different

2019 , 2020. ¹ This work is part of the first author’s Doctoral thesis. The (FAPESB) and the (CNPq) for the and last This study aimed to select genotypes resistant to witches’ broom (WB) and black pod (BP), major cacao diseases in Brazil, as well as incorporate resistance genes to moniliasis supplemented by clones EET75 and UF273, forming populations of second-cycle recurrent selection. Moniliophthora perniciosa (2 × 10 5 basidiospores/mL) was inoculated on 30-day-old seedlings from 72 different progenies, being assessed 60 days later, and a mixture of four isolates of Phytophthora palmivora (3 × 10 5 zoospores/mL) was inoculated on leaf discs from 58 progenies, observing lesions after seven days. Significant effects of progeny were observed in the tests of resistance to both diseases (p < 0.05). Scavina-6 expressed resistance to both pathogens, 26 crosses did not differ from free-pollinated progenies of Scavina-6 for WB, and ten crosses were higher and 27 similar for BP. Eight crosses were largely resistant to both diseases.

Obtaining genetic material resistant to these diseases with desirable agronomic characteristics, as well as organoleptic qualities that contribute to obtaining adequate chocolate quality, is the main objective of genetic improvement at present (Moreira et al., 2016;Pimenta Neto et al., 2018).
Cacao (Theobroma cacao L.) is a species of Neotropical origin in the Americas that occurs spontaneously from southern Mexico to Bolivia (Monteiro & Ahnert, 2012). This wide geographical range shows distinct edaphoclimatic conditions that allowed the development of vast genetic diversity with a varied population, representing genetic resources with the potential to obtain varieties resistant to diseases. Several cacao populations have been generated at the Cocoa Research Center (Cepec) in Ilhéus, Bahia, to obtain improved genotypes aiming at selecting clones with more durable resistance, which carry genes from different sources of resistance, as well as increasing the genetic basis in order to hinder pathogen evolution (Paim et al., 2006;Yamada et al., 2008;Lopes et al., 2011;Benjamin et al., 2016;Gramacho et al., 2016;Pimenta Neto et al., 2018).
Thus, this study aimed to form populations of secondcycle recurrent selection for resistance to WB using the North Carolina II design, crossing first-cycle selections with genetically distant and productive materials and until second leaf flushing was 15 mm (approximately 30 days), and then inoculated by depositing 30 µL of the suspension of 2 × 10 5 basidiospores/mL of M. perniciosa in an agar-water medium at 0.3% on the apical meristem. On the day before inoculation, leaves of first apical flushing were reduced to 1/3 to accelerate the growth of second flushing and better expose the area of the apical bud. After inoculation, seedlings were taken to the humid chamber at 25 °C and relative air humidity of 100% for 48 h. They were then transferred to a greenhouse, where they remained until the end of the assessments, i.e., 60 days after inoculation. Because seeds from the 72 progenies were obtained at different times, ten inoculations were performed at different times, but all of them with the three progenies of the controls.
Plants with presence and absence of WB were assigned with scores 01 and 00, respectively. The type of brooms B, i.e., terminal (TB), axillary (AB), dry (DB), and cotyledonary broom (CB), was also assessed. In addition, AB higher than 1 cm was quantified, and TB length was measured. For the data analysis, the disease index was calculated by the following Luz Index (Rodrigues et al., 2019): DI = TB + (0.1 × TBL) + AB + (0.2 × NAB) + CB + (4.3 × DB), where TB is the presence of terminal broom, TBL is the terminal broom length, AB is the presence of axillary broom, NAB = number of axillary brooms higher than 1 cm, CB is the presence of cotyledonary broom, and DB is the presence of dry broom. The coefficient that multiplies DB was defined to allow the plant with dry broom having a DI higher than all the others that did not die. The coefficient for TBL was defined to allow plants with larger terminal broom generating, together with TB, a value close to two, i.e., the double the DI presented by a plant with a very small terminal broom. Similarly, the coefficient for NAB was defined to allow plants with the highest number of large axillary broom having a DI corresponding to twice the DI of plants with only axillary broom lower than 1 cm. The randomized block design was used at each inoculation or test, with 14 plants per replications (56 plants per crossing and inoculation time, repeated once with an equal number of samples).
A model with the sources of variation test or inoculation and progeny was used to analyze differences between progenies in an incomplete block system. Comparisons between the corrected means of progeny for the effects of test or inoculation were performed by the T-test (SAS, 2002). We did not consider to which genetic design or which of the diallels the progeny belonged.
Because progenitor-corrected means are not estimable in the model with the sources of variation mother, father, and test or inoculation (i.e)., the means of Rev. Ceres, Viçosa, v. 67, n.5, p. 383-394, sep/oct, 2020 mother corrected by principle of incomplete blocks are not estimable at the same time for effects of father and test or father, corrected for mother and test because the tests mixed progenies of the three genetic designs), DI correction was applied for each test to analyze differences between progenitors. In this case, the corrected index for each plant is equal to the original index (DI) multiplied by the inverse of the sum of the means of the indices of the three controls in that test and divided by the sum of the overall means of the controls in all tests. Thus, DIs of each plant were corrected for the effect of the test to which they belong by the ratio between the mean DIs of the controls in that test and their overall mean DIs for all tests. The effects of progenitors were analyzed in the model with the sources of variation father and mother for each of the three diallels from the corrected DI. The previous model was used to compare fathers within mothers or mothers within fathers, with uncorrected DI and model with the sources of variation test or inoculation and progeny. After the assessments, diseased plants were incinerated, and healthy plants were selected to further assessment of BP resistance.

Assessment tests for black pod
Fifty-eight progenies among surviving plants and without the presence of WB symptoms from the previous experiment were selected to be tested for resistance to BP using the leaf disc method (Nyassé et al., 1995). The isolates of P. palmivora used were 1744, 1778, 1845, and 1913, obtained from the Arnaldo Medeiros collection at Cepec, originated from cacao pods samples collected in the following counties and years: Uruçuca (2011), Camacan (2011), Mutuípe (2011) and Belmonte (2010), respectively. These isolates were selected based on their high aggressiveness to cacao among 100 P. palmivora isolates tested in previous studies (Lessa, 2017). Healthy leaves of surviving plants from crossings were collected and taken to the laboratory of Phytophthora, where they were sanitized and 15-mm diameter discs were cut from the leaf blade. These discs were arranged with the abaxial part up in boxes containing foam moistened with sterile water to form a humid chamber and provide favorable conditions for pathogen development.
A 10-µm aliquot of zoospore suspension from the mixture of four isolates, obtained according to the protocol of the Luz et al. (2008) was adjusted to a concentration of 3 × 10 5 zoospores/mL and placed on the center of each leaf disc. The boxes were closed and incubated at 25 °C in the dark for seven days, when the assessment was performed using a scoring scale developed by Nyassé et al. (2002) with values varying from 0 to 5. The disease severity index (DI) was determined for each genotype from the scores using the equation Two experiments were set up with all the 58 progenies in a randomized block design with four replications containing ten discs per clone, totaling 40 discs inoculated per clone and experiment. The analysis of differences between means of progenies was conducted under the model with the sources of variation experiment and treatment, without considering the genetic designs. The model experiment, mother, and father was used to analyze differences between progenitors for each of the three diallels. The previous model was used to analyze mother within father or father within mother.

Assessment for witches' broom
The proposal presented here for defining the disease index for the early assessment of cacao seedlings took into account a very important factor: the dry broom. Therefore, plant death due to the disease was considered in this study. In addition, the methodology gives greater or lesser weight to the types of brooms formed according to the number of axillary broom and size of terminal broom. This formula also included the presence of cotyledonary broom because of the relatively high frequency of this type of symptom in plants from some genotypes.
Significant effects for inoculation test (p = 0.0415) and progeny (P < 0.0001) were observed by the F-test, with ten inoculations tests with different progenies in each test and three controls in all tests. As a primary and extremely important element, among the 72 crossings, 69 differed from the two susceptibility controls (Catongo and SIC23), showing the effectiveness of progenitor selection and prospects of gain with plant selection within these progenies (Table 2). Non-distinct crossings from one or both controls were [(MO20 x AMAZ15 (5A)) X UF273 (13)], [(CAB157 x MO20 (5)) X EET75 (12)], and (CAB148 x MO20 (4)) X UF273 (13)], all of them with one of the progenitors selected only for moniliasis and none of them with any Scavina ancestry.
At the other end of the list, 26 crossings did not differ from the control progeny of Scavina-6: [MCB09 (10) X EET75 (12)], [(CEPEC86 x RB36 (1A)) X RLF1938 (11) (6)) X EET75 (12)], and [(NA33 x RB39 (6)) X (SCA6 x RB36 (8))]. Scavina-6 is a clone resulting from the first studies aimed at resistance to cacao diseases started in the 1930s in Latin America and the Caribbean, and from Pound collections (Pound, 1938) in Peru. Per se and in the progeny was practically immune to WB in their first assessment in Trindade (Bartley, 2005), which would indicate possession of more than one resistance allele -all progeny practically immune -, with an exceptional behavior until today depending on fungus population. Scavina-6 is still widely used as a source of resistance to M. perniciosa in current breeding programs with cacao clones, but new sources of resistance need to be incorporated as a strategy to obtain lasting resistance (Pinto & Pires, 1998).
Thus, these various progenies have a mean behavior equivalent to that of a progeny known to have resistance alleles in all plants. Others, with means not so favorable but distinct from susceptible progenies and, therefore, carrying resistance alleles, may be derived from heterozygous parents for resistance even with more than one allele (in this case, in different loci), which could segregate and generate non-resistant plants, which would raise the mean DI of the progeny.
All progenies that differed from susceptible controls may provide resistant plants for selecting clones for assessment and indication of commercial varieties or generation of a new selection cycle. However, this selection will be carried out with resistance and productivity field data, especially from plants selected as resistant in this early selection phase, considering, in addition to the per se plant performance, the combining ability of each parent and the mean progeny performance.
Regarding the performance of fathers and mothers used in the three tests, MO20 x CCN34 (3) was the best mother in diallel 01, with no difference only from NA33 x RB39 (6) ( Table 3). Benjamin et al. (2016) used this last crossing in field assessments and concluded that RB39 is a highly promising source of resistance to WB, promoting the durability of this character when combined with other sources. This latter progenitor did not differ from the progenitors CSUL3 x CCN10 (1), CAB301 x CCN10 (2) and MO20 x CCN34 (3) which also showed low DI means.
The best father for crossings with MO20 x CCN34 (3) was SCA6 x P4B (7), but this crossing only differed significantly (p < 0.05) from the crossing with UF273 (13), which was, regarding resistance, selected only for moniliasis. The crossing [(MO20 x CCN34 (3)) X UF273 (13)] showed no ancestry of Scavina. The parents SCA6 x RB36 (8), RLF1938(11), and SJ02 (9), all with ancestry of Scavina, generated progenies with means close to that generated by the progenitor SCA6 x P4B (7) (means shown in Table 2 and probability of error for rejecting the hypothesis of equality between means not shown).

Controls
Continuation Table 2 Regarding the overall means of parents of diallel 01, the best parents were the mother SCA6 x RB36 (8) and clones RLF1938 (11) and UF273 (13), not distinct from each other and significantly different from all others.
The best mother in diallel 02 was CEPEC86 x RB36 (1A), which did not differ only from CCN10 x CAB324 (3A). The worst was the mother MO20 x AMAZ15 (5A), but the five did not present large differences in absolute values (Table 3).
The best fathers for mothers CEPEC86 x RB36 (1A) and CCN10 x CAB324 (3A) and overall means of diallel 2 were SCA6 x SGU114 (10A) and TSH1188 x CAB169 (9A). Clone RLF1938 (11), which is a selection carried out in a farm of the region, with probable ancestry of Scavina and also crossed with the first mother, as the two previous fathers, generated progenies with performance similar to that of the resistance pattern. In fact, the four crossings that had SCA6 x SGU114 (10A) Table 3: Mean performance of Moniliophthora perniciosa infection of fathers and mothers for the three studied genetic designs and the probability of error (P) for rejecting the hypothesis of equality between means by the T-test

P D C CM C (1) C (2) C (3) C (4) C (5) C (6)
CSUL3xCCN10 (1)  as father and four of the five with TSH1188 x CAB169 (9A) were as resistant as SCA6. The exception was observed for the crossing with the mother MO20 x AMAZ15 (5A), whose only resistant crossing was with 10A (Table 2). Diallel 03 showed no significant difference between the means of the mothers SJ02 (9), MCB09 (10), and RLF1938 (11), all clones selected in farms of the cacao region of Bahia due to their productivity and resistance to WB, as well as probable ancestry of Scavina. No significant differences were also observed between the two fathers of diallel 03, EET75(12) and UF273(13).

Assessment for black pod
The tests with leaf discs suggested by Nyassé et al. (1995) has been widely used for the assessment of resistance to cacao diseases (Santos et al., 2011;Bahia et al., 2015;Barreto et al., 2015), showing a high reliability regarding BP behavior in fruits (Pires et al., 1997;Santos et al., 2009). The species P. palmivora was used in resistance tests because it is a common cosmopolitan species in all cocoa producing regions (Luz et al., 2001). In addition, Risterucci et al. (2003) demonstrated that the selection to a single predominant species, such as P. palmivora in Bahia , provides significant genetic gains of resistance to the disease.
The means of infection caused by P. palmivora showed that differences regarding susceptible controls were not as clear as those found for M. perniciosa. This result was expected because the selection in the previous cycle for BP was primarily indirect by the priority selection for resistance to WB or moniliasis. In addition, many of the progenies did not differ or even surpassed, in average, the controls: 16 crossings were similar and 11 crossings had means higher when compared to those found for the susceptibility controls Catongo and SIC-23 (Table 4). Progenies more susceptible than susceptibility patterns were also observed in other studies (Santos et al., 2011;Bahia et al., 2015;Barreto et al., 2015).
Regarding the general combining ability in progenitors, the best mothers in diallel 01 were CAB148 x MO20 (4) and CSUL3 x CCN10 (1), not differing from each other and with significantly lower means than all other (1) Susceptibility control; (2) Resistance control.
Continuation Table 4 Rev. Ceres, Viçosa, v. 67, n.5, p. 383-394, sep/oct, 2020  (Table 5). The mother CAB148 x MO20 (4) was present in three crossings classified as resistant, with MCB09 (10), RLF1938 (11), and SCA6 x P4B (7), the first two not differing from each other and both different from the latter at 5% probability (means shown in Table 2 and probability of error for rejecting the hypothesis of equality between means not shown). The mother CSUL3 x CCN10 (1) was present in six crossings, five of them being the most resistant. The best combination of this mother was with clone SJ02 (9), not differing statistically from the combination with RLF1938 (11), UF273 (13), and SCA6 x P4B (7) (means shown in Table 4 and probability of error for rejecting the hypothesis of equality between means not shown).
For the fathers, the best performances were observed for clones SJ02 (9) and MCB09 (10), with means not statistically distinct and lower than those of the other progenitors. The third of the selections carried out in a farm, the clone RLF1938 (11), also presented a low mean infection, not differing from MCB09 (10). From the four crossings with SJ02 (9), two were among the best treatments, with the mothers CAB301 x CCN10 (2) and NA33 x RB39 (6).
In diallel 02, CCN34 x CAB301 (4A) was the best mother and EET75 (12) the best father, differing from the other progenitors. This mother, when combined with TSH1188 x CAB169 (9A), had a lower mean DI when compared to Scavina-6, with no statistical difference from each other. Rev. Ceres, Viçosa, v. 67, n.5, p. 383-394, sep/oct, 2020 EET75 (12) appears as the father in two of the most resistant treatments.
Clone SJ02 (9) contributed to the formation of three of the most resistant progenies to BP when crossed with mothers CSUL3 x CCN10 (1), CAB301 x CCN10 (2), and CAB157 x MO20 (5). The ancestry CSUL3 has been standing out as a progenitor in other tests for resistance to WB (Marita et al., 2001;Silva et al., 2010;Benjamin et al., 2016) and also for BP in field tests (Pires et al., 1997), as well as in the artificial inoculation on fruits with P. palmivora (Luz et al., 1996).

CONCLUSION
These results allow the early selection in the establishment of recurrent selection tests, future plant selection, which will be tested as clones and assessed regarding the possibility of becoming commercial varieties, and selection of progenitors for the next recurrent selection cycle. They also provide information on the potential of germplasm that can be used in other breeding programs. In addition to contributing to cacao farming in Bahia, they may also be useful for cacao farming in other regions.