Global climate change is considered one of the biggest and most complex challenges the mankind has faced. One effect that has been observed since 1970, which leads to severe yield losses, is the increased occurrence of erratic drought phenomena (Liu et al., 2019). Specifically, the agricultural sector is facing serious challenges since drought-stress is considered the most limiting factor in rain-fed cropping systems (Hu and Xiong, 2014). Based on calculations of the Intergovernmental Panel on Climate Change (IPCC), it is predicted that the global mean surface temperature will rise 2°C more in the 20-year period from 2046 to 2065, than in the comparable period between 1986 and 2005, with a total increase of 4.8°C by 2100 (IPCC, 2015). As a result, more frequent heat and drought events are to be expected and to be classified as a major threat to the primary production sector in general and the wheat production in particular (Steinfort et al., 2017). Based on crop modelling scenarios, it is predicted that global wheat production will fall by 6% per 1°C temperature increase (Asseng et al., 2013). In particular, the Australian wheat production region is expected to experience a strong increase in drought and heat events, with a yield decrease of up to 20% projected from a median temperature increase of 2°C (Asseng et al., 2015). The severe impact of strong drought events has already been observed during the “Millennium Drought” between 2001 and 2009, where major reductions in production were recorded. In southern Australia in particular, production was severely decreased due to the impact of drought during this phase (Dijk et al., 2013). Large areas of the southern and western wheat-cropping regions in Australia have a Mediterranean climate, which is defined by terminal droughts (Siddique et al., 1990; Whan et al., 1991; Botwright et al., 2002; Rebetzke et al., 2008; Sadras and Dreccer, 2015; Rebetzke et al., 2017).

Consequently, there has been a growing focus on commercial and public breeding programs to identify traits associated with water-use efficiency to increase the yield potential under water-limited conditions (Lopes and Reynolds, 2012; Passioura, 2012). Potential traits of interest include long coleoptiles, which enable deeper sowing in the soil profile and improved access to water reservoirs underneath a dry surface soil, as well as reduced tillering to lessen unnecessarily metabolism into the non-fertile emerging tillers (Richards et al., 2010). Another is early vigour (EV), defined as the rapid production of leaf area during the early development phase of the plant (López-Castañeda et al., 1996). The primary advantage of EV is the increased biomass production early in the season and the rapid closure of the canopy, which can reduce evaporation of soil water, which then increases water availability (López-Castañeda and Richards, 1994; Condon et al., 2004). Early canopy closure also leads to minimized solar radiation on the soil and to an enhanced competitiveness of the crop against weeds (Dingkuhn et al., 1999; Coleman et al., 2001; Lemerle et al., 2001; Bertholsson, 2005). Mediterranean growing areas are water-limited environments which are characterized by experiencing late seasons droughts. According to Richards et al. (1987) and López-Castañeda et al. (1995), EV offers great potential for increasing water-use efficiency in such drought-prone regions. Additional advantages that may be associated with EV include larger uptake of essential plant nutrients, superior tolerance to aluminium stress, as well as improved yield under high temperatures and elevated atmospheric CO₂ concentration (Coleman et al., 2001; Lemerle et al., 2001; Liao et al., 2004; Bertholsson, 2005; Ludwig and Asseng, 2010; Valle and Calderini, 2010; Ryan et al., 2015). Previous studies in wheat have already identified various physiological traits associated with EV, such as the embryo size (Moore and Rebetzke, 2015), coleoptile length (Clarke et al., 1991; Rebetzke et al., 2007), tiller size, and leaf characteristics (Rebetzke and Richards, 1999; Rebetzke et al., 2007, 2017).

Several studies have highlighted the beneficial effect of increased EV on yield performance in specific environments. Nevertheless, due to insufficient knowledge about genetic variation and lack of information on the economic value of the trait, EV has only been introduced into breeding programs to a limited degree (Rebetzke et al., 2017). However, Botwright et al. (2002) demonstrated the positive effect of EV on yield performance in medium and low rainfall regions in combination with favourable soil conditions, such as sandy soils. Early vigour has significant implications for water demand. For example, a slight increase in leaf area growth during the vegetative growth stages can lead to an increase in biomass, transpiration area, and water use (Asseng and van Herwaarden, 2003). This can result in rapid depletion of soil water prior to anthesis, which may have a negative impact on grain yield during flowering time and grain filling (Richards and Townley-Smith, 1987). According to Turner and Nicolas (1998), strong vigorous genotypes have deeper and greater water uptake compared to less vigorous genotypes and are subsequently considered to be advantageous in low rainfall environments, including Mediterranean growing regions. However, despite a clear genetic effect on EV, its interaction with the environment, soil type, and available fertilizer is substantial. For example, EV may adversely affect yield if managed unfavourably. Therefore, ensuring that the crop is provided with sufficient nitrogen (N) is essential in order to prevent premature N deficiency due to excessive biomass production (Asseng and van Herwaarden, 2003).

To determine the impact of EV on yield, a better understanding of the physiological mechanisms of the trait along with its genetic control is required. In addition, efficient and low-cost phenotyping procedures are needed to assess EV in wheat. The aim of this study was (i) to investigate the physiological characteristics that drive EV in wheat in the field and under controlled conditions, (ii) to assess the efficiency of phenotyping methodologies under greenhouse and field conditions, and (iii) to identify genomic regions influencing EV in wheat.

The first objective of this study was to provide information on physiological components that contribute to EV of wheat, both in the field and under controlled conditions. Several studies observed a strong contribution of leaf width to a specific leaf area, and consequently to EV (Rebetzke and Richards, 1999; Richards and Lukacs, 2002; Maydup et al., 2012). Our results showed that the variation in EV is strongly associated with the leaf length of both L1 and L2 in the field, as well as in the GH. For both leaves (L1 and L2), leaf length is the main contributor to a larger leaf area development, and, consequently, to an increased EV. Furthermore, leaf length of L2 showed a slightly greater impact on the area of L1 and L2 with an advancing development. This finding agrees with other studies reporting a significant positive correlation between the area of L2 and the leaf length (Nursinow et al., 2011; Boden et al., 2014; Moore and Rebetzke, 2015; Duan et al., 2016). The data revealed faster development of L2 in the GH compared to the field studies. This was also observed by Rebetzke et al. (2007) and leads to the conclusion that EV can be recorded at earlier stages under controlled conditions than in the field. Previous studies have reported that embryo size is a highly heritable trait that is strongly associated with leaf area (López-Castañeda et al., 1996; Rebetzke and Richards, 1999; Aparicio et al., 2002; Moore and Rebetzke, 2015). Since it has been established that embryo size increases with seed weight in wheat (Moore and Rebetzke, 2015) and barley (López-Castañeda et al., 1996), this study used seed weight to indirectly evaluate the impact of embryo size on EV. In the GH experiment, it was not possible to demonstrate the positive impact of embryo size on EV by using seed weight, since no significant correlation with any measured leaf parameter was observed. Nonetheless, several studies reported a positive effect of the embryo size on early vigour in wheat (Rebetzke and Richards, 1999; Richards and Lukacs, 2002; Moore and Rebetzke, 2015). In other studies, however, the total variation in EV could not be exclusively explained by the considering seed weight (Maydup et al., 2012), and the seed density has also been suggested as a potentially more decisive factor in determining EV (Ball et al., 2011). This partially explains why no correlation between seed weight and EV parameters could be found in the present study. In the GH experiment, biomass was recorded and was exhibited as a significant correlation with the area of L2, as well as area of L1 and L2, suggesting an increased above-ground biomass for genotypes with greater EV. Richard et al. (2019) reported that lines with increased above-ground biomass were strongly associated with an increased grain yield.

The second objective of this study was to evaluate high-throughput methods to precisely record physiological leaf characteristics in field and greenhouse trials. Mullan and Reynolds (2010) reported a significant correlation between NDVI, leaf area, and biomass. These PCA results from the current study tended to confirm that NDVI and the leaf area are related. Although the biomass in the field was not separately measured, a significant positive correlation between biomass and leaf area in the GH experiment suggested a putative relationship. In the correlation analysis, only one significant correlation between NDVI and leaf parameters could be established, which was NDVI17 at 21 DAS and leaf length L2 at 21 DAS. This confirms that leaf length L2 contributes more to an increased EV compared to leaf dimensions of L1. This is also supported by other studies (López-Castañeda et al., 1996; Richards and Lukacs, 2002; Duan et al., 2016). In terms of high-throughput phenotyping methods for EV in GH environments, this study tested the ability of a green pixel counter as a low-cost method. Since embryo size strongly affects EV, the method is not able to explain the trait completely. However, the results for PLA calculated by the green pixel counter indicate great potential to measure EV under controlled conditions. In particular, the leaf parameters of L2 showed a significant correlation with PLA, as well as with the biomass. Furthermore, the green pixel counter was successfully used to measure coleoptile tiller length, a trait which strongly affects EV and is also highly correlated to embryo size.

The third aim of this study was to achieve a better understanding of EV genetics in wheat. The GWAS for Set 1 and Set 2 revealed 41 SNP markers for NDVI and for PLA, which were linked to 60 protein-coding regions across 17 chromosomes. Consistent with previous studies, the present study shows that EV in wheat is a quantitative trait with numerous QTL located across several chromosomes (Li et al., 2014; Moore and Rebetzke, 2015; Boudiar et al., 2016). Numerous studies have reported the effect of dwarfing genes on coleoptile length (Rebetzke et al., 2007; Li et al., 2017), coleoptile width (Rebetzke et al., 2014), and leaf epidermal cell dimension (Botwright et al., 2005). In most of these studies, the influence of dwarfing genes was highlighted. The gibberellic acid (GA)-insensitive dwarfing genes Rht B1b on chromosome 4B and Rht D1db on chromosome 4D have been reported to reduce coleoptile length and, consequently, EV, since they decrease epidermal cell length in leaf tissue (Ellis et al., 2004; Rebetzke et al., 2007; Yu and Bai, 2010; Li et al., 2017; Rebetzke et al., 2017). In our study, no SNP markers were detected on either one of the 4B or 4D chromosomes. This suggests that NDVI and PLA are the less affected traits by the presence of Rht B1b and of Rht D1b. Comparable results were reported in Li et al. (2014), were none of the parental lines that carried Rht-B1b, and only one parental line contained Rht-D1b, while no QTL was identified on chromosome 4D. However, a significant correlation could be observed between Rht-D1 and NDVI and the ground cover in certain environments. The Rht 8, on the short arm of chromosome 2D, is a GA-responsive dwarfing gene reported to have a secondary effect of reducing epidermal cell length. Hence, it is more appropriate for achieving good canopy cover in combination with a semi-dwarf growth habit (Botwright et al., 2005). Chai et al. (2019) reported the WRKY transcription factor TraesCS2D01G051500 as a possible candidate gene for Rht8. We identified QTL QSG.qwr-2D.1 in the vicinity of TraesCS2D01G051500, and found this QTL to be significantly associated with NDVI17 at 29 DAS. Several studies have identified QTL associated with coleoptile length on chromosome 1B, including markers XpGTG-mTCGA294 (Yu and Bai, 2010) and wsnp_CAP11_c2596_1325540 (Ma et al., 2020). This study confirms that chromosome 1B is a region of interest for EV, since three significant QTL were detected on this chromosome. That applies particularly to QSG.qwr-1B, which is located at the same region on the long arm region of chromosome 1B as wsnp_CAP11_c2596_1325540, as reported in Ma et al. (2020). Another region of interest is chromosome 5A, which is also considered as the most important chromosome for stay-green traits (Shi et al., 2017; Liu et al., 2019), including the isopentenyl transferase gene (Gan and Amasino, 1995). Furthermore, chromosome 5A harbours several major developmental genes, such as the vernalisation gene Vrn1, frost resistance gene Fr1, as well as genes for ear emergence time and for the plant height (Sutka and Snape, 1989; Kato et al., 1999; Galiba et al., 1995).

The results of the current phenotypic investigation extended our insights into the EV trait in wheat. The key characteristics of EV and the relationship with other traits, such as biomass, were successfully identified. In addition, this study presents effective methods that can be used to detect EV in the field, as well as under controlled conditions. In particular, the connection between the leaf length parameters and the NDVI highlights the great potential of NDVI, especially if given recent advances in unmanned aerial vehicle or drone phenotyping platforms (Shi et al., 2016). Nevertheless, potential interactions due to environmental factors must be clarified by practical crop management for a better understanding, since factors such as sowing time, sowing depth, sowing rate, and row spacing may also influence EV. In particular, the interaction and value of EV in specific target environments must be clarified. Furthermore, it has to be considered that yield predictions based on EV can be very challenging. Since the trait is recorded at a very early developmental stage, its relationship to yield performance can subsequently be influenced by a multitude of complex environmental factors. For example, abiotic stress factors have a particularly decisive influence on yield and its yield components, especially during key developmental stages, such as tillering and flowering. Nevertheless, EV is essential for good crop establishment and, therefore, can impact yield even at this early stage. Hence, we suggest incorporating EV measurements into experiments by using the NDVI data in performance evaluations, such as stay-green trials. Our genomic analysis has identified QTL that is associated with EV, which are co-located or are closely linked to key genes controlling the plant development, such as plant height, coleoptile length, stay-green, and vernalisation. The results support the theory that EV is a trait regulated by pleiotropic genes. These findings may help identify the key drivers and determine potential trade-offs with important agronomic traits. Given that the trait is underpinned by many QTL with small effects, marker-assisted selection or gene-based approaches are likely to be challenging; however, genomic prediction approaches provide a suitable option for future breeding.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

JC and LTH conceived the study. SV, SA, and JC performed the experiments. SV and AS analysed the phenotypic data. SV wrote the manuscript with further input from AS, RJS, JC, SA, and LTH. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.