• The contribution of arbuscular mycorrhizas (AM) to plant community structure and diversity is reported here in an annual herbland in southern Australia.
• Mycorrhizal colonization was reduced in field plots by applying the fungicide benomyl as a soil drench. The mycorrhiza-responsiveness of plant species was assessed in intact soil cores containing the indigenous AM fungi and in a pot experiment using an isolate of Glomus mosseae.
• Glasshouse experiments showed that Medicago minima, Vittadinia gracilis and Velleia arguta were highly mycorrhiza-responsive, Salvia verbenaca became colonized but exhibited no growth response to AM, and Carrichtera annua remained uncolonized. There was no change in plant species richness in mycorrhiza-suppressed field plots but diversity increased owing to an increase in evenness. Treatment had no effect on community productivity and therefore there was no relationship between mycorrhizal effects on diversity and productivity.
• Mycorrhizal responsiveness was not a good predictor of species response to suppression of AM in the field. The mycorrhiza-responsive species V. gracilis and V. arguta were not affected by reduced mycorrhizal colonization in fungicide-treated plots, suggesting that competition from the mycorrhiza-responsive dominant M. minima offset the benefits of mycorrhizal association for these species.
Associations between plants and arbuscular mycorrhizal (AM) fungi are common in natural and agricultural ecosystems (Smith & Read, 1997). These associations contribute to plant nutrition, particularly plant uptake of poorly mobile soil nutrients such as phosphorus (Smith & Read, 1997) and to improved water relations where water is growth limiting (Sanchez-Diaz & Honrubia, 1994). However, not all plant species form AM associations, and not all AM plant species show clear nutritional benefits from colonization by mycorrhizal fungi under all growth conditions (Tester et al., 1985; Fitter, 1986; Francis & Read, 1995). Plant communities may often include species with varying response to the presence of AM fungi. This variation may range from nonmycorrhizal plants with benign or possibly antagonistic interactions with the fungi (Francis & Read, 1995), to highly responsive plant species with positive growth responses to AM association (Johnson et al., 1997). The composition of natural plant communities often includes plant species from across this response spectrum, and may be determined by differential contributions of the mycorrhizal fungi to individual plant species fitness.
In plant communities, the influence of mycorrhizas on individual plant growth is further affected by interactions between the individual plants. An established mycorrhizal mycelium is a potential resource and co-occurring plant species may differ in their ability to compete for this resource despite showing individual responsiveness to mycorrhizal association when grown separately in pots (Newman et al., 1992). The extent of host–plant benefit from mycorrhizas is also density dependent (Koide & Li, 1991; Facelli et al., 1999) and may be influenced by neighbor competition (Hartnett et al., 1993). Accordingly, changes in the abundance or size of individuals from one species may influence the abundance and size of individuals of the same or different species. The role of mycorrhizas in regulating community structure will therefore depend on the identity and functional characteristics of the species present, but can only be satisfactorily examined in intact communities.
Several studies have found that ecosystem processes such as primary productivity are linked to species diversity (Naeem et al., 1994; Tilman et al., 1996, 1997; Hector, 1999). While increasing species richness has been shown to increase primary productivity, species identity and functional group richness are also important factors controlling productivity (Hooper & Vitousek, 1997; Tilman et al., 1997; Symstad et al., 1998; Wilsey & Potvin, 2000). Results from experiments on the relationship between plant species richness and productivity have, however, been confounded by the ‘hidden treatment’ effects discussed by Huston (1997) and Wardle (1999). Some resolution to the question of whether increased productivity is a function of species identity or diversity has been offered by Wilsey & Potvin (2000). These authors suggest that diversity, measured as evenness (i.e. the distribution of abundance or biomass among species in a community), can have a direct effect on plant productivity. However, these results come from experiments where evenness was manipulated in nonnatural plots with very low species richness.
Grime et al. (1987) found that community structure could be significantly altered by mycorrhizal activity in species-rich mixtures of plants in experimental microcosms. The presence of mycorrhizas can increase floristic diversity (Grime et al., 1987) and species richness (Gange et al., 1990); however, this may depend on the identity and mycorrhizal responsiveness of the dominant plant species (Bergelson & Crawley, 1988). Hartnett & Wilson (1999) have recently shown that suppression of mycorrhizal fungi resulted in an increase in floristic diversity in a tallgrass prairie, probably because the dominant C4 grasses in that system are more strongly responsive to mycorrhizal colonization than the other species present.
Field-based studies of the involvement of mycorrhizas in structuring plant communities have been conducted in relatively few plant ecosystems. Extensive tracts of open woodland in semiarid Australia have been overgrazed, with the understory reduced to annual herbland often dominated by introduced species. The herb and grass species comprising these disturbed communities are thought to differ in their response to mycorrhizal fungi. The influence of mycorrhiza in controlling plant diversity may be vitally important in semiarid ecosystems, where communities can be subjected to dramatic seasonal and interannual fluctuations, and may rely on high biodiversity to maintain stability (Grime, 1997).
This study was designed to test experimentally the relationship between mycorrhizal responsiveness of plants from a semiarid herbland and the influence of mycorrhizas on the productivity and structure of this plant community. To test this relationship, the mycorrhizal responsiveness of plant species from a semiarid herbland was assessed in a pot experiment and changes in plant community structure in the field were measured after formation of mycorrhizas had been suppressed by fungicide application.
Materials and Methods
This experiment was conducted at the Brookfield Conservation Park, a 5527 ha reserve situated at the southern-most extension of the arid zone in South Australia (Lat. 34°19′ 18″; Long. 139°30′ 52″). The site has been protected since 1971 when it was purchased as a reserve for the Southern Hairy-nosed Wombat (Lasiorhinus latifrons). Long-term temperature data are available from a nearby meteorological station at Waikerie, South Australia. Temperatures range from an average maximum of 32.8°C and minimum of 15.2°C in January, to lowest monthly averages of 16.2°C maximum and 5.3°C minimum in July. The highest recorded temperature for the region is 46.5°C. Rainfall is low and irregular, averaging 248 mm but varying from 150 to 550 mm per annum. Rainfall in the year of this study (1998) was 292 mm. This site has a drought frequency of 78%, calculated using the measure of Trumble (1948) and is characterized by extreme variation in the understory condition between years.
The study site is situated in the understory of low open woodland and tall shrubland community dominated by sugarwood (Myoporum platycarpum R. Br.), sheep bush (Geijera linearifolia (DC.) J. Black) and bullock bush (Alectryon oleaefolium). The understory vegetation is characterized by annual herbs dominated by the introduced (‘weed’) species Medicago minima (L) Bartal. (Leguminosae; small erect herb with high palatability), Carrichtera annua (L) DC. (Cruciferae; small erect herb with low palatability) and Salvia verbenaca L. (Labiatae; small herb with basal rosette and low palatability), and the native species Erodium crinitum Carolin (Geraniaceae; small erect herb with low palatability), Velleia arguta R. Br. (Goodeniaceae; small erect herb, palatable) and Vittadinia gracilis (Hook. f.) N. Burb. (Compositae; small erect herb with low palatability). The vegetation is heavily grazed by wombats and by kangaroos (Macropus rufus and Macropus fuliginosus) and regeneration of previously dense spear-grass (Stipa nitida Summerh. & C. E. Hubb.) is limited.
The soil is a clay loam (27% clay, 15% silt and 58% sand) to a depth of 20–25 cm, overlying thick calcrete over Miocene limestone. The soil is generally low in nutrients with a pH of 7.85, 0.85% organic carbon, 9.9 µg g−1 available P (Colwell, 1963), 8.6 µg g−1 NO3-N and a cation exchange capacity (CEC) of 2.2 mEq 100 g−1. The dominant AM fungus species at this site is Glomus mosseae (Nicol. & Gerd.) Gerdemann & Trappe (O’Connor, 2001).
Field experiment design
Early in 1998, 21 1.5 m2 plots were established in seven blocks in open herbland (> 10 m to the nearest tree or shrub) by covering each with a wire cage (5 cm grid mesh, 1.5 × 1.0 × 0.5 m) to prevent grazing by large herbivores. Individuals of the short-lived perennial Salvia verbenaca were removed from all plots by cutting at the base of the stem and swabbing with glyphosate herbicide (100 g l−1). This was done to leave no live plants at the beginning of the growing season and germinating S. verbenaca seedlings could then be treated as annuals in this experiment. Each plot was separated from the others by a 1.5-m spacing and was divided into a central zone of 1 m2 and a surrounding buffer zone to reduce edge effects from surrounding untreated vegetation. Plots in each block were randomly assigned to one of the three treatments (mycorrhiza-suppressed, watered and unwatered control). Mycorrhiza-suppressed plots received the fungicide benomyl as a soil drench (9 g active ingredient (a.i.) in 15 l of water per plot). Watered plots received 15 l of water per plot and control plots received no amendments. Treatments began after the first rains in April and were repeated every 2 wks until the end of August.
Benomyl was chosen for the mycorrhiza-suppressed treatment as it has been shown to suppress colonization of plant roots by glomalean fungi in field plots (Merryweather & Fitter, 1996). The use of benomyl is thought to be a conservative approach to studying AM effects on plant communities (Wilson & Hartnett, 1997). The high rate of fungicide used in this experiment was justified as previous experiments had shown that at lower concentrations benomyl is sorbed to particles in the surface layer of this soil (O’Connor, 2001). Adequate suppression of the AM fungi requires percolation of the fungicide to greater than 5 cm below the soil surface. It was necessary to include a control treatment receiving only water, as well as an unwatered control, to determine effects of additional water on the vegetation at this low rainfall site.
To test for any effects of benomyl application on nontarget soil fungi, three soil samples (40 g) were collected on 8 August from the surface 10 cm of the buffer zone of each of the plots. Subsamples from each plot within a treatment were combined to produce a composite sample for assessing the population of fungi. Extraction of fungal propagules and dilution plating methods followed a modified method of Alef (1995). Moist soil (5 g) was mixed in 45 ml of 0.1% Na2H2P2O7 for 1 h on an end-over-end shaker. One milliliter of this solution was diluted 1 : 10 in 0.1% Na2H2P2O7. Serial dilutions were made using this method to produce dilutions 10−2–10−5 for solution plating. Plates were made using half-strength potato dextrose agar (PDA) + 7.5 g agar l−1. Antibiotics were added before pouring at rates of streptomycin 25 µg g−1 and ampicillin 37.5 µg g−1. Three replicate plates of each soil dilution from plots in each treatment were made by applying 100 µl of soil dilutions 10−2–10−5 to the sterile plates. Plates were incubated at 25°C and observed after 1 wk and fungal colonies counted. Roots collected from field plots and intact cores for assessment of mycorrhizal colonization were also examined for evidence of pathogenic fungi.
Assessment of the effectiveness of benomyl for suppression of AM fungi was made by collecting roots from field plots on 6 June and 8 August, 1998. On 8 August, five plants from each of the most abundant species –M. minima, C. annua, S. verbenaca and V. arguta– were collected from the buffer zone of each plot by excavating the whole root system (top 15 cm of soil). On 6 June, only roots of M. minima were collected as it was known to be colonized by AM fungi early in the growth season (O’Connor, 2001). Roots from each species in each plot were cut into approximately 1 cm lengths and cleared and stained in 0.05% trypan blue in lactic acid solution (Kormanick & McGraw, 1982). Root pieces were examined for mycorrhizal colonization at ×40–100 magnification by the grid intersect method (Giovannetti & Mosse, 1980) using a light microscope.
Plant productivity and diversity
On 9 August 1998,15 wks after treatments began (i.e. after seven applications), a 20 × 20 cm subplot within each field plot was sampled for number and above-ground biomass of all plant species. The shoots were removed by cutting at the soil surface and dried overnight at 80°C in an oven before measurement of shoot dry weights. Relative biomass of each species ((biomass per species ÷ total biomass of all species) × 100) and relative number of each species ((number per species ÷ total number of all species) × 100) were also calculated from the number and biomass of each species in each plot. Plant number and biomass per species were also used to calculate species richness (mean number of species per plot), diversity (Shannon H′) and evenness (Shannon J).
Mycorrhizal responsiveness of the most common plants in field plots was assessed in two different ways. At the start of the experiment (20 April 1998), polyvinylchloride (PVC) cylinders 15 cm tall and 10 cm in diameter were driven into the soil in an area adjacent to the experimental plot. The cylinders were positioned in such a way that 10 cores were collected for each of the major plant species (V. gracilis was not collected); cores had a PVC cap on the bottom, which had holes drilled in to allow free drainage. Seedlings were all at the cotyledon or first-leaf stage of growth at this early stage of the season. Cores were removed to a glasshouse and kept weeded and watered for 12 wks. At the end of the first and third weeks in the glasshouse, half the cores for each plant species were lowered into a solution of 1.5 g l−1 benomyl (a.i) for 5 min. These cores were designated as the mycorrhiza-suppressed treatment. The remaining cores were lowered into water for an equal length of time and were designated as the mycorrhizal treatment. Major plant species were also tested for mycorrhizal responsiveness (E. crinitum was not tested as seeds were unavailable) by growing them in 400-g closed pots of Mallala soil–sand mix (Dickson et al., 1999). Half of the pots for each species were designated as the mycorrhizal treatment and contained 10% soil and root inoculum from G. mosseae (isolate NBR 4-1) cultured on Trifolium subterraneum cv. Mt Barker in the glasshouse. Glomus mosseae NBR 4-1 has been shown to produce a growth response in a range of plant species (P. J. O’Connor unpublished). The other pots received an equal amount of nonmycorrhizal inoculum from pot cultures grown at the same time without mycorrhizal propagules. Pots were watered three times per week and received 10 ml of Long Ashton solution (without P) once a week. Plants were harvested after 6 wks’ growth and the extent of mycorrhizal colonization was assessed using the methods described above.
The unwatered control treatment was removed from all analyses of the field experiment, as it showed no difference from the watered control treatment with respect to the relative proportion of plant number and biomass per species. Watered control plots produced greater total above-ground biomass but did not differ from unwatered controls for any measures of community structure. All reference to controls hereafter is to the watered control treatment. Changes in community structure were analysed by manova on plant number and species biomass, for the mycorrhiza-suppressed and control treatments. Plots were used as replicates and fungicide application was used as the treatment. Plant species were included as variates but only the six most abundant species were included, since all other species represented less than 2% each of the total plot biomass and were irregularly distributed between the replicates. Because the manova indicated a significant treatment effect for both plant number (Wilk’s Lambda F = 4.17; df = 6,7; P = 0.051) and above-ground biomass (Wilk’s Lambda F = 43.88; df = 6,7; P < 0.0001), protected ANOVA was performed on species-by-number and species-by-biomass to assess the influence of treatment on individual species. Effects of suppression of mycorrhiza on total plot biomass, plant density, species richness, species diversity and evenness were analysed using one-way anova.
All percentage colonization data were arcsine transformed before analysis (Zar, 1999). Differences between treatments with respect to mycorrhizal colonization of plants from field plots were tested by two-way anova with species and treatment as factors. Differences in colonization of plants in field-collected cores and inoculated pots were tested using one-way anova for each species. The number of colony-forming units of fungi isolated from field plots was analysed for differences between treatments by one-way anova. Treatment differences were separated by Tukey’s HSD test for significantly different means in all experiments.
In the host plant responsiveness tests with field cores, average root length colonized was reduced to approximately 70% of controls when cores were treated with benomyl (Table 1). Root colonization was significantly reduced (P < 0.05) in all the cases except E. crinitum (although colonization was reduced to 72% of controls for this species). The species shown in Table 1 accounted for more than 96% of total plot biomass in the field experiment. Growth of the mycorrhizal host plants M. minima and V. arguta was significantly reduced in intact cores when colonization was suppressed. Growth of the host species S. verbenaca and E. crinitum was not significantly influenced by reduction in root colonization in benomyl-treated cores, while the nonhost C. annua showed a significant (P < 0.01) growth increase when benomyl was applied (Table 1). In the host plant responsiveness tests in inoculated pots (Table 1), there was no colonization of roots from any species in uninoculated pots. The growth of M. minima, V. arguta and V. gracilis was enhanced by inoculation with the mycorrhizal fungus G. mosseae (NBR 4-1). Salvia verbenaca showed no growth response to colonization by the mycorrhizal fungus and C. annua had a significant (P < 0.05) growth depression in inoculated pots (Table 1).
Table 1. Growth response of single plant species to mycorrhizal colonization in intact cores and inoculated pots in a glasshouse. Each value represents the mean (SE) of five replicates
The application of benomyl as a soil drench in field plots successfully suppressed mycorrhizal colonization with respect to controls throughout the growth period. By early June, colonization of M. minima roots had been reduced in benomyl-treated plots to approximately 66% of the level in controls (P < 0.001; Table 2). Total colonization of the roots of the most abundant host plant species, M. minima, S. verbenaca and V. arguta just before the onset of flowering (8 August 1998) was reduced in benomyl-treated plots to approximately 50% of that in control plots (Table 2). Carrichtera annua roots showed no signs of colonization by AM fungi.
Table 2. Per cent root length colonized with arbuscular mycorrhiza fungi in field plots treated with benomyl (fungicide) or water (control). Values represent the mean (SE) of seven replicates
Root length colonized (%)
NA, not applicable; ns, not significant (P > 0.05).
In vitro tests on the population of culturable fungi in field soils revealed that benomyl addition had a minor effect. The number of colony-forming units of culturable fungi was not significantly different (P = 0.128) between control plots and benomyl-treated plots (Table 3). Visual inspection of the roots of plants collected from field plots and intact cores revealed no signs of pathogenic fungi or their effects.
Table 3. Number of colony-forming units (cfu) of fungi from field plots treated with benomyl (fungicide) or water (control). Fungi cultured in vitro by dilution plating. Values represent the mean (SE) of seven replicates
cfu g−1 soil
Individual plant species response to suppression of AM fungi in field plots was only partly explained by mycorrhizal responsiveness, as determined in intact cores or pots inoculated with G. mosseae NBR 4-1 (Table 1). The dominant species, M. minima, showed a significant (P < 0.001) reduction in above-ground biomass in plots treated with benomyl relative to control plots (Fig. 1a). Carrichtera annua and S. verbenaca both showed significant (P < 0.001) increases in above-ground biomass when benomyl was added to field plots (Fig. 1a). Only S. verbenaca showed a significant (P < 0.05) change in plant density due to treatment. The number of S. verbenaca seedlings surviving was greater in benomyl-treated plots than in controls (Fig. 1b). Total plant density also increased in benomyl-treated plots (Table 4). None of the other species present at the site showed any significant alteration (P > 0.05) in abundance or biomass due to benomyl treatment.
Table 4. Total above-ground dry matter production, plant density and plant species richness, diversity and evenness in field plots treated with benomyl (fungicide) or water (control). Values represent the mean of seven replicates
Community indices Above-ground dry matter (g m−2)
Plant density (m−2)
Plant species richness (species m−2)
Plant species diversity (Shannon H′)
Plant species evenness(Shannon J)
Different letters indicate a significant (P < 0.05) difference between control and fungicide-treated plots as determined by Tukey’s honestly significant difference (HSD) test.
The structure of this semiarid herbland community was altered in benomyl-treated plots over the period of one growing season. There was no significant difference in total plot biomass between the benomyl-treated and control plots (P > 0.05; Table 4). Figure 2 shows the effects of mycorrhizal suppression on the relative contribution of each species to the abundance and biomass of the field plots. Benomyl applications resulted in a 62% reduction in the proportion of total plot biomass made up by the mycorrhizal plant M. minima (Fig. 2a,b). Benomyl treatment also resulted in an increase in the proportion of plot biomass made up by the nonmycorrhizal plant C. annua (159%) and the facultatively mycorrhizal species S. verbenaca (1540%). There was no significant increase (P > 0.05) in the relative contribution of any of the minor species (Fig. 2a,b).
The relative abundance of individual species was also altered by application of benomyl. The relative abundance of M. minima was reduced in benomyl-treated plots to 57% of controls (Fig. 2c,d). The relative abundance of C. annua remained approximately the same in different treatments, while S. verbenaca showed a 710% increase in relative abundance in benomyl-treated plots with respect to controls (Fig. 2c,d). Differences in relative abundance for minor species were not significantly affected by treatment (P > 0.05).
Alterations in the abundance and biomass of individual plant species caused by suppression of AM fungi did not translate into a net change in species richness (Table 4). Plant species diversity and evenness increased in mycorrhiza-suppressed plots relative to controls (Table 4). Diversity increased by approximately 29% and evenness by approximately 32% relative to controls. The minor species (all species excluding M. minima, C. annua and S. verbenaca) contributed equally (H′ = 0.41) to the diversity of both control and fungicide-treated plots. Differences in diversity between the treatments came from changes in the contribution of the three most abundant species. These species contributed 71% of H′ in benomyl-treated plots but only 63% in controls.
Community structure was significantly altered by suppression of mycorrhizal activity in this semiarid herbland. Plant species diversity increased when mycorrhizal colonization was suppressed because growth of the highly mycorrhiza-responsive dominant M. minima was reduced and subordinate plant species were released from competition. The mycorrhizal responsiveness of the subordinate plant species in the community was not as important in determining species diversity as the mycorrhiza-responsiveness and subsequent strength of competition from the dominant plant species.
Treatment of field plots with benomyl successfully reduced mycorrhizal colonization without significantly altering populations of other soil fungi. Benomyl treatments reduced total AM colonization in intact cores to approximately 70% of controls, and in field plots to approximately 50% of controls. Carey et al. (1992) showed that the effects of mycorrhizas were altered even when mycorrhizal root colonization was reduced to an average of approximately 70% of that in untreated controls. Smith et al. (1999) measured changes in individual plant growth, community structure and diversity when mycorrhizal root colonization was reduced in benomyl-treated plots to approximately 40% of that in untreated controls. The method used in this experiment to detect change in the levels of AM colonization was conservative and may not have detected dynamic changes throughout the growing season. The effect of benomyl on AM formation can be dose dependant (Schreiner & Bethlenfalvay, 1997) and may depend on the timing of application, soil heterogeneity and other site conditions (Pedersen & Sylvia, 1997). The use of root colonization levels as a surrogate measure of mycorrhizal activity may also result in overestimation of the capacity of the AM association to enhance P uptake. Larsen et al. (1996) showed that hyphal P-uptake was inhibited when benomyl was applied to an established hyphal network in soil.
Evidence linking reduction in the activity of AM fungi to changes in the plant community is further supported by the absence of symptoms of pathogenic fungi on the roots of plants grown in field plots and intact cores. Pathogenic fungi may interact with AM fungi to influence plant growth and fecundity (Carey et al., 1992; Newsham et al., 1994; Newsham et al., 1995). However, the consistent response to AM fungi of individual plant species found in pot and intact core experiments indicates that any phytotoxic effects of benomyl on plants, or fungitoxic effects on nontarget fungi, resulted in negligible effects on plant growth.
The decrease in mycorrhizal activity resulting from fungicide application caused significant changes in the abundance and productivity of the dominant species and in relative contributions of a number of species to community composition. The two most abundant species in control plots, M. minima and C. annua, were affected differently by the reduction in mycorrhizal activity. The dominant, highly mycorrhiza-responsive plant M. minima, had reduced productivity, while the nonmycorrhizal C. annua increased in biomass in mycorrhiza-suppressed plots. These results are readily explained as direct consequences of reduced mycorrhizal function in the system. The strongly mycorrhiza-responsive M. minima showed yield decline while the nonmycorrhizal C. annua was released from any antagonism with the AM fungi. At the same time, competition from M. minima was relieved. The increase in biomass and abundance of S. verbenaca is largely the result of increased survivorship in mycorrhiza-suppressed plots. This is presumably due to reduced competition from the highly mycorrhiza-responsive M. minima on the mycorrhizal but poorly responsive S. verbenaca.
Negative growth response to interaction with AM fungi has been previously observed for nonmycorrhizal plants (Francis & Read, 1995). The antagonism between C. annua and AM fungi was not large (> –34% mycorrhizal response) and is unlikely to account for all the increased biomass of this species in fungicide-treated plots. Sanders & Koide (1994) found a similar relationship between the decline in biomass of the mycorrhiza-responsive plant Abutilon theophrasti and the concomitant increase in biomass of the nonmycorrhizal species Amaranthus retroflexus. The nonmycorrhizal species C. annua, and the poorly mycorrhiza-responsive species S. verbenaca behave in the same way in our study as the poorly mycorrhiza-responsive plant species in the study of Hartnett & Wilson (1999). The productivity of species with low mycorrhizal dependency in our semiarid herbland was not restricted by loss of mycorrhizal function after competitive release from the highly mycorrhiza-responsive M. minima.
The change in biomass of the three most abundant species in our semiarid herbland accounted for all the increase in diversity (29%) in mycorrhiza-suppressed plots. However, equal contributions of the minor species to diversity in mycorrhiza-suppressed or control plots indicates that loss of mycorrhizal activity did not result in a significant decline in the highly mycorrhiza-responsive species V. gracilis and V. arguta. since neither of these minor species showed significant reduction in above-ground biomass or number in response to mycorrhiza-suppression it is concluded that release from competition with the highly mycorrhiza-responsive M. minima offset the cost of losing mycorrhizal function. This indicates that while some of the minor plant species at this site were highly responsive to mycorrhiza, they may not have been as competitive as M. minima in acquiring resources from the external mycorrhizal mycelium. The high root density of the dominant plant, M. minima, in control plots may also have reduced mycorrhizal benefit to subordinate species. The restricted growth of minor host-plant species could have resulted from reduced mycorrhizal benefit at high plant densities (Koide & Li, 1991; Koide, 1991). Competition for nutrients in this plant community may extend from competition between plants along the mutualism–parasitism continuum (Johnson et al., 1997) to competition between plants with equivalent mycorrhizal responsiveness but different capacities to exploit the symbiosis for nutrient uptake or alleviation of water stress.
Reduction in the activity of mycorrhizal fungi resulted in an increase in plant species diversity owing to an increase in species evenness in this semiarid plant community. This supports the findings of Hartnett & Wilson (1999) that diversity (both richness and evenness) increased when mycorrhizal fungi were suppressed over several growing seasons in a tallgrass prairie. Our results conflict with the microcosm experiment of Grime et al. (1987) who showed that mycorrhizal association increased plant species diversity (owing to increased evenness), and the field experiment of Gange et al. (1990), who showed that reduction in mycorrhizal activity was correlated to a decrease in plant species richness. Mycorrhizal associations were also shown to increase plant species diversity in macrocosms simulating North American old-field ecosystems and increase evenness in microcosms simulating a European calcareous grassland (Van der Heijden et al., 1998b). Our results support the hypothesis that mycorrhizal effects on plant species diversity are not absolute but depend on the mycorrhiza-responsiveness of the component plant species, especially the dominant species (Bergelson & Crawley, 1988; Hartnett & Wilson, 1999).
Further support for this hypothesis comes from our study, where plant species richness was unchanged by suppression of AM fungi, as was functional group richness. Decreased mycorrhizal activity resulted in a redistribution of biomass within the community, i.e. an increase in evenness, but no change in above-ground biomass of the community as a whole. This was also true in the studies of European calcareous grassland in microcosms (Van der Heijden et al., 1998b; but see Wardle, 1999), and tallgrass prairie in the field (Hartnett & Wilson, 1999). The fact that increasing diversity in our experiment did not coincide with increasing productivity suggests that productivity may be linked to species or functional group richness (Naeem et al., 1994; Tilman et al., 1996; Tilman et al., 1997; Symstad et al., 1998; Hector, 1999; Wilsey & Potvin, 2000), neither of which was changed in the current study. There is no evidence to suggest that the effect of increasing diversity on plant productivity found in other studies (Tilman et al., 1997; Hector, 1999) can be explained by increasing evenness (Wilsey & Potvin, 2000). Further study of the importance of functional group richness in plant community productivity and stability, should consider mycorrhizal responsiveness as a key functional characteristic of plant species, particularly in disturbed and early successional communities.
The behavior of individual plant species was a strong determinant of community structure. Changes in species composition following fungicide treatment of field plots altered plant density. Changes in plant density were largely attributable to increased survivorship of seedlings of S. verbenaca in mycorrhiza-suppressed plots. This highlights the different response of species to competitive release. There was an increase in above-ground biomass of S. verbenaca related to increased seedling survival, while C. annua showed increased plot biomass without significant adjustment of survival. These differences may result in long-term effects on community structure not seen in one growing season, especially since S. verbenaca is a short-lived perennial.
The seasonal fluctuations in aridity and levels of standing biomass may influence mycorrhizal activity in this semiarid herbland. Severe grazing pressure may also affect below-ground allocation of carbon and hence interactions between soil organisms and the plant community. The present plant community reflects previous grazing disturbance by large populations of kangaroos. Overgrazing has facilitated the establishment of several ruderal species including the nonmycorrhizal C. annua and the unresponsive host plant S. verbenaca, probably as a result of the decline of native herbs and grasses, and the associated soil flora. While not investigated in this study, it has been observed that some specificity between isolates of AM fungi and plant species from a grassland ecosystem (Van der Heijden et al., 1998a) exists. It is therefore possible that AM fungal communities enact a determinate role in plant community structure (Van der Heijden et al., 1998b). Restoration of the semiarid grasslands of southern Australia will require ecological approaches to the re-establishment of some plant species, and greater understanding of soil processes in disturbed and undisturbed systems.
We wish to thank Anna Ziersch, Justin O’Connor, Shaez Mortimer and Maria Roez for assistance in the field and Evelina Facelli for advice and assistance with statistical analysis. We also thank reviewers of this paper for their useful suggestions. This research was supported by an Australian Post-graduate Research Award to P. O’C.