Discussion

What determines the main vegetation gradient in Upper Guinea?

Our results show that rainfall is the single most important parameter determining the vegetation gradient for the Upper Guinean forests. However, several other factors are closely correlated with rainfall, notably soil characteristics like cation availability and soil water holding capacity. Also latitude and longitude are strongly correlated to rainfall and these geographical factors, more than soil parameters, account for the rest of the variation not accounted for by rainfall. Longitude and latitude are also the most important underlying parameters for the

Figure 4.5 Plot of the ordination axis 1 score (indicating the largest axis of variability in the species composition) as function of total amount of yearly rainfall (the parameter explaining most of the variability). The final site classifications (from Figure 4.1C) are given.

second ordination axis. This indicates that simply the location of a forest can give an indication of its position on the gradient. This is possible when location is well correlated to the most important environmental factors. This may be the case, with a strong overriding factor such as rainfall, which is rather unidirectional. Another reason may be that the gradient is a purely bio-geographical one, with a general slow turnover of species over a large distance.

Rainfall and soil characteristics strongly co-vary at regional scales (Hall & Swaine 1981, Van Rompaey 1993). Van Rompaey (1993) also showed that local situations (like the catena position) could compensate for larger scale conditions, leading to a wetter type forest in a drier area. Also for the whole Upper Guinea this strong relation between rainfall and soil characteristics holds. Swaine (1996) tried to separate the effects of rainfall and soil, using data for 155 sites in Ghana. He abstracted a composite soil parameter, based on 15 basic soil parameters. Of the 15 soil parameters 11 were significantly related to the major vegetation gradient, and so was the composite soil factor. He also showed that that composite soil factor was strongly related to rainfall. At a larger scale (over more countries) more variable soils could be expected and thus probably a less strong relation to rainfall. The rainfall gradient could also be longer however. In our case the relation between rainfall and soil holds as well, even along the much longer rainfall gradient. Possibly the major soil factors remain largely determined by rainfall. Another possibility is that the large tree species taken into account in this study are not susceptible to the extra part of the gradient (both soil and rainfall). Swaine (1996) argues that both nutrient limitation and moisture determine the distribution limits of large tree species. Of the soil parameters mineral cations seem to be the most important, and phosphorus seems not to be important for determining floristic composition (cf. Burslem et al. 1995). This is rather surprising as low phosphorus availability has been shown to strongly determine vegetation structure and species composition in Guyana (Ter Steege 1993, Raaimakers 1994) and other areas (Vitousek & Sandford 1986). Swaine (op. cit.) found a low impact of phosphorus on species composition in Ghana.

In this study we ignored the importance of the seasonality of rainfall. In an earlier study, for southeast Liberia and southwest Côte d'Ivoire, Bongers et al. (1999) showed that water availability, rather than rainfall alone was the most important parameter determining the abundance of a number of species. Length of the dry period and the intensity of that period (calculated as the cumulative water deficit) also had an effect, but less strong than the amount of rainfall. Here we have ignored these factors, but included others that are related to water availability, like soil water holding capacity. Surprisingly, the amount of rainfall alone accounts for 74% of the variation in species composition at the vegetation level.

Large trees generally are not the most sensitive plants with respect to gradual changes in environmental conditions. Including other woody species like lianas and shrubs (Hall & Swaine 1976) or herbs (De Rouw 1991) would refine the analysis and enhance the potential for indicating changes in vegetation composition at relatively small scale. De Rouw, for instance, showed that at local scale forest composition changed clearly in southwest Côte d'Ivoire, where Guillaumet (1967), working with larger plants only, did not find changes. Herbs do react to small changes in water availability while large trees probably do not. This indicates that herbs could be used to indicate gradients at small spatial scales.

In some areas the floristic gradient is steeper than in others, indicated by the intensity of change. This reflects larger changes in environment over smaller distances (cf. Van Rompaey 1993). For instance, along the line from Haut Sassandra in Côte d'Ivoire to the Krahn Bassa forests in Liberia the rainfall gradient is rather steep and thus the associated change in species composition. The same accounts for the southeast of Ghana. On the other hand, east-west gradients may be very shallow indicating that, for instance, the Haut Sassandra forests are similar to many of the forests in Ghana.

Altitude has a significant (negative) correlation to both ordination axes. The altitudinal range included is low however, as the highest forest occurs at 750 m above sea level. In general high altitude sites have high cation availability and are found at higher latitudes. High altitude sites are Lamco, North Lorma, Worobong South and Southern Scarp. In Lamco and North Lorma Heritiera is the most important species, in Worobong South and in Southern Scarp this is Celtis. In Atewa (very close to Worobong South), the data for the high plateau are mixed with data from lowland sites. The high altitude sites do not seem to take a special position in the overall gradient, probably because our data for these sites have been "polluted" by data from the adjacent lowlands. Because we had only very few high altitude sites we have indicated the high altitude areas directly on the vegetation map.

The merits of a regional map

For Upper Guinea several vegetation maps are available, but the scale is either too large or too small. Our map is one of the first for the whole Upper Guinean forests in one analysis (see also Van Rompaey 2001). We used forests from a large area (the two most distant forests are c. 1130 km apart, compare Hall & Swaine c. 530 km for Ghana). This gave rise to a longer environmental gradient. Our rainfall gradient is from 1200 to 3500 mm while the forest gradient in Ghana is from little under 1000 to 2000 mm. The differences in soil related environmental gradients between Upper Guinea and Ghana are of the same magnitude. However, the relations between the composition and environmental factors remained much the same.

Because of the larger area covered it is easier to interpolate for forests where no data are available. For example, in central Liberia hardly any detailed inventories have been done, and neither have botanical collections been made (see Wieringa & Poorter, Chapter 6). As our map is based on a selection of only 40 species, it is possible for the non-specialists (e.g. a forest manager) to locate a forest on the gradient, based on general inventory data. The fact that presence/absence data give similar results, a result found before by Hall and Swaine (1976), shows that not even data on species abundance are needed. This has been applied successfully before (e.g. Hawthorne 1995a, Hall & Swaine 1981). Another advantage is that a common map makes it possible and easy to compare forests in different countries, which was not possible before.

On the other hand, of course, we have to realise that with the large scale some of the local scale information is lost. Local environmental conditions can have a strong impact on the vegetation leading to deviations from the general pattern. A specific forest then can be of a drier or wetter type than expected based on interpolations from a larger area (see Hawthorne 1996 for examples in Ghana).

3000

S. 1500

2000

1000 100

CO CD

ds ms me we1 we2 we 3 hw

Oplan Termites

Oplan Termites

You Might Start Missing Your Termites After Kickin'em Out. After All, They Have Been Your Roommates For Quite A While. Enraged With How The Termites Have Eaten Up Your Antique Furniture? Can't Wait To Have Them Exterminated Completely From The Face Of The Earth? Fret Not. We Will Tell You How To Get Rid Of Them From Your House At Least. If Not From The Face The Earth.

Get My Free Ebook


Post a comment