CHARACTERIZATION AND EVALUATION OF TANNIA

(Xanthosoma sagittifolium (L.) Schott) GENOTYPES AT JIMMA,

SOUTHWEST ETHIOPIA

M.Sc. THESIS

SOLOMON FANTAW G/TSADIK

JUNE, 2014
JIMMA UNIVERSITY
 CHARACTERIZATION AND EVALUATION OF TANNIA
(Xanthosoma sagittifolium (L.) Schott) GENOTYPES AT JIMMA,
SOUTHWEST ETHIOPIA

A Thesis Submitted to the School of Graduate Studies

JIMMA UNIVERSITY

In Partial Fulfillment of the Requirements for the Degree of Master of

Science in Horticulture (Vegetable Science)

BY

SOLOMON FANTAW G/TSADIK

JUNE, 2014

JIMMA, ETHIOPIA
 Jimma University College of Agriculture and Veterinary Medicine
ThesisSubmissionRequestForm (F-05)
Name of Student: Solomon Fantaw ID No. 05510/05

Program of Study: M.Sc. in Horticulture (Vegetable)

Title: Characterization and Evaluation of Tannia (Xanthosoma sagittifolium (L.) Schott)

Genotypes at Jimma, Southwest Ethiopia

I have completed my thesis research work as per the approved proposal and it has been
evaluated and accepted by my advisers. Hence, I hereby kindly request the
Department to allow me to present the findings of my work and submit the thesis.

Solomon Fantaw
Name & signature of student

We, the thesis advisers have evaluated the contents of this thesis and found to be
satisfactory, executed according to the approved proposal, written according to the
standards and format of the University and is ready to be submitted. Hence, we
recommend the thesis to be submitted.

Major Advisor: Amsalu Nebiyu (PhD)

Name Signature Date
Co-Adviser: Mr. Tewodros Mulualem
Name Signature Date

Internal Examiner (If Depends on the Verdict)

Name _______________________ Signature Date

Decision/suggestion of Department Graduate Council (DGC)

--------------------------------------------------------------------------------------------------------------------
--------------------------------------------------------------------------------------------------------------------
--------
Chairperson, DGC Signature Date
------------------------------ ------------------------ --------------------
Chairperson, DGC Signature Date
------------------------------ ------------------------ --------------------
 DEDICATION

I dedicate this thesis manuscript to my father Mr. Fantaw G/Tsadik and my mother Mrs.
Yeshi Mengesha, who have shown what life is and creating great academic interest in my
mind and still inspiring me to realize my dream.

i
 STATEMENT OF AUTHOR

First, I declare that this thesis is my genuine work and that all sources of materials used for
this thesis have been duly acknowledged. This thesis has been submitted in partial
fulfillments of the requirements for an advanced MSc degree at Jimma University and is
deposited at the University Library to be made available to borrowers under rules of the
Library. I solemnly declare that this thesis is not submitted to any other institution
anywhere for the reward of any academic degree, diploma or certificate.

Brief quotations from this thesis are allowable without special permission provided that
accurate acknowledgement of source is made. Requests for permission for extended
quotation from or reproduction from this manuscript in whole or in part may be granted by
the head of the major department or the Dean of the School of Graduate Studies when in
his or her judgment the proposed use of the material is in the interests of scholarship. In all
other instances, however, permission must be obtained from the author.

Name: Solomon Fantaw Signature:________________________

Place: Jimma University, Jimma

Date of Submission: _________________________

ii
 BIOGRAPHY

The author was born from his father, Fantaw G/Tsadik and his mother, Yeshi Mengesha in
October, 1983 at Woreillu district, Kabie town, in South Wollo Administration Zone of
Ethiopia. He attended his elementary and junior classes at Kabie Elementary and Junior
School (1991-1999). He attended his high school education at W/Siheen technical and
vocational secondary school from 1999-2001 and Hottie preparatory secondary school
from 2001-2003 both at Dessie town. Then he joined Alemeya University (now Haramaya
University) in 2003 and graduated with a BSc degree in plant production and protection in
July, 2006.

After his graduation, he was employed by Amhara Regional Agricultural and Rural
Development Office in 2006, at Woreillu district and served as crop production and
protection expert for two years and as extension core process coordinator of the district for
one and half year. He was then employed by University of Gondar in 2010 and served for
two years and six months as an assistant lecturer in the department of plant sciences. He
then joined the School of Graduate Studies (SGS) of the Jimma University in September,
2012 to pursue his graduate studies in Horticulture.

iii
 ACKNOWLEDGMENTS

This work was made possible through the combined efforts and positive attitudes of many
kind-hearted persons all of whom I thank very much next to GOD. I shall only mention
few names among theses: I am very grateful to my research advisors; Dr. Amsalu Nebiyu
and Mr. Tewodros Mulualem whose continued guidance and constructive comments,
inspiration, encouragements and support throughout my study period made the completion
of this study smooth and successfully.

My special thanks go to the staff members of department of plant sciences of University of

Gondar, with especial mention of Mr. Asrat Ayalew, Zenebe G/medihin, Daniel Taddese,
Befekadu Abayneh and Kehali Jember for their dedicated assistance, support and
encouragement through the entire period of this work.

I am also deeply indebted to Root, Fruit and vegetable crops research case team staff of the
Jimma Agricultural Research Center (JARC) namely: Mr. Getachew W/michel, Mr. Neim
Semman, Haile Abshiro and all field assistants for their heartfull technical assistance,
encouragement and ever ready help through the entire period of this work.

Also I gratefully acknowledge the Jimma Agricultural Research Center for giving me the
opportunity to study in their site, providing the planting material and other field equipment
and transport access to the site. Also due acknowledgements are for crop production core
process, Agronomy laboratory, soil laboratory and all their staff members for their material
support and encouragement during the study period.

The University of Gondar is duly acknowledged for sponsoring me the MSc study. The
effort made by department of horticulture and plant sciences and the school of graduate
studies in general the collage of agriculture and veterinary medicine of Jimma University
(JUCAVM) in facilitating this work are also gratefully acknowledged.

I would like to express my sincere gratitude to my friend Tiegist Yisehak, Mehari

G/Michel, Wondosen Ayalew, Miuze Mehari, Destalem Gebremichael, and Girma Eshetu
for their sincere understanding, sociability, invaluable advices and encouragement.

I wish to express my deep, heartfelt gratitude to my father, Fantaw G/tsadik and my

mother, Yeshi Mengesha, my brothers (Rekik and Samuel), my sisters (Meseret, Genet and

iv
Hiwot), my mother in-law Abebu mekonen and sisters (Atsede, Seble, Desta and
Dagmawit) for their support and encouragement during the course of study.

Finally I wish to express my deep appreciation to my beloved wife w/o Helen Kibebu and
My sweet and lovely son Adonias Solomon for the affection understanding and patience
during the entire study period.

v
 TABLE OF CONTENTS

CONTENT PAGE

DEDICATION ...................................................................................................................... i

STATEMENT OF AUTHOR ............................................................................................. ii

BIOGRAPHY ..................................................................................................................... iii

ACKNOWLEDGMENTS .................................................................................................. iv

LIST OF TABLES .............................................................................................................. ix

LIST OF FIGURES OR LIST OF ILLUSTRATIONS ................................................... x

LIST OF TABLES IN THE APPENDIX ......................................................................... xi

ACRONYMS AND ABBREVIATIONS ......................................................................... xii

ABSTRACT ...................................................................................................................... xiii

1. INTRODUCTION ........................................................................................................ 1

2. LITERATURE REVIEW ............................................................................................ 5

2.1. Origin and Distribution .................................................................................................. 5

2.2. Botany ............................................................................................................................ 5

2.3. Taxonomic Classification and Genetic Diversity ........................................................... 7

2.4. Importance of the Crop................................................................................................... 8

2.4.1. Economical Value ........................................................................................... 8

2.4.2. Nutritional Importance .................................................................................... 9

2.5. Environmental Requirement ......................................................................................... 10

2.6. Germplasm Characterization ........................................................................................ 11

2.7. Genetic Variability ....................................................................................................... 12

2.8. Heritability (h2B) .......................................................................................................... 14

2.9. Expected Genetic Advance under Selection................................................................. 15

2.10. Association of Characters ........................................................................................... 16

2.10.1. Correlation Studies ........................................................................................ 16

2.10.2. Path Coefficient Analysis .............................................................................. 18

vi
 TABLE OF CONTENTS (Continued)

2.11. Multivariate Studies ................................................................................................... 18

2.11.1. Genetic Distance ............................................................................................ 19

2.11.2. Cluster Analysis............................................................................................. 20

2.12.3. Principal Component Analysis ...................................................................... 20

3. MATERIAL AND METHODS ................................................................................. 22

3.1. Description of the Study Area ...................................................................................... 22

3.2. Experimental Materials ................................................................................................ 22

3.3. Experimental Design .................................................................................................... 24

3.4. Data Collection ............................................................................................................. 24

3.4.1. Qualitative Data ............................................................................................. 25

3.4.2. Quantitative Data ........................................................................................... 27

3.5. Data Analysis................................................................................................................ 28

3.5.1. Analysis of Variances (ANOVA) .................................................................. 28

3.5.2. Analysis of Variance Components ................................................................ 29

3.5.3. Heritability in Broad Sense (h2B) .................................................................. 30

3.5.4. Expected Genetic Advance (GA) .................................................................. 30

3.5.5. Phenotypic and Genotypic Correction Coefficient Analysis......................... 31

3.5.6. Path Coefficient Analysis .............................................................................. 32

3.5.7. Cluster Analysis (CA) ................................................................................... 32

3.5.8. Genetic Divergence Analysis (D2) ................................................................ 33

3.5.9. Principal Component Analysis ...................................................................... 33

3.5.10. Shannon-Weaver Diversity Index (H’).......................................................... 34

4. RESULT AND DISCUSSION ................................................................................... 35

4.1. Analysis of Variance (ANOVA) .................................................................................. 35

vii
 TABLE OF CONTENTS (Continued)

4.2. Estimation of Variability .............................................................................................. 36

4.2.1. Estimation of Range and Mean ..................................................................... 36

4.2.2. Phenotypic and Genotypic Coefficient of Variation ..................................... 38

4.2.3. Heritability in Broad Sense ........................................................................... 39

4.2.4. Expected Genetic Advance ............................................................................ 40

4.3. Association of characters .............................................................................................. 43

4.3.1. Correlations of Total Yield with other Characters ........................................ 43

4.3.2. Correlation of Dry Matter Content with other Characters............................. 44

4.3.3. Genotypic and Phenotypic Correlation ......................................................... 45

4.3.4. Path coefficient analysis ................................................................................ 48

4.4. Multivariate Studies ..................................................................................................... 52

4.4.1. Cluster Analysis............................................................................................. 52

4.4.2. Genetic Divergence ....................................................................................... 55

4.4.3. Principal Component Analysis (PCA)........................................................... 57

4.5. Diversity of qualitative parameters .............................................................................. 64

4.5.1. Leaf Characteristics ....................................................................................... 64

4.5.2. Petiole Characteristics ................................................................................... 64

4.5.3. Cormel Characteristics .................................................................................. 68

4.5.4. Shannon-Weaver diversity index (H`) ........................................................... 68

5. SUMMARY AND CONCLUSION ........................................................................... 70

REFERENCES .................................................................................................................. 73

APPENDEX ....................................................................................................................... 86

viii
 LIST OF TABLES

TABLE PAGE

Table 1 List of Genotypes of Tannia Studied ...................................................................... 22

Table 2. Qualitative data and their descriptions (above ground) ......................................... 25

Table 3. Subterranean qualitative data and their descriptions ............................................. 27

Table 4. Analysis of Variance (ANOVA) for each quantitative character .......................... 29

Table 5. Mean squares of tannia genotypes for16 quantitative characters .......................... 35

Table 6. The range and the mean values of tannia genotypes for 16 characters ................. 37

Table 7. Estimates of phenotypic (σ2p), environmental variances (σ2e), genotypic variances

(σ2g), phenotypic coefficient of variation (PCV) and genotypic coefficient of variation
(GCV), heritability in broad sense (h2B), genetic advance (GA) and genetic advance as
percent of mean for characters ...................................................................................... 42

Table 8. Genotypic (above diagonal) and phenotypic (below diagonal) correlation

coefficients among 16 quantitative characters .............................................................. 47

Table 9. Path coefficient analysis showing direct effect (bold face and underlined) and
indirect effect (off diagonal) on tannia total yield per plant ......................................... 51

Table 10. Cluster means for 16 quantitative characters of tannia genotypes ...................... 55

Table 11 Inter (above diagonal) and Intra (bold and diagonals) cluster distances among 5
major clusters ................................................................................................................ 56

Table 12. Principal components and their loading values, eigen values and % of total
variances for 16 quantitative characters ........................................................................ 60

Table 13. Qualitative characteristics phenotypic classes descriptions and frequency

distribution .................................................................................................................... 65

Table 14. Shannon-Weaver Diversity Index of the 21 qualitative characters ..................... 69

ix
 LIST OF FIGURES OR LIST OF ILLUSTRATIONS

FIGURE PAGE
Figure 1. Dendnogram of 64 Tannia genotypes .................................................................. 53
Figure 2. Characters and their loading values for principal component 1 ........................... 58
Figure 3. Characters and their loading values for principal component 2 ........................... 59
Figure 4 principal components genotype by traits Bi-plot diagram of PCA 1 and PCA 2 . 63
Figure 5 Lamina characteristics among tannia genotypes: Acaulescent growth habit with
cup shaped lamina (a), Acaulescent growth habit with droopy lamina (b), purplish
color of lower side vein (c) light green vein than lamina color (d), smooth leaf margin
(e), undulated type of leaf margin (f) ............................................................................ 67
Figure 6 petiole characteristics purple color of petiole (a), green petiole and light green
edge of petiole (b), green staked with purplish edge of petiole (c) pink to purple
colored petiole edge (d), green staked petiole with red to purple petiole edge (e) ....... 67

x
 LIST OF TABLES IN THE APPENDIX
APPENDIX TABLE PAGE

Appendix Table 1. The means of tannia genotypes for the 16 quantitative characters.......86

xi
 ACRONYMS AND ABBREVIATIONS

ANOVA Analysis of Variance

CA Cluster Analysis
CEC Cation Exchange Capacity
CSA Central Statistical Agency
CV Coefficient of Variation
EIAR Ethiopian Institute of Agricultural Research
FAO Food and Agriculture Organization
GAM Genetic advance as percent of Mean
GCV Genotypic Coefficient of Variation
IBPGR International board for Plant Genetic Resources
JARC Jimma Agricultural Research Center
msal meter above sea level
MOA Ministry of Agriculture
PCA Principal Component Analysis
PCV Phenotypic Coefficient of Variation
RCBD Randomized Complete Block Design
SAS Statistical Analysis System
SNNP Southern Nations Nationalities and Peoples Region

xii
 CHARACTERIZATION AND EVALUATION OF TANNIA (Xanthosoma
sagittifolium (L.) Schott) GENOTYPES AT JIMMA, SOUTHWEST ETHIOPIA

By Solomon Fantaw
Advisors: Amsalu Nebiyu (PhD) and Mr. Tewodros Mulualem

ABSTRACT
Tannia is one of the most important tuber crop for food, feed and industrial applications worldwide. In
Ethiopia, tannia germplasm collection and introduction has been undertaken for the last few decades.
However, the progress to variety development is so slow due to lack of adequate germplasm
characterization and agronomic evaluation for yield and quality. Therefore, a total of 64 tannia
genotypes were studied to characterize and evaluate based on quantitative and qualitative agro-
morphological charactersaiming at determining the extent of genetic variability and relationship among
genotypes. the stusy was conducted at Jimma during the 2013/14 cropping season involving 62
genotypes collected from south, south western and western parts of Ethiopia and two introductions
from Cuba, laid out in 8 × 8 simple lattice design. Analysis of variance showed highly significant (P
<0.01) differences for most of the characters, indicating the existence of variability among the studied
genotypes. High phenotypic coefficient of variation (PCV) along with moderate to high genotypic
coefficient of variation (GCV) as well as high heritability coupled with high genetic advance as percent
of the mean were obtained for number of suckers per plant, number of cormels per plant, total yield per
plant, corm fresh weight per plant and cormel fresh weight per plant. This indicates that there is an
opportunity of selection to improve these characters. Majority of the characters were positively
correlated with each other. Total yield was positively and significantly correlated with corm weight and
cormel weight both at phenotypic and genotypic level (rp=0.957; rg=0.954). The path coefficient
analysis revealed that cormel fresh weight and corm weight exerted the maximum direct positive effect
on total yield per plant. Also, plant height, plant canopy, corm and cormel length, cormel fresh weight
and corm weight exerted positive direct and indirect effect which indicates that these characters could
be effective for selection and yield improvement of tannia. The genotypes were grouped into five
clusters with significant genetic distance, suggesting the possibility to develop better performing
varieties by selecting parents having high mean values for the characters of interest. Further, principal
component analysis resulted in four principal components which explained 70.5% of the variation
present among the genotypes. Shannon-Weaver diversity index (H`) showed also a high phenotypic
diversity for color of leaf margin, lower leaf surface, vein lower leaf surface, petiole color (lower 1/3),
cormel exterior and interior color, shape of cormel, cormel apex and surface texture color. In general,
the study showed that tannia has a wide genetic variability and diversity for qualitative and quantitative
characters, indicating the potential for further utilization of its genetic improvement through selection
and hybridization. However, the presence of morphological variation between genotypes is not a
guarantee for high genetic variation. Hence, there is a need to confirm genotype-environment
interactions and use molecular or biotechnological approaches as a complementary study. In addition
Calcium oxalate content of corm and cormel, effect of time of planting and harvesting also effect of type
of planting material on yield and dry matter content should be considered as future line work.

Key words: Genetic Diversity, Genetic Variability, Heritability, Principal Component Analysis, Tannia,
south west Ethiopia

xiii
 1. INTRODUCTION

Tannia is a herbaceous, monocotyledonous, perennial stem tuber crop that is widely

cultivated in tropical and subtropical regions of the world. Tannia belongs to the family
Araceae and originally came from tropical America (Facciola, 1998; Ramesh et al., 2007).
Globally, tannia is an important food for about 400 million people (Lebot, 2009) and
ranked sixth in planted area and production after cassava, potato, sweet potato, yam and
taro (Perez, 2010).Tannia is a commercial crop in many parts of the world, mainly for
smallholder farmers such as in Nicaragua (Castro et al., 2005), Cameroon (Bown, 2000),
Bangladesh (Paul and Bari, 2012) and Ghana (Opoku-Agyeman et al., 2004).

The challenges facing us in food and agriculture are enormous. According to the most
recent report on the state of food insecurity in the world, during 2011-2013 there were
about 842 million undernourished people from which 827 million (98.2%) were in
developing countries (FAO, 2013). In Sub-Saharan Africa and generally in the developing
countries including Ethiopia the demand for food is likely to rise significantly as a result of
population growth (FAO, 2010). To meet the ever increasing demand for food, root and
tuber crops can play multi-purpose roles in the global food system to address this issue and
feed millions of people (Ceballos, 2009; Lebot, 2009; Ndabikunze et al., 2011).

In Ethiopia root and tuber crops are part of the traditional food systems of the people
especially in the southern, southwestern and western part of the country. There is
enormous possibility for millions of poor farmers to boost production and their livelihood
using root and tuber crops perhaps highly neglected but strategic crops for the country’s
economy (Amsalu et al., 2008). Among the root and tuber crops, taro (Colocasia esculenta
(L.) Schott) and tannia (Xanthosoma sagitiffolium (L.) Schott), locally known as 'Godare',
are tuberous tropical food crops that supply high-energy food. Simon (1992); as cited by
Asfaw (2005) reported that godare has been grown in Ethiopia since time immemorial, but
how and when it was introduced to Ethiopia remains unclear. However, Amsalu et al.
(2008) reported that 120 taro and 87 tannia collections were introduced to Ethiopia in 1978
from Cuba. Taro and tannia are grown mostly as staple or subsistence crops throughout the
hot and humid areas of southwestern Ethiopia (Edossa, 1996; Amsalu and Tesfaye, 2006).

According to FAOSTAT (2012) world production for taro and tannia in 2011 was 10.37
million tones. Nigeria, Ghana, China and Cameron were the world top producer countries.
1
Africa as a continent produces 71% of this production. In Ethiopia a total of 1.5 million
farmers mainly in Southern Nations Nationalities and Peoples (SNNP) region (0.96
million) and in Oromia region (0.5 million) are dependent on taro and tannia as their food
source (CSA, 2012b). During 2011/2012 production year, taro and tannia production area
in Ethiopia reached 39,696 ha (CSA, 2012b) with total production of 315,242 tons of
which 81.2% is used for human consumption and 11.5% reserved for planting material.
From the total national production, SNNP accounted for 84.5% (266,293.5 tons), Oromia
region for 15.2% (48,015.1 tons), Benshangul-Gumuz for 0.05% (154.6 tons) and Gambela
Region for 0.25% (779 tons) (CSA, 2012a).

Even though tannia has many roles, it is un exploited and neglected crop. Lack of
improved varieties and the need of large quantity of planting material (Onokpise et al.,
1999), rare natural flowering and seed setting (Mbouobda et al., 2007), transmission of
pathogens specially dasheen mosaic virus (DsMV) via vegetative propagation which can
cause yield losses up to 90% (Onokpise et al., 1999; Reyes, et al., 2006; Mbouobda et al.,
2007) are major constraints of tannia production.

To overcome these constraints and improve production and productivity of tannia,

researches have been conducted in different countries. Saborio et al. (2004) implemented
mutation induction in order to generate genetic variability specifically resistance to the
pathogenic isolate of Fusarium sp. in Costa Rica. In Ghana, Blay et al. (2004) irradiated
tannia shoot tips with gamma ray to generate variability and to select for resistance to root
rot and leaf blight disease. On those irradiated genotypes, Danquah et al. (2006) reported
highest number of cormel and the highest weight of cormel yield than un-irradiated
genotypes and also reported morphological variation. In Bangladesh, protocol
establishment for micro propagation and vitro callus regeneration were experimented and
reported high percentage of direct regeneration within 25 days of culture (Paul and Bari,
2007). Similarly in Cameroon, Tsafack et al. (2008) reported best response of micro
propagation for White tannia cultivar followed by Red and Yellow cultivar which can
solve the problem of requiring large amount planting material during planting period and
dissemination of disease-free planting material.

In Ghana, random amplified polymorphic DNA (RAPDs) were used to determine genetic
diversity of 70 accessions of tannia, and found wide divergence between accessions
which offer the opportunity to generate variability by crossing (Blay et al., 2004). In
2
Cameroon, three different types of tannia cultivars namely white, red and yellow cultivars
were identified (Mbouobda et al., 2007). Also in Bangladesh 315 tannia genotypes were
studied (Paul & Bari, 2012).

In Ethiopia to improve production and productivity of various root and tuber crops,
researches have been going on since 1966. To mention some achievements two varieties of
Cassava, 28 varieties of Potato, 24 varieties of Sweet Potato, three varieties of Taro, six
varieties of Enset, one variety of Yam were released (Gebremedhin et al., 2008; MOA,
2010). Furthermore, recently 49 Anchote accessions were studied at Bako Agricultural
Research Center (Tilahun et al ., 2014) and 30 introduced sweet potato genotypes have
been studied in Awassa agricultural research center (Tsegaye et al., 2007), 47 Aerial Yam
accessions collected from south and south-western parts of Ethiopia have been studied at
Jimma Agricultural Research Center (Tewodros, 2013b), 75 taro accessions collected from
SNNP of Ethiopia have been studied at Areka Agriculture Research Center (Asfaw, 2005).
Also, since 1994 clones of Enset collected from six parts of Southern region: Wolaita (58),
Gamogofa (64) Gurage (40), Sidamo (20), Waka (69) and Kembata (68) were evaluated
for their kocho, bulla and fiber yield at Areka agriculture research center (Atnafua et al.,
2008).

However, despite the importance of tannia for food and feed and industrial application in
Ethiopia, so far no improved varieties are available due to lack of adequate
characterization, evaluation and genetic analysis works. Even though, the national average
yield level of tannia in Ethiopia is greater than the global average yield of 7.4 t/ha
(FAOSTAT, 2012), its productivity is far below the crop’s potential which is 30-60 t ha-
1
(Lebot, 2009; Mwenye et al., 2010). For crop improvement program, genetic variability is
an essential prerequisite for obtaining high yielding, quality, pest and disease resistance
varieties (Paul & Bari, 2012). This is usually possible by using the available genetic
resources especially the local gene pool is particularly important in the development of
regionally adapted cultivars (Amsalu and Tesfaye, 2006).

According to Amsalu and Tesfaye (2006) and Tewodros (2008) tannia has a large gene
pool in south and southwest Ethiopia in farmers’ field and homesteads. Some cultivars are
also known that are adapted to such varied conditions such as swamps, marshy areas,
flooded lands, bottomlands and dry uplands. But, in recent time some germplasm

3
collection and conservation works have been started by agricultural research centers
(Amsalu et al., 2008; Tewodros and Hussien, 2012). Nevertheless, the collected genotypes
have never been characterized or evaluated for desirable characteristics. So, there is
paucity of information in respect to their genetic variability and agronomic performance.
Therefore, the present study was conducted with the following objectives:

 To characterize and evaluate genotypes based on key agro-morphological

characters
 To determine the genetic relationship among genotypes based on quantitative
characters
 To determine the extent of genetic variability based on quantitative characters

4
 2. LITERATURE REVIEW

2.1. Origin and Distribution

Tannia was originated in tropical America, but currently grown widely as a subsistence
food crop in Asia, Africa and Polynesia (Bown, 2000). From five crops which are under
sub family aroid the only tannia originated from Central & South America others
originated from South-east Asia (Lebot, 2009).

When the Europeans arrived, it was further known to have been grown from Southern
Mexico to Bolivia in the Latin America. Only during the 19th century, it spread widely
throughout the tropical world. Currently, it is cultivated in tropical and subtropical zones,
between latitudes 30⁰ North and 15⁰ South. The main areas of distribution of the crop
include the Caribbean (Cuba, Dominican Republic, Puerto Rico, West Indies), Central and
South America; USA (Florida, Hawaii), West Africa (Cameroon, Ghana, Nigeria, Togo),
and tropical Asia (Indonesia, Malaysia, the South Pacific Islands) (Perez, 2010).

Tannia was introduced between the 16th and 17th centuries to Central and West Africa,
where it was given the common name of cocoyam because of its resemblance to
Colocasia. It was brought by Portuguese slavers into SaoTomé and Principe, where they
had important trading bases and was spread further by traders, missionaries and other
travelers (Bown, 2000; George, 2011).

2.2. Botany

Tannia is a herbaceous, monocotyledonous, perennial plant, but for practical purposes, it is

harvested after 6 -12 months of growth (Ramesh et al., 2007; Lebot, 2009). It is known that
photosynthesis in tannia follows the C-3 or Calvin cycle pathway (kay, 1987). It can reach
up to a height of about 2 m and have a short erect stem, having a corm or main
underground stem in the form of a rhizome from which swollen secondary shoots or
cormels sprout. Each plant has a central large corm surrounded by lateral cormels which
can vary in size between 12 and 25 cm in length and 12 to 15 cm in diameter. The flesh of
tuber may be white, yellow or sometimes pink and the white type is widely preferred (kay,
1987; Ramesh et al., 2007).
Tannia has close resemblance to taro in its general botanical characteristics. However,
tannia is more robust and taller than taro. The leaf of tannia is more or less heart-shaped;
5
there is a deep indentation which divides the base of the lamina in to the lobs. The petiole
is attached to the lamina at the indentation. The leaf is therefore hastate, in contrast to the
peltate leaves of most taro type. A thick mid-rib runs from the point of attachment to the
basal lobes of the leaf. Each of these main veins gives off branches, which in turn, branch
repeatedly to give a reticulate network of veins. Tannia has prominent vein on the leaves
which runs along the margin of the leaf but this marginal vein is absent in taro as well as
the other edible aroids and it serves as distinguishing feature for tannia (Onwueme and
Charles, 1994). Leaf petioles can be more than 2 m long, with blades more than 1 m long
and up to 0.7 m wide (Lebot, 2009).

The aroid family is characterized by protogyny flowering in which the female flowers
become receptive 2 - 4 days before pollen is shed. Protogyny induces cross-pollination and
most aroid inflorescences are adapted specifically to insect pollination (Lebot, 2009). The
inflorescence is born below the leaves, with a pale green spathe about 20 cm long; natural
flowering and seed setting is rare (kay, 1987; Mbouobda et al., 2007), but when it occurs,
the inflorescence consists of a cylindrical spadix of flowers enclosed in a 12-15 cm long
spathe (Ramesh et al., 2007).

The Inflorescences are large, usually from two to five together in a cluster, sometimes up
to ten depending on the genotype and the environment and appearing one after another
(Lebot, 2009). Each female flowers of tannia consists of an ovary on which lies a yellow
disc-like stigma. The ovary contains many ovules and the ovules are arranged in axile
placentation. Each male flower comprises six stamens whose anthers are united. The fruit
is berry, although fruits and seed production in tannia are extremely rare (Onwueme and
Charles, 1994).

Fruit maturity occurs 40-60 days after pollination: The fruit has a mean number of
berries/fruiting head ranging from 200 to 300, while the weight of each head is about 15-20
g. There are averages of 15 seeds per berry and one inflorescence could produce up to
several hundred seeds. 1000 seeds weight 0.2-0.3 g (Lebot, 2009).

Tannia is vegetatively propagated through pieces of the main corm or whole, small
cormels, setts cut from large cormels, stem cuttings consists of the apical portion of the
corm and lower 15-25 cm of petioles (Onwueme and Charles, 1994; Castro et al., 2005).
The root system of tannia is like that of taro, fibrous and superficial, being confined mostly

6
to the top meter of soil. The corm is more or less spherical; the cormels are flask-shaped
and usually larger than those of taro. Ten or more cormels may be produced on the corm
(Onwueme and Charles, 1994).

2.3. Taxonomic Classification and Genetic Diversity

Tannia belongs to genus Xanthosoma, which is a member of the family Araceae. The
family consists of about 110 genera with over 2500 species (Bown, 2000). Genus
Xanthosoma has about 40 species, grown as ornamentals and as food crops (Quero-Garcia
et al., 2010). The cultivated varieties have been allocated to four species: X. atrovirens, X.
caracu, X. nigrum (X. violaceum) and X. sagittifolium (Giacometti & Leon, 1994). Those
three species, X. sagittifolium, X. atrovirens and X. violaceum, have corm and cormel as
important as food, while X. brasiliense is grown for its edible leaves (Lebot, 2009).

The etymology of the genus Xanthosoma comes from the Greek, Xanthos = yellow, and
soma = body, due to the yellow or yellowish color of the corm and cormel pulp
characteristic of several species (Ramesh et al., 2007). The basic chromosome number is
n=13, Chromosome numbers of 2n = 26 (Onwueme and Charles, 1994; Lebot, 2009). The
common names for cocoyam are numerous and diverse, including Tannia, Tania, Tannier,
Yautia, Yautia des anglo saxons, Elephant's ear, Chou caraibe, Belembe, Calalu, Malanga,
Tayobe, Tayonne, Tayo Tyo, Mangaretto, Malanga, Macabo, Rascadera, Taioba, Kimpool,
Kong Kong Taro, Maduma, New cocoyam, Cocoyam (Giacometti & León 1994; Perez,
2010).

Xanthosoma species display a wide diversity of habit, leaf pigmentation, plant size, cormel
shape and number, cormel tip shape and pigmentation, spatial arrangement of cormels, and
cormel flesh pigmentation, which makes the taxonomic position unclear and in recent
years the tendency has been to give the name of X. sagittifolium to all cultivated
Xanthosoma (Lebot, 2009; Quero-Garcia et al., 2010; George, 2011). Among all,
Xanthosoma sagittifolium (L.) Schott is the most widely grown (kay, 1987). Red, white
and Yellow tuber flesh colored tannia cultivars each varying in yield and had a very dark
green, a dark green, light green and purple green leaf margin colors have been reported in
Cameroon (Mbouobda et al., 2007). Also Nurmiyati et al. (2009) observed morphological
differences on tannia accessions on the length of petiole, leaf blade length and width, and
size of cormel.

7
2.4. Importance of the Crop

2.4.1. Economical Value

Root & tuber crops, including tannia play multi-purpose roles, utmost important to the
poor, essential for food security. They represent an untapped potential for further economic
development, in the global food system as a starch supplier, source of cash income, as raw
material for food and processed products, different industries, as feed for livestock, and as
key components in small-scale agro-enterprise development (Onwueme & Charles, 1994;
Perez et al., 2007; Paul & Bari, 2012).

Tannia is one of the major Root & Tuber crops of the world. It has been commonly used as
a staple food since pre-Columbian times. Now a day, it is integrated in the staple diet of
several countries in the Americas, West Africa, Asia and the Pacific (Lima et al., 2010). It
provides about one third of food intake for approximately 200 million people in the tropics
and subtropics and for more than 400 million people worldwide (Onokpise et al., 1999). It
is the third most important starch food crop in Nicaragua (Castro et al., 2005) and the
second in Cameroon (Onokpise et al., 1999; Bown, 2000).

In Ethiopia root crops covered more than 1.51% of the area under all crops and contributed
12.58% to the production of all crops total in the country. Potatoes, Sweet potatoes and
godere added about 36.74%, 20.41% and 20.27% to the area of the root crop total. The
same crops contributed 23.78 %, 32.65 % and 30.79% of the root crop production total in
the same order (CSA, 2013)and about 1.5 million farmers in Ethiopia produce Godere
(CSA, 2012b).

About 10.37 million tones of taro and tannia produced in 2011 worldwide from which 71%
come from Africa continent (FAOSTAT, 2012). Ethiopia produces 315,242 tons from
39,696 ha of land during 2011/2012 production year (CSA, 2012b). From these, 81.2% is
used for human consumption, 11.5% reserved for planting material and 5.5% for local
market. From the total national production of taro and tannia, 84.5% (266293.5 tons)
accounted from SNNP, followed by Oromia region (15.2%) (CSA, 2012a). Tania can
produce 30-60 t/ha of yield (Lebot, 2009) and Optimum yields of 25-37 t/ha (Kay, 1987).

3+
Tannia is tolerant to soil acidity and it can give about 60% yield in soils with 50% Al
saturation (Ramesh et al., 2007). It can tolerate shading and provide efficient utilization of

8
space and light when intercropped between plantation crops. Also tolerant to salinity, being
able to grow in 25-50% sea water, Have high degree of resistance to the root-knot
nematodes (Onwueme and Charles, 1994).

As Adelekan (2012) reported tannia and taro have very good potential of bio-fuels (139
L/tone ethanol and methane) which is comparable with cassava (145 L/tone) and sweet
sorghum (135 L/tone), better from carrot (100 L/tone) and sugar cane (70 L/tone). The
byproduct mash can supply 59 calories energy per 100 g which can be good livestock feed.
Therefore, it will become a significant source of rural energy supply in many developing
areas of the world where this crop is grown.

2.4.2. Nutritional Importance

According to Eddy et al. (2012) roots and tubers contribute about 20-48% of the total
calories and about 7.1% protein to the diets of the people of Sub-Saharan Africa. Corms
and cormels of Aroids are mostly consumed; also leaves and petioles are part of the diet
(Quero-Garcia et al., 2010). Cormels of tannia are widely consumed in several countries of
Americas, West Africa, Asia and the Pacific (Perez et al., 2005; Lima et al., 2010).

The corm and cormels of tannia which are the major economic part contains about 15 to
39% carbohydrates, 2 to 3% protein and 70 to 77% water (Ndon et al., 2003). It has
nutritional value comparable to potato but easier to digest (Sefa-dedeh and Sackey, 2002).
Tannia is also rich in some elements including P, Mg, Zn, K, Fe, Ca, etc; fairly rich in
carotene, ascorbic acid, thiamine, riboflavin and nicotinic acid. There is less toxicant
character (oxalate, hydrocyanic acid and phytic acid) than taro (Ndabikunze et al., 2011;
Eddy et al., 2012). The young leaves contain 2% protein and other minerals (Ndon et al.,
2003).

Tannia has some nutritional advantages over the other root and tuber crops such as Yam
has less than 6% protein while cassava is a poor source (less than 3%) and tannia has fair
protein (7-9%) (Eddy et al., 2012). As Nishanthini & Mohan (2012) reported, tannia has
some medicinal values. It has good natural antioxidant potential, which showed higher
activity radical scavenging to superoxide anions, hydroxyl, nitric oxide and hydrogen
peroxide radicals than the standard ascorbic acid. Tannia prevent and treat bone diseases
such as osteoporosis, in traditional Brazilian medicine (Oliveira, 2012).

9
Tannia must be thoroughly cooked as some varieties contain high levels of calcium oxalate
crystals in the leaves and tubers. The presence of acridity factors cause sharp irritation and
burning sensation in the throat and mouth on ingestion of cooked corms. The main corms
usually contain oxalic acid and for this reason usually only the cormels are eaten which do
not contain much of this acid. The cormels can be eaten as boiled, baked, parboiled and as
oil fried. The tubers must be boiled for about fifteen minutes in water (sometimes with
baking soda added), the water discarded, rinsed, and boiled again in fresh water to rid the
tuber of its toxins (Ramesh et al., 2007). As Iwuoha and Kalu (1995) reported the acridity
of high calcium oxalate can be reduced by peeling, Steeping (43.3%), boiling (82.1%), and
roasting (61.9%) during processing. The reduction during steeping is due to leaching,
because some oxalate fractions are water-soluble.

2.5. Environmental Requirement

Tannia is originally a plant of the tropical rain forest. In their natural habitat, they grow
under the forest canopy, but under cultivation they are usually grown with full exposure to
sunlight (Ramesh et al., 2007; Nishanthini & Mohan, 2012). It grows well in shade, which
facilitates to intercrop with permanent plantations, with annual and perennial crops. The
crop is well known for its profuse sprouting as a volunteer crop. Most farmers take
advantage of volunteer plants in Ghana (Opoku-Agyeman et al., 2004).

Tannia grows best in tropical temperature conditions but can be grown over a fairly wide
range. They have grown successfully in areas where the mean annual temperature is 24°C
with maximum variations ranging from 13 to 29°C. The crop is suited to high rainfall areas
but can be grown with an annual rainfall as low as 1000 mm provided that this is evenly
distributed, although an average rainfall of 1400-2000 mm is preferable (kay, 1987;
Ramesh et al., 2007).

The time of planting depends on the water availability. If an irrigation system is available,
the planting can be made the whole year around. Where there are distinct and dry seasons,
planting done immediately at a time that the rains become regular (Onwueme and Charles,
1994). Field spacing is variable but 1 x 1 m is most commonly used and requires about 1
t/ha of planting material (cormels). However, in practice it ranges from about 60x60 cm to
180x 180 cm. Planting on ridges is often recommended: the corms or cormels are planted
7.5-10 cm deep, with the growth bud pointing downwards; if pieces of the main rootstock

10
are used about 2.5 cm is left above the ground (Kay, 1987). Plants require to be earthed up
several times; weeds are then controlled, the exposure of roots is prevented and the
development of the cormels favored (Castro et al., 2005).

Tannia can be grown on a wide variety of soils, except hard clays or pure sands but for
optimum yields they require a deep, well-drained, rich soil, preferably with a pH of 5.5-6.5
(Ramesh et al., 2007; Perez, 2010). But do not tolerate the presence of water lodging
(Nishanthini & Mohan, 2012). Tannia has good response to mulching and fertilizer
application, 20-40 t/ha of farm yard manure is recommended when available. In traditional
tannia cultivation in Africa, Central America and in the Pacific Islands, very little or no
chemical fertilizer is used especially when the crop is grown on land that has just been
cleared from bush fallow. However, some fertilizer recommendation as in Puerto Rico and
Pacific Islands, N and K at 100 kg ha-1 each in split application (Perez et al., 2007), in
Nigeria with the fertilizer rates of 200, 100 and 100 kg/ha N P K respectively combined
with 20 t/ha sawdust mulch (Shiyam et al., 2007).

2.6. Germplasm Characterization

Germplasm characterization is the recording of distinctly identifiable characteristics, which

are heritable. This needs to be distinguished from preliminary evaluation, which is the
recording of a limited number of agronomic traits considered to be important in crop
improvement (Upadhyaya et al., 2008). Germplasm is the total gene pool of a species or
the genetic materials that represent an organism (Xu, 2010), which consists of landraces,
advanced breeding lines, popular cultivars, wild and weedy relatives, accessions
(Upadhyaya et al., 2010).

According to Upadhyaya et al. (2008), adequate characterization of germplasm for

agronomic and morphological traits is necessary to facilitate utilization of germplasm by
breeders. i.e. to describe accessions, establish their diagnostic characteristics and identify
duplicates; Classify groups of accessions using sound criteria; identify accessions with
desired agronomic traits and select entries for more precise evaluation; develop
interrelationships between, or among traits and between geographic groups of cultivars;
and estimate the extent of variation in the collection. To achieve this, germplasm of all
crops are characterized for morphological and agronomic traits in batches over the years.

11
Germplasm screening against biotic and abiotic stresses and the estimations of food quality
are conducted jointly with various disciplinary scientists.

2.7. Genetic Variability

Any breeding program for improving the genetic pattern of crop plant depends upon the
nature and magnitude of variability and the extent to which the desirable characters are
heritable (Paul & Bari, 2012). Therefore, it is possible to say the knowledge of genetic
variation present within the germplasm is prerequisite to initiate the breeding program i.e.
to develop potential new cultivars aimed at increasing crop productivity either by using
the genetic variability select from the present assembled germplasm or by creating new
variability through crossing of selected parents, inducing mutations or using
biotechnological techniques and to a sustainable plant genetic resource conservation
(Upadhyaya et al., 2010; Xu, 2010; Acquaah, 2012).

Variability is the amount of variation present among the members of a population or

species, which may have references to one or more characters at a genotypic or phenotypic
levels and the variation is differences among individual belonging to a single species or
different species due to the difference in their genetic composition and or the environment
in which they are growing (Allard, 1999; Singh, 2001).

There are two fundamental sources of variation in phenotype: genetic makeup and the
environment. Phenotypic variability is the observable variation present in the character in a
population; it includes both genotypic and environmental components of variation. The
phenotypic value is variable because it depends on genetic differences among individuals,
as well as environmental factors and the interaction between genotypes and the
environment. Because genes are expressed in an environment, the degree of expression of
a heritable trait is impacted by its environment; as a result their magnitudes differ under
different environmental conditions (Acquaah, 2007, 2012). Such variations measure in
terms of phenotypic variance. In attempting to develop improved varieties, the plant
breeders base his or her observation often on the measurement of the phenotype. For plant
breeding to be effective there must be phenotypic variation on the desired traits and some
of the variation must be heritable from generation to generation (Stoskoph et al., 1999).

12
Genetic variability, which is due to genetic differences among individuals within a
population due to differences in their genetic composition, is the core of plant breeding as
its proper management can produce permanent gain in the performance of plant and can
buffer against seasonal fluctuations. Estimation of the magnitude of variation with in a
germplasm collection for important plant attributes will enable breeders to exploit genetic
diversity more efficiently (Sharma, 1998; Singh et al., 2003). But for improvement or
selection of genotype, not only the knowledge of variability, but also high range of
variability, heritability, genetic advance and positive correlation coefficient among traits
are an excellent tool (Akbar et al., 2003; Mwenye et al., 2010).

The response of traits to selection depends on the relative importance of the genetic and
environmental factors which contribute to phenotypic variation among genotypes in a
population (Xu, 2010). If the phenotypic variance is too small, indicates limited genetic
variability to select, resulting in a smaller genetic gain (Acquaah, 2012). High phenotypic
variations which composed of high genotypic variations and less of environmental
variations, that indicated the presence of high genetic variability for a traits and less
influence of environment, have well response for selection (Bisne et al., 2009).

According to Beeching et al. (1993) the extent of genetic variation present between
genotypes and the genetic distance between all closely related species can be achieved
through the characterization of the germplasm using either morphological, biochemical or
genetic markers. Similarly, Acquaah (2007) state genetic variation can be detected at the
molecular level (e.g., DNA markers) as well as the gross morphological level (as visible
variation in morphological traits e.g., height, color, size).

Several authors have widely used morphological and agronomical parameters in the
evaluation of various crops in many countries. In Cameron, 63 germplasm of tannia were
characterized by 18 morphological characteristics (Mbouobda et al., 2007), in Ethiopia,
100 Taro accessions were characterized based on 30 morphological characteristics
(Tewodros, 2013a), in Ghana 78 accessions of tannia germplasm were characterized based
on 31 morphological characteristics (Opoku-Agyeman et al.,2004), also in Bangladesh
Paul & Bari (2012) study 315 tannia genotypes.

The comparison of characters as regards to the extent of genetic variation could be better
judged by the estimation of genotypic coefficient of variation (GCV) in relation to their

13
respective phenotypic coefficient of variation (PCV) (Nechifor et al., 2011). Paul & Bari
(2012) reported a phenotypic coefficient of variances for leaf number (214.52), leaf area
index (97.43), cormel length (125.38), total fresh weight (137.66), total dry weight (92.34),
for cormel breadth (28.63), yield per plant (36.55), corm breadth (37.66), corm length
(19.90), plant height (38.34) and petiole length (37.98). On other hand they report high
genotypic coefficient of variances for total fresh weight (136.84), lormel length (122.50),
leaf number (214.20), cormel weight (73.01) yield per plant (17.70) cormel breadth (21.9),
total dry weight (15.62), corm length (9.15) for tannia genotypes in Bangladesh.

On another way, Opoku-Agyeman et al. (2004) performed Shannon weaver diversity index
and reported variations among tannia accessions for characters of cormel the cooking
quality (0.540), color of upper leaf surface (0.497), shape of cormel (0.495), petiole sheath
edge color (0.424), interior color of cormels (0.388), petiole color (lower 1/3) (0.348),
color of upper leaf surface (0.348), exterior cormel surface color (0.272), color of cormel
apex (0.217), leaf margin color (0.209) and the lowest diversity indices of cross section of
petiole sheath (0.118).

2.8. Heritability (h2B)

Heritability is the proportion of the observed variation in a progeny that is inherited. It

indicates the effectiveness and the scope of genetic improvement of these characters with
which selection of genotypes can be based on phenotypic performance and expressed as
the ration of genetic variance to the total variance (Singh, 2001; Bisne et al., 2009;
Acquaah, 2012).

The knowledge of magnitude and type of genetic variability and their corresponding
heritability of a trait helps to determine the most effective selection strategy to use in a
breeding program. This is because selection of superior genotypes is proportional to the
amount of genetic variability present and the extent to which the characters are inherited
(Nechifor et al., 2011). Selection which is the retention of desired genotypes and
elimination of undesirable ones is a major and important process in breeding for
improvement of one or more plant attributes. Thus, the utilization of any criterion for
selection is linked with genetic coefficient of variation, the magnitudes of heritability,
genetic advance and other genetic parameters for a character (Idahosa et al., 2010).

14
Breeding methods that use selection based on phenotype are effective when heritability is
high for the trait of interest (Sedeek et al., 2009; Acquaah, 2012).

Heritability percentage is estimated as a ratio between the total variance due to genetic
differences, without taking in to a consideration the component of genetic variance is
referred to as heritability in broad sense; because it estimates heritability on basis of all
genetic effects (Mittal & Sethi, 2005). It expresses the extent to which individuals
phenotypes are determined by their genotypes (Falconer, 1983). On the other hand,
heritability expressed as percentage of additive components of variances is referred as
narrow sense heritability (Mittal & Sethi, 2005). It expresses the extent to which
phenotypes are determined by the genes transmitted from the parents (Falconer, 1983;
Acquaah, 2012).

However, when breeding clonally propagated species in which both additive and non-
additive gene action are fixed and transferred from parent to offspring, broad sense
heritability is useful. The magnitude of narrow sense heritability cannot exceed, and is
usually less than, the corresponding broad sense heritability estimate (Wricke et al.,1986;
Acquaah, 2012).

Paul & Bari (2012) reported heritability of different traits in broad sense such as plant
height (96.98), petiole length (93.95), petiole breadth (95.54), leaf length (96.08), leaf
breadth (96.14), leaf number (99.63), cormel weight (85.13), cormel length (95.47), total
fresh weight (98.80), corm breadth (81.04) as high heritable, yield per plant (23.45),
cormel number (36.14), and corm length (20.78) as moderate heritable and low heritability
for characters total dry weight (2.86).

2.9. Expected Genetic Advance under Selection

Genetic advances measures the expected genetic advancement that would result from
selecting the best performing genotypes for a character being assessed (Allard, 1999). The
prediction of genetic advance is considerable importance because it indicates the
magnitude of the expected genetic gain especially when large populations are subjected to
selection (Sedeek et al., 2009), and the progress plant breeders make from one generation
to another (Acquaah, 2012).

15
Genetic advance is a function of heritability of the target trait, the total phenotypic
variation in the population in which selection will be conducted and selection pressure to
be imposed by the plant breeder (i.e. the proportion of the population that is selected for
the next generation) (Acquaah, 2007).

Genetic advance under selection indicates measure of the difference between the mean
genotypic values of the selected population over the mean genotypic value of the original
population for a given character (Allard, 1960). Heritability in itself provides no indication
of the amount of genetic progress that would result from selecting the best individuals.
High heritability value could be obtained with the accessions having a small or large
genotypic variance but genotypic progress would be larger with larger genotypic variance
(Allard, 1999). Heritability estimates along with genetic advance are normally more
helpful in predicting the gain under selection than heritability estimates (Bisne et al., 2009)

Paul & Bari (2012) found highest to moderate genetic advances for plant height (135),
petiole length (92.99), leaf length (71.91), leaf breadth (54.03) and low genetic advances
for cormel weight (4.47), cormel breadth (2.87), corm length (1.80), total fresh weight
(1.56), corm weight (0.76), yield per plant (0.20) and total dry weight (0.0516).

2.10. Association of Characters

2.10.1. Correlation Studies

Association of characters among yield, yield components and other economic traits is
important because yield is a complex character and governed by the number of characters.
Therefore, for rational approach towards the improvement of yield selection has to be
made for the components of yield by combining several desirable attributes (Paul & Bari,
2012). It suggests the advantage of a scheme of selection for more than one character at a
time (Simmonds, 1986). Knowledge of the nature and magnitude of genetic association
among agronomic traits with both yield and yield components as well as among each other
can help in improving the efficiency of selection by making possible use of suitable
combination of characters (Ahmad et al., 2013).

The correlations between each pair of traits are made up of three elements; genetic
correlation, phenotypic correlations and environmental correlation. Genetic correlation (rg)
is the associations of breeding values of the two characters. Genetic correlation measures
16
the extent to which degree the same genes or closely linked genes cause co-variation
(simultaneous variations) in two different characters (Falconer and Mackay, 1996).
Genotypic correlation coefficients provide a measure of genetic association between traits
and thus help in identifying the most important as well as the least important traits to be
considered in a breeding program (Sylva and Carvalho, 1997).

The association between two characters that can be directly observed is the correlation of
phenotypic values or phenotypic correlations (rp). Phenotypic correlations measure the
extent to which the two observed characters are linearly related. It is determined from
measurements of the two characters in a number of individuals of the populations. The
correlation of environmental deviations together with non-additive genetic deviations (i.e.,
dominance and epistatic genetic deviations) is referred to as environmental correlations
(re). Environmental correlations arise from the effect of overall environmental factors that
vary at different environments (Falconer and Mackay, 1996; Sharma, 1998).

The correlation coefficient gives a measure of the relationship between traits and provides
the degree to which various characters of a crop are associated with productivity (Younis et
al., 2008). The correlation coefficient is free of scale and measurement and has values that
lie between +1 and −1 (i.e., correlation can be positive or negative). If there is no linear
association between variables, the correlation is zero. However, a lack of significant linear
correlation does not mean that there is no association (the association could be non-linear
or curvilinear) (Acquaah, 2007).

Agueguia (1993) was reported significant negative genotypic correlation between number
of cormels per plant and cormel dry matter content (rg = -0.41), significant and positive
genotypic correlations between cormel dry matter content and corm dry matter content (rg
= 0.54) among 16 tannia genotypes. Pandey et al. (2009) reported that the total yield per
plant was positively and significantly correlated with number of cormel per plant and
positively and significantly associated with weight of cormel per plant in taro. Asumadu et
al. (2011) reported significant and positive correlation between cormel yield, plant height,
leaf area and number of leaves. Opoku-Agyeman et al. (2004) found high and positive
correlation between plant height and petiole sheath length (r = 0.773), petiole length (r =
0.850), corm circumference (r =0.748) and aboveground corm height (r = 0.692).

17
Mbouobda et al. (2007) found significant, positive and maximum correlation between
number of cormels per plant and weight of cormels per plant (r=0.824) followed by
number of leaves with number of cormels per plant (r=0.691), number of leaves with
weight of cormels per plant (r=0.676), weight of corm with weight of cormels (r= 0.613)
and weight of corm with number of cormels (r=0.433), non- significant correlations
between the number of shoot per plant and the average weight of cormels as well as the
number of shoots per plant and the number of corms.

2.10.2. Path Coefficient Analysis

Path coefficient analysis is a standardized partial regression coefficient that allows

partitioning of correlation coefficient into direct and indirect effects of various traits
towards dependent variable and also helps in assessing the cause-effect relationship as well
as effective selection (Bello et al., 2010). Path coefficient analysis further permits the
partitioning of the correlation coefficients in to components of direct and indirect factor of
association and provides an effective tool in finding out the direct and indirect contribution
of different contributing characters, a critical examination of the specific forces acting to
produce a given correlation and measuring the relative importance of the causal factors
towards yield (Falconar, 1989), while correlation analysis alone becomes insufficient to
explain the relationships among characters (Rao et al., 1997).

Paul & Bari (2011) reported positive maximum direct effect of cormel length (1.24)
followed by plant height (1.22), petiole length(0.3167), lamina length (0.306), cormel
weight (0.266), corm length (0.245) and positive indirect effect of corm weight, plant
height, cormel weight, corm length, cormel length on yield taro accessions. Fekadu et al.
(2013) also reported tuber yield were affected positive and directly by plant height (0.09)
and tuber diameter (0.024) but negative and directly by number of stem per plant (-0.016)
and tuber number per plant (-0.03).

2.11. Multivariate Studies

Multivariate analysis is the branch of statistics concerned with analyzing multiple

measurements that have been made on one or several samples of individuals. Because
these variables are interdependent among themselves it is best to considered them together
(Acquaah, 2007). Multivariate methods are useful for characterization, evaluation and
classification of plant genetic resources when a large number of accessions are to be
18
assessed for several characters (Peeters & Martinelli, 1989).This procedure permit to
establish the relationship among the variables and to determine how the plants vary in
terms of all variables considered together (Ortiz, 1996). Cluster analysis and principal
component analysis can be jointly used to explain the variations in breeding materials and
are most frequent genetic diversity assessing methods (Aremu, 2011, 2012; Ravishanker et
al., 2013).

2.11.1. Genetic Distance

The pattern and level of genetic diversity in a given crop gene pool can be measured in
terms of genetic distances. Genetic distances are measures of the average genetic
divergence between species or between populations within a species (Souza and Sorrells,
1991; Aremu, 2011, 2012). Smaller genetic distances indicate that the populations have
more similar genes. This indicates that they are closely related i.e. that they have a recent
common ancestor or recent interbreeding has taken place. Genetic distance is useful in
reconstructing the history of populations (http://en.wikipedia.org/wiki/Genetic_distance).

The study of genetic diversity to identify groups with similar genotypes is important for
conserving, evaluating and utilizing genetic resources for studying the diversity of pre-
breeding and breeding germplasm and for determining the uniqueness and distinctness of
the phenotypic and genetic constitution of genotypes with the purpose of protecting a
breeder intellectual property rights (Franco et al., 2001; Acquaah, 2007).

Mahalanobis (1936) generalized distance (D2) has been one of the important statistical
tools to provide a rational basis for selection of parents in breeding programs. This
approach adopting multiple measurements and subjected to multivariate analysis, provided
measure of the generalized distance as indicated by D2 statistics. D2 statistics helped in the
identification of genetically divergent genotypes that facilitated grouping and
characterization for agronomic and morphological characteristics. Mahalanobis’s
generalized distance approach has been widely used to assess genetic diversity in crop
plants Such as Asfaw (2005) found a maximum of inter cluster distance (D2 =2669) and
maximum of intra cluster distance (D2 = 8.6), among 75 taro accessions. Pandey & Dobhal
(1997) on 31 genotypes of taro and found a maximum of inter cluster distance (D2 =250)
and maximum of intra cluster distance (D2 = 27.4). Getachew et al. (2013) on coffee found
a maximum of inter cluster distance (D2 =184) among 49 coffee germplasm accessions

19
Ethiopia. Tilahun et al. (2014) reported inter-cluster distance ranging frpm D2 =23.74 to D2
=273.98 among 49 Anchote accessions in Ethiopia.

2.11.2. Cluster Analysis

Cluster analysis is a multivariate statistical procedure whose primary purpose is to group

individual based on the characteristics they possess, so that individuals with similar
performance gathered in to the same cluster (Crossa et al., 1995). Cluster analysis groups
genetically similar genotypes. Clustering can be done on a morphological or molecular
basis (Acquaah, 2007).The resulting clusters of individual should then exhibited high
internal (with in cluster) homogeneity and high external (between cluster) heterogeneity. If
the classification is successful, individuals with in a cluster shall be closer when plotted
genetically and different clusters shall be farther apart. Thus, clustering summarizes
complexity in the data with retention of the majority of the information by describing
performance with relatively few accession groups (Cooper and Delacy, 1994).

Multivariate analysis of morphological quantitative characters and qualitative characters

(using Cluster analysis) has been used previously to measure genetic relationships with in
tannia germplasms. Opoku-Agyeman et al. (2004) cluster 78 accessions based on petiole
sheath length, petiole length, lamina length, lamina width, corm circumference and overall
plant height. Mbouobda et al. (2007) in Cameroon, clustered 63 accessions of tannia based
on their color of main vein, color leaf margarine, color lower leaf surface, petiole color,
color of tuber/cormel flesh. Similarly Pandey & Dobhal (1997) grouped 31 genotypes of
taro in to 3 clusters.

2.12.3. Principal Component Analysis

Principal component analysis (PCA) is one of the multivariate statistical methods that can
be used to identify patterns in a data set and reduces the dimensions of multivariate data by
removing inter-correlations among the traits being studied. It can be utilized for genetic
diversity estimation and grouping of genotypes through bi-plot diagrams (Tabrizi et al.,
2011), and thereby enabling multi dimensional relationships to be plotted on two or three
principal axes. It performs simple reduction of the data set to a few components, for
plotting and clustering purposes, and can be used to hypothesize that the most important
components are correlated with some other underlying variables (Acquaah, 2007).

20
PCA accomplishes this reduction by identifying directions, called principal components,
along which the variation in the data is maximal. By using a few components, each sample
can be represented by relatively few numbers instead of by values for thousands of
variables. Samples can then be plotted, making it possible to visually assess similarities
and differences between samples and determine whether samples can be grouped (Ringnér,
2008).

Principal component analysis essentially restructures data sets containing many correlated
variables into smaller sets of components of the original variables. Each set is uncorrelated
with any other, but components within sets are related. The resulting combinations may
suggest a biological meaning for the grouping of variables or their components. In
addition, new values can be assigned to each orthogonal set and these can usually replace
several original correlated variables in uni-variate statistical procedures (Iezzoni & Pritts,
1991).
Principal component analysis can be performed by the SAS System’s PROC FACTOR.
Performing principal components analysis by factor analysis is preferable (Costello and
Osborne, 2005). Hence, Norman et al. (2014) perform principal component analysis by
factor analysis method in characterization of sweet potato germplasm, Felenji et al. (2011)
on 22 Potato cultivars, Avtar et al. (2014) on 92 germplasm accessions of Toria (Brassica
rapa (L.) var. toria) for 22 agro-morphological traits.

21
 3. MATERIAL AND METHODS

3.1. Description of the Study Area

The experiment was conducted at Jimma Agricultural Research Center (JARC) located at
366 km south west of Addis Ababa. The site is situated at a latitude 7o 46' N and longitude
36o E with an altitude of 1753 m.a.s.l. The soil of the study area is Eutric Nitisol with a pH
of 5.3. The area receives mean annual rainfall of 1432 mm with maximum and minimum
temperature of 29.2 0C and of 8.9 0C, respectively (Tewodros, 2013b).

3.2. Experimental Materials

A total of 64 tannia genotypes having same cormel size, 62 of them were collected from
different parts of south, south western and western parts of Ethiopia during 2004 – 2005,
the remaining two of which were introduced from Cuba and maintained at Jimma
Agricultural Research Center on field gene bank were used (Table 1).

Table 1 List of Genotypes of Tannia Studied

Gen. Name of Local Name of the

Zone Woreda Kebele/Village Altitude
No Genotype crop in the area
1 AAGT003 Keffa Chena Bobakrcha 2100 Sudan Kido
2 AAGT008 Keffa Bench Kochi 1380 Kub Jong
3 AAGT020 Keffa Bench Wachamaji Ferenji Jong
4 AAGT022 Keffa Bench Aman Gonji 1380 Kub Jong
5 AAGT030 Keffa Bench Mizan Kub Jong
6 AAGT031 Keffa Bench Koda 2040 Ferenji Jong
7 AAGT034 Keffa Chena Ralakocho Bacha 1960 Sudan Kido
8 AAGT035 Keffa Decha Chalta 1620 Sudan Kido
9 AAGT036 Keffa Decha Shapa 1840 Sudan Kido
10 AAGT043 Keffa Decha Deha 1880 Sudan Kido
11 AAGT045 Keffa Decha Chiri Sudan Kido
12 AAGT046 Keffa Decha Chiri Sudan Kido
13 AAGT051 Keffa Gimbo Kaiketa 1860 Sudan Kido
14 AAGT052 Keffa Gimbo Beyamo 1680 Sudan Kido
15 AAGT054 Keffa Gimbo Aman 1700 Sudan Kido
16 AAGT058 Keffa Gimbo Getoacho 1640 Sudan Kido
17 AAGT061 Keffa Gimbo Shamba 1500 Sudan Kido(Baka)

22
Table 1. List of genotypes (continued)

Gen. Name of Local Name of the

Zone Woreda Kebele/Village Altitude
No Genotype crop in the area
18 AAGT065 Keffa Decha Erma 1860 Sudan Kido
19 AAGT069 Keffa Decha Adaiminja 1860 Sudan Kido
20 AAGT077 Keffa Decha Muga 1900 Sudan Kido
21 AAGT080 Keffa Decha Gedam 1680 Sudan Kido
22 AAGT083 Keffa Telo Tura 2020 Sudan Kido
23 AAGT085 Keffa Telo Shadie 1640 Sudan Kido
24 AAGT088 Keffa Telo Felegeselam 2060 Ferenji Jong
25 AAGT092 Keffa Gimbo Beymo 1660 Sudan Kido
26 AAGT093 Keffa Gimbo Kicho 1720 Sudan Kido
27 AAGT094 Keffa Gimbo Kuti 1760 Goderie
28 AAGT097 Keffa Gimbo Emicho 1820 Sudan Kido
29 AAGT099 Keffa Gimbo Saja 2060 Sudan Kido
30 AAGT100 Keffa Gimbo Medaobo 1600 Sudan Kido
31 AAGT102 Keffa Gimbo Medaobo 1560 Sudan Kido
32 AAGT106 Keffa Gimbo Konda Sudan Kido
33 AAGT109 Keffa Gesha Hinigdo 1640 Sudan Kido
34 AAGT112 Keffa Gimbo Kaikelo 1600 Sudan Kido
35 AAGT116 Keffa Gimbo Kembo 1820 Gebiza
36 AAGT120 Keffa Chena Kutasheorai 1820 Sudan Kido
37 AAGT121 Keffa Chena Agaro 1980 Sudan Kido
38 AAGT127 Keffa Chena Culish Sudan Kido
39 AAGT132 Keffa Bench Aman Ferenji Jong
40 AAGT135 Keffa Bench Gerika 1460 Kub Jong
41 AAGT138 Keffa Sheka Bukita 1460 Kub Jong
42 AAGT144 Keffa Sheka Selale 1640 Kub Jong
43 AAGT148 Keffa Sheka Wesheka 1660 Kub Jong
44 AAGT152 Keffa Sheka Shimi 1320 Kub Jong
45 AAGT155 Keffa Sheka Gizm Kub Boka
46 AAGT159 Keffa Yeki Korech 1140 Sudan Kido
47 AAGT163 Keffa Yeki Korech 1380 Sudanboka
48 AAGT171 Keffa Mesha Tugri 1840 Ferenji Kido
49 AAGT176 Keffa Mesha Toba 2220 Boka
50 AAGT177 Keffa Mesha Keja 2140 Ferenji Kodo
51 AAGT178 Keffa Mesha Chewaka 1840 Ferenji Kodo

23
Table 1. List of Genotypes (continued)

Gen. Name of Local Name of the

Zone Woreda Kebele/Village Altitude
No Genotype crop in the area
52 AAGT180 Keffa Gesha Asho 2160 Ferenji Kodo
53 AAGT183 Keffa Gesha Yershiniti 2180 Ferenji Kodo
54 AAGT186 Keffa Mesha Gecha Ferenji Kodo
55 AAGT188 Keffa Yeki Chati 1820 Ferenji Kodo
56 AAGT193 Keffa Yeki Gendekore 1260 Ferenji Kodo
57 AAGT195 Keffa Yeki Sbosha 1220 Ferenji Kodo
58 AAGT199 Keffa Yeki Bechi 1180 Ferenji Baka
59 AAGT202 Keffa Yeki Kura Alamo 1220 Ferenj Kido
60 AAGT205 Keffa Yeki Alamo 1380 Ferenj Kido
61 AAGT208 Keffa Chena Tofa 1820 Sudan Kido
62 0002/07 Cuba
63 0003/07 Cuba
64 AAGT174 Keffa Mesha Gtimo 2250 ferenjikido

3.3. Experimental Design

The experiment was laid out in 8 × 8 simple lattice design using single row plots of 8.25
meter long each; planted on a ridges during the onset of the main rainy season (end of
May, 2013), spaced at 1 m apart between rows and 0.75 m between plants. There were 11
plants row-1 and the middle five plants were used for data collection. To avoid a border
effect, additional rows were planted at both ends of each block. After the crop established
well, earthing up and weeding were carried out when necessary.

3.4. Data Collection

Descriptor of tannia developed by International board for Plant Genetic Resources

(IBPGR, 1989) was followed for data collection. There were two sets of data, 21
qualitative (Table 2 and 3) and 16 quantitative, collected for characterization and assessing
the diversity. Most of which were distinguished as highly heritable traits.

The above ground morphological characters were recorded from each genotype on the
middle five plants during the 5th - 6th months after planting (Mwenye et al., 2010), while;
subterranean traits were evaluated at harvest (nine and half months after planting). Leaf
observation was carried out on three fully developed leaves per plant and the average of
five plants were recorded (Opoku-Agyeman et al., 2004).
24
3.4.1. Qualitative Data

Qualitative data and their descriptions are shown in Table 2 (above ground) and Table 2
(subterranean)

Table 2. Qualitative data and their descriptions (above ground)

S.No. Traits Code Descriptions

1 entire, smooth
2 entire, undulate
1 Leaf margin
3 lobed, partly divided
4 divided to or almost to base
1 concolorous(green) to the edge
2 clear edge
2 Leaf margin color
3 purple/red edge
4 pale yellow/creamy
1 in one plane-apex pointing upward(‘erect’)
Leaf position/
3 2 in one plane-apex pointing downward(‘droopy’)
orientation
3 three dimensional (‘cup-shaped’)
1 light green,
2 medium green,
Color of lower leaf
4 3 dark green,
surface
4 reddish/purplish,
5 other specified in the descriptor
1 light green,
2 medium green,
Color of upper leaf
5 3 dark green,
surface
4 reddish/purplish,
5 other specified in the descriptor
1 same colour as lamina,
Color of vein lower 2 lighter green than lamina,
6
leaf surface 3 darker green than lamina,
4 red/ purple
1 same colour as lamina,
Color of vein on upper 2 lighter green than lamina,
7
leaf surface 3 darker green than lamina,
4 red/ purple

25
Table 2. Qualitative data/above ground (continued)

S.No. Traits Code Descriptions

0 absent
8 Leaf variegation
1 Present
1 light green
Petiole color (upper 2 green,
9
2/3) 3 red/purple
4 green streaked with red/ purple
1 light green
Petiole color (lower 2 green,
10
1/3) 3 red/purple
4 green streaked with red/ purple
1 same color as rest of petiole and sheath
Color of edge of 2 lighter green than rest of petiole and sheath
11
petiole sheath 3 darker green than rest of petiole and sheath
4 pink/red/purple
1 Peltate
12 Petiole attachment 2 sub peltate
3 non peltate
1 White
2 Yellow
Interior color of
13 3 Orange
above ground stem
4 pink (or pale red)
5 Purple
1 Acaulescent
14 Growth habit 2 erect above ground stem
3 reclining above ground stem

26
Table 3. Subterranean qualitative data and their descriptions

S.No. Traits Code Descriptions

1 Color of cormel apex
2 Exterior color of cormels
3 Position of cormel apex
4 Shape of cormels
5 Cormel surface texture
6 Stolen formation
Interior color of cormel 3
7
1 white
2 pink/red
1 light or medium brown
2 dark brown
1 above ground
2 under ground
1 globose
2 ovate
3 cylindrical
4 elliptical
5 mixed (state which of the above)
1 smooth
2 fibrous rough
0 Absent
1 Present
1 White
2 Yellow
Orange
4 Pink/pale red
5 purple

3.4.2. Quantitative Data

Quantitative data measurements were carried out five to six months after planting when the
plants have reached their peak above ground vegetative growth. The middle five plants in
each plot were used for data collection.

1. Lamina length (cm): Three fully expanded leaves were selected from the middle
five plants and measured at the longest points on the lamina along the midrib
(Opoku-Agyeman et al., 2004).
2. Lamina width (cm): measured at the widest points of the lamina, perpendicular to
the midrib (Opoku-Agyeman et al., 2004).
3. Number of suckers per plant: the number of suckers emerging from the same plant
4. Petiole length (cm): The length between the base of the plant and point of insertion
or junction of leaf

27
 5. Plant height (cm): measured from soil surface to tip of the terminal leaf
6. Plant canopy diameter (cm): The average horizontal width/diameter of east-west
and north-south directions of the plant canopy measured at crop maturity.
7. Corm length (cm): average of five plants were measured through the vertical axis
using measuring tape
8. Corm breadth/ diameter (cm): average of five plants were measured horizontally
through the middle position of corm
9. Corm fresh weight per plant (kg): After harvesting the corms were weighted in
kilograms
10. Cormel length (cm): average of 4 cormels per plant measured through the vertical
axis using measuring tape and then average of five plants were recorded
11. Cormel diameter (cm): average of 4 cormels per plant measured horizontally
through the middle position of corm and then average of five plants were recorded
12. cormel fresh weight per plant (kg): After detaching the cormles from corm
cormels, washa and cleaned weighted in kilograms
13. Total root yield per plant (Kg): cormles and corm fresh weight were weighted
together in kilograms
14. Number of cormels per plant:
15. Corm and Cormel dry matter (%): Corm and cormel dry matter content were
estimated by slicing a 100g of corm and cormels followed by drying in forced air
circulation oven for 24 hour with 72⁰ C. The result was expressed in percentage as:

Dry matter (%) = weight of sample after drying (g) *100 (Zelalem et al., 2009).
Initial weight of sample (g)

3.5. Data Analysis

3.5.1. Analysis of Variances (ANOVA)

In order to identify the variability among genotypes, the data obtained for each quantitative
character were subjected to ANOVA using SAS version 9.2 (SAS, 2008). Before
performing ANOVA, normality test was performed using Minitab (2010) version16 and
SAS statistical software.

28
The ANOVA model for simple lattice design is:

Yijklm    ti    ( k )  yl   m   ijklm
-------------- (1)

Where: Yijklm = response of Y trait from the ith genotypes, jth replication,

μ= Overall mean effects,

ti= Effects of ith level of treatments,

β= Effects of jth level of replication,

χk= Effects of Kth level of blocks within replications (adjusted for treatments),

yl = Effects of lth level of intra block error,

πm= Effects of the mth randomized complete block error and

Oijklm= is a random error component.

Table 4. Analysis of Variance (ANOVA) for each quantitative character

Source DF MS EMS
Replications r-1 MSr
Block (rep) r (b-1) MSb
Genotypes g-1 MSg r 2g + 2e
Error r*g- [(r-1) + (g-1) + r(b-1)]-1 MSe 2e
Total r*g-1
r= number of replications; g= number of genotypes; b= number of blocks; MSr=mean
square of replications; MSb= mean square of blocks; MSg= mean square of genotypes;
MSe= mean square of error; 2g = genotypic variance and 2e = the error variance
Then the differences between genotypes mean were compared using LSD (Least
significance difference) at 5% probability level.

3.5.2. Analysis of Variance Components

Phenotypic coefficient of variation (PCV) and genotypic coefficient of variation (GCV),

heritability in broad sense (h2B) and genetic advance as percent of means (GA), phenotypic
and genotypic correction coefficient were computed as follows:
Variance component as (Singh, 2001)

Genotypic variance (2g)

MSg - MSe ---------------- (2)

 2g 
r
Where: MSg is genotypic mean square, MSe is error mean square and r is replication
29
Environmental variance component (On genotypic mean basis) (2 e)

 2 e  MSe ------------------------------------ (3)

Phenotypic variance component (2 p)

 2 p   2 g   2e ------------------------------- (4)

Phenotypic coefficient of variation (PCV) and genotypic coefficient of variation (GCV)

were estimated according to the method suggested by Burton and De Vane (1953), as:

Genotypic coefficients of variation (GCV)

 2g
GCV  __
 100 ------------------------------- (5)
x
Phenotypic coefficients of variation (PCV)
2p
PCV  __
 100 -------------------------------- (6)
__
x
Where: x - is the grand mean value of the trait

3.5.3. Heritability in Broad Sense (h2B)

Heritability in broad sense estimated as the ratio the genotypic variance to the phenotypic
variance, in each character using variance components as described by Allard (1960).
 2g
h 2B   100 -------------------------------- (7)
 p
2

3.5.4. Expected Genetic Advance (GA)

The expected gain or genetic advance (GA) with one cycle of selection, assuming the
selection intensity of 5%, was predicted as suggested by Johnson et al. (1955a).

---------------------------------------- (8)
GA  ( K ).(p).(h2 B)
Where: GA = expected genetic advance under selection
K = selection differential which varied with selection intensity (5% intensity was used at
which K = 2.06)
σp = phenotypic standard deviation
h2B = heritability

30
Expected genetic advance as percent of the mean (GAM) was calculated to compare the
extent of predicted genetic advance of different traits under selection, using the following
formula described by Johonson et al. (1955a)

GA
GAM  100 --------------------------------- (9)
x
__
Where: x = population mean

3.5.5. Phenotypic and Genotypic Correction Coefficient Analysis

Phenotypic correlation, the observable correlation between two variables, which includes
both genotypes and environmental components between two variables, was estimated
using the formula suggested by Johonson et al. (1955b).

Pcov xy
rp  -------------------------------- (10)
(Vp x . Vp y )

Gcov xy
rg 
(Vg x . Vg y ) ------------------------------------ (11)

Where: rp = phenotypic correlation coefficient

rg = genotypic correlation coefficient
Pcovxy = phenotypic covariance between variables x and y
Gcovxy = genotypic covariance between variables x and y
Vpx = Phenotypic variance for variables x
Vpy = Phenotypic variance for variables y
Vgx = genotypic variance for variables x
Vgy= genotypic variance for variables y

Significance of genotypic and phenotypic correlation coefficients were tested with the
following formula forwarded by Robertson (1959).
rgxy
t ------------------------------------ (12) for genotype
SErgxy

(1 - r 2gxy)2
SErgxy 
2h 2 x .h 2 y ------------------------------ (13) and

for phenotype rpxy

t
SErpxy
31
 (1 - r 2 pxy)2
SErpxy 
2h 2 x .h2 y
Where:
rgxy, rpxy = genotypic and phenotypic correlation coefficient respectively between character
x and y
SErgxy, Serpxy: Standard error of genotypic and phenotypic correlation respectively
coefficient between character x and y
h2x: Heritability for character x
h2y: Heritability for character y
The calculated absolute t-value was tested against the tabulated t-value at g-2 d.f. for both
correlation coefficients where, g is the number of genotypes.

3.5.6. Path Coefficient Analysis

The direct and indirect effect of yield related traits on total yield per plant were worked out
through path coefficient analysis. The analysis was made following the method suggested
by Dewey and Lu (1959) as follows:

r  P   r ik p kj -------------------------------------- (14)
ij ij
Where:
rij = association between the independent character (i) and dependent Character (j) as
measured by the correlation coefficient.
Pij = Component of direct effects of the independent character (i) on dependent character
(j) as measured by the path coefficient and,
Σrik pkj = Summation of components on indirect effect of a given independent character (i)
on the given dependent character (j) via all other independent character k).
Residual effects were estimated using the formula:

Where, R2 = ∑Pijrij

Total yield per plant were used as dependent character in separate path coefficient analysis
and the remaining 15 characters were used as Independent variables.

3.5.7. Cluster Analysis (CA)

Clustering procedure was performed to group sets of genotypes with similar performance
across morphological attributes using the proc cluster procedure of SAS version 9.2 (SAS,
32
2008) by employing the method of average linkage clustering strategy of the observations
and information summarized by constructing dendrograms.

Cubic clustering criterion (CCC), pseudo F (PSF), and pseudo t2 (PST2) statistics were
used in determining the number of clusters in the data. That is, local peaks of the CCC and
pseudo F statistic combined with a small value of the pseudo t2 statistic and a larger pseudo
t2 for the next cluster fusion.

3.5.8. Genetic Divergence Analysis (D2)

Genetic divergence between clusters was calculated using the generalized Mahalanobis's
D2 statistics using the equation

D2ij  ( X i  X j )S 1 ( X i  X j ) -------------------------- (15)

Where:
D2ij=is the distance between two groups i and j;
Xi and Xj are the two vectors mean of the traits for ith and jth genotypes, respectively
S = is the inverse of the pooled covariance

The D2 values obtained for pairs of clusters were tested for significance at the required
level of probability against the tabulated values of 2 for p degrees of freedom, where p is
the number of variables considered (Singh and Chaudhary, 1987). SAS 9.2 (SAS, 2008)
was employed for the analysis.

3.5.9. Principal Component Analysis

Principal component analysis was performed using correlation matrix by employing

procedure Proc FACTOR in SAS 9.2 (SAS, 2008) in order to examine the relationship
among 16 quantitative attributes that are correlated among each other by converting in to
uncorrelated traits called principal components. PAST Statistical Software (Hammer et al.,
2001) to draw bi-plot diagram.

Number of factors retained was decided by looking at the eigen values. The principal
components that had had eigen values > 1.0 were selected (Costello & Osborne, 2005).
Those traits that had load coefficient values > 0.40 (ignoring the sign) were considered as
relevant scores for the PCAs and were 0.40 flagged with an asterisk (“*”), which has been

33
considered as meaningful loadings and significant contributor to distinguish genotypes
(Costello & Osborne, 2005; Biabani & Pakniyat, 2008).

3.5.10. Shannon-Weaver Diversity Index (H’)

The qualitative data were subjected to analysis of the Shannon-Weaver diversity index (H’)
(Shannon and Weaver, 1949). H’ was calculated to compare phenotypic diversity among
qualitative character for each genotype using the phenotypic frequencies. H’ is defined as
follows:

H    pi ln( pi ) -------------------------------- (16)

H’ = the Shannon-Weaver Diversity Index

pi = the relative abundance of each trait
ln(pi) = the natural logarithm of abundance,
piln(pi)= relative abundance of trait, multiplied by the natural logarithm of the R.
abundance

 p ln( p ) = is the sum of piln(pi) product

i i

  p ln( p ) = the negative sign of the sum that was calculated

i i

To keep Shannon-Weaver diversity index between 0 and 1 or normalized divided by the

maximum value (ln (n)) in each case the formula suggested by Hennink and Zeven (1991)
was used as:

- p ln p
H' i i
----------------------------------- (17)
ln n
For an ‘n’ class trait
H’ of 0 indicates that is monomorphic, i.e. all individual belong to one and the same
category (clan), where as H’ of 1 indicates maximum diversity (Yu Li et al., 1996).

34
 4. RESULT AND DISCUSSION

4.1. Analysis of Variance (ANOVA)

The analysis of variance for the 16 characters studied is given in Table 5. the genotypes
showed highly significant variations (P <0.01) for character Lamina length, lamina width,
number of suckers per plant, plant canopy, petiole length, plant height, corm length, corm
weight, cormel diameter, fresh weight of cormel, number of cormels per plant, corm dry
matter, cormel dry matter and total yield per plant, and significant variation (P <0.05) for
those of cormel length and corm diameter.

Table 5. Mean squares of tannia genotypes for16 quantitative characters

Mean Square
CV (%) R2
Treatment Error
Character
SU 0.76** 0.15 16.73 88.4
LL 6.32** 1.4 4.87 91.89
LW 6.18** 1.5 5.42 89.11
PLC 61.01** 14.04 5.69 87.75
PTLg 35.87** 8.88 6.54 89.61
Ph 57.96** 15.91 7.08 88.78
COL 2.48* 1.43 11.52 77.89
CLD 0.44** 0.17 7.9 80.41
NCL 14.48** 4.9 18.84 83.98
CML 0.95** ad. 0.48 8.04 81.79
CMD 0.93* 0.59 9.27 75.5
COW 0.0226** 0.008 19.7 84.5
CMW 0.0064** 0.0014 16.09 91.62
CMDM 10.41** 4.09 7.24 81.78
CODM 11.23** 4.46 7.65 78.7
Tot Yi. 0.043** 0.0137 17.01 87.55

**, * significance at 0.01 and at 0.05 probability level; ad = adjusted treatment mean
SU = number of suckers per plant, LL = lamina length, LW = lamina width, PLC = plant canopy diameter,
PTLg = petiole length, Ph = plant height, COL = cormel length, CLD = cormel diameter, NCL = number of
cormels per plant, CML = corm length, CMD = corm diameter, COW = cormel fresh weight per plant, CMW
= corm fresh weight per plant, CMDM = corm dry matter, CODM = cormel dry matter, TotYi = total root
yield per plant

The significant variations among tested genotypes for the characters, which is the result of
combinations of genotypic and phenotypic effect (Acquaah, 2012), indicated the presence
of variability to have selection. The result goes with the finding of Paul & Bari (2012) that

35
indicates significant difference among tannia genotypes for characters such as plant height,
petiole length, cormel breadth, corm breadth, cormel weight, corm weight and yield per
plant. Similarly, Tewodros (2013a) reported that petiole length and plant canopy diameter
showed highly significant variation; but he also reported that non significant variation of
lamina length, lamina width, number of sucker per plant, plant height and tuber fresh
weight among taro genotypes in Ethiopia.

4.2. Estimation of Variability

4.2.1. Estimation of Range and Mean

The range and mean of the studied 16 quantitative morphological characters are presented
in Table 6. Wide ranges of variation were recorded for characters like number of sucker
per plant (1.19 to 4.42), lamina length (20.81 to 29.61 cm), lamina width (18.44 to 27.94
cm), plant canopy (56.33 to 82.03 cm), petiole length (37.72 to 57.52 cm), plant height
(44.68 to 72.72 cm), cormel length (7.83 to 13.13), number of cormels per plant (7.38 to
13.57), cormel weight (0.27 to 0.74 kg/plant), corm weight ( 0.13 to 0.43 kg/plant), corm
dry matter content (21.5 to 32.5%), cormel dry matter content (20 to 32.5 %) and total
yield per plant (0.44 to 1.10 kg/plant).

The range revealed the presence of high phenotypic variations among tannia genotypes.
Moreover, the difference between the minimum and the maximum mean values were high,
indicating the availability of variation for improvement through selection. Such a wide
range of variation in plant height, plant canopy, lamina width and lamina length gives good
opportunity for selection to have desired plant characters by selection or hybridization with
respect to spacing (plant population per hectare), leaf area with respect to other
physiological characters like transpiration and photosynthesis (light harvesting structure)
and availability of moisture to improve productivity per hectare base.

36
Table 6. The range and the mean values of tannia genotypes for 16 characters

Maximum Mean Minimum Mean Grand

Character
value Acc. value Acc. Meam
SU 4.42 AAGT102 1.19 AAGT127 2.36
LL 29.61 AAGT152 20.81 AAGT199 24.29
LW 27.94 AAGT152 18.44 AAGT199 22.65
PLC 82.03 AAGT102 56.33 AAGT135 65.51
PTLg 57.52 AAGT069 37.72 AAGT199 45.59
Ph 72.72 0003/07 44.68 AAGT171 56.33
COL 13.13 AAGT152 7.83 AAGT003 10.38
CLD 5.98 AAGT163 3.33 AAGT159 5.19
NCL 13.57 AAGT106 7.38 AAGT003 11.67
CML 9.64 0003/07 6.43 AAGT171 7.94
CMD 10.15 AAGT094 6.59 AAGT159 8.28
COW 0.74 AAGT183, 0003/07 0.27 AAGT171 0.454
AAGT199
CMW 0.43 AAGT069 0.13 AAGT176 0.24
AAGT109
CMDM 32.50 AAGT132 21.50 0002/07 27.94
CODM 32.50 AAGT202 20.00 0002/07 27.61
Tot. Yi 1.10 0003/07 0.44 AAGT127, AAGT159 0.69

SU = number of suckers per plant, LL = lamina length, LW = lamina width, PLC = plant canopy diameter,
PTLg = petiole length, Ph = plant height, COL = cormel length, CLD = cormel diameter, NCL = number of
cormels per plant, CML = corm length, CMD = corm diameter, COW = cormel fresh weight per plant, CMW
= corm fresh weight per plant, CMDM = corm dry matter, CODM = cormel dry matter, TotYi = total yield
per plant

Based on the mean values, the average value was almost twice that of the minimum mean
values for character cormel fresh weight and corm weight indicating that their maximum
contribution to the total variability observed among the genotypes was high. More than
half of the tested genotypes had mean corm and cormel dry matter content of above the
overall mean of the genotypes (27.94% and 27.61% respectively) (Appendix Table 2 and
Table 6). Similarly, 25%, 42% and 36% of the genotypes showed higher weight of cormel,
corm and total yield than the grand mean yield 0.454, 0.24, 0.69 kg/plant respectively.
Hence, there is an opportunity to find genotypes among the tested entries that give better
yield of cormel and corm as well as genotypes which have higher dry matter content of
corm and cormel.

Such a wide variation were observed before and the results were in agreement with the
findings of Opoku-Agyeman et al. (2004) who reported a wide ranges of variation of 78
tannia genotypes for characters like petiole length, plant height, number of cormel per
plant and cormel fresh weight in Ghana. Similarly, Castro et al. (2005) reported variation
37
between tannia genotypes for traits of plant height, number of suckers, number of cormels
per plant, cormel length, cormel weight and total yield. Asha & Nair (2003) reported wide
range of variation of yam (Dioscorea alata) germplasm in leaf length, petiole length, bulbil
length, number of bulbils per plant, tuber length and tuber yield per plant.

4.2.2. Phenotypic and Genotypic Coefficient of Variation

The magnitude of genotypic and phenotypic variability that exists in a species is essential
to initiate a breeding program. Estimated variance components, phenotypic coefficient of
variability (PCV) and genotypic coefficient of variation (GCV) of the characters studied
are presented in Table 7.

For the studied characters, genotypic variance and phenotypic variance ranged from 0.0025
to 23.49 and 0.0035 to 37.52, respectively. Consequently, the maximum phenotypic
variances were obtained for plant canopy (37.52) followed by plant height (36.94) and
petiole length (22.37). Similarly, the genotypic variances for these characters were also
high for plant canopy (23.49) followed by plant height (21.03) and petiole length (13.5),
indicating that the genotype could be reflected by the phenotype and the effectiveness of
selection based on phenotypic performances of these characters. Relatively lower variances
were observed for number of suckers per plant, cormel diameter, cormel length, corm
diameter, corm length, corm fresh weight cormel fresh weight and total root yield per
plant.

On the other hand, PCV ranged from 8.09 for lamina length to 28.64 for number of sucker
per plant and GCV ranged from 4.98 for cormel diameter to 23.48 for number of sucker
per plant. According to Deshmukh et al. (1986) PCV and GCV values of more than 20
percent are considered as high, values less than 10 percent as low and values between 10
and 20 as moderate. Hence, number of sucker per plant (GCV = 23.48; PCV = 28.64),
number of cormels per plant (GCV = 18.64; PCV = 26.49), cormel fresh weight (GCV =
18.81; PCV = 27.24), corm fresh weight (GCV = 21.13; PCV = 26.42) and total yield per
plant (GCV = 17.54; PCV = 24.40) showed moderate to high GCV along with higher PCV.
While the rest showed lower PCV and GCV.

In the present study, almost all characters exhibited higher phenotypic coefficient of
variation than their corresponding genotypic coefficient of variation, indicating the
apparent variations in the genotypes were not only due to genotypic effect but also due to
38
environmental influences, since phenotypic variances were contributed by the effect of
interaction of genotypes and environment (Acquaah, 2012). However, the difference
between phenotypic variances and genotypic variances as well as phenotypic coefficient of
variation and genotypic coefficient of variation were relatively narrow except corm
diameter and cormel length. These shows that the environmental variance had lower share
of variance on the expression of this trait. It indicates that the observed variations for the
trait were mostly due to genetic factors so, selection could be advanced. But, wider gap of
corm diameter and cormel length shows the importance of environmental influences on the
traits. This is in agreement with Bisne et al. (2009) who stated that high phenotypic
variations composed of high genotypic variations and less of environmental variations
indicate the presence of high genetic variability for different traits and less influence of
environment.

The result is in agreement with Choudhary et al. (2013) reported high value of phenotypic
coefficient of variation along with genotypic coefficient of variation for number of suckers
per plant, number of cormels and cormel yield on taro genotypes. Paul & Bari (2012)
obtained high PCV along with high GCV for number of cormels per plant, cormel weight,
corm weight and total fresh weight per plant for tannia genotypes. Similarly, Cheema et al.
(2006) reported high phenotypic and genotypic coefficients of variation for number of
cormels per plant, total yield per plant and corm fresh weight on taro genotypes. Regassa
& Basavaraj (2005) reported that moderate to high GCV and PCV for number of stem per
plant, number of tuber per plant and total tuber weight per plant, but they report moderate
GCV along with high PCV for plant height and tuber dry matter content which showed
low GCV and PCV in this study. Fekadu et al. (2013) reported that number of stem per
plant, number of tuber per plant and total tuber yield showed high genotypic and
phenotypic coefficients of variation on potato germplasm. Ntawuruhunga and Dixon
(2010) reported low GCV and low PCV along with moderate heritability for dry matter
content in cassava genotypes.

4.2.3. Heritability in Broad Sense

The estimate of broad sense heritability is presented in Table 7. Heritability values for the
16 characters ranged from 22.37 % to 67.18%. According to Verma and Agarawal (1982)
heritability values greater than 50% are considered as high, values between 20 to 50% as
medium and values less than 20 are as low heritable. Based on this, lamina length
39
(63.73%), lamina width (60.94%), number of suckers per plant (67.18%), plant canopy
diameter (62.58%), petiole length (60.31%), plant height (56.92%), corm weight (63.96%)
and total yield per plant (51.68%) exhibited high heritability estimates, suggesting that,
greater effectiveness of selection and improvement to be expected from these characters in
future breeding program.

This is in support of Sedeek et al. (2009) who stated that selection would be effective when
major portion of variability for different traits in the source population is heritable. This is
because there would be a close correspondence between the genotype and phenotype due
to relatively small contribution of the environment to the phenotype. Also moderate
heritability estimate of cormel diameter (44.26%), cormel fresh weight (47.68%), number
of cormels per plant (49.43%), corm dry matter (43.59%), cormel dry matter (43.15%),
cormel length (26.85%), corm diameter (22.37%) and corm length (32.87%) exhibited,
which indicates the presence of more environmental effect, because of this selection may
be difficult due to the masking effect of the environment on germplasm.

The present result is in agreement with Paul and Bari (2013) who reported high heritability
of plant height, petiole length, corm weight and total yield per plant for Elephant Foot Yam
accessions in Bangladesh. Also Paul et al. (2011a) reported high heritability of taro
genotypes for lamina length, lamina width, petiole length, plant height and corm weight.
Similarly, Regassa & Basavaraj (2005) reported moderate to high heritability and high
genetic advance for plant height, number of stem per plant, number of tuber per plant and
total yield per plant but differently reported high heritability along high genetic advance of
tuber dry matter content for 100 potato genotypes. Choudhary et al. (2013) found high
heritability along with high values of genetic advance on taro genotypes for cormel weight,
lamina length and lamina breadth.

4.2.4. Expected Genetic Advance

Heritability estimates though provide the basis for selection based on the phenotypic
performance, the estimates of heritability and genetic advance should always be considered
simultaneously; since heritability estimates along with genetic advance are normally more
helpful in predicting the gain under selection than heritability estimates alone (Bisne et al.,
2009). The expected genetic advance expressed as a percentage of the mean (GAM) is
shown in Table 7 and ranged from 4.85% for corm diameter to 39.64% for number of

40
suckers per plant. This indicated that selecting the top 5 percent of the base population
could result in an advance of 4.85% to 39.64% over the population mean.

According to Johnson et al. (1955a) genetic advance as percent of mean could be

considered as low if it ranges between 0-10%, as moderate 10-20% and high if it becomes
above 20%. Consequently, number of sucker per plant, number of cormel per plant, cormel
weight, corm weight and total yield per plant had high genetic advance. Selection based on
these traits will result in the improvement of the performance of the genotypes for the
traits.

Low values of genetic advance were recorded for corm diameter (4.85%), corm length
(7.21%), cormel diameter (9.70%) corm dry matter content (8.65%), cormel dry matter
content (9.02%) and cormel length (7.45%). These low genetic advances arise from
moderate heritability and low estimate of genotypic variances. Therefore, it is imperative
that selection of genotypes based on phenotypic performance for these characters would
not be effective for improvement. Low genetic advance as part of mean with moderate
heritability suggested the role of non additive gene action (dominance and epistasis) for the
control of these characters and most of the variation for these traits were environmental.

High to moderate genetic advances along with high heritability were found for all
characters except corm diameter, corm length and cormel length. This moderate to high
genetic advance indicates that selecting top 5 % of the genotypes could make an advance
of 39.64% in number of suckers per plant, 34.82% in corm weight, 26.98% in number of
cormels per plant, 26.75% in cormel fresh weight, 25.98 % in total yield per plant, 10.62 %
in lamina length, 10.86% in lamina width, 12% in plant canopy diameter, 12.89% in
petiole length and 12.65% in plant height (Table 7). The presence of high heritability
coupled with moderate to high genetic advance indicates that these are more inherited traits
and most likely the heritability was due to additive gene effects of these characters. Hence,
selection for these characters is likely to be more effective and have a greater scope of
improvement. This is in support of Johnson et al. (1955a) who stated that high heritability
along with high genetic advance is more reliable than heritability alone in predicting the
results of selection.

This finding is in harmony with that of Singh et al. (2003) who reported high heritability
estimates along with high genetic advance for weight of corms per plant, number of

41
cormels per plant and cormels fresh weight per plant among taro germplasms. Similarly,
Yadav et al. (2007) reported high heritability along with high genetic advance, high
phenotypic and genotypic coefficient of variation for corm weight per plant, cormel fresh
weight per plant and total yield in taro genotypes. Paul & Bari (2012) also reported high
genetic advance as a percent of mean along with high to moderate heritability for number
of cormels per plant, lamina length, lamina width, petiole length, plant height, cormel fresh
weight, corm weight and total yield per plant; and also low genetic advance for corm
length of tannia germplasm. Cheema et al. (2006) also reported high heritability coupled
with higher genetic advance for number of cormels per plant, total yield per plant and corm
weight on 24 taro genotypes collected from different parts of North India.

Table 7. Estimates of phenotypic (σ2p), environmental variances (σ2e), genotypic variances

(σ2g), phenotypic coefficient of variation (PCV) and genotypic coefficient of variation
(GCV), heritability in broad sense (h2B), genetic advance (GA) and genetic advance as
percent of mean for characters

Character σ2 g σ2 e σ2 p PCV GCV h2B GA GAM

SU 0.307 0.15 0.457 28.64 23.48 67.18 0.94 39.64
LL 2.46 1.4 3.86 8.09 6.46 63.73 2.58 10.62
LW 2.34 1.5 3.84 8.65 6.75 60.94 2.46 10.86
PLC 23.49 14.04 37.52 9.31 7.36 62.58 7.90 12.00
PLlg 13.5 8.88 22.37 10.38 8.06 60.31 5.88 12.89
PH 21.03 15.91 36.94 10.79 8.14 56.92 7.13 12.65
COL 0.525 1.43 1.955 13.47 6.98 26.85 0.77 7.45
CLD 0.135 0.17 0.305 10.64 7.08 44.26 0.50 9.70
NCL 4.79 4.9 9.69 26.49 18.64 49.43 3.17 26.98
CML 0.235 0.48 0.715 10.65 6.11 32.87 0.57 7.21
CMD 0.17 0.59 0.76 10.53 4.98 22.37 0.40 4.85
COW 0.0073 0.008 0.0153 27.24 18.81 47.68 0.12 26.75
CMW 0.0025 0.0014 0.0035 26.42 21.13 63.96 0.08 34.82
CMDM 3.16 4.09 7.25 9.64 6.36 43.59 2.42 8.65
CODM 3.385 4.46 7.845 10.14 6.66 43.15 2.49 9.02
Tot. Yi 0.0146 0.0137 0.0284 24.40 17.54 51.68 0.18 25.98

SU = number of suckers per plant, LL = lamina length, LW = lamina width, PLC = plant canopy diameter,
PTLg = petiole length, Ph = plant height, COL = cormel length, CLD = cormel diameter, NCL = number of
cormels per plant, CML = corm length, CMD = corm diameter, COW = cormel fresh weight per plant, CMW
= corm fresh weight per plant, CMDM = corm dry matter, CODM = cormel dry matter, TotYi = total yield
per plant

42
4.3. Association of characters

Correlation studies
Improvement for a target character can be achieved by indirect selection via other
characters that are more heritable and easy to select (Khayatnezhad et al., 2011).
Therefore, it requires understanding of the magnitude and the interrelationship of the
characters among themselves and with the target yield or quality character. The correlation
analysis helps in determining the direction and number of characters to be considered in
improving yield as well as quality. Traits may either be positively or negatively correlated
due to the mutual association with other characters.

Estimates of the phenotypic and genotypic correlation coefficients between each pair of the
characters in this study are presented in Table 8. The magnitudes of genotypic correlation
coefficients were mostly higher than the corresponding phenotypic correlations. This
revealed that association among these characters was under genetic control and indicating
the preponderance of genetic variance in expression of characters. This means the
influence of environmental factor is lower than the inherent genetic effect and has inherent
associations among various characters in tannia. As El-Mohsen et al. (2012), when value
of phenotypic correlation coefficient was greater than genotypic correlation coefficient, it
shows that apparent association of two traits was not only due to genes but also due to
favorable influence of environment. By contrast, if value of correlation coefficient was
zero, this showed that these two traits were independent.

4.3.1. Correlations of Total Yield with other Characters

Genotypic and phenotypic correlations of yield with other characters are presented in table
8. Total yield per plant had highly significant and positive correlation with cormel fresh
weight per plant and corm fresh weight per plant both at phenotypic and genotypic level.
The maximum significant positive correlation was with cormel fresh weight (rg = 0.954, rp
= 0.957) followed by corm fresh weight (rg = 0.859, rp = 0.825).The strong relationship is
therefore, indicative for that of total yield per plant can be improved by making
simultaneous improvement of those positively correlated characters which will advance
total yield. Also selection of genotypes will be effective based on characters which had
significant and positive correlation with total yield.

43
This result is in agreement with Pandey et al. (2009) who reported significant positive
correlation at phenotypic and genotypic levels for total yield with fresh weight of corm and
cormel in taro genotypes. Similarly, Yadav et al. (2007) reported high significant
correlation of yield with corm fresh weight and cormel fresh weight, also non significant
positive correlation with lamina width, lamina length, corm length and cormel length at
genotypic level for taro genotypes. Asfaw (2005) reported positive and highly significant
correlation of total yield with cormel fresh weight and corm yield. However, in contrast to
this study Paul et al. (2013) reported negative insignificant genotypic and non significant
positive phenotypic correlation of total yield with corm weight and corm length and also
positive highly significant phenotypic and insignificant negative genotypic correlation of
total yield with corm breadth.

4.3.2. Correlation of Dry Matter Content with other Characters

Among the characters studied, both corm and cormel dry matter content showed
insignificant and negative correlation with most of the characters. Cormel dry matter had
maximum positive correlation with cormel diameter (rg = 0.44) at genotypic level and with
corm dry matter content (rp = 0.246) at phenotypic level. The minimum positive correlation
was observed with number of sucker per plant at genotypic as well as at phenotypic level
(rg = 0.001, rp = 0.023, respectively). The maximum negative correlation was observed
between cormel dry matter and cormel fresh weight (rg = -0.318, rp = -0.178) at genotypic
and phenotypic levels, respectively. Similarly, corm dry matter content had maximum and
minimum negative genotypic correlation with plant height (rg = -0.44) and lamina width (rg
= -0.009) respectively. At phenotypic level, the minimum negative correlation (rp = -0.01)
was with number of sucker per plant. The negative association among characters indicates
that simultaneous improvement of characters cloud be quite difficult and independent
selection may have to be carried out to improve such characters.

Even though very week, dry matter content of corm and cormel had a negative correlation
with total yield per plant (rg = -0.093; rg = -0.228), corm fresh weight (rg = -0.042; rg = -
0.017) and cormel fresh weight (rg = - 0.114; rg = -0.318). Such a negative association of
dry matter content with yield and other characters was reported previously for sweet potato
at Jari and Sirinka in north and south Wollo zones (Yohannes et al., 2010). Also, Tsegaye
et al. (2006) reported highly significant negative phenotypic correlation and positive but
non significant genotypic correlation of root dry matter content with root fresh yield and

44
diameter of storage root; also negative non significant phenotypic and genotypic
correlation with number of storage root among 30 genotypes of sweet potato. However, in
contrary with this study, Burhan (2007) reported positive but non-significant correlation of
dry matter content with tuber yield per plant, plant height, number of tuber per plant and
number of stem per plant in potato. Also Khayatnezhad et al. (2011) reported positive non
significant correlation of tuber dry matter content with total yield, number of stem per
plant, plant height and number of tuber per plant among 10 potato genotypes in Iran.

4.3.3. Genotypic and Phenotypic Correlation

Phenotypic Correlation

Among the characters, the maximum positive and significant correlation was observed for
total yield per plant with cormel fresh weight (rp = 0.957) followed by plant height with
petiole length (rp = 0.903), lamina width with lamina length (rp = 0.825) and corm weight
with total yield per plant (rp = 0.825). The minimum positive insignificant correlation was
between corm dry matter and cormel length (rp = 0.018); while the maximum negative
phenotypic insignificant correlation was between cormel dry matter content and cormel
fresh weight (rp = -0.178) and the minimum was between corm dry matter content and
number of sucker per plant (rp = -0.01).

This result is in agreement with Pandey et al. (2009) who reported significant positive
correlation between petiole length and plant height, corm fresh weight and cormel fresh
weight and non significant positive correlation of cormel weight with number of cormel,
corm diameter, cormel length and cormel diameter; also plant height with weight of corm,
weight of cormel, cormel length and diameter, petiole length with number of cormel per
plant for taro accessions. Similarly, Paul et al. (2013) reported that corm length had
positive correlation with corm diameter, corm weight and yield per plant. Generally, the
association of characters at phenotypic level was less pronounced compared to that of
genotypic level in terms of significance, which may be attributed to the influence of
environment on the expression of traits.

Genotypic Correlation

The maximum positive correlation was observed for corm diameter with corm length (r g =
0.976) followed by total yield per plant with cormel fresh weight (rg = 0.954), plant height
with petiole length (rg = 0.941), lamina width with lamina length (rg = 0.87), while the
45
minimum correlation for cormel dry matter with number of sucker per plant (rg = 0.001).
The maximum negative correlation was between plant height and corm dry matter content
(rg = -0.44) and the minimum negative correlation was between lamina width with corm
dry matter content (rg = -0.009). Corm length with corm diameter (rg = 0.976) had highly
significant and plant height with lamina length (rg = 0.83) had significant correlation at
genotypic level but non significant at phenotypic level.

Petiole length had positive and significant correlation only with plant height (rg = 0.941)
and positively correlated with number of suckers per plant, lamina length, lamina width,
plant canopy diameter, cormel length, cormel diameter, corm length, corm diameter,
number of cormel per plant, corm weight, fresh weight of cormel and total yield per plant.
However, petiole length had negative and non significant correlation with corm and cormel
dry matter content. This is in agreement with Asfaw (2005) that reported significant and
positive correlation of petiole length with plant height both at phenotypic and genotypic
level but in opposite to this study he also reported that negative and highly significant
correlation with cormel fresh weight at both level.

Plant height had positive and high significant correlation with lamina length (rg = 0.838)
and petiole length (rg = 0.941). So, selection based on these characters will in advance with
all characters that had positive correlation including yield. This result is in agreement with
Yadav et al. (2007) who reported high significant and positive correlation of plant height
with lamina length; and also non significant correlation with cormel fresh weight and corm
fresh weight among 34 taro accessions at genotypic level. Similarly, Singh et al. (2005)
reported high significant positive correlation of plant height with leaf length and non
significant positive correlation with root diameter and number of leaves per plant both at
genotypic and phenotypic level among 32 genotypes of carrots in India. However, in
opposite to this study Asfaw (2005) reported negative correlation of plant height with
cormel length and cormel fresh weight among taro accessions at genotypic and phenotypic
level.

46
Table 8. Genotypic (above diagonal) and phenotypic (below diagonal) correlation coefficients among 16 quantitative characters

SU LL LW PLC PtLg PH COL CLD NCL CML CMD COWYi CMWYi CMDM CODM TOTYi
SU 0.264 0.15 0.534 0.318 0.303 0.204 -0.039 0.417 0.258 0.306 0.314 0.458 -0.225 0.001 0.395
LL 0.18 0.87** 0.501 0.767 0.838* 0.169 -0.017 0.196 0.519 0.803 0.375 0.554 -0.12 -0.306 0.484
LW 0.086 0.825* 0.391 0.72 0.758 0.324 0.093 0.166 0.568 0.756 0.328 0.495 -0.009 -0.11 0.426
PLC 0.469 0.53 0.439 0.486 0.489 0.085 0.076 0.594 0.124 0.306 0.44 0.349 -0.244 -0.143 0.445
PTLG 0.292 0.761 0.667 0.563 0.941** 0.182 0.159 0.214 0.447 0.681 0.522 0.649 -0.263 -0.208 0.621
PH 0.316 0.769 0.684 0.546 0.903* 0.233 0.167 0.167 0.519 0.763 0.616 0.595 -0.44 -0.35 0.663
COL 0.135 0.229 0.287 0.182 0.304 0.29 0.241 0.277 0.272 0.537 0.52 0.492 0.116 0.43 0.563
CLD 0.027 0.072 0.096 0.168 0.17 0.159 0.252 0.098 0.25 0.21 0.457 0.361 -0.074 0.44 0.46
NCL 0.404 0.151 0.122 0.425 0.196 0.182 0.317 0.138 -0.082 0.044 0.309 0.614 0.214 0.155 0.464
CML 0.176 0.378 0.356 0.229 0.392 0.439 0.225 0.14 0.099 0.976** 0.484 0.497 0.021 -0.07 0.527
CMD 0.147 0.515 0.491 0.378 0.457 0.447 0.236 0.183 0.12 0.503 0.293 0.547 0.111 0.012 0.423
COWYi 0.333 0.319 0.26 0.407 0.436 0.475 0.505 0.315 0.377 0.349 0.295 0.666 -0.114 -0.318 0.954**
CMWYI 0.494 0.489 0.399 0.409 0.591 0.566 0.335 0.288 0.489 0.39 0.372 0.626 -0.042 -0.017 0.859**
CMDM -0.01 -0.144 0.024 -0.117 -0.173 -0.19 0.044 0.072 0.146 0.018 -0.011 -0.09 -0.011 0.369 -0.093
CODM 0.023 -0.171 -0.022 -0.085 -0.082 -0.131 0.219 0.164 0.156 -0.034 -0.019 -0.178 -0.086 0.246 -0.228

TOTYi 0.423 0.414 0.336 0.444 0.535 0.554 0.49 0.333 0.457 0.398 0.351 0.957** 0.825* -0.066 -0.165

*Significant at 5% probability level; **Significant at 1% probability

SU = number of suckers per plant, LL = lamina length, LW = lamina width, PLC = plant canopy diameter, PTLg = petiole length, Ph = plant height, COL = cormel length,
CLD = cormel diameter, NCL = number of cormels per plant, CML = corm length, CMD = corm diameter, COW = cormel fresh weight per plant, CMW = corm fresh weight
per plant, CMDM = corm dry matter, CODM = cormel dry matter, TotYi = total yield per plant

47
4.3.4. Path coefficient analysis

Correlation analysis describes merely the mutual relationship between different pairs of
characters without providing the nature of the cause and effect relationship of characters.
Hence, the phenotypic and genotypic correlations were analyzed further by path-
coefficient technique, which involves partitioning of the correlation coefficients into direct
and indirect effects via alternative characters or pathways. This allows separation of direct
influence of each component on total yield of tannia from the indirect influence caused by
the mutual relationship among them.

Path coefficient analysis was carried out considering total tannia yield per plant as the
resultant variable since it is complex outcome of various characters (Yohannes et al.,
2010), and the rest of the variable viz., plant height, number of cormels per plant, cormel
length, cormel diameter, lamina length, lamina width, number of suckers per plant, plant
canopy diameter, petiole length, corm length, corm diameter, corm weight, fresh weight of
cormel and dry matter content as the causal variables. The estimates of genotypic direct
and indirect effects of those characters are presented in Table 9.

Path analysis revealed that cormel fresh weight (0.55) had maximum direct positive effect
on total tannia yield per plant followed by corm weight (0.51), plant canopy diameter
(0.095), cormel length (0.08) and plant height (0.065). The result of genotypic as well as
phenotypic correlation coefficient proved by path analysis, that cormel fresh weight and
corm fresh weight had positive, significant and direct effect on total yield. But this direct
effect was little bit reduced by the negative indirect effects of corm diameter, number of
cormel, number of sucker per plant, cormel diameter, petiole length, lamina length and
width. However, this negative indirect effect encounter balanced by positive indirect effect
of plant canopy, plant height, cormel length and corm length. Lowest magnitude and
positive direct effect was also exhibited by corm dry matter content (0.0091). On the other
hand, petiole length (-0.068), number of cormel (-0.0615), cormel dry matter content (-
0.044), lamina width (-0.038) and number of sucker per plant (-0.032) had negative direct
effect on total yield per plant. The indirect positive effect ranged 0.0001 to 0.325 while
indirect negative effect ranged -0.0001 to -0.1065.

Plant height, plant canopy, corm and cormel length, cormel fresh weight and corm weight
had positive indirect effect on total yield. Lamina length and width, number of sucker per

48
plant, corm and cormel diameter as well as corm and cormel dry matter content had low
negative indirect effect. This result is in agreement with Paul & Bari (2011) that found
maximum positive direct effect of corm weight; positive direct effect of plant height,
cormel weight, corm length, cormel length; positive indirect effect of corm weight, plant
height, cormel weight, corm length and cormel length. Contrary to this study they reported
lower but positive indirect effect of petiole length, lamina length and lamina breadth for
taro accessions.

Similarly, Paul & Bari (2013) reported positive direct effect of plant height and cormel
length on total yield per plant and also negative indirect effect of number of cormel per
plant on cormel length, petiole length, corm length, corm diameter, corm weight and plant
height; but in opposite to this study, corm length, corm weight and cormel fresh weight
showed negative direct effect on total yield of elephant yam accessions. Also Agueguia
(1993) reported negative direct effect of corm dry matter content and cormel dry matter
content and positive direct effect of number of cormel per plant but in opposite to this
study he reported that number of cormel had positive direct effect on yield of tannia.
Fekadu et al. (2013) reported that plant height had positive direct effect on yield, whereas
number of stem per plant, root dry matter content and number of tubers per plant showed
negative direct effect on potato germplasm. Tsegaye et al. (2006) reported negative direct
effect of tuber dry matter content on total fresh yield of sweet potato genotypes. Tuncturk
and Ciftc (2005) reported that plant height had positive indirect effect on number of tuber,
average tuber weight, tuber yield and dry matter content among 21 potato genotypes in
Turkey. However, Khayatnezhad et al. (2011) reported positive direct and indirect effect of
dry matter content, number of tuber per plant and number of stem per plant on yield of
potato.

As the result of correlation coefficient in Table 8 showed the positive correlation of total
yield per plant with plant height, cormel length, plant canopy diameter, corm length,
cormel fresh weight and corm fresh weight also explained by the path analysis which
indicates their positive direct effect on total yield per plant. Moreover, plant height, plant
canopy diameter, corm and cormel fresh weight have high heritability along with moderate
to high genetic advance as percent of mean. Therefore, direct selection would be effective,
i.e., selecting genotypes having tall plant height, wider plant canopy diameter could be
used to improve total yield of tannia. Similarly percentage of corm dry matter content had

49
direct positive effect, though it had no significant genotypic association with yield,
indicating the importance of these traits towards the genetic improvement of the crop.

Number of sucker per plant (rg=0.395), lamina length (rg=0.484), lamina width (rg=0.426),
petiole length (rg=0.621), cormel length (rg=0.563), number of cormel per plant (rg=0.464)
and corm diameter (rg=0.423) had positive genotypic correlation with total yield per plant,
but the path analysis showed negative direct effect. Hence, the causal factor of positive
correlation might be their respective positive indirect effects of other characters. Such as
the negative direct effect of lamina length and lamina width encountered by positive
indirect effect of plant canopy diameter, plant height, cormel length and width, corm length
corm as well as cormel weight; even if, slightly enhanced by the negative indirect effect of
number of cormel per plant and corm diameter. Similarly, the direct negative effect of
number of sucker corrected by the positive indirect effect of plant canopy diameter, plant
height, cormel length and width as well as corm dry matter content. So, indirect selection
could be used since they had positive correlation coefficient but direct effects were
negative which encountered by other traits.

The negative direct and indirect effect of lamina length, lamina width and petiole length on
total yield per plant as well as negative indirect effect on cormel length and diameter, corm
length and diameter and those fresh weights of cormel and corm may be related to the
transpiration of water, since they were grown on full sun light condition. The larger the
lamina length and lamina width and longer petiole length would lead the plant to transpire
more water and may result in decreasing of photo synthesis activity which leads lower
carbon fixation and lower translocation of photo assimilate. This is in support of
Valenzuela et al. (1991) who reported larger leaves and longer petioles of tannia grown in
the shade (30% to 50% daylight) leading to a greater light interception and to higher rates
of carbon fixation on a per plant basis which results the higher yields of underground
components for shade-grown than those grown in full sun.

The residual effect obtained (0.176) indicated that 16 characters included in this study
explained high percentage (82.4%) of the total variation in yield. Also, indicated that in
addition to these characters, there are some more components that contribute towards yield
of tannia.

50
Table 9. Path coefficient analysis showing direct effect (bold face and underlined) and indirect effect (off diagonal) on tannia total yield per plant

CMD rg of tot
SU LL LW PLC PtLg PH COL CLD NCL CML CMD COWYi CMWYi M CODM y
SU -0.032 -0.0027 -0.0043 0.0463 -0.0205 0.0211 0.0121 0.0000 -0.0252 0.0087 -0.0043 0.175 0.2419 -0.0008 -0.0007 0.395
LL -0.0069 -0.013 -0.032 0.0487 -0.0517 0.0527 0.0157 0.0000 -0.0103 0.0174 -0.0132 0.1567 0.267 -0.0012 0.0095 0.484
LW -0.0036 -0.0105 -0.038 0.0397 -0.0466 0.047 0.0226 -0.0001 -0.0085 0.0173 -0.0125 0.1354 0.2243 0.0001 0.0023 0.426
PLC -0.0156 -0.0064 -0.0158 0.0954 -0.0355 0.0356 0.0124 -0.0001 -0.0295 0.009 -0.0076 0.1963 0.1945 -0.0015 0.0045 0.445
PTLG -0.0097 -0.0095 -0.0261 0.0501 -0.068 0.059 0.0203 -0.0001 -0.0124 0.0177 -0.0115 0.2316 0.3196 -0.0019 0.0055 0.621
PH -0.0103 -0.0101 -0.0273 0.052 -0.0611 0.0653 0.0185 -0.0001 -0.0104 0.0204 -0.0117 0.2693 0.3056 -0.0025 0.0098 0.663
COL -0.0048 -0.0024 -0.0106 0.0146 -0.017 0.015 0.0808 -0.0001 -0.0182 0.0093 -0.0066 0.2753 0.187 0.0007 -0.0118 0.563
CLD -0.0001 -0.0005 -0.0036 0.0133 -0.0112 0.0147 0.0196 -0.0005 -0.0076 0.0064 -0.0042 0.1908 0.1666 0.0003 -0.0108 0.46
NCL -0.0132 -0.0021 -0.0052 0.0457 -0.0136 0.011 0.0239 -0.0001 -0.062 0.0014 -0.0022 0.2027 0.2755 0.0015 -0.0069 0.464
CML -0.0068 -0.0053 -0.0159 0.0209 -0.0291 0.0323 0.0183 -0.0001 -0.0022 0.0411 -0.0144 0.1959 0.2096 -0.0001 0.0029 0.527
CMD -0.0062 -0.0073 -0.021 0.032 -0.0344 0.0339 0.0237 -0.0001 -0.006 0.0263 -0.023 0.1322 0.214 0.0002 0.0005 0.423
COWYi -0.0102 -0.0036 -0.0093 0.0341 -0.0285 0.032 0.0405 -0.0002 -0.0227 0.0147 -0.0054 0.5493 0.3043 -0.0008 0.0085 0.954**
CMWYI -0.0151 -0.0065 -0.0165 0.0361 -0.042 0.0388 0.0294 -0.0002 -0.0329 0.0167 -0.0094 0.3249 0.5146 -0.0002 0.003 0.859**
CMDM 0.0029 0.0017 -0.0005 -0.0152 0.0138 -0.0179 0.006 0.0000 -0.0104 -0.0003 -0.0004 -0.0458 -0.0117 0.0091 -0.0125 -0.093
CODM -0.0005 0.0027 0.002 -0.0097 0.0085 -0.0146 0.0217 -0.0001 -0.0096 -0.0027 0.0003 -0.1065 -0.035 0.0026 -0.044 -0.228

RESIDUAL EFFECT= 0.175

SU = number of suckers per plant, LL = lamina length, LW = lamina width, PLC = plant canopy diameter, PTLg = petiole length, Ph = plant height, COL = cormel length,
CLD = cormel diameter, NCL = number of cormels per plant, CML = corm length, CMD = corm diameter, COW = cormel fresh weight per plant, CMW = corm fresh weight
per plant, CMDM = corm dry matter, CODM = cormel dry matter, TotYi = total yield per plant

51
4.4. Multivariate Studies

4.4.1. Cluster Analysis

Clustering is a multivariate technique that can conveniently show the pattern of genetic
relationship or proximity among genotypes. Cluster analysis based on the average values
of the 16 studied characters resulted in the grouping of the 64 tannia genotypes in to five
major clusters as shown in Figure 1. The Mahalanobis distances (Inter and intra cluster)
value based on the mean of the genotypes were computed for all possible pairs of clusters
as shown in Table 11.

As the cluster analysis showed the smallest cluster (C-V) consisted of two genotypes
(3.125%) while the broadest group (C-II) consisted of 28 genotypes (43.75%). In order to
determine the genetic structures between tannia genotypes in more detail dendrograms
were constructed (Figure1) using average linkage method which depicts the genetic
distances between all genotypes. It can be clearly observed major clusters be divided into
several sub clusters at the lower part of the dendrogram, especially major cluster one and
two. Cluster (C-I) was the second larger as well as tight clustered which comprised 26
genotypes accounted 40.63% of the total genotypes, collected from nine different districts,
Chena (4), Bench (2), Gimbo (6), Decha (1), Sheka (4), Yeki (4), Mesha (3), Telo (1) and
Gesha (1). This cluster was subdivided into different sub clusters based on their similarity
level which ranges from the 64.34% up to 87.92%.

The major cluster (C-II) was the broadest group which accounts 43.75% of all and consists
of 28 genotypes which were collected from Bench (5), Decha (5), Gimbo (6), Telo (2),
Gesha (2), Chena (2), Yeki (4) and Mesha (2) District. This cluster (II) was subdivided
into different sub clusters having range of similarity from 72.35% up to 90.5%. Likewise,
major cluster III further sub clustered based on similarity level that ranges 65.94 to
79.38%, consists of three genotypes from Decha and one from Sheka and accounts 6.25%
of the total genotypes. While that of cluster IV have three genotypes from Gimbo and one
from Mesha District, similarity of sub cluster varies from 64.84% to 85.51%. Cluster V
consists of 2 genotypes both imported from Cuba during 2007 and have 81.05% similarity
level.

52
 Similarity level

100.00
81.48
62.95
44.43

1
30
46
43
38
51
3
16
25
7
59
64
5
57
15
50
14
41
18
29
53
23
42
61

Figure 1. Dendnogram of 64 Tannia genotypes

55
45
2

53
36
40
8
33
17
9
39
24
49
4
48
Tannia Genotypes

58
6
28
12
47
20
37
56
21
26
52
60
22
32
35
34
10
11
44
19
62
63
13
31
27
54
In the present study most of the genotypes collected from different districts and different
kebele clustered together. For instance, genotypes collected from Gesha, Yeki Mesha,
Telo, Chena, Bench, Gimbo and Decha District clustered together in cluster one and two.
However, morphological variation is more important than geographical variation. In line
with this finding, Offei et al. (2004) in Ghana reported that 70 tannia accessions collected
from different geographical regions were clustered together in different cluster group.
Similarly, Opoku-Agyeman et al. (2004) clustered 78 tannia accessions collected from
seven regions of Ghana into eight different groups based on some morphological
characters. Pandey & Dobhal (1997) clustered 31 taro genotypes collected from three
regions of North Eastern India to eleven to different clusters. In other way Tilahun et al.
(2014) clustered 49 Anchote (Coccinia abyssinica) accessions in to 5 major clusters in
Ethiopia and reported the accessions collected from same location grouped to different
cluster. Mondal et al. (2007) clustered 31 potato genotypes to five major clusters in
Bangladesh.

Those genotypes, which clustered in one but from different district, might have originated
in localities different from those assigned based on their collection points. This suggests
that tannia genotypes may have been transported between localities at random as a result of
the normal farmer to farmer diffusion for suiting their local needs and the genetic drift as
vigorous plants are invariably saved by the farmers as seed for the next planting. This may
have been enhanced by the closeness of the districts to each other.

All clusters were also noted to have variability in the respective mean values for all 16
quantitative characteristics (Table 10). Genotypes in cluster one were characterized by the
highest mean value for corm dry matter (28.1 %) and minimum for cormel diameter (5.04).
Cluster two was unique of all in the mean of genotypes. In this cluster no high value even
for a single character and characterized by intermediate means for character of number of
cormels per plant, cormel diameter, cormel dry matter content and corm dry matter content
while the lowest mean for the rest 13 characters.

54
Table 10. Cluster means for 16 quantitative characters of tannia genotypes

character Cluster I Cluster II Cluster III Cluster IV Cluster V

SU 2.34 2.16* 2.53 3.31 3.4**
LL 24.76 22.83* 27.54 26.43 27.75**
LW 23.22 21.31* 26.23** 24.57 23.05
PLC 67.6 61.45* 68.28 78.87** 72.74
PTLg 47.05 41.56* 54.92 48.04 55.63**
PH 58.3 51.57* 66.87 59.22 70.51**
COL 10.38 10.14* 11.67** 10.83 10.21
CLD 5.04* 5.27 5.59** 5.37 5.22
NCL 12.31 10.68 12.4 16.22** 9.26*
CML 8.09 7.62* 8.78 7.93 8.89**
CMD 8.34 7.98* 8.85 9.04 9.1**
COW 0.44 0.41* 0.63 0.53 0.71**
CMW 0.23 0.21* 0.35** 0.28 0.34
CMDM 27.98 28.14 28.38 29.13** 21.5*
CODM 28.1** 27.82 27.75 26.63 20.25*
Tot. Yi 0.67 0.62* 0.98 0.81 1.06**

SU = number of suckers per plant, LL = lamina length, LW = lamina width, PLC = plant canopy diameter,
PTLg = petiole length, Ph = plant height, COL = cormel length, CLD = cormel diameter, NCL = number of
cormels per plant, CML = corm length, CMD = corm diameter, COW = cormel fresh weight per plant, CMW
= corm fresh weight per plant, CMDM = corm dry matter, CODM = cormel dry matter, TotYi = total yield
per plant

Cluster three was characterized by highest mean value of lamina width, cormel length,
cormel diameter, corm weight and intermediate for the rest characters. Likewise, cluster 4
was characterized with the highest mean value of corm dry matter content, number of
cormel and plant canopy diameter, while it was intermediate for the rest characters. Finally
cluster five characterized by highest mean value for total yield, cormel fresh weight, corm
length, corm diameter, plant height, petiole length, leaf length but had lowest mean value
for characters of number of cormel per plant, cormel dry matter content and corm dry
matter content .

In short, cluster analysis showed the presence of genetic variability among the tannia
genotypes and possibility for development of better performing varieties by selecting
parents, having high mean values for the characters, crossing from the divergent clusters or
by other suitable breeding methodology.

4.4.2. Genetic Divergence

Divergence analysis is usually performed to classify the diverse genotypes by using

Mahalanobis (1936) generalized distance D-square techniques which has been one of the
important statistical tools to provide a rational basis for selection of parents in breeding
55
program since the genetic improvement through hybridization and selection depends upon
the extent of genetic diversity between parents. The more divergent the two genotypes are
the more will be the probability of improving through selection and hybridization.

Inter and intra cluster genetic divergence (D2) values among the five clusters for the 64
tannia genotypes based on 16 quantitative characteristics is presented in Table 11. There
were highly significant (P <0.01) differences between clusters cluster II and V (D2 =
155.04) which is the maximum inter cluster distance observed, while the minimum and
insignificant distance was between cluster I and II (D2 = 16.06). Intra cluster distance was
being much lower than the inter cluster, suggested, heterogeneous and homogeneous
nature between and within groups, respectively. The distance between cluster IV and V (D2
= 108.62), Cluster I and Cluster V (D2 = 115.82), cluster III and V(D2 =68.84), cluster II
and IV(D2 = 53.37), cluster II and III (D2 = 67.14), clusters III and IV(D2 = 44.62) and
clusters I and III (D2 = 31.33) was highly significant. The differences between I and IV (D2
= 28.35) was significant (P<0.05), while the distance between cluster I and II (16.06) was
non-significant. The maximum intra cluster distance was found in cluster V (D2= 6.93)
followed by cluster III and IV (D2 = 5.55 for both) while the minimum was in cluster II (D2
= 1.65). In similar way significant distance were reported previously for other root crops
such as Tilahun et al. (2014) reported that highly significant inter-cluster distance among
49 Anchote accessions in Ethiopia. Also, Asfaw (2005) found highly significant
divergence among taro 70 accessions in Ethiopia.

Table 11 Inter (above diagonal) and Intra (bold and diagonals) cluster distances among 5
major clusters

Clusters I II III IV V
I 1.80 ns 16.06 ns 31.33** 28.35* 115.82**
II 1.65 ns 67.14** 53.37** 155.04**
III 5.55 ns 44.62** 68.84**
IV 5.55 ns 108.63**
V 6.93 ns

**= highly significant at p< 0.01(χ 2) = 30.57, * significant p< 0.05(χ 2) = 24.99, ns = non significant at 5%

According to Mondal et al. (2007), crossing of genotypes that have high inter cluster
distance is expected to produce maximum heterosis and generate wide variability in
genetic architecture than those with smaller inter cluster distances. Genotypes grouped
together are less divergent than genotypes that fall into different clusters (Yadav et al.,
56
2007), particularly clusters separated by largest statistical distance (such as, between
cluster I and V, cluster II and V, cluster IV and V) showed the maximum distance. Also,
the cluster mean (Table 10) reveals that cluster I was characterized by maximum cluster
mean of cormel dry matter content while cluster V had the lowest cluster mean for the
character. Similarly, for most character cluster II had the lowest cluster means (total yield
per plant, cormel weight, corm length and width, plant height, petiole length, lamina width
and length), while that of cluster V had the maximum cluster mean values. Also cluster IV
had the maximum cluster mean values for plant canopy diameter while cluster II had the
minimum cluster mean. On the other hand, crossing between genotypes of cluster I with II
and cluster I with IV may not produce desirable recombinants since they had low inter
cluster distance, indicating the close relationship and having recent common ancestors
between them.

4.4.3. Principal Component Analysis (PCA)

PCA was performed to assess the relative importance of each quantitative character to
characterize genotypes. PCA is a technique that identifies plant characters that contributed
more to the observed variation within a group of genotypes (Afuape et al., 2011). The
result of PCA showed that only the first four principal component axes (PCA1, PCA2,
PCA3 and PCA4) had eigen values up to 1.0, which explained 70.5% of the variation
present among genotypes (Table 12). This indicates that the identified characters within
these components exhibited great influence on the phenotype of the genotypes and could
effectively be used for selection among them.

The loading values of characters and their respective principal components are presented in
Table 12. Loading value closer to +1 indicates strong positive relationship while; closer to
-1 indicates a strong negative relationship. The sign on the loadings indicates the direction
of the relationship between the factor and the trait measured (Biabani & Pakniyat, 2008).
The higher the loading, regardless of the direction (positive or negative), the more effective
they will be in discriminating between accessions (Sanni et al., 2010).Two characters with
high weighting in the same principal component are expected to be highly correlated
(Biabani and Pakniyat, 2008). This may be attributed to their genetic relationship. The
principal component analysis showed the maximum loading value for plant height (0.86)
followed by petiole length (0.84) and total yield per plant (0.838) while the minimum
loading value was plant canopy diameter for principal factor two (0.004).
57
Figure 2. Characters and their loading values for principal component 1
SU = number of suckers per plant, LL = lamina length, LW = lamina width, PLC = plant canopy diameter,
PTLg = petiole length, Ph = plant height, COL = cormel length, CLD = cormel diameter, NCL = number of
cormels per plant, CML = corm length, CMD = corm diameter, COW = cormel fresh weight per plant, CMW
= corm fresh weight per plant, CMDM = corm dry matter, CODM = cormel dry matter, TotYi = total yield
per plant

The first principal component accounted for 40.3% of the total variability among
genotypes. As shown in Figure 2, principal component 1 had the maximum positive
loading value for plant height (0.86), followed by petiole length (0.84) and total yield per
plant (0.84). Number of sucker per plant (0.47), lamina length (0.78), lamina width (0.71),
number of cormel per plant (0.43), plant canopy diameter (0.65), corm weight (0.81),
cormel fresh weight (0.74), cormel length (0.47), corm length (0.60) and corm diameter
(0.64) contributed positive loading effect and were the discriminatory characters for this
principal component. On the other hand percentage of cormel and corm dry matter content
had negative loading effect.

This indicates that the first principal component absorbed and accounted for maximum
proportion of total variability in the set of all variables and it can be designated as
representative component. The present study supports the work of Tabachnick and Fidell
(2001) that showed a component with 5 or more strongly loading items are desirable and
indicate as solid factor while components with fewer than three items are generally weak.
This principal component had also higher eigen values of all components and according to
Costello & Osborne (2005) and Biabani & Pakniyat (2008), the higher the eigen-value of a
component, the more representative it is of the data.
58
Figure 3. Characters and their loading values for principal component 2
SU = number of suckers per plant, LL = lamina length, LW = lamina width, PLC = plant canopy diameter,
PTLg = petiole length, Ph = plant height, COL = cormel length, CLD = cormel diameter, NCL = number of
cormels per plant, CML = corm length, CMD = corm diameter, COW = cormel fresh weight per plant, CMW
= corm fresh weight per plant, CMDM = corm dry matter, CODM = cormel dry matter, TotYi = total yield
per plant

Principal component two accounted 12.8% of the total variation among genotypes and had
eigen value of 2.04. Characters such as cormel length (0.46), cormel diameter (0.42),
number of cormel per plant (0.57) and percentage of cormel dry matter content (0.49) with
strong positive loading values and lamina length (-0.45) with strong negative loading value
contributed to the variation of this principal component. Lamina length, lamina width (-
0.369), plant height (-0.334), corm length (-0.142) and corm diameter (-0.208) load this
principal component weakly (< 0.4) and negatively (Figure 3 and Table 12).

The third principal component explained 9.6% of total variability among genotypes. It was
strongly loaded by characters like corm diameter (0.43), corm dry matter content (0.51),
and cormel dry matter content (0.57) and number of suckers per plant (-0.39). Similarly,
traits like plant canopy diameter (0.4), number of cormel per plant (0.46), cormel weight
(0.4) number of suckers per plant (0.394) contributed more to the variation of principal
component four (7.9%).

59
Table 12. Principal components and their loading values, eigen values and % of total
variances for 16 quantitative characters

Character PCA 1 PCA 2 PCA 3 PCA 4

SU 0.474* 0.254 -0.392 0.394
LL 0.783* -0.445* 0.124 0.180
LW 0.714* -0.370 0.344 0.187
PLC 0.653* 0.004 -0.288 0.401*
PLlg 0.841* -0.287 0.025 0.065
PH 0.862* -0.335 -0.015 -0.012
COL 0.467* 0.457* 0.293 -0.173
CLD 0.303 0.418* 0.262 -0.387
NCL 0.433* 0.566* -0.205 0.465*
CML 0.597* -0.142 0.307 -0.208
CMD 0.647* -0.208 0.426* 0.016
COW 0.745* 0.342 -0.225 -0.403*
CMW 0.813* 0.276 -0.102 -0.029
CMDM -0.134 0.374 0.512* 0.296
CODM -0.137 0.478* 0.571* 0.318
Tot. Yi 0.838* 0.347 -0.198 -0.299
Eigen value 6.442 2.045 1.53 1.271
% variance 40.3 12.8 9.6 7.9
% cumulative 40.3 53 62.6 70.5
variance

SU = number of suckers per plant, LL = lamina length, LW = lamina width, PLC = plant canopy diameter,
PTLg = petiole length, Ph = plant height, COL = cormel length, CLD = cormel diameter, NCL = number of
cormels per plant, CML = corm length, CMD = corm diameter, COW = cormel fresh weight per plant, CMW
= corm fresh weight per plant, CMDM = corm dry matter, CODM = cormel dry matter, TotYi = total yield
per plant

In line with this result, many authors performed PCA to identify genetic variability of
different plant genotypes. For instance, Okpul et al. (2004) reported 6 principal
components for 276 taro accessions based on their eigen values, loading effect and
cumulative variations in New Guinea. Similarly, Afuape et al. (2011) obtained 3 principal
components which had eigen values up to 1.0 and explained 76% cumulative variation in
21 sweet potato genotypes in Nigeria. Further, Ahmadizadeh and Felenji (2011) reported
three principal components which had eigen value up to 1.0 and explained cumulative
variance of 80.1% among 22 potato cultivars in Iran.

Principal component bi-plot

After PCA was performed, bi-plot of first and second components, which accounted for
53% cumulative variability among tannia genotypes, was drawn to study the relationships
60
between characters (Figure 4). Hence, the horizontal axis on the bi-plot was related to first
component and the vertical axis was related to the second component. According to
Ahmadizadeh and Felenji (2011), Ghaffari et al. (2011), Tabrizi et al. (2011) if the angle
between vectors or lines are closer to each other, i.e. the angle between them is less than
90⁰, this represents a positive correlation; the more acute the angle the greater positive
correlation among related traits and vice versa, and if the angle between the lines is more
than 90⁰, this indicates the correlation is negative, also if the angle is equal or nearest to
90⁰ shows independency of traits.

Therefore, according to the bi-plot (Figure 4) there were some vectors of traits, which had
a very small angle with each other such as total yield per plant, cormel fresh weight and
corm weight. Also cormel fresh weight, corm yield and total yield per plant had angle
lower than 90⁰ with number of cormel per plant, corm length, cormel diameter, plant
canopy diameter, corm length and diameter, plant height, number of sucker per plant and
petiole length, that means they had positive correlations. Similarly, plant height had
smaller angle with petiole length, corm length and diameter, plant canopy diameter, lamina
length and lamina width that showed their positive correlation, but larger angle (>90⁰) with
cormel and corm dry matter content which indicates their negative correlation.

Hence, genotypes with higher cormel fresh weight and corm weight as well as the larger
cormel length and wider cormel diameter, wider plant canopy diameter, longer petiole and
plant height as well as more number of cormel had higher total yield. Similarly, the wider
and the longer lamina and the taller petiole length as well as the wider and longer corm had
higher plant height. The lamina length and lamina width had wider angle near to 90⁰ with
total yield per plant, cormel fresh weight as well as corm weight. Also, the dry matter
content for corm and cormel had wider angle near to 90⁰ with total yield per plant, cormel
fresh weight and corm weight, which showed that their correlation is low.

Genotypes scattered around the vectors in the bi-plot diagram indicates comprising of
distinct groups of genotypes for the nearby vectors, and selection for one of these traits
should be accompanied by the associated traits (Ghaffari et al., 2011). Therefore, this
association of traits would provide the opportunity to exert multi-traits selection in tannia
breeding programs. For example, genotype numbers 57, 53, 59, 31, 45, 19, 50 and 15
scattered around the cormel weight, corm weight, and total yield. Similarly, genotypes
numbers 64, 16, 43, 44, 54, 27, 10, 62 and 63 scattered around the plant height, lamina
61
length and width, corm length and diameter. Also genotypes; 60, 52, 39, 34, 50, 47 and 41
were scattered around corm and cormel dry matter content.

Genotypes with larger PCA1 and lower PCA2 scores gave high yields and genotypes with
lower PCA1 and PCA2 scores had lower yields. For example, genotype 63 and 62 had
larger load to PCA1 and lower to PCA2; and since these genotypes were located at positive
corner of PCA1 and negative corner of PCA2, they had the highest yield. Likewise,
genotype 46, 1 and 38 had the lowest yield because they had lower load for both PCA1 and
PCA2.

PCA bi-plots ordination generates ways to identify traits that could be used for efficient
selection Ghaffari et al. (2011). For example, if one wants to select higher yielder and
higher dry matter content genotype 31 as well as 53, 57, 45 and 50 would be a suitable
selection. Therefore, it is possible to say genotype number 31 is good since it had coupled
good yield with dry matter content and good cormel size. Even if genotype 63 had had
very large plant size (taller in height, wider and longer lamina) and the highest in yield
among all, lowest in corm and cormel dry matter and lower in number of cormels. On the
other hand genotype 39 had highest dry matter content and good in number of cormel per
plant but lower yield and lower cormel size smaller plant size. According to the bi-plot
diagram two genotypes (63 and 39) were found in opposite direction (nearly 180⁰).
Therefore, it is possible to ascertain that they had distinct variability between them and
crossing of these genotypes may advance to improve yield and dry matter content.

Regardless of the geographic sources, closeness of genotypes like 39 and 52; 47 and 41;
20, 26 and 29; 6, and 18; 40 and 35 showed strong similarities between them indicating
that genotypes with different names, could be shared the same parental background. So,
further study employing molecular tools may be needed to arrive at a good conclusion
about their relatedness. On the other hand weak similarities between genotypes 44, 27 and
10; 64, 43, 16, 7, 25, 42, 23, and 42; 53, 57, 59, 14, 13, 50 and 15; 4, 21, 32, 58, and 24;
48, 2 and 49 are depicted by the relative distance among the genotypes, is a measure of the
existence of exploitable variability existing among the genotypes. Genotypes such as 31,
45, 19, 11, 63, 62, 54, 46, 1, 38, 33 and 8 are completely distinct from others. Such a wide
variability gives good opportunity for the exploitation of the benefits of heterosis after
good parent selection.

62
 41, 47
52,39

40, 35

6, 18
20, 26, 29

Figure 4 principal components genotype by traits Bi-plot diagram of PCA 1 and PCA 2

63
Generally, the bi-plot showed wide variability between most of the genotypes, though few
genotypes exhibited strong similarities. Therefore, bi-plots provide more useful
information for breeders and many researchers had used to compare different genotypes,
e.g., Afuape et al. (2011) in sweet potato, Ahmadizadeh and Felenji (2011) in potato,
Tabrizi et al. (2011) in Sunflower hybrids in Turkey, Farshadfar et al. (2011) bread wheat
genotypes in Iran, Ngeve et al. (2005) in cassava genotypes in Nigeria.

In general, the presence of trait diversity among the tannia genotypes in the current study
suggests that there is ample opportunity for genetic improvement though selection directly
from the genotypes and selection of diverse parents for hybridization program and
conservation of germplasm for future utilization.

4.5. Diversity of genotypes for qualitative characterss

4.5.1. Leaf Characteristics

The Diversity of qualitative characteristics for different phenotypic classes is presented in

Table 13 also figure 5 presented some diffrences in leaf characterstics among tannia
genotypes. Genotypes showed predominantly cup-shaped lamina position (87.5%), the rest
7.8% were comprised of upward pointing (‘erect’) and 4.7% pointing downward
(‘droopy’). 71% of the genotypes showed light green color and 28% medium green when
observed from lower side of the leaf while 73.4% were dark green, 23.4% were medium
green and 3.1% were light green from upper side of the leaf. Three dominant color of vein
across the genotypes were 45.3% lighter green, 32.8% had same color as the lamina and
20.3% had dark green than lamina from the lower leaf side of leaf. While, on upper portion
of leaf lighter green vein color (92.2%) was dominated. With respect to leaf margin only
one genotypes (AAGT030) showed entire smooth leaf margin and the rest were entire
undulated type. Most of genotypes had pale yellow to creamy (68.88) color of vein and
purple to red edge color (21.9%). This is in agreement with Mbouobda et al. (2007)
reported variability of tannia leaf interms leaf margin colors i.e. very dark green, dark
green, light green and purple green and also variation in color of vein, color of upper and
lower leaf surface in Cameroon.

4.5.2. Petiole Characteristics

Among tannia genotypes there was difference in color of petiole on the lower and upper
portion (Figure 6): 57.8% of the genotypes had green staked petiole with red/purple color

64
on the lower 1/3 of the petiole while 53.1% of the genotypes had light green color of
petiole on the upper 2/3 of petiole. Similarly 78.1% of genotypes had darker green color
than rest of petiole and sheath, 12.5% of them had pink to purple and 9.4% had lighter
green color than rest of petiole and sheath. This is in harmony with Mbouobda et al.
(2007) who found variability of tannia for color of petiole. Also, Opoku-Agyeman et al.
(2004) reported a wide range of diversity of tannia genotypes in Petiole sheath color,
Petiole color lower 1/3, leaf margin color and color of leaf surface.

Table 13. Qualitative characteristics phenotypic classes descriptions and frequency

distribution

Sr. Number of
Traits Descriptions percentage
No Acc.
Entire, smooth 1 1.6
1 Leaf margin
Entire, undulate 63 98.4
Clear edge 6 9.4
2 Leaf margin color Purple/red edge 14 21.9
Pale yellow/creamy 44 68.8
In one plane-apex pointing
5 7.8
upward(‘erect’)
Leaf position/ In one plane-apex pointing
3 3 4.7
orientation downward(‘droopy’)
Three dimensional (‘cup-shaped’) 56 87.5
Color of lower leaf Light green 46 71.9
4
surface Medium green 18 28.1
Light green 2 3.1
Color of upper leaf
5 Medium green 15 23.4
surface
Dark green 47 73.4
Same colour as lamina 21 32.8
Color of vein lower Lighter green than lamina 29 45.3
6
leaf surface Darker green than lamina 13 20.3
Red/ purple 1 1.6

Color of vein on upper Same colour as lamina 5 7.8

7
leaf surface Lighter green than lamina, 59 92.2

8 Leaf variegation Absent 64 100

Light green 34 53.1

Petiole color (upper
9 Green 29 45.3
2/3)
Red/purple 1 1.6

65
Table 13. Qualitative characteristics (continues)

Sr. Number of
Traits Descriptions percentage
No Acc.
Green 6 9.4
10 Petiole color (lower 1/3) Red/purple 21 32.8
Green streaked with red/ purple 37 57.8
Lighter green than rest of
6 9.4
petiole and sheath
Color of edge of petiole Darker green than rest of
11 50 78.1
sheath petiole and sheath
Pink/red/purple 8 12.5
12 petiole attachment No peltate 64 100
White 9 14.1
Interior color of above Yellow 32 50
13
ground stem
Orange 23 35.9
14 Growth habit Acaulescent 64 100
15 Stolen formation Absent 64 100
White 36 56.3
16 Color of cormel apex
Pink/red 28 43.8
Exterior color of Light or medium brown 40 62.5
17
cormels Dark brown 24 37.5
18 Position of cormel apex Under ground 64 100
Globulose 4 6.3
Ovate 19 29.7
19 Shape of cormels Cylindrical 3 4.7
Elliptical 29 45.3
Mixed (state which of the
9 14.1
above)
Smooth 31 48.4
20 Cormel surface texture
Fibrous rough 33 51.6
White 14 21.9
Yellow 17 26.6
21 Interior color of cormel
Orange 6 9.4
Pink/pale red 27 42.2

66
 a b c

d e f

Figure 5 Lamina characteristics among tannia genotypes: Acaulescent growth habit with cup shaped
lamina (a), Acaulescent growth habit with droopy lamina (b), purplish color of lower side vein (c) light green
vein than lamina color (d), smooth leaf margin (e), undulated type of leaf margin (f)

a b c

d e e

Figure 6 petiole characteristics purple color of petiole (a), green petiole and light green edge of petiole (b),
green staked with purplish edge of petiole (c) pink to purple colored petiole edge (d), green staked petiole
with red to purple petiole edge (e)

67
4.5.3. Cormel Characteristics

In this study, it was observed that cormel exterior color of tannia varied from dark brown
(37.5%) to light or medium brown color (62.5%). Similarly interior color of cormel varied
from 42.2% pink to pale red, 9.4% with orange color, 26.6% with yellow color and 21.9%
white color. Also 45.3% genotypes had elliptical, 29.7% ovate, 6.3% globulose, 4.7%
cylindrical and the rest 14.1% had mixed shape of the cormel. In terms of texture 51.6%
had rough and 48.4% had smooth surface texture. This result is in agreements with
(kay,1987; Ramesh et al., 2007). Also Opoku-Agyeman et al. (2004) reported that there is
a wide range of variability of cormels in terms of color of cormel apex, exterior surface of
cormel, internal color of cormels, external surface color of cormels and shape of cormels
among 78 tannia accessions in Ghana. Similarly, Mbouobda et al. (2007) in Cameroon
observed red, white and yellow tuber flesh color of tannia.

4.5.4. Shannon-Weaver diversity index (H`)

Analysis of Shannon-Weaver diversity index (H`) as the measure of phenotypic diversity

for 21 qualitative morphological characters, revealed high diversity in tannia qualitative
morphological traits (Table 14). The lowest and the highest diversity index were recorded
for leaf margin (0.116) and cormel surface texture (0.99) respectively. As Jamago (2000)
that cited by Islam et al. (2012) the Shannon-Weaver Diversity Index (H´) classified as
high when H´ ≥ 0.75, moderate when H´ =0.50-0.75 and low when H´ H'<0.50.

Hence, color of lower leaf surface (0.857), color of vein lower leaf surface (0.803), petiole
color lower 1/3 (0.823), interior color of above ground stem (0.901), color of cormel apex
(0.989), exterior color of cormel (0.954), shape of cormels (0.815), cormel surface texture
(0.999) and interior color of cormel (0.917) showed higher H' values (>0.75). Leaf margin
color (0.739), leaf position/ orientation (0.663), color of upper leaf surface (0.614), petiole
color of upper 2/3 (0.692) and color of edge of petiole sheath (0.614) had moderate
diversity Low diversity index values was observed for leaf margin (0.116) and color of
vein on upper leaf surface (0.396). This wide diversity of tannia in color of lamina, cormel
shape, cormel color, cormel apex color, is in support of (Lebot, 2009; Quero-Garcia et al.,
2010; George, 2011). Also Yada et al. (2010) found diversity index ranged from 0.10 to
0.99 on Sweet potato using 40 morphological traits in Uganda. Opoku-Agyeman et al.
(2004) similarly report a diversity index ranging from 0.118 to 0.540 for tannia qualitative

68
characters. Monomorphism was found for characters such as position of cormel apex,
stolen formation, growth habit, petiole attachment and leaf variegation. A low H' estimate
(nearer to zero) indicates a low level of diversity and unevenness in the distribution and
vice versa (Hennink and Zevan, 1991).

Generally, from this study the level of morphological variation for the these 21 traits
estimated diversity index (H') ranged from 0.116 to 0.99, showed presence of diversity in
most of the traits in tannia germplasm and this may offer good chances for improving the
crop and use these traits for germplasm management.

Table 14. Shannon-Weaver Diversity Index of the 21 qualitative characters

Sr. No Traits H'

1 Leaf margin 0.116
2 Leaf margin color 0.739
3 Leaf position/ orientation 0.663
4 Color of lower leaf surface 0.857
5 Color of upper leaf surface 0.614
6 Color of vein lower leaf surface 0.803
7 Color of vein on upper leaf surface 0.396
8 Leaf variegation 0
9 Petiole color (upper 2/3) 0.692
10 Petiole color (lower 1/3) 0.823
11 Color of edge of petiole sheath 0.614
12 petiole attachment 0
13 Interior color of above ground stem 0.901
14 Growth habit 0
15 Stolen formation 0
16 Color of cormel apex 0.989
17 Exterior color of cormels 0.954
18 Position of cormel apex 0
19 Shape of cormels 0.815
20 Cormel surface texture 0.999
21 Interior color of cormel 0.917

69
 5. SUMMARY AND CONCLUSION

The objectives of the present study were to characterize tannia genotypes based on key
agro-morphological characters, determining the extent of genetic variability and the
genetic relationship among genotypes based on quantitative characters. To achieve these
objectives 64 tannia (Xanthosoma sagittifolium L.( Schott)) Genotypes were studied in
simple lattice design based on 16 quantitative and 21 qualitative morphological characters
at Jimma Agricultural Research Center during the year 2013/14.

ANOVA indicated that there was a highly significant difference among the genotypes for
the quantitative traits studied, indicating the existence of variability among the tested
genotypes. The range and mean performance showed the presence of considerable amount
of variability among the genotypes. For instance, cormel fresh weight varied from 0.27 to
0.74 kg/plant, corm yield varied from 0.13 to 0.43 kg/plant, total tuber yield varied from
0.44 to 1.1 kg/plant, and corm and cormel dry matter content ranges from 21.5 to 32.5 %
and 20 to 32.5%, respectively.

The phenotypic coefficient of variation (PCV) was found to be higher than genotypic
coefficients of variation (GCV) for all characters, indicating that the apparent variation in
the tannia genotypes was not only due to the genotypic effect but also due to
environmental influences. However, variations between PCV and GCV were relatively
small which showed that the environmental variance had the lowest share. PCV ranged
from 8.09 (lamina length) to 28.64 (sucker per plant). Likewise, GCV ranged from 4.98
(cormel diameter) to 23.48 (number of sucker per plant). Number of sucker per plant,
number of cormels per plant, total yield per plant, corm and cormel fresh weight per plant
showed higher PCV along with moderate to high GCV, indicating that these characters
could be important selection indices for future yield improvement schemes in tannia. High
broad sense heritability was noted in the range of 22.37 (corm diameter) to 67.18 (number
of sucker). High heritability coupled with high to moderate genetic advance as percent of
the mean was obtained for lamina length, lamina width, plant canopy, petiole length, plant
height, number of suckers per plant, number of cormels per plant, corm fresh weight per
plant, cormel fresh weight per plant and total yield per plant suggesting that selection for
these traits would be effective to improve yield of tannia.

70
The present study also showed that majority of the characters was positively correlated
with each other both at phenotypic and genotypic level. The maximum significant positive
correlation was between total yield per plant and cormel fresh weight per plant both at
phenotypic level and genotypic (rp=0.957; rg=0.954, respectively). Total yield per plant
was positively and significantly correlated with corm fresh weight per plant and cormel
fresh weight per plant. However corm and cormel dry matter showed non significant
negative but weak association with most characters there for independent selection could
be used.

The path coefficient analysis has shown that the maximum direct positive effect on tannia
yield per plant was exerted by cormel fresh weight per plant (0.55) followed by corm fresh
weight per plant (0.51). Also, plant canopy diameter, cormel length, plant height and corm
dry matter content had also exerted positive direct effect on tannia yield. However, petiole
length, number of cormel, cormel dry matter content, lamina width, and number of sucker
per plant had negative direct effect on yield. Positive correlation with yield of tannia, high
heritability coupled with high to moderate genetic advance as percent of the mean,
moreover, positive direct and indirect effect of plant height, plant canopy, corm and cormel
length, cormel fresh weight and corm fresh weight indicates that these traits could be
effective for selection and future yield improvement of tannia.

Cluster analysis grouped of 64 tannia genotypes into five major clusters as depicted clearly
by the dendrogram. Divergence analysis showed the presence of significant genetic
distances among the five clusters except the distance between I and II which was not
significant. The maximum distance was between cluster II and V (D2 = 155.04), followed
by cluster I and cluster V (D2 = 115.82), IV and V (D2 = 108.62). High genetic divergence
among the tannia genotypes indicates the possibility for development of better performing
varieties by selecting parents having high mean values for the characters of interest,
crossing of divergent clusters.

PCA revealed result four principal components which explained 70.5% of the variation
present among the genotypes. PCA one and two explained 53% of the cumulative
variability. PCA one which is loaded by more than five characters had highest eigen value
and accounted 40.3% of total variation hence, It can be considered as a representative
component. Principal component bi-plot drawn from the first and second components
showed variability between genotypes, the relationship of characters as well as the
71
association of genotypes with those characters studied. Genotypes with larger PCA1 and
lower PCA2 scores gave high yields and genotypes with lower PCA1 and PCA2 scores had
lower yields.

Shannon-Weaver diversity index (H`) analysis was performed for 21 qualitative characters
as measure of phenotypic diversity. Cormel surface texture (H’ = 0.99) and Leaf margin
(H’ = 0.116) were the maximum and the minimum H’ indexes, respectively. Most of the
qualitative characters showed a higher diversity index except for leaf margin and color of
vein on upper leaf surface which showed minimal H’ values. Also, monomorphism was
found for position of cormel apex, stolen formation, growth habit, petiole attachment and
leaf variegation. On the other hand 87.5% of genotypes had cup-shaped lamina, 71 % and
73.4 % of the genotypes showed light green and dark green color from lower and upper
side of the leaf respectively. 62.5% of the genotypes had light brown exterior cormel color,
26.6% of the genotypes showed yellow and 21.9% white interior cormel color and 45.3%
of genotypes had elliptical cormel shape.

In general, the study has shown that there is a wide genetic variability and diversity
between genotypes of tannia for further utilization in tannia improvement program.
However, the result of the present investigation may vary with location and season since
this study was conducted in single environment for one season. That means, the available
genotypes should be further studied with due emphasis on quantitative characters required
to determine further variations and observe the presence and magnitude of genotype-
environment interaction. Furthermore, the presence of morphological variation between
genotypes is not a guarantee for high genetic variation. Hence, molecular or biochemical
studies need to be considered as complementary to this study. Since simple selection of
superior types among the existing genotypes could result in identification of promising
lines, tannia accessions from other growing areas of Ethiopia need to be collected, so as to
broaden the base of existing breeding program. In addition way of inducing flowering and
seed propagation, Calcium oxalate content of corm and cormel, effect of time of planting
and harvesting also effect of type of planting material on yield and dry matter content
should be considered as future line work.

72
 REFERENCES

Acquaah, G., 2007. Principles of plant genetics and breeding, UK: Blackwell Publishing
Ltd. UK. pp. 89.

Acquaah, G., 2012. Principles of plant genetics and breeding. Second (edi.), by John Wiley
& Sons, Ltd. UK. pp.8.

Adelekan, B.A., 2012. An evaluation of the global potential of cocoyam (Colocasia and
Xanthosoma species) as an Energy Crop. British Journal of Applied Science &
Technology, 2(1): 1-15, 2012, 2(1) pp.1–15.

Afuape, S.O., P.I. Okocha, & D. Njoku, 2011. Multivariate assessment of the agromo-
rphological variability and yield components among sweet potato (Ipomoea batatas
( L .) Lam) landraces. African Journal of plant sciences, 5(2), pp.123–132.

Agueguia, A., 1993. Association of metric traits and path analysis in cocoyam,
Xanthosoma sagiitifolium (L.) schott. The Indian Journal of Genetics and Plant
Breeding, 53(3), pp.287–291. Available at:
http://www.indianjournals.com/ijor.aspx?target=ijor:ijgpb&volume=53&issue=3&a
rticle=012 [Accessed February 15, 2014].

Ahmad M., G. Zaffar, S.D. Mir, Z.A. Dar, S.H. Dar, S. Iqbal, S.A. Bukhari, G.H. Khan &
A. Gazal, 2013. Estimation of correlation coefficient in Oats (Avena sativa L.) for
forage yield, grain yield and their contributing traits. International Journal of Plant
Breeding and genetics, 7(3), pp.188–191. Available at:
http://docsdrive.com/pdfs/academicjournals/ijpbg/0000/51766-51766.pdf
[Accessed February 15, 2014].

Ahmadizadeh, M. & H. Felenji, 2011. Evaluating diversity among potato cultivars using
agro-morphological and yield components in fall cultivation of Jiroft area. Amer.
Eurasian J. Agric. Environ. Sci, 11(5), pp.655–662. Available at:
http://idosi.org/aejaes/jaes11(5)11/9.pdf [Accessed May 15, 2014].

Akbar, M., T. Mahmood, M. Yaqub, M. Anwar, M. Ali and N. Iqbal, 2003. Variability,
correlation and path coefficient studies in summer Mustard (Brassica juncea L.).
Asian J Pl. Sci., 2, 696-698.

Allard, R.W. 1960. Principles of plant breeding. John Willey and Sons Inc., New York,
London. p.485.

Allard, R.W., 1999. Principle of plant breeding. Jhon willeyand Sons, Inc, New York.
254p.

Amsalu N., Weyessa G., Assefa T., Wubishet A., Asefa k. and Edossa E., 2008. Other root
and tuber crops. pp 301-326. In: Gebremedhin et al., (eds.), Root and tuber crops,
the untapped resources, by Ethiopian Institute of Agricultural Research, EIAR,
Addis Abeba, Ethiopian.

73
Amsalu, N. and Tesfaye A., 2006. Exploration and collection of root and tuber crops in
south western Ethiopia: its implication for conservation and research. Pp 84-88. In:
Asfaw et al (eds.), proceedings of the eleventh conference of the crop science
society of Ethiopia, 26-28 April 2004, Addis Ababa, Ethiopia.

Aremu, C.O., 2011. Genetic diversity: A review for need and measurements for intraspecie
crop improvement. J. Microbiol. Biotech. Res, 1(2), pp.80–85. Available at:
http://scholarsresearchlibrary.com/JMB-vol1-iss2/JMB-2011-1-2-80-85.pdf
[Accessed February 19, 2014].

Aremu, C.O., 2012. Exploring statistical tools in measuring genetic diversity for crop
improvement. pp. 339–348. In M. Caliskan, (ed.) Genetic diversity in plants.
InTech,. Available at: http://www.intechopen.com/books/genetic-diversity-in-
plants/exploring-statistical-tools-in-measuring-genetic-diversity-for-crop-
improvement [Accessed February 19, 2014].

Asfaw K., 2005. Characterization and divergence analysis of some Ethiopian Taro
(Colocasia esculenta (L.) Schott.) Accessions. MSc thesis presented to the school
of graduate studies of Alemaya University.

Asha, K. & Nair, C.M., 2003. Characterisation and evaluation of an indigenous collection
of Greater Yam (Dioscorea alata L.). Indian Journal of Plant Genetic Resources,
16(1), pp.13–17. Available at:
http://www.indianjournals.com/ijor.aspx?target=ijor:ijpgr&volume=16&issue=1&a
rticle=005 [Accessed April 6, 2014].

Asumadu H., E.L. Omeny & F.Tetteh, 2011. Physiological and economical implication of
leaf harvesting on vegetative growth and cormel yield of Cocoyam (Xanthosoma
sagittifolium). Journal of Agronomy, 10(4), pp.112–117.

Atnafua B., Ermias T., Endale T., Mikias Y., Mulugeta D., And Yohannes T., 2008. Enset
Variety Development. pp 156-193. In: Gebremedhin et al. (eds.), Root and tuber
crops, the untapped resources, by Ethiopian Institute of Agricultural Research,
EIAR, Addis Abeba, Ethiopian.

Avtar R., Singh D., Thakral N.K., Singh Amit, Sangwan O., Rani Babita, Kumari Nisha,
2014. Multivariate analysis for evaluation and classification of toria germplasm
accessions. Research on Crops, 15(1), pp.129–134. Available at:
http://www.indianjournals.com/ijor.aspx?target=ijor:rcr&volume=15&issue=1&arti
cle=018 [Accessed April 13, 2014].

Beeching, J., Marmey, P., Garada, M., Nairot, M., Haysom, H., Hughes, M., and Charrier,
A., 1993. An assessment of genetic diversity within collections of Cassava
(Manihot esculenta Crantz) germplasm using molecular markers. Annals of Botany
72: 515-520.

Bello O.B., S.Y. Abdulmaliq, M.S. Afolabi, S.A. Ige, 2010. Correlation and path
coefficient analysis of yield and agronomic characters among open pollinated maize
varieties and their F 1 hybrids in a diallel cross. African Journal of Biotechnology,
9(18), pp.2633–2639.

74
Biabani A.R., Pakniyat H., 2008. Evaluation of seed yield-related characters in sesame
(Sesamum indicum L.) using Factor and Path Analysis.Pakistan Journal of
Biological Sciences 11:1157-1160.

Bisne, R., Sarawgi, A. & Verulkar, S., 2009. Study of heritability, genetic advance and
variability for yield contributing characters in rice. Bangladesh Journal of Agril.
Res., 34(2), pp.175–179. Available at:
http://www.banglajol.info/index.php/BJAR/article/viewArticle/5788 [Accessed
February 13, 2014].

Blay, E.T., S.K. Offei, And E.Y. Danquah, 2004. Genetic diversity in cocoyam as
revealed by Random Amplified Polymorphic DNA (RAPD) markers. In Proceedings
of a final Research Coordination Meeting organized by the Joint FAO/IAEA Division
of Nuclear Techniques in Food and Agriculture and held in Pretoria, South Africa,
19–23 May 2003. pp. 131–142.

Bown, D. 2000. Aroids. Plants of the Arum family. 2nd (ed.). Timber Press. Portland,
Oregon, USA. pp. 392.

Burhan, A., 2007. Relationships among yield and some yield characters in Potato (S.
tuberosum L.). Journal of Biological Sciences, 7(6), pp.973–976. Available at:
http://docsdrive.com/pdfs/ansinet/jbs/2007/973-976.pdf [Accessed April 6, 2014].

Burton, G.W. and Devane, E.H. 1953. Estimation of heritability in tall Festuca (Festuca
arundinacea) from replicated clonal material. Agron. J., 45: 478-481.

Castro G.R., Nyman, M. & Rönnberg-Wästljung, A. C., 2005. Agronomic performance of

three cocoyam (Xanthosoma violaceum Schott) genotypes grown in Nicaragua.
Euphytica, 142(3),pp.265–272. Available at:
http://link.springer.com/10.1007/s10681-005-2147-5 [Accessed November 12,
2013].

Ceballos, 2009. Root and tuber crops for feed and industry. 15th Symposium of
International Society for Tropical Root Crops (ISTRC), pp.1–11. Available at:
http://www.istrc.org/sites/default/files/files/symposium/Internal/2009_Peru/s9_ceba
llos.pdf [Accessed June 3, 2013].

Cheema DS, Singh H, Dhatt AS, Sidhu AS, and Garg (2006). Studies on genetic variability
and correlation for yield and quality traits in Arvi [Colocasia esculenta (L.) Schott].
In I International Conference on Indigenous Vegetables and Legumes. Prospectus
for Fighting Poverty, Hunger and Malnutrition 752, pp. 255-260.

Choudhary, V.K. P. Suresh Kumar, James George, M. Kanwat & R. Saravanan, 2013.
Genetic variability and character association in Taro (Colocasia esculenta (L.)
Schott.) under mid-hills of Arunachal Pradesh. Journal of root crops, 37(2), p.155.
Available at: http://isrc.in/ojs/index.php/jrc/article/view/39 [Accessed April 5,
2014].

Cooper, M. & Delacy, I.H., 1994. Relationships among analytical methods used to study
genotypic variation and genotype-by-environment interaction in plant breeding
multi-environment experiments. Theoretical and applied genetics, 88(5), pp.561–
75
 72. Available at: http://www.ncbi.nlm.nih.gov/pubmed/24186111 [Accessed
February 19, 2014].

Costello, A.B. & Osborne, J.W., 2005. Best practices in exploratory factor analysis: four
recommendations for getting the most from your analysis. Practical assessment
research & evaluation, 10(7), pp.1–10. Available at:
http://pareonline.net/getvn.asp?v=10&n=7 [Accessed April 16, 2014].

Crossa, J., Delacy, I.H. & Taba, S. 1995. The use of multivariate methods in developing a
core collection. pp. 77-79. In Hodgkin, J., Brown A.H.D., Van Hintum, TH.J.L. &
Morales, E.A.V(eds.) Core Collections of Plant Genetics Resources, John Willey &
Sons, UK.

CSA (The Federal Democratic Republic of Ethiopia Central, Statistical Agency), 2012a.
Agricultural sample survey on crop and livestock product utilization. Volume VII
Addis Abeba, Ethiopia. pp.11-97.

CSA (The Federal Democratic Republic of Ethiopia Central, Statistical Agency), 2013.
Agricultural sample survey on crop and livestock product utilization. Addis Abeba,
Ethiopia. pp.15-94. Available at:
http://www.csa.gov.et/images/general/news/2005%20crop%20prod%20report%20_
with%20segen.pdf).

CSA (The Federal Democratic Republic of Ethiopia Central, Statistical Agency), 2012b.
Agricultural sample survey report on land utilization (private peasant holdings,
meher season). volume IV Addis Abeba, Ethiopia. pp. 92 -169.

Danquah, A., S.K Offi, T. Blay, Eric Asare and E.Y. Danquah, 2006. Characterization of a
mutant population of Cocoyam (Xanthosoma sagittifolium L. Schott). International
Journal of Botany, 2(2), pp.128–132. Available at:
http://www.docsdrive.com/pdfs/ansinet/ijb/2006/128-132.pdf [Accessed June 30,
2014].

Deshmukh, S.N. Basu, M.S. and P.S. Reddy. 1986. Genetic variability, character
association and path coefficients of quantitative traits in Virginia bunch varieties of
groundnut. Indian J. Agri. Sci., 56:816-812.

Dewey, R.D. and K.H. Lu, 1959. A correlation and path coefficient analysis of components
of crested wheat grass seed production. Agron. J. (52), pp. 515–518.

Eddy, N. O., Emmanuel Essien, Eno E. Ebenso & Richard A. Ukpe, 2012. Industrial
potential of two varieties of Cocoyam in bread making. E-Journal of Chemistry,
9(1), pp.451–464.

Edossa E., 1996. Root and Tuber Crops: Potential as food crops in the humid areas of
Ethiopia. pp In: IAR News let- ter of Agric. Res. 2 (1). IAR, A.A, Ethiopia.

El-Mohsen, A.A., S. R. Abo Hegazy & M.H. Taha 2012. Genotypic and phenotypic
interrelationships among yield and yield components in Egyptian bread wheat
genotypes. Journal of Plant Breeding and crop science, 4(1), pp.9–16.

76
Facciola S.A. 1998. Source book of edible plants. In: Cornucopia II. Vista, Calif.:
Kampong Publications. pp. 23–5, 80–100.

Falconer, D. S., 1983 Problems on quantitative genetics. Longmans, London/New York.

Falconer, D.S. 1989. Introduction to quantitative Genetics, 3rd ed. Longman scientific and
technical. John Wiley and Sons, New York, 438p.

Falconer, D.S. and Trudy F.C. Mackay. 1996. Introduction to Quantitative Genetics. 4th
ed., Longman Group Limited, Malaysia. 464p.

FAO (Food and Agriculture Organization of the United Nations), 2010. The state of food
insecurity in the world addressing food insecurity in protracted crises.Rome, Italy.
pp 8-11.

FAO (Food and Agriculture Organization of the United Nations), 2013. The state of food
insecurity in the world, the multiple dimensions of food security. 2013, Rome,
Italy. pp. 8-15.

FAOSTAT, 2012 Available at:

http://faostat.fao.org/site/567/DesktopDefault.aspx?PageID=567#ancor [Accessed
May 22, 2014].

Farshadfar, E., Jamshidi, B. & Aghaee, M., 2012. Biplot analysis of drought tolerance
indicators in bread wheat lanraces of Iran. inernational Journal of Agriculture and
crop science, 4(5), pp.226–233.

Fekadu, A., Petros, Y. & Zelleke, H., 2013. Genetic variability and association between
agronomic characters in some potato (Solanum tuberosum L.) genotypes in
SNNPRS, Ethiopia. International Journal, 5(8), pp.523–528.

Felenji, H., S. Aharizad, G.R.Afsharmanesh and M. Ahmadizadeh, 2011. Evaluating

correlation and factor analysis of morphological traits in Potato cultivars in fall
cultivation of Jiroft area. American-Eurasian J. Agric. & Environ. Sci. , 11(5),
pp.679–684. Available at: http://www.idosi.org/aejaes/jaes11(5)11/12.pdf
[Accessed April 13, 2014].

Franco, J., J. Crossa, J.M. Ribaut, J. Betran, M.L. Warburton, & M. Khairallah. 2001. A
method for combining molecular markers and phenotypic attributes for classifying
plant genotypes. Theor. Appl. Genet.103: 944-952.

Gebremedhin, W., Endale, G. and Berga, L., 2008. potato variety development pp 15-32.
In: Gebremedhin et al., (eds.), Root and tuber crops, the untapped resources, by
Ethiopian Institute of Agricultural Research, EIAR, Addis Abeba, Ethiopian.

George, R.A.T., 2011. Tropical Vegetable Production. CAB International, UK. pp. 162 -
175.

Getachew, W., Sentayehu A., Taye K. and Tadesse B., 2013. Genetic diversity analysis of
some Ethiopian specialty Coffee (Coffea arabica L.) germplasm accessions based
on morphological traits. Time Journals of Agriculture and Veterinary Sciences, 1
(4): 47-54.
77
Ghaffari, M., Farrokhi, I., Mirzapour, M. 2011. Combining ability and gene action for
agronomic traits and oil content in sunflower (Helianthus annuus L.) using F1
hybrids. Crop breeding journal 1 (1): 75-87.

Giacometti, D.C. and León, J. 1994.Tannia, Yautia (Xanthosoma sagittifolium). pp. 253-
259. In: Neglected Crops: 1492 from a different perspective. Plant Production and
Protection Series No. 26. FAO, Rome, Italy. Available at:
http://www.fao.org/docrep/t0646e/t0646e0o.htm [Accessed April 28, 2013].

Hammer, O., D.A.T. Harper and P.D. Ryan, 2001. PAST (Paleontological Statistics
Software Package) for education and data analysis. Palaeontologia Electronica, 4:
1-9.

Hennink, S. and A.C. Zeven. 1991. The interpretation of Nei and Shannon- Weaver within
population variation indices. Euphytica 51: 235-240.

IBPGR ,1989. Descriptors for Xanthosoma. International Board for Plant Genetic
Resources, Rome, Italy.

Idahosa, D.O., Alika, J.E. & Omoregie, A.U., 2010. Genetic variability, heritability and
expected genetic advance as indices for yield and yield components selection in
Cowpea (Vigna unguiculata ( L .) Walp. Academia Arena, 2(5), pp.22–26.

Iezzoni, A. & Pritts, M., 1991. Applications of principal component analysis to

horticultural research. Hort.Science, 26(4), pp.334–338. Available at:
http://hortsci.ashspublications.org/content/26/4/334.full.pdf [Accessed April 7,
2014].

Islam, M. & Chowdhury, R., Afroz, R., Rahman, S. and Mamtazul H. MD., 2012.
Characterization and maintenance of yam (Dioscorea spp.) germplasm. Bangladesh
Journal of Agril. Res., 36(4), pp.605–621.

Iwuoha C.I. & Kalu F.A., 1995. Calcium oxalate and physico-chemical properties of
Cocoyam (Colocasia esculenta and Xanthosoma sagittifolium) tuber flours as
affected by processing. Food chemistry, 54(1), pp.61–66.

Jamago, J.M.2000. Morpho-agronomic and molecular diversity of the Philippine

mungbean (Vigna radiata L.) germplasm. Master of Science Thesis, University of
thePhilippines Los Banos, The Philippines.p.174.

Johonson, H.W., H.F. Robinson and Comostock R.E., 1955a. Estimates of genetics and
environmental variability in Soybean. Agronomy Journal, 47, 314-318.

Johonson, H.W., Robinson, H.F. and Comstock, R.E., 1955b. Genotypic and phenotypic
correlations in soybeans and their implication in selection. Agronomy Journal, 47:
477-483.

Kay, D.E. 1987. Root crops, 2nd (ed.). Crop and product digest number:2. Tropical
International and Research Institute, London. pp.380.

78
Khayatnezhad, M. & Shahriari, R., Gholamin, R., Jamaati-e-Somarin, S., and Zabihi-e-
Mahmoodabad, R., 2011. Correlation and path analysis between yield and yield
components in Potato (Solanum tubersum L.). Middle-east journal of scientific
research, 7(1), pp.17–21. Available at: http://idosi.org/mejsr/mejsr7(1)11/4.pdf
[Accessed May 14, 2014].

Lebot, V., 2009. Aroids. pp. 279–355. In Tropical Root and Tuber Crops: Cassava, Sweet
Potato, Yams and Aroids; Crop Production Science in Horticulture Series No 17.
CAB International, UK.

Lima, T.D.A., O. L. Franco, E.G. Goncalves, F. Borghetti, 2010. Evaluation of nutritional

and anti-nutritional compounds from tania (Xanthosoma sagittifolium Schott)
corms. Nutrition & Food Science, 40(4), pp.419–428.

Mahalanobis P.C., 1936. On the generalized distance in statistics. Proc. Natl. Sci. India B.
2: 49-55.

Mbouobda, H.D., T. Boudjeko, P.F. djocgoue, T.J.J. Tsafack & D.N. Omokolo, 2007.
Morpholofical characterization and agronomic evaluation of Cocoyam
(Xanthosoma sagittifolium ( L ) ( Schott ) germplasm in Cameroon. Journal of
Biological Science, 7(1), pp.27–33.

Minitab. 2010. Minitab® statistical software, version 16. Minitab Inc., State college, PA,
USA.

Mittal, R.K. & Sethi, G.S., 2005. Genetic variability in triticale x bread wheat derivatives
under normal and phosphorus stress regimes. Journal of environmental biology,
India, 26(1), pp.105–7. Available at:
http://www.ncbi.nlm.nih.gov/pubmed/16114468 [Accessed February 17, 2014].

MOA (Ministry of Agriculture), 2010. Animal and plant healath regulatory Directorate,
Crop Variety Registration,No.13, pp.135-175 Addis Abeba Ethiopia.

Mondal, M.A.A., Hossain, M.M., Rasul, M.G. And Uddin M. S. 2007. Genetic diversity in
potato (Solanum tuberosum L.). Bangladesh Journal of Botany, 36(2), pp.121–125.
Available at: http://www.banglajol.info/bd/index.php/BJB/article/viewArticle/1499
[Accessed May 16, 2014].

Mwenye, O., Labschagne, M.T., Herselman, L., Benesi, I.R.M. & Chipungu, F.P., 2010.
Ethno-botanical and morphological characterisation of Cocoyams (Colocasia
esculenta L. Schott and Xanthosoma sagittifolum L. Schott) germplasm in Malawi.
pp. 193–199. In Research application summary, second ruforum biennial meeting
20 - 24 September 2010. Entebbe, Uganda.

Ndabikunze B. K.,Talwana H. A. L., Mongi R. J., Issa-Zacharia A., Serem A. K., Palapala
V. and Nandi J. O. M.,2011. Proximate and mineral composition of cocoyam
(Colocasia esculenta L. and Xanthosoma sagittifolium L.) grown along the Lake
Victoria Basin in Tanzania and Uganda. African Journal of Food Science, 5(4),
pp.248–254.

79
Ndol, B.A., N.H. Ndulaka & N.U. Ndaeyo, 2003. Stabilization of yield parameters and
some nutrient components in cocoyam cultivars with time in Uyo, Southern
Nigeria. Global J. Agric. Sci., 2, pp.74-78.

Nechifor, B., Filimon, R. & Szilagyi, L., 2011. Genetic variability, heritability and
expected genetic advance as indices for yield and yield components selection in
common Bean (Phaseolus vulgaris L.). Scientific Papers, UASVM Bucharest,
Series A, LIV, pp.332–337.

Ngeve, J.M., A.G. Dixon and E.N. Nukenine, 2005. The influence of host Genotype X
Enviroment interactions on the responce of cassava Anthracnose disease in diverse
agro-ecologies in Nigeria. African crop science journal, 13(1), pp.1–11.

Nishanthini, A. & V.R., Mohan, 2012. Antioxidant activities of Xanthosoma sagittifolium

L. Schott using various in vitro assay models. Asian Pacific Journal of Tropical
Biomedicine,S, pp. 1701–1706.

Norman P.E., Beah A.A., Samba J.A., Tucker M.J., Benya M.T., Fomba S.N. 2014. Agro-
phenotypic characterization of Sweet potato (Ipomoea batatas L .) genotypes using
factor and cluster analyses. Agricultural science research journal, 4(2), pp.30–38.

Ntawuruhunga, P. & Dixon, A., 2010. Quantitative variation and interrelationship between
factors influencing cassava yield. Journal of Applied Biosciences, 26, pp.1594–
1602. Available at: http://m.elewa.org/JABS/2010/26/1.pdf [Accessed May 14,
2014].

Nurmiyati, Sugiyarto & Sajidan 2009. Kimpul ( Xanthosoma spp .) Characterization based
on morphological characteristic and isozymic analysis. Nusantara Bioscience,1(3),
pp.138–145.

Offei, S. K., Asante, I. K. and Danquah E. Y. 2004. Genetic structure of seventy cocoyam
(Xanthosoma sagittifolium , Linn, Schott) accessions in Ghana based on RAPD.
Hereditas 140: 123_/128.

Okpul, T., Singh, D., Gunua, T. & Wagih, M. E., 2004. Assessment of diversity using
agro-morphological traits for selecting a core sample of Papua New Guinea taro
(Colocasia esculenta (L.) Schott) collection. Genetic resources and crop evolution,
51, pp.671–678.

Oliveira D.G., 2012. Species used to prevent osteoporosis in Brazilian traditional medicine.
Pharm Biology, 50, pp.930-932.

Onokpise O.U., Wutoh J.G., Ndzana X., Tambong J.T., Mebeka M.M., Sama A.E.,
Nyochembeng L., Agueguia A., Nzietchueng S., Wilson J.G., Borns M.,1999.
Evaluation of Macabo Cocoyam germplasm in Cameroon. pp 394–396. In: Janick J.
(ed.) Perspectives on news crops and news uses. Ashs Press, Alexandra VA USA.

Onwueme, I.C. & W.B. Charles, 1994. Cultivation of cocoyam. pp. 136–161. In Tropical
Root and Tuber Crops production, perispective and future prospects, FAO Plant
Production and Protection Paper no.126. Rome, Italy.

80
Opoku-Agyeman M., S.O. Benneti-Lartey & C. Markwei, 2004. Agro-morphological and
sensory characterization of cocoyam (Xanthosoma sagittifolium (L) ( Schott )
germplasm in Ghana. Ghana Jnl avric. Sci., 37(1), pp.23–31. Available at:
http://gains.org.gh/articles/gjas_v37_p23_31.pdf.

Ortiz, R., 1996. Statistical analysis of DNA marker characterization of germplasm.

American journal of Bot. 34:40-49.

Pandey G., & V.K. Dobhal., 1997. Multivariate analysis in Taro (Colocasia esculenta L.).
Indian J. Genet, 57(3), pp.262–265.

Pandey, V.S., Ojha, M. & Singh, P., 2009. Genetic variability and correlation studies in
arvi (Colocasia esculenta L. var. Antiquarum). Vegetable Science, 36(3 Suppl),
pp.431–434. Available at:
http://vegsci.isvs.org.in/index.php/vegsci/article/download/757/pdf [Accessed April
6, 2014].

Paul K.K and Bari M.A., 2007. Protocol establishment for micro propagation and vitro
callus regeneration of Maulavi Kachu (Xanthosoma sagittifolium L. Schott ) from
cormel axillari bud meristem. Journal of plant sciences, 2(4) pp.398-406.

Paul, K. & Bari, M., 2011. Studies on direct and indirect effects of different plant
characters on yield of Taro (Colocasia esculenta L. Schott) Var. Antiquorum. The
Agriculturists, 9(1&2), pp.89–98. Available at:
http://www.banglajol.info/index.php/AGRIC/article/view/9483 [Accessed May 31,
2013].

Paul, K., Bari, M. & Debnath, S., 2011. Genetic variability of Colocasia esculenta (L.)
Schott. Bangladesh Journal of Botany, 40(2), pp.185–188. Available at:
http://www.banglajol.info/bd/index.php/BJB/article/view/9775 [Accessed April 11,
2014].

Paul, K., Bari, M. & Debnath, S., 2013. Correlation and path coefficient studies of yield
and yield attributing characters in Panikachu, Colocasia esculenta (L.) Schott.
Journal of Bangladesh Academy of sciences, 37(2), pp.131–137. Available at:
http://www.banglajol.info/index.php/JBAS/article/view/17552 [Accessed April 6,
2014].

Paul, K.K. & Bari, M.A., 2012. Estimates of genetic components for yield and related traits
in Cocoyam. The Agriculturists, a Scientific Journal of Krishi Foundation, 10(2),
pp.127–132.

Paul, K.K. and Bari, M.A., 2013. Genetic variability, correlation and path coefficient
studies in Elephant Foot Yam (Amorphophallus campanulatus Bl .). Journal of
scientific research, 5(2), pp.371–381.

Peeters, J.P. & J.A. Martinelli, 1989. Hierarchical cluster analysis as a tool to manage
variation in germplasm collections. Theor. Appl. Genet. 78: 42–48.

81
Perez E., F.S. Schultz & E.P. Delahaye, 2005. Characterization of some properties of
starches isolated from Xanthosoma sagittifolium (tannia) and Colocassia esculenta
(taro). Carbohydrate Polymers, 60, pp.139–145.

Perez Ponce, J., 2010. Cocoyam. pp. 41–48. In Quality declared planting material protocols
and standards for vegetatively propagated crops, FAO Plant Production and
Protection Paper No195. Rome, Italy,

Perez, E.E., Guti´Errez, M.E., Delahaye, E. P., Tovar, J. & Lares D. M. 2007. Production
and characterization of Xanthosoma sagittifolium and Colocasia esculenta flours.
Journal of food science, 72(6), pp.S367–72. Available at:
http://www.ncbi.nlm.nih.gov/pubmed/17995692 [Accessed December 6, 2013].

Quero-Garcia, J., Anton Ivancic, and Vincent Lebot., 2010. Taro and Cocoyam. pp. 149–
188. In J.E. Bradshaw (ed.). Handbook of plant, breeding, root and tuber crops.
Springer Science + Business Media. New York, USA.

Ramesh, V., K.S. John, C.S. Ravindran, S. Edison, 2007. Agro-techniques and plant
nutrition of tannia (Xanthosoma sp.): an overview. Journal of Root Crops, 33(1),
pp.1–11. Available at:
http://www.isrc.in/ojs/files/journals/5/articles/31/submission/review/31-75-1-
RV.pdf [Accessed November 24, 2013].

Rao, S.A., M.A. Khan, T. Mcneilly & A.A. Khan, 1997. Cause and effect relations of yield
components in rice (Oriza sativa L.). J. Genet. Breed, 51, pp. 1-5.

Ravishanker, S. Kumar, D.K. Baranwal, A. Chatterjee & S.S. Solankey, 2013. Genetic
diversity based on cluster and principal component analyses for yield and quality
attributes in Ginger (Zingiber officinale Roscoe). International Journal of journal
of plant breeding and Genetics, 7(3), pp.159–168.

Regassa D. & N. Basavaraj, 2005. Genetic variability studies in Potato (Solanum

tuberosum L.). Karnataka J. Agric. Sci., 18(1), pp.87–90.

Reyes, G., A. C. Ronnberg-Wa Stljung and M. Nyman, 2006. Comparison of field

performance between dasheen mosaic virus-free and virus-infected in vitro plants of
cocoyam (Xanthosoma spp.) in Nicaragua. Expl Agric., 42, pp.301–310.

Ringnér, M., 2008. What is principal component analysis? Nature biotechnology, 26(3),
pp.303–304. Available at: http://dx.doi.org/10.1038/nbt0308-303 [Accessed
February 16, 2014].

Robertson, A., 1959. The sampling variance of the genetic correlation coefficient.
Biometrics 15:469-485.

Saborio, F., G. Umana, W. Solano, P. Amador, G. Munos, A.T. Valerin, S. Torres, R.

Valverde, 2004. Induction of genetic variation in Xanthosoma SPP.In Proceedings of
a final Research Coordination Meeting organized by the Joint FAO/IAEA Division of
Nuclear Techniques in Food and Agriculture and held in Pretoria, South Africa, 19–
23 May 2003. pp. 143–155.

82
Sanni, K., Fawole, I. & Ogunbayo, A., Tia, D., Somado, E. A., Futakuchi, K., Sié,
M., Nwilene, F. E., and Guei R. G., 2010. Multivariate analysis of diversity of
landrace rice germplasm. Crop Science, 52(2), pp.494–504.

SAS (Statistical analysis system institute), 2008. Statistical analysis system software.
Version 9.2 SAS Institute Inc., cary., NC. USA.

Sedeek, S.E.M., S.A.A. Hammoud, M.H. Ammar & T.F. Metwally, 2009. Genetic
variability, heritability, genetic advance and cluster analysis for some physiological
traits and grain yield and its components in Rice (Oryza Sativa L.). Journal of
Gric.Res. Kafer El-Sheikh Univ., 35(3), pp.858–878.

Sefa-Dedeh, S. & E.K.A. Sackey, 2002. Starch structure and some properties of Cocoyam
(Xanthosoma sagittifolium and Colocasia esculenta) starch and raphides. Food
chem., 79, PP.435-444.

Shannon C.E. and Weaver W. 1949. The mathematical theory of communication. Urbana,
Illinois, USA: University of Illinois Press.

Sharma, J. R. 1998. Statistical and biometrical techniques in plant breeding. New Age
International Publication, New Delhi. pp.432.

Shiyam, J. O., Obiefuna, J. C., Ofoh, M. C., Oko, B. F. D. & Uko, A.E., 2007. Growth and
corm yield response of upland Cocoyam (Xanthosoma sagittifolium L) to Sawdust
Mulch and N P K 20 : 10 : 10 fertilizer rates in the humid forest zone of Nigeria.
Continental J. Agronomy, 1, pp.5–10.

Simmonds, N.W. 1986. Principles of crop improvement. Longman, Singapore. p.495.

Simone A., 1992. Taro root in North Omo. FPR Technical Pamphlet No.2. FARM Africa.
Addis Ababa, Ethiopia.

Singh, B. S. Pandey, A. K. Pal, J. Singh and Mathura Rai, 2005. Correlation and path
coefficient analysis in Asiatic carrot (Daucus carota var. sativa). Vegetable Science,
32(2), pp.136–139. Available at:
http://vegsci.isvs.org.in/index.php/vegsci/article/view/433 [Accessed May 14,
2014].

Singh, B.D. 2001. Plant Breeding. Hindu University, Varanasi, Kalayani Publisher,
Ludhiana, New Delhi.pp. 81.

Singh, R.K, and B.D. Chaudhury. 1987. Biometrical methods in quantitative genetic
analysis. Kalyani Publishers, New Delhi. pp. 318.

Singh, V., Singh, P.K., Kumar, K., Shahi, B.P. and Dwivedi, S.V., 2003. Genetic
variability, heritability and genetic advance for yield and its attributing triats in arvi
(Colocasia esculenta var. antiquorum). Indian Journal of Horticulture, 60(4),
pp.376–380. Available at:

Souza, E. and M.E. Sorrells. 1991. Relationship among 70 North American oat
germplasms: II. Cluster analysis using qualitative characters. Crop Sci. 31: 605-
612.
83
Stoskoph, N.C., D.T. Tomes & B.R. Christie, 1999. Plant breeding: Theory and Practise.
Scientific Publishers, India, pp.435.

Sylva, R.L.C. and Carvalho, A., 1997. Shoot and Root evaluation on seedlings from
Arabica coffee genotypes. Brahantia. 56. pp. 69-68.

Tabachnick, B. G., & Fidell, L. S., 2001. Using multivariate statistics. Boston: Allyn and
Bacon.

Tabrizi, H. Z., E. Şahin And K. Haliloğlu, 2011. Principal components analysis of some F1
Sunflower hybrids at germination and early seedling growth stage. Journal of
Agricultural faculty of Atatürk Univ., 42(2), pp.103–109.

Tewodros Mulualem. 2008. Morphological characterization and preliminary evaluation of

Aerial yam (Dioscorea bulbifera) collected from south and south-western Ethiopia.
MSc thesis, Awassa University, Ethiopia.

Tewodros, M. & Hussien M., 2012. Genetic variability and association among yield and
yield related traits in Aerial Yam / Dioscorea bulbifera (L.)/ Accessions at South-
western Ethiopia. Journal of natural sciences research, 2(9), pp.63–70.

Tewodros, M., 2013a. Morpho−Agronomical characterization of Taro (Colocasia

esculenta) Accessions in Ethiopia. J/ Plant, 1(1), pp.1–9.

Tewodros, M.B., 2013b. Diversity of Aerial Yam /Dioscorea bulbifera (L.) accessions in
Ethiopia based on agronomic traits. Agriculture, Forestry and Fisheries, 2(2),
pp.67.

Tilahun, W., Sentayehu A., Amsalu A., Weyessa G., 2014. Genetic Diversity Analysis
among Anchote (Coccinia abyssinica) Accessions in Western Ethiopia.
International Journal of Agricultural research, 9(3), pp.145–157. Available at:
http://scialert.net/abstract/?doi=ijar.2014.149.157 [Accessed May 13, 2014].

Tsafack, T. J. J., Gilbert P. Charles, A. Hourmant, N. D. Omokolo and M. Branchard

2008. Effect of photoperiod and thermoperiod on microtuberization and carbohydrate
levels in Cocoyam (Xanthosoma sagittifolium L. Schott). Plant Cell, Tissue and
Organ Culture (PCTOC), 96(2), pp.151–159. Available at:
http://link.springer.com/10.1007/s11240-008-9471-0 [Accessed April 26, 2013].

Tsegaye, E., Devakara, S. E.V. and & Dechassa, N., 2007. Genetic variablity for yield and
other agronomic traits in sweet potato. Journal of Agronomy, 6(1), pp.94–99.

Tsegaye, E., Devakara, S.E.V. and Nigussie D., 2006. Correlation and path analysis in
sweet potato and their implications for clonal selection. Journal of Agronomy, 5(3),
pp.391–395.

Tuncturk, M., and Ciftc V., 2005. Selection criteria for potato (Solanum tuberosum L.)
breeding. Asian journal of plant sciences, 4(1), pp.27–30.

Upadhyaya, H., Gowda, C. & Sastry, D., 2008. Plant genetic resources management:
collection, characterization, conservation and utilization. Journal of SAT

84
 Agricultural Research vol.6, pp.1–16. Available at: http://oar.icrisat.org/2744/
[Accessed February 17, 2014].

Upadhyaya, H.D., Devvart Yadav, Naresh Dronavalli, CLL Gowda & Sube Singh, 2010.
Mini core germplasm collections for infusing genetic diversity in plant breeding
programs. Electronic Journal of Plant Breeding, 1(4), pp.1294–1309.

Valenzuela, H.R., Stephen K., Hair, O. & Schaffer, B., 1991. Shading, growth, and dry-
matter partitioning of Cocoyam [ Xanthosoma sagittifolium ( L .) Schott ]. Journal
of American society of horticultural science, 116(6), pp.1117–1121.

Verma P.S and V.K. Agarawal (1982) Genetics. S.Chand and Co.Ltd., Ram Nagar, New
Dehlhi, pp:555.

Wricke, G., Weber, W.E. & Weber, E., 1986. Quantitative genetics and selection in plant
breeding, Walter de Gruyter. Available at:
http://books.google.com/books?id=ikteJNmsCp0C&pgis=1 [Accessed February 14,
2014].

Xu Yunbi, 2010. Molecular Plant Breeding. CAB International. UK. pp. 11-180.

Yada, B., P. Tukamuhabwa, A. Alajo and R. O. M. Mwanga, 2010. Morphological

characterization of Ugandan Sweet potato germplasm. Crop Science, 50(6),
pp.2364. Available at:
https://dl.sciencesocieties.org/publications/cs/abstracts/50/6/2364 [Accessed April
6, 2014].

Yadav, R K., Rai N., Yadav D.S. & Sanwal, S.K., 2007. Correlation, path-coefficient and
genetic diversity pattern in Colocasia (Colocasia fsculfnta) genotypes. Vegetable
Science, 34(2), pp.153–156.

Yohannes, G. Getachew, B. & Nigussie, D., 2010. Genotypic and phenotypic correlations
of root yield and other traits of orange-fleshed sweet potatoes [Ipomoea batatas (L.)
Lam.]. Journal of the drylands, 3(2), pp.208–213. Available at:
http://www.mu.edu.et/jd/pdfs/V3I2/5 Yohannes 208-213.pdf [Accessed April 5,
2014].

Younis N., M. Hanif1, S. Sadiq, G. Abbas, M. J. Asghar & M. Ahsanul Haq, 2008.
Estimates of genetic parameters and path analysis in lentil (Lens culinaris Medik).
Pak. J. Agri. Sci., 45(3), pp.44–48. Available at:
http://pakjas.com.pk/upload/5922.pdf [Accessed February 16, 2014].

Yu Li W., K. Shuzhi, C.A.O. Youngsheng and X. Zhang. 1996. A phenotypic diversity

analysis of foxtail millet (Setaria italica L.P. Beauv) landrace of Chinese origin.
Genetic Resources and Crop Evol. 43: 377-384.

Zelalem A., Tekalign T. and Nigussie D., 2009. Response of Potato (Solanum tuberosum
L.) to different rates of nitrogen and phosphorus fertilization on vertisols at Debre
Berhan, in the central highlands of Ethiopia. African Journal of Plant Science, 3(2),
16-24.

85
 APPENDEX

Appendix Table 1. The means of tannia genotypes for the 16 quantitative characters

Gen. Gen. Tot

SU LL Lw PLC PLlg PH COL CLD NCL CML CMD COWyi CMWyi CMDM CODM
No Name Yi
1 AAGT003 1.70 24.30 22.85 66.83 47.86 60.35 7.83 4.98 7.38 6.80 8.01 0.33 0.15 27.50 25.00 0.47
2 AAGT008 1.84 21.66 21.20 57.28 40.79 52.36 8.94 5.71 8.99 7.07 7.27 0.34 0.19 26.50 27.50 0.53
3 AAGT020 2.13 24.88 23.08 72.69 48.91 58.99 11.14 5.76 12.35 8.19 8.26 0.44 0.27 27.00 26.50 0.71
4 AAGT022 1.80 21.94 19.14 57.11 39.76 49.47 9.62 5.08 14.62 7.87 7.28 0.42 0.27 29.50 25.50 0.69
5 AAGT030 2.56 23.94 24.00 70.92 46.88 55.65 10.98 4.94 14.76 7.85 8.24 0.40 0.25 29.00 28.00 0.65
6 AAGT031 2.69 24.90 23.92 60.95 43.99 55.57 10.32 5.21 9.30 8.37 8.77 0.42 0.26 29.00 27.50 0.68
7 AAGT034 1.92 25.21 24.74 69.39 48.21 59.30 10.08 5.38 12.96 8.39 8.83 0.47 0.23 27.00 30.50 0.70
8 AAGT035 3.56 23.60 20.35 59.97 38.40 48.71 9.86 4.20 9.11 8.12 9.00 0.30 0.22 26.00 24.50 0.52
9 AAGT036 1.94 24.54 21.83 58.37 41.05 50.16 9.98 5.33 9.11 8.18 7.94 0.36 0.21 30.00 30.50 0.58
10 AAGT043 2.33 27.69 27.40 69.12 56.62 65.10 10.72 5.69 11.13 9.34 8.71 0.66 0.33 30.00 24.50 0.99
11 AAGT045 2.54 25.64 24.79 64.66 55.14 66.07 12.07 5.26 10.16 8.32 8.46 0.68 0.36 26.50 27.50 1.04
12 AAGT046 2.16 24.29 23.42 62.00 44.73 56.99 8.63 4.90 9.16 9.53 9.05 0.32 0.19 26.00 29.50 0.52
13 AAGT051 2.69 26.27 23.31 78.08 43.95 52.53 10.29 5.15 17.33 7.77 8.92 0.36 0.26 32.00 28.00 0.62
14 AAGT052 3.08 25.61 23.01 70.24 45.00 55.32 11.28 5.20 11.89 8.90 8.28 0.52 0.21 27.50 31.00 0.73
15 AAGT054 3.08 24.01 22.67 67.94 46.17 57.89 12.96 4.94 13.03 8.42 8.46 0.50 0.26 27.50 27.50 0.77
16 AAGT058 2.16 25.51 24.90 70.51 47.35 59.07 9.60 5.19 10.90 8.16 7.94 0.44 0.26 30.00 26.50 0.70
17 AAGT061 2.11 22.27 21.47 60.65 41.21 49.82 8.75 5.67 9.48 7.43 7.74 0.35 0.19 26.00 25.50 0.54
18 AAGT065 2.70 24.50 23.58 68.00 44.34 53.12 9.55 5.04 11.58 8.66 8.61 0.40 0.26 28.50 26.50 0.66
19 AAGT069 3.53 27.21 24.78 72.96 57.52 71.40 10.78 5.95 16.96 8.23 8.35 0.67 0.43 27.50 29.50 1.09
20 AAGT077 2.39 23.51 22.38 64.24 42.04 51.33 11.20 4.30 12.28 7.37 7.87 0.42 0.25 28.50 28.00 0.67
21 AAGT080 3.42 22.45 20.68 61.30 43.92 52.81 11.24 5.38 10.02 7.13 7.56 0.38 0.16 28.50 32.50 0.55
86
Appendix Table 1. The means of tannia genotypes (continued)

Gen. Gen. Tot

SU LL Lw PLC PLlg PH COL CLD NCL CML CMD COWyi CMWyi CMDM CODM
No Name Yi
22 AAGT083 2.30 22.24 20.55 68.17 40.32 51.09 9.79 5.25 11.37 7.48 7.87 0.43 0.19 26.50 29.00 0.62
23 AAGT085 1.78 25.41 23.71 63.07 49.99 59.99 9.60 5.13 8.85 8.31 8.44 0.41 0.22 30.50 28.50 0.62
24 AAGT088 2.22 22.05 19.47 62.10 39.66 48.46 9.42 5.45 10.44 7.32 7.88 0.36 0.17 31.00 29.00 0.54
25 AAGT092 2.36 24.70 22.17 68.34 47.97 62.81 9.62 5.27 9.76 9.12 8.80 0.36 0.22 29.00 28.50 0.58
26 AAGT093 2.13 22.13 20.61 63.46 45.64 53.43 9.59 5.27 12.85 8.16 8.73 0.43 0.21 30.50 27.50 0.64
27 AAGT094 2.63 27.75 27.25 77.11 49.97 64.97 11.55 5.48 13.05 8.59 10.15 0.63 0.23 29.00 25.00 0.86
28 AAGT097 1.53 25.82 23.59 62.82 44.42 55.74 10.17 5.62 10.73 6.74 8.38 0.47 0.29 30.50 27.50 0.76
29 AAGT099 3.23 23.17 22.19 73.17 42.46 54.78 8.96 4.85 12.08 7.26 7.41 0.43 0.19 30.00 29.50 0.62
30 AAGT100 1.97 26.75 22.72 66.92 47.80 56.53 8.46 4.82 9.17 8.22 8.23 0.37 0.18 28.50 26.50 0.55
31 AAGT102 4.42 24.55 22.37 82.03 48.74 58.28 11.27 5.47 21.12 8.31 8.56 0.71 0.36 27.50 28.50 1.07
32 AAGT106 2.27 21.03 19.86 65.09 42.18 49.61 10.70 5.37 13.57 7.04 7.67 0.48 0.19 27.50 27.00 0.67
33 AAGT109 1.53 21.98 21.51 57.73 38.64 49.10 8.75 4.60 8.01 7.34 7.58 0.46 0.13 26.00 25.00 0.59
34 AAGT112 2.06 23.02 21.52 67.19 43.52 53.01 10.94 5.94 12.39 7.25 7.28 0.44 0.25 22.00 30.00 0.69
35 AAGT116 1.95 20.86 19.43 65.84 41.05 51.57 8.79 5.39 10.96 8.30 7.94 0.43 0.21 27.50 26.50 0.64
36 AAGT120 1.48 24.91 21.92 56.82 41.78 53.89 10.47 5.34 9.99 6.79 7.85 0.41 0.20 27.00 28.00 0.62
37 AAGT121 2.20 23.67 23.10 61.64 43.08 53.33 11.58 5.51 11.54 7.73 8.22 0.48 0.20 26.00 29.00 0.68
38 AAGT127 1.19 25.49 23.85 69.95 46.42 57.03 9.51 4.34 9.96 7.39 7.96 0.28 0.15 23.50 24.50 0.44
39 AAGT132 2.81 22.12 21.80 56.83 41.13 49.96 9.31 5.41 11.23 8.44 8.55 0.44 0.32 32.50 29.00 0.76
40 AAGT135 1.91 22.65 22.22 56.33 42.50 50.09 11.48 5.47 8.62 7.56 7.59 0.41 0.19 27.50 28.00 0.60
41 AAGT138 2.17 22.69 21.70 70.09 44.30 53.13 10.85 5.33 11.90 9.10 9.00 0.52 0.21 29.00 30.00 0.73
42 AAGT144 2.30 24.52 23.42 58.48 47.85 58.64 11.62 4.65 10.87 8.79 8.39 0.39 0.23 27.00 27.50 0.63
43 AAGT148 2.00 26.21 26.01 62.74 44.98 60.35 9.82 4.97 13.09 8.83 8.99 0.44 0.24 29.00 25.50 0.69
44 AAGT152 1.72 29.60 27.93 66.40 50.41 64.89 13.13 5.46 11.36 9.23 9.88 0.51 0.28 29.50 29.50 0.79
45 AAGT155 1.83 26.75 24.39 67.94 48.42 61.72 11.61 5.61 21.83 8.47 8.36 0.69 0.36 30.00 26.00 1.05
87
Appendix Table 1. The means of tannia genotypes (continued)

Gen. Tot
Gen. Name SU LL Lw PLC PLlg PH COL CLD NCL CML CMD COWyi CMWyi CMDM CODM
No Yi
46 AAGT159 2.44 25.08 23.33 64.58 45.76 57.07 8.96 3.33 11.54 6.46 6.59 0.30 0.14 29.00 24.50 0.44
47 AAGT163 2.19 23.51 22.00 64.91 43.32 58.60 12.61 5.98 8.23 8.09 8.48 0.55 0.18 28.50 28.50 0.72
48 AAGT171 2.25 23.13 19.31 60.91 38.12 44.68 8.69 4.74 13.49 6.43 7.29 0.27 0.20 27.50 27.00 0.47
49 AAGT176 1.34 21.68 21.54 61.42 40.22 49.12 11.36 5.42 7.76 7.66 7.17 0.32 0.13 29.00 28.50 0.45
50 AAGT177 2.58 23.80 22.29 65.41 45.04 55.64 11.47 5.39 14.01 7.83 8.52 0.48 0.29 30.50 30.00 0.77
51 AAGT178 2.14 24.76 23.15 66.70 46.98 57.77 8.92 5.17 9.58 7.57 7.74 0.37 0.24 22.00 30.00 0.61
52 AAGT180 2.52 22.58 21.56 60.79 43.58 53.71 10.82 5.35 14.59 7.53 7.72 0.51 0.21 28.50 28.50 0.72
53 AAGT183 2.36 23.42 21.56 71.22 41.48 59.00 12.11 5.45 15.50 7.94 8.20 0.74 0.27 25.00 26.50 1.01
54 AAGT186 3.52 27.15 25.33 78.26 49.49 61.09 10.22 5.37 13.40 7.08 8.54 0.44 0.26 28.00 25.00 0.70
55 AAGT188 2.16 23.95 22.54 60.34 49.15 62.89 11.71 4.12 12.69 7.73 8.56 0.41 0.22 27.50 29.00 0.63
56 AAGT193 1.27 23.96 22.14 64.48 45.72 52.49 10.22 5.35 8.78 7.43 9.96 0.36 0.19 27.00 28.00 0.55
57 AAGT195 3.13 24.48 22.90 71.66 49.28 55.78 11.86 4.55 15.83 7.78 8.68 0.50 0.31 28.50 31.50 0.81
58 AAGT199 1.72 20.81 18.44 62.28 37.72 46.17 10.41 4.83 8.08 7.88 7.06 0.63 0.13 31.50 24.50 0.76
59 AAGT202 2.67 25.26 24.32 70.46 51.90 61.41 11.43 5.82 12.57 7.36 8.71 0.48 0.26 29.00 32.50 0.73
60 AAGT205 2.98 22.08 21.76 60.07 43.59 52.62 10.39 5.45 14.32 7.24 7.90 0.40 0.26 31.50 27.50 0.66
61 AAGT208 1.91 23.93 22.13 57.86 47.79 57.42 11.04 5.42 10.28 8.56 9.17 0.42 0.26 30.50 30.00 0.68
62 0002/07 2.40 26.94 21.49 71.62 56.91 68.30 10.36 5.25 9.51 8.33 8.80 0.68 0.33 21.50 20.00 1.01
63 0003/07 4.39 28.57 24.62 73.87 54.36 72.72 10.06 5.20 9.01 9.64 9.39 0.74 0.36 21.50 20.50 1.10
64 AAGT174 3.27 25.43 22.38 72.14 51.03 64.27 8.99 5.45 15.83 8.31 8.53 0.32 0.20 25.00 28.50 0.52

SU = number of suckers per plant, LL = lamina length, LW = lamina width, PLC = plant canopy diameter, PTLg = petiole length, PH = plant height, COL = cormel length,
CLD = cormel diameter, NCL = number of cormels per plant, CML = corm length, CMD = corm diameter, COW = cormel fresh weight, CMW = corm fresh weight, CMDM
= corm dry matter, CODM = cormel dry matter, TotYi = total yield per plant

88