1 INTRODUCTION

Drought is the most devastating natural disaster affecting the livelihood of developing world's farmers and economies. About 166 billion USD loss due to drought is estimated coming from the three-quarters of the global cropped area (454 million hectares) between 1983 and 2009 (Kim et al., 2019). In the future, long dry spells are predicted to be more frequent if the rising CO₂ is not curtailed (Ault, 2020). Moderate and severe drought events are predicted to increase by 20%–28% and 10%–16%, respectively, by 2080–2099 relative to 1970–1999 (Zhao & Dai, 2015). While many climatic modeling studies suggest a global increase in mean precipitation, these studies did not consider nutrient constraints and changes in host-pest/pathogen interactions, thus these predictions may not be very certain (Berg & Sheffield, 2018). Moreover, the increased precipitation belies that more sporadic, high-intensity rainfall events may still lead to more intermittent droughts and that decreased precipitation is not the major culprit of agricultural drought, but it is rather due to increased evapotranspiration (Orlowsky & Seneviratne, 2013). Securing crop yield under limited water availability is pivotal for future food and nutrition security as the irrigation water availability competes with the water demand from industry and the growing population. Over the past 100 years, global water use has increased by sixfold and will continue to rise by 1% every year (UNESCO UN-Water, 2020). Due to climate change, it is estimated that the global irrigation water demand will reach between 2900 and 9000 km³ year⁻¹ by 2100, over the baseline demand of 2500 km³ year⁻¹ observed in 2005 (IPCC, 2019).

Drought is defined here as an environmental condition in a farmer's field where water limitation over a period of time, ranging from weeks to months during a growing season, reduces crop yield significantly (Passioura, 2006). Generally, low rainfall, reduced soil water level, and limited availability of water sources cause the overall yield loss, depending upon the plant species and developmental stage. Drought resistance is the broad ability of plants to withstand drought and is often partitioned into drought escape, avoidance, and tolerance (Gupta et al., 2020). Here, we focus on the root system's ability to acquire water even in drying soil, which is an aspect of drought avoidance.

2 ROOT AND DROUGHT RESISTANCE

Plants evolved to sense the soil environment to regulate above and below ground growth by judicious partitioning of photosynthates to maintain the effectiveness of the roots and shoots. Now, there is experimental evidence and a general consensus that root signals modulate the shoot growth in response to water, nutrient availability, and soil properties. Either root-to-shoot and shoot-to-root signaling becomes the core regulator of shoot and root growth depending upon the environmental cues. MicroRNAs, peptides, ABA, trans-zeatin, cytokinin, strigolactone, and gibberellins are the major signaling molecules that regulate shoot growth in response to water and nutrient availability, while cytokinin, auxin, and sugar are the key signaling molecules from the shoot to the roots (Wheeldon & Bennett, inpress). Cortical cells within the root elongation zone sense the soil water status and, under drought, leaves respond by closing stomata. Stomata closing may be more sensitive to the water potential gradient near the roots than the high transpiration loss; thus, stomata close well before the drop in leaf water potential to avoid xylem cavitation (Carminati & Javaux, 2020). This long-distance communication is mediated by the CLE25 peptide, which is produced in roots and moves up to leaves where it induces ABA synthesis triggering stomata closing. Brassinosteroid (BR) is also synthesized in the vascular system during drought and initiates the accumulation of osmoprotectants in roots that help in the absorption of water. In Arabidopsis under fluctuating moisture availability, EXO70A3 (EXOCYST SUBUNIT EXO70 FAMILY PROTEIN A3), an auxin transport modulator, regulates the root system depth by rearranging the auxin efflux protein PIN4 in root columella cells, which can lead to a greater water absorption deeper in the soil. Similarly in rice, the auxin-inducible gene DEEPER ROOTING1 (DRO1) promotes a deeper root system (Gupta et al., 2020).

Root system architecture (RSA) refers to the three-dimensional placement of the roots and it is influenced by both the constitutive genetics and the environment. Introgression of DRO1 into a shallow-rooted rice line increased yield during drought due to a deep root system accessing deepwater reserves (Uga et al., 2013). Similarly, a wheat line harboring allele from a wild tetraploid wheat also exhibited deeper rooting and improved yield under drought (Merchuk-Ovnat et al., 2017). However, under rainfed or only intermittent drought conditions, the constitutive phenotypic expression of deep rooting with dense and long roots may be expensive in terms of carbon costs for tissue construction and maintenance (Lynch, 2007). For example, the root system for soil exploration may demand more than 50% of daily photosynthetic products (Lambers et al., 2002). Similarly, there is a tradeoff between different root phenotypes. For example, the RSA suitable for topsoil exploration is unfit for extracting water from deep soils under rainfed conditions where water becomes more available in subsoils as the season progresses (Ho et al., 2005). An overly branchy root system may not always be suitable for securing higher yield under low moisture availability, when water may be found deeper in the soil. Individual root tips all act as sinks for photosynthate, so root system with more branches will produce more shorter roots rather than few but deeper roots. In other words, the competition for carbon among growing root tips leads to a tradeoff between the extent soil exploration and local soil exploitation. Root features like less axial roots, reduced lateral root density, decreased responsiveness to resource availability, more aerenchyma, and reduced cortical cell number with a larger size affect the costs of roots and are suggested features to improve drought resistance under among growing root tips lead to a tradeoff between soil exploration and local soil exploitation resource-intensive agriculture (Lynch, 2018). Given these tradeoffs for any given root system trait in different experiments, root plasticity (the ability of a genotype to change phenotype in different environments) is hypothesized to be a valuable breeding target so that any variety will grow the ideal root system for a particular environment (Schneider & Lynch, 2020). Phenotyping these root traits led to the identification of genomic regions, yet their introgression into crop varieties is hindered due to the complex regulation of root traits interactions by genotypic and edaphic factors. Trait interactions can be neutral, synergistic, or antagonistic, and so the consideration of a root integrated phenotype is necessary (York et al., 2013). However, using functional phenomics, root traits can be linked to a function in a target environment and simulation modeling can be used to unravel the complex interactions between trait (root phenotype) and the environment (York, 2019).

Moreover, drought also changes the rhizosphere microbiome, the microbial community living on or near the root surface, which contributes to the reduction of the above-ground plant performance (Prudent et al., 2020). The microbial community plays a significant role in drought recovery or resistance by overcoming the oxidative stress through enhancing antioxidant enzymes and providing many microbial metabolites acting as a precursor to salicylic acid and other secondary metabolites (De Vries et al., 2020). However, the activity of microbial communities is more sensitive than the community structure to the effects of drought. These changes slow down the mineralization of nitrogen and carbon (Hueso et al., 2012). Roots also influence the microbiome of the rhizosphere, which provides an opportunity to manage the soil community. The effect of roots on the microbiome depends on cell wall structure, root exudate metabolic profile, and root surface area. Root hairs may be a place of higher nutrients and metabolite content, making them a favorable place for microbial colonization. In beans, thin roots accessions have a greater abundance of Bacteroidetes as compared to thick root accessions that have a greater abundance of Actinobacteria and Proteobacteria (Saleem et al., 2018). The root system also alters soil health properties such as drying and wetting, soil aggregation, carbon balance, biopores, and rhizosphere microbial communities. Optimizing root architecture, anatomy, and chemistry can, therefore, improve soil properties (Jin et al., 2017).

However, in this review, we do not focus on the interaction between the root system and soil properties and microbes, but we explore the opportunities to increase drought resistance by manipulating root plasticity by nutrients. We choose nutrients because fertilizer application is already a widely used ameliorative strategy against drought (Waraich et al., 2011). Moreover, similar mechanisms appear to benefit both water and nutrient capture and thus there may be crosstalk between the signaling molecules of water and nutrient sensing. For instance, phosphorus and nitrate supply was found to regulate the gene expression of aquaporins (Calderon-Vazquez et al., 2008; Liu et al., 2008). Water and nutrient stress also show similarity in long-distance hormonal signaling: increased ABA and decreased cytokinin synthesis and export from root, as well as increased auxin export from shoot to root that leads to greater sugar allocation to the roots for adaptive root growth resposes (Kudoyarova et al., 2015). The nutrient foraging responses of roots to nitrogen (N), phosphorus (P), iron (Fe), and boron (B) involved brassinosteroid signaling (Pandey et al., 2020). Brassinosteroids are also synthesized in the vascular system during drought (Gupta et al., 2020). These common hormonal responses imply that adaptations to either drought or nutrient stresses may be beneficial for other stresses at the same time.

We first discuss the RSA and root plasticity influenced by water and their significance in relation to drought resistance. We then explore the effects of nutrients on root architecture and plasticity and, finally, the potentiality of nutrients to achieve desired root system traits for drought resistance.

3 ROOT PLASTIC RESPONSES TO WATER

Roots respond in many ways to water deficit conditions. For instance, root xerotropism is where roots increase the gravitropic response upon water stress in order to grow more deeply in search of water. However, root penetration depends on both soil and root characteristics. In dry soil, root elongation is limited due to the soil's mechanical constraints and limited turgor pressure for root cell elongation. Hydropatterning is where roots branch toward moisture patches due to auxin signaling in response to a water potential gradient across the root. In contrast to hydropatterning, ABA-regulated xerobranching is the phenomenon where lateral roots are inhibited due to dry soil. Heterogeneous water availability in the topsoil induces root curvature toward moisture, termed hydrotropism (Fromm, 2019). Traits like decreased cortical cell file number, cortical senescence, and root cortical aerenchyma formation are found to provide drought resistance by increasing root depth with a low investment of carbon. Decreased root axial conductance by small xylem diameter and a smaller number of vessels substantially increase the yield potential in wheat under drought conditions at the end of the growing season in Australia (Richards & Passioura, 1989). Suberization and lignification of endodermis and exodermis, large rhizosheaths of agglutinated soil particles on the surface of the root bound by root hairs and mucilage, and root penetration capacity in dry hard soil are found to be associated with drought resistance (Lynch et al., 2014). In rice, drought-tolerant genotypes exhibit larger xylem vessels and less aerenchyma formation compared to drought-susceptible genotypes (Singh et al., 2013). In contrast to rice, maize genotypes with more aerenchyma had greater drought resistance (Souza et al., 2016), attained higher root length density and secured higher biomass under drought (Zhu et al., 2010). Many of these described traits have been studied in the context of genotypic differences. However, several are also expressed upon stress and appear to be plastic adaptations for drought resistance.

4 ROOT PLASTIC RESPONSES TO NUTRIENTS

Plants are dependent on acquiring nutrients from the soil; agricultural productivity is mostly limited by N, P, and K. Thus, the root sensing and signaling mechanisms are mostly known for these nutrients (Schachtman & Shin, 2007). Furthermore, much of our understanding of nutrient signaling is from model plants and our understanding of crop plants is limited (Jia & von Wirén, 2020). In the case of N, sensing and responses may be categorized as four mechanisms: (1) local perception and uptake of nutrients in roots, (2) root-to-shoot-to-root signaling in either a low nutrient zone or (3) a high nutrient zone, and (4) a systemic inhibitory signal from shoot-to-root to inhibit root foraging behavior when the shoot has sufficient nutrient content (Oldroyd & Leyser, 2020). Ammonium can suppress root elongation, induce lateral root branching, or elongate root hairs (Liu & Von Wirén, 2017). Nitrate can have either a suppressive or a stimulatory effect on lateral root growth depending on the N status of the plant. Under mild deficiency, lateral root growth is promoted, while lateral root growth is suppressed and new lateral roots emergence is stopped under severe deficiency. Unequal availability of nitrate also stimulates lateral root length and density in nitrate-rich zones (Sun et al., 2017).

In general, lateral roots are more responsive to nutrient variation than axial roots. Thicker roots are less prone to dehydration and have higher hydraulic conductance, but have high carbon cost and in many species, root diameter is less plastic to nutrient availability (Forde & Lorenzo, 2002). In spatially homogeneous low N condition, lateral growth is promoted, while lateral growth is suppressed under homogeneous high N condition. However, when a high N patch exists within an otherwise low N soil, lateral roots may proliferate in the patch. Higher phosphate availability increased primary root length with sparse and shortened laterals (Williamson et al., 2001). In common bean, P limitation induced a shallower basal root growth angle with higher root hair density, which may allow foraging in the top soil where P is typically found in greater amounts (Miguel et al., 2015). Nutrient availability also influences root hair density and length. In Arabidopsis, rapeseed, tomato, and spinach, low P produces long root hairs with high density relative to high P. Similar to P, NO₃ and Fe are also negative regulators of root hair development (Forde & Lorenzo, 2002). In Arabidopsis, deficiencies in Ca, K, Mn, B, and Zn caused increased lateral root density, whereas S, Fe, and Mg deficiencies lead to a decline in lateral root density (Gruber et al., 2013; Kellermeier et al., 2014). Nitrate was reported to increase the xylem vessel area in the lateral roots of tomato seedlings (Cohen et al., 2018). Nutrient deficiencies such as N, P, K, and S also induce aerenchyma formation in many crops, which may decrease the metabolic burden of root tissue and allow greater soil exploration to acquire more nutrients (Postma & Lynch, 2011). However, results from root architectural studies performed with non-soil media may not be the same as under field conditions. For instance, phosphate movement is diffusion-limited due to the presence of charged particles. Therefore, root architecture and its influence on uptake studied under phosphate-limited clay might be different from plants grown in phosphate-limited sand or glass beds (Sasse et al., 2020). However, in the near future, with the use of field-based root phenotyping methods, RSA responses to nutrients ratio and availability may be clearer (Wasaya et al., 2018). In Arabidopsis, adequate Fe availability is needed for RSA responses under low P conditions (Müller et al., 2015). Many studies focus on homogeneous conditions with only a single nutrient (Figure 1). In contrast, the RSA in the field is shaped by the availability of multiple nutrients distributed heterogeneously in the soil. Therefore, the way we are simplifying lab studies may cause deficits in our full understanding of complex interactions.

5 ROOT PLASTIC RESPONSES TO SIMULTANEOUS WATER AND NUTRIENT TREATMENTS

Despite that fertilizer application is a common agronomic practice to enhance plant growth and drought resistance (Garcia-Sanchez et al., 2016), very little attention has been given to how local and well-timed application of nutrients could affect the spatial and temporal distribution of roots to promote drought resistance. Generally, root plasticity responses were studied on either water or nutrient deficiencies with no interactions, and often in the context of intensive agriculture. These studies increased our understanding of how plants regulate RSA in space and time but lack critical insights into how multiple nutrients can optimize RSA for drought resistance. In wheat, yield reduction due to reduced water availability was alleviated using localized fertilizer application (NPK) at 8–10 cm depth as compared to homogeneous fertilizer distribution (Trapeznikov et al., 2003). Fertilizer's local application enhanced the wheat above-ground dry matter because the additional root acquisition of water and other nutrients were outside the nutrient-rich zone and reached deeper. Similarly, in conditions of high potential evapotranspiration, high K increased the total root volume, root surface area, the number of root forks and tips, and tomato seedlings attained higher biomass as compared to low K availability (Zhang et al., inpress). NH₄⁺-supplied rice seedlings showed relatively greater water uptake and xylem sap flow than those under NO₃⁻ nutrition. This increased water uptake under NH₄⁺ is likely due to lower aerenchyma formation, higher aquaporin activity and higher gene expression of root aquaporins (Ding et al., 2015; Yang et al., 2012). In rice roots, temporal and spatial changes were observed at seedling stage, which might be due to the different levels of moisture and K on the projected root area and average root density (Patel et al., 2020). In this study, root area and root density were found to be influenced by K and moisture interaction, where at higher K levels, reduction in root area and root density were minimal at higher K levels as compared to low K application at the same moisture levels. Higher K was found to improve the maximum width of the root system and rooting depth under moisture stress. However, the results suggested that at very low moisture availability, the effect of potassium was not as profound (Figure 2). Further, K was also found to reduce the aerenchyma formation in the rice variety NAUR-1 under water stress (Pandya, MN. (2020). Thesis, Navsari Agricultural University, Navsari, Gujarat, India) (Figure 3). K application also increases the stele diameter with high N (Harvey & van den Driessche, 1999), which may indicate mechanisms to increase xylem area and axial water flow.

6 PROMOTING DROUGHT RESISTANCE USING SMART NUTRIENT APPLICATION

In high-input agriculture, fertilizer management aims to ensure adequate and homogenous nutrient availability near the soil surface where roots emerge from the seed or stem. However, because roots tend to proliferate in areas of high nutrient content, this strategy may actually interfere with the ability of the plant to grow deep roots that would provide better drought resistance. At the same time, plants with adequate mineral nutrition invest less in the root system relative to the shoot (increase shoot:root ratio), which may be another way conventional fertilizer management could interfere with drought resistance. Instead, deep placement and split application of fertilizers could be used to encourage a deep root system. In Brassica, fertilizer placement at 10 cm depth significantly increased taproot length and lateral root mass compared to the surface application of fertilizers (0 or 5 cm) (Su et al., 2015). Deep placement of fertilizer is already use to increase the nutrient use efficiency and can be further adjusted to optimized the root system for water absorption. In maize, greater root length density with the proliferation of fine roots was attained with the application of localized placement of ammonia and phosphorus, leading to greater shoot mass and nutrient content (Jing et al., 2010; Ma et al., 2014). It is known that this type of increased root production may also be beneficial for drought resistance in maize (Rosa et al., 2019). Root proliferation due to localized nutrient application could be burdensome on the plant carbon economy, leading to reductions in total plant growth; however, lateral roots have a low carbon cost and are very efficient at nutrient and water acquisition (Li et al., 2016). Thus, such nutrient placement and optimization can be useful to increase crop resilience against drought (Figure 4). In soybean, deep banding placement of P led to a higher proportion of fine roots in deeper soil, which could access water during the critical stage of grain filling and secure greater yield as compared to the broadcast application (Hansel et al., 2017). However, the roots' responses to localized nutrient placement varied between species and genotypes because of varying degrees of root plasticity. In general, cereals express more root plasticity than leguminous species (Li et al., 2014). Moreover, field fertilization rate should be adjusted with anticipation of the moisture availability during the crop season as diffusion of ions from the deep placement of fertilizers may injure root apical meristems during drought and adversely affect the above-ground growth (Jacobs et al., 2004). Other points that should be taken into consideration are: fertilizer should have a low diffusion coefficient, stimulate root growth, have limited mobility from the site of placement, and stable in its chemical forms (Nkebiwe et al., 2016). Because over-fertilization will decrease the allocation to the root system, we also hypothesize that split applications of nutrients may not only increase nutrient use efficiency, but may also promote deeper rooting and drought resistance, but this needs further research. Smart nutrient application considering spatial and temporal aspects represents a substantial and largely unexplored opportunity for agronomic intervention to use plant root biology for drought resistance.

7 FUTURE PERSPECTIVES

Root system traits respond differently to water and nutrients or in response to their combined availability. However, understanding the interactions of water and nutrients in shaping the root system will be critical for optimizing fertilizer doses according to moisture availability. We propose four mechanisms for how smart application of nutrients could optimize root traits for water stress: (1) increase overall plant vigor and growth, including the root system, (2) increase the carbon allocation to the root system (decrease shoot:root ratio), (3) influence specific root traits such as angles, numbers, diameters, lengths, and xylem properties using specific nutrient amounts and ratios, and (4) use deep placement of fertilizer to drive root proliferation at depth. Future research needs to address the molecular regulatory mechanisms driving these plant responses and to devise effective agronomic practices and mechanical implements for economical smart fertilizer application. Smart application of nutrients for drought resistance will be an important component of the farmer's toolkit to increase agricultural sustainability and food security in the face of environmental degradation and climate change.

AUTHOR CONTRIBUTIONS

Kirti Bardhan, Larry M. York, and Mirza Hasanuzzaman conceptualized the framework of the article. Kirti Bardhan, Vipulkumar Parekh, and Suchismita Jena wrote the first draft. Suchismita Jena and Mansi N. Pandya created the figures. Larry M. York, Kirti Bardhan, and Mirza Hasanuzzaman edited and revised the first draft. All authors gave inputs and approve the final version.

ACKNOWLEDGMENTS

The authors are thankful to their respective institutes: Navsari Agricultural University, Noble Research Institute and Sher-e- Bangla Agricultural University for support. We apologize to researchers whose work we were not able to include due to limitations, it does not imply about the merit of their work. KB acknowledges his wife, Preeti, and mother as a mark of appreciation for unwavering support and faith. He also extends his thanks to his past students who helped his critical thinking by their unconventional questions inside or outside the classes.