Genetic and Phenotypic Diversity of Robusta Coffee (Coffea canephora L.)

GENETIC AND PHENOTYPIC DIVERSITY OF ROBUSTA COFFEE (COFFEA CANEPHORA L.)

3

Kahiu Ngugi⁎, Pauline Aluka† ⁎

Department of Plant Sciences and Crop Protection, College of Agriculture and Veterinary Sciences, University of Nairobi, Nairobi, Kenya †National Agricultural Research Organization (NARO), National Coffee Resources Research Institute (NaCORI), Mukono, Uganda

3.1  The Genetic Diversity of Robusta Coffee (Coffea canephora L.) Assessed by Simple Sequence Repeat Molecular Markers Robusta coffee germplasm that contribute 60% of Uganda’s foreign earnings is under threat from abiotic, biotic, and population pressures. This study assessed the genetic variability among cultivated Robusta coffee (Coffea canephora) that might be utilized to enhance productivity and quality and ultimately improve small-scale farmer earnings. A total of 349 diverse accessions of C. canephora were evaluated using 18, simple sequence repeats markers (SSRs) present in C. canephora linkage groups. Genetic diversity and the F-statistics were estimated with the GenAlEx 6.41 statistical package. Unweighted Pair Group Arithmetic Mean (UPGMA) was used to derive Darwin Neighbor Joining tree. The dissimilarity matrix table of accessions was calculated from principal component analysis (PCA). Analysis of molecular variance (AMOVA) was estimated by Arlequin software. Populations gave a mean allele range of 3–10 and the genetic diversity over loci range was 0.53–0.78. The accessions were grouped into three and had a mean genetic distance of 0.60. Within populations variability was 81.87%, whereas diversity within individuals was 54.05% and that between populations was 17.03%. The inbreeding index was lower than 50% suggesting that outcrossing was prevalent in C. canephora accessions. These results indicated that C. canephora from Uganda has tremendous genetic diversity, which is constantly being enriched Caffeinated and cocoa based beverages. https://doi.org/10.1016/B978-0-12-815864-7.00003-9 © 2019 Elsevier Inc. All rights reserved.

89

90  Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE

by the gene flow between the wild and cultivated ­populations. The ­emergence of two predominantly farmer selected landraces of nganda and erecta that differ distinctly in their plant types is also an indication that disruptive selection force of natural selection was key to the evolution of this diversity.

3.1.1 Introduction Robusta coffee contributes 80% of total production in Uganda and is grown on estimated 270,000 hectares (UCDA Annual reports, 2001– 2003). This production has earned the country about 388.4 million US dollars for its 8 million farmers (UCDA Annual reports, 2007–2008). Robusta coffee also known as Coffea canephora contributes 30% of the world’s production. C. canephora is a diploid parent hybridized with Coffea eugenioides to produce Coffea arabica, an allotetraploid (Combes et  al., 2000). There are two predominant forms of C. canephora found in Uganda: the erect type of Robusta coffee, known as erecta (also known as Coffea quillou) and nganda or Coffea ugandae, the spreading type. These two types of Robusta are cultivated together in mixtures and cross easily between themselves (Thomas, 1935). These semiwild forms of coffee with diverse phenotypic characteristics are reported to have tolerance to a number of pests and diseases, besides being high yielding (Prakash et  al., 2005). In recent times, C. canephora has undergone extensive genetic erosion imposed by biotic, abiotic, and human settlement factors which has led to decreased heterozygosity in the germplasm that now faces extinction and needs urgent conservation. DNA molecular markers such as microsatellites or simple sequence repeats (SSRs) are powerful tools that could be utilized to quicken the improvement of marketable traits such as yield and cup quality. SSRs are the molecular markers of choice in marker-assisted selection (MAS) of most crops because, they are widely found in the genome, are codominant, can be multiplexed and easily automated when compared to other marker systems such as AFLP (amplified fragment length polymorphism), RFLPs (restriction fragment length polymorphisms), or RADPs (random amplified DNA polymorphisms) (Aga et al., 2003; Leroy et al., 2005; Prakash et al., 2005). The most recent coffee genetic maps have been extensively constructed using SSR, RFLP, and SNP (single nucleotide polymorphism) markers (Dufour et  al., 2001). SSR markers linked to important agronomic traits of C. canephora would be useful tools in the development of coffee cultivars with superior market-driven traits that are urgently needed to raise coffee production in Uganda. In this study 18 SSR markers were used, of which 14 were from C. canephora clone 126 genetic library (Dufour et al., 2001; Pouncet

Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE   91

et al., 2007), three were from C. arabica var “Cattura” (Combes et al., 2000; Rovelli et al., 2000), and one was obtained from a bacterial artificial chromosome (BAC) library (Leroy et  al., 2005). These SSRs were developed at the Centre for International Agricultural Research Development, France (CIRAD) (Billotte et al., 1999) and the primers were synthesized by Eurogentec of South Africa.

3.1.2  Materials and Methods 3.1.2.1  Germplasm Collection In this study, the C. canephora L. germplasm screened included cultivated landraces and gene-bank accessions (Table 3.1). DNA from young leaves of 84 accessions, which included 19 erecta types, 20 nganda types, 18 from Entebbe Botanical Gardens, 15 from crosses, 7 parental materials of the crosses, and 5 controls were selected and screened with 9 SSRs markers as shown in Table 3.4 and in Fig.  3.5A. The accessions from Entebbe Botanical Gardens, the crosses, and their parental materials were regarded as improved germplasm. DNA from 231 leaf samples of Robusta local landraces was also extracted from 14 coffee growing districts as shown in Fig. 3.1. Also, eight elite commercial clones, seven accessions from West and Central Africa and two Ugandan wild types were included. One nganda type was used as the control DNA. Nganda accessions were identified in the field as the spreading trees with weaker upright stems, whereas trees erecta types were the trees with stronger stems and bigger berries. Landraces in districts near forests were sampled after every 5 km in order to minimize duplication (Fig. 3.1 and Table 3.2).

3.1.2.2  DNA Extraction and Amplification Leaf DNA was extracted using mixed alkyltrimethylammonium bromide (MATAB) buffer as explained by Risterucci et  al. (2000) at CIRAD laboratory. DNA quality was improved by the genomic DNA purification Promega Wizard Kit and readjusted to 0.5 ng/μL after calculating the concentration with cocoa standard DNA concentrations in agarose gel. The PCR amplification of SSR loci shown in Table 3.1 was conducted in a 10.0 μL final volume. The volume was made up of 5.0 μL of 0.5 ng/μL genomic DNA, and 5.0 μL of the master mix. The contents and concentrations of the mix were as follows: 10 mM tris-HCL, 50 mM KCl, 0.1% triton X-100, 1.5 nM of Mgcl2, 0.2 pmol of primer, and 0.2 nM of dNTPs (dCTP, dGTP, and dTTP). The primers included 0.01 nM of dATP (reverse primer) and 0.8 nCl [33P]-dATP (forward primer tailed with M13 sequence manufactured by Amersham Pharmacia, Piscataway, NJ). A 0.5 μL Taq DNA polymerase from

Table 3.1  Landraces and Gene-Bank Accessions Assessed for Genetic Diversity Source

Agroecological Zone/Germplasm

Landraces

Western highlands

Lake-Albert Crescent

Southern drylands South East Gene-bank and conserved collections

Lake-Victoria Crescent

Controls

Ugandan wild “nganda” types West and Central Africa controls

Total

Origin

Location Code

Gene-Bank and Controls

Landraces and Controls

Bundibugyo Kabarole Kamwenge Kyenjojo Hoima Kibale Kiboga Kayunga Mubende Mukono Mpigi Rakai Jinja Kamuli Kawanda Hybrids Entebbe Kawanda Kawanda Kibale forest Kawanda West and Central Africa

BU KB KW KJ HM KI KG KY MB MK MP RK JJ KM UC UH EB UE UN UW UN C&G

– – – – – – – – – – – – – – 15 6 12 19 20 7 – 5

11 31 34 37 24 20 8 8 13 4 11 15 3 12 16 8 – – – 2 1 7

84

265

Key: BU, Bundibudyo; HM, Hoima; JJ, Jinja; KB, Kabarole; KG, Kiboga; KI, Kibale; KJ, Kyenjojo; KM, Kamuli; KW, Kamwenge; KY, Kayunga; MB, Mubende; MK, Mukono; MP, Mpigi; RK, Rakai; UC, controlled crosses; UH, hybrids; EB, Entebbe botanical gardens; UE, erecta types; UN, nganda types; C&G, West and Central Africa controls; UW, UN, Ugandan wildand nganda-type controls. Reproduced with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE   93

Fig. 3.1  Map of Uganda showing sites where Robusta coffee germplasm was collected. Reproduced with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

Promega, Madison, WI, was finally added to the mixture. The whole PCR reaction was denaturated, annealed, and elongated in a PTC-200 thermocycler (MJ and Appendorf Research, Westbury, New York) using a touchdown.

3.1.2.3  Electrophoresis and Allele Coding The PCR products were loaded in a LICOR 4300 automated sequencer (LI-COR Biosciences, Lincoln, Nebraska) in the capillary comb radioactively labeled with a 10-base pair (bp) ladder that migrated in a 25 cm plate in the denaturing polyacrylamide gel. ­ The image bands were labeled as alleles based on fragment size (bp)

Table 3.2  Eighteen SSR Markers Used to Evaluate the Genetic Diversity of the 265 Cultivated C. canephora Accessions SSR Primer

Estimated Allele Position Min

Max

Repeat Type

DL026

No.

Motif

17 A

1 15

355 170

212

TG

265

321

CA

82

112

A

166

210

TG

263

276

AC

137

161

TG

181

206

A

211

245

CA

283

297

AC

358

2 11

364

2 21

368

1 13

384

2 10

394

2 9

429

2 13

442

1 19

445

2 10 2

Primer Sequences (5→3′) F:CGAGACGAGCATAAGAA R:GTGGAATGAAGAATGTAG F:CTATGATGTCTTCCAACCTTCTAAC R:GGTCCAATTCTGTTTCAATTTC F:CATGCACTATTATGTTTGTGTTTT R:TCTCGTCATATTTACAGGTAGGTT F:AGAAGAATGAAGACGAAACACA R:TAACGCCTGCCATCG F:CACATCTCCATCCATAACCATTT R:TCCTACCTACTTGCCTGTGCT F:ACGCTATGACAAGGCAATGA R:TGCAGTAGTTTCACCCTTTATCC F:GCCGTCTCGTATCCCTCA R:GAAGCCCAGAAAGTAGTCACATAG F:CATTCGATGCCAACAACCT R:GGGTCAACGCTTCTCCTG F:CGCAAATCTGAGTATCCCAAC R:TGGATCAACACTGCCCTTC F:CCACAGCTTGAATGACCAGA R:AATTGACCAAGTAATCACCGACT

Sequence Origin

Primer Origin

Species Origin

Leroy 2005

Leroy 2005

C. canephora, clone 126

Dufour 2001

Poncet 2007

C. canephora, clone 126

Dufour 2001

Poncet 2007

C. canephora, clone 126

Dufour 2001

Poncet 2007

C. canephora, clone 126

Dufour 2001

Poncet 2007

C. canephora, clone 126

Dufour 2001

Poncet 2007

C. canephora, clone 126

Dufour 2001

Poncet 2007

C. canephora, clone 126

Dufour 2001

Poncet 2007

C. canephora, clone 126

Dufour 2001

Poncet 2007

C. canephora, clone 126

Dufour 2001

Poncet 2007

C. canephora, clone 126

456

14 273

313

AC

88

124

AC

305

339

CT

135

181

TG

283

321

CA

126

158

GT TG, GA

116

138

230

290

461

2 9

471

2 12

501

2 8

753

2 15

790

2 21

837 477

AC

2 16, 11 16

2 2

F:TGGTTGTTTTCTTCCATCAATC R:TCCAGTTTCCCACGCTCT F:CGGCTGTGACTGATGTG R:AATTGCTAAGGGTCGAGAA F:TTACCTCCCGGCCAGAC R:CAGGAGACCAAGACCTTAGCA F:CACCACCATCTAATGCACCT R:CTGCACCAGCTAATTCAAGC F:GGAGACGCAGGTGGTAGAAG R:TCGAGAAGTCTTGGGGTGTT F:TTTTCTGGGTTTTCTGTGTTCTC R:TAACTCTCCATTCCCGCATT F:CTCGCTTTCACGCTCTCTCT R:CGGTATGTTCCTCGTTCCTC F:CGAGGGTTGGGAAAAGGT R:ACCACCTGATGTTCCATTTGT

Reproduced with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

Dufour 2001

Poncet 2007

C. canephora, clone 126

Dufour 2001

Poncet 2007

C. canephora, clone 126

Dufour 2001

Poncet 2007

C. canephora, clone 126

Dufour 2001

Poncet 2007

C. canephora, clone 126

Rovelli 2000

Poncet 2004

C. canephora, “Caturra”

Rovelli 2000

Poncet 2004

C. canephora, “Caturra”

Rovelli 2000

Poncet 2004

C. canephora, “Caturra”

Dufour 2001

Poncet 2007

C. canephora, clone 126

96  Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE

in the SAGA Generation 2 computer program as described by Combes et al. (2000). As a diploid C. canephora was expected to record a maximum of two alleles per individual.

3.1.2.4  Data Analysis The mean between and within population molecular diversity was estimated with Arlequin version 2 as shown in Tables 3.3 and 3.4 (Schneider et al., 1999). The mean population heterozygosity over loci and F-statistics as shown in Tables 3.3 and 3.4, respectively, were calculated with Gen-Al-Ex 6.41 statistical package (Peakall and Smouse, 2006). The genetic distance dissimilarity was calculated with Darwin4 software developed by CIRAD based on the Dice index (1945) which is equivalent Nei and Li (1979) genetic distance. This is calculated 2 n1,1 as: SG = , where n1,1 is the number of bands shared 2 n1,1 + n1,0 + n 0 ,1 ) by the individuals I and j, n1,0 is the number of bands observed for I and missing for j, and number n0,1 the number of bands observed for j and missing for I. When both I and j were missing, the information given was not considered. The genetic distance were represented graphically as a tree by the factorial analysis of dissimilarity table (FADT) calculated using the Darwin Neighbor Joining UPGMA method as described by Saitou and Nei (1987) shown in Fig. 3.3. The dendrogram shown in Fig.  3.4 was derived from FADT. FADT was also used to generate the geographical distribution of the genetic variation in the form of the linear and nonparametric patterns of PCA, (Fig. 3.5A and B) using the XLSTAT version 2011.2.05 statistical package. The reliability of the tree was evaluated by a bootstrap, 1000 repeated data analysis.

3.1.3 Results The 18 SSR markers were polymorphic as shown by the allelic polymorphism at two loci that differed in fragment size in Fig. 3.2. Kyenjojo (74) populations had the highest total gene copies, followed by Kamwenge (68) and Kabarole (62) districts but the lowest counts came from Jinja (6), Mukono (8), and from DNA controls (6–8) (Table  3.3). Kyenjojo (70), Kamwenge (64.11), and Kabarole (57.44) also had the highest usable gene copies which were also lowest among DNA controls (5.33–5.67). The lowest number of usable (4) and polymorphic loci (4) was recorded in Kamuli population. Jinja district accessions had the most usable loci of 17.0. The highest number of 16 polymorphic loci was found in Kibale district, despite the population having the lowest gene diversity of over 0.50. Kiboga accessions (0.78) had the highest gene diversity over loci, followed by hybrids (0.72) and Kamwenge accessions (0.71). Kamwenge had the highest number of

Table 3.3  Population Diversity Parameters of 265 C. canephora Accessions Estimated According to Nei and Li (1979) for 18 Loci at 50% Level of Missing Data Heterozygosity

Pop

Total Gene Copies

Usable Loci

Poly Loci

Gene Diversity Over Loci

Usable Gene Copies

Alleles

Obs (Ho)

Exp (He)

BU C G UG HM JJ KB KG KI KJ KM KW KY MB MK MP RK UC UH

22 8 6 6 48 6 62 16 40 74 24 68 16 26 8 22 30 32 16

14 16 13 15 15 17 14 7 16 12 4 11 14 8 15 13 11 12 5

14 15 13 12 14 13 14 7 16 12 4 11 14 8 12 13 11 12 5

0.61 0.70 0.67 0.53 0.59 0.53 0.53 0.78 0.50 0.68 0.63 0.71 0.61 0.68 0.56 0.62 0.61 0.62 0.72

20.11 7.33 5.33 5.67 44.56 5.67 57.44 14.33 38.67 70.00 20.78 64.11 14.89 23.78 7.56 21.00 27.78 29.44 13.89

5.11 3.72 2.61 2.56 6.39 2.44 5.78 4.94 6.22 7.94 5.44 9.56 4.39 6.06 2.94 5.50 5.67 5.44 4.72

0.40 0.44 0.23 0.42 0.39 0.33 0.28 0.38 0.38 0.37 0.47 0.42 0.47 0.41 0.47 0.46 0.44 0.41 0.50

0.56 0.73 0.58 0.55 0.58 0.50 0.52 0.70 0.56 0.67 0.66 0.72 0.59 0.66 0.60 0.64 0.62 0.60 0.68

Key: BU, Bundibudyo; HM, Hoima; JJ, Jinja; KB, Kabarole; KG, Kiboga; KI, Kibale; KJ, Kyenjojo; KM, Kamuli; KW, Kamwenge; KY, Kayunga; MB, Mubende; MK, Mukono; MP, Mpigi; RK, Rakai; UC, controlled crosses; UH, hybrids; C, G, UG, controls (Ugandan wild, nganda, Congolese). Reproduced with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

Table 3.4  Mean Population Heterozygosity Over Loci in 265 Cultivated Accessions and 77 Germplasm Collections Germplasm

Pop

N

Na

Ne

I

Ho

He

UHe

F

Pa

Cultivated C. canephora

C G UN UW BU HM JJ KB KG KI KJ KM KW KY MB MK MP RK UC UH Mean C EB UC UE UH UN UP Mean

4 3 1 2 11 24 3 31 8 20 37 12 34 8 13 4 11 15 16 8 13.25 4 10 13 18 5 20 7 11

3.83 2.89 1.50 1.83 5.33 6.67 2.50 6.06 5.56 6.61 8.50 6.22 10.06 4.61 6.61 3.11 5.78 6.06 5.78 5.44 5.25 4.15 4.65 4.35 5.35 3.35 5.55 4.70 4.59

2.97 2.61 1.50 1.67 3.23 3.19 2.14 2.76 3.85 3.32 4.12 3.63 4.88 3.03 3.97 2.58 3.39 3.08 3.15 3.66 3.14 3.43 3.14 2.96 3.26 2.74 3.49 3.44 3.21

1.11 0.93 0.35 0.48 1.17 1.24 0.71 1.11 1.43 1.25 1.55 1.44 1.71 1.16 1.45 0.93 1.32 1.29 1.24 1.41 1.16 1.29 1.24 1.17 1.32 1.03 1.35 1.32 1.25

0.36 0.20 0.50 0.31 0.40 0.39 0.33 0.28 0.35 0.36 0.36 0.40 0.40 0.44 0.38 0.43 0.43 0.39 0.40 0.43 0.38 0.47 0.53 0.55 0.59 0.55 0.59 0.66 0.56

0.59 0.56 0.25 0.32 0.55 0.57 0.42 0.52 0.69 0.55 0.69 0.68 0.73 0.59 0.67 0.53 0.62 0.61 0.60 0.69 0.57 0.69 0.64 0.62 0.67 0.58 0.67 0.68 0.65

0.67 0.67 0.50 0.43 0.58 0.59 0.50 0.53 0.74 0.57 0.69 0.71 0.74 0.63 0.70 0.60 0.65 0.63 0.62 0.74 0.62 0.80 0.68 0.65 0.69 0.65 0.69 0.73 0.70

0.39 0.68 −1.00 0.01 0.33 0.34 0.26 0.50 0.52 0.34 0.47 0.43 0.46 0.32 0.49 0.19 0.35 0.41 0.43 0.40 0.32 0.33 0.17 0.14 0.12 0.10 0.12 0.03 0.14

15 13 2 3 4 8 nd 4 2 3 11 6 22 nd 4 nd 3 3 2 4 6.41 15 7 2 1 1 8 nd 5.67

C. canephora germplasm collection

Key: N, no. of genotypes; Na, no. of alleles; Ne, no. of effective alleles; I, Shannon Diversity Index; Pa, rare alleles; Ho, observed heterozygosity; He, expected heterozygosity; UHe, unbiased expected heterozygosity; F, fixation index; nd, not detected; UB, Entebbe botanical gardens; UP, parents. Reproduced with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE   99

Allele

Ladder

94 accessions

Ladder

Fig. 3.2  SSR polymorphism for 96 accessions of C. canephora. Reproduced with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

alleles (9.56), followed by Kyenjojo (7.94), Hoima (6.39), Kibale (6.22), and Mubende (6.06) (Table 3.3). Populations with the lowest total gene copies, also had the lowest number of alleles and these were Jinja (2.44), Ugandan control (2.56), Guinea (2.61), and Mukono (2.94). Hybrids had the higher observed heterozygosity (Ho,0.50) than the Guinea population (Ho 0.23); however these values were less than the expected heterozygosity (He) in all the populations. The Congolese and Kamwenge accessions had the highest expected heterozygosity at 0.73 and 0.72, respectively, but these values ranged from 0.50 in Jinja to 0.70 in Kiboga. Table 3.4 shows a comparison of the mean heterozygosity of cultivated and germplasm collection of C. canephora population. Most alleles were found in cultivated populations of Kamwenge (10.06) and Kyenjojo (8.50) districts and the least were found among the nganda control (1.50) and Jinja (2.50) populations. In the germplasm, allelic values ranged between 3.35 in the hybrids and 5.55 in nganda types. In cultivated Robusta, effective alleles ranged between 1.50 in the nganda control to 4.88 in the Kamwenge accessions and between 2.74 in the hybrids to 3.49 in the nganda types in the germplasm collection (Table 3.4). The germplasm collections observed heterozygosity values ranged from 0.47 in the Congolese control to 0.66 in the parent materials but in the cultivated accessions the highest values of 0.50 was given by nganda DNA control while the lowest value of 0.20 was given by the Guinean DNA control. Expected heterozygosity in cultivated accessions was highest in the districts of Kamwenge (0.73), Kyenjojo (0.69), and Kiboga (0.69) whereas in the germplasm collections, He values ranged from 0.58 in the hybrids to 0.69 in the Congolese populations. Nganda control had low expected unbiased heterozygosity values of 0.25 and 0.43. Accessions with high fixation values found in Kiboga (0.52), Kabarole (0.50), Mubende (0.49), Kamwenge (0.46) were a

100  Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE

r­ eflection of high genetic diversity within these populations than was the case in the parent materials (UP) (0.03) and in the Entebbe botanical gardens collections (0.17). Most rare alleles of about 22 were found in the Kamwenge accessions but were lowest in the hybrids and erecta types. In the Jinja, Kayunga, Mukono, and parental (UP) populations (Table 3.4), no rare alleles were found. In the two land races, nganda types had more rare alleles than erecta types (Table 3.4). The Ugandan cultivated Robusta coffee accessions were clustered into three diversity groups (Figs. 3.3 and 3.4). Accessions from

Fig. 3.3  Neighbor Joining tree for 265 genotypes derived from factorial analysis of dissimilarity matrix. Key codes can be found in Table 3.1. Reproduced with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE   101

Fig. 3.4  Dendrogram of 265 on farm accessions and 24 elite selections derived from factorial analysis of dissimilarity table. Key codes can be found in Table 3.1. Reproduced with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

West and Central Africa formed a group of their own. As shown in Figs.  3.3 and 3.4, genotypes from Kabarole, Bundibugyo (Western highlands region), and Hoima (Lake Albert Crescent region) located in Western Uganda formed one group while genotypes from Kyenjojo, Kamwenge (Western highlands), Kiboga (Lake Victoria Crescent bordering Lake Albert Crescent region), and Kamuli formed the second group and a mixture of accessions from different locations; Kayunga, Mpigi, Mubende, Rakai, Mukono from Lake Victoria Crescent, and Kamwenge from Western highlands formed the third group. Table 3.5 shows that out of 81.87% within population variability, 54.05% was contributed by individual accessions, whereas 28.92% was by among individuals within populations, 17.03% of the variation was shared among individuals. The inbreeding index (FIS) was 0.35 indicating that there was 65% outcrossing. The FIT value of 0.46 indicated that there was 54% outcrossing among genotypes from different locations whereas the genetic differentiation value (FST) of 0.17 showed the level of polymorphism that occurred over generations.

102  Chapter 3  Genetic and Phenotypic Diversity of Robusta Coffee (Coffea canephora L.)

Table 3.5  Analysis of Molecular Variance (AMOVA) for 265 Cultivated Robusta Coffee Accessions Source of Variation Among populations Among individuals within populations Within individuals Within populations Total

df

Sums of Squares

Variance Components

Percentage of Variation

18 246

434.68 1161.60

0.72 1.22

17.03 28.92

265 511 529

604.50 1766.10 2200.78

2.28 3.46 4.22

54.05 81.87

FIS

FIT

FST

0.35

0.46

0.17

Reproduced with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

Table 3.6 shows that after 10,000 permutations, the population differentiation (FST) and inbreeding index (FIS) had an average ­population differentiation index of 0.18. Most of the populations had 0.18 or 0.19 differentiation indices rating except for Kamwenge where it was 0.17. Guinea control population had the highest inbreeding index value of 0.81 whereas the Ugandan control population had the lowest inbreeding value of 0.02. The other populations had inbreeding indices that ranged from 0.20 to 0.49. Observed inbreeding indices in 12 districts were more or less the same. There was no significant differences between randomly observed inbreeding indices for the Guinean and Ugandan controls and those of collections from Jinja and Mukono locations but there were significant differences among observed inbreeding indices for Congolese types, hybrids, and those of genotypes from Kayunga. Fig.  3.5A, PCA, showed that all genotypes were widely distributed and were not consolidated into distinct groups. Genotypes from Kabarole, Bundibugyo, and Hoima (Fig.  3.5B) were grouped with controlled crosses (UC in red). The hybrids (UH in blue) were grouped with accessions from Kayunga, Mpigi, Mubende, Rakai, Mukono from Lake Victoria Crescent, and Kamwenge from Western highlands, but were separated by a genetic distance of about 0.58 from controlled hybrids. Kyenjojo, Kamwenge, Kiboga, and Kamuli populations were separated from the control DNA (UC in black) by a genetic distance of 0.55. The nganda and erecta types were spread in all the three genetic diversity groups at an average genetic distance of 0.6.

Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE   103

Table 3.6  F-Statistics Indices in 19 Population Indices (After 10,100 Permutations With Significance Tests of 1023 Permutations) No.

Populations

FST

FIS

P (Random FIS≥Observed FIS

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

Bundibugyo Congolese controls Guinea controls Ugandan controls Hoima Jinja Kabarole Kiboga Kibale Kyenjojo Kamuli Kamwenge Kayunga Mubende Mukono Mpigi Rakai Crosses Hybrid

0.18 0.18 0.19 0.18 0.18 0.19 0.18 0.18 0.19 0.18 0.18 0.17 0.18 0.18 0.18 0.18 0.18 0.18 0.18

0.29 0.49 0.81 0.02 0.32 0.33 0.44 0.46 0.26 0.38 0.22 0.43 0.24 0.35 0.20 0.30 0.31 0.26 0.22

0 0.01 0.06 0.32 0 0.12 0 0 0 0 0 0 0.0008 0 0.13 0 0 0 0.01

0.18

0.33

Mean

Reproduced with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

3.1.4 Discussion The 265 accessions evaluated by 18 microsatellite markers were genetically diverse as shown in Fig. 3.2. Accessions from Kamwenge (KW), Kyenjojo (KJ), and Kabarole (KB) districts, respectively, had the highest diversity as shown by the high values in total gene copies (68, 74, 62), allele numbers (9.56, 7.94, 5.78), and expected gene diversity (0.72, 0.66, 0.52) (Table 3.3). Accessions from Jinja, Guinean, Ugandan, and Mukono populations gave lower genetic diversity parameters (Table  3.3). Populations from Kamwenge, Kyenjojo, and Kabarole close to Kibale forest exchanged gene flow

104  Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE

Fig. 3.5  Principal Component Analyses of dissimilarity matrix for (A) germ-plasm collections and (B) cultivated Robusta coffee. Key codes as in Table 3.1. Reproduced with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

(Continued)

between wild and cultivated C. canephora. In these populations, natural selection seems to be the major evolutionary forces. But for accessions from urban districts of Jinja and Mukono farmer selection pressures would be responsible for the higher fixation indices shown in Table  3.3 and reduced genetic diversity. Some accessions such as those from Kibale displayed a different scenario. These accessions had the highest number of polymorphic loci (16) but also the lowest gene diversity over the loci perhaps ­indicating that only a few of the polymorphic loci might have been fixed. In these accessions, it is likely that there was little gene flow from the wild populations but there was outcrossing within

Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE   105

Fig. 3.5, Cont’d

the populations. Gene flow between cultivated and wild Robusta forms evidently contributed to the higher genetic diversity in accessions that were cultivated closer to the forests since semiurban and elite cultivars had lower number of alleles such as those from Jinja (2.44), Ugandan controls (2.61), nganda, and Mukono (2.56, 2.94) as shown in Table 3.3. This pattern is clearly demonstrated by the diversity data presented in this study. Kamwenge and Kyenjojo accessions that are closer to the forests where wild C. canephora grows naturally, had higher different alleles with a mean of 5.25 and with a range of 1.50– 10.06 whereas gene bank collections had a mean of 4.29 as shown in Table 3.4. Cultivated accessions had expected (He) heterozygosity of 0.57 whereas that in the germplasm accessions was 0.65 suggesting that among the farmer accessions the rate of outcrossing was higher but among the germplasm accessions, there was more inbreeding. The FST and FIS mean values of 0.19 and 0.33, respectively, confirms

106  Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE

the outbreeding nature of Robusta populations as shown in Table 3.6. The presence of more rare alleles in the local landraces also reaffirmed that cultivated accessions may have acquired genes from wild populations through gene flow. The almost similar Shannon index and fixation values shown in Table 3.4 are an indication that both artificial and natural selections have influenced the genetic diversity found in cultivated Robusta. Within populations (81.87%) and within individuals (54.05%) variations contributed the highest diversity (Table 3.5) suggesting that this is the genetic variation that breeders should exploit. Other data such as the low inbreeding indices in Table 3.5, go a step further to confirm that outbreeding is about 65% in C. canephora as has always been suggested (Musoli et  al., 2009) and contributes higher diversity values as shown by landraces in Tables 3.3 and 3.4. Such accessions as from the gene bank at Kawanda, Entebbe botanical gardens, and Kituza that lost most of their trees due to old age, pest, and diseases attack had lesser genetic diversity (Table 3.4; Fig. 3.5A) (Wrigley, 1988). Populations from Kamwenge (1.71) and Kyenjojo (1.55) gave higher Shannon Diversity Index (I) values (Table 3.3) than those of nganda (0.35), Ugandan wild (0.48), Guinean (0.93) controls, and Mukono (0.93) suggesting that these populations are well established. Again, the presence of rare alleles among the Kamwenge (10), Kyenjojo (8), and bordering Kibale forest was a confirmation of the extent of gene flow. This constant gene flow between cultivated and wild types of Robusta is the most likely source for genes to improve bean quality, biotic, and abiotic stress in Robusta coffee. There is a likelihood that similar gene flow is present among unselected and random mating populations in the Congolese, Guinean populations, and nganda populations. The genotypes from the germplasm collections (Fig.  3.5A) were widely distributed and were not confined into distinct groups, an indication that they are a representative of the gene pool. The formation of three diversity groups shown in Figs. 3.3–3.5B, confirmed how reliable and precise the two diversity methods used in the analysis were and also how closely related accessions were as placed in their respective groups. Some nganda or erecta type accessions did not cluster separately but were distributed within the three genetic diversity groups. This is probably because of increased outcrossing among these types making it difficult to differentiate them clearly at genetic level (Fig. 3.5A). In Fig. 3.5B, the hybrids (UH) grouped with accessions from Kayunga, Mpigi, Mubende, Rakai, Mukono from Lake Victoria Crescent, and Kamwenge from Western highlands that possibly provided their parents. Controlled crosses and improved hybrids and varieties from research organizations grouped into separate clusters, with a genetic distance of 0.58 implying their genetic base was narrow.

Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE   107

The study showed there was genetic diversity in both nganda and erecta cultivated C. canephora populations (Tables 3.5 and 3.6 and Figs. 3.3–3.5A and B) but most of this diversity was found among the nganda local landraces which have continued to exchange gene flow with the wild forms. The cultivated nganda landraces have experienced both natural and artificial forms of selection. The genetic diversity that has accrued from gene flow between cultivated and wild Robusta types is closer or similar to that found in West Africa, Congo, and Guinea and there is reason to believe that these later populations could have been earlier progenitors of the C. canephora landraces found in Uganda. Genetic drift and other evolutionary forces of selection might have created nganda and erecta types that outcrosses more with wild Robusta populations.

3.1.5 Conclusion Most of the genetic variability in C. canephora in Uganda is within populations and within individuals, implying that diverse cultivars are widely distributed countrywide. Almost all the current C. canephora landraces in Uganda are panmictic, highly outcrossing and their adaptation in various agroecological zones is most likely under both natural selection through gene flow and human selection. The outcrossing nature of the cultivated Robusta coffee has also created an effective gene exchange mechanism between the cultivated nganda and erecta types with the wild types continuously enriching the local germplasm with genetic variability for traits such as disease resistance, yield, and coffee cup quality.

3.2  Phenotypic Diversity of Landraces of Robusta Coffee 3.2.1 Introduction In the past, the International Plant Genetic Resources Institute (IPGR, 1997) developed coffee descriptors that were used to identify highly productive cultivars commercial Robusta coffee varieties in Uganda (Kibirige-Sebunya et  al., 1996; Leakey, 1970; Wrigley, 1988). Aga et al. (2003) argued that there is considerable phenotypic variance that is heritable despite the combined effects of the genotype, growth stages, and environment. As in other food products, Robusta coffee bean has physical characteristics such as weight, volume, size, shape, color, solubility, moisture content, and texture that determine quality. The quality characteristics inside the coffee green bean are influenced by genotypic and

108  Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE

­environmental factors (Leroy et  al., 2006a,b). According to Charrier and Berthaud (1988) the size, shape, color, chemical composition, and flavor of the Robusta bean are controlled largely by the genotype. Farm management practices are the main environmental factors that influence the variability of physical characters and biochemical composition of the green bean (Lashermes et al., 2000; Clifford, 1985). Coffee beans from higher altitude for instance are known to be smaller, denser, and harder and have more acidity, quality aroma, and flavor than those from lower altitudes. Bean physical characteristics preferred by different markets are dictated by the size homogeneity, the regularity, and reliability of the production. Bean size from a commercial point of view is related to the coffee grade and therefore tied to the price (Leroy et al., 2006a,b). Roasted bean with moisture content of 12.5% and above are normally discarded. Phenotypic diversity displayed at the farm level is an important component of a Robusta coffee improvement program (Van der Vossen, 1985). Morphological and bean physical characters diversity is important in the design of conservation measures of the germplasm and in the mitigation of genetic erosion and climatic effects. Morphological traits are the tools exploited in the conventional plant breeding in the selection of parental materials or in the advancement of progenies arising from hybridization (Charrier and Berthaud, 1985). But even modern MAS and or marker-assisted breeding (MAB), do make use of both morphological and DNA markers to hasten improvement of targeted traits (Montagnon et al., 1998). Phenotypic and morphological traits have hardly been utilized in the breeding of Robusta coffee. Earlier efforts, to assess phenotypic variation in Ugandan Robusta coffee landraces by Thomas (1940) and Maitland (1926) did not have enough information to conduct the selection and the germplasm targeted was of narrow genetic base (Leakey, 1970). The few gene bank accessions available at Kawanda do not represent the wide genetic diversity in the country. There is need therefore to characterize the physical bean variability in hybrids and commercial cultivars and in erecta’ and nganda local landraces from across the agroecological zones where Robusta coffee is grown in Uganda (Fig. 3.6).

3.2.2  Materials and Methods 3.2.2.1  Tree and Site Selection Accessions from six traditional Robusta growing agro-ecological zones, ranging from, Lake Victoria crescent to Southern highlands were collected (Fig.  3.7). They included 21 on-farm landraces and 3 germplasm collections from Kawanda, Kituza, and Entebbe botanical gardens (Table 3.7). In the 21 districts, which included those closer to

Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE   109

Fig. 3.6  Map showing areas the various agro-ecological zones where Robusta coffee is grown.

the forests, collections of green beans were done at a distance of every 5 km. The sample size consisted of 25–30 trees per population and only trees that were older than 10  years were sampled as they were considered to be adapted as landraces. Field records were used to select 59 trees at Kawanda based on whether the coffee types were nganda, erecta or hybrid. In all, 10 trees were selected at Entebbe botanical gardens (Wrigley, 1988) whereas at Kituza, 16 controlled crosses planted were selected.

110  Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE

21 20.5 20 19.5 19 18.5 18 17.5 17 16.5 16

Girth (cm)

(A)

s rid

Age (years)

LL (cm)

(B)

PC (numbers)

Fruits (numbers)

IL (cm)

s rid H yb

a nd N ga

ta

al ci er C om

m

ec

nd

2

H yb

C om

H yb

rid

s

a nd N ga

ta ec Er

C om

m

er

ci

al

m

er

ci

5

a

3 al

15

4

N ga

25

5

ta

35

ec

45

6

Er

55

7

Er

75 65

Stems

(C)

Fig. 3.7  Comparison of Robusta coffee types morphological attributes of 476 genotypes. PC, production capacity; IL, inter node length (cm); LL, leaf length (cm). Reproduced with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

3.2.2.2  Morphological Data Scoring The quantitative descriptors scored are as shown in Table 3.8. Tree age was estimated by comparing the observed age with perceived age, where the farmer did not know the true age. The following morphological traits were measured from a mean of 25 trees; 10 randomly ripe cherry clusters randomly selected; inter node length was measured as the mean distance of two nodes in a primary branch in centimeters (cm) from 10 branches per tree. Leaf length was measured as the mean distance from the base of a leaf petiole to the apex in centimeters from 10 leaves per tree. Leaf width was measured as the distance between two widest points of a leaf at half-length in centimeters from 10 leaves per tree. Tree production capacity and vigor were scored on a 1–5 scale; where 1 was—very few berries present in the tree; 5 was—very many berries in the tree and 1 was—unhealthy tree growth; 5 was— healthy tree growth, respectively. Data were recorded by Geographical Position Systems (GPSs).

3.2.2.3  Harvesting and Drying Robusta Coffee Ripe Cherry All the 206 tagged Robusta coffee trees had at least 2 kg of their ripe cherry harvested and scored as shown in Table  3.1. The cherry was poured into a water container, the debris was removed and the beans sun dried in wire mesh boxes. The moisture content of about 50% in the cherry was reduced to less than 12.5% by evenly spreading the beans in a layer of about 1.5 cm thickness (Clifford, 1985). The cherry was sun dried for a month and later stored in a dry well-aerated room. After drying, the cherry was carefully hulled with a metal tray. Defective cherries and bean hulls were separated from the clean beans by use of density and the cleaned beans were later stored in polythene bags in a well-aerated room for physical evaluation.

Table 3.7  Agroecological Information From the 23 Locations Where the C. canephora Samples Were Collected for Phenotypic Evaluation Altitude Range (m a s l) Region

Agro-Ecology

District

Code

No. of Phenotypes

Min

Max

East

South east

Mayuge Bugiri Jinja Mubende Kamuli Iganga Mukono Kayunga Kiboga Masaka Kalangala Masindi Hoima Kibaal Kyenjojo Kabarole Kamwenge Bundibidyo Bushenyi Rakai Mbarara Rukungiri Kawanda Entebbe Kituza Controls

MY BG JJ MB KM IG MK KY KG MA KL MS HM KI KJ KB KW BU BS RK MR RG KA EB KZ CN

18 17 18

1181 1080 1124

1217 1120 1184

22 17 15 20 21 18 18 12 26 21 23 18 26 16 19 18 19 18 50 10 16

1080

1120

1122 1088 1136 1202 1028 1152 1101 1110 1372 1009 1242 687 1015 1212 1383 1015 1177 1177 1200

1176 1105 1373 1283 1239 1180 1295 1247 1517 1568 1325 1017 1654 1230 1478 1559 1177 1177 1200

Central

Lake Victoria crescent

West

Lake Albert crescent

Western highlands

Southwest

Southern Drylands S. highland On station germplasm

476 Reproduced with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

112  Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE

Table 3.8  Quantitative Morphological Characters Scored in the 206 Accessions of Robusta Coffee Parameter

Character Description

Measure

Acronym Use

1. Tree

Number of stems Tree age Tree girth Production capacity Vigor Primary internode length Leaf length Leaf width Fruit number/axil

Counts Centimeters Centimeters Rating Rating Centimeters Centimeters Centimeters Counts

SN TA TG PC VG IL LL LW FN

2. Primaries 3. Leaves 4. Fruits

Key: Rating for production capacity and vigor: PC: 1—few berries present i; 5—many berries. Vigor: 1—unhealthy growth; 5—healthy growth. Reproduced with permission from P. Aluka, PhD thesis, University of Nairobi, 2013.

3.2.2.4  Evaluating Green Bean Physical Characteristics In all, 300 g of cheery beans were randomly collected from each of the 206 farms in 21 Robusta growing districts, which included Kawanda and Kituza Robusta germplasm collections. Grading of the beans was done according to the size with specified screens of sizes that varied as: (A) ≥18 (7.0 mm), large; (B) ≥15–17 (6.0–6.75 mm), medium; and (C) 15 (6.0 mm), small. Roasted and green beans were weighed with an analytical balance and measured in grams while the volume in cubic milliliters was measured by a volumetric cylinder. Total roast time was divided with total weight of green beans roasted to produce, roast time per gram (RTPG) in seconds. Green bean percentage weight loss gave the percentage weight decrease of roasted beans. Roasted beans volume increase in percentage was calculated as a percentage, of the difference between green and roasted bean volumes.

3.2.2.5  Data Analysis Multivariate analysis was conducted using PCA and the geometric representation for the main factors plotted using XLSTAT version 2011.2.05 (Addinsoft, Paris, France). Before estimating the PCA, the Bartlett's test (XLSTAT version 2011.2.05 statistical program) was conducted to establish if significant differences from zero existed in correlated phenotypic variables. Genetic distances between accessions were estimated using the Euclidean straight-line method (Mohammadi

Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE   113

and Prasanna, 2003). In order to refine the principal component plot, Varimax rotation (XLSTAT version 2011.2.05) statistical program was applied (Mohammadi and Prasanna, 2003). Genotypes with the highest attribute variance were then assigned to the principal components. Mahalanobis and Fisher intergroup distances at 95% probability were calculated using factorial step discriminant analyses. The efficiency of genotype placement among groups was done using the confusion matrix. The distance of each phenotypic character from the average calculated similarity index between phenotypic characters of the various genotypes explained the origin of the accessions in relation to the populations. Descriptive statistics were used to compare the group means, medians, variances, and interquartile range in form of box plots.

3.2.3 Results 3.2.3.1  Morphological Traits As shown in Fig. 3.7A nganda and erecta types had older trees than the commercial and hybrid cultivars (Fig. 3.7A). The nganda types had the highest mean stem number (6.63), second was erecta (5.09), and the hybrids with a mean of 3.25 (Fig.  3.7C) had the lowest. The hybrids leaves were the longest with a mean of 19.94 cm and next to them were the commercial cultivars with a mean of 19.17 cm but nganda with a mean of 17.89 cm was the smallest. The hybrids had the most narrow leaves with a mean of 6.96 cm followed by nganda of 7.36 cm, and erecta types with a mean of 7.47 cm. Commercial cultivars had the highest number of fruits with a mean of 21.93 but hybrids had the lowest of 16.49 (Fig. 3.7B; Table 3.9). The ranges of all morphological values were highest in nganda and erecta types indicating high variability in all the traits measured in the two landraces. The overall mean values were almost similar among the nganda and erecta types. Tree girth increased with tree age (Fig. 3.8A) whereas vigorous trees were the most productive ones. Hybrids and landraces were unrelated in tree morphology, as is the case with nganda plant type (Fig. 3.8B). Hybrids and commercial cultivars were neither related to one another (Fig. 3.8B) but the commercial types were more closely related to erecta than to nganda types. However, the erecta and nganda types were closely related to each other. Commercial types had the longest, internodes and leaves, highest number of fruits but fewer stems. Hybrids had the youngest trees, were the most vigorous and highest producers. Most of the variability was defined by hybrids, followed by commercial types and then by erecta. The information shown in Fig.  3.8A and B is confirmed by the genetic distances in Table  3.10. Mahalanobis genetic distance was longest ­ between hybrids and nganda but was shorter between

Table 3.9  Morphological diversity of 476 cultivated Robusta coffee types shown in Fig 3.7 Types

Statistic

Age

Girth

PC

Vigour

Stems

IL

LL

LW

Fruits

Com No. 44

Range Mean s.e. Range Mean s.e. Range Mean s.e. Range Mean s.e. Range Mean s.e.

30 11.68 1.41 0 12 0 76 37.01 1.58 75 37.39 1.49 76 33.95 1.03

68 46.55 2.72 25 47 2.90 159 69.77 2.25 205 74.30 2.39 205 68.93 1.52

3 3.73 0.10 2 3.56 0.16 4 3.30 0.05 4 3.26 0.05 4 3.33 0.03

2 3.55 0.09 2 3.56 0.16 4 3.23 0.04 4 3.18 0.04 4 3.24 0.03

10 4.09 0.30 3 3.25 0.21 21 5.32 0.24 27 6.63 0.26 27 5.76 0.16

4.7 6.39 0.19 1.6 5.09 0.12 6.8 6.21 0.09 5.9 5.98 0.07 6.8 6.08 0.05

14.3 19.17 0.40 7.8 19.94 0.43 21.2 18.35 0.19 13.7 17.89 0.15 21.2 18.26 0.11

5.2 7.38 0.16 3 6.96 0.21 6.8 7.47 0.08 9.5 7.36 0.08 3.1 7.39 0.05

18.3 20.93 0.56 10.1 16.49 0.60 27.8 20.62 0.36 31.4 19.92 0.30 31.4 20.19 0.21

Hybrids no. 16 Erecta no. 191 nganda no. 224 All types No.476

Key: com-commercial types; s.e, standard error; PC, production capacity (scale 1–5); IL, inter node length (cm); LL, leaf length (cm); LW, leaf width (cm). Reproduced with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE   115

2

Stems

1

IL Fruits Vigour PC

Commercial

F2 (26.01%)

F2 (21.38%)

Tree age Girth

LL LW

erecta

0

–1

nganda

0

1

Hybrids

–1

(A)

F1 (24.48%)

(B)

2

–2 F1 (69.40%)

Centroids

Fig. 3.8  PCA relationships between (A) morphological characters and (B) cultivars. Reproduced with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

erecta and nganda. Hybrids were distinct populations from nganda and erecta landraces as they had significantly longer Mahalanobis genetic distances of 4.40 and 4.06, respectively, from the two landraces. Commercial types and hybrids had a Mahalanobis genetic distance with a value of 2.79 close to 3.0, which nearly categorized the two as distinct populations. Mahalanobis distances between the landraces erecta and nganda of 0.20, that between commercial and erecta types of 0.84 and that between commercial types and nganda of 1.50, were not significant and therefore, all the three types did not qualify to be classified as separate populations. Equally, Fisher estimates showed that the longest distance just as with Mahalanobis, was between the hybrids and nganda, while the shortest was between erecta and nganda. The probability that any two groups were significantly different from each other is indicated by the Fisher estimates (Table 3.10). Tree diameter increased with age (Fig. 3.9A). Younger trees of 11– 20 years old had slightly less longer leaves than those of 61–70-yearold trees. Throughout the tree life, stems increased with age but trees that were 60–70 years old had fewer leaves and the longest internode length. Trees grew faster at elevation 1101–1200 m above sea level (Fig.  3.9B). Productive and vigorous trees that produced more fruits were found at 1301–1400 m above sea level. At 1101–1200 m above sea level, trees were less productive and vigorous but had the longest leaves.

3

Table 3.10  Mahalanobis and Fisher Distances Estimated in Robusta Types Grown in Uganda Mahalanobis Distance Commercial erecta Hybrids nganda

p-Values for Fisher Distances

Fisher Distance

Commercial erecta Hybrids nganda Commercial erecta Hybrids nganda

Commercial

erecta

Hybrids nganda

0 0.84 2.79 1.50

1 <0.0001 <0.0001 <0.0001

<0.0001 1 <0.0001 0.003

<0.0001 <0.0001 1 <0.0001

0.84 0 4.06 0.20

2.79 4.06 0 4.40

1.50 0.20 4.40 0

0 4.98 5.40 9.09

Reproduced with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

4.98 0 9.89 3.33

5.40 9.89 0 10.84

9.09 3.33 10.84 0

<0.0001 0.003 <0.0001 1

Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE   117

75

3.8

65

3.7

8

75

6

55

5

35

Values

7 55 Values

Centimeters

95

3.4

35

3.3 3.2

25 3

4-10 Y 11-20 Y 21-30 Y 31-40 Y 41-50 51-60 61-70 71-80 Robusta coffee tree age range (years)

(A)

Girith

LL

Stems (numbers)

IL (cm)

3.5

45

4

15

3.6

Rating (numbers)

3.9

9

115

3.1

15 3 687–10001001–11001101–12001201–13001301–14001401–15001501–1654 Altitude range (masl) PC Age (years) Girth (cm) LL (cm) Vigor

(B)

Fig. 3.9  Influence of age and elevation on morphological traits at (A) tree age vs traits and (B) elevation vs traits. Reproduced with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

3.2.3.2  Multivariate Analyses of Phenotypic Parameters In Fig.  3.10A, PCA1 axis, on the positive side, factor 1 was composed of number of stems, whereas PCA2 axis was dominated by older trees with wider girth, longer, wider leaves and with longer internodes contributing 21.90% variance. On the negative side of PCA2 axis, factor 2, was dominated by higher producing vigorous trees with many fruits. Altitude appears to have no significant effects on morphological characters. The factorial discriminant analysis in Fig. 3.10B confirms the PCA information in Fig. 3.10A. Based on PCA information in Fig. 3.10A, three distinct groups were formed. Group 1 with a negative variance, comprised of many productive, young, and vigorous trees with fewer stems. Group 2 consisted of vigorous productive and young

Fig. 3.10  Principal component analysis of (A) morphological traits with altitude and (B) phenotypic groups. Key: PC, production capacity (scale 1–5); IL, inter node length (cm); LL, leaf length (cm); LW, leaf width (cm); (A) F1-Principal component 1; F2-Principal component 2; (B) 1,2,3- phenotypic groups. Reproduced with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

118  Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE

trees that had long internodes, broad leaves, and higher number of fruits and stems. Cluster 3 with a positive variance was dominantly made up of older trees with a greater girth. Group 2 positioned between groups 1 and 3 had the lowest variance. It is clear that groups 1 and 3 were the most distant, while groups 1 and 2 were the closest. There was close association between number of fruits and age with girth on one hand and altitude with production capacity, vigor on the other. Table 3.7 shows the placement of 476 accessions that were sampled initially. The placement was done using the confusion matrix which grouped the accessions according to coffee types where they ranged from 6.82% for commercial cultivars to 62.50% for nganda landraces with a mean of 36.73% (Table 3.11). In the correct placement, out of the initial 27 trees rated as commercial cultivars, only 3 were correctly placed, 27 were identified as erecta landraces, 11 were grouped as nganda while 3 types were placed as hybrids. Arranging the accessions into multivariate grouping, the number of accessions retained in the correct grouping, ranged from 91.77% to 97.77% with a mean of 95.27% (Table 3.11). Table 3.12 compares the correct placement of groups by coffee types with PCA multivariate arrangement. Most variables, starting with vigor, production capacity, leaf width to age, in both coffee types and multivariate grouping could be classified as distinct populations, with cumulative percent variance ranging from 100%–78.93% to 100%–83.64%, respectively (Table 3.12). The mean group abundance variance for vigor and production capacity was higher in hybrids and commercial types than in erecta and nganda types whereas abundance values for girth and age were higher in the landraces than in the commercial and hybrid cultivars, a fact also reported in Figs. 3.8–3.10A and B. Hybrids and commercial types, using production capacity and vigor could only be placed in group 1 as indicated by their higher mean abundance values for these two attributes in the multivariate grouping. Using girth and age as group formation criterion, nganda and erecta would most likely fall under groups 2 and 3. As regards morphological characters, girth, number of fruits per tree, and leaf length in both coffee type and multivariate groups were the most variable attributes.

3.2.3.3  Physical Bean Parameters and Multivariate Analysis A gram of green beans was roasted for 0.05 s at the lowest time but for 0.38 at the longest and had a mean of 0.11 (Table 3.13). After roasting, green beans lost weight that ranged from 4.30% to 51.59% and that had an average of 15.29%. Roasted beans increased from 11.80% to 180.0% and had an average of 70.98%. A total of 100 seed weight

Table 3.11  Genotype Placement of 476 Accessions With Confusion Matrix According to Type and Morphological Traits Coffee Types

Multivariate Grouping

From\To

Commercial

erecta

Hybrids

nganda

Total

Commercial erecta Hybrids nganda Total

3 1 0 0 4

27 89 6 84 206

3 2 5 0 10

11 100 5 140 256

44 192 16 224 476

Reproduced with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

% Correct 6.82 46.35 31.25 62.50 Mean: 36.73

From\To

1

2

3

Total

% Correct

1 2 3 Total

219 10 0 229

5 156 3 164

0 4 79 83

224 170 82 476

97.77 91.77 96.34 Mean=95.27

Table 3.12  Multivariate Phenotypic Traits Grouping Estimated From Bray-Curtis Distance Showing Similarity Percentages of the 476 Accessions Mean Group Abundance Cumulative % Variance Contribution

Girth Age Fruits Stems LL IL LW PC Vigor

Coffee Types

Multivariate Groups

Coffee Types

Multivariate Groups

Commercial

erecta

Hybrids

nganda

1

2

3

45.41 78.93 85.92 90.74 94.59 96.43 98.19 99.15 100

48.75 83.64 89.02 93 95.9 97.27 98.62 99.35 100

46.5 11.7 20.9 4.09 19.2 6.39 7.38 3.73 3.55

69.6 36.9 20.6 5.33 18.3 6.2 7.47 3.3 3.22

47 12 16.5 3.25 19.9 5.09 6.96 3.56 3.56

74.3 37.4 19.9 6.63 17.9 5.98 7.36 3.26 3.18

43.6 15.2 20.2 4.73 18.4 6.06 7.39 3.46 3.32

75 43.5 20 6.94 17.7 6.01 7.34 3.21 3.18

126 65.5 20.7 6.11 19 6.27 7.5 3.23 3.16

Key: PC, production capacity (scale 1–5); IL, inter node length (cm); LL, leaf length (cm); LW, leaf width (cm); and F, principal component. Reproduced with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE   121

Table 3.13  Green Bean Physical Characteristics of 204 Robusta Accessions Minimum Maximum Mean s.e.

RTPG

GB%WT dec

RB%VL inc

SDD

scr18

scr15

≥scr15

≤scr12

0.05 0.38 0.11 0

4.3 51.59 15.22 0.4

11.8 180 70.98 1.33

5.1 28.8 14.55 0.22

0 91.9 19.03 1.62

7.1 94.1 61.27 1.57

7.1 100 80.3 1.63

0 92.9 19.87 1.64

Key: RTPG, roast time per gram (seconds); GB, roasted green bean percentage weight decrease; RB, roasted green bean percentage weight increase; SSD, seed density; S18, screen size 18; S15, screen size 15; S≥15, screen size 15 and above; S≤12, screen size 12 and below; s.e., standard error. Printed with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

varied between 5.10 and 28.80 g and had an average of 14.55 g. Some green beans could not fit into of screen sizes of 18% and 12%. The average screen size 15 contained a minimum of 7.1% green beans and with a maximum of 94.10% and had a mean of 61.27%. Screen size 15 (>) contained a minimum of 7.1% of green beans and with a mean of 80.30. Screen size 12 contained the highest percentage of green beans at 92.90% and had a mean value of 19.87%. Fig. 3.11A–G shows that the means, median, variance, interquartile range within attributes of the three groups were variable. Group 1 had high variance values for screen sizes 15, 15 (>), medium seed density, and high volume increase but had the lowest roast time. In group 2, the variance and interquartile range for screen size 12 and for some screen size 15 beans was high, but the seed density was the lowest (Fig. 3.11A–G). In group 3 the mean, variance, and interquartile range for roast time, roast bean volume increase, seed density, and screen size 18 attributes were consistently high (Fig. 3.11A–G). The three groups had Mahalanobis distances of greater than 3 meaning that they were significantly different (Table 3.14). The Fisher probability values were also highly significant supporting the evidence given by Mahalanobis distances. Bean physical characteristics, Mahalanobis and Fisher distances showed that groups 2 and 3 were the most distant whereas groups 1 to 3 were the closest (Table 3.14). Fig.  3.12A and B shows the relationship between bean physical characters, altitude, and multivariate groups. Group 1 consisted of young trees that had high bean weight loss, lesser roast time, and had mainly a screen size of 15 (>). Group 2 consisted of older trees that had predominantly small beans of screen size 12 (<) but had a higher roast time and large volume increase. Group 3 beans consisted of larger beans of screen sizes 15 (>) and 18, which were denser, but with

122  Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE

(B)

(A)

(E)

(C)

(F)

(D)

(G)

Fig. 3.11  (A–G) Green bean physical character diversity analysis by use of box plots. Reproduced with permission from Aluka, P., 2013. PhD thesis, University of Nairobi.

Table 3.14  Green Bean Physical Characteristics as Separated by Mahalanobis and Fisher Distances Mahalanobis Distances

p-Values for Fisher Distances

Fisher Distances

Group

1

2

3

1

2

3

1

2

3

1 2 3

0 10.01 5.49

10.01 0 13.26

5.49 13.26 0

0 55.29 30.32

55.29 0 73.25

30.32 73.25 0

1 <0.0001 <0.0001

<0.0001 1 <0.0001

<0.0001 <0.0001 1

Reproduced with permission from Aluka, P., 2013. PhD Thesis, University of Nairobi.

F2 (29.69%)

-- axis F2 (25.62 %) -->

Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE   123

Centroids, -0.6586, 1.2395

Centroids, 2.1173, -0.2773

Centroids, -1.4587, -0.9622 F1 (34.26%)

(A)

-- axis F1 (25.90 %) -->

(B)

Centroids Series5 Series6 Series7

Fig. 3.12  (A,B) Bean physical character as explained by (A) principal component analysis and (B) step discriminant analysis. Key: RTPG, roast time per gram (seconds); GB%WT decrease, roasted green bean percentage weight decrease; RB%VL increase, roasted green bean percentage volume increase; SDD, seed density; S18, screen size 18; S≥15, screen size 15 and above; S15, screen size 15; S≤12, screen size 12 and below. Reproduced with permission from Aluka, P., 2013. PhD Thesis, University of Nairobi.

a ­medium roast time. Denser beans were positively associated with higher screen sizes. Groups 2 and 3 were the most distant while groups 1 and 3 were the closest. The groups that were correctly placed by confusion matrix were between 58.62% and 92.45% with a mean of 82.44% (Table  3.15). Fig.  3.12B confirms that groups 2 and 3 were the most distant while groups 1 to 2 were the closest. Table  3.16 shows similarity percentages of variance for the various bean physical attributes and collaborates the results of Fig.  3.12A–G and Table  3.15. The cumulative variance indicates that roasting characteristics were more important in separating the groups as their cumulative variances ranged from 100% for RTPG to 95.16% for roasted green bean percentage and volume increases (RB%VL) but variances of screen sizes ranged from 82.26% for screen size 18% to 22.27% for screen size 15 (>). RTPG in seconds was the most dominant character that could separate the groups, on the basis of their dissimilarity but screen size 15(>) with the lowest cumulative variance % was an indication of similarity of groups. Nevertheless, screen sizes, 15, 12 (<), and 15 (>) contributed the most variance. Group 1 was characterized by beans of screen sizes, 15 and 15(>) that had high percent volume increase as indicated by the mean group variance. Group 2 had beans of three screen sizes, namely, 15, 15(>), and 12(<) with that also had high percent volume increase, whereas group 3 was beans of sizes, 15 and 18 with a slightly longer roast time of 0.13 s (Table 3.16).

124  Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE

Table 3.15  Confusion Matrix Group Placement and Pair Wise Distances of Green Bean Physical Characters Confusion Matrix From\To

1

2

3

Total

% Correct

1 2 3 Total

115 10 3 128

0 45 1 46

10 1 19 30

125 56 23 204

92.45% 77.14% 58.62% 82.44%

Reproduced with permission from Aluka, P., 2013. PhD Thesis, University of Nairobi.

Table 3.16  Green Bean Physical Characteristics Estimated by Bray-Curtis Distance and Showing Similarity Percentage Pooled Groups

Variance

Mean Group Abundance

Taxon

Contribution

Cumulative %

1

2

3

Screen size ≥15 Screen size ≤12 Screen size 15 Screen size 18 RB%VL increase Seed density GB%WT decrease RTPG

6.46 6.46 5.76 5.18 3.74 0.80 0.60 0.01

22.27 44.54 64.39 82.26 95.16 97.91 99.98 100

92.2 7.8 72.3 19.9 67.6 15.1 16 0.10

51.4 48.6 49.5 1.88 73.6 11.6 14.4 0.12

86 14 29.9 56.1 83 18.6 13.2 0.13

Key: RB%-VL increase, roasted bean % volume increase; GB%WT decrease, roasted bean % weight decrease; RTPG, roast time per gram (seconds). Reproduced with permission from Aluka, P., 2013. PhD Thesis, University of Nairobi.

3.2.4 Discussion In coffee as in many other tree crops, the effects of genotype and environment interactions mask phenotypic differences between genotypes. To discern the variance effects of the different components, it becomes necessary to subject the PCA information to further

Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE   125

­ ultivariate analysis as can be seen when the data in Fig. 3.8A and B m is subjected to factorial step discriminant analysis in Fig. 3.10A and B. Montagnon et  al. (1998); Van der Vossen (1985) and Leroy et  al. (2006a,b) demonstrated how difficult it is to measure the intraspecific variations of vegetative traits such as leaf and fruit characters even with PCA because of the effects of the environment. Despite nganda and erecta types having a higher number of stems and longer internode length, the commercial and hybrid types had a much higher production capacity (Fig.  3.7C). Leaf length rather than the number of fruits per tree appears has been responsible for the higher production capacity in commercial and hybrid types, perhaps because of the increased leaf area and larger photosynthetic capacity (Fig. 3.7B). Commercial types also had the longest internodes which seemed to offer adequate space for fruit nodes and for formation of secondary branches. Girth width increased with age and was highest in nganda and erecta types. In the PCA analysis shown in Fig. 3.8A production capacity, vigor, and number of stems defined factor 1 variance whereas, leaf length, leaf width internode length, and number of fruits defined factor 2 variance. From PCA, analysis shown in Figs. 3.8–3.10A and B, it is clear that nganda and erecta types were closely related to each other with a D2 value of less than 3 and an insignificant Fisher value (Table 3.10) but both local landraces were significantly different from hybrid types. Commercial types appeared to be more closely related to the former two types than that to the hybrids. As has been reported in previous studies (Thomas, 1940; Leakey, 1970; Maitland, 1926) there is increased intercrossing between these two landraces and it has been suggested that erecta types arose as a result of farmer selection from nganda populations and commercial types may have arisen from hybridization between introduced cultivars from West-Africa and the land race cultivars or as reported by Kibirige-Sebunya et  al. (1996) they might have been progenies of a cross between Toro no. 9 tree with selected erecta types. The genetic variability in these landraces is enhanced especially in the Western highlands agroecological zones of Kamwenge, Kyenjojo, Bundibugyo, Kabarole, and Hoima in Lake Albert Crescent region where wild forms of Robusta grow alongside the cultivated types and where the practice of farmer sharing planting materials is common. With the nganda representing 62.50% and erecta 46.43% of landrace population (Table 3.11), the phenotypic variation in Robusta coffee would be expected to be dominated by these two coffee types. This confirms that earlier reports had indicated that the spreading nganda and the upright types were the predominant landraces of Robusta coffee varieties cultivated in Uganda (Leakey, 1970; Maitland, 1926; Musoli et al., 2009), which besides the indeterminate or d ­ eterminate habits, had similar morphological traits, and could be g­rouped ­genetically into one multivariate group (Fig.  3.8A and B; Tables 3.11

126  Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE

and 3.12). The indeterminate growth of nganda types is preferred by small-scale farmers because they require less pruning, weeding, and their large canopy maintains cool soil temperatures (Thomas, 1940). Preferential farmer selection for plant types has been accompanied by other traits such as increased vigor and tolerance to leaf diseases (Wrigley, 1988). At an elevation of 1301–1400 m above sea level, the trees were more vigorous and productive (Fig. 3.9B). At higher elevations, high leaf to fruit ratio may have resulted into a longer leaf life span which in turn may have increased bean nutrient supply (Musoli et al., 2009). The effect of low temperatures at higher elevations might have led to delayed ripening of flesh berries which in turn might have caused longer and better berry filling (Vaast et al., 2006). Physical bean characteristics namely, size homogeneity, density, shape, and color, affected the time it took to roast the beans. Larger and denser beans roasted for a longer period (Fig.  3.12A). Denser beans required a considerably longer roast time than lighter beans in order to achieve favorable flavor. Younger trees produced a screen size of 15 beans, which took a shorter time to roast and resulted in high bean weight loss (Fig. 3.12A) possibly because of inadequate bean filling. Prolonged roasting time might also have caused high bean weight loss due to excessive water loss and tissue burning (Bertrand et al., 2006). Overall, the mean green bean percentage weight decreases and increases in volume when roasted were 15.22% and 70.98%, respectively, as shown in Table 3.13 which is comparable to 16% weight loss and 50%–80% roast volume increase reported by Bertrand et al. (2006). The multivariate analysis of PCA produced three physical characteristics diversity groups as shown in Fig. 3.11A–G. Beans from group 1 were mainly of screen sizes, 15 and 15 (>), took the shortest time to roast and lost the least weight. These bean sizes also had high mean abundance variance, meaning there is a chance that selection would lead to a higher genetic advance compared to the other sizes (Table 3.16). The consistently lower mean and variance values shown by group 1 (Fig.  3.11A and B) was an indication of the high level of uniformity in the group accessions. Group 2 beans were from older trees that had screen size of 12 (<), was less denser, and had medium roasting time. Beans of smaller sizes might have arisen from inadequate farm management practices such as maintenance of old trees. Group 3 consisted of denser beans with a screen size of 18, a longer roast time, reduced roast weight loss but high roast volume increase. This screen size is ideal for markets given its weight, volume, and quality. The wide mean and variance in bean size in the multivariate group formation was an indication that wide variation for bean sizes that can be graded does exist in the local farms and could be exploited by the diverse markets.

Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE   127

The physical and chemical content of green beans are normally affected by the genotype, tree physiological age, harvesting period, and fruits to leaf ratio (Vaast et  al., 2006). Bean size development is faster in trees grown under the shade than in the sun grown coffee trees (Silva et al., 2005). Trees grown under the shade produce a high ratio of reducing sugars to sucrose than those grown in the sun, which stimulates cell division and elongation in young perisperm and endosperm tissues giving rise to larger beans (Vaast et al., 2006; Silva et al., 2005; Muschler, 2001). Seed development and physiology is influenced not only by the amount of light received but also by the position of the fruits within the branches (Vaast et al., 2006; Silva et al., 2005). Harvesting green cherries at the end of the season means a prolonged growing period that produces better quality and larger beans accruing from the accumulation of biochemical compounds needed for growth (Vaast et al., 2006; Silva et al., 2005). Trees grown under the shade for up to 1 month experience low temperatures that affect bean size, composition of chemical compounds, and cup quality positively (Vaast et al., 2006; Silva et al., 2005; Muschler, 2001). This means therefore, because the fruit continues to respire at a high rate for 2–3 weeks after harvesting, it is appropriate for coffee berries to be harvested once the pericarp is soft and is easily removable. The high genetic bean variability seen in the Ugandan landraces is most likely due to the gene flow between wild forms and the cultivated landraces (Musoli et al., 2009). Other processes such natural selection, farmer selection, and the highly diverse agroecological environments may have played a part in the evolution of this diversity. Physiological changes associated with old trees may have also been responsible for phenotypic variations in tree morphology and more so in the variability of physical bean characteristics.

3.2.5 Conclusion Most of the cultivated C. canephora in Uganda comprised of nganda and erecta landraces with hybrids and commercial types constituting only 10% of the genotypes. The local landraces, nganda and erecta were closely related to each other whereas the hybrid types had significantly larger genetic distances from the landraces. Girth diameter, production capacity, and vigor were the most important parameters defining the genotype characteristics. About 61% of the traded Robusta coffee beans had a mean screen size of 15 but 19% of the beans were of screen sizes larger than 15. Beans of screen sizes of 15, 15 (>), and 18, with a longer roast time, a lesser weight loss but a higher roast volume percent increase were the most abundant for the markets. Nevertheless, there is a high level of morphological and bean size variability in the landraces of Ugandan Robusta coffee that can be exploited for the crop improvement.

128  Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE

References Aga, E., Bryngelsson, T., Bekele, E., Bjorn, S., 2003. Genetic diversity of forest Arabica coffee (coffea arabica L.) in Ethiopia as revealed by random amplified polymorphic DNA (RAPD) analysis. Hereditas 138, 36–46. Bertrand, B., Vaast, P., Alpizar, E., Etienne, H., Davrieux, F., Charmetant, P., 2006. Comparison of bean biochemical composition and beverage quality of Arabica hybrids involving Sudanese- Ethiopian origins with traditional varieties at various elevations in Central America. Tree Physiol. 26 (9), 1239–1248. Billotte, N., Lagoda, P.J.L., Risterucci, A.M., Baurens, F.C., 1999. Microsatellite-enriched libraries: applied methodology for the development of SSR markers in tropical crops. Fruits 54, 277–288. Charrier, A., Berthaud, J., 1985. Botanical classification of coffee. In: Clifford, M.N., Wilson, K.C. (Eds.), Coffee: Botany, Biochemistry and Production of Beans and Beverages. Croom Helm, London, pp. 13–47. Charrier, A., Berthaud, J., 1988. Principles and methods in coffee plant breeding: Coffea canephora Pierre. In: Clarke, R.J., Macrae, R. (Eds.), Coffee: Agronomy. Elsevier Applied Science Publishers. vol. 4. London, New York, Great Britain, pp. 167–198. Clifford, M.N., 1985. Chemical and physical aspects of green coffee and coffee products. In: Clifford, M.N., Willson, K.C. (Eds.), Coffee: Botany, Biochemistry and Production of Beans and Beverage. Avi Publishing Company, Westport, CT, pp. 305–374. Combes, M.C., Andrzejewski, S., Anthony, F., Bertrand, B., Rovelli, P., Grasiozi, G., Lashermes, P., 2000. Characterization of microsatellite loci in Coffea arabica and related coffee species. Mol. Ecol. Notes 9, 1171–1193. Dice, L.R., 1945. Measures of the amount of ecologic association between species. Ecology 26, 297–302. Dufour, M., Hamon, P., Noirot, M., Risterucci, A.M., Brottier, P., Vico, V., Leroy, T., 2001. In: Potential use of SSR markers for Coffea spp. genetic mapping. Proc. of 19th International Conference on Coffee Science (ASIC). Trieste, Italy, CD-ROM. International Plant genetic Resources, 1997. Descriptors of Coffee (Coffea spp. and Psilanthus spp). IPGR, Rome. 206. Kibirige-Sebunya, I., Musoli, P., Aluka, P., Wetala, M.P.E., 1996. An update on a comparison among Robusta coffee (Coffea canephora Pierre) clonal materials and their seedling progenies at different levels of nitrogen. In: Tenywa, J.S., Adipala, E., Ogenga-Latigo, M.W. (Eds.), Improving Coffee Management Systems in Africa, Makerere University Press, Kampala, Uganda. pp. 79–87. Lashermes, P., Combes, M.C., Topart, P., Graziosi, G., Bertrand, B., Anthony, F., 2000. Molecular breeding in coffee (Coffee arabica L.). In: Sera, T., Soccol, C.R., Pandey, A., Roussos, S. (Eds.), Coffee Biotechnology and Quality. Kluwer Academic, Dordrecht, pp. 101–112. Leakey, C.L.A., 1970. The improvement of Robusta coffee in East Africa. In: Leakey, C.L.A. (Ed.), Crop Improvement in East Africa. Commonwealth Agricultural Bureaux, Farnham Royal, pp. 250–277. Leroy, T., Marraccini, P., Dufour, M., Montagnon, C., Lashermes, P., Sabau, X., Ferreira, L.P., Jourdan, I., Pot, D., Andrade, A.C., Glaszmann, J.C., Vieira, L.G.E., Piffanelli, P., 2005. Construction and characterization of a Coffea canephora BAC library to study the organization of sucrose biosynthesis genes. Theor. Appl. Genet. 111, 1032–1041. Leroy, T., Ribeyre, F., Bertrand, B., Charmetant, P., Dufour, M., Montagnon, C., Marraccini, P., Pot, D., 2006a. Genetics of coffee quality. Braz. J. Plant Physiol. 18 (1), 229–242. Leroy, T., Curbry, P., Durand, N., Dufour, M., De Bellis, F., Jourdan, I., Vieira, L.G., Musoli, P., Aluka, P., Marraccini, P., Pot, D., 2006b. In: Coffea spp. and Coffea canephora diversity evaluated with microsatellites and single nucleotide polymor-

Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE   129

phism. Lessons from comparative analysis. 21st International Conference on Coffee Science, Montpellier, France. pp. 53. Maitland, T.D., 1926. Coffee Robusta in Uganda. Uganda Protectorate Department of Agriculture Circular, No.14. Mohammadi, S.A., Prasanna, B.M., 2003. Analysis of genetic diversity in crop plants; salient statistical tools and considerations. Crop Sci. 43, 1235–1248. Montagnon, C., Guyot, B., Cilas, C., Leroy, T., 1998. Genetic parameters of several biochemical compounds from green coffee, Coffea canephora. Plant Breeding 117, 576–578. Muschler, R., 2001. Shade improves coffee quality in a sub-optimal coffee zone of Costa Rica. Agrofor. Syst. 51, 31–139. Musoli, P., Cubry, P., Aluka, P., Billot, C., Dufour, M., De Bellis, F., Pot, D., Bieysse, D., Charrier, A., Leroy, T., 2009. Genetic differentiation of wild and cultivated populations: diversity of Coffea canephora Pierre in Uganda. Genome 52 (7), 634–646. Nei, M., Li, W.H., 1979. Mathematical model for studying genetic variation in terms of restriction endonucleases. Proc. Natl. Acad. Sci. 76, 5269–5273. Peakall, R., Smouse, P.E., 2006. GenAlEx 6: genetic analysis in Excel. Population genetic software for teaching and research. Mol. Ecol. Notes 6, 288–295. Pouncet, V., Dufuor, M., Hamon, P., Hamon, S., de Kochko, A., Leroy, T., 2007. Development of microsattelites for Coffea Canephora and their transferability to other coffee species. Genome 50 (12), 1156–1161. Prakash, N.S., Combes, M.C., Dussert, S., Naveen, S., Lashermes, P., 2005. Analysis of genetic diversity in Indian robusta coffee genepool (Coffea canephora) in comparison with a representative core collection using SSRs and AFLPs. Genet. Resour. Crop. Evol. 52, 333–342. Risterucci, A.M., Grivet, L., N'Goran, J.A.K., Pieretti, I., Flament, M.H., Lanaud, C., 2000. A high-density linkage map of Theobroma cacao L. Theor. Appl. Genet. 101, 948–955. Rovelli, P., Mettulio, R., Anthony, F., Anzueto, F., Lasshermes, P., Graziosi, G., 2000. Microsatellites in Coffea arábica L. In: Sera, T., Socol, C.R., Pandey, A., Roussos, S. (Eds.), Coffee Biotechnology and Quality: Proceedings of the 3rd International Seminar on Biotechnology in the Coffee Agro-Industry. pp. 123–133. Londrina, Brazil. Saitou, N., Nei, M., 1987. The neighbour-joining method: a new method for reconstructing phylogenetic trees. Mol. Biol. Evol. 4, 406–425. Schneider, S., Roessli, D., Excoffier, L., 1999. Arlequin, Version 2.0: A Software for Population Genetics Data Analysis. Genetics and Biometry Laboratory, University of Geneva, Geneva. Silva, E.A., Mazzafera, P., Brunini, O., Sakai, E., Arruda, F.B., Mattoso, L.H.C., Carvalho, C.R.L., Pires, R.C.M., 2005. The influence of water management and environmental conditions on the chemical composition and beverage quality of coffee beans. Braz. J. Plant Physiol. 17, 229–238. Thomas, A.S., 1935. Types of Robusta coffee and their selection in Uganda. East Afr. Agr. J. 174. Thomas, A.S., 1940. Robusta coffee. In: Tothill, J.D. (Ed.), Agriculture in Uganda. Oxford University Press, London, pp. 289–313. UCDA, 2001–2003. 2007–2008. Uganda Coffee Development Authority Annual Reports. Vaast, P., Bertrand, B., Perriot, J.J., Guyot, B., Génard, M., 2006. Fruit thinning and shade influence on bean characteristics and beverage quality of C. arabica in optimal conditions. J. Sci. Food Agric. 86, 197–204. Van der Vossen, H.A.M., 1985. Coffee selection and breeding. In: Clifford, C., Wilson, K.C. (Eds.), Coffee: Botany, Biochemistry and Production of Beans and Beverage. Avi Publishing Company, Westport, CT, pp. 46–68. Wrigley, G., 1988. Coffee. Longman, New York. 639.

130  Chapter 3  GENETIC AND PHENOTYPIC DIVERSITY IN ROBUSTA COFFEE

Further Reading Leroy, T., De Bellis, F., Legnate, H., Kananura, E., Gonzales, G., Luiz Felipe Pereira, L.F., Andrade, A.C., Charmetant, P., Montagnon, C., Cubry, P., Marraccini, P., Pot, D., De Kochko, A., 2011. Improving the quality of African robustas: QTLs for yield and quality related traits in Coffea canephora. Tree Genet. Genomes. https://doi. org/10.1007/s11295-011-0374-6. Nei, M., 1987. Molecular Evolutionary Genetics. Columbia University Press, New York. Thomas, A.S., 1947. The cultivation and selection of Robusta coffee in Uganda. Empire J. Exp. Agric. 15, 65–81.