Drylands account for ~30-35% of terrestrial primary production and are an important component of the global carbon cycle. Changes in dryland vegetation thus have implications for carbon uptake and storage. One widely observed change is the conversion of grasslands to shrublands and woodlands. Although regarded as significantly impacting the North American terrestrial carbon sink, estimates are highly uncertain. Shrub proliferation typically increases aboveground carbon pools. However, the majority of organic carbon in drylands resides as soil organic carbon (SOC). Reports of impacts of shrub proliferation on SOC are equivocal and range from substantial decreases to no net change to substantial increases. 

We quantified changes in ecosystem carbon pools accompanying shrub (Prosopis velutina) proliferation in Arizona grasslands. Size and location of P. velutina plants in a 0.4 ha area were determined in 1932, 1948 and 2006. Aboveground C mass of P. velutina was calculated using allometric equations. SOC (0-20 cm) was quantified along bole-to-dripline transects for a range of shrub sizes. A predictive model of SOC mass as a function of shrub size (basal diameter) was developed, applied to each individual on each date and summed across all shrubs to estimate aboveground shrub and SOC stocks over time (g C m^-2). 

Results/Conclusions

Aboveground shrub C mass increased from 1932 (217 g m^-2) to 1948 (362 g m^-2) and then declined to 224 g m^-2 in 2006 as a result of an herbicide application in 1965/66. The herbicide application ostensibly shifted the population structure to smaller plants (mean plant canopy area declined from 6.3 [1948] to 3.8 m² [2006]), but was coupled with an increase in plant density (323 ha^-1 in 1948; 443 ha^-1 in 2006). SOC comprised 72-81% of the soil plus aboveground shrub carbon pool. In the absence of shrubs, SOC mass was 896 g C m^-2; accounting for the influence of shrubs increased SOC mass to 912 g C m^-2 in 1932 and to 940 kg C m^-2 by 2006, a 3% increase despite the herbicide application. Our results confirm Prosopis encroachment into semi-arid grasslands will increase carbon storage in both the aboveground and near-surface SOC pools. Because SOC accumulation increases with tree size, historically invaded landscapes with larger individuals will be greater carbon sinks than more recently invaded landscapes.  Land management practices (brush control) can dramatically reduce the aboveground C pool, but it recovers fairly quickly, with a potentially lagged, but poorly understood impact on the SOC pool.