Cannabis Use Among Aka Foragers of The

Home , Psychoactive plant, Spots (cannabis), Tobacco and other drugs

BEHAVIORAL ECOLOGY OF TOBACCO AND CANNABIS USE AMONG AKA FORAGERS OF THE

CONGO BASIN

CASEY JORDAN ROULETTE

A dissertation submitted in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

WASHINGTON STATE UNIVERSITY Department of Anthropology

MAY 2015

To the Faculty of Washington State University:

The members of the Committee appointed to examine the dissertation of CASEY JORDAN ROULETTE find it satisfactory and recommend that it be accepted.

______Edward H. Hagen, Ph.D., Chair

______Barry S. Hewlett, Ph.D.

______Robert J. Quinlan, Ph.D.

ACKNOWLEDGEMENT

I would like to thank my committee for their guidance, expertise, patience, and trust.

These are: Edward Hagen, my PhD chair, who introduced me to evolutionary approaches to substance

use and provided countless hours of mentoring and advice; Barry Hewlett, who opened my mind to anthropology and who introduced me to the Aka; Rob Quinlan, who introduced me to field work and urged

me to apply to the evolutionary anthropology program.

None of this would have been possible without the Aka, who graciously agreed to participate, who found

humor in the questions and tasks asked of them, and who provided assistance in the field. It has truly been a great experience working and spending time with them. I would also like to thank the people of the villages of Bagandou, Kenga, and Moboma, CAR, for their hospitality, especially my research assistants,

Mesmin Dopeningue, the late Nicaise Molende, Alaine Guy Kolet, Aubin Mongosso, and Eduard Mboula,

without who’s help the project never would have been completed.

There are several professors, colleagues, and associates who helped along the way.

These include Dr’s Marsha Quinlan, Bonnie Hewlett, Nicole Hess, Brian Kemp, Colin Grier, Courtney

Meehan, Cara Monroe, and Roger Sullivan, who provided additional support and guidance. Also, Haley

Mann, Jennifer Roulette, Mark Remiker, Mirdad Kazanji and the rest of the staff at Institute Pasteur in

Bangui, all of whom helped in data collection and/or analyses. Thanks also to Elsevier and John Wiley

and Sons for permission to reprint, as well as to the anonymous reviewers for their comments and

suggestions.

iii Finally, I would like to thank my family. These include my parents--Marie and Fred Snaza, Scott and

Debora Roulette, and Steve and Ellen Wilcox—as well as my sister, Saidi Roulette, for believing in and never giving up on me, and for all of their support. My wonderful daughter, Emma, who provided me with

a new focus and perspective that helped propel me through the finish line. And most importantly, my

amazing wife and colleague, Jennifer Ellen Wilcox Roulette, with whom I shared this journey and whose

love and support made it all possible.

Thank you all!

BEHAVIORAL ECOLOGY OF TOBACCO AND CANNABIS USE AMONG AKA FORAGERS OF THE

CONGO BASIN

Abstract

by Casey Jordan Roulette, Ph.D. Washington State University May 2015

Chair: Edward H.Hagen

Little is known about substance use among extant hunting-gathering populations. I therefore conducted one of the first biocultural, and biomarker validated, studies of tobacco and cannabis among the Aka foragers of the Congo Basin. Because tobacco and cannabis contain anthelmintic compounds, and because the Aka suffer from high rates of helminthiasis, I also tested a hypothesis that recreational use of neurotoxic plants helps defend against parasites.

Self- and peer-reports of tobacco and cannabis were collected from all Aka residing in the study area (n=379). Detailed questions about substance use were asked among a subset of these. Because female use was low, I restricted saliva, urine and stool sample collections to men. Saliva samples were assayed for cotinine, a nicotine metabolite; urine samples were assayed for THCA, a metabolite of THC; a subsample was genotyped for the CYP2A6 enzyme, which metabolizes nicotine. Stool samples were assayed for intestinal helminth eggs as an index of worm burden.

Aka men pay more for tobacco, yet have a higher smoking prevalence (95%) than men in most other populations, whereas Aka women have a low prevalence. Aka thus have one of the largest known gender differences in smoking. Tobacco is widely shared, and might play a central role in this defining aspect of Aka culture. Significant negative correlations between cotinine and worm burden and THCA and worm burden were found. Treating helminths with a commercial anthelmintic reduced cotinine concentration two weeks later, compared to placebo controls. Significant negative rank correlations were found between year 1 cotinine concentrations and reinfection by year 2 and between year 2 THCA

v concentrations and reinfection. Finally, younger and older participants with slow nicotine-metabolizing alleles had lower worm burdens compared to those with extensive metabolizing alleles.

Tobacco advertising cannot easily explain the high prevalence of smoking among Aka men, nor can socioeconomic disparities, proscriptions, or the addictiveness of nicotine easily explain the low prevalence among Aka women. Aka smoking might be better explained by internal, rather than external, cultural and political-economic factors. In addition, these results provide the first support for the hypothesis that substance use helps defend against parasites.

TABLE OF CONTENTS

Page

ACKNOWLEDGMENTS ...... iii

ABSTRACT ...... v

LIST OF TABLES ...... ix

LIST OF FIGURES ...... x

1. INTRODUCTION ...... 1

2. PUZZLE OF HIGH MALE TOBACCO USE, AND LOW FEMALE USE, AMONG CONGO

BASIN FORAGERS: IMPLICATIONS FOR ANTHROPOLOGICAL THEORIES OF DRUG USE

AND PUBLIC HEALTH ...... 7

ABSTRACT ...... 8

INTRODUCTION ...... 9

METHODS ...... 16

RESULTS ...... 18

DISCUSSION ...... 32

CONCLUSIONS ...... 40

ACKNOWLEDGEMENTS ...... 41

REFERENCES ...... 43

3. TOBACCO USE VS. HELMINTHS IN CONGO BASIN HUNTER-GATHERERS: SELF-

MEDICATION IN HUMANS? Roulette CJ, Mann H, Kemp BM, Remiker M, Roulette JW, Hewlett

BS, et al. (2014). Evolution and Human Behavior, 35(5):397-407. http://dx.doi.org/10.1016/j.evolhumbehav.2014.05.005...... 50

ABSTRACT ...... 51

INTRODUCTION ...... 52

STUDY ...... 58

MATERIALS AND METHODS ...... 63

vii RESULTS ...... 68

DISCUSSION AND LIMITATIONS ...... 74

ACKNOWLEDGEMENTS ...... 81

REFERENCES ...... 82

LIST OF TABLES ...... 96

FIGURE LEGENDS ...... 98

FIGURES ...... 103

4. HIGH PREVALENCE OF CANNABIS USE AMONG AKA FORAGERS OF THE CONGO

BASIN, AND ITS POSSIBLE RELATIONSHIP TO HELMINTHIASIS ...... 108

ABSTRACT ...... 109

INTRODUCTION ...... 110

STUDY AIMS ...... 115

MATERIALS AND METHODS ...... 117

RESULTS ...... 122

DISCUSSION ...... 128

CONCLUDING REMARKS ...... 135

ACKNOWLEDGEMENTS ...... 136

REFERENCES ...... 137

LIST OF TABLES ...... 143

FIGURE LEGENDS ...... 147

FIGURES ...... 152

5. CONCLUSION ...... 157

APPENDIX ...... 162

A. Chapter 2 Supplementary Material ...... 163

B. Chapter 3 Supplementary Material ...... 168

viii

LIST OF TABLES

Chapter 1 Tables

1. Structured Survey Questions ...... 17

2. Demographic Variables ...... 20

3. Epidemiology of tobacco based on self-report data ...... 22

4. Logistic regression models of female smoking vs. age and mother’s smoking ...... 24

5. Economics of tobacco smoking ...... 25

6. Cultural attitudes of tobacco ...... 28

7. Composite salience scores ...... 30

Chapter 2 Tables

1. Baseline sample characteristics based on interview data and samples s1-s3 ...... 96

2. ANCOVA model of follow-up cotinine concentration ...... 97

Chapter 3 Tables

1. Descriptive Statistics ...... 143

2. Logistic regression model of cannabis smoker status (THCA > 50 ng/ml) vs age and

material wealth score ...... 144

3. Negative binomial GLM models of worm burden score with log link ...... 145

4. OLS models of helminth reinfection score ...... 146

LIST OF FIGURES

Chapter 1 Figures

1. Projected numbers of tobacco-caused deaths for High-Income and Middle- plus Low-Income

countries, three scenarios, 2002-2030. Figure from Mathers and Loncar ...... 9

2. Female smoking prevalence vs. male smoking prevalence in 187 countries in 2012. Data

from Ng et al. 2014 ...... 11

3. Smoking status of Aka women vs. age and vs. the smoking status of her mother ...... 21

Chapter 2 Figures

1. Plots of male worm burden as smooth functions of age (A) and cotinine concentration (B)

controlling for linear functions of village, acculturation, and wealth ...... 103

2. A predicted effect of the intervention * baseline worm burden interaction on follow-up

cotinine concentration ...... 104

3. Scatterplot of reinfection scores vs. year 1 cotinine concentration ...... 105

4. The effect of CYP2A6 polymorphisms on worm ...... 106

5. Saliva and stool sample collection schedule and male participant attrition ...... 107

Chapter 3 Figures

1. THCA histogram for all participants and estimated density ...... 152

2. Box plot of THCA levels vs. self-reported cannabis smoker status ...... 153

3. Probability of smoking cannabis as a function of age and wealth ...... 154

4. Effect plot of worm burden vs THCA, village, and age ...... 155

5. Helminth reinfection score vs THCA concentration ...... 156

Dedication

This dissertation is dedicated to my family, and to the memory of my father, Scott Allen Roulette.

xi CHAPTER ONE

INTRODUCTION

Anthropology has a rich tradition of substance use research (e.g. Heath 1975; Furst 1976;

Marshall 1981; Page 1983; Stebbins 1990; Goodman 1993; Schultes et al., 1998; Room 2001; Singer

2004), but there are surprisingly few contributions from human behavioral ecologists and other evolutionary oriented anthropologists (but see Dudley 2000; Lende & Smith 2002). Evolutionary approaches are beginning to advance our understanding of an array of behavioral health issues, including nutrition, reproductive health, and chronic disease (Trevathan 2007). An evolutionary biocultural approach to substance use can add to this important body of research, as well as inform and challenge conventional approaches to substance use.

Conventional explanations for substance use focus on drug reward and other “external forces”, such as the campaigns of tobacco and anti-tobacco agencies and political-economic and sociocultural disparities. In contrast, this dissertation focuses on drug toxicity (see for example, Hagen et al. 2013) and highlights “internal forces”, such as personal choice, concern for one’s health and the health of a child, a need to attract and retain mates, and a (unconscious) motivation to prevent and treat helminth infections

(i.e. self-medicate).

Research was conducted among a group of Aka foragers residing in three communities along the

Lobaye River in southwestern Central African Republic (CAR). Compared to typical substance use research populations, the Aka have several unique cultural and ecological features. First, the Aka are a hunting-gathering population. Outside of the extensive ethnohistorical and archaeological literature on tobacco use in the Americas (e.g., Wilbert 1987; Winter 2000) and pituri use in Australia (e.g., Watson

1983; Ratsch et al. 2010), there are very few studies of substance use among extant hunting-gathering populations. Second, the Aka live in a remote region of the southwestern portion of the Central African

Republic and northern part of the Republic of the Congo, and have relatively little influence from the marketing efforts of transnational tobacco companies and the public health warnings of anti-tobacco campaigns. Third, they value autonomy and are egalitarian, with marked gender and age equality. And fourth, like other extant hunting-gathering populations (Hurtado et al. 2008), they are heavily infected with

1 soil-transmitted parasites (Lilly et al., 2002; Froment, 2014), which is important in light of theories linking helminth infections with substance use behavior (Rodriguez and Cavin 1982; Hagen et al. 2009 & 2013).

The frequent use of psychoactive substance by Congo Basin foragers has been noted by several ethnographers (e.g. Turnbull 1961; Bailey 1991; Hewlett 1991; Grinker 1994), but there are few systematic studies. These ethnographic observations stand in contrast to the relatively low prevalence of tobacco and cannabis use documented for other parts of Africa (for cross-national tobacco use rates see, e.g. Pampel 2008 and Ng et al., 2014. For cross-national cannabis use rates see, e.g. UN 2013). The initial aim of the study was therefore to describe tobacco and cannabis use, to validate use with biomarkers, and to determine the cultural models and economics of use in this unique population.

Results of this initial stage of the project laid the groundwork for the second phase. In particular, we found that the Aka also smoke the leaves of a tree that grows in the rain forest (Polyalthia suaveolens,

Annonaceae, also known as Greenwayodendron suaveolens Verdc., Annonaceae; Engl. & Diels) that the

Aka call motunga. This plant contains anthelmintic compounds (Nyasse et al., 2006), and is used traditionally by the Aka in a tea to treat parasitic infections. Interestingly, numerous other psychoactive drugs contain anthelmintic compounds including tobacco (nicotine), betel nut (arecoline) (Hammond et al.

1997), and cannabis (Roy and Tandon 1997), three of the world’s most widely used substances. This, along with the fact that the Aka are heavily infected with intestinal helminths, raised the intriguing possibility that the use of these substances helps defend against macroparasite infection (Hagen et al.,

2009 & 2013). The final two studies therefore investigate the link between substance use (tobacco and cannabis) and intestinal helminth infection.

All three of the studies in this dissertation are co-authored papers that have been submitted to or accepted by a peer-reviewed journal. The first study, titled “The puzzle of high male use, and low female use, among Congo basin foragers: implications for anthropological theories of drug use and public health”, by C Roulette, E Hagen, and B Hewlett, is formatted for the journal Human Nature, to which it was recently submitted. Author contributions were as follows. Study management: B.S.H. & E.H.H. Study design: C.J.R., B.S.H., E.H.H. Data and sample collection: C.J.R, E.H.H., B.S.H. Data analysis: C.J.R.,

E.H.H. Manuscript preparation: C.J.R., E.H.H. All authors discussed study execution and results and commented on the manuscript.

2 The second study was published in the journal, Evolution and Human Behavior (see, Roulette CJ,

Mann H, Kemp BM, Remiker M, Roulette JW, Hewlett BS, Kazanji M, Breurec S, Monchy D, Sullivan RJ

& Hagen EH, 2014. Tobacco use vs. helminths in Congo basin hunter-gatherers: self-medication in humans? Evolution and Human Behavior 35:397-407). Author contributions were as follows. Study management: E.H.H. Study design: C.J.R., E.H.H., R.J.S., B.S.H. M.K., D.M. Data and sample collection:

C.J.R, M.R., J.W., E.H.H., B.S.H. Cotinine assays: C.J.R., M.R., E.H.H. Helminth egg identification and counts: M.K, S.B., D.M. Genotyping: H.M., B.M.K. Data analysis: C.J.R., E.H.H. Manuscript preparation:

C.J.R., E.H.H. All authors discussed study execution and results and commented on the manuscript.

The third study was submitted to the journal, American Journal of Human Biology, and is currently under review. Included here is the pre-print version that has not been reviewed. Author contributions were as follows. Study management: E.H.H. Study design: C.J.R., E.H.H., M.K. Data and sample collection:

C.J.R. Cotinine assays: C.J.R., E.H.H. THCA assays: C.J.R. Helminth egg identification and counts: M.K,

S.B. Data analysis: C.J.R., E.H.H. Manuscript preparation: C.J.R., E.H.H. All authors discussed study execution and results and commented on the manuscript.

REFERENCES:

Bailey R. 1991. The Behavioral Ecology of Efe Pygmy Men in the Ituri Forest, Zaire. Anthropological

Papers, No. 86. Ann Arbor, Michigan: Museum of Anthropology, University of Michigan.

Dudley R. 2000. Evolutionary origins of human alcoholism in primate frugivory. The Quarterly Review of

Biology 75(1):3-15.

Froment A. 2014. Human biology and the health of African rainforest inhabitants. In: BS Hewlett, editor.

Hunter-gatherers of the Congo basin: Cultures, histories and biology of African Pygmies. New

Brunswick: Transaction Publishers. p. 117-164.

Furst PT 1976. Hallucogens And Culture. San Francisco, CA: Chandler & Sharp.

Goodman J. 1993. Tobacco in History. The Cultures of Dependence. London: Routledge.

Grinker R. R. 1994. Houses in the Rainforest. Ethnicity and Inequality among Farmers and Foragers in

Central Africa. Berkeley: University of California Press.

Hagen EH, Sullivan RJ, Schmidt R, Morris G, Kempter R, Hammerstain P. 2009. Ecology and

neurobiology of toxin avoidance and the paradox of drug reward. Neuroscience 160:69–84.

Hagen EH, Roulette CJ, Sullivan RJ 2013. Explaining human recreational use of ‘pesticides’: The

neurotoxin regulation model of substance use vs. the hijack model and implications for age and sex

differences in drug consumption. Front. Psychiatry 4:142. doi: 10.3389/fpsyt.2013.00142

Hammond JA, Fielding D, Bishop SC. 1997. Prospects for plant anthelmintics in tropical veterinary

medicine. Veterinary Research Communications 21:213–228.

Heath DB 1975. A critical review of ethnographic studies of alcohol use. Wiley.

Hewlett, B. S. 1991. Intimate Fathers: The Nature and Context of Aka Pygmy Paternal Infant Care. Ann

Arbor, MI: The University of Michigan Press.

Hurtado AM, Frey M, Hill K, Hurtado I, and Baker J. 2008. The role of helminthes in human evolution:

implications for global health in the 21st century. In: Elton S, O'Higgins P, editors. Medicine and

evolution: Current applications, 708 future prospects. CRC Press. p 151e-178e.

Lende DH, Smith EO. 2002. Evolution meets biopsychosociality: an analysis of addictive behavior.

Addiction 97(4), 447-458.

4 Lilly AA, Mehlman PT, Doran D. 2002. Intestinal Parasites in Gorillas, Chimpanzees, and Humans at

Mondika Research Site, Dzanga-Ndoki National Park, Central African Republic. International Journal

of Primatology 23:555–573.

Marshall, M. 1981. Tobacco Use in Micronesia: A Preliminary Discussion. Journal of Studies on Alcohol,

42(9), 885-893.

Ng M, Freeman MK, Fleming TD, Robinson M, et al. 2014. Smoking Prevalence and Cigarette

Consumption in 187 Countries, 1980-2012. JAMA 311(2):183-192. doi:10.1001/jama.2013.284692

Nyasse B, Ngantchou I, Nono, J-J, & Schneider B. 2006. Antifiliarial activity in vitro of polycarpol and 3-O-

acetylaleuritolic acid from Cameroonian medicinal plants against Onchocerca gutturosa. Natural

Product Research, 29(4):391-397.

Page JB 1983. The amotivational syndrome hypothesis and the Costa Rica study: Relationships between

methods and results. Journal of psychoactive drugs, 15(4), 261-267.

Pampel, F. 2008. Tobacco use in sub-Sahara Africa: Estimates from the demographic health surveys.

Social Science & Medicine, 66(8), 1772–1783, doi:10.1016/j.socscimed.2007.12.003

Ratsch A. Steadman KJ, & Bogossian F. 2010. The pituri story: a review of the historical literature

surrounding traditional Australian Aboriginal use of nicotine in Central Australia. Journal of

Ethnobiology and Ethnomedicine, 6(1), 26, doi:10.1186/1746-4269-6-26

Rodriguez E, Cavin C. 1982. The Possible Role of Amazonian Psychoactive Plants in the Chemotherapy

of Parasitic Worms—A Hypothesis. Journal of Ethnopharmacology 6:303-309.

Room R. 2001. Intoxication and bad behaviour: understanding cultural differences in the link. Social

Science & Medicine 53, 189–198

Roy B, Tandon V. 1997. In vitro fluckicidal effect of leaf extract Cannabis sativa Linn. on the trematode

Fasciolopsis buski. Indian J Exp Biol 35(1):80-82.

Schultes RE, Hofman A, Ratsch C. 1998. Plants of the Gods: Their sacred, healing and hallucinogenic

powers. Healing Arts Press, Rochester, NY.

Singer M. 2004. Tobacco Use in Medical Anthropological Perspective. In C. R. Ember & M. Ember (Eds.),

Encyclopedia of Medical Anthropology (pp. 518-528). New York: Kluwar Academic/Plenum

Publishers.

5 Stebbins KR 1990. Transnational Tobacco Companies and Health in Underdeveloped Countries:

Recommendations for Avoiding a Smoking Epidemic. Social Science and Medicine, 30, 227-235.

Trevathan WR 2007. Evolutionary Medicine. Annu. Rev. Anthropol. 2007. 36:139–54

Turnbull, CM 1961. The forest people. New York: Simon and Schuster.

UN. 2013. UN Office on Drugs and Crime. World drug report 2013. United Nations, New York.

www.unodc.org/unodc/secured/wdr/wdr2013/World_Drug_Report_2014.pdf (accessed on Dec 11,

2014).

Watson, PL 1983. This precious foliage: A study of the aboriginal psycho-active drug Pituri. Sydney:

University of Sydney.

Wilbert J. 1987. Tobacco and Shamanism in South America. New Haven, CT: Yale University Press.

Winter JC 2000. Tobacco Use By Native North Americans: Sacred Smoke and Silent Killer. Norman, OK:

University of Oklahoma Press.

6 CHAPTER TWO

THE PUZZLE OF HIGH MALE USE, AND LOW FEMALE USE, AMONG CONGO BASIN FORAGERS:

IMPLICATIONS FOR ANTHROPOLOGICAL THEORIES OF DRUG USE AND PUBLIC HEALTH

Casey J. Roulette*1, Edward Hagen1, Barry S. Hewlett1

1Department of Anthropology, Washington State University, 14204 Northeast Salmon Creek Avenue,

Vancouver, WA, 98686.

*Corresponding author: Department of Anthropology, Washington State University, 14204 Northeast

Salmon Creek Avenue, Vancouver, WA, 98686. Email: [email protected]. Telephone:

(509)592-8151. Fax: (360)546-9074

ABSTRACT

Purpose

Tobacco consumption is increasing in the developing world, and an epidemic of tobacco-related diseases is expected to follow. Current theories of tobacco use stress “external” forces, such as the addictiveness of nicotine, tobacco marketing, and socioeconomic disparities. Unfortunately, there is little research on tobacco use in the rural populations that comprise many developing nations, and almost none in small-scale societies, including hunter-gatherers. We therefore conducted one of the first systematic studies of tobacco use in a foraging society.

Methods

We used semi-structured group interviews to collect qualitative data on tobacco use among Aka hunter-gatherers of central Africa. To establish the prevalence of tobacco use in this population we conducted a structured interview among 106 Aka. A subset of the participants (n=40) were asked detailed questions about their tobacco use.

Results

Aka men pay more for tobacco, and yet have a higher smoking prevalence (95%) than men in almost any other population, whereas Aka women have a low prevalence. Aka, noted for their gender equality, thus have one of the largest known gender differences in smoking (odds ratio = 33.25, p <

0.001). Tobacco is widely shared, and appears to play a central role in this defining aspect of Aka culture.

Conclusions

Tobacco advertising cannot easily explain the high prevalence of smoking among Aka men, nor can socioeconomic disparities, proscriptions, or the addictiveness of nicotine easily explain the low prevalence among Aka women. Aka smoking might be better explained by endogenous, rather than exogenous, cultural and political-economic factors.

Key Words: smoking, substance use, addiction, hunter-gatherers, Africa

INTRODUCTION

Tobacco use is one of the world’s major health problems: it is responsible for 1 in 5 deaths in high income countries, and 1 in 10 deaths in low income countries (Ezzati and Lopez 2004). Globally, it is responsible for 16% of all male deaths and 7% of all female deaths (Eriksen et al. 2012). Whereas global tobacco use rates have declined over the last 30 years (Ng et al. 2014), the total number of smokers increased (41% for males and 7% for female) mostly due to population growth in the developing world.

Hence, most tobacco-related deaths are projected to occur in the developing world (figure 1).

Figure 1 Projected numbers of tobacco-caused deaths for High-Income and Middle- plus Low-Income countries, three scenarios, 2002-2030. Figure from Mathers CD and Loncar D (2006). Projections of global mortality and burden of disease from 2002 to 2030. PLoS Med 3(11):e442. Doi:

10.1371/journal.pmed.0030442. Reprinted under the terms of the Creative Commons Attribution License

2.5 (http://creativecommons.org/licenses/by/2.5/legalcode).

Unfortunately, little is known about the social, cultural, historical, political and economic contexts of tobacco use in the traditional rural populations that comprise many developing nations. One dramatic difference in smoking patterns in developed vs. developing countries is the sex difference. Male smoking prevalence is very similar in developed vs. developing countries (30.1% vs. 32.0 %, respectively). Female smoking, on the other hand, differs dramatically: in the developed world, 17.2% of women smoke, but in the developing world, only 3.7% smoke (Ng et al. 2014).

Despite the similarity in mean male smoking prevalence in the developing vs developed world, the prevalence of male smoking in fact varies widely across countries. Some of the highest rates of male smoking occur in Timor-Lese (61.1%), Indonesia (57%), Russia (51%), and China (45.1%). Some of the lowest male rates occur in central Africa (Nigeria: 7.5%; Ethiopia: 7.7%; Ghana: 8.2%; DRC: 15.3%; CAR:

15.9%) (Ng et al. 2014). Female prevalence also varies widely, although within each population it is almost always lower than male prevalence (figure 2), and it is especially low in Africa (Ng et al. 2014).

The low smoking rates in Africa might be bad news rather than good news because African nations appear to be entering, not leaving, a period of increasing smoking prevalence (Pampel 2008).

From 2000-2009, for instance, total cigarette consumption in Africa increased 57% (Eriksen et al. 2012).

Figure 2 Female smoking prevalence vs. male smoking prevalence in 187 countries in 2012. Solid line represents equal prevalence. Central African countries are in dark blue. Although there are more female smokers in Sweden than male smokers, more Swedish men use tobacco due to their use of snus, a smokeless tobacco product (Foulds et al. 2003). Country data from Ng et al. (2014). Aka data as reported here, and in Roulette et al. (2014).

This study had two aims. The first was to conduct one of the few quantitative investigations of tobacco use in a population of hunter-gatherers, in this case Congo basin foragers. The second was to evaluate popular theories of tobacco use in light of the patterns seen in this unique population. Before describing our methods and results, we first briefly review the existing literature on tobacco use in hunter- gatherers, and then sketch the influential political economic and cultural approaches to tobacco use in the developing world.

Tobacco use in hunter-gatherers

Most Nicotiana spp. are native to the Americas, and the first tobacco users might well have been

Paleo-Indian hunter-gatherers in North America, c. 15-20 kya (Winter 2000). Tobacco was domesticated c. 3000-2000 BP, probably in the Andean region of South America (Pearsall 1992). Archaeological evidence has confirmed use by hunter-gatherers in the Pacific Northwest c. AD 860 (Tushingham et al.

2013), and tobacco use by more recent Native American hunter-gatherers is well-documented ethnohistorically (Tushingham et al. 2013; Wilbert 1987; Winter 2000). Aside from a few arctic groups who had essentially no access to it, all Native North American societies incorporated tobacco into their religious systems (Winter 2000). However, the secular, recreational use of tobacco by Native Americans was also widespread, and almost certainly preceded the arrival of European observers (von Gernet 2000, pp. 74-78).

Some Nicotiana spp. are also native to Australia and its neighboring islands. Pituri, a substance

11 widely chewed by Australian Aborigines, comprises Duboisia hopwoodii and Nicotiana spp., both of which contain nicotine. At the time of European conquest, pituri had great political, social and economic importance, and social control was exerted over both supply and demand (Watson 1983). As a trading commodity, it was considered equivalent in value to boomerangs, spears, shields, and ochre (Ratsch et al. 2010), and it continues to be extremely important in Aboriginal life.

Outside the Americas and Australia, there is relatively little research on tobacco use among extant hunter-gatherers. Tobacco is widely used by North Asian hunter-gatherer populations (cf Munro

1963; Schrenck 1964). Among the Koryaks, a population of Siberian hunter-gatherers well known for their traditional use of the hallucinogenic fly-agaric, for example, tobacco is an important imported product

(Antropova 1964; Jochelson 1905). Cipriani and Cox (1966) note that the Onge of the Andaman Islands smoke tobacco, a recent introduction, and that they might have smoked aromatic leaves in hollowed-out crab legs for thousands of years.

Tobacco was introduced to Africa as early as the 1500s (e.g., Jeffreys 1963) and its use was widespread throughout the African continent by the nineteenth century (Laufer et al. 1930). In southern

Africa the !Kung smoked as much tobacco as possible, unless forbidden by religion, and had favorable attitudes towards smoking, which served an important social function (Damon 1973). In east Africa the

Hadza “smoke as much tobacco as they can get their hands on”, and, based on their historical use of stone pipes, have probably had access to it, or cannabis, for centuries (Marlowe 2004).

Tobacco use in Congo Basin hunter-gatherers

Tobacco use by central African foragers has been frequently noted by ethnographers (e.g., Bailey

1991; Grinker 1994; Hewlett 1991; Turnbull 1961), but there are few systematic studies. One exception is

Oishi and Hayashi (2014) who mention that tobacco use pervades the daily life of the Baka foragers in

Cameroon. The Baka say that they cannot hunt without it and will often move their forest camps closer to villages in order to maintain access to it. They also extensively share tobacco. Efe foragers of the Ituri forest have an intense craving for tobacco that produces a dependence upon Lese villagers. While Efe foragers trade labor and meat to villagers for tobacco and cannabis (cf. Terashima 1998), it often comes at the expense of acquiring other important resources. In fact, Efe smokers were significantly more likely to be poor than non-smokers (Bailey 1991). Hewlett (1977) mentions that the Mbuti, Efe, and Aka forest-

12 foragers smoke tobacco and cannabis to keep warm and to increase their energy, courage, and skill in hunting. (For more on Aka cannabis use, see Roulette et al., under review.)

Theories of tobacco use

The global popularity of tobacco is usually ascribed to the effects of nicotine on mesolimbic dopamine, a neurotransmitter that plays an important role in rewarding and/or reinforcing behaviors (the

“reward” model; Hyman et al. 2006). However, these neurophysiological effects are presumably similar in different populations, and in men and women (and indeed, in many non-human animals), and therefore cannot explain the dramatic differences in smoking across populations, and due to sex (Hagen et al.

2013). These differences are instead usually linked to socioeconomic factors.

Political economy of tobacco use

The political economy framework typically highlights two exogenous forces. First, to protect emerging markets and take advantage of rapid population growth, transnational tobacco companies aim to derail public health initiatives in developing countries and foster dependence in youth. Tobacco companies file lawsuits against low- and middle-income countries that enact tough tobacco regulations, fund sophisticated ad campaigns to defeat anti-smoking legislation, spend billions of dollars annually on advertising in developing countries, and, to attract new young smokers, use all the marketing gimmicks that have been banned in the US. Many developing countries have no limits on tar and nicotine levels in tobacco products, no required health warnings on packaging, no general public awareness of potential health risks associated with smoking, and their populations often have limited access to health care and education (Goodman 1993; Marshall 1991, 1993; Nichter and Cartwright 1991; Singer 2004; Stebbins

1990, 2001).

Second, tobacco and other “drug foods”, such as coca leaves, sugar, coffee and tea, have played a key role in facilitating trade within the expanding global economy. Agents of colonialism, such as traders, merchants, and settlers, initially lacked the means to motivate indigenous populations to perform new forms of labor. Members of the indigenous populations, perhaps including some hunter-gatherer populations, were willing to exchange their labor for novel and highly desirable drug foods, which therefore served as effective labor inducers and promoted social as well as chemical dependency

13 (Jankowiak and Bradburd 1996, 2003; Mintz 1985).

Cultural attitudes towards tobacco and smoking

The limited ethnographic research on tobacco (e.g. Black 1984; Brady and Long 2003; Haddon

1946; Hays 1991; Marshall 1981; Vallance et al. 1987; Watson 1983), highlights tobacco’s endogenous social and cultural roles and functions, as well as how cultural norms and institutions -- including ideas about gender, age, aesthetics, and health -- affect patterns of use (for review, see Singer 2004). Black

(1984), for example, investigated the socio-cultural ecology of tobacco use in Tobi, a small, isolated island in Micronesia, where tobacco is an important cultural item, is associated with competency, and is symbolic of the role of "senior person." Damon (1973) explored personal and social reasons for smoking among four Solomon Island populations and three African populations, including the !Kung. Personal gratification, rather than social factors, was the main reason for smoking among all the populations, although the !Kung and the pastoral Herero also smoked for social reasons. This contrasts with most societies in the Americas, where tobacco was incorporated into traditional myths and rituals, and was often smoked for both social and personal reasons (Wilbert 1987, Winter 2000).

Gender and tobacco smoking

Gender differences in tobacco use are usually ascribed to women’s reduced social power and economic status relative to men and/or religious proscriptions on their use of tobacco (Waldron et al.

1988; Kaplan et al. 1990; Eriksen et al. 2012; WHO 2007), all of which are particularly common in the developing world, and also apply to alcohol and other drugs (e.g. Almedom and Abraham 1994; Heath

1991; McDonald 1994; Room 1996; Waldron et al. 1988). Among traditional African populations (Maasai,

Samburu, Kisii, and Gikuyu), for example, Kaplan et al. (1990) found that gender differences in tobacco use was related to a general pattern of social restrictions on women’s behavior. Ng et al. (2007), on the other hand, found that although cultural proscriptions against women’s smoking remained strong in rural

Java, the main factor accounting for the gender difference was that tobacco was an important symbol of manhood and was essential to the formation of a male identity. Women are being increasingly targeted by tobacco advertising campaigns (Amos and Haglund 2000), however, as they represent one of the world’s largest untapped markets.

14 Study objectives

In Roulette et al. (2014) we reported a study testing the hypothesis that Aka tobacco use might, in part, be an unconscious form of self-medication against helminths. Here, we report qualitative and quantitative results of an investigation of the personal, socio-cultural and political-economic factors involved in tobacco use among Aka foragers of the Central African Republic, one of the few detailed investigations of smoking in a hunter-gatherer population outside of the Americas.

STUDY POPULATION

Aka (also called BaAka, Biaka and Bayaka) are a group of hunter-gatherers residing in the western Congo Basin, a tropical forest region encompassing southwestern Central African Republic

(CAR) and the northern part of the Republic of the Congo (ROC). Aka are generally peaceful and egalitarian with pronounced gender equality (Bahuchet 1984; Hewlett 1991). (Congo basin foragers are commonly referred to as “pygmies,” a term that some consider derogatory, and that we will avoid.)

Most of the estimated 30,000 Aka live in small camps scattered throughout the western Congo

Basin (Bahuchet 1992; Hewlett 1996). They are transitional foragers who maintain important social and economic relations with their farming neighbors, consuming agricultural products on an almost daily basis. While they spend a majority of the year in the tropical forest hunting and gathering, Aka live near farming villages for four to six months each year to trade forest products, such as meat and honey, and labor, for agricultural goods, clothes, salt, cigarettes, alcohol, axes and knives, and occasionally money.

Aka also work on villager coffee plantations in exchange for food, money, and tobacco.

Aka were introduced to tobacco during the height of colonialism, a period that brought many sweeping changes to the Aka way of life (e.g., Hewlett 1991). Currently, Aka smoke a variety of substances, including a leaf that grows on trees in the rain forest (Polyalthia suaveolens, Annonaceae, also known as Greenwayodendron suaveolens Verdc., Annonaceae; Engl. & Diels) that the Aka call motunga; cannabis; and cigarettes and locally grown tobacco. This study focuses on tobacco use

(detailed descriptions of motunga and cannabis use will be reported in a future publication). Although Aka have ready access to commercial and locally grown tobacco during parts of the year, due to their remote location they have only limited daily exposure to Western and urban cultural influences (e.g. lack of

15 television, limited access to radios), including limited exposure to tobacco advertising and anti-tobacco campaigns.

Research was conducted in the summer of 2008 among the approximately 300 Aka associated with a single trail near Bagandou, a village in southwestern CAR. Most of the nearly 800 villagers in this district are Ngandu farmers (Hewlett 1996). Participants were nearly all adult Aka residing in 12 camps located along the first two kilometers of a trail leading from the district to the forest. Because the Aka do not keep track of ages, adults were defined as married Aka with at least one child. The final sample comprised 40 male and 66 female adults (106 total).

METHODS

The study was approved by the Institutional Review Board of Washington State University.

Informed consent was obtained from all participants prior to data collection. All interviews and surveys were administered with the aid of a local research assistant who translated English into DiAka, the Aka language. Participants were paid 500 Central African Francs (CFA), the equivalent of $1 USD, for participating in the survey. Statistical analyses were conducted using Stata/IC 11.0 for Mac.

Semi-Structured Group Interviews

In order to understand cultural attitudes and beliefs of tobacco, about half a dozen open-ended, semi-structured interviews (Bernard 2006; Schensul et al. 1999) were conducted with adult Aka; these took place in several camps with small groups of men and women interviewed separately. We collected information on the types of tobacco used, proscriptions and prescriptions for its use, beliefs about the effects of tobacco use, and knowledge about its medicinal uses.

Structured Survey

A structured survey was conducted (Bernard 2006; Schensul et al. 1999) to further assess cultural attitudes and to establish the prevalence of tobacco use in the population. All 106 participants were asked if they smoked tobacco, if their mothers smoked tobacco, and if their fathers smoked tobacco.

Ages were estimated by one of the authors (BSH), based on 30 years of fieldwork at this site.

A subsample of 40 Aka, ages 25-45 (female n=20, male n=20) were asked additional questions

16 about their tobacco use (see table 1). The assessment of Aka income is rudimentary, not only because it was based on self-report, but because when Aka provide labor to the villagers they are infrequently paid with money and typically compensated with goods (e.g., manioc, salt, soap, cigarettes). However, the estimates illustrate how much money an average Aka can expect to earn on a given day if he/she were paid in cash while working for villagers (termed “daily income”).

Table 1 Structured survey questions

Questions asked of entire sample (n=106)

Do you smoke tobacco?

Does (did) your mother smoke tobacco?

Does (did) your father smoke tobacco?

Questions asked of subsample (n=40)

What type of tobacco do you prefer to smoke (e.g., commercial, local)?

At what age did you start using tobacco (e.g. mona, ngondoa/bokala)?

Who you first smoked with (e.g., parent, sibling, friend)?

Do you prefer a spouse who smokes tobacco? Why/why not?

Have you had diarrhea, vomiting and/or coughing and chest pain from smoking?

Is it OK for pregnant women to smoke? Why/why not?

Is smoking ekila? Why/why not?

Are there restrictions on who can use tobacco?

Your daily income

How much tobacco do you purchase each day?

How much tobacco do you share each day?

How do you acquire tobacco?

If you do not smoke, why not?

If you smoke, why do you smoke?

Have you learned from a doctor or missionary that smoking is bad for your health?

Have you learned from a doctor or missionary that smoking is bad for the health of the fetus?

We also asked if smoking while pregnant is ekila. Ekila is a term used by Aka and Baka to refer to practices or rules concerning hunting, eating, sex, and menstruation (Bahuchet 1984; Lewis 2008). One is free to follow these prohibitions but if they do not (except by accident) there could be terrible consequences to one’s self or even to one’s child (Motte-Florac et al. 1993). For instance, in an Aka causes-of-death study, Hewlett et al. (1986) found that ekila (i.e. eating a taboo food) was the second leading cause of death of infants.

Finally, participants were asked to free list reasons why they smoke, why they prefer or do not prefer a spouse who is a smoker, the effects of smoking on the fetus, and how they acquire their tobacco.

Composite salience scores (∑) were calculated by summing all individual salience scores (a statistic accounting for rank and frequency) for an item and then dividing by the number of informants (cf. Quinlan

2005). Items that are mentioned first by many informants have a composite salience score closer to 1 than items mentioned second or third, and by fewer informants.

RESULTS

Semi-Structured Group Interviews

Here we summarize our detailed notes from the semi-structured group interviews.

The Aka word for tobacco is ndako. There are two types of ndako used by the Aka, blancs and gbangaya. The Aka prefer and primarily smoke blancs, or manufactured cigarettes. The word “blanc”

(white) refers to the color of the cigarette packaging currently used.1 A pack of cigarettes today costs 500

CFA but most Aka purchase only one or two cigarettes at a time. At the time of the study, a single cigarette cost 25 CFA. Aka usually obtain blancs either in direct exchange for garden labor, or by working in villagers’ gardens for money, which they then use to buy them. They also trade forest products to villagers for cigarettes. While in the forest (6-8 months out of the year) Aka obtain blancs by trading forest products to villagers, loggers, and miners whom they infrequently encounter.

Gbangaya are dried tobacco leaves usually grown by villagers in their fields. The villagers sell the

1 Manufactured cigarettes in the 1970s-1980s were called bleu (blue) or jaune (yellow) because cigarettes were sold in blue and yellow boxes. Aka preferred bleu, which was stronger than jaune, whereas villagers preferred jaune.

18 leaves in small, baseball-sized balls for 50 CFA (see figure A1 in Online Resource 1). Most Aka get gbangaya by working for villagers in their gardens or by trading forest products to them. Although it is rare, a few Aka plant gbangaya in the forest. Some Aka use pipes (makundu) to smoke gbangaya, but rolling it in a dried leaf is more common. Aka refer to a rolled ndako cigarette as a mobinza (a term used to refer to being rolled in a leaf, such as gbangaya mobinza or motunga mobinza). Gbangaya mobinzas are considered more difficult to smoke than blancs.

Motunga is the substance the Aka might have been using the longest (see figure A2 in Online

Resource 1). Some Aka call it “forest tobacco” or the “ndako of the ancestors.” Aka say that when they are in the forest it is easy to get because the trees grow everywhere. Like gbangaya, Aka roll motunga in a leaf. Most Aka feel that too much motunga at one time will make one cough a lot and some Aka consider it to be stronger than gbangaya. When near the village Aka rarely smoke motunga, and do so only when there are no blancs or gbangaya around. Even then one must go into the forest to collect some. When in the forest blancs and gbangaya are more difficult to get, which makes one rely more on motunga. Aka use motunga for a variety of other utilitarian purposes: the leaves are often used, in conjunction with a variety of other leaves, to cover the base of huts (huma), and the wood is considered a good word to burn for heating (Motte 1980).

Aka believe that tobacco makes one warmer and stronger. This has several effects. First, if one is warmer and stronger then one is a better hunter and forager. The forest, which is largely covered by canopy, is often very cool, especially during morning and evening hours. Smoking tobacco while in the forest to stay warm, according to the Aka, thus helps one become a better hunter. Being stronger and warmer also allows one to work harder for villagers. The more work an Aka does on a villager’s plantation or the more forest products they collect and trade with them, the more cultivated foods, Western goods, and/or money they get. Finally, being stronger makes a person a better dancer/singer. The forest spirit, dzengi, likes it when people sing and dance, so smoking is something that dzengi enjoys. Smoking is particularly prevalent at dances (such as funeral dances), where dzengi often appears. In general, Aka attitudes towards tobacco emphasize tobacco’s motivational and social effects. Aka are using it to help them be better hunters and foragers, to work harder, and to dance and sing better. Nevertheless, Aka do not want their children to use ndako and believe that if they do they will get pains or get sick, and/or they

19 will grow up to not respect their parents.

(For information on the rare use of snuff, and on leaves used for rolling cigarettes, see Online

Resource 2. For Aka cannabis use, see Roulette et al. under review.)

Structured Surveys

Sample demographics are summarized in table 2.

Table 2 Demographic variables

N Age mean Age median Age range

Total Sample

Both genders 106 33.8 32.0 18-70

Male 40 34.4 32.5 18-70

Female 66 33.5 32.0 18-70

Subsample

Both genders 40 32.6 30.0 25-45

Male 20 32.3 30.0 25-45

Female 20 32.9 29.5 25-43

Patterns of tobacco use

Sample sizes for some questions vary slightly because not all participants responded to every question. Ninety-five percent of Aka men were smokers compared to only 36% of Aka women (table 3).

The gender difference in smoking was large and significant, with an odds ratio of 33.25 (se = 25.58, z =

4.55, p < 0.001). Some women (10%) reported trying tobacco but not developing a regular smoking habit. In contrast, all men were habitual smokers at some point.

Female smoking increased with age. Whereas only 27.8% (15/54) of women between 18 and 44 smoked ndako, 75% (9/12) of women 45 and older smoked ndako, a statistically significant difference (z =

-2.33, p = 0.02). A logistic regression model of female smoker status vs. age found that each one-year increase in age increased the odds of being a smoker by 1.06 (table 4, model 1). A logistic regression

20 model of female smoker status vs. both age and maternal smoker status found that women whose mothers smoked were 12 times more likely to be smokers than women whose mothers did not smoke, after controlling for age (table 4, model 2; figure 3). Nearly all men smoked, so determining predictors of male smoking was not possible.

Figure 3 Smoking status of Aka women vs. age and vs. the smoking status of her mother. A: Mother does not smoke (n=34). B: Mother smokes (n=32). Histograms represent the age distribution of nonsmokers (bottom of each plot) and smokers (top of each plot). Red lines are the probabilities of smoking computed from a logistic regression (table 4, model 2).

Self-reported indices of daily cigarette consumption were somewhat inconsistent (table 3).

Women and men smokers reported having smoked 0.8 and 1.8 cigarettes, respectively, on the day of the study; 2.4 cigarettes each on the day prior to the study; and 6.2 and 9.3 cigarettes, respectively, on a typical day. Cotinine values reported in Roulette et al. (2014) are more consistent with the “typical” daily values, at least for men; the amount of money men report spending on tobacco would purchase about 5 cigarettes (see below).

Nearly all of the participants reported first smoking tobacco as an adolescent (bokala/ngondoa), although four men reported starting to smoke in middle childhood (or mona) (table 3). Most participants started smoking with their fathers, e.g. the child lit the cigarette for their father or their father passed them a lit cigarette (table 4).

Table 3 Epidemiology of tobacco based on self-report data

Female Male Total

Total sample

Smoking prevalence dataa

Smoker 24 (36%)* 38 (95%)* 63 (59%)

Mean age of smokers 40.6 33.4 36.2

Former smoker 1 (2%) 2 (5%) 3 (3%)

Mean age of former smokers 40.0 54.0 49.3

Parental smoking datab

Mother Smokes 32 (48%) 11 (28%) 43 (41%)

22 Father Smokes 59 (89%) 38 (97%) 97 (92%)

Mother quit 3 (5%) 2 (5%) 5 (5%)

Father quit 6 (9%) 1 (3%) 7 (7%)

Subsample

Smoke motungac 5 (25%) 13 (65%) 17 (43%)

Plant gbangayac 3 (15%) 4 (20%) 7 (18%)

Number cigs/dayd

Mean blancs today 0.6 1.3 1.2

Mean gbangaya today 0.2 0.5 0.4

Mean blancs/gbangaya yesterday 2.4 2.4 2.4

Mean blancs/gbangaya typical day 6.2 9.3 8.6

Age start smokinge

Mona 0 (0%) 4 (21%) 4 (16%)

Bokala/ngondoa 6 (100%) 15 (79%) 21 (84%)

First smoked withf

Father 3 (75%) 11 (64%) 14 (66%)

Older Sibling/Friend 0 4 (23%) 4 (19%)

23 Spouse 1 (25%) 0 (0%) 1 (4%)

Other 0 2 (11%) 2 (9%) aN=106 for all variables bN=105 for all variables. cN=40, 20 male and 20 female dSmokers only, N=24 eSmokers only, N=25 fSmokers only, N=21

*Significant gender difference at p=0.00

Table 4 Logistic regression models of female smoking vs. age and mother’s smoking

ß se z p

Model I

Age 0.06 0.02 2.90 0.004

Model II

Age 0.07 0.03 2.90 0.004

Mother’s smoking 2.50 0.72 3.47 0.001

The estimated average daily income across both genders was 173.57 CFA (approximately $0.35

US) (table 5). (Traditional healers -- ngangas -- earn significantly more money and were therefore excluded from this analysis.) There was a significant gender difference in daily income, with males earning more (M = 242.19, sd = 167.26) than females (M = 115.79, sd=62.48); t(18.9) = -2.86, p = 0.01.

There was no significant difference in the daily income of female non-smokers (M = 121.4, SD=72.6) vs.

24 smokers (M =100.0, SD=0.0); t(13) = 1.1, p = 0.29.

Based on the mean daily income (173.57 CFA) and the price of one pack of the cheapest locally available manufactured cigarettes (500 CFA) over 280% of daily income is needed to purchase one pack of cigarettes.

Aka estimated spending almost half of their daily income on tobacco (73.57 CFA, or $0.15 US)

(table 5). Men (M = 0.64, SD = 0.34) spend a marginally significant greater proportion of their daily income on tobacco than do women (M = 0.36, SD = 0.59); t = -1.72, p = 0.095).

Of the purchased tobacco, Aka estimated that they share almost half of it with others (table 5). In terms of daily income, Aka give away over a quarter of their earnings to other Aka in the form of tobacco.

In terms of CFA, men share significantly more tobacco (M = 51.41, SD = 25.87) than women (M = 19.08,

SD = 33.43); t = -3.15, p = 0.0034). However, women give away a marginally significant greater proportion of their tobacco (M = 0.62, SD = 0.30) than men (M = 0.43, SD = 0.13); t = 1.98, p = 0.0628).

Women distribute most of the tobacco they purchase to husbands and male relatives.

Table 5 Economics of tobacco smokinga, b

Female Male Total

Daily income

Mean 115.8* 242.2* 173.6

Tobacco purchasing

Mean spent on tobacco 30.3 125.0 73.6

Mean spent on blancs 14.5 81.3 45.0

Mean spent on gbangaya 15.8 43.8 28.6

25 Percent spent on tobacco 36%** 64%** 49%

Tobacco sharing

Mean tobacco shared 19.1* 51.4* 33.9

Percent purchased tobacco shared 62%** 44%** 49%

Percent income shared as tobacco 24% 29% 26% aAll estimates are in Central African Francs, CFA. b N = 35; three highest wage earners, all male ngangas, were excluded.

* Significant gender difference at p<0.01

** Significant gender difference at p<0.1

Cultural attitudes towards tobacco and smoking

Most Aka obtain their tobacco by working for villagers (either to obtain tobacco directly or to obtain money to purchase tobacco) (table 6). Some Aka sell forest products (e.g. firewood or dry leaves) to villagers and then purchase tobacco, or they ask their spouse for tobacco. No one mentioned cultivation as the primary way of obtaining tobacco, although 18% of Aka reported growing tobacco

(gbangaya) (table 3).

The most preferred item to smoke was blancs (table 7). Aka were asked to list reasons why they smoke. The most salient response for why they smoke was their tobacco use satisfies a “hunger” or

“desire” (ndjala), although some mentioned that they use it because it is their usual thing or that it gives them strength or helps them keep warm (table 7).

All males (100%, n=19) preferred a woman who does not use tobacco whereas most females

(79%, n=15) preferred a mate who does (table 6). The proportion of women who preferred a smoking spouse (79%) was significantly greater than the proportion of men who preferred a smoking spouse (0%), z = 4.98, p = 0.0000.

Aka men did not like women who smoke for several reasons, but the most salient response was

26 that the "spouse might become lazy, or may not listen or obey" (table 7). Other reasons were it is not for women/it is for men, spouse speaks too much/quickly, spouse gets promiscuous, spouse gets dizzy/jumpy, or that spouse steals money for tobacco.

Most females preferred a mate who smokes ndako because it gives them “strength” for subsistence hunting and foraging or for working for the villagers. Some Aka preferred a smoking spouse because if both smoke it can be a good thing, or together they will have more tobacco. But it could also be a bad thing that can cause a “struggle” between husband and wife (table 7). Although most women preferred a man who smokes, reasons why a few women preferred a non-smoker mate were that they did not like ndako because it smelled bad and that ndako made men get angry/crazy and get into fights.

Despite the fact that most males prefered a non-smoking spouse, the most salient reasons given by women for not smoking had to do with personal preference (i.e. do not like tobacco, it made them sick). However, proscriptive reasons (i.e. it is not for women/it is for men) were still very salient and accounted for the remainder of the responses (table 7).

Negative health consequences of smoking

Aka associated tobacco smoking with negative health effects (table 6). Over 80% of participants mentioned coughing and/or chest pain from smoking blancs, gbangaya and/or motunga. Treatments for coughing mentioned by the Aka include bloodletting the chest, putting mbaso flowers in water and drinking it; making a mongi leaf tea; making a moondango bark tea; and making a mokata (Garcinia

Punctata, Clusiacaees) and motunga leaf tea. Headaches were also frequently mentioned. Treatments for headaches include bloodletting the head, using an ekama leaf snuff, moondango bark tea, or using enzembe bark mixed with water and snorted through a leaf (possibly also for coughing) (species names of plants listed unknown).

Most Aka agree that it is less common to smoke tobacco, cannabis, or motunga when one is sick.

The exception is snuff, which is often used to treat headaches. In general, Aka prefer to smoke these substances when they are healthy enough to work and hunt.

Over 90% of Aka participants mentioned that it was bad to smoke while pregnant. The most salient health risks mentioned were that it causes a baby to cough either inside the womb or once it is born. Other risks included unspecified sickness, baby cannot breath, baby dizzy/eyes dizzy, baby gets

27 diarrhea, baby turns very black, baby will not listen when older, smoke into eyes of the baby, women get tired/slow, stomach turns, and baby dies. Of the few who thought it was okay to smoke while pregnant, one person stated that it was okay for women to smoke in later trimesters but not early on. Another informant stated that it was not okay to smoke during later trimesters because ndako causes warmth and warmth makes birth difficult. Others thought it depended on the women and that the baby was fine if the mother did choose to smoke (table 7). About a third of Aka reported hearing from a doctor or missionary that smoking is bad for the fetus, and about half had heard it was bad for their own health (table 6).

Although it was generally agreed that maternal smoking while pregnant (MSP) was harmful for the baby, there was no cultural taboo against MSP per se. When asked if MSP or breastfeeding is ekila

(taboo), only 22% of participants mentioned it was (table 6). However, a marginally significant greater proportion of women (33%) mentioned that smoking is ekila than did men (11%); z = 1.77, p = 0.09.

Besides smoking tobacco, motunga, and cannabis, Aka also use them medicinally (as described in Online Resource 3).

Table 6 Cultural attitudes of tobacco

Female Male Total

n (%)

Prefer spouse who smokesa 15 (79%)* 0 (0%)* 15 (39%)

Perceived health effectsb

Cough/Chest pain from

Blancs 6 (100%) 16 (84%) 22 (88%)

Gbangaya 6 (100%) 16 (84%) 22 (88%)

Motunga 6 (100%) 15 (79%) 21 (84%)

28 Vomiting from

Blancs 2 (33%) 8 (42%) 10 (40%)

Gbangaya 2 (33%) 7 (37%) 9 (36%)

Motunga 1 (17%) 7 (37%) 8 (32%)

Diarrhea from

Blancs 1 (17%) 3 (16%) 4 (16%)

Gbangaya 0 (0%) 4 (21%) 4 (16%)

Motunga 1 (17%) 2 (11%) 3 (12%)

Maternal smoking while pregnantc

is bad for baby 16 (89%) 18 (95%) 3 (92%)

is ekila 6 (33%)** 2 (11%)** 8 (22%)

Learn from doctor and/or missionary that smokinga

is bad for your own health 7 (37%) 11 (58%) 18 (47%)

is bad for health of fetus 4 (21%) 7 (37%) 11 (29%) aN=38 bSmokers only, N=25 cN=37

*Significant gender difference at p=0.00

29 **Significant gender difference at p<0.1

Table 7 Composite salience scores

Female Male Total

∑ (# of ∑ (# of ∑ (# of

mentions) mentions) mentions)

Most preferred plant to smoke (N=24)

Blanc 0.93 (5) 0.93 (18) 0.93 (23)

Gbangaya 0.20 (2) 0.17 (4) 0.17 (6)

Cannabis 0.27 (2) 0.10 (3) 0.13 (5)

Motunga 0.13 (1) 0.02 (1) 0.04 (2)

Blanc w/ Cannabis 0.00 (0) 0.05 (1) 0.04 (1)

Least preferred plant to smoke (N=24)

Gbangaya 0.75 (3) 0.47 (8) 0.52 (11)

Cannabis 0.25 (1) 0.47 (8) 0.43 (9)

Motunga 0.25 (1) 0.21 (4) 0.21 (5)

How acquire tobacco (N=23)

Works for villagers 0.88 (4) 0.95 (18) 0.94 (22)

Sells forest products 0.00 (0) 0.08 (2) 0.07 (2)

Ask spouse 0.25 (1) 0.00 (0) 0.04 (1)

Why not smoke (N=11)

Personal reasons, combined 0.55 (6) 1.00 (1) 0.59 (7)

Do not like it 0.40 (4) 0.00 (0) 0.36 (4)

Made them sick 0.15 (2) 1.00 (1) 0.23 (3)

Proscriptive reasons

For men, not for women 0.50 (5) 0.00 (0) 0.46 (5)

Why started smoking (N=21)

Father influence 0.75 (3) 0.65 (11) 0.67 (14)

30 Saw friend/sibling 0.00 (0) 0.24 (4) 0.19 (4)

Spouse smoked 0.25 (1) 0.00 (0) 0.05 (1)

Saw other smoke 0.00 (0) 0.06 (1) 0.05 (1)

Desired to 0.00 (0) 0.06 (1) 0.05 (1)

Reason smokes now (N=15)

Desire/hunger (ndjala) 0.88 (4) 0.64 (7) 0.70 (11)

Usual thing 0.25 (1) 0.18 (2) 0.20 (3)

Warmth/strength/pleasure 0.13 (1) 0.18 (2) 0.17 (3)

Reason prefers nonsmoker spouse (N=23)

Spouse is lazy; doesn’t obey/listen 0.13 (1) 0.53 (10) 0.48 (11)

Spouse get angry/crazy/fight 0.42 (2) 0.10 (3) 0.15 (5)

Not normal gender role 0.00 (0) 0.16 (3) 0.13 (3)

Spouse speak too much/quickly 0.25 (1) 0.09 (2) 0.12 (3)

Does not like smoke 0.50 (2) 0.00 (0) 0.09 (2)

Spouse get promiscuous 0.00 (0) 0.11 (2) 0.09 (2)

Spouse gets dizzy/jumpy 0.08 (1) 0.08 (2) 0.08 (3)

Spouse steals tobacco money 0.13 (1) 0.05 (1) 0.07 (2)

Reason prefers smoker spouse (N=15)

Spouse get strength/work harder 0.80 (12) 0.00 (0) 0.80 (12)

Compatible if both smoke 0.13 (2) 0.00 (0) 0.13 (2)

More tobacco if both smoke 0.07 (1) 0.00 (0) 0.07 (1)

Effects of maternal smoking while pregnant

(N=33)

Baby coughs 0.40 (6) 0.33 (6) 0.36 (12)

Baby sick (unspecified) 0.13 (2) 0.33 (6) 0.24 (8)

Baby can not breath 0.20 (3) 0.00 (0) 0.09 (3)

Do not know why, but it is 0.07 (1) 0.06 (1) 0.06 (2)

31 Baby dizzy, eyes dizzy 0.00 (0) 0.11 (2) 0.06 (2)

Baby gets diarrhea 0.07 (1) 0.03 (1) 0.05 (2)

Turns baby very black 0.07 (1) 0.00 (0) 0.03 (1)

Baby will not listen when older 0.00 (0) 0.06 (1) 0.03 (1)

Smoke into eyes of baby 0.07 (1) 0.00 (0) 0.03 (1)

Women get tired/slow 0.00 (0) 0.06 (1) 0.03 (1)

Stomach turns 0.00 (0) 0.06 (1) 0.03 (1)

Baby dies 0.03 (1) 0.00 (0) 0.02 (1)

Reason why maternal smoking while pregnant is ekila (N=8)

Baby coughs when born 0.33 (1) 0.33 (2) 0.33 (3)

Baby coughs inside womb 0.00 (0) 0.33 (2) 0.25 (2)

Bad for breastfeeding/baby coughs 0.00 (0) 0.25 (2) 0.19 (2)

Baby sick (unspecified) 0.00 (0) 0.17 (2) 0.13 (1)

Baby still inside womb 0.50 (1) 0.00 (0) 0.13 (1)

Pregnancy eats child (Ekila of yama) 0.50 (1) 0.00 (0) 0.13 (1)

Diarrhea 0.17 (1) 0.00 (0) 0.04 (1)

DISCUSSION

Compared to global patterns, Aka tobacco use stands out in four important ways. First, tobacco is extraordinarily expensive for Aka. Second, despite the high cost, the prevalence of male tobacco use is extraordinarily high, leading to, third, a very large gender difference in tobacco smoking. Fourth, most purchased tobacco is shared with others.

Tobacco expense

The estimated daily income for Aka was about 175 CFA (approximately $0.35 US) (table 5). This is much lower than in most of the developing world, where the average income is about $34 USD per day, and low compared to sub-Saharan Africa where it is about $2 USD per day (The World Bank 2012). Our

32 assessment of Aka income is rudimentary, however, because it is based on self-report, and because when Aka provide labor to the villagers they are often compensated with goods, such as manioc, salt, soap, and cigarettes.

At our field site, one pack of commercial cigarettes costs 500 CFA (about $1 USD). Locally grown tobacco (gbangaya) is somewhat cheaper. A standard method to compare the real cost of tobacco across countries is to compare the percentage of annual income required to purchase 100 packs of the cheapest local cigarettes. If the Aka worked 250 days out of the year, they would earn about 43,750 CFA, compared to the 50,000 CFA required to buy 100 packs of cigarettes. Thus, Aka would require about

114% of their annual income to buy 100 packs of cigarettes, compared to the global median and mean of

3.4% and 7.6%, respectively, and to 14.5% in the Central African Republic (Eriksen et al. 2012).

On a daily basis, Aka estimated spending 73.6 CFA ($0.15 US) on tobacco, or almost half of their daily income (table 5). Again, this is a limited estimate of the cost of tobacco to Aka because they can also steal it from villager gardens, and trade meat, other forest products, and labor for it at an unknown rate of exchange. Nevertheless, our data indicate that, compared to its cost in other populations in the developing world, tobacco is very expensive for the Aka. Similar results are reported for the Efe, a foraging group in the eastern Congo, who expend considerable time and effort to obtain tobacco and cannabis at the expense of obtaining food and material items (Bailey 1991). Twenty nine percent of Efe exchanges with Lese villagers involved receipt of tobacco and cannabis, second only to exchanges for food (Terashima 1998).

High male prevalence

Ninety five percent of Aka men smoke. In contrast, the prevalence of male smoking in the Congo basin is 13-16%, 15.9% in CAR (Ng et al. 2014), and in sub-Saharan Africa ranges from 10% to 27%

(Pampel 2008). The highest national male smoking prevalences range from 60-70% in e.g., Russia,

Indonesia, and some Pacific islands. The high rate of smoking among Aka men is only partially matched by the 36% prevalence rate among Aka women, which is much higher than elsewhere in sub-Saharan

Africa (<10%, Ng et al. 2014), but not so different from the female rates seen in, e.g., Europe. In Roulette et al. (2014) we report assays of Aka salivary cotinine, a nicotine biomarker. There is a very close correspondence between the cotinine-based male prevalence in Roulette et al. (94%) and that reported

33 here, but a marked discrepancy between the cotinine-based female prevalence (5%) and the self- reported rate we report here. Aka women who self-report as smokers are smoking much less than Aka men who self-report as smokers.

Although the prevalence of male smoking is high, daily cigarette consumption is low. Aka men report smoking 9.3 cigarettes per day and women smokers 6.2 per day. In contrast, globally, smokers smoke an average of 18.8 cigarettes a day, and 16.6 per day in the developing world (Ng et al. 2014).

Low cigarette consumption is probably due to their high cost. (However, Roulette et al. 2014 found that

Aka men’s mean cotinine levels were only slightly lower than those seen in Western populations, suggesting that Aka tobacco consumption might be higher than reported here.)

In the Western biomedical model, tobacco use is explained by nicotine’s effects on brain dopamine levels, but this dopaminergic effect presumably does not vary much across human populations, whereas the prevalence of tobacco use varies widely. Thus, to explain the exceptionally high rate of Aka men’s tobacco use, we must examine population-specific factors. Theories from political economy that link tobacco use to tobacco marketing also do not easily explain the high male rate in the Aka, a population with limited daily exposure to tobacco advertising. Only 10% of Aka men in this region possessed a radio (Roulette et al. 2014). Of course, men without radios listen to the radios of others, but radio shows are typically broadcast in either French or Sango (the national languages), Aka men do not speak French and only a few speak Sango. Aka men also have limited exposure to health warnings about tobacco, however, which could help explain the high rate. Yet the Aka indigenous cultural models recognize many health hazards from smoking, and about 1/3 reported hearing from a doctor or missionary that smoking was bad for their health (table 6).

A second theory from political economy argues that agents of colonialism use tobacco to induce labor from indigenous peoples. This might partially account for the high prevalence of male Aka tobacco use. The Aka have a subsistence system common to several forest foraging populations around the world where forager’s forest products and garden labor are traded for farmer’s agricultural foods. Like colonial powers, Ngandu villagers might be using tobacco to induce Aka labor. Villagers often trade tobacco and other psychoactive substances to Aka in exchange for garden labor, a pattern also documented in similar groups in the Republic of Congo (Hanawa 2004) and among the Efe (Bailey 1991). Moreover, when Aka

34 are paid in cash, they use about half of it to buy tobacco (table 5). Because the labor comes at the expense of missing the best net-hunting season of the year (Hewlett 1991) it undermines the Aka’s autonomy, which depends on their ability to hunt. Thus, tobacco-for-labor exchanges are one possible avenue through which the Aka were, and still are, being drawn into the expanding global economy.

The labor inducement perspective also recognizes that tobacco and other drugs could serve as labor enhancers. That is, the neurophysiological effects of tobacco might, e.g., reduce fatigue, increase attention, and so forth. On this view, colonial powers, and Ngandu farmers, would encourage tobacco use to increase worker productivity.

Missing from the labor inducement theory is an account of why tobacco is appealing to the Aka in the first place, and how it accounts for the gender difference in smoking. After all, women do more garden labor than men (Hewlett 1991), yet use far less tobacco than men. And labor enhancement serves

Ngandu interests, not Aka interests.

Labor enhancement might partially explain Aka men’s attraction to tobacco, albeit not quite in the way envisioned in earlier work. Because, according to the Aka, smoking increases strength, warmth and courage, which increase hunting and foraging success, and because Aka women are attracted to men who are capable of providing food, most Aka believe that, to get a wife, it is important for young men to smoke. Though Aka are net hunters, and men and women hunt together, Aka men also climb trees to collect honey. Climbing trees for honey was described by one Aka camp as the “first reason for a marriage”—the more an Aka man climbs trees and provides honey the more attractive he becomes to a woman and her family. Aka men also sometimes hunt alone with a spear or, more frequently today, a shotgun. In support of this idea, men start to smoke about the time they are interested in women or when they are becoming “more like a man.” More importantly, most women prefer a husband who smokes

(79%, table 6) because it gives them strength for subsistence hunting and foraging or for working for villagers.

Similarly, Aka men smoke heavily at dances because smoking helps one become strong and dance better, which appeases the forest spirit, dzengi. Good dancers, in turn, are attractive to women.

One Aka informant said that when he goes to dances and smokes it is attractive to women and they approach him. Some men mentioned that smoking is attractive and makes women not able to control

35 themselves. Tobacco is expensive, so smoking might be a status symbol that indicates a man had money to buy tobacco, and thus presumably money for clothes, food, and other items for the family. Though dances are open to all Aka of all ages many Aka recognize that young men attend dances more frequently than older men, apparently with the intention of finding a wife.

Another possible explanation, which we develop and test elsewhere (Hagen et al. 2009, Hagen et al. 2013, Roulette et al. 2014) is that substance use is linked to parasite infections. In brief, nicotine and other plant drugs are harmful to human parasites, such as intestinal worms. Populations with high parasite prevalence, and little access to Western medicine, might consume toxic plants as an unconscious form of self-medication. In our population, virtually all Aka men are infected with intestinal worms (99%; Roulette et al. 2014), which might contribute to the high rate of tobacco use as an unconscious form of anti-worm treatment.

Despite these theoretical accounts of tobacco use, most Aka smoke for personal gratification: ndjala, or a “desire” or “hunger”, is the most salient response given for why smokers use tobacco (table

7).

Low female prevalence

Globally, fewer women than men smoke tobacco, although the magnitude of the difference varies by level of economic development: about 30% of men smoke in both developed vs. developing countries, whereas 17.2% of women smoke in developed countries but only 3.7% smoke in developing countries

(Ng et al. 2014). In sub-Saharan Africa, women’s smoking rates range from 0.6% in Cameroon to 9.9% in

Namibia, with a median of 4.5%. In Congo basin countries, age-adjusted female smoking rates are <

2.6%, and 1.5% in CAR (Ng et al. 2014).

The low prevalence of smoking among Aka women, compared to Aka men, conforms to this global pattern. Less clear is whether Aka women’s smoking rate is more similar to women in developed countries, as it seems to be based on self-reported smoking (36%), or more similar to women in developing countries and sub-Saharan Africa, as indicated by the cotinine biomarker data (5%; Roulette et al. 2014).

One theory of low female use, especially pertinent to the developing world, argues that social and/or economic proscriptions on women’s behavior impose limits on how much women can smoke.

36 Indeed, throughout most of Africa long established cultural traditions forbid women from smoking. In West

Africa, for instance, Islamic religious beliefs prevent women’s smoking (Laufer et al. 1930). In East Africa among Kenyans, as well as pastoralist Maasai and Samburu, decreased social power for women is related to fewer women smoking (Kaplan et al. 1990). In a cross-cultural study of smoking, Waldron et al.

(1988) found that in many parts of Africa, Asia, the Pacific and Latin America Western influences, such as the Church, have helped foster the idea that smoking is morally wrong, and hence smoking has been restricted to men.

There is some evidence in support of these ideas among the Aka. There are clear economic disparities in daily income, for instance, with men earning twice as much as women earn per day of work

(table 5). It is not clear, however, that a wage difference, by itself, explains the gender difference in tobacco use. There was no significant difference in income among women smokers vs nonsmokers, for example, and even though Aka men earn more, they also spend a greater fraction of their income on tobacco than women, which suggests that they value tobacco more than women do.

There is also some evidence that women might be socially proscribed from using tobacco. Like some other activities among the Aka, tobacco use is gendered, which is reflected in statements like,

“tobacco is for men, not for women”, a view commonly expressed by our informants. Social proscriptions are unlikely to account for the low female smoking rates among Aka women, however. The Aka are one of the most egalitarian societies known to anthropology, with essentially no ascribed gender hierarchy.

Although there is some sexual division of labor, Aka men and women often forage together and are not ridiculed for participating in activities usually designated for the opposite sex. Men carry baskets and digging sticks and women carry the men’s nets, spears and crossbows. While males usually retain all the named positions, such as kombeti, tuma (great elephant hunter) and nganga (healer), their power and authority is subdued and women tend to have pronounced political power and prestige (Hewlett 1991).

Contrary to theories emphasizing economic or social disparities, most Aka women provided personal reasons for why they do not smoke, such as they did not like it or it made them sick (table 7), indicating that Aka women are indeed free to try tobacco even in the presence of a gender norm. Many women have tried tobacco, in fact, but chose not to continue smoking (e.g., because it made them sick), and over a third of our female sample claimed they were smokers, despite the fact that they probably

37 didn’t smoke that often (judging from the cotinine biomarker data).

One reason why women might choose to not smoke is because they are trying to attract or retain a spouse. Virtually all Aka men preferred a spouse who does not use tobacco (table 6). Therefore Aka women, especially young women, might be avoiding tobacco because men find female smoking to be unattractive.

A complimentary reason is that both sexes view smoking as harmful to the fetus (table 6). The

Aka are a natural fertility population, i.e., do not use condoms or oral contraceptives. Their total fertility rate is 5-6 (Hewlett 1992; Fouts and Brookshire 2009), which means that from their late teens to their late

30’s, most Aka women are pregnant or lactating. As a consequence, women in this age who smoked would be exposing their fetuses or infants to tobacco constituents. Although the consequences of tobacco use mentioned in the scientific literature differ from those mentioned by the Aka, there is abundant evidence from the former that maternal smoking is harmful to fetuses, nursing infants, and children exposed to environmental tobacco smoke (Eriksen et al. 2012). It is possible that toxin avoidance mechanisms deter pregnant and lactating women from using tobacco and other drugs (Hagen et al.

2013).

It is interesting to note that while the Aka were uniformly in agreement that maternal smoking is bad for the developing fetus, there was a lack of cultural consensus when it came to whether or not maternal smoking is ekila. In our study, less than a quarter of the respondents mentioned that maternal smoking is ekila. However, significantly more women (33%) than men (11%) believed it is ekila.

Under both the mate attraction and fetal protection hypotheses, reproductive-aged women should avoid smoking whereas postmenopausal women would be free to smoke if they chose. In support of this prediction, the likelihood of being a female smoker increased as age increased, albeit with an interesting interaction: the only reproductive-aged women who smoked had mothers who smoked (figure 3). More strikingly, Roulette et al. (2014) found that none of 33 reproductive-aged women had levels of cotinine consistent with recent smoking. Perhaps reproductive aged women who claim to be smokers do smoke the occasional cigarette, which might not show up in the cotinine data. We can only speculate about why such women all had mothers who smoked. Perhaps tobacco was more available to such women, they were more comfortable smoking, or they were more willing to admit smoking.

38 After menopause, many Aka women, having divorced a husband or been widowed, often move back in with their brothers or sons, which might provide additional liberties (Hewlett 1991). Older women, for example, tend to have fewer restrictions or prohibitions. Women also claim that after menopause they become “more like a man” (Bonnie Hewlett, personal communication). Finally, the female age trend could be a cohort effect. A few Aka said that in the past both male and female Aka smoked. Tobacco might have become increasingly gendered as Aka became increasingly acculturated and settled.

Tobacco sharing and Aka society

Aka give away nearly half of all the tobacco they purchase. As a fraction of their daily income,

Aka thus give away one-quarter of their income to others in the form of tobacco (table 5). Tobacco sharing occurs in all economic and social settings, including hunting, leisure time with others, and dances, where Aka often light a cigarette and pass it around with friends and family members. Women, interestingly, gave away a greater proportion of their purchased tobacco (62%) than men (44%).

Virtually all ethnographers of foraging societies have emphasized the importance of sharing and generalized reciprocity (e.g., Sahlins 1972; Woodburn 1982; Bird-David 1990). Theoretical models of foraging societies usually emphasize the sharing of meat and other valued foods (e.g., Gurven 2004).

Aka, too, almost always share meat (e.g., Kitanishi 2000). Given the number of hours Aka must work to obtain tobacco, and the large fraction they then share with others, tobacco might play an unexpectedly large role in the development and maintenance of Aka social relationships and cooperation.

Indeed, the social importance of tobacco and of tobacco sharing among small-scale societies is noted by several observers (e.g., Black 1984, Brady and Long 2003, Haddon 1946; Hays 1991; Kaplan et al. 1990; Marshall 1981; Vallance et al. 1987; Waldron et al. 1988). In one of the only other studies to systematically investigate drug use in an African hunting-gathering community, the !Kung, Damon (1973) found that although personal gratification had a strong effect on why the !Kung smoked, tobacco also played a prominent social role in that the !Kung prefer to smoke with other people and often pass around a pipe during discussions.

The importance of sharing could also help explain the high prevalence of Aka male smoking.

Tobacco, though expensive, is lightweight and easy to transport, does not spoil, and can therefore be enjoyed by Aka in the village and in the forest. It is an ideal “good” upon which to ground a sharing-based

39 society. Because a reputation for sharing is so critical to the Aka, and because tobacco is so widely shared, it might simply be impossible for a man to not smoke. Women, too, could participate in tobacco sharing. Women obtain tobacco by working in villager gardens, and then share it with their husbands, male relatives, older female relatives, and friends.

We conjecture that valuable “recreational” drugs have played an important but largely unrecognized role in the sharing-based social organization seen in many foraging societies, such as central African foragers, San, Hadza, and the many Australian foraging groups. If so, tobacco would be a relatively recent such drug, at least in Africa, where it arrived from the New World in the fifteenth century.

Cannabis, however, probably arrived in Africa by the twelfth or thirteenth century, and perhaps as early as the first millenium (van der Merwe 2005), and is widely smoked by central African foragers today (e.g.,

Roulette et al. under review). Motunga, a native plant, has been smoked for at least the last three or four generations, according to our interviews. Baka foragers in Cameroon also smoke motunga (which they term botunga). Oishi and Hayashi (2014) speculate that motunga might have been used since before the

Aka-Baka split, about 250-300 years ago. It is not out of the question that central African foragers have smoked motunga, or other native psychoactive plants, for many centuries or more (it is also used by Ituri foragers for, e.g., firewood, houses and arrow poison, but recreational or medicinal uses do not seem to be recorded; Terashima and Ichikawa 2003). Although motunga is widely available in the forest, our Aka informants suggest that it is less convenient to smoke than tobacco because the leaves must first be gathered and then dried in a fire. Hence, it is only smoked when tobacco is not available. In some sense, motunga might be more “expensive” than tobacco.

CONCLUSIONS

Aka men pay more for tobacco, and yet are more likely to be smokers, than men in almost any other population. And despite their marked gender equality, Aka also have one of the largest documented gender differences in smoking prevalence. Although it is costly and desirable, tobacco is widely shared, and appears to play a central role in this defining aspect of Aka culture. Some popular explanations for variation in tobacco smoking, such as tobacco advertising, cannot easily explain the high prevalence of smoking among Aka men, nor can socioeconomic disparities and proscriptions easily explain the low prevalence among Aka women. Labor inducement captures important elements of Aka-farmer relations,

40 yet incorrectly portrays Aka as lacking agency (cf. Brady and Long 2003). Aka men do use tobacco to enhance their work and hunting productivity (labor enhancement), yet they appear to do so to advance their own agendas (e.g. attracting wives, provisioning families) rather than the villagers’.

Reproductive-aged women avoid using tobacco, but postmenopausal women use it more often.

The Aka themselves note the risk that smoking poses to their infants’ health, which perhaps explains why men prefer wives who do not smoke. Women also earn less than men when working for villagers, leaving less money to purchase tobacco, yet income was similar among women smokers vs. non-smokers.

Despite some exceptional features, Aka tobacco use has a number of similarities to the recreational use of tobacco seen in other societies: Aka smoke on a daily basis but especially during social events, such as dances; they do so because they enjoy it; men smoke more than women; and children generally do not smoke. Even the role of tobacco in attracting mates is seen in other societies

(e.g., Kurzban et al. 2010; Jones and Figueredo 2007), as is its role in enhancing physical and mental performance (Groark 2010; Wilbert 1987; Winter 2000).

Although the global prevalence of tobacco use has decreased over the last 30 years, this decrease seems to have stalled (Ng et al. 2014). Even if prevalence continues to decline, exceptional population growth in the developing world ensures that tobacco-caused deaths will impose an enormous burden on these countries for decades to come (figure 1). One bright spot in tobacco statistics in the developing world is the very low female prevalence (Ng et al. 2014). Aka gender differences, and their cultural model of women’s avoidance of tobacco, raise the concern that as women in the developing world adopt birth control and reduce their fertility, they will increase their use of tobacco and other drugs.

Current theories of tobacco use emphasize external “forces”: either nicotine’s neurophysiological

“addictive” effects, or the power and reach of transnational tobacco companies. Our results, in contrast, highlight the importance that local and personal factors have for tobacco use, such as Aka-farmer relations, the preferences of marriage partners and the demands of being a good provider, a desire to dance well, sharing relationships, and the well-being of one’s children.

ACKNOWLEDGEMENTS

The authors thank Aubin Mongosso and the late Nicaise Molende for assistance in the field, and the Aka

41 for generously agreeing to participate in our study. This investigation was supported in part by funds provided to EHH and CJR for medical and biological research by the State of Washington Initiative

Measure no. 171. The authors declare they have no conflict of interest.

REFERENCES

Almedom, A. M., & Sembatu, A. (1994). Women, Moral Virtue and Tchat-chewing. M. McDonald (Ed.),

Gender, Drink and Drugs (pp. 249-259). Oxford: BERG.

Amos, A., & Haglund, M. (2000). From social taboo to “torch of freedom”: the marketing of cigarettes to

women. Tobacco Control, 9(1), 3.

Antropova, V. V. (1964). The Koryaks. In M.G. Levin & L. P. Potapov (Eds.), The peoples of Siberia (pp.

851-875). Chicago: University of Chicago Press. As seen in eHRAF World Cultures in 7 Jan

2015.

Bahuchet, S. (1984). Les Pygmées Aka et la Forêt Centrafricaine. Paris: Selaf.

Bahuchet, S. (1992). Dans la Forêt D’Afrique Centrale les Pygmées Aka et Baka. Paris: Peeters-Selaf.

Bailey, R. C. (1991). The Behavioral Ecology of Efe Pygmy Men in the Ituri Forest, Zaire. Ann Arbor, MI:

Regents of the University of MIchigan.

Bernard, H. R. (2006). Research Methods in Anthropology, Qualitative and Quantitative Approaches,

Fourth Edition. Lanham, MD: AltaMira Press.

Bird-David, N. (1990). The giving environment: another perspective on the economic system of gatherer-

hunters. Current Anthropology, 31(2), 189-196.

Black, P. W. (1984). The Anthropology of Tobacco Use: Tobian Data and Theoretical Issues. Journal of

Anthropological Research, 40(4), 475-503.

Brady, M. & Long, J. (2003). Mutual Exploitation? Aboriginal Australian Encounters with Europeans,

Southeast, Asians, and Tobacco. In W. Jankowiak & D. Bradbury (Eds.), Drugs, Labor, and

Colonial Expansion (pp. 31-58). Tucson, AZ: The University of Arizona Press.

Cipriani, L. & Cox, D. T. (1966). The Andaman Islanders. New York: Frederick A. Praeger, Inc. As seen in

eHRAF World Cultures in 7 Jan 2015.

Damon, A. (1973). Smoking Attitudes and Practices in Seven Preliterate Societies. In W. L. Dunn, Jr.

(Ed.), Smoking Behavior: Motives and Incentives (pp. 219-230). Washington, D.C.: V. H. Winston

& Sons

Eriksen, M., Mackay, J., & Ross, H. (2012). The Tobacco Atlas, 4th Edition. Atlanta, GA: American

Cancer Society.

43 Ezzati, M., & Lopez, A. D. (2004). Smoking and oral tobacco use. In M. Ezzati, A. D. Lopez, A. Rogers, &

C. J. L. Murray (Eds.), Comparative quantiﬁcation of health risks: Global and regional burden of

disease attributable to selected major risk factors (pp. 883–957). Geneva: World Health

Organization.

Foulds, J., Ramstrom, L., Burke, M., & Fagerström, K. (2003). Effect of smokeless tobacco (snus) on

smoking and public health in Sweden. Tobacco Control, 12(4), 349–359, doi:10.1136/tc.12.4.349

Fouts, H. N., & Brookshire, R. A. (2009). Who feeds children? A child's-eye-view of caregiver feeding

patterns among the Aka foragers in Congo. Social science & medicine, 69(2), 285-292.

Goodman, J. (1993). Tobacco in History. The Cultures of Dependence. London: Routledge.

Grinker, R. R. (1994). Houses in the Rainforest. Ethnicity and Inequality among Farmers and Foragers in

Central Africa. Berkeley: University of California Press.

Groark, K. P. (2010). Angel in the Gourd: Social, Therapeutic, and Ritual Uses of Tobacco (Nicotiana

tabacum) among the Tzeltal and Tzotzill Maya of Chiapas, Mexico. Journal of Ethnobiology,

30(1), 5-30.

Gurven, M. (2004). To give and to give not: the behavioral ecology of human food transfers. Behavioral

and Brain Sciences, 27(4), 543-559.

Haddon, A. C. (1946). Smoking and tobacco pipes in New Guinea. Philosophical Transactions of the

Royal Society of London, Series B, 232, 1-278.

Hagen, E. H., Sullivan, R. J., Schmidt, R., Morris, G., Kempter, R., & Hammerstein, P. (2009). Ecology

and neurobiology of toxin avoidance and the paradox of drug reward. Neuroscience, 160(1), 69-

84.

Hagen, E. H., Roulette, C. J., & Sullivan, R. J. (2013). Explaining human recreational use of ‘pesticides’:

the neurotoxin regulation model of substance use vs. the hijack model and implications for age

and sex differences in drug consumption. Frontiers in psychiatry, 4.

Hanawa, R. (2004). The dynamics of the relationship between shifting cultivators and hunter-gatherers

along the Motaba of northern Congo. Journal of African Studies, 64, 19-42.

Hays, T. (1991). "No tobacco, No halleluja": Missions and the early history of tobacco in Eastern Papua.

Pacific Studies, 14, 91-112.

44 Heath, D. B. (1991). Women and alcohol: Cross-cultural perspectives. Journal of substance abuse, 3(2),

175-185.

Hewlett, B. S. (1977). Notes on the Mbuti and Aka Pygmies of Central Africa. Master’s thesis

(Anthropology). Chico, CA: California State University Chico.

Hewlett, B. S. (1991). Intimate Fathers: The Nature and Context of Aka Pygmy Paternal Infant Care. Ann

Arbor, MI: The University of Michigan Press.

Hewlett, B. S. (1992). Father-Child Relations: Cultural and Biosocial Contexts. New York: Aldine.

Hewlett, B. S. (1996). Cultural Diversity Among African Pygmies. In S. Kent (Ed.), Cultural Diversity

Among Twentieth-Century Foragers: An African Perspective (pp. 215-244). Cambridge:

Cambridge University Press.

Hewlett, B. S., van de Koppel, J. M. H., & van de Koppel, M. (1986). Causes of Death Among Aka

Pygmies of the Central African Republic. In L. L. Cavalli-Sforza (Ed.), African Pygmies (pp. 45-

63). London: Academic Press.

Hyman, S. E., Malenka, R. C., & Nestler, E. J. (2006). Neural mechanisms of addiction: the role of

reward-related learning and memory. Annu. Rev. Neurosci., 29, 565–598.

Jankowiak, W. M., & Bradburd, D. (1996). Using drug foods to capture and enhance labor performance:

a cross-cultural perspective. Current Anthropology, 37, 717-720.

Jankowiak, W. M., & Bradburd, D. (2003). Drugs, Labor, and Colonial Expansion. Tucson, AZ: University

of Arizona Press.

Jeffreys, M. D. W. (1963). Smoking in Africa. The South African Archaeological Bulletin, 18(69), 33.

Jochelson, W. (1905) The Koryak: Religion and Myths. In Publications of the Jesup North Pacific

Expedition. New York: E. J. Bril; G.E. Stechert. As seen in eHRAF World Cultures in 7 Jan 2015.

Jones, D. N., & A. J. Figueredo (2007). Mating effort as a predictor of smoking in a college sample.

Current Research in Social Psychology, 12, 186-195.

Kaplan, M., Carriker, L., & Waldron, I. (1990). Gender differences in tobacco use in Kenya. Social

Science and Medicine, 30, 305-10.

Kitanishi, K. (2000). The Aka and Baka: Food sharing among two central Africa hunter-gatherer groups.

Senri ethnological studies, 53, 149-169.

45 Kurzban, R., Dukes, A., & J. Weeden (2010). Sex, drugs and moral goals: reproductive strategies and

views about recreational drugs. Proceedings of the Royal Society B Biological Sciences, 277,

3501-3508.

Laufer, B., Hambley, W. D., & Linton, R. (1930). Tobacco and Its Use in Africa. Chicago: Field Museum of

Natural History.

Lewis, J. (2008). Ekila: blood, bodies, and egalitarian societies. Journal of the Royal Anthropological

Institute, 14, 297-315.

Marlowe, F. (2004). The Hadza. In C. R. Ember & M. Ember (Eds.), Encyclopedia of Medical

Anthropology (pp. 689-696). New York: Springer.

Marshall, M. (1981). Tobacco Use in Micronesia: A Preliminary Discussion. Journal of Studies on Alcohol,

42(9), 885-893.

Marshall, M. (1991). The 2nd Fatal Impact: Cigarette-Smoking, Chronic Disease, and the Epidemiologic

Transition in Oceania. Social Science and Medicine, 33(12), 1327-1342.

Marshall, M. (1993). A Pacific haze: Alcohol and drugs in Oceania. In V. J. Lockwood, T. Harding, & B.

Wallace (Eds.), Contemporary Pacific Society: Studies in development and change (pp. 260-272).

Englewood Cliffs, NJ: Prentice-Hall.

Mathers, C. D. & Loncar, D. (2006). Projections of Global Mortality and Burden of Disease from 2002 to

2030. PLoS Medicine, 3(11), e442.

McDonald, M. (1994). Gender, drink and drugs. Oxford, UK: Berg.

Mintz, S. W. (1985). Sweetness and Power: The Place of Sugar in Modern History. New York: Penguin.

Motte, E. (1980). Les Plantes Chez Les Pygmées Aka et les Monzombo de la Lobaye (Centrafrique):

contribution a une etude ethnobotanique comparative chez des chasseurs-cueilleurs et des

pecheurs-cultivateurs vivant dans un meme milieu vegetal. Paris: SELAF.

Motte-Florac, E., Bahuchet, S., & Thomas, J. M. C. (1993). The Role of Food in the Therapeutics of the

Aka Pygmies of the Central African Republic. In C. M. Hladik, A. Hladik, O.F. Linares, H. Pagezy,

A. Semple & M. Hadley (Eds.), Tropical Forests People and Food: Biocultural Interactions and

Applications to Development (pp. 549-560). Paris: Parthenon.

Munro, N. G. (1963). Ainu Creed and Cult. New York: Columbia University Press. As seen in eHRAF

46 World Cultures in 7 Jan 2015.

Ng, N., Weinehall, L., & Öhman, A. (2007). “If I don’t smoke, I’m not a real man”—Indonesian teenage

boys’ views about smoking. Health Education Research, 22(6), 794–804, doi:10.1093/her/cyl104

Ng, M., Freeman, M. K., Fleming, T. D., Robinson, M., Dwyer-Lindgren, L., Thomson, B., et al. (2014).

Smoking prevalence and cigarette consumption in 187 countries, 1980-2012. JAMA, 311(2), 183–

192, doi:10.1001/jama.2013.284692

Nichter, M., & Cartwright, E. (1991). Saving the children for the tobacco industry. Medical Anthropology

Quarterly, 5(3), 236-256.

Oishi, T., & Hayashi, K (2014). From Ritual Dance to Disco: Change in Habitual use of Tobacco and

Alcohol among the Baka Hunter-Gatherers of Southeastern Cameroon. African study

monographs, S47, 143-163.

Pampel, F. (2008). Tobacco use in sub-Sahara Africa: Estimates from the demographic health surveys.

Social Science & Medicine, 66(8), 1772–1783, doi:10.1016/j.socscimed.2007.12.003

Pearsall, D. M. (1992). The origins of plant cultivation in South America. In C. W. Cowan & P. J. Watson

(Eds.), Origins of Agriculture. An International Perspective (pp. 173-205). Washington, D.C.:

Smithsonian Institution Press.

Quinlan, M. (2005). Considerations for Collecting Freelists in the Field: Examples from Ethnobotany. Field

Methods, 17(3), 1-16.

Ratsch, A., Steadman, K. J., & Bogossian, F. (2010). The pituri story: a review of the historical literature

surrounding traditional Australian Aboriginal use of nicotine in Central Australia. Journal of

Ethnobiology and Ethnomedicine, 6(1), 26, doi:10.1186/1746-4269-6-26

Room, R. (1996). Gender roles and interactions in drinking and drug use. Journal of substance abuse,

8(2), 227-239.

Roulette, C. J., Mann, H., Kemp, B. M., Remiker, M. Roulette, J. W., Hewlett, B. S., et al. (2014). Tobacco

use vs. helminths in Congo basin hunter-gatherers: self-medication in humans? Evolution and

Human Behavior. http://dx.doi.org/10.1016/j.evolhumbehav.2014.05.005

Sahlins, M. D. (1972). Stone age economics. New York: Aldine.

Schensul, S., Schensul, J. J., & LeCompte, M. D. (1999). Essential Ethnographic Methods:

47 Observations, Interviews, and Questions. Walnut Creek, CA: AltaMira Press.

Schrenck, L. V. (1964). The peoples of the Amur region. In Reisen und Forschungen im Amur-Lande in

den Jahren 1854-1856. St. Petersburg: Kaiserliche Akademie der Wissenschaften. As seen in

eHRAF World Cultures in 7 Jan 2015.

Singer, M. (2004). Tobacco Use in Medical Anthropological Perspective. In C. R. Ember & M. Ember

(Eds.), Encyclopedia of Medical Anthropology (pp. 518-528). New York: Kluwar

Academic/Plenum Publishers,.

Stebbins, K. R. (1990). Transnational Tobacco Companies and Health in Underdeveloped Countries:

Recommendations for Avoiding a Smoking Epidemic. Social Science and Medicine, 30, 227-235.

Stebbins, K. R. (2001). Going Like Gangbusters: Transnational Tobacco Companies “Making a Killing” in

South America. Medical Anthropology Quarterly, 15(2), 147-170.

Terashima, H. (1998). Honey and holidays: The interactions mediated by honey between Efe hunter-

gatherers and Lese farmers in the Ituri forest. African study monographs, S25, 123-134.

Terashima, H., & Ichikawa, M. (2003). A comparative ethnobotany of the Mbuti and Efe hunter-gatherers

in the Ituri forest, Democratic Republic of Congo. African Study Monographs, 24(1/2), 1-168.

The World Bank (2012). Regional Highlights, World Development Indicators 2012. Development Data

Group. The World Bank. http://data.worldbank.org/sites/default/files/wdi-regional-highlights2012-

web.pdf. Accessed 8 Jan 2015.

Turnbull, C. M. (1961). The forest people. New York: Simon and Schuster.

Tushingham, S., Ardura, D., Eerkens, J. W., Palazoglu, M., Shahbaz, S., & Fiehn, O. (2013). Hunter-

gatherer tobacco smoking: earliest evidence from the Pacific Northwest Coast of North America.

Journal of Archaeological Science, 40(2), 1397–1407, doi:10.1016/j.jas.2012.09.019

Vallance, P.J., Anderson, H.R., & Alpers, M. P. (1987). Smoking habits in a rural community in the

highlands of Papua New Guinea in 1970 and 1984. Papua New Guinea Medical Journal, 30, 277-

80.

Van der Merwe, N. J. (2005). Antiquity of the smoking habit in Africa; human behaviour and culture.

Transactions of the Royal Society of South Africa, 60, 147-150.

Von Gernet, A. (2000). North American indigenous nicotiana use and tobacco shamanism: the early

48 documentary record, 1520–1660. In J. C. Winter (Ed.), Tobacco Use by Native Americans:

Sacred Smoke and Silent Killer (pp. 59-80). Norman, OK: University of Oklahoma Press.

Waldron, I., Bratelli, G., Carriker, L., Sung, W.-C., Vogeli, C., & Waldman, E. (1988). Gender differences

in tobacco use in Africa, Asia, the Pacific, and Latin America. Social Science and Medicine, 27(1),

269-75.

Watson, P. L. (1983). This precious foliage: A study of the aboriginal psycho-active drug Pituri. Sydney:

University of Sydney.

WHO (2007). Gender and tobacco control: A Policy Brief. World Health Organization.

http://www.who.int/tobacco/resources/publications/general/policy_brief.pdf. Accessed 8 Jan 2015.

Wilbert, J. (1987). Tobacco and Shamanism in South America. New Haven, CT: Yale University Press.

Winter, J. C. (2000). Tobacco Use By Native North Americans: Sacred Smoke and Silent Killer. Norman,

OK: University of Oklahoma Press.

Woodburn, J. (1982). Egalitarian Societies. Man, 17(3), 431–451.

49 CHAPTER THREE

TOBACCO USE VS. HELMINTHS IN CONGO BASIN HUNTER-GATHERERS: SELF-

MEDICATION IN HUMANS? Evolution and Human Behavior, 35(5):397-407 (2014).

http://dx.doi.org/10.1016/j.evolhumbehav.2014.05.005

Roulette CJ, Mann H, Kemp BM, Remiker M, Roulette JW, Hewlett BS, Kazanji M, Breurec S,

Monchy D, Sullivan RJ, and Hagen EH

Abstract

We tested a novel hypothesis that recreational use of neurotoxic plants helps defend against parasites. Specifically, we investigated the relationship between smoking and helminthiasis among the Aka, a remote population of Central African foragers who are avid tobacco smokers, suffer high rates of helminthiasis, and have little-to-no access to commercial anthelmintics. Two hundred and six healthy Aka men provided saliva and stool samples. Saliva samples were assayed for cotinine, a nicotine metabolite; a subsample was genotyped for the

CYP2A6 enzyme, which metabolizes nicotine. Stool samples were assayed for intestinal helminth eggs as an index of worm burden. After 1 year, a subsample of participants was located and provided additional saliva and stool samples. We found (1) an exceptionally high prevalence of tobacco use, (2) a strong negative correlation between cotinine (a nicotine metabolite) and worm burden, (3) that treating helminths with albendazole, a commercial anthelmintic, reduced cotinine concentration two weeks later, compared to placebo controls, (4) among treated participants, higher cotinine concentrations in year 1 predicted less reinfection by year 2, and (5) younger and older participants with slow nicotine-metabolizing CYP2A6 alleles had lower worm burdens compared to those with extensive metabolizing alleles. These results provide the first evidence of a link between helminthiasis and smoking. They also suggest that, in populations where intestinal helminths are endemic, tobacco use might protect against helminth infection and reduce worm burden among infected individuals, and that individuals modulate nicotine exposure in response to infection. The results thus support the hypothesis that substance use helps defend against parasites.

51 1. Introduction

The ancient coevolutionary relationship between vertebrates and helminths (parasitic worms) has potentially profound implications for human health. The fundamental structure and control of the immune system was likely shaped by vertebrate-helminth coevolution, for instance, and the evolutionarily novel absence of helminth infections in many populations might therefore be responsible for the increasing prevalence of allergic and autoimmune diseases (the hygiene hypothesis) (Jackson et al., 2009).

It is possible that behavioral anti-parasite strategies, also referred to as non- immunological defenses, were also shaped by vertebrate-helminth coevolution. Plants have evolved an enormous variety of toxins to deter herbivores, and herbivores have co-evolved to use plant toxins to defend against their own parasites, a phenomenon referred to as self- medication, zoopharmacognosy, or pharmacophagy (Boppré 1984; Rodriguez & Wrangham

1993; Glander 1994). Woolly bear caterpillars (Grammia incorrupta), for example, ingest pyrrolizidine alkaloids to protect themselves from tachinid flies (Singer et al., 2009). Tobacco hornworm (Manduca sexta) larvae co-opt ingested nicotine to defend against the wolf spider

(Camptocosa parallela) (Kumar et al., 2013) and the endoparasitoid, Cotesia congregata

(Barbosa et al., 1991; Thorpe and Barbosa, 1986). In an interesting use of plant toxins among urban sparrows and finches, cigarette butts containing residual nicotine are placed in nests to defend against nest-dwelling ectoparasites such as mites (Suárez-Rodríguez et al., 2013).

Some species ingest plant toxins to specifically defend against helminths. Domesticated lambs, for instance, respond to gastrointestinal parasite infections by increasing their intake of alfalfa tannins (Villalba et al., 2010). Wild chimpanzees (P. troglodytes), which are known to consume a variety of medicinal foods (Huffman 1997), ingest Aspilia (Compositae) and the bitter pith of

Vernonia amygdalina to treat intestinal helminth infections (Wrangham and Nishida 1983;

Huffman 2001).

In humans, there is a large literature on the deliberate use of plants for medicinal purposes (ethnomedicine). In addition, there is increasing evidence that the plants that are incorporated into many diets provide protection against parasites (Etkin 1986; Etkin and Ross

1991). Indeed, in many cultures, the distinction between plant “foods” and plant “medicines” is not always clear—i.e. many of the plants that are used as medicines are also consumed as foods (Etkin and Ross 1982, 1991; Moerman 1996). For example, 90 percent of the plants commonly used by the Hausa of Nigeria to treat malaria are also consumed as foods (Etkin and

Ross 1982, 1991), and cassava (Manihot esculenta) and fava bean (Vicia faba) contain glycosides that, when consumed as part of the normal diet, protect against Plasmodium falciparum malaria among West African and Mediterranean populations, respectively (Jackson

1990, 1996; Katz and Schall 1986). Similarly, Billing and Sherman (1998) argue that the addition of spices to food functions to defend against meat-borne pathogens (see also Sherman and Billing 1999).

It is possible that the human “recreational” use of psychoactive plant drugs is explained, in part, by the antiparasitic properties of these plant compounds. Wyatt (1977), for example, noted the anthelmintic (anti-worm) properties of betel nut, which is widely chewed in Asia and the Pacific. Rodriguez and Cavin (1982) proposed that the origin of tropical hallucinogenic plant use in Amazonia might be explained by the anthelmintic and antimicrobial properties of the plants. Intriguingly, three of the world’s most popular recreational drugs are effective treatments against helminths: nicotine (the psychoactive component of tobacco) and arecoline (the psychoactive component of betel-nut) have been used as commercial anthelmintics in animals

(Hammond et al., 1997), and cannabis is toxic to plant-parasitic nematodes (Mukhtar et al.,

53 2013). We have proposed that human recreational drug use might involve neurophysiological mechanisms that evolved to regulate the intake of plant neurotoxins to prevent or reduce infections of helminthes and other macroparasites (Hagen et al., 2009; Hagen et al. 2013).

There need be no conscious awareness of the antiparastic properties of “recreational” drugs, however, just as there is no common awareness of the antiparasitic properties of spices.

1.1. The chemoprophylaxis and chemotherapy hypotheses

Based on an analogy with plant defenses, there are at least two pharmacophagy hypotheses for the human use of psychoactive plant toxins. First, even when they are not under attack, plants maintain a basal level of plant toxins to deter attacks, a form of chemoprophylaxis known as a constitutive defense. Second, when they detect an herbivore attack, plants often up- regulate energetically expensive toxin production, down-regulating production after the attack ceases, a form of chemotherapy (in its non-cancer specific sense) known as an inducible defense (Baldwin 1999).

Human psychoactive substance use might similarly be explained, at least in part, by both chemoprophylaxis and chemotherapy. Under the chemoprophylaxis hypothesis, the human propensity to consume neurotoxic plants prevents or limits infection by helminths, one hypothesis we test here (populations with virtually no risk of helminth infection, such as the US and other high-income countries, might thus still exhibit considerable use of such substances; if so, this would be an example of an evolutionary mismatch). Under the chemotherapy hypothesis, consumption of psychoactive substances should be up-regulated by infection, serve to limit infection levels (we test this hypothesis here), and be down-regulated if and when the infection is cleared (we test this hypothesis here).

1.2. Nicotine as a model drug

There is an especially solid physiological basis for believing that nicotine could serve as an anthelmintic in humans. Modern anthelmintics such as levamisole and the tetrahydropyrimidines achieve their effects by targeting the same receptors as nicotine – the nicotinic acetylcholine receptors (nAChRs) on somatic muscle cells – which induces spastic paralysis of the parasite, leading to its expulsion (Kohler 2001). Further, nicotine is an extremely potent neurotoxin and the quantities absorbed via smoking and chewing are surprisingly high.

Nausea and vomiting are typically induced by doses of 2-5 mg. The lethal dose for an adult human has been estimated to be 30-60 mg (Gosselin et al., 1984; but see Mayer, 2014, who argues that the lethal oral dose of nicotine might be 10x higher). Because smokers typically absorb 1-2 mg per cigarette, and chewers up to 4.5 mg per “wad” (Hukkanen et al., 2005), tobacco users are regularly absorbing quantities of nicotine just below that which induces acute toxic response.

We are not proposing that humans evolved to consume nicotine specifically. Humans evolved in the Old World, and there is currently little evidence that Old World plants contained high amounts of nicotine (although some Old World foods, such as eggplant, contain low levels of nicotine). Instead, we argue that humans evolved to regulate intake of broad classes of psychoactive plant toxins (Sullivan et al. 2008; Hagen et al. 2009; Hagen et al. 2013).

Nevertheless, nicotine consumption is not simply an artifact of modern life. Nicotiana species were widely used by hunter-gatherers in the Americas (i.e. tobacco) and Australia (i.e. pituri) for thousands of years. Nicotiana tabacum, the most well known and widely distributed tobacco species, was domesticated by Andean farmers ca. 1000 YBP (Pickersgill 2007), but was being consumed by hunter-gatherers in the south-central Andes at least as early as 100 BC

55 (Echeverría and Niemeyer 2013). By the time of European contact tobacco was being used in a variety of contexts throughout North and South America—it played an important function in many rituals, it was often a medicinal panacea, yet it was also consumed in more mundane, daily, recreational contexts (Tushinham et al. 2013; Wilbert; Winter 2000). Soon after European contact, tobacco was brought to the Old World. Today over 20 percent of the global population smokes tobacco (Erickson, Mackay, Ross 2012).

1.3. Population variation in nicotine exposure

1.3.1 Age and sex differences

Cross-nationally, users of popular psychoactive substances, including tobacco, report virtually no use prior to the age of 10. Starting about the age of 12 there is a rapid increase in substance use, so that almost everyone who will ever use a substance has done so by age 20.

We argue elsewhere that during childhood the risk that plant neurotoxins would disrupt development of the nervous system outweighs any anti-parasitic benefits, so children have an evolved aversion to drug consumption. During adolescence the balance shifts, with increasing benefits outweighing diminishing costs, leading to substantial drug use (Hagen et al. 2013).

Similarly, cross-nationally, men almost always have a higher prevalence of tobacco and other drug use than women (Degenhardt et al. 2008; Hagen et al. 2013). Although proscriptions on women’s behavior and lack of access to drugs likely play a role in some of the sex differences in drug use (Kaplan 1990, Mackay and Eriksen 2002, Waldron 1988), the risk of exposing their fetuses and nursing infants to plant teratogens means that drug use entails a higher fitness cost for women than for men. Therefore, it is possible that the sex difference in

substance use is, at least in part, explained by the sex difference in the costs versus antiparasitic or other benefits of toxin consumption (Hagen et al. 2013).

1.3.2 Differences in drug metabolism

Drug exposure (i.e. the level of a psychoactive substance in the body) is related to both intake and elimination (e.g. drug metabolism). Most drugs, including nicotine, are metabolized by a superfamily of detoxification enzymes known as cytochrome P450 (CYP) haemoproteins.

Mammalian CYP enzymes are concentrated in the liver, where the majority of them function to metabolize endogenous fatty acids and perform steroidogenesis. However, a subset of CYPs evolved to also detoxify xenobiotics (e.g. dietary phytochemicals) (Lewis 2001; Nelson 1999).

About a dozen of the 57 human P450 enzymes are primarily responsible for xenobiotic metabolism. Compared to other CYP enzymes, the xenobiotic metabolizing enzymes are highly polymorphic (Nelson 1999; 2004). These polymorphisms, some of which occur at high frequencies in some populations, can have a profound influence on drug elimination rates.

Nicotine is primarily metabolized by CYP2A6, which has over 30 currently recognized alleles (Nelson et al., 2008). Compared to the wild type, some 2A6 alleles increase nicotine metabolism and some reduce metabolism, or are non-functional. Most of the alleles have low frequencies, but a few have high frequencies in some populations. For example, approximately

20% of Japanese have a slow or non-functioning 2A6 allele (Sullivan et al., 2008).

We are proposing that, compared to normal or fast metabolizers, slow metabolizers might have increased protection against infection because serum nicotine concentrations remain higher for a longer period of time, possibly increasing the prophylactic and/or therapeutic effects.

2. Study

The impact of tobacco use on human helminthiasis has never been investigated, so far as we know, yet one third of the world’s adult population, or 1.3 billion individuals, are tobacco users (Guindon & Boisclair 2003), smoking about 15 billion cigarettes every day, and approximately one billion people in developing regions of sub-Saharan Africa, Asia, and the

Americas are infected with one or more helminths (Hotez et al., 2008). Hence, the regular intake of an anthelmintic – nicotine – by a broad swath of the global population could be critical to, both, the epidemiology of helminth diseases, which comprise a substantial fraction of disease burden (Hotez et al., 2008), as well as understanding the alarming increase in smoking in the developing world, where tobacco-attributable deaths are expected to double in the next 20 years (Mathers & Loncar 2006).

To test the pharmacophagy hypothesis for the human ‘recreational’ use of tobacco

(Hagen et al., 2009; Sullivan et al., 2008) we required a population with a high prevalence of helminthiasis yet with little-to-no access to commercial anthelmintics, whose members might therefore be motivated (consciously or unconsciously) to seek out and consume locally available substances, like nicotine, that are potentially effective against helminths. We also needed a population with ready access to tobacco.

We chose a remote population of African hunter-gatherers, the Aka of the Central

African Republic (CAR), which met each requirement. Aka “pygmies” number between 15,000 and 30,000. (The term pygmy is now viewed as derogatory in some contexts, but no suitable replacement has yet emerged.) The Aka share several traits with other foragers across the

Central African rain forest, such as a strong identity with (and preference for) forest life, relative egalitarianism, polyphonic music, and an association with farmer populations. The Aka spend part of the year camped near villages working in the fields for these farmers, and the rest of the year in the forests hunting and gathering. Aka camps are dispersed along trails that radiate out from farming villages into the forest, with about 6–12 adult Aka per camp (Hewlett 1996).

Previous studies found a high rate of helminthiasis in Aka and other Congo basin hunter- gatherers (Lilly et al., 2002; Froment, 2014).

2.1 Aka recreational drug use

Like other central African foragers, Aka smoke tobacco, marijuana, and the leaves of a native tree they call motunga (Polyalthia suaveolens, Annonaceae, a.k.a. Greenwayodendron suaveolens Verdc., Annonaceae; Engl. & Diels) (Hewlett 1977; Oishi and Hayashi 2014). The

Aka gather motunga leaves from the trees in the forest, dry them, and roll them in a leaf to smoke. The psychoactive compounds in Polyalthia suaveolens have not yet been identified.

Motunga smoking dates back at least three or four generations, based on our interviews, and the Aka refer to it as the “tobacco of the ancestors,” as do the Baka foragers of Cameroon

(Oishi and Hayashi 2014). Given the Aka’s rich ethnobotanical knowledge, they probably used motunga, and perhaps other psychoactive plant drugs, long before cannabis and tobacco were introduced into the region. Alternatively, because motunga is typically used when Aka do not have tobacco, such as when they are in the forest, they might have only begun to smoke it as a replacement for tobacco and cannabis after those drugs were introduced.

Cannabis, which contains the psychoactive compound THC, is native to Asia. The approximate date that cannabis was introduced to Africa is disputed, although it was grown in

Egypt for as many as a thousand years (Du Toit 1975). Cannabis spread down the east coast

59 with the Arab caravan trade and might have reached both southern Africa and the Congo Basin before European colonization in the 1500’s. The Aka and their trading partners do not smoke cannabis as frequently as they smoke tobacco, perhaps because it is currently illegal in the

Central African Republic, which might limit availability.

Tobacco is native to the Americas and was introduced to the African west coast by

Europeans as early as the 1500’s (Jeffreys 1963). It quickly spread throughout the continent and into the interior. There are two forms of tobacco the Aka smoke. The first is called bangaya, and is grown locally by neighboring Ngandu farmers. The second are manufactured cigarettes, sold by the Ngandu and other villager merchants in the market. Aka provide labor or forest products to their village-trading partner in return for cannabis and tobacco.

Aka perceive smoking these plant drugs to be both beneficial and harmful. On the one hand, tobacco is believed to increase strength and warmth, making one a better worker, hunter, and dancer. The Aka also highlight ndjala, or a “desire,” as a main reason why they continue to use tobacco. On the other hand, most Aka recognize that long term use can lead to negative health consequences, such as prolonged coughing and chest pain or birth defects (unpublished data, Roulette, C.J., Hagen, E.H., and Hewlett, B.S.).

Although the Ngandu farmers control access to most of the psychoactive compounds used by the Aka (i.e. tobacco, cannabis, and alcohol) we currently have little data on Ngandu use of plant drugs. There has also been surprisingly little systematic study of the recreational use of tobacco and other drugs in extant hunter-gatherers or similar small-scale societies (Black

1984). Damon (1973) presents results from a survey of smoking in seven “preliterate” societies, including the !Kung. Although the sample sizes were small (e.g., 14 !Kung), he found that adults smoked as much as possible, unless forbidden by religion, and that, compared to the other

groups, the !Kung had more favorable attitudes towards smoking, which served an important social function, similar to what we observed among the Aka.

2.2 Aka medicinal uses of recreational drugs

The Aka do not associate smoking tobacco, cannabis or motunga with any antiparasite or medicinal benefits. It is intriguing, however, that they make a tea with the roots and bark of motunga to treat helminths (unpublished data, Roulette et al.). According to laboratory studies,

P. suaveolens contains numerous pharmacological compounds (Abad et al., 2007; Lamidi et al.,

2005), including the antiparasitic compound polycarpol (Nyasse et al., 2006), and the antibacterial compound suaveolindole (Yoo et al., 2005). Local peoples throughout Africa have taken advantage of these compounds by incorporating the plant into their pharmacopeia. In some parts of Africa, for example, it is used for rheumatism and toothache, and even as an aphrodisiac (Okorie 1980). The Baka, a closely related group of forest foragers (Bahuchet 1992,

1993), use P. suaveolens (which they term botunga) for a variety of medicinal purposes including treating headaches, snake-bites and malaria (Betti 2004; Hattori 2010). Virtually every

Aka we interviewed used motunga as a treatment for parasites. However, the Aka also have numerous other medicinal applications of motunga, most of which involve making a tea with the roots and bark of the plant.

Although Cannabis has anti-parasitic activity (Mukhtar et al. 2013), the Aka do not use it to treat or prevent helminth infection. The only medicinal use of cannabis mentioned by the Aka was as a treatment for yellow fever (consumed as a tea).

Of the three recreational plants drugs used by the Aka, tobacco is the only known source of nicotine; it also contains numerous other bioactive compounds, such as nornicotine and

61 anabasine (Hukkanen et al., 2005). Tobacco is included in traditional pharmacopeias throughout

Africa, including as a treatment for epilepsy and the Guinea worm (Neuwinger, 1996). In the southwest of C.A.R. a tobacco tea is consumed to treat hemorrhoids. The only medicinal use of tobacco mentioned by the Aka we interviewed was an epicutaneous treatment of an unidentified skin infection, called dombo. Although Aka use tea infusions to treat numerous ailments, we found no evidence of them using tobacco tea. The fact that there are many medicinal uses of motunga, including as an anthelmintic, but few medicinal applications of cannabis and tobacco, might suggest a long history of motunga use, but a relatively recent use of cannabis and tobacco.

The Aka are one of the most egalitarian societies known to anthropology. Nevertheless,

95% of Aka men self-report tobacco use compared to only about 15-30% of Aka women, depending on the population surveyed. The high male prevalence might be partially explained by a pro-tobacco enculturation of young Aka males—i.e. children grow up with adult males who smoke tobacco, they learn that tobacco is associated with success in male subsistence activities, and women prefer male partners who use tobacco because it is associated with success in subsistence activities. The low female use is consistent with fetal protection. Aka men and women mention that tobacco causes harm to a developing fetus and/or breastfed child. While women are not proscribed from using tobacco, the fear of harming their children is a strong motivation preventing many reproductive-aged women from using tobacco (in comparison, a significantly greater proportion of post-reproductive aged women smoke tobacco than do reproductive aged women) (unpublished data, Roulette et al.).

2.3 Predictions

Although the hypothesis we propose applies to all recreational plant drugs, here we only investigate the use of tobacco. We tested four predictions, the first three derived from the chemotherapy hypothesis and one from the chemoprophylaxis hypothesis. First, in a population with endemic helminth infections, worm burden should be inversely correlated with nicotine exposure from tobacco use (i.e., nicotine treats helminth infections). Second, treating helminth infections with a commercial anthelmintic should reduce smoking (i.e., drug intake is down- regulated when it is no longer needed) relative to placebo controls, with the biggest reduction in those with the highest worm burden. Third, individuals with CYP2A6 alleles that reduce nicotine metabolism (slow metabolizers) should have higher nicotine levels for a given level of smoking, and thus lower worm burdens than individuals with normal or fast metabolizing alleles. Fourth, among individuals whose helminth infections have been treated with a commercial anthelmintic, higher smoking levels should limit reinfection with helminths (chemoprophylaxis).

3. Materials and methods

This study was approved by the Washington State University Institutional Review Board

(IRB) and the IRB of the Medical School of the University of Bangui, CAR. Informed consent was obtained from all participants. Consent was oral, and not written, because the Aka are non- literate. Both IRBs approved oral consent, and consent was documented by a local translator.

Permission to conduct this research was obtained from the Ministere de l’Education Nationale, de l’Enseignement superieur et de la Recherche, CAR.

Because so few women self-report as smokers, and because many variables in the study vary by sex, we only tested our predictions in male participants. However, to confirm the low prevalence of female smoking, we also recruited a small sample of women.

Participants were recruited from three regional subpopulations along the Lobaye river,

CAR, with a combined Aka population of about 300 adult men and 300 adult women. We visited all camps within 1 km of a logging road and asked all healthy men present to participate. The sample comprises the 206 men from 69 camps on 23 trails who provided informed consent and a self-report of tobacco use, and for whom we had an age estimate. Nearly all remaining men were absent on extended hunting trips (see the CONSORT diagram in supplemental materials).

We did not pre-screen participants for tobacco use, health complaints, or on any other criteria.

We recruited 44 adult women from the largest of the 3 regional subpopulations.

3.1. Sample collection

Using provided kits, participants supplied 2–5 ml of saliva and one stool sample every other day for about six days, for a total of three baseline samples each (s1-s3). We instructed participants to provide saliva immediately upon wakening and before smoking their first cigarette

(after first rinsing their mouths with water). Stool was collected using ParaPak vials with formalin. Saliva was stored at -20C in a solar-powered freezer. Three post-intervention samples

(s4-s6) and three one-year samples (s7-s9) were collected similarly. Participants received the equivalent of 1 day’s wage for every 3 samples returned (see fig. 5).

3.2. Cotinine assay

Nicotine plasma half-life is 2-3 hours and cotinine half-life approximately 17 hours.

Cotinine is thus a superior biomarker of recent nicotine exposure (Benowitz 1996). Saliva samples were assayed in the Bioanthropology Laboratory at WSU Vancouver using a

Salimetrics enzyme immunoassay kit according to the manufacturer’s protocol.

3.3. Semi-quantification of worm burden

Stool samples were analyzed by The Institut Pasteur in Bangui, CAR. A total of three species of geohelminths were assessed for the worm burden score—Trichuris trichiura (round worm); Ascaris lumbricoides (giant roundworm); and hookworm (including both Ancylostoma duodenale and Necator americanus because their eggs cannot be distinguished). Worm burden was estimated per-species using (1) a wet smear that was microscopically examined for helminth eggs; (2) the merthiolate iodine formaldehyde (MIF) concentration technique (Blagg et al., 1955); and (3) the Kato method (Komiya & Kobayashi 1966). Worm burden comprised the average number of eggs on each x40 microscope field. There were thus 3 measurements for each of the 3 helminth species (hookworm, whipworm, Ascaris), for a total of 9 measurements per stool sample, which were summed to produce the sample worm burden score. The MIF technique is the most sensitive, so this count was weighted by a factor of 2. The sum was then adjusted for the volume of stool examined.

3.4. Control variables

The negative relationship between cotinine levels and worm burden predicted by the pharmacophagy hypothesis could instead result if, e.g., wealth, acculturation, location, other substances, or age confounded tobacco use and access to commercial anthelmintics, health care, exposure to helminths, or immunity. To help rule out some of these alternative hypotheses, we measured 13 control variables in 5 domains: wealth, acculturation, use of commercial anthelmintics, subpopulation, and age. Material wealth was the self-reported number of clothes, watches, flashlights, radios, and batteries owned, with the ‘wealth score’ equaling the total sum. Acculturation was self-reported attendance in school as a child and as

65 an adolescent, church attendance, and preference for living in the village vs. the forest (yes=1, no=0), with the ‘acculturation score’ equaling the total sum. Participants provided a self-report of use of worm pills or traditional treatment for worms in the last year. The worm pills are available in the market and are widely distributed commercial anthelmintics like albendazole and mebendazole. The traditional medicine was often motunga tea. All individuals were coded as belonging to one of the 3 regional subpopulations. Aka do not keep track of age, so age was estimated by Ngandu research assistants who had lifelong associations with these Aka populations.

3.5. Randomized control trial (RCT)

Eligibility for the RCT required male sex, returning at least one saliva and one stool sample, providing a tobacco use self-report, having an age estimate, and being camped near the road when participants were randomized into equal-sized treatment and control groups

(double-blind).

Packets containing two pills were prepared by an individual not otherwise involved in the research. In treatment packets the first pill contained 400 mg albendazole (a full course) and the second was a placebo of identical appearance, whereas in control packets the first pill was placebo and the second albendazole. Packets were code-labeled and the list kept secret.

Researchers blind to packet contents assigned them to participants and then observed participants take pill #1 (along with a fatty pastry to improve absorption). The immediate psycho- physiological effects of albendazole are difficult to distinguish from baseline symptom rates

(Horton 2000), which should have minimized study participants’ ability to distinguish albendazole from placebo. About two weeks later, participants provided 3 follow-up saliva and

stool samples (s4-s6), and then took pill #2, concluding the RCT (all participants thus ultimately received albendazole treatment).

3.6. Cotinine vs. reinfection

Aka shift residence every few weeks or months. After 1 year, 22 men in the treatment group were located; these men provided saliva and stool samples (s7-s9). A reinfection score was computed by subtracting mean post-treatment year 1 worm burden scores (s4-s6) from mean year 2 worm burden scores (s7-s9).

3.7. CYP2A6 genotyping

Due to limited funds, we could only genotype a random sub-sample of 39 males for common CYP2A6 polymorphisms. Compared to the wild type, CYP2A6*1B is associated with faster in vivo nicotine metabolism, CYP2A6*9 and CYP2A6*17 with slower metabolism, and

CYP2A6*20 lacks enzymatic activity (Mwenifumbo & Tyndale 2009). We obtained DNA samples from cheek swabs and Whatman FTA cards or dried blood spots stored at -20 C.

We extracted DNA using Qiagen reagents and blood kit materials. CYP2A6*1B was tested using a two-step allele-specific PCR method; CYP2A6*17 by a PCR-RFLP method; and

CYP2A6*9 and CYP2A6*20 by PCR amplification followed by sequencing. Unclassified alleles were assigned to the CYP2A6*1A wild type allele category (Nakajima et al., 2006).

A predicted metabolic phenotype was computed by assigning fast, normal, and slow alleles scores of 1, 0, and -1, respectively, and then summing the two scores for each individual.

Because only 1 individual had a predicted phenotype > 0, he was assigned to the ‘Normal’

67 category (n=27) and individuals with a phenotype < 0 were assigned to the ‘Slow’ category

(n=12).

3.8. Data analysis

Statistics were computed with R version 2.15.0 (2012-03-30). The worm burden score was a sum of egg counts, and was overdispersed, and was therefore modeled by a negative binomial distribution. Cotinine concentration was square-root transformed to better approximate a normal distribution when used as a response variable. To assess the effects of population structure (camps and trails), models with and without random effects for these groups were compared by AIC when technically feasible. ΔAIC was < 1 in each case, so only models without random effects are reported. All data will be made available to qualified researchers upon request.

4. Results

Table 1 presents summary statistics of the sample at baseline (samples s1-s3; see also fig. 5). The sample size for cotinine concentration and worm burden is reduced because some participants did not return any saliva or stool samples. Sample size for wealth and acculturation is reduced because these data were collected during a second interview upon the receipt of sample s3, and some attrition had occurred (fig. 5).

4.1. Helminth infection patterns

Helminth infections were widespread, with 97.9% of the total sample (male and female) testing positive for at least 1 species of helminth. Hookworm was found in 95.4% of the sample, whipworm in 52.7%, and Ascaris in 54.8%.

Polyparasitism was common in our sample, with 47.7% infected with two helminth species, and 28.6% infected with three. There were 7.1% uninfected females and 1% uninfected males, a difference that was marginally significant (χ2= 3.8, p = 0.052). Otherwise, there were no significant sex differences in worm burden, either per species, or in aggregate.

4.2. Smoking patterns

By self-report, 95% of the men were tobacco smokers, and 70% were cannabis smokers, whereas only 16% of the women were tobacco smokers and 5% were cannabis smokers. Using a 5 ng/ml threshold concentration of salivary cotinine to distinguish regular smokers from nonsmokers exposed to environmental tobacco smoke (ETS) (see Supporting information), 94% men were recent tobacco smokers, which accords well with their self-reported use of tobacco, but only 5% of women had recently smoked, a marked discrepancy with their

16% self-reported smoking rate (we did not collect urine samples necessary to assay THC). The geometric mean cotinine concentration among recent Aka male smokers was 100.1 ng/ml. In comparison, the geometric mean cotinine concentration in a large, nationally representative sample of US male smokers was 122.1 ng/ml (Benowitz et al., 2009).

4.3. Cross-sectional relationship between smoking and worm burden

Missing interview data for some cases yielded a final sample size of 177 men for the analysis of worm burden as a function of cotinine concentration, controlling for age, acculturation, wealth, and subpopulation. Worm burden was higher in the subpopulation located

69 near a large farming village, and lower in the other two, so subpopulation was re-coded as

‘village’ vs. ‘not village.’ Participants who had been treated for worms at the local health clinic

(n=7) had significantly higher worm burden scores than those who had not (W = 319.5, p =

0.019). However, they represented only 3% of the male sample. In addition, the hypothesis applies only to individuals without access to commercial anthelmintics. These participants were therefore excluded from this particular analysis only (including them with a dummy variable yielded qualitatively similar results).

A generalized additive model (GAM) (Wood 2006) of worm burden as a smooth function of cotinine, controlling for a smooth function of age and linear functions of village, acculturation, and wealth, found that cotinine was a significant predictor (p = 0.00042; see fig. 1; for model parameters, see table S1).

Examination of figure 1B shows that, at high levels, cotinine concentration is strongly negatively correlated with worm burden, as predicted, but at low levels the relationship is essentially flat.

Figure 1 about here

4.4. Randomized controlled trial

Of the original sample of 206 men, 179 (86.9%) participated in the RCT. See figure 5.

The number of days between the intervention and the start of follow-up (between collection of sample s3 and s4) ranged from 4 to 30, with a mean of 13.6 (SD = 3.9).

4.4.1 Randomization check

To confirm that the sample was successfully randomized, treatment and control groups were compared on size, age, baseline cotinine concentrations, baseline worm burden scores, and distribution among the regional subpopulations. Mean values were compared with t-tests

(transforming variables, if necessary), the nonparametric Wilcoxon rank-sum test, a bootstrap of mean values, and the Kolmogorov-Smirnov test, which is sensitive to both location and shape.

No significant differences were found between the treatment and control groups (results not reported).

4.4.2 Attrition

Attrition in an RCT reduces statistical power, can bias estimates of the treatment effect, and can undermine the generalizability of the study. In this study, attrition was primarily caused by participants leaving for extended trips to forest to hunt and gather after receiving the first pill in their packets, and not returning prior to the departure of the research team from the field

(such trips are typical for Aka). The primary concern is that dropout differed between the two arms of the RCT, and was related to the outcome measure, nicotine exposure.

There was some evidence of biased attrition: Two-way ANOVA’s of age and baseline cotinine and worm burden values as functions of intervention group and lost status, and their interaction, found that, compared to the remaining participants, participants lost to follow-up were significantly younger (M=32.7 vs. M=38.1, F(1, 175)=11.1, p=0.001) and had higher mean worm burden scores (M=28.2 vs. M=20.9, F(1, 175)=4.8, p=0.03), but did not have significantly different mean cotinine concentrations (M=129.6 vs. M=159.7, F(1, 175)=0.5, p=0.47). However, there were no significant interactions of intervention group with lost status, i.e., no significant differences between treatment and control groups in the lost vs. remaining groups for age (F(1,

175)=0.73, p=0.39), cotinine (F(1, 175)=0.04, p=0.85), or worm burden (F(1, 175)=0.02,

71 p=0.89). Despite the biased loss, remaining participants did not differ significantly from the original sample on mean age, baseline cotinine concentration or baseline worm burden. See table S2.

4.4.3 Manipulation check

According to a summary of the existing literature, the mean cure rate and egg reduction rates for a single 400 mg dose of albendazole are, respectively, 77.7% and 87.8% for

Ancylostoma duodenale/Necator americanus (hookworm), 94.6% and 98.6% for Ascaris lumbricoides, and 47.7% and 75.4 % for Trichuris trichiura (whipworm) (Horton 2000).

Consistent with the existing literature, the albendazole treatment group experienced a dramatic reduction in worm burden, Mpre = 21.1 vs. Mpost = 6.7, Cohen’s d = 0.96 (W=918.5, p=0), whereas the placebo control group experienced no significant reduction in worm burden Mpre =

24.9 vs. Mpost = 27.1, Cohen’s d = -0.09 (W=2795, p=0.84). See figure S1.

Whereas there was 1 individual in the treatment group who exhibited no evidence of worm infection at baseline, there were 16 such individuals (24.6%) at follow-up. Albendazole treatment failed to reduce worm burden in 10 individuals (15.4%), which is consistent with previous studies of albendazole treatment of hookworm and whipworm (Horton 2000).

4.4.4. Effect of albendazole treatment on cotinine concentration

Follow-up cotinine values were highly correlated with baseline cotinine values (r=0.8, p<.001). We therefore fit an ANCOVA with follow-up cotinine as the outcome, baseline cotinine and baseline worm burden as covariates, and intervention (treatment/control) as a factor variable, plus its interaction with worm burden (table 2 and figure 2).

Table 2 about here

Figure 2 about here

Results supported the main prediction: compared to controls, albendazole treatment caused cotinine concentrations to decrease, moderated by baseline worm burden scores (fig. 2 bottom). This supports the idea that nicotine exposure is regulated in response to helminth infection, although other causal mechanisms cannot be ruled out.

4.5. Cotinine vs. reinfection

As predicted, there were significant negative Spearman rank correlations between reinfection scores and both year 1 cotinine concentration (r=-0.48, p=0.012) and year 2 cotinine concentration (r=-0.48, p=0.012). Due to the small sample size, multivariate analysis with control variables was not feasible. These results suggest that nicotine protects against reinfection. See figure 3.

Figure 3 about here

4.6. CYP2A6 alleles vs. cotinine concentrations

Individuals with slow-metabolizing CYP2A6 alleles should have higher cotinine concentrations than individuals with normal or fast metabolizing alleles. CYP2A6 allele frequencies from both this study and (to increase sample size) a previously unpublished study

73 of the same population are reported in table S3. The analysis of baseline cotinine concentration of recent smokers vs. predicted metabolic phenotype included 2 females and 13 males from the previous study, and 39 males from the current study. Although a trend of higher cotinine concentrations in slower phenotypes is apparent in figure S2, an ANOVA of square-root transformed cotinine concentration with a first-order polynomial contrast found it was not significant (F(1, 52)=1.9, p=0.18), perhaps due to the small sample size.

4.7. CYP2A6 polymorphisms vs. worm burden

Regular smokers with slow-metabolizing CYP2A6 alleles should have higher nicotine exposure and thus lower worm burden than those with normal-metabolizing alleles. The predicted metabolic phenotype (Normal/Slow) of participants in this study was added to the

GAM of baseline worm burden as a function of age and baseline cotinine concentration analyzed above (fig. 1). There was no significant main effect of phenotype, but there was a significant curvilinear interaction with age, such that slow metabolizers, relative to normal metabolizers, had a lower worm burden at older ages, and perhaps also at younger ages (fig. 4; see table S4 for model parameters). (The relationship between baseline cotinine concentration and worm burden in this model was very similar to that in fig. 1).

Figure 4 about here

5. Discussion and limitations

5.1. Helminth infection patterns

Aka suffer very high rates of helminth infection, with 99% of the men in our sample infected with at least one species. Rates of helminth infection were similar to those found in another Aka population working at a national park about 230 km to the southwest of the study field site, where 93.1% were infected with hookworms or nodular worms, 75.9% with whipworms, and 10.3% with Ascaris. Interestingly, those Aka had higher worm burden than local farmers and sympatric gorillas and chimpanzees (Lilly et al., 2002, cf. Froment, 2014).

5.2. Smoking patterns

The prevalence of male smoking in the Congo basin is 10-14%, and in sub-Saharan

Africa ranges from 10% to 27% (Pampel 2008). The highest national male smoking prevalences are in the range of 60-70% in e.g., Russia, Indonesia, and some Pacific islands. By any measure, the 94% prevalence of male smoking among the Aka is extraordinarily high.

Moreover, Aka men report earning the equivalent of USD 0.50 a day working for the farmers (a day’s work being about 4-5 hours of physical labor), and spending half of that, about USD

0.25/day, on tobacco (unpublished data, Roulette et al.). Smoking is obviously very important to

Aka men.

Cotinine values confirm a low prevalence of tobacco use in Aka women (5% were recent smokers), consistent with rates seen in women in other developing countries (Guindon and

Boisclaire 2003). Hence, the sex difference in Aka smoking is also dramatic, with an odds ratio of 94.5 based on self-report and 311.7 based on salivary cotinine concentration.

The pharmacophagy hypothesis received support from all studies, with some limitations.

None of the four studies included a behavioral measure of smoking. Thus, differences in

75 cotinine concentration in these studies, including the treatment effect in the RCT, could be a consequence of differences in nicotine intake (e.g., differences in smoking) and/or differences in clearance of nicotine and cotinine.

5.3. Observational studies of smoking vs. worm burden

The three observational studies all found that higher nicotine exposure was correlated with lower worm burden. The cross-sectional relationship between nicotine and worm burden

(figure 1B) appears similar to a dose-response curve, with worm burden beginning to drop for cotinine concentrations exceeding about 200 ng/ml (roughly the level found in 1 pack/day smokers), controlling for age, region, wealth, and acculturation, which supports the chemotherapy hypothesis.

Individuals with slow metabolizing CYP2A6 alleles had marginally significantly higher cotinine values (figure S2) and significantly lower worm burden scores than those with normal or fast metabolizing alleles, the latter result holding only at younger and older ages (figure 4). The marginal significance of the metabolic phenotype effect on cotinine concentration is not surprising, given the small sample size and limited number of saliva samples. Over years of smoking, the cumulative effect of increased exposure to nicotine could be substantial.

Regarding the curvilinear interaction with age: in populations with endemic worm infections, worm burden usually peaks during middle childhood or adolescence (Anderson & May, 1985), and again in old age (see, e.g., fig. 1A). One interpretation, therefore, is that male tobacco use is relatively constant across metabolic phenotypes, leading to higher lifetime levels of nicotine exposure in slow metabolizers. This higher lifetime exposure protects against the increase in worm burden that typically occurs in younger and older men.

Finally, in a sample treated with albendazole in year 1, individuals with higher cotinine concentrations in year 1 and/or year 2 had lower reinfection by year 2, supporting the chemoprophylaxis hypothesis.

5.3.1 Limitations of observational studies

It is possible that differences in worm burden caused differences in cotinine concentration, a direction of causation contrary to the hypotheses we tested here, or that cotinine concentration and/or worm burden were influenced by confounding factors that we did not control for, including the consumption of dietary or medicinal plants that contain nicotine. It is unlikely that plant foods would contribute much to the overall cotinine concentrations, however, because the nicotine concentrations of dietary plants are very low (Domino et al. 1993).

The effects of grapefruit juice, other citrus juices, and a variety of plant compounds on drug metabolism are well documented. It is possible that some Aka plant foods or traditional medicines alter nicotine metabolism, leading to higher or lower cotinine concentrations. It is also possible some dietary foods prevent and/or treat infection (cf. Etkin 1986; Etkin and Ross 1982;

Jackson 1990, 1996; Moerman 1996), altering worm burden values.

Both the reinfection and CYP2A6 allele studies had small sample sizes, which limited our ability to control for confounds with multivariate statistical models. In addition, CYP2A6 metabolizes other xenobiotic substrates, such as coumarin. It is possible that some other

CYP2A6 xenobiotic substrate in the Aka diet is protective against helminths. Further, CYP2A6 plays some role in estrogen metabolism, is induced by estrogen via the estrogen receptor, and is inhibited by other steroids and neurotransmitters (Higashi et al., 2007). As there is bidirectional interaction between gonadal steroids and the immune system (Grossman 1985),

77 CYP2A6 polymorphisms could, in principle, directly influence immunity to helminths via their effect on estrogen metabolism (although this effect appears small), or another physiological pathway. Nevertheless, because slow metabolizing alleles of other P450 enzymes occur at relatively high frequencies in some populations, it is worth considering that these alleles provided protection against parasites by slowing the clearance of dietary toxins.

5.4. Randomized controlled trial

Treating Aka men for helminthiasis with albendazole reduced cotinine concentrations 2 weeks later, compared to placebo controls (controlling for baseline cotinine concentration and worm burden scores). This causal effect was moderated by baseline worm burden score, with a larger reduction in cotinine concentration seen in men who, at baseline, had higher worm burden scores (figure 2 and table 2). Other aspects of the model in table 2 and figure 2 largely support the pharmacophagy hypothesis, with some caveats. In the sample as a whole, cotinine concentration decreased from baseline to follow-up (coef = 0.78). Although unexpected, the decrease is not too surprising as there are numerous factors that affect access to tobacco in this population. For instance, 40 men reported spending time hunting in the forest during the two- week period between collection of baseline and follow-up samples, which would decrease access to tobacco.

Despite this overall decrease in cotinine concentration, in the control group there was a positive relationship between baseline worm burden and follow-up cotinine (coef = 0.03; fig 2B), whereas this coefficient was predicted to be 0 (i.e., in the control group, follow-up cotinine should not depend on baseline worm burden; fig. 2A). However, this coefficient, although positive, was not significantly greater than 0 (p = 0.13).

Another prediction was that because treatment would have caused little change for individuals with low baseline worm burden, the follow-up cotinine concentration of such individuals in the treatment group should have been similar to comparable individuals in the control group. Instead, individuals in the treatment group with low worm burden had slightly higher follow-up cotinine than individuals in the control group (left hand side of the panels in fig.

2B). Again, however, this difference was not significant.

5.4.1 Limitations of the RCT

The RCT design does not elucidate the underlying causal mechanism. The treatment effect could be due to mechanisms unrelated to self-medication, or to population specific mechanisms, contrary to the hypothesis tested here. For instance, treatment improved health, which could have encouraged some men to spend more time in the forest, reducing access to tobacco (but our data rule out this specific alternative explanation: there were no significant differences in time spent in the forest in the two arms of the RCT). Alternatively, albendazole slightly induces CYP2A6 in the rat (Souhaili-El Amri et al., 1988), so there is a possibility that albendazole itself influences nicotine or cotinine metabolism. It is unlikely that such P450 induction explains our treatment effect, however, because participants received a single 400 mg dose of albendazole; albendazole and its active metabolite have a short half-life of about 12 hours whereas follow-up cotinine was measured about two weeks after receiving albendazole; and the treatment effect was moderated by baseline worm burden, which is not clearly consistent with an induction-mediated pathway. Finally, cannabis is known to contain anthelmentic compounds that are potentially toxic to nematodes. In the RCT study, treatment could have affected cannabis smoking or THC metabolism. However we did not quantify cannabis use and THC, which is a limitation of our study.

79 5.7. Candidate self-medication mechanisms

An evolved mechanism to self-medicate with psychoactive substances should up- and down-regulate consumption and elimination of such substances in response to infection and/or infection risk, either by increasing consumption of a plant substance and/or by physiologically down-regulating drug metabolism. Intriguingly, infection and inflammation are associated with a broad down-regulation of xenobiotic-metabolizing enzymes and transporters in humans and laboratory animals (albeit with complications for CYP2A6), which often results in a pronounced increase in plasma concentrations of various drugs. This well-documented but poorly understood phenomenon (Morgan et al., 2008) could be evidence for a chemotherapy mechanism.

The “proinflammatory hypothesis of drug abuse” has emerged from growing evidence of immune involvement in drug reinforcement. Bidirectional neural-immune interactions are also well established (Wrona 2006). More specifically, opioids, perhaps acting as xenobiotic- associated molecular patterns, activate toll-like receptor 4 (TLR4) signaling, which surprisingly reinforces opium consumption via the mesolimbic dopamine reward pathway (Hutchinson et al.,

2012). Especially intriguing is direct evidence that the immune system modulates intake of the psychoactive drug ethanol (Blednov et al., 2011; Blednov et al., 2012). One genome-wide association study found that smoking behavior might be regulated by IL-15 (Liu et al., 2009).

Such results indicate an intimate relationship between psychoactive drug use and immunity, and importantly that central immune signals can mediate drug consumption. (The prevailing view, however, is that drug abuse increases infection risk by impairing immunity [Friedman et al.,

2003].)

Despite limitations and uncertainty about mechanisms, if tobacco use in this population is motivated, in part, by therapeutic benefits, it suggests that signaling pathways exist between the immune and nervous systems that might be exploited to reduce smoking. The project also provides a novel model system to test evolutionary theories of self-medication by humans and other primates. Lastly, these results, to our knowledge, are the first to suggest an important relationship between smoking and helminthiasis, two of the world’s most pressing health problems.

Acknowledgments

We thank Mesmin Dopeningue and Niquez Molende for assistance in collecting samples and translating, Lester Cordes for medical advice, and the Aka for participating in our study.

Competing Interests: The authors declare that they have no competing financial interests.

References

Abad, Maria José, Maria Ansuategui, and Paulina Bermejo (2007). Active antifungal substances

from natural sources. ARKIVOC vii:116-145.

Anderson, R. M., and May, R. M. (1985). Herd immunity to helminth infection and implications

for parasite control. Nature 315, 493–496.

Bahuchet, Serge (1992). Dans la Forêt D’Afrique Centrale les Pygmées Aka et Baka. Peeters-

Selaf, Paris.

Bahuchet, Serge (1993). History of the Inhabitants of the Central African Rain Forest:

Perspectives from Comparative Linguistics. In Tropical Forests People and Food:

Biocultural Interactions and Applications to Development, eds. C. M. Hladik, A. Hladik,

O.F. Linares, H. Pagezy, A. Semple and M. Hadley, pp. 37-54. Parthenon, Paris.

Baldwin, I. T. (1999). Inducible nicotine production in native Nicotiana as an example of

adaptive phenotypic plasticity. Journal of Chemical Ecology 25, 3–30.

Barbosa, P., Gross, P., & Kemper, J. (1991). Influence of Plant Allelochemicals on the Tobacco

Hornworm and its Parasitoid, Cotesia Congregata. Ecology, 72(5), 1567–1575.

Benowitz, N. L. L. (1996). Cotinine as a Biomarker of Environmental Tobacco Smoke Exposure.

Epidemiologic Reviews 18, 188–204.

Benowitz, N. L., Bernert, J. T., Caraballo, R. S., Holiday, D. B., and Wang, J. (2009). Optimal

serum cotinine levels for distinguishing cigarette smokers and nonsmokers within

different racial/ethnic groups in the United States between 1999 and 2004. American

Journal of Epidemiology 169, 236.

Betti, Jean L. (2004). An Ethnobotanical Study of Medicinal Plants Among the Baka Pygmies in

the Dja Biosphere Reserve, Cameroon. African Study Monographs 25(1):1-27.

Billing, J., & Sherman, P. W. (1998). Antimicrobial functions of spices: why some like it hot.

Quarterly Review of Biology, 3-49.

Black, P. W. (1984). The Anthropology of Tobacco Use: Tobian Data and Theoretical Issues.

Journal of Anthropological Research, 40(4), 475–503.

Blagg, W., Schloegel, E. L., Mansour, N. S., and Khalaf, G. I. (1955). A new concentration

technic for the demonstration of protozoa and helminth eggs in feces. The American

Journal of Tropical Medicine and Hygiene 4, 23–28.

Blednov, Y. A., Benavidez, J. M., Geil, C., Perra, S., Morikawa, H., and Harris, R. A. (2011).

Activation of inflammatory signaling by lipopolysaccharide produces a prolonged

increase of voluntary alcohol intake in mice. Brain, Behavior, and Immunity 25, S92–

S105.

83 Blednov, Y. A., Ponomarev, I., Geil, C., Bergeson, S., Koob, G. F., and Harris, R. A. (2012).

Neuroimmune regulation of alcohol consumption: behavioral validation of genes

obtained from genomic studies. Addiction Biology 17, 108–120.

Boppré, M. (1984). Redefining “Pharmacophagy.” Journal of Chemical Ecology 10, 1151–1154.

Damon, A. (1973). Smoking attitudes and practices in seven preliterate societies. In W. Dunn

(Ed.), Smoking behavior: motives and incentives (pp. 219–230). Winston and Sons.

Degenhardt L, Chiu WT, Sampson N, Kessler RC, Anthony JC, Angermeyer M, et al. (2008).

Toward a global view of alcohol, tobacco, cannabis, and cocaine use: _findings from the

WHO World Mental Health Surveys. PLoS medicine 5 (2008) e141.

Domino, E. F., Hornbach, E., & Demana, T. (1993). The nicotine content of common

vegetables. New England Journal of Medicine, 329(6), 437-437.

Du Toit, BM (1975). Dagga: The history and ethnographic setting of Cannabis Sativa in

southern African. Cannabis and Culture. V. Rubin, ed. (The Hague: Mouton) 81-116

Echeverría, J., & Niemeyer, H. M. (2013). Nicotine in the hair of mummies from San Pedro de

Atacama (Northern Chile). Journal of Archaeological Science, 40(10), 3561–3568.

doi:10.1016/j.jas.2013.04.030

Eriksen, Michael, Judith Mackay, and Hana Ross (2012). The Tobacco Atlas, 4th Edition.

American Cancer Society, Atlanta, GA.

Etkin, N. L. (Ed.) (1986). Plants in Indigenous Medicine and Diet: Biobehavioral Approaches

(Vol. 1). Psychology Press.

Etkin, N. L. and Ross, P. J. (1982). Food as Medicine and Medicine as Food: An Adaptive

Framework for the Interpretation of Plant Utilization among the Hausa of Northern

Nigeria. Social Science and Medicine 16:1559-1573.

Etkin, N. L. and Ross, P. J. (1991). Recasting Malaria, medicine and meals: A Perspective on

Disease Adaptation. In L Romanucci-Ross, D. E. Moerman, and L. R. Tancredi (Ed.’s),

The anthropology of medicine: From culture to method (Second Edition), (pp. 230-259).

New York: Bergin and Garvey.

Friedman, H., Newton, C., and Klein, T. W. (2003). Microbial Infections, Immunomodulation, and

Drugs of Abuse. Clinical Microbiology Reviews 16, 209–219.

Froment, A. (2014). Human biology and the health of African rainforest inhabitants. In B. S.

Hewlett (Ed.), Hunter-gatherers of the Congo basin (pp. 117-164). New Brunswick:

Transaction Publishers.

Glander, K. E. (1994). Nonhuman Primate Self-Medication with Wild Plant Foods. In Eating on

the Wild Side: The Pharmacologic, Ecologic, and Social Implications of Using

Noncultigens. N. Etkin (ed.). pp 227-239. University of Arizona Press: Tucson.

Grossman, C. J. (1985). Interactions Between the Gonadal Steroids and the Immune System.

Science 227, 257–261.

85 Gosselin, R. E., Smith, R. P., and Hodge, H. C. (1984). Clinical Toxicology of Commercial

Products 5 ed. (Baltimore, MD: Williams and Wilkin).

Grotenhermen, F. (2003). Pharmacokinetics and Pharmacodynamics of Cannabinoids. Clin

Pharmacokinet, 42(4), 327-360.

Guindon, G. E., and Boisclair, D. (2003). Past, Current and Future Trends in Tobacco Use.

Washington, DC: World Bank. Health, Nutrition and Population (HNP) Discussion

Paper. Economics of Tobacco Control Discussion Papers, (6).

Hagen, E. H., Roulette, C.J., Sullivan, R.J. (2013). Explaining human recreational use of

‘pesticides’: The neurotoxin regulation model of substance use vs. the hijack model and

implications for age and sex differences in drug consumption. Front. Psychiatry 4:142.

doi: 10.3389/fpsyt.2013.00142

Hagen, E. H., Sullivan, R. J., Schmidt, R., Morris, G., Kempter, R., and P., H. (2009). Ecology

and neurobiology of toxin avoidance and the paradox of drug reward. Neuroscience 160,

69–84.

Hammond, J. A., Fielding, D., and Bishop, S. C. (1997). Prospects for plant anthelmintics in

tropical veterinary medicine. Veterinary Research Communications 21, 213–228.

Hattori, Shiho (2010). “My Medicine (Ma a le)”: Variability of Medicinal Plant Knowledge among

Adult Baka Hunter-Gatherers of Southeast Cameroon. Paper presented at the Workshop

for Recent Research among Congo Basin Hunter-Gatherers and Farmers. University of

New Mexico, Albuquerque, NM.

Hewlett, Barry S. (1977). Notes on the Mbuti and Aka Pygmies of Central Africa. Master’s thesis

(Anthropology), California State University Chico, Chico, CA.

Hewlett, B. S. (1996). Cultural diversity among African pygmies. In Cultural Diveristy among

Twentieth-Century Foragers: An African Perspective, S. Kent, ed. (Cambridge University

Press), pp. 215–244.

Higashi, E., Nakajima, M., Katoh, M., Tokudome, S., and Yokoi, T. (2007). Inhibitory Effects of

Neurotransmitters and Steroids on Human CYP2A6. Drug Metabolism and Disposition

35, 508–514.

Horton, J. (2000). Albendazole: a review of anthelmintic efficacy and safety in humans.

Parasitology 121, 113–132.

Hotez, P. J., Brindley, P. J., Bethony, J. M., King, C. H., Pearce, E. J., and Jacobson, J. (2008).

Helminth infections: the great neglected tropical diseases. The Journal of Clinical

Investigation 118, 1311–1321.

Huffman, M. A. (1997). Current evidence for self-medication in primates: A multidisciplinary

perspective. Yearbook of Physical Anthropology 104, 171–200.

Huffman, M. A. (2001). Self-Medicative Behavior in the African Great Apes: An Evolutionary

Perspective into the Origins of Human Traditional Medicine. BioScience, 51, 8, 651-661.

87 Hukkanen, J., Jacob, P., and Benowitz, N. L. (2005). Metabolism and Disposition Kinetics of

Nicotine. Pharmacological Reviews 57, 79–115.

Hutchinson, M. R., Northcutt, A. L., Hiranita, T., Wang, X., Lewis, S. S., Thomas, J., Steeg, K.

van, Kopajtic, T. A., Loram, L. C., Sfregola, C., et al. (2012). Opioid Activation of Toll-

Like Receptor 4 Contributes to Drug Reinforcement. The Journal of Neuroscience 32,

11187–11200.

Jackson, F. L. C. (1990). Two evolutionary models for the interactions of dietary organic

cyanogens, hemoglobins , and falciparum malaria. American Journal Human Biology, 2,

521-532.

Jackson, F. L. C. (1996). The coevolutionary relationship of humans and domesticated plants.

Yearbook of Physical Anthropology, 39, 161-176.

Jackson, J. A., Friberg, I. M., Little, S., & Bradley, J. E. (2009). Review series on helminths,

immune modulation and the hygiene hypothesis: immunity against helminths and

immunological phenomena in modern human populations: coevolutionary legacies?

Immunology, 126, 18-27.

Jarvis, M. J., Fidler, J., Mindell, J., Feyerabend, C., and West, R. (2008). Assessing smoking

status in children, adolescents and adults: cotinine cut-points revisited. Addiction 103,

1553–1561.

Jeffreys, Mervyn David Waldgrave (1963) Smoking in Africa. The South African Archaeological

Bulletin 18(69):33.

Kaplan, Mara, Laura Carriker and Ingrid Waldron (1990). Gender differences in tobacco use in

Kenya. Social Science and Medicine 30:305-10.

Katz, S. H. and J. I. Schall (1986). Favism and Malaria: A model of Nutrition and Biocultural

Evolution. In N. L. Etkin (Ed.), Plants in Indigenous Medicine and Diet (Volume 1) (pp.

211-238). Psychology Press.

Kohler, P. (2001). The biochemical basis of anthelmintic action and resistance. International

Journal for Parasitology 31, 336–345.

Komiya, Y., and Kobayashi, A. (1966). Evaluation of Kato’s thick smear technic with a

cellophane cover for helminth eggs in feces. Japanese Journal of Medical Science &

Biology 19, 59–64.

Kumar, P., Pandit, S. S., Steppuhn, A. and Baldwin, I. T. (2013). Natural history-driven, plant-

mediated study reveals CYP6B46’s role in a nicotine-mediated antipredator herbivore

defense.

Lamidi, Marow, Carole DiGiorgio, Florence Delmas, A. Favel, C. Eyele Mve-Mba, M.L. Rondi, E.

Ollivier, L. Nze-Ekekang, and G. Balansard (2005). In vitro cytotoxic, antileishmanial and

antifungal activities of ethnopharmacologically selected Gabonese plants. Journal of

Ethnopharmacology 102:185-190.

Lewis, D. F. V. (2001). Guide to cytochrome P450: structure and function. London, UK: Taylor

and Francis.

Lilly, A. A., Mehlman, P. T., and Doran, D. (2002). Intestinal Parasites in Gorillas, Chimpanzees,

and Humans at Mondika Research Site, Dzanga-Ndoki National Park, Central African

Republic. International Journal of Primatology 23, 555–573.

Liu, Y. Z., Pei, Y. F., Guo, Y. F., Wang, L., Liu, X. G., Yan, H., Xiong, D. H., Zhang, Y. P., Levy,

S., Li, J., et al. (2009). Genome-wide association analyses suggested a novel

mechanism for smoking behavior regulated by IL15. Molecular Psychiatry 14, 668–680.

Mackay, Judith and Michael Eriksen (2002). The Tobacco Atlas. World Health Organization,

Geneva, Switzerland.

Mathers, C. D., and Loncar, D. (2006). Projections of Global Mortality and Burden of Disease

from 2002 to 2030. PLoS Med 3, e442.

Mayer, Bernd (2014). How much nicotine kills a human? Tracing back the generally accepted

lethal dose to dubious self-experiments in the nineteenth century. Archives of

Toxicology, 88(1):5-7.

Moerman, D. E. (1996). An analysis of the food plants and drug plants of native North America.

Journal of Ethnopharmacology, 52(1), 1–22. doi:10.1016/0378-8741(96)01393-1

Morgan, E. T., Goralski, K. B., Piquette-Miller, M., Renton, K. W., Robertson, G. R., Chaluvadi,

M. R., Charles, K. A., Clarke, S. J., Kacevska, M., Liddle, C., et al. (2008). Regulation of

drug-metabolizing enzymes and transporters in infection, inflammation, and cancer.

Drug Metabolism and Disposition 36, 205–216. Available at:

http://dmd.aspetjournals.org/content/36/2/205.abstract

Mukhtar, T., Kayani, M. Z., and Hussain, M. A. (2013). Nematicidal activities of Cannabis sativa

L. and Zanthoxylum alatum Roxb. against Meloidogyne incognita. Industrial Crops and

Products 42, 447–453.

Mwenifumbo, J. C., and Tyndale, R. F. (2009). Molecular genetics of nicotine metabolism. In

Handbook of Experimental Pharmacology, J. E. Henningfield, E. D. London, and S.

Pogun, eds. (Springer), pp. 235–259.

Nakajima, M., Fukami, T., Yamanaka, H., Higashi, E., Sakai, H., Yoshida, R., Kwon, J. T.,

McLeod, H. L., and Yokoi, T. (2006). Comprehensive evaluation of variability in nicotine

metabolism and CYP2A6 polymorphic alleles in four ethnic populations. Clinical

Pharmacology & Therapeutics 80, 282–297.

Nelson, D.R. (1999). Cytochrome P450 and the individuality of species. Arch. Biochem.

Biophys. 369, 1-10.

Nelson, D. R., Zeldin, D. C., Hoffman, S. M. G., Maltais, L. J., Wain, H. M., & Nebert, D. W.

(2004). Comparison of cytochrome P459 (CYP) genes from the mouse and human

genomes, including nomenclature recommendations for genes, pseudogenes and

alternative-splice variants. Pharmacogenetics 14, 1-18.

Nelson, D. R. et al. (2008). Cytochrome P450 Homepage,

http://drnelson.utmem.edu/CytochromeP450.html

Neuwinger, Hans Dieter (1996). African Ethnobotany: Poisons and Drugs. Chapman & Hall

GmbH, Weinheim (Bundesrepublik Deutschland).

Nyasse, B., I. Ngantchou, J.-J. Nono, and B. Schneider (2006). Antifilarial activity in vitro of

polycarpol and 3-O-acetylaleuritolic acid from Cameroonian medicinal plants against

Onchocerca gutturosa. Natural Product Research 29(4):391-397.

Oishi, T. and Hayashi, K. (2014). From Ritual Dance to Disco: Change in habitual use of

tobacco and alcohol among the Baka Hunter-Gatherers of Southeastern Cameroon.

African Study Monographs, 47, 143-163.

Okorie, Dominic A. (1980). A new class of indolosesquiterpene alkaloids from Polyalthia

suaveolens. Tetrahedron 36:2005-2008.

Pampel, F. (2008). Tobacco use in sub-Sahara Africa: estimates from the demographic health

surveys. Social Science & Medicine 66, 1772–1783.

Pickersgill, B. (2007). Domestication of Plants in the Americas: Insights from Mendelian and

Molecular Genetics. Annals of Botany, 100(5), 925–940. doi:10.1093/aob/mcm193

Rodriguez, E. and Cavin, C. (1982) The Possible Role of Amazonian Psychoactive Plants in the

Chemotherapy of Parasitic Worms—A Hypothesis. Journal of Ethnopharmacology 6,

303-309.

Rodriguez, E., and Wrangham, R. (1993). Zoopharmocognosy: The Use of Medicinal Plants by

Animals. Recent Advances In Phytochemistry 27, 89–89.

Sherman, P. W., & Billing, J. (1999). Darwinian Gastronomy: Why We Use Spices Spices taste

good because they are good for us. BioScience, 49(6), 453-463.

Singer, M. S., Mace, K. C., & Bernays, E. A. (2009). Self-Medication as Adaptive Plasticity:

Increased Ingestion of Plant Toxins by Parasitized Caterpillars. PLoS ONE, 4(3), e4796.

doi:10.1371/journal.pone.0004796

Souhaili-El Amri, H. S. E., Fargetton, X., Benoit, E., Totis, M., and Batt, A. M. (1988). Inducing

effect of albendazole on rat liver drug-metabolizing enzymes and metabolite

pharmacokinetics. Toxicology and applied pharmacology 92, 141–149.

Suárez-Rodríguez, M., López-Rull, I., & Garcia, C. M. (2013). Incorporation of cigarette butts

into nests reduces nest ectoparasite load in urban birds: new ingredients for an old

recipe? Biology Letters, 9(1). doi:10.1098/rsbl.2012.0931

Sullivan, R. J., Hagen, E. H., and Hammerstein, P. (2008). Revealing the paradox of drug

reward in human evolution. Proceedings of the Royal Society B 275, 1231–1241.

Thorpe, K. W., & Barbosa, P. (1986). Effects of comsumption of high and low nicotine tobacco

by Manduca sexta (Lepidoptera: Sphingidae) on survival of gregarious endoparasitoid

Cotesia congregata (Hymenoptera: Braconidae). Journal of Chemical Ecology, 12(6),

1329–1337.

93 Tushingham, S., Ardura, D., Eerkens, J. W., Palazoglu, M., Shahbaz, S., & Fiehn, O. (2013).

Hunter-gatherer tobacco smoking: earliest evidence from the Pacific Northwest Coast of

North America. Journal of Archaeological Science, 40(2), 1397–1407.

doi:10.1016/j.jas.2012.09.019

Villalba, J. J., Provenza, F. D., Hall, J. O., & Lisonbee, L. D. (2010). Selection of tannins by

sheep in response to gastrointestinal nematode infection. Journal of Animal Science,

88(6), 2189-2198.

Waldron, Ingrid, Gary Bratelli, Laura Carriker, Wei-Chin Sung, Christine Vogeli and Elizabeth

Waldman (1988). Gender differences in tobacco use in Africa, Asia, the Pacific, and

Latin America. Social Science and Medicine 27:1,269-75.

Wood, S. (2006). Generalized additive models: an introduction with R (Chapman & Hall/CRC).

Wrangham, R. W., & Nishida, T. (1983). Aspilia spp. leaves: a puzzle in the feeding behavior of

wild chimpanzees. Primates, 24(2), 276–282.

Wrona, D. (2006). Neural-immune interactions: an integrative view of the bidirectional

relationship between the brain and immune systems. Journal of Neuroimmunology 172,

38–58.

Wyatt, G.B. (1977). Health in Melanesia. In J. H. Winslow (Ed.), The Melanesian Environment

(pp 459-462). Canberra, Australia: National University Press.

Yoo, H. D., et al. (2005) Suaveolindole, a New Mass-Limited Antibacterial Indolosesquiterpene

from Greenwayodendron suaveolens Obtained via High-Throughput Natural Products

Chemistry Methods. Journal of Natural Products, 68, 1, 122-124.

Males (206) N Range Median Mean SD

Age 206 18–70 35 36.3 9.8

Cotinine (ng/ml) 199 0.8–715.9 120.2 148.5 139.5

Worm burden score 199 0–95.3 16.3 23.1 19.3

Material wealth score 191 1–8 2 2 1.26

Acculturation score 191 0–4 2 2.2 0.85

Females (44)

Age 44 20–55 30 32.2 9.4

Cotinine (ng/ml) 42 0.1–332 0.9 10 51.3

Worm burden score 42 0–128 19.5 29.3 28.5

Table 1: Baseline sample characteristics based on interview data and samples s1-s3.

Estimate Std. Error t value Pr(t)

(Intercept) 2.300 0.697 3.298 0.001

Baseline cotinine 0.779 0.051 15.134 0.000

Baseline worm 0.035 0.023 1.528 0.129

Intervention(treatment) -0.091 0.619 -0.147 0.884

Baseline worm:intervention(treatment) -0.069 0.034 -2.014 0.046

Table 2: ANCOVA model of follow-up cotinine concentration. Baseline worm burden is centered at the mean.

Figure 1: Plots of male worm burden as smooth functions of age (A) and cotinine concentration

(B), controlling for linear functions of village, acculturation, and wealth). Lines fit by a negative binomial GAM with log-link function (the y-axis is thus in units of log worm burden score). For model parameters and p-values, see table S1. Dashed lines represent +/- 2 SE.

Figure 2: A. Predicted effect of the intervention * baseline worm burden interaction on follow-up cotinine concentration. B. Study results. Baseline worm burden is centered at the mean. Plotted at mean . Dashed lines represent +/- SE.

Figure 3. Scatterplot of reinfection scores vs. year 1 cotinine concentration. Reinfection score = mean year 2 worm burden (s7-s9) minus mean year 1 worm burden (s4-s6). Line fit by linear regression. Shading represents +/- SE.

100

Figure 4: The effect of CYP2A6 polymorphisms on worm burden. Plots of male worm burden as a smooth function of age, controlling for cotinine concentration. Top: Normal metabolizers.

Bottom: Slow metabolizers. Lines fit by a negative binomial GAM with log-link function; y-axis in log units; dotted lines represent +/- 2 SE. For model parameters and p-values, see table S4.

101

Figure 5: Saliva and stool sample collection schedule and male participant attrition. `I1' = initial interview. Wealth and acculturation data were collected during a second interview (I2) upon receipt of sample s3. There is an increase in s2 relative to s1 because some men joined the study after collection of s1.

102

A B 4 2 Worm burden Worm 0 2 −

20 40 60 0 200 400 600

Age (years) Cotinine (ng/ml)

103

104

105

106

107 CHAPTER FOUR

HIGH PREVALENCE OF CANNABIS USE AMONG CONGO BASIN FORAGERS,

AND ITS POSSIBLE RELATIONSHIP TO HELMINTHIASIS

Casey J. Roulette1, Mirdad Kazanji2, Sébastien Breurec2, Edward H. Hagen1*

1Washington State University, Department of Anthropology, Vancouver, WA, 98686; 2Institute Pasteur, Bangui, CAR

*Corresponding author: EHH, 14204 Northeast Salmon Creek Avenue, Vancouver, WA 98686, 1(360)546-9257, [email protected]

Grant information: This investigation was supported in part by funds provided to EHH and CJR for medical and biological research by the State of Washington Initiative Measure No. 171.

108

Abstract Objectives Little is known about cannabis use in hunter-gatherers. We therefore investigated cannabis use in the Aka, a population of Congo Basin foragers. Because cannabis contains anthelminthic compounds, and the Aka have a high prevalence of helminthiasis, we also tested the hypothesis that cannabis use might be an unconscious form of self-medication against helminths. Methods We collected self- and peer-reports of cannabis use from all adult Aka in the Lobaye district of the Central African Republic (n=379). Because female cannabis use was low, we restricted sample collection to men. Using an immunoassay for THCA, a urinary biomarker of recent cannabis consumption, we validated cannabis use in men currently residing in camps near a logging road (n=62). We also collected stool samples to assay worm burden. A longitudinal reinfection study was conducted among a subsample of the male participants (n=23) who had been treated with a commercial anthelmintic one year previously. Results The prevalence of self- and peer-reported cannabis was 70.9% among men and 6.1% among women, for a total prevalence of 38.6%. Using a 50 ng/ml threshold for THCA, 67.7% of men used cannabis. Cannabis users were significantly younger and had less material wealth than non-cannabis users. There were significant negative associations between THCA levels and worm burden, and reinfection with helminths one year after treatment with a commercial anthelmintic. Conclusions The prevalence of cannabis use among adult Aka men was high compared to most global populations. THCA levels were negatively correlated with parasite infection and reinfection, supporting the self-medication hypothesis.

Key words: Recreational psychoactive substance use, marijuana, hunter-gatherers, self-medication, infectious disease, evolutionary medicine

109

Introduction

Cannabis is the world's most widely used illicit drug (Hall and Degenhardt,

2007), with a global prevalence of use estimated at 2.9% - 4.3% of all individuals between 15-64 years (UN, 2010). Research on cannabis use is limited, however, due to the potential legal and social consequences of admitting use. Consequently, little is known about cannabis use in the developing world (e.g. Degenhardt and Hall,

2012). Even less is known about cannabis use among the many small-scale populations scattered throughout remote parts of the developing world, such as in the Congo Basin. Here we investigate patterns of cannabis use in a population of

Congo basin foragers. This is one of the first systematic studies of cannabis use in a hunting-gathering population, and the first, to our knowledge, to validate use with a biomarker. It is also one of the first studies to explore the relationship between substance use and intestinal helminth infection, two of the developing world's great health problems.

Cannabis

Cannabis spp. are indigenous to Asia (Hillig, 2005), where they have been used as fiber, food, oil, medicine, and intoxicant since prehistory. They belong to the small family Cannabaceae, which also includes another socially and economically important species, humulus (hops) (Clark and Merlin, 2013). Cannabis contains the psychoactive cannabinoid, Δ9-tetrahydrocannabinol (THC), as well as hundreds of

110 other chemicals, including over 60 cannabinoids (Turner et al., 1980; Huestis, 2007).

C. sativa is the most widely distributed of the cannabis species today, due in large part to its important economic role as a source of fiber.

Arrival of cannabis in Africa

It is uncertain when cannabis was brought to Africa. Schultes (1970) argues that cannabis arrived in Africa as early as 4000-3000 BP, but du Toit (1980), using linguistic, historical and archaeological evidence, concludes that cannabis arrived via Moslem sea traders from the Indian subcontinent around the 1st century A.D. It then spread west and south via the Arab caravan trade (Du Toit, 1976). Philips

(1983; pg 308), however, argues that Islamic Arabs were not involved and that even if cannabis was in Africa prior to European colonization, it was likely not smoked until the 17th century. Today, the prevalence of cannabis use in Africa is estimated at 5% - 12.5%, with the highest rate in West and Central Africa, although data are very scarce (UN, 2013).

Cannabis use among Congo Basin foragers

Congo Basin foragers, also called "pygmies", are a large and diverse population of hunter-gatherers comprising at least 15 ethnolinguistic groups, of which the most well studied are the Aka, Baka, Efe, and Mbuti (Bahuchet, 2014).

These foragers are characterized by a preference for forest life, polyphonic music, an association with farmer populations, and are generally peaceful and egalitarian with marked gender equality. Most are transitional foragers in that they spend part

111 of the year camped near villages to work in the farmers' fields, and the rest of the year in the forest hunting and gathering. When camped near the villages, they trade labor and forest products for agricultural foods, clothes, salt, axes and knives, and psychoactive substances like alcohol, tobacco, and cannabis (Hewlett, 2014).

Congo Basin foragers smoke a variety of psychoactive substances, including commercial cigarettes, locally grown tobacco, cannabis, and native plants. It is probable that Congo Basin foragers have always known of plants with psychoactive properties, such as the forest plant medeaka smoked by the Mbuti (Schebesta,

1941), which might be a species of Mitragyna (Rätsch, 2005), as well the forest plant motunga (Polyalthia suaveolens, Annonaceae, a.k.a. Greenwayodendron suaveolens

Verdc., Annonaceae; Engl. and Diels) which is smoked by the Aka (Roulette et al.,

2014).

Although the Mbuti say that they have smoked cannabis since the begining of time (Hallet, 1975; pg 409), there is little evidence for forager use prior to European colonization. Ethnographically, elephant hunting specialists have been described as the heaviest users of cannabis (Schebesta, 1933; Hewlett, 1977), which might indicate that it was adopted at the height of the ivory trade, in the 19th century

(Hewlett, 1977).

Linguistic evidence suggests that cannabis spread from eastern Africa west into the interior of the Congo Basin. For example, bangui, the common forager word for cannabis, is related to the popular east African root word, -mbangi; and the west

Congo Basin Baka word, njama (Oishi and Hayashi, 2014), and the east Congo Mbuti

112 word, djemu, are both similar to the word, njemu, which is used by the Nyamwezi, an east African Bantu group (Hewlett, 1977).

Ethnographic descriptions of cannabis use among Congo Basin foragers are limited, and those that exist are often contradictory. Turnbull (1968; 1976), for instance, rarely mentions cannabis use among the Mbuti, whereas Schebesta (1933;

1936) notes its frequent use. Schebesta (1933) mentions that cannabis gives archers the "power" to kill elephants, but that it also makes them "dazed and trembling all over" (Schebesta, 1936; pg 214). In one of the only reviews of cannabis use among

Congo Basin pygmies, Hewlett (1977) notes that cannabis plays important sociocultural and ecological functions and that it is associated with motivational effects. As he states, they "smoke to increase their courage on a hunt, dance better, increase their vital force, or to increase their work capacity when working for

Europeans or village people" (Hewlett, 1977; pg 10). However, Turnbull (1983) notes that cannabis can also be detrimental to hunting (and mating) success. Bailey

(1991) focuses on the negative economic impact of smoking cannabis (and tobacco) among Mbuti foragers, noting that Mbuti smokers are more likely to trade labor and goods for cannabis and tobacco, thus decreasing their overall material wealth.

Despite these few ethnographic descriptions, to our knowledge there are no systematic studies of cannabis use among Congo basin foragers. Thus, the current extent of cannabis use in these populations is unknown.

113 Cannabis versus helminth infection

Human behavioral ecologists have rarely investigated psychoactive substance use (but see Dudley, 2000; Hill and Chow, 2002; Lende and Smith, 2002).

In non-human animals, including many vertebrate, mammalian, and primate species, there is increasing evidence that consumption of toxic plants is a form of self-medication against pathogens (Forbey et al., 2009; Huffman, 1997; Suárez-

Rodríguez et al., 2013; Villalba et al., 2014; Wrangham and Nishida, 1983).

Interestingly, all major human recreational plant drugs -- coffee, tea, tobacco, cannabis, and betel-nut -- are toxic to helminths. Indeed, nicotine, the principal psychoactive compound in tobacco, as well as arecoline, the psychoactive compound in betel nut, have been widely used as commercial anthelmintics (Hammond et al.,

1997). This raises the possibility that human "recreational" substance use might be an unconscious form of self-medication against helminths and other macroparasites

(Hagen et al., 2009; Roulette et al., 2014; Rodriguez and Cavin, 1982; Wyatt, 1977).

The global disease burden of helminth infections is greater than that of malaria and tuburculosis, and is associated with anemia, growth stunting, protein- calorie undernutrition, fatigue, and poor cognitive development (Hotez et al., 2008).

Most Congo Basin foragers are heavily parasitized (Lilly et al., 2002; Froment, 2014;

Roulette et al., 2014), similar to other hunter-gatherer populations (Hurtado et al.,

2008).

Cannabis contains anthelminthic compounds and its use might therefore protect against parasitic infection. C. sativa, for example, has been used as a pest

114 repellent and pesticide as well as a companion crop to help control plant nematodes

(McPartland, 1997). Aqueous extracts of C. sativa are, indeed, toxic to plant nematodes (Mukhtar et al., 2013). Cannabis extracts also have demonstrated activity against human pathogens and parasites. All five major cannabinoids

(cannabidiol, cannabichromene, cannabigerol, THC, and cannabinol) have antibacterial activity against Staphylococcus aureus, a coccal bacterium often found in the human respiratory tract and on the skin (Appendino et al., 2008). Crude extracts of C. sativa leaves also cause paralysis and death of the intestinal trematode

Fasciolopsis buski, and the effect is more lethal than the commercial flukicide,

Oxyclozanide (Roy and Tandon, 1997).

Study aims

This study had two aims. First, using self-reports and a urinary biomarker, we wanted to provide a definitive prevalence of cannabis use for one Congo basin forager population: Aka residing in the Lobaye district of the Central African

Republic (CAR). Our previous research on the prevalence of tobacco use among the

Aka found that 95% of Aka men smoked tobacco (Roulette et al., 2014), compared to

~17% of men in sub-Saharan Africa and 31% of men globally (Ng et al., 2014). This raised the possibility that the prevalence of Aka men's cannabis use might also be high. In contrast, less than 15% of Aka women self-reported smoking tobacco, and only 5% were smokers according to their levels of cotinine, a nicotine metabolite

115 (Roulette et al., 2014). This resembled sex differences in substance use seen in other developing countries (e.g., Ng et al., 2014; Degenhardt et al., 2008).

We also wanted to determine if variation in cannabis use was associated with variation in: wealth, as previously reported (Baily, 1991); acculturation; Aka social roles, such as traditional healers and camp leaders (elephant hunting is now very rare); residential proximity to a large village; and age and sex, as seen in other populations (Degenhardt et al., 2008).

Our second aim was to test the hypothesis that psychoactive drug use, in part, is a form of self-medication against helminths. We therefore investigated the relationship between cannabis use and helminth infections. In our previous study on Aka tobacco use, Roulette et al. (2014) found that, in Aka men, (1) the heaviest tobacco smokers had the lowest worm burden; (2) treatment with albendazole, a commercial anthelmintic, reduced tobacco smoking (or increased nicotine metabolism) compared to placebo controls; (3) after treatment, salivary cotinine levels were negatively correlated with helminth reinfection one year later; and (4) individuals with cytochrome P450 alleles that slowed nicotine metabolism had lower worm burden compared to those with normal metabolizing alleles.

Here we attempt to replicate results (1) and (3) for cannabis (we did not have funding to replicate (2) or (4)). Specifically, our first study investigated the cross-sectional relationship between cannabis use and worm burden. Worm burden should be negatively correlated with THCA concentrations. Our second study was longitudinal and explored cannabis use versus helminth reinfection. Among

116 individuals whose helminth infections have been treated with a commercial anthelmintic in year 1, higher levels of cannabis use should limit reinfection with helminths in year 2.

Materials and methods

This study was approved by the Washington State University Institutional

Review Board (IRB) and the IRB of the Medical School of the University of Bangui,

CAR. Informed consent was obtained from all participants. Consent was oral, and not written, for two reasons. First, the Aka are non-literate. Second, cannabis is illegal in CAR, although in rural villages this law is only lightly enforced. Oral consent helped ensure that privacy would not be breached. Both IRBs approved oral consent, and consent was documented by a local translator. Permission to conduct this research was obtained from the Mistere de l'Eduation Nationale, de l'Enseignement superieur et de la Recherche, CAR.

Study population and participant recruitment

The Aka (also called BaAka, Biaka and Bayaka) are a group of hunter- gatherers (forest foragers) residing in the western Congo Basin. They number about

30,000, but live in small camps (approximately 6-12 adult Aka per camp) scattered throughout southwestern CAR and the northern part of the Republic of the Congo

(Hewlett, 1996).

117 Participants were recruited from three regional subpopulations along the

Lobaye river in CAR with a combined Aka population of 164 adult men and 215 adult women residing in 36 camps. Aka camps are located on trails radiating out from the main logging road into the forest. All participants resided in camps within

~1 km of the main logging road.

We first surveyed cannabis and tobacco use in all men and women in all camps. Not everyone in each camp was present at the time of the survey. According to self and peer reports, 70.9% of males and 6.1% of females used cannabis, for a total prevalence of 38.6%. Due to the low percentage of female cannabis users, we restricted our remaining data collection to men.

We recruited all adult men who were currently residing in camp to participate in the study (N=62). We did not prescreen these participants for cannabis or tobacco use, health complaints, or any other criteria. Most of these men

(54) had been treated with albendazole one year previously.

Specimen collection

Using provided kits, all 62 male participants were asked to provide stool specimens every other day for about six days, for a total of three year-2 worm specimens each, as well as three saliva and urine samples. Stool was collected using

ParaPak vials with formalin. Participants were instructed to provide 2-5 ml of saliva immediately upon waking and before smoking their first cigarette. Urine was collected in sterile collection cups and then pippetted into 5 ml centrifuge tubes.

118 Urine and saliva were stored at -20C in a solar-powered freezer and then shipped to the Bioanthropology Laboratory at Washington State University (WSU) Vancouver for analysis.

A subsample of the albendazole-treated men (n=23) had also provided 1-3 stool samples two weeks after treatment (see Roulette et al., 2014 for details); these men comprised the sample for the longitudinal reinfection study reported here. We refer to these immediate post-treatment stool samples as the "year 1 baseline" samples, s1-s3, and to the new samples collected specifically for this study as "year

2 samples", s4-s6.

Interview

All male participants were interviewed, using a semi-structured questionnaire (Schensul et al., 1999; Bernard, 2006), to assess the sociodemographic characteristics of cannabis use and to control for potential confounds in the cross-sectional and reinfection studies. The surveys produced 14 demographic variables in 6 domains: kombeti, nganga, wealth, acculturation, subpopulation, and age. Kombeti are the head of camps and are the oldest male member of each camp (1=kombeti, 0=not kombeti). Nganga are traditional healers

(1=nganga, 0=not nganga). The 'wealth score' was the sum of self-reported number of clothes, watches, flashlights, radios, and batteries owned. The 'acculturation score' was the sum of self-reported attendance in school as a child and as an adolescent, church attendance, and preference for living in the village vs. the forest

(yes=1, no=0). Subpopulation was either coded as belonging to one of the three

119 regional subpopulations, or as village (the largest subpopulation) versus non-village

(the other two, smaler, subpopulations). Because Aka do not keep track of age, age was estimated by an Ngandu research assistant who had lifelong associations with these Aka populations.

For each trail, we also recruited three adults to peer-rate the frequency of cannabis use of all participants who resided on that trail (0=less than average,

1=more than average). We then summed those 3 ratings to create a peer-rated frequency of use score, which ranged from 0-3.

THCA and cotinine assays

Delta 9-THC-11-oic acid (THCA) is a metabolic product of THC. It is predominantly excreted in urine as the ester-linked glucuronide conjugate (Foltz,

2007). THCA is a superior biomarker to THC because it can be detected for several weeks following a single dose of THC (Agurell et al., 1986; Ellis et al., 1985) and it is more soluble and absorbs less onto the walls of storage containers (Rogers et al.,

1978).

Nicotine, the principal psychoactive alkaloid in tobacco, has a half-life of 2-3 hours whereas cotinine, the principal metabolite of nicotine, has a half-life of approximately 17 hours. Cotinine is thus a superior biomarker of recent nicotine exposure (Benowitz, 1996).

Urine and saliva were assayed in the Bioanthropology Laboratory at WSU

Vancouver using THCA and cotinine enzyme linked immunosorbent assay (ELISA)

120 kits, respectively, according to the manufacturer's (Salimetrics and Immunalysis) protocols. THCA and cotinine concentrations were the mean of s4-s6.

Semi-quantification of worm burden and reinfection score

Stool samples were analyzed by the Institute Pasteur in Bangui, CAR. A total of three species of geohelminths were assessed for worm burden score: Trichuris trichiura (roundworm); Ascaris lumbricoides (giant roundworm); and hookworm

(including both Ancylostoma duodenale and Necator americanus because their eggs cannot be distinguished). Worm burden was estimated per-species using (1) a wet smear that was microscopically examined for helminth eggs; (2) the merthiolate iodine formaldehyde (MIF) concentration technique (Blagg et al., 1955); and (3) the

Kato method (Komiya and Kobayashi, 1966). Worm burden comprised the average number of eggs on each x40 microscope field. There were thus 3 measurements for each of the 3 helminth species (hookworm, whipworm, Ascaris), for a total of 9 measurements per stool sample, which were averaged to produce the sample worm burden score. The MIF technique is the most sensitive, so this count was weighted by a factor of 2. The average was then adjusted for the volume of stool examined. A worm reinfection score was computed by substracting the mean of year-1 baseline worm burden scores (s1-s3) from the mean of year-2 worm burden (s4-s6).

Data analysis

Statistics were computed with R version 3.1.0 (2014-04-10) for Mac. The worm burden score was an average of egg counts, and was overdispersed. We

121 therefore modeled worm burden using negative binomial generalized linear models

(GLMs) and generalized additive models (GAMs) with a log link. Cannabis smoker status was dichotomous and was modeling using logistic regression. Among smokers, we modeled log(THCA) with ordinary least squares (OLS) regression. All models were evaluated using standard diagnostics, such as residual plots used to assess homogeneity and independence, and to identify possible outliers. We used the corrected Akaike information criterion (AICc) for model selection (Burnham and

Anderson 2004). Sample size for the reinfection study was small, limiting our ability to use control variables. THC concentrations were log transformed in some of the figures for visual clarity. Data will be available at https://thedata.harvard.edu/dvn/dv/wsuv-bioanthro-lab.

Results

Of our 62 participants, 42 resided in camps near a large village, and 12 and 8 resided near one of two smaller outlying villages; 18 were kombeti (camp leaders); and 12 were nganga (healers). Descriptive statistics for worm burden, THCA and cotinine concentrations, age, wealth, and acculturation are presented in table 1.

Table 1 about here

THCA vs. self- and peer-reported cannabis use

THCA concentrations were bimodally distributed, with 67.7% (n=42) of scores above the conventional cutoff of 50 ng/ml THCA (SAMHSA 2008) and the rest

122 below the cutoff (Figure 1). In comparison, 60.7% of the sample self-reported as cannabis users. All self-reported cannabis users had THCA concentrations above the

50 ng/ml cutoff, whereas 19 of 24 self-reported non-cannabis users (79.2%) had

THCA concentrations below the 50 ng/ml cutoff (Figure 2). Self-reported non- cannabis users had a median THCA concentration of 8.3 ng/ml, and self-reported cannabis users had a median THCA concentration of 832 ng/ml. Unless stated otherwise, in the remainder of the study cannabis smoker status is defined by THCA concentration >50 ng/ml. THCA values below the cutoff represent either second hand exposure to cannabis smoke or less recent use of cannabis.

Figure 1 about here.

Figure 2 about here

Peer-rated frequency of use ranged from 0 to 3. Non-cannabis smokers were reliably distinguished from cannabis smokers by the raters: only one of the fourteen participants with a peer-rating of 0 had THCA concentrations above 50 ng/ml, whereas only two of the 42 participants with a rating of 1 or greater had THCA concentrations below 50 ng/ml. Among the peer-rated smokers, there was a marginally significant trend for THCA levels to increase with higher peer-ratings of use (Z = 1.29, p = 0.099).

THCA vs. demographic and social role variables

Cannabis smokers (THCA > 50 g/ml) were younger than nonsmokers (mean age = 33.6 vs. 43.1, respectively), and had lower wealth scores (mean wealth = 1.92

123 vs. 2.95, respectively). A logistic regression model of cannabis smoker status vs. the demographic variables (age, wealth, acculturation, village) found that both age and wealth were independent, significant negative predictors (the other demographic variables were not significant predictors alone or in combination with other variables, and were therefore omitted). See table 2 and figure 3.

Table 2 about here

Figure 3 about here

Of the social roles, ngangas were not significantly more or less likely to be cannabis smokers (X2 = 2.14, p=0.14). Kombeti were significantly less likely to be cannabis smokers than other men, but this was due to a confound with age.

"Kombeti" literally means "older brother," and they are typically the oldest male in camp. After controlling for age in a logistic regression model, kombeti status was no longer a signficant predictor of smoker status (test not reported).

We then investigated, among cannabis smokers, whether our demographic variables predicted log THCA levels (using OLS). According to AIC scores, no model with any combination of demographic variables outperformed an intercept-only model (results not reported). In addition, using permutation tests, neither ngnanga

(Z=-1.21, p=0.23) or kombeti status (Z=1, p=0.32) was associated with THCA levels among cannabis smokers. Finally, among cannabis smokers, cotinine levels were positively rank correlated with THCA levels (rs=0.36, p=0.02).

124 Worm burden vs. demographic and social role variables

Of the 62 participants, 3 (4.8%) had no evidence of helminth infection, 20

(32.3%) were infected with at least one species of helminth, 22 (35.5%) were infected with two species of helminths, and 17 (27.4%) were infected will all three species of helminths. By species, 56 (90.3%) participants were infected with hookworm (including both Ancylostoma duodenale and Necator americanus), 31

(50%) were infected with T. trichiura, and 28 (45.2%) were infected with A. lumbricoides.

We investigated worm burden scores vs. our demographic and social role variables using permutation tests, correlation tests, and negative binomial GLMs and GAMs with a log link. There were no convincing associations between worm burden and any of these variables either alone or in combination. Although there was a negative rank correlation between worm burden and age, as seen in other populations, it was only marginally significant (rs = -0.23, p=0.076). In addition, there was a significant, u-shaped relationship between acculturation and worm burden in a GAM, but inspection revealed it was due to high worm burden in 2 individuals with the minimum acculturation score (0), and 3 individuals with the maximum acculturation score (4). For the large majority of participants with intermediate acculturation scores (1-3), the relationship was essentially flat.

125 Worm burden vs. THCA

To test if THCA was a negative predictor of worm burden after controlling for potentially confounding variables, we first computed the same GAM as we did for the tobacco vs. helminths study in Roulette et al. (2014), except here we included

THCA concentration. This model thus fit worm burden in all particpants as a function of village, wealth, acculturation, age, cotinine, and THCA. We found significant negative effects but they were not curvilinear, negating the need to use a

GAM. We therefore fit several GLMs with the negative binomial family and log link, and used AICc for model selection. Model 1 (Table 3) best approximates the GAM model used in Roulette et al. (2014), but model 2 (Table 3, Figure 4), which does not include cotinine, wealth, or acculturation, had the lowest AICc score. Because some data were missing for some variables we used multiple imputation to test whether imputing missing data would make a difference in the coefficients and standard errors, but it did not (results not shown). Among cannabis smokers, peer-reported frequency of use was negatively rank correlated with worm burden, though only marginally significantly (rs = -0.28, p = 0.052).

Table 3 about here

Figure 4 about here

To investigate why cotinine was not a significant negative predictor of worm burden in year 2 data, contrary to results from year 1 data (Roulette et al. 2014), we examined the relationship separately in each subpopulation. Whereas THCA was

126 negatively associated with worm burden in all subpopulations, cotinine was negatively associated in the largest subpopulation and in one smaller subpopulation, but positively in the other small population. Also, our sample size here was about 1/3 that in Roulette et al. (2014).

Reinfection study

The reinfection study involved a subsample of the cross-sectional sample that included all participants with year 1 baseline worm burden scores (s1-s3, n=23). Summary statistics for the reinfection sample did not differ from the total sample, except for mean cotinine, which decreased slightly (301 ng/ml vs 241 ng/ml). Most men had been reinfected with helminths: 60.9% had positive delta worm burden scores (year 2 burden minus year 1 burden), and 39.1% had negative scores.

Delta worm burden was significantly negatively rank correlated with THCA concentrations (rs = -0.43, p = 0.02) (Figure 5). Our ability to control for potential confounds was limited due to the small sample size. Keeping in mind that use of multiple regression with such a small sample is questionable, we centered and scaled all variables so that no intercept would be estimated, saving one degree of freedom. We then fit two multiple regression models of reinfection. The first aimed to test if THCA and cotinine had independent effects on reinfection (table 4, model

1). The second aimed to test if THCA had an effect on reinfection controlling for age, another potential confound (table 4, model 2). THCA was a significant negative

127 predictor in both models, as were cotinine and age. The AICc of model 2 was lower

(55.8) than that of model 1 (60.8).

Figure 5 about here

Table 4 about here

Discussion

This, to our knowledge, is the first biomarker-validated study of cannabis use in a hunting-gathering population, and also the first to explore the relationship between the use of cannabis, which is toxic to helminths, and intestinal helminth infection. We first discuss the patterns and sociodemographic characteristics of Aka cannabis use. We then discuss the results of the cross-sectional and longitudinal worm burden studies.

High rate of cannabis use among Aka men, and the low rate among women

According to prospective population-based cohorts and/or retrospective studies, an estimated 3.9% of the global population aged 15-64 used cannabis in

2011 (i.e. used at least once in the last year), with the highest rate in West and

Central Africa (12.4%) (UN, 2013). In Africa as a whole, 7.5% of the population aged

15-64 years used cannabis in 2011. The highest lifetime prevalence is 42.4% in the

US (Degenhardt et al., 2008).

128 Based on our demographic survey of self- and peer-reported cannabis use in the entire population, 38.6% of all adolescent and adult Aka were current cannabis smokers, and most of these were males: the adolescent and adult male prevalence was 70.9%. Using THCA biomarker data, and a cut-off of 50 ng/ml, 67.7% of men in our study were classified as cannabis smokers. The THCA levels of the cannabis smokers were comparable to, though somewhat higher than, the THCA levels of chronic cannabis smokers in the West (cf. Lowe et al., 2009). The prevalence of cannabis use among Aka men is indeed exceptionally high.

Previously, we found that the prevalence of tobacco use among Aka men is also high: 95% of adult Aka men smoked tobacco (validated with a cotinine assay;

Roulette et al., 2014), whereas the highest nation-level male tobacco smoking prevalence rates are, e.g., Timor-Leste (61.1%) and Indonesia (57%), Russia (51%), and China (45.1%) (Ng et al., 2014). Thus, cannabis and tobacco smoking rates among Aka men exceed the highest nation-level prevalence rates.

Aka adolescent and adult women's cannabis smoking prevalence was 6.1%.

The low rate relative to men is similar to that seen in other developing countries, although Aka women's use might still be high compared to other African women

(Degenhardt et al., 2008). The prevalence of Aka women tobacco smoking was 5%

(Roulette et al., 2014), compared to a median national female prevalence of tobacco smoking in sub-Saharan Africa of 1.95%, and 6.2% globally (Ng et al., 2014).

Regarding tobacco, Aka women explain that, although they can smoke if they want, they avoid it because it harms the fetus. Interestingly, this seemed to be an

129 indigenous cultural model rather than a Western medical model, because the fetal problems included "making the baby black" and "making the fetus cough." For tobacco and cannabis, they also said that smoking is for men, and that they simply did not like it because it makes them sick (Roulette et al. under review). Tobacco is a teratogen, and cannabis is associated with low birth weight and mild developmental abnormalities (Hall and Degenhardt, 2009). Hagen et al. (2013) argue that the large sex differences in substance use, particularly in the developing world, might be a consequence, in part, of women's avoidance of toxic and teratogenic substances during the childbearing years.

There is a notable association between tobacco use and cannabis use

(Agrawal & Lynskey, 2009). In the US population aged 12 and older, for example,

90% of cannabis users reported being a cigarette smoker at some point during their life (compared to 57.9% of cigarette smokers who reported ever using cannabis)

(Agrawal et al., 2012). Among the Aka, there was a positive correlation between cotinine and THCA. However, based on self-reports, very few participants actually preferred to mix tobacco and cannabis together. This contrasts with earlier reports that the Mbuti and Aka often mixed cannabis with tobacco before smoking it

(Hewlett, 1977).

The high prevalence of Aka men's use of cannabis and tobacco is remarkable considering their high cost in this population. Aka men earn about USD 0.50/day. A cannabis or tobacco cigarette costs about USD 0.10, and Aka men report smoking one-to-three cigarettes/day. In many cases, they obtain cigarettes in trade for labor

130 or forest products, or via sharing with other Aka. Nevertheless, about half of Aka

"wages" are used to obtain cannabis or tobacco (Roulette et al. under review). This could explain the negative association we found between cannabis smoking and wealth (table 2 and figure 3). It also accords with a study of Efe foragers in the eastern Congo that found a negative relationship between smoker status and material wealth (Bailey, 1991).

Aka men who smoked cannabis were significantly younger than those who did not, which is in agreement with national and cross-national studies that have found that cannabis use rates peak in young adulthood and then decline as people age, enter the workforce, get married, and/or have children (Bachman et al., 1997;

Degenhardt and Hall, 2012). Alternatively, It is possible the high male prevalence of cannabis use is a relatively recent phenomenon: over thirty years ago, Hewlett

(1977) noted that most Aka men "rarely" smoke cannabis. Thus, the relative youth of Aka cannabis users might be a cohort effect.

The cause of high, almost universal, smoking among Aka men is unclear.

Speculatively, it might be due to a lack of effective social restrictions on smoking cannabis and tobacco, combined with several population-specific benefits.

According to most Congo Basin foragers, smoking increases strength, vital force, courage, and warmth (Hewlett, 1977). It is also possible that villagers, who control access to these substances, encourage Aka dependency on cannabis and tobacco to obtain and secure labor (see, e.g., Jankowiak and Bradburd, 1996 and 2003). Aka women also express a preference for husbands that smoke, which could encourage

131 male smoking (Roulette et al., under review). Finally, if smoking does help control helminthiasis, the high rate of helminth infections among the Aka, combinded with their lack of access to commerical anthelmintics, could unconsciously encourage high consumption of more readily available anthelmintics.

Cannabis vs. helminths

Most men (95.2%) were infected with helminths, the majority (62.9%) with two or more species, similar to findings of earlier studies (Lilly et al., 2002; Froment,

2014; Roulette et al. 2014).

Worm burden was significantly negatively correlated with THCA, which is consistent with the chemotherapeutic hypothesis of drug use (e.g., Roulette et al.,

2014). With THCA in the model, cotinine was no longer a significant predictor of worm burden, contrary to our earlier study of tobacco vs. helminths (Roulette et al.,

2014). However, our sample size here was only about 1/3 that of the previous study. Inspection of the relationship between cotinine and worm burden by subregion in the current study showed the predicted negative relationship in the largest and smallest subregions (albeit with wide confidence intervals for the small subregion).

We found support for the chemoprophylaxis hypothesis: of those who were treated with a commercial anthelmintic in year 1, THC concentrations in year 2 were significantly negatively associated with reinfection levels. Furthermore, whereas Roulette et al. (2014), using the same sample, found that cotinine

132 concentrations were significantly negatively correlated with worm reinfection, here we found that THC and cotinine were both independently and significantly negatively correlated with reinfection. However, a model that controlled for age instead of cotinine had a lower AICc. The sample size of the reinfection study was small (n=23), so the mulitple regression results must be treated with caution.

The Aka do not associate smoking cannabis or tobacco with anti-parasitic properties. Cannabis and tobacco are also not used medicinally as anthelmintics.

This contrasts with the indigenous plant, motunga, which is smoked recreationally, and also consumed in a tea to treat parasitic infections (Roulette et al., 2014).

Whereas there are numerous medicinal uses of motunga, there are relatively few uses of cannabis and tobacco. However, cannabis is consumed in a tea to treat yellow fever and tobacco is placed on the skin to treat an unidentified skin infection.

Candidate self-medication mechanisms

Although the conventional view is that drug abuse impairs immunity, thus increasing susceptibility to infection (Friedman et al., 2003), if recreational drug use is explained (at least in part) by the drugs' anti-parasitic properties, this would suggest that the immune system plays a key role in regulating drug use. Indeed, there is increasing evidence that central immune signals like toll-like receptor 4

(TLR4) and interleuken 15 (IL-15) mediate use of tobacco, opiates, and alcohol

(Blednov et al., 2011 and 2012; Hutchinson et al., 2012; Liu et al., 2009). Complex effects of THC on cellular and humoral immunity have also been observed in animal and cell experiments (Cabral, 2002; Melamede, 2002; Sofia et al., 1973).

133 Limitations

Our observational research design could not establish causation. Moreover, although we were able to control for some confounds, it is possible that unmeasured variables, or perhaps sampling variation (type I error), were responsible for the effects we found. This is a particular concern for the reinfection study, which had a small sample size that limited the use of controls. One of our important controls, age, could not be measured accurately because Aka do not know their ages.

For the cross-sectional study of THCA vs. worm burden, although most of our participants had been treated with albendazole 1 year before, a few had not. We added a dummy variable indicating treatment status to our model, and the coefficients, standard errors, and p-values of our predictor variables were essentially the same (results not reported).

THC and other cannabinoids in cannabis are lipophilic, and bind readily to body fat. Consequently, and unlike nicotine (which is rapidly eliminated from the body), these compounds accumulate in body fat and other tissues, slowly rediffusing to plasma. Hence, they continue to be present in urine for an extended period -- days to weeks -- after the last consumption of cannabis. This might account for the seemingly stronger negative impact of cannabis on worm burden relative to tobacco.

In addition, urinarly levels of THCA do not decrease monotonically in time after last exposure (Grotenhermen, 2003). This means that THCA levels have a

134 variable relationship to recent cannabis smoking behavior. Other sources of uncontrolled variation include common polymorphisms in metabolic enzymes that alter rates of elimination.

Although cannabis has been shown to be toxic to plant and human parasitic helminths in vitro, the responsible compounds have not yet been identified (Roy and

Tandon, 1997; Mukhtar et al., 2013). Further, it is not known if cannabis is toxic to any of the three helminth species measured here, nor if the levels of THC and other cannabis toxins consumed by smoking cannabis would be effective against any human parasite in vivo.

Finally, we did not validate the low female cannabis smoking prevalence with urinary THCA. We note, however, that there was a close correspondence between men's self-reported cannabis use and their THCA levels, and Roulette et al. (2014) found that, if anything, Aka women's self-reports overstated their tobacco use relative to that inferred from their cotinine levels.

Concluding remarks

Sixty-seven percent of Aka men were recent cannabis users, a very high rate relative to other populations, especially considering the high cost of cannabis in this population. Similar to other populations, Aka male cannabis users were younger than non-cannabis users, and tended to smoke more tobacco. Like an earlier study, we found a negative relationship between cannabis use and material wealth, but no

135 relationship to acculturation or regional subpopulation. The low female prevalence of cannabis use corresponds to the low rates seen in other developing countries.

Most Aka men were infected with one or more species of helminth. We found a negative relationship between men's THCA levels and worm burden, and between

THCA levels and reinfection with helminths one year after treatment with the commercial anthelmintic albendazole. It is worth considering that the exceptionally high prevalence of cannabis and tobacco use among Aka men is linked to their exceptionally high rate of helminthiasis and relative lack of access to commercial anthelmintics (Roulette et al., 2014). More generally, these results provide further evidence of a link between parasite infection and drug use, two of the world's great health problems, and highlight the need for more research on the high rate of male substance use in this population.

Acknowledgements The authors thank Nicaise Molende for assistance in the field, and the Aka for generously agreeing to participate in our study. The authors declare no conflicts of interest.

136

Literature Cited Agrawal A, Budney AJ, Lynskey MT. 2012. The co-occurring use and misuse of cannabis and tobacco: a review. Addiction 107(7):1221-1233. Agrawal A, Lynskey MT. 2009. Tobacco and cannabis co-occurrence: does route of administration matter?. Drug and alcohol dependence 99(1):240-247. Agurell S, Halldin M, Lindgren JE, et al. 1986. Pharmacokinetics and metabolism of Δ-1-tetrahydrocannabinol and other cannabinoids with emphasis on man. Pharmacol Rev 38:21-43. Appendino G, Gibbons S, Giana A, Pagani A, Grassi G, Stavri M, Smith E, Rahman MM. 2008. Journal of Natural Products 71(8):1427-1430. doi:10.1021/np8002673 Bachman JG, Wadsworth KN, O'Malley P , Johnston L, Schulenberg J. 1997. Smoking, drinking and drug use in young adulthood: the impacts of new freedoms and responsibilities. Mahwah, NJ: Lawrence Erlbaum. Bahuchet S. 2014. Cultural Diversity of African Pygmies. In: Hewlett BS, editor: Hunter-gatherers of the Congo basin: Cultures, histories and biology of African Pygmies. New Brunswick: Transaction Publishers. p 1-30. Bailey R. 1991. The Behavioral Ecology of Efe Pygmy Men in the Ituri Forest, Zaire. Anthropological Papers, No. 86. Ann Arbor, Michigan: Museum of Anthropology, University of Michigan. Benowitz NL. 1996. Cotinine as a Biomarker of Environmental Tobacco Smoke Exposure. Epidemiologic Reviews 18(2):188-204. Bernard HR. 2006. Research Methods in Anthropology: Qualitative and Quantitative Approaches, Fourth Edition. Lanham, MD: AltaMira Press. Blagg W, Schloegel EL, Mansour NS, Khalaf GI. 1955. A new concentration technic for the demonstration of protozoa and helminth eggs in feces. The American Journal of Tropical Medicine and Hygiene 4:23–28. Blednov YA, Benavidez JM, Geil C, Perra S, Morikawa H, Harris RA. 2011. Activation of inflammatory signaling by lipopolysaccharide produces a prolonged increase of voluntary alcohol intake in mice. Brain, Behavior, and Immunity 25:S92–S105. Blednov YA, Ponomarev I, Geil C, Bergeson S, Koob GF, Harris RA. 2012. Neuroimmune regulation of alcohol consumption: behavioral validation of genes obtained from genomic studies. Addiction Biology 17:108–120.

137 Borgwardt S, Allen P, Bhattacharyya S, et al. 2008. Neural basis of delta-9- tetrahydrocannabinol and cannabidiol: effects during response intervention. Biol Psychiatry 64:966–73. Burnham KP, Anderson DR. 2004. Multimodel inference: understanding AIC and BIC in model selection. Sociological methods & research 33(2):261-304. Cabral G. 2002. Immune system. In: Grotenhermen F, Russo E, editors. Cannabis and cannabinoids: pharmacology, toxicology, and therapeutic potential. Binghamton, NY: Haworth Press. p 279-288. Clarke R, Merlin M. 2013. Cannabis: Evolution and ethnobotany. Oakland, CA: Univ of California Press. Degenhardt L, Hall W. 2012. Extent of illicit drug use and dependence, and their contribution to the global burden of disease. Lancet 379:55-70. Degenhardt L, Chiu W-T, Sampson N, Kessler RC, Anthony JC, et al. 2008. Toward a Global View of Alcohol, Tobacco, Cannabis, and Cocaine Use: Findings from the WHO World Mental Health Surveys. PLoS Med 5(7):e141. doi:10.1371/journal.pmed.0050141 du Toit BM. 1976. Man and cannabis in Africa: a study of diffusion. African Economic History I 17:3-5. du Toit BM. 1980. Cannabis in Africa: a survey of its distribution in African and a study of cannabis use in multi-ethnic South Africa. Rotterdam: A. A. Balkema. Dudley R. 2000. Evolutionary origins of human alcoholism in primate frugivory. The Quarterly Review of Biology 75(1):3-15. Ellis Jr GM, Mann MA, Judson BA, et al. 1985. Excretion patterns of cannabinoid metabolites after last use in a group of chronic users. Clin. Pharmacol. Ther. 38(5):572-8. Foltz RL. 2007. Mass Spectrometric Methods for Determination of Cannabinoids in Physiological Specimens. In: ElSohly MA, editor. Marijuana and the Cannabinoids. Totowa, NJ: Humana Press. p 179-204. Forbey JS, Harvey AL, Huffman MA, Provenza FD, Sullivan R, Tasdemir D. 2009. Exploitation of secondary metabolites by animals: A response to homeostatic challenges. Integrative and Comparative Biology 49(3):314-328. doi:10.1093/icb/icp046 Friedman H, Newton C, Klein TW. 2003. Microbial Infections, Immunomodulation, and Drugs of Abuse. Clinical Microbiology Reviews 16:209–219. Froment A. 2014. Human biology and the health of African rainforest inhabitants. In: BS Hewlett, editor. Hunter-gatherers of the Congo basin: Cultures, histories and biology of African Pygmies. New Brunswick: Transaction Publishers. p. 117-164.

138 Grotenhermen F. 2003. Pharmacokinetics and Pharmacodynamics of Cannabinoids. Clin. Pharmacokinet. 42(4):327-360. Hagen EH, Sullivan RJ, Schmidt R, Morris G, Kempter R, Hammerstain P 2009. Ecology and neurobiology of toxin avoidance and the paradox of drug reward. Neuroscience 160:69–84. Hagen EH, Roulette CJ, Sullivan RJ 2013. Explaining human recreational use of ‘pesticides’: The neurotoxin regulation model of substance use vs. the hijack model and implications for age and sex differences in drug consumption. Front. Psychiatry 4:142. doi: 10.3389/fpsyt.2013.00142 Hall W, Degenhardt L 2007. Prevalence and correlates of cannabis use in developed and developing countries. Current Opinion in Psychiatry 20:393-397. Hall W, Degenhardt L 2009. Adverse health effects of non-medical cannabis use. Lancet 374:1383-91. Hallet JP. 1975. Pygmy Kitabu. USA: Fawcett Crest Books. Hammond JA, Fielding D, Bishop SC. 1997. Prospects for plant anthelmintics in tropical veterinary medicine. Veterinary Research Communications 21:213–228. Harrell, F. E. (2001). Regression modeling strategies: with applications to linear models, logistic regression, and survival analysis. Springer. Hewlett BS. 1977. Notes on the Mbuti and Aka Pygmies of Central Africa. Master’s thesis (Anthropology). Chico, CA: California State University Chico. Hewlett BS. 1996. Cultural diversity among African pygmies. In: Kent S, editor. Cultural Diveristy among Twentieth-Century Foragers: An African Perspective. Cambridge: Cambridge University Press. p 215–244. Hewlett BS, editor. 2014. Hunter-gatherers of the Congo basin: Cultures, histories and biology of African Pygmies. New Brunswick: Transaction Publishers. Hill EM, Chow K. 2002. Life history theory and risky drinking. Addiction 97:401-413. Hillig KW. 2005. Genetic evidence for speciation in Cannabis (Cannabaceae). Genetic Resources and Crop Evolution 52(2):161-180. Hotez PJ, Brindley PJ, Bethony JM, King CH, Pearce EJ, Jacobson J. 2008. Helminth infections: the great neglected tropical diseases. The Journal of Clinical Investigation 118:1311–1321. Huestis MA. 2007. Human Cannabinoid Pharmacokinetics. Chem Biodivers 4(8):1770-1804. Huffman MA. 1997. Current evidence for self-medication in primates: A multidisciplinary perspective. Yearbook of Physical Anthropology 104:171–200.

139 Hurtado AM, Frey M, Hill K, Hurtado I, and Baker J. 2008. The role of helminthes in human evolution: implications for global health in the 21st century. In: Elton S, O'Higgins P, editors. Medicine and evolution: Current applications, future prospects. CRC Press. p 151e-178e. Hutchinson MR, Northcutt AL, Hiranita T, Wang X, Lewis SS, Thomas J, van Steeg K, Kopajtic TA, Loram LC, Sfregola C, et al. 2012. Opioid Activation of Toll-Like Receptor 4 Contributes to Drug Reinforcement. The Journal of Neuroscience 32:11187–11200. Jankowiak W, Bradburd D. 1996. Using Drug Foods to Capture and Enhance Labor Performance: A Cross-Cultural Perspective. Current Anthropology 37(4):717-720. Jankowiak W, Bradburd D, editors. 2003. Drugs, labor, and colonial expansion. Tucson, AZ: University of Arizona Press. Komiya Y, Kobayashi A. 1966. Evaluation of Kato’s thick smear technic with a cellophane cover for helminth eggs in feces. Japanese Journal of Medical Science & Biology 19:59–64. Lende DH, Smith EO. 2002. Evolution meets biopsychosociality: an analysis of addictive behavior. Addiction 97(4), 447-458. Lilly AA, Mehlman PT, Doran D. 2002. Intestinal Parasites in Gorillas, Chimpanzees, and Humans at Mondika Research Site, Dzanga-Ndoki National Park, Central African Republic. International Journal of Primatology 23:555–573. Liu YZ, Pei YF, Guo YF, Wang L, Liu XG, Yan H, Xiong DH, Zhang YP, Levy S, Li J, et al. 2009. Genome-wide association analyses suggested a novel mechanism for smoking behavior regulated by IL15. Molecular Psychiatry 14:668–680. Lowe RH, Abraham TT, Darwin WD, Herning R, Cadet JL, Huestis MA. 2009. Extended urinary Δ9-tetrahydrocannabinol excretion in chronic cannabis users precludes use as a biomarker of new drug exposure. Drug and alcohol dependence 105(1):24-32. McPartland JM. 1997. Cannabis as repellent and pesticide. Journal of the International Hemp Association 4(2):87-92. http://www.druglibrary.net/olsen/HEMP/IHA/jiha4210.html Melamede R. 2002. Possible mechanisms in autoimmune diseases. In: Grotenhermen F, Russo E, editors. Cannabis and cannabinoids: pharmacology, toxicology, and therapeutic potential. Binghamton, NY: Haworth Press. p 111-22. Mukhtar T, Kayani MZ, Hussain MA. 2013. Nematicidal activities of Cannabis sativa L. and Zanthoxylum alatum Roxb. against Meloidogyne incognita. Industrial Crops and Products 42:447–453.

140 Oishi T, Hayashi K. 2014. From ritual dance to disco: change in habitual use of tobacco and alcohol among the Baka hunter-gatherers of Southeastern Cameroon. African Study Monographs Suppl 47:143–163. Ng M, Freeman MK, Fleming TD, Robinson M, et al. 2014. Smoking Prevalence and Cigarette Consumption in 187 Countries, 1980-2012. JAMA 311(2):183-192. doi:10.1001/jama.2013.284692 Philips JE. 1983. African Smoking and Pipes. The Journal of African History 24(3):303-319. Rätsch C. 2005. The encyclopedia of psychoactive plants: ethnopharmacology and its applications. Rochester, VT: Inner Traditions/Bear & Co. Rodriguez E, Cavin C. 1982. The Possible Role of Amazonian Psychoactive Plants in the Chemotherapy of Parasitic Worms—A Hypothesis. Journal of Ethnopharmacology 6:303-309. Rogers R, Crowl CP, Eimstad WM, Hu MW, Kam JK, Ronald RC, Rowley GL, Ullman EF. 1978. Homogenous Enzyme Immunoassay for Cannabinoids in Urine. Clin. Chem. 24(1):95-100. Roulette CJ, Mann H, Kemp BM, Remiker M, Roulette JW, Hewlett BS, et al. 2014. Tobacco use vs. helminths in Congo basin hunter-gatherers: self-medication in humans? Evolution and Human Behavior 35(5):397-407. doi:10.1016/j.evolhumbehav.2014.05.005 Roulette CJ, Hewlett BS and Hagen EH (under review) Tobacco use among Aka foragers of the Congo basin: implications for anthropological theories of drug use and public health. Roy B, Tandon V. 1997. In vitro fluckicidal effect of leaf extract Cannabis sativa Linn. on the trematode Fasciolopsis buski. Indian J Exp Biol 35(1):80-82. SAMHSA. 2008. Mandatory Guidelines for Federal Workplace Drug Testing Programs. Federal Register 73(228):71858-71907. www.gpo.gov/fdsys/pkg/FR- 2008-11-25/pdf/E8-26726.pdf Schebesta P. 1933. Among Congo Pygmies. London: Hutchinson and Co. Schebesta P. 1936. My Pygmy and Negro Hosts. London: Hutchinson and Co. Schebesta P. 1941. Die Bambuti-Pygmäen von Ituri, vol. 2: Ethnographie der Ituri- Bambuti. Brussels: Goerge van Campenhout. Schensul S, Schensul JJ, LeCompte MD. 1999. Essential Ethnographic Methods: Observations, Interviews, and Questions. Walnut Creek, CA: AltaMira Press. Schultes, RE. 1970. Random thoughts and queries on the botany of cannabis. In: Joyce CRB, Curry SH, editors. The botany and chemistry of cannabis. London: J. & A. Churchill. p 11-33.

141 Sofia RD, Nalepa SD, Harakal JJ, et al. 1973. Anti-edema and analgesic properties of Δ9-tetrahydrocannabinol (THC). J Pharmacol Exp Ther 186(3):646-55. Suárez-Rodríguez M, López-Rull I, Garcia CM. 2013. Incorporation of cigarette butts into nests reduces nest ectoparasite load in urban birds: new ingredients for an old recipe? Biology Letters 9(1). doi:10.1098/rsbl.2012.0931 Turnbull CM. 1968. The forest people. New York: Touchstone. Turnbull CM. 1976. Wayward servants: The two worlds of the African pygmies. Westport, CT: Greenwood Press. Turnbull CM. 1983. The Mbuti pygmies: change and adaptation. New York: Holt, Rinehart, and Winston. Turner CE, Elsohly MA, Boeren EG. 1980. Constituents of Cannabis sativa L. XVII. A review of the natural constituents. J. Natural Products 43:169-234. UN. 2010. UN Office on Drugs and Crime. World drug report 2010. United Nations, New York. www.unodc.org/documents/wdr/WDR_2010/World_Drug_Report_2010_lo_res.pdf (accessed on Dec 11, 2014). UN. 2013. UN Office on Drugs and Crime. World drug report 2013. United Nations, New York. www.unodc.org/unodc/secured/wdr/wdr2013/World_Drug_Report_2014.pdf. (accessed on Dec 11, 2014). Villalba JJ, Miller J, Ungar ED, Landau SY, Glendinning J. 2014. Ruminant self- medication against gastrointestinal nematodes: evidence, mechanism, and origins. Parasite 21:31 doi:10.1051/parasite/2014032 Wrangham, RW, Nishida T. 1983. Aspilia spp. leaves: a puzzle in the feeding behavior of wild chimpanzees. Primates 24(2):276–282. Wyatt GB. 1977. Health in Melanesia. In: Winslow JH, editor. The Melanesian Environment. Canberra, Australia: National University Press. p 459-462.

142

Table 1. Descriptive Statistics

N Min Max Median Mean SD

Age (years) 61 18.0 60 35.0 36.70 10.00

THCA (ng/ml) 62 1.3 4100 370.0 663.00 830.00

Cotinine (ng/ml) 61 0.0 980 240.0 301.00 270.00

Worm burden score 62 0.0 14 2.1 3.27 3.10

Wealth score 56 1.0 8 2.0 2.27 1.40

Acculturation score 56 0.0 4 2.0 2.21 0.89

143

Table 2: Logistic regression model of cannabis smoker status (THCA > 50 ng/ml) vs. age (year) and material wealth score. N=56. Likelihood ratio test of model significance: X2 = 22, df = 2, p = 1.3 × 10-5. Null deviance: 71.7 on 55 df. Residual deviance: 49.3 on 53 df. Comparing predicted vs. observed smoker status, and using cutoff=0.50, the sensitivity (true positives) was 92% and the specificity (true negatives) was 63%. The area under the receiver operating curve (AUC), corrected for bias using 1000 bootstrap replications (Harrell 2001), was AUC = 0.85.

Dependent variable:

Smoker

Age -0.134*** (0.040)

Wealth -0.890*** (0.315)

Constant 8.000*** (2.109)

Observations 56 Log Likelihood -24.630 Akaike Inf. Crit. 55.270

Note: p<0.1; p<0.05; p<0.01

144

Table 3. Negative binomial GLM models of worm burden score with log link. N=56. (1) Model that best approximates model in Roulette et al. (2014). (2) Lowest AICc value of a priori models.

Dependent variable:

Worm.burden.score (1) (2)

Village -0.69*** -0.55** (0.24) (0.24)

Wealth 0.15* (0.08)

Acculturation -0.22 (0.14)

Age -0.03** -0.02** (0.01) (0.01)

THCA -0.0003** -0.0004** (0.0002) (0.0002)

Cotinine 0.0002 (0.0005)

Constant 3.04*** 2.66*** (0.67) (0.52)

Observations 56 56 Log Likelihood -121.50 -124.20 3.02** theta 2.44*** (0.86) (1.18) Akaike Inf. Crit. 257.10 256.40

Note: p<0.1; p<0.05; p<0.01

145

Table 4. OLS models of helminth reinfection score. (1) Reinfection vs. THCA and cotinine (AICc=60.8). (2) Reinfection vs. THCA and age (AICc=55.8).

Dependent variable:

scale(Delta worm) (1) (2) scale(THC) -0.400** -0.725*** (0.178) (0.170) scale(Cotinine) -0.359* (0.175) scale(Age) -0.488*** (0.152)

Observations 23 23 R2 0.372 0.494 Adjusted R2 0.312 0.446 Residual Std. Error (df = 21) 0.811 0.728 F Statistic (df = 2; 21) 6.210*** 10.250***

Note: p<0.1; p<0.05; p<0.01

146

Figure 1: THCA histogram for all participants (blue) and estimated density (red). The vertical dotted line represents the conventional cutoff of 50 ng/ml used to identify recent cannabis smokers. X-axis is on a log scale. Dotted vertical line is conventional cutoff for determining recent consumption of cannabis.

147

Figure 2. Box plot of THCA levels vs. self-reported cannabis smoker status. Y-axis is on a log scale. Horizontal dotted line represents the conventional threshold of 50 ng/ml for recent consumption of cannabis.

148

Figure 3: Probability of smoking cannabis as a function of age and wealth (1=low, 3=high). Bars are 95% confidence intervals. See table 2 for model coefficients and indices of fit. Wealth=3 was chosen as "high" wealth because although wealth scores ranged from 1 to 8, 78.6% of men had scores less than 3. To display overlapping points, a small amount of jitter was added to age.

149

Figure 4. Effect plot of worm burden vs. THCA, village, and age (model 2, table 3).

150

Figure 5. Helminthes reinfection score vs. THCA concentration. Line fit by linear regression.

151

152

153

154

155

156 CHAPTER FIVE

CONCLUSION

There are five characteristics of Aka tobacco and cannabis use that are highlighted in these studies: (1) there is a very large sex difference in use; (2) this large difference is due to low female use, but most notably, very high male use; (3) the high male use is surprising given the high cost of tobacco and cannabis in this population; (4) tobacco is extensively shared; and (5) among men, intestinal helminth infections are negatively associated with tobacco and cannabis use.

Many conventional explanations for substance use fail to explain the large sex difference in use.

For example, the neurobiological perspective focuses on the rewarding effects of psychoactive drugs, but these are presumably the same in men and women, so therefore cannot easily explain the sex difference in drug use. Gender disparities that result from political-economic and sociocultural factors (e.g. Waldron et al., 1988; Kaplan et al. 1990; Heath 1991; McDonald 1994; Room 1996; WHO 2007; Eriksen et al.

2012) also cannot easily explain the large sex difference in use because the Aka are known for having pronounced gender equality and also value autonomy. Indeed, the gender disparities hypothesis fails to explain why, cross-nationally, men almost universally use more than women (see for example,

Degenhardt et al. 2008). Other explanations focus on the effects of transnational tobacco companies and anti-tobacco campaigns. For example, the prevalence of current daily smoking among adults in the US declined 24.3% between 1965, a year after the first surgeon general warning linking smoking and health, and 2012 (US Department of Health and Human Services, 2014). Today, the increased use of tobacco in the developing world is due, in part, to the marketing aims of transnational tobacco corporations (Lee et al., 2012). However, because the Aka have relatively little access to public health messages and tobacco marketing gimmicks these explanations cannot easily explain the high male use or the low female use.

One reason for low female use might be that women are protecting their fetuses from teratogens.

The Aka themselves mentioned several negative effects of tobacco use on the developing fetus, and many women mentioned they did not like tobacco because it made them sick. Because most drugs are neurotoxins, and the Aka are a natural fertility population in which women are pregnant or lactating for much of their reproductive years, Aka women might be avoiding psychoactive substances as a fetal

157 protection strategy (see for example, Hagen et al. 2013). The finding that post-reproductive aged Aka women were more likely to use tobacco than reproductive aged Aka women offers indirect support for this hypothesis. Alternatively, because most men prefer a non-smoking spouse, younger women might not smoke as a mate attraction/retention strategy. It is difficult to tease apart the relative contributions of these factors, however, because we have limited data on female substance use.

Compared to most global populations, the prevalence of tobacco and cannabis use among adult

Aka men is exceptionally high. This is surprising given the high cost of tobacco and cannabis in this population. The perceived labor and hunting enhancing effects of tobacco and cannabis make these substances highly valuable to the Aka, which might explain the high male use. Male smoking might also be a mate attraction/retention strategy, as most women prefer a spouse who smokes. Speculatively, high male use might also be explained, in part, by self-medication. Although Aka men and women suffer intestinal helminth infections, adult men and post-reproductive women might be the only one’s capable of affording (in the fitness sense) to use psychoactive substances. In contrast, reproductive aged women and children cannot afford to self-medicate with drugs because the costs of neurotoxin exposure outweigh the benefits.

Support for the self-medication hypothesis was found across all tests, albeit with some caveats.

There were also hints that THCA (a metabolite of THC) might have a greater effect on worms than cotinine (a metabolite of nicotine), but further tests comparing the two substances are needed. In addition, the cross-sectional relationship between cotinine and worm burden appeared similar to a dose-response curve, but there was no such curvilinear relationship between THCA and worm burden.

There are many features of Aka substance use that were left unexplored. For example, self- and peer-reports of tobacco and cannabis use (of all Aka in the three communities) found that childhood smoking was virtually absent, which is consistent with global use patterns (e.g. Degenhardt 2008), More systematic data are therefore needed on the transition into smoking, which appears to begin in adolescence. One reason that Aka adolescent males might smoke is to attract women. According to self- reports, most women prefer men who can hunt and provide honey (and who smoke), and unpublished data indicate that risky subsistence activities (i.e. climbing trees for honey and hunting with a spear) are positively correlated with substance use. Furthermore, according to one focus group interview with a

158 group of young Aka men, it is important to smoke at dances because women like it and because there are numerous potential mating opportunities at dances. Alternatively, Aka adolescent smoking might be mediated by risk of helminth infection, which in many populations peak in middle childhood or adolescence (e.g. Woolhouse 1998; Blackwell et al., 2011).

We also did not collect behavioral measures of substance use, which was a major limitation of the self-medication studies in particular. In addition, we did not collect data on alcohol, another important psychoactive substance used by the Aka. Finally, because the Aka acquire most of their tobacco, cannabis, and alcohol via labor-for-drug exchanges with farmers, more data are needed comparing Aka and farmer substance use and exploring the effect of these exchanges on Aka patterns of use.

159 REFERENCES:

Blackwell AD,Gurven MD, Sugiyama LS,Madimenos FC, Liebert MA,Martin MA,et al. 2011. Evidence for

a peak shift in a humoral response to helminths: age profiles of IgE in the Shuar of Ecuador, the

Tsimane of Bolivia, and the U.S. NHANES. PLoS Negl Trop Dis 5(6):e1218.

doi:10.1371/journal.pntd.0001218

Degenhardt L, Chiu W-T, Sampson N, Kessler RC, Anthony JC, et al. 2008. Toward a Global View of

Alcohol, Tobacco, Cannabis, and Cocaine Use: Findings from the WHO World Mental Health

Surveys. PLoS Med 5(7):e141. doi:10.1371/journal.pmed.0050141

Eriksen M, Mackay J, & Ross H. 2012. The Tobacco Atlas (4th ed.). Atlanta, GA: American Cancer

Society.

Hagen EH, Roulette CJ, Sullivan RJ 2013. Explaining human recreational use of ‘pesticides’: The

neurotoxin regulation model of substance use vs. the hijack model and implications for age and sex

differences in drug consumption. Front. Psychiatry 4:142. doi: 10.3389/fpsyt.2013.00142

Heath D. B. 1991. Women and alcohol: Cross-cultural perspectives. Journal of substance abuse, 3(2),

175-185.

Kaplan M, Carriker L, & Waldron I. 1990. Gender differences in tobacco use in Kenya. Social Science and

Medicine, 30, 305-10.

Lee S, Ling PM & Glantz SA 2012. The vector of the tobacco-epidemic: tobacco industry practices in low

and middle-income countries. Cancer Causes Control, 23(0 1):117–129. doi:10.1007/s10552-012-

9914-0.

McDonald, M. 1994. Gender, drink and drugs. Oxford, UK: Berg.

Room, R. 1996. Gender roles and interactions in drinking and drug use. Journal of substance abuse, 8(2),

227-239.

U.S. Department of Health and Human Services. 2014. The Health Consequences of Smoking: 50 Years

of Progress. A Report of the Surgeon General. Atlanta, GA: U.S. Department of Health and Human

Services, Centers for Disease Control and Prevention, National Center for Chronic Disease

Prevention and Health Promotion, Office on Smoking and Health. Printed with corrections, January

2014.

160 Waldron I., Bratelli G., Carriker L., Sung W.-C., Vogeli C., & Waldman E. 1988. Gender differences in

tobacco use in Africa, Asia, the Pacific, and Latin America. Social Science and Medicine, 27(1), 269-

75.

WHO 2007. Gender and tobacco control: A Policy Brief. World Health Organization.

Woolhouse M. 1998. Patterns in parasite epidemiology: the peak shift. Parasitol Today, 14(10):428–34.

doi:10.1016/S0169-4758(98) 01318-0

161

APPENDIX

Title: The Puzzle of High Male Tobacco Use, and Low Female Use, among Congo Basin

Foragers: Implications for Anthropological Theories of Drug Use and Public Health.

Journal: Human Nature.

Authors: Roulette, C. J., Hagen, E. H., & Hewlett, B. S.

Affiliation: Department of Anthropology, Washington State University.

Email: [email protected]

Online Resource 1. Figure A1: Gbangaya (locally grown tobacco) rolled into balls and ready to be sold. Figure A2: Leaves of motunga (Polyalthia suaveolens, Annonaceae, also known as Greenwayodendron suaveolens).

Figure A1

163 Figure A2

164 Title: The Puzzle of High Male Tobacco Use, and Low Female Use, among Congo Basin

Foragers: Implications for Anthropological Theories of Drug Use and Public Health.

Journal: Human Nature.

Authors: Roulette, C. J., Hagen, E. H., & Hewlett, B. S.

Affiliation: Department of Anthropology, Washington State University.

Email: [email protected]

Online Resource 2. Description of the use of snuff, chewing tobacco, and cannabis, among the Aka, and of their smoking materials.

Dry snuff, a powdered tobacco inhaled through the nose or taken by mouth, is also used.

Called ndako wa mio by the Aka, it is used much more often by the Ngandu farmers than the Aka.

It is sold in a small plastic bag for 50 CFA and can last up to one week. Older adults use snuff more often than younger adults do. One reason for this is that these individuals used to smoke but quit smoking for personal health reasons and instead started using snuff. Snuff is also used medicinally to treat headaches and older adults have more headaches than younger adults.

Some farmers also chew gbangaya, much like chewing tobacco is used in the West. However, chewing tobacco is rare among Aka foragers. Finally, Aka also smoke cannabis, bangui. It is illegal to use and/or grow cannabis in the Central African Republic, which complicated our ability to collect data on its use. It is apparent that many Aka, especially men, smoke marijuana (see

Roulette et al. under review, for results of our study of Aka cannabis use). The cultural beliefs of marijuana (Hewlett 1977) are congruent with the attitudes towards tobacco and motunga.

The Aka use several types of rolling leaves, most commonly moseti (Renealmia sp.) leaves. They also use mongongo (Megaphrynium sp.) and bendjo (species name unknown), but the use of esanja, undolu/mutongelenge, palm leaf, and likoka (species names unknown) is not as frequent. Hewlett (1977) mentioned that Aka also use a leaf called ndilingbey (species name unknown). Aka occasionally smoke ndako with a pipe. They place tobacco in the bowl and inhale through the stem, sometimes through water. Pipes are common among central-African forest

165 foragers. Efe archers, for instance, use bamboo and gourd water pipes, and Mbuti net-hunters use four-foot banana-stem pipes (Hewlett 1977). Aka pipes are not as elaborate as Efe and Mbuti pipes, and it appears that the use of them is declining. Over the last several field seasons (the summers of 2008, 2010, 2011, and 2012) for instance, we never saw one being used. Pipes are currently known as makundu. Some Aka still refer to pipes by their old name, pokbo. Gbangaya and blancs (when they are broken), motunga and cannabis are all smoked in pipes. Aka use pipes both in the forest and near the village. They were probably adopted from Bantu villagers

(Hewlett 1977), who might have adopted the technology from Arab traders (Laufer et al. 1930).

166 Title: The Puzzle of High Male Tobacco Use, and Low Female Use, among Congo Basin

Foragers: Implications for Anthropological Theories of Drug Use and Public Health.

Journal: Human Nature

Authors: Roulette, C. J., Hagen, E. H., & Hewlett, B. S.

Affiliation: Department of Anthropology, Washington State University.

Email: [email protected]

Online Resource 3: Description of the medicinal uses of tobacco, motunga, and cannabis among the Aka.

Aka treat dombo, an unidentified skin-infection, by placing gbangaya (locally grown tobacco) directly on the infected skin. Gbangaya is also used to cover a wound after it bleeds to reduce swelling. There were no mentions of using manufactured cigarettes medicinally.

Aka use motunga to treat stomach parasites (madjembe). They make a tea of the roots and bark of the plant. They shave the roots and bark into ½ to 1-cup water where it soaks for five minutes (If one is really sick he or she can scrape the bark off of the tree and eat it right away).

Children (mona) take ¼ to ½ the dose of adults. Aka say it is very bitter tasting. A motunga tea was also listed as a treatment for prolonged coughing and chest pain caused by smoking.

Cannabis roots are used in a tea for yellow fever (i.e. when one “has red urine, yellow eyes, a fever, and is very tired”). Cannabis is also rubbed on the face to treat pimples.

Snuff is used to treat headaches, enlarged testes (caused by working too hard), and snake bites. To treat the latter two, snuff is mixed with water into a tea. Eating the snuff can also treat snakebites.

167 Supporting information Determining cotinine threshold distinguishing smokers from non-smokers To determine which Aka had smoked tobacco recently, it was necessary to identify a threshold concentration of salivary cotinine that would distinguish regular smokers from nonsmokers exposed to environmental tobacco smoke (ETS). In Western populations, estimates of this threshold have ranged from 3-20 ng/ml, with 15 ng/ml a widely used standard in the past; newer estimates have been revised downwards, however, which probably reflects recent restrictions on public smoking [37, 38].

Aka live outdoors, which might reduce ETS exposure, but sleep in small huts, which might increase ETS exposure. To determine an optimal cutoff using a receiver operating characteristic curve (ROC) analysis, female Aka (most of whom do not smoke) who (1) self- reported that they did not smoke tobacco and (2) whose cotinine values did not clearly indicate recent smoking (i.e., were less than 50 ng/ml) were classified as true nonsmokers.

The maximum cotinine value in this group (n=34) was 4.2 ng/ml (M = 1.2, SD=0.98). Men

(most of whom smoke) who self reported that they smoked tobacco were classified as smokers. The area under the ROC curve was .995. A threshold of 5 ng/ml had a specificity of 100% and a sensitivity of 95%.

Author contributions: Study concept and management: E.H.H. Study design: E.H.H., C.J.R., R.J.S., B.S.H. M.K., D.M. Data and sample collection: C.J.R, M.R., J.W., E.H.H., B.S.H. Cotinine assays: C.J.R., M.R., E.H.H. Helminth egg identification and counts: M.K, S.B., D.M. Genotyping: H.M., B.M.K. Data analysis: E.H.H., C.J.R. Manuscript preparation: E.H.H., C.J.R. All authors discussed study execution and results and commented on the manuscript.

168

Figure S1: Manipulation check. The effect of intervention on the distribution of worm burden scores in the placebo control group (top) vs. the albendazole treatment group

(bottom). Red dots are the median values. The dashed vertical line represents the administration of placebo (top) and albendazole (bottom) after collection of sample 3.

169

Figure S2: Boxplot of mean baseline cotinine concentration vs. predicted metabolic phenotype.

170

CONSORT 2010 Flow Diagram

Assessed for eligibility (n=247) Enrollment

Excluded (n= 68) ♦ Left for forest prior to providing any samples (n=41) ♦ Declined to participate (n=0) ♦ Not meeting inclusion criteria (n=27) )

Randomized (n=179)

Allocation Allocated to albendazole treatment (n=87) Allocated to placebo (n=92) ♦ Received allocated intervention (n=87) ♦ Received allocated intervention (n=92) ♦ Did not receive allocated intervention (n=0) ♦ Did not receive allocated intervention (n=0)

Follow-Up Lost to follow-up (in forest) (n=22) Lost to follow-up (in forest) (n=30)

Discontinued intervention (n=0) Discontinued intervention (n=0)

Analysis Analysed (n=65) Analysed (n=62) ♦ Excluded from analysis (give reasons) ♦ Excluded from analysis (give reasons) (n=0) (n=0)

171

Family: Negative Binomial(1.511) Link function: log

Formula: worm_burden ~ village + acculturation + wealth + s(age) + s(cotinine) + offset(log(stool_count))

Parametric coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) 3.06088 0.23074 13.265 <2e-16 *** villageTRUE 0.18167 0.13514 1.344 0.181 acculturation 0.01102 0.08450 0.130 0.896 wealth -0.06126 0.05228 -1.172 0.243 ---

Approximate significance of smooth terms: edf Ref.df F p-value s(age) 1.90 2.399 1.115 0.337547 s(cotinine) 2.07 2.573 6.956 0.000422 *** ---

R-sq.(adj) = 0.0488 Deviance explained = 10.4% UBRE score = 1.0472 Scale est. = 1 n = 177

Table S1: Generalized additive model of worm burden score as smooth functions of age and cotinine concentration, controlling for linear functions of subpopulation, acculturation, wealth, and age.

172

All participants Lost to follow-up Remaining participants

Treatment Control Treatment Control Treatment Control

N 87 92 22 30 65 62

Age 36.8 (10) 36.3 (10.1) 31.6 (8.5) 33.5 (9) 38.5 (9.9) 37.6 (10.4)

Cotinine 163.3 (147.8) 139.3 (135.6) 139.2 (109.5) 122.6 (105.2) 171.4 (158.6) 147.3 (148.2)

Worm score 21.1 (17.2) 24.9 (21.7) 26.7 (16.8) 29.4 (25.4) 19.2 (17) 22.7 (19.6)

Table S2: Attrition vs. participant characteristics. Cotinine and worm burden are the arithmetic mean (SD) of baseline values (s1-s3).

173

Allele Activity Frequency

1A Normal 97 1B Fast 20

9 Slow 1 17 Slow 16 20 Slow 0

Table S3: Aka CYP2A6 allele frequencies.

174

Family: Negative Binomial(2.67) Link function: log

Formula: worm.pretreatment ~ phenotype + s(baseline_cotinine) + s(age_c, by = phenotype) + offset(log(stool_count.pretreatment))

Parametric coefficients: Estimate Std. Error z value Pr(>|z|) (Intercept) 2.64620 0.19987 13.240 <2e-16 *** phenotype2Normal 0.06876 0.24070 0.286 0.775 ---

Approximate significance of smooth terms: edf Ref.df Chi.sq p-value s(baseline_cotinine) 2.437 2.964 18.630 0.000313 *** s(age_c):phenotypeSlow 1.963 2.352 6.904 0.044431 * s(age_c):phenotypeNormal 4.379 5.366 26.759 9.04e-05 *** ---

R-sq.(adj) = 0.316 Deviance explained = 55.9% UBRE score = 0.67242 Scale est. = 1 n = 39

Table S4: Generalized additive model of worm burden as a function of CYP2A6 genotype.

175