Introduction

Mangroves are carbon-rich ecosystems1, serving as hotspots of biodiversity, as well as a nature-based solution for mitigating carbon emissions1,2,3 and protecting the coastal community from natural disasters4. Despite their significance, global mangrove forests have decreased alarmingly over the past few decades4,5,6,7,8,9,10,11, with approximately 62% of mangrove loss worldwide resulting from human-induced land-use changes from 2000 to 201612, further generating substantial carbon emissions7,8,10,13. Among various pressures from natural and anthropogenic forces that result in mangrove change, urbanization with continuously intensifying trends profoundly affects natural habitats14,15. Urbanization-driven economic activities have led to land conversion from mangroves to aquaculture, agriculture, and urban infrastructure4,12,16,17,18,19,20,21, with accompanying pollution problems17,22 in mangrove habitats. Gradually, with growing recognition of the value of blue carbon and the profound implications of its loss, numerous actions aimed at mangrove restorations have been initiated worldwide23,24,25,26. Nevertheless, a comprehensive understanding of the direct and indirect roles played by urbanization in the global mangrove cover changes and carbon storage capacity needs to be clarified. Therefore, it is urgent to investigate the impacts of urbanization on the mangrove ecosystem to formulate sound conservation policies and promote sustainable development of tropical and subtropical coasts.

Simultaneously, climate change-induced alterations in temperature, precipitation regime, extreme weather events, and sea level have been identified as major stressors on mangroves by altering the sediment supply, salinity, carbon stock, and productivity of these ecosystems2,27,28,29,30,31,32,33. Projections showed that future climate change would increase mangrove carbon stock by 7% globally34, through increased temperature, photosynthesis rate, and nutrient load. It is worth noting that urbanization reinforces warming35 and exacerbates precipitation and flooding36, leading to local, regional, and even global climate alterations37,38,39. Furthermore, the impacts of urbanization on surrounding flora growth and phenology are climate-dependent40,41. Hence, the potential hidden impacts of urbanization on the mangrove ecosystem can be mediated by climate-related factors.

Despite extensive previous studies examining the individual impacts of urbanization or climate on mangrove ecosystems separately, there remains a lack of integrated assessments about the synergistic and interactive impacts of urbanization and climate change on mangrove ecosystems at the global scale. The potential indirect impacts of urbanization mediated by climate change can be easily hidden by the widely proven linkage between climate change and urbanization and have been mistakenly attributed as part of individual impacts of climate on mangroves. However, these indirect impacts may get exacerbated under the context of global climate change and rapid urbanization and can even bring unexpected shocks to global mangroves. The joint effects of urbanization and climate are complex and need to be sufficiently explored. Overlooking these effects may lead to low efficiency of mangrove conservation and a biased understanding of the roles played by urbanization in mangrove conservation in the era of climate change. Addressing this knowledge gap would directly benefit the implementation of blue carbon and mangrove restoration projects by integrating and anticipating the impacts of urbanization and climate change on them. Therefore, an integrated assessment that unveils a complete picture of the combined effects of urbanization and climate change on mangroves is urgently needed, since both urbanization and climate tend to continuously change as predicted regionally and globally42,43. These insights can provide valuable information for uncovering neglected interplay among urbanization, climate, and mangroves, and balancing urban development and coastal conservation in light of local conditions.

To address these gaps, we developed a systematic framework to assess urbanization impacts on mangroves at a global scale. The framework incorporates two main components: direct urbanization effects, and indirect urbanization effects mediated by climate pathways. The joint effect of urbanization and climate on mangroves was further investigated. Specifically, to quantify the hidden impacts, the Urbanization-Climate Change-Mangrove (UCM) model and fine-scale spatial datasets collected at 19717 plots (each 0.1° × 0.1°) were incorporated to test the theoretical feasibility at a global scale, which achieved a good model fit (CFI = 0.928, RMSEA = 0.058, SRMR = 0.041) (Fig. 1, Supplementary Fig. 1, Methods). The proposed framework comprises three interconnected systems: urbanization, climate, and mangrove. Several representative variables were employed to quantify each system. For the urbanization system, variables were selected to represent urbanization from the land, population, and economic perspectives. The climate system incorporates several climate factors that mangroves are sensitive to. In addition, mangrove cover and carbon stock are employed as important ecological indicators for the mangrove system. Notably, due to the widely proven linkage between urbanization and climate, urbanization can indirectly affect mangroves via the climate-mediated pathway. Thus, the total urbanization impacts on mangroves were measured by 1) direct urbanization effects (EUM) and 2) indirect urbanization effects via the climate-mediated pathway (EUC × ECM) (Fig. 1, Methods). The joint effect of urbanization and climate on mangroves was investigated by comparing the direction and magnitude of total urbanization effects (EUM + EUC × ECM) and climate effects (ECM) on mangroves (Fig. 1, Methods). Finally, spatial heterogeneity and regional disparity were investigated through the same model, at a fine scale across 924 grids (each 1° × 1°) and at the country scale across most countries with mangrove cover, to ensure the generalizability of the proposed framework.

Fig. 1: Systematic framework of urbanization impacts on mangroves based on the UCM model.
figure 1

Several representative variables for urban (impervious surface, population, GDP), climate (sea level, winter air temperature, precipitation), and mangrove (mangrove area, mangrove carbon stock) were employed for quantifying three latent variables: urbanization, climate, and mangrove. The framework enables the direct, indirect, and total effects of urbanization on mangroves to be measured by means of different pathways and equations that are listed in the figures. In this model, the direct impacts of urbanization and climate change on mangroves are EUM and ECM; the indirect impacts of urbanization on mangroves are (EUC × ECM). The total urbanization impacts on mangroves were measured by (1) direct urbanization effects (EUM) and (2) indirect urbanization effects via the climate-mediated pathway (EUC × ECM). These effects are standardized within the range of (−1,1), whereby the positive and negative effects are represented by plus and minus signs, respectively. Icons made by Freepik, Icon Pond, Sittipat Tojarean, cah nggunung, Nualnoi Kinkaeo from www.flaticon.com.

Full size image

Results

Positive effects of urbanization on mangroves indirectly mediated by climate

Our findings suggest that urbanization has both positive and negative indirect impacts on mangrove carbon stock and coverage, as mediated by climate effects. Our results indicate that approximately 28% of grids exhibited positive indirect impacts (indirect effect score>0 and <1) of urbanization on mangroves, indicating that urbanization can affect the local climate (e.g., winter air temperature, rainfall regime, and sea level) and which then indirectly increase carbon stock and promote mangrove expansion, producing beneficial influences on mangroves (Fig. 2a). Many mangrove species, being temperature-sensitive and preferring warm habitats, may benefit from positive urbanization’s climate-mediated indirect impacts. This could alleviate climate-related growth limits of mangroves, expand their natural habitats for colonization, and further increase carbon stock. A similar proportion (27%) of grids showed negative indirect impacts (indirect effect score<0 and >−1) of urbanization on mangroves. The regions with negative indirect impacts were mainly observed in Southeast Asia and Central America, with the highest negative effects found in Costa Rica, and North America, whereas the highest positive indirect impacts were observed in Ghana, and Western Africa. At the country scale (Fig. 2b), more countries experienced positive indirect effects of urbanization on mangroves. Countries with stronger positive indirect effects were mainly distributed in North America and North Africa.

Fig. 2: Global patterns of the indirect effect of urbanization on the mangroves.
figure 2

a Global pattern of the indirect effect of urbanization on mangroves at the 1-degree grid level. The effect score of the indirect effect ranged from −1 to 1. The NA value represents areas that are unaffected by urbanization, where the relationship between variables in our model is not statistically significant. b Global pattern of the indirect effect of urbanization on mangroves at the country level (displayed within coastal regions of each country). c Comparison of indirect effects by different continents. d Comparison of indirect effects by different climate zones.

Full size image

For direct impacts of urbanization on mangroves, nearly 45% of the grids exhibited positive impacts (direct effect score>0 and <1), while approximately 10% of the grids exhibited negative direct impacts (direct effect score<0 and >−1) (Fig. 3a). The regions with negative direct impacts of urbanization were mainly observed in Africa, Southeast Asia, and Central America, especially in Indonesia and the Philippines, with the highest impact observed in Mozambique (−0.99), southern Africa. Negative urbanization impacts lead to mangrove loss and carbon stock decline due to human disturbances stemming from land use, population dynamics, and economic aspects. In addition, positive direct effects of urbanization were mainly concentrated in western Southeast Asia, southern China, and the East coast of South America, where the largest positive direct impact is found in Mexico (0.79) and North America. Along with the urbanization process, increasing conservation awareness and availability of colonization spaces, could positively facilitate increases in mangrove expansion and carbon stocks. Additionally, at the country scale (Fig. 3b), more countries experience negative direct effects of urbanization on mangroves. Countries with stronger positive direct effects were mainly distributed in Africa and Central America. Negative direct impacts of urbanization were observed in 55% of the countries.

Fig. 3: Global patterns of the direct effect of urbanization on the mangroves.
figure 3

a Global pattern of the direct effect of urbanization on the mangroves at the 1-degree grid level. The effect score of the direct effect ranged from −1 to 1. The NA value represents areas that are unaffected by urbanization, where the relationship between variables in our model is not statistically significant. b Global pattern of the direct effect of urbanization on mangroves at the country level (displayed within coastal regions of each country). c Comparison of direct effects by different continents. d Comparison of direct effects by different levels of economic development.

Full size image

Regional variations in positive and negative urbanization impact on mangroves

Distinct regional variations in urbanization impacts were observed by different continents, climate zones, and economic development levels. In tropical and subtropical/temperate zones, the direct effects on mangroves were negative, while the indirect effects were positive, though negative direct impacts were larger in the subtropical/temperate zone. Notably, positive indirect impacts are more distinct in the subtropical/temperate zone, particularly in Southeast China and the east coast of America (Fig. 2a, d, Supplementary Fig. 2), suggesting the alleviation of mangrove’s growth limiting factor and potential for habitat expansion in less-adapted climate conditions. Moreover, in arid regions, the direct and indirect effects of urbanization on mangroves were negative but relatively low (Fig. 2d, Supplementary Fig. 3b). Furthermore, in developing and the least developed countries, the direct and indirect impacts of urbanization were relatively low, whereas urbanization exhibited larger negative direct impacts on mangroves in developed countries (Fig. 3d, Supplementary Fig. 3c). It is observed that in South America, both direct, indirect total effect of urbanization were relatively positive (Figs. 2c,  3c, Supplementary Fig. 3a). Asia experienced larger positive direct effects and smaller negative indirect effects. While other continents represented distinct negative direct, indirect and total urbanization effects.

Fifteen percent of the grids indicated that positive indirect urbanization effects could offset the negative direct effects of urbanization by 1% to 100% (with an average of 91%). Among these grids, the positive indirect effects of 79% of the grids could completely offset the negative direct effects of urbanization, representing positive total urbanization effects. In contrast, a negative indirect effect could offset an average of 28% of the benefits of urbanization’s direct effect. It is found that urbanization has positive total effects in Asia and South America (Supplementary Fig. 3). The region with the highest positive total effect was located in the southern part of Indonesia, with direct and indirect effects of 0.54 and 0.16, respectively (Supplementary Fig. 4). The total effects of urbanization are negative in different climate zones and economic development levels, with larger effects observed in subtropical/temperate zones and developing countries.

The joint effect of urbanization and climate on mangroves

23% of the total regions showed a joint detrimental effect of urbanization and climate’s negative impacts, with several clusters of grids mainly located in North America, South Africa, and Southeast Asia. This suggests that urbanization and climate change jointly pose potential threats to mangroves (Fig. 4a–c). Countries with positive impacts from both urbanization and climate change were mainly distributed in Central America and the Middle East, indicating that improvements in mangroves were jointly brought by urbanization and climate. The majority of grids (51%) and countries exhibited negative/positive relationships, suggesting that urbanization and climate have opposite effects on mangroves (Supplementary Fig. 5).

Fig. 4: Global patterns of the joint effects of urbanization and climate on mangroves.
figure 4

a Global pattern of the joint effect of urbanization and climate on mangroves at the 1-degree grid level. b Global pattern of joint effect at the country level (displayed within coastal regions of each country). c Illustration of category and percentage of joint effects. The x-axis represents the total urbanization effects (EUM + EUC × ECM), and the y-axis represents the total climate effects (ECM). d Relationship between joint effect and urbanization degree. The degree of urbanization corresponds to the stages of urbanization (early stage: urbanization degree≤3; medium stage: 3<urbanization degree<7; late stage: urbanization degree≥7), representing the full spectrum of the urbanization process. The red curve represents a cubic nonlinear fit of the mean value of the joint effect index in each urbanization degree bin, with a p-value < 0.01 and an R-squared of 0.85. The gray shading represents the 95% confidence interval.

Full size image

The joint effect of urbanization and climate on mangroves showed a nonlinear response to the degree of urbanization, with an overall decrease as the degree of urbanization increased (Fig. 4d). In regions with low degrees of urbanization regions, the impacts of urbanization and climate on mangroves are in the same direction (both positive/negative). However, the joint effect is dominated by climate effects during the early urbanization stages, where the climate effects are 10–60 times greater than urbanization effects. During the medium urbanization stages, the impacts of urbanization and climate are in opposite directions but with slight differences in magnitude. As the degree of urbanization further increases to the late urbanization stages, the joint effect index gradually increases and tends to be stable. The joint effect is still dominated by climate effects during the late urbanization stages. However, the differences between urbanization and climate effects are reduced, indicating that urbanization could partially compensate for climate effects in highly urbanized regions.

Discussion

This study presented the fine-scale assessment of the hidden impacts of urbanization on mangrove cover and carbon stock affected or mediated by climate, as well as the joint effects of urbanization and climate at the global scale. Our findings suggest that mangroves are not only affected by the individual impacts of urbanization and climate change but are also highly influenced by their joint effects, which have been ignored and underestimated in previous research. Both global and local climates can bring strong or slight disturbances to mangroves through winter air temperature, sea surface temperature, rainfall regime, and sea level changes, which are partially affected by the local urbanization process. It is necessary to comprehensively understand the impacts of influencing factors that have complex interplay, such as urbanization and climate, which are continuously changing and interrelated. This insight provides a reference for other coastal ecosystems, as well as various terrestrial and marine ecosystems that suffer from multiple external disturbances, further promoting global sustainable development.

Surprisingly, our study found that both urbanization and climate change had positive impacts on the mangrove ecosystems in some regions (Central America and the Middle East, etc.), which is different from the general belief of adverse impacts from previous research16,17,18,19,20,21. This suggests the feasibility of facilitating mangrove conservation through effective urban planning to achieve coastal sustainability. The positive effect of urbanization in our study is in agreement with previous studies that reveal urbanization’s positive impacts on surrounding vegetation40,44,45. Historically, mangroves were converted to other land uses12 for economic development, such as urban infrastructures, fish/shrimp ponds, and palm trees, leading to substantial mangrove habitat loss and fragmentation. However, the adverse impacts of urbanization and climate change on mangroves are not a universal and sole pattern. In our study, we have also observed positive urbanization impacts in certain regions. As shown in the satellite images in Supplementary Fig. 6, several places experienced distinct mangrove expansions where both the direct, indirect and total urbanization impacts were identified as positive. The potential reasons for positive urbanization impacts include three aspects. Firstly, urbanization has the potential to alter local climatic conditions46,47,48. Notably, rising temperatures associated with urbanization may promote the growth and colonization of mangrove ecosystems27, particularly in subtropical and temperate regions. Mangroves are well-adapted to thrive in warm environments, and the warming trends driven by urbanization could create conditions more favorable for mangrove colonization, enhancing mangrove growth and enabling more carbon stocks49. Secondly, urbanization and associated human activities can lead to increased sediment load in coastal areas, which can result in the formation of mudflats50,51. These newly formed mudflats can provide a suitable substrate for mangrove colonization and expansion52. Thirdly, urbanization can also facilitate mangrove expansion through increased conservation awareness and restoration efforts. Mangrove restoration efforts, both reforestation and afforestation, provide substantial blue carbon benefits24. For example, studies have highlighted how Singapore incorporates mangrove restoration as an important component of its urban space planning53. A study5 demonstrated a positive correlation between mangrove gain and factors associated with urbanization, such as increased night-time lighting (indicative of economic development) and community forestry initiatives (indicative of conservation effort). Urbanization also affects people’s pro-environmental behavior in different dimensions (economic, spatial, social, and population urbanizations)54. Our findings suggest that urbanization if carefully planned and managed, may offer opportunities for enhancing the sustainability and resilience of mangrove ecosystems in these regions. The recognition of the positive aspects of human-induced stressors in certain regions is essential for achieving harmonious coexistence between humans and nature without limiting human development.

Low-degree urbanization areas exhibit synergistic relationships between the joint effect of urbanization and climate. The joint effect degree decreases with increasing urbanization degree and then increases and tends to be stable in high-degree urbanization regions. Many countries have attached great importance to and invested heavily in mangrove conservation and restoration projects for a long time. Our findings provide valuable implications for making effective and targeted policies regarding mangrove conservation and restoration. Previous restoration strategies have primarily prioritized suitability assessment based on the biological and ecological attributes of mangroves, while it is also imperative to consider the impacts of surrounding human pressures to mitigate substantial anthropogenic disturbances. Specifically, regions where mangroves received positive impacts from both urbanization and climate indicate mangroves’ good adaptability to the current coordinated situation. For these regions, efforts should be made to maintain the current conservation policy and urbanization progress and further advance the restoration of deforested mangroves and the plantation of new mangroves in appropriate locations. In areas where mangroves suffer from detrimental situations, the priority is to improve the habitat of the remaining mangroves by strengthening protection and controlling the disturbance of artificial coastlines to local climate and mangrove habitats before investing substantially in restoration projects for either blue carbon24 or biodiversity conservation.

The interactions and mechanisms underlying these effects are complex and require further investigation. This study provides insights into the observed positive (direct or indirect) influence of urbanization in certain areas. Promoting sustainable urban development practices or integrating blue carbon considerations into urban planning processes could help recognize and capitalize on the positive aspects of urbanization for blue carbon. It also highlights the importance of identifying areas where conservation efforts should be concentrated to mitigate the negative impacts. We hope these insights could contribute to the achievement of the United Nations Decade on Ocean Science and Decade on Ecosystem Restoration. There are still some limitations in this study. We can further improve the consistency and compatibility of the multi-source input datasets used in our analysis, as well as improve the model, by incorporating a more comprehensive set of mangrove change drivers and attributes of the mangrove ecosystem. In addition, the mechanisms underlying the observed positive impacts of urbanization on mangrove ecosystems require deeper investigation in future work. Future research could also explore the tipping points of positive and negative effects under different urbanization and climate scenarios.

Methods

Defining urbanization

Previous studies have defined urbanization in different ways, such as the process of population shift from rural to urban areas, urban area expansion, population growth, and economic development55,56. We have characterized urbanization through three key dimensions: land, population, and economics. This allows us to capture the various human activities and accompanying environmental phenomena that can influence mangrove ecosystems, including not only urban expansion/population growth/economic development but also associated changes in the environment and ecology. Specifically, in our definition, urbanization can represent processes such as deforestation caused by land use conversion and relevant human activities, as these changes can be driven by the need to accommodate growing urban populations and economic activities. For example, as indicated by a study57, the expansion of aquaculture has been driven by growing urban demand. A report from FAO58 also stated that the increase in per capita consumption and population growth has led to a rising demand for fish produced through aquaculture. Urbanization can involve hydrological changes and pollution caused by urbanization. By adopting this broader perspective, we aim to explore the complex and multifaceted impacts of urbanization on mangrove ecosystems, recognizing that it encompasses a range of interrelated processes and phenomena.

Direct and indirect effects analysis

As shown in Fig.1, a conceptual framework of the structural equation model was built to estimate the causal relationships between two essential stressors and the mangrove ecosystems. The framework incorporates two main components: direct urbanization effects (EUM), and indirect urbanization effects (EUC × ECM) mediated by climate pathways. Urbanization (EUM) and climate change (ECM) directly impact on mangroves individually. EUC represents the direct impact of urbanization on climate change, indicating its influence on local climate. Due to the widely proven linkage between urbanization and climate, urbanization can indirectly affect mangroves via the climate-mediated pathway. Urbanization (U), climate change (C) and mangrove ecosystems (M) are three complex and abstract systems, which can be represented by multiple specific indicators. Several representative variables for urbanization (impervious surface, population, GDP), climate change (sea level, winter air temperature, precipitation), and mangrove ecosystems (mangrove area, mangrove carbon stock) were employed for quantifying three systems. In this model, urbanization and climate change may affect the mangrove ecosystems via (1) individual direct effects of urbanization (EUM) and climate (ECM), (2) indirect effects of urbanization (EUC × ECM) on the mangrove ecosystems mediated by urbanization-climate linkage. EUM, EUC, and ECM are the effect index (standardized path coefficients) to be investigated. Thus, the total urbanization impacts (EUM + EUC × ECM) on mangroves were measured by the sum of 1) direct urbanization effects (EUM) and 2) indirect urbanization effects via the climate-mediated pathway (EUC × ECM).

SEM is an analytical approach for evaluating complex causal networks of cause-effect relationships59, integrated by factor analysis and path analysis. SEM has been widely applied in psychology, social science, and ecology research to provide a valuable tool for analyzing abstract concepts that cannot be directly measured and quantified. One important advantage of SEM is the ability to infer or estimate the relations between latent variables, which are unobservable or unmeasured variables inferred from observed indicator variables. In our study, mangroves and two stressors (i.e., urbanization and climate) functioned as abstract and complex concepts that could not be directly measured and quantified. We identified potential indicators influencing the mangroves. These indicators should be observable and measurable and are called observed variables in SEMs. We selected variables based on literature reviews and expert experiences to build an SEM framework that could indicate the drivers of mangrove change. In our SEM, mangrove (\({{{\boldsymbol{\eta }}}}\)) is an endogenous latent variable, while urbanization (\({{{{\boldsymbol{\xi }}}}}_{1}\)) and climate (\({{{{\boldsymbol{\xi }}}}}_{2}\)) are exogenous latent variables. Y and X are endogenous and exogenous observed variables, respectively. In our model, Y represent mangrove’s observed variables (mangrove area, mangrove carbon stock), X represents urbanization’s observed variables (impervious surface, population, GDP), and climate’s observed variables (sea level, winter air temperature, precipitation). \({{{\boldsymbol{\varepsilon }}}}\), \({{{\boldsymbol{\delta }}}}\) and \({{{\boldsymbol{\zeta }}}}\) represent the measurement errors of Y, X, and \({{{\boldsymbol{\eta }}}}\). Path coefficients and factor loadings are the primary parameters measured by SEM, which elucidate the causal relationships and interdependencies among the study variables. SEM includes the measurement model and structural model. The measurement model estimates the relationship between latent and observed variables, and the structural model evaluates the causal relationship between latent variables60. Confirmatory factor analysis is applied to ensure that the measurement model fits the data. The measurement model is given by

$${{{\rm{Y}}}}={\Lambda }_{{{{\rm{Y}}}}}{{{\rm{\eta }}}}+{{{\rm{\varepsilon }}}}$$
(1)
$${{{\rm{X}}}}={\Lambda }_{{{{\rm{X}}}}}{{{\rm{\xi }}}}+{{{\rm{\delta }}}}$$
(2)

where Y and X are endogenous and exogenous observed variables, \(\eta\) and \({{{\rm{\xi }}}}\) are endogenous and exogenous latent variables, \({{{\boldsymbol{\varepsilon }}}}\) and \({{{\boldsymbol{\delta }}}}\) are the measurement error of Y and X. ΛY and ΛX are factor loading matrices indicating the relationship between latent and observed variables. In our model, they can reveal how much the observed variables explain the latent variables (mangrove/urbanization/climate system). The structural model for assessing the relationship between \({{{\boldsymbol{\eta }}}}\) and \({{{\boldsymbol{\xi }}}}\) is given by:

$${{{\boldsymbol{\eta }}}}={{{{{\bf{\Gamma }}}}}_{11}{{{\boldsymbol{\xi }}}}}_{1}+{{{{{\bf{\Gamma }}}}}_{12}{{{\boldsymbol{\xi }}}}}_{2}{{{\boldsymbol{+}}}}{{{\boldsymbol{\zeta }}}}$$
(3)

where Γ are unknown parameter matrices of regression coefficients (EUM, EUC, and ECM), representing the impact of the exogenous latent variable on endogenous latent variables, and \({{{\boldsymbol{\zeta }}}}\) is a vector of error measurements of \({{{\boldsymbol{\eta }}}}\).

SEM was conducted using the AMOS software and R package lavaan61 with a generalized least squares estimator. To evaluate the UCM model performance, we obtained the overall result on a global scale, and all the observable variables were measured within a 0.1° grid where mangroves exist. The overall result of the global model was generally good, with a good model fit (Comparative Fit Index (CFI) = 0.928, Root Mean Square Error of Approximation (RMSEA) = 0.058, Standardized Root Mean Square Residual (SRMR) = 0.041). All the path coefficients and factor loadings were significant. U and M showed a positive relationship with positive standardized path coefficients EUM (0.004), which was the direct effect of U on M. The indirect effect of urbanization can reveal the underestimated effect hidden by the climate-mediated pathway, represented by the multiplication of standardized path coefficients of U to C (EUC) and C to M (ECM). The total effect of urbanization on the mangrove ecosystems is the sum of direct and climate-mediated indirect effects. Supplementary Fig. 7 and Supplementary Fig. 8 display the comparison of the three path coefficients employed in the effect estimation in different continents/climate zones/economic development levels. Supplementary Fig. 10 visualizes the correlation among all variables, using color to indicate the Pearson correlation coefficients. All correlation coefficients are significant (p < 0.001). There were moderate negative relationships between the mangrove and urbanization variables. The correlations between sea level and all other variables were negative. Temperature and total precipitation were significantly positively correlated with mangrove variables but negatively correlated with urbanization variables. All variables showed strong positive relationships with other variables within the same system.

Evaluation of joint effects between urbanization and climate on mangroves

To investigate the joint effect of urbanization and climate on mangroves, we compared the total urbanization effects (direct + indirect effects; EUM + EUC × ECM) (Supplementary Fig. 4) and the total climate effects (ECM) (Supplementary Fig. 9). Based on their direction and magnitude, the joint effect of urbanization and climate can be classified into several patterns (Fig. 4c): Positive-Positive (PP), Negative/Positive (NP), and Negative-Negative (NN). A PP pattern indicates that urbanization and climate have mutually positive impacts on mangroves, suggesting that they jointly improve mangroves when acting together. Conversely, an NN pattern indicates that urbanization and climate have mutually negative impacts, jointly posing potential threats to mangroves. An NP pattern indicates that urbanization and climate have opposite effects on mangroves. The joint effect index is calculated as the ratio of total climate effects to total urbanization effects. A joint effect index greater than zero indicates a synergistic relationship (i.e., both positive/negative) between the effects of urbanization and climate, while an index less than zero represents the opposite impacts of urbanization and climate on mangroves. We utilized the Global Human Settlement Layer (GHSL) project’s GHS-DUG product62 to evaluate the relationship between urbanization degree and joint effect index. The GHS-DUG, a measure of urbanization degree based on population and settlement clusters obtained from the GHSL, was rescaled to range from 0 to 10. The mean value of the joint effect index in each urbanization degree bin was fitted to a non-linear curve.

Spatial variability

To reveal the spatial variability of urbanization-induced impacts on mangroves, the effect analysis was conducted at global, country and grid scales. At the global scale, all the data collected worldwide were incorporated into the inputs in the UCM model. We also examined the spatial pattern of countries with sufficient data to conduct regional analysis. However, it is essential to note that the effects might be scale-dependent. To explore the local effects, we separated the study area into 924 1° × 1° grids as study units and conducted the same effect analysis in each grid. We measured the overall spatial autocorrelation of the direct and indirect effects at the grid level using the global Moran’s I index. A positive Moran’s I index indicated a clustered pattern of effects and consistency with neighboring grids. The global Moran’s I index of direct (0.06) and indirect (0.01) effects was slightly larger than 0, with a reliable z score and P value, indicating clustered patterns of direct and indirect effects generally consistent with the surroundings.

Data selection

Mangroves are distributed along sheltered coastlines across the tropics and subtropics around the world. The distribution of mangroves was analyzed by dividing the area into equally sized 0.1° grid cells, resulting in 19717 study plots. Mangrove and climate variables were collected within 0.1° study plots, and urbanization variables were collected within a 100 km buffer zone of 0.1° study plots. In the UCM model, several representative variables U1-3 (impervious surface, population, GDP), C1-3 (sea level, winter air temperature, precipitation), and M1-2 (mangrove area, mangrove carbon stock) were employed to quantify three latent variables (urbanization, climate, and mangrove). These variables were chosen based on their demonstrated relevance to the model’s predictions and the availability of corresponding data at different scales. Geospatial datasets consisting of raster and vector formats derived from satellite images and open-source datasets were collected for 17 different years including 1996, 2000-2012, 2015, 2016, and 2020. The mangrove cover distribution was obtained from Global Mangrove Watch (GMW) (1996-2016)63, Global Database of Continuous Mangrove Forest Cover for the 21st Century (CGMFC-21) (2000 to 2012)64 and European Space Agency (ESA) WorldCover 10 m v100 (2020)65. Detailed data consistency and compatibility of mangrove datasets are provided in the Supplementary Methods (Supplementary Fig. 11). We conducted comparisons to assess the data consistency across multiple sources. Specifically, we compared the mangrove extent data in 2010 from the GMW and the CGMFC-21 dataset. Additionally, we examined the mangrove extent data in 2020, comparing the ESA dataset with the newly released GMW_202066 and HGMF_202067 datasets. Our analysis revealed that these datasets are generally consistent, with an R-squared value exceeding 0.9. This indicates a strong correlation between the different data sources, suggesting a high degree of consistency and compatibility in the mangrove extent information. Mangrove carbon stock was collected from the Global Mangrove Carbon Stocks Database (GMCSD) (2000 to 2012)68. The impervious surface datasets were extracted from the global artificial impervious area (GAIA) (1996–2018)69. Population datasets were obtained from the WorldPop project (2000–2020)70 and Gridded Population of the World (GPW) v3 (1995)71. Gross Domestic Product (GDP) datasets were collected from Gridded global datasets for Gross Domestic Product (1996–2015)72. Data for sea levels, winter air temperature and total precipitation were obtained from the Global Ocean Gridded L4 Sea Surface Heights73, and ERA574 by ECMWF/Copernicus Climate Change Service, respectively. The study also collected the distribution of climate zones based on Köppen-Geiger climate classification75, including tropical, subtropical/temperate and arid climate zones. Data preprocessing was conducted using Google Earth Engine and ArcGIS Pro.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.