Genetic Resources and Crop Evolution

, Volume 61, Issue 1, pp 7–22 | Cite as

Identification of mango (Mangifera indica L.) landraces from Eastern and Central Kenya using a morphological and molecular approach

  • A. Sennhenn
  • K. Prinz
  • J. Gebauer
  • A. Whitbread
  • R. Jamnadass
  • K. Kehlenbeck
Research Article

Abstract

Local mangos (Mangifera indica L.) are highly valued for home consumption in rural Kenya and are regarded by the local population to be comparatively drought tolerant and less susceptible to pests and diseases than the improved varieties. These are characteristics which make them interesting for improvement and breeding purposes. To date, research on Kenyan mangos has mainly focused on introduced and commercial varieties, whereas information on local varieties and landraces is lacking. We present the first comprehensive morphological and molecular characterisation of local mango landraces from Eastern and Central Kenya. Thirty-eight local mango trees were sampled and characterized by 75 selected qualitative (44) and quantitative (31) morphological descriptors selected from the descriptor list developed for mango by Bioversity International (former IPGRI). Hierarchical cluster analysis was performed using first all variables and finally only 10 selected key descriptors. Additionally, dried mango leaves from the same 38 trees were used for molecular classification with 19 simple sequence repeat markers. Genetic relatedness between the mango samples was visualized using a dendrogram based on Nei’s genetic distance and Neighbor Joining methods. Morphological characterisation resulted in six distinct clusters, and molecular analysis in eight clusters, which partly supported the morphological classification. Four of the eight molecular clusters were consistent and molecular results confirmed morphological classification in these cases. Identification of local mango landraces using morphological traits can be considered as satisfying under field conditions, e.g. for rootstock identification in nurseries, but environmental conditions may influence the results. Molecular marker analysis is more expensive, but independent from environmental influences and thus suitable for identification of landraces in field genebanks and for breeding purposes. Outcomes of the present study can form the basis for urgently needed future conservation efforts, including circa situ conservation on farms and the development of ‘conservation through use’ strategies for local mango landraces in Kenya.

Keywords

Characterisation Conservation Genetic diversity Mangifera indica L. Morphological descriptor SSR 

Introduction

Mango (Mangifera indica L.) is of great importance in the tropics and subtropics (Singh 1996). Its social and economic impacts are most relevant in developing and emerging countries, where mango is a valued component in the diet, being rich in vitamins and minerals (Rocha Ribeiro et al. 2007). In Kenya, mango has been shown to have a positive effect on nutrition security and health (Gathambiri et al. 2010; Gitonga et al. 2010). Mango production in Kenya is an important part of the horticultural sector, and in terms of production, Kenya is ranked twelfth in the world and is the leading mango producer in Eastern Africa (FAOSTAT 2011). In 2010, about 540,000 t of mango fruits were produced (Horticultural Crops Development Authority, HCDA and Ministry of Agriculture, MoA 2011). The mango sub-sector provides employment and income opportunities for rural and peri-urban communities and mango farming has a promising potential to drive rural development in Kenya in particular in the semi-arid areas (ABD 2011).

Although mango is not native to Kenya, the trees have been cultivated in the country for centuries. Polyembryonic mango varieties were most probably first introduced to Kenyan’s Coastal region from India in the fourteenth century (Griesbach 2003). Local and farmer varieties (hereafter referred to as ‘landraces’; Zeven 1998) found in Kenya nowadays may have been derived from these early introduced varieties. Further landraces may have been introduced during colonial times, e.g. by Indian immigrants (Kehlenbeck et al. 2012). Improved cultivars from Florida (USA), Australia, Israel and other countries only reached Kenya during the 1980s (Griesbach 2003; ABD 2011).

Today, only about seven mango cultivars are frequently grown on a commercial scale in Kenya, these being the local landraces “Apple”, “Ngowe” and “Boribo”, which produce large, fibre-poor fruits, and the introduced cultivars ‘Kent’, ‘Sensation’, ‘Tommy Atkins’ and ‘Van Dyke’ (Kehlenbeck et al. 2012). In Eastern Kenya, mango production is based on the local landrace “Apple” and the improved cultivars ‘Kent’, ‘Tommy Atkins’ and ‘Van Dyke’ (ABD 2011). This low varietal diversity found on farms contrasts with the cultivar richness found in the motherblocks of national research stations and prison farms in the same region of Kenya where as many as 50 different cultivars (including some landraces) have been maintained (Kehlenbeck et al. 2010). Due to the difficulties in accessing grafted seedlings of the above mentioned commercial mango cultivars, small-scale farmers generally cultivate the traditional small-fruited landraces, which account for some 20–30 % of all mango trees in Eastern Kenya (ABD 2011; Kehlenbeck et al. 2012). Until now, however, no detailed information is available about the diversity and abundance of these local mango landraces in Kenya. The large number of local languages in the country leads to confusion about clear identification of mango landraces as many different names may exist for the same landrace. From the literature available in Kenya and the wider region, the focus is on the few economically important commercial cultivars mentioned above.

While the traditional landraces generally produce small and fibrous fruits with low market value, advantageous biological attributes such as high and stable yields, low management requirements, low susceptibility to pests and diseases and high drought tolerance were noted by Kehlenbeck et al. (2012). Because of these characteristics, rootstocks are usually sourced from local landraces for grafting commercial mango cultivars. The facts that commercial mango production in Kenya is dominated by a few commercial cultivars and that the numbers of small-fruited local mango landraces was found to significantly decrease (Sennhenn 2011) indicate that the local landraces in Kenya are threatened by genetic erosion. It is therefore necessary to conduct an inventory of this genetic diversity to explore, utilize and protect available genetic resources, which may be an important component of future mango improvement and breeding. Reliable and reproducible identification and characterization methods are prerequisite to develop conservation strategies for Kenyan local mango landraces and to make use out of their valuable genetic resources.

The main objective of the present study was to identify and characterize local mango landraces from Eastern and Central Kenya using morphological and molecular approaches which may contribute insights into the regional diversity and to develop future genetic resource conservation strategies. Further we aimed to identify morphological key descriptors and discuss the suitability of different identification approaches for different applications.

Materials and methods

Sampling

From a survey carried out on 90 farms in Eastern and Central Kenya (Fig. 1) as described by Sennhenn (2011), 38 mango trees were included in the sampling procedure for characterization. Sampling and morphological characterization was conducted during the main mango harvest season from January to March 2011. Sampling sites were selected according to the recommendations of district agricultural officers from the Ministry of Agriculture and key informants from Kenyan Agricultural Research Institute (KARI) research stations and farmer cooperatives. The geographic location of each of the sampled trees was recorded using a hand-held global positioning system (GPS) along with location information and local names of the surveyed trees (Table 1).
Fig. 1

Locations of surveyed mango farms (red circles, n = 90) and of sampled local mango trees (black crosses, n = 38) within different agro-ecological zones in Eastern (Machakos and Kitui Districts) and Central (Maragua and Murang’a Districts) Kenya. (Color figure online)

Table 1

Sample identifications (Sample ID), local names, their meanings (direct translation in square brackets) and the locations (GPS data) of the 38 local mango samples from Eastern and Central Kenya

Sample ID

Local mango name

Meaning

Province

District

Latitude [S]

Longitude [E]

Kav1

Kavumbu

[Local]

Eastern

Machakos

01.51743°

37.42172°

Kim2

Kimbole

[From Kitui]

Eastern

Machakos

01.51740°

37.43950°

Uga3_1

Uganda

Refers to origin

Eastern

Machakos

01.52012°

37.43794°

Mun3_2

Munyolo

Shape describing

Eastern

Machakos

01.52012°

37.43794°

Mom4

Mombasa

Refers to origin

Eastern

Machakos

01.50463°

37.44746°

Mun6

Munyolo

Shape describing

Eastern

Machakos

01.45788°

37.45338°

Kim8

Kimbole

[From Kitui]

Eastern

Machakos

01.44666°

37.46053°

Mun10

Munyolo

Shape describing

Eastern

Machakos

01.43276°

37.34212°

Kit13

Kitui

Refers to origin

Eastern

Machakos

01.37974°

37.23278°

Mun14

Munyolo

Shape describing

Eastern

Machakos

01.38278°

37.23447°

Kit15

Kitui

Refers to origin

Eastern

Machakos

01.39599°

37.28198°

Kit17

Kitui

Refers to origin

Eastern

Machakos

01.40835°

37.31725°

Mun19

Munyolo

Shape describing

Eastern

Machakos

01.38335°

38.26311°

Don20

Sikio la punda

[Donkey Ear]

Eastern

Kitui

01.35694°

37.94837°

Kim22

Kimunyi

[Local]

Eastern

Kitui

01.35316°

37.97176°

Wes23

Western

Refers to origin

Eastern

Kitui

01.34869°

37.99953°

Nai24

Nairobi prison

Refers to origin

Eastern

Kitui

01.34546°

37.98524°

Don25

Sikio la punda

[Donkey Ear]

Eastern

Kitui

01.40676°

37.98719°

Bor27

Boribo Mombasa

Boribo like plus origin

Eastern

Kitui

01.37912°

37.97685°

Bor28

Boribo Zanzibar

Boribo like plus origin

Eastern

Kitui

01.37030°

37.99565°

Bor29

Boribo small

Boribo like

Eastern

Kitui

01.34510°

37.99362°

Ndo31

Ndodo

[Dreaming]

Eastern

Machakos

01.42078°

37.43778°

Mun33

Munyolo

Shape describing

Eastern

Machakos

01.28183°

37.33985°

Kit39

Kituitamu

Origin plus [sweet]

Eastern

Machakos

01.37471°

37.36983°

Thi42

Thiururi

[Round]

Central

Murang’a

00.72557°

37.08424°

Ndo45

Ndodo

[Dreaming]

Central

Murang’a

00.72968°

37.16164°

Nju46

Njurufi

NA

Central

Murang’a

00.69597°

37.06177°

Nic51

Nice Thing

General description

Central

Maragua

00.68452°

37.01430°

Nja52

Njamba

NA

Central

Maragua

00.69036°

37.06169°

Thi53

Thiururi

[Round]

Central

Murang’a

00.70112°

37.11885°

Ngu58

Ngungugu

Refers to the smell

Central

Murang’a

00.72291°

37.13524°

Ngu64

Ngungugu

Refers to the smell

Central

Murang’a

00.69891°

36.96531°

Ngu66

NgunguguYaDodo

Refers to the smell plus dodo-like

Central

Murang’a

00.70862°

36.98944°

Ndo67

Ndodo

[Dreaming]

Central

Murang’a

00.67372°

37.11597°

Nun69

Nungungu

Refers to the smell

Central

Murang’a

00.92722°

37.12405°

Kag72_1

Kagege small

[Round] plus size

Central

Maragua

00.92722°

37.12045°

Bla72_2

Black Ndodo

Color plus [Dreaming]

Central

Maragua

00.92722°

37.12045°

Kag72_3

Kagege big

[Round] plus size

Central

Maragua

00.92722°

37.12045°

NA not available

Morphological approach

A total of 75 morphological characters including 31 quantitative and 44 qualitative traits were analyzed following the guidelines for mango descriptors published by Bioversity International (formerly IPGRI) for tree, leaf and fruit traits (IPGRI 2006). Characterization was conducted on-farm for tree and leaf attributes and in the laboratory of the World Agroforestry Centre (ICRAF) in Nairobi for further leaf and fruit characteristics. Ten leaves and 20 ripe fruits were randomly collected from each sampled tree as recommended by Bioversity International (IPGRI 2006). Only healthy and undamaged leaves and fruits were sampled.

Genetic approach

For molecular analysis, additional leaf material was collected from the 38 sampled trees. About five healthy and undamaged young, but fully developed leaves were picked from each of the trees. The leaves were washed gently with distilled water to remove all surface particles and stored in silica gel away from direct sunlight (Kit and Chandran 2010). In the lab, veins of the dried leaves were removed and genomic DNA in the remaining leaf material was extracted using the Cetyltrimethyl-ammoniumbromid (CTAB) method slightly modified after the protocol described by Kit and Chandran (2010). DNA samples were stored at −20 °C. The final DNA quantity was determined using size standards and gel electrophoresis.

Genetic analyses were performed using 19 microsatellite primers (Table 4) developed by Duval et al. (2005), Schnell et al. (2005) and Viruell et al. (2005). Forward primers were labeled with the fluorescent dyes 6-FAM (blue), VIC (green), PET (red) and NED (yellow) to allow coloading of PCR products in the capillary electrophoresis. Amplifications were carried out using 0.25 ng/μl 1x AmpliTaq Gold® PCRMaster Mix (Applied Biosystems) and 0.2 mM of each primer forward and reverse in a total volume of 20 μl with a thermocyler engine. Protocols and annealing temperatures were adjusted according to the first primer release publications (Duval et al. 2005; Schnell et al. 2005; Viruell et al. 2005). Capillary electrophoresis was undertaken in an ABI Prism 3100 Genetic Analyzer (Applied Biosciences, Carlsbad, California). The output was used to determine fragment lengths with GeneMapper 3.0 (Applied Biosystems, Carlsbad, California).

Data analysis

Morphological data were statistically analyzed using SPSS 19.0 and R 2.15.1 (IBM 2010; R 2008). Variations among the samples for each morphological trait were computed using the analysis of variance (ANOVA). Coefficients of variation (CV) were calculated for each quantitative parameter. Similarly, the relevance of qualitative variables was established with the Chi squared values which tested the independence between values. A hierarchical cluster analysis was conducted in order to group samples according to their morphological similarities. Clustering was performed after z-score standardization of qualitative and quantitative variables using squared Euclidean distances and the Ward method (1963).

Principal component analysis (PCA) and discriminant analysis were carried out for the identified clusters in order to find main variation trends and to evaluate their correlation which helps to identify those variables that contribute most to the fixed cluster formation of local mango samples (Abdi 2007; Abdi and Williams 2010). Such variables were selected as key descriptors. ANOVA and post hoc Tukey tests were performed for quantitative key descriptors and Chi squared tests for qualitative ones to detect significant differences between the clusters (Everitt and Hothorn 2011). Finally, the above described cluster analysis was repeated using only the selected key descriptors. The final cluster was compared with the first cluster to reveal usefulness of the key descriptors for classification purposes. Suitable key descriptors should lead to the same clustering results as including all morphological descriptors together. Significant differences of mean quantitative and qualitative parameters were again detected by ANOVA followed by post hoc Tukey-tests and Chi squared tests, respectively.

Standard genetic measures for each locus were assessed by number of alleles (A), effective number of alleles (Ae), observed (Ho), expected heterozygosity (He) and Wright’s inbreeding coefficient (FIS) calculated with GenePop 3.4 (Raymond and Rousset 1995) and Microsatellite Analyzer (Dieringer and Schlötterer 2003; Hardy 1908; Weinberg 1908; Wright 1951). Visualization of genetic relatedness was achieved by a multivariate PCA. The calculations based on quantitative data (fragments lengths for each marker and sample) were performed with GenAlEx 6.41 (Peakall and Smouse 2006). In addition, phylogenetic networks were constructed using the equal angle algorithm (Gambette and Huson 2008) based on Nei’s genetic distances between the local mango samples (Nei 1972, 1978). Neighbor Joining (NJ; Saitou and Nei 1987) and Neighbor-Net (Bryant and Moulton 2004) were used as agglomerative methods for the construction of phylogenetic trees to visualize the genetic relationships among the individual mango samples using the software packages Treeview (Page 1996) and Splitstree (Huson and Bryant 2006). An Analysis of Molecular Variance (AMOVA) was performed with Arlequin 3.5.1.2 (Excoffier and Lischer 2010) to assess the genetic variation within and among the groups of local mango samples based on squared Euclidean distances (Sneath and Sokal 1973).

Results

Morphological approach

Applying the morphological characterization method devised by IPGRI (2006) to the mangos collected from 38 trees, high variation in the selected morphological descriptors was found. While the method devised by IPGRI was usually sufficient to describe the fruit, additional descriptors were needed for adequate description of stone and fruit shapes. For ‘stone shape’, two additional levels ‘roundish shape’ and ‘heart shape’ were added, while for seed shape, ‘cashew nut shape’ was added. Characteristics related to the growth form of sampled trees showed limitations in their application for identification and characterization purposes in the field. Characteristics such as tree height and crown shape were highly influenced by the age of the tree and pruning activities of the farmers. Therefore, all seven tree-related descriptors suggested by IPGRI were excluded from the analysis. Morphological leaf characteristics did not show relevant differences between the phenotypes with low variation amongst all samples and were also excluded from the analysis. The most variable and hence suitable fruit-related quantitative descriptors were weight parameters such as fruit, pulp, skin, stone and seed weight while more homogenous characteristics included the thickness and width of fruits, stones and seeds. The cluster analysis conducted with all 56 finally selected descriptors (Table 2) (33 qualitative and 23 quantitative ones) resulted in the formation of six distinct clusters (results not shown). The number of clusters was chosen in alignment with the possibility of morphological discrimination reflected in the following results.
Table 2

56 selected morphological descriptors (33 qualitative and 23 quantitative ones) and their scale of measurement

Morphological descriptor

Scale of measurement

Fruit

Stalk attachment

Ordinal: weak = 1, intermediate = 2, strong = 3

Quantity of latex oozing from peduncle

Ordinal: absent = 0, low = 1, medium = 2, high = 3

Length

Scale: cm

Width

Scale: cm

Thickness

Scale: cm

Weight

Scale: g

Length/width ratio

Scale: [−]

Number of lenticels per cm²

Scale: [counts]

Shape

Nominal: oblong = 1, elliptic = 2, roundish = 3, ovoid = 4, obovoid = 5

Apex shape

Nominal: acute = 1, obtuse = 2, round = 3

Stalk insertion

Nominal: vertical = 1, slightly oblique = 2, oblique = 3

Depth of fruit stalk cavity

Ordinal: absent = 0, shallow = 1, medium = 2, seep = 3, very deep = 4

Neck prominence

Ordinal: absent = 0, slightly prominent = 1, prominent = 2, very prominent = 3

Slope of fruit ventral shoulder

Nominal: slopping abruptly = 1, ending in a long curve = 2, rising and then rounded = 3

Beak type

Nominal: perceptible = 1, pointed = 2, prominent = 3, mammiform = 4, reverse = 5

Sinus type

Nominal: absent = 0, shallow = 1, deep = 2

Skin color ripe fruit (ground color)

Nominal: green = 1, yellow green = 2, yellow = 3, yellow orange = 4, orange = 5, orange red = 6, red = 7, purple = 8

Skin color ripe fruit (flush)

Nominal: none = 0, yellow = 1, yellow orange = 2, orange = 3, orange red = 4, red = 5, purple = 6

Skin color unripe fruit

Nominal: green = 1, yellowgreen = 2, yellow = 3,

Skin surface texture

Nominal: smooth = 1, rough = 2

Skin waxiness

Nominal: waxy = 1, non-waxy = 2

Pulp

Weight

Scale: g

Fiber content (ratio of fiber to pulp)

Scale: % (w/w) of total pulp

Pulp content (ration of pulp to total fruit)

Scale: % (w/w) of total fruit

Juiciness

Ordinal: slightly juicy = 1, juicy = 2, very juicy = 3

Texture

Ordinal: soft = 1, intermediate = 2, firm = 3

Aroma

Ordinal: mild = 1, intermediate = 2, strong = 3

Presence of turpentine flavor

Ordinal: absent = 0, mild = 1, intermediate = 2, strong = 3

Color

Nominal: light yellow = 1, golden yellow = 2, yellow orange = 3, orange = 4, greenish yellow = 5, yellow = 6, light orange = 7, dark orange = 8

Adherence of fruit skin to pulp

Ordinal: absent = 0, weak = 1, intermediate = 2, strong = 3

Adherence fiber to fruit skin

Ordinal: low = 1, medium = 2, high = 3

Quantity of fiber

Ordinal: absent = 0, low = 1, medium = 2, high = 3

Fiber length

Ordinal: short = 1, medium = 2, long = 3

Pulp content (ration of pulp to skin plus stone)

Scale: [ration of pulp to skin plus stone]

Skin

Thickness

Scale: mm

Weight

Scale: g

Skin proportion (ration of skin to total fruit)

Scale:  % (w/w) of total fruit

Stone

Length

Scale: cm

Width

Scale: cm

Thickness

Scale: cm

Weight

Scale: g

Stone proportion (ration of stone to total fruit)

Scale: % (w/w) of total fruit

Shape

Nominal: ellipsoid = 1, oblong = 2, reniform = 3, heart shape = 4, roundish = 5

Veins

Nominal: level with surface = 1, depressed = 2, slightly elevated = 3, elevated = 4

Pattern of venation

Nominal: parallel = 1, forked = 2

Quantity of fiber on the stone

Ordinal: low = 1, medium = 2, high = 3

Texture of stone fiber

Nominal: soft = 1, coarse = 2

Length of stone fibers

Ordinal: short = 1, medium = 2, long = 3

Adherence of fiber to stone

Ordinal: weak = 1, intermediate = 2, strong = 3

Space occupied by seed inside the stone

Ordinal: <25 % = 1, 26–50 % = 2, 51–75 % = 3, 76–100 % = 4

Seed

Length

Scale: cm

Width

Scale: cm

Thickness

Scale: cm

Weight

Scale: g

Seed proportion (ratio of seed to stone)

Scale: % (w/w) of total fruit

Shape

Nominal: ellipsoid = 1, oblong = 2, reniform = 3, cashew = 4, tooth like = 5

The eigenvalues obtained by the PCA for the quantitative and qualitative morphological traits indicated that five components provided a good approximation of the data explaining 73 % of the total variance. Four variables showed high loadings on the first Eigenvectors, namely fruit weight, pulp weight, skin weight and seed length. The qualitative parameter ‘slope of fruit shoulder’ and ‘fruit ground colour’ had high loadings in the second and third Eigenvectors, respectively. The discriminant analysis of the cluster results revealed that the most differentiating morphological variables were fruit shape, fruit neck prominence, fruit slope of shoulder, fruit ground colour of ripe fruit, fruits stalk depth, fruit weight, fruit apex, fruit width to length ratio, fruit sinus, pulp weight, seed length, seed weight and stone proportion. Out of the seven important quantitative variables, fruit weight, seed length and fruit width to length ratio showed the most significant differences among clusters according to ANOVA followed by post hoc Tukey-tests (P < 0.05). Regarding qualitative variables, fruit shape, fruit apex, fruit stalk depth, fruit neck prominence, fruit slope of shoulder, fruit sinus and fruit ground colour were found to be most different according to contingency tables and Chi squared test (P < 0.05). Combining results of the PCA, the discriminant analysis and the ANOVA/Chi squared tests the three quantitative variables fruit weight, seed length and fruit length to width ratio and the seven qualitative traits fruit shape, fruit apex, fruit stalk depth, fruit neck prominence, fruit slope of shoulder, fruit sinus and fruit ground colour were selected as key descriptors and used for the repeated cluster analysis. This final analysis resulted in six clusters (Fig. 2) as in the first cluster analysis. However, four mango samples changed their groups. The finally identified six local mango types showed significant differences regarding the 10 key descriptors (Table 3). Cluster 1 grouped eight mango samples with small and egg-shaped fruits, which mostly had a completely yellow-orange skin when ripe. Most samples in cluster 1 had a shallow stalk depth, contrasting to the other clusters. In cluster 2, four mango samples with round and completely green fruits were grouped together (Table 3). Thirteen mango samples with round, but yellow-green fruits were grouped in cluster 3. In cluster 4, only three mango samples with the biggest and heaviest fruits of yellow-green skin color were included. Their weight was almost 3 times as much as those of cluster 1. Four samples with long fruits having curved fruit shapes and a typical deep sinus were grouped in cluster 5. Fruits from this mango type had the highest length to width ratio and were of intermediate weight. Unique was the acute shape of the fruit apex (Table 3). Cluster 6 grouped five samples characterized by oblong, almost quadratic shaped fruits without sinus. The apex of fruits in this cluster was obtuse and the skin color green without any flush (Table 3).
Fig. 2

Final dendrogram showing six clusters as a result of cluster analysis (Ward method, squared Euclidean distances, z-score standardization of variables) using ten morphological key descriptors on 37 local mango samples (sample Boribo Zanzibar was excluded from the cluster analysis because of outstanding quantitative morphological characteristics) collected in Eastern and Central Kenya. The cutting line for the cluster formation is marked as a dotted line

Table 3

Means and most frequent levels of three quantitative and seven qualitative key descriptors used for morphological classification of 37 local mango samples collected in Eastern and Central Kenya separately for the six identified clusters

Key descriptor

Cluster 1 (n = 8)

Cluster 2 (n = 4)

Cluster 3 (n = 13)

Cluster 4 (n = 3)

Cluster 5 (n = 4)

Cluster 6 (n = 5)

“Small and obovoid”

“Round and green”

“Round and small”

“Big and heavy”

“Long and curved”

“Long and straight”

Quantitative

Fruit weight (g)

108.6c (80–158)

159.0bc (113–235)

131.5bc (78–189)

295.1a (235–353)

168.2b (145–188)

178.6bc (158–231)

Seed length (cm)

4.7cd (4.2–5.1)

4.2d (3.7–4.9)

4.1d (3.8–4.4)

7.0a (6.3–9.0)

5.5bc (5.0–6.0)

6.0ab (5.5–6.9)

Fruit length/width ratio

1.3b (1.3–1.4)

1.0c (0.9–1.1)

1.0c (0.9–1.1)

1.4b (1.2–1.7)

1.6a (1.5–1.7)

1.5ab (1.2–1.8)

Qualitative

Fruit shape

Obovoid (100 %)

Roundish (100 %)

Roundish (100 %)

Oblong (100 %)

Oblong (100 %)

Oblong (60 %)

Fruit apex

Obtuse (100 %)

Round (100 %)

Round (100 %)

Round (100 %)

Acute (100 %)

Obtuse (100 %)

Fruit stalk depth

Shallow (100 %)

Absent (67 %)

Medium (77 %)

Absent (100 %)

Absent (100 %)

Absent (100 %)

Fruit neck prominence

Absent (100 %)

Absent (50 %) or slightly prominent (50 %)

Absent (92.3 %)

Slightly prominent (75 %)

Slightly prominent (75 %)

Slightly prominent (60 %)

Fruit slope of shoulder

Ending in a long curve (88 %)

Rising then rounded (100 %)

Rising then rounded (100 %)

Rising then rounded (100 %)

Slopping abruptly (50 %) or prominent (50 %)

Ending in a long curve (100 %)

Fruit sinus

Shallow (88 %)

Absent (100 %)

Absent (92 %)

Shallow (75 %)

Deep (75 %)

Absent (80 %)

Fruit ground colour

Yellow orange (88 %)

Green (100 %)

Yellow green (92 %)

Yellow green (100 %)

Green (75 %)

Green (80 %)

In brackets, the ranges for the quantitative descriptors and the percentages of occurrence for qualitative descriptors are given. One sample (Boribo Zanzibar) was identified as an outlier and was thus excluded from the cluster analysis

Means in a row followed by different letters are significantly different at P < 0.05 determined by an ANOVA followed by post hoc Tukey test

Genetic approach

In total, 91 alleles were found in 38 local mango samples using 19 SSR loci. While some of the amplified SSR loci sequences appeared conserved among the local mango samples, the majority of the sequences amplified by the primers were not conserved as shown by the presence and absence of some SSR loci sequences from one genotype to another. The number of alleles (A) ranged from three to nine per locus; and the effective number of alleles (Ae) ranged from 1.9 to 5.7 (Table 4). The fact that the observed heterozygosities (Ho) were lower than those expected (He) at most loci (Table 4) indicates a departure from the Hardy–Weinberg equilibrium (HWE)—a hypothesis that is supported by the high inbreeding coefficients (FIS) for some loci. However, the divergence was still small and relatively balanced between the different loci and possibly only a consequence of the existence of null alleles and/or the low sample size.
Table 4

Genetic diversity parameters for 19 loci estimated from 38 local mango samples collected in Eastern and Central Kenya

SSR locus

A

Ae

Size range

Ho

He

FIS

MiSHRS-1a

9

5.7

191–219

0.895

0.840

0.126

MiSHRS-32a

5

3.3

202–218

0.316

0.720

0.403

MiSHRS-37a

3

2.4

127–131

0.605

0.600

−0.078

MiSHRS-39a

3

2.4

343–359

0.895

0.600

−0.377

MiSHRS-44a

4

1.9

242–254

0.490

0.480

0.154

LMMA1b

4

2.7

198–208

0.840

0.640

−0.025

LMMA7b

6

4.0

199–213

0.840

0.760

−0.190

LMMA8b

4

3.2

254–266

0.790

0.730

0.011

LMMA9b

6

2.9

171–185

0.370

0.660

0.549

LMMA10b

6

2.5

153–177

0.680

0.600

−0.152

LMMA11b

4

2.7

230–248

0.740

0.630

−0.066

LMMA12b

5

3.3

198–206

1.000

0.710

−0.082

LMMA13b

4

3.3

179–199

0.790

0.710

0.038

LMMA15b

4

2.4

210–224

0.500

0.600

0.137

LMMA16b

5

3.5

232–243

0.740

0.720

0.000

MMiCIR14c

3

1.9

153–163

0.710

0.480

−0.216

mMiCIR22c

5

4.1

145–181

0.820

0.770

0.170

mMiCIR29c

5

2.4

175–191

0.580

0.600

0.095

mMiCIR32c

5

4.6

159–201

1.000

0.790

0.138

Total/mean

9

2.5

 

0.716

0.665

0.033

The following parameters are given: locus name, number of alleles (A), effective number of alleles (Ae), size range, observed heterozygosity (Ho), expected heterozygosity (He) and Wright’s inbreeding coefficient (FIS)

aSchnell et al. (2005)

bViruel et al. (2005)

cDuval et al. (2005)

A cluster analysis based on Nei’s genetic distance and the Neighbor Net performed with 38 local mangos revealed eight different clusters with two to 11 samples per cluster (Fig. 3). Two samples were identified as genetically identical in cluster I (samples Kav1 and Mun3_2) and another two samples in cluster V (Thi53 and Ngu64).
Fig. 3

Molecular clustering results (Nei’s genetic distance, NJ) for 38 local mango samples from Eastern and Central Kenya (sample names per cluster in the blackellipses) overlaid by the morphological clustering results (z-score standardization, squared Euclidean distance, Ward method; sample names in color are marking different clusters, see legend for the cluster names

An AMOVA found the highest molecular variance among groups (66.7 %). The molecular variance within groups was lower (33.3 %) and not significant (Table 5), which proves that the genetic consistency within the groups was relatively high but the groups themselves were genetically different to each other.
Table 5

Analysis of molecular variance (AMOVA) with grouping accordant to the previously defined eight clusters (Fig. 3) based on distance measures (** P ≤ 0.001)

Source of variation

d.f.

Sum of squares

Variance components

Variation (%)

P

Among groups

7

302.119

18.746

66.72

**

Within groups

30

130.854

14.362

33.27

NS

Total

37

432.974

13.252

  

NS no significance (P > 0.05), d.f. degrees of freedom

** = significance (P < 0.01)

Additionally, the expected heterozygosities (He) determined for the identified eight local mango clusters were fairly high and ranged from 0.414 in cluster VII to 0.667 in cluster V, which indicates a relative high genetic variability (Table 6).
Table 6

Expected heterozygosity (He) of eight different mango clusters and number of local mango samples per cluster (n) revealed by distance-based clustering across 19 SSR loci from 38 local mango samples

Distance-based cluster

n

He

I

7

0.440

II

3

0.589

III

3

0.498

IV

8

0.564

V

11

0.414

VI

2

0.553

VII

2

0.588

VIII

2

0.667

Comparison of morphological and molecular classifications

Classification of the sampled local mangos based on SSR markers (Fig. 3) partly supported the morphological classification as shown in Fig. 2. The morphological method grouped the local mango samples into six distinct landraces, whereas the molecular approach differentiated them into eight landraces. Some of the identified clusters seemed to be fairly consistent for both classification methods. The highest consistency was reached in the morphology-based cluster 1 (‘small and obovoid’), where 87.5 % of the samples also grouped in one single cluster by the molecular-based clustering approach (cluster I). A similarity of 75 % was reached each for the morphology-based clusters 4 (‘big and heavy’) and 5 (’long and curved’), where only one out of four samples each were missing in the clusters II and III, respectively, after applying the distance-based molecular approach (Fig. 3). Even the morphology-based clusters 3 (‘round and small’) and 2 (‘round and green’) were relatively stable as about 71 and 67 % of their samples grouped similarly in the molecular-based clusters V and IV, respectively. However, in the morphology-based cluster 6 (‘long and straight’), only 40 % of the samples grouped together in cluster VI after applying the molecular method (Fig. 3).

Discussion

Local mango names are not useful for identification purposes as mango samples with similar names appeared in different clusters, e.g. the six samples named “Munyolo” in three different morphology-based clusters (1, 5 and 6; Fig. 2) and even five molecular-based ones (clusters I, III, VI, VII and VIII; Fig. 3). However, some local names described important characteristics such as the supposed origin of the landrace or the shape of the mango fruit. Similar findings were reported by Maundu et al. (1999) and Nesbitt et al. (2010) who stated that local names of fruit landraces are of ethnobotanical or anthropological interest but their importance and usefulness for scientific identification purposes is limited.

The local mango samples characterized in this study displayed a considerable diversity for most of the selected morphological characters evaluated. Therefore, the mango descriptors published by Bioversity International (IPGRI 2006) can be considered useful for morphological characterization, particularly the fruit-related descriptors. However, tree characteristics heavily depend on pruning activities of farmers and environmental conditions rather than being characteristic for certain local mango landraces or farmers varieties—a fact also observed by other authors (Allard and Bradshaw 1964; Bradshaw 1965; Schlichting 1986). Furthermore, leaf characteristics showed a very high variation within one tree in this study, thus, limiting their differentiation potential. Consequently, fruit characteristics were identified to have the strongest discriminating power (Illoh and Olorode 1991; Rajwana et al. 2011; Mussane 2010; Gálvez-López et al. 2010).

Furthermore in the present study, fruit characteristics showed significant variation among the different local mango samples and were therefore considered as useful for the identifying particular landraces or morphotypes. Morphological diversity within local mango samples from Eastern and Central Kenya was very high and for some qualitative characteristics, such as stone and seed shape, even beyond the phenotypic classes given by Bioversity International (IPGRI 2006). The statistical analysis proved that in our study fruit related characteristics were most powerful to differentiate among local mango landraces (Table 3), therefore, a reduced number of descriptors (key descriptors) were considered to be sufficient for clear landrace identification. This approach is labor and resource efficient and allows easy on-farm identification. In addition to the scientific usefulness, fruit characteristics are of great importance for mango variety identification for commercial purposes. Morphological fruit descriptors are easy to assess and widely applied by farmers, traders, processors and consumers. Even nursery operators can easily identify fruits of desired landraces suitable as rootstocks. One disadvantage of using fruit characteristics for variety identification is the relatively limited time during the harvest season suitable for applying this method. Another problem is related to the fact that morphological observations alone are not considered effective enough for genotype identification because the phenotype depends on environmental and developmental factors (Vieira et al. 2007). Many factors such as climate and soil fertility of the location, as well as nutritional status, age and management of the plant individual influence the appearance of a certain plant and its parts (Allard and Bradshaw 1964; Bradshaw 1965). In particular, parameters related to sizes (e.g. tree size, fruit length and width, seed length) and weights are highly influenced by environmental conditions and therefore, their application for variety identification is considered critically (Hamrick and Godt 1989; Schlichting 1986). On the contrary, qualitative characteristics such as fruit shape, fruit apex and fruit stalk depth are less prone to influences from environmental factors but are considered to be subjective to a certain extent (Morell et al. 1995). As seven out of the 10 selected key descriptors in our study are qualitative ones (Table 3), the results of this study in respect to morphological classification are not strongly biased by environment.

Molecular characterization is more sensitive and unaffected by environmental conditions (Morell et al. 1995; Krishna and Singh 2007; Kumar et al. 2009; Kalia et al. 2011). In particular, SSR markers are very useful for variety identification in plant genetics and breeding because of their reproducibility, multiallelic nature, codominant inheritance, relative abundance and good genome coverage (Sharma and Majumdar 1988; Lavi et al. 1989; Krishna and Singh 2007; Varshney et al. 2005; Kumar et al. 2009).

In the present study, the identification of local mango types based on SSR markers partly supported the morphological classification as shown in Fig. 3. Only in one morphology-based cluster (cluster 6; Fig. 2) most of the samples grouped differently after applying molecular methods (Fig. 3). Discrepancies between morphological and genetic classifications can be explained by the impact of genotype × environment interactions on the phenotype (Vieira et al. 2007). The so-called phenotypic plasticity describes the ability of an organism to change its physiology and morphology in response to the prevalent environmental conditions (Schlichting 1986). However, estimation of environmental bias is difficult and morphological traits cannot serve as indicators for genetic distances since the degree of divergence is not necessarily correlated between genotypes and phenotypes (Hamrick and Godt 1989, Schlichting 1986). In addition, morphological characteristics such as fruit size and fruit skin color usually have a multigenic character, and the morphological expression is determined by multiple or additive genes (Sharma and Majumdar 1988; Lavi et al. 1989). On the contrary, molecular markers can be considered neutral and free of selection bias (Krishna and Singh 2007; Kumar et al. 2009; Varshney et al. 2005). Results from the present study support the suitability of the 19 selected SSR markers developed and used by different authors (Duval et al. 2005; Schnell et al. 2005, 2006; Viruell et al. 2005; Gálvez-López et al. 2009) for genetic analysis of local mangos from Central and Eastern Kenya. The combination of morphological and molecular characterization methods, however, provides a more detailed and clear description of new germplasm (Wortley and Scotland 2006). SSR molecular markers have the ability to reveal additional insights when morphological descriptors are insufficient to distinguish between landraces or varieties even when genetically close (Dos Santos Ribeiro et al. 2012). Many studies showed a variable degree of fitting of morphological and molecular findings depending on the methodology used and the studied organism; and there is not always a clear consensus of morphological and molecular identification results (Campos et al. 2005; Smykal et al. 2008; Mussane 2010).

Selecting the most suitable method for variety identification depends not only on the desired level of accuracy but also on the suitability for a certain application. Based on the results presented we recommend using the mentioned morphological key descriptors for fast and easy identification of landraces including desired rootstock types on farms, markets and nurseries. Molecular identification techniques such as SSR markers need to be applied for detecting relationships among landraces, for breeding purposes or genebank maintenance in order to guarantee reproducibility of results and independence of environmental influences. Genetic markers are also useful for eventually fingerprinting varieties/cultivars for indexing or naming purposes. This is most useful especially if new varieties are developed, to ensure that farmers or institutes have rights over their material. An advanced level of expertise is however required as well as relatively expensive laboratory equipment, and thus, molecular identification techniques are of limited relevance not only for farmers, nursery managers, traders or extension officers but often also for national research institutions in developing countries that lack access to well-equipped laboratories and financial funds to cover the costs related to the analysis.

In the present study the term landrace is applied to each cluster of local mango samples despite the slight inconsistencies in morphological and molecular clustering results. In the past, many researchers have attempted to define the term landrace. Harlan (1975) and Hawkes (1983) described a landrace as a highly diverse mixture of different genotypes which grow in a certain region and are adapted to a particular environment. The complex nature of landraces was recognized by Frankel and Soulé (1981), who differentiated the genetic diversity of landraces into two dimensions: between and within sites and populations. In the present study, each cluster of local mango samples can be considered as at least one single landrace characterized as a rather genetically dynamic unit representing a changing environment. Some of the eight molecular-based clusters particularly cluster II, IV and VIII may have been separated into more than one landrace each (Fig. 3) if more local mango samples had been included. The borders between the different landraces are rather dynamic than exclusive, which explains the slight discrepancies in morphological and molecular clustering results. The continuous introduction of new genetic material of mango—both local landraces and introduced cultivars—in the research area within the last decades and the ongoing vital exchange of germplasm by farmers further contributed to the mix of different mango landraces and the potential emergence of new ones. Landraces are results of many years of natural and artificial selection and can therefore be described as ‘farmers varieties’ or ‘local varieties’ as well (Bellon and Brush 1994). Characteristics such as open pollinating and outbreeding and high heterozygosity of mangos (Bally et al. 2009; Krishna and Singh 2007) result in a great diversity of locally adapted mango landraces favored by the diverse environmental conditions in the study area. These landraces—including the studied ones—have a great capacity to tolerate biotic and abiotic stresses resulting in high yield stability even under low agricultural input levels (Harlan 1975). These attributes are of great importance for future plant breeding and commercial use, e.g. for the development of rootstock varieties. Identification and characterization of local mango landraces are a prerequisite for conservation and exploitation of these genetic resources. The present study is the basis for future conservation efforts, including circa situ conservation on farms and the development of ‘conservation through use’ strategies. Circa situ conservation has been proven to be an effective strategy to link scientific research outcome with farmers needs in order to improve, maintain and use local varieties (Leakey and Akinnifesi 2008). Possible uses for local mango varieties could include drought-tolerant, locally adapted rootstocks for commercial mango production, aromatic fruits for processing into juice and jam, and late maturing landraces to broaden the harvest window and gain higher prices.

In addition the presented study has the potential to function as model for other naturalized fruit species such as guava (Psidium guajava), passionfruit (Passiflora edulis), avocado (Persea americana) and tree tomato (Solanum betaceum). It can be used as a methodological framework to further characterize valuable fruit genetic resources in the tropics to make better use out of their available potential in the future. Furthermore the methodology presented can lay the basis to standardize morphological characterization efforts on tropical fruit species to build up databases of genetic resources or to catalogue genetic diversity.

Notes

Acknowledgments

We thank SegoliLab at the International Livestock Research Institute (ILRI) in Nairobi, Kenya, for the technical support with the molecular marker analysis. Great thanks to Fatuma Ghelle from the Kenyan Agricultural Research Institute (KARI) Katumani in Machakos and the affiliated field assistants for the good cooperation. This study would not have been possible without the financial support of the GIZ-BEAF (Gesellschaft für Internationale Zusammenarbeit—Advisory Service on Agricultural Research for Development) that enabled the first author to perform the field work in Kenya.

References

  1. ABD (2011) The mango sub-sector in eastern region. The results of the mango tree census and baseline survey for eastern region. Final report. Institution Development & Management Services, Mombasa, KenyaGoogle Scholar
  2. Abdi H (2007) Discriminant correspondence analysis. In: Salkind NJ (ed) Encyclopedia of measurement and statistics. Sage, Thousand Oaks, pp 270–275Google Scholar
  3. Abdi H, Williams LJ (2010) Principal component analysis. Wiley Interdiscip Rev Comput Stat 2:433–459CrossRefGoogle Scholar
  4. Allard RW, Bradshaw AD (1964) Implications of genotype–environmental interactions in applied plant breeding. Crop Sci 4:503–508CrossRefGoogle Scholar
  5. Bally ISE, Ping L, Johnson P (2009) Mango breeding. In: Jain SM, Priyadarshan PM (eds) Breeding plantation tree crops. Springer, New York, pp 51–83CrossRefGoogle Scholar
  6. Bellon MR, Brush SB (1994) Keepers of maize in Chiapas, Mexico. Econ Bot 48:196–209CrossRefGoogle Scholar
  7. Bradshaw AD (1965) Evolutionary significance of phenotypic plasticity in plants. Adv Genet 13:115–156CrossRefGoogle Scholar
  8. Bryant D, Moulton V (2004) Neighbor-net: an agglomerative method for the construction of phylogenetic networks. Mol Biol Evol 21:255–265PubMedCrossRefGoogle Scholar
  9. Campos E, Espinosa MAG, Warburton ML, Varela AS, Villegas AM (2005) Characterization of mandarin (Citrus spp.) using morphological and AFLP markers. Interciencia 30:687–693Google Scholar
  10. Dieringer D, Schlötterer C (2003) Microsatellite analyser (MSA): a platform independent analysis tool for large microsatellite data sets. Mol Ecol Notes 3:167–169CrossRefGoogle Scholar
  11. Dos Santos Ribeiro IC, Lima Neto FP, Santos CA (2012) Allelic database and accession divergence of a Brazilian mango collection based on microsatellite markers. Genet Mol Res 11:4564–4574PubMedCrossRefGoogle Scholar
  12. Duval MF, Bunel J, Sitbon C, Risterucci AM (2005) Development of microsatellite markers for mango (Mangifera indica L.). Mol Ecol Notes 4:824–826CrossRefGoogle Scholar
  13. Everitt B, Hothorn T (2011) An introduction to applied multivariate analysis with R (use R), 1st edn. Springer, New YorkCrossRefGoogle Scholar
  14. Excoffier L, Lischer HEL (2010) ARLEQUIN suite ver 3.5: a new series of programs to perform population genetics analyses under Linux and Windows. Mol Ecol Res 10:564–567CrossRefGoogle Scholar
  15. Frankel OH, Soulé ME (1981) Conservation and evolution. London, CambridgeGoogle Scholar
  16. Gálvez-López D, Hernández-Delgado S, González-Paz M, Becerra-Leor EN, Salvador-Figueroa M, Mayek-Pérez N (2009) Genetic analysis of mango landraces from Mexico based on molecular markers. Plant Genet Res 7:244–251CrossRefGoogle Scholar
  17. Gálvez-López D, Salvador-Figueroa M, Adriano-Anaya ML, Mayek-Pérez N (2010) Morphological characterisation of native mangos from Chiapas, Mexico. Subtrop Plant Sci J 62:18–26Google Scholar
  18. Gambette P, Huson DH (2008) Improved layout of phylogenetic networks. IEEE/ACM Trans Comput Biol Bioinform 5:472–479PubMedCrossRefGoogle Scholar
  19. Gathambiri CW, Gitonga JG, Kamau M, Njuguna JK, Kiiru SN, Muchui MN, Gatambia EK, Muchira DK (2010) Assessment of potential and limitations of post-harvest value addition of mango fruits in Eastern Province: A case study in Mbeere and Embu Districts. In: Transforming agriculture for improved livelihoods through agricultural product value chains. Proceedings of the 12th KARI biennial scientific conference, Kenya Agricultural Research Institute, Nairobi, Kenya, pp 564–566Google Scholar
  20. Gitonga KJ, Gathambiri C, Kamau M, Njuguna K, Muchui M, Gatambia E, Kiiru S (2010) Enhancing small scale farmers’ income in mango production through agro-processing and improved access to markets. In: Transforming agriculture for improved livelihoods through agricultural product value chains. Proceedings of the 12th KARI biennial scientific conference, Kenya Agricultural Research Institute, Nairobi, Kenya, pp 1336–1342Google Scholar
  21. Griesbach J (2003) Mango growing in Kenya. ICRAF, NairobiGoogle Scholar
  22. Hamrick JL, Godt MJ (1989) Allozyme diversity in plant species. In: Brown ADH, Clegg MT, Kahler AL, Weir BS (eds) Plant population genetics, breeding and genetic resources. Sinaver Association, Sunderland, pp 43–63Google Scholar
  23. Hardy GH (1908) Mendelian proportions in a mixed population. Science 28:49–50PubMedCrossRefGoogle Scholar
  24. Harlan JR (1975) Our vanishing genetic resources. Science 188:618–621CrossRefGoogle Scholar
  25. Hawkes JG (1983) The diversity of crop plants. Cambridge, London, p 184Google Scholar
  26. Huson DH, Bryant D (2006) Application of phylogenetic networks in evolutionary studies. Mol Biol Evol 23:254–267PubMedCrossRefGoogle Scholar
  27. IBM Corp. Released 2010. IBM SPSS Statistics for Windows, Version 19.0. IBM Corp, Armonk, NY, USAGoogle Scholar
  28. Illoh HC, Olorode O (1991) Numerical taxonomic studies of mango (Mangifera indica L.) varieties in Nigeria. Euphytica 51:197–205CrossRefGoogle Scholar
  29. IPGRI (2006) Descriptors for mango (Mangifera indica). International Plant Genetic Resources Institute, RomeGoogle Scholar
  30. Kalia RK, Rai MK, Kalia S, Singh R, Dhawan AK (2011) Microsatellite markers: an overview of the recent progress in plants. Euphytica 177:309–334CrossRefGoogle Scholar
  31. Kehlenbeck K, Rohde E, Njuguna JK, Omari F, Wasilwa L, Jamnadass R (2010) Mango cultivar diversity and its potential for improving mango productivity in Kenya. In: Transforming agriculture for improved livelihoods through agricultural product value chains. Proceedings of the 12th KARI biennial scientific conference, Kenya Agricultural Research Institute, Nairobi, Kenya, pp 657–665Google Scholar
  32. Kehlenbeck K, Rohde E, Njuguna JK, Jamnadass R (2012) Mango production in Kenya. In: Valavi SG, Rajmohan K, Govil JN, Peter KV, Thottappilly G (eds) Mango, vol 2., Cultivation in different countriesStudium Press LLC, Houston, pp 186–207Google Scholar
  33. Kit YS, Chandran S (2010) A simple, rapid and efficient method of isolating DNA from Chokanan mango (Mangifera indica L.). Afr J Biotechnol 9:5805–5808Google Scholar
  34. Krishna H, Singh SK (2007) Biotechnological advances in mango (Mangifera indica L.) and their future implication in crop improvement—a review. Biotechnol Adv 25:223–243PubMedCrossRefGoogle Scholar
  35. Kumar P, Gupta VK, Misra AK, Modi DR, Pandey BK (2009) Potential of molecular markers in plant biotechnology. Plant Omics J 2:141–162Google Scholar
  36. Lavi U, Tomer E, Gazit S (1989) Inheritance of agriculturally important traits in mango. Euphytica 4:5–10CrossRefGoogle Scholar
  37. Leakey RRB, Akinnifesi FK (2008) Towards a domestication strategy for indigenous fruit trees in the tropics. In: Akinnifesi FK, Leakey RRB, Ajayi OC, Sileshi G, Tchoundjeu Z, Matakala P, Kwesiga FR (eds) Indigenous fruit trees in the tropics: domestication, utilization and commercialization. CAB International, WallingfordGoogle Scholar
  38. Maundu PM, Ngugi GW, Kabuye CHS (1999) Traditional food plants of Kenya. Kenya Resource Centre for Indigenous Knowledge, National Museums of Kenya, NairobiGoogle Scholar
  39. MoA , HCD (2011) Horticultural crops production report 2010. HCDA, MoA, Nairobi, Kenya. http://www.hcda.or.ke/Statistics/2010/2010%20Horticulture%20Validated%20 Report.pdf. Accessed 17 Nov 2012
  40. Morell MK, Peakall R, Apels R, Preston LR, Lloyd HL (1995) DNA profiling techniques for plant variety identification. Aust J Exp Agric 35:807–819CrossRefGoogle Scholar
  41. Mussane CRB (2010) Morphological and genetic characterisation of mango (Mangifera indica L.) varieties in Mozambique. M.Sc. Thesis, Faculty of Natural and Agricultural Sciences at the University of the Free State, South AfricaGoogle Scholar
  42. Nei M (1972) Genetic distance between populations. Am Nat 106:283–392CrossRefGoogle Scholar
  43. Nei M (1978) Estimation of average heterozygosity and genetic distance from a number of individuals. Genetics 89:538–590Google Scholar
  44. Nesbitt M, McBurney RPH, Broin M, Beentje HJ (2010) Linking biodiversity, food and nutrition: the importance of plant identification and nomenclature. J Food Compos Anal 23(6):486–498CrossRefGoogle Scholar
  45. Page RDM (1996) TreeView: an application to display phylogenetic trees on personal computers. Comput Appl Biosci 12:357–358Google Scholar
  46. Peakall R, Smouse PE (2006) GENALEX 6: genetic analysis in Excel. Population genetic software for teaching and research. Mol Ecol Notes 6:288–295CrossRefGoogle Scholar
  47. Rajwana IA, Khan IA, Malik AU, Saleem BA, Khan AS, Ziaf K, Anwar R, Amin M (2011) Morphological and biochemical markers for varietal characterization and quality assessment of potential indigenous mango (Mangifera indica) germplasm. Int J Agric Biol 13:151–158Google Scholar
  48. Raymond M, Rousset F (1995) GENEPOP Version 1.2: population genetics software for exact tests and ecumenicism. J Hered 86:248–249Google Scholar
  49. R Development Core Team (2008) R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.URL http://www.R-project.org. Accessed 3 May 2012
  50. Rocha Ribeiro SM, Queiroz JH, Lopes Ribeiro Queiroz ME, de Campos FM, Pinheiro Sant’ana HM (2007) Antioxidant in mango (Mangifera indica L.) pulp. Plant Foods Hum Nutr 6:13–17CrossRefGoogle Scholar
  51. Saitou N, Nei M (1987) The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol Biol Evol 4:406–525PubMedGoogle Scholar
  52. Schlichting C (1986) The evolution of phenotypic plasticity in plants. Annu Rev Ecol Syst 17:677–693CrossRefGoogle Scholar
  53. Schnell RJ, Olano CT, Quintanilla WE, Meerow AW (2005) Isolation and characterisation of 15 microsatellite loci from mango (Mangifera indica L.) and cross-species amplification in closely related taxa. Mol Ecol Notes 5:626–627CrossRefGoogle Scholar
  54. Schnell RJ, Brown JS, Olano CT, Meerow AW, Campbell RJ, Kuhn DN (2006) Mango genetic diversity analysis and pedigree inferences for Florida cultivars using microsatellite markers. JASHS 131:214–224Google Scholar
  55. Sennhenn A (2011) Morphological and molecular characterization of local mango varieties in Eastern and Central Kenya. M.Sc. Thesis, Georg-August University Goettingen, GermanyGoogle Scholar
  56. Sharma DK, Majumdar PK (1988) Further studies in inheritance in mango. Acta Hortic 231:106–111Google Scholar
  57. Singh RN (1996) Mango. ICAR, New DelhiGoogle Scholar
  58. Smykal P, Horacek J, Dostalova R, Hybl M (2008) Variety discrimination in pea (Pisum sativum L.) by molecular, biochemical and morphological markers. Theor Appl Genet 49:155–166Google Scholar
  59. Sneath PHA, Sokal RR (1973) Principles of numerical taxonomy. Freeman Publishing, San FranciscoGoogle Scholar
  60. Varshney RK, Graner A, Sorrells ME (2005) Genetic microsatellite markers in plants: their features and applications. Trends Biotechnol 23:48–55PubMedCrossRefGoogle Scholar
  61. Vieira EA, de Carvalho FIF, Bertran I, Kopp MM, Zimmer PD, Benin G, da Silva JAG, Hartwig I, Malone G, de Oliveira AC (2007) Association between distances genetic in wheat (Triticum aestivum L.) as estimated by AFLP and morphological markers. Genet Mol Biol 30:392–399CrossRefGoogle Scholar
  62. Viruell MA, Escribano P, Barbieri M, Ferri M, Hormaza JI (2005) Fingerprinting, embryo type and geographic differentiation in mango (Mangifera indica L., Anacardiaceae) with microsatellites. Mol Breed 15:383–393CrossRefGoogle Scholar
  63. Ward JH (1963) Hierarchical grouping to optimize an objective function. J Am Stat Assoc 58:236–244CrossRefGoogle Scholar
  64. Weinberg W (1908) Über den Nachweis der Vererbung beim Menschen. Jahreshefte des Vereins für vaterländische Naturkunde in Württemberg 64:368–382Google Scholar
  65. Wortley AH, Scotland RW (2006) The effect of combining molecular and morphological data in published phylogenetic analyses. Syst Biol 55:677–685PubMedCrossRefGoogle Scholar
  66. Wright S (1951) The genetical structure of populations. Ann Eugen 15:323–354CrossRefGoogle Scholar
  67. Zeven A (1998) Landraces: a review of definitions and classifications. Euphytica 104:127–139CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media Dordrecht 2013

Authors and Affiliations

  • A. Sennhenn
    • 1
  • K. Prinz
    • 2
  • J. Gebauer
    • 3
  • A. Whitbread
    • 1
  • R. Jamnadass
    • 4
  • K. Kehlenbeck
    • 4
  1. 1.Department for Crop Sciences, Crop Production Systems in the TropicsGeorg-August University GöttingenGöttingenGermany
  2. 2.Department for Forestry, Forest Genetics and Forest Tree BreedingGeorg-August University GöttingenGöttingenGermany
  3. 3.Sustainable Agricultural Production Systems with Special Focus on Horticulture, Faculty of Life SciencesRhine-Waal University of Applied SciencesKleveGermany
  4. 4.Domestication and Delivery, World Agroforestry Centre ICRAFTree DiversityNairobiKenya

Personalised recommendations