Change of soil carbon storage in monoculture tree plantations across wide environmental gradients: Argentina as a case study

Forest plantations (through afforestation or reforestation) have been proposed as a valuable option to mitigate carbon (C) emissions. However, high spatiotemporal variation has been observed in soil organic carbon (SOC) pools associated to these systems. It has been proposed that SOC stock (Mg/ha) changes under forest plantations (Delta SOC) are related to the SOCstock in the natural system being replaced with positive Delta SOC in the sites with lowest initial SOCstock and negative Delta SOC in sites with initially high SOCstock. Although there is some debate related with statistical artifacts, the slope of this relationship may depend on environmental, biological and anthropic variables. In this context, we took advantage of a recent effort to quantify soil organic C stocks in forest plantations and contiguous land uses along 136 sites across a wide climatic and edaphic gradient in Argentina, South-America, to explore the patterns and drivers of SOC change due to this land-use change. We also added 183 data from a systematic bibliographic survey. Average Delta SOC of all the studied paired sites was low (average of -3.85 +/- 29.97 Mg ha(-1)), with 57 % of the paired sites showing a negative change. After applying a diagnosis method to detect statistical regression-to-the-mean bias and considering the pooled data and the different forest plantation groups (Pinus spp., Eucalyptus spp., Salicaceae spp, and native species), only Eucalyptus plantations showed a negative relationship between Delta SOC and SOCstock in the control situation (i.e. a significant and negative impact on SOC in high-C soils). In general, most of the sites where forest plantations were introduced in Argentina have a relatively low baseline SOC (SOCstock < 200 Mg ha(-1)), and 98 % of the studied sites showed SOC deficit (Csat-def) estimated as the difference between the observed vs. the theoretical maximal C storage potential based on silt and clay content. SOCstock and the Csat-def variation in forest plantations across regions were explained mainly by the SOCstock of the natural situation where the plantation was installed, followed by edaphic and some climatic variables. Within the silvicultural variables evaluated, only the forest species cultivated had a significant effect at the scale evaluated and within the range of stand densities (medium to high) and ages (close to rotation period in each region) considered. Our results indicate that -along a broad range of environmental situations but within a limited set of silvicultural conditions- in average forests plantations decrease soil carbon stock. However, there are situations -that could be coarsely predicted with a model with multiple variables and their interactions- where the change is positive bringing opportunities to increase the C sequestration service of planted forests.