Understanding temporal variability in net primary production (NPP) is essential for developing ecosystem models to predict regional carbon budgets. Although estimates of aboveground NPP are increasingly available, long-term measurements of belowground production are rarer despite representing a large proportion of NPP in water-limited ecosystems. Moreover, recent work suggests above- and belowground production may respond asymmetrically to environmental drivers in these systems. To examine the long-term dynamics of net primary production, we analyzed thirteen years (2005-2017) of above- and belowground NPP data collected from five sites associated with the Sevilleta Long Term Ecological Research program. Sites span a semi-arid grassland-shrubland ecotone in central New Mexico, USA, where dominant vegetation ranges from mixed blue and black grama grassland to creosote shrubland. One grassland site received 10 g m-2 yr-1 nitrogen prior to the summer monsoon, while the other grassland site underwent a management burn in 2003, and both grassland sites were paired with complementary control sites. Aboveground production was measured allometrically twice per year in 30-40 1 m2 permanent plots to determine the maximum biomass for each species. Belowground production was measured in the same plots using the root ingrowth method (0-30 cm), and samples were harvested annually following the growing season.
Results/Conclusions
From 2005-2017, average aboveground NPP ranged from 71.7 g m-2 y-1 in burned grassland to 104.0 m-2 y-1 in fertilized grassland, while average belowground NPP ranged from 81.5 g m-2 y-1 in burned grassland to 102.3 g m-2 y-1 in unfertilized grassland. Temporal variability in aboveground NPP was generally low, ranging from a CV of 31.7% in shrubland to 53.8% in fertilized grassland. Aboveground NPP was significantly correlated among all sites, ranging from R2=0.34 (p<0.05) to R2=0.86 (p<0.001). In contrast, temporal variability in belowground NPP was much higher, ranging from a CV of 53.4% in burned grassland to 73.7% in fertilized grassland. Fewer sites were correlated with respect to belowground NPP, particularly when comparing grasslands with shrubland, with significant correlations ranging from R2=0.35 (p<0.05) to R2=0.97 (p<0.001). Thus, while aboveground NPP generally followed similar patterns, belowground NPP was less temporally and spatially consistent across this semi-arid grassland-shrubland ecotone. Within sites, the temporal correlation of above- and belowground NPP ranged from R2=0.00 in unfertilized grassland to R2=0.19 in burned grassland. However, none of these correlations were significant. Since above- and belowground production is not correlated, our results suggest the need for long-term above- and belowground NPP measurements for ecosystem model development.