Author for correspondence: Erik A. Hobbie Tel: +1603 8623581 Fax: +1603 8620188 Email: Erik.Hobbie@unh.edu
• Nitrogen isotope (δ 15 N) patterns in plants may provide insight into plant N dynamics. Here, two analytical models of N-isotope cycling in plants and mycorrhizal fungi were tested, as dominant plants in many forest ecosystems obtain most of their N through intereactions with mycorrhizal fungi.
• Fungi were treated either as a single well-mixed N pool, or as two N pools (one available, plus one not available, for transfer to the host). Models were compared against complete biomass and 15 N budgets from culture studies of nonmycorrhizal and ectomycorrhizal Pinus sylvestris (colonized with Suillus luteus or Thelephora terrestris ) grown exponentially at low and high N supply.
• Fungal biomass and N increased at low N relative to high N supply, whereas needle δ 15 N decreased. Needle δ 15 N correlated strongly and negatively with biomass of extraradical hyphae. Our data and models suggest that low plant δ 15 N values in low productivity and N-limited environments result partly from high retention of 15 N-enriched N by mycorrhizal fungi; this retention was driven by increased C flux to fungi under N-limited conditions. The two-pool model of fungal N accounted for greater variability in plant δ 15 N than the one-pool model.
• Plant δ 15 N patterns may indicate relative allocation of fixed C from plants to mycorrhizal fungi under some conditions. Studies are needed on whether patterns observed in culture can be applied to interpret field measurements of δ 15N.
In contrast to the rich body of theoretical and experimental work that has developed to explain plant carbon isotope patterns, few comparable efforts have been undertaken to explain plant nitrogen isotope (δ15N) patterns (Robinson et al., 1998; Hobbie et al., 2000; Comstock, 2001) and yet, observations that plant δ15N changes over gradients of N deposition (Högberg et al., 1996), N availability (Hobbie et al., 1999a, 2000), precipitation (Handley et al., 1999), and primary succession (Vitousek et al., 1989) suggest that a better mechanistic understanding of how these patterns are created could illuminate plant N cycles. Further progress in interpreting plant δ15N will require developing theoretical models of plant δ15N and testing those models under controlled conditions (Evans, 2001), before validating models in the field.
Recent field studies have suggested that mycorrhizal fungi may influence plant 15N abundances (Michelsen et al., 1998), with δ15N values of mycorrhizal plants depleted up to 8‰ relative to co-occurring nonmycorrhizal plants (Nadelhoffer et al., 1996; Schmidt & Stewart, 1997; Michelsen et al., 1998, foliage is usually the only tissue measured in field studies). The few studies that have concurrently measured the δ15N of mineral N have found these sources to be insufficiently depleted in 15N to explain the low δ15N of mycorrhizal plants (Schmidt & Stewart, 1997; Hobbie et al., 1999a). This suggests that mycorrhizal fungi may create 15N-depleted compounds that are subsequently transferred to host plants. Of the three main types of mycorrhizal associations (ecto-, arbuscular, and ericoid mycorrhiza), ectomycorrhizal fungi have been particularly studied because they produce easily collected sporocarps and because they associate with many dominant tree families in boreal, temperate, and tropical regions (e.g. Pinaceae, Betulaceae, Fagaceae, Myrtaceae, and Dipterocarpaceae). Field studies of sporocarps have shown that ectomycorrhizal fungi are isotopically enriched in 15N relative to potential host plants, co-occurring saprotrophic fungi, and available N (Hobbie et al., 1999a; Kohzu et al., 1999; Gebauer & Taylor, 1999).
Based on field and modeling studies (Hobbie et al., 1999a,b), we proposed a simple analytical model which predicts that plant δ15N declines with greater retention of N in fungal symbionts (Hobbie et al., 2000) (Fig. 1a). This model assumes that fungal N consists of a single well-mixed pool that is equally available for transfer to plants. The enrichment in 15N of fungal tissues and the consequent depletion of host plant tissues is probably due to isotopic discrimination during synthesis of transfer compounds within fungi or during transfer reactions themselves (represented by a fractionation factor, Δf). The proportion of fungal-assimilated N that is transferred to plant hosts (referred to here as the transfer ratio, Tr) will therefore determine plant and fungal δ15N (Hobbie et al., 2000). Because foliage is generally measured in field and laboratory studies and is assumed to be representative of whole plant δ15N, we have throughout compared the δ15N of foliage to the δ15N of supplied N and to various measures of N budgets.
δ15Nfoliage = δ15Navailable N − (1 − Tr) ·▵f(Eqn 1)
An alternate model acknowledges that not all fungal N is equally available for transfer. In this model, a fraction (Ti(isolated)) of assimilated N is isolated in unavailable forms such as chitin and excreted proteins (Wessels, 1993) (Fig. 1b), with the potential for isotopic fractionation (Δi) during creation of such isolated compounds. Only assimilated N not diverted to unavailable pools joins a metabolically active pool (1 −Ti) that is available for transfer to plants, with Ti + Tf + Tr = 1.
Eqns 1 and 3 are equivalent at Ti = 0. Under plausible values of Δ i and Δ f , both scenarios predict that: mycorrhizal plants will have lower δ 15 N than nonmycorrhizal plants; ectomycorrhizal plants will have lower δ 15 N at low N availability than at high N availability if relative fungal N retention increases at low N availability (e.g. if Tf increases and Tr decreases at low N relative to high N availability); and the δ 15 N of nonmycorrhizal plants will not change with shifts in N availability. However, Fig. 1(a) does not provide any mechanism for N to be recycled from plants and fungi to the external environment.
To explicitly test the models expressed in Equations 1 and 3 against δ15N patterns of foliage, we cultured Pinus sylvestris at two levels of N supply. Seedlings were either uncolonized, colonized by Thelephora terrestris, or colonized by Suillus luteus. To create conditions of stable internal nutrient concentrations and to minimize the potential for isotopic fractionation during uptake, culture conditions were used in which N supply was closely matched to N uptake (Ingestad & Kähr, 1985). With this approach, we could treat our experimental units as open systems (quasi steady-state) (Hayes, 1982), and thereby avoid the mathematical difficulties of dealing with isotopic fractionation in closed systems using Rayleigh kinetics (Robinson et al., 1998). Because identical amounts of total N were supplied at the two N levels, possibilities for ontogenetic effects on isotopic patterns were likewise minimized. Complete biomass and N isotope budgets were constructed for the culture systems.
Materials and Methods
Surface-sterilised Scots pine seeds (P. sylvestris L.) were sown in a perlite/vermiculite (2/1, v/v) mixture moistened with a balanced nutrient solution for P. sylvestris (Ingestad & Kähr, 1985). The stock solution contained (in µM): K2SO4 (56), KNO3 (77), KH2PO4 (50), K2HPO4 (46), NH4NO3 (585), Ca(NO3)2· 4H2O (29), Mg(NO3)2·6H2O (49), H3BO3 (4), Mn(NO3)2· 4H2O (1.5), Fe(NO3)3·9H2O (2.5), Zn(NO3)2·4H2O (0.1), CuCl2·2H2O (0.1), Na2MoO4·2H2O (0.01). The pH was adjusted to 4.5. Nitrogen was the growth-limiting element and consisted of 41% ammonium and 59% nitrate. The experiment was carried out in a growth chamber with 300 µmol m−2 s−1 PAR, at least 70% relative air humidity, and with a day/night rhythm of 18/6 h and 22/15°C. 35 uniform seedlings were selected for the experiment 28 d after sowing. A sandwich technique was used to inoculate 20 seedlings (Van Tichelen & Colpaert, 2000) either with Suillus luteus (L. Fr.) Roussel or Thelephora terrestris (Ehrh.) Fr. Ten nonmycorrhizal seedlings followed the same procedure in the absence of fungal inoculum and five plants were immediately harvested to determine %N, δ15N, N amounts, and biomass at inoculation. Plants were transferred to 70-cm3 containers filled with 4.6 g acid-washed, sieved perlite (2–4 mm particles) (Colpaert et al., 1999) 3 d later. Perlite has a low nutrient buffering capacity so plants are actually growing in a semihydroponic environment where it is possible to match nutrient addition and nutrient uptake in the plants.
Immediately after inoculation two different nutrient supply rates were applied. Nutrients were either added at a constant relative addition rate of 3% day−1 (low N treatment) or 5% day−1 (high N treatment). These nutrient regimes are suboptimal so that seedlings will adjust their relative growth rate to nutrient addition rates (Ingestad & Kähr, 1985). Nutrient weight proportions supplied to the plants in both treatments were identical (100 N: 15P: 67K: 6 Ca: 6 Mg: 19S).
Plants were harvested once the cumulative N added to each plant during the experimental period was 8.0 mg, after either 45 d (high N treatment) or 70 d (low N treatment). Plant shoots were cut. Nonabsorbed N was washed from the perlite with 200 ml of nitrogen-free nutrient solution. Subsequently, containers were centrifuged at 135 g for 30 s in order to remove most of the solution retained in the perlite. Roots and perlite were pulled out of the containers and roots separated from the perlite (Colpaert et al., 1999). Fine roots were detached from coarse roots, mycorrhizas being included in the former fraction. Some samples of pure mycelia were also collected from drainholes in the low N treatments. Subsamples of perlite were frozen in liquid nitrogen for ergosterol determinations. Ergosterol was analysed by HPLC and results were converted to fungal biomass using conversion factors of 3.0 and 5.9 mg ergosterol/g biomass for T. terrestris and S. luteus, respectively (Colpaert et al., 1999). D. wt of foliage, stems, and roots were determined. Dried plant and perlite material was ground with a ball mill at 200 Hz for 2 min. Samples were stored at room temperature prior to stable isotope analyses.
Nitrogen isotope and concentration measurements were determined at the Max Planck Institute for Biogeochemistry in Jena, Germany, using a Finnigan Delta-Plus linked to a continuous-flow Carlo Erba elemental analyser. Samples were analysed in single-element mode (just N), and were referenced against concurrently run caffeine and acetanilide standards. The precision of standards was ± 0.1‰. The δ15N of available N was calculated as the weighted average of the nitrogen-containing salts in the nutrient medium. For ammonium nitrate, the ammonium and nitrate were separated by passing a solution of ammonium nitrate through a cation exchange resin (AG 50 W-X8, Bio-Rad Laboratories, Hercules, California, USA), where ammonium was retained. The ammonium-containing ion exchange resin beads were dried and used directly for continuous-flow isotope ratio mass spectrometry (CF-IRMS) according to techniques described by Lehmann et al. (2001). Nitrate in the remaining solution was passed through an anion exchange resin (AG 1-X8, Bio-Rad Laboratories) and converted to silver nitrate using techniques described by Silva et al. (2000). The silver nitrate was also directly analyzed using CF-IRMS (B. Mayer, pers. comm.).
We calculated the amount of fungal N on fine roots (Nfungi) by assuming that the N concentrations were equivalent in needles and in the plant component of colonized fine roots (as seen in nonmycorrhizal pines in our study). Therefore, increases in N concentration of fine roots compared to needles with colonization were attributed to fungal N. The total N in fine roots (Nfineroots) equals the fungal N plus N in the root component of fine roots (Nroot):
Nfineroots = Nfungi + Nroot(Eqn 4)
Because Nx = %Nx· Biomassx, (4) can be expressed as:
We assumed that the N concentration in fungi on fine roots was the same as measured on pure mycelia collected from low N treatments.
Nfungi = Biomassfungi· %Nfungi(Eqn 10)
The root N in fine roots is the difference between the total N in fine roots and the N in fungi.
Nroot = Nfineroots − Nfungi(Eqn 11)
Similarly, the plant biomass in fine roots is the difference between total fine root biomass and the calculated fungal biomass (Eqn 6). The isotopic signature of fungi on fine roots is calculated from an isotopic mass balance. We assume that the isotopic signature of root material in fine roots of mycorrhizal plants was the same as foliage (as seen in nonmycorrhizal pine in our study) (Eqn 12).
All reported results (%N, δ15N, N amounts, and biomass) were corrected for initial values at inoculation. Results were compared with t-tests or with a two-factor ANOVA (N × mycorrhizal associate) with a Tukey post-hoc test at P= 0.05.
High N treatments significantly increased the %N of all plant pools compared to low N treatments except for fine roots (Table 1), and had the opposite effect on %N of perlite. Higher %N in perlite indicated either increased N retention in extraradical hyphae or increased N loss from fungal or plant cells.
Table 1. Effects of mycorrhizal associate and nitrogen availability on system %N and δ 15 N by ANOVA. A post-hoc Tukey-Kramer test was used at P = 0.05
Abbreviations: N = nitrogen, F = fungal associate, H = high nitrogen treatment, L = low nitrogen treatment, non = nonmycorrhizal, Sl=Suillus luteus, Tt=Thelephora terrestris.
H > L
H > L
H > L
L > H
Tt > non
Sl , Tt > non
Sl , Tt > non
H > L
H > L
non > Sl,Tt
non > Sl,Tt
Sl , Tt > non
Sl > Tt > non
Patterns of %N across plant pools differed for mycorrhizal and nonmycorrhizal treatments. In nonmycorrhizal treatments, nitrogen concentrations in fine roots and needles were similar (at low N, 1.55 ± 0.04% for fine roots and 1.50 ± 0.13% for needles, and at high N, 1.76 ± 0.04% for fine roots and 1.68 ± 0.03% for needles) and considerably higher than %N in stems (0.60 ± 0.03% and 1.33 ± 0.10% at low and high N, respectively) and coarse roots (0.82 ± 0.07% and 0.95 ± 0.02% at low and high N, respectively). By contrast, fine roots in mycorrhizal treatments were elevated in %N by 0.6% to 1.0% relative to needles. Overall, colonization by mycorrhizal fungi increased %N of fine roots and perlite (Table 1), but did not affect %N of needles or stems.
It was only possible to collect pure mycelia from low N treatments. The %N of mycelia did not differ between Suillus and Thelephora (mean, 3.8% N), but were significantly higher for both fungal species than for any plant tissue.
Biomass and N budgets
A greater proportion of biomass was allocated below-ground in low N treatments than in high N treatments, and this effect was independent of mycorrhizal association (Table 2). Fungal biomass on fine roots was calculated using Eqn 9, whereas fungal biomass in perlite (extraradical hyphae) was estimated from ergosterol concentrations. At high N supply, allocation to fungi (in perlite and on fine roots) was higher for Thelephora than for Suillus, but this effect disappeared at low N supply. Fungal biomass in fine roots varied twofold across the four mycorrhizal treatments and extraradical hyphal biomass varied sixfold.
Table 2. Biomass budget for Pinus sylvestris
Fungi on fine roots
Fungi in perlite
Data are means (mg) ± SE (n = 5). Designations as in Table 1. Fungi and plant biomass in mycorrhizal fine roots separated using eqns 9 and 10 in Materials and Methods. Fungi in perlite calculated from ergosterol measurements and appropriate conversion factors for T. terrestris or S. luteus. Values followed by different letters within a treatment indicate significant differences at P < 0.05 using Tukey's post-hoc test for plant data and using t-tests for fungal data.
High N (45)
226 ± 13a
228 ± 19a
186 ± 10ab
184 ± 8ab
29 ± 3a
4 ± 1a
185 ± 8b
173 ± 9b
51 ± 5b
11 ± 1b
Low N (70)
217 ± 13a
345 ± 11a
169 ± 3b
230 ± 5b
57 ± 7a
26 ± 3a
147 ± 8b
229 ± 11b
64 ± 7a
18 ± 2a
On average, between 96% (high N treatments) and 98% (low N treatments) of the applied N was recovered. Patterns in N budgets were generally similar to those of biomass budgets, with a shift of N below-ground in low N treatments compared to high N treatments (Table 3). Allocation of N above-ground and to roots was higher in uncolonized than in colonized treatments. With the assumption that the %N of all extraradical hyphae was the same as measured values (3.8%) in low N treatments, we estimated that the proportion of total perlite N derived from extraradical hyphae for Suillus was 24% at high N and 84% at low N. Equivalent values for Thelephora were 59% and 72%. The remaining perlite N could be attributed to exudates, excreted proteins, debris of root cortex cells and dead hyphae. This pool accounted for between 2.4% and 6.0% of total added N in mycorrhizal treatments, and between 6% (high N) and 10% (low N) in nonmycorrhizal treatments. The quantity of degraded root hairs and cortex cells was visually higher in perlite of nonmycorrhizal treatments than in mycorrhizal ones. We calculated the proportion of system N that was in plant components (Tr) for Thelephora at 70 ± 2% at high N and 62 ± 6% at low N and for Suillus at 79 ± 4% (high N) and 58 ± 8% (low N).
Table 3. Nitrogen budget for Pinus sylvestris
Fungi on fine roots
Data are means (mg) ± standard error (SE) (n = 5). Figures in parentheses give the number of days since the N treatment. Above-ground is needles plus stems, roots is coarse roots plus plant portion of fine roots. Nitrogen in mycorrhizal fungi on fine roots calculated from eqn 7, nitrogen in plants of mycorrhizal fine roots calculated from eqn 8. Values followed by different letters within a treatment indicate significant differences at P < 0.05 using Tukey's posthoc test for plant and perlite data and using t-tests for fungal data.
High N (45)
3.74 ± 0.16a
3.34 ± 0.27a
0.51 ± 0.02a
3.13 ± 0.18b
2.51 ± 0.14b
1.11 ± 0.12a
0.63 ± 0.04b
3.09 ± 0.13b
2.34 ± 0.12b
1.86 ± 0.17b
0.71 ± 0.03b
Low N (70)
3.05 ± 0.16a
4.41 ± 0.10a
0.82 ± 0.08a
2.28 ± 0.06b
2.20 ± 0.10b
2.24 ± 0.28a
1.18 ± 0.08b
2.36 ± 0.11b
2.90 ± 0.23c
2.34 ± 0.25a
0.95 ± 0.03ab
Colonization by mycorrhizal fungi significantly altered plant δ15N patterns (Table 4). In addition, N availability altered δ15N patterns in mycorrhizal plants but not in nonmycorrhizal plants. Relative to nonmycorrhizal plants, needles of mycorrhizal plants were depleted in 15N in high N treatments by 0.7‰ (P = 0.011) and depleted under low N treatments by 2.4‰ (P < 0.001). Needles of Thelephora-colonized plants were 1.2‰ more depleted under low N treatments than under high N treatments (P = 0.002), whereas needles of Suillus-colonized plants were 2.5‰ more depleted (P < 0.001). The δ15N of needles of uncolonized plants only changed 0.2‰ between the two N levels (P = 0.162). In contrast to results for needles, mycorrhizal fine roots were enriched in 15N relative to nonmycorrhizal fine roots by an average of 0.6‰ at high N availability (P = 0.023) and 1.3‰ at low N availability (P < 0.001). The greatest difference between the two low N mycorrhizal treatments was in the δ15N of mycelia, with mycelia high in δ15N in Suillus (3.4 ± 0.6‰, n= 3) but low in Thelephora (−1.0 ± 0.4‰, n= 5) (t-test, df = 6, P < 0.001, difference = 4.4‰). Perlite was enriched about 2.4‰ in mycorrhizal compared to nonmycorrhizal treatments, and was enriched by 1.1‰ in high N compared to low N treatments.
Table 4. Patterns in d 15 N of system components of Pinus sylvestris cultures, as affected by N supply rate (3% day −1 , low N; 5% day −1 , high N) and mycorrhizal status. Data are means (‰) ± SE ( n = 5). Values followed by different letters within a treatment indicate significant differences using Tukey's posthoc test at P < 0.05
2.15 ± 0.10a
2.07 ± 0.25
1.83 ± 0.42
1.95 ± 0.12a
4.23 ± 0.30a
1.27 ± 0.22b
1.39 ± 0.39
2.45 ± 0.30
2.89 ± 0.12b
4.63 ± 0.19a
1.62 ± 0.22ab
1.17 ± 0.37
1.53 ± 0.37
2.18 ± 0.14a
5.52 ± 0.11b
1.97 ± 0.06a
2.67 ± 0.51a
3.05 ± 0.88
1.35 ± 0.09a
1.61 ± 0.34a
0.07 ± 0.13b
0.57 ± 0.87ab
2.31 ± 0.62
2.28 ± 0.08b
2.38 ± 0.18ab
−0.84 ± 0.31c
−0.38 ± 0.44b
1.97 ± 0.76
3.09 ± 0.13c
3.09 ± 0.08c
Fungal δ15N on fine roots
We calculated δ15N values for the fungal component of mycorrhizal roots assuming that the δ15N of the plant component of fine roots equaled that of needles. The δ15N of Thelephora on fine roots was calculated at 5.2 ± 0.2‰ at high N and 4.8 ± 0.7‰ at low N. The equivalent values for Suillus were 5.9 ± 0.6‰ and 5.9 ± 0.7‰. In high N treatments, the calculated δ15N of fungal matter on fine roots was 0.6‰ (Thelephora) and 0.4‰ (Suillus) depleted relative to perlite N.
Fungal mycelia collected from drainholes in low N treatments provided the only direct estimate of fungal δ15N. Mycelia of Suillus were 2.5‰ depleted relative to estimated δ15N of fungal matter on fine roots, and were 0.3‰ enriched relative to the δ15N of perlite. Mycelia of Thelephora were 5.8‰ depleted relative to estimated δ15N of fungal matter on fine roots, and were 3.4‰ depleted relative to the δ15N of perlite. Because a large proportion of perlite N in low N treatments was probably fungal hyphae (estimated at 72% for Thelephora and 84% for Suillus), the δ15N of these extraradical hyphae should resemble that of bulk perlite δ15N, unless the residual perlite N dramatically differed in δ15N from fungal N. Our results therefore suggest that bulk extraradical hyphae of Thelephora probably differed in δ15N by several per mille (‰) from drainhole-collected hyphae, whereas Suillus-collected hyphae were probably representative in δ15N of bulk extraradical hyphae.
Correlations and regressions
In comparisons across treatments, several calculated values were strongly correlated. In mycorrhizal treatments, fungal biomass in perlite was positively correlated with total N in perlite (r2 = 0.77, n= 20, P < 0.001, data not shown) and was negatively correlated with needle δ15N (r2 = 0.90, n= 20, P < 0.001, Fig. 2). Fungal N in fine roots was also negatively correlated with needle δ15N but the relationship was considerably weaker (r2 = 0.41, n= 20, P= 0.002). Fine root δ15N and needle δ15N were negatively correlated (r2 = 0.43, n= 30, P < 0.001, Fig. 3), with the correlation significant at low N (r2 = 0.92, n= 15, P < 0.001) but not at high N (r2 = 0.25, n= 15, P= 0.057).
Finally, to test our two possible mechanisms for how mycorrhizal fungi and nitrogen budgets control plant δ15N patterns (Fig. 1), we regressed needle δ15N against the proportion of system N that was in plants (Tr, Eqn 1, Fig. 4). We also did a multiple regression of needle δ15N against values for the proportion of system N in perlite (assumed to be Ti, the unavailable fraction), and the derived variable Tf/(Tf + Tr), where Tf is estimated as the fungal tissue on fine roots (assumed to be potentially available for transfer), and where Tr is the proportion of system N in plants (Eqn 3). Both regressions were highly significant (P < 0.001), but the multiple regression explained a greater proportion of the variance (adjusted r2 = 0.71 vs 0.59 for the simple regression). From the simple regression, the isotopic discrimination in making transfer compounds (▵f) was estimated at 9‰ (Fig. 4), in close agreement with previous estimates of 10‰ for this fractionation in a modeling study of foliar δ15N patterns in mycorrhizal spruce (Hobbie et al., 2000). From the multiple regression using Eqn 3, the calculated value of ▵f was 4.33 ± 0.57‰, the calculated value of ▵i was −23.07 ± 5.75‰, and the intercept was 4.25 ± 0.57‰.
Mycorrhizal colonization generally increases retention of N in root systems (Ingestad & Kähr, 1985; Colpaert et al., 1996). In our study, N retention varied between the two fungal symbionts, depending on N availability. The fungal biomass determined the extent of this effect. Both fungi grew equally well in the low N treatment, whereas Suillus grew considerably less than Thelephora at high N supply. The factors controlling sensitivity of mycorrhizal taxa to shifts in N supply are not fully known, although the ability to retain a large proportion of assimilated N in fungal tissues appears important for maintaining fungal biomass (Wallander et al., 1999).
Differences in N cycling of fungal taxa are of obvious importance in understanding shifts in community composition with N deposition (Lilleskov & Bruns, 2001). Several studies have previously observed differential sensitivity of mycorrhizal taxa to N availability in both field and laboratory studies. In a culture study of mycorrhizal P. sylvestris, Wallander & Nylund (1992) reported that Suillus was more sensitive than Laccaria or Hebeloma to increases in N availability, as measured by decreased extraradical hyphal abundance and decreased colonization on fine roots. In agreement with our results, they also observed that extraradical hyphae were more sensitive to shifts in N availability than mycorrhizal fungi directly associated with fine roots. Numerous field studies have also reported that fruiting of ectomycorrhizal fungi declines with increased N availability, in particular in protein-using taxa (Lilleskov et al., 2002). However, difficulties in determining whether hyphae in the field are mycorrhizal or saprotrophic has largely precluded direct assessment of declines in production of extraradical hyphae (Wallenda & Kottke, 1998).
Confounding factors affecting δ15N patterns
Because of isotopic fractionation during nitrate reduction, partial reduction of nitrate in roots followed by export of the remaining nitrate to shoots increases foliar δ15N relative to root δ15N (Robinson et al., 1998; Evans, 2001). This did not appear to influence our results, as roots and foliage were similar isotopically in nonmycorrhizal treatments. Since conifers usually assimilate nitrate in their roots (Smirnoff et al., 1984; Kronzucker et al., 1997), most nitrate is probably reduced to organic N either by the fungal symbiont or in the roots, and therefore plant components should not differ in δ15N because of nitrate reduction.
The desire to have balanced levels of all nutrients and constant pH in the culture solution required avoiding the use of large amounts of mineral N except as ammonium nitrate. As a result, we cannot completely rule out the possibility that some of the observed differences between plant components arose from preferential routing of ammonium or nitrate (of different δ15N) to particular components. However, the δ15N of the supplied ammonium (1.87‰) and nitrate (1.04‰, weighted average of mineral N 1.38‰) differed little, so differential retention of ammonium vs nitrate per se cannot explain most of the isotopic differences observed.
In culture studies that have not linked N supply rates to plant N uptake demands, isotopic fractionation against 15N on assimilation has probably affected plant δ15N values, particularly when plant δ15N is lower than source δ15N (Högberg et al., 1999; Emmerton et al., 2001). For example, a review by Handley & Raven (1992) reported that fractionation on ammonium assimilation varied between 9 and 20‰ in culture studies. However, such fractionation cannot explain our isotopic patterns since N was supplied and taken up at the same rate. In addition, such fractionation cannot explain the low δ15N signatures generally observed in mycorrhizal plants in N-limited ecosystems. Such environments should also be characterized by little fractionation on uptake because of low N concentrations at the soil-root or soil–fungal interface (Nadelhoffer & Fry, 1994; Evans, 2001).
To calculate the δ15N of fungal matter on fine roots we made two assumptions. We first assumed that under our N-limited conditions, plant tissue in mycorrhizas and fine roots had the same %N as needles. This was true here for nonmycorrhizal pines at high and low N availability, and has been previously reported in pine culture (Kohzu et al., 2000). It therefore seemed a reasonable assumption for our culture conditions. It could be tested by either an indirect measure of fungal N on fine roots (e.g. through measuring ergosterol on fine roots and establishing a relationship between fungal N and ergosterol in pure culture), or alternately, through direct isolation of plant matter from fine roots using careful dissection.
Our second assumption, that the δ15N of plant matter on fine roots equaled the δ15N of needles, is more problematic. The similar %N on coarse roots across all treatments indicated that fungal matter was unlikely to have been present on coarse roots, whereas much higher %N on mycorrhizal fine roots than on nonmycorrhizal fine roots indicated the presence of fungal matter. The δ15N of needles, coarse roots and fine roots did not differ in nonmycorrhizal treatments, although coarse roots were rather variable. Because fine roots and foliage should both be stronger sinks for N than coarse roots, they will probably be closer isotopically to each other than to coarse roots. If the plant component of fine roots were assumed to equal coarse roots rather than foliage in δ15N, then the calculated fungal δ15N using Eqn 12 would be 2.6 ± 0.9‰ at high N and 4.3 ± 1.3‰ at low N for Thelephora, and 3.9 ± 1.1‰ at high N and 3.7 ± 1.5‰ at low N for Suillus. These values are from 0.1‰ to 1.9‰ enriched relative to values calculated under the assumption that fine root and foliar δ15N are identical.
Our assumption that fungal N in fine roots corresponded to Tf and N in perlite corresponded to Ti is obviously a simplification. In particular, a substantial fraction of the fungal N on fine roots (here assigned to Tf) may be in unavailable forms such as chitin or bound proteins. In Fig. 5, we show a sensitivity analysis of the shifts in Tf and Ti as the calculated value of Ti for each of the 20 mycorrhizal cultures was systematically increased by 10–100%. The calculated value of Δf from regression analyses decreased slightly over this range (from 4.3‰ to 3.8‰), whereas the value of Δi increased substantially, from −23.1‰ to −14‰. Separating fungal pools into different chemical fractions (e.g. protein, amino acids, and chitin) may improve our ability to partition fungal N between unavailable vs available pools.
We have also neglected direct N uptake by plants in our calculations. We are unsure how uptake could be partitioned between plants and fungi without substantial disturbance to the experimental system and loss of uniform exponential growth conditions for both plant and fungal components.
Interpreting δ15N patterns
Our experiment clearly demonstrated that N availability can influence plant δ15N under conditions of constant source δ15N when plants are colonized by mycorrhizal fungi. Furthermore, the results demonstrated close links among plant δ15N, fungal biomass and N allocation patterns of the symbiotic partners.
Lower rates of N supply in our culture experiments resulted in greater allocation to mycorrhizal fungi and a corresponding decrease in needle δ15N. Our results clearly showed that increased mycorrhizal colonization increased the isotopic depletion of needles relative to source N. In contrast to our maximum depletion of needles in mycorrhizal plants relative to nonmycorrhizal plants of 2.5‰ (Suillus at low N), 15N depletions of greater than 5‰ in foliage of mycorrhizal plants relative to nonmycorrhizal plants are common in field studies at sites of low N availability (Michelsen et al., 1996, 1998; Nadelhoffer et al., 1996). It is possible that direct N uptake by plants is more important in the culture experiment than in the field, in particular during the first weeks after inoculation when the hyphal network is not yet fully developed. Sequential harvests over a longer period than used here would be useful to study temporal development of 15N depletion in shoots. Another explanation for the discrepancy is that rates of N supply are sometimes lower in the field than in the low N treatment of our culture study. For example, our low N supply rate (3% day−1) produced foliage of about 1.5% N compared to values as low as 1% in boreal ecosystems (Schulze et al., 1994). An alternate explanation is that large differences in δ15N in field studies reflect in part uptake of different N forms by mycorrhizal vs nonmycorrhizal plants (e.g. organic N vs mineral N) of differing δ15N signatures (Michelsen et al., 1996, 1998). Although this hypothesis has some merit and has been favorably viewed within the mycorrhizal community (Aerts, 2002; van der Heijden & Sanders, 2002), no data on the δ15N of simple organic N in soils currently exist.
The negative correlation between fine root and foliage δ15N (Fig. 3) suggested that mycorrhizal transfer processes simultaneously created opposite isotopic patterns in mycorrhizal fine roots and foliage. In a similar fashion, the strongly negative correlation between needle δ15N and fungal biomass in perlite (Fig. 2) suggested that retention of 15N-enriched N by fungi in extraradical hyphae was highly correlated with the ultimate mechanism controlling foliar δ15N.
As previously discussed, a substantial fraction of the fungal N on fine roots may actually be unavailable for transfer to plants. Because needle δ15N in Eqn 2 varies as the quantity Ti· Δi varies, underestimating Ti results in overestimating the magnitude of Δi. The negative value of Δi means that N entering unavailable pools (e.g. excreted proteins) is enriched in 15N relative to the source N. The rather low predicted value for Δi of −23‰ compared to isotopic fractionation during mycorrhizal transfer, compared to isotopic differences among different N-containing compounds, or compared to isotopic fractionation during enzymatic reactions involving N transformations (Table 5) suggests that our values for Ti and Δi may be too low by a factor of two or more. A sensitivity analysis (Fig. 5) of shifts in Δi and Δf as Ti is increased and Tf is decreased indicates that increasing Ti will decrease the calculated value for Δi to more plausible values.
Table 5. Nitrogen isotope fractionations likely to affect δ 15 N patterns in plant-mycorrhizal systems. Abbreviations: oxoglu, 2-oxoglutarate; oxal, oxaloacetate
The high 15N enrichment of perlite in high N treatments, including nonmycorrhizal treatments, indicated that N lost to the media under these conditions was isotopically enriched relative to all measured biomass pools. Similar losses of 15N-enriched substances to the media could account for δ15N patterns in cultured nonmycorrhizal barley (Robinson et al., 2000). Secretion of proteins or amino acids is a probable source of isotopically enriched N. Such secretion by mycorrhizal fungi can be substantial: Ahmad et al. (1990) reported that 20–25% of N assimilated in ammonium- and nitrate-supplied Laccaria cultures was returned as amino acids to the media. This source of 15N-enriched nitrogen, together with 15N-enriched fungal matter from mycorrhizae (Högberg et al., 1996), may contribute to the enrichment of deeper soil horizons in 15N relative to surface litter layers. From our data, secretion of 15N-enriched N appears probable in both plants and mycorrhizal fungi.
From our results, including N fluxes unavailable to plants (Ti) appears desirable to predict system-wide isotopic patterns. For example, because Eqn 1 includes no mechanism for N loss to the media, it cannot predict δ15N patterns of pools such as perlite that include exudates. Here, foliar δ15N was strongly correlated with the proportion of fungal N that was in extraradical hyphae and therefore presumably unavailable for transfer. Chitin, structurally bound proteins, excreted enzymes (Wessels, 1993) and excreted amino acids are probable main constituents of N unavailable to plants.
What are the implications for interpreting δ15N measurements? Our results suggest that shifts in foliar δ15N may be a good marker of shifts in nutrient and biomass allocation within mycorrhizal symbioses under certain conditions. We accordingly explain field reports of higher foliar δ15N with higher N availability and higher levels of N deposition (Högberg et al., 1996; Jung et al., 1997; Michelsen et al., 1998; Hobbie et al., 2000) as decreased retention of N by the fungal partner in mycorrhizal symbioses. Conversely, foliar δ15N of mycorrhizal plants is generally low in low productivity and N-limited systems (Nadelhoffer et al., 1996; Michelsen et al., 1998; Hobbie et al., 1999a). Below-ground carbon allocation as a fraction of total net primary production increases in low productivity systems (Chapin, 1980; Ingestad & Kähr, 1985; Raich & Nadelhoffer, 1989). Combined, these two facts suggest that much of the relative increase in below-ground allocation under these conditions reflects increased carbon flux to and N retention by mycorrhizal fungi, and not necessarily increased carbon flux to roots alone.
The results indicate that δ15N patterns in plant-mycorrhizal systems are explained better with a two-pool model of fungi than with a single pool. Values predicted for Δf differed between simple and multiple regressions of foliar δ15N, presumably because the patterns of Ti and Tf among the different treatments were perhaps qualitatively correct, but not quantitatively correct. From the sensitivity analysis (Fig. 5), it appears probable that a portion of the fungal N on fine roots was actually unavailable for transfer, so should more properly be assigned to Ti. Therefore, Ti is probably underestimated and Tf overestimated. The universal depletion in 15N of chitin relative to protein and amino acids in fungi, insects, and other organisms (Table 5), and the very different roles of chitin and protein in fungi, suggests that compound-specific analyses will be needed to better understand how bulk δ15N patterns in plant-mycorrhizal systems are created. Such compound-specific analyses have much to teach us about the potential insights into C and N dynamics of δ15N measurements in the field.
This research was fostered by postdoctoral support for Erik Hobbie from Dave Schimel and the Max Planck Institute for Biogeochemistry. The contribution of Bernhard Mayer in isolating mineral N forms for isotopic analysis is gratefully acknowledged. Samples were analyzed for %N and 15N content by Willy Brand and Heike Geilmann. John Aber, Hormoz BassiriRad, Knute Nadelhoffer and two anonymous reviewers contributed constructive comments on the manuscript. This work was supported in part by the Fund for Scientific Research-Flanders (Belgium) (project G. 0033. 03).