Assessing the effects of agricultural intensification on natural habitats and biodiversity in Southern Amazonia

3.1. Area conversion in the land-use change scenarios In the following, the main LULCC characteristics of each scenario are described in form of a land-change matrix (Table 3). The simulation of the Trend Scenario led to a strong conversion of rainforest to pasture area in PA while a smaller share of rainforest was converted into cropland. Only relatively small shares of other natural vegetation had to be converted into managed land. In MT, a compara- tively large area covered by natural vegetation had to be converted into pastures and a smaller area into cropland. Rainforest was only converted into pastures. An intensification of pasture management is not part of this scenario. In contrast, the Legal Intensification Scenario led to no conversion of rainforest into pas- tures in both Federal States. This is explained by the intensification of pasture management as described in section 2.3. Nonetheless, in PA the conversion of rainforest into cropland was increased by 221% in comparison with the Trend Scenario due to an increasing demand for, on the one hand, fodder crops (soybean) and, on the other hand, staple crops used for human consumption in Asian countries [61]. In addition to the rainforest that is converted to crop- land, 216,020 km2 of rainforest are classified as Legal Reserve in PA. This land is not converted from its original vegetation cover but is considered disturbed due to being fragmented by small-scale agriculture. Thus, it is considered with a population impact of 0.83 (see Table 3)

when calculating the BII for PA. In MT, no rainforest was converted into managed land while the conversion of natural vegetation into managed land could be decreased by 3.1% compared to the Trend Scenario. In the case of the Illegal Intensification Scenario, in PA the conversion of rainforest to crop- land could be reduced by 99% and the conversion of rainforest to pastures could be decreased by 26% in comparison with the Trend Scenario. In PA, particularly pastures are relocated into protected areas covered by rainforest due to higher net primary productivity. In MT, a strong relocation of cropland into protected areas is discernible. Here, possible crop yields are higher than in other regions of the Federal State. In the case of the Sustainable Development Scenario, the conversion of rainforest into man- aged land was not allowed (see section 2.3). The shift of the consumption behavior towards a Table 2. Values used as population impact to calculate BII. land-use impact source Cropland 0.15 [74] weighted by proportion of high input agriculture to low input agriculture in Latin America Pasture extensive (PA) 0.6 [75] intensive (MT) 0.3 Mosaic Agricultural Area/rainforest (Legal Reserve in the transition matrix) 0.83� 20% cropland impact as calculated above and 80% undisturbed forest impact [74] Rainforest 1.0 [74] Grassland, Savannah, Shrubland, Wetland (natural vegetation in the transition matrix) 0.94 [39, 76] Fallow land 0.5 [74] Urban 0.05 [74] � A population impact value of 0.83 has been assumed for areas in the Amazon biome that are made up of 20%

cropland and 80% rainforest, the so called “Legal Reserve”, in which any kind of deforestation is prohibited according to the Brazilian Forest Code [77, 78]. This population impact value is considered as an expression for the fragmentation of rainforest. https://doi.org/10.1371/journal.pone.0225914.t002 PLOS ONE Agricultural intensification effects on natural habitats and biodiversity PLOS ONE | https://doi.org/10.1371/journal.pone.0225914 November 25, 2020 8 / 21