In11-unresponsive accessions from Argentina are genetically related

We previously identified two In11-unresponsive Phaseolus vulgaris accessions from USDA-GRIN germplasm by quantifying ethylene production following In11 treatment in 91 varieties [36]. Elsewhere, we described characterization of a null allele termed inr-1, identified in an In11-unresponsive accession from Honduras (W6 13807) [36]. A separate variety from Jujuy, Argentina (W6 17491) similarly showed a lack of In11-induced ethylene when applied to scratch-wounded leaves (w + In11) relative to wounding alone (w + H₂O) (Fig 1A). To further investigate this pattern, we screened 20 additional USDA-GRIN germplasm accessions from the same region of northwestern Argentina and discovered a broader subset of related landraces that were also In11-unresponsive (8 of 21 tested varieties) (Fig 1A & B, in S1 File.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. In11-unresponsive accessions from Argentina are genetically related.

A) Ethylene production in Argentinian common bean accessions after scratch-wounding and treatment with H₂O or 1 μM In11 peptide. Boxes show the median, 1st and 3rd quartiles and whiskers extending to the minimum and maximum values of replicate leaflets for individual In11 treatments (n = 2 or 3). Means were compared to H₂O and were statistically significant for responsive lines (Student’s t-test, p < 0.05) unless noted (n.s.), and not significant for all unresponsive lines. Responsive accessions (In11/H₂O ratio ≥1.9) are shown in teal, and unresponsive accessions (In11/H₂O ratio <1.9) are shown in red. Colors in bars represent treatment, w + H₂O (white) or w + In11 (green). B) Geographical distribution of Argentinian landraces and wild material using geotagged coordinates from USDA National Plant Germplasm System on a map of South America. Bubble size indicates number of accessions in a location. C) Seed coat variation is shown between select accessions. D) Hierarchical clustering of TASSEL (v 5.2.79) genetic distance values calculated from whole-genome single nucleotide polymorphisms (SNPs).

https://doi.org/10.1371/journal.pone.0343332.g001

To explore the potential genetic basis for In11 insensitivity among these 8 individuals, we performed low coverage genome resequencing of the 21 varieties (S2 Table in S1 File). Analysis of relatedness through single nucleotide variation indicated that unresponsive varieties were generally more related to each other than to other In-11 responsive individuals (Fig 1C & D).

Natural variation at the INR promoter correlates with differential expression

To identify the potential basis for In11 insensitivity, we crossed the insensitive line PI 638858 with the genome-sequenced reference accession G19833 (S1 Fig in S1 File). F₂ individuals showed segregation of In11-induced ethylene response consistent with a genetic basis at a single locus, and the phenotype was fully associated with a marker single-nucleotide polymorphism (SNP) at Chr07:7,408,876, within the INR coding sequence (Fig 2A).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Natural variation at the INR locus and correlation with INR expression and promoter strength.

A) In11 sensitivity corresponds with INR locus genotype in F₂ plants. Ethylene production after scratch-wounding and treatment with H₂O or 1 μM In11 peptide, with ratio greater than 1 indicating induced response. N = 11 F₂ plants were tested and displayed according to genotype at position 7,408,876, with plants homozygous or heterozygous for G19833 allele (G19833/-) or homozygous for PI 638858 allele (PI 638858). Boxplot shows the median, 1st and 3rd quartiles (box) and whiskers extending to the minimum and maximum values. Statistical significance between allele types was assessed by an unpaired two-tailed Student’s t-test, with p-values indicated above the brackets. Significance levels: *** p < 0.001; ** p < 0.01; * p < 0.05; ns, not significant. B) Relative expression of INR in In11 responsive accessions from PI 638852 and PI 638856 compared to non-responsive lines PI 661803 and PI 638858. Relative expression was normalized to PI 638858 expression level. C) Luciferase assay data showing the Z-score of relative luminescence units (RLU) of pINR from common bean accessions driving luciferase relative to PI 638858 pINR. Colors indicate biological replicates, plotted as technical replicate leaf discs (n = 324). Different letters represent significant differences (Welch’s ANOVA p < 0.05; Games-Howell post-hoc test, α = 0.05). D) Alignment of pINR sequences and Transcription Factor (TF) binding prediction of SNPs in TF binding sites upstream of the INR Translation Start Site (TSS) (0 bp = TSS). Only SNPs predicted to alter TF binding are shown: 1 – HD-Zip, 2 – Dof, 3 – MIKC-MADS, 4 – ERF and MYB, 5 - bZIP, bHLH, and Nin-like, 6 – MYB. TF binding was predicted with PlantRegMap parameterized with the Arabidopsis thaliana reference dataset [37]. Full analysis of SNPs, including those with no predicted TF binding site, is shown in S4A Fig in S1 File.

https://doi.org/10.1371/journal.pone.0343332.g002

To analyze potential causal variation, we reanalyzed the INR locus (Chr07:7,407,895–7,413,204) across the 21 accessions from our WGS dataset and identified 64 SNPs at the INR locus. The majority of the SNPs were found in the 1.8kb promoter region (68.8%), while the rest were distributed in the coding sequence (CDS) of INR (S2A Fig in S1 File). We resolved six promoter haplotypes (P1-P6) spanning the −800 region upstream of INR. Lines responsive to In11 (13/21) predominantly possessed allele P5 with a mix of SNPs from P1, P2, and P3, while unresponsive lines (8/21) possessed mainly P2 and P4. P5 was noteworthy in that it was present in only one unresponsive individual, whereas it was widespread in the responsive individuals. Likewise, P4 was present in 5 of 8 unresponsive individuals and only one responsive individual. In summary, In11 insensitivity generally corresponded with one of two alleles (P2 and P4) based on observable SNPs in the INR promoter region.

In contrast, coding sequence variation did not correspond with In11 sensitivity. Three nonsynonymous CDS SNPs were observed, two isolated individuals in the leucine-rich repeat region, and one nearly fixed substitution in the C-terminus near the transmembrane domain (S2A Fig in S1 File) [38]. Despite one amino acid substitution in the island domain, a key structural component of an LRR-RLP which facilitates binding to coreceptor BAK1 [5,39], accession W6 17017 still strongly responded to In11 treatment (Fig 1A and S2A Fig in S1 File). Similarly, we detected no 3’ UTR polymorphisms and therefore ruled out coding sequence and 3’-UTR variation as potential bases for In11 insensitivity.

We hypothesized that low INR expression from certain promoter alleles may mediate In11 insensitivity. We tested INR expression in six selected varieties. INR expression was 3.1 to 11.1-fold higher in PI 638852, W6 17024, and PI 638856 than in three unresponsive varieties PI 661803, PI 638858, and W6 17019 (Fig 2B and S3 Fig in S1 File). To test if promoter variation from these lines corresponded with In11 sensitivity, we created promoter-luciferase fusions. We selected two responsive accessions, PI 638852 and PI 638856 and two unresponsive accessions, PI 661803 and PI 638858, and tested the ability of their pINR alleles to drive expression in Agrobacterium-transformed Nicotiana benthamiana. Promoters from the two In11-responsive lines, pINR^{PI 638852} and pINR^{PI 638856}, drove 48.0% higher luciferase activity than the pINR^{PI 638858} from an unresponsive line, while a second promoter pINR^{PI 661803} was similarly low in activity (Fig 2C). In summary, the strength of INR expression across P. vulgaris varieties corresponded with promoter strength in the heterologous expression system.

We compared the tested pINR sequences and found additional SNPs relative to those called from short-read resequencing, as well as an additional deletion from −387 to −448 bp from the TSS for pINR^{PI 638856} (Fig 2D and S4A Fig in S1 File). Many SNPs were embedded in predicted transcription factor binding motifs. While unresponsive promoters contained several SNPs relative to the high-expression promoter pINR^{PI 638852}, none were uniquely present in unresponsive promoters. Motif enrichment analysis of these polymorphisms also did not clearly pinpoint a single transcription factor binding site (TFBS) corresponding with promoter strength (S4B-F Fig in S1 File). We conclude that a combination of promoter changes likely functions to alter expression strength in the pINR::Luc constructs.

In11 insensitivity in low INR expression lines does not correspond with herbivore resistance

To test the effects of the low-expression INR allele, we generated recurrent backcross lines varying at the INR locus with a weak promoter allele as characterized above. We crossed PI 638858 with the recurrent parent G19833 for a total of four backcrosses, each time selecting for the PI 638858 allele (here termed inr-2), and selfed to select sibling near-isogenic lines (NILs) lines homozygous for inr-2/inr-2 or the G19833 allele (INR/INR) (S1 Fig). INR expression level was reduced in inr-2/inr-2 plants, comparable to differences between parent lines G19833 and PI 638858. (Fig 3A). In11 sensitivity as measured by In11-induced ethylene response was confirmed in INR/INR plants, while inr-2/inr-2 lines remained insensitive (Fig 3B). In contrast, response to an unrelated damage-associated elicitor peptide PvPep1 [40,41] was not affected, indicating that inr-2 specifically disrupts In11-dependent signaling. In addition, INR genotype corresponded with sensitivity of an In11-responsive marker gene encoding a MYB transcription factor previously shown to be highly induced by In11 [36], consistent with rapid In11-induced transcriptional reprogramming dependent on INR expression level (Fig 3C).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. INR genotype associates with lack of In11 sensitivity, but not herbivore susceptibility or In11-induced herbivore protection A) Relative expression of INR in the first trifoliate leaf tissue 2 weeks post germination.

Colors represent INR genotype in parent accessions or near isogenic lines (NILs) (G19833, INR vs PI 638858, inr-2). B) BC4F5 In11 response corresponds to INR genotype. Ethylene production in BC4F5 INR/INR and inr-2/inr-2 lines after treatment with H₂O, 1 μM PvPep1 peptide, or 1 μm In11 peptide. Boxes show the median, 1st and 3rd quartiles and whiskers extending to the minimum and maximum values of replicate leaflets for individual water or peptide treatments (n = 6). Colors signify peptide or H₂O treatment. C) Relative expression of PvMYB (Phvul.001G215100) marker gene in the primary leaf tissue. Colors in bars represent treatment, w + H₂O (white), w + In11 (green). Different letters in A), B) and C) represent significant differences (One-way ANOVA p < 0.05; Tukey Honestly Significant Difference (HSD), α = 0.05). Pairings that did not have letters did not reach the threshold for statistical evaluation. Statistical significance in the second facet of A) was assessed by an unpaired two-tailed Student’s t-test. Significance levels: *** p < 0.001; ** p < 0.01; * p < 0.05; ns, not significant. D) Herbivore mass gain on BC4F5 plants. Final masses of 2^nd instar Spodoptera exigua larvae feeding on common bean lines for 7 days after trifoliates were treated with inceptin-11 or left untreated. Each dot represents one larva (n = 13 - 25 per group) feeding on an individual plant. Colors represent responsive vs unresponsive lines; difference in shading represents treatment with peptide. Data is pooled from two independent experiments. Different letters indicate groups that differ significantly based on Games–Howell post-hoc comparisons (α = 0.05), following a Welch’s ANOVA that indicated a significant overall effect (p < 0.05).

https://doi.org/10.1371/journal.pone.0343332.g003

In order to test the effect of the inr-2 allele on direct defenses against herbivory, we performed herbivore growth experiments using beet armyworm (Spodoptera exigua) on the sibling lines. No significant difference in larval weight gain between inr-2 and INR plants was observed (Fig 3D). We also assessed sensitivity to In11-induced protection against herbivores, as In11 pretreatment was previously shown to reduce subsequent larval weight gain in a related legume species, cowpea (Vigna unguiculata) [28]. Surprisingly, pretreatment reduced caterpillar mass gain by as much as 79% for both inr-2 and INR sibling lines, indicating that wounding and In11 can still elicit defense responses on inr-2 plants in the pretreatment setup. (Fig 3D). These results indicate that In11 recognition can potentially still occur in inr-2 lines and disrupt caterpillar physiology enough to decrease mass gain. We hypothesized that INR expression might increase after wounding, explaining In11 sensitivity in the pretreatment assay. However, INR expression was similarly low relative to PI 638856 in wounded plants at 6 h and 24 h after wounding, as well as after treatment with the damage-induced elicitor peptide PvPep1-treated plants (S5 Fig in S1 File). These results demonstrate that differences in promoter profiles differentially drive INR expression, potentially diminishing the immune phenotype, but unlike an INR null allele, the inr-2 allele can still mediate an immune response to In11.

Plant growth

Seeds were sown on Sunshine Mix No. 4 and germinated in reach-in PGC-FLEX growth chambers (Argus/Conviron) at 26ºC, 50% relative humidity, and 500 μMol of light with a 12/12 h light/dark photoperiod. On day twelve, plants with a fully expanded first trifoliate were transferred to a walk-in PGW40 (Conviron) at 26º, 50% relative humidity, and 500 μMol of light with a 12/12 h light/dark photoperiod prior to experiments. For Nicotiana benthamiana, plants were transplanted 1 week after sowing in five parts Sunshine Mix No. 4 and one part Premier Sphagnum Peat moss. Starts were fertilized with 50g Jack’s Professional 20-20-20 General Purpose mixed in with 4L of water. Transplants were grown at 20°C under 12-h light and dark cycles and 160−200 μMol of light. The seedlings were grown under humidity domes for three weeks, after which the domes were removed, and the plants were grown an additional week before infiltrations. Seedlings were fertilized every other week starting at 3 weeks old with 25g Jack’s Professional 20-20-20 General Purpose mixed with 4L water. Fully expanded, mature leaves of five-week-old plants were used for all transient expression experiments.

Resequencing of Phaseolus vulgaris varieties

DNA was prepared from trifoliate leaflets using Macherey-Nagel Plant DNA Kit (Macherey-Nagel Inc.) Illumina libraries were synthesized and 150 bp paired-end reads were sequenced with the HiSeq X instrument by Illumina, producing ≈237.9 million paired-end read fragments in ≈68.7 Gb of data. 94.72% of reads were mapped to the Phaseolus vulgaris G19833 reference genome (Pvulgaris_442 v2.0) (Phvul 1.0 using BWA-MEM (v 1.2.3) [34,56] using default settings representing 100% genome coverage. The average sequence depth was 5.57-fold (2.98–9.49-fold). Genotype coverages at ≥4x, ≥ 10x and ≥20x averaged 85.89%, 62.48% and 52.55%, respectively (S2 Table in S1 File). Variants were called using Freebayes (v 0.9.14) [57]. Distance matrix was calculated using TASSEL (v 5.2.79) [58]. The TASSEL pairwise genetic distance matrix was imported into R (v 4.4.2) and clustered with average-linkage hierarchical clustering (hclust), and the resulting clustered distance heatmap and dendrogram were generated with the pheatmap package (v1.0.13) [59].

For analysis of INR promoter variation, a different pipeline was used. The Pvulgaris_442 v2.0 reference genome was first indexed with samtools faidx (v1.17) and the 6.8 kb INR interval on Chr07 (7,407,744–7,413,510) was extracted into a standalone FASTA [60]. For each accession, sorted, duplicate‐marked binary alignment maps (BAMs) were then processed with bcftools (v1.17) mpileup and call (base Q ≥ 13, θ = 1.1 × 10 ⁻ ³) to generate VCFs, and sample-specific consensus sequences were built with bcftools consensus, masking any site with depth < 10, alternate count < 3, or QUAL < 30 for downstream alignment and SNP visualization [61]. We have marked SNP positions where there was lack of coverage in S2B Fig in S1 File. Variant and coverage data were processed in R (Version 4.4.2). Sequencing quality control statistics were obtained using samtools and mosdepth [62] and parsed through custom R scripts using tidyverse [63]. Alignment was performed with Benchling alignment tool [64] and visualization was performed in Adobe Illustrator CC 29.7.1 and 30.0.

Generation of inr-2 introgression lines

PI 638858 was selected by virtue of its sexual compatibility with the Phaseolus vulgaris reference genome G19833. Mature stigmata from opened flowers of Phaseolus vulgaris PI 638858 containing the lone SNP in the promoter region of the INR locus (inr-2) were excised and used to pollinate the immature stigma from unpollinated, unopened host flowers of Phaseolus vulgaris G19833 that did not harbor the SNP. The resulting seeds were planted and genotyped as described. Heterozygous plants were kept and crossed with the recurrent line, G19833. This procedure was repeated until four backcross lines were generated. For the fourth backcross line, heterozygous plants were allowed to self-pollinate and their offspring genotyped to select homozygous individuals with INR/INR or inr-2/inr-2 genotypes.

Transcription factor binding site enrichment analysis

Promoter sequences were organized into a fasta file grabbing 0 bp from the TSS to −1852 from the TSS. This fasta file was uploaded to the PlantRegMap Binding site prediction tool [37] to identify all possible TF binding sites on each sequence. TF binding sites were filtered for loci that are changed from G19833 and then annotated on an alignment chart. Alignment was done using the Benchling alignment tool [64]. Herbivore-inducible transcription factor binding site predictions were obtained using FIMO [65], part of the MEME suite. We supplied FIMO with position-specific frequency matrices for herbivore-inducible motifs and scanned the target sequence, producing a tab-delimited output file, which was imported into R (Version 4.4.2) and combined with SNP coordinates from PlantRegMap and visualized with ggplot2 [66] and Adobe Illustrator CC 29.7.1.

Generation of Luciferase reporter constructs

INR promoter regions (pINR) of four common bean accessions (PI 638852, PI 638856, PI 661803, and PI 638858) were amplified from genomic DNA of Phaseolus vulgaris using gene-specific primers and using AccuPrime Taq High Fidelity MM polymerase (ThermoFisher Scientific) (S3 Table in S1 File). These primers amplified from the start codon to approximately 1.8 kb upstream. Promoter regions were then cloned into the promoter + 5U acceptor backbone pICH41295 from the Golden Gate MoClo Plant Toolkit using type IIS restriction enzymes (BbsI, New England Biolabs) [67]. Constructs were selected in E. coli DH5α on LB agar supplemented with spectinomycin (50 µg/mL), Isopropyl β-D-thiogalactopyranoside (IPTG, 100 mM), and 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal, 20 mg/mL). Promoter, luciferase, and terminator L0 parts were assembled into the pGII acceptor backbone using BsaI HF®v2 restriction enzyme (New England Biolabs). Constructs were selected in E. coli DH5α on LB agar supplemented with kanamycin (50 µg/mL). Reporter constructs were transformed by electroporation into Agrobacterium tumefaciens GV3101 (pMP90, pSOUP). Constructs were selected on LB agar supplemented with kanamycin (50 µg/mL), gentamycin (50 µg/mL), rifampicin (50 µg/mL), and tetracycline (10 mg/mL). All sequences were verified by Whole Plasmid Sequencing performed by Plasmidsaurus using Oxford Nanopore Technology with custom analysis and annotation.

Luciferase assays

Agrobacterium tumefaciens GV3101 harboring Luciferase reporter plasmids [67] driven by the different promoters (PI 638852, PI 638856, PI 638858, PI 661803) was cultured for 24h and then resuspended in infiltration media containing 10 mM 2-(N619 morpholino) ethanesulfonic acid (MES) (pH 5.6), 10 mM MgCl2, and 150 μM acetosyringone and diluted to a matched OD600 of 0.6 [68]. Cultures were incubated at RT for 3h in IM and then infiltrated into fully expanded leaves of 6-week-old N. benthamiana plants. 48h post infiltration 8 circular leaf punches (4 mm) were taken per infiltration strain per leaf and floated stomata side down on 100 µL of H₂O containing 1 mM Luciferin-D in a white BioRad 96-well plate. The plate was immediately placed in a Tecan SPARK® Multimode Microplate reader following collection and relative luminescence units (RLU) were measured with 5000 ms reads per well for 90 minutes. The max RLU was taken for each well and normalized to the mean of PI 638858 bio-replicates for each test day.

Genotyping

Samples were taken from primary leaves of common bean plants and flash frozen in liquid nitrogen. Frozen samples were then ground to a fine powder using a mixer mill (Retsch MM400, 2008, Part No. 20.745.0001) and genomic DNA extraction was performed using the NucleoSpin® Plant II kit (Machery-Nagel, 740770.520) according to the manufacturer’s instructions. The concentration and quality of the extracted DNA was assessed by NanoDrop One (ThermoFisher Scientific ND-ONE-W) and gel electrophoresis. Common bean accessions were genotyped using gene-specific primers (S3 Table in S1 File) flanking a ~ 1000 bp region that contained a single nucleotide polymorphism (SNP) in the PvINR locus at position 7,408,876 in the PvINR locus polymorphic in the parent accessions G19833 and PI 638858. PCR reactions were performed using gene-specific primers with DreamTaq PCR MasterMix (Thermo Fisher Cat. No. K1081).

Peptide-induced ethylene gas production and gene expression in common bean

The In11 peptide (ICDINGVCVDA) was synthesized (Genscript) based on the cATP synthase sequence from cowpea (Vigna unguiculata) and reconstituted in water. The PvPep1 peptide (STSLIARSGRRSTVSHGSGPQHD) was synthesized (SynPeptide) based on the Phaseolus vulgaris gene Phvul.007G086700 (Phytozome).

For ethylene gas production, the left and middle leaflet of the first trifoliate or the two primary leaves of a common bean plant were lightly wounded on both sides of the main vein using a fresh razor blade to remove the cuticle, and 10 μL water, 1 μM In11, or 1 μM PvPep1 were distributed between the wounds using the tip of a pipette. After 1 h, leaflets or primary leaves were excised at the petiole using a razor blade and placed in sealed tubes for 1h before sampling 1 mL of the headspace. Ethylene gas was measured with a gas chromatograph (HP 5890 series 2, supelco #13018-U, 80/100 Hayesep Q 3FT x 1/8IN x 2.1MM nickel) with flame ionization detection and quantified using a standard curve (Scott, 99.5% ethylene, Cat. No 25881-U) [69].

For gene expression analyses, the middle leaflet of the first trifoliate or the primary leaves of a common bean plant were lightly wounded on both sides of the main vein using a fresh razor blade to remove the cuticle, and 10 μL water or 1 μM In11 were distributed between the wounds using the tip of a pipette. After 1h, leaf samples were collected by removing the treated area with a 0.6 cm² cork borer and flash frozen in liquid nitrogen. All samples were stored at −80ºC until RNA extraction.

RNA extraction, cDNA synthesis and qRT-PCR

Total RNA was extracted using the TRIzol™ Reagent (Invitrogen, Cat. No. 15596026) method. Briefly, frozen samples were ground to a fine powder using a mixer mill (Retsch MM400, 2008), and the rest of the extraction procedure was carried out according to manufacturer’s instructions. The concentration and quality of the extracted RNA was assessed by NanoDrop One (ThermoFisher Scientific ND-ONE-W) and gel electrophoresis, respectively. As DNA contamination was evident from the gel electrophoresis, the RNA was treated with TURBO DNA-free kit (Invitrogen, Cat. No AM1907) following manufacturer’s instructions. cDNA was synthesized using 1 μg of RNA using the SuperScript IV Reverse Transcriptase Kit (ThermoFisher, Cat. No. 18090050). qPCR reactions were performed using gene-specific primers with PowerUp SYBR Green PCR Master Mix (Thermo Fisher, Cat. No. 4367659). Relative expression was quantified using the ∆∆Cq method using ∆Cq values normalized to the common bean UBQ gene and made relative to wound (S3 Table in S1 File).

Herbivory experiments

First instar beet armyworm (Spodoptera exigua) larvae were obtained from Benzon Research Inc (Carlisle, PA). Upon receipt, larvae were incubated in their original packaging (artificial diet) under dark at 28ºC for 48 h to accelerate molting. Second instar larvae were weighed using a digital scale (Mettler Toledo), and ~1.5 mg caterpillars were placed in plastic cups without artificial diet. Once all masses were recorded, one single larva was placed on the first trifoliate or primary leaf of INR/INR and inr-2/inr-2 plants, and leaves were enclosed in a transparent mesh bag to contain the larva. Infested plants were kept in a growth chamber at 26°C, 50% RH, and 500 μMol light, and larvae weights were recorded seven days later. For plants that were pretreated with In11, 24 hours before the experiment, the first trifoliate that the caterpillar would be placed on was scratch-wounded with a razor to remove the cuticle and 10 uL of 1 μM of In11 peptide was applied.

Statistical analysis

All statistical analyses were conducted in R (Version 4.4.2) and manipulated using base R functions and dplyr [70]. We first tested whether residuals fulfilled the assumption of normality using a Shapiro-Wilk Test, Levene Test, and qq plots using rstatix, ggpubr, and car [71–73]. For normally distributed data, various statistical evaluations were conducted, including the Student’s T-test, one-way analysis of variance (ANOVA) followed by Tukey’s Honestly Significant Difference (HSD) post-hoc test (base R, multcompView [74]). In the cases where data did not conform to normal distribution and group variances, Welch’s ANOVA with Games-Howell’s post-hoc test (rstatix) was conducted to determine differences in relative expression between treatments or the mean mass of beet armyworm caterpillars feeding on inr-2/inr-2 or INR/INR lines. Scripts used to analyze data and generate figures are available at github.com/BQBehnken/Argentinian_Beans_Project_Scripts.

Geoanalysis

All mapping and spatial‐data processing was performed in R (Version 4.4.2). The script used to generate the figures is available at github.com/BQBehnken/Argentinian_Beans_Project_Scripts. Geographic coordinates for each bean accession were retrieved from the USDA-ARS Germplasm Resources Information Network (GRIN) database and aggregated by location and response type to compute point counts. Country boundaries were obtained from rnaturalearth [75]. A bounding box was constructed around the points and passed to elevatr [76] to download ~1 km DEM data, which was converted to a terra SpatRaster [77]. Monthly mean temperature and precipitation layers (WorldClim v2.1, 10′ resolution, and 30s resolution) were read from local GeoTIFFs into two SpatRasters, cropped to the bounding box, and renamed with month abbreviations [78]. Plots were built with ggplot2 [66], exported as an encapsulated postscript (.EPS) file, and arranged in Adobe Illustrator CC 29.5.1. Visual delineation of a selection seeds from responsive and unresponsive accessions were photographed by a Nikon D750 and annotated in Adobe Photoshop CC 26.7.