Soil microbes influence the global carbon cycle via their role in the decomposition and formation of soil organic matter. Thus, rates of ecosystem processes such as primary production, soil respiration, and pedogenesis are sensitive to changes in the aggregate functional traits of the entire microbial community. To predict the magnitude and direction of microbial feedbacks on climate change, it is necessary to identify the physiological, ecological, and evolutionary mechanisms that underlie microbes' responses to altered temperature and rainfall. Therefore, I examined microbial community composition and function in relation to manipulations of resource availability and precipitation in two contrasting ecosystems: a tropical rainforest at La Selva Biological Station, Costa Rica, and a semi-arid grassland in central Texas. I conducted a leaf litter decomposition experiment at La Selva to identify the physiological constraints on microbial allocation to extracellular enzymes, which degrade organic matter. I found strong evidence that microbial enzyme production is decoupled from foliar stoichiometry, consistent with weak links between leaf litter nutrients and decomposition rates at the pan-tropical scale. Next, to examine whether ecological trade-offs within microbial communities may drive shifts in carbon cycling at local spatial scales, I quantified changes in soil fungal and bacterial community composition in response to an in situ precipitation exclusion experiment I established at La Selva. Although drought-induced shifts in community structure were small, large increases in biomass-specific respiration rates were observed under dry conditions. These findings suggest that physiological adjustments to drought may constitute an important feedback on climate change in wet tropical forests. Finally, I focused on microbial community responses to climate change within a meta-community framework, using a reciprocal transplant experiment to investigate how dispersal shapes bacterial community structure along a natural rainfall gradient in central Texas. I found that soils from the wet end of the precipitation gradient exhibited more plastic functional responses to altered water availability. However, soil bacterial community composition was resistant to changes in rainfall and dispersal, preventing functional acclimatization to precipitation regime. Together, the results of these experiments emphasize the potential for physiological plasticity or microevolutionary shifts within microbial populations to drive ecosystem carbon cycling under climate change.