Habitats: transects to sample the local environmental variability in French Guiana since 2005.

Habitats inventories have been installed over the years in French Guiana using a standardized multi-scale protocol. In each site, we established two to four transects, about 2.5 to 3-km long in different directions to optimally sample the local environmental variability. Each 20-m width transect was divided into 100 m segments so that the basic sampling units are 0.2-ha plots. All the plots were geo-referenced using a Garmin 76CSx GPS receiver and delineated in the field using a Vertex laser clinometer. In each plot, all trees (including palm-trees) greater than 20 cm diameter at breast height (DBH) or above the buttresses were identified by local forest-plotters from ONF. They referred to a vernacular nomenclature whose names correspond either to a botanical species, genus or family. Tested against true botanical data on 4279 trees (Guitet et al., 2014), this nomenclature proved 83% effective accuracy at the family level and 74% at the level of the most precise botanical equivalent of our vernacular names (i.e. the species, genus or family level depending on the precision reached by forest-plotters). Soil samples were also collected in 490 selected plots representative of the different topographical positions along each transect. Chemical and textural analyses are available. Plants of the under-storey have been collected by IRD in several 5m x 5m sub-plots on different sites : Kourouaï, Piton Baron, Aïmara, Saut Parasol, Mont Itoupé, Waki, Toponowini, Bagote, Haute Beiman, Roche Koutou, Regina. Standardized line transects surveys were conducted by ONCFS in 25 sites on the same transects in order to record large mammals and terrestrial birds. Transects were walked at less than 1 kmh-1 speed every morning (7:00-11:00) and afternoon (14:30-18:00) by only one observer by trail, systematically alternating transects on consecutive days to avoid observer bias. All encounters of focus species and their localization on the trail were systematically recorded and the perpendicular distance between animal and transect was measured with a laser ranger finder to the nearest meter.