Introduction

Aluminium is a common metal in the Earth’s crust that is usually found in insoluble forms when the soil pH is neutral or slightly acidic. However, when the soil pH drops below 5.0, the solubility of aluminium increases and toxic aluminium ions (especially the trivalent cation Al3+) can limit root growth and negatively interfere with water and nutrient uptake, consequently leading to reduced crop yields due to drought and mineral deficiencies (Kochian et al. 2015). Plant species have developed different mechanisms to cope with Al3+ toxicity. The most studied mechanism is the exudation of organic acids from the root apex. Once in the rhizosphere, the organic acids chelate Al3+, thereby reducing its toxicity (Ryan et al. 2001).

Although increments in organic acid production by plant cells have been associated with increased Al3+ resistance in plants (de la Fuente et al. 1997; Barone et al. 2008), organic acid transport out of the root cells has a greater association with Al3+ resistance (Sasaki et al. 2004, 2006; Raman et al. 2008; Pereira et al. 2010; Zhou et al. 2013, 2014). In wheat (Triticum aestivum L.), two genes (TaALMT1 and TaMATE1B) that encode organic acid transporters were demonstrated to play a role in Al3+ resistance. TaALMT1 is located in chromosome 4D and encodes a malate transporter belonging to the ALMT (aluminium-activated malate transporter) family that is expressed in the wheat root tip (Sasaki et al. 2004). TaMATE1B is located on chromosome 4B and encodes a citrate transporter of the MATE (multi-drug and toxin extrusion) family that is expressed in the root tip depending on the allele (Ryan et al. 2009; Tovkach et al. 2013). Both genes exhibit a correlation between the level of expression, the amount of organic acid exudate by the roots and the level of Al3+ resistance. The gene expression level is regulated by sequences in the promoter region (Sasaki et al. 2004, 2006; Ryan et al. 2009; Tovkach et al. 2013). For instance, the level of TaALMT1 expression is correlated with tandem repeats in the promoter (named here I to VII). Promoters containing a higher number of repeats (V and VI) have higher levels of gene expression and are usually Al3+ resistant, whereas promoters without repeats (I and II) have low levels of gene expression and are typically Al3+ sensitive (Sasaki et al. 2006; Raman et al. 2008). The TaMATE1B gene expression level is correlated with two polymorphisms in the promoter region (Tovkach et al. 2013). The first polymorphism is a single nucleotide polymorphism (SNP) that is responsible for a two-fold increase in gene expression and the second is a transposon-like insertion that is associated with a 20-fold increase in TaMATE1B expression. In fact, the transposon-like insertion not only increases but also changes the location of the expression to the root apex, thereby increasing the citrate efflux and consequently Al3+ resistance (Tovkach et al. 2013).

The repeats located in the TaALMT1 promoter and the insertion in the TaMATE1B promoter can be easily detected by PCR. The allelic variations in these promoter regions have been studied in a number of genotypes and have revealed interesting aspects of Al3+ resistance in wheat (Sasaki et al. 2006; Raman et al. 2008; Han et al. 2013; Garcia-Oliveira et al. 2014; Pereira et al. 2015). For instance, wheat lines of Japanese origin do not show a clear correlation between the number of repeats in the TaALMT1 promoter region and Al3+ resistance; this finding is in contrast to the effects observed with lines of non-Japanese origin (Sasaki et al. 2006). An Al3+-tolerant Brazilian cultivar named Trintecinco was shown to possess TaALMT1 allele II (usually associated with sensitivity) (Raman et al. 2008), whereas the majority of the Brazilian wheat cultivars contains promoters V and VI (typically associated with Al3+ resistance) (Pereira et al. 2015). TaALMT1 allele V was less frequently identified in a collection of cultivars and landraces of Chinese wheat, whereas promoter VI was not detected and allele III was identified only in Chinese and Brazilian wheat to date (Han et al. 2013; Pereira et al. 2015). TaALMT1 promoter VII is rarely found; indeed, it has been detected in only three genotypes to date (Raman et al. 2008; Pereira et al. 2015). Moreover, the presence of the transposon-like insertion in the TaMATE1B promoter is less frequent in Brazilian wheat genotypes but is widespread among Portuguese wheat cultivars (Garcia-Oliveira et al. 2014; Pereira et al. 2015). This insertion appears to be important for the root growth of some genotypes analysed in short-term soil experiments. However, most genotypes presenting the insertion in the TaMATE1B promoter did not outperform the genotypes without the insertion that possessed the same TaALMT1 allele (Pereira et al. 2015). Therefore, evaluating the allelic distribution of the TaALMT1 and TaMATE1B promoters is crucial to increase our knowledge concerning the contribution of the different alleles to wheat Al3+ resistance. Moreover, the TaALMT1 and TaMATE1B alleles have been associated with only Al3+ resistance in hydroponics or in short-term soil experiments to date. No correlation has been described between these alleles and wheat plants growing in acidic soils under field conditions.

Phenotyping using hydroponics is widely used to evaluate a number of traits including Al3+ resistance because this method is fast, non-destructive and allows greater control over the environmental conditions (Shavrukov et al. 2012). Indeed, phenotyping for Al3+ resistance in Brazilian wheat has typically been performed under hydroponic conditions (Camargo et al. 1987, 1998, 1999, 2006). Studies assessing Al3+ resistance under field conditions for a large number of Brazilian wheat genotypes are rare (de Sousa 1998) even though field evaluation in acidic soils is more realistic. The downside is that soil evaluation presents issues related to non-uniform nutrient distribution, uncontrollable factors (i.e., drought or excess heat) and Al3+ toxicity occurring together with calcium deficiency and manganese toxicity. A high correlation for Al3+ resistance phenotyping between field and hydroponic conditions has been reported for wheat (Baier et al. 1995); however, the genotypes evaluated under the field and hydroponic conditions differed in that study. To obtain a clear correlation, the same genotypes need to be used for the comparison of Al3+ resistance phenotyping between different methods.

In Brazil, acidic soils are predominant in wheat-growing areas (Echart and Cavalli-Molina 2001), and Brazilian wheat genotypes have been considered good sources of Al3+ resistance. Some genotypes have been used in different breeding programmes worldwide to develop new lineages, including BH 1146, Carazinho, Cotiporã, Frontana, IAC 5—Maringá and Trintecinco (Hettel 1989). BH 1146, Frontana, IAC 5—Maringá, and Trintecinco are among the most common genotypes used by Brazilian breeders (de Sousa and Caierão 2014). Moreover, a number of the Brazilian cultivars exhibiting Al3+ tolerance have PG1 as a common ancestor. PG1 is a selection of Polyssú that is one of the parents of Fronteira, which is also used as a good source of Al3+ resistance (de Sousa 1998). The constant use of the same or similar sources in breeding programmes can result in a lower genetic basis for Al3+ resistance. Phylogenetic analysis using microsatellite markers revealed similarities among some of the Brazilian cultivars used as Al3+ resistance sources, and these genotypes were most likely to share similar resistance/tolerance mechanisms (Pereira et al. 2015). Thus, the phenotyping of a large collection of wheat materials can result in the identification of sources containing potential new alleles/mechanisms that could be useful for breeders.

In this context, the goal of this work was to correlate the TaALMT1 and TaMATE1B allelic variability with wheat resistance to toxic aluminium under hydroponic and field conditions. We used a diverse wheat collection representing different countries and more than 90 years of breeding. Our goal also extended to the identification of resistant sources not commonly used by breeders, the comparison of hydroponic and field data and the evaluation of the TaALMT1 and TaMATE1B alleles over time to associate their frequency with Al3+ resistance over the past nine decades.

Materials and methods

Plant materials

A total of 338 wheat accessions were used in this study, including commercial cultivars, landraces, synthetic hexaploid wheats and varieties from the Embrapa Trigo (Brazilian Agricultural Research Corporation in Passo Fundo, Rio Grande do Sul, Brazil) breeding program; these accessions represented 40 countries belonging to five continents and are presented in Table 1 and Table S1. All analyses, including the molecular and phenotypic evaluations, were performed after accession purification and multiplication of one selected seed. The Al3+ resistance evaluation was performed in nutrient solution and in the field using cultivar IAC 5-Maringá as the resistant control and Anahuac 75 as the sensitive control.

Table 1 Geographical origin of the 338 wheat genotypes evaluated in this work
Full size table

Evaluation of aluminium resistance in nutrient solution

Al3+ resistance was evaluated by measuring the relative root growth (RRG) after 4 days. Wheat seeds were surface-sterilized with a 0.2 % NaClO solution for 5 min and incubated in the dark at 23 °C for 48 h. Twelve germinated seeds with roots that were approximately the same size (5–10 mm) and had similar viability were transferred to an aerated nutrient solution (pH 4.0) consisting of 0.40 mM CaCl2·2H2O, 0.65 mM KNO3, 0.25 mM MgCl2·6H2O, 0.04 mM NH4NO3, and 0.01 mM (NH4)2SO4 and grown for 4 days at 23 °C under a 12/12 h light/dark cycle in a growth cabinet. Two treatments were used: control (without AlCl3·6H2O) and stressed (74 µM Al added in the form of AlCl3·6H2O) as described by Voss et al. (2006). The nutrient solution was changed every 24 h to minimize changes in the pH and Al3+ concentration. The length of the main root was measured at the end of the experiment. The RRG for each genotype was estimated to be 100× (RLT/RLC), where RLT represent the root length under the Al3+ treatment and RLC was the root length in the control solution. Errors associated with deriving the RRG were calculated as SERRG = RRG [(SEx/x)2 + (SEy/y)2]1/2, where x represented the mean RLC and y was the mean RLT. For all experiments, the RRG of each genotype was compared to the RRG of the resistant and sensitive controls to determine the classification index of aluminium toxicity in hydroponics (CIATH); this step allowed the genotypes to be separated into four classes: sensitive (S); moderately sensitive (MS); moderately resistant (MR); and resistant (R). The limits for each class were calculated as follows: limit inferior for R and superior for MR [RRG resistant control—(0.5 × ((RRG resistant control—RRG sensitive control)/3))]; limit inferior for MR and superior for MS [RRG resistant control—(1.5 × ((RRG resistant control—RRG sensitive control)/3))]; and limit inferior for MS and superior for S [RRG sensitive control + (0.5 × ((RRG resistant control—RRG sensitive control)/3))].

Evaluation of aluminium resistance in the field

The reaction of the genotypes to Al3+ toxicity under field conditions was evaluated over a 2 year period (2012 and 2013) at Embrapa Trigo’s experimental field (latitude 28o15′S, longitude 52o24′W). The soil pH of the test areas, measured in water, ranged from 4.3 to 4.7 and the soil contained 29 to 54 mmol/dm3 Al3+ during the 2 years of the test. The soil was classified as Haplorthox and contained ~45 % clay. The chemical soil analysis showed the following nutrient contents: P, 17.4 mg/dm3; K, 118.5 mg/dm3; Ca, 5.98 mmol/dm3; Mg, 2.7 mmol/dm3; and organic matter, 25.25 g/dm3. The percentage of saturation of CTC-Al was 75 to 87.9 %. The genotypes were planted in a randomized complete block design with plots consisting of six rows that were 3 m in length with three replications containing five different material lines and one line formed by one of the controls placed alternately in every plot. The plant response was evaluated visually at different growth stages (tillering, silking and maturation). The evaluation method was based on de Sousa (1998) with some modifications. The field scores were as follows: 0.5, outstanding resistance; 1, resistant: good vegetative development, normal and abundant tillers, and normal spikes; 2, moderately resistant: normal plant, slightly less vigorous and fewer tillers, and slightly impaired development compared with the previous group; 3, moderately susceptible: intermediate or deficient plant growth and normal spikes; 4, susceptible: deficient plant growth, no tillering and small spikes; and 5, highly susceptible: dead plant or very deficient development and no spikes. We estimated the overall average (AVE) based on the overall average of the three observations over the 2 year period; the AVE was used to determine the classification index of aluminium toxicity in the field (CIATF) as follows: 0.05–1.25, resistant (R); 1.26–2.5, moderately resistant (MR); 2.51–3.75, moderately susceptible (MS) and 3.76–5.00 susceptible (S) as proposed by de Sousa (1998).

DNA extraction and TaALMT1 and TaMATE1B allelic variability

Leaves were collected from all genotypes (~150 mg), transferred to a 2 ml Eppendorf tube containing three stainless steel beads (2.3 mm diameter), immediately frozen in liquid nitrogen and triturated in a Mini-Beadbeater™ (Biospec Products) for 2 min. Total DNA was extracted using a CTAB-based protocol (Doyle and Doyle 1987). After quantification in an agarose gel, 100–150 ng of DNA was used for the PCR. The TaALMT1 promoter alleles were detected using the primers LPF-F (CCTGGTTTTCTTGATGGGGGCACA) and LPF-R (TGCCCACCATCTCGCCGTCGCTCTCTCT) as described by Sasaki et al. (2006). Another amplification with the primers SPF-F (GCTCCTACCACTATGGTTGCG) and SPF-R (CCAGGCCGACTTTGAGCGAG) was performed in cases where allele I or II was identified. To detect the TaMATE1B alleles, the primers TaMATE1-4B-SLT-F (ATCCATCCTCCTTCCCTCAC) and TaMATE1-4B-SLT-R (ATGAATGCTGTGTCCACCAA) were used according to Garcia-Oliveira et al. (2014); the presence of the PCR fragment was associated with the presence of the transposon insertion in the TaMATE1B promoter region. All amplification reactions were performed in a final volume of 20 µl containing 1 × PCR buffer, 0.30 mM of each dNTP, 1 × Q solution, 0.3 µM of each primer, 1 U Taq DNA polymerase (Qiagen) and 100 ng of DNA. The amplification programs were performed as described by Sasaki et al. (2006) and Garcia-Oliveira et al. (2014). The amplicons were separated by electrophoresis and visualized in a 1 % agarose gel (LPF primers), 1.25 % agarose gel (SPF primers) or 1.5 % agarose gel (TaMATE1B). The gels were stained with ethidium bromide.

Statistical analyses

The data obtained from the different growth stages (tillering, silking and maturation) and the overall average (AVE) were subjected to analysis of variance (ANOVA). The means were compared by the Scott and Knott (P < 0.01) test. Data averages from tillering, silking, maturation, AVE, CIATF, RRG, CIATH, the different TaALMT1 and TaMATE1B alleles and the combinations of alleles (haplotypes) were employed in the correlation analysis and principal components analysis (PCA) to estimate the relative contribution of characteristics to the diversity according to the method of Singh (1981) using the GENES program (Cruz 2006) and SAS package (SAS Institute 1997).

Results

Al3+ resistance in nutrient solution

We used hydroponics to evaluate the RRG of the genotypes. The duration (4 days) and stress level (74 µM AlCl3) of the treatment reduced the root length in some genotypes but did not substantially reduce the root length in others. On average, the RRG was 52.18 % with a range from 23.3 % in Ocepar 7—Batuira to 95.3 % in Fundacep 30 (see the frequency distribution in Fig. 1). The genotypes used as the tolerant and sensitive controls were evaluated in each experiment to allow the comparison of the RRG between them and the tested plants. In this way, the RRG of each genotype was compared to the RRG of the resistant and sensitive controls to determine the classification index of aluminium toxicity in hydroponics (CIATH). This index allowed the separation of the genotypes into four categories: 106 resistant (R), 64 moderately resistant (MR), 84 moderately susceptible (MS) and 84 susceptible (S) (Table 2). The genotypes classified as R included 91 materials from Brazil representing 85.8 % of the resistant materials and 15 materials (14.2 %) from different origins. Among the MR genotypes, 47 (73.4 %) were from Brazil and 17 (26.6 %) were from different continents. For the MS genotypes, 49 (58.3 %) were Brazilian and 35 (41.7 %) had different origins. Among the S genotypes, 26 (31.0 %) were from Brazil and 58 (69.0 %) had different origins (Table 2).

Fig. 1
figure 1

Frequency distribution of grades assigned to the tillering, silking, maturation, general average (AVE), classification index of aluminium toxicity in field (CIATF), relative root growth (RRG), classification index of aluminium toxicity in hydroponics (CIATH), and haplotypes (based on the combination of the TaALMT1 and TaMATE1B alleles). Arrows indicate the two materials used as controls: CS indicates the susceptible control (Anahuac 75) and CR indicates the resistant control (IAC 5—Maringá)

Full size image
Table 2 Distribution of the 338 wheat genotypes accordingly to the geographical origin the Al3+ resistance in field (CIATF) and hydroponics (CIATH)
Full size table

Al3+ resistance in acidic soil

Wheat plants were evaluated for two successive years in acidic soil. The evaluation was based on visual characteristics in three different growth stages (tillering, silking and maturation). A highly significant difference (P < 0.001, ANOVA) was obtained for the genotypes, which demonstrated the high rate of diversity in the responses among the genotypes as observed in the groups for the Scott-Knott test (P < 0.01) (Supplementary file 1). The two wheat cultivars used as the controls showed consistency in their reactions (see the frequency distribution in Fig. 1). The averages of the grades were 2.97 [ranging from 0.82 (IAC-21—Iguaçú) to 5.41 (Wadhanak)] for tillering, 3.02 [ranging from 0.75 (Toropi) to 5.25 (CIGM 93.395)] for silking, 2.88 [ranging from 0.57 (Minuano 82) to 5.02 (Emb_07105)] for maturation, and 2.96 [ranging from 0.78 (Toropi) to 4.89 (CIGM 92.1712)] for the AVE (Supplementary file 1). The high Pearson correlation coefficient (r > 0.86; P < 0.001) obtained between these traits indicated that the Al3+ resistance evaluation could be practiced at any of these stages (Table 3).

Table 3 Pearson correlation obtained by wheat traits evaluated in 338 wheat genotypes cultivated in field and hydroponics
Full size table

Using the overall average of the field evaluation (AVE), a classification index for aluminium toxicity in the field (CIATF) was obtained to allow the separation of the genotypes into four classes. In this index, scores closer to zero represented genotypes with higher Al3+ resistance. The CIATF showed a high correlation (r > −0.88; P < 0.001) with tillering, silking, maturation and the AVE (Table 3). Only eight genotypes were classified as resistant (Trintecinco, Toropi, PG1, BH 1146, Aceguá, C33, IAC-21—Iguaçú and PAT 7219), all of which were from Brazil; with the exception of Trintecinco and Aceguá, all cultivars shared Polyssú as their common ancestor. Among these genotypes, the AVE ranged from 0.78 to 1.25 with an average of 1.06. The numbers of moderately resistant, moderately susceptible and susceptible genotypes were much higher than the numbers observed in the hydroponic experiments, with 122 genotypes classified as MR, 117 as MS and 91 as S (Table 2). The materials classified as MR included 114 genotypes from Brazil (93.4 % of the MR materials) and only eight (6.6 %) with different origins, including one for each of the following countries: Japan (Aburakomugi, grade 2.0), Israel (Bet Dagan 131, grade 1.7), USA (Escondido 41, grade 1.8), Egypt (Giza, grade 1.3), Argentina (Klein Lucero, grade 1.3), Ukraine (Mironovskaya Yubileinaya 50, grade 1.9), Guatemala (Maya 74, grade 1.7) and Turkey (Menceki, grade 1.5). The MS genotypes contained 82 (70.0 %) Brazilian materials and 35 (30.0 %) materials with different origins; Mexico and Japan were the most represented countries with 10 and four genotypes, respectively. The materials categorized as S included nine (9.9 %) Brazilian genotypes and 82 (90.1 %) materials from different origins; Mexico was the most represented origin with 35 genotypes (31 were synthetic wheat). The Brazilian genotypes represented 63.0 % of the genotypes evaluated in this study and corresponded to 93.8 % of the total number of R/MR genotypes from the field evaluation, indicating the importance and superiority of these resistance sources for breeding programmes (Table 2).

The distribution obtained in the different CIATF classes showed that the soil classification exerted a greater selection pressure for Al3+ resistance because the number of R genotypes in the field (eight) was lower than the 106 R genotypes obtained from the hydroponic experiments. Indeed, all R genotypes from the field test were also classified as R in the hydroponic experiments; the other genotypes classified as R in the hydroponic experiments were distributed in the four classes of the field response (Supplementary file 2A and 2B). Similarly, 31 among the 122 materials classified as MR in the field maintained their classification in the hydroponic experiments, whereas 65 genotypes were classified as R and 26 as MS or S. This variation in the classification between the field (CIATF) and hydroponic (CIATH) tests resulted in a moderate Pearson correlation of 0.56 (P < 0.001) (Table 3). Among the 26 materials classified as MS or S in the hydroponic experiments but as MR in the field, the scores in the first growth stage (tillering) were usually higher (indicating more susceptibility) than those in the following stages (silking and maturation), indicating that some genotypes possessed the ability to recover from the initial stress in acidic soil with a high Al+3 concentration. This finding indicates that these genotypes recovered from Al3+ stress and may be one reason why the correlation between the performances in hydroponics and the field was not higher.

TaALMT1 and TaMATE1B allelic variability

PCR markers that can be used to discriminate between different alleles at the promoter region of the TaALMT1 and TaMATE1B genes have been previously reported by others (Sasaki et al. 2006; Raman et al. 2008; Han et al. 2013; Garcia-Oliveira et al. 2014; Pereira et al. 2015). These markers were used here to evaluate the allelic distribution among the genotypes (Table 4). Six TaALMT1 alleles (here named I to VI) were detected as follows: V in 183 (54.1 %) genotypes, followed by allele I in 78 (23.1 %), II in 25 (7.4 %), VI in 24 (7.1 %), IV in 23 (6.8 %), and III in 5 (1.5 %). Only allele VII was not found. The presence of the insertion in the TaMATE1B promoter [here named allele TaMATE1(+)] was observed in 37 genotypes (Table 4). Thus, the allele TaMATE1B(−) did not present the insertion in 89.1 % of the materials.

Table 4 Distribution of haplotypes formed by the combination of six alleles in the TaAMLT1 promoter and two alleles in the TaMATE1B promoter in 338 wheat genotypes with different origins
Full size table

The distribution of the TaALMT1 alleles in the genotypes from different geographic origins is shown in Table 4. Six TaALMT1 promoters were found out of the 288 genotypes from the Americas, with allele Type V being found most frequently (present in 58.3 % of the genotypes), followed by Type I (present in 20.8 % of the genotypes) and less frequently Type III (present in two genotype). In the eleven genotypes that originated from Africa and the Middle East, promoters Type III and Type VI were not found and the most frequent promoters were Type I and Type V. In the thirteen European genotypes, only three promoters were found (Type I, IV and V), with Type V the most common. In the 26 Asian and Oceania genotypes, all six TaALMT1 promoters were found, with Type I the most common and Type IV and Type VI present in only one genotype each. TaMATE1B(+) was not found in any of the genotypes that originated from Asia and Oceania. This allele was found in only one genotype from the Middle East and Europe and in two genotypes from the Americas (not including Brazil). Most of the TaMATE1B(+) alleles were detected in Brazilian genotypes, which represented 89.2 % of the genotypes that contained that allele.

When we investigated the different combinations of the TaALMT1 and TaMATE1B alleles, 11 haplotypes were detected among the 338 genotypes from different origins (Table 4). When Brazil was excluded, genotypes showing the haplotype I(−) were the most common with detection in 64 of 125 genotypes (51.2 %), followed by V(−), which was present in 30 genotypes (24.0 %). When only the Brazilian genotypes were considered, three haplotypes [I(+), III(− and +)] were not found. The most frequent combination in Brazil was V(−), which was detected in 133 of 213 genotypes. The TaALMT1 alleles Type V and Type VI and the insertion in the TaMATE1B promoter usually represent genotypes with higher gene expression and organic acid efflux (Sasaki et al. 2006; Tovkach et al. 2013); this superior combination was found in 32 genotypes, most of which (90.6 %) were from Brazil. When Brazil was excluded, the superior alleles were found in one genotype from Guatemala (Maya 74), one from Argentina (Klein Lucero), and one from Israel (Bet Dagan 131). The other five genotypes containing the haplotype with TaMATE1B(+) but presenting TaALMT1 alleles different from V or VI were Galego Rapado [haplotype I(+)], Fortaleza and Horto [haplotype II(+)], and IAS 14—Contestado and Emb_04101 [haplotype IV(+)]. Among all geographical regions, none of the materials showed haplotype III(+) and the majority (86.5 %) presented the TaMATE1B(+) allele in association with the TaALMT1 alleles Type V and Type VI.

Relationships between Al3+ resistance and the TaALMT1 and TaMATE1B alleles

The controls used here showed alleles corresponding to the previously published association (Sasaki et al. 2006; Tovkach et al. 2013) where the susceptible control Anahuac75 presented haplotype I(−) and the resistant control IAC 5—Maringá presented haplotype VI(+) (Supplementary file 1). The proportion of the TaALMT1 alleles I and II in categories S and MS identified by the hydroponic and field tests were 79.6 and 94.2 %, respectively, which confirmed the association of these promoters with the Al3+-susceptible genotypes (Table 5). The proportions of alleles V and VI in categories R and MR were 67.1 and 58.0 % in the hydroponic and soil tests, respectively, which also confirmed their association with the Al3+-resistant genotypes (Table 5). When only haplotypes V(+) and VI(+) evaluated in the hydroponic tests were considered, 87.5 % of the genotypes were classified as R or MR; among the 20 genotypes carrying haplotype V(+), 16 were R or MR; all genotypes with haplotype VI(+) were R or MR. Additionally, the proportion under field conditions increased to 90.6 %, with 18 genotypes with haplotypes V(+) and 11 VI(+) classified as R or MR (Table 5).

Table 5 Distribution of the haplotypes formed by the combination of the six TaALMT1 alleles and two TaMATE1B alleles in different Al3+-resistance classes in the hydroponics and field experiments
Full size table

When only the two TaMATE1B promoter alleles (presence or absence of the insertion) were considered, the Pearson correlation with Al3+ resistance was higher for the field (−0.43> r > −0.41, P < 0.001) than the hydroponic experiments (0.28 with RRG and 0.25 with CIATH, P < 0.001) (Table 3). Similarly, the correlation of all TaALMT1 alleles with the Al3+ resistance phenotype in the field (−0.73> r > −0.68, P < 0.001) was higher than the hydroponic experiments (0.47 with RRG and 0.49 with CIATH, P < 0.001). When both genes were compared, the correlation with Al3+ resistance in with CIATH and CIATF was higher for the TaALMT1 (0.49 and 0.69, respectively) than for the TaMATE1B allele (0.25 and 0.37, respectively). When considering the haplotypes, the correlation with CIATH and CIATF was similar (0.50 and 0.71, respectively) when only TaALMT1 was considered (0.49 and 0.69, respectively) (Table 5). In this context, the phenotyping in the field presented a high association between the TaALMT1 and TaMATE1B alleles and Al3+ resistance, although the additive effect of the citrate efflux was not significant.

Relative contribution of traits to variability in Al3+ resistance

The analysis of the relative contribution of each trait allowed us to assess their contribution according to their impact on genetic diversity. The contribution was estimated based on the methodology described by Singh (1981). The highest contributions were obtained for CIATH (18.5 %), TaALMT1 (15.5 %) and haplotypes (12.6 %), followed by the characteristics TaMATE1B, maturation, CIATF, AVE, silking, RRG and tillering (Fig. 2). The combined effect of the traits CIATH, TaALMT1, haplotypes, TaMATE1B, maturation and CIATF represented 77.8 % of the assessment of diversity among the genotypes. Thus, the high correlation among the characteristics evaluated in the field (silking, tillering and maturation) made these variables redundant. In the future, the analysis of the later stages (tillering and maturation) should not be performed.

Fig. 2
figure 2

Relative contribution of eight characteristics to the diversity in the Al3+ resistance in 338 genotypes according to the method of Singh (1981). Tillering, relative root growth (RRG), silking, overall average (AVE), classification index of aluminium toxicity in the field (CIATF), maturation, TaMATE1B promoter alleles, haplotypes (based on the combination of the TaALMT1 and TaMATE1B alleles), TaALMT1 promoter alleles, and classification index of aluminium toxicity in hydroponic (CIATH)

Full size image

Principal component analysis

The results of the principal component analysis showed that the first two components explained 78.8 % of the diversity among the genotypes (Supplementary file 2 and 3). The genotype distribution when both components were considered did not allow differentiation into separate clear groups. However, the genotypes could be identified based on the knowledge of the characteristics CIATH, CIATF, haplotypes and origin, which allowed better separation of the groups. The distribution based on these variables confirmed that the field classification (CIATF) better discriminated the genotypes without overlap among the groups (R, MR, MS and S) (Supplementary file 2).

Year of release for the wheat cultivars and association with the TaALMT1 and TaMATE1B alleles

The distribution of the superior TaALMT1 and TaMATE1B alleles over time can be evaluated based on the year of release/obtainment for each cultivar/variety and the presence of the alleles. The materials can be considered to be under positive selection if the increase in Al3+ tolerance is associated with the presence of specific alleles. Here, we were able to obtain the year of release or obtainment for 327 genotypes, which represented more than 90 years of wheat breeding. In general, the frequency of the TaALMT1 alleles Type V and Type VI increased over time, whereas the TaMATE1B(+) allele showed a constant lower frequency (Fig. 3a). The oldest cultivars (1921–1930) presented haplotypes I/V/VI with or without TaMATE1B(+) [from Tunisia: Cailloux–haplotype V(−); from Brazil: Floreana–haplotype V(−) and PG1–haplotype VI(+); and from Portugal: Galego Rapado–haplotype I(+)]. Over the following three decades (1931–1960, Fig. 3b), 15 of the 24 cultivars released (62.5 %) showed haplotype V/VI(−) or haplotype V/VI(+), whereas the nine (37.5 %) remaining cultivars presented other haplotypes with TaMATE1B(−) or TaMATE1B(+). Over the next 40 years (1961–2000), 247 genotypes among the varieties and cultivars were released or obtained of which 147 (59.5 %) presented haplotype V/VI with TaMATE1B(−) (83.7 %) or TaMATE1B(+) (16.3 %), followed by 67 (27.1 %) possessing haplotype I/II(−) (100 %). The remaining genotypes showed different haplotypes that always contained TaMATE1B(−). A total of 52 genotypes were released or obtained between 2001 and 2013, with 37 (71.2 %) presenting haplotype V/VI(−) (96 %) or V/VI(+) (4 %) and the remaining 15 presenting haplotype I/II/IV(−) (93.3 %) or I/II/IV(+) (6.7 %). Curiously, although the presence of TaMATE1B(+) associated with the TaALMT1 allele Type V or Type VI usually increased the RRG and decreased the AVE, there was a tendency for an average reduction in the RRG and average increase in the AVE over time (Fig. 3c, d).

Fig. 3
figure 3

Distribution of TaALMT1 and TaMATE1B alleles and plant performance against Al3+, in hydroponics and field, based on more than 90 years of wheat breeding. a TaALMT1 and TaMATE1B alleles, b haplotypes, c relative root growth—RRG, and d overall average of the field grades—AVE. In c and d the red line is the average value of the characteristic in each sampled period

Full size image

Discussion

Aluminium (Al3+) toxicity is one of the main constraints that limits crop production worldwide. Therefore, the identification of resistant sources and molecular markers linked to Al3+ resistance are needed to increase plant performance under acidic soil conditions. In this work, we used hydroponic and field tests to evaluate the Al3+ resistance of 338 wheat genotypes and correlated the resistance with markers linked to two genes encoding organic acid transporters. A wide diversity in plant performance was detected, which allowed the identification of genotypes from different geographic origins containing high variations in Al3+ resistance (Tables 2, 4). Although the allelic variability in TaALMT1 and TaMATE1B in wheat was previously published (Sasaki et al. 2006; Raman et al. 2008; Han et al. 2013; Garcia-Oliveira et al. 2014; Pereira et al. 2015), we extended the understanding of that variation by demonstrating for the first time a correlation of these alleles with field data. A high correlation (r = 0.71, P < 0.001) between the TaALMT1 and TaMATE1B haplotypes with Al3+ resistance in the field was obtained (Table 3).

The selection pressures under the field conditions (soil pH from 4.3 to 4.7 with Al representing 75 to 87.9 % of the CTC) were higher than the stress applied in the hydroponic experiments because only eight materials were considered to be resistant in the field conditions (Tables 2, 5). The field evaluation described here was relatively fast based on the time required to reach the tillering stage. There was a high correlation (from 0.86 to 0.94, P < 0.001) between the phenotyping in the three growth stages (tillering, silking and tillering), indicating that only one of these stages needed be used for the evaluation of wheat Al3+ resistance. Additionally, the method of Singh (1981) indicated redundancy and thus a low contribution of the AVE, silking and tillering to the diversity in Al3+ resistance (Fig. 2), suggesting that only the evaluation at tillering was needed to facilitate the field test; this finding will reduce the laborious process of evaluating plants in the field. Moreover, the use of the controls alternately placed in every plot also facilitated the field evaluations by reducing the interference of the heterogeneous Al3+ distribution.

Al3+ resistance has been evaluated in wheat using different methods, with hydroponics being the preferred choice. The correlation between the hydroponic method and the plant performance in acidic soil has been reported to be as high as the 0.77 (P < 0.001) index obtained by Baier et al. (1995). However, is important to note that these authors used different genotypes to compare plant growth between the hydroponic and soil tests. We observed an intermediate correlation (r = 0.56, P < 0.001) between the hydroponic and field data (Table 3). Discrepancies between different phenotyping methods have also been reported for wheat (Hayes et al. 2004), barley (Szira et al. 2008) and oats (Nava et al. 2015). Moreover, the hydroponic and field tests showed different levels of stress; for instance, Al3+ could be tested in isolation in the hydroponic experiments whereas there was a mix of different factors in the field. The inconsistencies between these two methods could be partially explained by the possible non-uniformity of the nutrient distribution and other nutritional deficiencies or toxicities (e.g., phosphorus, calcium and manganese) in addition to the Al3+ toxicity. Additionally, the different physical/chemical characteristics in each environment (hydroponics and soil) could lead to the development of different morphological traits in the roots. The differences in root traits could impact water and nutrient uptake by the plant and thus its performance in the soil (Shavrukov et al. 2012). Moreover, the amount of Al3+ used in our hydroponics experiments (74 µM) suggested by Voss et al. (2006) might not be sufficient to discriminate the wheat genotypes; thus, a wider genetic basis and adjustments in the Al3+ concentrations are required to improve the correlation with the field tests.

The different TaALMT1 and TaMATE1B alleles were previously shown to be associated with Al3+ resistance (Sasaki et al. 2006; Raman et al. 2008; Tovkach et al. 2013; Garcia-Oliveira et al. 2014; Pereira et al. 2015). The TaALMT1 alleles V and VI were correlated with higher gene expression and malate efflux by the root tip, whereas an insertion upstream of the TaMATE1B promoter region extended and increased the gene expression to the root apex, thereby enhancing the citrate efflux (Sasaki et al. 2006, Tovkach et al. 2013). Here, six TaALMT1 promoter alleles were found in addition to the two TaMATE1B alleles. The majority of the TaALMT1 and TaMATE1B alleles detected here were consistent with the alleles detailed in previous reports (Raman et al. 2008; Han et al. 2013; Garcia-Oliveira et al. 2014; Pereira et al. 2015). However, as related by Raman et al. (2008) and Pereira et al. (2015), we found inconsistent results for the TaALMT1 alleles in the Siete Cerros, INIA F66 and Morocco cultivars compared with the results published by Raman et al. (2008). These inconsistencies could be explained by the incorrect labelling of the accessions or multilines. Most of the genotypes with alleles I/II or V/VI evaluated in this study were characterized as susceptible (S/MS) or resistant (MR/R), respectively, although the proportions were not the same (Table 5). For instance, 79.6 and 94.2 % of the genotypes classified as S/MS in the hydroponics and soil, respectively, possessed the TaALMT1 I/II alleles, whereas 67.1 and 58.0 % of the genotypes with alleles V/VI were discriminated as MR/R. Eighty-seven genotypes in the field (42.0 %) and 68 in the hydroponic (32.9 %) experiments presented alleles V/VI and were classified as S/MS. This study is the first report of non-Japanese wheat lines carrying TaALMT1 promoters V and VI presenting the Al3+-sensitive or moderately sensitive phenotype, suggesting that some factor(s) in addition to the level of TaALMT1 expression is (are) involved in the control of the malate efflux in these genotypes. Conversely, 21 genotypes presenting alleles I/II were classified as MR/R in the hydroponic experiments and six of them (Fortaleza, Dom Marco, Horto, IAS c46—Curitiba, Minuano and Emb_05104) were also identified as MR in the field (Supplementary file 1). Among these six genotypes, Fortaleza and Horto presented the allele TaMATE1B(+), which could partially explain the good performance. However, the other four genotypes lacked the insertion in the TaMATE1B promoter; thus, other Al3+ resistance mechanisms may be involved in the Al3+ resistance presented by these genotypes. These other mechanism could be related to a number of genes and metabolic pathways involved in Al3+ resistance (Raman et al. 2010; Ryan et al. 2011; Delhaize et al. 2012). More analyses should be performed to elucidate why the lower number of repeats in the TaALMT1 promoter of these genotypes was correlated with Al3+ resistance or moderate resistance. TaALMT1 alleles I/II were previously detected in Al3+-resistant genotypes (Raman et al. 2008), but the mechanisms that contributed to the resistance that were not related to the organic acid efflux were not described.

In terms of the TaMATE1B gene, we identified the lack of the insertion in the promoter as the most frequent allele. Only 37 of 338 genotypes showed the insertion (Table 4): 33 of them originated from Brazil, two from another American country (Klein Lucero from Argentina and Maya 74 from Guatemala), one from Europe (Galego Rapado from Portugal) and one from the Middle East (Bet Dagan 131 from Israel). The insertion in the TaMATE1B promoter was widespread among Portuguese wheat cultivars (Garcia-Oliveira et al. 2014), and it is possible that the Portuguese immigrants who established in Brazil transported the varieties carrying that allele. However, although the majority of the 37 genotypes containing the TaMATE1B(+) allele were from Brazil, the frequency was low (15.5 % representing 33 genotypes among the 213 Brazilian accessions). This finding was in agreement with the report by Pereira et al. (2015) that detected the TaMATE1B(+) allele in 80 of the 302 Brazilian genotypes evaluated. Here, more than 86 % of the TaMATE1B(+) genotypes were classified as R/MR in the field or hydroponic conditions, and only one genotype (Onix) containing the TaMATE1B(+) allele was susceptible in the hydroponic and field experiments. We found a high association of the TaMATE1B(+) allele with the TaALMT1 promoters V/VI, with 32 of 37 genotypes containing the TaMATE1B(+) allele showing the alleles V/VI. Among these 32 genotypes, 29 were classified as R/MR in the field. The combination of these superior alleles is interesting because they were proposed to possess higher malate and citrate efflux by the root tip. In fact, the Brazilian genotypes IAC 5—Maringá and Toropi, which are recognized as important Al3+-resistant sources, showed the combination of the superior TaALMT1 and TaMATE1B alleles. Moreover, the best combinations for root growth of the 33 Brazilian wheat cultivars in a short-term soil experiment were reported to be the TaALMT1 promoters V/VI and the insertion in the TaMATE1B promoter (Pereira et al. 2015).

The combination of the TaALMT1 and TaMATE1B alleles detected here allowed the identification of 11 haplotypes associated with higher relative contributions to the diversity of Al3+ resistance (Fig. 2). The correlation of the TaALMT1 and TaMATE1B alleles with the hydroponic and field data was similar (0.71) to the correlation of the TaALMT1 alleles alone (0.69). In a recent survey, the only consistent QTL across the tested environments in an RIL population used to map genes for acidic soil tolerance in the field was centred on the TaALMT1 gene on chromosome 4DL, which accounted for up to 38.5 % of the total phenotypic variation (Liu et al. 2015). That report and the data described here show the importance of TaALMT1 for acid soil tolerance under field conditions. In contrast, small and not significant difference was detected between TaALMT1 alleles I and V under mildly acid soils although no substantial disadvantage was observed for cultivars with TaALMT1 allele V (Eagles et al. 2014). To further evaluate the impact of the TaMATE1B(+) allele on Al3+ resistance, we analysed its frequency over 90 years of breeding (the oldest cultivar Galego Rapado was obtained in 1922 and the newest cultivar BRS Marcante was obtained in 2013). An increase in the TaMATE1B(+) allele frequency was not detected; conversely, the number of TaALMT1 alleles V/VI was significantly increased (Fig. 3a). This finding demonstrated that the TaALMT1 alleles were selected by the breeding programmes but that there was little selection pressure for the TaMATE1B(+) allele, indicating the low adaptive power of that allele. Results recently published by our group revealed that the transposon insertion in the TaMATE1B promoter was significantly advantageous for a few materials; however, no clear correlation was observed with higher root growth under short-term soil experiments in 33 Brazilian wheat genotypes (Pereira et al. 2015). Despite the lower frequency of the allele (+), we confirmed that a clear correlation could not be established between the TaMATE1B(+) allele and higher root growth in hydroponics or better plant performance in field conditions even using a wider genetic basis. Recently, it has also been showed that TaMATE1B(+) conferred higher Al3+ tolerance when associated with an inferior TaALMT1 allele but, when combined with a superior TaALMT1 allele, the increase in tolerance was small (Han et al. 2016). The TaMATE1B(+) allele appeared to not have a good association with other traits because no benefits of citrate efflux for phosphorus nutrition were observed in studies with near-isogenic wheat lines (Ryan et al. 2014). Based on these results, the effective role of the insertion in the TaMATE1B promoter for traits related to the citrate efflux (e.g., Al3+ resistance or phosphorus nutrition) appeared to be limited. Reasons for that limited additive effect of the citrate efflux possibly include one or more of the different assumptions detailed by (Ryan et al. 2014) as, for example, low concentration and/or diffusion of citrate in the soil due to lower citrate efflux by mature roots, microbial degradation and/or soil chemistry. Clearly, these assumptions should be experimentally evaluated especially because the limited role of the citrate efflux contradicts the theoretical potential of citrate that is approximately eightfold more effective than malate in chelating Al3+ (Zheng et al. 1998).

Most of the Brazilian wheat varieties analysed by others were considered resistant (de Sousa 1998; Raman et al. 2008). In this study, 64.8 and 57.3 % of the Brazilian genotypes were Al3+-resistant or moderately resistant in the hydroponic and soil tests, respectively (Table 2), thereby confirming the importance of Brazilian wheat as an Al3+-resistant source. However, although acidic soils are common in the wheat breeding regions in Brazil, some of the Brazilian cultivars detected here as Al3+-sensitive or moderately sensitivity were recommended for growth in our country and presented good yields in the past (Castro et al. 2011). Six moderately resistant foreign genotypes identified here were interesting and should be considered for further characterization. Among them, only Maya 74 (from Guatemala) and Bet Dagan 131 (from Israel) possessed Frontana as an ancestor. The other non-Brazilian genotypes classified as MR were Aburakomugi (from Japan), Escondido 41 (from USA), Giza (from Egypt) and Klein Lucero (from Argentina). These materials are good candidates for incorporation into Brazilian wheat breeding programmes. Stodart et al. (2007) found landraces that originated from different countries and represented different Al3+-tolerant haplotypes compared with the materials actually used in breeding programmes. In this way, new sources can contribute with additive alleles and potentially improve the current level of Al3+ tolerance in modern wheat cultivars.

In conclusion, for the first time markers linked to the TaALMT1 and TaMATE1B promoter regions were associated with Al3+ resistance in field conditions. The high correlation detected here is interesting for wheat breeding programmes targeting better wheat growth in fields containing very acidic soil. Only eight of the 338 genotypes were classified as resistant in field conditions, revealing the high selection pressures in our condition. Among these eight genotypes, Aceguá, C33, IAC-21—Iguaçú and PAT 7219 were not widely recognized as Al3+-resistant sources. Six materials (Dom Marco, Fortaleza, Horto, IAS c46—Curitiba, Minuano and Emb_05104) showed TaALMT1 alleles that were not usually associated with Al3+ resistance but presented good performances in the hydroponic and field tests. These results indicate the potential for new Al3+ resistance mechanisms, and thus these genotypes should be further characterized.

Author contribution statement

L.C. and J.P.S. Jr designed the research. J.G.A., J.A.D.M., C.M.F. and D.B. performed the hydroponic experiments. J.G.A., J.P.S. Jr and J.A.D.M. performed the field tests. J.G.A. and J.F.P. evaluated the PCR markers. J.G.A., J.F.P., L.C. and J.P.S. Jr analysed the data and wrote the paper.