2.4 Experimental methodology
2.4.1 Plant materials
2.4.1.1
PROCEDURES TO SAMPLE PLANT POPULATIONS
The same procedure as for creating a working or core collection (Harlan, 1972 loc.
cit. Spagno1etti-Zeuli and Qualset, 1993; Frankel and Brown, 1984
1oc. cit. Spagno1etti-Zeuli
and Qualset, 1993) was applied when a representative sample of plants was necessary to
characterise populations for genetical studies. Details are presented in the corresponding
Material and Methods sections in this thesis and a brief review of different methodologies is
presented here. The objective is to minimise the cost of germplasm conservation while
ensuring
maximumgenetic diversity containing most of the alleles present in the whole
collection (Crossa et al. , 1993: Holbrook et al. , 1993; Spagno1etti-Zeuli and Qualset, 1993;
Diwan et al. , 1994).
To select a core collection from the U.S. Germplasm Collection of peanut (Arachis
hypogaea L.) consisting of 7432 accessions, two methods were used (Holbrook et al. , 1993).
When information was poor or unavailable, a random sample of 10% of the accessions was
chosen. When the information was available, the data was sorted by country of origin and
then the cluster procedure (SAS Institute, 1988) was used to sort the data and 10% of the
accessions in each cluster were randomly selected. The means and ranges for the six variables
considered were very similar for the entire collection and the core collection (Holbrook et al. ,
1993).
Diwan et al. (1994) obtained a core collection for the United States annual Medicago
Germplasm collection containing 3 159 accessions from 36 species. A SAS macro (Jacobs,
1990 loCo cit. Diwan et al. , 1994) calculated a distance matrix for each of
theMedicago
species based on Euclidean distances between all 1 4 traits, to conduct cluster analysis using
an unweighted pair group method with arithmetic averages. Euclidean distance of 3.0 was
Literature Review
4 4
used to obtain the desired core collection size ( 15% of the accessions). Means, variances and
ranges of each trait were compared between the core collection and the main collection using
a Wilcoxon rank-sum non-parametric test (SAS Institute, 1988). Differences between means
of three or fewer traits were found significant for eight of the 36 species. Only two species
were significantly different for each trait for variances and ranges.
Crossa et
al.(1993) suggested that a useful strategy for fonning a core collection
would be to use a stratified sampling strategy subdividing the accessions into non-overlapping
groups based on ecogeographical criteria. Classification techniques such as cluster analysis
and ordination methods such as principal components analysis have proved to be useful for
assessing genetic diversity.
Spagnoletti-Zeuli and Qualset ( 1993) evaluated five strategies for obtaining a core
collection of 500 accessions from a collection of 3000
accessions of durum wheat (Triticum
turgidum L. d
urumgroup). The strategies were ( 1) random-sampling without replacement;
(2) random-systematic by chronology - sample every fourth accession in the order in which
the accessions were accepted by the gene
bank;(3) random-stratified by geographical origin
and frequency-selecting at random 16% of the accessions of each country; (4) random
stratified by log frequency of accessions by geographical origin - same as (3) but countries
with large number of accessions contributed proportionally fewer accessions to the core
collection and the opposite for countries with few accessions; and (5) random-stratified by
canonical variables - based on the concept that pre-existing information is available; the first
three canonical variables were plotted and about 10% were randomly selected.
The fIrst three strategies produced representative samples, but strategies four and fIve
produced the desired effect of increasing frequencies from less-represented countries of origin
for several traits. The fifth strategy was the best and was effective in increasing the phenotypic
variances in the sample for most characters, due mostly to the increase of the less-frequent
accessions and a decrease in the most-frequent ones, thus flattening the frequency
distnbution.
Literature Review
4 5If necessary information is not available for the :fifth strategy the others are adequate to each level of information available.
2.4.1.2
CLONING PROCEDURE
Identical copies of the same genotype of Red clover plants can be produced by cloning. Such clones can be useful in testing some issues discussed here.
Scerbakova
( 1 936)
compared the rooting and development of cuttings taken fromthe
upper-stem, middle-stem and basal-stem The upper cuttings rooted better than the middle ones, but both regenerated plants with only a single stem Basal cuttings had the poorest rooting but they regenerated plants with normal stem numbers. Cuttings from
the
crown were also tried, and they showed the best rooting. Rooting of allthe
cuttings was better in sand(83.33%)
than in soil(66.26%)
or in water(48.33%).
Hanson
( 1 950)
found that stage of growth was important and that cuttings from actively growing vegetative plants were easier to root than cuttings from plants in the reproductive stage. He also found that temperatures between20°C
and30°C
were optimum for rooting. The influence of the length of the internode below the last node was also studied by the author, who concluded that internode length should be shorter than1 .5
cm to produce best results.Barrales and Ludwig