Woody plant proliferation in grasslands and savannas over the past 150 years is thought to comprise a significant, but highly uncertain component of the North American terrestrial carbon budget. Assessments of belowground carbon stocks in ecosystems undergoing this transformation are typically based on soil and root samples obtained via coring techniques.  We contend that such assessments fail to account for two major belowground shrub carbon pools: shallow lateral roots and lignotubers.  To determine how much is missed by not accounting for these pools, we excavated 26 velvet mesquite (Prosopis velutina) shrubs in a desert grassland and determined the abundance and mass of shallow primary, lignified lateral roots arising from lignotubers; and the mass of the lignotuber giving rise to these roots. Shrubs were selected to cover a broad range of aboveground biomass (0.56 to 435.41 kg/plant).

Results/Conclusions

Mean (SE) lignotuber biomass [15.4kg (3.3); range = 0.05-66.00 kg/plant) scaled linearly with aboveground biomass (R² = 0.86).  These structures gave rise to an average of 15.8 (1.4) primary lateral roots with average diameter of 2.8 (0.2) cm (range = 0.07-15.7 cm) at the bole. The density of lateral roots was low [2.1 (0.6) roots/m² canopy area; range = 0.2-12.0)]. Roots tapered as they extend outward from the bole, their mass varying as a function of distance and initial diameter (R²=0.83). Total under-canopy primary lateral root biomass was 0.90 (0.11) kg m² (range = 0.34-2.66). The combined under-canopy lignotuber and lateral root biomass scaled linearly with aboveground biomass (R²= 0.91); and for plants with stems >11 cm basal diameter these structures comprised 29 (9)% of the measured total (above + below ground) plant mass. Soil coring approaches commonly used to obtain estimates of belowground carbon mass completely miss the substantial biomass contained in shrub lignotubers; and given the scattered distribution of primary lateral roots the probability of encountering them with soil core sampling approaches is low. Furthermore, when large, lignified roots are encountered, they often obstruct penetration of the core sampler, requiring it to be moved. In addition, our estimates of primary lateral root contributions to shrub biomass are highly conservative in that sampling was confined to the sub-canopy area, while other studies indicate laterals can extend 3-8 times the canopy radius. Ground-penetrating radar, coupled with strategic excavations and allometric modeling, may be an alternative approach for generating stand- and landscape-scale estimates of these components of the ecosystem carbon pool.