1. Plant productivity varies though time in response to environmental fluctuations. Reducing variability in productivity requires an improved understanding of how plant community attributes interact with environmental fluctuations to influence plant growth dynamics. We evaluated links between two community attributes, species diversity and abundance-weighted values of specific leaf area (SLA), and temporal variability in grassland productivity at patch (local) and aggregate (multi-patch) spatial scales. 2. Aggregate communities were created by combining patches of spatially-distinct communities of perennial plant species from grassland biodiversity experiments in Texas, USA. Inter-annual variability in aboveground net primary productivity (ANPP) of aggregate communities was analyzed as a function of two multiplicative components, mean temporal variability in the ANPP of patches and temporal synchrony in ANPP dynamics among patches. We used regression analyses to determine whether temporal variability in aggregate ANPP and its components were correlated with either species diversity or community-weighted SLA over 5 years. 3. Temporal variability in ANPP of aggregate communities was strongly correlated with temporal variability in patch ANPP. Increasing mean SLA reduced ANPP variability of aggregate communities by increasing mean productivity. Increased temporal changes in patch-scale SLA further reduced temporal variability in aggregate ANPP by reducing effects of precipitation fluctuations on productivity. Conversely, increasing species diversity over the narrow range measured increased temporal variability in aggregate ANPP. High diversity was associated with reduced dominance of temporally-stable C4 grasses. 4. Our results implicate means and patch-scale temporal dynamics in community SLA as potential indicators of variability in grassland primary productivity through time.