Successional dynamics of VAMfungal communities

We observed a significant change in the composition of the VAM fungal communities across the Cedar Creek chronosequence. Relative spore densities of some VAM fungi were clearly more abundant in the youngest sites while other species were more abundant in the older sites. This contrasts with the findings of Benjamin et al. (1989) who observed no change in VAM fungal species throughout an Illinois prairie-forest chronosequence. A major difference between our two studies is that there was little edaphic variation in the prairie-forest gradient studied by Benjamin et al. (1989) while there was a significant accrual of organic C and N across the Cedar Creek chronosequence. This suggests that edaphic factors may be very important in structuring VAM fungal communitles.

Spore abundance of late successional VAM fungi tended to be positively related to soil C and N while spore abundance of early successional fungi tended to be inversely related to these parameters. It should be noted that closely related species of fungi often responded similarly to soil factors. For example, members of the genus Scutellospora were consistently inversely related to soil pH, soil C and soil N. These fungi are common sand dune inhabitants (Koske and Walker 1985), and thus may be expected to do well in early successional soils at Cedar Creek. Abundance of all four Acaulospora species were also inversely related to soil pH. High densities of Acaulospora spores have been reported in acid environments (Morton 1986; Porter et al. 1987) suggesting that members of this genus tend to be well adapted to low pH soils.

An alternative interpretation of our results might be that VAM fungi were not early or late successional species, but were specialists on different soil series. In this study, field age was somewhat confounded with soil series because the youngest sites were aggregated in one part of Cedar Creek and occurred on Nymore sands, while the older fields and forest sites were interspersed throughout the natural history area and occurred on Sartell, Nymore and Zimmerman sands. This interpretation seems less likely than the successional interpretation however because the 53 year old field and forest sites occurred on the Nymore sands, yet they contained VAM fungal communities which were more similar to the 35, 46, and 60 year old fields than to the younger Nymore sand sites. Furthermore, the 19 and 30 year old fields occurred on Sartell sand but had VAM fungal communities more similar to the 1-13 year old fields on Nymore sand, than to the other sites on Sartell sand.

VAM fungi do not tend to be host specific. For example, Glomus mosseae has been shown to colonize roots of twenty different plant species belonging to twelve different families (Mosse 1973). However, there is growing evidence that VAM fungal species differ in their ability to proliferate in rhizospheres of different crop plants (Schenck and Kinlock 1980; McGraw and Hendrix 1984; Johnson et al. 1991). It is possible that VAM fungal species differ in their ability to proliferate in early versus late successional plant species, and consequently, Iate successional VAM fungi must await the arrival of late successional host plants before they can become abundant in the VAM fungal community. The present study illustrates that the relative cover of non-host plant taxa is negatively correlated with total counts of VAM fungal spores. Whether or not the composition of the host community influences the composition of the VAM fungal community at Cedar Creek remains to be experimentally addressed.

Successional dynamics of VAM density

We suggest that there is a close interrelationship between successional dynamics of soil properties, plant productivity and VAM density as measured by infectivity and total spore counts. Water soluble C was the only factor which correlated to both infectivity and total spore count in the 14 field sites. VAM fungi are obligate symbionts and thus it is not surprising that their densities were highly correlated with H2O soluble C since it is a measurement of the most labile C fraction, including exudates of living roots.

The finding that infectivity was positively correlated with H2O soluble C and negatively correlated with soil P has a physiological basis and might be expected from the results of several greenhouse studies. Colonization of roots by VAM fungi have been shown to increase with increasing quantities of soluble carbohydrates in roots and root exudates and decrease with increasing concentrations of tissue P (e.g. Schwab et al. 1983; Same et al. 1983; Thompson et al. 1986). It is thought that this P induced reduction in VAM colonization is associated with a membrane mediated decrease in root exudation (Ratnayake et al. 1978; Graham et al. 1981).

The negative correlation between total P and infectivity became insignificant when pH was held constant. This suggests that soil pH and P interact in their affects on the VAM infection process. It is well known that P solubility in the soil is highly pH dependent (Mengel and Kirkby 1982). Riley and Barber (1971) showed that P uptake by soybean decreased with increasing pH. Could plants growing in high pH soils form more VAM because they take up less P, and consequently exude more carbohydrate from their roots, compared to plants grown

in low pH soils? Unlike infectivity, total spore density was not significantly correlated with soil pH, suggesting that VAM colonization of roots may be more sensitive to pH than sporulation of VAM fungi. Alternatively, the highly significant correlation between infectivity and soil

pH may be an artifact of the bioassay and not a true reflection of propagule density in the soil. Such an artifact could occur if VAM colonization of the corn roots was sensitive to the pH of the soil in the Conetainers.

Plants vary greatly in the degree to which they form VAM associations, ranging from non-host species, which never form mycorrhizae, to species which always form VAM associations. Reeves et al. (1979) hypothesized that VAM fungal populations will be low and non-host plant taxa will dominate early in succession, while VAM fungal populations will be high and mycorrhizal plant taxa will dominate late in succession. Furthermore Janos (1980) hypothesized that following disturbance in temperate regions. VAM fungal populations increase and then decrease as ectomycorrhizal fungi come to predominate with the onset of a forest community. Results from the Cedar Creek chronosequence support these hypotheses.

Both infectivity and total spore count generally increased with age in the 14 field sites but then decreased with successional rank in the 3 forest sites. Benjamin et al. (1989) also found infectivity and spore counts decreased as tree basal area increased in a sand prairie to oak hickory forest succession. We suggest that the reduced VAM densities in the savanna, oak forest and hardwood forest resulted from a shift in the plant community dominants from herbaceous, primarily VAM hosts, to woody, primarily ectomycorrhizal hosts.

Our two indices of VAM density. infectivity and total spore count, were not significantly correlated. There are several reasons to expect this result, and it agrees with the findings of others (Powell 1977; Hayman and Stovlod 1979; Abbott and Robson 1982; Scheltema et al. 1987). Spore counts assess only one type of propagule, while infectivity indirectly measures all types: spores, hyphae, and VAM roots. Consequently spore counts probably do not measure total VAM density as accurately as infectivity. Furtherrnore, spore formation may be unrelated to the total fungal biomass (including hyphae, vesicles and arbuscules) since different VAM fungal species do not all sporulate to the same degree. For example, Glomus aggregatum produce copious tiny spores in loose sporocarps, while G. tenuis rarely, or never, form soil-borne spores (Hall 1977). Finally, dead spores may accumulate in the soil and are often impossible to distinguish from viable spores. Consequently, the dominance of G. aggregatum in the spore community may or may not reflect its actual abundance within plant roots.

During old field succession at Cedar Creek, changes in plant community composition, primary production, and nutrient accrual occur in relatively predictable patterns. Previous research has demonstrated that a close interrelationship exists between the pattern of plant and

microbial biomass owing to the reciprocal nature of C and N cycles (Zak et al. 1990). The results we present here suggest a concomitant interrelationship exists between soil properties, plant productivity and VAM density.