Table 1 Summary of covariates, data sources and processing methods

RAINMEAN

Community average annual rainfall (mm/year)

Isohyetal maps of mean annual rainfall in Trinidad from 1961–1990, 1991, 1999, 2000, 2001 and 2002 (Trinidad & Tobago Meteorological Office).

1. Isohyetal maps georectified and digitized to produce vector lines of equal mean annual rainfall.

2. Vector lines were interpolated onto a raster grid using a natural neighbor method and the statistical mean across years was then calculated using a weighted average.

3. Mean annual total rainfall for each community was then extracted (Figure 4a).

SOILFREE, SOILIMPD

Proportion of soil with free (SOILFREE) or imperfect/impeded (SOILIMPD) drainage in each community (% area)

Trinidad 1:25,000 soils vector polygon map and the land capability survey of Trinidad & Tobago (Brown & Bally 1970a; Brown & Bally 1970b.

1. Codes of dominant soils in the island were extracted from the soils map and, from these, soil and drainage type were identified (free draining, imperfect/impeded drainage).

2. Soils data were then intersected with communities and the percentage area coverage of free-drainage (Figure 4b) and imperfect/ impeded-drainage (Figure 4c) soils for each community was calculated.

RIVRDENS

Average river density by community (m/ha)

1:25,000 vector line data of rivers in Trinidad.

1. Using vector overlay, river vectors were split at the community borders and assigned community codes.

2. The vector length of each river segment was calculated and the total length of the segments was summarized by community.

3. Finally, the river drainage density for each community (average river length per hectare) was calculated by dividing the total river length in each community by the community area (Figure 4d).

WETINDEX

Average wetness index by community, In (α/tan S₀)

Raster Digital Elevation Model (DEM) derived from photogrammetry and processed to remove topographic “sinks”.

1. Upslope flow accumulation (α) and local topographic slope (S₀) in degrees were calculated in GIS.

2. Wetness index was then derived for each pixel using a map algebra calculation of In (α/tan S₀).

3. The average topographic wetness index for each community was then extracted (Figure 4e).