Spatial Characteristics and Driving Forces Analysis of Water Conservation Services in Coastal Plain Cities—Yancheng as an Example

Abstract

The stability of ecosystems in coastal plain cities is fragile, and the interaction between humans and the land is complex, making the region’s natural water cycle capabilities more vulnerable to destruction. Quantitatively assessing the water conservation services in coastal areas and revealing its spatial characteristics and driving factors play a crucial role in the construction of regional ecological barriers and the assurance of ecological security. In this study, based on the water balance model, the spatial dynamics of water in the ecosystems of Yancheng from 2019 to 2021 were assessed in two dimensions: ecosystem and administrative subdivision. The response of the influencing factors to the water conservation was examined using a geographical detection. The results show the following: (1) Yancheng’s water conservation services increased (2019–2021), averaging 1.188 × 10⁹ m³/a. Spatially, it was higher in southeastern/northern sectors and lower in western/central regions, with wetlands and croplands contributing 93.76% collectively while others each accounted for <1%. (2) NDVI was the strongest driver of spatial heterogeneity (q = 0.736), followed by per capita water use, population density (q = 0.642), and DEM (q = 0.638); GDP per capita and annual precipitation exerted the weakest influences. (3) Factor interactions exceeded individual factors in explanatory power, dominated by population density synergies with per capita water use and NDVI, which most strongly controlled spatial patterns. (4) Optimization thresholds were identified: peak water conservation occurred at DEM 5.34–5.47 m and NDVI 0.37–0.42. This study provides a new perspective on water conservation in coastal areas, supplying serves as a reference for crafting specific water preservation strategies in the coming years.

1. Introduction

Ecosystem services encompass the goods and functions derived from ecosystems that sustain human survival, well-being, and development [1,2]. In response to escalating anthropogenic pressures-including rapid urbanization, economic globalization, and climate change-the valuation of ecosystem services has garnered increasing scientific and policy attention. According to China’s Guidelines for Accounting for Gross Value of Ecological Services, these services are systematically classified into three primary categories: provisioning services, regulating services, and cultural services [3,4]. Water conservation constitutes a critical regulating service [5], primarily characterized by the interception, infiltration, and storage of precipitation, thereby facilitating the regulation of hydrological flows and the water cycle [6]. Key manifestations include attenuation of surface runoff, replenishment of groundwater aquifers, moderation of river discharge seasonality, flood mitigation, drought alleviation, and water purification. This service plays a fundamental role in maintaining regional ecosystem stability, preventing reservoir sedimentation, and regulating local water cycles and surface hydrologic processes [7]. Coastal cities concentrate economic activities, thereby amplifying anthropogenic impacts and developmental pressures. Their ecosystems remain intrinsically vulnerable due to unique geomorphological and climatic constraints. These challenges are exacerbated by global marine degradation—including intensifying pollution, saltwater intrusion, and sea-level rise—which increasingly threatens the sustainability of low-lying coastal plains [8]. Consequently, elucidating the spatial distribution patterns of water conservation services within coastal plain cities [9] and identifying the key drivers influencing these patterns are essential prerequisites [10] for formulating effective strategies to enhance the ecological resilience of coastal regions [11].

The concept of water conservation originated in forest hydrology studies, primarily emphasizing the capacity of forest ecosystems to regulate surface water quality. Early research established water purification as an integral component of this functional framework [12,13,14,15,16,17,18]. Recent advancements have positioned water conservation research as a critical interdisciplinary frontier bridging ecology and hydrology, with significant progress in both theoretical frameworks and practical methodologies [19,20]. The spatial scale of such studies has expanded from local watersheds to inter-provincial regions and national strategic levels, while methodological approaches have evolved from single-indicator measurements to multi-model integration and multi-source data synthesis [21]. Concurrently, research scope has diversified to encompass foundational functional assessments, spatiotemporal evolutionary mechanisms, driver identification, and systemic analyses of complex ecosystem service interactions. Current research prioritizes ecologically vulnerable inland regions experiencing severe soil erosion and ecosystem degradation.

For instance, Lv et al. combined the InVEST model with Geodetector to quantify water conservation patterns in the Yellow River Basin (YRB), revealing a distinct “high southeast-low northwest” spatial distribution [22]. Their analysis further identified precipitation variability and the Grain-to-Green Program as dominant drivers of increased conservation capacity. Zheng et al. revealed human activity intensity exhibits a significantly adverse spatial association with water conservation service at the regional scale, as empirically demonstrated [23]. Their findings suggest that regions undergoing rapid urbanization require mitigation measures-such as ecological corridor construction-to counteract functional degradation. In Southwest China’s ecologically fragile zone, Liu et al. demonstrated that Yunnan Province’s water conservation capacity is influenced by climate variability and rubber forest expansion, as quantified through system dynamics and PLUS model simulations [24]. They proposed land-use structure optimization to enhance ecological resilience. Coastal cities, situated at marine-terrestrial ecotones, exhibit distinct characteristics: advanced economies, abundant natural resources, fragile ecosystems, and frequent anthropogenic disturbances. The prevalent perception of coastal regions as “water-rich areas” has led to an underestimation of their ecological challenges, resulting in limited research focus on water conservation [25]. Growing evidence indicates that escalating global warming, resource overconsumption, and coastal wetland contraction have severely compromised coastal ecosystem functions [26,27,28]. This degradation adversely impacts water conservation capacity and jeopardizes regional sustainable development. Current research on coastal water conservation primarily targets large-scale watersheds or broad ecosystem service assessments [29], with scant analysis at coastal plain or municipal scales.

This study employs quantitative methods to evaluate water conservation services in Yancheng (2019–2021) across ecosystem and administrative dimensions, supplemented by geodetector analysis to identify drivers of spatial heterogeneity. It is hypothesized that water conservation services in coastal plains (Yancheng) are primarily driven by nonlinear synergies between anthropogenic pressure and terrain-mediated hydrology, rather than linear effects of isolated factors. Based on this hypothesis, we spatially quantified the water conservation services and driving forces in Yancheng to untangle complex anthropogenic-terrain interactions, aiming to bridge ecosystem service dynamics with coastal sustainability. The findings are expected to inform the design of conservation strategies, ecosystem management, and policy implementation for coastal cities globally.

2. Materials and Methods

2.1. Study Area

Geographically anchored at 32°34′–34°28′ N, 119°27′–120°54′ E on Jiangsu’s central coast, Yancheng interfaces with the Yellow Sea through a 582 km littoral corridor constituting 93% of the provincial coastline [30], a typical coastal plain city (Figure 1). This transitional positioning within the subtropical–warm temperate ecotone induces distinct oceanically moderated bioclimatic signatures: (1) delayed vernal warming with persistent thermal depression. (2) attenuated autumnal cooling yielding higher fall means than spring. (3) elevated mean annual precipitation (1009.5 mm, 2019–2023) within a constrained thermal regime (15.3–16.3 °C) [31]. Yancheng coastal wetlands represent a high-value resource convergence zone, integrating extensive aquatic, terrestrial, biological, landscape, mineral, and energy reserves. This internationally significant mudflat wetland harbors 12 nationally first-class protected animal species (e.g., Elaphurus davidianus, Grus japonensis) and 84 second-class protected species, constituting a critical sanctuary for global biodiversity conservation [32].

2.2. Data Sources

The dataset utilized in this research consists primarily of Digital Elevation Model (DEM) data, boundary vector data, and driving factor datasets, as specified in

Table 1. Introduction of data sources.

Data Name	Data Source (Accessed on 8 February 2025)	Resolution
DEM (Digital Elevation Model)	Geospatial Data Cloud https://www.gscloud.cn/	30 m
Boundary data	Yancheng Geographic Information Public Service Platform https://jiangsu.tianditu.gov.cn/yancheng/	—
Annual mean temperature	Yancheng Meteorological Bureau http://js.cma.gov.cn	1 km
Annual precipitation
Annual evaporation
Population density	Yancheng Bureau of Statistics https://tjj.yancheng.gov.cn/col/col1779/index.html	1 km
Population size
GDP(Gross Domestic Product)
Per capita water consumption	Yancheng Water Affairs Bureau https://slj.yancheng.gov.cn/col/col1471/index.html	1 km
LULC (land-use/land cover) type		30 m
Park green space area	Yancheng Natural Resources and Planning Bureau https://zrzy.jiangsu.gov.cn/yc/	—
Hydrological information
NDVI (Normalized Difference Vegetation Index)	MOD13Q1 data in the MODIS product of NASA https://ladsweb.modaps.eosdis.nasa.gov/	250 m

. DEM elevation data was acquired from the Geospatial Data Cloud platform and processed subsequently through cropping and mosaicking using ArcMap 10.8.2. Factors selected as determinants for water conservation capacity include per capita Gross Domestic Product (GDP), population density, water consumption per capita, DEM elevation, annual precipitation, Normalized Difference Vegetation Index, green space area within parks, and groundwater volume. All datasets were standardized to the WGS_1984 coordinate system, with raster data resampled at a 30 m × 30 m spatial resolution to maintain uniformity in projection and scale. The resolution unification was achieved exclusively through resampling techniques, with continuous grids using bilinear interpolation and classified data using the nearest neighbor method.

2.3. Water Conservation Model

Water conservation quantification applies the water balance principle [33], treating a delineated area as a hydrologically isolated system. This methodology focuses on tracking water flux inputs and outputs across the system boundary. Under the law of mass conservation, the net change in stored water volume within the system equates to the water conservation capacity [34]. This research was calculated the quantity of water conservation in Yancheng for the years 2019, 2020, and 2021.

$$Q_{wr}={\sum }_{i=1}^n{A}_i\times \left(P_i-R_i-E{T}_i\right)\times {10}^3$$

(1)

where

Q_wr—The annual water conservation volume (m³/a);

n/i—the ecosystem types;

A_i—Spatial extent of ecosystem i (km²);

P_i—Total annual precipitation depth over ecosystem i (mm/a);

R_i—Surface runoff yield from ecosystem i (mm/a);

ET_i—Evapotranspiration flux from ecosystem i (mm/a);

The following formula is employed to calculate surface runoff (R_i):

$$R_i=P_i\times \alpha$$

(2)

In the calculation formula, α denotes the mean surface runoff coefficient, which was predominantly derived from the National Development and Reform Commission [35]. Exhaustive coefficients related to surface runoff and pertinent to this research are comprehensively detailed in

Table 2. Average runoff coefficient for each ecosystem type.

Level 1 Ecosystem Type	Level 2 Ecosystem Type	Average Runoff Coefficient/%
Forest	Evergreen broad-leaved forest	4.65%
	Deciduous broad-leaved forest	2.70%
	Evergreen coniferous forest	4.52%
	Deciduous coniferous forest	0.88%
	Mixed forest	3.52%
Shrub	Deciduous broad-leaved shrub	4.17%
Grassland	Grassland	9.37%
Plowland	Cropland	34.70%
	Cropland/natural	9.57%
Urban construction land	Green space	19.20%

.

2.4. Geographical Detector

The geographic detector is a statistical methodology employed for the detection of heterogeneity in the spatial stratification of elements, it can be downloaded for free (URL: http://www.geodetector.org/, accessed on 12 August 2025) [36], with the objective of elucidating the driving forces that give rise to these variations. It can capture nonlinear interactions influencing water conservation services, enabling robust driver analysis. Spatial dependency emerges when a predictor critically affects an outcome variable, necessitating aligned distributional geometries between driver and target across geographic space. The magnitude of spatial heterogeneity can be measured using the q value of the geographic detector. The geographic detector method determines interactive dynamics between paired factors through computational analysis comparing the q-statistic of individual explanatory variables with the superimposed q-value of their combination [37]. Interaction types are classified by comparing individual and superimposed q-values. The geographic detector’s four modules assess the following: (1) factor influence (q-statistic); (2) inter-variable synergy; (3) risk detector; and (4) ecological detector. This study applies modules (1), (2), and (4) [38].

The factor detector serves to detect spatial heterogeneity in the dependent variable (Y), thereby quantifying the extent to which an influencing factor (X) explains this heterogeneity through the q-statistic. The extent of explanation is measured by the q value, and the relevant formulas are as follows:

$$q=1-{\displaystyle \frac{1}{N{\sigma }^2}}{\sum }_{h=1}^L{N}_h\sigma \begin{array}{c}2\\ h\end{array}=1-{\displaystyle \frac{SSW}{SST}}$$

(3)

$$SSW={\sum }_{h=1}^L{N}_h\sigma \begin{array}{c}2\\ h\end{array}$$

(4)

$$SST=N{\sigma }^2$$

(5)

The q-statistic quantifies spatial heterogeneity and causal influence through variance decomposition. Where L denotes strata count, N_h and N denote the total number of cells contained within stratum h and the entire region, respectively, and σh² are the variances of the values of the dependent variable in stratum h and the whole region, respectively [39]. The q-statistic quantifies spatial autocorrelation and causal influence, where SSW denotes cumulative within-stratum variance and SST represents global variance. Higher q values (range [0, 1]) indicate stronger spatial self-organization in Y and greater explanatory power of X over Y since q directly measures the proportion of Y’s variance attributable to X’s stratification. A higher q value corresponds to more pronounced spatial differentiation in the dependent variable. When independent variable X generates the stratification, this q value demonstrates that X explains 100 × q% of Y. As the q value approaches 1, the explanatory ability of X regarding Y intensifies; conversely, lower values indicate diminished explanatory capacity.

This study utilized the GeoDetector method to quantify the spatial heterogeneity of water conservation services. With water conservation (QWR) as the dependent variable, the method assessed the explanatory power of driving factors on spatial variations in soil conservation capacity through q-statistic analysis. Factor interaction detectors elucidated interdependencies among these variables. Empirical studies consistently identify natural drivers, encompassing climatic variables, vegetation dynamics, and topographic attributes, alongside anthropogenic drivers, including population distribution, gross domestic product, and land-use patterns, as primary regulators of water conservation services across most biogeographical regions. However, the respective contributions of natural and anthropogenic drivers to water conservation services within coastal zones remain inadequately quantified. This investigation incorporated climatic variables (mean annual precipitation, temperature, and evaporation), terrain characteristics (slope and DEM), vegetation dynamics (NDVI), and anthropogenic elements (land-use type) as potential influencing factors.

3. Results

3.1. Interannual Characterization in Water Conservation

From 2019 to 2021, the total water conservation capacity in Yancheng reached 1.170, 1.186, and 1.208 × 10⁹ m³, respectively, with a triennial mean of 1.188 × 10⁹ m³. This metric demonstrated a consistent upward trend, reflecting a 0.376 billion m³ (or 3.21%) increase in 2021 compared to 2019. Significant disparities in water conservation capacity were observed across distinct ecosystem types (Figure 2). Wetland ecosystems dominated, exhibiting the highest multi-year average water conservation (0.689 billion m³), accounting for 58.04% of the total. Contributions from other ecosystems each constituted <1% of the regional water conservation.

3.2. Spatial Distributions of Water Conservation

The spatial distribution of water conservation capacity in Yancheng exhibits pronounced regional heterogeneity, delineated into high-capacity, medium-capacity, and low-capacity zones (see Figure 3). High-capacity areas (2.0–3.8 × 10⁸ m³/a) are predominantly clustered in central and southern regions, driven by intact wetland ecosystems, interconnected river networks, and dense vegetation cover that collectively enhance water retention efficiency. Medium-capacity zones (0.5–2.0 × 10⁸ m³/a) encompass extensive transitional landscapes, characterized by moderate ecosystem functionality interspersed with agricultural and urban land-uses. Low-capacity regions (0–0.1 × 10⁸ m³/a) occur along northern and western peripheries, where soil permeability constraints, intensive agricultural practices, and urban expansion substantially diminish natural water retention capabilities. This spatial patterning underscores the synergistic interplay of biophysical drivers (topographic gradients, soil properties, vegetative biomass) and anthropogenic forcing (land-use conversion, industrial encroachment).

A pronounced spatial heterogeneity in water conservation capacity exists across districts and counties, driven by topographic, climatic, vegetation, and land-use disparities (see Figure 4). Quantifying these variations identifies regions with robust versus compromised ecological functions, enabling tailored management strategies. Analysis of water conservation distribution during 2019–2021 reveals that Dongtai City and Dafeng District (southern Yancheng) exhibit the highest water conservation volumes, averaging >230 × 10⁶ m³. This prominence stems primarily from Dongtai’s extensive mudflats, which provide substantial rainwater and surface runoff storage capacity, and Dafeng’s vast wetlands, where vegetation and soils enhance natural water retention through filtration and subsurface storage processes. Sheyang County and Binhai County (northeastern Yancheng) show moderate water conservation levels (120–180 × 10⁶ m³), attributable to their low-lying terrain within the plains river-network zone. The perennial flow of the Sheyang River sustains elevated regional water tables. Meanwhile, Xiangshui, Funing, and Jianhu Counties demonstrate comparatively lower water conservation capacities (60–100 × 10⁶ m³), reflecting constrained natural retention mechanisms in these areas. Owing to the region’s elevated topography interspersed with low-relief hills, precipitation interception efficiency is markedly reduced, thereby accelerating surface runoff generation and restricting deep soil infiltration. Consequently, the resultant impairment of water conservation capacity is most acute in Yancheng’s urban core, specifically Yandu and Tinghu districts, where construction land coverage reached 35–40% in 2020. This urbanization effect manifests as (1) extensive impervious surfaces limiting rainwater retention and (2) replacement of natural ecosystems-wetland and forest cover here is <10%-with artificial lawns and street trees offering minimal hydrological functionality. Collectively, these drivers depress water conservation levels to ≤0.6 × 10⁸ m³ in the urban zone, while conversely, southern and coastal regions sustain maximal water conservation capacity (>2.3 × 10⁸ m³) through intact mudflats and wetlands. Spatially, this bifurcation reveals a pronounced functional gradient: high-performance water conservation in southern/coastal interfaces, moderate water-nutrient synergy in northern/western plains, and critically deficient capacity in the anthropogenically altered urban center.

3.3. Factors Identification of Spatial Heterogeneity of Water Conservation

3.3.1. Single Factor Analysis

Water conservation capacity in Yancheng is modulated by climatic and anthropogenic drivers. This study quantified correlations between water conservation (dependent variable) and key factors, including mean annual precipitation, elevation (DEM), normalized difference vegetation index (NDVI), and groundwater reserves (natural factors), alongside GDP per capita, population density, per capita water consumption, and urban green space coverage (anthropogenic factors). All independent variables were discretized using standardized methods and analyzed via geographical detector to derive explanatory power (q-values;

Table 3. The q values of influencing factors of the water conservation in Yancheng.

Influencing Factor	GDP per Capita (X1)	Population Density (X2)	per Capita Water Consumption (X3)	DEM (X4)	Annual Precipitation (X5)	NDVI (X6)	Green Space in Parks (X7)	Groundwater Volume (X8)
q value	0.28	0.642 **	0.655 **	0.638 **	0.27	0.736 **	0.539 **	0.490 **

Note: ** represents significance level p < 0.001.

).

Results demonstrate the following q-values for water conservation spatial heterogeneity, in descending order: NDVI (0.736) > per capita water consumption (0.655) > population density (0.642) > DEM (0.638) > urban green space (0.539) > groundwater reserves (0.490) > GDP per capita (0.279) > annual precipitation (0.272). The NDVI exhibited the dominant influence on water conservation variability, with a q-value exceeding 0.70, indicating its paramount role in governing spatial distribution. Per capita water consumption, population density, urban green space, and DEM also exerted substantial explanatory power.

High per capita water consumption typically reflects intensive demand for agricultural irrigation, industrial operations, and domestic use. Concurrently, construction land expansion in densely populated zones reduces natural vegetation cover, accelerating precipitation runoff and diminishing groundwater recharge efficiency. These dynamics underscore the profound sensitivity of Yancheng’s water conservation capacity to anthropogenic pressures. Urban green infrastructure mitigates these impacts by enhancing interception and infiltration capacity through vegetation and soil matrices, thereby reducing flood risks and augmenting localized water retention—evidence of positive feedback from urban ecological restoration. Groundwater reserves reflect long-term storage dynamics but show moderated explanatory power (q = 0.490) due to equilibrium between anthropogenic extraction and natural recharge. Spatial heterogeneity in GDP per capita and annual precipitation exerted minimal influence on water conservation (q < 0.30), attributable to homogeneous economic development and rainfall distribution across the study area.

3.3.2. Double-Factor Interaction Analysis

To decode the multifactorial drivers of water conservation spatial heterogeneity, we employed Geodetector’s interaction detection module to quantify synergistic effects between variables. Results unequivocally demonstrate that two-factor interactions universally amplify water conservation explanatory power (q-values) relative to isolated factors (Figure 5), indicating that Yancheng’s water conservation distribution is dominantly governed by cross-factor synergies rather than univariate controls. Critically, interactions achieving q > 0.99 include the following: population density × (per capita water use, annual precipitation, NDVI, park green space), DEM × (annual precipitation, NDVI), annual precipitation × (park green space, groundwater volume), and GDP per capita × NDVI. This hierarchy reveals population density as the paramount synergistic agent, followed sequentially by DEM, annual precipitation, and GDP per capita-a progression implicating anthropogenic pressure and bio-physical couplings as the dual engines of water conservation spatial patterning in coastal urban systems. Further analysis will be discussed in Section 3.3.3.

3.3.3. Suitability Partitioning Analysis

Risk detector analysis (95% confidence level) confirms significant spatial heterogeneity in mean water conservation (

Table 4. Suitable ranges or types of different influencing factors.

Influencing Factor	Suitable Range/Type	Trends in the Scope of Applicability	Mean Value of Water Conservation (Billion m3)
GDP per capita (million RMB/person/a)	9.20–10.03	Increasing applicability with classification	2.34
Population density (million people/km2)	0.022–0.038	Decreasing from classification 1 to 5	2.34
Per capita water consumption (m3/person/a)	711.49–836.67	Fluctuating upward (Classification 2 peaks)	2.34
DEM (m)	5.34–5.47	Decreasing applicability with increasing classification	2.56
Annual precipitation (mm)	982.77–1065.21	Classification 2 has the highest applicability	1.9
NDVI	0.37–0.42	Classification 4–5 applicability jumped	2.56
Green area of parks (ha)	1293.69–2420.33	Outstanding applicability of classifications 3 and 5	2.34
Groundwater volume (billion m3)	1.55–2.10	Classification 4 highest applicability	1.42

), driven by divergent responses to environmental and anthropogenic variables: while water conservation increases with rising GDP per capita (92,000–103,300 yuan/person/year), NDVI (peak stratum), and parkland area (1293–24,020.33 ha), it declines with elevated population density (0.022–0.038 × 10⁴ people/km²) and DEM values-with optima observed at minimal DEM (5.34–5.47 m) and maximal NDVI, signifying topography and vegetation as primary biophysical regulators. Crucially, socioeconomic thresholds simultaneously sustain high water conservation (e.g., per capita water use: 711.49–836.67 m³/year), revealing human activity as a co-dominant control. Conversely, annual precipitation and groundwater volume exhibit negligible influence-a consequence of spatially uniform rainfall distribution in topographically homogeneous terrain and anthropogenic homogenization of aquifers via centralized water infrastructure, which decouples traditional groundwater-recharge linkages, thereby attenuating their explanatory power for water conservation variability.

The applicability analysis further elucidates the drivers of strong interactions (q > 0.99) observed in Section 3.3.2. Spatial influence ranges of population density and DEM concurrently decreased. High-population areas exhibit water consumption and green space demand that overwhelm supply capacity, amplifying spatial mismatches in resource allocation. In high-elevation zones, terrain-driven rainfall redistribution directly governs vegetation patterns, triggering terrain-precipitation-vegetation feedback loops. Rainfall infiltration also sustains groundwater recharge, confirming a robust replenishment-storage feedback in regional water systems. Meanwhile, expanding influence ranges of per capita GDP and NDVI suggest that economic growth enables greater investment in ecological infrastructure. Their interaction reveals economic-ecological co-benefits.

4. Discussion

This study investigates water conservation capacity in Yancheng during 2019–2021, examining its spatiotemporal dynamics. The multi-year total water conservation value remained relatively stable, with a spatial distribution characterized by high values in the southeastern and northern regions and low values in the western and central areas. This pattern corresponds to the spatial configuration of natural features such as forests, wetlands, and river networks. Geographical detector analysis was employed to comprehensively assess driving factors. Natural factors (e.g., NDVI and DEM) exhibited the highest explanatory power, while anthropogenic factors (e.g., population density and human water consumption) also demonstrated significant influence. These results indicate that coastal cities’ water conservation capacity reflects a complex interplay between natural background conditions and anthropogenic interventions. Collectively, our findings reveal that balancing ecological protection and economic development is critical for optimizing water conservation functionality. This can be achieved via dual pathways: Natural Restoration Combined with Anthropogenic Regulation and Natural Restoration. The former approach—integrating human management with natural processes—is essential to harmonize ecological and socioeconomic objectives, thereby enhancing water conservation efficacy.

4.1. Trend Analysis of the Spatial and Temporal Evolution of Water Conservation in Yancheng

The multi-year average water conservation in Yancheng was 1.188 × 10⁹ m³, exhibiting minimal interannual variation. However, the contributions of different ecosystems to water conservation varied substantially. Wetlands and croplands contributed disproportionately, accounting for 93.76% of the total water conservation collectively, while all other ecosystems each accounted for less than 1%.

Most studies on the spatial characteristics and temporal evolution of water conservation have focused on larger watersheds. For example, Zhao and Gaolei et al. employed the Soil and Water Assessment Tool (SWAT) to develop a large-scale water conservation model for analyzing water conservation dynamics in the Yellow River Basin [40]. Their results indicated higher water conservation values in the northwestern regions relative to the southeast, with Gansu, Shaanxi, and Qinghai ranking among the top three provinces in average water conservation out of nine studied. Zhou Luyao et al. investigated spatiotemporal patterns of water conservation in China’s coastal zones, identifying a distinct sea–land gradient [33]. They further quantified water conservation using meteorological, land-use, and soil data to establish linkages between coastal land-use changes and water conservation services. J. Xu et al. analyzed spatial-temporal patterns of water conservation in Beijing for 2005 and 2010 across multiple spatial scales, including municipal, major functional areas, and key districts/counties [41].

Research on the spatiotemporal evolution of water conservation in localized coastal areas remains scarce, particularly regarding dynamic water conservation functional changes across diverse geomorphic features and raster scales. This study demonstrates that land-use categories significantly influence the spatial distribution of water conservation capacities in Yancheng. Southern and coastal regions-specifically Dongtai and Dafeng-represent high-value water conservation zones, where rich tidal flat and wetland ecosystems enhance vegetation-driven soil stabilization, soil water retention, and runoff regulation, thereby augmenting water conservation. The northeastern plains (encompassing Sheyang and Binhai), characterized by dense river networks, constitute medium-value water conservation areas, with the Sheyang River identified as a key facilitator of water conservation and infiltration. Conversely, low-value water conservation areas include the northern hilly terrain (Xiangshui, Funing, and Jianhu) and the central urban core (Yandu and Tinghu). This pattern arises primarily from topographic runoff losses and limited precipitation interception capacity in hilly regions, further exacerbated by concentrated constructed land cover and insufficient natural vegetation. These findings indicate that while the natural background of low-elevation coastal cities like Yancheng underpins water conservation, anthropogenic interventions can substantially amplify or diminish this function.

4.2. Analysis of Factors Affecting Water Conservation in Yancheng

Yancheng, influenced by oceanic conditions, exhibits a mild, humid climate with abundant rainfall. However, its topographic characteristics-particularly its low elevation and flat coastal terrain-result in rainfall patterns that differ significantly from watershed- and large-scale studies. In contrast to these studies reporting a strong rainfall influence on water conservation capacity (q < 0.3), rainfall in Yancheng plays a less dominant role due to uniform spatial distribution and gentle slopes. The primary natural drivers of water conservation here are vegetation cover (NDVI) and topography (DEM).

A robust correlation exists between high population density (q = 0.642) and per capita water usage (q = 0.638), likely attributed to diminished water conservation capacity from rapid urban expansion and groundwater extraction. Two-factor interaction analysis revealed that all paired interactions exceeded individual factor impacts, with the highest q-value (0.998) occurring for population density interacting with per capita water use and NDVI. This indicates that water conservation capacity declines sharply when intensive anthropogenic activities (e.g., high population density) coincide with degraded natural conditions (e.g., low vegetation cover). Conversely, coupling high NDVI with moderate population density substantially enhances water conservation capacity. Furthermore, applicability zoning analysis confirms that regions with the highest NDVI and lowest DEM exhibit maximal water conservation capacity, which substantiates the pivotal role of topography and vegetation.

Water conservation capacity is more pronounced in regions with higher per capita GDP, suggesting that moderate economic growth provides a foundation for ecological investments. Population densities exhibited a marked increase within the subsequent lower category (Category 2), followed by a substantial decline beyond this threshold. In summary, the spatial heterogeneity of water conservation capacity in coastal cities, exemplified by Yancheng, results from complex interactions between natural background conditions and human activities. For Yancheng, NDVI and topography dominate the natural dimension, signifying that vegetation cover is pivotal for water conservation and must be prioritized in protection and restoration efforts. The primary anthropogenic risks are population density and water consumption. The interaction strength (q), quantified by the Pearson correlation coefficient between population density and per capita water consumption, reaches 0.998. This indicates that high population density coupled with high water consumption significantly diminishes water conservation effectiveness. Simultaneously, moderate economic development can align with ecological protection. A per capita GDP of 90,000–100,000 yuan (≈12,500–13,900 USD) correlates with high water conservation capacity. Per capita GDP exceeding this range may be associated with resource overexploitation, leading to declining capacity. Thus, economic resources should prioritize integrated water source protection, pollution control, and ecological restoration initiatives. This approach ensures dynamic equilibrium between urbanization and water resource carrying capacity.

5. Conclusions

This study evaluates the spatial and temporal distribution characteristics of water conservation in Yancheng from 2019 to 2021, analyzing spatial variation trends across ecosystems and district/county dimensions. Using Geodetector, we investigated influencing factors and their interactions, yielding the following key findings:

(1) The water conservation services of Yancheng’s ecosystems increased during 2019–2021, with an average annual total of 1.188 × 10⁹ m³ yr⁻¹. Spatially, water conservation exhibited higher levels in the southeastern and northern sectors and lower levels in the western and central regions. Wetland and cropland ecosystems contributed predominantly to water conservation (93.76% combined), while other ecosystems each accounted for <1%.

(2) NDVI emerged as the most significant driver of water conservation spatial heterogeneity, followed by per capita water use, population density, and Digital Elevation Model (DEM). In contrast, GDP per capita and annual precipitation exerted the weakest influences.

(3) The explanatory power of factor interactions exceeded that of individual factors. Interactions involving population density dominated, with its synergy with per capita water use and NDVI exhibiting the strongest control over water conservation spatial patterns.

(4) Thresholds for optimizing water conservation function were identified: average water conservation peaked at DEM values of 5.34–5.47 m and NDVI values of 0.37–0.42.

This study, centered on Yancheng, provides insights for similar coastal plain cities. By integrating the water balance method and geographic detectors, it quantifies spatial patterns and key drivers of water conservation services, offering a practical framework for coastal ecosystem assessment. These findings support evidence-based water resource policies and ecological management. However, spatiotemporal dynamics and underlying mechanisms remain unexplored. To address this, future work should enhance data and model precision, leveraging neural networks and deep learning to simulate temporal dynamics and predict long-term trends. Such advances would optimize water-use planning and climate adaptation strategies.