In arid and semi-arid ecosystems, dominant tree species create dramatic mosaics of plant islands of fertility and relatively barren plant interspaces that exert immense pressure on ecosystem processes and offers an ideal opportunity to explore the impact of bacterial communities. We evaluated potential links between soil respiration and N mineralization, and community co-occurrence networks and predicted gene function across three tree island microsites (i.e., beneath tree canopies, at the canopy edge, and in interspaces) in a replicated field experiment in thirty-eight woodlands sites in the Great Basin Desert in UT, USA. Additionally, we potentially intensified the effects of tree islands by creating a treatment where whole trees were shredded and the resulting fine woody debris (FWD) was deposited onto the soil surface and measured a suite of characteristics relating to the metabolic functional state of communities (i.e., microbial efficiency as the microbial quotient, C substrate quality, biomass, and dissolved organic C) to improve our interpretation of potential links between function and structure. We found that tree islands were the predominant driver, creating highly complex and connected assemblies of bacterial populations and easily discernable differences in abundance and composition of predicted functional genes. Specifically, communities directly beneath Juniperus and Pinus canopies were comprised of at least 5.2-times more connections between bacterial taxa than present in networks from interspace and edge. Using Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt) to predict the gene expression, differences in the functional potential mirrored shifts in network complexity. Tree island communities expressed 236 genes with many related to the degradation of polyaromatic or polycyclic compounds, while interspace communities expressed only 66 genes associated with the decomposition of more labile C substrates. We observed a robust tree island microsite effect on all ecosystem processes, with soil respiration rates increasing 12% and N mineralization decreasing 29% in canopy than interspace soils demonstrating that a more recalcitrant substrate from a sole C source provided high amounts of low quality of DOC and lead to a decrease in metabolic efficiency, but ultimately selected for a specific community assembly. Alternatively, communities at the edge of canopies, experiencing both tree island and interspace soil conditions generated similar levels of soil respiration as canopy soils regardless of not selecting for a highly connected community and/or specific genes suggesting that a diverse composition of labile and recalcitrant C substrates from multiple sources (e.g., trees, perennial grasses, annual grasses, and forbs) potentially elevates function by promoting the activity of a wide range of taxa. Our results identify that tree islands exert enough pressure to create distinct interactions between bacteria and alter gene expression resulting in changes in ecosystem function, but the link between structure and function is mediated through the diversity and quality of C substrates.

Multiple mechanisms control bacterial community structure but two in particular, the "legacy" of past environmental conditions, and the "priming" of bacteria to respond to seasonal or reoccurring fluctuations in resources, have the potential to determine both bacterial communities, as well as, temporal shifts in active bacterial taxa. To begin to evaluate the legacy effects of resources on microbial communities, we added four limiting resources annually (i.e., water only; C-mannitol + water; N-NH4NO3 + water; and C, N + water) and measured shifts in bacterial community composition after seven years in a cold desert ecosystem in the McMurdo Dry Valleys, Antarctica. Further, to investigate the ecological significance of priming, we conducted a series of stable isotope probing experiments (i.e., 18O-DNA SIP with 18O-labeled water, 13C-DNA SIP with 13C-labeled mannitol, 15N-DNA with 15N- NH4NO3, and a combined C and N SIP) and characterized the responding (i.e., isotopically labeled) and seed bank (i.e., unlabeled) bacterial communities. We performed each of the SIPs in soil microcosms corresponding to a single resource manipulation (e.g., 13C-labeled mannitol in C addition soils). We hypothesized that all long-term additions of nutrients and water will lead to a distinct bacterial community—a legacy effect due to the nutrient and water impoverished state of Antarctica soils. We also hypothesized that the stronger the legacy effects demonstrated by a specific community the more adapted or primed bacterial species will be to take advantage of the resource and respond. As hypothesized, resource additions created distinct bacterial legacy but to different degrees among the treatments. The extent of the resource legacy effects was greatest in the CN, intermediate in water and N, and lowest in C communities relative to the control communities, suggesting that C induced changes in communities were intensified by tandem N additions and that water alone created a more distinct legacy than water and C additions combined. Contrary to our hypothesis, the stronger the legacy effects, the less adapted or primed the community was to take advantage of resource additions. For example, the CN treatment that induced the greatest effect on bacterial communities had the lowest number of species (20.9%) in common between the responding and seed bank communities. This inverse relationship may be due to only two species (i.e., Arthrobacter, Actinobacteria and Massilia, Betaproteobacteria) really being primed to take advantage of CN and these species constituting over 75% of the seed bank community. Water, N, and C additions had similar levels of priming with 38.4%, 41.4%, and 36.3% of the responding species being present in the seed bank community, respectively. But of these three treatments, only the priming with water resulted in a unique responding community, suggesting that water, a universal bacterial resource, was enough to prime bacteria. Furthermore, water generates the most diverse responding community of all the resources with stemming from all of the fourteen dominant phyla. We did find patterns of ecological coherence among the responders, especially in the major responders (i.e., responders that increased in relative recovery by at least ten-fold). These responders were predominantly found in only three phyla (i.e., Actinobacteria, Bacteriodetes, and Gammaproteobacteria) regardless of resource addition. Alternatively minor responders (i.e., responders that increased in relative recovery at least two-fold) were contained in fourteen different phyla with specific taxa stimulated by CN (i.e., Betaproteobacteria) and N and water (i.e., Deltaproteobacteria). Further, resource additions elicited responses from 37% of bacterial species with species specializing on a specific resource (e.g., Chloroflexi) or being a generalist (e.g., Planctomycetes and Gammaproteobacteria). Our results offer the first direct links between legacy and priming effects on bacterial community composition and demonstrate that these mechanisms are not always complimentary leading to the formation of similar communities but may both be essential to maintain the high levels of bacterial diversity. Further, all resources produced elicited responders that were either specialists of generalists demonstrating that even bacteria in the extreme environment of Antarctica respond to pulses of resources.

Biocrusts and the ecosystem services they provide are becoming more susceptible to fire as exotic annual grass invasions facilitate the spread of desert wildfires. Further, precipitation patterns across the western United States are predicted to change over the next century, and have the potential to dramatically influence fire regimes and the recovery of burned biocrusts. Despite these changes to desert fire and precipitation cycles, our understanding of post-fire biocrust recovery is limited, especially regarding the first two years after fire. To investigate biocrust recovery, we created burn manipulations (i.e., unburned and burned) and tracked crust form and function over two years in one cold and one hot desert ecosystem (UT, USA). We evaluated the entire bacterial community, but focused on Cyanobacteria species that confer soil stability and N fixation capabilities to biocrusts. Specifically, we quantified shifts in biocrust bacterial community composition using target metagenomics of 16S rDNA; monitored biocrust moss and lichen cover; measured N fixation potential; and assessed soil infiltration rates and soil stability. We found little evidence that biocrust form or function recovered from fire within two years. Based on pyrosequencing results, fire altered biocrust community composition in interspace and shrub biocrusts. Cyanobacteria species were almost completely eliminated by fire, constituting 9-21% of unburned plots and less than 0.01% of burned interspace and shrub biocrust communities. Based on cover estimates, no lichen or moss species survived the fire or recovered within two years. N fixation potentials decreased by at least six-fold in burned interspace biocrusts, representing a reduction in soil N inputs into already N-limited desert soils. Soil infiltration rates also drastically declined in burned biocrusts and remained depressed, but only remained depressed for one year. To investigate the interactions between biocrust recovery, fire, and precipitation, we nested precipitation treatments manipulating the amount of monthly rainfall (i.e., ambient, plus 30% and minus 30%) within burn treatments during the second year. Soil NH4+ was the only parameter to be affected by precipitation, and exhibited a positive relationship with precipitation magnitude at the end of one year. Our results demonstrate that fire is a strong destabilizer of the bacterial components of biocrust communities and that the ecosystem services provided by crusts recover at different rates, with N dynamics recovering more slowly than soil ecohydrology.