Uncertainties in current Earth System Model (ESM) predictions of terrestrial carbon-climate feedbacks over the 21^st century are as large as, or larger than, any other reported natural system uncertainties. Soil Organic Matter (SOM) decomposition and photosynthesis, the dominant fluxes in this regard, are tightly linked through nutrient availability, and the recent Coupled Model Inter-comparison Project 5 (CMIP5) used for climate change assessment had no credible representations of these constraints. In response, many ESM land models (ESMLMs) have developed dynamic and coupled soil and plant nutrient cycles. Here we quantify terrestrial carbon cycle impacts from well-known observed plant nutrient uptake mechanisms ignored in most current ESMLMs. In particular, we estimate the global role of plant root nutrient competition with microbes and abiotic process at night and during the non-growing season using the ACME land model (ALMv1-ECA-CNP) that explicitly represents these dynamics. We first demonstrate that short-term nutrient uptake dynamics and competition between plants and microbes are accurately predicted by the model compared to ¹⁵N and ³³P isotopic tracer measurements from more than 20 sites. We then show that global nighttime and non-growing season nitrogen and phosphorus uptake accounts for ~46 and 45%, respectively, of annual uptake, with large latitudinal variation. Model experiments show that ignoring these plant uptake periods leads to large positive biases in annual N leaching (globally ~58%) and N₂O emissions (globally ~68%). Biases these large will affect modeled carbon cycle dynamics over time, and lead to predictions of ecosystems that have overly open nutrient cycles and therefore lower capacity to sequester carbon.