Transcription factors (TFs) and transcription factor binding sites directly affect the transcriptional regulation of plant genes. Several studies have elucidated the roles of TFs in maize adaptation to abiotic stresses. For example, ZmNF-YB16 has been introduced into the inbred maize line (B104). The overexpression of this TF increased the expression of genes encoding antioxidant enzymes, antioxidant synthase, and molecular chaperones associated with the endoplasmic reticulum stress response. As a result, lines sustained greater photosynthesis, improved dehydration, and drought stress tolerance during vegetative and reproductive stages, increasing maize grain production under drought-stressed conditions (Wang et al., 2018a). Similarly, the WRKY genes in maize have been well associated with abiotic stress resistance. For example, the ZmWRKY40 and ZmNF-YB2 gene encoding a transcription factor has improved drought tolerance in maize through functional genomics studies (Nelson et al., 2007; Zhang et al., 2008; Wang et al., 2018b; Gangola and Ramadoss, 2020). The ABP9 gene in maize encodes the bZIP transcription factor in Arabidopsis, improving salt, drought, and cold tolerance by increasing oxidative enzyme levels (Zong et al., 2018). In addition, several MYB-related proteins in the maize genome have been identified, and 46 of them were already characterized in relation to different abiotic stresses. Among them, the expression of ZmMYB30 was observed to have increased remarkably under drought and high salinity conditions (Nelson et al., 2007; Wang et al., 2018a; Zong et al., 2018). The MYB TF, ZmMYB31, repressed isopalmitate biosynthesis, increased UV exposure sensitivity, decreased plant height, and activated several stress-responsive genes (Zm5H, C3H, ZmActin, and ZmCOMT) in transgenic maize plants (Fornalé et al., 2010). Interestingly, ZmSAPK8, an SnRK2 phosphokinase, was cloned from maize that confers salinity tolerance with transcriptional upregulation of stress-linked genes, including DREB2A, P5CS1, RAB18, RD29A, and RD29B under salt stress indicating ZmSAPK8 is involved in diverse signal transduction and has the potential to improve salt tolerance in crops (Ying et al., 2011). Other transcription factors like ZmNF-YB2 and ZmNAC3 were previously characterized through functional genomics and were described as stress-induced genes in maize that responds to physiological variations in photosynthesis and polysaccharide metabolism (Table 1).

Stress can also cause genome-wide transcriptome reprogramming in plants to respond to environmental stimuli. As a result, groups of genes linked to various physiological features and response pathways are controlled to counteract their negative impacts (Shao et al., 2021). Several transcriptomic studies have identified regulatory networks crucial to stress tolerance in maize. For example, a microarray-based technique was used to explore gene expression dynamics throughout seed development, which found 3445 genes with variable expression across samples from six different time points (Liu et al., 2008). Wang B. et al. (2019) demonstrated that drought-induced transcriptomic changes were strongly linked to developmental and physiological adaptation, affecting maize production. Similarly, the transcriptome sequencing of the maize roots grown in N, P, and K deficient environments identified a total of 2555 (N), 2340 (P), and 1173 (K) differentially expressed genes that are involved in nutrient utilization, plant hormones, and transcription factors (Ma et al., 2020).

Cis-acting elements function as stress signaling factors at terminal points in signal transduction pathways. In general, they impart several auxiliary functions to the plant systems, such as developmental regulation of growth-associated processes, morphological modifications, regulating senescence, and DNA damage repair mechanisms. Like TFs, Cis elements and signal transduction genes were also associated with abiotic stress response in maize. For example, in a recent study, the exogenous application of ABA-induced CAT1 (Catalase1) expression via the cis-element ABRE2 affects the enzymes that are related to ROS (H₂O₂) in maize (Guan et al., 2000). Similarly, several ectopic expressions of genes such as AtNHX1 and NPK1 were found to regulate transporters, ion channels, and oxidative signaling in response to salinity cold, heat, and salinity tolerance (Yin et al., 2004; Lin et al., 2014). Tejeda et al. (2019) characterized autophagy-related genes (ATG genes; ZmATG8 and ZmATG12) in maize landraces under osmotic stress and found them potential targets for functional characterization and development of osmotic-tolerant maize genotypes using molecular breeding strategies (Table 1).

MicroRNAs (miRNAs) were also documented to control abiotic stress tolerance in plants by binding to cis-regulatory elements or genes/transcripts. For example, miRNAs regulate target gene expression post-transcriptionally and play important roles in seed germination, ear development, and root architecture in maize (Spollen et al., 2008; Peng et al., 2018; Wani et al., 2020). Table 2 shows various miRNAs that are involved in maize abiotic stress tolerance.

Short tandem target mimics (STTM) is a technology that develops a resource for producing miRNA inactivation vectors and transgenic lines in model and crop plants (Peng et al., 2018). A recent study on a series of maize STTM166 transgenic plants identified 178 differentially expressed genes (60 downregulated and 118 upregulated genes). Most were involved in the cell membrane system, cell components, oxidation-reduction process, oxidoreductase activity, and carbohydrate metabolic processes. Several studies were also carried out to identify novel miRNA and mRNA interaction and their association with abiotic stress tolerance in maize. For example, a microarray expression analysis of a drought-associated study revealed 13 miRNAs families regulating 42 novel mRNAs target and 65 in maize (Li et al., 2020). Another study showed that 23 drought-responsive cis-regulatory elements and three TFs (GAMYB, HD-Zip III, and NAC) were associated with the target mRNAs (Aravind et al., 2017; Sepúlveda-García et al., 2020). The knock-down maize mutants of miR166 showed its association with adaptation through various phenotypic variations. It was observed that knock-down mutants had rolled leaves, inferior yield-related traits, epidermis structures, vascular patterns, enhanced abiotic stress resistance, elevated ABA levels, and reduced levels of indole acetic acid. The results shed light on the importance of ABA and auxin interaction in monocots and suggested that the specific mechanism differs from dicots (Aydinoglu, 2020; Li et al., 2020).

Du et al. (2016) performed overexpression studies to understand and improve the adaptive response of phosphorus (Pi) deficiency in maize and found a phosphorus-deficiency-induced long non-coding RNA1 (PILNCR1) that inhibits miR399 and thus miR399-mediated cleavage of ZmPHO2 [Zea mays PHOSPHATE2 (PHO2) pathway]. These findings indicate interactions among PILNCR1 and miR399 that could be potential targets for improving phosphorus efficiency in maize. Maize mutant lines for miR166 and its target gene Rld1 are involved in Rolled leaf 1 (Rld1) (a homologous gene of Arabidopsis HD-ZIP III transcription factor). The interaction of these two components is responsible for leaf polarity and exhibits various developmental defects such as delayed flowering, reduced stature, and curled leaf. Other mutants related to leaf polarity have been documented, including leaf-bladeless1 (lbl1), Arabidopsis SGS3 homolog, ragged seedling2 (rgd2), Arabidopsis ago7 homolog, milkweed pod1 (mwp1), and Arabidopsis KANADI homolog (Juarez et al., 2004; Douglas et al., 2010). Intriguingly, miR390-lbl1 and miR166-rld1 have been implicated synthesis, transport, and action in maintaining leaf developmental patterns (Nogueira et al., 2007, 2009; Itoh et al., 2008).

The current advances in sequencing have led to the development of genetic tools to understand complex traits in crops. Over the years, these tools were successfully used to develop new genetic resources that could be used to mine beneficial genes and marker-assisted breeding (Kitony et al., 2021; Reyes et al., 2021b). Investigating complex traits like abiotic stresses requires quantitative genetic approaches such as linkage mapping and association analysis using different mapping populations or association panels (Almeida et al., 2013; Samantara et al., 2021).

To date, QTL governing various abiotic stresses have been mapped using backcrossing, F₂, RILs, introgression lines, and natural populations (Qiu et al., 2007; Mano and Omori, 2008; Osman et al., 2013; Zaidi et al., 2015; Frey et al., 2016). For cold tolerance, Hund et al. (2011) mapped important putative meta-QTL (MQTL Rt-6, MQTL Ax-2, MQTL Rt-7, and MQTL Ax-15) for root architecture traits during cold stress. Similarly, Hu et al. (2016) identified 12 QTL for germination rate and primary root length in maize under cold stress using B73 × Mo17 (IBM) derived population. Interestingly, the candidate gene GRMZM2G398807 governs cortical cell-delineating protein expression in the primary root, assisting in radicle protrusion. Goering et al. (2021) identified two QTL with high additive impact on chromosomes 1 and 4 using a panel of IBM Syn4 recombinant inbred lines. The genes with putative function related to auxin and gibberellin response were identified in the QTL region. Recently, Han et al. (2022) mapped five QTL clusters on chromosomes 1, 2, 3, 4, and 9 using an B73 x Mo17 (IBM) Syn10 double haploid population. From these clusters, 39 genes were extracted, and through RNA-seq, upregulated genes from B73 and Mo17 were identified.

For heat stress, Frey et al. (2016) identified two significant QTL for grain yield and one QTL for the leaf scorching trait. Similarly, McNellie et al. (2018) identified three major QTL, of which two major QTL were mapped on chromosome 3 (for plant death and leaf firing) and chromosome 1 (for leaf firing) in B73 × CML103 under heat stress at the late vegetative stage. The co-localized QTL for plant death and leaf firing reveals chromosome 3 as a potential region for heat stress tolerance. Inghelandt et al. (2019) identified QTL related to heat susceptibility index (HIS) of five traits on chromosome 2, 5, 9, and 10 using dent and flint maize inbred line populations. The study revealed an antagonistic pleiotropy between heat tolerance at seedling stage and adult stage. However, a low PVE was observed in these QTL making it not suitable for marker-assisted breeding applications.

Quantitative trait loci mapping under drought condition is associated with multiple traits such as anthesis–silking interval (ASI), root architecture, and grain yield traits. Giuliani et al. (2005) elucidated the role of Root-ABA1 on root architecture and leaf ABA concentration in response to different water deficit conditions. And in Landi et al. (2006) further validated the effect of this QTL under two environments (China and Italy). Almeida et al. (2013) mapped two significant constitutive QTL (chromosomes 2 and 6) and two adaptive QTL (chromosomes 5 and 7) for grain yield under drought stress. Semagn et al. (2013) identified four meta-QTL (mQTL2.2, mQTL6.1, mQTL7.5, and mQTL9.2) for grain yield under water-deficit and well-watered conditions. Zhao et al. (2019) identified three major QTL for drought stress tolerance (qKR-Ch.1-1 for kernel ratio, qEHPH-Ch.1-1 for ear height to plant height ratio, qGW-J1-1 for grain weight per ear, and qGW-Ch.4-1 for grain weight per ear) that could be targeted for introgression. Similarly, genetic regions that are associated with plant height, root length, and dry weight traits under salinity stress were mapped in the maize genome (Qiu et al., 2007; Osman et al., 2013).

Genome-wide associated studies (GWAS) is a powerful tool for dissecting genetic loci that significantly influence agronomic traits based on distinct phenotypes and extensive coverage of molecular markers. Many initiatives related to abiotic stress that include stress-linked genes in agricultural plants have used genome-wide association studies. For example, Liu et al. (2013) conducted GWAS using 368 maize varieties at the seedling stage and identified connections between the genetic variation in ZmDREB2.7 and the degree of drought tolerance in different maize varieties is highly dependent on DNA polymorphisms in the promoter region of ZmDREB2.7. In a study conducted by Thatcher et al. (2016), the genome-wide analysis showed significant changes in ears and leaves, magnifying that developmental splicing is linked to developmental stage, tissue type, and stress conditions. Song et al. (2016) identified cysteine-rich poly-comb-like (CPP) proteins in maize, and different expression levels of the ZmCPP gene under abiotic stresses (cold, salt, heat, and drought), indicated that ZmCPP functional characterization in maize could help understand maize growth and development. Using a genome-wide analysis of the AP2/ERF TF family, Zhou et al. (2012) showed that AP2/ERF gene family could participate in various stress responses such as drought and salinity. The same approach was carried out by Luo et al. (2019), wherein two promising genes, SAG4 and SAG6 (salt-tolerance-associated gene), were identified and have a great potential to develop salt-tolerant maize lines. Similarly, the GWAS approach was used to identify genetic regions associated with root-related traits, drought tolerance, and nitrogen use in maize (Wang N. et al., 2019; Guo et al., 2020). The success in GWAS utilization showed its great importance for studying multiple traits under drought, salt, and temperature stresses. GWAS studies in maize have identified novel gene candidates or genes responsible for abiotic stress, which can be used for maize breeding and designing climate-smart maize lines.

Maize (family Poaceae) comprises seven genera, including Chionachne, Sclerachne, Coix, Trilobachne, Polytoca, Tripsacum, and Zea. The genus Zea contains four species, but only Zea mays L. (2n = 20) is economically valuable. The other Zea species, referred to as teosintes, are essentially wild grasses native to Mexico and Central America (Doebley, 1990). The origin of maize (Zea mays L.) dates back more than 7,000 years and seems to have developed from annual teosinte (Zea mexicana) through gradual selection. The wild ancestor of modern maize has comparable plant architecture and growth forms to maize, while Tripsacum has higher chromosome numbers (2n = 36, 64, or 72), making it more challenging to hybridize with maize. These wild ancestors of maize have considerable genetic diversity and could be a new resource for germplasm enhancement (Maazou et al., 2016). For example, molecular evolution studies on maize have included genetic and genomic tools developed for Tripsacum (Blakey et al., 2007). Data from orthologous genes in maize and the Tripsacum genus revealed a distinct collection of genes with frequent non-synonymous substitutions in Tripsacum, which were prevalent when domestic maize was adapted to temperate regions through artificial selection. An intermediate metabolic route, phospholipid metabolism, was linked to cold and freezing tolerance. Anatomical descriptions such as aerenchyma tissue in roots and other properties in Tripsacum have contributed to survival under drought stress. Similar studies on root penetration and increased biomass in Tripsacum have revealed drought resistance. Physiological data revealed that the exceptional drought resistance of the Tripsacum genus is due to its high-water use efficiency and photosynthetic levels (Kemper et al., 1998). Later research revealed that the Tripsacum-introgressed cultivar grows better than modern maize under drought stress. Tripsacum introgression lines appear to have better-rooting systems that penetrate deeper into the soil and higher grain yields than modern maize cultivars (Gilker et al., 2002; Eubanks, 2006; Gitz et al., 2013). An evaluation of maize (Zea mays ssp. mays) × Tripsacum dactyloides L. hybrid (eastern gamagrass) calli revealed that hybrid plantlets had higher fresh weight than maize plants under salt stress and thus improved salt tolerance (Shavrukov and Sokolov, 2015). These hybrids retain sodium in their leaves, decreasing water potential and maintaining turgor pressure, vital for vegetative development (Pesqueira et al., 2003). Therefore, tapping the genetic potential of wild relatives is necessary to create new genetic resources that can serve as future accessions to develop maize lines with improved abiotic stress tolerance.

Traditional maize breeding can be used to develop high-yielding hybrids from selective parental mating based on specific combining abilities. The development of parental lines require 4–6 years and another 1–2 years for their hybrids, resulting in a long and somewhat complicated breeding cycle. To address this, speed breeding approaches were introduced (Hickey et al., 2017; Samantara et al., 2021). Several approaches can be used to shorten the breeding cycle in maize including (a) off-season nurseries, (b) double haploid (DH) technology, and (c) in vitro nurseries. The off-season nursery approach grows crops in a different location suited to their photoperiodism. DHs can be produced in two generations, significantly reducing the breeding cycle (Geiger and Gordillo, 2009). The most common method for maize DH uses the R1-nj color marker. In vitro nurseries are another option for breeding homozygous and homogenous lines, with a mix of DH technology (homozygosity per generation) and off-season nurseries (generations per year). These methods have several advantages and disadvantages. Off-season nurseries require quarantine clearance or other permissions to grow crops in a different location, extra time to transport genetic material and could introduce new pests and diseases. DH technology can have low success rates, hindering application for haploid induction and tissue culture adaptation. In addition, doubling is genotypic dependent and uses a carcinogenic agent (colchicine).

Genomic selection helps accumulate favorable genes with minor effects and improves multigenic traits, especially in stress-prone environments with a high environmental variation. The genomic selection focuses on estimating breeding values using many molecular markers that ideally cover the entire genome to predict the genetic value of the candidate for selection and using genetically estimated breeding values (GEBVs) to advance populations through rapid-cycle genomic selection without phenotyping in each cycle (Massman et al., 2013). Figure 4 is a schematic diagram integrating genomic selection and speed breeding for varietal and line development.

Genomic selection improves polygenic traits, governed by small-effect genetic loci, such as plant yield. This tool uses marker effects across the genome to estimate GEBVs. Genomic selection produced higher stover and grain yield than MAS in a bi-parental maize population (Massman et al., 2013). Similarly, two fast cycles each year via genomic selection increased grain production by about 2% in a multi-parental maize population (Zhang et al., 2016). The accuracy of genomic predictions correlated with the test cross’s marker-predicted genotypic and phenotypic values in the validation population. The model’s genomic selection prediction accuracy was more accurate than genomic selection and marker-assisted selection. It could be used to capture alleles with fewer additive effects to increase genetic gain under drought. The predicted results differed slightly between the training and validation sets or linkage disequilibrium with causal variants in another study. The mean performance of the breeding populations differed from the underlying predicted traits (Windhausen et al., 2012). Genomics selection combined with MAS (marker-assisted selection, GS-MAS) had advantages over the QTL-MAS approach for drought-stressed growth characteristics (grain yield, anthesis, and plant height), evident from the differences between R2 (QTL-MAS) and prediction accuracies (GS-MAS) in the analysis (Tilman et al., 2011). Combining molecular marker data and phenotypic data as input variables delivered higher-quality estimates for grain production on average than phenotypic data alone (Trachsel et al., 2019).

Maize is susceptible to abiotic stresses, especially drought and waterlogging, allowing a higher degree of genotype × environmental interaction in a breeding program. To identify those with higher yield potential in the target environment, maize lines/hybrids were evaluated under targeted or managed stress conditions. However, most of the parents/hybrids/crosses were discarded due to poor performance in the field, which could be due to the low genetic value of the parents selected for crossing. Thus, parents of high GEBV should be used for crossing to accumulate the maximum number of favorable alleles for desirable traits in the progenies. Further, genetic gain is limited under abiotic stresses and high environmental error (Trachsel et al., 2019). Thus, combining genomic selection and speed breeding for abiotic stress tolerance would enhance the selection accuracy of parents to increase genetic gain and abiotic stress tolerance.

Genome editing tools are some of the major breakthroughs in molecular biology. These techniques offer a precise and efficient way to edit an organism’s genome. Some of the well-known genome editing tools are ZFNs (zinc-finger nucleases), TALENs (transcription activator-like effector nucleases), and clustered CRISPR (regularly interspaced short palindromic repeats)/Cas systems (Miller et al., 1985; Jinek et al., 2012; Joung and Sander, 2013). These tools use common sequence-specific nucleases to identify specific DNA sequences in the genome and generate desired double-stranded breaks (DSBs) (Mushtaq et al., 2021). ZFNs have been used to modify various crop plants, including maize. The ZFN-mediated targeted transgene integration was used to stack characteristic traits, especially for stress tolerance, in maize by combining several beneficial features (Ainley et al., 2013; Mushtaq et al., 2021). On the other hand, TALENs are a promising genetic tool for targeted gene mutagenesis in maize and other crops. This approach has been used to produce stable mutations at the maize glossy2 (gl2) locus that are heritable. The transgenic maize lines with mono-di-tri- allelic mutations conferred the glossy phenotype (Char et al., 2015). In most cases, integrated TALEN T-DNA separated independently from the new loss of function alleles, giving rise to null-segregant offspring in the T1 generation; thus, TALENs are an efficient tool for maize genome mutagenesis for identifying gene function and improving abiotic stress-related characteristics. Constructing TALEN repetitions is complex and TALEN gene targeting effectiveness varies.

CRISPR/Cas is a flexible genome editing technology with potential applications. Compared with ZFNs and TALENs, the CRISPR/Cas system has been adopted rapidly in plants due to its high efficiency, simplicity, cost-effectiveness, and ability to target multiple genes (Cong et al., 2013; Hsu et al., 2014; Razzaq et al., 2019). Several abiotic stress tolerance studies have been carried out using CRISPR/Cas. For example, drought-resistant characteristics have been incorporated into maize using CRISPR/Cas9 gene insertion and substitution techniques. To explore the targeted use of ARGOS8 native expression variation in drought tolerance breeding, CRISPR-Cas was used to generate novel variants of ARGOS8 (Shi et al., 2017). The low-expression gene ARGOS8 in maize, which negatively regulates ethylene responses, is crucial for plant growth. Field evaluation of lines showed that the ARGOS8 variations produced five extra bushels of grain production per acre under drought stress compared to the wild type. These findings show that CRISPR-Cas9 technology can create new allelic diversity in crops by substituting breeding drought-tolerant crops. In another study, CRISPR/Cas9 was used to unravel the abilities of the ZmWRKY40 gene encoding a transcription factor (Wang et al., 2018b), which improved drought tolerance, and resultant lines imparted drought tolerance in maize.

Therefore, genome editing techniques have a great potential to improve crops. After carefully selecting the genome-editing tool, target sequences can be designed and introduced into the most appropriate vectors. Suitable genetic cargo (DNA, RNA, or RNPs) is then selected for delivery by (i) modifying the targeted sequence, (ii) regenerating the edited calli, and (iii) producing the edited plants. Combining earlier genome editing tools with newly developed breakthroughs is speeding up genome editing in crop breeding to meet the world’s exponentially increasing need for food. In addition, climate change necessitates flexibility and ingenuity regarding crop resilience and production methods. Moreover, we must consider government restrictions and the public acceptability of these new breeding methods.

Numerous research digging into its systems biology, maize has benefited from the introduction of omics techniques. Because maize and its derivatives are used as both food and bioethanol, the metabolism of maize has received significant study. As reviewed by Medeiros et al. (2021) several aspects of maize metabolism have received significant attention, including (1) the role of the metabolome in basic molecular processes, responses to biotic and abiotic stresses, and beneficial biotic interactions; (2) the nutritional composition of maize kernels and the molecular mechanisms that underpin the production of specific metabolites; and (3) the mechanisms by which the metabolome and metabolic models link to leaf physiology. (4) The metabolic changes caused by genetic modifications; and (5) the degree of natural metabolic variance and its potential utility in breeding efforts.

In a study by Alvarez et al. (2008), changes in stomatal hormones like as ABA and cytokinins were discovered, underlining the importance of root-to-shoot signaling in maize under drought condition. Furthermore, an increase in the content of phenylpropanoid pathway intermediates was found. In a study by Casati et al. (2011), metabolomics was used to document the candidate signaling molecules associated with temporal effects of irradiation of canopy leaves in maize. The study identified myoinositol as a candidate molecule for UV-B responses. Witt et al. (2012) used the same approach to assess metabolome changes under drought condition. However, differential changes were not detected in their experiment which could possibly be attributed to experimental design such as the use of greenhouse instead of actual field setup. Metabolite profiling under salt stress conditions was carried out by Richter et al. (2015). Using two differently resistant maize hybrids, they were able to deduce that TCA cycle and sugar metabolism are highly affected biochemical pathways during salt stress. Sun et al. (2016) investigated the metabolic growth and response of maize to cyclic drought using a metabolomic approach. Their study demonstrated that distinct metabolic pathways in maize plants returned to normal at different speeds during recovery. However, metabolic study revealed quantitative differences between the two cycles, indicating the intricacy of metabolic processes launched by water cycle alterations.

Other than pathways, metabolomic studies have aided in showing which organs in maize are mostly affected by abiotic stress. For example, during drought condition, leaf blades were identified to have the greatest metabolic changes (Witt et al., 2012; Obata et al., 2015). And under high-salt conditions, metabolic changes were found to be greater in shoots than in roots (Gavaghan et al., 2011). In a study conducted by Ganie et al. (2015), metabolomic approach highlighted that under phosphorus starvation conditions, leaves are the main site for metabolic changes.

Conventional phenotyping uses destructive approaches to measure drought, heat, and salt tolerance characteristics. However, the development of high throughput phenotyping (HTP) platforms has greatly reduced the phenotyping issues that limit breeding programs. To date, several high-throughput phenotyping methods have been used to characterize maize phenomes, including penetrometers, electrical conductivity sensors to assess soil mapping variability, thermal imaging to monitor plant canopies, and spectral reflectance (Masuka et al., 2012). In addition, satellite imagery, mobile cameras, UAV imaging, and ground-based imaging platforms are widely used in maize plant stress assessment.

Typically, high-throughput plant phenotyping is accomplished by collecting photos that quantify attributes across a crop plant’s whole life cycle as traditional phenotyping is inefficient, expensive, and inaccurate (Arya et al., 2022). Nowadays, satellite imaging technology has become a powerful tool for collecting data in large agricultural practices, but few apply to small experimental breeding plots. For instance, satellite-driven NDVI (normalized differential vegetation index) strongly correlated with UAS multispectral imagery in maize plots (24 m2) (Sankaran et al., 2020). Unmanned aerial vehicle-red, green, and blue (UAV-RGB) images supplemented UAV thermal images for precise extraction of maize canopy temperature (Zhang et al., 2019a). In addition to general data measurement methods, advanced tools such as digital imagery, stable isotopes, spectral reflectance, and thermal imagery are used to improve data accuracy for soil and climate measurements. Maize studies have used SPAD meter, NDVI, and infrared gas analyzer (IRGA) to quantify environmental effects, digital imagery to measure early biomass in response to water stress, and thermal imagery to quantify leaf temperature during transpiration (Marti et al., 2007; Jones et al., 2009; Rorie et al., 2011). In several studies, infrared thermography (IRT) has been used to measure temperature differences between leaf, air, and canopies under heat and drought stress, spectroscopy to monitor the photosynthetic rate of leaves and canopies in response to the early onset of water stress, and fluorescence imaging to scrutinize plant growth during drought and low-temperature stress (Zia et al., 2013; Lee et al., 2015; Boote et al., 2016). Recently, fluorescence imaging was used to investigate plant chemical composition via spectral absorption and reflectance from cell to canopy level (Zhang et al., 2019a). The multispectral imaging techniques like X-ray-CT, MRI, and ultraviolet spectra (UV) can report changes in ionic balance, stomatal conductance, and transpiration, contributing to drought and high-temperature stress resistance in maize (Huang et al., 2016).

The fast-paced development of these HTPs have generated large number of datasets. And to date, one of the major challenges is the analytical approaches for these datasets. As previously reviewed, machine learning and deep learning techniques are some of the emerging techniques that can be used in the identification of hidden relationships between large datasets (Singh et al., 2018, 2021; Arya et al., 2022; Gill et al., 2022). In a review presented by Singh et al. (2018), both ML and DL are identified to be seamlessly integrated into data acquisition, data preprocessing, and data analytics for real-time HTP of plant traits in the field. Recently, Gill et al. (2022) highlighted that the current ML-based techniques focuses on a single stress or disease on a leaf or canopy, but in real-world situations, numerous diseases and stresses may appear on a single leaf or canopy. Therefore, ML platforms must be flexible and robust, with the ability to distinguish multiple disease symptoms on a single leaf or within the same plant canopy. Collectively, the quantity of data needed to train ML models is determined by the complexity of the problem and the complexity of the learning algorithm; thus, for wider applicability, the training data should really be continuously updated using techniques such as artificial learning to reflect the complexity of stress symptoms for the targeted crop.