Tree cover and tree traits affects soil carbon and soil compaction in parklands in central Burkina Faso

Abstract: Burkina Faso is a country with an insufficient agricultural production and this is partly due to the fact that the population is growing and also because of environmental factors such as drought, erosion and poor soils. The most common agricultural system in this part of the world is the parkland which can be defined as farmlands with scattered well-grown trees. Trees contribute with soil organic material (SOM) which is the most common limiting factor for crop growth and which in turn improves soil fertility and water infiltration. This study mainly aimed at identifying the correlation between tree cover and soil organic carbon (SOC) and dry bulk density (DBD). Different tree variables such as breast height diameter (dbh), crown width and species were also taken into account. There were indeed correlations between the proximity to trees and SOC, even though the correlations were rather weak. The correlation strength varied between plots with low(>0-10%), medium(>10-40%) and high(>40%) vegetation densities based on normalized difference vegetation index (NDVI). The medium dense plots had the overall highest significant correlations between distance to trees and SOC (R2=0.25 for both closest and next closest tree). The low density strata had no significant correlations between SOC and distance to trees. The high density plots had only significant correlations between second closest tree and SOC (R2=0.086) and between “the dbh of the second closest tree divided by the distance to the same tree” and SOC (R2=0.075). The correlation between SOC and NDVI was only almost significant with a P-value of 0.052. However, strata 4 had significantly higher SOC than strata 2 and 3. DBD in general had higher correlation with distance to trees than SOC. The medium density vegetation plots however, had no significant correlations with DBD. There were significant negative correlations between DBD and both vegetation density (R2=0.26) and tree density (R2=0.25) respectively. In conclusion, there are significant correlations between both SOC and DBD with proximity to trees and different tree parameters even though they are rather weak. Since strata 4 had significantly higher NDVI, one can suggest that there some correlation between SOC and NDVI. This could in theory be used for SOC assessment with remote sensing techniques but more studies are needed on this topic.