Introduction

The calcifuge vs calcicole syndrome has been known for many decades (Tansley 1917) and explained in term of pH or pH-buffering capacity. Some plant species avoid calcareous soils (calcifuges) as opposed to preferring such soils (calcicoles), whereas some are soil-indifferent (Bugbee and Salisbury 1985; De Silva 1934; Zohlen and Tyler 2000). Many Lupinus species prefer acid soils rather than alkaline, calcareous soils (White 1990). This is remarkable, since they have the capacity to release carboxylic acids and strongly acidify their rhizosphere (Lambers et al. 2013; Neumann et al. 2000). High pH, and high bicarbonate (HCO3−) and calcium (Ca) availability are characteristics of calcareous soils, and are presumably the main factors explaining why some Lupinus species grow poorly on calcareous soils (Jessop et al. 1990; Tang and Robson 1995; Tang et al. 1995; Tang and Thomson 1996; White 1990). Some calcifuge Lupinus species show signs of Ca toxicity when grown at a high Ca supply (De Silva et al. 1994; Ding et al. 2018b; Jessop et al. 1990). Banksia and Hakea species (Proteaceae) are particularly sensitive to high Ca supply, when combined with a high phosphorus (P) supply (Nichols and Beardsell 1981). This is associated with the cellular allocation of P and Ca in their leaves. However, in our previous study, we found that a high Ca supply only partly explains the calcifuge habit of some Lupinus species. Therefore, other factors, such as high pH and high [HCO3−], need to be explored (Ding et al. 2018b).

Soil solution pH influences nutrient availability for root uptake (Neumann and Römheld 2012; White and Broadley 2009). At high pH (>7), the availability of P, iron (Fe), manganese (Mn), zinc (Zn), copper (Cu) and boron (B) is very low. As the pH decreases, the availability of Fe, Mn, Zn, B and aluminium (Al) increases due to increased solubility, desorption, or reduction (George et al. 2012; Parker et al. 1999). However, if it is too acidic, P can be sorbed onto soil particles, and mobile cations, such as potassium (K), Ca, and magnesium (Mg) can be leached as well (Lambers et al. 2008). A low nutrient availability related with a high pH has been the classic explanation why soil pH is a limiting factor for calcifuge species, explaining why they are sensitive to calcareous soils (Tyler and Ström 1995).

High [HCO3−], which ranges from 0.4 to 3 mM in calcareous soil solutions depending on soil water content (Mengel et al. 1984), can raise the soil pH. It also acts as a strong pH-buffer, and causes leaf chlorosis, both in calcareous soils (Coulombe et al. 1984b; Mengel et al. 1984) and in nutrient solution (Coulombe et al. 1984a; Romera et al. 1992; Tang and Thomson 1996; Waters and Troupe 2012). Bicarbonate can either stimulate or decrease root ferric chelate reductase (FCR) activity, depending on Fe supply (Hsieh and Waters 2016). It also inhibits Fe absorption and translocation to leaves (Bertoni et al. 1992; Fleming et al. 1984; Karimi and Tari 2017; Lucena et al. 2007; Romera et al. 1991; Rutland and Bukovac 1971) and causes Fe immobilisation in leaves (Kosegarten et al. 1999; Mengel et al. 1994). Similarly, high [HCO3−] can affect the availability and uptake of Mn, Cu and Zn (Coulombe et al. 1984a; Dogar and Van Hai 1980; Yue Ao et al. 1987).

High [HCO3−] not only negatively affects nutrient status, but also inhibits root growth of calcifuge species (Lee and Woolhouse 1969; Peiter et al. 2001; Tang et al. 1993b). In a hydroponic experiment, Tang and Thomson (1996) found that high pH and [HCO3−] decreases root elongation of a calcifuge Lupinus species, L. angustifolius, and this response was even observed after exposing it to nutrient solution pH ( ≥ 6) for one hour (Tang et al. 1992). However, the most pH-tolerant Lupinus species, L. pilosus, produces more cluster roots at high [HCO3−] (Tang et al. 1996) which release carboxylates and allow L. pilosus to access P and other nutrients that are otherwise poorly available at high pH (Dinkelaker et al. 1989; Lambers et al. 2013). It is obvious that there is a relationship between root morphology and nutrient status in plants when they grow in calcareous soils or nutrient solution high in [HCO3−].

There have been a number of studies investigating the effect of high [HCO3−] and/or high pH on the growth of Lupinus species (Brand et al. 2000; Kerley and Huyghe 2002; Tang and Robson 1993; White and Robson 1989a). However, it is difficult to separate the effects of high [HCO3−] from those of high pH, and then distinguish the effect of high [HCO3−] itself from that of high pH. This is because high pH will inevitably increase [HCO3−]. The best way to distinguish the difference of high [HCO3−] itself and that of high pH is by comparing differences between high [HCO3−] together with high pH and high pH alone. Those studies investigating the effects of high [HCO3−] and/or high pH alone focused on root growth or leaf Fe deficiency, and the causal relationship of root growth and nutrient deficiency under high pH is not clear. More importantly, other nutrients in Lupinus species were not assessed. Our early study showed that Ca supply has no effect on the leaf cell types where P and Ca are allocated (Ding et al. 2018b), but we do not know if high [HCO3−] and/or high pH can change these allocation patterns. Therefore, the aims of this study were to examine the effects of high [HCO3−] and/or high pH on root growth, nutrient availability, P- and Ca-allocation patterns, and their relationship in Lupinus species. We hypothesised that nutrient uptake of calcicole L. pilosus and calcifuge L. angustifolius and L. cosentinii will be altered differently by pH increase, giving useful insights on chlorosis tolerance mechanism. We also hypothesised that the pH increase resulting from high [HCO3−], rather than HCO3− itself, is the main limiting factor regarding the calcifuge habitat of some Lupinus species. In addition, we hypothesised that any nutrient deficiencies in plants grown at high pH will be as a result of impaired nutrient uptake, rather than the effects of high pH on nutrient availability.

Materials and methods

Plant growth

Three Lupinus species (six genotypes, L. angustifolius cv L. Mandelup, L. angustifolius L. P25741, L. angustifolius L. P26723, L. angustifolius L. P22721, L. pilosus Murr. P27440 and L. cosentinii Guss. P27225) (Table 1) were chosen. All seeds were obtained from the Australian Lupin Collection (Department of Primary Industries and Regional Development, Western Australia). The seeds of L. pilosus and L. cosentinii were scarified and then all seeds were sterilised in 5% (v/v) sodium hypochlorite for 20 mins and rinsed with deionised (DI) water at least three times. All seeds were then soaked in deionised (DI) water overnight and sown in pots filled with sterilised river sand in a glasshouse (20 °C/15 °C day/night) to germinate. Ten days later, seedlings of similar size were chosen and washed free of sand. Each seedling was fixed in the centre of a foam lid and two plants were immersed in a 4 L black plastic pot with continuously aerated nutrient solution of the following composition: 200 μM KNO3, 10 μM KH2PO4, 100 μM CaCl2, 54 μM MgSO4, 0.24 μM MnSO4, 0.1 μM ZnSO4, 2.4 μM H3BO3, 0.03 μM Na2MoO4, 0.018 μM CuSO4 and 10 μM Fe-Na. EDTA (Ding et al. 2018a, b). After two weeks of acclimation in the glasshouse, all plants were placed under treatment. The four treatments were, pH 5, 6.5, 8a (high pH, low [HCO3−], around 0.52 mM, as calculated by Geochem-EZ (Shaff et al. 2010)), and 8b (high pH, high [HCO3−], around 1.4 mM, as calculated by Geochem-EZ). Treatments pH 5, 6.5 and 8a were adjusted by KOH and buffered with MES and TES (0.5 mM each). Treatment pH 8b was adjusted by KHCO3 with no MES or TES added, and the resulting [HCO3−] in the nutrient solution was 1.5 mM. The ion free activities of different treatments predicted by GeoChem-EZ are shown in Table S1. The average daily pH of the nutrient solution in which the six Lupinus genotypes were grown including manual pH adjustments were recorded (Fig. S1). In total, there were 6 genotypes (3 species) × 3 replicates (3 pots) × 2 plants in each pot × 4 pH treatment =144 plants. The pH adjusted by KOH (5, 6.5 and 8a) was adjusted three times daily, early morning, noon and night. When the nutrient solutions were changed, [K+] in all the treatments were balanced to be 1.71 mM. Between each nutrient solution change, [K+] was not balanced, as the amount of KOH added to adjust the pH was hard to calculate. The pH of treatment pH 8b was stable during the experiment, so there was no need to adjust it. The nutrient solution was changed three times per week to minimise depletion of nutrients and bicarbonate accumulation. Pots were placed in a temperature-controlled, root-cooling tank maintained at 18 °C.

Table 1 Lupinus species and genotypes used in this study with collection site soil pH and origin, their calcicole and calcifuge habits are based on their natural occurrence and other studies (Brand et al. 1999; Gladstones 1974; Kerley 2000; White 1990)
Full size table

Gas exchange and chlorophyll fluorescence measurements

Net photosynthesis rate (Amax) and stomatal conductance (gs) were measured on the youngest fully-expanded leaves using a LI-6400 portable gas exchange system (Li-Cor, Lincoln, NE, USA) at 1500 μmol quanta m−2 s−1. Stomatal limitation (Ls) was calculated according to: Ls=1\( -\frac{C_i}{C_a} \) (Jones 1985). The same leaf used to measure photosynthesis was dark-adjusted in a leaf clip for 30 min, and then the maximum fluorescence yield (Fm) and the minimum fluorescence yield (Fo) were measured by a Hansatech Plant Efficiency Analyser (PEA, King’s Lynn, Norfolk, UK). The maximum photochemical quantum yield of PSII (Fv/Fm) was calculated according to Fv/Fm = (Fm– Fo)/Fm (Maxwell and Johnson 2000).

Net P- and Ca-uptake rate

Four individual plants from all genotypes under each treatment were placed in a basal nutrient solution with no P or Ca added for >16 h (overnight) prior to measurement. The next day, the root surfaces of each plant were carefully blotted with paper towels and placed in a separate fresh 1 L nutrient solution with either 10 μM P or 0.1 mM Ca; the pH of the nutrient solution was kept the same in each treatment. A sample of solution (1 mL for P and 5 mL for Ca) was withdrawn every 30 mins for P, and every hour for Ca to determine the uptake rate, over a four-hour period. The solution was continuously aerated throughout the whole experiment. Phosphorus concentration in the solution was measured by the malachite green method (Motomizu et al. 1983), and [Ca] was measured by inductively coupled plasma optical emission spectrometry (ICP-OES; School of Agriculture and Environment, University of Western Australia, Perth, Australia). The [P] and [Ca] were then used to calculate the P- and Ca-depletion rates and thus the net P- and Ca-uptake rates per unit root dry weight for each plant.

Plant harvest

After five weeks of treatment, plants were carefully and thoroughly washed with deionised (DI) water. Plants were then separated into mature leaves (totally expanded leaves), immature leaves (partly expanded leaves), stems (including petioles) roots (including non-cluster and cluster roots). Cluster roots, also called proteoid roots, comprise many lateral rootlets (0.5–35 mm long) along a root axis of 0.5–4 cm (Dinkelaker et al. 1995; Shane and Lambers 2005). Leaf area was measured by a Li-Cor Ll- 3000 leaf area meter with a LI-3050A transparent belt conveyor (Li-Cor, Lincoln, NE, USA). Dry weights of all the plant parts were recorded after drying them in an oven at 70 °C for one week. Leaf symptoms and root visual appearance were recorded by taking photos prior to harvest.

Whole leaf nutrient concentrations, leaf + root nutrient content and root-to-leaf nutrient translocation ratio

Dry leaves and roots were ground into a fine powder using plastic vials and ceramic beads in a GenoGrinder vertical ball mill (SPEX SamplePrep LLC, Metuchen, New Jersey, USA). A subsample of 100 mg dried ground material was then digested with concentrated HNO3 and HClO4 (3:1) and the concentration of Ca, Cu, Fe, Mg, Mn, P, S, and Zn determined using inductively coupled plasma optical emission spectrometry (ICP-OES). Leaf and root nutrient content was calculated as leaf and root nutrient concentration × leaf and root dry biomass. Leaf + root nutrient content was calculated as leaf nutrient content + root nutrient content; the root-to-leaf ratio was expressed as the ratio (%) of leaf nutrient content to leaf + root nutrient content (Impa et al. 2013).

Distribution of P, Ca and K within leaf cells

Prior to harvest, a small section (~ 3 × 3 mm) was cut from the youngest fully-expanded leaves of each treatment to measure leaf cellular P, Ca and K concentrations, which are average P, Ca and K concentrations (based on fresh weight) in each leaf cell (including all the organelles and cytosol). Samples were then carefully stored and prepared to be measured with X-ray microanalysis; the details of sample preparation and analytical system can be found in Ding et al. (2018b). This method has been shown to be highly suitable for cellular element analyses of biological samples (Guilherme Pereira et al. 2018; Hayes et al. 2018; Huang et al. 1994; Marshall 2017; McCully et al. 2010).

For cellular analyses, oblique cells and airspaces were avoided and only cells that were clearly identifiable and had a flat surface were chosen. Different Lupinus species possess similar anatomy and cell types as noted before (Ding et al. 2018b), including upper epidermal cells (UE), palisade mesophyll cells (PM), spongy mesophyll cells (SM), lower epidermal cells (LE), bundle sheath cells, and veins (Fig. S2). The concentrations of P, Ca, and K in UE, PM, SM, and LE were thus obtained by selecting these cells in the elemental maps and quantifying the resulting summed spectra from each region of interest.

Statistics

Data were analysed with the R software platform (R Core Team 2017). General linear mixed-effect models were used to test the differences in nutrient-uptake rate, nutrient concentrations, leaf area, and biomass among species, pH and the interaction between them. We tested the differences of gas exchange parameters and chlorophyll fluorescence with general linear mixed-effect models among species, pH and the interaction between these two factors. Differences in leaf cell [nutrient] were tested using general linear mixed-effect models among different pH, cell types and the interactions of treatments and cell types, with individual plants as the random effect. Based on Akaike Information Criterion (AIC), models with different variance structures were compared to find the one with the residuals most closely meeting homoscedasticity (Burnham and Anderson 2003; Zuur et al. 2009). The mean values and 95% confidence intervals (Cl) were determined by the effects package and plotted (Fox 2003). The significant differences were defined based on Tukey’s post-hoc analysis (P < 0.05). Model results (P < 0.05) are shown in Table S2-S6.

Results

Leaf symptoms

Leaf chlorosis was observed in the youngest leaves at pH 8 (both 8.0a, i.e. adjusted by KOH, and 8.0b, i.e. adjusted by KHCO3) for all L. angustifolius genotypes and L. cosentinii, but it was less severe for L. cosentinii. Interestingly, leaf chlorosis was much worse for all L. angustifolius genotypes at pH 8.0a than at 8.0b (Fig. 1). We observed P-deficiency symptoms at pH 8 for all genotypes, with older leaves of L. angustifolius wild genotypes exhibiting a red colour, and older leaves of L. angustifolius cv, L. cosentinii and L. pilosus exhibiting a yellow colour first, and then a dried appearance (Figs. 1 and S3).

Fig. 1
figure 1

Leaf symptoms of six Lupinus genotypes when grown at different pH. The pH 5, 6.5 and 8a were adjusted by KOH and buffered with MES and TES (0.5 mM each), whereas 8b was adjusted by KHCO3 without MES or TES added. Scale bars are 2 cm

Full size image

Root appearance

The lateral root growth of all genotypes was inhibited at higher pH, but the least for L. pilosus. In addition, root disintegration was observed in all of the L. angustifolius genotypes at pH 8 (Fig. 2).

Fig. 2
figure 2

Root morphology of six Lupinus genotypes when grown with different pH. From left to right in each photo, the pH is 5, 6.5, 8a and 8b individually. The pH 5, 6.5 and 8a were adjusted by KOH and buffered with MES and TES (0.5 mM each), whereas 8b was adjusted by KHCO3 without MES or TES added. Scale bars are 5 cm

Full size image

Net P- and Ca-uptake rates

The net P-uptake rates of all L. angustifolius genotypes and L. cosentinii were significantly and negatively affected at higher pH, while pH had no effect on the P-uptake rate of L. pilosus. The net Ca-uptake rate of calcifuge L. angustifolius cv Mandelup was significantly faster at pH 8 than that at pH 5 and 6.5, whereas pH only had a minor effect on the net Ca-uptake rate of calcicole L. pilosus (Fig. 3).

Fig. 3
figure 3

Net phosphorus (P)-uptake rate of six Lupinus genotypes (a) and calcium (Ca)-uptake rate of two Lupinus genotypes, one calcifuge and one calcicole genotype (b) when grown at different pH. The pH 5, 6.5 and 8a were adjusted by KOH and buffered with MES and TES (0.5 mM each), whereas 8b was adjusted by KHCO3 without MES or TES added. The rates shown were calculated from P- and Ca-depletion in the external solution, which was the same for all plants, irrespective of prior growing conditions; for details, refer to Materials and Methods. Error bars represent 95% confidence intervals (Cl). Letters show significant differences of different pH within each genotype (based on Tukey’s post-hoc analysis, P < 0.05)

Full size image

Whole leaf nutrient concentrations

Compared with those at pH 5, leaf [P], [Fe] and [Zn] at pH 8 were significantly lower for all genotypes; leaf [P] of all genotypes at pH 8 was below the concentration considered deficient for whole shoots for L. albus (1.3 mg g−1) (Loneragan and Snowball 1969); leaf [P] of all L. angustifolius genotypes and leaf [Zn] of all genotypes at pH 6.5 were also significantly lower (Fig. 4a, b and c). Compared with that at pH 5, leaf [Mn] of all L. angustifolius wild genotypes was significantly lower at pH 8.0b, whereas there was no significant difference for L. angustifolius cv Mandelup, L. pilosus and L. cosentinii; however, leaf [Mn] of all genotypes was significantly lower at pH 8.0a (Fig. 4d). Only some leaf [Mn] of L. angustifolius genotypes at pH 5 and 6.5 were around the critical shoot [Mn] (277 mg g−1); in other words, meeting the requirement for near-maximum growth (Snowball and Robson 1986). Compared with that at pH 5 and 6.5, leaf [K] of L. angustifolius cv Mandelup and L. angustifolius P22721 were significantly lower at pH 8, while those of L. pilosus and L. cosentinii were significantly higher; pH had no effect on the leaf [K] of L. angustifolius P25741 and P26723 (Fig. 4e). Among all pH treatments, leaf [Ca] of all L. angustifolius genotypes at pH 6.5 were the highest, while those of L. pilosus and L. cosentinii were greater at higher pH (Fig. 4f). Leaf Mg, Cu and sulfur (S) concentrations are shown in Fig. S4, and root nutrient concentrations in Fig. S5.

Fig. 4
figure 4

Concentrations of a range of nutrients in whole leaves of six Lupinus genotypes when grown at different pH. The pH 5, 6.5, and 8a were adjusted by KOH and buffered with MES and TES (0.5 mM each), whereas 8b was adjusted by KHCO3 without MES or TES added. Error bars represent 95% confidence intervals (Cl). Letters show significant differences of different pH within each genotype (based on Tukey’s post-hoc analysis, P < 0.05). The grey dashed line represents the deficient (P, Ca, K, Fe, Zn) or critical (Mn) nutrient concentration in dry matter of whole shoots or pairs of youngest open leaves (YOL) of L. angustifolius or L. albus harvested on different dates; critical concentration means that the concentration required for maximum growth (Snowball and Robson 1986)

Full size image

Leaf + root nutrient content

Compared with those at pH 5 and 6.5, total P content of all L. angustifolius genotypes and L. cosentinii were significantly lower at pH 8, while total P content of L. pilosus at pH 8 showed no significant difference with that at pH 6.5 (Fig. 5a). Total Fe and Zn content of all the genotypes showed a similar trend with total P content (Fig. 5b and c). Compared with those at pH 5 and 6.5, total Mn content of all the L. angustifolius genotypes was significantly lower at pH 8, while total Mn content of L. cosentinii at pH 8.0b showed no significant difference with those at pH 5 and 6.5. The total Mn content of L. pilosus showed no significant differences between different treatments (Fig. 5d). Total K and total Ca content of all the genotypes showed a similar trend with total Mn content (Fig. 5e and f). Total Mg, Cu and S contents are shown in Fig. S6. Whole leaf and root nutrient contents are shown in Figs. S7 and S8.

Fig. 5
figure 5

Leaf + root nutrient content of six Lupinus genotypes when grown at different pH. The pH 5, 6.5 and 8a were adjusted by KOH and buffered with MES and TES (0.5 mM each), whereas 8b was adjusted by KHCO3 without MES or TES added. Error bars represent 95% confidence intervals (Cl). Letters show significant differences of different pH within each genotype (based on Tukey’s post-hoc analysis, P < 0.05)

Full size image

Root-to-leaf nutrition translocation ratio

For all the L. angustifolius genotypes and L. cosentinii, when the pH was higher, the P translocation index tended to be lower, while there was no significant difference for L. pilosus among different treatments. The Mn translocation ratios of all the species also tended to be lower at pH 8 than those at pH 5 and 6.5. The Fe translocation ratios of L. angustifolius genotypes (except L. angustifolius P22721) and L. cosentinii at pH 6.5 were significantly lower than those at pH 5; and the trend for the Zn translocation indices were very similar to those of Fe translocation ratios. The trend of K and Ca translocation ratios for all the L. angustifolius genotypes were not very clear, but for L. cosentinii and L. pilosus, the K and Ca translocation ratios at pH 8 were significantly higher than those at pH 5 and 6.5 (Table 2).

Table 2 Root-to-leaf nutrient translocation ratio (mean ± 95% confidence intervals) of different Lupinus species when grown at different pH
Full size table

For all the L. angustifolius genotypes, the P, Fe, Zn, Mn and K translocation ratios at pH 8.0b were generally higher than those at pH 8a (Table 2). The other nutrient (Mg, Cu and S) root-to-leaf translocation indices are shown in Table S7.

Leaf cellular nutrient concentration in different cell types

In all genotypes, most P was located in epidermal cells (both UE and LE) at pH 5 and 6.5, while [P] in all cell types were very low and largely consistent across all cell types at pH 8. The [P] in the epidermal cells (both UE and LE) were significantly lower at pH 8 than at pH 5 and 6.5 (Fig. 6a).

Fig. 6
figure 6

Concentrations of phosphorus (P) (a) calcium (Ca) (b) and potassium (K) (c) in individual leaf cell types of four Lupinus genotypes when grown at different pH. The pH 5, 6.5 and 8a were adjusted by KOH and buffered with MES and TES (0.5 mM each), whereas 8b was adjusted by KHCO3 without MES or TES added. Error bars represent 95% confidence intervals. UE, upper epidermal cells; LE, lower epidermal cells; PM, palisade mesophyll cells; SM, spongy mesophyll cells. Uppercase letters show significant differences among different cell types at the same pH, while lowercase letters show significant differences at different pH within each genotype (based on Tukey’s post-hoc analysis, P < 0.05)

Full size image

For all genotypes, most Ca accumulated in the mesophyll cells and tended to accumulate more in PM than SM. The [Ca] in the mesophyll cells of L. angustifolius cv Mandelup were significantly higher at pH 8 than at pH 5 and 6.5. The [Ca] in the PM of L. angustifolius P26723 were higher at pH 6.5 and pH 8.0a than at pH 5, while they did not differ for SM among these pH treatments; however, it was the lowest for both PM and SM at pH 8.0b among all pH treatments. The [Ca] in PM of L. pilosus at pH 6.5 and 8.0a were significantly higher than those at pH 5 and 8.0b, while the [Ca] in SM at pH 8.0a were the highest among all pH treatments. The [Ca] in PM of L. cosentinii at pH 8.0a were the lowest among all pH treatments, while they were the highest at pH 8.0b. The [Ca] in the SM of L. cosentinii were higher at higher pH (Fig. 6b).

At pH 5 and 6.5, L. angustifolius cv Mandelup tended to accumulate more K in epidermal cells; at pH 8, more K was allocated to LE and SM. In L. angustifolius P26723, at pH 5, 6.5 and 8.0b, mesophyll cells tended to accumulate more K than epidermal cells. However, at pH 8.0a, LE and SM of L. angustifolius P26723 were the cell types that exhibited the largest [K], followed by PM, then UE. Lupinus pilosus and L. cosentinii accumulated more K in mesophyll cells at all pH treatments. Cellular [K] of L. pilosus and L. cosentinii at pH 8 were much higher than at pH 5 and 6.5 for all cell types (Fig. 6c). Cellular Mg, S and chlorine (Cl) concentrations are shown in Fig. S9.

Gas exchange and chlorophyll fluorescence

Net photosynthetic rates of all genotypes at pH 8.0a were significantly slower than those at pH 5 and 6.5, and were even slower than those at pH 8.0b for L. angustifolius P25741 and P22721, L. pilosus and L. cosentinii. Rates of all L. angustifolius genotypes and L. cosentinii at pH 8.0b were significantly lower than those at pH 5 and 6.5, with no significant difference for L. pilosus (Fig. 7a). Stomatal conductance (gs) of all genotypes at pH 8.0a was the lowest, except for L. angustifolius cv Mandelup, showing the smallest gs at pH 8.0b; there was a large difference in gs for all genotypes between pH 8.0a and 8.0b (Fig. 7b). Stomatal limitation (Ls) of all L. angustifolius genotypes at pH 8.0b and Ls of all the L. angustifolius wild genotypes at pH 8.0a were significantly higher than at pH 5 and 6.5, while Ls of L. angustifolius cv Mandelup did not differ between pH 5, 6.5 and 8.0a. Stomatal limitation of L. pilosus at pH 8.0a was the highest among all pH treatments, with no significant difference for L. cosentinii (Fig. 7c). Fv/Fm of all genotypes at pH 8 were smaller than those at pH 5 and 6.5 (Fig. 7d).

Fig. 7
figure 7

Net photosynthetic rate (Amax) (a), stomatal conductance (gs) (b), stomatal limitation (Ls) (c), and Fv/Fm (d) of six Lupinus genotypes when grown with different pH. The pH 5, 6.5 and 8a were adjusted by KOH and buffered with MES and TES (0.5 mM each), whereas 8b was adjusted by KHCO3 without MES or TES added. Error bars represent 95% confidence intervals. Letters show significant differences between different pH within each genotype (based on Tukey’s post-hoc analysis, P < 0.05)

Full size image

Biomass

Leaf area of all L. angustifolius genotypes and L. cosentinii were negatively affected by pH 8, and that of L. angustifolius P26723 was less at pH 6.5, with no effect for L. pilosus (Fig. 8a). A similar trend was found for shoot and total root biomass (Fig. 8b and c). The two cluster-root forming species, L. pilosus and L. cosentinii, showed no significant differences in cluster-root biomass among pH treatments (Fig. 8d).

Fig. 8
figure 8

Leaf area (a), shoot (b), and total root (c) dry biomass of six Lupinus genotypes and cluster-root (d) dry biomass of two Lupinus species when grown at different pH. The pH 5, 6.5 and 8a were adjusted by KOH and buffered with MES and (0.5 mM each), whereas 8b was adjusted by KHCO3 without MES or TES added). Error bars represent 95% confidence intervals. Letters show significant differences of different pH within each genotype (based on Tukey’s post-hoc analysis, P < 0.05)

Full size image

Discussion

All L. angustifolius genotypes and L. cosentinii P27225 were sensitive to pH 8 (both 8.0a, i.e. adjusted by KOH, and 8.0b, i.e. adjusted by KHCO3), while L. pilosus P27440 was relatively tolerant. The growth of all pH-sensitive genotypes was significantly less at pH 8 than that at pH 5 and 6.5, while the growth of the pH-tolerant genotype did not differ among pH treatments. All L. angustifolius genotypes and L. cosentinii exhibited P-deficiency and leaf chlorosis symptoms, while L. pilosus only exhibited P-deficiency symptoms at pH 8.

The effect of pH on root morphology

A high [HCO3−] and/or high pH negatively affects the root growth of L. angustifolius and L. albus (Kerley and Huyghe 2002; Tang et al. 1996; Tang and Robson 1993; White and Robson 1990). In this study, the lateral root growth of all genotypes was inhibited at pH 8, but it was least severe for L. pilosus. In addition, root disintegration was observed in the L. angustifolius genotypes at pH 8, and this confirmed what Tang et al. (1993a) found. Decreased root elongation has also been observed in L. angustifolius in alkaline soils (Tang et al. 1992; Tang et al. 1993b), as well as decreased root elongation after exposure to a nutrient solution of pH ≥ 6 for one hour (Tang et al. 1992). Tang et al. (1993b) also found decreased root cell elongation in L. angustifolius in nutrient solution of pH ≥ 6. Yu and Tang (2000) suggested that the difference in H+-buffering and Ca2+-exchange capacities in the root apoplast is related to the sensitivity of root growth of L. angustifolius and Pisum sativum to high pH, and that the greater sensitivity of L. angustifolius roots to high pH can be partly explained by a higher H+ requirement for cell-wall loosening. According to this explanation, the better root growth of L. pilosus might be the result of higher H+-buffering capacity and a high proportion of easily exchangeable Ca2+. There are specific PM H+-ATPases, as discussed below, that improve the root growth of L. pilosus at high pH.

Solution pH and leaf chlorosis

Leaf chlorosis in relation to Fe or Mn deficiency has been observed in Lupinus species growing in calcareous soils (Brand et al. 2002; George et al. 2012; Moraghan 1991; White and Robson 1989a), in nutrient solution with a high [HCO3−] and/or high pH (Bertoni et al. 1992; White and Robson 1990), or in nutrient solution with a low [Fe] (White and Robson 1989b) or [Mn] (Zornoza et al. 2010). In the present study, all L. angustifolius genotypes and L. cosentinii exhibited leaf chlorosis at pH 8, although the whole leaf [Fe] of these genotypes were similar to those at pH 6.5, and the leaf [Mn] were above the concentration considered adequate for crop species (Kirkby 2012). However, total [Fe] or [Mn] in shoot dry weight are unreliable predictors of deficiency (Brennan 1999; Hannam et al. 1985; Jessop et al. 1990), because Mn is available to plants only as Mn2+ and plants predominantly require Fe2+, rather than Fe3+ (Broadley et al. 2012; Grillet and Schmidt 2017; Pittman 2005). Increased pH (from 6.5 to 8) in the leaf apoplast restricts Fe3+- or Mn4+- reductase activity (Kosegarten et al. 2004; Mengel 1994; Römheld 2000; Zohlen and Tyler 2000), reducing Fe and/or Mn availability. In contrast with L. angustifolius and L. cosentinii, leaf chlorosis of L. cosentinii was very light, and we did not observe any chlorosis in L. pilosus at pH 8. The increased leaf concentrations of K, together with the greater translocation of K to leaves in L. cosentinii and L. pilosus may be related to the activity of high pH-activated membrane-bound proton-pumping ATPases, resulting in a more acidic apoplastic pH and improving Fe2+ or Mn2+ utilisation (Jolley et al. 2004; Tagliavini and Rombolà 2001).

The effect of pH on nutrient uptake and translocation

The net P-uptake capacity of all the L. angustifolius genotypes and L. cosentinii was less at pH 8, while there was no difference in the net P-uptake capacity of L. pilosus. This is probably because lateral root growth of all the L. angustifolius genotypes and L. cosentinii was inhibited at pH 8, and the root surface area was decreased as well, likely accounting for a lower P-uptake capacity (Niu et al. 2013). The uptake of Fe, Zn, Mn, K and Ca of all the L. angustifolius genotypes and L. cosentinii was inhibited at pH 8 as well, because of decreased lateral root growth and reduced nutrient solubility in nutrition solution at high [HCO3−] and/or high pH (George et al. 2012; Yang et al. 1994). However, we only observed lower uptake of P, Fe and Zn at pH 8 which could be caused by decreased P, Fe and Zn solubility at pH 8 (George et al. 2012; Yang et al. 1994; Table S1).

A high pH in the root apoplast could restrict Fe3+- and/or Mn4+-reductase activity and decrease Fe and/or Mn translocation to leaves, and hence more accumulated in the roots (Kosegarten and Koyro 2001; Mengel 1994; Millaleo et al. 2010; Rengel 2000; Zribi and Gharsalli 2002). This agrees with our results showing that root [Mn] of all the genotypes was generally greater at pH 8 than at pH 5 and 6.5, and the Fe and Mn translocation to leaves of all genotypes was inhibited at high pH. In addition, the root Fe3+-reducing capacity requires the availability of metal ions (e.g., Fe2+, Zn2+, Mn2+) (Romera et al. 1997), while the availability of these ions is relatively low at high pH which might result in decreased root Fe3+-reducing capacity. Interestingly, pH 8.0a, adjusted by KOH, had a more severe negative effect on the translocation of P, Fe, Zn, Mn and K in L. angustifolius than pH 8.0b, adjusted by KHCO3. This is probably because high HCO3− supply enhanced root PEP-carboxylase activity, producing organic acids (Abadía et al. 2002; Miller et al. 1990) to be transported to the leaves (López-Millán et al. 2000) or increase the availability of nutrients to be transported to leaves (Jones 1998).

Low leaf [P] or [Fe] can trigger cluster-root formation, and/or the capacity to release protons and carboxylates in Lupinus species, thus increasing the solubility of P and Fe (Dinkelaker et al. 1989; Lambers et al. 2013; Liang and Li 2003; White and Robson 1989b). However, we did not observe an increase in cluster-root formation in L. pilosus and L. cosentinii from pH 5 to pH 8. A high [HCO3−] and/or high pH may have inhibited cluster-root formation or carboxylate release in response to endogenous [P] or [Fe].

The effect of pH on P and Ca allocation

As previously observed (Ding et al. 2018b), P was preferentially allocated to epidermal cells, and Ca to mesophyll cells, especially palisade mesophyll cells. This has been found before in a range of dicots (Conn and Gilliham 2010), but is not typical for all dicots (Hayes et al. 2018; Guilherme Pereira et al. 2018). At higher pH, the [P] in epidermal cells of all genotypes was significantly lower; this agrees with the P-uptake capacity and whole leaf [P] results. The change of [Ca] in the mesophyll cells of all genotypes was complex. The most interesting observation was the [Ca] of L. angustifolius P26723 and L. pilosus, whose [Ca] in the mesophyll cells at pH 8.0b were much lower than those at pH 8.0a. We did not find any increase in [Ca] of the epidermal or bundle sheath cells in L. angustifolius P26723 and L. pilosus at pH 8.0b. However, their whole leaf [Ca], should be higher at pH 8.0a than those at pH 8.0b according to leaf cellular [Ca], which was very similar at pH 8.0a and 8.0b. This means there might be some Ca in the leaf apoplast at pH 8.0b.

The relationship between nutrient status and photosynthesis

Photosynthesis requires P, Fe and Mn (Broadley et al. 2012; Hawkesford et al. 2012; Mengel and Kirkby 2001), and the rate of photosynthesis of all L. angustifolius genotypes and L. cosentinii decreased at pH 8, while that of L. pilosus was only slightly decreased at pH 8.0a. The photosynthetic rate of L. angustifolius P25741, L. angustifolius P22721, L. pilosus, and L. cosentinii at pH 8.0a was even slower than that at pH 8.0b. In addition, the quantum yield of PSII (Fv/Fm) of all genotypes were significantly lower at pH 8 than at pH 5 and 6.5, indicating that the PSII activity of all genotypes was reduced at pH 8 (Maxwell and Johnson 2000). We observed P-deficiency and chlorosis symptoms in young leaves of all of the L. angustifolius genotypes and L. cosentinii at pH 8, and the leaf chlorosis may be related with either Fe- or Mn-deficiency, while we only observed P-deficiency symptoms in L. pilosus at pH 8. Therefore, the decreased photosynthetic capacity of all the L. angustifolius genotypes was caused by P, Fe or Mn deficiency, or a combination thereof. However, the low Amax of L. pilosus was consistent with its low leaf [Mn], which is much lower than the critical shoot required for the growth of L. angustifolius [Mn] (277 mg kg−1), so the decreased Amax of this species was likely due to a decreased leaf [Mn] caused by pH 8.

Similarities and differences between the results in this study and those in other studies

Generally, our findings agree with the natural occurrence of the species; for example, L. angustifolius naturally grows on acidic soils, and L. pilosus on more alkaline soils (Clements and Cowling 1990). Other studies also showed similar nutrient imbalances at high [HCO3−] and/or high pH. For example, Kerley et al. (2001) reported that leaf [P], [Fe], [Mn] and [Zn] of L. albus, L. angustifolius and L. pilosus are lower when plants grow in calcareous soils. In the same study, Kerley et al. (2001) also found that leaf [P] was below the critical concentration in L. albus, L. angustifolius and L. pilosus when grown in calcareous soils. In addition, Fe translocation of L. albus, L. angustifolius and L. pilosus from stems to leaves was negatively affected by high [HCO3−] (Kerley et al. 2001). A negative effect of high [HCO3−] on leaf [Zn] and [Fe] was also found for other species (Dogar and Van Hai 1980; Fleming et al. 1984; Forno et al. 1975; Gharsalli and Hajji 2002). Our results indicate that pH is the main reason why some Lupinus species are sensitive to calcareous soils. Interestingly, comparing effects of pH 8.0a and 8.0b, pH 8.0a gave a stronger inhibition of Mn translocation to leaves in all L. angustifolius genotypes and L. cosentinii. Also, leaf cellular [Ca] of L. angustifolius P26723 and L. pilosus were lower at pH 8.0b than at pH 8.0a. This probably reflects increased HCO3− assimilation and formation of organic acids in roots at high HCO3− supply which resulted in greater Mn translocation to leaves. At the same time, bicarbonate-enhanced Mn2+-translocation would compete with Ca for the plasma-membrane Ca2+ channel in L. angustifolius P26723 (White 2000).

We found that leaf K concentrations of L. pilosus and L. cosentinii at pH 8 were higher than those at pH 5 and 6.5 which could be the reason why these two species had no or less leaf chlorosis symptoms. This is probably because K is involved in activating membrane-bound proton-pumping ATPases which resulted in a more acidic apoplastic pH and improved Fe and Mn availability (Barak and Chen 1984; Jolley et al. 2004). We also found that more Ca was translocated to leaves at pH 8 than at pH 5 and 6.5 in these two species which suggests that Ca could be related to the alleviation of negative effects caused by high pH, as Ca could be involved in decreasing K loss through the inhibition of the K outward-rectified channels, thus maintaining K concentrations (Gómez-Pérez et al. 2014; Murata et al. 2000). Further studies are needed to test this.

Conclusions

A high pH inhibited lateral root growth of L. angustifolius genotypes and L. cosentinii, decreased photosynthetic rates, caused leaf chlorosis and inhibited P uptake. It also inhibited uptake of Fe and Zn in all genotypes and that of Mn, K and Ca in all sensitive species, as well as translocation of P, Fe, Zn, Mn and Ca of most sensitive species from roots to leaves. However, a high pH increased K translocation to leaves in L. pilosus and L. cosentinii which may be related with the activity of H+-ATPases and then their relative tolerance or lower sensitivity to pH 8. Bicarbonate decreased the negative effect of pH 8.0 on nutrient translocation to leaves in most L. angustifolius genotypes. Decreased growth of all L. angustifolius genotypes and L. cosentinii was associated with decreased lateral-root growth, photosynthetic rate and P, Fe or Mn deficiencies. High pH did not affect the leaf cell types that accumulated P and Ca, but the leaf cellular [P] decreased at high pH. This knowledge provides critical insights into the calcicole or calcifuge habits of Lupinus species and other plants which can be used to guide breeding of calcicole plants to improve their production and use.