Spatial distribution of soil organic carbon and macronutrients in the deep soil across a chronosequence of tea agroforestry

Due to the rapid growth of the tea industry, the tea plantation area has steadily increased worldwide. In some regions, croplands and natural forests have been converted into tea plantations causing habitat fragmentation, reducing landscape connectivity and loss of biodiversity. On the other hand, ancient tea forests, rustic tea and tea agroforestry have been shown to support higher native biodiversity. However, there is a dearth of information on the potential of tea agroforestry systems (TAFS) to store soil organic carbon (SOC) and nutrients in deeper soil layers. For the first time, here we provide estimates of concentrations and stocks of SOC and available nitrogen (N), phosphorus (P) and potassium (K) in deep soils in a chronosequence of tea agroforestry and adjacent natural forest stands. The concentrations of SOC, available N, P and K steadily decreased with soil depth across all the chronosequence of TAFS and natural forest stands. The SOC stocks in the 0-500 cm soil depth varied from 174.5 Mg C/ha in 20 years old TAFS to 213.2 Mg C/ha in 60 years old TAFS and 311 Mg C/ha in natural forest stands. Based on the empirical data, we propose the following predictive models for estimation of deep SOC concen-trations (Y): Y = 0.55 +14.67e -0.013D in TAFS and Y = 1.04 +31.53e -0.014D in natural forests. The highest available N stocks were found in 60 years old TAFS (181.7 kg ha-1) while the available P and K stocks were highest in the forest soils (22.42 kg ha-1 and 56.86 kg ha-1). As in SOC, clay content and the SOC: clay ratios declined with soil depth in all the land uses studied. The SOC: clay ratios indicated a good soil structure in the study sites. It is concluded that TAFS store 39-46 % the SOC stocks and over 50 % of the available N, P and K stocks in the 100-500 cm soil depths. The predictive models developed here could be used for estimating deep soil SOC stocks after validation with independent datasets.