The synthetic allotetraploid BatSten1 PI 695418 {[Arachis batizocoi PI298639/K9484 x A. stenosperma PI666100/V10309]^{(2n = 4x = 40)}} (Bertioli et al., 2021a) was used to introgress the nematode resistance QTL from A. stenosperma into tetraploid peanut. An F₂ population was created by selfing an F₁ derived from a cross between A. hypogaea cv. Runner IAC-886 (herein called Runner-886) and BatSten1, and then used for QTL mapping (Ballén-Taborda et al., 2019). From this population, four superior F₂ lines (F₂-7, F₂-13, F₂-34, and F₂-73) were selected based on (1) better vigor, as visually more leaf biomass; (2) good agronomic traits; (3) late leaf spot (LLS) and PRKN resistance; (4) harboring QTL in A02 and A09 for nematode resistance per molecular genotyping. Six F₂-derived F₃ (F_2:3) homozygous progeny from F_2:3-7 and F_2:3-34, with validated resistance to PRKN (Ballén-Taborda et al., 2021), were used as initial donor parents in a backcrossing scheme (Figure 1, red boxes). For incorporation of resistance from A. stenosperma into peanut, we used three susceptible recurrent parents, which are TifGP-2, 5-646-10, and 13-1014. For pyramiding resistances from both wild species A. stenosperma and A. cardenasii, we used two resistant lines, which are 13-2113 and 13-1125. (1) TifGP-2 is a breeding line with good yield and grade, and normal oleic content (Holbrook et al., 2012); (2) 5-646-10 is derived from the cross Florida-07 x Tifguard, with good yield and grade, and high oleic/linoleic fatty acid ratio (Holbrook, CC, unpublished data); (3) 13-1014 is derived from [C1805-617-1 (Florida-07 x Tifguard) x GA-06G], with high oleic/linoleic fatty acid ratio (Holbrook, CC, unpublished data); (4) 13-2113 is derived from [C1805-2-9-16 (Florida-07 x Tifguard) x TifGP-2], a high oleic/linoleic fatty acid ratio, that was included in the second cycle only (Holbrook, CC, unpublished data); (5) 13-1125 breeding line was included in the fourth cycle only (Holbrook, CC, unpublished data) (Figure 1, blue boxes).

A marker-assisted backcrossing (MABC) approach was used to incorporate PRKN resistance from A. stenosperma into cultivated peanut. Four cycles of backcrossing were performed in two different locations: Athens, GA and Tifton, GA under greenhouse conditions. In the first, second, and third cycles, 16 SNP markers linked to the QTL in A02 and A09 (Leal-Bertioli et al., 2016) were used for foreground selection (Supplementary Table 1, gray-shaded markers). For the fourth cycle, 10 new markers were used to finely target these chromosome segments based on high-throughput genotyping of backcross (BC) lines (Ballén-Taborda et al., 2021) (Supplementary Table 2, gray-shaded markers). 60 additional SNPs markers were also developed and are available here for genotypic selection in breeding programs (Supplementary Tables 1, 2, not shaded markers). Progeny from each cycle that harbored segments associated with PRKN resistance were used as male parents for the next backcross cycle (Figure 1, orange boxes). BC₄F₁s were germinated for seed increase.

Kompetitive Allele-Specific PCR assays (KASP, LGC Biosearch Technologies, Hoddesdon, United Kingdom)¹ assays were used for the selection of PRKN resistance alleles. For KASP reactions, genomic DNA was isolated from a small section of the peanut cotyledon opposite to the embryo (seed chip, ∼50-100 mg). DNA was extracted using the DNeasy Plant Mini Kit (QIAGEN, Hilden, DE) according to the manufacturer’s instructions. Each KASP marker consisted of three primers per SNP position (two allele-specific and one common flanking primer). Primers were designed using the web-based program BatchPrimer3² (USDA-ARS, Albany, CA, United States) (You et al., 2008) with the “Allele-specific primers and allele flanking primers” option. Parameters used were 60–120 bp in fragment size, GC content of 30–80%, and Tm between 58 and 60°. KASP primer assay mix per SNP position consisted of 12 ul (100 uM) of each allele-specific primer, 30 ul (100 uM) of the flanking primer, and 46 ul of H₂O. Single KASP reactions (5 ul) consisted of 2.5 ul of KASP 2x Master Mix (Low Rox 5000 V4.0), 0.07 ul of KASP primer assay mix, 1.93 ul of water, and 0.5 ul of DNA (10 ng/ul). Two replicates per primer per sample were included in each reaction, as well as no-template controls (NTCs). A C100 touch Thermal Cycler (BIO-RAD, Hercules, CA, United States) was used with the following conditions: 94°C for 15 min; 8 cycles of 94°C for 20 s and touchdown starting at 61°C for 1 min (dropping 6° per cycle); 31 cycles of 94°C for 10 s and 55°C for 1 min; 9 cycles of 94°C for 20 s, and 57°C for 1 min; 4°C hold. Fluorescence was read with a LightCycler^® 480 Instrument II (Roche Life Science, Switzerland) and analyzed using the LightCycler^® 480 Software (v.1.5.1.62) (Roche Life Science, Switzerland). Finally, data were exported into Microsoft Excel for analysis.

Genomic DNA of 271 BC₃F₁ lines and controls (BatSten1, Runner-886, 5-646-10, 13-1014, TifGP-2, Tifguard, and Tifrunner) were extracted from lyophilized leaves using the DNeasy 96 Plant Kit (QIAGEN, Hilden, DE, United States) according to the manufacturer’s instructions. DNAs were genotyped with the Axiom_Arachis2 SNP array (Clevenger et al., 2018; Korani et al., 2019). Genotypic data was extracted and processed using the Axiom™ Analysis Suite software (v.4.0.3.3) (ThermoFisher Scientific, Waltham, MA, United States). Output was analyzed using custom shell scripts (see below) and resulting data were visualized as a color map in Microsoft Excel. The physical positions of the A-genome markers were determined according to the position of their orthologs in the A. duranensis pseudomolecules and the K-genome markers based on the A. ipaensis pseudomolecules (Bertioli et al., 2016).

The strategy to identify polymorphic SNP markers included two main steps. First, a set of polymorphic SNP markers between parental genotypes (BatSten1 ≠ Runner-886) was extracted. Original genotyping calls were replaced by numbers (“1” for BatSten, “2” for Runner-886, and “3” for a different genotype). Second, SNPs present in the genetic map previously identified (A. stenosperma-specific and A. batizocoi-specific markers) (Ballén-Taborda et al., 2019) were retrieved (Supplementary Script 1). Genotypic data were used to perform the principal component analysis (PCA) using the “dist” function in R.

BC₃F₃ segregating lines from five BC₃F₁ families and controls were evaluated for PRKN resistance to further validate the QTL in A02 and/or A09 (Table 1). The experiment included 72 BC₃F₃ lines, which were selected based on the genotypic information of the BC₃F₁ generation, by focusing on high cultivar genome recovery (89.1–95.9%) and with superior field performance (data not shown) of the BC₃F₂ generation during summer 2020. Selected lines included two BC₃F₃ lineages with the A02-QTL, two with the A09-QTL, one with both A02-QTL and A09-QTL, and one with a large A09-QTL. The synthetic allotetraploid BatSten1 and the cultivar A. hypogea TifNV-High O/L (Holbrook et al., 2017) were used as resistant controls and A. hypogea 5-646-10 and 13-1014 as susceptible controls. To confirm the presence of the QTL, BC₃F₃s and controls were genotyped using KASP markers (as described in the “Marker-assisted breeding” section). Six KASP markers targeting the bottom of A02 (A02-83,464,195, A02-92,077,207, and A02-92,983,792) and A09 (A09-16,516,448, A09-112309,231 and A09-114,515,959) (Supplementary Table 2) were used. Additionally, Axiom_Arachis2 SNP array genotyping was performed for genome-wide characterization, and data were filtered as described above.

Peanut root-knot nematode (PRKN) populations were cultured and extracted from eggplant (Solanum melongena) to be used as inoculum for bioassays. Second stage juveniles (J₂s) were collected from infected roots in a mist chamber every 2–3 days over a week and stored at 10°C until inoculation. Peanut seeds were grown in nursery pots filled with steam-sterilized sandy soil in the greenhouse. Bioassay was performed under greenhouse conditions, in a randomized complete block design with 12 replicates per genotype. Furthermore, 40-day-old plants were inoculated with 6,000 J₂s by adding the inoculum in two 2-cm deep holes at the base of the plant. Two months later, plants were uprooted, rinsed to remove soil, assessed for galling, and weighed. Eggs were extracted from roots using 0.5% NaOCl and counted (Hussey and Barker, 1973; Holbrook et al., 2003). Two different traits were measured: (1) galling index (GI), where 0 = no galls, 1 = 1-2 galls, 2 = 3-10 galls, 3 = 11-30 galls, 4 = 31-100 galls and 5 = more than 100 galls (Taylor and Sasser, 1978) and (2) number of eggs. Galling index and number of eggs per root weight (GI/g and eggs/g) were used for resistance assessment. A highly resistant plant was defined as such when the reproduction of nematodes was less than 20% of the reproduction in a susceptible plant (Taylor and Sasser, 1978).

Association between genome-wide genotypic data of BC₃F₁ lines and phenotypic data of seed weight, length and width, leaf spot incidence, fertility, architecture, and flower color were used to identify candidate wild introgressions that could be controlling these traits. For seed size, leaf spot incidence, fertility, and architecture a Pearson correlation was performed. Flower color was analyzed using a mixed linear model (MLM) in Tassel (v.5) (Ithaca, NY, United States) (Bradbury et al., 2007). A Manhattan plot was created in R using the “qqman” package (Turner, 2014) and thresholds calculated using the “CalcThreshold” package with the Bonferroni method (Hamazaki and Iwata, 2020).

Phenotypic data for seed weight, length and width, pollen viability, and nematode resistance bioassay were analyzed using the package R. A Shapiro-Wilk test was used to test for normal distribution. Non-parametric Kruskal-Wallis one-way analysis of variance (Kruskal and Wallis, 1952) was used to evaluate differences at a 5% level of significance (P < 0.05). For the seed weight, length, and width (transformed to Log10 when needed) the Welch t-test was used to perform pair-wise comparisons between wild accessions, cultivated genotypes, and backcross generations. Additionally, the non-parametric Skillings-Mack test (Chatfield and Mander, 2009) was used to evaluate significant differences for the RCBD nematode bioassay (P < 0.05). Further analysis included the Wilcoxon signed-rank test for pairwise comparisons using false discovery rate (FDR) correction to group samples by significant similarity (P < 0.05).

Four generations of MABC for introgression of PRKN resistance from A. stenosperma were completed in two locations, Athens, GA, and Tifton, GA under greenhouse conditions (Figure 1). KASP genotyping was performed using 16 (for first, second and third cycles) and 10 SNP markers (fourth cycle) (Supplementary Tables 1, 2, gray-shaded markers). For the first cycle of backcrosses, 38 cross combinations were used, with 19 F_2:3 plants as donor male parents and two cultivated peanut female parents, namely, TifGP-2 and 5-646-10. In this cycle, 1,008 potential BC₁F₁ seeds were harvested, and based on KASP genotyping, 17 were selected for harboring the nematode resistance segments in A02 and A09 (Supplementary Table 3).

For the second cycle, 14 cross combinations were made with four cultivated peanut female parents, TifGP-2, 5-646-10, 13-1014 and 13-2113, and 11 BC₁F₁ male parents. Here, 61 potential BC₂F₁ seeds were obtained and genotyped with KASP markers. A total of 21 were selected as they harbored resistance segments in A02 and A09 PRKN resistance QTL (Supplementary Table 4).

For the third cycle of backcrossing, a total of 21 cross combinations were made using 5-646-10 and 13-1014 as female parents and 21 BC₂F₁ male parents selected the previous year, six of which were shown to be resistant to PRKN (Ballén-Taborda et al., 2021). This resulted in 397 potential BC₃F₁ seeds (Supplementary Table 5), that were genotypically characterized in two different ways. First, 81 BC₃F₁ seeds were randomly selected and evaluated for the presence of resistance segments using the KASP assays. From this group, 52 BC₃F₁s harbored the segments in A02 and A09 from A. stenosperma (Supplementary Table 5). Second, 271 BC₃F₁s were genome-wide genotyped (see next section).

For the fourth cycle, six cross combinations were used with 5-646-10, 13-1014, and 13-1125 as females and two BC₃F₂ male parents harboring A02 and A09 loci. Here, 27 potential BC₄F₁ seeds were obtained and genotyped with KASP markers. A total of 25 were confirmed to harbor PRKN resistance chromosome segments at the bottom of both A02 and A09 (Supplementary Table 6). Three of them could have combined both sources of resistance in A02 from A. stenosperma and A09 from A. cardenasii present in parent 13-1125.

To characterize the wild introgressions in the BC₃F₁ population, 271 lines and controls were genotyped with the Axiom_Arachis2 SNP array. A total of 930 informative polymorphic SNP markers, previously identified and assigned to A and B subgenomes (Ballén-Taborda et al., 2019), were recovered. Among these, 527 markers were located in A-subgenome (A. stenosperma-specific markers) and 403 to B/K-subgenome (A. batizocoi-specific markers). Of the 271 genotyped lines, 253 (93.4%) were true progeny from hybridization and 18 (6.6%) were products of self-pollination (Supplementary Table 7).

Examples of Axiom clustering plots of SNPs linked to QTL in A02 (Supplementary Figures 1A,B) and A09 (Supplementary Figures 1C–E) show the distribution of BC₃F₁ lines and controls. Red clusters (A02 SNPs) and blue clusters (A09 SNPs) comprise genotypes without the A. stenosperma-derived allele. A. stenosperma was genotyped with the alternate allele. Yellow clusters indicate the backcross lines with incorporated PRKN resistance from A. stenosperma.

The principal component analysis (PCA) performed on the genotyping data, allowed us to observe that the BC₃F₁s have recovered a high percentage of the A. hypogaea genome. Backcross lines showed a recurrent parent genome recovery between 80.2 and 98.8%, while still carrying between 1.1 and 19.1% of the wild donor genome (Supplementary Figure 2). Additionally, to visualize the distribution of the backcross population according to the proportion of wild introgression (%) in each chromosome, the data was displayed in violin plots. These plots allowed us to observe that the lines were harboring more wild alleles in the A-subgenome than in B-subgenome, especially in chromosomes A02 and A09 where foreground selection was applied during the backcrossing process (Figure 3 and Supplementary Table 7).

To validate the PRKN resistance controlled by QTL in A02 and/or A09, 72 BC₃F₃ segregating lines, and resistant and susceptible controls were evaluated using a greenhouse pot nematode bioassay and genotyped with both KASP and Affymetrix to confirm the presence of the QTL. Specifically, Affymetrix was completed for genome-wide characterization. Galling index (GI) and number of eggs in relation to root weight (GI/g and eggs/g) allowed us to assess the resistance to M. arenaria within the backcross lines (Table 1).

Resistant controls (BatSten1 and TifNV-High O/L) and susceptible genotypes (5-646-10 and 13-1014) exhibited the expected phenotype. BatSten1 and TifNV-High O/L showed strong resistance, with no or low gall/egg production. In contrast, for 5-646-10 and 13-1014, GI/g fluctuated between 0.31 ± 0.26 and 0.42 ± 0.11 and eggs/g varied between 579.18 ± 855.26 and 1268.64 ± 1046.49 (Figure 4, Table 1 and Supplementary Table 8). BC₃F₃ lines showed significant differences for GI/g and eggs/g (Kruskal-Wallis, Skillings-Mack and Wilcoxon tests, P < 0.05). Since the BC₃F₃s were still recombining and segregating for PRKN QTL, to better summarize the results, the lines were grouped according to the segments they were carrying as follows: • Group 1: bottom A02 (A02) (A. stenosperma allele at A02-83,464,195, A02-92,077,207 and A02-92,983,792 → 81.0 – 93.8 Mb); • Group 2: bottom A09 (A09) (A. stenosperma allele at A09-112,309,231 and A09-114,515,959 → 104.6 – 119.8 Mb); • Group 3: large A09 (A09+) (A. stenosperma allele at A09-16,516,448, A09-112,309,231 and A09-114,515,959 → 3.4 – 118.7 Mb); and • Group 4: Both bottom small A02 (A. stenosperma allele at A02-92,077,207 and A02-92,983,792 → 91.6 – 93.8 Mb) and bottom A09 (A02- and A09). According to this grouping, all the backcross materials belonging to groups 1, 3, and 4 exhibited high levels of resistance to PRKN. No or few galls (0.01 ± 0.03–0.17 ± 0.20) and low egg production (0.00 ± 0.00–36.30 ± 83.17) was observed in the infected roots. For groups 3 and 4 galls were observed in the roots, but the production of eggs was inhibited. In contrast, lines in group 2 were susceptible to PRKN by having GI/g and eggs/g values of 0.26 ± 0.19 and 1050.62 ± 1005.38, respectively (Figure 4). For more details of GI/g and eggs/g values for the BC₃F₃ lines, see Table 1 and Supplementary Table 8.

A wide variation of morphological and agronomic traits was observed in the backcross populations (BC₃F₁s and BC₃F₂s), including seed size, pollen viability, leaf spot incidence, fertility (number of pegs), plant architecture, flower color, branching, and extra leaves (Figure 2). Association analysis between phenotypic information and wild introgressions in the BC₃F₁ population was performed to identify candidate regions associated with these traits.

In this study, resistance to PRKN was successfully validated in a set of BC₃F₃ lines. The bottom of A02 (81.0 – 93.8 Mb) (Figure 4, group 1) provided strong resistance as previously described and validated (Leal-Bertioli et al., 2016; Ballén-Taborda et al., 2019, 2021). In the case of the QTL in chromosome A09, we observed that the small introgression at the bottom (104.6 – 119.8 Mb) was insufficient to stop nematode development (group 2) and that a larger segment at the top-bottom of A09 (3.4 – 118.7 Mb) was required to provide resistance, especially for preventing eggs production (group 3). It is possible that the presence of wild alleles in A04 associated with susceptibility in group 2 could be acting against resistance as previously described (Ballén-Taborda et al., 2019). When the plants were harboring both bottom small A02 (91.6 – 93.8 Mb) and bottom A09 (104.6 – 119.8 Mb) (group 4) we would expect to observe inhibition of both galls and egg production since A02 was present; however, galls were present in the roots. Field testing is in progress to test the stability of the resistance and for allele fixation through selfing.

Genetic maps, quantitative trait loci (QTL), and marker-phenotype associations have been reported for numerous crops and traits (Collard and Mackill, 2008). Despite this, examples of QTL incorporation in plant breeding programs are lower than expected (Bernardo, 2016). This work represents a successful example of QTL introgression from a wild relative into an elite peanut despite the genetic incompatibilities. This provides an alternative to the only source of root-knot resistance currently deployed in the peanut crop, derived from A. cardenasii (Simpson and Starr, 2001).

The population of advanced peanut backcross lines that we have developed during this work has wild chromosome segments through much of the genome, distributed in different ways in different lines. They are being tested and advanced in several locations, and the best performing lines are being selected for germplasm release. The PRKN resistance alleles have been successfully validated and DNA markers are now available to facilitate the marker-assisted selection. Furthermore, because of the diverse wild chromosome segments in this population, we also anticipate that it has other disease resistances and traits of value to the peanut crop. We anticipate that, over time, these backcrossed lines will impact peanut production by delivering several new traits to the peanut crop, similar to the case of North Carolina peanut lines with A. cardenasii segments that have provided resistance to late leaf spot, rust, and web blotch in numerous countries around the world (Bertioli et al., 2021b).