We chose the estuaries in Quang Ninh and Quang Nam Provinces to build BBN models. The selected areas contain typical inland and wetlands ecosystems to a depth of - 6 m. The estuary region of Quang Ninh Province has four large estuaries: Bach Dang, Ka Long, Tien Yen and Ba Che (Fig. 1). The estuary region in Quang Ninh Province has the Nature Reserve - Dong Rui Wetland with 400 species of fish and 140 species of seaweed (Hà 2022). Economic and cultural activities, industrial parks, coastal urban areas of the Province are concentrated in estuaries and coastal areas. The number of tourists in 2019 reached 4.4 million and fee collection reached more than 1,294 billion VND (50.8 million USD) (Lịch 2020). Total output of exploited and aquaculture seafood in 2021 will reach 149,890 tonnes, reaching 13,009.59 billion VND (512.7 million USD), accounting for 52.5% of the value of the entire industry (Hà 2021). However, the ecological environment faces many problems from the increased sedimentation entering the port. It has narrowed the distribution area of estuarial ecosystems from 1990 to 2008, such as mangrove forests by 7,253 hectares (or 25%), tidal flats by about 10,425 hectares (or 21%) and seagrass beds by 650 hectares (or over 80%) (Lan et al. 2013). The main causes of the decline come from the intensive farming of aquaculture and salt levelling and the increase in storms and floods. Coal-mining activities impact the average sedimentation rate in coastal areas, accounting for about 12.5% of the total sedimentation value (Vinh et al. 2012).

Figure 1.

The location of two chosen large estuaries for ecosystem service assessment in Vietnam. The figure clearly demonstrates the rapid and clear expansion of the urban area from 1985 to present.

Figure 2.

The contents of the study to assess estuarial ecosystem services.

In Quang Nam Province, there are two large estuaries, including Thu Bon Estuary and Truong Giang River Estuary (Fig. 1). With low-lying terrain, coastal cities in Quang Nam, including Hoi An, Vinh Dien, Tam Ky, Nui Thanh and Tam Hoa face many challenges from river and sea bank erosion and flooding. More than 3,000 houses in Tam Ky experienced deep flooding at the end of 2021. However, the estuary area of Quang Nam Province has a long coastline with many beautiful beaches for the service industry. In 2019, the total number of visitors and stays reached more than 7.8 million and income reached 14,570 billion VND. 

The Thu Bon Estuary is the downstream area of the river that flows into the East Sea at Cua Dai and is the key economic region of the country. The mangrove coconut forest ecosystem not only has cultural and historical value, but also provides a favourable environment for the livelihood and development of various aquatic species. The nucleus of Hoi An urban centre, the Old Town, which has an area of 5 km², has been recognised by UNESCO as a World Cultural Heritage Site (4/12/1999). The Cu Lao Cham-Hoi An Biosphere Reserve identified the Thu Bon Estuary as its buffer zone in 2009. The process of urbanisation and coastal tourism development has taken place rapidly, making it challenging to control emerging problems. From 2004 to 2016, approximately 112.5 hectares of some wetland ecosystems in the Biosphere Reserve were lost, including 77.1 hectares of mangrove forests, 34.6 hectares of seagrass beds and 0.8 hectares of coral reefs (Nguyen Van and Tong Phuoc 2021). The conversion of land-use purposes and the construction of infrastructure put pressure on conservation efforts, degrading natural ecosystems. Many animals and plants lost their habitats, disrupting the integrity of estuarine ecosystems and reducing the supply of ES.

In the Thu Bon Estuary area, six main landscape types have been identified, including:

1. River - swamp landscape;
2. Sea - wind landscape;
3. Sand-dune landscape;
4. Shallow water landscape (0-5 m depth);
5. Deep water landscape (5-30 m depth);
6. Alluvial landscape.

Alluvial, shallow water and deep water landscapes include the ecosystem of the upper surface water and underlying marine life ecosystems (aquaculture, seagrass and coral reef). In addition, sandy dunes appeared in the “shallow water landscape” in the north of the Thu Bon River in the years 1992-1997 and in recent years, in the “deep water landscape” where the river discharges into the sea. In the “alluvial landscape”, aquaculture structures for seafood farming established by residents are forming gradually at the old riverbed. The “river - swamp landscape” encompasses the most comprehensive range of ecosystems considered, such as residential areas, agricultural areas, bare land, aquaculture (mainly intensive farming) and mangrove coconut forests. Notably, the gradually expanding residential areas rank second in size after agricultural land (rice paddies and grasslands), followed by the forest ecosystem. The “sand-dune landscape” concentrates in residential areas, coastal tourism and construction works in the north of the Thu Bon Estuary and protective forests and rural and agricultural communities in the south of the Estuary. This landscape has a stretch of sandy beaches directly affected by the waves. The “sea - wind landscape”, accounts for most of the area of agricultural land and human settlement land. Land with sparse vegetation, bare land cover a small area. There are aquaculture ecosystems whose water surface is not large, occupying less than 5% of the total area.

To assess estuarial ecosystem services, based on the BBN model, the study processed four steps: the first was an overview of documents and fieldwork to identify estuary ecosystems and related ES (presented in the Introduction). In the second step, the ecosystems of estuary area can be extracted from LULC maps (in the "Database for model development" and Fig. 3). An ES matrix table was developed, based on expert knowledge, then put into the regression model include GLM (Generalised Linear Model), SEM (Structural Equation Model) (Fan et al. 2016) to calculate the relationships between ES in the third step (Fig. 4). The results of the matrix table are summarised in Table 1. The links and data were transferred to the BBN model before applying to particular regions. The results of the study at step three are presented in "Statistical analysis of interdependence as a basis for BBN parameterisation" and displayed in Appendix 2. In the fourth step, semi-quantitative values of ES were calculated, based on a developed BBN model during the last 30 years and for the scenarios of regional planning and natural hazards up until 2030. The analysis of the data calculated by the model compared to reality is in "Value of ecosystem services in the Thu Bon estuary area" and "Scenario results". The "Development of the BBN Model” is described in detail in steps 2 to 4. Fig. 2 clearly illustrates these steps:

Figure 3.

Land uses/covers in the Thu Bon Estuary from 1991 to 2020.

Figure 4.

The relationship between ecosystem services in the Thu Bon Estuary.

The development of scenarios is an essential function of the BBN model. As per the strategic plan of Hoi An City spanning the years 2020 to 2030, rice-growing land, natural forest land and forest land will be converted to non-agricultural purposes. The Provinces People's Committee in Quang Nam Province expects Hoi An City to become an exceptionally intelligent metropolis with a 40% urbanisation rate by 2030. Consequently, in accordance with the planning, the study formulates two scenarios: (1) Urbanisation and tourism development; and (2) Preservation of natural landscapes (Table 2). To assess the impact of future policies on ES, the two scenarios were constructed by reallocating the area ratio between different categories of LULC. The data were gathered in accordance with the Decree (province 2023a, province 2023b) of the Province of Quang Nam's general planning and the actual data were compared to the LULC area in 2020.

Scenario 1: “Urbanisation and tourism development” considers expanding settlements, tourism and investment in the direction of modernisation - industrialisation and rapid urbanisation. Some 400ha of housing area have been expanded due to the conversion of agricultural land, forest and bare soil. Agricultural land, annual crops and rice cultivation land were changed to aquaculture land. The scenario selected a forest area to be reduced by about 40% compared to 2020, equivalent to more than 250 ha. This area is partly a coastal forest for the construction of tourism areas, partly a forest around residential areas converted to housing land or other technical infrastructure development and the rest is due to the natural degradation of protection forests along mudflats. The water surface of rivers and seas narrows by 50 ha, mainly converted to coastal aquaculture; and this ecosystem is easily eroded every year at a high rate, so the area is narrowed.

Scenario 2: “Preservation of natural landscapes” focuses on the expansion of natural ecosystems, reducing pressure from socio-economic development. This scenario's main focus is to increase the regulating ES. The rating of the change in land-use types is at a stable level and towards the expansion of the protected forest ecosystem. The forest area expanded by 20%, equivalent to more than 100 ha. The focus is mainly on the coastal protection forests, coconut forests and other natural forests to help regulate storms and coastal erosion. Urban land or other infrastructure areas are stabilised or narrowed down to 100 ha compared to the urbanisation growth rate of the region (the urbanisation rate of Hoi An City is 74.5% (2021)). The beach area is expanded by 5 ha (especially the Cua Dai beach area) to help protect the inner coast area and attract tourism (Scenario results analysed in "Scenario results").

Fig. 3 illustrates the trend of changes in the Thu Bon Estuary, based on classification results from remote sensing images. In the early 1990s, the population was partly concentrated along the coast and on both sides of the Thu Bon River, serving marine livelihoods and partly concentrated in the old town area. By the 2020s, the population has concentrated more around the old town, expanding to the coastal strip for tourism development and forming small residential clusters in the other localities. Agricultural land for crops, such as rice and crops, has gradually stabilised with more organised planning compared to scattered distribution in many places. This type is concentrated in the floating areas of Duy Vinh and Cam Kim communes and the Cam Thanh area has also long had a large area of agricultural land (Table 3).

The coastline is experiencing significant erosion and landward shifts and tends to shift towards the south of the Estuary. For the downstream flow of the Thu Bon River: the river channel meanders, dividing into branches with numerous sandbanks, underwater areas and large and small mudflats on both banks, as in Cam Kim, Cam Nam and Duy Vinh communes. About 30 years ago, the river had many small branches and the process of conduction development was relatively “free”. After 30 years, some small flows disappeared and the main flows gradually grew larger. Large-scale erosion and accretion occur on both sides of the river. In the area of Hoi An ancient town, flow path fluctuations are related to the flow redirection on the main flow path, with many floating islands seemingly gradually “drifting” towards the estuary due to the combined impact of the river flow and tidal currents. Currently, people have gradually used these mudflats for settlement, cultivation and aquaculture purposes (Table 3).

The forest ecosystem/mangrove forests have developed steadily in the area of Cam Thanh, in river creeks. This ecosystem tends to change more continuously than other types of ecosystems. The most stable part is the Cam Thanh coconut forest area, which is currently under conservation. More stable is the forest ecosystem/protection forest distributed in Duy Hai and Duy Vinh communes, although the coverage is not high. The least modified type is barren/bare soils, which are mostly sandy soils difficult to cultivate and devoid of vegetation.

The value ES is an important result to assess the change in ecosystem quality over the last 30 years. In the Delphi method, the assessment includes 16 landscape unit/LULC (node 02) divided into four main groups: residential, agricultural, mangrove and wetland. In the research area, the authors list six large landscape types (node 01), each of which includes one or more different landscape unit/LULC. LULC changes easily across landscape types, while landscape types are less susceptible to change. Therefore, the area of a landscape type is equal to the sum of the LULC. The distribution area ratio of LULC in the landscape, when included in the BBN model, is based on GIS. Specifically, when evaluating the alluvial landscape (node 1) over a one-year period, the landscape's proportion is 100% and the area proportion of LULC (node 2) corresponds with the GIS data when incorporated into the model. Finally, the BBN model's parameters automatically calculate the value to provide each ecosystem service.

The values of ES in the research area can be classified into three different stages: the first stage from 1991 to 2001 was less affected by humans, the value fluctuating slightly. The second stage, up until 2015, was clearly influenced by urbanisation. Therefore, the value of ecosystem services has fluctuated significantly. The third stage has the remaining time. Although this period is relatively short to separate into a new stage, it has relatively reflected the immediate stable progress in value, gradually adapting to the stable development of urbanisation.

Overall, the values of the three ES groups have been relatively stable, only fluctuating in certain years. The total value that ecosystems provide to human life fluctuated from 627 points to 670 points. The provisioning ES is a group with a stable value of over 130 points per year and is the group with the smallest value of ES, in which, “fish and other seafood” and “crops” have the highest values (accounting for more than 52%). The regulating ES tends to decrease, from 234 points to 222 points, the rate of decrease compared to 1991 was 4.36% and fluctuated between the years. Specifically, in the three years from 1994 to 1996, the value of services increased (249 points), then the value began to decrease gradually and intermittently, the lowest value was 218 points in 2013 (down 31 points). In particular, ecosystem activities for “groundwater recharge” have the highest value (about 55 points) and “air quality regulation” have the lowest value (less than 30 points). The regulating ES have a similar relationship with each other; all six types of services have the same increase or decrease in value over the year. The cultural ES are the highest, but not stable group in 30 years, increasing in value from 280 points to 292 points, the growth rate being 4.4%. The highest value is “regional identity” (medium of 50-52 points) and the lowest value is “knowledge systems” (from 42 to 45 points) (Figs 6, 7).

Figure 6.

Total value of ecosystem services of each landscape in Thu Bon Estuary, Quang Nam Province.

Figure 7.

Fluctuation of ecosystem services in “Urbanisation and tourism development” and “Preservation of natural landscapes”.

In Fig. 6, The total ES value of six landscape types from 1991 to 2020 tended to increase from 3,375 points to 3,480 points, but the growth rate of only 3% was not high. In particular, the “river - swamp landscape” with the largest acreage is the “alluvial landscape”; the smallest area is the “sand-dune landscape” and the “shallow water landscape (0-5 m depth)”. The “alluvial landscape” has the highest ES value, the average value over the last 30 years reached 772 points. The “river - swamp landscape” has the second highest value. These are two types of ecosystems that provide much higher values than the other four landscape types. However, both of these types tend to decrease in value gradually and not continuously. The second group includes the “sand-dune landscape” and the “sea - wind landscape”. Both types of landscapes have continuously fluctuating values, especially the “sand-dune landscape” which has the strongest fluctuation. The remaining two landscape types have constant values and a low index range of 445 to 450 points. Although the area did not change much, the value of ES shows a relative difference. Due to the type of landscape, there has been a change in land use/cover and ecosystem function. For example, when people change from a mangrove coconut forest ecosystem to an aquaculture ecosystem, the value will be reduced. Mangrove coconut forests not only provide regulating values, but also include the value of providing seafood, especially cultural values. Meanwhile, the aquaculture ecosystem only provides natural or farmed seafood, the value of regulating and cultural ES being very low.

After two scenarios are selected, the data on land-use change continue to be included in the BBN model and the results of the changes in the ES values until 2030 are shown in Fig. 7. In the urbanisation and tourism development scenario, cultural ES values are experiencing a significant increase by 2030, accompanied by a noticeable decrease, at a faster rate, in regulating ES values. Ecosystems change ecological functions, increasing the risk of natural hazards. Rises in sea level during storms, combined with tides, cause the total water level in the coastal plains to reach 150 to 200 cm and increase the frequency of flooding in low-lying areas (urban areas, creeks and coastal areas and aquaculture areas). In the second scenario, which focuses on the preservation of natural landscapes, ES values will decrease slightly compared to 2020. The inherent values of an ecosystem remain stable without urban expansion. Preserving and enhancing heritage and relics is a must. There is a more noticeable increase in the values of the regulating ES group than those of the cultural ES group when nature is conserved. Naturally, this represents a gradual recovery of the environment, contrasting with the rapid destruction caused by urbanisation. Both scenarios have a tendency to reduce the value of provided ES. However, scenario 1 experiences a faster decline due to the dominant role of the tourism and service industries in the economic structure, while scenario 2 sees a milder decline as the affected areas still maintain the ability to provide services. In Figs 6, 7, the wrong number is based on the percentage (%) of each node's confidence level. The BBN model is not calculated, based on accuracy, but on model reliability, based on approaches to the probability of that outcome occurring.

The application of the integrated probability model in the BBN network for the assessment of multiple types of ES has proven successful in this study. Accordingly, the different strengths, weaknesses, opportunities and challenges of BBN in assessing ecosystem services can be explained in Table 7. The model effectively addressed the data scarcity issue in the assessment of Thu Bon Estuary, especially regulating ES, by leveraging the expertise of 21 scientists, LULC maps and statistical data at the provincial, district and city levels, as done by Dang et al. (2021). It can be seen that the model has successfully evaluated multiple types of ES and the study provides detailed and specific results for 17 values according to the applicable directives compared to Duarte et al. (2016), Guerry et al. (2012). The model has high transparency, capable of bridging the gap between qualitative and quantitative methods of assessing ES values in the study by combining the calculated data on monetary values (providing ES and cultural ES) with the potential for objective provision (regulating ES).

This study addresses some of the limitations of previous BBN models. First of all, the study has given three indicators for the assessment of ES in terms of provision, regulation and culture. The expert-based matrix table has a wider scoring range, from 0 to 100. This offers experts more options to assign scores corresponding to ES on the one hand and the benefits from ecosystems can be understood in more detail on the other hand. The use of LULC as a representation for the provision of ES will better illustrate the direct human impact on the ability to create such services. Additionally, the model has provided a complex link between the functions and services of ecosystems. Integrating the correlation matrix and the SEM regression model has made it possible to simplify interactions and validate linkages between nodes in the developed BBN. Instead of ignoring the interactions between nodes or reducing the number of connections, the SEM regression model optimised between input variables, while minimising interaction in connections. SEM has reduced the limitation of modelling complex processes in the absence of feedback loops in BBN, as pointed out by Nyberg et al. 2006, McCann et al. 2006 and Castelletti and Soncini-Sessa 2007. Henceforth, the BBN model illustrates the relationship and structure of the network clearly, showing the interactions between ES to researchers without obscuring them, thus reducing the complexity of the provision processes of ES compared to Barton et al. (2008). For example, node 18_“Cultural heritage” has a strong correlation with nodes 19 and 16. The remaining relationships are weaker. All of the services have such correlations, demonstrating different connections (as shown in Table 4). Three appendices of Table 4 make the results easily analysable, assessable and reviewable for accuracy/applicability to reality. In addition, the model also uses conditional probabilities to express the relationship between nodes.

Moreover, the BBN model requires minimal updates or may only necessitate additional data to align with the current situation during the study period. Therefore, the model consistently reflects the current state of knowledge. The data on the model are already representative and easily transferable from one location to another when quick information references are needed for decision-making processes related to the management and utilisation of ecosystems, as demonstrated by Xu et al. (2022). The model is also easily accessible by many different user groups. In addition, the developed model provides direct results on ES values on the gain and loss when changing land-use scenarios to planners and managers for direct use, without the need for additional applications or through any other intermediary. Moreover, the combined model includes analogue data and digital data to serve research purposes.

The result of the change in ecosystem service values was calculated using the BBN model in the Thu Bon Estuary, consistent with the sectoral development structure of the region and the whole province. It explains two trade-offs between estuarial ecosystem services through the change in LULC types.

The first is the process of tourism and urbanisation development. According to Table 1, the value of the cultural ES group is rated highest amongst the 3 ES groups (average 50–70). Most LULC types have experts showing high appreciation for cultural ES (KDL, RPT, SON, A1 etc.). The value of cultural ES increases as tourism develops. This is evidenced in Hoi An, right after the old town became a World Cultural Heritage Site in 1999. The significant development of large resorts along the coast has greatly enhanced the cultural environment. The number of visitors to Hoi An has increased rapidly; in 1999, there were only nearly 100,000 visitors and 17 hotel and motel facilities. By 2019, Hoi An had more than 7.6 million visitors and 624 accommodation facilities with more than 10,000 rooms, tourist villas and homestays. However, the Hoi An urban heritage is still preserved quite intact in terms of landscapes and architecture, as mentioned by Nguyen et al. (2024). 

The second is natural hazards. The results of Table 1 of the Delphi method indicate a high assessment of the regulating ES for forests and wetlands (70-90). This demonstrates their significance in mitigating natural hazards. However, in Thu Bon Estuary, the area of forests and wetlands is small and easily subject to changes, leading to rapid degradation that affects life, such as extensive and prolonged floods (which can reach a depth of one metre (Hieu 2007). Storm Nari (2013) destroyed more than 80,000 hectares of fruit and forestry trees. The expansion of urban and tourism types (ODT, KDL, CDG, A2 etc.) has a low assessment value (5–20) in the regulating ES group. In the Thu Bon Estuary, erosion has continuously penetrated inland at a rate of 2-3 m to 20-30 m/year; even in the rainy season, floods can be as high as 40-50m/year, but there is a process of rapid backfilling and accumulation of a large amount of sludge in the dry season according to Thien and No (2012). In 2020, the Cua Dai area is strongly eroded to significantly strong from 3-10 m/year, even over 10 m/year according to Liem et al. (2020). The rest is due to the urbanisation process in Hoi An City and the expansion of residential areas in Duy Xuyen District. 

The two scenarios give a more objective view of the trade-offs of ES values for different types of land use/cover. In scenario 1, the trade-offs and losses are more apparent when moving towards tourism development and urbanisation. When prioritising urbanisation and tourism development, the ability to ensure the supply and demand for cultural ES is stable because these are products that humans can easily create and utilise and “long-established” resources offer many advantages. However, when the emissions of an urban area increase, environmental pollution, population carrying capacity in a territory and other problems increase; the demand is insufficient for environmental treatment and protection due to the concretisation of forest systems and green cover. The clear evidence is the substantial reduction in the regulating ecosystem services. As a result, excess supply leads to socio-economic conditions that exceed the limits of the sustainable development circle. Therefore, the key issue is protecting the natural environment.

Indicator table for assessing estuarine ecosystem services in Vietnam (Table 8).

The SEM model to set up a BBN for assessing esturian ecosystem service (Fig. 8).

Figure 8.

The SEM model to set up a BBN for assessing esturian ecosystem service.

A linear chart showing the correlation between ecosystem services (Fig. 9).

Figure 9.

A linear chart showing the correlation between ecosystem services.

Field pictures of four major ecosystem groups (Fig. 10).

Figure 10.

Field pictures of four major ecosystem groups.