Dear Ecologgers,

I would like to ask for advice concerning the design and statistical 
analysis of an experiment. My problem is described below.

I have made crosses by applying the same mixture of pollen from two pollen 
donors (provenance 1 and 2) on one mother plant of each of the two same 
provenances (a mother couple), with a total of 16 mother plants (8 per 
provenance). I now want to investigate the performance of the resulting 
offspring (hybrids, pure provenance 1, pure provenance 2). A difficulty is 
that the genotypes are unknown at the time of the start of the experiment, 
and will only be known after a genetic analysis of the leaf material. I 
will use two experimental conditions: wet and dry. For practical reasons, 
it is easier to vary the water factor for the whole germination trays. I’ve 
got 6 experimental tables that can hold each a maximum of 3 germination 
trays with 104 seeds each. Considering the tables as replicates, I will 
assign 2 trays to each table and have the wet and dry treatments randomly 
assigned to trays within each table.

Thus, the ingredients I have got are:
max. 6 tables [replicates or blocks]
2 water treatments
2 provenances
8 families or mothers (nested in provenance)
2 genotypes [pure and hybrid]

Two possible options are: a) I grow 6 seeds of each of the 16 mother plants 
(completely randomized within tray) in each of 8 trays on 4 tables. Given 
that I don’t know the genotypes, this will lead to unequal sample sizes for 
genotype and, in the worst case, to missing genotypes in certain trays. b) 
I assign all seeds of two mother couples (4 x 24) to one randomly selected 
tray and thus, confound mothers and block.

Given that I’m interested in estimating differences in performance related 
to genotype and provenance (hybrids, pure provenance 1, pure provenance 2), 
but not in variability between families,
1) which design is the more suitable/optimal (confounding/ power) for the 
question I am most concerned about? Or is there still a better one?
2) Do I have a split-split plot design with an additional nested factor? 
What about the statistical evaluation of the results?

Please feel free to email me off list ([EMAIL PROTECTED]).

Sincerely,
Madlen


*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~**~*~*~*~*~*~*~*~*
Madlen Denoth, PhD
Département de Biologie/Ecologie & Evolution
Université de Fribourg/Pérolles
Chemin du Musée 10
1700 Fribourg
Suisse
Tel. ++41 26 300 88 48
email: [EMAIL PROTECTED]


"Too often we enjoy the comfort of opinion without
the discomfort of thought. " -- John F. Kennedy

*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~*~**~*~*~*~*~*~*~*~*

Reply via email to