Spatiotemporal Evolution Characteristics and Prediction of Habitat Quality Changes in the Poyang Lake Region, China

Abstract

The terrestrial spatial patterns were affected by human activities, primarily on regional land use (LU) changes, with habitat quality (HQ) serving as a prerequisite for achieving regional sustainable development. Assessing and predicting the spatiotemporal evolution characteristics of regional LU changes and HQ is critical for formulating regional LU strategies and enhancing ecosystem service functions. Using the Poyang Lake Region as our research object, this research employs LU data and utilizes the ‘InVEST’ model and hot-spot analysis to quantitatively evaluate the spatiotemporal changes in HQ during 2000–2020. The PLUS model is then applied to predict LU and HQ trends from 2020 to 2050. The findings are as follows: (1). From 2000 to 2020, the areas of forestland, shrubland, sparse woodland, paddy fields, and dryland in the Poyang Lake Region showed a decreasing trend, with reductions mainly occurring in urban expansion zones such as Nanchang City and largely converted into urban construction land. (2). Since 2000, HQ in the Poyang Lake Region has shown a slight retrogressive evolution, with significant spatial heterogeneity. HQ spatially exhibits a pattern of improvement radiating outward from major cities. (3). Predictions for 2030 to 2050 indicate that HQ in the Poyang Lake Region will continue to decline, with the most significant downward trends occurring in urban built-up areas and their peripheries. The spatiotemporal characteristics reveal an expansion ring around Poyang Lake and an east–west urban expansion corridor linking Pingxiang, Yichun, Xinyu, Nanchang, Fuzhou, Yingtan, and Shangrao. This study provided a research basis for LU direction and urban planning policies in the Poyang Lake Region and its surrounding areas, while also contributing to the construction of agrarian security patterns and the enhancement of ecosystem service levels in the region.

1. Introduction

Land is the most essential basis for the functioning of human society and different ecosystems. Its changes effectively illustrate the interaction between human actions and the ecological surroundings, making it a core focus in the study of spatiotemporal dynamics in geography [1,2,3]. Rational optimization of land use (LU) patterns can enhance the diversity of LU types and purposes, playing a crucial role in attaining efficient zoning management and promoting sustainable development [4]. However, under the influence of swift urban growth and industrial development, significant changes in LU patterns have occurred in many countries, including developing nations [5,6,7,8]. A notable phenomenon has been the substantial changes in regional habitat quality (HQ) [9,10]. HQ is important for regional steady development and human well-being [11,12]. Therefore, in the current context of pursuing high-quality regional development and ecological civilization, regional HQ has become a focal point in LU pattern research [13,14].

Evaluating the past, present, and future trends in regional HQ changes provides crucial references for policymakers to develop strategies for high-quality economic and ecological development and to study the construction of land security patterns [15,16,17,18]. As an essential reference for assessing a region’s ecological environment, research on HQ originated in the 1960s [19]. Initially, these investigations were limited in scale, primarily focusing on the impact of habitat on species reproduction, growth, migration, and biodiversity [20,21,22]. However, with advances in computing and the widespread application of 3S technologies, large-scale HQ assessments have become feasible [23,24]. Since 2007, the model ‘InVEST’ has provided significant theoretical and technical guidance for HQ evaluation [25,26]. For example, Bhagabati et al. [27] and Di Febbraro et al. [28] used it to accurately monitor the HQ of bird communities in Central Italy, demonstrating that HQ maps generated by ‘InVEST’ with SDM parameterization are highly reliable. Similarly, Kim et al. [29] explained the degree and causes of habitat degradation on Jeju Island using ‘InVEST’, providing valuable insights for homeland security planning in the zone.

Classifying and evaluating the spatiotemporal evolution of HQ also allows for predictions of LU/land-cover (LC) dynamic variation rules and HQ trends [30]. Recently, researchers have coupled HQ models with prediction models such as Cellular Automata, Markov, and others to forecast LU and HQ changes [31,32,33]. For example, models like CA-Markov [34,35], CLUE-S [36,37], and FLUS [38,39] have been applied, though the recently developed PLUS model has demonstrated greater advantages in large-scale and multi-class simulations and predictions [40]. Accurate studies of LU changes give a scientific basis for evaluating the spatiotemporal evolution characteristics of regional HQ [41,42].

China, being one of the world’s largest nations with a diverse range of topographical features, inevitably encounters tensions between the necessity of environmental protection and the demands of human development as a result of its swift urban expansion and economic progression [43]. The Poyang Lake Region, situated in Jiangxi Province, is an integral area within the middle Yangtze River urban agglomeration and the most intensively populated and economically vibrant area in the province [44]. As China’s largest freshwater resource, an internationally representative wetland, and Asia’s largest bird habitat, the ecological environment surrounding Poyang Lake has drawn significant international attention, especially regarding its environmental fragility and recovery trends [45]. Since 2000, the region has undergone rapid economic growth and urban development, resulting in significant LU/LC changes and heightened conflicts between economic growth and land protection. China approved the “Poyang Lake Ecological Economic Zone Plan” in 2009, emphasizing coordinated ecological and economic development, making ecological governance and LU optimization critical alongside economic growth [44]. Under the Chinese government’s promotion of “sustainable development” and “ecological civilization”, studying and predicting the time and space evolution is essential for clarifying regional ecological and economic development, reducing ecological risks, and supporting the development of the Yangtze River basin [38,46,47].

Given the region’s unique and critical environmental characteristics, researchers have conducted studies on vegetation indices and degradation causes [45], farmland utilization intensity [48], and urban ecological security patterns [49] in the Poyang Lake Region. However, existing research lacks a comprehensive analysis of the spatiotemporal evolution characteristics of HQ and future trends, which limits the general planning of LU and hinders the sustainable development of its ecological environment and economy. This study addresses these gaps by analyzing LU data, climatic factors, and boundary vector data (since 2000). Using ‘InVEST’ and hot-spot analysis, it quantitatively evaluates the spatiotemporal evolution of HQ. Furthermore, a ‘PLUS’ model is employed to predict LU and HQ trends over the next 30 years, aiming to provide research experience for land design, economic strategies, and ecological protection in the lake region.

2. Study Area

The Poyang Lake Region is in northern Jiangxi Province, China, linked to the Yangtze River. Its geographic range spans 113°34′ E–118°28′ E and 26°57′ N–30°04′ N, covering more than 9 million hectares (Figure 1). The region includes 64 cities of various sizes, with a permanent population of approximately 30.67 million, accounting for about 68% of Jiangxi Province’s total population, and an urbanization rate of 61%. As a significant strategic economic development area [44], the Poyang Lake Region features a subtropical monsoon climate with distinct seasons, an annual average temperature of 18 °C, and a multi-year average precipitation of 1700 mm.

Topographically, the central Poyang Lake Plain is flat, surrounded by mountains to the east, west, and south, forming a “basin-like” terrain with low central areas and higher peripheries [49]. The region is abundant in water and wetland resources, featuring rivers the Gan, Xin, Fu, Xiu, and Rao, along with lakes including Poyang, Junshan, and Qinglan. Poyang Lake, China’s largest freshwater lake, holds significant ecological value for water conservation, flood and drought regulation, wetland preservation, and biodiversity protection [48].

3. Materials and Methods

3.1. Data Acquisition and Preprocessing

Our data and products required for this study were obtained through open-source channels. Soil physicochemical property data at a 1 km resolution were from the World Soil Database. Digital elevation model (DEM) products were from the USGS website (https://earthexplorer.usgs.gov/, accessed on 5 February 2025), and slope data were from the DEM. All data of Meteorological were collected from the NMSDCC (National Meteorological Science Data Center of China) (http://data.cma.cn) and spatialized into grid datasets by the Kriging interpolation method. The boundary vector data for natural protected areas were sourced from the Jiangxi Provincial Forestry Bureau, while hydrological systems, road networks, and administrative boundaries were from the Resource and Environment Science Data Center of the Chinese Academy of Sciences (https://www.resdc.cn). Coordinates of highway toll stations, railway stations, city centers, and county centers were collected via Python web (version 3.13.1) scraping from mainstream commercial maps, and distances to railway stations and toll stations were calculated using Euclidean distance analysis based on the ROI coordinates from the map data.

Socio-economic information was retrieved from two official documents (Jiangxi Statistical Yearbook (2000–2020) and Jiangxi Province Seventh Population Census Report). All spatialized data visualizations were preprocessed using QGIS software (version 3.34). Raster data were standardized to ensure uniform rows and columns to accommodate model calculations, with all rasterized data recollected to a pixel size of 30 m. Finally, the projection coordinate for all images was set to CGCS2000 to meet the geographic alignment requirements during model computations.

3.2. HQ Assessment

We used the HQ module of ‘InVEST’ (3.10.2) to assess HQ. The model captures spatial distribution and patterns of HQ and habitat degradation (HD). Following a method proposed by Bhagabati et al. [27], they were performed using the following formulas:

$$D_{xj}=\sum _1^{r}r\sum _1^y\left({\displaystyle \frac{{\omega }_w}{\sum _{r=1}^n{\omega }_w}}\right)\times r_y\times i_{rxy}\times {\beta }_x\times S_{jr}$$

$$i_{rxy}=\left\{\begin{array}{l}1-\left({\displaystyle \frac{d_{xy}}{d_{rmax}}}\right) \text{Exponential decay}\\ \exp \left[-\left({\displaystyle \frac{2.99}{d_{rmax}}}\right)\right] \text{Linear decay}\end{array}\right.$$

$$Q_{xj}=H_j\left[1-\left({\displaystyle \frac{D_{xj}^z}{D_{xj}^z+k^z}}\right)\right]$$

With D_xj the HD of grid cell x for habitat type j; define r as the number of threat factors (Tf), with y indicating the grid cell containing Tf r; w the weight of each Tf, while r_y signifies the stress value of grid cell y for Ts r; β_x the disturbance resistance level of habitat at grid cell x, and S_jr indicates the sensitivity of habitat type j to Tf r; i_rxy the influence of grid cell y, containing Tf r, on grid cell x; d_xy the distance between grid cells x and y; d_rmax the max impact distance of Ts r, while Q_xj corresponds to the HQ of grid cell x for habitat type j; H_j signifies the suitability of habitat type j, z is the normalization constant (default value 2.5), and k is a constant typically set to half the max HD (default value 0.5).

Following the ‘InVEST’ model manual and related research [29], we identified major Tf and their weights based on the characteristics of the Poyang Lake region. Specifically, paddy fields, drylands, urban areas, rural settlements, other construction lands, and main roads were chosen as Tf. We list the relevant information in

Table 1. Weight and max impact distant of threat factors.

Threat Factors	Max Distance of Influence (km)	Weight	Decay Type
Paddy field	1.0	0.6	Linear decay
Dry land	1.0	0.6	Linear decay
Urban land	8.0	1.0	Exponential decay
Rural settlements	6.0	0.8	Exponential decay
Other construction land	5.0	0.6	Exponential decay
Unused land	3.0	0.5	Linear decay

and

Table 2. Habitat suitability and sensitivity to threat factors of various LU types.

Land Use Types		Habitat	Paddy Field	Dry Land	Urban Land	Rural Settlements	Other Construction Land	Unused Land
Cultivated Land	Paddy field	0.5	0.0	0.0	0.8	0.7	0.6	0.4
	Dry land	0.3	0.0	0.0	0.8	0.7	0.6	0.4
Forest land	Forest	1.0	0.6	0.7	0.9	0.8	0.8	0.5
	Shrub	0.8	0.6	0.7	0.8	0.6	0.7	0.5
	Sparse woodland	0.7	0.5	0.6	0.7	0.7	0.8	0.4
	Other woodland	0.6	0.5	0.6	0.7	0.7	0.8	0.4
Grassland	High covered grassland	0.7	0.6	0.7	0.9	0.8	0.7	0.6
	Medium covered grassland	0.6	0.5	0.6	0.9	0.8	0.7	0.6
	Low covered grassland	0.5	0.5	0.6	0.9	0.8	0.7	0.6
Water area	River	0.9	0.4	0.5	0.9	0.8	0.7	0.3
	Lake	1.0	0.4	0.5	0.9	0.8	0.6	0.3
	Reservoir	0.6	0.4	0.5	0.9	0.8	0.6	0.3
	Beach land	0.7	0.55	0.6	0.7	0.6	0.7	0.3
Construction Land	Urban land	0.0	0.0	0.0	0.0	0.0	0.0	0.0
	Rural settlements	0.0	0.0	0.0	0.0	0.0	0.0	0.0
	Other construction land	0.0	0.0	0.0	0.0	0.0	0.0	0.0
Unused land	Unused land	0.3	0.3	0.4	0.5	0.4	0.5	0.0

.

3.3. LU Prediction

The PLUS model introduces a rule-mining framework that is grounded in land expansion analysis and a cellular automata (CA) model using multi-type random seeds. It shows higher accuracy and more realistic landscape distributions in LU change simulations [32]. The LU prediction process based on PLUS generally involves the following steps:

3.3.1. LU Type Development Probability

Referring to the method proposed by Rashmi and Lele [34], the study utilizes the expansion analysis strategy module in ‘InVEST’ to extract the LU changes for 2010 and 2020 in the study area. Six environmental factors—elevation, slope, temperature, precipitation, proximity to water systems, and soil class—and nine socio-economic factors—GDP, population density, and distance to railways, train stations, highways, highway entrances, major roads, city centers, and county centers—are selected. Using the Random Forest algorithm, the changes in different LU types are sampled and calculated to obtain the impact weights of each factor on LU type changes and the development probabilities for different LU types. The development probability is expressed as follows:

$$P_{ix}^d={\displaystyle \frac{\sum _{n=1}^{M}I\left[h_n\left(x\right)=y_n\right]}{M}}$$

With d represents the transition conditions among different LU types, where ‘0’ indicates no allowable transition and ‘1’ means a specific LU type can transition to another; x refers to the vector that encompasses various driving factors influencing these transitions; h_nx the result of the decision tree at n; Y_n the classification result of the decision tree at n; the I[h_n(x) = y_n] serves as an indicator function for the classification result; M represents numbers of decision tree; $P_{ix}^d$ the development probability of the i-th LU type at grid x under condition d.

3.3.2. Markov Model

Our study uses past LU data from the research area and follows the method proposed by Twisa et al. [21] to predict LU changes using the Markov model. The equation for predicting LU types is as follows:

$$S^{\left(t+1\right)}=S^{\left(t\right)}{P}_{ij}=S^{\left(0\right)}{P}_{ij}^{\left(t+1\right)}$$

With S^(t+1) the predicted LU type at time t + 1; S⁽⁰⁾ the initial LU type at the starting time; S^(t) the LU type at time t; P_ij the probability of LU converting from type i to type j.

3.3.3. CA Model

Based on the Markov model, the areas and transition probabilities of various LU types are predicted. Using a geographic cellular automata (CA) model, which incorporates mechanisms such as adaptive inertia, random seed generation, threshold decrement, neighborhood weight tables, and transition matrices, dynamic LU changes can be automatically imitated. The transition matrix is set according to past LU changes in the study area and real-world conditions [20,32]. The calculated neighborhood weights for different LU types are presented in

Table 3. Realm weight parameters of different LU types.

Land Use Types	Realm Weight	Land-Use Types	Realm Weight	Land-Use Types	Realm Weight
Paddy field	1	Dry land	0.4478	Forest	0.8773
Shrub	0.3381	Sparse woodland	0.4346	Other woodland	0.0348
High covered grassland	0.1646	Medium covered grassland	0.0572	Low covered grassland	0
River	0.0682	Lake	0.2177	Reservoir	0.1774
Beach land	0.1701	Other construction land	0.1145	Rural settlements	0.1757
Urban land	0.7889	Unused land	0.0203

.

3.3.4. Accuracy Test

(1) Kappa Coefficient

The Kappa coefficient serves to evaluate the accuracy of predictions. The calculation formula is as follows:

$$Kappa={\displaystyle \frac{\left(P_0-P_c\right)}{\left(P_p-P_c\right)}}$$

P₀ represents the accurate proportion of simulation; P_c denotes the correct proportion of prediction in random conditions; and P_p indicates the accurate proportion of simulation in ideal conditions. This coefficient varies from −1 to 1. A higher value signifies improved prediction accuracy.

(2) Verification of Historical Trend Consistency

This study established a dual qualitative validation approach, integrating historical backtracking and regional comparative analysis, to assess the reliability of historical trends simulated by the PLUS model. On one hand, long-term land use datasets (1980–2020) were used to identify land use change trends in the region from a broader temporal and spatial perspective, under a shared historical context and consistent policy conditions. On the other hand, the spatial patterns of the predicted results were examined to determine whether they align with the spatiotemporal logic of regional land use evolution and the fundamental principles of geography.

4. Results

4.1. LU Spatiotemporal Dynamics

Overall, throughout the timeframe spanning from 2000 to 2020, the LU area distribution ranked as follows: Forest land > Cultivated land > Water area > Grassland > Construction land > Unused land (

Table 4. Area and percentage of LU types in study area (km², %).

LU Types		2000		2005		2010		2015		2020
		Area	Pct	Area	Pct	Area	Pct	Area	Pct	Area	Pct
Cultivated Land	Paddy field	22,846.9	24.8	22,583.9	24.5	22,554.7	24.5	22,338.1	24.2	22,167.0	24.1
	Dry land	7337.2	8.0	7314.1	7.9	7245.3	7.9	7137.4	7.7	7063.6	7.7
Forest land	Forest	33,307.9	36.1	33,330.1	36.2	34,226.6	37.1	33,993.1	36.9	33,918.5	36.8
	Shrub land	6876.7	7.5	6825.9	7.4	6444.7	7.0	6421.3	7.0	6413.7	7.0
	Sparse woodland	9496.1	10.3	9442.1	10.2	8861.0	9.6	8738.9	9.5	8666.6	9.4
	Other Forest Land	365.9	0.4	396.4	0.4	364.8	0.4	378.2	0.4	376.3	0.4
Grassland	High covered grassland	1716.2	1.9	1654.1	1.8	1523.0	1.7	1604.1	1.7	1602.3	1.7
	Medium covered grassland	1381.9	1.5	1402.1	1.5	1275.2	1.4	1260.5	1.4	1248.9	1.4
	Low covered grassland	27.3	0.0	28.4	0.0	27.3	0.0	27.5	0.0	27.2	0.0
Water area	River	988.5	1.1	907.7	1.0	1183.7	1.3	1183.6	1.3	1184.7	1.3
	Lake	1947.7	2.1	2824.5	3.1	1080.9	1.2	1772.5	1.9	1188.2	1.3
	Reservoir	1392.7	1.5	1290.2	1.4	1316.9	1.4	1385.5	1.5	1404.8	1.5
	Beach land	1547.4	1.7	1055.6	1.2	2576.2	2.8	1840.5	2.0	2414.3	2.6
Construction Land	Urban land	450.7	0.5	704.5	0.8	914.8	1.0	988.4	1.1	1023.6	1.1
	Rural settlements	1440.7	1.6	1569.7	1.7	1624.1	1.8	1622.1	1.8	1625.5	1.8
	Other construction land	129.5	0.1	224.2	0.2	411.6	0.5	953.1	1.0	1316.2	1.4
Unused land	Unused land	920.1	1.0	622.9	0.7	546.0	0.6	531.7	0.6	531.8	0.6

). The overall extent of Cultivated land exhibited a minor decline, falling from 32.8% in 2000 to 31.7% in 2020. Specifically, both Paddy field and Dry land experienced a slight reduction in area. The area of Forest land exhibited a different trend, reaching its peak in 2010 before slightly declining, with Shrub land and Sparse woodland continuously decreasing. The area of Grassland remained relatively stable, with its proportion decreasing slightly from 3.4% in 2000 to 3.1% in 2020. The area of Water area showed an initial increase followed by stabilization, with significant fluctuations in River and Lake areas, while the Reservoir area remained relatively stable, and Beach land outspread significantly. The area of Construction land nearly doubled over the 20 years, increasing from 2.2% in 2000 to 4.3% in 2020. Since 2000, the area of Unused land decreased from 1.0% to 0.6%.

From a spatial perspective, the overall LU pattern in the Poyang Lake Region remained largely unchanged between 2000 and 2020 (Figure 2). Specifically, Cultivated land was predominantly located in the Poyang Lake Plain and low-altitude flat valleys within the region, which are conducive to agricultural practices, exhibiting no significant spatial distribution changes over the 20 years. Forest land was extensively distributed across the eastern, western, and southern mountain/hilly areas of the study region, predominantly comprising Forest and Sparse woodland dominated by various tree species. Grassland was concentrated in the plains surrounding the Poyang Lake shoreline, with slight changes following interannual variations in the lake’s waterline. Changes in the Water area were primarily influenced by the fluctuations in lakes and rivers, with notable oscillations in the lake area from 2000 to 2020, attributed to the dual impacts of climate change and water conservancy projects. Construction land exhibited a marked trend of scattered expansion, with the fastest growth occurring between 2010 and 2015, mainly resulting from the transformation of forest and cultivated land at urban edges and along urban connectivity corridors. Unused land was mainly sandy areas along the southwestern shores of Poyang Lake, characterized by scattered patches. Since 2000, it has continuously declined slightly, with most of the land being used for Construction land.

4.2. Spatial and Temporal Dynamics of HQ

The results of HQ calculations indicate that the average HQ scores for the Poyang Lake urban agglomeration in 2000, 2005, 2010, 2015, and 2020 were 0.7224, 0.7215, 0.7163, 0.7131, and 0.7096, respectively. On the whole, the HQ score was relatively high (Figure 3). The change rates every five years were −0.13%, −0.71%, −0.45%, and −0.49%, with a total change rate of −1.77%. Generally, HQ showed a declining trend.

The HD scores calculated by ‘InVEST’ were categorized into seven levels: strong degradation (HQD < −0.5), obvious degradation (−0.5 ≤ HQD < −0.2), slight degradation (−0.2 ≤ HQD < 0), keep stable (HQD = 0), slight improvement (0 < HQD ≤ 0.2), obvious improvement (0.2 < HQD ≤ 0.5), and strong improvement (HQD > 0.5). The areas and proportions of these categories were analyzed (

Table 5. Area and percentage (Pct) change of the HQ from 2000 to 2020 (km², %).

HQ Changes	2000–2005		2005–2010		2010–2015		2015–2020		2000–2020
	Area	Pct	Area	Pct	Area	Pct	Area	Pct	Area	Pct
Strong degradation	183.76	0.20	373.11	0.40	321.05	0.35	183.76	0.20	1131.22	1.23
Obvious degradation	754.63	0.82	2313.16	2.51	507.75	0.55	754.63	0.82	3784.56	4.11
Slight degradation	23,281.87	25.27	30,166.68	32.74	33,543.58	36.40	23,281.87	25.27	42,889.95	46.55
Keep stable	61,407.44	66.64	53,136.41	57.66	56,269.58	61.06	61,407.44	66.64	36,754.01	39.89
Slight improvement	5293.22	5.74	4759.22	5.16	712.76	0.77	5293.22	5.74	4663.62	5.06
Obvious improvement	863.91	0.94	1290.09	1.40	782.68	0.85	863.91	0.94	2606.10	2.83
Strong improvement	363.36	0.39	112.25	0.12	14.33	0.02	363.36	0.39	311.31	0.34

) along with their spatial distribution (Figure 4).

Over the 20 years, the areas of HQ categorized as degraded, stable, or improved experienced some fluctuations. However, the extent of degraded regions consistently exceeded that of improved regions. The disparity was most pronounced during 2010–2015, with degraded areas accounting for 37.3%, improved areas only 1.64%, and 61.06% remaining stable. During this period, degraded regions were mainly concentrated around urban areas. Overall, 39.89% of the region’s HQ remained stable, primarily located in the western section; 51.88% experienced degradation, with 46.55% classified as slight degradation in this period. This degradation was most notable around major cities such as Nanchang, Yichun, Jiujiang, and Shangrao, as well as along the connecting highways. Only 8.23% of the region showed improvement in HQ, of which 5.06% experienced slight improvement, mainly concentrated around the Poyang Lake area.

Taking 2020 as an example, the hotspot analysis of HQ and HD in the Poyang Lake Region demonstrates notable spatial variation in HQ (Figure 5a). HQ hotspots are gathered in the northeastern, southeastern, and western areas, indicating that these regions generally exhibit higher HQ. In contrast, cold spots are mainly located in the Poyang Lake plain, particularly in Nanchang’s urban zone and its neighboring counties. From the perspective of HD (Figure 5b), degradation hotspots are concentrated in the mid-eastern, southwestern, and central parts.

4.3. Prediction of HQ

4.3.1. Prediction of LU

This study uses the LU scenarios for 2030, 2040, and 2050 predicted by the PLUS model as the basis to forecast the spatiotemporal characteristics in HQ in the Poyang Lake Region over the next 30 years. To validate the reliability of the LU predictions generated by the PLUS model, we first predicted the 2020 LU patterns based on pre-2020 LU data. After comparing the predicted and actual 2020 LU patterns, the Kappa coefficient was 0.827. This shows that the PLUS model predictions are accurate and reliable, making them suitable for use in forecasting HQ.

To evaluate the consistency of historical trends in land use predictions for the Poyang Lake area, this study extracted the land use structural evolution sequence of Jiangxi Province from 1980 to 2020 using a homologous land use dataset. The results were then compared with the projected changes in land use proportions for the Poyang Lake Region from 2030 to 2050 (Figure 6). The predicted changes in land use area showed strong consistency with the province-level evolution trends, particularly for land use types sensitive to human activities, such as cultivated land and construction land. As the core development area in northern Jiangxi, the Poyang Lake area shares high consistency with the entire province in terms of economic foundation, policy orientation, and sociocultural context. This regional-to-local trend inheritance strongly supports the numerical reliability of the PLUS model’s predictions.

From the perspective of spatial patterns, the prediction results strictly follow the spatial proximity effect of Tobler’s First Law of Geography. The expansion of construction land exhibits an edge-permeation growth pattern, while the intensification of cultivated land and the morphological evolution of water systems both demonstrate typical characteristics of spatial autocorrelation (Figure 7). Such predictions, which align with the spatial evolution rules of geographic entities, provide spatial evidence supporting the trend rationality of the model simulation results.

4.3.2. HQ Prediction Based on PLUS and InVEST

Using the predicted LU data for the years 2030, 2040, and 2050, ‘InVEST’ and ‘PLUS’ were employed to forecast the HQ spatiotemporal distribution in the next 30 years (Figure 8). The area and percentage of different HQ levels for different periods were also calculated (

Table 6. Area and Pct of the HQ levels in the future.

HQ Levels	2030		2040		2050
	Area (km2)	Pct (%)	Area (km2)	Pct (%)	Area (km2)	Pct (%)
Low (0–0.2)	4910.82	5.33	5852.22	6.35	6734.95	7.31
Poor (0.2–0.4)	7447.14	8.08	7322.49	7.95	7207.52	7.82
Medium (0.4–0.6)	24,438.07	26.52	24,162.82	26.22	23,892.04	25.93
Good (0.6–0.8)	18,267.46	19.83	18,075.06	19.62	17,950.73	19.48
High (0.8–1.0)	37,076.16	40.24	36,725.90	39.86	36,353.25	39.46

).

The results show that in 2030, 2040, and 2050, the mean HQ values in the Poyang Lake Region will be 0.7026, 0.6956, and 0.6888, respectively. Similar to the trend observed from 2000 to 2020, the HQ in the region will generally remain at a relatively good level over the next 30 years, but it will continue to show a clear downward trend. This suggests that if the current LU practices and protection intensities are maintained, and the economic development model remains unchanged, the ecological environment of the Poyang Lake Region will face increasing pressure, and HQ will continue to decline.

From a spatial perspective, the predicted results further underscore the characteristics of HQ, revealing that the most significant degradation occurs in the Nanchang urban area. In contrast, the regions surrounding Jiujiang, Shangrao, and Yichun exhibit less significant degradation. The pattern of HQ degradation parallels the LU change simulation, with the expansion of construction land identified as a key contributing factor. Notably, the degradation zones correspond to the Poyang Lake HD circle and the east–west HD belt spanning from Pingxiang, Yichun, Nanchang, Fuzhou, Yingtan, to Shangrao.

5. Discussion

For an extended period, the Earth’s surface has been affected by human activities, which are primarily manifested through changes in regional LU patterns [50]. LU transformation serves as one of the most direct indicators of urban development [51]. As research on LU/LC change advances, LU change is considered an important cause of global environmental change [52]. The study of LU can more comprehensively reveal the dominant and latent characteristics of LU changes during different periods [53]. Overall, the rapid growth of activities like urban development and agricultural land reclamation unavoidably results in the ongoing reduction of natural ecosystems, including forests, grasslands, and aquatic systems, alongside a decrease in the services provided by these ecosystems. This calls for researchers and government managers to devise effective measures to reverse this trend [54].

This study found that from 2000 to 2020, the expanse of construction land around Poyang Lake nearly doubled, while the areas of cultivated land, forest land, grassland, and unused land remained generally stable or decreased to varying degrees. This indicates that the urbanization process has intensified the exploitation and utilization of land resources in the Poyang Lake Region. This outcome aligns with the results presented by Deng et al. [55], who proposed that the spatiotemporal displacement of urban development and ecological initiatives in China would change the availability and requirements of ecosystem services. Among them, the changes in cultivated land were generally stable, which is mainly due to China’s strict farmland protection policies and the “balance of occupation and compensation” strategy since the 21st century [56]. In contrast to cultivated land, the reduction in forest land was smaller, and the grassland area remained relatively stable, which may be related to China’s emphasis on ecological environmental protection, strengthening urban green space construction, and sustainable forest development [57,58]. The study also found that during 2000–2020, the area of rivers and lakes fluctuated significantly, while the area of reservoirs remained relatively stable, and the area of tidal flats significantly increased. This could be linked to climate change (including increased temperatures, frequent extreme rainfall, and drought events) and water conservancy projects (such as the construction of the Poyang Lake water conservancy hub) [59,60].

From a spatial distribution perspective (2000–2020), cultivated land and grassland primarily occupied the plains (with a small amount in low-altitude flat valleys), and forest land was concentrated in mountainous regions. This aligns with the findings of Wang [61]. However, due to the effects of various factors such as dam water storage, droughts, floods, streamflow size, and human activities [62], the scope of the Poyang Lake area experienced significant fluctuations during this period. These findings provide new insights for further studying the toppling effect of the outflow from Poyang Lake. Additionally, during 2010–2015, the area of construction land grew significantly, with many previously unused lands and reduced cultivated and forest lands being converted into construction land. This is mainly due to the strong infrastructure and facilities during China’s 12th Five-Year Plan and the accelerated urbanization process [63]. Overall, the study further demonstrates that the increasing demands for social development and technological progress, along with natural factors, can disrupt the LU change patterns, and the conversion of LU types often exhibits dispersive and diverse characteristics in spatial distribution [64].

Given the swift pace of urban growth and intense land development, changes in LU can effectively evaluate the spatial and temporal shifts of regional HQ [65,66]. Through HQ research, the ecological condition of a region or even the global environment at different times and spaces can be evaluated [19,41,67]. Studies show that HD can reflect the degree of stress on habitats, with higher degradation indicating higher threats and a greater likelihood of HD [68]. This study found an overall decline in HQ over 20 years, with the greatest degradation between 2010 and 2015. The most severe degradation occurred in the urban belt extending from Nanchang city and the surrounding cities of Pingxiang, Yichun, Xinyu, Fuzhou, Yingtan, and Shangrao, further indicating that the rapid economic growth, urbanization, and industrialization have negative impacts on regional HQ [68,69]. Hotspot and cold spot analyses showed that HQ hotspots were concentrated in the northeast, southeast, and western mountainous areas, while cold spots were concentrated in urban and suburban areas. This is mainly due to the fact that the rugged terrain of the mountainous area preserves rich natural resources, which include nature reserves like Lushan National Nature Reserve, Jiuling Mountain National Nature Reserve, and Taohongling National Nature Reserve. Urbanization and industrial development are mainly concentrated in the cities and their adjacent areas, which was also found by Wang et al. [70] on HQ evolution in Shenzhen.

An analysis by the Jiangxi Provincial Bureau of Statistics on the economic development trends in 2021 indicates that the GDP has increased approximately 13 times from 2000 to 2020 (https://tjj.jiangxi.gov.cn/jxstjj/), accompanied by significant spatial restructuring in the Poyang Lake region. The <Poyang Lake Ecological Economic Zone Plan>, approved as a national strategy by the State Council of China in 2009, aims to reshape the regional economic geography through industrial upgrading while constraining development intensity with institutional innovations such as ecological compensation mechanisms and green GDP accounting. The focus is on promoting the construction of a lakeside urban cluster centered around Nanchang and Jiujiang. This has been achieved through the expansion of transportation networks, industrial park agglomeration, and land transfer policies, forming a “policy-investment-spatial expansion” transmission chain. However, intensive policy interventions and human activities inevitably alter the region’s land-use patterns [71,72]. This study shows that from 2000 to 2020, the area of built-up land in the Poyang Lake Region increased by 113%, with spatial distribution highly coupled with the “One Lake, Two Belts” urban cluster in the plan.

Although the farmland “dynamic equilibrium” policy maintains a balance in total cultivated land through off-site compensation, the compensated farmland is often located in ecologically sensitive hilly areas, potentially leading to habitat fragmentation [73]. We hypothesize that “policy-driven spatial displacement” might lead to three types of ecological risks: (1) the urban expansion areas experience increased heat island effects and altered hydrological cycles due to the surge in impervious surfaces; (2) the farmland compensation areas face accelerated soil erosion due to intensified cultivation; and (3) aquatic habitats (e.g., for species like Neophocaena asiaeorientalis) are threatened. Additionally, internal contradictions in the policy system may exacerbate the complexity of HQ evolution [74]. For instance, while the ecological economic zone plan establishes wetland protection boundaries, local governments, under GDP assessment pressures, tend to adopt the model of “bundling ecological restoration projects with industrial development” (e.g., commercial real estate accompanying lakeside landscape belts) to increase land value. Thus, under a multi-level governance framework, the implementation of ecological policies may be partially neutralized by economic rationality, ultimately impacting ecosystems through LU changes and influencing HQ evolution. Therefore, it is particularly important to predict regional LU patterns and HQ evolution trends as early as possible to guide government policy-making.

Currently, the global trend of urban expansion is expected to continue, and its impact on the natural environment and ecosystems will inevitably lead to serious biodiversity loss and habitat fragmentation [75,76]. Therefore, predicting future LU change patterns and the scale of urbanization is crucial for alleviating regional ecological pressure and improving HQ [77]. In our paper, we simulated the trends of LU and HQ changes in the future based on ‘PLUS’ and ‘InVEST’ models. The results demonstrate particularly pronounced HQ degradation in urbanizing zones surrounding major cities like Nanchang, which should be prioritized as key control areas within ecological protection redlines. We recommend that local governments implement stricter development restrictions in territorial spatial planning and explore mechanisms integrating ecological restoration with construction land quota exchanges to balance development and conservation. Furthermore, the ongoing urban expansion trend is projected to persist, potentially exacerbating increased surface runoff and diminished hydrological connectivity in wetlands [78]. To mitigate these impacts, the comprehensive Poyang Lake basin planning should incorporate enhanced sponge city requirements and restore natural buffer zones in critical catchment areas. The predicted patterns mirror similar trends observed in Yellow River floodplains [79,80], while the ecological risks (e.g., accelerated soil erosion) associated with farmland compensation zones resulting from urban expansion necessitate optimized spatial implementation rules for the “compensation balance” policy to prevent overexploitation of marginal lands. These findings collectively suggest that the delineation of urban growth boundaries should incorporate dynamic adjustments based on HQ modeling outcomes to facilitate more scientifically grounded territorial spatial planning.

However, we must acknowledge that the design of the HQ module initially focused more on the interaction between human factors and the environment, and the relationship between the internal interactions of ecosystems and HQ patterns was relatively lacking. This could result in a degree of subjectivity in the outcomes produced by the HQ module of ‘InVEST’. Furthermore, while our current framework does not explicitly parameterize socio-political dynamics—including negotiation processes among local governments, communities, and conservation agencies regarding ecological compensation standards or land exchange disputes—these latent variables may indirectly influence land use patterns through planning adjustments. Similarly, resident perceptions of habitat degradation and adaptation strategies (e.g., modifications to traditional farming practices or ecological migration demands) constitute critical feedback mechanisms whose exclusion may introduce systematic biases in ecological impact assessments. These important dimensions, currently beyond the scope of this study, will be prioritized in future investigations to strengthen the socio-ecological validity of our models. Additionally, this study has not yet conducted multi-scenario simulations considering possible future limiting factors. Since LU changes are strongly influenced by sudden policy and economic factors, the LU change predicted by the PLUS model reflects the future trends under conditions of inertial development. Therefore, we plan to incorporate the constraints of long-term macro policies into the assessment model in the future to better guide policy formulation, HQ improvement, and the sustainable development of ecosystem services in more research areas, including the Poyang Lake Region.

6. Conclusions

This paper analyzes the spatial and temporal evolution characteristics and prediction of HQ in the Poyang Lake area. The findings are outlined as follows:

(1) In the twenty years after 2000, forest land, shrub land, and sparse woodland in the Poyang Lake urban agglomeration continued to decrease, primarily converted into urban land and other construction land. Paddy fields and dry land were mainly concentrated in the Poyang Lake plain, experiencing severe loss, with the loss concentrated along the edges of cities such as Nanchang. Construction land showed a rapid expansion trend, with the expansion directly corresponding to the loss of forest and cultivated land. The PLUS model predicts that from 2020 to 2050, areas of forest land, shrub land, paddy fields, and dry land will continue to decline, while urban land and other construction land will expand further.

(2) From 2000 to 2020, the HQ in the Poyang Lake urban agglomeration remained generally good, with a slight decreasing trend in the average value. There were significant regional differences. High-quality habitat areas are mainly distributed in the northeastern, northwestern, eastern, and southern mountainous and central hilly regions of Jiangxi Province. Low-quality habitat areas were found in cities such as Nanchang and the Jitai Plain. Generally, it is lower in the center, higher around the edges, lower in the north, and higher in the south, with cities like Nanchang at the center. The spatial patterns of degraded areas and low-quality habitat areas showed commonalities, indicating that low-quality habitat areas are expanding.

(3) The habitat quality in the Poyang Lake urban agglomeration is projected to maintain a declining trend from 2030 to 2050 compared to 2020 levels, with the most pronounced degradation occurring in peri-urban areas surrounding Nanchang city and other urban centers. Spatiotemporal analysis reveals this deterioration follows two distinct patterns: concentric urban expansion rings around Poyang Lake, and an east–west urban development corridor connecting Pingxiang–Yichun–Xinyu–Nanchang–Fuzhou–Yingtan–Shangrao cities.

These findings provide valuable references for policymakers to formulate ecological conservation policies aimed at improving habitat quality in the Poyang Lake Region. Based on the research outcomes, we recommend implementing concrete measures including rigorous enforcement of cultivated land protection policies with enhanced regulatory oversight to curb disorderly development, systematic vegetation restoration initiatives in key ecologically vulnerable areas to improve ecosystem stability, and prioritized development of ecological corridor networks along urban expansion pathways to maintain and enhance intra-ecosystem connectivity. In future research, we will focus on operationalizing these findings to explore habitat quality improvement models for the Poyang Lake Region, thereby providing more targeted guidance for optimizing ecological compensation mechanisms, refining land use governance systems, and implementing scientifically grounded ecological restoration policies.