There are several methods of studying genetic diversity. These include morphometric, biochemical and molecular methods. For this manual, only the morphometric method will be discussed as procedures for use of other methods are yet to be standardized.
Studying genetic diversity in situ and the exploration and survey of new sources of germplasm are two of the four main activity areas arising from the principal objectives of the International Coconut Genetic Resources Network (COGENT). These activities include taking palm and nut samples within the identified sampling sites following the International Plant Genetic Resources Institute's (IPGRI) Coarse Grid Sampling Strategy (Chapter 10). The most ideal practice is for the identified populations to be observed for six months to one year in situ following a regular schedule. In situ work requires the observation of the populations based on an individual palm performance. Hence, all the observed sample palms could also serve as source palms for collecting seednuts for the genebank. The latter is described in Chapter 3.
Guide to sampling
An accurate description of the populations in situ is crucial because this will be the only guide for their use in breeding programmes in the intervening period from the time the surveyed populations are collected to over a decade of complete evaluation and characterization. This description involves two aspects:
(a) Identification of the populations in accordance with the IBPGR list of descriptors as discussed in Chapter 11, andDepending on field situation, it is necessary to consider the following sampling techniques:
(b) Characterization of a certain number of plant and fruit components. These measurements should be taken from a random sample of 30 normal palms (discarding diseased or atypical palms) in the chosen site. Wherever and whenever possible, one ripe nut should be analyzed per studied palm. Otherwise, 30 nuts are taken from the heap.
Random palm sampling - This is the most ideal technique. The collector chooses at random 30 sample palms and obtains one representative nut per palm.Characters to be observed
Random heap sampling - This is done when nuts from the sampled palms are already harvested and heaped or piled at the time of the visit. Sample nuts are picked at random from the heap. Only undamaged, fully ripened ungerminated nuts are taken.
The following list of parameters is not exhaustive and should be seen as a basis for minimum comparison. Details of procedure are found in Chapter 7.
(a) Stem morphology (figs. 1 & 2)
· Girth measurement at 20 cm above soil level (cm)(b) Overall appearance/shape of crown (fig. 3)
· Girth measurement at 1.5 m height (cm)
· Length (m) of stem with 11 leaf scars, measured starting from the bottom of the first leaf scar to the bottom of the 11th leaf scar
1 Spherical(c) Leaf morphology (figs. 4 & 5)
3 X-shaped 'silhouette'
5 Other (specify)
The observation is normally made on leaf #14, which is the leaf subtending the bunch with fist-size nuts. For practical reasons, however, in old coconut palms and if age is not precisely known, take the oldest fully mature green frond by detaching the entire leaf (petiole included) from the stem using a sharp machete or bolo. The following data should be gathered from the sample leaf:
· Colour of petiole
1 Yellow· Petiole length (cm) - from base to the most proximal leaflet
11 Other (specify)
· Petiole thickness (cm) - measure at insertion of first leaflet
· Petiole width (cm) - measure as above
· Rachis length (cm) - from the base of the petiole to the tip
· Number of leaflets - count on one side of the frond that has the first leaflet closest to the base
· Leaflet length (cm) - use four leaflets (two on each side) near the middle of the rachis and record average of four measurements
· Leaflet width (mm) - use the same leaflets as above and record average (at maximum width) of four measurements
(d) Inflorescence and flower morphology (fig. 6)
Preferred samples are inflorescences with male flowers open (may be used for pollen collection if necessary), one inflorescence per palm.
1 Normal· Overlapping of male and female phases:
2 Spicata (full or partial)
4 Additional spathes or bracts
5 Other (specify)
- Presence of receptive female flowers on palm· Length of peduncle (cm) - distance between the point where the bunch is attached to the palm and the base of the first spikelet
- Presence of open male flowers on palm
- If both are present, are they in the same inflorescence?
· Length of the central axis (cm) - measure from the first spikelet to the end of the axis
· Diameter of the peduncle (cm) - at the insertion of first spikelet
· Number of spikelets with female flowers
· Number of spikelets without female flowers
· Length of first spikelet bearing female flower
· Total number of spikelets
· Number of female flowers can be counted from scars if flowers have already been shed
· Female flower distribution per spikelet: number of female flowers divided by total of number of spikelets
(e) Fruit appearance
All fruits analyzed should be mature, i.e. when changing from fresh to dry fruit and water sloshes when fruit is shaken.
· Fruit set (visually estimate number of fruits bigger than a fist in each sample palm)
1 0 to 10· Fruit colour (less than 6 months old) - follow colour code description for petiole
2 11 to 20
3 21 to 50
4 51 to 80
5 81 and above
· Shape of fruit (polar view, figs. 7a & 7b)
1 Round· Shape of fruit (equatorial view, fig. 8)
5 Other (specify)
1 Round· Appearance/shape of husked nut (fig. 9)
4 Other (specify)
1 Flat(f) Fruit component analysis (FCA)
4 Almost round
· Fruit weight (g)(g) Endosperm
· Nut weight (g)
· Weight of split nut (g)
· Shell weight (g)
· Meat weight (g) - difference of shell weight and the weight of the split nut
Measure (mm) the endosperm thickness, at the equatorial portion of 10 mature fruits (more or less 12 months old) and take the mean. Do not include makapuno (gel-endosperm) types.
Statistical analysis of the palm and nut/fruit characters observed in the survey situation is complicated because the characters observed may have been influenced by the environment. Hence, any information available on the soil and climatic conditions and the probable origin of the populations has to be taken into account when interpreting these results.
It should be borne in mind that the characters to be studied are those that are highly heritable. In some cases it may be even better to consider the ratio between two measurements, as in number of nuts, copra per nut and copra per palm, rather than the basic data. Studies conducted within the experimental station could provide clearer answers to these questions.
Simple analysis of variance (ANOVA) is sufficient for determining coefficient of variance when dealing with effect of treatments (e.g. cultivars) on individual parameters (e.g. yield). However, to compare several parameters or characters, a multivariate analysis is required.
There are several methods of conducting multivariate data analysis. The choice of a method depends on the problem to be worked out. If the problem is to represent a set of individuals without preliminary grouping, principal component analysis can be used. The purpose of this technique is to condense the information given by the original variables into a smaller set of new independent variables.
Comparison of populations can be made with homogeneity tests. Hotelling's T2 test, for instance, can be used to test the equality of the means of two populations with several variables. For more than two populations, analysis of dispersion (multiple analysis of variance) can be used. Discriminant analysis can be used to plot points representative of the populations on main axes taking into account within-population variability, and to compute Mahalanobis distances between populations. Mahalanobis generalized distance between two populations is presented in Appendix 2.
Representation of distances between individuals can be defined with classification methods. Dendrograms built with clustering techniques can help to define groups of individuals. Additive trees can be interpreted as phylogenetic representations to relations between individuals. All these methods can be applied to populations as well as individuals.
Another problem is to allocate individuals to groups previously defined. In this case, discriminant analysis or Fisher's discriminant functions can be used.
A biometrician is particularly helpful in analyzing quality data (colour, fruit shape, etc.) or discontinuous variables (e.g. fruit set) which require special coding. The use of a data management system software to facilitate analysis should be explored.