Genome-Wide Association Analysis of Traits Related to Nitrogen Deficiency Stress in Potato

Abstract

Potato (Solanum tuberosum L.) crop yields may be reduced by nitrogen deficiency stress tolerance. An evaluation of 144 tetraploid potato genotypes was carried out during two consecutive seasons (2019 and 2020), with the objective of characterizing their variability in key physiological and agronomic parameters. Physiological parameters included chlorophyll content and fluorescence, stomatal conductance, NDVI, leaf area, and perimeter, while agronomic characteristics such as yield, tuber fresh weight, tuber number, starch content, dry matter, and reducing sugars were evaluated. To genotype the population, the GGP V3 Potato array was used, generating 18,259 high-quality SNP markers. Marker–trait association analysis was conducted using the GWASpoly package in R, applying Q + K linear mixed models to enhance precision. This methodology enabled the identification of 18 SNP markers that exhibited statistically significant associations with the traits analyzed in both trials and periods, relating them to genes whose functional implication has already been described. Genetic loci associated with chlorophyll content and tuber number were detected across non-stress and stress treatments, while markers linked to leaf area and leaf perimeter were identified specifically under nitrogen deficiency stress. The genomic distribution of these markers revealed that genetic markers or single-nucleotide polymorphisms (SNPs) correlated with phenotypic traits under non-stress conditions were predominantly located on chromosome 11, whereas SNPs linked to stress responses were mainly identified on chromosomes 2 and 3. These findings contribute to understanding the genetic mechanisms underlying potato tolerance to nitrogen deficiency stress, offering valuable insights for the development of future marker-assisted selection programs aimed at improving nitrogen use efficiency and stress resilience in potato breeding.

1. Introduction

Potato (Solanum tuberosum L.) is one of the most significant crops worldwide, providing an annual global production of 383 Mt of tubers, according to FAOSTAT data (2023). It ranks as the fourth most important crop globally, following rice, wheat, and maize, and is the leading non-grain agricultural product. Due to their rich nutritional profile, potatoes play a crucial role in human diets, and they are also widely used for starch production in industries such as paper, adhesives, textiles, and food processing [1].

Nitrogen is a key component of various cellular structures and is an essential nutrient for plant growth and development. As a result, nitrogen application is a vital agricultural practice in potato cultivation [2,3].

A major challenge in modern agriculture is striking a balance between minimizing environmental harm while maintaining high yields and quality produce. Although nitrogen availability boosts plant growth, excessive application can lead to overly vigorous vegetative growth at the expense of root and tuber development. The proper management of nitrogen fertilizers is essential for fostering sustainable farming systems, as nitrogen is also a potential environmental pollutant. Therefore, optimizing nitrogen use is critical for ensuring both environmental protection and productive crop yields [4].

Breeding for nitrogen deficiency stress is challenging, but essential for the expected upcoming climatic changes. Improving nitrogen use efficiency (NUE) can help reduce water and soil pollution and increase crop yield.

However, nitrogen use efficiency (NUE) is a highly complex trait, regulated by an extensive network of genes and strongly influenced by environmental conditions. To understand its genetic basis, it is essential to establish connections between plant physiological function, agronomic traits, and DNA markers, which would allow for the development of more precise and sustainable breeding strategies [5].

The genetic foundation of nitrogen use efficiency (NUE) in potato remains largely unknown, with its complexity stemming from various quantitative trait loci (QTLs) that influence phenotypic traits related to stress adaptation. Conducting QTL analysis using diverse molecular markers allows researchers not only to pinpoint the key loci associated with specific traits, but also to explore the connections between different phenotypic characteristics. This approach enhances our understanding of genetic regulation, ultimately supporting more efficient breeding strategies [6].

Recent advancements in sequencing technology have led to lower costs and a higher volume of sequence reads [7]. Genetic association, also known as linkage disequilibrium (LD) analysis, allows for the establishment of links between particular genetic variants and observable characteristics, facilitating the identification of the functional genes or genomic regions involved [8]. One of the main strengths of this method lies in its independence from the creation of mapping populations, allowing researchers to analyze historical recombination events within a broader population [9,10]. Additionally, the absence of biparental crosses for QTL identification makes association mapping more efficient and cost-effective [11]. The effectiveness of identifying QTLs in marker-assisted selection (MAS) studies is limited by a low mapping resolution, resulting from a reduced number of recombination events and segregations [12]. To address the limitations of traditional breeding methods, such as classical breeding and biparental mapping, Next-Generation Sequencing (NGS) emerges as a highly promising alternative. NGS facilitates the rapid development of genome-wide markers (SNPs), providing an unprecedented level of resolution for exploring the connection between genetic variation and phenotypic diversity.

Association Mapping (AM) utilizes high-throughput, densely spaced DNA markers like SNPs to effectively uncover connections between genetic markers and desirable traits in crops. This approach enables the identification of specific functional variants (alleles), which can then serve as selection markers for targeted trait improvement. AM works on the principle of linkage disequilibrium (LD) [13].

For a few years now, the complete potato genome sequence has been available, allowing for the development of single-nucleotide polymorphism (SNP) arrays by the potato community [14]. Several generations of SNP arrays have been generated, building on the original Infinium 8303 SNP array [15].

Genome-Wide Association Studies (GWASs) offer a powerful approach to identifying the QTLs linked to specific traits, providing deeper insights into genetic variations. When many molecular markers are used across diverse genotypes, this method not only detects significant loci, but can also pinpoint causal polymorphisms within genes, explaining the differences between alternative phenotypes. Essentially, GWASs enhance genetic mapping precision, supporting the identification of key genetic factors underlying trait variation [16,17,18,19].

In recent years, several QTL analysis studies have been conducted on different potato traits such as flower color, maturity, dry matter, starch content, yield [20], agronomic and quality traits [21], tuber dormancy [22], tuber shape, skin color, and flesh color [23,24], as well as studies related to drought stress [25,26,27]. However, the number of QTL studies for nitrogen deficiency is still very limited.

In potato, QTL studies for NUE are still very limited. Only recently were QTLs affecting traits related to NUE under contrasting N regimes reported [28,29,30].

Ospina et al. [30] and Getahum et al. [29], in more recent GWASs under nitrogen deficiency conditions, found some interesting QTLs related to yield, chlorophyll content, and NUE.

Getahum et al. [29] identified several interesting QTLs, highlighting the marker SPUD 237 related to tuber number at both nitrogen levels they tried.

Ospina et al. [30] found some QTLs common to both nitrogen levels they tried, related to yield and NUE and located on chromosome 5, while the QTLs detected for yield exclusively under high nitrogen levels were located on chromosomes 9 and 12.

In our case, QTLs related to chlorophyll content (SPAD) and yield factors were found at both nitrogen levels we tried, with some of these chlorophyll-content-related QTLs (SPAD) concentrated on chromosome 11.

Moreover, major nitrogen-responsive genes such as nitrate reductase, nitrite-reductase, glutamine synthetase, glutamate dehydrogenase ammonium transporters, asparagine synthase, chloroplast pigment binding, association with chlorophyll a-b, ATP synthase, and cell wall kinase genes have been identified in different potato genotypes under different N levels. These findings evidence the close connection between nitrogen metabolism and photosynthesis, highlighting the complex interaction among plant physiological processes [2,31,32,33].

Molecular-marker-assisted selection, utilizing high-density SNP markers identified through GWASs, has significantly enhanced breeding efficiency. Unlike traditional pedigree-based selection, this approach does not rely on pinpointing specific gene loci controlling traits or conducting extensive field phenotyping. Instead, it allows for faster genetic improvements by leveraging molecular data, accelerating genetic gain over shorter time periods [34,35].

In this study, we design and identify molecular markers associated with physiological and agronomic traits of interest for potato breeding under nitrogen-deficiency-stressed and unstressed conditions, using a Genome-Wide Association Study. We find interesting QTLs related to many of the agronomic traits measured, particularly related to tuber number, chlorophyll content, and leaf area. Therefore, our findings are consistent with the other studies cited here, although the QTLs found are not the same.

2. Materials and Methods

2.1. Vegetal Samples and Experimental Site

A collection of 144 tetraploid cultivars of Solanum tuberosum ssp. tuberosum—representing varied breeding backgrounds—was tested in field conditions. The trials took place at NEIKER’s experimental station (42°51005.700 N, 2°37013.200 W) in Spain across two years, 2019 and 2020.

2.2. Experimental Design

The agronomic evaluations were carried out between May and September in each of the years considered, under the specific meteorological conditions of the experimental site. The following two experimental sets were implemented in both years of study: fertilized plots representing the control and unfertilized plots as a contrast. The control plot was fertilized with sowing at 1000 kg/ha NPK (9-18-27) and in mulch (300 kg/ha NAC 27%), while the N-deficient plot was fertilized with 50% of the dose of sowing. In each block, the genotypes were planted in a completely randomized experimental design with two replicates of 5 plants each. Plants were spaced at intervals of 30 cm within rows, with 75 cm separating each row

2.3. Phenotypic Data Collection

This study evaluated four physiological traits in different potato genotypes 70 days after planting (DAP). Chlorophyll content (SPAD value, unitless index ranging from 0 to 60) was measured using a SPAD-502 chlorophyll meter (Konica Minolta, Osaka, Japan) in the last fully expanded leaf of three plants per replicate. The photochemical efficiency of PSII (Fv’/Fm’, unitless index ranging from 0 to 1) was analyzed using a fluorimeter (FluorPen FP 100, Photon Systems Instruments, Drasov, Czech Republic), also in the last fully expanded leaf of three plants of each replicate and treatment. Stomatal conductance (gs, mmol H₂O m⁻² s⁻¹) was measured using a porometer (Leaf Porometer, Decagon Devices, Pullman, Washington, EEUU) in the last fully expanded leaf of one plant of each replicate and treatment. The normalized difference vegetation index (NDVI values, unitless index ranging from −1 to +1) was assessed with a Rapidscan (RapidScan CS-45, Holland Scientific, Lincoln, EEUU), scanning five samples per replicate located 50 m above the top of the plants and oriented in the direction of the rows. At 70 days after planting (DAP), three leaves per genotype and replicate were sampled from each treatment to assess leaf surface area (cm²) and perimeter (cm) using ImageJ (version 1.8.0).

To accommodate the maturation of late varieties, plants were collected at 127 DAP (2019) and 126 DAP (2020). The entire experiment was collected in a single harvest event, and yield (g), tuber quantity per plant (number of tubers), and average fresh weight (g) were measured across replicates. Five plants per replicate were harvested collectively, and their combined output was averaged to determine the individual yield and tuber count per plant. The average tuber mass was obtained by dividing the total yield by the number of tubers. Two samples per genotype and treatment were analyzed to determine dry matter (DM, %) percentage. Tubers were initially weighed right after harvesting to determine their fresh mass (FW). Subsequently, samples were dried at 80 °C for 72 h, after which dry weight (DW) was measured. Starch concentration (mg g⁻¹ FW) was estimated through the following standardized formula [36]:

$$Starch=\left({\displaystyle \frac{DW}{FW}} \times 100 - 6.0313\right) \times 10$$

The quantification of reducing sugars in the samples was performed using spectrophotometry, relying on the reduction of dinitrosalicylic acid [37]. Each variety and treatment were analyzed in duplicate to ensure accuracy. Potato samples were peeled and homogenized into a homogenous juice. From this, 0.3 g of the mixture was weighed, followed by the addition of 1 mL of distilled water and 2 mL of dinitrosalicylic acid. The solution was then heated to 100 °C in a water bath under constant stirring for 10 min. After heat treatment, the samples were diluted with distilled water and their absorbance was measured at 546 nm using a UV–VIS spectrophotometer. The percentage of reducing sugars was calculated based on the obtained absorbance values, as follows:

$$\%reducing sugars=\left(absorbance - 0.00385\right)\times 1.07893$$

Nitrogen use efficiency (NUE, %) was calculated as the ratio of tuber dry matter to nitrogen applied, following the methodology described by Meise et al. (2019) [1].

Data collected over both years underwent analysis of variance (ANOVA) for each trait using RStudio version 2022.12.0 (R Core Team, 2017, Vienna, Austria). ANOVA was performed using the aov() function from base R. This approach was selected because the experimental design corresponds to a completely randomized design (CRD), which is appropriate given the homogeneous conditions under which the treatments were applied. Subsequently, mean trait values were employed to establish statistical correlations.

Pearson’s correlation analysis was performed to evaluate the linear relationships between physiological and agronomic traits across different nitrogen conditions. The analysis was conducted using base RStudio version 2022.12.0 (R Core Team, 2017), applying the cor() function to obtain pairwise correlation coefficients.

2.4. Genotypic Data Collection

Genomic DNA was isolated from 144 freshly harvested potato leaves using the innuPREP Plant DNA Kit (AnJenaalytik, Jena, Germany), in accordance with the manufacturer’s protocol. The concentration and purity of the DNA were evaluated with a NanoDrop 2000 spectrophotometer (NanoDrop Technologies, Inc., Wilmington, DE, USA). The DNA samples were then submitted to Neogen (Scotland, UK) for genotyping using the GGPv3 Potato 35K SNP array. Genotype identification was carried out with Genome Studio software (Illumina MiSeq plataform, San Diego, CA, USA), which enabled the detection of four alleles at each genetic locus. To ensure high-quality SNP data, markers with over 10% missing values and those with a minor allele frequency below 0.05 were excluded from the dataset.

2.5. Characterization of Population Genetics, LD Metrics, and Genome-Wide Analysis

The population structure matrix (Q-matrix) was assessed across the entire population using Structure v.2.4 software, considering K-values from 1 to 10. This analysis incorporated 18,259 SNP markers, allowing for a comprehensive evaluation of genetic diversity and subpopulation differentiation within the dataset [38].

The burn-in period was set to 100,000 iterations, followed by 100,000 MCMC replications to refine the estimation. The optimal K-value was determined using the delta K (∆K) method [39], a previously established approach, which was applied through the Structure Harvester website [40]. The relationship between genotypes and genetic diversity in the population was calculated from the SNP marker data using TASSEL software [40].

The evaluation of linkage disequilibrium was conducted on selected SNPs with high fidelity through TASSEL software [41]. Pairwise linkage estimates (r²) were determined across SNPs using a 50-SNP sliding window approach. Subsequently, the data were visualized as a function of genomic distance, and a logarithmic regression model was fitted in R to quantify the extent of linkage disequilibrium decay throughout the genome [42].

Association mapping was conducted utilizing phenotypic and genotypic data, using the GwasPoly [43] statistical library integrated into R software (R Core Team, 2017). The analysis was based on a mixed model, incorporating corrections for kinship (K) and sub-population structure (Q). To correct for multiple testing, the Bonferroni threshold of 5% was applied, setting a significance cutoff at −log10(P) = 5.01. A total of 31,190 markers were initially screened to validate SNP quality, with those exhibiting a missing data rate above 10% or a minor allele frequency below 0.05 being excluded. This filtering process resulted in 18,259 high-quality SNP markers, which offered comprehensive genome-wide coverage across the 12 chromosomes of tetraploid potatoes.

3. Results

3.1. Phenotypic Data Analysis

Analysis of variance (ANOVA) revealed highly significant differences between the varieties in both years (

Table 1. ANOVA of physiological and agronomic parameters evaluated in the field. SPAD70: Soil Plant Analysis Development after 70 days after planting (SPAD value, unitless index ranging from 0 to 60); NDVI70: Normalized difference vegetation index (NDVI values, unitless index ranging from −1 to +1); SC70 = stomatal conductance (mmol H2O m⁻² s⁻¹); FLUOR70 = chlorophyll fluorescence (Fv’/Fm’, unitless index ranging from 0 to 1); Yield = Yield (g); TubNum = number of tubers; TubWeight = tuber weight (g); DM: dry matter (%); RD: reducing sugars (%); Starch: starch (mg g⁻¹ FW); NUE: Nitrogen Use Efficiency (%).

	SPAD70	NDVI70	SC70	FLUOR70	Yield	TubNum	TubWeight	DM	RS	Starch	NUE
YEAR	5307.97 **	0.2628 **	13,285.38 ns	0.2878 **	202.84 **	24,733.92 **	1280.07 ns	1163.12 **	0.08 **	1227.07 **	0.16 **
VAR	133.67 *	0.0091 **	104,986.25 **	0.0085 **	22.00 **	2369.87 **	6824.98 **	8676.68 **	12.70 **	9160.06 **	1.82 **
STRESS	577.13 **	0.0108 ns	274,893.98 *	0.3528 **	27.71 **	1786.08 *	3957.53 *	74.53 **	0.06 *	78.61 **	28.07 **
REP YEAR	4.04 ns	0.0015 ns	6841.68 ns	0.0028 ns	6.35 *	224.72 ns	2576.15 ns	7.50 ns	0.07 *	7.50 ns	0.00 ns
YEAR * VAR	26.74 **	0.0037 **	90,423.26 **	0.0035 **	5.53 **	410.09 **	1963.76 **	2497.69 **	2.59 **	2636.43 **	0.60 **
STRESS * VAR	15.93 *	0.0026 ns	42,024.97 *	0.0019 **	1.71 **	191.12 **	691.64 *	2781.60 **	3.01 **	2935.46 **	1.09 **
YEAR * STRESS	52.22 *	0.0162 **	533,012.04 **	0.0003 ns	10.45 **	1130.72 *	0.144 ns	55.68 **	0.14 **	58.95 **	0.03 **
YEAR * STRESS * VAR	15.91 **	0.0029 **	56,095.79 **	0.0013 ns	1.23 *	169.16 **	578.93 ns	17.38 **	0.02 **	18.35 **	0.01 **

^ns = not significant; *, ** significant at p < 0.05; p < 0.001, respectively.

). Summary statistics of the evaluated traits are presented in Supplementary Tables S1.1 and S1.2.

The results reveal that indicators of plant health, including SPAD70 and NDVI70, are strongly affected by factors such as environmental conditions (YEAR) and genetic variation (VAR). Moreover, stress response is particularly evident in traits like stomatal conductance and chlorophyll fluorescence, suggesting that different varieties exhibit variable physiological adaptations under nitrogen deficiency stress.

The relationships between physiological parameters and yield components under stress conditions during the 2019 and 2020 seasons were analyzed (Figure 1). Among these traits, yield stands out as a key indicator for evaluating tolerance to stress. The intensity of the color shading reflects the strength of the correlation coefficients, ranging from light to dark blue, where darker tones indicate stronger positive associations. A side legend helps interpret the gradient scale. The correlation analysis reveals that several physiological traits—such as SPAD values, NDVI, and fluorescence—are moderately to strongly associated with key agronomic parameters like tuber number, tuber weight, and starch content under nitrogen-deficient conditions (Figure 1).

3.2. Population Structure Analysis and Linkage Disequilibrium

STRUCTURE software analysis identified two subpopulations within the study group, consisting of 133 and 11 genotypes, respectively, based on the delta K value of 2 (Supplementary Figure S1a). A probabilistic assignment was carried out to classify each genotype into its designated group, generating the corresponding Q matrix values (Supplementary Figure S1b). To further assess genetic variation, a genetic distance matrix was constructed for all genotypes. The maximum genetic distance between two varieties was determined to be 0.4, the minimum was 0.26, and the average genetic distance was 0.37.

LD decay was estimated from high-quality SNPs. Genetic distances among markers were derived by locating the crossover between the genome-wide r² half-decay value and the smoothing spline regression used to model LD patterns (Supplementary Figure S2).

3.3. Genome-Wide Association Analysis

To minimize the incidence of false positives in the association mapping, a kinship-based correction was employed. The Q + K mixed linear model was implemented using 18,259 rigorously filtered SNPs across a diverse panel of 144 genotypes. The Q-Q plot outcomes demonstrated strong concordance between observed and expected −log₁₀(P) values, supporting the robustness of the applied statistical method (Supplementary Figure S3).

Marker–trait associations were used to illustrate the population’s response to nitrogen deficiency stress. Eighteen QTLs showed statistically significant effects, exceeding the Bonferroni-adjusted cutoff. Four of these QTLs were associated with two of the traits measured under control conditions, while the rest were associated with traits measured in plants under nitrogen deficiency.

The marker solcap_snp_c2_15287, located on chromosome 11, was associated with chlorophyll content measured at 70 DAP in control plants. Two other QTLs were also found on chromosome 11, which, in this case, were associated with the number of tubers in control plants (solcap_snp_c2_37217 and ST4.03ch11_2070850). The marker solcap_snp_c2_15676, located on chromosome 5, was also associated with this trait (

Table 2. Significant SNPs associated with evaluated physiological and agronomical traits under control and nitrogen deficiency stress conditions in 144 potato varieties. SPAD70_C: Soil Plant Analysis Development after 70 days after planting under control condition; TubNum_C: number of tubers under control condition; SPAD70_N: Soil Plant Analysis Development after 70 days after planting under nitrogen deficiency stress; TubNum_N: number of tubers under nitrogen deficiency stress; Area_N: leaf area under nitrogen deficiency stress; Perim_N: leaf perimeter under nitrogen deficiency stress.

Trait	Marker	Chrom	Position	Ref	Alt	Effect	R2	p-Value	FDR	Biological Function
SPAD70_C	solcap_snp_c2_15287	11	41743380	A	G	−4.11	0.0075	2.26 × 1010	0.0472	P-loop containing nucleoside triphosphate hydrolases superfamily protein
TubNum_C	solcap_snp_c2_37217	11	1818959	A	G	39.96	6.00 × 1011	0.6786	0.0361	Disease resistance protein (NBS-LRR class) family
TubNum_C	ST4.03ch11_2070850	11	2070850	A	T	−24.22	0.0153	0.0355	0.0222	Di-glucose binding protein with Kinesin motor domain
TubNum_C	solcap_snp_c2_15676	5	18718517	G	T	−32.25	0.0438	3.44 × 1011	0.05	RNA-binding CRS1/YhbY (CRM) domain-containing protein
SPAD70_N	ST4.03ch01_72837229	1	72837229	A	G	5.88	0.0228	0.0101	0.0083	Glycosyl transferase family 1 protein
SPAD70_N	solcap_snp_c2_20505	1	67223465	C	T	−16.22	0.0227	0.0103	0.0111	6-phosphogluconate dehydrogenase family protein
SPAD70_N	solcap_snp_c2_32462	2	22381719	G	T	17.13	0.00087	0.1122	0.0277	Beta HLH protein
SPAD70_N	ST4.03ch02_48054928	2	48054928	A	G	5.44	0.0013	0.5315	0.0305	Hypothetical protein
SPAD70_N	ST4.03ch02_48088436	2	48088436	A	G	20.31	1.79 × 1012	0.8208	0.3888	YELLOW STRIPE like
SPAD70_N	PotVar0075324	4	67822196	G	T	−18.33	0.0184	0.0210	0.0194	WRKY DNA-binding protein
SPAD70_N	ST4.03ch05_51888861	5	51888861	A	T	15.04	0.0209	0.0138	0.0138	Eukaryotic aspartyl protease family protein
SPAD70_N	ST4.03ch05_51733937	5	51733937	C	T	18.13	0.0010	0.5753	0.0333	Mitochondrial acyl carrier protein
SPAD70_N	solcap_snp_c2_35078	7	49839630	A	G	17.74	0.0130	0.0524	0.0250	RAB GTPase homolog A1F
SPAD70_N	ST4.03ch11_2770569	11	2770569	C	T	14.77	0.0272	0.0049	0.0027	3-deoxy-D-arabino-heptulosonate 7-phosphate synthase
TubNum_N	solcap_snp_c2_15676	5	18718517	G	T	2.2	0.0963	7.33 × 107	0.0416	RNA-binding CRS1/YhbY (CRM) domain-containing protein
Area_N	PotVar0055568	3	51605633	C	T	2.48	0.0259	0.0061	0.0055	Aldolase-type TIM barrel family protein
Area_N	ST4.03ch03_51965651	3	51965651	A	G	−24.6	0.0195	0.0174	0.0166	Calcineurin-like metallo-phosphoesterase superfamily protein
Perim_N	ST4.03ch03_51965651	3	51965651	A	G	1.34	0.0750	2.31 × 108	0.0444	Calcineurin-like metallo-phosphoesterase superfamily protein

, Figure 2).

If we observe the physiological parameters under nitrogen deficiency conditions, we can see that most of the associations occurred with the measurements taken at 70 DAP.

Ten SNP markers associated with chlorophyll content measured at 70 DAP in plants under nitrogen deficiency conditions were found, two on chromosome 1 (ST4.03ch01_72837229 and solcap_snp_c2_20505), three on chromosome 2 (solcap_snp_c2_32462, ST4.03ch02_48054928 and ST4.03ch02_48088436), one on chromosome 4 (PotVar0075324), two on chromosome 5 (ST4.03ch05_51888861 and ST4.03ch05_51733937), another on chromosome 7 (solcap_snp_c2_35078), and the other one on chromosome 11 (ST4.03ch11_2770569) (Table 2, Figure 2).

Although almost all associations were found in measurements taken at 70 DAP, we found one QTL associated with tuber number on chromosome 5 (solcap_snp_c2_15676), two on chromosome 3 associated with leaf area (PotVar0055568 and ST4.03ch03_51965651), and the other one, the marker ST4.03ch03_51965651 on chromosome 3, was also found to be associated with leaf perimeter (Table 2, Figure 2).

4. Discussion

The ANOVA results underscore the significant effects of both genotype (VAR) and nitrogen deficiency stress (STRESS), as well as their interaction, across several agronomic and physiological traits. These findings highlight the complex interaction between genetics and environmental factors, offering valuable insights to enhance crop yield and resilience to nitrogen deficiency stress in potato breeding programs.

The Pearson’s correlation results indicate positive associations between photosynthetic parameters—such as SPAD and NDVI—and traits related to reserve accumulation, including tuber weight, dry matter, and starch content. These results suggest that even under nitrogen deficiency, plants exhibiting better photosynthetic development tend to accumulate more reserves, potentially buffering the negative impact of nutrient stress on yield-related traits.

The results from the Population Structure Analysis and linkage disequilibrium evaluation reveal notable genetic diversity within the studied population, characterized by different structural dimensions, which were incorporated into the association analysis.

The development of new genotyping and sequencing techniques in the last decade has facilitated the identification of many SNPs and the whole genome, selecting trait-associated regions and key genetic elements. Of all these techniques, the GGP Potato 35K array was used in this study, which allowed us to explore the whole potato genome and identify genes of interest [18].

When testing numerous genetic markers simultaneously, the risk of detecting false associations increases. Thus, p-value correction was applied to reduce spurious associations among markers during analysis [43]. We observed that FDR-adjusted values were generally higher than those derived via the Bonferroni method. Although Bonferroni is widely utilized for controlling type I errors, its strictness can lead to missed associations, making FDR a more adaptable alternative [44,45].

This investigation identified the quantitative trait loci (QTLs) associated with chlorophyll concentration measured at 70 days after planting (DAP), as well as tuber number, under non-stressed conditions. A significant positive association was observed between the two variables, suggesting that leaf chlorophyll concentration plays a role in determining the number of tubers.

Under conditions of nitrogen deficiency stress, plants may exhibit reduced photosynthetic activity as a tolerance response, leading to a decline in chlorophyll concentration. In plants, nitrogen deficiency often leads to a reduction in chlorophyll content and inhibits photosynthesis. Nitrogen is a key component of chlorophyll molecules, so, when plants suffer nitrogen deficiency, they attempt to conserve it by breaking down chlorophyll and transferring it to younger parts of the plant.

Chlorophyll content was significantly associated with solcap_snp_c2_15287 SNP under control conditions. solcap_snp_c2_15287 (Soltu.DM.11G023130) was co-localized with a gene encoding for a “P-loop containing nucleoside triphosphate hydrolases superfamily protein”, which is a type of hydrolase that catalyzes the hydrolysis of the beta–gamma phosphate bond of a bound nucleoside triphosphate (NTP), and the obtained energy from this reaction is used to make conformational changes in other molecules [46].

In a study of Arabidopsis, it was found that nucleoside hydrolases (NSHs) are crucial in both intracellular and extracellular purine metabolism. These enzymes facilitate the recycling and degradation of purines, which may also influence the plant’s defense responses [47]. An independent study in rice identified a novel DEAD-box helicase, known as OsABP—an ATP-binding protein belonging to the P-loop NTPase family—as being significantly upregulated under various abiotic stress conditions [48].

Tuber numbers under control conditions were associated with three SNP markers, solcap_snp_c2_37217, ST4.03ch11_2070850, and solcap_snp_c2_15676. solcap_snp_c2_37217 was co-localized with the “Disease resistance protein (NBS-LRR class) family” gene (Soltu.DM.08G003430.1). NBS-LRR proteins are essential components of plant disease resistance, detecting pathogen-derived molecules and initiating immune responses. A key feature of the NBS domain is its NTPase activity. Mutations in the NBS domain can lead to autoactivation of the immune response without the presence of an elicitor, often resulting in an increased susceptibility to diseases [49].

Also associated with tuber number was ST4.03ch11_2070850, which co-localized with the “Di-glucose binding protein with Kinesin motor domain” gene. Although the precise roles of most kinesins remain unclear, their importance in plant cell division is evident. Kinesins contribute to microtubule organization by facilitating bundling and influencing polymerization dynamics. Additionally, they are responsible for the directional transport of cellular components along microtubules [50]. Among the kinesins regulated by the cell cycle, two contain an N-terminal malectin domain, a protein domain known for its ability to bind polysaccharides and peptides when located extracellularly in receptor-like kinases. Sergio Galindo-Trigo et al., 2020, showed that the absence of Malectin Domain Kinesin 2 (MDKIN2) leads to random developmental defects affecting pollen, embryo, and endosperm formation [51].

The solcap_snp_c2_15676 SNP was also related to tuber number. This SNP was associated with the “RNA-binding CRS1/YhbY (CRM) do-main-containing protein”, as well as the solcap_snp_c2_15676 SNP found under nitrogen deficiency stress. The function of this domain remains unclear; however, structural analysis and its occurrence in multiple proteins associated with RNA suggest that it may have an RNA-binding function [52].

When plants were under nitrogen deficiency stress, chlorophyll content was significantly associated with 10 SNPs. One of these QTL was ST4.03ch01_72837229 and was co-localized with glycosyl transferase family 1 protein (Soltu.DM.01G034390.1). Glycosyltransferases are enzymes responsible for transferring sugar molecules to different substrates, playing a key role in glycosylation. In plants, these enzymes contribute to the stability and modification of photosynthetic proteins and pigments, optimizing light absorption and energy conversion [53]. Another SNP, solcap_snp_c2_20505, was associated with 6-phosphogluconate dehydrogenase family protein, which is involved in the pentose phosphate pathway. Studies indicate that 6PGDH activity plays an important role in plant responses to phosphate deficiency, which, in turn, affects photosynthesis and chlorophyll levels. In soybean, for instance, increasing the expression of cytosolic 6PGDH (Gm6PGDH1) has been associated with enhanced root system development and antioxidant defense mechanisms, potentially improving phosphate uptake and overall plant vitality. Given phosphorus’s essential role in photosynthesis, this could indirectly support chlorophyll synthesis and stability [54].

The solcap_snp_c2_32462 and PotVar0075324 markers were co-localized with two different transcription factors. The first one was co-localized with the basic helix–loop–helix (bHLH) transcription factor family. In plants, bHLH transcription factors are involved in various biological functions, including stress responses, hormone signaling, and secondary metabolism. Some bHLH proteins regulate iron homeostasis, while others influence flower development and anthocyanin biosynthesis. A Genome-Wide Association Study revealed that these proteins are expressed in various tissues to respond to abiotic stresses, including salt, drought, and heat [55,56]. The second one was co-localized with WRKY DNA-binding transcription factor. This has a vital role in controlling gene expression, particularly under environmental stress conditions. It helps plants to respond to both biotic and abiotic challenges, such as pathogen invasion, drought, and high salinity. In potato (Solanum tuberosum), WRKY proteins are key regulators of heat, drought, and salt stress tolerance, influencing various defense mechanisms. Gene expression studies reveal that multiple WRKY genes are activated in response to harsh environmental conditions, with cis-regulatory elements linked to drought, heat, and salicylic acid signaling found in their promoter regions. This suggests their involvement in adaptive stress responses and resilience [57,58].

The remaining four QTLs were co-localized with enzymes that are involved in metabolism. The ST4.03ch05_51888861 SNP was related to eukaryotic aspartyl protease family protein. In potato (Solanum tuberosum), aspartyl proteases are involved in protein breakdown, stress adaptation, and defense mechanisms. A key example is Aspartic Protease Inhibitor 5 (StAPI5), which has been investigated for its role in resistance against Potato Virus Y (PVY) and Potato Virus A (PVA) in transgenic potato plants, improving photosynthetic efficiency and stomatal conductivity [59]. The ST4.03ch05_51733937 QTL was co-localized with mitochondrial acyl carrier protein. These proteins are essential for fatty acid synthesis. In potato (Solanum tuberosum), ACPs are responsible for generating lipoic acid, a key cofactor required for various enzymatic activities. Research on Arabidopsis has identified multiple ACP isoforms, indicating that similar proteins likely exist in potatoes, where they contribute to energy metabolism, coenzyme production, and stress resilience [60]. Another enzyme was RAB GTPase homolog A1F. Although studies on this enzyme in potato are rare, findings from Arabidopsis thaliana indicate its involvement in protein transport and signal transduction, affecting cellular structure and stress responses. Additionally, research on Rop GTPases, a subset of Rab proteins, has identified 11 members in potato, with several exhibiting variable expression patterns under biotic stress conditions, highlighting their role in environmental resilience [61,62]. The last SNP, ST4.03ch11_2770569, was co-localized with 3-deoxy-D-arabino-heptulosonate 7-phosphate synthase enzyme, which is the first enzyme in the shikimate pathway, fundamental for the synthesis of aromatic amino acids in plants. It contributes to carbohydrate metabolism with the formation of phenylpropanoids, which participate in plant defense, growth, and secondary metabolic processes [63].

To conclude, the leaf area and perimeter under nitrogen deficiency stress were also associated with three QTLs, and just one associated with the perimeter coincided with one associated with the leaf area. PotVar0055568 was co-localized with aldolase-type TIM barrel family protein. In potato (Solanum tuberosum), fructose-1,6-bisphosphate aldolase (FBA) plays a crucial role in carbohydrate metabolism and photosynthesis. A comprehensive genetic analysis found that most FBA genes possess cis-regulatory elements linked to light exposure and stress adaptation, indicating their potential role in enhancing photosynthetic performance and contributing to microtuber development in plants [64]. An SNP matching both traits, ST4.03ch03_51965651, was co-localized with calcineurin-like metallo-phosphoesterase superfamily protein. In a study with Arabidopsis thaliana, it was found that these enzymes participate in metal-dependent phosphatase activity, influencing protein regulation and stress responses [65].

We detected marker–trait associations for the measured physiological parameters. Prior studies have documented genomic regions and QTLs associated with chlorophyll levels, fluorescence efficiency, leaf dimensions, and tuber production in response to nitrogen-limited stress. Like us, Ospina et al. [22] used an Illumina infinium array in their work. Like them, we also found markers related to the measured agronomic traits, but the related markers were not the same in the two studies [30,66].

Considering both the magnitude of their effect and their statistical values, there are four SNPs that deserve special attention. On the one hand, solcap_snp_c2_20505 is associated with a marked decrease in chlorophyll content. Since it is linked to an enzyme of the phosphorus metabolism, its alteration could compromise the recycling of nitrogenous compounds or limit the production of NADPH, essential for antioxidant defense, which would aggravate the effects of stress due to low nitrogen availability. The second SNP to highlight would be PotVar0075324, also related to SPAD reduction. This marker affects a WRKY-like gene, known for its role in gene regulation under stress. It is possible that it influences pathways related to nitrogen uptake, leaf aging, or hormonal signaling. Thirdly, ST4.03ch11_2070850 shows a considerable decrease in tuber number. Being associated with a kinesin-like motor protein, its alteration could affect the cellular transport of key signals and hormones such as auxins or cytokinins, thus altering tuber initiation and formation. Finally, solcap_snp_c2_15287 has a slight negative effect on the SPAD index. Since it is linked to P-loop NTPase type proteins, its variation could compromise energy-dependent cellular mechanisms, hindering the plant’s ability to adapt to nitrogen restriction.

5. Conclusions

This study aimed to identify SNPs linked to nitrogen deficiency stress across 144 tetraploid potato genotypes using genome-wide association analysis. The detected SNPs were associated with genes involved in plant metabolism, defense mechanisms, cell differentiation, and resistance to PVY (Potato Virus Y). Once validated through genotyping techniques such as kompetitive allele-specific PCR (KASP) markers, these SNPs could play a key role in future potato breeding programs, accelerating marker-assisted selection (MAS) and helping to address the growing population’s agricultural demands.