Previous Page Table of Contents Next Page


Characterisation and preliminary evaluation of accessions of Zornia species from the ILCA collection

J.J. Hakiza, J.R. Lazier and A.R. Sayers
ILCA, P.O. Box 5689, Addis Ababa, Ethiopia


Introduction
Methodology
Results
Discussion
Acknowledgements
References

Abstract

The genus, Zornia has a tropical distribution mainly in America and Africa. A collection of Zornia species is used to provide natural grazing and hay during different seasons.

The genus has maintained its taxonomic notoriety. The significance of classification of the species in this genus would possibly be in the application of various groupings to agronomic problems, particularly selection and/or breeding for specific purposes.

As a guide to the understanding of the range of variation available in the Zornia accessions at ILCA, a study was made on the plants already growing at Soddo, Ethiopia, based on morphological and agronomic characters. Analysis of the collected data revealed that with the use of morphological characters, it was possible to obtain groupings belonging to particular species as exemplified by Zornia setosa, Z. glochidiata, Z. glabra and Z. latifolia. Based on agronomic characters a different pattern was obtained, but this narrowed the collection into two major groups. One group was characterised by prostrate to decumbent accessions while the other group comprised accessions which ranged from decumbent to erect in their growth forms. Other smaller groups with particular agronomic attributes could also be identified.

It seemed imperative, however, that further studies involving more attributes than those used here would provide more systematic groupings of accessions with succinct characteristics.

Introduction

The genus Zornia is indigenous to savannas and similar areas in tropical and subtropical regions of the world (Mohlenbrock, 1961: McIvor and Bray, 1983). The number of species in this genus is variable but commonly estimated to be around 80 and distributed predominantly in America and Africa.

Some Zornia species (Bogdan, 1977) are of importance in providing natural grazing and hay, especially in Brazil and the neighbouring countries and also in the drier parts of West Africa. Species like Z. glochidiata (synonym Z. diphylla) provides some scant forage during the wet season and early in the season but does not survive long into the dry season (Skerman, 1977).

The genebank at the International Livestock Centre for Africa (ILCA) currently holds at least 234 accessions of Zornia species some of which were grown for seed production and agronomic evaluation during mid-1985 at Soddo in Sidamo, Ethiopia. In order to provide some quick information about these collections, a study of these accessions was superimposed on the existing evaluation programme. Using 22 attributes on 161 accessions of Zornia species data were collected on field-grown plants that had been growing for just over one year. The 22 attributes comprised both morphological and agronomic characters (Table 1) whose choice was partially based on taxonomic character descriptions of Zornia species (Mohlenbrock, 1961) and some of the forage evaluation characters used at ILCA. The purpose of this work was to try and group these accessions on the basis of their similarities and dissimilarities using both morphological and agronomic characters. This was to assist in picking out closely related accessions and also to obtain those accessions of good agronomic characters that could be subjected to further evaluation while awaiting the results from the ongoing evaluation trial.

The collection data were classified using principal component analysis (PCA), cluster analysis (CA) and K-means clustering.

Table 1. Codes for the characters (descriptors and descriptor states) recorded on the Zornia sp. growing at Soddo, Ethiopia.

1.

Number of leaflets per leaf: 1 = 2 leaflets; 2 = 4 leaflets

2.

Flower size: 1 = flowers shorter than the bracts; 2 = flowers longer than or equal to the bracts

3.

Shape of the leaflets of the lowest leaves: 1 = linear; 2 = filiform; 3 = lanceolate; 4 = ovate; 5 = elliptic; 6 = elliptic/lanceolate; 7 = oblanceolate; 8 = obovate

4.

Size of the lowest leaflets: 1 = leaflets more than twice as long as broad. 2 = leaflets not more than twice as long as broad.

5.

Shape of upper leaflets: 1 = linear; 2 = filiform; 3 = lanceolate; 4 = ovate; 5 = elliptic; 6 = elliptic/lanceolate; 7 = oblanceolate; 8 = obovate

6.

Flower colour: 1 = corolla yellow with no red marks 2 = corolla yellow with few red marks 3 = corolla yellow with many red marks 4 = corolla red

7.

Flower (corolla) size: 1 = very small; 2 = small; 3 = small-medium; 4 = medium; 5 = medium-large; 6 = large.

8.

Plant growth form: 1 = creeping; 2 = prostrate; 3 = semi-prostrate; 4 = decumbent; 5 = semi-erect; 6 = semi-erect/erect; 7 = erect

9.

Bract and leaf pubescence: 1 = glabrous; 2 = pilose

10.

Colour of the bracts: 1 = green; 2 = green with yellow marks; 3 = green with red marks; 4 = purple

11.

Internode length: 1 = very short; 2 = short; 3 = medium; 4 = long

12.

Internode thickness: 1 = very slender; 2 = slender; 3 = medium; 4 = thick

13.

Inflorescence length of lowest inflorescence: 1 = very short; 2 = short; 3 = medium; 4 = long

14.

Spread of the bracts on the inflorescence: 1 = very sparse; 2 = sparse; 3 = medium; 4 = large

15.

Bract size: 1 = very small; 2 = small; 3 = medium; 4 = large

16.

Leafiness: 1 = negligible; 2 = very poor; 3 = poor; 4 = fair; 5 = good; 6 = very good

17.

Plot cover: 1 = negligible; 2 = very poor; 3 = poor; 4 = fair; 5 = good; 6 = very good

18.

Inflorescence density: 1 = negligible; 2 = very poor; 3 = poor; 4 = fair; 5 = high; 6 = very high

19.

Branching intensity: 1 = negligible; 2 = very poor; 3 = poor; 4 = fair; 5 = high; 6 = very high

20.

Vigour = 1 = negligible; 2 = very poor; 3 = poor; 4 = fair; 5 = good; 6 = very good

21.

Productivity: 1 = negligible; 2 = very low; 3 = low; 4 = medium; 5 = high; 6 = very high

22.

Adaptability: 1 = negligible; 2 = very poor; 3 = poor; 4 = fair; 5 = good; 6 = very good

Methodology

A total of 161 Zornia accessions which were growing at Soddo in Sidamo Province of Ethiopia were observed for characterisation and preliminary evaluation. Soddo is at an elevation of 1850 m with acid Nitosols of pH 5.5 to 6.0. Rainfall is above 700 mm per year while the minimum and maximum temperatures are 12.6°C and 24.5°C respectively. The total length of the growing season is 245 days.

Each accession has been planted on unreplicated raised plots each containing six plants. The plot size was 2 m by 1.5 m with 75 cm pathways across the slope and 1 m pathways or drains down the slope.

Twenty-two characters - morphological and agronomic - that could be recorded at the time were selected based on their taxonomic applicability. Each of the 22 characters was recorded once for each accession.

An HP 3000 computer and BMDP software were used for analysis of the collected data. For PCA the BMDP4M program was employed while the CA used the BMDP2M program to produce the vertical and horizontal dendrograms. The K-means clustering used BMDPKM program (Engelman, 1980; Frane and Hill, 1974).

Only components with Eigenvalues greater than unity were retained in the principal component analysis, and they were rotated by the Varimax procedure to aid interpretation.

The BMDPKM program for K-means clustering used the hierarchical agglomerative method (Engelman, 1980). All data were standardised before clustering. The BMDPKM program iteratively reallocates cases into clusters until each case is in the cluster whose centre is closest to the case. This procedure is followed for each number of clusters specified. In this study the specified clusters were 8, 10 and 12. The criterion of clustering was to minimise the maximum distance of a case from the centre of the cluster to which the case belonged. The word "case" here is synonymous to "accession".

In all the above analyses the characters were considered under morphological, agronomic and morphological-agronomic categories.

Results

Principal Component Analysis

Before carrying out principal component analysis it was necessary to consider the correlation matrix for the characters under study. There existed a number of high correlations among the characters which confirmed the justification for carrying out principal component analysis on the data (Tables 2, 3 and 4). The level of significance at 5% and 1% were respectively >0.15 and >0.211 for morphological characters; >0.156 and >0.204 for agronomic characters; and >0.214 for morphological-agronomic (all) characters.

On application of the principal component analysis to the data, four factors (components) were found to have eigenvalues greater than unit for morphological characters, three for agronomic characters and four for morphological-agronomic characters. The four factors explained 71.6% of the total variance for morphological characters with the eigenvalues of 3.249, 2.421, 1.658 and 1.373 respectively. Three factors explained 84.97% of the total variance for agronomic characters with their respective eigenvalues of 2.897, 2.606 and 1.295. For the combined morphological-agronomic characters the four factors accounted for 72.09% of the total variance with the eigenvalues of 5.045, 1.932, 1.739 and 1.376 respectively. The sorted rotated factor loadings for morphological, agronomic and morphological-agronomic characters are respectively presented in Tables 5, 6 and 7.

Table 2. Correlation matrix for morphological characters in Zornia sp.


NLEAFPLF

FLOWERSZ

SHAPELL

SIZELL

FLOWERCO

FLOWERSZ

BRACTSCO

ININDEL

ININDETH

INFLLENG

SPBRINFL

BRACTSZE

3

4

5

7

8

9

12

13

14

15

16

17

NLEAFPLF

3

1.000












FLOWERSZ

4

0.064

1.000











SHAPELL

5

0.742**

0.052

1.000










SIZELL

7

0.661**

0.001

0.765**

1.000









FLOWERCO

8

-0.699**

-0.093

-0.393**

-0.388**

1.000








FLOWERSZ

9

-0.096

0.126

-0.197*

-0.184*

-0.245**

1.000







BRACTSCO

12

-0.486**

-0.048

-0.431**

-0.461**

0.277**

0.259**

1.000






ININDEL

13

-0.266**

0.154

-0.096

-0.063

0.155

-0.018

0.109

1.000





ININDETH

14

-0.180**

0.090

-0.090

-0.002

-0.033

0.302**

-0.009

0.302**

1.000




INFLLENG

15

-0.306**

-0.173*

-0.046

-0.064

0.308**

-0.113

-0.040

0.374**

0.142

1.000



SPBRINFL

16

-0.202**

-0.150

-0.315**

-0.317**

-0.079

0.514**

0.257**

-0.147

0.023

-0.016

1.000


BRACTSZE

17

0.146

-0.286**

-0.006

-0.011

-0.310**

0.533**

0.074

-0.291**

0.101

-0.045

0.643*

1.000

Level of significance:

5% : >0.161 (*)
1 % : >0.211 (**)

Table 3. Correlation matrix for agronomic characters in Zornia sp.

 

PLGRFORM

LEAFNESS

PLOTCOVR

INFLDENS

BRANDENS

VIGOUR

PRODUCTV

ADAPTBTY

10

18

19

20

21

22

23

24

PLGRFORM

10

1.000








LEAFNESS

18

-.279**

1.000







PLOTCOVR

19

.140

.453**

1.000






INFLDENS

10

.319**

.086

.659**

1.000





BRANDENS

21

.285**

.264**

.746**

.798**

1.000




VIGOUR

22

.202**

.354**

.587**

.458**

.501**

1.000



PRODUCTV

23

.259**

.403**

.785**

.676**

.690**

.701**

1.000


ADAPTBTY

24

.309**

.397**

.767**

.655**

.700**

.724**

.940**

1.000

Level of significance:

5% : >0.156 (*)
1% : >0.204(**)

Table 4. Correlation matrix for morphological-agronomic characters in Zornia sp.

 

FLOWERSZ

SHAPELL

SIZEUL

FLOWERSZ

PLGRFORM

BRACTSCO

BRACTSZE

LEAFNESS

PLOTCOVR

INFLDENS

BRANDENS

VIGOUR

PRODCTV

ADAPTBTY

4

5

7

9

10

12

17

18

19

20

21

22

23

24

FLOWERSZ

4

1.000














SHAPELL

5

.069

1.000













SIZEUL

7

-.007

.551**

1.000












FLOWERSZ

9

.154

-.182*

-.160

1.000











PLGRFORM

10

.233**

-.054

-.168*

.257**

1.000










BRACTSCO

12

.025

-.114

-.212*

.281**

.421**

1.000









BRACTSZE

17

-.399**

-.210*

-.177*

.587

.057

.181*

1.000








LEAFNESS

18

.022

-.095

-.039

.183*

-.314**

-.191*

.196*

1.000







PLOTCOVR

19

.098

-.210*

-.269**

.325**

.070

.054

1.450**

.517**

1.000






INFLDENS

20

-.132

-.249**

-.321**

.325**

.275**

.232**

.500**

.052

.603**

1.000





BRANDENS

21

-.023

-.275**

-.371**

.308**

.232**

.173*

.408**

.346**

.720**

.757**

1.000




VIGOUR

22

.180*

-.103

-.140

.501**

.189*

.326**

.402**

.359**

.587**

.454**

.478**

1.000



PRODUCTV

23

-.007

-.203*

-.293**

.398**

.181*

.109

.538**

.466**

.792**

.654**

.681**

.684**

1.000


ADAPTBTY

24

.032

-.267**

-.297**

.407**

.215**

.153

.496**

.460**

.772**

.619**

.673**

.713**

.933**

1.000

Level of significance

5%: >0.164(*)
1 %: >0.214 (**)

In Tables 5, 6 and 7 the factor loading matrix has been rearranged 60 that the columns appear in decreasing order or variance explained by factors. The rows have been rearranged so that for each successive factor, loadings greater than .5000 appear first. Loadings less than. 2500 have been replaced by zero.

The interpretations attached to these factors for the morphological characters are that the positive end of factor one is characterised by attributes responsible for the higher number of leaflets per leaf, obovate in shape, yellow flowers with few or no red marks and green bracts. Factor two represents characters associated with large bracts and corollas, the corollas being yellow in colour. The inflorescences are compact having very many flowers and bracts. The third factor expresses thick and long internodes and long inflorescences. The fourth factor is associated with characters responsible for the flowers that are longer than the bracts.

For the agronomic characters the positive end of factor one is associated with characters responsible for bushy plants which are highly productive and adaptable while factor two indicates plants which are vigorous and leafy with high productivity. Factor three is associated with erect plants having very few leaves.

Table 5. Sorted rotated factor loadings for morphological characters in Zornia sp.



Factor 1

Factor 2

Factor 3

Factor 4

NLEAFPLF

3

0.878

0.000

-0.273

0.000

SHAPELL

5

0.864

0.000

0.000

0.000

SIZELL

7

0.858

0.000

0.000

0.000

BRACTISCO

12

-0.663

0.000

0.000

0.000

FLOWERCO

8

-0.643

-0.422

0.000

-0.325

BRACTSZE

17

0.000

0.878

0.000

0.000

FLOWERSZ

9

0.000

0.789

0.000

0.298

SPBRINFL

16

-0.297

0.782

0.000

0.000

ININDEL

13

0.000

0.000

0.769

0.000

ININDETH

14

0.000

0.261

0.740

0.000

FLOWERSZ

4

0.000

0.000

0.000

0.813

INFLLENG

15

0.000

0.000

0.564

-0.624


VP

3.249

2.421

1.658

1.373

Table 6. Sorted rotated factor loadings for agronomic characters in Zornia sp.



Factor 1

Factor 2

Factor 3

INFLDENS

20

.905

.000

.000

BRANDENS

21

.880

.295

.000

PLOTCOVR

19

.721

.532

.000

VIGOUR

22

.259

.837

.000

ADAPTBTY

24

.570

.751

.000

PRODUCTV

23

.600

.721

.000

PLGRFORM

10

.000

.000

.887

LEAFNESS

18

.000

.607

-.655


VP

2.897

2.606

1.295

Table 7. Sorted rotated factor loadings for morphologicalagronomic characters in Zornia sp.



Factor 1

Factor 2

Factor 3

Factor 4

PRODUCTV

23

.910

.000

.000

.000

ADAPTBTY

24

.900

.000

.000

.000

PLOTCOVR

19

.869

.000

.000

.000

VIGOUR

22

.797

.000

.000

.000

BRANDENS

21

.733

.000

-.364

.000

INFLDENS

20

.632

.295

-.308

-.295

LEAFNESS

18

.613

-.567

.000

.000

FLOWERSZ

9

.538

.428

.000

.000

PLGRFORM

10

.000

.782

.000

.000

BRACTSCO

12

.000

.748

.000

.000

SIZEUL

7

.000

.000

.843

.000

SHAPELL

5

.000

.000

.817

.000

FLOWERSZ

4

.000

.000

.000

.892

BRACTSZE

17

.577

.000

.000

-.618


VP

5.045

1.932

1.739

1.376

Considering the morphological-agronomic characters combined the positive end of factor one has some resemblance to factor one for agronomic characters. It is characterised by high productivity, adaptability, plot cover, vigour, branching intensity, inflorescence density, leafiness and large flowers (corollas) and bracts. Factor two, on the other hand, is mainly characterised by plants having reddish or purple bracts, semi-erect to erect but scanty foliage and flowers of medium size. Factor three is associated with plants having obovate leaflets with low branching intensity and inflorescence density. Factor four is mainly characterised by flowers that are larger than the bracts.

The factor scores of individual accessions on the first four factors were plotted against one another under morphological, agronomic and all characters combined (Figures 1, 2 and 3). Clustering could be identified especially at the ends of the axes with most of the accessions scattered randomly at the point of origin.

Based on the morphological characters four clusters, (A, B. C and D in Figure 1) on the plot of factor 1 versus factor 2 were formed. Cluster A comprised Zornia setosa accessions, cluster B was composed of Z. brasiliensis while cluster C comprised accessions 11411 and 11424. These two accessions have not been identified, but they could possibly belong to Z. lanata. Cluster D was composed mainly of Z. glabra. Accessions belonging to Z. diphylla and Z. pratensis tended to form a continuum, which is not surprising since Z. diphylla is normally applied to a complex of species, both perennial and annual (Bogdan, 1977). This complex of species usually embraces such species as Z. latifolia, Z. gracilis, Z. perforate, Z. pratensis and several others.

Groupings of accessions or clusters based on agronomic characters were almost similar to those based on the combined morphological-agronomic characters (Figures 2 and 3). Other species formed their own groups based on their stature and plant growth forms. Based on the clustering pattern of these accessions and the factor scores responsible for some characters, particular groups with particular characteristics can be picked out for further evaluation.

Cluster Analysis

As in the principal component analysis, the data were categorised into morphological, agronomic and morphological-agronomic characters. Using morphological characters, a minimum of ten clusters was formed with some non-conformist groups which later joined the rest of the clusters at higher amalgamation distances. The amalgamation distances ranged from 0.000 to 10.330 thus producing simple to complex fusions of accessions that produced a very large vertical tree dendrogram. Table 8 shows the composition of various clusters by accessions.

Some characteristic features of some clusters had a definitive role for particular species. Cluster 1, for example, was characterized by having four leaflets per leaf, obovate upper and lower leaflets, small yellow flowers with few or no red marks, very short and slender internodes and very short inflorescences and bracts. This cluster comprised all accessions of Z. setosa and one accession, ILCA 11476, of Z. brasiliensis. Cluster 2 had accessions whose flowers were shorter than the bracts. They had ovate to elliptic leaflets with a few accessions having lanceolate leaflets. They also tended to have medium to long inflorescences, medium to large spread of bracts or inflorescences and medium sized bracts. Accessions in this cluster belonged to Z. glochidiata and Z. orbiculata. Cluster 10, which was the largest, comprising 28 accessions, was characterised by medium to large flowers with inflorescences of medium length, leaflets that were more than twice as long as they were broad and medium-to-large spread of bracts on inflorescences which were either purple in colour or had red marks. Cluster 10 was composed of Z. glabra and Z. latifolia. Other clusters tended to be an assortment of various species, namely Z. diphylla, Z. latifolia, Z. pratensis and other unidentified Zornia spp.

Figure 1. PCA morphological characters.

Scale is from -3 to +3. Factor scores greater than 3 are plotted as 3. Factor scores less than -3 are plotted as -3.

Figure 2. PCA agronomic characters.

Scale is from -3 to +3. Factor scores greater than 3 are plotted as 3. Factor scores less than -3 are plotted as -3.

Figure 3. PCA all characters.

Scale is from -3 to +3. Factor scores greater than 3 are plotted as 3. Factor scores less than -3 plotted as -3.

Table 8. Accessions of Zornia species constituting various clusters in cluster analysis.

Table 8. Accessions of Zornia species constituting various clusters in cluster analysis. - Cont. a

Table 8. Accessions of Zornia species constituting various clusters in cluster analysis. - Cont. b

For clustering based on agronomic and morphological agronomic characters, ten clusters could be identified in each case. However, based on agronomic characters the whole population could be divided into two major units. One group comprised accessions that ranged from creeping to decumbent and were of small stature (clusters 1 to 7). The second group was characterised by being decumbent to erect, of fair to very good leafiness, good to very good branching intensity, vigour, productivity and plot cover. The first group comprised such species as Z. setosa, Z. glochidiata, Z. diphylla and some unidentified species while the second group comprised mainly Z. glabra and Z. latifolia.

K-means Clustering

In the final K-means clustering analysis, divisions into 8, 10 and 12 clusters were specified under morphological, agronomic and morphological-agronomic characters (Appendix 1). This number was arrived at on the basis of the previous observation which indicated that during cluster analysis at least 10 clusters could be identified while in the principal component analysis a minimum of six clusters including the continuum of accessions scattered at the point of origin could be identified.

The characters or variables which were important in determining clusters were those with relatively high "F-ratios" of the between to the within cluster mean squares. The characters which had high "F-ratios" (>30) were adaptability, productivity, inflorescence density, branching intensity and plot cover. These characters determined clusters under agronomic characterisation while those under morphological characters were flower size, size of the lowest leaflets and shape of the leaflets of the lowest leaves. For morphological-agronomic characters were flower size, size of the lowest leaflets and shape of the leaflets of the lowest leaves. For morphological-agronomic characters the important variables responsible for clustering included the number of leaflets per leaf, flower (corolla) size, size of the lowest leaflets, productivity and adaptability.

Some clusters were either wholly or partially composed of accessions belonging to particular or closely related species. This was exemplified by the clusters under morphological characters. Cluster 1 is, for instance, composed of accessions belonging to Z. brasiliensis and Z. setosa only. Z. brasiliensis is represented in the ILCA collection by accessions 11347 and 11476 only. Cluster 2 is dominated by Z. glabra and Z. latifolia while cluster 3 is composed of the accessions 11411 and 11424 (possibly belonging to Z. lanata). Cluster 6 comprises accessions belonging to Z. glochidiata while cluster 7 has an assortment of species including Z. diphylla, Z. pratensis and Z. glochidiata.

Discussion

Zornia is a genus of known notorious taxonomic difficulty. The significance of classifications of the species in this genus would possibly be in the application of various groupings to agronomic problems, particularly selection and/or breeding for specific purposes. The results of classification do not establish but merely suggest the potential of a given accession.

Plant evaluation is expensive in terms of time and effort, and it is known that certain species or genera are more likely to be of value than others. Selection of the materials to be used in the improvement of pasture germplasm depends on the available information about the germplasm. Coupled with the information on preliminary evaluation is the knowledge on characterisation of these materials both of which help in grouping the germplasm on the basis of morphological and agronomic characters.

Agronomic information is, of course, essential for an understanding of the range of variation available in the accessions as a guide to existing desirable combination of characters and to the beneficial recombinations that may be possible with genetic manipulation. A pure morphological or morphological-agronomic classification is liable to be insufficiently sensitive to agronomic heterogeneity. A pure agronomic classification is too crude to permit identification, owing to the limited numbers of attributes that can be devised for collection. There is ample evidence that classification provides the single most profitable technique for obtaining a preliminary synoptic view of a complex system (Williams et al., 1973). In a study of this nature the groups of accessions may be regarded as centres of variation and allow the information to be presented in a more simple form than enumerating the properties of all individuals. So if a collection like this Zornia one is too large to permit succinct description, a classification in groups of known mean characteristics will still be necessary if the information is to be summarised and communicated to others.

Under morphological ordination and classification, accessions belonging to particular species or closely related species tended to form groups. Zornia setosa and Z. brasiliensis formed their own group; so did Z. glabra and Z. latifolia (with a few exceptions); and Z. glochidiata. A continuum of accessions belonging to Z. diphylla, Z. pratensis, some Z. latifolia and other unidentified Zornia spp. was formed. This was consistently so under the three types of analyses (PCA, CA and K-means clustering). Since morphological classification is basically very valuable for identification and description, such classification has provided a useful tool for that purpose. Whereas a mixed morphological-agronomic classification is useful for description, it also provides some guide to the agronomic potentiality of some of the groups. The groupings in this study tended to conform to this observation. Agronomic classification tended to differ from the other two. Here groupings were purely based on agronomic similarities. However, the clusters comprising some accessions of Z. glabra and Z. latifolia were similar in composition for the three categories of classification. Accessions of Z. diphylla, Z. pratensis, Z. setosa and Z. glochidiata tended to exhibit characteristics of poor agronomic potential while Z. glabra and Z. latifolia generally showed better agronomic potentiality. For example, in principal component analysis and cluster analysis accessions of Z. glabra, Z. latifolia, some Z. glochidiata and some unidentified Zornia spp. showed a high expressivity of branching intensity, inflorescence density, plot cover, productivity and adaptability. Good performance in a spaced-plant trial, however, does not guarantee good performance in a grazed pasture. Grazing of Z. brasilensis accessions, for example, had to be terminated at any point in time (CIAT, 1983-1985), at Guilichao, CIAT, although the preliminary evaluation of these materials had been promising.

However, because it is often uneconomic or even impracticable to carry out field trials of all available material, the function of methods such as those outlined in this paper is to select those accessions most likely to justify more intensive trials. Groupings of accessions and/or species have been obtained in this study and some of their characteristics spelt out. Choice of particular groupings will depend upon the use to which these materials are to be put. Groups comprising Z. setosa, Z. glabra, Z. latifolia and Z. glochidiata including their characteristics could give a guideline as to what use these selections could be put should they be advanced to further evaluation trials. The numerical methods of groupings are of particular value in the tropical genera which may require taxonomic revision like Zornia. Groupings defined on the basis of overall similarities of agronomic importance and morphology could well be of more value for practical purposes than stringently defined species and could be used equally well for communication purposes.

Acknowledgements

We wish to acknowledge the technical assistance of Mr. James Ochang in growing and looking after the plants at Soddo, Ethiopia and members of the Forage Agronomy Group (FLAG) who provided transport and other necessary assistance.

One of the authors (J.J. Hakiza) wishes to acknowledge the financial support and authorisation to carry out the study by the International Board for Plant Genetic Resources (IBPGR) and the International Livestock Centre for Africa (ILCA). He is equally grateful to IBPGR for allowing him to attend the PANESA workshop at Arusha, Tanzania.

References

Bogdan, A.V. 1977. Tropical Pasture and Fodder Plants. Longman Group Limited, London.

CIAT, Tropical Pastures Programme, 1983-1985. Annual Reports, Cali, Colombia.

Engelman, Laszlo. 1980. Annotated computer output for BMDPKM/K-means clustering. BMDP Technical Report No. 71. University of California, Los Angeles, U.S.A.

Frane, J.M. and Hill, M.A. 1974. Annotated computer output for factor analysis: A supplement to the write up for computer program BMDP4M. BMDP Technical Report No. 8. University of California, Los Angeles, U.S.A.

McIvor, J.G. and Bray, R.A. (eds). 1983. Genetic resources of forage plants. CSIRO, Melbourne, Australia.

Mohlenbrock, R.J. 1961. A monograph of the leguminous genus Zornia. Webbia 16:1141.

Skerman, P.J. 1977. Tropical forage legumes. FAO Plant Production and Protection series No. 2. FAO, Rome.

Williams, W.T., Edye, L.A., Burt, R.L., and Grof, B. 1973. The use of ordination techniques in the preliminary evaluation of Stylosanthes accessions. Aust. J. Agric. Res. 24:715-731.

Appendix 1. K-means of clustering for Zornia sp. Constitution by accessions of different clusters for various groups of clusters and categories of charecters.

Appendix 1. K-means of clustering for Zornia sp. Constitution by accessions of different clusters for various groups of clusters and categories of charecters.

Appendix 1. K-means of clustering for Zornia sp. Constitution by accessions of different clusters for various groups of clusters and categories of charecters. Cont.1

Appendix 1. K-means of clustering for Zornia sp. Constitution by accessions of different clusters for various groups of clusters and categories of charecters. Cont. 2

Appendix 1. K-means of clustering for Zornia sp. Constitution by accessions of different clusters for various groups of clusters and categories of charecters. Cont. 3

Appendix 1. K-means of clustering for Zornia sp. Constitution by accessions of different clusters for various groups of clusters and categories of charecters. Cont. 4


Previous Page Top of Page Next Page