For decades, the main selection criteria among maize breeders were to improve grain yield and grain yield stability across environments, and the D used to select superior genotypes varied from 4 pl m− 2 in 1960´s to 8 pl m− 2 in 1990´s (Eyherabide et al. 1994). Moreover, during the last decades superior genotypes are tested at 16 pl m− 2 during the advanced phases prior to the commercialization (Lee and Tollenaar 2007). Thus, these breeding strategies have forced to select genotypes with compact plant type, and restricted vegetative and reproductive PP that would be the best physiological response to avoid bareness and grain yield penalties at high D (Haro et al. 2013). However, the paradigm that governed maize breeding during the last decades should change in the short term due to three current events: i) several regions of the world maize production area involve limiting environments in the supply of resources (Grassini et al. 2015; Dimes et al. 2015; Rodriguez et al. 2016; Rotili et al.), ii) the adoption of technology for seeding variable rates has increased specially in farms with high environmental variability (Melchiori and Garcia 2018), and iii) the current context of climate change and its impact on rainfalls (IPCC 2014) would generate environments that would become increasingly extreme and variable (Kusmec et al. 2017). Consequently, breeders are challenged to exploit PP (Monforte 2020) to develop high-yielding genotypes, capable of responding to contrasting supply of resources with high PP for traits related to resource capture and grain yield. Our study was focused on PP of plant architecture traits (i.e., those traits related to capture of aerial resources). Results showed a high variation in PP of these traits which would allow to select genotypes with contrasting PP for different target environments. The focus was not only placed on PP of architecture traits but also on the environmental modulation of the relationship between the trait per se and its PP. Such information would allow to improve genotypes for a collection of environments (de Felipe and Alvarez Prado 2021). Following this purpose, some plant architecture traits (AZ, EH, PH, EHPH, and SD) had a similar capacity (symmetrical) of response to PP under favorable (e.g. low D) and unfavorable (e.g. high D) environmental conditions, but others showed an asymmetrical response (LA, AV, LFLL). Interestingly symmetrical or asymmetrical responses of 90th and 10th percentile traits to their PPs may fit well with the better performance of plant architecture across environments. Thus, selection of lines with higher LAP (asymmetrical response, with a higher LA response under low D than high D), AVP (asymmetrical response, with more planophile leaf habit at low D but a erectrophile leaf habit at high D), AZP (symmetrical response, with leaves close to 0° or 90° at low and high D; respectively) and EHPHP (symmetrical response with higher and lower values at low and high D; respectively) seem to be possible. By contrast, selection of lines with high SDP (symmetrical response) would be undesirable, especially for high D crops, because thinner stems could generate lodging (Flint-Garcia et al. 2003).
Some studies have postulated a negative relationship between the heritability of the trait per se and its PP, which would limit the efficiency of selection of both traits simultaneously. This hypothesis was not rejected for grain yield and grain yield components of several cereals (Sadras and Slafer 2012), and kernel weight in maize (Alvarez Prado et al. 2014), and was tested for leaf traits of several species including maize (Donovan et al. 2011). However, in our work this hypothesis was partially rejected since no relationship between heritability of the trait per se and PP was found for most traits with low PP, while a negative relationship was observed for two traits (SD and AZ) with PP above 0.84, i.e., those traits strongly controlled by the environment. These differences could not be attributed to the approximation used to calculate heritability of traits per se, but to the number and type of trait and the methodologies for PP estimation (Finlay and Wilkinson 1963; Sadras et al. 20092010). Moreover, comparing PP values of plant architecture traits obtained in this work, with PP values of kernel weight related traits using the same methodology for PP estimation (Alvarez Prado et al. 2014), we obtained higher PP values (maximum PP values of 0.96 vs 0.55).
Despite of a “plastic phenotype” was considered rather a nuisance initially (Bradshaw 1965), PP was lately (Bradshaw 2006) considered a trait per se with a complex genetic basis that can be selected to adapt genotypes to certain environments. Hence, studies of the genetic control of PP would be relevant. Some studies showed that in maize, the genetic control of several traits and their PPs were independent, e.g. foliar expansion (Reymond et al. 2003), kernel weight and its physiological determinants (Alvarez Prado et al. 2014) and several traits related to phenology, ear and plant morphology (Kusmec et al., 2017). We found that PP of plant architecture traits presented an independent genetic control of the trait per se, not rejecting the hypothesis postulated by Bradshaw (1965). This statement would also be supported by the low correlation values between the traits per se and their PPs and because no QTLs of PP of plant architecture traits co-located with QTLs for traits per se detected by Incognito et al. (2020) for the same RILs.
Regarding the study of genomic regions related to PP, some QTLs co-located, suggesting pleiotropy in the genetic control of PP of these traits. In three regions located at 2.27 cM of chromosome 5, at 1.16 cM of chromosome 8 and at 0.97 cM of chromosome 10, a QTL of LAP co-located with one of LLP, in contrast to the phenotypic relationships that indicate that LW was the most important trait by which plants regulate their LA (Sonohat and Bonhomme 1998 ; Incognito et al. 2020). The region located at 2.27 cM of chromosome 5 was the most important, because it explained a high percentage of variance explained (PVE) (9.8%) and because it contains genes directly involved in leaf growth such as glossyN681A (Neuffer and England) and leaf function as chloroplast RNA processing2 (Barkan et al. 1993), together with another gene that could indirectly affect LA regulation by intra-plant shading such as the dwarf candidate6 (Todesco et al. 1990). Another QTL for LLP located at 1.06 cM of chromosome 8 is of interest, since it explained 10.6% of PVE and contains the candidate gene compact plant1 (Nelson Jr. and Ohlrogge 1961) that would proportionally reduce all vegetative organs of the plant.
For leaf orientation, an interesting QTL for AZP is the one located at 1.21 cM of chromosome 10, because it presented a high PVE (9.1%) and contains the candidate gene dwarf candidate3685 (Winkler and Helentjaris 1993) that determine the elongation of plant organs mediated by the synthesis of gibberellin. Probably a change in the architecture of the plant promoted by gibberellin, such as twisting of leaf sheaths or of internodes triggered by the presence of neighbors could produce the reorientation of leaves towards inter-row spacing (Girardin and Tollenaar 1994), reducing shading among plants of the same row (Maddonni et al. 2001a). For LFP, the QTL of greatest interest which presented the highest PVE (15.7%) of all the detected QTLs in this study, was the one located at 1.27 cM of chromosome 5. This QTL in addition to including the gene glossyN681A (Neuffer and England 1995) previously mentioned, would contain the gen dwarf candidate6 (Todesco et al. 1990) and the extended auricle1 (Osmont et al. 2003) that with a synergistic effect with the liguless1 (Becraft et al. 1990) and liguless2 (Walsh et al. 1998) would affect the establishment of the blade-sheath boundary potentially modifying the future growth of the leaf. Additionally, the region where a QTL was located with a high PVE (11.2%) for VAP at 1.41 cM on chromosome 1, would constitute another region of interest containing the gen humpback1 (Schneeberger et al. 1996). This gene would generate the proliferation of the sheath just below the auricle resulting in a bulging sheath that could potentially cause changes in the VA especially for leaves above of the ear node, with potential impact on carbon assimilation (Cagnola et al. 2021).
Finally, for the PP of stem architecture related traits, the QTL detected for EHP at 1.15 cM on chromosome 9, generates the greatest interest because it presents a high PVE of 13.1% and because it is on a chromosome with the highest density of QTLs for stem architecture associated traits found by Incognito et al. (2020) for the same population used in this study. Genes of interest that could affect EHP were described in this region, like yellow-green2 (Neuffer and England 1995) and semaphore1 (Scanlon et al. 1996) which could produce shorter or brachytic plants, respectively, avoiding penalties on grain promoted by stalk lodging.
The study of the genetic bases of PP would strongly contribute to the success for future maize breeding in a rapidly changing environment (Monforte 2020). Although some causal genes controlling PP of traits have been studied, this field of research is in early stages (Laitinen and Nikoloski 2018). In addition to the availability of high-throughput genotyping technologies, new genomic tools together with the increasing use of automated phenotyping will help to increase knowledge about the genetic and molecular mechanisms underlying PP. In agreement to this, our results provide useful information about genomic regions involved in the PP of plant architecture traits in maize. More QTL analysis of PP on plant architecture traits using different genetic backgrounds are needed, in order to find stable QTLs capable of being used in marker assisted selection to develop genotypes with greater adaptation to environmental changes.