Introduction
The practice of agroforestry on agricultural land has been the subject of considerable research (Akinnifesi et al., Reference Akinnifesi, Ajayi, Sileshi, Chirwa and Chianu2010; Schroth, Reference Schroth1998; Luedeling et al., Reference Luedeling, Smethurst, Baudron, Bayala, Huth, van Noordwijk, Ong, Mulia, Lusiana, Muthuri and Sinclair2016; Dupraz et al., Reference Dupraz, Wolz, Lecomte, Talbot, Vincent, Mulia, Bussière, Lafontaine, Andrianarisoa, Jackson, Lawson, Dones, Sinoquet, Lusiana, Harja, Domenicano, Reyes, Gosme and van Noordwijk2019; Sida et al., Reference Sida, Baudron, Ndoli, Tirfessa and Giller2020; Cardinael et al., Reference Cardinael, Hoeffner, Chenu, Chevallier, Béral, Dewisme and Cluzeau2019b; Clivot et al., Reference Clivot, Petitjean, Marron, Dalle, Genestier, Blaszczyk, Santenoise, Laflotte and Piutti2020). Trees modify soil properties through numerous processes (Rhoades, Reference Rhoades1997; Isaac and Borden, Reference Isaac and Borden2020; van Noordwijk et al., Reference Van Noordwijk, Barrios, Shepherd, Bayala and Oborn2019). Tree and crop root systems play a crucial role in the belowground interactions (Bayala and Prieto, Reference Bayala and Prieto2020; Bardgett et al. Reference Bardgett, Mommer and De Vries2014). Biological nitrogen fixation by tree, crop, or both components of agroforestry systems has received a lot of attention in the tropics (Nair et al., Reference Nair, Buresh, Mugendi and Latt1999). There is a sense that resource use is increased under agroforestry systems (Vandermeer et al., 1998); for example, belowground transfer of nitrogen among plants has been reported, from a N2-fixing source plant to a non-fixing plant sink (Haystead et al., Reference Haystead, Malajczuk and Grove1988; Arnebrant et al., Reference Arnebrant, Finlay and Söderström1993; He et al., Reference He, Xu, Qiu and Zhou2009). The nitrogen transfer from N2-fixing trees can be a major source for the associated crops in a low-inputs farming system. According to Lal (Reference Lal2004), the N is the most important nutrient limiting crop production in the tropical small-scale farming; therefore integrating the N2-fixing trees into farmland may improve the N supply of the cropping system.
Plants are known to exude nitrogenous compounds from living roots (Jalonen et al., Reference Jalonen, Nygren and Sierra2009). These compounds can be absorbed as such by adjacent plants without prior mineralization by soil microorganisms (Jones et al., Reference Jones, Hodge and Kuzyakov2004). Several mechanisms involving N transfer from N2 fixation by plants have been reported by different authors, via: roots grafts or root-to-root contacts (Caldwell and Richards, Reference Caldwell, Richards and Givnish1986); roots exudations (Teste et al., Reference Teste, Veneklaas, Dixon and Lambers2014); and mycorrhizal networks (He et al., Reference He, Critchley and Bledsoe2003; Barea et al., Reference Barea, Pozo, Azcon and Azcon-Aguilar2005; Battie-Laclau et al., Reference Battie-Laclau, Taschen, Plassard, Dezette, Abadie, Arnal, Benezech, Duthoit, Laure Pablo, Jourdan, Laclau, Bertrand, Taudiere and Hinsinger2020). But it is not clear how important it is quantitatively. It is also argued that after root exudation, two basic mechanisms may be transporting ions and simple amino acids to neighboring plants; the first is mass flow that moves ions along with the flow of water and the second is the diffusion of ions along concentration gradient without the flow of water (Teste et al., Reference Teste, Veneklaas, Dixon and Lambers2014).
According to Evans (Reference Evans2001), whole plant and leaf N isotope composition is determined by the isotope ratio of the external N sources and physiological mechanisms within the plant. The physiological mechanisms that influence plant N isotopic signature have been reviewed by Evans (Reference Evans2001), and earlier by Högberg (Reference Högberg1997) and Handley and Raven (Reference Handley and Raven1992).
The study of N transfer among plants could have potential in agro-farming under low external inputs (Jensen, Reference Jensen2005; Wichern et al., Reference Wichern, Eberhardt, Mayer, Joergensen and Müller2008). Scientists initially hoped that quantifying δ15N could be used to trace the relative contribution of N 2 fixation to plants and soils (Michener and Lajtha, Reference Michener and Lajtha2007). Many authors have used δ15N data to draw inferences regarding N sources (Garten et al., Reference Garten, Hanson, Todd, Lu, Brice, Michener and Lajtha2007; Evans et al., Reference Evans, Michener and Lajtha2007; Phillips and Greggs, Reference Phillips and Gregg2003). For example, Schulze et al. (Reference Schulze, Chapin and Gebauer1991) employed the natural abundance of δ15N values in Acacia savannas to estimate the nitrogen fixation by the Acacia melifera trees on aridity gradient in Namibia, and found that about 71 % of nitrogen was fixed. Similarly, Shearer et al. (Reference Shearer, Virginia, Bryan, Skeeters, Nilsen, Sharifi and Rundel1983) used the natural abundance of δ15N in tissues of Prosopis grandulosa to estimate the N2-fixation by these trees, and concluded it is feasible to use variation in natural abundance of δ15N as an index of N2-fixation. Additionally, studies in the California Sonaran desert indicated Prosopis woodland fixed a significant amount of N2 based on soil N accumulation beneath (Virginia and Jarrell, Reference Virgina and Jarrell1983); since desert soil is often deficient in nitrogen, they argued that N2-fixing by prosopis tree might be the basis of that soil N accumulation. However, the novel in this research is coupling the isotopic mixing modeling “IsoSource” to the plant natural abundance of δ15N to determine the proportional contribution of N by N2 fixing plant to a non-fixing intercrop.
There would be a great deal to be learned from δ15N of plant tissues and their sources of N. The aim of this research was to experiment if the natural abundance of 15N coupled to isotopic mixing modeling can be applied to determine the relative transfer of N from a N 2 fixing tree to a non-fixing intercrop in the field.
Material and methods
Description of research area
The research described here was conducted at three sites located in Northern Rwanda (Supplementary Material Fig. S1). The site at Rurembo was located at 01° 53′S; 29° 57′E and elevation 2245 m. The second site at Kirezi was at an elevation of 2269 m and at 01° 54′ S; 29° 57′ E. The third site at Cyansure was located at 01° 55′ S; 29° 57′ E, and elevation of 2248 m.
The rain distribution across the whole region is bimodal, characterized by long and short rain seasons that allow two cropping seasons a year. Based on climatic data of the local weather station in the study area, the average annual rainfall during the experiment was 1640 mm, with an average annual temperature of 15 °C. The soils of the area are Alfisols (Ultic Tropudalf), Inceptisols (Andic Eutropept; Typic Dystrandept; Entic Eutrandept), and Mollisols (Cumulic Hapludoll) (Supplementary Material Fig. S2). The pH (in water) of the soil was 6.7, and the dominant crops are wheat, potatoes, maize, and peas. Potatoes and wheat are the main cash crops in the area.
Experimental setting and data collection
The experiment was conducted on the farmers’ fields’ terraces that were established in 2010 with transplants of Alnus acuminata trees on the terrace risers (Supplementary Material Fig. S3). The spacing between trees was 5 m, and the terrace width was 8 m. The trees on the terrace risers were pruned three times a year. Wheat (variety Bisagi) was grown as a test crop on terrace’s bench in all fields of the study. The planting density for wheat was 150 plants m−2.
Sampling and measurement
Thirty leaves per field (90 for the whole experiment) were collected on the twigs of Alnus acuminata trees for δ15N measurement. Soil was sampled using a soil auger at the depth of 0–20 cm in each plot at four points across the terrace’s bench at 1 m, 3 m, 5 m, and 7 m from the trees, and three replicates. A total number of 40 composite soil samples were taken, and air-dried and sieved to less than 2 mm for δ15N determination. At anthesis (GS65), the flag leaf wheat sample (144 in total) was collected at four points further away from the trees (1 m, 3 m, 5 m, and 7 m) in the bench along the slope and three replicates for measurements of specific leaf area (SLA), and Carbon discrimination (Δ1 3C) and δ15N. At each of the four points, the chlorophyll content index (CCI) of the flag leaf was measured using the chlorophyll meter SPAD-502 (Konika Minolta Sensing Inc., Osaka, Japan). SLA (cm2 g−1) was obtained by measuring the leaf area (LA), one side of the leaf, with ImageJ (version 1.42q, National Institute of Health, USA). Then, the sample was oven-dried in a paper envelope at 75 °C for 24 h. The leaf dry weight (DW) was obtained by reweighing the sample on micro-balance after oven drying.
The SLA was calculated as the ratio of LA to DW:
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20220311130055498-0575:S0014479721000284:S0014479721000284_eqn1.png?pub-status=live)
For the isotopic measurement and analysis, the dry leaf and soil samples were ground separately using ball mill (MM200 Mixer Mill, Glen Creston Ltd, UK). For each sample, a subsample of 1 mg for leaf and 10 mg for soil was weighed into a tin capsule and analyzed for δ15N at the Godwin Laboratory (University of Cambridge) using a Costech elemental analyzer attached to a Thermo Delta V mass spectrometer in continuous flow mode. The δ15N data were analyzed for the proportional contribution sources to the crop 15N signatures, using an isotopic mixing model ‘IsoSource’ (Phillips and Gregg, Reference Phillips and Gregg2003).
The Carbon discrimination (Δ13C) was obtained through the measurement of the value of δ13C in the dried ground leaf samples weighed (1 mg) into a tin capsule, using Costech elemental analyzer attached to a Thermo Delta V mass spectrometer in continuous flow mode. The δ 13 C value was used to compute the Δ13C following Farquhar et al. (Reference Farquhar, O’Leary and Berry1982):
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20220311130055498-0575:S0014479721000284:S0014479721000284_eqn2.png?pub-status=live)
where the δ13 C a is the delta value of C in the air and the δ13 C p is the delta value of C in the sample.
Statistical analysis
The statistical analysis of the data was performed using SPSS 16.0 for windows (SPSS Inc., Chicago, IL, USA). Firstly, the data were explored for parametric assumptions of normal distribution and homogeneity of variance using Kolmogorrov–Smirnov (K–S) and Levene’s tests, respectively. The graphing of means was performed using bar charts, and the data were subjected to one-way independent ANOVA at p < .01, followed by Bonferroni test at significance level p < .01.
Results
The δ15N of the wheat intercrop indicated consistent gradient declining with the distance from the N2 fixing Alnus acuminata trees (Table 1). The wheat nearest to trees of Alnus acuminata showed the δ15N signature values closer to that of the tree; at 1 m from the trees, the wheat signature was 7·23 ± 0·52 ‰ relative to the tree (7·50 ± 0·13 ‰), while further at 7 m, the crop 15N signature was 3·22 ± 0·83 ‰. The isotopic mixing model indicated that tree N could theoretically have provided 33·6 ± 4·3 % of the crop N at 1 m (Figure 1a). Additionally, the CCI of wheat leaf was significantly higher at 1 m, 3 m, and 5m from the trees, which declined moving further out into the bench (Figure 1b). The results also revealed a gradient decline both in the value of SLA and Δ13C of wheat leaf intercrop toward the tree lines of Alnus acuminata (Figure 1c, d).
Table 1. The mean isotopic values of wheat leaf at different distances from the Alnus acuminata trees across the bench terrace and of the sources (Alnus acuminata tree; soil) of N
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20220311130055498-0575:S0014479721000284:S0014479721000284_tab1.png?pub-status=live)
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20220311130055498-0575:S0014479721000284:S0014479721000284_fig1.png?pub-status=live)
Fig. 1. The proportional contribution of Alnus acuminata to wheat intercrop N (a), Chlorophyll content of wheat leaf (b), Specific leaf area of wheat leaf (SLA, in c) and Δ13C of wheat leaf (d) at different distances from the A. acuminata trees across the bench terrace. Values are means ± se (N 144). The bars showing the same letters indicate that their mean values do not statistically differ significantly (p < 0·01).
Discussion
The understanding of the physiological basis of isotope signatures of the plant nitrogen may be an approach to encapsulate the plant interaction and resources acquisition in agroforestry systems. The key to this research was that the distinct isotopic signatures of various sources of plant nitrogen can be identified, and their relative contribution to plant N could be determined by an isotopic mixing model ‘IsoSource’. The results indicated that Alnus acuminata, a N2-fixing tree, exhibits nitrogen transfer.
The transfer of nitrogen has been reported to exist where plants with contrasting nutrients acquisition strategies (N2-fixing and non-fixing) co-occur (Dawson et al., Reference Dawson, Mambell, Plamboeck, Templer and Tu2002; He et al., Reference He, Xu, Qiu and Zhou2009). However, there remains controversy about whether belowground N transfer occurs (Ikram et al., Reference Ikram, Jensen and Jakobsen1994; Johansen and Jensen, Reference Johansen and Jensen1996); and often research on N transfer in agroforestry has conventionally assumed that N transfer occurs via the decomposition of legume litter and pruning residues in soil (Jalonen et al., Reference Jalonen, Nygren and Sierra2009).
We addressed these questions with the assessment of natural abundance of δ15N and an isotopic mixing model to determine the proportional contribution of N by Alnus acuminata to the wheat intercrop N. The literature has been highlighting the need to study N transfer among plants toward agro-farming under low external inputs (Hauggaard and Jensen, Reference Hauggaard and Jensen2005; Wichern et al., Reference Wichern, Eberhardt, Mayer, Joergensen and Müller2008); therefore we aimed to inform the design of intercropping of N2 fixing trees with wheat as an intercrop for enhanced transfer and distribution of N to the non-fixing N2 intercrop.
The data from this study provided an indication that N transfer from N2-fixing trees can be a considerable N source for the associated intercrop in agroforestry farming (Figure 1a and Table 1). Alnus have been reported to have the potential to provide N-fixation benefits in temperate agroforestry systems (Sharifi et al., Reference Sharifi, Nilsen and Rundel1982; Seiter et al., Reference Seiter, Ingham, William, Hibbs and Ehrenreich1995). Seiter et al. (Reference Seiter, Ingham, William, Hibbs and Ehrenreich1995) demonstrated this potential in a red alder (Alnus rubra Bong.) – maize alley cropping system in Oregon.
Similarly, in this study, a comparison of the δ15N signatures of the tree and the wheat revealed that the crops in the proximity of trees exhibited value closer to the tree δ15N and declined as moving further in the terrace (Table 1), and these are supported with the value of both SLA and Δ13C wheat intercrop. In agreement with our data, Seiter et al. (Reference Seiter, Ingham, William, Hibbs and Ehrenreich1995) observed, using an 15N injection technique, that 32–58% of the total N in maize was obtained from N2 fixed by red alder and that nitrogen transfer increased by shortening the distance between the trees and crops. Our results are also consistent with the works of many authors (Handley and Scrimgeour, Reference Handley and Scrimgeour1997; Robinson, Reference Robinson2001; Evans, Reference Evans2001; Stewart, Reference Stewart and Unkovich2001) who suggested that the δ15N of leaf tissues reflect the net effect of δ15N of the sources used by that plant. Our findings are in accordance with the works of both Moyer et al. (Reference Moyer, Burton, Israel and Rufty2006) and Lu et al. (Reference Lu, Kang, Sprent, Xu and He2013) who showed that N transfer among plants can occur through the release of N compounds from the N 2 -fixing plant leading to uptake by a non N 2 -fixing plant. The relative N transfer of 33.6 ± 4.3 % at 1 m from the trees observed in this study agrees with Sierra and Daudin (Reference Sierra and Daudin2010) who assessed in situ the 15N transfer from stem-labeled trees to associated grass and found that the transfer of the added 15N was limited in space (up to 1 m from trees) and was on average 33 %. Similarly, Snoeck et al. (Reference Snoeck, Zapata and Domenach2000) noted 13–42 % of 15N transfer from the legume trees to coffee. These findings are also in agreement with Handley and Raven (Reference Handley and Raven1992) who suggested that there is no evidence of fractionation either of δ1 4N or δ15N during its physical movement (passive and active uptake) across living membranes of plants.
Nevertheless, such data should be interpreted with caution; for example, Daudin and Sierra (Reference Sierra and Daudin2010) observed that grass presented a preferential uptake of N released by the tree; if that is the case, then this preferential N uptake may cause discrepancy in isotopic mixing model results. Similarly, Sierra et al. (Reference Sierra, Daudin, Domenach and Nygren2007) argued that N transfer from N2 fixing trees may involves direct and indirect pathways; i.e. N transfer could be indirect if N exudates from the roots of tree were taken by soil microorganisms and passed through microbial turnover (Høgh-Jensen, Reference Høgh-Jensen2006); in that case, the isotopic mixing model could not resolve such system because it takes into account only N sources. The pruning regime (frequency and intensity) was also argued to be another factor that may affect N transfer by limiting the rate of N2 fixation (Nygren et al., Reference Nygren, Cruz, Domenach, Vaillant and Sierra2000).
Moreover, Sanchez et al. (Reference Sanchez, Shepherd, Soule, Place, Buresh and Izac1997) argued that the roots of trees are often able to capture nutrients at the depths beyond the reach of most crop and redistribute them into topsoil, and this can be an additional nutrients input in an agroforestry system.
Finally, the question of which mechanisms (and their importance) that drive the N transfer between plants remains unsolved up to date; several mechanisms have been reported; release of N in exudates (Høgh-Jensen, Reference Høgh-Jensen2006), roots-grafts (Caldwell and Richards, Reference Caldwell, Richards and Givnish1986), and mycorrhizal netwoks (He et al., Reference He, Critchley and Bledsoe2003), but the extent of their contribution to N transfer remains unsolved. We, therefore, recommend further research into the molecular mechanisms by which plants transfer N to their neighbors.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S0014479721000284
Conflict of interest
None.