Evaluating long term soil organic matter dynamics of cocoa farms in Indonesia.

Abstract

In Sulawesi, Indonesia, cocoa (Theobroma cacao L.) yields are suboptimal. Most cocoa farms are agroforestry systems, thought to be efficient at storing large carbon stocks (C), protecting the soil from degradation, and recycling nutrients. Despite this, inappropriate management practices can lead to the progressive deterioration of soil fertility and constrain cocoa productivity. One critical component of soil fertility is soil organic matter (SOM). Although organic additions are available to producers, the SOM dynamics of cocoa farms are poorly understood, precluding the development of evidence-based practices for SOM and fertilizer management. Hence, this research was conducted to determine the relationship between organic matter additions, soil fertility, and cocoa production, using meta-analysis, field experimentation, a chronosequence study, and modelling. The meta-analysis, which incorporated 37 references from 14 countries, showed that the mean C stock of 15 to 35-year old cocoa systems (including shade trees, and soil to 10 cm depth) was ~85 Mg ha⁻¹. For this age range, the mean C stocks for aboveground cocoa, shade trees, litter, and roots were approximately9.8, 37.4, 1.0, and 11.4 Mg ha⁻¹ , respectively. The mean soil C stock (0-10 cm) was ~24 Mg ha⁻¹. If taken from deeper soil layers, soil C stocks can be substantial and may exceed plant C. Large differences observed within the same age classes suggest that modified designs and practices can increase C storage for a particular pedoclimatic context. The continuation of an already established field experiment (a randomized block experiment with 16 cocoa trees for each four repetitions, including applications of mineral fertiliser, compost and dolomite alone and in combinations) indicated that compost application (locally made of 60% cow manure, 15% empty oil palm bunches, 10% rice straw, 10% diverse leaves (banana, grass, Gliricidia, and maize), 5% cocoa pod husks, and a EM4 micro-organism mix; 10 kg tree⁻¹ year⁻¹) increased cocoa yields (over four years) to 1.8 Mg dry bean ha⁻¹, three times that of a control treatment with no additions. The four-year cumulated yield of the fertiliser-only treatment was 0.98 Mg dry bean ha⁻¹. The tree survival rate was low in the fertiliser-only blocks (on average 41% after 7 years). No additional yield effect was observed by adding fertiliser or dolomite to the compost treatment. Soil responses were variable. For example, measured 25% HCl extractable P declined across all treatments, and a loss of soil organic C (SOC) occurred across all treatments with composts. This suggests that the maintenance of SOM through compost additions requires a systematic understanding of their losses and inputs. A chronosequence study across 13 Sulawesian cocoa farms (0.5-31 years old) indicated significant SOM losses within cocoa plantations in the first 1-5 years after planting, as SOM mineralisation was greater than the rate of new SOM addition. Soil samples (0 - 100 cm) were collected in 20 cm increments to determine SOM, SOC, and N contents, clay-adjusted SOM, SOC, and N contents, and SOM, SOC and N stocks. The observed decline between 0.5 and 2 years in SOM (-46%) was also associated with declines in SOM per unit clay (-40%). These findings suggest that from the moment a plot is cleared in preparation for planting, the high temperatures and precipitation found in Sulawesi can result in rapid soil degradation through fast SOM mineralisation. Future research should focus on the first years after planting, and farm practices, such as strategic organic additions, should target this sensitive period. The modelling study provides a framework to predict SOM variations on cocoa farms. The model combined the AMG soil model (Andriulo et al., 1999; Clivot et al., 2019; Saffih-Hdadi & Mary, 2008) with a cocoa growth curve from the chronosequence dataset. An annual SOM mineralisation rate of 0.125 (unitless) was calculated using the characteristics of a representative farm of the chronosequence dataset (averaging the local variables of each farm) and represent a relatively high rate compared to other world locations. Backward modelling was used by optimisation to simulate SOM dynamics in each of the 13 farms. The simulations indicated that SOM could deplete rapidly after planting, and the long-term trend can either be a decline or a build-up and even exceed planting levels. In general, farms with a high initial SOM content tended to lose SOM, whereas farms with a low initial SOM content tended to gain SOM in the long term (after 20-30 years of cultivation). The model was also applied to calculate the amounts of various organic inputs required to offset SOM losses fully. This model was programmed in R, and an RStudio Shiny app was developed to allow for user-friendly simulations. Future research should include further calibration of model parameters, improved modelling of pruning and shade trees in residue deposition, and making crop growth responsive to environmental parameters. The above results highlight that Sulawesian cocoa farms are particularly at risk of SOM losses in the initial years after planting. This is a critical period during which organic additions could support cocoa productivity and provide other environmental benefits. Recommendations for SOM management and future research are proposed to limit soil degradation and improve the C balance of cocoa farms.