Forest soils in the warm-humid tropics significantly contribute to the regional greenhouse gas (GHG) budgets. However, spatial heterogeneity of GHG fluxes is often overlooked. Here, we present a study of N₂O and CH₄ fluxes over 1.5 years, along a topographic gradient in a rainforest catchment in Xishuangbanna, SW China. From the upper hillslope to the foot of the hillslope, and further to the flat groundwater discharge zone, we observed a decrease of N₂O emission associated with an increase of soil water-filled-pore-space (WFPS), which we tentatively attribute to more complete denitrification to N₂ at larger WFPS. In the well-drained soils on the hillslope, denitrification at anaerobic microsites or under transient water-saturation was the potential N₂O source. Negative CH₄ fluxes across the catchment indicated a net soil CH₄ sink. As the oxidation of atmospheric CH₄ is diffusion-limited, soil CH₄ consumption rates were negatively related to WFPS, reflecting the topographic control. Our observations also suggest that during dry seasons N₂O emission was significantly dampened (<10 μg N₂O-N m^-2 h^-1) and CH₄ uptake was strongly enhanced (83 μg CH₄-C m^-2 h^-1) relative to wet seasons (17 μg N₂O-N m^-2 h^-1 and 56 μg CH₄-C m^-2 h^-1). In a post-drought period, several rain episodes induced exceptionally high N₂O emissions (450 μg N₂O-N m^-2 h^-1) in the groundwater discharge zone, likely driven by flushing of labile organic carbon accumulated during drought. Considering the global warming potential associated with both GHGs, we found that N₂O emissions largely offset the C sink contributed by CH₄ uptake in soils (more significant in the groundwater discharge zone). Our study illustrates important topographic controls on N₂O and CH₄ fluxes in forest soils. With projected climate change in the tropics, weather extremes may interact with these controls in regulating forest GHG fluxes, which should be accounted for in future studies.

Clarifying the response of soil microbial communities to vegetation restoration is essential to comprehend biogeochemical processes and ensure the long-term viability of forest development. To assess the variations in soil microbial communities throughout the growth of Pinus armandii plantations in the karst region, we utilized the “space instead of time” approach and selected four P. armandii stands with ages ranging from 10 to 47 years, along with a grassland control. The microbial community structure was determined by conducting Illumina sequencing of the 16 S rRNA gene and the ITS gene, respectively. The results demonstrated that afforestation with P. armandii significantly influenced soil microbial communities, as indicated by notable differences in bacterial and fungal composition and diversity between the plantations and the control. However, soil microbe diversity did not display significant variation across stand ages. Moreover, the bacterial community exhibited higher responsiveness to age gradients compared to the fungal community. Soil physicochemical factors play a critical role in elucidating microbial diversity and community composition variations during restoration processes. TN, AN, TP, AP, SOC, AK, and pH were the most significant influencing factors for the composition of bacterial community, while TC, SOC, pH, and TC_a were the most significant influencing factors for the composition of fungal community. Our findings indicate substantial changes in soil bacterial and fungal communities across successive stages of development. Additionally, the changes in dominant bacteria and fungi characteristics across the age gradient were primarily attributed to variations in the prevailing soil conditions and chemical factors.

Differences in traits between lianas and trees in tropical forests have been studied extensively; however, few have compared the ecological strategies of lianas from different habitats. Here, we measured 25 leaf and stem traits concerning leaf anatomy, morphology, physiology and stem hydraulics for 17 liana species from a tropical seasonal rainforest and for 19 liana species from a valley savanna in south-west China. We found that savanna lianas had higher vessel density, wood density and lower hydraulically weighted vessel diameter and theoretical hydraulic conductivity than tropical seasonal rainforest lianas. Compared with tropical seasonal rainforest lianas, savanna lianas also showed higher leaf dry matter content, carbon isotope composition (δ¹³C), photosynthetic water use efficiency, ratio of nitrogen to phosphorus, photosynthetic phosphorus use efficiency and lower leaf size, stomatal conductance and nitrogen, phosphorus and potassium concentrations. Interestingly, no differences in light-saturated photosynthetic rate were found between savanna and tropical seasonal rainforest lianas either on mass or area basis. This is probably due to the higher water and nutrient use efficiencies of savanna lianas. A principal component analysis revealed that savanna and tropical seasonal rainforest lianas were significantly separated along the first axis, which was strongly associated with acquisitive or conservative resource use strategy. Leaf and stem functional traits were coordinated across lianas, but the coordination or trade-off was stronger in the savanna than in the tropical seasonal rainforest. In conclusion, a relatively conservative (slow) strategy concerning water and nutrient use may benefit the savanna lianas, while higher nutrient and water use efficiencies allow them to maintain similar photosynthesis as tropical seasonal rainforest species. Our results clearly showed divergences in functional traits between lianas from savanna and tropical seasonal rainforest, suggesting that enhanced water and nutrient use efficiencies might contribute to the distribution of lianas in savanna ecosystems.

Patterns in functional trait variation associated with the ecological strategies of lianas and trees in subtropical montane forests remain poorly understood due to lack of trait comparisons. Here, we filled this gap by investigating trait divergence between lianas and trees with different leaf habits for 13 traits of 33 species (7 deciduous and 4 evergreen liana species, and 10 deciduous and 12 evergreen tree species) in a subtropical montane forest in southwestern China. We found that lianas had significantly larger stem xylem vessel diameter, higher theoretical hydraulic conductivity, specific leaf area, leaf nitrogen, and phosphorus concentrations, but lower leaf dry-matter content and N/P ratio than sympatric trees, indicating that lianas employ an acquisitive ecological strategy, with a more efficient stem hydraulic system and more productive leaves. In contrast to trees, lianas exhibited a larger variation in vessel diameter, with a few wide and many narrow vessels in the xylem, indicating a greater vessel dimorphism. Growth form explained 48.7% of the total trait variation, while leaf habit only explained 3.8% of trait variation, without significant interaction between growth form and leaf habit. In addition, significant stem–leaf trait relationships were only found in trees, but not in lianas, indicating decoupling of stem hydraulic function and leaf traits in subtropical lianas. These results suggest that subtropical montane lianas and trees differ strikingly in stem and leaf functional traits. Further studies are warranted to strengthen our understanding of the mechanisms underlying the strong divergence in ecological strategies between lianas and trees in subtropical forest ecosystems.

Rocky desertification (RD), a natural and human-induced process of land degradation in karst areas, has become the primary ecological disaster and one of the obstacles to sustainable ecological development in southwest China. Nevertheless, the variation of soil physical and chemical properties, bacterial and fungal communities, and their relationships in RD forests remains limited. Therefore, soil samples were collected from forests under four degrees of RD (NRD, non-RD; LRD, light RD; MRD, moderate RD; and SRD, severe RD) and subjected to high-throughput sequencing of 16S rRNA and ITS1 genes. The results showed a significant reduction in bacterial richness and diversity, while fungal richness and diversity decreased markedly and then showed a balanced trend with the increase in RD degree, indicating that bacteria and fungi did not present the same dynamics in response to the process of RD. The bacterial communities were dominated by Proteobacteria, Actinobacteria, Acidobacteria, and Chloroflexi, while the fungal communities were dominated by Basidiomycota, Ascomycota, and Mortierellomycota. The PCoA and NMDS demonstrated significant differences in microbial communities in study sites, among which the fungal communities in non-RD forest and LRD forest clustered together, suggesting that fungal communities were more stable than bacteria in RD forest. The db-RDA, Mantel test, and random forest model confirmed the important role of soil BD, pH, SOC, AN, and AP in driving microbial diversity and communities. The IndVal analysis suggested that Chloroflexi, Patescibacteria, Atheliales, and Cantharellales with high indicator values were identified as potential bio-indicators for RD forests. This study could not only improve our understanding of bacterial and fungal community dynamics across RD gradients, but also could provide useful information for the further use of microorganisms as indicators to reflect the environmental changes and ecosystem status during forest RD.