[vsni.co.uk] Contact us
Author Message
Post new topic

  ASReml  ~  bivariate spatial analysis

Christine Hansen
Posted: Fri Dec 10, 2010 7:05 pm Reply with quote
Joined: 05 Aug 2009 Posts: 6
I am analysing height and diameter in a lodgepole pine MET on two sites; individual site univariate analyses indicate that a spatial approach provides a better fit for the data than analyses using experimental design effects only. Is it possible to combine a bivariate and spatial analysis in ASReml?
View user's profile Send private message Send e-mail
Arthur
Posted: Sun Dec 12, 2010 11:16 pm Reply with quote
Joined: 05 Aug 2008 Posts: 471 Location: Orange, NSW
Dear Christine, it is not easy to define a completely general model for a bivariate spatial model, but if the spatial correlations have the same sign across traits,
then AR1(rows) x AR1(cols) x US
will be a better model than ignoring the spatial correlation, since it will use an average spatial correlation (whyich is not zero).

My personal seasonal greetings are on
www.cargovale.com.au/Christmas2010/Christmas2010.html

May God bless you all through His grace

Arthur

_________________
Arthur Gilmour

Retired Principal Research Scientist (Biometrics)
View user's profile Send private message Send e-mail Visit poster's website
Christine Hansen
Posted: Mon Dec 13, 2010 6:54 pm Reply with quote
Joined: 05 Aug 2009 Posts: 6
Dear Arthur,
Thank you for replying to my question. I had thought that the AR1 x AR1 x US would be the way to do the analysis but I am at a loss as to how to code it in ASReml. This is what I have tried:

!ASUV
h14 d14 ~ Trait !r Trait.rep Trait.type.acc Trait.rep.acc !f mv
2 2 0
row row AR 0.5
col col AR 0.5
0 0 IDEN
Trait 0 US
3*0

The .asr file says that the error structures are the wrong size.
View user's profile Send private message Send e-mail
Arthur
Posted: Tue Dec 14, 2010 10:27 am Reply with quote
Joined: 05 Aug 2008 Posts: 471 Location: Orange, NSW
Dear Christine,

If both trials are the same spatial dimensions,
and the data is ordered columns within rows,
analysing just one site

!ASUV
h14 d14 ~ Trait !r Trait.rep Trait.type.acc Trait.rep.acc !f mv
1 3 0
row 0 AR 0.5
col 0 AR 0.5
Trait 0 US
3*0

should work *maybe it needs starting values for US
obtained by running

1 2
0
Trait 0 US
3*0

first.

Then when you have results from both sites, you could combine with

!ASUV
h14 d14 ~ Trait Trait.site !r at(site).Trait.rep at(site).Trait.type.acc at(site).Trait.rep.acc !f mv
2 3 0
row 0 AR 0.5
col 0 AR 0.5
Trait 0 US
3*0 Trait 0 US
3*0
row 0 AR 0.5
col 0 AR 0.5
Trait 0 US
3*0 Trait 0 US
3*0

adapted to the specific site dimensions.

G structures need adding since you need trait and site specific variances for
the random components

I trust you can take it from here.

_________________
Arthur Gilmour

Retired Principal Research Scientist (Biometrics)
View user's profile Send private message Send e-mail Visit poster's website

Display posts from previous:  

All times are GMT
Page 1 of 1
Post new topic

Jump to:  

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You can attach files in this forum
You can download files in this forum