Agriculture is one of the survival factors for humans on this planet. However, our intensive agriculture practices severely endangered our climate. Agriculture and related activities emitted 9.3 billion tons of CO₂ eq. in 2018 (FAO, 2020). Additionally, the agriculture sector contributed around 9% of the US greenhouse gas emission and was the largest supplier of the US N₂O emission in 2021 (EPA, 2023). Notably, fertilizer application aiming for higher nitrogen availability was the reason for higher N₂O emission, and the contribution of urea fertilization in CO₂ emission was 5.2 MMT CO₂ eq. (EPA, 2023). The extensive use of chemical fertilizers and pesticides also harms the ecosystem by contaminating groundwater and other natural resources. Although these practices ensure higher yields, they are destroying our environment.

Thus, we need a transition to a sustainable as well as low-input (LI) agriculture system to avoid further environmental problems in the future.

We can plan a sustainable agriculture system by ensuring economic profitability, a healthy environment, and social development while safeguarding our natural resources (Horrigan et al., 2002). One way to build a sustainable agriculture system is to modify our modern farming systems into low-input systems. For instance, researchers described that low-input systems could lead to sustainable agriculture by achieving efficiency in conventional practices (Hill and MacRae, 1996). A profitable yield can be achieved with lower fertilizer input with appropriate nutrient management (Good and Beatty, 2011; Chen et al., 2014). Popular high-yielding varieties require a constant supply of synthetic fertilizers and pesticide. For this reason, the success of these LI systems depends on the development of high yielding cultivars adapted to low-input conditions. Further, Evans (1996) stated that the success of any cropping system depends on the synergistic interaction between all inputs, such as fertilizers, irrigation, weed control, etc. Therefore, agronomic practices that ensure positive interaction between all these inputs will be crucial for LI systems.

Rice is one of the most extensively consumed cereals globally. It is the number one staple food among developing countries and also the most vulnerable crop to climate change (Wassmann et al., 2010). Since a large portion of world population consume rice, rice production needs to be increased to ensure food security across the globe. Further, the global population is forecasted to cross 9 billion by 2050 and more than 10 billion by 2080 (UN, 2022). For this reason, maintaining rice acreage across the globe is very important. However, extensive use of chemical fertilizers and pesticides significantly contributes to agricultural pollution. In this context, growing rice cultivars suited for LI systems is vital in practicing sustainable agriculture.

However, the most challenging part of developing crop cultivars for the LI system is incorporating tolerance to various biotic and abiotic factors, nutrient use efficiency, and higher yields. Therefore, rice breeders should target multiple traits to develop extremely resource-efficient cultivars. From the era of the green revolution, several genotypes were characterized with specific traits through conventional breeding approaches. The genomics era has opened many windows for rice breeders to understand the function of different pathways with traits of interest, which directly support rice improvement. It is challenging for rice breeders to combine all the traits into single cultivars, thus integrated breeding approaches help to develop multiple-traits cultivars (Singh et al., 2024). Fortunately, modern genomics tools increased the efficacy of multi-trait breeding and reduced the time required for integrating numerous traits. Besides, the low-cost genotyping techniques have enabled plant breeders to decipher multiple characteristics of a population in a cheaper and faster way. Speed breeding can be incorporated with various modern breeding tools in numerous steps of a breeding pipeline to reduce the breeding cycle and enhance selection accuracy and efficiency (Gudi et al., 2022). However, more effort must be made to include all the modern genomic tools to develop varieties suitable for LI agriculture. As far as we know, none of the suitable articles are available discussing the LI system for sustainable agriculture regarding the breeding perspectives. Keeping this in mind, our goal is to prepare a precise review on rice breeding for the LI system. Hence, we reviewed the requisite traits and genomics-based strategies for breeding rice for a low-input agriculture system.

Following the “green revolution,” chemical fertilizer and pesticide use in modern agriculture has increased tremendously worldwide ( Figure 1 ). Although high-input systems ensure rapid fiscal growth, these systems damage the environment. High-input rice cultivation is practiced around the globe to feed the massive population. Rice has higher greenhouse gas emission potential than other major cereals. For example, rice has a 467% higher GWP (Global Warming Potential) than wheat and 169% higher GWP than maize (Linquist et al., 2012). Rice fields are significant contributors to CH₄ and N₂0. Previous data suggested that rice fields are responsible for 30% and 11% of world agricultural CH₄ and N₂O emissions, respectively (IPCC, 2007). For this extensive rice cultivation, carbon dioxide and other greenhouse gases are increasing quickly, endangering all living beings on Earth.

Furthermore, the heavy dosage of pesticides and fertilizers exert residual effects on foods, creating carcinogenic impact on the consumer’s body. Besides, pesticides contaminate natural resources like water and alter the harmony between many biological processes. Excessive input is closely related to the environmental damage caused by rice cultivation (Ahmad et al., 2023). Therefore, an alternative low-input system is needed to safeguard our nature while sustaining agricultural growth.

Efficient use of fertilizer and exploring plant’s inherent resource use efficiency will reduce greenhouse gas emissions and remove the risk of contamination from chemical pesticides. A previous study reported that low input farming will help in better management of soil fertility in the long run. Since LI farming increases the amount of organic C and stored nutrients in the soil, LI farming will gradually increase soil health (Clark et al., 1998). Management practices like crop rotation will establish proper nutrient cycling and preserve soil productivity. The LI system will also indirectly positively affect biodiversity by minimizing water and air pollution. Moreover, the LI system is the most feasible remedy for the negative impact of conventional agriculture systems.

The ongoing modern high-input agriculture system needs an alternate solution due to the fast-changing climate and various agricultural pollutions. Despite having wide cultivation areas across the globe, rice production can be intensified sustainably (Yuan et al., 2021). Though improved varieties are being developed worldwide, significant yield gaps exist in many rice cropping regions. Inefficient fertilizer use and lack of proper management are the prime causes of this discrepancy (Yuan et al., 2021). Moreover, it is essential to simultaneously increase both yield and resource use efficiency for a sustainable farming system. Breeding for better nutrient use efficiency will help to reduce the yield gap present in certain rice growing regions.

Some rice cropping systems have lower yields despite having higher N input (Yuan et al., 2021). In those systems, N (Nitrogen) input can be reduced while having higher yield by increasing crop resource use efficiency (Yuan et al., 2021). Therefore, the development of resource-use-efficient rice cultivars that can be cultivated in LI systems is recommended. Researchers began to emphasize the value of LI systems around the turn of the twenty-first century. These systems use the least production input while management procedures are upheld to guarantee a successful crop output (Parr et al., 2020). The primary focus of low-input systems is minimizing off-farm resources such as pesticides and fertilizer to reduce environmental pollution and improve soil health.

Additionally, LI system relies on on-farm resources like management practices to generate maximum yield output. Notably, LI systems offer a plethora of environmental benefits over conventional methods. For instance, the LI system ensures lower rates of N leaching and mineralization, which helps maintain future sustainability (Poudel et al., 2002). A study conducted for eight years with crop rotation with different crops, including rice, showed that low-input and organic farming improved soil chemical properties (Clark et al., 1998). Although organic farming has long been touted as a sustainable agriculture practice, it produces lower yield per acre than conventional farming (Seufert et al., 2012).

In contrast, the LI system can reduce the trade-offs between environmentally benign farming methods and financial success. Crops in LI systems must rely on effective resource utilization and innate defensive mechanisms since they get minimal external input. Consequently, LI systems yield less than traditional systems. Two significant issues that restrict profitability under low input systems are the lack of N supply and weed competition (Clark et al., 1999). Cultivars suited for low input circumstances should be designed considering multiple traits and a range of selection environments to address different yield-limiting issues. In summary, LI agriculture needs inclusions of two major factors for sustainable rice farming: selection of resource efficient crops and adaptation of proper management practices to minimize the wastage of natural resources. Initially, a goal might be set up for 20% resources reduction during rice crop management i.e. nitrogen and/or and water use, while maintaining similar or enhanced yield.

Plant characteristics are the decisive factor for adaptation in a particular environment. Similarly, the ability of the plant to survive in a pesticide-free and nutrient-scarce system is essential for LI system. Therefore, traits that improve nutrient-usage-efficiency, weed competency, and multiple stress tolerance are critical for LI systems. Breeding for ideal rice plant ensures the incorporation of important traits into a single genotype to develop cultivars for low-input rice. Genes that control seedling establishment, vigor and multiple stress tolerance must be utilized to enhance seedling establishment to the pre-booting stage ( Figure 2 ). For better yield under LI systems, genes controlling grain yield and nutrient use efficiency should be explored and utilized ( Figure 2 ).

The safe food movement made organic farming (OF) a practical alternative to crop production, as people became increasingly cautious about their food choices. As a corollary, growers across the world are becoming more interested in organic rice. Like the low-input systems, varieties for organic systems are also being tested in the conventional systems. However, a combination of evolutionary and participatory breeding will be helpful to develop superior varieties for both organic and low-input systems. Many farmers choose to prefer the standard high-input methods over the organic systems since the yield of OF is significantly lower than that of conventional systems. However, the low yield of organic rice is offset by a better benefit-cost ratio, which compensates for the reduced yield. For instance, research done in Thailand showed that despite the poor yield, OF had higher economic benefits due to the cheap production cost and higher market value (Arunrat et al., 2022). Furthermore, organic rice may be a financially feasible alternative for farmers in developing nations that have limited resources since organic amendments combined with management practices can establish a profitable organic farm (Mishra et al., 2018). In addition to the health advantages, organic rice farming has a positive impact on the environment. Soil organic carbon and soil carbon sequestration, for instance, will be less affected by organic farming in the context of climate change (Arunrat et al., 2021). Further, sustainable crop production methods should safeguard the complex biological interactions between many organisms within a given environment. Fortunately, rice fields’ biodiversity may be preserved and increased via the use of organic agriculture practices (Katayama et al., 2019). Nevertheless, due to the current interconnectedness of other stakeholders, only OF practices cannot maintain an entirely sustainable system. Accordingly, a combination of OF and other agricultural techniques is required for sustainability (Reganold and Wachter, 2016). For instance, an efficient organic fertilization system combined with crop rotation is the most effective method for variable climatic conditions (Arnés et al., 2013).

Nature has endowed wild plants with a unique association capability with microbes, and this association benefits crops in many ways. However, the domestication of wild plants adversely affected the beneficial plant-microbe interaction of wild species (Pérez-Jaramillo et al., 2016). Modern plant breeding should emphasize utilizing maximum symbiosis from crop cultivars and select materials hospitable to soil microbial composition. In this context, using beneficial microbes from wild plants and seed endophytic bacteria can be a positive way of achieving better sustainability (L’Hoir and Duponnois, 2021).

Non-leguminous crops like rice can be improved through the exploitation of arbuscular mycorrhizal associations. Microbes help plants in many ways, including adverse effects on disease-forming pathogens, abiotic stress tolerance, and nutrient uptake efficiency (Bulgarelli et al., 2013). In particular, Arbuscular mycorrhizal fungi (AMF) helps plants to maintain proper growth under P deficiency (Harrison et al., 2002; Smith et al., 2003). Arbuscular mycorrhizal symbiosis can also be a potential way of improving rice yield and stress tolerance (Mbodj et al., 2018). Previous studies have reported the transfer of nutrients such as N, P, S, and Fe from mycorrhiza-like fungi to plants. A study showed that Piriformospora indica transfers Fe to rice (Verma et al., 2022). Very few efforts have been made to identify genetic factors controlling the root microbiome of rice under stress conditions. A recent study identified 10 SNPs and ten candidate genes related to the root microbiome of rice under drought stress. They identified more major fungal clusters. One of the fungal clusters, Pleosporales, is a crucial member of the rice seed microbiome (Eyre et al., 2019) and has previously been shown to improve plant growth and nitrogen content (Vergara et al., 2019). Another study identified 23 putative QTLs related to root colonization in rice (Davidson et al., 2019). In addition, substituting OsCERK1DY allele from wild rice to an indica variety ZH11 improved the phosphorus uptake by increasing AMF colonization (Huang et al., 2020). One study also found four QTLs related to associative N2 fixation in rice (Wu et al., 1995; Kaeppler et al., 2000).

Though soil is the primary determinant of microbial composition, plant species also play a crucial role in determining microbiome. Researchers have documented the effect of various genotypes in maintaining microbiome composition (Bergelson et al., 2019; Schmid et al., 2019; Jacoby et al., 2020). For example, Lundberg et al. (2012) documented that different Arabidopsis accessions have distinct microbiome compositions. In rice, researchers also found genotypic variation in microbiome composition. For example, root endophytes like Azoarcus spp. and Acremonium preferentially prefer wild species for colonization (Engelhard et al., 2000). Researchers found differences in nitrogen derived from air (ndfa) by comparing diverse rice lines (Shrestha and Ladha, 1996). Traditional varieties with high ndfa value showed high associative fixation and sustained 12 months on N-free medium (Rolfe et al., 1997).

For this reason, screening genotypes for microbiome friendliness should be considered. Two things need to be appropriately considered for comparing the genotypes’ feasibility in plant-microbiome interactions: root morphology and root exudates.

Root morphology can influence microbiome composition by affecting nutrient availability. For example, longer roots reduce microbial biomass by decreasing nitrogen availability (Pérez-Jaramillo et al., 2017; Wan et al., 2021). Root traits such as root length, root hair, and root branching patterns are essential in determining microbial diversity (Eisenhauer et al., 2017; Robertson-Albertyn et al., 2017; Wang et al., 2017). Previous studies reported that roots having less root diameter attract more microbes (Szoboszlay et al., 2015; Wang et al., 2020b). Bacteria also can modify root traits by exerting certain phytohormones (Grover et al., 2021). AMF preferred larger lateral roots in rice for colonization (Gutjahr et al., 2009).

Further, root exudates are one of the prime factors for studying the rhizosphere. Metabolomics is the perfect means of characterizing rhizosphere based on metabolites. Metabolites can facilitate communication between beneficial microbes for adapting to adverse conditions or protecting plants against pathogens (Jacoby et al., 2020). Root exudates create a favorable environment for microbes by providing a proper nutritional source (Sasse et al., 2018). In addition, utilizing ‘holo-omic’ and ‘exometabolomics’ in the breeding pipeline can decipher useful plant-microbe interactions (Sasse et al., 2018; Nerva et al., 2022).

LI systems can be an effective solution for fostering and maintaining healthy associations between microorganisms. In low-input systems, management techniques and landscape variation impact the soil microbiota and enzyme activity (Wickings et al., 2016). However, the judicious application of different management approaches will determine the extent of the association.

A shift to LI agriculture would reduce pollution, reward farmers financially, and protect the environment. The current practice of developing high-yielding varieties fails to consider low-input cultivars (Vanloqueren and Baret, 2008). So, an integrative breeding approach should be designed to cover every facet of sustainable agriculture ( Figure 4 ). Transforming modern agriculture to a low-input sustainable system requires new varieties that can thrive under extreme conditions. Varieties for the LI system should be equipped with multiple traits, making it a cumbersome task. Fortunately, the advent of modern technologies can be incorporated in the breeding program to reduce the timeframe required for developing a variety of LI systems. The first step will be exploring germplasms for enhanced resource use efficiency and multiple stress resistance. Since most of our materials are adapted to high input conditions, we must screen many materials for starting any breeding program for LI systems. Recent phenomics developments will help breeders rapidly screen several germplasms to identify the perfect candidate that can be included in the crossing program. The flexibility of machine learning algorithms can infer multiple parameters just by analyzing field images. Important traits such as early vigor and nutrient use efficiency can be tested within a shorter time with adjustment in image processing pipeline.

Further, genetic factors behind traits like early vigor, nutrient use efficiency, and stress resistance have already been explored. For instance, genes controlling nitrogen assimilation and utilizations like OsNPF3.1, OsNPF7.1, OsNRT2.1 can be targeted for better nitrogen use efficiency. Rice diseases like blast, and bacterial leaf blight are vastly studied, and multiple genes are identified, for instance, OsCPK4 for blast resistance, Xa23 for bacterial leaf blight resistance, OsOSN1 for sheath blight resistance. Pyramiding multiple genes will be an excellent way to develop advanced lines for low-input agriculture. Another critical aspect of sustainable agriculture is root traits. The modern phenomics pipeline has lessened the difficulty in phenotyping root traits. Traits such as root length, diameter, and hair density would have been complex to study without phenomics platforms. Incorporation of root phenomics in the screening program will efficiently identify good lines for resource use efficiency. Various metabolomics studies can be done simultaneously to identify microbe-friendly genotypes since the microbiome is crucial for maintaining sustainability. Also, we must explore new genes, QTLs, and markers under low-input conditions by utilizing cheaper genotyping data. A genome-wide association study should be conducted for all the essential traits for identifying new markers and genomic regions associated with desired traits. Searching for these genes or alleles to introgress into existing lines will be beneficial for developing an ideal variety for the LI system. In addition, identified genes can be validated easily by incorporating genome editing tools. This validation step will increase the success rate before starting to pyramid multiple genes. Since genotyping has become cheaper than phenotyping, genomic selection will be a handy tool in any breeding program. While training the genomic selection model, we should incorporate all the relevant traits for the LI system. Finally, genomic selection will be helpful in rapidly testing advanced lines.

We must modify our conventional agricultural methods into resource-efficient systems to attain sustainability in agricultural practices. In this context, modern development in genomic research can facilitate rice breeding for low-input systems. A successful rice breeding program for a low-input system will require fast identification of traits that promote resource use efficiency, better protection against pathogens, and incorporating these traits into one cultivar. Utilizing all modern tools throughout the process will increase selection efficiency and reduce the number of breeding cycles required. Since rice is cultivated across a range of environments, breeding practices should consider precise estimation of genotype-environment interaction. Recent developments in environmental modeling have created new opportunities to predict the performance of certain genotypes by including multiple environmental covariates in the prediction model. Besides, developing countries may face adverse environmental conditions due to climate change. Testing the performance of LI rice in stressed environments will provide better insights into its suitability in developing countries.

Initially, we must identify early vigorous rice genotypes that can compete with weeds. Tolerance to multiple stress factors will also play a vital role. Moreover, phenotyping for root traits will be the most effective way to search for nutrient use efficiency. Genes related to nitrogen transportation and assimilation, like OsNLP4, OsNPF7.7, and gs3 allele, can be advantageous. In general, nutrient use efficiency traits are polygenic in nature. QTLs associated with nitrogen, phosphorus, and potassium use efficiency have already been mapped in rice. For early vigor OsIPMS1, OsHIPL1 genes can be utilized. Root traits such as root lengths, diameter, and surface area are essential for nutrient uptake. Fortunately, many high-throughput platforms have been developed to facilitate these phenotyping processes. Although people are using many of these platforms regularly, we should be careful to limit the misinterpretation of the data. In addition, breeders should cross-validate their results for better accuracy. Besides, researchers should continuously improve machine learning and deep learning algorithms for better accuracy in high-throughput phenotyping. Also, cross-validation should be a routine task associated with all these methods.

Further, cheaper NGS data can be used in genome-wide association mapping for identifying genes, QTLs, and markers for multiple traits. Proper selection of genotyping platforms, marker system, and accurate phenotyping will determine the efficiency of this method. Besides, the genomic selection model can help evaluate many lines for multiple traits in a shorter time. Also, the ability to use a multi-trait model or incorporation of phenomics data made GS a powerful tool in breeding for low-input systems. In predictive breeding, minimizing the error rate is a vital step. For this reason, data analysts should carefully manage the statistical packages to reduce the error rate. Breeders should also focus on newer statistical models that can generate more accurate predictions of their dataset.

Furthermore, genome editing can also help modify target genes for nutrient use efficiency or stress tolerance. However, we must be careful with gene editing tools and transgenic approaches since these techniques have acceptability issues. Gene editing can reduce the time for introducing the desired variation, but transforming plants for editing often becomes difficult. Fortunately, people are working on DNA and tissue culture-free methods for better efficiency in genome editing (Altpeter et al., 2016). Sometimes, getting a homozygous mutation becomes complex with the increased number of target genes in multiplex editing. More work on multiplex editing can render this a helpful approach for targeting multiple genes for low-input systems.

For developing a sustainable farming system, the soil microbiome must be taken care of. Microbes not only improve soil health but also improve nutrient availability to plants. For this reason, all lines should be checked thoroughly for microbiome compatibility. Metabolomic and phenomics studies can be used to study the genotypic effect on soil microbiome. Attaining sustainability relies on covering every facet of environmental safety. Hence, breeding rice for the LI system should be carefully crafted to ensure maximum protection of all natural resources.

SJ: Conceptualization, Data curation, Formal analysis, Methodology, Software, Visualization, Writing – original draft, Writing – review & editing. GS: Methodology, Writing – review & editing, Visualization. AP: Visualization, Writing – review & editing. LT: Writing – review & editing. ES: Writing – review & editing. ST: Conceptualization, Funding acquisition, Methodology, Supervision, Writing – review & editing, Resources.