6.4.1 Drought impacts on livestock
6.4.2 Wealth effects on herd losses
6.4.3 Decline in terms of trade
6.4.4 Traditional drought-mitigation tactics
6.4.5. Equilibrial versus non-equilibrial population dynamics
Compared to the dry season in an average rainfall year, the net production impact of the 1983-84 drought on Borana households found by Donaldson (1986) was a 92% reduction in milk offtake for human consumption due to a 90% reduction in the number of cows in milk and a 65% reduction in offtake per lactating cow. The major effect was on calving percentage owing to poor nutrition. Around 60% of the pre-drought inventory of cattle was lost due to modality (largely cows and immatures) and from the sale of males and other followers to purchase grain. Drought-related animal losses appeared to vary according to region (Cossins and Upton, 1988a). It is assumed that calves are highly vulnerable because of restricted milk supply and immobility. Milk cows may be more vulnerable owing to the energy demands of lactation and because they are kept near encampments where forage is commonly depleted. Prime cows are often the last animals to be sold because of their high future production value and pastoralists retain them in the hope that the next rainy season will be good (Coppock, 1992b).
The lower death rates for males is a testament to their value as a risk-mitigating investment (de Leeuw and Wilson, 1987). Seemingly nonproductive, mature male cattle held in large numbers have been traditionally regarded by pastoral development agents as a commodity of little apparent value (D. L. Coppock, ILCA, personal observation). This view is inappropriate, however, given the high value of males for offsetting asset losses during drought and as item to trade for cows during post-drought recovery on the Borana Plateau (see Section 5.4.5: Cattle growth and implications for breed persistence).
Small ruminants appear to have fared better than cattle during the 1983-84 drought. In part this is probably because of their lower absolute nutritional requirements and mixed food habits that include drought-resistant browse (see Section 3.3.5.1: Livestock food habits).
There was no evidence that camel mortality rates were appreciably lower than those for cattle in the Beke Pond area. There was evidence that camels were superior to cattle in terms of maintaining milk production. Such conclusions, however, must be tempered in light of the accuracy of verbal recall data. Overall, there is some support for these findings concerning drought impacts on livestock from the literature, but researchers also suggest that there may be significant system-specific differences in response to drought.
The marked loss of cattle and the apparent greater susceptibility of cows and immatures to starvation and disease during droughts has been variously reported for other pastoral systems in East Africa in the 1980s. Rodgers and Homewood (1986) found a net 52% loss of cattle among the Maasai in semi-arid Tanzania, a 75% loss of lactating cows and a 20% decline in calving rate. Homewood and Lewis (1987) working in semi-arid Baringo, Kenya, revealed regionally variable loss of cattle, sheep and goats during the first nine months of drought that averaged 50% overall for each species. In the next four months of drought per cent loss of remaining animals was less than one-half or one-third of that for the first period for cattle and small stock overall, and there was also evidence that some herds had begun to recover. Calving rates ranged from 70 to 90% before the drought, but declined to 64 to 77% during the drought; these relatively high rates may be related to the moderate density of stock in the study area before the drought. A case study revealed that cow mortality was 62%, with 88% for calves <1 year old (Homewood and Lewis, 1987). The percentage loss of came for a given region during the first nine months of drought was positively correlated (P<0.01) with cattle density at the start of the survey, suggesting that density-dependent factors were operating. This relationship was not found for small ruminants (Homewood and Lewis, 1987). In a study of five Turkana herd owners in very arid northwest Kenya, McCabe (1987) calculated net drought losses of 63% for cattle, 45% for camels and 55% for smallstock. Most of these animals died of starvation and disease. Calving rates declined to 24% for cattle at the peak of the drought and lactating cows were particularly vulnerable to starvation. Male cattle were depleted as a result of slaughter and sale to purchase grain. Relatively more small ruminants than cattle or camels reportedly died of starvation and this was attributed to their reduced mobility in this patchy, arid environment. Fratkin and Roth (1990) noted that camel keepers tended to lose fewer stock than keepers of cattle or small ruminants among the Ariaal of arid Kenya. Grandin et al (1989: p 256-258) reported for Maasailand in semi-arid Kenya that household cattle herds declined by over 40% on average as a result of drought and most of these deaths were caused by nutritionally mediated diseases. Cows and immatures -1 year of age died most readily while older immatures and adult males had the highest survival rates. Aside from an outbreak of Nairobi sheep disease which reduced some flocks, small ruminants were largely unaffected by drought compared to cattle and played valuable roles as food and marketable commodity (Grandin et al, 1989). The situation observed by Grandin et al (1989) seems to be similar to that for the Boran and this may be related to general similarity in the semi-arid environment and production systems between the Kenya Maasai and the Boran (see Section 4.4.2: Economic comparisons among pastoral systems).
That the cattle herds in the southern rangelands may have recovered from drought much more quickly than predicted from computer models is not surprising. Herd growth rates on the order of 10% per annum have been observed in favourable situations elsewhere (see review in Mulugeta Assefa, 1990). Assuming that cow mortality rates reported by Donaldson (1986) and Cossins and Upton (1988a) were accurate the discrepancy probably stems from two main issues. First, the models used production values for an average rainfall year in 1981-82. It is now known, however, that although 1981-82 was a time of average rainfall, it was also a time of high stocking rates and these high stocking rates probably reduced cow productivity through forage competition (see Section 7.2: A theory of local system dynamics). Therefore, actual calving rates in 1985-87 during the time of average rainfall but lower cattle density could have been markedly higher than the 75% recorded in 1981-82. Second, in post-drought periods the Boran reportedly actively trade mature males for milk cows from the southern highlands. Although these cows may be somewhat less durable over the long term than the range-adapted Boran (see Section 5.4.5: Cattle growth and implications for breed persistence), they may give the system a considerable boost when resource availability is high. It is this practice of trading for presumably inferior highland genotypes that has led to official concern about dilution of the valuable Boran breed (see Chapter 8: Synthesis and conclusions).
That drought has a greater impact on the assets of poor Borana and Gabra households compared to those of the middle and wealthy classes has been reported for other groups. Grandin et al 1989 (pp 256-258) found that the wealthy Maasai had a net loss of 40% of cattle holdings (range 18 to 60%), while the poor lost 70% (range 30 to 90%). Fratkin and Roth (1990) also noted that the 1984 drought exacerbated wealth differentiation among the Ariaal of arid Kenya. Despite large animal losses the wealthy tended to remain so while the middle class and poor became poorer. Although the need for poorer households to sell or slaughter a higher percentage of stock than the wealthy is straight forward and is because of lower per capita food production among the poor, it is less clear why the herds of the poor should have higher mortality rates. Sperling (1989) speculated that the herds of wealthier households should have lower mortality rates because the animals would be milked less intensively and thus have more body reserves to endure stress. She also noted that the wealthy can better afford veterinary care and may distribute their animals to a greater extent during drought than poorer households.
The findings of Grandin et al (1989) and Fratkin and Roth (1990) were interpreted to indicate that the strategy of attempting to maximise herd size so as to increase the likelihood that they will survive in the system after drought is valid. Although the wealthy suffer larger absolute losses compared to their poorer counterparts, they usually retained a sufficiently large nucleus herd to rebound in an efficient manner while the poor may lose enough to be pushed out. This is also a valid hypothesis for the motivations of the Boran, who reportedly desire large cattle herds (Coppock, 1992b). Their stated primary motivation for cattle accumulation, however, is to accrue social and economic status and not to mitigate drought (see Section 4.3.4.7: Marketing attitudes). Importantly, most of the Borana pastoralists have only livestock assets to buffer them from drought, in contrast to the farmers who can sell assets such as farm implements and personal effects in addition to animals when in dire need of cash (Corbett, 1988; Webb et al, 1992).
The decline in terms of trade between live animals and grain observed by researchers in the central region of the Borana Plateau equates to a 70% reduction in the monetary value of cattle, a 150% increase in the price of grain and 90% net loss in pastoral purchasing power. This may represent the worst local terms of trade on the plateau, as cattle losses in the central region were the highest reported (Cossins and Upton, 1988a: p 124).
Solomon Desta et al (nd) recorded prices for livestock products and cereals in five to seven markets throughout the southern rangelands during March and April 1984 when the drought was entering the second year. They compared these findings to pre-drought prices and found that, overall, cattle prices declined by 26 to 45% and grain prices increased from 20 to 103%. In contrast, the price of milk and butter (which by the middle of the drought were in very short supply) increased by an average of 142%. Price changes over the same period for animals were -40% (bulls), -38% (heifers), -30% (steers), -46% (COWS), -34% (goats) and -35% (sheep). Detail price changes for cereals were +86% (maize), +82% (sorghum), +42% (teff), +57% (barley) and +45% (wheat). For animals these data can be interpreted as suggesting that cows suffered the greatest relative fall in price, possibly because of their increased vulnerability to drought. No prices were reported for calves and it is likely that supply was small and there was no effective demand. Somewhat surprising was the large drop in small ruminant prices which belles the notion that they would maintain their market value during drought because they could persist better in the local environment. There were no prices reported for camels. Although some Boran probably wanted to have milk camels at this time, there may not have been sufficient demand for these animals which normally cost at least twice as much as cows (see Section 4.3.4.6: Prices). The Gabra may also have restricted the flow of camels to market. It has been reported that if a Gabra were forced to sell a camel, his peers would try to assist him so he would not have to sell, or would ensure the camel was purchased by a wealthier individual and kept within the Gabra clan network (Coppock, 1988). Thus, except for small quantities of milk and butter, the majority of pastoralists had no animal resources that did not drop markedly in value during the drought. Dairy sales may have been very important in pert-urban locations during this period as they partially substitute for live animals sales in average rainfall years (Holder and Coppock, 1992). Paradoxically milk is commonly sold in time of milk scarcity because favourable terms of trade permit the purchase of a survival ration of energy in the form of grain (see Section 4.4.10: Dairy marketing). Regional differences may have given some pastoralists an advantage in procuring cheaper sources of carbohydrate compared to grains. Households in the upper semi-arid zone near Beke Pond commonly mentioned that enset (or false banana) was cheaper than maize during the drought (D. L. Coppock, ILCA, unpublished data).
In summary, patterns observed here support the axiom that a decline in pastoral terms of trade is a consequence of drought (Toulmin, 1986: p 2; Moris, 1988: p 291). More severe drought situations have been reported elsewhere in East Africa where grain is only sporadically available and in these cases pastoralists are commonly unable to procure grain regardless of the low value of their livestock (Sperling, 1989).
As one of the first management responses to deficient rainfall, the dispersal of cattle from home-based warra herds to satellite forra herds is an attempt to expand grazing area in relation to a decline in net primary production (Donaldson, 1986: Cossins and Upton, 1988a). Herd segregation and reliance on fallback regions or drought reserves is a common traditional response to drought in African pastoral systems (McCabe, 1983, 1987; Homewood and Lewis, 1987; Moris, 1988; Grandin et al, 1989; Sperling, 1989). As on the Borana Plateau, came and camels are usually widely dispersed while small ruminants tend to be kept nearer to home areas (McCabe, 1987; Homewood and Lewis, 1987; Grandin et al, 1989). Moris (1988) noted the reduction in traditional fallback areas owing to population growth and land alienation, and the threat this poses to pastoral drought endurance and subsequent recovery. Grandin et al (1989: pp 258259) considered improved management of fallback areas in terms of tick and disease control as a research priority to reduce cattle losses during drought. The overall pattern of use by the Boran of internal fallback areas at the periphery of the plateau during 1983, and of external areas in the southern Ethiopian highlands and northern Kenya during 1984, suggests that constraints on use of fallback areas will be affected differently by internal and external forces. The use of fallback areas for settlement and unregulated grazing by the Boran is potentially dangerous as cattle herd crashes could occur if fallback areas are not "reclaimed". High population growth rates among neighbouring peoples (exceeding 3% per annum in many cases (EMA, 1988)) along with a proliferation of weapons suggests that use of external fallback regions will be more constrained in future. This loss of fallback areas may constitute the most immediate threat to growth and sustainability of the Borana system (see Chapter 8: Synthesis and conclusions).
At the household level, Donaldson (1986), Cossins and Upton (1988a), Coppock (1988) and Webb et al (1992) observed that internal adjustments by the Boran in response to restricted resources included the following tactics (1) Youths were dispersed for forra herding, which decreased food demand at encampments; (2) young children were given priority to receive milk while older youths and adults consumed more grain, meat, blood and other commodities; (3) more income was allocated to food purchases, along with attempts made to diversify income-earning activities, intensify social networking, and increase efforts to collect bush food and consume other unusual food items; (4) older adults voluntarily received restricted food rations; and (5) some older individuals emigrated to famine relief camps. At the encampment level, 27% of the households in 60 encampments moved locally in response to drought, but no mass migrations were observed (Coppock and Mulugeta Mamo, 1985). Opportunistic cultivation of maize was commonly observed as a post-drought response in 1985 to compensate for a lag in milk production (see Section: 4.4.1.1: Pastoralism and cultivation).
Most of the adjustments listed above have been observed in other pastoral systems. That the milk-dominated diet of pastoralists is subsidised with nonpastoral grain and an increased use of such ancillary pastoral foods as blood, meat and native plant material during is well known (Ellis et al, 1986; Sperling, 1989). Sperling (1989: p 269) noted that use of collected forage has declined among the Samburu in the last few decades because of loss of land, livestock pressure and purchased grain which has gradually replaced these items in the diet. Traditional knowledge of foraging has thus been lost to a large extent and taste preferences have moved towards grain. One example of the effects of livestock pressure is the use of dry dehiscent acacia fruits. Before the 1940s these fruits were a dry-season staple while today they are allocated only as a critical dry-season feed for livestock (Sperling, 1989: p 269). Dessalegn Rahmato's (1987) study of farmers in Wollo, Ethiopia, during the 1984-85 famine (cited in Corbett, 1988: p 1104-1105) noted a reduction in: (1) food variety and quality and (2) number of daily meals. Finally, the need for pastoralists to seek employment or engage in otherwise unusual economic activities during drought is also common (Campbell, 1984; Mortimore, 1987; Hay, 1986; see Section 4.4.1.1: Pastoralism and cultivation). Gifting and social networking among pastoralists is frequent and regarded as a social tactic to ameliorate risks; networking may increase among households during times of stress (Campbell, 1984; Galvin, 1985; Nestel, 1985).
There does not appear to be much literature on food allocation within pastoral households during drought. It is thus unclear whether the Borana practice of buffering young children and sacrificing elders is unusual or not. One implication of this practice, however, is that children less than five years of age may not be a suitable group to monitor in nutrition or anthropometric surveys that seek to detect indicators of famine onset or drought recovery (D. L. Coppock, ILCA, personal observation).
Human response to drought can be characterised as a hierarchy of adjustments over time. Corbett (1988: pp 1107-1108) proposed a framework for subsistence farmers which consists of three major stages:
1) Insurance stage: households first attempt to buffer themselves by selling small ruminants or other "less essential" and more readily replace able animals, reducing food demand; collecting wild foods; conducting inter-household transfers of assets and loans; increasing production of "petty commodities" for sale; migrating in search of employment; and selling personal possessions.2) Crisis stage: households then begin to dispose of productive assets which may include larger, more durable livestock such as cattle sell agricultural tools; seek credit; and initiate further reductions in food demand.
3) Distress migration stage: people embark on mass migrations in search of food. This is a stage at which numerous deaths can occur.
Based on this model and observations in 1983-84, it is hypothesised that, on average, pastoral families in Borana probably spent most of the drought in the insurance stage, with the crisis stage encountered only when households were forced to sell prime cattle or camels, commonly during the second drought year. There was no distress migration stage in Borana during 1983-84 (Webb et al, 1992). In contrast to Borana, drought impact in the Ethiopian highlands commonly appeared to elicit distress migration among farming peoples (RRC, 1985). It is likely that, simply because of a lower number and diversity of livestock holdings, poor pastoralists entered the crisis stage considerably sooner than the middle class or wealthy. A third consecutive year of drought would probably elicit a distress-migration stage in the Borana system. An analysis of the numbers of animals the various wealth classes would need to sell to endure two and three-year droughts is reported in Section 7.3.3.7: Mitigation of drought impact.
As discussed in Section 4.4.4: Traditional marketing rationale, the Boran appear to sell cattle mostly when they have an acute need for money. Their ultimate goal is not income generation but animal accumulation for status. They thus try to endure periods of stress so they do not have to make a "withdrawal" from their pool of herd capital. That households with marketable animals restrict food intake and undergo other severe hardships during drought (Webb et al, 1992) is hypothesised to be a manifestation of this behaviour. When given no other income-generating option, the Boran reportedly prefer to sell mature male cattle because the gross income received allows them to purchase food plus replacement calves; and thus attain two goals, namely commodity procurement and herd building. Because of a lower diversity and number of cattle, the poor are often forced to sell more immature cattle which permits only commodity procurement. Diversification of the traditional economy into small ruminants and cultivation, in part, reflects attempts to substitute other products and food-procurement activities for sales of cattle (see Section 4.4.4: Traditional marketing rationale).
The predominant management behaviour may thus be described as "optimistic gambling", in which the hope is that the next wet season will be good and that households can survive without having to sell cattle. The Boran do not seem to sell in anticipation of a future crisis, but instead wait until they have no other recourse. To illustrate, they appear to understand the implications of seasonal fluctuations in terms of trade of livestock for grain, but do not sell cattle at the time of year which would maximise their returns. They merely wait until they have no choice, regardless of how detrimental the terms of trade happen to be (Coppock, 1992b). Overall, this behaviour probably has implications for delaying cattle offtake during the initial stages of drought, which promotes the maintenance of higher than appropriate densities of cattle that rapidly consume forage in fallback areas and accelerate negative, density-dependent feedback on the system. These effects manifest themselves in lower milk production, poorer animal conditions and higher mortality rates (see below). That per cent loss of cattle to mortality during drought was at least three to four times higher than per cent cattle sold and slaughtered. During drought, most dead cattle are speculated to be completely lost to households in economic terms because animals on forra die far from encampments and the meat spoils rapidly. This large loss of capital assets would justify some degree of investment alternatives other than in male cattle for risk mitigation, at least for the wealthy and middle class (see Section 7.3.3.6: Cattle marketing). The "optimistic gambling" mentality also presents important constraints for interventions such as grain storage in normal rainfall years (see Section 7.3.3.7: Mitigation of drought impact). Certainly, low livestock prices are another disincentive for selling cattle during drought. Although the Boran hope to avoid cattle sales, when they sell they seek higher prices which enable them to sell fewer head over the long term (see Section 4.4.4: Traditional marketing rationale).
The scope of the famine in the southern rangelands during 1983-84 was far less than that observed within the mixed farming systems of the northern Ethiopian highlands (RRC, 1985: Webb et al, 1992). Livestock assets thus served effectively to buffer most pastoralists during the early stages of drought (Webb et al, 1992). The counter to this, however, is that purely pastoral systems probably have a longer drought-recovery period; the best situation may be an opportunistic mix of farming and pastoralism in drought-prone environments (Campbell, 1984). The size of the famine relief camps in the southern rangelands was relatively small in 1983-86, with about 12000 occupants in total for two camps near Yabelo and Mega. This may have represented 18% of the local population, but assessments are complicated because of immigrants from as far as Kenya resided in these camps (D. L. Coppock, ILCA, personal observation). Again, mortality due to famine appeared quite low at all 65 encampments studied by ILCA. This small sample may not be representative of the southern rangelands overall, however; human mortality may have been more pronounced in remote locations (B. Lindtjørn, University of Bergen, personal communication).
The apparent widespread occurrence of diet restriction and human morbidity belies the reported increase in family size due to childbirth among a high percentage of sample families at Beke Pond (Coppock, 1988). However, human reproduction under stress is not unusual. For example, conception and births commonly occur in famine relief camps (C. Toulmin, University of East Anglia, personal communication). Research is required to ascertain what social factors contribute to increased reproduction during drought, as the Boran have traditionally practiced various forms of population regulation under stress in the past (Asmarom Legesse, 1973: pp 154-155; Helland, 1980b; see Section 2.4.3: Human population growth). Such practices have been found to decrease with the decline in traditional social values that accompanies commercialization (Swift, 1977: p 276). Perhaps a related explanation deals with privileges afforded families with infants during drought. Informants reported that it was easier to obtain food relief on the Borana Plateau if a person had a young infant (D. L. Coppock, ILCA, personal observation).
There has been recent debate concerning the degree to which climate, pastoralists and their livestock cause apparent degradative trends in rangeland resources. The mainstream view has been that heavy grazing by livestock, commonly exacerbated by inappropriate development such as proliferation of water points or extensive veterinary campaigns in the absence of stimuli for animal offtake has served to cause erosion and detrimental changes in vegetation, reducing system productivity and sustainability over the long term (Lamprey, 1983). Another view is that changes in vegetation are more commonly driven by medium and long-term rainfall trends and that livestock play a relatively minor role in vegetation change; i.e. livestock are merely "along for the ride" and are also victimised by the vagaries of rainfall (Horowitz and Little, 1987; Ellis and Swift, 1988). Implicit in the first model is the assumption that livestock populations are tightly coupled to the vegetation.
These systems are regarded as equilibrial in the sense that heavy grazing by livestock can affect a negative trend in forage production, which eventually loops back to reduce animal performance and numbers. Animal performance equates to primary and thus internal interactions are important. Concepts such as carrying capacity were developed for managing such systems. Berryman (1989) argues that the defining process for equilibrial systems is the occurrence of density-induced negative feedback.
Implicit in the second model is the assumption that livestock never are able to reach densities at which they fundamentally affect vegetation. Plant production, cover and species composition are more affected by annual rainfall or other external events than livestock and vegetation. Feedback loops are thus tenuous but may change in intensity depending on sequences of higher or lower rainfall years. The system is more chaotic and nonequilibrial. Concepts such as carrying capacity are less relevant for the management of these systems because stocking rates have little influence on vegetation performance from year to year.
Behnke and Scoones (1991) attempted to synthesise the results from several papers that examined equilibrial and non-equilibrial system phenomena in African rangelands. Coppock (1987b) concluded that the Borana system appeared to be equilibrial in light of cattle population dynamics observed from 1982-1989. Loss of internal and external fullback regions for use during drought should increase density-induced negative feedback on the cattle populations. He also hypothesised that both equilibrial and non-equilibrial pastoral systems exist in East Africa. The equilibrial system in the semi-arid southern Ethiopian rangelands is fundamentally defined by a more favourable rainfall regime (i.e. 600 mm/year) in which frequency of severe drought is low enough (i.e. once in 10-20 years) compared to cattle generation time (i.e. four years). Epidemic diseases are also apparently under control. There is thus a high probability that cattle can reach population levels that ultimately make them susceptible to modest fluctuations in rainfall in dry years, resulting in competition for forage. Perennial grasses and woody plants dominate the system on fine-textured soils and their prevalence over the medium to long term is probably dependent on cattle-stocking rates, especially during episodic high-density phases of the cattle population (see Chapter 3: Vegetation dynamics and resource use and Section 7.2: A theory of local system dynamics).
In contrast, the very arid pastoral system of north-western Kenya in South Turkana has a non-equilibrial character (Ellis and Swift, 1988). Droughts may occur at a high frequency which commonly disrupts growth and survival of slow-maturing larger species such as cattle and camels. The low annual rainfall in South Turkana (350 mm) has resulted in an annual community of grasses produced from seed in a "boom or bust" mode depending, almost exclusively, on annual rainfall. Linking cattle density to annual plants production would therefore be tenuous. Even if livestock were periodically to attain high densities in South Turkana, the dominance of flat terrain and sandy soils at lower elevations would make the system relatively immune to livestock-induced erosion and degradation. The key concept is that, considering the contingencies of landscape and soil erodibility, systems with an annual rainfall of more than 450 mm on hilly, fine-textured soils with perennial vegetation should exhibit equilibrial properties; those having less rainfall on flat terrain with coarse soils and annual vegetation should be somewhat less equilibrial (Coppock, 1987b). Equilibrial systems behaviour occurs when people and livestock are spatially confined (as in Borana), essentially creating an unstable "pressure cooker".
In this chapter cattle population dynamics in response to drought were interpreted to illustrate several important interactions involving pastoral households, regional populations and ecosystem function. A lack of empirical information concerning resource allocation among households, or response of regional marketing networks meant that these important issues could not be addressed. The total picture is thus incomplete.
Any attempt to understand trends towards increased frequency of famine, poverty and environmental degradation in the southern rangelands must include consideration of the population dynamics of people in conjunction with those of cattle. That the human population may have remained stable, or even have grown slowly, during the 1983-84 drought, while there was dramatic loss of livestock assets and production resources, is a crucial finding. The sustainability of the system is utimately linked to an inappropriately high density of people and this threshold may have been exceeded within the context of our study period during 1980-91. A global synthesis that unites these concepts is provided in Section 7.2: A theory of local system dynamics.
