When rainforests are being converted to commercial land-use systems such as jungle rubber (a local agroforestry system where rubber trees are planted in a logged-over forest), or rubber, or oil palm plantations, carbon is lost. In each of these examples (jungle rubber, rubber or oil palm), carbon losses resulted as follows: 116 Mg C ha−1 (tonnes of carbon per hectare), 159 Mg C ha−1, and 174 Mg C ha−1. Most of the carbon loss originated from above-ground carbon pools and represented a loss of 41%, 56% and 61% in carbon respectively (Guillaume et al. 2018; Fig. 1B). Oil palm cultivation resulted in the highest carbon storage losses, but due to the high oil palm yield it was actually the most efficient land use, i.e. oil palm had the highest yield per unit of carbon storage loss (Guillaume et al. 2018; Fig. 1A; see also Dislich et al. 2017). Ecosystem water cycling was also affected by forest transformation. Compared to selectively logged rainforests, plant water use (‘transpiration’) was 12% lower in jungle rubber agroforests, 43% lower in rubber monocultures and 19% lower in smallholder oil palm plantations. In commercial, intensively managed oil palm plantations, transpiration was 40% higher than in forests (Röll et al. 2019). Total water loss from commercial oil palm plantations (‘evapotranspiration’), as assessed by eddy covariance measurements, were also substantial (Meijide et al. 2017). Soil structure degradation and associated erosion after forest conversion (Guillaume et al. 2015, 2016) strongly reduced soil water infiltration and increased surface run-off, more so in oil palm plantations than in rubber plantations (Tarigan et al. 2018). Together with high water losses from (evapo)transpiration, this can lead to additional reductions of available water and even occasional water scarcity in oil palm dominated landscapes (Merten et al. 2016). Converting forest to plantation also affected the local climate. The air inside oil palm and rubber monocultures was warmer (by about 2.2°C) and drier (about 12% less relative humidity) than inside forests, and forest conversion amplified the diurnal range of all microclimatic variables studied (Meijide et al. 2018). Remote sensing data further indicated a local warming effect after forest conversion (Sabajo et al. 2017), with land surface temperature differences of up to 10°C between natural forest and land after clear-cut (Sabajo et al. 2017). The analysis of the land surface temperature trend of the past 16 years indicated that the average daytime surface temperature in Jambi province increased by 1.1°C exceeding the effects of climate warming (Sabajo et al. 2017).


Figure 1. Multiple aggregate ecosystem functions and their indicators. Yield/harvested biomass (A), carbon stocks (B), naturalness (C), observed local species richness (D), gross margin per hectare (E), and gross margin per hour labor (F). Indicators for naturalness (C) are: proportion forest species among bird communities, proportion indigenous tree species, proportion common weed species present. Indicators for biodiversity/species richness (D) are: number of species/operational taxonomic unit of trees, understory plants, birds, litter invertebrates, termites, ants, testate amoebae, archaea and bacteria recorded per plot. Variables in a-d were standardised to allow joint plotting. Details are given in Clough et al. (2016).