The problem is that you haven't read the output from R! When you print 
'RWA.all', one of the things it prints is:

> Some constraints were aliased because they were collinear (redundant)

This means that some of the predictor variables in the data set are fully 
explained by one or more combinations of the the predictor variables. It is not 
possible to include them in the analysis hence only 9 RWA axes were fitted and 
hence why not all variables show up in the step output.

HTH

Gavin

Sent from my HTC

----- Reply message -----
From: "carolina monmany" <[email protected]>
Date: Sun, Feb 5, 2012 17:10
Subject: [R-sig-eco] vegan's ordistep function
To: "[email protected]" <[email protected]>

Hi all,

Another question on ordistep() in vegan.  I am running a RDA, have 16
explanatory variables and five of them were not even listed in the Biplot
scores for constraining variables. When I analyzed the VIF, it also appeard
a NA in their place. This happened also when data were first standardized.

Now I want to choose the best model using ordistep().
Here, the five explanatory variables that I mentioned before are included
(the last five in the formula).

>RDA.all <- rda(main ~ ., data=second)
> RDA.all
> Call: rda(formula = main ~ MeanAsp + SDAsp + MenaSlo + SDSlo + Urban +
> Treelines + TreeGroups +
> SolTress + IntermGrass + TallGrass + Highways + Canals + DistDCA +
> MeanTrackDens + SDTrackDens +
> MeanMagnetism, data = second)
>
>                Inertia Proportion Rank
> Total         0.005999   1.000000
> Constrained   0.005999   1.000000    9
> Unconstrained 0.000000   0.000000    0
> Inertia is variance
> Some constraints were aliased because they were collinear (redundant)
>
> Eigenvalues for constrained axes:
>      RDA1      RDA2      RDA3      RDA4      RDA5      RDA6      RDA7
>  RDA8      RDA9
> 4.421e-03 1.406e-03 8.930e-05 5.354e-05 1.413e-05 8.836e-06 4.612e-06
> 1.324e-06 3.500e-07

# Here's the problem

> step.forward <- ordistep(rda(main ~ 1, data=second), RDA.all,
direction="forward", pstep=1000)

Start: main ~ 1

>

The process stops in less than a second and it just does not give any
output. So I asked for:

> step.forward$anova
NULL

# Why null?
# And then,

>summary(step.forward)

Call:
rda(formula = main ~ 1, data = second)

Partitioning of variance:
               Inertia Proportion
Total         0.005999          1
Unconstrained 0.005999          1

Eigenvalues, and their contribution to the variance

Importance of components:
                           PC1      PC2       PC3       PC4       PC5
PC6       PC7       PC8     PC9
Eigenvalue            0.004421 0.001406 0.0000893 5.354e-05 1.413e-05
8.836e-06 4.612e-06 1.324e-06 3.5e-07
Proportion Explained  0.736960 0.234350 0.0148800 8.920e-03 2.360e-03
1.470e-03 7.700e-04 2.200e-04 6.0e-05
Cumulative Proportion 0.736960 0.971320 0.9862000 9.951e-01 9.975e-01
9.990e-01 9.997e-01 9.999e-01 1.0e+00

Scaling 2 for species and site scores
* Species are scaled proportional to eigenvalues
* Sites are unscaled: weighted dispersion equal on all dimensions
* General scaling constant of scores:  0.5068435


Species scores

              PC1        PC2        PC3        PC4        PC5        PC6
T30      0.020230  0.0260141  1.640e-03  0.0009593  0.0016298  1.494e-03
T60      0.029968 -0.0416358  5.082e-02 -0.0033992 -0.0033624  7.331e-03
T90      0.040460  0.0520283  3.280e-03  0.0019185  0.0032597  2.988e-03
T120     0.138776  0.1828691  1.381e-02 -0.0063556  0.0100248  6.255e-06
T150     0.083839  0.0991670 -1.549e-03  0.0246916 -0.0144196 -4.801e-03
T180     0.017788 -0.0189122  4.420e-03 -0.0192773  0.0054548 -1.322e-02
T210     0.032552  0.0531892 -1.255e-02 -0.0353441 -0.0138663  3.979e-03
T240    -0.002133  0.0006171 -2.533e-06 -0.0001967  0.0002564 -3.974e-04
X.1dia   0.398124 -0.0909282 -8.897e-03  0.0008550  0.0005249  6.968e-04
X.2dias  0.014241 -0.0188583  2.790e-02 -0.0021515 -0.0070527 -9.956e-03


Site scores (weighted sums of species scores)

         PC1       PC2        PC3      PC4      PC5      PC6
D1N -0.02813  0.012849 -0.0937696 -0.32021  0.04749 -0.19677
D1S  0.29325 -0.300213  0.1317071 -0.08004  0.05188 -0.07038
D1E  0.29622  0.380916  0.0240131  0.01405  0.02387  0.02188
D1W -0.08521  0.011027  0.3446805  0.05417 -0.18557 -0.09601
D2N  0.06379 -0.071989 -0.2057112  0.28033 -0.24678 -0.12687
D2S -0.02582 -0.053152  0.1201434  0.04442  0.05756  0.35846
D2E -0.13478  0.039000 -0.0001601 -0.01243  0.01620 -0.02512
D2W -0.05436 -0.004211 -0.1337934 -0.22111 -0.28460  0.19936
D4N  0.04743 -0.095968 -0.1891478  0.08302  0.18306  0.09654
D4S -0.10152  0.011007 -0.0243409  0.05661  0.12162 -0.03389
D4E -0.13543  0.035367  0.0131894  0.05059  0.10763 -0.06359
D4W -0.13543  0.035367  0.0131894  0.05059  0.10763 -0.06359

I would appreciate any help in understanding why I am not given an output.
I also tried turning OFF the "Buffered output" option in Menu/Misc in
Windows.

Thanks!

Carolina

2012/2/5 carolina monmany <[email protected]>

> That was the mistake. Now it worked. Thanks a lot, Roman and Olmo and
> thanks for the recommendation on the code editor!
>
> Carolina
>
>
> 2012/2/5 Olmo <[email protected]>
>
>>  Hi Carolina, it could be a syntaxis ommition indeed, I see a space
>>  that I think it must not be there, between < and the minus - ,
>>  (probably = is simpler but I use <- too):
>>
>> > step.forward < - ordistep(rda(main ~ 1, data=second),
>>  scope=formula(RDA.all), direction="forward", pstep=1000)
>>
>> > step.forward <- ordistep(rda(main ~ 1, data=second),
>>  scope=formula(RDA.all), direction="forward", pstep=1000)
>>
>>  Check that, just in case.
>>
>> Hope this helps,
>> Olmo.
>>
>>
>> El Sat, 4 Feb 2012 15:03:08 -0400
>> carolina monmany <[email protected]> escribió:
>> > Dear all,
>> >
>> > I am running a redundancy analysis in vegan and I am using ordistep to
>> > perform a forward selection of variables in my model. I'm following
>> > Borcard et al. (2011) for this. I got stuck in one of the steps and
>> > couldn't find out why.
>> >
>> > > RDA.all <- rda(main ~ ., data=second)
>> > > RDA.all
>> > Call: rda(formula = main ~ MeanAsp + SDAsp + MenaSlo + SDSlo + Urban +
>> > Treelines + TreeGroups +
>> > SolTress + IntermGrass + TallGrass + Highways + Canals + DistDCA +
>> > MeanTrackDens + SDTrackDens +
>> > MeanMagnetism, data = second)
>> >
>> >                Inertia Proportion Rank
>> > Total         0.005999   1.000000
>> > Constrained   0.005999   1.000000    9
>> > Unconstrained 0.000000   0.000000    0
>> > Inertia is variance
>> > Some constraints were aliased because they were collinear (redundant)
>> >
>> > Eigenvalues for constrained axes:
>> >      RDA1      RDA2      RDA3      RDA4      RDA5      RDA6      RDA7
>> >  RDA8      RDA9
>> > 4.421e-03 1.406e-03 8.930e-05 5.354e-05 1.413e-05 8.836e-06 4.612e-06
>> > 1.324e-06 3.500e-07
>> >
>> > # And here's the problem,
>> >
>> > > step.forward < - ordistep(rda(main ~ 1, data=second),
>> > scope=formula(RDA.all), direction="forward", pstep=1000)
>> > Error: object 'step.forward' not found
>> >
>> > I hope it's a syntaxis ommition or error. Any clue?
>> >
>> > Thanks a lot,
>> >
>> > Carolina
>> >
>>
>>
>> _______________________________________________
>> R-sig-ecology mailing list
>> [email protected]
>> https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
>>
>
>
>
> --
> ---------------------------
> A. CAROLINA MONMANY
> Universidad de Puerto Rico
> Departamento de Biologia - CN 235
> POBOX 23360
> San Juan, Puerto Rico 00931-3360
> Tel: +1 787 764 0000 x2847
> Fax: +1 787 764 2610
>



--
---------------------------
A. CAROLINA MONMANY
Universidad de Puerto Rico
Departamento de Biologia - CN 235
POBOX 23360
San Juan, Puerto Rico 00931-3360
Tel: +1 787 764 0000 x2847
Fax: +1 787 764 2610

        [[alternative HTML version deleted]]


        [[alternative HTML version deleted]]

_______________________________________________
R-sig-ecology mailing list
[email protected]
https://stat.ethz.ch/mailman/listinfo/r-sig-ecology

Reply via email to