4.1. Crossing Certification
A crossing certification and contaminant detection in breeding populations are essential for the efficiency of plant breeding programs. This is particularly true for perennial and long cycle species such as coffee. Crossing certifications determine whether the genetic material selected for future generations is indeed the desired target since contaminating hybrids can affect gain in subsequent generations.
As
C. arabica is an autogamous cleistogamous species, the hybridization process is manual, involving the emasculation of the flower to be used as a female parent. However, despite due care, self-fertilization may occur before emasculation and crossing, generating undesirable populations for the program. This self-fertilization contamination was the case with C1-3, C1-6, C1-7 and C1-8 coffee trees of the study population in this report. Self-fertilization contamination was previously reported in coffee [
36]. Valencia et al. [
36] analyzed the BA-124-Kf marker associated with the
SH3 gene in a F
1 population and detected some coffee trees with no marker. They considered these plants contaminated and attributed this result to the fact that the respective hybridization had not occurred.
Hybrid coffees with genetic markers different from their parents were also detected in our work. This contamination may have occurred due to a crossing with pollen from other coffee trees than those selected as parents, or due to mixed seeds. In breeding programs, seed mixing can occur during the harvest period, as a large number of genotypes are handled. Thus, the use of molecular markers is essential to confirm plant identity. Yashitola et al. [
37] used SSR markers to confirm the genetic purity of rice hybrids, concluding that this practice is considerably simpler than standard growth tests, which involve developing plants to maturity and evaluating their morphological and floral characteristics.
This study confirmed that the use of molecular markers for crossing certification is an important tool for recurrent selection breeding programs. This approach reduces labor, time, and financial resources by allowing the identification of undesirable plants.
4.2. Genetic Parameters and Components of Variance
The analysis of nine desired morphological and agronomic traits in coffee trees revealed genetic variability with a low magnitude between populations. This low variability was previously detected, especially in Arabica coffee cultivars [
38,
39], which were used in the crosses for the present study. These results highlight the importance of using MAS in Arabica coffee breeding populations to identify and increase genetic gain. The association of phenotypic and genotypic data allowed the selection of superior hybrids to compose the subsequent recurrent selection cycles and allowed hybrid selection with greater genetic variance (GV) and traits of economic relevance.
The populations studied exhibited low heritability for most analyzed traits. For yield (Y), a low value was expected as this trait is polygenic. However, vegetative vigor (Vig) showed the highest heritability (0.13), indicating the possibility of successful selection and the positive correlation of this trait with Y. According to Severino et al. [
40], the selection of coffee with high Vig can increase Y since vegetative vigor has a direct effect on coffee productivity.
For data repeatability, most of the evaluated traits showed values below 10% in the studied populations. Resende [
41] classified the repeatability as high (r ≥ 0.60), medium (0.30 ≤ r < 0.60), and low (r < 0.30). The low repeatability values observed indicate that a large number of repetitions are required to achieve a satisfactory determination value. The use of this coefficient increases the plant evaluation efficiency and reduces the time and labor [
42], thereby decreasing costs and optimizing the breeding program.
On the other hand, according to Guerra et al. [
43], the coefficients were used to determine environmental effects (c
2parc) to quantify variability within blocks. The low values observed indicate a low level of environmental variation between the plots and within blocks, which suggests that the experimental design was adequate and that environmental homogeneity remained within the blocks.
Three analyzed traits were significant by the ANADEV approach; vegetative vigor (Vig), ripening fruit size (RFS), and cercosporiosis incidence (Cer). This significant effect indicates the existence of genetic variability in the progeny, confirming the genetic variance (GV) results.
4.3. Marker-Assisted Selection for H. vastatrix and C. kahawae Resistance
The populations evaluated in this study were formed by parent plants with traits of agronomic interest and alleles conferring resistance to the main diseases currently affecting the coffee crop. Thus, UFV 311-63 was used as a source for the
SH3 resistance gene. This coffee corresponds to an F
3 plant derived from the Indian selections S.26 and S.31 backcrossed with the Kent or Coorg cultivars carrying the
SH2,
SH3 and
SH5 alleles. The
SH3 gene has been associated with long-lasting resistance and confers resistance to a variety of
H. vastatrix types. This gene, derived from
C. liberica, was introduced into
C. arabica using tetraploid coffee trees from the Indian selection and natural
C. arabica ×
C. liberica [
25]. UFV 311-63 was used in C2T, C4T and C12T crosses, justifying the identification of molecular markers associated with
SH3 only in coffee originated from crosses and the allocation to group 2 in the diversity analysis.
Although homozygous-resistant hybrids are ideal for generation progression, it was not possible to obtain this type of hybrid for the SH3 gene because the UFV 311-63 parent plant was crossed with non-carrier coffee trees. However, coffee trees identified as heterozygous can be used in the program, because molecular markers can be used to select progeny with the desired target gene in subsequent generations.
Valencia et al. [
36] evaluated eight F
1 populations that contained the parental plants S288/23 and BA-2 (
SH3 gene carriers) in their genealogy using the BA-124-Kf marker alone. In the present study, four molecular markers linked to and flanking the
SH3 gene were used to improve selection accuracy. Hybrids with the four markers linked to the target gene only were selected as resistant hybrids. This strategy minimized the error that can be introduced by selecting hybrids that may have contained the marker but lost the allele by recombination.
Additional resistance sources were also used in the 12 crosses since the resulting populations contained various genes conferring resistance to other
H. vastatrix types
, in addition to hybrids with the
Ck-1 resistance gene to the
C. kahawae pathogen. These alternative sources of resistance corresponded to HdT derivatives, which are carriers of the gene combinations
SH6,
SH7,
SH8,
SH9 and
SH? originating from
C. canephora [
23].
Molecular markers flanking the two QTL associated with resistance to the three different pathotypes were also analyzed within the populations to identify hybrids containing other
H. vastatrix resistance alleles. The markers revealed that progeny from the C6T population (
Table S6) contained the two QTL, therefore expressing the BBC genotype (
Table S6). As they already presented pyramided alleles, the selection of these hybrids for future generations is promising. In the other populations, hybrids with the BBcc, Bbcc and bbC_ genotypes were identified (
Table S6), which did not have pyramided alleles but were also resistant and could be selected for the next generation. According to Pestana et al. [
35], the resistance of coffee trees to races I and II and pathotype 001
H. vastatrix is conferred by two independent dominant loci. Thus, the presence of a dominant allele in one of these two loci is sufficient for the plant material to be resistant and thus selected for the next generation.
The populations were also analyzed for the presence of resistance allele to
C. kahawae, as HdT derivatives were used in several crosses. The identification of hybrids containing the
C. kahawae resistance gene via molecular markers was especially important for
C. arabica breeding programs as it allows the selection of resistant plants in the absence of the pathogen. This strategy promotes the preventive management of this disease since CBD has not been reported in Latin America. However, there are concerns regarding the introduction of this pathogen to this region due to the damage caused by the disease in Africa. CBD is considered the main production limiting factor in coffee-producing countries in Africa [
11].
The analysis of the two molecular markers flanking the
Ck-1 gene showed the possibility of recombination events between several hybrids, which could have resulted in the loss of the desired resistance alleles. As some hybrids contained CBD-Sat207 or CBD-Sat235 marker alone, hybrids of both markers for the
Ck-1 resistance locus were considered resistant. According to Frisch and Melchinger [
44], the efficiency of MAS results from the distance and orientation of the markers in relation to the gene. To obtain high efficiency, it is essential that the distance between the markers and the genes is as short as possible and that the markers used are flanking the gene.
Considering the molecular markers associated with the four loci of resistance to the two diseases (CLR and CBD), 11 hybrids belonging to the C2T and C4T populations exhibited four pyramided resistance alleles. These hybrids corresponded to the AaBbC_Dd genotype (
Table S7). In addition, there were other 10 resistant homozygous hybrids for the
Ck-1 gene in the C5T population with the genotypes aaBbC_DD or aaBbccDD, as well as 39 other homozygous-resistant hybrids for LG2 QTL (C1T, C6T, C8T, C9T, C10T and C11T populations). Among the latter, seven hybrids from the C6T population carried LG5 QTL, corresponding to the genotype aaBbC_Dd (
Table S7). These hybrids are thus important parent plants for breeding programs that seek to develop cultivars with multiple and long-lasting resistance since their pyramided alleles can be readily used by the next generations.
Both genotypic and phenotypic data were used to select the best hybrids (
Table 6). The phenotypic data with a selection index of 30% ranked 29 hybrids as superior. The genotypes of these superior hybrids were obtained using the corresponding molecular markers, allowing the observation of pyramided alleles of the four loci associated with
H. vastatrix and
C. kahawae resistance. The C4-10 hybrid, the first in the mean rank, presented resistance alleles from all four loci. The C2-10 and C4-9 hybrids also showed pyramided alleles originating from all four alleles and were in the mean rank at positions 13 and 24, respectively. The other hybrids showed different combinations of the four allele loci, which hence could be considered for subsequent crosses to ensure the pyramiding of the loci.
This study proposes a method to develop a recurrent selection program using molecular markers for the coffee breeding program (
C. arabica), as presented in
Figure 1. Such a program allows the formation of a base population followed by the production of new cultivars containing pyramided resistance alleles to
H. vastatrix and
C. kahawae, in addition to improving other traits of economic importance.