How Pollution and Climate Change Affect the Future of Mangrove Forest—A Simulation Study on the Mangrove Area in the Thi Vai Catchment, Vietnam

Abstract

Environmental pollution and climate change have been reported to severely affect the growth and productivity of mangroves. However, it is still unclear how the mangroves will fare if stressed by these adverse conditions, and how the mangroves might fare if these conditions improve. In this study, the trends of mangrove forests in the Thi Vai catchment (Vietnam) were assessed using mathematical models, addressing the polluted environment under climate change conditions. This simulated study was conducted based on the analysis of different types of data. Data on 18 elements’ concentrations accumulated in mangrove tissues in this catchment were analyzed in relation to the states of tree growth rates. Data on the economic productivity and water quality of the Thi Vai River in the five years from 2017 to 2021 were analyzed to detect the main sources of pollution that induced damage to mangrove forests. The results achieved from data analysis are the linear and nonlinear interactions between the concentrations of tissue-accumulated substances and the growth rates of trees. Concentrations of P, Mg, and Sr in mangrove leaves have a linear relationship with plant growth while Cr, Cu, and Ni accumulated in roots have a nonlinear relationship. The mining industry and accommodation and food services are the main contributing sources of Cr and Cu, which affect mangrove health. Information supplied from the data analysis helped in designing the scenarios of different combined environmental conditions for model simulations. Our previously developed mangrove dynamics model was applied to predict the trajectory of the mangrove forest in this area under a total of 16 combined environmental condition scenarios.

1. Introduction

Mangroves are widely known as one of the most productive and biologically diverse ecosystems in the world [1,2,3]. However, mangrove ecosystems are alarmingly threatened by activities of humans such as land-use conversion, deforestation, greenhouse gas emission, waste dumping, overpopulation [4], pollutants discharged [5,6], and climate change [7,8]. In Asia, due to the expansion of industrial zones in coastal areas, heavy metals (HMs) are enriched in the environment and become pollutants that affect mangrove ecosystems. Mangrove plants are believed to play key roles in metal removal [9,10] by absorbing metals in the soil into their roots, transporting these elements, and concentrating them in their tissues [11]. However, the excess of essential and non-essential elements could affect the growth, metabolism, and cell structure of plants [12,13]. HMs bioaccumulation is detrimental to the ecosystems since these metals can only concentrate over time without natural destruction [14]. Plant responses when stressed by HMs were examined by experimental studies for single HM and applied to young mangroves [15,16,17,18]. These authors highlighted that excess HMs inhibited plant growth. Heavy metals can cause harmful effects on mangrove tissues and the root epidermis [19]; for example, the inhibition of photosynthetic ability and osmotic stress were observed in excessive concentrations of trace metals in mangrove seedlings [20].

Another stressor for mangrove survival is climate change. Climate change is an emerging issue, which makes changes to the habitat of mangrove ecosystems through changes in salinity, water level, and the quality and quantity of sediment loading [21]. Sea level rise (SLR) changes the hydrodynamics in estuaries [22] and causes a salinity increment, which not only affects households, industry, and agriculture [23] but also diminishes marsh productivity including mangrove forests, or even causes mortality [24,25]. It was said that the interactive impacts of climate change and pollution cause the degradation of mangroves and the failure of restoration efforts [26].

The evaluation of mangrove forest status in responding to pollution and climate change is a difficult task. For this problem, mathematical models are an effective tool that can tackle this question because they can simulate complex systems with the interactions of many driving factors. Richter et al. (2016) [17] developed a phytoremediation model with Cr uptake for the mangrove species R. apiculata coupled to the plant growth dynamics model and addressed the adverse effects of Cr in the plant. They then developed a compartment model of the phytoremediation system embedded into the hydrodynamics model [27]. This model was later applied to the Thi Vai catchment in Vietnam combined with the Delft3D model for hydrodynamics and substance transport [28]. Mandal et al. (2019) [29] considered matters up-taken as the budget for mangroves of the Sundarbans using the box model. Recently, A. Nguyen, Richter, et al. (2020) [30] simulated the effects of long-term HMs enrichment on the growth of mangroves in the Can Gio Mangrove Forest in Vietnam under different pollution scenarios. For mangrove responses to climate change, Doyle et al. (2003) [31] conducted modeling work to study the migration of mangrove forests along the southwest coast of Florida under climate change and concluded that mangrove habitats will increase over the next century due to climate change. Alizad et al. (2018) [32] applied a coupled hydrodynamics–marsh model (Hydro-MEM) to simulate SLR scenarios in Grand Bay, Mississippi and Weeks Bay, Alabama. They reported an expansion of the bays along with marsh migration to higher land. Nevertheless, up to now, the evaluation of the response of mangroves to pollution in climate change situations by time and space has not been addressed.

The mangroves at the Thi Vai catchment belong to the Can Gio Mangrove Forest in Vietnam (Figure 1). This mangrove forest was nominated as a “Biosphere Reserve” by UNESCO in 2000. The mangroves in the Thi Vai catchment suffer from the wastewater released from industrial zones on the right bank of the Thi Vai River (cf. Figure 1). This river was a “hotspot” since it was heavily contaminated during the period of ~2003 to 2013 with an alert about the long-term pollution of Cr, Cu, and Ni in this area [33,34,35]. Recently, Lorenz et al. (2021) [36] indicated that pollution remains a problem for the health of this river. Nevertheless, in 2018, A. Nguyen, Richter, et al. (2020) [30] conducted another sampling campaign and reported that the pollution situation in this region had improved. Like other coastal regions in the world, this area is also vulnerable to climate change impacts. Thuong and Thach (2019) [37] determined that the sea level has shown an increment trend in the last 31 years in this area, salinity intrusion has happened as was reported by Thu et al., (2020) [38], and other authors also discussed that climate change is putting a strain on the Thi Vai catchment [39,40,41].

The question addressed here is how the mangrove forest would cope when this area faces risks of environmental pollution due to increasing economic activities along with climate change?

In this study, the trends of mangrove forests in the Thi Vai catchment were assessed using mathematical models. Based on the fact that this mangrove area was heavily polluted and then was restored, scenarios of pollution combined with climate change were proposed for the simulations. Different types of data were used for the analysis in this work. Data on the concentrations of 18 elements in plant tissues, which were sampled in 2013, were analyzed to examine their contributions to the health status of mangrove trees. Data on the economic activities and monitoring water quality for a five-year period from 2017 to 2021 were analyzed to determine the contributing sources of the pollutants in the Thi Vai catchment. The simulations were conducted using the mangrove dynamics model of H. A. Nguyen (2011) [42]. This model was coupled with the phytoremediation compartment model from H. A. Nguyen and Richter (2016) [27] to predict the health of mangrove forests in the Thi Vai catchment under different scenarios of HMs pollution and climate change.

2. Materials and Methods

2.1. Study Area

Thi Vai River is located at the boundary between Ho Chi Minh City on the left bank, Dong Nai province on the northeast bank, and Ba Ria Vung Tau province on the southeast bank (Figure 1), from latitude 10°22′ to 10°44′ N and from longitude 106°46′ to 107°01′ E. The mangrove forest mainly occupies the left bank of the river. This forest belongs to the Can Gio Mangrove Forest, one of the most successfully rehabilitated mangrove forests in the world. The mangrove species R. apiculata is the main planted species and covers more than 60% of the forest, among other naturally occurring species. The Thi Vai catchment is subject to asymmetric semidiurnal tidal regimes and a tropical monsoon climate with two distinct seasons, the dry season (from November to April) and the rainy season (from May to October). Industrial zones and international ports have been established on the right bank of the river, operating since the year 2000 resulting from the industrialization process. The severe pollution in this river was caused by uncontrolled wastewater discharges from industrial activities. It was determined that before 2006, over 60% of industrial zones released untreated wastewater directly into the river [33], causing heavy pollution of the waters and sediments in this catchment, especially HMs pollution [34].

2.2. Data Collection

2.2.1. Data on HM Concentrations in Plant Organs

The field sampling data (in 2013) used in this study are from Project EWATEC-COAST (https://www.tu-braunschweig.de/en/ewatec, accessed on 1 August 2023) [43]. These data were partly published in Hoang Anh Nguyen et al., (2014) [44]. The data are composed of HM concentrations in tissues of the mangrove species R. apiculata, including the main root, core, bark, and leaves. The tree organs were collected from one random adult tree in each measured forest stand. These data are shown in Supplementary Table S1.

The samples were analyzed at the Institute of Environmental and Sustainable Chemistry–Technische Universität Braunschweig, Germany. All plant samples were rinsed with pure water to eliminate residues of seawater and soil attached. These samples were air-dried at room temperature. Aliquots of 3 g to 10 g of the air-dried and milled plant samples were reduced to ashes according to DIN 19684-3 (2000) at 550 °C, and the residues were digested with aqua regia according to DIN EN 13346 (2000). The concentrations of elements (Al, Ba, Ca, Co, Cr, Cu, Fe, K, Mg, Mn, Mo, Na, Ni, P, S, Sr, Ti, and Zn) in the aqua regia extracts were determined with an ICP OES (Vista MPX, Varian, Darmstadt, Germany).

2.2.2. Data on Water Quality Monitoring in Thi Vai River

The Surface Water Quality Monitoring data from the provincial sampling sites at 10 locations along the Thi Vai River (sampling 6 times per year) were collected for the analysis of water quality in recent years. It should be noted that this river was severely polluted before 2013. These regular measurement data include physical-chemical parameters such as major ions, nutrients, and metals. Data were collected under the authorities of the Department of Natural Resources and Environment of Ba Ria Vung Province. Supplementary Table S2 shows the mean and standard deviation of these monitoring data and the threshold values regulated by the Vietnam National technical regulations on surface water quality (QCVN 08:2023/BTNMT) [45].

2.2.3. Socio-Economic Data

Data on the economic production of different sectors in five years (2017–2021) were also collected, including agriculture, forestry and fishery production, industrial production, construction production, and service production. Data were collected at the Ba Ria-Vung Tau Investment, Trade, and Tourism Promotion Center. These data were analyzed to quantify the trend of economic growth in relation to environmental quality (using water quality monitoring data). Analysis of these data could help to determine some key variables and to choose the driving forces ruled by socioeconomic pathways for setting up the scenarios for model simulations. Supplementary Table S3 shows data on the production of economic sectors.

2.3. Data Analysis

2.3.1. Statistical Analysis

Descriptive data analysis (mean and standard deviation) for the concentrations in plant tissues and in water was conducted to summarize the ranges of concentrations in plant compartments (roots, cores, barks, and leaves; see Supplementary Table S1) and the surface water of Thi Vai River (Supplementary Table S2). Under the assumption that the data have a linear relationship and are normally distributed, Pearson correlations (p < 0.05) were calculated to determine the relationships among elements in different plant compartments. In addition, principal component analysis (PCA) was employed using MATLAB programming software v.R.2021b to determine the sources of the pollutants in the Thi Vai River.

2.3.2. Regression Analysis

Linear and nonlinear regression was carried out using the data on plant organ concentrations to examine the form of mathematical relationships between the tree growth rate (G, cf. Equation (3)) and the element concentrations accumulated in plant organs. MATLAB programming software version R2021b was employed for this work.

2.4. Modelling

2.4.1. Delft-Flow and D-WAQ Models

The two modules hydrodynamics (Delft3D Flow) and water quality (D-WAQ) from Delft3D were applied to model the flow and salinity of the Thi Vai catchment. Delft3D is open-source software that was developed by the Netherlands WL|Delft Hydraulics. Delft3D is one of the most widely used hydrodynamic models for water flow simulation (e.g., [46,47]) and water quality simulation (e.g., [48]). The hydrodynamic module Delft3D-Flow numerically solves the unsteady-state shallow-water equations in two (depth-averaged) or three dimensions based on the finite element method. The water quality module (D-WAQ) solves the convection diffusion equation for pollutant transport. This study used the Delft3D package to predict the water level and transportation of substances in the Thi Vai River (in this case, the salinity).

In this study, the calibrated and validated water flow and substance transport Delft3D model from the work of A. Nguyen, Le, et al. (2020) [28] was used to run the set-up scenarios. For more information on the data used for calibrating and validating the models Delft3D Flow and WAQ for this area, please see A. Nguyen, Le, et al. (2020) [28]. Based on this calibrated model, the water flow and substance transport in the past and future scenarios were produced. These modeling results were used as input data for the simulations in this work.

2.4.2. Mangrove Dynamics and Phytoremediation Models

The tree growth model based on the classical Botkin model [49] embedding into a tree life cycle model that comprises the processes “Establishment–Growth–Competition–Reproduction–Mortality” from the work of H. A. Nguyen (2011) [42] was applied. This model was extended with a chromium pollution multiplier as presented in A. Nguyen, Richter, et al. (2020) [30]. In this study, the chromium pollution multiplier is modified by applying the three HMs Cu, Cr, and Ni. The form of this multiplier equation is shown in the nonlinear relationship of HMs and the tree growth rate $\mathrm{G}$, as will be described later in this paper. Full Equations and parameter values are expressed in Appendix A and Appendix B.

The growth equations in terms of tree diameter and tree biomass are expressed in Equations (1) and (2).

$$\frac{d\left(dbh\right)}{dt}=G_{opt}\times \frac{dbh·(1-dbh·\frac{H}{D_{max}·H_{max}})}{2b_1+3b_2·dbh-4b_3·{dbh}^2}\times MUL$$

(1)

$$Biom=a_1·{dbh}^{c_1}$$

(2)

where dbh is the tree diameter at breast height (cm); H is the tree height (cm); $H_{max}$ and $D_{max}$ are the species-specific maximum values of tree height and tree diameter; $b_1$, $b_2$, and $b_3$ are the scaling parameters for the relationship between H and dbh; $a_1$ and $c_1$ are species-specific constants for calculating tree biomass $Biom$ from dbh value; and MUL is the growth multiplier. Its value ranges from 0 to 1 and comprises multipliers from factors such as salinity, ground elevation, tree density, and HM pollutants accumulated in plant roots. The ground elevation multiplier is an indirect indicator of water level. $G_{opt}$ is the species-specific growth rate under optimal environmental conditions.

Under the influences of driving factors (MUL), the actual value of the tree growth rate G of a tree is always smaller than the optimum $G_{opt}$ value and is reduced by multipliers MUL, which gives: $G=G_{opt}·MUL$. The parameter $G$ is a species-specific growth rate and can be extracted via the integration of the Botkin model (Equation (1)) using the tree diameter data at the time of planting, ${dbh}_0$, and at the time of measuring (tree age at the time of measurement), ${dbh}_{age}$ [42]. A detailed description of this growth rate constant $\left(G\right)$ is described in the work of A. Nguyen, Richter, et al. (2020) [30].

$$G=\frac{2·H_{max}\left[\int \frac{b_1+a\left(3e^x-2e^{2x}\right)}{\mathrm{a}+b_1-e^x(b_1+a(2e^x-e^{2x}\left)\right)}dx-C\right]}{\mathrm{t}}$$

(3)

where $a=H_{max}-b_1$, $x0=\mathrm{l}\mathrm{n}({dbh}_0/D_{max})$, $x1=\mathrm{l}\mathrm{n}({dbh}_{age}/D_{max})$.

The mathematical model for the uptake of substances by mangroves from the works of H. A. Nguyen and Richter (2016) [27] and A. Nguyen, Le, et al. (2020) [28] was coupled to the growth model from Equation (1). Then, the whole model was embedded into the Delft3D model to simulate the absorption of HMs from soil to plant media. Within each computational cell, the uptake of HMs is modeled by a first-order process. Equations (4)–(6) describe the process of substances (in this case, the HMs) entering plant roots.

$$\frac{\partial P_l}{\partial t}=k S_l$$

(4)

$$k=k_{max}·tsr·Density$$

(5)

$$tsr={\left(\frac{dbh}{D_{max}}\right)}^c$$

(6)

where $P_l$ is the concentration of HMs in plant roots; $S_l$ is the concentration of HMs in soil; $k_{max}$ represents the maximum absorption rate; the rate constant $k$ governing the uptake depends on the root surface, which is related to $dbh$ via the allometric relation $tsr$ in Equation (6) and the $Density$ of trees in a computational cell.

2.4.3. Computational Grid, Topographic and Bathymetric Data

The computational grid is composed of 978 × 493 cells with a rectangular cell size of 30 × 30 m². Topographic and bathymetric data for the simulation were inherited from the work of A. Nguyen, Le, et al. (2020) [28]. These data originated from a variety of sources, comprising 10 × 10 m² bathymetric data of the riverbed, 100 × 100 m² resolution bathymetric data at Ganh Rai Bay, and land topographic data (50 × 50 m²) of the surrounding areas.

2.5. Scenarios Setting

In order to study the trends of mangrove forests in this area with respect to socio-economic activities and under sea level rise and salinity intrusion, scenarios related to these problems need to be set up. Past simulation and future scenario prediction of long-term mangrove dynamics will provide valuable information on how mangroves react to driving factors. In this study, based on the situation and the sources of pollution in the region through the analysis of water quality monitoring data and socio-economic data, scenarios for simulations were set up. The heavy metals Cr, Cu, and Ni were chosen as the major HMs for the simulations because these HMs were found to have effects on mangrove growth, as will be discussed later in the paper.

In total, 16 scenarios were set up for the simulations. The scenarios are summarized in

Table 1. Scenarios set for the simulations.

Pollution Conditions Setting for Model Run	Physical Conditions Setting for Model Run
	Current Condition of Water Level	Current Condition of Water Level + 0.25 m (From the IPCC RCP 8.5 Scenario at Year 2050)	Current Condition of Water Level + 0.53 m (From the IPCC RCP 4.5 Scenario at Year 2100)	Current Condition of Water Level + 0.73 m (From the IPCC RCP 8.5 Scenario at Year 2100)
Baseline (no HMs pollution)	S1	S2	S3	S4
HMs pollution with concentration in 2013	S5	S6	S7	S8
Intense HMs pollution due to economic growth (twofold concentration of those in 2013)	S9	S10	S11	S12
Management control: decrease of pollution after the year 2013, with HMs concentration as those measured in 2018	S13	S14	S15	S16

. Since the forest along the Thi Vai River was planted in 1990 and the region was industrialized from 2000 onwards, the assumption for the simulation was that the environment in this area was clean and stable in the period from 1990 to 2000 with an HM concentration of <20 ppm. After the year 2000, industrial activities started to increase, and the environment in this area was polluted until ~ 2013. Based on this fact, the scenarios were set up such that mangroves developed in healthy conditions from the beginning of 1990 to 2000, and then they suffered from pollution stress in the time period from 2001 to 2013. In scenarios S5 to S8 (Table 1), the polluting situation in 2013 is continued for the next fifty years to 2063. Another group of scenarios assumed apt development in economics, thus releasing large amounts of wastewater (twofold compared to the HM concentration in 2013) in the time period of 2014 to 2063 (scenarios S9 to S12). The timely intervention in management practice to reduce pollution was introduced in scenarios S13 to S16. These scenarios are based on the declining trend of HM concentrations from 2013 to 2018 [30]. Scenarios S1 to S4 are the baseline conditions, which assumed no environmental pollution from 1990 to 2063.

The factors of climate change that have impacts on mangroves include increasing temperature, increasing storm frequency, ocean acidification, and changing evaporation and precipitation, among others [21]. In this study, the two main factors of climate change included in the simulations are sea level rise and saltwater intrusion because these two factors directly affect the tree life cycle and reproduction of mangroves during their lifetime. Scenarios set for sea level rise (SLR) in this study were at 4 levels: current water level condition, current condition of water level + 0.25 m (water level from IPCC RCP 8.5 in the year 2050), current condition of water level + 0.53 m (from IPCC RCP 4.5 in the year 2100), and current condition of water level + 0.73 m (from IPCC RCP 8.5 in the year 2100).

Other assumptions were that (i) there was only one planted mangrove species, R. apiculata, in the forest; (ii) there were no disturbances after the forest was planted (e.g., felling or breaking due to extreme weather); and (iii) all scenarios had the same initial forest conditions in accordance with the recorded reforestation history of this forest. Figure 2 shows the distribution of the mangrove area (Figure 2a) and salinity distribution under sea level rise (Figure 2b–e).

3. Results and Discussion

3.1. Effects of Bioaccumulated Concentrations on Tree Growth Rate

Supplementary Table S1 shows the element concentrations in mangrove tissues. In plant compartments, macronutrients accumulated abundantly in the roots, leaves, and barks with concentrations larger than 1000 mg kg⁻¹, particularly for Na, K, Mg, and Ca (supplementary Table S1). Sulfur (S) and P concentrations are also larger than 1000 mg kg⁻¹ except in the roots (812.75 mg kg⁻¹ for S and 374.67 mg kg⁻¹ for P). For tree cores, these elements are highly present with concentrations in the range between 100 and 1000 mg kg⁻¹. The micronutrients Al, Fe, and Mn are highly accumulated in roots, leaves, and barks with concentrations in the range of 100 to 1000 mg kg⁻¹ (Supplementary Table S1). Other micronutrients such as Zn, Cu, Mo, Ni, and Ti accumulated with low concentrations (<50 mg kg⁻¹). The non-essential elements for plants, Cr, Co, and Ba, accumulated in tree organs with concentrations smaller than 50 mg kg⁻¹ while Sr accumulated highly in roots, leaves, and barks with concentrations in the range of 100 to 1000 mg kg⁻¹.

The total concentrations that accumulate in each organ correlate with each other, reflecting the transport and exchange of matter between plant organs. This is indicated by the moderate to significant correlations between the total concentrations in the leaves and roots with other organs (cores and barks, with r values of >0.6,

Table 2. Pearson correlation (r) of total element concentrations within the compartments (r > 0.55 and p-value < 0.05).

Elements in:	Root	Core	Bark	Leaves
Root
Core	0.573
Bark	0.631	0.907
Leaves	0.707	0.610	0.632

). The data also show the role of the wooden parts of plants (cores and barks) in accumulating substances. This is shown by the significant correlations among the concentrations in the cores and barks in Table 2 (r ≥ 0.9). This high correlation coefficient value indicates the detection of the signal of long-term pollution in the area since substances are highly stored in heartwood (tree core). Tree sapwood is periodically converted to heartwood (the cores), which constitutes the layers of dead sapwood cells. Therefore, with time, plants accumulate elements and immobilize them in their dead cells (tree cores).

It is shown from the data that tree growth is retarded by high concentrations of Mg, P, and Sr in the leaves in a linear way. These concentrations have a negative correlation with the tree growth rate G (r equals −0.86, −0.62, and −0.65, respectively; Figure 3). It is also noted that the tree growth retardation is strengthened along the increasing salinity gradient. Among those three concentrations, Sr is not an essential element for plants, but plant cells have entry mechanisms for Sr such as plasma membrane transporters for Ca and K [50]. The mobility and availability of Sr to plant roots from the soil are controlled by external factors such as the chemical composition of the soil (salinity, pH, and temperature). Sr in mangrove tissues has not been studied yet; however, Storey and Leigh, (2004) [51] found that in the leaves of citrus, the accumulation of Sr in palisade and spongy mesophyll was accompanied by the loss of K and its accumulation in the bundle sheath. P and Mg are plant macronutrients with major impacts on plant productivity. Nevertheless, there are studies that show that excessive concentrations of these nutrients may inhibit plant growth, especially in saline soil. Cerda et al. (1977) [52] in their research on sesame found that the increment in P concentrations increased yields only at low salinity levels; at higher salinity, yields decreased progressively as P increased. For Mg, Trolove and Reid (2012) [53] found that there was a relationship in the Meyer lemon between leaf Mg concentration and leaf stomatal conductance, and there was also a trend for starch to decrease in leaves from plants grown at Mg concentrations above 0.5 mM.

The nonlinear relationship between HMs and tree growth was described in the work of A. Nguyen, Richter, et al., (2020) [30] and is also shown here in this dataset. The data show that the dependence of the tree growth rate G on root concentrations of Cr, Cu, and Ni exhibits a distinctive optimum. This behavior is described by Equation (7) and Figure 4, which shows the data together with a fit to Equation (7). Below a threshold value of the root concentration (20 mg kg⁻¹), the growth rate increases, and above this threshold, it decreases. Equation (7) was embedded into the tree growth model (MUL in Equation (1)) for the scenario’s simulations in this study.

$$G=485.95·\left(1-e^{{-0.623 x}^{0.272}}\right)·e^{-0.004 x}$$

(7)

where x is the total concentration of Cr, Ni, and Cu.

Similar to the linear relationship of Mg, P, and Sr, salinity plays a role in strengthening the inhibition effect of Cr, Cu, and Ni on the tree growth rate, as can be seen in Figure 4. The trees at sites upstream (low salinity) have the highest G values, followed by the sites in the middle and downstream (high salinity), exhibiting the influence of salinity on tree growth. In the quasi-same salinity zone, the effect of contaminant concentrations is shown. For instance, for sites M3 and DK6 located in nearly the same salinity zone, their difference in G values was thus caused by their root concentration (concentration in M3 < DK6; cf. Figure 4). The inhibition of tree growth by this form of nonlinear relationship was discussed before by A. Nguyen, Richter, et al. (2020) [30]. Mechanisms of plant inhibition by the HMs Cu, Cr, and Ni were also discussed by other authors; for example, Oliveira, (2012) [54] discussed that Cr displaces the nutrients from physiological binding sites to enter plant tissues, Ni is extremely toxic to plants when present at excessive levels in the soil or in nutrient solutions to which plants are exposed [55], and Cu accumulated in root tissue at concentrations of 200μg/g and higher will inhibit root growth [16].

3.2. Contribution of Economic Sectors to Water Quality in Thi Vai River

The pollution that occurred in the area was due to economic activities, mainly from industrial companies [35]. The economy in this area was reported to have been increasing during industrialization. Economic data from the recent five years (from 2017 to 2021) show a considerable increase in production (see Supplementary Table S3). The mining industry had brought about the highest production among all sectors, ranging from 82,716.9 to 126,715.4 billion VND, but had an unstable trend from 2017 to 2021. The second highest revenue was the manufacturing industry with an increasing trend ranging from 28,903.68 to 46,429.92 billion VND in five years, with the main activities in the tannery, mechanical engineering, electrical engineering, textile and dyes industries, and oil and cement industries. The number of industrial companies in the Thi Vai catchment increased from 10 companies in 2002 to 308 companies in 2021 (data from the Ba Ria Vung Tau Department of Natural Resources and Environment). The lowest and most unstable production was accommodation and food services, ranging from 528.6 to 1328.62 billion VND.

However, water quality in the Thi Vai catchment has increased in recent years, and monitoring data in 2017–2021 show that most of the monitored parameters were under the threshold of the Vietnam National technical regulation on surface water quality (QCVN 08:2023/BTNMT) [45]. Some locations were slightly contaminated by total suspended solids (TSS), Coliform bacteria, and Escherichia coli (E. coli). The parameter nitrite (NO₂⁻) was higher than QCVN at almost all locations, signaling the pollution from nutrients in this river.

The sources of recorded water quality were determined as three components (PC) using principal component analysis (PCA), which explained 95.69% of the data (cf.

Table 3. Principal components show relationships between economic activities and parameters of water quality. Bold print indicates correlation higher than 0.55. Bold italic print indicates correlation lower than −0.55. Blurred print indicates no correlation.

Parameters	PC1 (50.35%)	PC2 (29.94%)	PC3 (15.4%)
Agriculture and forestry	0.96	0.05	−0.27
Mining industry	−0.34	−0.21	0.84
Manufacture industry	0.92	−0.22	−0.30
Construction	0.94	−0.12	−0.32
Retail sales of consumer goods	0.97	0.16	−0.12
Accommodation and food services	−0.25	0.45	0.83
Turbidity	−0.01	0.93	0.12
DO	−0.02	0.99	−0.02
TSS	0.39	0.90	0.19
BOD5	−0.92	0.25	0.17
COD	−0.91	0.18	0.38
NO3−	0.26	−0.17	0.24
NO2−	0.53	−0.76	−0.35
NH4+	0.13	0.97	−0.05
PO42−	0.10	0.31	0.92
Fe	−0.09	0.94	0.15
Cr	−0.43	0.70	0.55
Cu	−0.60	0.75	0.20
Ni	−0.81	0.08	0.57
Oils and grease	−0.96	−0.02	−0.18
Coliform	−0.88	−0.11	−0.39
E. coli	−0.91	−0.12	−0.31

). The mining industry and the accommodation and food services have positive correlations with the phosphate (PO₄²⁻), Cr, and Ni parameters in component three (PC3). This means these two economic activities are in charge of these contaminated parameters (PO₄², Cr, and Ni). The sectors agriculture and forestry, manufacturing, construction, retail sales, and consumer goods in PC1 have negative correlations with the parameters biochemical oxygen Demand (BOD₅), chemical oxygen demand (COD), Cu, Ni, oil and grease, Coliform bacteria, and E. coli. This relationship means these economic activities involved but did not induce the increment in these contaminated parameters in the period of 2017–2021. Component 2 (PC2) is the component for the group of parameters turbidity, dissolved oxygen (DO), total suspended solids (TSS), cations (NH₄⁺), Fe, Cr, and Cu. These parameters have a slightly positive correlation with the accommodation and food services in PC2, showing that this economic activity has contributed to the environmental quality with respect to these contaminants. In general, heavy metals (Cr and Ni) and PO₄²⁻ (in PC3) are mainly caused by the mining industry (the highest production sector) and accommodation and food services, while other sectors are in sufficient control of their wastewater. Meanwhile, Cr and Ni are the heavy metals that have inhibitory effects on mangrove growth, as shown in Figure 4.

Although economic production increased in the five-year period (2017–2021), water quality was in good control. This evidences the efforts in recovering from the severely polluted situation in this river. This fact motivated the idea of the design for scenarios concerning concentrations in discharge wastewater under economic growth with sufficient and insufficient environmental management practices.

3.3. Mangrove Dynamics under Disturbances (Scenarios Simulation Result)

In the following section, simulation results for the 16 scenarios are presented. Figure 5a shows the average biomass and Figure 5b shows the average concentration of HMs accumulated in root tissues from the HM pollution scenarios (S1, S5, S9, and S13 in Table 1).

The highest biomass in 2063 is in S1 (5.123 tons ha⁻¹), and then S13 (3.251 tons ha⁻¹), S5 (1.825 tons ha⁻¹), and S9 (1.2 tons ha⁻¹). As can be seen in Figure 5a, the biomass of the two uncontrolled pollution scenarios (S5 and S9) decreases rapidly after 2030, especially in S9 with a biomass decrease of 76.56% compared to S1 (baseline scenario) in 2063, indicating the devastating impact of HM pollution on the growth of mangroves. The timely rectification in environmental management in scenario S13 shows the increment in biomass, even though it was inhibited at the beginning like the trend of growth in S5 when HM pollution was still happening. This can be explained by the average HM concentration stored in mangrove roots (Figure 5b). HM storage in roots increased in S5, S9, and S13 during the period in which the environment was polluted (from 2001 to 2013). In S13, due to good pollution control practice, HMs in plant tissues then decreased the rate of accumulation (67.345 ppm in 2063) while these concentrations still significantly increased in S5 and S9 with a steep slope and reach values of 239.715 ppm for S5 and 473.75 ppm for S9 in 2063, respectively (Figure 5b). In a clean environment (S1), tissue-accumulated HM concentrations increase gradually (an average of 56.476 ppm in 2063).

The biomass of mangrove forests not only differs temporally but also dramatically differs spatially. Figure 6a illustrates the spatial distribution of biomass in the year 2063 of scenario S1 (baseline scenario). The most distinctive feature of Figure 6a is the large spatial heterogeneity of biomass concentrates upstream. The low biomass areas are those at very low or high ground elevations or those distributed near the ocean with high salinity. In the HM pollution scenarios S5 and S9, the biomass decreases in the whole region. In the upstream area, biomass is significantly low compared to scenario S1, and it is even worse in scenario S9 (Figure 6b,c). Biomass of S13 in 2063 (Figure 6d) on the other hand is greatly improved compared to S5 and S9 with fewer areas of negative differences when compared to S1 (colored by yellow to blue). The distribution of biomass of these four scenarios in 2063 (Figure 6e) highlights the effectiveness of pollution control in recovering forest health. Mangroves in clean environmental conditions (S1) have the highest biomass, reaching over 120 tons ha⁻¹, while in polluted conditions (S5 and S9), mangroves have biomass below 50 tons ha⁻¹. The timely convention in stopping the release of pollutants causes the biomass of S13 to reach a value of 100 tons ha⁻¹, which is much higher compared to scenarios S5 and S9 (Figure 6e).

In another aspect, climate change is now a phenomenon that attracts deep interest from people across the world. Consequences caused by climate change have been discussed widely, especially its effects on coastal zone ecosystems, which are the most valuable but also the most vulnerable ecosystems in the world. The interaction between climate change and environmental pollution is exacerbating the consequences. Figure 7 shows the differences in biomass of each climate change scenario (by rows) compared to the baseline within the same HM pollution condition (by columns). Figure 7a–c provide the behaviors of mangroves in responding to SLR and salinity intrusion without the stress from pollution (scenarios S1, S2, S3, and S4). Mangrove biomass tends to migrate upstream and to higher ground elevations (Figure 7a,b). This behavior can be clearly recognized in extreme climate change conditions in scenario S4 (Figure 7c) when most of the area is submerged by increasing water levels and high salinity, which causes biomass to dramatically decrease compared to S1. Mangroves tend to migrate to higher ground conditions and move upstream in the North and to the Northwest of the study area under SLR and increasing salinity.

When climate change and pollution occur together, the trends of mangrove migration are different, as can be seen in Figure 7d–l. Mangroves migrate upstream and avoid the heavily polluted areas upstream. In S6, S7, and S8, mangroves migrate to the middle-stream areas and higher land where there is suitable salinity and less pollution instead of moving upstream (Figure 7d–f) but their health status is low indicated by low biomass differences when compared to the biomass of S5. Under severe pollution in scenarios S9, S10, S11, and S12 (Figure 7g–i), the expanse of biomass loss is large and there is no sign of an increase in biomass, implying that mangroves have failed to reallocate. This is because the upper part of the area is heavily polluted while the mangroves in the lower part are wiped out by the rising sea level. This is the worst-case situation, in which mangrove health collapses and they lose their ability to survive. In contrast, with the controlled pollution scenarios (S13, S14, S15, and S16), the re-establishment of mangroves can be seen as shown in Figure 7j–l. Although the areas near the shore also suffer from climate change indicated by some negative biomass differences, mangroves migrate away from upstream and downstream and move to middle-stream areas (Figure 7j). Mangrove biomass continues to increase in S15 and S16 compared to the biomass in S13 (Figure 7k,l).

The temporal average biomass of all scenarios is illustrated in Figure 8 The most distinctive trend is the impact of pollution governing the slopes of the four groups of scenarios S1–4, S5–8, S9–12, and S13–S16. Under clean environmental conditions (S1 to S4), biomass increases rapidly (group of blue curves in Figure 8). The highest average biomass reaches a value of 5.123 tons ha⁻¹ in scenario S1 in 2063. Biomass in S2 to S4 decreases, signaling the effects of climate change, which induces the biomass in S2 > S3 > S4 with values of 4.977 tons ha⁻¹, 4.811 tons ha⁻¹, and 4.696 tons ha⁻¹, respectively. In scenarios of 2013 pollution (S5–S8), average biomass increases in the first 37 years with the maximum value of 2461 tons ha⁻¹ (in S5), while in double-polluted situation (S9–S12), average biomass increases to 1.867 tons ha⁻¹ (in S9) after 31 years; but then biomass decreases in the year 2063 to the values of 1.825 tons ha⁻¹ and 1.2 tons ha⁻¹ for S5 and S9, respectively. When climate change is taken into account in scenarios S6 to S8 and S10 to S12, the average biomass decreases even more. The average biomass at the end of the simulation in scenarios S6, S7, and S8 are 1.782 tons ha⁻¹, 1.758 tons ha⁻¹, and 1.739 tons ha⁻¹, respectively. In the intense pollution scenarios (S10–S12), the impact of climate change is not as strong as that of pollution. Their biomass values are 1.175 tons ha⁻¹, 1.162 tons ha⁻¹, and 1.149 tons ha⁻¹ in 2063, respectively. In the scenarios with pollution control and climate change (scenarios S14–S16), the biomass of mangroves can still increase after having suffered from HM pollution before the year 2018, which is indicated by the overlapping values of biomass in S13–S16 and S5–S8 (Figure 8). After 2018, there is a split in the trend of biomass between the two groups of scenarios (S13–S16 and S5–S8, Figure 8). In the context of climate change, the average biomass of S14 > S15 > S16 is 3.138 tons ha⁻¹, 3.06 tons ha⁻¹, and 3.008 tons ha⁻¹, respectively.

3.4. Discussion

The most important result from this simulation study is that the impact of pollution is more critical to the health and the survival of mangroves (Figure 8). Under polluted conditions (S5–S16), HMs in the polluted soil can be absorbed by mangroves [11,17,30], thus HM concentrations in plant organs increase over time. In the early stages of tree growth, the concentration is still below the critical threshold of 20 ppm; therefore, these HMs support the growth of trees. When HM concentrations reach the toxicity threshold, biomass starts to decrease Figure 5a and Figure 8). The upstream area was extremely polluted and the simulation results in this work show that the most degradation in forest biomass happens in the upstream area (Figure 6 and Figure 7). High concentrations of HMs accumulated in mangrove tissues cause plant growth inhibition and mortality [20,56]. It is clearly expressed in the intense pollution scenarios (S9–S12) of this study that biomass damage expands to the middle stream (Figure 6).

For climate change situations, mangroves tend to reallocate toward upstream and higher lands, which have become new suitable habitats for the dispersal of seedling recruitment [57], to keep pace with the inundation from the rising sea levels and high salinity intrusion. The expansion of the mangrove migration increases with the increase in SLR as can be seen in Figure 7a–c. The same behavior is also observed in the simulation study of Doyle et al. (2003) [31]. The biomass of mangroves is not significantly affected under climate change conditions. This simulation result is consistent with the experimental results from the work of Ellison and Farnsworth (1997) [58] who studied the growth (in plant pots) of the mangrove species Rhizophora mangle to levels of inundation expected for the next 100 years. This simulation results also agreed well with the opinions from previous works [21,25,59].

With the interactive impacts of both pollution and climate change, the mangrove ecosystem in the Thi Vai catchment is under the threat of severe damage (Figure 7i and Figure 8) as the mangroves cannot keep up with rising sea levels and cannot find space to migrate inland due to pollution [60]. Climate change can reduce the suitable habitats for mangroves to develop and, thus, decrease the mangrove areas. Mangroves under climate change may migrate to other new suitable land. Moreover, climate change occurs globally, its occurrence is inevitable, and controlling climate change locally is not an easy task. Pollution, on the other hand, is of local anthropogenic origins.

Therefore, environmental quality can be controlled and managed by strictly practicing the environmental policies and administration guidelines as were implemented in the Thi Vai catchment. Scenario S13 mimics the real-world conditions that occurred in the area when good-practice management was implemented to control the pollution situation. The highest biomass of S13 reaches 52.018 tons ha⁻¹. This result is comparable to our previous work and to the observations of other authors such as T. Van Vinh et al. (2019) [61] and Hoan (2019) [62]. Hence, based on the simulation results in this study, pollution control can be a very effective means to protect mangrove ecosystems, especially upstream of the study area.

Simulation studies, which are based on a variety of data sources and information, have supplied more reliable results. The use of mathematical hydro-ecological models can help in predicting the trends and dynamics of phenomena in different conditions. Nowadays, integrated models addressing the co-development of environmental, ecological, and socioeconomic systems are becoming favorable tools to simulate complex processes under the interference of human activities and determine the states of key natural and socioeconomic elements in the future. These variables are the major parameters used to quantify ecosystem health and are closely related to the co-development of ecological and socioeconomic pathways. For this study, sectors of economic activities characterize the typical types of pollutants that may influence mangrove ecosystems. Efforts to recover the degraded environment in this region have brought about a healthier ecosystem while still leading to an increasing trend of economic activities. However, the optimal economic development that the ecosystems in this river catchment can afford is still unclear, given that each ecosystem has its own carrying capacity. With the increasing influence of human activities in the Anthropocene era, future development of this kind of simulation model should be coupled with socioeconomic models. Using this coupled hydro-ecosystem and socioeconomic model not only enhances the understanding of the phenomenon between the human and natural systems but also helps to improve management and regulation capacities.

4. Conclusions

This work showed a simulation study on the trajectory of mangrove forests in the Thi Vai catchment under different scenarios of environmental conditions. Mangrove growth is severely inhibited in heavy-metal-contaminated conditions and is even worse if combined with climate change impacts. Between a polluted environment and an environment impacted by climate change, the growth of mangrove forests is severely inhibited in a polluted environment. Mangrove tends to move landward in climate-change-impacted conditions.

High concentrations of elements in mangrove tissues led to high ecological degradation. Concentrations of P, Mg, and Sr in the mangrove leaves have a negative linear relationship with plant growth, while Cr, Cu, and Ni in roots have a nonlinear relationship. Hence, pollution control is a prior solution to deal with the degradation of mangroves. The simulation results from this study also indicated that restoring polluted environments can help to protect mangroves. The economic growth following the sustainable development goals requires apt environmental management.

The findings from this study are relevant to ecosystem managers since protecting these vulnerable mangrove ecosystems is important from environmental and ecological safety perspectives. Further development of this modeling tool is also required to advance the simulations and predictions and analyze the impacts of both natural and anthropogenic factors on the sustainability of this mangrove ecosystem.