Genetic difference of FOV4 resistance among the 11 founder parents
On a genotype basis, DI ranged from 16.7 to 100% in Test 1 and from 11.8 to 87.5% in Test 2, while DSR ranged from 0.2 to 3.3 in Test 1 and from 0.2 to 2.6 in Test 2, among the 11 founder parents. Significant genotypic variation in FOV4 resistance was detected (Table 2). Tests 1 and 2 showed relatively consistent results in FOV4 resistance among the 11 founder lines, as indicated by a significant positive correlation in DSR between the two tests (r = 0.715, P < 0.05). As shown in Table 2, when Deltapine Acala 90 was used as the susceptible check because of its highest average DSR (2.84) and DI among the 11 parents, Acala Ultima, M-240RNR, and Stoneville 474 were significantly lower in DSR in both tests and across the two tests (0.09–0.78), and they were therefore considered as the resistant parent group. In addition, FiberMax 966, Paymaster HS 26, and SureGrow 747 had significantly lower DSR in Test 1 and also on average across the two tests (0.64–1.26). On the other hand, the other four parental lines (Coker 315, Phytogen PSC 355, Stoneville 825, and Tamcot Pyramid) were equally susceptible to FOV4 as the susceptible check in both tests and on average (1.74–2.59), and therefore, these four genotypes were considered as the susceptible parent group.
Table 2
Fusarium wilt (Fusarium oxysporum f. sp. vasinfectum) race 4 resistance of 11 founder Upland cotton lines (Gossypium hirsutum) used in the development of the MAGIC population, evaluated in the greenhouse, Las Cruces, NM, USA, 2018.
Parent | DSR | | Incidence (%) |
| Test 1 | Test 2 | Mean | Test 1 | Test 2 | Mean |
Acala Ultima | 0.50* | 0.20* | 0.35* | 50.00 | 20.00* | 35.00* |
Coker 315 | 2.30 | 2.19 | 2.25 | 92.86 | 66.67 | 79.76 |
Deltpine Acala 90 (CK) | 3.30 | 2.38 | 2.84 | 100.00 | 81.25 | 90.63 |
FberMax 966 | 0.17* | 1.11 | 0.64* | 16.67* | 39.29* | 27.98* |
M-240RNR | 1.28* | 0.29* | 0.78* | 90.00 | 14.29* | 52.14* |
Paymaster HS 26 | 0.63* | 1.90 | 1.26* | 46.43 | 65.97 | 56.20 |
Phytogen PSC 355 | 2.50 | 1.29 | 1.90 | 100.00 | 50.79 | 75.40 |
Stoneville 474 | 0.00* | 0.18* | 0.09* | 0.00* | 11.81* | 5.90* |
Stoneville 825 | 2.53 | 2.64 | 2.59 | 100.00 | 87.50 | 93.75 |
SureGrow 747 | 0.83* | 0.94 | 0.89* | 75.00 | 37.50* | 56.25 |
Tamcot Pyramid | 2.10 | 1.38 | 1.74 | 80.00 | 49.21 | 64.60 |
LSD (0.05) | 1.76 | 1.95 | 1.36 | 69.47 | 43.03 | 37.46 |
* Significantly different from the susceptible check parent-Deltapine Acala 90. DSR, disease severity ratings. |
These 11 cultivars were developed under non-FOV4 infested field conditions in a span of 24 years between 1979 and 2002. Based on a pedigree analysis (Table 1), seven parents can trace their parentages to Deltapine (DP) 15 and/or Stoneville (ST) 7, while three parental lines had Acala or New Mexico germplasm in their parentages. However, Tamcot Pyramid was developed in the MAR program and its parentage could not be traced to any specific germplasm lines. It appears that the genetic difference in FOV4 resistance detected in this study was not due to the inclusion of the backbone germplasm lines (DP 15 and ST 7) in their pedigrees. For example, the three more resistant lines- M-240RNR, ST 474, and SureGrow 747 and the four more susceptible lines- Coker 315, DP Acala 90, Phytogen PSC 355, and ST 825, all had DP 15 and/or ST 7 in their pedigrees. The more resistant Acala Ultima and Paymaster HS 26 and the more susceptible DP Acala 90 and FiberMax 966 all had Acala or New Mexico germplasm in their parentages. Tamcot Pyramid did not show any FOV4 resistance, although it was developed in the MAR program for enhancement of cotton resistance to VW, FW and RKN. Therefore, the genetic variation in FOV4 resistance within the 11 founder parents may be the consequence of random recombination and fixation of resistance genes and alleles in breeding under non-selective conditions for FOV4 resistance. In this study, the existence of the quantitative genetic variation in FOV4 resistance among the 11 founder parents allowed the following analysis of the genetic and genomic basis of FOV4 resistance in the MAGIC population derived from them.
Analysis of variance and broad-sense heritability estimates for FOV4 resistance in the MAGIC population
As shown in Table 3, a combined ANOVA for the entire MAGIC population indicates that DI and DSR were significant (at P < 0.001) in test, genotype, and also genotype × test interaction. Therefore, a further analysis was performed on an individual test basis. In both Tests 1 and 2, DI and DSR showed a significant genotypic variation (at P < 0.001, except for DI in Test 1 at P < 0.05). The two tests were grown in the same greenhouse during different time periods with two different inoculum preparations and inoculations, in addition to other uncontrollable factors in greenhouse and crop management. These factors may have contributed to genotype × test interactions, as seen from different results between the two tests for some parental genotypes (Table 2).
Table 3
Analysis of variance (mean square) of Fusarium wilt (Fusarium oxysporum f. sp. vasinfectum) race 4 resistance in the MAGIC population of 550 recombinant inbred lines (Gossypium hirsutum) tested two replicated tests in the greenhouse, Las Cruces, NM, USA.
Source | DSR | Incidence (%) |
Combined | | |
Test | 46.51*** | 404820.86*** |
Genotype | 1.35*** | 1187.87*** |
Genotype × Test | 1.28*** | 1073.11*** |
Error | 0.88 | 793.19 |
h2 | 0.605 | 0.599 |
Test 1 | | |
Replication | 9.33** | 907.67** |
Genotype | 0.99*** | 2712.70* |
Error | 0.66 | 659.38 |
h2 | 0.600 | 0.804 |
Test 2 | | |
Replication | 21.68*** | 2298.50 ns |
Genotype | 1.70*** | 1396.68*** |
Error | 0.99 | 893.20 |
h2 | 0.632 | 0.601 |
* P < 0.05; ** P < 0.01; *** P < 0.001; ns, not significant. DSR, disease severity ratings. |
Because DSR takes into consideration of incidence and leaf symptom severity including plant death, it was therefore further used to quantitatively measure the overall FOV4 resistance in this study. Broad-sense heritability estimates for DSR in the two tests were similar, ranging from 0.60 in Test 1 to 0.63 in Test 2, and the estimate was 0.62 in a combined ANOVA from the two tests (Table 3). The results indicate that approximately 60% of the phenotypic variation in FOV4 resistance was due to genetic variation in this MAGIC population. Therefore, a further detection of genetic factors for FOV4 resistance was warranted.
Genetic difference of FOV4 resistance in the MAGIC population of 550 RILs
In the MAGIC population, the DI ranged from 0 to 100% in both tests with an average of 89.4 and 83.6% in Tests 1 and 2, respectively; and the DSR ranged from 0 to 4.2 with an average of 1.8 in Test 1 and from 0 to 4.9 with an average of 1.5 in Test 2. An average of only 1.0 and 5.0% plant mortality was observed in Tests 1 and 2, respectively. Averaging across the two tests, many RILs had significantly and consistently lower DSR than the susceptible check (DP Acala 90) in both tests, and Table 4 shows 22 of them. As a comparison, the table also lists 23 most susceptible lines. It appears that the resistance level of the most resistant RILs was within the level of the most resistant parents (Table 2), while the most susceptible RILs were more susceptible than the most susceptible parent (DP Acala 90). The above results suggest that transgressive segregation toward susceptibility occurred during the intermating process among the progeny from the hybrids of the 11 founders. However, because responses of some of the RILs are likely different to FOV4 infections under low temperature conditions, RILs with significantly higher FOV4 resistance than the parents may be identified.
Table 4
Most resistant and most susceptible lines (Gossypium hirsutum) in responses to Fusarium wilt (Fusarium oxysporum f. sp. vasinfectum) race 4 infections in the MAGIC population based on disease severity ratings (DSR).
Line | Test 1 | Test 2 | Mean | | Line | Test 1 | Test 2 | Mean |
Most resistant lines | | | | Most susceptible lines | |
MG150 | 1.00 | 0.00 | 0.50 | | MG64 | 2.25 | 4.50 | 3.38 |
MG202 | 1.00 | 0.17 | 0.59 | | MG32 | 2.24 | 3.46 | 2.85 |
MG207 | 0.17 | 0.89 | 0.53 | | MG427 | 2.00 | 3.42 | 2.71 |
MG217 | 0.75 | 0.31 | 0.53 | | MG72 | 2.05 | 3.75 | 2.90 |
MG22 | 0.88 | 0.93 | 0.91 | | MG326 | 2.03 | 3.55 | 2.79 |
MG236 | 0.75 | 0.91 | 0.83 | | MG444 | 3.67 | 1.94 | 2.81 |
MG272 | 0.50 | 0.25 | 0.38 | | MG320 | 4.00 | 2.75 | 3.38 |
MG274 | 0.60 | 0.19 | 0.40 | | MG421 | 2.25 | 2.36 | 2.36 |
MG283 | 0.43 | 0.76 | 0.60 | | MG388 | 3.25 | 2.93 | 3.09 |
MG286 | 0.83 | 0.93 | 0.88 | | MG108 | 3.58 | 1.93 | 2.76 |
MG313 | 0.67 | 0.79 | 0.73 | | MG305 | 1.92 | 2.53 | 2.23 |
MG374 | 0.80 | 0.37 | 0.59 | | MG516 | 3.08 | 2.60 | 2.84 |
MG398 | 0.33 | 0.69 | 0.51 | | MG399 | 3.25 | 3.36 | 3.31 |
MG407 | 0.34 | 0.67 | 0.51 | | MG513 | 1.75 | 3.25 | 2.50 |
MG416 | 0.20 | 0.68 | 0.44 | | MG435 | 2.67 | 3.29 | 2.98 |
MG419 | 0.29 | 0.67 | 0.48 | | MG288 | 2.25 | 3.81 | 3.03 |
MG425 | 0.25 | 0.69 | 0.47 | | MG465 | 1.92 | 3.96 | 2.94 |
MG426 | 0.17 | 0.00 | 0.09 | | MG67 | 2.70 | 4.17 | 3.44 |
MG498 | 0.00 | 0.21 | 0.11 | | MG390 | 3.00 | 4.13 | 3.57 |
MG528 | 1.00 | 0.20 | 0.60 | | MG123 | 2.19 | 3.42 | 2.81 |
MG547 | 0.20 | 0.42 | 0.31 | | MG31 | 3.25 | 2.20 | 2.73 |
MG60 | 0.14 | 0.96 | 0.55 | | MG394 | 3.54 | 2.07 | 2.81 |
Acala 90 (S) | 3.30 | 2.38 | 2.84 | | MG101 | 2.33 | 3.05 | 2.69 |
| | | | | LSD (0.05) | 1.76 | 1.95 | 1.36 |
Qtls Mapped For Fov4 Resistance In The Magic Population
Mean DSRs for each MAGIC line from Tests 1 and 2 and the average from the two tests were used for GWAS based on 473,516 polymorphic SNP markers in this study. Based on means across the two tests, 8 QTLs including 2 (one each on A10 and A12) on the A subgenome and 6 QTLs (5 on D02 and 1 on D04) on the D subgenome were detected (Table 5, Fig. 2). In Test 1, 10 QTLs were detected, including 3 QTLs with one each on A06, A10 and A13 of the A subgenome and 7 QTLs on the D-subgenome (5 on D02 and one each on D03 and D11). Notably, qFOV4-D02-3a detected between the 0.7 and 1.2 Mb region represented a major QTL with the highest delta(SNP-index) value and LOD score (Table 5, Fig. 2). In Test 2, 13 QTLs were detected, including 4 QTLs with one each on A05, A08, A12, and A13 of the A-subgenome and 9 QTLs with 6 on D02 and one each on D01, D11, and D13 of the D-subgenome. Three common QTLs were detected between Test 1 and the overall means across the two tests, while 3 QTLs were common between Test 2 and the overall means across the two tests. Therefore, a total of 6 common QTLs were detected between an individual test and the overall means across the two tests. Taken together, 23 QTLs (7 on the A-subgenome and 16 on the D- subgenome) for FOV4 resistance were detected on 12 of the 26 pairs of Upland cotton chromosomes.
Table 5
QTL detected for Fusarium wilt (Fusarium oxysporum f. sp. vasinfectum) race 4 resistance based on severity disease ratings in the MAGIC population of 550 recombinant inbred lines (Gossypium hirsutum), evaluated in two replicated tests with artificial inoculations in the greenhouse, Las Cruces, NM, USA.
QTL locus | Test | Chr | Start (bp) | End (bp) | Peak (bp) | Mean SNP Index | -Log10 (p) | Allelic Effect |
qFOV4-A05-1 | T2 | A05 | 90,432,574 | 90,753,953 | 90,552,418 | 0.25 | 4.14 | 0.21 |
qFOV4-A06-1 | T1 | A06 | 1,679,590 | 2,204,558 | 2,204,558 | 0.08 | 3.51 | 0.18 |
qFOV4-A08-1 | T2 | A08 | 90,521,160 | 90,636,368 | 90,600,982 | 0.01 | 3.45 | -0.15 |
qFOV4-A10-1a | T1 | A10 | 95,558,764 | 95,606,868 | 95,580,590 | 0.16 | 3.58 | 0.13 |
qFOV4-A10-1b | Mean | A10 | 95,606,868 | 96,054,240 | 95,635,324 | 0.16 | 4.81 | -0.16 |
qFOV4-A12-1a | T2 | A12 | 84,239,091 | 84,422,787 | 84,420,565 | 0.15 | 5.34 | -0.19 |
qFOV4-A12-1b | Mean | A12 | 84,260,095 | 84,420,565 | 84,286,590 | 0.15 | 3.78 | -0.12 |
qFOV4-A12-2 | T1 | A12 | 86,835,123 | 87,046,181 | 87,032,949 | 0.15 | 3.42 | 0.10 |
qFOV4-A13-1 | T2 | A13 | 59,513,985 | 59,961,534 | 59,745,124 | 0.11 | 3.29 | -0.21 |
qFOV4-D01-1 | T2 | D01 | 54,347,064 | 54,684,752 | 54,663,115 | 0.17 | 4.17 | 0.17 |
qFOV4-D02-1 | T1 | D02 | 247,093 | 409,056 | 409,056 | 0.11 | 5.79 | 0.18 |
qFOV4-D02-2 | T1 | D02 | 508,090 | 681,893 | 539,793 | 0.15 | 6.32 | -0.21 |
qFOV4-D02-3a | T1 | D02 | 714,520 | 1,249,921 | 933,851 | 0.35 | 9.64 | 0.28 |
qFOV4-D02-3b | Mean | D02 | 731,052 | 1,344,425 | 979,730 | 0.33 | 4.12 | 0.13 |
qFOV4-D02-4a | T1 | D02 | 1,303,697 | 1,526,555 | 1,303,697 | 0.33 | 6.69 | 0.23 |
qFOV4-D02-4b | Mean | D02 | 1,386,109 | 1,478,616 | 1,392,442 | 0.33 | 3.71 | -0.11 |
qFOV4-D02-5a | T1 | D02 | 1,707,229 | 2,336,933 | 2,097,862 | 0.22 | 6.72 | 0.20 |
qFOV4-D02-5b | Mean | D02 | 1,787,329 | 2,143,747 | 1,804,734 | 0.19 | 3.89 | -0.11 |
qFOV4-D02-6a | T2 | D02 | 18,042,939 | 18,725,840 | 18,251,131 | 0.15 | 3.60 | -0.18 |
qFOV4-D02-7a | T2 | D02 | 18,799,231 | 19,572,947 | 19,572,947 | 0.15 | 4.32 | 0.20 |
qFOV4-D02-7b | Mean | D02 | 19,319,469 | 19,919,667 | 19,458,775 | 0.15 | 3.55 | 0.12 |
qFOV4-D02-7c | T2 | D02 | 19,802,857 | 20,438,501 | 19,908,316 | 0.13 | 3.87 | 0.19 |
qFOV4-D02-8a | T2 | D02 | 21,472,314 | 21,995,058 | 21,682,186 | 0.19 | 5.35 | -0.25 |
qFOV4-D02-8b | Mean | D02 | 21,541,506 | 22,199,752 | 21,541,506 | 0.17 | 4.01 | -0.15 |
qFOV4-D02-8c | T2 | D02 | 22,119,747 | 22,794,370 | 22,697,227 | 0.19 | 4.07 | -0.22 |
qFOV4-D02-9a | T2 | D02 | 22,843,353 | 23,250,735 | 22,988,244 | 0.19 | 3.65 | 0.20 |
qFOV4-D03-1 | T1 | D03 | 2,857,159 | 2,894,575 | 2,866,837 | 0.15 | 3.93 | 0.19 |
qFOV4-D04-1 | Mean | D04 | 1,197,145 | 1,248,176 | 1,219,821 | 0.15 | 4.92 | -0.14 |
qFOV4-D11-1 | T2 | D11 | 16,201,083 | 16,568,089 | 16,237,920 | 0.01 | 5.32 | 0.40 |
qFOV4-D11-2 | T1 | D11 | 52,964,907 | 53,368,876 | 53,302,570 | 0.09 | 3.71 | 0.19 |
qFOV4-D13-1 | T2 | D13 | 45,707,408 | 46,211,141 | 45,957,468 | 0.41 | 3.49 | -0.15 |
Two QTLs were detected on each of A12 and D11. However, the common QTL (qFOV4-A12-1) on A12 between Test 2 and the overall means across the two tests was not in proximity (9 Mb apart) to the QTL (qFOV4-A12-2) detected in Test 1. The 2 QTLs on D11 detected from Tests 1 and 2 were also different, as they were 37 Mb apart. Interestingly, two QTL clusters with a total of 9 QTLs were declared on D02. The two QTLs (qFOV4-D02-3 and qFOV4-D02-4) were overlapped within a 1.4 Mb interval (at the chromosome position between 0.4–1.8 Mb), which were only less than 35 kb away from qFOV4-D02-2 on the left and 400 kb apart from qFOV4-D02-5 on the right. Therefore, these 5 QTLs were clustered. The other 4 QTLs were 5 Mb apart (within the chromosome position 18–23 Mb) and were considered another QTL cluster on the same chromosome.
A meta-analysis of QTL reported for FW resistance in cotton
A summary of all reported FW resistance QTLs (excluding the ones detected in this current study) on chromosomes is shown in Table 6 and the detailed information from each study is listed in Supplementary Table 1. Of 101 QTLs reported, the D subgenome has significant more (by 59%) FW resistance QTLs than the A subgenome (62 vs. 39; χ2 = 5.24; χ2 = 3.84, P < 0.05).
Table 6
Distribution of QTL over the 26 tetraploid cotton chromosomes for resistance to Fusarium wilt (Fusarium oxysporum f. sp. vasinfectum) race 4.
Chromosome | Zhang et al. (2015) | Abdelraheem et al. (2017) | This study, meta-analysis |
c1/A01 | 0 | 0 | 3 |
c2/A02 | 0 | 1 | 3 |
c3/A03 | 1 | 1 | 6 |
c4/A04 | 0 | 0 | 0 |
c5/A05 | 0 | 0 | 0 |
c6/A06 | 3 | 6 | 8 |
c7/A07 | 1 | 1 | 2 |
c8/A08 | 2 | 4 | 8 |
c9/A09 | 0 | 0 | 2 |
c10/A10 | 0 | 0 | 1 |
c11/A11 | 2 | 4 | 0 |
c12/A12 | 1 | 2 | 6 |
c13/A13 | 0 | 0 | 0 |
A-subgenome | 10 | 19 | 39 |
c15/D01 | 1 | 1 | 6 |
c14/D02 | 1 | 2 | 4 |
c17/D03 | 4 | 5 | 10 |
c22/D04 | 2 | 3 | 3 |
c19/D05 | 4 | 8 | 8 |
c25/D06 | 1 | 2 | 3 |
c16/D07 | 1 | 1 | 6 |
c24/D08 | 1 | 2 | 4 |
c23/D09 | 1 | 0 | 3 |
c20/D10 | 0 | 0 | 0 |
c21/D11 | 0 | 0 | 4 |
c26/D12 | 0 | 0 | 6 |
c18/D13 | 1 | 4 | 5 |
D-subgenome | 17 | 28 | 62 |
Total | 27 | 47 | 101 |
On an individual chromosome basis, A04, A05, A11, and A13 had no FW resistance QTLs detected, and A07, A09, and A10 had only 1–2 QTLs reported, while A03, A06, A08, and A12 had the most FW resistance QTLs within the A subgenome. The resistance to FOV4 infections in the two A genome species- G. arboreum and G. herbaceum is inherited as a Mendelian trait (Zhang et al. 2015a); however, thus far, no major gene or QTL for FOV4 resistance has been reported on the A subgenome. Du et al. (2018) reported a main-effect QTL on chromosome c11 (A11) for FOV7 resistance based on GWAS using 215 Chinese G. arboreum accessions. However, it is currently unclear whether this QTL confers resistance to FOV4. On the D subgenome, only D10 did not have reported QTLs for FW resistance, and all other chromosomes carried 3 or more FW resistance QTLs including D01, D03, D05, D07, D12, and D13 with 5 or more QTLs detected. Therefore, overall, the D subgenome harbors more QTLs for resistance to FW, and the FW resistance QTLs were not evenly distributed on the tetraploid cotton chromosomes. Furthermore, the chromosomes with more QTLs detected are not necessarily the ones possessing a major gene/QTL for FW resistance including FOV1 on D07 (c16), FOV4 on D02 (c14), FwR on D03 (c17), and an unnamed gene on D11 (c21).
On a homeologous chromosome pair basis, the two pairs (A03/D03 and A12/D12) had more QTLs detected, while A10/D10 had almost no QTL reported among the 13 pairs. A06 and A08 had 2-2.7 times more QTLs detected than their homeologous counterparts (D06 and D08), respectively. However, D01, D03, D04, D05, D07, D11, and D13 carried 1.7 times or more QTLs than their homeologous A subgenome counterparts, respectively. Overall, there was no significant positive correlation (r = 0.153, P > 0.05) in the number of FW resistance QTLs between the 13 pairs of homeologous chromosomes, suggesting different mechanisms evolved from the two subgenomes to fight FOV infections in tetraploid cotton.
Identification Of Candidate Genes For Fw Resistance Qtl
In this study, a total of 532 genes were identified within the 23 QTL regions declared, of which 198 genes were found to have non-synonymous SNPs among the 11 parents. Based on the concept that these parental SNP should be reflected in RILs with different responses to FOV4, SNP indexes were further calculated between two groups (resistant vs. susceptible) of the selected RILs, as shown in Table 4. An average SNP index within each QTL region is shown in Table 5. Only these genes with a SNP index of 0.3 or higher were considered as candidates for the QTL. As a result, 48 genes met all the criteria, most of which were located on D02 and 15 genes were related to defenses to fungi (Supplementary Table 2). Interestingly, the major FOV4-resistance QTL locus qFOV4-D02-3 (with peak LOD scores greater than the FDR of 1.056 × 10− 7 at the 0.92–0.98 Mb region) contained 14 candidate genes, three of which (Gh_D02G0105, Gh_D02G0115, and Gh_D02G0119) responded to fungal infections (Supplementary Table 2). More studies are needed to narrow candidate genes for the QTL.