Fig 1.
SIFs/SIEs are slaughterhouses with federal/state inspection. See S1 Table for definitions of Empresômetro, GTA, Imazon, LAPIG, Sintegra, MAPA, and MPF.
Fig 2.
Study area, active slaughterhouses (2016) and clusters of closed plants (2002–2016).
The red clouds are heat areas estimated from kernel densities of plants closed (all years) over a radius of 50 km around each plant location; darker red indicates locations with a higher historical frequency of plant closings. Most areas with closings are also where there are active plants today, except for the area just north of Cuiabá, where soy has largely replaced pastures, and an area west of Alta Floresta. A plant was classified as active if it had positive slaughter activity in 2016. All plants without federal (SIF) or state (SIE) inspection were classified as uninspected. For seven plants with a closing date but without a starting date, we estimated the starting date using the average life-cycle of the plants in the same inspection category. Sources: [23, 24, 58]; company registry (CNPJ), Empresômetro; Sintegra; Taxpayer Central Registry; Ministry of Agriculture (see S1 Table for more details on sources).
Table 1.
Slaughter volume shares in Mato Grosso by inspection system, 2013–2016.
Fig 3.
Slaughterhouses in Mato Grosso by starting year, 1967–2015.
Circles represent individual plants. The opening (closing) date is inferred from the company registry where available, from the earliest (latest) slaughter transaction date, or from the Ministry of Agriculture’s records. For years before 2013, only companies that were formally registered as slaughterhouses are accounted for, so the evolution from 1967 to 2013 indicates both actual births and deaths of slaughterhouses and the increased formalization of companies. From 2013, all slaughter transactions recorded at the state’s sanitation agency are mapped. For seven plants with a closing date but without a starting date, we estimated the starting date using the average life-cycle of the plants in the same inspection category. Sources: same as in Fig 2.
Fig 4.
Count of estimated openings and closings of slaughterhouses, 1967–2016.
The opening (closing) date is inferred from the company registry where available, from the earliest (latest) slaughter transaction date, or from the Ministry of Agriculture’s records. For years before 2013, only companies that were formally registered as slaughterhouses are accounted for, so the evolution from 1967 to 2013 indicates both actual births and deaths of slaughterhouses and the increased formalization of companies. From 2013, all slaughter transactions recorded at the state’s sanitation agency are mapped. Seven plants that have closing dates in 2006 but no opening dates are accounted for in the closings but not in the total active plants. Sources: same as in Fig 2.
Fig 5.
Share of uninspected slaughters to total municipality slaughters, 2016.
The slaughter of animals without inspection for food safety is concentrated in a few municipalities, while most of the others have very low levels of uninspected slaughter. The locations with a high incidence of uninspected slaughter tend to have high cattle densities but a poor road connection to the main urban centers. Sources: same as in Fig 2.
Fig 6.
Evolution of slaughterhouses, pastures and cattle densities, 2000–2016.
Cattle densities (head per hectare of pasture) calculated from [32] and the maximum pasture area in each municipality. Pastures include all pixels classified as ‘pastures’ or ‘pastures or agriculture’ by [54] or [56]. For seven plants with a closing date but without a starting date, we estimated the starting date using the average life-cycle of the plants in the same inspection category. Sources: same as in Fig 2.
Fig 7.
Average distance to three nearest slaughterhouses and cattle densities by biome, 2000–2016.
To obtain distances between pasture areas and slaughterhouses, we generated 15,000 random points over the pasture areas and calculated the distance between each point and each slaughterhouse. The random points were replicated three times to assess robustness. Two lines are thus presented for each biome, one with the minimum and one with the maximum distance values. Cattle densities (head per hectare of pasture) were calculated from [32] and the maximum pasture area in each municipality. Pastures include all pixels classified as ‘pastures’ and ‘pastures or agriculture’ by [54] or [56]. For seven plants with a closing date but without a starting date, we estimated the starting date using the average life-cycle of the plants in the same inspection category. Sources: same as in Fig 2.