Heterogeneity in citizens’ valuation of ecosystem services resulting from a lagoon public restoration project: Evidence from Tunisia

1 Heterogeneity in Citizens' Valuation of Ecosystem Services from a Mediterranean Sea-Connected Lagoon 1 Restoration: Evidence from Tunisia 2 Abstract 3 The North African region's lagoon ecosystems face numerous challenges, including urbanization, wastewater 4 pollution, and overexploitation. The negative impact of these challenges on the ecological health of lagoons is 5 evident, leading to socioeconomic consequences that have proven to be detrimental. Therefore, we conducted a 6 choice experiment to assess citizens' preferences for Mediterranean Sea-Connected lagoon restoration in Tunisia, 7 as a case study and their willingness to support the EcoPact endeavor to enhance the prevailing circumstances and 8 halter environmental degradation. This research devised two improvement scenarios and utilized a Latent Class 9 Model to gauge citizens' utility in a lagoon restoration. The results revealed two citizen classes, "Pro-restoration" 10 and "Reluctant to Restore". The majority are pro-restoration citizens willing to voluntarily pay (WTP) up to 11 $165.58 for one-year contribution for a high-impact scenario. The other class, Reluctant to Restore, appear to 12 recognize the value of the project's attributes, as evidenced by their significant WTP for high-level attributes. 13 However, they still prefer to maintain the current situation for other reasons, resulting in an insignificant WTP for 14 the overall high or medium scenario. The results showed that the aggregated benefits is very close to the required 15 project cost for the high-impact scenario, suggesting that the project is almost viable only if the improvement is 16 highly significant. Hence, further evaluation is required to validate these results. 17 Keywords: Bizerte citizens’ preferences, latent heterogeneity, choice experiment, lagoon restoration, developing 18 countries 19 JEL codes : Q01, Q2, Q56, Q57, Q58 20 21 22 23 24 25 26 27 28 29 30 2 1. Introduction 31 Lagoons are among the most productive and ecologically significant ecosystems, offering many goods 32 and services vital to local communities' well-being (El Zrelli et al. 2021). These dynamic environments serve as 33 crucial food sources, supporting fisheries and contributing to local economies and food security. Beyond their 34 direct provisioning of resources, lagoons also offer recreational opportunities, including water sports, bird 35 watching, and scenic spaces that enhance community leisure and tourism activities (Conde et al. 2015; Jorge et 36 al. 2023). Moreover, lagoons are essential in supporting various industrial processes, such as salt production and 37 aquaculture, which provide employment and stimulate economic growth in coastal regions. They deliver critical 38 ecosystem services, including water purification, climate regulation through carbon sequestration, and protection 39 against storm surges, thereby mitigating the impacts of climate change. Additionally, lagoons possess rich cultural 40 legacies, deeply embedded in the history and traditions of local communities, fostering a sense of identity and 41 heritage (Lopes and Videira. 2013; Newton et al. 2018). Their unique biodiversity (Tonin 2018), also makes them 42 prime sites for ecotourism, attracting visitors interested in experiencing the natural beauty and diverse wildlife of 43 these Mediterranean coastal landscapes (Wilson et al. 2023). However, despite their immense value, lagoons face 44 significant vulnerability and threats, primarily from pollution caused by human activities, including 45 industrialization, agriculture, and climate change, which profoundly affect water bodies. This susceptibility 46 manifests by introducing various organic, inorganic, and nutrient pollutants (Wilson et al. 2023). Excessive 47 amounts of pollutants, particularly nitrate and phosphate, can lead to the eutrophication of marine ecosystems 48 (Elizabeth and Joy 2018; Owa 2013). 49 Globally, the adverse impacts of marine pollution and freshwater contamination on human health and 50 their contribution to the loss of local sense of place and cultural identity have been extensively documented 51 (Mechler et al. 2019). For example, recent research by McNamara et al. (2021) investigated stakeholder 52 perceptions of non-economic loss and damage in Pacific island contexts, aligning with prior studies and 53 elucidating anticipated risks and impacts such as escalating temperatures, fish protein shortages, and substantial 54 biodiversity threats. Consequently, ecological restoration is essential for preserving marine ecosystem services, 55 fostering conservation efforts, and enhancing human welfare (Stainback et al. 2020; Paramana et al., 2023). 56 The body of literature highlights the necessity of lagoon restoration (Tuan et al., 2014; Clara., 2018; 57 Beharry-Borg and Scarpa. 2010; Smyth et al. 2009; Eggert and Olsson. 2009; Beharry-Borg and Scarpa, 2010; 58 Wang et al. 2013; Newton et al. 2018; Pacifico et al. 2025) prompting policymakers to enact directives to protect 59 ecological well-being, animal welfare, and water standards by aiming for sustainability. However, restoration has 60 financial costs to society. While pollution is a pressing issue in developing and developed countries, its 61 significance is particularly pronounced in developing nations due to the wide use of unsustainable industrial waste 62 management systems. Tunisia is a pertinent example of a developing country grappling with pollution challenges. 63 Croitoru and Sarraf (2010) documented the order of magnitude of marine pollution externalities in 64 Tunisia. They conducted a study to measure the cost of water degradation in Tunisia due to inappropriate 65 agricultural practices, transport, industry, and power generation. The estimated total cost was approximately 66 $165.8 million1 in 2004. 67 1 0.6% of the GDP in 2004 (INSTM, 2004) 3 At a regional level, Bizerte Lagoon, located in northern Tunisia, represents one of the most productive 68 marine ecosystems. It contributes to the country’s GDP through fishery products,2 aquaculture, shellfish farming, 69 and industrial activities. However, the region is suffering the consequences of misusing the lagoon. Untreated 70 water sewage, lack of treatment plants, wastewater discharges, and gas emissions from factories installed around 71 the lagoon represent the major causes of environmental degradation, making the lagoon unsuitable for reaction 72 activities and commercial fishing. Moreover, the fish and seafood caught in the lagoon are unsafe for human 73 consumption (El Zrelli et al. 2021). 74 Because of these concerns, the EcoPact3 project was implemented to enhance the socioeconomic and 75 environmental situation in the Bizerte Lagoon. In particular, the project aimed at (1) reducing industrial pollution 76 (i.e., atmospheric emission, liquid effluents, solid waste, and wastewater collection and treatment), (2) extending 77 the artisanal port4 of Manzel Abderrahmèn to reduce exposure to storms and increase its boat accommodation 78 capacity, and (3) developing an esplanade to the east of Manzel Abderrahmèn port to improve the lagoon frontage 79 of the region. Throughout an initial 5-year period, the project plans to significantly reduce indirect pollution 80 impacting the Mediterranean Sea through an integrated and concerted approach. 81 The existing literature indicates that citizens in developed countries are generally willing to pay for ecosystem 82 restoration (Anaya-Romero et al., 2016; Birol et al., 2006; Blasi et al., 2023; Martínez-Paz et al., 2013; Perni et 83 al., 2011; Xu et al., 2020). This study contributes to the limited body of research on estimating the economic 84 values of lagoons in developing countries (Ahmed et al., 2005; Ali et al., 2013; Diop et al., 2016; Ghermandi & 85 Nunes, 2013; Kairo et al., 2008; Lal et al., 2024; Rönnbäck et al., 2007). It aims to enhance our understanding of 86 how local communities perceive and prioritize various ecosystem attributes. Focusing on Tunisia, this study 87 expands the geographic scope of discrete choice experiment (DCE) applications in environmental valuation, an 88 area that remains underexplored in North Africa. It is the first lagoon valuation study conducted in Tunisia. It 89 provides insights into the trade-offs people are willing to make for improved lagoon management, better water 90 quality, and biodiversity conservation. 91 To our knowledge, no previous study has evaluated the benefits of restoring lagoons in Tunisia, particularly in 92 Bizerte. This gap can be attributed to two main factors: first, there is a lack of economic information on lagoon 93 restoration; second, few valuation studies have assessed the benefits or non-use values of lagoon restoration, 94 especially in developing countries. 95 Additionally, using traffic light indicators as a communication tool enhances accessibility for non-expert 96 respondents, ensuring that valuation outcomes reflect broader societal preferences. This study examines 97 preference heterogeneity, offering deeper insights into variations in public attitudes toward lagoon conservation. 98 These findings are particularly relevant for policymakers in developing countries, where financial and institutional 99 constraints often limit effective environmental management. They highlight the support of Bizerte citizens for 100 lagoon restoration, providing concrete evidence of the project's feasibility. The insights gained can help secure 101 further backing and ensure a thorough evaluation of the project's potential viability. By offering valuable support 102 to policymakers and illuminating social demand in Tunisia, this study may also guide future investments from 103 2 The total annual fishery production in 2020 was 62417 kg ≅ 62 tons (INSTM, 2020). 3 Details of the project are available at http://ecopact.tn. 4 The port is situated inside the Bizerte lagoon, approximately 5 km from Manzel Jemil and 4 km from Bizerte. 4 international organizations, such as the Food and Agriculture Organization (FAO) and the World Bank (WB), 104 fostering opportunities for development and investment in Tunisia. 105 The article is structured as follows. The next section overviews the Bizerte Lagoon, followed by a description of 106 the design and the survey’s conduction. Subsequently, the results are presented in Section 3, while the discussion 107 and conclusions are offered in Section 4. 108 2. Methods 109 2.1. Study Area 110 The Bizerte Lagoon is located on the northern coast of Tunisia. It has an area of 150 km² and an average depth of 111 7 m (Fig. 1). The lagoon is connected to the Mediterranean Sea by a 7-km-long channel and Lake Ichkeul by the 112 Tinja River, which supplies it with irregular fresh water. According to the National Institute of Statistics (INS,5 113 2023) census, approximately 300,0006 people live around it (90% are in the city of Bizerte). 114 Bizerte is recognized for its industrial heritage, tourism, agriculture, fishing, and a significant commercial port, 115 supported by free trade zones that encourage investment with tax exemptions (Ministry of Environment7, 2016). 116 The city is home to around 274 industrial companies, including 172 exporters, with 33.6% in textiles. 117 The region's agriculture produces 67,280 tons of fodder, 238,409 tons of cereals, and significant quantities of 118 market gardening products, red meat, milk, white meat, and eggs. Additionally, it generates 6,229 tons of fishery 119 products from five ports and 1,280 fishing fleets (Tunisian National Institute of Statistics8, 2015). 120 The lagoon has an urbanized and industrialized shoreline. Moreover, it features 10 industrial zones 121 spanning 250 hectares, with notable industrial activities are readily present, including heavy industries like 122 SOTULUB and Bizerte Cement Factories employing over 45,858 people (located in the vicinity of Bizerte), dye 123 works, and metallurgy (located in Menzel Jemil), making it one of the most threatened lagoons in Tunisia (Alves 124 Martins et al. 2015). 125 126 127 128 129 130 131 132 133 134 5 For more information, see: https://www.ins.tn/sites/default/files/publication/pdf/Estimation%20de%20%20la%20population%201er%20Janvier%20202 3.pdf 6 Represents approximately 187,000 in Bizerte center, almost 66,000 in Manzel Bourguiba, 31,000 in Manzel Jmil, and 21,000 in Manzel Abderrahmen. The total population of Bizerte was 600,012 inhabitants in 2023. 7 https://www.ins.tn/en/statistiques/45 8 https://www.ins.tn/en/statistiques/45 5 135 Fig. 1 Bizerte Lagoon, Main Industrial Agglomeration, source: Nasri et al. (2022) 136 The Bizerte Lagoon has been used as the principal dumping place for water discharge directly or after 137 primary treatment for several decades. However, it does not meet the standards set by the central government for 138 used water treatment. The three wastewater treatment plants built in 1997 to treat the water before being dumped 139 in the lagoon have decayed. Previous studies conducted in Bizerte (Barhoumi 2014; Boukef et al. 2010; Toumi et 140 al. 2019) showed that the level of water treatment is low and the quality of treated water is non-compliant with 141 discharge standards, particularly in terms of the levels of nitrate, phosphorus, polycyclic aromatic hydrocarbons, 142 and heavy metals. These pollutants can be present in the atmosphere or the soil or especially dissolved in aquatic 143 environments that are much more contaminated than others (Barhoumi 2014). Moreover, the city of Bizerte has 144 experienced atmospheric pollution due to industrial emissions in the region (Barhoumi 2014), along with human 145 population growth around the lagoon and maritime traffic that have increased wastewater discharge into the 146 lagoon. 147 The lagoon is historically known for artisanal fishing activities, aquaculture, and, specifically, shellfish 148 farming. The production from shellfish farming in the lagoon has decreased for over a decade (General Direction 149 of Fishery Products and Aquaculture [DGPA] 2018) due to the rise in water temperature and the decrease in the 150 level of dissolved O2 (Bousbih9 2015). 151 2.2. Data Collection: Choice Experiment-Based Survey 152 This study used a DCE to investigate citizens’ valuation of the benefits and the costs of restoration plans that 153 EcoPact is considering implementing in the Bizerte Lagoon. A DCE is a quantitative research technique that 154 presents individuals with alternative scenarios and asks them to state their preferred scenario (Hensher et al., 155 2015). Several attribute levels describe each alternative scenario. Individuals’ responses are then used to determine 156 whether their preferences are significantly influenced by the attribute levels considered in the DCE. The responses 157 9 Press at: https://inkyfada.com/fr/2015/07/27/bizerte-pollution-lac-peche-tunisie/ 6 are also used to determine the relative importance of the attributes. Individuals’ WTP for the attribute levels of 158 the alternative scenarios can also be estimated if a monetary attribute (e.g., the price or cost of each alternative) is 159 considered in the choice experiment. 160 An initial and exhaustive list of attributes was prepared based on an extensive literature review on 161 economic values of lagoons and wetlands (Hanley et al. 1998; Brouwer et al. 1999; Brander and Schuyt 2004; 162 Birol et al. 2006; Christie et al. 2006; Campbell et al. 2008; Meyerhoff et al. 2009, Barhoumi 2014; Dang et al. 163 2022, Pacifico et al. 2024, 2025) to identify the most relevant attributes to consider in the choice experiment. The 164 literature-based list of attributes and their levels was then refined using three focus groups (totaling 30 persons) 165 of residents, experts, and scientists where the main objective is scoring the most relevant attributes according to 166 the context, the urgent public intervention, and local priorities. This attribute selection and refinement process by 167 focus groups is crucial for our choice experiment design, as it ensures that the attributes chosen are both relevant 168 to the specific wetland or lagoon context and meaningful to the local population. Moreover, this participatory 169 approach also helps to capture any unique local factors that might not be apparent from the literature review alone. 170 The attributes typically considered in such studies include environmental factors such as biodiversity and 171 water quality and socio-economic aspects like fisheries production and recreational opportunities (Birol et al. 172 2006; Tsegaw 2012; Dixon et al. 2021; Pacifico et al. 2024). 173 After the focus group discussions, the attributes are typically scored and ranked based on their perceived 174 importance and relevance. This process helps narrow the list to a manageable number of attributes for the choice 175 experiment, usually between 4 to 6 attributes (Bliemer & Collins, 2016). The final selection of attributes 176 considered in this study were water quality, biodiversity, recreational facilities, reduced gas emissions, and the 177 cost of improvement which represent a one-payment cost to be voluntary paid. A description of the attributes and 178 their corresponding levels is displayed in Table 1 below. 179 Water quality, biodiversity, reduced gas emissions, and recreational facilities were defined as categorical 180 attributes with three levels each. Each was color-coded:10 red for no improvement (i.e., the status quo), amber for 181 moderate improvement, and green for high improvement. To further ease the understanding of the attribute levels 182 and avoid confusion (Osman and Thornton 2019), qualitative terms (i.e., low, medium, and high) were used to 183 describe the traffic light color-coded levels (Fig. 2). Many studies supported the use of visual saliency instead of 184 only including texts or values to avoid the hypothetical bias (Shr et al., 2019; Delong et al., 2021). 185 Respondents were invited to contribute voluntarily to support a five-year project through a one-time 186 payment. We based our assumption on the belief that the improvement cost would be covered through voluntary 187 payments. These contributions are likely more effective in Tunisia due to cultural norms and a social familiarity 188 with charitable giving. Additionally, the proposed cost levels were informed by feedback from three focus groups, 189 which helped refine the list of attributes considered for the questionnaire. 190 191 192 10 This study adopted the concept of visual saliency, where both images and text were presented to respondents. This approach aimed to mitigate hypothetical bias and prevent confusion by providing a clear and tangible representation of the attributes evaluated. Recent research by Netusil et al. (2023) and Shr et al. (2019) showed that the respondents strongly preferred attributes represented by both image and text. 7 Table 1 Description of Attribute used in the DCE and Their Levels 193 194 The combination of the five attributes and their levels resulted in 405 possible combinations (i.e., 3^4 * 195 5). Ngene software was used to generate a Bayesian D-optimal (i.e., fractional) design that allowed robust 196 estimation of all main effects (ChoiceMetrics, 2018). The priors for the fractional design were obtained from a 197 pilot of 30 respondents. The final experimental design (D-error = 0.07) consisted of 18 choice cards, each 198 comprising three alternatives—two pairs of hypothetical options of improvements and the status quo (SQ). The 199 final design had three blocks (i.e., each respondent was presented with only six of 18 choice sets). An example of 200 one of the choice example cards used in this study is shown in Figure 2. 201 202 11 TND: Tunisian dinar Attributes Attribute Levels Variable Description in Model Estimation (Effect Coding) Water Quality High (Green) It refers to a high improvement level where water quality and clarity respond to the standards without a nasty smell and rehabilitated sanitation networks (1 = high water quality [green color], 0 otherwise). Medium (Amber) It refers to a medium level of improvement where water quality, clarity, and smell improved compared to the current situation (1 = medium water quality [amber color], 0 otherwise). Low (Red) It refers to poor water quality and clarity, with a nasty smell (–1 = low water quality [red color]). Biodiversity High (Green) It refers to a high level of improvement where fauna (well-diversified species) and marine flora (nutritive algae) are diversified, valorisation of invasive species (1 = high biodiversity [green color], 0 otherwise). Medium (Amber) It refers to a medium level of improvement where marine fauna and flora are more diversified compared to the current situation (1 = medium biodiversity level [amber color], 0 otherwise). Low (Red) It refers to a situation where some species have disappeared while invasive species have appeared (–1 = low biodiversity [red color]). Recreational Facilities High (Green) It refers to a high level of improvement where all activities are accessible (i.e., fishing, swimming, and walking) due to the development of the coastal infrastructure (1 = high recreational facilities [green color], 0 otherwise). Medium (Amber) It refers to a medium improvement level where coastal infrastructure is moderately developed and at least one activity is possible (1 = medium recreational facilities [amber color], 0 otherwise). Low (Red) It refers to a situation where the coastal infrastructure is not developed, and accessibility to activities is almost limited (–1 = low recreational facilities [red color]). Decreased Gas Emissions High (Green) It refers to a high improvement level where air quality is good (filters installed), good visibility (i.e., almost clear emissions, decreased CO2 emissions), and no smell (1 = high decreased gas emissions [green color], 0 otherwise). Medium (Amber) It refers to a medium improvement level where average air quality (average CO2 level in the atmosphere), average visibility (less gray emissions than before), and less smell compared to the current situation (1 = medium decreased gas emissions level [amber color], 0 otherwise.). Low (Red) It refers a situation with poor air quality is poor (significant CO 2 in the atmosphere does not meet discharge standards), poor visibility (gray emissions), and bad smell (–1 = low decreased gas emissions [red color]). Voluntary Improvement Cost TND11 0 (no payment), TND 30, TND 60, TND 90, TND 120 (40 EUR), TND 150 (continuous variable). 8 203 204 205 206 207 208 209 210 Fig. 2 Choice Example Cards 211 The survey was designed with a structured approach, divided into four sections to ensure comprehensive 212 data collection and accuracy. The first section introduced the study's framework, outlining its objectives and 213 methodology while emphasizing the voluntary nature of participation. Participants were required to review and 214 sign an informed consent form, which assured them of anonymity and clarified that the research had no 215 commercial purposes. The second section focused on the choice task and began with a detailed explanation of the 216 survey’s context, objectives, and the importance of providing unbiased responses. To minimize hypothetical bias, 217 reminders and "cheap talk" techniques were included. Respondents were then presented with randomly assigned 218 choice cards, designed to enhance realism and reduce potential bias. Each choice question included three 219 alternatives: two hypothetical improvement scenarios and a status quo (SQ) option. The first two alternatives were 220 described using five key attributes detailed in Table 1, with variations across different choice questions, while the 221 SQ option remained unchanged, reflecting the actual condition of the lagoon at the time of data collection. The 222 final section gathered socioeconomic information from respondents to provide context for the findings. 223 The survey was conducted face-to-face in Bizerte between February and April 2023, engaging 371 citizens. 224 However, 10 respondents were identified as zero protestors and were excluded from the final dataset based on the 225 criteria12 stated by De Jong et al. (2023) and Xu et al. (2020). As a result, the final data used in the analysis was 226 based on the responses of 361 respondents. 227 2.3. Econometric Modeling of DCE Data 228 According to Hensher et al. (2015), the total utility 𝑈𝑈𝑗𝑗𝑗𝑗 derived by an individual n from choosing an alternative j 229 is equal to the sum of two components, a deterministic component 𝑉𝑉𝑗𝑗𝑗𝑗 and an indeterministic component εnj 230 assumed to be independent and identically distributed: 231 𝑈𝑈𝑗𝑗𝑗𝑗 =𝑉𝑉𝑗𝑗𝑗𝑗 +𝜀𝜀𝑗𝑗𝑗𝑗 (1) 232 12 They choose the SQ in all the choice sets. 9 233 Assuming that the deterministic component of the utility is linear-in-parameter, equation (1) can be written 234 as follows: 235 𝑈𝑈𝑗𝑗𝑗𝑗𝑛𝑛 =𝛽𝛽𝛽𝛽𝑗𝑗𝑗𝑗𝑛𝑛 +𝜀𝜀𝑗𝑗𝑗𝑗𝑛𝑛 (2) 236 where 𝛽𝛽 denotes the K × 1 vector of unknown utility parameters and, 𝛽𝛽𝑗𝑗𝑗𝑗𝑛𝑛 represents the attribute levels. 237 Average respondents’ preferences and WTP were estimated using the conditional logit model (CLM). 238 A Hausman test was performed to evaluate whether the IIA assumption holds; our test results indicated that this 239 assumption does not hold, suggesting that the choices among alternatives are interdependent and necessitating 240 alternative models that can account for this interdependence (See Table 2). 241 Table 2. Hausman Test to Test for IIA Assumption 242 Hausman Test of Independence of Irrelevant Alternatives for sample Dropped Alternatives 𝝌𝝌𝟐𝟐 Degree of freedom Probability Alternative 1 66.19 10 0.00 Alternative 2 48.31 10 0.00 Alternative 3 (SQ) ND13 10 Could not carry out Hausman test for the IIA However, to investigate respondents’ preferences and WTP heterogeneity14 a latent class model (LCM) 243 was also estimated. In the LCM, respondents were grouped in a finite number of identifiable classes, allowing the 244 respondents’ preferences to be heterogeneous across classes but homogeneous within each class (Greene and 245 Hensher 2003; Sinha et al. 2021). 246 In the LCM, the probability of individual n in class c choosing alternative i from a particular set of 247 alternatives J was as follows: 248 Pns|c= exp (β′cXni) ∑exp (β′cXn𝑛𝑛𝑛𝑛) J j=1 (3) 249 where βs is the parameter vector of class s associated with the vector of explanatory choice attributes Xni. 250 Additionally, the analyst considered a classification model as a function of some individual-specific characteristics 251 to determine the allocation of individuals to the c classes. 252 According to Greene and Hensher (2003), the probability that an individual n belongs to latent class c is given by 253 Pn∈c = exp (θ′cZn) ∑exp (θ′cZn) C C=1 (4) 254 255 where 𝑧𝑧𝑖𝑖𝑗𝑗 is the vector of respondents’ socioeconomic characteristics, and θ is a set of parameters to be 256 estimated. Moreover, the log-likelihood of the LCM can be expressed as follows: 257 13 Undefined 14 The estimation of the Random Parameter Logit (RPL) model showed a non-significant price coefficient, likely due to the exponential function amplifying the standard deviation and leading to a low z-value. Using the "logitr" R package, we found that treating price as a random parameter complicated model convergence in willingness to pay (WTP) space. To assess the price parameter's impact, we tested several distributions, including normal and lognormal. We fixed the scale parameter and treated other parameters as normally distributed. However, assuming a constant price implies uniform sensitivity across individuals, which may bias WTP estimates by ignoring variations in price sensitivity. 16 sustainability. This finding aligns with a study conducted by Nahar et al. (2023) in Dhaka, Bangladesh, which 397 emphasized that respondents with high environmental awareness exhibited a greater understanding of 398 environmental laws and regulations, indicating a positive relationship between awareness and environmental 399 concerns. Third, this finding may also be justified by the importance that respondents assign to recreational 400 facilities where citizens are willing to contribute TND 58.20 ($19.40) (Class 1) to TND 79.46 ($26.48) (Class 2) 401 for a high-impact scenario, representing the most preferred attribute. Hence, enhancing coastal infrastructure could 402 serve as an asset for the citizens of Bizerte by allowing them to reclaim the values and activities they have lost 403 due to pollution over time. Reviving activities such as swimming, artisanal fishing, and walking through improved 404 coastal infrastructure would enhance the residents’ quality of life while helping restore cultural and recreational 405 practices that hold significance for the community. Moreover, the study highlighted the importance that citizens 406 assigned to reducing gas emissions. They were willing to pay TND 89.01 ($29.67) (Class 1) to TND 104.31 407 ($34.77) (Class 2) to reduce the maximum of gas emissions. 408 The WTP results for Class 1, where members were willing to pay up to TND 55.82 ($18) to enhance 409 biodiversity, align with comparative WTP analysis results from similar studies (Owuor et al., 2019; Chen & Ting 410 Cho, 2019) where citizens were willing to pay TND 73.6 ($23) to improve biodiversity. However, the findings 411 regarding biodiversity in Class 2 were unexpected and differed from the existing literature. Specifically, 412 biodiversity was not deemed important at the medium and high levels. People may be more focused on short-413 term, tangible results that they can see and touch immediately (O’Donoghue and Rabin 2000). Furthermore, 414 drawing from López et al. (2007), non-economic motives for biodiversity WTP, such as familiarity and biophilia, 415 may play a significant role. Individuals may be less inclined to contribute to biodiversity conservation efforts if 416 they lack a personal connection or appreciation for nature. Thus, it may reflect how to increase in biodiversity 417 awareness among citizens. 418 Results from the viability study, in the present value, suggest that voluntary payments could almost fully 419 fund the project, assuming the population perceives a high benefit from the intervention, suggesting that the 420 EcoPact project is almost viable only if the improvement is highly significant, which may be because respondents 421 are willing to pay much more for the high-improvement scenario since it entails more substantial changes and 422 interventions than the medium-improvement scenario. These interventions will likely result in more 423 significant environmental, social, and economic benefits, such as improved water quality, enhanced biodiversity, 424 and increased recreational opportunities. 425 The difference between the medium and high-impact scenarios is significant. For the medium impact 426 scenario, the weighted WTP is only USD 37.42 million, far below the project cost. This highlights 427 the sensitivity of the population to contribute based on their perception of the benefits. Suppose the perceived 428 impact of the project is low or moderate. In that case, the funding gap will be much larger, affecting the project's 429 viability or requiring additional funding sources such as government subsidies. Otherwise, governmental efforts 430 will be needed to reduce project implementation costs and cover the required average investment and operational 431 expenses so that aggregated WTP for medium level can cover the costs because medium improvements are more 432 attainable and less costly than a high-impact scenario. Consequently, the medium-impact scenario may not recover 433 its total costs without a government subsidy. 434 The article adds to the existing literature by providing new insights into the increasing public concern 435 over the degradation of the Bizerte Lagoon’s ecosystem, indicating a social demand for restoration. The results 436 17 emphasize the substantial net benefits derived from both ecological and social aspects of restoring the Bizerte 437 Lagoon in a way that by providing the economic values of ecosystem services, this study can serve as a baseline 438 that policy-makers are based on to conduct the benefit transfer method and assess the TEV or precisely ecosystem 439 services values in other similar locations. It can serve also as an input to assess the ongoing projects or research 440 aiming to set best management for marine ecosystems such lagoons. 441 Successful implementation relies on harmonious collaboration between institutions and effective 442 government involvement. 443 Research indicates that individuals place significant value on public goods and often demonstrate a 444 willingness to contribute financially to their preservation. This finding suggests a strong potential for community 445 support for public policies aimed at the conservation or restoration of natural spaces that provide essential 446 ecosystem services. Moreover, citizens can play a pivotal role in financing these initiatives, which addresses a 447 major challenge—financial constraints—often encountered in the implementation of public policies. These 448 insights can be instrumental in designing effective programs; by ensuring they achieve a meaningful impact; we 449 can enhance public support and engagement. For public authorities aiming to promote the protection of public 450 goods, such as lagoons, it is crucial to understand how to effectively engage different segments of the population. 451 Tailoring approaches for Class 1 and Class 2 citizens will be key to raising awareness and fostering a collaborative 452 environment for conservation efforts. In this way, policies should emphasize the organization of campaigns to 453 effectively raise awareness about environmental issues among older individuals using spoken local dialects to 454 ensure accessibility and the ability to understand the technical terms for those with limited literacy skills and 455 strengthen collaboration with non-governmental organizations to facilitate information-sharing and dissemination 456 of environmental knowledge, as emphasized through the results obtained from Class 2. It is also crucial to maintain 457 a solid foundation in environmental education for young students to ensure similar results in WTP for Class 1 and 458 to achieve better outcomes for Class 2. Disseminating environmental education in schools can also be a solution. 459 Furthermore, by teaching kids the environmental importance of the lagoon, they can grow up with greater 460 awareness, which will make it easier to influence their behavior positively when they become adults. Tunisian 461 government must supplement existing environmental legislation with regular monitoring of wastewater treatment 462 plants and gas filtration to ensure that factories surrounding the lagoon comply with standard norms. Deviations 463 from these norms must be met with appropriate penalties to deter non-compliance. 464 In conclusion, this enchanting lagoon holds immense significance not just for the residents of Bizerte but 465 for the entire Mediterranean region. Its intricate connection to the sea underscores the importance of its restoration; 466 by rejuvenating this vital ecosystem, we actively contribute to the preservation of the Mediterranean—a lifeblood 467 for countless communities surrounding its shores. “This lagoon is more than a body of water; it is a sanctuary of 468 ecological diversity and a source of cultural heritage, deeply intertwined with the lives of those who inhabit this 469 beautiful coastal landscape”. 470 471 472 473 474 475 476 18 References 477 Ahmed M, Chong C K, Cesar H S J (2005) Economic valuation and policy priorities for sustainable management 478 of coral reefs (Rev. ed) WorldFish Center. 479 Ali M, Fjeldstad, O H, Sjursen I H (n.d.) CITIZENS’ ATTITUDES TOWARDS TAXATION IN KENYA TANZANIA 480 UGANDA AND SOUTH AFRICA. 481 Alves Martins, M V, Zaaboub N Aleya, Frontalini F, Pereira E, Miranda P, Mane M, Rocha F, Laut L, El Bour, 482 M (2015) Environmental Quality Assessment of Bizerte Lagoon (Tunisia) Using Living Foraminifera 483 Assemblages and a Multiproxy Approach. PLOS ONE, 10(9), e0137250. 484 https://doi.org/10.1371/journal.pone.0137250 485 Anaya Romero, M, Muñoz Rojas M, Ibáñez B, Marañón T (2016) Evaluation of forest ecosystem services in 486 Mediterranean areas A regional case study in South Spain. Ecosystem Services, 20, 82–90. 487 https://doi.org/10.1016/j.ecoser.2016.07.002 488 Balcombe K, Fraser I, Falco S D (2010) Traffic lights and food choice: A choice experiment examining the 489 relationship between nutritional food labels and price. Food Policy, 35(3), 211–220. 490 https://doi.org/10.1016/j.foodpol.2009.12.005 491 Barhoumi B (2014) Biosurveillance de la pollution de la lagune de Bizerte (Tunisie) par l’analyse comparée des 492 niveaux de contamination et de l’écotoxicité des sédiments et du biote. 349. 493 Birol E, Karousakis K, Koundouri P (2006) Using a choice experiment to account for preference heterogeneity in 494 wetland attributes: The case of Cheimaditida wetland in Greece. Ecological Economics, 60(1), 145–156. 495 https://doi.org/10.1016/j.ecolecon.2006.06.002 496 Bliemer M C J , and Collins A T (2016) On determining priors for the generation of efficient stated choice 497 experimental designs. Journal of Choice Modelling, 21, 10–14. 498 https://doi.org/10.1016/j.jocm.2016.03.001 499 Boukef I, Mejri S, Mraouna R, Bejaoui B, Belhassan M, Harzallah A, Boudabous A, Elbour M (n.d.) Distribution 500 spatiale des populations de Vibrionaceae thermotolérants dans une lagune côtière (lagune de Bizerte: 501 Nord-Tunisie). 502 Boxall P C, and Adamowicz W L (2002) Understanding heterogeneous preferences in random utility models: a 503 latent class approach. Environmental and resource economics, 23, 421-446. 504 Brander L, and Schuyt K (2004) The economic values of the world's wetlands. 505 Conde D, Vitancurt J, Rodríguez Gallego L, De Álava D, Verrastro N, Chreties C, Solari S, Teixeira L, Lagos X, 506 Piñeiro G, Seijo L, Caymaris H, Panario D (2015) Solutions for Sustainable Coastal Lagoon 507 Management. In Coastal Zones (pp. 217–250). Elsevier. https://doi.org/10.1016/B978-0-12-802748-508 6.00013-9 509 Croitoru, L, and Sarraf, M (2010). The Cost of Environmental Degradation. 192. 510 Dang K B, Phan T T H, Nguyen T T, Pham T P N, Nguyen M H, and Hoang T T H (2022) Economic valuation 511 of wetland ecosystem services in northeastern part of Vietnam. Knowledge & Management of Aquatic 512 Ecosystems, (423), 12. https://doi.org/10.1051/kmae/2022010 513 De Jong L, Schmidt T, Stahmeyer J T, Eberhard S, Zeidler J, Damm K (2023) Willingness to provide informal 514 care to older adults in Germany : A discrete choice experiment. The European Journal of Health 515 Economics, 24(3), 425–436. https://doi.org/10.1007/s10198-022-01483-5 516 19 Diederich J, Goeschl T (2014) Willingness to Pay for Voluntary Climate Action and Its Determinants: Field-517 Experimental Evidence. Environmental and Resource Economics, 57(3), 405–429. 518 https://doi.org/10.1007/s10640-013-9686-3 519 Dixon A, Wood A. and Hailu A Wetlands in Ethiopia: Lessons From 20 years of Research, Policy and 520 Practice. Wetlands 41, 20 (2021). https://doi.org/10.1007/s13157-021-01420-x 521 El Zrelli R, Yacoubi L, Wakkaf T, Castet S, Grégoire M, Mansour L, Courjault Radé P, Rabaoui L (2021) Surface 522 sediment enrichment with trace metals in a heavily human-impacted lagoon (Bizerte Lagoon, Southern 523 Mediterranean Sea): Spatial distribution, ecological risk assessment, and implications for environmental 524 protection. Marine Pollution Bulletin, 169, 112512. https://doi.org/10.1016/j.marpolbul.2021.112512 525 Elahi E, Khalid Z, and Zhang Z (2022) Understanding farmers’ intention and willingness to install renewable 526 energy technology: A solution to reduce the environmental emissions of agriculture. Applied Energy, 527 309, 118459. https://doi.org/10.1016/j.apenergy.2021.118459 528 Ghermandi A, Nunes P A L D (2013) A global map of coastal recreation values: Results from a spatially explicit 529 meta-analysis. Ecological Economics, 86, 1–15. https://doi.org/10.1016/j.ecolecon.2012.11.006 530 Greene W H, Hensher D A (2003) A latent class model for discrete choice analysis: Contrasts with mixed logit. 531 Transportation Research Part B: Methodological, 37(8), 681–698. https://doi.org/10.1016/S0191-532 2615(02)00046-2 533 Haile T A (2012) Application of Choice Experiment to Value Improved Wetland Attributes: The Case of Wetlands 534 around Jimma Town South Western Ethiopia. Doctoral thesis, 535 https://repository.ju.edu.et/bitstream/handle/123456789/2019/TSEGAW%20%20ABEBE%20HAILE%536 20(MSc.THESIS).pdf?sequence=1&isAllowed=y 537 Hensher D A, Rose J M, Greene W H (2015) Applied Choice Analysis (2nd ed.). Cambridge University Press. 538 https://doi.org/10.1017/CBO9781316136232 539 Kairo J G, Lang’at J K S, Dahdouh Guebas F, Bosire J, Karachi M (2008) Structural development and productivity 540 of replanted mangrove plantations in Kenya. Forest Ecology and Management, 255(7), 2670–2677. 541 https://doi.org/10.1016/j.foreco.2008.01.031 542 Lal P, Lim Applegate H, Scoccimarro M (2024) The Adaptive Decision-Making Process as a Tool for Integrated 543 Natural Resource Management: Focus, Attitudes, and Approach. 544 Lopes R, Videira N (2013) Valuing marine and coastal ecosystem services: An integrated participatory 545 framework. Ocean & Coastal Management, 84, 153–162. 546 https://doi.org/10.1016/j.ocecoaman.2013.08.001 547 Martínez Paz J M, Perni A, Martínez Carrasco F (2013) Assessment of the Programme of Measures for Coastal 548 Lagoon Environmental Restoration Using Cost–Benefit Analysis. European Planning Studies, 21(2), 549 131–148. https://doi.org/10.1080/09654313.2012.722923 550 McNamara K E, Westoby R, Clissold R, Chandra A (2021) Understanding and responding to climate-driven non-551 economic loss and damage in the Pacific Islands. Climate Risk Management, 33, 100336. 552 https://doi.org/10.1016/j.crm.2021.100336 553 Mechler R, Bouwer L M, Schinko T, Surminski S, Linnerooth Bayer J (Eds) (2019) Loss and Damage from 554 Climate Change: Concepts, Methods and Policy Options. Springer International Publishing. 555 https://doi.org/10.1007/978-3-319-72026-5 556 20 Meyer A (2015) Does education increase pro-environmental behavior? Evidence from Europe. Ecological 557 Economics, 116, 108–121. https://doi.org/10.1016/j.ecolecon.2015.04.018 558 Missaoui S, Rahmani D, Thabet C, Feás J, Gíl J M, and Akaichi F (2023) Assessment of social demand for 559 environmental and cultural heritage preservation: Evidence from a discrete choice experiment in Tunisia. 560 Frontiers in Environmental Economics, 2, 1234249. https://doi.org/10.3389/frevc.2023.1234249 561 Nahar N, Hossain Z, and Mahiuddin S (2023) Assessment of the environmental perceptions, attitudes, and 562 awareness of city dwellers regarding sustainable urban environmental management: A case study of 563 Dhaka, Bangladesh. Environment, Development and Sustainability, 25(8), 7503–564 7531. https://doi.org/10.1007/s10668-022-02354-y 565 Nasri A, Mhadhbi T, Allouche M, Hannachi A, Demey H, Aïssa P, and Mahmoudi E (2022) Persistent Organic 566 Pollutants in the Bizerte Lagoon Ecosystem: Occurrence, Distribution, and Ecotoxicological Assessment 567 Using Marine Organism. IntechOpen. doi: 10.5772/intechopen.100549 568 Netusil N R, Dissanayake S T M, Lavelle L, Ando A W, Wells K K (2023) Does presentation matter? An analysis 569 of images and text in a choice experiment of green roofs. Q Open, 3(1), qoad010. 570 https://doi.org/10.1093/qopen/qoad010 571 Newton, A., Brito, A. C., Icely, J. D., Derolez, V., Clara, I., Angus, S., ... & Khokhlov, V. (2018). Assessing, 572 quantifying and valuing the ecosystem services of coastal lagoons. Journal for nature conservation, 44, 573 50-65. https://doi.org/10.1016/j.jnc.2018.02.009 574 O’Donoghue T, Rabin M (2000) The economics of immediate gratification. Journal of Behavioral Decision 575 Making, 13(2), 233–250. https://doi.org/10.1002/(SICI)1099-0771(200004/06)13:2<233::AID-576 BDM325>3.0.CO;2-U 577 Osman M, Thornton K (2019) Traffic light labelling of meals to promote sustainable consumption and healthy 578 eating. Appetite, 138, 60–71. https://doi.org/10.1016/j.appet.2019.03.015 579 Pacifico AM, Mulazzani L and Malorgio G (2025) Are the economic valuations of marine and coastal ecosystem 580 services supporting policymakers? A systematic review and remaining gaps and challenges. Front. Mar. 581 Sci. 11:1501812. doi: 10.3389/fmars.2024.1501812 582 Pacifico A M, Brigolin D, Mulazzani L, Semeraro M, and Malorgio, G (2024) Managing marine aquaculture by 583 assessing its contribution to ecosystem services provision: The case of Mediterranean mussel, Mytilus 584 galloprovincialis. Ocean & Coastal Management, 259, 107456. 585 Paramana T, Dassenakis M, Bassan N, Dallangelo C, Campostrini P, Raicevich S, et al (2023). Achieving 586 coherence between the marine strategy framework directive and the maritime spatial planning directive. 587 Mar. Policy 155, 105733. doi: 10.1016/j.marpol.2023.105733 588 Perni A, Martínez Carrasco F, Martínez-Paz J (2011) Economic valuation of coastal lagoon environmental 589 restoration: Mar Menor (SE Spain). Ciencias Marinas, 37(2), 175–190. 590 https://doi.org/10.7773/cm.v37i2.1889 591 Plavsic S (n.d) An Investigation of Gender Differences in Pro-environmental Attitudes and Behaviors. 592 Rahmani D, Loureiro M L (2019) Assessing drivers’ preferences for hybrid electric vehicles (HEV) in Spain. 593 Research in Transportation Economics, 73, 89–97. https://doi.org/10.1016/j.retrec.2018.10.006 594 Ramstetter L, Habersack F (2020) Do women make a difference? Analysing environmental attitudes and actions 595 of Members of the European Parliament. Environmental Politics, 29(6), 1063–1084. 596 https://doi.org/10.1080/09644016.2019.1609156 597 21 Rodrigues-Filho J L, Macêdo R L, Sarmento H et al From ecological functions to ecosystem services: linking 598 coastal lagoons biodiversity with human well-being. Hydrobiologia 850, 2611–2653 (2023). 599 https://doi.org/10.1007/s10750-023-05171-0 600 Rönnbäck P, Crona B, Ingwall L (2007) The return of ecosystem goods and services in replanted mangrove 601 forests: Perspectives from local communities in Kenya. Environmental Conservation, 34(04). 602 https://doi.org/10.1017/S0376892907004225 603 Schaak H, and Musshoff O (2020) Public preferences for pasture landscapes in Germany—A latent class analysis 604 of a nationwide discrete choice experiment. Land Use Policy, 91, 104371. 605 Shr YH (Jimmy), Ready R, Orland B, Echols S (2019) How Do Visual Representations Influence Survey 606 Responses? Evidence from a Choice Experiment on Landscape Attributes of Green Infrastructure. 607 Ecological Economics, 156, 375–386. https://doi.org/10.1016/j.ecolecon.2018.10.015 608 Stainback G A, Lai J H, Pienaar E F, Adam D C, Wiederholt R, Vorseth C (2020) Public preferences for ecological 609 indicators used in Everglades restoration. PLOS ONE, 15(6), e0234051. 610 https://doi.org/10.1371/journal.pone.0234051 611 Tianyu J, Meng L (2020) Does education increase pro-environmental willingness to pay? Evidence from Chinese 612 household survey. Journal of Cleaner Production, 275, 122713. 613 https://doi.org/10.1016/j.jclepro.2020.122713 614 Tonin S (2018). Economic value of marine biodiversity improvement in coralligenous habitats. Ecological 615 indicators, 85, 1121-1132. https://doi.org/10.1016/j.ecolind.2017.11.017 616 Toumi H, Abidli S, Bejaoui M (2019) Microplastics in freshwater environment: The first evaluation in sediments 617 from seven water streams surrounding the lagoon of Bizerte (Northern Tunisia). Environmental Science 618 and Pollution Research, 26(14), 14673–14682. https://doi.org/10.1007/s11356-019-04695-0 619 Weller B E, Bowen N K, and Faubert S J (2020) Latent class analysis: a guide to best practice. Journal of black 620 psychology, 46(4), 287-311. DOI: 10.1177/0095798420930932 621 Whittington, D (2010) What have we learned from 20 years of stated preference research in less-developed 622 countries? Annual Review of Resource Economics, 2, 209–236. 623 https://doi.org/10.1146/annurev.resource.012809.103908 624 Wilson W W, Haiges R, Christe K (2023) Assessing, quantifying and valuing the ecosystem services of coastal 625 lagoons. https://doi.org/10.2139/ssrn.4417034 626 Xu L, Ao C, Mao B, Cheng Y, Sun B, Wang J, Liu B, Ma J (2020) Which is more Important Ecological 627 Conservation or Recreational Service? Evidence from a Choice Experiment in Wetland Nature Reserve 628 Management. Wetlands, 40(6), 2381–2396. https://doi.org/10.1007/s13157-020-01348-8 629 630