Asynchronous responses of microbial CAZymes genes and the net CO2 exchange in alpine peatland following 5 years of continuous extreme drought events

The effects of extreme drought on soil biochemical properties

As shown in Fig. 1A, the range of SOC during the early, midterm and late extreme drought experiments, were 73.53–251.44 g kg⁻¹, 54.75–256.16 g kg⁻¹, and 66.37–282.16 g kg⁻¹, respectively. Concomitantly, DOC was 171.85–323.74 mg kg⁻¹, 158.15 – 504.62 mg kg⁻¹, and 166.63–418.43 mg kg⁻¹, MBC was 247.80 – 461.69 mg kg⁻¹, 257.90–450.98 mg kg⁻¹, and 264.10–458.15 mg kg⁻¹, respectively (Fig. 1B, C). The variation ranges of soil TN were 3.50–16.60 g kg⁻¹, 4.70–34.5 g kg⁻¹, and 6.70–32.50 g kg⁻¹, respectively (Fig. 1D). Similarly, the variation ranges of NH₄⁺ were 5.96–12.03 g kg⁻¹, 5.39–12.59 g kg⁻¹, and 5.74–13.03 g kg⁻¹, NO₃⁻ were 2.27–8.79 mg kg⁻¹, 5.07–9.62 mg kg⁻¹, and 5.09–9.52 mg kg⁻¹, respectively (Fig. 1E, F). The changes of SOC and NH₄⁺ with soil depth were significantly different in different extreme drought periods and decreased significantly with the increase of soil depth (Table 1, P < 0.05). DOC and TN had significant differences at different depths under early and late extreme drought, which also decreased significantly with the increase in soil depth (Table 1, P < 0.05). The content of MBC in 10–20 cm soil was significantly lower than that in 0–10 cm soil by 22.89% (P < 0.05). However, soil depth had no significant effect on NO₃^– in three extreme drought periods (Table 1). In this study, only the ED treatment significantly increased soil TN content by 17% (P < 0.05) compared with E_CK, and other soil biochemical indexes did not change significantly under extreme drought conditions (Table 1).

The effects of extreme drought on NEE

Figure 2 shows the NEE under different periods of extreme drought and Table 2 shows the difference between control and extreme drought treatments.

In early extreme drought, among all plots, the NEE variation range of E_CK was from −946.72 to −3093.78 mg C m⁻² h⁻¹, while the range of NEE change in ED treatment was −138.42 to −1688.88 mg C m⁻² h⁻¹ (Fig. 2A). During drought treatment, the average value of ED decreased 22–68% compared with that of E_CK. After rewetting, the gap gradually narrowed, and the range change was limited to 1–29% (Fig. 2A). The analysis of the variance of repeated measurements showed that during the drought treatment, both ED treatment and sampling time had significant effects on NEE, and the two factors showed significant interaction (Table 2, P < 0.05).

In midterm extreme drought, the NEE variation range of M_CK and MD were −1375.93–−2400.74 and −988.15–−1858.06 mg C m⁻² h⁻¹ respectively (Fig. 2B). During drought treatment, the average value of MD decreased 7–37% compared with that of M_CK after rewetting. The resilience of MD was found weaker than that of ED, and the carryout effect of extreme drought treatment was larger. Compared with M_CK, MD treatment decreased by 64% and 46% on the 6th and 10th day of rewetting, respectively (Fig. 2B). Even though MD had a significant effect on NEE (P < 0.05), the sampling time and their interaction were not significant (Table 2).

In late extreme drought, among all plots, the NEE range of L_CK varied from −464.55 to −2143.41 mg C m⁻² h⁻¹, and ranged from −496.28 to −1581.03 mg C m⁻² h⁻¹ in MD treatment (Fig. 2C). Compared with L_CK, the average value of LD per sampling decreased 4–32% during drought treatment, except 2% increase on September 8 (Fig. 2C). Compared with the previous two periods of drought, NEE was affected by LD in a lesser extent but not significantly (Fig. 2C). Even though sampling time revealed changes in NEE, time and treatment interactions did not appear significantly correlated (Table 2).

Figure 3 revealed that the change of NEE was significantly correlated with soil hydrothermal properties. SWC_5 and SWC_20 were significantly positively correlated with the change of NEE, with Pearson correlation coefficients (R) being 0.825 (P < 0.01) and 0.672 (P < 0.01), respectively. Soil temperature and NEE also showed a significant positive correlation. The correlation coefficients between Ts_5 with NEE were 0.554 (P < 0.05).

Variations in CAZymes genes classes and families involved in the degradation of SOM following different periods of extreme drought

Among the functional groups of CAZymes at the class level, the gene abundances of glycoside hydrolase (GHs), Glycosyl transferases (GTs), polysaccharide lyases (PLs), Carbohydrate esterases (CEs), auxiliary activities (AAs) and carbohydrate-binding modules (CBMs) are the sum of the relative abundance of each specific gene that belongs to appointed modules (Fig. 4A−F, Supplementary Fig. S1). The number of CAZymes families in GHs, GTs, PLs, CEs, AAs and CBMs were 111, 69, 18, 16, 10 and 40, respectively. Although GHs has a great advantage in quantity, GTs relative abundance was the largest, ranging from 38.40% to 40.04% (Fig. 4C), while CBMs varied only from 2.09% to 2.15% (Supplementary Fig. S1) for all the plots. At the family level, GT41 (peptide beta-N-acetylglucosaminyltransferase: EC 2.4.1.225), GT4 (sucrose synthase: EC 2.4.1.13), GT2 (cellulose synthase: EC 2.4.1.12), CE1 (acetyl xylan esterase: EC 3.1.1.72) and CE10 (arylesterase: EC 3.1.1.2) presented the top five highest relative abundance in both control and extreme drought treatments.

In all samples of this study, taxon statistics and distribution for GHs, GTs, PLs, CEs, AAs, and CBMs were shown in Supplementary Table S1. In general, the relative abundance of microorganisms for GHs and GTs at phylum, class, order, family, genus, and species levels were higher than other CAZymes families. Figure 5 showed that the regression between microbial species and functional genes was significantly correlated (P < 0.001) for α and β diversity under extreme drought in the alpine peatland.

We selected specific CAZyme families which are involved in the decomposition of SOM, including plant-derived biomass, such as starch, hemicellulose, carbohydrate esters, pectin, and lignin, fungal-derived biomass such as chitin and glucan, and bacteria-derived biomass such as peptidoglycan. In general, the recalcitrance of organic C components derived from plants follows the order: starch, which is the most easily mineralizable, hemicellulose, carbohydrate esters, pectin, and lignin, which is the most resistant. Enzymatic depolymerization of the different soil organic C components requires the synergistic action of a spectrum of enzymes as shown in Supplementary Table S2. Enzymes contributing to plant-derived biomass decomposition are classified into GHs, CEs, AAs, and PLs families. Enzymes contributing to both the fungal- and bacteria-derived biomass are classified into the GHs families.

As shown in Fig. 6, for the decomposition of different SOM, the CAZymes families involved in the enzymes had the same variation trend. They decreased under ED and MD, while increased under LD compared to control. For the decomposition of starch, the alpha-amylase and glucoamylase had a higher abundance of specific microbial CAZymes families compared to alpha-glucosidase and beta-amylase (Fig. 6A). The cellobiose dehydrogenase, acetyl xylan esterase and pectate lyase were the microbial CAZymes families with the highest abundance of cellulose, hemicellulose and pectin (Fig. 6B, C, D). The three main enzymes that participated in the decomposition of lignin were oxidase, manganese peroxidase, and laccase, with manganese peroxidase being the least abundant among them (Fig. 6E). For the decomposition of fungal-derived biomass such as chitin and glucan, chitinase was the most abundant, while endo-1,3-glucanase was the least abundant (Fig. 6F, G). From the four main enzymes involved in the bacteria-derived biomass, the higher gene number of microbial CAZymes families was related to lysozyme type G and peptidoglycan lytic transglycosylase (Fig. 6H).

Family level CAZymes genes involved in the selected different types of SOM decomposition were attributed to the bacterial community, mainly to Proteobacteria and Actinobacteria (Fig. 7). In addition to the two bacterial communities above, the other microbial communities contributed to different types of SOM decomposition genes existed differences as expected. For starch, hemicellulose, and pectin, Acidobacteria and Bacteroidetes were the main species (Fig. 7A, C, D). For cellulose and chitin, the main species contributing were Chloroflexi and Acidobacteria (Fig. 7B, E). Besides the top two communities that contributed to the genes, unclassified bacteria were also present in lignin decomposition (Fig. 7F). Deinococcus-Thermus and Acidobacteria control glucan and peptidoglycan decomposition (Fig. 7G, H).

The role of microbial CAZymes genes of SOM degradation in characterizing ecosystem CO₂ fluxes

We found that ED and MD significantly decreased NEE (Table 2, P < 0.05), while LD significantly decreased the Rm [18]. Moreover, Rm greatly varied with the specific CAZymes involved in the decomposition of starch, cellulose, hemicellulose and pectin, while NEE was only affected by soil hydro-thermal Factors (Fig. 3, Fig. 8A). In detail, Rm was positively and significantly correlated with alpha-amylase (R = 0.31, P < 0.05), beta-amylase (R = 0.26, P < 0.05), endoglucanase (R = 0.43, P < 0.05), acetyl xylan esterase (R = 0.37, P < 0.05), endo beta-1.4-xylanase (R = 0.41, P < 0.05), beta-xylosidase (R = 0.37, P < 0.05), pectin-acetylesterase (R = 0.38, P < 0.05) and pectate-lyase (R = 0.43, P < 0.05). Both NEE and Rm were negatively and significantly correlated with the soil water content. Surprisingly, soil C and N components were not significantly correlated with the CAZymes (Fig. 8A).

We assessed the correlation between NEE and GPP, Rs, Rm, and Re by the polynomial Fit. GPP is not only one of the major determinants of carbon exchange between the atmosphere and terrestrial ecosystems, but also a crucial gauge to describe plant activities and functions.a strong and significant correlation was found between NEE and GPP (n = 108, R² = 0.847, P  <  0.001, Fig. 8B). A weak but significant correlation was found between NEE and Re (n = 108, R² = 0.122, P  <  0.01, Fig. 8C). For CO₂ fluxes from soils, both the Rs and Rm were nonsignificantly correlated with NEE (Rs: n = 108, R² = 0.021, P = 0.331, and Rm: n = 108, R² = 0.022, P = 0.309, Fig. 8D, E).

Source: Ecology - nature.com