1. Arbuscular mycorrhizal fungi (AMF) have been proposed as a mechanism to reduce nutrient inputs in agriculture, thereby reducing costs and increasing environmental sustainability. However, before this can be achieved, we need to gain a better understanding of the importance of the prolonged selective pressures acting on indigenous AMF communities.
2. Much research concentrates on short-term ecological soil × plant × AMF interactions. However, we have little understanding of how long-term manipulations of abiotic conditions can be strong selection agents for AMF communities. Here, we ask how the long-term management of soil fertility and fertilizer use can influence the AM symbiosis. More specifically, we investigated whether 70 years of consistently imposed nutrient limitations affected the structure and symbiotic functioning of indigenous AMF communities.
3. Using the long-term Static Nutrient Deficiency Experiment carried out since 1937 in Thyrow, Germany, with and without nitrogen (N) and phosphorus (P) additions, we addressed the following questions: (i) Do different soil fertilizer treatments affect the overall abundance and diversity of indigenous AMF in an agricultural field; and (ii) Does the depletion of a nutrient select for an enhanced AMF ability to supply the deficient nutrient?
4. We assessed AMF spore diversity in the field and established a common garden experiment where soil nutrient treatments were calibrated against those in the long-term field experiment. For each soil nutrient treatment, we compared the growth responses of barley plants to the indigenous AMF communities isolated from the different soil fertilization treatments in the field.
5. We found that the long-term use of specific soil fertilization treatments altered the effects of the AMF symbiosis on plant and fungal growth. Consistent with the optimal foraging theory, AMF from N- or P-deficient soils grew larger but reduced plant growth more in those conditions relative to AMF isolated from non-deficient soils. This could result from both community-level changes and/or adaptations within species.
6. Thus, we propose that the ongoing agronomic management of abiotic selective pressures such as soil fertility needs to be considered as a strong determinant of AMF symbiotic functioning.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
In this study, we investigated whether long-term nutrient limitations can result in both community-level changes and adaptations within indigenous AMF. Using a long-term field experiment established in 1937, we addressed the following questions: (i) Do different soil nutrient treatments affect the overall abundance and diversity of indigenous AMF in an agricultural field?; and (ii) Does the depletion of a nutrient select for an enhanced AMF ability to supply the limiting nutrient?
We measured AMF diversity in the field and established a common garden experiment where soil nutrient treatments were calibrated against those in the field to test for AMF and plant growth responses to indigenous AMF communities isolated from the field. Based on the trade balance model (Johnson 2010), which predicts the outcome of the AMF symbiosis along the parasitism–mutualism continuum depending on how the interactions between N and P availability affect C supply and demand between partners, we hypothesized that: (i) AMF diversity is affected by different nutrient-deficient fertilization regimes; and that (ii) AMF communities from soil under long-term N or P deficiency are better able to enhance plant growth under N- or P-deficient conditions relative to AMF isolated from non-deficient soils.
Materials and methods
Field site characteristics and soil sampling
The study site was the long-term Static Nutrient Deficiency Experiment established by Kurt Opitz in 1937 in Thyrow (south of Berlin, Germany; 52°15′08·12″N 13°14′08·26″E) and in which fertilization treatments were left unchanged (i.e. managed continuously) ever since (Ellmer et al. 2000). The long-term experiment consists of a Latin square design with eight fertilizer treatments comprised of (1) Unfertilized, (2) Cattle manure, (3) NPKCa + Cattle manure, (4) NPKCa, (5) NPK (Minus Ca), (6) NPCa (Minus K), (7) NKCa (Minus P) and (8) PKCa (Minus N) (Table 1). Treatments were replicated four times in 10 × 7·2 m plots. The crop rotation consists of potato–barley–maize–barley applied equally to all treatments and plots since the onset in 1937. Nitrogen, P and K were added at the rate of 60, 24 and 100 kg ha−1, respectively (fertilizers supplied by K + S Aktiengesellschaft, Kassel, Germany). Calcium was applied as CaCO3 at a rate of 250 kg ha−1 every 3 or 4 years. The soil in Thyrow is an Albic Luvisol whose physicochemical properties were reported in detail by Langer & Klimanek (2006) and Baumecker, Ellmer & Köhn (2009). The background soil texture is sandy (83·1, 16·2 and 2·7% sand, silt and clay, respectively), and the nutrient concentrations are considered limiting for plant growth (Table S1, Supporting Information). Total N (Kjeldhal) was reported by Langer & Klimanek (2006) as being 0·04% in the control plots. On 25 June 2008, five soil cores (top 38·5 cm2 × 30 cm) were randomly collected from each plot of the Unfertilized, NPKCa, Minus P and Minus N treatments. Soil samples were air-dried and immediately used in the controlled environment (CE) experiment described elsewhere.
Table 1. Experimental layout of the Static Nutrient Deficiency experiment established in Thyrow, Germany, in 1937
Diversity and abundance of AMF communities in the field (Hypothesis 1)
To evaluate the diversity and abundance of AMF spore communities present in field soils, additional soil cores were collected from the treatments Unfertilized, Cattle manure, NPKCa + Cattle manure, NPKCa, Minus P and Minus N. These samples were air-dried before being analysed for AMF hyphal length as a measure of fungal abundance in the soil (Tennant 1975; Jakobsen, Abbott & Robson 1992). There is no specific stain for AMF hyphae; therefore, identification relies on morphological criteria, described elsewhere (Rillig et al. 1998; Rillig, Field & Allen 1999). These include absence of regular septation, knobby appearance of hyphal surfaces and connection to AMF spores for AMF hyphae. Non-AMF hyphae in our soil had the following traits: melanization, large diameters (∼ >30 μm), regular septation and clamp connections.
Diversity of AMF communities was determined through morphological separation (i.e. by morphotypes) and quantification (Gerdemann & Nicolson 1963) of field-collected spores. Accurate identification of field spores was difficult owing to accumulated debris and degradation of spore walls, which are the main characters used for identification. Therefore, morphotypes were grouped according to colour, size and surface ornamentation under a 50× magnification. The number of intact and non-parasitized spores per morphotype was used to calculate rank–abundance curves for each soil fertilization treatment.
As taxonomic identification of the field-collected spores was difficult, we established trap cultures for each of the six treatments. The objective was to use cleaner spores from these cultures to help identify morphotypes isolated from the field soil and to further evaluate differences in richness among treatments (i.e. presence/absence; Table S2, Supporting Information). It was also our objective to establish pure cultures and eliminate putative environmental maternal effects for future experiments. Trap cultures were prepared by layering 150 g of field soil between sterilized sand (autoclaved twice at 121 °C for 60 min) in each of four 1-L pots per treatment. Four sterile controls were established to test for potential contamination. Sixty seeds of Sorghum vulgare Pers. were surface sterilized (immersed in 70% alcohol for 3 min) and then added to each pot and left to grow for 13 weeks. Water and nutrients were supplied as needed using deionized water and a low-P Hoaglands solution. Spores extracted as described previously served as voucher specimens for identification. They were mounted in polyvinyl–lactoglycerol (PVLG) and PVLG/Melzer’s reagent (1 : 1) (Largent, Johnson & Watling 1977) and examined using a compound microscope (Leica Microsystems CMS GmbH, Wetzlar, Germany). Taxonomic identification was carried out following keys obtained from the International Culture Collection of Arbuscular Mycorrhizal Fungi (INVAM; http://invam.caf.wvu.edu/index.html).
Common garden experiment (Hypothesis 2)
We established an experiment in a CE chamber at the Freie Universität Berlin, Germany, from 10 July to 2 September 2008, to test the interaction between AMF Origin in terms of symbiotic functioning and long-term soil nutrient conditions. It comprised 80 experimental units consisting of 4 × 20·5 cm ‘cone-tainers’ (Stuewe and Sons Inc., Corvallis, OR, USA). These were divided into two crossed factorial treatments consisting of ‘Field-Simulated Nutrient Regime’ (Unfertilized, NPKCa, Minus P, Minus N) and ‘AMF Origin’ (non-mycorrhizal control, Unfertilized, NPKCa, Minus P, Minus N). Each treatment was replicated four times following the field design.
The CE chamber was set to 18·0 : 25·0 °C mean night/day temperatures and a 16-h photoperiod. Light conditions were attained using a combination of incandescent lamps and compact fluorescent bulbs (photosynthetically active radiation of approximately 400 μmol m−2 s−1).
Each cone-tainer was first filled with cotton pressed tightly to cover the openings at the bottom, followed by 200 mL of sterilized sand (121 °C for 20 min) and an AMF fraction, which was covered with more sterilized sand. The AMF fraction was obtained following Klironomos (2002); suspending 50 mL of field soil in 500 mL water and decanting twice through a stack of 1-mm, 250 and 38-μm sieves. Spores retained in the 38-μm sieve were rinsed with a 100 mL 10% bleach solution and water before being added to the pots. To correct for differences in non-AM microbial communities, each cone-tainer received a 1 mL filtered (20-μm mesh) washing comprised of extract from a mixture of subsamples from all field-collected soil. To compare our simulated fertilization treatments against the field conditions, for each fertilization treatment, four cone-tainers were filled with field soil from each respective field plot (hereafter field-soil treatments).
Summer barley (Hordeum vulgare L.) seeds were surface sterilized by immersing in 50% alcohol for 2 min. Three seeds were placed in each cone-tainer and covered with approximately 20 mL of sterile sand to conserve moisture. Three days later, germinated seedlings were thinned to one plant per cone-tainer, and the nutrient treatments were applied 5 days after that.
Using the same commercial fertilizers used in the field, we prepared a baseline aqueous solution that was added to all cone-tainers except to those assigned to the field-soil treatments used for comparison purposes. This baseline, consistent with the limiting nutrient conditions of the Unfertilized field plots, consisted of 4·4, 1·7 and 7·4 mg kg−1 of N, P and K, respectively. We also applied 26·6, 10·6 and/or 44·4 mg kg−1 of N, P and/or K, respectively, to containers in each appropriate nutrient treatment solutions. These concentrations simulated 60, 24 and 100 kg ha−1, respectively. Calculations on a per hectare basis assumed 1500 m3 (i.e. top 15 cm) and a soil bulk density of 1·5 g cm−3.
Plants were watered daily to near field capacity with care being taken to avoid leaching. Cone-tainers were destructively harvested on 2 September 2008 when the barley plants had reached head emergence stage. Total shoot and root dry weight were measured (dried at 60 °C for 48 h), and plant material from the NPKCa and Minus P treatments was analysed for P concentration by dry-ashing followed by the ammonium heptamolybdate–ammonium vanadate method (Jones, Wolf & Mills 1991). To determine whether the AMF communities from P-limited soils colonized roots more extensively than those isolated from soil fertilized with all nutrients, we measured the percentage of root length colonized by AMF in NPKCa, Minus P and Field-Soil treatments according to methodology developed by McGonigle et al. (1990).
Field responses of AMF sporulation and hyphal length to fertilization regime were analysed by univariate analysis of variance (anova). Spore data were log transformed to satisfy the anova assumptions of normality and equality. Field diversity of AMF communities (i.e. of spore morphotypes) was assessed by rank–abundance curves (with relative abundance (y– axis) expressed on a log scale), which provide a visual representation of species richness and evenness. Evenness was calculated from the slope of the line fitting the ranked species; the slope steepness is inversely proportional to evenness. Analysis of rank-abundance among treatments (i.e. test whether slopes and intercepts are significantly different) was performed by linear regression using graphpad prism version 5·00 (graphpad Software, San Diego California USA, http://www.graphpad.com), which follows a method equivalent to an analysis of covariance (ancova) described by Zar (1984). Data were entered onto an XY table, with X in one column, and each of the four replicate Y columns for each of the six fertilization treatments. In addition, the Simpson’s Index of diversity (1-D) (Simpson 1949) was also calculated for these data and tested for the differences among treatments using one-way anova.
For the growth chamber experiment, shoot dry weight of the barley plants was significantly correlated with root dry weight (R =0·7236, P <0·0001) and total dry weight (R =0·9138, P <0·0001), thus suggesting that shoot weight is a good estimator of total plant dry weight (data not shown). As such, of these response variables, we only report results obtained for shoot dry weight. Data were analysed by anova, using a factorial model with the factors ‘Field-Simulated Nutrient Regime’ and ‘AMF Origin’. Shoot dry weight and P uptake data were Box–Cox- and log transformed, respectively, to satisfy the anova assumptions of normality and equality. Conversely, examination of the residuals after arcsine transforming the AMF colonization data suggested that anova assumptions were not met. Therefore, these data were analysed using nonparametric tests (Zar, 1984).
Where appropriate and unless otherwise specified, least-square means contrasts within treatments were performed or means were compared using the Tukey honestly significant difference test (P <0·05). Statistical analyses were performed with the software StatSoft, Inc. (2010).
Diversity of AMF communities in the field (Hypothesis 1)
According to our first hypothesis, AMF diversity would be affected by different nutrient-deficient fertilization regimes. The type of fertilization caused significant variation in overall AMF abundance (F5,18 = 12·74, P <0·0001 and F5,18 = 43·35, P <0·0001, for AMF hyphal length and spore number, respectively) (Fig. 1). AMF, which were found in all treatments, produced significantly more hyphae under deficient soil nutrient conditions, particularly of N, compared to non-deficient treatments. Conversely, AMF spore abundance was significantly smaller in the N-deficient treatment than in all other treatments. Overall, AMF sporulation appeared to be favoured in soil that received organic matter through the addition of manure.
Despite the morphological similarities, we were unable to identify the field-collected spores using those isolated from the trap cultures as a template. As such, we report these data separately: (i) a total of 11 distinct spore morphotypes was found in soil of the trap cultures (Table S2, Supporting Information). With the exception of Acaulospora koskei Blaszk., which was only found in the Unfertilized treatment, all other morphotypes were members of the family Glomeraceae; (ii) as for the direct field-soil spore extractions, seven distinct morphotypes were detected across all treatments and, with few exceptions, each morphotype was consistently found in all four replicate plots of each treatment (data not shown). Overall, the slopes and elevations of the rank–abundance curves (i.e. a measure of AMF diversity) were similar among treatments (F5,116 = 2·11, P <0·07 and F5,121 = 1·43, P <0·22, respectively). It was therefore possible to calculate one pooled slope and intercept for all data, which equalled −0·292 and 1·94, respectively. Such a steep slope indicates that communities were dominated by very few species (Fig. 2). The similarity of AMF diversity among treatments was also supported by the Simpson’s Index (F5,18 = 0·9447, P <0·476), whose overall value was 0·66 ± 0·0217 (Mean ± 1 SE). We observed four morphotypes in the Unfertilized treatment whereas Minus P was the most AMF rich. Generally, the same morphotype dominated in all treatments and morphotypes ranked in the same approximate order among treatments.
Common garden experiment (Hypothesis 2)
The shoot dry weight of barley was significantly affected by ‘Field-Simulated Nutrient Regime’ (F3,60 = 405·29, P <0·0001) (Table 2). Plants under deficient N conditions grew significantly less compared to the other ‘Field-Simulated Nutrient Regime’ treatments. Phosphorus-deficient conditions also significantly affected plant growth, although not to same degree as N deficiency. Identical responses were observed for the field-soil treatments, indicating that our simulated fertilizations were consistent with those found in the field (data not shown).
Table 2. Effect of Field-Simulated Nutrient Regime and AMF Origin from the Tyrow Static Nutrient Deficiency experiment on shoot biomass of barley plants at harvest
Field-Simulated Nutrient Regime
Dry shoot biomass (g)
Mean** (n = 16)
*Means (n =4) ± 1 SE followed by the same letter are not significantly different (P <0·05).
**For each row or column, means ± 1 SE followed by the same letter are not significantly different (P <0·05).
0·50 ± 0·030hi
1·60 ± 0·045a
1·52 ± 0·10ab
0·56 ± 0·023h
1·05 ± 0·135a
0·49 ± 0·018hij
1·46 ± 0·040abc
1·18 ± 0·09de
0·55 ± 0·062h
0·92 ± 0·109b
0·44 ± 0·038ijk
1·18 ± 0·025de
1·01 ± 0·11f
0·37 ± 0·027k
0·75 ± 0·095d
0·40 ± 0·025jk
1·27 ± 0·040cde
0·82 ± 0·060g
0·36 ± 0·033k
0·71 ± 0·097d
0·47 ± 0·012hij
1·31 ± 0·051bcd
1·10 ± 0·082ef
0·41 ± 0·018ijk
0·82 ± 0·103c
Mean** (n = 20)
0·46 ± 0·013c
1·36 ± 0·037a
1·13 ± 0·064b
0·45 ± 0·025c
There was a significant effect of AMF origin on barley growth (F4,60 = 22·41, P <0·0001). Plants grew significantly less in the presence of AMF communities isolated from field plots under the Minus P and Minus N fertilization regimes (Table 2). Moreover, we observed a significant ‘Field-Simulated Nutrient Regime’ × ‘AM Origin’ interaction (F12,60 = 1·93, P <0·048). In treatments where N was not deficient (i.e. long-term NPKCa and Minus P), the growth of barley plants colonized by AMF in their fertilizer conditions of origin was significantly smaller relative to the other fertilizer conditions (Fig. 3a). Such pattern was also consistent for data on P concentration and uptake, particularly for the Minus P treatments (Table 3, Fig. 3b).
Table 3. Effect of Field-Simulated Nutrient Regime and AMF Origin from the Tyrow Static Nutrient Deficiency experiment on P concentration and uptake of barley plants at harvest
P concentration (mg g−1)
P uptake (mg per plant)
Mean** (n = 8)
Mean** (n = 8)
*Means (n =4) ± 1 SE followed by the same letter are not significantly different (P <0·05).
**For each row or column, means ± 1 SE followed by the same letter are not significantly different (P <0·05).
†anova on Field-Simulated Nutrient Regime (A) and AMF Origin (B) for P concentration (A –F1,30 = 101·3, P <0·0001, B –F4,30 = 0·79, P <0·54, A × B –F4,30 = 1·97, P <0·12) and uptake (A –F1,30 = 137·3, P <0·0001, B –F4,30 = 11·44, P <0·0001, A × B –F4,30 = 5·47, P <0·002).
14·4 ± 0·68a
11·4 ± 0·67b
12·9 ± 0·72a
22·9 ± 0·46a
17·5 ± 1·92b
20·2 ± 1·36a
15·6 ± 0·30a
10·1 ± 1·05bc
12·8 ± 1·16a
22·8 ± 0·96a
11·7 ± 0·53c
17·2 ± 2·16b
15·3 ± 1·20a
11·3 ± 0·46bc
13·3 ± 0·97a
18·1 ± 1·68b
11·3 ± 1·31c
14·7 ± 1·62bc
15·4 ± 0·33a
9·44 ± 0·50c
12·4 ± 1·15a
19·6 ± 0·99ab
7·7 ± 0·66d
13·7 ± 2·31c
13·8 ± 0·46a
10·7 ± 0·53bc
12·2 ± 0·67a
18·2 ± 1·15b
11·6 ± 0·70c
14·9 ± 1·38b
Mean** (n = 20)
14·9 ± 0·31a
10·6 ± 0·32b
20·3 ± 0·66a
12·0 ± 0·85b
The levels of AMF colonization observed in plants inoculated with AMF fractions were significantly smaller compared with the pooled field-soil treatments (Fig. 4). We observed a marginally significant ‘Field-Simulated Nutrient Regime’ effect on AMF colonization (Mann–Whitney U test, Z = −1·57, P <0·1) (Fig. 4). There was a significant effect of AMF Origin on the level of AMF colonization (Kruskal–Wallis test, H4, N = 40 = 19·11388 P <0·0007). Moreover, AMF isolated from the Minus P plots produced the largest amount of colonization relative to other treatments.
Contrary to our first hypothesis, we did not find a striking difference in AMF diversity among treatments despite 70 years of different soil fertilization regimes. The overall AMF morphotype richness was low, consistently dominated by the same morphotype and diversity, both as evaluated by the slopes of rank–abundance curves and the Simpson’s Index, was similar among treatments. However, despite these similarities, since we only considered AMF that sporulate, it is possible that the AMF communities could be either more or less different if other measures were taken. Recent molecular work using next generation sequencing shows that there is considerably more AMF diversity than might be detected from spore sampling alone (e.g. Öpik et al. 2009). Furthermore, we did find significant differences in total spore and extraradical hyphal abundance among fertilization treatments, which may affect symbiotic function. Changes in AMF community structure among fertilization treatments have been found in field and greenhouse studies, showing either population- or community-level sensitivities and shifts in response to P and N additions (e.g. Johnson 1993; De Miranda & Harris 1994; Egerton-Warburton & Allen 2000). Conversely, Johnson (1993) found similar levels of AMF spore diversity in fertilized and unfertilized soils from an 8-year field experiment. Consistent with our data, eleven AMF were observed, with more than 96% of the spores belonging to a single species (i.e. Glomus aggregatum N.C. Schenck & G.S. Sm). In our study, soil left unfertilized for 70 years contained a viable AMF community, which might have been supported either by incidental weeds occurring in the plots or by the crop seedlings that year after year die shortly after seeding because of nutrient deficiency.
The common garden experiment established to test our second hypothesis provides evidence that AMF communities may either have structurally changed or contain individual isolates that adapted as a result of 70 years of continuous manipulations of soil fertility. This is consistent with the recent results reported by Johnson et al. (2010), showing local adaptations of the AMF symbiosis in different P-deficient natural grassland sites. However, opposite to our hypothesis, deficient soil nutrient conditions did not select for the increased ability of AMF to supply that nutrient to its host. On the contrary, AMF were more parasitic in their respective soil nutrient conditions relative to other fertilization regimes. In soil treatments where N was not deficient (i.e. NPKCa and Minus P), plants grown with an AMF community isolated from field soil under the same fertilizer treatment grew significantly less than with AMF from other treatments, which were not exposed to long-term selection for those nutrient conditions. This effect was more pronounced for P-deficient soil than for soil receiving all nutrients (i.e. NPKCa). Conversely, in N-deficient soils, conditions were so limiting for plant growth, and biomass reductions so high, that they might have hindered the capacity to detect any functional adaptations.
Why were the AMF symbioses more parasitic in their own fertilization conditions? According to the trade balance model, parasitic outcomes in the AMF symbiosis might be related to the balance of N, P and C trade between partners (Johnson 2010). If N and P are supplied in sufficient amounts, there is plenty of C for the fungal partner but no benefit to be gained by the plant host in terms of AMF-mediated P uptake, thereby potentially leading to parasitism. The parasitic effect observed in the NPKCa treatment, in which macronutrients were supplied in adequate amounts, is consistent with this hypothesis. Other studies also provide evidence for the reductions in AMF diversity and abundance because of soil fertilization as well as selection for less mutualistic fungi, which succeed under conditions of low necessity to their hosts (Johnson 1993; Treseder & Allen 2002).
The trade balance model predicts a mutualistic outcome whenever N is sufficient and P is not, which has been empirically supported (e.g. Valentine, Osborne & Mitchell 2001; Johnson et al. 2010). However, although Johnson et al. (2010), found that soil nutrient deficiencies can, indeed, select for a more mutualistic AMF symbiosis, that effect was not universal. If long-term selective pressures can result in AMF better able to draw C from their host under impoverished P conditions, they will continue to grow at the expense of the host, scavenging for the limited available resources to meet their C/N/P ratio demands. This is consistent with the optimal foraging theory, which states that the conditions of essential resource deficiency select for individuals with increased capacity of acquiring the deficient resources (Pyke 1984). Indeed, we found significantly larger amounts of extraradical AMF mycelium produced under deficient inorganic nutrient conditions, especially P, in the field as well as the higher AMF root colonization levels and parasitic effects observed in the Minus P treatment. Treseder & Allen (2002) also found that AMF biomass increased as a result of N and P additions to N- and P-deficient soils, respectively, but was reduced under non-deficient soil nutrient conditions.
While it is possible that symbiotic functional specificities of AMF communities from the NPKCa and Minus P treatments might have resulted from shifts in community structure, because the AMF communities were generally similar among treatments (i.e. species poor and dominated by the same morphotype), it is possible that adaptations by isolates forming those communities have occurred. This is not surprising if we consider recent studies showing that AMF species contain high genetic diversity (Stockinger, Walker & Schussler 2009; Borstler et al. 2010), which can be exchanged between genetically different isolates of the same species (Croll et al. 2009). Indeed, manipulations of P concentration in artificial media lead to genetic and phenotypic change in individual Glomus intraradices N.C. Schenck & G.S. Sm. (Ehinger, Koch & Sanders 2009). Moreover, recent reports provide evidence for the existence of ecotypes within AMF species adapted to temperature conditions (Antunes et al. 2011).
Here, we show that the long-term management of abiotic factors can act as an important driver of functional and/or evolutionary processes in AMF communities. Our results suggest that incessant severe nutrient deficiencies might promote AMF with increased capacity to cheat the host plant (i.e. communities that include AMF that access C without providing the deficient resources). However, it is possible that these AMF can confer to their hosts other benefits different than nutrient uptake, for instance protection against pathogens (Wehner et al. 2010). On the other hand, our results obtained from the fully fertilized treatment also contribute to support the model that conventional use of chemical fertilizers and consequent sufficient availability of N and P select for parasitic outcomes of the AMF symbiosis (Johnson 1993). Given the need for a more sustainable agriculture that makes better use of naturally occurring mutualisms (Hart & Trevors 2005), future work should focus on gauging multiple long-term selection pressures to maximize the use or the search for the most effective indigenous AMF.
We thank the Freie Universität Berlin for funding the study. We also thank the Humboldt University for providing access to the long-term nutrient deficiency experiment in Thyrow. We are also grateful to Sabine Artelt and Sabine Buchert for their help with P analyses. The Natural Sciences and Engineering Research Council of Canada and the Ontario Ministry of Natural Resources currently fund PMA. AL was supported by the Dahlem Center of Plant Sciences at FU Berlin.