Within arid ecosystems primary productivity and standing biomass are lowest in deserts and highest in the savannah woodlands. Productivity (P) is highly correlated with rainfall (R) and dP/dR, the marginal productivity, can be used to compare sites (Noy-Meir,1985; Le Houerou, 1984). A constant dP/dR can be interpreted as water limiting plant growth. Where this slope declines indicates that nutrients (particularly N and P) are limiting (Fig. 4.3a). Thereafter, even though biomass and production may be higher, the quality, in terms of N and P content, and therefore the value as a resource for herbivores, is lower. This is of great significance to secondary (herbivore) production, (Fig. 4.3b).

   The marginal productivity (dP/dR) declines at low values of R and there is usually a threshold of rainfall below which there is no production; i.e. this threshold value defines an effective rainfall. However, small rainfall events often occur as intense showers and are redistributed to run-on areas where production is enhanced well above that had the rainfall not be redistributed. Spatial patterning of vegetation in response to this redistribution of water is common throughout arid ecosystems (Mabbutt and Fanning,1987; Shmida et al.,1986; Werger,1986).

   The life forms of the vegetation also influence dP/dR. Annual plants have a higher dP/dR than, say, perennial grasses and, particularly, perennial shrubs, which have a large metabolic investment in non-photosynthetic, woody tissues. Conversely perennial plants are able to use the small rainfalls which annuals cannot, and they are able to respond more rapidly to larger, effective rainfalls (Westoby,1980; Sinclair and Fryxell,1985; Walker, 1987).