Fst. I am not familiar with any method that uses raw genotype data to perform an RDA analysis. 

I generally utilize R packages to convert my data for RDA, but it does really depend what type of dependent variable you plan to use. I primarily use sPCA in adegenet because it is a great way to remove noise and gives nice results.  

However, to have power to detect landscape effects on gene flow between populations, I think you will have to use pair-wise data. Most genetic programs will give an output of pair-wise Fst tables. I like the R package DiveRsity, but programs such as Spagedi, Genalex, Genepop, and Fstat also work. Programs such as convert (as mentioned above) can create input files for programs. Both DiveRsity and adegenet (R packages) take genepop files. If you don't want to use Fst, Spagedi gives a lot of alternatives, and outputs decent tables. 

I hope this helps,

Liz K

Thank you for the answer Elizabeth! I'm using allele frequencies as my genetic input data. I've managed to run the RDA by now, and I get significant effects of my explanatory variables on the genetic pattern after the ANOVA. Now I'm trying to run a partial RDA, conditioned on geography, i.e. I want to disentangle the effects of the coordinates of my sampling stations from the other explanatory variables (e.g. salinity, nutrients etc.). This is causing me problems, I suppose I don't have the condition matrix in the right format. Should that information be included as an own third matrix, or should it be included in the explanatory matrix? I'd appreciate any advise on this.

I had to check back at my simulations (been awhile). I had everything in a columnar format in separate files for genetic and landscape data (spatial coordinates and landscape data in one). I used the package vegan in R. You just have to make sure that all the numbers match up in R (same number of columns for all files), and for normal RDA, everything must be columnar format. 

As long as you define the correct variables within the RDA model, there is no need for the spatial coordinates to be in a separate file. I have had them separate and together, and it makes no difference. 

My code typically looks like this:

rda

I really appreciate the tip, thank you! Comparing your code to mine it appears I did not define the condition variables appropriately. This should help!

Hi Vivek! I got everything to work now so I can send you an example of the input files if you give me your email address.

Dear Conny,

I'm also working with SSR and I'm trying to analyse my data using RDA. Can you also share your input file for genetic and geographical data with me please? My email is [email protected]

Thank you

Dear Conny,

Could you please send me the genetic data input file for the RDA?

my email address: [email protected]

Thanks

Hi Conny, I am trying to do something similar. Could you kindly share your inputfiles (genetic and geographic data?

Thanks in advance Conny Sjöqvist

Hello Coony,

Could you kindly share your inputfiles genetic and geographic data for RDA ?

Thanks a lot,

Monika