Locating the PCQ Transects

The aim o f these vegetation surveys was to distinguish the effects o f cattle on the above vegetation parameters. It was therefore important to minimise any confounding variables, particularly the effects o f other human activities. Saanata was chosen as the most appropriate site for these surveys on the basis o f discussion with the local

communities and the results o f human disturbance surveys (Chapter two). The results o f the faecal count surveys had also shown that parts o f Saanata were extensively used by cattle during the dry season, but cattle use did not extend over the entire Saanata area (Chapter three).

The transects were located in a stratified random manner. Three strata were identified, based on cattle grazing indices: ‘moderately grazed’, ‘ungrazed’’ and ‘forest edge’ (intensively grazed). Two informants, both o f whom had extensive knowledge o f the forest and local plants and plant use, used large scale aerial

photographs to establish areas o f Saanata forest where livestock had and had not been taken for grazing. The older o f the two informants was roughly 55 years o f age so it

’ areas not grazed by cattle will be referred to here as ungrazed, although o f course the area will be subject to grazing by wildlife

was possible to be sure that the ‘ungrazed’ areas had not been used by cattle at least in the last 40 years.

Once the overall ‘grazed’ and ‘ungrazed’ areas had been identified on the

photographs, a 5km x 5km grid was drawn onto a map o f the area. Ten squares from both the grazed and ungrazed areas were selected at random, and each square was then located with the help o f the aerial photographs and the informants and surveyed using the plotless techniques described above. The highest grazing intensities were located close to the forest edge, but there were no comparable forest edge areas where cattle had not grazed at all. Three transects were surveyed within 200m o f the forest edge. However, direct comparison between these transects and the moderately grazed and ungrazed transects is difficult given the very different micro-climate found at the forest edge.

Unfortunately it was not possible to quantify grazing pressure in the areas chosen. A pellet count along each transect would not have been practical given that the surveys took place at the end o f the rainy season, when cattle use o f the forest was practically non-existent and would anyway apply only to current levels o f use and not to

historical levels.

The location o f the transects are mapped and shown in chapter two (Figure 2-8). The distribution o f the moderately grazed transects and the ungrazed transects tended to co-vary with distance from the forest edge, due to accessibility by the herds and the greater risk from wild animals. By spreading the transects over as great an area as possible and ensuring none o f the moderately grazed transects were less than 500m from the forest edge, it was hoped that this problem would be minimised. It would be difficult to rule out this distributional effect, however, if a significant difference was found between the two sets o f transects.

Variation is inherent in a forest ecosystem, particularly in an area such as this with steep slopes, ridges and river courses. Different altitude, aspect and slope gradient adds to the list o f common variables cited, such as soil structure, water and nutrient availability and light intensity. As far as possible, the transects were kept

homogenous by avoiding ridges, steep slopes, valley bottoms and large glades within each grid square, by choosing area o f roughly the same altitude, and by selecting for

areas o f forest which looked similar in terms o f overall structure (Greig-Smith, 1982). It may be argued that the selection o f areas which look similar would bias the results towards showing no difference between grazed and ungrazed sites. However, if the long term presence o f cattle in the area is having an effect, it should be evident from the data. Altitude was recorded for each transect and used in multivariate analysis o f the data.

The transects were each 500m in length which ensured a total o f 20 points along each transect, the number recommended for the PCQ method in forest vegetation (Mueller- Dombois & Ellenberg, 1974). For each transect, therefore, measurements were taken for 80 trees except for once, where this would have meant recording the same tree twice (i.e. from the previous point). Given the varied topography o f the area, the transects were laid out in a zigzag pattern: 150m south, 50m west, 150m north, 50m west and 100m south (Figure 4-2). Straight transects, would have made it impossible to avoid the transect following ridges or valley bottoms or crossing glades, which would have increased the variance within each transect considerably.

The transect layout is shown below in Figure 4-2.

25m 50m 75m 100m 125m 150m 175m

400m

200m

375m 350m 325m 300m 275m 250m 225m

425m 450m 475m 500m

I = PCQ point and Seedling/Sapling quadrat

Figure 4-2 Transect layout for PCQ and seedling plot surveys