Abandoned forest ecosystem: Implications for Japan's Oak Wilt disease

Journal of Forest Economics 29 (2017) 56–61

Contents lists available at ScienceDirect

Journal of Forest Economics journal homepage: www.elsevier.com/locate/jfe

Abandoned forest ecosystem: Implications for Japan’s Oak Wilt disease夽 Kohei Imamura a,∗ , Shunsuke Managi b , Shoichi Saito c , Tohru Nakashizuka d a Graduate School of Agriculture, Kyoto University, Division of Natural Resource Economics, Forest Policy and Economics, Kitashirakawa, Oiwakecho, Sakyoku, Kyoto, Kyoto 606-8502, Japan b Urban Institute & Departments of Urban and Environmental Engineering, School of Engineering, Kyushu University, 744 Motooka, Nishi-Ku, Fukuoka, Japan c Yamagata Prefectural Forest Research & Instruction Center, 2707 Sagae Hei, Sagae City, Yamagata, Japan d Research Institute for Humanity and Nature, Kyoto, Kyoto 603-8047, Japan

a r t i c l e

i n f o

Article history: Received 4 February 2017 Received in revised form 3 June 2017 Accepted 16 August 2017 Available online 29 September 2017 Keywords: Abandoned coppice forests Japanese Oak Wilt Discrete choice experiment Generalized multinomial logit model WTP-space

a b s t r a c t This study determined values for the ecosystem services of abandoned coppice forests that are threatened by a forest disease known as Japanese Oak Wilt. We applied a discrete choice experiment to value these ecosystem services. The results indicated that ecosystem services were highly valued in the order of biodiversity conservation, water and soil regulation, timber provision, and climate change mitigation. This study suggests that people expect abandoned coppice forests to be protected from Japanese Oak Wilt and to become rich in biodiversity. However, public preference for biodiversity conservation services had high heterogeneity among people. On the other hand, water and soil regulation services were widely ranked as important among people. Furthermore, traditional management method is most preferred than other forest-change scenarios in JOW countermeasures. ˚ © 2017 Department of Forest Economics, Swedish University of Agricultural Sciences, Umea. Published by Elsevier GmbH. All rights reserved.

Introduction The outbreak of an epidemic tree disease, Japanese Oak Wilt (JOW), has severely damaged coppice oak forests in many area of Japan since the 1980s (Ito et al., 1998; Kubono and Ito, 2002; Ito et al., 2009). JOW infections have been observed by Ida and Takahashi (2010) and, for over one hundred years, by the Forestry Agency (2013), although the geographic range was limited until several decades ago (e.g., Miyazaki, Kochi, Hyogo, and Yamagata prefectures). In the past, such outbreaks used to last for only five to ten years; however, recent outbreaks in large areas have continued for more than ten years (Ito and Yamada, 1998). In 2016, thirty prefectures were damaged by JOW (Forest Agency 2016). JOW damages in Akita and Nara prefectures were particularly higher than other prefectures (Forestry Agency, 2016). This difference between earlier and recent infections can be traced to differences in Japan’s forest management system. JOW

夽 This article is part of a special issue entitled: Land use, Forest Preservation and Biodiversity in Asia published at the Journal of Forest Economics 29C, part A. ∗ Corresponding author. E-mail addresses: [email protected] (K. Imamura), [email protected] (S. Managi), [email protected] (S. Saito), [email protected] (T. Nakashizuka).

infections were observed in trees with large trunk diameters, since the vector, Platypus quercivorus, a wood boring ambrosia beetle, prefers large-trunk trees (Kobayashi and Ueda, 2005). Oak coppice forests in Japan used to be frequently logged, and they were generally felled before they reached the size large enough for a beetle mass attack. Logging was frequent because the tree size required for charcoal and firewood was about 10–20 cm in diameter, which is much smaller than the size of the trees currently maturing. However, energy shifts to fossil fuels caused a drastic decrease in the demand for coppice woods from 1960 to 1970. As a result, many coppice forests were abandoned without management, and oak trees have grown to 40–50 cm in diameter, which is an attractive size for the disease vector. The public lacks a clear incentive to protect these forests from infections because the forests have much less commercial value than they did several decades ago. However, a lack of protection has left the forests susceptible to forest degradation. The oak ecosystem damaged by JOW may lose a significant amount of its ecosystem services (the processes by which an environment produces natural resources) because these forests are sometimes degraded into scrub forests or bamboo grasslands without tall tree species (Ito et al., 2009, 2011). Such forests may then trap less carbon, suffer more erosion, or display less biodiversity than healthy forests. Therefore, to develop strategic measures against JOW, understand-

http://dx.doi.org/10.1016/j.jfe.2017.08.005 ˚ Published by Elsevier GmbH. All rights reserved. 1104-6899/© 2017 Department of Forest Economics, Swedish University of Agricultural Sciences, Umea.

K. Imamura et al. / Journal of Forest Economics 29 (2017) 56–61 Table 1 Attributes and levels of forest change scenarios in choice experiment questions.

57

Table 2 Attributes and levels of status quo scenario in choice experiment questions.

Attributes

Levels

Attributes

S1 level

S2 level

S3 level

TYPE SPECIES CARBON WATER WOOD COST (in 10,000 JPY)

TFM, BRF, PCF +15%, +25%, +35%, +45%, −15%, −25%, −35% −10%, +0%, +10% −10%, +0%, +10% +10%, +20%, +40% 0.1, 0.2, 0.5, 1

SPECIES CARBON WATER WOOD COST (in JPY)

+0% +0% +0% +0% 0

−10% −45% +10% +0% 0

−20% −90% +20% +0% 0

ing the value of ecosystem services that can be lost to this disease is important. This study estimates the values of ecosystem services in terms of protecting abandoned coppice forests from JOW. Many studies have been conducted on the valuation of forest ecosystem services. The Science Council of Japan (2001) used the replacement-cost approach to value all Japanese forests, and reported that the value of the CO2 storage function was JPY 1,239 billion and that the flood water mitigation function was JPY 6,468 billion. Meyerhoff et al. (2009) value benefits from changed levels of biodiversity due to nature-oriented silviculture in Lower Saxony, Germany using a choice experiment. They value habitat for endangered and protected wildlife, wildlife diversity, forest stand structure, and landscape diversity. Juutinen et al. (2011) value different trade-offs between biodiversity and recreational services that emerged in national park development scenarios in Oulanka National Park (Finland) using a choice experiment. Tyrväinen et al. (2014) value enhanced forest amenities, in particular landscape values and biodiversity, in private forests in the Ruka-Kuusamo tourism area in northern Finland using a choice experiment. Mogas et al. (2009) use a contingent valuation and choice experiment to estimate non-market values from alternative afforestation programs in the Northeast of Spain, in particular recreation value, sequestered CO2 , and erosion. Chang et al. (2011) estimate the willingness to pay (WTP) for conserving forests from insect outbreaks using the contingent valuation method and compare forest types (recreational, ecological, and productive forests) to determine which type should be conserved in New Brunswick and Saskatchewan, Canada. Shoyama et al. (2013) use a choice experiment to estimate WTPs for managing natural forests, wetlands, productive forests, and agricultural land in Kushiro, Japan. However, few studies have valued abandoned coppice forests degraded forest disease. This study evaluated four forest ecosystem services, including biodiversity conservation, climate change mitigation, water and soil regulation, and timber provision, all over the abandoned secondary growth forests threatened by JOW in Japan. To value these ecosystem services, this study used four quantitative indicators: “number of wildlife species in forests”, “CO2 storage of forests”, “amount of flood water”, “forestry benefits”. This study addresses the following two questions: Q1: What ecosystem services are highly valued in Japanese abandoned coppice forests? Q2: What types of management methods are highly valued in JOW countermeasures? This study is structured as follows. Section 2 explains the empirical methods, Section 3 reports the results, Section 4 provides a discussion, and Section 5 presents the conclusion. Methods Survey design To value abandoned coppice forests, we applied discrete choice experiment. Discrete choice experiments are a kind of question-

naire surveys on modelling preferences for goods, where goods are described in terms of their attributes and the levels they achieve (Kumar, 2010). Respondents are presented a choice set including alternatives for goods, differentiated by their attribute levels, and are asked to choose one of the alternatives. By analyzing the results of the questionnaire surveys, researchers can obtain values for the attributes of the goods. Our questionnaire comprised three sections. The first section explained JOW and asked respondents about their perspectives on JOW. The second section conducted discrete choice experiments and the third section asked the socioeconomic status of a respondent. In the second section, we provided each respondent the following hypothetical situation, before the discrete choice experiments: (1) Abandoned coppice forests are, and will be, threatened by JOW. Forest ecosystem services will deteriorate by JOW. (2) To conserve forest ecosystem services, it is necessary to manage abandoned coppice forests. (3) However, management of abandoned coppice forests require public donation. (4) Respondents were asked to choose their most preferred scenario from each scenario set four times. Donation for the management of the forests is only a one-time payment. A choice set in discrete choice experiment comprised three forest-change scenarios and a status quo scenario. The forestchange scenarios are those that convert current forests (all forests vulnerable to JOW) into new types of forests (forests tolerant against JOW). The status quo scenario is the scenario that does not apply any measure to forests. The attributes of these scenarios are “forest management type” (TYPE), “number of wildlife species in forests” (SPECIES), “CO2 storage of forests” (CARBON), “amount of flood water” (WATER), “forestry benefits” (WOOD), and “cost of scenario per person” (COST). The attribute TYPE represents kinds of forest-change scenarios. The attribute TYPE are (a) the traditional forest management (TFM) scenario, which converts current forests into forests following the traditional management system with periodical logging for charcoal and firewood, (b) the biodiversityrich forest (BRF) scenario, which converts current forests into biodiversity rich forests escaping disturbance by outside forces like storms, disease or logging and (c) the productive conifer forest (PCF) scenario, which converts current forests into plantations of more commercially important conifer trees. The attributes SPECIES, CARBON, WATER, WOOD represent the level of ecosystem services of forests. The levels of these attributes are represented as a percentage change compared with current attribute levels. The attribute COST represents amount of donation for a forest-change scenario. The level of COST is represented in JPY. Respondents choose one preferred scenario from these four by comparing their attributes. An example of a choice set is presented in Fig. 1. The levels of forest-change scenarios were assigned from Table 1. However, the level of SPECIES was determined by the level of TYPE. When TYPE was TFM, SPECIES was one of +15%, +25%, or +35%. When TYPE was BRF, SPECIES was one of +25%, +35%, or +45%. When TYPE was PCF, SPECIES was one of −15%, −25%, or −35%.

58

K. Imamura et al. / Journal of Forest Economics 29 (2017) 56–61

Fig. 1. Example of a discrete choice experiment question.

The levels of the status quo scenario are presented in Table 2. We also set up three kinds of the status quo scenarios representing different levels of JOW damage because potential JOW damage is uncertain. The scenarios are (d) slight JOW damage (S1), (e) intermediate JOW damage (S2), and (f) serious JOW damage (S3). The levels of the forest-change scenarios and the status quo scenarios are determined based on the expert opinion of JOW researchers. During the survey, respondents were presented four different choice sets. Among them, the levels of status quo scenario are common, but the three forest-change scenarios are different. Development of choice sets was conducted as follows: first, using CBC System (Sawtooth Software, Inc., 2013), we developed 15 kinds of forest-change-scenario-set combinations constituted by four different forest-change-scenario-sets. Each forest-changescenario-set is constituted by three different forest-change scenarios. Next, we combined the 15 kinds of forest-changescenario-set combinations and the three levels of the status quo scenario, S1, S2, and S3. Finally, we obtained 45 kinds of choice set combinations. Our questionnaire differed by the choice-set combinations in the second section. Respondents were randomly assigned to one of the 45 choice-set combinations. Data collection An Internet survey titled “Questionnaire about Japanese Forests” was conducted from December 20 to 26, 2011, by Nikkei Research, Inc., on our behalf. The respondents were men and women aged

20–60 years from all Japanese prefectures. From the 28,562 monitors of Nikkei Research, we obtained data from 6,440 respondents, representing a response rate of 22.5%. The survey population was representative of Japan in terms of gender, age, and demographic spread by prefecture. Descriptive statistics of respondent’s characteristics are showed in Table 3.

Analysis Theory To analyze public preference for management of abandoned coppice forests, we used the generalized multinomial logit model II (GMNL-II) proposed by Fiebig et al. (2010). In estimating marginal WTPs for ecosystem services of abandoned coppice forests, we followed Hensher and Greene (2011) and applied WTP-space specification to GMNL-II. In this paper, we describe a summary of GMNL-II in WTP space. Formulation of preference in a discrete choice experiment is based on the random utility theory. We assume that a respondent chooses an alternative that maximizes his or her utility. We also assume that the utility is represented by a linear utility function that is separable into observable and unobservable components. Then, the utility an individual i obtains by choosing a scenario j from a choice set c is represented as follows: Uijc = X ij ˇi + εijc ,

(1)

K. Imamura et al. / Journal of Forest Economics 29 (2017) 56–61 Table 3 Descriptive statistics of respondent’s characteristics. Rate (%) Gender Man Woman Age

51.93 48.07

20s

17.28

30s 40s 50s 60s

21.86 17.42 20.67 22.76

11.80 34.55 18.21 15.98 8.90

10- to 11-million JPY range 11- to 12-million JPY range 12- to 13-million JPY range 13- to 14-million JPY range

1.74 2.25 0.75 0.98

10.56

14- to 15-million JPY range

2.78

No answer

10.47

Area Hokkaido or Tohoku Kanto Chubu Kinki Chugoku or Shikoku Kyushu or Okinawa

Rate (%) Income Under one million JPY One- to two-million JPY range Two- to three-million JPY range Three- to four-million JPY range Four- to five-million JPY range Five- to six-million JPY range Six- to seven-million JPY range Seven- to eight-million JPY range Eight- to nine-million JPY range Nine- to 10-million JPY range

7.95 6.40 11.72 12.45 10.99 8.70 7.03 6.69 5.23 3.87

exp(X ij ˇi )

k=1

exp(X ik ˇi )

,

(2)

where J is the number of alternatives in a choice set. In the analysis of GMNL-II, ˇi is represented as follows: ˇi = i (ˇ + i ).

(3)

In Eq. (3), ˇ is a constant vector representing the mean of individual parameters. i is a random vector distributed MVN(0,˙) (Gu et al., 2013). The parameter  i is the individual-specific scale of the idiosyncratic error (Gu et al., 2013). To complete the model specification, we followed Fiebig et al. (2010) and assumed that ␴i is lognormally distributed with standard deviation  and mean  + zi , where  is a normalising constant, zi is a vector of characteristics of individual i, and  is a parameter of zi . A marginal WTP for an attribute k is calculated by the following equation: WTPk = −

ˇk . ˇCOST

(4)

The value ˇk is the mean parameter for a scenario attribute k except COST, and the value ˇCOST is the mean parameter for COST. However, in the estimation using GMNL-II, the ratio of attribute parameters cannot be directly calculated because WTP distribution is heavily skewed and may not even have defined moments (Train and Weeks, 2005). Therefore, the authors suggested revising Eq. (1) as follows: Uijc = −ˇCOST (−COST + W ij ˛i ) + εijc .

Coefficient

Standard Error

z

P>|z|

Mean COST SPECIES CARBON WATER WOOD

1 0.0203 0.0073 −0.0151 0.0075

(constrained) 0.0005 0.0004 0.0006 0.0004

44.52 16.58 −23.49 19.72

0.000 0.000 0.000 0.000

SD SPECIES CARBON WATER WOOD Het Tau N McFadden’s R2 Log likelihood

0.0144 0.0156 0.0169 0.0127 0.7557 −0.6205 103,040 0.220 −27,851.475

0.0004 0.0006 0.0013 0.0006 0.0309 0.0358

32.27 27.64 13.34 20.84 24.48 −17.35

0.000 0.000 0.000 0.000 0.000 0.000

distribution of WTP (Hole and Kolstad, 2012). The WTPs for the attributes are given by:

where Uijc is the utility, Xij is the vector corresponding to the scenario attributes, ˇi is the vector of individual parameters for the scenario attributes, and εijc is the unobservable component of the utility. If the εijc is distributed independently and identically with an extreme value distribution, the probability of an individual i choosing scenario j from a choice set c is represented as follows:

J

Table 4 Estimates of marginal willingness to pay for each attribute of forest management scenarios.

Note: SPECIES, “number of wildlife species in forests”; CARBON, “CO2 storage of forests”; WATER, “amount of flood water”; WOOD, “forestry benefits”; COST, “cost of scenario per person”. Coefficients represent marginal willingness to pay (in 10,000 JPY per person) for attributes of a forest management scenario respectively.

Note: Sample size is equal to 6440.

Pijc =

59

(5)

Train and Weeks (2005) call Eq. (5) the model in the WTP-space. Eqs. (1) and (5) are behaviorally equivalent, but standard assumptions regarding the distribution of ˇCOST and ˇk can lead to unusual

˛i = −

ˇi . ˇCOST

(6)

In WTP-space specification in GMNL-II, ˇCOST has to be cancelled out in the expression for WTP coefficients (Hole and Kolstad, 2012), i.e. ˇCOST is incorporated in the scale parameter  i in Eq. (3) (Hole and Kolstad, 2012). By applying this specification, we can avoid the problem of skewing in the distribution. The log likelihood for G-MNL is represented as follows:

LL(˛, , , ˙) =

N  i=1

ln

⎧ J C  ⎨ 

⎫ ⎬

⎩

⎭

yicj Pijc p(˛i |˛, , , ˙)d˛i

c=1 j=1

,

(7)

where N is a sample size, yicj is a dummy variable representing observed choice, p(˛i |˛,,,˙) is implied by Eq. (3). We followed Train (2009) and approximated Eq. (7) with simulation. In the simulation, we used the maximum simulated likelihood method to estimate parameters referring to Thiene and Scarpa (2009). Statistical software and settings We estimated two models, Model 1 and 2. Model 1 clarifies the values of ecosystem services of abandoned coppice forests, and Model 2 clarifies preferred method for JOW countermeasure. Model 1 and 2 correspond to our question Q1 and Q2, respectively. The explanatory variables of Model 1 were SPECIES, CARBON, WATER, WOOD, and COST. On the other hand, the explanatory variables of Model 2 were CARBON, WATER, WOOD, COST, and dummy variables representing TYPE: TFM, BRF, and PCF. The dependent variable was a dummy variable of the answer for the discrete choice experiment. Because SPECIES is correlated with TFM, BRF, and PCF, we allocated these variables to two model. We used the statistical operation software Stata version 13.2 (StataCorp LP, Texas, U.S.A.) in all analyses. The data was regarded as panel data in all analysis. The simulation method was Halton sequences. The number of replications was 500. We used the generalized multinomial logit commands developed by Gu et al. (2013) referring to the work of Gu et al. (2013).

60

K. Imamura et al. / Journal of Forest Economics 29 (2017) 56–61

Table 5 Estimates of marginal willingness to pay for each attribute of forest management scenarios. Coefficient

Standard error

z

P > |z|

Mean COST TFM BRF PCF CARBON WATER WOOD

1 1.2267 0.9596 0.0939 0.0055 −0.0131 0.0022

(constrained) 0.0298 0.0266 0.0278 0.0005 0.0006 0.0004

41.21 36.06 3.38 11.79 −21.51 5.20

0.000 0.000 0.001 0.000 0.000 0.000

SD TFM BRF PCF CARBON WATER WOOD Het Tau N McFadden’s R2 Log likelihood

0.6055 0.2084 0.5970 0.0167 0.0147 0.0149 0.8249 −0.5124 103,040 0.242 −27,061.627

0.0224 0.0370 0.0287 0.0006 0.0013 0.0007 0.0291 0.0397

27.05 5.62 20.83 30.24 11.30 21.71 28.35 −12.91

0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

Note: TFM, BRF, and PCF means forest management types. TFM, “traditional forest management”; BRF, “biodiversity-rich forest”; PCF, “productive conifer forest”; CARBON, “CO2 storage of forests”; WATER, “amount of flood water”; WOOD, “forestry benefits”; COST, “cost of scenario per person”. Coefficients represent marginal willingness to pay (in 10,000 JPY per person) for attributes of a forest management scenario respectively.

Results Marginal WTP for each ecosystem service of abandoned forests All estimates of the attribute variables were statistically significant in Model 1 (Table 4). The mean parameters of SPECIES, CARBON, and WOOD had significantly positive effects (Table 4). The mean parameter of WATER had significantly negative effects (Table 4). The standard deviation of all variables was significantly positive (Table 4). The mean marginal WTPs of SPECIES, CARBON, WATER, and WOOD were 0.0203, 0.0073, 0.0151, and 0.0075 (10,000 JPY per person), respectively (Table 4). The size of mean parameters was SPECIES > WATER > WOOD > CARBON. The z-value size of standard deviation parameters was SPECIES > CARBON > WOOD > WATER. Marginal WTP for each management method in JOW countermeasure All estimates of the attribute variables were statistically significant in Model 2 (Table 5). The mean parameters of TFM, BRF, PCF, CARBON, and WOOD had significantly positive effects. WATER had significantly a negative effect (Table 5). The standard deviation of all variables was significantly positive. The marginal WTPs of TFM, BRF, PCF, CARBON, WATER, and WOOD were 1.2267, 0.9596, 0.0939, 0.0055, 0.0131, and 0.0022 (10,000 JPY per person), respectively (Table 5). The size of mean parameters was TFM > BRF > PCF. The z-value size of standard deviation was TFM > PCF > BRF. Discussion What ecosystem services are highly valued? Our respondents preferred an increase in SPECIES (number of wildlife species in forests), CARBON (CO2 storage of forests), and WOOD (forestry benefits) because these mean parameters were significantly correlated with positive values (Table 4). They also preferred a decrease in WATER (amount of flood water) because

the mean parameter was significantly correlated with negative value (Table 4). These results are reasonable because an increase in SPECIES, CARBON, and WOOD and a decrease in WATER lead to human well-being. People’s valuations can be ranked in the following order: biodiversity conservation, water and soil regulation, timber provision, climate change mitigation (Table 4). The mean value of biodiversity conservation was highest among forest ecosystem services, whereas the z-value of standard deviation was also highest (Table 4). This result indicates high heterogeneity in preference for biodiversity conservation among respondents. On the other hand, the z-value of standard deviation of water and soil regulation was smallest (Table 4). Namely, heterogeneity in preference for water and soil regulation is relatively low. This result indicates that the value of water and soil regulation is widely recognized among people, however, the perspectives for biodiversity conservation is differ among people. The Ministry of the Environment (2011) showed that over 60 percent of Japanese people did not know the word “biodiversity” in an opinion poll. Some persons of our respondents may not have recognized the importance of biodiversity conservation. The Science Council of Japan (2001) showed that the flood water mitigation function was valued higher than the CO2 storage function, which matches our results. Because valuation method used in Science Council of Japan (2001) is the aversive expenditure method, the valuation method is different with our study. However, our respondents’ perspectives were consistent with the results of Science Council of Japan (2001). Chang et al. (2011) revealed the highest values for forests with ecologically important features, followed by forests used for timber production, and then, by forests used as recreational sites, assuming an interest in protecting them from insect outbreaks. Shoyama et al. (2013) placed a higher value on natural forests and wetlands than on productive forests. These results are similar to ours. Our study valued the ecosystem services of abandoned coppice forests that have a low economic value. Respondents assigned low values for the timber provision service in the current forestry situation. The large difference between this study and others, such as Chang et al. (2011) and Shoyama et al. (2013), concerns the question of whether forests remain in use or are abandoned; however, our results are similar and the issue of active forest use did not significantly affect the valuation of ecosystem services. What types of management methods are highly valued? Because the coefficients of TFM (traditional forest management), BRF (biodiversity-rich forest), and PCF (productive conifer forest) had significantly positive effect (Table 5). This result means that our respondents prefer forest-change scenarios to the status quo scenario. PCF deteriorates forest biodiversity than the status quo scenario (Tables 1 and 2), nevertheless our respondents regard that PCF is better than the status quo scenario. People’s valuations can be ranked in the following order: TFM, BRF, and PCF (Table 5). The mean value of TFM was highest among forest ecosystem services, whereas the z-value of standard deviation was also highest (Table 5). This result indicates high heterogeneity in preference for TFM among respondents. On the other hand, the z-value of standard deviation of BRF was smallest (Table 5). Namely, heterogeneity in preference for BRF is relatively low. This result indicates that the value of BRF is widely recognized among people, however, the perspectives for TFM is differ among people. The TFM (traditional forest management) scenario conducts the periodical logging of coppice forests in about 10–20 year interval. To success the TFM scenario under current social and economic systems, it is necessary to increase commercial benefits of woods provided from coppice forests. To solve this problem, the produc-

K. Imamura et al. / Journal of Forest Economics 29 (2017) 56–61

tion and use of woody chip fuels are proposed (e.g. Saito et al., 2015), whereas further studies are needed in order to promote use of coppice forest woods. Our study has a serious problem that the effect of TYPE and SPECIES cannot be divided, because these variables are correlated. However, we are convinced that there are effects of TYPE at least, because the mean parameter of TFM is larger than one of BRF, furthermore, the mean parameter of PCF was significantly positive. If there is no effect of TYPE, the mean parameter of BRF should be larger than TFM, and the mean parameter of PCF should be negative. Conclusion This study evaluated the economic values of ecosystem services provided by abandoned coppice forests by utilizing the case of disease expansion. Our results showed that the marginal WTP per capita for biodiversity conservation was the highest and that CO2 storage was the lowest. The valuation of ecosystem services in an abandoned forest ecosystem did not greatly differ from that given to forests in use as estimated by previous studies. This study has three unique aspects. First, biodiversity conservation is determined to be the most important, even in valuations of abandoned forests. Second, water and soil regulation services are widely ranked as important among people. Third, traditional forest and biodiversity-rich forest management are preferred in JOW countermeasure. This study provides suggestions for how people should manage abandoned coppice forests so as to maximize their ecosystem services in the future. The SATOYAMA Initiative, proposed in the Conference of the Parties (COP) 10, is the activity to realize the sustainable use of natural resources. The SATOYAMA Initiative introduces the traditional forest management in Kawanishi City in Hyogo prefecture as the case of sustainable management (Ministry of the Environment, 2010). The case in Kawanishi City is one of “the traditional forest managements” mentioned in this study. This study also showed that public people preferred the forest management introduced in the SATOYAMA Initiative. The further promotion of the activity as the SATOYAMA Initiative can be regarded as political implication of this study. Acknowledgements This manuscript was revised by Enago, English Editing & Proofreading of Scientific Manuscripts for ESL Authors. References Chang, W.Y., Lantz, A.V., MacLean, A.D., 2011. Social benefits of controlling forest insect outbreaks: a contingent valuation analysis in two Canadian provinces. Can. J. Agr. Econ. 59, 383–404. Fiebig, D.G., Keane, M.P., Louviere, J., Wasi, N., 2010. The generalized multinomial logit model: accounting for scale and coefficient heterogeneity. Market. Sci. 29 (3), 393–421. Gu, Y., Hole, A.R., Knox, S., 2013. Fitting the multinomial logit model in Stata. Stata J. 13 (2), 382–397. Hensher, D.A., Greene, W.H., 2011. Valuation of travel time savings in WTP and preference space in the presence of taste and scale heterogeneity. J. Transp. Econ. Policy 45 (3), 505–525. Hole, A.R., Kolstad, J.R., 2012. Mixed logit estimation of willingness to pay distributions: a comparison of models in preference space and WTP space using data from a health-related choice experiment. Empir. Econ. 42, 445–469.

61

Ida, H., Takahashi, S., 2010. Mass mortality of oak trees had already occurred at the Edo period [1603-1863 in Japan]. J. Jpn. Forest Soc. 92 (2), 115–119. Ito, H., Igarashi, T., Kinuura, H., 2009. Changes in stand structure after the mass mortality of oak trees in Keihoku area, Kyoto City, Japan. J. Jpn. Forest Soc. 91 (1), 15–20. Ito, H., Kinuura, H., Oku, H., 2011. Stand structure of a broadleaved forest with sasa-type floor after mass mortality by oak wilt disease. J. Jpn. Forest Soc. 93 (2), 84–87. Ito, S., Kubono, T., Sahashi, N., Yamada, T., 1998. Associated fungi with the mass mortality of oak trees. J. Jpn. Forest Soc. 80 (3), 170–175. Ito, S., Yamada, T., 1998. Distribution and spread of the mass mortality of oak trees. J. Jpn. Forest Soc. 80 (3), 229–232. Juutinen, A., Mitani, Y., Mäntymaa, E., Shoji, Y., Siikamäki, P., Svento, R., 2011. Combining ecological and recreational aspects in national park management: a choice experiment application. Ecol. Econ. 70, 1231–1239. Kobayashi, M., Ueda, A., 2005. Wilt disease of Fagaceae trees caused by Platypus quercivorus (Murayama) (Coleoptera: Platypodidae) and the associated fungus: aim is to clarify the damage factor. J. Jpn. Forest Soc. 87 (5), 435–450. Kubono, T., Ito, S., 2002. Raffaelea quercivora sp. Nov. associated with mass mortality of Japanese oak, and ambrosia beetle (Platypus quercivorus). Mycoscience 43, 255–260. Kumar, P., 2010. The Economics of Valuing Ecosystem Services and Biodiversity. The Economics of Ecosystems and Biodiversity: Ecological and Economic Foundations, Earthscan, Oxford, UK, pp. 183–256. Meyerhoff, J., Liebe, U., Hartje, V., 2009. Benefits of biodiversity enhancement of nature-oriented silviculture: evidence from two choice experiments in Germany. J. Forest Econ. 15, 37–58. Mogas, J., Riera, P., Bennett, J., 2009. A comparison of contingent valuation and choice modeling with second order interactions. J. Forest Econ. 12, 5–30. Saito, S., Miguchi, H., Shoda-Kagaya, E., 2015. Attracting effect of mass accumulated oak logs for Platypus quercivorus. J. Jpn. For. Soc. 97, 100–106. Shoyama, K., Managi, S., Yamagata, Y., 2013. Public preferences for biodiversity conservation and climate change mitigation: a choice experiment using ecosystem services indicators. Land Use Policy 34, 282–293. Thiene, M., Scarpa, R., 2009. Deriving and testing efficient estimates of WTP distributions in destination choice models. Environ. Resour. Econ. 44 (3), 379–395. Train, K.E., Weeks, M., 2005. Discrete choice models in preference space and willingness-to-pay space. In: Scarpa, R., Alberini, A. (Eds.), Application of Simulation Methods in Environmental and Resource Economics. Springer, Dordrecht, pp. 1–16. Train, K.B., 2009. Discrete Choice Methods with Simulation, second ed. Cambridge University Press, Cambridge. Tyrväinen, L., Mäntymaa, E., Ovaskainen, V., 2014. Demand for enhanced forest amenities in private lands: the case of the Ruka-Kuusamo tourism area, Finland. Forest Policy Econ. 47, 4–13.

Web references Forestry Agency, Current Situation of Japanese Oak Wilt, 2011. http://www.rinya.maff.go.jp/j/hogo/higai/naragare.html (accessed 17.05.16). Ministry of the Environment, Practices for the Sustainable Use and Management of Natural Resources, 2010. http://www.env.go. jp/nature/satoyama/syuhourei/practices.html (accessed 03.06.17). Ministry of the Environment, Annual Report on the Environment, the Sound Material-Cycle Society and Biodiversity in Japan 2011, 2011. http://www.env.go.jp/policy/hakusyo/h23/pdf/full.pdf (accessed 30.05.17). Sawtooth Software Inc., The CBC System for Choice-Based Conjoint Analysis Version 8, 2013, Sawtooth Software Technical Paper Series. https://www.sawtoothsoftware.com/download/ techpap/cbctech.pdf (accessed 04.02.17). Science Council of Japan, Valuation of Ecosystem Services from Agriculture and Forests, 2001. http://www.scj.go.jp/ja/info/kohyo/ 2001.html (accessed 17.05.16).