In forest ecosystems, fire alters the soil microbial community composition and activity in the form of direct heat-induced microbial mortality [1
]. Furthermore, loss of forest by fire and grass invasion may alter nutrient cycling and quality of organic matter [3
], which can indirectly affect the soil microbial community. Soil bacterial and fungal biomass is lower in fire-induced grassland areas than unaffected forest areas [4
]. Post-fire tree replanting can accelerate the recovery of forest function in an ecosystem [5
], thereby leading to greater recovery of the soil microbial community.
In Taiwan, grasslands are generally distributed above the tree line, at 3100 to 3400 m above sea level (asl). Approximately 50 years ago, a wildfire led to a vegetation shift from forest to grassland at Tatajia, the saddle (2600 to 2800 m asl) of Yushan, one of the highest mountains (3950 m) in East Asia. This region provides a unique environment for observing long-term ecological changes in vegetation succession in a subalpine area.
Soil microorganisms are critical for ecosystem function and contribute substantially to organic matter decomposition and maintenance of soil structure [6
]. Fungi and bacteria often occupy separate ecological niches in the soil environment and play critical roles in nutrient cycling [8
]. Microbial biomass increases quickly in the early forest succession stage and then decreases and remains at a constant level with long-term forest restoration [10
]. Hedo et al. [11
] reported that post-fire silvicultural treatment did not have a direct effect on soil microbial properties and soil enzyme activities. However, little is known about the response of microbial communities to the succession of vegetation after fire [12
]. A decade after a forest fire in a tropical rainforest, the dominant soil bacteria and actinobacteria did not differ between unburnt and burnt areas [13
]. On the other hand, with the construction of 16S rRNA gene clone libraries, the diversity of bacterial communities was significantly lower in planted or natural forest than in fire-induced grassland soil in subalpine area after 3 decades [14
]. Soil microbial biomass, activity and microbial community may increase with vegetation succession.
Analysis of phospholipid fatty acids (PLFA) is useful for determining the active soil microbial community structure [15
]. It provides quantitative data on microbial communities [16
]. As another simple, rapid and molecular technique, denaturing gradient gel electrophoresis (DGGE) can provide information on dominant microbial species such as bacteria, fungi, and target groups by the use of different primers in the same soil DNA extract [17
With the hypothesis that replanting of trees in a wildfire-induced grassland might enhance the recovery of microbial communities and function, we examined microbial activities, biomass and community structure in a subalpine forest ecosystem after a fire disturbance using PLFA and DGGE.
The planting of trees can have direct effects on soil properties [33
], mostly because of increased soil organic matter (SOM) through the supply of litter and root exudates. In our study of a Pinus
plantation in central Taiwan (2600–2800 m asl) after a fire event, the trend we found in soil SOM, total N, microbial activities and microbial community structure may be related to the vegetation and recovery history.
The result of the PCA carried out with soil biochemical parameters clearly indicated a differentiation of chemical properties and soil enzyme activities in soils under different types of vegetation (Figure 1
). Soil pH was more acidic in the mature forest than in the re-planted forest and grassland in our study area, which may be attributed to accumulation of more organic acids under the older forest [35
]. We found the soil organic C content increased from the grassland to the mature forest in our study site. SOM is highly correlated with the structure of the microbial community of forest soils because of the quantity and quality of organic matter input [36
]. Soil microbial biomass is significantly decreased after fire events in coniferous forests [38
]. Furthermore, microbial biomass in forest soils takes decades to recover to the original state once it has been depleted [40
]. In our previous study at this site, both levels of microbial biomass Cmic
and ratios of fungal to bacterial biomass were higher in the mature forest than the grassland soil [4
], but after 15 years of succession at this study site, microbial biomass Cmic
did not differ among the vegetation zones. Our current study suggests that the establishment of a young forest, the Pinus
plantation, caused the change in soil Cmic
ratios. The soil Cmic
ratio may reflect the potential for soil organic mineralization after fresh input of organic materials [41
]. The grassland soil in our study site showed the greatest potential for mineralization, with a high Cmic
ratio. Furthermore, the re-planted forest gradually provided more recalcitrant SOM into soil and decreased the Cmic
ratio. The re-planted forest still had a higher Cmic
ratio than the mature forest, which may indicate that the re-planted forest is still under succession.
Recovery of microbial biomass in burned forest–grassland ecosystems is linked to plant community recovery, providing an important interaction between plants and soil microbes [42
]. In general, enzyme activities in soil are closely related to the content of organic matter [43
] and Cmic
]. High SOM usually sustains high Cmic
and enzyme activities [45
]. Xylanase is produced and released mostly by fungi in the forest environment [46
]. Ali et al. [47
] showed that in coniferous tree species such as Pinus taeda
L., ectomycorrhizal fungi produce extracellular phosphatase to improve phosphorus uptake. In our study, the mature forest soil had higher fungal biomass than other soils and provided more xylanase and phosphatase (Table 3
). In addition, fungal biomass, particularly of vesicular arbuscular mycorrhizal (VAM) fungi, is a good predictor of phosphatase in forest soil [48
] which coincides with our results, with high fungal biomass in the mature forest soil (Table 1
and Table 3
). On the other hand, acid phosphatase activity decreased in soil with severe wildfires [49
], which is consistent with our results in the grassland and re-planted forest.
Although soil microbial biomass (both Cmic
) was similar among the three types of vegetation, urease activity was significantly higher in the grassland than in the re-planted and mature forest. Increased communities of gramineous and herbaceous species of grasses more likely provide higher inputs of labile C and N than do perennial forest species. Scott and Binkley [50
] showed that the quality of substrate (low lignin to N ratios) provided by existing plants can directly influence N turnover rates.
Our PCA of PLFA revealed different microbial communities in soils under different vegetation (Figure 2
). Other studies have also shown that associated changes in vegetation rather than direct changes in soil properties may lead to shifts in the soil microbial community structure [51
We found no differences in bacterial biomarker features between grassland and re-planted forest soil. However, fungal and VAM levels were higher in the re-planted forest than the grassland soil. Fungal communities play a dominant role in fresh organic matter decomposition [53
]. Djukic et al. [54
] showed similar results with higher amounts of fungi in forest soils than in grassland soils.
Fungal biomass contributes considerably to total biomass in coniferous forest soils [55
]. By comparison, high availability of labile substrate (root exudates), for which bacteria are effective competitors, may constrain fungal biomass in grassland sites [54
]. Allison et al. [56
] found that C inputs in grasslands were mainly derived from root exudates: the rapidly metabolized C limits the development of fungi because fungi use more recalcitrant sources of C.
VAM fungi play an important role in improving plant growth and uptake of nutrients, especially phosphorus [57
]. PLFA 16:1ω5 is present in VAM fungi [58
] and in G− bacteria [59
]. Jonasson et al. [60
] indicated that 20% to 30% of total organic P is immobilized by microbes, and Thoms et al. [61
] showed a high association of soil P content and PLFA 16:1ω5. The alterations in the 16:1ω5 pattern we found support results from previous studies showing higher extractable P in mature forest than other vegetation soils (Table 1
The ratio of fungi to bacteria was increased in the re-planted forest soil (Table 3
). The increased ratio of fungi to bacteria can be an important indicator of the amount and composition of litter that enters soils because fungi are the dominant decomposers of plant cell-wall polymers in the litter [62
]. In this study, the switch from grassland to forest ecosystems was critical for the source of SOM because these ecosystems differ in the quantity and quality of dead plant biomass input. Although bacterial biomass was similar in the grassland and re-planted forest soils, fungal biomass increases from grassland to re-planted and mature forests. Our previous study found that both soil ergosterol and the respiration rate of fungi to bacteria decreased from the forest to grassland soil [4
]. In this study, the ratio of fungi to bacteria PLFA patterns was similar in both the mature forest and re-planted forest soil, which suggests the recovery of the microbial community structure by reforestation after wildfire events.
The composition of the soil microbial community was mainly controlled by the pH and ratio of C to N of the substrate [63
]. Bacterial biomass has been reported to increase in soils with a lower C to N ratio [64
]. Ingham et al. [65
] stated that grasslands more strongly dominated by bacteria than a coniferous forest, which supports our results. In grasslands, fast-growing plant species, especially those with highly branched fine root systems, supply large quantities of exudates [66
], which are favored by bacteria.
A low ratio of G+/G− bacteria, found in grassland soils, may be due to better growth of G− bacteria under substrate-rich conditions [27
]. By contrast, with progressive succession from a grassland to a forest system, the detritus food webs become more complex and recalcitrant. Slow-growing specialists, such as G+ bacteria, are more competitive than G− bacteria in resource-limited areas [67
] because of effective cell metabolism and effective use of recalcitrant substances such as cellulose and lignin in a coniferous environment [68
]. This situation is similar to the decrease in the G+/G− ratio for the soils in a bamboo-invaded cedar forest, as herbaceous litter favors the growth of G− bacteria [70
Our dendrograms of soil bacteria and fungi from PCR-DGGE showed primary differences in the microbial communities related to the succession of vegetation (Figure 3
and Figure 4
). By using 16S rRNA gene clone libraries for this same study site, Lin et al. [14
] found that the soil bacterial communities differed between the grassland and the mature forest 30 years after the fire. These findings support our results. The Pinus
plantation can reform soil microbial communities. Although the microbial community structure was similar in grassland and re-planted forest soil, our DGGE findings agreed with the PCA results of PLFA, showing a distinct composition of bacterial and fungal communities with forest succession.