Cluster Analysis Using Quantitative, Qualitative and Molecular Traits

for the Study of the Genetic Diversity in Pineapple Genotypes

C. de Fatima Machadoa, F.V.D. Souza, J.R.S. Cabral, C.A. da Silva Ledo,
A.P. de Matos and R. Ritzinger
Embrapa Mandioca e Fruticultura Tropical
CP 07, Cruz das Almas
Bahia, 44.380-000
Brazil
Keywords: Ananas spp., multivariate analysis, Gower algorithm, genetic distance,
variability
Abstract
Cluster analysis using quantitative, qualitative and molecular variables is a
useful tool in estimating genetic diversity between genotypes in germplasm
collections. The objective of this study was to carry out a simultaneous analysis of
quantitative, qualitative and molecular variables followed by clustering to study the
genetic diversity between pineapple genotypes using the Gower algorithm. Eleven
quantitative, five qualitative and forty three molecular characteristics in ninety
pineapple genotypes were evaluated. The cophenetic correlation coefficient of the
joint analysis was higher when compared with the individual analysis coefficients.
Ten groups were observed which indicates higher variability among genotypes
evaluated. The simultaneous analysis of the quantitative, qualitative and molecular
variables was efficient in the expression of the genetic diversity between pineapple
genotypes when compared to the individual analysis for each type of variable.
INTRODUCTION
The characterization of germplasm can be accomplished through the use of
phenotypic traits (morphologic and agronomic descriptors), passport data and molecular
markers. The characterization of the accessions is important to determine the genetic
variability, to identify duplicated accessions and to establish nuclear collections. The use
of multivariate techniques is one of the factors that has urged the increase in the studies
about genetic diversity among genotypes (Gonalves et al., 2008, 2009; Mohammadi and
Prasanna, 2003; Podany and Schmera, 2006; Sudr et al., 2007).
Cluster analysis of those traits are done individually, because the genetic distances
are calculated according to the type of the trait. It has been reported by Cruz (2008) that
the procedures to estimate dissimilarity measures can be based on quantitative traits
(Euclidean distances or distances of Mahalanobis), binary traits (Jaccard index, Nei and
Li index) and multicategorical traits (Cole-Rodgers et al., 1997). In that sense, several
discrepancies are observed in relation to the clusters and to the inferences in relation to
the quantification of the variability among accessions of a germplasm bank. A technique
that allows the combined analysis of quantitative and qualitative data was proposed by
Gower (1971).
The objective of this effort was to accomplish the characterization of pineapple
accessions of the Active Germplasm Bank of Embrapa Cassava and Tropical Fruits
through the combined analysis of quantitative, qualitative and molecular data.
MATERIALS AND METHODS
Ninety pineapple accessions of the Active Germplasm Bank of Embrapa Cassava
and Tropical Fruits were evaluated. Each accession was evaluated with regard to 11
quantitative traits (fruit weight, crown weight, fruit length and diameter, crown length,
axis diameter, soluble solids and acidity of the pulp, plant height, stalk length and
a

cristina.machado@cnpmf.embrapa.br

Proc. 7th International Pineapple Symposium

Eds.: H. Abdullah et al.
Acta Hort. 902, ISHS 2011

159

diameter), 5 qualitative traits (leaf color, presence of spines along the borders of the
leaves, fruit shape, colors of the fruit and the pulp) and 43 RAPD (Random Amplified
Polymorphic - DNA) molecular markers.
A combined analysis of the quantitative, qualitative and molecular data was
accomplished for determination of the genetic distance, based on the algorithm described
by Gower (1971). The hierarchical clusters of the accessions were obtained by the
UPGMA - Unweighted Pair-Group Method With Arithmetic Averages (Sneath and Sokal,
1973). The validation of the clusters was determined by the cophenetic correlation
coefficient (Sokal and Rohlf, 1962) and its significance was calculated by the test of
Mantel with 10,000 permutations (Mantel, 1967).
A statistical software was used (R Development Core Team, 2006) for the
calculation of the algorithm of Gower (Gower, 1971). The cophenetic correlation
coefficient was calculated using the Genes software (Cruz, 2008) and the dendrogram was
generated based on the matrix distance using the software Statistica (Statsoft, 2005).
RESULTS AND DISCUSSION
The dendrogram of dissimilarity obtained through the algorithm of Gower
(Gower, 1971) in the evaluation of 11 quantitative traits, 5 qualitative traits and 43
molecular markers is represented in Figure 1. The dissimilarity matrix mean calculated
from the UPGMA method (0.15) provided the formation of 10 groups. On Table 1 the
listed accessions belong to each one of the 10 formed groups.
The cophenetic correlation coefficient (CCC) was 0.92** (significant by the test
of Mantel with 10,000 permutations). The high CCC value indicates high correlation
between the original distance matrices and the cluster matrix; Mohammadi and Prasanna
(2003), Podani and Schmera (2006) and Gonalves et al. (2008) came to the same
conclusion.
The analysis through the algorithm of Gower (Gower, 1971) was efficient in
expressing the degree of genetic diversity among the pineapple accessions evaluated,
demonstrating that the combined analysis of quantitative and qualitative data with the
molecular markers provides a greater efficiency in the knowledge of the genetic diversity.
According to Gonalves et al. (2009), the type and number of the variables chosen can
jeopardize the efficiency of the combined analysis, especially in the case of using a large
number of data from molecular markers, in the quantification of the genetic diversity.
CONCLUSION
The pineapple accessions of the Active Germplasm Bank of Embrapa Cassava and
Tropical Fruits show high genetic variability based on quantitative, qualitative and
molecular traits.
Literature Cited
Cole-Rodgers, P., Smith, D.W. and Bosland, P.W. 1977. A novel statistical approach to
analyze genetic resource evaluations using Capsicum as an example. Crop Sci.
37:1000-1002.
Cruz, C.D. 2008. Programa genes (verso Windows): aplicativo computacional em
gentica e estatstica. Viosa: UFV.
Gonalves, L.S, Rodrigues, R., Amaral, A.T.Jr., Karasawa, M. and Sudr, C.P. 2009.
Heirloom tomato gene bank: assessing genetic divergence based on morphological,
agronomic and molecular data using a Ward-modified location model. Genet. Mol.
Res. 8:364-374.
Gonalves, L.S.A., Rodrigues, R. and Amaral Jnior, A.T. 2008. Comparison of
multivariate statical algorithms to cluster tomato heirloom accessions. Genetics and
Molecular Research 7:1289-1297.
Gower, J.C. 1971. A general coefficient of similarity and some of its properties.
Biometrics, Arlington 27:857-874.
Mantel, N. 1967. The detection of disease clustering and generalized regression approach.
160

Cancer Research, Birmingham 27:209-220.

Mohammadi, S.A. and Prasanna, B.M. 2003. Analysis of genetic diversity in crop plants salient statistical tools and considerations. Crop Sci. 43:1235-1248.
Podani, J. and Schmera, D. 2006. On dendrogram-based measures of functional diversity.
Oikos 115:179-185.
R Development Core Team. 2006. A language and environment for statistical computing.
Vienna: R Foundation for Statistical Computing.
Sneath, P.H. and Sokal, R.R. 1973. Numerical taxonomy: the principles and practice of
numerical classification. San Francisco: W.H. Freeman. 573p.
Sokal, R.R. and Rohlf, F.J. 1962. The comparison of dendrograms by objective methods.
Taxon. 11:33-40.
Statsoft, Inc. 2005. Statistica for Windows (data analysis software system), version 7.1.
Statsoft, Tulsa, Oklahoma (USA).
Sudr, C.P., Leonardez, E., Rodrigues, R., Amaral Jnior, A.T. et al. 2007. Genetic
resources of vegetable crops: a survey in the Brazilian germplasm collections pictured
through papers published in the journals of the Brazilian Society for Horticultural
Science. Hortic. Bras. 25:496-503.
Tables
Table 1. Groups and accessions within groups formed in the cluster analysis obtained by
the algorithm of Gower in the evaluation of 11 quantitative traits, 5 qualitative traits
and 43 molecular markers.
Groups
1
2
3
4
5
6
7
8
9
10

Accessions
FRF-632
Guiana
Perolera
RBR-1, SNG-3
SNG-2 (Quinari)
FRF-678, FRF-762, FRF-1399, Pao de Acucar
LBB-1384, LBB-1374
LBB-1385
Hbrido-3607, Natal Queen, Puerto Rico I-67, Smooth Cayenne,
Smooth Cayenne 3, Smooth Cayenne 2
AltoTuri, AUST-2480, BGA-25, Boituva, Champaka F-153, Codazzi.
Comum, Fazenda Barreiro, Fernando Costa, FFR-1200, FRF-11, FRF-1202,
FRF-1208, FRF-1220, FRF-1221, FRF-1226, FRF-1358, FRF-1369,
FRF-1396, FRF-1403, FRF-150, FRF-154, FRF-156, FRF-160, FRF-162,
FRF-168, FRF-235, FRF-250, FRF-297, FRF-351, FRF-585, FRF-609,
FRF-634, FRF-640, FRF-652, FRF-680, FRF-684, FRF-7, FRF-733,
FRF-737, FRF-770, FRF-8, FRF-800, FRF-820, FRF-845, G-79, G-80,
Inerme de Rondnia, Jupi, Jupi 2, Jupi-TO, LBB-1383, LBB-1386,
LBB-1396, LBB-1413, LBB-1444, LBB-1450, LBB-569, LBB-612,
LocaldeTef, MD-2, Muito Rstico, Prola, Primavera, Rondon 2, Rondon 3,
Roxo deTef, Semi-Selvagem, SNG-1, SNG-4, TD-240

161

FRF-632
Guiana
Perolera
LBB-569
RBR-1
SNG-3
SNG-2 (Quinari)
FRF-678
FRF-762
FRF-1399
Pao de Acucar
LBB-1384
LBB-1374
LBB-1385
Hbrido-3607
Natal Queen
Puerto Rico I-67
Smooth Cayenne2
Smooth Cayenne 3
Smooth Cayenne
Jupi-TO
FRF-845
LBB-1386
FRF-1221
Fazenda Barreiro
AUST-2480
Inerme de Rondnia
TD-240
LBB-1383
FRF-770
Roxo deTefe
Semi-Selvagem
Muito Rstico
FRF-800
FRF-1220
LBB-1413
FFR-1200
BGA-25
FRF-737
G-79
LBB-1450
Fernando Costa
AltoTuri
Primavera
LBB-1444
Rondon 3
Rondon 2
FRF-820
FRF-684
LocaldeTef
SNG-4
FRF-1396
Champaka F-153
MD-2
FRF-1403
Codazzi
FRF-351
G-80
Boituva
LBB-612
FRF-235
SNG-1
FRF-7
FRF-156
FRF-634
FRF-160
FRF-250
FRF-585
FRF-652
FRF-640
FRF-168
FRF-154
FRF-162
FRF-150
LBB-1396
FRF-609
Comum
Jupi 2
FRF-1202
FRF-1358
FRF-1369
FRF-733
FRF-680
Jupi
FRF-1208
FRF-1226
FRF-11
FRF-297
FRF-8
Prola

Linkage Distance

Figures

0.4

0.3

162
0.2

0.1

0.0

Fig. 1. Dendrogram of dissimilarity of 90 pineapple accessions obtained through the

algorithm of Gower in the evaluation of 11 quantitative traits, 5 qualitative traits
and 43 molecular markers.