Introduction
India is the second-largest producer of potatoes in the world after China, having increased its area, production and productivity by 5.75, 18.9 and 3.27 times, respectively, in the last 6 decades (FAO 2022). However, the productivity of the crop remains low at 23.7 tonnes/ha, due to factors including stagnation in yield and lack of availability of quality seed. The unique agro-climate of India allows more than 90% of the potato produce to be cultivated in the north-western plains of India (lying in the sub-tropical region), during the winter months (short day conditions) with a short growing period (mostly 75–90 days). The crop is harvested during the short spring season and is subjected to intense heat conditions during the succeeding summer months. The semi-perishable nature of the crop results in post-harvest losses in quality and quantity contributing towards economic loss to all stakeholders. Losses are further amplified when large stocks of potatoes reach markets in a very short span of time, leading to crashing of prices and distress sale, where growers resort to dumping of produce. Management of the potato crop by way of transportation and cold storage incurs additional charges, as the produce cannot be stored on farms due to soaring summer temperatures. Depending on various factors approximately 10–40% of the produce is reportedly lost in on-farm storage (Mehta and Ezekiel Reference Mehta and Ezekiel2010). Deterioration mainly occurs in the form of sprouting, weight loss and rotting. Changes in firmness and organoleptic properties, affect the quality of the produce reducing its economic value and making it unsuitable for processing.
Post-harvest treatments like radiation, use of sprout retardant chemicals and storage at low temperature are currently being used to reduce post-harvest losses mainly by inhibiting sprouting, which has been reported to cause major weight losses in tubers. However, these increase the price of the crop in the market, raise food safety issues and are not readily available to the farmers for use. Low temperature storage at 2–4°C (90–95% RH), which is the most adopted strategy for long-term storage of potato produce, causes unacceptable sweetening of potatoes, which are neither preferred by consumers nor suitable for processing (Sonnewald Reference Sonnewald2001).
Given the current scenario, the most sustainable strategy to enhance potato production in sub-tropical regions is to breed improved cultivars with superior keeping quality. Breeding such varieties would require evaluation of variability for this character in the existing germplasm, stability in the keeping character and breeding methods suitable for incorporation of the character. Limited reliable information on potato genetic resources, including landraces, breeding lines, imported varieties and wild accessions having good shelf-life attributes is presently available and has not been evaluated for this character in the prevailing conditions. Previously available studies have mostly been restricted to keeping evaluation of only a few released varieties or advanced hybrids for their keeping quality under ambient or non-refrigerated storage like heaps, pits and evaporatively cooled potato stores (Das et al. Reference Das, Ezekiel, Pandey and Singh2004; Mehta et al. Reference Mehta, Singh, Pandey and Ezekiel2006; Pande et al. Reference Pande, Singh, Pandey and Singh2007; Gupta et al. Reference Gupta, Luthra and Singh2015).
Augmented designs are a valuable tool in breeding programs, particularly in potato, where the limited availability of tubers in conserved germplasm collections necessitates multi-year evaluations for various traits. In such cases, storage losses – often varying across years due to differences in pre-storage and storage conditions can be excluded as block effects, using replicated checks. This enables breeders to minimize their impact on trait evaluation and supports more reliable comparisons among genotypes. Adjusted means derived from the augmented design facilitate robust selection decisions and can be effectively utilized in multivariate analyses to identify promising parental lines (Flores et al. Reference Flores, Gutierrez, Lopez, MT and JI1997; Lotti et al. Reference Lotti, Marcotrigiano, De Giovanni, Resta, Ricciardi, Zonno, Fanizza and Ricciardi2008). Principal component analysis (PCA) (Karl Pearson et al. Reference Pearson1901) has been widely adopted as the method of choice by breeders for evaluating their germplasm collections and making useful selections based on phenotypes (Bharadwaj et al. Reference Bharadwaj, Satyavathi and Subramanyam2001; Hailu et al. Reference Hailu, Merker, Harjit-Singh, Belay and Johansson2006; Maji Reference Maji2012; Zia et al. Reference Zia, Demirel, Nadeem, Ali, Dawood, Ijaz and Çalişkan2022). It has also been widely used for genetic diversity analysis using biochemical and molecular data.
Current breeding programs in the region primarily focus on developing high-yielding, early to medium maturing (75–90 days) table potato varieties suited for cultivation under short-day (winter) conditions in the subtropical plains. Most studies focus on potato yield and other characters regulating biotic and abiotic stresses. However, there are relatively few investigations on the keeping quality of potatoes, which is the focus of this investigation. There is a significant untapped potential to enhance productivity by improving the keeping quality, which is a key attribute determining the overall productivity of a potato variety under subtropical conditions, and there is a lack of large-scale germplasm evaluations for this key trait. This novel study addresses this shortcoming by evaluating diverse genotypes including S. tuberosum ssp tuberosum landraces, primitive types, varieties and improved breeding lines from the germplasm collection to identify Solanum tuberosum accessions with superior keeping quality, which can be channelized to breeding programs to enhance the post-harvest resilience of future varieties, better equipping them to withstand the prevailing storage conditions in the region and ultimately increasing the profitability for potato growers.
Materials and methods
Plant Material and crop management: The experimental material consisted of 540 germplasm accessions (Supplementary Table S1) belonging to the Indian Solanum tuberosum ssp tuberosum germplasm collection maintained at ICAR-Central Potato Research Institute Regional Station Jalandhar, Punjab, India (30ºN and 75ºE at an elevation of 237 metre above mean sea level) under field multiplication. The said collection is the national active germplasm site for potato in India maintaining over 4,500 accessions comprising approximately 3,200 lines of S. tuberosum, 762 of S. andigena and 540 clones of semi-cultivated and wild species, conserved in situ and in vitro. These accessions have been procured from about 30 countries, predominantly from Peru, which is the recognized centre of origin of potato and home to the largest diversity of wild species and primitive forms. In the present study, the evaluated 540 S. tuberosum ssp tuberosum accessions comprised of landraces, primitive types, S. tuberosum varieties and advanced breeding lines introduced in India, with probable wild species parentage. The genotypic evaluation was carried out under ambient conditions over a 10-year period (2010–2020), involving 72–103 accessions each year. The collection included potato varieties originated from European countries, North and South America, Australia, Africa and Asia.
The evaluated tubers were harvested from crop planted during the first week of October of the respective year. All recommended agronomic practices for cultivation of the crop were followed throughout the growing season. Dehaulming was carried out 90 days after planting, with harvesting done 15–20 days later to ensure proper skin maturity and firmness. After harvest, clean and dry tubers were skin-cured under controlled conditions (10–20°C and 85–95% relative humidity) for 15–20 days to promote wound healing and enhance skin set.
Each test accession consisted of a uniform set of 20 tubers (50–70 mm in size, free from bruises or skin damage) were selected. These tubers were stored in small bags at room temperature and evaluated alongside control varieties Kufri Dewa, Kufri Pukhraj and Kufri Ashoka. Keeping quality parameters of these accessions were evaluated from March to May of respective years. The major parameters included sprouting percentage (SP), weight loss percentage (WL), rotting percentage (R), Total weight loss (TWL) and firmness (F). Observations were taken at 30, 60 and 90 days interval from initial date of storage.
Weight loss percentage (WL): The representative set of 10–20 freshly harvested, soil-free tubers was selected and weighed at the initiation of the experiment. Physiological weight loss (%) was calculated using the following formula:
\begin{equation*}WL\left( \% \right) = \frac{{Wi - Wf}}{{Wi}} \times 100\end{equation*}where Wi is the initial weight at the start of storage and Wf is the final weight of the tubers at each observation period.
Sprouting percentage (SP): Sprouting was evaluated on a numerical basis from the set of stored tubers. A tuber was considered sprouted when at least one visible sprout attained a minimum length of 0.2 cm. Sprouting percentage was calculated as percent sprouted tubers out of total tubers stored (Fig. 1).
Differentiating sprouted tubers A from unsprouted B (<2 mm sprouts mild chitting or no visible sprouts).
Figure 1 Long description
The image shows two groups of tubers placed on a textured surface. The group labeled A consists of six tubers, while the group labeled B contains seven tubers. Both groups are arranged closely together, with each tuber displaying a similar size and shape within its respective group.
Rotting percentage (R): Rotting percentage was calculated as weight of rotted tubers out of total, where the stored tubers were examined externally for symptoms of rotting such as softness, discoloration or tissue breakdown. Tubers exhibiting such symptoms were cut open to confirm internal rotting. The weight of rotten tubers was recorded, and rotting percentage was also expressed as percent weight of rotten tubers out of total weight of stored tubers.
Total weight loss: The sum total of physiological (WL) and rotting percentage (R) was taken as the total weight loss. The character is a better determinant of keeping quality parameter of the accession. It was calculated similar to the weight loss (WL).
Firmness: Firmness was evaluated by manual pressing of the tubers against flat hard surface. Firmness of freshly harvested tubers were recorded as 100% (scale 5) while firmness of tubers which can be pressed up to < 2 mm as 81–100% firmness (scale 5); 1–2 mm were recorded as 61–80% (scale 4); 2–3 mm as 41–60% (scale 3); 3–5 mm as 21–40% (scale 2) and 5–10 mm were recorded as 1–20% (scale 1) (Fig. 2). None of the stored tubers were given firmness score of 5.
Scoring firmness in potatoes on the scale of 1–5 based on tuber quality and compression (where 1 is most firm and 5 is least).
Figure 2 Long description
Five rows of potatoes are displayed, each labeled with numbers from 1 to 5. The potatoes in row 1 appear most firm, while those in row 5 appear least firm. Each row contains four potatoes, arranged horizontally. The labels indicate a scale of firmness, with 1 being the most firm and 5 being the least firm.
In order to derive conclusive results, evaluation at 90 days was used in analysis of keeping characters of these germplasm accessions. Firmness was further converted to percent data. All data were transformed using square roots (√x + 0.5) and arc sine transformations based on the spread and type of observations (Gomez and Gomez Reference Gomez and Gomez1984).
Statistical analysis: Data compiled across years were analyzed using an augmented design to account for climatic and environmental variability across seasons, with block effects adjusted through replicated standard control varieties (Kufri Dewa, Kufri Pukhraj and Kufri Ashoka). This approach allowed the estimation of adjusted means and derivation of analysis of variance (ANOVA) across years and accessions, thereby facilitating meaningful comparisons among test accessions and controls.
To explore complex trait interrelationships, multivariate analysis was performed using PCA. The number of relevant principal components (PCs) was determined through a scree plot using the elbow method. Biplots of the selected PCs were generated in MS Excel, enabling visualization of accessions along with eigenvectors and their respective loadings. Test genotypes were labelled according to their spatial distribution within the most informative biplots.
A Venn diagram, based on the best biplot combination of keeping quality PCs, was constructed to identify genotypes exhibiting the most favourable combination of storage-related traits. The shortlisted genotypes were further validated against base data to confirm consistency of results.
All augmented design and PCA analysis were conducted using the freely available OPSTAT software (http://14.139.232.166/opstat/; Kumar et al. Reference Kumar, Rani, Niwas, Sheoran and Malik2020; Sheoran et al. Reference Sheoran, Kumar, Poonia and Malik2020), while MS Excel was employed for graphical representation and interpretation of results.
Results
The keeping quality parameters of accessions were evaluated: firmness ranged from 1 to 4.2, sprouting from 0 to 100% rottage from 0 to 74%, weight loss from 0.19 to 63.64% and total weight loss from 5.17 to 97.9% variation.
The mean and descriptive statistics of untrasformed data provided a brief outline of the spread of the data over years. Coefficient of variation for base data and augmented data was used to evaluate changes in the spread of the data after augmentation, where minor changes in the spread of augmented (adjusted means) data as compared to the base transformed data were observed (Supplementary Table S2). Maximum variance was observed in case of F and TWL.
The analysis of variance for augmented design showed significant variation among the germplasm accessions for TWL and SP (Supplementary Table S3). The block effects or yearly variation were observed to be significant for SP and F. The check vs varieties variation was significant for WL, TWL and F.
Critical difference (CD) values were obtained in the study (Supplementary Table S4) to evaluate significant variation between check means, between adjusted variables in the same block, between adjusted variables in different blocks and between adjusted variety mean against check means.
PCA was carried out to further resolve the data and identify good combination of keeping characters. The scree plot generated from the data depicted the first three eigen vectors to contribute 86.3% to the total observed variability (Supplementary Fig. 1). Eigen vectors suitably depicted the coefficients which associated positively or negatively with the PCs and indicated the contribution of each variable to a selected PC (Supplementary Table S5).
The correlation matrix of loadings with PCs is shown in Supplementary Fig. 2. PC1 contributed positively to TWL (0.68), R (0.50) and WL (0.37) and negatively with F (−0.35). PC2 showed a negative correlation with SP (−0.76) and positive association with F (0.54). For PC3, there was a negative correlation for R (−0.65) and a high positive association for WL (−0.74).
The three biplots BP1 (PC1 X PC2), BP2 (PC1 X PC3) and BP3 (PC2 X PC3) were used as the graphical tools to understand and interpret the results of the PC analysis. The different zones of the biplot with good keeping germplasm accessions were demarcated based on the correlation of PC with eigen vectors (Figs. 3–5). The data points lying in the demarcated zones of the three biplots were labelled in MS Excel (not shown in the figure) to identify the good keeping accessions. It was observed that 114, 152 and 75 accessions could be shortlisted based on the three biplots BP1, BP2 and BP3, respectively, falling in the demarcated zones. A Venn diagram for these identified accessions was generated to further filter the best keeping accessions, where 18 accessions having the best combination of characters were identified (Supplementary Fig. 3). The reported keeping characters of these accessions were validated with the base information available. These along with the other attributes in relation to the breeding behaviour of the identified accessions have been depicted in Table 1.
Biplot BP1 displaying eigenvectors and scores for principal components 1 (PC1) and principal components 2 (PC2).
Figure 3 Long description
The biplot displays principal components PC1 on the x-axis and PC2 on the y-axis, with percentages 38.9 percent and 24.1 percent respectively. Data points are represented by triangles scattered across the plot. Several eigenvectors are drawn from the origin, pointing in different directions. An ellipse encloses a cluster of data points, indicating a specific grouping. The plot includes labels for S, R, W over L, T over W over L and F, each represented by different line styles. The axes are marked with numerical values ranging from negative to positive, providing a scale for the components.
Biplot BP2 displaying eigenvectors and scores for principal components 1 (PC1) and principal components 3 (PC3).
Figure 4 Long description
The biplot displays eigenvectors and scores for principal components 1 (PC1) and principal components 3 (PC3). The x-axis is labeled PC1 with a percentage of 38.8 percent and the y-axis is labeled PC3 with a percentage of 23.3 percent. The plot includes a large number of triangular data points scattered across the graph. An ellipse is drawn around a cluster of points on the left side of the plot. Several vectors are labeled: S, R, WL, TVWL and F, each represented by arrows pointing in different directions. The vectors are positioned among the data points, indicating their influence on the principal components.
Biplot BP3 displaying eigenvectors and scores for principal components 2 (PC2) and principal components 3 (PC3).
Figure 5 Long description
The biplot displays eigenvectors and scores for principal components 2 (PC2) and principal components 3 (PC3). The x-axis is labeled PC3 with a percentage of 23.3 and the y-axis is labeled PC2 with a percentage of 24.1. Two zones, Zone A and Zone B, are marked with ellipses. Various data points are scattered across the plot, represented by triangles. The legend on the right side lists variables: S, R, WL, TWL and F, each associated with an arrow indicating direction. The arrows in the plot point towards different directions, representing the influence of these variables on the principal components.
Untransformed keeping attributes of the selected germplasm accessions and important characteristics of selected germplasm related to breeding for good keeping quality for the northern plains region
Table 1 Long description
The table provides a comprehensive overview of germplasm accessions from countries like Peru, Canada, and India, highlighting attributes relevant to breeding for good keeping quality in the northern plains. Key data points include tuber shape, eye depth, tuber and flesh color, flowering intensity, pollen fertility, and adaptability to short day conditions. Notable entries include 'HUACHAPA' with purple-red tubers and 'RUSSET BURBANK' with brown tubers, both showing different adaptability and firmness levels. The table also measures dry matter content, sprouting, rottage, weight loss, and firmness, with 'KUFRI DEWA' showing high dry matter and firmness. Some entries lack complete data, indicating areas for further evaluation.
NE: not evaluated; PER: Peru; Can: Canada; USA: United States of America; BGD: Bangladesh; IND: India; EP: Eastern plains; NP: Northern plains; P: Poor; A: Average; G: Good; LB: Late Blight; IV: In vitro; TU: tuber form; OB: Oblong; RO: Round; OV: Ovoid; M: Medium; SH: Shallow; D: Deep; P: Protruding; H: High; L: Low.
Source: ICAR-Central Potato Research Institute Technical bulletin no. 70.
Discussion
The study clearly identified 18 accessions having the best combination of characters for post-harvest resilience, which have delayed sprouting, reduced weight loss and rotting. These accessions represent immediate candidates for breeding programs and longer-term strategic resources for subtropical potato production. Their utility lies not only in improving the storage and marketability of potatoes but also in contributing to the broader objectives of sustaining cultivation in stress-prone subtropical regions, enhancing climate resilience and strengthening food security.
Extended dormancy, stronger periderm integrity, higher firmness and resistance to storage rottage are key contributors to good keeping quality in potato (Singh and Ezekiel Reference Singh and Ezekiel2003). In this study, the strong association between weight loss (WL) and sprouting (S) agrees with earlier findings that sprout initiation greatly enhances water loss due to the high permeability of sprout epidermis (Singh and Ezekiel Reference Singh and Ezekiel2003). Pande et al. (Reference Pande, Singh, Pandey and Singh2007) also reported significant WL–S correlations. Thicker, suberized periderm reduces both moisture loss and pathogen entry, thereby influencing storability (Ezekiel et al. Reference Ezekiel, Singh, Sharma, Garg and Khurana2004). Firmness was negatively correlated with WL and S, supporting the view that delayed sprouting helps preserve tissue integrity (Ezekiel et al. Reference Ezekiel, Singh, Sharma, Garg and Khurana2004). Genotypes combining low WL and S with high firmness consistently perform as superior keepers. Such genotypes with lower respiration and transpiration thereby maintain higher marketable weight and visual quality (Luthra et al. Reference Luthra, Gupta, Lal and Tewari2018).
The practical implications of these findings are substantial. Subtropical regions are increasingly emerging as new potato production frontiers, reflecting both rising demand and the crop’s adaptability to non-traditional ecologies. However, such environments impose stresses of elevated temperature, humidity and pathogen pressure, which accelerate deterioration. In these regions, ambient storage remains the predominant practice and access to refrigerated facilities is limited, post-harvest losses pose significant constraints to profitability. The identification of genotypes with reduced WL and TWL, delayed S and improved F directly addresses these challenges by providing cultivars with inherently superior keeping quality. High phenotypic variation for the character TWL and S observed in the germplasm collection is supported by previous reports by Luthra et al. (Reference Luthra, Gupta, Lal and Tewari2018), who also reported higher variability for keeping parameters weight loss and sprouting during their evaluation of 30 advanced hybrids. These characters have also been defined as the primary determinants of storage performance in potato by Burton (Reference Burton and Burton1989) and Ezekiel et al. (Reference Ezekiel, Singh, Kumar, Paul and Das2013).
The study demonstrates considerable potential for improving varieties by reducing losses during potato storage. The inclusion of these shortlisted 18 accessions in the breeding programme will contribute to the development of potato varieties with good keeping attributes. Farmers adopting such new varieties would benefit from extended marketability windows, reduced storage-related losses and a more stable market supply. At a global level, the findings contribute to ongoing efforts to enhance climate resilience and food security. The results could also contribute to further studies to enhance knowledge about the regulation of mechanisms controlling this character.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1479262125100415.
Author contributions
All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Ashiv Mehta, Arvind Jaiswal and Brajesh Singh. The data analysis and interpretation were carried out by Ratna Preeti Kaur. The first draft of the manuscript was written by Ratna Preeti Kaur, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.
Funding statement
This work was supported by a research grant received from the Indian Council of Agricultural Research, India.
Competing interests
The authors have no relevant financial, non-financial or competing interests to disclose.
Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request
Statement detailing use of AI
The authors used ChatGPT, an AI language model developed by OpenAI, to improve the readability and clarity of the manuscript draft. All intellectual content, data interpretation and conclusions are solely those of the authors
