Assessing the effectiveness of the TaMATE1B and TaALMT1 genes to enhance the Al3+ tolerance of durum wheat (Triticum turgidum) grown under controlled conditions and in the �eld.

Purpose Durum wheat is sensitive of acid soils because it lacks effective genes for Al 3+ tolerance. Previous research showed introgression of the bread wheat (Triticum aestivum) TaMATE1B and TaALMT1 genes individually increased the Al 3+ tolerance of durum wheat. Here we aimed to (a) combine the genes into a single durum line, (b) compare the various introgression lines to each another and (c) establish the effectiveness of the introgressions in improving tolerance of acid soils in the �eld.


Introduction
The high protein content of durum wheat (Triticum turgidum) is suitable for pasta production which makes it a lucrative crop and an important option for alleviating poverty in countries such as Ethiopia (Sall et al. 2019;Sissons 2008). However, current durum wheat cultivars are sensitive of acid soils because they lack genes that confer Al 3+ tolerance. Acid soils are widespread around the world and a common toxin found in these soils is the Al 3+ released from normally innocuous forms by the low pH (von Uexküll and Mutert 1995). The use of Al 3+ tolerant germplasm to breed crops able to grow effectively on acid soils forms part of the strategy for managing these soils (Ryan 2018). Acid soils constrain the effective production of durum wheat and Ethiopia in particular has many regions where acid soils restrict the growth of this preferred crop (Wayima 2022). Inhibition of root growth is the earliest obvious symptom of Al 3+ toxicity amongst the small grain cereals (Kochian et al. 2015). When root growth is inhibited by Al 3+ toxicity, plants are compromised in their ability to take up nutrients and water. Durum wheat is one of the most Al 3+ -sensitive of the small-grained cereals (Bona et al. 1991) and screens of T. turgidum germplasm comprising both cultivars and landraces have failed to identify useful levels of Al 3+ tolerance ; Wayima et al. 2019). This contrasts with bread wheat (Triticum aestivum) which shows a large variation in Al 3+ tolerance within its germplasm. Although Al 3+ tolerance in bread wheat is largely conferred by the major genes TaALMT1 and TaMATE1B (Delhaize et al. 2012), other genes of smaller effect confer additive tolerance ). TaALMT1 encodes an Al 3+activated anion channel that is permeable to malate. In the presence of Al 3+ , the TaALMT1 protein, which resides on the plasma membrane of root apical cells, is activated allowing the e ux of malate down its electrochemical gradient (Delhaize et al. 2012; Ryan et al. 1997). The malate acts to chelate the Al 3+ to protect root apices which are particularly susceptible of Al 3+ toxicity. Although TaMATE1B also encodes a membrane protein it transports citrate and, unlike TaALMT1, TaMATE1B function is constitutive such that the protein does not require Al 3+ for citrate to be released Tovkach et al. 2013). Like malate, citrate chelates Al 3+ to render it non-toxic. Tolerant alleles of both genes are due to enhanced expression of the respective coding regions. In the case of TaALMT1 enhanced expression is due to repeated sequences in the promoter whereas for TaMATE1B a large transposable element inserted into the promoter confers enhanced expression (Delhaize et al. 2012).
To enhance the acid-soil tolerance of durum wheat, Al 3+ -tolerant alleles of both TaALMT1 and TaMATE1B were introgressed from bread wheat into durum wheat. Starting with a chromosome 4D substitution line of durum wheat, the pH1c mutant of durum was used to introgress a chromosomal fragment of TaALMT1 of bread wheat to chromosome 4B of durum wheat (Han et al. 2014;Han et al. 2016). Since TaMATE1B maps to chromosome 4B, it was possible to undertake direct crosses between durum and bread wheat to introgress TaMATE1B into the durum genetic background (Han et al. 2016).
Analysis of durum lines with the introgressed genes found that TaMATE1B was more effective gene than TaALMT1 in conferring acid soil tolerance in short term experiments of about a week (Han et al. 2016). Based on this nding, subsequent research focused on the TaMATE1B line which was able to develop an effective root system in an acid subsoil that the control was unable to achieve in a large rhizobox where plants were grown to maturity and grain yield (Pooniya et al. 2020). Durum wheat is grown in regions that often experience a terminal drought (Boussakouran et al. 2019;Habash et al. 2014) and climate change will likely exacerbate the extent and severity of these events (Habash et al. 2010). A subsequent growth study mimicked this situation by growing plants in a reconstituted eld soil with an acid subsoil and exposing them to a terminal drought. The results showed that the TaMATE1B line grew larger and yielded signi cantly more grain than the control line (Lui et al. 2020).
Here we describe the development of a new durum line that combines TaALMT1 and TaMATE1B. This "double gene" line was compared to the parental cultivar and durum lines that possessed the individual genes separately. Growth of the lines were compared by hydroponics and in soil pots grown under the controlled conditions of a cabinet. We further compared their performance in eld trials on acid soils and demonstrate the bene t of these introgressions to grain yields under eld conditions.

Germplasm
The development of durum wheat lines with TaALMT1 and TaMATE1B introgressed separately into a Jandaroi genetic background has been described previously (Han et al. 2016;Pooniya et al. 2020). These lines had been backcrossed three times to the cultivar Jandaroi. Here we refer to the various lines as Null (Jandaroi parental line), TaALMT1-Jandaroi and TaMATE1B-Jandraoi. Since a small chromosomal fragment with TaALMT1 had been introgressed into the 4B chromosome and TaMATE1B is located on chromosome 4B of bread and durum wheat, we needed to identify a recombinant line that possessed both genes on the same chromosome. To develop a line with both genes introgressed we crossed the backcross three lines with the separate genes in the Jandaroi background to one another. To identify suitable recombinants, we rst screened seedlings with a co-dominant marker for TaMATE1B (Tovkach et al. 2013) to identify seedlings that were homozygous for TaMATE1B. Once identi ed, we screened those lines with a dominant marker for TaALMT1 (Sasaki et al. 2004). The dominant marker for TaALMT1 did not allow us to identify seedlings homozygous for this gene but having identi ed the presence of at least one 4B chromosome that possessed TaALMT1 in a homozygous TaMATE1B background ensured that a recombination had occurred. Selected recombinants were self-fertilised to generate F 2 and resulting F 3 populations were screened for lines that possessed homozygous Al 3+ -tolerant alleles of both TaALMT1 and TaMATE1B. We named a line identi ed from the screening as the TaALMT1 + TaMATE1B-Jandaroi line or "double gene" line.

Hydroponics
For both long and short-term experiments seedlings were initially grown using a basket (28 cm diameter × 12 cm height with 2 mm mesh size) that was positioned on a top of a bucket (28 cm diameter × 28 cm height) lled with approximately 10 L of 0.2 mM CaCl 2 solution so that the bottom of the basket touched the surface of the solution. Seedling pre-germinated on Petri dishes for 2 days in the dark were placed in the basket so that the roots passed through the mesh into the solution. The CaCl 2 solution was aerated with an aquarium air pump, and the basket covered with aluminium foil to keep the plants in the dark. The seedlings were grown in the dark for another 6 days at room temperature until the coleoptiles develop to about10 cm. After the 6 days growth seedlings were transferred to a plastic tank (38.5 cm length × 23.5 cm width × 32.3 cm height) covered with a lid. A total of 12 slits (7 cm length × 1.5 cm width) were made around the lid to hold the plants. The tank was lled with 22 L of a basal nutrient solution comprising of 500 µM CaCl 2 , 150 µM MgSO 4 , 1000 µM KNO 3 , 500 µM NH 4 Cl, 2 µM FeEDTA, 10 µM KH 2 PO 4 , 11 µM H 3 BO 3 , 2 µM MnCl 2 , 0.35 µM ZnSO 4 and 0.2 µM CuCl 2 . The seedling coleoptiles were wrapped with a piece of sponge and carefully inserted into the slits, so that the roots grew into the nutrient solution while the shoot grew above the tank. The Al 3+ treatments were prepared by addition of stock 1 M AlCl 3 to the nal concentration. The tanks were aerated with an air compressor at the rate of 8.8 L air min − 1 tank − 1 . The plants were grown for a further 7 days (short-term experiment) or 21 days (long-term experiment) in a growth cabinet (Conviron, Canada) set to 16 h day/8 h night cycle (24°C/20°C) and 600 µmol photon m − 2 s − 1 . The pH of solutions was monitored and adjusted daily to 4.3 for all treatments. Solutions were not changed for the short-term experiment whereas for the long-term experiment nutrient solutions were replaced on days 8, 11, 14 and 19 after seedlings had been planted in the tanks. At harvest, roots were cut away from shoots and stored in 70% ethanol (v/v) until they were scanned and analysed with WinRHIZO™ software (Regent Instruments Inc., Quebec, Canada). The length of axial roots (roots without lateral roots) was measured with a ruler and this was subtracted from the length of the total root system to obtain the lateral root length.

Soil Experiments
An acid ferrosol described previously ) was collected from a farm in Robertson, NSW (latitude 34.63° S; longitude 150.48° E). The soil was air dried, sieved (4 mm) and stored at ambient temperature. The initial soil pH was 4.2, as determined by extraction with 10 mM CaCl 2 . Prior to use for the experiment, soil moisture was adjusted to 324 mL H 2 O kg -1 dry soil (90% of the eld capacity) with deionised water. Lime (CaCO 3 ) was added (4 g kg -1 dry soil) to neutralise the acidity for control treatments. For the long-term soil experiment, nutrients were added as described below. No nutrients were added in the short-term soil experiment as nutrients in the seed were su cient to supply the developing seedlings over 10 days.
Short-term (10 days) and long-term (3 weeks) pot trials were established to enable comparison of root morphology of young seedlings and more mature plants on acid and limed (control) soils. For the shortterm soil experiment, 15 cm tall plastic tubes with an 9 cm diameter were used. A mass of 620 g of moistened soil was packed to 0.88 g cm -3 with no added nutrients. The genotypes were tested on the acid and limed (control) soil with 4 replicates per genotype per treatment. Pots were watered daily to weight to maintain the starting moisture content.
For the long-term soil experiment, 50 cm tall plastic pots with a 9 cm diameter were used. Soil was added as two strati ed layers. The bottom layer (39 cm) contained 2150 g of soil, compacted to about 0.69 g cm -3 , whilst the top 10 cm layer contained 550 g of moist soil compacted to about 0.86 g cm -3 . A thin layer of polyethylene pellets was added between the top and bottom layers to allow the layers to be differentiated at subsequent sampling. Since this was a long-term experiment, nutrients were added to sustain plant growth and 400 mg of P as KH 2 PO 4 was added only to the top 10 cm layer of the soil pro le to mimic eld conditions where the P is typically found in the top 5 to 10 cm of the soil pro le. The following nutrients soil (mg kg -1 ) were added to both layers: Ca(NO 3 ) 2 .4H 2 O (400), (NH 2 ) 4 SO 4 (250),  (Table 1). Each genotype was sown with four (Gunning and Tarcutta) or three (Merredin) replications in randomised complete blocks in trials that included other germplasm. For Tarcutta and Merredin plots were 6 m long by 1.75 m wide (6 rows at 25 cm inter-row; approximately 120 plants m − 2 ) that were cut back to 5 m long at harvest. Only Jandaroi-TaMATE1B was available for the Tarcutta trial run in 2018 and, due to limited grain, single rows 3 m long were sown at Gunning in 2019.
Su cient grain was available for the Merredin trial in 2019 to sow 6 m by 1.8 m plots of each genotype but the trial experienced severe drought. Nitrogen was supplied as urea prior to sowing and was topdressed again at stem elongation with urea (timed with rainfall), or liquid N fertilisers (e.g. urea and ammonium nitrate liquid formulations). Muriate and sulphate of potash (KCl and K 2 SO 4 ) were applied as required to ensure adequate K and S, as were micronutrients. Foliar diseases were managed with prophylactic fungicide and pesticide applications. Weeds were managed with district herbicide application practices at the recommended rates. Quadrat cuts comprising of 1 m 2 from each plot were collected close to anthesis for the Tarcutta and Merredin trials whereas total shoot biomass of plants collected at Gunning was measured at harvest. The Tarcutta and Merredin trials were machine-harvested whereas the Gunning trial was hand-harvested and grain cleaned with a table-top thresher.

Hydroponics
The Al 3+ tolerance of the various lines was rst assessed in a hydroponic experiment over ten days growth in acidic nutrient solutions that contained Al 3+ . Figure 1 shows that all the lines that possessed either TaALMT1 or TaMATE1B individually or both genes together had roots that were more tolerant of Al 3+ than the parental cultivar Jandaroi (Null line). The enhanced tolerance measured as root length was particularly evident for axial seminal roots (Fig. 1B) whereas lateral roots for all lines were more sensitive of the Al 3+ although, even here, the lines still differed from one another and maintained their relative tolerance rankings (Fig. 1C). Lines that possessed TaALMT1 either by itself or in combination with TaMATE1B performed best with the double gene line showing the greatest Al 3+ tolerance of lateral roots at 2.5 and 5.0 µM AlCl 3 .
We then compared the three Al 3+ -tolerant lines to each other in a longer-term 21 day growth experiment in solutions that contained Al 3+ . In this experiment, Jandaroi was omitted as its extreme sensitivity of Al 3+ would have resulted in very short roots and the aim was to establish if the Al 3+ -tolerant lines differed from one another. The three Al 3+ -tolerant lines performed similarly to one another for total root length ( Fig. 2A) and axial root length (Fig. 2B)  were largely driven by differences in lengths of lateral roots (Fig. 2C). In this experiment the double gene line did not perform better than the single gene line containing TaALMT1 only.

Pot trials
In contrast to the hydroponics experiments, lines with TaMATE1B only or in combination with TaALMT1, generally performed best when grown on acid soil in pots (Figs. 3 and 4). In the short-term experiment, both TaMATE1B-containing lines had the most tolerant roots although the TaALMT1 line still performed better than the Null line (Fig. 3). The double gene line did not perform better than the line with TaMATE1B alone. When the soil had been limed to detoxify the Al 3+ , root growth of all four lines, including the Null line, was similar to one another (Fig. 3).
For the long-term pot experiments a pro le with an acidic subsoil was constructed that approximated the eld. The pots were 50 cm tall and P fertiliser was applied only in the top 10 cm re ecting the strati cation of P in the soil pro le that is commonly found in soils. Phosphate can bind with Al 3+ and the application of K phosphate in the top 10 cm served to detoxify toxic Al 3+ . This was clearly demonstrated by measuring root weights in the upper soil layer as they were the same for all lines with or without lime addition (Fig. 5A). By contrast, the TaMATE1B lines had the longest roots in the lower acid layer with the double gene line tending to have the longest roots (Fig. 4A). All lines had similar root lengths to one another in the limed soil and although the Al 3+ tolerance genes enabled roots to grow in the acid layer they did not grow as well as in soil that had lime incorporated throughout the pot as determined by root length (Fig. 4A and 4B). The TaMATE1B lines had the greatest shoot dry weights (Fig. 5B) in the acid soil treatment although this was not as marked as the differences in root growth found in the bottom 39 cm of the tube (Fig. 4A).

Field experiments
Although we expected that pot experiments would more closely approximate the eld than hydroponics experiments, they still only approximate eld conditions. For instance, water was applied to pots daily and Al 3+ -sensitive roots of the Null line grew well in the top 10 cm of the P-fertilised layer which had been detoxi ed for Al 3+ as discussed above and stayed moistened for the duration of growth (Fig. 5A). To assess the lines in the eld, Jandaroi-TaMATE1B and Jandaroi lines were rst compared in a moderately acidic site at Tarcutta in 2018 and all lines were subsequently grown together at two other sites where acidity was more severe (Merredin and Gunning in 2019).
The TaMATE1B line produced about 40% more grain than Jandaroi when grown at Tarcutta in 2018 (Fig. 6A). Shoot biomass estimated with a quadrat cut pre-anthesis showed a similar trend to grain yield although it was not statistically signi cant at P < 0.05 (Fig. S1A). A pronounced bene t of TaMATE1B alone or in combination with TaALMT1 was evident at Gunning in 2019 ( Fig. 6B; note that data for Gunning are from single rows whereas data for Tarcutta and Merredin are from whole plots). There was a tendency for the double gene line to produce the most grain whereas the TaALMT1 line was similar to the Null line. Lines that possessed TaMATE1B (lines TaMATE1B and TaALMT1 + TaMATE1B) had a signi cantly greater grain yield than lines that lacked TaMATE1B (lines Null and TaALMT1) when assessed by Student's t-test (P < 0.001). For the Gunning trial the Null and TaALMT1 lines produced about 50% of the shoot biomass of the TaMATE1B lines (Fig. S1B) although they produced almost no grain (Fig. 6B). Merredin in 2019 was severely droughted with the Null line producing little or no grain (Fig. 6C).
Although grain yields were considerably lower than would have been expected for a normal year, the Al 3+ tolerance genes still enabled plants to produce some grain with an indication that line Jandaroi TaMATE1B + TaALMT1 produced the most grain. Although the Al 3+ tolerance genes enabled durum to set grain whereas the Jandaoi Null line produced almost no grain, the large variability and only three replicates meant grain data for all lines were not signi cantly different from the Null line at P < 0.05 (Fig. 6C). Total biomass prior to anthesis showed similar trends to grain yields in the Merredin trial although in this case the TaMATE1B + TaALMT1 line was signi cantly larger than the Null line (Fig. S1C).

Discussion
Here we show that TaMATE1B and TaALMT1 derived from bread wheat conferred Al 3+ tolerance to root growth of durum wheat when grown in hydroponics and acid soils. Although TaALMT1 was more effective than TaMATE1B in conferring Al 3+ tolerance when grown in hydroponics ( Figs. 1 and 2), the opposite was apparent when the lines were grown in soil (Figs. 3 and 4). This inverse relationship of the tolerance of Jandaroi-TaMATE1B and Jandaroi-TaALMT1 in hydroponics and soil was previously described for young seedlings (Han et al. 2016) and here we extend the work to nd a similar relationship in more mature plants. Lateral roots were generally more sensitive of the Al 3+ than the axial roots for all lines when grown in both hydroponics and acid soil (Figs. 1 to 3) even though Al 3+ -tolerant wheat lines are capable of citrate and malate release from tips of lateral roots (Kawasaki et al. 2018). Lateral roots comprised the largest proportion of root system length and had the greatest in uence on total root length for both hydroponics and soil experiments. The bene t in shoot growth of lines having the TaMATE1B gene was evident for the long-term soil experiment but it was a relatively small advantage (~ 1.3-fold) when compared to the bene t on root growth in the acid layer of the pot experiment (~ 4-fold). This subdued effect on shoots can be attributed to (a) the top 10 cm of the soil having been detoxi ed by the application of P-fertiliser and (b) water being applied daily to the surface of the pots. The daily watering ensured that the surface soil remained moist and allowed root growth to be maintained for all lines including the Null line (Fig. 5A). Similarly, previous work (Pooniya et al.) found that when the TaMATE1B line was grown to maturity in a severely acid soil in large rhizoboxes, grain yield was not increased compared to a Null line even though root growth was considerably improved. Similar to the current experiment, the top 10 cm of soil was not Al 3+ toxic and plants were watered daily on the surface. When the TaMATE1B line was grown in large pots and experienced a terminal drought the bene t of TaMATE1B on grain yield became apparent as a result of an ability to access soil moisture at depth (Lui et al. 2020). Importantly, there was no evidence of introgression of deleterious genes linked to TaALMT1 or TaMATE1B on the basis of the long-term pot trial. Here we show that shoot biomass and root growth (Fig. 5) on the limed treatments where Al 3+ toxicity was eliminated were similar for all lines.
The bene t on grain yield of durum lines with introgressed Al 3+ tolerance genes was apparent in eld trials at several sites with acid soils. The TaMATE1B gene by itself conferred a greater grain yield than Jandaroi on the moderately acid site of Tarcutta (Fig. 6A). The better performance of TaMATE1B than TaALMT1 was evident in the eld trials undertaken at the Gunning 2019 and Merredin 2019 sites. There was a tendency for the double gene line to produce the most grain yield at these two very acidic sites suggesting an additive effect of the two genes on Al 3+ tolerance which was most evident in the severelydroughted site in Merredin. Previous work comparing the TaMATE1B and TaALMT1 lines in young seedlings grown for a week found that TaMATE1B was more effective in conferring Al 3+ tolerance to root growth in acid soil than TaALMT1  Here we show that that the TaMATE1B line performed better for root growth than the TaALMT1 line in pot trials with acid soil indicating that short term experiments were able to predict performance in longer term experiments and in the eld. However, we suggest the double gene line warrants further research as it may be the best option on severely acid soils particularly when experiencing drought. An improved drought tolerance of the double gene line compared with the single gene lines might be a consequence of longer roots than the single gene lines in enabling access deep water as suggested by root lengths in the acid layer of the long-term pot trial (Fig. 4A).
From a breeding perspective it is simpler to introgress a single genetic locus for the desired trait instead of two or more loci. The TaMATE1B-Jandaroi line performed best under controlled conditions and this was also evident in the eld indicating that introgression of this gene alone could be su cient to obtain effective Al 3+ tolerance in the eld. Although we suggest the double gene line could be the best option for production on acid soils but this shouldn't complicate breeding efforts. Both genes have been introgressed onto the 4B chromosome of durum wheat and are linked such that they can be transferred together to generate new lines. Molecular markers are available for each gene to verify that a recombination event in progeny has not dissociated the genes from one another. Although the Al 3+tolerant durum lines enabled improved growth on acid soil, liming practices should still be maintained. Note that the most effective root and shoot growth was found in the limed treatments of the long-term pot trials (Figs. 4B and 5B). In that experiment the soils were well-mixed with the lime and the acid pH was effectively neutralised throughout the pro le. Liming is likely to provide the most effective solution to soil acidity in the eld if it can be incorporated throughout the soil pro le which may take years with repeated applications (Azam and Gazey 2021). Lime application can be costly particularly if it needs to be transported long distances and is typically applied to the surface soil. Incorporating the lime to depth incurs additional cost that can be prohibitive. The cost of transporting and incorporating lime is a particular challenge for developing countries such as Ethiopia where durum wheat is grown. Furthermore, previous research on bread wheat has shown that combining liming practices with Al 3+ -tolerant bread wheat germplasm provides an effective strategy for managing acid soils (Scott et al. 2001;Tang et al. 2003) and a similar argument can be made for durum wheat. The lines generated from this study warrant further assessment in eld trials on a range of acid soils. Our initial eld trials provide supportive evidence that the TaALMT1 and TaMATE1B genes can be used to enhance the Al 3+ tolerance of commercial durum germplasm. When combined with liming practices, these lines will have the potential of greatly improving durum production on acid soils.  TaALMT1 and TaMATE1B introgressed from bread wheat confer Al 3+ tolerance to durum wheat in shortterm hydroponic assays. Seedlings were grown in hydroponic culture for seven days after which roots were collected and scanned to determine total root length (a), axial root length (b) and lateral root length (c). Data were analysed by one way ANOVA for each concentration with different letters indicating signi cant differences at P < 0.05 (n = 3; grey circles for individual values, error bars indicate the standard error of the mean and in some cases error bars are obscured by the circles). The null line indicates Jandaroi as the parental genotype for all lines. Comparison of Al 3+ tolerance in long-term hydroponic assays of durum wheat lines with TaALMT1, TaMATE1B and TaALMT1 + TaMATE1B introgressed from bread wheat. Seedlings were grown in hydroponic culture for 21 days after which roots were collected and scanned to determine total root length (a), axial root length (b) and lateral root length (c). Data were analysed by one way ANOVA for each concentration with different letters indicating signi cant differences at P < 0.05 (n = 8; grey circles for individual values and error bars indicate the standard error of the mean). The null line Jandaroi was not included due to its extreme sensitivity of Al 3+ as determined in the short-term assay.

Figure 3
TaALMT1 and TaMATE1B introgressed from bread wheat confer Al 3+ tolerance to durum wheat in shortterm soil assays. Seedlings were grown in soil culture for ten days after which roots were washed out and analysed to determine total root length (a), axial root length (b) and lateral root length (c). The acid soil was a highly acid ferrosol that was treated with CaCO 3 to neutralise the acidity for the limed treatment.
Data were analysed by a two-way ANOVA for factors genotype and treatment with different letters indicating signi cant differences at P < 0.05 (n = 8; grey circles for individual values and error bars indicate the standard error of the mean). The null line indicates Jandaroi as the parental genotype for all lines.

Figure 4
TaMATE1B + TaALMT1 introgressed from bread wheat confers Al 3+ tolerance to durum wheat in a longterm soil assay. Seedlings were grown in soil culture with two layers comprising of soil supplemented with phosphate in the top 10 cm while the lower layer (39 cm) had all nutrients supplied except for P. After 27 days roots were washed out from the lower layer and scanned to determine total root length in the acid soil (A) and the same soil that had been limed (B). Data were analysed by one way ANOVA for each soil type (acid or limed) with different letters indicating signi cant differences at P < 0.05 (n = 6 for the acid soil; n = 4 for the limed soil; grey circles show individual values and error bars indicate the standard error of the mean). The null line indicates Jandaroi as the parental genotype for all lines and there were no signi cant differences between genotypes found in the limed soil. Root growth in a surface soil that had been supplemented with phosphate did not differ between lines whereas TaMATE1B lines had improved shoot growth. Seedlings were grown in soil culture with two layers comprising of soil supplemented with phosphate in the top 10 cm while the lower layer (39 cm) had all nutrients supplied except for P. After 27 days roots were washed out from the upper layer and shoots collected. Roots (A) and shoots (B) were dried and weighed. Data for roots (A) and shoots (B) were analysed by two-way ANOVAs with no signi cant differences between genotypes at P < 0.05 for either soil (n = 6; grey circles show individual values and error bars indicate the standard error of the mean). Shoot growth of the lines possessing the TaMATE1B gene were combined and compared with Student's t-test to the lines that did not possess TaMATE1B (null and TaALMT1) and found to be signi cantly greater at P < 0.05 whereas analysis with a one-way ANOVA where each soil was treated separately did not detect differences. The null line indicates Jandaroi as the parental genotype for all lines.

Figure 6
Effect of TaALMT1 and TaMATE1B introgressed from bread wheat into durum on grain yields when grown on acid soils in the eld. Jandaroi-TaMATE1B and Jandaroi were grown on an acid site at Tarcutta (A) and grain yield data of genotypes compared with Student's t-test (n = 4, * P < 0.05). All lines were grown on acid sites at Gunning (B) and Merredin (C). A one-way ANOVA detected differences between genotypes at P < 0.05 for Gunning with different letters indicating signi cant differences (n = 4 except for TaMATE1B which had n = 3). No signi cant differences between genotypes were detected for Merredin at P < 0.05. For all experiments grey circles show individual values and error bars indicate the standard error of the mean.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. KawasakietalFigS1.docx