Symptom severity
The first disease symptoms were noted 30 days post-planting in one Passiflora edulis genotype and six BC3 generation genotypes. By the 150-day mark, nearly all P. edulis genotypes exhibited symptoms, with the exception of plant 542, which only showed symptoms at 225 days. At the end of the assessments, some genotypes displayed reduced symptom severity.
Due to the quantitative nature of inheritance, identifying symptom-free genotypes becomes challenging as backcross generations progress. This is attributed to the increasing genetic similarity among genotypes since the likelihood of recombination between donor and recurrent genomes diminishes with each backcrossing cycle. This results in a higher contribution of the recurrent genome in the progeny (93.75%), leading to greater similarity between progeny and the recurrent parent (Souza et al. 2022). Given that the evaluated families are in the BC3 stage, the P. edulis genome, which is susceptible, has already been almost entirely restored.
Passiflora setacea, the resistant species, exhibited mild virus symptoms in some genotypes after 240 days of evaluation. By the 300-day mark, all genotypes of this species were affected. Nine months post-planting, all P. edulis individuals showed disease symptoms. During this period, 96.2% of BC3 families and 9.52% of P. setacea individuals displayed mild disease characteristics. A year later, at the end of the assessments, all plants exhibited disease symptoms with varying severity levels.
The variability in symptom severity caused by CABMV can also be attributed to the number of genes involved in resistance inheritance. Santos et al. (2019) identified seven loci with minor effects (QTLs) linked to virus resistance.
Estimates of genetic parameters and gains
The AUDPCM values for the evaluated genotypes varied significantly, ranging from 371.25, indicative of resistance, to 1076.25, signaling susceptibility. The lowest AUDPCM results were observed in P. setacea (from 371.25 to 393.75) and in BC3 segregating population plants 450, 340, and 30 (397.5, 412.5, and 420, respectively). These genotypes exhibited only mild disease symptoms, such as slight mosaic patterns without leaf distortion and few leaves with symptoms, and were thus classified as resistant.
Conversely, the highest AUDPCM values were recorded not only in P. edulis but also in BC3-501 family plants 533 (1076.25) and 365 (990), as well as in BC3-17 family plant 417 (975). Among these, P. edulis plant 426 stood out with an AUDPCM value of 948.75. These genotypes demonstrated more severe disease manifestations, including intense mosaic patterns, blistering, and leaf distortions across practically the entire area occupied by the plant, leading to their classification as highly susceptible.
The genetic parameters for AUDPCM were estimated (Table 3) and significant genotypic differences were detected at a 1% probability level using the Chi-square test, highlighting the presence of genetic diversity within the study population. This diversity presents an opportunity for the selective breeding of superior genotypes in subsequent backcrossing phases.
Table 3
Analysis of deviance and estimates of genetic parameters via the REML procedure, for resistance to cowpea aphid-borne mosaic virus based on the mean area under the disease progress curve (AUDPCM), in passion fruit genotypes from the BC3 segregating population.
AUDPCM |
Effect | Deviance | LRT (χ2) |
Genotype | 4815.64 | 18.71** |
Full model | 4796.93 | |
Genetic parameters | AUDPCM |
σ2g | 2426.47 |
σ2 plot | 399.06 |
σ2 within | 8961.38 |
σ2p | 11786.92 |
h2a | 0.41 +- 0.12 |
h2ad | 0.27 |
h2mp | 0.95 |
C2plot | 0.03 |
Acprog | 0.97 |
Mean | 621.63 |
** significant at 1% probability using the Chi−square test. LRT(x2): likelihood ratio test. σ2g: genotypic variance between full−sib progenies; σ2parc: between−plot environmental variance; σ2within: within−plot residual variance; σ2p: individual phenotypic variance; h2a: individual narrow−sense heritability; h2ad: within−plot additive heritability; h2mp: progeny mean heritability; C2plot: coefficient of determination of plot effects; Acprog: progeny selection accuracy.
Table 3 reveals mean estimates for narrow-sense individual heritability (h2a), and relatively low within-plot additive heritability (h2ad). These outcomes can largely be attributed to the polygenic nature and complex inheritance patterns of the traits, which are typically influenced by numerous genes each contributing minor effects and are substantially affected by environmental factors (Cruz et al., 2014).
When examining heritability based on the family mean (h2mp), the value for AUDPCM was notably high (0.95). This is anticipated since environmental effects are reduced by averaging between the number of replicates and the number of plants per plot, thereby aligning phenotypic values more closely with genotypic values.
The heritability values for AUDPCM suggest that selection based on family performance is the most effective approach for achieving significant genetic progress for the trait in question (Ramalho et al., 2012a).
In terms of selection accuracy, Resende et al. (2014) recommend aiming for values over 70%. This study achieved an accuracy estimate of 97%, indicating a high level of precision and reliability in the predicted values.
A ranking of the 30 genotypes with the lowest genetic values for the trait under evaluation was established. The efficacy of a breeding program depends on the selection of the best genotypes that will be used as parents for future generations, based on the additive genetic values of the individuals (Cruz et al., 2014). Consequently, genetic gains were predicted and new means were estimated (Table 4).
Table 4
Ranking of the 30 genotypes with the lowest estimates for additive genetic value, genetic gains, and new predicted means, estimated via REML/BLUP, in passion fruit genotypes from the third backcross generation, for the mean area under the disease progress curve (AUDPCM).
| | | | AUDPCM | |
Rank | Genotype | Block | Family | Phenotypic value | Genetic value | Gain (%) | New mean |
1 | 450 | 4 | 293 | 397.50 | -103.73 | -7.85 | 613.78 |
2 | 675 | 6 | 293 | 423.75 | -101.94 | -7.65 | 613.98 |
3 | 163 | 2 | 293 | 420.00 | -94.63 | -7.45 | 614.18 |
4 | 340 | 3 | 293 | 412.50 | -94.12 | -7.26 | 614.37 |
5 | 476 | 5 | 293 | 438.75 | -90.03 | -7.08 | 614.55 |
6 | 475 | 5 | 293 | 446.25 | -88.01 | -6.91 | 614.73 |
7 | 673 | 6 | 293 | 476.25 | -87.73 | -6.73 | 614.91 |
8 | 334 | 3 | 293 | 442.50 | -85.99 | -6.56 | 615.07 |
9 | 483 | 5 | 293 | 453.75 | -85.97 | -6.38 | 615.24 |
10 | 183 | 2 | 293 | 457.50 | -84.47 | -6.22 | 615.42 |
11 | 441 | 4 | 293 | 476.25 | -82.41 | -6.05 | 615.58 |
12 | 474 | 5 | 293 | 468.75 | -81.91 | -5.88 | 615.75 |
13 | 434 | 4 | 293 | 480.00 | -81.39 | -5.72 | 615.91 |
14 | 482 | 5 | 293 | 472.50 | -80.89 | -5.56 | 616.07 |
15 | 481 | 5 | 293 | 472.50 | -80.89 | -5.39 | 616.24 |
16 | 674 | 6 | 293 | 506.25 | -79.60 | -5.23 | 616.40 |
17 | 445 | 4 | 293 | 487.50 | -79.36 | -5.07 | 616.56 |
18 | 389 | 4 | 153 | 453.75 | -79.12 | -4.91 | 616.73 |
19 | 677 | 6 | 293 | 510.00 | -78.58 | -4.75 | 616.88 |
20 | 446 | 4 | 293 | 491.25 | -78.34 | -4.58 | 617.05 |
21 | 396 | 4 | 153 | 457.50 | -78.09 | -4.42 | 617.21 |
22 | 106 | 1 | 153 | 442.50 | -76.95 | -4.26 | 617.37 |
23 | 41 | 1 | 293 | 468.75 | -76.01 | -4.11 | 617.53 |
24 | 35 | 1 | 293 | 468.75 | -76.01 | -3.94 | 617.69 |
25 | 579 | 5 | 153 | 468.75 | -75.76 | -3.78 | 617.85 |
26 | 384 | 4 | 153 | 468.75 | -75.05 | -3.63 | 618.01 |
27 | 29 | 1 | 293 | 472.50 | -74.99 | -3.46 | 618.17 |
28 | 486 | 5 | 293 | 495.00 | -74.81 | -3.31 | 618.33 |
29 | 479 | 5 | 293 | 495.00 | -74.81 | -3.15 | 618.48 |
30 | 448 | 4 | 293 | 506.25 | -74.28 | -2.98 | 618.65 |
Phenotypic value: intensity of damage to the plant caused by CABMV at the end of the evaluations; New mean: expected mean of the next generation regarding the intensity of damage to the plant caused by CABMV; AUDPCM: mean area under the disease progress curve.
AUDPCM quantifies the numerical interaction between disease proportion and time, enabling the assessment of disease severity over time. Thus, a lower AUDPCM value is indicative of greater resistance in an individual (Mendes et al. 2022). The ranking for AUDPCM was therefore organized in ascending order based on the values of predicted additive genetic effects and gains.
The additive genetic values ranged from − 103.73 to -74.28, with predicted gains ranging from − 7.85% to -2.98%. The new mean for the selected genotypes was below the overall mean, reflecting the nature of genetic gain estimates derived from the area under the disease progress curve, where negative values are expected for superior genotypes.
Plant 450, among the 30 selected genotypes, exhibited the highest resistance with an AUDPCM of 397.5. It was also distinguished for its superior genetic merit within the BC3-293 family, which was the family that demonstrated the most significant advancements.
Despite the quantitative inheritance of the resistance trait, employing the backcrossing technique proved effective for transferring resistance genes against CABMV. This success might be attributed to the extensive evaluation of numerous genotypes and the high heritability coefficients for the trait in question. For traits governed by multiple genes, it is advisable to assess a wide array of genotypes in each backcross generation and conduct a thorough examination of the disease-induced symptoms, enhancing the precision of selecting genotypes resistant to the virus (Preisigke et al., 2021).
Dissimilarity through UPGMA
Differences among the studied genotypes were estimated. As expected, Passiflora setacea emerged as the most distinct genotype within the population due to its species differentiation. Following the generation of the dissimilarity matrix, the genotypes were classified into four discrete clusters via the UPGMA approach (Fig. 1).
Descriptive captions: In this dendrogram, 4 distinct groups are noted, one group consisting only of Passiflora setacea, the second group only of genotype 1, the third group only of genotype 69 and the fourth group composed of all the remaining genotypes that were part of the analyzes
The cutoff point in the dendrogram was established using the criterion established by Mojena (1977), setting the threshold at 80% dissimilarity. This statistical method relies on the comparative magnitude of distance levels in the dendrogram, dispensing with prior group formation knowledge.
Passiflora setacea (93 S), with an AUDPCM of 405, was allocated to the first group, marked in red (Fig. 1), signifying its significant separation from the remaining genotypes. As expected, the most pronounced dissimilarity, recorded at 0.57, was between Passiflora setacea and Passiflora edulis.
Genotype 1, corresponding to plant 5 in the field and displaying an AUDPCM of 495, singularly constituted the second group, depicted in lilac. It showed a distance divergence from Passiflora setacea with a dissimilarity of 0.54.
Group III, illustrated in blue (Fig. 1), comprised solely of genotype 69 (plant 520 in the field) with an AUDPCM of 952.5. This genotype displayed a 0.50 dissimilarity from genotype 44 (plant 337 in the field), which had an AUDPCM of 498.75.
The fourth group, represented in black (Fig. 1), included the largest number of individuals with 92 genotypes, accounting for 96.8% of the total. This group included Passiflora edulis (94 E), which registered an AUDPCM of 948.75. The aggregation of such a large number of genotypes into a single group suggests that these individuals share the majority of alleles for the assessed loci. Individuals 9 (plant 40, AUDPCM = 435) and 17 (plant 108, AUDPCM = 543.75) exhibited the highest genetic affinity within this group, with a 0.98 similarity. The overall mean for AUDPCM of this group was 648.99, with values ranging from 397.5 (genotype 62) to 990 (genotype 48).
The genotypes identified as most resistant were predominantly from the BC3-293 family, with genotype 62 (plant 450, AUDPCM = 397.5) leading, followed by genotype 88 (plant 675, AUDPCM = 423.75). Genotype 62 displayed a 0.77 similarity to the commercial cultivar Passiflora edulis, marking it as a promising candidate for future backcross generations, especially given the aim of the study to identify resistant genotypes resembling the commercial variety in traits. Genotype 88 showed a 0.68 similarity to P. edulis. Additional genotypes from the top 30 selected based on their AUDPCM values include genotype 45 (plant 340, AUDPCM = 412.5), genotype 65 (plant 475, AUDPCM = 446.25), and genotype 66 (plant 476, AUDPCM = 438.75), with similarities to P. edulis of 0.74, 0.73, and 0.68, respectively.
Analysis of parental genome contribution and population genetic structure
An individual analysis of the contribution of the genome of the recurrent parent, cultivar 'UENF Rio Dourado' was carried out, considering the loci identical to those of the parent and the total allele count.
In the BC3 population, it is anticipated that 93.75% of the parental genome will be present in the progeny. However, the analysis revealed that the average genomic contribution from the recurrent parent 'UENF Rio Dourado' within the population was only 72.36%, which is lower than expected.
Considering all amplified alleles among the five evaluated families, BC3-153 showed the highest contribution of recurrent parents at 74.48%. Every genotype within this family had at least 68.52% of the genome of the recurrent parent, with individual percentages ranging from 68.52–81.48%. The BC3-293 family followed, with an average genomic recurrence of 72.93% and individual genotype contributions ranging from 61.11–81.48%.
The BC3-17 family had an average recurrence of 72.15%, with individual genotypes showing a range from 62.96–77.77%. For the BC3-355 family, the average was slightly lower at 71.66%, with a range from 51.85–85.18% per genotype. The BC3-501 family had the lowest average at 69.48%, with genotype contributions varying between 51.85% and 79.63%.
Additionally, a Bayesian analysis was employed, suggesting the formation of two distinct groups. This approach offers a more rigorous and less subjective method of classifying population structure into defined groups compared to hierarchical methods like UPGMA.
Based on the methodology of Evanno et al. (2005), the optimal delta K value was identified at K = 2, suggesting optimal structuring when the sample was segmented into two distinct groups, which were well-structured. A membership probability threshold of 60% was established for group classification. Therefore, Group I (Fig. 2, red) comprised the majority, with 67 genotypes (70.5%), including the commercial variety and the sole parent from the evaluated family. The AUDPCM for this group was 642.75, with values ranging from 397.5 (genotype 62) to 990 (genotype 48). Group II (Fig. 2, green) included 28 genotypes and encompassed the wild species, with an AUDPCM of 656.8, extending from 405 for P. setacea to 963.75 for genotype 82.
The formation of two groups indicates a unique set of alleles in the green group, distinguishing it from the red group. The emergence of few groups was anticipated due to the high genomic region overlap among the evaluated individuals, which is a reflection of the genetic structure of the population.
In the context of this study, identifying two distinct groups sufficed to inform the subsequent phases of the UENF passion fruit breeding program, given that the 95 assessed genotypes represent a well-defined genetic composition. Considering the different analytical approaches, the genotypes recommendable for future generations should be those most similar to the commercial cultivar, as well as those which exhibited the best means for resistance.