Influencing Factors and Measurement of “Willingness to Accept” Living with Alligators in a Nature Reserve: A Case Study in National Chinese Alligator Nature Reserve, China

Abstract

The establishment of nature reserves is an important measure to protect the wild population of Chinese alligators. Due to the overlap of nature reserves and human living areas, there is a certain conflict between economic development and ecological protection. How to formulate a feasible eco-compensation scheme and accurately analyze the influencing factors of eco-compensation willingness is of great significance to alleviate the contradiction between local residents and Chinese alligators. In this study, the contingent valuation method (CVM) was used to measure and analyze the residents’ willingness to accept (WTA) at National Chinese Alligator Nature Reserve (NCANR) located in Anhui province, China. Furthermore, decision tree modeling and logistic regression were used to analyze the influencing factors of residents’ WTA in NCANR, which provides a new insight to the influencing factors of eco-compensation. The results indicate that: (1) 93% of the residents living in NCANR have a WTA compensation, and the amount of WTA is CNY 25,542 (USD 3659.36) per household per year; and (2) individual characteristics, ecological protection cognition and external impact affect the WTA, and external impact on local residents is the most important factor affecting WTA. Therefore, it is necessary to make differential compensation to ensure the fairness of eco-compensation. In addition, the propaganda of eco-compensation should be strengthened, and the boundary of NCANR needs to be further clarified. The sources of funds for eco-compensation are supposed to be broadened, and poverty alleviation can be combined with eco-compensation policies.

1. Introduction

Nature reserves (NRs) are geographical spaces that have been recognized and managed by legislation or other effective measures to safeguard nature and its accompanying ecosystem services [1]. Establishing NRs is the most widely accepted and effective strategy for conserving biodiversity and preventing the deterioration of ecological environment [2,3]. China is one of the major players in biodiversity conservation across the world [4]. In China, the first NR was established in 1956. By the end of 2018, China had established 2750 NRs, covering over 14.8% of the land area and protecting over 90% of the land’s natural ecosystem types in China, as well as approximately 89% of the national key protected wildlife species, and most importantly, natural relics [5]. Around the world, biodiversity conservation in NRs often comes at the expense of local communities’ interests, especially for rural communities in developing countries [6]. The management of NRs in China is mandatory, which limits the use of natural resources by local residents in a certain space. Since the use of resources in NRs is restricted, communities around NRs usually struggle with poverty, if there are no reasonable policy supports [7]. NRs have changed the traditional production activities and lifestyles in neighboring communities, and the conflict between development and ecological protection has become increasingly acute [8,9]. Many residents of NRs have a negative attitude towards the construction of NRs, because they believe that the economic benefits they directly derive from NRs are much lower than the economic costs they pay [10,11]. As the main body of biodiversity conservation, residents’ behaviors and attitudes are the key affecting factors [12]. Conservation objectives are nearly impossible to achieve without considering the needs and concerns of local residents [13]. How to address the contradiction between the management of NRs and the development of local communities is a severe challenge.

“Eco-compensation” is a unique term in China [14]. Although the terms “payment for ecological services” or “payment for environmental services (PES)” and “eco-compensation” are often used interchangeably, eco-compensation is a broader term that includes a growing range of payment approaches based on incentive mechanism [4,8,14]. Considering of ecological protection costs, opportunity costs and ecosystem service values, eco-compensation provides reasonable compensation to ecological protectors through transfer payments or market transactions [15]. In 2020, China’s government made it clear to establish a compensation mechanism that is compatible with the level of local economic development by 2022. At present, studies on eco-compensation of NRs are few and the development of eco-compensation theory of NRs lags behind the practice. NRs usually have multiple ecosystems, such as forests, wetlands, grasslands, rivers and watersheds, thus the eco-compensation of NRs is more complicated than for a single ecosystem [16]. China has made many attempts at eco-compensation to NRs, and these practices have recognized that low compensation levels and complex compensation policies with unclear compensation objects will compromise the efficiency of eco-compensation [17,18].

Chinese alligator is the only crocodile species distributed in China. Among 23 kinds of crocodiles, it is the smallest crocodile species in the world. The ancestors of Chinese alligator, one of the earliest “residents” on the earth, appeared in Mesozoic from Triassic to Cretaceous 200 million years ago, and once dominated the earth with dinosaurs. With the change of living environment, many reptiles, such as dinosaurs, have become extinct, while some crocodiles such as Chinese alligator, have miraculously survived because of their unique living habits and learning to live in water to adapt to environmental changes. They are called “living fossils” by people and have extremely high research value. Chinese alligator plays an important role at the top of the food chain in the ecological chain of wetlands, and has an important value and role in maintaining both the ecology of wetlands and the balance of species. In the 1950s and 1960s, Chinese alligators were widely distributed in southern Anhui. However, with the rapid development of China’s economy, the habitat of Chinese alligator is shrinking dramatically. Their wild population is also decreasing [19,20]. In 1999, there were less than 150 alligators in the wild, and the annual rate of decline was 4–6%. Chinese alligator has become one of the most endangered species in the world. NCANR is the last refuge and the only concentrated distribution area of wild Chinese alligators, which has irreplaceable significance for protecting and restoring the wild Chinese alligator population. Therefore, it is particularly important to develop a feasible ecological compensation scheme for NCANR.

Quantitative research on the amount of eco-compensation is one of the key issues [21]. The research methods mainly include the ecosystem service value method, opportunity cost method, and contingent valuation method (CVM) [22,23,24,25]. CVM is a typical declarative preference method, which is the most widely used to measure the economic value of environmental goods [26,27,28]. The core of the CVM is to directly investigate and inquire about the stakeholders’ willingness to pay (WTP) or WTA for ecosystem services in the hypothetical market situation, and to express the economic value of ecosystem services by WTP or WTA [29]. The CVM has been applied to the eco-compensation of NRs worldwide [30,31,32,33,34,35,36] (

Table 1. Contingent valuation method (CVM) studies.

Country	Site	Activies	Author(s)
China	Crested Ibis National Nature Reserve	Measured local farmers’ WTA rice loss in 2008 and 2011	Wang et al. (2012) [35]
Nigeria	Yankari Game Reserve	Estimated the value that visitors would give to the Yankari animal shelter	Adamu et al. (2015) [33]
Sri Lanka	Minneriya National Park	Investigated tourists’ WTP for ‘elephant watching’ tickets	Rathnayake (2016) [31]
New Zealand	Pekapeka Swamp	Studied the economic value of the restoration and maintenance of swamp	Ndebele et al. (2017) [32]
China	Yancheng Rare Bird Nature Reserve	Estimated WTA of local farmers	Liu et al. (2018) [34]
China	Ecological function area in Lishui City	Estimated local farmers’ willingness to acccept welfare forest construction	Li (2018) [36]
Taiwan, China	Huisun National Forest Park	Assessed visitors’ WTP for maintaining the quality of forests	Liu et al. (2019) [30]

).

In addition to estimating the specific amount of payment, the study on eco-compensation level can also reflect the real preferences behind stakeholders, and it is also important to determine the causal relationship that affects their preferences [37]. The existing studies mostly apply the logistic regression model to analyze the influencing factors of eco-compensation willingness. Logistic regression model is a generalized regression algorithm widely used in classification problems, and it has a good fitting effect for data with strong linear relationship [21]. However, logistic regression model only includes the main effect of the influencing factors, which can neither express the interaction between variables nor visualize the influencing process [38]. Since the factors that affect the WTA are multifaceted and inter-linked, the interactions between variables cannot be ignored. Decision tree modelling is a non-parametric supervised learning method to represent the classification prediction situation through a tree diagram, which can clearly show the interaction between variables, breaking away from the traditional linear processing [39]. However, it cannot identify the main effect of influencing factors. According to existing research, selection of the optimal solution should not be limited to the choice between the two models, but to achieving an effective combination and complement of two models [40,41,42,43]. It has been suggested that screening out the main effect variables through logistic regression models and further analyzing the interactions between variables in combination with decision tree algorithms can enhance the interpretability of the models [44].

Therefore, this paper first calculates the amount of WTA, then logistic regression and decision tree models are introduced to study influencing factors of eco-compensation intention, to comprehensively analyze the factors and their interaction. Understanding local residents’ WTA and their influencing factors can contribute to theoretical development in this field. An appropriate eco-compensation scheme could improve local residents’ satisfaction with government conservation work and ease the conflict between local residents and Chinese alligators. This paper aims to provide theoretical and practical references for the establishment of a sound eco-compensation mechanism in NCANR, and assist in the overall establishment of NRs’ eco-compensation policy for China. In addition, this paper introduces a machine learning method as a mechanism of identifying factors that influence WTA, which provides a new perspective for the study of identification of influencing factors.

The rest of this paper is organized as follows. Section 2 details the case study site, explains the data sources, and describes the methodology used. Section 3 presents the results, and Section 4 discusses the results. Section 5 presents the conclusion and policy implications.

2. Materials and Methods

2.1. Study Area

The NCANR (Figure 1) is located at the junction of the low hills in southern Anhui province and the plains of the lower Yangtze River in China, and consists of eight zones: Zhucun, Gaojingmiao, Yanglin, Hongxing, Xiadu, Shuanghang, Zhongqiao, and Changle. The areas of the core zone, buffer zone and experimental zone are, 5188 hm², 2506 hm² and 10,871 hm², respectively. In addition, NCANR is located in traditional farming areas and low hilly forest areas, with dense populations, resulting in more severe human disturbance. In accordance with the type of forest land in NCANR, collectively-owned commercial forest land, state-owned commercial forests and public welfare forests, and collective public welfare forests account for 66%, 13% and 8% of the overall area, respectively. Considerable upfront costs are needed in the process of contracting of barren hills, planting trees, forest management and protection, before the commercial forest reaches maturity. However, according to the current policy, all trees in NRs are prohibited from being cut down, which leads to huge income loss for forest farmers.

Chinese alligators’ wetland habitat is frequently occupied by locals around NRs, who also conduct farming operations in the reserve’s central area. This habitat loss affects Chinese alligators’ breeding and hatching in a significant way. Especially since the implementation of the household contract responsibility system, the extensive use of pesticides and fertilizers and the excessive capture of wild fish and shrimp have resulted in a serious shortage of food resources for wild Chinese alligators. On the other hand, the Chinese alligator often brings damage to agricultural production in its activity area, and the strict restrictions on production and life in the reserve also bring serious economic losses and inconvenience to local residents and enterprises. Therefore, the habitat situation of the existing wild Chinese alligator is still grim, and the problems of human-alligator conflict and economic backwardness in NCANR are very prominent. In addition, the current eco-compensation mechanism in NCANR is lacking and the compensation mechanism for wildlife damage is not perfect. Dissatisfaction of local residents with the status quo has made it difficult to carry out field protection and management smoothly.

2.2. Data Sources

We conducted a week-long field survey in Anhui province to properly address the eco-compensation issue of NCANR. The team held separate symposiums and visited sample sites in Hongxing, Yanglin, Zhucun, Gaojingmiao, Shuangkeng, Zhongqiao, and Changle districts. The sample data were collected by randomly distributing questionnaires in various districts, and mainly involve three parts: (1) basic statistical characteristics of the interviewees, (2) cognitive assessment on Chinese Alligator conservation, and (3) willingness to accept (WTA). Relevant investigations were conducted on WTA of residents living in the NCANR, including whether there is a willingness to accept compensation, acceptable methods and expected amount. In order to reduce the deviation of CVM, we adopt the combination of face-to-face interview and random questionnaires. We conducted a pre-survey by using an open questionnaire, and based on the pre-survey, we used a Payment Card questionnaire to study the core valuation issues. There is a sequential range of amounts from low to high presented in the Payment Card questionnaire (Figure 2), from which the respondents choose their own acceptable amount range.

As shown in

Table 2. Sources of sample data.

Region	Count	Proportion
Hongxing	71	20.82%
Yanglin	15	4.40%
Zhucun	70	20.53%
Gaojingmiao	35	10.26%
Shuangkeng	45	13.20%
Zhongqiao	35	10.26%
Changle	37	10.85%
Other areas in Anhui	31	9.09%
The rest of the country (other than Anhui)	2	0.59%
Total	341	100%

, 355 questionnaires were distributed and received, while 341 valid questionnaires were obtained after excluding those with incomplete information and unqualified responses. The efficiency of the questionnaires was 96.06%. Different districts in Chinese Alligator protection areas have both commonalities and divergences, so the sample has certain representativeness.

2.3. WTA Evaluation Method

Consumers are generally influenced by the prices of market goods (p) and non-market goods (q), personal endowment (y), individual preference (s), and random factors (ε) caused by individual preference and measurement errors. We expressed the utility function of consumers as $U\left(p, q, y, s, \epsilon \right)$ [40]. Time before the implementation of protection measures is period 0, and time after the implementation of protection measures is period 1. According to our survey and related research, farmers will lose income because of environmental protection, thus reducing its effect. That is $q_0<q_1$, $U\left(p, q_0, y, s, \epsilon \right)>U\left(p, q_1, y, s, \epsilon \right)$. Therefore, it is necessary to make certain economic compensation for those whose interests are damaged.

$$U\left(p, q_0, y, s, \epsilon \right)=U\left(p, q_1, y+WTA, s, \epsilon \right)$$

Because WTA is influenced by personal income and preference and other factors, the WTA function can be built as follows:

$$WTA=F\left(D, T, S\right)$$

where D is economic damage, S is individual preference, and T represents a personal economic characteristic.

In this paper, we used the non-parametric estimation method. Applying the mathematical expectations of discrete variables, the expectations of the WTA are calculated by multiplying the survey values with the interviewed probabilities as follows:

$$E\left(WTA\right)={\displaystyle \sum }_{i=1}^n{A}_i\times P_i$$

where E(WTA) is the expected value of willingness to accept, $A_i$ is the amount of willingness to accept (CNY), $P_i$ is the probability that the interviewees determine the amount, and n is the sample size of the interviewees who are willing to accept the amount.

2.4. Method for Analyzing WTA Impacting Factors

Taking the WTA of residents in Chinese alligator nature reserve as the explained variable, the logistic regression model and decision tree model were established by using the potential influencing factors of willingness to pay, and the statistically significant variables were screened out. The mechanism of the influencing factors was demonstrated by tree diagram.

2.4.1. Variables Setting

There are many factors that affect the WTA of residents in NCANR. As the main body of agricultural production, farmers are constrained by their own internal resource limits and external environmental conditions [45]. This paper defines 16 independent variables from three aspects [46]: individual characteristics, ecological protection cognition and external factors. The varibles are: (1) individual characteristics: region, gender, age, education, occupation type, family size, average annual household income, main source of income, and distance; (2) ecological protection cognition: understanding of the functions of NCANR, attitudes toward protection, recognition of NCANR values, and understanding of eco-compensation policies; and, (3) external factors: impact of NCANR on household income, extent of positive impact from the establishment of NCANR, and extent of negative impact from the establishment of NCANR. In order to better study the individual’s WTA, the above variables, such as region, distance, extent of favorable impact from the establishment of NCANR, extent of negative impact from the establishment of NCANR, are all indicators drawn up according to the special conditions of NCANR. Because of the wide range of NCANR, great differences among various regions, and different policies in the core area and buffer area, the region and distance are included in the variables. Farmers’ economic behaviors are the feedback to external information in pursuit of maximum benefit in a given social environment. Due to the individual differences of local residents, the impact of the establishment of NCANR on residents is multiple and different. Extent of favorable impact from the establishment of NRs and extent of negative impact from the establishment of NCANR are designed to classify the direction and ranking of external impacts on local residents. The information related to the variables is shown in

Table 3. Selection and definition of variables.

Variables		Definition	Content
Basic statistical characteristics	X1	Region	HongXing = 1; YangLin = 2; ZhuCun = 3; GaoJingMiao = 4; ShuangKeng = 5;
			ZhongQiao = 6; ChangLe = 7; Other areas in Anhui = 8; The rest of the country (other than Anhui) = 9
	X2	Gender	Male = 1; Female = 0
	X3	Age	0–18 = 1; 19–30 = 2; 31–45 = 3; 46–65 = 4; 66+ = 5
	X4	Education	Never attended primary school = 1; Primary school = 2; Junior high school = 3; Senior high school = 4; College and above = 5
	X5	Occupation	Local residents = 1; Village cadres/civil servants = 2; Employees of enterprises nearby = 3; Enterprise representatives = 4; Others = 5
	X6	Household population	Actual population
	X7	Average annual household income	≤12,000 = 1; 12,001–36,000 = 2; 36,001–60,000 = 3; 60,001–90,000 = 4; 90,001–120,000 = 5; 120,001–150,000 = 6; 150,001–180,000 = 7; 180,001–200,000 = 8; >200,000 = 9
	X8	Main sources of income	Planting = 1; Breeding = 2; Government subsidy = 3; Migrant workers = 4; Self-employed = 5; Salary = 6
	X9	Distance	0~5km = 1; 6~15km = 2; >15km = 3
Ecological cognitive assessment	X10	Familiarity of the function of the reserve	Not at all familiar = 1; Slightly familiar = 2; Somewhat familiar = 3; Moderately familiar = 4; Extremely familiar = 5
	X11	Attitude toward nature reserve areas	Strongly oppose = 1; Somewhat oppose = 2; Neutral = 3; Somewhat favor = 4; Strongly favor = 5
	X12	Value identification	Low = 1; Medium = 2; High = 3
	X13	Familiarity of the policies on eco-compensation	Not at all familiar = 1; Slightly familiar = 2; Somewhat familiar = 3; Moderately familiar = 4; Extremely familiar = 5
External factors	X14	Impacts on household income	Reduced income = 1; No effect = 2; Increased income = 3
	X15	Positive effects	No affect = 1; Minor affect = 2; Neutral = 3; Moderate affect = 4; Major affect = 5
	X16	Adverse effects	No affect = 1; Minor affect = 2; Neutral = 3; Moderate affect = 4; Major affect = 5

. The theoretical framework is shown in Figure 3.

2.4.2. Logistic Regression Model

The equation of logistic regression model was established based on existing relevant studies [46,47,48].

$$ln\frac{P}{1-P}={\beta }_0+{\beta }_1{X}_1+\cdots +{\beta }_n{X}_n+\mu$$

where P is the probability of WTA, β is the estimate of variables, μ is the random error, $X_n$ is the independent variable.

Residents’ WTA is taken as an explanatory variable. To ensure that the model estimation results are accurate and valid, the variance inflation factor (VIF) is used preemptively to test the covariance of the independent variables. The test results show that the maximum value of VIF is 1.598. There is no multicollinearity between variables. Univariate analysis of independent variables shows that all independent variables influencing WTA are statistically significant. The logistic regression models were developed using SPSS. All independent variables were analyzed by the backward stepwise regression method based on maximum likelihood estimation with an entry criterion of 0.05 and an exclusion criterion of 0.1.

2.4.3. Decision Tree Model

Similar to the flowchart, Decision tree has a hierarchical tree structure. Each path from a root node to the leaf node corresponds to a specific decision rule. While the whole tree corresponds to a set of classification rules, including the selection of features of the nodes, tree construction, and pruning. In the selection of features, information gain ratio and Gini coefficient are used as the selection criteria. Python was employed to invocate scikit-learn to model decision trees on influencing factors of WTA using CART algorithms.

3. Results

3.1. Respondents’ Characteristics

The results of descriptive statistics of the samples are shown in

Table 4. Demographic data and characteristics of respondents.

Survey Items	Options	Proportion
Gender	Male	79.71%
Age	0–18	0.58%
	19–30	10.14%
	31–45	31.01%
	46–65	55.07%
	≥65	3.19%
Education	Never attended primary school	0.87%
	Primary school	6.09%
	Junior high school	41.74%
	Senior high school (polytechnic school)	19.13%
	College and above	32.17%
Average annual household income (CNY)	≤12,000	8.41%
	12,001–36,000	21.16%
	36,001–60,000	33.04%
	60,001–90,000	15.94%
	90,001–150,000	12.76%
	≥150,000	8.70%

. Most of the residents interviewed were in the age group of 31 to 45 and 45 to 65, which accounted for 86.08% of interviewees. The groups are generally the main labor force and the important support of the family income, so it strongly represents the overall opinions of the respondents’ families. The respondents’ educational level is mainly junior high school, or college and above, accounting for 41.74% and 32.17%, respectively. The characteristics such as the low proportion of young people and low education level for the remainders in the villages are in line with the current situation of rural China. The annual household income curve basically conforms to the right-skewed distribution. In 2020, the per capita disposable income of rural residents in Xuancheng City, Anhui province was CNY 18,928 (USD 2711.79), which is basically consistent with the survey results in this paper.

3.2. Residents’ Awareness of Protecting Chinese Alligator

In order to understand the residents’ WTA, we must first understand the residents’ attitude and awareness regarding Chinese alligator protection. Results of the questionnaire showed that 84.5% of residents support the establishment of NCANR and are willing to protect Chinese alligators within their power. On the contrary, 15.5% of residents oppose the establishment of a reserve, believing that it causes a lot of inconvenience to normal life. About 30% of residents think that economic development is more important than ecological protection; 56.9% believe that both are equally important; and only 13.1% think that economic development is more important. More than 80% of the residents affirmed the value of NCANR for the protection of Chinese alligator. Based on the results of the field research, we learned that the residents have a strong bond with the crocodile and feel proud of having a “national treasure” in their hometown. Older villagers recalled that when they were young they could often see Chinese alligators sunning themselves on the riverbank. In summary, most residents of the reserve can recognize the importance of the NCANR, but there are still some residents whose ecological awareness needs to be improved.

3.3. Preference of Residents’ WTA Methods

Understanding residents’ preference for WTA methods is helpful to improve their overall satisfaction with eco-compensation. Questionnaire results show that 59% of residents prefer to receive cash compensation, 5% of residents prefer to accept other forms of compensation, 36% of residents have no specific preference for compensation methods. Among the choices of non-cash compensation methods, the residents’ choices with higher frequencies are land compensation, employment arrangement or employment guidance, relocation, and infrastructure construction. As for the frequency of compensation payment, 24% of residents choose regular compensation, 44% of residents choose one-time compensation, and 32% of residents choose both.

3.4. Measurement of Residents’ WTA

Survey results show 93% of the residents believe they should be compensated, and 59% of residents want to be compensated more than CNY 3000 (USD 428) per household per year, 13.8% of residents expect to be paid between CNY 2400 CNY (USD 342) per household per year and CNY 3000 (USD 428) per household per year, whereas 9.5% of residents want to be paid between CNY 1800 (USD 257) per household per year and CNY 2400 (USD 342) per household per year, 6.4% of residents want to be paid between CNY 1000 (USD 143) per household per year and CNY 1800 (USD 257) per household per year, the rest of residents choose a compensation amount below CNY 1000 (USD 143) per household per year. The median value of the amounts chosen by the respondents is used as their preferred WTA (Figure 4). According to the distribution frequency data and the non-parametric estimation model of WTA, it is estimated that WTA of residents around NCANR is CNY 25,542 (USD 3659) per household per year.

3.5. Analysis of Influencing Factors of Residents’ WTA

3.5.1. Results Based on Logistic Regression Model

In the logistic regression model, the likelihood ratio is used to test the irrelevant hypothesis, and the likelihood ratio statistics approximately obey the χ² distribution. The –2 log likelihood of this model is 71.067; Cox & Snell R² is 0.260; Nagelkerke R² is 0.651; the likelihood ratio chi-square value is 102.587, and the p value is 0.000. Therefore, the model is overall significant and has good interpretability.

Results are shown in

Table 5. Logistic regression results of WTA.

Variables	B	S.E.	Wald	P	OR	OR 95% CI
Age (X3)	1.083	0.386	7.858	0.005	2.952	1.384–6.293
Household population (X6)	−0.386	0.230	2.823	0.093	0.680	0.433–1.066
Main sources of income (X8)
Planting	3.717	1.372	7.342	0.007	41.13	2.795–604.967
Going out to work	−2.050	1.541	1.776	0.183	0.128	0.006–2.628
Individual business	0.983	1.185	0.689	0.407	2.673	0.262–27.245
Wage	2.210	0.260	5.698	0.067	9.111	1.485–55.912
Aquaculture					1.000
Degree of adverse effects (X16)	1.605	0.333	23.270	0.000	4.977	2.592–9.553
Constant	−3.860	1.765	4.7840	0.029	0.021

. There are four variables selected by logistic regression model. “Age (X3)”, “main sources of income (X8)” and “adverse effects (X16)” were significant at the level of 0.05. “Age (X3)” is a risk factor (OR > 1), which is significant because older people have a greater need for eco-compensation due to their reduced labor ability, lacking of working skills, and narrow employment opportunities. “Main sources of income (X8)—Planting” is a risk factor (OR > 1). Families whose main source of income is planting have lost their important sources of income due to the regulations of protected areas, so they have a strong demand and dependence on eco-compensation. “Main sources of income (X8)—Going out to work” is a protective factor (OR < 1). Migrant workers are less dependent on the income in NCANR, so their WTA will be relatively reduced. “Adverse effects (X16)” is a risk factor (OR > 1), which is the most significant factor in the explanation. NCANR will restrict residents’ production and business activities. The greater the damage, the stronger the residents’ WTA.

3.5.2. Results Based on Decision Tree Model

The influencing factors listed in Table 3 were used to build the decision tree model and the tree diagram is shown as Figure 5. The prediction accuracy of the decision tree model is 94.23%, and the area under ROC curve (AUC) is 0.739. The prediction effect is good. The decision tree model screened out three variables: impacts on household income (X14), adverse effects (X16), and attitude towards protected areas (X11).

Figure 5 illustrates the decision tree, from which a set of if–then classification criteria, namely a decision list, is generated as follows:

Class 1: if X14 > 2, then WTA = No.

Class 2: if X14 ≤ 2, X16 > 1.32, then WTA = Yes

Class 3: if X14 ≤ 2, X16 ≤ 1.32, X11 ≤ 3.5 then WTA = Yes

Class 4: if X14 ≤ 2, X16 ≤ 1.32, X11 > 3.5 then WTA = No

The root node of the decision tree is “Impacts on household income (X14)”, which is the significant influencing factor. “Impacts on household income (X14)” is a comprehensive index that considers both the positive and adverse effects brought by the establishment of NCANR. As long as NCANR generally provides benefits to the residents’ income, then they do not have WTA. Residents whose interests have been damaged by the NCANR do not necessarily have the will to be compensated. Residents who have suffered great losses are generally willing to be compensated. Residents with a few compromised interests who have a supportive attitude toward NCANR do not have WTA. On the contrary, if people do not support NCANR, even those with little damage want to be compensated.

4. Discussion

4.1. Higher WTA for Residents around NCANR

Of the farmers living in Yancheng Rare Bird Nature Reserve, 88% chose a non-zero WTA, with an average minimum WTA compensation of CNY 7727.7 per hm² [34]. WTA amounts of rice loss of farmers in Crested Ibis Nature Reserve in 2008 and 2011 are, CNY 3560.56 per hm² and CNY 3679.83 per hm², respectively [35]. Of the local farmers, 97.77% have WTA for public welfare forest construction in ecological function area in Lishui city, and the average WTA is CNY 96 per acre per year, which is far higher than the current compensation standard in Zhejiang province [36]. Compared with the above and other eco-compensation cases in China [34,35,36,49,50,51], it is found that the residents living in the NCANR have higher WTA and expect a higher amount of compensation.

The policy of turning farmland into forests and turning farmland into grasslands in China follows the principle of voluntary participation of farmers [52]. Unlike these voluntary projects, the restrictions on the production activities of residents living in NCANR are mandatory, and a great number of local residents have suffered inevitable losses to some extent. Thus, the residents living in NCANR have a higher WTA. According to the investigation and interview, people who have no desire to be compensated could be grouped into two categories. One represents those who do not suffer or suffer very little, and the other is those who voluntarily protect Chinese alligators. This is also consistent with the results of the empirical evidence in this paper. In fact, private organizations began to protect wild Chinese alligators earlier than government organizations. For example, the villager Jinyin Zhang and his family are known as “guardians of wild Chinese alligator”. They have voluntarily undertaken the work of protecting wild Chinese alligators for more than 40 years. While feeding wild Chinese alligators, they often fight with the “bad guys” who illegally steal fish, electrify earthworms and pollute the habitat environment. Finally, under the guardianship of several generations, the breeding population of wild Chinese alligators in Changle village where they live, has become the largest currently in the world [53].

Because NRs contain many ecosystems, its eco-compensation is more complicated than it would be for a single type of ecosystem. Most NRs in China are located in remote areas with low population density. Due to their remote location, local people retain their traditional ways of production, living and resource utilization. Compared to other densely populated areas, human activities and exploitation of resources in NRs are relatively low. As a result, local residents will suffer relatively less from the establishment of the reserve and the amount of ecological compensation would be lower. Different from other NRs, the NCANR is densely populated, with good land resources, developed water system and abundant natural resources, and the living areas of people and Chinese alligators overlap highly. The industry within and around NCANR is underdeveloped, with emphasis on the development of forest, tea, fruit and some organic agricultural products. Livestock is also important in the economic structure of the community, accounting for about 25% of the total community economy. The constraints on production activities due to NCANR have caused heavy economic losses to the farmers living there.

In addition, the collective-owned forest lands within NCANR are dominated by artificial commercial forests. This is different from other NRs with a high proportion of public welfare forest. The commercial forests in NCANR are the result of several generations of effort, and farmers have invested a lot so far. Those commercial forests in NCANR are now reaching maturity and timber prices are at high levels. However, they are not allowed to be cut, which has caused great loss to the farmers. Considering the actual loss of residents living in NCANR, the WTA level calculated in this study is reasonable.

4.2. Individual Characteristics Have a Significant Influence on WTA

Age (X3) and Main sources of income (X8)—planting both belong to individual characteristics. Older people are more likely to have WTA than younger people. Since the local industry is underdeveloped, many young and middle-aged laborers leave the village in search of work, which is one of their major sources of income. Older people have to return home to work, due to the decline of working ability, lack of working skills and limited employment channels. As a result, older people are more dependent on the resources of NRs and suffer more loss. Main sources of income (X8)—planting, a factor that reflects the families whose primary source of income is planting, implying that their ability to resist risks is weak. The restrictions on production in NCANR make the income of these families plummet, and even the basic life is difficult to maintain. Therefore, they have a strong need and dependence on eco-compensation.

4.3. Ecological Protection Cognition Has a Significant Effect on WTA

Attitude towards protected areas (X11) belongs to the ecological protection cognition variable, and residents’ awareness of protecting Chinese alligators has a significant effect on WTA. Residents who understand the ecological functions of NRs have a higher tolerance for the loss of benefits from NRs. They usually can recognize the value of NRs. Local residents who do not support the establishment of the NRs are less likely to cooperate with the government in the protection of the Chinese alligator and less likely to agree with the government on the amount of compensation. Because of the negative impact of protected areas on local residents, not all residents are willing to accept local governments’ eco-compensation scheme. According to the results of the questionnaire in this paper, 81% of the respondents do not understand ecological compensation. This situation will undoubtedly increase the difficulty of implementing the eco-compensation policy and local residents to be dissatisfied with the government.

4.4. External Impact Has a Significant Effect on WTA

The eco-compensation for local residents in this paper is more like a type of compensation for their losses, and the external impact on local residents is the most important factor affecting WTA. NCANR is widely distributed, and there are great differences in land value and resource endowment among different regions. The livelihood capital and specific damages of different residents are also very different. Too simple a compensation scheme would obviously be difficult to meet the requirements of farmers. If the losses suffered by the residents cannot be reasonably compensated, and the residents’ satisfaction with the ecological compensation policy is low, the residents will have strong negative emotions. These negative emotions will not only make grassroots work difficult, but may even further intensify the conflict between people and Chinese alligator. The local residents are the main force in the protection of Chinese alligator. Without the support and cooperation of local residents, it would be difficult to protect the Chinese alligators.

5. Conclusions and Policy Implications

5.1. Conclusions

In this paper, the WTA of the residents living in NCANR is measured, and its influencing factors are analyzed by employing the logistic regression model and the decision tree model. The main conclusions are as follows:

(1) Of the residents living in NCANR, 93% have WTA, and the amount of WTA is CNY 25,542 (USD 3659.36) per household per year. The proportion of the residents who have WTA and the overall level of amount of WTA are both relatively high.

(2) Factors affecting residents’ WTA include: Age (X3), Main sources of income (X8)–planting and adverse effects (X16), Impacts on household income (X14), Attitude towards protected areas (X11). Among them, Age (X3) and Adverse effects (X16) are positively correlated with WTA, Impacts on household income (X14) and Attitude towards protected areas (X11) are negatively correlated with WTA.

5.2. Policy Implications

In light of the actual situation of the construction of NCANR, as well as the results of questionnaire survey and field survey, the study provides several suggestions as follows:

(1) Fairness should be taken into account in the eco-compensation mechanism. It is suggested that local governments make differential compensation through field measurement. Factors such as land endowment, forest management, loss of breeding industry, damage of crops, location land price, etc. can all be taken into consideration in the measurement, and the appraisal proposal refers to the market price.

(2) The grassroots government can further strengthen their propaganda work for the eco-compensation policy and the ecological function of the NCANR, and establish an incentive mechanism for the collectives and individuals with outstanding protection results, to encourage increased participation of non-governmental parts in the protection of Chinese alligators.

(3) The source of eco-compensation funds in the Chinese Alligator Reserve is single, mainly relying on local and central finance, and the government is under great financial pressure. Compensation can be carried out in stages, with diversified compensation schemes and methods. In the early stage, cash compensation, land circulation and ecological relocation can be the main methods, and in the later stage, technical support, skills training, policy preferential treatment and other methods are more reasonable. Cash compensation can be a combination of lump-sum buyouts and periodic compensation to alleviate the financial pressure for local government.

(4) Eco-compensation policies and poverty alleviation policies can be combined to establish special local financial funds for providing subsidies to vulnerable groups, with a focus on elderly groups. The grassroots government can consider establishing a regular return visit mechanism to provide targeted assistance to families at risk of returning to poverty in response to the establishment of protected areas.

(5) The current situation of the NCANR is that the production activities in the core areas and the experimental area are both prohibited, and the boundary of the reserve is not clear. Some data should be collected such as the number of Chinese alligators, geographical condition, and the number of farmers in different areas in the reserve to reasonably determine the attribute and boundary of the NCANR. It is suggested to further clarify the scope of productive activities allowed in different areas of the reserve, strengthen the management of the core area, and implement ecological relocation and land circulation as soon as possible. Without destroying the original ecosystem, it is permitted in the experimental area to formulate practical ecological tending schemes, introduce advanced agricultural technology, and develop organic agriculture.

This paper has some limitations. This paper only considers the WTA of residents living in NCANR, and ignores the WTP of local residents. In fact, local residents are the main force to protect Chinese alligators. To overcome this limitation, a future study may investigate the WTP of local residents. By combining WTA and WTP, we may have a deeper discussion. In addition, a future study may discuss the proportion of eco-compensation funds contributed by the central government and local government.