{"paper_id":"040d231d-8165-4b94-9e04-2e8403fed9a9","body_text":"1 \n \nSouthern South American Maize Landraces: A Source of Phenotypic Diversity \nDudzien, Tatiana L uján1; Freilij, Damián 2,3; Defacio, Raquel  Alicia4; Fernández, Mariana 4; \nPaniego, Norma Beatriz3; Lia, Verónica Viviana1,2; Dominguez, Pia Guadalupe3* \n1Instituto de Ecología, Genética y Evolución de Buenos Aires (IEGEBA, UBA-CONICET), Buenos Aires, \nArgentina \n2Facultad de Ciencias Exactas y Naturales, Universidad de Buenos Aires (UBA), Buenos Aires, \nArgentina \n3Instituto de Agrobiotecnología y Biología Molecular (IABIMO, INTA -CONICET), Buenos Aires, \nArgentina \n4Estación Experimental Agropecuaria Pergamino, INTA, Pergamino, Buenos Aires, Argentina \n* Corresponding author, piagdominguez@gmail.com \n \nAbstract \nMaize (Zea mays L.) is a globally important crop for food, feed, and industrial uses. Modern \nbreeding increasingly targets traits beyond yield, including stress tolerance, nutritional \nquality, and pest resistance. Progress toward these goals is constrained by the narrow \ngenetic diversity of commercial varieties, a consequence of the repeated use of a limited \nnumber of inbred lines. Maize landraces therefore represent valuable reservoirs of genetic \nand phenotypic diversity. \nNorthern Argentina is one of the southernmost regions of maize landrace cultivation and \ncomprises two main centers of diversity: Northeastern Argentina (NEA; <2000 m.a.s.l.) and \nNorthwestern Argentina (NWA; >2000 m.a.s.l.). Despite their potential, phenoty pic \ncharacterization of these landraces remains limited, particularly for biochemical traits, \nwhich, although less visible, play key roles in biomass accumulation, defense against \npathogens and herbivores, tolerance to environmental stress, and quality att ributes such \nas flavor. \nHere, we evaluated 17 phenotypic traits, including morphological traits, biochemical \ncompounds (such as pigments, carbohydrates, and phenolics), and salt stress tolerance, \nin 19 maize landrace accessions from Northern Argentina. Substantial variation was \ndetected across all traits, both within and among accessions, indicating that each \naccession harbors a distinct phenotypic profile. While no si gnificant differences were \nobserved between regions, redundancy analysis revealed associations between \nphenotypic variation and collection-site altitude. \nThese findings highlight the value of Argentine maize landraces as sources of biochemical \nand stress -related traits and support their conservation and use in breeding programs \naimed at broadening the genetic base of cultivated maize. \n \nKey words \nMaize Landrace-Biochemistry-Salt Tolerance-Variability-RDA-Altitude \n \n \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n2 \n \nIntroduction \nMaize (Zea mays L.) is one of the most important crops worldwide, serving as a staple food \nfor humans and livestock and as a raw material for numerous industrial products. Despite \na global grain production of approximately 1.23 billion metric tons in 2024/2025 ( USDA-\nFAS, 2025 ), the growing human population and increasing demand for animal -derived \nproducts continue to drive the need for higher maize yields (Andorf et al. 2019). Moreover, \nmodern breeding goals extend beyond yield improvement to include enhanced tolerance \nto abiotic stresses, improved nutritional quality, pest resistance, and suitability for \nindustrial uses (Liu et al. 2020; Prasanna et al. 2020 ). One of the main limitations in maize \nbreeding is the narrow genetic base of commercial varieties, a consequence of the \nrecurrent use of a limited set of ancestral races in breeding programs (Hufford et al. 2012; \nSmith et al. 2017). Indeed, commercial maize is estimated to contain only 5-10% of the total \ngenetic diversity available in the species (Hoisington et al. 1999). \nMaize was domesticated in Mexico approximately 9000 years before present (BP) \n(Matsuoka et al. 2002; Piperno et al. 2009) and introduced into South America around 8000 \nyears BP (Piperno 2011; Bonavia 2013; Aceituno and Loaiza 2014 ). As it spread across the \ncontinent through human-mediated exchange, maize underwent substantial demographic \nand selective changes that led to the development of numerous local varieties, known as \nlandraces. The term landrace refers to populations or groups of individuals sharing \ncommon morphological, ecological, and genetic characteristics linked to their cultivation \nhistory, which distinguish them as a recognizable group (Anderson and Cutler 1942). In \ngeneral, landraces are char acterized by a shared histori cal origin, high genetic diversity, \nlocal adaptation, identifiable phenotype, absence of formal breeding, and association with \ntraditional farming systems (Camacho Villa et al. 2005; Guzzon et al. 2021).  \nNorthern Argentina constitutes one of the southernmost regions for maize landrace \ncultivation. This area has been proposed as a historical zone of interaction between Andean \nand tropical lowland maize varieties (Vigouroux et al. 2008; Tenaillon and Charcos set \n2011). It harbors around 57 distinct maize landraces grouped into two major genetic \nclusters corresponding to contrasting agroecosystems: the Northwestern (NWA) and \nNortheastern (NEA) Argentine maize groups (Bracco et al. 2012; Melchiorre et al. 2017; \nRealini et al. 2018; López et al. 2021; Rivas et al. 2022; Dominguez et al. 2024 a). \nNorthwestern Argentina is a mountainous region reaching up to 4000 m above sea level, \ncharacterized by low precipitation, high solar radiation, and large daily temperature \namplitudes (Rivas et al. 2022) . E ntisols, alfisols, mollisols, and rock outcrops  are the  \npredominant soil types (Panigatti 2010) . In contrast, Northeastern Argentina lies near sea \nlevel, with subtropical temperatures and high annual rainfall (Heck et al. 2020 ), and \noxisols/ultisols, alfisols, mollisols, and vertisols as predominant soil types (Panigatti 2010).  \nAssessing phenotypic variation in crops is fundamental for identifying traits relevant to \nbreeding (Geiler -Samerotte et al. 2013). Such variation encompasses not only \nmorphological differences but also physiological, biochemical, and stress -related traits. \nAmong these, resistance to abiotic and biotic stresses is increasingly important under \nchanging climate conditions. Biochemical traits, though less visible, play a key role in \nagronomically important processes, including biomass accumulation, defense agai nst \npathogens and herbivores, tolerance to environmental stress, and organoleptic properties \nsuch as flavor (Pott et al. 2019). Phenotypic variation in any population arises from both \ngenetic and environmental factors. While morphological, biochemical and stress -related \ntraits are environmentally influenced, several studies have demonstrated that their \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n3 \n \nunderlying genetic basis in maize is important (Reynolds et al. 2005; Mladenović et al. 2014; \nKumar et al. 2015; Das et al. 2019; Santiago et al. 2025).  \nMaize landraces from Northern Argentina have been extensively characterized at the \nmorphological and phenological levels (Cámara -Hernández et al. 2012; Melchiorre et al. \n2017, 2020; Rivas et al. 2022; Realini et al. 2023; Defacio et al. 2025). However, studies \naddressing other phenotypic dimensions remain limited. For example, Heck et al. (2019)  \nreported wide variation in oil and fatty acid content among 16 NEA landraces, while \nMansilla et al. (2021) found differential carbohydrate and phenolic levels in tw o open -\npollinated populations derived from Argentine landraces. Similarly, Sampietro et al. (2013) \nidentified variation in phenylpropanoid levels associated with biotic stress resistance in \nthree Argentine landraces. Together, these findings suggest substantial biochemical \ndiversity within local maize varieties, underscoring their potential value for br eeding \napplications. \nIn this context, the present study aims to expand the morphological, biochemical and salt-\nstress tolerance characterization of maize landraces from Northern Argentina to assess the \nextent of variability among them. To this end, 19 landrace accessions origi nating from the \nNWA and NEA regions were evaluated under common garden conditions. The analyses \nincluded morphological and agronomically relevant leaf biochemical traits -such as \nphenolic compounds, total sugars, and pigments- as well as assessments of seedling salt \nstress tolerance under three different salt conditions. The results provide new insights into \nthe phenotypic diversity of Argentine maize landraces, contributing to their conservation \nand to breeding efforts aimed at broadening the genetic base of cultivated maize. \n \nMaterials and Methods \nPlant Material \nA set of 19 maize landrace accessions from Argentina was obtained from the “Banco Activo \nde Germoplasma INTA Pergamino” (BAP; Active Germplasm Bank of the National Institute \nof Agricultural Technology, Pergamino, Buenos Aires, Argentina), representing two distinct \ngeographical regions, NWA and NEA (Figure 1; Supplementary Table 1).  \nCultivation Conditions \nPlants were grown in a greenhouse (80% relative humidity; 200 µmol PAR s ⁻¹ m⁻²; 16 h \nlight/8 h dark). Five individuals per accession were cultivated in each of two independent \nexperiments: Experiment 1 (January -April 2023) and Experiment 2 (October 2023 -January \n2024). \nAgromorphological Analyses \nLeaf number, plant height, and chlorophyll content (Dualex® optical leaf clip meter, ForceA) \nwere measured 30- and 37-days post-sowing in Experiments 1 and 2, respectively. Because \nthese traits were measured at different growth points, they were analyzed independently for \neach experiment (see statistical analysis section).  Stem diameter and dry biomass (oven-\ndried at 80°C for 3 days) were recorded for each plant at the end of each experiment at the \nsame growth stage (R5) and were analyzed collectively (see statistical analysis section). \nSample Collection for Biochemical Analyses \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n4 \n \nFully expanded leaves from V12 -stage plants were collected  in each experiment , \nimmediately frozen in liquid nitrogen, stored at −80°C, and ground to a fine powder under \nliquid nitrogen.  Data from both experiments were analyzed collectively (see statistical \nanalysis section). \nSpectrophotometric Measurements \nAll biochemical measurements were performed using a Thermo Scientific Multiskan \nSkyHigh spectrophotometer. \nTotal Protein Measurement \nProtein extraction was performed by adding 1 mL extraction buffer (100  mM NaCl, 50 mM \nTris pH 7.5 -8, 10 mM DTT -dithiothreitol-, 20% v/v glycerol, 2 mM PMSF-\nphenylmethylsulfonyl fluoride ) to 100 mg ground tissue. Protein concentration was \ndetermined using the Bradford assay (Bradford, 1976) adapted for microplates (He, 2011). \nEach well contained 200 µL of 1:4 diluted Bradford reagent (Biorad, catalogue #500-0006) \nand 4 µL of sample. A BSA calibration curve (0.25-10 µg/µL) was included, and absorbance \nwas measured at 600 nm. \nChlorophyll and Carotenoid Measurement \nPigments were extracted by adding 250 µL of 95% ethanol to 10 mg of sample and \nincubating at 80°C for 20 min. The supernatant was transferred to a clean tube, and the \nprocess was repeated twice to obtain pooled extracts. Absorbance was recorded at 664.2, \n648.6, and 470 nm, and pigment concentrations were calculated using Lichtenthaler’s \nequations (Lichtenthaler 1987). \nTotal Sugar Measurement \nTotal soluble sugars were quantified following Laurentin and Edwards (2003) using the same \nethanolic extract as for pigment analysis. A glucose calibration curve (0.025-1 mg/mL) was \nused. The reaction was performed with 40 µL of sample and 100 µL of 0.2 % anthrone \nreagent (Biopack, catalogue #9939.02), vortexed, incubated at 90°C for 17 min, cooled on \nice, and measured at 620 nm. \nStarch Measurement  \nStarch content was quantified following Smith and Zeeman (2006) with modifications. The \nethanol-insoluble pellet from pigment and sugar extractions was resuspended in 400 µL of \n0.1 M NaOH, incubated at 95°C for 30 min, and neutralized with 80 µL of 0.1 M sodium \nacetate buffer. Aliquots (40 µL) were incubated overnight at 37°C with shaking in 100 µL of \nenzyme solution containing α-amylase (0.5 U) (Sigma -Aldrich, catalogue #A3176) and \namyloglucosidase (0.45 U) (Sigma-Aldrich, catalogue #ROAMYGL) in 50 mM acetate buffer. \nControl samples rece ived buffer only. The glucose r eleased was quantified in the \nsupernatant using the anthrone method as described for total sugars. \nTotal Phenolic Compounds Measurement \nTotal phenolic content was determined following Ainsworth and Gillespie (2007) using the \nFolin-Ciocalteu assay with a gallic acid calibration curve (0.05-1 mM). For extraction, 1.5 mL \nof 95% methanol was added to 20 mg of sample and shaken at 4°C for 2 h. The supernatant \nwas reacted with 200 µL of 10% Folin-Ciocalteu reagent (Biopack, catalogue #0891.05) for \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n5 \n \n6 min, followed by 800 µL of 700 mM sodium carbonate and incubation at 45°C for 15 min. \nAbsorbance was measured at 765 nm. \nAntioxidant Activity Measurement \nAntioxidant activity was determined using the ABTS assay (Re et al. 1999). For extraction, 1 \nmL of hexane:acetone:ethanol (2:1:1) was added to 20 mg of sample and shaken for 15 min. \nAfter addition of 150 µL of water and further mixing, the hydrophobic (upper) and hydrophilic \n(lower) phases were separated. ABTS (Abcam, catalogue #ab142041) was diluted in ethanol \nto an absorbance of 0.7, and reactions were carried out with 10 µL of extract and 200 µL of \nABTS solution. A gallic acid calibration curve (0.025-0.0025 mM) was used, and absorbance \nwas read at 734 nm in both phases. \nSalinity Stress Test \nAccessions were evaluated under three NaCl treatments: 0 mM (T0), 17.5 mM (T1), and 35 \nmM (T2). Seeds (20 per accession) were sown in sand moistened with the respective NaCl \nsolution and incubated for 21 days in germination chambers (20-30°C; 16 h light/8 h dark). \nSeedlings were harvested, washed, and measured for root length (from stem base to \nlongest root tip) and shoot length (from root insertion to leaf apex). Roots and shoots from \nthe same accession were pooled, dried at 65°C for 48 h, and weighed together to obtain a \nsingle root and shoot dry weight per accession.  Three independent replicates of the \nexperiment were performed. \nUnivariate Statistical Analyses \nMorphological and biochemical traits in adult plants were fitted with linear mixed-effects \nmodels to assess differences at two hierarchical levels: (1) Region, and (2) Accession. For \nthe regional analyses (1), region was treated as a fixed factor, while accession (nested within \nregion) and experiment were included as random effects: Y ∼ Region + \n(1∣Region/Accession) + (1∣Experiment). Models were fitted using the lmer() function from \nthe lme4 package (Bates et al. 2015). For the accession-level analyses (2), accession was \nmodeled as a fixed effect and experiment as a random effect: Y  ∼ Accession + \n(1∣Experiment). These models were fitted with the lme() function from nlme (Pinheiro et al. \n2025), allowing for heterogeneous variances among accessions or experiments using the \nvarIdent() or varPower() variance structures when appropriate.  For the three phenotypic \ntraits measured at different plant stages in each experiment (plant height, leaf number, and \nchlorophyll content measured with Dualex®), analyses were conducted separately for each \ntrial. For continuous variables (height and chlorophyll), we applied linear models to each \ndataset as before, with the term \" Experiment\" being excluded from the models. For count \ndata (leaf number), a Conway-Maxwell-Poisson distribution (family = compois) was applied \nusing glmmTMB (Brooks et al. 2017). \nVariables associated with salt stress in seedlings were analyzed using linear mixed-effects \nmodels. Two analytical approaches were employed: (1) Region and Treatment were \nconsidered fixed factors, while Accession and Experiment Repetition were included as \nrandom factor s; (2) Accession and Treatment were considered fixed factors , while \nExperiment Repetition was considered a random factor. Leaf and root length were analyzed \nusing both approaches, while leaf and root dry weight were analyzed only with approach (1), \nas individual accession -level data were not available. In approach (1), the region × \ntreatment interaction was initially tested but found non-significant, and models without the \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n6 \n \ninteraction term were subsequently fitted. The general model structure was: Y ∼ Region + \nTreatment + (1 ∣ Region/Accession) + (1|Repetition). Models were fitted using the lmer \nfunction from the lme4 package (Bates et al. 2015). In approach (2), a significant accession \n× treatment interaction was detected; therefore, models including the interaction term were \nfitted as follows: Y ∼ Accession×Treatment + (1|Repetition). Models were fitted using the \nlme function from the nlme  package (Pinheiro et al. 2025), applying appropriate variance \nstructures to account for heteroscedasticity among groups. \nFor all the analyses, model selection was based on Akaike’s Information Criterion (AIC) and \ninspection of diagnostic plots. Normality was evaluated with Q -Q plots and \nhomoscedasticity with residuals versus fitted values plots. Logarithmic  transformations \nwere applied when necessary to meet model assumptions. Estimated marginal means \n(EMMs) were obtained with the emmeans package (Lenth and Piaskowski 2025) , and \npairwise comparisons were adjusted for multiple testing using Tukey’s method. Back -\ntransformations to the original scale were performed with the \"response\" option. Statistical \nsignificance was set at p < 0.05. \nMultivariate Statistical Analyses \nCoefficients of variation (CV) and Spearman correlations were computed for the 17 \nmeasured traits in Experiment 1 and 2 in adult plants using the stats and rcorr functions of \nthe Hmisc package, respectively (R Core Team, 2023). Pairwise comparisons were adjusted \nfor multiple testing using the sequential Bonferroni correction applied by  the p.adjust \nfunction of stats.  For Accession-level analyses, data were averaged per accession and then \nstandardized and centered within each experiment to avoid potential scale-related biases. \nPrincipal Component Analysis (PCA) was conducted with the prcomp function (stats), and \nresults were visualized using the fviz_pca_ind and fviz_pca_var functions from factoextra \n(Kassambara and Mundt 2020). The structure of the morphological and biochemical data \nwas further examined through Discriminant Analysis of Principal Components (DAPC) with \nadegenet (Jombart 2008), which identifies clusters by maximizing between-group variation \nwhile minimizing within -group variation (Jombart et al. 2010) . Up to 10 clusters were \nevaluated with find.clusters(), and cross-validation (xvalDapc, 90% training set) was used \nto determine the number of PCs (Principal Components) to retain, selecting the value that \nminimized the mean squared error (MSE). Based on this, the optimal number of groups was \nchosen according to the Bayesian Information Criterion (BIC). Cluster assignments were \nvisualized with scatterplots of the first di scriminant function and membership probability \nbarplots (compoplot). Additionally, hierarchical cluster analysis was performed on \nadjusted means derived from the linear models performed previously. Euclidean distances \nbetween accessions were computed, and the agglomerative coefficient, calculated with \nagnes() (cluster, Maechler et al. 2022)  across different linkage methods (average, single, \ncomplete, Ward’s), indicated Ward’s minimum variance method as  the most suitable \nclustering approach. The optimal number of clusters was determined with the average \nsilhouette method using the fviz_nbclust function ( factoextra), and the resulting \ndendrogram was generated with the ggdendro package (de Vries and Ripley 2024). Finally, \nto explore potential geographical patterns, we performed a Redundancy Analysis (RDA) to \nquantify the proportion of multivariate trait variation ( employing adjusted means derived \nfrom the univariate linear models ) explained by Altitude, Latitude and Longitude \n(Supplementary table 1)  as environmental predictors. The analysis was conducted using \nthe vegan package (Oksanen et al. 2022) , fitting one model with Altitude as the sole \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n7 \n \npredictor and another also including Latitude and Longitude. Significance was assessed \nthrough permutation tests with 999 replicates. \n \nResults \nAccessions Show High Variability with Low Coefficients of Variation at the \nMorphological Level \nPlants were evaluated in two independent experiments at the morphological level (Tables 1 \nand 2, Supplementary table 2). All traits showed higher CV (coefficient of variation) in \nExperiment 1 in comparison to Experiment 2, indicating a higher variability i n the former \nassay (Supplementary figure 1 ) probably due to environmental differences  between \nexperiments combined with the high plasticity of maize. \nThe analysis of phenotypic traits between accessions from NEA and NWA showed slight \ndifferences when compared at the region of origin level, with only two traits exhibiting \nstatistically significant variation (Supplementary table 2). Leaf number differed significantly \nin Experiment 1, with NWA accessions producing more leaves (8.79) than NEA accessions \n(8.07), while NWA accessions had significantly greater plant height (64.1 cm) compared to \nthose from NEA (56.8 cm)  in Experiment 2.  Other measured parameters, such as total \nchlorophyll content (measured with Dualex®), total biomass, leaf biomass, stem biomass, \nand stem diameter, showed no significant differences  at the Region level  across both \nexperiments (p-values > 0.05). These results suggest that while some differences in plant \nheight and leaf number were observed between the regions, most measured phenotypic \ntraits remained largely unaffected by regional origin. \nNext, variables were analyzed at the Accession level (Tables 1 and 2). Height marginal \nmeans ranged from approximately 34 to 58 cm (average CV 39%) in Experiment 1 and 10 to \n12 cm (average CV 51%) in Experiment 2 (Table 1), with three statistical groups in each \nexperiment. ARZM05118 exhibited the greatest mean height in both experiments, while \nARZM06060 and ARZM03042 showed the lowest. The number of leaves ranged from 7 to 9.2 \nin Experiment 1 and from 9.5 to 11.6 in Experiment 2, with three and two statistical groups, \nrespectively (Table 1). B oth experiments showed small coefficients of variation (<10%). \nTotal chlorophyll (Dualex®) ranged from 26.6 to 47.4 μg/cm² in Experiment 1 and 34.2 to 42.5 \nμg/cm² in Experiment 2 (Table 1). In both experiments, chlorophyll values differed \nmoderately among accessions, with t wo statistical groups each, and CV <35 %. The \nmaximum fold-changes between minimum and maximum values varied between 1.22 and \n1.78 for each trait (Table 1). \nTotal biomass ranged from 26.9 g (ARZM06060) to 103.7 g (ARZM05118) with four statistical \ngroups, although most accessions clustered within the intermediate abc-bcd groups, while \nCV varied between 5 -43% (Table 2) . Leaf biomass ranged from 11.1 -50.4g (CV < 14%, 4 \nstatistical groups) while stem biomass ranged from 8.77 -31.57g (CV < 16%, 6 statistical \ngroups) (Table 2). Consistently with total biomass, ARZM06060 presented the lowest values \nand ARZM05118 the highest. Stem diameter varied less across landraces, ranging from 0.86 \ncm ( ARZM06060) to 1.31 cm  (ARZM08178), with generally low coefficients of variation  \n(<0.74%) and three statistical groups (Table 2) . Overall, the ranking of accessions was \ncomparable across traits: high -biomass landraces (e.g., ARZM05118, ARZM10076, \nARZM08118, ARZM08018) also produced greater leaf and stem biomass and tended to have \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n8 \n \nlarger stem diameters, while low -biomass accessions (e.g., ARZM06060, ARZM08020 , \nARZM03042) were consistently grouped among the lowest performers. This can be \nexplained by the high correlation between these traits (Supplementary figure 2). On average, \ncoefficients of variation were low to moderate for all traits, except for total biomass, \nindicating low variation within accessions.  However, the fold -change between the \nmaximum and minimum values for each trait was consistently high ( range: 1.52-1.78), \nindicating substantial variation among the accessions. \nOverall, although no differences were generally observed at the Region level between NWA \nand NEA groups (Supplementary table 2), morphological analysis showed variability among \naccessions with several significantly different groups for each trait (Tables 1 and 2) . \nAlthough some accessions showed high variability within the accession (such as height in \nExperiment 2, average CV: 51 .16%), most traits showed moderate or low coefficients of \nvariation (<30% on average). It is worth highlighting that for all trait s significant Accession \neffects were observed (p < 0.05), indicating a genetic influence linked to the accessions. \nAccessions Show Large Fold-Changes and High Within-Group Variation in Biochemical \nVariables  \nThe metabolic content of plants influences several physiological processes with \nagronomical impact (Pott et al. 2019) . We measured leaf metabolites involved in primary \nmetabolism that affects plant growth and yield (total protein, total soluble sugars, starch, \npigments) and secondary metabolites involved in general stress responses (total phenolics, \ntotal antioxidant activity). In coincidence with what was observed for morphological traits, \nbiochemical traits showed higher CV in Experiment 1 than in Experiment 2, except for starch \ncontent and antioxidant activity (Supplementary figure 1). \nThe biochemical analysis of accessions from NEA and NWA revealed some regional trends \nin various traits, though none of them were statistically significant (Supplementary table 3). \nAccessions from NWA generally exhibited higher total proteins (1.52 mg/g), total soluble \nsugars (13.6 mg/g) and total carotenoids (0.957 g/g) compared to those from NEA ( 1.45 \nmg/g, 11.2 mg/g and 0.46 g/g, respectively). In contrast, NEA accessions had higher starch \n(11.79 mg/g), total chlorophyll (1.943 g/g) and chlorophyll a (1. 55 g/g) compared to NWA \naccessions (8.53 mg/g, 0.987 g/g  and 0.845 g/g , respectively). Chlorophyll b  and total \ncarotenoid levels were greater in NWA accessions (0.914 g/g  and 0.957 g/g) than in NEA \naccessions (0.495 g/g  and 0.46 g/g ). Total phenolics and antioxidant activity (in both \nhydrophilic and hydrophobic phases ) showed minimal regional variation . Beyond these \ntrends, the variation within each region was considerable  (CV 25-105%), reflecting a high \ndegree of heterogeneity among accessions. These findings suggest that while regional \ndifferences may exist in certain biochemical traits , variability within regions may play a \nsignificant role in shaping these traits. \nNext, variables were compared at the Accession level  (Tables 3a and b). Total proteins \nranged from approximately 2.2 (ARZM05067) to 7.3 mg/g  (ARZM04013), with low \ncoefficients of variation and two statistical clusters (Table 3a). Total soluble sugars varied \nmore widely , from 7.7 (ARZM03042) to 16.2 mg/g  (ARZM05007), while all landraces \nbelonged to a single statistical group (Table 3a). Starch content showed high variability (CV \nfrequently > 70%), with means spanning 6.7 (ARZM08020) to 17.9 mg/g (ARZM06109), while \nno significant differences were observed among accessions  (Table 3a) . Total phenols \nranged from 1.86 (ARZM10076) to 3. 03 mg GA/g (ARZM08020 ), forming two statistical \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n9 \n \ngroups with CV lower than 1.84% (Table 3a). Both measures of antioxidant activity \n(hydrophilic and hydrophobic phases) showed very limited differentiation among \naccessions (Table 3a). Antioxidant activity in hydrophilic phase values ranged  from 0.076 \n(ARZM04029) to 0.104 mg GA/g (ARZM08020), while  antioxidant activity in hydrophobic \nphase ranged from 0.053 (ARZM08020) to 0.146 mg GA/g (ARZM12205), all within a single \nsignificance group , with high CV for hydrophilic phase (>46%) and low CV for the \nhydrophobic pha se measurements (<0.1%) . In contrast, maximum fold -changes when \ncomparing maximum and minimum values for each trait were high (range: 1.37 -3.33), \nindicating high variability between accessions. \nTotal chlorophyll ranged from 1.01 to 2.68 g/g, with ARZM05120 displaying the highest mean \nvalue, forming two statistical clusters (Table 3b) . Several accessions  -including \nARZM08020, ARZM08178, and ARZM12205 , and ARZM05067- showed the lowest total \nchlorophyll levels. For chlorophyll a, values ranged from 0.72 (ARZM08020 and ARZM12205) \nto 2.48 g/g  (ARZM05120), while chlorophyll b ranged from 0.72 (ARZM05067) to 1.15 g/g  \n(ARZM10082) (Table 3b) . Chlorophyll a had three  statistical groups . On the contrary, \naccessions did not show  statistically significant differences  for chlorophyll b. Total \ncarotenoids showed a narrower range -between 0.86 (ARZM05007) and 1.26 g/g  \n(ARZM05118), while accessions did not show statistical differences (Table 3b) . A ll the \nchlorophylls as well as  the carotenoids showed high  CV values (2 4-107%), suggesting \nsubstantial within -accession variability.  Maximum fold -changes when comparing \nmaximum and minimum values for each trait were high ( range: 1.45-3.46), indicating high \nvariability between accessions. Chlorophylls were highly correlated with each other \n(Supplementary figure 2). In turn, they were positively correlated with growth -related \nparameters (biomass, height, number of leaves), possibly due  to their relationship with \nphotosynthesis. \nTotal proteins, total phenols, total chlorophyll, and chlorophyll a exhibited significant \nAccession effects (p < 0.05), indicating a genetic influence linked to the accessions. While \nthe analysis of biochemical traits did not reveal many statistical groups  or clusters -likely \ndue to the limited sample size per accession- there was considerable variability in each trait \nobserved both between and within accessions. Between accessions, the maximum value \nwas as high as ≈3.5-fold higher than the minimum (for chlorophyll a and total proteins) and \nas low as ≈1.5-fold higher than the minimum (for total carotenoids and antioxidant activity \nin hydrophilic phase) (Table 3a and b). Additionally, significant variation was seen within \neach accession, as evidenced by the high CV in some accessions, with starch showing the \nhighest average CV of 96% (Table 3a). This variability between and within accessions might \nhold biological significance.  \nSalt Stress Tolerance Varies by Accessions and Regions \nWith the aim of extending the phenotypic characterization of the 19 accessions, we \nassessed the performance of their seedlings to salt stress, considering that the likelihood \nof soil salinity increases with climate change (Corwin 2021), affecting many crop species \nincluding maize (Farooq et al. 2015; Hu et al. 2022) . When compared at the Region level, \nthe Region × Treatment interaction was non -significant, so Region was averaged across \nTreatment (NaCl concentration) . NWA plants showed higher root weight and length in \ncomparison to NEA plants, while no differences were observed at the leaf level (Figure 2a \nand b). This suggests that there is a difference in the response of seedlings according to \ntheir region of origin. \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n10 \n \nWhen compared at the Accession level, Accession × Treatment interaction was significant \nfor both root and leaf length ; thus, the effect of each level of Treatment was evaluated \nseparately across Accessions (Figure 2c and d). Overall, root length decreased over time \nfrom T0 to T2 across all accessions (Figure 2c). It showed some variation among accessions \nunder control conditions (T0), with mean values ranging from 17.78 to 30.09 cm. Under the \nT1 and T2 treatments, differentiation among accessions was also evident, with mean values \nranging from 8.11 to 17.06 cm (T1) and 3.12 to 6 cm (T2). In all cases, ARZM05069 and \nARZM04029 showed consistently lower values, while ARZM10076, ARZM08178 and \nARZM08096 showed higher values. \nLeaf length also decreased from T0 to T2 across Accessions (Figure 2d). Under control \nconditions (T0), it showed variation among accessions, with mean values ranging from \n22.68 to 31.44 cm. Under treatment T1, accession means ranged from 14.57 to 21.13 cm, \nwith most genotypes not differing significantly. Under treatment T2, values were markedly \nlower, ranging from 1.49 cm  in ARZM 05069 to 8.07 cm  in ARZM05007, with seven \nstatistically significant groups . ARZM05067 appeared consistently among the highest \nmeans. \nAltogether, both T1 and T2 treatments had a marked effect across accessions, significantly \nreducing the measured variable s compared with control conditions (T0). Both T1 and T2 \ntreatments revealed clear differences among accessions  at both root and leaf levels, \nindicating genotype-specific responses. Across both measured variables (Figure 2c and d), \nARZM05069 and ARZM04029 showed lower means at T2, indicating higher sensitivity, while \nARZM05067 and ARZM10082 showed higher means at T2, indicating higher tolerance. When \nconsidering regional and accession differences, the root phenotype seems to be more \nresponsive to the treatments than the leaf phenotype, indicating a more prominent role in \nthe response to salt stress in these plants. \nRegional Differences at the Phenotypic Level Can Be explained by the Altitudinal Cline \nTo explore global differences among regions and accessions, we examined the relationship \nbetween morphological and biochemical traits measured in adult plants (Tables 1, 2, and \n3) using a combination of multivariate analyses. The Principal Components Analysis (PCA) \nrevealed high phenotypic variation among the different maize Accessions and different \nstability of these accessions across the two experiments (Figure  3a). Accessions with the \nhighest overall variability, indicated by a wide dispersion of points and greater distance from \ntheir centroids included ARZM10076, ARZM05007 and ARZM05120 (Figure 3a). Conversely, \npopulations with the lowest variability and highest stability were ARZM08018, ARZM05069 \nand ARZM08020 , which demonstrated consistent performance between the two \nexperiments (Figure 3a) . Regarding explained variance and variables contribution, PC1 \nexplained 31.40% of the variance, mainly influenced by biomass and pigments, while PC2 \nexplained 22.80% of the variance, with pigments and biomass  contributing most as well \n(Figure 3b). The PCA biplot shows that accessions are distributed along both axes, with no \nclear pattern of separation by experiment or region of origin (Figure 3a; Supplementary \nFigure 3). In turn, based on the BIC criterion, the DAPC identified k = 4 as the most probable \nnumber of clusters (Figure 3c, d), and eleven accessions were consistently grouped in both \nexperiments, although no clustering by region was observed in agreement with the PCA. The \ngrouping was highly concordant with the major variation axes identified by the PCA (Figure \n3a, b). Accessions such as ARZM3042 in Experiment 2, and ARZM05120 and ARZM10082 in \nExperiment 1 , we re grouped together into Cluster 1  in the DAPC (Figure 3c, d) and had \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n11 \n \nnegative PC1 scores (Figure 3a, b), characterized by high-moderate values for pigments and \nmoderate-low biomass values (Tables 2, 3b). A second group (Cluster 2 of the DAPC, Figure \n3c and d) including ARZM04013, ARZM05067, ARZM05069, ARZM06109, and ARZM08020, \nwere distinctively defined by moderate-low pigment values and low biomass values (Tables \n2 and 3b) and had positive scores on PC2 of the PCA (Figure 3a, b). In contrast, accessions \nARZM04029, ARZM05118, ARZM08096, and ARZM8178 were grouped in Cluster 3  of the \nDAPC (Figure 3c, d), had positive scores on PC1 of the PCA (Figure 3a, b), and were \ncharacterized by high-moderate biomass and moderate-low pigment values (Tables 2 and \n3b). Finally, populations including ARZM08018 and ARZM10076 were grouped into Cluster \n4 of the DAPC (Figure 3c, d) and had negative scores on the PC2 of the PCA (Figure 3a, b), \nbased on their high biomass values, and moderate pigment content (Tables 2 and 3b). \nHierarchical cluster analysis , which was performed on marginal means from univariate \nlinear models, identified five groups as the optimal, according to the average silhouette \ncriterion (Figure 4a). Group 1 comprised accessions ARZM10076 and ARZM05118; Group 2 \nincluded ARZM04060, ARZM04013, and ARZM05120; Group 3 contained ARZM04029, \nARZM03042, and ARZM05067; Group 4 consisted solely of ARZM08020; and Group 5 \nencompassed the remaining ten accessio ns. No evident and strong clustering pattern \nassociated with the region of origin was observed. Pairwise distances highlighted that these \ngroups display marked internal similarity and strong divergence from the others (Figure 4b). \nGroup 4 (ARZM08020) was the most divergent, which showed high distance values relative \nto nearly all other accessions. Additionally,  groups 1 and 2 also appeared highly distinct, \nexhibiting very low internal distances but strong divergence from the other accessions. In \ncontrast, the larger group containing the remaining ten accessions (Group 5) displayed \nmore moderate and mixed distance values, indicating some internal structure.  \nThe clustering similarity between the dendrogram and the PCA/DAPC results was \nmoderate, with a stronger alignment to the PCA outcomes  (Figures 3, 4) . Given that the \nclustering was based on marginal means from univariate linear models, whereas the \nPCA/DAPC analyses used raw means, this suggests that the observed differences between \nunits at the multivariate level are, at least in part, influenced by their varying responses to \nthe environment. \nRedundancy Analysis (RDA) used for the association between multivariate phenotypic \nvariation and geographic predictors revealed an altitudinal pattern. The first RDA model \nincluding Altitude as the sole predictor showed that elevation accounted for a signi ficant \nportion of multivariate phenotypic variation (13.5%, F = 2.49, p = 0.02). This indicated that \nAltitude alone captures a detectable ecological gradient associated with trait differentiation \namong populations. The full model including Altitude, Latitude, and Longitude, explained \n~25.3% of total variance, and only Altitude explained a significant  portion of the trait \nvariation (sequential test: F = 2.53, p = 0.029), whereas Latitude and Longitude showed no \nsignificant effects ( p > 0.29). Marginal tests revealed that the three variables capture \noverlapping spatial structure and cannot be statistically disentangled due to spatial  \nautocorrelation and collinearity. The differentiation is associated with Altitude, but it is \npartly a consequence of the spatial structure (Lat itude/Longitude). The ordination biplot \nreflected this pattern (Figure 5), with a strong axis of variation aligned with Altitude. \nAccessions occurring at higher elevations (ARZM8096, ARZM8018, ARZM10082, \nARZM12205), all from NWA, were displaced toward the positive end of RDA1, showing a \nclear separation from NEA low‐elevation populations, which cluster on the opposite side of \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n12 \n \nthe axis. Latitude and Longitude vectors had lower contribution relative to Altitude. High -\naltitude samples (positive RDA1 scores) were associated with higher concentrations of total \nsugars, total phenolics, carotenoids, and chlorophyll b (Supplementary ta ble 4). In \ncontrast, low-altitude samples (negative RDA1 scores) were characterized by higher levels \nof chlorophyll a, total chlorophyll, starch, and hydrophilic antioxidant capacity \n(Supplementary table 4). This aligns with tendencies in univariate analyses (Supplementary \ntable 3). \nIn conclusion, more than half of the accessions showed consistency in phenotypic \ncharacteristics across Experiments (Figure 3, 4), while the variables that contributed most \nto distinguishing the accessions were biomass and pigment content (Figure 3). Although the \nmultivariate analyses did not reveal a clear regional separation (Figures 3c, 4a, and \nSupplementary figure 3), the  RDA, which examined the contribution of environmental \nvariables (Figure 5), indicated that altitude significantly influenced the phen otypic \ncharacteristics of the accessions. Therefore, it can be concluded that phenotypic variation \namong accessions followed an altitudinal gradient, driven largely by differences in pigment \ncomposition. \n \nDiscussion \nStudying population diversity is essential for the conservation and sustainable use of crop \ngermplasm. In this work, we analyzed the morphological, biochemical, and salinity -stress \ntolerance variability of 19 native maize accessions from Northwestern and Northeastern \nArgentina, the southernmost regions in the world where maize  landraces are cultivated, \nwith the aim of providing information on agronomically relevant traits for maize breeding \nand conservation programs. \nMaize Landrace Accessions as a Source of Phenotypic Variation \nBoth univariate and multivariate statistical analyses revealed high variability among the 19 \naccessions evaluated in terms of morphology, biochemistry, and salt stress tolerance \n(Tables 1-3, Figures 2-4). The phenotypic variability observed is likely a reflection of the high \ngenetic variability present in native maize varieties from Argentina (Bracco et al. 2016; Rivas \net al. 2022; Dominguez et al. 2024a) and the Americas more broadly (Hufford et al. 2012). \nThe origin of this extensive genetic variability in maize  landraces can be attributed to \nmultiple biocultural factors, including germplasm exchange among farmers, crosses \nbetween different local varieties due to their open -pollinated nature, farmer selection to \nmeet quality and diversity requirements, adaptation to diverse climatic conditions, and \ngene flow with wild relatives (teosintes) in Mesoamerica (Camacho Villa et al. 2005; Arteaga \net al. 2016; Guzzon et al. 2021).  \nThe fact that the univariate analyses conducted in this study showed the accession factor \nto be statistically significant for all morphological variables and for several biochemical \nvariables further supports the idea that the observed phenotypic variabili ty has a genetic \nbasis associated with each accession. This finding highlights the potential use of these \naccessions in crop improvement, which is particularly relevant given that commercial \nvarieties harbor only between 5 and 10% of the available genetic variability in maize \n(Hoisington et al. 1999). In this regard, numerous examples underscore the importance of \nincreasing the genetic diversity in commercial maize, such as the impact of fall armyworm \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n13 \n \n(FAW; Spodoptera frugiperda) following the development of resistance to insecticides in \nsome maize lines (Blanco et al. 2016) , as well as the severe impact of drought on maize \nproduction in the United States in 2012 (Boyer et al. 2013) . Nevertheless, despite their \npotential as sources of novel genetic variation and agronomically valuable traits, maize \nlandraces have not been widely incorporated into breeding programs (Dominguez et al. \n2024b). Maize landraces from other regions of the Americas have been used to improve \nyield under drought conditions (Medeiros Barbosa et al. 2021)  and increase resistance to \nthe tar spot complex (Willcox et al. 2022). Together, these studies highlight the importance \nof introducing the diversity present in landraces into breeding populations and emphasize \nthe need to characterize the phenotypic variability of native maize as a resource for crop \nimprovement. \nMany of the variables measured in this study, for which differences among accessions were \ndetected both statistically and in terms of magnitude as measured by fold-changes, have \nhigh agronomic relevance and could be of interest for breeding programs. At the phenotypic \nlevel, a larger stem diameter may be desirable for increasing water and nutrient transport \ncapacity or for providing greater mechanical resistance, thereby reducing the risk of lodging \n(García et al. 2001). Leaf number and biomass are traits associated with the photosynthetic \ncapacity of maize (Mehta and Sarkar 1992) and its potential use as forage or for bioenergy \nproduction (Klopfenstein et al. 2013) . At the biochemical level, phenolic compounds play \nimportant roles in responses to both biotic and abiotic stress (Tak and Kumar 2021). Sugars, \nproteins, and starch influence leaf quality in maize used as a forage species  (Liimatainen \net al. 2022), while starch also plays a key role in seed germination and early seedling growth \nuntil photosynthetic maturity is reached, among other functions (Li et al. 2025). In addition, \nsalinity tolerance has important implications for food security, not only in regions with \nsaline soils  (Egea et al. 2023)  but also in relation to other stress conditions, given the \nextensive crosstalk between stress signaling pathways (Mishra et al. 2016; Kesawat et al. \n2023). \nOverall, the accessions exhibited high phenotypic variability across nearly all measured \nvariables, with each accession displaying a unique combination of traits, as well as \nsubstantial variability within accessions (Tables 1 -3). These characteristics demonstrate \nthe considerable potential of these accessions as a source of phenotypic variation of \nagronomic interest and suggest that these populations may be capable of responding to a \nwide range of environmental pressures and future challenges. \nAltitudinal Cline as a Key Variable Driving Differences Between Regions \nAlthough no significant differences were detected when variables were compared at the \nregional level (NWA vs. NEA) (Supplementary Tables 2 and 3), the RDA indicated that \nmultivariate phenotypic variability among accessions could be explained by altitude (Figure \n5). All accessions from the NEA region originate from altitudes below 345 m.a.s.l., whereas \nall accessions from the NWA region originate from altitudes above 782 m .a.s.l., with one \nexception (accession ARZM10076, originating at 322 m .a.s.l.) (Supplementary Table 1). \nTherefore, the phenotypic differentiation between regions appears to be driven primarily by \naltitude. RDA may have been more effective than linear mixed models  in detecting these \nphenotypic differences because it captures the gradual altitudinal cline, whereas the linear \nmixed models may have overlooked this more subtle environmental effect by treating region \nas a categorical variable. \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n14 \n \nPhenotypic differences, both morphological and biochemical, within and among species \nas a function of altitude have long been documented (Körner 1989; Terashima et al. 1995; \nSundqvist et al. 2013; Ratier Backes et al. 2022) . In maize, numerous studies report \nphenotypic differences -most often morphological -associated with altitude in Mexican \n(Pace et al. 2024), African (Asare et al. 2016), Brazilian (Ribeiro e Souza et al. 2008), and \nNorth and South American landraces more broadly (Romero -Navarro et al. 2017 ; López-\nValdivia et al. 2025). A comprehensive overview of these studies is provided in Salve et al. \n(2023a). Notably, Janzen et al. (2022) identified altitude-associated phenotypic traits, such \nas increased anthocyanin content in highland maize from Mexico, as evidence of local \nadaptation, suggesting that altitudinally structured phenotypic variation may contribute to \nadaptive processes. \nIn Argentina, fewer studies have addressed altitude-related variation in maize, but existing \nwork has reported morphological ( Salve et al. 2023b) and cytological (Realini et al. 2018)  \ndifferences along altitudinal gradients. In our study, the RDA showed that several \nphenotypical traits contributed to the differentiation of the accessions including pigments \nand antioxidants. Because altitude-related phenotypic differences are driven by factors \nsuch as temperature, precipitation, solar irradiation, and other environmental variables  \n(Sundqvist et al. 2013) , a more detailed analysis of the relationships between phenotypic \ntraits and specific environmental factors may reveal additional, previously undetected \npatterns in these populations. \nMaize Landraces as a Source of Salinity Stress Tolerance \nNearly half of the NWA individuals were collected from sites with Entisol soils, which are \ncharacterized by salinity in the upper 50 cm, while the remaining accessions originated from \nMollisol soils, which can become saline under certain environmental cond itions (Lavado \n2007) (Supplementary Table 1). The observed association between regions of origin and \ngreater tolerance to salt stress (Figure 2a and b ) suggests that geographic origin may be \nlinked to a process of local adaptation. With respect to their phenotype, NWA accessions \nexhibited longer and heavier roots (Figure 2a and b ), a pattern consistent with previous \nreports showing that salt -tolerant maize often develops more extensive root systems \n(Farooq et al. 2015; Hu et al. 2022).  \nPratt et al. (202 2) evaluated 11 maize landraces from the Southwestern United States \noriginating from regions with saline soils and identified one landrace that maintained high \nyield under saline field conditions. Together with our results, this finding suggests that \nnative maize landraces represent a promising source of salinity stress tolerance and that \nfurther evaluation under field conditions, including comparisons with commercial lines, is \nwarranted. In addition, tolerance to saline stress is known to exhibit substantial crosstalk \nwith other stress -response pathways through shared signaling mechanisms, particularly \nhormonal pathways (such as abscisic acid signaling) and antioxidant responses, which can \nconfer resistance to multiple stressors (Mishra et al. 2016; Kesawat et al. 2023 ). This \ninteraction suggests the presence of additional stress tolerances that could be evaluated in \nfuture studies using these same accessions. \nGiven the increasing extent of soil salinization worldwide (Egea et al. 2023) , together with \nthe progression of climate change, the phenotypic variation observed in these landraces \nmay be of considerable value for the development of maize varieties tolerant to salinity and \npotentially to other abiotic stresses. \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n15 \n \n \nConclusions \nPhenotypic variability, including diversity in biochemical composition and stress tolerance, \nacquires strategic relevance when correlated with genotype, as it contributes to the \nsustainability and resilience of agricultural systems in the face of climate change and other \nglobal challenges (Guzzon et al. 2021; Teixeira et al. 2021). The 19 maize accessions from \nNorthern Argentina evaluated in this study exhibited high variability across the assessed \ntraits at the individual level (Tables 1 -3). The combined u se of univariate and multivariate \nanalyses revealed that each accession possesses a unique combination of morphological, \nbiochemical, and salinity stress tolerance traits, with overall phenotypic differentiation \nobserved among accessions associated with the altitude of collection. \nThese results underscore the importance of considering the high variability among \naccessions and the singularity of each accession when designing both in situ and ex situ \nconservation strategies. Such an approach is essential not only for addressing agronomic \nchallenges but also for preserving the cultural heritage associated with maize landraces. \nMoreover, the phenotypic diversity documented here provides a valuable foundation for \nfuture studies, including quantitative trait locus (QTL) mapping and genome -wide \nassociation studies (GWAS), aimed at identifying genetic loci linked to agronomically \nimportant traits. \n \nConflicts of Interest \nThe authors declare no conflicts of interest. \n \nData Availability Statement \nAll data supporting the findings of this study are included in the main manuscript, available \nin the Supplementary materials, or can be obtained directly from the corresponding author \nupon request. \n \nFunding  \nThis research was funded by the Fondo para la Investigación Científica y Tecnológica \n(FONCYT), grants PICT 2020 -1790 and PICT 2021 -1286, and the Instituto Nacional de \nTecnología Agropecuaria (INTA), grant 2023-PD-L01-I085. \n \nReferences \nAceituno FJ, Loaiza N (2014) Early and Middle Holocene evidence for plant use and \ncultivation in the Middle Cauca River Basin, Cordillera Central (Colombia). Quat Sci \nRev 86:49–62. http://dx.doi.org/10.1016/j.quascirev.2013.12.013 \nAinsworth EA, Gillespie KM (2007) Estimation of total phenolic content and other oxidation \nsubstrates in plant tissues using Folin – Ciocalteu reagent. Nat Protoc 2:875 –877. \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n16 \n \nhttps://doi.org/10.1038/nprot.2007.102 \nAnderson E, Cutler H (1942) Races of Zea mays: I. Their recognition and classification. Ann \nMissouri Bot Gard 29:69–88. https://doi.org/10.2307/2394331.  \nAndorf C, Beavis WD, Hufford M, Smith S, Suza WP, Wang K, Woodhouse M, Yu J, \nLübberstedt T (2019) Technological advances in maize breeding : past , present and \nfuture. Springer Berlin Heidelberg. https://doi.org/10.1007/s00122-019-03306-3 \nArteaga MC, Moreno -Letelier A, Mastretta -Yanes A, Vázquez -Lobo A, Breña -Ochoa A, \nMoreno-Estrada A, Eguiarte LE, Piñero D (2016) Genomic variation in recently \ncollected maize landraces from Mexico. Genomics Data 7:38 –45. \nhttp://dx.doi.org/10.1016/j.gdata.2015.11.002 \nAsare S, Tetteh AY, Twumasi P, Adade KB, Akromah RA (2016) Genetic diversity in lowland, \nmidaltitude and highland African maize landraces by morphological trait evaluation. \nAfrican J Plant Sci 10:246–257. https://doi.org/10.5897/AJPS2016.1448 \nBates D, Mächler M, Bolker BM, Walker SC (2015) Fitting linear mixed-effects models using \nlme4. J Stat Softw 67:1–48. \nBlanco CA, Chiaravalle W, Rodrı JC, Tera H (2016) Current situation of pests targeted by Bt \ncrops in Latin America. Curr Opin Insect Sci 15:131 –138. \nhttps://doi.org/10.1016/j.cois.2016.04.012 \nBonavia D (2013) Maize. Origin, domestication, and its role in the development of culture. \nCambridge University Press, Cambridge, UK. \nBoyer JS, Byrne P, Cassman KG, Cooper M, Delmer D, Greene T, Gruis F, Habben J, \nHausmann N, Kenny N, La R, Paszkiewicz S, Porter D, Schlegel A, Schussler J, Setter \nT, Shanahan J, Sharp RE, Vyn TJ, Warner D, Gaffney J (2013) The U.S. drought of 2012 \nin pers pective : A call to action. Glob Food Sec 2:139 –143. \nhttps://doi.org/10.1016/j.gfs.2013.08.002 \nBracco M, Cascales J, Hernández JC, Poggio L, Gottlieb AM, Lia V V (2016) Dissecting maize \ndiversity in lowland South America: Genetic structure and geographic distribution \nmodels. BMC Plant Biol 16. https://doi.org/10.1186/s12870-016-0874-5 \nBracco M, Lia V V., Hernández JC, Poggio L, Gottlieb AM (2012) Genetic diversity of maize \nlandraces from lowland and highland agro -ecosystems of Southern South America: \nImplications for the conservation of native resources. Ann Appl Biol 160:308 –321. \nhttps://doi.org/10.1073/pnas.96.11.5937 \nBradford MA (1976) Rapid and sensitive method for the quantitation of microgram \nquantities of protein utilizing the principle of protein-dye binding. Anal Biochem \n72:248–254. https://doi.org/10.1016/0003-2697(76)90527-3 \nBrooks ME, Kristensen K, Benthem KJ van, Magnusson A, Berg CW, Nielsen A, Skaug HJ, \nMächler M, Bolker BM (2017) glmmTMB balances speed and flexibility among \npackages for zero-inflated generalized linear mixed modeling. R J 9:378–400. \nCamacho Villa T, Maxted N, Scholten M, Ford -Lloyd B (2005) Defining and identifying crop \nlandraces. Plant Genet Resour 3:1321–1332. https://doi.org/10.1079/PGR200591 \nCámara Hernández JC, Miante Alzogaray AM, Bellón R, Galmarini AJ (2012) Razas de maíz \nnativas de la Argentina. Editorial Facultad de Agronomía, Universidad de Buenos \nAires. \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n17 \n \nCorwin DL (2021) Climate change impacts on soil salinity in agricultural areas. Eur J Soil Sci \n72:842–862. https://doi.org/10.1111/ejss.13010. \nDas AK, Kumar S, Vignesh J, Rajkumar M, Zunjare U (2019) Molecular diversity and genetic \nvariability of kernel tocopherols among maize inbreds possessing favourable \nhaplotypes of c-tocopherol methyl transferase (ZmVTE4). J Plant Biochem Biotechnol \n28:253–262. https://doi.org/10.1007/s13562-018-0470-x \nDefacio R, Paz N, Schlater A, Ferrer M, Bramardi S (2025) Agro -morphological and \nmolecular characterization of Argentine maize ( Zea mays L.) landraces of ‘Cristalino \nColorado’ race. Genet Resour 6:14–25. https://doi.org/10.46265/genresj.TSJG3884 \nDominguez PG, Freilij D, Lia VV (2024 a) A genomic journey across the past, present, and \nfuture of South American maize. J Exp Bot 75:7006 –7018. \nhttps://doi.org/10.1093/jxb/erae383 \nDominguez PG, Gutierrez AV, Fass MI, Filippi CV, Vera P, Puebla A, Defacio RA, Paniego NB, \nLia VV (2024b) Genome-wide diversity in lowland and highland maize landraces from \nSouthern South America: population genetics insights to assist conservation. Evol \nAppl 17:e70047. https://doi.org/10.1111/eva.70047 \nEgea I, Estrada Y, Faura C, Bolarin MC, Flores FB, Mar A (2023) Salt -tolerant alternative \ncrops as sources of quality food to mitigate the negative impact of salinity on \nagricultural production. Front Plant Sc 14:1 –17. \nhttps://doi.org/10.3389/fpls.2023.1092885 \nFarooq M, Hussain M, Wakeel A, Siddique KHM (2015) Salt stress in maize : Effects, \nresistance mechanisms, and management. A review. Agron Sustain Dev 35:461–481. \nhttps://doi.org/10.1007/s13593-015-0287-0 \nGarcía M, Watson C, Salcedo F (2001) Evaluación de métodos para determinar resistencia \nal acame de raíces en maíz dulce (Zea mays L.). Bioagro 13:22–31. \nGeiler-Samerotte K, Bauer C, Li S, Ziv N, Gresham D, Siegal M (2013) The details in the \ndistributions: Why and how to study phenotypic variability. Curr Opin Biotechnol \n24:752–759. https://doi.org/10.1016/j.copbio.2013.03.010 \nGuzzon F, Arandia Rios LW, Caviedes Cepeda GM, Céspedes Polo M, Chavez Cabrera A, \nMuriel Figueroa J, Medina Hoyos AE, Jara Calvo TW, Molnar TL, Narro León LA, Narro \nLeón TP, Mejía Kerguelén SL, Ospina Rojas JG, Vázquez G, Preciado -Ortiz RE, \nZambrano JL, Pa lacios Rojas N, Pixley K V. (2021) Conservation and use of Latin \nAmerican maize diversity: Pillar of nutrition security and cultural heritage of humanity. \nAgronomy 11:172. https://doi.org/10.3390/agronomy11010172 \nHe F (2011) Bradford protein assay. Bio-Protocol 1:4 –5. \nhttps://doi.org/10.21769/BioProtoc.45 \nHeck MI, Defacio RA, Ferrer ME, Cirilo AG, Fariza SI, Lucia AD De, Blaszchik JA (2020) \nEvaluación de la variabilidad agromorfológica de poblaciones nativas de maíz de \nMisiones, Argentina. RECyT:6–13. \nHeck M, Defacio R, Ferrer M, Cirilo A, Fariza S, De Lucia A, Blaszchik J (2019) Evaluación de \nla calidad nutricional de variedades nativas de maíz de Misiones, Argentina. Rev \nInvestig la Fac Ciencias Agrar - UNR 34. https://doi.org/10.35305/agro34.266 \nHoisington D, Khairallah M, Reeves T, Ribaut JM, Skovmand B, Taba S, Warburton M (1999) \nPlant genetic resources: What can they contribute toward increased crop \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n18 \n \nproductivity? Proc Natl Acad Sci U S A 96:5937 –5943. \nhttps://doi.org/10.1073/pnas.96.11.5937 \nHu D, Li R, Dong S, Zhang J, Zhao B, Ren B, Ren H, Yao H, Wang Z, Liu P (2022) Maize ( Zea \nmays L.) responses to salt  stress in terms of root anatomy , respiration and \nantioxidative enzyme activity. BMC Plant Biol 22:1 –17. \nhttps://doi.org/10.1186/s12870-022-03972-4 \nHufford MB, Xu X, Van Heerwaarden J, Pyhäjärvi T, Chia JM, Cartwright RA, Elshire RJ, \nGlaubitz JC, Guill KE, Kaeppler SM, Lai J, Morrell PL, Shannon LM, Song C, Springer \nNM, Swanson-Wagner RA, Tiffin P, Wang J, Zhang G, Doebley J, McMullen MD, Ware D, \nBuckler ES, Yang S, Ross -Ibarra J (2012) Comparative population genomics of maize \ndomestication and improvement. Nat Genet 44:808 –811. \nhttps://doi.org/10.1038/ng.2309 \nJanzen GM, Aguilar -Rangel MR, Cíntora -Martínez C, Blöcher -Juárez KA, González-Segovia \nE, Studer AJ, Runcie DE, Flint -Garcia SA, Rellán -Álvarez R, Sawers RJH, Hufford MB \n(2022) Demonstration of local adaptation in maize landraces by reciprocal \ntransplantation. Evol Appl 15:817–837. https://doi.org/10.1111/eva.13372 \nJombart T (2008) Adegenet: A R package for the multivariate analysis of genetic markers. \nBioinformatics 24:1403–1405. https://doi.org/10.1093/bioinformatics/btn129 \nJombart T, Devillard S, Balloux F (2010) Discriminant analysis of principal components: A \nnew method for the analysis of genetically structured populations. BMC Genet 11:94. \nhttps://doi.org/10.1186/1471-2156-11-94 \nKassambara A, Mundt F (2020) factoextra: Extract version, visualize the results of \nmultivariate data analyses. R package. https://cran.r-project.org/package=factoextra \nKesawat MS, Satheesh N, Kherawat BS, Kumar A, Kim H, Chung S, Kumar M (2023) \nRegulation of reactive oxygen species during salt stress in plants and their crosstalk \nwith other signaling molecules — Current perspectives and future directions. Plants \n12. https://doi.org/10.3390/plants12040864 \nKlopfenstein TJ, Erickson GE, Berger LL (2013) Maize is a critically important source of food, \nfeed, energy and forage in the USA. F Crop Res 153:5 –11. \nhttp://dx.doi.org/10.1016/j.fcr.2012.11.006 \nKörner C (1989) The nutritional status of plants from high altitudes : A worldwide \ncomparison. Oecologia 89:379–391. https://doi.org/10.1007/BF00377088 \nKumar A, Kumari J, Rana JC, Chaudhary DP, Kumar R, Singh H, Singh TP, Dutta M (2015) \nDiversity among maize landraces in North West Himalayan region of India assessed \nby agro -morphological and quality traits . Indian J Genet 75:188 –195. \nhttps://doi.org/10.5958/0975-6906.2015.00029.2 \nLaurentin A, Edwards CA (2003) A microtiter modification of the anthrone -sulfuric acid \ncolorimetric assay for glucose -based carbohydrates. Anal Biochem 315:143 –145. \nhttps://doi.org/10.1016/S0003-2697(02)00704-2 \nLavado RS (2007) Visión sintética de la distribución y magnitud de los suelos afectados por \nsalinidad en la Argentina. In: Taleisnik E, Grunberg K, Santa Maria G (eds) La \nsalinización de suelos en la Argentina. Editorial de la Universidad Católica de \nCórdoba, Córdoba, Argentina, pp 1–6. \nLenth R, Piaskowski J (2025) emmeans: Estimated Marginal Means, aka Least -Squares \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n19 \n \nMeans. https://rvlenth.github.io/emmeans/ \nLi C, Ding Z, Li E, Xu R, Lv C, Zhang C, Huang L, Gilbert RG (2025) The molecular structure \nof leaf starch from three cereal crops. Carbohydr Polym 351. \nhttps://doi.org/10.1016/j.carbpol.2024.123099 \nLichtenthaler H (1987) Chlorophylls and carotenoids: Pigments of photosynthetic \nbiomembranes. In: Packer L, Douce R (eds) Methods in Enzymology. Academic Press, \npp 350–382. https://doi.org/10.1016/0076-6879(87)48036-1 \nLiimatainen A, Sairanen A, Jaakkola S, Kokkonen T, Kuoppala K (2022) Yield, quality and \nnitrogen use of forage maize under different nitrogen application rates in two boreal \nlocations. Agronomy 12:1–13. https://doi.org/10.3390/agronomy12040887 \nLiu J, Fernie AR, Yan J (2020) The past, present, and future of maize improvement: \ndomestication, genomics, and functional genomic routes toward crop enhancement. \nPlant Commun 1:100010. https://doi.org/10.1016/j.xplc.2019.100010 \nLopez-valdivia I, Rangarajan H, Vallebueno -estrada M, Lynch JP (2025) Broad \nenvironmental adaptation is associated with root anatomical phenotypes in maize \nlandraces : An in silico study. Ann Bot mcaf179:1 –17. \nhttps://doi.org/10.1093/aob/mcaf179 \nLópez MG, Fass M, Rivas JG, Carbonell-Caballero J, Vera P, Puebla A, Defacio R, Dopazo J, \nPaniego N, Hopp HE, Lia VV (2021) Plastome genomics in South American maize \nlandraces: Chloroplast lineages parallel the geographical structuring of nuclear gene \npools. Ann Bot 128:115–125. https://doi.org/10.1093/aob/mcab038 \nMaechler M, Rousseeuw P, Struyf A, Hubert M, Hornik K (2022) cluster: Cluster analysis \nbasics and extensions. https://cran.r-project.org/web/packages/cluster/index.html \nMansilla P, Bongianino N, Nazar M, Perez G (2021) Agronomic and chemical description of \nopen-pollinated varieties of opaque -2 and purple maize ( Zea mays  L.) adapted to \nsemiarid region of Argentina. Genet Resour Crop Evol. \nhttps://doi.org/10.1007/s10722-021-01133-4 \nMatsuoka Y, Vigouroux Y, Goodman MM, G JS, Buckler E, Doebley J (2002) A single \ndomestication for maize shown by multilocus microsatellite genotyping. Proc Natl \nAcad Sci U S A 99:6080–6084. https://doi.org/10.1073/pnas.052125199 \nMedeiros Barbosa PA, Fritsche -Neto R, Andrade MC, Petroli CD, Burgueño J, Galli G, \nWillcox MC, Sonder K, Vidal -Martínez VA, Sifuentes -Ibarra E, Molnar TL (2021) \nIntrogression of maize diversity for drought tolerance: Subtropical maize landraces as \nsource of new positive variants. Front Plant Sci 12:1 –19. \nhttps://doi.org/10.3389/fpls.2021.691211 \nMehta H, Sarkar K (1992) Heterosis for leaf photosynthesis, grain yield and yield \ncomponents in maize. Euphytica 61:161–168. https://doi.org/10.1007/BF00026807 \nMelchiorre P, Bartoloni N, Cámara Hernández JA (2017) Relaciones fenéticas entre razas \nde maíz (Zea mays ssp. mays) nativas de la provincia de Jujuy (Argentina). Boletín de \nla Soc Argentina Botánica 52:717–735. \nMishra S, Kumar S, Saha B, Awasthi J, Dey M, Panda S, Sahoo L (2016) Crosstalk between \nsalt, drought, and cold stress in plants: Toward genetic engineering for stress \ntolerance. In: Tuteja N, Gill S (eds) Abiotic stress response in plants. Wiley -VCH, \nWeinheim. https://doi.org/10.1002/9783527694570.ch4 \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n20 \n \nMladenović Drinić, S Andjelkovic V, Ignjatovic D (2012) Genetic diversity of maize landraces \nas sources of favorable traits. In: Çalışkan M (ed) The molecular basis of plant genetic \ndiversity. InTech, London, pp 89–112. https://doi.org/10.5772/33057 \nOksanen J, Simpson G, Blanchet F, Kindt R, Legendre P, Minchin P, O’Hara R, Solymos P, \nStevens M, Szoecs E, Wagner H, Barbour M, Bedward M, Bolker B, Borcard D, Carvalho \nG, Chirico M, De Caceres M, Durand S, Evangelista H, FitzJohn R, Friendly M, Furneaux \nB, Hannigan G, Hill M, Lahti L, McGlinn D, Ouellette M, Ribeiro Cunha E, Smith T, Stier \nA, Ter Braak C, Weedon J (2022) vegan: Community Ecology Package. https://cran.r -\nproject.org/package=vegan \nPace BA, Perales HR, Id NG, Id LM (2024) Physiological traits contribute to growth and \nadaptation of Mexican maize landraces. PLoS One 19:e0290815. \nhttp://dx.doi.org/10.1371/journal.pone.0290815 \nPanigatti JL (2010) Argentina: 200 años, 200 suelos. Ediciones INTA, Buenos Aires. \nPinheiro J, Bates D, Team RDC (2025) nlme: Linear and Nonlinear Mixed Effects Models. \nhttps://cran.r-project.org/package=nlme \nPiperno DR (2011) The origins of plant cultivation and domestication in the New World \ntropics. Patterns, process, and new developments. Curr Anthropol 52. \nhttps://doi.org/10.1086/659998 \nPiperno DR, Ranere AJ, Holst I, Iriarte J, Dickau R (2009) Starch grain and phytolith evidence \nfor early ninth millennium B.P. maize from the Central Balsas River Valley, Mexico. \nProc Natl Acad Sci U S A 106:5019–5024. https://doi.org/10.1073/pnas.0812525106 \nPott DM, Osorio S, Vallarino JG (2019) From central to specialized metabolism: An overview \nof some secondary compounds derived from the primary metabolism for their role in \nconferring nutritional and organoleptic characteristics to fruit. Front Plant Sci 1 0. \nhttps://doi.org/10.3389/fpls.2019.00835 \nPrasanna BM, Palacios-rojas N, Hossain F, Muthusamy V, Menkir A, Dhliwayo T, Ndhlela T, \nVicente FS (2020) Molecular breeding for nutritionally enriched maize: Status and \nprospects. Front Genet 10:1–16. https://doi.org/10.3389/fgene.2019.01392 \nPratt RC, Montgomery R, Velasco -cruz C, Grant L, Darapuneni M (2022) Southwest -\nadapted maize germplasm as a potential genetic resource for selection of salinity \ntolerant cultivars. Crop Sci 62:286–300. https://doi.org/10.1002/csc2.20654 \nRatier Backes A, Römermann C, Alexander JM, Ramón J, Petr A, Padrón - MA, Trogisch S, \nHaider S (2023) Mechanisms behind elevational plant species richness patterns \nrevealed by a trait -based approach. J Veg Sci 34:1 –15. \nhttps://doi.org/10.1111/jvs.13171 \nRe R, Pellegrini N, Proteggente A, Pannala A, Yang M, Rice -Evans C (1999) Antioxidant \nActivity applying an improved ABTS radical cation decoloration assay. Free Radic Biol \nMed 26:1231–1237. https://doi.org/10.1016/s0891-5849(98)00315-3%09 \nRealini MF, Poggio L, Hernández JC, González GE (2018 ) Exploring karyotype diversity of \nArgentinian guaraní maize landraces: relationship among South American maize. \nPLoS One 13:1–18. https://doi.org/10.1371/journal.pone.0198398 \nRealini MF, Poggio L, María A, Alzogaray M, González GE (2023) Morphological and \nphenological variability of Guaraní maize germplasm from Northeastern Argentina \n(NEA) Rodriguésia 74:e00292022.2023. https://doi.org/10.1590/2175 -\n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n21 \n \n7860202374012%09 \nReynolds T, Nemeth M, Glenn K, Ridley W, Astwood J (2005) Natural variability of \nmetabolites in maize grain: Differences due to genetic background. J Agric Food Chem \n53:10061–10067. https://doi.org/10.1021/jf051635q%09 \nRibeiro e Souza AR, Vieira Miranda G, Gonzaga Pereira M, Vagno de Souza L, Lopes Ferreira \nP (2008) Agronomic performance of white maize landrace in different environmental \nconditions. Rev Ceres 55:497–503. \nRivas JG, Gutierrez AV, Defacio RA, Schimpf J, Vicario AL, Hopp HE, Paniego NB, Lia VV \n(2022) Morphological and genetic diversity of maize landraces along an altitudinal \ngradient in the Southern Andes. PLoS One 17:e0271424. \nhttp://dx.doi.org/10.1371/journal.pone.0271424 \nRomero Navarro JA, Willcox M, Burgueño J, Romay C, Swarts K, Trachsel S, Preciado E, \nTerron A, Delgado HV, Vidal V, Ortega A, Banda AE, Montiel NOG, Ortiz -Monasterio I, \nVicente FS, Espinoza AG, Atlin G, Wenzl P, Hearne S, Buckler ES (2017) A study of \nallelic diversity underlying flowering -time adaptation in maize landraces. Nat Genet \n49:476–480. http://dx.doi.org/10.1038/ng.3784 \nSalve DA, Ferreyra MJ, Defacio RA, Maydup ML, Lauff DB, Tambussi EA, Antonietta M \n(2023a) Technology in Agronomy Andean maize in Argentina: Physiological effects \nrelated with altitude, genetic variation, management practices and possible avenues \nto improve yield. Technol Agron 3. https://doi.org/10.48130/TIA-2023-0014 \nSalve DA, Maydup ML, Salazar GA, Tambussi EA, Antonietta M (2023 b) Canopy \ndevelopment, leaf traits and yield in high -altitude Andean maize under contrasting \nplant densities in Argentina. Exp Agric 59:1 –18. \nhttps://doi.org/10.1017/S0014479723000194 \nSampietro DA, Fauguel CM, Vattuone MA, Presello DA, Catalán CAN (2013) \nPhenylpropanoids from maize pericarp : Resistance factors to kernel infection and \nfumonisin accumulation by Fusarium verticillioides. Eur J Plant Pathol 135:105 –113. \nhttps://doi.org/10.1007/S10658-012-0069-3 \nSantiago R, Malvar R, Revilla P, Butrón A (2025) Maize landraces as useful donors of genetic \ndiversity for resilience to drought. J Agric Food Res 23. \nhttps://doi.org/10.1016/j.jafr.2025.102297 \nSmith JS (2017) Ensuring the genetic diversity of maize and its wild relatives. In: Watson D \n(ed) Achieving sustainable cultivation of maize Volume 1: From improved varieties to \nlocal applications. Burleigh Dodds Science Publishing, Cambridge, UK, pp 1 –48. \nhttps://doi.org/10.19103/AS.2016.0001.02 \nSmith AM, Zeeman SC (2006) Quantification of starch in plant tissues. Nat Protoc 1:1342 –\n1345. https://doi.org/10.1038/nprot.2006.232 \nSundqvist MK, Sanders NJ, Wardle DA (2013) Community and ecosystem responses to \nelevational gradients: Processes, mechanisms, and insights for global change. Annu \nRev Ecol Evol Syst 44:261 –280. https://doi.org/10.1146/annurev -ecolsys-110512-\n135750 \nTak Y, Kumar M (2021) Phenolics: A key defence secondary metabolite to counter biotic \nstress. https://doi.org/10.1007/978-981-15-4890-1_13 \nTeixeira M, Feij E, Matos AR, Figueiredo A, Marques J (2021) Exploring local maize diversity \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n22 \n \nfor increased agricultural sustainability: New insights into drought stress response \nand recovery of Guinea -Bissau landraces. Sustainability 13. \nhttps://doi.org/10.3390/su13105441 \nTenaillon MI, Charcosset A (2011) A European perspective on maize history. C R Biol \n334:221–228. https://doi.org/10.1016/j.crvi.2010.12.015 \nTerashima I, Masuzawa T, Ohba H, Yokoi Y (1995) Is Photosynthesis Suppressed at Higher \nElevations Due to Low CO 2 Pressure? Ecology 76:2663 –2668. \nhttps://doi.org/10.2307/2265838 \nVigouroux Y, Glaubitz JC, Matsuoka Y, Goodman MM, G. JS, Doebley J (2008) Population \nstructure and genetic diversity of New World maize races assessed by DNA \nmicrosatellites. Am J Bot 95:1240–1253. https://doi.org/10.3732/ajb.0800097 \nde Vries A, Ripley B (2024) ggdendro: Create dendrograms and tree diagrams using \n‘ggplot2’. https://cran.r-project.org/web/packages/ggdendro/index.html \nWillcox MC, Burgueño JA, Jeffers D, Rodriguez-Chanona E, Guadarrama-Espinoza A, Kehel \nZ, Chepetla D, Shrestha R, Swarts K, Buckler ES, Hearne S, Chen C (2022) Mining \nalleles for tar spot complex resistance from CIMMYT’s maize Germplasm Bank. Front \nSustain Food Syst 6. https://doi.org/10.3389/fsufs.2022.937200 \n \n \nSupplementary Figures \nSupplementary Figure 1. Coefficients of variation of agromorphological and biochemical \ntraits of Experiment 1 (pink) and 2 (green). AA: Antioxidant Activity. \nSupplementary Figure 2.  Spearman correlations  between agromorphological  and \nbiochemical traits of the 19 accessions, adjusted for multiple testing using the sequential \nBonferroni correction. Bold font represents significant results (p<0.05). \nSupplementary Figure 3. Principal Components Analysis.  Data are classified by Region \nof origin (NWA: Northwestern Argentina; NEA: Northeastern Argentina) and Experiment (1 \nand 2). Data were averaged per accession and standardized and centered within each \nexperiment. \n \nSupplementary Tables \nSupplementary table 1.  Characteristics of the accessions evaluated from the “ Banco \nActivo de Germoplasma INTA Pergamino ” (BAP; Active Germplasm Bank of the National \nInstitute of Agricultural Technology, Pergamino, Buenos Aires, Argentina). The classification \nof soils was retrieved from the soil charter of the Instituto Geográfico Nacional (National \nGeographic Institute, IGN). \nSupplementary table 2. Agromorphological evaluation of landraces in Experiment 1 and \nExperiment 2 (Height, Leaf Number, Total Chlorophyll) or combined Experiments 1 and 2 \n(Biomass and Stem Diameter) classified by Region. Data represent Estimated Marginal \nMeans or EMM (Standard Errors or SE) from Linear Mixed Effects Models, with bold letters \nindicating significant differences (p<0.05), and Coefficients of Variation (CV). N=3-5. \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\n23 \n \nSupplementary table 3. Leaf metabolite content of fully expanded leaves from V12 stage \nplants in combined Experiments 1 and 2 classified by Region. Data represent Estimated \nMarginal Means or EMM (Standard Errors or SE) from Linear Mixed Effects Models, with bold \nletters indicating significant differences (p<0.05), and Coefficients of Variation (CV). N=3-5.  \nSupplementary table 4. Scores for phenotypic d ata obtained in the Redundant Analysis \n(RDA). Bold letters indicate the major contributors to the RDA axes (scores >0.8 or < -0.8). \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\nFigure 1. Collection sites of the individuals evaluated in this\nstudy. The list of accessions is in Table S1. NEA: Northeastern\nArgentina. NWA: Northwestern Argentina.\n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\na a\nb\nFigure 2. Salt stress tolerance assay. (a) Dry weight of roots (upper panel) and leaves (lower panel) by Region. (b) Length of\nroots (upper panel) and leaves (lower panel) by Region. Length of roots (c) and leaves (d) by Accession. Data represent Estimated\nMarginal Means or EMM (Standard Errors or SE) from Linear Mixed Effects Models, with different letters indicating significant\ndifferences (p<0.05). Three independent repetitions were performed with N=19 (a,b) and N=20 (c,d). T0 (control, 0 mM NaCl), T1\n(17.5 mM NaCl), T2 (35 mM NaCl). NEA: Northeastern Argentina. NWA: Northwestern Argentina.\na\na\na\nb\na\nb\nc\nT0\nT1\nT2\nT0\nT1\nT2\nd\na\n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\nFigure 3. Multivariate ordination and clustering analyses . a. PCA biplot based on accession and experiment mean\nvalues. Each accession is represented by two smaller points corresponding to the means of Experiment 1 and 2. Larger\npoints indicate centroids. b. Variable contributions to the first two principal components of the PCA. c. Membership\nprobabilities of each accession in Experiments 1 and 2 for the four DAPC clusters (lower panel) selected according to BIC\n(upper panel). d. Scatterplot of the four DAPC clusters. AA: Antioxidant Activity. NWA: Northwestern Argentina. NEA:\nNortheastern Argentina.\n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\nFigure 4. Hierarchical clustering . a. Dendrogram based on Euclidean distances, showing cluster assignment and region\nof origin of each accession. b. Heatmap of pairwise Euclidean distances among accessions. Dashed lines indicate the\nassigned cluster number. The clustering was done employing marginal means (Tables 1-3). NWA: Northwestern\nArgentina. NEA: Northeastern Argentina.\n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\nFigure 5. Redundancy analysis (RDA). Model\nbiplot displaying the first two axes of the model\nincluding biochemical–morphological traits\nand the three environmental predictors\n(Elevation, Latitude, and Longitude). Elevation\nis highlighted in black, as it was the only\nsignificant predictor in the full model (p<0.05).\n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\nTable 1. Height, Leaf number, and Total Chlorophyll measured with Dualex® in Experiment 1 (30 Days After Sowing) and Experiment 2 (37 Days After \nSowing).  Data represent Estimated Marginal Means or EMM (Standard Errors or SE) from Linear Mixed Effects Models, with different letters indicating \nsignificant differences (p<0.05), and Coefficients of Variation (CV). N=3-5. \nAccession \nHeight (cm) Leaf Number Total Chlorophyll Dualex® (μg/cm2) \nExperiment 1 Experiment 2 Experiment 1 Experiment 2 Experiment 1 Experiment 2 \nEMM (SE) CV (%) EMM (SE) CV (%) EMM (SE) CV (%) EMM (SE) CV (%) EMM (SE) CV (%) EMM (SE) CV (%) \nARZM03042 46 (3.43)abc 34 9.5 (0.35)abc 46 8 (0.34)abc 8.84 9.5 (0.35)a 13.59 37.1 (3.12)ab 34.2 35.1 (0.89)a 33.7 \nARZM04013 37.2 (2.77)ab 33 10.4 (0.33)abc 51 7.6 (0.33)abc 7.21 10.4 (0.33)ab 8.6 35.7 (3)ab 29 41.3 (0.62)b 40 \nARZM04029 43.5 (3.24)abc 33 11 (0.75)abc 63 7.8 (0.34)abc 10.73 11 (0.75)ab 0 38.8 (3.26)ab 34.6 37.4 (1.2)ab 36.2 \nARZM04060 50.1 (4.17)abc 45 10.7 (0.37)ab 44 8.75 (0.40)abc 5.71 10.7 (0.37)ab 4.65 44 (3.7)b 27.2 40 (3.31)ab 33.2 \nARZM05007 44.4 (3.3)abc 34 10.2 (0.32)abc 45 8.2 (0.35)abc 10.2 10.2 (0.32)ab 4.38 33.3 (2.8)ab 24.6 42.2 (2.82)ab 34 \nARZM05067 49.8 (3.71)abc 37 10.2 (0.32)abc 51 8.4 (0.35)abc 6.52 10.2 (0.32)ab 4.38 26.6 (2.23)a 20.2 37.6 (2.07)ab 32 \nARZM05069 47.6 (3.54)abc 38 10.4 (0.33)bc 55 9.2 (0.36)c 9.09 10.4 (0.33)ab 5.27 36.5 (3.06)ab 28.3 34.2 (1.39)a 30.7 \nARZM05118 58.2 (4.34)c 48 11.6 (0.34)bc 55 8.4 (0.35)abc 13.57 11.6 (0.34)b 13.07 41.7 (3.5)b 38.4 40.4 (2.16)ab 33.5 \nARZM05120 44.9 (3.34)abc 33 10.2 (0.32)abc 48 7.4 (0.33)ab 7.4 10.2 (0.32)ab 10.74 45.4 (3.82)b 40.7 40.6 (2.69)ab 34.2 \nARZM06060 34.6 (2.57)a 22 10 (0.36)abc 45 7 (0.32)a 10.1 10 (0.36)ab 8.16 35.2 (2.96)ab 27.3 41.9 (2.01)ab 36.6 \nARZM06109 46.7 (3.48)abc 39 10.3 (0.36)a 39 8.2 (0.35)abc 13.36 10.3 (0.36)ab 4.88 47.4 (3.99)b 41 40.3 (2.17)ab 33.3 \nARZM08018 48.6 (4.05)abc 44 10.8 (0.33)bc 60 9 (0.40)bc 18.14 10.8 (0.33)ab 7.75 42.3 (3.55)b 37.3 40.9 (1.7)ab 34.7 \nARZM08020 41.7 (3.11)abc 33 10 (0.32)abc 32 8.6 (0.35)abc 13.26 10 (0.32)ab 7.07 42.7 (4.02)b 35.6 42.5 (2.53)ab 35.7 \nARZM08096 53.2 (3.96)bc 49 11 (0.33)abc 55 9 (0.36)bc 7.86 11 (0.33)ab 11.13 34 (2.86)ab 23.2 40 (1.73)ab 34 \nARZM08178 53.2 (3.96)bc 48.5 10.6 (0.33)c 54 9 (0.36)bc 7.86 10.6 (0.33)ab 5.17 38.2 (3.21)ab 33.5 40.7 (4.25)ab 30.2 \nARZM10036 45.6 (3.4)abc 39.5 10.4 (0.33)bc 60 9 (0.36)bc 13.61 10.4 (0.33)ab 5.27 39.5 (3.32)ab 34 39.2 (2.654)ab 33.3 \nARZM10076 50.8 (3.78)bc 43 11 (0.33)abc 49 8.8 (0.36)bc 9.51 11 (0.33)ab 6.43 43.8 (3.68)b 37.6 41.2 (3.26)ab 31.1 \nARZM10082 50 (3.73)abc 47 10.6 (0.33)bc 61 8.2 (0.35)abc 10.2 10.6 (0.33)ab 5.17 35.9 (3.02)ab 25.8 41.2 (1.62)ab 35.5 \nARZM12205 56.5 (4.71)c 44 10 (0.32)bc 59 8.75 (0.40)abc 5.71 10 (0.32)ab 7.07 34.7 (2.91)ab 23.5 42.1 (1.37)b 39.3 \nRange 34.6–58.2  9.5–11.6  7–9.2  9.5–11.6  26.6–47.4  34.2–42.5  \nMaximum  \nfold-change 1.68  1.22  1.31  1.22  1.78  1.24  \n \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\nTable 2. Biomass and Stem Diameter in combined Experiments 1 and 2 . Data represent Estimated Marginal Means or EMM (Standard Errors or SE) \nfrom Linear Mixed Effects Models, with different letters indicating significant differences (p<0.05), and Coefficients of Variation (CV). N=3-5. \nAccession \n \nTotal Biomass (g) \n  \n \nLeaf Biomass (g) \n  \n \nStem Biomass (g) \n  \nStem Diameter (cm)  \nEMM (SE) CV (%) EMM (SE) CV (%) EMM (SE) CV (%) EMM (SE) CV (%) \nARZM03042 42.8 (23.8)abc 16.7 17.1 (11.22)ab 7.25 15.43 (8.16)abcde 5.65 1.05 (0.61)abc 0.6 \nARZM04013 46.6 (25.5)abc 7.28 21 (13.72)abc 3.55 14.59 (7.66)abcde 2.38 1.22 (0.71)c 0.35 \nARZM04029 72.8 (40.2)bcd 37.09 32.8 (21.69)bcd 14.25 25.97 (14.08)cdef 12.39 1.28 (0.75)c 0.4 \nARZM04060 86.5 (47)cd 42.58 33 (21.6)bcd 13.37 27.35 (14.04)ef 12.1 1.16 (0.68)abc 0.42 \nARZM05007 73.2 (39.7)bcd 20.37 24.6 (16.07)abcd 9.03 19.8 (10.05)de 7.31 1.10 (0.64)abc 0.24 \nARZM05067 52.2 (28.5)abcd 11.58 21.8 (14.23)abcd 5.85 19.52 (10.09)bcdef 5.36 1.06 (0.62)abc 0.32 \nARZM05069 53.7 (29.3)abcd 21.38 22.7 (14.83)abcd 9.34 18.15 (9.21)cde 5.54 1.10 (0.64)abc 0.32 \nARZM05118 103.7 (56)d 32.08 50.4 (32.89)d 10.43 31.57 (16.12)f 15.42 1.26 (0.73)c 0.38 \nARZM05120 71 (38.5)bcd 18.42 31.7 (20.69)bcd 6.08 17.07 (8.68)bcd 7.78 1.11 (0.65)abc 0.53 \nARZM06060 26.9 (14.9)a 5.44 11.1 (7.24)a 2.43 8.77 (4.51)a 3.22 0.86 (0.50)a 0.22 \nARZM06109 48.2 (26.3)abc 13.37 16.2 (10.59)ab 6.91 14.38 (7.28)bc 3.6 1.05 (0.61)abc 0.3 \nARZM08018 83 (44.9)cd 29.13 44.8 (29.23)cd 8.68 25.97 (13.27)ef 15.91 1.30 (0.76)c 0.68 \nARZM08020 35.6 (19.7)ab 7.35 15.5 (10.13)ab 4.24 11.38 (5.85)ab 3.05 0.86 (0.50)ab 0.2 \nARZM08096 68 (36.8)bcd 21.85 32.4 (21.04)bcd 9.9 25.66 (13.34)def 7.39 1.24 (0.72)c 0.62 \nARZM08178 72.7 (39.9)bcd 12.57 34.6 (22.77)bcd 3.22 20.94 (11.66)abcdef 3.87 1.31 (0.77)c 0.3 \nARZM10036 65.5 (36)bcd 27.3 29.7 (19.55)bcd 9.4 19.41 (9.82)de 10.15 1.16 (0.68)abc 0.74 \nARZM10076 85.1 (45.9)cd 35.53 45.6 (29.66)cd 10.95 27.18 (14.31)def 14.18 1.15 (0.67)bc 0.65 \nARZM10082 64.1 (34.8)bcd 14.32 23 (14.99)abcd 5.93 15.8 (8.12)bcd 7.03 1.25 (0.73)c 0.36 \nARZM12205 62.5 (34)bcd 21.76 32.3 (21.05)bcd 7.74 17.27 (8.92)bcde 8.8 1.10 (0.64)abc 0.34 \nRange 26.9–103.7  11.1–50.4  8.77–31.57  0.86–1.31  \nMaximum  \nfold-change 3.86  4.54  3.6  1.52  \n \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\nTable 3a. Leaf metabolite content of fully expanded leaves from V12 stage plants in combined Experiments 1 and 2 (Part 1). Data represent Estimated \nMarginal Means or EMM (Standard Errors or SE) from Linear Mixed Effects Models, with different letters indicating significant differences (p<0.05), and \nCoefficients of Variation (CV). N=3-5.  \nAccessions \nTotal Proteins \n(mg/g) \nTotal Soluble Sugars \n(mg/g) Starch (mg/g) Total Phenols  \n(mg GA/g) \nAntioxidant  \nActivity \n(Hydrophilic  \nphase) \n(mg GA/g) \nAntioxidant  \nActivity \n(Hydrophobic  \nphase) \n(mg GA/g) \nEMM (SE) CV (%) EMM (SE) CV (%) EMM (SE) CV (%) EMM (SE) CV (%) EMM (SE) CV (%) EMM (SE) CV (%) \nARZM03042 6.33 (2.3)ab 4.34 7.73 (1.49)a 5.21 8.6 (10.12)a 113.2 1.94 (0.28)ab 1.2 0.0967 (0.0607)a 52.41 0.078 (0.0219)a 0.03 \nARZM04013 7.27 (2.61)b 3.17 11.43 (2.2)a 5.9 11.43 (13.37)a 56.71 1.86 (0.26)a 1.17 0.0924 (0.0604)a 65.86 0.1026 (0.0263)a 0.06 \nARZM04029 3.5 (1.29)a 2.86 9.63 (2.14)a 6 11.72 (13.66)a 65.8 1.89 (0.29)ab 1.11 0.076 (0.0624)a 60.52 0.0604 (0.0189)a 0.03 \nARZM04060 5.92 (2.29)ab 2.07 9.63 (1.85)a 4.03 10.99 (12.86)a 72.82 2.08 (0.28)ab 1.36 0.0913 (0.0605)a 57.83 0.0682 (0.0194)a 0.004 \nARZM05007 3.09 (1.25)a 1.1 16.23 (2.79)a 5.06 17.73 (20.94)a 76.5 2.18 (0.25)ab 1.54 0.085 (0.0607)a 64.24 0.0965 (0.0235)a 0.03 \nARZM05067 2.18 (1.41)ab 0.05 10.49 (1.9)a 7.37 8.44 (9.78)a 120.5 2.1 (0.26)ab 1.26 0.1017 (0.0612)a 46.55 0.1207 (0.0395)a 0.1 \nARZM05069 3.93 (1.51)ab 2.18 13.9 (2.39)a 4.54 11.42 (13.26)a 77.1 2.33 (0.26)ab 1.28 0.092 (0.0605)a 75.49 0.0921 (0.0249)a 0.01 \nARZM05118 2.73 (1.15)a 0.37 8.16 (1.48)a 2.93 11.36 (13.19)a 72.55 1.92 (0.25)ab 1.04 0.0932 (0.0605)a 63.7 0.0859 (0.023)a 0.03 \nARZM05120 6.63 (2.44)ab 3.26 13.36 (2.57)a 7.47 10.18 (11.88)a 104.75 2.06 (0.25)ab 1.65 0.0903 (0.0605)a 72.07 0.0824 (0.021)a 0.01 \nARZM06060 5.89 (2.15)ab 2.89 12.45 (2.26)a 6.02 12.65 (14.72)a 103.29 2.34 (0.26)ab 1.6 0.0989 (0.0605)a 63.48 0.1124 (0.0269)a 0.02 \nARZM06109 3.12 (1.26)a 1.16 11.8 (2.03)a 6.8 17.9 (20.98)a 94.44 2.03 (0.26)ab 0.87 0.0997 (0.0606)a 61.28 0.0952 (0.0232)a 0.03 \nARZM08018 5.57 (2.07)ab 3.2 14.75 (2.54)a 4.83 8.28 (9.66)a 137.59 2.7 (0.25)ab 1.84 0.0907 (0.0605)a 72.59 0.0984 (0.0258)a 0.03 \nARZM08020 3.44 (1.36)ab 1.91 14.22 (3.16)a 4.5 6.74 (8.24)a 166.26 3.03 (0.27)b 1.49 0.1044 (0.0612)a 54.29 0.0533 (0.0174)a 0.01 \nARZM08096 4.67 (1.68)ab 2.11 11.75 (2.13)a 6.37 8.13 (9.42)a 90.02 2.14 (0.25)ab 1.34 0.0861 (0.0605)a 78.65 0.0922 (0.0282)a 0.01 \nARZM08178 3.41 (1.5)ab 0.71 13.81 (2.66)a 8.3 9.37 (10.94)a 104.5 2.24 (0.26)ab 1.5 0.0854 (0.0605)a 67.96 0.1055 (0.0271)a 0.03 \nARZM10036 5.2 (1.87)ab 3.46 13.09 (2.37)a 1.39 8.99 (10.47)a 101.48 1.98 (0.25)ab 1.33 0.0888 (0.0605)a 70.5 0.0898 (0.0217)a 0.02 \nARZM10076 5.27 (1.97)ab 2.68 13.42 (2.58)a 5.87 13.84 (16.14)a 88.04 1.84 (0.26)a 1.31 0.0925 (0.0606)a 77.23 0.1087 (0.0298)a 0.04 \nARZM10082 5.23 (1.94)ab 2.69 13.37 (2.42)a 7.12 7.21 (8.52)a 108.31 2.12 (0.25)ab 1.51 0.0951 (0.0605)a 67.26 0.0806 (0.0202)a 0.01 \nARZM12205 4.3 (1.76)ab 0.89 14.61 (2.65)a 8.27 8.78 (10.22)a 85.74 2.1 (0.25)ab 1.53 0.0919 (0.0619)a 59.03 0.1459 (0.0393)a 0.01 \nRange 2.18–7.27  7.73–16.23  6.74–17.9  1.84–3.03  0.08–0.1  0.05–0.15  \nMaximum  \nfold-change 3.33  2.1  2.66  1.65  1.37  2.74  \nGA= gallic acid. \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint \n\nTable 3b. Leaf metabolite content of fully expanded leaves from V12 stage plants in combined Experiments 1 and 2 (Part 2). Data represent Estimated \nMarginal Means or EMM (Standard Errors or SE) from Linear Mixed Effects Models, with different letters indicating significant differences (p<0.05), and \nCoefficients of Variation (CV). N=3-5.  \nAccessions Total Chlorophyll (g/g) Chlorophyll a (g/g) Chlorophyll b (g/g) Total Carotenoids (g/g) \nEMM (SE) CV (%) EMM (SE) CV (%) EMM (SE) CV (%) EMM (SE) CV (%) \nARZM03042 1.6 (0.45)ab 41.32 1.554 (0.80)abc 41.45 0.806 (0.48)a 42.21 0.951 (0.65)a 41.47 \nARZM04013 1.78 (0.52)ab 33.39 1.737 (0.86)abc 33.69 0.872 (0.48)a 32.83 0.934 (0.65)a 33.37 \nARZM04029 1.52 (0.44)ab 24.16 1.269 (0.61)abc 24.37 0.838 (0.48)a 26.05 1.104 (0.65)a 25.77 \nARZM04060 2.24 (0.64)ab 24.02 1.836 (0.84)bc 26.25 0.992 (0.48)a 16.18 0.944 (0.65)a 26.28 \nARZM05007 1.55 (0.43)ab 43.35 1.606 (0.83)abc 44.64 0.815 (0.48)a 39.55 0.864 (0.64)a 40.57 \nARZM05067 1.15 (0.32)a 47.24 1.3 (0.70)abc 48.79 0.723 (0.48)a 41.66 0.932 (0.65)a 45.32 \nARZM05069 1.39 (0.39)ab 40.91 1.42 (0.69)abc 42.31 0.759 (0.48)a 35.84 0.951 (0.65)a 37.36 \nARZM05118 1.48 (0.41)ab 32.28 1.373 (0.64)abc 32.54 0.834 (0.48)a 32.46 1.257 (0.65)a 28.36 \nARZM05120 2.68 (0.75)b 59.87 2.479 (1.18)c 59.38 1.15 (0.49)a 61.9 0.983 (0.65)a 54.88 \nARZM06060 1.41 (0.39)ab 56.79 0.962 (0.41)ab 55.25 0.928 (0.48)a 81.76 1.045 (0.65)a 89.02 \nARZM06109 1.47 (0.41)ab 50.74 1.234 (0.60)abc 49.82 0.849 (0.48)a 73.07 1.115 (0.65)a 91.69 \nARZM08018 1.47 (0.41)ab 66.27 1.035 (0.44)ab 60.22 0.943 (0.48)a 75.65 0.953 (0.65)a 80.18 \nARZM08020 1.01 (0.30)a 65.45 0.717 (0.32)a 59.66 0.813 (0.48)a 98.22 0.969 (0.65)a 105.03 \nARZM08096 1.18 (0.33)ab 74.69 0.774 (0.34)a 69.49 0.928 (0.48)a 71.62 0.946 (0.65)a 85.23 \nARZM08178 1.02 (0.30)a 51.73 0.757 (0.32)a 46.75 0.745 (0.48)a 89.25 0.98 (0.65)a 90.75 \nARZM10036 1.17 (0.34)ab 70.59 0.868 (0.38)a 60.78 0.943 (0.48)a 73.95 1.065 (0.65)a 82.57 \nARZM10076 1.39 (0.39)ab 83.48 0.914 (0.40)ab 81.79 0.997 (0.48)a 75.45 1.169 (0.65)a 85.63 \nARZM10082 1.63 (0.45)ab 106.92 1.02 (0.46)abc 99.01 1.228 (0.50)a 58.44 0.942 (0.65)a 74.1 \nARZM12205 1.07 (0.30)a 65.72 0.722 (0.31)a 62.35 0.826 (0.48)a 81.41 0.944 (0.65)a 89.51 \nRange 1.01–2.68  0.72–2.48  0.72–1.23  0.86–1.26  \nMaximum \nFold-change 2.65  3.46  1.7  1.45  \n \n.CC-BY 4.0 International licenseperpetuity. It is made available under a \npreprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in \nThe copyright holder for thisthis version posted January 3, 2026. ; https://doi.org/10.64898/2026.01.02.697242doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}