A new phenotyping method for root growth studies in compacted soil validated by GWAS in barley

Abstract

Background

Soil compaction is defined as the reduction of air-filled pore space affecting soil density, water conductivity and nutrient availability. These conditions negatively influence root morphology, root development and plant growth leading to yield loss. To date, the ability of roots to penetrate compacted soil has been investigated using high density agar or wax-petrolatum layers as a proxy for compaction. Nevertheless, these methods are not realistic and fail to account for the root-soil interaction that influences root growth ability.

Results

Artificially compacted soil lumps were prepared using natural field soil mixed with sand and vermiculite in a 1:1:0.2 ratio and adjusted to a final water content of 31%. A Genome Wide Association Study (GWAS) was performed to validate this new methodology, combining a panel of 139 barley cultivars with a Single Nucleotide Polymorphism (SNP) dataset of 5,317 polymorphic markers. The panel was evaluated at seedling stage for four traits: total root length, average of diameter width, seminal root number, shoot: root weight ratio and two novel Quantitative Trait Loci (QTLs) associated with total root length were identified on Chr 4 H and 5 H. Four genes (a Nitrate Transporter1 (NRT1)/Peptide Transporter (PTR) family protein 2.2, a Hedgehog-interacting-like protein, an expansin and a cyclic nucleotide-gated channel) were hypothesized as plausible candidates for further investigation, given their implication in root development. In addition, the new phenotyping method revealed an altered plagiogravitropism phenomenon in barley during root emergence in compact substrates. In uncompacted soil, only the primary root exhibits vertical gravitropic set-point angle while a variable number of embryonic seminal roots develop with a shallower growth angle. In contrast, in compacted substrate all roots developed vertically to restore the growth angle after reaching a length of 4–5 millimetres.

Conclusions

A methodology based on root-soil interaction is presented as a new method for root growth evaluation and genomic studies in seedlings growing in compacted soil.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13007-025-01408-2.

Background

Soil compaction causes a rearrangement of soil particles which in turn disrupts and reduces pore sizes. This soil condition is due to natural and anthropogenic factors such as high clay content, limited water availability and machinery tracks. The alteration in soil texture affects changes in the rate of heat transfer, reduces the pore sizes and consequently the air/oxygen content, decreases water and nutrients availability besides increasing the soil mechanical resistance and strength [1]. All together, these modifications lead to root rot, hinder cellular respiration and cause a stunted plant growth [2]. Moreover, soil oxygen depletion alters the redox potential of soil solutions reducing the availability of the nutrients such as N, P, and K and the root uptake capability [3, 4]. Hypoxia impedes also water infiltration and retention modifying the mechanical impedance of soil and compromising the interaction forces between soil particles [5]. At the same time, the increased cementation within substrate micro-aggregates combined with a poorly developed root system in both the topsoil and the subsoil leads to water lodging, mainly in heavy loamy soil [6]. A reduced water availability can alter the composition and the function of soil microbial communities, altering symbiotic associations between endophytic fungi and plants [7, 8]. There is evidence that plants grown in compacted soils have a reduced number of root-associated microbes [9] and reduced ability to form nodules upon rhizobium colonization [10].

It has been estimated that soil compaction affects more than 70 Mha worldwide [11], resulting in crop production losses of more than $ 300 million per year [12]. Some regional and more recent estimates show that 23% of European soils are subjected to critical high soil density [13].

Increased bulk density and poorly connected and discontinuous pore spaces affect negatively root morphology, root development and plant growth, leading to evident modification in Root System Architecture (RSA). Reduction of root length, increase of their diameter and tortuosity, increase in root hairs and lateral roots formation are the most evident modifications because of physical impedance imposed by the soil [5, 14, 15].

Genetic diversity for the ability of roots to overcome the mechanical resistance of soil compaction has been observed in several species such as Arabidopsis, maize, rice and durum wheat [16–19] although the genetic bases underlying these belowground traits are still largely unknown. The ability to penetrate the soil is an adaptive form of plasticity, allowing roots either to avoid impenetrable obstacles or to break through hard soil layers [20]. This ability has been extensively investigated in recent years and it is associated to the modulation of physiological mechanisms that shape RSA. Overall, plants can adopt two main strategies to cope with soil compaction. The first strategy relies on the obstacle-overcoming mechanisms, which in turn exert sufficient root growth pressure for breaking through micropores or cracks present in the subsoil. The secondary strategy is an obstacle-avoidance strategy, whereby roots reorient their growth direction to circumvent impenetrable obstacles such as rocks or hardpans. When roots fail to overcome soil obstacles, they explore new soil aggregates with lower mechanical resistance through different growth patterns (i.e., bending, bucking, waving, circumnavigating) that facilitate root penetration into compacted soil layers [21].

In recent years several genes associated with root growth in compacted soil have been identified. In a recent work in Arabidopsis, Xu et al. [22] have demonstrated that the soil emergence-related transcription factor PHYTOCHROME-INTERACTING FACTOR 3 (PIF3) controls root penetration via transducing external signals perceived by the receptor kinase FERONIA (FER). The loss of FER function in mutants resulted in a severe defect in root growth. In maize, a MEI2-like RNA-binding protein that regulates the multiseriate cortical sclerenchyma (MCS) in roots has been found to be strongly associated with root penetrability in compacted soils [17]. In rice the histidine kinase 1 (HK1) gene was shown to modulate root circumnnavigation and growth of the root system into complex substrates [18]. Quantitative trait locus (QTL) analyses in rice and durum wheat have shown that root development in compacted substrates is associated with greater root diameter and bending stiffness in superior genotypes [19, 23].

Till now, the investigation of root growth in compacted substrates at seedling stage has been conducted using wax-petrolatum layers or high-density agar as proxy for compaction. Wax-petrolatum layers consist of thick discs made of wax and white petrolatum mixtures in different proportions, such as 20% and 80%, 60% and 40%, and 80% and 20%, corresponding to a low, medium and high mechanical impedance, determined using a soil penetrometer [24]. Beyond wax-petrolatum based methods, to mimic different levels of soil compaction, researchers have also used agar, adjusted to different concentrations. For example, a high-density agar gel layer (3%) has been used to create a partial barrier that simulated the hardpan (high-strength layer) typically found in compacted soils, thereby mimicking the resistance encountered by roots, and was able to reveal root altered response in rice [25].

Although these two methods have proven useful for studying the root growth ability, they suffer from certain limitations that can be overcome by adopting natural soil instead. Root penetration in compacted soil is a complex phenomenon involving not only the intrinsic and mechanical capacity of roots, but also the interaction between roots and soil. Consequently, root growth is the result of the balance between the root-soil adhesion strength (the force required for roots to detach from soil particles) and the root elongation strength (the ability of roots to grow and to extend through the soil). Compaction and drought conditions increase soil adhesion and limit the root elongation, leading to a reduced development of root system [20, 26]. Both high-density agar and wax layers make it impossible to assess the ability of plant roots to encompass the altered strength of the soil to grow.

Another major limitation is the inability to consider the interaction between plants and microbes, bypassing the dynamics of the microbiome and the changes in microbial communities that occur in compacted soil. The physical arrangement of roots can affect the rhizosphere influencing microbial recruitment by modifying the plant’s uptake of soil resources [27].

To overcome these limitations, the present study developed a new protocol based on reproducible compacted soil for phenotyping RSA-related traits has been developed and tested on a panel of 139 barley accessions. Ground natural silty clay soil was pulverised and mixed with sand and vermiculite to produce lumps with homogeneous compacted soil texture.

The new methodology was validated using a genome wide association study (GWAS) to detect two novel QTLs controlling root growth under compacted soil.

Methods

Preparation of soil lumps artificially compacted

The soil used in this work was classified as fine silty, mixed, mesic Udic Ustochrepts with a composition in texture classes of sand 14%, silt 50%, and clay 36% [28]. To eliminate the variability of physical characteristics of natural field soils, ten kilograms were collected in March from the experimental farm of the CREA Research Centre for Genomics and Bioinformatics in Fiorenzuola d’Arda (PC), Northern Italy (44°56’ N 9°54’ E, 80 m. a.s.l.), dried at 60 °C for 6 h, manually powdered using a pottery mortar. The finest part was collected using a sieve (Endecott Ltd, England, aperture 1.0 mm), Fig. 1(a). Dried sand and vermiculite were added to the powdered soil in a ratio of 1:1:0.2 (soil: sand: vermiculite, respectively) and manually mixed to create a homogeneous texture. Sand and vermiculite were added to the natural soil to facilitate the subsequent root washing step without excessive manipulation of the root system. Water was then added to this mixed dry soil in a ratio of 1:2.3 (mixed dry soil: water) to achieve a hydration of 43% (w/w). An aliquot of 300 g +/-0.3 of the resulting hydrated soil was placed in plastic baskets (7 cm x 5.4 cm x 4.5 cm, Anelli Company) and incubated at 60 °C (Votsch, Industrietechnik, VT3050) to reach the desired level of compaction through a dehydration step Fig. 1(b). The duration of the dehydration step determined the amount of water content into the lumps and indirectly the target compaction degree. The most suitable compaction degree for the evaluation of the RSA-related traits in barley seedlings was determined with two preliminary trials with three cultivars (Formula, Ketos and Nure). The final compacted lumps showed a dimension of 6.5 cm x 5 cm x 3.5 cm, Fig. 1(c).

Fig. 1.

Main steps of the new phenotyping method for root evaluation under compacted soil conditions. In a) the powdered soil was mixed with sand and vermiculite (1:1:0.2 ratio), in b) the mixed soil hydrated with 43% of water was aliquoted in plastic baskets and incubated at 60°C for compaction phase. In c) the final compacted lumps ready to be sowed with 3 barley seeds. In d) plants in compacted lumps at 5 Days After Germination (DAG), isolated with plastic film to avoid the soil dehydration. In e) the washing phase, where all lumps were inserted in the plastic baskets, placed in containers and then covered with water for 10’ to discard rapidly the soil from the roots. In f) three seedlings of each cultivar were sown in a single lump after a coarse washing. In g) Seedlings after a manual accurate washing, ready for root imaging

Preliminary trials for assessment of compaction degree: trial 1 (water content of 22%) and trial 2 (water content of 31%)

Particular attention was given to avoid drought stress in barley seedlings sowed in the compacted lumps to ensure that the phenotypic effects on root development were mainly due to soil compaction. For this reason, two trials with different water contents (22% and 31%) were carried out to identify the appropriate compaction degree of lumps, considering the 20% of water content as the minimum moisture to avoid drought stress (www.eos.com). The water content in lumps was calculated using the gravimetric method described in the following equation:

where represented the gravimetric water content, W_water the mass of water and the W_dry the mass of the dry soil.

Trial 1 was carried out using 15 compacted lumps with a final water content of 22% obtained after 5 h 30’ +/-15’ of dehydration at 60 °C, while trial 2 was carried out using 15 compacted lumps with a final water content of 31% achieved after 4 h 30’ +/-15’ of dehydration at 60 °C. Additional twenty-seven non-compacted lumps with a final water content of 43% were produced by dehydrating for 3 h +/-15’ at 60 °C and used as controls. After the dehydration step, all lumps were placed at room temperature for 1 h to cool down. Subsequently, three lumps of each trial and three uncompacted lumps were cut to measure the hardness of the soil in the internal part using a digital durometer (53215 TP–Turoni, Forlì, Italy) which return the hardness values on a Shore (sh) scale (10–100). The Shore scale considers the following ranges: soft (10–20sh), medium soft (21–35 sh), medium hard (36–70 sh), hard (71–90 sh), extra hard (91–100).

The remaining lumps from trial 1 and 2 as well as the uncompacted lumps were used to evaluate the root growth of three seeds of three barley varieties: Ketos, Nure and Formula. Three biological replicates (lumps) were sowed for each variety, each containing three seeds (technical replicates). Five days after sowing, three phenotypic parameters were considered to establish the appropriate degree of compaction: (i) the effective development of roots; (ii) the absence of visible symptoms of drought stress in the aerial part such as chlorosis and/or leaf wilting; (iii) changes in root morphology compared to the control, such as the presence of tortuosity, reduction in length, increase of diameter width.

Plant material, plant growth, root wash

A barley collection composed by 139 European winter 2-rows and 6-rows cultivars, part of the germplasm collection assembled in the framework of the Exbardiv project [29] and conserved at the CREA Research Centre for Genomics and Bioinformatics was used (Supplementary Table 1, Additional file 1) to validate the new phenotyping method described in this work. To manage the experiment, the entire collection was randomly divided in two subsets of 69 and 70 genotypes arbitrary named Exp A and Exp B, respectively. Three biological replicates were conducted for each subset (Exp 1–2 A-3 A, Exp 1B-2B-3B). Four genotypes (Canoro, Fanfare, Franka, Glenan) were considered as checks and added to each experiment. Seeds of each cultivar were calibrated (Endecotts Octagon 200 Sieve machine, aperture 2.50) and pregerminated in Petri dishes on wet (1 ml of Millipore water) filter paper for 6 days at 6 °C in the dark. At the stage of rupture of the caryopsis coat (chitting stage) three germinated seeds (technical replicates) of each cultivar were sown in each soil lump, which was wrapped with transparent film to keep the water content and the appropriate degree of compaction. All lumps were placed in a growth chamber under controlled conditions (photoperiod 16 h/8 h and temperature 23 °C/15°C) for plant growth.

Plants were removed from the soil and collected when the first root among all cultivars reached the base of the lump. This stage corresponded to 5 days after sowing (Fig. 1 (d). Each lump was repositioned in the plastic basket (the same previously used to create the lump), placed in a container, dipped in water for 10 min (Fig. 1(e)) and then roots were washed manually to totally remove the soil, Fig. 1 (f-g).

Root scan and phenotyping

The aerial part of the plantlets was separated from the seminal root system by cutting above and below the seed. Each seminal root was separated from the others to avoid their overlapping during the scanning (EPSON Expression 10000XL). Then images were analysed using the software WinRHIZO^® (Regent Instruments, Inc.) to measure the total root length and the average of root diameter. Finally, the aerial part of each plant and the corresponding seminal roots were collected in paper bag, dried at 70 °C for 6 h and weighted. The three plantlets grown in each individual lump were collected separately in paper bags and individually analysed.

Statistical models for computing the adjusted means of TRL, ADW, S/R traits

Raw phenotypic traits including total root length (TRL), average of diameter width (ADW), seminal root number (SRN), shoot/root weight ratio (S/R) were examined using the R 4.0.5 (R Core Team, 2021) statistical environment. The adjusted means (BLUEs) of each trait was calculated and used as phenotypic data for all further analysis and genomic investigations.

To compute the adjusted means of TRL, ADW and S/R across the experimental design the following linear mixed models were fitted:

where is the response variable, that is one of the root phenotypic root traits, is the effect of the i^th experiment, is the effect of the j^th replicate within the i^th experiment, while Gen_k is the random effect of the k^th genotype. In these models, the random effects of Gen_k follow a normal distribution with mean 0 and variance , and similarly, the residual terms e_ijk are normally distributed with mean 0 and variance equals to σ². All models were fitted using lme4 and the resulting variance components were used to compute broad sense heritability of all traits in according to the following formula:

where is genotypic variance component and is the error variance component and nRep is the number of replicates.

SNP markers information

The barley panel examined in this study was previously genotyped using the Illumina 9 K iSelect Chip, which identified 7,864 SNPs [30]. All monomorphic SNP markers and markers with a minor allele frequency (MAF) below 5% were removed from the data set and excluded from further analysis. Imputation with Beagle 5.1 [31] was used to reduce the number of missing data points and then those with more than 10% missing data were eliminated, resulting in a final dataset of 5,317 informative SNPs. The genomic coordinates of the SNP markers were projected over the reference sequence of barley Morex V.3 mapping probes of the Illumina 9 K iSelect Chip to barley pseudomolecule sequences [32].

Genetic diversity, population structure and linkage disequilibrium

To assess the phenotypic diversity of the barley collection with respect to the target root traits, Principal Component Analysis (PCA) was performed on the phenotypic data using prcomp R package version 4.0.5. The first and the second principal components were used to visualize the genetic clusters of the barley cultivars.

The population structure of the collection, based on SNP genotyping data, was determined by estimating their degree of admixtures and the most likely number of clusters (K) into which the accessions were grouped, using the program STRUCTURE 2.3.4 [33]. To determine the most probable number of clusters (K), 10 independent simulations were performed for each 10 Ks [1–10] with a burn-in period of 10,000 iterations and a Markov Chain Monte Carlo (MCMC) run length of 20,000 iterations. After the first round of analysis, the best Ks were selected for further simulations with a burn-in period of 50,000 iterations and a MCMC run length of 100,000 iterations. The resulting log-likelihood estimates for the Ks were used to find the most probable K using the delta K method [34]. The resulting population structure matrix (Q matrix) was then generated using the membership coefficients of 139 barley accessions based on the best K.

In addition, hierarchical clustering was performed using the R packages. The Nei’s nucleotide distance, which explains the allele frequency divergence between two sub-populations, was estimated between barley accessions [35] and the hierarchical clustering analysis was carried out with a neighbour joining (NJ) method.

Linkage Disequilibrium (LD) of SNP markers of each chromosome was determined as the correlation between marker-pairs calculated and expressed through the Pearson correlation coefficient (r²). LD decay was calculated using TASSEL 3.0 [36]. LD pairwise values were plotted as a function of physical distance using a custom R script developed by Remington et al. [37].

Genome wide association analysis and transcript investigation

Association analyses between markers and traits were performed running TASSEL 3.0 using Mixed Linear Models (MLM) with kinship matrix (K) to control for the effects of genotype relatedness. To define significant association, a threshold P-value was calculated considering Bonferroni correction and the false discovery rate (FDR) method [38]. The MLM stats were plotted in R using qqman package. Candidate genes for Marker-Trait Associations (MTAs) detected in this study were investigated using the annotation of barley reference genome v.3 (Barley cv. MorexV3 (2021) - GrainGenes (https://www.usda.gov); [39] To establish the interval confidence for QTL detection, the local LD decay into a genetic distance of 1 Mbp around the SNP was calculated.

Results

Preliminary trials for compaction degree identification

To establish the appropriate degree of soil compaction and to evaluate the root growth capacity, two preliminary trials were carried out with three barley varieties (Formula, Ketos and Nure) sown in compacted lumps having different degrees of compaction: 22% of water content (trial 1) and 31% of water content (trial 2), compared to the 43% of water content of the control condition (non-compacted). The firmness value of the lumps, measured with a digital durometer, ranged around 93.4+/-0.3 sh for the lumps with 22% of water content with a bulk density of 0.47 g/cm³ and around 88.8 +/-0.3 sh for those with 31% of water content and a bulk density of 0.44 g/cm³, in respect to 19.1+/-0.1 sh for the non-compacted lumps and a bulk density of soil of 0.29 g/cm³. As expected, the bulk density values are not in line with the typical evaluation, referring to artificially compacted soil.

The effects of soil compaction on seedling root growth were evaluated comparing the root morphology of plants grown in compacted lumps to those grown in non-compacted ones. In particular, the following parameters were considered in the evaluation: the presence of tortuosity, the reduction of length, the increase of diameter as well as other traits such as the effective root development and visible symptoms of drought stress in the aerial part (chlorosis and/or leaf wilting). Plants germinated in lumps with 22% water content did not respect the established parameters developing an altered root system after 5 days of germination (DAG) (Fig. 2). Therefore, this degree of compaction was considered excessive and not appropriate for the aims of the present study.

Fig. 2.

Results of trial 1 (compacted lumps with 22% of water content) at 5 DAG carried out using three barley varieties: Formula (a), Ketos (b) and Nure (c)

In contrast, plants germinated in lumps with 31% of water content showed a root system characterized by some morphological modifications already reported in the literature and associated to growth in compacted soil. All cultivars showed a reduction in total length (30.4%, 20.1% and 23.3% for Ketos, Nure and Formula, respectively), an increase in diameter width (29.7%, 34.1% and 24.4% for Ketos Nure and Formula, respectively), tortuosity and emergency of lateral roots as compensatory growth when the main axes were mechanically impeded (Fig. 3). In particular, growth conditions in compacted lumps revealed significant differences in TRL between plants grown in non-compacted and compacted soils for Formula and Ketos. In Nure cultivar, on the other hand, the differences were not statistically significant due to greater phenotypic variability between replications, which led to higher SD values compared to Formula and Ketos. ADW, as expected, showed a significant increase in plants grown in compacted substrate and a low variability among replicates in all genotypes.

Fig. 3.

Morphological changes (reduction of TRL, increase of ADW, tortuosity, lateral root emergency) in Formula, Ketos and Nure at 5 DAG during trial 2 in uncompacted (43% of water content) and compacted (31% of water content) lumps. a) and b) report the values of TRL and ADW in the three cultivars. The scans of roots grown for 5 days in non-compacted (c) and compacted soil (d) are shown

Root growth in compacted soil manifested an altered plagiogravitropism in all seminal roots during root emergence (Fig. 4). Typically, only the primary root exhibits orthogravitropism that is a vertical Gravitropic Set-point Angles (GSAs) [40]. In addition, a variable number of embryonic seminal roots develop with a shallower growth angle (plagiogravitropism). While the root growth in uncompacted soil (Fig. 4A) showed this pattern, plants grown in compacted substrate (Fig. 4B) exhibited an orthogravitropism in all seminal roots, suggesting an alteration in the balance between gravitropic signals and anti-gravitropic responses.

Fig. 4.

Seedlings of barley cv. Ketos grown in uncompacted lumps (a) and in compacted soil (31% water content) (b). The yellow arrow indicates the altered plagiogravitropism effect in all seminal roots which exhibit vertical Gravitropic Set-point Angles (GSAs) to grow under compacted soil

Phenotypic evaluation of RSA and broad sense heritability (H²) of traits

To test the effectiveness of the new method for phenotyping root growth in compacted soil, a GWAS experiment was performed with a panel of 139 barley varieties. A total of 1,251 seedlings and 7,051 seminal roots (corresponding to 3 biological and 3 technical replicates per genotype) were processed, measured for TRL, ADW, SRN and S/R and the broad sense heritability (H²) was calculated. A Shapiro-Wilk normality test was applied to assess the normality of the distribution of phenotypic data for all traits considered. All traits, except S/R, showed a normal distribution, therefore a logarithmic transformation of S/R data was calculated and used for the following analysis. The broad-sense heritability ranged from 10.5% for ADW to 84.5% for S/R (Table 1), indicating that environmental factors play a greater role in increasing root diameter, whereas genetic bases are the predominant source of variation for shoot: root ratio.

Table 1.

Traits measured using WinRHIZO software, their symbol, description, the distribution of values, the normality test results and the level of broad sense heritability (H²)

Trait Name	Symbol	Trait description	Distribution of values (min mean max)	Shapiro-Wilk normality test (p-value)	H2
Total Root Length	TRL	Cumulative length of all the roots (cm)	5.682 23.517 40.670	0.509	0.472
Average Diameter Width	ADW	Average of the diameter of all the roots (mm)	0.514 0.579 0.652	0.9952	0.105
Shoot: Root ratio	S/R	Shoot to root dry weight ratio	1.326 3.083 8.651	0.02282*	0.846
Seminal Root Number	SRN	Number of seminal roots for each plant	2 6 9	0.5183	0.448

*Logarithmic transformation of data was performed

Population diversity, population structure and linkage disequilibrium analysis

After stepwise filtering, a dataset of 5,317 informative SNPs was retained from the preliminary dataset of 7,864 SNPs. The markers were spread across the seven barley chromosomes ranging from 571 (chromosome 1 H) to 915 (chromosome 5 H). Principal Component Analysis (PCA) showed that PC1 and PC2 explained 76.4% of variability (PC1 52.5% and PC2 23.9%, Fig. 5a) with PC1 mainly associated to row-type, as reported in the Supplementary Table 2 (Additional file 2). In contrast, we did not identify any known germplasm variables associated with PC2. The population structure of the collection was assessed using the Bayesian clustering method implemented in STRUCTURE software following the admixtures analysis which resulted in 2 clusters (K), thereafter namely sub-Pop1 and sub-Pop2 (Fig. 5b). Sub-Pop1 mainly included 2-rows cultivars, while the sub-Pop2 cluster mainly grouped 6-rows accessions. The estimated proportion of the membership (Q) of individuals to each sub-population is illustrated in Fig. 5B and summarized in 57.9% of accessions assigned to sub-Pop1 (red colour) and 42.1% to sub-Pop2 (green colour), with a Net nucleotide distance between the two population of 0.12. The fixation index (F_st) calculated for the two clusters was 0.399 for Sub-Pop1 and 0.202 for sub-Pop2. The genetic divergence within the populations (expected heterozygosity, He) was 0.2576 for Sub-Pop1 and 0.3262 for sub-Pop2.

Fig. 5.

Results of population diversity analysis. a) PC analysis highlighting two clusters: the green cluster contains mainly 6-rows cultivars, while the red cluster is composed mostly of 2-rows genotypes. PC1 explains the main clustering (52.5%) compared to PC2 (23.9%). b) Stratification in K = 2 of the collection considering 5,317 SNPs. The colour bars point out the membership coefficients of each genotype to the two clusters. The red colour refers to sub-Pop1, while the green colour addresses sub-Pop2

LD was estimated by calculating the squared allele frequency correlation (r²) between all possible marker pairs for each of the seven chromosomes. The r² values obtained were then plotted against the physic distance (bp) across the whole genome. The LD critical distance was calculated at a threshold of r² = 0.2. LD decayed in a range between 1.5 Mb in chromosome 2H and 5.5 Mb in chromosome 3 H (Supplementary Fig. 1, Additional file 3).

Genome-wide association studies for RSA-related traits

A GWAS analysis was carried out using Mixed Linear Model (MLM). The Manhattan plot and QQ plots resulting from the analyses are shown in Fig. 6. Due to the limited number of genotypes and markers, a significant SNP association was only found for the TRL trait, while for ADW, S/R ratio, SRN traits any SNPs exceed the significance FDR threshold (-log₁₀(p) = 4.83). However, a few putative genomic regions could be considered for future investigations. In detail, three SNPs on 4 H located between 53.4 and 54.8 Mbp showed association with ADW with -log₁₀(p) = 3.70. Two SNPs on 5 H located at 30.4 Mbp were associated with the S/R ratio with -log₁₀(p) value of 3.85 and 3.58, while two other SNPs at 14.1 Mbp on 7 H showed a -log₁₀(p) value of 3.68. The GWAS for the SRN highlighted two SNPs on 3 H (-log₁₀(p) = 3.86 and -log₁₀(p) = 3.38) mapped at 591.2 and 592.4 Mbp.

Fig. 6.

Genome-wide association studies for the TRL. (a) Manhattan plot obtaining fitting MLM model for carrying out GWAS. Negative log₁₀-transformed P values from a genome-wide scan are plotted against a physical position on each of 7 chromosomes. The orange horizontal dashed line indicates the FDR genome-wide significance threshold (-log10(p) = 4.83), while the black line represents the Bonferroni threshold (-log10(p) = 5.03). Two SNPs on chromosomes 4 H and 5 H showed significant association with TRL trait. (b) Quantile-quantile plot of the MLM model, showing for two SNPs a clear deviation of the observed values from the expected ones, confirming the results highlighted in the Manhattan plot. c) Details of the significant SNPs for the TRL trait showing SNP ID, chromosome and position where the markers are physically located, significance association values, phenotypic variance explained

Considering the TRL trait, two SNP markers were found to be significantly associated with this trait, BOPA1_3432_290 on chromosome 4 H and BOPA1_272_944 on chromosome 5 H. Both markers exhibit P-values beyond the FDR threshold, but only the SNP mapping on chromosome 4 H exceeded the Bonferroni threshold. The corresponding QQ plot showed a perfect match between the observed and expected P-values for most SNPs, except for the two significant SNPs which showed a clear deviation of the observed values from the expected ones suggesting a realistic positive association between the SNPs and the trait. The proportion of phenotypic variance (R²) explained by the model was of 13.7% and 12.8% for BOPA1_3432_290 and BOPA1_272_944, respectively.

To identify the candidate genes underlying the significant SNPs, a confidence interval was calculated based on the local LD decay (at a genetic distance of 1 Mb around the SNP). The local LD around the significant SNP located on chromosome 4 H measures circa ~ 689.8 kb, while for the significant SNP located on chromosome 5 H a value of about ~ 821.1 kb was reported (Supplementary Fig. 2, Additional file 4). Searching for candidate genes within 690 kb up- and down-stream of the associated SNPs on 4 H, 51 annotated sequences were identified (19 and 32, low and high confidence respectively). Twelve sequences were classified as transposon/retrotransposon related, while the remaining genes [40] were involved in response to stimulus, signal transduction and cellular component organization (Supplementary Table 3, Additional file 5). Within 830 kb up- and down-stream of the associated SNPs on 5 H, 65 candidate genes were mapped (33 and 32, low and high confidence respectively). Fifty-two of the total annotated sequences accomplished molecular functions like anatomical structure, nucleotide binding, RNA binding, signal recognition particle, translational initiation, hydrolase activity, DNA binding. Thirteen sequences were classified as transposon/retrotransposon related.

Discussion

To date, phenotyping methods for the root system investigation are already described, but none of them simultaneously fulfils the following characteristics: high-throughput method, low cost, natural soil based, compaction degree, homogeneous texture and evaluation of root growth in seedlings.

Rhizotrons, for example, are largely used for root phenotyping in soil in a 2-D system but they are not suitable for compaction studies and the evaluation of RSA is partial (not the whole root system can be visualised).

Shovelomics is a high-throughput phenotyping method that allows the investigation of root system in open field conditions, mainly in adult plants. Although this method is realistic, it lacks some relevant elements for assessing root growth capacity in compacted soil. Firstly, it can only be used to measure crown root traits. Secondly, it is not suitable for evaluating the root system in seedlings and, lastly, root morphology is strongly influenced by environmental variability (soil texture and climatic conditions), adding complexity to root genomics studies.

For this reason, an innovative, cheap and cost-effective phenotyping method was developed to investigate the growth ability of seminal roots under compacted soil condition. Previously, this ability has been studied by sowing seeds in high density agar or using wax-petrolatum layers as proxy for compaction. These methods have some limitations as they ignore the root-soil interactions in compacted environments which play pivotal role for root/plant growth. Roots are in fact significantly influenced by the biological and physicochemical properties of the soil [26]. For example, the strenght required to penetrate the soil is due to the cohesive forces exerted between individual soil particles and the frictional resistance during the penetration of growing roots [41]. An increase in mechanical impedance, unbalance the equilibrium between the adhesive force of.

soil and the elongation force of the root, affecting root growth, rooting depth and root system morphology. Furthermore, high-density agar or wax-petrolatum layers do not reproduce the root-microbe interaction which is a relevant element in root adaptability to compacted substrates as described by Xu et al. [42].

For these reasons, the new approach for root growth studies developed in the present work and based on soil lumps artificially compacted looks to be more realistic and to contribute at filling a methodology gap. The innovative method proposed in this study revealed also a new phenomenon never described before, namely an altered response to gravitropic signals. During emergence seminal roots grew straight down, vertically, avoiding the growth angle determination and the usual spatial distribution of roots in the soil. In general, only the primary roots exhibit vertical GSAs, while a variable number of seminal roots develop with a shallower growth angle, a phenomenon known as plagiogravitropism [43]. No data are available about this phenomenon in seminal roots under compacted soil, but Rostamza et al. [44] investigating the effects of drought conditions on nodal roots in pearl millet and sorghum, observed a more vertical root growth pattern in dry soil. The authors suggested that the downward direction of growth was in response to gravity and other signals to tropism.

This phenomenon typically is observed in lateral roots, which, usually, do not grow straight down like primary roots (orthogravitropism] but develop at different angles, suggesting some mechanisms behind plagiogravitropism that involves a balance between gravitropic signals and anti-gravitropic responses. In Arabidopsis, auxin has been implicated in this process by regulating cell elongation and division in response to gravity [45].

While under dry soil pearl millet sorghum roots kept gradually the vertical GSA for at least 5 centimetres, barley roots grown in compacted soil restored the plagiogravitropism after a few millimetres after emergence. When the seminal roots reached a length of 4–5 mm, they were able to drastically restore a shallower growth angle, as shown in our results (Fig. 4). At this stage, no hypothesis can be formulated without developing further experimental studies.

To validate the new phenotyping methodology for root genomics approaches a case study was developed using a panel of 139 barley cultivars to identify favourable alleles associated with root growth ability in compacted soil. Four traits were considered for the GWAS, although only for TRL two independent significantly associated SNPs located on chromosomes 4 H and 5 H were detected.

Very few QTLs and genes associated with root penetration in compacted soil have been described. A single work has been published about root penetration during seedling germination in Arabidopsis [22], while several QTLs associated with root growth in compacted soil have been discovered in adult plants or in seedling using hard agar layers [21]. Ray et al. [45] reported two QTLs after testing the root penetration of 202 recombinant inbred lines in rice grown in wax-petrolatum layers. In 2007 Kubo et al. [19] found a QTL on chromosome 6 A associated with the number of roots penetrating paraffin-vaseline discs in durum wheat. In 2014, Botwright Acuñaet al. [46] used matrix columns containing wax layers to evaluate root and shoot traits in a wheat doubled haploid population and identified 29 QTLs associated with root traits, most of them with a small phenotypic effect. Bello-Bello et al. [21] demonstrated that auxin dynamics have a critical role in facilitating root penetration into hard agar layers in seedlings, and mapped a large-effect QTL, ROOT PENETRATION INDEX 3, associated with the primary root penetrability in Arabidopsis thaliana, without identifying any candidate genes. Several papers describe the involvement of auxin, ethylene and abscisic acid (ABA] in root growth and development in compacted soil. Huang et al. [47] explained how the auxin accumulation in epidermal cells restricted their expansion and the root growth in compacted soil conditions. Pandey et al. [48] discovered the role of the hormone ethylene in sensing soil compaction by the roots. The ethylene accumulation around the root tips caused by the compaction, inhibits root elongation and promotes radial expansion. Qin et al. [49] demonstrated that also ABA, through an auxin biosynthesis-mediated process, also causes suppression of root elongation and promotes root swelling in response to compacted soil.

In this work, we detected two QTLs (Chr. 4 H, marker BOPA1_3432 − 290, and 5 H marker BOPA1_272–944) associated with the TRL explaining 13.7% and 12.8% of the phenotypic variance (R²), respectively. These values are seemingly low, but not surprising given the modest heritability of the trait (47.2%) and the assumption that this trait is controlled by many genes with small/minor effects.

The confidence interval of 1.38 Mb around the SNP BOPA1_3432 − 290 reported 51 annotated genes (Supplementary Table 3, Additional file 5), two out of them potentially related to the observed phenotype. A first candidate is HORVU.MOREX.r3.4HG0334860 which encodes a Nitrate Transporter1 (NRT1)/Peptide Transporter (PTR) family protein 2.2 located at 457.55 Kb upstream the significant SNP. Besides the well-known involvement of this family proteins in nitrogen uptake, Mondal et al. [50] demonstrated the ability of compacted soil to modify the expression of genes encoding N-assimilation enzymes in wheat plants at the active tillering stage.

The second candidate, HORVU.MOREX.r3.4HG03351609, is located at 657.55 kb downstream BOPA1_3432 − 290, encodes a Hedgehog-interacting-like protein (HIPL) and has previously been described to be associated with seed vigour enhancement in rice [51]. In rice again, Xu et al. [52] detected a QTL cluster for seed vigour with a significant effect (p < 0.01) on root length, suggesting a correlation between the two traits.

The only association between the higher seedling vigour and a greater root penetration into compacted soil layers was reported in wheat by Albert et al. [53]. Studying embryo orientation during seed germination, the authors found that seeds with embryo-down orientation exhibited a great seed vigour which caused a stronger root penetration ability in compacted soil.

On chromosome 5 H, the confidence interval (1.65Mbp) around the significant SNP BOPA1_272–944, underlined 65 annotated sequences (Supplementary Table 3, Additional file 5). Among them, HORVU.MOREX.r3.5HG0508130, located 540 kB downstream of the significant SNP and encoding a Cyclic nucleotide-gated (CNG) channel. Close to their activation during plant development, stress response and immunity [54, 55], CNG channels have also been implicated in the regulation of root hair growth in Arabidopsis. In compacted soil, rice plants developed longer root hairs, induced by elevated auxin response upon interaction with compacted substrate, which enhanced the root ability to penetrate harder soil layers [25]. The authors hypothesised that the elongation of root hairs might increase the surface area of root-soil contact, providing the required anchoring force to support root penetration into compacted layers.

At 742.40 kb from SNP BOPA1_272–944 a second candidate gene associated in literature with RSA, was HORVU.MOREX.r3.5HG0508200, which encodes an expansin protein. Expansins (EXPAs) are a family of proteins that have been implicated in cell wall relaxation and expansion [56]. In Arabidopsis, soybean and bread wheat, EXPAs were found highly expressed in primary roots [57–59]. In Arabidopsis seedlings, the expression of AtEXP7 and AtEXP18 was associated with root hair formation and elongation via ethylene perception. Blocking ethylene perception by gain-of-function mutations of the ethylene receptors markedly inhibited root hair elongation. Cho and Cosgrove [60] proposed that the signals from the developmental and environmental/hormonal pathways converge at or before the transcriptional regulators that control the hair cell specificity of the expansin genes.

In barley, Liu et al. [61] identified 46 expansins distributed unevenly on the seven chromosomes using an RNA-SEQ approach under drought stress. In the distal region of the long arm of chromosome 5 H, the authors mapped 7 expansins (HvEXPB10, HvEXPA15, HvEXPA16, HvEXPA17, HvEXPA18, HvEXPA19, HvEXPA20) on the Morex v.1 genome which were differentially expressed under drought. These seven genes were located between 521Mbp and 578Mbp on the Morex v.3 genome. The candidate expansin HORVU.MOREX.r3.5HG0508200 identified in this work and located at 521Mbp was found to correspond to HvEXPB10 (HORVU5Hr1G092910.9, mapped on chr5H at 589Mbp on Morex v.1], which was reported to be down regulated under drought stress conditions.

Studies have shown that expansins are associated with cell wall modifications of roots in response to stress conditions and via ethylene induction, playing a role in plant adaptation and plasticity [62, 63]. Nothing is known about the involvement of expansins in root growth in compacted soil, but their role could be considered with a regard to the rearrangement of root morphology in response to compaction and to induction via ethylene perception. In compacted soil it is well described how ethylene accumulation around the root tips stops the root growth in plants, but the underlined molecular mechanism is not known. So far, it is not clear how ethylene accumulation can stop or reduce root growth, but it is well known that expansin proteins are involved in cell wall modifications via ethylene induction during plant growth and stress conditions [61]. Considering our results, the root growth ability in compacted soil could be orchestrate by four independent biological/biochemical processes: the nitrogen uptake and the mobilization of structurally compounds in plants which involved the NRT1/PTR family proteins; the seed vigour controlled by HIPL protein; the elongation of root hairs regulated by the CNG channel; the cell wall modification in response to ethylene accumulation due to the involvement of expansin proteins.

Our two significant SNPs explain only the 26.5% of the phenotypic variance, suggesting the need of further investigation to obtain an exhaustive and complete description of all the genetic mechanisms underlying the tolerance to this adaptive phenomenon. The two loci here identified are promising candidates for breeding programmes in response to the compaction tolerance in the earlier developmental stages in barley.

Conclusions

In the present study, we introduced and validated a cheap and cost-effective phenotyping method to examine the growth ability of seminal roots under compacted soil. This method takes into consideration the root-soil interaction in compacted soil, underline for example an altered response to gravitropic signals in seminal roots, never described before. To our knowledge, this is the first study exploiting soil lumps artificially compacted to assess the development of root traits. The validation of this method using a panel of barley cultivars demonstrated the existence of variability for root penetration in compacted soil and allowed two markers statistically associated to the ability of root growth in compacted soil. Our methodology along with significant markers detected using GWAS might be of potential interest to sustain barley breeding programmes to improve belowground traits in the context of climate change. Finally, the functional characterization of candidate genes identified within the confidence interval of significant SNPs paves the way to dissect their possible role in the regulation of root plasticity.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

GWAS

Genome Wide Association Studies

SNP

Single Nucleotide Polymorphism

QTLs

Quantitative Trait Loci

RSA

Root System Architecture

PIF3

PHYTOCHROME-INTERACTING FACTOR 3

FER

FERONIA

MCS

Multiseriate Cortical Sclerenchyma

HK1

Histidine Kinase 1

TRL

Total Root Length

ADW

Average of diameter width

SRN

Seminal Root Number

S/R

Shoot/Root weight ratio

MAF

Minor Allele Frequency

PCA

Principal Component Analysis

K

Number of clusters

MCMC

Markov Chain Monte Carlo

NJ

Neighbour Joining

LD

Linkage Disequilibrium

MLM

Mixed Linear Models

K

Kinship matrix

FDR

False Discovery Rate

MTAs

Marker-Trait Associations

DAG

Days After Germination

SD

Standard Deviation

GSAs

Gravitropic Set-point Angles

H²

Broad sense heritability

PCA

Principal Component Analysis

ABA

ABscisic Acid

PTR

Nitrate Transporter1 (NRT1)/Peptide Transporter

EXPB

Expansins EXPAs

Author contributions

The study was designed by GC and LC. Experimental work and data analyses were performed by GC and AF. The first draft of the manuscript was written by GC, AF and EM. All authors revised all versions of the manuscript. All authors have read and approved the final manuscript.

Funding

This research was not funded.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.