Differential modulation of host plant δ13C and δ18O by native and nonnative arbuscular mycorrhizal fungi in a semiarid environment

Authors

  • J. I. Querejeta,

    Corresponding author
    1. Departamento de Conservación de Suelos y Aguas, Centro de Edafología y Biología Aplicada del Segura-Consejo Superior de Investigaciones Científicas (CEBAS-CSIC), PO Box 4195, Campus de Espinardo E−30100 Murcia, Spain;
    2. Center for Conservation Biology, The University of California, Riverside, CA 92521, USA
    • Author for correspondence: J. I. Querejeta Tel: +34 96 8396257 Fax: +34 96 8396213 Email: querejeta@cebas.csic.es

    Search for more papers by this author
  • M. F. Allen,

    1. Center for Conservation Biology, The University of California, Riverside, CA 92521, USA
    Search for more papers by this author
  • F. Caravaca,

    1. Departamento de Conservación de Suelos y Aguas, Centro de Edafología y Biología Aplicada del Segura-Consejo Superior de Investigaciones Científicas (CEBAS-CSIC), PO Box 4195, Campus de Espinardo E−30100 Murcia, Spain;
    Search for more papers by this author
  • A. Roldán

    1. Departamento de Conservación de Suelos y Aguas, Centro de Edafología y Biología Aplicada del Segura-Consejo Superior de Investigaciones Científicas (CEBAS-CSIC), PO Box 4195, Campus de Espinardo E−30100 Murcia, Spain;
    Search for more papers by this author

Summary

  • • Native, drought-adapted arbuscular mycorrhizal fungi (AMF) often improve host-plant performance to a greater extent than nonnative AMF in dry environments. However, little is known about the physiological basis for this differential plant response.
  • • Seedlings of Olea europaea and Rhamnus lycioides were inoculated with either a mixture of eight native Glomus species or with the nonnative Glomus claroideum before field transplanting in a semiarid area.
  • • Inoculation with native AMF produced the greatest improvement in nutrient and water status as well as in long-term growth for both Olea and Rhamnus. Foliar δ18O measurements indicated that native AMF enhanced stomatal conductance to a greater extent than nonnative AMF in Olea and Rhamnus.δ13C data showed that intrinsic water-use efficiency in Olea was differentially stimulated by native AMF compared with nonnative AMF.
  • • Our results suggest that modulation of leaf gas exchange by native, drought-adapted AMF is critical to the long-term performance of host plants in semiarid environments. δ18O can provide a time-integrated measure of the effect of mycorrhizal infection on host-plant water relations.

Introduction

Soil degradation limits the potential for re-establishment of native vegetation in semiarid areas affected by desertification processes (Agnew & Warren, 1996). In particular, desertification reduces the diversity and abundance of key mutualistic microbial symbionts such as arbuscular mycorrhizal fungi (AMF) which enhance the ability of plants to establish in the face of drought and low fertility conditions (Allen, 1989). The introduction of native shrub species in combination with a managed community of microbial symbionts can help guarantee the successful establishment of vegetation in degraded semiarid areas (Requena et al., 2001; Caravaca et al., 2002; Querejeta et al., 2003). A critical step in this integral approach to ecosystem restoration is the selection of an appropriate AMF inoculum capable of maximizing plant survival and growth under drought conditions.

Despite their low host specificity (e.g. Klironomos, 2000), AMF are known to vary widely in their ability to take up phosphorus and stimulate the growth of host plants (Ravnskov & Jakobsen, 1995; Streitwolf-Engel et al., 1997; van der Heijden et al., 1998a,b; Helgason et al., 2002; Klironomos, 2003; van der Heijden, 2004). Requena et al. (1996) reported that inoculation with a mixture of indigenous AMF failed to stimulate the growth of the naturally co-occurring shrub Anthyllis cytisoides in AMF screening trials in the glasshouse, whereas inoculation with an exotic AMF (Glomus intraradices) significantly enhanced growth. In a subsequent field study, Requena et al. (2001) found that inoculation with the exotic AMF Glomus intraradices promoted faster Anthyllis growth than inoculation with a mixture of native AMF during the first year after seedling transplanting in a degraded semiarid area. However, 4 yr later the shrubs inoculated with the native AMF were much larger than the shrubs inoculated with the exotic AMF. These studies showed that the ability of nonnative AMF to stimulate plant growth in the glasshouse or during the early stages after outplanting may not be a reliable indicator of good long-term performance under semiarid field conditions. However, the nutritional and/or nonnutritional mechanisms behind the differential ability of native and nonnative AMF to promote host plant growth in the long term were not identified by Requena and coworkers.

Arbuscular mycorrhizal fungi have been shown to provide various nonnutritional benefits to the host plant (Newsham et al., 1995; Allen et al., 2003), although these roles are generally considered to be subsidiary roles of the AM symbiosis. In particular, AMF native to semiarid environments take up water and nutrients more efficiently in drying soil, therefore conferring better drought resistance to the host plant (Tobar et al., 1994; Ruiz-Lozano & Azcón, 1995; Ruiz-Lozano et al., 2001; Marulanda et al., 2003; Porcel & Ruiz-Lozano, 2004; Ruiz-Lozano et al., 1995a,b). Different AMF species have been shown to modulate the physiological response of host plants to drought differently, including stomatal conductance, photosynthetic rates or water use efficiency (Allen et al., 1981; Allen & Boosalis, 1983; Dixon et al., 1994; Mathur & Vyas, 1995; Ruiz-Lozano et al., 1995a,b; Shrestha et al., 1995).

The δ13C values in plant organic matter respond to the interplay among all aspects of plant carbon and water relations and are thereby more useful than single-time gas exchange measurements as integrators of whole plant function throughout the period the plant tissue was synthesized (Dawson et al., 2002). The δ13C signature of plants is linearly linked to the ratio between the partial pressure of CO2 in the leaf intercellular spaces and that of the ambient air (Farquhar et al., 1989). δ13C reflects the relative magnitudes of net photosynthesis (A) and stomatal conductance (gs) and therefore provides an estimate of mean growing season intrinsic water use efficiency (WUE; A/gs) in C3 plants (Farquhar et al., 1989).

The amount of δ18O in plant tissues is determined by the integrated leaf-to-air vapor pressure gradient during photosynthetic gas exchange (Farquhar et al., 1998). This gradient is influenced by the plant physiological response to changes in environmental conditions such as atmospheric humidity or soil moisture. The simultaneous measurement of δ13C and δ18O in plant organic matter can help separate the independent effects of carbon fixation and stomatal conductance on δ13C, as δ18O shares dependence on stomatal conductance with δ13C but is not dependent on Rubisco activity (Scheidegger et al., 2000; Keitel et al., 2003). While δ13C measurements have been successfully used to assess the physiological response of host plants to AM colonization (Di & Allen, 1991; Handley et al., 1999; Querejeta et al., 2003), mycorrhizal effects on plant δ18O signature yet remain to be addressed.

Most published studies reporting differential effects of various AMF on host plant physiological response to drought were carried out in the glasshouse using agricultural plants species. We designed a field experiment in which seedlings of two native wild shrub species (Olea europaea and Rhamnus lycioides) belonging to different plant families were inoculated with either native or nonnative Glomus species before outplanting in an abandoned agricultural land. Our goal was to evaluate the nutritional and/or nonnutritional basis for any differential growth advantage conferred to the host plants by the drought-adapted native AMF under semiarid conditions. For that purpose, we measured the nutrient status, water content, growth, root AM colonization and shoot δ13C, δ15N and δ18O of mycorrhizal Olea and Rhamnus seedlings 1 yr after transplanting in the field.

Materials and Methods

Study site

The experimental area was located in the foothills of the El Picarcho range in the province of Murcia, south-east Spain (1°10′ W, 38°23′ N, 400 m above sea level). The climate is semiarid Mediterranean, with an average annual rainfall of 290 mm, average annual potential evapotranspiration of 827 mm and a mean annual temperature of 15.9°C. The topography of the area is mostly flat, with slopes not exceeding 6%. The plant cover in the experimental area is sparse (< 20% canopy cover), degraded by secular logging, grazing and farming, and dominated by dwarf shrubs (< 1 m high) such as rosemary (Rosmarinus officinalis) and alpha grass (Stipa tenacissima). The soil is a Petrocalcic Xerosol (FAO, 1998) developed from limestone, with a silt loam texture. Some characteristics of the soil are shown in Table 1.

Table 1. Some characteristics of the soil in the experimental area
Parameter
  1. MPN, most probable number; AM, arbuscular mycorrhiza.

  2. Each value is the mean of five soil samples (± SE).

pH (H2O)  7.6 ± 0.0
Electrical conductivity (1 : 5) (µS cm−1) 144 ± 3
Total organic carbon (g kg−1) 20.8 ± 0.9
Water soluble carbon (µg g−1) 134 ± 6
Total carbohydrates (µg g−1)1956 ± 82
Water-soluble carbohydrates (µg g−1)   1 ± 0
Total nitrogen (g kg−1)  0.7 ± 0.1
Available P (µg g−1)  20 ± 4
Extractable K (µg g−1) 177 ± 47
Aggregate stability (%) 19.5 ± 3.0
AM infective propagules (MPN g−1 dry soil) 0.24 ± 0.01

Plant material and mycorrhizal treatments

The shrub species used in the experiment are key components of the climax plant communities of semiarid south-east Spain. The shrub species were O. europaea L. ssp. sylvestris (Mill.) Lehr. (Oleaceae) and R. lycioides L. (Rhamnaceae). Both are well adapted to water stress conditions and are frequently used in revegetation of disturbed semiarid lands. The actinorhizal species R. lycioides forms symbiotic associations with nitrogen fixing microorganisms (Frankia spp.), while O. europaea is a nonfixer species.

The mycorrhizal inoculum used was either a nonnative isolate of Glomus claroideum Schenk & Smith (EEZ 24), or a mixture of native AMF isolated from a nearby semiarid area where the target plants grow naturally. The mixture of native endophytes included Glomus geosporum (Nico. and Gerd.) Walker (EEZ 31), Glomus albidum Walker and Rhodes (EEZ 39), Glomus microaggregatum Koske, Genma & Olexia (EEZ 40), Glomus constrictum Trappe (EEZ 42), Glomus mosseae (Nicol. & Gerd.) Gerd. & Trappe (EEZ 43), Glomus coronatum Giovannetti (EEZ 44), G. intraradices Schenk & Smith (EEZ 45), and an unidentified Glomus sp. (EEZ 46). The acronym EEZ refers to the AMF collection of the Estación Experimental Zaidín-CSIC in Granada, Spain.

The AMF inoculum consisted of a mixture of rhizospheric soil from trap cultures (using Sorghum sp.), containing spores, hyphae and mycorrhizal root fragments. Once germinated, seedlings were transplanted into the growing substrate, consisting of peat and cocopeat (1 : 1, v : v, Paisajes del Sur, Granada). The corresponding AMF inoculum was applied at a rate of 5% (v : v). The same amount of autoclaved inoculum was added to control plants, supplemented with a filtrate (< 20 µm) of the inoculum to provide the microbial populations accompanying the mycorrhizal fungi. Inoculated and noninoculated seedlings were grown for 8 months under nursery conditions without any fertilizer addition.

Experimental design and layout

The experiment was a two-factor factorial design with five replicate blocks. The first factor was shrub species (two levels). The second factor was mycorrhizal treatment, including three levels: no inoculation (control), inoculation with G. claroideum and inoculation with the mixture of eight native Glomus species. In January 2000, an area of 1200 m2 was mechanically prepared with a subsoiler. Three rows per plant species (1 m wide, 3 m apart) were established in each block, and mycorrhizal treatments were randomly assigned to them. In early November 2000, seedlings were planted in individual holes 1 m apart from each other. Fifteen seedlings per mycorrhizal treatment per replicate block were planted for each shrub species.

The field experiment was carried out under strictly natural conditions, without any watering or fertilizer treatments. Total rainfall during the 18 month duration of the field experiment was 466 mm.

Sampling and laboratory procedures

Five plants of each shrub species × mycorrhizal treatment combination were harvested just before field transplanting, and again 12 months and 18 months after outplanting (one seedling per replication block). During seedling harvesting in the field, we tried to minimize damage to roots by carefully excavating a soil rhizosphere volume of 40 × 40 × 40 cm. Shoot fresh weights were measured in the laboratory within 2 h of seedling harvesting in order to estimate shoot water contents. Dry (65°C, 24 h) weights of shoots were measured afterwards. Oven-dried plant tissues were finely ground before chemical analysis. Only new shoots and leaves produced after field transplanting were used for nutrient and isotopic analyses. Foliar concentrations of phosphorus and potassium were determined after digestion in nitric–perchloric acid for 6 h at 210°C. Foliar phosphorus was determined colorimetrically, nitrogen was determined by the Kjeldahl method and potassium was estimated by flame photometry as described in Caravaca et al. (2002).

Three subsamples from the upper, middle and lower root system of each seedling were taken, and the percentage of root length colonized by AMF was calculated by the gridline intersect method (Giovannetti & Mosse, 1980) after staining with Trypan blue (Philips & Hayman, 1970).

The δ13C and δ15N analyses were conducted at the University of California Davis Stable Isotope Facility, using a continuous flow, isotope ratio mass spectrometer (CF-IRMS; Europa Scientific, Crewe, UK) in the dual-isotope mode, interfaced with a CN sample converter. Concentrations of 13C and 15N are reported using differential notation, showing differences between the observed concentration and that of a common standard. The standard for δ13C was Pee Dee Belemnite, and for δ15N, atmospheric nitrogen (N2).

Foliar in δ18O tissues was determined at the Earth and Planetary Sciences Department, University of New Mexico. Finely ground leaf material was placed in silver capsules and dropped into a high-temperature (1450°C) reduction furnace (MAT TC-EA; Finnigan, Bremen, Germany) in a helium flow. The pyrolysis unit was interfaced with a Finnigan Mat Delta Plus XL mass spectrometer. The δ18O values are given relative to VSMOW (Vienna standard mean oceanic water).

Statistical analyses

Plant species and mycorrhizal inoculation effects on measured variables were tested by a two-way analysis of variance. Comparisons among means within plant species were made using the least significant difference (LSD) multiple range test. Statistical procedures were carried out with the software package SPSS 13.0 (SPSS Inc., Chicago, IL, USA).

Results

Glasshouse conditions

Before field transplanting, M (mixed native Glomus)- and G (G. claroideum)-inoculated seedlings showed very similar levels of AM root colonization in Olea, whereas M-inoculated Rhamnus plants had higher percentage colonization than G-inoculated ones (Table 2).

Table 2. Percentage arbuscular mycorrhizal (AM) fungal colonization of roots, shoot dry biomass, shoot nutrient concentrations and shoot water content of 8-month-old nursery grown Olea and Rhamnus seedlings
  Olea europaea ssp. sylvestris Rhamnus lycioides
CMGCMG
  1. C, Uninoculated plants; M, plants inoculated with a mixture of eight native Glomus species; G, plants inoculated with Glomus claroideum. Within plant species, values sharing the same letter are not significantly different (P < 0.05) according to the LSD test.

AM root colonization (%)0a  74 ± 1b  69 ± 2b0a  66 ± 2c  43 ± 1b
Shoot dry biomass (g)0.29 ± 0.02a0.74 ± 0.11b0.93 ± 0.08b0.14 ± 0.04a0.35 ± 0.03b0.13 ± 0.04a
Nitrogen (mg g−1) 7.7 ± 0.3a10.4 ± 0.4b14.6 ± 0.8c 2.9 ± 0.3a14.6 ± 0.2b12.4 ± 0.7b
Phosphorus (mg g−1)0.23 ± 0.04a0.75 ± 0.02b0.73 ± 0.03b0.38 ± 0.03a0.75 ± 0.04b0.63 ± 0.05b
Potassium (mg g−1) 5.9 ± 0.3a 8.2 ± 0.1b 7.5 ± 0.3ab 5.4 ± 0.5a10.3 ± 0.3b13.8 ± 0.8c
Shoot water content (%)54.1 ± 1.3a  56 ± 3.2a55.3 ± 1.3a58.2 ± 1.2a56.6 ± 1.3a58.5 ± 1.8a

Both M- and G-inoculation enhanced seedling growth in Olea compared with the uninoculated controls (Table 2). The M- and G-inoculated plants had similar size in Olea. Inoculation with G. claroideum failed to stimulate growth compared with the uninoculated controls in Rhamnus. Only the native Glomus inoculum enhanced Rhamnus growth in the glasshouse.

Both AMF inoculation treatments improved the nutrient status of Olea and Rhamnus seedlings under glasshouse conditions (Table 2). The G-inoculated seedlings showed similar phosphorus concentrations but higher nitrogen concentrations than M-inoculated ones in Olea. Foliar nitrogen and phosphorus concentrations were similar in M- and G-inoculated Rhamnus. The G-inoculated plants had higher potassium concentrations than M-inoculated ones in Rhamnus.

Mycorrhizal inoculation treatments did not affect the shoot water content of Olea or Rhamnus seedlings under glasshouse conditions (Table 2).

Twelve months after field transplanting

Control seedlings showed low AM root colonization levels (< 10%) in both Olea and Rhamnus (Table 3). The M- and G-inoculated seedlings had much higher AM colonization than their respective controls in both shrub species, although M-inoculated seedlings showed higher colonization than G-inoculated ones.

Table 3. Percentage arbuscular mycorrhizal (AM) fungal colonization of roots, shoot dry biomass, shoot nutrient concentrations, shoot water content and carbon, nitrogen and oxygen stable isotope compositions of Olea and Rhamnus seedlings 1 yr after field transplanting
  Olea europaea ssp. sylvestris Rhamnus lycioides
CMGCMG   
  1. C, Uninoculated plants; M, plants inoculated with a mixture of eight native Glomus species; G, plants inoculated with Glomus claroideum. Within plant species, values sharing the same letter are not significantly different (P < 0.05) according to the LSD test. *C and G are different at P < 0.073. Significance of effects of shrub species and mycorrhizal treatments on the measured variables are also shown. NS, not significant; AMF, mycorrhizal inoculation treatment.

AM root colonization (%)    9 ± 2a   72 ± 10c   48 ± 7b    6 ± 5a   68 ± 7c   31 ± 6b   
Shoot dry biomass (g) 0.64 ± 0.05a 4.92 ± 0.56b 4.42 ± 0.89b 0.56 ± 0.05a 2.22 ± 0.16c 0.81 ± 0.12b   
Nitrogen (mg g−1)  6.8 ± 1.2a   14 ± 0.9b 14.5 ± 0.6b  8.5 ± 0.8a 15.2 ± 1.1b  9.1 ± 1.1a   
Phosphorus (mg g−1) 0.60 ± 0.05a 0.82 ± 0.03b 0.67 ± 0.06a 0.45 ± 0.08a 0.62 ± 0.05b 0.54 ± 0.02ab   
Potassium (mg g−1)  5.4 ± 0.4a  6.8 ± 0.3a  6.2 ± 0.5a  3.8 ± 0.1b  2.9 ± 0.1a  2.9 ± 0.1a   
Shoot water content (%) 41.3 ± 0.2a   49 ± 1.7c 45.6 ± 1.1b 35.6 ± 0.9a 41.9 ± 0.7b 36.6 ± 0.8a   
δ13C ()−29.3 ± 0.1a−28.5 ± 0.4b−29.2 ± 0.3a−29.1 ± 0.2a−29.6 ± 0.1a−29.3 ± 0.2a   
δ15N ()  2.1 ± 0.2a  3.3 ± 0.4b  2.8 ± 0.3ab  −0.4 ± 0.3a  0.2 ± 0.2a  0.1 ± 0.4a   
δ18O ()   28 ± 0.5b 26.8 ± 0.5a 28.3 ± 0.7b 28.2 ± 0.5b 26.7 ± 0.4a 29.4 ± 0.3b*   
 AM colonizationShoot biomassNitrogenPhosphorusPotassiumWater contentδ13Cδ15Nδ18O
Shrub speciesNS0.0000.0000.0000.0000.000NS0.000NS
AMF0.0000.0000.0000.001NS0.000NS0.0030.009
Shrub × AMFNS0.0010.000NSNSNS0.024NSNS

The M- and G-inoculated seedlings were significantly larger than control ones for both Olea and Rhamnus (Table 3). Shoot dry biomass of M- and G-inoculated plants was similar in Olea, whereas M-inoculated plants were larger than G-inoculated ones in Rhamnus.

The M-inoculated seedlings had higher nitrogen foliar concentrations than their controls in both Olea and Rhamnus, while G-inoculated plants showed higher nitrogen concentrations than their controls in Olea only. The M- and G-inoculated seedlings showed very similar nitrogen foliar concentrations in Olea. Inoculation with the mixture of native Glomus species improved nitrogen status in Rhamnus over inoculation with Glomus claroideum.

The M-inoculated plants (but not G-inoculated plants) showed significantly higher phosphorus foliar concentrations than their controls in Olea and Rhamnus (Table 3). Phosphorus concentrations in M- and G-inoculated plants were not significantly different for Rhamnus, while for Olea the M-inoculated plants had higher values.

Potassium foliar concentration was not affected by the inoculation treatments in Olea (Table 3). M- and G-inoculated seedlings showed lower potassium concentrations than their controls in Rhamnus. M- and G-inoculated seedlings showed similar potassium foliar concentrations for both shrub species.

The M-inoculated seedlings had higher shoot water content than their controls for both Olea and Rhamnus (Table 3). The G-inoculated plants had higher water content than their controls in Olea only. Inoculation with the mix of native Glomus species improved shoot water content to a greater extent than inoculation with Glomus claroideum in both Olea and Rhamnus.

The M-inoculated Olea plants used water more efficiently than their uninoculated controls during the first year under field conditions, as shown by δ13C data (Table 3). Intrinsic WUE was also higher in M- than in G-inoculated plants, as indicated by higher δ13C values. By contrast, mycorrizal treatments did not affect shoot δ13C significantly in Rhamnus.

Seedlings inoculated with the mix of native Glomus species showed lower shoot δ18O signatures than control or G-inoculated seedlings for both Olea and Rhamnus (Table 3). M-inoculated shrubs showed higher δ15N values than their respective controls, although differences were significant for Olea only (Table 3).

Eighteen months after field transplanting

The survival rates of M- and G-inoculated Olea seedlings were higher than that of their uninoculated controls (Table 4). Survivorship was not affected by mycorrhizal inoculation treatments in Rhamnus. Survival rates of M- and G-inoculated plants were nearly identical for both Olea and Rhamnus.

Table 4. Percentage arbuscular mycorrhizal (AM) fungal colonization of roots, shoot dry biomass, relative growth rate and survival rate of Olea and Rhamnus seedlings 18 months after field transplanting
  Olea europaea ssp. sylvestris Rhamnus lycioides
CMGCMG
  1. C, Uninoculated plants; M, plants inoculated with a mixture of eight native Glomus species; G, plants inoculated with Glomus claroideum. Within plant species, values sharing the same letter are not significantly different (P < 0.05) according to the LSD test.

AM root colonization (%) 10 ± 1a  85 ± 2b 88 ± 3b  8 ± 1a 67 ± 2b 74 ± 6b
Shoot dry biomass (g)1.4 ± 0.1a16.5 ± 2.7c6.7 ± 0.2b0.5 ± 0.1a8.1 ± 2.1c1.8 ± 0.2b
Relative growth rate (g g−1)3.921.26.22.922.212.5
Survival rate (%)45a95b92b55a62a64a

Uninoculated control plants still showed negligible levels (8–10%) of AM root colonization in Olea and Rhamnus 18 months after field transplanting (Table 4). By contrast, M- and G-inoculated plants showed similarly high (67–88%) levels of AM root colonization in both Olea and Rhamnus.

Shoot biomass of M-inoculated shrubs was between 12 (Olea) and 16 (Rhamnus) times larger than that of their respective uninoculated controls (Table 4). The G-inoculated plants were also significantly larger than control plants in Olea and Rhamnus. The M-inoculated plants had much greater shoot biomass than their G-inoculated counterparts in both Olea and Rhamnus.

Over the entire 18-month period in the field, plants inoculated with the mixture of native Glomus fungi showed by far the highest relative growth rates in both shrub species (Table 4).

Discussion

Large differences in AM percentage colonization between control and nursery-inoculated seedlings persisted throughout the 18-month field experiment. The local AMF community showed little capacity to colonize shrub roots, likely because of low fungal species diversity (Azcón-Aguilar et al., 2003) and low abundance of mycorrhizal propagules in the soil (Table 1).

The local AMF community was much less effective than the added (native or nonnative) Glomus inoculum at stimulating host plant growth for both Olea and Rhamnus. Inoculation with the mixture of native Glomus species conferred a clear growth advantage to Rhamnus over inoculation with G. claroideum. This growth advantage increased after field transplanting, indicating strong specificity of response to different Glomus species in Rhamnus. Inoculation with a mixture of eight AMF species increases the probability of plant–fungus matches that stimulate optimal plant growth compared with inoculation with a single AMF species (van der Heijden et al., 1999). Moreover, coevolution and co-occurrence of the symbiotic partners, better adaptation of native AMF to local semiarid conditions and potential functional complementarity among AMF species would all be expected to provide a growth advantage to the seedlings inoculated with the mixture of native Glomus fungi (Klironomos, 2003).

Remarkably, Olea shrubs inoculated with a single nonnative Glomus species had comparable size to those inoculated with a mixture of eight native Glomus species, both in the glasshouse and 12 months after field transplanting. Only in the longer term (18 months) did the native Glomus inoculum provide a clear growth advantage over Glomus claroideum for Olea. Overall, these results support the idea that plant growth response to different AMF is highly dependent on environmental conditions (Johnson et al., 1997). However, as illustrated by Olea, diverging plant growth responses to different AMF may only become evident in the long term under the wide range of extreme environmental conditions that characterize semiarid ecosystems.

The putative mechanisms underlying the growth advantage provided in the longer term by the mixture of native Glomus sp. varied between shrub species. Inoculation with native AMF enhanced shoot concentrations of either nitrogen (Rhamnus) or phosphorus (Olea) to a greater extent than inoculation with nonnative AMF during the first year after field outplanting. Functional complementarity among the native Glomus species added as inoculum may have contributed to increased nutrient uptake in M-inoculated plants compared with G-inoculated ones (Koide, 2000). Differing spatial abilities to acquire nutrients has been cited as one possible reason why colonization by multiple AMF can be more beneficial than colonization by a single species (Smith et al., 2000). However, it is interesting to note that, under glasshouse conditions, Olea and Rhamnus seedlings inoculated with G. claroideum had similar or even higher nutrient concentrations than those inoculated with the mixture of eight native Glomus. Greater enhancement of nutrient uptake by the native compared with the nonnative AMF occurred only after field transplanting, indicating that the former were more efficient at absorbing nutrients in a xeric natural environment, but not under more mesic glasshouse conditions. Drought-adapted native Glomus species have been shown to be particularly effective at scavenging for nutrients in dry soil (Tobar et al., 1994; Ruiz-Lozano et al., 1995a).

Lower shoot δ18O values in Olea and Rhamnus seedlings inoculated with the native Glomus species than in their respective uninoculated controls could just reflect a plant size effect. Evaporative isotopic enrichment of soil water near the surface is very pronounced in semiarid environments (Barnes & Allison, 1983; Allison & Hughes, 1983), creating sharp gradients in soil water δ18O with depth. Larger M-inoculated plants likely had access to deeper, less isotopically enriched soil water than control plants. Since the oxygen isotope ratio of plant cellulose reflects the signature of source water (with an enrichment of 27; Sternberg et al., 1986), uptake of water with dissimilar δ18O signatures would lead to differences in shoot δ18O among mycorrhizal treatments (Barbour et al., 2002). However, this cannot explain the large difference in shoot δ18O found between M- and G-inoculated Olea seedlings, as they had similar size and presumably were extracting water from similar depths. Moreover, control and G-inoculated Olea seedlings showed similar shoot δ18O values despite large size differences between them, suggesting that depth of water uptake was not the major determinant of δ18O variability among mycorrhizal treatments.

The source water signal can be modified by large variability in evaporative enrichment in drought-adapted Mediterranean species with tight stomatal regulation of transpiration (Ferrio & Voltas, 2005). The ratio of water vapor pressure of the air outside and inside the leaf is the parameter controlling evaporative enrichment of O18 in leaf water and therefore in plant organic matter (Farquhar et al., 1998; Keitel et al., 2003). Since greater stomatal conductance cools the leaf and reduces internal water pressure, the δ18O signature of plant tissues is a useful tool to characterize gs, independent from effects of carbon fixation (Keitel et al., 2003). Lower δ18O values indicated enhanced gs in seedlings inoculated with the native Glomus species compared with those inoculated with G. claroideum, as the δ18O signature of plant organic matter decreases in response to increased gs (Scheidegger et al., 2000; Barbour et al., 2002). More efficient hyphal water uptake and transport in dry soil by drought-adapted, native AMF may have contributed to greater gs in their host plants (Augé, 2001; Marulanda et al., 2003). Higher shoot water content in plants inoculated with the native Glomus inoculum is consistent with and supports the aforementioned interpretation of δ18O differences among mycorrhizal treatments. Shoot water content was negatively correlated with δ18O in both Olea (Pearson correlation coefficient = −0.589; P < 0.05) and Rhamnus (Pearson correlation coefficient = −0.655; P < 0.01). To our knowledge, this is the first study showing evidence of a mycorrhizal effect on the δ18O signature of host plants. Our results indicate that δ18O can provide an integrated measure of the effect of mycorrhizal infection on host plant water relations over the period when the plant tissue was formed.

Mycorrhizal treatments had contrasting effects on shoot δ13C depending on the specific plant–fungus combination. δ13C data showed differential ability of native and nonnative Glomus fungi to modulate intrinsic water use efficiency in Olea. Greater WUE in M-inoculated Olea seedlings compared with G-inoculated seedlings was likely the result of better phosphorus status in the former, as improved nutrition can lead to specific stimulation of photosynthetic capacity over stomatal conductance (Koide, 1993; Querejeta et al., 2003). Higher δ13C but lower δ18O in M-inoculated Olea seedlings compared with G-inoculated seedlings suggests a very strong enhancement of photosynthetic rate capable of increasing plant WUE despite increased stomatal conductance (Scheidegger et al., 2000). Nonnutritional mechanisms may have also played a role in the differential modulation of host plant photosynthetic capacity by different AMF species in Olea (e.g. differences in carbon sink strength; Wright et al., 1998).

Foliar δ13C was not significantly affected by the mycorrhizal treatments in Rhamnus, although inoculation with the native AMF tended to decrease δ13C compared with the uninoculated controls (P = 0.08). Lower δ18O values in seedlings inoculated with the native AMF further suggest enhancement of stomatal conductance over photosynthetic capacity in Rhamnus (Scheidegger et al., 2000).

Differences in δ13C between Olea and Rhamnus were largest when comparing seedlings inoculated with the native AMF, which supports a previous study indicating that much of the interspecific variability in foliar δ13C found in semiarid plant communities results from dissimilar patterns of physiological response to AMF infection (Querejeta et al., 2003).

Greater enhancement of shoot nitrogen concentration by native than by nonnative AMF was found in Rhamnus but not in Olea, suggesting specific stimulation of atmospheric nitrogen fixation rather than of soil nitrogen uptake in the former. Rhamnus showed δ15N values around 0 which reflected substantial atmospheric nitrogen biological fixation, as soil nitrogen is typically enriched in 15N compared with atmospheric nitrogen (δ15N = 0; Shearer et al., 1983). Ruiz-Lozano et al. (2001) showed that the AM symbiosis enhanced the nitrogen status of legume plants by protecting them against drought-induced nodule senescence, and partially attributed this protective effect to better hydration of AM plants. Arbuscular mycorrhizal symbiosis may exert a similar protective effect against drought in actinorhizal plants. Improved water status in Rhamnus seedlings inoculated with native AMF may have favoured the activity of their associated nitrogen-fixing Frankia symbionts, thus leading to much higher shoot nitrogen concentration than in those inoculated with the nonnative AMF.

The interpretation of natural abundance δ15N in plants in the field is complex as it reflects the net effect of a wide range of processes, including soil nitrogen sources, mycorrhizal infection, internal fractionations and rooting depth (Evans, 2001). The 15N isotopic enrichment in M-inoculated Olea plants could reflect greater uptake of soil nitrogen in the field, as the δ15N values of mycorrhizal shrubs were closer to that of the surface soil layer (δ15N = 4.6 ± 0.3 at 0–20 cm depth, n = 5). This 15N enrichment might also reflect the larger size and presumably greater rooting depth of M-inoculated seedlings, since there is usually a steep gradient of increasing δ15N down the soil profile (Handley et al., 2001).

In conclusion, inoculation with a mixture of native AMF greatly improved the nutrient and water status as well as the long-term growth of Olea and Rhamnus seedlings in a semiarid environment. Oxygen and carbon isotopic measurements showed that native AMF enhanced stomatal conductance in both Olea and Rhamnus, and stimulated photosynthetic capacity and WUE in Olea. These results suggest that modulation of leaf gas exchange parameters by drought-adapted, native AMF is critical to the long-term performance of host plants in semiarid environments.

Acknowledgements

We thank María del Mar Alguacil for help during field and laboratory work and Viorel Atudorei for help with δ18O analyses. This research was supported by the EC + CICYT cofinanced FEDER programme (1FD97-0507 FOREST) and by the Biocomplexity Program (DEB 9981548) of the US National Science Foundation. J. I. Q. acknowledges a Fulbright postdoctoral fellowship from the Spanish Ministry of Education and Science.

Ancillary