Our first hypothesis was that prescribed burning in the spring would have no significant effect on soil C, N, mineral soil bulk density, C:N ratios, pH, and δ13C/15N isotopic signatures of soil samples collected 3 yr post-burn, as compared to unburned control units. We expected some reduction in surface fuels (litter and CWD) in the spring burn units. The fall burn units were expected to have lower levels of soil C and N, higher mineral soil bulk density, lower C:N ratios, higher pH, unchanged δ13C/15N isotopic signatures, and reduced levels of surface fuels. The treatment effects on soil properties and fuel levels were as expected with the exception of mineral soil bulk density (Table 1). Bulk density might be anticipated to increase with fire severity as a function of increased consumption of organic soil components; however, here the bulk density was lowest in the spring burn units and was not correlated with total C (Tables 1 and 2). This pattern is difficult to explain until analyzed spatially: the units with the highest soil bulk densities were at the lower (eastern) end of the project area and concentrated adjacent to Highway 62. An artifact of random treatment assignment was that all of the early spring burn treatment units were placed in the western (least dense) end of the project area (Figure 1), and half of the fall burn treatments (as well as several control units) were located in the densest end of the geographic gradient. We observed a similar spatial pattern with soil N and δ13C depletion. This gradient helped us to distinguish the effects of the burn treatments from the effects of the soil chemistry.
Our second hypothesis was that burning in the spring would not significantly change mycorrhizal fruiting patterns, but we expected a shift in fruiting patterns in the fall burned units as compared to controls. We expected a decline overall in fungal diversity in the fall burned units compared to unburned control or spring burned units. The fall burned units did not differ significantly from controls in either numbers of collections or numbers of species, but did differ from spring burn units (Table 1).
We identified guilds of indicator mycorrhizal fungal species that co-occurred under similar sets of soil attributes. This pattern was more closely correlated with soil C:N ratios than burn treatment (Figures 3 and 4). While all of the units with above-mean C:N ratios were either control or early spring burn treatments, three of the units with below-mean C:N ratios were unburned controls that produced fungal fruiting patterns similar to the fall burned units (units N, S, and U).
The fall ordination also suggested a transitional group of treatment units: A (early spring burn), E (late spring burn), H (fall burn), K (late spring burn), and O (late spring burn). These units tended to group with the fall burn units in habitat attribute space due to their moderate to low fuel levels, but none of them produced the low C:N indicators Boletus chrysenteron or B. zelleri, and all three produced the high C:N indicators, Suillus punctatipes and S. tomentosus.
Although mineral soil bulk density was negatively correlated with the C:N ratio (Table 2), several units that had above-mean bulk densities also had above-mean C:N ratios and produced high C:N (control units F and P) or intermediate fungal guilds (unit O). Conversely, units B, N, U, and X had below-mean bulk densities and C:N ratios, and produced the low C:N fungal guilds, indicating that C:N ratios were more consistently correlated with fruiting patterns than bulk density.
Likewise, coarse woody debris levels were positively correlated with C:N ratios, and all of the high C:N fungal guild-producing units had above-mean CWD levels. However, four of the units producing the low C:N fungal guild (units N, S, T, and U) had above-mean CWD levels, and three of them (N, S, and U) were unburned control units and also had above-mean FWD and litter levels. Again, fungal fruiting patterns correlate more closely with C:N ratios than fuel levels or prescribed burn timing.
The suites of attributes indicated by these groupings are consistent with what we might expect would separate burned from non-burned sites, however these fungi were collected over a continuum of burn severities from non-burned controls to fall burns, suggesting that the relationships are not as simple as burned versus not burned. For example, Gautieria monticola, associated with higher C:N ratios, was collected on four control units but also on five spring burn units. Conversely, Sarcosphaera coronaria, associated with lower C:N ratios, was collected on two control units as well as on nine burned units. Only Morchella angusticeps consistently fruited on burned units to the exclusion of control units.
These patterns are further supported by logistic and indicator species analysis of each species against each habitat attribute (Table 3). Indicator species analysis tended to identify more variables as significant than did logistic regression. All of the species identified in the ordination had significant correlations with habitat attributes, most frequently the C:N ratio.
Categorizing Units by Fungal Guilds
The units can be grouped into three categories based on the fungal guild indicator species they supported (Figures 5, 6): the low C:N guild in units B, J, L, M, N, Q, R, S, T, U, and X; intermediate or transitional units A, E, H, K, and O (inconsistent or without either high or low C:N indicator species); and the high C: N guild in units C, D, F, G, I, P, V, and W. Four of the intermediate units (A, E, K, and O) produced the high C:N guild in the spring (Figure 3) and the low C:N guild in the fall (Figure 4). Unit H was only sampled in the fall and did not produce any C:N ratio indicator species.
Of the late spring burn units, only unit T produced a clearly low C:N guild both spring and fall, while the other late spring burn units (E, K, and O) were intermediate. Three of the four early spring burn treatment units (units C, V, and W) produced high C:N-associated guilds, and the fourth (unit A) was intermediate. In total, more spring burn units produced the high C:N guild than did control units.
No fall burn units produced high C:N fungal guilds. However, three of the control units produced the low C:N fungal guild. All but one of the units (unit G; control) having above-mean C:N ratios produced the high C:N guild. Unit G is spatially transitional between high and low C:N units, and the apparent inconsistency between its C:N ratio and the fungal guild produced may be an artifact of the random locations from which soil cores were taken within the unit. All of the intermediate units also had above-mean C:N ratios. Only one of the low C:N guild producing units had an above-mean C:N ratio (unit J, a fall burn). The three control units (units S, N, and U) that produced the low C:N guild all had below-mean C:N ratios. The correlation between the C:N ratio (Figure 5) and fungal guilds (Figure 6) is much closer than that of burn treatment (Figure 1) and fungal guild (Figure 6), and explains the occurrence of low C:N fungal guilds in control units N, S, and U.
The seven units at the east end of the study area all had low C:N ratios and produced low C:N guilds, irrespective of burn treatment. One clue to the effect of the fall burn treatment on the units is to compare the C:N ratios from control units G and S to those of proximate fall burn units L, M, and R. The C:N ratios of control units G and S were 22.1 and 23.2, respectively, and fall burn units L, M, and R ranged from 18.6 to 19.7. If we assume that the C:N ratio of control units G and S did not change appreciably from before the burn treatments, then we can infer that the fall burn treatment itself reduced the C:N ratio by 2.4 to 4.6 in units L, M, and R. By this estimate, it is quite possible that these units were producing the low C:N fungal guild even before the burn prescriptions were applied. The contrast in C:N ratio between fall burn units B, J, and X, and their neighboring control and spring burn units is also striking (Figure 5), suggesting direct influence by the fall burn treatment.
Possible explanations for the spatial pattern of bulk density, total N, δ13C depletion, and C:N ratios include the adjacent Highway 62, historic human use, or natural causes. Isotopic patterns do not support the effect of motor vehicle traffic as a source of C or N deposition. Both petrocarbon deposition (Wilkes et al. 2000) and N fertilization (Temple et al. 2005) would tend to increase δ13C depletion, and at our site the low C:N units are less δ13C depleted.
From about 1925 to 1932, there was a park entry station and maintenance camp in the vicinity of units Q, R, S, and T (S. Mark, National Park Service, personal communication). All of these units had above-mean bulk density, and unit S (a control) had the highest bulk density of the entire project area. It is possible that this area is still responding to an intense and prolonged disturbance from 75 years ago, either from the camp itself or from related highway construction activities.
All five of the units intermediate between the high and low C:N fungal guilds were in a line between unit O and unit A (Figure 6). This line marked the transition from high C:N (to the west) to low C:N soils (to the east). Four of these five units were spring burn treatments; one (unit H) was a fall burn treatment. Unit H was the only fall burn unit that did not clearly produce a low C:N fungal guild; it had the highest C:N ratio and CWD levels of the fall burn treatment units.
Fall burn treatment units B, J, and X all had below-mean C:N ratios and produced the low C:N fungal guild, but in their cases the lower C:N ratios were due to lower levels of total C, rather than to higher levels of total N. The fall burns may have reduced total C but the difference is non-significant (P = 0.123). These units were surrounded by control and spring burn treatment units (Figure 1) that maintained higher C:N ratios (Figure 5), suggesting that the fall burn treatment changed the soil C:N ratio enough to shift mycorrhizal fungus fruiting patterns.
All fall-burned units produced low C:N guilds except unit H. The burn treatment may have reduced the C:N ratio enough to suppress fruiting of high C:N guild species, but not enough to produce the low C:N guild. Of all the spring burn units, only unit T produced the low C:N fungal guild. It was among the band of low C:N ratio units along the highway, and based on the spatial pattern of soil properties, also may have produced the low C:N fungal guild before the burn treatment was applied.
While the development of sporocarps in saprobic (wood-decaying) fungi can be very sensitive to substrate chemistry (Moore 1998), we know very little about sporocarp initiation factors in mycorrhizal fungi. Primary elements in mycorrhizal morphogenesis are thought to be available energy, temperature patterns, and moisture availability (Smith and Read 1997). Mycorrhizal fungi presumably have steady access to photosynthetically fixed C, but relative levels of organic and inorganic forms of N may be differentially influential to sporocarp morphogenesis between species.
The meaning of the C:N ratio threshold of 26 as a divide for the fruiting of some fungi and not others at this site is unknown, but may be a consequence of the varying abilities of different species to access the energy required to produce fruiting bodies under differing soil chemistry or water potential conditions. Although it is unknown whether the fruiting patterns we observed are a consequence of spatial patterns of mycorrhizal fungi across the landscape or fruiting responses to environmental conditions, it seems unlikely that mycorrhizal fungi could colonize an area where they were previously absent and initiate fruiting in the three years since the burn treatments were applied. It seems more likely that those fungi we detected were there before the treatments and what we observed is a fruiting response, rather than newly arrived fungal individuals. Conversely, however, it is easy to imagine sensitive species being extirpated by the effects of more severe fires.
With the exception of Morchella angusticeps, which responded more to the treatment itself than to the effects on soil properties, the timing and consequent severity of prescribed burn treatments influenced fungal communities only indirectly as a function of their effects on soil attributes. Burn treatments may adjust pre-existing soil chemistry that in turn influences fungal community composition. Although the differences in the number of fungal collections and species between control and fall burn treatments were not significant, it is likely that some mortality of fungal individuals occurred in the fall burn units. In no unit or treatment was mycorrhizal fungal fruiting suppressed entirely. The observed indicator species tended to be mutually exclusive, and given the short duration of time since treatments, it is more likely that the patterns observed represent sporocarp morphogenesis rather than mycorrhizal colonization. This site offers and excellent opportunity for studying the long-term effects of the prescribed burn treatments on the stand structure, species composition, and soil chemistry on mycorrhizal fruiting patterns.