The current study showed that drought stress significantly increased the ASI (from 4.4 to 5.6 days) and reduced the PH (from 127.2 cm to 111.4 cm), indicating that silk extrusion is significantly delayed and plant architecture is greatly affected under drought stress. These results are in pipeline with the previous reports, which also reported significant extension in ASI and reduction in plant and ear height under WS condition (Messmer et al. 2009; Xue et al. 2013). Water deficit at the flowering stage delays or inhibits the growth and female ear development leading to reduced ear biomass, from 1.71 g to 1.41 g (Table 1). It was reported that osmotic stress limits the dry matter accumulation by approximately 50% during serious water shortage (Hejnák et al. 2015; Zafar et al. 2020). Thus, considering the higher estimated heritability and vulnerable to drought stress, ear biomass could be an option for improving maize selection under water scarce condition (Edmeades et al. 2020; Setter 2012). In this study, PH was significantly correlated with yield related traits under both WW and WS conditions, plant height decreases less when exposed to drought can ensure sufficient "source" and exhibit better drought resistance. PH was positively correlated with EBM, suggesting a potential role of EBM in drought tolerance. No significant correlation was found between the ASI and ear biomass indicating the delayed silk extrusion has no relationship with the ear development. Therefore, the three traits may not be tightly correlated with each other or the correlation may be disturbed due to the variation in the fluctuating environment in the field. Based on our funding, we proposed that the selection of taller plants, shorter ASI and bigger EBM might boost yield under water deficit.
GWAS is a powerful strategy for genetic dissection of complex traits in plants. In this study, a total of 71, 159 and 21 SNPs were significantly (P < 10−5) associated with ASI, EBM, and PH by GWAS, which were located in 36, 81 and 16 genes, respectively. To date, only few GWAS studies have been conducted in maize drought tolerance under complex field condition, and few stable genomic regions were detected across various mapping populations and environments (Li et al. 2016; Xue et al. 2013). Multiple maize nested association mapping populations were tested under two contrasting water regimes for seven drought-related traits including ASI, PH, and yield related traits, and resulted in hundreds of promising QTLs and candidate genes through GWAS and linkage mapping (Li et al. 2016). In addition, many other candidate genes were detected to be associated with drought tolerance correlated yield and agronomic traits (Farfan et al. 2015; Hao et al. 2011; Lu et al. 2010; Xue et al. 2013), however, none of them were found in this study, which may be due to the different growth and climate conditions, and drought treatment. Fortunately, few candidate genes co-localized with reported QTLs in our study. Uncharacterized gene GRMZM2G173084, associated with ASI-WW in 2017, overlapped in the QTL for both ASI-WW and ASI-WS which was detected by joint linkage analysis in a CN-NAM population (Li et al. 2016). Zm00001d003939, encoding a 11-ß-hydroxysteroid dehydrogenase 2, is the candidate gene for PH-WS-2017 located in a consistent QTL for WW-PH which was identified in both CN-NAM and US-NAM population. In addition, several overlapping genes were identified under water treatments (Table 3) among different years, while only Zm00001d013992 was commonly identified in two-year environment, which implied maize drought tolerance is a complex trait, highly affected by environment and treatment.
Members of zinc finger family protein play critical roles in plants growth and developmental processes, including flowering, senescence, and also abiotic stress responses (de Lorenzo et al. 2007; Yan et al. 2017). A C2H2 zinc finger transcription factor determines stomatal closure by regulating genes related to H2O2 homeostasis, such as peroxidases, glutathione S-transferase and cytochrome P450s, thereby modulate drought response in rice (Huang et al. 2009). Here, we identified Zm00001d042997, encoding a HIT-type Zinc finger family protein was associated with ASI under both WW and WS conditions, indicating a potential conserved abiotic stress tolerant role of Zinc finger family protein. F-box domain containing protein ARABIDILLO-1 is conserved in plants, involved in root architecture development and functions during rice abiotic stress mainly through regulating root branching and lateral root development (Mu et al. 2010; Sharma et al. 2014). It was reported that Arabidillo-1 mediated protein degradation most probably through modulating the GA3 signalling pathway (Mu et al. 2010). However, another research revealed that ARABIDILLO 1 knock out and overexpression plants responded normally to auxin and abscisic acid (Nibau et al. 2011). Zm00001d029938, encoding ARABIDILLO 1 in maize, was associated with ASI in this study. However, whether it contributes to drought tolerance through a hormone dependent or independent way is still unknown. For EBM, a SNP (S5_27121944) which annotated as Pyridoxal phosphate dependent transferase (Zm00001d013992) was consistently associated with EBM for consecutive two years under drought regime (Table 3). Pyridoxal phosphate (PLP) is an active form of pyridoxine (vitamin B6) which functions as coenzyme in several reactions such as decarboxylation, deamination as well as transamination. The PLP dependent enzymes mainly perform in amino acid biosynthesis and the metabolism of its derived metabolites. Therefore, it is interesting to speculate that Zm00001d013992 might involve in amino acid metabolism and then promotes ear development in maize under drought stress condition.
The drought responsive patterns of these genes suggest that Zm00001d013992, Zm00001d029938, and Zm00001d039319 might positively while Zm00001d029937 might negatively correlate with drought tolerance, suggesting their potential roles in the drought tolerance. Based on the haplotype analysis, we identified favourable and rare allele for candidate genes. The lead SNP S5_27121944 (A/G) in Zm00001d013992 separates the association panel into two groups, and only around 6% lines (BY855 and BY4960, etc) carries the favourable haplotype exhibits higher ear biomass, 3 g under WS (Fig. 6A). These lines and causal SNPs could be selected to cultivate drought tolerant lines with big ear biomass. The abovementioned results might be meaningful for genetic improvement of drought tolerant maize aimed at shorten ASI under both WW and WS. Haplotype analysis displayed potential causal SNPs of candidate drought tolerant genes and thereby will benefit maize breeding in both genome selection and genome editing.
The findings of this study provide insights into the genetic basis of drought tolerance at the flowering stage especially for the female inflorescence’s development. Those overlapping genes are proposed as candidate genes for drought tolerance in maize. Moreover, it also provides genetic resources which could be used for drought-tolerance marker development and benefit for future marker assisted or genome-wide selection for drought tolerant maize breeding. Future research might explore the association signals in depth and the role of nonsynonymous SNPs, candidate genes function through mutation and gene editing method as well as the underlying molecular mechanism of maize ear and silk development under water deficit condition.