The production and consumption of acacia browse and its potential for animal protein production*

Robin A. Pellew

Physiological laboratory, University of Cambridge

1. Introduction

2. Definitions and methods

2.1 The woodland types

3. Results

3.1 Woodland structure, cover density, and canopy volume

3.2 Browse biomass

3.3 Browse production

3.4 Browse consumption

3.5 Nutritional quality of Acacia browse

4. The utilization of acacia browse

Acknowledgements

References

1. Introduction

This paper considers the potential of Acacia browse as a food resource for meat production either by domestic livestock or by a free-ranging ungulate population cropped to provide a sustained annual protein yield.

In many areas of Africa, especially the Sahelian and Sudanian arid and semi-arid areas of West and Central Africa, traditional livestock management, faced with the combined problem of overgrazing and drought, has been shown to be inadequate to meet human protein needs. As Le Houérou (this volume) states in his paper on the browse in these ecological zones, these rangelands do, however, support reasonable densities of shrubs and trees (density range of between 100 and 2000 woody plants per hectare), with an estimated productivity of about 1 kg consumable dry matter/ha/year/mm of rainfall.

Although this forage potential of browse has been exploited (and over-exploited in most areas) by pastoralists, until recently it has been largely ignored by rangeland ecologists. In East Africa, no browse biomass or production measurements have been published, and only qualitative statements have been made of offtake rates by browsers, or browse impact assessments. Compared with the research effort and commitment allocated to grassland communities, the dynamics of browse has been seriously neglected.

A contributing factor to this neglect is the lack of any standardized techniques for the determination of browse production and its consumption. The browse material is often inaccessible without special equipment, and because of the extreme variability of the resource, grassland techniques cannot, in general, be adapted.

It is not the intention of this paper to discuss techniques, which have been excellently reviewed by Rutherford (1979). Only a very superficial outline is provided of the methods employed, as these are discussed elsewhere in detail (Pellew, in prep.); instead, attention is concentrated on the results which, it is hoped, will at least partly remedy the deficiency of published data of browse dynamics in East Africa.

All the data presented in this paper were collected as part of a four-year study in the Serengeti National Park, Tanzania, of browse yields in the Acacia dominated woodlands and their consumption by giraffe.

2. Definitions and methods

"Available browse", as referred to in this paper, is defined as the sum total of the plant material produced by a variety of woody species that is potentially edible to ungulates. It comprises all the green leaf and all the young unlignified (i.e. before secondary thickening) shoots of the current season's growth. Available browse is also defined by its accessibility to ungulate consumers. All the data in this paper refer to browse material occurring between ground level and a height of 5.75 m, the maximum reach of a bull giraffe. Browse above this height is considered to be unavailable as in a non-destructive way it is inaccessible to unglates. The extremely dense foliage of certain broad-leaf bushes, especially Grewia bicolor and G. fallax, can impede access by browsers, thus reducing availability. Therefore, for dense bushes with a foliage density class 4 or 5, either the outside 50 cm or the surface area only of the bush, respectively, is included as available biomass.

This measure of availability is thus a gross quantitative estimate with no qualitative restriction in the form of palatability. Standing crop biomass figures thus tend to be overestimates of actual available browse, as a proportion of the forage included in the estimates would be rejected by the consumers as unpalatable. No attempt has been made to quantify this proportion, although a subjective judgment based on giraffe feeding behaviour would suggest that it was small.

Species composition, height structure, tree density, and available canopy volume were determined for each woodland type by the Point-Centred Quarter (PCQ) technique of vegetation sampling (Cottam & Curtis, 1956).

Available canopy volume was converted to available biomass by a direct estimation technique. Each PCQ sample tree was categorized by species, and visually, by foliage density class on a scale of 0-5. Every tenth tree of each category was photographed and then harvested to determine available browse biomass/m³ of canopy volume. PCQ sample trees that were not harvested were then assigned a biomass/m³ figure by comparison with the photographs of the relevant category and best photographic fit. This Successive Photographic Approximation technique produced quick, cheap, and consistently reliable results. Trees estimated this way before harvesting showed an error factor of less than 10%. Harvesting was terminated when the standard error of the sample of a category was less than 10% of the mean. Biomass estimates could then be made in a non-destructive way. The drawback of the method is that techniques based on direct estimation cannot be treated by normal statistical analyses. Regression techniques (Rutherford, 1979) were tried and rejected due to poor correlations and excessive variances.

Productivity was estimated using tagged sample shoots on three Acacia species, A. tortilis and A. hockii of the ridge-top woodland type and A. xanthophloea of the riverine woodland type. Browsing was excluded by fencing, and at three-monthly intervals the shoots were clipped and weighed. These weights were converted to production estimates per hectare from estimates, derived from sample counts, of the total number of shoots per hectare.

Consumption was determined as an "index of consumption" by comparison of browse productivity rates with and without giraffe browsing. The index demonstrates the proportion of primary shoot production consumed by giraffe, and is expressed as a percentage of unbrowsed shoot production. Actual consumption was calculated from feeding trials of measured quantities of browse using captive giraffe.

2.1 The woodland types

The study was conducted in the Acacia dominated woodlands of the catchments area of the Upper Seronera River in the Serengeti National Park, Tanzania. Four woodland types were recognised, the distribution of these associations depending upon the topography and drainage. They were classified according to their position within the drainage catena, as follows:

a)  Ridge-top (and upper slope) Acacia regeneration thicket.
b)  Mid-slope open mature woodland.
c)  Valley bottom woodland, which sub-divided on the basis of the permanence of the drainage water flow:

c.i)Korongo (stream-line) thicket of the seasonal water-courses.
c.ii)Acacia xanthophloea riverine woodland of the permanently flowing Seronera River.

These four woodland types were assessed separately for estimation of available browse biomass. The species and available canopy volume composition of each type are shown in Table 1.

Table 1.The composition of the main woodland types of the upper seronera river catchment, serengeti national park

The mean annual rainfall of the area is about 800 mm, of which 17% falls in the five-month dry season of June to October. December (16%) and April (18%) are the wettest months coinciding with the short and long rains. Mean monthly rainfall data for the ten year period 1962–72 is shown in Figure 1 (Norton-Griffiths, Herlocker & Pennycuick, 1975). The area lies in eco-climatic zone Pas defined for East Africa by Pratt and Gwynne (1977).

Figure 1. Monthly variations in available biomass per hectare for sample area of ridge-top acacia regeneration woodland.

3. Results

3.1 Woodland structure, cover density, and canopy volume

Table 1 shows the composition of the four woodland types in terms of absolute canopy cover density, species composition, and available canopy volume. The figures give a general picture of the woodland types in the context of which their dynamics can be discussed. A. tortilis dominates the ridge-top and mid-slope types (49% and 62% of the available canopy volume respectively), whilst the Grewia bushes, G. bicolor and G. fallax dominate the korongos (43%). With 90% of the volume, the riverine woodland can be considered essentially as a pure stand of A. xanthophloea.

The relative seasonal biomass of available browse is also shown in Table 1 by species for each woodland type. Interseasonal variance is largely due to the deciduous dry season characteristic of certain browse species, including A. tortilis, A. senegal, A. hockii, and particularly Commiphora trothae. Seasonality is less manifest in the valley bottom woodland types, which have a higher ground water table.

3.2 Browse biomass

The biomass per hectare of available browse forage is shown in Table 2 for the four woodland types. The figures refer to the biomass of consumable material only, comprising young shoots and leaves, and exclude all lignified branches and stems.

Table 2. Biomass of available browse less than 5.75 m above ground level

The korongos provide the greatest quantity of browse forage per hectare, especially in the dry season. This is reflected in the dry season giraffe distribution, for at this time, when the drier ridge-top and upper slope woodlands are shedding their leaves, green browse biomass in still available in the korongo and riverine woodlands, and here the giraffe concentrate their browsing.

The high korongo biomass is due to the dominance of the broad-leaf shrubs, Grewia bicolor, G. fallax, Cordia ovalis, and Gardenia jovis-tonantis. Although collectively they represent only 9% of the species composition, they contribute about 75% of the total available biomass of the woodland type. Although it has often been used for the purpose, frequency of occurrence is a very poor indicator of availability of browse (Leuthold, 1972).

Collectively, the mean wet season biomass/m³ of all broad-leaf shrubs is 0.36 kg/ m³ (S.D.±0.17), compared with a mean figure for all Mimosaceae browse of 0.15±0.05 kg/m³. In the dry season these figures fall to 0.27±0.14 kg/m³ and 0.05±0.02 kg/m³ respectively. These woodland types dominated by Acacia species thus produce a small biomass/m' of available browse, due to the lower foliage density of the canopy, and in the dry season with leaf loss and no new shoot production, the biomass/m³ of Acacia canopy falls to a low level.

3.2.1 Biomass and rainfall

The effect of seasonality upon browse biomass is demonstrated in Figure 1 for a sample area of about 33 hectares of ridge-top mixed Acacia regeneration. Intra-specific variation in available biomass is shown on a monthly basis, together with mean monthly rainfall for the ten-year period 1962–72.

Consumable browse mass accumulates throughout the wet season reaching a peak at the end of the rains of 712 kg/ha. The dry season fall in biomass is due initially to shoot dormancy and the lignification of existing shoots, and then to leaf loss. The biomass decline follows about two months after rainfall decline, reaching a minimum in September of 167 kg/ha. The extreme interseasonal range in biomass is thus greater than a four times factor. The Acacias flush again in advance of the rains, and although the timing of this flush varies interspecifically, it is phased out over a four-week period in October. This flush is of new leaves only, and presumably is the result of the translocation of metabolites stored in the roots. New shoot production resumes with the onset of the early rains in November, and the browse mass again accumulates.

For this sample area, the mean seasonal biomass of available browse was 594 kg/ha (S.D.±101, range 410 to 712) and 304 kg/ha (S.D±136, range 166 to 524) for wet and dry season respectively.

Thus at the end of the dry season when grass in scarce and at its lowest nutritional level, the browse plants are actively producing new forage. The significance of this production at the most critical time of year is important in any browse utilisation scheme.

3.2.2 Height distribution biomass

The availability of this browse biomass to a variety of different ungulates, both domestic and wild, depends upon its vertical distribution. Table 3 shows the height stratification of the browse biomass up to 5.75 m for this sample ridge top area. The vegetation comprised ten year old mixed Acacia regeneration of about 28% canopy cover density, which has developed as a result of the almost total removal of the overstorey by elephants, combined with the introduction of effective fire protection.

Table 3. Vertical distribution of canopy volume and browse biomass below 5.75 m in the ridge-top Acacia regeneration thickets

Height Strata	Canopy Volume m3/ha	Available browse biomass				Biomass/m3 of canopy volume
		Wet season		Dry season		Wet season		Dry season kg/m3
		%	kg/ha	%	kg/ha	%	kg/m3
Below 1m	185.7	6	10.7	2	2.1	1	0.06	0.01
1-2 m	953.3	3	99.0	19	38.2	15	0.10	0.04
2-4 m	1447.9	47	272.5	52	133.5	53	0.19	0.09
5-5.75 m	491.6	16	141.0	27	78.8	31	0.29	0.16
Total
below 5.75 m	3078.5 m3/ha		523.1 kg/ha		252.6 kg/ha		0.17 kg/m3	0.08 kg/m3

Only some 20% of the total browse biomass lies below 2m. Most browse is thus above the reach of domestic livestock, except camels, unless the material is cut. And uncontrolled cutting can weaken and eventually kill the tree, in arid areas resulting in accelerated erosion and desertification

The biomass density increases with height. A cubic metre of canopy at 5 m produces five times the consumable biomass of a cubic metre at 1 m. The low canopy below 2 m represents about 37% of the total volume, yet in the dry season it produces only 16% of the total browse biomass.

The low browse accessible to livestock thus represents only a small proportion of the browse available to giraffe. Even the camel can exploit only 70% of the giraffe's food resource.

These results are comparable with figures from Southern Africa. Dayton (1978), in mixed Combretum apiculatum /C. zeyheri bushwillow veld in the Kruger National Park, estimated the standing crop biomass of currentshoots considered potentially edible for the two species combined at 1527 kg/ha. The biomass was then stratified into three height classes according to the maximum reach of impala (1.5m), kudu (2.5m), and giraffe (5.5m). The vertical distribution of the edible biomass was 4% below 1.5m, 5% between 1.6 – 2.5m, 67% between 2.6–5.5m, and 24% above 5.5m. The total edible biomass below 5.5m was 1156 kg/ha. The height stratification suggested by Dayton is very similar to the Table 3 data.

In another study in riverine vegetation in north-western Zimbabwe, Goodman (1975) estimated the twig and leaf mass at 2240 kg/ha, of which 50% was above 5 m height and only 33% below 2.5m. These results are again comparable with the Table 3 figures for the korongo woodland type.

3.3 Browse production

Rates of above ground production of browse available to giraffe are shown in Table 4 for the three Acacia species, A. xanthophloea, A. tortilis, and A. hockii.

Table 4 Rates of net primary production of consumable browse below 5.75 m

A. xanthophlea is the dominant species of the riverine woodland, representing 90% of the canopy volume below 5.75m. In the relatively dense regeneration areas (canopy cover density of 30.7%) it has an unbrowsed production of nearly 5000 kg/ha/annum. A. tortilis contributes nearly half the available canopy volume of the ridge-top woodland type and has an unbrowsed production of about 900 kg/ha/annum. A. hockii is a minor ingredient of the ridge-top woodlands, but is important in the giraffe's diet. The annual production of browse forage by the ridge-top woodland type as a whole can be estimated at about 1725 kg/ha.

In Table 5, these production rates are compared with production estimates for savanna woody vegetation reported by Rutherford (1978) in Southern Africa. An estimate of 1725 kg/ha/annum for the ridge-top Acacia regeneration thickets is not dissimilar to these production figures. However, a yield of some 5000 kg/ha/annum for the A. xanthophloea riverine woodland is substantially greater than has been previously recorded for savanna browse.

Table 5. Productivity rates of browse from a variety of savanna woodland types

In fact, the annual browse production of the riverine woodland compares very favourably with the above ground production of most savanna grasslands; Sinclair (1975) estimated productivities of about 4000 kg/ha for the Serengeti short grasslands and 2500 kg/ha for the long grasslands during one growing season of about 8 months at 500 mm rainfall. A similar general value of 4250 kg/ha/annum is given by Phillipson (1973). Such comparisons indicate that annual browse production available to ungulates can in certain instances match or even exceed that of grasslands. But the A. xanthophloea is exceptional in that it occupies a fertile site with permanently flowing ground water, and thus continues to produce browse forage throughout the year.

A comparison of shoot increments during the four quarterly periods that the tagged shoots were monitored, shows that the production of A. xanthophloea browse continues throughout the year. The quarterly production estimates for the three Acacia species are shown in Table 6, these figures being derived from the relative shoot increments during each quarter.

Table 6 Estimated rates of quarterly production of available browse below 5.75 m

Even during the dry season period, A. xanthophloea continues to produce substantial quantities of browse (894 kg/ha), at a time when the productivity of the ridge-top woodland is very low (approximately 90 kg/ha). It is this dry season production that concentrates the giraffe in the riverine woodland during the dry season. Of the tagged A. tortilis shoots, 82% were dominant during the July–September dry period as compared with 38% dormant in A. xanthophloea during the same period.

A. tortilis does show some dry season shoot growth, but it comprises a very small proportion of total annual production. If there was no dry season production, then the available A. tortilis biomass would soon fall almost to zero as the last wet season's shoots became lignified and unavailable. In fact, even in the driest month, some A. tortilis browse is available, derived from the very low level of shoot production

3.4 Browse consumption

Consumption rates of browse by giraffe are shown in Table 7. Two measures are presented, the "index of consumption" expressed as a percentage of browse production, and the actual rate of consumption expressed in kg/ha/day and determined from feeding trials.

Table 7.Rate of consumption by giraffe of browse below 5.75 m

Comparison of the actual rate derived using the index suggests that the index is an over-estimate of actual offtake. For A. xanthophloea, an index of 85% of primary production implies a consumption rate of 11.6 kg/ha/day as compared with an actual rate of 10.6 kg/ha/day. The actual rate of offtake is 78% of production. For A. tortilis and A. hockii, actual consumption represents 58% and 54% of production.

The significance of these figures lies in the high proportion of the browse production being removed, and the apparent resilience of the Acacia to such high offtake rates. Data of browse offtake rates in savanna ecosystems have not been published, so comparisons are not possible, but Lamprey (this volume) assumes a 50% rate of browse offtake in determining browse impact in the Tarangire National Park, Tanzania.

The proportion of the annual browse production removed by giraffe is certainly very high, especially in the case of A. xanthophjloea. And the browsing pressure exerted by the giraffe is increasing as the population expands beyond its present density of 1.6 giraffe/km². In the absence of crowding or resource limitations, the intrinsic rate of population increaser m, equals 0.09, with a population doubling time of only eleven years (Pellew, in prep.).

The browsing pressure upon the tree regeneration has been consistently heavy for at least the last six years, for the increase in giraffe density during this time has been compensated for by the development and increasing biomass of the food resource.

Yet despite such high offtake rates over several years, no evidence of reduced vigour can be detected in the regeneration. The browsing certainly stunts growth producing unusual topiary shapes, yet the trees appear to be highly resilient to this level of offtake, with no obvious decline in browse production. Despite appearances, there is no evidence of an "overbrowsing" situation. The term "overbrowsing" has in the past been used in an unquantified way on the basis of casual observation (Musoke, 1980). Because of the resilience of the woody vegetation to heavy browsing impact, the undefined use of the term is misleading, implying as it does a reduction in plant vigour caused by unsustainable offtake.

In many African Acacias, including A. xanthophloea, A. tortilis, and A. hockii, a significant proportion of the tree's total chlorophyll is located in the epidermal layers of lignified shoots which are not ingested by browsers, except by the destructive feeding of elephants. It is possible that this old shoot chlorophyll represents a response to browsing pressure and may partly explain the tree's tolerance to sustained high rates of offtake for defoliation by browsing, or leaf shedding in the dry season, and may not preclude photosynthesis.

The ratio of shoot to leaf chlorophyll varies considerably both inter- and intra-specifically, and even between shoots on the same tree, but can be as high as 60% shoot, (Pellew, in prep.). The photosynthetic effectiveness of old shoot chlorophyll has not been assessed, but due to problems of gas exchange, reduced light penetration through bark tissues, and translocation of metabolites, it is likely to be considerably less than leaf chlorophyll.

The grassland-herbivore dynamics in the Serengeti have been investigated by McNaughton (1979), who concludes that the ability of the grasslands to support such a high ungulate biomass can be explained by the opportunistic feeding behaviour of the consumers and by the grazing adaptations of the vegetation resource. Describing some of the grasses as "obligate grazophils", he demonstrates that grazing stimulates above ground productivity, that the grasses show considerable resistance to heavy grazing, and that grazed genotypes show definite adaptations to high offtake in the form of shorter internodes, smaller leaves, and prostrate culms. Offtake spread over only a few days, may reach as high as 85% of green forage biomass. McNaughton concludes that these grasses are able to maintain significant levels of net productivity under such severe defoliation regimes, which few cultivated grasses could tolerate, as a direct consequence of selection for grazing resistance.

It is probable that a similar situation exists in the browse-herbivore system, and that the resilience of the Acacia trees to such high browsing pressure is the result of positive selection for resistant genotypes. Unbrowsed shoot chlorophyll obviously confers considerable potential photosynthetic advantage to an individual subjected to severe defoliation. The maintenance of adequate photosynthetic material to replenish root reserves is the critical factor determining browsing resistance, especially the removal of new leaf at the end of the dry season before the onset of the rains when shoot production resumes. It is conceivable that browsing pressure has encouraged the selection for genotypes with an increased threshold of root replenishment and consequently a higher resistance to defoliation.

When extreme browsing impact, even this high threshold will be exceeded so that the plant becomes a net consumer of metabolites with insufficient replacement to maintain adequate root reserves. This condition would be manifest by a decline in plant vigour and an increased susceptibility to other environmental factors, such as drought or fire. I have observed such a development on a cattle ranch in Northern Kenya, where a very high giraffe density has reduced the vigour of A. gerrardii and A. drepanolobium resulting in widespread tree mortality during the present drought. And in southern Africa, giraffe are actively managed to deliberately suppress bush encroachment. Both these cases are examples of overbrowsing.

3.5 Nutritional quality of Acacia browse

Table 8 shows some of the results of the chemical analyses of six important Acacia browse species in the Serengeti. "Very new shoots" and "very new leaves" refers to material actively flushing and only a few days old. Such shoots are soft and droopy with the thorns still unhardened. "Young shoots" refers to shoots more than about two weeks old, but of the current growth season. Such shoots are erect and firm with hardened thorns, but do not yet show any extensive secondary thickening. "Old leaves" are more than two weeks old, have a darker green hue than new leaves, and are borne an old lignified shoots.

Shoot and leaf material was collected by a replicated sampling procedure at three times of the year, in December during the short rains, in April at the end of the wet season, and in September at the height of the dry season. The Table 8 data show the means of these collections, expressed as a percentage of dry weight.

Of particular significance are the high crude protein figures of the very new shoot and leaf material. For all six Acacia species, the mean crude protein figures are 23.5 ± 5.5% and 22.8 ± 4.7% respectively, as compared with 10.8 ± .1.5% and 16.2 ± 3.1% for young shoots and old leaves. Protein levels of very new shoots are significantly greater than young shoots (P<0.01), and those of very new leaves significantly greater than old leaves (P<0.05). The high protein levels persist for longer in old leaves than shoots (P<0.05).

The highest crude protein contents are shown by A. senegal, a species which is strongly selected for in the diet of both giraffe and elephants (Pellew, in prep). The soft and floppy flushing shoots have a mean crude protein content of 33.5 ± 3.1%, with a maximum recorded figure of 37.7%. Even in those species with the lowest protein levels, A. xanthophloea and A. hockii, the crude protein contents, especially of the very new material, are still relatively high compared with other browse species, (Dougal and Drysdale, 1964).

Table 8. Mean chemical composition (with standard deviations) of acacia browse in the serengeti, expressed as percentage dry matter

Plant part	Crude protein	Acid detergent lignin	Acid detergent fibre	Cell wall consituents	Ether extract	N-Free extract	Total ash
Acacia tortilis
Very new shoots	20.5+4.2	16.2+2.9	38.3+6.2	50.7+  7.9	3.1+0.7	32.8+ 5.2	5.3+0.8
Young shoots	10.3+0.7	20.5+4.1	46.1+2.7	60.0+  4.2	3.0+0.4	34.44+4.7	6.2+0.8
Very new leaves	21.0+1.7	13.2+2.8	23.5+3.7	44.5+10.2	4.0+0.7	43.5+ 9.6	8.0+1.6
Old leaves	15.3+0.9	15.0+3.5	24.5+0.9	39.9+  2.6	4.6+0.6	45.1+ 8.0	9.9+2.0
Acacia Senegal:
Very new shoots	33.6+3.1	11.1+4.8	29.0+5.9	47.9+ 8.7	3.0+0.3	28.8+ 5.7	5.6+0.4
Young shoots	14.1+1.4	22.2+8.1	43.1+4.8	60.3+ 3.2	3.4+0.5	33.7+ 4.9	5.7+0.9
Very new leaves	32.9+4.2	8.9+2.0	20.3+2.8	40.1+16.3	3.9+0.3	35.3+ 9.7	7.6+1.2
Old leaves	21.9+0.8	10.3+0.7	24.8+1.0	44.8+ 8.6	4.7+0.7	39.1+ 1.6	9.5+2.2
Acacia xanthophloea:
Very new shoots	18.2+2.5	12.6+3.7	33.1+3.0	54.6+ 4.5	2.4+0.3	40.2+ 2.1	6.1+0.8
Young shoots	10.0+0.8	17.9+1.3	45.2+4.8	61.0+ 7.9	2.5+0.1	37.6+ 5.3	4.7+0.5
Very new leaves	23.2+1.9	14.9+9.9	24.8+6.4	49.9+10.2	3.2+0.6	41.2+ 4.4	7.6+0.5
Old leaves	18.2+0.4	14.1+2.9	31.4+12.3	46.1+ 2.6	4.0+1.2	37.9+ 4.0	8.4+1.1
Acacia hockii:
Very new shoots	18.0+2.7	8.3+1.4	26.9+5.6	37.2+ 6.7	2.2+0.2	49.0+ 4.3	3.9+0.6
Young shoots	10.5+1.1	15.1+3.2	38.0+4.8	51.1+ 5.0	3.2+0.4	44.3+ 3.6	4.0+0.2
Very new leaves	19.4+2.6	7.0+2.8	13.9+5.4	28.4+ 5.1	3.2+0.4	58.6+ 8.4	4.9+1.2
Old leaves	15.3+0.3	8.6+3.9	15.1+2.1	39.1+11.8	4.3+0.7	58.2+12.3	7.1+2.3
Acacia clavigera:
Very new shoots	24.2+3.7	10.2+2.3	26.8+2.5	45.2+ 4.9	3.1+0.3	40.4+ 3.1	5.5+1.7
Young shoots	10.2+0.5	14.3+0.7	35.9+1.0	49.2+ 2.8	2.9+0.5	45.1+ 1.6	5.9+0.8
Very new leaves	20.2+2.0	14.2+4.4	22.8+3.3	39.9+ 4.3	4.0+1.2	45.4+ 3.7	7.6+2.2
Old leaves	12.6+0.9	10.6+0.6	21.3+0.9	35.3+ 3.5	4.2+0.8	51.5+ 2.8	9.9+3.2
Acacia gerrardii:
Very new shoots	26.5+4.5	18.7+6.0	27.9+5.7	59.4+ 6.2	2.2+0.6	36,2+ 81	7.2+2.6
Young shoots	9.9+0.8	17.4+1.4	38.9+7.0	59.3+ 1.6	2.7+0.8	43.2+ 8.6	5.3+0.8
Very new leaves	20.2+3.1	16.1+4.8	27.6+4.3	46.2+10.5	3.1+0.8	41.9+ 9.6	7.2+2.4
Old leaves	13.5+1.5	17.4+3.8	37.3+4.7	50.3+  5.3	4.5+1.1	37.4+ 4.9	7.3+2.8

But these high protein levels persist for only a few days. The soft and droopy new shoot is very vulnerable to browsing, but within about ten days, the shoot has stiffened and hardened off. This hardening is accompanied by an increase in lignin and fibre content, mainly laid down in the cell walls, so the proportion of cell wall constituents also rises. Crude protein content is inversely correlated with lignin and fibre contents (P> 0.01 in both cases), and during the first two weeks both protein levels and digestibility fall rapidly with age. After these first two weeks, there is no significant further reduction in protein content of the current season's growth.

The Acacia species have developed a strategy of growth spurts, possibly as a response to browsing pressure. By cell division at the meristem, an apical bud swells and then presumably by water uptake and cell elongation, a spurt of very rapid shoot growth follows. A shoot may show as much as 200 mm linear increment in less than 48 hours. The protein content of this shoot is very high, but it soon falls as the lignin and fibre contents rise, and the shoot hardens. A new apical bud develops, and about four weeks later, a further growth spurt begins. An individual shoot may show several such spurts during a growth season.

These growth spurts are not synchronized, and at any given time the proportion of flushing shoots on an Acacia tree is usually very low (probably less than 5%). But these high protein shoots are actively selected for by feeding giraffe, which explain their pernickety browsing behaviour. Confronted with a dense green A. tortilis bush with a substantial available browse biomass, a giraffe might take no more than a few bites before moving on. In fact, the animal is selecting the flushing shoots, which it seems able to identify visually. One result of this is an even distribution of browsing pressure. And for at least half the year, the giraffe are exploiting an extremely nutritious food resource.

4. The utilization of acacia browse

It has been demonstrated in this paper that Acacia browse can attain high rates of productivity, that the trees are resilient to high rates of offtake, and that the browse forage is highly nutritious. Acacia browse clearly has a major potential in any savanna system for improving animal protein production.

But with traditional livestock management techniques, this potential cannot be exploited because of its inaccessibility to the consumers, unless cut browse is supplied. In many arid and semi-arid regions of Africa, the cutting of browse to provide a protein rich supplement, especially in the dry season, has become an essential livestock practice.

But unless carefully regulated, this cutting will debilitate and eventually kill the trees, intensifying the problems of soil erosion in areas threatened with desertification. Exploiting cut browse may temporarily relieve the dry season food shortage, but as livestock populations expand to accommodate this resource, so the necessity for more cutting increases.

Instead, it is suggested that the browse potential inaccessible to livestock can best be utilized by the giraffe. This animal can efficiently convert the high yields of nutritious browse into animal protein, which can then be made available for human consumption. The natural browse/consumer system is a more efficient and less destructive exploitation technique than is the lopping of branches, when the bulk of the cut material is inedible.

Experience from the Serengeti and other giraffe areas of East Africa suggests that open bushland savannas with about a 20% canopy cover density produce a browse biomass sufficient to support a population density of about 2 giraffe/km² without overexploiting the range. This represents a liveweight animal biomass of some 1350 kg/km². By manipulating the population structure to maximise productivity, and by removing juveniles at about 18 months old at the time of maximum weight gain, such a population could produce a sustained annual yield of some 280 kg liveweight/km², equivalent after carcass dressing to a meat yield of about 150 kg/ km² annum. This annual protein supply derived from giraffe is equivalent to slaughtering one adult cow/km².

The main produce of cattle in a pastoral economy is milk, the meat requirement being supplied by the small stock, sheep and goats. FAO census data in East Africa (reported in Pratt and Gwynne,1977) shows that an average family of pastoralists comprises 8 individuals half of these being under the age of 14 years. In terms of nutritional requirements, the average family may thus be regarded as comprising 6.5 adults, with a daily food need of about 15,000 calories per family.

This requirement can be provided by any combination of milk meat and blood, although milk is the primary component of the diet because of its regular availability. Assuming a diet based 75% on milk, the daily food requirement of this family has been calculated at 16 litres of milk and 2.4 kg of treat, or 5840 litres and 876 kg per annum, respectively (Pratt and Gwynne, 1977).

To supply this food demand, a cattle herd of at least 40 animals, equal to about 30 adult equivalents, is required by each family, together with a herd of at least 100 sheep and goats, including a minimum of 40 breeding females. This represents a total livestock biomass for year to year survival of approximately 9000 kg liveweight per family, equivalent to at least 20 standard stock units (450 kg/unit). In terms of biomass, cattle make up about 80% of the total, although in arid areas they may be replaced as milk producers by camels. A good general figure is three standard stock units per head of population, comprising about six head of cattle and fifteen sheep and goats, although in arid areas this requirement will increase (Pratt and Gwynne, 1977).

In the thorn-bushland savannas of the semi-arid and arid areas of East Africa (Pratt and Gwynne eco-climatic zone V), the rangeland carrying capacity for livestock without serious overgrazing is in the region of about 8 hectares of productive grazing land per standard stock unit. Each pastoral family must have exclusive use of about 1.6 km² of useful grazing land to satisfy their year round subsistence. In practice, the land requirement is probably nearer 2 km² per family.

A giraffe population with a density of about 2 animals/ km², managed for sustained meat production, could yield about 280 kg fresh meat/annum on the same 2 km² required by each family. This represents some 32% of the family's meat needs.

Experience on ranches in East Africa where livestock and giraffe graze and browse together suggest that giraffe can become habituated and very tolerant of the human presence. Their loose and flexible social organization with no stable social units renders them most suitable for cropping. The provision of a large supply of meat in one location, as opposed to scattered over a wide area as is the case with small bodyweight game species (unless artificially concentrated at great expense), would simplify the marketing. The cropping of a free-range giraffe resource would appear to present no major ecological or logistical problems. By diversifying the consumer resource protein yields can be significantly increased without obviously competing with the production of milk and meat from domestic livestock on the same land.

Considerable attention is now being devoted to techniques for up-grading the productivity of the browse resource, including artificial propagation and the introduction of exotic browse plants—indeed, this has been the orientation of many papers submitted to this conference. But the browse production/consumption interface is a two component system, and improved efficiency of the consumer trophic level would increase the efficiency of the whole system. One obvious means of achieving this is to diversify the "acceptable" consumer resource base to incorporate not just traditional domestic livestock, but also wild ungulates where they have a potentially beneficial role to play. And by exploiting a portion of the browse resource at present unused except in a shortsighted destructive way, the giraffe has an obvious niche in the browse consumption system which potentially could be very valuable to man. It is hoped than an organization such as the International Livestock Centre for Africa will recognise this potential, and encourage its development.

Acknowledgements

My thanks are due to the Director and Trustees of Tanzania National Parks, and to the Director of the Serengeti Research Institute, for permission to reside and to work within the Serengeti National Park. My thanks also to the International Livestock Centre for Africa for inviting me to participate in the browse symposium and present this paper. My fieldwork in the Serengeti was financially supported by grants from the Research Awards Advisory Committee of the Leverhulme Trust, from the East African Wildlife Society, and from the Central Research Fund of the University of London. Without their help this study could not have been undertaken. This paper comprises the Serengeti Research Institute publication no. 238.

References

Cottam, G., Curtis, J.T. (1956). 'The use of distance measures in phytosociological sampling'. Ecology, 37, 451–456.

Dayton, B.T. (1978). 'Standing crop of dominant Combretum species at three browsing levels in the Kruger National Park'. Koedoe, 21, 67–76.

Dougall, H.W., Drysdale, V.M., Clover, P.F. (1964). 'The chemical composition of Kenya browse and pasture herbage'. E. Afr Wild. J., 2, 86–121.

Goodman, P.S. (1975). 'The relation between vegetation structure and its use by wild herbivores in a riverine habitat'. M. Sc. thesis, University of Rhodesia.

Kelly, R.D. (1973). 'A comparative study of primary production under different kinds of land use in South-eastern Rhodesia'. Ph.D. thesis University of London.

Leuthold, N.M., Leuthold, W. (1972). 'Food habits of giraffe in Tsavo National Park, Kenya'. E. Afr Wildl. J., 10,129–141.

McNaughton, S.J. (1979). Grassland-herbivore dynamics. In: Sinclair, A.R.E. Norton-Griffiths, M. (ed.): Serengeti, dynamics of an ecosystem. University of Chicago Press.

Musoke, M.B. (1980). 'Overbrowsing of Capparis tomentosa bushes by goats in Rwenzori National Park, Uganda'. Afr. J. of Ecol., 18, 7–10.

Norton-Griffiths, M., Herlocker, D., Pennycuick (1975). `The patterns of rainfall in the Serengeti ecosystem, Tanzania'. E. Afr. Wildl, J., 13, 347–374.

Phillipson, J. (1973). 'The biological efficiency of protein production by grazing and other land-based systems'. In: Jones, J.G.W. (ed.). The biological efficiency of protein production. Cambridge University Press.

Pratt, D.F., Gwynne, M.D. (1977). Rangeland management and ecology in East Africa. London, Hodder and Stoughton.

Rutherford, M.C. (1978). 'Primary production ecology in Southern Africa'. In: Werger, M.J.A. (ed.). Biogeography and Ecology of Southern Africa. The Hague, Dr. W. Junk.

Rutherford, M.C. (1979). 'Plant-based techniques for determining available browse and browse utilization: a review'. Botanical Review, 45, 2, 203–228.

Sinclair, A.R.E. (1975). 'The resource limitaton of trophic levels in tropical grassland ecosystems'. J. Anim. Ecol., 44; 497–520.