Fig 1.
Protected areas (polygons) and land cover in the study region.
Fig 2.
Flowchart of methodology used in the study.
1) Pre-selection of the best covariates groups. 2) Realization and evaluation of the matching quality.
Fig 3.
Regular grid with cells of 1000 m by 1000 m.
The map depicts the sampling units in protected areas, the sampling units excluded from the analysis, and the sampling units from the 10-km buffer.
Table 1.
Percentage of different anthropic classes that were classified correctly for the 100 best groups of pre-selected data.
Fig 4.
The ATT (a) and ATT% (b) for the 15 best models (box plots) and for the best model (red dots) of buffer and non-buffer subgroups.
PA–Protected Area, SP–Strictly Protected Areas, SU–Sustainable Use Areas, Fed–Federal Units, Sta–State Units, >Sz–Larger size PAs, <Sz–Smaller size PAs, <86—Before 1986, <96—Between 1986–1996, <02—Between 1996–2002, <08—Between 2002–2008.
Table 2.
Average treatment effect for SNUC Protected Areas as estimated from the best data group.
Fig 5.
The absolute mean ATT and ATT% for the 15 best models (box plots) and for the best model (red dots) regarding restriction and government sphere groups.
a) restriction group (a.1 –ATT, a.2 ATT%), b) government sphere group (b.1 –ATT, b.2 ATT%). PA–Protected Areas, SP–Strictly Protected Areas, SU–Sustainable Use Areas, Fed–Federal Units, Sta–State Units, >Sz–Larger sized PAs, <Sz–Smaller sized PAs, <86—Before 1986, <96—Between 1986–1996, <02—Between 1996–2002, <08—Between 2002–2008.
Fig 6.
The absolute mean ATT and ATT% for the 15 best models (box plots) and for the best model (red dots) regarding size group and cohorts.
a) size group (a.1 –ATT, a.2 ATT%), b) cohorts (b.1 –ATT, b.2 ATT%). PA–Protected Areas, SP–Strictly Protected Areas, SU–Sustainable Use Areas, Fed–Federal Units, Sta–State Units, >Sz–Larger sized PAs, <Sz–Smaller sized PAs, <86—Before 1986, <96—Between 1986–1996, <02—Between 1996–2002, <08—Between 2002–2008.