UC Davis

Net primary productivity and carbon sequestration are dependent on an adequate source of nitrogen (N) in terrestrial ecosystems. Biological nitrogen fixation, the microbially-mediated conversion of atmospheric N2 to biochemically reactive ammonia (NH3) serves as an important source of new N to many natural ecosystems. Despite its crucial role in global biogeochemistry, controls over the rate of this microbially mediated process remain poorly understood. This is particularly true in the context of land use (LU) alteration, a major driver of climate change and ecological disruption. In the Amazon Basin, LU change from primary tropical forest to agricultural operations, particularly cattle pasture, has been prevalent and ongoing for decades. To better understand shifts in N inputs resulting from LU change, this dissertation investigates the activity and diversity of free-living soil diazotrophic communities across the LU dichotomy of forest and pasture in the Amazon Basin. First, to establish a broader perspective on the microbial response to LU change in the Amazon, we performed a comprehensive review of all studies investigating microbial communities or microbially mediated processes across variable LUs and successional states in the region, highlighting both generalized and repeated findings, as well as points of disagreement which may indicate regional heterogeneity or methodological biases. This meta-analysis revealed that although trends in the alpha-diversity of microbial communities vary somewhat across studies, spatial biotic homogenization in pastures compared to primary forests has been a near-universal finding among prokaryotes, fungi, and several other taxonomic or functional subgroups. Additionally, a large proportion of studies have identified soil Al3+ content and extractable acidity to be major factors shaping community structure, emphasizing the importance of tropical mineralogy. Significant shifts in microbially-driven biogeochemical cycling with LU change have been identified as well. Most studies have found that soil C stocks increase slightly with pasture conversion, simultaneous with elevated methane emissions and a decline in the community proportion and taxonomic richness of methanotrophs. However, the most drastic and consequential nutrient cycle shift has been to the N cycle. Studies have found that rates of net mineralization and nitrification decline sharply in maturing pastures, coincident with reductions in nitrous oxide emissions and inorganic N pools. Recent studies have indicated that free-living diazotrophs increase in abundance and diversity with pasture conversion, but measurements of asymbiotic nitrogen fixation (ANF) have not been made across LUs to bolster these community findings. Addressing this knowledge gap in conjunction with community profiling was a primary focus of the original research presented here.To fulfill this aim, we surveyed three primary forests and three pastures (converted in or around 1972) in the state of Rondônia, Brazil near the end of the wet season in April of 2017. At each site, we established a 100 m2 quadrat and collected four replicate soil cores (0-10 cm in depth) from seven locations each. After homogenization, fresh soil was utilized for 15N2 gas incubations to calculate soil N incorporation attributable to ANF. Additionally, preserved soil samples were used for nucleic acid extraction and analysis, quantifying a suite of physicochemical measurements, and profiling the bulk soil metabolome. Community nucleic acids were utilized for marker-gene targeted amplification to serve as a proxy of absolute abundance and functional community structure, as well as to obtain profiles of potential diazotrophs within the broader soil metagenome. From this analysis, we concluded that soil ANF is indeed stimulated (47x increase) in active cattle pastures coincident with an augmentation in soil nifH copy number (18x increase; in line with previous observations), but that the two are not directly related within LUs. Using soil physicochemical parameters (including various pools of C and N, natural isotopic abundance, soil texture, pH, P, cation exchange capacity, and enzyme metallocluster constituents including Mo and V) for variable selection in multiple linear regression, we were unable to identify variables strongly associated with nifH community augmentation from forest to pasture. Substantially lower soil NO3- concentrations provided significant, but modest value in explaining the stimulation of ANF in mature, active cattle pastures compared to primary forests. Additionally, when forests were considered separately from pastures, nifH copy number and ANF rates (which were consistently near zero) were poorly explained by physicochemical parameters. Together these findings indicate that factors such as high inorganic N concentration as well as alternative, more productive N input pathways (i.e., canopy lichens, nodules, or the surface litter layer), which were not measured in this study, could largely suppress the activity of potential diazotrophic bacteria in primary forests. Within pasture soils, however, we found that nifH copy number was primarily associated with C pools. We found a modest, but significant association between pasture soil ANF rate and the ratio of low molecular weight extractable organic C to N, as well as the fraction of total N in the dissolved organic form, suggesting that this energy-intensive reaction is stimulated by a limited availability of N combined with sufficient fuel in the form of low molecular weight C compounds. Additionally, using a multigene approach to profile soil diazotrophs within the broader microbial community revealed that the genetic potential for asymbiotic diazotrophy is one of the most (if not the most) enriched soil microbial functions accompanying pasture conversion. This realization speaks to the immense influence that large-scale LU change can have on microbial communities, and the strong pressure for N replenishment in grazed (but unfertilized) pastures of the Amazon Basin. After observing that the absolute abundance of potential diazotrophs (based on DNA copy number) did not scale with ANF measurements in either LU, we aimed to further explore any meaningful relationships between diazotrophic communities and this crucial biogeochemical process they mediate. A previous analysis of the potential diazotroph community in forest and pasture soils of Rondônia found significant shifts in its structure. However, given the phylogenetic and trophic breadth of potential diazotrophs (i.e., microorganisms bearing nitrogenase-encoding genes, but not necessarily contributing to ANF), it is not clear whether these shifts have occurred independent of the larger soil prokaryotic community. By comparing potential diazotroph community structure (using DNA-based amplification of the nitrogenase marker gene, nifH) with that of all prokaryotes (by amplifying the 16S rRNA gene), we found that potential diazotroph community alpha-diversity was significantly higher in pasture soils, while the overall prokaryotic community did not reflect an increase. However, both profiles reflected a similar degree of community compositional dissimilarity with respect to LU change. Additionally, both communities exhibited significantly lower groupwise compositional dispersion in pasture compared to forest soils, agreeing with several earlier studies observing biotic homogenization of soil microbial communities with LU change. The compositional dissimilarities of both communities were associated with a similar subset of physicochemical conditions including clay content, pH, and total sulfur content, as well as the proportion of nitrogen in inorganic forms. We further investigated how these potential diazotrophic communities relate to the subset of taxa that are both metabolically active, and actively transcribing the nitrogenase enzyme (via RNA-based sequencing of nifH), to determine if the latter community exhibits a community structural response to LU change differing from that of DNA-based communities, and whether active communities may better explain ANF activity. In contrast to our expectations, we found that active diazotrophs did not reflect an increase in alpha-diversity with LU change and exhibited only a modest compositional response with no environmental correlates. Surprisingly, active, RNA-based communities across LUs were more similar to each other than they were to their DNA-based counterparts, showing an opposite trend in community dispersion with respect to LU. This increase in dispersion was related to ANF rates; we identified 17 of 882 taxa whose relative abundance scaled linearly and significantly (Pearson r = 0.88) with ANF. Of these, two OTUs, annotated as Bradyrhizobium and Enterobacteriaceae, reflected strong correlations with ANF on individual bases and were found to have the third and twentieth highest relative abundances among active pasture taxa, respectively. While paired potential and active diazotrophic communities (i.e., derived from the same sample) reflected a high degree of compositional dissimilarity from each other overall, by far the most drastic shift observed was the more than 100x enrichment of the photosynthetic cyanobacterial family Aphanizomenonaceae in active diazotroph profiles, irrespective of LU. Therefore, soil surface diazotrophs may also play an important role in providing N to pastures and forests alike, suggesting further work is needed to capture measurements of ANF activity under lighted conditions.