Bushmeat represents an important source of animal protein for humans in tropical Africa. Unsustainable bushmeat hunting is a major threat to wildlife and its consumption is associated with an increased risk of acquiring zoonotic diseases, such as Ebola virus disease (EVD). During the recent EVD outbreak in West Africa, it is likely that human dietary behavior and local attitudes toward bushmeat consumption changed in response to the crisis, and that the rate of change depended on prevailing socio-economic conditions, including wealth and education. In this study, we therefore investigated the effects of income, education, and literacy on changes in bushmeat consumption during the crisis, as well as complementary changes in daily meal frequency, food diversity and bushmeat preference. More specifically, we tested whether wealthier households with more educated household heads decreased their consumption of bushmeat during the EVD crisis, and whether their daily meal frequency and food diversity remained constant. We used Generalized Linear Mixed Models to analyze interview data from two nationwide household surveys across Liberia. We found an overall decrease in bushmeat consumption during the crisis across all income levels. However, the rate of bushmeat consumption in high-income households decreased less than in low-income households. Daily meal frequency decreased during the crisis, and the diversity of food items and preferences for bushmeat species remained constant. Our multidisciplinary approach to study the impact of EVD can be applied to assess how other disasters affect social-ecological systems and improve our understanding and the management of future crises.
The consumption of wild animal meat, commonly known as bushmeat, is widespread throughout tropical regions. Bushmeat provides an essential source of protein and income for human livelihoods. However, its consumption is linked to the transmission of zoonotic diseases, such as Ebola, and its over-harvest is a major threat to many wildlife species. The bushmeat trade therefore encompasses a broad range of socio-economic and ecological issues. As such, we think it is highly important to use an interdisciplinary approach to investigate the drivers of bushmeat consumption, to improve our understanding and management of future crises. Our analysis of household interview data collected during two surveys across Liberia shows that there was an overall decrease in bushmeat consumption during the recent Ebola crisis. However, the consumption of bushmeat in wealthier households decreased less than in poorer households. In addition, we found that daily meal frequency decreased during the crisis, and the diversity of food items and preferences for bushmeat species remained constant.
Citation: Ordaz-Németh I, Arandjelovic M, Boesch L, Gatiso T, Grimes T, Kuehl HS, et al. (2017) The socio-economic drivers of bushmeat consumption during the West African Ebola crisis. PLoS Negl Trop Dis 11(3): e0005450. doi:10.1371/journal.pntd.0005450
Editor: Oladele B. Akogun, Common Heritage Foundation, NIGERIA
Received: October 31, 2016; Accepted: March 2, 2017; Published: March 10, 2017
Copyright: © 2017 Ordaz-Németh et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: This study was funded by the Robert Bosch Foundation, the Max-Planck Society, and Erasmus+. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
The recent Ebola virus disease (EVD, ) epidemic that emerged in March 2014 in West Africa [2, 3] was the largest recorded, resulting in over 28,600 cases and 11,300 deaths in Guinea, Liberia and Sierra Leone . EVD is a deadly zoonotic disease that is transmitted to humans through contact with the blood and other bodily fluids of infected wildlife, such as fruit bats, forest antelopes, and nonhuman primates [5, 6]. Although there are several contributing factors that caused the outbreak to rapidly expand, such as poverty [7, 8, 9] and weak medical infrastructure [8, 10], the harvest and butchering of bushmeat (i.e., wild animal meat) have been suspected as potential sources for initial spillover events in this [2, 11] and other EVD epidemics [12, 13].
Despite its health risks, bushmeat consumption is widespread throughout tropical regions and common in both rural and urban areas [14, 15], although the reasons for its consumption tend to vary between and within areas. In remote, impoverished, rural areas, bushmeat is often an essential source of animal protein that may contribute to food security, particularly where livestock and fish are inaccessible [16, 17, 18] or unaffordable . In contrast, urban consumers are likely to choose bushmeat from a number of interchangeable animal protein sources, and for a variety of reasons, such as its low cost, preference of taste, or perception of prestige [16, 20]. Bushmeat also provides an important source of cash income for rural and forest dwellers who may depend on wildlife to alleviate periods of economic hardship (e.g. crop failures), or supplement their primary source of income, which is often agriculture [21, 22].
Along with the rapid growth of human populations, the extraction of wildlife for subsistence and commercial use has become a major biodiversity threat . Since the 1970s, the abundance of large mammals in African protected areas has halved, largely due to over-hunting . To make more profit, hunters prefer large animals, such as large antelopes, elephants, and great apes, whose low intrinsic rates of population growth make them extremely vulnerable to intensive hunting. Regardless of their scarcity, they are continued to be targeted and have already been hunted to the point of extirpation in a number of places [20, 25]. These unsustainable hunting practices ultimately lead to the defaunation of otherwise undisturbed forests and create “empty forests” . The ecological impacts of over-hunting include the direct effect on the hunted populations (e.g. Tweh et al.  have documented this across Liberia), as well as indirect effects on ecosystem function and structure, which are more difficult to measure [20, 28]. For instance, Effiom et al.  found that the seedling layer in forest sites with widespread bushmeat hunting was significantly different from that of protected sites, suggesting that the loss of seed-dispersing primates in particular might impact forest regeneration processes and ultimately forest composition.
Indeed, the bushmeat trade encompasses a broad range of socio-economic and ecological issues, which highlights the need to use interdisciplinary approaches to better understand the links between the exploitation of natural resources and human socio-economic status [19, 30]. It is therefore imperative to identify the underlying drivers of bushmeat consumption to develop more effective, targeted conservation management strategies. Such strategies must aim to reduce the unsustainable harvest of bushmeat whilst improving human livelihoods and lowering the risk of zoonotic disease transmissions, such as EVD.
It is likely that due to social pressure and risk aversion, human behavior and local attitudes toward bushmeat consumption changed during the most recent and largest-ever recorded Ebola outbreak in West Africa . Furthermore, the magnitude of these changes may have been influenced by different socio-economic factors. Here, we analyze household-level data on socio-economic status, wildlife consumption and eating habits, to answer the following research question: were wealthier, more educated or literate people more likely to change (1) bushmeat consumption, (2) number of meals per day, (3) food diversity, and (4) bushmeat preference during the Ebola crisis?
Our surveys were conducted across the entire country of Liberia. Following 14 years of civil conflict, the country’s economy had been growing rapidly in recent years . However, Liberia is still among the poorest countries in the world (HDI rank 177/188, ), and was one of the countries hardest hit by the Ebola crisis [10, 31]. It is also found within the richest 5% of land area for threatened bird, amphibian and mammal species . In addition, it is home to one of the most viable chimpanzee populations in West Africa, which is primarily threatened by hunting .
Field data collection
We used two different data sources: (1) interview data collected in Liberia during a nationwide chimpanzee and large mammal survey from 2010 to 2012, and (2) interview data from a follow-up nationwide interview survey on socio-economic status and natural resource use of Liberian households during the Ebola crisis in 2015.
Between August 2010 and May 2012 a nationwide chimpanzee and large mammal survey was conducted on line transects that were systematically distributed across the country . When travelling to these transects to record large mammal abundance, survey teams visited nearby villages to conduct interviews to collect data that served as the basis of this study (for more details see Tweh et al. ). In each location (i.e., village), one to ten household heads were interviewed. If the head of the household was absent at the time, the person otherwise responsible for the household was interviewed instead. A total of 275 household heads from 70 locations were interviewed during this survey.
A follow-up interview survey was conducted between January and June 2015 to gather information on the impact of Ebola on socio-economic status and natural resource use of Liberian households. The geographical distribution of interview locations during the follow-up survey in 2015 was based on the sampling locations from the 2010–2012 survey (Fig 1), but the respondents and households were not necessarily the same across surveys. The majority of the interview questions were paired with identical retrospective questions, which made it possible to collect information about the time period before the Ebola crisis. This was an important feature of the questionnaire because the 2010–2012 survey questionnaires were far less extensive (see S2 Appendix for the complete questionnaires used in the surveys). For this reason, data from the 2015 follow-up survey were mainly used in our data analysis. Overall, there was an overlap of 60 interview locations that were sampled during both household surveys.
We ran Generalized Linear Mixed Models (GLMMs, ) for eight response variables to assess the changes in eating habits before vs. during the Ebola crisis. Specifically, we assessed the changes in (1) the number of meals eaten per day, (2) bushmeat consumption, (3) food diversity, and (4) bushmeat species preference (Table 1). In addition, we ran two Linear Mixed Models (LMMs, ) to investigate whether bushmeat and fish prices changed during the crisis, and one GLMM to investigate whether domestic meat prices changed. All but one response variable (number of meals per day) had to be processed before fitting them to the models; for more details see S1 Appendix.
To test for an influence of our socio-economic factors on changes in eating habits, the following test predictors were included in our GLMMs: the interaction of time period (i.e., before vs. during the Ebola crisis) with: (1) monthly income, (2) number of years of education, and (3) literacy, as well as (4) time period alone (Table 2). The interaction was included for each test predictor because we hypothesized that the influence of the time period on our response variables would vary depending on each socio-economic factor. We included time period alone as a test predictor to confirm that changes in eating habits that could not be explained by the other test predictors were associated with the incidence of the EVD outbreak.
Their measurements, the type of data, data sources, the names of the models they were included in, the hypothesized effects on the response variables, and references to studies that have examined their effects before.
To control for other potential effects that may influence dietary changes, fixed-effects control predictors were included (Table 2). Household ID, location ID, interviewer ID and interview date were included as random-effects control predictors to account for variance clustered in these groups .
Literacy and level of education were included as test predictors in our GLMMs because it has been shown that education may have an effect on people’s dietary choices . According to the Human Development Index Report, the expected number of years of schooling for Liberians is 9.5, and adult literacy is under 43% ; we therefore included literacy in addition to education, in case the variability of the latter was not large enough. We expected bushmeat consumption to decrease during the Ebola crisis in households with more educated and/or literate household heads, because they may be more knowledgeable about the risks of consuming bushmeat.
Assuming that bushmeat represented the main source of animal protein in rural households before the Ebola outbreak due to its relatively low cost , we predicted that wealthier households would more likely decrease their consumption of bushmeat during the crisis compared to poorer households, because they could afford alternative sources of animal protein. Furthermore, if high-income households were able to replace bushmeat with alternative meat sources, then they would be less likely to experience a reduction in their number of meals per day or in the diversity of food items they consumed.
To test whether the Ebola crisis had an influence on meat prices, time period was included as a test predictor in our models. Household ID, location ID and type of meat (i.e. animal species) were included as random-effects control predictors in our models for changes in local bushmeat and domestic meat prices. Prices were collected for five different types of domestic meat, and ten different types of bushmeat (S1 Appendix). For our model assessing the change in local fish prices, location ID was included as a random-effect control predictor (type of meat and household ID were not included in the model because only prices for one type of meat were available, i.e., fish).
We analyzed the data using GLMMs and LMMs. All models were fitted in R (version 3.2.2, ) using the following functions of the lme4 package : ‘glmer’ for GLMMs, ‘lmer’ for LMMs, and ‘glmer.nb’ for a GLMM with a negative binomial error structure. All possible random slopes components were included in the models to keep type I error rate at the nominal level of 5% [35, 55]. Numeric predictor variables were z-transformed before running the models. Collinearity among the predictor variables was assessed by computing Variance Inflation Factors (VIFs; ) using the function ‘vif’ of the package ‘car’ .
To test the overall effect of our test predictors, we compared full models to null models using likelihood ratio tests  in which the test predictors were omitted from the null models, and all other control fixed effects, random effects and random slopes were the same as in the full models . Non-significant interactions (including the test predictors that were included as interactions) were dropped from the full model before running the final model. The significance of each test predictor was assessed by using a likelihood ratio test  that compared the final model to a reduced model by dropping one test predictor at a time using the R function ‘drop1’. For details on model stability, see S1 Appendix.
Number of meals per day.
We fitted a GLMM with a Poisson error structure and log link function to investigate the change in number of meals per day. The sample size for this model was 399 households from 80 locations. Collinearity was not an issue (maximum VIF: 1.966).
We fitted two GLMMs with Poisson error structure and log link function to analyze the changes in (1) frequency of bushmeat consumption, and (2) the proportion of people in the community who preferred to eat bushmeat. The sample size for the bushmeat consumption frequency model was 277 households from 75 locations, and the sample size for the model analyzing the proportion of the community that preferred bushmeat was 267 households from 75 locations. Collinearity was not an issue (maximum VIF for bushmeat consumption frequency model: 1.929; maximum VIF for proportion of the community that preferred bushmeat model: 1.872).
We fitted two GLMMs with Poisson error structure and log link function to analyze the changes in (1) the number of different food items consumed, and (2) the number of different food groups consumed. The sample size for both models was 267 households from 73 locations. Collinearity was not an issue (maximum VIF: 1.808 for both models). With these models we analyzed one aspect of food diversity, which was the number of food items and food groups consumed, but not the change in specific foods; hence, to further investigate changes in food diversity, we also used descriptive statistics to compare the proportions of respondents that consumed each food item before and during the crisis, as well as the food groups to which these items belonged.
We fitted three GLMMs with binomial error structure and logit link function to analyze the changes in the likelihood of preferring to eat three of the most frequently mentioned bushmeat species: (1) duiker, (2) monkey, and (3) pangolin. Time period was included as a test predictor and location ID as a random-effect control predictor in the models (Table 2). Monthly income, level of education, and literacy were not included as test predictors in these models because the 2010–2012 survey did not gather data on these variables. The sample size for the models was 271 households from 58 locations.
We ran two LMMs to assess whether local bushmeat and fish prices changed during the crisis. The sample size for the bushmeat prices model was 146 households from 18 locations, and included the bushmeat prices for 10 species (see S1 Appendix). The sample size for the local fish prices model was 591 households from 55 locations. The assumptions of normally distributed and homogeneous residuals were checked by visually examining a QQ-plot and the residuals plotted against the fitted values. To allow for a likelihood ratio test, the models were fitted using Maximum Likelihood (ML) rather than Restricted Maximum Likelihood (REML, ).
To test whether domestic meat prices changed, we ran a GLMM with a negative binomial error structure and log link function. The sample size for the model was 650 households from 58 locations, and prices were specified for 5 types of meat (see S1 Appendix). A GLMM with a negative binomial distribution was chosen based on its ability to deal with overdispersion . To fit this model, prices were converted into count data by rounding them to the nearest integer.
Number of meals per day
The full-null model comparison was significant (likelihood ratio test [LRT]: χ2 = 11.138, df = 4, p = 0.025). The number of meals that respondents consumed per day decreased significantly during the Ebola (LRT: χ2 = 10.369, df = 1, p = 0.001). Income, literacy, and the respondent’s level of education did not influence the response (Table 3, Fig 2).
The bold lines show the medians, which are equal to the first quartiles for before and during the Ebola crisis, and also equal to the minimum value during the crisis. The dashed lines depict the expected values based on the model.
The full model testing for changes in frequency of bushmeat consumption fitted the data significantly better than its corresponding null model (LRT: χ2 = 21.029, df = 4, p <0.001). The interaction of time period with income was a trend (LRT: χ2 = 3.119, df = 1, p = 0.077; Table 4, Fig 3). In addition, the control predictor for local bushmeat prices was a trend (LRT: χ2 = 3.813, df = 1, p = 0.051; Table 4), indicating a decrease in bushmeat consumption frequency where local bushmeat prices were high. Only the control predictor for perceived risk of bushmeat consumption had a significant influence on the frequency of bushmeat consumption; households were more likely to decrease their consumption frequency if the household head believed that Ebola could be contracted from bushmeat (LRT: χ2 = 8.731, df = 1, p = 0.003; Table 4).
The size of each circle corresponds to the proportion of households and the dashed lines depict the fitted regressions for each time period.
We found a similar pattern regarding changes in the proportion of the community that preferred to eat bushmeat. There was a significant difference between the full model and its respective null model (LRT: χ2 = 12.964, df = 4, p = 0.012). The interaction of time period with income had a significant effect (LRT: χ2 = 4.073, df = 1, p = 0.044; Table 5, Fig 4). Heads of low-income households thought that a smaller proportion of their community continued to prefer eating bushmeat during the crisis, whereas heads of high-income households perceived a smaller decrease in bushmeat consumption in their community.
The size of each circle corresponds to the proportion of households and the dashed lines depict the fitted regressions for each time period.
Both full models testing for changes in food diversity were not different from their respective null models (change in number of food items consumed, LRT: χ2 = 3.841, df = 4, p = 0.428; change in number of food groups consumed, LRT: χ2 = 2.411, df = 4, p = 0.661). As a complement to our models, we used descriptive statistics to further investigate changes in food diversity, and we found differences in the proportions of individual food items and food groups that were typically consumed in a meal before and during the crisis. Notably, the consumption of bushmeat dropped from 81% to 16.5%, while chicken and fish consumption increased from 11.3% to 46.3% and from 47.1% to 86.5% respectively (Table 6, Fig 5).
None of the three full models were different from their respective null models (preference for monkey meat, LRT: χ2 = 2.317, df = 1, p = 0.128; duiker meat, LRT: χ2 = 0.525, df = 1, p = 0.469; and pangolin meat, LRT: χ2 = 0.106, df = 1, p = 0.74). Thus, the likelihood of respondents preferring to eat these three species during the Ebola crisis did not change.
The full model for changes in local bushmeat prices was not significantly different from the null model (LRT: χ2 = 2.260, df = 1, p = 0.133); hence, prices for bushmeat did not significantly change during the Ebola crisis. Fish prices significantly increased during the crisis (Table 7) by 19.5%, from an estimated 71.9 to 85.9 Liberian dollars per piece (LRT: χ2 = 4.804, df = 1, p = 0.028).
Likewise, domestic meat prices increased during the crisis (Table 7) by 34%, from 3757.0 to 5050.0 Liberian dollars per body (LRT: χ2 = 8.864, df = 1, p = 0.003).
People in Liberia consumed less bushmeat during the EVD crisis than before. However, our study suggests that wealthier people reduced their consumption of bushmeat less than those with a lower income. People also suffered from food shortages as they ate fewer meals per day during, as compared to before the crisis. Although the diversity of food items that made up people’s meals did not change, it seems likely that the items themselves changed and bushmeat was mainly replaced by chicken and fish. However, people’s preference for specific bushmeat species remained the same.
There was a clear effect of the Ebola crisis on daily meal frequency, which decreased across all levels of income and education. Several factors could have contributed to this food shortage; quarantine measures and border closures greatly restricted the movement of people and goods, disrupting agricultural activities and businesses, leading to aggravated food insecurity [47, 62]. Furthermore, the areas with a high incidence of EVD infections were also the most productive regions in the country, where a shortage of labor during the crisis caused a drop in both food and cash-crop production . Consequently, household incomes were negatively impacted and food accessibility was further inhibited . In addition, the limited distribution of imported foods from Monrovia’s sea-port to rural markets during the crisis resulted in price increases of some food items .
Our study shows that the frequency of bushmeat consumption during the Ebola crisis was influenced by household income. In addition, we found in our control predictors an influence of (1) bushmeat prices, and (2) the perceived risk of bushmeat consumption. We interpreted these as two components of the costs associated with bushmeat consumption: (1) the monetary costs and (2) the health risk costs. Bushmeat price is therefore not only monetary, but also contains a health component. Both poor and rich households were subject to health risk costs, and people were therefore more likely to consume bushmeat less frequently if they thought that EVD could be contracted from bushmeat. This may partly explain the overall decrease in bushmeat consumption during the crisis in Liberia across all levels of household income. However, the monetary costs had a greater influence on the consumption habits of poor households, which were more affected by the prices of bushmeat compared to wealthy households, as the latter could compensate higher prices with higher incomes. This coincides with the economic law of demand, which predicts that the demand for a good decreases if the price for it increases .
The demand for bushmeat is partially driven by bushmeat prices and the prices of similar substitutes . This dynamic is also supported by our model results on individually reported bushmeat consumption frequency: bushmeat consumption decreased with increasing prices. To further investigate the changes in bushmeat consumption, we therefore tested whether prices for domestic meats and bushmeat changed between the two surveys. We found that fish and domestic meats became more expensive during the crisis, which can be explained by an increase in their demand (reflected in our food diversity analysis) but a decrease or stability in their production , and possibly also a limited accessibility to them due to the travel restrictions during the crisis  and the high costs of transportation to some parts of the country . In contrast, our results indicated that prices for bushmeat remained stable during the crisis. This finding is puzzling and does not match the model results or theoretical predictions. However, a likely explanation is that this price stability was caused by opposing trends in bushmeat demand and bushmeat hunting that may have occurred as a result of the EVD crisis. A very likely decrease in hunting rates during the crisis would result in a reduced supply, which would presumably increase bushmeat prices; however, the much lower demand for bushmeat during the crisis (demonstrated by the decrease in bushmeat consumption) may have counteracted this effect, resulting in price stability. An additional explanation may be that our sample size on bushmeat prices was small. This may have complicated the detection of a change in bushmeat prices.
Our analysis for the change in the proportion of people in the community that preferred to eat bushmeat showed a similar effect of income on bushmeat consumption. However, this model was problematic due to the difference in scale of the response and the predictors; i.e., the response was at the community level, but the predictors were at the household level. Hence, our interpretation of this model is that the perception of respondents about the consumption of bushmeat in the community was influenced by their income level. We do not intend to interpret the other predictors in the model due to the described problem of scale in the model.
The number of food groups and food items consumed remained constant during the crisis; nevertheless, there were differences in diet composition, reflected by the proportions of different food items and food groups consumed. Our results suggest that chicken and fish were important substitutes for bushmeat for Liberians during the Ebola crisis.
The drop in bushmeat consumption during the crisis does not imply that people did not like bushmeat anymore; indeed, the likelihood of respondents preferring to eat duiker, pangolin or monkey meat did not change. This finding indicates that factors such as taste preference or tradition may play an influential role in human dietary choices. This may have important implications for future conservation management. For example, if taste preference is a major driver of bushmeat consumption, the demand for bushmeat may be less responsive to price changes, and finding an acceptable substitute may be difficult . Furthermore, due to the consistent preference for bushmeat during the crisis, people may continue to consume it as soon as the memories of the impacts of the crisis have faded sufficiently.
Implications for conservation
The overall decrease in bushmeat consumption associated with the Ebola crisis may have had a short-term positive effect on vulnerable wildlife populations. Needless to say, however, that this should not advocate the use of fear for the disease as a medium for accomplishing conservation goals . It is problematic to suggest that the epidemic presented a “silver lining” for conservation  because of the catastrophic impacts on human livelihoods and food security. Furthermore, spreading fear about the disease could backfire in that this may lead to attempts to eradicate the vectors of the deadly disease .
Conservation efforts should instead focus on developing strategies that are compatible with human livelihoods and food security . Due to the complexity and variability of bushmeat consumption drivers that need to be addressed, multiple interventions may be required [51, 70]. Based on our results, conservation strategies that aim to reduce bushmeat consumption in Liberia may be more effective by making a distinction between the consumption patterns of high-income households to those of low-income households. In addition, although we did not find an effect of literacy or education on bushmeat consumption during the Ebola crisis, environmental education should not be disregarded as an important component in conservation strategies, as it has been shown to correlate with environmental health [19, 71]. Indeed, household heads were more likely to consume bushmeat less frequently if they thought that Ebola could be contracted from bushmeat, suggesting that knowledge about the disease had an impact on bushmeat consumption patterns. This also means that there is a difference in the effect of general education and specific knowledge on human behavioural patterns.
Law enforcement is often called for as an indispensable means of mitigating the illegal bushmeat trade (e.g., [70, 72, 73]). The enforcement of laws that prohibit the sale and consumption of protected and endangered species in urban markets is crucial for reducing the demand for bushmeat of high-income, urban households . However, a complete ban on bushmeat is unrealistic, given that poor households that rely heavily on bushmeat as a source of nutrition will be negatively impacted; especially if alternative sources of protein are not provided [49, 70, 74]. It is therefore important to distinguish between resilient species that may still be hunted sustainably from those that are too vulnerable to be harvested .
Providing alternative income and protein sources could reduce the reliance on bushmeat of low-income households [19, 70, 75]. Similar to other studies [17, 19, 76] our results indicate that fish is an important alternative protein source to bushmeat. However, if fish represents a direct replacement for bushmeat, it is necessary to improve the management of domestic fisheries to help increase the sustainability of fish stocks [76,77]. Similarly, meat from domestic animals may be an acceptable replacement for bushmeat; however, the negative environmental impacts associated with increased livestock production must be reduced through proper management [70, 77]. Furthermore, it is also important to secure the availability of staple foods such as grains, roots and tubers throughout the year  and expand the use of plant proteins, such as dried beans and other pulses, which have a long shelf life when stored properly and could provide a readily-available source of protein in times of crisis .
S1 Appendix. Details on data preparation for response and predictor variables, and model stability.
S2 Appendix. Complete questionnaires used during the 2010–212 and 2015 Liberia nationwide surveys.
S1 Data. The dataset contains all interview data used in our analysis that were collected from 623 Liberian households during the 2015 survey.
S2 Data. The dataset contains all interview data used in our analysis that were collected from 275 Liberian households during the 2010–2012 survey.
We thank the Forestry Development Authority, the Wild Chimpanzee Foundation, Fauna & Flora International, for their support and collaboration. We also thank Roger Mundry for statistical support, as well as all survey team members who collected the data, and Madlen Schliephake and Enrique Sulbaran Romero for their help entering the data. For their valuable comments on the manuscript, we thank Mona Bachman and Sergio Marrocoli.
- Conceptualization: ION LB TGa CS HSK JJ.
- Data curation: ION.
- Formal analysis: CS ION.
- Funding acquisition: HSK JJ.
- Investigation: TGr CT ML.
- Methodology: MA TGa TGr CT ML CS HSK JJ.
- Resources: HSK JJ.
- Supervision: HSK JJ.
- Writing – original draft: ION.
- Writing – review & editing: MA LB TGa CS HSK JJ.
- 1. Johnson KM, Lange JV, Webb AP, Murphy FA. Isolation and partial characterisation of a new virus causing acute haemorrhagic fever in Zaire. Lancet. 1977; 1:569–571. pmid:65661
- 2. Mari Saez A, Weiss S, Nowak K, Lapeyre V, Zimmermann F, Dux A, et al. Investigating the zoonotic origin of the West African Ebola epidemic. EMBO Molecular Medicine. 2015; 7(1):17–23.
- 3. Baize S, Pannetier D, Oestereich L, Rieger T, Koivogui L, Magassouba N, et al. Emergence of Zaire Ebola virus disease in Guinea—preliminary report. New England Journal of Medicine. 2014; 371:1418–1425. doi: 10.1056/NEJMoa1404505. pmid:24738640
- 4. World Health Organization (WHO). Situation report. 2016b. http://apps.who.int/iris/bitstream/10665/208883/1/ebolasitrep_10Jun2016_eng.pdf?ua=1
- 5. Leroy EM. Multiple Ebola Virus Transmission Events and Rapid Decline of Central African Wildlife. Science. 2004;303(5656):387–90. doi: 10.1126/science.1092528. pmid:14726594
- 6. Judson SD, Fischer R, Judson A, Munster VJ. Ecological contexts of index cases and spillover events of different Ebola viruses. PLOS Pathogens. 2016; 12(8):e1005780. doi: 10.1371/journal.ppat.1005780. pmid:27494600
- 7. Fallah MP, Skrip LA, Gertler S, Yamin D, Galvani AP. Quantifying poverty as a priver of Ebola transmission. PLOS Neglected Tropical Diseases. 2015; 9(12):e0004260. doi: 10.1371/journal.pntd.0004260. pmid:26720278
- 8. Bausch DG, Schwarz L. Outbreak of Ebola virus disease in Guinea: where ecology meets economy. PLOS Neglected Tropical Diseases. 2014; 8(7):e3056. doi: 10.1371/journal.pntd.0003056. pmid:25079231
- 9. Hotez PJ. Neglected Tropical Diseases in the Anthropocene: The Cases of Zika, Ebola, and Other Infections. PLOS Neglected Tropical Diseases. 2016; 10(4): e0004648. doi: 10.1371/journal.pntd.0004648. pmid:27058728
- 10. World Health Organization (WHO). Ebola virus disease. Fact sheet No. 103. 2016a. http://www.who.int/mediacentre/factsheets/fs103/en/
- 11. Alexander KA, Sanderson CE, Marathe M, Lewis BL, Rivers CM, Shaman J, et al. What factors might have led to the emergence of Ebola in West Africa? PLOS Neglected Tropical Diseases. 2015; 9(6):e0003652. doi: 10.1371/journal.pntd.0003652. pmid:26042592
- 12. Pigott DM, Golding N, Mylne A, Huang Z, Henry AJ, Weiss DJ, et al. Mapping the zoonotic niche of Ebola virus disease in Africa. eLife. 2014; 3.
- 13. Mann E, Streng S, Bergeron J, Kircher A. A review of the role of food and the food system in the transmission and spread of Ebola virus. PLOS Neglected Tropical Diseases. 2015; 9(12):e0004160. doi: 10.1371/journal.pntd.0004160. pmid:26633305
- 14. Davies G. Bushmeat and international development. Conservation Biology. 2002;16(3):587–9.
- 15. Milner-Gulland EJ, Bennett EL. Wild meat: the bigger picture. Trends in Ecology & Evolution. 2003; 18(7):351–7.
- 16. Bowen-Jones E, Brown D, Robinson EJZ. Economic commodity or environmental crisis? An interdisciplinary approach to analysing the bushmeat trade in central and West Africa. Area. 2003; 35(4):390–402.
- 17. Brashares JS. Bushmeat hunting, wildlife declines, and fish supply in West Africa. Science. 2004; 306(5699):1180–3. doi: 10.1126/science.1102425. pmid:15539602
- 18. Fa John, Nasi Robert. The role of bushmeat in food security and nutrition. 2015.
- 19. Junker J, Boesch C, Mundry R, Stephens C, Lormie M, Tweh C, et al. Education and access to fish but not economic development predict chimpanzee and mammal occurrence in West Africa. Biological Conservation. 2015; 182:27–35.
- 20. Nasi R, Taber A, Van Vliet N. Empty forests, empty stomachs? Bushmeat and livelihoods in the Congo and Amazon Basins. International Forestry Review. 2011; 13(3):355–68.
- 21. Loibooki M, Hofer H, Campbell KLI, East ML. Bushmeat hunting by communities adjacent to the Serengeti National Park, Tanzania: the importance of livestock ownership and alternative sources of protein and income. Environmental Conservation [Internet]. 2002; 29(3). Available from: http://www.journals.cambridge.org/abstract_S0376892902000279
- 22. Schulte-Herbrüggen B, Cowlishaw G, Homewood K, Rowcliffe JM. The Importance of Bushmeat in the Livelihoods of West African Cash-Crop Farmers Living in a Faunally-Depleted Landscape. Noë R, editor. PLOS ONE. 2013; 8(8):e72807. doi: 10.1371/journal.pone.0072807. pmid:23977355
- 23. Jeffries MJ. Biodiversity and conservation. 2nd ed. London; New York: Routledge; 2006.
- 24. Craigie ID, Baillie JEM, Balmford A, Carbone C, Collen B, Green RE, et al. Large mammal population declines in Africa’s protected areas. Biological Conservation. 2010; 143(9):2221–8.
- 25. Bennett E, Eves H, Robinson J, Wilkie D. Why is eating bushmeat a biodiversity crisis? Conservation in Practice. 2002; 3:28–29.
- 26. Redford KH. The empty forest. Bioscience. 1992; 42(6):412–422.
- 27. Tweh CG, Lormie MM, Kouakou CY, Hillers A, Kühl HS, Junker J. Conservation status of chimpanzees Pan troglodytes verus and other large mammals in Liberia: a nationwide survey. Oryx. 2014; 1–9.
- 28. Abernethy KA, Coad L, Taylor G, Lee ME, Maisels F. Extent and ecological consequences of hunting in Central African rainforests in the twenty-first century.
- 29. Effiom EO, Nunez-Iturri G, Smith HG, Ottosson U, Olsson O. Bushmeat hunting changes regeneration of African rainforests. Proceedings of the Royal Society B: Biological Sciences. 2013; 280(1759):20130246–20130246.
- 30. Berkes F, Colding J, Folke C, editors. Navigating social-ecological systems: building resilience for complexity and change. Cambridge; New York: Cambridge University Press; 2003.
- 31. World Bank Group (WBG). Update on the economic impacts of the 2014 Ebola epidemic on Liberia, Sierra Leone, and Guinea. 2014. http://www.worldbank.org/content/dam/Worldbank/document/Economic%20Impact%20Ebola%20Update%202%20Dec%202014.pdf
- 32. United Nations Development Program (UNDP). Human development report, Liberia. 2015a. http://hdr.undp.org/sites/all/themes/hdr_theme/country-notes/LBR.pdf
- 33. Jenkins CN, Pimm SL, Joppa LN. Global patterns of terrestrial vertebrate diversity and conservation. Proceedings of the National Academy of Sciences. 2013; 110(28):E2602–E2610.
- 34. Baayen R. H. Analyzing linguistic data: a practical introduction to statistics using R. New York: Cambridge University Press; 2008.
- 35. Barr DJ, Levy R, Scheepers C, Tily HJ. Random effects structure for confirmatory hypothesis testing: Keep it maximal. Journal of Memory and Language. 2013; 68(3):255–78.
- 36. United Nations Development Program (UNDP). Human development indicators. 2015b. http://hdr.undp.org/en/countries/profiles/LBR
- 37. Albrechtsen L, Fa JE, Barry B, Macdonald DW. Contrasts in availability and consumption of animal protein in Bioko Island, West Africa: the role of bushmeat. Environmental Conservation. 2005; 32(4):340.
- 38. Fa JE, Albrechtsen L, Johnson PJ, Macdonald DW. Linkages between household wealth, bushmeat and other animal protein consumption are not invariant: evidence from Rio Muni, Equatorial Guinea. Animal Conservation. 2009; 12(6):599–610.
- 39. Faber M, Schwabe C, Drimie S. Dietary diversity in relation to other household food security indicators. International Journal of Food Safety, Nutrition and Public Health. 2009; 2(1):1.
- 40. Kamins AO, Rowcliffe JM, Ntiamoa-Baidu Y, Cunningham AA, Wood JLN, Restif O. Characteristics and Risk Perceptions of Ghanaians Potentially Exposed to Bat-Borne Zoonoses through Bushmeat. EcoHealth. 2015; 12(1):104–20. doi: 10.1007/s10393-014-0977-0. pmid:25266774
- 41. LeBreton M, Prosser AT, Tamoufe U, Sateren W, Mpoudi-Ngole E, Diffo JLD, et al. Patterns of bushmeat hunting and perceptions of disease risk among central African communities. Animal Conservation. 2006; 9(4):357–63.
- 42. Gandiwa E., Zisadza-Gandiwa P., Mango L., and Jakarasi J.. Law enforcement staff perceptions of illegal hunting and wildlife conservation in Gonarezhou National Park, southeastern Zimbabwe. Tropical Ecology. 2014; 55(1):119–127.
- 43. Brashares JS, Golden CD, Weinbaum KZ, Barrett CB, Okello GV. Economic and geographic drivers of wildlife consumption in rural Africa. Proceedings of the National Academy of Sciences. 2011; 108(34):13931–6.
- 44. Wolfe ND, Daszak P, Kilpatrick AM, Burke DS. Bushmeat hunting, deforestation, and prediction of zoonotic disease. Emerging Infectious Diseases. 2005; 11(12):1822–7. doi: 10.3201/eid1112.040789. pmid:16485465
- 45. Fa JE, Olivero J, Farfán MÁ, Márquez AL, Duarte J, Nackoney J, Hall A, Dupain J, Seymour S, Johnson PJ, Macdonald DW, Real R, Vargas JM. Correlates of bushmeat in markets and depletion of wildlife. Conservation Biology. 2015; 29(3):805–815. doi: 10.1111/cobi.12441. pmid:25580729
- 46. Alexander JS, McNamara J, Rowcliffe JM, Oppong J, Milner-Gulland EJ. The role of bushmeat in a West African agricultural landscape. Oryx. 2015; 49(4):643–651.
- 47. World Bank Group (WBG). The Socio-Economic Impacts of Ebola in Liberia. Results from a high frequency cell phone survey, Round 3. 2015. http://www.worldbank.org/content/dam/Worldbank/document/Poverty%20documents/Socio-Economic%20Impacts%20of%20Ebola%20in%20Liberia,%20Jan%2012%20%28final%29.pdf
- 48. Koppert G. J. A., Dounias E., Froment A., and Pasquet P.. 1993. Food consumption in three forest populations of the southern coastal areas of Cameroon: Yassa-Mvae-Bakola. In: Tropical Forests, People and Food. Biocultural Interactions and Applications to Development. Man and Biosphere Series No. 13. Paris: UNESCO/ The Parthenon Publishing Group.
- 49. Foerster S, Wilkie DS, Morelli GA, Demmer J, Starkey M, Telfer P, et al. Correlates of bushmeat hunting among remote rural households in Gabon, Central Africa. Conservation Biology. 2012; 26(2):335–44. doi: 10.1111/j.1523-1739.2011.01802.x. pmid:22182047
- 50. United Nations Statistics Development (UNSD). Population Density and Urbanization. 2016. http://unstats.un.org/unsd/demographic/sconcerns/densurb/densurbmethods.htm
- 51. Friant S, Paige SB, Goldberg TL. Drivers of Bushmeat hunting and perceptions of zoonoses in Nigerian hunting communities. PLOS Neglected Tropical Diseases. 2015; 9(5):e0003792. doi: 10.1371/journal.pntd.0003792. pmid:26001078
- 52. Wilkie DS, Godoy RA. Income and price elasticities of bushmeat demand in lowland Amerindian societies. Conservation Biology. 2001; 15(3):761–9.
- 53. R Core Team. R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing; 2015. https://www.R-project.org/
- 54. Bates D, Mächler M, Bolker B, Walker S. Fitting Linear Mixed-Effects Models Using lme4. Journal of Statistical Software. 2015; 67(1). Available from: http://www.jstatsoft.org/v67/i01/
- 55. Schielzeth H, Forstmeier W. Conclusions beyond support: overconfident estimates in mixed models. Behavioral Ecology. 2009; 20(2):416–20. doi: 10.1093/beheco/arn145. pmid:19461866
- 56. Field A. Discovering statistics using SPSS for Windows. London: Sage; 2005.
- 57. Fox J, Weisberg S, Fox J. An R companion to applied regression. 2nd ed. Thousand Oaks, Calif: SAGE Publications; 2011.
- 58. Dobson A. J. An introduction to Generalized Linear Models. Boca Raton, Florida: Chapman & Hall/CRC; 2002.
- 59. Forstmeier W, Schielzeth H. Cryptic multiple hypotheses testing in linear models: overestimated effect sizes and the winner’s curse. Behavioral Ecology and Sociobiology. 2011; 65(1):47–55. doi: 10.1007/s00265-010-1038-5. pmid:21297852
- 60. Bolker BM, Brooks ME, Clark CJ, Geange SW, Poulsen JR, Stevens MHH, et al. Generalized linear mixed models: a practical guide for ecology and evolution. Trends in Ecology & Evolution. 2009; 24(3):127–35.
- 61. Hilbe JM. Negative binomial regression. 2nd ed. Cambridge, UK; New York: Cambridge University Press; 2011.
- 62. United Nations Development Program (UNDP). Socio-economic Impact of Ebola Virus Disease in West African Countries. 2015c. http://www.africa.undp.org/content/dam/rba/docs/Reports/ebola-west-africa.pdf
- 63. Food and Agriculture Organization of the United Nations (FAO). Global Information and Early Warning System on Food and Agriculture, special report No. 333. 2014. http://www.fao.org/3/a-i4003e.pdf
- 64. Food and Agriculture Organization of the United Nations (FAO). Global Information and Early Warning System on Food and Agriculture Country Briefs. 2016a. http://www.fao.org/giews/countrybrief/country.jsp?code=LBR
- 65. Marshall A. Principles of Economics. London: MacMillan; 1890.
- 66. Action Against Hunger International (ACF). Ebola in Liberia: impact on food security and livelihoods. 2015. https://www.actionagainsthunger.org.uk/sites/default/files/publications/ebola_in_liberia_impact_on_food_security_and_livelihoods.pdf
- 67. Schenck M, Nsame Effa E, Starkey M, Wilkie D, Abernethy K, Telfer P, et al. Why People Eat Bushmeat: Results from two-choice, taste tests in Gabon, Central Africa. Human Ecology. 2006; 34(3):433–45.
- 68. Pooley S, Fa JE, Nasi R. No conservation silver lining to Ebola. Conservation Biology. 2015; 29(3):965–7. doi: 10.1111/cobi.12454. pmid:25581265
- 69. Williams T. Ebola’s silver lining: we can clamp down on bushmeat. New Scientist. 2014; 2985. https://www.newscientist.com/article/mg22329850-300-ebolas-silver-lining-we-can-clamp-down-on-bushmeat/
- 70. Lindsey PA, Balme G, Becker M, Begg C, Bento C, Bocchino C, et al. The bushmeat trade in African savannas: Impacts, drivers, and possible solutions. Biological Conservation. 2013; 160:80–96.
- 71. De Boer WF, van Langevelde F, Prins HHT, de Ruiter PC, Blanc J, Vis MJP, et al. Understanding spatial differences in African elephant densities and occurrence, a continent-wide analysis. Biological Conservation. 2013; 159:468–76.
- 72. Critchlow R, Plumptre AJ, Alidria B, Nsubuga M, Driciru M, Rwetsiba A, Wanyama F, Beale CM. Improving law-enforcement effectiveness and efficiency in protected areas using ranger-collected monitoring data. Conservation Letters. 2016; 00(0):1–9.
- 73. Ripple WJ, Abernethy K, Betts MG, Chapron G, Dirzo R, Galetti M, Levi T, Lindsey PA, Macdonald DW, Machovina B, Newsome TM, Peres CA, Wallach AD, Wolf C, Young H. Bushmeat hunting and extinction risk to the world's mammals. Royal Society Open Science. 2016; 3:160498. doi: 10.1098/rsos.160498. pmid:27853564
- 74. Challender DWS, MacMillan DC. Poaching is more than an enforcement problem. Conservation Letters. 2014; 7(5):484–494.
- 75. Schoneveld-de Lange N, Meijaard E, Löhr A. South to south learning in great ape conservation: South to South Learning for Great Apes. American Journal of Primatology. 2016; 78(6):669–78. doi: 10.1002/ajp.22531. pmid:26828200
- 76. Wilkie DS, Starkey M, Abernethy K, Effa EN, Telfer P, Godoy R. Role of Prices and Wealth in Consumer Demand for Bushmeat in Gabon, Central Africa. Conservation Biology. 2005; 19(1):268–74.
- 77. Rentsch D, Damon A. Prices, poaching, and protein alternatives: An analysis of bushmeat consumption around Serengeti National Park, Tanzania. Ecological Economics. 2013; 91(1):1–9.
- 78. Fao. Pulses (english). S.l.: Food & Agriculture Org. 2016. http://www.fao.org/3/a-i5528e.pdf
- 79. Christie T, Steininger MK, Juhn D, Peal A. Fragmentation and clearance of Liberia’s forests during 1986–2000. Oryx. 2007; 41(4): 539–543.