Achieving balance between socioeconomic development and ecosystem conservation via policy adjustments in Guangdong Province of southeastern China

Abstract

Rapid urbanization improves socioeconomic development but challenges ecosystem sustainability. Meanwhile, the gradient responses of ecosystem services (ESs) to landscape structures and associated regime shifts of the agriculture–ecosystem–economy nexus (AEEN) have not been sufficiently addressed, preventing an effective balance between socioeconomic prosperity and ecosystem conservation. To bridge this knowledge gap, this study selected the Guangdong Province of southeastern China to explore landscape dynamics from 1985 to 2020 and their spatially heterogeneous impacts on ESs and the AEEN, based on Integrated Valuation of Ecosystem Services and Trade-offs approach and other biophysical models as well as statistical records about socioeconomic factors. AEEN elements, including ESs, responded directly to policy adjustments in terms of ecosystem restoration and landscape management and presented remarkable regime shifts (i.e., phase changes) and spatial heterogeneity. Aggressive agricultural reclamation before 1999 increased crop productivity but caused vegetation degradation and biomass decline. Accelerated urban expansion and ecosystem restoration efforts have improved economic and ecological benefits but have substantially reduced crop productivity and threatened food security. However, timely policy adjustments since 2009 reversed the declining trend and maintained the grain supply. Landscape composition presented patterns of gradual decline along the urban–rural gradient, which in turn determined ES gradient patterns. For instance, water yield and nitrogen export positively correlated with each other (p < 0.0001) but negatively correlated with other ESs. Our study enriches the understandings of social–ecological systems’ response to man-made interventions from AEEN perspective allowing for spatial variabilities and regime shifts, which support policy formulation for coordinating ecological and economic benefits.

Graphical Abstract

Appendix

Model parameterization and validation

WY validation

In the WY assessment, meteorological variables including precipitation, potential evapotranspiration (PET), and actual evapotranspiration (AET) normally introduce uncertainties and errors (Sharp et al. 2018), which should be validated by independent data sources. Therefore, the annual precipitation and PET in grid format were derived from the Chinese Academy of Sciences (http://www.resdc.cn/) and NASA (https://modis.gsfc.nasa.gov/), respectively. The raster format precipitation dataset is interpolated by site-based in situ observations, while the PET is directly obtained from MOD16 product. Owing to the direct contributions of precipitation and AET to runoff generation, the AET is a feasible proxy among the model output variables for hydrological model validation, as the WY is difficult to be measured at the regional scale (Yang et al. 2021). This study cross-compared the model simulated AET with the data from published reports on the same study site and research period (Fan and Zhang 2013; Li et al. 2016). The results showed that the regional average model simulated AET and observed AET for eight sub-regions basically presented a consistent trend and significant linear correlation, with a R² value of 0.9389 (p < 0.0001; Fig. 10a).

Fig. 10

A comparison of estimated results in this study with those in existing literatures at regional or city scale: (a) AET, (b) NPP, and (c) soil loss

NPP validation

In the carbon sequestration assessment indicated by NPP, the most important input parameter that largely determines simulation accuracy is maximum light-use efficiency of vegetation under ideal conditions (${\varphi }_{max}$; Li et al. 2021), which was derived from Zhu et al. (2006) in this study. Owing to the scale difference between site observation and model simulation, this study simply compared the annual mean level of city scale NPP with that reported in existing literatures (Chen et al. 2017, 2019b, a) to validate the result accuracy. The results showed that the regional average model simulated NPP and results given by Chen (2019) for all the cities in Guangdong Province basically presented a consistent trend and significant linear correlation, with a R² value of 0.8695 (p < 0.0001; Fig. 10b). Although the estimated results are not identical because of different input parameters and spatial extents, they do not influence the reliability of AEEN assessment and related analyses.

Soil loss validation

For the validation of the soil loss estimated by RUSLE, this study concurrently conducted the validation based on model-estimated and actual-observed soil erosion modulus. In the comparison between erosion modulus estimated by Ouyang et al. (2016) and current study, the results indicated that the modeled values for the two studies were consistent (i.e., a significant linear correlation), with a R² value of 0.9466 (Fig. 10c). Meanwhile, for validation purposes, the model-estimated erosion moduli were compared against sediment load observed by gauging stations. Although the two indicators have different definitions and reflect different aspects of soil erosion, they presented similar change trends between 1985 and 2020, which demonstrated that the RUSLE performed well in soil loss monitoring.