C U R R E N T A N T H R O P O L O G Y Volume 44, Supplement, December 2003© 2003 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved ooii-32O4/2OO3/44supp-oooi$3.oo

Cattle Raiding,Cultural Survival, andAdaptability of EastAfrican Pastoralists^

by Sandra Gray, Mary Sundal,Brandi Wiebusch,Michael A. Little,Paul W. Leslie, and Ivy L. Pike

Since the late 1970s, cattle raiding with automatic weapons hasescalated among nomadic herding societies in northern East Af-rica. We examine the impact of AK-47 raiding on the adaptabilityof Karimojong agropastoralists in northern Uganda. Most notably,raiding is linked to a loss of population resilience in Karamoja,measured in increased mortality of young children and of adultmales in their prime reproductive years and decreased female fer-tility. AK-47 raiding has acted both directly and indirectly as aDarwinian stressor in this population, compromising long-stand-ing adaptive strategies and intensifying selection pressure. Webriefly discuss similar effects of recently altered patterns of raid-ing among related Turkana pastoralists in Kenya. We then con-sider the process by which this traditional cultural institutionwas modified in the interests of preserving cultural identity. Weconclude nonetheless that cattle raiding with automatic weaponsconstitutes singularly maladaptive cultural behavior in contem-porary pastoralist societies. Indeed, it represents the singlegreatest threat to their biobehavioral resilience and ultimatelymay have profound evolutionary costs in terms of pastoralists'survival.

SANDRA GRAY is Associate Professor of Anthropology at theUniversity of Kansas |Lawrence, Kans. 66045-2110, U.S.A.[sjgray@ku.edu]), where she is also interim director of the HumanBiology Program and a member of the executive committee ofthe African Studies Resource Center.

MARY SUNDAL and BRANDI WIEBUSCH are graduate studentsin anthropology at the University of Kansas.

MICHAEL A. LITTLE is Distinguished Professor of Anthropol-ogy at Binghamton University and past president of the HumanBiology Association and the American Association of PhysicalAnthropologists.

I. We are indebted to the Karimojong women who spoke with el-oquence and honesty of the deaths of so many of their parents,husbands, siblings, and children during the terrible closing decadesof the 20th century. Their stories were united by stark themes ofviolence that overshadowed all other considerations, compelling usto examine our data in a different, darker light. We thank Rada andNeville Dyson-Hudson and several anonymous reviewers for theircomments on an earlier draft of this paper. This research was fundedby Wenner-Gren Grant # 6276, National Geographic Society Grant

PAUL w. LESLIE is Professor of Anthropology at the Universityof North Garolina, Ghapel Hill, and a demographer at its Popula-tion Studies Center.

IVY L. PIKE is Assistant Professor of Anthropology at the OhioState University.

The present paper was submitted 4 in 02 and accepted 6 v 03.

Beginning in the last decade of the colonial era (1952-62),the Karamoja region of northeastern Uganda was the tar-get of a series of ill-conceived economic developmentprograms that interfered with long-standing strategies forherd management, land use, and resource allocation ina highly unpredictahle and seasonally arid savannaecosystem (Mamdani, Kasoma, and Katende 1992). Theseprojects effectively uncoupled indigenous livestock andrangeland hushandry, and the resulting overgrazing andenvironmental degradation were blamed for an ecologi-cal catastrophe in the late 1970s that reportedly left thehuman population increasingly vulnerable to food short-ages and disease outbreaks. A few reports from the dis-trict in the ensuing decades suggested that Karimojongpastoralists led a far more precarious existence at theend of the 20th century than they had 50 years earlier.Therefore, a study was initiated by Gray in 1998 to doc-ument changes in the structure and functioning of anAfrican pastoralist ecosystem as a result of economicdevelopment and modernization and to quantify the ef-fects of this transformation on human biobehavioraladaptability.

During the year-long field season, however, a modernform of cattle raiding with automatic weapons—whatwe designate here as "AK-47 raids"—appeared to be thechief impediment to established adaptive behavioral re-sponses of Karimojong pastoralists to environmentalstress. This perception was validated as our biologicaland demographic data took shape during subsequentanalyses. Indeed, their adoption of this new form of raid-ing appears to have caught the Karimojong on the hornsof a dilemma of Darwinian proportions: while AK-47raiding may have augmented their collective cattlewealth and thus ensured their cultural survival as pas-toralists, at the same time it gave rise to increasinglyunequal access to the herds within the Karimojong pop-ulation and severely constrained the ability of the mostvulnerable segment of the population—women andyoung children—to buffer themselves from abrupt fluc-tuations in environmental quality. Since the 1970s,armed cattle raiding has exacted a mounting toll on thehuman population in terms of mortality and morbidityand impaired the capacity of the Karimojong to ensurethe health and survival of their children. In Darwinianterms, human agency in the form of collective violenceappears to have acted as the principal agent of selectionamong Karimojong pastoralists in the second half of the

# 6048-97, and the General Research Fund of the University ofKansas.

S3

S4 I CURRENT ANTHROPOLOGY Volume 44, Supplement, December 2003

Sudan %tom

Toposa

Kotido District(NorthernKaramoja)

Mmvtp District(SauthmiKaramoja)

Ethiopia

International boundaries

District boundaries

Dodoth Karimojong duster

/^ ^ Approximate boundaries ofthe Karimojong home areain 1998-99

FIG. I. Karamoja, showing approximate locations of the homelands of the Karimojong-speaking cluster of theEastern Nilotes (Atekerj (Gray 2000).

20th century. In this paper we examine its effects onexplicitly Darwinian parameters, mortality and fertility.

Secondarily, we are concerned with the dual role ofculture as adaptation and as stressor. Eschewing for themoment the dehate over hiological and environmentaldeterminism versus historical and sociocultural process,we assume at the outset that cattle raiding arose as acultural institution in East Africa at the interface of hi-ological, environmental, and sociopolitical imperatives.Moreover, we suspect that it was at one time adaptive,an integral component of a profoundly contingent, flex-ihle, and successful suhsistence strategy—one that hashad remarkahle staying power in the harsh ecosystemsof the Rift Valley (Gifford-Gonzalez 1998, Marshall 1990,McDonald 1998). Ironically, hy the 1990s, the same in-stitution may have effectively ahrogated the value of spe-cialized pastoralism as an adaptive strategy in these no-toriously unpredictahle savanna systems. Indeed, in itspresent form cattle raiding appears to he suhlimely mal-adaptive. In the course of our discussion here, we alsoconsider this premise.

Background

The Karimojong are memhers of a group of specializedpastoralist societies identified collectively as the Kari-mojong Cluster of the Eastern Nilotes (Gulliver 1955) or,alternately, as the Ateker, or "clan," cluster (Lamphear1992). Currently they include the Jiye and Toposa, insouthern Sudan, the Nyangatom, in southern Sudan andEthiopia, the Turkana, in northwestern Kenya, and theIteso, Karimojong, Jie, and Dodoth, in northwesternUganda (fig. i). As far as can he determined on the hasisof scant archeological data and oral traditions, they are alldescendants of Sudanese peoples who migrated southsome 500 years ago to a location in northern Karamojathat they recognize as their ancestral homeland. A seconddispersal, beginning some 300 years ago, gave rise to theirmodern territorial configuration (Lamphear 1976).

Investigation of the impact of economic developmenton the Karimojong was initiated as a direct outgrowthof an earlier study of human ecology in Karamoja (N.Dyson-Hudson 1966) and a later, long-term study of hu-

GRAY ET AL. Cattle Raiding, Cultural Survival, and Adaptability \ S5

man adaptation among nomadic Turkana (the South Tur-kana Ecosystem Project, STEP [Little and Leslie 1999]).The Karimojong and the Turkana diverged from a com-mon ancestral population sometime in the mid-1800s(Barber 1968, Lamphear 1976), and as recently as the1960s the two populations still demonstrated markedcultural and biological affinities reflecting their sharedancestry (N. Dyson-Hudson 1966; R. Dyson-Hudsonr96o; Gulliver 1951, 1955; Lamphear 1976). Adults inboth populations exhibit the classic Nilotic physique(Roberts and Bainbridge 1963), combining tall stature andextreme thinness, but earlier investigators found neithergroup to be acutely at risk of malnutrition and their over-all health remarkably good. In a pioneering study carriedout in the early 1960s, Jelliffe et al. (1964) suggested thatthe growth, nutritional status, and health of Karimojongchildren at that time reflected successful adaptation toa harsh environment and nomadic lifestyle. Their con-clusions would eventually be validated by similar flnd-ings in South Turkana, where between 1979 and 1995members of the STEP team documented a human pop-ulation whose biological and cultural adaptations wereprototypes of human biobehavioral plasticity (Little andLeslie 1999).

During the period of the Turkana study, however, avery different picture began to emerge from Karamoja.The Karimojong suffered a devastating famine during asevere drought in 1980 (Dodge and Wiebe 1985:127-89),during which child mortality was upwards of 60% (Biel-lik and Henderson 1981, 1985) and female fertility re-portedly dropped to zero (Alnwick 1981, Akol and Grayn.d.). The drought was equally severe in Turkana Dis-trict, but its effects on the human population there weredamped relative to its effects in Karamoja. Furthermore,whereas the Turkana steadily recouped their livestocklosses from the drought during several years of good rain-fall in the middle to late 1980s, the same period in Kar-amoja was characterized by chronic food shortages anda continuous threat of famine,- reportedly, food aid hasbeen required there in every year since 1980 (Cisternino1985). A series of health and nutrition surveys conductedamong the Karimojong in the 1980s and 1990s reportedmalaria in epidemic proportions (Devlin 1998), recurringoutbreaks of measles, cholera, and meningitis, and wast-ing and stunting among Karimojong children symptom-atic of chronic moderate-to-acute nutritional stress(Alnwick 1981, Mears and Riley 1987, UNICEF r993;see also Moroto Diocese 1995). In critical indicators ofpopulation resilience, the Karimojong appeared to haveveered sharply away from a more positive situation a fewdecades earlier.

At the outset of her research. Gray shared the viewthat attributed this loss of population resilience directlyto unsustainable economic development in Karamoja inthe form of agricultural intensiflcation, coercive reset-tlement of nomads, and concomitant restrictions on pas-toralists' access to large areas of traditional grazing land(Mamdani, Kasoma, and Katende 1992, Okudi r992). Byreducing the mobility of people and herds in response toenvironmental stress, these initiatives promoted over-

grazing and undermined local strategies that facilitatedrecovery of the rangelands during the intervals betweendroughts. Effects of development policies appeared tohave precipitated an ecological crisis (Wilson 1962, citedin Mamdani, Kasoma, and Katende 1992) that culmi-nated in the collapse of the cattle herds during thedrought of 1979-81 (Cisternino 1985, Gray 2000, Gray,Leslie, and Akol 2002, Mamdani, Kasoma, and Katende1992). As a consequence, many Karimojong had becomedependent on a profoundly unreliable agricultural sub-sistence, heightening their vulnerability to undernutri-tion in the dry season and to famine during droughts.

Thus, the situation of the Karimojong in the 1990sappeared to be a textbook example of the impact of un-sustainable economic development on a pastoralist sys-tem (Gray, Leslie, and Akol 2002). Given the availabilityof comparative data for closely related pastoralists inTurkana, where at that time economic development hadbeen limited, we were presented with a natural labora-tory in which to quantify effects of economic develop-ment and modernization on human health, adaptability,and fltness in a traditional society (Garruto et al. 1999).

Guns and Cattle in Karamoja, 1998-99

The Karimojong have traditionally engaged in opportun-istic sorghum cultivation in years of adequate rainfall(approximately one out of every three years), and theymaintain semipermanent homesteads (ngireria, sing, ere]in the agricultural zone to and from which adult males,herd boys, and unmarried girls migrate seasonally withthe herds. Women and young children remain in thehomesteads year-round when the sorghum harvest isgood but typically follow the herds during drought years.The research plan for the flrst field season in Karamojacalled for the collection of baseline data on fertility andmortality, health and nutritional status, and childgrowth in a random sample of homesteads in each ofthree major territorial sections of the Karimojong: Pian,Bokora, and Matheniko (flg. 2). The goal was to collectreproductive histories for all adult Karimojong womenin each homestead sampled and to conduct anthropo-metric examinations of the women, their husbands, andtheir children. Fieldwork was to be restricted to the ag-ricultural homesteads because the focus of the study wason women and children and because most people wereassumed to have very few livestock as a result of lossessuffered in the 1980s and early 1990s. Cattle migrationswere expected to be relatively small-scale if they oc-curred at all.

In August 1998, when the study began, the Karimojongwere heavily armed: in each of the 33 homesteads sam-pled, at least one adult male was in possession of one ormore automatic weapons. In the majority of homesteads,all of the younger herders were armed. The supply ofguns in the district seemed to be bottomless, and thethriving black market in automatic weapons was linkedinextricably to the acquisition of cattle (Gray 2000, Mir-zeler and Young 2000; see also Hutchinson 1996:109),

S6 I CURRENT ANTHROPOLOGY Volume 44, Supplement, December 2003

acholirkfina

Matheniko

tCSO Non-Karimojong involved in raiding asvictims, aggressors or both

Pian Agricultural zones ofthe three majorKarimojong territorial sections

Administrative and trading centers

Sites of research in 1998-99

SOROTI Neighboring districts in Uganda

^ ^ Contested grazing areas and locales^ ofraiding in 1998-99

FIG. 2. Approximate locations of the home areas, cultivation zones, and cattle camps of the three major terri-torial sections of the Karimojong, ipp8—pp (Gray 2000). Also shown are contested grazing areas in which inter-ethnic raiding was concentrated in ipp8—pp and the groups involved.

which, contrary to initial assumptions, still shaped everyaspect of Karimojong society and culture in 1998 just asthey had in the 1950s (N. Dyson-Hudson 1966). In fact,in 1998-99 the Karimojong were in possession of sizahlemixed herds of cattle and small stock, hut these appearedto change hands regularly as a result of AK-47 raids, themost frequent and most lethal form of armed violenceat that time. Raids were on a massive scale, involvinglarge groups of marauders of different homestead andterritorial affiliations who had formed alliances to as-sault other clusters of homesteads or cattle camps. As adefensive measure against these raids, herders had hegunto migrate collectively: although the great mass of ani-mals presented a larger target, the greater numher ofarmed herders enhanced their ahility to fend off attack-ers. On the first day of travel to homesteads in Bokora,the research team found itself in the middle of the rainy-

season migration, and its vehicle was surrounded by tensof thousands of livestock migrating toward the agricul-tural zone. The dry-season migrations were of the samemagnitude.

In addition to cattle raids, isolated attacks on home-steads, vehicles, and individuals also were common. In-deed, guns were used for resolving disputes at every levelof Karimojong society, from the homestead to the ter-ritorial section. Assaults were often directed at non-Kar-imojong, especially neighboring Teso, Pokot, and Tepeth,hut they were just as likely to he intraethnic, pittingmemhers of the Matheniko territorial section againstmemhers of the Pian or Bokora sections. Much of theviolence in 1998-99 was hetween sections, and casualtravel from one home area to another was risky hecauseit required travel through targeted grazing areas or onlong and especially dangerous roads. Fieldwork in Pian

GRAY ET AL. Cattle Raiding, Cultural Survival, and Adaptability \ S7

was especially prohlematic, involving travel along thenotoriously insecure trunk road that extended southfrom Lorwengedwat to Mhale, where vehicle hijackingshy armed Karimojong, Pokot, and Tepeth bandits werefrequently reported. In the end, work was concentratedin four clusters of agricultural homesteads in Matheniko(Nadunget) and Bokora (Lokopo and Apeitolim). Theteam also made a brief visit to cattle camps in north-western Bokora but, given the logistics and the risks in-volved, was unahle to incorporate a camp componentinto the study in any systematic manner.

Results of the 1998-99 Karamoja Study

In spite of the instahility in the region, hetween Augustand March Gray and her team interviewed over 300 Kar-imojong women and 33 senior herd owners and carriedout anthropometric examinations and nutritional as-sessments of the women, their hushands and children,and all other relatives in residence. In all, nearly 900individuals participated in the study. Here we summa-rize results for child growth (Wiebusch 2002), adult andsubadult mortality (Sundal 2002), and female fertility(Gray and Akol 2000, Gray and Pike 1998) and comparethem with our findings in Turkana.

CROSS-SECTIONAL GROWTH OF KARIMOJONGCHILDREN, 1 9 9 8 - 9 9

Mean height and weight of Karimojong children youngerthan age five in 1998-99 were helow the National Genterfor Health Statistics third percentiles, and these childrenwere on average shorter and lighter than hoth their coun-terparts from the 1960s and the Turkana children mea-sured in the 1980s, who tracked more closely to the tenthpercentile (fig. 3). By age five, however, mean height ofKarimojong children in the 1990s was virtually indistin-guishahle from that of Turkana children, and they wereslightly heavier than those children. Inasmuch as theseare cross-sectional data, it is highly unlikely that con-vergence of growth curves for Turkana and Karimojongchildren in middle childhood is a result of catch-upgrowth (Tanner 1986) in the Ugandan group. Gray (2002)has argued that their apparent acceleration in middlechildhood is in fact a measure of the intensity of selec-tion pressure operating on Karimojong children in in-fancy and early childhood: by age five, the smallest Kar-imojong children have died. Since growth in this periodis a measure of a child's experience of environmentalstress, we assume that the smallest children died hecausethey were most frequently exposed and least resistant toinfection and malnutrition. Surviving children werelarger and presumahly healthier because they were moreresistant to nutritional and disease insults. The ahsenceof a similar correction in the Turkana data suggests lowermortality of young Turkana children, presumahly he-cause they were either less exposed or less susceptihleto these stresses. A later deceleration in growth of heightand weight of Turkana children, after age four, prohahly

1 2 0 -

1 0 0 -

80

60-

40 -

20

Weight

Karimojong, 1960sTurkanaKarimojong, 1998NCHS 3rdNCHS 10tli

20

- 15

- 10

- -5

ICO

3 4 5

Age, years

FIG. 3. Comparison of growth in height (recumbentlength) and weight of Karimojong children in igpS-ppstudy, Turkana children (Little, Galvin, and Mugambirp83. Little and Gray rppo), and Karimojong childrenmeasured in the ip6os (Rutishauser and Whitehead1969). The 1969 Karimojong sample was restricted tochildren age five years and younger. NCHS third andtenth percentiles (Kuczmarski et al. 2000) are shownas the average of the curves for boys and girls.

represents a true adjustment in velocity of growth tochanges in diet and activity that occur after weaning inthis population (Gray 1999). These hypotheses are givensome support hy patterns of infant and child mortalityin the two populations.

SUBADULT MORTALITY IN KARAMOJA, 194O-99

In marked contrast to the Turkana (Leslie et al. 1999^),Karimojong women interviewed in the late 1990s weresurprisingly forthcoming regarding their children whohad died. Their willingness to provide this informationmay be a function of their shared experience of childdeaths, inasmuch as 64% of these women reported atleast one such death. Exclusion of prima gravidas andwomen experiencing secondary sterility increased theproportion who had experienced the death of a child to70%; 45% cited two deaths, and one-quarter reportedthat three or more of their livehorn children had died.

Of 1,457 live births of Karimojong women hetween1940 and 1999, 348 (24%) died hefore the age of five and393 (2.7%) before age ten (estimates up to age ten are19% and 28% for females and males, respectively [Sun-dal 2002]). Probabilities of dying between consecutiveage classes {qj for subadults from i960 to 1999 are pre-sented in table i for the comhined sample and for Math-eniko and Bokora sections. Availahle estimates for Tur-

S8 I CURRENT ANTHROPOLOGY Volume 44, Supplement, December 2003

kana also are shown. The Turkana estimates have beencorrected for underreporting (because women were re-luctant to discuss child deaths or, in the case of elderlyinformants, because the deaths had occurred so long agothat women failed to recall them), whereas the Kari-mojong data are uncorrected estimates hased on the di-rect reports of informants. They therefore may be con-servative, even given the willingness of Karimojongwomen to discuss child deaths, because of the faultyrecollections of the oldest women. Nevertheless, evenwithout corrections for unreported deaths, the lowestestimates of mortality of Karimojong children in everydecade from i960 to 1990 are higher than the highestcorrected estimates for Turkana children. For the threedecades hetween i960 and 1990, combined mortalityrisks in infancy d̂ o) in Karamoja range from approxi-mately 13% to 20%. The prohahility of dying hetweenages one and five years (4̂ 1) ranges from 13% to 18%.Mortality varies hy decade, with the highest mortalityreported in the 1960s and 1970s. In the two decades fromi960 to 1980, the probability that Karimojong childrenwould die hefore age five was 30% (5̂ 0) ^^d before theage of ten (,o(Jo) nearly 40%. For all age-classes and inevery decade except the 1990s, child mortality in Kara-moja exceeded that recorded for nomadic Turkana, forwhom corrected estimates of q^ were 13% to 21 % and20%, respectively, for under-fives and under-tens. More-over, on a continent in which infant and child mortalityrates exceed those of most other regions of the world(Barton and Wamai 1994:34), the rates for Karamoja areequal to some of the highest.

ADULT MORTALITY, 192O—99

In addition to information on the deaths of their children,women were asked to provide mortality data for theirmothers, fathers, sihlings, husbands, and co-wives. In all.

0,8 - Females, post-1940Females, pre-1940Males, post-1940Males, pre-1940

0.0 -

0 10 20 30 40 50 60 70 80

Age at the beginning of the decade in which death occurred

FIG. 4. The probability of dying between ages x and x+ n („q J among Karimojong males and females bornbefore and after rp4o. X is presented as s-year inter-vals between age 4 and 19 years and as ro-year inter-vals beginning at age 20. „<\^ is calculated as a propor-tion of the number of individuals surviving to thebeginning of each interval. To control for censoringbias, only those who were old enough to have livedthrough a specified interval were used in the calcula-tions for the post-r940 cohort.

Gray obtained data for a total of 2,235 individuals bornbetween i860 and 1999. Figure 4 compares age-specificmortality risks [q^] for males and females in this sample.The data are presented for two cohorts: those horn heforeand those horn after 1940. Birth and death years for mostadults are approximations (early 1940s, middle 1940s,and so forth); therefore, after age 19, q^ was calculatedfor ten-year intervals. Because the post-1940 cohort issuhject to censoring hias at each age x, our computations

TABLE I

Probability of Dying between Ages x and x + n(nq,, j in the Karimojong Sample for the Decades from r96o to1999 for Bokora and Matheniko Territorial Sections and Comparative Data for Turkana (Leslie, Dyson-Hudson,and Pry '

Sample

1990-99

BokoraMathenikoCombined

1980-89

BokoraMathenikoComhined

1970-79BokoraMathenikoCombined

1960-69BokoraMathenikoCombined

Turkana

Number of Births

339157496

276

117389

15890

248

119

77196

-

.106

•191•133

.129

.128

.129

.146

.167

•153

•193.208.199.09-. 14

.040

.102

.058

.105

.176

.127

.185

.173

.iSi

.114

.180

.140-

0

.009

.002

.028

.048

.034

.091,097.093

.094

.060

.081-

5^10

-

_

-

0

.013•003

.040

.036

.038

.026

.021

.024

-

5<?0

.142

.274

.183

.221

.282

.240

•304.311.310

.286

.351•311• I 3 - 2 1

loiJo

.142

.280

.185

.243

.316

.265

•367•378•371

•353.390

•367.200

_

_

_

.028

.060•037

.127

.129

.128

.118

.080

.104-

GRAY ET AL. Cattle Raiding, Cultural Survival, and Adaptability | S9

TABLE 2Leading Causes of Subadult and Adult Deaths (Percentage) in Moroto and Bokora Territorial Sections, Moroto,Uganda, 1860-1999, Determined from Maternal Recall and Verbal Autopsy during

Cause of Death

Acute respiratory infectionDiarrhea"MalariaFamineMeaslesMeningitis or hepatitisTuherculosis"Chest"MaternalOld ageViolenceSuicide

Infants|N = 209)

141 2

155

176r

_-_--

Age 1-4[N = 121)

8i 317

525

9r

_----

Age 5-9{N = 60)

1 0

2 0

68

2 2

66

---2"

-

Age 10-19(N = 21)

6-1 2-

6-—--

19

Adult Females(N = 172)

2 1

2

1 0-

4886

1 0

34

Adult Males{N = 268)

13I

1 2-

34

1 0-

628

2

SOURCE: Sundal (2002).NOTE: Deaths of "unknown" etiology (approximately 25% of all deaths reported) are not included in these calculations. A number ofdeaths were from miscellaneous causes. Among infants, 10 of these were neonatal or perinatal deaths. Among subadults, 19 deathswere attrihuted to nonspecific sickness [edeke], including 9 from "evil eye." Among adults, 30 deaths were attributed to nonspecificsickness and 13 resulted from a variety of infections and abscesses."Includes diarrheal deaths from all causes.""Combined percentage for ages 1-9 years.

for that group are based only on those individuals oldenough to have cotnpleted each age-interval (who v^ereat risk of dying or who died during each ten-year period).Individuals horn hefore 1940 were either older inform-ants or informants' parents; thus, we have no data forsuhadult deaths in this cohort. Conversely, the post-1940cohort provides ample data on mortality in childhoodand in early to middle adulthood, hut data for individualsolder than age 59 are lacking. However, the overlap ofthe two cohorts in their early and middle adult decades(age 20 to 50 years) allows cross-cohort comparison ofmortality risks in those age-classes.

Mortality of females in the post-1940 cohort is slightlyhigher hetween age 20 and 40 years than among femaleshorn hefore 1940 (survivorship to age 50 of females whoattained age 20 is 84% and 80% for the pre- and post-1940 cohorts, respectively), hut mortality differences he-tween the two male cohorts are dramatic. In their fourthand fifth decades, the prohahility of dying was .34 and.28 among males in the younger cohort, whereas amongmales in the earlier cohort the risks at the same ageswere .04 and .08, respectively. Of those males in the post-r94o cohort who attained age 20, only 45% survived toage 50, compared with 87% of males in the pre-1940cohort, for whom a steep increase in q^ occurred afterage 59.

Human mortality is typically highest at the oldest andyoungest ages in all populations and is lowest in thesecond decade; in the third and fourth decades, mortalitytends to he slightly higher than hetween ages 10 and 20,and the pattern is suhject to variation among populations(Coale and Demeny 1983:11). The prohahility of dyingalso tends to he higher among males than among femalesat all ages. However, the exceedingly high prohahility ofdying hetween ages 30 and 49 among Karimojong males

horn after 1940 is striking; furthermore, sex differencesohserved in the Karimojong sample are far greater thanwhat is generally ohserved (Coale and Demeny 1983).

We do not have comparahle adult data for the Turkana,hut the indirect estimate of survivorship to the middleof the third decade (to age 25) in that population is 81%(Leslie, Dyson-Hudson, and Fry 1999), compared with78% and 70% among females and males, respectively,in the post-1940 Karimojong sample. Mortality thus ap-pears to he higher in their first two decades in the Ugan-dan group among hoth males and females. Again, thedifference is pronounced for Karimojong males.

CAUSES OF MORTALITY IN KARAMOTA (194O-99)AND TURKANA

Tahle 2 summarizes the leading causes of suhadult andadult mortality in Karimoja from 1940 to 1999, hased onmaternal recall and verhal autopsies conducted duringinterviews. Percentages are estimates hased on the sam-ple of individuals for whom a cause of death was iden-tified, approximately 75% of the total reported deaths.We assume that these estimates are representative ofoverall morhidity patterns in this population. Measles[puuru], diarrhea, acute respiratory infections, and ma-laria were the principal causes of child deaths up to age10 years (Sundal 2002). Famine (including akoro, "hun-ger," and lobute, "swelling") accounted for 10% of allrecalled deaths of children older than age 5 years. Only21 deaths were reported of suhadults aged 10 to 19, hutmost individuals in this age-group died of causes otherthan sickness, including famine (three) and violence(four).

Adult females in this sample died of diarrheal disease,famine, and a variety of other illnesses, including tu-

Sio I CURRENT ANTHROPOLOGY Volume 44, Supplement, December 2003

berculosis and an ambiguous ailment referred to simplyas "chest," meningitis, and a number of ailments asso-ciated witb pain in tbe head and gut (Sundal 2002:104).Adult males died from a host of similar diseases, butnoninfectious causes accounted for 44% of all maledeaths, including famine (12%), violence (28%), and su-icide and accidents combined (4%). Only 6% of men and10% of women were identified as having died of causestypically associated with old age.

The leading causes of child morbidity among nomadicTurkana are consistent witb tbe findings in Karamoja(Sbelley 1985, Brainard 1991, Shell-Duncan, Shelley, andLeslie 1999), refiective of a common pattern throughoutsub-Sabaran Africa (Barton and Wamai 1994). On tbe ba-sis of the results of antbropometric examinations, bow-ever, the overall nutrition and health status of Karimo-jong infants and toddlers in the late 1990s was poor incomparison witb tbeir Turkana counterparts as well aswitb Karimojong cbildren examined in tbe 1960s (Ru-tisbauser and Wbitebead 1969; Gray 1998, 1999). Tbiswas a puzzling finding inasmuch as the three years pre-ceding the study had seen abundant rainfall in Karamoja,good-to-excellent sorgbum barvests, and a steady in-crease in the size of the herds. Moreover, all of the Kar-imojong homesteads visited during the field season werewithin 5 km or less of a hospital, health clinic, or dis-pensary or were visited monthly by a mobile clinic op-erating out of St. Kizito's Hospital in Bokora (reputed tobe one of the best hospitals in Uganda). In addition toits mobile clinics, St. Kizito's operates an ambulanceservice for rural bomesteads and antenatal and cbild im-munization programs (~50% of Bokora children in thesample had received at least one immunization [Wie-busch 2002]).

In contrast, in the 1980s all functioning health unitsin Turkana were located in settlements that were onaverage 10 km or more distant from tbe cattle camps oftbe nomadic population. Nearly 70% of nomadic home-steads were outside of a radius of 20 km from a healthcare facility. The few mobile clinics in tbe district werefor tbe most part out of operation because of inadequatemaintenance of vebicles (TRDP 1989). Thus, althoughTurkana District boasted four bospitals, eight health cen-ters, and 35 dispensaries in the 1980s, their utilizationby pastoralists was minimal. As a consequence, altbougbimmunization and bealtb coverage of women and cbil-dren in Turkana settlements may bave paralleled cov-erage in Bokora, it bad met witb limited success in thenomadic sector (TRDP 1989).

One notable difference in patterns of adult morbidityin tbe two populations emerges from the comparison ofoutpatient reports for Karamoja (Devlin 1998) and Tur-kana (TRDP 1989). References to trauma are relativelyscarce in the Turkana data, whereas tbey figured promi-nently among the causes of outpatient visits and adultdeaths in the Karimojong sample (Devlin 1998:65); amongthe latter, trauma was frequently a consequence of gun-shot wounds.

FERTILITY OF KARIMOJONG WOMEN

Our preliminary estimate of completed family sizeamong postreproductive Karimojong women inter-viewed in 1998-99 is 6.4. This approximates estimatesfor nomadic Turkana women (Campbell et al. 1999, Les-lie et al. i999fl). However, fertility appears to bave de-creased among postreproductive women born after 1940in comparison witb women born prior to that year. Me-dian completed family size of women born after 1940was 6, compared witb 7 for women in tbe older cobort(p = .004, Kruskal-Wallis test).

Approximately 40% of Karimojong women inter-viewed in 1998-99 reported at least one failed pregnancy:we include in this category all miscarriages and still-birtbs as well as cbildren who died perinatally (beforethey were given a name, which typically occurs within24 hours after birth). Fifteen percent of the women hadexperienced two or more pregnancy losses. Tbe propor-tion of failed pregnancies was 11% among postmeno-pausal women in tbe sample (tbe average number of preg-nancy failures was .8), and tbe proportion and numberof pregnancy failures did not vary significantly by cobort.Estimates of failed pregnancies are markedly higher thancomparable estimates for nomadic Turkana women(Campbell et al. 1999:339; Brainard i99r; Leslie et aL

Mean age at the birtb of their last child was youngerby one year among postmenopausal Karimojong womenthan among nomadic Turkana (39 years versus 40 years[Leslie et al. i999a[), and for this variable significant in-tercohort differences were observed in tbe Karimojongsample. Termination of reproduction was tbree years ear-lier among Karimojong interviewees born after 1940 tbanamong tbose born prior to tbat year (37.4 years comparedwitb 41 years, t = -3.75, p = .0003). The range in ageat termination of childbearing was 23-47 years in theformer group, compared with 27-48 years in the oldercohort, suggesting that some of tbe variation is ac-counted for by the greater risk of terminating reproduc-tion in the first decade of the effective reproductive spanin the younger cohort. Indeed, the majority of reproduc-tive pathologies recorded in this study were amongwomen born after 1940. Tbese included tbree cases ofbiological termination of reproduction prior to age 30,all in women wbo reported three or fewer pregnancies;five probable cases of secondary sterility (premenopausalwomen for whom more than 11 years had lapsed sincethe birth of their last child; for tbree of tbem, over 20years had lapsed) and two unequivocal cases (one in awoman who was postmenopausal in 1998-99 and theother in a woman in her mid-2os who had been diagnosedclinically after her only pregnancy ended in miscarriage).In the pre-1940 cohort, cessation of reproduction priorto age 30 was ascertained reliably for only one woman,who reported two live births and whose last two preg-nancies had ended in a stillbirth and a miscarriage. Oneother woman in this cobort was barren, and no womenreported experiencing secondary sterility. In all, 12 (4%)of 302 women in the study had experienced premature

GRAY ET AL. Cattle Raiding, Cultural Survival, and Adaptability \ Sii

hiological cessation of reproduction, and io of these werein the post-1940 cohort. If we include in the sample ofpostreproductive women the three women who had notconceived in over 20 years since the hirth of their firstchild, early termination of reproduction reaches 12% inthis group (8 of 66), compared with 5% (2 of 41) in thepre-1940 cohort.

We cannot discount the possihility that women hornhefore 1940 who experienced impaired fecundity weremore likely to have died or to have migrated and thusare not represented in our sample. Considered alongsidemortality trends in Karamoja in the 1960s and 1970s,however, lowered fertility of the post-1940 cohort isprohahly a real demographic phenomenon. We proposea causal link hetween these population trends and anescalation in the frequency and ferocity of Karimojongcattle raids. The effects of raiding were mediated hy apurposeful modification of the sociopolitical structure ofKarimojong pastoralism, a change that undermined hu-man hehavioral responsiveness and in effect intensifiedselection pressure. In the next sections we examine thepersonal experience of this transformation and descriheits historical origins and chronology. We then track itsimpact on the Karimojong population in relation to thetrends we have identified.

The Experience of Violence in Karamoja

DEATH AND REMEMBRANCE

The majority of female informants who participated inthe 1998-99 study were nonliterate, possessed only avague understanding of the Western calendar, and keptno records of vital events. Instead, the timing of hirths,deaths, and other milestones in women's reproductivehistories was estimated from a calendar of local eventscompiled hy Akol and Gray (n.d.) during the course ofthe field season. A response to the question of when achild was horn or died typically hegan with a woman'srecollection of what she was doing or where she was atthe time of the event, information which was used firstto locate her experience in relation to the local calendarand then to calihrate it with Gregorian time. Here wesummarize three of these accounts.

Born in the mid-1940s, the first informant entered herreproductive years in the mid-1960s. Of her nine live-horn children, only three survived in 1998. She herselfwas one of ten children, two of whom had died—oneduring childhood and the other as he fled west towardTeso with his wife and children to escape famine (proh-ahly in 1968). The informant recalled:

All [of her dead hrother's family] were killed there,hy Iteso. My father died of hunger during famine. Hehad three official wives and one inherited wife. Hissecond wife also died during the famine, in Loriki-tae. The inherited woman died of cholera, during thefamine. Her son was a hig hoy when he was killedhy Jie in the same year, at Kamarumor. My mother

. . . is the first wife and is still living. I had ninechildren and one miscarriage. . . . I left my first hus-hand, hecause five of the six children hy him died.This hride price never was paid in full, hecause wemigrated to Jie hefore it was completed: only thechild price [ekucul] was paid. Two children died ofmeasles in the same year: one is huried on thegrounds of St. Kizito's hospital. One child, a toddler,was scaring hirds in the garden, and he fell from theplatform and was killed. Three others died in earlyinfancy; all three were twins. One fell sick with fe-ver,- we took her to [St. Kizito's Hospital in[ Ma-thany. She was given medicine and injected, thensent home. That night the condition worsened andthe child died.

The second informant was also horn in the 1940s andwas one of eight children, four of whom survived in 1998.Of her six live-horn children, three were living:

One son died when he was ahout ten years old. Iwas hreastfeeding that child when cattle were seizedat Angoro, when they were giving out relief food[i967-68[. He was killed hy Bokora during the raidon the kraals at Naitae, heyond Rupa. . . . My father. . . was killed in a raid hy the Turkana. I was agirl—he was middle-aged. The woman [her mother[died of diarrhea, as an old woman.

In her late 40s, the third informant had no living chil-dren. Three had died in childhood and one as a neonate;another was stillhorn. She was one of four children, oneof whom was dead hy violence. Her parents and hushandwere all dead:

My hushand . . . was killed in a horder skirmish hy"rehels" [prohahly Acoli] in the time of Ohote, dur-ing the trouhle with Kenya [the early i98os]. My fa-ther . . . was a young man when he was killed inthe kraals, hy Pokot: this was around Amudat, inthe south [Pian[. My mother died of swelling [ofhunger, lobute]. This was sometime after Akoro [thefamine of 1980]: when they hurned Kangole, ourheads still were shaved [in mourning for hermother's death]. You mourn your parents a longtime, even now. Loneliness makes me suicidal, hutGod told me it wasn't time. Satan tempted me, hutGod is there and saved me even as I took medicineto die. In quarrels, they [neighhor women] ask me,"How many children do you have?" and I tell them,"Oh, you ask God!"

These three accounts are a sampling from over 300similar stories recorded during the field season. The cen-tral motifs of these recollections and of the local calendaritself are famine, sickness, and death, linked hy runningthemes of violence and flight. The summary that followsis informed hoth hy this local perspective and hy a morefamiliar sequence of historical, political, and environ-mental events. Tahle 3, an ahbreviated version of theevent calendar, summarizes the local perspective on theKarimojong experience in the late 20th century, when

Si2 I CURRENT ANTHROPOLOGY Volume 44, Supplement, December 2003

TABLE 3Critical Events in Karamoja from 1940 to1998-99

and Events Commonly Associated with Them by Informants in

Date and Name of Event Description

1940s, Lorengalaga, "marriage bands rusted"

Nataparin, "dams"

Ngaduruko, "thorn trees"

Maring, "hysteria"

1950s, Lotira, "the sun was stubborn"Apule, "where the Karimojong began"

Ngigetei, "gazelles"

1960s, Logogwotho, "luggage"

Lolibakipi, "green water"

Namongo, alluding either to cows or tosweetness

Mukuki, "grabbing"

Nagilgil, "helicopter"

Kalameriaputh

Lopetun, "widespread"

1970s, Lobulbul, "cattle disease,""swelling"

Amin, "the time of Amin," "Aminstripped the people," aslo Nawaikorot

Kolera, "cholera"

Ngamatidae, "homemade guns"Ebuta, "didn't seed"Ngimatheniko, "Matheniko"

1980s, Akoro, "hunger"

Apaloris

Amilica, "militia"

Nakoko, "theft"

Three-piece, "the three-piece-suits"; alter-nately, "hung by three pieces"

1990s, Apuno

Matakul, "where people drink urine"

Ateregaegae, "meningitis"

The second year of a drought that began sometime in the early 1940s, probably ~I943.The bands rust because women do not have ghee with which to grease them.

Construction of dams by the colonial administration begins in Karamoja, probablysometime in the late 1930s.

The last recorded women's age-set initiates its members in the same year as the previ-ous age-set, Ngakomomwa, "ant hills," is closed. Thorn trees are associated withrain, ant hills with the dry season. ~i945-49.

A hysterical disease in which people assume the identities of people from other com-munities becomes widespread, 1945-49.

A drought begins in 1952 or 1953 and is followed by a very good harvest, -1954.The last official transfer of authority from the senior male generation-set to the junior

set. Probably 1955 or 1956 (recorded for Pian by N, Dyson-Hudson [1966]),The first age-sets of this generation-set are initiated, sometime between 1956 and

1958,A period of much raiding and retaliation in Karamoja, There is no food in Bokora, and

people migrate widely in search of assistance and wage labor, ~i965-67.There is so much rain that sorghum sprouts before it can be harvested. The rains con-

tinue from the wet season of 1961 until the late spring of 1962, Called Achakaeki-pul, "the rain unlocked," in the Turkana calendar |Leslie et al, i999ti).

Name of a woman who inspired a movement in which people collected their troublesin a vessel passed from homestead to homestead and cast them out into the bush,Prohably 1962-63, Female informants still remember the song that was sung at thisritual.

Government soldiers raid homesteads in search of spears, and people are punished orimprisoned if they had them or hid them, 1962-63,

Government helicopters are used to confiscate cattle in retaliation for the killing of agovernment-appointed Bokora chief, Apanyanginyang, 1961-63,

Bokora raid Jie at this place; the Jie and Turkana retaliate, and more Bokora are dis-persed, ~r966,

A cattle epidemic decimates Bokora herds. Food relief is provided for the first time,1968-69,

An epidemic of hemorrhagic septicemia in Bokora kills the remaining cattle, leavingpeople destitute and without food because they also have been unable to plant,1974-75-

Amin's troops slaughter people at Nawaikorot protesting his prohibition of the wear-ing of traditional dress, 1971,

An epidemic and immunization campaign across the pastoralist zone in northernUganda and Kenya, 1971-72,

The Matheniko begin to raid the Bokora with homemade weapons, 1974,The sorghum planted flowers but does not seed. Probably 1976-77,Amin is ousted and the Matheniko raid his barracks in Moroto, hauling off thousands

of AK-47S, 1979,The famine of 1980, probably the defining event of the modern Karomojong

experience,A respected Matheniko elder by this name is murdered by the Teso on his way to

peace talks, setting off a bloody sequence of raids and retaliations in Teso and Kara-moja, 1982,

Bokora acquires guns, Acoli troops sent in to confiscate cattle as punishment for theongoing violence are waylaid by the Bokora and killed and their weapons taken,1983,

Government helicopters bomb homesteads, confiscate cattle, and set fire to all of themajor trading centers and cattle markets in Karamoja, Obote is overthrown, andmany more guns are obtained in another raid on the Moroto barracks. All of theKarimojong are now heavily armed with automatics, 1984-85,

Museveni assumes power in 1986, Men in suits appear in the district in anticipation ofparliamentary elections, Museveni's soldiers descend on the district to confiscateweapons. People who refuse to give them up are tied together at their knees and el-bows and hung from poles. Raiding resumes between Matheniko and Bokora,1987-88,

A Matheniko spokesman by this name is ambushed by Bokora en route to peace meet-ings. Several elders are killed in this attack, January 1992,

A major raid on Bokora is mounted by Somali, Turkana, and Matheniko raiders. TheBokora go after them and slaughter them at this desolate place where there is nowater, ~i99i I1987 is an alternate date). The decade is marked by a series of similar,large-scale raids.

An outbreak and immunization campaign, 1991,

GRAY ET AL. Cattle Raiding, Cultural Survival, and Adaptability \ S13

Nakadanya A ceremony in which power is transferred from the senior to the junior generation-set,scheduled in late 1998 without the consensus of the elders. The ritual is not carriedout according to established custom, and elders predict that it will increase intraeth-nic violence.

SOURCE: Akol and Gray (n,d,),NOTE: Events in the late 1990s have yet to be named,

pastoralists saw themselves forced to the hrink of cul-tural extinction (calendar events are indicated in paren-theses in the summary that follows). Their response wasa purposeful elahoration of institutionalized cattle raid-ing made possible hy their fortuitous acquisition of au-tomatic weapons. While this strategy may have guar-anteed their cultural survival—at least temporarily—itwas embraced at great cost to human survival and well-heing.

CATTLE RAIDING AND CULTURAL SURVIVAL INKARAMOTA, 1 9 5 0 - 9 9

The establishment of colonial authority in East Africain the early 20th century interrupted a long-standing tra-dition of interethnic cattle raiding among the pastoralistpeoples of the arid north (see, for example, Fukui andTurton 1979, Kurimoto and Simonse 1998). Hostilitieswere typically concentrated in periods of extreme en-vironmental stress, and raiding may have been integralto the success of nomadic pastoralism in this profoundlyuncertain context because it provided a means to recouplivestock losses to drought and disease outbreaks and toredistribute people and herds quickly in response toabrupt fluctuations in the resource base (Gray 2000; Hen-drickson, Armon, and Mearns 1998; Lamphear 1976,1988; Leslie et al, i999£'; Little et al. 1999 )̂.̂ Its adaptivesignificance for pastoralists notwithstanding, the insta-hility it created made institutionalized raiding untenahleunder British colonialism. Indeed, hoth raiding and no-madism posed real threats to the colonial economy. Mostproblematic was the threat of encroachment by nomadsand their herds on rangelands that either were claimedby European settlers, had been set aside as hunting pre-serves, or had been designated by the colonial adminis-tration as the permanent territory of some other "tribe,"Typically, the latter were those who had establishedclose ties to the colonial administration and receivedmost of the benefits of colonial economic development(Barber 1968, Lamphear 1992), In the Ugandan and Ken-yan Protectorates, these groups occupied the fertile ag-ricultural and grazing lands in the south and in the high-lands, and their assimilation into the colonial polity wascritical to British economic interests (Mamdani, Kasoma,and Katende 1992; Bouckaert 1999:28-29). In contrast,the northern pastoralist zone was important economi-cally only briefiy, as a source of ivory. By the 1920s theelephant herds had disappeared, and thereafter the northwas seen as a no-man's-land hetween the developingsouth and the southern extent of the Ethiopian empire

2, As Krystina Rigdon has suggested, it may also have enhancedthe resilience of the herds through genetic admixture.

(Barber 1968), serving only as a route for an expandingand troublesome northern gun trade (Barher 1968, Mam-dani, Kasoma, and Katende 1992),

To safeguard its economic and political interests in thesouth, the colonial administration pursued a policy ofdisarmament, coercive containment, and forcible settle-ment of pastoralists in the north. In Karamoja, trans-humant migrations were restricted hy periodic militaryoperations to enforce new district and trihal houndaries,by the designation of vast areas of the rangelands as forestor game preserves that were off-limits to herders (alsocontributing to the southern expansion of the tsetse flyzone [Mamdani, Kasoma, and Katende 1992]), and hy theestahlishment of "closed" districts in Karamoja and inneighboring Turkana in which rights to unrestrictedtravel and trade were limited to representatives of thecolonial administration (Barher 1968, Cisternino 1979,Mamdani, Kasoma, and Katende 1992). Cattle raidingwas checked by heavy-handed punitive campaigns inwhich livestock were confiscated in large numbers bygovernment troops (Barber 1968, Lamphear 1992), ameans hoth to compel recalcitrant herders to submit tocolonial authority and to implement the government'spolicy of destocking.

Initially, restrictions on pastoralists' mohility and ac-cess to rangelands had limited impact, inasmuch as theirherds had been decimated in the early 1900s by the com-bined effects of the rinderpest epidemic and governmentconfiscations (Mamdani, Kasoma, and Katende 1992).Karimojong herds were reduced further hy disease anddrought in the 1930s and 1940s, remembered by olderinformants as a time of unremitting hardship that waspunctuated by Lokwakoit, in the 1930s, named for the"white bones" of dead cattle (probably from East Coastfever [Mamdani, Kasoma, and Katende i992[), and Lor-engalaga, "when women's marriage bands rusted," in the1940s. The last known female generation-set (Ngadu-ruko) was initiated in the late 1940s, reflecting a Kari-mojong belief that social discord was the root cause oftheir ongoing misfortunes (according to informants in1998-99, women's generation-sets were traditionally as-sociated with the maintenance of domestic harmony andpeace: it is of interest that this tradition has virtuallydisappeared in recent decades).

Conditions improved in the 1950s, apparently a decadeof good rainfall. As the herds recovered, however, live-stock needs for water and forage increased proportion-ately, and colonial restrictions on livestock migrationsbecame more burdensome to herders. The few economicdevelopment initiatives undertaken in Karamoja by thecolonial and postcolonial governments—largely to facil-itate taxation of herders—only exacerbated the effects of

Si4 I CURRENT ANTHROPOLOGY Volume 44, Supplement, December 2003

prohibitive policies on pastoralists in that they entailedeither intensive development of irrigated agriculture,which accelerated environmental degradation in the dis-trict and the depletion of scarce water and forage re-sources (Mamdani, Kasoma, and Katende 1992), or ranch-ing schemes on the Western model, which assumedecological equilihrium and called for destocking to thelevel of hypothetical carrying capacity (Barher 1968,Cleave 1996, Fleay 1996, Mamdani, Kasoma, and Ka-tende 1992; see also Ellis and Swift 1988, Gray, Leslie,and Akol 2002).

After Ugandan independence in 1962, interethnic re-lations quickly deteriorated throughout the fledgling re-puhlic. Policies toward northern pastoralists, already re-pressive, hecame increasingly capricious as Ugandahegan its descent into civil war. Concurrently, environ-mental conditions worsened, heightening competitionamong pastoralists for rapidly dwindling resources. Ac-cording to the local calendar of events, the 1960s was aperiod of hunger and social upheaval. The killing of aBokora chief, Apanyanginyang, triggered the last of thepunitive campaigns carried out hy the colonial admin-istration, during which large numbers of cattle were con-fiscated (the use of helicopter gunships to bomb pastor-alists is reported in the calendar for the first time in thisdecade, immediately prior to independence, hut in factmay have occurred after 1962 [Crawford Young, personalcommunication, 1999]). Animal confiscations continuedin the postcolonial era against a hackdrop of drought,poor harvests, and livestock epidemics. In the mid-1960sthe threat of famine forced impoverished Bokora to mi-grate in large numhers to neighboring districts to escapeinsecurity and starvation, initiating a diaspora thatwould continue into the 1980s (Logogwotho). Manywomen reported that they never saw members of theirfamilies again. There was widespread hunger in 1967,and this appears to have heen the first of a series of foodshortages in which traditional strategies to avert large-scale famine broke down: a large-scale international aidintervention was mounted in Karamoja for the first timeduring this crisis, and by far the most extreme solutionto an imminent pastoralist disaster in the north was pro-posed. Calling for resettlement of the entire Karimojongpopulation on comhined ranching and farming schemesin the west of the district, the outrageous project wassubmitted for consideration by the minister of regionaladministration in 1968 but wisely dropped before it wasbegun (Mamdani, Kasoma, and Ketende 1992). Never-theless, it underscored a widening disjunction hetweenpastoralists and government after the collapse of colon-ialism.

The 1970s appear to have been worse as a consequenceof drought and the oppressive policies of Idi Amin, whoheld power for most of the decade. Women recalled hav-ing had their heads and clothing cut off their bodies bysoldiers who were enforcing Amin's prohibition of tra-ditional dress (recalled in the calendar as "Amin strippedthe people"); old women remembered burning goatskinsin great fires and pounding piles of beads into dust. Peo-ple were compelled to abandon traditional subsistence

activities to plant cash crops such as cotton. There wasan outbreak of cholera. Asian traders, who supplied theKarimojong with many hasic goods, such as cloth, sugar,and soap, were expelled (Wilson 1985). For the Bokora,the decade was horrific, bringing recurring famines, ep-idemics, and dispersals. People reported having sold theirchildren in Mathany market during Lohulbul [a livestockdisease that ravaged the herds in 1974-75] hecause theyhad no food to feed them. Throughout the decade, Kar-imojong migrated in large numbers to Teso, Masindi,Iganga, and Mbale and to Busia, in Kenya.

The impact of the environmental disasters and polit-ical repression of the 1960s and 1970s was aggravated bya virtual pandemic of cattle raiding, which had eruptedagain in the mid-i9sos as herds recovered and availahlerangelands shrank (N. Dyson-Hudson 1966, Barber 1968).During the 1960s Jie, Turkana, Pokot, and Karimojongmounted increasingly aggressive raids against theirneighhors throughout the north (Thomas 1965, Leslie etal. 1999b). Local political relations at this time werecomplicated hy the constant presence of government andrehel troops and hy the manipulation of pastoralist pol-itics hy outsiders (including Sudanese factions) whoseallegiances shifted opportunistically with prevailingwinds in the hrutal civil war (Kidon-Unyang 1987; Pi-rouet 1991; Woodward 1991; Mamdani, Kasoma, and Ka-tende 1992; Brett 1996; Bouckaert 1999:30-35). Modernweapons, now procured with relative ease in a contextof civil war in Uganda and southern Sudan, were in de-mand among pastoralists across the north, and the sav-agery of raids intensified in corresponding fashion.

The pivotal event in recent Karimojong history cameon the eve of Amin's ouster in 1979 (Ngimatheniko),when Matheniko raiders ransacked his ahandoned ar-mory in Moroto and reportedly made off with tens ofthousands of automatic weapons (Alnwick 1985, Wilson1985). In an effort to overcome the reversals of the 1960s,the Matheniko had joined forces in the early 1970s withtheir former foes, the Turkana. With them they had in-itiated raiding on other Karimojong sections, sometimesusing homemade guns (Ngamatidae). After their acqui-sition of automatic weapons, they raided with the Tur-kana indiscriminately in Karamoja and across othernorthern districts. In the early 1980s (Amilica), the Bo-kora also acquired automatic weapons and now engagedin retaliatory raiding against the Matheniko. By am-bushing government troops sent in to quell the violence,other Karimojong sections procured automatic weapons,and by the middle of the decade everyone was armed.The Karimojong now raided widely outside of Karamoja,particularly in Teso. In retaliation, the Teso killed aMatheniko elder in 1982 (Apaloris), setting off a partic-ularly savage cycle of Teso-Karimojong raids (Nakicu-met). A series of revolving-door regimes during the sav-age last years of the Ugandan civil war sent troops topunish and disarm the Karimojong; between 1984 and1987 they managed to subdue the pastoralists by burningmarket centers to the ground and bombing homesteadsfrom helicopter gunships (Nakoko). After Yoweri Mu-seveni's victory in 1986 and the suhsequent conclusion

GRAY ET AL. Cattle Raiding, Cultural Survival, and Adaptability | S15

of the war in the south of the country, the new NationalResistance government immediately launched a disar-mament campaign ("three-piece") in Karamoja in an ef-fort to pacify the north and consolidate power. Remem-bered by informants as especially brutal, the campaignapparently succeeded only in intensifying the hostilityof northern pastoralists toward the government in thesouth. Subsequently, armed looting of government andnongovernment facilities and convoys became the chiefstrategy for recovery and resistance (Knutsson 1985).

The violence continued unabated in the early 1990s,when the Bokora, hy resorting to large-scale raids on theTeso, in particular, recovered from the devastating lossesof the previous decades. People recalled raids at Nathe-perwae, Turutuko, Batanga, "Alinga's kraal," and, es-pecially, Matakul, where a comhined raiding force of So-mali, Matheniko, and Turkana was slaughtered by theBokora. There were desultory attempts to restore peace:in one infamous meeting the Bokora agreed to meet inKangole for talks, where instead they ambushed andkilled the Matheniko representatives, including an in-fluential herder (Apuno), and set off another wave of re-taliation. People in surrounding districts were, however,the chief targets of cattle raids, and under pressure fromrepresentatives of Teso, Sehei, Lango, and Acoli the gov-ernment called a peace conference in Kampala in 1994(GOU 1994). In the period that followed, herders werecalled upon to disarm and cattle whose ownership wasdisputed as a consequence of raids were confiscated andheld in government facilities. The peace held for twoyears, then collapsed in the period hefore the parliamen-tary elections in 1996. At that time, members of parlia-ment who had enforced the government's policies forpeace retreated under threats of death, and politics asusual resumed in 1996. With the exception of this briefhiatus, the decades between 1980 and the present timehave been marked hy unahated armed violence in a con-text of recurring drought and continuous food insecurity.In late 1999, several hundred raiders were reported tohave been killed hy helicopter gunships (Gray 2000; seealso Dawes 2001), and in an attempt forcibly to suppressthe raiding in 2,000-2001 Kampala armed people in ad-jacent districts at the same time as it carried out yetanother largely unsuccessful disarmament operation inKaramoja.̂

Cattle Raiding and Biobehavioral Adaptability

Widespread adoption of automatic weapons and the re-sultant transformation of cattle raiding occurred againsta backdrop of unrelenting drought in the north. Almostimmediately, AK-47 raids compromised the ahility of thelocal population to respond effectively to steadily dete-

3. In addition to the inaccessibility of much of Karamoja, a steadyflow of arms across the Sudanese border continually underminesgovernment efforts to disarm pastoralists at tbe present time. Infact, tbe major constraint on raiding appears to be tbe bigb cost ofammunition.

riorating environmental conditions: in the spring of 1980Matheniko raids compounded the effects of drought, pre-cipitating the famine known by the Karimojong simplyas Akoro, "hunger." In the years since, raiding has he-come the principal constraint on Karimojong biobehav-ioral adaptability.

EFFECTS OF AK-47 RAIDING ON MORBIDITY ANDMORTALITY IN KARAMOJA

Increasing mortality of adult Karimojong males after1940 is linked directly and unequivocally to the inten-sification of cattle raiding. In the 1950s, when cattle raid-ing resumed in the north, violence accounted for an es-timated 22% of all adult male deaths, an increase ofapproximately 10% from the preceding two decades(Sundal 2002; fig. 5). By the 1970s, when automatic weap-ons were adopted by the Matheniko and their neighbors,raiding accounted for 3 5 % of adult male deaths, and theproportion has remained constant at approximately 33 %in the ensuing decades. Between 37% and 44% of malesin this sample who died between age 30 and 60 diedviolently, and 25% of deaths of men between ages 60and 69 were attributed to violence (fig. 6). Additionally,a significant proportion of male deaths in all age-classeswere attributed to famine-related causes (diarrheal in-fections, hunger, and "swelling"), which are indirect con-sequences of modern raiding. In all, direct and indirecteffects of AK-47 raids accounted for an estimated45-70% of adult male deaths in this sample. For malesaged 30 to 49, these combined estimates exceed 70%. Inthe 30-39 age-group, the number of deaths by violenceapproaches 50%, whereas for men aged 40-49 cause ofdeath is more evenly distributed between violence andfamine (35% to 40%). These results most certainly ex-plain the high probability of mortality during these dec-ades among men born after 1940 (fig. 4). Karimojongmales in their 20s appear to be at somewhat lower riskof dying. We hypothesize that this is an effect of thetiming of their formal attainment of adult status throughtheir initiation into the age-set system: traditionally,raiders were initiated males of the junior generation-set(N. Dyson-Hudson 1966:185-86). Today, Karimojongmales are initiated in their middle to late 20s. For boysin school, initiation and the assumption of their role asherders and warriors may he postponed even longer.Thus, while uninitiated herd hoys may die as a conse-quence of raids on cattle camps, they are less likely tohe actively involved in carrying out raids.

Whereas many women reported having lost their par-ents, spouses, and adult siblings directly to AK-47 raid-ing, changing patterns of child mortality and morbidityin Karamoja are associated with its indirect effects onsubsistence activities and food production, residence pat-terns, and access to health care as a consequence of re-duced population mobility. In 1998-99 the threat of raidsinduced herd owners to move their livestock to remotegrazing areas, beyond the range of marauders, during thedry season, while heightened risk of amhushes duringseasonal migrations compelled them to forgo movement

Si6 I CURRENT ANTHROPOLOGY Volume 44, Supplement, December 2003

I(D

CL

\ -

•Diarrhea, females iMDiarrhea, males raDiarrhea, SABFamine, females HFamine, males SFamine, SAEaViolence, males ^Measles, SA ^Malaria, SA

>^'

Decade of deaths

FIG, 5. Patterns of morbidity and mortality in the Karimoiong sample from the 1930s to the 1990s. Onlymajor causes of death are shown. SA, subadults. Concurrent social and political events from the Karimojongevent calendar: L, Logogwotho; M, Ngimatheniko; A, Akoro; GOU, the government's peace initiative in themid-1990s.

back to the agricultural zone in v̂ rhat was an exception-ally good rainy season, when returning herders otherwisemight have assisted women in preparing fields forplanting.

The loss of men's lahor and the threat of attacks onindividuals curtailed women's agricultural activity andsubsequently diminished sorghum yields in spite of goodrainfall. At the same time, women, young children, andthe elderly were forced to remain in the agriculturalhomesteads regardless of the inadequacy of the harvestbecause they were unable to undertake the long, dan-gerous trek to distant cattle camps alone: informants alsoreported numerous attacks on solitary women, and raid-ers have been reported to steal wives as well as cattle(GOU 1994). Loss of access to the herds deprives childrenof a critical source of protein and micronutrients andmay be the most important factor in the epidemiologyof toddler kwashiorkor in this population (Wiebusch2002), Indeed, a frequent request from women inter-viewed in 1998-99 was for transport to the cattle camps,because Karimojong mothers were well aware that their

children were healthier and grew more rapidly when theyhad access to the herds.

Since the famine of 1980, relief food has been availableat a number of distribution centers in the district. Home-steads nearest the relief centers have become the loci ofdense concentrations of displaced and nutritionally vul-nerahle people, where crowding and contamination ofalready inadequate water supplies ensure that epidemicswill spread quickly (see Allen 1996, CDC 1992, de Waal1987, Mohamed 1999), The establishment of these set-tlements corresponds with changing trends in morbidityfrom i960 to the 1990s (fig, 5), Prior to i960, malaria,gastrointestinal disease, and miscellaneous "other"causes accounted for most deaths. The pattern changedin the ensuing decades, when measles mortality domi-nated: measles outhreaks were responsible for 22%, 26%,and 23% of subadult deaths from known causes in the1960s, 1970s, and 1980s, respectively, compared with15% to 17% in earlier decades. Simultaneously, malaria-related deaths fell from over 30% in the 1940s to lessthan 12% in the 1960s, The timing of the peak in measles

GRAY ET AL. Cattle Raiding, Cultural Survival, and Adaptability \ S17

t—

"cDCD

Ti

male

(t

of

adult

rcen

Q.

80

70

60

50

40

30

20

10

0

: : : • • .

II

•Violence^Diarrhea and famine•Combined

: ' - • : • : " * ' . *

Age at death

FIG. 6. Proportion of adult male deaths attributed to direct and indirect effects of AK-47 raids, by decade ofdeath.

deaths, in the 1970s, was synchronous with the Bokoradispersals and the estahlishment of relief centers in thedistrict. Measles deaths declined somewhat but re-mained high in the next decade, prohahly reflecting sus-tained high mortality in settlements established duringthe famine. Subsequent introduction of immunizationprograms by aid agencies in response to increasing mea-sles mortality most certainly accounts for the decline inmeasles deaths after the 1980s (Wiehusch 2002). Deathsfrom diarrheal diseases showed a corresponding increaseand decline: in the local calendar, 1971-72 is referred toas Kolera (cholera). Subadult mortality from meningitisand hepatitis, recognized diseases of crowded, unsanitaryconditions, also increased in Karamoja after relief centersbecame established in the district: the proportion ofsubadult deaths attributed to meningitis and hepatitisoutbreaks increased from approximately 5% prior to1980 to 15% in the 1990s (Sundal 2002:102). In the localevent calendar, 1991 is known as Ateregege, "meningi-tis," named for an epidemic and vaccination campaignin that year.

Subadult deaths attributed to famine (synonymouswith hunger) declined simultaneously with the increasesin measles, meningitis, and diarrhea deaths. The mostlikely interpretation of this pattern is that the contin-

uous presence of relief operations after 1980 reduced theprevalence of malnutrition at the same time as it height-ened exposure to infectious diseases. A spike in deathsfrom malaria appeared in the 1970s, but generally thenumber of malaria-related deaths also decreased afteri960 (fig. s). Attenuation of mortality from malaria mayhe linked to improved prophylaxis after relief centerswere established in the district or to the drought con-ditions of the late 1970s and 1980s. A more plausibleexplanation is that the most vulnerable children weremore likely to die of measles or diarrheal infections: thedramatic decline in measles deaths in the 1990s is con-current with a slight increase in malaria mortality.

Raiding also interferes with the transport of essentialcommodities and with the maintenance of Karamoja'sminimal infrastructure. Access to clean water and soapis unreliable in rural homesteads across the district, andthe lack of hygiene exposes very young children chron-ically to multiple secondary infections, such as scabiesand worms, further damping their resistance to acuteinfections (Wiebusch 2002). The formal health care sec-tor is particularly vulnerable: with few exceptions, gov-ernment-operated health care facilities in Moroto werepoorly supplied and understaffed in 1998-99 in spite ofthe habitual presence of numerous international aid

Si8 I CURRENT ANTHROPOLOGY Volume 44, Supplement, December 2003

agencies in the district (including the Danish aid organ-ization DANIDA, Doctors Without Borders, the Sistersof Charity, Guinea Worm 2,000, the Comboni Mission-aries, UNICEF, Save the Children, the World Food Pro-gram, and the Italian medical missionary programCUAMM, among others). The health care system hashecome a prime target of raiders, and attacks on healthunits and on vehicles transporting health care workers,drugs, and supplies are increasingly common. In the1980s, drugs stolen from the health sector in Karamojaappeared for resale in markets in Mhale, according toWhyte (1991). Doctors Without Borders, which was par-ticipating in the rehabilitation of the health care sectorin 1998, withdrew in late 1999 in response to attacks onhealth care workers. Consequently, Karimojong are atonce more exposed and more vulnerahle to infection andunlikely to he seen hy a trained professional or to receiveadequate treatment after diagnosis even if they do haveaccess to a medical facility.

Three narratives summarize the local perception of thequality of health care in the district. One woman toldus, "If you go to the hospital [district hospital in Moroto],you wait and wait. They take hlood, your temperature,hut the child is sick and you have to wait. You alwayswait a long time, until [that child[ dies in your lap." Asecond informant reported that all of her children weredead, even though she had sought medical treatment foreach of them:

The first child was crawling when he died of pneu-monia in the time of Amin's takeover [early r97os[.He was sick one week, then I took him to hospital[in Moroto[, and he died there. The second childdied of measles—she was old enough to collect fire-wood. She went to the hospital and recovered afterone week and came home, hut then she got measlesand died: the family had said not to take her to hos-pital." Nothing can he done for measles. Grief forthis girl caused me to miscarry: it was horn earlyand died the next day. The third child died whenshe was crawling—she had a fever hut the hospitalwouldn't admit her, and she died in three days.

A third woman tersely recalled the consequences of thehreakdown of the health care system in Karamoja duringthe exceedingly violent years of the early 1980s: "I wasnursing a child—he was ahout one month old. The childdied of fever—it cried and then it died. I couldn't get tothe dispensary, hecause we were running, running, fromthe soldiers. Anyway, the medical staff had run away,too."

EFFECTS OF AK-47 RAIDING ON THE PROXIMATEDETERMINANTS OF FERTILITY

The association hetween cattle raiding and damped fer-tility of women horn after 1940 appears to have heen

4. Consensus among informants was that people went to the gov-ernment hospital "to die." A similar attitude toward the districthospital in Moroto was articulated by Turkana subjects in the

mediated hy two proximal effects (Davis and Blake 1956;Wood 1994:69-72): (i) the changing structure and dura-tion of sexual unions and (2) heightened risks to women'sreproductive health. The first is a consequence of in-creasing male mortality in raids. If the fertility patternsof Karimojong males are similar to those of Turkanamales, then over 50% of the men in this sample died intheir reproductive prime (approximately age 25-54 [Les-lie, Dyson-Hudson, and Fry 1999]), and the majority ofthese deaths resulted in the premature termination of asexual union. Since Karimojong men are older than theirwives hy ten years or more (men typically marry first intheir 30s, whereas women marry in their late teens toearly 20s), deaths hy violence of men in their 30s and40s produced a numher of widows still in their repro-ductive years. Ironically, that men marry in their fourthdecade may amplify their mortality risks from violenceat that age, if, as we suspect, AK-47 raiding has hecomethe principal means to procure hridewealth.

In the past, widows from marriages that had heen for-malized hy the payment of hride price were remarried tomale relatives of their dead hushands, generally withinthe year. In this sample, remarriage to an in-law stillappeared to he the rule for women in their 20s and early-to-mid-3os. For older premenopausal women, remarriagewas no longer a given. We surmise that the inequitahledistrihution of livestock created hy AK-47 raiding (Ger-tzel 1988) has made payment of hride price increasinglyhurdensome for many men. A concurrent decline in theratio of marriageahle men to marriageahle women as aresult of high male mortality in raids (Sundal 2002:93)may have necessitated a relaxation of proscriptionsagainst informal sexual unions,- consequently, more Kar-imojong women are "unmarried wives" in unions notofficially sanctioned through the exchange of cattle. Ayoung wife is effectively less expensive, and a union withsuch a woman may present a more attractive reproduc-tive option to a young man who might otherwise hecompelled to accept responsihility for an older widowand her dependent children. Certainly, many women inthis sample who were widowed in their 30s or early 40shad failed to remarry, a phenomenon contributing to ear-lier social termination of reproduction in the post-1940cohort.

Both formal and informal marriages appear to havebecome less stable as a consequence of AK-47 raids. In-formal unions may be costly for males, but they also areless binding and may be dissolved readily. Both the manand the woman are then free to enter into unions withother individuals from other lineages: unofficially mar-ried women in the study commonly reported participat-ing in two or three such unions. One Matheniko inform-ant explained her daughter's confusing reproductivehistory: "That daughter has heen married three times.She had three children by her first hushand, then he waskilled by Turkana. She had one child hy her second hus-band, who was killed by Tepeth in 1995. Then she mar-ried again and has one child by that man." At the sametime, some women for whom an official marriage hadheen negotiated recalled that the marriage had broken

GRAY ET AL. Cattle Raiding, Cultural Survival, and Adaptability \ S19

down because the husband's cattle were stolen beforethe bride-price payments had been completed (in somecases, during the exchange). Other Matheniko womenreported having moved to the home of a formally con-tracted spouse and, indeed, having borne him severalchildren only to have the contract nullified before thefull bride price had been paid when another man steppedforward with the full payment, acquired recently in araid. Finally, displacement of the Karimojong as they fledto escape food shortages and raiding led to the termi-nation or temporary disruption of both official and in-formal unions, when spouses and often their childrenwent in different directions in search of wage labor oralternative support systems. Many were separated for adecade or more, and the adults entered into informalsexual unions with other individuals. A cassava devel-opment project in Apeitolim is populated largely by re-turnees, most of whom found that they had no spousesor families to which to return.

Our data show an emerging marriage pattern of serialinformal unions of short duration that deviates from thepattern prior to the 1950s. Women in the post-1940 co-hort were significantly more likely to be unmarriedwives—married without the exchange of cattle and a for-mal marriage contract between two homesteads (oddsratio 6.0; 95% confidence limits 3.56-10.03). Seventypercent of the women in this cohort were unofficialwives. Given such a high percentage, intercohort differ-ences are unlikely to be an artifact of ascertainment biasamong older women as a consequence of more of the pre-

cohort's having died or left the pastoralist system.

INDIRECT EFFECTS OF AK-47 RAIDING ON WOMEN'S

REPRODUCTIVE HEALTH

At present we have few data on the effects of these un-stable unions on women's reproductive health, particu-larly on the incidence of sexually transmitted diseasesand HIV/AIDS. However, AK-47 raids have affected theirutilization of the formal health care sector and their ac-cess to antenatal and obstetric care. Further, amid thesocial turmoil generated by raiding, indigenous institu-tions that ensured preferential care of pregnant andbreast-feeding women (Gray 1994) appear to have brokendown (see also Madut Jok 1999). In the intervals betweeninformal sexual unions, whether they are terminated asa result of "divorce" or the death of the male, an un-married wife's status in relation to her social networksis ambiguous. Neither her agnatic nor her affinal kin areunder obligation to provide for her, and this compromisesher access to critical resources (sorghum fields as wellas livestock) and consequently her health and nutritionalstatus. Termination of a union when the woman is preg-nant—a situation frequently reported by women in thestudy—may therefore jeopardize pregnancy outcomes aswell as maternal and infant survival.

On the basis of data compiled from reports by districthealth units in Karamoja for the years from 1992 to 1996,we estimated maternal mortality in this population toexceed 1,700 deaths/100,000 pregnancies (Gray and Pike

1998; see also Barton and Wamai 1994). During the1998-99 field season, we recorded 33 pregnancies in oursample. One of these most certainly would have resultedin the death of the mother during childbirth without theintervention of Gray and her team. A second childbirthin which intervention was not possible did end in ma-ternal death (a third woman, who was taken by the teamto St. Kizito's in her third trimester and was subse-quently diagnosed with severe malaria, would otherwisehave miscarried, at considerable risk to her own sur-vival). These outcomes denote a maternal mortality rateof I in 17 pregnancies among women who do not haveaccess to obstetric care: childbearing in Karamoja rep-resents an appallingly high risk to women by any stan-dard or definition.

DIFFERENTIAL EFFECTS OF AK-47 RAIDING IN

BOKORA AND MATHENIKO

Perhaps the most compelling evidence of the human im-pact of AK-47 raiding in Karamoja is found in divergentpopulation trends in Bokora and Matheniko. First, Math-eniko women were significantly more likely to report achild death than Bokora women (odds ratio 1.5; 95%confidence limits r.03-2.21), and in every decade sincei960 mortality prior to age ten years tends to be higherin the Matheniko sample than among the Bokora (tablei). The difference is marked after 1980, when the prob-ability of dying before age ten in the Bokora sampledropped from a high of 37% in 1970-79 to 24%. De-creases in Bokora rates are apparent for all age-classes inthis period but are most pronounced between ages oneand five years (— 8 %) and between ages five and ten years( — 6%). There is some suggestion that the decreasingtrend in child mortality continued into the 1990s in Bo-kora (the infant mortality risk reached a low of 11% inthis decade, whereas mortality between ages one and fiveyears dropped to 4%, the lowest rates since the 1950s).In contrast, in Matheniko the probability of dying beforeage ten never dropped below 30% in any decade, and itfioated at or above 40% from 1950 to 1980. Furthermore,infant mortality in this section actually was lowest inthe 1970s and 1980s, when it was highest in the Bokorasample. Infant deaths in Matheniko increased in the1990s, whereas mortality rates of older subadults (agesfive to ten years) never fell below 6%.

These differences may be explained by distinct re-sponses of the two territorial sections to the crises of the1960s and 1970s (Gray n.d.). To ensure their own sur-vival, the Matheniko emphasized armed diplomacy—first by realigning themselves with their former ad-versaries in Turkana, who were heavily armed at thetime, and second by embracing AK-47 raiding. In sub-sequent decades, this militarism has assumed an elab-orate, almost stylized quality that is synonymous withMatheniko cultural identity as "warriors" (Gray 2000).Today they are recognized throughout northwesternUganda as the most conservative, most bellicose sectionof the Karimojong, and they are more likely to be in-volved in cattle raids both as perpetrators and as the

S20 I CURRENT ANTHROPOLOGY Volume 44, Supplement, December 2003

targets of retaliation. They are also more openly defiantof government interventions of any kind. Consequently,Matheniko children are more vulnerahle to direct andindirect effects of violence, are less likely to go to school,and have more limited access to health care: few of theMatheniko children in our study had heen immunized(12% of Matheniko children in the study, compared with50% of Bokora children), even in homesteads within 5km of Moroto District Hospital.

This contrasts with the Bokora response to the crisis,in which they initially were victims rather than aggres-sors. Left destitute hy the droughts, famines, and raidsof the 1970s, they were more likely to he recipients ofinternational aid after 1980. Their close association withdonor agencies and enhanced access to education andhealth care may explain the steady decline in suhadultmortality in Bokora after 1979. Increasing immunizationcoverage probahly explains the decrease in under-fivedeaths, whereas the declining mortality of school-agechildren is most likely an effect of their enrollment ingovernment and mission schools. Older children tend tohe at higher risk of nutrition-related mortality (tahle 2),and parents in Bokora may have opted to enroll childrenin school, where they could take advantage of donor-suhsidized meals programs. Attendance at schools ac-tually may have afforded children dual protection fromraids and starvation. Today, the majority of Karimojongcivil servants, husiness people, and educators are Bokora,who are well integrated into Ugandan national cultureand who navigate smoothly hetween their Ugandan andKarimojong identities.

Marriage structure also differs hy territorial section. InMatheniko, the emphasis on AK-47 raiding facilitatedthe accumulation of disproportionately large herds hysome men after the 1970s. This in turn supported anincrease in hride price in that section from 20-30 headof cattle in the early 19SOS to 100-150 head hy the early1980s, according to informants. Bride-price inflation it-self may now he hoth a cause and an effect of Mathenikoraiding. On the one hand, it is an exceedingly difficultand slow process for many men to complete such steeppayments hy means of traditional mechanisms (N. Dy-son-Hudson 1966), hut on the other hand the high costof marriages is an added incentive for herders with fewcattle to invest at least some of their animals in modernweapons, which can then he used mount raids to procurethe requisite cattle. Livestock of less well-armed Bokoraherders have heen a favorite target, and a dramatic re-duction in the numher and size of their herds after 1975has made it increasingly difficult for many Bokora fam-ilies to complete hride-price payments. Accordingly, agreater proportion of official marriages might he antici-pated in the Matheniko sample than in the Bokora sam-ple. Our data support this hypothesis: 61% (125 of 204)of the Bokora women were married without cattle com-pared with 40% (38 of 95) of the Matheniko women (oddsratio = 2.4; 95% confidence limits 1.44-3.90). Moreover,this is not simply an effect of premarital liaisons amongyoung women who eventually may he expected to hecontracted formally in marriage: when the sample is re-

stricted to women aged 30 to 59 years, intersectionaldifferences in the percentages of official and unmarriedwives are unchanged.

Human Adaptability in South TurkanaReconsidered

when the STEP research hegan in 1979, cattle raidingwas acknowledged as one of the principal stressors shap-ing Turkana hehavioral responses. Although their mainadversaries at that time were the Pokot, the Turkanathemselves had heen involved in armed raiding againstthe Karimojong, Dodoth, and Toposa. The Karimojongevent calendar provides considerahle detail regardingTurkana aggression, documenting Turkana raids againstBokora, in particular, in 1966-67, 1974-75, and 1978 (seealso Thomas 1965). In the late 1970s the government ofKenya disarmed the southern Turkana, leaving them es-sentially defenseless against the heavily armed Pokot,who suhsequently perpetrated a numher of hrutal raidsin South Turkana hetween 1977 and 1984 (R. Dyson-Hudson 1999, Leslie et al. 1999^). Theft and raiding infour Turkana families (Dyson-Hudson and McCahe 1985,McCahe, Dyson-Hudson, and Wienpahl 1999) accountedfor 15 % of their total livestock losses in the second yearof the drought of 1979-81 and 35% of losses in the firstyear of recovery. During the drought, the herd ownersmoved their main camps on average 15 times per year,and over 20% of their moves were hased on the proha-hility of attacks by Pokot or other raiders (McCahe 1987,Pike n.d.).

Nevertheless, concerns for forage and water still ex-ceeded security concerns during most of the period ofSouth Turkana study, and during a three-year period ofgood rainfall in the late 1980s the threat of raids wasgreatly diminished. However, the late 1990s saw an es-calation in AK-47 raiding across the whole of the north-ern pastoralist zone. In the last decade of the 20th cen-tury insecurity in Turkana paralleled that in Karamoja,with a familiar pattern of AK-47 raids and counterraids,government retaliation, and ineffective disarmamentcampaigns that perpetually shifted control of the guns(and with it the advantage in raiding) hack and forth.Virtually every pastoralist group in northern Kenya andUganda was caught up in the devastation (Little et al.1999a), although most of the violence involved Jie, Kar-imojong, Pokot, Nyangatom, Turkana, and Tepeth.' Be-tween 1992 and 1999, the Pokot carried out 14 large-scale AK-47 raids in critical dry-season grazing areas inSouth Turkana (Pike n.d.). In 1991 a massive operationagainst the Bokora hy a comhined force of Turkana, So-mali, and Matheniko ended disastrously for the raidersin their slaughter at Matakul (tahle 3) and triggered pre-dictahle retaliation. In spite of continued efforts hy hoth

5. Formerly hunters and gatherers, the Tepeth of the mountainouszone along the Kenya/Uganda escarpment have acquired cattle inrecent decades and are now engaged in raiding against the Kari-mojong, reportedly in alliance with the Pokot.

GRAY ET AL. Cattle Raiding, Cultural Survival, and Adaptability \ S21

Ugandan and Kenyan governments to effect an end tothe violence, hy late 1998 hoth districts were in chaos.In that year schools were attacked in hoth Turkana andKaramoja and schoolchildren were killed; at least 100people were killed in a Pokot raid at Kainuk in early1999 (Pike n.d.). In 1993 none of the Turkana herdersparticipating in a study conducted hy Pike (1999) werearmed; in a follow-up study in 1998, every herder wasin possession of an automatic weapon.

In Turkana, mohility is the most effective strategy fordamping the impact of raids on the human population.Herders respond to the continuous threat of raids hymoving their main camps (which include women andchildren as well as adult males and herd hoys) more fre-quently and into more remote grazing areas. During 1998the average numher of camp moves of herders inter-viewed hy Pike (n.d.) was 24, and 100% of these moveswere in response to the threat of raids. Moreover, manyof the herders in South Turkana, where Pokot raids wereconcentrated, moved farther north, where they came intoconflict with other Turkana territorial sections. By thelate 1990s AK-47 raids had escalated to such a degreethat Turkana herders, who previously had migrated insmall, independent groups (Dyson-Hudson and McCahe1985), also hegan to move their herds in large-scale massmigrations, a defensive strategy adopted from the Kari-mojong (N. Dyson-Hudson 2000, Pike n.d.).

As in Karamoja, the escalation of AK-47 raids in Tur-kana was associated with heightened stress in the humanpopulation. In 1994 the South Turkana landscape wasdenuded of cattle, whereas in 1990 it had featured vastherds wherever one looked. In the space of less than fouryears, the herds had virtually disappeared, and Turkanainformants attrihuted the collapse chiefly to raiding hyPokot and Jie rather than to the recurring droughts andgenerally dry conditions from 1990 to 1994. More im-portant, the losses appear to have heen universal, andduring the drought of 1993-94 emergency food relief wasdistrihuted in South Turkana for the first time (Pike1999). Long-standing strategies for damping the impactof drought and averting famine in Turkana depend ondifferential fitness of family herds within networks of"cattle kin" (Dyson-Hudson and Dyson-Hudson 1999,McCahe 1990), hut no such variahility exists in relationto large-scale assaults with automatic weapons. Whereall herds are affected equally, mechanisms for deflectingor redistrihuting stress hreak down.

In response to looming famine, pastoralists were com-pelled to curtail their migrations and move closer to fooddistrihution centers, where they remained until 1995,when the intervention was terminated (Pike n.d.). In in-terviews in 1994 many people said that access to foodrelief as well as fears for their personal safety influencedthem to remain close to the distrihution centers. As inthe case of Karamoja, morhidity was high in these densesettlements, where there were outhreaks of measles andhepatitis, as well as of chloroquine-resistant malaria.The magnitude of the health crisis was such that somedistrict health units closed their doors in 1993 hecauseno drugs were availahle at the time to treat the epidemics

and hecause health care workers were themselves fearfulof hecoming infected.

Shell-Duncan et al. (1999) conducted a study of healthand immune function in South Turkana during the firsttwo years of the calamitous period from 1990 to 1994.In contrast to the findings of Gray (1992, 1998), whostudied maternal and infant health and nutrition in theperiod prior to the escalation in AK-47 raiding (1989-90),the families sampled in the Shell-Duncan et al. studywere largely sedentary, and their children were at height-ened risk of acute malnutrition and multiple infections.The contradictory results of these two studies, carriedout within two years of each other, underscore not onlythe dramatic effects of spirals of raiding and drought inTurkana hut also the speed with which human resiliencemay shift in response to social as well as environmentalconditions. In this regard, the lack of seasonal effect inShell-Duncan's study is particularly revealing (Little etal. 1999(3:325-26; see also Galvin 1985): among someherding units, customary responses to drought may haveheen secondary to responses to the escalation in cattleraiding.

The most telling evidence of the impact of AK-47 raidson the human population in South Turkana is providedhy Pike (n.d.) in her 1998 study. In a preliminary inves-tigation of the psychosocial effects of raiding, 82% ofherd owners interviewed cited raids as their principalsource of anxiety, whereas 72% and 45 % of women citedfood shortages and raids, respectively. In hoth Turkanaand Karamoja, these two are linked unequivocally.

Culture as Stressor and the Persistence ofPastoralist Systems

Even at an early stage of fieldwork and analysis, it isclear that AK-47 raiding seriously constrained the hio-hehavioral adaptahility of the Karimojong in the last dec-ades of the 20th century. Through its effects on seasonalmigrations, suhsistence activity, the structure of mar-riage, and the stahility of social networks of the Kari-mojong, as well as on the operation of trade routes inand out of the district and on health care delivery, armedcattle raiding emerges as the critical factor in recurringfamines and epidemics in Karamoja since the 1970s. Itsimpact on human hiological resilience is incipient in theTurkana case, hut raiding was implicated unequivocallyin a series of population crises there in the 1990s (Hen-drickson, Armon, and Mearns 1998, Pike n.d.).

It might reasonahly he argued that AK-47 raids cur-rently represent the single greatest risk to the persistenceof the pastoralist systems considered here and to thecontinued survival of pastoralists themselves; if so, thepersistence of such a seemingly maladaptive institutiondemands further investigation. We are not in agreementwith Hendrickson, Armon, and Mearns (1998) andFleisher (1998, 1999), who argue that AK-47 raiding is apredatory strategy carried out for personal gain and re-sulting from the erosion of the moral economy. Rather,

S22 I CURRENT ANTHROPOLOGY Volume 44, Supplement, December 2003

we argue that its genesis lay in the elaboration of anadaptive cultural response to environmental uncertaintythat was altered in response to increasing sociopoliticaluncertainty (see also Mamdani, Kasoma, and Katende1992:47). The greatest risk to pastoralists' survival in thelate 20th century was not drought or disease but theirvulnerable position vis-a-vis governments that eitherpunished them or exploited them according to prevailingpolitical currents, ultimately marginalizing them andendangering their subsistence base. In this process, cattleraiding, transformed by a combination of serendipity andproximity to the expanding illicit trade in automaticweapons, became the chief mechanism for cultural sur-vival. This hypothesis requires consideration of the na-ture of Ateker cultural identities.

Broch-Due (1999) and Gray (2000, n.d.) have argued theprimary cultural identification of Ateker pastoralists isas pastoralists, and it is determined by the possession ofcattle: people with cattle are Turkana or Karimojong orJie, whereas people without are not. Cultural identity istherefore contingent on the preservation of the pastor-alist subsistence base. Cattle keeping itself has no guar-antees, but given the striking failure of alternative sub-sistence strategies in the pastoralist zone, enumeratedin detail in Majok and Schwabe (1996), this is a pragmaticideology. Not surprisingly, pastoralist families identifythemselves primarily with their herds, and insofar asother associations promote the survival of those herdsthey may identify themselves secondarily with groupsat different levels of the territorial hierarchy. None ofthese allegiances takes precedence over the primary iden-tity with cattle.* Indeed, without cattle, cultural identitybecomes irrelevant because survival of the family—theprincipal herding unit—is at best improbable. In otherwords, cultural survival hinges less on the persistenceof a group of people than on the persistence of a groupof herds—a proposition that receives support from theshallow genealogical memory of informants in Turkanaand Karamoja. Seen from this perspective, the elabora-tion of raiding becomes the chief mechanism for bothfamily and cultural survival, and militarism becomessynonymous with cultural identity.

Pastoralists thus find themselves trapped betweencultural and Darwinian imperatives. Whereas AK-47raiding may be the most effective strategy by which tomaintain their pastoralist subsistence and identity, itendangers their biological resilience and threatens thepersistence of human and livestock lineages in the pas-toralist system (see also N. Dyson-Hudson 1966:247).In Darwinian terms, it has intensified selection pres-sure on the Karimojong through its impact on mortalityof adult males and children and on female fertility. Thisproposition raises questions regarding the position ofcultural behaviors in models of human ecosystems (Lit-

6. It is significant in this respect that, in contrast to other herdingsystems, East African pastoralism appears to have emerged as adistinct cultural adaptation well hefore the introduction of sed-entary agriculture and the rise of state societies in the region (seeGifford-Gonzalez 1998).

tie et al. 1990, 1999^), in which cultural institutionstypically have been viewed as parts of a smoothly work-ing homeostatic system. Clearly, such a simplistic,functional analysis is inappropriate in relation to mod-ern cattle raiding, regardless of its probable adaptivefunction in prehistory. Like the famous spandrels of SanMarco (Gould and Lewontin 1979), AK-47 raiding arisesat the intersection of set of environmental, political,and social vectors. In contradistinction to its architec-tural analogue, however, raiding with automatic weap-ons appears to have assumed a distinct trajectory, inthe process becoming an increasingly capricious driverin a nonequilibrial, nonhomeostatic system (Gray, Les-lie, and Akol 2002). In this view, AK-47 raiding operatesas a positive feedback effect in pastoralist societies; bydefinition, it can be aborted only by a change in stateeither of the effect itself or of its target. While thisunderstanding may illuminate the mechanisms bywhich maladaptive cultural behaviors become wide-spread, the outcome of this intrinsically evolutionaryprocess remains obscure. Gray (n.d.) has argued thattheir stubborn adherence to modern raiding does notbode well for the persistence of pastoralist cultures inAfrica in the long term, given the negative impact ofAK-47 raids on human welfare, economic development,and pastoralists' integration into the national andglobal economies.

The present escalation in AK-47 raiding does not de-rive from familiar environmental stressors such as cat-tle disease and droughts but rather promotes them andaggravates their impact on the human population (Hen-drickson, Armon, and Mearns 1998). This disjunctionbetween cultural imperatives and environmental con-straints does not negate a fundamental role of the en-vironment in the shaping of culture itself (broadly de-fined as a human behavioral character), nor does itimply in any way that the human capacity for cultureis distinct from our biological nature. Rather, it pre-sumes that conditions under which erstwhile adaptivecultural institutions become maladaptive may arise, atleast in part, from the internal structures of cultureitself (see also Madut Jok 1999). In this way, the processof cultural selection may preserve behavioral traits ina population that ultimately set the stage for its demise.With specific reference to AK-47 raiding, we are sug-gesting that cultural rationalizations may temporarilyoverride environmental and evolutionary sense. In as-serting that a cultural institution may constitute anindependent agent of selection, we confront the dualpotential of culture as both adaptation and Darwinianstressor. In effect we come full circle: although culturalstress may emerge independently of classic evolution-ary constraints, cultures, in the end, are only as per-manent as the populations that bear them and ulti-mately are themselves vulnerable to the vagaries ofevolutionary change.

GRAY ET AL. Cattle Raiding, Cultural Survival, and Adaptability \ S23

Comments

MONIQUE BORGERHOFF MULDER

Department of Anthropology, Graduate Group inEcology, and Center for Population Biology, Universityof California, Davis, Calif 95616, U.S.A.(mborgerhoffmulder@ucdavis.edu). 9 vii 03

Gray and colleagues have done a remarkable job in re-viving field studies in this war-torn corner of East Africa.They offer a wonderful brew of demography, ethnogra-phy, and sociopolitical history for the Karimojong, rep-resenting the very best tradition of anthropology as amultidisciplinary enterprise, and their paper deftly linksobservations on marital practices, political develop-ments, and NGO interventions into a story that is bothconvincing and shocking. While many of the elements—settlement as a cause of environmental degradation anddisease, bridewealth infiation as a cause and conse-quence of raiding, and unstable wealth investments pre-cipitating poor outcomes for morbidity and mortality—are familiar, all these pieces fit into the picture of chaosthat emerges for northern Uganda from newspaper sto-ries and brave NGO workers. While one might want toquibble with the explicit causality that Gray et al. at-tribute to AK-47 cattle raiding per se rather than to thebroader circumstances of what has been effectively beena civil war of several decades or, alternatively, to ques-tion some of the causal arrows in their argument (forinstance, unstable marriage not only causes but resultsfrom low fertility), I would far prefer to draw attentionto the high quality of the ethnographic, demographic,and historical fieldwork.

I am not overly impressed, however, with the theo-retical framework, which seems neither novel nor in-cisive. Gray et al. are certainly correct in claiming thatonce-adaptive hehaviors or institutions can become mal-adaptive in novel circumstances and that culturallytransmitted institutions can impose constraints on po-tential behavioral or institutional shifts that would seemto offer greater adaptive value. I have argued, for example,that Kipsigis agropastoralists who are running out of landin Kenya's Kericho highlands "should" adopt unigeni-ture to prevent total fragmentation of their landholdingsor dowry to attract wealthy sons-in-law (BorgerhoffMulder 1995) but do not do so because of deeply heldviews pertaining to equality among sons (and the insig-nificance of daughters) that derive from their pastoralistideology. Such arguments can be broadened to the studyof historical constraints on cultural variability in EastAfrica (Borgerhoff Mulder et al. 2001) and lie at the heartof major theoretical models regarding the relationshipbetween cultural and biological evolution (Boyd andRicherson 1985). In this respect Gray et al.'s theoreticalconclusions are hardly new.

Rather than explore further these intriguing dynamics,they seem content to emphasize that cultural institu-tions cannot be viewed as part of a smoothly working

homeostatic system (surely something we have beenaware of since the demise of structural-functionalism)and to stress the true but, again, hardly novel point thatfor pastoralists raiding and militarism are practicallysynonymous with cultural identity. These are all validobservations and are beautifully illustrated with the datafrom Karamoja but are already widely if not uniformlyaccepted. Furthermore, Gray et al. reveal some confusionbetween individual and group selection insofar as theyare puzzled that armed raiding is both "the chief mech-anism for cultural survival" and devastating for fertility,health, and survival. Regular Darwinian evolutionaryprocesses clearly have the potential to drive aggregateoutcomes into suboptimal states (Rogers 1988), and allindividually adaptive strategies come with costs (often,in the case of pastoralists, incurred by other householdmembers [Borgerhoff Mulder and Sellen 1994]); this iswhy behavioral ecologists are always talking about trade-offs and how these shift under new social, political, andenvironmental conditions. Karimojong armed cattleraiding comes with terrible costs, but how do these com-pare with the costs experienced by hypothetical clans orfamilies that abjure raiding altogether? Framed in thisway, armed raids are surely best seen as an adaptive re-sponse to novel sociopolitical circumstances and oppor-tunities—the punitive campaigns of governments, themisguided initiatives of development agencies, and thefiood of automatic weapons into the area—and areclearly motivated by a combination of personal protec-tion and private gain, both of which in this cultural andecological context require the amassing of livestock. Inmy view Fleisher's (2000) perspective remains un-scathed. The empirical content of this paper neverthelessremains a major anthropological contribution.

GiJNTHER SCHLEE

Max Planck Institute for Social Anthropology, P. O.Box 110351, 06017 Halle/Saale, Germany (schlee@eth.mpg.de). 27 vi 03

Some of the terms used by Gray et al. need translation;a reader who takes the Darwinian terms too literally mayget it all wrong. What they mean by "selection" is notselection for specific traits and thereby selection as anagent of biological evolution but simply mortality. And,correspondingly, the "Darwinian imperative" simplytells us to survive and to reproduce if we want to bearound in the next generation. Gray et al. do not discusswhether under the combined effects of cattle raiding anddrought the bearers of certain genes are more likely tosurvive than others. One might think that smaller in-dividuals, the ones with cheaper bodies in terms of nu-tritional needs, would under certain conditions have anadvantage, but that is not Gray et al.'s argument. Theydiscuss whether pastoralism among the Ateker will dis-appear because the bearers of this culture, obeying itsnorms, destroy each other physically. If this were to hap-pen, those who have managed to quit the system by be-coming urban elites or agriculturalists might manage to

S24 I CURRENT ANTHROPOLOGY Voluuie 44, Supplement, December 2003

survive even though they bear the same genetic traits.Gray et al. are writing, then, about a cultural system andthe chances of survival of its bearers, not about biologicalevolution. That they do so in Darwinian terminologymay be confusing.

Given these clarifications, the argument is convincingand substantiated by compelling empirical detail. Atekerculture has become self-destructive. Ateker cattle raidershurt their communities not by being free riders or byacting against the norms of their culture but by obeyingcultural norms and thereby initiating a process of violentescalation that may lead to the end of Ateker culture bythe physical annihilation of its bearers. This point iscounterintuitive. The popular assumption is that cultureis good and mostly adaptive. Cattle raids have alwaysbeen a part of Ateker life, and not much has changedapart from the technology of killing, which was revo-lutionized by the mass influx of AK-47S. The formerlyadaptive level of violence—just enough to keep the pred-atory state at a distance—appears to have escalated be-yond a level that can be regulated by local forces. Becausethis point has been made well, I shall just pose somequestions that Gray et al. have not addressed in thispaper and that might form the topic of the next one.

How do Ateker explain what is going on? Culturalsystems that are self-destructive are not new to us. In ahighly competitive market economy people may beforced to use nonrenewable resources faster than theywould like to and may knowingly destroy the basis oftheir economic activities, regretting the weakness or ab-sence of overarching institutions that would preventthem and their competitors from doing so. Are Atekeractors aware that what they are doing is unsustainable?Is there an emic theory that corresponds to Gray et al.'s?If not, it might be interesting to hear alternative Atekertheories about Ateker.

If the Ateker do not develop institutions capable ofstopping the vicious circle that Gray et al. describe,might outside intervention achieve this? Can any policyadvice be derived from this paper? Much effort is beingdevoted to small-arms control, and the UN spends con-siderable money on it. I have been asked in my role asan "expert" in the Somali peace conference in Kenya topersuade my fellow Germans or Europeans to send anarmy to disarm the Somali, because they are unable todo it themselves. At the same time, Knighton (n.d.), writ-ing about the same area as Gray et al., warns againstunbalanced disarmament and points to the possibility ofmarauding state agents' attacking unarmed civilians.

Another question is whether there are thresholds ofescalation beyond which irreversible processes of mutualdestruction set in and deterrence and the preservation ofgroup identities are no longer possible. Turton (1994:27-30) seems to describe a case in which an infiux ofAK-47S helped the Mursi, not far from Karamoja insouthwestern Ethiopia, to keep the Nyangatom (who arepart of the Ateker) at a distance and preserve their groupidentity for the time being. This case clearly deviatesfrom the model described by Gray et al., which may haveto be expanded to accommodate it.

DAN SELLENDepartment of Anthropology, Emory University, is57Pierce Dr. NE, Atlanta, Ga. 30322, U.S.A. (dsellen®emory.edu). 24 vi 03

Gray and colleagues are to be commended for raisingsome important questions about the negative impact onEast African pastoralist lifeways of the militarization ofboth government and peasantry and the increased avail-ability of modern weapons. The retrospective evidencepresented for an association between a shift in raidingpractice and reduced livelihood security, reduced fertil-ity, reduced marital stability, and age- and sex-specificincreases in mortality is certainly plausible, if hardlydirect and not subjected to formal epidemiological test.I hope that their paper will stimulate further research onthis urgent humanitarian, environmental, and social is-sue and suggest four questions that might be pursued.

First, how has the increased stochasticity of maritalarrangements (and, presumably, of livestock holdingsand wealth differentials) attributed to AK-47 raiding un-dermined the status of women? Observations from pas-toral populations subsisting in more peaceful circum-stances do lend indirect support to the hypothesis thatKarimojong demographic shifts may be linked tochanged patterns of household food production, caregiv-ing, and health seeking, but the changing role of womenin more violent societies (particularly as agents ratherthan victims) remains understudied. For example, pas-toral women's work is often highly valued when the pro-duction returns from animal husbandry or alternativesare high (Brockington 2001, Horowitz and Jowkar 1992),and this influences household formation strategies (Sel-len 1999, Sellen, Borgerhoff Mulder, and Sieff 2000) andincreases women's economic autonomy (Fratkin 1989,Talle 1988). This may not be so when household herdsize and persistence depend mainly on herd-ownerfirepower.

Second, is AK-47 raiding really cultural maladapta-tion? Natural selection occurs only when excess deathsare linked to heritable phenotypic differences. The highcontribution of violence to deaths among men and teen-agers, high maternal mortality, and very high subadultmortality rates in comparison with those in other pas-toralists and agropastoralists (Sellen and Mace 1999) arecause for concern. However, the most commonly re-ported causes of death remain those related to infectionand undernutrition, and the data presented show an un-derlying secular decrease in infant and child mortality.Thus, unless there is something genetically differentabout those killed by gun violence, the data do not wellsupport the interpretation that increased raiding alterseither the direction or the intensity of selection. From ahuman behavioral ecological perspective, the salientevolutionary issues may be how survivors find strategiesto cope with the threat of violence and negotiate newsocial relations.

Third, how are we to tease out the independent effectsof more lethal weaponry from the accompanying social,political, and economic upheaval the Karimojong have

GRAY ET AL. Cattle Raiding, Cultural Survival, and Adaptability \ S25

experienced? Few details are presented about the specificforms of cattle raiding and their historical role in com-munity organization and ecological adaptation prior tothe acquisition of AK-47S in 1979 or ahout subsequentchanges in the social organization of raiding parties, thefrequency of cattle raids, and the numher of deaths perraid. It would he useful to know how an increased threatfrom armed raiders influences decisions ahout herd man-agement, mobility, and allocation of household lahor todefense and other activities (Borgerhoff Mulder and Sel-len 1994, Coppolillo 2000).

Fourth, can understanding of hoth the individual mo-tives for specific raiding activities and the geopolitics oflivestock markets help color the sketch of what is goingon in Karimojong? Surely detailed comparative studiesare needed hefore we can reject suggestions (Fleisher1998, 1999; Hendrickson, Armon, and Mearns 1998) thatshifts in raiding practice are hest understood in terms ofindividual motivations and changed patterns of conflictand interest among kin, affines, and others. For example,among the Datoga (Barahaig) and Maasai resident in theMhulu, Eyasi, and Ndutu regions of Tanzania in the early1990s, young, unmarried men in the warrior age-setssometimes sought to ohtain handguns and semiauto-matics in order to increase their ahility to kill "enemiesof the people" (natural predators and raiders from otherethnic groups) with the aim of more quickly acquiringcattle, through the usual traditional mechanisms of cat-tle gifts, and thus hrides and prestige. Most of them werefrustrated in this hy the high cost, illegality, and limitedsupply of such weaponry, the stern opposition of mostelders, and a reasonahly effective system of local lawenforcement in response to internecine murders. With-out guns and with severe consequences likely, attemptsat "hit-and-run" killings appeared far more commonthan successful ones, and few men acquired livestockdirectly through raiding. Nevertheless, attitudes at thecommunity level were changing in response to the ap-preciahle losses of livestock to organized raiding hy out-side groups armed with semiautomatic weaponry (andeven helicopter gunships, hy some accounts), which isone of several factors that contrihuted to the gradualimpoverishment of Datoga families (Sellen 2003, Sieff1999). We heard that some armed raiders stole cattle toorder for wealthy traders hased in Nairohi and elsewhereand that the cattle were rehranded and rustled to marketsin Kenya with the assistance or collusion of police, cus-toms officers, politicians, and businessmen in hoth coun-tries. We were unahle to verify these reports, and it ispossihle that many were emhellished or were "ruralmyths" reflecting general concerns ahout the failure ofgovernment to represent and protect the interests of thenomads and traditional hostilities to other clans and eth-nic groups. In any case, the lower frequency of armedraiding in northern Tanzania is prohahly linked as muchto increasing economic growth and diversification andthe relative stahility of traditional local and national po-litical systems as to continued difficulties in ohtainingnontraditional weapons.

Reply

SANDRA GRAYLawrence, Kans., U.S.A. 14 viii 03

All three commentaries point to weaknesses in the the-oretical argument of the paper, and I concede that thisis somewhat tentative as it stands. We have argued thatAK-47 raiding negates pastoralism as a suhsistence strat-egy in East Africa—a conclusion apparently more tenahlethan our corollary hypothesis that it acts as a Darwinianstressor in pastoralist populations. Demonstration ofspecific Darwinian effects of the violence is prohlematic,to he sure. In addition to the perpetual challenge of dis-tinguishing hetween group and individual effects, it re-quires that we track associations hetween distal effectsof violence and the proximal variahles that mediatethem. As Sellen notes, the demographic effects of cattleraiding discussed in this paper are indirect, apart fromthe deaths of (mainly male) victims of gunshot wounds.Distal effects are expected to he mediated hy a complexinteraction of intermediate variahles ranging from cul-tural responses—altered marriage practices, for exam-ple—to individual physiological responses that may heunder strong genetic control, such as the function of theendocrine system in response to chronic psychosocialstress. Any of these may promote intergenerationalchanges in gene frequencies as a result of selectionagainst phenotypes that are less hiologically resilient orless likely to reproduce (less marriageahle or more likelyto die in their prereproductive years).

Both Borgerhoff Mulder and Sellen note the need toquantify presumptive selection against individual phe-notypes. The questions who dies in raids and whetherthis mortality correlates with specific phenotypes areapropos. The existing evidence is suggestive. In 1998-99I ohserved that the physical maturation of Karimojongschoolhoys appeared to he late relative to that of herdhoys, although the small sample sizes of the latter pre-clude statistical analysis of the differences using existingdata. Some schoolhoys in the sample who were in theirlate teens showed no signs of the onset of puherty, andtheir age-set initiations—a necessary prelude to activeparticipation in raids—were ohserved to he similarly de-layed. Aside from their ahsence from the seasonal mi-grations and cattle camps that are the targets of raids, Ihypothesize that schoolhoys are unlikely to engage inraiding until after their third decade, if at all. Those whoremain in the pastoral system, however, hegin raiding intheir late teens or twenties and are more likely to die asa consequence of their herding activities, residence atcattle camps, and active participation in raids. Theirhigher risk of mortality at all ages after approximatelyage ten may result in their leaving fewer offspring thantheir schoolhoy counterparts. If hoys who hecome herd-ers are more physically rohust, such selection may in-fluence several biological parameters in the populationin succeeding generations, including hut not limited to

S26 I CURRENT ANTHROPOLOGY Volume 44, Supplement, December 2003

stature, hody composition, and endocrine profiles. Aquestion with ohvious relevance for our argument iswhether the segregation of fast and slow developers or,alternatively, of hig and small hoys is culturally dictated.

Which of these raider-herders ultimately survive alsohas implications for microevolutionary changes in thepopulation. In addition to ohvious physical traits suchas rate of maturation, size, physique, and physical prow-ess, personality traits such as aggression, response tostress, intelligence, and emotional resiliency may he ma-jor predictors of individual survival in a society anchoredin violence. Maternal hiological and hehavioral traits areexpected to infiuence all of these hypothetical outcomes.Caretaking hehavior, which appears to have heen alteredhy the violent social context (Gray n.d.), may have directeffects on child health, child development, and child sur-vival. More important, however, may he maternal traitsthat underlie women's apparent acquiescence in raidingand killing and related maternal strategies that reinforceaggressive traits in their children and are effective inenculturating them to violence. Heritability of such he-havioral phenotypes has direct hearing on the mecha-nisms hy which a militant ethos is transmitted cross-generationally and on its population consequences insucceeding generations.

The second part of our theoretical argument—that AK-47 raiding is maladaptive in the strict Darwinian sense—assumes that if unchecked these hypothesized effectsmay precipitate hoth cultural and population extinction.My purpose in putting forward a hypothesis that is onthe surface suhlimely untestahle is to shift the focusaway from cultures as an adaptive response to environ-mental stress to culture as a kind of stressor. A widelytheorized but inadequately quantified feature of culturalsystems merits consideration here—that cultural insti-tutions may be utterly dissociated from evolutionary im-peratives and may perversely undermine adaptive re-sponses. In fact, culture is a mechanism hy which humangroups may collectively choose to preserve maladaptivetraits which otherwise would he eliminated hy naturalselection. Once separated from their ecological and hi-ological hrakes—that is, from their classic Darwinianconstraints—these institutions may assume their owndestructive trajectories, with the result that they cometo exist "simply" for culture's sake and to satisfy somepeculiar human need for "cultural" identity, an inher-ently ideological trait. Cultural selection does not op-erate according to prescrihed Darwinian rules, given thatit involves human manipulation. Nor does it conformto definitions of artificial selection, since the changes inphenotype frequencies hypothesized here are presuma-bly not purposive. Furthermore, cultural hehaviors withthe potential to eliminate entire herding lineages wouldappear to be irreconciahle with individual interests, yetthey are, inexplicahly, individually compelling. The pro-cess is analogous to "runaway" selection (Fisher 1930),since hoth phenomena involve individuals' adhering toa particular hiohehavioral strategy beyond the point ofsustainahility simply hecause that is what other mem-bers of the group do.

What checks the rate of culture change in most casesis that cultural behaviors are parts of a complex of in-terconnected institutions. With regard to the singularcase of the Karimojong, I hypothesize that AK-47 raidingdislodged or nullified other cultural structures that mayonce have modulated its effects (the transformation ofKarimojong raiding itself has heen examined at lengthin Gray 2000 and n.d.), and Schlee also wonders whetherthe adaptive transformation of cattle raiding by the adop-tion of automatic weapons has crossed a hypotheticalthreshold heyond which the cultural system is no longereither effective or sustainahle. Having assumed its ownindependent and nonhomeostatic trajectory, raiding nowoperates as a distinct stressor independently of or, morecorrectly, in opposition to hoth environmental and othercultural pressures.

Borgerhoff Mulder rightly ohserves that the revelationthat cultural systems are nonhomeostatic is old news,but awareness of a concept is not synonymous with itsintegration into prevailing models. In fact, both she andSellen lean toward a view of AK-47 raiding as adaptive,and adaptation is inherently homeostatic: the Red Queen(Van Valen 1973) runs as fast as she can to stay roughlyin the same place while the environment shifts abouther. Culture, in contrast, is predicated on the capacityfor rapid hehavioral change in response to an unpredict-ahle environment. The divergence of rates of genetic andcultural change gave humans a distinct evolutionary ad-vantage hecause in most instances it allowed them tocircumvent the Red Queen's dilemma altogether. Evi-dently, however, there are evolutionary limits to themagnitude of this rate disparity.

Schlee ohserves that AK-47 raiding need not result inthe hleak evolutionary scenario we predict if pastoralistssimply adopt some other form of suhsistence that is moreadaptive in their current environmental, socioeconomic,and political situation. That most Karimojong are strik-ingly averse to doing so requires that we entertain othercontingencies. Forcihle dispersal of the entire Karimojongpopulation if they persist in their present strategy has heenhaldly proposed hy hoth memhers of Parliament and otherinfiuential ohservers in Uganda (Gray 2000, Mamdami etal. 1994). Such a draconian measure is equivalent to pop-ulation extinction in the genetic sense, and any autoch-thonous hehavior that promotes this outcome must heviewed as maladaptive in the strict biological sense. So-cioeconomic alternatives have heen presented to the Kar-imojong, hut these programs have done little more thanreinforce intra- and interethnie differences. Ultimatelythey appear to have been exploited by the Karimojong toenhance or to alter access to guns and cattle (Gray 2,000).In this case ideology has effectively prohihited the trade-offs that Borgerhoff Mulder hypothesizes.

Long-term investigation of these hypotheses is sched-uled to begin in early 2004, Proposals pending at variousfunding agencies include a study of women's roles andwomen's agency in Karimojong society since the adop-tion of AK-47 raiding, one of the four problems enu-merated by Sellen, The research will examine how thechanging reproductive and social value of Karimojong

GRAY ET AL. Cattle Raiding, Cultural Survival, and Adaptability \ S27

women has infiuenced their reproductive ideologies, re-sulting in their complicity in the violence as a means toenhance their own reproductive success, social status,and survival. A proposed study of the women's genera-tion-set system among the Karimojong focuses on therole of other cultural institutions in moderating inter-and intraethnic hostilities. This system, which shouldnot he confused with more informal female age-grades,may have functioned to restrain the militancy of its malecounterpart, for which organizing and executing raidswas a major occupation (Lamphear 1998). The femalesystem appears to have fallen into disuse sometime inthe 1970s, however; its ahandonment thus coincideswith widespread adoption of AK-47 raiding. Anotherstudy is in preparation on the structure of raiding: whoraids, why, and when (and the related question of whohas access to the guns).' The ohjective of this study isto measure costs and henefits of individuals' complicityin raiding in terms of survival, marriageahility, and re-productive success. Examination of the relationships he-tween the individual and the group effects of these strat-egies will explore hypothesized tradeoffs. A longitudinalstudy of maternal strategies and of child survival, childmorbidity, and child growth and development will heundertaken to ascertain how women's hehavior andthinking are infiuenced by the violence and to quantifyrelationships hetween maternal hehavior, personal char-acteristics of mothers and their children, and childoutcomes.

Schlee is concerned with how this research may beused to assist efforts to stem the violence in eastern Af-rica and asks what is known of the opinion of the Kar-imojong themselves regarding the sustainahility of theirpresent strategy for cultural survival. Whereas the emo-tional impact of individuals' experience of violence waspatent during their interviews, data on the effects of col-lective experience on ideological structures are lackingat present. In 1998-99 a few Karimojong expressed theirconcern that the guns would result in the total annihi-lation of their people, hut these were all memhers of aneducated elite that is largely alienated from the pastor-alist system. It seems to me that collective awarenessthat their individual actions imply costs for populationsurvival in the long term, regardless of ostensible short-term benefits, may compel herders to consider alterna-tive strategies more seriously than has been the case withpeace initiatives to date.

The overarching ohjective of the proposed fieldworkis to collect empirical hiobehavioral evidence of culturalhehavior as maladaptation. Specifically, the goal is toquantify the role of collective violence as an agent ofcultural selection that may shape hiohehavioral varia-bility in a population. Unlike many studies of humanwarfare and aggression in hiological anthropology or he-

I, Differential access to weapons and ammunition may be a majorfactor influencing both individual and group behaviors. However,the routes of small-arms flow and the means by which they areprocured are closely guarded information in Karamoja, Their in-vestigation raises thorny ethical and methodological issues as wellas obstacles to funding.

havioral ecology (see, for example, Betzig 1997 and Mar-tin and Frayer 2000), this work is less concerned withwhat or why human violence is than with what it does.I have elected, for the present, to treat the origins ofhuman violence as distinct from its impact on the struc-ture of modern human populations. Ultimately we mayfind ourselves concerned with the essence of human vi-olence itself, but such a discussion is largely irrelevantunless preceded hy the empirical study of its effects onthe persistence of populations. It is precisely in the areaof the latter that our methodology, data, and theory arelacking.

References Cited

AKOL, H, A,, AND s, J, GRAY, Event Calendar for Bokoraand Matheniko counties, Moroto District, Uganda, 1910-1999,MS,

ALLEN, T, 1996. "A flight from refuge," in In search of coolground: War, flight, and homecoming in Northeast Africa. Ed-ited by T Allen, pp, 220-61, Oxford: James Currey,

ALNWiCK, D, J, 1981, Towards a food aid policy for Kara-moja: A report to UNICEF, fune 2, 1981. Kampala: UNICEFKaramoja Emergency Relief Program,

, 1985, "The 1980 famine in Karamoja," in Crisis inUganda: The breakdown of health services. Edited by C, P,Dodge and P, D, Wiebe, pp, 127-44, Oxford: Pergamon Press,

BARBER, J, 1968, Imperial frontier. Nairobi: East African Pub-lishing House.

BARTON, T,, AND G. WAMAi, 1994. Equity and Vulnerabil-ity: A situation analysis of women, adolescents, and childrenin Uganda, 1994. Kampala: Government of Uganda, UgandaNational Council for Children,

BETZIG, L, Editor,i997, Human nature: A critical reader. NewYork: Oxford University Press,

BIELLIK, R, J,, AND P, L, HENDERSON, 1981, Mortality,nutritional status, and dietary conditions in a food deficit re-gion: North Teso District, Uganda, December, 1980, Ecology ofFood and Nutrition 11:163-70,

, 1985, "Mortality, nutrition status, and diet during thefamine in Karamoja, Uganda, 1980," in Crisis in Uganda: Thebreakdown of health services. Edited by C, P, Dodge and P. D,Wiebe, pp, 145-54, Oxford: Pergamon Press,

BORGERHOFF MULDER, M. 1995, Bridewealth and its Corre-lates: Quantifying changes over time, CURRENT ANTHROPOLOGY36:573-603, [MB]

BORGERHOFF MULDER, M, , M, G E O RG E - CRAMER, J,ESHELMAN, AND A. ORTOLANi, 2001, A Study of East Af-rican kinship and marriage using a phylogenetically basedcomparative method, American Anthropologist 103:1059-82,[MB]

BORGERHOFF MULDER, M, , AND D, W, SELLEN, I994 ,"Pastoralist decision-making: A behavioral ecological perspec-tive," in African pastoralist systems: An integrated approach.Edited by E, Fratkin, K, Galvin, and E, Roth, pp, 205-30, Boul-der: Lynne Rienner, |MB, DS]

BOUCKAERT, p, i9C)ci. HostHe to democracy: The movementsystem and political repression in Uganda. New York: HumanRights Watch,

BOYD, R,, AND p, J, RicHERSON, 198$. Culture and evolu-tionary progress. Chicago: University of Chicago Press, [MB]

BRAiNARD, J, 1991, Health and development in a rural Ken-yan community. New York: Peter Lang Publishing,

BRETT, E, A, 1996, "Rebuilding War damaged Communities inUganda," in In search of cool ground: War, flight, and home-

S28 I CURRENT ANTHROPOLOGY Volume 44, Supplement, December 2003

coming in Northeast Africa. Edited by T. Allen, pp. 203-19.Oxford: James Currey.

BROCH-DUE, V. 1999. "Remembered Cattle, forgotten people:The morality of exchange and exclusion of the Turkana poor,"in The poor are not us: Poverty and pastoralism in Eastern Af-rica. Edited by V. Broch-Due and D. M. Anderson, pp. 50-88.Oxford: James Currey.

BROCKINGTON, D. 2001. Women's income and the livelihoodstrategies of dispossesed pastoralists near the Mkomazi GameReserve, Tanzania. Human Ecology 29:307-38. [DS]

CAMPBELL, B. C , P. W. LESLIE, M. A. LITTLE, J. M.BRAINARD, AND M. A. DELUCA. 1999. "Settled Tur-kana," in Turkana herders of the dry savanna: Ecology andbiobehavioral response of nomads to an uncertain environ-ment. Edited by M. A. Little and R W. Leslie, pp. 333-52. Ox-ford: Oxford University Press.

CDC (CENTERS FOR DISEASE C O N T R O L ) . 1992. Famine-af-fected, refugee, and displaced populations: Recommendationsfor public health issues. Morbidity and Mortality Weekly Re-view 41 (RR-13).

CISTHRNINO, M. 1979. Karamoja, the human zoo: The historyof planning for Karamoja vtfith some tentative counterplanning.Ph.D. diss., Post-Graduate School for Development Studies,University of Swansea, Swansea, Wales.

. 1985. "Famine and food relief in Karamoja," in Crisis inUganda: The breakdown of health services. Edited by C. P.Dodge and P. D. Wiebe, pp. 155-61. Oxford: Pergamon Press.

CLEAVE, J. 1996. "First posting," in Looking back at the Ugan-dan Protectorate: Recollections of district officers. Edited byD. Brown and M. V. Brown, pp. 30-38. Dalkeith: DouglasBrown.

COALE, A. J., AND P. DEMENY, 1983. i d edition. Regio^flimodel life tables and stable populations. New York: AcademicPress.

COPPOLILLO, p. B. 2000. The landscape ecology of pastoralherding: Spatial analysis of land use and livestock productionin East Africa. Human Ecology 28:527-60. [DS]

DAVIS, K., AND J. BLAKE. 1956. Social Structure and fertil-ity: An analytic framework. Economic Development and Cul-ture Change 4:211-35.

DAWES, M. 2001. Hundreds die in Uganda cattle raid. BritishBroadcasting Company Online Network. http//news.bbc.co.uk/hi/english/world/africa/newsid_446ooo/446o4i.stm.

DEVLIN, H. 1998. Patterns of morbidity in Karamoja, Uganda,1992-1996. M.A. thesis. Department of Anthropology, Univer-sity of Kansas, Lawrence, Kans.

DE WAAL, A. 1987. Famine mortality: A case study of Darfur,Sudan 1984-5. Population Studies 43:5-24.

DODGE, c. p., AND P. D. WIEBE. Editors. 1985. Crisis inUganda: The breakdown of health services. Oxford: PergamonPress.

DYSON-HUDSON, N. 1966. KarimojongpoUtics. Oxiotd. Clar-endon Press.

. 2000. "Processes of discovery and modes of commentarythrough time: Karamoja and Turkana as illustrative cases, "inAbstracts of the 99th Annual Meeting of the American An-thropological Association, p. 190. Arlington: American Anthro-pological Association.

D Y S O N - H U D S O N , N . , AND R. D YS O N - H U D S O N . I 9 9 9 ."The social organization of resource exploitation," in Turkanaherders of the dry savanna: Ecology and biobehavioral re-sponse of nomads to an uncertain environment. Edited by M.A. Little and P. W. Leslie, pp. 69-86. Oxford: Oxford UniversityPress.

DYSON-HUDSON, R. 1960. Men, women, and work in a pas-toral society. Natural History 69:42-56.

. 1999. "Turkana in time perspective," in Turkana herdersof the dry savanna: Ecology and biobehavioral response of no-mads to an uncertain environment. Edited by M. A. Little andP. W. Leslie, pp. 24-40. Oxford: Oxford University Press.

D Y S O N - H U D S O N , R. , AND J. T. MC CABE. 1985. SoUthTurkana nomadism: Coping with an unpredictably varying en-vironment. New Haven: Human Relations Area Files, Inc.

ELLIS, j . M., AND D. M. SWIFT. 1988. Stability of Africanpastoral ecosystems: Alternate paradigms and implications fordevelopment. Journal of Range Management 41:450-59.

FISHER, R. A. I9io. The genetical theory of natural selection.Oxford: Clarendon Press.

FLEAY, M. 1996. "Karamoja's district team and developmentscheme," in Looking back at the Ugandan Protectorate: Recol-lections of district officers. Edited by D. Brown and M. V.Brown, pp. 20-25. Dalkeith: Douglas Brown.

FLEISHER, M. L. 1998. Cattle raiding and its Correlates: Thecultural-ecological consequences of market-oriented raidingamong the Kuria of Tanzania. Human Ecology 26:547-72.

. 1999. Cattle raiding and household demography amongthe Kuria of Tanzania. Africa 69:238-55.

-. 2000. Kuria cattle raiders: Violence and vigilantism onthe Tanzania/Kenya frontier. Ann Arbor: University of Michi-gan Press, [MB]

FRATKIN, E. 1989. Household Variation and gender inequalityin Ariaal pastoral production: Results of a stratified time-allo-cation survey. American Anthropologist 91:430-40.

FUKUi, K., AND D. TURTON. Editors. 1979. Wflr/flre flmongEast African herders. Senri Ethnological Studies 3.

GALVIN, K. 1985. Food procurement, diet, activities, and nutri-tion of Ngisonyoka Turkana pastoralists in an ecological andsocial context. Ph.D. diss.. State University of New York, Bing-hamton, N.Y.

G A R R U T O , R. , M. A. L I T T L E , G. D . TAMES, AND D . E.BROWN. 1999. Natural experimental models: The globalsearch for biomedical paradigms among traditional, moderniz-ing, and modern populations. Proceedings of the NationalAcademy of Sciences, U.S.A. 96:10536-43.

GERTZEL, c. 1988. The poHtics of uneven development. Flin-ders University of South Australia Discussion Paper 20. Kam-pala: UNICEF.

GIFFORD-GONZALEZ, D. 1998. Early pastoralists in East Af-rica: Ecological and social dimensions. Journal of Anthropologi-cal Archeology 17:166-200.

GOU ( G O V E R N M E N T OF UGANDA]. 1994.. All eyes and na-tional attention to Karamoja: Proceedings of a national confer-ence on peace and sustainable development for Karamoja andneighboring districts, r8-22 July rpp4. Kampala: Ministry ofState for Karamoja Affairs.

GOULD, s. J., AND R. c. LEWONTiN. 1979. The Spandrelsof San Macro and the Panglossian paradigm: A critique of theadaptationist programme. Proceedings of the Royal Society ofLondon B 205:581-98.

GRAY, s. J. 1992. Infant care and feeding among nomadic Tur-kana pastoralists: Implications for child survival and fertility.Ph.D. diss.. State University of New York, Binghamton, N.Y.

. 1994. Correlates of dietary intake of lactating women inSouth Turkana. American Journal of Human Biology 6:369-83.

-. 1998. Butterfat feeding in early infancy in African popula-tions: New hypotheses. American Journal of Human Biology10:163-78.

. 1999. "Infant care and feeding," in Turkana herders ofthe dry savanna: Ecology and biobehavioral response of no-mads to an uncertain environment. Edited by M. A. Little andP. W. Leslie, pp. 165-85. Oxford: Oxford University Press.

-. 2000. The memory of loss: Ecological politics, local his-tory, and the evolution of Karimojong violence. Human Organ-ization 59:401-18.

-. 2002. Growth in puberty of pastoralist children in Tur-kana, Kenya, and Moroto, Uganda: A life history perspective.Poster presented at the annual meeting of the Human BiologyAssociation, Buffalo, N.Y., April 14-16.

-. n.d. "The experience of violence and pastoralist identityin southern Karamoja," in Changing identifications and alli-ances in North-Eastern Africa. Edited by G. Schlee. London:James Currey.

GRAY, s. I . , AND H. A. AKOL. 2000. Reproductive historiesand life history of Bokora and Matheniko Karimojong womenof northeast Uganda. American Journal of Physical Anthropol-ogy supp. 30:191.

GRAY ET AL. Cattle Raiding, Cultural Survival, and Adaptability \ S29

GRAY, s. J., P. w. LESLIE, AND H. A. AKOL. 2OO2. "Un-certain disaster: Environmental instability, colonial policy, andthe resilience of East African pastoralist systems," in The lastof the nomadic herders. Edited by W. R. Leonard and M. H.Crawford. Cambridge: Cambridge University Press. In press.

GRAY, s. J., AND I. L. PIKE. 1998. Morbidity, pregnancyoutcomes, and fitness costs of sedentarization among pastoral-ist women in Uganda. American Journal of Physical Anthro-pology supp. 26:100.

GULLIVER, p. H. ig$i. A preliminary survey of the Turkana:A report compiled for the Government of Kenya. University ofCape Town Communication from the School of African Stud-ies, n.s., 26.

. 1955. The family herds. London: Routledge and KeganPaul.

HENDRIGKSON, D., J. ARMON, AND R. MEARNS. 1998.The changing nature of conflict and famine vulnerability: Thecase of livestock raiding in Turkana District, Kenya. Disasters22:185-99.

HOROWITZ, M. A., AND F. JOWKAR. l^^l. PaStOralwomen and change in Africa, the Middle East, and CentralAsia. Institute for Development Anthropology 91. [DS]

HUTCHINSON, s. E. 1996. Nuer dilemmas: Coping withmoney, war, and the state. Berkeley: University of CaliforniaPress.

JELLIFFE, D. B., F. J. BENNETT, E. F. P. JELLIFFE,AND R. H. R. WHITE. 1964. Ecology of childhood disease inthe Karamojong of Uganda. Archives of Environmental Health9:25-36.

KiDON-UNYANG, M. 1987. War fOr Karamoja. New A/ricfliJ,February, pp. 20-31.

KNIGHTON, BEN. n.d. The state as raider among the Karimo-jong. Africa. In press. [GS|

KNUTSSON, K.-E. 1985. "Preparedness for disaster operations,"in Crisis in Uganda: The breakdovm of health services. Editedby C. P. Dodge and P. D. Wiebe, pp. 183-89. Oxford PergamonPress.

KUGZMARSKI, R. J., C. L. OGDEN, L. M. GRUMMER-STRAWN, ET AL. 2000. CDC growth charts: United States,Advance data from vital and health statistics, no. 3r4. Hyatts-ville: National Center for Health Statistics.

KURIMOTO, E., AND s. s iMONSE. Editors. 1998. Conflict,age, and power in North East Africa. Oxford: James Currey.

LAMPHEAR, j . 19J6. The traditional history of the Jie ofUganda. Oxford: Clarendon Press.

. 1988. The people of the grey bull: The origin and expan-sion of the Turkana. Journal of African History 29:27-39.

-. 1992. The scattering time: Turkana responses to colonialrule. Oxford: Clarendon Press.

-. 1998. "Brothers in arms: Military aspects of East Africanage-class systems in historical perspective," in Conflict, age,and power in North-East Africa. Edited by E. Kurimoto and S.Simonse, pp. 79-97. Oxford: James Currey.

LESLIE, p. W., K. L. GAMPBELL, B. C. CAMPBELL, C.s. KiGONDu, AND L. H. KiRUMBi. 19990. "Eecundityand fertility," in Turkana herders of the dry savanna: Ecologyand biobehavioral response of nomads to an uncertain envi-ronment. Edited by M. A. Little and P. W Leslie, pp. 249-78.Oxford: Oxford University Press.

LESLIE, p. W., R. DYSON-HUDSON, AND P. H. FRY.1999. "Population replacement and persistence," in Turkanaherders of the dry savanna: Ecology and biobehavioral re-sponse of nomads to an uncertain environment. Edited by M.A. Little and P. W. Leslie, pp. 281-301. Oxford: Oxford Univer-sity Press.

LESLIE, p. W., R. DYSON-HUDSON, E. A. LOWOTO,AND J. MUNYESi. I999t>. "Ngisonyoka event Calendar," inTurkana herders of the dry savanna: Ecology and biobehav-ioral response of nomads to an uncertain environment. Editedby M. A. Little and P. W. Leslie, pp. 375-78. Oxford: OxfordUniversity Press.

LITTLE, M. A., R. DYSON-HUDSON, N. DYSON-HUD-SON, J. E. ELLIS, K. A. GALVIN, P. W. LESLIE, AND

D. M. SWIFT. 1990. "Ecosystem approaches in human biol-ogy: Their history and a case study of the South Turkana Eco-system Project," in The ecosystem approach in anthropology:From concept to practice. Edited by E. F. Moran, pp. 389-434.Ann Arbor: University of Michigan Press.

LITTLE, M. A., R. D Y S O N-HUD S ON, N. DYSON-HUD-SON, AND N. L. wiNTERBAUER, I999fl. "Environmentalvariations in the South Turkana ecosystem," in Turkana herd-ers of the dry savanna: Ecology and biobehavioral response ofnomads to an uncertain environment. Edited by M. A. Littleand P. W. Leslie, pp. 317-30. Oxford: Oxford University Press.

LITTLE, M. A., R. D Y S O N-HUD S ON, P. W. LESLIE, ANDN. DYSON-HUDSON. 1999b. "Framework and theory," inTurkana herders of the dry savanna: Ecology and biobehav-ioral response of nomads to an uncertain environment. Editedby M. A. Little and P. W. Leslie, pp. 2-23. Oxford: Oxford Uni-versity Press.

LITTLE, M. A., K. A. GALVIN, AND M. MUGAMBI.1983. Cross-sectional growth of nomadic Turkana pastoralists.Human Biology 55:811-30.

LITTLE, M. A., AND S. T. GRAY. I99O. GrOWth of yOUngnomadic and settled Turkana children. Medical AnthropologyQuarterly 4:296-314.

LITTLE, M. A., AND P. w. LESLIE. Editor. 1999. Turkanaherders of the dry savanna: Ecology and biobehavioral re-sponse of nomads to an uncertain environment. Oxford: Ox-ford University Press.

MC CABE, J. T. 1987. Drought and recovery: Livestock dy-namics among the Ngisonyoka Turkana of Kenya. HumanEcology 18:81-103.

. 1990. Success and failure: The breakdown of traditionaldrought coping institutions among the pastoral Turkana ofKenya. Journal of Asian and African Studies 25:146-60.

MC CABE, J. T., R. D YS ON-HUD S O N, AND J. WIEN-PAHL. 1999. "Nomadic movements," in Turkana herders ofthe dry savanna: Ecology and biobehavioral response of no-mads to an uncertain environment. Edited by M. A. Little andP. W. Leslie, pp. 109-21. Oxford: Oxford University Press.

MC DONALD, M. M. A. 1998. Early African pastoralism:View from Dakhlel oasis (South Central Egypt). Journal of An-thropological Archeology 17:124-42.

MADUT JOK, J. 1999. Militarism, gender, and reproductive suf-fering: The case of abortion in western Dinka. Africa 69:194-212.

MAjoK, A. A., AND c. w. scHWABE. 19^6. Developmentamong Africa's migratory pastoralists. Westport: Bergin andGarvey.

M A M D A N I , M . , P. M. B. KASOMA, AND A. B. KA-TE NDE. 1992. Karamoja: Ecology and history. Centre for Ba-sic Research Working Paper 22.

MARSHALL, F. 1990. Origins of Specialized pastoral productionin East Africa. American Anthropologist 92:873-94.

MARTIN, D. L., AND D. w. FRAYER. Editors. 1997. Trou-bled times: Violence and warfare in the past. Amsterdam:Gordon and Breach.

MEARS, c , AND I. RiLEY. 198J. Preliminary report: Nutri-tional survey of Dodoth County, Kotido District, Karamoja,i6th November to 4th December. Kampala: UNICEF.

MiRZELER, M., AND c . YOUNG. 2000. Pastoral politics inthe northeast periphery in Uganda: AK-47 as change agent.Journal of Modern African Studies 38:407-29.

MOHAMED, J. 1999. Epidemics and public health in early colo-nial Somaliland. Social Science and Medicine 48:507-21.

MOROTO DIOCESE. 199S. ipp4 annual report of Saint Kizito'sHospital, Mathany. Kampala: Moroto Diocese/Karamoja, Com-boni Missionaries.

OKUDi, B. 1992. Causes and effects of the 1^80 famine in Kar-amoja. Kampala: Makerere University, Center for BasicResearch.

PIKE, I. L. 1999. Age, reproductive history, seasonality, andmaternal hody composition during pregnancy for nomadic Tur-kana pastoralists of Kenya. American Journal of Human Biol-ogy 11:658-72.

S30 I CURRENT ANTHROPOLOGY Volume 44, Supplement, December 2003

. n.d. The biosocial consequences of life on the run: A casestudy of the Turkana of Kenya, Human Organization. In press,

piROUET, M, L, 1991, "Human rights issues in Museveni'sUganda," in Changing Uganda. Edited by H, B, Hansen and M,Twaddle, pp, 197-209, London: James Currey,

ROBERTS, D, E,, AND D. R. BAINBRIDGE, I963, Niloticphysique, American Journal of Physical Anthropology 21:341-70,

ROGERS, A. 1988. Does hiology constrain culture? AmericanAnthropologist 90:813-91, [MB]

R U T I S H A U S E R , I , H, E , , AND R, G, W H I T E H E A D , I 9 6 9 ,Field evaluation of two biochemical tests which may reflectnutritional status in three areas of Uganda. British lournal ofNutrition 23:1-13,

SELLEN, D. w. 1999, Polygyny and child growth in a tradi-tional pastoral society: The case of the Datoga of Tanzania,Human Nature: An Interdisciplinary Biosocial fournal io:329-71, [DSJ

. 2003. Nutritional consequences of wealth differentialsamong pastoral Datoga households in northern Tanzania. Hu-man Ecology. In press, [DS]

SELLEN, D, W,, M. BORGERHOFF MULDER, AND D, F.SIEFF, 2000, "Fertility, offspring quality, and wealth in Datogapastoralists: Testing evolutionary models of intersexual selec-tion," in Adaptation and human behavior: An anthropologicalperspective. Edited by L. Cronk, N. Chagnon, and W, Irons, pp.87-110, Hawthorne, N,Y,: Aldine de Gruyter, [DS]

SELLEN, D, w., AND R, MACE, 1999, A phylogenetic analy-sis of the relationship between sub-adult mortality and modeof subsistence, fournal of Biosocial Science 31:1-16. [DS]

S H E L L - D U N C A N , B , , J, K, SHELLEY, AND P, W. LESLIE .1999. "Health and morbidity: Ethnomedical and epidemiologi-cal perspectives," in Turkana herders of the dry savanna: Ecol-ogy and biobehavioral response of nomads to an uncertain en-vironment. Edited hy M. A. Little and P, W, Leslie, pp. 207-29,Oxford: Oxford University Press,

SHELLEY, J, K, 1985, Medicines for misfortune: Diagnosis andhealth care among southern Turkana pastoralists of Kenya,Ph,D, diss,. University of North Carolina, Chapel Hill, N,C,

SIEFF, D, F, 1999, The effects of wealth on livestock dynamicsamong the Datoga pastoralists of Tanzania. Agricultural Sys-tems 59:1-25, ]DS]

SUNDAL, M. 2002, Mortality and causes of death among Kari-

mojong agropastoralists of northeastern Uganda: 1940-1999,M,A, thesis. Department of Anthropology, University of Kan-sas, Lawrence, Kans,

TALLE, A. 1988. Women at a loss: Changes in Masaaipastoral-ism and their effects on gender relations. Stockholm: Univer-sity of Stockholm Press, ]DS]

TANNER, J. M, 1986, "Growth as a target-Seeking function:Catch-up and catch-down in man," in Human growth: A com-prehensive treatise, 2d edition, vol. i. Edited hy F, Falkner andJ, M. Tanner, pp, 167-79, New York: Plenum Press,

THOMAS, E, M. 1965. Wamor ierdsmeii. New York: Alfred A.Knopf.

TRDP (TURKANA RURAL DEVELOPMENT PROGRAMME).1989. Report of the health sector review, Turkana District,April, 1989. Nairobi: TRDP.

TURTON, DAVID, 1994, "Mursi political identity and warfare,"in Ethnicity and conflict in the Horn of Africa. Edited by Kat-suyoshi Fukui and John Markakis, pp, 15-31, London: JamesCurrey. [GS]

UNICEF, 1993, Moroto District nutrition survey, November1993. Kampala: UNICEF Health Section,

VAN VALEN, L, 1973, A new evolutionary law. fvo/utionflryTheory 1:1-30,

WHYTE, s, R, 1991, "Medicines and self-help: The privatiza-tion of health care in eastern Uganda," in Changing Uganda.Edited by H, B. Hansen and M, Twaddle, pp. 130-48, London:James Currey,

WIEBUSCH, B, 2002. Environmental effects on mixed-longitu-dinal growth in weight of immunized Karimojong children.M.A, thesis. Department of Anthropology, University of Kan-sas, Lawrence, Kans,

WILSON, J, G. 1962, Opportunities for pasture improvementin various ecological zones of Karamoja, East African Agricul-tural and Forestry Journal 27(special issue):47-48.

, 1985. "Resettlement in Karamoja," in Crisis in Uganda:The breakdown of health services. Edited by C, P, Dodge andP. D, Wiebe, pp, 163-70, Oxford: Pergamon Press,

WOOD, J. w. 1994.. Dynamics of human reproduction: Biology,biometry, demography. New York: Aldine de Gruyter,

WOODWARD, p, 1991, "Uganda and southern Sudan 1986-9:New regimes and peripheral politics," in Changing Uganda.Edited hy H. B, Hansen and M. Twaddle, pp, 178-86. London:James Currey,