Factors influencing the pattern of wildlife product consumption in Indochina: case study of Cambodia

. This study determines socio-demographic factors and knowledge of consumers on wildlife animals to the pattern of wildlife consumption in Cambodia. A sample of 200 consumers from major markets in Phnom Penh (capital), Koh Kong (border with Gulf of Thailand), Stung Treng (border with Laos) Kratie and Mondulkiri (border with Vietnam). A structured questionnaire was used to collect information on gender, age, ethnicity, religion, level of education, occupation, income, wildlife consumption, frequency of wildlife consumption, kind of wildlife animal used for local/domestic or international trade, etc. The results of multinominal logistic regression showed that consumers' place of birth and education level have a significant relationship with the pattern of wildlife consumption. The results indicate that local consumers working for the government are more likely to buy wildlife products for special party than those working at other sectors, but they seem not to choose wildlife products as medicine. Local consumers working as trader are more likely to consume wildlife products for their family gathering. Regarding the use of wildlife products as medicine, local consumers working for government are less likely to consume wildlife products than those working for company/NGO. The findings suggest that the government should target local people and specify different propaganda for people with varying sectors of occupation to reduce the use of wildlife products.


Introduction
Demand for wild animals and their products has been increasing in many countries.Wild species are used as the source of a wide variety of goods, including foods, medicines, pets, displays, fashion and cultural items, industrial resins and extracts, and household items.Use may be local to the resource itself, for example, hunting for meat for direct consumption, or take place many thousands of miles away, the wildlife products passing along a complex processing and trade chain from harvester to end-consumer.
Southeast Asia is both a center for the consumption of wildlife products and a key supplier of wildlife products to the world.Cambodia is Southeast Asian country that acts as significant wildlife sources in trade.The trade involves a wide variety of native species, which, in many cases, are declining due to unsustainable and often illegal harvest .
Overexploitation to supply domestic and international trade has been recognized as the most significant threat to biodiversity in many Southeast Asian countries [2,15,22].Over the past decade, Southeast Asia has become an important center for regional wildlife trade activities.Wildlife and related products from Cambodia, Lao PDR, Myanmar, Thailand, Malaysia, and Indonesia are channeled through Vietnam to mainland China, Taiwan, Korea, Hong Kong, and Singapore [5,13,17,18,25].It is probably because wildlife and wildlife products can pass through the border region to China more easily with support from a wellestablished trade network [17,20].
Cambodia is one of the most biodiverse countries in South-east Asia [6], however, overexploitation of natural resources is widespread, and laws are weakly enforced [27].Cambodia has a high prevalence of hunting and reliance on wild meat, with an estimated 83 percent of rural households engaged in some form of harvest of wild animals at least once per year [14].
Cambodia has been considered a part of an international transit system for wildlife trade, such as elephant ivory, tiger's skin, and rhino horns from Africa and Asia to supply China, Vietnam, and Thailand [4,34].
In Southeast Asia, the demand of wildlife species consumption is increasingly as foods and medicines [16].There are perceived attributes of status conferred by consuming the rareness of meat and therefore the expensiveness of wild species making the meat of these species a high-end product [9].In contrasts, the situation in other tropical areas in Congo where bushmeat is a crucial source of protein for the poor who cannot pay for processing meat [32].In Cambodia, 89 percent of rural people obtain the recommended human protein intake through the consumption of fish, which is the most widely-consumed source of animal protein [33].Illegal wildlife trade has been increasing in Cambodia and threatening the native biodiversity [11,24].
Cambodia has emerged as a transit country for shipments of elephant ivory, rhino horn, pangolin scales and big cat bones, which are being trafficked by criminal groups from Africa to Asia.It is also a source country for many wildlife species, including ungulates, turtles, pangolins primates, cats and carnivores, that may be consumed domestically or trafficked across nearby international [29].
Rapid economic development in the region has also increased purchasing in Southeast Asian countries.Domestic markets have continued to expand to meet the soaring demand for bushmeat and wildlife products [15,22].It is expected to rise in the coming years as the domestic economy grows [30,31].Trade in wildlife involves many different taxonomic groups.Turtles, pangolins, and snakes have been most traded internationally.Other groups, which are often consumed domestically, include civet, muntjac, bear, primate, sambar, otter, and serow [17,31].
Nowadays, wildlife meat and products are consumed not only for the poor's survival but also used as a luxury food for city people [12,21].Wildlife trade has increased in the urban areas as the urban economy has developed and expanded [35].Wildlife consumption varies in regions, politics, and cultures [3].
A study the impact of income on wildlife consumption in Bolivia and found that income was positively correlated and strongly influenced by wildlife consumption [12].Another research studied the relationship between economics and characteristics of rural wildlife consumption in the United States, and the study showed that high income is associated with wildlife consumption [3].
A study in Hanoi, Vietnam, found that income, occupations, and gender are the most important relationships with wildlife consumption while age and education level have no significant relationship to the consumption of wildlife products [10].Another study to understand wildlife consumers in Ho Chi Minh City, Vietnam, showed that education levels and occupations have an important relationship with the consumption of wildlife meat [21].
The study on hunting for food and consumption of rats in Lao PDR stated that age, occupation, and ethnic group have statistically significant to consumption [28].In addition to socio-demographic factors, consumers' knowledge about wildlife conservation has a significant relationship with the consumption of wildlife [35].
Many issues related to the pattern of wildlife consumption in Indochina, such as high demand, low levels of awareness among consumers, and traditional beliefs, have not been elucidated.Therefore, the pattern, scale, and drivers of wildlife consumption should be well understood to improve and better target biodiversity conservation in Indochina, particularly in Cambodia.

Conceptual framework
Deciding to consume or trade wildlife animals or their products is often perceived as an individual's choice.While all such decisions are bounded by social context, they are probably even more for those whose society and culture frame many choices.
The theory of planned behavior (TPB) developed by Ajzen considers the individual's attitude, social norms, and perceived control as accurate predictors of behavioral intentions [1].Behavioral intention is then mainly affected by attitude, subjective norm, and perceived behavioral control; therefore, TPB is selected as the basic theoretical framework of this study.The theoretical framework of planned behavior is an open analytical framework.This study reviewed the existing literature, expanded on the theoretical framework of TPB, and constructed an analytical framework to study socioeconomic factors influencing the consumption or trade of wildlife meat or its products.
The conceptual framework as follows is developed based on literature reviews and TPB.

Fig. 1. The relationship between independent and dependent variables
The conceptual framework describes the characteristics of respondents and their knowledge concerning the popularity level of wildlife consumption, rareness of wildlife species (independent variables) that may influence the pattern of wildlife products (dependent variable).
The description and measurement scale of variables have been given in the following table.

Knowledge of consumers on wildlife trade Experience in the past
A pattern of wildlife consumption

Questionnaire design and data collection
We follow Drury [9] in designing structured questionnaires for market surveys.The questionnaire contained mostly closed questions to complete in a relatively short period of time, allowing for larger overall sample sizes.Information such as kinds of bush meat consumption, socioeconomic backgrounds of consumers, and the scale of the trade were obtained.Structured questionnaires were used to survey major markets in Phnom Penh, Koh Kong, Stung Treng, Kratie, and Mondulkiri of Cambodia in 2019.Individuals were approached.A face-to-face interview was carried out to collect information on gender, age, ethnicity, religion, level of education, occupation, income, consumption of wildlife, frequency of wildlife consumption, the purpose of consumption, the market price for wildlife animals, kind of wildlife animal used for local/domestic or international trade.The responses also covered around ten animals traded in and from Cambodia under various markets.
We performed 200 interviews in total.All interviews were anonymous, and interviewees were asked for their verbal consent and told they could stop the interview at any time.
The questionnaire data were analyzed at an aggregate level to give a picture of the wildlife trade overall, and detailed case studies were produced for Lorises, Muntjacs, Turtles, Deer, Squirrel, Porcupine, Large bamboo rat, Bangal monitor, Snake, and Wild boar to elucidate the socioeconomic factors underlying the trade.
In addition to interviewing, we also gathered data through observation.The data are compared with those available in the literature to infer the trade trend through time.Our surveys will also help to understand the patterns, scales, and trends of wildlife trade in Cambodia.

Demographic characteristics of the sample
The individuals interviewed were equal for male and female, counting for 41.5 percent male and 51.5 percent of female respondents; the surveyed people were mainly under 30 years old (37%), and 31-40 years old was the leading group, accounting for 26.5 percent, while those 41-50 accounted for 20 percent, following by groups of over 51-year-old were 16.5 percent.
The most predominant occupation was as trader (45.5%).The second most common occupation was farmer/fisher/worker (29.5%).The third most common occupation was as government staff (16%).
Among the respondents, 40 percent was graduated secondary or high school.The proportion of primary schoolers was 39 percent, the proportion of bachelor's or higher degrees was 14.5 percent, and 6.5 percent of respondents having no education (illiterate).Regarding the yearly income of surveyed people, 39.5% of the respondents' income was around 2,100-5,000 USD (middle level) while 35.5% of them living in very low-income level (2,000 USD).Compared with the characteristics of Statistical Yearbook of Cambodia 2020 [25], the sample data of this study were similar in terms of age and education, but there were some inconsistencies in gender and income structure.That may reflect demographic differences among wildlife product consumers.

The relationship between general characteristics and the decision to buy wildlife product
The relationship between independent variables, namely socio-demographic characteristics of consumers, consumers' experience as when the last time they consumed wild meat or wildlife product, consumers' knowledge of the popularity of wildlife trade, on the rareness of species such as muntjac, lorises/monkey (independent variables), and their decision to buy wildlife product in what context (dependent variable).
A Chi-square test significant level at .001, .01,and .05 is used to examine the association between independent variables and dependent variable.From the results of the bivariate analyses above, a strong relationship between income, experience from the past, knowledge of the rareness of lorises, and buying wildlife products have been found in this study (p< .001).Other factors such as age, place of birth, occupation, knowledge of the popularity of wildlife products, and knowledge of the rareness of muntjac are statistically significant in buying wildlife products (p< .005).The result indicates that people at aged above 51 more likely to purchase wildlife products for Special Party (54.5%) than for other purposes, while many people aged under 30 buy wildlife products for Family gathering (47.3%), and 40% people aged 41~50 buy wildlife products as Medicine.Local people are more likely to buy wildlife products for Special party (48.8%), while people from other provinces buy that for Family gathering (50%).
Regarding occupation, people work as farmer/fisher/worker seem to buy wildlife products for Special party (52.5%).A few people working for Government buy wildlife products as medicine (9.4%), while many of them (59.4%)buy wildlife products for Special party.Traders are more likely to purchase wildlife products for Family gathering (40.6%) than other purposes.Similar occupation, the income of consumers also has a strong relationship with the pattern of wildlife trade.The proportion of high-income people buying wildlife products for their Special party is very high (76.5%),while 36.6 percent of lowincome people have the same decision.
Consumers' experience, how often they bought a wildlife product, has statistical significance with the pattern of wildlife products they buy recently.Around 60.9 percent of consumers consume wildlife products very often (several times a week) usually buy wildlife products for Special party.
Similarly, there is a significant relationship (p< .005) between consumers' knowledge about the popularity of wildlife products with buying wildlife products.Consumers who consume wildlife products for Special party are more likely to answer that wildlife product consumption becomes less popular in recent years at the local and national levels (50%).An in-depth interview explains that the reasons are the influence of Chinese and Vietnamese consumption in Cambodia.Even consuming wildlife products for Special party, many of them (51.7%)realized that lorises is becoming rarer in the market compared to a few years ago, while 75.8 percent of them answered that muntjac are more common in the market than a few years ago.

Multivariate results
Bivariate analyses of wildlife consumption patterns by socio-demographic characteristics are examined, as is the nature of the association among these factors and the pattern of wildlife trade.Several associations are significant in the bivariate analysis.However, the bivariate association does not present a strong relationship.Therefore, a multivariate analysis was also applied to specify which factors best explain and predict the pattern of wildlife meat and its products.
The multinomial logistic regression results predicting the pattern of wildlife product consumption are interpreted.The model for the dependent variable as buying wildlife products was fitted for all respondents.Coefficients, standard errors, and odds ratios are presented in the model.Moreover, multinomial logistic regression is also used for groups of independent variables to predict the change of probabilities of buying wildlife products.
The logistic regression model estimates a model of the form: where   is the estimated probability of a particular event occurring to an individual with a given set of characteristics     is a constant that defines the probability   for an individual with all   set to zero,   is the estimated coefficients Many individual characteristics appear to be associated with the pattern of wildlife product consumption.Thus, the multinominal regression model was used to examine the specific effect of independent variables on the pattern of wildlife consumption.All independent variables having significant association in the bivariate analysis were included in the model.
The results of multinominal logistic regression showed that consumers' age, place of birth, occupation, experience in past, and knowledge regarding the rareness of wildlife species have a significant relationship with the pattern of wildlife consumption.A separate model was run to test the significance of these independent variables on the pattern of wildlife consumption.Surprisingly, the significant relationship between dependent variable and income, experience in the past, knowledge on the rareness of muntjac disappeared.Occupation and place of birth remain the most effective factors in the model.
The final model, which included selected independent variables such as occupation and place of birth were analyzed to measure the strength of independent variables that influence the probability of buying wildlife products for specific purposes.The "Farmer/Fisher/Worker", "Trader", "Government staff", and "Origin" predictors are positive and significant association with the pattern of wildlife consumption.
Exponentiating a beta parameter provides the multiplicative effect of that predictor on the odds, controlling for the other variables.The formula for the probability itself is: Based on the results of multinomial logistic regression and the formula above, the predicted probability of wildlife consumption pattern by occupation for local consumers has been calculated.Among local consumers, people who work as farmer/fisher/worker have a 19.18 percent probability of buying wildlife products as medicine.In comparison, people who work as company or NGO staff and trader have a 52.41 percent and 21.26 percent probability of buying wildlife products for their medicine.Buying wildlife products for family gathering purpose is the least likely among local consumers who work for company/NGO compared to 37.56 percent probability in the group of traders.Local consumers work as farmer/fisher/worker have a 56.38 percent probability of consuming wildlife products for special party.It is much more than traders (41.18%) and company/NGO staff (28.56%).

Conclusion
Despite the biodiversity conservation programs implemented in Cambodia to improve the knowledge on wildlife protection in recent decades, the use of wildlife products remains high.While many factors contribute to biodiversity conservation, the use of wildlife products is widely recognized as a significant causal factor.This study has identified the sociodemographic characteristics of consumers that influence the pattern of wildlife product consumption as occupation and place of birth.
The results indicate that local consumers working for the government are more likely to buy wildlife products for special party than those working at other sectors, but they seem not to choose wildlife products as medicine.Local consumers working as trader are more likely to consume wildlife products for their family gathering.Regarding the use of wildlife products as medicine, local consumers working for government are less likely to consume wildlife products than those working for company/NGO.
The findings suggest that the government should target local people and specify different propaganda for people with varying sectors of occupation to reduce the use of wildlife products.

Table 1 .
Operational definition of variables

Table 2 .
Percentage distribution of independent variables by the dependent variable

Table 3 .
Multinomial logistic regression results: Factors influencing the pattern of wildlife consumption This parameter is set to zero because it is redundant Based on the Likelihood Ratio Test, we can say that the model containing the full set of predictor represents a significant improvement in fit relative to a null model [LRX 2 (8)=28.285,p<.001].The Pearson and Deviance Chi-square tests suggest a good fitting model (p=.687 and p=.522).

Table 4 .
Predicted probability of wildlife consumption pattern by Occupation of consumers "Origin"