Understanding Public Intentions to Participate in Protection Initiatives for Forested Watershed Areas Using the Theory of Planned Behavior: A Case Study of Cameron Highlands in Pahang, Malaysia

: The heavy emphasis on land-use changes to meet the needs for gross domestic product growth often causes deforestation, affecting forests’ capability to function as watershed areas properly. While land-use changes generate socioeconomics success, they also lead to environmental deteriora-tion that puts public welfare at greater risk. This study employs the Theory of Planned Behavior (TPB) to evaluate the public’s behavioral intentions towards participation in the protection initiatives for the forested watershed areas in the mountainous region of Cameron Highlands in Pahang, Malaysia. Survey data were used to analyze the effects of TPB constructs on the public’s behavioral intentions. The results show that the public demonstrated readiness to comply with governmental rules concerning environment protection and were motivated to participate in the protection initiatives when there is social encouragement. This study ﬁnds that attitude signiﬁcantly inﬂuences the public’s behavioral intention. This, therefore, indicates the importance of creating conditions to encourage the public’s behavioral beliefs towards protection initiatives that would ensure the sustainability of forested watershed areas. Overall, this study offers information on public participation that is useful to be integrated into a meaningful institutional framework when addressing challenging environmental issues caused by land-use changes that could imperil public welfare.


Introduction
The role of forests in ensuring water availability is of the utmost importance as forests yield the highest quality of water of any ecosystem, which is crucial for ecological needs and human survival [1]. However, land-use changes alter forest cover or land cover and are viewed as significant drivers for hydrological variations in forested watershed areas [2]. Land-use change is driven by deforestation [3] and attributes to 27% of global forest loss [4]. Deforestation is further intensified by agricultural activity [5,6], in which it reduces forest cover and disturbs the biophysical aspects of forests [4]. Although this land-use change has facilitated socio-economic growth, to some extent, it is adversely impacting the fundamental ecological function of forests such as precipitation, water yield and the hydrologic cycle [7]. The precipitation rate decreases as deforestation progresses [8]. At the same time, variations in forest cover due to land-use change affect annual water yield [2], and forest conversion into pasture or cropland extensively alters the hydrologic cycle [9]. Land-use change also induces erosion and other landscape processes that reduce the capabilities of forests to function as watershed areas and therefore limit the generation of critically important water-based ecosystem services [10,11].
The effects of these land-use changes on forested watershed areas have been widely debated in searching for sustainable solutions that preserve the linkages between water 1.
How would the public participate in the protection initiatives (i.e., attitude)? 2.
How would the public perceive social pressure related to their participation in the protection initiative (i.e., subjective norm)? 3.
How would the public recognize their capability to perform the behavior that could protect the forested watershed areas (i.e., perceived behavioral control)?
The findings of this study show that attitude is the most decisive variable that influences public participation in protection initiatives, relative to subjective norms and perceived behavioral control. Attitude, which is based on complex moral and social values [21], could encourage the public to participate in protection initiatives for forested watershed areas, and therefore help in sustaining water-based ecosystems which are vital for human welfare. However, this study's results vary from other studies where various factors besides attitude could also become the significant variables that affect behavioral intention. For example, a study on tourists' intention to participate in the Zero Litter Initiative in mountainous tourism areas in China shows that environmentally responsible intention, moral obligation and perception responsibility shaped tourists' personal norms and became the significant variables affecting their intention [23]. Another study on young people's intention toward municipal solid waste sorting shows that the rankings of significance, personal moral obligation, perceived behavioral control, and subjective norm had positive influences on young people's intention in addressing this predicament of waste management [24]. Nevertheless, several significant results in this study illustrate proenvironmental behaviors and indicate public behavioral and control beliefs also influencing their intention to participate in the protection initiatives. It corroborates with the findings of other studies e.g., [25][26][27], in which TPB is suitable for predicting pro-environmental behavioral intentions. The present study contributes information from the perspective of public participation, which is informative for resource managers to use in safeguarding the Sustainability 2021, 13, 4399 3 of 18 critically important forested watershed areas. This study also enriches the TPB literature in the context of the water-based ecosystem that is relevant to forest resource management.

The Case Study of Cameron Highlands in Pahang, Malaysia
The site for this study is Cameron Highlands in Pahang, which is located in Peninsular (West) Malaysia ( Figure 1). The allocated sampling points for the face-to-face survey interview include five towns in Cameron Highlands (i.e., Ringlet, Tanah Rata, Brinchang, Tringkap and Terla). Cameron Highlands shares its boundary with the states of Kelantan and Perak and consists of three subdistricts (mukim), namely Mukim Ulu Telom, Mukim Ringlet and Mukim Tanah Rata. The total land area is estimated at 71,218 ha and is inhabited by approximately 40,700 people, in which the largest community is Indian with a total of 14,100 people. It is followed by 7800 Chinese and 7500 Malays. The communities of Orang Asli (indigenous people) and non-citizens comprise 6500 and 4800 people, respectively [28]. Although being the smallest district in the largest state, i.e., Pahang, in Peninsular Malaysia, Cameron Highlands is one of the most economically important highlands in Malaysia. Located within the Main Range of Peninsular Malaysia at a mean of elevation between 1000 m and 1830 m above sea level, it is a premier agriculture area for sub-tropical vegetables and flower farming. It hosts the prestigious tea estates of BOH Tea Plantation and Bharat Tea Estate. The area is also well known as a tourism site and it consistently received more than 1 million visitors from 2016 to 2018 [29].
vironmental behavioral intentions. The present study contributes information from the perspective of public participation, which is informative for resource managers to use in safeguarding the critically important forested watershed areas. This study also enriches the TPB literature in the context of the water-based ecosystem that is relevant to forest resource management.

The Case Study of Cameron Highlands in Pahang, Malaysia
The site for this study is Cameron Highlands in Pahang, which is located in Peninsular (West) Malaysia ( Figure 1). The allocated sampling points for the face-to-face survey interview include five towns in Cameron Highlands (i.e., Ringlet, Tanah Rata, Brinchang Tringkap and Terla). Cameron Highlands shares its boundary with the states of Kelantan and Perak and consists of three subdistricts (mukim), namely Mukim Ulu Telom, Mukim Ringlet and Mukim Tanah Rata. The total land area is estimated at 71,218 ha and is inhabited by approximately 40,700 people, in which the largest community is Indian with a tota of 14,100 people. It is followed by 7800 Chinese and 7500 Malays. The communities of Orang Asli (indigenous people) and non-citizens comprise 6500 and 4800 people, respectively [28]. Although being the smallest district in the largest state, i.e., Pahang, in Peninsular Malaysia, Cameron Highlands is one of the most economically important highlands in Malaysia. Located within the Main Range of Peninsular Malaysia at a mean of elevation between 1000 m and 1830 m above sea level, it is a premier agriculture area for sub-tropical vegetables and flower farming. It hosts the prestigious tea estates of BOH Tea Plantation and Bharat Tea Estate. The area is also well known as a tourism site and it consistently received more than 1 million visitors from 2016 to 2018 [29].   As Cameron Highlands is a mountainous area, approximately 74% of the total area has an elevation of more than 1000 m [31]. Due to its location, the forest classifications are ranging from the upper dipterocarp, lower montane to upper montane forests. With different forest types and a variety of elevation and slope aspects, forest areas of Cameron Highlands serve as watershed and catchment areas. The area channels water into three main rivers, namely Telom, Bertam and Lemoi, where these rivers flow across Cameron Highlands and head eastward to form the Jelai River, which is then fused with Tembeling River and forms the longest river in Peninsular Malaysia, i.e., Pahang River (459 km), which subsequently drains into the South China Sea [32,33]. The lowest annual average rainfall in the area is not less than 2000 mm, with relative humidity between 70% and 90%, and the average temperature is 18 • C [34].
With a higher rainfall intensity, the area stores and channels rainwater into three major water catchments; Telom catchment in the north with 14,749.18 ha of forest cover, Bertam catchment in the middle (6903.37 ha of forest cover) and Lemoi catchment (covered by 11,092.55 ha of the forest) in the south [35]. These water resources are critically important for upstream activities such as for the generation of hydroelectric power allocated under the Cameron Highlands Hydroelectric Scheme. For example, in one of the water reservoirs, i.e., the Ringlet reservoir, it utilizes 183 km 2 (18,300 hectares, ha) of the forest as catchment areas and uses headwater from two major rivers, i.e., Sungai Telom and Sungai Bertam, to generate electricity [36]. Water resources are also critical for irrigation in approximately 7508 ha of agricultural areas around Cameron Highlands, where 2300 farmers produce more than 800 metric tons of vegetables daily for domestic and export consumptions, which are annually estimated at MYR 2,945,645 million (USD 707,662) [37,38]. For downstream users, it supplies water for domestic consumption and recreational use [35].
Despite all these critically important water-based ecosystem services, intense land-use change predominantly for agriculture has threatened the sustainability of the forested watershed areas in Cameron Highlands. In river reserve areas, the water bodies have decreased approximately 25%, from 237 ha in 1984 to 177 ha in 2002, due to continuous encroachment [35]. The intense land-use pattern also affects the size of riverbanks of Bertam and Telom and the surrounding forest areas, where it correlated with extensive vegetable farming and urban development [39]. Observation on forests and natural areas indicated a 13% loss between 1966 and 2010 [39], while there was approximately an 8% decrease in forest and wetland areas from 1997 to 2014 [40]. In 2015, agriculture drove a 3% reduction in forest cover [31], while a total of 532.04 ha, out of the 38,419 ha of Permanent Forest Reserve (PRF), have been encroached for illegal agricultural farming [41].
Intensive land-use change driven by deforestation for agricultural activities has severely reduced the size of the forested watershed areas in Cameron Highlands. It has further caused environmental issues including river pollution, slope failure, temperature rising and flooding [39]. The accumulative effects of these environmental issues could affect the sustainability of this critically important water-based ecosystem. Therefore, public support is needed to develop effective environmental policy implementation [42]. The meaningful public input that could be channeled into constructive protection measures is required to ensure sustainable resource use [43]. In the case of the forested watershed areas in Cameron Highlands, it is crucial to mitigate the continued pressures from environmental threats that are produced by rampant land-use change.

Theoretical Framework
The socio psychological TPB theory has been widely used to predict and explain human behavior. In this study, the significance of using the socio-psychological framework is understanding public behavioral intention toward protection initiatives which could help reduce pressure from environmental threats which have been induced by land-use change in the critically important water-based ecosystem in Cameron Highlands. The theory, which was developed from the Theory of Reasoned Action (TRA), highlights that attitude towards performing specific actions should be measured to understand an individual's specific behavior [44]. However, TRA cannot entirely predict the capability of an individual to enact an action. Hence, it is further modified as TPB by including the perceived behavioral control construct [45].
TPB described that actual behavior is explained by behavioral intention, which depends on psychological constructs comprised of attitude towards the specific behavior, subjective norm, and perceived behavioral control. Attitude refers to the individual's positive or negative evaluation of performing the behavior and is preceded by the beliefs concerning the specific behavior and the evaluation of their outcomes [45]. Subjective norm is the individual's perception of social pressure to perform or not perform the behavior, with the causal chain comprised of normative beliefs and the motivation to comply with specific referents [46]. The newly added construct of perceived behavioral control is a measure of an individual's perception of their ability to perform the behavior in question, with components of control beliefs and the perceived power of the particular control factor to facilitate or inhibit behavior [45,47]. The TPB can be mathematically expressed as follows [45,48,49]: where B = behavior, BI = behavioral intention, A = (global) attitude towards the specific behavior, SN = subjective norm, PBC = perceived behavioral control,~denotes to explanation, ∑ k i=1 b i e i = sum of belief-evaluation products, ∑ m i=1 nb i mc i = sum of normative belief-motivation-to-comply products, ∑ n i=1 cb i p i = sum of control belief-perceived-powerto = facilitate/inhibit products, k = number of beliefs, m = number of normative beliefs, n = number of control beliefs, w j = weighting factor, j = 1, 2, 3.
To date, the TPB is applied to examine multifaceted factors affecting environmentally friendly behavior. From the context of protection and conservation, the TPB was used to investigate the behavioral intention of Florida residents in the water conservation program for landscape irrigation [50]. The TPB has also been applied to determine visitor's willingness-to-pay for urban park conservation in the Monte San Pedro Park, Spain [51]. The TPB framework was also implemented in the protection of forests for biodiversity conservation, water resource management and carbon storage among farmers in North Selangor Peat Swamp Forest, Malaysia [21]. In Finland, the TPB was used to examine the private forester willingness to allocate their private forests as wildlife habitats [52]. The wide range of TPB applications in both protection and conservation contexts has ascertained that the theory could address behavioral intention in environmentally friendly behavior.
The theoretical framework of this study is summarized in Figure 2. The public behavioral intention to participate in protection initiatives for the forested watershed areas in Cameron Highlands is determined based on attitude towards participation in protection initiatives (A), subjective norm (SN) and perceived behavioral control (PBC). These TPB constructs can be further divided into their respective components, in which the sums of the component products are estimated according to the theory equivalent to the overall concepts [53]. In this study, the specific action identified is protection initiative participation, the target of the behavior is to increase the participation in protection initiatives, while the context is the public's perception. The time factor is not included as the participation in protection initiatives is viewed as infinite, where continuous effort is needed to reduce pressures and mitigate environmental threats from land-use change to maintain the sustainability of the forested watershed areas in Cameron Highlands.

Sampling Population and Frame
This study is based on data obtained from a face-to-face survey interview conducted in April 2019. The survey interview covered the main population areas of Ringlet, Tanah Rata, Brinchang, Tringkap and Terla in Cameron Highlands (see Figure 1). These areas were selected as study sites as all are often frequented by visitors and occupied by residents, hence they are practical as study sites that represent the targeted sampled population. The data collection was administered to both visitors and residents, aged 18 years and above, using systematic random sampling to ensure samples were well distributed throughout the population. The inclusion of visitors and residents was because both groups represent the human factor (i.e., the initiators of land-use change and recipients of its impacts) in Cameron Highlands. Based on the population data of 40,700 [28], the sample size was determined using the standard formula of the Cochran method [54] as follows: ( . )( . ) .
= 384 respondents when = sample size, = actual sample size, = actual number of populations, = 95% confidence interval, = estimate of variance (0.25) and = 0.05 (acceptable margin of error). Following the Cochran formula, the estimation for the desired sampling size was:

Sampling Population and Frame
This study is based on data obtained from a face-to-face survey interview conducted in April 2019. The survey interview covered the main population areas of Ringlet, Tanah Rata, Brinchang, Tringkap and Terla in Cameron Highlands (see Figure 1). These areas were selected as study sites as all are often frequented by visitors and occupied by residents, hence they are practical as study sites that represent the targeted sampled population. The data collection was administered to both visitors and residents, aged 18 years and above, using systematic random sampling to ensure samples were well distributed throughout the population. The inclusion of visitors and residents was because both groups represent the human factor (i.e., the initiators of land-use change and recipients of its impacts) in Cameron Highlands. Based on the population data of 40,700 [28], the sample size was determined using the standard formula of the Cochran method [54] as follows: n 0 = 384 respondents when n 0 = sample size, n 1 = actual sample size, N = actual number of populations, t = 95% confidence interval, pq = estimate of variance (0.25) and d = 0.05 (acceptable margin of error). Following the Cochran formula, the estimation for the desired sampling size was: Although the sample size was determined at 380 respondents, this study was unable to obtain the target number as many desired respondents were occupied with their daily activity, hence they were unable to participate in this survey. Nevertheless, a total of 143 questionnaires were completed. The original amount of the questionnaires gathered was 200, but 57 questionnaires were not included in data compilation as they were incomplete. Overall, the response rate of this study is 72%.

Survey Design and Analysis
Prior to data collection, a pilot study involving 33 respondents using similar survey methods was conducted in October 2018. It was conducted to assess the overall content of the questionnaire by improving the flow of questions and sequence for each section, identifying environmental issues and checking appropriate TPB constructs concerning behavioral intention on protection initiatives for the forested watershed areas in Cameron Highlands. The pilot study ensured the elimination of repetitive items in the questionnaire, improving any other deficiencies and making sure the final version of the questionnaire fitted with the research objectives [55].
The questionnaire was written in Malay and English and organized by placing instructions consistently above each section, to guide and minimize errors when respondents filled up the questionnaire. Findings from the pilot study assisted in the improvement of the questionnaire before the implementation of the larger scale final data collection. The questionnaire was divided into four main sections: (i) sociodemographic information and related characteristics, (ii) attitude towards protection initiatives, (iii) influence from normative referents on participation in protection initiatives (subjective norms), and (iv) perception of the ability to perform protection initiatives (perceived control). For each questionnaire item in sections (ii), (iii) and (iv), the respondents were asked to indicate the level of agreement using a five-point psychometric Likert scale (1 = Strongly disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly agree). The reliability of the coefficients for these items was analyzed using the Alpha Cronbach method and was found acceptable, as coefficients values for attitude α = 0.70, subjective norms α = 0.78, and perceived behavioral control α = 0.71 [56].
Data from this study were analyzed using The Statistical Package for Social Sciences for Windows (SPSS, v22). Descriptive statistics were used to describe frequencies of respondents' sociodemographic and TPB constructs. The empirical analysis of Spearman's rho correlation was used to measure the strength and direction of the association between variables (see Table 1). Furthermore, multiple regression analysis was applied to determine which factors have the most influence on the respondents' behavioral intention on protection initiatives for forested watershed areas. Table 1. Interpretation of the size of a correlation coefficient.

Size of Correlation
Interpretation of the Effect Size

TPB Variables
The rating scale used in this study was unipolar to ensure variables are unambiguous to the respondents [50]. It allows the respondents to distinguish the difference between the variable categories [58]. The variables were extracted from statements comprised of salient beliefs, referents as well as control beliefs relevant to the protection initiatives for the forested watershed areas in Cameron Highlands. These variables were developed from the findings of the elicitation interview during the pilot study and supplemented with information obtained from previous studies ( Table 2). [55]

Subjective norm (SN)
Perceptions of the influential group sn1 My family gives support when I oppose using chemical fertilizers and pesticides in farming practices. sn2 People who are visiting with me think that I should make an official report to the responsible authority when seeing forest encroachment activities.

Elicitation interview sn3
People who are important to me support my commitment towards the establishment of conservation funds for the encroached forests. [59,60] sn4 The government encourages my involvement in the sustainable agriculture-oriented program. [61] Perceived behavioral control (PBC) Level of the ability to participate in the protection initiatives pbc1 By attending seminars on sustainable agricultural management practices, I am confident it will improve my knowledge of agriculture. [62] pbc2 By accepting the improvement of manure plan, I think it will improve the quality of agricultural produce.

Elicitation interview
The measurement for attitude (A) comprised of belief strengths (b) and outcome evaluations (e). Three items assessed to illustrate belief strengths (b) were public participation in conservation campaign for the protection of the forested watershed areas, public engagement in the good management of fertilizers and pesticides in agricultural activity to reduce river pollution, and public support for the establishment of a watershed conservation fund to be used in rehabilitation of the encroached forests. Meanwhile, three items representing the outcome evaluations (e) were: the importance of conservation campaigns to protect the forested watershed areas, the necessity of good management of chemical fertilizers and pesticides in agricultural activity to reduce river pollution, and the significance of mutual responsibility from all forest beneficiaries to support the establishment of conservation funds for the encroached forests.
For subjective norm (SN) variables, the identified four referents were family members, people who were visiting Cameron Highlands with the respondents, people who are important to them and the forestry department of local government. The norm pressures on the respondents' intention to participate in the protection initiatives, from this influential referent group, were assessed based on the opinions of this normative referent and the respondents' motivation to comply with their suggestions.
The component of normative beliefs (nb) was assessed based on four items: the availability of family support when opposed to excessive use of chemical fertilizers and pesticides in farming practices, receiving encouragement to report forest encroachment activities from people who are visiting together, obtaining encouragement to commit to the establishment of conservation funds for the encroached forests from people who are important to them as well as attaining government support in their involvement in the sustainable agriculture program.
The motivation to comply with the specific referents (mc) was also assessed using four items, including the opposition of farming practices that use excessive chemical fertilizers and pesticides when getting support from family, reporting forest encroachment activities to responsible authorities when encouraged by people who are visiting together, engaging in the establishment of conservation funds for encroached forests when supported by people who are important to them and getting involved in sustainable agriculture programs when there is support from the government.
In the case of perceived behavioral control (PBC), it concerned the possibility to participate in the protection initiatives. The estimation includes the component of control beliefs (cb), which are associated with opportunity, resources or knowledge, and the component of perceived power to facilitate or inhibit behavior (p). The components were measured using two items, comprised of attendance in seminars on sustainable agricultural management practices to improve the knowledge about agriculture, and acceptance in the improvement of the manure management plan to improve the quality of agriculture practice. Meanwhile, two items assessed to indicate perceived power to facilitate or inhibit behavior (p) were attendance in seminars on sustainable agricultural management practices (as it is considered a move that should be embraced by farmers)), and the implementation of sustainable agriculture practice (to reduce pollution risk on forest area and water quality).

Respondents Demographic Information
As shown in Table 3, females constituted 57.3% of the sample, as compared to the 42.7% of male respondents. Higher female participation in this survey might be due to the role of males as the breadwinner of their family, so they were working during the survey interview. Hence, they are less involved in this survey. Nonetheless, females are reported to have stronger environmental attitudes and behaviors as compared to the opposite gender due to higher levels of social orientation and social responsibility [63]. The majority of the respondents surveyed were aged between 18 and 41 years old (89.6%). Only 0.7% of the respondents who participated in this survey were aged between 50 and 59 years old. Meanwhile, the most interviewed were Malays (51%), Chinese and Indian at 21.7% and 20.3%, respectively, and other ethnic groups at 7%. The majority of the respondents have a secondary school education level (46.9%). The income distribution is varying, where most of the gross monthly income is between MYR 900-MYR 2000 (USD 216-481) (84%). More than half of the respondents are locals (58%), and the rest are visitors (42%).

Formation of Attitude, Subjective Norm and Perceived Behavioral Intention
The expectation of the public concerning the salience of belief statements on the protection initiatives for the forested watershed areas is remarkable ( Table 4). The strong agreement on belief strengths and outcome evaluation could increase behavioral intention to participate in initiatives to protect forested watershed areas. The aspect of belief-evaluation products (b i e i ) was all positive and correlated with the attitude (A) ( Table 5). Following interpretation in Table 1, the high positive correlation with the attitude towards participation in the protection initiatives was the establishment of the watershed conservation fund. In contrast, moderate positive correlation and negligible yet positive correlation were shown by public participation in conservation campaigns and public engagement in the good management of fertilizers and pesticides, respectively. The regression values showed that these variables significantly influenced public attitude towards participation in protection initiatives.  The estimations seen in Table 6 could be summarized as unanimous with respect to normative beliefs as they concerned beliefs related to attitude towards participation in protection initiatives for forested watershed areas. The majority of the respondents considered family members, people who are visiting Cameron Highlands with them, and people who are important to them had a favorable opinion with respect to their intention to participate in protection initiatives for forested watershed areas. Respondents also believed the government substantially supports public participation in protection initiatives for forested watershed areas. Concerning the public's motivation to comply, the same four normative referents were the influential advisory groups, but family members were considered the most significant advisers. Table 5. Formation of attitudes towards the participation in protection initiatives for forested watershed areas (A) predicted from belief-evaluation products (b i e i ).

β-Coefficients (t-Values)
Conservation campaign × Importance of public participation in the conservation campaign 0.508 ** 0.554 ** (7.910) Good management of fertilizers and pesticides × Importance of public engagement in the good management of fertilizers and pesticides in agricultural activity  In Table 7, all normative belief-motivation-to-comply products (nb i mc i ) correlated with the subjective norm (SN). The source of norm pressure is similar as in Table 6, but the effect of these norm pressures was further verified using the regression model. Within the context of normative belief-motivation-to-comply products (nb i mc i ), both correlations and regression estimations show the government was notably the most important source of norm pressure relative to other normative referents. Table 7. Formation of the subjective norm towards the participation in protection initiatives for forested watershed areas (SN) predicted from normative belief-motivation-to-comply products (nb i mc i ). The two control belief statements concerning attendance in seminars on sustainable agricultural management practices and accepting the improvement of a manure management plan received significant support from the respondents (Table 8). Respondents also regarded having an improved knowledge on sustainable agriculture practice and engaging in agriculture activities that avoid the excessive use of chemical fertilizers and pesticides had the power to inhibit or facilitate their participation in protection initiatives for forested watershed areas. Both their attendance in seminars on sustainable agriculture and their acceptance of manure improvement were perceived positively and correlated with the global measure of perceived behavioral control (PBC) ( Table 9). The correlations and regression estimations results suggest that these two statements were the components of the global measure of perceived behavioral control.  Table 9. Formation of the perceived behavioral control (PBC) towards the participation in protection initiatives for forested watershed areas predicted from control belief-perceived-power-tofacilitate/inhibit products (cb i p i ).

Models Predicting Intention to Participate in the Protection Initiatives
The global measures of attitude (A), subjective norm (SN) and perceived behavioral control (PBC) had positive correlations with the behavioral intention to participate in protection initiatives for forested watershed areas (BI) ( Table 10). The sum of normative belief-motivation-to-comply products (∑ nb i mc i ) and the sum of belief-evaluation products (∑ b i e i ) were positively correlated with intention, and so did was the sum of the components of the perceived behavioral control (∑cb i p i ). Table 11 presents the estimation results of the regression models explaining the intention to participate in protection initiatives. The standardized regression coefficients (β-coefficients) of the Basic model and its corresponding t-values (in parenthesis) are presented in the first column. The explanatory variables were comprised of attitude, subjective norm, and perceived behavioral control. The results for the Basic model show 29% of the total variation in the intention to participate in protection initiatives (Adj. R 2 ), and it was statistically significant (F-value 19.8). The comparison between β-coefficients of the three TPB constructs shows that attitude was the most powerful attribute explaining the intention followed by the norm, while perceived controlling factors were statistically insignificant.  Model 1, including the attitude, the sums of norm and control, was significant in explaining the intention (Adj. R 2 = 0.64). The signs of the variables were as expected with the attitude component, and the sums of norm and control factors were all statistically significant. The model was improved in Model 2 (Adj. R 2 = 0.73), with attitude remaining positive and statistically significant, although perceived behavioral control turned negative but statistically insignificant. The sums of attitude and control were significant in explaining the intention of the respondents to participate in protection initiatives.
Both estimations for attitude and subjective norm turned negative in Model 3 (Adj. R 2 = 0.64), but only the attitude component was statistically significant. Like its effect in Model 2, the sum of attitude was positive and statistically significant. The sum of norm also remained positive and statistically significant. In the Extended model, only the sum of control was estimated positive and statistically significant. However, the explanatory power of this model substantially decreased (Adj. R 2 = 0.08) relative to other estimated models, suggesting that the Extended model inadequately explains the intention of the respondents to participate in protection initiatives.
Based on the results in Table 11, the interaction effects of global measures of attitude, subjective norm, and perceived behavioral control were tested by multiplying each construct with each other and by adding the respective products as an independent variable. Across the tested models, global measures of attitude predominantly affect the intention, followed by the sum of norm. The sum of control somewhat explained the intention, but the variable of the subjective norm caused the smallest effect.

Concluding Discussion
The empirical results confirm the utility of the TPB as a framework that explains the behavioral intention of the public. The measures of attitude, subjective norm and perceived behavioral control explained the self-reported intention behavior that the public has towards protection initiatives for the forested watershed areas in Cameron Highlands. The sums of components of the attitude, subjective norm, and perceived behavioral control also functioned as expected. The estimated models have proved the sufficiency of the TPB framework in this application. Statistical analyses show that data fit the estimated models, highlighting that while the TPB construct of subjective norm and perceived behavioral control affects intention, attitude has emerged as the most dominant explanatory factor in explaining behavioral intention. The results are like other TPB applications of forestry context, including in, e.g., forest owners' choice of reforestation method [64], farmers' behavior towards tree planting [65] and professional foresters' intentions to conserve habitats beyond what is the minimum legally defined requirement when planning forestry operations [52]. This supports the claim that intention is often preceded by attitudes [25,66] and suggests that measures which emphasize the positive outcomes of behaving favorably towards protection initiatives would be highly effective in protecting forested watershed areas, as compared to efforts that focus on the subjective norm and perceived behavioral control.
The examination of the evaluation of the outcome affecting intention revealed three protection initiatives that the public expects would increase the protection of forested watershed areas. These protection initiatives include (i) conservation campaigns for the protection of forested watershed areas, (ii) good management of fertilizers and pesticides in agricultural activity, and (iii) the establishment of a watershed conservation fund for the encroached forests. These findings provide useful information for establishing measures that would encourage behavioral intention by means of increasing the public's beliefs concerning the protection of forested watershed areas. In particular, the establishment of a watershed conservation fund appeared as the most significant protection initiative as the public perceived the funding would strengthen the linkage and effort between the public, the government, and non-governmental agencies in safeguarding forests [67,68].
From the perspective of subjective norm, four normative referents were the influential advisory groups. The extent of influence from each normative referent may differ. Still, the results indicate that the public would adopt behavior that supports forest protection when there is encouragement from groups of people that they trust [69]. Among these normative referents, the government was identified as the most important source of norm pressure. This is associated with the fact that the government plays a significant role in decision-making processes relevant to the policies that promote a positive impact on the environment [70]. When examining normative beliefs, the public's intention towards protection initiatives would be insignificant if the public perceived the expected outcome to be negative. Hence, the influence of normative referents, such as the government, is seen as the catalyst that motivates the public to comply with behaviors that would generate a positive outcome for forest protection. In the case of Cameron Highlands, the government plays a significant role in supporting the public's intention towards the forested watershed areas' protection initiatives. This is because environmental treats to the forested watershed areas could affect the vitality of agriculture and tourism activities in the area, which are essential for economic growth. The government's role in Cameron Highlands can be seen through certain measures that support the sustainable practice that generally generates positive outcomes that the public welcomes. For example, through the collaboration of the Department of Environment with non-governmental organizations such as the Regional Environmental Awareness of Cameron Highlands (REACH), the government provides training for farmers to avoid and minimize the use of chemical fertilizers and pesticides in their farms [71]. The government also offers extension funds to support farmers that have low income [72].
Apart from assessing the effect of attitude and the influence of social norms, the ability of the public to perceive their behavior is also examined in this study. In this regard, the public perception of whether (or not) they participate in protection initiatives depends on a few factors that help to facilitate or inhibit their behavioral intention. The identified factors are their attendance in seminars on sustainable agriculture practice and acceptance of the improvement in manure. As agricultural activity predominantly threatens the sustainability of forests in Cameron Highlands, the public revealed that these two factors could facilitate their behavioral intention towards participation in protection initiatives to protect forested watershed areas. The improvement in the manure management plan is deemed necessary to reduce biological pollution in Cameron Highlands [73]. Meanwhile, seminars on sustainable agriculture practice are viewed as a platform to educate the public on their role in implementing appropriate measures for environmental protection [74,75].
The findings of this study show that the respondents' behavioral intention is influenced by positive beliefs, favorable attitude and supportive social norms, in addition to the ability to perform protective behavior towards forested watershed areas. The TPB components are mostly positive and significantly correlated with behavioral beliefs, normative beliefs, and control beliefs. These findings indicate that the public supports the protection initiatives for the forested watershed areas in Cameron Highlands. As this study analyzes the public's behavioral intention, it highlights public tendency to participate in environmental protection. It is a timely and relevant public input to be integrated into the institutional management framework, particularly when addressing pressures from land-use change that threaten the sustainability of the critically important forested watershed areas.
However, this research is not without limitations. A few aspects need to be considered for further research, including the importance of environmental attributes of the forest, e.g., biodiversity conservation and habitat provisioning. These aspects are among the primary characteristics of the forested watershed areas critical to ecological balance and livelihood, yet very vulnerable to threats from land-use change. Hence, their inclusion as research attributes is highly relevant and recommended to be prioritized, especially when considering initiatives that would ensure the sustainability of forested watershed areas. Future research could also emphasize a better sampling procedure and sampling size to capture a more comprehensive range of behavioral intention, or to allow heterogeneity estimation of sociodemographic aspects. Lastly, the limited capacity of the measured variables had limited results and implications. The variables of attitude and subjective norm were seen to have a significant effect as compared to the perceived behavioral control. Therefore, it is recommended to widen the use of the perceived behavioral control variable in future studies to acquire more information about the public's capability to perceive their behavior.