-
Cocoa (Theobroma cacao L.) is cultivated in the tropical regions of the American, African, Asian, and Oceanian continents to produce dried beans, which are roasted, shelled, and ground in mills to produce liquor, fat (butter), cake, and/or powder[1]. Cocoa butter is one of the most valuable products in the market and is of great industrial interest, particularly for the manufacture of chocolate, pharmaceutical, and cosmetic products[2]. It is predominantly composed of saturated fatty acids, palmitic and stearic acids, unsaturated fatty acids, oleic acids, and linoleic acids[3].
T. cacao has a great genetic variability for agronomic and industrial features that can be used by breeders to improve cultivars[4,5]. Generally, breeding programs in different cocoa-producing regions worldwide have focused on increasing productivity, pest resistance, larger pod and seed sizes, thinner pod shells, and self-compatibility[6,7]. However, programs that use recurrent selection, such as those in Côte d'Ivoire and Brazil (Comissão Executiva do Plano da Lavoura Cacaueira [CEPLAC]), are also trying to increase the fat percentage in seeds[8,9]. These programs produce hybrid and clonal cultivars, which are grown by farmers; however, selecting cultivars with higher fat content and better organoleptic quality has been facilitated by the increasing use of clonal cultivars[3,10].
The amount of butter in cocoa beans ranges from 45.4% to 60.3% (average = 53.2%), as shown by a study conducted in Brazil involving 490 cocoa accessions of different geographical origins[11]. Similar results were observed in 188 cocoa clones from Malaysia[12] and 35 accessions from the Brazilian Amazon[13]. This demonstrates a wide variation in fat content in cocoa accessions, allowing for the selection of superior genotypes.
The frequency distribution of fat content approximated a normal curve, suggesting that the inheritance of this trait is quantitative[11]. Fat content has low correlations with disease resistance, productivity, seed size, and pod size, making it an independent trait during selection work. Fat content exhibits a xenia effect, in which the origin of the pollen affects the expression of the the trait; therefore, selection can be made in the seed generation under analysis[14,15]. Regarding the genetic inheritance of fat content in cocoa, a study using an incomplete diallel design that evaluated 94 progenies in four locations in Malaysia demonstrated heritability ranging from 0.48% to 0.81% and significant additive genetic effects in all experiments, with dominant genetic effects being significant in only one[16]. In addition, a quantitative trait loci (QTL) of major effect, explaining 22% of the variation in fat content, was identified in a self-fertilized population of the clone TSH 516[17].
In general, studies of genetic inheritance of agronomc traits of interest in cacao are controlled by genes of additive effects with high general combining ability[18−21]. The aims of this study were to characterize the percentage of fat content found in the seeds of 16 parental clones and 62 hybrid combinations resulting from a partial diallel mating scheme, to obtain estimates of the general combining ability (GCA) of the parental clones, and to estimate the specific combining ability (SCA) of hybrid combinations resulting from partial diallel crossing among the 16 parents.
-
In this study, cocoa seed fat was extracted from 16 parental clones and 62 progenies. The 16 parental clones were part of a reciprocal recurrent selection program at the Cocoa Research Center (Centro de Pesquisa do Cacau [CEPEC]), CEPLAC, Ilhéus-BA. These parents were selected according to several criteria, including geographical origin, genetic resistance to diseases such as witches' broom and black pod, pod and seed size, fat content, and molecular DNA patterns[22]. These 16 parental clones were grouped into two sets: Eight as mothers and eight as fathers (Table 1). They were crossed according to a partial diallel strategy, namely Design II of Comstock and Robinson[23], resulting in the production of 62 of 64 progenies. Two crosses were lost during this process.
Table 1. Cocoa clones used as parents in Group I (mothers) and Group II (fathers), with their respective codes, geographical groups, and origins.
Clone Code Geographical group Origin Group I (mothers) Belém 4 BE 4 Forastero: Lower Amazon Belém-PA, Brazil Centro de Pesquisa do Cacau 89 CEPEC 89 Forastero Itabuna-BA, Brazil Cruzeiro do Sul 07 CSUL 7 Forastero: Upper Amazon Cruzeiro do Sul-AC, Brazil Estação Experimental de Goitacazes 29 EEG 29 Forastero: Lower Amazon Espírito Santo, Brazil Imperial College Selections ICS 98 Trinitario Trinidad and Tobago Ocumare 67 OC 67 Criollo Ocumare de la Costa-Aragua, Venezuela Rio Branco RB 39 Forastero Rio Branco-AC, Brazil Scavina 6 SCA 6 Forastero: Upper Amazon Ucayali River, Peru Group II (fathers) Cacao Castro Naranjal 10 CCN 10 Forastero × Criollo (Hybrid) Napo, Ecuador Cacao Castro Naranjal 51 CCN 51 Forastero × Criollo (Hybrid) Napo, Ecuador Centro de Pesquisa do Cacau 86 CEPEC 86 Forastero Itabuna-BA, Brazil Imperial College Selections ICS 9 Trinitario Trinidad and Tobago Iquitos Mixed Calabacillo 76 IMC 76 Forastero: Upper Amazon Iquitos, Peru Nanay 33 NA 33 Forastero: Upper Amazon Nanay River, Peru Pound 4/B P4B Forastero − Selecciones Guatemaltecas 54 SGU 54 Forastero: Guianas Los Brillantes, Guatemala In total, 16 clonal parents and their 62 progenies were evaluated in a randomized complete block design with two replicates of 20 plants per plot in an experiment at CEPEC. The plants were grown at a spacing of 3 m × 3 m, and data were collected during the productive phase of the plants, approximately 8 years after planting. The agronomic maintenance of the field trials followed the management scheme, fertilization, and disease control protocols adopted by CEPLAC[24].
Ten pods from different plants under the same treatment were collected from each plot, from which 10 seeds per pod (simple sampling) were collected, totaling a composite sample of 100 seeds per plot used for data collection. The pods originated from open pollination during the main crop season from September to November in 2011.
Fat percentage
-
The extraction of cocoa seed fat was based on a method described by the Association of Analytical Chemists[25], adopting the following procedure:
(1) We ground 100 cocoa seeds without testa from each genotype until a homogeneous fine mass was formed.
(2) The mass was dehydrated in an oven at 105 °C until a constant mass was obtained.
(3) For each genotype, a sample (5 g) was placed in a 250 mL beaker, 100 mL of 70 °C hydrochloric acid solution was added, and the beaker was placed on a hot plate for gentle boiling for 20 min.
(4) The digestion product was filtered through a medium-grade filter paper with a diameter of 15 cm, previously moistened and free of fat, using a glass funnel with a capacity of 150 mL.
(5) The filter paper with the moist sample was transferred to a Soxhlet cartridge and covered with degreased cotton. The cartridges were dried in an oven at 105 °C for 4 h.
(6) A 250 mL flask from the Soxhlet extractor set, previously dried and tared with glass beads to standardize boiling, was filled with 150 mL of petroleum ether or hexane (30–60 °C). The cartridge containing the sample was placed in the extractor and subjected to extraction for 4 h under continuous flow.
(7) Subsequently, the solvent was evaporated in a rotary evaporator, and the flask containing the fat was dried in an oven at 105 °C until complete evaporation of the solvent.
(8) The flask containing the dry matter was weighed, and the percentage of fat was calculated.
Statistical analysis
-
The analysis of variance (ANOVA) of the data was performed according to the statistical model:
$ {Y}_{{ij}} =\mu+g_i+b_j+e_{ij} $ where, Yij is the value of the ijk-th observation relative to genotype i in the j-th block,
$\mu $ For comparisons between means, the Scott–Knott test was used at a 5% probability level.
Diallel analysis was performed according to the model proposed by Geraldi and Miranda[26]. This model is an adaptation of Griffing's Method IV[27], in which the effects of the GCA of each parent and the effects of the SCA of the hybrid combinations are estimated using data obtained from p(p – 1)/2 F1 hybrids[28].
The ANOVA of the diallel analysis was performed according to the following model:
$ {Y}_{{ij}} =m+ 1/2(d_{{1}} +{d}_{{2}}) + g_{{i}} + g'_{{j}} +s_{{ij}} +e_{{ij}} $ where, Yij is the mean of the treatment involving the i-th parent from Group I (mother clones) and the j-th parent from Group II (father clones), m is the overall mean of the diallel, d1 and d2 are the contrasts involving the means of Groups I and II, gi is the effect of the general combining ability of the i-th parent in Group I, g'j is the effect of the general combining ability of the j-th parent from Group II, sij is the effect of the specific combining ability of the i-th parent from Group I and the j-th parent from Group II, and eij is the experimental error.
-
The ANOVA revealed significant differences in seed fat content among the treatments (Table 2). A significant difference between treatments was expected because the parents were heterozygous clones selected for their differences in seed fat content and other characteristics. The average fat content of the genotypes was 54.19%, which is close to the minimum fat content required by the industry (55%). These results are similar to those observed by Pires et al.[11], who analyzed 490 cacao genotypes from different geographic origins. Tucci et al.[29] identified fat content within a commercially interesting range only in the ICS 39, UF 677, and IAC 1 clones, attributing the observed differences to the genetic material used, although they also considered climatic conditions to be a factor influencing the fat content.
Table 2. Summary of the ANOVA for the percentage of fat content in cocoa seeds, based on data from hybrids and parent clones.
Source of variation Degrees of freedom Fat (mean square) Blocks 1 4.79* Genotypes 77 8.92* Clones 15 15.73* Hybrids 61 7.00* Clones × hybrids 1 24.04* Error 77 0.46 CV (%) 1.25 Overall mean 54.21 * Significant at 1% level by the F-test; CV: coefficient of variation. Regarding the breakdown of treatments, significant differences (p < 0.01) were observed for hybrids, clones, and hybrid versus clone interactions. The existence of significant differences between clones and hybrids indicates the possibility of identifying genotypes that differ from others in terms of seed fat content in both groups. The hybrid versus clone contrast indicated that they behaved differently under different treatments. This heterogeneity of the material allows the implementation of selection practices, which, in turn, can result in genetic gains by selecting superior genotypes.
The mean fat content percentages of cocoa seeds of the parental clones and hybrids are presented in Tables 3 and 4, respectively. The variation in fat content among clones ranged from 47.51% (OC 67) to 57.20% (CCN 51), and from 48.71% (EEG 29 × CCN 10) to 57.51% (CEPEC 89 × CEPEC 86). Significant differences between the means were detected using the Scott–Knott test at a 5% probability level. This indicated a variation of approximately 10% in fat content for both clones and hybrids, demonstrating a wide range for this trait within the evaluated population. The coefficient of variation (CV) was 1.25%, reflecting a high experimental precision.
Table 3. Mean values for fat content (%) in cocoa seed clones.
Clone Fat (%) Group I (mothers) CEPEC 89 56.10 a CSUL 7 55.91 a ICS 98 53.94 b BE 4 53.87 b EEG 29 53.20 b SCA 6 52.86 b RB 39 52.01 b OC 67 47.51 d Group II (fathers) CCN 51 57.19 a P4B 55.84 a ICS 9 55.59 a CEPEC 86 55.44 a SGU 54 53.97 b CCN 10 53.03 b IMC 76 50.60 c NA 33 47.99 d Means followed by the same letters belong to the same group according to the Scott–Knott test at 5% probability. Table 4. Mean values for fat content (%) in seeds of hybrids originating from interclonal crosses of cocoa.
Hybrid Fat (%) Hybrid Fat (%) BE 4 × ICS 9 56.22 a ICS 98 × IMC 76 57.18 a BE 4 × CCN 10 55.93 a ICS 98 × NA 33 55.96 a BE 4 × NA 33 55.39 b ICS 98 × CCN 51 55.93 a BE 4 × P4B 55.17 b ICS 98 × CCN 10 55.27 b BE 4 × CCN 51 55.06 b ICS 98 × ICS 9 54.04 b BE 4 × IMC 76 54.97 b ICS 98 × CEPEC 86 53.34 c BE 4 × CEPEC 86 54.22 b ICS 98 × P4B 52.44 c BE 4 × SGU 54 51.48 d ICS 98 × SGU 54 51.39 d CEPEC 89 × CEPEC 86 57.50 a OC 67 × CCN 51 56.49 a CEPEC 89 × IMC 76 54.55 b OC 67 × NA 33 54.87 b CEPEC 89 × NA 33 54.37 b OC 67 × IMC 76 54.82 b CEPEC 89 × P4B 54.06 b OC 67 × SGU 54 54.15 b CEPEC 89 × CCN 51 53.46 c OC 67 × ICS 9 54.13 b CEPEC 89 × ICS 9 53.32 c OC 67 × P4B 53.57 c CEPEC 89 × SGU 54 53.10 c OC 67 × CEPEC 86 49.98 e CSUL 7 × CCN 51 56.77 a RB 39 × CEPEC 86 57.29 a CSUL 7 × NA 33 55.09 b RB 39 × SGU 54 56.17 a CSUL 7 × SGU 54 54.92 b RB 39 × P4B 56.08 a CSUL 7 × IMC 76 53.46 c RB 39 × NA 33 55.87 a CSUL 7 × ICS 9 52.57 c RB 39 × CCN 51 55.16 b CSUL 7 × CCN 10 52.30 c RB 39 × ICS 9 54.18 b CSUL 7 × P4B 51.97 c RB 39 × CCN 10 53.69 c CSUL 7 × CEPEC 86 51.83 c RB 39 × IMC7 6 50.35 e EEG 29 × CCN 51 56.72 a SCA 6 × IMC 76 57.46 a EEG 29 × NA 33 55.28 b SCA 6 × ICS 9 56.61 a EEG 29 × P4B 54.88 b SCA 6 × CCN 51 56.34 a EEG 29 × IMC 76 54.04 b SCA 6 × P4B 56.03 a EEG 29 × ICS 9 53.91 b SCA 6 × CCN 10 55.48 b EEG 29 × SGU 54 53.04 c SCA 6 × NA 33 55.14 b EEG 29 × CEPEC 86 52.65 c SCA 6 × SGU 54 54.56 b EEG 29 × CCN 10 48.70 f SCA 6 × CEPEC 86 52.66 c Means followed by the same letters belong to the same group according to the Scott–Knott test at 5% probability. In general, genotypes collected from the Amazon region (CSUL, SPA, NA, PA, CJ, POUND, and RB) tended to have fat contents within the commercially desirable range (> 54%), whereas genotypes from the Trinitario/Criollo collection (UF, SGU, OC, ICS, P, and CC) commonly exhibited lower values (< 53%). In addition, low fat content was observed in Ecuadorian genotypes (EET) and traditional cocoa populations from Bahia and Espírito Santo[11].
The physical and chemical characteristics of cocoa seeds affect the final quality of beans processed to produce cocoa liquor and its derivatives. The clonal cultivar CCN 51 is highly productive, disease-resistant, and widely cultivated in Ecuador, Brazil, and Peru. Although it has a very acidic and watery pulp, resulting in beans with low organoleptic quality after traditional fermentation, its high fat content makes it a viable option for cocoa fat production.
GCA and SCA for seed fat content percentage
-
In a cocoa breeding program, evaluating only the parental genotypes is insufficient, particularly for traits such as fat content, which have quantitative genetic inheritance. Therefore, it is essential to obtain information on the combined abilities of these genotypes[30]. According to the diallel analysis, significant differences (p < 0.01) were observed for GCA in Groups I and II, as well as for SCA in hybrid combinations for fat percentage (Table 5). This suggests that both additive and nonadditive genetic effects are important for the expression of this trait.
Table 5. Summary of the ANOVA for GCA and SCA of cocoa clone parents from Groups I and II for fat content (%) in cocoa seeds based on analysis of the mean data from diallel crosses.
Source of variation Degrees of freedom Fat (mean square) Treatments 79 8.7871** Groups 1 2.2578* GCA Group I 7 9.9113** GCA Group II 7 11.8887** SCA I × II 64 8.4269** Error 77 0.4608 Effect of the mean 54.1872 Effect of d1 −0.2656 Effect of d2 0.2656 ** and *: Significant at the 1% and 5% levels, respectively, by the F-test. In other words, the values observed in the hybrids for each evaluated trait resulted from the parents' GCA and SCA. These findings align with those reported by Lockwood and Pang[16], who identified both additive and dominant genetic effects on fat content.
In the present study, the quadratic genetic components of the SCA and GCA for fat content were found to be similar. This proximity between GCA and SCA suggests that additive, dominant, and epistatic gene effects are involved in the expression of fat content in cocoa. Other studies have reported significant mean squares for GCA and SCA in production traits[19], with GCA effects being superior for traits related to pod size and seed weight[30,31], as well as pod production[32,33]. The same pattern was observed for the average seed weight per pod[34], and seed shape and size[35].
The parental clones that exhibited the highest GCA estimates were SCA 6, CEPEC 89, and BE 4 in Group I and CCN 51, ICS 9, and P4B in Group II (Table 6). Given the importance of increasing the fat content in cocoa breeding, these genotypes can be used for crosses in ongoing reciprocal recurrent selection programs to develop improved populations in Groups I and II with high frequencies of alleles or allele combinations that are favorable for increased fat content.
Table 6. Estimates of GCA of parental clones from Groups I and II for fat content (%) of cocoa seeds, based on the mean diallel cross data.
Clone Fat Group I (mothers) BE 4 0.4050* CEPEC 89 0.4188* CSUL 7 −0.0484 EEG 29 −0.4751 ICS 98 0.1750 OC 67 −1.3240 RB 39 0.1245 SCA 6 0.7241* Group 2 (fathers) CCN 10 −0.7303 CCN 51 1.4939* CEPEC 86 −0.1698 ICS 9 0.3156* IMC 76 −0.3623 NA 33 −0.3690 P4B 0.2922* SGU 54 −0.4703 * Clones showing high GCA. Clones EEG 29, CCN 10, and OC 67 presented the lowest estimates of GCA. These findings align with the results reported by Pires et al.[11], who reported a lower average fat content (< 53%) in the Trinitario/Criollo and Ecuadorian genotypes.
Among the hybrid combinations, ICS 98 × IMC 76, CEPEC 89 × CEPEC 86, and SCA 6 × IMC 76 had values of around three, whereas OC 67 × NA 33, RB 39 × SGU 54, OC 67 × IMC 76, OC 67 × CCN 51, EEG 29 × NA 33, BE 4 × CCN 10, ICS 98 × NA 33, RB 39 × NA 33, and OC 67 × SGU 54 had values of around two (Table 7). Hybrid individuals from these progenies, characterized by a high fat content, productivity, and disease resistance, can be selected and multiplied through vegetative propagation and subsequently utilized in competitive clone trials aimed at developing superior clonal cultivars.
Table 7. Estimates of SCA for fat content of cocoa seeds in hybrid combinations formed by parental genotypes from Groups I and II, derived from the mean of diallel cross data.
Hybrid Fat Hybrid Fat BE 4 × CCN 10 2.0731* ICS 98 × CCN 10 1.6431 BE 4 × CCN 51 −1.0261 ICS 98 × CCN 51 0.0789 BE 4 × CEPEC 86 −0.2023 ICS 98 × CEPEC 86 −0.8523 BE 4 × ICS 9 1.3173 ICS 98 × ICS 9 −0.6377 BE 4 × IMC 76 0.7402 ICS 98 × IMC 76 3.1802* BE 4 × NA 33 1.1718 ICS 98 × NA 33 1.9668* BE 4 × P4B 0.2856 ICS 98 × P4B −2.2144 BE 4 × SGU 54 −2.6369 ICS 98 × SGU 54 −2.5019 CEPEC 89 × CCN 10 −0.0990 OC 67 × CCN 10 0.2750 CEPEC 89 × CCN 51 −2.6400 OC 67 × CCN 51 2.1329* CEPEC 89 × CEPEC 86 3.0688* OC 67 × CEPEC 86 −2.7133 CEPEC 89 × ICS 9 −1.6016 OC 67 × ICS 9 0.9563 CEPEC 89 × IMC 76 0.3113 OC 67 × IMC 76 2.3192* CEPEC 89 × NA 33 0.1379 OC 67 × NA 33 2.3808* CEPEC 89 × P4B −0.8333 OC 67 × P4B 0.4196 CEPEC89 × SGU54 −1.0358 OC 67 × SGU 54 1.7571 CSUL 7 × CCN 10 −1.1035 RB 39 × CCN 10 0.1135 CSUL 7 × CCN 51 1.1423 RB 39 × CCN 51 −0.6457 CSUL 7 × CEPEC 86 −2.1339 RB 39 × CEPEC 86 3.1531 CSUL 7 × ICS 9 −1.8843 RB 39 × ICS 9 −0.4423 CSUL 7 × IMC 76 −0.3114 RB 39 × IMC 76 −3.5944 CSUL 7 × NA 33 1.3202 RB 39 × NA 33 1.9322* CSUL7 × P4B −2.4560 RB 39 × P4B 1.4760 CSUL7 × SGU54 1.2565 RB 39 × SGU 54 2.3285* EEG 29 × CCN 10 −4.2769 SCA 6 × CCN 10 1.2990 EEG 29 × CCN 51 1.5139 SCA 6 × CCN 51 −0.0652 EEG 29 × CEPEC 86 −0.8923 SCA 6 × CEPEC 86 −2.0814 EEG 29 × ICS 9 −0.1177 SCA 6 × ICS 9 1.3882 EEG 29 × IMC 76 0.6952 SCA 6 × IMC 76 2.9161* EEG 29 × NA 33 1.9368* SCA 6 × NA 33 0.6027 EEG 29 × P4B 0.8756 SCA 6 × P4B 0.8315 EEG 29 × SGU 54 −0.2019 SCA 6 × SGU 54 0.1190 * Hybrids showing high SCA. In a reciprocal recurrent selection program, it is crucial to obtain parental genotypes with high and positive GCA estimates for the trait. In addition, these parents should exhibit high SCA when crossing Groups I and II to achieve superior hybrids. Several parental genotypes listed in Table 7 met this criterion, demonstrating the potential for good allele complementarity in their combinations.
Fat content can be increased by enhancing the seed yield per hectare, increasing the amount of fat in the seeds, or both simultaneously. The latter approach must consider that seed yield and fat percentage are not correlated and, in principle, are independent traits[14]. Therefore, during the selection process, it is important to identify parents that possess a high frequency of favorable alleles for these traits, either within the same genotype or across different genotypes, to develop segregating populations for selection.
Moreover, given the significant role of SCA in both fat content and yield, the development of improved populations through reciprocal recurrent selection schemes that provide good allele complementarity should be explored to obtain superior cultivars.
-
The evaluated genotypes, comprising 16 parental clones and 62 progenies, exhibited significant differences in seed fat content. These differences benefit breeding programs because genetic gains can be achieved by selecting superior genotypes. Additive effects were superior to dominant effects on fat content in cacao. The parents SCA 6, CEPEC 89, BE 4, CCN 51, ICS 9, and P4B can be recommended for crosses when the breeding program's goal is to increase seed fat content. Conversely, EEG 29, CCN 10, and OC 67 should not be used, as they tend to reduce the values of the trait under analysis.
-
The authors confirm their contributions to the paper as follows: study conception and design: Almeida AAFd, Pires JL, Santos EA, and Ahnert D; data collection: Melo HL; analysis and interpretation of results: Melo HL, Almeida AAFd, Barretto WS, Santos EA, and Ahnert D; draft manuscript preparation: Melo HL, Almeida AAFd, Silva GS, Zhang D, and Ahnert D. All authors reviewed the results and approved the final version of the manuscript.
-
All data generated or analyzed during this study are included in this published article.
-
The first author thanks the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), Brazil, for granting a fellowship for Master's study. The second author gratefully acknowledges the CNPq for granting a fellowship of scientific productivity. We thank CEPLAC for conducting the field experiment and laboratory analyses, and Editage (www.editage.com) for English language editing.
-
The authors declare that they have no conflict of interest.
- Copyright: © 2026 by the author(s). Published by Maximum Academic Press, Fayetteville, GA. This article is an open access article distributed under Creative Commons Attribution License (CC BY 4.0), visit https://creativecommons.org/licenses/by/4.0/.
-
About this article
Cite this article
Melo HL, Almeida AAFd, Pires JL, Silva GS, Barretto, et al. 2026. Estimates of combining ability for cocoa fat content in seeds of parental clones and their hybrids in diallel crosses. Beverage Plant Research 6: e002 doi: 10.48130/bpr-0025-0032 
Estimates of combining ability for cocoa fat content in seeds of parental clones and their hybrids in diallel crosses
- Received: 07 February 2025
- Revised: 23 July 2025
- Accepted: 03 September 2025
- Published online: 15 January 2026
Abstract: Cocoa butter or fat is one of the most valuable products extracted from cocoa beans. It is used for the manufacture of chocolate, pharmaceutical, and cosmetic products. Breeding programs in the world are trying to develop cocoa varieties with a high fat content and productivity to increase the production of fat per hectare. In a laborary, we extracted cocoa fat from seeds of 16 parental clonal genotypes and their 62 hybrid combinations tested in a field experiment. The hybrids plants were obtained by controlled pollination using a partial diallel mating design. Analysis of variance, Scott–Knott mean comparison tests, a test of the general combining ability (GCA) of parental clones, and tests of the specific combining ability (SCA) of hybrid combinations were performed. Significant differences were obtained for fat content in seed of the parental plants. Genotypes from the Amazon region demonstrated a higher fat content, whereas genotypes from Trinidad (Trinitario) and Criollo exhibited lower values. Diallel analysis indicated that both additive and nonadditive effects were relevant to fat content, with the additive effects having a greater influence. Regarding the fat content, it is recommended to use the parents SCA 6, CEPEC 89, BE 4, CCN 51, ICS 9, and P4B for crosses aiming to increase fat, whereas the genotypes EEG 29, CCN 10, and OC 67 should be avoided because of their tendency to reduce fat content.
-
Key words:
- Additive effect /
- Cocoa breeding /
- Genotype selection /
- Theobroma cacao L.