Introduction

Calcareous soils are alkaline and contain free CaCO3, and high concentrations of, calcium (Ca) and bicarbonate (HCO3−). The availability of phosphorus (P) and some micronutrients, including iron (Fe), zinc (Zn), manganese (Mn) and copper (Cu) in calcareous soils is very low (Tyler 2003). Most Lupinus species grow poorly on calcareous soils, and this has been attributed to various factors, including Ca toxicity, sensitivity to high [HCO3−] (HCO3− concentration) and high pH and Fe deficiency, and poor nodulation (Abd-Alla 1999; Ding et al. 2018b; Jessop et al. 1990; Kerley 2000; Tang et al. 1995b; Tang and Thomson 1996; White 1990).

Calcium toxicity has been considered one of the major causes for why some Lupinus species are sensitive to calcareous soils (De Silva and Mansfield 1994; Jessop et al. 1990; Kerley et al. 2001). For example, a high Ca supply may disturb the lamellar structure of chloroplasts, and consequently negatively affect net photosynthetic rates (Chevalier and Paris 1980; Chevalier and Paris-Pireyre 1984; De Silva et al. 1994). Excessive Ca may precipitate with P as Ca3(PO4)2 in plant tissues which makes both Ca and P unavailable (Ding et al. 2018b; McLaughlin and Wimmer 1999; Zohlen and Tyler 2004). We found that Ca-tolerant Lupinus species tended to have tight control over Ca uptake and/or translocation from roots to leaves, or better Ca compartmentation at the cellular level (Ding et al. 2018a). These abilities likely play an important role in the tolerance of some Lupinus species to calcareous soils, as suggested for other calcicole species (He et al. 2014; Jefferies and Willis 1964; Raza et al. 2000; Valentinuzzi et al. 2015; Webb 1999; Wu et al. 2011).

In addition, high [HCO3−] and high pH in calcareous soils are considered the main factors affecting the sensitivity of some Lupinus species to calcareous soils (Coulombe et al. 1984; Mengel et al. 1984; Romera et al. 1992; Tang and Thomson 1996; Waters and Troupe 2012). For example, a high rhizosphere pH caused by a high [HCO3−] in calcareous soils reduces nutrient availability, especially that of P, Fe, Mn, Zn, Cu and boron (B) for root uptake, and thus limits plant growth (George et al. 2012; Neumann and Römheld 2012; Parker et al. 1999; Tyler and Ström 1995; Yue Ao et al. 1987). In addition, a high [HCO3−] and high pH can also inhibit root growth of some Lupinus species (Peiter et al. 2001). Tang and Thomson (1996) found that root elongation of L. angustifolius is inhibited by high [HCO3−] and high pH in nutrient solution, and this was even observed in nutrient solution with pH ≥ 6 (Tang et al. 1992). Our comparative study showed that there is a correlation between the decreased lateral root growth and nutrient availability of some Lupinus species grown in nutrient solution with high [HCO3−] and/or high pH (Ding et al., personal observations).

In a series of hydroponic experiments, we found that a high Ca supply caused P deficiency in some Lupinus species, but its effects on the growth, especially the leaf symptoms of Lupinus species were inconsistent with what has been reported for those grown in calcareous soils (Ding et al. 2018b). In contrast, the effects of high pH (buffered by MES/TES or caused by high [HCO3−]) on different Lupinus species are generally consistent with those of plants grown in calcareous soils, for example, leaf chlorosis were developed in the calcifuge Lupinus species under high pH (Ding et al., personal observations). Considering the results of additional studies (Kerley and Huyghe 2002; White and Robson 1990), we concluded that pH is the main reason why some Lupinus species are sensitive to nutrient solutions mimicking calcareous soils. However, a hydroponic system is quite different from soil systems, and plants may respond differently to these two growing conditions. Therefore, in this study, we aimed to test if pH is the main factor determining why some Lupinus species are sensitive to calcareous soils, by growing different Lupinus species in soils with different forms of Ca, pH levels, and [HCO3−]. We compared leaf symptoms, root nodulation, root and shoot biomass, and root and shoot nutrient status. We hypothesised that pH, rather than Ca, is the main reason why Lupinus species respond differently to calcareous soils. In addition, we hypothesised that root surface area, root length, and fine root growth of calcifuge species will be inhibited significantly in calcareous soils, while those of calcicole species will be slightly inhibited or not inhibited at all.

Materials and methods

Plant growth

Four Lupinus species were selected (two calcifuge species: L. angustifolius L. cv. Mandelup and L. hispanicus ssp. bicolor Boiss. and Reut. P22999; two calcicole species: L. pilosus Murr. P27440 and L. cosentinii Guss. P27225) (Table 1). All seeds were obtained from the Australian Lupin Collection (Department of Primary Industries and Regional Development, WA, Australia). Except for L. angustifolius, seeds were scarified. Seeds were then sterilised in 5% (v/v) sodium hypochlorite for 20 min and rinsed thoroughly with deionised (DI) water. All seeds were pre-germinated in sterilised river sand for 4 days, and then three or four seeds were inoculated with Group G® (Bradyrhizobium sp. (Lupinus) WU425), which is effective for the inoculation of Lupinus species (Tang and Robson 1993), sown in prepared pots, and then later thinned to one plant. All sands were washed, sterilised and then dried first. After drying, CaCO3 or CaSO4 was mixed evenly with sand, as shown in Table 2. Each pot was then filled with 1.2 kg dry river sand.

Table 1 Lupinus species and genotypes used in this study, and the collection site soil pH
Full size table
Table 2 Calcium (Ca), exchangeable Ca2+, and pH of the different treatments used in the experiment
Full size table

All essential nutrients other than Ca were provided as: 189.5 mg kg−1 Ca(NO3)2.4H2O, 28.6 mg kg−1 NH4Cl, 65.9 mg kg−1 KH2PO4, 101.2 mg kg−1 MgSO4.7H2O, 12.3 mg kg−1 MnSO4.H2O, 8.8 mg kg−1 ZnSO4.7H2O, 0.7 mg kg−1 H3BO3, 1.0 mg kg−1 Na2MoO4, 2.0 mg kg−1 CuSO4.5H2O and 32.9 mg kg−1 FeNaEDTA. This nutrient composition has been used in our lab and proved to be suitable for the growth of legumes (Pang et al. 2010, 2011). The N was 80 mg kg−1, and this was to supply some N for young plants; when they grow bigger, they will need to rely on biological N2 fixation. Besides, and the amount of Ca(NO3)2 added to all the treatments was very small compared with the amount of CaSO4 or CaCO3 added, so it hardly affected the experiment design. All these nutrients were added as a nutrient solution with DI water, and KOH was used to adjust solution pH to 6 and 8 according to different treatments shown in Table 2. K2SO4 was used to balance [K], and the final [K] was 80 mg kg−1 for all the treatments, the final [S] was 133.4 mg kg−1 for control and pH 6, the final [S] was 40 mg kg−1 for high Ca and high pH + high Ca treatment. After adding basic nutrition solution for 1 week, all pots were ready to use, and sands from three plain pots of each treatment were air-dried and passed through a 2-mm sieve to measure the actual pH (10 g dry sand in 50 ml CaCl2 as shown in Table 2) and exchangeable Ca2+ of each treatment. Exchangeable Ca2+ was then extracted by 1 M ammonium acetate at pH 7 (Rayment and Lyons 2011) and analysed by inductively coupled plasma optical emission spectrometry (ICP-OES; School of Agriculture and Environment, University of Western Australia, Perth, Australia).

The experiment was run in a glasshouse (20 °C day/15 °C night). All pots were watered with DI water to weight, three times per week, to 75% field capacity.

Gas exchange and chlorophyll fluorescence measurements

The day before harvest, the youngest fully-expanded leaves were chosen to measure net photosynthesis rate (Amax), stomatal conductance (gs) and intercellular carbon dioxide concentration (Ci) with a LI-6400 portable gas exchange system (Li-Cor, Lincoln, NE, USA) at 1500 μmol quanta m−2 s−1. The same leaf used to measure photosynthesis was dark-adjusted in a dark room for 1 h, and then the maximum photochemical quantum yield of PSII (Fv/Fm) was measured using a LI-6400 portable gas exchange system as described above.

Plant harvest and carboxylate extraction

All the pots were watered before harvest to make sure the roots were easily removed and the root damage was minimal. The intact plants were removed from the pots, and the bulk sand around the roots was removed gently. The roots were then placed into a 100- or 150-ml beaker, and a certain amount of CaCl2 (0.2 mM), ranging from 25 to 100 ml, depending on root size, was added to the beaker to remove the rhizosphere sand. After removing the roots, the beaker was shaken by hand, and the pH of the extract was measured by a portable pH meter (pH 300/310, Eutech Instruments Pte Ltd., Singapore). A subsample of the extract was filtered through a 0.2 μm syringe filter into a 1-ml HPLC vial and acidified with one drop of concentrated phosphoric acid. The sample was then stored at −20 °C until HPLC analysis. The whole process for each plant was about 5 min.

Nodulation was assessed according to the British Columbia Ministry of Forests field guide (British Columbia Ministry of Forests 1991), where a ranking of 0–5 is given to each of five categories, including plant growth and vigour (1 for very chlorotic plants, 2 for slightly chlorotic plants, 3 for green and relatively small plants and 5 for green and vigorous plants); nodule number (0 for no nodules, 2 for few nodules (<5) with slight pigmentation, 3 for few nodules (<5) with pink pigmentation, 3 also for many nodules (>50) mainly white or smaller, 5 for many nodules (>50) with pink pigmentation, 5 for 5–50 nodules as well); nodule position (1 for nodules on lateral roots only, 3 for nodules on both crown and lateral roots, 5 for nodules predominantly on the crown); nodule colour (0 for white or green colour, 2 for some pink or whitish with green areas, 5 for predominantly pink); and nodule appearance (0 for ineffective, 3 for intermediate, 5 for effective). A total score was summed to evaluate the effectiveness of nodulation, and a score between 20 and 25 is considered very effective, a score between 15 and 20 less effective, while a score between 0 and 14 is considered unsatisfactory nodulation.

The whole roots were then thoroughly washed to remove the remaining sand. After that, plants were separated into mature leaves, immature leaves, stems (including petioles), and roots. Leaf area was measured and fresh biomass was recorded. Leaf symptoms were recorded by taking photos prior to harvest. Roots of all plants were scanned by an Epson Perfection V700 Photo Scanner and analysed by WinRHIZO 2009a, b (Regents Instruments Canada lnc, Quebec City, Quebec, QC, Canada). After that, all the plant parts were dried in an oven at 70 °C for 7 days, and dry biomass was recorded.

Carboxylate analysis

The pre-frozen extracts were brought back to room temperature, and a 100-μl sample of each extract was analysed by HPLC with a 600E pump, 717 plus autosampler and a 996 photodiode array detector (Waters, Milford, MA, USA). A reversed-phase column liquid chromatography (RPLC) was used to separate and quantify organic acids as described by Cawthray (2003). The carboxylic acid working standards included acetic, citric, cis-aconitic, fumaric, lactic, malic, malonic, maleic, succinic and trans-aconitic acids.

Determination of whole leaf and root nutrient concentrations

To determine whole leaf and root nutrient concentrations, all the dry mature leaves and roots were ground to a fine powder with a GenoGrinder (SPEX SamplePrep LLC, Metuchen, New Jersey, USA), and 0.1 g of ground material was digested with concentrated HNO3 and HClO4 (3:1) and analysed by inductively coupled plasma optical emission spectrometry (ICP-OES). Nitrogen was determined by a CNS analyser (Elementar Vario Macro Elemental Analyzer, Elementar Analysensytem GmbH, Hanau, Germany).

Statistics

Data and statistical analyses were performed with the R software platform (R Core Team 2017). As data of L. hispanicus grown under high pH + high Ca treatment were not included, data structure of the whole experiment was unbalanced. General linear mixed-effect models were used to test the differences in whole leaf and root nutrient concentrations, root surface area, average root diameter, total root length, root nodulation, carboxylates, leaf area, shoot biomass, root biomass and total biomass among different treatments within each species. We tested the differences of gas exchange parameters and chlorophyll fluorescence with general linear mixed-effect models among different treatments within each species; individual plant was considered as the random effect.

The residuals of each model were inspected for heteroscedasticity, and in the presence of heteroscedasticity, models with different variance structures were compared, and a suitable variance structure significantly improved the model were specified based on Akaike Information Criterion (AIC) (Burnham and Anderson 2003; Zuur et al. 2009). The effects package was used to determine the mean values and 95% confidence intervals (Cl) (Fox 2003). Significant differences were defined based on Tukey’s post-hoc analysis (P < 0.05).

Results

Leaf symptoms

Leaf chlorosis was observed in the youngest leaves under high pH and high pH + high Ca for all Lupinus species from the appearance of seedlings. However, after 2 weeks, the leaf chlorosis symptoms of all Lupinus species under high pH started to disappear, with calcicole species fully recovering, and calcifuge species partly recovering (Fig. 1). The leaf chlorosis symptoms of calcicole species under high pH + high Ca partly decreased as well, while those of calcifuge species did not decline at all (Fig. 1).

Fig. 1
figure 1

Leaf symptoms of four Lupinus species 1 day prior to harvest when grown under different pH and calcium (Ca) treatments. Scale bar at the bottom of each photo is 1 cm. marks calcifuge species, marks calcicole species

Full size image

Leaf nutrients

Compared with control plants, leaf N concentrations of L. hispanicus under either high Ca or high pH, and those of L. pilosus and L. cosentinii under high pH and high pH + high Ca were significantly lower. However, leaf N concentration of L. angustifolius under high Ca was the lowest among all treatments, and there were no significant differences among the other three treatments (Fig. 2a).

Fig. 2
figure 2

Concentrations of a range of nutrients in leaves of four Lupinus species when grown under different pH and calcium (Ca) treatments. No data are shown for L. hispanicus grown under high pH + high Ca, as they died in this treatment. Error bars represent 95% confidence intervals (Cl). Letters show significant differences of different treatments within each species (based on Tukey’s post-hoc analysis, P < 0.05). The grey dashed line represents the corresponding nutrient concentration in plant shoot dry matter considered adequate for crop growth (Kirkby 2012). marks calcifuge species, marks calcicole species

Full size image

Leaf P concentrations of L. angustifolius and L. pilosus under high Ca, high pH, and high pH + high Ca treatments and those of L. hispanicus under either high Ca or high pH were significantly lower than those of control plants. For L. cosentinii, only the leaf P concentration under the high pH + high Ca treatment was significantly lower than that under other treatments (Fig. 2b).

Leaf Ca concentrations of all Lupinus species under high Ca, high pH, and high pH + high Ca treatments were significantly higher than those of control plants. There were no significant differences for leaf Ca concentrations of L. angustifolius, L. cosentinii, and L. pilosus between these three treatments. However, the leaf Ca concentration of L. hispanicus under high pH was significantly lower than that under high Ca (Fig. 2c).

Leaf Mg concentrations of L. angustifolius and L. pilosus were significantly lower under high pH and high pH + high Ca treatments than those of control plants and plants of high Ca treatment. However, leaf Mg concentrations of L. hispanicus under high Ca were the lowest among all the treatments. There was no significant difference for L. cosentinii between different treatments (Fig. 2d).

Leaf Mn and Fe concentrations of all Lupinus species were significantly lower under high pH and/or high pH + high Ca treatments than those of control plants and plants of high Ca treatment (Fig 2e and f).

Leaf potassium (K), sulfur (S), zinc and copper concentration are shown in Fig. S1.

Root nutrients

Compared with control plants, root N concentrations of L. angustifolius, L. hispanicus and L. pilosus at high Ca supply were significantly lower. Root N concentrations of L. hispanicus, L. pilosus and L. cosentinii under high pH and high pH + high Ca treatments were significantly lower than those of control plants (Fig. 3a).

Fig. 3
figure 3

Concentrations of a range of nutrients in roots of four Lupinus species when grown under different pH and calcium (Ca) treatments. No data are shown for L. hispanicus grown under high pH + high Ca, as they died in this treatment. Error bars represent 95% confidence intervals (Cl). Letters show significant differences of different treatments within each species (based on Tukey’s post-hoc analysis, P < 0.05). marks calcifuge species, marks calcicole species

Full size image

Root P concentration of L. hispanicus under high Ca supply was the lowest among all treatments, whereas the P concentration of L. pilosus under high pH was the lowest among all treatments. There were no significant differences for the root P concentrations of L. angustifolius and L. cosentinii among all treatments (Fig. 3b).

Compared with control plants, root Ca concentrations of all Lupinus species in the treatments were significantly higher, except for L. hispanicus under a high Ca supply. There were no significant differences for either L. angustifolius or L. pilosus between these three treatments. However, the root Ca concentrations of L. cosentinii under high pH + high Ca treatment was the highest among all treatments (Fig. 3c).

Compared with control plants, root Mg concentrations of all Lupinus species at high pH and/or high pH + high Ca treatments were significantly higher. The root Mg concentration of L. angustifolius under high Ca supply was also significantly higher than that of control plants. However, the root Mg concentration of L. cosentinii under high Ca supply was significantly lower than that of the control. There were no significant differences in root Mg concentrations of L. hispanicus and L. pilosus between control and the high-Ca treatments (Fig. 3d).

Root Mn concentration of L. angustifolius at a high Ca supply was the highest among all treatments, while that of L. cosentinii under a high Ca supply was the lowest among all treatments. Root Mn concentrations of L. hispanicus under high pH were significantly higher than those of the control and the high-Ca treatments, whereas root Mn concentrations of L. pilosus under high pH and high pH + high Ca treatments were significantly lower than those of the control and the high-Ca treatments (Fig. 3e).

Root Fe concentrations of L. angustifolius and L. cosentinii under high pH and high pH + high Ca treatments were significantly lower than those of the control and plants of the high-Ca treatments. Root Fe concentration of L. hispanicus under high pH was significantly lower than that of the control. Root Fe concentration of L. pilosus under high Ca was the highest, while that under high pH + high Ca was the lowest among all treatments (Fig. 3f).

Root potassium (K), sulfur (S), zinc and copper concentration are shown in Fig. S2.

Gas exchange parameters

Net photosynthetic rate (Amax), stomatal conductance (gs), and the maximum photochemical quantum yield of PSII (Fv/Fm) of L. angustifolius under high pH + high Ca were the smallest among all treatments, while the intercellular carbon dioxide concentration (Ci) was the largest among all treatments. There were no significant differences in these values among the other three treatments (Fig. 4).

Fig. 4
figure 4

Net photosynthetic rate (Amax) (a), stomatal conductance (gs) (b), intercellular carbon dioxide concentration (Ci) (c) and the maximum photochemical quantum yield of PSII (Fv/Fm) (d) of four Lupinus species when grown under different pH and calcium (Ca) treatments. No data are shown for L. hispanicus grown under high pH + high Ca, as they died in this treatment. Error bars represent 95% confidence intervals (Cl). Letters show significant differences of different treatments within each species (based on Tukey’s post-hoc analysis, P < 0.05). marks calcifuge species, marks calcicole species

Full size image

Lupinus hispanicus had the smallest Amax,gs and Ci under high pH, while the largest Fv/Fm was under high pH. There were no significant differences in these values between control and high-Ca treatments (Fig. 4).

The Amax, gs and Ci of L. pilosus under high pH + high Ca were the smallest among all treatments, while the Amax under high pH was the largest. There were no significant differences for Fv/Fm between different treatments (Fig. 4).

The Amax,gs and Fv/Fm of L. cosentinii under high pH and high pH + high Ca were significantly smaller than those under control or high Ca, while there were no significant differences for Ci of L. cosentinii between different treatments (Fig. 4).

Root morphology

Root surface area of L. angustifolius under high pH and high pH + high Ca was significantly less than that of control plants and plants in the high-Ca treatment. Root surface area of L. pilosus and L. cosentinii under high pH + high Ca was the smallest among all treatments. There was no significant difference for the root surface area of L. hispanicus among treatments (Fig. 5a).

Fig. 5
figure 5

Root surface area (a), average root diameter (b), total root length (c) and root nodulation score (d) of four Lupinus species when grown under different pH and calcium (Ca) treatments. The score above the upper grey dashed line in d represents nodulation considered as very effective, the score between upper and lower grey dashed line in d represents nodulation considered as less effective, while the score below the lower grey dashed line in d is considered as ineffective nodulation. Nodulation was assessed according to the British Columbia Ministry of Forests field guide (BCMF 1991); details are included in Materials and Methods (British Columbia Ministry of Forests 1991). No data are shown for L. hispanicus grown under high pH + high Ca, as they died in this treatment. Error bars represent 95% confidence intervals (Cl). Letters show significant differences of different treatments within each species (based on Tukey’s post-hoc analysis, P < 0.05). marks calcifuge species, marks calcicole species

Full size image

Average root diameter of all Lupinus species under high pH and high pH + high Ca treatments was significantly greater than that under control or high Ca. Average root diameter of L. angustifolius under a high Ca supply was significantly greater than that of control plants (Fig. 5b).

Root length of L. angustifolius and L. cosentinii under high pH and high pH + high Ca was significantly less than that of control and high-Ca plants. Root length of L. pilosus under high pH and high pH + high Ca was significantly less than that of high-Ca plants. Root length of L. angustifolius, L. pilosus and L. cosentinii under high pH + high Ca was even less than that under high pH. However, there were no significant differences in root length of L. hispanicus among treatments (Fig. 5c).

Root nodulation of L. angustifolius was only effective under control conditions. The root nodulation of L. hispanicus was effective under control conditions and at high pH, while that under high pH was less effective. Root nodulation of L. pilosus and L. cosentinii under high pH + high Ca was not effective. Root nodulation was more effective for L. pilosus under high Ca and high pH than under control conditions, while there was no significant difference for L. cosentinii under these three treatments (Fig. 5d).

Carboxylates

The amounts of rhizosphere citrate for all Lupinus species under high Ca, high pH, and high pH + high Ca were significantly greater than those of control plants (Fig. 6a). The rhizosphere malate amounts in L. angustifolius and L. pilosus under high pH were the greatest among all the treatments, while there were no significant differences for L. hispanicus and L. cosentinii among treatments (Fig. 6b). Rhizosphere fumarate amounts of all Lupinus species showed no significant difference among treatments (Fig. 6c). Rhizosphere cis-aconitate amounts in L. angustifolius and L. pilosus under high pH were the greatest among all treatments, while we found no significant difference for L. cosentinii among treatments. No cis-aconitate was detected for L. hispanicus under any treatment (Fig. 6d).

Fig. 6
figure 6

Amounts of rhizosphere citrate (a), malate (b), fumarate (c) and cis-aconitate (d) acid of four Lupinus species when grown under different pH and calcium (Ca) treatments. Rhizosphere carboxylates were extracted with 0.2 mM CaCl2 and amounts are expressed per unit total root surface area. No data are shown for L. hispanicus grown under high pH + high Ca, as they died in this treatment. Error bars represent 95% confidence intervals (Cl). Letters show significant differences of different treatments within each species (based on Tukey’s post-hoc analysis, P < 0.05). marks calcifuge species, marks calcicole species

Full size image

Biomass

The total, root and shoot biomass, and leaf area of L. angustifolius under high pH and high pH + high Ca were significantly less than those under control conditions and high Ca treatments. The total biomass, root biomass, and shoot biomass of L. hispanicus under high pH + high Ca were the lowest among all treatments, and there was no significant difference among control, high Ca, and high pH treatments. However, the leaf area of L. hispanicus under high pH was significantly less than that of control plants. The total biomass, shoot biomass and leaf area of L. pilosus were the greatest under high-Ca conditions among all treatments, while the root biomass was greatest under high pH. There were no significant differences for the total and root biomass of L. cosentinii among treatments. However, the shoot biomass and leaf area of L. cosentinii under high pH + high Ca were significantly less than under control conditions (Fig. 7).

Fig. 7
figure 7

Total (a), root (b), shoot (c) dry biomass and leaf area (d) of four Lupinus species when grown under different pH and calcium (Ca) treatments. No data are shown for the leaf area of L. hispanicus grown under high pH + high Ca, as they died in this treatment. Error bars represent 95% confidence intervals (Cl). Letters show significant differences of different treatments within each species (based on Tukey’s post-hoc analysis, P < 0.05). marks calcifuge species, marks calcicole species

Full size image

A detailed shoot biomass, including mature leaf, immature leaf and stem biomass of different Lupinus species, are shown in Fig. S3.

Discussion

The present study demonstrates that Lupinus species responded very differently to high pH and high pH + high Ca treatments, with the calcicole species L. pilosus being the most tolerant, followed by the calcicole species L. cosentinii. The other two calcifuge species, L. angustifolius and L. hispanicus, were sensitive to high pH and high pH + high Ca treatments, and L. hispanicus even died in the high pH + high Ca treatment. “Tolerant” means the species can grow healthily or the growth is only slightly inhibited in calcareous soils, and “sensitive” means the growth of the species can be inhibited from an intermediate to the worst level. In this study, tolerance was mainly based on the biomass data. This is consistent with results on these species when grown in nutrient solution with high pH (buffered by MES/TES or caused by high [HCO3−]) (Ding et al., personal observations), some field and glasshouse studies, and their natural occurrence on acid or alkaline soils (Brand et al. 2002; Clements and Cowling 1990; Tang et al. 1993a, 1995b; White 1990). However, this result is inconsistent with the Ca sensitivity of different Lupinus species grown in a hydroponic system, where L. pilosus and L. angustifolius were tolerant of high Ca, whereas L. cosentinii and L. hispanicus were sensitive (Ding et al. 2018b). This discrepancy might be related to different growth systems (i.e. sand vs hydroponics) or different [Ca2+] used in various experiments (4.8 g kg−1 vs 6 mM). Most likely, Ca is not the main reason why some Lupinus species are sensitive to calcareous soils (Tang et al. 1995a). This also explains why there was no correlation between Ca concentration in Lupinus species grown in Wangary calcareous soil and their leaf chlorosis or growth (Brand et al. 2000).

Leaf symptoms

The pH of the high pH and high pH + high Ca treatments was similar. Therefore, the difference in leaf chlorosis and its gradual disappearance with time among Lupinus species under these two treatments is likely explained by different [Ca2+], [HCO3−] and pH buffer capacity associated with the different [CaCO3]. In addition, high [Ca2+] showed no relationship with leaf chlorosis in this study. Therefore, we conclude that the gradual disappearance of the leaf chlorosis symptoms was due to a lower [HCO3−].

Similarly, Brand et al. (1999) also found that leaf chlorosis of L. angustifolius correlated with CaCO3 content (which can react with CO2 and H2O to produce bicarbonate) in a range of calcareous soils; the pH of these calcareous soils was similar. Except for leaf chlorosis, the negative effects of high pH + high Ca treatment on the growth, as shown by the biomass, of all Lupinus species were generally consistent with those observed under high pH, rather than with those observed under a high Ca treatment. The effects of high pH (buffered by MES/TES or caused by high [HCO3−]) on the growth of Lupinus species were consistent in a hydroponic study as well (Ding et al., personal observations). MES/TES or bicarbonate rather than CaCO3 were used in the hydroponic study to maintain a stable pH is because CaCO3 is not soluble in a hydroponic system. Taken together, this indicates that high pH is the actual cause why some Lupinus species are sensitive to calcareous soils, while the strong buffering capacity of HCO3− determines if Lupinus species can recover from leaf chlorosis and then survive in calcareous soils or not. As the strong buffering capacity of bicarbonate could restrict rhizosphere acidification of carboxylic acids, and then limit nutrient availability. This also explains why active lime (high Ca, high HCO3− and high pH) tends to have greater negative effects on the growth of some calcifuge Lupinus species than alkalinity (high pH) (Liu and Tang 1999).

Plant growth

Among all the species, the greatest tolerance of L. pilosus to high pH and high pH + high Ca treatments is evidenced by the biomass under those treatments. In addition, the shoot and total biomass of L. pilosus under a high Ca supply was actually significantly greater than that of control plants. This confirms our finding that Ca improves the shoot growth of species that grow well under an extremely high Ca concentration (Ding et al. 2018a). The negative effects of high pH and/or high pH + high Ca on total root length of Lupinus species agree with the decreased lateral root growth of some Lupinus species grown in either a high pH (buffered by MES/TES or caused by high [HCO3−]) nutrient solution (Ding et al., personal observations) or limed soil (Kerley and Huyghe 2001). This is also consistent with the decreased root length of L. angustifolius grown in nutrient solution with high pH (>6) (Tang et al. 1992, 1993b). Thus, we can conclude that high pH is the direct reason why root growth was inhibited.

Root nodulation of all Lupinus species was not effective under high pH + high Ca treatment, as evidenced by nodulation scores. However, nodulation of Lupinus species responded differently to high pH treatment, with that in L. angustifolius being ineffective under high pH, while that of calcicole species not being negatively affected by high pH. The nodule numbers of L. angustifolius, L. albus and L. pilosus were inhibited when grown in nutrient solution with high pH (buffered by MES/TES or caused by high [HCO3−]) (Tang and Robson 1993; Tang and Thomson 1996). This is inconsistent with what we found; however, as pointed out above, high pH + high Ca treatment contains larger [HCO3−] and [Ca2+] than the high pH treatment. Therefore, the buffering capacity of HCO3−, that is the ability to maintain a high pH, was closely related with root nodulation of the calcicole species, while for the calcifuge species, either high [Ca] or high [HCO3−] inhibited root nodulation.

Carboxylates

The exudation of citrate increased in the high pH + high Ca treatment, and the amount of citrate was much greater than that of the other carboxylates under any treatment. The relatively large amounts of malate and cis-aconitate in the rhizosphere of L. angustifolius and L. pilosus under high-pH treatment compared with those under high pH + high Ca treatment partly explain why leaf chlorosis symptoms under a high-pH treatment decreased over time more than they did under a high pH + high Ca treatment. This is because increased carboxylate release would decrease rhizosphere pH, subsequently increasing the availability of P and micronutrients such as Fe, Mn and Zn (Dinkelaker et al. 1989; Lambers et al. 2013; Liang and Li 2003). However, the extent of rhizosphere acidification would be limited by the strong buffering capacity of HCO3− under a high pH + high Ca treatment. The extent of rhizosphere acidification is also shown by the rhizosphere pH (Table S1).

Leaf and root nutrient concentrations

As discussed above, root surface area, root length, and fine root growth of L. cosentinii, L. angustifolius and L. hispanicus were significantly reduced at high pH and/or a high pH + high Ca treatment, which would have resulted in less nutrient uptake, including Fe. In contrast, a high pH did not inhibit Fe uptake of L. pilosus, as the root surface area and total root length of L. pilosus under high pH were not affected. However, translocation of Fe and Mn from root to shoot in all Lupinus species was inhibited by high pH and high pH + high Ca treatments. Probably because Fe3+- and/or Mn4+-reductase activity was restricted by a high pH in the root apoplast which resulted in reduced Fe and Mn availability and translocation to leaves (Kosegarten and Koyro 2001; Mengel 1994; Millaleo et al. 2010; Rengel 2000; Zribi and Gharsalli 2002). Leaf Fe concentrations in these two treatments were even below the concentration in shoot dry matter considered adequate for crop growth (Kirkby 2012). This is consistent with the observed leaf chlorosis in this study, and also agrees with previous findings (Bertoni et al. 1992; White and Robson 1989).

Based on the comparison of leaf and root P concentrations of different treatments, we found the P translocation of calcifuge species from roots to shoots was inhibited by either high soil [Ca] or high pH, and this was similar for L. cosentinii, while P translocation in L. pilosus was only negatively affected by high pH + high Ca treatment. This shows only under a high pH + high Ca treatment, the strong buffering capacity of HCO3− could limit the extent of rhizosphere carboxylate acidification in L. pilosus and then result in less P availability and translocation from roots to shoots. This also explains why, compared with other species in this study, L. pilosus was more tolerant to calcareous soils.

Nitrogen concentration and content were clearly decreased in L. angustifolius, L. albus and L. pilosus grown in solution with high pH (buffered with MES and TES), especially L. angustifolius (Tang and Robson 1993). However, in the present study, leaf N concentrations of L. angustifolius were not affected by high pH and high pH + high Ca treatments. The reason for this disagreement is not yet clear; further research is needed to explain this. In addition, leaf N concentration was not consistent with root nodulation. These results suggest that the role of nodulation (rather than N2 fixation) related with Bradyrhizobium sp. (Lupinus) WU425 in the growth of Lupinus species in calcareous soils is not as important as that reported before (Tang and Robson 1995).

Gas exchange

The decreased Amax of L. hispanicus under high pH and L. pilosus under high pH + high Ca treatment was due to reduced stomatal conductance, as it was associated with a corresponding decrease of gs and Ci. The Amax and Fv/Fm of L. angustifolius under the high pH + high Ca treatment and of L. cosentinii under the high pH and high pH + high Ca treatments were significantly lower than those of plants under other treatments. In addition, the change of Amax and Fv/Fm of L. angustifolius under high pH and L. cosentinii under high pH + high Ca treatments was correlated with leaf Fe concentration. This indicates that the decreased PSII photochemical capacity of L. angustifolius under high pH + high Ca treatment and that of L. cosentinii under high pH and high pH + high Ca treatments was likely due to Fe deficiency caused by high [HCO3−] and/or high pH (Maxwell and Johnson 2000).

Conclusions

A high pH (resulting from either KOH or HCO3−) inhibited root growth of Lupinus species which resulted in less nutrient uptake (except N) and reduced shoot growth. However, the strong buffering capacity of bicarbonate determines if Lupinus species can survive in calcareous soils or not. Among all studied Lupinus species, L. pilosus was the most tolerant, followed by L. cosentinii and L. angustifolius, while L. hispanicus was the most sensitive.