Soil pathogenic fungi have the potential to affect the co-existence of two tallgrass prairie species


Present address and correspondence: Department of Biology, Indiana University, Bloomington, IN 47405, USA (fax 812 855 6705; e-mail


1 Negative feedback may exist between plant species and their biotic soil communities. Two co-occurring native tallgrass species (Andropogon gerardii, a perennial grass, and Chamaecrista fasciculata, an annual legume) were reciprocally transplanted into pots containing soil from areas dominated by one of the species. Half of the soil derived from each area was microwaved to reduce soil fungi, resulting in four different ‘soil types’.

2Chamaecrista fasciculata plants were smaller when grown in their native soil vs. that from under A. gerardii, but were unaffected by microwaving (i.e. fungal reduction). Andropogon gerardii plants were shorter with fewer inflorescences in non-microwaved soil, with the poorest growth occurring in non-microwaved C. fasciculata soil.

3 Fungi were isolated from roots of A. gerardii growing in the different soil types. We tested whether the fungi found differed between the four soil types and whether any species characteristic of C. fasciculata soil were responsible for the poor growth of A. gerardii in this medium.

4 Fungi unique to the non-microwaved C. fasciculata soil type reduced tillering and caused an early reduction in growth of A. gerardii. These effects were partially ameliorated when potentially mycoparasitic fungi associated with A. gerardii were also included. By the end of the experiment, both fungal treatments increased above-ground biomass but greatly reduced below-ground biomass of A. gerardii compared with controls, suggesting that the exposure to potential fungal pathogens from C. fasciculata soil altered biomass allocation within plants.

5 There was no evidence of negative feedback between the prairie species and their own soils. However, fungi associated with Chamaecrista fasciculata were detrimental to A. gerardii, one of the dominant perennials in the surrounding area, and may facilitate the annual’s long-term persistence. Arbuscular mycorrhizae did not ameliorate the pathogenic effects of these fungi as there was little colonization of A. gerardii roots in C. fasciculata soil.


Many mechanisms have been proposed whereby a large number of plant species can co-exist in a single community ( Harper 1977; Tilman 1982; Silvertown & Law 1987; Tilman & Pacala 1993). Native herbivores may contribute to the maintenance of plant species diversity ( Janzen 1970; Hulme 1996), and organisms that cause disease may play a similar role, as well as determining species distribution and the dynamics of individual populations ( Augspurger 1988; Harper 1990; Alexander 1992). Although, compared to agricultural systems, natural systems may harbour a much more diverse and equally devastating collection of pathogens relative to the diversity of hosts ( Kranz 1990), they have rarely been studied. Indigenous diseases are often deemed ‘insignificant’ to the supposed resilience of natural communities ( Schmidt 1978; Dinoor & Eshed 1984) and studies have traditionally focused on the more dramatic effects of introduced pathogens, such as Chestnut bight, Dutch Elm disease and in Australia the Jarrah forest dieback ( Harper 1990; Alexander 1992). Research is limited because diseases are caused by microscopic organisms that are difficult to see and study directly, diverse plant communities can easily ‘mask’ disease effects as other species may compensate for damage done to neighbours, and current plant distributions may reflect the actions of past pathogens (the ‘ghost of diseases past’) rather than any current disease presence ( Harper 1990).

Most studies on native plant diseases have focused on those fungi that affect the above-ground portions of plants and cause visible lesions or deformities ( Tiffany et al. 1990 ; Alexander et al. 1996 ; Burdon & Silk 1997). However, many of the most devastating diseases in agricultural systems, as well as in some natural systems ( Weste & Ashton 1994), are caused by below-ground pathogens ( Agrios 1988). Recent research on coastal foredune vegetation ( van der Putten et al. 1993 ; van der Putten & Peters 1997) and old-field communities ( Bever 1994) shows that plant species grow better in soil associated with other plant species than in their own soil. This suggests that there is a ‘negative feedback’ between the biotic soil community of a particular plant species and the growth of that plant species, and that this feedback may be due to long-term accumulation of host-specific fungal soil pathogens. This could prevent, or delay, competitive exclusion of a particular species, and thus affect community characteristics such as diversity, successional rates and co-existence of species ( van der Putten et al. 1993 ; Bever 1994). However, differential effects of disease on plant species that are stronger competitors, as has been shown with predator–prey interactions ( Paine 1974; Louda et al. 1990 ), could also maintain diversity. For instance, two species, A and B, may co-exist even if B is the stronger competitor if B is the species that is more affected by pathogen(s) associated primarily with species A ( Chilvers & Brittain 1972).

Most living plant biomass in tallgrass prairie is underground ( Weaver 1968) and we evaluated negative feedback between plants and soil-borne pathogens in species differing in life histories. Chamaecrista fasciculata, an annual legume, and Andropogon gerardii, a dominant perennial grass, have co-existed in the study area for a long period of time. Seedlings were grown in soils from areas dominated by one of the two prairie species and the effects of reduced levels of both pathogenic and mutualistic fungi (via microwaving) were determined. Although our primary goal was to assess if there was an interaction between soil origin and treatment (i.e. is plant response in a specific soil type dependent on the fungal reduction treatment), we could also determine whether pathogens or mutualists dominated the fungal soil community by comparing growth in microwaved vs. non-microwaved soil. The role of soil pathogens particular to soil types was explored by isolating fungi from within roots and assessing their effects on growth. We also examined whether the presence of arbuscular mycorrhizae (AM) was associated with an amelioration of disease effects, as suggested by Newsham et al. (1995) . The possible impact of fungal pathogens and mutualists on species co-existence in this community could then be assessed.

Materials and methods

Study organisms and study site

Andropogon gerardii Vitman (big bluestem) is a C4 perennial grass (Poaceae) that attains heights of 2 m or more. It is one of the dominant tallgrasses on the prairie in north-eastern Kansas, USA, as well as throughout much of the Great Plains, and historically has been the major constituent of the biomass of the lowland prairie and much of the upland area ( Weaver & Fitzpatrick 1934). Vegetative reproduction of this perennial is common and involves production of tillers (new shoots) at the base of mature plants. Chamaecrista fasciculata (Michx.) Greene-Rydberg (showy partridge pea, formerly known as Cassia chamaecrista and Cassia fasciculata) is an annual legume (Caesalpiniaceae), 10–120 cm tall, that blooms in the early autumn in north-eastern Kansas. Although it is native to tallgrass prairies, C. fasciculata commonly grows in disturbed areas and is often found on roadsides.

The study area was located on the DogLeg Prairie Preserve at the Kansas Ecological Reserves (Jefferson County, T11S, R20E), approximately 13 km north of Lawrence, Kansas, USA. The area consists of 0.8 ha of unploughed prairie as well as a more disturbed strip (0.02 ha) alongside a road. Chamaecrista fasciculata dominates this strip (approximately 75 m long and 3 m wide), where it has been present since records of the vegetation began (i.e. at least 8 years; W.D. Kettle, personal communication). The area further from the road comprises species more typical of undisturbed prairie, notably A. gerardii, which forms dense clumps here.

Soil collection and treatment

Twenty soil cores, 20 cm deep and 2.5 cm in diameter, were taken from beneath large clumps of C. fasciculata and another 18 from beneath A. gerardii in an area of approximately 900 m2 within the study site during August 1995. Each soil core was taken at a position where the surrounding 1 m2 was dominated by, but did not contain exclusively, either C. fasciculata or A. gerardii. Cores from areas dominated by either C. fasciculata or A. gerardii were bulked, yielding 1600–1700 g of soil of each type. Half of each type of soil was microwaved on high power (950 watts/2450 MHz) for 300 s. Ferriss (1984) found that microwaving 1 kg of soil (between 7% and 37% water content) for 150 s or more eliminated, or at least reduced, the population sizes of many common soil-borne pathogenic fungi without destroying the integrity of the prokaryotic and nutrient components of the soil. The elimination of pathogens from soil using this method does not depend on soil type. Fungi found at all depths of the core should have been affected equally by microwaving in our bulked samples, and all soil invertebrates, such as nematodes, many of which are probably parasitic on plants, were destroyed. Four different ‘soil types’ were thus available: A-NM (soil from areas dominated by A. gerardii, with no microwaving), A-M (as A-NM but microwaved), C-NM (from areas dominated by C. fasciculata, with no microwaving), and C-M (as C-NM but microwaved).

Experiment 1: reciprocal transplanting in different soil types

For each of the four soil types, 10 500-ml (100-cm2) pots were filled with a mixture of vermiculite and 74–80 g of one of the four soil types. This allowed even water retention despite the high clay content of the field soils. Chamaecrista fasciculata seeds were pretreated with sulphuric acid and water to enhance germination (M. Christianson, personal communication), and 25 seeds were then planted in five pots of each soil type. There had been a series of dry late summers, so that A. gerardii seeds were largely unviable, and seed (from Kansas/Missouri stock) was obtained from Sharp Bros. Seed Co. (Clinton, Missouri, USA). Even then, germination was poor compared with C. fasciculata and seeds were germinated in trays before transferring 25 10-day-old seedlings to the five remaining pots of each soil type. Due to a planting error, only four pots contained C-NM soil sown with C. fasciculata seeds. Plants were kept in a greenhouse and watered daily, but not fertilized.

Germination of C. fasciculata and survival of A. gerardii seedlings were recorded on day 23 and day 32 after planting. On day 57, C. fasciculata survivorship in each pot was recorded together with the number of leaves > 7 cm per pot (a measure correlated with the overall biomass of C. fasciculata plants; H. Alexander, personal observation). Height and tiller number of each A. gerardii plant were recorded on days 57, 110 and 135 after planting. Height is highly correlated with biomass (dry weight = –0.0161 + 0.0134 × height, R2 = 0.90; J. Holah, unpublished data). Inflorescence number was counted for A. gerardii for the last data collection.

Two-way analysis of variance was used to analyse the data using SAS ( SAS Institute Inc. 1988). Type III sums of squares were interpreted as they are not affected by ordering of model parameters and take into account unequal sample sizes.

Assessing biotic soil communities associated with a. gerardii

Because the effect of microwaving on plant performance was statistically significant only for A. gerardii (see the Results), we restricted our analysis to this species. At the conclusion of the reciprocal transplanting experiment, roots were separated from above-ground parts and cleaned; those from each pot were divided into 12 subsamples of approximately equal size, surface sterilized for 5 min in a 10% hypochlorite solution, and then washed in distilled water. Ten to 12 root segments (approximately 2.5 cm long) from each subsample were placed together on a potato dextrose agar plate (Sigma, St Louis, Missouri, USA; 39 g l–1 water). At intervals of approximately 18–20 h, plated root segments were checked for hyphal growth under a microscope. Areas of growing hyphae were marked and hyphae with similar characteristics were grouped and assigned a code and subcultured to allow characterization: hyphal types that occurred only once were assigned a unique number. The fungi were generally not identified into known taxonomic groups, but distinct species could be separated that differed in cultural characteristics such as hyphal type, growth rate, growth on different media, and presence–absence and type of fruiting bodies. Specimens were regrouped accordingly and each group was assumed to represent a unique fungal taxon.

Andropogon gerardii roots were also stained for AM using trypan-blue staining methods ( Phillips & Hayman 1970), and percentage colonization was assessed by counting vesicles under a dissecting microscope ( Daniels et al. 1981 ). This was done for half of the collected roots from three pots with A-M, A-NM, and C-NM soils, and two pots with C-M soil. This estimate was conservative, as any non-vesicle-forming mycorrhizal species would have been missed. A small sample of root biomass (approximately 1.0–1.5 g) from the same pots was treated to assess the presence of Oomycete fungi, which, although often important plant pathogens, can be slow-growing in culture and hence missed in routine sampling where they are easily outcompeted by other fungi ( Carlile & Watkinson 1994). A baiting technique was used: fruit (pears) were placed in distilled water and water agar (2%) containing 250 p.p.m. streptomycin was used to isolate fungi from any lesions that formed ( Mitchell & Kannwischer-Mitchell 1992).

A Multi-Response Permutation Procedure (MRPP on PC-ORD) ( McCune & Mefford 1995) was used to assess the randomness of fungal occurrence by soil type. MRPP is a distribution-free test and can be used to assess how often the grouping of objects from some collection could be expected by chance. Presence–absence data for all fungal species found more than once were used, rather than the number of times a fungal species was recovered from a soil type, because unequal growth rates in culture may mean that the amount of inoculum present in roots may not be measured accurately by the recovery data. Arcsine square root transforms were used on percentage colonization data of AM fungi for anova, and type III sums of squares were used.

experiment 2: replanting A. gerardii with recovered fungi

Andropogon gerardii experienced the poorest growth on C-NM soil (see the Results). All five fungal species found only in roots from this soil type, as well as a Rhizoctonia spp. found primarily in this type of soil, were raised in pure culture. Cultures were raised for inoculation by the technique of Martin (1992) as modified by Larkin et al. (1995). Samples were placed in separate plastic bags containing a mixture of autoclaved vermiculite (500 ml), cornmeal (10 ml), distilled water (100 ml) and vegetable juice (V8) (100 ml) and allowed to grow for approximately 2 weeks in the bags, by which time the medium appeared to be fully colonized. A greenhouse experiment was established with four treatments. To create the main fungal treatment (F pots), 10 ml of each type of inoculated medium was mixed in a 100-cm2 pot with a 1 : 2 mixture of microwaved potting soil and vermiculite, a ratio previously determined to allow the best assessment of pathogenicity of the fungal isolate (J. Holah, unpublished data). So as not to confuse pathogen effects with effects of potential mutualists, microwaved potting soil was used to reduce the chances of mycorrhizal colonization from any spores present in the greenhouse soil. Two types of control treatments were used: pots without any added fungal inoculum (CC pots), and pots in which the fungal inoculum had been autoclaved (C pots). The fourth treatment consisted of mixing the fungal mixture with an additional 10 ml of each of two species of Trichoderma (F + T pots), a mycoparasitic fungi often used for biocontrol in agriculture ( Agrios 1988; Carlile & Watkinson 1994) that occurred frequently only in A soil (i.e. soil only from A. gerardii areas).

Five seedlings of A. gerardii approximately 10 days old were transplanted into the pots. Pots were kept in the same greenhouse as before, watered daily, and fertilized weekly with Technigrow (20-18-18 plus micros; Sun Grow Horticulture Inc., Bellevue, WA) at a concentration of 400 p.p.m. There were five replicates per treatment. Initial heights were recorded and data on tiller number, heights and inflorescence number per pot were recorded on days 13, 26, 45, 64 and 128 after planting. Although the rate of increase in height had slowed by 64 days, there was no indication that pot size was severely restricting plant size. After flowering, which occurred during the last time period, plants began to senesce. Data were taken on a per plot basis because individual plants were often indistinguishable due to extensive tillering. At the conclusion of the experiment, grass plants were removed from the soil, above- and below-ground biomass separated, and roots thoroughly washed. Above- and below-ground biomasses were dried in a 65 °C oven and weighed.

Results were analysed using a one-way analysis of variance, planned comparisons between pots with fungal treatments and the controls, and a repeated measures anova using SAS ( SAS Institute Inc. 1988). Type III sums of squares were used.


Experiment 1: reciprocal transplanting in different soil types

By 57 days after initial seeding neither survival nor size of C. fasciculata was significantly different in microwaved and unmicrowaved soil of the same type ( Fig. 1a,b) although survival tended to be higher in untreated soil ( Fig. 1a). The number of leaves larger than 7 cm was significantly greater when C. fasciculata was grown in ‘foreign’ soil (A soil) rather than in soil from its own sites (C soil), regardless of the microwaving treatment ( Fig. 1b).

Figure 1.

Survival and growth after 57 days in experiment 1. (a) The proportion of surviving Chamaecrista fasciculata (Chfa) seeds (± SE); (b) the average number of C. fasciculata leaves that are greater than 7 cm (± SE); (c) the average number (± SE) of Andropogon gerardii (Ange) tillers; (d) the average height of A. gerardii (± SE) in soils that were microwaved (M) or not (NM) from sites that were dominated by either A. gerardii (A soil origin) or C. fasciculata (C soil origin). F-statistics with associated P-values are given for a two-way anova. Degrees of freedom are (1, 15) for (a) and (b) and (1, 16) for (c) and (d).

By 57 days after transplanting A. gerardii seedlings had developed the greatest number of tillers in the A-M soil (mean 7.8) and least in the C-NM soil (mean 5.6), but neither of the main effects (soil origin and treatment) nor their interaction was significant ( Fig. 1c). The average height was higher in microwaved than in untreated soil but the effect of the microwaving treatment depended on soil origin (interaction marginally significant; Fig. 1d). The interaction term between treatment and soil origin was still nearly significant after 110 days (F116 = 3.47, P = 0.081), but the effect had disappeared by the end of the experiment (135 days, treatment × soil F1,15 = 1.71, P = 0.21). At all times the height of A. gerardii was less in untreated than in microwaved soil types and the mean value in C-NM soil was always at least 10 cm less than in A-NM soil. Mean height varied from 38 to 54 cm between treatments at 57 days, 57–77 cm at 110 days and 65–86 cm at 135 days.

The average number of A. gerardii inflorescences per pot in the C-NM soil was less than one-quarter of the number in any of the other soil types [mean = 0.6 ± 0.26 (SE) vs. means ranging from 3.8 to 5.2 inflorescences pot–1]. The interaction term (soil type × treatment), however, was not significant, probably due to small sample size (F1,15 = 1.29, P = 0.27). Microwaving significantly increased the average number of inflorescences (F1,15 = 5.99, P = 0.03), while the effect of soil origin was marginally significant for inflorescence number (F1,15 = 3.38, P = 0.097).

Assessing biotic soil communities associated with a. gerardii

Fungal collection results

A total of 21 distinct fungi (‘species’) occurred more than once on the general plating media. Five of these were unique to C-NM soil (i.e. were recovered only from this soil type) and comprised an Oomycete, a Rhizoctonia, a stage of Monascus, and two unidentified fungal species. C-NM soil contained more ‘unique’ fungi than all other soil types combined. Plants grown in C. fasciculata areas generally had more unique species associated with them than when grown in A. gerardii areas, and, as would be expected, those from non-microwaved field soil contained more unique species than microwaved soil ( Table 1).

Table 1.  Number of fungal species found in each soil type, as well as the number of unique fungal species found for each type, i.e. species found only for that particular soil type. See text for abbreviations
Soil typeNumber of fungal speciesNumber of unique fungal species
All A soil225
All C soil228
All NM soil258
All M soil193

Overall the presence or absence of fungal species in particular soil types had a 20% chance of occurring at random (MRPP, T = –0.799). There was a 25.2% probability (MRPP = –0.574) of isolate groups being random if grouped according to microwaved and non-microwaved categories, but only 10% (MRPP, T = –1.33) if soil origin (C vs. A soil) was the criterion for defining groups. These results indicate that soil origin was the most meaningful variable to explain the distribution of fungal species across the four experimental soil types.

Results for mycorrhizae and Oomycete fungi

The only soil type in which A. gerardii roots were significantly colonized by AM was A-NM (15.9% vs. < 5% in any other type). Both the main effects of treatment (M vs. NM) and soil origin (C vs. A) were significant (F1.7 = 8.11, P = 0.025 and F1.7 = 16.87, P = 0.005, respectively), and the interaction term was marginally significant (F1.7 = 4.90, P = 0.06).

Only A. gerardii roots from two pots, both containing A-NM soil, yielded any Oomycetes using pear baiting techniques.

Experiment 2: replanting a. gerardii with recovered fungi

The height of A. gerardii was significantly less in the basic fungal mixture compared with all other treatments for day 26 after planting ( anova contrast, F1,16 = 13.86, P = 0.002) and for day 45 after planting ( anova contrast, F1,16 = 9.04, P = 0.008) ( Fig. 2). By 125 days however, the height of A. gerardii in both types of fungal treatment pots significantly exceeded that of the control treatments ( anova contrast, F1.,16 = 15.22, P = 0.001). Over time, plants grew differently with regards to the different treatments (repeated measures anova, Wilks’ Lambda 0.125, P = 0.002).

Figure 2.

The average height of Andropogon gerardii (Ange) per pot (± SE) in the control treatments (trt) (CC, control treatment with no addition to the soil; C, control treatment with autoclaved fungal mixture added to soil) and fungal treatments (F, live fungal collection added to soil; F + T, live fungal collection with Trichoderma spp. added to the soil). Asterisks indicate degree of significance with a one-way anova: *,** the P-value associated with the F-statistic (with 1, 16 degrees of freedom) is < 0.02 and < 0.007, respectively.

Sixty-four days from the initial planting, the number of tillers per pot was significantly less in the pots inoculated with the fungal mixture alone compared with all other treatments ( anova contrast, F1,16 = 7.47, P = 0.015), and remained so until the conclusion of the experiment, 128 days after the initial planting ( anova contrast, F1,16 = 6.05, P = 0.026) ( Fig. 3). By day 64, tiller number was slightly higher if the Trichoderma species were part of the fungal mixture, but the average tiller number of plants for both fungal treatments was lower than the control treatment ( anova contrast, F1,16 = 6.85, P = 0.019). By day 128, this difference was less ( anova contrast, F1,16 = 3.41, P = 0.08).

Figure 3.

The average number of Andropogon gerardii (Ange) tillers per pot (± SE) in the fungal treatment and control pots during the course of the experiment (see Fig. 2 for abbreviations).

Plants in different treatments produced different numbers of tillers throughout the course of the experiment (repeated measures anova, Wilks’ Lambda 0.422, P = 0.033).

Although the average below-ground biomass was slightly higher if the fungal mixture had the Trichoderma spp. added, both types of fungal treatments were significantly lower in biomass than control treatments ( Table 2). In contrast to these results, the final A. gerardii above-ground biomass was significantly higher in both the fungal treatments compared with the control treatments, and the fungal mixture treatment alone was significantly higher than both the controls ( anova contrast, F1,16 = 5.32, P = 0.035). Also, the average number of inflorescences in both fungal mixture treatments was significantly higher than in controls ( Table 2).

Table 2.  Average above- and below-ground biomass of A. gerardii and number of inflorescences per pot for the four treatments (CC, control treatment with no addition to soil; C, control treatment with autoclaved fungal mixture added to soil; F, live fungal collection added to soil; F + T, live fungal collection with Trichoderma spp. added to the soil) are listed with standard errors. F-statistics are given for anova contrasts between the two control treatments and the two fungal treatments. **Results significant at < 0.01; ***results significant at < 0.002
 CCCFF + Tanova contrast (controls vs. fungal treatments)
Above-ground biomass (g)14.16 (± 0.34)12.06 (± 1.24)16.84 (± 1.38)17.66 (± 1.85)F1,16 = 9.83 **
Below-ground biomass (g)9.78 (± 0.60)9.68 (± 0.44)6.10 (± 0.22)6.52 (± 0.45)F1,16 = 58.48 ***
Average number of
inflorescences pot–1
4.20 (± 0.58)4.00 (± 1.38)8.20 (± 1.24)10.00 (± 1.76)F1,16 = 14.53 ***


Differentiation of biotic soil communities

The biotic soil communities associated with two co-existing prairie species had variable effects on the growth of those species under greenhouse conditions. The seed survival and biomass, as indicated by leaf number, for C. fasciculata was reduced in its own soil by microwaving, suggesting a positive association between C. fasciculata and its own biotic soil community. Because C. fasciculata is a nodulating legume species ( Foote & Jackobs 1966) and microwaving of soil may affect some prokaryotes ( Ferriss 1984), this may not be surprising. A height reduction for A. gerardii in its own untreated soil compared with microwaved soil, together with a similar trend for tiller number, provide evidence for the opposite relationship in this species. Thus, for A. gerardii the effects of potential soil mutualists, such as mycorrhizae, may be outweighed by the effects of pathogenic soil organisms. Although this may be somewhat surprising given reports of the positive effects of mycorrhizae on A. gerardii growth on the Konza Prairie in central Kansas ( Hetrick et al. 1988 ; Hartnett et al. 1994 ), this grass does not seem to be nearly as dependent on mycorrhizae in other, more nutrient-rich prairies ( Anderson et al. 1994 ). Because A. gerardii height is greater on the Kansas Ecological Reserve study site compared with heights attained on the Konza prairie and the soil horizon is deeper, the relationship between A. gerardii and its mycorrhizal symbionts may be similar to that found in more eastern prairies ( Anderson et al. 1994 ).

Andropogon gerardii growth, specifically its below-ground biomass and tillering, is inhibited by C. fasciculata soil. The lack of this inhibition when the same soil is microwaved suggests that it is due to some biotic component of the legume’s soil. Species are known to differ in their response to the biotic community associated with any one soil type ( Martin et al. 1956 ; van der Putten et al. 1993 ; Bever 1994; Bever et al. 1996 ). Differential responses to mycorrhizal communities ( Hartnett et al. 1993 ; Sanders 1993; Bever et al. 1996 ) and bacterial communities ( Westover et al. 1997 ) in natural systems may be commonplace. In all of the few systems examined, the biotic soil community associated with a plant species suppresses that plant’s growth more than the biotic community associated with other plant species ( van der Putten et al. 1993 ; van der Putten & Peters 1997; Bever 1994). In models assuming equal competitive ability between species, negative feedback between plant species and their soil communities can lead to maintenance of species diversity ( Bever et al. 1997 ). However, if the competitive abilities of plant species are very different, as is likely between a tillering perennial such as A. gerardii and a weedy annual like C. fasciculata, negative feedback may not be necessary to explain co-existence. A competitively dominant species may not exclude a less competitive neighbour if that neighbouring species is associated with pathogens, or predators, that preferentially attack the competitive species. This is analogous to Paine’s landmark study ( Paine 1974) of a predator maintaining competing prey species in the intertidal zone and other studies of plant fungal pathogens that have demonstrated the role of disease in maintaining competing plant species in a community ( Burdon et al. 1984 ; Paul & Ayres 1990). Competition trials between the perennial grass and annual legume would be necessary to assess whether competition does indeed exist between these species and whether it is asymmetric.

The effects of the biotic community may normally be directly or indirectly mediated by the associated plant species, but we found evidence that the growth of A. gerardii is inhibited by non-microwaved C-type soil even if no C. fasciculata is present. Soil pathogens are probable candidates for causing this response, and we therefore tried to pinpoint those responsible.

The role of mycorrhizae, fungal pathogens and other members of the biotic soil community

The fungi isolated from A. gerardii in the different soil types were taken from within roots after surface sterilization and not from the soil itself, and were therefore likely to be pathogenic rather than saprophytic. Those fungi that were unique to the soil in which A. gerardii did worst (C-NM) did indeed have a negative effect on tillering, below-ground biomass and, initially, on above-ground growth, suggesting that they, rather than another component of the soil community, caused the observed inhibition of growth. However, soil invertebrates, which were eliminated by the microwaving treatment, and potential allelopathic chemicals associated with C. fasciculata, may also have contributed to the results of the first experiment.

It is interesting that the negative effects of C-NM soil on A. gerardii were severe in the first experiment, particularly in terms of above-ground growth and flowering, whereas by the end of the second experiment these measures had been enhanced by the fungal collection derived from this soil. This could be due to regular fertilization applied during the second experiment. Mycorrhizae should not have impacted the second experiment, as only specific fungal cultures were added to a sterile potting medium. AM fungi can alter allocation patterns (Streitwolf-Engel et al. 1997), and this may have contributed to the discrepancies in above-ground biomass results for the first and second experiments.

Microwaving soil did appear to reduce fungal inoculum: there was very little colonization of A. gerardii roots by AM fungi in microwaved A soil whereas, in untreated A soil, colonization was comparable to rates found previously for warm season grasses ( Hetrick et al. 1990 ). The low colonization observed in both treated and untreated C. fasciculata soil could be due either to absence of AM propagules or low AM colonization, even in untreated soil, due to the action of pathogenic organisms. A protective role for AM fungi, as suggested by Newsham et al. (1995) , cannot be discounted for A. gerardii. However, the absence of AM cannot by itself explain the poor growth of A. gerardii in C-NM soil because growth was not inhibited in microwaved soil, which contained few AM propagules.

An alternative explanation for the better growth of A. gerardii in its own soil compared to that of the legume (at least when untreated) is differences in mycorrhizal community rather than the pathogenic soil community between the two species due to specificity of plant–mycorrhizal associations ( Sanders 1993; Wetzel & van der Valk 1996). However, isolations from A. gerardii roots clearly showed differences between the fungal pathogen communities from the two types of soils. Soil origin (C vs. A soil) was also the strongest predictor of fungal classification in the randomization test (MRPP) used, which would not be the case if the potential pathogenic communities had been homogeneous.

The Trichoderma spp. often found in A soil were initially thought to act as potential mutualists to A. gerardii owing to the known mycoparasitic characteristics of these fungi ( Agrios 1988; Carlile & Watkinson 1994). Although the addition of these fungi did seem to ameliorate the initial decrease in growth and tillering experienced by plants exposed to the fungal mixture derived from C-NM soil alone, the significant reduction in below-ground biomass occurred regardless of Trichoderma presence. Thus, any protection may be short-term during the plant/tiller establishment phase, not long-term protection in the allocation of resources to perennating structures over time. However, the activity of Trichoderma could be dependent on pH and soil nutrients, which were not measured in this study.

The baiting techniques used to assess the presence of Oomycete fungi in the soil yielded few species. This suggests that populations of these fungi within A. gerardii roots were not significant, although one of the fungi unique to C-NM soil was a potentially pathogenic Oomycete. It should be noted that not all potential fungal pathogens would have grown on the general fungal media used, and other components of the biotic soil community, such as invertebrates, nematodes, protozoa and bacteria, were beyond the scope of this study.

The fact that C. fasciculata plants are associated with a biotic soil community that inhibits A. gerardii may be important for the persistence of an annual species with perennial neighbours. Although the in situ physical characteristics of the soils associated with A. gerardii in more intact prairie and C. fasciculata in more disturbed areas may have been very different, when microwaved they supported similar biomasses of A. gerardii (in terms of height). Untreated C. fasciculata soil inhibited tillering of A. gerardii and so germinating seedlings of this annual may not be threatened by encroaching tillers of A. gerardii. The hypothesis that the association of C. fasciculata with pathogenic organisms that primarily affect A. gerardii increases the likelihood of the annual persisting in the community needs to be tested by competition studies between the species.

This study, as well as previous studies on negative feedback, point to the fine-scale below-ground heterogeneity possible within plant communities. The heterogeneity is driven in part by the presence of the particular plant species themselves, and the interactions between species that result are probably complex and species-specific. Just as previous research has found that plant-specific herbivores ( Hulme 1996) could underpin the presence or absence of a species in the community, we need heightened awareness of the potential role of microscopic ‘predators’ that are ubiquitous both above-and below-ground.


The authors would lie to thank Drs Robert Holt and Norman Slade for their insight and helpful comments on the manuscript, and Dr Jeanne Mihail for her expertise in fungal matters. The project was funded in part by Graduate Research fund 3852 (1994–95) of the University of Kansas to H.M.Alexander.

Received 2 February 1998revision accepted 1 December 1998