Discussion
Understanding fungal response to climate warming is important for revealing their roles in nutrient cycling processes and evaluating the ecological consequences of climate warming on high-latitude tundra. In this study, we hypothesized that climate warming would affect fungal community composition and potential interactions. Our hypotheses were partially upheld: fungal functional gene composition and potential interactions among genes/species were significantly altered, whereas fungal community composition was unchanged (Table 1 & Fig. 2). Different responses of functional gene composition and community composition were similar to the findings of a previous warming study in a tall-grass prairie ecosystem (Cheng et al., 2017), suggesting that GeoChip is more sensitive for detecting subtle changes in functional genes. Due to horizontal gene transfer among microbial species, species identities detected in GeoChip do not reliably represent phylogenetic species in environmental samples. However, if both sequencing and GeoChip data point to the same species, the chance of disconnection between sequencing and GeoChip is slim. For example, we detected both functional genes and OTUs belonging toHypocrea in GeoChip data and amplicon sequencing data. Similar to the overall changes of fungal composition change, we found significant increases in gene abundances from Hypocrea (e.g., the genes encoding endochitinase, exopolygalacturonase, and xylanase) (P< 0.05), whereas no changes of OTUs from this genus. Nonetheless, the lack of detectable changes in fungal community composition should be interpreted with caution because 28S rRNA gene has a relatively low taxonomic resolution (Brown, Rigdon-Huss, & Jumpponen, 2014), which might fail to detect subtle changes in fine taxonomic levels (Hultman et al., 2015).
Ascomycota was the dominant fungal phylum in our study and other studies conducted in sedge-occupied tussock tundra (Christiansen et al., 2016; Deslippe, Hartmann, Simard, & Mohn, 2012; Timling, Walker, Nusbaum, Lennon, & Taylor, 2014). However, this phylum was less dominant in shrub-occupied tundra (Wallenstein, McMahon, & Schimel, 2007), indicating that vegetation pattern is probably closely related to fungal community composition. Climate warming has caused vegetation shifts in the tundra, with diminishing sedges and bryophytes but expanding shrubs and trees (Fraser, Lantz, Olthof, Kokelj, & Sims, 2014; Sturm et al., 2001). Considering the potential relationship between fungi and plants, it is likely that dominant fungi in high-latitude tundra will shift accompanying the shrub expansion.
The increase in fungal C degradation capacities, based on higher relative abundances of functional genes encoding invertase, xylose reductase, and vanillin dehydrogenase (Fig. 1), might imply that fungi-mediated C degradation was accelerated under warming. Those genes are associated with the degradation of complex plant-derived saccharides (Culleton, Mckie, & de Vries, 2013), probably resulting from higher plant productivity (Table S1). Moreover, higher vanillin dehydrogenase gene abundance might result from the “priming effect,” i.e., labile C input can stimulate recalcitrant C degradation (de Graaff, Classen, Castro, & Schadt, 2010; Mau, Dijkstra, Schwartz, Koch, & Hungate, 2018). Warming-induced increases of fungal C degradation capacities were also found in other ecosystems including grasslands (Cheng et al., 2017), subtropical freshwater wetlands (Wang et al., 2012), deserts (Weber et al., 2011), showing a consistent response pattern of fungal communities across ecosystems. For example, in the biological soil crusts of a desert, soil warming increased the abundance of fungal functional gene cbhI , which encodes cellobiohydrolase for cellulose degradation (Weber et al., 2011).
Higher average path distance but lower average clustering coefficient of the warmed fMEN (Table 3) suggested that the complexity of potential interactions decreased under warming. As the percentage of negative links could represent competition within community members (Fuhrman, 2009), a lower percentage of negative links in warmed fMEN concurred with higher soil nutrient availability that alleviates resource competition (Banerjee et al., 2016). Higher nutrient availability appears to stimulate C degradation, as almost all key genes, the most important members in maintaining network structure and strongly influence community stability and functions (Banerjee et al., 2018; Shi et al., 2016), in warmed fMEN were associated with C degradation (Table S4). However, it is important to bear in mind a caveat that most networks cannot distinguish true ecological interactions based on positive or negative correlations between nodes since it remains largely intractable to analyze in situ microbial interaction experimentally in most communities of natural environments (Faust & Raes, 2012). When interpreting networks in ecological terms, topological properties (e.g., average node connectivity, average path length, clustering coefficient, and modularity), which reflect whole-network changes, could be more reliable.
Legacy effects of winter warming included increases in thaw depth, soil moisture, and GPP in the growing season (Table S1). The increase of thaw depth, a sign of permafrost degradation, was also observed in previous experimental warming studies using snow fences (Hinkel & Hurd, 2006; Nowinski, Taneva, Trumbore, & Welker, 2010). Since we excluded snow addition effects on soil hydrological conditions by removing snow before melting, the most likely cause of soil moisture increase was ice-wedge melting in permafrost (Liljedahl et al., 2016). Soil temperature, soil moisture, thaw depth, and GPP imposed the strongest influence on fungal functional gene composition (Fig. 3), suggesting that they were very important in affecting fungal functional capacities. Oxygen availability could be affected by soil hydrology and temperature, which is shown to change species diversity, ecological functions, and survival of most aerobic fungi (Wang et al., 2012; Zak & Kling, 2006).
Permafrost thawing increases the potential of old C degradation, leading to stronger heterotrophic respiration and CO2 emission (Nowinski et al., 2010; Schuur et al., 2009). Consistently, we found a significant warming-induced increase in R eco(Table S1 & Fig. 4), primarily resulting from the increase in winterR eco that mainly represents heterotrophic respiration (by 103.2%, P< 0.05, Table S1). Our winter R eco data were derived from a site-specific model that assumes winter R ecoincreases with soil temperature exponentially, which is commonly used when simulating the temperature dependence of heterotrophic soil respiration. The model fitted in situ measured data well (R 2 = 0.70, P < 0.001), which was consistent with observations elsewhere (Tuomi, Vanhala, Karhu, Fritze, & Liski, 2008).
Increased R eco shared positive regressions with fungal functional genes for C degradation (Fig. 4), suggesting that fungal functional capacities in C degradation might be very important in mediating R eco, and thus the C stability of tundra ecosystems. According to the central dogma, DNA has to be transcribed into RNA and translated into protein before displaying enzymatic activity, which is the missing link in our study since enzyme activities were not measured. However, a large number of papers have justified the strong, positive relationship between enzyme activities and abundances of their encoding genes (Blackwood, Waldrop, Zak, & Sinsabaugh, 2007; Fan, Li, Wakelin, Yu, & Liang, 2012; Trivedi et al., 2016). Notably, the positive correlations between R eco and fungal functional genes should be interpreted with caution since GPP was also increased by warming (Table S1). Therefore, the increase of R eco could be explained in terms of the increased photosynthesis and corresponding increased respiration of new photosynthate. Additionally, higher bacterial C degradation capacities under warming conditions at this site have been previously documented (Xue et al., 2016), which could also account for the R eco increase. Therefore, future studies that distinguish plant respiration and heterotrophic soil respiration are needed.