Ranking of the 140 final raw traits from the literature review base on trait type: root, flower (FLWR), seed, fruit, and vegetative (VEG).
Ranking of the 140 final raw traits from the literature review based on frequency of appearance in the articles, with High being the most frequently found traits and Niche appearing in five articles or less.
Crop ontology workflow beginning with the literature search and ending an ontology based on ∼280 traits.
Examples of seed traits included in the bottle gourd crop ontology, including seed tip shape (top row), seed color pattern (middle row), and Seed wings - excess seed coat protruding from the margin of the seed (bottom row).
Examples of the diversity of bottle gourd fruit. From standardized phenotyping methods, fruit metrics such as shape, color, pattern, dimension, luster, and texture can be extracted from a single image.
Click on author name to view affiliation information
Underutilized crops have considerable cultural, culinary, and historical value but lack widespread cultivation or extensive research. With these gaps in commercialization efforts, maximizing available information is crucial for breeding adapted and improved cultivars, conserving genetic resources, and, ultimately, promoting broader adoption. Crop ontologies provide a framework for describing a crop’s relevant attributes by standardizing data collection protocols across various research endeavors. These ontologies enhance the value of the broader pool of genetic resources and facilitate greater interoperability among collaborative conservation and improvement efforts. To maximize impact, a crop ontology should prioritize the inclusion of traits from markets, cultures, and regions with connections to the given crop. The bottle gourd [Lagenaria siceraria (Molina) Standl.] is an underutilized, under-researched crop originating in Africa, but has regional significance across Europe, Asia, and the Americas. We developed a crop ontology for bottle gourds via literature reviews using 35 previous characterization studies from 10 countries, related cucurbit ontologies, and multisite and multiyear collaborative phenotyping efforts with Kasetsart University (Thailand), the US Department of Agriculture National Plant Germplasm System (United States), Cornell University (United States), and the University of Erciyes Melikgazi Kayseri-Türkiye (Turkey). The crop ontology emphasizes traits important for localized use and includes 300 traits that describe vegetative, floral, fruit, and seed phenotypes critical to horticultural variation and culturally diverse uses. Furthermore, our bottle gourd ontology provides a foundation for future conservation and improvement efforts, as well as a framework for creating ontologies that could be applied to other underutilized crops.
Lagenaria siceraria (Molina) Standl., commonly known as the bottle gourd or calabash, is a vining crop in the Cucurbitaceae family, which broadly includes gourds, watermelons, melons, squash, and cucumbers. Bottle gourds are considered one of the earliest domesticated crops, with their use dating to more than 10,000 years ago (Chomicki et al. 2020). They have been widely adapted to diverse climates and production systems, including nonirrigated deserts, and high-humidity tropical and temperate conditions, with production as far north as Canada and northern Europe. Evidence suggests that bottle gourds originated in southern Africa but traveled across continents and oceans, leading to two additional independent centers of domestication in Eurasia and South America, helping expand their cultural significance to nearly every continent (Kistler et al. 2014; Zhao et al. 2024). Bottle gourds are commonly referred to as underutilized crops because they fit the common definition of crops that are regionally important, often with strong historical and cultural ties, but lack the research and resources for improvement that many major crops such as corn and soy have received (Padulosi et al. 2001).
Despite their modern classification as an underutilized crop, bottle gourds have been used historically by various groups of people for diverse purposes, as demonstrated by its use as a food crop and as a raw material to create a wide array of practical, everyday objects (Fig. 1). As food, the immature fruits, leaves, stems, seeds, and flowers are consumed by humans and animals, and used in traditional medicine (Ahmad et al. 2022; Dilshad and Sunhail 2022; Islam et al. 2021). The mature, dried fruits are used to make durable goods and containers, including water vessels, musical instruments, relics, and art. In addition to bottle gourds’ cultural use and use as a food crop, they have also been recognized as a valuable source of rootstock. Bottle gourds have been used to confer abiotic stress tolerance, such as tolerance to salinity (Huang et al. 2009; Yan et al. 2018), drought conditions (Morales et al. 2023), low temperature (Jang et al. 2022), and heavy metals (Nawaz et al. 2018); biotic stress tolerance, which includes resistance to pest and diseases such as powdery mildew (Kousik et al. 2018), gummy stem blight (Mahapatra et al. 2023), root-knot nematodes (Özarslandan et al. 2011), fusarium wilt (Huh et al. 2012), and crown rot (Kousik et al. 2012); increased yield (Karaca et al. 2012; Suárez-Hernández et al. 2022; Yetişir and Sari 2018); and improved nutritional value to other cucurbits (Çandir et al. 2013; Karaca et al. 2012; Mahapatra et al. 2023), most notably watermelons, when used as a rootstock. These resistance traits have additional implications for using bottle gourds as a climate-smart crop, especially in regions affected by drought and increasing temperatures (Mkhize et al. 2024).
Citation: HortScience 60, 10; 10.21273/HORTSCI18689-25
Despite the multifaceted importance of bottle gourds culturally and historically, and their source of resistance, global acreage devoted to them is a fraction of major vegetable crops. For example, during 2023, the aggregated group of pumpkins, squash, and gourds—which includes bottle gourds along with a plethora of cucurbit crops from other genera, such as winter squash and pumpkins (Cucurbita sp.), winter melons (Benincasa hispida), snake gourds (Trichosanthes cucumerina), and bitter gourds (Momordica charantia)—had a harvested area of only 1.5 million ha, whereas tomatoes alone had an estimated total global area harvested of 13.3 million ha (FAOSTATS 2025). India is one of the largest producers of bottle gourds (Islam et al. 2021). Between 2018 and 2019, bottle gourd production in India comprised ∼186,000 ha of a total estimated 10.1 million ha used for producing vegetable crops, which equates to only ∼1.8% of production (National Horticulture Board Ministry of Agriculture & Farmers Welfare Government of India 2019).
This lack of broad-scale production is a result, in part, of the scarcity of high-performing cultivars. Although many bottle gourd landraces are still used in small-scale subsistence farming and home gardening, few high-performing cultivars have been commercialized for the market. In a survey of several prominent international vegetable seed companies, we found East-West Seed (Nonthaburi, Thailand) has three varieties (East-West Seed 2024), HM Clause (Davis, California, USA) has four varieties in their India program and one variety in their Vietnam breeding program (HM Clause 2025), and Bayer’s Indian branch (Maharashtra, India) currently has only a single variety available (Bayer AG 2024). This is in stark contrast to the variety availability in related crops such as squash and pumpkins (Cucurbita sp.). Johnny’s Selected Seeds (Winslow, Maine, USA), a popular US-based seed company, alone carries 82 squash varieties and 78 pumpkin varieties (Johnny’s Selected Seeds 2025).
The issue of germplasm accessibility is also present in germplasm repositories worldwide. Although there are germplasm repositories that hold hundreds of bottle gourd germplasm, many of the accessions are unavailable for distribution, and those that can be requested for research purposes have limited characterization data available (Table 1). Currently, the US Department of Agriculture (USDA) National Plant Germplasm System (NPGS) Germplasm Resources Information Network (GRIN)–Global (USDA-ARS 2025a) holds the most known available accession of bottle gourd with 197 accessions. Of these 197 accessions, only three have seed images, and 97 accessions have limited fruit and maturity metrics. Similar trends were observed across other bottle gourd collections, with only a handful of accessions having available information. This lack of data makes it difficult for germplasm curators, growers, plant breeders, and other researchers to make informed choices about which accessions capture specific aspects of diversity or target particular market types. Whether it be creating a core collection of the accessions that capture the maximum amount of diversity, breeding better regionally adapted cultivars, or better understanding pathways of secondary metabolites, without characterization data, picking accessions that are ideal candidates for the research is like drawing names blindly from a hat.
Underutilized crops such as bottle gourds face significant barriers to research and conservation because of a lack of information, be that for characterized germplasm, standardized and accessible passport, phenotypic data, and a limited research community. This dynamic creates a challenging, Catch-22–type cycle that underutilized crops face toward improvement efforts. Despite their cultural importance and potentially high societal or economic value, data on underutilized crops are frequently limited. In some cases, no data may be available, whereas in others, data are poorly documented and stakeholders have limited familiarity with the crop or a poor understanding of its contextual use. This paucity of preliminary research can, in turn, limit new public investment in research, creating additional barriers to knowledge acquisition that are required to drive broad adoption.
This is incredibly challenging for crops with traits that vary by market, region, or culture. For instance, the lack of consistency in terminology for descriptors such as crunchy or hard to describe fruit texture can lead to vastly different interpretations and perceptions of desirability, depending on linguistic and cultural contexts. A single trait may go by multiple synonyms or name variations. Across crops, traits to measure seed weight may be called “100 grain weight,” “weight of 100 seeds,” or “hundred-seed dry weight” (Andrés-Hernández et al. 2021). All three could be listed separately, but essentially measure the same underlying phenotype. When researchers work with underutilized crops, even using existing terminology for other crops can create confusion, as one trait may be assigned multiple names depending on the resource used to select the terminology.
Ontologies play a crucial role in addressing these challenges. Crop ontologies (sometimes called trait ontologies) provide a systematic framework and vocabulary for standardizing data collection and making information more findable, accessible, interoperable, reusable (FAIR) and, in cases of culturally important crops, include collective benefit, authority to control, responsibility, ethics—or CARE—principles (Carroll et al. 2020; Hauschke et al. 2021). By creating universal crop descriptors that cover attributes such as fruit shape, texture, and taste, researchers can ensure that data from different studies are comparable and useful across regions, disciplines, and market types. These standardized descriptors serve as a foundation for further research, promoting collaboration and unlocking the full potential of crops such as bottle gourds.
We developed a bottle gourd crop ontology with three core objectives: 1) to strive to capture comprehensively all relevant crop attributes to all stakeholders across market classes and end uses, 2) to generate a standardized and well-defined data structure to maximize data comparison and utility across regions and disciplines, and 3) to provide a foundational framework for further research, promote collaboration, and improve the accessibility of bottle gourds on a global scale. Through the development and application of a comprehensive and precise crop ontology, bottle gourds can transition from an underutilized crop to one with renewed global significance, bridging the gap between traditional knowledge and modern agricultural research with FAIR public data.
Globally, only a few breeding programs focus on bottle gourds, with limited international cooperation through consortia or agency-based collaboration. Therefore, a literature review was conducted to generate an initial, broad, and inclusive preliminary list of traits. Google Scholar was used to find articles because of its ability to search for articles across both major and minor journals. Search terms related to crop descriptors, characterization, diversity, and variation were used to identify papers containing trait information, including Lagenaria, bottle gourd, gourd, germplasm, characterization, crop description, trait description, morph*, diversity, and variet*. Searches included articles written at any time, and articles were sorted by relevance to the search terms. Searches were limited to articles written in or translated into English. The articles identified from the initial search were filtered to only those that contained characterization data and/or trait descriptions, from which “raw” traits were extracted.
Each raw trait was compared with other similarly described raw traits and merged where overlap occurred. For traits with multiple names, the most commonly used name was maintained as the primary name, with any other variant names recorded in the Synonyms column of the crop ontology template. Each raw trait was added to the crop ontology format created by Breeding Insight, which was simplified based on the full template from Pietragalla et al. (2024). Raw traits that appeared to assess more than one trait were broken into multiple, separate traits to improve clarity and specificity. To annotate trait frequency and market-specificity more comprehensively, raw traits were ranked by frequency across the selected articles used for trait extraction and grouped by trait type (e.g., seed, vegetative, flower, root, fruit).
Additional trait information was collected directly from stakeholders from several countries to increase the global scope. The American Gourd Society (Kokomo, IN, USA), curators from the USDA NPGS (Geneva, NY, USA & Griffin, GA, USA), the Tropical Vegetable Research Center at Kasetsart University Kamphaeng Sean, Thailand and the Department of Horticulture Faculty of Agriculture of the Kayseri, Türkiye provided feedback on the ontology. Traits in the ontology were also compared with existing crop descriptor guides for bottle gourds and related cucurbit (e.g., Cucurbita sp.) species from the Protection of Plant Varieties and Farmers’ Rights Authority Government of India, NPGS GRIN-Global, European Cooperative Program of Plant Genetic Resources, and the Tropical Vegetable Research Center at Kasetsart University Kamphaeng Saen.
The initial literature search yielded ∼70 articles with search term hits. All initial articles found were written or translated into English. After reviewing each article, 35 contained characterization data and/or trait descriptions (Table 2). Together, these articles represented the phenotypic evaluation of bottle gourd germplasm from 10 different countries across four continents: Africa, Asia, Europe, and North America.
A total of 154 raw traits were initially extracted from the literature. Overlapping traits were merged, and one trait was removed, because of a lack of description of what the trait measured, which resulted in a final total of 140 raw traits. Each trait was grouped by trait type, with the categories being seed, vegetative, flower, root, and fruit (Fig. 2). Approximately 40% of the raw traits (55 of 140) were categorized under vegetative traits, which included metrics on the leaves, vines, tendrils, peduncles, cotyledons, and hypocotyls. The next largest category was fruit traits, which was 30% (n = 43) of the total traits, and included measures of fruit dimensions, coloring and pattern of the skin, flesh texture and nutritional qualities, and yield. Root, flower, and seed categories represented the smallest percentage of traits: ∼2%, ∼12%, and ∼16%, respectively. Flower traits included 17 traits that represented flower dimensions, flowering date, number of flowers, and flower color. The 23 seed traits spanned whole seed and embryo dimensions, color, texture, and germination metrics. The root trait category included only three traits: root length, root fresh mass, and root dry mass.
Citation: HortScience 60, 10; 10.21273/HORTSCI18689-25
The 140 raw traits were also ranked based on the number of appearances across the selected literature to understand more fully the utility of traits (general vs. niche) (Fig. 3).
Citation: HortScience 60, 10; 10.21273/HORTSCI18689-25
High-general traits were found in at least ∼43% of the articles and included only 11 of the 140 raw traits. These high-general traits included basic dimensions of seeds, vines, and fruit, as well as flowering dates (Table 3). The rank of moderate-general traits included 16 traits found in ∼17% of the articles. Moderate-general traits included additional fruit phenotypes, such as shape and color, and leaf phenotypes (Table 3). The remaining 113 traits were categorized as niche traits, with limited appearances (in five articles or fewer). These niche traits addressed the diverse uses of bottle gourds, including disease resistance pertinent to locale, nutritional quality, and more detailed morphology such as the shape, color pattern, and texture of fruits, seeds, and leaves. The high number of niche traits was anticipated because of the vast diversity of market types and uses of bottle gourds at different life stages and across cultures.
After further assessment of trait and method descriptions, additional refinement was necessary to resolve more overlapping traits and create new, separate traits that measured multiple phenotypes. This refinement resulted in a total of 198 traits that were sent to stakeholders for more feedback. From these responses, and by comparing to existing bottle gourd and related cucurbit trait descriptor guides, additional traits were added for a grand total of 300 traits. The full workflow process of annotation is summarized in Fig. 4.
Citation: HortScience 60, 10; 10.21273/HORTSCI18689-25
The bottle gourd crop ontology includes traits for vegetative structures, flowers, fruits, seeds, and roots, and covers a broad range of measurements such as size, color, olfactory, density, nutrition, texture, and disease and pest resistance (Figs. 5 and 6). The ontology is included in Supplemental File 1, and method description citations in Supplemental File 2. The ontology will be made available through the US Department of Agriculture–Agricultural Research Service (2025b).
Citation: HortScience 60, 10; 10.21273/HORTSCI18689-25
Citation: HortScience 60, 10; 10.21273/HORTSCI18689-25
Species that have been able to adapt to a wide range of areas and have historical use, but have been neglected in recent times, are among our most essential resources in the fight against climate change and its consequences for human and animal nutrition. Bottle gourds, which have a cultural age of 12,000 years (Erickson et al. 2005), are one of the most widely recognized underutilized species, with their vibrant history and diversity of cultural uses that span across continents. Creating common ontological standards and a unified scientific language for the identification and evaluation of underutilized species is a valuable and impactful effort in terms of public data and scientific communication.
Our bottle gourd crop ontology provides a critical resource to help renew the bottle gourd’s global significance by leveraging data across available studies and small but active research programs. This comprehensive ontology encompasses 300 traits that relate to all stages of plant life, and represents diverse uses and market types. Although the most common traits included in studies focus on metrics often collected on plants targeting high fruit yield, there is a vast range of aims for bottle gourd breeding programs. These potential breeding targets can be identified by the majority of their traits in the ontology classified as niche. These niche traits represent regionally important breeding goals, including the production of edible seed, improving the nutritional quality of edible greens, and preserving diversity for cultural uses. By encompassing traits of all market types of bottle gourds, the ontology can be used to meet the needs of breeders, researchers, farmers, and other stakeholders across markets, regions, and cultures.
The ontology can take its impact a step further by facilitating the creation of a common practice for data collection. Collecting data by adhering to a crop ontology provides a standardized procedure for measuring traits and replicating those measurements from season to season. Standardization of collection renders the resulting data more valuable, allowing for the aggregation of data from studies across disciplines, regions, and time periods to perform more in-depth analyses. Without this standardization, crop data are disjointed, requiring researchers to spend excessive time aligning data from different studies, even in the best cases, or having to discard or ignore uninterpretable data in the worst cases. Data standardization is crucial for overcoming research barriers and improving accessibility in underutilized crops. As many studies currently done on underutilized crops such as the bottle gourd collect data on a small pool of germplasm grown locally, having the capacity to pool data from many small-scale studies can provide more power to accomplish large-scale research objectives. Instead of having intelligible, accessible data on only 10 to 30 varieties of the crop, combining data collected using a standardized procedure across multiple studies allows for the creation of a richer dataset that contains hundreds of varieties across various locations.
Although the current state of global bottle gourd research has been fragmented, future studies can be aligned easily by using our ontology. Because the ontology was built on current and past research, it may also serve as a helpful tool in deciphering synonymous traits to make the dataset more compatible and to create larger, accessible datasets. Larger compatible datasets lay the foundation for creating digital resources that combine phenotypic and genomic data for researchers, breeders, and other stakeholders, such as the Sol Genomic Network for Solanaceae family crops, which holds genetic and trait data, and has analysis tools for Basic Local Alignment Search Tool (BLAST), genome browsers, and quantitative trait loci (QTL) analysis all combined into an online user-friendly interface (Fernandez-Pozo et al. 2015). When working with a larger dataset, more substantial objectives can be achieved. Compiled datasets from ontologies can provide the foundation for integrating phenotypic data with genetic and genomic datasets. Integrating these datasets catalyzes breeding programs, enabling trait–marker associations, QTL mapping, and genome-wide association studies. By using these methodologies, improved cultivars that are disease resistant, high yielding, and locally adapted can be developed more efficiently, facilitating widespread accessibility of the crop and connecting cultural knowledge with genomic data, while exploring new market opportunities for using the crop. Some of the groundwork has already been completed, with a range of smaller scale international studies already published on resistance to stressors present in bottle gourd populations. These studies can be combined using the bottle gourd ontology as a reference and by adding to the ontology when needed. When combined with genomic data, we can begin to build a unified phenotypic and genomic resource for bottle gourds.
These compiled datasets are also foundational resources for gene banks. Germplasm accessions that have more characterization data are correlated with higher “useful”(-ness) to requestors (Rubenstein et al. 2006). Standardized phenotypic data collected on accessions allows researchers to narrow the scope of which accessions to request for a project (Byrne et al. 2018). Instead of picking accessions randomly from a database, researchers can target accessions that have, or do not have, the trait of interest, because the phenotypic data from screening of the germplasm collection has been collected and made available. Unified phenotypic data can also be used to understand more fully the diversity of the collection and the creation of core collections that capture the maximum amount of diversity in the collection as a whole in an easy-to-work-with, smaller subset of accessions (Byrne et al. 2018). Using the crop ontology developed for bottle gourds, germplasm repositories with large collections of bottle gourd germplasm, such as those held by the World Vegetable Center and USDA NPGS, could characterize germplasm and create core collections that allow for data-informed decisions on what germplasm to prioritize to improve conservation of genetic and phenotypic diversity.
Despite the potential opportunities derived from using crop ontologies, currently there are a lack of crop ontologies or other unified resources available for evaluating underutilized crops (Ali and Bhattacharjee 2023). The methods detailed here for creating the bottle gourd ontology can serve as a template or starting point for other underutilized crop species and cucurbit species that currently lack a crop ontology. For example, Cucurbita sp. (winter squash) does not have a crop ontology, but has similar life stage end uses as bottle gourds, such as the use of immature fruit for eating and the use of mature fruit for art. Although there will be traits that will not overlap, the abundance of uses for bottle gourds creates an ontology with a far greater diversity of traits across all plant life stages that may not or need not be represented in other crops.
Crop ontologies are essential resources for underutilized crops, but further work is needed to optimize the use of these species. Crop ontologies are often viewed as a living document, such that they can be changed and adapted over time as new technological advances, methods, and breeding objectives or market types necessitate the addition of new traits and the depreciation of outdated or obsolete traits. In some cases, refinement of a trait description might be warranted, because a better understanding of the trait’s underlying biology may change how the trait should be measured. As these shifts occur in research, crop ontologies must also be allowed to adapt and expand. This requires an infrastructure to be implemented to ensure crop ontologies are updated routinely. Currently, there is no existing organization that oversees this needed periodic review, with the exception of cropontology.org, which only oversees cop ontologies for 36 crops.
Additional resources can help optimize even further the use of crop ontologies. Crop ontologies typically prioritize computer readability to support statistical analysis, data validation, integration with other “-omics” data classes, and applications of high-throughput phenotyping approaches such as remote sensing and Field Book App (Rife and Poland 2014). Although the ontology contains human-readable information (e.g., trait descriptions, brief descriptions of the methods used to collect the trait, citations), the ontology itself may not provide sufficient details to understand and replicate the phenotyping protocol comprehensively. Crop descriptor guides are another helpful tool to supplement a crop ontology, emphasizing human readability. Compared with the spreadsheet format of a crop ontology, a crop descriptor guide is generally prepared in a text-rich document format, such as PDF, HTML, or markdown, with in-depth descriptions or discussion, photographs, and supporting URLs. These guides train scientists and technicians on how to break down the evaluation process into steps, with photos or vector images to help the user replicate complex protocols or discern different phenotypes, although they are cumbersome to manage or reference in the field. Combined with a crop descriptor guide, crop ontologies provide a comprehensive set of instructions on how to collect data for a crop that adheres to FAIR data principles.
Ranking of the 140 final raw traits from the literature review base on trait type: root, flower (FLWR), seed, fruit, and vegetative (VEG).
Ranking of the 140 final raw traits from the literature review based on frequency of appearance in the articles, with High being the most frequently found traits and Niche appearing in five articles or less.
Crop ontology workflow beginning with the literature search and ending an ontology based on ∼280 traits.
Examples of seed traits included in the bottle gourd crop ontology, including seed tip shape (top row), seed color pattern (middle row), and Seed wings - excess seed coat protruding from the margin of the seed (bottom row).
Examples of the diversity of bottle gourd fruit. From standardized phenotyping methods, fruit metrics such as shape, color, pattern, dimension, luster, and texture can be extracted from a single image.
Contributor Notes
Funding for this work was provided by the US Department of Agriculture (USDA)–Agricultural Research Service (ARS) Crop Germplasm Committee grant (agreement no. 58-8060-2-008) “Characterization of Bottle Gourd (Lagenaria siceraria) Accessions” and USDA Foreign Language and Area Studies Fellowship stipend funding. Breeding Insight is a USDA-ARS initiative hosted by Cornell University and supported by the USDA (agreement nos. 8062-21000-043-004-A, 8062-21000-052-002-A, and 8062-21000-052-003-A).
We thank the American Gourd Society members for reviewing the ontology and providing feedback, as well as Graham Ottoson for allowing us to use her images of bottle gourd uses for Fig. 1.
M.E.L. is the corresponding author. E-mail: ml946@cornell.edu/marlielukach@gmail.com.
Ranking of the 140 final raw traits from the literature review base on trait type: root, flower (FLWR), seed, fruit, and vegetative (VEG).
Ranking of the 140 final raw traits from the literature review based on frequency of appearance in the articles, with High being the most frequently found traits and Niche appearing in five articles or less.
Crop ontology workflow beginning with the literature search and ending an ontology based on ∼280 traits.
Examples of seed traits included in the bottle gourd crop ontology, including seed tip shape (top row), seed color pattern (middle row), and Seed wings - excess seed coat protruding from the margin of the seed (bottom row).
Examples of the diversity of bottle gourd fruit. From standardized phenotyping methods, fruit metrics such as shape, color, pattern, dimension, luster, and texture can be extracted from a single image.
