1 Introduction

The 15N natural abundance (NA) method is an important technique to measure N2 fixation, N uptake and N transfer in the soil-plant system. This method is used in plant physiology and ecology (Handley and Raven 1992; Högberg 1997; Robinson 2001), but it can also be applied at the farm level as well as at the landscape level (Unkovich and Pate 2000). The 15N NA method has become more commonly used to quantify N2 fixation (Okito et al. 2004).

However, to adequately and correctly quantify N2 fixation, we need to know the δ15N of soil N; if a legume or actinorhizal plant was totally reliant on soil mineral and organic N, its δ15N should resemble that of the soil mineral N taken up (Unkovich et al. 2008). Hence, the δ15N of a legume or actinorhizal plant using a combination of atmospheric N2, soil mineral and organic N for growth should range between the values of the two possible N sources: soil and atmospheric N2 (Andrews et al. 2011). Usually, to calculate the %Ndfa (percentage of N derived from the atmospheric N2), non-N2-fixing ‘reference’ plants are used to obtain the δ15N of soil mineral and organic N and the N2 fixation by the legume or actinorhizal plant can be calculated using the following formula (Eq. 1):

$$ \%\mathrm{Ndfa}=\left(\frac{\delta^{15}\mathrm{N}\ \mathrm{of}\ \mathrm{reference}\mathrm{plant}-{\delta}^{15}\mathrm{N}\ \mathrm{of}\ {\mathrm{N}}_2\ \mathrm{fixingplant}}{\backslash {\mathrm{updelta}}^{15}\mathrm{N}\ \mathrm{of}\ \mathrm{reference}\ \mathrm{plant}-\backslash {\mathrm{updelta}}^{15}\mathrm{N}\ \mathrm{of}\ {\mathrm{N}}_2}\right)\times 100 $$

In 15N enrichment studies, isotopic discrimination during N2 fixation is generally considered to be zero; however, isotopic discrimination should be considered when using 15N natural abundance (Unkovich et al. 2008). The natural abundance of atmospheric N2 will have a δ15N of 0‰, by definition. The above equation assumes that N2 fixing plants relying only on N2 fixation do not have any isotope discrimination, hence they have a δ15N similar to the δ15N of atmospheric N2. However, N2 fixing plants have a certain discrimination between 14N and 15N, even when relying only on N2 fixation. Hence, to correctly quantify N2 fixation, the 15N NA method demands a constant, namely the “B” value, which is the δ15N value of the N2 fixing plants relying only on N2 fixation. Therefore, the above Eq. 1, including the B value, can be written as follows (Eq. 2):

$$ \%\mathrm{Ndfa}=\left(\backslash {\mathrm{updelta}}^{15}\mathrm{N}\ \mathrm{of}\ \mathrm{reference}\ \mathrm{plant}-\backslash {\mathrm{updelta}}^{15}\mathrm{N}\ \mathrm{of}\ {\mathrm{N}}_2\ \mathrm{fixing}\ \mathrm{plant}/\backslash {\mathrm{updelta}}^{15}\mathrm{N}\ \mathrm{of}\ \mathrm{reference}\ \mathrm{plant}-\mathrm{B}\right)\times 100 $$

The use of the equation including the B value will result in a more accurate N2 fixation contribution and hence a more precise calculation of the percentage of N derived from the atmosphere. Several studies have emphasised the importance of measuring B values (Kurdali 2009; Andrews et al. 2011; Chalk et al. 2016). Previous studies on legumes have shown that the B value can vary with plant species and age (Carlsson et al. 2006; Unkovich et al. 1993), bacterial strains, and growing conditions (Carlsson et al. 2006; Ledgard 1989; Unkovich and Pate 2000). Much less is known about actinorhizal plants; however, Domenach et al. (1989) have investigated δ15N in Alnus incana and A. glutinosa inoculated with four different Frankia strains. Tjepkema et al. (2000) studied differences in δ15N values among parts of the actinorhizal plants Alnus glutinosa (L.) Gaertn., Casuarina cunninghamiana Miq., Datisca glomerata (Presl.) Baill. and Myrica gale L grown with N-free solution. However, plants were inoculated mainly with crushed nodule suspensions without testing different strains (Tjepkema et al. 2000).

The group of actinorhizal plants includes eight families of mainly woody plants that are infected with N2-fixing actinomycetes of the genus Frankia (Huss-Danell 1997). In the present work, we studied the actinorhizal plants Alnus incana L. Moench (grey alder) and Hippophaë rhamnoides L. (sea buckthorn). H. rhamnoides is a multipurpose plant that has been exploited in East Europe and Asia for several years (Li and Schroeder 1996; Zielinska and Nowak 2017). Alnus incana is a fast-growing tree that thrives in alluvial nutrient-poor soils, and it is used to stabilize damaged forest topsoils (Flores Fernández et al. 2019). These two actinorhizal plants have the following different Frankia infection pathways: via root hairs in A. incana and via intercellular penetration in H. rhamnoides (Berry and Torrey 1979; Gentili et al. 2006). Generally, each of actinorhizal plant species has a different infection pathway (Pawlowski et al. 2011), and is infected by a specific Frankia strain; however some Frankia strains such as UFI 132715 are able to form nodules in actinorhizal plant species having different infections pathways (Lumini and Bosco 1996).

The legumes studied in the present work were Hedysarum coronarium (Italian “sainfoin” or “sulla”) infected by Rhizobium sullae (de Diego-Diaz et al. 2018) and Robinia pseudoacacia (black locust) infected by Mesorhizobium and Rhizobium (Andrews and Andrews 2017). Hedysarum coronarium is a good forage plant for Mediterranean regions, and it can grow in arid environments and clay soils (Chisci et al. 2001), while R. pseudoacacia is a widespread tree used in agroforestry and land reclamation (Groninger et al. 1997).

The main aim of this work was to determine the B values in the four plant species mentioned above at different plant ages and with different Frankia strains for the actinorhizal plants and with crushed nodule or soil suspension for the legumes.

2 Materials and methods

2.1 Plant cultivation

Alnus incana seeds were obtained from a clone (Huss-Danell 1991), while H. rhamnoides from Hungary and R. pseudoacacia from Poland were purchased from Sandeman Seeds (UK). Hedysarum coronarium seeds were collected in central Italy (Santa Sofia, Forlì, Italy). Seeds of H. rhamnoides and R. pseudoacacia were surface-sterilised as previously described by Gentili and Huss-Danell (2002). Alnus incana seeds were washed in water containing a few drops of detergent, immersed in hydrogen peroxide (30%v/v) for 1 h, shaken a few times manually, and then carefully rinsed with sterile water. Hedysarum coronarium seeds were not surface-sterilised. Hippophaë rhamnoides and R. pseudoacacia were sown in Petri dishes containing perlite, He. coronarium was sown in 8 × 8 cm plastic pots containing perlite and A. incana was sown in Magenta jars (Sigma, St Louis, Mo, USA) containing glass beads. All seeds were watered with modified N-free Evans solution (Huss-Danell 1978) diluted to 1/10 of full strength. The perlite, glass beads and nutrient solution were sterile. The seedlings were transferred to larger pots twice for the legumes He. coronarium and R. pseudoacacia or three times for A. incana and H. rhamnoides (Table 1).

Table 1 Plants age at different experimental phases. WAS stands for weeks after sowing
Full size table

During the entire experiment, plants were watered with N-free Evans solution diluted to 1/10 of full strength. During the whole growth period (end of February to beginning of October), the plants were kept in a greenhouse in Umeå, Sweden (63°45’N), with 17 h of supplemental light set to about 25 °C and 7 h of darkness set to about 15 °C. Philips HPI/T 400 W lamps were used. Relative air humidity was set to 40%.

2.2 Inoculation and harvest of plants

Some of the A. incana seedlings were inoculated with the Frankia strain ArI3 (Berry and Torrey 1979) grown on P + N medium (Van Ghelue et al. 1997). The Frankia cells were collected by centrifugation at 7000 g for 10 min at 22 °C, and the supernatant was discarded. The cell pellet was washed twice by centrifugation (as above-mentioned), using N-free Evans solution diluted to 1/10 of full strength. Each seedling was inoculated with approximately 6 mg (wet weight) of Frankia cells added into the Magenta jars. Another part of the A. incana seedlings was inoculated with the crushed root nodule suspension from A. incana, the local source of Frankia (lsF) (Huss-Danell 1991).

Hippophaë rhamnoides seedlings were inoculated with the strains E15b (Lumini and Bosco 1996) and T1 (Chaia 1998), both of which were grown in K-medium with casein hydrolysate and yeast extract as N source (Lumini et al. 1996). Both strains were harvested as described above for the Frankia strain, ArI3. However, each H. rhamnoides seedling was inoculated with 30 mg (wet weight) of Frankia cells the first time and 16 mg of cells the second time (Table 1).

Hedysarum coronarium was inoculated with a soil suspension from an Italian area where this legume grows naturally, while R. pseudoacacia was inoculated with a crushed root nodule suspension from plants of R. pseudoacacia growing in an Italian nursery garden. For inoculation times, see Table 1. All plant species were harvested twice at the different plant ages indicated in Table 1. Once harvested, the plants were immediately dried at 60 °C for 48 h at the first harvest time and for 72 h at the second harvest time. This difference between the drying times was because at the second harvest, plants were much larger and more lignified. The dried shoots and the dried roots, including the nodules, were milled and analysed for N and 15N as previously described (Gentili 2006). All seeds were dried at 60 °C for 24 h, milled and analysed as above-mentioned.

2.3 Nitrogenase activity

Acetylene-reducing activity (ARA) was measured in five randomly chosen H. rhamnoides plants for the two Frankia strains at 21.5 weeks after sowing. The measurements were done three times a week, with one day in between measurements. Each intact individual plant was placed in a gastight chamber (Huss-Danell 1978) during exposure to 10% (v/v) C2H2 in air. During these incubations, the plants were kept in the greenhouse. The formation of C2H4 was measured by gas chromatography as described previously (Valverde et al. 2000).

2.4 Statistical method

One-way ANOVA was used to identify statistically significant differences. Statistical calculations were performed using the software package Minitab (Minitab Inc., State College Pennsylvania, PA, USA, 2000), with P < 0.05 as the significance level.

3 Results

The legume seeds had a higher nitrogen concentration than the actinorhizal seeds (Table 2). Only Robinia pseudoacacia seeds had a positive δ15N value when compared to the other seeds analysed (Table 2). However, regarding the weight of a single seed in comparison with the plant dry matter and leaf plus root dry matter (Tables 2 and 3), the influence of the N and δ15N of the seeds had a negligible impact on the N and δ15N of the plants. There was a great difference in N concentration among the plants analysed, ranging from 0.94% in roots of Alnus incana to 3.39% in leaves of Hippophae rhamnoides. However, these differences were not only due to differences in plant species, but also there were significant differences among plant parts, time of harvest and Frankia strains (Table 4). In particular, A. incana inoculated with Frankia strain ArI3 had a significantly higher N concentration in both leaves and roots at first harvest than at second harvest (Table 4). However, A. incana inoculated with Frankia strain lsF had a significantly higher N concentration in leaves at second harvest than at first harvest (Table 4).

Table 2 Dry seeds mass, nitrogen concentration (% of seed dry mass) and δ15N values in dry seeds of the studied plant species. The N and δ15N values are the mean ± SE of 5 replicates
Full size table
Table 3 Plant dry matter (g) and leaf+root (g) of the two actinorhizal plant species inoculated with two different Frankia strains and of the two leguminous plants inoculated. All the plants were harvested at two different time. Values are the mean ± SE of 5 replicates. na stands for not available
Full size table
Table 4 Nitrogen concentration in leaves and in roots with nodules at different harvest time and with different Frankia strains
Full size table

In H. rhamnoides, there were no effects on N concentration due to harvest time or strains used (Table 4). Even in Hedysarum coronarium, harvest time had no effect on leaf N concentrations (Table 4). However, in R. pseudoacacia, roots had higher N concentrations than leaves at both harvest times, while harvest time did not have any clear effect on N concentrations (Table 4).

The δ15N values of the N2 fixing plants relying only on N2 fixation or B values are shown in Table 5. The δ15N ranged from −2.62 in leaves of A. incana inoculated with the “lsF” to 0.77 in roots of R. pseudoacacia (Table 5). The δ15N was negative in the two actinorhizal plants studied with all the Frankia strains, while the two legume species studied showed negative (He. coronarium) and both negative and positive δ 15N (R. pseudoacacia). On the one hand, in A. incana inoculated with the ArI3 strain, the only relevant difference among plant parts and time of harvest was a slight, but significantly more negative, δ15N value in the roots at the second harvest (Table 5). On the other hand, A. incana inoculated with the “lsF” did not show any relevant differences among plant parts and times of harvest (Table 5). Hippophaë rhamnoides was even more interesting because, although the two strains did not affect N concentrations and δ 15N signatures among plant parts and time of harvest, they greatly affected plant biomass (compare Tables 3, 4 and 5) and acetylene-reducing activity (Fig. 1). Under these experimental conditions, the strain T1 fixed N2 much more efficiently than the strain E15b.

Table 5 δ15N values (B values) of leaves and roots at different harvest times and in plants with different Frankia strains
Full size table
Fig. 1
figure 1

Acetylene reducing activity in five randomly chosen H. rhamnoides plants measured at three different date during one week. Empty bars show ARA for plants infected by the Frankia strain E15 b while bars with grid show ARA for plants infected by the T1 Frankia strain. Bars stand for mean ± SD of five replicates. Different letters on the top of the bars represent statistically differences among plants infected by different strains at the same measurement date at p ≤ 0.05 as significance level

Full size image

Interestingly, in H. rhamnoides, there were clear differences in δ15N values due to the time of harvest and the plant parts; roots had a more negative δ15N than leaves. However, Frankia strains did not significantly affect the δ 15N signature of H. rhamnoides (Table 5).

He. coronarium showed a significant difference between δ 15N values of leaves at different harvesting times and between leaves and roots (Table 5). In He. coronarium, N concentrations and δ 15N values had a different trend (compare Tables 4 and 5).

R. pseudoacacia δ 15N signatures followed a trend similar to that of N concentrations (Tables 4 and 5). In this leguminous species, roots had δ 15N signatures higher than leaves at both harvesting times, while harvesting time did not have any clear effect on N concentrations (Table 4).

4 Discussion

It is particularly interesting that in A. incana, the two Frankia strains affected not only N concentration and δ15N signatures of leaves and roots, but also had an impact on plant growth at first harvest (compare Tables 3, 4 and 5). This could be due to a faster root colonisation and nodule formation by the strain ArI3 compared to the lsF. At the second harvest, the plants inoculated with the lsF had similar dry matter values compared to the plants inoculated with the ArI3 strain (Table 3) and a higher nitrogen concentration in both leaves and roots (Table 4). This could be due to a higher N2 fixation of the plants inoculated with lsF than plants inoculated with ArI3. Remarkably, plants infected with the lsF strain had the δ15N value of the leaves at second harvest about 0.5 δ units more negative than plants inoculated with the ArI3 Frankia strain. A previous study had shown that the choice of the Rhizobium strain significantly influenced the B values in shoots of Trifolium hybridum and Trifolium repens and the N contents in all Trifolium species tested (Carlsson et al. 2006). Furthermore, in legumes, considerable variation was found in δ15N values of plant herbage (−4.5 to +0.8); the extent of isotopic discrimination was dependent on both the host plant and the infecting rhizobial strain (Steele et al. 1983). In a previous study, A. incana had a leaf δ15N of approximately −1.4, hence less negative than in the present study. However, the inoculating strain and the age of the plants were not specified (Domenach et al. 1989). Working on actinorhizal plants, Tjepkema et al. (2000) found that whole plant δ15N for the four actinorhizal plants examined ranged from −1.41 to −1.90‰. However, they harvested the plants when they reached a dry mass of between 0.5 and 4 g, which was similar to our results from the first harvest, but significantly smaller than our results from the second harvest, where actinorhizal plant dry matter values ranged from 10.5 to 17.6 g.

Hippophaë rhamnoides plants inoculated with the T1 Frankia strain showed a greater biomass and higher ARA than plants inoculated with the E15b strain (Table 3, Fig. 1). Consequently, under these experimental conditions, the T1 strain had a significantly higher efficiency to fix N2 than the E15b strain. However, for both N concentrations and δ15N values, plants infected with the E15b and T1 strain, respectively, did not differ at any harvest time (Tables 4 and 5). This means that although T1 fixed N2 more efficiently than E15b, as also confirmed by the ARA measurements (Fig. 1), which resulted in larger plants at both harvesting times, it had the same 15N discrimination as the less productive E15b. On the one hand, concerning N concentrations in H. rhamnoides, there was no significant difference among Frankia strains, time of harvest and plant parts. On the other hand, in terms of δ15N, there were differences at time of harvest in leaves and among plant parts, with more negative values for nodulated roots than for leaves. The fact that leaves showed differences over time is extremely relevant, because leaves are generally used to measure N2 fixation for both actinorhizal plants and legumes. Hence, it would be important in future studies to follow δ15N values of actinorhizal and legume woody plants grown with N2 as the only nitrogen source during several growing seasons.

In the present study, leguminous plants showed a higher δ15N of roots compared to leaves, as had been previously shown (Unkovich et al. 2008). In contrast, the two actinorhizal plants inoculated with four different Frankia strains generally had more negative δ15N values of nodulated roots than leaves (Table 5). It seems that actinorhizal plants have a great variability of 15N fractionation among plant parts, as had also been shown by Tjepkema et al. (2000), where some plants grown on N2-free medium had more positive δ15N values for roots than for leaves, while the opposite was true for other actinorhizal plant species.

In conclusion, for the calculation of the %Ndfa, the δ15N value of N2 fixing plants relying only on N2 as nitrogen source (B value, see above Eq. 2) should be used instead of the δ15N value of atmospheric N2. This will produce more accurate estimations of %Ndfa. In fact, the use of δ15N of atmospheric N2 (0‰) as the B value could result in an underestimation or overestimation of the %Ndfa if the real B value is positive or negative, respectively. Based on our results, the use of the B value of different plant parts, plant age and infecting strains should be carefully considered.