Variation in agro-morphological characteristics of cacao, Theobroma cacao L., in farmers' fields in Nigeria

Abstract

Knowledge of available variability in a crop species provides a guide for the utilization and conservation of useful variation. Seventeen agro-morphological traits were studied in 184 accessions of cacao, Theobroma cacao L. collected from farmers' fields (138) and field genebank collections (46). Descriptive statistics—univariate analysis of variance (ANOVA)—were used to determine variation among accessions. Multivariate analysis with principal component analysis (PCA) was done and a cluster dendrogram was generated based on the unweighted pair group mean with arithmetic average (UPGMA). The first three principal component axes accounted for 43.6% of total variation observed among accessions studied. Quantitative bean traits—dry bean weight, nib weight, fresh weight, cotyledon length and cotyledon width—and qualitative fruit traits—basal constriction, apex form, ridge colour, fruit shape, flush colour and cotyledon shape—were the most important traits accounting for the variability observed. Six groups of accessions were obtained from cluster analysis. Fruit and bean traits of Upper Amazon Forasteros observed in farmers' accessions provided evidence of a shift from previously grown local ‘West African Amelonado’ from the Lower Amazon Forastero population. In conclusion, this study revealed some accessions with useful trait combinations in farmers' plantations and indicated a need to conserve the local ‘Amelonado’ landrace variety to avoid extinction and to preserve the unique flavour of ‘West African Amelonado’ in cocoa breeding programmes.

Keywords:

• cacao
• conservation
• germplasm
• morphological variation
• on-farm
• West African ‘Amelonado’

Introduction

It is estimated that the world's cocoa exports, which amount to US$5–6 billion/year and provide cocoa butter, powder and liquor used in the manufacture of chocolate, cosmetics and other products, drive a market worth approximately US$70 billion and provide more than 60,000 jobs in the USA alone (Guiltinan 2007). Cocoa is a major foreign exchange-earning commodity crop for West and Central African countries such as Ivory Coast, Ghana, Nigeria and Cameroon which together produce more than 70% of the world's cocoa export of 3.592 million metric tons (ICCO 2008). In West Africa, more than 96% of cocoa production is carried out by smallholder farmers who rely on proceeds from the sale of cocoa beans as a major source of family income. Proceeds from export also provide substantial revenue for the governments’ capital development projects (Aikpokpodion 2007). The presence of significant genetic diversity in genetic resources maintained in farmers' fields and field genebanks could provide insurance against any present or future threat to cultivation and contribute to the sustainability of the world's cocoa economy.

Cacao is an understorey tree crop species native to the Amazon Basin belonging to the Sterculiaceae family (Purseglove 1974). However, recent molecular studies have revealed that Theobroma cacao belongs to the Malvaceae family (Alverson et al. 1999) and was introduced into Africa by Spanish and Portuguese seafarers—first to São Tome in 1822 and then to Fernando Po (now Bioko in Equatorial Guinea) in 1855 (Wood and Lass 1985; Bartley 2005). In São Tome, two cacao trees, which were successfully established from a batch of plants brought in from Bahia, Brazil in 1822, became the parents of the subsequent cacao trees of the island (Toxopeus 1973). Swiss missionaries also made other introductions, from Suriname, and the first cocoa seeds were sown on the African mainland in 1857 (Lanaud et al. 2000).

Cacao was first introduced by Chief Squiss Ibaningo into Bonny in present-day Rivers State of Nigeria in 1874 from Fernando Po (Opeke 1969). Apart from Chief Squiss Ibaningo, missionaries, slave traders and the former Nigeria Department of Agriculture also introduced several cacao germplasm materials. As a result of these introduction efforts, a fair range of variability was then present but all within the ‘Amelonado’ population (Aikpokpodion 2007). However, the cultivation and development of the cocoa industry in Nigeria only began in the 1890s with 95% of the production export of 21 tonnes in 1895 coming from the Western Region (Opeke 1969). The first materials planted were of the Lower Amazon Forastero population (‘Amelonado’), but this was followed by the introduction of the hybrid ‘Trinitario’ and ‘Criollo’ types in 1920 which formed hybrids with the original ‘Amelonado’ type (Toxopeus 1972). In 1944, several accessions from Upper Amazon Forastero and ‘Trinitario’ populations were introduced from Trinidad to widen the genetic base of the germplasm which was lacking in variability at the time (Posnette & Todd 1951). Between 1965 and 1967, a large-scale introduction of Upper Amazon cacao materials was made from Trinidad as part of the Trinidad–Nigeria Cacao Introduction Scheme sponsored by the Cocoa Alliance (Atanda 1977). This was also documented in the Annual Reports of Cacao Research, ITCA, 1966 and 1967 . In this scheme, some 701 progenies and 313 clones of intra-Nanay, intra-Parinari, intra-Iquitos and inter-P (Pound's selections) crosses from a total of 350 crosses were introduced (Olatoye and Esan 1992). Some materials were also acquired from Costa Rica, Indonesia, Fernando Po, Kew Gardens (UK), Wageningen (The Netherlands) and Miami (USA) (Jacobs et al. 1971). These materials formed the basis of cultivars developed in the breeding programmes and planting materials distributed to farmers.

Although molecular markers are more commonly used in the assessment of genetic diversity of cacao (N'Goran et al. 1994; Sounigo et al. 2005; Cryer et al. 2006; Aikpokpodion et al. 2009), highly heritable morphological and agronomic characters are still useful for germplasm characterization (Klug & Cummings 1994; Sounigo et al. 1997). In a recent study, Bekele et al. (2006) showed that classification of accessions in the International Cocoa Genebank, Trinidad (ICG,T) based on morphological variation was congruent with the traditional classification of the cacao types.

Morpho-agronomic characteristics of the pods, seeds and flowers have also been used to evaluate relationships among cacao genotypes (Engels 1986; Bekele & Bekele 1996; Lachenaud et al. 1999; Lachenaud & Oliver 2005). Previous diversity studies have often concentrated on field genebank collections (Bekele et al. 2006; Engels 1986; Lachenaud & Oliver 2005), however, this study focused on cacao accessions and diversity maintained on-farm in field plantations by farmers.

As an introduced tree crop species, cacao has been grown as an important cash crop by the West African farmers for more than a century and cultivation has witnessed several challenges including the devastating effect of the cocoa swollen shoot virus disease particularly in the 1930s and 1940s (Posnette 1947). Since the mid-1950s, new varieties of cacao, largely of Upper Amazon Forastero population, had been released and distributed for planting both as a replacement for the genetically uniform local ‘Amelonado’ types that had been cut out as a remedy for the virus disease attack and as new plantings for production expansion. In Nigeria, Toxopeus (1964) reported that up to 1961, more than 20 million seedlings of the newly developed synthetic variety had been distributed. Since then, there has been little or no information on the extent of diversity in field-cultivated cacao. This study was therefore necessary to determine the level of variation in field-grown cacao in order to provide information that could be useful for future breeding strategies, on-farm selection and conservation efforts.

The main objective of this study therefore, was to determine variation in agro-morphological characteristics of cacao grown in farmers' fields and determine their relationships with some genebank accessions used in the development of varieties distributed for planting.

Materials and methods

Plant materials

A total of 184 samples used for this study was comprised of 138 accessions collected from farmers' plantations in cocoa growing areas of Nigeria (Table 1) and 46 accessions from seed gardens and multiplication units maintained in the field genebank of the Cocoa Research Institute of Nigeria, Owena in Ondo State (07.20° N, 5.03° E; 267 m above sea level) that are used to develop hybrid varieties for distribution to farmers.

Table 1  Accession codes, local government area (LGA) and geographical locations of cocoa plantations used in this study

Accession code	Farm location	LGA & State	Longitude (°E)	Latitude (°N)	Altitude (masl)
ETG	Effraya	Etung, Cross Rivers	8.9698	5.8878	167
	Bendeghe-Ekim	Etung, Cross Rivers	8.8381	6.0162	177
	Ekimayang	Etung, Cross Rivers	8.7573	5.8870	134
	Agbokim Cocoa Estate	Etung, Cross Rivers	8.8733	5.9214	156
	Etomi	Etung, Cross Rivers	8.8559	5.9290	163
	Bijah-Abia	Etung, Cross Rivers	8.8825	5.9443	151
IKM	James & Harrison Farm	Ikom, Cross Rivers	8.7138	5.9536	101
	Akparabong	Ikom, Cross Rivers	8.7931	6.0255	125
	Adijikpor	Ikom, Cross Rivers	8.7512	5.9991	113
OBR	Iyamonyong	Obubra, Cross Rivers	8.3510	5.9694	119
	Ochon	Obubra, Cross Rivers	8.4456	5.9445	120
AJS	Mbe Ndifon Farm, Ajassor	Etung, Cross Rivers	8.8604	5.8846	153
BOK	Orimekpang	Boki East, Cross Rivers	8.8990	6.0632	122
	Ekumba-Kakwagom	Boki East, Cross Rivers	8.7929	6.4714	168
OBD	Ukwutia-Utugwang	Obudu, Cross Rivers	9.0270	6.6362	170
	Bighe-Aleghe	Obudu, Cross Rivers	9.0467	6.5661	199
AKP	New Ndebeji	Akamkpa, Cross Rivers	8.7929	6.4714	168
BND	Ntumbuzor	Bende, Abia	7.6664	5.4880	36
IDR	Aponmu-Okemoye	Idanre, Ondo	5.0412	7.1731	266
	Ayetoro-Owena	Idanre, Ondo	5.0295	7.1724	247
	Ipoba-Ojomu	Idanre, Ondo	5.0891	7.1761	253
BAMK	Bamikemo-Oja	Ile-Oluji/Okeigbo, Ondo	4.9035	7.3123	233
USEN	Usen	Ovia South-West, Edo	7.5910	5.4008	397

Twenty-three cocoa plantations were sampled and the number of trees sampled ranged from 4 to 12 per farm, depending on the tree population and criteria defined with farmers. Samples were randomly selected from trees considered by farmers as their ‘best trees’ in terms of productivity; ‘choice’ trees from which pods were collected to raise new seedlings for planting (‘mother tree’); and ‘worst trees’ that would not be used for planting materials.

From the field genebank, accessions with small bean sizes (less than 1.0 g) like C77 (T85/799) and large bean sizes such as T17/11 derived from ‘Iquitos Mixed Calabacillo’ (IMC) population were included with other hybrid genotypes distributed to farmers such as CRIN Selected Accessions (CSA), CRIN Elite Selected Accessions (CESA), CRIN Series 2 hybrids (CS2H), Trinidad Introduction (CTIS and T-Clones).

Data collection

Data were collected on cacao accessions following the Short Descriptor List of the Cocoa Research Unit, Trinidad and Tobago (Bekele & Butler 2000) adapted from IBPGR Cacao descriptor list (IBPGR Secretariat 1981). The morphological descriptors used were carefully selected from the International Board for Plant Genetic Resources descriptor list for cacao (IBPGR Secretariat 1981) and have been found to be most discriminative, taxonomically useful and precluded redundancy (Engels 1986; Lachenaud et al. 1999; Bekele & Butler 2000; Bekele et al. 2006). Each farmer's selection was characterized in terms of the 17 traits listed in Table 2. Only healthy, mature pods were selected for characterization. Fruit shape, surface texture (rugosity), anthocyanin pigmentation (colour) on the ridge of mature pods, ridge disposition, basal constriction and apex form were scored, while fruit length and width were measured. Fruits were considered mature when their colour had turned (or were about to turn) yellow (in the case of green pods) or orange (in the case of red pods). Wet bean mass was separated from the central placenta and number of beans per pod was counted. For each accession, three fresh beans were collected from the middle section of each of the 10 pods and bulked. Cotyledon length and width were measured and the fresh weight of 20 randomly selected beans was determined using a Sartorius balance (BP 610 model). Dry weight was determined after samples had been oven-dried at 105°C for 24–30 hours and nib weight was recorded after the seed-coat had been removed from the dried beans.

Table 2  Characters and states of 17 agro-morphological traits studied in cacao accessions from farmers' plantation and the field genebank of the Cocoa Research Institute of Nigeria (CRIN)

Character		State
1	Leaf, flush colour (anthocyanin)	0 = absent; 3 = slight; 5 = intermediate; 7 = intense (n = 10)
2	Fruit shape	1 = Angoleta (elongate with broad base); 2 = Cundeamour (elongate with bottleneck); 3 = Amelonado (small with slight bottleneck) (n = 10)
3	Fruit, apex form	1 = attenuate; 2 = acute; 3 = obtuse; 4 = rounded; 5 = mammelate; 6 = indented (n = 10)
4	Fruit, basal constriction	0 = absent; 1 = slight; 2 = intermediate; 3 = strong (n = 10)
5	Fruit, ridge colour (anthocyanin)	0 = absent; 1 = slight; 3 = intermediate; 5 = intense (n = 10)
6	Fruit, rugosity (wartiness)	0 = absent; 3 = slight; 5 = intermediate; 7 = intense (n = 10)
7	Fruit, ridge pair disposition	1 = equidistant; 2 = paired (n = 10)
8	Cotyledon shape	1 = oblong; 2 = elliptical; 3 = ovate (n = 20)
9	Cotyledon colour	1 = white; 2 = grey; 3 = light purple; 4 = medium purple; 5 = dark purple; 6 = mottled (n = 20)
10	Fruit length (cm)	Fruit length from base to apex (n = 10)
11	Fruit width (cm)	Widest point at the middle section (n = 10)
12	Bean number	Number of beans per fruit (n = 10)
13	Cotyledon length (cm)	Length of bean (n = 20)
14	Cotyledon width (cm)	Width of bean (n = 20)
15	Bean fresh weight (g)	Weight of cleaned fresh bean (n = 20)
16	Total dry bean weight (including testa) testa)	Constant weight of dried bean with testa (n = 20)
17	Nib weight (testa removed)	Constant weight of dried bean without testa (n = 20)

Data analysis

Descriptive statistics were derived for all traits observed and all eight quantitative variables used for analysis showed no departure from normal distribution. Principal component analysis (PCA) with all 17 traits was carried out to determine the relative importance of the traits responsible for variation among cacao accessions. The aforementioned statistical analyses were conducted using SAS v.9.1 (SAS Institute Inc, NC, USA). Cluster analysis with all 17 traits was used to explore relationships among the accessions based on the unweighted pair group method with the arithmetic mean (UPGMA) implemented with the NTSYSpc v. 2.02 package (Applied Biostatistics, Inc). The general linear model (GLM) procedure was specified for the analysis of variance (ANOVA) and mean separation was performed on the a priori clusters to determine cluster characteristics.

Results

The descriptive statistics of the quantitative traits of cacao accessions from field genebank and farmers' plantations are shown in Table 3 while the qualitative traits are shown in Table 4. Test of goodness of fit showed no significant deviation from normal distribution in all the quantitative traits. Correlation analysis (Table 5) showed significant positive correlation among all traits except between number of beans per pod on the one hand, and cotyledon length, cotyledon width, dry weight and nib weight, on the other.

Table 3  Mean, standard error (in parenthesis), coefficient of variation (%) and levels of significance from analysis of variance (ANOVA) of 184 cacao accessions from field genebank and farmers plantation in Nigeria

	Genebank (n = 46)		Plantation (n = 138)
Character	Mean (SE)	CV	Mean (SE)	CV	ANOVA Pr > F-value
Fruit length (cm)	15.3 (0.32)	10.6	15.5 (0.17)	14.9	ns
Fruit width (cm)	8.3 (0.09)	5.2	8.2 (0.06)	9.2	ns
Number of beans/pod	35.0 (1.16)	16.6	39.5 (0.45)	14.0	***
Cotyledon length (cm)	2.5 (0.03)	8.0	2.4 (0.02)	8.0	ns
Cotyledon width (cm)	1.4 (0.01)	6.6	1.3 (0.01)	7.5	ns
Fresh bean weight (g)	2.0 (0.04)	13.3	1.7 (0.03)	17.8	***
Dry bean weight (g)	1.2 (0.02)	12.7	1.1 (0.02)	17.2	***
Nib weight (g)	1.1 (0.02)	13.4	1.0 (0.03)	36.0	ns
P-level: *** = P<0.001, respectively; ns = not significant

Table 4  Frequency distribution of nine qualitative traits in cacao accessions from CRIN's field genebank and farmers' plantation

Character	State	Genebank (%) n = 46	Plantation (%) n = 138	Overall (%)
Leaf, flush colour (anthocyanin)	0 = absent	4.0	12.3	11.4
	3 = slight	76.0	48.7	51.8
	5 = intermediate	20.0	30.3	29.1
	7 = intense	0	8.7	7.7
Fruit shape	1= Angoleta (elongate with broad base)	20.0	14.7	15.3
	2= Cundeamour (elongate with bottleneck)	80.0	76.2	76.7
	3 = Amelonado (small with slight bottleneck)	0.0	9.1	8.0
Fruit, apex form	1 = attenuate	0.0	15.1	13.3
	2 = acute	28.0	32.8	32.2
	3 = obtuse	72.0	46.3	49.3
	4 = rounded	0.0	1.7	1.5
	5 = mammelate	0.0	3.6	3.2
	6 = indented	0.0	0.5	0.5
Fruit, basal constriction	0 = absent	22.4	10.0	11.4
	1 = slight	77.6	75.6	75.9
	2 = intermediate	0.0	11.7	10.3
	3 = strong	0.0	2.7	2.4
Fruit, ridge colour (anthocyanin)	0 = absent	76.0	81.o	80. 5
	1 = slight	8.0	7.7	7.7
	3 = intermediate	0.0	3.9	3.4
	5 = intense	16.0	7.4	8.4
Fruit, rugosity (wartiness)	0 = absent	4.0	7.1	6.8
	3 = slight	8.0	6.8	6.9
	5 = intermediate	24.0	64.0	59.4
	7 = intense	64.0	22.1	26.9
Fruit, ridge pair disposition	1 = equidistant	24.0	33.9	32.8
	2 = paired	76.0	66.1	67.2
Cotyledon shape	1 = oblong	2.1	3.6	3.2
	2 = elliptic	27.0	31.2	30.1
	3 = ovate	70.9	65.2	66.7
n = number of samples

Table 5  Correlation matrix of eight quantitative traits of cacao accessions

	Fruit length (cm)	Fruit width (cm)	Number of beans/pod	Fresh bean weight (g)	Cotyledon length (cm)	Cotyledon width (cm)	Dry bean weight (g)
Fruit width (cm)	0.441***
Number of beans/pod	0.165***	0.123***
Fresh bean weight (g)	0.189***	0.218***	−0.131***
Cotyledon length (cm)	0.125***	0.270***	0.055^ns	0.525***
Cotyledon width (cm)	0.095***	0.209***	0.033^ns	0.469***	0.522***
Dry bean weight (g)	0.339***	0.267***	–0.019^ns	0.619***	0.538***	0.492***
Nib weight (cm)	0.250***	0.284***	–0.014^ns	0.629***	0.540***	0.502***	0.909***
P-level: *** =P<0.001, respectively; ns = not significant

Principal component analysis of all 17 morphological traits revealed that the first three principal component (PC) axes accounted for 43.6% of total variation among cacao accessions (Table 6). Examination of the first PC axis which accounted for 24.5% of total variation showed that it was composed mainly of bean traits—dry bean weight, fresh bean weight, cotyledon length and cotyledon width. The second PC axis accounting for 10.8% of total variation was mainly composed of fruit traits—fruit length, fruit shape, rugosity and apex form. Although the fruit length and rugosity were positively correlated, fruit shape and apex form were negatively correlated to this axis. The third PC axis which made up 8.3% of total variation was composed mainly of qualitative traits of flush colour, ridge colour and cotyledon shape. The fourth PC axis which accounted for 8.0% of variation observed was composed mainly of fruit rugosity, basal constriction, apex form and cotyledon colour. Qualitative characteristics including fruit shape, basal constriction, apex form, cotyledon shape, rugosity (wartiness), ridge pair disposition and cotyledon colour were important traits which accounted for 23.29% of total variation in PC5, PC6, PC7 and PC8 axes. Cotyledon length, fruit width and number of beans per pod were also important quantitative traits accounting for the variance found in these PC axes.

Table 6  Eigenvectors of 17 agro-morphological traits of cacao accessions in first three principal component (PC) axes explaining 43.6% of total variation

	PC1 4.16^# 24.5*	PC2 1.83 10.8	PC3 1.41 8.3
Dry bean weight (g)	0.457	−0.020	−0.061
Fresh bean weight (g)	0.444	−0.040	−0.122
Cotyledon width (cm)	0.391	−0.192	0.080
Cotyledon length (cm)	0.400	−0.180	−0.032
Fruit length (cm)	0.235	0.390	−0.002
Fruit width	0.276	−0.019	−0.051
Fruit shape	−0.115	−0.401	0.053
Basal constriction	0.005	0.202	0.155
Apex form	−0.082	−0.334	0.018
Flush colour	0.056	0.045	0.654
Ridge colour	0.165	0.070	0.497
Cotyledon shape	0.041	−0.144	0.362
Number of beans/fruit	0.058	0.231	0.210
Cotyledon colour	−0.019	0.457	−0.166
Ridge pair disposition	−0.107	0.178	0.214
Rugosity	0.003	0.371	−0.040
Nib weight (g)	0.288	−0.065	−0.139
# = Eigen value of PC axis, * = percentage of total variation explained by PC axis

Scatter plots of the first two principal component axes (Fig. 1) showed the importance of some characteristics in showing relationships among the cacao accessions. Cluster analysis showing relationships among all accessions (Fig. 2) showed three distinct groups corresponding to six clusters due to sub-clustering. Accessions with the largest nib, dry and fresh weights, fruit length and fruit width, cotyledon length and cotyledon width were grouped into cluster 3 (Table 7). Although fresh bean weight was greater in cluster 3A, cotyledon length was greater in cluster 3B. Accessions with the least bean weight, cotyledon length, cotyledon width and smallest fruit size were grouped into cluster 2. Fruits of accessions in cluster 3 were largely elongate in shape with a broad base (‘Angoleta’) and usually have slight to intense anthocyanin pigmentation in the ridge of their mature fruits. On the other hand, accessions in clusters 1 and 2 appeared to be mostly ‘Cundeamour’ and ‘Amelonado’ as evident from the fruit length and fruit width. Fruits in these clusters were always green and rarely had anthocyanin pigmentation present on the ridges of their mature fruits. Nib weight, fresh and dry bean weights of accessions in cluster 3 were almost twice those of accessions in cluster 2. There was however, no significant difference in the number of beans per pod, flush colour, basal constriction, apex form, rugosity, shape and colour of cotyledon among the six clusters.

Fig. 1  Scatter plot of cacao accessions on principal component axes 1 and 2 (PC1 and PC3) accounting for 35.3% of total variation. i – cotyledon fresh weight (g), ii – cotyledon dry weight (g), iii – nib weight (g), iv – fruit length (cm), v – cotyledon colour, vi – ridge pair disposition, vii – fruit shape, viii – apex form, ix – cotyledon width (cm), x – cotyledon length (cm).

Fig. 2  Dendrogram based on the unweighted pair group arithmetic mean average (UPGMA) showing six clusters of on-farm accessions and field genebank accessions (in boxes) of cacao in Nigeria.

Table 7  Group characteristics of the six clusters based on 17 morphological traits of cacao accessions

Cluster	n	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17
1A	40	14.9b	8.0c	38.8a	1.58e	2.33c	1.29d	0.96d	0.90d	3.5a	1.5a	1.1a	2.6a	4.9a	0.5a	1.7ab	3.8a	2.6a
1B	42	15.5ab	8.2bc	39.4a	1.73d	2.39c	1.33c	1.08c	1.00c	3.3a	1.1ab	1.0a	2.4a	5.1a	0.3b	1.6ab	4.0a	2.7a
2	27	14.6b	8.0c	39.3a	1.31f	2.17d	1.21e	0.82e	0.77e	3.7a	1.5a	0.9a	2.4a	4.8a	1.0b	1.7a	4.2a	2.6a
3A	11	16.7a	8.6a	40.2a	2.31a	2.54a	1.47a	1.36a	1.27a	4.5a	1.0b	0.9a	2.4a	5.7a	2.9a	1.3b	4.1a	2.8a
3B	23	16.7a	8.7a	38.1a	2.18b	2.64a	1.44a	1.32a	1.24a	3.8a	1.0b	0.9a	2.4a	4.8a	1.6b	1.6ab	3.9a	2.6a
3C	41	15.9ab	8.4ab	37.3a	1.92c	2.48b	1.39b	1.19b	1.11b	3.8a	1.2ab	1.0a	2.6a	5.0a	2.0ab	1.7ab	3.9a	2.6a
F (5,178)		***	***	ns	***	***	***	***	***	ns	***	ns	ns	ns	***	***	ns	ns
n = number of accessions; 1 = fruit length (cm), 2 = fruit width (cm), 3 = number of beans/pod, 4 = fresh bean weight (g), 5 = cotyledon length (cm), 6 = cotyledon width (cm), 7 = dry bean weight (g), 8 = nib weight (g), 9 = flush colour, 10 = fruit shape, 11 = basal constriction, 12 = apex form, 13 = fruit rugosity, 14 = ridge colour, 15 = ridge pair disposition, 16 = cotyledon colour, 17 = cotyledon shape; means followed by the same letter within the column are not significantly different (P < 0.001); ***: F-ratio level of significance at P < 0.001 and ns = not significant

The genebank clones included in the analysis were useful as checks to confirm observations made. Clones T17/11 (a progeny of Imperial Mixed Calabacillo ‘IMC’ IMC53 known for large beans and pod size), PA35 (‘Parinari’) and T7/12 (progeny of M 8 [SUR] from the Maraboen River, Suriname) were classified into cluster 3, which represented 40.8% of samples studied. Cluster 3A was mainly composed of red fruited ‘Trinitarios’ also known for large beans, and clusters 3B and 3C most likely represented ‘Upper Amazon’ hybrid cocoa with large beans and fruits. In contrast, clone C77 (T85/799) with small beans was classified into cluster 2. Cluster 2, which made up 14.7% of the accessions studied consisted of accessions with small beans (0.82 g mean dry bean weight) and included many of the accessions known as ‘local Amelonado’ by the farmers. Accessions in cluster 1 had intermediate trait values and made up the rest of the 44.6% of the samples studied.

ANOVA of the a priori groups showed significant differences (P < 0.001) among clusters in all quantitative traits except number of beans per pod (Table 7). There were also significant differences (P < 0.001) in fruit shape, ridge colour and ridge pair disposition among the six clusters. However, ANOVA also showed no significant differences between field genebank and farmers' plantation accessions among all quantitative traits except in fresh bean weight (P < 0.001), dry bean weight (P < 0.001) and number of beans per fruit (P < 0.001) (Table 3).

Discussion

In this study, the variation in morphological traits among cacao accessions grown in farmers' fields and their relationship with accessions from a field genebank collection was evaluated. This is unlike previous studies where the emphasis was on morphological variation of cacao in field genebank collections (Engels 1983; Bekele et al. 1994; Bekele & Bekele 1996; Iwaro et al. 2003; Bekele et al. 2006). Analysis of variance showed significant morphological variation in cacao accessions maintained in farmers' field collections and also between accessions from the field genebank and farmers' fields in bean and fruit traits. While nib, fresh and dry bean weights were greater in the field genebank accessions, the number of beans per pod was greater in the farmers' field accessions. It appeared that the higher number of beans per pod was related to the lower fresh bean weight in farmers' accessions as shown from the correlation values between these variables (r = − 0.131, P < 0.0001). A similar relationship (r = − 0.19; P < 0.001) was also reported among cacao accessions from a genebank in Trinidad (Iwaro et al. 2003). However, there was no significant difference among accessions in fruit length and cotyledon width.

The observed variation permitted the accessions to be classified into six clusters. From the classification obtained, it appeared that accessions with large bean and fruit size in cluster 3 displayed the characteristics of Upper Amazon derived varieties and ‘Trinitario’ materials distributed to farmers from seed gardens. On the other hand, the small bean and fruit size of accessions in cluster 2 were characteristic of the local ‘Amelonado’ cacao. Intermediate values for bean weight and fruit characteristics obtained for accessions in cluster 1 indicated possible hybridization that could have taken place in the field as a result of the introduction of Upper Amazon materials to farmers' fields which were initially dominated by the old ‘local Amelonado’ variety. In their study of some 600 accessions of cacao maintained in the International Cocoa Germplasm, Trinidad (ICG,T), Bekele et al. (2006) found that phenotypic data showed distinct classification of accessions that were congruent with recognized genetic classes and were able to differentiate among the ‘Trinitarios’, ‘Forasteros’ and ‘Refractarios’. Engels (1983, 1986) also showed the usefulness of bean characteristics in showing affinity between ‘Trinitario’ and ‘Criollo’ cacao populations.

Some accessions with large dry bean weight and number of beans per pod: ETG24 (1.6 g, 36.3/pod), ETG29 (1.5 g, 46/pod), AKP5 (1.5 g, 54/pod), IKM11 (1.4 g, 40.7/pod), ETG 40 (1.4 g, 42.8/pod), IKM7 (1.4 g, 43.9/pod) and OBR44 (1.4 g, 43.7/pod) identified in cluster 3 showed promise for selection as clones with a low pod index. On the other hand, accessions in cluster 2 with low dry bean weight such as OBR3 (0.67 g), BOK1 (0.68 g), OBR59 (0.68 g), AJS3 (0.75), ETG41 (0.76 g) and OBR22 (0.85) could be useful for further study as their characteristic low bean weight and size reflected the trait of ‘local Amelonado’, a traditional variety grown by farmers before the release of introduced Upper Amazon derived materials around 1956. Selection of some of these individual genotypes as individual clones could be justified; Bartley (2005) also noted that cultivated populations that are products of recombination usually have greater genetic diversity, and consequently, a larger number of individual genotypes than the source populations.

The large variation observed in this study for cacao grown by farmers indicates a high level of heterogeneity in materials maintained on-farm. The current field composition of cacao indicates a shift from the situation preceding the 1950s, when uniform ‘Amelonado’ cacao types were mainly grown. This would have resulted from the use Upper Amazon-derived cacao varieties distributed to farmers through the seed gardens. As observed by Bartley (2005), the introduction of materials from another source into an original population could result in hybridization between them, and subsequent recombination among these hybrids could lead to genotypes genetically distinct from the parental varieties but possibly having phenotypic ranges encompassing those of the parents.

Quantitative bean traits—dry bean weight, nib weight, fresh weight, cotyledon length and cotyledon width—and qualitative fruit traits—basal constriction, apex form, ridge colour and fruit shape—were among the most important traits accounting for the variation observed among the samples studied. The quantitative bean traits were however most useful for classification. These nine characteristics were also among the traits found by Bekele et al. (2006) to explain more than 50% of phenotypic variation in the germplasm collection studied. However, the dispersion of variation among the first nine principal component axes, which explained 76.7% of total variation, indicated a need to use the full complement of the recommended morphological traits rather than these nine characters alone. Significant variation observed for bean and fruit characteristics among cacao accessions in this study indicated the importance of on-farm collections as a valuable reservoir of genetic diversity. Some of these traits are of commercial importance and have been used as selection criteria by farmers in the choice of parent trees for raising seedlings to make new plantings and used for farm expansion.

From this study, the complete absence or slight anthocyanin pigmentation on the ridge of the mature fruit of more than 88% of accessions indicated that most of the cacao now grown in Nigeria was apparently derived from the Amazonian Forastero origin. This observation may further confirm the report that Amazonian Forastero provides more than 95% of the world's cocoa output (Bekele et al. 2006). The preponderance of ‘Cundeamour’ fruit shape (>76%) with slight to strong basal constriction (>88%), obtuse to attenuate apex forms (95%) and intermediate to intense rugosity (86%) showed that, possibly, the Upper Amazon Forastero (UAF) ‘Parinari’ population, characterized by pronounced bottleneck, conspicuous apex form and the intermediate to intensely warty fruit (Bartley 2005; Bekele et al. 2005) and to a lesser extent, the ‘Nanay’ population, had the most impact on cacao grown in Nigeria. This could be explained by the fact that clones from these populations have been largely used in Nigerian cacao breeding programmes and form the basis of the released F₃ Amazon variety and others (Toxopeus 1964; Olatoye & Esan 1992). It has also been reported that UAF accounted for about 60% of cacao cultivation worldwide (Wood & Lass 1985; Eskes & Lanaud 2001).

On the other hand, the low percentage (less than 15%) of red pigmentation in fruits, a trait known to be under the control of a single dominant gene (Bartley 2005) associated with some ‘Criollo’ populations indicated that ‘Criollo’ and red-podded ‘Trinitario’ populations have, at present, only a minimal influence on field-grown cacao in Nigeria. This could be related to the limited use of ‘Trinitario’ clones in the Nigerian cocoa breeding programme in developing improved materials distributed to farmers, a situation similar to that observed by Lockwood (1976) in Ghana.

The low percentage of fruit traits that are typical of ‘Amelonado’ and ‘Trinitario’ types provides some evidence of variety replacement of ‘West Africa Amelonado’ (WAA) cacao types in farmers' fields with Upper Amazon-derived types and hybrids between them. This reported erosion of West African ‘Amelonado’ presents a serious case for the conservation of this population which may be completely lost if the present trend continues in farmers' fields. In a study of factors influencing farmers' choice of mother trees as sources of new planting on farms, Aikpokpodion et al. (2003) showed farmers' preference for Upper Amazon Forastero-derived types as sources of seeds for new planting due to their good vigour allowing for easier field establishment, enhanced precocity (flowering three years after field establishment, compared with five to seven years for WAA), year-round fruit production and higher yields. These factors could have been responsible for this observation.

Although of low vigour, the WAA trees are amenable to high-density plant ing and the beans possess flavour attractive to chocolate manufacturers. Aikpokpodion et al. (2003) also showed that farmers in Nigeria prefer green-podded trees as ‘mother trees’ or sources of planting materials rather than red-podded trees. An exception is in areas where farm sizes are small (e.g., marginal cocoa-growing areas) and the known high-yielding ability of the red-podded ‘Trinitario’ trees becomes the most important factor in spite of the high Phytophthora pod rot incidence associated with ‘Trinitario’.

The WAA or ‘bulk cocoa’ cocoa, known for its strong chocolate flavour, is commonly used for the production of high-volume chocolate lines. The shift in cacao types cultivated on farms observed in this study indicates the need to maintain the flavour quality profile characteristic of WAA in cocoa breeding programmes. In a flavour assessment done on UAF families used in developing the ‘F₃ Amazon’ variety, Wadsworth (1953) had shown that it was within the WAA flavour range, although it was harsher than a really good WAA cocoa. It is important to ensure that the WAA flavour profile used as standard for bulk cocoa is not compromised with particular reference to the Nigerian cocoa, often regarded as closest to the Ghanaian cocoa used as reference for ‘bulk cocoa’ on the world market (Anon 1991).

The shape of the cotyledon, largely ovate and elliptical, indicates a high level of uniformity for the shape of cocoa beans produced in Nigeria. The beans were purple coloured, varying from light to dark with more than half, medium purple. The mean dry weight of cocoa beans from Nigeria was slightly higher than the minimum weight of 1.0 g required in the international market (Wood & Lass 1985). In order to exploit variation in these cacao collections, more studies are needed to evaluate the quality characteristics such as butter fat content and flavour of accessions within each group. This should have emphasis on the preservation of the flavour profile of WAA bulk cocoa.

Acknowledgements

The field assistance of Messrs. Lateef Raji, Marcus Efunla, Segun Adeyanju, Sunday Njoku, Victor Enagu and Sunday Taiwo; financial support from the USAID/USDA-ARS, Mars Inc and three anonymous referees are gratefully acknowledged.