Experimental Design of the SAFE Project

For a detailed presentation of the SAFE Project experimental design, read: A large-scale forest fragmentation experiment: the Stability of Altered Forest Ecosystems Project.

Forest Modification Gradient

The SAFE project encompasses an extensive gradient of land use intensity in the province of Sabah, Malaysia. The gradient includes old growth forest, logged forest and intensive agriculture, representing a progressive increase in the degree of forest modification. Moreover, between the logged forest and agriculture land uses, SAFE will experimentally fragment a landscape of logged forest that is being converted to oil palm plantation. This experiment will leave behind six landscape blocks that will vary in the total amount of forest cover.

  1. Continuous Old Growth Forest
  2. Sites are located in, and immediately adjacent to, the Maliau Basin Conservation Area. Two of the three sites are located within the Conservation Area itself and have never been logged. The third site (OG3) is located within the Water Catchment for the Maliau Basin Field Centre. This water catchment was lightly logged in the 1970s and again in the mid-1990s to provide timber for the field centre, but the vertical structure and species composition of the canopy and undergrowth communities in this area remains representative of primary forest in the wider region. Maliau Basin Conservation Area is embedded within a one million hectare expanse of logged forest.

  3. Continuous Logged Forest
  4. Located directly north of the experimental fragments and contiguous with the Maliau Basin Conservation Area (an area that will not be logged), the continuous logged forest is a forest concession area of approximately one million hectares. The three sites are located approximately four kilometres northeast of the experimentally fragmented area, and are situated one kilometre or more away from the nearest forest edge. The part of the forest concession that includes the sites has been logged twice.

  5. Logged and Fragmented Forest
  6. The experimental fragments will be created within an 8,000 ha block of logged forest that is currently contiguous with the one million hectare forest concession. This area has also been logged twice, including all sites that will eventually become fragments. Further details on the spatial layout of the fragments can be found at the description of the experimental design.

  7. Oil Palm Agriculture
  8. Oil palm sites are located seven kilometres west of the experimental fragments, with the intervening area currently comprised of Acacia plantations. The three oil palm sites are located in blocks of oil palm that vary in age from 4 to 10 years. The youngest site is very open and has a herbaceous ground cover dominated by legumes. The oldest site is very different: the palms form a canopy that shades out ground cover, and the palms themselves host a variety of epiphytes.

Together, the SAFE Project includes more than ten levels of habitat modification, spanning a massive range of human impacts from no impact, through resource extraction into partial habitat conversion, across progressively more intensive land uses and culminating in one of the world’s most intensive forms of agriculture.

Safe Project Experimental Fragments

Figure 1. Map of the SAFE Project experimental stations and fragments . The experiment comprises six blocks (A-F), each with seven fragments (1 ha fragments are obscured by the sampling points). Sampling points show the locations of the 2nd order fractals (See Sampling design). In addition to the experimental fragments themselves, sampling across edge gradients will occur in a Virgin Jungle Reserve (VJR) adjacent to the experimental area and at the edge of continuous logged forest (LFE) to the north of the experimental area.


The experiment has a split-plot experimental design. There will be six experimental blocks, each containing four plots: (1) 1 x 100 ha fragment, (2) 2 x 10 ha fragments, (3) 4 x 1 ha fragments; and (4) forest that will be converted to oil palm. Across all blocks, there will be a total of 42 fragments (Figure 2).

Fragments will be arranged in space such that the total amount and the spatial distribution of sampling effort within the different fragment sizes is equal (see Sampling design). Sampling points within the matrix will be aligned with the sampling points in the 100 ha fragment to form a large-scale transect crossing the fragment edge.

The six experimental blocks are placed within the experimental area in such a way as to maximise the range of forest cover that will remain in the landscapes surrounding sampling points. Average forest cover around sampling points within the six blocks ranges from 18 to 54 % when a landscape is described as a circle with a 3 km radius (see Forest modification gradient).

Experimental blocks are also located and oriented within the study area to take advantage of the local topography. All blocks and control sites are placed so that most sampling points are located within a single altitudinal band between 400 and 500 m, removing the potentially confounding effect of altitude from the experiment. Blocks are also oriented to remove other potentially confounding effects such as slope, latitude, longitude, and distance to forest edges prior to the forest conversion.

In addition to the sampling sites in the experimental fragments, SAFE we will take advantage of a 2,200 ha Virgin Jungle Reserve (VJR) that will become isolated during the forest conversion process (Figure 1). A single transect of sampling points will be placed from the edge to interior of the VJR. The transect extends twice the distance of the transects in the experimental fragments, representing the larger size of the VJR, but the distribution of sampling points will still be based on the fractal pattern used elsewhere. A similar, long transect will be placed from the edge to interior of the continuous logged forest habitat (LFE).

The forest clearance, and hence the creation of the fragments, began in July 2011.


The SAFE Project will experimentally manipulate the width of riparian strips bordering permanent streams. Within the experimental area, we have identified six micro-watersheds of approximately equal size (260 ha ± SD 10) and slope (16˚ ± SD 2). Watersheds vary in the amount of forest cover that will remain (range 8 – 39 %).

The width of riparian strips in the experimental watersheds will be 0, 5, 15, 30, 60 and 120 m. Below all experimental watersheds, riparian widths will revert to the Malaysian legal requirement (30 m, subject to confirmation). Riparian widths are assigned to basins in a way that ensures width is not confounded with watershed size, slope or forest cover.

In addition to the experimental watersheds, we have identified a further three watersheds to act as controls (1 × old growth forest, 1 × logged forest and 1 × oil palm plantation). Control watersheds were chosen to match the experimental watersheds as closely as possible in terms of size and slope.

Sampling Design

The sampling design at SAFE is a heirarchical one based on a fractal pattern and is explicitly designed to allow for the unification of data from ecological patterns and processes that operate at different scales (Figure 2a). The basic design consists of sets of three sampling points arranged in a triangle. The distance between these ‘1st-order’ sampling points is metres and is relevant to taxa such as invertebrates. For other taxa such as birds and tree communities, such a fine distribution of sampling points makes little sense, so these will be sampled in the middle of each triangle of 1st-order sampling points. These 2nd-order sampling points are separated by metres. Further increases in scale occur with sites identified at the 3rd-order (samples separated by m) and 4th-order (samples separated by m). A final sampling point is designated that falls in the centre of the sampling triangle.

We have identified all sampling sites on this fractal pattern in the old growth forest, logged forest and oil palm landscapes (each with 81 × 1st-order sites, 27 × 2nd-order, 9 × 3rd-order and 3 × 4th-order). For samples within the fragments, we use a transect that is extracted from the fractal geometry of the control sites (Figure 2b). Each of the experimental blocks has 48 × 1st-order sampling points (16 × 2nd order, 8 × 3rd-order and 4 × 4th order), with the number of points and spatial distribution of points the same in each of the four plots.

Similar to the transects used in the fragments, we have established long transects extending from the edge to interior of a 2,200 ha Virgin Jungle Reserve (VJR) and at the edge of the continuous logged forest habitat (LFE). These transects extend twice the distance of the transects in the experimental fragments, but the distribution of sampling points is still based on the fractal pattern used elsewhere.

For more information about the overall sampling design, see this paper: A fractal-based sampling design for ecological surveys quantifying beta-diversity. To as large an extent as possible, we would prefer all independent researchers to base their sampling on this spatial design. This won’t always work for everyone, but is strongly encouraged and a component of our requirements to work at SAFE.

Sampling Design Diagram

Figure 2. Structure of fractal sampling design.