By
J.I. House & D.O. Hall
Division of Life Sciences, King's College London

There is a paucity of data on the biomass and productivity of different types of savannas and grasslands under the considerable range of climatic and soil conditions they experience. Of the studies that have measured savanna biomass and/or productivity, very few have been sufficiently long-term to capture the interannual variations (which can be considerable), few have measured both tree and grass components, even fewer have measured roots, and almost none have looked at insectivorous and microbial populations, although herbivores are sometimes considered. Thus there are no estimates for total ecosystem biomass/productivity. This chapter considers plants only.

Table 1 presents several estimates of biomass each based on different published data. Biomass values reported for individual savanna sites and compiled in Scholes and Hall (1996) are shown in Table 3, with the original class names and groupings, although the calculations have been re-done and totals differ slightly from the original paper. Due to the paucity of root data, total biomass was calculated from average aboveground biomass using estimates belowground biomass percentage, ranging between 20% and 30% in different savanna types. Jackson et al. (1996,1997) reported a higher average belowground biomass of 40% for tropical savannas and grasslands in their compilation and analysis of root data for all global biomes.

It is difficult to match up the biomass classes given by Scholes and Hall (1996) in Table 3 with the land use classes of Olsen et al. (1983) in Table 1. Scholes and Hall (1996) assigned their woodlands biomass class (originally 83 t/ha) to the land class drought-deciduous woodlands. In fact, the original Olson et al. (1983) description of tropical dry forest and woodland class includes drought deciduous forest types such as Australian Brigalow as well as the Miombo woodlands, therefore it seems more reasonable to combine these two classes in Table 3. Scholes and Hall (1986) assigned their savannas biomass class value (originally 25 t/ha) to all other land classes as it included examples from all of them. The Scholes and Hall (1996) biomass values are all lower than the Olson et al. (1983) values which were also based on reported data. For the purposes of this chapter it seems reasonable to take an average of the two sets of estimates, using the re-calculated Scholes and Hall (1996) values in Table 3. This gives an average biomass value for the savanna biome of 5.75 kg DM m- ², or a total biomass of 158.5 Pg (DM). Assuming 30% of biomass in roots, the average belowground biomass is therefore 1.73 kg DM m- ², and aboveground is 4.02 kg DM m- ². The Jackson et al. (1996, 1997) analysis of root data calculated a total (tree and grass) root biomass of 1.4 kg m- ² (of which fine roots account for 0.99 kg m- ²). Using his value of 40% root biomass this would equal a total biomass of 3.5 kg m- ², much lower than the other estimates.

Woody biomass production is non-linear, declining with maturity which can be in the order of 30 to 100 years (Scholes & van der Merwe, 1996). It is not possible to estimate average annual production from tree rings as growth may be negligible in drought years, and certain regions experience two growing seasons in some years. Therefore production is typically estimated from change in biomass alone, which is an underestimation as it does not take account of death and turnover. Litter traps fail to account for dead branches. Removals for fuelwood, building materials, browsing, etc. are common but poorly recorded in these largely "unmanaged" areas.

Estimates of tropical grassland production vary by almost five-fold depending on the techniques employed (Long et al, 1989). Herbaceous NPP has generally been estimated using annual maximum standing-crop (aboveground), usually at the end of the growing season, see Bourlière & Hadley (1970) (figures for 22 tropical grasslands) and Rutherford (1978). Importantly, this method does not account for belowground productivity, the effects of grazing and trampling, mortality before or growth after peak standing-crop has been attained, the differences in time at which species attain their peak standing-crop (especially mixtures of perennials and annuals), or litter turnover (Rutherford, 1978; Solbrig, 1996). Milner & Hughes (1968) proposed a method for the International Biological Programme (IBP) which measures positive increments in aboveground live biomass, see Singh & Joshi (1979) (review of 21 studies in tropical grasslands in India and Africa), but this still leads to underestimates.

A more robust methodology was developed under a United Nations Environment Program (UNEP) project for measuring NPP in tropical grasslands (Long et al, 1989, 1992). This measured monthly increments in aboveground live and dead biomass, roots, and monthly decomposition rates for standing dead material, litter and roots. NPP was defined as the sum of net monthly increases in live biomass plus losses due to death and decomposition. NPP at three of these grassland sites was two to five times higher than that obtained using the standard IBP procedure which ignores mortality and two to ten times higher than previous methods which ignore belowground NPP (Long et al, 1989, 1992). However, the UNEP studies still probably underestimate NPP as they do not account for losses due to root exudation or removals due to pest attack.

Studies accounting for belowground productivity, phenology, death and decay have NPP values that approximate the figures for tropical forests (Solbrig, 1996). Measurements of worldwide "grassland" NPP have been incorporated into an Internet database site managed by the U.S. Oak Ridge National Laboratory (ORNL) Distributed Active Archive Centre - http://www-eosdis.ornl.gov/npp/npp_home.html (Scurlock et al, 1999; Cramer et al, this volume). The sites are often mixed tree-grass systems, although for many of the studies only grass NPP was measured.

Belowground production can be as high as or higher than aboveground production, but it is difficult to measure as it is hard to distinguish between live and dead roots, and different species, and assumed root:shoot ratios are often used. However, the ratio is highly variable depending on the vegetation formation, maturity, prevailing environmental conditions and disturbances. For example, in the UNEP study, belowground NPP varied from 40% to 70% of total NPP in different years at some sites (Long et al, 1992). While root :shoot ratios tend to increase with stress, severe drought or frequent fire can lead to negative belowground NPP (Long et al, 1992). Root turnover and exudates are rarely considered. Menaut & Cesar (1979) observed root turnover up to 100% in an open humid savanna and 70% in wooded savannas. It is common in savanna studies to assume belowground production is equal to aboveground production.Some estimates of NPP in the savanna biome as a whole can be found in Table 1. Whittaker and Likens (1973, 1975) used measured production and phytomass values and extrapolated them by their estimated area. Atjay et al (1979) used this as a basis for their own study, but they evaluated more recent data on NPP and took account of the role of organic matter. They cite improved root production data as the main factor accounting for their much higher NPP value. In Atjay et al’s assessment, savanna has the second highest share of total production after forest systems. Lieth (1973) suggested mean production of 800 g/m²/y which Long et al (1992) multiplied by 3.5 to take account of methodological underestimates to give a value of 2800 g/m²/y. Both Olson et al (1983) and Scholes and Hall (1996) give much lower estimated rates of productivity. It is not clear where Olson et al obtained their figures, but the Scholes and Hall values were derived as explained below:

Rather than apply one estimate of NPP (based on studies at a small number of sites) to an entire vegetation class that incorporates a range of environmental conditions, Scholes & Hall (1996) used data from individual studies to develop a relationship between water availability and NPP, and then applied this to a number of points in each class. Table 4, compiled by Scholes & Hall (1996), lists a series of NPPs reported for tropical savannas and grasslands. To be included, studies must have measured NPP over a period of at least a year, and used the sum of positive increments method plus some assessment of losses.

Several studies reported belowground NPP for tree and/or grass components, enabling the calculation of a relationship between total NPP and aboveground NPP for the trees and grasses:

total NPP = 1.01 (aboveground NPP) + 853

This equation predicts that belowground production accounts for between 5 and 70% of the total NPP, the proportion decreasing with increasing productive potential of the site (Scholes & Hall, 1996). However, this equation would seem to have an unrealistic intercept and is driven by an outlier - the floodplain site Manaus in Brazil which can be considered an exception due to high water and nutrient availability, although it does show the potential of grassland vegetation. This relationship is redrawn in Figure 2 without the outlier, and, sticking to the simple linear relationship, the equation becomes:

total NPP = 1.42 (aboveground NPP) + 284

Scholes & Hall used their original equation to calculate total NPP for all the sites, then plotted total NPP (tree and grass, above and belowground) against a calculated water availability index (WAI) based on the monthly ratio of rainfall to evaporation at each site (equivalent to the number of days per year without water stress), presented in Figure 3. The derived relationship was then applied to a large number of randomly selected locations in each of the vegetation classes to calculate more representative average NPP values, presented in Table 1. While water and nutrient limitation do constrain savanna productivity, the inherent capacity for production by savanna plants is comparable to forest systems i.e. they can produce just as much or more biomass per unit of water available. Referring to Figure 3, if the upper range of data is projected to a water availability of 365 days (i.e. no water limitation) the predicted annual NPP is 4000 g/m², which is well within the average-to-high range for other natural ecosystems (Scholes & Hall, 1996).

Many thanks to Bob Scholes (CSIR, South Africa), Xavier Le Roux (INRA, Clermont-Ferrand, France), Jonathan Scurlock (ORNL, USA) and Joe Scanlan (Department of Natural Resources, Queensland, Australia) for providing information and making corrections to the manuscript. Dale Kaiser & Sonja Jones (ORNL, USA) for calculating tropical % of the Olson et al (1983) "grasslands" category.

Sadly, David Hall passed away in August 1999 before this chapter was published. His knowledge and love of savannas was only surpassed by his eagerness to learn and teach.

     If you have entered this site from UNEP.NET or SCOPE, then feel free to use the link below to view the full site.