Morphological Characterization of Five Newly Developed Lines of Arabica Coffee as Compared to Commercial Cultivars in Kenya

MATERIALS AND METHODS

Study Site
The field trial was carried out at Tatu Estate in Ruiru District, Kenya from 2005 to 2009. The site lies within the upper Midland 2 agro-ecological zone (UM 2) at latitude 01°06’S and longitude 36°45’E and is approximately 1620 m above the sea level. The area receives a bimodal mean annual rainfall of 1063 mm and the mean annual temperature is 19°C (minimum 12.8°C and maximum 25.2°C). The soils are classified as a complex of humic nitisols and plinthic ferrasols. They are welldrained, deep reddish brown, slightly friable clays with murram sections occasionally interrupting. The soil pH ranges between 5 and 6 (Jaetzold and Schimidt, 1983).

Test Materials
The test materials included five advanced breeder’s lines coded CR8, 22, 23, 27 and 30 which were evaluated alongside two commercial Arabica cultivars, SL28 and Ruiru 11 as check cultivars. The five breeder’s lines have been developed by Coffee Research Foundation in Ruiru, Kenya. They are true breeding with tall stature and have been tested both in the lab and in the field and proven to be resistant to two major fungal diseases of coffee namely Coffee Berry Disease (CBD) and Coffee Leaf Rust (CLR).

Experimental Layout
The site was laid out in a Randomized Complete Block Design (RCBD) with twenty trees per plot planted on a spacing of 2x1.5 M and replicated three times. Field establishment was done in April/May 2005 and morphological data scored in October 2009.

Data Collection
Data were collected on morphological (both qualitative and quantitative) characters using a 25 character descriptors adopted from UPOV. Each qualitative descriptor was scored by observing twenty tagged trees per genotype taking six to seven plants from every block (replicate) and then calculating the mode to get an overall figure per replicate. Quantitative descriptors were taken as the mean value of three measurements made on six to seven trees per replicate and then calculating the mean to get an overall figure per replicate.

Data Analysis
The data was organized into a matrix and subjected to cluster analysis using R Statistical Software (Venables et al., 2006). Variables were segregated into discrete factors (e.g., Fruit colour - light red, pink, yellow); rank-ordered factors (e.g., fruit size - very small, small, medium, large and very large); integers (e.g., number of flowers per inflorescence) and numerical variable (e.g., average internode length). The clustering was done using DAISY (dissimilarity matrix calculation) function and unweighted pair-group method with arithmetic average [UPGMA] (Venables et al., 2006). Quantitative data were subjected to ANOVA using SAS version 9.1. Students-Newman-Keuls (SNK_5%) was used to separate the means. The procedure PRINCOMP was then used to perform a principle component (PC) analysis using the quantitative variables. In this procedure, first a similarity matrix was calculated and was used to calculate eigen values and scores for the accessions. The accessions were then plotted on two dimensions using the first two principle components (PC1 and PC2).

RESULTS

Variation in qualitative characters is presented in Table 1; there were minimal variations among all genotypes in most of the qualitative characters such as leaf shape, anthocyanin colouration and undulation of the margin, depth of the secondary veins, leaf domatia, fruit shape, fruit colour and adherence to the branch. Significant variations were found in plant shape and fruit size. CR8, 30 and SL28 were conical in shape while the rest were cylindrical. The fruit size ranged medium for CR27 to large for CR22, 23, SL28 and Ruiru 11 and very large for CR8 and CR30. The new lines were morphologically very similar to SL28 especially for plant height, canopy diameter and intensity of ramification. They were tall with large canopy diameter and medium intensity of ramification unlike Ruiru 11 which was short with medium canopy diameter and strong intensity of ramification.

Table 1:	Variation of qualitative characters in the seven coffee genotypes

Fig. 1:	Cluster dendrogram illustrating morphological diversity between the seven coffee genotypes characterized using 24 morphological descriptors

Fig. 2:	Kelley-Gardner-Sutcliffe (KGS) penalty function showing the similarity index for clustering

Results of the cluster analysis are presented in Fig. 1. Degree of similarity varied from 25 to 20% and four main groups were formed when the similarity index was considered for clustering (Fig. 2). The improved cultivar, Ruiru 11 and the traditional cultivar SL28 were clustered alone in separate clusters, while two new lines CR8 and CR30 clustered together. The other three new lines CR22, 23 and 27 also clustered together in shared sub-clusters. The R command g clus was used to reorder the genotypes within a cluster keeping them contiguous to each other and therefore Ruiru 11 and SL28 were the most distantly related.

Table 2:	Variation of quantitative characters in the seven coffee genotypes
NB: Fruit ripening was scored when some of the tagged branches had attained 100% ripening

Table 3:	The first two Principle Components (PC) of the morphological characters

The cultivar Ruiru 11 was found to be the most uniform recording intra-variation of less than 5%. On the other hand, SL28 was the most variable cultivar with an intra-variation of close to 15%. The new lines portrayed an intra-variation of less than 10%.

Quantitative morphological data are presented in Table 2, all the test materials were not significantly different in leaf length, leaf width, flower number, seed width and seed thickness. However, significant variations were obtained in internode length, weight of 100 dry fruits, seed length, seed lenth/width ratio, weight of 100 dry seeds and period between flowering and harvesting. Ruiru 11 had significantly shorter internode length than the rest of the test materials. On the other hand, CR8 had a bigger seed size as evident from its significantly bigger seed length and seed length/width ratio as compared to the rest of the genotypes. All the test materials were significantly different from each other in weight of 100 dry fruits while SL28 recorded significantly more weight of 100 dry seeds than Ruiru 11, CR22, CR 23 and CR27 though similar to CR8 and CR30. SL28, Ruiru 11 and CR30 took significantly longer time between flowering and harvesting than CR8, 22, 23 and 27.

Results of the principle component analyses indicated that the first two PCS explained 37 and 23% (a total of 60%) of the total variation (Table 3). The most principle component was weight of 100 dry seeds followed by the period between flowering and ripening, plant height and internode length in that order (Table 3). Plant height and internode length contributed most to PC1 while seed width and period between flowering and ripening contributed most to PC2.

Fig. 3:	Diversity among genotypes as determined by morphological variables

The traits leaf shape, anthocyanin colouration, undulation of the margin, depth of the secondary veins, presence or absence of leaf domatia, fruit shape, fruit colour and adherence to the branch did not contribute at all to both PC1 and 2.

The two dimensional presentation of the seven genotypes is presented in Fig. 3. Improved commercial cultivar, Ruiru 11, separated clearly from other genotypes and was located on the lower part of the PCA graph (Fig. 3). On the other hand, the traditional commercial cultivar, SL28 and the five advanced breeder’s lines were located on the upper part. Among the advanced breeder’s lines, CR22, 23 and 27 were plotted closely together, a distance from CR8 and 30 which were plotted on the uppermost part of the PCA graph. CR30 was plotted closer to SL28 than other advanced breeder’s lines.

DISCUSSION

Genetic variation in Arabica coffee has previously been characterised using morphological and yield-related traits and phylogenetic relationships established (Lashermes et al., 1996). Morphological markers are a classical method to distinguish variation based on the observation of the external morphological differences such as the size and shape of the leaf and of the plant form, the colour of the shoot tip, the characteristics of the fruit, the angle of branching and the length of the internodes (De Vienne et al., 2003). In this study, the cluster analysis depicted less than 25% morphological variation between the test materials indicating a narrow genetic base. This contradicted the report of Kumar et al. (2008) that coffee trees differ greatly in morphology, size and ecological adaptation. Ruiru 11 and SL28 were the most distantly related. The observed difference could be attributed to hybridization programme used in the development of Ruiru 11 unlike in SL28 where single tree selection procedures were used.

There was a high degree of similarity in qualitative traits of SL28 and the five new lines. The principle component analysis also produced SL28 as being closely related to the five new lines than to Ruiru 11. However, considering that the five new lines were selected from the male parents of Ruiru 11, it was expected that the new lines would be more genetically similar to Ruiru 11 than to SL28. The otherwise results could be attributed to inefficiency of morphological markers in assessing genetic variation. This concurs to the report of De Vienne et al. (2003) that assessing genetic variation with morphological markers can be inefficient as they are generally dominant traits, they often exhibit epistatic interactions with other genetic traits and can also be influenced by the environment. Lashermes et al. (1996) reported that genetic factors are more accurately tested by molecular markers. The observed similarity was also attributed to the single tree selection procedures that were used to develop these varieties. Agwanda et al. (1997) also reported that single tree selection procedures used to develop most Arabica coffee cultivars have contributed to high level of uniformity among them.

The divergence of Ruiru 11 from the other varieties studied could have resulted from its female parent Catimor, which has also been shown (Agwanda et al., 1997) as being genetically divergent from other Arabica genotypes owing to the interspecific origin of one of its progenitor, Timor Hybrid. Ruiru 11 being an F1 hybrid of the variety Catimor, as the female parent and outstanding male selections most of which has Timor Hybrid in their pedigree must contain a considerable amount of Timor Hybrid genes. In addition all Ruiru 11 male progenitors contain the semi-wild variety, Rume Sudan in their parentage which further accentuates its divergence from other varieties. Agwanda et al. (1997) reported that the use of semi-wild varieties such as Rume Sudan, or interspecific hybrids such as Catimor, as resistance donors results into the associated introgression of other un-targeted and agronomically undesirable genes. Silveira et al. (2003) also reported that coffee genotypes derived from Timor Hybrid demonstrate notable genetic diversity. Although both Rume Sudan and Timor Hybrid are contained in the pedigree of the five new lines, the magnitude of their genes may be smaller than in Ruiru 11.

As expected there was very little morphological variation within varieties. The mean dissimilarity values between varieties were higher than within varieties which underscores the low variability and the inbreeding nature of cultivated Arabica coffee. The results concurs with those of Anthony et al. (2001), who also demonstrated low genetic variation within Arabica coffee genotypes. Masumbuko and Bryngelsson (2006) also found similar results when comparing diploid coffee species and cultivated Coffea arabica L. from Tanzania. The low genetic variability within varieties further strengthened the evidence of the narrow genetic base of Arabic coffee. The observed minimal intra-variation within varieties was attributed to ecological instability as the trees were less than five years old. This is in line with the report of Hue (2005) that variability within varieties may result from environmental and/or management factors such as mis-labelling or mis-planting. However, other hypothesis, such as gene introgression from a gene pool alien to C. arabica, cannot be excluded.

CONCLUSION

The study demonstrated minimal morphological variation among the cultivars that were tested indicating low genetic variation and narrow genetic base. There is therefore need to widen the genetic base of Arabica coffee in Kenya by having more introductions, initiate hybridisation programmes to create variability and use of diploid species as a source of desirable genes. The study also demonstrated minimal morphological variation within varieties indicative of high genetic consistency within Kenyan Arabica coffee varieties.

ACKNOWLEDGMENT

This research was co-financed by Coffee Research Foundation (CRF) and the Common Fund for Commodities (CFC) through the Coffee Leaf Rust Project (CFC/ICO/40). Additional financial support was provided the European Union through the Quality Coffee Production and Commercialization Programme (QCPCP). Thanks are due to Messrs S.M. Njeruh and M.M. Musembi of Plant Breeding section who were responsibile for the management of experimental site and data collection. This work is published with the permission of the Director of Research, CRF, Kenya.